Structural basis of instability of the nucleosomecontaining a testis-specific histone variant, human H3THiroaki Tachiwanaa, Wataru Kagawaa, Akihisa Osakabea, Koichiro Kawaguchia, Tatsuya Shigaa,Yoko Hayashi-Takanakab, Hiroshi Kimurab, and Hitoshi Kurumizakaa,1

aLaboratory of Structural Biology, Graduate School of Advanced Science and Engineering, Waseda University, 2-2 Wakamatsu-cho, Shinjuku-ku, Tokyo162-8480, Japan; and bGraduate School of Frontier Biosciences, Osaka University, 1-3 Yamada-oka, Suita, Osaka 565-0871, Japan

Edited by Timothy J. Richmond, Swiss Federal Institute of Technology, Zurich, Switzerland, and approved May 3, 2010 (received for review March 9, 2010)

A histone H3 variant, H3T, is highly expressed in the testis, suggest-ing that it may play an important role in the chromatin reorganiza-tion required for meiosis and/or spermatogenesis. In the presentstudy, we found that the nucleosome containing human H3T issignificantly unstable both in vitro and in vivo, as compared tothe conventional nucleosome containing H3.1. The crystal structureof the H3T nucleosome revealed structural differences in the H3Tregions on both ends of the central α2 helix, as compared to thoseof H3.1. The H3T-specific residues (Met71 and Val111) are thesource of the structural differences observed between H3T andH3.1. A mutational analysis revealed that these residues areresponsible for the reduced stability of the H3T-containing nucleo-some. These physical and structural properties of the H3T-contain-ing nucleosomemay provide the basis of chromatin reorganizationduring spermatogenesis.

During spermatogenesis, dramatic chromatin reorganizationoccurs, and most histones are eventually replaced by prota-

mines (1). Several histone variants are highly expressed in thetestis and are considered to be incorporated into the chromatinin the early stage of spermatogenesis.

In humans, about 4% of the haploid genome in the sperm isreportedly retained in nucleosomes, some containing the testis-specific histone H2B, hTSH2B/TH2B (2). Interestingly, the nu-cleosomes retained in the sperm are significantly enriched in locithat contain developmentally important genes. In addition,histone modifications, such as acetylation and methylation, arelikely to occur after the incorporation of the histone variantsduring spermatogenesis (1). These observations suggest thatnucleosomes containing testis-specific histone variants, with orwithout chemical modifications, may function as epigeneticmarkers in the sperm chromatin.

H3T is a variant of histone H3 that is robustly expressed in thehuman testis (3–5). We previously reported that H3T, like theconventional H3.1, can be assembled into nucleosomes withH2A, H2B, and H4 (H3T nucleosome) (6). A histone chaperone,Nap2, with 3-fold higher expression in the testis than in othersomatic tissues (7), was found to be a more efficient chaperonefor H3T nucleosome assembly than the ubiquitously expressedhistone chaperone, Nap1 (6). Therefore, H3T may be assembledinto the chromatin by a specific chaperone-mediated pathway inthe testis. Comprehensive proteome analyses of nuclear extractsfrom HeLa cells suggested that H3T also exists in somatic cells(8, 9). However, the nucleosomes containing H3T probably com-prise only a small proportion of the bulk chromatin in somaticcells, because the amount of H3T in HeLa cells is extremelylow. Therefore, H3T may have a limited function in somatic cellsthat is currently unknown.

In the present study, we found that the H3T nucleosome issignificantly unstable, as compared to the conventional H3.1 nu-cleosome, both in vitro and in vivo. The crystal structure of theH3T nucleosome was determined at 2.7 Å resolution, revealingthat, although the overall structure was similar to that of the con-ventional H3.1 nucleosome, structural differences were observedat both ends of the central α2 helix of H3Tand H3.1. The unique

physical and structural characteristics of the H3T nucleosomewere attributed to the Val111 andMet71 residues that are specificto H3T.

