helical structure and circular dichroism spectra of dna: a ... · b-dna, have right-handed...

14
Helical Structure and Circular Dichroism Spectra of DNA: A Theoretical Study Tomoo Miyahara, Hiroshi Nakatsuji,* ,and Hiroshi Sugiyama Quantum Chemistry Research Institute, JST, CREST, Kyodai Katsura Venture Plaza, North Building 107, 1-36 Goryo-Oohara, Nishikyo-ku, Kyoto 615-8245, Japan Department of Chemistry, Graduate School of Science, Kyoto University, Kitashirakawa-Oiwakecho, Sakyo-Ku, Kyoto 606-8502, Japan * S Supporting Information ABSTRACT: The helical structure is experimentally determined by circular dichroism (CD) spectra. The sign and shape of the CD spectra are dierent between B-DNA with a right-handed double-helical structure and Z-DNA with a left-handed double-helical structure. In particular, the sign at around 295 nm in CD spectra is positive for B-DNA, which is opposite to that of Z-DNA. However, it is dicult to determine the helical structure from the UV absorption spectra. Three important factors that aect the CD spectra of DNA are (1) the conformation of dG monomer, (2) the hydrogen-bonding interaction between two helices, and (3) the stacking interaction between nucleic acid bases. We calculated the CD spectra of (1) the dG monomer at dierent conformations, (2) the composite of dG and dC monomers, (3) two dimer models that simulate separately the hydrogen-bonding interaction and the stacking interaction, and (4) the tetramer model that includes both hydrogen-bonding and stacking interactions simultaneously. The helical structure of DNA can be claried by a comparison of the experimental and SAC-CI theoretical CD spectra of DNA and that the sign at around 295 nm of the CD spectra of Z-DNA reects from the strong stacking interaction characteristic of its helical structure. 1. INTRODUCTION All living organisms carry genetic information. 1 Eukaryotic organisms store the genetic information in the DNA in the cell nucleus. DNA is a polymer composed of purine or pyrimidine bases, deoxyriboses, and phosphate groups and forms a characteristic helical structure. DNA can exist in many dierent conformations. 2 In solution, the most common forms, A- and B-DNA, have right-handed double-helical structure, while Z- DNA is left-handed and appears only when DNA has a special sequence. The transition from B- to Z-DNA or from Z- to B- DNA can be induced by changes in the buer concentration or the temperature. 38 DNA is a complex dynamic system and exible in solution. It is dicult to make a simple model for DNA in solution, and also, it is dicult to analyze the transition processes from B- to Z-DNA. So, basic questions may arise: Is it possible to use a simple model to study DNA in solution? Can we understand the important aspects of DNA from the calculations using such a small simple model? Circular dichroism (CD) spectroscopy is widely used to study the helical structure of DNA in solution. 9 The experimental ultraviolet (UV) spectra are similar between B- and Z-DNA because the electronic excitations of DNA are mainly intramolecular excitations within the nucleic acid bases. However, the experimental circular dichroism (CD) spectra of B- and Z-DNA are very dierent, even when the B- and Z-DNA are composed of the same sequences of nucleotides. 48 These dierences appear even in the CD spectra of a DNA sequence composed of only six base pairs. The sign of the CD spectrum is used to identify the helical structure of DNA. However, only a few theoretical studies have investigated why the sign of the CD spectrum of DNA changes depending on the type of its helical structure. The time-dependent density functional theory (TDDFT) 10,11 was applied to the UV spectra of the DNA bases 12,13 and to the CD spectra of double-helical DNA using adenine-thymine dimer models to distinguish cross-strand base pairs and WatsonCrick base pairs 14 and of single-strand DNA using adenine dimer and tetramer to validate the structure of molecular dynamics calculations. 15 However, the TDDFT has some serious problems in Rydberg excitations, 16 in charge transfer excitations, 17 in double excitations, 18 etc. In addition, the TDDFT calculations of the CD spectra did not show sucient accuracy. 19 To overcome these problems, the improvement of TDDFT were done. 20,21 For example, the long-range corrected (LC)-TDDFT theory 22,23 improved some of these problems but not the problems of the CD spectra. However, the symmetry-adapted cluster (SAC)-conguration interaction (CI) theory 2427 has been applied to many dierent systems and gave highly reliable results for the excited states of many dierent types and properties 28,29 including the CD spectra. 30 This is so as long as proper active space and proper Received: August 29, 2012 Revised: December 10, 2012 Published: December 12, 2012 Article pubs.acs.org/JPCA © 2012 American Chemical Society 42 dx.doi.org/10.1021/jp3085556 | J. Phys. Chem. A 2013, 117, 4255

Upload: others

Post on 16-Jul-2020

2 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Helical Structure and Circular Dichroism Spectra of DNA: A ... · B-DNA, have right-handed double-helical structure, while Z-DNA is left-handed and appears only when DNA has a special

Helical Structure and Circular Dichroism Spectra of DNA ATheoretical StudyTomoo Miyaharadagger Hiroshi Nakatsujidagger and Hiroshi SugiyamaDagger

daggerQuantum Chemistry Research Institute JST CREST Kyodai Katsura Venture Plaza North Building 107 1-36 Goryo-OoharaNishikyo-ku Kyoto 615-8245 JapanDaggerDepartment of Chemistry Graduate School of Science Kyoto University Kitashirakawa-Oiwakecho Sakyo-Ku Kyoto 606-8502Japan

S Supporting Information

ABSTRACT The helical structure is experimentally determined bycircular dichroism (CD) spectra The sign and shape of the CD spectra aredifferent between B-DNA with a right-handed double-helical structure andZ-DNA with a left-handed double-helical structure In particular the signat around 295 nm in CD spectra is positive for B-DNA which is oppositeto that of Z-DNA However it is difficult to determine the helical structurefrom the UV absorption spectra Three important factors that affect theCD spectra of DNA are (1) the conformation of dG monomer (2) thehydrogen-bonding interaction between two helices and (3) the stackinginteraction between nucleic acid bases We calculated the CD spectra of (1) the dG monomer at different conformations (2) thecomposite of dG and dC monomers (3) two dimer models that simulate separately the hydrogen-bonding interaction and thestacking interaction and (4) the tetramer model that includes both hydrogen-bonding and stacking interactions simultaneouslyThe helical structure of DNA can be clarified by a comparison of the experimental and SAC-CI theoretical CD spectra of DNAand that the sign at around 295 nm of the CD spectra of Z-DNA reflects from the strong stacking interaction characteristic of itshelical structure

1 INTRODUCTION

All living organisms carry genetic information1 Eukaryoticorganisms store the genetic information in the DNA in the cellnucleus DNA is a polymer composed of purine or pyrimidinebases deoxyriboses and phosphate groups and forms acharacteristic helical structure DNA can exist in many differentconformations2 In solution the most common forms A- andB-DNA have right-handed double-helical structure while Z-DNA is left-handed and appears only when DNA has a specialsequence The transition from B- to Z-DNA or from Z- to B-DNA can be induced by changes in the buffer concentration orthe temperature3minus8 DNA is a complex dynamic system andflexible in solution It is difficult to make a simple model forDNA in solution and also it is difficult to analyze the transitionprocesses from B- to Z-DNA So basic questions may arise Isit possible to use a simple model to study DNA in solutionCan we understand the important aspects of DNA from thecalculations using such a small simple modelCircular dichroism (CD) spectroscopy is widely used to

study the helical structure of DNA in solution9 Theexperimental ultraviolet (UV) spectra are similar between B-and Z-DNA because the electronic excitations of DNA aremainly intramolecular excitations within the nucleic acid basesHowever the experimental circular dichroism (CD) spectra ofB- and Z-DNA are very different even when the B- and Z-DNAare composed of the same sequences of nucleotides4minus8 Thesedifferences appear even in the CD spectra of a DNA sequence

composed of only six base pairs The sign of the CD spectrumis used to identify the helical structure of DNA However onlya few theoretical studies have investigated why the sign of theCD spectrum of DNA changes depending on the type of itshelical structure The time-dependent density functional theory(TDDFT)1011 was applied to the UV spectra of the DNAbases1213 and to the CD spectra of double-helical DNA usingadenine-thymine dimer models to distinguish cross-strand basepairs and WatsonminusCrick base pairs14 and of single-strand DNAusing adenine dimer and tetramer to validate the structure ofmolecular dynamics calculations15 However the TDDFT hassome serious problems in Rydberg excitations16 in chargetransfer excitations17 in double excitations18 etc In additionthe TDDFT calculations of the CD spectra did not showsufficient accuracy19 To overcome these problems theimprovement of TDDFT were done2021 For example thelong-range corrected (LC)-TDDFT theory2223 improved someof these problems but not the problems of the CD spectraHowever the symmetry-adapted cluster (SAC)-configurationinteraction (CI) theory24minus27 has been applied to many differentsystems and gave highly reliable results for the excited states ofmany different types and properties2829 including the CDspectra30 This is so as long as proper active space and proper

Received August 29 2012Revised December 10 2012Published December 12 2012

Article

pubsacsorgJPCA

copy 2012 American Chemical Society 42 dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus55

basis sets are used Actually the SAC-CI method was used asthe reference data for confirming the accuracy of the LC-TDDFT method2223

Actually it is difficult to model DNA in solution becauseDNA is a complex dynamic system However when we use asmall DNA system whose UV and CD spectra are well studiedexperimentally it would be possible to calculate the UV andCD spectra of such DNA in high accuracy by the SAC-CImethod and by comparing with the experimental data wewould be able to elucidate some basic aspects of the complexphenomena In this article we study the UV and CD spectra ofDNA and their relationships to the helical structures using thesmall models taken from B- and Z-DNA composed of the samesequence of only six deoxyguanosine (dG) and deoxycytidine(dC) We want to show that even such a small model can clarifysome aspects of the relationships between the CD spectra andthe helical structure of DNA when we use a highly reliablecomputational method like the SAC-CI methodThe difference between B- and Z-DNA is the helical-

structure as well as the conformation of dG The purine basedG has the anti conformation in B-DNA while it has synconformation in Z-DNA3 The anti- and syn-dG differ in thedihedral angle between the guanine of the purine base anddeoxyribose by approximately 180deg However the pyrimidinebase dC exists as only the anti conformation in both B- and Z-DNA because of the steric hindrance between the pyrimidinebase cytosine and deoxyriboseThere are three main interactions in DNA (1) the

interaction through the bond between the nucleic acid basesand the deoxyribose (2) hydrogen-bonding interactionsbetween two nucleic acid bases and (3) stacking interactionsbetween two nucleic acid bases Our goal is to understand theroles of these interactions in forming the helical structure andhow these interactions affect the CD spectra of B- and Z-DNAWe have applied the symmetry-adapted cluster (SAC)-

configuration interaction (CI) theory24minus27 to elucidate theunderlying relationships between the helical structure of DNAand the sign of their CD spectra The SAC-CI theory is areliable method for studying the UV and CD spectra in relationto the electronic and geometrical structures of the biologicalsystem and chiral compounds30minus34 We calculated the UV andCD spectra of the monomer of dG in various conformationsincluding both anti and syn conformations For the dCmonomer however we performed calculations only for theanti geometries observed in B- and Z-DNA because theconformation of dC is rigid in DNA For purposes ofcomparison we used an experimentally acquired spectrum ofdG obtained in aqueous solution35 We also studied the UV andCD spectra of DNA First we approximated the spectra as asimple sum of the spectra of dG and dC that have the structurein B- and Z-DNA Next we calculated the UV and CD spectrausing dimer models of B- and Z-DNA that represented eitherthe hydrogen-bonding interaction or the stacking interactionsFinally we calculated the UV and CD spectra using tetramermodels of B- and Z-DNA that included both of the hydrogen-bonding and stacking interactions within the same model All ofthe models include the through-bond interactions between thenucleic acid base and the deoxyribose

2 MODELING AND COMPUTATIONAL DETAILSThe SAC-CI theory is a useful coupled-cluster type electroncorrelation theory for studying ground excited ionized andelectron-attached states of molecules published in 197824minus27

The so-called EOM-CC method published about a decadelater3637 is very similar to the SAC-CI method although calleda different name This equivalence was shown even numericallyin the previous publications3839 For the calculations of the CDspectra of large molecules the gauge-invariant velocity formmust be used since the position of the gauge origin is notknown The independence of the SAC-CI method from thegauge origin has already been determined for the three-membered ring compounds30

First we studied the dependence of the CD spectrum on theconformation of the dG monomer The ground stategeometries with both anti and syn conformations of dG wereoptimized with Gaussian0340 using the density functionaltheory (DFT)41minus44 with the B3LYP functional4546 for the 6-31G(dp) basis4748 sets All of the geometrical parameters ofdG except for the dihedral angle φ were optimized at every 20degin the potential energy for the ground state of dG Fulloptimization was done for anti and syn conformations Thesolvent effects of water were considered using the polarizablecontinuum model (PCM)49 that was connected recently withthe SAC-CI method in the Gaussian suite of programs50

We then analyzed the UV and CD spectra of B- and Z-DNAobtained using the dimer model which contains either thehydrogen-bonding interaction or the stacking interaction andnext using the tetramer model which contains both thehydrogen-bonding and stacking interactions within the samemodel (Figure 1) The geometries were taken from the X-raycrystallography structure 9BNA and 1DCG were used for B-and Z-DNA respectively In B-DNA the GC pair weakly stackswith both of the lower and upper GC pairs (diamond-shaped

Figure 1 (a) Hydrogen-bonding (Z-Hbond) and stacking (Z-Stack)models taken from the X-ray crystallography structure (1DCG) of Z-DNA (b) Hydrogen-bonding (B-Hbond) and stacking (B-Stack)models taken from the X-ray crystallography structure (9BNA) of B-DNA (c) Tetramer model (Z-Tetra) taken from the X-raycrystallography structure (1DCG) of Z-DNA (d) Tetramer model(B-Tetra) taken from the X-ray crystallography structure (9BNA) ofB-DNA

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5543

box in Figure 1bd) but in Z-DNA the guanine-cytosine (GC)pair strongly stacks with only one GC pair (denoted by thesquare box Figure 1ac) The stacking model of Z-DNA (Z-Stack model) has strong stacking interactions As in thegeometry optimization for dimer and tetramer models only thehydrogen atom positions in these models were optimized usingthe B3LYP6-31G(dp) basis with the PCM of waterFor the SACSAC-CI calculations the basis functions

employed were D9551 sets for deoxyribose and D95(d)51 setsfor the nucleic acid bases The core orbitals were treated asfrozen orbitals and all singles and selected doubles wereincluded Perturbation selection52 was carried out for thedouble excitation operators using the threshold sets of 5 times 10minus7

and 1 times 10minus7 hartree for SAC and SAC-CI calculationsrespectively In the calculations for the tetramer model thebasis functions employed were D95 sets (without polarizationfunctions) because the calculations were heavier than thecalculations for the dimer model and further the orbitalswhose energies were within minus11 to +11 au were chosen as theactive orbitals The SAC-CI UV and CD spectra wereconvoluted using Gaussian envelopes to describe the FranckminusCondon widths and the resolution of the spectrometer The fullwidth at half-maximum (fwhm) of the Gaussian was 08 eVwhich was estimated from the criterion that the two bands (253and 278 nm) looks like a single band in the experimental UVspectrum of dG the lowest two bands were not divided clearlyThis value (08 eV) was also used for the convolution of thetheoretical CD spectra Since DNA is a polymer composed ofmonomers we used the same value in dimer and tetramercalculations The UV and CD spectra of (1) dG with severalconformations (2) only dG and only dC in the X-raycrystallography structures of B- and Z-DNA (3) the hydro-gen-bonding and stacking models of B- and Z-DNA and (4)the tetramer models of B- and Z-DNA were calculated usingthe SAC-CI method

3 RESULTS AND DISCUSSION

31 Structure of the Ground State dG Monomer inWater We calculated the changes in the potential energy forthe ground state of the dG monomer in water as a function ofchanges in the dihedral angle (φ) between the deoxyribose andthe guanine (Figure 2) All of the geometrical parameters of dGwere optimized and φ was varied by incremental increases of20deg The optimized dihedral angle was calculated to be 3103degin the anti region and at 1295deg for the syn region The

deoxyribose has C3prime-endo conformation for both anti- and syn-dG with the present calculations although for anti-dG C2prime-endo was calculated to be more stable than C3prime-endo5354Thus in dG the conformation of the deoxyribose seems to bevery flexible and its energy difference very small However theconformation of deoxyribose is rigid in DNA the deoxyribosein dG is C2prime-endo for B-DNA and C3prime-endo for Z-DNA basedon the X-ray crystallography structures (9BNA and 1DCG)The anti-dG was calculated to be more stable than the syn-dGby 047 kcalmol The conformation of dG converts easily fromanti to syn (or from syn to anti) because the energy barrierbetween the anti and syn forms is only sim5 kcalmol The anti-dG rotates more easily in the clockwise direction (the directionof the arrow in Figure 2) than in the counterclockwise directionbecause the energy barrier is less than 4 kcalmol in theclockwise direction but is more than 5 kcalmol in thecounterclockwise direction The potential energy curve is flat inthe anti region while it is sharper in the syn region becausethere is a hydrogen bond interaction between the hydroxylgroup of the deoxyribose and the nitrogen atom of the guaninenear the syn conformation However the anti-dG can rotatemore freely between φ = 270deg and 330deg because of the lack ofhydrogen-bonding interaction

32 UV and CD Spectra of dG Monomer in WaterUsing the SAC-CI method we calculated the excitationenergies oscillator strengths rotatory strengths and the naturesof the excited states for both of the anti- and syn-dG (Table 1)The difference in the UV excitation spectra between anti- andsyn-dG is very small within 007 eV in energy and within 008au in oscillator strength for all the calculated states Howeverfor the CD spectra whose shapes are described by the rotatorystrengths the spectra of anti-dG differ markedly from thespectra of syn-dG not only in the absolute value but also in thesign of the rotatory strength Below we show these theoreticalresults in the spectral forms to facilitate the comparisons withthe experimentally obtained data

321 UV Spectra of dG Monomer The SAC-CI UV spectra(red lines) of anti- and syn-dG were compared with theexperimental spectrum (black lines) of dG above 200 nm(Figure 3) There are two main bands (276 and 250 nm) above200 nm and one band below 200 nm The SAC-CI UV spectraof both anti- and syn-dG are similar to the experimental UVspectrum and therefore it is unclear whether the experimentalUV spectrum of dG is due to the anti or the syn conformationThe first band at about 276 nm was assigned to the π rarr πhighest occupied molecular orbital (HOMO)minuslowest unoccu-pied molecular orbital (LUMO) transition (11A state at 446 eV(278 nm) for anti-dG and 440 eV (282 nm) for syn-dG) Thesecond band at approximately 250 nm was assigned to the 31Astate (at 510 eV (243 nm) for anti-dG and 506 eV (245 nm)for syn-dG) of π(HOMO) rarr π(next-LUMO) and 21A state(at 508 eV (244 nm) for anti-dG and 503 eV (247 nm) forsyn-dG) of the n rarr π nature As expected from its nature theoscillator strength of 21A is weak (005 au) for anti-dG and veryweak (001 au) for syn-dG The states 41A 51A and 61A (at591 597 and 608 eV (210 208 and 204 nm) for anti-dG and590 598 and 608 eV (210 207 and 204 nm) for syn-dG)were calculated to be in the region of the third band atapproximately 214 nm The oscillator strength is almost zerofor the n rarr π 41A and 61A states but is not zero for the π rarrπ 51A state However these states were not observed in theexperimental UV spectra because these three states are in themedian of the two strong peaks (31A and 71A states) and their

Figure 2 Potential energy curve of the ground state of dG determinedas dihedral angle φ is varied in increments of 20deg Other geometricalparameters were optimized at each φ

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5544

oscillator strengths are comparatively small The 71A state (at662 eV (187 nm) for anti-dG and 654 eV (189 nm) for syn-dG) was π rarr π (next-HOMO to LUMO and next-LUMO)excitation which was assigned to the fourth band comparedwith the experimental UV spectrum of deoxyguanosinemonophosphate (dGMP)55 Thus the SAC-CI UV spectra ofboth anti- and syn-dG are similar to the experimental UVspectrum The differences between the anti- and syn-dG werevery small because the HOMO next-HOMO LUMO andnext-LUMO of the main configuration for 11A 31A and 71Astates with large oscillator strengths are localized on theguanine site Hence we cannot determine if the experimentalUV spectrum reflects that of either anti- or syn-dG322 CD Spectra of dG Monomer in Water The SAC-CI

CD spectra (red lines) of the dG monomer at several dihedralangles φ were compared with the experimental CD spectrum(black line) above 200 nm (Figure 4) In the experimentallyacquired spectra the first band is at 276 nm which is strong inthe UV spectrum but very weak in the CD spectrum Thesecond band is at 250 nm which is strong in both UV and CDspectra but has a negative sign in the CD spectrum The thirdband at 214 nm is weak in the UV spectrum but stronglypositive in the CD spectrumThe SAC-CI CD spectra of dG with φ = 10minus150 and 350deg

(Figure 4aminushr) are quite different from the experimental CDspectrum The SAC-CI CD spectrum of syn-dG (1295deg) isopposite in sign for the second band as compared with theexperimental CD spectrum (Figure 4g) For dG with φ = 170210minus290deg (Figure 4ikminuso) the second band of the SAC-CI CDspectra agrees with the experimental CD spectrum but thethird band of the SAC-CI CD spectra has a weak intensitycompared with the experimental CD spectrum The SAC-CICD spectrum of the anti-dG and dG with φ = 330deg (Figure4pq) is in good overall agreement with the experimental CD

spectrum with the second band being strongly negative and thethird band being strongly positiveAs shown in Figure 2 the potential energy curve is flat

around the anti conformer and the energy barrier is not largeSo we calculated the ratio of the existence of each conformerassuming the Boltzmann distribution at 275 K a normalexperimental condition and using the ground state B3LYP6-31G(dp) energies In Table 2 the column labeled ldquoallrdquo givesthe ratio of each conformer when averaging included all 18conformers ldquoantirdquo gives the ratio of each conformer when onlyseven conformations around anti are used and ldquosynrdquo gives theratio of each conformer when only five conformations aroundsyn are used More than 70 of dG exists near the anticonformation between 270deg and 330deg and about 11 existsnear the syn-dG at 1295deg Within the syn region (90minus170deg)more than 50 dG exist as the syn-dG (1295deg) However inthe anti region (250minus10deg) dG exists mostly between 270deg and330deg as anti-dG The gradient of the potential energy is sharparound the syn conformer but flat around the anti conformerbecause dG forms a hydrogen bond with the deoxyribose onlyin the syn conformation Therefore for the CD spectra we mayconsider only the conformer at 1295deg for the syn conformerbut we have to consider several conformers around 3103deg forthe anti conformerWe also carried out the Boltzmann averaging for the spectra

of the anti (Figure 4p) and syn (Figure 4g) conformations TheSAC-CI CD spectra of ldquoallrdquo (green) ldquoantirdquo (blue) and singleanti-dG (red) are similar to the experimental CD spectrum asshown in Figure 4p while the SAC-CI CD spectra of ldquosynrdquo(blue) and single syn-dG (red) are opposite in sign to theexperimental CD spectrum (Figure 4g) These results confirmthat anti-dG is the main conformation under the experimentalcondition We note that the SAC-CI CD spectrum of dG at thesingle geometry of φ = 330deg is more similar to the experimentalCD spectrum of that of anti-dG (3103deg) This may indicatethat dG with φ = 330deg is more stable than the anti-dG insolutionThe dependence on the dihedral angle φ is very small in the

UV spectrum of Figure 3 because only π rarr π has strongintensity However the CD spectra of Figure 4 strongly dependon the dihedral angle φ because both π rarr π and n rarr π havestrong intensities and their MOs mainly localized at theguanine site are somewhat extended to the deoxyribose The11A state assigned to the first band at 276 nm was calculated tohave a small rotatory strength for both anti- and syn-dG butthis state was not observed in the experimental CD spectrumIn the second band region (250 nm) the rotatory strengths of

Table 1 SAC-CI UV and CD Spectra of anti- and syn-dG

SAC-CI

anti-dG syn-dG

EEa oscb rotc EEa oscb rotc

state (eV) (nm) (au) (10minus40 cgs) (eV) (nm) (au) (10minus40 cgs) nature exptld (eVnm)

11A 446 278 016 453 440 282 016 244 π rarr π 44927621A 508 244 005 minus7631 503 247 001 3812 n rarr π 49625031A 510 243 020 6820 506 245 022 minus3461 π rarr π 49625041A 591 210 000 1731 590 210 000 minus292 n rarr π 57921451A 597 208 002 minus225 598 207 005 830 π rarr π 57921461A 608 204 000 minus215 608 204 000 minus270 n rarr π 57921471A 662 187 044 minus782 654 189 038 minus078 π rarr π

aExcitation energy bOscillator strength cRotatory strength dRef 35

Figure 3 SAC-CI UV spectra of (a) anti- and (b) syn-dG (red)compared with the experimentally determined UV spectrum35 of dG(black)

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5545

both the 21A and 31A states are large and opposite in sign(Table 1) The 21A state is calculated to lie at 508 eV (244nm) for anti-dG and at 503 eV (247 nm) for syn-dG The 31Astate is calculated to lie at 510 eV (243 nm) for anti-dG and at506 eV (245 nm) for syn-dG The difference between 21A and31A in the excitation energy is very small so that their rotatorystrengths cancel each other giving relatively small rotatorystrengths From Table 1 we see that the total rotatory strengthof the second band is negative for anti-dG and positive for syn-dG To reproduce the experimental CD spectra the theoreticalrotatory strength must be reliable since otherwise reliablepositive or negative CD peaks cannot be obtained The SAC-CItheory seems to give reliable rotatory intensity since thetheoretical CD spectra agree well with the experimental one Inthe third band region (214 nm) the sum of the rotatorystrengths is positive for both anti- and syn-dG although thestrongest rotatory strength comes from the 41A (n rarr π) statefor anti-dG but from the 51A (π rarr π) state for syn-dG As inthe experimental CD spectrum the third band has largerintensity than the second band in the SAC-CI CD spectrum ofanti-dG However for syn-dG the third band in the SAC-CI

CD spectrum is smaller than the second band The 71A stateassigned to the fourth band with a strong negative intensity at190 nm in the experimental CD spectrum of dGMP55 isnegative for anti-dG and almost zero for syn-dG

33 Composite CD Spectra of dG and dC Comparedwith the CD Spectra of DNA We then extended our studiesto the spectra of B- and Z-DNA and we compared the SAC-CICD spectra of anti- and syn-dG (Figure 4) with theexperimental CD spectra of B- and Z-DNA (Figure 5)4 Thesecond band at 250 nm in the SAC-CI CD spectrum of anti-dGis similar to the experimental CD spectrum of B-DNA whilethe SAC-CI CD spectrum of syn-dG is similar to theexperimental CD spectrum of Z-DNA This result may indicatethat in DNA the second band at 250 nm reflects the dihedralangle φ of dG between the deoxyribose and the guanineTherefore we calculated the composite SAC-CI CD spectra ofthe noninteracting dG and dC monomer from the available X-ray crystallographic structures in B- and Z-DNA to clarify if thedifferent conformations of dG andor dC in different types ofDNA explain the differences in the CD spectra

Figure 4 CD spectra of dG at several conformation angles φ The experimental CD spectrum35 (black line) of dG is compared with the SAC-CI CDspectra (red lines) of (a) φ = 10deg (b) φ = 30deg (c) φ = 50deg (d) φ = 70deg (e) φ = 90deg (f) φ = 110deg (g) φ = 1295deg (syn) (h) φ = 150deg (i) φ = 170deg(j) φ = 190deg (k) φ = 210deg (l) φ = 230deg (m) φ = 250deg (n) φ = 270deg (o) φ = 290deg (p) φ = 3103deg (anti) (q) φ = 330deg and (r) φ = 350deg TheSAC-CI CD Boltzmann averaged spectra of (g) syn (blue line) (p) anti (blue line) and (p) all (green line) are also shown

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5546

The SAC-CI CD spectra of dG and dC alone (Figure5abde) having the structures of those in B- and Z-DNA andthe composite CD spectra (Figure 5cf) obtained as (a + b andd + e) were compared to the experimental CD spectra of B-and Z-DNA The SAC-CI excitation energies oscillatorstrengths rotatory strengths and natures of the excited statesfor the monomer model (dG and dC) were also determined(Table 3)The conformations of dG and dC are the same as those in

the ideal B-DNA or Z-DNA in which dG and dC are arrangedalternately (Figure 1) However the X-ray structure wasobtained for the DNA composed of only 6 sequences in thecase of Z-DNA Since the geometry is affected by the effect ofthe edge of DNA the conformation is different for eachmonomer Since the conformation of dG is in particular moreflexible than that of dC in DNA we calculated two kinds of dG(model I and II) in B- and Z-DNA respectively The models Iand II correspond to dG of the hydrogen-bonding models (B-Hbond and Z-Hbond models) and the stacking models (B-Stack and Z-Stack models) of the dimer models respectivelyThe dihedral angle (φ) between the deoxyribose and theguanine is 2763deg and 3007deg respectively for models I and IIof B-DNA and 1113deg and 1175deg respectively for models Iand II of Z-DNA The dihedral angle between the deoxyriboseand cytosine is 3232deg for B-DNA and 3329deg for Z-DNA Therotatory strength of dG is very different between models I andII for both B- and Z-DNA as shown in Table 3 Therefore fordG we prepared an averaged CD spectra for models I and II(Figure 5ad) For Z-DNA the SAC-CI CD spectrum of dG(Figure 5a) is negative for the first band and positive for thesecond band as in the experimental spectrum of Z-DNAalthough the excitation energies are calculated at a higherregion However the SAC-CI result is opposite in sign to theexperimental CD spectrum for the first band of B-DNA (Figure5d) These results show that in the CD spectra of DNA the

conformation of dG (monomer) is related to the sign of thesecond band but has no effect on the sign of the first bandThe CD spectra of dC have four peaks (Figure 5be) The

first peak represents the πminusπ of HOMO to LUMO Thesecond peak is the excitation from the nonbonding orbital tothe LUMO The third and fourth peaks are composed of boththe πminusπ and nminusπ from next-HOMO and nonbondingorbitals to LUMO The third and fourth peaks are mainly πminusπand nminusπ respectively The SAC-CI CD spectra of dC aresimilar in B- and Z-DNA with two positive and two negativepeaks The rotatory strength is positive for the first and thirdpeaks and negative for the second and fourth peaks Thereforethe CD spectra of dC are always positive at around 295 nmcorresponding to the conformation of dC which is always antiin DNAAs shown in Figure 5cf the composite SAC-CI CD spectra

obtained from dG and dC alone are similar between B- and Z-DNA The sign at 295 nm is positive for both B- and Z-DNAbecause the rotatory strength of dC which is the lowest peak ismuch stronger than that of dG Therefore the composite CD

Table 2 Boltzmann Distribution of dG along theConformational Angle φ

angle (φ) all ()a anti ()b syn ()c

10 086 1130 01450 00470 01090 170 91110 279 1431295 (syn) 1131 517150 454 224170 041 25190 008210 000230 010250 325 41270 1534 195290 1700 2163103 (anti) 2669 339330 1307 166350 258 33total 1000 1000 1000

aldquoallrdquo gives the ratio of each conformer when averaging included all 18conformers bldquoantirdquo gives the ratio of each conformer when only sevenconformations around anti are used cldquosynrdquo gives the ratio of eachconformer when only five conformations around syn are used

Figure 5 SAC-CI CD spectra of monomer ((a) dG in Z-DNA (b) dCin Z-DNA (d) dG in B-DNA and (e) dC in B-DNA) and thecomposite CD spectra of (c) Z-DNA and (f) B-DNA obtained fromthe SAC-CI CD spectra of dG and dC compared with theexperimental CD spectra4 (black lines) of Z-DNA (aminusc) and B-DNA (dminusf) The red and blue lines represent the excited state of dGand dC respectively The magenta line in panels c and f represents thecomposite spectra

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5547

spectrum is similar to that of dC for both B- and Z-DNA As aresult in the composite CD spectra of B- and Z-DNA the signat around 295 nm is independent of the geometry of themonomer (dG or dC) even though the CD spectrum of dGdepends on the dihedral angle between the guanine and thedeoxyribose because the lowest peak originates from dCwhose geometry is always anti in both B- and Z-DNA Thisindicates that the noninteracting dG and dC cannot explain theexperimental spectra of DNA and that the hydrogen-bondingor stacking interactions with the neighboring nucleic acid basesmust be important for the CD spectra of DNA34 UV and CD Spectra of Dimer Models In the studies

of the structure spectroscopy and functions of DNA it isgenerally recognized that the hydrogen-bonding interaction andthe stacking interaction are the two most important interactionsthat characterize the properties of DNA So we expect thatthese interactions might also be the important factors that affectthe features of the UV and CD spectra of DNA To investigatethe possible origin we considered the dimer model composedof dG and dC These models contained either the hydrogen-bonding interaction or the stacking interaction (Figure 1)However we note that the SAC-CI results using these modelsneed not be in agreement with the experimental spectrabecause these models lack either the stacking interaction or thehydrogen-bonding interaction We must use the tetramer

model at least to compare with the experimental spectra asnoted below

341 UV Spectra of Dimer Models Figure 6 shows theSAC-CI UV spectra of both hydrogen-bonding and stackingmodels of both B- and Z-DNA compared with the experimentalUV spectrum The experimental UV spectra of Z-DNA displaya main peak at 256 nm and a shoulder peak at around 290 nmThe experimental UV spectrum of B-DNA is very similar butthe shoulder peak at around 290 nm is weaker Table 4 showsthe excitation energies oscillator strengths rotatory strengthsnatures and the types of the excited states of the SAC-CIresults for both hydrogen-bonding and stacking models of thedimer When the excitation in the dimer model is not withindG or dC but corresponds to the electron transfer (ET)between dG and dC we assigned it as the ET typeThe SAC-CI UV spectrum of the hydrogen-bonding model

of Z-DNA (Z-Hbond model) is in good agreement with theexperimental UV spectrum of Z-DNA (Figure 6a) Threestrong peaks of the πminusπ nature are calculated The loweststrong peak corresponds to the shoulder peak at 428 eV (290nm) of the experimental UV spectrum of Z-DNA Theremaining two strong peaks correspond to the main peak at484 eV (256 nm) of the experimental UV spectrum of Z-DNARed blue and green colors represent the excited states of dGdC and ET respectively

Table 3 SAC-CI UV and CD Spectra of dG and dC in Z- and B-DNA

Z-DNA B-DNA

dG (model I φ = 1113deg) dG (model I φ = 2763deg)

EEa oscb rotc EEa oscb rotc

state (eV) (nm) (au) (10minus40 cgs) nature (eV) (nm) (au) (10minus40 cgs) nature

11A 441 281 017 125 π rarr π 443 280 016 minus1035 π rarr π21A 474 262 000 minus1099 n rarr π 488 254 000 1099 n rarr π31A 493 251 023 2066 π rarr π 513 242 025 minus3355 π rarr π41A 578 214 000 minus864 n rarr π 595 208 002 minus2619 π rarr π51A 591 210 004 minus656 π rarr σ 599 207 001 3571 n rarr π61A 611 203 000 minus438 n rarr π 618 201 000 983 n rarr π71A 628 197 000 359 π rarr σ 675 184 008 1235 π rarr σ81A 659 188 030 634 π rarr π 678 183 028 minus2402 π rarr π

dG (model II φ = 1175deg) dG (model II φ = 3007deg)

EEa oscb rotc EEa oscb rotc

state (eV) (nm) (au) (10minus40 cgs) nature (eV) (nm) (au) (10minus40 cgs) nature

11A 437 284 016 minus451 π rarr π 434 286 016 minus247 π rarr π21A 459 270 000 minus322 n rarr π 493 251 000 minus1467 n rarr π31A 491 252 022 minus319 π rarr π 518 239 021 minus200 π rarr π41A 581 213 003 323 π rarr π 595 209 001 2884 n rarr π51A 598 207 000 minus738 π rarr π 599 207 003 minus647 π rarr π61A 621 200 000 796 n rarr π 617 201 000 minus3033 n rarr π71A 641 193 000 353 π rarr σ 660 188 048 978 π rarr π81A 647 191 031 minus077 π rarr π 666 186 001 326 π rarr σ

dC (φ = 3329deg) dC (φ = 3232deg)

EEa oscb rotc EEa oscb rotc

state (eV) (nm) (au) (10minus40 cgs) nature (eV) (nm) (au) (10minus40 cgs) nature

11A 413 300 009 2385 π rarr π 409 303 009 1403 π rarr π21A 442 280 000 minus1548 n rarr π 452 274 000 minus915 n rarr π31A 507 245 011 8392 π + n rarr π 499 249 007 5975 π + n rarr π41A 527 235 011 minus8639 n + π rarr π 527 235 015 minus8017 π + n rarr π51A 573 216 000 minus1385 n rarr π 559 222 001 393 n rarr π61A 632 196 039 2349 π rarr π 625 198 032 1098 π rarr π

aExcitation energy bOscillator strength cRotatory strength

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5548

The shoulder peak in the experimental UV spectrum of Z-DNA is composed of 11A (439 eV) and 21A (453 eV) statesThe 11A state (in blue color) of the Z-Hbond modelcorresponds to the 11A state of dC at 413 eV which is theexcitation from the HOMO to the LUMO of dC This state haslarge oscillator strength of the πminusπ nature In the Z-Hbondmodel this state is blue-shifted by 026 eV because theHOMOminusLUMO gap (0449 au) of the Z-Hbond model islarger than that (0438 au) of dC This state is in goodagreement with the first (shoulder) peak of the experimentalUV spectrum of Z-DNA However the 21A state (453 eV redcolor) of the Z-Hbond model is assigned to the 31A state (493eV) of dG which is the excitation from the HOMO to the next-LUMO of dG This large shift of the excitation energy is causedby the MO-order change The MO of the Z-Hbond modelcorresponding to the next-LUMO of dG becomes lower thanthe MO of the Z-Hbond model corresponding to the LUMOof dG by the formation of hydrogen bonds with cytosine in theZ-Hbond model Therefore this state is red-shifted by 04 eVHowever the oscillator strength of this state is small despite itsπminusπ nature The transition moment is canceled because themain configuration includes not only the excitation from theHOMO to the next-LUMO in dG but also the excitation fromthe next-HOMO to the LUMO in dC and the electron transfertype excitation from the HOMO of dG to the LUMO and next-LUMO of dCThe main peak of the experimental UV spectrum of Z-DNA

is composed of 31A (484 eV) and 51A (493 eV) states havinglarge oscillator strengths (Figure 6a) The 31A state (so blue

color) of the Z-Hbond model is assigned to the 31A state (507eV) of dC which reflects the strong excitation from the next-HOMO to the LUMO of dC The 51A state (red color) of theZ-Hbond model is assigned to the 11A state (441 eV) of dGwhich is the excitation from the HOMO to the LUMO in dGThe excitation energy increases by 052 eV because theHOMOminusLUMO gap becomes large which is the opposite tothat of the 21A state but again due to the hydrogen-bondingeffectFor B-DNA however the SAC-CI UV spectrum of the

hydrogen-bonding model (B-Hbond model Figure 6c) also hasboth main and shoulder peaks in which the main peak is higherand the shoulder peak is lower as compared with theexperimental UV spectrum of B-DNA The main peak (41Aand 51A states) of the SAC-CI spectra of the B-Hbond model ishigher than that of the experimental spectra similarly to theSAC-CI spectrum of the Z-Hbond model The 31A state of dGin B-DNA is also higher than that in Z-DNA (Table 3) Theseresults may be related to the fact that dG is more flexible thandC because the conformation of dG is anti for B-DNA but synfor Z-DNA Since the stacking interaction of B-DNA is weakerthan that of Z-DNA the X-ray structure of B-DNA may bemore flexible than that of Z-DNA However the result that theoscillator strength of the lowest excited state of the B-Hbondmodel is weaker than that of the Z-Hbond model is reflected bythe fact that the shoulder peak of the experimental UV spectrais stronger in Z-DNA but weaker in B-DNAFor the SAC-CI UV spectra of the stacking models of both

B- (B-Stack model) and Z-DNA (Z-Stack model) the firstexcited states originated from dC were calculated at the lower-energy region (higher wavelength region) but the excitationenergy of the main peak is almost the same between the B-Hbond and B-Stack models or between the Z-Hbond and Z-Stack models If there is no hydrogen-bonding interaction inDNA the lowest excited state would be observed in the lower-energy region of the experimental spectra separately from themain peak The excitation energy of the lowest excited state ofthe stacking model is lower than that of the hydrogen-bondingmodel by 044 eV for Z-DNA and 033 eV for B-DNArespectively This result may correspond to the fact that thelowest excitation energy of cytosine differs between the gasphase (428 eV)56 and in solution (464 eV)57

For both B- and Z-DNA the SAC-CI results of thehydrogen-bonding models (B-Hbond and Z-Hbond) are inbetter agreement with the experimental spectra than those ofthe stacking models (B-Stack and Z-Stack) Thus thehydrogen-bonding interaction is reflected in the UV spectraof DNA On the other hand the shoulder peak observed for Z-DNA may reflect the stronger shoulder nature of the Z-Stackmodel in Z-DNA However the UV spectra of DNA providelittle information about stacking interactions between nucleicacid bases

342 CD Spectra of Dimer Models Figure 7 shows theSAC-CI CD spectra of both hydrogen-bonding and stackingmodels of B- and Z-DNA compared with the experimental CDspectra In the experimental CD spectra the first peak isnegative at 296 nm and the second peak is positive at 275 nmfor Z-DNA However B-DNA displays peaks that are oppositein sign the first peak is positive at 280 nm and the second peakis negative at 255 nmIt is worth nothing that the SAC-CI spectrum has a negative

peak at 295 nm only for the stacking model of Z-DNA (Z-Stack) but for the other models the SAC-CI spectrum is

Figure 6 SAC-CI UV spectra for the dimer models The SAC-CIspectra (magenta line) of (a) Z-Hbond model (b) Z-Stack model (c)B-Hbond model and (d) B-Stack model are compared with theexperimental UV spectra4 (black line) of Z-DNA (ab) and B-DNA(cd) The red and blue lines represent the intramolecular excitedstates of dG and dC respectively The green lines represent theelectron transfer (ET) excited state from dG to dC

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5549

Table

4SA

C-CIUVandCD

Spectraof

theHydrogen-Bon

ding

andStacking

Dim

erMod

elsof

Z-andB-D

NA

Z-DNA

B-DNA

SAC-CIof

Z-H

bond

exptle

SAC-CIof

B-H

bond

exptle

EEa

oscb

rotc

UV

EEa

oscb

rotc

UV

state

(eV)

(nm)

(au)

(10minus

40cgs)

nature

typed

(eVnm)

(eV)

(nm)

(au)

(10minus

40cgs)

nature

Typed

(eVnm)

11A

439

282

020

3513

πrarr

πdC

(11 A)

428290

437

284

017

1143

πrarr

πdC

(11 A)

21A

453

274

004

884

πrarr

πdG

(31 A)

456

272

003

minus1481

πrarr

πdG

(31 A)

31A

484

256

018

912

πrarr

πdC

(31 A)

493

251

006

583

πrarr

πET

41A

485

255

005

minus1145

πrarr

πET

498

249

021

1767

πrarr

πdC

(3+41A)

51A

493

252

042

1182

πrarr

πdG

(11 A)

484256

518

240

043

minus1390

πrarr

πdG

(11 A)

484256

61A

512

242

000

057

nrarr

πdC

(21 A)

525

236

000

minus1197

nrarr

πdC

(21 A)

71A

523

237

000

minus1384

nrarr

πdG

(21 A)

529

234

000

minus579

nrarr

πdG

(2+51A)

81A

623

199

000

minus034

nrarr

πdG

(2+51A)

SACminusCIof

Z-Stack

exptle

SACminusCIof

B-Stack

exptle

EEa

oscb

rotc

CD

EEa

oscb

rotc

CD

state

(eV)

(nm)

(au)

(10minus

40cgs)

nature

typed

(eVnm)

(eV)

(nm)

(au)

(10minus

40cgs)

nature

typed

(eVnm)

11A

395

314

009

minus4511

πrarr

πdC

(11 A)

419296

404

307

007

4358

πrarr

πdC

(11 A)

443280

21A

421

294

000

minus3349

πrarr

πdG

(11 A)

419296

440

282

006

minus1032

πrarr

πdG

(11 A)

31A

445

279

001

2967

nrarr

πdC

(21 A)

461

269

000

minus503

nrarr

πdC

(21 A)

41A

472

263

000

minus1335

nrarr

πdG

(21 A)

493

251

013

9407

πrarr

πdC

(3+41A)

51A

485

256

019

minus1796

πrarr

πdG

(31 A)

451275

507

245

005

minus15115

nrarr

πdG

(21 A)

486255

61A

499

248

006

4861

πrarr

πET

451275

512

242

014

minus11056

πrarr

πdG

(31 A)

486255

71A

526

236

007

2721

πrarr

πET

+dC

(31 A)

451275

521

238

014

19475

πrarr

πET+dG

(31 A)

81A

565

219

000

minus231

nrarr

πdC

(41 A)

619

200

000

minus805

nrarr

πdG

(41 A)

aExcitatio

nenergybOscillator

strengthcRotatorystrengthddG

representsintram

olecular

excitedstates

ofdG

dCrepresentsintram

olecular

excitedstates

ofdC

and

ETrepresentselectron

transfer

excitedstates

from

dGto

dCeRef

4

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5550

positive at 295 nm Therefore the strong negative sign of theCD spectra at 295 nm indicates that a strong stackinginteraction exists in Z-DNA The SAC-CI CD spectra of bothstacking models (B-Stack and Z-Stack models) more closelyresemble the experimental CD spectra of B- and Z-DNA thanthose of the hydrogen-bonding models (B-Hbond and Z-Hbond models) For the SAC-CI CD spectrum of the Z-Stackmodel the first and second peaks are negative and positiverespectively but they are the opposite of those obtained fromthe Z-Hbond model For the B-Stack and B-Hbond modelshowever the first peak is positive in the SAC-CI CD spectrabut the second and third peaks are negative and positiverespectively for the B-Stack model in opposition to the B-Hbond model Since the stacking models do not have thehydrogen-bonding interactions the excitation energy of the firstpeak is lower than that of the experimental spectrum as notedaboveIn the Z-Stack model the first peak with a negative sign is

assigned to the 11A (395 eV) and 21A (421 eV) states (Table4) The 11A state of the Z-Stack model corresponds to the 11Astate (413 eV) of dC which is the excitation of the πminusπ fromthe HOMO to the LUMO in dC A large oscillator strength isthe origin of the first (shoulder) peak but the excitation energyis lower than the experimental value because there is nohydrogen bond in the Z-Stack model The rotatory strength ofthe 11A state of dC changes the sign from positive to negativeby the stacking interaction with dG Therefore the stackinginteraction between dG and dC accounts for the strongnegative sign of the CD spectra in Z-DNA

The excitation energy of the 21A state (421 eV) is also closeto the first peak of the experimental CD spectrum This state isassigned to the 11A state (437 eV) of dG which is the HOMOto LUMO πminusπ excitation in dG but also includes thecontribution of the excitation from the next-HOMO to LUMOin dC The oscillator strength in the UV spectrum is smallbecause this state differs from the 11A state of dG by theinclusion of the excitation in dC Although the excitationenergy is shifted by the formation of hydrogen bonds betweendG and dC the rotatory strength of a large negative maycontribute to the strong negative sign of the CD spectra in Z-DNA In the 31A state (445 eV) which is the excitation fromnonbonding orbital to the LUMO in dC the rotatory strengthchanges to positive from the negative sign of the monomer dCHowever this state is not the origin of the second peak as willbe noted later in the tetramer model section Thus the rotatorystrength is strongly affected by the stacking interactionThree important excited states are calculated in the higher-

energy region of the second peak The 51A state (485 eV) hasthe largest oscillator strength and is assigned to the 31A (491eV) state of dG which is the excitation from HOMO to next-LUMO This state is assigned to the main peak (484 eV) of theexperimental UV spectrum of Z-DNA The 6 and 71A states(499 and 526 eV) with a large rotatory strength are the ETstate from the HOMO of dG to the LUMO of dC The mainconfiguration of the 71A state also includes the excitation fromthe next-HOMO to the LUMO in dCIn the B-Stack model the first peak with a positive sign is

assigned to the 11A state (404 eV) This state corresponds tothe 11A state (409 eV) of dC which is the excitation from theHOMO to the LUMO The 21A state (440 eV) is assigned tothe 11A state (434 eV) of dG which reflects the excitationfrom the HOMO to the LUMO Since the B-Stack model doesnot have hydrogen bonds the excitation energies are close tothose of dG or dC Both states have the oscillator strength butthe 11A state is positive while the 21A state is slightly negativefor the rotatory strength which is the same as for themonomers dC and dG respectively Therefore the stackinginteraction of B-DNA has little effect on the sign of the CDspectra at 295 nm because the stacking interaction is weaker inB-DNA than in Z-DNAThe second peak with a negative sign is assigned to the 5 and

61A states with a large rotatory strength The 5 and 61A states(507 and 512 eV in Table 4) are assigned to the 2 and 31Astates (493 and 518 eV in Table 3) of dG which are theexcitations from the nonbonding orbital and the HOMO to thenext-LUMO The main configuration is nminusπ for the 51A stateand πminusπ for the 61A state Both states have strong oscillatorand negative rotatory strengths which means that they can beassigned to the main peaksThe first ET excited states (71A state 521 eV) of B-Stack

model are calculated at a higher region than that (61A state499 eV) of the Z-Stack model because the stacking interactionis weaker in B-DNA than in Z-DNAThe SAC-CI results show that both hydrogen-bonding and

stacking interactions affect the UV and CD spectra of DNA Inparticular the excitation energy of the first excited stateincreases by the formation of hydrogen bonds and the stackinginteraction changes the sign of the CD spectrum of the firstexcited state from positive to negative The strong stackinginteraction characteristic of the helical structure of Z-DNA isthe origin of the strong negative sign at around 295 nm in theCD spectra The hydrogen-bonding and stacking interactions

Figure 7 SAC-CI CD spectra for the dimer models The SAC-CIspectra (magenta line) of (a) Z-Hbond model (b) Z-Stack model (c)B-Hbond model and (d) B-Stack model are compared with theexperimental CD spectra4 (black line) of Z-DNA (ab) and B-DNA(cd) The red and blue lines represent the intramolecular excitedstates of dG and dC respectively The green lines represent theelectron transfer (ET) excited state from dG to dC

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5551

also confer other smaller effects on the UV and CD spectra ofDNA In the higher energy region the SAC-CI CD spectra arenot in good agreement with the experimental CD spectrabecause these models have either hydrogen-bonding interactionor stacking interaction but not both Therefore we need toinvestigate a better model that includes both hydrogen bondingand stacking interactions within the same model A minimumsuch model is the tetramer model of Figure 1 on which weexplain in the next section35 Tetramer Model The SAC-CI calculations of the

dimer models were suitable to investigate the effects of thehydrogen-bonding interaction and the stacking interactionseparately For example the SAC-CI UV spectra of thehydrogen-bonding models were in good agreement with theexperimental UV spectra implying the importance of thehydrogen-bonding interaction The SAC-CI CD spectra for thestacking model clarified that the stacking interaction is thesource of the negative CD peak at around 295 nm Howevergenerally speaking the SAC-CI CD spectra were markedlydifferent from the experimental spectra The SAC-CI UVspectra of the stacking models were composed of two peaksdifferent from the experimental UV spectra Thus theinformation that we gained from the dimer models still didnot fully resolve these differences in the spectra acquired by twodifferent processes So we used the tetramer model (Figure 1)which included both hydrogen-bonding and stacking inter-actions within the same modelFigure 8 shows the SAC-CI UV and CD spectra of the

tetramer model of Z-DNA (Z-Tetra model) and B-DNA (B-Tetra model) in comparison with the experimental UV andCD spectra In order to compare the experimental spectra theSAC-CI results were shifted to lower energy by 05 eV Thecomputational conditions were poor as compared with theother SAC-CI calculations In the tetramer calculations the d-polarization functions were not added and the small activespace was used The SAC-CI calculations using the small activespace and without the d-polarization functions gave higherexcitation energies by about 04minus05 eV (see SupportingInformation Table S1) for the hydrogen-bonding dimermodels In particular the excited states corresponding to themain peak of the experimental UV spectra were shifted by 048eV for both the Z-Hbond model (61A state) and the B-Hbondmodel (71A state)Table 5 shows the excitation energies oscillator strengths

rotatory strengths natures and type of the excited states of theSAC-CI results for tetramer modelsThe SAC-CI UV spectrum of the Z-Tetra model is in good

agreement with the experimental UV spectrum of Z-DNA Themain peak as well as the shoulder peak is reproduced by theSAC-CI method The shoulder peak is assigned to the 4 and51A states which is the πminusπ excitation (HOMO to LUMO)in dC and the main peak is assigned to the 101A state which isthe πminusπ (HOMO to next-LUMO) excitation in dGThe SAC-CI UV spectrum of the B-Tetra model is also in a

reasonable agreement with the experimental UV spectrum of B-DNA although the shape of the shoulder is slightly different Inthe B-Tetra model the shoulder peak is assigned to the 21Astate which is the πminusπ (HOMO to LUMO) excitation in dCand the main peak is assigned to the 81A state which is theπminusπ (HOMO to next-LUMO) excitation in dGThe sign of the rotatory strength is the same between the

SAC-CI CD spectrum of the Z-Tetra model and theexperimental CD spectrum of Z-DNA although the negative

rotatory strength at 295 nm is very weak because thepolarization functions were not included in the tetramermodels It is well-known that the basis set must be good for anadequate description of the stacking interaction5859 In thedimer model the lack of polarization d-functions and thesmaller active space changed the intensity and sign of therotatory strength (see Supporting Information Table S1)Since the stacking interaction is small in B-DNA the SAC-CI

CD spectrum of the B-Tetra model is in good agreement withthe experimental CD spectrum of B-DNA and reproduces notonly the sign of the rotatory strength but also the overall shapeof the spectrumIn Z-DNA the 11A state is assigned to the first peak of the

experimental CD spectrum The 11A state is the ET excitationfrom the HOMO of dG to the LUMO of dC through thestacking conformation The 3 and 51A states with a largenegative rotatory strength are the excitation from the HOMOto the LUMO in dG and dC respectively These states alsocontribute to the first negative peak as in the 1 and 21A stateswith a negative rotatory strength in the Z-Stack model The 8and 101A states are assigned to the second peak of the CDspectrum The 101A state which is the origin of the main peakof the UV spectrum and the πminusπ of dC (81A state) alsocontribute to the second peak of the CD spectrumIn B-DNA the first peak of the CD spectrum is assigned to

the 21A state of πminusπ (HOMO to LUMO) in dC which is thesame as the origin of the first peak of the UV spectrum The ETstates are calculated at the higher energy region than the 21A

Figure 8 SAC-CI UV and CD spectra for the tetramer models TheSAC-CI spectra (magenta line) of Z-Tetra model (ab) and B-Tetramodel (cd) as compared with the experimental UV and CD spectra4

(black line) of Z-DNA (ab) and B-DNA (cd) The red and blue linesrepresent the intramolecular excited states of dG and dC respectivelyThe green lines represent the electron transfer (ET) excited state fromdG to dC All excited states of SAC-CI calculations have been shiftedto the lower values by 05 eV

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5552

Table

5SA

C-CIUVandCD

Spectraof

theTetramer

Mod

elsof

Z-andB-D

NA

Z-DNA

B-DNA

SACminusCIof

Z-Tetra

exptle

SAC-CIof

B-Tetra

exptle

EEa

Oscb

Rotc

UV

CD

EEa

Oscb

Rotc

UV

CD

state

(eV)

(nm)

(au)

(10minus

40cgs)

nature

typed

(eVnm)

(eVnm)

(eV)

(nm)

(au)

(10minus

40cgs)

nature

typed

(eVnm)

(eVnm)

11A

470

264

000

minus663

πrarr

πET

419296

479

259

003

minus041

πrarr

πdG

21A

481

258

002

6611

πrarr

πET

482

257

033

16891

πrarr

πdC

443280

31A

484

256

004

minus5326

πrarr

πdG

419296

484

256

001

minus7454

πrarr

πdC

41A

488

254

015

4553

πrarr

πdC

428290

498

249

007

minus130

πrarr

πET

51A

492

252

012

minus11712

πrarr

πdC

428290

419296

511

243

004

minus622

πrarr

πET

61A

502

247

003

minus3034

πrarr

πET

525

236

000

minus046

πrarr

πET

71A

513

242

001

2737

nrarr

πdG

527

235

001

332

πrarr

πET

81A

515

241

006

14772

πrarr

πdC

451275

545

227

045

minus15609

πrarr

πdG

484256

486255

91A

522

238

001

1978

nrarr

πdG

552

225

002

minus1795

nrarr

πdG

101 A

538

230

071

7299

πrarr

πdG

484256

451275

556

223

013

5929

πrarr

πdG

111 A

541

229

001

minus6858

πrarr

πdG

558

222

009

minus525

πrarr

πdG

121 A

548

226

002

minus1519

nrarr

πdC

563

220

008

minus7436

nrarr

πdG

131 A

553

224

001

minus924

nrarr

πdC

609

204

001

1379

nrarr

πdG

141 A

601

206

001

minus4159

nrarr

πdG

626

198

002

minus464

nrarr

πdG

aExcitatio

nenergyA

llexcitedstates

ofSA

C-CIcalculations

have

been

shifted

tothelower

values

by05eV

inFigure

8bOscillator

strengthcRotatorystrengthddG

representsintram

olecular

excited

states

ofdG

dCrepresentsintram

olecular

excitedstates

ofdC

and

ETrepresentselectron

transfer

excitedstates

from

dGto

dCeRef

4

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5553

state because the stacking interaction is weaker in B-DNA thanin Z-DNA (Figure 1) Therefore the first peak of theexperimental CD spectra originates from the ET excitationfrom dG to dC for Z-DNA but corresponds to theintramolecular excitation in dC for B-DNA The second peakwith a negative sign is assigned to the 81A state The 81A statewhich is the πminusπ of dG is the origin of both the main peaks ofthe UV spectrum and the second peak of the CD spectrumThe ET excited states of the tetramer models were calculated

at a lower energy region than those of the dimer models forboth B- and Z-DNA This result may be related to the overlapbetween the nucleic acid bases The overlap exists only in partof the nucleic acid bases in the dimer models but in thetetramer models the overlap covers the hydrogen bonds as wellas the nucleic acid bases The ET excited states are importantfor the CD spectrum of Z-DNA because the lowest excitedstate is the ET type and is negative for rotatory strength but forB-DNA the ET excited states are not important because thesestates are weak in both oscillator and rotatory strengthsThe SAC-CI results of the tetramer models are calculated at

a higher region by about 05 eV because of the small activespace Therefore better SAC-CI calculations are required forcomparison with the experimental UV and CD spectra of DNAHowever these results indicate that the strong stackinginteraction characteristic of the helical structure of Z-DNAchanges the sign of the rotatory strength for the lowest excitedstates of dG and dC and produces the emergence of the ETexcited states

4 CONCLUSIONSDNA chemistry is affected by three important factors (1) theconformation of the monomer (dG) which is related to thedihedral angle between deoxyribose and guanine (2) thehydrogen-bonding interactions and (3) the stacking inter-actions between two nucleic acid bases To verify whether thesethree factors are important for the CD spectra of DNA weexamined four different models using the SAC-CI method(1) dG monomer model We first applied the SAC-CI

method to the dG monomer to examine the effect ofconformation on the CD spectra of dG and from this resultto identify the structure of the dG monomer in solution Ourcalculations showed that the shape and the sign of the CDspectra depend strongly on the dihedral angle between thedeoxyribose and the guanine but that the SAC-CI UV spectraof both anti- and syn-dG are very similar because the UVspectra are not strongly dependent on the conformation Theexperimental CD spectrum agreed with the theoreticalspectrum of dG only for the anti conformation The resultingthermally averaged spectra also support the anti geometry ofdG as a stable conformation in solution However the SAC-CICD spectrum of syn-dG is opposite in sign from that of theexperimental spectrum Thus the SAC-CI CD spectra of dG inboth anti and syn conformations do not explain the negativesign at 295 nm of the CD spectra of the DNA composed of dGand dC(2) Composit model of dG and dC monomers A

comparison of the calculated composite spectra of dG anddC with the experimental spectra of the two-component DNAshowed similarities between B- and Z-DNA because theconformation of dC is always anti in both B- and Z-DNAThe SAC-CI results led us to confirm that that theconformation of dG is not the origin of the negative peak at295 nm in Z-DNA

(3) Dimer model Our examination of the dimer modelpermitted us to investigate the separate importance of thehydrogen-bonding and the stacking interactions in thestructural chemistry of B- and Z-DNA The SAC-CIcalculations showed that the hydrogen-bonding interactionschange the excitation energies and that the stacking interactionsare responsible for the changes in the sign of the CD rotatorystrength The SAC-CI UV spectra of the hydrogen-bondingmodels (B-Hbond and Z-Hbond models) are similar to theexperimental UV spectra in which the main peak as well as theshoulder peak is well reproduced The SAC-CI CD spectra ofthe stacking models (B-Stack and Z-Stack models) have thesame sign as the experimental CD spectra We concluded thatthe stacking interaction is especially important for the CDspectrum of the strongly stacked Z-DNA because the CD signalat 295 nm changed to a negative sign only for the Z-Stackmodel(4) Tetramer model Finally our calculations with the

tetramer model that contain both hydrogen-bonding andstacking interactions revealed that the shape of the SAC-CIUV and CD spectra compared well with the experimental onesThe ET excitations as well as the lowest intramolecularexcitation of dC and dG are the origins of the negative sign ofthe experimental CD spectrum of Z-DNA The ET excited stateis calculated at a lower energy region than the intramolecularexcited states of dG or dC for the strongly stacked Z-DNA butat a higher energy region for the weakly stacked B-DNA Thenegative sign of the first peak of the experimental CD spectra ofZ-DNA is due to the strong stacking effect characteristic of itshelical structureThe dynamic behavior of the dG monomer in solution was

partially considered by taking the Boltzmann averaging of thecalculated spectra over the various conformations Fortunatelythe Boltzmann averaged SAC-CI CD spectrum was in goodagreement with the experimental spectra although the SAC-CICD spectra largely depend on the conformation of dGHowever in this study we used only the static model for dimerand tetramer models The SAC-CI CD results of the dimer andtetramer models explained the difference between theexperimental CD spectra of B- and Z-DNA However sincethe structure of DNA would also be flexible in solution it isnecessary to take into account the dynamic behavior of DNAThis is an important and interesting topic

