J. Biochem. 117, 860-855 (1995)

Structural Composition of Hammerhead Ribozymes1

Akio Takenaka,*-2 Osamu Matsumoto,' Yixin Chen,* Sei-ichi Hasegawa,*Toshiyuki Chatake,* Masaru Tsunoda,* Tsutomu Ohta,f Yasuo Komatsu/Makoto Koizumi,' and Eiko Ohtsuka1

'Faculty of Bioscience and Biotechnology, Tokyo Institute of Technology, Nagatsuda 4259, Midori-ku, Yokohama226; and ^Faculty of Pharmaceutical Sciences, Hokkaido University, Kita-12, Nishi-6, Kita-ku, Sapporo 060

Received for publication, November 7, 1994

Eleven kinds of hammerhead ribozymes were designed and synthesized to investigate thestructural composition of the complexes and to find suitable crystallization conditions, thesubstrate chains having been modified to prevent hydrolysis. Electrophoresis patternsindicated that the two strands except for the substrate chain form a binary complex and thatthey form a ternary complex when mixed with the substrate chain. Both complexes werecrystallized. The crystal of the binary complex belongs to a Laue symmetry of 32 (spacegroup of P321, P3i21, or P3221) with cell dimensions of a = 6 = 53.4 and c = 59.4 A. Thevolume per one nucleotide allows the asymmetric unit to contain one binary complex. Ourresults suggest that the catalytic part forms a rigid ribozyme structure which induces ascissile reaction when the substrate is bound, in a similar manner to an enzyme protein.

Key words: crystallization, electrophoresis, ribozyme, X-ray work.

Discovery of functional RNA molecules that behave likeenzymes has changed our concept of the role of nucleic acidsin biological systems (1). Examples include hammerheadribozymes, which participate in "molecular recognition"and "catalysis" during mRNA processing (2). They areclassified into several types according to the source and thesecondary structure characteristics (3). It has been pos-tulated that hammerhead ribozymes have a common struc-ture which is basically composed of three RNA segments,the consensus sequences of which form a core structuresupported by the three stems (4-6). Several oligonucleo-tides have been synthesized and examined to identify theessential structure required for the ribozyme function,which catalyzes a frans-esterification of a 3',5'-phospho-diester bond to yield two chains with a 2',3'-cyclic phospho-diester and a free 5'-hydroxyl group in an Mg^-dependentreaction (7).

To investigate the tertiary structure of ribozymes, NMRexperiments were attempted and the stem parts wereidentified (8). As a preliminary to X-ray analysis, weconducted crystallographic studies on eleven kinds ofhammerhead ribozymes with different nucleotide se-quences. All the ribozymes contain the RNA substratechain with chemical modification at the reaction site toprevent hydrolysis. Recently other catalytic RNAs havebeen crystallized (9-11). Our ribozymes, derived originallyfrom transcripts of the satellite DNA2 of newt (12), are thesmallest ones, designed so as to allow identification of theminimum essential structure.

The image of hammerhead ribozyme has been always1 This work was partially supported by Grants-in-Aid for ScientificResearch on Priority Areas (No. 03242104 and No. 06258205) fromthe Ministry of Education, Science and Culture of Japan, and also inpart by a grant from the Kihara Memorial Yokohama Foundation forthe advancement of Life Sciences.1 To whom correspondence should be addressed.

described including the substrate chain, though the struc-tural composition in solution is still ambiguous. The ques-tion arises, (i) does the hammerhead ribozyme alwaysrequire the substrate chain to assemble a rigid structure, or(ii) can the two chains form an enzymatic structure withoutsubstrate. In case (i), the two strands would not form anystructure without the substrate chain and its addition wouldcause formation of a ternary complex, leading to scission ofthe substrate chain. In the latter case, the two strandswould always form a rigid ribozyme structure which causesa catalytic reaction when the substrate is bound, like anenzyme protein in principle.

In this paper we will report the structural composition ofhammerhead ribozymes revealed by electrophoresis andcrystallographic studies.