ResultsH3T Nucleosome Is Less Stable than the Conventional Nucleosome.The nucleosome containing human H3T was reconstituted by asalt-dialysis method, using human histones H3T, H2A, H2B,and H4, and a 146 base-pair DNA. To prepare a structurallyhomogeneous nucleosome, the reconstituted H3T nucleosomewas incubated for 2 h at 55 ˚C to disrupt the inappropriatehistone–DNA interactions (Fig. 1A) and was purified from thefree DNA by gel electrophoresis (Fig. 1B). Conventional H3.1nucleosomes were also prepared with the same procedure(Fig. 1B). Histone compositions of the H3T and H3.1 nucleo-somes prepared in this procedure were confirmed by SDS-PAGE(Fig. 1C). We next compared the stabilities of the H3Tand H3.1nucleosomes by examining the gel migration distances of thenucleosomes exposed to different NaCl concentrations. Exposureto 0.4 M NaCl had no apparent effect on the migration distancesof both the H3.1 and H3T nucleosomes (Fig. 1D, lanes 1 and 5,respectively). This observation indicates that the nucleosomeswere intact at this salt concentration. The H3.1 nucleosome ap-peared stable even when exposed to 0.8 M NaCl (Fig. 1D, lane 4),and only a small fraction of the nucleosome migrated slower.These nucleosomes are probably multimers formed by enforcedhydrophobic interactions from higher salt concentrations. By con-trast, the band corresponding to the intact nucleosome was nearlyabsent for the H3T nucleosome that was exposed to 0.6 M NaCl(Fig. 1D, lane 6). Instead, multiple, nonnucleosomal bands, con-taining only H2A/H2B (Fig. S1), were detected (Fig. 1D, lanes6–8). These results indicate that the H3T nucleosome is lessstable than the H3.1 nucleosome under high salt concentrations.

Weaker Association of H3T/H4 Tetramer to H2A/H2B Dimer. To inves-tigate the stability of the H3T nucleosome at physiological ionicstrengths, we examined the Nap1-mediated H2A/H2B disassem-bly from the H3T nucleosome. It is known that excess amount ofthe histone chaperone, human Nap1 (hNap1), promotes H2A/H2B disassembly from the nucleosome (10). The nucleosomalDNA used in this assay contains two PstI sites that are palindro-mically located in regions 13–18 bases away from both ends of the146 base-pair DNA (Fig. 2A). These PstI sites are close to the

Author contributions: H.T. and H. Kurumizaka designed research; H.T., W.K., A.O., K.K.,T.S., Y.H.-T., and H. Kimura performed research; H.T., W.K., H. Kimura, and H. Kurumizakaanalyzed data; W.K., H. Kimura, and H. Kurumizaka wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Data deposition: The crystallography, atomic coordinates, and structure factors havebeen deposited in the Research Collaboratory for Structural Bioinformatics (RCSB) ProteinData Bank, www.pdb.org (RCSB ID codes 3A6N and 3AFA).1To whom correspondence should be addressed. E-mail: kurumizaka@waseda.jp.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1003064107/-/DCSupplemental.

10454–10459 ∣ PNAS ∣ June 8, 2010 ∣ vol. 107 ∣ no. 23 www.pnas.org/cgi/doi/10.1073/pnas.1003064107

Dow

nloa

ded

by g

uest

on

June

10,

202

0

binding sites of H2A/H2B within the nucleosome (Fig. 2A). TheDNA also contains a single EcoRI site at the nucleosomal dyad,which is a binding site for H3/H4 (Fig. 2A). Therefore, if hNap1disassembles H2A/H2B, the DNA is predicted to become moresusceptible to digestion by PstI.