ASSOCIATED CONTENT

S Supporting InformationSAC-CI results of the hydrogen-bonding and stacking dimermodels of Z- and B-DNA using small active space and withoutthe polarization d-functions This material is available free ofcharge via the Internet at httppubsacsorg

AUTHOR INFORMATION

Corresponding AuthorPhone +81-75-634-3211 Fax +81-75-634-3211 E-mail hnakatsujiqcriorjp

NotesThe authors declare no competing financial interest

ACKNOWLEDGMENTS

The computations were performed at the Research Center forComputational Science Okazaki Japan

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5554

REFERENCES(1) Voet D Voet J G Eds Biochemistry John Wiley amp Sons IncNew York 1995(2) Ghosh A Bansal M Acta Crystallogr 2003 D59 620minus626(3) Rich A Nordheim A Wang A H-J Annu Rev Biochem 198453 791minus846(4) Tran-Dinh S Taboury J Neumann J-M Huynh-Dinh TGenissel B Langlois drsquoEstaintot B Igolen J Biochemistry 1984 231362minus1371(5) Sugiyama H Kawai K Matsunaga A Fujimoto K Saito IRobinson H Wang A H-J Nucleic Acids Res 1996 24 1272minus1278(6) Kawai K Saito I Sugiyama H J Am Chem Soc 1999 1211391minus1392(7) Oyoshi T Kawai K Sugiyama H J Am Chem Soc 2003 1251526minus1531(8) Xu Y Ikeda R Sugiyama H J Am Chem Soc 2003 12513519minus13524(9) Berova N Nakanishi K Woody R W Eds Circular DichroismPrinciples and Applications Wiley-VCH New York 2000(10) Bauernschmitt R Ahlrichs R Chem Phys Lett 1996 256454minus464(11) Casida M E Jamorski C Casida K C Salahub D R JChem Phys 1998 108 4439minus4449(12) Tsolakidis A Kaxiras E J Phys Chem A 2005 109 2373minus2380(13) Varsano D Di Felice R Marques M A L Rubio A J PhysChem B 2006 110 7129minus7138(14) Nielsen L M Holm A I S Varsano D Kadhane UHoffmann S V Di Felice R Rubio A Nielsen S B J Phys ChemB 2009 113 9614minus9619(15) Tonzani S Schatz G C J Am Chem Soc 2008 130 7607minus7612(16) Casida M E Salahub D R J Chem Phys 2000 113 8918minus8935(17) Dreuw A Weisman J L Head-Gordon M J Chem Phys2003 119 2943minus2946(18) Maitra N T Zhang F Cave R J Burke K J Chem Phys2004 120 5932minus5937(19) McCann D M Stephens P J J Org Chem 2006 71 6074minus6098(20) Casida M E J Mol Struct 2009 914 3minus18(21) Niehaus T A J Mol Struct 2009 914 38minus49(22) Tawada Y Tsuneda T Yanagisawa S Yanai T Hirao K JChem Phys 2004 120 8425minus8433(23) Chiba M Tsuneda T Hirao K J Chem Phys 2006 124144106-1-11(24) Nakatsuji H Hirao K J Chem Phys 1978 68 2053minus2065(25) Nakatsuji H Chem Phys Lett 1978 59 362minus364(26) Nakatsuji H Chem Phys Lett 1979 67 329minus333(27) Nakatsuji H Chem Phys Lett 1979 67 334minus342(28) Nakatsuji H Bull Chem Soc Jpn 2005 78 1705minus1724(29) Ehara M Hasegawa J Nakatsuji H SAC-CI Method Appliedto Molecular Spectroscopy In Theory and Applications of Computa-tional Chemistry The First 40 Years A Vol of Technical and HistoricalPerspectives Dykstra C E Frenking G Kim K S Scuseria G EEds Elsevier Oxford UK 2005 pp 1099minus1141(30) Miyahara T Hasegawa J Nakatsuji H Bull Chem Soc Jpn2009 82 1215minus1226(31) Bureekaew S Hasegawa J Nakatsuji H Chem Phys Lett2006 425 367minus371(32) Nakatani N Hasegawa J Nakatsuji H J Am Chem Soc2007 129 8756minus8765(33) Fujimoto K Hasegawa J Nakatsuji H Bull Chem Soc Jpn2009 82 1140minus1148(34) Miyahara T Nakatsuji H Collect Czech Chem Commun2011 76 537minus552(35) Sugiyama H Unpublished(36) Geertsen J Rittby M Bartlett R J Chem Phys Lett 1989164 57minus62

(37) Stanton J F Bartlett R J J Chem Phys 1993 98 7029minus7039(38) Ehara M Ishida M Toyota K Nakatsuji H SAC-CIGeneral-R Method Theory and Applications to the Multi-ElectronProcesses In Reviews in Modern Quantum Chemistry (A Tribute toProfessor Robert G Parr) Sen K D Ed World Scientific Singapore2003 pp 293minus319(39) SAC-CI GUIDE httpwwwqcriorjpsacci(40) Frisch M J Trucks G W Schlegel H B et al Gaussian 03development revision revision E05 Gaussian Inc Wallingford CT2004(41) Hohenberg P Kohn W Phys Rev 1964 136 B864minusB871(42) Kohn W Sham L J Phys Rev 1965 140 A1133minusA1138(43) Salahub D R Zerner M C Eds The Challenge of d and fElectrons The American Chemical Society Washington DC 1989(44) Parr R G Yang W Density-Functional Theory of Atoms andMolecules Oxford University Press Oxford UK 1989(45) Becke A D J Chem Phys 1993 98 5648minus5652(46) Lee C Yang W Parr R G Phys Rev 1988 37 785minus789(47) Hariharan P C Pople J A Theor Chim Acta 1973 28 213minus222(48) Francl M M Pietro W J Hehre W J Binkley J SDeFrees D J Pople J A Gordon M S J Chem Phys 1982 773654minus3665(49) Tomasi J Mennucci B Cammi R Chem Rev 2005 1052999minus3093(50) Fukuda R Ehara M Nakatsuji H Cammi R J Chem Phys2011 134 104109minus1minus11(51) Dunning T H Jr Hay P J In Modern Theoretical ChemistrySchaefer H F III Ed Plenum New York 1976 Vol 3 pp 1minus28(52) Nakatsuji H Chem Phys 1983 75 425minus441(53) Hocquet A Leulliot N Ghomi M J Phys Chem B 2000 1044560minus4568(54) Mishra S K Mishra P C J Comput Chem 2002 23 530minus540(55) Sprecher C A Johnson W C Biopolymers 1977 16 2243minus2264(56) Clark L B Peschel G G Tinoc I Jr J Phys Chem 1965 693615minus3618(57) Voet D Gratzer W B Cox R A Doty P Biopolymers 19631 193minus208(58) Hobza P Mehlhorn A Carsky P Zahradnik R J Mol Struct1986 138 387minus399(59) Hobza P Sponer J Chem Rev 1999 99 3247minus3276

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5555

Page 2: Helical Structure and Circular Dichroism Spectra of DNA: A ... · B-DNA, have right-handed double-helical structure, while Z-DNA is left-handed and appears only when DNA has a special

basis sets are used Actually the SAC-CI method was used asthe reference data for confirming the accuracy of the LC-TDDFT method2223

Actually it is difficult to model DNA in solution becauseDNA is a complex dynamic system However when we use asmall DNA system whose UV and CD spectra are well studiedexperimentally it would be possible to calculate the UV andCD spectra of such DNA in high accuracy by the SAC-CImethod and by comparing with the experimental data wewould be able to elucidate some basic aspects of the complexphenomena In this article we study the UV and CD spectra ofDNA and their relationships to the helical structures using thesmall models taken from B- and Z-DNA composed of the samesequence of only six deoxyguanosine (dG) and deoxycytidine(dC) We want to show that even such a small model can clarifysome aspects of the relationships between the CD spectra andthe helical structure of DNA when we use a highly reliablecomputational method like the SAC-CI methodThe difference between B- and Z-DNA is the helical-

structure as well as the conformation of dG The purine basedG has the anti conformation in B-DNA while it has synconformation in Z-DNA3 The anti- and syn-dG differ in thedihedral angle between the guanine of the purine base anddeoxyribose by approximately 180deg However the pyrimidinebase dC exists as only the anti conformation in both B- and Z-DNA because of the steric hindrance between the pyrimidinebase cytosine and deoxyriboseThere are three main interactions in DNA (1) the

interaction through the bond between the nucleic acid basesand the deoxyribose (2) hydrogen-bonding interactionsbetween two nucleic acid bases and (3) stacking interactionsbetween two nucleic acid bases Our goal is to understand theroles of these interactions in forming the helical structure andhow these interactions affect the CD spectra of B- and Z-DNAWe have applied the symmetry-adapted cluster (SAC)-

configuration interaction (CI) theory24minus27 to elucidate theunderlying relationships between the helical structure of DNAand the sign of their CD spectra The SAC-CI theory is areliable method for studying the UV and CD spectra in relationto the electronic and geometrical structures of the biologicalsystem and chiral compounds30minus34 We calculated the UV andCD spectra of the monomer of dG in various conformationsincluding both anti and syn conformations For the dCmonomer however we performed calculations only for theanti geometries observed in B- and Z-DNA because theconformation of dC is rigid in DNA For purposes ofcomparison we used an experimentally acquired spectrum ofdG obtained in aqueous solution35 We also studied the UV andCD spectra of DNA First we approximated the spectra as asimple sum of the spectra of dG and dC that have the structurein B- and Z-DNA Next we calculated the UV and CD spectrausing dimer models of B- and Z-DNA that represented eitherthe hydrogen-bonding interaction or the stacking interactionsFinally we calculated the UV and CD spectra using tetramermodels of B- and Z-DNA that included both of the hydrogen-bonding and stacking interactions within the same model All ofthe models include the through-bond interactions between thenucleic acid base and the deoxyribose

2 MODELING AND COMPUTATIONAL DETAILSThe SAC-CI theory is a useful coupled-cluster type electroncorrelation theory for studying ground excited ionized andelectron-attached states of molecules published in 197824minus27

The so-called EOM-CC method published about a decadelater3637 is very similar to the SAC-CI method although calleda different name This equivalence was shown even numericallyin the previous publications3839 For the calculations of the CDspectra of large molecules the gauge-invariant velocity formmust be used since the position of the gauge origin is notknown The independence of the SAC-CI method from thegauge origin has already been determined for the three-membered ring compounds30

First we studied the dependence of the CD spectrum on theconformation of the dG monomer The ground stategeometries with both anti and syn conformations of dG wereoptimized with Gaussian0340 using the density functionaltheory (DFT)41minus44 with the B3LYP functional4546 for the 6-31G(dp) basis4748 sets All of the geometrical parameters ofdG except for the dihedral angle φ were optimized at every 20degin the potential energy for the ground state of dG Fulloptimization was done for anti and syn conformations Thesolvent effects of water were considered using the polarizablecontinuum model (PCM)49 that was connected recently withthe SAC-CI method in the Gaussian suite of programs50

We then analyzed the UV and CD spectra of B- and Z-DNAobtained using the dimer model which contains either thehydrogen-bonding interaction or the stacking interaction andnext using the tetramer model which contains both thehydrogen-bonding and stacking interactions within the samemodel (Figure 1) The geometries were taken from the X-raycrystallography structure 9BNA and 1DCG were used for B-and Z-DNA respectively In B-DNA the GC pair weakly stackswith both of the lower and upper GC pairs (diamond-shaped

Figure 1 (a) Hydrogen-bonding (Z-Hbond) and stacking (Z-Stack)models taken from the X-ray crystallography structure (1DCG) of Z-DNA (b) Hydrogen-bonding (B-Hbond) and stacking (B-Stack)models taken from the X-ray crystallography structure (9BNA) of B-DNA (c) Tetramer model (Z-Tetra) taken from the X-raycrystallography structure (1DCG) of Z-DNA (d) Tetramer model(B-Tetra) taken from the X-ray crystallography structure (9BNA) ofB-DNA

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5543

box in Figure 1bd) but in Z-DNA the guanine-cytosine (GC)pair strongly stacks with only one GC pair (denoted by thesquare box Figure 1ac) The stacking model of Z-DNA (Z-Stack model) has strong stacking interactions As in thegeometry optimization for dimer and tetramer models only thehydrogen atom positions in these models were optimized usingthe B3LYP6-31G(dp) basis with the PCM of waterFor the SACSAC-CI calculations the basis functions

employed were D9551 sets for deoxyribose and D95(d)51 setsfor the nucleic acid bases The core orbitals were treated asfrozen orbitals and all singles and selected doubles wereincluded Perturbation selection52 was carried out for thedouble excitation operators using the threshold sets of 5 times 10minus7

and 1 times 10minus7 hartree for SAC and SAC-CI calculationsrespectively In the calculations for the tetramer model thebasis functions employed were D95 sets (without polarizationfunctions) because the calculations were heavier than thecalculations for the dimer model and further the orbitalswhose energies were within minus11 to +11 au were chosen as theactive orbitals The SAC-CI UV and CD spectra wereconvoluted using Gaussian envelopes to describe the FranckminusCondon widths and the resolution of the spectrometer The fullwidth at half-maximum (fwhm) of the Gaussian was 08 eVwhich was estimated from the criterion that the two bands (253and 278 nm) looks like a single band in the experimental UVspectrum of dG the lowest two bands were not divided clearlyThis value (08 eV) was also used for the convolution of thetheoretical CD spectra Since DNA is a polymer composed ofmonomers we used the same value in dimer and tetramercalculations The UV and CD spectra of (1) dG with severalconformations (2) only dG and only dC in the X-raycrystallography structures of B- and Z-DNA (3) the hydro-gen-bonding and stacking models of B- and Z-DNA and (4)the tetramer models of B- and Z-DNA were calculated usingthe SAC-CI method

3 RESULTS AND DISCUSSION

31 Structure of the Ground State dG Monomer inWater We calculated the changes in the potential energy forthe ground state of the dG monomer in water as a function ofchanges in the dihedral angle (φ) between the deoxyribose andthe guanine (Figure 2) All of the geometrical parameters of dGwere optimized and φ was varied by incremental increases of20deg The optimized dihedral angle was calculated to be 3103degin the anti region and at 1295deg for the syn region The

deoxyribose has C3prime-endo conformation for both anti- and syn-dG with the present calculations although for anti-dG C2prime-endo was calculated to be more stable than C3prime-endo5354Thus in dG the conformation of the deoxyribose seems to bevery flexible and its energy difference very small However theconformation of deoxyribose is rigid in DNA the deoxyribosein dG is C2prime-endo for B-DNA and C3prime-endo for Z-DNA basedon the X-ray crystallography structures (9BNA and 1DCG)The anti-dG was calculated to be more stable than the syn-dGby 047 kcalmol The conformation of dG converts easily fromanti to syn (or from syn to anti) because the energy barrierbetween the anti and syn forms is only sim5 kcalmol The anti-dG rotates more easily in the clockwise direction (the directionof the arrow in Figure 2) than in the counterclockwise directionbecause the energy barrier is less than 4 kcalmol in theclockwise direction but is more than 5 kcalmol in thecounterclockwise direction The potential energy curve is flat inthe anti region while it is sharper in the syn region becausethere is a hydrogen bond interaction between the hydroxylgroup of the deoxyribose and the nitrogen atom of the guaninenear the syn conformation However the anti-dG can rotatemore freely between φ = 270deg and 330deg because of the lack ofhydrogen-bonding interaction

32 UV and CD Spectra of dG Monomer in WaterUsing the SAC-CI method we calculated the excitationenergies oscillator strengths rotatory strengths and the naturesof the excited states for both of the anti- and syn-dG (Table 1)The difference in the UV excitation spectra between anti- andsyn-dG is very small within 007 eV in energy and within 008au in oscillator strength for all the calculated states Howeverfor the CD spectra whose shapes are described by the rotatorystrengths the spectra of anti-dG differ markedly from thespectra of syn-dG not only in the absolute value but also in thesign of the rotatory strength Below we show these theoreticalresults in the spectral forms to facilitate the comparisons withthe experimentally obtained data

321 UV Spectra of dG Monomer The SAC-CI UV spectra(red lines) of anti- and syn-dG were compared with theexperimental spectrum (black lines) of dG above 200 nm(Figure 3) There are two main bands (276 and 250 nm) above200 nm and one band below 200 nm The SAC-CI UV spectraof both anti- and syn-dG are similar to the experimental UVspectrum and therefore it is unclear whether the experimentalUV spectrum of dG is due to the anti or the syn conformationThe first band at about 276 nm was assigned to the π rarr πhighest occupied molecular orbital (HOMO)minuslowest unoccu-pied molecular orbital (LUMO) transition (11A state at 446 eV(278 nm) for anti-dG and 440 eV (282 nm) for syn-dG) Thesecond band at approximately 250 nm was assigned to the 31Astate (at 510 eV (243 nm) for anti-dG and 506 eV (245 nm)for syn-dG) of π(HOMO) rarr π(next-LUMO) and 21A state(at 508 eV (244 nm) for anti-dG and 503 eV (247 nm) forsyn-dG) of the n rarr π nature As expected from its nature theoscillator strength of 21A is weak (005 au) for anti-dG and veryweak (001 au) for syn-dG The states 41A 51A and 61A (at591 597 and 608 eV (210 208 and 204 nm) for anti-dG and590 598 and 608 eV (210 207 and 204 nm) for syn-dG)were calculated to be in the region of the third band atapproximately 214 nm The oscillator strength is almost zerofor the n rarr π 41A and 61A states but is not zero for the π rarrπ 51A state However these states were not observed in theexperimental UV spectra because these three states are in themedian of the two strong peaks (31A and 71A states) and their

Figure 2 Potential energy curve of the ground state of dG determinedas dihedral angle φ is varied in increments of 20deg Other geometricalparameters were optimized at each φ

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5544

oscillator strengths are comparatively small The 71A state (at662 eV (187 nm) for anti-dG and 654 eV (189 nm) for syn-dG) was π rarr π (next-HOMO to LUMO and next-LUMO)excitation which was assigned to the fourth band comparedwith the experimental UV spectrum of deoxyguanosinemonophosphate (dGMP)55 Thus the SAC-CI UV spectra ofboth anti- and syn-dG are similar to the experimental UVspectrum The differences between the anti- and syn-dG werevery small because the HOMO next-HOMO LUMO andnext-LUMO of the main configuration for 11A 31A and 71Astates with large oscillator strengths are localized on theguanine site Hence we cannot determine if the experimentalUV spectrum reflects that of either anti- or syn-dG322 CD Spectra of dG Monomer in Water The SAC-CI

CD spectra (red lines) of the dG monomer at several dihedralangles φ were compared with the experimental CD spectrum(black line) above 200 nm (Figure 4) In the experimentallyacquired spectra the first band is at 276 nm which is strong inthe UV spectrum but very weak in the CD spectrum Thesecond band is at 250 nm which is strong in both UV and CDspectra but has a negative sign in the CD spectrum The thirdband at 214 nm is weak in the UV spectrum but stronglypositive in the CD spectrumThe SAC-CI CD spectra of dG with φ = 10minus150 and 350deg

(Figure 4aminushr) are quite different from the experimental CDspectrum The SAC-CI CD spectrum of syn-dG (1295deg) isopposite in sign for the second band as compared with theexperimental CD spectrum (Figure 4g) For dG with φ = 170210minus290deg (Figure 4ikminuso) the second band of the SAC-CI CDspectra agrees with the experimental CD spectrum but thethird band of the SAC-CI CD spectra has a weak intensitycompared with the experimental CD spectrum The SAC-CICD spectrum of the anti-dG and dG with φ = 330deg (Figure4pq) is in good overall agreement with the experimental CD

spectrum with the second band being strongly negative and thethird band being strongly positiveAs shown in Figure 2 the potential energy curve is flat

around the anti conformer and the energy barrier is not largeSo we calculated the ratio of the existence of each conformerassuming the Boltzmann distribution at 275 K a normalexperimental condition and using the ground state B3LYP6-31G(dp) energies In Table 2 the column labeled ldquoallrdquo givesthe ratio of each conformer when averaging included all 18conformers ldquoantirdquo gives the ratio of each conformer when onlyseven conformations around anti are used and ldquosynrdquo gives theratio of each conformer when only five conformations aroundsyn are used More than 70 of dG exists near the anticonformation between 270deg and 330deg and about 11 existsnear the syn-dG at 1295deg Within the syn region (90minus170deg)more than 50 dG exist as the syn-dG (1295deg) However inthe anti region (250minus10deg) dG exists mostly between 270deg and330deg as anti-dG The gradient of the potential energy is sharparound the syn conformer but flat around the anti conformerbecause dG forms a hydrogen bond with the deoxyribose onlyin the syn conformation Therefore for the CD spectra we mayconsider only the conformer at 1295deg for the syn conformerbut we have to consider several conformers around 3103deg forthe anti conformerWe also carried out the Boltzmann averaging for the spectra

of the anti (Figure 4p) and syn (Figure 4g) conformations TheSAC-CI CD spectra of ldquoallrdquo (green) ldquoantirdquo (blue) and singleanti-dG (red) are similar to the experimental CD spectrum asshown in Figure 4p while the SAC-CI CD spectra of ldquosynrdquo(blue) and single syn-dG (red) are opposite in sign to theexperimental CD spectrum (Figure 4g) These results confirmthat anti-dG is the main conformation under the experimentalcondition We note that the SAC-CI CD spectrum of dG at thesingle geometry of φ = 330deg is more similar to the experimentalCD spectrum of that of anti-dG (3103deg) This may indicatethat dG with φ = 330deg is more stable than the anti-dG insolutionThe dependence on the dihedral angle φ is very small in the

UV spectrum of Figure 3 because only π rarr π has strongintensity However the CD spectra of Figure 4 strongly dependon the dihedral angle φ because both π rarr π and n rarr π havestrong intensities and their MOs mainly localized at theguanine site are somewhat extended to the deoxyribose The11A state assigned to the first band at 276 nm was calculated tohave a small rotatory strength for both anti- and syn-dG butthis state was not observed in the experimental CD spectrumIn the second band region (250 nm) the rotatory strengths of

Table 1 SAC-CI UV and CD Spectra of anti- and syn-dG

SAC-CI

anti-dG syn-dG

EEa oscb rotc EEa oscb rotc

state (eV) (nm) (au) (10minus40 cgs) (eV) (nm) (au) (10minus40 cgs) nature exptld (eVnm)

11A 446 278 016 453 440 282 016 244 π rarr π 44927621A 508 244 005 minus7631 503 247 001 3812 n rarr π 49625031A 510 243 020 6820 506 245 022 minus3461 π rarr π 49625041A 591 210 000 1731 590 210 000 minus292 n rarr π 57921451A 597 208 002 minus225 598 207 005 830 π rarr π 57921461A 608 204 000 minus215 608 204 000 minus270 n rarr π 57921471A 662 187 044 minus782 654 189 038 minus078 π rarr π

aExcitation energy bOscillator strength cRotatory strength dRef 35

Figure 3 SAC-CI UV spectra of (a) anti- and (b) syn-dG (red)compared with the experimentally determined UV spectrum35 of dG(black)

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5545

both the 21A and 31A states are large and opposite in sign(Table 1) The 21A state is calculated to lie at 508 eV (244nm) for anti-dG and at 503 eV (247 nm) for syn-dG The 31Astate is calculated to lie at 510 eV (243 nm) for anti-dG and at506 eV (245 nm) for syn-dG The difference between 21A and31A in the excitation energy is very small so that their rotatorystrengths cancel each other giving relatively small rotatorystrengths From Table 1 we see that the total rotatory strengthof the second band is negative for anti-dG and positive for syn-dG To reproduce the experimental CD spectra the theoreticalrotatory strength must be reliable since otherwise reliablepositive or negative CD peaks cannot be obtained The SAC-CItheory seems to give reliable rotatory intensity since thetheoretical CD spectra agree well with the experimental one Inthe third band region (214 nm) the sum of the rotatorystrengths is positive for both anti- and syn-dG although thestrongest rotatory strength comes from the 41A (n rarr π) statefor anti-dG but from the 51A (π rarr π) state for syn-dG As inthe experimental CD spectrum the third band has largerintensity than the second band in the SAC-CI CD spectrum ofanti-dG However for syn-dG the third band in the SAC-CI

CD spectrum is smaller than the second band The 71A stateassigned to the fourth band with a strong negative intensity at190 nm in the experimental CD spectrum of dGMP55 isnegative for anti-dG and almost zero for syn-dG

33 Composite CD Spectra of dG and dC Comparedwith the CD Spectra of DNA We then extended our studiesto the spectra of B- and Z-DNA and we compared the SAC-CICD spectra of anti- and syn-dG (Figure 4) with theexperimental CD spectra of B- and Z-DNA (Figure 5)4 Thesecond band at 250 nm in the SAC-CI CD spectrum of anti-dGis similar to the experimental CD spectrum of B-DNA whilethe SAC-CI CD spectrum of syn-dG is similar to theexperimental CD spectrum of Z-DNA This result may indicatethat in DNA the second band at 250 nm reflects the dihedralangle φ of dG between the deoxyribose and the guanineTherefore we calculated the composite SAC-CI CD spectra ofthe noninteracting dG and dC monomer from the available X-ray crystallographic structures in B- and Z-DNA to clarify if thedifferent conformations of dG andor dC in different types ofDNA explain the differences in the CD spectra

Figure 4 CD spectra of dG at several conformation angles φ The experimental CD spectrum35 (black line) of dG is compared with the SAC-CI CDspectra (red lines) of (a) φ = 10deg (b) φ = 30deg (c) φ = 50deg (d) φ = 70deg (e) φ = 90deg (f) φ = 110deg (g) φ = 1295deg (syn) (h) φ = 150deg (i) φ = 170deg(j) φ = 190deg (k) φ = 210deg (l) φ = 230deg (m) φ = 250deg (n) φ = 270deg (o) φ = 290deg (p) φ = 3103deg (anti) (q) φ = 330deg and (r) φ = 350deg TheSAC-CI CD Boltzmann averaged spectra of (g) syn (blue line) (p) anti (blue line) and (p) all (green line) are also shown

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5546

The SAC-CI CD spectra of dG and dC alone (Figure5abde) having the structures of those in B- and Z-DNA andthe composite CD spectra (Figure 5cf) obtained as (a + b andd + e) were compared to the experimental CD spectra of B-and Z-DNA The SAC-CI excitation energies oscillatorstrengths rotatory strengths and natures of the excited statesfor the monomer model (dG and dC) were also determined(Table 3)The conformations of dG and dC are the same as those in

the ideal B-DNA or Z-DNA in which dG and dC are arrangedalternately (Figure 1) However the X-ray structure wasobtained for the DNA composed of only 6 sequences in thecase of Z-DNA Since the geometry is affected by the effect ofthe edge of DNA the conformation is different for eachmonomer Since the conformation of dG is in particular moreflexible than that of dC in DNA we calculated two kinds of dG(model I and II) in B- and Z-DNA respectively The models Iand II correspond to dG of the hydrogen-bonding models (B-Hbond and Z-Hbond models) and the stacking models (B-Stack and Z-Stack models) of the dimer models respectivelyThe dihedral angle (φ) between the deoxyribose and theguanine is 2763deg and 3007deg respectively for models I and IIof B-DNA and 1113deg and 1175deg respectively for models Iand II of Z-DNA The dihedral angle between the deoxyriboseand cytosine is 3232deg for B-DNA and 3329deg for Z-DNA Therotatory strength of dG is very different between models I andII for both B- and Z-DNA as shown in Table 3 Therefore fordG we prepared an averaged CD spectra for models I and II(Figure 5ad) For Z-DNA the SAC-CI CD spectrum of dG(Figure 5a) is negative for the first band and positive for thesecond band as in the experimental spectrum of Z-DNAalthough the excitation energies are calculated at a higherregion However the SAC-CI result is opposite in sign to theexperimental CD spectrum for the first band of B-DNA (Figure5d) These results show that in the CD spectra of DNA the

conformation of dG (monomer) is related to the sign of thesecond band but has no effect on the sign of the first bandThe CD spectra of dC have four peaks (Figure 5be) The

first peak represents the πminusπ of HOMO to LUMO Thesecond peak is the excitation from the nonbonding orbital tothe LUMO The third and fourth peaks are composed of boththe πminusπ and nminusπ from next-HOMO and nonbondingorbitals to LUMO The third and fourth peaks are mainly πminusπand nminusπ respectively The SAC-CI CD spectra of dC aresimilar in B- and Z-DNA with two positive and two negativepeaks The rotatory strength is positive for the first and thirdpeaks and negative for the second and fourth peaks Thereforethe CD spectra of dC are always positive at around 295 nmcorresponding to the conformation of dC which is always antiin DNAAs shown in Figure 5cf the composite SAC-CI CD spectra

obtained from dG and dC alone are similar between B- and Z-DNA The sign at 295 nm is positive for both B- and Z-DNAbecause the rotatory strength of dC which is the lowest peak ismuch stronger than that of dG Therefore the composite CD

Table 2 Boltzmann Distribution of dG along theConformational Angle φ

angle (φ) all ()a anti ()b syn ()c

10 086 1130 01450 00470 01090 170 91110 279 1431295 (syn) 1131 517150 454 224170 041 25190 008210 000230 010250 325 41270 1534 195290 1700 2163103 (anti) 2669 339330 1307 166350 258 33total 1000 1000 1000

aldquoallrdquo gives the ratio of each conformer when averaging included all 18conformers bldquoantirdquo gives the ratio of each conformer when only sevenconformations around anti are used cldquosynrdquo gives the ratio of eachconformer when only five conformations around syn are used

Figure 5 SAC-CI CD spectra of monomer ((a) dG in Z-DNA (b) dCin Z-DNA (d) dG in B-DNA and (e) dC in B-DNA) and thecomposite CD spectra of (c) Z-DNA and (f) B-DNA obtained fromthe SAC-CI CD spectra of dG and dC compared with theexperimental CD spectra4 (black lines) of Z-DNA (aminusc) and B-DNA (dminusf) The red and blue lines represent the excited state of dGand dC respectively The magenta line in panels c and f represents thecomposite spectra

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5547

spectrum is similar to that of dC for both B- and Z-DNA As aresult in the composite CD spectra of B- and Z-DNA the signat around 295 nm is independent of the geometry of themonomer (dG or dC) even though the CD spectrum of dGdepends on the dihedral angle between the guanine and thedeoxyribose because the lowest peak originates from dCwhose geometry is always anti in both B- and Z-DNA Thisindicates that the noninteracting dG and dC cannot explain theexperimental spectra of DNA and that the hydrogen-bondingor stacking interactions with the neighboring nucleic acid basesmust be important for the CD spectra of DNA34 UV and CD Spectra of Dimer Models In the studies

of the structure spectroscopy and functions of DNA it isgenerally recognized that the hydrogen-bonding interaction andthe stacking interaction are the two most important interactionsthat characterize the properties of DNA So we expect thatthese interactions might also be the important factors that affectthe features of the UV and CD spectra of DNA To investigatethe possible origin we considered the dimer model composedof dG and dC These models contained either the hydrogen-bonding interaction or the stacking interaction (Figure 1)However we note that the SAC-CI results using these modelsneed not be in agreement with the experimental spectrabecause these models lack either the stacking interaction or thehydrogen-bonding interaction We must use the tetramer

model at least to compare with the experimental spectra asnoted below

341 UV Spectra of Dimer Models Figure 6 shows theSAC-CI UV spectra of both hydrogen-bonding and stackingmodels of both B- and Z-DNA compared with the experimentalUV spectrum The experimental UV spectra of Z-DNA displaya main peak at 256 nm and a shoulder peak at around 290 nmThe experimental UV spectrum of B-DNA is very similar butthe shoulder peak at around 290 nm is weaker Table 4 showsthe excitation energies oscillator strengths rotatory strengthsnatures and the types of the excited states of the SAC-CIresults for both hydrogen-bonding and stacking models of thedimer When the excitation in the dimer model is not withindG or dC but corresponds to the electron transfer (ET)between dG and dC we assigned it as the ET typeThe SAC-CI UV spectrum of the hydrogen-bonding model

of Z-DNA (Z-Hbond model) is in good agreement with theexperimental UV spectrum of Z-DNA (Figure 6a) Threestrong peaks of the πminusπ nature are calculated The loweststrong peak corresponds to the shoulder peak at 428 eV (290nm) of the experimental UV spectrum of Z-DNA Theremaining two strong peaks correspond to the main peak at484 eV (256 nm) of the experimental UV spectrum of Z-DNARed blue and green colors represent the excited states of dGdC and ET respectively

Table 3 SAC-CI UV and CD Spectra of dG and dC in Z- and B-DNA

Z-DNA B-DNA

dG (model I φ = 1113deg) dG (model I φ = 2763deg)

EEa oscb rotc EEa oscb rotc

state (eV) (nm) (au) (10minus40 cgs) nature (eV) (nm) (au) (10minus40 cgs) nature

11A 441 281 017 125 π rarr π 443 280 016 minus1035 π rarr π21A 474 262 000 minus1099 n rarr π 488 254 000 1099 n rarr π31A 493 251 023 2066 π rarr π 513 242 025 minus3355 π rarr π41A 578 214 000 minus864 n rarr π 595 208 002 minus2619 π rarr π51A 591 210 004 minus656 π rarr σ 599 207 001 3571 n rarr π61A 611 203 000 minus438 n rarr π 618 201 000 983 n rarr π71A 628 197 000 359 π rarr σ 675 184 008 1235 π rarr σ81A 659 188 030 634 π rarr π 678 183 028 minus2402 π rarr π

dG (model II φ = 1175deg) dG (model II φ = 3007deg)

EEa oscb rotc EEa oscb rotc

state (eV) (nm) (au) (10minus40 cgs) nature (eV) (nm) (au) (10minus40 cgs) nature

11A 437 284 016 minus451 π rarr π 434 286 016 minus247 π rarr π21A 459 270 000 minus322 n rarr π 493 251 000 minus1467 n rarr π31A 491 252 022 minus319 π rarr π 518 239 021 minus200 π rarr π41A 581 213 003 323 π rarr π 595 209 001 2884 n rarr π51A 598 207 000 minus738 π rarr π 599 207 003 minus647 π rarr π61A 621 200 000 796 n rarr π 617 201 000 minus3033 n rarr π71A 641 193 000 353 π rarr σ 660 188 048 978 π rarr π81A 647 191 031 minus077 π rarr π 666 186 001 326 π rarr σ

dC (φ = 3329deg) dC (φ = 3232deg)

EEa oscb rotc EEa oscb rotc

state (eV) (nm) (au) (10minus40 cgs) nature (eV) (nm) (au) (10minus40 cgs) nature

11A 413 300 009 2385 π rarr π 409 303 009 1403 π rarr π21A 442 280 000 minus1548 n rarr π 452 274 000 minus915 n rarr π31A 507 245 011 8392 π + n rarr π 499 249 007 5975 π + n rarr π41A 527 235 011 minus8639 n + π rarr π 527 235 015 minus8017 π + n rarr π51A 573 216 000 minus1385 n rarr π 559 222 001 393 n rarr π61A 632 196 039 2349 π rarr π 625 198 032 1098 π rarr π

aExcitation energy bOscillator strength cRotatory strength

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5548

The shoulder peak in the experimental UV spectrum of Z-DNA is composed of 11A (439 eV) and 21A (453 eV) statesThe 11A state (in blue color) of the Z-Hbond modelcorresponds to the 11A state of dC at 413 eV which is theexcitation from the HOMO to the LUMO of dC This state haslarge oscillator strength of the πminusπ nature In the Z-Hbondmodel this state is blue-shifted by 026 eV because theHOMOminusLUMO gap (0449 au) of the Z-Hbond model islarger than that (0438 au) of dC This state is in goodagreement with the first (shoulder) peak of the experimentalUV spectrum of Z-DNA However the 21A state (453 eV redcolor) of the Z-Hbond model is assigned to the 31A state (493eV) of dG which is the excitation from the HOMO to the next-LUMO of dG This large shift of the excitation energy is causedby the MO-order change The MO of the Z-Hbond modelcorresponding to the next-LUMO of dG becomes lower thanthe MO of the Z-Hbond model corresponding to the LUMOof dG by the formation of hydrogen bonds with cytosine in theZ-Hbond model Therefore this state is red-shifted by 04 eVHowever the oscillator strength of this state is small despite itsπminusπ nature The transition moment is canceled because themain configuration includes not only the excitation from theHOMO to the next-LUMO in dG but also the excitation fromthe next-HOMO to the LUMO in dC and the electron transfertype excitation from the HOMO of dG to the LUMO and next-LUMO of dCThe main peak of the experimental UV spectrum of Z-DNA

is composed of 31A (484 eV) and 51A (493 eV) states havinglarge oscillator strengths (Figure 6a) The 31A state (so blue

color) of the Z-Hbond model is assigned to the 31A state (507eV) of dC which reflects the strong excitation from the next-HOMO to the LUMO of dC The 51A state (red color) of theZ-Hbond model is assigned to the 11A state (441 eV) of dGwhich is the excitation from the HOMO to the LUMO in dGThe excitation energy increases by 052 eV because theHOMOminusLUMO gap becomes large which is the opposite tothat of the 21A state but again due to the hydrogen-bondingeffectFor B-DNA however the SAC-CI UV spectrum of the

hydrogen-bonding model (B-Hbond model Figure 6c) also hasboth main and shoulder peaks in which the main peak is higherand the shoulder peak is lower as compared with theexperimental UV spectrum of B-DNA The main peak (41Aand 51A states) of the SAC-CI spectra of the B-Hbond model ishigher than that of the experimental spectra similarly to theSAC-CI spectrum of the Z-Hbond model The 31A state of dGin B-DNA is also higher than that in Z-DNA (Table 3) Theseresults may be related to the fact that dG is more flexible thandC because the conformation of dG is anti for B-DNA but synfor Z-DNA Since the stacking interaction of B-DNA is weakerthan that of Z-DNA the X-ray structure of B-DNA may bemore flexible than that of Z-DNA However the result that theoscillator strength of the lowest excited state of the B-Hbondmodel is weaker than that of the Z-Hbond model is reflected bythe fact that the shoulder peak of the experimental UV spectrais stronger in Z-DNA but weaker in B-DNAFor the SAC-CI UV spectra of the stacking models of both

B- (B-Stack model) and Z-DNA (Z-Stack model) the firstexcited states originated from dC were calculated at the lower-energy region (higher wavelength region) but the excitationenergy of the main peak is almost the same between the B-Hbond and B-Stack models or between the Z-Hbond and Z-Stack models If there is no hydrogen-bonding interaction inDNA the lowest excited state would be observed in the lower-energy region of the experimental spectra separately from themain peak The excitation energy of the lowest excited state ofthe stacking model is lower than that of the hydrogen-bondingmodel by 044 eV for Z-DNA and 033 eV for B-DNArespectively This result may correspond to the fact that thelowest excitation energy of cytosine differs between the gasphase (428 eV)56 and in solution (464 eV)57

For both B- and Z-DNA the SAC-CI results of thehydrogen-bonding models (B-Hbond and Z-Hbond) are inbetter agreement with the experimental spectra than those ofthe stacking models (B-Stack and Z-Stack) Thus thehydrogen-bonding interaction is reflected in the UV spectraof DNA On the other hand the shoulder peak observed for Z-DNA may reflect the stronger shoulder nature of the Z-Stackmodel in Z-DNA However the UV spectra of DNA providelittle information about stacking interactions between nucleicacid bases

342 CD Spectra of Dimer Models Figure 7 shows theSAC-CI CD spectra of both hydrogen-bonding and stackingmodels of B- and Z-DNA compared with the experimental CDspectra In the experimental CD spectra the first peak isnegative at 296 nm and the second peak is positive at 275 nmfor Z-DNA However B-DNA displays peaks that are oppositein sign the first peak is positive at 280 nm and the second peakis negative at 255 nmIt is worth nothing that the SAC-CI spectrum has a negative

peak at 295 nm only for the stacking model of Z-DNA (Z-Stack) but for the other models the SAC-CI spectrum is

Figure 6 SAC-CI UV spectra for the dimer models The SAC-CIspectra (magenta line) of (a) Z-Hbond model (b) Z-Stack model (c)B-Hbond model and (d) B-Stack model are compared with theexperimental UV spectra4 (black line) of Z-DNA (ab) and B-DNA(cd) The red and blue lines represent the intramolecular excitedstates of dG and dC respectively The green lines represent theelectron transfer (ET) excited state from dG to dC

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5549

Table

4SA

C-CIUVandCD

Spectraof

theHydrogen-Bon

ding

andStacking

Dim

erMod

elsof

Z-andB-D

NA

Z-DNA

B-DNA

SAC-CIof

Z-H

bond

exptle

SAC-CIof

B-H

bond

exptle

EEa

oscb

rotc

UV

EEa

oscb

rotc

UV

state

(eV)

(nm)

(au)

(10minus

40cgs)

nature

typed

(eVnm)

(eV)

(nm)

(au)

(10minus

40cgs)

nature

Typed

(eVnm)

11A

439

282

020

3513

πrarr

πdC

(11 A)

428290

437

284

017

1143

πrarr

πdC

(11 A)

21A

453

274

004

884

πrarr

πdG

(31 A)

456

272

003

minus1481

πrarr

πdG

(31 A)

31A

484

256

018

912

πrarr

πdC

(31 A)

493

251

006

583

πrarr

πET

41A

485

255

005

minus1145

πrarr

πET

498

249

021

1767

πrarr

πdC

(3+41A)

51A

493

252

042

1182

πrarr

πdG

(11 A)

484256

518

240

043

minus1390

πrarr

πdG

(11 A)

484256

61A

512

242

000

057

nrarr

πdC

(21 A)

525

236

000

minus1197

nrarr

πdC

(21 A)

71A

523

237

000

minus1384

nrarr

πdG

(21 A)

529

234

000

minus579

nrarr

πdG

(2+51A)

81A

623

199

000

minus034

nrarr

πdG

(2+51A)

SACminusCIof

Z-Stack

exptle

SACminusCIof

B-Stack

exptle

EEa

oscb

rotc

CD

EEa

oscb

rotc

CD

state

(eV)

(nm)

(au)

(10minus

40cgs)

nature

typed

(eVnm)

(eV)

(nm)

(au)

(10minus

40cgs)

nature

typed

(eVnm)

11A

395

314

009

minus4511

πrarr

πdC

(11 A)

419296

404

307

007

4358

πrarr

πdC

(11 A)

443280

21A

421

294

000

minus3349

πrarr

πdG

(11 A)

419296

440

282

006

minus1032

πrarr

πdG

(11 A)

31A

445

279

001

2967

nrarr

πdC

(21 A)

461

269

000

minus503

nrarr

πdC

(21 A)

41A

472

263

000

minus1335

nrarr

πdG

(21 A)

493

251

013

9407

πrarr

πdC

(3+41A)

51A

485

256

019

minus1796

πrarr

πdG

(31 A)

451275

507

245

005

minus15115

nrarr

πdG

(21 A)

486255

61A

499

248

006

4861

πrarr

πET

451275

512

242

014

minus11056

πrarr

πdG

(31 A)

486255

71A

526

236

007

2721

πrarr

πET

+dC

(31 A)

451275

521

238

014

19475

πrarr

πET+dG

(31 A)

81A

565

219

000

minus231

nrarr

πdC

(41 A)

619

200

000

minus805

nrarr

πdG

(41 A)

aExcitatio

nenergybOscillator

strengthcRotatorystrengthddG

representsintram

olecular

excitedstates

ofdG

dCrepresentsintram

olecular

excitedstates

ofdC

and

ETrepresentselectron

transfer

excitedstates

from

dGto

dCeRef

4

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5550

positive at 295 nm Therefore the strong negative sign of theCD spectra at 295 nm indicates that a strong stackinginteraction exists in Z-DNA The SAC-CI CD spectra of bothstacking models (B-Stack and Z-Stack models) more closelyresemble the experimental CD spectra of B- and Z-DNA thanthose of the hydrogen-bonding models (B-Hbond and Z-Hbond models) For the SAC-CI CD spectrum of the Z-Stackmodel the first and second peaks are negative and positiverespectively but they are the opposite of those obtained fromthe Z-Hbond model For the B-Stack and B-Hbond modelshowever the first peak is positive in the SAC-CI CD spectrabut the second and third peaks are negative and positiverespectively for the B-Stack model in opposition to the B-Hbond model Since the stacking models do not have thehydrogen-bonding interactions the excitation energy of the firstpeak is lower than that of the experimental spectrum as notedaboveIn the Z-Stack model the first peak with a negative sign is

assigned to the 11A (395 eV) and 21A (421 eV) states (Table4) The 11A state of the Z-Stack model corresponds to the 11Astate (413 eV) of dC which is the excitation of the πminusπ fromthe HOMO to the LUMO in dC A large oscillator strength isthe origin of the first (shoulder) peak but the excitation energyis lower than the experimental value because there is nohydrogen bond in the Z-Stack model The rotatory strength ofthe 11A state of dC changes the sign from positive to negativeby the stacking interaction with dG Therefore the stackinginteraction between dG and dC accounts for the strongnegative sign of the CD spectra in Z-DNA

The excitation energy of the 21A state (421 eV) is also closeto the first peak of the experimental CD spectrum This state isassigned to the 11A state (437 eV) of dG which is the HOMOto LUMO πminusπ excitation in dG but also includes thecontribution of the excitation from the next-HOMO to LUMOin dC The oscillator strength in the UV spectrum is smallbecause this state differs from the 11A state of dG by theinclusion of the excitation in dC Although the excitationenergy is shifted by the formation of hydrogen bonds betweendG and dC the rotatory strength of a large negative maycontribute to the strong negative sign of the CD spectra in Z-DNA In the 31A state (445 eV) which is the excitation fromnonbonding orbital to the LUMO in dC the rotatory strengthchanges to positive from the negative sign of the monomer dCHowever this state is not the origin of the second peak as willbe noted later in the tetramer model section Thus the rotatorystrength is strongly affected by the stacking interactionThree important excited states are calculated in the higher-

energy region of the second peak The 51A state (485 eV) hasthe largest oscillator strength and is assigned to the 31A (491eV) state of dG which is the excitation from HOMO to next-LUMO This state is assigned to the main peak (484 eV) of theexperimental UV spectrum of Z-DNA The 6 and 71A states(499 and 526 eV) with a large rotatory strength are the ETstate from the HOMO of dG to the LUMO of dC The mainconfiguration of the 71A state also includes the excitation fromthe next-HOMO to the LUMO in dCIn the B-Stack model the first peak with a positive sign is

assigned to the 11A state (404 eV) This state corresponds tothe 11A state (409 eV) of dC which is the excitation from theHOMO to the LUMO The 21A state (440 eV) is assigned tothe 11A state (434 eV) of dG which reflects the excitationfrom the HOMO to the LUMO Since the B-Stack model doesnot have hydrogen bonds the excitation energies are close tothose of dG or dC Both states have the oscillator strength butthe 11A state is positive while the 21A state is slightly negativefor the rotatory strength which is the same as for themonomers dC and dG respectively Therefore the stackinginteraction of B-DNA has little effect on the sign of the CDspectra at 295 nm because the stacking interaction is weaker inB-DNA than in Z-DNAThe second peak with a negative sign is assigned to the 5 and

61A states with a large rotatory strength The 5 and 61A states(507 and 512 eV in Table 4) are assigned to the 2 and 31Astates (493 and 518 eV in Table 3) of dG which are theexcitations from the nonbonding orbital and the HOMO to thenext-LUMO The main configuration is nminusπ for the 51A stateand πminusπ for the 61A state Both states have strong oscillatorand negative rotatory strengths which means that they can beassigned to the main peaksThe first ET excited states (71A state 521 eV) of B-Stack

model are calculated at a higher region than that (61A state499 eV) of the Z-Stack model because the stacking interactionis weaker in B-DNA than in Z-DNAThe SAC-CI results show that both hydrogen-bonding and

stacking interactions affect the UV and CD spectra of DNA Inparticular the excitation energy of the first excited stateincreases by the formation of hydrogen bonds and the stackinginteraction changes the sign of the CD spectrum of the firstexcited state from positive to negative The strong stackinginteraction characteristic of the helical structure of Z-DNA isthe origin of the strong negative sign at around 295 nm in theCD spectra The hydrogen-bonding and stacking interactions

Figure 7 SAC-CI CD spectra for the dimer models The SAC-CIspectra (magenta line) of (a) Z-Hbond model (b) Z-Stack model (c)B-Hbond model and (d) B-Stack model are compared with theexperimental CD spectra4 (black line) of Z-DNA (ab) and B-DNA(cd) The red and blue lines represent the intramolecular excitedstates of dG and dC respectively The green lines represent theelectron transfer (ET) excited state from dG to dC

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5551

also confer other smaller effects on the UV and CD spectra ofDNA In the higher energy region the SAC-CI CD spectra arenot in good agreement with the experimental CD spectrabecause these models have either hydrogen-bonding interactionor stacking interaction but not both Therefore we need toinvestigate a better model that includes both hydrogen bondingand stacking interactions within the same model A minimumsuch model is the tetramer model of Figure 1 on which weexplain in the next section35 Tetramer Model The SAC-CI calculations of the

dimer models were suitable to investigate the effects of thehydrogen-bonding interaction and the stacking interactionseparately For example the SAC-CI UV spectra of thehydrogen-bonding models were in good agreement with theexperimental UV spectra implying the importance of thehydrogen-bonding interaction The SAC-CI CD spectra for thestacking model clarified that the stacking interaction is thesource of the negative CD peak at around 295 nm Howevergenerally speaking the SAC-CI CD spectra were markedlydifferent from the experimental spectra The SAC-CI UVspectra of the stacking models were composed of two peaksdifferent from the experimental UV spectra Thus theinformation that we gained from the dimer models still didnot fully resolve these differences in the spectra acquired by twodifferent processes So we used the tetramer model (Figure 1)which included both hydrogen-bonding and stacking inter-actions within the same modelFigure 8 shows the SAC-CI UV and CD spectra of the

tetramer model of Z-DNA (Z-Tetra model) and B-DNA (B-Tetra model) in comparison with the experimental UV andCD spectra In order to compare the experimental spectra theSAC-CI results were shifted to lower energy by 05 eV Thecomputational conditions were poor as compared with theother SAC-CI calculations In the tetramer calculations the d-polarization functions were not added and the small activespace was used The SAC-CI calculations using the small activespace and without the d-polarization functions gave higherexcitation energies by about 04minus05 eV (see SupportingInformation Table S1) for the hydrogen-bonding dimermodels In particular the excited states corresponding to themain peak of the experimental UV spectra were shifted by 048eV for both the Z-Hbond model (61A state) and the B-Hbondmodel (71A state)Table 5 shows the excitation energies oscillator strengths

rotatory strengths natures and type of the excited states of theSAC-CI results for tetramer modelsThe SAC-CI UV spectrum of the Z-Tetra model is in good

agreement with the experimental UV spectrum of Z-DNA Themain peak as well as the shoulder peak is reproduced by theSAC-CI method The shoulder peak is assigned to the 4 and51A states which is the πminusπ excitation (HOMO to LUMO)in dC and the main peak is assigned to the 101A state which isthe πminusπ (HOMO to next-LUMO) excitation in dGThe SAC-CI UV spectrum of the B-Tetra model is also in a

reasonable agreement with the experimental UV spectrum of B-DNA although the shape of the shoulder is slightly different Inthe B-Tetra model the shoulder peak is assigned to the 21Astate which is the πminusπ (HOMO to LUMO) excitation in dCand the main peak is assigned to the 81A state which is theπminusπ (HOMO to next-LUMO) excitation in dGThe sign of the rotatory strength is the same between the

SAC-CI CD spectrum of the Z-Tetra model and theexperimental CD spectrum of Z-DNA although the negative

rotatory strength at 295 nm is very weak because thepolarization functions were not included in the tetramermodels It is well-known that the basis set must be good for anadequate description of the stacking interaction5859 In thedimer model the lack of polarization d-functions and thesmaller active space changed the intensity and sign of therotatory strength (see Supporting Information Table S1)Since the stacking interaction is small in B-DNA the SAC-CI

CD spectrum of the B-Tetra model is in good agreement withthe experimental CD spectrum of B-DNA and reproduces notonly the sign of the rotatory strength but also the overall shapeof the spectrumIn Z-DNA the 11A state is assigned to the first peak of the

experimental CD spectrum The 11A state is the ET excitationfrom the HOMO of dG to the LUMO of dC through thestacking conformation The 3 and 51A states with a largenegative rotatory strength are the excitation from the HOMOto the LUMO in dG and dC respectively These states alsocontribute to the first negative peak as in the 1 and 21A stateswith a negative rotatory strength in the Z-Stack model The 8and 101A states are assigned to the second peak of the CDspectrum The 101A state which is the origin of the main peakof the UV spectrum and the πminusπ of dC (81A state) alsocontribute to the second peak of the CD spectrumIn B-DNA the first peak of the CD spectrum is assigned to

the 21A state of πminusπ (HOMO to LUMO) in dC which is thesame as the origin of the first peak of the UV spectrum The ETstates are calculated at the higher energy region than the 21A

Figure 8 SAC-CI UV and CD spectra for the tetramer models TheSAC-CI spectra (magenta line) of Z-Tetra model (ab) and B-Tetramodel (cd) as compared with the experimental UV and CD spectra4

(black line) of Z-DNA (ab) and B-DNA (cd) The red and blue linesrepresent the intramolecular excited states of dG and dC respectivelyThe green lines represent the electron transfer (ET) excited state fromdG to dC All excited states of SAC-CI calculations have been shiftedto the lower values by 05 eV

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5552

Table

5SA

C-CIUVandCD

Spectraof

theTetramer

Mod

elsof

Z-andB-D

NA

Z-DNA

B-DNA

SACminusCIof

Z-Tetra

exptle

SAC-CIof

B-Tetra

exptle

EEa

Oscb

Rotc

UV

CD

EEa

Oscb

Rotc

UV

CD

state

(eV)

(nm)

(au)

(10minus

40cgs)

nature

typed

(eVnm)

(eVnm)

(eV)

(nm)

(au)

(10minus

40cgs)

nature

typed

(eVnm)

(eVnm)

11A

470

264

000

minus663

πrarr

πET

419296

479

259

003

minus041

πrarr

πdG

21A

481

258

002

6611

πrarr

πET

482

257

033

16891

πrarr

πdC

443280

31A

484

256

004

minus5326

πrarr

πdG

419296

484

256

001

minus7454

πrarr

πdC

41A

488

254

015

4553

πrarr

πdC

428290

498

249

007

minus130

πrarr

πET

51A

492

252

012

minus11712

πrarr

πdC

428290

419296

511

243

004

minus622

πrarr

πET

61A

502

247

003

minus3034

πrarr

πET

525

236

000

minus046

πrarr

πET

71A

513

242

001

2737

nrarr

πdG

527

235

001

332

πrarr

πET

81A

515

241

006

14772

πrarr

πdC

451275

545

227

045

minus15609

πrarr

πdG

484256

486255

91A

522

238

001

1978

nrarr

πdG

552

225

002

minus1795

nrarr

πdG

101 A

538

230

071

7299

πrarr

πdG

484256

451275

556

223

013

5929

πrarr

πdG

111 A

541

229

001

minus6858

πrarr

πdG

558

222

009

minus525

πrarr

πdG

121 A

548

226

002

minus1519

nrarr

πdC

563

220

008

minus7436

nrarr

πdG

131 A

553

224

001

minus924

nrarr

πdC

609

204

001

1379

nrarr

πdG

141 A

601

206

001

minus4159

nrarr

πdG

626

198

002

minus464

nrarr

πdG

aExcitatio

nenergyA

llexcitedstates

ofSA

C-CIcalculations

have

been

shifted

tothelower

values

by05eV

inFigure

8bOscillator

strengthcRotatorystrengthddG

representsintram

olecular

excited

states

ofdG

dCrepresentsintram

olecular

excitedstates

ofdC

and

ETrepresentselectron

transfer

excitedstates

from

dGto

dCeRef

4

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5553

state because the stacking interaction is weaker in B-DNA thanin Z-DNA (Figure 1) Therefore the first peak of theexperimental CD spectra originates from the ET excitationfrom dG to dC for Z-DNA but corresponds to theintramolecular excitation in dC for B-DNA The second peakwith a negative sign is assigned to the 81A state The 81A statewhich is the πminusπ of dG is the origin of both the main peaks ofthe UV spectrum and the second peak of the CD spectrumThe ET excited states of the tetramer models were calculated

at a lower energy region than those of the dimer models forboth B- and Z-DNA This result may be related to the overlapbetween the nucleic acid bases The overlap exists only in partof the nucleic acid bases in the dimer models but in thetetramer models the overlap covers the hydrogen bonds as wellas the nucleic acid bases The ET excited states are importantfor the CD spectrum of Z-DNA because the lowest excitedstate is the ET type and is negative for rotatory strength but forB-DNA the ET excited states are not important because thesestates are weak in both oscillator and rotatory strengthsThe SAC-CI results of the tetramer models are calculated at

a higher region by about 05 eV because of the small activespace Therefore better SAC-CI calculations are required forcomparison with the experimental UV and CD spectra of DNAHowever these results indicate that the strong stackinginteraction characteristic of the helical structure of Z-DNAchanges the sign of the rotatory strength for the lowest excitedstates of dG and dC and produces the emergence of the ETexcited states

4 CONCLUSIONSDNA chemistry is affected by three important factors (1) theconformation of the monomer (dG) which is related to thedihedral angle between deoxyribose and guanine (2) thehydrogen-bonding interactions and (3) the stacking inter-actions between two nucleic acid bases To verify whether thesethree factors are important for the CD spectra of DNA weexamined four different models using the SAC-CI method(1) dG monomer model We first applied the SAC-CI

method to the dG monomer to examine the effect ofconformation on the CD spectra of dG and from this resultto identify the structure of the dG monomer in solution Ourcalculations showed that the shape and the sign of the CDspectra depend strongly on the dihedral angle between thedeoxyribose and the guanine but that the SAC-CI UV spectraof both anti- and syn-dG are very similar because the UVspectra are not strongly dependent on the conformation Theexperimental CD spectrum agreed with the theoreticalspectrum of dG only for the anti conformation The resultingthermally averaged spectra also support the anti geometry ofdG as a stable conformation in solution However the SAC-CICD spectrum of syn-dG is opposite in sign from that of theexperimental spectrum Thus the SAC-CI CD spectra of dG inboth anti and syn conformations do not explain the negativesign at 295 nm of the CD spectra of the DNA composed of dGand dC(2) Composit model of dG and dC monomers A

comparison of the calculated composite spectra of dG anddC with the experimental spectra of the two-component DNAshowed similarities between B- and Z-DNA because theconformation of dC is always anti in both B- and Z-DNAThe SAC-CI results led us to confirm that that theconformation of dG is not the origin of the negative peak at295 nm in Z-DNA

(3) Dimer model Our examination of the dimer modelpermitted us to investigate the separate importance of thehydrogen-bonding and the stacking interactions in thestructural chemistry of B- and Z-DNA The SAC-CIcalculations showed that the hydrogen-bonding interactionschange the excitation energies and that the stacking interactionsare responsible for the changes in the sign of the CD rotatorystrength The SAC-CI UV spectra of the hydrogen-bondingmodels (B-Hbond and Z-Hbond models) are similar to theexperimental UV spectra in which the main peak as well as theshoulder peak is well reproduced The SAC-CI CD spectra ofthe stacking models (B-Stack and Z-Stack models) have thesame sign as the experimental CD spectra We concluded thatthe stacking interaction is especially important for the CDspectrum of the strongly stacked Z-DNA because the CD signalat 295 nm changed to a negative sign only for the Z-Stackmodel(4) Tetramer model Finally our calculations with the

tetramer model that contain both hydrogen-bonding andstacking interactions revealed that the shape of the SAC-CIUV and CD spectra compared well with the experimental onesThe ET excitations as well as the lowest intramolecularexcitation of dC and dG are the origins of the negative sign ofthe experimental CD spectrum of Z-DNA The ET excited stateis calculated at a lower energy region than the intramolecularexcited states of dG or dC for the strongly stacked Z-DNA butat a higher energy region for the weakly stacked B-DNA Thenegative sign of the first peak of the experimental CD spectra ofZ-DNA is due to the strong stacking effect characteristic of itshelical structureThe dynamic behavior of the dG monomer in solution was

partially considered by taking the Boltzmann averaging of thecalculated spectra over the various conformations Fortunatelythe Boltzmann averaged SAC-CI CD spectrum was in goodagreement with the experimental spectra although the SAC-CICD spectra largely depend on the conformation of dGHowever in this study we used only the static model for dimerand tetramer models The SAC-CI CD results of the dimer andtetramer models explained the difference between theexperimental CD spectra of B- and Z-DNA However sincethe structure of DNA would also be flexible in solution it isnecessary to take into account the dynamic behavior of DNAThis is an important and interesting topic

ASSOCIATED CONTENT

S Supporting InformationSAC-CI results of the hydrogen-bonding and stacking dimermodels of Z- and B-DNA using small active space and withoutthe polarization d-functions This material is available free ofcharge via the Internet at httppubsacsorg

AUTHOR INFORMATION

Corresponding AuthorPhone +81-75-634-3211 Fax +81-75-634-3211 E-mail hnakatsujiqcriorjp

NotesThe authors declare no competing financial interest

ACKNOWLEDGMENTS

The computations were performed at the Research Center forComputational Science Okazaki Japan