EXPERIMENTAL

Materials—Several kinds of hammerhead ribozymes

Fig. 1. A postulated secondary structure for hammerheadribozymes. Nucleotides enclosed are the consensus sequences andthe cleavage site is indicated by an arrow.

850 J. Biochem.

Hammerhead Ribozyme 851

were designed by changing nucleotides except for the core,with the consensus sequence shown in Fig. 1. Each oligonu-cleotide was synthesized separately by the phosphoram-idite method (13). Their catalytic properties have alreadybeen tested (13). For the present work, the substratechains were modified at the scissile site by substitution of

the 02 ' hydroxyl group with a methoxyl, an o-nitroben-zoxyl, or a fluoro group to prevent cleavage (13, 14).

Crystallization—Synthesized oligonucleotides were dis-solved in 20 mM Na-cacodylate buffer and mixed inequimolar ratio to form a hammerhead ribozyme includingthe modified substrate chain. Crystallization conditions

R 1

CL-4 C ' GC-GC-GA-U

CL-3Cf

s- GACCCU

Ci

CUAGG

UAGUGGGA. CL-2 _GAUCC,

A _ CG U A G U

R 3

ras15 mer Ui

ras7

A Q

cA-U

UAQCCACAUC

R 5 CL-3CmC-GC-G

CL-11 A-U

M

CUAGG

R 7

CL-10

ACCCU

A CL-3CmC-GC-G M

.A-UCrn

CUAGG

GAUCC,UGGGA A r 1 1 C C a A U C O G 5G U A G U

R 9

lnoslne-21c-Q] M

CUAGGA

U -lnoslno-18 -̂ . . ,-^-^uCGAUCCG

UAGR11

CL-4

ACCCUG

.UGGGAA

R2 A «"C-G

«. - C-GCL-4 A .u

AA EfK " I T i l 1

5ACCCU

CL-3C«.

ONB

CUAGG

UAG U G G G A A G CL-2 C G A U C C G S .* f a U A G U

R4

ras8

ras15 mer Ut

in.<SGuac)A,o

CCACAUfr

R 6cL-12

C-GC-G

.A-U

•GCGGC

y CGCCGyj

CL-3C™

M

CUAGG A

GAUCC <~

A G U

R 8

CR-11C-GA-U

r ACCCU G

3.UGGGAA

CR-9M

^ CUAGGA

c G A U C C G j

RIB c l-3 C m

CL-4

ACCCUU

UGGGA._UAG A G

M

CUAGG'

GAUCCU A G

Fig. 2. Eleven kinds of hammerhead ribozymes designed and synthe-sized for the present studies. Nucleotide sequences enclosed in boxes aredifferent from those of Rl. The substrate chains are modified at the 02' hydroxylgroup of the scissile nucleotide (indicated by an arrow) by adding fluoride (F),o-nitrobenzoxyl (ONB), or methoxyl group (M).

Vol. 117, No. 4, 1995

§52 A. Takenaka et at.

were surveyed by using the hanging- and the sitting-dropvapor diffusion methods at different temperatures and inthe pH range of 6-7.8 with 10-200 mM MgCl2 and 10-100mM spermine 4HC1, and with 5-30% 2-methyl-2,4-pentanediol as a precipitant. The eleven kinds of hammer-head ribozymes examined are shown in Fig. 2. To ensurecomplex formation, the ternary mixture (CL-2, CL-3m,and CL-4) of R10 was first incubated without magnesiumby changing the molar ratio of the substrate from 1 to 5, orby heating for 2 min at 80°C, followed by slow cooling toroom temperature over 2 h, and then crystallized under thesame conditions as described above. Crystallizations ofbinary mixtures of the catalytic strands of R10 or Rl 1 werealso performed in an incubator at 25'C by controlling theionic concentration finely between the drop and the reser-voir solution.