In the absence of hNap1, PstI partially digested the nucleoso-mal DNA of both H3T and H3.1 nucleosomes (Fig. 2 B and C,lane 6), whereas EcoRI only slightly digested the nucleosomalDNA (Fig. 2 B and C, lane 4). Therefore, the DNA segmentlocated near the exit of the nucleosome is more susceptible to re-striction nucleases than that located near the dyad. Interestingly,hNap1 substantially increased the PstI susceptibility of the DNAin theH3T nucleosome (Fig. 2B, compare lanes 6 and 7), probablyby its H2A/H2B disassembly function. In contrast, this hNap1-dependent enhancement of the PstI susceptibility was notobserved in the H3.1 nucleosome under the experimental condi-tions used in this study (Fig. 2C, compare lanes 6 and 7). Theseresults suggested that H2A/H2B in the H3T nucleosome is less

stably incorporated, as compared with those in the conventionalH3.1 nucleosome. The instability of the H3T nucleosome may beprimarily caused by a weaker association of the H2A/H2B dimerto the H3T/H4 tetramer. It should be noted that the DNA in theH3.1 nucleosome was more susceptible than that in the H3Tnucleosome in the absence of hNap1, and that the PstI digestionpatterns of the H3.1 and H3T nucleosomes were slightly different(Fig. 2 B and C). Differences in the DNA flexibility of the H3.1andH3Tnucleosomes were observed (Fig. S2B), whichmay be thereason for the distinct digestion patterns.

We then tested the interaction between H2A/H2B and H3T/H4 or H3.1/H4 in the absence of DNA. Gel filtration analysis inthe presence of 2 M NaCl revealed that the H2A/H2B dimers didnot tightly associate with the H3T/H4 tetramer (Fig. 2D). Thisresult sharply contrasts with the fact that two H2A/H2B dimersstably associates with the conventional H3.1/H4 tetramer andforms a stable H2A/H2B/H3.1/H4 octamer (Fig. 2E) (11). There-fore, the weaker association between H2A/H2B and H3T/H4maybe responsible for the instability of the H3T nucleosome. Apossible explanation for this difference could be the fact that theself-association of the H3T/H4 tetramer prevents the interactionbetween the H3T/H4 tetramer and the H2A/H2B dimer. A dy-namic light scattering analysis revealed that the Stokes radiusof the H3T/H4 tetramer in the presence of 2MNaCl was approxi-

Fig. 1. Instability of the H3T nucleosome. (A) H3T nucleosomes, reconsti-tuted using 1.2 mg∕mL total histones and 0.7 mg∕mL DNA, were analyzedby nondenaturing 6% PAGE. Lane 1 indicates naked DNA. Lanes 2 and 3 in-dicate the H3T nucleosomes before and after a 55 °C incubation, respectively.DNA was visualized by ethidium bromide staining. Asterisks indicate bandscorresponding to nonnucleosomal DNA–histone complexes. (B) The H3T andH3.1 nucleosomes were purified using a Prepcell apparatus, and wereanalyzed by nondenaturing 6% PAGE with ethidium bromide staining.(C) Histone compositions of the purified H3T and H3.1 nucleosomes wereanalyzed by 18% SDS-PAGE with Coomassie brilliant blue staining. (D) Salttitration. The nucleosomes were incubated in the presence of 0.4 M (lanes1 and 5), 0.6 M (lanes 2 and 6), 0.7 M (lanes 3 and 7), and 0.8 M NaCl (lanes4 and 8) at 42 °C for 2 h. The samples were analyzed by nondenaturing 6%PAGE with ethidium bromide staining. Lanes 1–4 and 5–8 indicate experi-ments with H3.1 and H3T nucleosomes, respectively. Bands correspondingto nucleosome monomers and nucleosome-nucleosome aggregates areindicated. Asterisks represent bands corresponding to nonnucleosomalDNA-histone complexes.