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5554

REFERENCES(1) Voet D Voet J G Eds Biochemistry John Wiley amp Sons IncNew York 1995(2) Ghosh A Bansal M Acta Crystallogr 2003 D59 620minus626(3) Rich A Nordheim A Wang A H-J Annu Rev Biochem 198453 791minus846(4) Tran-Dinh S Taboury J Neumann J-M Huynh-Dinh TGenissel B Langlois drsquoEstaintot B Igolen J Biochemistry 1984 231362minus1371(5) Sugiyama H Kawai K Matsunaga A Fujimoto K Saito IRobinson H Wang A H-J Nucleic Acids Res 1996 24 1272minus1278(6) Kawai K Saito I Sugiyama H J Am Chem Soc 1999 1211391minus1392(7) Oyoshi T Kawai K Sugiyama H J Am Chem Soc 2003 1251526minus1531(8) Xu Y Ikeda R Sugiyama H J Am Chem Soc 2003 12513519minus13524(9) Berova N Nakanishi K Woody R W Eds Circular DichroismPrinciples and Applications Wiley-VCH New York 2000(10) Bauernschmitt R Ahlrichs R Chem Phys Lett 1996 256454minus464(11) Casida M E Jamorski C Casida K C Salahub D R JChem Phys 1998 108 4439minus4449(12) Tsolakidis A Kaxiras E J Phys Chem A 2005 109 2373minus2380(13) Varsano D Di Felice R Marques M A L Rubio A J PhysChem B 2006 110 7129minus7138(14) Nielsen L M Holm A I S Varsano D Kadhane UHoffmann S V Di Felice R Rubio A Nielsen S B J Phys ChemB 2009 113 9614minus9619(15) Tonzani S Schatz G C J Am Chem Soc 2008 130 7607minus7612(16) Casida M E Salahub D R J Chem Phys 2000 113 8918minus8935(17) Dreuw A Weisman J L Head-Gordon M J Chem Phys2003 119 2943minus2946(18) Maitra N T Zhang F Cave R J Burke K J Chem Phys2004 120 5932minus5937(19) McCann D M Stephens P J J Org Chem 2006 71 6074minus6098(20) Casida M E J Mol Struct 2009 914 3minus18(21) Niehaus T A J Mol Struct 2009 914 38minus49(22) Tawada Y Tsuneda T Yanagisawa S Yanai T Hirao K JChem Phys 2004 120 8425minus8433(23) Chiba M Tsuneda T Hirao K J Chem Phys 2006 124144106-1-11(24) Nakatsuji H Hirao K J Chem Phys 1978 68 2053minus2065(25) Nakatsuji H Chem Phys Lett 1978 59 362minus364(26) Nakatsuji H Chem Phys Lett 1979 67 329minus333(27) Nakatsuji H Chem Phys Lett 1979 67 334minus342(28) Nakatsuji H Bull Chem Soc Jpn 2005 78 1705minus1724(29) Ehara M Hasegawa J Nakatsuji H SAC-CI Method Appliedto Molecular Spectroscopy In Theory and Applications of Computa-tional Chemistry The First 40 Years A Vol of Technical and HistoricalPerspectives Dykstra C E Frenking G Kim K S Scuseria G EEds Elsevier Oxford UK 2005 pp 1099minus1141(30) Miyahara T Hasegawa J Nakatsuji H Bull Chem Soc Jpn2009 82 1215minus1226(31) Bureekaew S Hasegawa J Nakatsuji H Chem Phys Lett2006 425 367minus371(32) Nakatani N Hasegawa J Nakatsuji H J Am Chem Soc2007 129 8756minus8765(33) Fujimoto K Hasegawa J Nakatsuji H Bull Chem Soc Jpn2009 82 1140minus1148(34) Miyahara T Nakatsuji H Collect Czech Chem Commun2011 76 537minus552(35) Sugiyama H Unpublished(36) Geertsen J Rittby M Bartlett R J Chem Phys Lett 1989164 57minus62

(37) Stanton J F Bartlett R J J Chem Phys 1993 98 7029minus7039(38) Ehara M Ishida M Toyota K Nakatsuji H SAC-CIGeneral-R Method Theory and Applications to the Multi-ElectronProcesses In Reviews in Modern Quantum Chemistry (A Tribute toProfessor Robert G Parr) Sen K D Ed World Scientific Singapore2003 pp 293minus319(39) SAC-CI GUIDE httpwwwqcriorjpsacci(40) Frisch M J Trucks G W Schlegel H B et al Gaussian 03development revision revision E05 Gaussian Inc Wallingford CT2004(41) Hohenberg P Kohn W Phys Rev 1964 136 B864minusB871(42) Kohn W Sham L J Phys Rev 1965 140 A1133minusA1138(43) Salahub D R Zerner M C Eds The Challenge of d and fElectrons The American Chemical Society Washington DC 1989(44) Parr R G Yang W Density-Functional Theory of Atoms andMolecules Oxford University Press Oxford UK 1989(45) Becke A D J Chem Phys 1993 98 5648minus5652(46) Lee C Yang W Parr R G Phys Rev 1988 37 785minus789(47) Hariharan P C Pople J A Theor Chim Acta 1973 28 213minus222(48) Francl M M Pietro W J Hehre W J Binkley J SDeFrees D J Pople J A Gordon M S J Chem Phys 1982 773654minus3665(49) Tomasi J Mennucci B Cammi R Chem Rev 2005 1052999minus3093(50) Fukuda R Ehara M Nakatsuji H Cammi R J Chem Phys2011 134 104109minus1minus11(51) Dunning T H Jr Hay P J In Modern Theoretical ChemistrySchaefer H F III Ed Plenum New York 1976 Vol 3 pp 1minus28(52) Nakatsuji H Chem Phys 1983 75 425minus441(53) Hocquet A Leulliot N Ghomi M J Phys Chem B 2000 1044560minus4568(54) Mishra S K Mishra P C J Comput Chem 2002 23 530minus540(55) Sprecher C A Johnson W C Biopolymers 1977 16 2243minus2264(56) Clark L B Peschel G G Tinoc I Jr J Phys Chem 1965 693615minus3618(57) Voet D Gratzer W B Cox R A Doty P Biopolymers 19631 193minus208(58) Hobza P Mehlhorn A Carsky P Zahradnik R J Mol Struct1986 138 387minus399(59) Hobza P Sponer J Chem Rev 1999 99 3247minus3276

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5555

Page 3: Helical Structure and Circular Dichroism Spectra of DNA: A ... · B-DNA, have right-handed double-helical structure, while Z-DNA is left-handed and appears only when DNA has a special

box in Figure 1bd) but in Z-DNA the guanine-cytosine (GC)pair strongly stacks with only one GC pair (denoted by thesquare box Figure 1ac) The stacking model of Z-DNA (Z-Stack model) has strong stacking interactions As in thegeometry optimization for dimer and tetramer models only thehydrogen atom positions in these models were optimized usingthe B3LYP6-31G(dp) basis with the PCM of waterFor the SACSAC-CI calculations the basis functions

employed were D9551 sets for deoxyribose and D95(d)51 setsfor the nucleic acid bases The core orbitals were treated asfrozen orbitals and all singles and selected doubles wereincluded Perturbation selection52 was carried out for thedouble excitation operators using the threshold sets of 5 times 10minus7

and 1 times 10minus7 hartree for SAC and SAC-CI calculationsrespectively In the calculations for the tetramer model thebasis functions employed were D95 sets (without polarizationfunctions) because the calculations were heavier than thecalculations for the dimer model and further the orbitalswhose energies were within minus11 to +11 au were chosen as theactive orbitals The SAC-CI UV and CD spectra wereconvoluted using Gaussian envelopes to describe the FranckminusCondon widths and the resolution of the spectrometer The fullwidth at half-maximum (fwhm) of the Gaussian was 08 eVwhich was estimated from the criterion that the two bands (253and 278 nm) looks like a single band in the experimental UVspectrum of dG the lowest two bands were not divided clearlyThis value (08 eV) was also used for the convolution of thetheoretical CD spectra Since DNA is a polymer composed ofmonomers we used the same value in dimer and tetramercalculations The UV and CD spectra of (1) dG with severalconformations (2) only dG and only dC in the X-raycrystallography structures of B- and Z-DNA (3) the hydro-gen-bonding and stacking models of B- and Z-DNA and (4)the tetramer models of B- and Z-DNA were calculated usingthe SAC-CI method

3 RESULTS AND DISCUSSION

31 Structure of the Ground State dG Monomer inWater We calculated the changes in the potential energy forthe ground state of the dG monomer in water as a function ofchanges in the dihedral angle (φ) between the deoxyribose andthe guanine (Figure 2) All of the geometrical parameters of dGwere optimized and φ was varied by incremental increases of20deg The optimized dihedral angle was calculated to be 3103degin the anti region and at 1295deg for the syn region The

deoxyribose has C3prime-endo conformation for both anti- and syn-dG with the present calculations although for anti-dG C2prime-endo was calculated to be more stable than C3prime-endo5354Thus in dG the conformation of the deoxyribose seems to bevery flexible and its energy difference very small However theconformation of deoxyribose is rigid in DNA the deoxyribosein dG is C2prime-endo for B-DNA and C3prime-endo for Z-DNA basedon the X-ray crystallography structures (9BNA and 1DCG)The anti-dG was calculated to be more stable than the syn-dGby 047 kcalmol The conformation of dG converts easily fromanti to syn (or from syn to anti) because the energy barrierbetween the anti and syn forms is only sim5 kcalmol The anti-dG rotates more easily in the clockwise direction (the directionof the arrow in Figure 2) than in the counterclockwise directionbecause the energy barrier is less than 4 kcalmol in theclockwise direction but is more than 5 kcalmol in thecounterclockwise direction The potential energy curve is flat inthe anti region while it is sharper in the syn region becausethere is a hydrogen bond interaction between the hydroxylgroup of the deoxyribose and the nitrogen atom of the guaninenear the syn conformation However the anti-dG can rotatemore freely between φ = 270deg and 330deg because of the lack ofhydrogen-bonding interaction

32 UV and CD Spectra of dG Monomer in WaterUsing the SAC-CI method we calculated the excitationenergies oscillator strengths rotatory strengths and the naturesof the excited states for both of the anti- and syn-dG (Table 1)The difference in the UV excitation spectra between anti- andsyn-dG is very small within 007 eV in energy and within 008au in oscillator strength for all the calculated states Howeverfor the CD spectra whose shapes are described by the rotatorystrengths the spectra of anti-dG differ markedly from thespectra of syn-dG not only in the absolute value but also in thesign of the rotatory strength Below we show these theoreticalresults in the spectral forms to facilitate the comparisons withthe experimentally obtained data

321 UV Spectra of dG Monomer The SAC-CI UV spectra(red lines) of anti- and syn-dG were compared with theexperimental spectrum (black lines) of dG above 200 nm(Figure 3) There are two main bands (276 and 250 nm) above200 nm and one band below 200 nm The SAC-CI UV spectraof both anti- and syn-dG are similar to the experimental UVspectrum and therefore it is unclear whether the experimentalUV spectrum of dG is due to the anti or the syn conformationThe first band at about 276 nm was assigned to the π rarr πhighest occupied molecular orbital (HOMO)minuslowest unoccu-pied molecular orbital (LUMO) transition (11A state at 446 eV(278 nm) for anti-dG and 440 eV (282 nm) for syn-dG) Thesecond band at approximately 250 nm was assigned to the 31Astate (at 510 eV (243 nm) for anti-dG and 506 eV (245 nm)for syn-dG) of π(HOMO) rarr π(next-LUMO) and 21A state(at 508 eV (244 nm) for anti-dG and 503 eV (247 nm) forsyn-dG) of the n rarr π nature As expected from its nature theoscillator strength of 21A is weak (005 au) for anti-dG and veryweak (001 au) for syn-dG The states 41A 51A and 61A (at591 597 and 608 eV (210 208 and 204 nm) for anti-dG and590 598 and 608 eV (210 207 and 204 nm) for syn-dG)were calculated to be in the region of the third band atapproximately 214 nm The oscillator strength is almost zerofor the n rarr π 41A and 61A states but is not zero for the π rarrπ 51A state However these states were not observed in theexperimental UV spectra because these three states are in themedian of the two strong peaks (31A and 71A states) and their

Figure 2 Potential energy curve of the ground state of dG determinedas dihedral angle φ is varied in increments of 20deg Other geometricalparameters were optimized at each φ

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5544

oscillator strengths are comparatively small The 71A state (at662 eV (187 nm) for anti-dG and 654 eV (189 nm) for syn-dG) was π rarr π (next-HOMO to LUMO and next-LUMO)excitation which was assigned to the fourth band comparedwith the experimental UV spectrum of deoxyguanosinemonophosphate (dGMP)55 Thus the SAC-CI UV spectra ofboth anti- and syn-dG are similar to the experimental UVspectrum The differences between the anti- and syn-dG werevery small because the HOMO next-HOMO LUMO andnext-LUMO of the main configuration for 11A 31A and 71Astates with large oscillator strengths are localized on theguanine site Hence we cannot determine if the experimentalUV spectrum reflects that of either anti- or syn-dG322 CD Spectra of dG Monomer in Water The SAC-CI

CD spectra (red lines) of the dG monomer at several dihedralangles φ were compared with the experimental CD spectrum(black line) above 200 nm (Figure 4) In the experimentallyacquired spectra the first band is at 276 nm which is strong inthe UV spectrum but very weak in the CD spectrum Thesecond band is at 250 nm which is strong in both UV and CDspectra but has a negative sign in the CD spectrum The thirdband at 214 nm is weak in the UV spectrum but stronglypositive in the CD spectrumThe SAC-CI CD spectra of dG with φ = 10minus150 and 350deg

(Figure 4aminushr) are quite different from the experimental CDspectrum The SAC-CI CD spectrum of syn-dG (1295deg) isopposite in sign for the second band as compared with theexperimental CD spectrum (Figure 4g) For dG with φ = 170210minus290deg (Figure 4ikminuso) the second band of the SAC-CI CDspectra agrees with the experimental CD spectrum but thethird band of the SAC-CI CD spectra has a weak intensitycompared with the experimental CD spectrum The SAC-CICD spectrum of the anti-dG and dG with φ = 330deg (Figure4pq) is in good overall agreement with the experimental CD

spectrum with the second band being strongly negative and thethird band being strongly positiveAs shown in Figure 2 the potential energy curve is flat

around the anti conformer and the energy barrier is not largeSo we calculated the ratio of the existence of each conformerassuming the Boltzmann distribution at 275 K a normalexperimental condition and using the ground state B3LYP6-31G(dp) energies In Table 2 the column labeled ldquoallrdquo givesthe ratio of each conformer when averaging included all 18conformers ldquoantirdquo gives the ratio of each conformer when onlyseven conformations around anti are used and ldquosynrdquo gives theratio of each conformer when only five conformations aroundsyn are used More than 70 of dG exists near the anticonformation between 270deg and 330deg and about 11 existsnear the syn-dG at 1295deg Within the syn region (90minus170deg)more than 50 dG exist as the syn-dG (1295deg) However inthe anti region (250minus10deg) dG exists mostly between 270deg and330deg as anti-dG The gradient of the potential energy is sharparound the syn conformer but flat around the anti conformerbecause dG forms a hydrogen bond with the deoxyribose onlyin the syn conformation Therefore for the CD spectra we mayconsider only the conformer at 1295deg for the syn conformerbut we have to consider several conformers around 3103deg forthe anti conformerWe also carried out the Boltzmann averaging for the spectra

of the anti (Figure 4p) and syn (Figure 4g) conformations TheSAC-CI CD spectra of ldquoallrdquo (green) ldquoantirdquo (blue) and singleanti-dG (red) are similar to the experimental CD spectrum asshown in Figure 4p while the SAC-CI CD spectra of ldquosynrdquo(blue) and single syn-dG (red) are opposite in sign to theexperimental CD spectrum (Figure 4g) These results confirmthat anti-dG is the main conformation under the experimentalcondition We note that the SAC-CI CD spectrum of dG at thesingle geometry of φ = 330deg is more similar to the experimentalCD spectrum of that of anti-dG (3103deg) This may indicatethat dG with φ = 330deg is more stable than the anti-dG insolutionThe dependence on the dihedral angle φ is very small in the

UV spectrum of Figure 3 because only π rarr π has strongintensity However the CD spectra of Figure 4 strongly dependon the dihedral angle φ because both π rarr π and n rarr π havestrong intensities and their MOs mainly localized at theguanine site are somewhat extended to the deoxyribose The11A state assigned to the first band at 276 nm was calculated tohave a small rotatory strength for both anti- and syn-dG butthis state was not observed in the experimental CD spectrumIn the second band region (250 nm) the rotatory strengths of

Table 1 SAC-CI UV and CD Spectra of anti- and syn-dG

SAC-CI

anti-dG syn-dG

EEa oscb rotc EEa oscb rotc

state (eV) (nm) (au) (10minus40 cgs) (eV) (nm) (au) (10minus40 cgs) nature exptld (eVnm)

11A 446 278 016 453 440 282 016 244 π rarr π 44927621A 508 244 005 minus7631 503 247 001 3812 n rarr π 49625031A 510 243 020 6820 506 245 022 minus3461 π rarr π 49625041A 591 210 000 1731 590 210 000 minus292 n rarr π 57921451A 597 208 002 minus225 598 207 005 830 π rarr π 57921461A 608 204 000 minus215 608 204 000 minus270 n rarr π 57921471A 662 187 044 minus782 654 189 038 minus078 π rarr π

aExcitation energy bOscillator strength cRotatory strength dRef 35

Figure 3 SAC-CI UV spectra of (a) anti- and (b) syn-dG (red)compared with the experimentally determined UV spectrum35 of dG(black)

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5545

both the 21A and 31A states are large and opposite in sign(Table 1) The 21A state is calculated to lie at 508 eV (244nm) for anti-dG and at 503 eV (247 nm) for syn-dG The 31Astate is calculated to lie at 510 eV (243 nm) for anti-dG and at506 eV (245 nm) for syn-dG The difference between 21A and31A in the excitation energy is very small so that their rotatorystrengths cancel each other giving relatively small rotatorystrengths From Table 1 we see that the total rotatory strengthof the second band is negative for anti-dG and positive for syn-dG To reproduce the experimental CD spectra the theoreticalrotatory strength must be reliable since otherwise reliablepositive or negative CD peaks cannot be obtained The SAC-CItheory seems to give reliable rotatory intensity since thetheoretical CD spectra agree well with the experimental one Inthe third band region (214 nm) the sum of the rotatorystrengths is positive for both anti- and syn-dG although thestrongest rotatory strength comes from the 41A (n rarr π) statefor anti-dG but from the 51A (π rarr π) state for syn-dG As inthe experimental CD spectrum the third band has largerintensity than the second band in the SAC-CI CD spectrum ofanti-dG However for syn-dG the third band in the SAC-CI

CD spectrum is smaller than the second band The 71A stateassigned to the fourth band with a strong negative intensity at190 nm in the experimental CD spectrum of dGMP55 isnegative for anti-dG and almost zero for syn-dG

33 Composite CD Spectra of dG and dC Comparedwith the CD Spectra of DNA We then extended our studiesto the spectra of B- and Z-DNA and we compared the SAC-CICD spectra of anti- and syn-dG (Figure 4) with theexperimental CD spectra of B- and Z-DNA (Figure 5)4 Thesecond band at 250 nm in the SAC-CI CD spectrum of anti-dGis similar to the experimental CD spectrum of B-DNA whilethe SAC-CI CD spectrum of syn-dG is similar to theexperimental CD spectrum of Z-DNA This result may indicatethat in DNA the second band at 250 nm reflects the dihedralangle φ of dG between the deoxyribose and the guanineTherefore we calculated the composite SAC-CI CD spectra ofthe noninteracting dG and dC monomer from the available X-ray crystallographic structures in B- and Z-DNA to clarify if thedifferent conformations of dG andor dC in different types ofDNA explain the differences in the CD spectra

Figure 4 CD spectra of dG at several conformation angles φ The experimental CD spectrum35 (black line) of dG is compared with the SAC-CI CDspectra (red lines) of (a) φ = 10deg (b) φ = 30deg (c) φ = 50deg (d) φ = 70deg (e) φ = 90deg (f) φ = 110deg (g) φ = 1295deg (syn) (h) φ = 150deg (i) φ = 170deg(j) φ = 190deg (k) φ = 210deg (l) φ = 230deg (m) φ = 250deg (n) φ = 270deg (o) φ = 290deg (p) φ = 3103deg (anti) (q) φ = 330deg and (r) φ = 350deg TheSAC-CI CD Boltzmann averaged spectra of (g) syn (blue line) (p) anti (blue line) and (p) all (green line) are also shown

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5546

The SAC-CI CD spectra of dG and dC alone (Figure5abde) having the structures of those in B- and Z-DNA andthe composite CD spectra (Figure 5cf) obtained as (a + b andd + e) were compared to the experimental CD spectra of B-and Z-DNA The SAC-CI excitation energies oscillatorstrengths rotatory strengths and natures of the excited statesfor the monomer model (dG and dC) were also determined(Table 3)The conformations of dG and dC are the same as those in

the ideal B-DNA or Z-DNA in which dG and dC are arrangedalternately (Figure 1) However the X-ray structure wasobtained for the DNA composed of only 6 sequences in thecase of Z-DNA Since the geometry is affected by the effect ofthe edge of DNA the conformation is different for eachmonomer Since the conformation of dG is in particular moreflexible than that of dC in DNA we calculated two kinds of dG(model I and II) in B- and Z-DNA respectively The models Iand II correspond to dG of the hydrogen-bonding models (B-Hbond and Z-Hbond models) and the stacking models (B-Stack and Z-Stack models) of the dimer models respectivelyThe dihedral angle (φ) between the deoxyribose and theguanine is 2763deg and 3007deg respectively for models I and IIof B-DNA and 1113deg and 1175deg respectively for models Iand II of Z-DNA The dihedral angle between the deoxyriboseand cytosine is 3232deg for B-DNA and 3329deg for Z-DNA Therotatory strength of dG is very different between models I andII for both B- and Z-DNA as shown in Table 3 Therefore fordG we prepared an averaged CD spectra for models I and II(Figure 5ad) For Z-DNA the SAC-CI CD spectrum of dG(Figure 5a) is negative for the first band and positive for thesecond band as in the experimental spectrum of Z-DNAalthough the excitation energies are calculated at a higherregion However the SAC-CI result is opposite in sign to theexperimental CD spectrum for the first band of B-DNA (Figure5d) These results show that in the CD spectra of DNA the

conformation of dG (monomer) is related to the sign of thesecond band but has no effect on the sign of the first bandThe CD spectra of dC have four peaks (Figure 5be) The

first peak represents the πminusπ of HOMO to LUMO Thesecond peak is the excitation from the nonbonding orbital tothe LUMO The third and fourth peaks are composed of boththe πminusπ and nminusπ from next-HOMO and nonbondingorbitals to LUMO The third and fourth peaks are mainly πminusπand nminusπ respectively The SAC-CI CD spectra of dC aresimilar in B- and Z-DNA with two positive and two negativepeaks The rotatory strength is positive for the first and thirdpeaks and negative for the second and fourth peaks Thereforethe CD spectra of dC are always positive at around 295 nmcorresponding to the conformation of dC which is always antiin DNAAs shown in Figure 5cf the composite SAC-CI CD spectra

obtained from dG and dC alone are similar between B- and Z-DNA The sign at 295 nm is positive for both B- and Z-DNAbecause the rotatory strength of dC which is the lowest peak ismuch stronger than that of dG Therefore the composite CD

Table 2 Boltzmann Distribution of dG along theConformational Angle φ

angle (φ) all ()a anti ()b syn ()c

10 086 1130 01450 00470 01090 170 91110 279 1431295 (syn) 1131 517150 454 224170 041 25190 008210 000230 010250 325 41270 1534 195290 1700 2163103 (anti) 2669 339330 1307 166350 258 33total 1000 1000 1000

aldquoallrdquo gives the ratio of each conformer when averaging included all 18conformers bldquoantirdquo gives the ratio of each conformer when only sevenconformations around anti are used cldquosynrdquo gives the ratio of eachconformer when only five conformations around syn are used

Figure 5 SAC-CI CD spectra of monomer ((a) dG in Z-DNA (b) dCin Z-DNA (d) dG in B-DNA and (e) dC in B-DNA) and thecomposite CD spectra of (c) Z-DNA and (f) B-DNA obtained fromthe SAC-CI CD spectra of dG and dC compared with theexperimental CD spectra4 (black lines) of Z-DNA (aminusc) and B-DNA (dminusf) The red and blue lines represent the excited state of dGand dC respectively The magenta line in panels c and f represents thecomposite spectra

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5547

spectrum is similar to that of dC for both B- and Z-DNA As aresult in the composite CD spectra of B- and Z-DNA the signat around 295 nm is independent of the geometry of themonomer (dG or dC) even though the CD spectrum of dGdepends on the dihedral angle between the guanine and thedeoxyribose because the lowest peak originates from dCwhose geometry is always anti in both B- and Z-DNA Thisindicates that the noninteracting dG and dC cannot explain theexperimental spectra of DNA and that the hydrogen-bondingor stacking interactions with the neighboring nucleic acid basesmust be important for the CD spectra of DNA34 UV and CD Spectra of Dimer Models In the studies

of the structure spectroscopy and functions of DNA it isgenerally recognized that the hydrogen-bonding interaction andthe stacking interaction are the two most important interactionsthat characterize the properties of DNA So we expect thatthese interactions might also be the important factors that affectthe features of the UV and CD spectra of DNA To investigatethe possible origin we considered the dimer model composedof dG and dC These models contained either the hydrogen-bonding interaction or the stacking interaction (Figure 1)However we note that the SAC-CI results using these modelsneed not be in agreement with the experimental spectrabecause these models lack either the stacking interaction or thehydrogen-bonding interaction We must use the tetramer

model at least to compare with the experimental spectra asnoted below

341 UV Spectra of Dimer Models Figure 6 shows theSAC-CI UV spectra of both hydrogen-bonding and stackingmodels of both B- and Z-DNA compared with the experimentalUV spectrum The experimental UV spectra of Z-DNA displaya main peak at 256 nm and a shoulder peak at around 290 nmThe experimental UV spectrum of B-DNA is very similar butthe shoulder peak at around 290 nm is weaker Table 4 showsthe excitation energies oscillator strengths rotatory strengthsnatures and the types of the excited states of the SAC-CIresults for both hydrogen-bonding and stacking models of thedimer When the excitation in the dimer model is not withindG or dC but corresponds to the electron transfer (ET)between dG and dC we assigned it as the ET typeThe SAC-CI UV spectrum of the hydrogen-bonding model

of Z-DNA (Z-Hbond model) is in good agreement with theexperimental UV spectrum of Z-DNA (Figure 6a) Threestrong peaks of the πminusπ nature are calculated The loweststrong peak corresponds to the shoulder peak at 428 eV (290nm) of the experimental UV spectrum of Z-DNA Theremaining two strong peaks correspond to the main peak at484 eV (256 nm) of the experimental UV spectrum of Z-DNARed blue and green colors represent the excited states of dGdC and ET respectively

Table 3 SAC-CI UV and CD Spectra of dG and dC in Z- and B-DNA

Z-DNA B-DNA

dG (model I φ = 1113deg) dG (model I φ = 2763deg)

EEa oscb rotc EEa oscb rotc

state (eV) (nm) (au) (10minus40 cgs) nature (eV) (nm) (au) (10minus40 cgs) nature

11A 441 281 017 125 π rarr π 443 280 016 minus1035 π rarr π21A 474 262 000 minus1099 n rarr π 488 254 000 1099 n rarr π31A 493 251 023 2066 π rarr π 513 242 025 minus3355 π rarr π41A 578 214 000 minus864 n rarr π 595 208 002 minus2619 π rarr π51A 591 210 004 minus656 π rarr σ 599 207 001 3571 n rarr π61A 611 203 000 minus438 n rarr π 618 201 000 983 n rarr π71A 628 197 000 359 π rarr σ 675 184 008 1235 π rarr σ81A 659 188 030 634 π rarr π 678 183 028 minus2402 π rarr π

dG (model II φ = 1175deg) dG (model II φ = 3007deg)

EEa oscb rotc EEa oscb rotc

state (eV) (nm) (au) (10minus40 cgs) nature (eV) (nm) (au) (10minus40 cgs) nature

11A 437 284 016 minus451 π rarr π 434 286 016 minus247 π rarr π21A 459 270 000 minus322 n rarr π 493 251 000 minus1467 n rarr π31A 491 252 022 minus319 π rarr π 518 239 021 minus200 π rarr π41A 581 213 003 323 π rarr π 595 209 001 2884 n rarr π51A 598 207 000 minus738 π rarr π 599 207 003 minus647 π rarr π61A 621 200 000 796 n rarr π 617 201 000 minus3033 n rarr π71A 641 193 000 353 π rarr σ 660 188 048 978 π rarr π81A 647 191 031 minus077 π rarr π 666 186 001 326 π rarr σ

dC (φ = 3329deg) dC (φ = 3232deg)

EEa oscb rotc EEa oscb rotc

state (eV) (nm) (au) (10minus40 cgs) nature (eV) (nm) (au) (10minus40 cgs) nature

11A 413 300 009 2385 π rarr π 409 303 009 1403 π rarr π21A 442 280 000 minus1548 n rarr π 452 274 000 minus915 n rarr π31A 507 245 011 8392 π + n rarr π 499 249 007 5975 π + n rarr π41A 527 235 011 minus8639 n + π rarr π 527 235 015 minus8017 π + n rarr π51A 573 216 000 minus1385 n rarr π 559 222 001 393 n rarr π61A 632 196 039 2349 π rarr π 625 198 032 1098 π rarr π

aExcitation energy bOscillator strength cRotatory strength

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5548

The shoulder peak in the experimental UV spectrum of Z-DNA is composed of 11A (439 eV) and 21A (453 eV) statesThe 11A state (in blue color) of the Z-Hbond modelcorresponds to the 11A state of dC at 413 eV which is theexcitation from the HOMO to the LUMO of dC This state haslarge oscillator strength of the πminusπ nature In the Z-Hbondmodel this state is blue-shifted by 026 eV because theHOMOminusLUMO gap (0449 au) of the Z-Hbond model islarger than that (0438 au) of dC This state is in goodagreement with the first (shoulder) peak of the experimentalUV spectrum of Z-DNA However the 21A state (453 eV redcolor) of the Z-Hbond model is assigned to the 31A state (493eV) of dG which is the excitation from the HOMO to the next-LUMO of dG This large shift of the excitation energy is causedby the MO-order change The MO of the Z-Hbond modelcorresponding to the next-LUMO of dG becomes lower thanthe MO of the Z-Hbond model corresponding to the LUMOof dG by the formation of hydrogen bonds with cytosine in theZ-Hbond model Therefore this state is red-shifted by 04 eVHowever the oscillator strength of this state is small despite itsπminusπ nature The transition moment is canceled because themain configuration includes not only the excitation from theHOMO to the next-LUMO in dG but also the excitation fromthe next-HOMO to the LUMO in dC and the electron transfertype excitation from the HOMO of dG to the LUMO and next-LUMO of dCThe main peak of the experimental UV spectrum of Z-DNA

is composed of 31A (484 eV) and 51A (493 eV) states havinglarge oscillator strengths (Figure 6a) The 31A state (so blue

color) of the Z-Hbond model is assigned to the 31A state (507eV) of dC which reflects the strong excitation from the next-HOMO to the LUMO of dC The 51A state (red color) of theZ-Hbond model is assigned to the 11A state (441 eV) of dGwhich is the excitation from the HOMO to the LUMO in dGThe excitation energy increases by 052 eV because theHOMOminusLUMO gap becomes large which is the opposite tothat of the 21A state but again due to the hydrogen-bondingeffectFor B-DNA however the SAC-CI UV spectrum of the

hydrogen-bonding model (B-Hbond model Figure 6c) also hasboth main and shoulder peaks in which the main peak is higherand the shoulder peak is lower as compared with theexperimental UV spectrum of B-DNA The main peak (41Aand 51A states) of the SAC-CI spectra of the B-Hbond model ishigher than that of the experimental spectra similarly to theSAC-CI spectrum of the Z-Hbond model The 31A state of dGin B-DNA is also higher than that in Z-DNA (Table 3) Theseresults may be related to the fact that dG is more flexible thandC because the conformation of dG is anti for B-DNA but synfor Z-DNA Since the stacking interaction of B-DNA is weakerthan that of Z-DNA the X-ray structure of B-DNA may bemore flexible than that of Z-DNA However the result that theoscillator strength of the lowest excited state of the B-Hbondmodel is weaker than that of the Z-Hbond model is reflected bythe fact that the shoulder peak of the experimental UV spectrais stronger in Z-DNA but weaker in B-DNAFor the SAC-CI UV spectra of the stacking models of both

B- (B-Stack model) and Z-DNA (Z-Stack model) the firstexcited states originated from dC were calculated at the lower-energy region (higher wavelength region) but the excitationenergy of the main peak is almost the same between the B-Hbond and B-Stack models or between the Z-Hbond and Z-Stack models If there is no hydrogen-bonding interaction inDNA the lowest excited state would be observed in the lower-energy region of the experimental spectra separately from themain peak The excitation energy of the lowest excited state ofthe stacking model is lower than that of the hydrogen-bondingmodel by 044 eV for Z-DNA and 033 eV for B-DNArespectively This result may correspond to the fact that thelowest excitation energy of cytosine differs between the gasphase (428 eV)56 and in solution (464 eV)57

For both B- and Z-DNA the SAC-CI results of thehydrogen-bonding models (B-Hbond and Z-Hbond) are inbetter agreement with the experimental spectra than those ofthe stacking models (B-Stack and Z-Stack) Thus thehydrogen-bonding interaction is reflected in the UV spectraof DNA On the other hand the shoulder peak observed for Z-DNA may reflect the stronger shoulder nature of the Z-Stackmodel in Z-DNA However the UV spectra of DNA providelittle information about stacking interactions between nucleicacid bases

342 CD Spectra of Dimer Models Figure 7 shows theSAC-CI CD spectra of both hydrogen-bonding and stackingmodels of B- and Z-DNA compared with the experimental CDspectra In the experimental CD spectra the first peak isnegative at 296 nm and the second peak is positive at 275 nmfor Z-DNA However B-DNA displays peaks that are oppositein sign the first peak is positive at 280 nm and the second peakis negative at 255 nmIt is worth nothing that the SAC-CI spectrum has a negative

peak at 295 nm only for the stacking model of Z-DNA (Z-Stack) but for the other models the SAC-CI spectrum is

Figure 6 SAC-CI UV spectra for the dimer models The SAC-CIspectra (magenta line) of (a) Z-Hbond model (b) Z-Stack model (c)B-Hbond model and (d) B-Stack model are compared with theexperimental UV spectra4 (black line) of Z-DNA (ab) and B-DNA(cd) The red and blue lines represent the intramolecular excitedstates of dG and dC respectively The green lines represent theelectron transfer (ET) excited state from dG to dC

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5549

Table

4SA

C-CIUVandCD

Spectraof

theHydrogen-Bon

ding

andStacking

Dim

erMod

elsof

Z-andB-D

NA

Z-DNA

B-DNA

SAC-CIof

Z-H

bond

exptle

SAC-CIof

B-H

bond

exptle

EEa

oscb

rotc

UV

EEa

oscb

rotc

UV

state

(eV)

(nm)

(au)

(10minus

40cgs)

nature

typed

(eVnm)

(eV)

(nm)

(au)

(10minus

40cgs)

nature

Typed

(eVnm)

11A

439

282

020

3513

πrarr

πdC

(11 A)

428290

437

284

017

1143

πrarr

πdC

(11 A)

21A

453

274

004

884

πrarr

πdG

(31 A)

456

272

003

minus1481

πrarr

πdG

(31 A)

31A

484

256

018

912

πrarr

πdC

(31 A)

493

251

006

583

πrarr

πET

41A

485

255

005

minus1145

πrarr

πET

498

249

021

1767

πrarr

πdC

(3+41A)

51A

493

252

042

1182

πrarr

πdG

(11 A)

484256

518

240

043

minus1390

πrarr

πdG

(11 A)

484256

61A

512

242

000

057

nrarr

πdC

(21 A)

525

236

000

minus1197

nrarr

πdC

(21 A)

71A

523

237

000

minus1384

nrarr

πdG

(21 A)

529

234

000

minus579

nrarr

πdG

(2+51A)

81A

623

199

000

minus034

nrarr

πdG

(2+51A)

SACminusCIof

Z-Stack

exptle

SACminusCIof

B-Stack

exptle

EEa

oscb

rotc

CD

EEa

oscb

rotc

CD

state

(eV)

(nm)

(au)

(10minus

40cgs)

nature

typed

(eVnm)

(eV)

(nm)

(au)

(10minus

40cgs)

nature

typed

(eVnm)

11A

395

314

009

minus4511

πrarr

πdC

(11 A)

419296

404

307

007

4358

πrarr

πdC

(11 A)

443280

21A

421

294

000

minus3349

πrarr

πdG

(11 A)

419296

440

282

006

minus1032

πrarr

πdG

(11 A)

31A

445

279

001

2967

nrarr

πdC

(21 A)

461

269

000

minus503

nrarr

πdC

(21 A)

41A

472

263

000

minus1335

nrarr

πdG

(21 A)

493

251

013

9407

πrarr

πdC

(3+41A)

51A

485

256

019

minus1796

πrarr

πdG

(31 A)

451275

507

245

005

minus15115

nrarr

πdG

(21 A)

486255

61A

499

248

006

4861

πrarr

πET

451275

512

242

014

minus11056

πrarr

πdG

(31 A)

486255

71A

526

236

007

2721

πrarr

πET

+dC

(31 A)

451275

521

238

014

19475

πrarr

πET+dG

(31 A)

81A

565

219

000

minus231

nrarr

πdC

(41 A)

619

200

000

minus805

nrarr

πdG

(41 A)

aExcitatio

nenergybOscillator

strengthcRotatorystrengthddG

representsintram

olecular

excitedstates

ofdG

dCrepresentsintram

olecular

excitedstates

ofdC

and

ETrepresentselectron

transfer

excitedstates

from

dGto

dCeRef

4

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5550

positive at 295 nm Therefore the strong negative sign of theCD spectra at 295 nm indicates that a strong stackinginteraction exists in Z-DNA The SAC-CI CD spectra of bothstacking models (B-Stack and Z-Stack models) more closelyresemble the experimental CD spectra of B- and Z-DNA thanthose of the hydrogen-bonding models (B-Hbond and Z-Hbond models) For the SAC-CI CD spectrum of the Z-Stackmodel the first and second peaks are negative and positiverespectively but they are the opposite of those obtained fromthe Z-Hbond model For the B-Stack and B-Hbond modelshowever the first peak is positive in the SAC-CI CD spectrabut the second and third peaks are negative and positiverespectively for the B-Stack model in opposition to the B-Hbond model Since the stacking models do not have thehydrogen-bonding interactions the excitation energy of the firstpeak is lower than that of the experimental spectrum as notedaboveIn the Z-Stack model the first peak with a negative sign is

assigned to the 11A (395 eV) and 21A (421 eV) states (Table4) The 11A state of the Z-Stack model corresponds to the 11Astate (413 eV) of dC which is the excitation of the πminusπ fromthe HOMO to the LUMO in dC A large oscillator strength isthe origin of the first (shoulder) peak but the excitation energyis lower than the experimental value because there is nohydrogen bond in the Z-Stack model The rotatory strength ofthe 11A state of dC changes the sign from positive to negativeby the stacking interaction with dG Therefore the stackinginteraction between dG and dC accounts for the strongnegative sign of the CD spectra in Z-DNA

The excitation energy of the 21A state (421 eV) is also closeto the first peak of the experimental CD spectrum This state isassigned to the 11A state (437 eV) of dG which is the HOMOto LUMO πminusπ excitation in dG but also includes thecontribution of the excitation from the next-HOMO to LUMOin dC The oscillator strength in the UV spectrum is smallbecause this state differs from the 11A state of dG by theinclusion of the excitation in dC Although the excitationenergy is shifted by the formation of hydrogen bonds betweendG and dC the rotatory strength of a large negative maycontribute to the strong negative sign of the CD spectra in Z-DNA In the 31A state (445 eV) which is the excitation fromnonbonding orbital to the LUMO in dC the rotatory strengthchanges to positive from the negative sign of the monomer dCHowever this state is not the origin of the second peak as willbe noted later in the tetramer model section Thus the rotatorystrength is strongly affected by the stacking interactionThree important excited states are calculated in the higher-

energy region of the second peak The 51A state (485 eV) hasthe largest oscillator strength and is assigned to the 31A (491eV) state of dG which is the excitation from HOMO to next-LUMO This state is assigned to the main peak (484 eV) of theexperimental UV spectrum of Z-DNA The 6 and 71A states(499 and 526 eV) with a large rotatory strength are the ETstate from the HOMO of dG to the LUMO of dC The mainconfiguration of the 71A state also includes the excitation fromthe next-HOMO to the LUMO in dCIn the B-Stack model the first peak with a positive sign is

assigned to the 11A state (404 eV) This state corresponds tothe 11A state (409 eV) of dC which is the excitation from theHOMO to the LUMO The 21A state (440 eV) is assigned tothe 11A state (434 eV) of dG which reflects the excitationfrom the HOMO to the LUMO Since the B-Stack model doesnot have hydrogen bonds the excitation energies are close tothose of dG or dC Both states have the oscillator strength butthe 11A state is positive while the 21A state is slightly negativefor the rotatory strength which is the same as for themonomers dC and dG respectively Therefore the stackinginteraction of B-DNA has little effect on the sign of the CDspectra at 295 nm because the stacking interaction is weaker inB-DNA than in Z-DNAThe second peak with a negative sign is assigned to the 5 and

61A states with a large rotatory strength The 5 and 61A states(507 and 512 eV in Table 4) are assigned to the 2 and 31Astates (493 and 518 eV in Table 3) of dG which are theexcitations from the nonbonding orbital and the HOMO to thenext-LUMO The main configuration is nminusπ for the 51A stateand πminusπ for the 61A state Both states have strong oscillatorand negative rotatory strengths which means that they can beassigned to the main peaksThe first ET excited states (71A state 521 eV) of B-Stack

model are calculated at a higher region than that (61A state499 eV) of the Z-Stack model because the stacking interactionis weaker in B-DNA than in Z-DNAThe SAC-CI results show that both hydrogen-bonding and

stacking interactions affect the UV and CD spectra of DNA Inparticular the excitation energy of the first excited stateincreases by the formation of hydrogen bonds and the stackinginteraction changes the sign of the CD spectrum of the firstexcited state from positive to negative The strong stackinginteraction characteristic of the helical structure of Z-DNA isthe origin of the strong negative sign at around 295 nm in theCD spectra The hydrogen-bonding and stacking interactions

Figure 7 SAC-CI CD spectra for the dimer models The SAC-CIspectra (magenta line) of (a) Z-Hbond model (b) Z-Stack model (c)B-Hbond model and (d) B-Stack model are compared with theexperimental CD spectra4 (black line) of Z-DNA (ab) and B-DNA(cd) The red and blue lines represent the intramolecular excitedstates of dG and dC respectively The green lines represent theelectron transfer (ET) excited state from dG to dC

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5551

also confer other smaller effects on the UV and CD spectra ofDNA In the higher energy region the SAC-CI CD spectra arenot in good agreement with the experimental CD spectrabecause these models have either hydrogen-bonding interactionor stacking interaction but not both Therefore we need toinvestigate a better model that includes both hydrogen bondingand stacking interactions within the same model A minimumsuch model is the tetramer model of Figure 1 on which weexplain in the next section35 Tetramer Model The SAC-CI calculations of the

dimer models were suitable to investigate the effects of thehydrogen-bonding interaction and the stacking interactionseparately For example the SAC-CI UV spectra of thehydrogen-bonding models were in good agreement with theexperimental UV spectra implying the importance of thehydrogen-bonding interaction The SAC-CI CD spectra for thestacking model clarified that the stacking interaction is thesource of the negative CD peak at around 295 nm Howevergenerally speaking the SAC-CI CD spectra were markedlydifferent from the experimental spectra The SAC-CI UVspectra of the stacking models were composed of two peaksdifferent from the experimental UV spectra Thus theinformation that we gained from the dimer models still didnot fully resolve these differences in the spectra acquired by twodifferent processes So we used the tetramer model (Figure 1)which included both hydrogen-bonding and stacking inter-actions within the same modelFigure 8 shows the SAC-CI UV and CD spectra of the

tetramer model of Z-DNA (Z-Tetra model) and B-DNA (B-Tetra model) in comparison with the experimental UV andCD spectra In order to compare the experimental spectra theSAC-CI results were shifted to lower energy by 05 eV Thecomputational conditions were poor as compared with theother SAC-CI calculations In the tetramer calculations the d-polarization functions were not added and the small activespace was used The SAC-CI calculations using the small activespace and without the d-polarization functions gave higherexcitation energies by about 04minus05 eV (see SupportingInformation Table S1) for the hydrogen-bonding dimermodels In particular the excited states corresponding to themain peak of the experimental UV spectra were shifted by 048eV for both the Z-Hbond model (61A state) and the B-Hbondmodel (71A state)Table 5 shows the excitation energies oscillator strengths

rotatory strengths natures and type of the excited states of theSAC-CI results for tetramer modelsThe SAC-CI UV spectrum of the Z-Tetra model is in good

agreement with the experimental UV spectrum of Z-DNA Themain peak as well as the shoulder peak is reproduced by theSAC-CI method The shoulder peak is assigned to the 4 and51A states which is the πminusπ excitation (HOMO to LUMO)in dC and the main peak is assigned to the 101A state which isthe πminusπ (HOMO to next-LUMO) excitation in dGThe SAC-CI UV spectrum of the B-Tetra model is also in a

reasonable agreement with the experimental UV spectrum of B-DNA although the shape of the shoulder is slightly different Inthe B-Tetra model the shoulder peak is assigned to the 21Astate which is the πminusπ (HOMO to LUMO) excitation in dCand the main peak is assigned to the 81A state which is theπminusπ (HOMO to next-LUMO) excitation in dGThe sign of the rotatory strength is the same between the

SAC-CI CD spectrum of the Z-Tetra model and theexperimental CD spectrum of Z-DNA although the negative

rotatory strength at 295 nm is very weak because thepolarization functions were not included in the tetramermodels It is well-known that the basis set must be good for anadequate description of the stacking interaction5859 In thedimer model the lack of polarization d-functions and thesmaller active space changed the intensity and sign of therotatory strength (see Supporting Information Table S1)Since the stacking interaction is small in B-DNA the SAC-CI

CD spectrum of the B-Tetra model is in good agreement withthe experimental CD spectrum of B-DNA and reproduces notonly the sign of the rotatory strength but also the overall shapeof the spectrumIn Z-DNA the 11A state is assigned to the first peak of the

experimental CD spectrum The 11A state is the ET excitationfrom the HOMO of dG to the LUMO of dC through thestacking conformation The 3 and 51A states with a largenegative rotatory strength are the excitation from the HOMOto the LUMO in dG and dC respectively These states alsocontribute to the first negative peak as in the 1 and 21A stateswith a negative rotatory strength in the Z-Stack model The 8and 101A states are assigned to the second peak of the CDspectrum The 101A state which is the origin of the main peakof the UV spectrum and the πminusπ of dC (81A state) alsocontribute to the second peak of the CD spectrumIn B-DNA the first peak of the CD spectrum is assigned to

the 21A state of πminusπ (HOMO to LUMO) in dC which is thesame as the origin of the first peak of the UV spectrum The ETstates are calculated at the higher energy region than the 21A

Figure 8 SAC-CI UV and CD spectra for the tetramer models TheSAC-CI spectra (magenta line) of Z-Tetra model (ab) and B-Tetramodel (cd) as compared with the experimental UV and CD spectra4

(black line) of Z-DNA (ab) and B-DNA (cd) The red and blue linesrepresent the intramolecular excited states of dG and dC respectivelyThe green lines represent the electron transfer (ET) excited state fromdG to dC All excited states of SAC-CI calculations have been shiftedto the lower values by 05 eV

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5552

Table

5SA

C-CIUVandCD

Spectraof

theTetramer

Mod

elsof

Z-andB-D

NA

Z-DNA

B-DNA

SACminusCIof

Z-Tetra

exptle

SAC-CIof

B-Tetra

exptle

EEa

Oscb

Rotc

UV

CD

EEa

Oscb

Rotc

UV

CD

state

(eV)

(nm)

(au)

(10minus

40cgs)

nature

typed

(eVnm)

(eVnm)

(eV)

(nm)

(au)

(10minus

40cgs)

nature

typed

(eVnm)

(eVnm)

11A

470

264

000

minus663

πrarr

πET

419296

479

259

003

minus041

πrarr

πdG

21A

481

258

002

6611

πrarr

πET

482

257

033

16891

πrarr

πdC

443280

31A

484

256

004

minus5326

πrarr

πdG

419296

484

256

001

minus7454

πrarr

πdC

41A

488

254

015

4553

πrarr

πdC

428290

498

249

007

minus130

πrarr

πET

51A

492

252

012

minus11712

πrarr

πdC

428290

419296

511

243

004

minus622

πrarr

πET

61A

502

247

003

minus3034

πrarr

πET

525

236

000

minus046

πrarr

πET

71A

513

242

001

2737

nrarr

πdG

527

235

001

332

πrarr

πET

81A

515

241

006

14772

πrarr

πdC

451275

545

227

045

minus15609

πrarr

πdG

484256

486255

91A

522

238

001

1978

nrarr

πdG

552

225

002

minus1795

nrarr

πdG

101 A

538

230

071

7299

πrarr

πdG

484256

451275

556

223

013

5929

πrarr

πdG

111 A

541

229

001

minus6858

πrarr

πdG

558

222

009

minus525

πrarr

πdG

121 A

548

226

002

minus1519

nrarr

πdC

563

220

008

minus7436

nrarr

πdG

131 A

553

224

001

minus924

nrarr

πdC

609

204

001

1379

nrarr

πdG

141 A

601

206

001

minus4159

nrarr

πdG

626

198

002

minus464

nrarr

πdG

aExcitatio

nenergyA

llexcitedstates

ofSA

C-CIcalculations

have

been

shifted

tothelower

values

by05eV

inFigure

8bOscillator

strengthcRotatorystrengthddG

representsintram

olecular

excited

states

ofdG

dCrepresentsintram

olecular

excitedstates

ofdC

and

ETrepresentselectron

transfer

excitedstates

from

dGto

dCeRef

4

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5553

state because the stacking interaction is weaker in B-DNA thanin Z-DNA (Figure 1) Therefore the first peak of theexperimental CD spectra originates from the ET excitationfrom dG to dC for Z-DNA but corresponds to theintramolecular excitation in dC for B-DNA The second peakwith a negative sign is assigned to the 81A state The 81A statewhich is the πminusπ of dG is the origin of both the main peaks ofthe UV spectrum and the second peak of the CD spectrumThe ET excited states of the tetramer models were calculated

at a lower energy region than those of the dimer models forboth B- and Z-DNA This result may be related to the overlapbetween the nucleic acid bases The overlap exists only in partof the nucleic acid bases in the dimer models but in thetetramer models the overlap covers the hydrogen bonds as wellas the nucleic acid bases The ET excited states are importantfor the CD spectrum of Z-DNA because the lowest excitedstate is the ET type and is negative for rotatory strength but forB-DNA the ET excited states are not important because thesestates are weak in both oscillator and rotatory strengthsThe SAC-CI results of the tetramer models are calculated at

a higher region by about 05 eV because of the small activespace Therefore better SAC-CI calculations are required forcomparison with the experimental UV and CD spectra of DNAHowever these results indicate that the strong stackinginteraction characteristic of the helical structure of Z-DNAchanges the sign of the rotatory strength for the lowest excitedstates of dG and dC and produces the emergence of the ETexcited states

4 CONCLUSIONSDNA chemistry is affected by three important factors (1) theconformation of the monomer (dG) which is related to thedihedral angle between deoxyribose and guanine (2) thehydrogen-bonding interactions and (3) the stacking inter-actions between two nucleic acid bases To verify whether thesethree factors are important for the CD spectra of DNA weexamined four different models using the SAC-CI method(1) dG monomer model We first applied the SAC-CI

method to the dG monomer to examine the effect ofconformation on the CD spectra of dG and from this resultto identify the structure of the dG monomer in solution Ourcalculations showed that the shape and the sign of the CDspectra depend strongly on the dihedral angle between thedeoxyribose and the guanine but that the SAC-CI UV spectraof both anti- and syn-dG are very similar because the UVspectra are not strongly dependent on the conformation Theexperimental CD spectrum agreed with the theoreticalspectrum of dG only for the anti conformation The resultingthermally averaged spectra also support the anti geometry ofdG as a stable conformation in solution However the SAC-CICD spectrum of syn-dG is opposite in sign from that of theexperimental spectrum Thus the SAC-CI CD spectra of dG inboth anti and syn conformations do not explain the negativesign at 295 nm of the CD spectra of the DNA composed of dGand dC(2) Composit model of dG and dC monomers A

comparison of the calculated composite spectra of dG anddC with the experimental spectra of the two-component DNAshowed similarities between B- and Z-DNA because theconformation of dC is always anti in both B- and Z-DNAThe SAC-CI results led us to confirm that that theconformation of dG is not the origin of the negative peak at295 nm in Z-DNA

(3) Dimer model Our examination of the dimer modelpermitted us to investigate the separate importance of thehydrogen-bonding and the stacking interactions in thestructural chemistry of B- and Z-DNA The SAC-CIcalculations showed that the hydrogen-bonding interactionschange the excitation energies and that the stacking interactionsare responsible for the changes in the sign of the CD rotatorystrength The SAC-CI UV spectra of the hydrogen-bondingmodels (B-Hbond and Z-Hbond models) are similar to theexperimental UV spectra in which the main peak as well as theshoulder peak is well reproduced The SAC-CI CD spectra ofthe stacking models (B-Stack and Z-Stack models) have thesame sign as the experimental CD spectra We concluded thatthe stacking interaction is especially important for the CDspectrum of the strongly stacked Z-DNA because the CD signalat 295 nm changed to a negative sign only for the Z-Stackmodel(4) Tetramer model Finally our calculations with the

tetramer model that contain both hydrogen-bonding andstacking interactions revealed that the shape of the SAC-CIUV and CD spectra compared well with the experimental onesThe ET excitations as well as the lowest intramolecularexcitation of dC and dG are the origins of the negative sign ofthe experimental CD spectrum of Z-DNA The ET excited stateis calculated at a lower energy region than the intramolecularexcited states of dG or dC for the strongly stacked Z-DNA butat a higher energy region for the weakly stacked B-DNA Thenegative sign of the first peak of the experimental CD spectra ofZ-DNA is due to the strong stacking effect characteristic of itshelical structureThe dynamic behavior of the dG monomer in solution was

partially considered by taking the Boltzmann averaging of thecalculated spectra over the various conformations Fortunatelythe Boltzmann averaged SAC-CI CD spectrum was in goodagreement with the experimental spectra although the SAC-CICD spectra largely depend on the conformation of dGHowever in this study we used only the static model for dimerand tetramer models The SAC-CI CD results of the dimer andtetramer models explained the difference between theexperimental CD spectra of B- and Z-DNA However sincethe structure of DNA would also be flexible in solution it isnecessary to take into account the dynamic behavior of DNAThis is an important and interesting topic

ASSOCIATED CONTENT

S Supporting InformationSAC-CI results of the hydrogen-bonding and stacking dimermodels of Z- and B-DNA using small active space and withoutthe polarization d-functions This material is available free ofcharge via the Internet at httppubsacsorg

AUTHOR INFORMATION

Corresponding AuthorPhone +81-75-634-3211 Fax +81-75-634-3211 E-mail hnakatsujiqcriorjp

NotesThe authors declare no competing financial interest

ACKNOWLEDGMENTS

The computations were performed at the Research Center forComputational Science Okazaki Japan

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5554

REFERENCES(1) Voet D Voet J G Eds Biochemistry John Wiley amp Sons IncNew York 1995(2) Ghosh A Bansal M Acta Crystallogr 2003 D59 620minus626(3) Rich A Nordheim A Wang A H-J Annu Rev Biochem 198453 791minus846(4) Tran-Dinh S Taboury J Neumann J-M Huynh-Dinh TGenissel B Langlois drsquoEstaintot B Igolen J Biochemistry 1984 231362minus1371(5) Sugiyama H Kawai K Matsunaga A Fujimoto K Saito IRobinson H Wang A H-J Nucleic Acids Res 1996 24 1272minus1278(6) Kawai K Saito I Sugiyama H J Am Chem Soc 1999 1211391minus1392(7) Oyoshi T Kawai K Sugiyama H J Am Chem Soc 2003 1251526minus1531(8) Xu Y Ikeda R Sugiyama H J Am Chem Soc 2003 12513519minus13524(9) Berova N Nakanishi K Woody R W Eds Circular DichroismPrinciples and Applications Wiley-VCH New York 2000(10) Bauernschmitt R Ahlrichs R Chem Phys Lett 1996 256454minus464(11) Casida M E Jamorski C Casida K C Salahub D R JChem Phys 1998 108 4439minus4449(12) Tsolakidis A Kaxiras E J Phys Chem A 2005 109 2373minus2380(13) Varsano D Di Felice R Marques M A L Rubio A J PhysChem B 2006 110 7129minus7138(14) Nielsen L M Holm A I S Varsano D Kadhane UHoffmann S V Di Felice R Rubio A Nielsen S B J Phys ChemB 2009 113 9614minus9619(15) Tonzani S Schatz G C J Am Chem Soc 2008 130 7607minus7612(16) Casida M E Salahub D R J Chem Phys 2000 113 8918minus8935(17) Dreuw A Weisman J L Head-Gordon M J Chem Phys2003 119 2943minus2946(18) Maitra N T Zhang F Cave R J Burke K J Chem Phys2004 120 5932minus5937(19) McCann D M Stephens P J J Org Chem 2006 71 6074minus6098(20) Casida M E J Mol Struct 2009 914 3minus18(21) Niehaus T A J Mol Struct 2009 914 38minus49(22) Tawada Y Tsuneda T Yanagisawa S Yanai T Hirao K JChem Phys 2004 120 8425minus8433(23) Chiba M Tsuneda T Hirao K J Chem Phys 2006 124144106-1-11(24) Nakatsuji H Hirao K J Chem Phys 1978 68 2053minus2065(25) Nakatsuji H Chem Phys Lett 1978 59 362minus364(26) Nakatsuji H Chem Phys Lett 1979 67 329minus333(27) Nakatsuji H Chem Phys Lett 1979 67 334minus342(28) Nakatsuji H Bull Chem Soc Jpn 2005 78 1705minus1724(29) Ehara M Hasegawa J Nakatsuji H SAC-CI Method Appliedto Molecular Spectroscopy In Theory and Applications of Computa-tional Chemistry The First 40 Years A Vol of Technical and HistoricalPerspectives Dykstra C E Frenking G Kim K S Scuseria G EEds Elsevier Oxford UK 2005 pp 1099minus1141(30) Miyahara T Hasegawa J Nakatsuji H Bull Chem Soc Jpn2009 82 1215minus1226(31) Bureekaew S Hasegawa J Nakatsuji H Chem Phys Lett2006 425 367minus371(32) Nakatani N Hasegawa J Nakatsuji H J Am Chem Soc2007 129 8756minus8765(33) Fujimoto K Hasegawa J Nakatsuji H Bull Chem Soc Jpn2009 82 1140minus1148(34) Miyahara T Nakatsuji H Collect Czech Chem Commun2011 76 537minus552(35) Sugiyama H Unpublished(36) Geertsen J Rittby M Bartlett R J Chem Phys Lett 1989164 57minus62

(37) Stanton J F Bartlett R J J Chem Phys 1993 98 7029minus7039(38) Ehara M Ishida M Toyota K Nakatsuji H SAC-CIGeneral-R Method Theory and Applications to the Multi-ElectronProcesses In Reviews in Modern Quantum Chemistry (A Tribute toProfessor Robert G Parr) Sen K D Ed World Scientific Singapore2003 pp 293minus319(39) SAC-CI GUIDE httpwwwqcriorjpsacci(40) Frisch M J Trucks G W Schlegel H B et al Gaussian 03development revision revision E05 Gaussian Inc Wallingford CT2004(41) Hohenberg P Kohn W Phys Rev 1964 136 B864minusB871(42) Kohn W Sham L J Phys Rev 1965 140 A1133minusA1138(43) Salahub D R Zerner M C Eds The Challenge of d and fElectrons The American Chemical Society Washington DC 1989(44) Parr R G Yang W Density-Functional Theory of Atoms andMolecules Oxford University Press Oxford UK 1989(45) Becke A D J Chem Phys 1993 98 5648minus5652(46) Lee C Yang W Parr R G Phys Rev 1988 37 785minus789(47) Hariharan P C Pople J A Theor Chim Acta 1973 28 213minus222(48) Francl M M Pietro W J Hehre W J Binkley J SDeFrees D J Pople J A Gordon M S J Chem Phys 1982 773654minus3665(49) Tomasi J Mennucci B Cammi R Chem Rev 2005 1052999minus3093(50) Fukuda R Ehara M Nakatsuji H Cammi R J Chem Phys2011 134 104109minus1minus11(51) Dunning T H Jr Hay P J In Modern Theoretical ChemistrySchaefer H F III Ed Plenum New York 1976 Vol 3 pp 1minus28(52) Nakatsuji H Chem Phys 1983 75 425minus441(53) Hocquet A Leulliot N Ghomi M J Phys Chem B 2000 1044560minus4568(54) Mishra S K Mishra P C J Comput Chem 2002 23 530minus540(55) Sprecher C A Johnson W C Biopolymers 1977 16 2243minus2264(56) Clark L B Peschel G G Tinoc I Jr J Phys Chem 1965 693615minus3618(57) Voet D Gratzer W B Cox R A Doty P Biopolymers 19631 193minus208(58) Hobza P Mehlhorn A Carsky P Zahradnik R J Mol Struct1986 138 387minus399(59) Hobza P Sponer J Chem Rev 1999 99 3247minus3276