HPLC Analysis—Single crystals obtained from theternary mixture of R10 were separated from the motherliquor by centrifugation and dissolved in the same buffersolution after washing. This solution, the mother liquor andthe initial ternary mixture were charged independently on

CL-2:CL-4

CL-3Cm

CL-4CL-2

• \ ' f

CL-2:CL-3Cm

\

CL-4 iCL-3CmCL-2:CL-4:CL-3Cm

' T t

•

a high performance liquid chromatography (HPLC) column[TSK gel DEAE 2SW (^4.6x250 mm)]. The oligonu-cleotides were eluted with 20% CH3CN in 2 M NrL,COOHand detected by UV absorption measurement at 280 nm.

X-Ray Measurement—All X-ray experiments were doneusing a synchrotron radiation source (A. =1.0 A) at thePhoton Factory of KEK in Tsukuba. A crystal obtainedfrom the binary mixture of CL-2 and CL-4 was sealed in aglass capillary and mounted on a goniometer head, the longaxis of the crystal being along the spindle axis. Thediffraction patterns were recorded on imaging plates usinga Weissenberg camera for macromolecules (15) with acollimator of 0.1x0.1 mm square and crystal-to-imagingplate distance of 286.7 mm. The oscillation angle for eachshot was 9.5' in o> and the coupling constant betweenrotation of the specimen and the movement of the filmcassette was 27mm. In the same way, diffraction patternsof a crystal obtained from the ternary mixture of R10 weretaken. By using the computer program WEIS (16), diffrac-tion spots recorded on imaging plates were indexed andintegrated to evaluate their intensities, and then they weremerged on a common scale.

Electrophoresis—Gel electrophoresis was carried out at25"C in 20% polyacrylamide gel using 500 mM Tris-boratebuffer solution (pH 8.2) containing 100 mM Mg-acetate and100 mM Na-acetate. The component oligonucleotides ofR10 and Rl 1 were dissolved in 20 mM Na-cacodylate buffer(pH 7.8) separately, and they, their binary mixtures andthe ternary mixture were charged on the gel. All electro-phoresis experiments were done at room temperaturewithout heat treatment of the samples. The ternary com-plex formation was examined by heating the mixtures for 2min at 80*C followed by slow cooling to room temperature

1 2 3 4 5 6 7

(a)

CL-3CmN

CL-4

CL-2K

CL-2N:CL-4

CL-2N:CL-3CmH

CL-4:CL-3CmN

I CL-2N:CL-4:CL-3CmN

i | t t

1 2 3 4 5 6 7

(b)

Fig. 3. Polyacrylamide gel electrophoresis of the componentnucleotide chains in different combinations for R10 (a) and forRl l (b). Lanes 1, 2, and 3 are for individual component chains. Lanes4, 5, and 6 are for the binary (1:1) mixtures in different combinations,and lane 7 is for the ternary (1:1:1) mixture.

CL-3Cm5GGUCmCUAGG A3

3,AGGAUCCmUGG5,CL-3Cm

Fig. 4. Possible duplex formation of CL-3Cm.

CL-2N:CL-4:CL-3CmN

CL-2:CL-4:CL-3Cm(H)

CL-2:CL-4:CL-3Cm

CL-2N:CL-4:CL-3CmN(H)

V V V

A A

1 2 3 4Fig. 6. Polyacrylamide gel electrophoresis of ternary mix-tures. Lanes 1 and 2 are for R10 with and without pre-heating (H),respectively, and lanes 3 and 4 for R l l in the same order.

J. Biochem.

Hammerhead Ribozyme 853

over 2h, in order to release each oligonucleotide fromadopting secondary structure and to ensure complex forma-tion.