Fig. 2. H2A/H2B associates weakly with H3T/H4. (A) The crystal structure ofthe H3.1 nucleosome determined in this study. Locations the PstI and EcoRIsites are indicated. The H2A/H2B and H3.1/H4 molecules are colored in purpleand in dark blue, respectively. (B and C) H2A/H2B disassembly assay withhNap1. The nucleosomes were treated with PstI or EcoRI in the presenceor absence of excess amount of hNap1 (6.5 μM). The resulting DNA fragmentswere extracted by Phenol/chloroform, and were analyzed by 10% PAGE withethidium bromide staining. Arrows indicate the DNA fragment produced bycomplete PstI digestion. These results were confirmed to be reproduced inthree independent experiments. (B) The H3T nucleosome. (C) The H3.1 nu-cleosome. (D and E) Interaction between H2A/H2B and H3T/H4 or H3.1/H4.H2A, H2B, H4, and H3T (D) or H3.1 (E) were incubated without DNA inthe presence of 2 M NaCl. The samples were then subjected to HiLoad26∕60 Superdex 200 prep grade gel filtration column chromatography. His-tone compositions of the peak fractions were analyzed by 18% SDS-PAGEwith Coomassie brilliant blue staining. The peak fractions denoted as a, b,and c correspond to H2A/H2B/H3/H4 octamer, H3/H4 tetramer, and H2A/H2Bdimer, respectively.

Tachiwana et al. PNAS ∣ June 8, 2010 ∣ vol. 107 ∣ no. 23 ∣ 10455

BIOCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

June

10,

202

0

mately twice the size of that of the H3.1/H4 tetramer, which wasjudged to be monodisperse (Table S1), suggesting that the H3T/H4 tetramer aggregation competes with the H3T/H4–H2A/H2Binteraction.

Rapid Exchange of H3T in Nucleosomes of Living Cells. We next com-pared the mobility of H3T with H3.1, as GFP-fusion proteins inliving cells, by fluorescence recovery after photobleaching(FRAP) (12). Because GFP-H3.1 was stably incorporated intonucleosomes in living cells, only a subtle recovery was observed,even at 20 min after photobleaching (Fig. 3 A and B). By contrast,the fluorescence of GFP-H3T recovered within several minutes(Fig. 3 A and B), suggesting that H3T in the nucleosome is morerapidly exchanged compared to the conventional H3.1 in livingcells. GFP-H3Twas detected in the mono- and oligo-nucleosomalfractions prepared by micrococcal nuclease digestion followedby sucrose gradient centrifugation (Fig. 3C), indicating thatGFP-H3Twas actually incorporated into chromatin. These in vivoresults are consistent with the instability of the H3T nucleosome.

Crystal Structure of the H3T Nucleosome. To understand the struc-tural basis of the instability observed in the H3T nucleosome, thecrystal structure of the nucleosome core particle containing H3Twas solved at 2.7 Å resolution (Table S2). The overall structurewas essentially similar to that of the H3.1 nucleosome (13)(Fig. 4A). There are four amino acid differences between H3T(Val24, Met71, Ser98, and Val111) and the conventional H3.1(Ala24, Val71, Ala98, and Ala111), and three of them (Met71,Ser98, and Val111) are located in the visible, histone-fold domain(Fig. 4B). The Val24 residue, which is located in the N-terminaltail outside of the histone-fold domain, was not visible in the pre-sent H3T nucleosome structure (Fig. 4 A and B). H3.1 and its

variants contain a central helix (α2) with two shorter α-helices(α1 and α3) flanking both ends of α2. The Met71 (α1) andVal111 (α2) residues of H3Tare located near the L1 and L2 loops,respectively, which connect the flanking helices (α1 and α3) toα2 (Fig. 4B). The Ser98 residue is located near the middle ofthe α2 helix (Fig. 4B).

To identify the finer structural differences between the H3Tand H3.1 nucleosomes, we next compared the structures ofthe H3T and H3.1 histones. Although several H3 structuresare available for comparison, differences in crystallization condi-tions and crystal packing may obscure the fine structural differ-ences. To minimize these possibilities, the H3.1 nucleosome wasprepared and crystallized under identical procedures as those ofthe H3T nucleosome, and its structure was determined at a simi-lar resolution (2.5 Å) (Fig. S2 and Table S2). All residues in thecrystal structures, including the H3T-specific residues, were wellstructured, as judged from their B factors (Table S3). The H3Tand H3.1 histones were then superimposed, and the rmsd foreach residue pair was calculated and plotted against each other.The largest deviations were found near both ends of the centralα2 helix (Fig. 4B). At the N terminus of the H3T α2 helix, the vander Waals radii of the Met71 residue side chain come in close