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5555

Page 4: Helical Structure and Circular Dichroism Spectra of DNA: A ... · B-DNA, have right-handed double-helical structure, while Z-DNA is left-handed and appears only when DNA has a special

oscillator strengths are comparatively small The 71A state (at662 eV (187 nm) for anti-dG and 654 eV (189 nm) for syn-dG) was π rarr π (next-HOMO to LUMO and next-LUMO)excitation which was assigned to the fourth band comparedwith the experimental UV spectrum of deoxyguanosinemonophosphate (dGMP)55 Thus the SAC-CI UV spectra ofboth anti- and syn-dG are similar to the experimental UVspectrum The differences between the anti- and syn-dG werevery small because the HOMO next-HOMO LUMO andnext-LUMO of the main configuration for 11A 31A and 71Astates with large oscillator strengths are localized on theguanine site Hence we cannot determine if the experimentalUV spectrum reflects that of either anti- or syn-dG322 CD Spectra of dG Monomer in Water The SAC-CI

CD spectra (red lines) of the dG monomer at several dihedralangles φ were compared with the experimental CD spectrum(black line) above 200 nm (Figure 4) In the experimentallyacquired spectra the first band is at 276 nm which is strong inthe UV spectrum but very weak in the CD spectrum Thesecond band is at 250 nm which is strong in both UV and CDspectra but has a negative sign in the CD spectrum The thirdband at 214 nm is weak in the UV spectrum but stronglypositive in the CD spectrumThe SAC-CI CD spectra of dG with φ = 10minus150 and 350deg

(Figure 4aminushr) are quite different from the experimental CDspectrum The SAC-CI CD spectrum of syn-dG (1295deg) isopposite in sign for the second band as compared with theexperimental CD spectrum (Figure 4g) For dG with φ = 170210minus290deg (Figure 4ikminuso) the second band of the SAC-CI CDspectra agrees with the experimental CD spectrum but thethird band of the SAC-CI CD spectra has a weak intensitycompared with the experimental CD spectrum The SAC-CICD spectrum of the anti-dG and dG with φ = 330deg (Figure4pq) is in good overall agreement with the experimental CD

spectrum with the second band being strongly negative and thethird band being strongly positiveAs shown in Figure 2 the potential energy curve is flat

around the anti conformer and the energy barrier is not largeSo we calculated the ratio of the existence of each conformerassuming the Boltzmann distribution at 275 K a normalexperimental condition and using the ground state B3LYP6-31G(dp) energies In Table 2 the column labeled ldquoallrdquo givesthe ratio of each conformer when averaging included all 18conformers ldquoantirdquo gives the ratio of each conformer when onlyseven conformations around anti are used and ldquosynrdquo gives theratio of each conformer when only five conformations aroundsyn are used More than 70 of dG exists near the anticonformation between 270deg and 330deg and about 11 existsnear the syn-dG at 1295deg Within the syn region (90minus170deg)more than 50 dG exist as the syn-dG (1295deg) However inthe anti region (250minus10deg) dG exists mostly between 270deg and330deg as anti-dG The gradient of the potential energy is sharparound the syn conformer but flat around the anti conformerbecause dG forms a hydrogen bond with the deoxyribose onlyin the syn conformation Therefore for the CD spectra we mayconsider only the conformer at 1295deg for the syn conformerbut we have to consider several conformers around 3103deg forthe anti conformerWe also carried out the Boltzmann averaging for the spectra

of the anti (Figure 4p) and syn (Figure 4g) conformations TheSAC-CI CD spectra of ldquoallrdquo (green) ldquoantirdquo (blue) and singleanti-dG (red) are similar to the experimental CD spectrum asshown in Figure 4p while the SAC-CI CD spectra of ldquosynrdquo(blue) and single syn-dG (red) are opposite in sign to theexperimental CD spectrum (Figure 4g) These results confirmthat anti-dG is the main conformation under the experimentalcondition We note that the SAC-CI CD spectrum of dG at thesingle geometry of φ = 330deg is more similar to the experimentalCD spectrum of that of anti-dG (3103deg) This may indicatethat dG with φ = 330deg is more stable than the anti-dG insolutionThe dependence on the dihedral angle φ is very small in the

UV spectrum of Figure 3 because only π rarr π has strongintensity However the CD spectra of Figure 4 strongly dependon the dihedral angle φ because both π rarr π and n rarr π havestrong intensities and their MOs mainly localized at theguanine site are somewhat extended to the deoxyribose The11A state assigned to the first band at 276 nm was calculated tohave a small rotatory strength for both anti- and syn-dG butthis state was not observed in the experimental CD spectrumIn the second band region (250 nm) the rotatory strengths of

Table 1 SAC-CI UV and CD Spectra of anti- and syn-dG

SAC-CI

anti-dG syn-dG

EEa oscb rotc EEa oscb rotc

state (eV) (nm) (au) (10minus40 cgs) (eV) (nm) (au) (10minus40 cgs) nature exptld (eVnm)

11A 446 278 016 453 440 282 016 244 π rarr π 44927621A 508 244 005 minus7631 503 247 001 3812 n rarr π 49625031A 510 243 020 6820 506 245 022 minus3461 π rarr π 49625041A 591 210 000 1731 590 210 000 minus292 n rarr π 57921451A 597 208 002 minus225 598 207 005 830 π rarr π 57921461A 608 204 000 minus215 608 204 000 minus270 n rarr π 57921471A 662 187 044 minus782 654 189 038 minus078 π rarr π

aExcitation energy bOscillator strength cRotatory strength dRef 35

Figure 3 SAC-CI UV spectra of (a) anti- and (b) syn-dG (red)compared with the experimentally determined UV spectrum35 of dG(black)

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5545

both the 21A and 31A states are large and opposite in sign(Table 1) The 21A state is calculated to lie at 508 eV (244nm) for anti-dG and at 503 eV (247 nm) for syn-dG The 31Astate is calculated to lie at 510 eV (243 nm) for anti-dG and at506 eV (245 nm) for syn-dG The difference between 21A and31A in the excitation energy is very small so that their rotatorystrengths cancel each other giving relatively small rotatorystrengths From Table 1 we see that the total rotatory strengthof the second band is negative for anti-dG and positive for syn-dG To reproduce the experimental CD spectra the theoreticalrotatory strength must be reliable since otherwise reliablepositive or negative CD peaks cannot be obtained The SAC-CItheory seems to give reliable rotatory intensity since thetheoretical CD spectra agree well with the experimental one Inthe third band region (214 nm) the sum of the rotatorystrengths is positive for both anti- and syn-dG although thestrongest rotatory strength comes from the 41A (n rarr π) statefor anti-dG but from the 51A (π rarr π) state for syn-dG As inthe experimental CD spectrum the third band has largerintensity than the second band in the SAC-CI CD spectrum ofanti-dG However for syn-dG the third band in the SAC-CI

CD spectrum is smaller than the second band The 71A stateassigned to the fourth band with a strong negative intensity at190 nm in the experimental CD spectrum of dGMP55 isnegative for anti-dG and almost zero for syn-dG

33 Composite CD Spectra of dG and dC Comparedwith the CD Spectra of DNA We then extended our studiesto the spectra of B- and Z-DNA and we compared the SAC-CICD spectra of anti- and syn-dG (Figure 4) with theexperimental CD spectra of B- and Z-DNA (Figure 5)4 Thesecond band at 250 nm in the SAC-CI CD spectrum of anti-dGis similar to the experimental CD spectrum of B-DNA whilethe SAC-CI CD spectrum of syn-dG is similar to theexperimental CD spectrum of Z-DNA This result may indicatethat in DNA the second band at 250 nm reflects the dihedralangle φ of dG between the deoxyribose and the guanineTherefore we calculated the composite SAC-CI CD spectra ofthe noninteracting dG and dC monomer from the available X-ray crystallographic structures in B- and Z-DNA to clarify if thedifferent conformations of dG andor dC in different types ofDNA explain the differences in the CD spectra

Figure 4 CD spectra of dG at several conformation angles φ The experimental CD spectrum35 (black line) of dG is compared with the SAC-CI CDspectra (red lines) of (a) φ = 10deg (b) φ = 30deg (c) φ = 50deg (d) φ = 70deg (e) φ = 90deg (f) φ = 110deg (g) φ = 1295deg (syn) (h) φ = 150deg (i) φ = 170deg(j) φ = 190deg (k) φ = 210deg (l) φ = 230deg (m) φ = 250deg (n) φ = 270deg (o) φ = 290deg (p) φ = 3103deg (anti) (q) φ = 330deg and (r) φ = 350deg TheSAC-CI CD Boltzmann averaged spectra of (g) syn (blue line) (p) anti (blue line) and (p) all (green line) are also shown

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5546

The SAC-CI CD spectra of dG and dC alone (Figure5abde) having the structures of those in B- and Z-DNA andthe composite CD spectra (Figure 5cf) obtained as (a + b andd + e) were compared to the experimental CD spectra of B-and Z-DNA The SAC-CI excitation energies oscillatorstrengths rotatory strengths and natures of the excited statesfor the monomer model (dG and dC) were also determined(Table 3)The conformations of dG and dC are the same as those in

the ideal B-DNA or Z-DNA in which dG and dC are arrangedalternately (Figure 1) However the X-ray structure wasobtained for the DNA composed of only 6 sequences in thecase of Z-DNA Since the geometry is affected by the effect ofthe edge of DNA the conformation is different for eachmonomer Since the conformation of dG is in particular moreflexible than that of dC in DNA we calculated two kinds of dG(model I and II) in B- and Z-DNA respectively The models Iand II correspond to dG of the hydrogen-bonding models (B-Hbond and Z-Hbond models) and the stacking models (B-Stack and Z-Stack models) of the dimer models respectivelyThe dihedral angle (φ) between the deoxyribose and theguanine is 2763deg and 3007deg respectively for models I and IIof B-DNA and 1113deg and 1175deg respectively for models Iand II of Z-DNA The dihedral angle between the deoxyriboseand cytosine is 3232deg for B-DNA and 3329deg for Z-DNA Therotatory strength of dG is very different between models I andII for both B- and Z-DNA as shown in Table 3 Therefore fordG we prepared an averaged CD spectra for models I and II(Figure 5ad) For Z-DNA the SAC-CI CD spectrum of dG(Figure 5a) is negative for the first band and positive for thesecond band as in the experimental spectrum of Z-DNAalthough the excitation energies are calculated at a higherregion However the SAC-CI result is opposite in sign to theexperimental CD spectrum for the first band of B-DNA (Figure5d) These results show that in the CD spectra of DNA the

conformation of dG (monomer) is related to the sign of thesecond band but has no effect on the sign of the first bandThe CD spectra of dC have four peaks (Figure 5be) The

first peak represents the πminusπ of HOMO to LUMO Thesecond peak is the excitation from the nonbonding orbital tothe LUMO The third and fourth peaks are composed of boththe πminusπ and nminusπ from next-HOMO and nonbondingorbitals to LUMO The third and fourth peaks are mainly πminusπand nminusπ respectively The SAC-CI CD spectra of dC aresimilar in B- and Z-DNA with two positive and two negativepeaks The rotatory strength is positive for the first and thirdpeaks and negative for the second and fourth peaks Thereforethe CD spectra of dC are always positive at around 295 nmcorresponding to the conformation of dC which is always antiin DNAAs shown in Figure 5cf the composite SAC-CI CD spectra

obtained from dG and dC alone are similar between B- and Z-DNA The sign at 295 nm is positive for both B- and Z-DNAbecause the rotatory strength of dC which is the lowest peak ismuch stronger than that of dG Therefore the composite CD

Table 2 Boltzmann Distribution of dG along theConformational Angle φ

angle (φ) all ()a anti ()b syn ()c

10 086 1130 01450 00470 01090 170 91110 279 1431295 (syn) 1131 517150 454 224170 041 25190 008210 000230 010250 325 41270 1534 195290 1700 2163103 (anti) 2669 339330 1307 166350 258 33total 1000 1000 1000

aldquoallrdquo gives the ratio of each conformer when averaging included all 18conformers bldquoantirdquo gives the ratio of each conformer when only sevenconformations around anti are used cldquosynrdquo gives the ratio of eachconformer when only five conformations around syn are used

Figure 5 SAC-CI CD spectra of monomer ((a) dG in Z-DNA (b) dCin Z-DNA (d) dG in B-DNA and (e) dC in B-DNA) and thecomposite CD spectra of (c) Z-DNA and (f) B-DNA obtained fromthe SAC-CI CD spectra of dG and dC compared with theexperimental CD spectra4 (black lines) of Z-DNA (aminusc) and B-DNA (dminusf) The red and blue lines represent the excited state of dGand dC respectively The magenta line in panels c and f represents thecomposite spectra

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5547

spectrum is similar to that of dC for both B- and Z-DNA As aresult in the composite CD spectra of B- and Z-DNA the signat around 295 nm is independent of the geometry of themonomer (dG or dC) even though the CD spectrum of dGdepends on the dihedral angle between the guanine and thedeoxyribose because the lowest peak originates from dCwhose geometry is always anti in both B- and Z-DNA Thisindicates that the noninteracting dG and dC cannot explain theexperimental spectra of DNA and that the hydrogen-bondingor stacking interactions with the neighboring nucleic acid basesmust be important for the CD spectra of DNA34 UV and CD Spectra of Dimer Models In the studies

of the structure spectroscopy and functions of DNA it isgenerally recognized that the hydrogen-bonding interaction andthe stacking interaction are the two most important interactionsthat characterize the properties of DNA So we expect thatthese interactions might also be the important factors that affectthe features of the UV and CD spectra of DNA To investigatethe possible origin we considered the dimer model composedof dG and dC These models contained either the hydrogen-bonding interaction or the stacking interaction (Figure 1)However we note that the SAC-CI results using these modelsneed not be in agreement with the experimental spectrabecause these models lack either the stacking interaction or thehydrogen-bonding interaction We must use the tetramer

model at least to compare with the experimental spectra asnoted below

341 UV Spectra of Dimer Models Figure 6 shows theSAC-CI UV spectra of both hydrogen-bonding and stackingmodels of both B- and Z-DNA compared with the experimentalUV spectrum The experimental UV spectra of Z-DNA displaya main peak at 256 nm and a shoulder peak at around 290 nmThe experimental UV spectrum of B-DNA is very similar butthe shoulder peak at around 290 nm is weaker Table 4 showsthe excitation energies oscillator strengths rotatory strengthsnatures and the types of the excited states of the SAC-CIresults for both hydrogen-bonding and stacking models of thedimer When the excitation in the dimer model is not withindG or dC but corresponds to the electron transfer (ET)between dG and dC we assigned it as the ET typeThe SAC-CI UV spectrum of the hydrogen-bonding model

of Z-DNA (Z-Hbond model) is in good agreement with theexperimental UV spectrum of Z-DNA (Figure 6a) Threestrong peaks of the πminusπ nature are calculated The loweststrong peak corresponds to the shoulder peak at 428 eV (290nm) of the experimental UV spectrum of Z-DNA Theremaining two strong peaks correspond to the main peak at484 eV (256 nm) of the experimental UV spectrum of Z-DNARed blue and green colors represent the excited states of dGdC and ET respectively

Table 3 SAC-CI UV and CD Spectra of dG and dC in Z- and B-DNA

Z-DNA B-DNA

dG (model I φ = 1113deg) dG (model I φ = 2763deg)

EEa oscb rotc EEa oscb rotc

state (eV) (nm) (au) (10minus40 cgs) nature (eV) (nm) (au) (10minus40 cgs) nature

11A 441 281 017 125 π rarr π 443 280 016 minus1035 π rarr π21A 474 262 000 minus1099 n rarr π 488 254 000 1099 n rarr π31A 493 251 023 2066 π rarr π 513 242 025 minus3355 π rarr π41A 578 214 000 minus864 n rarr π 595 208 002 minus2619 π rarr π51A 591 210 004 minus656 π rarr σ 599 207 001 3571 n rarr π61A 611 203 000 minus438 n rarr π 618 201 000 983 n rarr π71A 628 197 000 359 π rarr σ 675 184 008 1235 π rarr σ81A 659 188 030 634 π rarr π 678 183 028 minus2402 π rarr π

dG (model II φ = 1175deg) dG (model II φ = 3007deg)

EEa oscb rotc EEa oscb rotc

state (eV) (nm) (au) (10minus40 cgs) nature (eV) (nm) (au) (10minus40 cgs) nature

11A 437 284 016 minus451 π rarr π 434 286 016 minus247 π rarr π21A 459 270 000 minus322 n rarr π 493 251 000 minus1467 n rarr π31A 491 252 022 minus319 π rarr π 518 239 021 minus200 π rarr π41A 581 213 003 323 π rarr π 595 209 001 2884 n rarr π51A 598 207 000 minus738 π rarr π 599 207 003 minus647 π rarr π61A 621 200 000 796 n rarr π 617 201 000 minus3033 n rarr π71A 641 193 000 353 π rarr σ 660 188 048 978 π rarr π81A 647 191 031 minus077 π rarr π 666 186 001 326 π rarr σ

dC (φ = 3329deg) dC (φ = 3232deg)

EEa oscb rotc EEa oscb rotc

state (eV) (nm) (au) (10minus40 cgs) nature (eV) (nm) (au) (10minus40 cgs) nature

11A 413 300 009 2385 π rarr π 409 303 009 1403 π rarr π21A 442 280 000 minus1548 n rarr π 452 274 000 minus915 n rarr π31A 507 245 011 8392 π + n rarr π 499 249 007 5975 π + n rarr π41A 527 235 011 minus8639 n + π rarr π 527 235 015 minus8017 π + n rarr π51A 573 216 000 minus1385 n rarr π 559 222 001 393 n rarr π61A 632 196 039 2349 π rarr π 625 198 032 1098 π rarr π

aExcitation energy bOscillator strength cRotatory strength

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5548

The shoulder peak in the experimental UV spectrum of Z-DNA is composed of 11A (439 eV) and 21A (453 eV) statesThe 11A state (in blue color) of the Z-Hbond modelcorresponds to the 11A state of dC at 413 eV which is theexcitation from the HOMO to the LUMO of dC This state haslarge oscillator strength of the πminusπ nature In the Z-Hbondmodel this state is blue-shifted by 026 eV because theHOMOminusLUMO gap (0449 au) of the Z-Hbond model islarger than that (0438 au) of dC This state is in goodagreement with the first (shoulder) peak of the experimentalUV spectrum of Z-DNA However the 21A state (453 eV redcolor) of the Z-Hbond model is assigned to the 31A state (493eV) of dG which is the excitation from the HOMO to the next-LUMO of dG This large shift of the excitation energy is causedby the MO-order change The MO of the Z-Hbond modelcorresponding to the next-LUMO of dG becomes lower thanthe MO of the Z-Hbond model corresponding to the LUMOof dG by the formation of hydrogen bonds with cytosine in theZ-Hbond model Therefore this state is red-shifted by 04 eVHowever the oscillator strength of this state is small despite itsπminusπ nature The transition moment is canceled because themain configuration includes not only the excitation from theHOMO to the next-LUMO in dG but also the excitation fromthe next-HOMO to the LUMO in dC and the electron transfertype excitation from the HOMO of dG to the LUMO and next-LUMO of dCThe main peak of the experimental UV spectrum of Z-DNA

is composed of 31A (484 eV) and 51A (493 eV) states havinglarge oscillator strengths (Figure 6a) The 31A state (so blue

color) of the Z-Hbond model is assigned to the 31A state (507eV) of dC which reflects the strong excitation from the next-HOMO to the LUMO of dC The 51A state (red color) of theZ-Hbond model is assigned to the 11A state (441 eV) of dGwhich is the excitation from the HOMO to the LUMO in dGThe excitation energy increases by 052 eV because theHOMOminusLUMO gap becomes large which is the opposite tothat of the 21A state but again due to the hydrogen-bondingeffectFor B-DNA however the SAC-CI UV spectrum of the

hydrogen-bonding model (B-Hbond model Figure 6c) also hasboth main and shoulder peaks in which the main peak is higherand the shoulder peak is lower as compared with theexperimental UV spectrum of B-DNA The main peak (41Aand 51A states) of the SAC-CI spectra of the B-Hbond model ishigher than that of the experimental spectra similarly to theSAC-CI spectrum of the Z-Hbond model The 31A state of dGin B-DNA is also higher than that in Z-DNA (Table 3) Theseresults may be related to the fact that dG is more flexible thandC because the conformation of dG is anti for B-DNA but synfor Z-DNA Since the stacking interaction of B-DNA is weakerthan that of Z-DNA the X-ray structure of B-DNA may bemore flexible than that of Z-DNA However the result that theoscillator strength of the lowest excited state of the B-Hbondmodel is weaker than that of the Z-Hbond model is reflected bythe fact that the shoulder peak of the experimental UV spectrais stronger in Z-DNA but weaker in B-DNAFor the SAC-CI UV spectra of the stacking models of both

B- (B-Stack model) and Z-DNA (Z-Stack model) the firstexcited states originated from dC were calculated at the lower-energy region (higher wavelength region) but the excitationenergy of the main peak is almost the same between the B-Hbond and B-Stack models or between the Z-Hbond and Z-Stack models If there is no hydrogen-bonding interaction inDNA the lowest excited state would be observed in the lower-energy region of the experimental spectra separately from themain peak The excitation energy of the lowest excited state ofthe stacking model is lower than that of the hydrogen-bondingmodel by 044 eV for Z-DNA and 033 eV for B-DNArespectively This result may correspond to the fact that thelowest excitation energy of cytosine differs between the gasphase (428 eV)56 and in solution (464 eV)57

For both B- and Z-DNA the SAC-CI results of thehydrogen-bonding models (B-Hbond and Z-Hbond) are inbetter agreement with the experimental spectra than those ofthe stacking models (B-Stack and Z-Stack) Thus thehydrogen-bonding interaction is reflected in the UV spectraof DNA On the other hand the shoulder peak observed for Z-DNA may reflect the stronger shoulder nature of the Z-Stackmodel in Z-DNA However the UV spectra of DNA providelittle information about stacking interactions between nucleicacid bases

342 CD Spectra of Dimer Models Figure 7 shows theSAC-CI CD spectra of both hydrogen-bonding and stackingmodels of B- and Z-DNA compared with the experimental CDspectra In the experimental CD spectra the first peak isnegative at 296 nm and the second peak is positive at 275 nmfor Z-DNA However B-DNA displays peaks that are oppositein sign the first peak is positive at 280 nm and the second peakis negative at 255 nmIt is worth nothing that the SAC-CI spectrum has a negative

peak at 295 nm only for the stacking model of Z-DNA (Z-Stack) but for the other models the SAC-CI spectrum is

Figure 6 SAC-CI UV spectra for the dimer models The SAC-CIspectra (magenta line) of (a) Z-Hbond model (b) Z-Stack model (c)B-Hbond model and (d) B-Stack model are compared with theexperimental UV spectra4 (black line) of Z-DNA (ab) and B-DNA(cd) The red and blue lines represent the intramolecular excitedstates of dG and dC respectively The green lines represent theelectron transfer (ET) excited state from dG to dC

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5549

Table

4SA

C-CIUVandCD

Spectraof

theHydrogen-Bon

ding

andStacking

Dim

erMod

elsof

Z-andB-D

NA

Z-DNA

B-DNA

SAC-CIof

Z-H

bond

exptle

SAC-CIof

B-H

bond

exptle

EEa

oscb

rotc

UV

EEa

oscb

rotc

UV

state

(eV)

(nm)

(au)

(10minus

40cgs)

nature

typed

(eVnm)

(eV)

(nm)

(au)

(10minus

40cgs)

nature

Typed

(eVnm)

11A

439

282

020

3513

πrarr

πdC

(11 A)

428290

437

284

017

1143

πrarr

πdC

(11 A)

21A

453

274

004

884

πrarr

πdG

(31 A)

456

272

003

minus1481

πrarr

πdG

(31 A)

31A

484

256

018

912

πrarr

πdC

(31 A)

493

251

006

583

πrarr

πET

41A

485

255

005

minus1145

πrarr

πET

498

249

021

1767

πrarr

πdC

(3+41A)

51A

493

252

042

1182

πrarr

πdG

(11 A)

484256

518

240

043

minus1390

πrarr

πdG

(11 A)

484256

61A

512

242

000

057

nrarr

πdC

(21 A)

525

236

000

minus1197

nrarr

πdC

(21 A)

71A

523

237

000

minus1384

nrarr

πdG

(21 A)

529

234

000

minus579

nrarr

πdG

(2+51A)

81A

623

199

000

minus034

nrarr

πdG

(2+51A)

SACminusCIof

Z-Stack

exptle

SACminusCIof

B-Stack

exptle

EEa

oscb

rotc

CD

EEa

oscb

rotc

CD

state

(eV)

(nm)

(au)

(10minus

40cgs)

nature

typed

(eVnm)

(eV)

(nm)

(au)

(10minus

40cgs)

nature

typed

(eVnm)

11A

395

314

009

minus4511

πrarr

πdC

(11 A)

419296

404

307

007

4358

πrarr

πdC

(11 A)

443280

21A

421

294

000

minus3349

πrarr

πdG

(11 A)

419296

440

282

006

minus1032

πrarr

πdG

(11 A)

31A

445

279

001

2967

nrarr

πdC

(21 A)

461

269

000

minus503

nrarr

πdC

(21 A)

41A

472

263

000

minus1335

nrarr

πdG

(21 A)

493

251

013

9407

πrarr

πdC

(3+41A)

51A

485

256

019

minus1796

πrarr

πdG

(31 A)

451275

507

245

005

minus15115

nrarr

πdG

(21 A)

486255

61A

499

248

006

4861

πrarr

πET

451275

512

242

014

minus11056

πrarr

πdG

(31 A)

486255

71A

526

236

007

2721

πrarr

πET

+dC

(31 A)

451275

521

238

014

19475

πrarr

πET+dG

(31 A)

81A

565

219

000

minus231

nrarr

πdC

(41 A)

619

200

000

minus805

nrarr

πdG

(41 A)

aExcitatio

nenergybOscillator

strengthcRotatorystrengthddG

representsintram

olecular

excitedstates

ofdG

dCrepresentsintram

olecular

excitedstates

ofdC

and

ETrepresentselectron

transfer

excitedstates

from

dGto

dCeRef

4

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5550

positive at 295 nm Therefore the strong negative sign of theCD spectra at 295 nm indicates that a strong stackinginteraction exists in Z-DNA The SAC-CI CD spectra of bothstacking models (B-Stack and Z-Stack models) more closelyresemble the experimental CD spectra of B- and Z-DNA thanthose of the hydrogen-bonding models (B-Hbond and Z-Hbond models) For the SAC-CI CD spectrum of the Z-Stackmodel the first and second peaks are negative and positiverespectively but they are the opposite of those obtained fromthe Z-Hbond model For the B-Stack and B-Hbond modelshowever the first peak is positive in the SAC-CI CD spectrabut the second and third peaks are negative and positiverespectively for the B-Stack model in opposition to the B-Hbond model Since the stacking models do not have thehydrogen-bonding interactions the excitation energy of the firstpeak is lower than that of the experimental spectrum as notedaboveIn the Z-Stack model the first peak with a negative sign is

assigned to the 11A (395 eV) and 21A (421 eV) states (Table4) The 11A state of the Z-Stack model corresponds to the 11Astate (413 eV) of dC which is the excitation of the πminusπ fromthe HOMO to the LUMO in dC A large oscillator strength isthe origin of the first (shoulder) peak but the excitation energyis lower than the experimental value because there is nohydrogen bond in the Z-Stack model The rotatory strength ofthe 11A state of dC changes the sign from positive to negativeby the stacking interaction with dG Therefore the stackinginteraction between dG and dC accounts for the strongnegative sign of the CD spectra in Z-DNA

The excitation energy of the 21A state (421 eV) is also closeto the first peak of the experimental CD spectrum This state isassigned to the 11A state (437 eV) of dG which is the HOMOto LUMO πminusπ excitation in dG but also includes thecontribution of the excitation from the next-HOMO to LUMOin dC The oscillator strength in the UV spectrum is smallbecause this state differs from the 11A state of dG by theinclusion of the excitation in dC Although the excitationenergy is shifted by the formation of hydrogen bonds betweendG and dC the rotatory strength of a large negative maycontribute to the strong negative sign of the CD spectra in Z-DNA In the 31A state (445 eV) which is the excitation fromnonbonding orbital to the LUMO in dC the rotatory strengthchanges to positive from the negative sign of the monomer dCHowever this state is not the origin of the second peak as willbe noted later in the tetramer model section Thus the rotatorystrength is strongly affected by the stacking interactionThree important excited states are calculated in the higher-

energy region of the second peak The 51A state (485 eV) hasthe largest oscillator strength and is assigned to the 31A (491eV) state of dG which is the excitation from HOMO to next-LUMO This state is assigned to the main peak (484 eV) of theexperimental UV spectrum of Z-DNA The 6 and 71A states(499 and 526 eV) with a large rotatory strength are the ETstate from the HOMO of dG to the LUMO of dC The mainconfiguration of the 71A state also includes the excitation fromthe next-HOMO to the LUMO in dCIn the B-Stack model the first peak with a positive sign is

assigned to the 11A state (404 eV) This state corresponds tothe 11A state (409 eV) of dC which is the excitation from theHOMO to the LUMO The 21A state (440 eV) is assigned tothe 11A state (434 eV) of dG which reflects the excitationfrom the HOMO to the LUMO Since the B-Stack model doesnot have hydrogen bonds the excitation energies are close tothose of dG or dC Both states have the oscillator strength butthe 11A state is positive while the 21A state is slightly negativefor the rotatory strength which is the same as for themonomers dC and dG respectively Therefore the stackinginteraction of B-DNA has little effect on the sign of the CDspectra at 295 nm because the stacking interaction is weaker inB-DNA than in Z-DNAThe second peak with a negative sign is assigned to the 5 and

61A states with a large rotatory strength The 5 and 61A states(507 and 512 eV in Table 4) are assigned to the 2 and 31Astates (493 and 518 eV in Table 3) of dG which are theexcitations from the nonbonding orbital and the HOMO to thenext-LUMO The main configuration is nminusπ for the 51A stateand πminusπ for the 61A state Both states have strong oscillatorand negative rotatory strengths which means that they can beassigned to the main peaksThe first ET excited states (71A state 521 eV) of B-Stack

model are calculated at a higher region than that (61A state499 eV) of the Z-Stack model because the stacking interactionis weaker in B-DNA than in Z-DNAThe SAC-CI results show that both hydrogen-bonding and

stacking interactions affect the UV and CD spectra of DNA Inparticular the excitation energy of the first excited stateincreases by the formation of hydrogen bonds and the stackinginteraction changes the sign of the CD spectrum of the firstexcited state from positive to negative The strong stackinginteraction characteristic of the helical structure of Z-DNA isthe origin of the strong negative sign at around 295 nm in theCD spectra The hydrogen-bonding and stacking interactions

Figure 7 SAC-CI CD spectra for the dimer models The SAC-CIspectra (magenta line) of (a) Z-Hbond model (b) Z-Stack model (c)B-Hbond model and (d) B-Stack model are compared with theexperimental CD spectra4 (black line) of Z-DNA (ab) and B-DNA(cd) The red and blue lines represent the intramolecular excitedstates of dG and dC respectively The green lines represent theelectron transfer (ET) excited state from dG to dC

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5551

also confer other smaller effects on the UV and CD spectra ofDNA In the higher energy region the SAC-CI CD spectra arenot in good agreement with the experimental CD spectrabecause these models have either hydrogen-bonding interactionor stacking interaction but not both Therefore we need toinvestigate a better model that includes both hydrogen bondingand stacking interactions within the same model A minimumsuch model is the tetramer model of Figure 1 on which weexplain in the next section35 Tetramer Model The SAC-CI calculations of the

dimer models were suitable to investigate the effects of thehydrogen-bonding interaction and the stacking interactionseparately For example the SAC-CI UV spectra of thehydrogen-bonding models were in good agreement with theexperimental UV spectra implying the importance of thehydrogen-bonding interaction The SAC-CI CD spectra for thestacking model clarified that the stacking interaction is thesource of the negative CD peak at around 295 nm Howevergenerally speaking the SAC-CI CD spectra were markedlydifferent from the experimental spectra The SAC-CI UVspectra of the stacking models were composed of two peaksdifferent from the experimental UV spectra Thus theinformation that we gained from the dimer models still didnot fully resolve these differences in the spectra acquired by twodifferent processes So we used the tetramer model (Figure 1)which included both hydrogen-bonding and stacking inter-actions within the same modelFigure 8 shows the SAC-CI UV and CD spectra of the

tetramer model of Z-DNA (Z-Tetra model) and B-DNA (B-Tetra model) in comparison with the experimental UV andCD spectra In order to compare the experimental spectra theSAC-CI results were shifted to lower energy by 05 eV Thecomputational conditions were poor as compared with theother SAC-CI calculations In the tetramer calculations the d-polarization functions were not added and the small activespace was used The SAC-CI calculations using the small activespace and without the d-polarization functions gave higherexcitation energies by about 04minus05 eV (see SupportingInformation Table S1) for the hydrogen-bonding dimermodels In particular the excited states corresponding to themain peak of the experimental UV spectra were shifted by 048eV for both the Z-Hbond model (61A state) and the B-Hbondmodel (71A state)Table 5 shows the excitation energies oscillator strengths

rotatory strengths natures and type of the excited states of theSAC-CI results for tetramer modelsThe SAC-CI UV spectrum of the Z-Tetra model is in good

agreement with the experimental UV spectrum of Z-DNA Themain peak as well as the shoulder peak is reproduced by theSAC-CI method The shoulder peak is assigned to the 4 and51A states which is the πminusπ excitation (HOMO to LUMO)in dC and the main peak is assigned to the 101A state which isthe πminusπ (HOMO to next-LUMO) excitation in dGThe SAC-CI UV spectrum of the B-Tetra model is also in a

reasonable agreement with the experimental UV spectrum of B-DNA although the shape of the shoulder is slightly different Inthe B-Tetra model the shoulder peak is assigned to the 21Astate which is the πminusπ (HOMO to LUMO) excitation in dCand the main peak is assigned to the 81A state which is theπminusπ (HOMO to next-LUMO) excitation in dGThe sign of the rotatory strength is the same between the

SAC-CI CD spectrum of the Z-Tetra model and theexperimental CD spectrum of Z-DNA although the negative

rotatory strength at 295 nm is very weak because thepolarization functions were not included in the tetramermodels It is well-known that the basis set must be good for anadequate description of the stacking interaction5859 In thedimer model the lack of polarization d-functions and thesmaller active space changed the intensity and sign of therotatory strength (see Supporting Information Table S1)Since the stacking interaction is small in B-DNA the SAC-CI

CD spectrum of the B-Tetra model is in good agreement withthe experimental CD spectrum of B-DNA and reproduces notonly the sign of the rotatory strength but also the overall shapeof the spectrumIn Z-DNA the 11A state is assigned to the first peak of the

experimental CD spectrum The 11A state is the ET excitationfrom the HOMO of dG to the LUMO of dC through thestacking conformation The 3 and 51A states with a largenegative rotatory strength are the excitation from the HOMOto the LUMO in dG and dC respectively These states alsocontribute to the first negative peak as in the 1 and 21A stateswith a negative rotatory strength in the Z-Stack model The 8and 101A states are assigned to the second peak of the CDspectrum The 101A state which is the origin of the main peakof the UV spectrum and the πminusπ of dC (81A state) alsocontribute to the second peak of the CD spectrumIn B-DNA the first peak of the CD spectrum is assigned to

the 21A state of πminusπ (HOMO to LUMO) in dC which is thesame as the origin of the first peak of the UV spectrum The ETstates are calculated at the higher energy region than the 21A

Figure 8 SAC-CI UV and CD spectra for the tetramer models TheSAC-CI spectra (magenta line) of Z-Tetra model (ab) and B-Tetramodel (cd) as compared with the experimental UV and CD spectra4

(black line) of Z-DNA (ab) and B-DNA (cd) The red and blue linesrepresent the intramolecular excited states of dG and dC respectivelyThe green lines represent the electron transfer (ET) excited state fromdG to dC All excited states of SAC-CI calculations have been shiftedto the lower values by 05 eV

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5552

Table

5SA

C-CIUVandCD

Spectraof

theTetramer

Mod

elsof

Z-andB-D

NA

Z-DNA

B-DNA

SACminusCIof

Z-Tetra

exptle

SAC-CIof

B-Tetra

exptle

EEa

Oscb

Rotc

UV

CD

EEa

Oscb

Rotc

UV

CD

state

(eV)

(nm)

(au)

(10minus

40cgs)

nature

typed

(eVnm)

(eVnm)

(eV)

(nm)

(au)

(10minus

40cgs)

nature

typed

(eVnm)

(eVnm)

11A

470

264

000

minus663

πrarr

πET

419296

479

259

003

minus041

πrarr

πdG

21A

481

258

002

6611

πrarr

πET

482

257

033

16891

πrarr

πdC

443280

31A

484

256

004

minus5326

πrarr

πdG

419296

484

256

001

minus7454

πrarr

πdC

41A

488

254

015

4553

πrarr

πdC

428290

498

249

007

minus130

πrarr

πET

51A

492

252

012

minus11712

πrarr

πdC

428290

419296

511

243

004

minus622

πrarr

πET

61A

502

247

003

minus3034

πrarr

πET

525

236

000

minus046

πrarr

πET

71A

513

242

001

2737

nrarr

πdG

527

235

001

332

πrarr

πET

81A

515

241

006

14772

πrarr

πdC

451275

545

227

045

minus15609

πrarr

πdG

484256

486255

91A

522

238

001

1978

nrarr

πdG

552

225

002

minus1795

nrarr

πdG

101 A

538

230

071

7299

πrarr

πdG

484256

451275

556

223

013

5929

πrarr

πdG

111 A

541

229

001

minus6858

πrarr

πdG

558

222

009

minus525

πrarr

πdG

121 A

548

226

002

minus1519

nrarr

πdC

563

220

008

minus7436

nrarr

πdG

131 A

553

224

001

minus924

nrarr

πdC

609

204

001

1379

nrarr

πdG

141 A

601

206

001

minus4159

nrarr

πdG

626

198

002

minus464

nrarr

πdG

aExcitatio

nenergyA

llexcitedstates

ofSA

C-CIcalculations

have

been

shifted

tothelower

values

by05eV

inFigure

8bOscillator

strengthcRotatorystrengthddG

representsintram

olecular

excited

states

ofdG

dCrepresentsintram

olecular

excitedstates

ofdC

and

ETrepresentselectron

transfer

excitedstates

from

dGto

dCeRef

4

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5553

state because the stacking interaction is weaker in B-DNA thanin Z-DNA (Figure 1) Therefore the first peak of theexperimental CD spectra originates from the ET excitationfrom dG to dC for Z-DNA but corresponds to theintramolecular excitation in dC for B-DNA The second peakwith a negative sign is assigned to the 81A state The 81A statewhich is the πminusπ of dG is the origin of both the main peaks ofthe UV spectrum and the second peak of the CD spectrumThe ET excited states of the tetramer models were calculated

at a lower energy region than those of the dimer models forboth B- and Z-DNA This result may be related to the overlapbetween the nucleic acid bases The overlap exists only in partof the nucleic acid bases in the dimer models but in thetetramer models the overlap covers the hydrogen bonds as wellas the nucleic acid bases The ET excited states are importantfor the CD spectrum of Z-DNA because the lowest excitedstate is the ET type and is negative for rotatory strength but forB-DNA the ET excited states are not important because thesestates are weak in both oscillator and rotatory strengthsThe SAC-CI results of the tetramer models are calculated at

a higher region by about 05 eV because of the small activespace Therefore better SAC-CI calculations are required forcomparison with the experimental UV and CD spectra of DNAHowever these results indicate that the strong stackinginteraction characteristic of the helical structure of Z-DNAchanges the sign of the rotatory strength for the lowest excitedstates of dG and dC and produces the emergence of the ETexcited states

4 CONCLUSIONSDNA chemistry is affected by three important factors (1) theconformation of the monomer (dG) which is related to thedihedral angle between deoxyribose and guanine (2) thehydrogen-bonding interactions and (3) the stacking inter-actions between two nucleic acid bases To verify whether thesethree factors are important for the CD spectra of DNA weexamined four different models using the SAC-CI method(1) dG monomer model We first applied the SAC-CI

method to the dG monomer to examine the effect ofconformation on the CD spectra of dG and from this resultto identify the structure of the dG monomer in solution Ourcalculations showed that the shape and the sign of the CDspectra depend strongly on the dihedral angle between thedeoxyribose and the guanine but that the SAC-CI UV spectraof both anti- and syn-dG are very similar because the UVspectra are not strongly dependent on the conformation Theexperimental CD spectrum agreed with the theoreticalspectrum of dG only for the anti conformation The resultingthermally averaged spectra also support the anti geometry ofdG as a stable conformation in solution However the SAC-CICD spectrum of syn-dG is opposite in sign from that of theexperimental spectrum Thus the SAC-CI CD spectra of dG inboth anti and syn conformations do not explain the negativesign at 295 nm of the CD spectra of the DNA composed of dGand dC(2) Composit model of dG and dC monomers A

comparison of the calculated composite spectra of dG anddC with the experimental spectra of the two-component DNAshowed similarities between B- and Z-DNA because theconformation of dC is always anti in both B- and Z-DNAThe SAC-CI results led us to confirm that that theconformation of dG is not the origin of the negative peak at295 nm in Z-DNA

(3) Dimer model Our examination of the dimer modelpermitted us to investigate the separate importance of thehydrogen-bonding and the stacking interactions in thestructural chemistry of B- and Z-DNA The SAC-CIcalculations showed that the hydrogen-bonding interactionschange the excitation energies and that the stacking interactionsare responsible for the changes in the sign of the CD rotatorystrength The SAC-CI UV spectra of the hydrogen-bondingmodels (B-Hbond and Z-Hbond models) are similar to theexperimental UV spectra in which the main peak as well as theshoulder peak is well reproduced The SAC-CI CD spectra ofthe stacking models (B-Stack and Z-Stack models) have thesame sign as the experimental CD spectra We concluded thatthe stacking interaction is especially important for the CDspectrum of the strongly stacked Z-DNA because the CD signalat 295 nm changed to a negative sign only for the Z-Stackmodel(4) Tetramer model Finally our calculations with the

tetramer model that contain both hydrogen-bonding andstacking interactions revealed that the shape of the SAC-CIUV and CD spectra compared well with the experimental onesThe ET excitations as well as the lowest intramolecularexcitation of dC and dG are the origins of the negative sign ofthe experimental CD spectrum of Z-DNA The ET excited stateis calculated at a lower energy region than the intramolecularexcited states of dG or dC for the strongly stacked Z-DNA butat a higher energy region for the weakly stacked B-DNA Thenegative sign of the first peak of the experimental CD spectra ofZ-DNA is due to the strong stacking effect characteristic of itshelical structureThe dynamic behavior of the dG monomer in solution was

partially considered by taking the Boltzmann averaging of thecalculated spectra over the various conformations Fortunatelythe Boltzmann averaged SAC-CI CD spectrum was in goodagreement with the experimental spectra although the SAC-CICD spectra largely depend on the conformation of dGHowever in this study we used only the static model for dimerand tetramer models The SAC-CI CD results of the dimer andtetramer models explained the difference between theexperimental CD spectra of B- and Z-DNA However sincethe structure of DNA would also be flexible in solution it isnecessary to take into account the dynamic behavior of DNAThis is an important and interesting topic

ASSOCIATED CONTENT

S Supporting InformationSAC-CI results of the hydrogen-bonding and stacking dimermodels of Z- and B-DNA using small active space and withoutthe polarization d-functions This material is available free ofcharge via the Internet at httppubsacsorg

AUTHOR INFORMATION

Corresponding AuthorPhone +81-75-634-3211 Fax +81-75-634-3211 E-mail hnakatsujiqcriorjp

NotesThe authors declare no competing financial interest

ACKNOWLEDGMENTS

The computations were performed at the Research Center forComputational Science Okazaki Japan

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5554

REFERENCES(1) Voet D Voet J G Eds Biochemistry John Wiley amp Sons IncNew York 1995(2) Ghosh A Bansal M Acta Crystallogr 2003 D59 620minus626(3) Rich A Nordheim A Wang A H-J Annu Rev Biochem 198453 791minus846(4) Tran-Dinh S Taboury J Neumann J-M Huynh-Dinh TGenissel B Langlois drsquoEstaintot B Igolen J Biochemistry 1984 231362minus1371(5) Sugiyama H Kawai K Matsunaga A Fujimoto K Saito IRobinson H Wang A H-J Nucleic Acids Res 1996 24 1272minus1278(6) Kawai K Saito I Sugiyama H J Am Chem Soc 1999 1211391minus1392(7) Oyoshi T Kawai K Sugiyama H J Am Chem Soc 2003 1251526minus1531(8) Xu Y Ikeda R Sugiyama H J Am Chem Soc 2003 12513519minus13524(9) Berova N Nakanishi K Woody R W Eds Circular DichroismPrinciples and Applications Wiley-VCH New York 2000(10) Bauernschmitt R Ahlrichs R Chem Phys Lett 1996 256454minus464(11) Casida M E Jamorski C Casida K C Salahub D R JChem Phys 1998 108 4439minus4449(12) Tsolakidis A Kaxiras E J Phys Chem A 2005 109 2373minus2380(13) Varsano D Di Felice R Marques M A L Rubio A J PhysChem B 2006 110 7129minus7138(14) Nielsen L M Holm A I S Varsano D Kadhane UHoffmann S V Di Felice R Rubio A Nielsen S B J Phys ChemB 2009 113 9614minus9619(15) Tonzani S Schatz G C J Am Chem Soc 2008 130 7607minus7612(16) Casida M E Salahub D R J Chem Phys 2000 113 8918minus8935(17) Dreuw A Weisman J L Head-Gordon M J Chem Phys2003 119 2943minus2946(18) Maitra N T Zhang F Cave R J Burke K J Chem Phys2004 120 5932minus5937(19) McCann D M Stephens P J J Org Chem 2006 71 6074minus6098(20) Casida M E J Mol Struct 2009 914 3minus18(21) Niehaus T A J Mol Struct 2009 914 38minus49(22) Tawada Y Tsuneda T Yanagisawa S Yanai T Hirao K JChem Phys 2004 120 8425minus8433(23) Chiba M Tsuneda T Hirao K J Chem Phys 2006 124144106-1-11(24) Nakatsuji H Hirao K J Chem Phys 1978 68 2053minus2065(25) Nakatsuji H Chem Phys Lett 1978 59 362minus364(26) Nakatsuji H Chem Phys Lett 1979 67 329minus333(27) Nakatsuji H Chem Phys Lett 1979 67 334minus342(28) Nakatsuji H Bull Chem Soc Jpn 2005 78 1705minus1724(29) Ehara M Hasegawa J Nakatsuji H SAC-CI Method Appliedto Molecular Spectroscopy In Theory and Applications of Computa-tional Chemistry The First 40 Years A Vol of Technical and HistoricalPerspectives Dykstra C E Frenking G Kim K S Scuseria G EEds Elsevier Oxford UK 2005 pp 1099minus1141(30) Miyahara T Hasegawa J Nakatsuji H Bull Chem Soc Jpn2009 82 1215minus1226(31) Bureekaew S Hasegawa J Nakatsuji H Chem Phys Lett2006 425 367minus371(32) Nakatani N Hasegawa J Nakatsuji H J Am Chem Soc2007 129 8756minus8765(33) Fujimoto K Hasegawa J Nakatsuji H Bull Chem Soc Jpn2009 82 1140minus1148(34) Miyahara T Nakatsuji H Collect Czech Chem Commun2011 76 537minus552(35) Sugiyama H Unpublished(36) Geertsen J Rittby M Bartlett R J Chem Phys Lett 1989164 57minus62

(37) Stanton J F Bartlett R J J Chem Phys 1993 98 7029minus7039(38) Ehara M Ishida M Toyota K Nakatsuji H SAC-CIGeneral-R Method Theory and Applications to the Multi-ElectronProcesses In Reviews in Modern Quantum Chemistry (A Tribute toProfessor Robert G Parr) Sen K D Ed World Scientific Singapore2003 pp 293minus319(39) SAC-CI GUIDE httpwwwqcriorjpsacci(40) Frisch M J Trucks G W Schlegel H B et al Gaussian 03development revision revision E05 Gaussian Inc Wallingford CT2004(41) Hohenberg P Kohn W Phys Rev 1964 136 B864minusB871(42) Kohn W Sham L J Phys Rev 1965 140 A1133minusA1138(43) Salahub D R Zerner M C Eds The Challenge of d and fElectrons The American Chemical Society Washington DC 1989(44) Parr R G Yang W Density-Functional Theory of Atoms andMolecules Oxford University Press Oxford UK 1989(45) Becke A D J Chem Phys 1993 98 5648minus5652(46) Lee C Yang W Parr R G Phys Rev 1988 37 785minus789(47) Hariharan P C Pople J A Theor Chim Acta 1973 28 213minus222(48) Francl M M Pietro W J Hehre W J Binkley J SDeFrees D J Pople J A Gordon M S J Chem Phys 1982 773654minus3665(49) Tomasi J Mennucci B Cammi R Chem Rev 2005 1052999minus3093(50) Fukuda R Ehara M Nakatsuji H Cammi R J Chem Phys2011 134 104109minus1minus11(51) Dunning T H Jr Hay P J In Modern Theoretical ChemistrySchaefer H F III Ed Plenum New York 1976 Vol 3 pp 1minus28(52) Nakatsuji H Chem Phys 1983 75 425minus441(53) Hocquet A Leulliot N Ghomi M J Phys Chem B 2000 1044560minus4568(54) Mishra S K Mishra P C J Comput Chem 2002 23 530minus540(55) Sprecher C A Johnson W C Biopolymers 1977 16 2243minus2264(56) Clark L B Peschel G G Tinoc I Jr J Phys Chem 1965 693615minus3618(57) Voet D Gratzer W B Cox R A Doty P Biopolymers 19631 193minus208(58) Hobza P Mehlhorn A Carsky P Zahradnik R J Mol Struct1986 138 387minus399(59) Hobza P Sponer J Chem Rev 1999 99 3247minus3276

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5555

Page 5: Helical Structure and Circular Dichroism Spectra of DNA: A ... · B-DNA, have right-handed double-helical structure, while Z-DNA is left-handed and appears only when DNA has a special

both the 21A and 31A states are large and opposite in sign(Table 1) The 21A state is calculated to lie at 508 eV (244nm) for anti-dG and at 503 eV (247 nm) for syn-dG The 31Astate is calculated to lie at 510 eV (243 nm) for anti-dG and at506 eV (245 nm) for syn-dG The difference between 21A and31A in the excitation energy is very small so that their rotatorystrengths cancel each other giving relatively small rotatorystrengths From Table 1 we see that the total rotatory strengthof the second band is negative for anti-dG and positive for syn-dG To reproduce the experimental CD spectra the theoreticalrotatory strength must be reliable since otherwise reliablepositive or negative CD peaks cannot be obtained The SAC-CItheory seems to give reliable rotatory intensity since thetheoretical CD spectra agree well with the experimental one Inthe third band region (214 nm) the sum of the rotatorystrengths is positive for both anti- and syn-dG although thestrongest rotatory strength comes from the 41A (n rarr π) statefor anti-dG but from the 51A (π rarr π) state for syn-dG As inthe experimental CD spectrum the third band has largerintensity than the second band in the SAC-CI CD spectrum ofanti-dG However for syn-dG the third band in the SAC-CI

CD spectrum is smaller than the second band The 71A stateassigned to the fourth band with a strong negative intensity at190 nm in the experimental CD spectrum of dGMP55 isnegative for anti-dG and almost zero for syn-dG

33 Composite CD Spectra of dG and dC Comparedwith the CD Spectra of DNA We then extended our studiesto the spectra of B- and Z-DNA and we compared the SAC-CICD spectra of anti- and syn-dG (Figure 4) with theexperimental CD spectra of B- and Z-DNA (Figure 5)4 Thesecond band at 250 nm in the SAC-CI CD spectrum of anti-dGis similar to the experimental CD spectrum of B-DNA whilethe SAC-CI CD spectrum of syn-dG is similar to theexperimental CD spectrum of Z-DNA This result may indicatethat in DNA the second band at 250 nm reflects the dihedralangle φ of dG between the deoxyribose and the guanineTherefore we calculated the composite SAC-CI CD spectra ofthe noninteracting dG and dC monomer from the available X-ray crystallographic structures in B- and Z-DNA to clarify if thedifferent conformations of dG andor dC in different types ofDNA explain the differences in the CD spectra

Figure 4 CD spectra of dG at several conformation angles φ The experimental CD spectrum35 (black line) of dG is compared with the SAC-CI CDspectra (red lines) of (a) φ = 10deg (b) φ = 30deg (c) φ = 50deg (d) φ = 70deg (e) φ = 90deg (f) φ = 110deg (g) φ = 1295deg (syn) (h) φ = 150deg (i) φ = 170deg(j) φ = 190deg (k) φ = 210deg (l) φ = 230deg (m) φ = 250deg (n) φ = 270deg (o) φ = 290deg (p) φ = 3103deg (anti) (q) φ = 330deg and (r) φ = 350deg TheSAC-CI CD Boltzmann averaged spectra of (g) syn (blue line) (p) anti (blue line) and (p) all (green line) are also shown

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5546

The SAC-CI CD spectra of dG and dC alone (Figure5abde) having the structures of those in B- and Z-DNA andthe composite CD spectra (Figure 5cf) obtained as (a + b andd + e) were compared to the experimental CD spectra of B-and Z-DNA The SAC-CI excitation energies oscillatorstrengths rotatory strengths and natures of the excited statesfor the monomer model (dG and dC) were also determined(Table 3)The conformations of dG and dC are the same as those in

the ideal B-DNA or Z-DNA in which dG and dC are arrangedalternately (Figure 1) However the X-ray structure wasobtained for the DNA composed of only 6 sequences in thecase of Z-DNA Since the geometry is affected by the effect ofthe edge of DNA the conformation is different for eachmonomer Since the conformation of dG is in particular moreflexible than that of dC in DNA we calculated two kinds of dG(model I and II) in B- and Z-DNA respectively The models Iand II correspond to dG of the hydrogen-bonding models (B-Hbond and Z-Hbond models) and the stacking models (B-Stack and Z-Stack models) of the dimer models respectivelyThe dihedral angle (φ) between the deoxyribose and theguanine is 2763deg and 3007deg respectively for models I and IIof B-DNA and 1113deg and 1175deg respectively for models Iand II of Z-DNA The dihedral angle between the deoxyriboseand cytosine is 3232deg for B-DNA and 3329deg for Z-DNA Therotatory strength of dG is very different between models I andII for both B- and Z-DNA as shown in Table 3 Therefore fordG we prepared an averaged CD spectra for models I and II(Figure 5ad) For Z-DNA the SAC-CI CD spectrum of dG(Figure 5a) is negative for the first band and positive for thesecond band as in the experimental spectrum of Z-DNAalthough the excitation energies are calculated at a higherregion However the SAC-CI result is opposite in sign to theexperimental CD spectrum for the first band of B-DNA (Figure5d) These results show that in the CD spectra of DNA the

conformation of dG (monomer) is related to the sign of thesecond band but has no effect on the sign of the first bandThe CD spectra of dC have four peaks (Figure 5be) The

first peak represents the πminusπ of HOMO to LUMO Thesecond peak is the excitation from the nonbonding orbital tothe LUMO The third and fourth peaks are composed of boththe πminusπ and nminusπ from next-HOMO and nonbondingorbitals to LUMO The third and fourth peaks are mainly πminusπand nminusπ respectively The SAC-CI CD spectra of dC aresimilar in B- and Z-DNA with two positive and two negativepeaks The rotatory strength is positive for the first and thirdpeaks and negative for the second and fourth peaks Thereforethe CD spectra of dC are always positive at around 295 nmcorresponding to the conformation of dC which is always antiin DNAAs shown in Figure 5cf the composite SAC-CI CD spectra

obtained from dG and dC alone are similar between B- and Z-DNA The sign at 295 nm is positive for both B- and Z-DNAbecause the rotatory strength of dC which is the lowest peak ismuch stronger than that of dG Therefore the composite CD

Table 2 Boltzmann Distribution of dG along theConformational Angle φ

angle (φ) all ()a anti ()b syn ()c

10 086 1130 01450 00470 01090 170 91110 279 1431295 (syn) 1131 517150 454 224170 041 25190 008210 000230 010250 325 41270 1534 195290 1700 2163103 (anti) 2669 339330 1307 166350 258 33total 1000 1000 1000

aldquoallrdquo gives the ratio of each conformer when averaging included all 18conformers bldquoantirdquo gives the ratio of each conformer when only sevenconformations around anti are used cldquosynrdquo gives the ratio of eachconformer when only five conformations around syn are used

Figure 5 SAC-CI CD spectra of monomer ((a) dG in Z-DNA (b) dCin Z-DNA (d) dG in B-DNA and (e) dC in B-DNA) and thecomposite CD spectra of (c) Z-DNA and (f) B-DNA obtained fromthe SAC-CI CD spectra of dG and dC compared with theexperimental CD spectra4 (black lines) of Z-DNA (aminusc) and B-DNA (dminusf) The red and blue lines represent the excited state of dGand dC respectively The magenta line in panels c and f represents thecomposite spectra

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5547

spectrum is similar to that of dC for both B- and Z-DNA As aresult in the composite CD spectra of B- and Z-DNA the signat around 295 nm is independent of the geometry of themonomer (dG or dC) even though the CD spectrum of dGdepends on the dihedral angle between the guanine and thedeoxyribose because the lowest peak originates from dCwhose geometry is always anti in both B- and Z-DNA Thisindicates that the noninteracting dG and dC cannot explain theexperimental spectra of DNA and that the hydrogen-bondingor stacking interactions with the neighboring nucleic acid basesmust be important for the CD spectra of DNA34 UV and CD Spectra of Dimer Models In the studies

of the structure spectroscopy and functions of DNA it isgenerally recognized that the hydrogen-bonding interaction andthe stacking interaction are the two most important interactionsthat characterize the properties of DNA So we expect thatthese interactions might also be the important factors that affectthe features of the UV and CD spectra of DNA To investigatethe possible origin we considered the dimer model composedof dG and dC These models contained either the hydrogen-bonding interaction or the stacking interaction (Figure 1)However we note that the SAC-CI results using these modelsneed not be in agreement with the experimental spectrabecause these models lack either the stacking interaction or thehydrogen-bonding interaction We must use the tetramer

model at least to compare with the experimental spectra asnoted below

341 UV Spectra of Dimer Models Figure 6 shows theSAC-CI UV spectra of both hydrogen-bonding and stackingmodels of both B- and Z-DNA compared with the experimentalUV spectrum The experimental UV spectra of Z-DNA displaya main peak at 256 nm and a shoulder peak at around 290 nmThe experimental UV spectrum of B-DNA is very similar butthe shoulder peak at around 290 nm is weaker Table 4 showsthe excitation energies oscillator strengths rotatory strengthsnatures and the types of the excited states of the SAC-CIresults for both hydrogen-bonding and stacking models of thedimer When the excitation in the dimer model is not withindG or dC but corresponds to the electron transfer (ET)between dG and dC we assigned it as the ET typeThe SAC-CI UV spectrum of the hydrogen-bonding model

of Z-DNA (Z-Hbond model) is in good agreement with theexperimental UV spectrum of Z-DNA (Figure 6a) Threestrong peaks of the πminusπ nature are calculated The loweststrong peak corresponds to the shoulder peak at 428 eV (290nm) of the experimental UV spectrum of Z-DNA Theremaining two strong peaks correspond to the main peak at484 eV (256 nm) of the experimental UV spectrum of Z-DNARed blue and green colors represent the excited states of dGdC and ET respectively

Table 3 SAC-CI UV and CD Spectra of dG and dC in Z- and B-DNA

Z-DNA B-DNA

dG (model I φ = 1113deg) dG (model I φ = 2763deg)

EEa oscb rotc EEa oscb rotc

state (eV) (nm) (au) (10minus40 cgs) nature (eV) (nm) (au) (10minus40 cgs) nature

11A 441 281 017 125 π rarr π 443 280 016 minus1035 π rarr π21A 474 262 000 minus1099 n rarr π 488 254 000 1099 n rarr π31A 493 251 023 2066 π rarr π 513 242 025 minus3355 π rarr π41A 578 214 000 minus864 n rarr π 595 208 002 minus2619 π rarr π51A 591 210 004 minus656 π rarr σ 599 207 001 3571 n rarr π61A 611 203 000 minus438 n rarr π 618 201 000 983 n rarr π71A 628 197 000 359 π rarr σ 675 184 008 1235 π rarr σ81A 659 188 030 634 π rarr π 678 183 028 minus2402 π rarr π

dG (model II φ = 1175deg) dG (model II φ = 3007deg)

EEa oscb rotc EEa oscb rotc

state (eV) (nm) (au) (10minus40 cgs) nature (eV) (nm) (au) (10minus40 cgs) nature

11A 437 284 016 minus451 π rarr π 434 286 016 minus247 π rarr π21A 459 270 000 minus322 n rarr π 493 251 000 minus1467 n rarr π31A 491 252 022 minus319 π rarr π 518 239 021 minus200 π rarr π41A 581 213 003 323 π rarr π 595 209 001 2884 n rarr π51A 598 207 000 minus738 π rarr π 599 207 003 minus647 π rarr π61A 621 200 000 796 n rarr π 617 201 000 minus3033 n rarr π71A 641 193 000 353 π rarr σ 660 188 048 978 π rarr π81A 647 191 031 minus077 π rarr π 666 186 001 326 π rarr σ

dC (φ = 3329deg) dC (φ = 3232deg)

EEa oscb rotc EEa oscb rotc

state (eV) (nm) (au) (10minus40 cgs) nature (eV) (nm) (au) (10minus40 cgs) nature