RESULTS AND DISCUSSION

Figure 3 shows the electrophoresis patterns of RIO andRl l . Lanes 1, 2, and 3 are for individual segments. Lanes4, 5, and 6 are for the binary (1:1) mixtures in differentcombination, and 7 is for the ternary (1:1:1) mixture. Thesubstrate chain, CL-3Cm in Fig. 3a, moves more slowlythan CL-4 despite its shorter length, suggesting thatCL-3Cm exists as a dimer, in which eight of 10 bases canform base pairings as shown in Fig. 4. Therefore anothersubstrate chain, CL-3CmN of the same length as CL-3Cm,was designed so as not to form such a duplex and used tomake Rl 1. It ran at the expected position in Fig. 3b (slightlyfaster than CL-4). This also strongly supports the duplex

CL-2

CL-4

Fig. 6. HPLC analysis of oligonucleotides in the RIO crystalli-zation. HPLC elution patterns are shown for (a) the single crystalsobtained from a ternary mixture of CL-2, CL-4, andCL-3Cm, (b) themother liquor after removal of the crystals, and (c) the initial ternarymixture. The other peaks in (b) which are not observed in fresh (c)suggest some degradation of the nucleotides during crystallization.

formation of CL-3Cm.In either combination of CL-3Cm with CL-2 or with

CL-4, the binary mixtures showed two bands correspond-ing to each chain, indicating that the two chains in thesecombinations exist separately without any interaction. Incontrast, the combination of CL-2 and CL-4 showed a singleband of slow mobility, indicating complex formation be-tween them. The ternary mixture showed two bands justcorresponding to the duplex of CL-3Cm, and the binarycomplex of CL-2 and CL-4. If CL-3Cm's are associatedwith themselves in solution, it would be possible to dissoci-ate them by heating. Figure 5 shows the electrophoresispatterns of the ternary mixtures, without heating (lane 1)and with heating followed by annealing (lane 2). Both havetwo bands, but the upper band in lane 2 is slightlybroadened upward, suggesting the coexistence of a ternarycomplex together with the CL-3Cm duplex and the CL-2:CL-4 binary complex after heat treatment.

In the combinations of Rl l (Fig. 3b), only the binarymixture of CL-2N and CL-4 shows a single band of slowmobility, indicating complex formation similar to that ofRIO. The single band in lane 6 is interpreted to be thesuperimposed bands of CL-3CmN and CL-4. The ternarymixture of CL-2N, CL-4, and CL-3CmN shows a singleband (lane 7). Even when heated, it showed a single bandonly, with no change in its mobility (compare lanes 3 and 4in Fig. 5). Such behavior indicated that the substrate chain,

Fig. 7. Single crystals grown in the binary mixture of thecatalytic strands (CL-2:CL-4) of the hammerhead ribozyme,RIO.

TABLE I. Precipitates and crystallizationRibozymesPrecipitates'Conditions

Sample (AiM/ml)Buffer (mM)Spermine (mM)Mg*+ (mM)pHMPDb(%)Temp. CO

RlA

302010156.5304

R2A

302010156.5304

conditions for eleven ribozymes.R3A

302010156.5304

R4A

302010156.5304

R5i

302010157.625

4, 25

R6

1

302010157.625

4, 25

R71

302010157.625

4, 25

R8A

302010157.5

5.3-6.24

R9A

302010157.53025

R10O

302001001507.8525

R l l

1

30100751507.57

25

*O, single crystals; A, microcrystals; J , amorphous. b2-Methyl-2,4-pentanediol.

Vol. 117, No. 4, 1995

8S4 A. Takenaka et al.

U AG

CL-4

ACCCU

UGGGA . X3 A U C CC] G

CL-2Fig. 8. Possible secondary structure (?) for the binary com-plex (catalytic part) between CL-2 and CL-4.

Fig. 9. A diffraction pattern of the crystal obtained from thebinary mixture of CL-2 and CL-4.

CL-3mN, which does not form a duplex, immediately bindsto the catalytic strands to form the ternary complex. It ispossible to assume that the substrate chain binds to thebinary complex, when added, by inducing a conformationalchange of the catalytic strands. Although the substrate inthis work was blocked as regards trans-esterification reac-tion by the methyl group, the real substrate might becleaved after binding to the catalytic binary complex.

As another approach, we undertook crystallographicstudies of these ribozymes. The results of crystallizationare summarized in Table I. Among the eleven ribozymes,only RlO gave single crystals from the ternary mixture ofCL-2, CL-4, and CL-3Cm, which were suitable for X-raywork. Some of the other ribozymes, which were precipitat-ed as microcrystals, might afford single crystals afterfurther studies. The composition of the RIO crystals thusobtained was analyzed by HPLC. In Fig. 6, it can be seenthat the crystal contained only CL-2 and CL-4 in equimolarratio and CL-3Cm remained in the mother liquor. This isconsistent with the results of the electrophoresis experi-ments mentioned above.