Fig. 3. FRAP. HeLa cells expressing GFP-H3.1 or GFP-H3T were subjected tothe FRAP analysis. (A) The mobility of GFP-H3.1 or GFP-H3T in living cells wasanalyzed by bleaching one-half of the nucleus. (B) The averages of the rela-tive fluorescence intensity of bleached area were plotted with the standarddeviations (n ¼ 5). (C) GFP-H3T was incorporated into the HeLa cell chroma-tin. (Upper ) DNA fragments of mono-, di-, and trinucleosomes fractionatedby sucrose gradient centrifugation were analyzed by agarose gel electro-phoresis with ethidium bromide staining. (Right and Left) The nucleosomesamples from the HeLa cells with and without GFP-H3T expression, respec-tively. The sucrose gradient fraction numbers are indicated at the top of eachpanel. Middle panel. Histone compositions of the purified nucleosomes wereanalyzed by 16% SDS-PAGE with Coomassie brilliant blue staining staining.(Lower) GFP-H3T was detected with anti-GFP monoclonal antibody.

Fig. 4. Crystal structure of the H3T nucleosome. (A) Two views of theH3T-nucleosome structure are represented. The H3T molecules are shownin red. Locations of the Met71 and Val111 residues are indicated. (B) Struc-tural differences between H3T and H3.1 in the nucleosomes. The H3T andH3.1 structures are superimposed, and the rmsd values for each residue pairis calculated and plotted. The secondary structure of H3T in the nucleosome isshown in the top of the panel. Arrows indicate the locations of the H3T-specific amino acid residues, Met71, Ser98, and Val111. (C and D) Comparisonof the H3T structure (red) with the H3.1 structure (green). The side chains ofthe H3T-M71, H3.1-V71, H3T-V89, H3.1-V89, H3T-V111, H3.1-A111, H3T-R116,H3.1-R116, H3T-D123, and H3.1-D123 residues are represented by space-filling models. The H3T and H3.1 regions containing the amino acid residues71 (C) and 111 (D) are shown. Arrows in C and D indicate the locations of H3Tand H3.1 that are structurally different from each other.

10456 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1003064107 Tachiwana et al.

Dow

nloa

ded

by g

uest

on

June

10,

202

0

contact with that of the Val89 residue (Fig. 4C, Left). This closecontact is absent in the corresponding location of the H3.1 struc-ture (Fig. 4C, Right). At the other end of the α2 helix, the sidechain of the Val111 residue contacts the side chain of theAsp123 residue, which forms a salt bridge with Arg116 (Fig. 4D,Left). This van der Waals interaction between 111 and 123residues is not observed in the H3.1 structure (Fig. 4D, Right).This difference between H3T and H3.1 appears important forthe structural differences observed near the C terminus of theα2 helix (Fig. 4 B and D). All other locations of H3T, includingthe region around the H3T-specific Ser98 residue, were essen-tially identical in the main chain structure to those of H3.1(Fig. 4B). Therefore, the Met71 and Val111 residues in H3Tappear to be important for the formation of the specific structureof the H3T nucleosome and may contribute to the reducedstability of the H3T nucleosome.