11A 413 300 009 2385 π rarr π 409 303 009 1403 π rarr π21A 442 280 000 minus1548 n rarr π 452 274 000 minus915 n rarr π31A 507 245 011 8392 π + n rarr π 499 249 007 5975 π + n rarr π41A 527 235 011 minus8639 n + π rarr π 527 235 015 minus8017 π + n rarr π51A 573 216 000 minus1385 n rarr π 559 222 001 393 n rarr π61A 632 196 039 2349 π rarr π 625 198 032 1098 π rarr π

aExcitation energy bOscillator strength cRotatory strength

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5548

The shoulder peak in the experimental UV spectrum of Z-DNA is composed of 11A (439 eV) and 21A (453 eV) statesThe 11A state (in blue color) of the Z-Hbond modelcorresponds to the 11A state of dC at 413 eV which is theexcitation from the HOMO to the LUMO of dC This state haslarge oscillator strength of the πminusπ nature In the Z-Hbondmodel this state is blue-shifted by 026 eV because theHOMOminusLUMO gap (0449 au) of the Z-Hbond model islarger than that (0438 au) of dC This state is in goodagreement with the first (shoulder) peak of the experimentalUV spectrum of Z-DNA However the 21A state (453 eV redcolor) of the Z-Hbond model is assigned to the 31A state (493eV) of dG which is the excitation from the HOMO to the next-LUMO of dG This large shift of the excitation energy is causedby the MO-order change The MO of the Z-Hbond modelcorresponding to the next-LUMO of dG becomes lower thanthe MO of the Z-Hbond model corresponding to the LUMOof dG by the formation of hydrogen bonds with cytosine in theZ-Hbond model Therefore this state is red-shifted by 04 eVHowever the oscillator strength of this state is small despite itsπminusπ nature The transition moment is canceled because themain configuration includes not only the excitation from theHOMO to the next-LUMO in dG but also the excitation fromthe next-HOMO to the LUMO in dC and the electron transfertype excitation from the HOMO of dG to the LUMO and next-LUMO of dCThe main peak of the experimental UV spectrum of Z-DNA

is composed of 31A (484 eV) and 51A (493 eV) states havinglarge oscillator strengths (Figure 6a) The 31A state (so blue

color) of the Z-Hbond model is assigned to the 31A state (507eV) of dC which reflects the strong excitation from the next-HOMO to the LUMO of dC The 51A state (red color) of theZ-Hbond model is assigned to the 11A state (441 eV) of dGwhich is the excitation from the HOMO to the LUMO in dGThe excitation energy increases by 052 eV because theHOMOminusLUMO gap becomes large which is the opposite tothat of the 21A state but again due to the hydrogen-bondingeffectFor B-DNA however the SAC-CI UV spectrum of the

hydrogen-bonding model (B-Hbond model Figure 6c) also hasboth main and shoulder peaks in which the main peak is higherand the shoulder peak is lower as compared with theexperimental UV spectrum of B-DNA The main peak (41Aand 51A states) of the SAC-CI spectra of the B-Hbond model ishigher than that of the experimental spectra similarly to theSAC-CI spectrum of the Z-Hbond model The 31A state of dGin B-DNA is also higher than that in Z-DNA (Table 3) Theseresults may be related to the fact that dG is more flexible thandC because the conformation of dG is anti for B-DNA but synfor Z-DNA Since the stacking interaction of B-DNA is weakerthan that of Z-DNA the X-ray structure of B-DNA may bemore flexible than that of Z-DNA However the result that theoscillator strength of the lowest excited state of the B-Hbondmodel is weaker than that of the Z-Hbond model is reflected bythe fact that the shoulder peak of the experimental UV spectrais stronger in Z-DNA but weaker in B-DNAFor the SAC-CI UV spectra of the stacking models of both

B- (B-Stack model) and Z-DNA (Z-Stack model) the firstexcited states originated from dC were calculated at the lower-energy region (higher wavelength region) but the excitationenergy of the main peak is almost the same between the B-Hbond and B-Stack models or between the Z-Hbond and Z-Stack models If there is no hydrogen-bonding interaction inDNA the lowest excited state would be observed in the lower-energy region of the experimental spectra separately from themain peak The excitation energy of the lowest excited state ofthe stacking model is lower than that of the hydrogen-bondingmodel by 044 eV for Z-DNA and 033 eV for B-DNArespectively This result may correspond to the fact that thelowest excitation energy of cytosine differs between the gasphase (428 eV)56 and in solution (464 eV)57

For both B- and Z-DNA the SAC-CI results of thehydrogen-bonding models (B-Hbond and Z-Hbond) are inbetter agreement with the experimental spectra than those ofthe stacking models (B-Stack and Z-Stack) Thus thehydrogen-bonding interaction is reflected in the UV spectraof DNA On the other hand the shoulder peak observed for Z-DNA may reflect the stronger shoulder nature of the Z-Stackmodel in Z-DNA However the UV spectra of DNA providelittle information about stacking interactions between nucleicacid bases

342 CD Spectra of Dimer Models Figure 7 shows theSAC-CI CD spectra of both hydrogen-bonding and stackingmodels of B- and Z-DNA compared with the experimental CDspectra In the experimental CD spectra the first peak isnegative at 296 nm and the second peak is positive at 275 nmfor Z-DNA However B-DNA displays peaks that are oppositein sign the first peak is positive at 280 nm and the second peakis negative at 255 nmIt is worth nothing that the SAC-CI spectrum has a negative

peak at 295 nm only for the stacking model of Z-DNA (Z-Stack) but for the other models the SAC-CI spectrum is

Figure 6 SAC-CI UV spectra for the dimer models The SAC-CIspectra (magenta line) of (a) Z-Hbond model (b) Z-Stack model (c)B-Hbond model and (d) B-Stack model are compared with theexperimental UV spectra4 (black line) of Z-DNA (ab) and B-DNA(cd) The red and blue lines represent the intramolecular excitedstates of dG and dC respectively The green lines represent theelectron transfer (ET) excited state from dG to dC

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5549

Table

4SA

C-CIUVandCD

Spectraof

theHydrogen-Bon

ding

andStacking

Dim

erMod

elsof

Z-andB-D

NA

Z-DNA

B-DNA

SAC-CIof

Z-H

bond

exptle

SAC-CIof

B-H

bond

exptle

EEa

oscb

rotc

UV

EEa

oscb

rotc

UV

state

(eV)

(nm)

(au)

(10minus

40cgs)

nature

typed

(eVnm)

(eV)

(nm)

(au)

(10minus

40cgs)

nature

Typed

(eVnm)

11A

439

282

020

3513

πrarr

πdC

(11 A)

428290

437

284

017

1143

πrarr

πdC

(11 A)

21A

453

274

004

884

πrarr

πdG

(31 A)

456

272

003

minus1481

πrarr

πdG

(31 A)

31A

484

256

018

912

πrarr

πdC

(31 A)

493

251

006

583

πrarr

πET

41A

485

255

005

minus1145

πrarr

πET

498

249

021

1767

πrarr

πdC

(3+41A)

51A

493

252

042

1182

πrarr

πdG

(11 A)

484256

518

240

043

minus1390

πrarr

πdG

(11 A)

484256

61A

512

242

000

057

nrarr

πdC

(21 A)

525

236

000

minus1197

nrarr

πdC

(21 A)

71A

523

237

000

minus1384

nrarr

πdG

(21 A)

529

234

000

minus579

nrarr

πdG

(2+51A)

81A

623

199

000

minus034

nrarr

πdG

(2+51A)

SACminusCIof

Z-Stack

exptle

SACminusCIof

B-Stack

exptle

EEa

oscb

rotc

CD

EEa

oscb

rotc

CD

state

(eV)

(nm)

(au)

(10minus

40cgs)

nature

typed

(eVnm)

(eV)

(nm)

(au)

(10minus

40cgs)

nature

typed

(eVnm)

11A

395

314

009

minus4511

πrarr

πdC

(11 A)

419296

404

307

007

4358

πrarr

πdC

(11 A)

443280

21A

421

294

000

minus3349

πrarr

πdG

(11 A)

419296

440

282

006

minus1032

πrarr

πdG

(11 A)

31A

445

279

001

2967

nrarr

πdC

(21 A)

461

269

000

minus503

nrarr

πdC

(21 A)

41A

472

263

000

minus1335

nrarr

πdG

(21 A)

493

251

013

9407

πrarr

πdC

(3+41A)

51A

485

256

019

minus1796

πrarr

πdG

(31 A)

451275

507

245

005

minus15115

nrarr

πdG

(21 A)

486255

61A

499

248

006

4861

πrarr

πET

451275

512

242

014

minus11056

πrarr

πdG

(31 A)

486255

71A

526

236

007

2721

πrarr

πET

+dC

(31 A)

451275

521

238

014

19475

πrarr

πET+dG

(31 A)

81A

565

219

000

minus231

nrarr

πdC

(41 A)

619

200

000

minus805

nrarr

πdG

(41 A)

aExcitatio

nenergybOscillator

strengthcRotatorystrengthddG

representsintram

olecular

excitedstates

ofdG

dCrepresentsintram

olecular

excitedstates

ofdC

and

ETrepresentselectron

transfer

excitedstates

from

dGto

dCeRef

4

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5550

positive at 295 nm Therefore the strong negative sign of theCD spectra at 295 nm indicates that a strong stackinginteraction exists in Z-DNA The SAC-CI CD spectra of bothstacking models (B-Stack and Z-Stack models) more closelyresemble the experimental CD spectra of B- and Z-DNA thanthose of the hydrogen-bonding models (B-Hbond and Z-Hbond models) For the SAC-CI CD spectrum of the Z-Stackmodel the first and second peaks are negative and positiverespectively but they are the opposite of those obtained fromthe Z-Hbond model For the B-Stack and B-Hbond modelshowever the first peak is positive in the SAC-CI CD spectrabut the second and third peaks are negative and positiverespectively for the B-Stack model in opposition to the B-Hbond model Since the stacking models do not have thehydrogen-bonding interactions the excitation energy of the firstpeak is lower than that of the experimental spectrum as notedaboveIn the Z-Stack model the first peak with a negative sign is

assigned to the 11A (395 eV) and 21A (421 eV) states (Table4) The 11A state of the Z-Stack model corresponds to the 11Astate (413 eV) of dC which is the excitation of the πminusπ fromthe HOMO to the LUMO in dC A large oscillator strength isthe origin of the first (shoulder) peak but the excitation energyis lower than the experimental value because there is nohydrogen bond in the Z-Stack model The rotatory strength ofthe 11A state of dC changes the sign from positive to negativeby the stacking interaction with dG Therefore the stackinginteraction between dG and dC accounts for the strongnegative sign of the CD spectra in Z-DNA

The excitation energy of the 21A state (421 eV) is also closeto the first peak of the experimental CD spectrum This state isassigned to the 11A state (437 eV) of dG which is the HOMOto LUMO πminusπ excitation in dG but also includes thecontribution of the excitation from the next-HOMO to LUMOin dC The oscillator strength in the UV spectrum is smallbecause this state differs from the 11A state of dG by theinclusion of the excitation in dC Although the excitationenergy is shifted by the formation of hydrogen bonds betweendG and dC the rotatory strength of a large negative maycontribute to the strong negative sign of the CD spectra in Z-DNA In the 31A state (445 eV) which is the excitation fromnonbonding orbital to the LUMO in dC the rotatory strengthchanges to positive from the negative sign of the monomer dCHowever this state is not the origin of the second peak as willbe noted later in the tetramer model section Thus the rotatorystrength is strongly affected by the stacking interactionThree important excited states are calculated in the higher-

energy region of the second peak The 51A state (485 eV) hasthe largest oscillator strength and is assigned to the 31A (491eV) state of dG which is the excitation from HOMO to next-LUMO This state is assigned to the main peak (484 eV) of theexperimental UV spectrum of Z-DNA The 6 and 71A states(499 and 526 eV) with a large rotatory strength are the ETstate from the HOMO of dG to the LUMO of dC The mainconfiguration of the 71A state also includes the excitation fromthe next-HOMO to the LUMO in dCIn the B-Stack model the first peak with a positive sign is

assigned to the 11A state (404 eV) This state corresponds tothe 11A state (409 eV) of dC which is the excitation from theHOMO to the LUMO The 21A state (440 eV) is assigned tothe 11A state (434 eV) of dG which reflects the excitationfrom the HOMO to the LUMO Since the B-Stack model doesnot have hydrogen bonds the excitation energies are close tothose of dG or dC Both states have the oscillator strength butthe 11A state is positive while the 21A state is slightly negativefor the rotatory strength which is the same as for themonomers dC and dG respectively Therefore the stackinginteraction of B-DNA has little effect on the sign of the CDspectra at 295 nm because the stacking interaction is weaker inB-DNA than in Z-DNAThe second peak with a negative sign is assigned to the 5 and

61A states with a large rotatory strength The 5 and 61A states(507 and 512 eV in Table 4) are assigned to the 2 and 31Astates (493 and 518 eV in Table 3) of dG which are theexcitations from the nonbonding orbital and the HOMO to thenext-LUMO The main configuration is nminusπ for the 51A stateand πminusπ for the 61A state Both states have strong oscillatorand negative rotatory strengths which means that they can beassigned to the main peaksThe first ET excited states (71A state 521 eV) of B-Stack

model are calculated at a higher region than that (61A state499 eV) of the Z-Stack model because the stacking interactionis weaker in B-DNA than in Z-DNAThe SAC-CI results show that both hydrogen-bonding and

stacking interactions affect the UV and CD spectra of DNA Inparticular the excitation energy of the first excited stateincreases by the formation of hydrogen bonds and the stackinginteraction changes the sign of the CD spectrum of the firstexcited state from positive to negative The strong stackinginteraction characteristic of the helical structure of Z-DNA isthe origin of the strong negative sign at around 295 nm in theCD spectra The hydrogen-bonding and stacking interactions

Figure 7 SAC-CI CD spectra for the dimer models The SAC-CIspectra (magenta line) of (a) Z-Hbond model (b) Z-Stack model (c)B-Hbond model and (d) B-Stack model are compared with theexperimental CD spectra4 (black line) of Z-DNA (ab) and B-DNA(cd) The red and blue lines represent the intramolecular excitedstates of dG and dC respectively The green lines represent theelectron transfer (ET) excited state from dG to dC

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5551

also confer other smaller effects on the UV and CD spectra ofDNA In the higher energy region the SAC-CI CD spectra arenot in good agreement with the experimental CD spectrabecause these models have either hydrogen-bonding interactionor stacking interaction but not both Therefore we need toinvestigate a better model that includes both hydrogen bondingand stacking interactions within the same model A minimumsuch model is the tetramer model of Figure 1 on which weexplain in the next section35 Tetramer Model The SAC-CI calculations of the

dimer models were suitable to investigate the effects of thehydrogen-bonding interaction and the stacking interactionseparately For example the SAC-CI UV spectra of thehydrogen-bonding models were in good agreement with theexperimental UV spectra implying the importance of thehydrogen-bonding interaction The SAC-CI CD spectra for thestacking model clarified that the stacking interaction is thesource of the negative CD peak at around 295 nm Howevergenerally speaking the SAC-CI CD spectra were markedlydifferent from the experimental spectra The SAC-CI UVspectra of the stacking models were composed of two peaksdifferent from the experimental UV spectra Thus theinformation that we gained from the dimer models still didnot fully resolve these differences in the spectra acquired by twodifferent processes So we used the tetramer model (Figure 1)which included both hydrogen-bonding and stacking inter-actions within the same modelFigure 8 shows the SAC-CI UV and CD spectra of the

tetramer model of Z-DNA (Z-Tetra model) and B-DNA (B-Tetra model) in comparison with the experimental UV andCD spectra In order to compare the experimental spectra theSAC-CI results were shifted to lower energy by 05 eV Thecomputational conditions were poor as compared with theother SAC-CI calculations In the tetramer calculations the d-polarization functions were not added and the small activespace was used The SAC-CI calculations using the small activespace and without the d-polarization functions gave higherexcitation energies by about 04minus05 eV (see SupportingInformation Table S1) for the hydrogen-bonding dimermodels In particular the excited states corresponding to themain peak of the experimental UV spectra were shifted by 048eV for both the Z-Hbond model (61A state) and the B-Hbondmodel (71A state)Table 5 shows the excitation energies oscillator strengths

rotatory strengths natures and type of the excited states of theSAC-CI results for tetramer modelsThe SAC-CI UV spectrum of the Z-Tetra model is in good

agreement with the experimental UV spectrum of Z-DNA Themain peak as well as the shoulder peak is reproduced by theSAC-CI method The shoulder peak is assigned to the 4 and51A states which is the πminusπ excitation (HOMO to LUMO)in dC and the main peak is assigned to the 101A state which isthe πminusπ (HOMO to next-LUMO) excitation in dGThe SAC-CI UV spectrum of the B-Tetra model is also in a

reasonable agreement with the experimental UV spectrum of B-DNA although the shape of the shoulder is slightly different Inthe B-Tetra model the shoulder peak is assigned to the 21Astate which is the πminusπ (HOMO to LUMO) excitation in dCand the main peak is assigned to the 81A state which is theπminusπ (HOMO to next-LUMO) excitation in dGThe sign of the rotatory strength is the same between the

SAC-CI CD spectrum of the Z-Tetra model and theexperimental CD spectrum of Z-DNA although the negative

rotatory strength at 295 nm is very weak because thepolarization functions were not included in the tetramermodels It is well-known that the basis set must be good for anadequate description of the stacking interaction5859 In thedimer model the lack of polarization d-functions and thesmaller active space changed the intensity and sign of therotatory strength (see Supporting Information Table S1)Since the stacking interaction is small in B-DNA the SAC-CI

CD spectrum of the B-Tetra model is in good agreement withthe experimental CD spectrum of B-DNA and reproduces notonly the sign of the rotatory strength but also the overall shapeof the spectrumIn Z-DNA the 11A state is assigned to the first peak of the

experimental CD spectrum The 11A state is the ET excitationfrom the HOMO of dG to the LUMO of dC through thestacking conformation The 3 and 51A states with a largenegative rotatory strength are the excitation from the HOMOto the LUMO in dG and dC respectively These states alsocontribute to the first negative peak as in the 1 and 21A stateswith a negative rotatory strength in the Z-Stack model The 8and 101A states are assigned to the second peak of the CDspectrum The 101A state which is the origin of the main peakof the UV spectrum and the πminusπ of dC (81A state) alsocontribute to the second peak of the CD spectrumIn B-DNA the first peak of the CD spectrum is assigned to

the 21A state of πminusπ (HOMO to LUMO) in dC which is thesame as the origin of the first peak of the UV spectrum The ETstates are calculated at the higher energy region than the 21A

Figure 8 SAC-CI UV and CD spectra for the tetramer models TheSAC-CI spectra (magenta line) of Z-Tetra model (ab) and B-Tetramodel (cd) as compared with the experimental UV and CD spectra4

(black line) of Z-DNA (ab) and B-DNA (cd) The red and blue linesrepresent the intramolecular excited states of dG and dC respectivelyThe green lines represent the electron transfer (ET) excited state fromdG to dC All excited states of SAC-CI calculations have been shiftedto the lower values by 05 eV

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5552

Table

5SA

C-CIUVandCD

Spectraof

theTetramer

Mod

elsof

Z-andB-D

NA

Z-DNA

B-DNA

SACminusCIof

Z-Tetra

exptle

SAC-CIof

B-Tetra

exptle

EEa

Oscb

Rotc

UV

CD

EEa

Oscb

Rotc

UV

CD

state

(eV)

(nm)

(au)

(10minus

40cgs)

nature

typed

(eVnm)

(eVnm)

(eV)

(nm)

(au)

(10minus

40cgs)

nature

typed

(eVnm)

(eVnm)

11A

470

264

000

minus663

πrarr

πET

419296

479

259

003

minus041

πrarr

πdG

21A

481

258

002

6611

πrarr

πET

482

257

033

16891

πrarr

πdC

443280

31A

484

256

004

minus5326

πrarr

πdG

419296

484

256

001

minus7454

πrarr

πdC

41A

488

254

015

4553

πrarr

πdC

428290

498

249

007

minus130

πrarr

πET

51A

492

252

012

minus11712

πrarr

πdC

428290

419296

511

243

004

minus622

πrarr

πET

61A

502

247

003

minus3034

πrarr

πET

525

236

000

minus046

πrarr

πET

71A

513

242

001

2737

nrarr

πdG

527

235

001

332

πrarr

πET

81A

515

241

006

14772

πrarr

πdC

451275

545

227

045

minus15609

πrarr

πdG

484256

486255

91A

522

238

001

1978

nrarr

πdG

552

225

002

minus1795

nrarr

πdG

101 A

538

230

071

7299

πrarr

πdG

484256

451275

556

223

013

5929

πrarr

πdG

111 A

541

229

001

minus6858

πrarr

πdG

558

222

009

minus525

πrarr

πdG

121 A

548

226

002

minus1519

nrarr

πdC

563

220

008

minus7436

nrarr

πdG

131 A

553

224

001

minus924

nrarr

πdC

609

204

001

1379

nrarr

πdG

141 A

601

206

001

minus4159

nrarr

πdG

626

198

002

minus464

nrarr

πdG

aExcitatio

nenergyA

llexcitedstates

ofSA

C-CIcalculations

have

been

shifted

tothelower

values

by05eV

inFigure

8bOscillator

strengthcRotatorystrengthddG

representsintram

olecular

excited

states

ofdG

dCrepresentsintram

olecular

excitedstates

ofdC

and

ETrepresentselectron

transfer

excitedstates

from

dGto

dCeRef

4

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5553

state because the stacking interaction is weaker in B-DNA thanin Z-DNA (Figure 1) Therefore the first peak of theexperimental CD spectra originates from the ET excitationfrom dG to dC for Z-DNA but corresponds to theintramolecular excitation in dC for B-DNA The second peakwith a negative sign is assigned to the 81A state The 81A statewhich is the πminusπ of dG is the origin of both the main peaks ofthe UV spectrum and the second peak of the CD spectrumThe ET excited states of the tetramer models were calculated

at a lower energy region than those of the dimer models forboth B- and Z-DNA This result may be related to the overlapbetween the nucleic acid bases The overlap exists only in partof the nucleic acid bases in the dimer models but in thetetramer models the overlap covers the hydrogen bonds as wellas the nucleic acid bases The ET excited states are importantfor the CD spectrum of Z-DNA because the lowest excitedstate is the ET type and is negative for rotatory strength but forB-DNA the ET excited states are not important because thesestates are weak in both oscillator and rotatory strengthsThe SAC-CI results of the tetramer models are calculated at

a higher region by about 05 eV because of the small activespace Therefore better SAC-CI calculations are required forcomparison with the experimental UV and CD spectra of DNAHowever these results indicate that the strong stackinginteraction characteristic of the helical structure of Z-DNAchanges the sign of the rotatory strength for the lowest excitedstates of dG and dC and produces the emergence of the ETexcited states

4 CONCLUSIONSDNA chemistry is affected by three important factors (1) theconformation of the monomer (dG) which is related to thedihedral angle between deoxyribose and guanine (2) thehydrogen-bonding interactions and (3) the stacking inter-actions between two nucleic acid bases To verify whether thesethree factors are important for the CD spectra of DNA weexamined four different models using the SAC-CI method(1) dG monomer model We first applied the SAC-CI

method to the dG monomer to examine the effect ofconformation on the CD spectra of dG and from this resultto identify the structure of the dG monomer in solution Ourcalculations showed that the shape and the sign of the CDspectra depend strongly on the dihedral angle between thedeoxyribose and the guanine but that the SAC-CI UV spectraof both anti- and syn-dG are very similar because the UVspectra are not strongly dependent on the conformation Theexperimental CD spectrum agreed with the theoreticalspectrum of dG only for the anti conformation The resultingthermally averaged spectra also support the anti geometry ofdG as a stable conformation in solution However the SAC-CICD spectrum of syn-dG is opposite in sign from that of theexperimental spectrum Thus the SAC-CI CD spectra of dG inboth anti and syn conformations do not explain the negativesign at 295 nm of the CD spectra of the DNA composed of dGand dC(2) Composit model of dG and dC monomers A

comparison of the calculated composite spectra of dG anddC with the experimental spectra of the two-component DNAshowed similarities between B- and Z-DNA because theconformation of dC is always anti in both B- and Z-DNAThe SAC-CI results led us to confirm that that theconformation of dG is not the origin of the negative peak at295 nm in Z-DNA

(3) Dimer model Our examination of the dimer modelpermitted us to investigate the separate importance of thehydrogen-bonding and the stacking interactions in thestructural chemistry of B- and Z-DNA The SAC-CIcalculations showed that the hydrogen-bonding interactionschange the excitation energies and that the stacking interactionsare responsible for the changes in the sign of the CD rotatorystrength The SAC-CI UV spectra of the hydrogen-bondingmodels (B-Hbond and Z-Hbond models) are similar to theexperimental UV spectra in which the main peak as well as theshoulder peak is well reproduced The SAC-CI CD spectra ofthe stacking models (B-Stack and Z-Stack models) have thesame sign as the experimental CD spectra We concluded thatthe stacking interaction is especially important for the CDspectrum of the strongly stacked Z-DNA because the CD signalat 295 nm changed to a negative sign only for the Z-Stackmodel(4) Tetramer model Finally our calculations with the

tetramer model that contain both hydrogen-bonding andstacking interactions revealed that the shape of the SAC-CIUV and CD spectra compared well with the experimental onesThe ET excitations as well as the lowest intramolecularexcitation of dC and dG are the origins of the negative sign ofthe experimental CD spectrum of Z-DNA The ET excited stateis calculated at a lower energy region than the intramolecularexcited states of dG or dC for the strongly stacked Z-DNA butat a higher energy region for the weakly stacked B-DNA Thenegative sign of the first peak of the experimental CD spectra ofZ-DNA is due to the strong stacking effect characteristic of itshelical structureThe dynamic behavior of the dG monomer in solution was

partially considered by taking the Boltzmann averaging of thecalculated spectra over the various conformations Fortunatelythe Boltzmann averaged SAC-CI CD spectrum was in goodagreement with the experimental spectra although the SAC-CICD spectra largely depend on the conformation of dGHowever in this study we used only the static model for dimerand tetramer models The SAC-CI CD results of the dimer andtetramer models explained the difference between theexperimental CD spectra of B- and Z-DNA However sincethe structure of DNA would also be flexible in solution it isnecessary to take into account the dynamic behavior of DNAThis is an important and interesting topic

ASSOCIATED CONTENT

S Supporting InformationSAC-CI results of the hydrogen-bonding and stacking dimermodels of Z- and B-DNA using small active space and withoutthe polarization d-functions This material is available free ofcharge via the Internet at httppubsacsorg

AUTHOR INFORMATION

Corresponding AuthorPhone +81-75-634-3211 Fax +81-75-634-3211 E-mail hnakatsujiqcriorjp

NotesThe authors declare no competing financial interest

ACKNOWLEDGMENTS

The computations were performed at the Research Center forComputational Science Okazaki Japan

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5554

REFERENCES(1) Voet D Voet J G Eds Biochemistry John Wiley amp Sons IncNew York 1995(2) Ghosh A Bansal M Acta Crystallogr 2003 D59 620minus626(3) Rich A Nordheim A Wang A H-J Annu Rev Biochem 198453 791minus846(4) Tran-Dinh S Taboury J Neumann J-M Huynh-Dinh TGenissel B Langlois drsquoEstaintot B Igolen J Biochemistry 1984 231362minus1371(5) Sugiyama H Kawai K Matsunaga A Fujimoto K Saito IRobinson H Wang A H-J Nucleic Acids Res 1996 24 1272minus1278(6) Kawai K Saito I Sugiyama H J Am Chem Soc 1999 1211391minus1392(7) Oyoshi T Kawai K Sugiyama H J Am Chem Soc 2003 1251526minus1531(8) Xu Y Ikeda R Sugiyama H J Am Chem Soc 2003 12513519minus13524(9) Berova N Nakanishi K Woody R W Eds Circular DichroismPrinciples and Applications Wiley-VCH New York 2000(10) Bauernschmitt R Ahlrichs R Chem Phys Lett 1996 256454minus464(11) Casida M E Jamorski C Casida K C Salahub D R JChem Phys 1998 108 4439minus4449(12) Tsolakidis A Kaxiras E J Phys Chem A 2005 109 2373minus2380(13) Varsano D Di Felice R Marques M A L Rubio A J PhysChem B 2006 110 7129minus7138(14) Nielsen L M Holm A I S Varsano D Kadhane UHoffmann S V Di Felice R Rubio A Nielsen S B J Phys ChemB 2009 113 9614minus9619(15) Tonzani S Schatz G C J Am Chem Soc 2008 130 7607minus7612(16) Casida M E Salahub D R J Chem Phys 2000 113 8918minus8935(17) Dreuw A Weisman J L Head-Gordon M J Chem Phys2003 119 2943minus2946(18) Maitra N T Zhang F Cave R J Burke K J Chem Phys2004 120 5932minus5937(19) McCann D M Stephens P J J Org Chem 2006 71 6074minus6098(20) Casida M E J Mol Struct 2009 914 3minus18(21) Niehaus T A J Mol Struct 2009 914 38minus49(22) Tawada Y Tsuneda T Yanagisawa S Yanai T Hirao K JChem Phys 2004 120 8425minus8433(23) Chiba M Tsuneda T Hirao K J Chem Phys 2006 124144106-1-11(24) Nakatsuji H Hirao K J Chem Phys 1978 68 2053minus2065(25) Nakatsuji H Chem Phys Lett 1978 59 362minus364(26) Nakatsuji H Chem Phys Lett 1979 67 329minus333(27) Nakatsuji H Chem Phys Lett 1979 67 334minus342(28) Nakatsuji H Bull Chem Soc Jpn 2005 78 1705minus1724(29) Ehara M Hasegawa J Nakatsuji H SAC-CI Method Appliedto Molecular Spectroscopy In Theory and Applications of Computa-tional Chemistry The First 40 Years A Vol of Technical and HistoricalPerspectives Dykstra C E Frenking G Kim K S Scuseria G EEds Elsevier Oxford UK 2005 pp 1099minus1141(30) Miyahara T Hasegawa J Nakatsuji H Bull Chem Soc Jpn2009 82 1215minus1226(31) Bureekaew S Hasegawa J Nakatsuji H Chem Phys Lett2006 425 367minus371(32) Nakatani N Hasegawa J Nakatsuji H J Am Chem Soc2007 129 8756minus8765(33) Fujimoto K Hasegawa J Nakatsuji H Bull Chem Soc Jpn2009 82 1140minus1148(34) Miyahara T Nakatsuji H Collect Czech Chem Commun2011 76 537minus552(35) Sugiyama H Unpublished(36) Geertsen J Rittby M Bartlett R J Chem Phys Lett 1989164 57minus62

(37) Stanton J F Bartlett R J J Chem Phys 1993 98 7029minus7039(38) Ehara M Ishida M Toyota K Nakatsuji H SAC-CIGeneral-R Method Theory and Applications to the Multi-ElectronProcesses In Reviews in Modern Quantum Chemistry (A Tribute toProfessor Robert G Parr) Sen K D Ed World Scientific Singapore2003 pp 293minus319(39) SAC-CI GUIDE httpwwwqcriorjpsacci(40) Frisch M J Trucks G W Schlegel H B et al Gaussian 03development revision revision E05 Gaussian Inc Wallingford CT2004(41) Hohenberg P Kohn W Phys Rev 1964 136 B864minusB871(42) Kohn W Sham L J Phys Rev 1965 140 A1133minusA1138(43) Salahub D R Zerner M C Eds The Challenge of d and fElectrons The American Chemical Society Washington DC 1989(44) Parr R G Yang W Density-Functional Theory of Atoms andMolecules Oxford University Press Oxford UK 1989(45) Becke A D J Chem Phys 1993 98 5648minus5652(46) Lee C Yang W Parr R G Phys Rev 1988 37 785minus789(47) Hariharan P C Pople J A Theor Chim Acta 1973 28 213minus222(48) Francl M M Pietro W J Hehre W J Binkley J SDeFrees D J Pople J A Gordon M S J Chem Phys 1982 773654minus3665(49) Tomasi J Mennucci B Cammi R Chem Rev 2005 1052999minus3093(50) Fukuda R Ehara M Nakatsuji H Cammi R J Chem Phys2011 134 104109minus1minus11(51) Dunning T H Jr Hay P J In Modern Theoretical ChemistrySchaefer H F III Ed Plenum New York 1976 Vol 3 pp 1minus28(52) Nakatsuji H Chem Phys 1983 75 425minus441(53) Hocquet A Leulliot N Ghomi M J Phys Chem B 2000 1044560minus4568(54) Mishra S K Mishra P C J Comput Chem 2002 23 530minus540(55) Sprecher C A Johnson W C Biopolymers 1977 16 2243minus2264(56) Clark L B Peschel G G Tinoc I Jr J Phys Chem 1965 693615minus3618(57) Voet D Gratzer W B Cox R A Doty P Biopolymers 19631 193minus208(58) Hobza P Mehlhorn A Carsky P Zahradnik R J Mol Struct1986 138 387minus399(59) Hobza P Sponer J Chem Rev 1999 99 3247minus3276

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5555

Page 6: Helical Structure and Circular Dichroism Spectra of DNA: A ... · B-DNA, have right-handed double-helical structure, while Z-DNA is left-handed and appears only when DNA has a special

The SAC-CI CD spectra of dG and dC alone (Figure5abde) having the structures of those in B- and Z-DNA andthe composite CD spectra (Figure 5cf) obtained as (a + b andd + e) were compared to the experimental CD spectra of B-and Z-DNA The SAC-CI excitation energies oscillatorstrengths rotatory strengths and natures of the excited statesfor the monomer model (dG and dC) were also determined(Table 3)The conformations of dG and dC are the same as those in

the ideal B-DNA or Z-DNA in which dG and dC are arrangedalternately (Figure 1) However the X-ray structure wasobtained for the DNA composed of only 6 sequences in thecase of Z-DNA Since the geometry is affected by the effect ofthe edge of DNA the conformation is different for eachmonomer Since the conformation of dG is in particular moreflexible than that of dC in DNA we calculated two kinds of dG(model I and II) in B- and Z-DNA respectively The models Iand II correspond to dG of the hydrogen-bonding models (B-Hbond and Z-Hbond models) and the stacking models (B-Stack and Z-Stack models) of the dimer models respectivelyThe dihedral angle (φ) between the deoxyribose and theguanine is 2763deg and 3007deg respectively for models I and IIof B-DNA and 1113deg and 1175deg respectively for models Iand II of Z-DNA The dihedral angle between the deoxyriboseand cytosine is 3232deg for B-DNA and 3329deg for Z-DNA Therotatory strength of dG is very different between models I andII for both B- and Z-DNA as shown in Table 3 Therefore fordG we prepared an averaged CD spectra for models I and II(Figure 5ad) For Z-DNA the SAC-CI CD spectrum of dG(Figure 5a) is negative for the first band and positive for thesecond band as in the experimental spectrum of Z-DNAalthough the excitation energies are calculated at a higherregion However the SAC-CI result is opposite in sign to theexperimental CD spectrum for the first band of B-DNA (Figure5d) These results show that in the CD spectra of DNA the

conformation of dG (monomer) is related to the sign of thesecond band but has no effect on the sign of the first bandThe CD spectra of dC have four peaks (Figure 5be) The

first peak represents the πminusπ of HOMO to LUMO Thesecond peak is the excitation from the nonbonding orbital tothe LUMO The third and fourth peaks are composed of boththe πminusπ and nminusπ from next-HOMO and nonbondingorbitals to LUMO The third and fourth peaks are mainly πminusπand nminusπ respectively The SAC-CI CD spectra of dC aresimilar in B- and Z-DNA with two positive and two negativepeaks The rotatory strength is positive for the first and thirdpeaks and negative for the second and fourth peaks Thereforethe CD spectra of dC are always positive at around 295 nmcorresponding to the conformation of dC which is always antiin DNAAs shown in Figure 5cf the composite SAC-CI CD spectra

obtained from dG and dC alone are similar between B- and Z-DNA The sign at 295 nm is positive for both B- and Z-DNAbecause the rotatory strength of dC which is the lowest peak ismuch stronger than that of dG Therefore the composite CD

Table 2 Boltzmann Distribution of dG along theConformational Angle φ

angle (φ) all ()a anti ()b syn ()c

10 086 1130 01450 00470 01090 170 91110 279 1431295 (syn) 1131 517150 454 224170 041 25190 008210 000230 010250 325 41270 1534 195290 1700 2163103 (anti) 2669 339330 1307 166350 258 33total 1000 1000 1000

aldquoallrdquo gives the ratio of each conformer when averaging included all 18conformers bldquoantirdquo gives the ratio of each conformer when only sevenconformations around anti are used cldquosynrdquo gives the ratio of eachconformer when only five conformations around syn are used

Figure 5 SAC-CI CD spectra of monomer ((a) dG in Z-DNA (b) dCin Z-DNA (d) dG in B-DNA and (e) dC in B-DNA) and thecomposite CD spectra of (c) Z-DNA and (f) B-DNA obtained fromthe SAC-CI CD spectra of dG and dC compared with theexperimental CD spectra4 (black lines) of Z-DNA (aminusc) and B-DNA (dminusf) The red and blue lines represent the excited state of dGand dC respectively The magenta line in panels c and f represents thecomposite spectra

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5547

spectrum is similar to that of dC for both B- and Z-DNA As aresult in the composite CD spectra of B- and Z-DNA the signat around 295 nm is independent of the geometry of themonomer (dG or dC) even though the CD spectrum of dGdepends on the dihedral angle between the guanine and thedeoxyribose because the lowest peak originates from dCwhose geometry is always anti in both B- and Z-DNA Thisindicates that the noninteracting dG and dC cannot explain theexperimental spectra of DNA and that the hydrogen-bondingor stacking interactions with the neighboring nucleic acid basesmust be important for the CD spectra of DNA34 UV and CD Spectra of Dimer Models In the studies

of the structure spectroscopy and functions of DNA it isgenerally recognized that the hydrogen-bonding interaction andthe stacking interaction are the two most important interactionsthat characterize the properties of DNA So we expect thatthese interactions might also be the important factors that affectthe features of the UV and CD spectra of DNA To investigatethe possible origin we considered the dimer model composedof dG and dC These models contained either the hydrogen-bonding interaction or the stacking interaction (Figure 1)However we note that the SAC-CI results using these modelsneed not be in agreement with the experimental spectrabecause these models lack either the stacking interaction or thehydrogen-bonding interaction We must use the tetramer

model at least to compare with the experimental spectra asnoted below

341 UV Spectra of Dimer Models Figure 6 shows theSAC-CI UV spectra of both hydrogen-bonding and stackingmodels of both B- and Z-DNA compared with the experimentalUV spectrum The experimental UV spectra of Z-DNA displaya main peak at 256 nm and a shoulder peak at around 290 nmThe experimental UV spectrum of B-DNA is very similar butthe shoulder peak at around 290 nm is weaker Table 4 showsthe excitation energies oscillator strengths rotatory strengthsnatures and the types of the excited states of the SAC-CIresults for both hydrogen-bonding and stacking models of thedimer When the excitation in the dimer model is not withindG or dC but corresponds to the electron transfer (ET)between dG and dC we assigned it as the ET typeThe SAC-CI UV spectrum of the hydrogen-bonding model

of Z-DNA (Z-Hbond model) is in good agreement with theexperimental UV spectrum of Z-DNA (Figure 6a) Threestrong peaks of the πminusπ nature are calculated The loweststrong peak corresponds to the shoulder peak at 428 eV (290nm) of the experimental UV spectrum of Z-DNA Theremaining two strong peaks correspond to the main peak at484 eV (256 nm) of the experimental UV spectrum of Z-DNARed blue and green colors represent the excited states of dGdC and ET respectively

Table 3 SAC-CI UV and CD Spectra of dG and dC in Z- and B-DNA

Z-DNA B-DNA

dG (model I φ = 1113deg) dG (model I φ = 2763deg)

EEa oscb rotc EEa oscb rotc

state (eV) (nm) (au) (10minus40 cgs) nature (eV) (nm) (au) (10minus40 cgs) nature

11A 441 281 017 125 π rarr π 443 280 016 minus1035 π rarr π21A 474 262 000 minus1099 n rarr π 488 254 000 1099 n rarr π31A 493 251 023 2066 π rarr π 513 242 025 minus3355 π rarr π41A 578 214 000 minus864 n rarr π 595 208 002 minus2619 π rarr π51A 591 210 004 minus656 π rarr σ 599 207 001 3571 n rarr π61A 611 203 000 minus438 n rarr π 618 201 000 983 n rarr π71A 628 197 000 359 π rarr σ 675 184 008 1235 π rarr σ81A 659 188 030 634 π rarr π 678 183 028 minus2402 π rarr π

dG (model II φ = 1175deg) dG (model II φ = 3007deg)

EEa oscb rotc EEa oscb rotc

state (eV) (nm) (au) (10minus40 cgs) nature (eV) (nm) (au) (10minus40 cgs) nature

11A 437 284 016 minus451 π rarr π 434 286 016 minus247 π rarr π21A 459 270 000 minus322 n rarr π 493 251 000 minus1467 n rarr π31A 491 252 022 minus319 π rarr π 518 239 021 minus200 π rarr π41A 581 213 003 323 π rarr π 595 209 001 2884 n rarr π51A 598 207 000 minus738 π rarr π 599 207 003 minus647 π rarr π61A 621 200 000 796 n rarr π 617 201 000 minus3033 n rarr π71A 641 193 000 353 π rarr σ 660 188 048 978 π rarr π81A 647 191 031 minus077 π rarr π 666 186 001 326 π rarr σ

dC (φ = 3329deg) dC (φ = 3232deg)

EEa oscb rotc EEa oscb rotc

state (eV) (nm) (au) (10minus40 cgs) nature (eV) (nm) (au) (10minus40 cgs) nature

11A 413 300 009 2385 π rarr π 409 303 009 1403 π rarr π21A 442 280 000 minus1548 n rarr π 452 274 000 minus915 n rarr π31A 507 245 011 8392 π + n rarr π 499 249 007 5975 π + n rarr π41A 527 235 011 minus8639 n + π rarr π 527 235 015 minus8017 π + n rarr π51A 573 216 000 minus1385 n rarr π 559 222 001 393 n rarr π61A 632 196 039 2349 π rarr π 625 198 032 1098 π rarr π

aExcitation energy bOscillator strength cRotatory strength

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5548

The shoulder peak in the experimental UV spectrum of Z-DNA is composed of 11A (439 eV) and 21A (453 eV) statesThe 11A state (in blue color) of the Z-Hbond modelcorresponds to the 11A state of dC at 413 eV which is theexcitation from the HOMO to the LUMO of dC This state haslarge oscillator strength of the πminusπ nature In the Z-Hbondmodel this state is blue-shifted by 026 eV because theHOMOminusLUMO gap (0449 au) of the Z-Hbond model islarger than that (0438 au) of dC This state is in goodagreement with the first (shoulder) peak of the experimentalUV spectrum of Z-DNA However the 21A state (453 eV redcolor) of the Z-Hbond model is assigned to the 31A state (493eV) of dG which is the excitation from the HOMO to the next-LUMO of dG This large shift of the excitation energy is causedby the MO-order change The MO of the Z-Hbond modelcorresponding to the next-LUMO of dG becomes lower thanthe MO of the Z-Hbond model corresponding to the LUMOof dG by the formation of hydrogen bonds with cytosine in theZ-Hbond model Therefore this state is red-shifted by 04 eVHowever the oscillator strength of this state is small despite itsπminusπ nature The transition moment is canceled because themain configuration includes not only the excitation from theHOMO to the next-LUMO in dG but also the excitation fromthe next-HOMO to the LUMO in dC and the electron transfertype excitation from the HOMO of dG to the LUMO and next-LUMO of dCThe main peak of the experimental UV spectrum of Z-DNA

is composed of 31A (484 eV) and 51A (493 eV) states havinglarge oscillator strengths (Figure 6a) The 31A state (so blue

color) of the Z-Hbond model is assigned to the 31A state (507eV) of dC which reflects the strong excitation from the next-HOMO to the LUMO of dC The 51A state (red color) of theZ-Hbond model is assigned to the 11A state (441 eV) of dGwhich is the excitation from the HOMO to the LUMO in dGThe excitation energy increases by 052 eV because theHOMOminusLUMO gap becomes large which is the opposite tothat of the 21A state but again due to the hydrogen-bondingeffectFor B-DNA however the SAC-CI UV spectrum of the

hydrogen-bonding model (B-Hbond model Figure 6c) also hasboth main and shoulder peaks in which the main peak is higherand the shoulder peak is lower as compared with theexperimental UV spectrum of B-DNA The main peak (41Aand 51A states) of the SAC-CI spectra of the B-Hbond model ishigher than that of the experimental spectra similarly to theSAC-CI spectrum of the Z-Hbond model The 31A state of dGin B-DNA is also higher than that in Z-DNA (Table 3) Theseresults may be related to the fact that dG is more flexible thandC because the conformation of dG is anti for B-DNA but synfor Z-DNA Since the stacking interaction of B-DNA is weakerthan that of Z-DNA the X-ray structure of B-DNA may bemore flexible than that of Z-DNA However the result that theoscillator strength of the lowest excited state of the B-Hbondmodel is weaker than that of the Z-Hbond model is reflected bythe fact that the shoulder peak of the experimental UV spectrais stronger in Z-DNA but weaker in B-DNAFor the SAC-CI UV spectra of the stacking models of both

B- (B-Stack model) and Z-DNA (Z-Stack model) the firstexcited states originated from dC were calculated at the lower-energy region (higher wavelength region) but the excitationenergy of the main peak is almost the same between the B-Hbond and B-Stack models or between the Z-Hbond and Z-Stack models If there is no hydrogen-bonding interaction inDNA the lowest excited state would be observed in the lower-energy region of the experimental spectra separately from themain peak The excitation energy of the lowest excited state ofthe stacking model is lower than that of the hydrogen-bondingmodel by 044 eV for Z-DNA and 033 eV for B-DNArespectively This result may correspond to the fact that thelowest excitation energy of cytosine differs between the gasphase (428 eV)56 and in solution (464 eV)57

For both B- and Z-DNA the SAC-CI results of thehydrogen-bonding models (B-Hbond and Z-Hbond) are inbetter agreement with the experimental spectra than those ofthe stacking models (B-Stack and Z-Stack) Thus thehydrogen-bonding interaction is reflected in the UV spectraof DNA On the other hand the shoulder peak observed for Z-DNA may reflect the stronger shoulder nature of the Z-Stackmodel in Z-DNA However the UV spectra of DNA providelittle information about stacking interactions between nucleicacid bases

342 CD Spectra of Dimer Models Figure 7 shows theSAC-CI CD spectra of both hydrogen-bonding and stackingmodels of B- and Z-DNA compared with the experimental CDspectra In the experimental CD spectra the first peak isnegative at 296 nm and the second peak is positive at 275 nmfor Z-DNA However B-DNA displays peaks that are oppositein sign the first peak is positive at 280 nm and the second peakis negative at 255 nmIt is worth nothing that the SAC-CI spectrum has a negative

peak at 295 nm only for the stacking model of Z-DNA (Z-Stack) but for the other models the SAC-CI spectrum is

Figure 6 SAC-CI UV spectra for the dimer models The SAC-CIspectra (magenta line) of (a) Z-Hbond model (b) Z-Stack model (c)B-Hbond model and (d) B-Stack model are compared with theexperimental UV spectra4 (black line) of Z-DNA (ab) and B-DNA(cd) The red and blue lines represent the intramolecular excitedstates of dG and dC respectively The green lines represent theelectron transfer (ET) excited state from dG to dC

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5549

Table

4SA

C-CIUVandCD

Spectraof

theHydrogen-Bon

ding

andStacking

Dim

erMod

elsof

Z-andB-D

NA

Z-DNA

B-DNA

SAC-CIof

Z-H

bond

exptle

SAC-CIof

B-H

bond

exptle

EEa

oscb

rotc

UV

EEa

oscb

rotc

UV

state

(eV)

(nm)

(au)

(10minus

40cgs)

nature

typed

(eVnm)

(eV)

(nm)

(au)

(10minus

40cgs)

nature

Typed

(eVnm)

11A

439

282

020

3513

πrarr

πdC

(11 A)

428290

437

284

017

1143

πrarr

πdC

(11 A)

21A

453

274

004

884

πrarr

πdG

(31 A)

456

272

003

minus1481

πrarr

πdG

(31 A)

31A

484

256

018

912

πrarr

πdC

(31 A)

493

251

006

583

πrarr

πET

41A

485

255

005

minus1145

πrarr

πET

498

249

021

1767

πrarr

πdC

(3+41A)

51A

493

252

042

1182

πrarr

πdG

(11 A)

484256

518

240

043

minus1390

πrarr

πdG

(11 A)

484256

61A

512

242

000

057

nrarr

πdC

(21 A)

525

236

000

minus1197

nrarr

πdC

(21 A)

71A

523

237

000

minus1384

nrarr

πdG

(21 A)

529

234

000

minus579

nrarr

πdG

(2+51A)

81A

623

199

000

minus034

nrarr

πdG

(2+51A)

SACminusCIof

Z-Stack

exptle

SACminusCIof

B-Stack

exptle

EEa

oscb

rotc

CD

EEa

oscb

rotc

CD

state

(eV)

(nm)

(au)

(10minus

40cgs)

nature

typed

(eVnm)

(eV)

(nm)

(au)

(10minus

40cgs)

nature

typed

(eVnm)

11A

395

314

009

minus4511

πrarr

πdC

(11 A)

419296

404

307

007

4358

πrarr

πdC

(11 A)

443280

21A

421

294

000

minus3349

πrarr

πdG

(11 A)

419296

440

282

006

minus1032

πrarr

πdG

(11 A)

31A

445

279

001

2967

nrarr

πdC

(21 A)

461

269

000

minus503

nrarr

πdC

(21 A)

41A

472

263

000

minus1335

nrarr

πdG

(21 A)

493

251

013

9407

πrarr

πdC

(3+41A)

51A

485

256

019

minus1796

πrarr

πdG

(31 A)

451275

507

245

005

minus15115

nrarr

πdG

(21 A)

486255

61A

499

248

006

4861

πrarr

πET

451275

512

242

014

minus11056

πrarr

πdG

(31 A)

486255

71A

526

236

007

2721

πrarr

πET

+dC

(31 A)

451275

521

238

014

19475

πrarr

πET+dG

(31 A)

81A

565

219

000

minus231

nrarr

πdC

(41 A)

619

200

000

minus805

nrarr

πdG

(41 A)

aExcitatio

nenergybOscillator

strengthcRotatorystrengthddG

representsintram

olecular

excitedstates

ofdG

dCrepresentsintram

olecular

excitedstates

ofdC

and

ETrepresentselectron

transfer

excitedstates

from

dGto

dCeRef

4

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5550

positive at 295 nm Therefore the strong negative sign of theCD spectra at 295 nm indicates that a strong stackinginteraction exists in Z-DNA The SAC-CI CD spectra of bothstacking models (B-Stack and Z-Stack models) more closelyresemble the experimental CD spectra of B- and Z-DNA thanthose of the hydrogen-bonding models (B-Hbond and Z-Hbond models) For the SAC-CI CD spectrum of the Z-Stackmodel the first and second peaks are negative and positiverespectively but they are the opposite of those obtained fromthe Z-Hbond model For the B-Stack and B-Hbond modelshowever the first peak is positive in the SAC-CI CD spectrabut the second and third peaks are negative and positiverespectively for the B-Stack model in opposition to the B-Hbond model Since the stacking models do not have thehydrogen-bonding interactions the excitation energy of the firstpeak is lower than that of the experimental spectrum as notedaboveIn the Z-Stack model the first peak with a negative sign is

assigned to the 11A (395 eV) and 21A (421 eV) states (Table4) The 11A state of the Z-Stack model corresponds to the 11Astate (413 eV) of dC which is the excitation of the πminusπ fromthe HOMO to the LUMO in dC A large oscillator strength isthe origin of the first (shoulder) peak but the excitation energyis lower than the experimental value because there is nohydrogen bond in the Z-Stack model The rotatory strength ofthe 11A state of dC changes the sign from positive to negativeby the stacking interaction with dG Therefore the stackinginteraction between dG and dC accounts for the strongnegative sign of the CD spectra in Z-DNA

The excitation energy of the 21A state (421 eV) is also closeto the first peak of the experimental CD spectrum This state isassigned to the 11A state (437 eV) of dG which is the HOMOto LUMO πminusπ excitation in dG but also includes thecontribution of the excitation from the next-HOMO to LUMOin dC The oscillator strength in the UV spectrum is smallbecause this state differs from the 11A state of dG by theinclusion of the excitation in dC Although the excitationenergy is shifted by the formation of hydrogen bonds betweendG and dC the rotatory strength of a large negative maycontribute to the strong negative sign of the CD spectra in Z-DNA In the 31A state (445 eV) which is the excitation fromnonbonding orbital to the LUMO in dC the rotatory strengthchanges to positive from the negative sign of the monomer dCHowever this state is not the origin of the second peak as willbe noted later in the tetramer model section Thus the rotatorystrength is strongly affected by the stacking interactionThree important excited states are calculated in the higher-

energy region of the second peak The 51A state (485 eV) hasthe largest oscillator strength and is assigned to the 31A (491eV) state of dG which is the excitation from HOMO to next-LUMO This state is assigned to the main peak (484 eV) of theexperimental UV spectrum of Z-DNA The 6 and 71A states(499 and 526 eV) with a large rotatory strength are the ETstate from the HOMO of dG to the LUMO of dC The mainconfiguration of the 71A state also includes the excitation fromthe next-HOMO to the LUMO in dCIn the B-Stack model the first peak with a positive sign is

assigned to the 11A state (404 eV) This state corresponds tothe 11A state (409 eV) of dC which is the excitation from theHOMO to the LUMO The 21A state (440 eV) is assigned tothe 11A state (434 eV) of dG which reflects the excitationfrom the HOMO to the LUMO Since the B-Stack model doesnot have hydrogen bonds the excitation energies are close tothose of dG or dC Both states have the oscillator strength butthe 11A state is positive while the 21A state is slightly negativefor the rotatory strength which is the same as for themonomers dC and dG respectively Therefore the stackinginteraction of B-DNA has little effect on the sign of the CDspectra at 295 nm because the stacking interaction is weaker inB-DNA than in Z-DNAThe second peak with a negative sign is assigned to the 5 and

61A states with a large rotatory strength The 5 and 61A states(507 and 512 eV in Table 4) are assigned to the 2 and 31Astates (493 and 518 eV in Table 3) of dG which are theexcitations from the nonbonding orbital and the HOMO to thenext-LUMO The main configuration is nminusπ for the 51A stateand πminusπ for the 61A state Both states have strong oscillatorand negative rotatory strengths which means that they can beassigned to the main peaksThe first ET excited states (71A state 521 eV) of B-Stack

model are calculated at a higher region than that (61A state499 eV) of the Z-Stack model because the stacking interactionis weaker in B-DNA than in Z-DNAThe SAC-CI results show that both hydrogen-bonding and

stacking interactions affect the UV and CD spectra of DNA Inparticular the excitation energy of the first excited stateincreases by the formation of hydrogen bonds and the stackinginteraction changes the sign of the CD spectrum of the firstexcited state from positive to negative The strong stackinginteraction characteristic of the helical structure of Z-DNA isthe origin of the strong negative sign at around 295 nm in theCD spectra The hydrogen-bonding and stacking interactions

Figure 7 SAC-CI CD spectra for the dimer models The SAC-CIspectra (magenta line) of (a) Z-Hbond model (b) Z-Stack model (c)B-Hbond model and (d) B-Stack model are compared with theexperimental CD spectra4 (black line) of Z-DNA (ab) and B-DNA(cd) The red and blue lines represent the intramolecular excitedstates of dG and dC respectively The green lines represent theelectron transfer (ET) excited state from dG to dC

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5551

also confer other smaller effects on the UV and CD spectra ofDNA In the higher energy region the SAC-CI CD spectra arenot in good agreement with the experimental CD spectrabecause these models have either hydrogen-bonding interactionor stacking interaction but not both Therefore we need toinvestigate a better model that includes both hydrogen bondingand stacking interactions within the same model A minimumsuch model is the tetramer model of Figure 1 on which weexplain in the next section35 Tetramer Model The SAC-CI calculations of the

dimer models were suitable to investigate the effects of thehydrogen-bonding interaction and the stacking interactionseparately For example the SAC-CI UV spectra of thehydrogen-bonding models were in good agreement with theexperimental UV spectra implying the importance of thehydrogen-bonding interaction The SAC-CI CD spectra for thestacking model clarified that the stacking interaction is thesource of the negative CD peak at around 295 nm Howevergenerally speaking the SAC-CI CD spectra were markedlydifferent from the experimental spectra The SAC-CI UVspectra of the stacking models were composed of two peaksdifferent from the experimental UV spectra Thus theinformation that we gained from the dimer models still didnot fully resolve these differences in the spectra acquired by twodifferent processes So we used the tetramer model (Figure 1)which included both hydrogen-bonding and stacking inter-actions within the same modelFigure 8 shows the SAC-CI UV and CD spectra of the

tetramer model of Z-DNA (Z-Tetra model) and B-DNA (B-Tetra model) in comparison with the experimental UV andCD spectra In order to compare the experimental spectra theSAC-CI results were shifted to lower energy by 05 eV Thecomputational conditions were poor as compared with theother SAC-CI calculations In the tetramer calculations the d-polarization functions were not added and the small activespace was used The SAC-CI calculations using the small activespace and without the d-polarization functions gave higherexcitation energies by about 04minus05 eV (see SupportingInformation Table S1) for the hydrogen-bonding dimermodels In particular the excited states corresponding to themain peak of the experimental UV spectra were shifted by 048eV for both the Z-Hbond model (61A state) and the B-Hbondmodel (71A state)Table 5 shows the excitation energies oscillator strengths

rotatory strengths natures and type of the excited states of theSAC-CI results for tetramer modelsThe SAC-CI UV spectrum of the Z-Tetra model is in good

agreement with the experimental UV spectrum of Z-DNA Themain peak as well as the shoulder peak is reproduced by theSAC-CI method The shoulder peak is assigned to the 4 and51A states which is the πminusπ excitation (HOMO to LUMO)in dC and the main peak is assigned to the 101A state which isthe πminusπ (HOMO to next-LUMO) excitation in dGThe SAC-CI UV spectrum of the B-Tetra model is also in a

reasonable agreement with the experimental UV spectrum of B-DNA although the shape of the shoulder is slightly different Inthe B-Tetra model the shoulder peak is assigned to the 21Astate which is the πminusπ (HOMO to LUMO) excitation in dCand the main peak is assigned to the 81A state which is theπminusπ (HOMO to next-LUMO) excitation in dGThe sign of the rotatory strength is the same between the

SAC-CI CD spectrum of the Z-Tetra model and theexperimental CD spectrum of Z-DNA although the negative

rotatory strength at 295 nm is very weak because thepolarization functions were not included in the tetramermodels It is well-known that the basis set must be good for anadequate description of the stacking interaction5859 In thedimer model the lack of polarization d-functions and thesmaller active space changed the intensity and sign of therotatory strength (see Supporting Information Table S1)Since the stacking interaction is small in B-DNA the SAC-CI

CD spectrum of the B-Tetra model is in good agreement withthe experimental CD spectrum of B-DNA and reproduces notonly the sign of the rotatory strength but also the overall shapeof the spectrumIn Z-DNA the 11A state is assigned to the first peak of the

experimental CD spectrum The 11A state is the ET excitationfrom the HOMO of dG to the LUMO of dC through thestacking conformation The 3 and 51A states with a largenegative rotatory strength are the excitation from the HOMOto the LUMO in dG and dC respectively These states alsocontribute to the first negative peak as in the 1 and 21A stateswith a negative rotatory strength in the Z-Stack model The 8and 101A states are assigned to the second peak of the CDspectrum The 101A state which is the origin of the main peakof the UV spectrum and the πminusπ of dC (81A state) alsocontribute to the second peak of the CD spectrumIn B-DNA the first peak of the CD spectrum is assigned to

the 21A state of πminusπ (HOMO to LUMO) in dC which is thesame as the origin of the first peak of the UV spectrum The ETstates are calculated at the higher energy region than the 21A

Figure 8 SAC-CI UV and CD spectra for the tetramer models TheSAC-CI spectra (magenta line) of Z-Tetra model (ab) and B-Tetramodel (cd) as compared with the experimental UV and CD spectra4

(black line) of Z-DNA (ab) and B-DNA (cd) The red and blue linesrepresent the intramolecular excited states of dG and dC respectivelyThe green lines represent the electron transfer (ET) excited state fromdG to dC All excited states of SAC-CI calculations have been shiftedto the lower values by 05 eV

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5552

Table

5SA

C-CIUVandCD

Spectraof

theTetramer

Mod

elsof

Z-andB-D

NA

Z-DNA

B-DNA

SACminusCIof

Z-Tetra

exptle

SAC-CIof

B-Tetra

exptle

EEa

Oscb

Rotc

UV

CD

EEa

Oscb

Rotc

UV

CD

state

(eV)

(nm)

(au)

(10minus

40cgs)

nature

typed

(eVnm)

(eVnm)

(eV)

(nm)

(au)

(10minus

40cgs)

nature

typed

(eVnm)

(eVnm)