Such a binary complex formation was further confirmedby crystallization. Prom the binary mixture of CL-2 andCL-4, single crystals (Fig. 7) were grown to 0.05 X 0.05 X

<0

uc

ucaa.t>E

600 '

500 "

•> 4000 20 40 60

Number of nucleotldes

Fig. 10. A correlation curve between the volume for onenucleotide in the crystalline state and the number of nucleotldesforming the molecule.

0.12 mm in size by controlling the ionic strength betweenthe hanging drop containing the sample and the reservoirsolution. These crystals have a similar shape to thoseobtained from the ternary mixture of RlO. Furthermore,X-ray diffraction indicated that they are crystallographical-ly the same.

For the binary complex between CL-2 and CL-4, it isdifficult to predict the secondary structure except in thestem region (Fig. 8). X-ray analysis will give us knowledgeof the interactions between bases or between base andribose phosphate backbone to form the complex structure.We therefore collected X-ray data of the binary complex tosolve the structure.

The crystal obtained from the binary mixture of CL-2and CL-4 gave reflections up to 4.5 A resolution at roomtemperature (Fig. 9). From the diffraction patterns, it wasdetermined that the unit cell had a Laue symmetry of 32with a space group of P321, P3,21, or P3221, and that thelattice constants were a=6=53.4 and c = 59.4A. Thenumber of molecules in the asymmetric unit can be esti-mated from the volume per nucleotide, VN. For the presentdiscussion, we calculated the volumes from 131 availabledata in the Protein Data Bank of 1993 (17). They variedslightly in the range of 500-980 A3 according to the size ofthe molecules, as can be seen in Fig. 10. Such a correlationmay reflect the packing effect. Similar values were reportedby several authors in reviews (18-20). The value of VN forthe present crystal was evaluated as 740 A3, when it wasassumed that the asymmetric unit contained one unit of thebinary complex. This value is close to the correlation linewithin the range mentioned above. In all, 3,016 reflectionswith F0>3<y were collected, 519 of which were indepen-dent. iJsym and -flange were 0.046 and 0.049, respectively.Crystals of heavy-atom derivatives are in preparation tosolve the structure by the multiple isomorphous replace-ment method.

In order to crystallize the ternary complex of RIO, it isnecessary to release the substrate chain from the duplex.Pre-heating and annealing a solution of the ternary mixtureshould be an effective way to ensure complex formation (8).From crystallizations based on this approach, we obtainedsmall crystals with different shape, suggesting ternary

J. Biochem.

Hammerhead Ribozyme 855

complex formation. They are still too small for X-rayanalysis, but a suitable crystal should become availablesoon.

Hammerhead ribozymes have been defined including thesubstrate chain, without any evidence for the structuralcomposition. It was not clear whether they always requirethe substrate chain to assemble a rigid structure. Ourresults show that the two catalytic strands formed a rigidcomplex, and it is possible that the catalytic binary complexforms a ternary complex when mixed with the substratechain in solution. Hammerhead ribozymes may be thusdefined in two parts, the catalytic and the substrate parts,like an enzyme protein in principle. If this is the case,comparative studies of the three-dimensional structures ofthe binary and the ternary complexes will be most interest-ing.

We are grateful to Prof. N. Sakabe and his colleagues for theopportunity to use the Photon Factory facilities, to Prof. K. Nishi-kawa and Dr. T. Yokogawa for assistance : i electrophoresis experi-ments and helpful discussions, and finally to Prof. M. Sekine forpermitting us to use a freeze-drying apparatus.