Contribution of the V71M and A111V Substitutions in the H3T Nucleo-some Stability. In the crystal structure of the H3T nucleosome, wefound that Met71 and Val111 are the key residues affecting theH3T-specific conformation (Fig. 4). This structural property ofthe H3T nucleosome may be responsible for its reduced stability.Therefore, we then sought to determine whether the nucleosomestability is affected by substituting H3T-specific amino acids forthe H3.1 types, by creating the H3T-M71V, -S98A, and-V111A mutants. All of the mutants were competent in nucleo-some formation (Fig. 5A), and the histone compositions of thesenucleosomes were confirmed (Fig. 5B). The nucleosome stability,as judged by the resistance to high NaCl concentrations, was en-hanced drastically by the H3T-V111A mutation (Fig. 5C, lanes13–16) and moderately by the H3T-M71V mutation (Fig. 5C,lanes 5–8). By contrast, the nucleosome stability was not affectedby the S98A mutation (Fig. 5C, lanes 9–12). An H3T-M71V/V111A double mutant exhibited essentially similar stability tothe H3.1 nucleosome (Fig. 5C, lanes 17–20). Consistent resultswere obtained by reciprocal experiments using H3.1 mutants, inwhich Val71, Ala98, and Ala111 were replaced by the correspond-ing H3T amino acids. The nucleosome stability was decreaseddrastically and moderately by H3.1-A111Vand H3.1-V71Mmuta-tions, respectively (Fig. 5D, lanes 13–16 and lanes 5–8). The H3.1-V71M/A111V double mutant, like the H3T nucleosome, was veryunstable (Fig. 5D, lanes 17–20). As expected, the H3.1-A98Smutation did not affect the stability of the H3.1 nucleosome(Fig. 5D, lanes 9–12). Consistently, the H2A/H2B disassemblyassay revealed that the H3T-V111A and H3.1-A111V mutationssuppressed and enhanced, respectively, theH2A/H2Bdisassemblyby hNap1 (Fig. S3). These results indicate that the reducedstability of the H3T nucleosome is attributable mainly to theV111A substitution and partly to the M71V substitution.

To confirm whether the Val111 residue is responsible for theweaker H2A/H2B association to H3T/H4, we performed gelfiltration analyses with these H3 mutants. As shown in Fig. 6C,H3.1/H4 containing the H3.1-A111V mutant was significantlydefective in the H2A/H2B binding in the absence of DNA, likeH3T/H4 (Fig. 2D). The H3.1-V71M and H3.1-A98S mutationsdid not significantly affect the H2A/H2B binding of H3.1/H4(Fig. 6A andB).Reciprocal results were obtainedwithH3/H4 con-taining H3T-M71V, H3T-S98A, and H3T-V111A (Fig. 6 D–F).These biochemical results are consistent with the differencesobserved in the crystal structure, and therefore, the V111A substi-tution in H3Tmay be mainly responsible for the specific structureand function of the H3T nucleosome.

DiscussionIn the present study, we found that a prominent property of theH3T nucleosome is its instability. The crystal structure of the H3Tnucleosome revealed that the H3T-specific amino acid residues,Met71 and Val111, corresponding to Val71 and Ala111 in H3.1,

are responsible for the structural differences observed betweenH3T and H3.1. Consistently, mutational analyses revealed thatthe reduced stability of the H3T nucleosome is caused mainlyby Val111 and partly by Met71. Therefore, we conclude thatthe Met71 and Val111 residues of H3T are essential for theH3T-specific structure and function.

Our structural and biochemical analyses revealed that the sin-gle amino acid substitution of H3, at position 111, significantlyaffected the nucleosome stability, probably by weakening theassociation of the H3T/H4 tetramer with H2A/H2B dimers.Whereas the Val111 of H3T does not directly interact with

Fig. 5. Mutational analysis of H3T and H3.1 nucleosomes. (A) Nucleosomescontaining H3T and H3.1 mutants were purified using a Prepcell apparatusand were analyzed by nondenaturing 6% PAGE with ethidium bromidestaining. (B) Histone compositions of the purified nucleosomes containingH3T and H3.1 mutants were analyzed by 18% SDS-PAGE with Coomassiebrilliant blue staining. (C and D) Salt titration. Nucleosomes were incubatedin the presence of 0.4 M (lanes 1, 5, 9, 13, and 17), 0.6 M (lanes 2, 6, 10, 14, and18), 0.7 M (lanes 3, 7, 11, 15, and 19), and 0.8MNaCl (lanes 4, 8, 12, 16, and 20)at 42 °C for 2 h. The samples were analyzed by nondenaturing 6% PAGE withethidium bromide staining. Bands corresponding to nucleosome monomersand nucleosome–nucleosome aggregates are indicated. Asterisks representbands corresponding to nonnucleosomal DNA-histone complexes. (C) Thestability of H3T mutants. Lanes: 1–4, H3T; 5–8, H3T-M71V; 9–12, H3T-S98A;13–16, H3T-V111A; and 17–20, H3T-M71V/V111A nucleosomes. (D) Thestability of H3.1 mutants. Lanes: 1–4, H3.1; 5–8, H3.1-V71M; 9–12, H3.1-A98S;13–16, H3.1-A111V; and 17–20, H3.1-V71M/A111V.