11A

470

264

000

minus663

πrarr

πET

419296

479

259

003

minus041

πrarr

πdG

21A

481

258

002

6611

πrarr

πET

482

257

033

16891

πrarr

πdC

443280

31A

484

256

004

minus5326

πrarr

πdG

419296

484

256

001

minus7454

πrarr

πdC

41A

488

254

015

4553

πrarr

πdC

428290

498

249

007

minus130

πrarr

πET

51A

492

252

012

minus11712

πrarr

πdC

428290

419296

511

243

004

minus622

πrarr

πET

61A

502

247

003

minus3034

πrarr

πET

525

236

000

minus046

πrarr

πET

71A

513

242

001

2737

nrarr

πdG

527

235

001

332

πrarr

πET

81A

515

241

006

14772

πrarr

πdC

451275

545

227

045

minus15609

πrarr

πdG

484256

486255

91A

522

238

001

1978

nrarr

πdG

552

225

002

minus1795

nrarr

πdG

101 A

538

230

071

7299

πrarr

πdG

484256

451275

556

223

013

5929

πrarr

πdG

111 A

541

229

001

minus6858

πrarr

πdG

558

222

009

minus525

πrarr

πdG

121 A

548

226

002

minus1519

nrarr

πdC

563

220

008

minus7436

nrarr

πdG

131 A

553

224

001

minus924

nrarr

πdC

609

204

001

1379

nrarr

πdG

141 A

601

206

001

minus4159

nrarr

πdG

626

198

002

minus464

nrarr

πdG

aExcitatio

nenergyA

llexcitedstates

ofSA

C-CIcalculations

have

been

shifted

tothelower

values

by05eV

inFigure

8bOscillator

strengthcRotatorystrengthddG

representsintram

olecular

excited

states

ofdG

dCrepresentsintram

olecular

excitedstates

ofdC

and

ETrepresentselectron

transfer

excitedstates

from

dGto

dCeRef

4

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5553

state because the stacking interaction is weaker in B-DNA thanin Z-DNA (Figure 1) Therefore the first peak of theexperimental CD spectra originates from the ET excitationfrom dG to dC for Z-DNA but corresponds to theintramolecular excitation in dC for B-DNA The second peakwith a negative sign is assigned to the 81A state The 81A statewhich is the πminusπ of dG is the origin of both the main peaks ofthe UV spectrum and the second peak of the CD spectrumThe ET excited states of the tetramer models were calculated

at a lower energy region than those of the dimer models forboth B- and Z-DNA This result may be related to the overlapbetween the nucleic acid bases The overlap exists only in partof the nucleic acid bases in the dimer models but in thetetramer models the overlap covers the hydrogen bonds as wellas the nucleic acid bases The ET excited states are importantfor the CD spectrum of Z-DNA because the lowest excitedstate is the ET type and is negative for rotatory strength but forB-DNA the ET excited states are not important because thesestates are weak in both oscillator and rotatory strengthsThe SAC-CI results of the tetramer models are calculated at

a higher region by about 05 eV because of the small activespace Therefore better SAC-CI calculations are required forcomparison with the experimental UV and CD spectra of DNAHowever these results indicate that the strong stackinginteraction characteristic of the helical structure of Z-DNAchanges the sign of the rotatory strength for the lowest excitedstates of dG and dC and produces the emergence of the ETexcited states

4 CONCLUSIONSDNA chemistry is affected by three important factors (1) theconformation of the monomer (dG) which is related to thedihedral angle between deoxyribose and guanine (2) thehydrogen-bonding interactions and (3) the stacking inter-actions between two nucleic acid bases To verify whether thesethree factors are important for the CD spectra of DNA weexamined four different models using the SAC-CI method(1) dG monomer model We first applied the SAC-CI

method to the dG monomer to examine the effect ofconformation on the CD spectra of dG and from this resultto identify the structure of the dG monomer in solution Ourcalculations showed that the shape and the sign of the CDspectra depend strongly on the dihedral angle between thedeoxyribose and the guanine but that the SAC-CI UV spectraof both anti- and syn-dG are very similar because the UVspectra are not strongly dependent on the conformation Theexperimental CD spectrum agreed with the theoreticalspectrum of dG only for the anti conformation The resultingthermally averaged spectra also support the anti geometry ofdG as a stable conformation in solution However the SAC-CICD spectrum of syn-dG is opposite in sign from that of theexperimental spectrum Thus the SAC-CI CD spectra of dG inboth anti and syn conformations do not explain the negativesign at 295 nm of the CD spectra of the DNA composed of dGand dC(2) Composit model of dG and dC monomers A

comparison of the calculated composite spectra of dG anddC with the experimental spectra of the two-component DNAshowed similarities between B- and Z-DNA because theconformation of dC is always anti in both B- and Z-DNAThe SAC-CI results led us to confirm that that theconformation of dG is not the origin of the negative peak at295 nm in Z-DNA

(3) Dimer model Our examination of the dimer modelpermitted us to investigate the separate importance of thehydrogen-bonding and the stacking interactions in thestructural chemistry of B- and Z-DNA The SAC-CIcalculations showed that the hydrogen-bonding interactionschange the excitation energies and that the stacking interactionsare responsible for the changes in the sign of the CD rotatorystrength The SAC-CI UV spectra of the hydrogen-bondingmodels (B-Hbond and Z-Hbond models) are similar to theexperimental UV spectra in which the main peak as well as theshoulder peak is well reproduced The SAC-CI CD spectra ofthe stacking models (B-Stack and Z-Stack models) have thesame sign as the experimental CD spectra We concluded thatthe stacking interaction is especially important for the CDspectrum of the strongly stacked Z-DNA because the CD signalat 295 nm changed to a negative sign only for the Z-Stackmodel(4) Tetramer model Finally our calculations with the

tetramer model that contain both hydrogen-bonding andstacking interactions revealed that the shape of the SAC-CIUV and CD spectra compared well with the experimental onesThe ET excitations as well as the lowest intramolecularexcitation of dC and dG are the origins of the negative sign ofthe experimental CD spectrum of Z-DNA The ET excited stateis calculated at a lower energy region than the intramolecularexcited states of dG or dC for the strongly stacked Z-DNA butat a higher energy region for the weakly stacked B-DNA Thenegative sign of the first peak of the experimental CD spectra ofZ-DNA is due to the strong stacking effect characteristic of itshelical structureThe dynamic behavior of the dG monomer in solution was

partially considered by taking the Boltzmann averaging of thecalculated spectra over the various conformations Fortunatelythe Boltzmann averaged SAC-CI CD spectrum was in goodagreement with the experimental spectra although the SAC-CICD spectra largely depend on the conformation of dGHowever in this study we used only the static model for dimerand tetramer models The SAC-CI CD results of the dimer andtetramer models explained the difference between theexperimental CD spectra of B- and Z-DNA However sincethe structure of DNA would also be flexible in solution it isnecessary to take into account the dynamic behavior of DNAThis is an important and interesting topic

ASSOCIATED CONTENT

S Supporting InformationSAC-CI results of the hydrogen-bonding and stacking dimermodels of Z- and B-DNA using small active space and withoutthe polarization d-functions This material is available free ofcharge via the Internet at httppubsacsorg

AUTHOR INFORMATION

Corresponding AuthorPhone +81-75-634-3211 Fax +81-75-634-3211 E-mail hnakatsujiqcriorjp

NotesThe authors declare no competing financial interest

ACKNOWLEDGMENTS

The computations were performed at the Research Center forComputational Science Okazaki Japan

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5554

REFERENCES(1) Voet D Voet J G Eds Biochemistry John Wiley amp Sons IncNew York 1995(2) Ghosh A Bansal M Acta Crystallogr 2003 D59 620minus626(3) Rich A Nordheim A Wang A H-J Annu Rev Biochem 198453 791minus846(4) Tran-Dinh S Taboury J Neumann J-M Huynh-Dinh TGenissel B Langlois drsquoEstaintot B Igolen J Biochemistry 1984 231362minus1371(5) Sugiyama H Kawai K Matsunaga A Fujimoto K Saito IRobinson H Wang A H-J Nucleic Acids Res 1996 24 1272minus1278(6) Kawai K Saito I Sugiyama H J Am Chem Soc 1999 1211391minus1392(7) Oyoshi T Kawai K Sugiyama H J Am Chem Soc 2003 1251526minus1531(8) Xu Y Ikeda R Sugiyama H J Am Chem Soc 2003 12513519minus13524(9) Berova N Nakanishi K Woody R W Eds Circular DichroismPrinciples and Applications Wiley-VCH New York 2000(10) Bauernschmitt R Ahlrichs R Chem Phys Lett 1996 256454minus464(11) Casida M E Jamorski C Casida K C Salahub D R JChem Phys 1998 108 4439minus4449(12) Tsolakidis A Kaxiras E J Phys Chem A 2005 109 2373minus2380(13) Varsano D Di Felice R Marques M A L Rubio A J PhysChem B 2006 110 7129minus7138(14) Nielsen L M Holm A I S Varsano D Kadhane UHoffmann S V Di Felice R Rubio A Nielsen S B J Phys ChemB 2009 113 9614minus9619(15) Tonzani S Schatz G C J Am Chem Soc 2008 130 7607minus7612(16) Casida M E Salahub D R J Chem Phys 2000 113 8918minus8935(17) Dreuw A Weisman J L Head-Gordon M J Chem Phys2003 119 2943minus2946(18) Maitra N T Zhang F Cave R J Burke K J Chem Phys2004 120 5932minus5937(19) McCann D M Stephens P J J Org Chem 2006 71 6074minus6098(20) Casida M E J Mol Struct 2009 914 3minus18(21) Niehaus T A J Mol Struct 2009 914 38minus49(22) Tawada Y Tsuneda T Yanagisawa S Yanai T Hirao K JChem Phys 2004 120 8425minus8433(23) Chiba M Tsuneda T Hirao K J Chem Phys 2006 124144106-1-11(24) Nakatsuji H Hirao K J Chem Phys 1978 68 2053minus2065(25) Nakatsuji H Chem Phys Lett 1978 59 362minus364(26) Nakatsuji H Chem Phys Lett 1979 67 329minus333(27) Nakatsuji H Chem Phys Lett 1979 67 334minus342(28) Nakatsuji H Bull Chem Soc Jpn 2005 78 1705minus1724(29) Ehara M Hasegawa J Nakatsuji H SAC-CI Method Appliedto Molecular Spectroscopy In Theory and Applications of Computa-tional Chemistry The First 40 Years A Vol of Technical and HistoricalPerspectives Dykstra C E Frenking G Kim K S Scuseria G EEds Elsevier Oxford UK 2005 pp 1099minus1141(30) Miyahara T Hasegawa J Nakatsuji H Bull Chem Soc Jpn2009 82 1215minus1226(31) Bureekaew S Hasegawa J Nakatsuji H Chem Phys Lett2006 425 367minus371(32) Nakatani N Hasegawa J Nakatsuji H J Am Chem Soc2007 129 8756minus8765(33) Fujimoto K Hasegawa J Nakatsuji H Bull Chem Soc Jpn2009 82 1140minus1148(34) Miyahara T Nakatsuji H Collect Czech Chem Commun2011 76 537minus552(35) Sugiyama H Unpublished(36) Geertsen J Rittby M Bartlett R J Chem Phys Lett 1989164 57minus62

(37) Stanton J F Bartlett R J J Chem Phys 1993 98 7029minus7039(38) Ehara M Ishida M Toyota K Nakatsuji H SAC-CIGeneral-R Method Theory and Applications to the Multi-ElectronProcesses In Reviews in Modern Quantum Chemistry (A Tribute toProfessor Robert G Parr) Sen K D Ed World Scientific Singapore2003 pp 293minus319(39) SAC-CI GUIDE httpwwwqcriorjpsacci(40) Frisch M J Trucks G W Schlegel H B et al Gaussian 03development revision revision E05 Gaussian Inc Wallingford CT2004(41) Hohenberg P Kohn W Phys Rev 1964 136 B864minusB871(42) Kohn W Sham L J Phys Rev 1965 140 A1133minusA1138(43) Salahub D R Zerner M C Eds The Challenge of d and fElectrons The American Chemical Society Washington DC 1989(44) Parr R G Yang W Density-Functional Theory of Atoms andMolecules Oxford University Press Oxford UK 1989(45) Becke A D J Chem Phys 1993 98 5648minus5652(46) Lee C Yang W Parr R G Phys Rev 1988 37 785minus789(47) Hariharan P C Pople J A Theor Chim Acta 1973 28 213minus222(48) Francl M M Pietro W J Hehre W J Binkley J SDeFrees D J Pople J A Gordon M S J Chem Phys 1982 773654minus3665(49) Tomasi J Mennucci B Cammi R Chem Rev 2005 1052999minus3093(50) Fukuda R Ehara M Nakatsuji H Cammi R J Chem Phys2011 134 104109minus1minus11(51) Dunning T H Jr Hay P J In Modern Theoretical ChemistrySchaefer H F III Ed Plenum New York 1976 Vol 3 pp 1minus28(52) Nakatsuji H Chem Phys 1983 75 425minus441(53) Hocquet A Leulliot N Ghomi M J Phys Chem B 2000 1044560minus4568(54) Mishra S K Mishra P C J Comput Chem 2002 23 530minus540(55) Sprecher C A Johnson W C Biopolymers 1977 16 2243minus2264(56) Clark L B Peschel G G Tinoc I Jr J Phys Chem 1965 693615minus3618(57) Voet D Gratzer W B Cox R A Doty P Biopolymers 19631 193minus208(58) Hobza P Mehlhorn A Carsky P Zahradnik R J Mol Struct1986 138 387minus399(59) Hobza P Sponer J Chem Rev 1999 99 3247minus3276

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5555

Page 7: Helical Structure and Circular Dichroism Spectra of DNA: A ... · B-DNA, have right-handed double-helical structure, while Z-DNA is left-handed and appears only when DNA has a special

spectrum is similar to that of dC for both B- and Z-DNA As aresult in the composite CD spectra of B- and Z-DNA the signat around 295 nm is independent of the geometry of themonomer (dG or dC) even though the CD spectrum of dGdepends on the dihedral angle between the guanine and thedeoxyribose because the lowest peak originates from dCwhose geometry is always anti in both B- and Z-DNA Thisindicates that the noninteracting dG and dC cannot explain theexperimental spectra of DNA and that the hydrogen-bondingor stacking interactions with the neighboring nucleic acid basesmust be important for the CD spectra of DNA34 UV and CD Spectra of Dimer Models In the studies

of the structure spectroscopy and functions of DNA it isgenerally recognized that the hydrogen-bonding interaction andthe stacking interaction are the two most important interactionsthat characterize the properties of DNA So we expect thatthese interactions might also be the important factors that affectthe features of the UV and CD spectra of DNA To investigatethe possible origin we considered the dimer model composedof dG and dC These models contained either the hydrogen-bonding interaction or the stacking interaction (Figure 1)However we note that the SAC-CI results using these modelsneed not be in agreement with the experimental spectrabecause these models lack either the stacking interaction or thehydrogen-bonding interaction We must use the tetramer

model at least to compare with the experimental spectra asnoted below

341 UV Spectra of Dimer Models Figure 6 shows theSAC-CI UV spectra of both hydrogen-bonding and stackingmodels of both B- and Z-DNA compared with the experimentalUV spectrum The experimental UV spectra of Z-DNA displaya main peak at 256 nm and a shoulder peak at around 290 nmThe experimental UV spectrum of B-DNA is very similar butthe shoulder peak at around 290 nm is weaker Table 4 showsthe excitation energies oscillator strengths rotatory strengthsnatures and the types of the excited states of the SAC-CIresults for both hydrogen-bonding and stacking models of thedimer When the excitation in the dimer model is not withindG or dC but corresponds to the electron transfer (ET)between dG and dC we assigned it as the ET typeThe SAC-CI UV spectrum of the hydrogen-bonding model

of Z-DNA (Z-Hbond model) is in good agreement with theexperimental UV spectrum of Z-DNA (Figure 6a) Threestrong peaks of the πminusπ nature are calculated The loweststrong peak corresponds to the shoulder peak at 428 eV (290nm) of the experimental UV spectrum of Z-DNA Theremaining two strong peaks correspond to the main peak at484 eV (256 nm) of the experimental UV spectrum of Z-DNARed blue and green colors represent the excited states of dGdC and ET respectively

Table 3 SAC-CI UV and CD Spectra of dG and dC in Z- and B-DNA

Z-DNA B-DNA

dG (model I φ = 1113deg) dG (model I φ = 2763deg)

EEa oscb rotc EEa oscb rotc

state (eV) (nm) (au) (10minus40 cgs) nature (eV) (nm) (au) (10minus40 cgs) nature

11A 441 281 017 125 π rarr π 443 280 016 minus1035 π rarr π21A 474 262 000 minus1099 n rarr π 488 254 000 1099 n rarr π31A 493 251 023 2066 π rarr π 513 242 025 minus3355 π rarr π41A 578 214 000 minus864 n rarr π 595 208 002 minus2619 π rarr π51A 591 210 004 minus656 π rarr σ 599 207 001 3571 n rarr π61A 611 203 000 minus438 n rarr π 618 201 000 983 n rarr π71A 628 197 000 359 π rarr σ 675 184 008 1235 π rarr σ81A 659 188 030 634 π rarr π 678 183 028 minus2402 π rarr π

dG (model II φ = 1175deg) dG (model II φ = 3007deg)

EEa oscb rotc EEa oscb rotc

state (eV) (nm) (au) (10minus40 cgs) nature (eV) (nm) (au) (10minus40 cgs) nature

11A 437 284 016 minus451 π rarr π 434 286 016 minus247 π rarr π21A 459 270 000 minus322 n rarr π 493 251 000 minus1467 n rarr π31A 491 252 022 minus319 π rarr π 518 239 021 minus200 π rarr π41A 581 213 003 323 π rarr π 595 209 001 2884 n rarr π51A 598 207 000 minus738 π rarr π 599 207 003 minus647 π rarr π61A 621 200 000 796 n rarr π 617 201 000 minus3033 n rarr π71A 641 193 000 353 π rarr σ 660 188 048 978 π rarr π81A 647 191 031 minus077 π rarr π 666 186 001 326 π rarr σ

dC (φ = 3329deg) dC (φ = 3232deg)

EEa oscb rotc EEa oscb rotc

state (eV) (nm) (au) (10minus40 cgs) nature (eV) (nm) (au) (10minus40 cgs) nature

11A 413 300 009 2385 π rarr π 409 303 009 1403 π rarr π21A 442 280 000 minus1548 n rarr π 452 274 000 minus915 n rarr π31A 507 245 011 8392 π + n rarr π 499 249 007 5975 π + n rarr π41A 527 235 011 minus8639 n + π rarr π 527 235 015 minus8017 π + n rarr π51A 573 216 000 minus1385 n rarr π 559 222 001 393 n rarr π61A 632 196 039 2349 π rarr π 625 198 032 1098 π rarr π

aExcitation energy bOscillator strength cRotatory strength

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5548

The shoulder peak in the experimental UV spectrum of Z-DNA is composed of 11A (439 eV) and 21A (453 eV) statesThe 11A state (in blue color) of the Z-Hbond modelcorresponds to the 11A state of dC at 413 eV which is theexcitation from the HOMO to the LUMO of dC This state haslarge oscillator strength of the πminusπ nature In the Z-Hbondmodel this state is blue-shifted by 026 eV because theHOMOminusLUMO gap (0449 au) of the Z-Hbond model islarger than that (0438 au) of dC This state is in goodagreement with the first (shoulder) peak of the experimentalUV spectrum of Z-DNA However the 21A state (453 eV redcolor) of the Z-Hbond model is assigned to the 31A state (493eV) of dG which is the excitation from the HOMO to the next-LUMO of dG This large shift of the excitation energy is causedby the MO-order change The MO of the Z-Hbond modelcorresponding to the next-LUMO of dG becomes lower thanthe MO of the Z-Hbond model corresponding to the LUMOof dG by the formation of hydrogen bonds with cytosine in theZ-Hbond model Therefore this state is red-shifted by 04 eVHowever the oscillator strength of this state is small despite itsπminusπ nature The transition moment is canceled because themain configuration includes not only the excitation from theHOMO to the next-LUMO in dG but also the excitation fromthe next-HOMO to the LUMO in dC and the electron transfertype excitation from the HOMO of dG to the LUMO and next-LUMO of dCThe main peak of the experimental UV spectrum of Z-DNA

is composed of 31A (484 eV) and 51A (493 eV) states havinglarge oscillator strengths (Figure 6a) The 31A state (so blue

color) of the Z-Hbond model is assigned to the 31A state (507eV) of dC which reflects the strong excitation from the next-HOMO to the LUMO of dC The 51A state (red color) of theZ-Hbond model is assigned to the 11A state (441 eV) of dGwhich is the excitation from the HOMO to the LUMO in dGThe excitation energy increases by 052 eV because theHOMOminusLUMO gap becomes large which is the opposite tothat of the 21A state but again due to the hydrogen-bondingeffectFor B-DNA however the SAC-CI UV spectrum of the

hydrogen-bonding model (B-Hbond model Figure 6c) also hasboth main and shoulder peaks in which the main peak is higherand the shoulder peak is lower as compared with theexperimental UV spectrum of B-DNA The main peak (41Aand 51A states) of the SAC-CI spectra of the B-Hbond model ishigher than that of the experimental spectra similarly to theSAC-CI spectrum of the Z-Hbond model The 31A state of dGin B-DNA is also higher than that in Z-DNA (Table 3) Theseresults may be related to the fact that dG is more flexible thandC because the conformation of dG is anti for B-DNA but synfor Z-DNA Since the stacking interaction of B-DNA is weakerthan that of Z-DNA the X-ray structure of B-DNA may bemore flexible than that of Z-DNA However the result that theoscillator strength of the lowest excited state of the B-Hbondmodel is weaker than that of the Z-Hbond model is reflected bythe fact that the shoulder peak of the experimental UV spectrais stronger in Z-DNA but weaker in B-DNAFor the SAC-CI UV spectra of the stacking models of both

B- (B-Stack model) and Z-DNA (Z-Stack model) the firstexcited states originated from dC were calculated at the lower-energy region (higher wavelength region) but the excitationenergy of the main peak is almost the same between the B-Hbond and B-Stack models or between the Z-Hbond and Z-Stack models If there is no hydrogen-bonding interaction inDNA the lowest excited state would be observed in the lower-energy region of the experimental spectra separately from themain peak The excitation energy of the lowest excited state ofthe stacking model is lower than that of the hydrogen-bondingmodel by 044 eV for Z-DNA and 033 eV for B-DNArespectively This result may correspond to the fact that thelowest excitation energy of cytosine differs between the gasphase (428 eV)56 and in solution (464 eV)57

For both B- and Z-DNA the SAC-CI results of thehydrogen-bonding models (B-Hbond and Z-Hbond) are inbetter agreement with the experimental spectra than those ofthe stacking models (B-Stack and Z-Stack) Thus thehydrogen-bonding interaction is reflected in the UV spectraof DNA On the other hand the shoulder peak observed for Z-DNA may reflect the stronger shoulder nature of the Z-Stackmodel in Z-DNA However the UV spectra of DNA providelittle information about stacking interactions between nucleicacid bases

342 CD Spectra of Dimer Models Figure 7 shows theSAC-CI CD spectra of both hydrogen-bonding and stackingmodels of B- and Z-DNA compared with the experimental CDspectra In the experimental CD spectra the first peak isnegative at 296 nm and the second peak is positive at 275 nmfor Z-DNA However B-DNA displays peaks that are oppositein sign the first peak is positive at 280 nm and the second peakis negative at 255 nmIt is worth nothing that the SAC-CI spectrum has a negative

peak at 295 nm only for the stacking model of Z-DNA (Z-Stack) but for the other models the SAC-CI spectrum is

Figure 6 SAC-CI UV spectra for the dimer models The SAC-CIspectra (magenta line) of (a) Z-Hbond model (b) Z-Stack model (c)B-Hbond model and (d) B-Stack model are compared with theexperimental UV spectra4 (black line) of Z-DNA (ab) and B-DNA(cd) The red and blue lines represent the intramolecular excitedstates of dG and dC respectively The green lines represent theelectron transfer (ET) excited state from dG to dC

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5549

Table

4SA

C-CIUVandCD

Spectraof

theHydrogen-Bon

ding

andStacking

Dim

erMod

elsof

Z-andB-D

NA

Z-DNA

B-DNA

SAC-CIof

Z-H

bond

exptle

SAC-CIof

B-H

bond

exptle

EEa

oscb

rotc

UV

EEa

oscb

rotc

UV

state

(eV)

(nm)

(au)

(10minus

40cgs)

nature

typed

(eVnm)

(eV)

(nm)

(au)

(10minus

40cgs)

nature

Typed

(eVnm)

11A

439

282

020

3513

πrarr

πdC

(11 A)

428290

437

284

017

1143

πrarr

πdC

(11 A)

21A

453

274

004

884

πrarr

πdG

(31 A)

456

272

003

minus1481

πrarr

πdG

(31 A)

31A

484

256

018

912

πrarr

πdC

(31 A)

493

251

006

583

πrarr

πET

41A

485

255

005

minus1145

πrarr

πET

498

249

021

1767

πrarr

πdC

(3+41A)

51A

493

252

042

1182

πrarr

πdG

(11 A)

484256

518

240

043

minus1390

πrarr

πdG

(11 A)

484256

61A

512

242

000

057

nrarr

πdC

(21 A)

525

236

000

minus1197

nrarr

πdC

(21 A)

71A

523

237

000

minus1384

nrarr

πdG

(21 A)

529

234

000

minus579

nrarr

πdG

(2+51A)

81A

623

199

000

minus034

nrarr

πdG

(2+51A)

SACminusCIof

Z-Stack

exptle

SACminusCIof

B-Stack

exptle

EEa

oscb

rotc

CD

EEa

oscb

rotc

CD

state

(eV)

(nm)

(au)

(10minus

40cgs)

nature

typed

(eVnm)

(eV)

(nm)

(au)

(10minus

40cgs)

nature

typed

(eVnm)

11A

395

314

009

minus4511

πrarr

πdC

(11 A)

419296

404

307

007

4358

πrarr

πdC

(11 A)

443280

21A

421

294

000

minus3349

πrarr

πdG

(11 A)

419296

440

282

006

minus1032

πrarr

πdG

(11 A)

31A

445

279

001

2967

nrarr

πdC

(21 A)

461

269

000

minus503

nrarr

πdC

(21 A)

41A

472

263

000

minus1335

nrarr

πdG

(21 A)

493

251

013

9407

πrarr

πdC

(3+41A)

51A

485

256

019

minus1796

πrarr

πdG

(31 A)

451275

507

245

005

minus15115

nrarr

πdG

(21 A)

486255

61A

499

248

006

4861

πrarr

πET

451275

512

242

014

minus11056

πrarr

πdG

(31 A)

486255

71A

526

236

007

2721

πrarr

πET

+dC

(31 A)

451275

521

238

014

19475

πrarr

πET+dG

(31 A)

81A

565

219

000

minus231

nrarr

πdC

(41 A)

619

200

000

minus805

nrarr

πdG

(41 A)

aExcitatio

nenergybOscillator

strengthcRotatorystrengthddG

representsintram

olecular

excitedstates

ofdG

dCrepresentsintram

olecular

excitedstates

ofdC

and

ETrepresentselectron

transfer

excitedstates

from

dGto

dCeRef

4

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5550

positive at 295 nm Therefore the strong negative sign of theCD spectra at 295 nm indicates that a strong stackinginteraction exists in Z-DNA The SAC-CI CD spectra of bothstacking models (B-Stack and Z-Stack models) more closelyresemble the experimental CD spectra of B- and Z-DNA thanthose of the hydrogen-bonding models (B-Hbond and Z-Hbond models) For the SAC-CI CD spectrum of the Z-Stackmodel the first and second peaks are negative and positiverespectively but they are the opposite of those obtained fromthe Z-Hbond model For the B-Stack and B-Hbond modelshowever the first peak is positive in the SAC-CI CD spectrabut the second and third peaks are negative and positiverespectively for the B-Stack model in opposition to the B-Hbond model Since the stacking models do not have thehydrogen-bonding interactions the excitation energy of the firstpeak is lower than that of the experimental spectrum as notedaboveIn the Z-Stack model the first peak with a negative sign is

assigned to the 11A (395 eV) and 21A (421 eV) states (Table4) The 11A state of the Z-Stack model corresponds to the 11Astate (413 eV) of dC which is the excitation of the πminusπ fromthe HOMO to the LUMO in dC A large oscillator strength isthe origin of the first (shoulder) peak but the excitation energyis lower than the experimental value because there is nohydrogen bond in the Z-Stack model The rotatory strength ofthe 11A state of dC changes the sign from positive to negativeby the stacking interaction with dG Therefore the stackinginteraction between dG and dC accounts for the strongnegative sign of the CD spectra in Z-DNA

The excitation energy of the 21A state (421 eV) is also closeto the first peak of the experimental CD spectrum This state isassigned to the 11A state (437 eV) of dG which is the HOMOto LUMO πminusπ excitation in dG but also includes thecontribution of the excitation from the next-HOMO to LUMOin dC The oscillator strength in the UV spectrum is smallbecause this state differs from the 11A state of dG by theinclusion of the excitation in dC Although the excitationenergy is shifted by the formation of hydrogen bonds betweendG and dC the rotatory strength of a large negative maycontribute to the strong negative sign of the CD spectra in Z-DNA In the 31A state (445 eV) which is the excitation fromnonbonding orbital to the LUMO in dC the rotatory strengthchanges to positive from the negative sign of the monomer dCHowever this state is not the origin of the second peak as willbe noted later in the tetramer model section Thus the rotatorystrength is strongly affected by the stacking interactionThree important excited states are calculated in the higher-

energy region of the second peak The 51A state (485 eV) hasthe largest oscillator strength and is assigned to the 31A (491eV) state of dG which is the excitation from HOMO to next-LUMO This state is assigned to the main peak (484 eV) of theexperimental UV spectrum of Z-DNA The 6 and 71A states(499 and 526 eV) with a large rotatory strength are the ETstate from the HOMO of dG to the LUMO of dC The mainconfiguration of the 71A state also includes the excitation fromthe next-HOMO to the LUMO in dCIn the B-Stack model the first peak with a positive sign is

assigned to the 11A state (404 eV) This state corresponds tothe 11A state (409 eV) of dC which is the excitation from theHOMO to the LUMO The 21A state (440 eV) is assigned tothe 11A state (434 eV) of dG which reflects the excitationfrom the HOMO to the LUMO Since the B-Stack model doesnot have hydrogen bonds the excitation energies are close tothose of dG or dC Both states have the oscillator strength butthe 11A state is positive while the 21A state is slightly negativefor the rotatory strength which is the same as for themonomers dC and dG respectively Therefore the stackinginteraction of B-DNA has little effect on the sign of the CDspectra at 295 nm because the stacking interaction is weaker inB-DNA than in Z-DNAThe second peak with a negative sign is assigned to the 5 and

61A states with a large rotatory strength The 5 and 61A states(507 and 512 eV in Table 4) are assigned to the 2 and 31Astates (493 and 518 eV in Table 3) of dG which are theexcitations from the nonbonding orbital and the HOMO to thenext-LUMO The main configuration is nminusπ for the 51A stateand πminusπ for the 61A state Both states have strong oscillatorand negative rotatory strengths which means that they can beassigned to the main peaksThe first ET excited states (71A state 521 eV) of B-Stack

model are calculated at a higher region than that (61A state499 eV) of the Z-Stack model because the stacking interactionis weaker in B-DNA than in Z-DNAThe SAC-CI results show that both hydrogen-bonding and

stacking interactions affect the UV and CD spectra of DNA Inparticular the excitation energy of the first excited stateincreases by the formation of hydrogen bonds and the stackinginteraction changes the sign of the CD spectrum of the firstexcited state from positive to negative The strong stackinginteraction characteristic of the helical structure of Z-DNA isthe origin of the strong negative sign at around 295 nm in theCD spectra The hydrogen-bonding and stacking interactions

Figure 7 SAC-CI CD spectra for the dimer models The SAC-CIspectra (magenta line) of (a) Z-Hbond model (b) Z-Stack model (c)B-Hbond model and (d) B-Stack model are compared with theexperimental CD spectra4 (black line) of Z-DNA (ab) and B-DNA(cd) The red and blue lines represent the intramolecular excitedstates of dG and dC respectively The green lines represent theelectron transfer (ET) excited state from dG to dC

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5551

also confer other smaller effects on the UV and CD spectra ofDNA In the higher energy region the SAC-CI CD spectra arenot in good agreement with the experimental CD spectrabecause these models have either hydrogen-bonding interactionor stacking interaction but not both Therefore we need toinvestigate a better model that includes both hydrogen bondingand stacking interactions within the same model A minimumsuch model is the tetramer model of Figure 1 on which weexplain in the next section35 Tetramer Model The SAC-CI calculations of the

dimer models were suitable to investigate the effects of thehydrogen-bonding interaction and the stacking interactionseparately For example the SAC-CI UV spectra of thehydrogen-bonding models were in good agreement with theexperimental UV spectra implying the importance of thehydrogen-bonding interaction The SAC-CI CD spectra for thestacking model clarified that the stacking interaction is thesource of the negative CD peak at around 295 nm Howevergenerally speaking the SAC-CI CD spectra were markedlydifferent from the experimental spectra The SAC-CI UVspectra of the stacking models were composed of two peaksdifferent from the experimental UV spectra Thus theinformation that we gained from the dimer models still didnot fully resolve these differences in the spectra acquired by twodifferent processes So we used the tetramer model (Figure 1)which included both hydrogen-bonding and stacking inter-actions within the same modelFigure 8 shows the SAC-CI UV and CD spectra of the

tetramer model of Z-DNA (Z-Tetra model) and B-DNA (B-Tetra model) in comparison with the experimental UV andCD spectra In order to compare the experimental spectra theSAC-CI results were shifted to lower energy by 05 eV Thecomputational conditions were poor as compared with theother SAC-CI calculations In the tetramer calculations the d-polarization functions were not added and the small activespace was used The SAC-CI calculations using the small activespace and without the d-polarization functions gave higherexcitation energies by about 04minus05 eV (see SupportingInformation Table S1) for the hydrogen-bonding dimermodels In particular the excited states corresponding to themain peak of the experimental UV spectra were shifted by 048eV for both the Z-Hbond model (61A state) and the B-Hbondmodel (71A state)Table 5 shows the excitation energies oscillator strengths

rotatory strengths natures and type of the excited states of theSAC-CI results for tetramer modelsThe SAC-CI UV spectrum of the Z-Tetra model is in good

agreement with the experimental UV spectrum of Z-DNA Themain peak as well as the shoulder peak is reproduced by theSAC-CI method The shoulder peak is assigned to the 4 and51A states which is the πminusπ excitation (HOMO to LUMO)in dC and the main peak is assigned to the 101A state which isthe πminusπ (HOMO to next-LUMO) excitation in dGThe SAC-CI UV spectrum of the B-Tetra model is also in a

reasonable agreement with the experimental UV spectrum of B-DNA although the shape of the shoulder is slightly different Inthe B-Tetra model the shoulder peak is assigned to the 21Astate which is the πminusπ (HOMO to LUMO) excitation in dCand the main peak is assigned to the 81A state which is theπminusπ (HOMO to next-LUMO) excitation in dGThe sign of the rotatory strength is the same between the

SAC-CI CD spectrum of the Z-Tetra model and theexperimental CD spectrum of Z-DNA although the negative

rotatory strength at 295 nm is very weak because thepolarization functions were not included in the tetramermodels It is well-known that the basis set must be good for anadequate description of the stacking interaction5859 In thedimer model the lack of polarization d-functions and thesmaller active space changed the intensity and sign of therotatory strength (see Supporting Information Table S1)Since the stacking interaction is small in B-DNA the SAC-CI

CD spectrum of the B-Tetra model is in good agreement withthe experimental CD spectrum of B-DNA and reproduces notonly the sign of the rotatory strength but also the overall shapeof the spectrumIn Z-DNA the 11A state is assigned to the first peak of the

experimental CD spectrum The 11A state is the ET excitationfrom the HOMO of dG to the LUMO of dC through thestacking conformation The 3 and 51A states with a largenegative rotatory strength are the excitation from the HOMOto the LUMO in dG and dC respectively These states alsocontribute to the first negative peak as in the 1 and 21A stateswith a negative rotatory strength in the Z-Stack model The 8and 101A states are assigned to the second peak of the CDspectrum The 101A state which is the origin of the main peakof the UV spectrum and the πminusπ of dC (81A state) alsocontribute to the second peak of the CD spectrumIn B-DNA the first peak of the CD spectrum is assigned to

the 21A state of πminusπ (HOMO to LUMO) in dC which is thesame as the origin of the first peak of the UV spectrum The ETstates are calculated at the higher energy region than the 21A

Figure 8 SAC-CI UV and CD spectra for the tetramer models TheSAC-CI spectra (magenta line) of Z-Tetra model (ab) and B-Tetramodel (cd) as compared with the experimental UV and CD spectra4

(black line) of Z-DNA (ab) and B-DNA (cd) The red and blue linesrepresent the intramolecular excited states of dG and dC respectivelyThe green lines represent the electron transfer (ET) excited state fromdG to dC All excited states of SAC-CI calculations have been shiftedto the lower values by 05 eV

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5552

Table

5SA

C-CIUVandCD

Spectraof

theTetramer

Mod

elsof

Z-andB-D

NA

Z-DNA

B-DNA

SACminusCIof

Z-Tetra

exptle

SAC-CIof

B-Tetra

exptle

EEa

Oscb

Rotc

UV

CD

EEa

Oscb

Rotc

UV

CD

state

(eV)

(nm)

(au)

(10minus

40cgs)

nature

typed

(eVnm)

(eVnm)

(eV)

(nm)

(au)

(10minus

40cgs)

nature

typed

(eVnm)

(eVnm)

11A

470

264

000

minus663

πrarr

πET

419296

479

259

003

minus041

πrarr

πdG

21A

481

258

002

6611

πrarr

πET

482

257

033

16891

πrarr

πdC

443280

31A

484

256

004

minus5326

πrarr

πdG

419296

484

256

001

minus7454

πrarr

πdC

41A

488

254

015

4553

πrarr

πdC

428290

498

249

007

minus130

πrarr

πET

51A

492

252

012

minus11712

πrarr

πdC

428290

419296

511

243

004

minus622

πrarr

πET

61A

502

247

003

minus3034

πrarr

πET

525

236

000

minus046

πrarr

πET

71A

513

242

001

2737

nrarr

πdG

527

235

001

332

πrarr

πET

81A

515

241

006

14772

πrarr

πdC

451275

545

227

045

minus15609

πrarr

πdG

484256

486255

91A

522

238

001

1978

nrarr

πdG

552

225

002

minus1795

nrarr

πdG

101 A

538

230

071

7299

πrarr

πdG

484256

451275

556

223

013

5929

πrarr

πdG

111 A

541

229

001

minus6858

πrarr

πdG

558

222

009

minus525

πrarr

πdG

121 A

548

226

002

minus1519

nrarr

πdC

563

220

008

minus7436

nrarr

πdG

131 A

553

224

001

minus924

nrarr

πdC

609

204

001

1379

nrarr

πdG

141 A

601

206

001

minus4159

nrarr

πdG

626

198

002

minus464

nrarr

πdG

aExcitatio

nenergyA

llexcitedstates

ofSA

C-CIcalculations

have

been

shifted

tothelower

values

by05eV

inFigure

8bOscillator

strengthcRotatorystrengthddG

representsintram

olecular

excited

states

ofdG

dCrepresentsintram

olecular

excitedstates

ofdC

and

ETrepresentselectron

transfer

excitedstates

from

dGto

dCeRef

4

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5553

state because the stacking interaction is weaker in B-DNA thanin Z-DNA (Figure 1) Therefore the first peak of theexperimental CD spectra originates from the ET excitationfrom dG to dC for Z-DNA but corresponds to theintramolecular excitation in dC for B-DNA The second peakwith a negative sign is assigned to the 81A state The 81A statewhich is the πminusπ of dG is the origin of both the main peaks ofthe UV spectrum and the second peak of the CD spectrumThe ET excited states of the tetramer models were calculated

at a lower energy region than those of the dimer models forboth B- and Z-DNA This result may be related to the overlapbetween the nucleic acid bases The overlap exists only in partof the nucleic acid bases in the dimer models but in thetetramer models the overlap covers the hydrogen bonds as wellas the nucleic acid bases The ET excited states are importantfor the CD spectrum of Z-DNA because the lowest excitedstate is the ET type and is negative for rotatory strength but forB-DNA the ET excited states are not important because thesestates are weak in both oscillator and rotatory strengthsThe SAC-CI results of the tetramer models are calculated at

a higher region by about 05 eV because of the small activespace Therefore better SAC-CI calculations are required forcomparison with the experimental UV and CD spectra of DNAHowever these results indicate that the strong stackinginteraction characteristic of the helical structure of Z-DNAchanges the sign of the rotatory strength for the lowest excitedstates of dG and dC and produces the emergence of the ETexcited states

4 CONCLUSIONSDNA chemistry is affected by three important factors (1) theconformation of the monomer (dG) which is related to thedihedral angle between deoxyribose and guanine (2) thehydrogen-bonding interactions and (3) the stacking inter-actions between two nucleic acid bases To verify whether thesethree factors are important for the CD spectra of DNA weexamined four different models using the SAC-CI method(1) dG monomer model We first applied the SAC-CI

method to the dG monomer to examine the effect ofconformation on the CD spectra of dG and from this resultto identify the structure of the dG monomer in solution Ourcalculations showed that the shape and the sign of the CDspectra depend strongly on the dihedral angle between thedeoxyribose and the guanine but that the SAC-CI UV spectraof both anti- and syn-dG are very similar because the UVspectra are not strongly dependent on the conformation Theexperimental CD spectrum agreed with the theoreticalspectrum of dG only for the anti conformation The resultingthermally averaged spectra also support the anti geometry ofdG as a stable conformation in solution However the SAC-CICD spectrum of syn-dG is opposite in sign from that of theexperimental spectrum Thus the SAC-CI CD spectra of dG inboth anti and syn conformations do not explain the negativesign at 295 nm of the CD spectra of the DNA composed of dGand dC(2) Composit model of dG and dC monomers A

comparison of the calculated composite spectra of dG anddC with the experimental spectra of the two-component DNAshowed similarities between B- and Z-DNA because theconformation of dC is always anti in both B- and Z-DNAThe SAC-CI results led us to confirm that that theconformation of dG is not the origin of the negative peak at295 nm in Z-DNA

(3) Dimer model Our examination of the dimer modelpermitted us to investigate the separate importance of thehydrogen-bonding and the stacking interactions in thestructural chemistry of B- and Z-DNA The SAC-CIcalculations showed that the hydrogen-bonding interactionschange the excitation energies and that the stacking interactionsare responsible for the changes in the sign of the CD rotatorystrength The SAC-CI UV spectra of the hydrogen-bondingmodels (B-Hbond and Z-Hbond models) are similar to theexperimental UV spectra in which the main peak as well as theshoulder peak is well reproduced The SAC-CI CD spectra ofthe stacking models (B-Stack and Z-Stack models) have thesame sign as the experimental CD spectra We concluded thatthe stacking interaction is especially important for the CDspectrum of the strongly stacked Z-DNA because the CD signalat 295 nm changed to a negative sign only for the Z-Stackmodel(4) Tetramer model Finally our calculations with the

tetramer model that contain both hydrogen-bonding andstacking interactions revealed that the shape of the SAC-CIUV and CD spectra compared well with the experimental onesThe ET excitations as well as the lowest intramolecularexcitation of dC and dG are the origins of the negative sign ofthe experimental CD spectrum of Z-DNA The ET excited stateis calculated at a lower energy region than the intramolecularexcited states of dG or dC for the strongly stacked Z-DNA butat a higher energy region for the weakly stacked B-DNA Thenegative sign of the first peak of the experimental CD spectra ofZ-DNA is due to the strong stacking effect characteristic of itshelical structureThe dynamic behavior of the dG monomer in solution was

partially considered by taking the Boltzmann averaging of thecalculated spectra over the various conformations Fortunatelythe Boltzmann averaged SAC-CI CD spectrum was in goodagreement with the experimental spectra although the SAC-CICD spectra largely depend on the conformation of dGHowever in this study we used only the static model for dimerand tetramer models The SAC-CI CD results of the dimer andtetramer models explained the difference between theexperimental CD spectra of B- and Z-DNA However sincethe structure of DNA would also be flexible in solution it isnecessary to take into account the dynamic behavior of DNAThis is an important and interesting topic

ASSOCIATED CONTENT

S Supporting InformationSAC-CI results of the hydrogen-bonding and stacking dimermodels of Z- and B-DNA using small active space and withoutthe polarization d-functions This material is available free ofcharge via the Internet at httppubsacsorg

AUTHOR INFORMATION

Corresponding AuthorPhone +81-75-634-3211 Fax +81-75-634-3211 E-mail hnakatsujiqcriorjp

NotesThe authors declare no competing financial interest

ACKNOWLEDGMENTS

The computations were performed at the Research Center forComputational Science Okazaki Japan

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5554

REFERENCES(1) Voet D Voet J G Eds Biochemistry John Wiley amp Sons IncNew York 1995(2) Ghosh A Bansal M Acta Crystallogr 2003 D59 620minus626(3) Rich A Nordheim A Wang A H-J Annu Rev Biochem 198453 791minus846(4) Tran-Dinh S Taboury J Neumann J-M Huynh-Dinh TGenissel B Langlois drsquoEstaintot B Igolen J Biochemistry 1984 231362minus1371(5) Sugiyama H Kawai K Matsunaga A Fujimoto K Saito IRobinson H Wang A H-J Nucleic Acids Res 1996 24 1272minus1278(6) Kawai K Saito I Sugiyama H J Am Chem Soc 1999 1211391minus1392(7) Oyoshi T Kawai K Sugiyama H J Am Chem Soc 2003 1251526minus1531(8) Xu Y Ikeda R Sugiyama H J Am Chem Soc 2003 12513519minus13524(9) Berova N Nakanishi K Woody R W Eds Circular DichroismPrinciples and Applications Wiley-VCH New York 2000(10) Bauernschmitt R Ahlrichs R Chem Phys Lett 1996 256454minus464(11) Casida M E Jamorski C Casida K C Salahub D R JChem Phys 1998 108 4439minus4449(12) Tsolakidis A Kaxiras E J Phys Chem A 2005 109 2373minus2380(13) Varsano D Di Felice R Marques M A L Rubio A J PhysChem B 2006 110 7129minus7138(14) Nielsen L M Holm A I S Varsano D Kadhane UHoffmann S V Di Felice R Rubio A Nielsen S B J Phys ChemB 2009 113 9614minus9619(15) Tonzani S Schatz G C J Am Chem Soc 2008 130 7607minus7612(16) Casida M E Salahub D R J Chem Phys 2000 113 8918minus8935(17) Dreuw A Weisman J L Head-Gordon M J Chem Phys2003 119 2943minus2946(18) Maitra N T Zhang F Cave R J Burke K J Chem Phys2004 120 5932minus5937(19) McCann D M Stephens P J J Org Chem 2006 71 6074minus6098(20) Casida M E J Mol Struct 2009 914 3minus18(21) Niehaus T A J Mol Struct 2009 914 38minus49(22) Tawada Y Tsuneda T Yanagisawa S Yanai T Hirao K JChem Phys 2004 120 8425minus8433(23) Chiba M Tsuneda T Hirao K J Chem Phys 2006 124144106-1-11(24) Nakatsuji H Hirao K J Chem Phys 1978 68 2053minus2065(25) Nakatsuji H Chem Phys Lett 1978 59 362minus364(26) Nakatsuji H Chem Phys Lett 1979 67 329minus333(27) Nakatsuji H Chem Phys Lett 1979 67 334minus342(28) Nakatsuji H Bull Chem Soc Jpn 2005 78 1705minus1724(29) Ehara M Hasegawa J Nakatsuji H SAC-CI Method Appliedto Molecular Spectroscopy In Theory and Applications of Computa-tional Chemistry The First 40 Years A Vol of Technical and HistoricalPerspectives Dykstra C E Frenking G Kim K S Scuseria G EEds Elsevier Oxford UK 2005 pp 1099minus1141(30) Miyahara T Hasegawa J Nakatsuji H Bull Chem Soc Jpn2009 82 1215minus1226(31) Bureekaew S Hasegawa J Nakatsuji H Chem Phys Lett2006 425 367minus371(32) Nakatani N Hasegawa J Nakatsuji H J Am Chem Soc2007 129 8756minus8765(33) Fujimoto K Hasegawa J Nakatsuji H Bull Chem Soc Jpn2009 82 1140minus1148(34) Miyahara T Nakatsuji H Collect Czech Chem Commun2011 76 537minus552(35) Sugiyama H Unpublished(36) Geertsen J Rittby M Bartlett R J Chem Phys Lett 1989164 57minus62

(37) Stanton J F Bartlett R J J Chem Phys 1993 98 7029minus7039(38) Ehara M Ishida M Toyota K Nakatsuji H SAC-CIGeneral-R Method Theory and Applications to the Multi-ElectronProcesses In Reviews in Modern Quantum Chemistry (A Tribute toProfessor Robert G Parr) Sen K D Ed World Scientific Singapore2003 pp 293minus319(39) SAC-CI GUIDE httpwwwqcriorjpsacci(40) Frisch M J Trucks G W Schlegel H B et al Gaussian 03development revision revision E05 Gaussian Inc Wallingford CT2004(41) Hohenberg P Kohn W Phys Rev 1964 136 B864minusB871(42) Kohn W Sham L J Phys Rev 1965 140 A1133minusA1138(43) Salahub D R Zerner M C Eds The Challenge of d and fElectrons The American Chemical Society Washington DC 1989(44) Parr R G Yang W Density-Functional Theory of Atoms andMolecules Oxford University Press Oxford UK 1989(45) Becke A D J Chem Phys 1993 98 5648minus5652(46) Lee C Yang W Parr R G Phys Rev 1988 37 785minus789(47) Hariharan P C Pople J A Theor Chim Acta 1973 28 213minus222(48) Francl M M Pietro W J Hehre W J Binkley J SDeFrees D J Pople J A Gordon M S J Chem Phys 1982 773654minus3665(49) Tomasi J Mennucci B Cammi R Chem Rev 2005 1052999minus3093(50) Fukuda R Ehara M Nakatsuji H Cammi R J Chem Phys2011 134 104109minus1minus11(51) Dunning T H Jr Hay P J In Modern Theoretical ChemistrySchaefer H F III Ed Plenum New York 1976 Vol 3 pp 1minus28(52) Nakatsuji H Chem Phys 1983 75 425minus441(53) Hocquet A Leulliot N Ghomi M J Phys Chem B 2000 1044560minus4568(54) Mishra S K Mishra P C J Comput Chem 2002 23 530minus540(55) Sprecher C A Johnson W C Biopolymers 1977 16 2243minus2264(56) Clark L B Peschel G G Tinoc I Jr J Phys Chem 1965 693615minus3618(57) Voet D Gratzer W B Cox R A Doty P Biopolymers 19631 193minus208(58) Hobza P Mehlhorn A Carsky P Zahradnik R J Mol Struct1986 138 387minus399(59) Hobza P Sponer J Chem Rev 1999 99 3247minus3276

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5555

Page 8: Helical Structure and Circular Dichroism Spectra of DNA: A ... · B-DNA, have right-handed double-helical structure, while Z-DNA is left-handed and appears only when DNA has a special

The shoulder peak in the experimental UV spectrum of Z-DNA is composed of 11A (439 eV) and 21A (453 eV) statesThe 11A state (in blue color) of the Z-Hbond modelcorresponds to the 11A state of dC at 413 eV which is theexcitation from the HOMO to the LUMO of dC This state haslarge oscillator strength of the πminusπ nature In the Z-Hbondmodel this state is blue-shifted by 026 eV because theHOMOminusLUMO gap (0449 au) of the Z-Hbond model islarger than that (0438 au) of dC This state is in goodagreement with the first (shoulder) peak of the experimentalUV spectrum of Z-DNA However the 21A state (453 eV redcolor) of the Z-Hbond model is assigned to the 31A state (493eV) of dG which is the excitation from the HOMO to the next-LUMO of dG This large shift of the excitation energy is causedby the MO-order change The MO of the Z-Hbond modelcorresponding to the next-LUMO of dG becomes lower thanthe MO of the Z-Hbond model corresponding to the LUMOof dG by the formation of hydrogen bonds with cytosine in theZ-Hbond model Therefore this state is red-shifted by 04 eVHowever the oscillator strength of this state is small despite itsπminusπ nature The transition moment is canceled because themain configuration includes not only the excitation from theHOMO to the next-LUMO in dG but also the excitation fromthe next-HOMO to the LUMO in dC and the electron transfertype excitation from the HOMO of dG to the LUMO and next-LUMO of dCThe main peak of the experimental UV spectrum of Z-DNA

is composed of 31A (484 eV) and 51A (493 eV) states havinglarge oscillator strengths (Figure 6a) The 31A state (so blue

color) of the Z-Hbond model is assigned to the 31A state (507eV) of dC which reflects the strong excitation from the next-HOMO to the LUMO of dC The 51A state (red color) of theZ-Hbond model is assigned to the 11A state (441 eV) of dGwhich is the excitation from the HOMO to the LUMO in dGThe excitation energy increases by 052 eV because theHOMOminusLUMO gap becomes large which is the opposite tothat of the 21A state but again due to the hydrogen-bondingeffectFor B-DNA however the SAC-CI UV spectrum of the

hydrogen-bonding model (B-Hbond model Figure 6c) also hasboth main and shoulder peaks in which the main peak is higherand the shoulder peak is lower as compared with theexperimental UV spectrum of B-DNA The main peak (41Aand 51A states) of the SAC-CI spectra of the B-Hbond model ishigher than that of the experimental spectra similarly to theSAC-CI spectrum of the Z-Hbond model The 31A state of dGin B-DNA is also higher than that in Z-DNA (Table 3) Theseresults may be related to the fact that dG is more flexible thandC because the conformation of dG is anti for B-DNA but synfor Z-DNA Since the stacking interaction of B-DNA is weakerthan that of Z-DNA the X-ray structure of B-DNA may bemore flexible than that of Z-DNA However the result that theoscillator strength of the lowest excited state of the B-Hbondmodel is weaker than that of the Z-Hbond model is reflected bythe fact that the shoulder peak of the experimental UV spectrais stronger in Z-DNA but weaker in B-DNAFor the SAC-CI UV spectra of the stacking models of both

B- (B-Stack model) and Z-DNA (Z-Stack model) the firstexcited states originated from dC were calculated at the lower-energy region (higher wavelength region) but the excitationenergy of the main peak is almost the same between the B-Hbond and B-Stack models or between the Z-Hbond and Z-Stack models If there is no hydrogen-bonding interaction inDNA the lowest excited state would be observed in the lower-energy region of the experimental spectra separately from themain peak The excitation energy of the lowest excited state ofthe stacking model is lower than that of the hydrogen-bondingmodel by 044 eV for Z-DNA and 033 eV for B-DNArespectively This result may correspond to the fact that thelowest excitation energy of cytosine differs between the gasphase (428 eV)56 and in solution (464 eV)57

For both B- and Z-DNA the SAC-CI results of thehydrogen-bonding models (B-Hbond and Z-Hbond) are inbetter agreement with the experimental spectra than those ofthe stacking models (B-Stack and Z-Stack) Thus thehydrogen-bonding interaction is reflected in the UV spectraof DNA On the other hand the shoulder peak observed for Z-DNA may reflect the stronger shoulder nature of the Z-Stackmodel in Z-DNA However the UV spectra of DNA providelittle information about stacking interactions between nucleicacid bases

342 CD Spectra of Dimer Models Figure 7 shows theSAC-CI CD spectra of both hydrogen-bonding and stackingmodels of B- and Z-DNA compared with the experimental CDspectra In the experimental CD spectra the first peak isnegative at 296 nm and the second peak is positive at 275 nmfor Z-DNA However B-DNA displays peaks that are oppositein sign the first peak is positive at 280 nm and the second peakis negative at 255 nmIt is worth nothing that the SAC-CI spectrum has a negative

peak at 295 nm only for the stacking model of Z-DNA (Z-Stack) but for the other models the SAC-CI spectrum is

Figure 6 SAC-CI UV spectra for the dimer models The SAC-CIspectra (magenta line) of (a) Z-Hbond model (b) Z-Stack model (c)B-Hbond model and (d) B-Stack model are compared with theexperimental UV spectra4 (black line) of Z-DNA (ab) and B-DNA(cd) The red and blue lines represent the intramolecular excitedstates of dG and dC respectively The green lines represent theelectron transfer (ET) excited state from dG to dC

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5549

Table

4SA

C-CIUVandCD

Spectraof

theHydrogen-Bon

ding

andStacking

Dim

erMod

elsof

Z-andB-D

NA

Z-DNA

B-DNA

SAC-CIof

Z-H

bond

exptle

SAC-CIof

B-H

bond

exptle

EEa

oscb

rotc

UV

EEa

oscb

rotc

UV

state

(eV)

(nm)

(au)

(10minus

40cgs)

nature

typed

(eVnm)

(eV)

(nm)

(au)

(10minus

40cgs)

nature

Typed

(eVnm)

11A

439

282

020

3513

πrarr

πdC

(11 A)

428290

437

284

017

1143

πrarr

πdC

(11 A)

21A

453

274

004

884

πrarr

πdG

(31 A)

456

272

003

minus1481

πrarr

πdG

(31 A)

31A

484

256

018

912

πrarr

πdC

(31 A)

493

251

006

583

πrarr

πET

41A

485

255

005

minus1145

πrarr

πET

498

249

021

1767

πrarr

πdC

(3+41A)

51A

493

252

042

1182

πrarr

πdG

(11 A)

484256

518

240

043

minus1390

πrarr

πdG

(11 A)

484256

61A

512

242

000

057

nrarr

πdC

(21 A)

525

236

000

minus1197

nrarr

πdC

(21 A)

71A

523

237

000

minus1384

nrarr

πdG

(21 A)

529

234

000

minus579

nrarr

πdG

(2+51A)

81A

623

199

000

minus034

nrarr

πdG

(2+51A)

SACminusCIof

Z-Stack

exptle

SACminusCIof

B-Stack

exptle

EEa

oscb

rotc

CD

EEa

oscb

rotc

CD

state

(eV)

(nm)

(au)

(10minus

40cgs)

nature

typed

(eVnm)

(eV)

(nm)

(au)

(10minus

40cgs)

nature

typed

(eVnm)

11A

395

314

009

minus4511

πrarr

πdC

(11 A)

419296

404

307

007

4358

πrarr

πdC

(11 A)

443280

21A

421

294

000

minus3349

πrarr

πdG

(11 A)

419296

440

282

006

minus1032

πrarr

πdG

(11 A)

31A

445

279

001

2967

nrarr

πdC

(21 A)

461

269

000

minus503

nrarr

πdC

(21 A)

41A

472

263

000

minus1335

nrarr

πdG

(21 A)

493

251

013

9407

πrarr

πdC

(3+41A)

51A

485

256

019

minus1796

πrarr

πdG

(31 A)

451275

507

245

005

minus15115

nrarr

πdG

(21 A)

486255

61A

499

248

006

4861

πrarr

πET

451275

512

242

014

minus11056

πrarr

πdG

(31 A)

486255

71A

526

236

007

2721

πrarr

πET

+dC

(31 A)

451275

521

238

014

19475

πrarr

πET+dG

(31 A)

81A

565

219

000

minus231

nrarr

πdC

(41 A)

619

200

000

minus805

nrarr

πdG

(41 A)

aExcitatio

nenergybOscillator

strengthcRotatorystrengthddG

representsintram

olecular

excitedstates

ofdG

dCrepresentsintram

olecular

excitedstates

ofdC

and

ETrepresentselectron

transfer

excitedstates

from

dGto

dCeRef

4

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5550

positive at 295 nm Therefore the strong negative sign of theCD spectra at 295 nm indicates that a strong stackinginteraction exists in Z-DNA The SAC-CI CD spectra of bothstacking models (B-Stack and Z-Stack models) more closelyresemble the experimental CD spectra of B- and Z-DNA thanthose of the hydrogen-bonding models (B-Hbond and Z-Hbond models) For the SAC-CI CD spectrum of the Z-Stackmodel the first and second peaks are negative and positiverespectively but they are the opposite of those obtained fromthe Z-Hbond model For the B-Stack and B-Hbond modelshowever the first peak is positive in the SAC-CI CD spectrabut the second and third peaks are negative and positiverespectively for the B-Stack model in opposition to the B-Hbond model Since the stacking models do not have thehydrogen-bonding interactions the excitation energy of the firstpeak is lower than that of the experimental spectrum as notedaboveIn the Z-Stack model the first peak with a negative sign is

assigned to the 11A (395 eV) and 21A (421 eV) states (Table4) The 11A state of the Z-Stack model corresponds to the 11Astate (413 eV) of dC which is the excitation of the πminusπ fromthe HOMO to the LUMO in dC A large oscillator strength isthe origin of the first (shoulder) peak but the excitation energyis lower than the experimental value because there is nohydrogen bond in the Z-Stack model The rotatory strength ofthe 11A state of dC changes the sign from positive to negativeby the stacking interaction with dG Therefore the stackinginteraction between dG and dC accounts for the strongnegative sign of the CD spectra in Z-DNA

The excitation energy of the 21A state (421 eV) is also closeto the first peak of the experimental CD spectrum This state isassigned to the 11A state (437 eV) of dG which is the HOMOto LUMO πminusπ excitation in dG but also includes thecontribution of the excitation from the next-HOMO to LUMOin dC The oscillator strength in the UV spectrum is smallbecause this state differs from the 11A state of dG by theinclusion of the excitation in dC Although the excitationenergy is shifted by the formation of hydrogen bonds betweendG and dC the rotatory strength of a large negative maycontribute to the strong negative sign of the CD spectra in Z-DNA In the 31A state (445 eV) which is the excitation fromnonbonding orbital to the LUMO in dC the rotatory strengthchanges to positive from the negative sign of the monomer dCHowever this state is not the origin of the second peak as willbe noted later in the tetramer model section Thus the rotatorystrength is strongly affected by the stacking interactionThree important excited states are calculated in the higher-

energy region of the second peak The 51A state (485 eV) hasthe largest oscillator strength and is assigned to the 31A (491eV) state of dG which is the excitation from HOMO to next-LUMO This state is assigned to the main peak (484 eV) of theexperimental UV spectrum of Z-DNA The 6 and 71A states(499 and 526 eV) with a large rotatory strength are the ETstate from the HOMO of dG to the LUMO of dC The mainconfiguration of the 71A state also includes the excitation fromthe next-HOMO to the LUMO in dCIn the B-Stack model the first peak with a positive sign is

assigned to the 11A state (404 eV) This state corresponds tothe 11A state (409 eV) of dC which is the excitation from theHOMO to the LUMO The 21A state (440 eV) is assigned tothe 11A state (434 eV) of dG which reflects the excitationfrom the HOMO to the LUMO Since the B-Stack model doesnot have hydrogen bonds the excitation energies are close tothose of dG or dC Both states have the oscillator strength butthe 11A state is positive while the 21A state is slightly negativefor the rotatory strength which is the same as for themonomers dC and dG respectively Therefore the stackinginteraction of B-DNA has little effect on the sign of the CDspectra at 295 nm because the stacking interaction is weaker inB-DNA than in Z-DNAThe second peak with a negative sign is assigned to the 5 and