REFERENCES

1. Cech, T.R., Zaug, A.J., and Grabowski, P.J. (1981) In vitrosplicing of the ribosomal RNA precursor of Tetrahymena:Involvement of a guanosine nucleotide in the excision of theintervening sequence. Cell 27, 487-496

2. Prody, G.A., Bakos, J.T., Buzayan, J.M., Schneider, I.R., andBruening, G. (1986) Autolytic processing of dimeric plant virussatellite RNA. Science 231, 1577-1580

3. Buzayan, J.M., Gerlach, W.L., and Bruening, G. (1986> Non-enzymatic cleavage and ligation of RNAs complementary to aplant virus satellite RNA. Nature 323, 349-353

4. Huchins, C.J., Rathjen, P.D., Forster, A.C., and Symons, R.H.(1986) Self-cleavage of plus and minus RNA transcripts ofavocado sunblotch viroid. Nucleic Acids Res. 14, 3627-3640

6. Symons, R.H. (1989) Self-cleavage of RNA in the replication ofsmall pathogens of plants and animals. Trends Biochem. Sci. 14,445-450

6. Uhlenbeck, O.C. (1987) A small catalytic oligoribonucleotide.Nature 328, 596-600

7. Koizumi, M., Iwai, S., and Otsuka, E. (1988) Construction of aseries of several self-cleaving RNA duplexes using synthetic21-mers. FEBS Lett. 288, 228-230

8. Heus, H.A., Uhlenbeck, O.C, and Pardi, A. (1990) Sequence-dependent structure variations of hammerhead RNA enzymes.Nucleic Acids Res. 18, 1103-1108

9. Pley, H.W., Lindes, D.S., DeLuca-Flaherty, C, and Mckay, D.B.(1993) Crystals of a hammerhead ribozyme. J. Biol. Chem. 268,19656-19658

10. Doudna, J.A., Grosshans, C, Gooding, A., and Kundrot, C.E.(1993) Crystallization of ribozymes and small RNA motifs by asparse matrix approach. Proc. Natl. Acad. Sci. USA 90, 7829-7833

11. Kim, R., Holbrook, E.L., Jancarik, J., Pandit, J., Weng, X.,Bohm, A., and Kim, S.H. (1994) High-resolution crystals andpreliminary X-ray diffraction Btudies of a catalytic RNA. ActaCryst. D50, 290-292

12. Epstein, L.M. and Gall, J.G. (1987) Self-cleaving transcripts ofsatellite DNA from the Newt. Cell 48, 535-543

13. Koizumi, M., Hayase, Y., Iwai, S., Kamiya, H., Inoue, H., andOhtsuka, E. (1989) Design of RNA enzymes distinguishing asingle base mutation in RNA. Nucleic Acids Res. 17, 7059-7071

14. Ohtsuka, E., Ohta, T., and Koizumi, M. (1993) Sequence-specificcleavage of RNA by designed ribozymes. Supramol. Chem. 2,197-200

15. Sakabe, N. (1983) A focusing Weissenberg camera with multi-layer-line screen for macromolecular crystallography. J. Appl.Cryst. 16, 542-547

16. Higashi, T. (1989) The processing of diffraction data taken on ascreenless Weissenberg camera for macromolecular crystallogra-phy. J. Appl. Cryst. 22, 9-18

17. Bernstein, F.C., Koetzle, T.F., and Tasumi, M. (1977) TheProtein Data Bank: A computer-based archival file for macro-molecular structures. J. MoL Biol. 112, 535-542

18. Kennard, 0. and Hunter, W.N. (1989) Oligonudeotide structure:A decade of results from single crystal X-ray diffraction studies.Q. Rev. Biophys. 22(3), 327-379

19. Dickerson, R.E. (1990) What do we really know about B-DNA? inStructure & Methods: DNA & RNA (Sarma, R.H. and Sarma,M.H., eds.) Vol. 3, pp. 1-38, Adenine Press, New York

20. Bingman, C.A., Zon, G., and Sundaralingam, M. (1992) Crystaland molecular structure of the A-DNA dodecamer d(CCGTA-CGTACGG): Choice of fragment helical axis. J. Mol. Biol. 227,738-756

Vol. 117, No. 4, 1995