61A states with a large rotatory strength The 5 and 61A states(507 and 512 eV in Table 4) are assigned to the 2 and 31Astates (493 and 518 eV in Table 3) of dG which are theexcitations from the nonbonding orbital and the HOMO to thenext-LUMO The main configuration is nminusπ for the 51A stateand πminusπ for the 61A state Both states have strong oscillatorand negative rotatory strengths which means that they can beassigned to the main peaksThe first ET excited states (71A state 521 eV) of B-Stack

model are calculated at a higher region than that (61A state499 eV) of the Z-Stack model because the stacking interactionis weaker in B-DNA than in Z-DNAThe SAC-CI results show that both hydrogen-bonding and

stacking interactions affect the UV and CD spectra of DNA Inparticular the excitation energy of the first excited stateincreases by the formation of hydrogen bonds and the stackinginteraction changes the sign of the CD spectrum of the firstexcited state from positive to negative The strong stackinginteraction characteristic of the helical structure of Z-DNA isthe origin of the strong negative sign at around 295 nm in theCD spectra The hydrogen-bonding and stacking interactions

Figure 7 SAC-CI CD spectra for the dimer models The SAC-CIspectra (magenta line) of (a) Z-Hbond model (b) Z-Stack model (c)B-Hbond model and (d) B-Stack model are compared with theexperimental CD spectra4 (black line) of Z-DNA (ab) and B-DNA(cd) The red and blue lines represent the intramolecular excitedstates of dG and dC respectively The green lines represent theelectron transfer (ET) excited state from dG to dC

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5551

also confer other smaller effects on the UV and CD spectra ofDNA In the higher energy region the SAC-CI CD spectra arenot in good agreement with the experimental CD spectrabecause these models have either hydrogen-bonding interactionor stacking interaction but not both Therefore we need toinvestigate a better model that includes both hydrogen bondingand stacking interactions within the same model A minimumsuch model is the tetramer model of Figure 1 on which weexplain in the next section35 Tetramer Model The SAC-CI calculations of the

dimer models were suitable to investigate the effects of thehydrogen-bonding interaction and the stacking interactionseparately For example the SAC-CI UV spectra of thehydrogen-bonding models were in good agreement with theexperimental UV spectra implying the importance of thehydrogen-bonding interaction The SAC-CI CD spectra for thestacking model clarified that the stacking interaction is thesource of the negative CD peak at around 295 nm Howevergenerally speaking the SAC-CI CD spectra were markedlydifferent from the experimental spectra The SAC-CI UVspectra of the stacking models were composed of two peaksdifferent from the experimental UV spectra Thus theinformation that we gained from the dimer models still didnot fully resolve these differences in the spectra acquired by twodifferent processes So we used the tetramer model (Figure 1)which included both hydrogen-bonding and stacking inter-actions within the same modelFigure 8 shows the SAC-CI UV and CD spectra of the

tetramer model of Z-DNA (Z-Tetra model) and B-DNA (B-Tetra model) in comparison with the experimental UV andCD spectra In order to compare the experimental spectra theSAC-CI results were shifted to lower energy by 05 eV Thecomputational conditions were poor as compared with theother SAC-CI calculations In the tetramer calculations the d-polarization functions were not added and the small activespace was used The SAC-CI calculations using the small activespace and without the d-polarization functions gave higherexcitation energies by about 04minus05 eV (see SupportingInformation Table S1) for the hydrogen-bonding dimermodels In particular the excited states corresponding to themain peak of the experimental UV spectra were shifted by 048eV for both the Z-Hbond model (61A state) and the B-Hbondmodel (71A state)Table 5 shows the excitation energies oscillator strengths

rotatory strengths natures and type of the excited states of theSAC-CI results for tetramer modelsThe SAC-CI UV spectrum of the Z-Tetra model is in good

agreement with the experimental UV spectrum of Z-DNA Themain peak as well as the shoulder peak is reproduced by theSAC-CI method The shoulder peak is assigned to the 4 and51A states which is the πminusπ excitation (HOMO to LUMO)in dC and the main peak is assigned to the 101A state which isthe πminusπ (HOMO to next-LUMO) excitation in dGThe SAC-CI UV spectrum of the B-Tetra model is also in a

reasonable agreement with the experimental UV spectrum of B-DNA although the shape of the shoulder is slightly different Inthe B-Tetra model the shoulder peak is assigned to the 21Astate which is the πminusπ (HOMO to LUMO) excitation in dCand the main peak is assigned to the 81A state which is theπminusπ (HOMO to next-LUMO) excitation in dGThe sign of the rotatory strength is the same between the

SAC-CI CD spectrum of the Z-Tetra model and theexperimental CD spectrum of Z-DNA although the negative

rotatory strength at 295 nm is very weak because thepolarization functions were not included in the tetramermodels It is well-known that the basis set must be good for anadequate description of the stacking interaction5859 In thedimer model the lack of polarization d-functions and thesmaller active space changed the intensity and sign of therotatory strength (see Supporting Information Table S1)Since the stacking interaction is small in B-DNA the SAC-CI

CD spectrum of the B-Tetra model is in good agreement withthe experimental CD spectrum of B-DNA and reproduces notonly the sign of the rotatory strength but also the overall shapeof the spectrumIn Z-DNA the 11A state is assigned to the first peak of the

experimental CD spectrum The 11A state is the ET excitationfrom the HOMO of dG to the LUMO of dC through thestacking conformation The 3 and 51A states with a largenegative rotatory strength are the excitation from the HOMOto the LUMO in dG and dC respectively These states alsocontribute to the first negative peak as in the 1 and 21A stateswith a negative rotatory strength in the Z-Stack model The 8and 101A states are assigned to the second peak of the CDspectrum The 101A state which is the origin of the main peakof the UV spectrum and the πminusπ of dC (81A state) alsocontribute to the second peak of the CD spectrumIn B-DNA the first peak of the CD spectrum is assigned to

the 21A state of πminusπ (HOMO to LUMO) in dC which is thesame as the origin of the first peak of the UV spectrum The ETstates are calculated at the higher energy region than the 21A

Figure 8 SAC-CI UV and CD spectra for the tetramer models TheSAC-CI spectra (magenta line) of Z-Tetra model (ab) and B-Tetramodel (cd) as compared with the experimental UV and CD spectra4

(black line) of Z-DNA (ab) and B-DNA (cd) The red and blue linesrepresent the intramolecular excited states of dG and dC respectivelyThe green lines represent the electron transfer (ET) excited state fromdG to dC All excited states of SAC-CI calculations have been shiftedto the lower values by 05 eV

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5552

Table

5SA

C-CIUVandCD

Spectraof

theTetramer

Mod

elsof

Z-andB-D

NA

Z-DNA

B-DNA

SACminusCIof

Z-Tetra

exptle

SAC-CIof

B-Tetra

exptle

EEa

Oscb

Rotc

UV

CD

EEa

Oscb

Rotc

UV

CD

state

(eV)

(nm)

(au)

(10minus

40cgs)

nature

typed

(eVnm)

(eVnm)

(eV)

(nm)

(au)

(10minus

40cgs)

nature

typed

(eVnm)

(eVnm)

11A

470

264

000

minus663

πrarr

πET

419296

479

259

003

minus041

πrarr

πdG

21A

481

258

002

6611

πrarr

πET

482

257

033

16891

πrarr

πdC

443280

31A

484

256

004

minus5326

πrarr

πdG

419296

484

256

001

minus7454

πrarr

πdC

41A

488

254

015

4553

πrarr

πdC

428290

498

249

007

minus130

πrarr

πET

51A

492

252

012

minus11712

πrarr

πdC

428290

419296

511

243

004

minus622

πrarr

πET

61A

502

247

003

minus3034

πrarr

πET

525

236

000

minus046

πrarr

πET

71A

513

242

001

2737

nrarr

πdG

527

235

001

332

πrarr

πET

81A

515

241

006

14772

πrarr

πdC

451275

545

227

045

minus15609

πrarr

πdG

484256

486255

91A

522

238

001

1978

nrarr

πdG

552

225

002

minus1795

nrarr

πdG

101 A

538

230

071

7299

πrarr

πdG

484256

451275

556

223

013

5929

πrarr

πdG

111 A

541

229

001

minus6858

πrarr

πdG

558

222

009

minus525

πrarr

πdG

121 A

548

226

002

minus1519

nrarr

πdC

563

220

008

minus7436

nrarr

πdG

131 A

553

224

001

minus924

nrarr

πdC

609

204

001

1379

nrarr

πdG

141 A

601

206

001

minus4159

nrarr

πdG

626

198

002

minus464

nrarr

πdG

aExcitatio

nenergyA

llexcitedstates

ofSA

C-CIcalculations

have

been

shifted

tothelower

values

by05eV

inFigure

8bOscillator

strengthcRotatorystrengthddG

representsintram

olecular

excited

states

ofdG

dCrepresentsintram

olecular

excitedstates

ofdC

and

ETrepresentselectron

transfer

excitedstates

from

dGto

dCeRef

4

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5553

state because the stacking interaction is weaker in B-DNA thanin Z-DNA (Figure 1) Therefore the first peak of theexperimental CD spectra originates from the ET excitationfrom dG to dC for Z-DNA but corresponds to theintramolecular excitation in dC for B-DNA The second peakwith a negative sign is assigned to the 81A state The 81A statewhich is the πminusπ of dG is the origin of both the main peaks ofthe UV spectrum and the second peak of the CD spectrumThe ET excited states of the tetramer models were calculated

at a lower energy region than those of the dimer models forboth B- and Z-DNA This result may be related to the overlapbetween the nucleic acid bases The overlap exists only in partof the nucleic acid bases in the dimer models but in thetetramer models the overlap covers the hydrogen bonds as wellas the nucleic acid bases The ET excited states are importantfor the CD spectrum of Z-DNA because the lowest excitedstate is the ET type and is negative for rotatory strength but forB-DNA the ET excited states are not important because thesestates are weak in both oscillator and rotatory strengthsThe SAC-CI results of the tetramer models are calculated at

a higher region by about 05 eV because of the small activespace Therefore better SAC-CI calculations are required forcomparison with the experimental UV and CD spectra of DNAHowever these results indicate that the strong stackinginteraction characteristic of the helical structure of Z-DNAchanges the sign of the rotatory strength for the lowest excitedstates of dG and dC and produces the emergence of the ETexcited states

4 CONCLUSIONSDNA chemistry is affected by three important factors (1) theconformation of the monomer (dG) which is related to thedihedral angle between deoxyribose and guanine (2) thehydrogen-bonding interactions and (3) the stacking inter-actions between two nucleic acid bases To verify whether thesethree factors are important for the CD spectra of DNA weexamined four different models using the SAC-CI method(1) dG monomer model We first applied the SAC-CI

method to the dG monomer to examine the effect ofconformation on the CD spectra of dG and from this resultto identify the structure of the dG monomer in solution Ourcalculations showed that the shape and the sign of the CDspectra depend strongly on the dihedral angle between thedeoxyribose and the guanine but that the SAC-CI UV spectraof both anti- and syn-dG are very similar because the UVspectra are not strongly dependent on the conformation Theexperimental CD spectrum agreed with the theoreticalspectrum of dG only for the anti conformation The resultingthermally averaged spectra also support the anti geometry ofdG as a stable conformation in solution However the SAC-CICD spectrum of syn-dG is opposite in sign from that of theexperimental spectrum Thus the SAC-CI CD spectra of dG inboth anti and syn conformations do not explain the negativesign at 295 nm of the CD spectra of the DNA composed of dGand dC(2) Composit model of dG and dC monomers A

comparison of the calculated composite spectra of dG anddC with the experimental spectra of the two-component DNAshowed similarities between B- and Z-DNA because theconformation of dC is always anti in both B- and Z-DNAThe SAC-CI results led us to confirm that that theconformation of dG is not the origin of the negative peak at295 nm in Z-DNA

(3) Dimer model Our examination of the dimer modelpermitted us to investigate the separate importance of thehydrogen-bonding and the stacking interactions in thestructural chemistry of B- and Z-DNA The SAC-CIcalculations showed that the hydrogen-bonding interactionschange the excitation energies and that the stacking interactionsare responsible for the changes in the sign of the CD rotatorystrength The SAC-CI UV spectra of the hydrogen-bondingmodels (B-Hbond and Z-Hbond models) are similar to theexperimental UV spectra in which the main peak as well as theshoulder peak is well reproduced The SAC-CI CD spectra ofthe stacking models (B-Stack and Z-Stack models) have thesame sign as the experimental CD spectra We concluded thatthe stacking interaction is especially important for the CDspectrum of the strongly stacked Z-DNA because the CD signalat 295 nm changed to a negative sign only for the Z-Stackmodel(4) Tetramer model Finally our calculations with the

tetramer model that contain both hydrogen-bonding andstacking interactions revealed that the shape of the SAC-CIUV and CD spectra compared well with the experimental onesThe ET excitations as well as the lowest intramolecularexcitation of dC and dG are the origins of the negative sign ofthe experimental CD spectrum of Z-DNA The ET excited stateis calculated at a lower energy region than the intramolecularexcited states of dG or dC for the strongly stacked Z-DNA butat a higher energy region for the weakly stacked B-DNA Thenegative sign of the first peak of the experimental CD spectra ofZ-DNA is due to the strong stacking effect characteristic of itshelical structureThe dynamic behavior of the dG monomer in solution was

partially considered by taking the Boltzmann averaging of thecalculated spectra over the various conformations Fortunatelythe Boltzmann averaged SAC-CI CD spectrum was in goodagreement with the experimental spectra although the SAC-CICD spectra largely depend on the conformation of dGHowever in this study we used only the static model for dimerand tetramer models The SAC-CI CD results of the dimer andtetramer models explained the difference between theexperimental CD spectra of B- and Z-DNA However sincethe structure of DNA would also be flexible in solution it isnecessary to take into account the dynamic behavior of DNAThis is an important and interesting topic

ASSOCIATED CONTENT

S Supporting InformationSAC-CI results of the hydrogen-bonding and stacking dimermodels of Z- and B-DNA using small active space and withoutthe polarization d-functions This material is available free ofcharge via the Internet at httppubsacsorg

AUTHOR INFORMATION

Corresponding AuthorPhone +81-75-634-3211 Fax +81-75-634-3211 E-mail hnakatsujiqcriorjp

NotesThe authors declare no competing financial interest

ACKNOWLEDGMENTS

The computations were performed at the Research Center forComputational Science Okazaki Japan

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5554

REFERENCES(1) Voet D Voet J G Eds Biochemistry John Wiley amp Sons IncNew York 1995(2) Ghosh A Bansal M Acta Crystallogr 2003 D59 620minus626(3) Rich A Nordheim A Wang A H-J Annu Rev Biochem 198453 791minus846(4) Tran-Dinh S Taboury J Neumann J-M Huynh-Dinh TGenissel B Langlois drsquoEstaintot B Igolen J Biochemistry 1984 231362minus1371(5) Sugiyama H Kawai K Matsunaga A Fujimoto K Saito IRobinson H Wang A H-J Nucleic Acids Res 1996 24 1272minus1278(6) Kawai K Saito I Sugiyama H J Am Chem Soc 1999 1211391minus1392(7) Oyoshi T Kawai K Sugiyama H J Am Chem Soc 2003 1251526minus1531(8) Xu Y Ikeda R Sugiyama H J Am Chem Soc 2003 12513519minus13524(9) Berova N Nakanishi K Woody R W Eds Circular DichroismPrinciples and Applications Wiley-VCH New York 2000(10) Bauernschmitt R Ahlrichs R Chem Phys Lett 1996 256454minus464(11) Casida M E Jamorski C Casida K C Salahub D R JChem Phys 1998 108 4439minus4449(12) Tsolakidis A Kaxiras E J Phys Chem A 2005 109 2373minus2380(13) Varsano D Di Felice R Marques M A L Rubio A J PhysChem B 2006 110 7129minus7138(14) Nielsen L M Holm A I S Varsano D Kadhane UHoffmann S V Di Felice R Rubio A Nielsen S B J Phys ChemB 2009 113 9614minus9619(15) Tonzani S Schatz G C J Am Chem Soc 2008 130 7607minus7612(16) Casida M E Salahub D R J Chem Phys 2000 113 8918minus8935(17) Dreuw A Weisman J L Head-Gordon M J Chem Phys2003 119 2943minus2946(18) Maitra N T Zhang F Cave R J Burke K J Chem Phys2004 120 5932minus5937(19) McCann D M Stephens P J J Org Chem 2006 71 6074minus6098(20) Casida M E J Mol Struct 2009 914 3minus18(21) Niehaus T A J Mol Struct 2009 914 38minus49(22) Tawada Y Tsuneda T Yanagisawa S Yanai T Hirao K JChem Phys 2004 120 8425minus8433(23) Chiba M Tsuneda T Hirao K J Chem Phys 2006 124144106-1-11(24) Nakatsuji H Hirao K J Chem Phys 1978 68 2053minus2065(25) Nakatsuji H Chem Phys Lett 1978 59 362minus364(26) Nakatsuji H Chem Phys Lett 1979 67 329minus333(27) Nakatsuji H Chem Phys Lett 1979 67 334minus342(28) Nakatsuji H Bull Chem Soc Jpn 2005 78 1705minus1724(29) Ehara M Hasegawa J Nakatsuji H SAC-CI Method Appliedto Molecular Spectroscopy In Theory and Applications of Computa-tional Chemistry The First 40 Years A Vol of Technical and HistoricalPerspectives Dykstra C E Frenking G Kim K S Scuseria G EEds Elsevier Oxford UK 2005 pp 1099minus1141(30) Miyahara T Hasegawa J Nakatsuji H Bull Chem Soc Jpn2009 82 1215minus1226(31) Bureekaew S Hasegawa J Nakatsuji H Chem Phys Lett2006 425 367minus371(32) Nakatani N Hasegawa J Nakatsuji H J Am Chem Soc2007 129 8756minus8765(33) Fujimoto K Hasegawa J Nakatsuji H Bull Chem Soc Jpn2009 82 1140minus1148(34) Miyahara T Nakatsuji H Collect Czech Chem Commun2011 76 537minus552(35) Sugiyama H Unpublished(36) Geertsen J Rittby M Bartlett R J Chem Phys Lett 1989164 57minus62

(37) Stanton J F Bartlett R J J Chem Phys 1993 98 7029minus7039(38) Ehara M Ishida M Toyota K Nakatsuji H SAC-CIGeneral-R Method Theory and Applications to the Multi-ElectronProcesses In Reviews in Modern Quantum Chemistry (A Tribute toProfessor Robert G Parr) Sen K D Ed World Scientific Singapore2003 pp 293minus319(39) SAC-CI GUIDE httpwwwqcriorjpsacci(40) Frisch M J Trucks G W Schlegel H B et al Gaussian 03development revision revision E05 Gaussian Inc Wallingford CT2004(41) Hohenberg P Kohn W Phys Rev 1964 136 B864minusB871(42) Kohn W Sham L J Phys Rev 1965 140 A1133minusA1138(43) Salahub D R Zerner M C Eds The Challenge of d and fElectrons The American Chemical Society Washington DC 1989(44) Parr R G Yang W Density-Functional Theory of Atoms andMolecules Oxford University Press Oxford UK 1989(45) Becke A D J Chem Phys 1993 98 5648minus5652(46) Lee C Yang W Parr R G Phys Rev 1988 37 785minus789(47) Hariharan P C Pople J A Theor Chim Acta 1973 28 213minus222(48) Francl M M Pietro W J Hehre W J Binkley J SDeFrees D J Pople J A Gordon M S J Chem Phys 1982 773654minus3665(49) Tomasi J Mennucci B Cammi R Chem Rev 2005 1052999minus3093(50) Fukuda R Ehara M Nakatsuji H Cammi R J Chem Phys2011 134 104109minus1minus11(51) Dunning T H Jr Hay P J In Modern Theoretical ChemistrySchaefer H F III Ed Plenum New York 1976 Vol 3 pp 1minus28(52) Nakatsuji H Chem Phys 1983 75 425minus441(53) Hocquet A Leulliot N Ghomi M J Phys Chem B 2000 1044560minus4568(54) Mishra S K Mishra P C J Comput Chem 2002 23 530minus540(55) Sprecher C A Johnson W C Biopolymers 1977 16 2243minus2264(56) Clark L B Peschel G G Tinoc I Jr J Phys Chem 1965 693615minus3618(57) Voet D Gratzer W B Cox R A Doty P Biopolymers 19631 193minus208(58) Hobza P Mehlhorn A Carsky P Zahradnik R J Mol Struct1986 138 387minus399(59) Hobza P Sponer J Chem Rev 1999 99 3247minus3276

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5555

Page 9: Helical Structure and Circular Dichroism Spectra of DNA: A ... · B-DNA, have right-handed double-helical structure, while Z-DNA is left-handed and appears only when DNA has a special

Table

4SA

C-CIUVandCD

Spectraof

theHydrogen-Bon

ding

andStacking

Dim

erMod

elsof

Z-andB-D

NA

Z-DNA

B-DNA

SAC-CIof

Z-H

bond

exptle

SAC-CIof

B-H

bond

exptle

EEa

oscb

rotc

UV

EEa

oscb

rotc

UV

state

(eV)

(nm)

(au)

(10minus

40cgs)

nature

typed

(eVnm)

(eV)

(nm)

(au)

(10minus

40cgs)

nature

Typed

(eVnm)

11A

439

282

020

3513

πrarr

πdC

(11 A)

428290

437

284

017

1143

πrarr

πdC

(11 A)

21A

453

274

004

884

πrarr

πdG

(31 A)

456

272

003

minus1481

πrarr

πdG

(31 A)

31A

484

256

018

912

πrarr

πdC

(31 A)

493

251

006

583

πrarr

πET

41A

485

255

005

minus1145

πrarr

πET

498

249

021

1767

πrarr

πdC

(3+41A)

51A

493

252

042

1182

πrarr

πdG

(11 A)

484256

518

240

043

minus1390

πrarr

πdG

(11 A)

484256

61A

512

242

000

057

nrarr

πdC

(21 A)

525

236

000

minus1197

nrarr

πdC

(21 A)

71A

523

237

000

minus1384

nrarr

πdG

(21 A)

529

234

000

minus579

nrarr

πdG

(2+51A)

81A

623

199

000

minus034

nrarr

πdG

(2+51A)

SACminusCIof

Z-Stack

exptle

SACminusCIof

B-Stack

exptle

EEa

oscb

rotc

CD

EEa

oscb

rotc

CD

state

(eV)

(nm)

(au)

(10minus

40cgs)

nature

typed

(eVnm)

(eV)

(nm)

(au)

(10minus

40cgs)

nature

typed

(eVnm)

11A

395

314

009

minus4511

πrarr

πdC

(11 A)

419296

404

307

007

4358

πrarr

πdC

(11 A)

443280

21A

421

294

000

minus3349

πrarr

πdG

(11 A)

419296

440

282

006

minus1032

πrarr

πdG

(11 A)

31A

445

279

001

2967

nrarr

πdC

(21 A)

461

269

000

minus503

nrarr

πdC

(21 A)

41A

472

263

000

minus1335

nrarr

πdG

(21 A)

493

251

013

9407

πrarr

πdC

(3+41A)

51A

485

256

019

minus1796

πrarr

πdG

(31 A)

451275

507

245

005

minus15115

nrarr

πdG

(21 A)

486255

61A

499

248

006

4861

πrarr

πET

451275

512

242

014

minus11056

πrarr

πdG

(31 A)

486255

71A

526

236

007

2721

πrarr

πET

+dC

(31 A)

451275

521

238

014

19475

πrarr

πET+dG

(31 A)

81A

565

219

000

minus231

nrarr

πdC

(41 A)

619

200

000

minus805

nrarr

πdG

(41 A)

aExcitatio

nenergybOscillator

strengthcRotatorystrengthddG

representsintram

olecular

excitedstates

ofdG

dCrepresentsintram

olecular

excitedstates

ofdC

and

ETrepresentselectron

transfer

excitedstates

from

dGto

dCeRef

4

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5550

positive at 295 nm Therefore the strong negative sign of theCD spectra at 295 nm indicates that a strong stackinginteraction exists in Z-DNA The SAC-CI CD spectra of bothstacking models (B-Stack and Z-Stack models) more closelyresemble the experimental CD spectra of B- and Z-DNA thanthose of the hydrogen-bonding models (B-Hbond and Z-Hbond models) For the SAC-CI CD spectrum of the Z-Stackmodel the first and second peaks are negative and positiverespectively but they are the opposite of those obtained fromthe Z-Hbond model For the B-Stack and B-Hbond modelshowever the first peak is positive in the SAC-CI CD spectrabut the second and third peaks are negative and positiverespectively for the B-Stack model in opposition to the B-Hbond model Since the stacking models do not have thehydrogen-bonding interactions the excitation energy of the firstpeak is lower than that of the experimental spectrum as notedaboveIn the Z-Stack model the first peak with a negative sign is

assigned to the 11A (395 eV) and 21A (421 eV) states (Table4) The 11A state of the Z-Stack model corresponds to the 11Astate (413 eV) of dC which is the excitation of the πminusπ fromthe HOMO to the LUMO in dC A large oscillator strength isthe origin of the first (shoulder) peak but the excitation energyis lower than the experimental value because there is nohydrogen bond in the Z-Stack model The rotatory strength ofthe 11A state of dC changes the sign from positive to negativeby the stacking interaction with dG Therefore the stackinginteraction between dG and dC accounts for the strongnegative sign of the CD spectra in Z-DNA

The excitation energy of the 21A state (421 eV) is also closeto the first peak of the experimental CD spectrum This state isassigned to the 11A state (437 eV) of dG which is the HOMOto LUMO πminusπ excitation in dG but also includes thecontribution of the excitation from the next-HOMO to LUMOin dC The oscillator strength in the UV spectrum is smallbecause this state differs from the 11A state of dG by theinclusion of the excitation in dC Although the excitationenergy is shifted by the formation of hydrogen bonds betweendG and dC the rotatory strength of a large negative maycontribute to the strong negative sign of the CD spectra in Z-DNA In the 31A state (445 eV) which is the excitation fromnonbonding orbital to the LUMO in dC the rotatory strengthchanges to positive from the negative sign of the monomer dCHowever this state is not the origin of the second peak as willbe noted later in the tetramer model section Thus the rotatorystrength is strongly affected by the stacking interactionThree important excited states are calculated in the higher-

energy region of the second peak The 51A state (485 eV) hasthe largest oscillator strength and is assigned to the 31A (491eV) state of dG which is the excitation from HOMO to next-LUMO This state is assigned to the main peak (484 eV) of theexperimental UV spectrum of Z-DNA The 6 and 71A states(499 and 526 eV) with a large rotatory strength are the ETstate from the HOMO of dG to the LUMO of dC The mainconfiguration of the 71A state also includes the excitation fromthe next-HOMO to the LUMO in dCIn the B-Stack model the first peak with a positive sign is

assigned to the 11A state (404 eV) This state corresponds tothe 11A state (409 eV) of dC which is the excitation from theHOMO to the LUMO The 21A state (440 eV) is assigned tothe 11A state (434 eV) of dG which reflects the excitationfrom the HOMO to the LUMO Since the B-Stack model doesnot have hydrogen bonds the excitation energies are close tothose of dG or dC Both states have the oscillator strength butthe 11A state is positive while the 21A state is slightly negativefor the rotatory strength which is the same as for themonomers dC and dG respectively Therefore the stackinginteraction of B-DNA has little effect on the sign of the CDspectra at 295 nm because the stacking interaction is weaker inB-DNA than in Z-DNAThe second peak with a negative sign is assigned to the 5 and

61A states with a large rotatory strength The 5 and 61A states(507 and 512 eV in Table 4) are assigned to the 2 and 31Astates (493 and 518 eV in Table 3) of dG which are theexcitations from the nonbonding orbital and the HOMO to thenext-LUMO The main configuration is nminusπ for the 51A stateand πminusπ for the 61A state Both states have strong oscillatorand negative rotatory strengths which means that they can beassigned to the main peaksThe first ET excited states (71A state 521 eV) of B-Stack

model are calculated at a higher region than that (61A state499 eV) of the Z-Stack model because the stacking interactionis weaker in B-DNA than in Z-DNAThe SAC-CI results show that both hydrogen-bonding and

stacking interactions affect the UV and CD spectra of DNA Inparticular the excitation energy of the first excited stateincreases by the formation of hydrogen bonds and the stackinginteraction changes the sign of the CD spectrum of the firstexcited state from positive to negative The strong stackinginteraction characteristic of the helical structure of Z-DNA isthe origin of the strong negative sign at around 295 nm in theCD spectra The hydrogen-bonding and stacking interactions

Figure 7 SAC-CI CD spectra for the dimer models The SAC-CIspectra (magenta line) of (a) Z-Hbond model (b) Z-Stack model (c)B-Hbond model and (d) B-Stack model are compared with theexperimental CD spectra4 (black line) of Z-DNA (ab) and B-DNA(cd) The red and blue lines represent the intramolecular excitedstates of dG and dC respectively The green lines represent theelectron transfer (ET) excited state from dG to dC

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5551

also confer other smaller effects on the UV and CD spectra ofDNA In the higher energy region the SAC-CI CD spectra arenot in good agreement with the experimental CD spectrabecause these models have either hydrogen-bonding interactionor stacking interaction but not both Therefore we need toinvestigate a better model that includes both hydrogen bondingand stacking interactions within the same model A minimumsuch model is the tetramer model of Figure 1 on which weexplain in the next section35 Tetramer Model The SAC-CI calculations of the

dimer models were suitable to investigate the effects of thehydrogen-bonding interaction and the stacking interactionseparately For example the SAC-CI UV spectra of thehydrogen-bonding models were in good agreement with theexperimental UV spectra implying the importance of thehydrogen-bonding interaction The SAC-CI CD spectra for thestacking model clarified that the stacking interaction is thesource of the negative CD peak at around 295 nm Howevergenerally speaking the SAC-CI CD spectra were markedlydifferent from the experimental spectra The SAC-CI UVspectra of the stacking models were composed of two peaksdifferent from the experimental UV spectra Thus theinformation that we gained from the dimer models still didnot fully resolve these differences in the spectra acquired by twodifferent processes So we used the tetramer model (Figure 1)which included both hydrogen-bonding and stacking inter-actions within the same modelFigure 8 shows the SAC-CI UV and CD spectra of the

tetramer model of Z-DNA (Z-Tetra model) and B-DNA (B-Tetra model) in comparison with the experimental UV andCD spectra In order to compare the experimental spectra theSAC-CI results were shifted to lower energy by 05 eV Thecomputational conditions were poor as compared with theother SAC-CI calculations In the tetramer calculations the d-polarization functions were not added and the small activespace was used The SAC-CI calculations using the small activespace and without the d-polarization functions gave higherexcitation energies by about 04minus05 eV (see SupportingInformation Table S1) for the hydrogen-bonding dimermodels In particular the excited states corresponding to themain peak of the experimental UV spectra were shifted by 048eV for both the Z-Hbond model (61A state) and the B-Hbondmodel (71A state)Table 5 shows the excitation energies oscillator strengths

rotatory strengths natures and type of the excited states of theSAC-CI results for tetramer modelsThe SAC-CI UV spectrum of the Z-Tetra model is in good

agreement with the experimental UV spectrum of Z-DNA Themain peak as well as the shoulder peak is reproduced by theSAC-CI method The shoulder peak is assigned to the 4 and51A states which is the πminusπ excitation (HOMO to LUMO)in dC and the main peak is assigned to the 101A state which isthe πminusπ (HOMO to next-LUMO) excitation in dGThe SAC-CI UV spectrum of the B-Tetra model is also in a

reasonable agreement with the experimental UV spectrum of B-DNA although the shape of the shoulder is slightly different Inthe B-Tetra model the shoulder peak is assigned to the 21Astate which is the πminusπ (HOMO to LUMO) excitation in dCand the main peak is assigned to the 81A state which is theπminusπ (HOMO to next-LUMO) excitation in dGThe sign of the rotatory strength is the same between the

SAC-CI CD spectrum of the Z-Tetra model and theexperimental CD spectrum of Z-DNA although the negative

rotatory strength at 295 nm is very weak because thepolarization functions were not included in the tetramermodels It is well-known that the basis set must be good for anadequate description of the stacking interaction5859 In thedimer model the lack of polarization d-functions and thesmaller active space changed the intensity and sign of therotatory strength (see Supporting Information Table S1)Since the stacking interaction is small in B-DNA the SAC-CI

CD spectrum of the B-Tetra model is in good agreement withthe experimental CD spectrum of B-DNA and reproduces notonly the sign of the rotatory strength but also the overall shapeof the spectrumIn Z-DNA the 11A state is assigned to the first peak of the

experimental CD spectrum The 11A state is the ET excitationfrom the HOMO of dG to the LUMO of dC through thestacking conformation The 3 and 51A states with a largenegative rotatory strength are the excitation from the HOMOto the LUMO in dG and dC respectively These states alsocontribute to the first negative peak as in the 1 and 21A stateswith a negative rotatory strength in the Z-Stack model The 8and 101A states are assigned to the second peak of the CDspectrum The 101A state which is the origin of the main peakof the UV spectrum and the πminusπ of dC (81A state) alsocontribute to the second peak of the CD spectrumIn B-DNA the first peak of the CD spectrum is assigned to

the 21A state of πminusπ (HOMO to LUMO) in dC which is thesame as the origin of the first peak of the UV spectrum The ETstates are calculated at the higher energy region than the 21A

Figure 8 SAC-CI UV and CD spectra for the tetramer models TheSAC-CI spectra (magenta line) of Z-Tetra model (ab) and B-Tetramodel (cd) as compared with the experimental UV and CD spectra4

(black line) of Z-DNA (ab) and B-DNA (cd) The red and blue linesrepresent the intramolecular excited states of dG and dC respectivelyThe green lines represent the electron transfer (ET) excited state fromdG to dC All excited states of SAC-CI calculations have been shiftedto the lower values by 05 eV

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5552

Table

5SA

C-CIUVandCD

Spectraof

theTetramer

Mod

elsof

Z-andB-D

NA

Z-DNA

B-DNA

SACminusCIof

Z-Tetra

exptle

SAC-CIof

B-Tetra

exptle

EEa

Oscb

Rotc

UV

CD

EEa

Oscb

Rotc

UV

CD

state

(eV)

(nm)

(au)

(10minus

40cgs)

nature

typed

(eVnm)

(eVnm)

(eV)

(nm)

(au)

(10minus

40cgs)

nature

typed

(eVnm)

(eVnm)

11A

470

264

000

minus663

πrarr

πET

419296

479

259

003

minus041

πrarr

πdG

21A

481

258

002

6611

πrarr

πET

482

257

033

16891

πrarr

πdC

443280

31A

484

256

004

minus5326

πrarr

πdG

419296

484

256

001

minus7454

πrarr

πdC

41A

488

254

015

4553

πrarr

πdC

428290

498

249

007

minus130

πrarr

πET

51A

492

252

012

minus11712

πrarr

πdC

428290

419296

511

243

004

minus622

πrarr

πET

61A

502

247

003

minus3034

πrarr

πET

525

236

000

minus046

πrarr

πET

71A

513

242

001

2737

nrarr

πdG

527

235

001

332

πrarr

πET

81A

515

241

006

14772

πrarr

πdC

451275

545

227

045

minus15609

πrarr

πdG

484256

486255

91A

522

238

001

1978

nrarr

πdG

552

225

002

minus1795

nrarr

πdG

101 A

538

230

071

7299

πrarr

πdG

484256

451275

556

223

013

5929

πrarr

πdG

111 A

541

229

001

minus6858

πrarr

πdG

558

222

009

minus525

πrarr

πdG

121 A

548

226

002

minus1519

nrarr

πdC

563

220

008

minus7436

nrarr

πdG

131 A

553

224

001

minus924

nrarr

πdC

609

204

001

1379

nrarr

πdG

141 A

601

206

001

minus4159

nrarr

πdG

626

198

002

minus464

nrarr

πdG

aExcitatio

nenergyA

llexcitedstates

ofSA

C-CIcalculations

have

been

shifted

tothelower

values

by05eV

inFigure

8bOscillator

strengthcRotatorystrengthddG

representsintram

olecular

excited

states

ofdG

dCrepresentsintram

olecular

excitedstates

ofdC

and

ETrepresentselectron

transfer

excitedstates

from

dGto

dCeRef

4

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5553

state because the stacking interaction is weaker in B-DNA thanin Z-DNA (Figure 1) Therefore the first peak of theexperimental CD spectra originates from the ET excitationfrom dG to dC for Z-DNA but corresponds to theintramolecular excitation in dC for B-DNA The second peakwith a negative sign is assigned to the 81A state The 81A statewhich is the πminusπ of dG is the origin of both the main peaks ofthe UV spectrum and the second peak of the CD spectrumThe ET excited states of the tetramer models were calculated

at a lower energy region than those of the dimer models forboth B- and Z-DNA This result may be related to the overlapbetween the nucleic acid bases The overlap exists only in partof the nucleic acid bases in the dimer models but in thetetramer models the overlap covers the hydrogen bonds as wellas the nucleic acid bases The ET excited states are importantfor the CD spectrum of Z-DNA because the lowest excitedstate is the ET type and is negative for rotatory strength but forB-DNA the ET excited states are not important because thesestates are weak in both oscillator and rotatory strengthsThe SAC-CI results of the tetramer models are calculated at

a higher region by about 05 eV because of the small activespace Therefore better SAC-CI calculations are required forcomparison with the experimental UV and CD spectra of DNAHowever these results indicate that the strong stackinginteraction characteristic of the helical structure of Z-DNAchanges the sign of the rotatory strength for the lowest excitedstates of dG and dC and produces the emergence of the ETexcited states

4 CONCLUSIONSDNA chemistry is affected by three important factors (1) theconformation of the monomer (dG) which is related to thedihedral angle between deoxyribose and guanine (2) thehydrogen-bonding interactions and (3) the stacking inter-actions between two nucleic acid bases To verify whether thesethree factors are important for the CD spectra of DNA weexamined four different models using the SAC-CI method(1) dG monomer model We first applied the SAC-CI

method to the dG monomer to examine the effect ofconformation on the CD spectra of dG and from this resultto identify the structure of the dG monomer in solution Ourcalculations showed that the shape and the sign of the CDspectra depend strongly on the dihedral angle between thedeoxyribose and the guanine but that the SAC-CI UV spectraof both anti- and syn-dG are very similar because the UVspectra are not strongly dependent on the conformation Theexperimental CD spectrum agreed with the theoreticalspectrum of dG only for the anti conformation The resultingthermally averaged spectra also support the anti geometry ofdG as a stable conformation in solution However the SAC-CICD spectrum of syn-dG is opposite in sign from that of theexperimental spectrum Thus the SAC-CI CD spectra of dG inboth anti and syn conformations do not explain the negativesign at 295 nm of the CD spectra of the DNA composed of dGand dC(2) Composit model of dG and dC monomers A

comparison of the calculated composite spectra of dG anddC with the experimental spectra of the two-component DNAshowed similarities between B- and Z-DNA because theconformation of dC is always anti in both B- and Z-DNAThe SAC-CI results led us to confirm that that theconformation of dG is not the origin of the negative peak at295 nm in Z-DNA

(3) Dimer model Our examination of the dimer modelpermitted us to investigate the separate importance of thehydrogen-bonding and the stacking interactions in thestructural chemistry of B- and Z-DNA The SAC-CIcalculations showed that the hydrogen-bonding interactionschange the excitation energies and that the stacking interactionsare responsible for the changes in the sign of the CD rotatorystrength The SAC-CI UV spectra of the hydrogen-bondingmodels (B-Hbond and Z-Hbond models) are similar to theexperimental UV spectra in which the main peak as well as theshoulder peak is well reproduced The SAC-CI CD spectra ofthe stacking models (B-Stack and Z-Stack models) have thesame sign as the experimental CD spectra We concluded thatthe stacking interaction is especially important for the CDspectrum of the strongly stacked Z-DNA because the CD signalat 295 nm changed to a negative sign only for the Z-Stackmodel(4) Tetramer model Finally our calculations with the

tetramer model that contain both hydrogen-bonding andstacking interactions revealed that the shape of the SAC-CIUV and CD spectra compared well with the experimental onesThe ET excitations as well as the lowest intramolecularexcitation of dC and dG are the origins of the negative sign ofthe experimental CD spectrum of Z-DNA The ET excited stateis calculated at a lower energy region than the intramolecularexcited states of dG or dC for the strongly stacked Z-DNA butat a higher energy region for the weakly stacked B-DNA Thenegative sign of the first peak of the experimental CD spectra ofZ-DNA is due to the strong stacking effect characteristic of itshelical structureThe dynamic behavior of the dG monomer in solution was

partially considered by taking the Boltzmann averaging of thecalculated spectra over the various conformations Fortunatelythe Boltzmann averaged SAC-CI CD spectrum was in goodagreement with the experimental spectra although the SAC-CICD spectra largely depend on the conformation of dGHowever in this study we used only the static model for dimerand tetramer models The SAC-CI CD results of the dimer andtetramer models explained the difference between theexperimental CD spectra of B- and Z-DNA However sincethe structure of DNA would also be flexible in solution it isnecessary to take into account the dynamic behavior of DNAThis is an important and interesting topic

ASSOCIATED CONTENT

S Supporting InformationSAC-CI results of the hydrogen-bonding and stacking dimermodels of Z- and B-DNA using small active space and withoutthe polarization d-functions This material is available free ofcharge via the Internet at httppubsacsorg

AUTHOR INFORMATION

Corresponding AuthorPhone +81-75-634-3211 Fax +81-75-634-3211 E-mail hnakatsujiqcriorjp

NotesThe authors declare no competing financial interest

ACKNOWLEDGMENTS

The computations were performed at the Research Center forComputational Science Okazaki Japan

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5554

REFERENCES(1) Voet D Voet J G Eds Biochemistry John Wiley amp Sons IncNew York 1995(2) Ghosh A Bansal M Acta Crystallogr 2003 D59 620minus626(3) Rich A Nordheim A Wang A H-J Annu Rev Biochem 198453 791minus846(4) Tran-Dinh S Taboury J Neumann J-M Huynh-Dinh TGenissel B Langlois drsquoEstaintot B Igolen J Biochemistry 1984 231362minus1371(5) Sugiyama H Kawai K Matsunaga A Fujimoto K Saito IRobinson H Wang A H-J Nucleic Acids Res 1996 24 1272minus1278(6) Kawai K Saito I Sugiyama H J Am Chem Soc 1999 1211391minus1392(7) Oyoshi T Kawai K Sugiyama H J Am Chem Soc 2003 1251526minus1531(8) Xu Y Ikeda R Sugiyama H J Am Chem Soc 2003 12513519minus13524(9) Berova N Nakanishi K Woody R W Eds Circular DichroismPrinciples and Applications Wiley-VCH New York 2000(10) Bauernschmitt R Ahlrichs R Chem Phys Lett 1996 256454minus464(11) Casida M E Jamorski C Casida K C Salahub D R JChem Phys 1998 108 4439minus4449(12) Tsolakidis A Kaxiras E J Phys Chem A 2005 109 2373minus2380(13) Varsano D Di Felice R Marques M A L Rubio A J PhysChem B 2006 110 7129minus7138(14) Nielsen L M Holm A I S Varsano D Kadhane UHoffmann S V Di Felice R Rubio A Nielsen S B J Phys ChemB 2009 113 9614minus9619(15) Tonzani S Schatz G C J Am Chem Soc 2008 130 7607minus7612(16) Casida M E Salahub D R J Chem Phys 2000 113 8918minus8935(17) Dreuw A Weisman J L Head-Gordon M J Chem Phys2003 119 2943minus2946(18) Maitra N T Zhang F Cave R J Burke K J Chem Phys2004 120 5932minus5937(19) McCann D M Stephens P J J Org Chem 2006 71 6074minus6098(20) Casida M E J Mol Struct 2009 914 3minus18(21) Niehaus T A J Mol Struct 2009 914 38minus49(22) Tawada Y Tsuneda T Yanagisawa S Yanai T Hirao K JChem Phys 2004 120 8425minus8433(23) Chiba M Tsuneda T Hirao K J Chem Phys 2006 124144106-1-11(24) Nakatsuji H Hirao K J Chem Phys 1978 68 2053minus2065(25) Nakatsuji H Chem Phys Lett 1978 59 362minus364(26) Nakatsuji H Chem Phys Lett 1979 67 329minus333(27) Nakatsuji H Chem Phys Lett 1979 67 334minus342(28) Nakatsuji H Bull Chem Soc Jpn 2005 78 1705minus1724(29) Ehara M Hasegawa J Nakatsuji H SAC-CI Method Appliedto Molecular Spectroscopy In Theory and Applications of Computa-tional Chemistry The First 40 Years A Vol of Technical and HistoricalPerspectives Dykstra C E Frenking G Kim K S Scuseria G EEds Elsevier Oxford UK 2005 pp 1099minus1141(30) Miyahara T Hasegawa J Nakatsuji H Bull Chem Soc Jpn2009 82 1215minus1226(31) Bureekaew S Hasegawa J Nakatsuji H Chem Phys Lett2006 425 367minus371(32) Nakatani N Hasegawa J Nakatsuji H J Am Chem Soc2007 129 8756minus8765(33) Fujimoto K Hasegawa J Nakatsuji H Bull Chem Soc Jpn2009 82 1140minus1148(34) Miyahara T Nakatsuji H Collect Czech Chem Commun2011 76 537minus552(35) Sugiyama H Unpublished(36) Geertsen J Rittby M Bartlett R J Chem Phys Lett 1989164 57minus62

(37) Stanton J F Bartlett R J J Chem Phys 1993 98 7029minus7039(38) Ehara M Ishida M Toyota K Nakatsuji H SAC-CIGeneral-R Method Theory and Applications to the Multi-ElectronProcesses In Reviews in Modern Quantum Chemistry (A Tribute toProfessor Robert G Parr) Sen K D Ed World Scientific Singapore2003 pp 293minus319(39) SAC-CI GUIDE httpwwwqcriorjpsacci(40) Frisch M J Trucks G W Schlegel H B et al Gaussian 03development revision revision E05 Gaussian Inc Wallingford CT2004(41) Hohenberg P Kohn W Phys Rev 1964 136 B864minusB871(42) Kohn W Sham L J Phys Rev 1965 140 A1133minusA1138(43) Salahub D R Zerner M C Eds The Challenge of d and fElectrons The American Chemical Society Washington DC 1989(44) Parr R G Yang W Density-Functional Theory of Atoms andMolecules Oxford University Press Oxford UK 1989(45) Becke A D J Chem Phys 1993 98 5648minus5652(46) Lee C Yang W Parr R G Phys Rev 1988 37 785minus789(47) Hariharan P C Pople J A Theor Chim Acta 1973 28 213minus222(48) Francl M M Pietro W J Hehre W J Binkley J SDeFrees D J Pople J A Gordon M S J Chem Phys 1982 773654minus3665(49) Tomasi J Mennucci B Cammi R Chem Rev 2005 1052999minus3093(50) Fukuda R Ehara M Nakatsuji H Cammi R J Chem Phys2011 134 104109minus1minus11(51) Dunning T H Jr Hay P J In Modern Theoretical ChemistrySchaefer H F III Ed Plenum New York 1976 Vol 3 pp 1minus28(52) Nakatsuji H Chem Phys 1983 75 425minus441(53) Hocquet A Leulliot N Ghomi M J Phys Chem B 2000 1044560minus4568(54) Mishra S K Mishra P C J Comput Chem 2002 23 530minus540(55) Sprecher C A Johnson W C Biopolymers 1977 16 2243minus2264(56) Clark L B Peschel G G Tinoc I Jr J Phys Chem 1965 693615minus3618(57) Voet D Gratzer W B Cox R A Doty P Biopolymers 19631 193minus208(58) Hobza P Mehlhorn A Carsky P Zahradnik R J Mol Struct1986 138 387minus399(59) Hobza P Sponer J Chem Rev 1999 99 3247minus3276

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5555

Page 10: Helical Structure and Circular Dichroism Spectra of DNA: A ... · B-DNA, have right-handed double-helical structure, while Z-DNA is left-handed and appears only when DNA has a special

positive at 295 nm Therefore the strong negative sign of theCD spectra at 295 nm indicates that a strong stackinginteraction exists in Z-DNA The SAC-CI CD spectra of bothstacking models (B-Stack and Z-Stack models) more closelyresemble the experimental CD spectra of B- and Z-DNA thanthose of the hydrogen-bonding models (B-Hbond and Z-Hbond models) For the SAC-CI CD spectrum of the Z-Stackmodel the first and second peaks are negative and positiverespectively but they are the opposite of those obtained fromthe Z-Hbond model For the B-Stack and B-Hbond modelshowever the first peak is positive in the SAC-CI CD spectrabut the second and third peaks are negative and positiverespectively for the B-Stack model in opposition to the B-Hbond model Since the stacking models do not have thehydrogen-bonding interactions the excitation energy of the firstpeak is lower than that of the experimental spectrum as notedaboveIn the Z-Stack model the first peak with a negative sign is

assigned to the 11A (395 eV) and 21A (421 eV) states (Table4) The 11A state of the Z-Stack model corresponds to the 11Astate (413 eV) of dC which is the excitation of the πminusπ fromthe HOMO to the LUMO in dC A large oscillator strength isthe origin of the first (shoulder) peak but the excitation energyis lower than the experimental value because there is nohydrogen bond in the Z-Stack model The rotatory strength ofthe 11A state of dC changes the sign from positive to negativeby the stacking interaction with dG Therefore the stackinginteraction between dG and dC accounts for the strongnegative sign of the CD spectra in Z-DNA

The excitation energy of the 21A state (421 eV) is also closeto the first peak of the experimental CD spectrum This state isassigned to the 11A state (437 eV) of dG which is the HOMOto LUMO πminusπ excitation in dG but also includes thecontribution of the excitation from the next-HOMO to LUMOin dC The oscillator strength in the UV spectrum is smallbecause this state differs from the 11A state of dG by theinclusion of the excitation in dC Although the excitationenergy is shifted by the formation of hydrogen bonds betweendG and dC the rotatory strength of a large negative maycontribute to the strong negative sign of the CD spectra in Z-DNA In the 31A state (445 eV) which is the excitation fromnonbonding orbital to the LUMO in dC the rotatory strengthchanges to positive from the negative sign of the monomer dCHowever this state is not the origin of the second peak as willbe noted later in the tetramer model section Thus the rotatorystrength is strongly affected by the stacking interactionThree important excited states are calculated in the higher-

energy region of the second peak The 51A state (485 eV) hasthe largest oscillator strength and is assigned to the 31A (491eV) state of dG which is the excitation from HOMO to next-LUMO This state is assigned to the main peak (484 eV) of theexperimental UV spectrum of Z-DNA The 6 and 71A states(499 and 526 eV) with a large rotatory strength are the ETstate from the HOMO of dG to the LUMO of dC The mainconfiguration of the 71A state also includes the excitation fromthe next-HOMO to the LUMO in dCIn the B-Stack model the first peak with a positive sign is

assigned to the 11A state (404 eV) This state corresponds tothe 11A state (409 eV) of dC which is the excitation from theHOMO to the LUMO The 21A state (440 eV) is assigned tothe 11A state (434 eV) of dG which reflects the excitationfrom the HOMO to the LUMO Since the B-Stack model doesnot have hydrogen bonds the excitation energies are close tothose of dG or dC Both states have the oscillator strength butthe 11A state is positive while the 21A state is slightly negativefor the rotatory strength which is the same as for themonomers dC and dG respectively Therefore the stackinginteraction of B-DNA has little effect on the sign of the CDspectra at 295 nm because the stacking interaction is weaker inB-DNA than in Z-DNAThe second peak with a negative sign is assigned to the 5 and

61A states with a large rotatory strength The 5 and 61A states(507 and 512 eV in Table 4) are assigned to the 2 and 31Astates (493 and 518 eV in Table 3) of dG which are theexcitations from the nonbonding orbital and the HOMO to thenext-LUMO The main configuration is nminusπ for the 51A stateand πminusπ for the 61A state Both states have strong oscillatorand negative rotatory strengths which means that they can beassigned to the main peaksThe first ET excited states (71A state 521 eV) of B-Stack

model are calculated at a higher region than that (61A state499 eV) of the Z-Stack model because the stacking interactionis weaker in B-DNA than in Z-DNAThe SAC-CI results show that both hydrogen-bonding and

stacking interactions affect the UV and CD spectra of DNA Inparticular the excitation energy of the first excited stateincreases by the formation of hydrogen bonds and the stackinginteraction changes the sign of the CD spectrum of the firstexcited state from positive to negative The strong stackinginteraction characteristic of the helical structure of Z-DNA isthe origin of the strong negative sign at around 295 nm in theCD spectra The hydrogen-bonding and stacking interactions

Figure 7 SAC-CI CD spectra for the dimer models The SAC-CIspectra (magenta line) of (a) Z-Hbond model (b) Z-Stack model (c)B-Hbond model and (d) B-Stack model are compared with theexperimental CD spectra4 (black line) of Z-DNA (ab) and B-DNA(cd) The red and blue lines represent the intramolecular excitedstates of dG and dC respectively The green lines represent theelectron transfer (ET) excited state from dG to dC

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5551

also confer other smaller effects on the UV and CD spectra ofDNA In the higher energy region the SAC-CI CD spectra arenot in good agreement with the experimental CD spectrabecause these models have either hydrogen-bonding interactionor stacking interaction but not both Therefore we need toinvestigate a better model that includes both hydrogen bondingand stacking interactions within the same model A minimumsuch model is the tetramer model of Figure 1 on which weexplain in the next section35 Tetramer Model The SAC-CI calculations of the

dimer models were suitable to investigate the effects of thehydrogen-bonding interaction and the stacking interactionseparately For example the SAC-CI UV spectra of thehydrogen-bonding models were in good agreement with theexperimental UV spectra implying the importance of thehydrogen-bonding interaction The SAC-CI CD spectra for thestacking model clarified that the stacking interaction is thesource of the negative CD peak at around 295 nm Howevergenerally speaking the SAC-CI CD spectra were markedlydifferent from the experimental spectra The SAC-CI UVspectra of the stacking models were composed of two peaksdifferent from the experimental UV spectra Thus theinformation that we gained from the dimer models still didnot fully resolve these differences in the spectra acquired by twodifferent processes So we used the tetramer model (Figure 1)which included both hydrogen-bonding and stacking inter-actions within the same modelFigure 8 shows the SAC-CI UV and CD spectra of the

tetramer model of Z-DNA (Z-Tetra model) and B-DNA (B-Tetra model) in comparison with the experimental UV andCD spectra In order to compare the experimental spectra theSAC-CI results were shifted to lower energy by 05 eV Thecomputational conditions were poor as compared with theother SAC-CI calculations In the tetramer calculations the d-polarization functions were not added and the small activespace was used The SAC-CI calculations using the small activespace and without the d-polarization functions gave higherexcitation energies by about 04minus05 eV (see SupportingInformation Table S1) for the hydrogen-bonding dimermodels In particular the excited states corresponding to themain peak of the experimental UV spectra were shifted by 048eV for both the Z-Hbond model (61A state) and the B-Hbondmodel (71A state)Table 5 shows the excitation energies oscillator strengths

rotatory strengths natures and type of the excited states of theSAC-CI results for tetramer modelsThe SAC-CI UV spectrum of the Z-Tetra model is in good

agreement with the experimental UV spectrum of Z-DNA Themain peak as well as the shoulder peak is reproduced by theSAC-CI method The shoulder peak is assigned to the 4 and51A states which is the πminusπ excitation (HOMO to LUMO)in dC and the main peak is assigned to the 101A state which isthe πminusπ (HOMO to next-LUMO) excitation in dGThe SAC-CI UV spectrum of the B-Tetra model is also in a

reasonable agreement with the experimental UV spectrum of B-DNA although the shape of the shoulder is slightly different Inthe B-Tetra model the shoulder peak is assigned to the 21Astate which is the πminusπ (HOMO to LUMO) excitation in dCand the main peak is assigned to the 81A state which is theπminusπ (HOMO to next-LUMO) excitation in dGThe sign of the rotatory strength is the same between the

SAC-CI CD spectrum of the Z-Tetra model and theexperimental CD spectrum of Z-DNA although the negative

rotatory strength at 295 nm is very weak because thepolarization functions were not included in the tetramermodels It is well-known that the basis set must be good for anadequate description of the stacking interaction5859 In thedimer model the lack of polarization d-functions and thesmaller active space changed the intensity and sign of therotatory strength (see Supporting Information Table S1)Since the stacking interaction is small in B-DNA the SAC-CI

CD spectrum of the B-Tetra model is in good agreement withthe experimental CD spectrum of B-DNA and reproduces notonly the sign of the rotatory strength but also the overall shapeof the spectrumIn Z-DNA the 11A state is assigned to the first peak of the

experimental CD spectrum The 11A state is the ET excitationfrom the HOMO of dG to the LUMO of dC through thestacking conformation The 3 and 51A states with a largenegative rotatory strength are the excitation from the HOMOto the LUMO in dG and dC respectively These states alsocontribute to the first negative peak as in the 1 and 21A stateswith a negative rotatory strength in the Z-Stack model The 8and 101A states are assigned to the second peak of the CDspectrum The 101A state which is the origin of the main peakof the UV spectrum and the πminusπ of dC (81A state) alsocontribute to the second peak of the CD spectrumIn B-DNA the first peak of the CD spectrum is assigned to

the 21A state of πminusπ (HOMO to LUMO) in dC which is thesame as the origin of the first peak of the UV spectrum The ETstates are calculated at the higher energy region than the 21A

Figure 8 SAC-CI UV and CD spectra for the tetramer models TheSAC-CI spectra (magenta line) of Z-Tetra model (ab) and B-Tetramodel (cd) as compared with the experimental UV and CD spectra4

(black line) of Z-DNA (ab) and B-DNA (cd) The red and blue linesrepresent the intramolecular excited states of dG and dC respectivelyThe green lines represent the electron transfer (ET) excited state fromdG to dC All excited states of SAC-CI calculations have been shiftedto the lower values by 05 eV

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5552

Table

5SA

C-CIUVandCD

Spectraof

theTetramer

Mod

elsof

Z-andB-D

NA

Z-DNA

B-DNA

SACminusCIof

Z-Tetra

exptle

SAC-CIof

B-Tetra

exptle

EEa

Oscb

Rotc

UV

CD

EEa

Oscb

Rotc

UV

CD

state

(eV)

(nm)

(au)

(10minus

40cgs)

nature

typed

(eVnm)

(eVnm)

(eV)

(nm)

(au)

(10minus

40cgs)

nature

typed

(eVnm)

(eVnm)

11A

470

264

000

minus663

πrarr

πET

419296

479

259

003

minus041

πrarr

πdG

21A

481

258

002

6611

πrarr

πET

482

257

033

16891

πrarr

πdC

443280

31A

484

256

004

minus5326

πrarr

πdG

419296

484

256

001

minus7454

πrarr

πdC

41A

488

254

015

4553

πrarr

πdC

428290

498

249

007

minus130

πrarr

πET

51A

492

252

012

minus11712

πrarr

πdC

428290

419296

511

243

004

minus622

πrarr

πET

61A

502

247

003

minus3034

πrarr

πET

525

236

000

minus046

πrarr

πET

71A

513

242

001

2737

nrarr

πdG

527

235

001

332

πrarr

πET

81A

515

241

006

14772

πrarr

πdC

451275

545

227

045

minus15609

πrarr

πdG

484256

486255

91A

522

238

001

1978

nrarr

πdG

552

225

002

minus1795

nrarr

πdG

101 A

538

230

071

7299

πrarr

πdG

484256

451275

556

223

013

5929

πrarr

πdG

111 A

541

229

001

minus6858

πrarr

πdG

558

222

009

minus525

πrarr

πdG

121 A

548

226

002

minus1519

nrarr

πdC

563

220

008

minus7436

nrarr

πdG

131 A

553

224

001

minus924

nrarr

πdC

609

204

001

1379

nrarr

πdG

141 A

601

206

001

minus4159

nrarr

πdG

626

198

002

minus464

nrarr

πdG

aExcitatio

nenergyA

llexcitedstates

ofSA

C-CIcalculations

have

been

shifted

tothelower

values

by05eV

inFigure

8bOscillator

strengthcRotatorystrengthddG

representsintram

olecular

excited

states

ofdG

dCrepresentsintram

olecular

excitedstates

ofdC

and

ETrepresentselectron

transfer

excitedstates

from

dGto

dCeRef

4

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5553

state because the stacking interaction is weaker in B-DNA thanin Z-DNA (Figure 1) Therefore the first peak of theexperimental CD spectra originates from the ET excitationfrom dG to dC for Z-DNA but corresponds to theintramolecular excitation in dC for B-DNA The second peakwith a negative sign is assigned to the 81A state The 81A statewhich is the πminusπ of dG is the origin of both the main peaks ofthe UV spectrum and the second peak of the CD spectrumThe ET excited states of the tetramer models were calculated

at a lower energy region than those of the dimer models forboth B- and Z-DNA This result may be related to the overlapbetween the nucleic acid bases The overlap exists only in partof the nucleic acid bases in the dimer models but in thetetramer models the overlap covers the hydrogen bonds as wellas the nucleic acid bases The ET excited states are importantfor the CD spectrum of Z-DNA because the lowest excitedstate is the ET type and is negative for rotatory strength but forB-DNA the ET excited states are not important because thesestates are weak in both oscillator and rotatory strengthsThe SAC-CI results of the tetramer models are calculated at

a higher region by about 05 eV because of the small activespace Therefore better SAC-CI calculations are required forcomparison with the experimental UV and CD spectra of DNAHowever these results indicate that the strong stackinginteraction characteristic of the helical structure of Z-DNAchanges the sign of the rotatory strength for the lowest excitedstates of dG and dC and produces the emergence of the ETexcited states

4 CONCLUSIONSDNA chemistry is affected by three important factors (1) theconformation of the monomer (dG) which is related to thedihedral angle between deoxyribose and guanine (2) thehydrogen-bonding interactions and (3) the stacking inter-actions between two nucleic acid bases To verify whether thesethree factors are important for the CD spectra of DNA weexamined four different models using the SAC-CI method(1) dG monomer model We first applied the SAC-CI

method to the dG monomer to examine the effect ofconformation on the CD spectra of dG and from this resultto identify the structure of the dG monomer in solution Ourcalculations showed that the shape and the sign of the CDspectra depend strongly on the dihedral angle between thedeoxyribose and the guanine but that the SAC-CI UV spectraof both anti- and syn-dG are very similar because the UVspectra are not strongly dependent on the conformation Theexperimental CD spectrum agreed with the theoreticalspectrum of dG only for the anti conformation The resultingthermally averaged spectra also support the anti geometry ofdG as a stable conformation in solution However the SAC-CICD spectrum of syn-dG is opposite in sign from that of theexperimental spectrum Thus the SAC-CI CD spectra of dG inboth anti and syn conformations do not explain the negativesign at 295 nm of the CD spectra of the DNA composed of dGand dC(2) Composit model of dG and dC monomers A

comparison of the calculated composite spectra of dG anddC with the experimental spectra of the two-component DNAshowed similarities between B- and Z-DNA because theconformation of dC is always anti in both B- and Z-DNAThe SAC-CI results led us to confirm that that theconformation of dG is not the origin of the negative peak at295 nm in Z-DNA

(3) Dimer model Our examination of the dimer modelpermitted us to investigate the separate importance of thehydrogen-bonding and the stacking interactions in thestructural chemistry of B- and Z-DNA The SAC-CIcalculations showed that the hydrogen-bonding interactionschange the excitation energies and that the stacking interactionsare responsible for the changes in the sign of the CD rotatorystrength The SAC-CI UV spectra of the hydrogen-bondingmodels (B-Hbond and Z-Hbond models) are similar to theexperimental UV spectra in which the main peak as well as theshoulder peak is well reproduced The SAC-CI CD spectra ofthe stacking models (B-Stack and Z-Stack models) have thesame sign as the experimental CD spectra We concluded thatthe stacking interaction is especially important for the CDspectrum of the strongly stacked Z-DNA because the CD signalat 295 nm changed to a negative sign only for the Z-Stackmodel(4) Tetramer model Finally our calculations with the

tetramer model that contain both hydrogen-bonding andstacking interactions revealed that the shape of the SAC-CIUV and CD spectra compared well with the experimental onesThe ET excitations as well as the lowest intramolecularexcitation of dC and dG are the origins of the negative sign ofthe experimental CD spectrum of Z-DNA The ET excited stateis calculated at a lower energy region than the intramolecularexcited states of dG or dC for the strongly stacked Z-DNA butat a higher energy region for the weakly stacked B-DNA Thenegative sign of the first peak of the experimental CD spectra ofZ-DNA is due to the strong stacking effect characteristic of itshelical structureThe dynamic behavior of the dG monomer in solution was

partially considered by taking the Boltzmann averaging of thecalculated spectra over the various conformations Fortunatelythe Boltzmann averaged SAC-CI CD spectrum was in goodagreement with the experimental spectra although the SAC-CICD spectra largely depend on the conformation of dGHowever in this study we used only the static model for dimerand tetramer models The SAC-CI CD results of the dimer andtetramer models explained the difference between theexperimental CD spectra of B- and Z-DNA However sincethe structure of DNA would also be flexible in solution it isnecessary to take into account the dynamic behavior of DNAThis is an important and interesting topic

ASSOCIATED CONTENT

S Supporting InformationSAC-CI results of the hydrogen-bonding and stacking dimermodels of Z- and B-DNA using small active space and withoutthe polarization d-functions This material is available free ofcharge via the Internet at httppubsacsorg

AUTHOR INFORMATION

Corresponding AuthorPhone +81-75-634-3211 Fax +81-75-634-3211 E-mail hnakatsujiqcriorjp

NotesThe authors declare no competing financial interest

ACKNOWLEDGMENTS

The computations were performed at the Research Center forComputational Science Okazaki Japan

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5554

REFERENCES(1) Voet D Voet J G Eds Biochemistry John Wiley amp Sons IncNew York 1995(2) Ghosh A Bansal M Acta Crystallogr 2003 D59 620minus626(3) Rich A Nordheim A Wang A H-J Annu Rev Biochem 198453 791minus846(4) Tran-Dinh S Taboury J Neumann J-M Huynh-Dinh TGenissel B Langlois drsquoEstaintot B Igolen J Biochemistry 1984 231362minus1371(5) Sugiyama H Kawai K Matsunaga A Fujimoto K Saito IRobinson H Wang A H-J Nucleic Acids Res 1996 24 1272minus1278(6) Kawai K Saito I Sugiyama H J Am Chem Soc 1999 1211391minus1392(7) Oyoshi T Kawai K Sugiyama H J Am Chem Soc 2003 1251526minus1531(8) Xu Y Ikeda R Sugiyama H J Am Chem Soc 2003 12513519minus13524(9) Berova N Nakanishi K Woody R W Eds Circular DichroismPrinciples and Applications Wiley-VCH New York 2000(10) Bauernschmitt R Ahlrichs R Chem Phys Lett 1996 256454minus464(11) Casida M E Jamorski C Casida K C Salahub D R JChem Phys 1998 108 4439minus4449(12) Tsolakidis A Kaxiras E J Phys Chem A 2005 109 2373minus2380(13) Varsano D Di Felice R Marques M A L Rubio A J PhysChem B 2006 110 7129minus7138(14) Nielsen L M Holm A I S Varsano D Kadhane UHoffmann S V Di Felice R Rubio A Nielsen S B J Phys ChemB 2009 113 9614minus9619(15) Tonzani S Schatz G C J Am Chem Soc 2008 130 7607minus7612(16) Casida M E Salahub D R J Chem Phys 2000 113 8918minus8935(17) Dreuw A Weisman J L Head-Gordon M J Chem Phys2003 119 2943minus2946(18) Maitra N T Zhang F Cave R J Burke K J Chem Phys2004 120 5932minus5937(19) McCann D M Stephens P J J Org Chem 2006 71 6074minus6098(20) Casida M E J Mol Struct 2009 914 3minus18(21) Niehaus T A J Mol Struct 2009 914 38minus49(22) Tawada Y Tsuneda T Yanagisawa S Yanai T Hirao K JChem Phys 2004 120 8425minus8433(23) Chiba M Tsuneda T Hirao K J Chem Phys 2006 124144106-1-11(24) Nakatsuji H Hirao K J Chem Phys 1978 68 2053minus2065(25) Nakatsuji H Chem Phys Lett 1978 59 362minus364(26) Nakatsuji H Chem Phys Lett 1979 67 329minus333(27) Nakatsuji H Chem Phys Lett 1979 67 334minus342(28) Nakatsuji H Bull Chem Soc Jpn 2005 78 1705minus1724(29) Ehara M Hasegawa J Nakatsuji H SAC-CI Method Appliedto Molecular Spectroscopy In Theory and Applications of Computa-tional Chemistry The First 40 Years A Vol of Technical and HistoricalPerspectives Dykstra C E Frenking G Kim K S Scuseria G EEds Elsevier Oxford UK 2005 pp 1099minus1141(30) Miyahara T Hasegawa J Nakatsuji H Bull Chem Soc Jpn2009 82 1215minus1226(31) Bureekaew S Hasegawa J Nakatsuji H Chem Phys Lett2006 425 367minus371(32) Nakatani N Hasegawa J Nakatsuji H J Am Chem Soc2007 129 8756minus8765(33) Fujimoto K Hasegawa J Nakatsuji H Bull Chem Soc Jpn2009 82 1140minus1148(34) Miyahara T Nakatsuji H Collect Czech Chem Commun2011 76 537minus552(35) Sugiyama H Unpublished(36) Geertsen J Rittby M Bartlett R J Chem Phys Lett 1989164 57minus62

(37) Stanton J F Bartlett R J J Chem Phys 1993 98 7029minus7039(38) Ehara M Ishida M Toyota K Nakatsuji H SAC-CIGeneral-R Method Theory and Applications to the Multi-ElectronProcesses In Reviews in Modern Quantum Chemistry (A Tribute toProfessor Robert G Parr) Sen K D Ed World Scientific Singapore2003 pp 293minus319(39) SAC-CI GUIDE httpwwwqcriorjpsacci(40) Frisch M J Trucks G W Schlegel H B et al Gaussian 03development revision revision E05 Gaussian Inc Wallingford CT2004(41) Hohenberg P Kohn W Phys Rev 1964 136 B864minusB871(42) Kohn W Sham L J Phys Rev 1965 140 A1133minusA1138(43) Salahub D R Zerner M C Eds The Challenge of d and fElectrons The American Chemical Society Washington DC 1989(44) Parr R G Yang W Density-Functional Theory of Atoms andMolecules Oxford University Press Oxford UK 1989(45) Becke A D J Chem Phys 1993 98 5648minus5652(46) Lee C Yang W Parr R G Phys Rev 1988 37 785minus789(47) Hariharan P C Pople J A Theor Chim Acta 1973 28 213minus222(48) Francl M M Pietro W J Hehre W J Binkley J SDeFrees D J Pople J A Gordon M S J Chem Phys 1982 773654minus3665(49) Tomasi J Mennucci B Cammi R Chem Rev 2005 1052999minus3093(50) Fukuda R Ehara M Nakatsuji H Cammi R J Chem Phys2011 134 104109minus1minus11(51) Dunning T H Jr Hay P J In Modern Theoretical ChemistrySchaefer H F III Ed Plenum New York 1976 Vol 3 pp 1minus28(52) Nakatsuji H Chem Phys 1983 75 425minus441(53) Hocquet A Leulliot N Ghomi M J Phys Chem B 2000 1044560minus4568(54) Mishra S K Mishra P C J Comput Chem 2002 23 530minus540(55) Sprecher C A Johnson W C Biopolymers 1977 16 2243minus2264(56) Clark L B Peschel G G Tinoc I Jr J Phys Chem 1965 693615minus3618(57) Voet D Gratzer W B Cox R A Doty P Biopolymers 19631 193minus208(58) Hobza P Mehlhorn A Carsky P Zahradnik R J Mol Struct1986 138 387minus399(59) Hobza P Sponer J Chem Rev 1999 99 3247minus3276

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5555

Page 11: Helical Structure and Circular Dichroism Spectra of DNA: A ... · B-DNA, have right-handed double-helical structure, while Z-DNA is left-handed and appears only when DNA has a special

also confer other smaller effects on the UV and CD spectra ofDNA In the higher energy region the SAC-CI CD spectra arenot in good agreement with the experimental CD spectrabecause these models have either hydrogen-bonding interactionor stacking interaction but not both Therefore we need toinvestigate a better model that includes both hydrogen bondingand stacking interactions within the same model A minimumsuch model is the tetramer model of Figure 1 on which weexplain in the next section35 Tetramer Model The SAC-CI calculations of the

dimer models were suitable to investigate the effects of thehydrogen-bonding interaction and the stacking interactionseparately For example the SAC-CI UV spectra of thehydrogen-bonding models were in good agreement with theexperimental UV spectra implying the importance of thehydrogen-bonding interaction The SAC-CI CD spectra for thestacking model clarified that the stacking interaction is thesource of the negative CD peak at around 295 nm Howevergenerally speaking the SAC-CI CD spectra were markedlydifferent from the experimental spectra The SAC-CI UVspectra of the stacking models were composed of two peaksdifferent from the experimental UV spectra Thus theinformation that we gained from the dimer models still didnot fully resolve these differences in the spectra acquired by twodifferent processes So we used the tetramer model (Figure 1)which included both hydrogen-bonding and stacking inter-actions within the same modelFigure 8 shows the SAC-CI UV and CD spectra of the

tetramer model of Z-DNA (Z-Tetra model) and B-DNA (B-Tetra model) in comparison with the experimental UV andCD spectra In order to compare the experimental spectra theSAC-CI results were shifted to lower energy by 05 eV Thecomputational conditions were poor as compared with theother SAC-CI calculations In the tetramer calculations the d-polarization functions were not added and the small activespace was used The SAC-CI calculations using the small activespace and without the d-polarization functions gave higherexcitation energies by about 04minus05 eV (see SupportingInformation Table S1) for the hydrogen-bonding dimermodels In particular the excited states corresponding to themain peak of the experimental UV spectra were shifted by 048eV for both the Z-Hbond model (61A state) and the B-Hbondmodel (71A state)Table 5 shows the excitation energies oscillator strengths

rotatory strengths natures and type of the excited states of theSAC-CI results for tetramer modelsThe SAC-CI UV spectrum of the Z-Tetra model is in good

agreement with the experimental UV spectrum of Z-DNA Themain peak as well as the shoulder peak is reproduced by theSAC-CI method The shoulder peak is assigned to the 4 and51A states which is the πminusπ excitation (HOMO to LUMO)in dC and the main peak is assigned to the 101A state which isthe πminusπ (HOMO to next-LUMO) excitation in dGThe SAC-CI UV spectrum of the B-Tetra model is also in a

reasonable agreement with the experimental UV spectrum of B-DNA although the shape of the shoulder is slightly different Inthe B-Tetra model the shoulder peak is assigned to the 21Astate which is the πminusπ (HOMO to LUMO) excitation in dCand the main peak is assigned to the 81A state which is theπminusπ (HOMO to next-LUMO) excitation in dGThe sign of the rotatory strength is the same between the

SAC-CI CD spectrum of the Z-Tetra model and theexperimental CD spectrum of Z-DNA although the negative

rotatory strength at 295 nm is very weak because thepolarization functions were not included in the tetramermodels It is well-known that the basis set must be good for anadequate description of the stacking interaction5859 In thedimer model the lack of polarization d-functions and thesmaller active space changed the intensity and sign of therotatory strength (see Supporting Information Table S1)Since the stacking interaction is small in B-DNA the SAC-CI

CD spectrum of the B-Tetra model is in good agreement withthe experimental CD spectrum of B-DNA and reproduces notonly the sign of the rotatory strength but also the overall shapeof the spectrumIn Z-DNA the 11A state is assigned to the first peak of the

experimental CD spectrum The 11A state is the ET excitationfrom the HOMO of dG to the LUMO of dC through thestacking conformation The 3 and 51A states with a largenegative rotatory strength are the excitation from the HOMOto the LUMO in dG and dC respectively These states alsocontribute to the first negative peak as in the 1 and 21A stateswith a negative rotatory strength in the Z-Stack model The 8and 101A states are assigned to the second peak of the CDspectrum The 101A state which is the origin of the main peakof the UV spectrum and the πminusπ of dC (81A state) alsocontribute to the second peak of the CD spectrumIn B-DNA the first peak of the CD spectrum is assigned to

the 21A state of πminusπ (HOMO to LUMO) in dC which is thesame as the origin of the first peak of the UV spectrum The ETstates are calculated at the higher energy region than the 21A

Figure 8 SAC-CI UV and CD spectra for the tetramer models TheSAC-CI spectra (magenta line) of Z-Tetra model (ab) and B-Tetramodel (cd) as compared with the experimental UV and CD spectra4

(black line) of Z-DNA (ab) and B-DNA (cd) The red and blue linesrepresent the intramolecular excited states of dG and dC respectivelyThe green lines represent the electron transfer (ET) excited state fromdG to dC All excited states of SAC-CI calculations have been shiftedto the lower values by 05 eV

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5552

Table

5SA

C-CIUVandCD

Spectraof

theTetramer

Mod

elsof

Z-andB-D

NA

Z-DNA

B-DNA

SACminusCIof

Z-Tetra

exptle

SAC-CIof

B-Tetra

exptle

EEa

Oscb

Rotc

UV

CD

EEa

Oscb

Rotc

UV

CD

state

(eV)

(nm)

(au)

(10minus

40cgs)

nature

typed

(eVnm)

(eVnm)

(eV)

(nm)

(au)

(10minus

40cgs)

nature

typed

(eVnm)

(eVnm)

11A

470

264

000

minus663

πrarr

πET

419296

479

259

003

minus041

πrarr

πdG

21A

481

258

002

6611

πrarr

πET

482

257

033

16891

πrarr

πdC

443280

31A

484

256

004

minus5326

πrarr

πdG

419296

484

256

001

minus7454

πrarr

πdC

41A

488

254

015

4553

πrarr

πdC

428290

498

249

007

minus130

πrarr

πET

51A

492

252

012

minus11712

πrarr

πdC

428290

419296

511

243

004

minus622

πrarr

πET

61A

502

247

003

minus3034

πrarr

πET

525

236

000

minus046

πrarr

πET

71A

513

242

001

2737

nrarr

πdG

527

235

001

332

πrarr

πET

81A

515

241

006

14772

πrarr

πdC

451275

545

227

045

minus15609

πrarr

πdG

484256

486255

91A

522

238

001

1978

nrarr

πdG

552

225

002

minus1795

nrarr

πdG

101 A

538

230

071

7299

πrarr

πdG

484256

451275

556

223

013

5929

πrarr

πdG

111 A

541

229

001

minus6858

πrarr

πdG

558

222

009

minus525

πrarr

πdG

121 A

548

226

002

minus1519

nrarr

πdC

563

220

008

minus7436

nrarr

πdG

131 A

553

224

001

minus924

nrarr

πdC

609

204

001

1379

nrarr

πdG

141 A

601

206

001

minus4159

nrarr

πdG

626

198

002

minus464

nrarr

πdG

aExcitatio

nenergyA

llexcitedstates

ofSA

C-CIcalculations

have

been

shifted

tothelower

values

by05eV

inFigure

8bOscillator

strengthcRotatorystrengthddG

representsintram

olecular

excited

states

ofdG

dCrepresentsintram

olecular

excitedstates

ofdC

and

ETrepresentselectron

transfer

excitedstates

from

dGto

dCeRef

4

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5553

state because the stacking interaction is weaker in B-DNA thanin Z-DNA (Figure 1) Therefore the first peak of theexperimental CD spectra originates from the ET excitationfrom dG to dC for Z-DNA but corresponds to theintramolecular excitation in dC for B-DNA The second peakwith a negative sign is assigned to the 81A state The 81A statewhich is the πminusπ of dG is the origin of both the main peaks ofthe UV spectrum and the second peak of the CD spectrumThe ET excited states of the tetramer models were calculated

at a lower energy region than those of the dimer models forboth B- and Z-DNA This result may be related to the overlapbetween the nucleic acid bases The overlap exists only in partof the nucleic acid bases in the dimer models but in thetetramer models the overlap covers the hydrogen bonds as wellas the nucleic acid bases The ET excited states are importantfor the CD spectrum of Z-DNA because the lowest excitedstate is the ET type and is negative for rotatory strength but forB-DNA the ET excited states are not important because thesestates are weak in both oscillator and rotatory strengthsThe SAC-CI results of the tetramer models are calculated at

a higher region by about 05 eV because of the small activespace Therefore better SAC-CI calculations are required forcomparison with the experimental UV and CD spectra of DNAHowever these results indicate that the strong stackinginteraction characteristic of the helical structure of Z-DNAchanges the sign of the rotatory strength for the lowest excitedstates of dG and dC and produces the emergence of the ETexcited states

4 CONCLUSIONSDNA chemistry is affected by three important factors (1) theconformation of the monomer (dG) which is related to thedihedral angle between deoxyribose and guanine (2) thehydrogen-bonding interactions and (3) the stacking inter-actions between two nucleic acid bases To verify whether thesethree factors are important for the CD spectra of DNA weexamined four different models using the SAC-CI method(1) dG monomer model We first applied the SAC-CI

method to the dG monomer to examine the effect ofconformation on the CD spectra of dG and from this resultto identify the structure of the dG monomer in solution Ourcalculations showed that the shape and the sign of the CDspectra depend strongly on the dihedral angle between thedeoxyribose and the guanine but that the SAC-CI UV spectraof both anti- and syn-dG are very similar because the UVspectra are not strongly dependent on the conformation Theexperimental CD spectrum agreed with the theoreticalspectrum of dG only for the anti conformation The resultingthermally averaged spectra also support the anti geometry ofdG as a stable conformation in solution However the SAC-CICD spectrum of syn-dG is opposite in sign from that of theexperimental spectrum Thus the SAC-CI CD spectra of dG inboth anti and syn conformations do not explain the negativesign at 295 nm of the CD spectra of the DNA composed of dGand dC(2) Composit model of dG and dC monomers A

comparison of the calculated composite spectra of dG anddC with the experimental spectra of the two-component DNAshowed similarities between B- and Z-DNA because theconformation of dC is always anti in both B- and Z-DNAThe SAC-CI results led us to confirm that that theconformation of dG is not the origin of the negative peak at295 nm in Z-DNA

(3) Dimer model Our examination of the dimer modelpermitted us to investigate the separate importance of thehydrogen-bonding and the stacking interactions in thestructural chemistry of B- and Z-DNA The SAC-CIcalculations showed that the hydrogen-bonding interactionschange the excitation energies and that the stacking interactionsare responsible for the changes in the sign of the CD rotatorystrength The SAC-CI UV spectra of the hydrogen-bondingmodels (B-Hbond and Z-Hbond models) are similar to theexperimental UV spectra in which the main peak as well as theshoulder peak is well reproduced The SAC-CI CD spectra ofthe stacking models (B-Stack and Z-Stack models) have thesame sign as the experimental CD spectra We concluded thatthe stacking interaction is especially important for the CDspectrum of the strongly stacked Z-DNA because the CD signalat 295 nm changed to a negative sign only for the Z-Stackmodel(4) Tetramer model Finally our calculations with the

tetramer model that contain both hydrogen-bonding andstacking interactions revealed that the shape of the SAC-CIUV and CD spectra compared well with the experimental onesThe ET excitations as well as the lowest intramolecularexcitation of dC and dG are the origins of the negative sign ofthe experimental CD spectrum of Z-DNA The ET excited stateis calculated at a lower energy region than the intramolecularexcited states of dG or dC for the strongly stacked Z-DNA butat a higher energy region for the weakly stacked B-DNA Thenegative sign of the first peak of the experimental CD spectra ofZ-DNA is due to the strong stacking effect characteristic of itshelical structureThe dynamic behavior of the dG monomer in solution was

partially considered by taking the Boltzmann averaging of thecalculated spectra over the various conformations Fortunatelythe Boltzmann averaged SAC-CI CD spectrum was in goodagreement with the experimental spectra although the SAC-CICD spectra largely depend on the conformation of dGHowever in this study we used only the static model for dimerand tetramer models The SAC-CI CD results of the dimer andtetramer models explained the difference between theexperimental CD spectra of B- and Z-DNA However sincethe structure of DNA would also be flexible in solution it isnecessary to take into account the dynamic behavior of DNAThis is an important and interesting topic

ASSOCIATED CONTENT

S Supporting InformationSAC-CI results of the hydrogen-bonding and stacking dimermodels of Z- and B-DNA using small active space and withoutthe polarization d-functions This material is available free ofcharge via the Internet at httppubsacsorg

AUTHOR INFORMATION

Corresponding AuthorPhone +81-75-634-3211 Fax +81-75-634-3211 E-mail hnakatsujiqcriorjp

NotesThe authors declare no competing financial interest

ACKNOWLEDGMENTS

The computations were performed at the Research Center forComputational Science Okazaki Japan

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5554

REFERENCES(1) Voet D Voet J G Eds Biochemistry John Wiley amp Sons IncNew York 1995(2) Ghosh A Bansal M Acta Crystallogr 2003 D59 620minus626(3) Rich A Nordheim A Wang A H-J Annu Rev Biochem 198453 791minus846(4) Tran-Dinh S Taboury J Neumann J-M Huynh-Dinh TGenissel B Langlois drsquoEstaintot B Igolen J Biochemistry 1984 231362minus1371(5) Sugiyama H Kawai K Matsunaga A Fujimoto K Saito IRobinson H Wang A H-J Nucleic Acids Res 1996 24 1272minus1278(6) Kawai K Saito I Sugiyama H J Am Chem Soc 1999 1211391minus1392(7) Oyoshi T Kawai K Sugiyama H J Am Chem Soc 2003 1251526minus1531(8) Xu Y Ikeda R Sugiyama H J Am Chem Soc 2003 12513519minus13524(9) Berova N Nakanishi K Woody R W Eds Circular DichroismPrinciples and Applications Wiley-VCH New York 2000(10) Bauernschmitt R Ahlrichs R Chem Phys Lett 1996 256454minus464(11) Casida M E Jamorski C Casida K C Salahub D R JChem Phys 1998 108 4439minus4449(12) Tsolakidis A Kaxiras E J Phys Chem A 2005 109 2373minus2380(13) Varsano D Di Felice R Marques M A L Rubio A J PhysChem B 2006 110 7129minus7138(14) Nielsen L M Holm A I S Varsano D Kadhane UHoffmann S V Di Felice R Rubio A Nielsen S B J Phys ChemB 2009 113 9614minus9619(15) Tonzani S Schatz G C J Am Chem Soc 2008 130 7607minus7612(16) Casida M E Salahub D R J Chem Phys 2000 113 8918minus8935(17) Dreuw A Weisman J L Head-Gordon M J Chem Phys2003 119 2943minus2946(18) Maitra N T Zhang F Cave R J Burke K J Chem Phys2004 120 5932minus5937(19) McCann D M Stephens P J J Org Chem 2006 71 6074minus6098(20) Casida M E J Mol Struct 2009 914 3minus18(21) Niehaus T A J Mol Struct 2009 914 38minus49(22) Tawada Y Tsuneda T Yanagisawa S Yanai T Hirao K JChem Phys 2004 120 8425minus8433(23) Chiba M Tsuneda T Hirao K J Chem Phys 2006 124144106-1-11(24) Nakatsuji H Hirao K J Chem Phys 1978 68 2053minus2065(25) Nakatsuji H Chem Phys Lett 1978 59 362minus364(26) Nakatsuji H Chem Phys Lett 1979 67 329minus333(27) Nakatsuji H Chem Phys Lett 1979 67 334minus342(28) Nakatsuji H Bull Chem Soc Jpn 2005 78 1705minus1724(29) Ehara M Hasegawa J Nakatsuji H SAC-CI Method Appliedto Molecular Spectroscopy In Theory and Applications of Computa-tional Chemistry The First 40 Years A Vol of Technical and HistoricalPerspectives Dykstra C E Frenking G Kim K S Scuseria G EEds Elsevier Oxford UK 2005 pp 1099minus1141(30) Miyahara T Hasegawa J Nakatsuji H Bull Chem Soc Jpn2009 82 1215minus1226(31) Bureekaew S Hasegawa J Nakatsuji H Chem Phys Lett2006 425 367minus371(32) Nakatani N Hasegawa J Nakatsuji H J Am Chem Soc2007 129 8756minus8765(33) Fujimoto K Hasegawa J Nakatsuji H Bull Chem Soc Jpn2009 82 1140minus1148(34) Miyahara T Nakatsuji H Collect Czech Chem Commun2011 76 537minus552(35) Sugiyama H Unpublished(36) Geertsen J Rittby M Bartlett R J Chem Phys Lett 1989164 57minus62

(37) Stanton J F Bartlett R J J Chem Phys 1993 98 7029minus7039(38) Ehara M Ishida M Toyota K Nakatsuji H SAC-CIGeneral-R Method Theory and Applications to the Multi-ElectronProcesses In Reviews in Modern Quantum Chemistry (A Tribute toProfessor Robert G Parr) Sen K D Ed World Scientific Singapore2003 pp 293minus319(39) SAC-CI GUIDE httpwwwqcriorjpsacci(40) Frisch M J Trucks G W Schlegel H B et al Gaussian 03development revision revision E05 Gaussian Inc Wallingford CT2004(41) Hohenberg P Kohn W Phys Rev 1964 136 B864minusB871(42) Kohn W Sham L J Phys Rev 1965 140 A1133minusA1138(43) Salahub D R Zerner M C Eds The Challenge of d and fElectrons The American Chemical Society Washington DC 1989(44) Parr R G Yang W Density-Functional Theory of Atoms andMolecules Oxford University Press Oxford UK 1989(45) Becke A D J Chem Phys 1993 98 5648minus5652(46) Lee C Yang W Parr R G Phys Rev 1988 37 785minus789(47) Hariharan P C Pople J A Theor Chim Acta 1973 28 213minus222(48) Francl M M Pietro W J Hehre W J Binkley J SDeFrees D J Pople J A Gordon M S J Chem Phys 1982 773654minus3665(49) Tomasi J Mennucci B Cammi R Chem Rev 2005 1052999minus3093(50) Fukuda R Ehara M Nakatsuji H Cammi R J Chem Phys2011 134 104109minus1minus11(51) Dunning T H Jr Hay P J In Modern Theoretical ChemistrySchaefer H F III Ed Plenum New York 1976 Vol 3 pp 1minus28(52) Nakatsuji H Chem Phys 1983 75 425minus441(53) Hocquet A Leulliot N Ghomi M J Phys Chem B 2000 1044560minus4568(54) Mishra S K Mishra P C J Comput Chem 2002 23 530minus540(55) Sprecher C A Johnson W C Biopolymers 1977 16 2243minus2264(56) Clark L B Peschel G G Tinoc I Jr J Phys Chem 1965 693615minus3618(57) Voet D Gratzer W B Cox R A Doty P Biopolymers 19631 193minus208(58) Hobza P Mehlhorn A Carsky P Zahradnik R J Mol Struct1986 138 387minus399(59) Hobza P Sponer J Chem Rev 1999 99 3247minus3276

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5555

Page 12: Helical Structure and Circular Dichroism Spectra of DNA: A ... · B-DNA, have right-handed double-helical structure, while Z-DNA is left-handed and appears only when DNA has a special

Table

5SA

C-CIUVandCD

Spectraof

theTetramer

Mod

elsof

Z-andB-D

NA

Z-DNA

B-DNA

SACminusCIof

Z-Tetra

exptle

SAC-CIof

B-Tetra

exptle

EEa

Oscb

Rotc

UV

CD

EEa

Oscb

Rotc

UV

CD

state

(eV)

(nm)

(au)

(10minus

40cgs)

nature

typed

(eVnm)

(eVnm)

(eV)

(nm)

(au)

(10minus

40cgs)

nature

typed

(eVnm)

(eVnm)

11A

470

264

000

minus663

πrarr

πET

419296

479

259

003

minus041

πrarr

πdG

21A

481

258

002

6611

πrarr

πET

482

257

033

16891

πrarr

πdC

443280

31A

484

256

004

minus5326

πrarr

πdG

419296

484

256

001

minus7454

πrarr

πdC

41A

488

254

015

4553

πrarr

πdC

428290

498

249

007

minus130

πrarr

πET

51A

492

252

012

minus11712

πrarr

πdC

428290

419296

511

243

004

minus622

πrarr

πET

61A

502

247

003

minus3034

πrarr

πET

525

236

000

minus046

πrarr

πET

71A

513

242

001

2737

nrarr

πdG

527

235

001

332

πrarr

πET

81A

515

241

006

14772

πrarr

πdC

451275

545

227

045

minus15609

πrarr

πdG

484256

486255

91A

522

238

001

1978

nrarr

πdG

552

225

002

minus1795

nrarr

πdG

101 A

538

230

071

7299

πrarr

πdG

484256

451275

556

223

013

5929

πrarr

πdG

111 A

541

229

001

minus6858

πrarr

πdG

558

222

009

minus525

πrarr

πdG

121 A

548

226

002

minus1519

nrarr

πdC

563

220

008

minus7436

nrarr

πdG

131 A

553

224

001

minus924

nrarr

πdC

609

204

001

1379

nrarr

πdG

141 A

601

206

001

minus4159

nrarr

πdG

626

198

002

minus464

nrarr

πdG

aExcitatio

nenergyA

llexcitedstates

ofSA

C-CIcalculations

have

been

shifted

tothelower

values

by05eV

inFigure

8bOscillator

strengthcRotatorystrengthddG

representsintram

olecular

excited

states

ofdG

dCrepresentsintram

olecular

excitedstates

ofdC

and

ETrepresentselectron

transfer

excitedstates

from

dGto

dCeRef

4

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5553

state because the stacking interaction is weaker in B-DNA thanin Z-DNA (Figure 1) Therefore the first peak of theexperimental CD spectra originates from the ET excitationfrom dG to dC for Z-DNA but corresponds to theintramolecular excitation in dC for B-DNA The second peakwith a negative sign is assigned to the 81A state The 81A statewhich is the πminusπ of dG is the origin of both the main peaks ofthe UV spectrum and the second peak of the CD spectrumThe ET excited states of the tetramer models were calculated

at a lower energy region than those of the dimer models forboth B- and Z-DNA This result may be related to the overlapbetween the nucleic acid bases The overlap exists only in partof the nucleic acid bases in the dimer models but in thetetramer models the overlap covers the hydrogen bonds as wellas the nucleic acid bases The ET excited states are importantfor the CD spectrum of Z-DNA because the lowest excitedstate is the ET type and is negative for rotatory strength but forB-DNA the ET excited states are not important because thesestates are weak in both oscillator and rotatory strengthsThe SAC-CI results of the tetramer models are calculated at

a higher region by about 05 eV because of the small activespace Therefore better SAC-CI calculations are required forcomparison with the experimental UV and CD spectra of DNAHowever these results indicate that the strong stackinginteraction characteristic of the helical structure of Z-DNAchanges the sign of the rotatory strength for the lowest excitedstates of dG and dC and produces the emergence of the ETexcited states

4 CONCLUSIONSDNA chemistry is affected by three important factors (1) theconformation of the monomer (dG) which is related to thedihedral angle between deoxyribose and guanine (2) thehydrogen-bonding interactions and (3) the stacking inter-actions between two nucleic acid bases To verify whether thesethree factors are important for the CD spectra of DNA weexamined four different models using the SAC-CI method(1) dG monomer model We first applied the SAC-CI

method to the dG monomer to examine the effect ofconformation on the CD spectra of dG and from this resultto identify the structure of the dG monomer in solution Ourcalculations showed that the shape and the sign of the CDspectra depend strongly on the dihedral angle between thedeoxyribose and the guanine but that the SAC-CI UV spectraof both anti- and syn-dG are very similar because the UVspectra are not strongly dependent on the conformation Theexperimental CD spectrum agreed with the theoreticalspectrum of dG only for the anti conformation The resultingthermally averaged spectra also support the anti geometry ofdG as a stable conformation in solution However the SAC-CICD spectrum of syn-dG is opposite in sign from that of theexperimental spectrum Thus the SAC-CI CD spectra of dG inboth anti and syn conformations do not explain the negativesign at 295 nm of the CD spectra of the DNA composed of dGand dC(2) Composit model of dG and dC monomers A

comparison of the calculated composite spectra of dG anddC with the experimental spectra of the two-component DNAshowed similarities between B- and Z-DNA because theconformation of dC is always anti in both B- and Z-DNAThe SAC-CI results led us to confirm that that theconformation of dG is not the origin of the negative peak at295 nm in Z-DNA

(3) Dimer model Our examination of the dimer modelpermitted us to investigate the separate importance of thehydrogen-bonding and the stacking interactions in thestructural chemistry of B- and Z-DNA The SAC-CIcalculations showed that the hydrogen-bonding interactionschange the excitation energies and that the stacking interactionsare responsible for the changes in the sign of the CD rotatorystrength The SAC-CI UV spectra of the hydrogen-bondingmodels (B-Hbond and Z-Hbond models) are similar to theexperimental UV spectra in which the main peak as well as theshoulder peak is well reproduced The SAC-CI CD spectra ofthe stacking models (B-Stack and Z-Stack models) have thesame sign as the experimental CD spectra We concluded thatthe stacking interaction is especially important for the CDspectrum of the strongly stacked Z-DNA because the CD signalat 295 nm changed to a negative sign only for the Z-Stackmodel(4) Tetramer model Finally our calculations with the

tetramer model that contain both hydrogen-bonding andstacking interactions revealed that the shape of the SAC-CIUV and CD spectra compared well with the experimental onesThe ET excitations as well as the lowest intramolecularexcitation of dC and dG are the origins of the negative sign ofthe experimental CD spectrum of Z-DNA The ET excited stateis calculated at a lower energy region than the intramolecularexcited states of dG or dC for the strongly stacked Z-DNA butat a higher energy region for the weakly stacked B-DNA Thenegative sign of the first peak of the experimental CD spectra ofZ-DNA is due to the strong stacking effect characteristic of itshelical structureThe dynamic behavior of the dG monomer in solution was

partially considered by taking the Boltzmann averaging of thecalculated spectra over the various conformations Fortunatelythe Boltzmann averaged SAC-CI CD spectrum was in goodagreement with the experimental spectra although the SAC-CICD spectra largely depend on the conformation of dGHowever in this study we used only the static model for dimerand tetramer models The SAC-CI CD results of the dimer andtetramer models explained the difference between theexperimental CD spectra of B- and Z-DNA However sincethe structure of DNA would also be flexible in solution it isnecessary to take into account the dynamic behavior of DNAThis is an important and interesting topic

ASSOCIATED CONTENT

S Supporting InformationSAC-CI results of the hydrogen-bonding and stacking dimermodels of Z- and B-DNA using small active space and withoutthe polarization d-functions This material is available free ofcharge via the Internet at httppubsacsorg

AUTHOR INFORMATION

Corresponding AuthorPhone +81-75-634-3211 Fax +81-75-634-3211 E-mail hnakatsujiqcriorjp

NotesThe authors declare no competing financial interest

ACKNOWLEDGMENTS

The computations were performed at the Research Center forComputational Science Okazaki Japan

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5554

REFERENCES(1) Voet D Voet J G Eds Biochemistry John Wiley amp Sons IncNew York 1995(2) Ghosh A Bansal M Acta Crystallogr 2003 D59 620minus626(3) Rich A Nordheim A Wang A H-J Annu Rev Biochem 198453 791minus846(4) Tran-Dinh S Taboury J Neumann J-M Huynh-Dinh TGenissel B Langlois drsquoEstaintot B Igolen J Biochemistry 1984 231362minus1371(5) Sugiyama H Kawai K Matsunaga A Fujimoto K Saito IRobinson H Wang A H-J Nucleic Acids Res 1996 24 1272minus1278(6) Kawai K Saito I Sugiyama H J Am Chem Soc 1999 1211391minus1392(7) Oyoshi T Kawai K Sugiyama H J Am Chem Soc 2003 1251526minus1531(8) Xu Y Ikeda R Sugiyama H J Am Chem Soc 2003 12513519minus13524(9) Berova N Nakanishi K Woody R W Eds Circular DichroismPrinciples and Applications Wiley-VCH New York 2000(10) Bauernschmitt R Ahlrichs R Chem Phys Lett 1996 256454minus464(11) Casida M E Jamorski C Casida K C Salahub D R JChem Phys 1998 108 4439minus4449(12) Tsolakidis A Kaxiras E J Phys Chem A 2005 109 2373minus2380(13) Varsano D Di Felice R Marques M A L Rubio A J PhysChem B 2006 110 7129minus7138(14) Nielsen L M Holm A I S Varsano D Kadhane UHoffmann S V Di Felice R Rubio A Nielsen S B J Phys ChemB 2009 113 9614minus9619(15) Tonzani S Schatz G C J Am Chem Soc 2008 130 7607minus7612(16) Casida M E Salahub D R J Chem Phys 2000 113 8918minus8935(17) Dreuw A Weisman J L Head-Gordon M J Chem Phys2003 119 2943minus2946(18) Maitra N T Zhang F Cave R J Burke K J Chem Phys2004 120 5932minus5937(19) McCann D M Stephens P J J Org Chem 2006 71 6074minus6098(20) Casida M E J Mol Struct 2009 914 3minus18(21) Niehaus T A J Mol Struct 2009 914 38minus49(22) Tawada Y Tsuneda T Yanagisawa S Yanai T Hirao K JChem Phys 2004 120 8425minus8433(23) Chiba M Tsuneda T Hirao K J Chem Phys 2006 124144106-1-11(24) Nakatsuji H Hirao K J Chem Phys 1978 68 2053minus2065(25) Nakatsuji H Chem Phys Lett 1978 59 362minus364(26) Nakatsuji H Chem Phys Lett 1979 67 329minus333(27) Nakatsuji H Chem Phys Lett 1979 67 334minus342(28) Nakatsuji H Bull Chem Soc Jpn 2005 78 1705minus1724(29) Ehara M Hasegawa J Nakatsuji H SAC-CI Method Appliedto Molecular Spectroscopy In Theory and Applications of Computa-tional Chemistry The First 40 Years A Vol of Technical and HistoricalPerspectives Dykstra C E Frenking G Kim K S Scuseria G EEds Elsevier Oxford UK 2005 pp 1099minus1141(30) Miyahara T Hasegawa J Nakatsuji H Bull Chem Soc Jpn2009 82 1215minus1226(31) Bureekaew S Hasegawa J Nakatsuji H Chem Phys Lett2006 425 367minus371(32) Nakatani N Hasegawa J Nakatsuji H J Am Chem Soc2007 129 8756minus8765(33) Fujimoto K Hasegawa J Nakatsuji H Bull Chem Soc Jpn2009 82 1140minus1148(34) Miyahara T Nakatsuji H Collect Czech Chem Commun2011 76 537minus552(35) Sugiyama H Unpublished(36) Geertsen J Rittby M Bartlett R J Chem Phys Lett 1989164 57minus62

(37) Stanton J F Bartlett R J J Chem Phys 1993 98 7029minus7039(38) Ehara M Ishida M Toyota K Nakatsuji H SAC-CIGeneral-R Method Theory and Applications to the Multi-ElectronProcesses In Reviews in Modern Quantum Chemistry (A Tribute toProfessor Robert G Parr) Sen K D Ed World Scientific Singapore2003 pp 293minus319(39) SAC-CI GUIDE httpwwwqcriorjpsacci(40) Frisch M J Trucks G W Schlegel H B et al Gaussian 03development revision revision E05 Gaussian Inc Wallingford CT2004(41) Hohenberg P Kohn W Phys Rev 1964 136 B864minusB871(42) Kohn W Sham L J Phys Rev 1965 140 A1133minusA1138(43) Salahub D R Zerner M C Eds The Challenge of d and fElectrons The American Chemical Society Washington DC 1989(44) Parr R G Yang W Density-Functional Theory of Atoms andMolecules Oxford University Press Oxford UK 1989(45) Becke A D J Chem Phys 1993 98 5648minus5652(46) Lee C Yang W Parr R G Phys Rev 1988 37 785minus789(47) Hariharan P C Pople J A Theor Chim Acta 1973 28 213minus222(48) Francl M M Pietro W J Hehre W J Binkley J SDeFrees D J Pople J A Gordon M S J Chem Phys 1982 773654minus3665(49) Tomasi J Mennucci B Cammi R Chem Rev 2005 1052999minus3093(50) Fukuda R Ehara M Nakatsuji H Cammi R J Chem Phys2011 134 104109minus1minus11(51) Dunning T H Jr Hay P J In Modern Theoretical ChemistrySchaefer H F III Ed Plenum New York 1976 Vol 3 pp 1minus28(52) Nakatsuji H Chem Phys 1983 75 425minus441(53) Hocquet A Leulliot N Ghomi M J Phys Chem B 2000 1044560minus4568(54) Mishra S K Mishra P C J Comput Chem 2002 23 530minus540(55) Sprecher C A Johnson W C Biopolymers 1977 16 2243minus2264(56) Clark L B Peschel G G Tinoc I Jr J Phys Chem 1965 693615minus3618(57) Voet D Gratzer W B Cox R A Doty P Biopolymers 19631 193minus208(58) Hobza P Mehlhorn A Carsky P Zahradnik R J Mol Struct1986 138 387minus399(59) Hobza P Sponer J Chem Rev 1999 99 3247minus3276

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5555

Page 13: Helical Structure and Circular Dichroism Spectra of DNA: A ... · B-DNA, have right-handed double-helical structure, while Z-DNA is left-handed and appears only when DNA has a special

state because the stacking interaction is weaker in B-DNA thanin Z-DNA (Figure 1) Therefore the first peak of theexperimental CD spectra originates from the ET excitationfrom dG to dC for Z-DNA but corresponds to theintramolecular excitation in dC for B-DNA The second peakwith a negative sign is assigned to the 81A state The 81A statewhich is the πminusπ of dG is the origin of both the main peaks ofthe UV spectrum and the second peak of the CD spectrumThe ET excited states of the tetramer models were calculated

at a lower energy region than those of the dimer models forboth B- and Z-DNA This result may be related to the overlapbetween the nucleic acid bases The overlap exists only in partof the nucleic acid bases in the dimer models but in thetetramer models the overlap covers the hydrogen bonds as wellas the nucleic acid bases The ET excited states are importantfor the CD spectrum of Z-DNA because the lowest excitedstate is the ET type and is negative for rotatory strength but forB-DNA the ET excited states are not important because thesestates are weak in both oscillator and rotatory strengthsThe SAC-CI results of the tetramer models are calculated at

a higher region by about 05 eV because of the small activespace Therefore better SAC-CI calculations are required forcomparison with the experimental UV and CD spectra of DNAHowever these results indicate that the strong stackinginteraction characteristic of the helical structure of Z-DNAchanges the sign of the rotatory strength for the lowest excitedstates of dG and dC and produces the emergence of the ETexcited states

4 CONCLUSIONSDNA chemistry is affected by three important factors (1) theconformation of the monomer (dG) which is related to thedihedral angle between deoxyribose and guanine (2) thehydrogen-bonding interactions and (3) the stacking inter-actions between two nucleic acid bases To verify whether thesethree factors are important for the CD spectra of DNA weexamined four different models using the SAC-CI method(1) dG monomer model We first applied the SAC-CI

method to the dG monomer to examine the effect ofconformation on the CD spectra of dG and from this resultto identify the structure of the dG monomer in solution Ourcalculations showed that the shape and the sign of the CDspectra depend strongly on the dihedral angle between thedeoxyribose and the guanine but that the SAC-CI UV spectraof both anti- and syn-dG are very similar because the UVspectra are not strongly dependent on the conformation Theexperimental CD spectrum agreed with the theoreticalspectrum of dG only for the anti conformation The resultingthermally averaged spectra also support the anti geometry ofdG as a stable conformation in solution However the SAC-CICD spectrum of syn-dG is opposite in sign from that of theexperimental spectrum Thus the SAC-CI CD spectra of dG inboth anti and syn conformations do not explain the negativesign at 295 nm of the CD spectra of the DNA composed of dGand dC(2) Composit model of dG and dC monomers A

comparison of the calculated composite spectra of dG anddC with the experimental spectra of the two-component DNAshowed similarities between B- and Z-DNA because theconformation of dC is always anti in both B- and Z-DNAThe SAC-CI results led us to confirm that that theconformation of dG is not the origin of the negative peak at295 nm in Z-DNA

(3) Dimer model Our examination of the dimer modelpermitted us to investigate the separate importance of thehydrogen-bonding and the stacking interactions in thestructural chemistry of B- and Z-DNA The SAC-CIcalculations showed that the hydrogen-bonding interactionschange the excitation energies and that the stacking interactionsare responsible for the changes in the sign of the CD rotatorystrength The SAC-CI UV spectra of the hydrogen-bondingmodels (B-Hbond and Z-Hbond models) are similar to theexperimental UV spectra in which the main peak as well as theshoulder peak is well reproduced The SAC-CI CD spectra ofthe stacking models (B-Stack and Z-Stack models) have thesame sign as the experimental CD spectra We concluded thatthe stacking interaction is especially important for the CDspectrum of the strongly stacked Z-DNA because the CD signalat 295 nm changed to a negative sign only for the Z-Stackmodel(4) Tetramer model Finally our calculations with the

tetramer model that contain both hydrogen-bonding andstacking interactions revealed that the shape of the SAC-CIUV and CD spectra compared well with the experimental onesThe ET excitations as well as the lowest intramolecularexcitation of dC and dG are the origins of the negative sign ofthe experimental CD spectrum of Z-DNA The ET excited stateis calculated at a lower energy region than the intramolecularexcited states of dG or dC for the strongly stacked Z-DNA butat a higher energy region for the weakly stacked B-DNA Thenegative sign of the first peak of the experimental CD spectra ofZ-DNA is due to the strong stacking effect characteristic of itshelical structureThe dynamic behavior of the dG monomer in solution was

partially considered by taking the Boltzmann averaging of thecalculated spectra over the various conformations Fortunatelythe Boltzmann averaged SAC-CI CD spectrum was in goodagreement with the experimental spectra although the SAC-CICD spectra largely depend on the conformation of dGHowever in this study we used only the static model for dimerand tetramer models The SAC-CI CD results of the dimer andtetramer models explained the difference between theexperimental CD spectra of B- and Z-DNA However sincethe structure of DNA would also be flexible in solution it isnecessary to take into account the dynamic behavior of DNAThis is an important and interesting topic

ASSOCIATED CONTENT

S Supporting InformationSAC-CI results of the hydrogen-bonding and stacking dimermodels of Z- and B-DNA using small active space and withoutthe polarization d-functions This material is available free ofcharge via the Internet at httppubsacsorg

AUTHOR INFORMATION

Corresponding AuthorPhone +81-75-634-3211 Fax +81-75-634-3211 E-mail hnakatsujiqcriorjp

NotesThe authors declare no competing financial interest

ACKNOWLEDGMENTS

The computations were performed at the Research Center forComputational Science Okazaki Japan

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5554

REFERENCES(1) Voet D Voet J G Eds Biochemistry John Wiley amp Sons IncNew York 1995(2) Ghosh A Bansal M Acta Crystallogr 2003 D59 620minus626(3) Rich A Nordheim A Wang A H-J Annu Rev Biochem 198453 791minus846(4) Tran-Dinh S Taboury J Neumann J-M Huynh-Dinh TGenissel B Langlois drsquoEstaintot B Igolen J Biochemistry 1984 231362minus1371(5) Sugiyama H Kawai K Matsunaga A Fujimoto K Saito IRobinson H Wang A H-J Nucleic Acids Res 1996 24 1272minus1278(6) Kawai K Saito I Sugiyama H J Am Chem Soc 1999 1211391minus1392(7) Oyoshi T Kawai K Sugiyama H J Am Chem Soc 2003 1251526minus1531(8) Xu Y Ikeda R Sugiyama H J Am Chem Soc 2003 12513519minus13524(9) Berova N Nakanishi K Woody R W Eds Circular DichroismPrinciples and Applications Wiley-VCH New York 2000(10) Bauernschmitt R Ahlrichs R Chem Phys Lett 1996 256454minus464(11) Casida M E Jamorski C Casida K C Salahub D R JChem Phys 1998 108 4439minus4449(12) Tsolakidis A Kaxiras E J Phys Chem A 2005 109 2373minus2380(13) Varsano D Di Felice R Marques M A L Rubio A J PhysChem B 2006 110 7129minus7138(14) Nielsen L M Holm A I S Varsano D Kadhane UHoffmann S V Di Felice R Rubio A Nielsen S B J Phys ChemB 2009 113 9614minus9619(15) Tonzani S Schatz G C J Am Chem Soc 2008 130 7607minus7612(16) Casida M E Salahub D R J Chem Phys 2000 113 8918minus8935(17) Dreuw A Weisman J L Head-Gordon M J Chem Phys2003 119 2943minus2946(18) Maitra N T Zhang F Cave R J Burke K J Chem Phys2004 120 5932minus5937(19) McCann D M Stephens P J J Org Chem 2006 71 6074minus6098(20) Casida M E J Mol Struct 2009 914 3minus18(21) Niehaus T A J Mol Struct 2009 914 38minus49(22) Tawada Y Tsuneda T Yanagisawa S Yanai T Hirao K JChem Phys 2004 120 8425minus8433(23) Chiba M Tsuneda T Hirao K J Chem Phys 2006 124144106-1-11(24) Nakatsuji H Hirao K J Chem Phys 1978 68 2053minus2065(25) Nakatsuji H Chem Phys Lett 1978 59 362minus364(26) Nakatsuji H Chem Phys Lett 1979 67 329minus333(27) Nakatsuji H Chem Phys Lett 1979 67 334minus342(28) Nakatsuji H Bull Chem Soc Jpn 2005 78 1705minus1724(29) Ehara M Hasegawa J Nakatsuji H SAC-CI Method Appliedto Molecular Spectroscopy In Theory and Applications of Computa-tional Chemistry The First 40 Years A Vol of Technical and HistoricalPerspectives Dykstra C E Frenking G Kim K S Scuseria G EEds Elsevier Oxford UK 2005 pp 1099minus1141(30) Miyahara T Hasegawa J Nakatsuji H Bull Chem Soc Jpn2009 82 1215minus1226(31) Bureekaew S Hasegawa J Nakatsuji H Chem Phys Lett2006 425 367minus371(32) Nakatani N Hasegawa J Nakatsuji H J Am Chem Soc2007 129 8756minus8765(33) Fujimoto K Hasegawa J Nakatsuji H Bull Chem Soc Jpn2009 82 1140minus1148(34) Miyahara T Nakatsuji H Collect Czech Chem Commun2011 76 537minus552(35) Sugiyama H Unpublished(36) Geertsen J Rittby M Bartlett R J Chem Phys Lett 1989164 57minus62

(37) Stanton J F Bartlett R J J Chem Phys 1993 98 7029minus7039(38) Ehara M Ishida M Toyota K Nakatsuji H SAC-CIGeneral-R Method Theory and Applications to the Multi-ElectronProcesses In Reviews in Modern Quantum Chemistry (A Tribute toProfessor Robert G Parr) Sen K D Ed World Scientific Singapore2003 pp 293minus319(39) SAC-CI GUIDE httpwwwqcriorjpsacci(40) Frisch M J Trucks G W Schlegel H B et al Gaussian 03development revision revision E05 Gaussian Inc Wallingford CT2004(41) Hohenberg P Kohn W Phys Rev 1964 136 B864minusB871(42) Kohn W Sham L J Phys Rev 1965 140 A1133minusA1138(43) Salahub D R Zerner M C Eds The Challenge of d and fElectrons The American Chemical Society Washington DC 1989(44) Parr R G Yang W Density-Functional Theory of Atoms andMolecules Oxford University Press Oxford UK 1989(45) Becke A D J Chem Phys 1993 98 5648minus5652(46) Lee C Yang W Parr R G Phys Rev 1988 37 785minus789(47) Hariharan P C Pople J A Theor Chim Acta 1973 28 213minus222(48) Francl M M Pietro W J Hehre W J Binkley J SDeFrees D J Pople J A Gordon M S J Chem Phys 1982 773654minus3665(49) Tomasi J Mennucci B Cammi R Chem Rev 2005 1052999minus3093(50) Fukuda R Ehara M Nakatsuji H Cammi R J Chem Phys2011 134 104109minus1minus11(51) Dunning T H Jr Hay P J In Modern Theoretical ChemistrySchaefer H F III Ed Plenum New York 1976 Vol 3 pp 1minus28(52) Nakatsuji H Chem Phys 1983 75 425minus441(53) Hocquet A Leulliot N Ghomi M J Phys Chem B 2000 1044560minus4568(54) Mishra S K Mishra P C J Comput Chem 2002 23 530minus540(55) Sprecher C A Johnson W C Biopolymers 1977 16 2243minus2264(56) Clark L B Peschel G G Tinoc I Jr J Phys Chem 1965 693615minus3618(57) Voet D Gratzer W B Cox R A Doty P Biopolymers 19631 193minus208(58) Hobza P Mehlhorn A Carsky P Zahradnik R J Mol Struct1986 138 387minus399(59) Hobza P Sponer J Chem Rev 1999 99 3247minus3276

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5555

Page 14: Helical Structure and Circular Dichroism Spectra of DNA: A ... · B-DNA, have right-handed double-helical structure, while Z-DNA is left-handed and appears only when DNA has a special

REFERENCES(1) Voet D Voet J G Eds Biochemistry John Wiley amp Sons IncNew York 1995(2) Ghosh A Bansal M Acta Crystallogr 2003 D59 620minus626(3) Rich A Nordheim A Wang A H-J Annu Rev Biochem 198453 791minus846(4) Tran-Dinh S Taboury J Neumann J-M Huynh-Dinh TGenissel B Langlois drsquoEstaintot B Igolen J Biochemistry 1984 231362minus1371(5) Sugiyama H Kawai K Matsunaga A Fujimoto K Saito IRobinson H Wang A H-J Nucleic Acids Res 1996 24 1272minus1278(6) Kawai K Saito I Sugiyama H J Am Chem Soc 1999 1211391minus1392(7) Oyoshi T Kawai K Sugiyama H J Am Chem Soc 2003 1251526minus1531(8) Xu Y Ikeda R Sugiyama H J Am Chem Soc 2003 12513519minus13524(9) Berova N Nakanishi K Woody R W Eds Circular DichroismPrinciples and Applications Wiley-VCH New York 2000(10) Bauernschmitt R Ahlrichs R Chem Phys Lett 1996 256454minus464(11) Casida M E Jamorski C Casida K C Salahub D R JChem Phys 1998 108 4439minus4449(12) Tsolakidis A Kaxiras E J Phys Chem A 2005 109 2373minus2380(13) Varsano D Di Felice R Marques M A L Rubio A J PhysChem B 2006 110 7129minus7138(14) Nielsen L M Holm A I S Varsano D Kadhane UHoffmann S V Di Felice R Rubio A Nielsen S B J Phys ChemB 2009 113 9614minus9619(15) Tonzani S Schatz G C J Am Chem Soc 2008 130 7607minus7612(16) Casida M E Salahub D R J Chem Phys 2000 113 8918minus8935(17) Dreuw A Weisman J L Head-Gordon M J Chem Phys2003 119 2943minus2946(18) Maitra N T Zhang F Cave R J Burke K J Chem Phys2004 120 5932minus5937(19) McCann D M Stephens P J J Org Chem 2006 71 6074minus6098(20) Casida M E J Mol Struct 2009 914 3minus18(21) Niehaus T A J Mol Struct 2009 914 38minus49(22) Tawada Y Tsuneda T Yanagisawa S Yanai T Hirao K JChem Phys 2004 120 8425minus8433(23) Chiba M Tsuneda T Hirao K J Chem Phys 2006 124144106-1-11(24) Nakatsuji H Hirao K J Chem Phys 1978 68 2053minus2065(25) Nakatsuji H Chem Phys Lett 1978 59 362minus364(26) Nakatsuji H Chem Phys Lett 1979 67 329minus333(27) Nakatsuji H Chem Phys Lett 1979 67 334minus342(28) Nakatsuji H Bull Chem Soc Jpn 2005 78 1705minus1724(29) Ehara M Hasegawa J Nakatsuji H SAC-CI Method Appliedto Molecular Spectroscopy In Theory and Applications of Computa-tional Chemistry The First 40 Years A Vol of Technical and HistoricalPerspectives Dykstra C E Frenking G Kim K S Scuseria G EEds Elsevier Oxford UK 2005 pp 1099minus1141(30) Miyahara T Hasegawa J Nakatsuji H Bull Chem Soc Jpn2009 82 1215minus1226(31) Bureekaew S Hasegawa J Nakatsuji H Chem Phys Lett2006 425 367minus371(32) Nakatani N Hasegawa J Nakatsuji H J Am Chem Soc2007 129 8756minus8765(33) Fujimoto K Hasegawa J Nakatsuji H Bull Chem Soc Jpn2009 82 1140minus1148(34) Miyahara T Nakatsuji H Collect Czech Chem Commun2011 76 537minus552(35) Sugiyama H Unpublished(36) Geertsen J Rittby M Bartlett R J Chem Phys Lett 1989164 57minus62

(37) Stanton J F Bartlett R J J Chem Phys 1993 98 7029minus7039(38) Ehara M Ishida M Toyota K Nakatsuji H SAC-CIGeneral-R Method Theory and Applications to the Multi-ElectronProcesses In Reviews in Modern Quantum Chemistry (A Tribute toProfessor Robert G Parr) Sen K D Ed World Scientific Singapore2003 pp 293minus319(39) SAC-CI GUIDE httpwwwqcriorjpsacci(40) Frisch M J Trucks G W Schlegel H B et al Gaussian 03development revision revision E05 Gaussian Inc Wallingford CT2004(41) Hohenberg P Kohn W Phys Rev 1964 136 B864minusB871(42) Kohn W Sham L J Phys Rev 1965 140 A1133minusA1138(43) Salahub D R Zerner M C Eds The Challenge of d and fElectrons The American Chemical Society Washington DC 1989(44) Parr R G Yang W Density-Functional Theory of Atoms andMolecules Oxford University Press Oxford UK 1989(45) Becke A D J Chem Phys 1993 98 5648minus5652(46) Lee C Yang W Parr R G Phys Rev 1988 37 785minus789(47) Hariharan P C Pople J A Theor Chim Acta 1973 28 213minus222(48) Francl M M Pietro W J Hehre W J Binkley J SDeFrees D J Pople J A Gordon M S J Chem Phys 1982 773654minus3665(49) Tomasi J Mennucci B Cammi R Chem Rev 2005 1052999minus3093(50) Fukuda R Ehara M Nakatsuji H Cammi R J Chem Phys2011 134 104109minus1minus11(51) Dunning T H Jr Hay P J In Modern Theoretical ChemistrySchaefer H F III Ed Plenum New York 1976 Vol 3 pp 1minus28(52) Nakatsuji H Chem Phys 1983 75 425minus441(53) Hocquet A Leulliot N Ghomi M J Phys Chem B 2000 1044560minus4568(54) Mishra S K Mishra P C J Comput Chem 2002 23 530minus540(55) Sprecher C A Johnson W C Biopolymers 1977 16 2243minus2264(56) Clark L B Peschel G G Tinoc I Jr J Phys Chem 1965 693615minus3618(57) Voet D Gratzer W B Cox R A Doty P Biopolymers 19631 193minus208(58) Hobza P Mehlhorn A Carsky P Zahradnik R J Mol Struct1986 138 387minus399(59) Hobza P Sponer J Chem Rev 1999 99 3247minus3276

The Journal of Physical Chemistry A Article

dxdoiorg101021jp3085556 | J Phys Chem A 2013 117 42minus5555