Tachiwana et al. PNAS ∣ June 8, 2010 ∣ vol. 107 ∣ no. 23 ∣ 10457

BIOCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

June

10,

202

0

H2A/H2B and DNA, a mutation at Arg116 (to His) close toVal111 reportedly destabilized the nucleosome (14, 15). ThisH3-R116H mutation in Saccharomyces cerevisiae has beenidentified as a Sin mutation that alleviates the requirement forthe nucleosome-remodeling factor, Swi/Snf, which activates tran-scription (14, 16). A structural study revealed that the H3-Arg116residue may not directly interact with the DNA backbone andH2A/H2B, but instead forms a salt bridge with H3-Asp123 (17).This salt bridge may be very important to arrange the residuesappropriately around position 116 for hydrogen bond formationwith the DNA backbone and H2A/H2B (18). Therefore, thisSin mutation at Arg116 of H3 may destabilize the nucleosomeby allosterically reducing the histone-DNA and/or histone-histoneinteractions. In the present study, we found that theVal111 residueof H3Tsterically affected the side chain orientation of the Arg116and Asp123 residues within the H3T monomer. The H3T-Val111residue may destabilize nucleosomes by a similar mechanism tothat proposed for the H3-R116H mutation. Consistent with thisidea, a histone H3-A111G mutant exhibited the Swi/Snf-indepen-dent phenotype in S. cerevisiae (19).

In the present study, we showed that the H3T nucleosomestructurally and biochemically differs from the conventionalH3.1 nucleosome. The histone H3T variant is highly expressedin the testis (5). Therefore, H3T is anticipated to have a specificfunction in chromatin reorganization during meiosis and/or thepostmeiotic maturation of male germ cells. In germ cells, drasticchromatin reorganization occurs by histone replacement withhistone variants, and the histones are eventually replaced by pro-tamines during sperminogenesis (1, 20). The reduced stability ofthe H3T nucleosome may be favorable to promote this globaltransition of the chromosome architecture during meiotic and/or postmeiotic events. Intriguingly, a mouse testis-specific H2Avariant, H2AL2, reportedly formed a specific nucleosome that

was distinct from the conventional nucleosome. The H2AL2 nu-cleosome was quite susceptible to nucleases, suggesting that thesenucleosomes have different structural properties, as compared tothose of the conventional nucleosome (21). A possible humancounterpart of H2AL2, H2A.Bbd, which is highly expressed inthe testis, was also suggested to form a specific nucleosome struc-ture with reduced stability (22–24). Furthermore, nucleosomescontaining a testis-specific H2B variant, hTSH2B/TH2B, were re-portedly unstable, as compared to the conventional nucleosome(25). Therefore, the instability of these nucleosomes formed bythe testis-specific histone variants may be a common property re-quired for chromatin reorganization during spermatogenesis.

H3T may also be a constituent of sperm chromatin, in whichabout 4% of the human sperm genome retains nucleosomes (2).These sperm nucleosomes are suggested to have a specific epige-netic function, because they are significantly enriched arounddevelopmentally important genes, such as the imprinted geneclusters, microRNA clusters, and HOX gene clusters. Testis-spe-cific H2B variants, hTSH2B/TH2B and H2BFWT, were found inthe sperm nuclei (2, 26). H2BFWTwas stably incorporated intonucleosomes in vitro and in vivo (27) and was suggested to havetelomere-associated function (26). Whereas it remains to berevealed whether sperm chromatin indeed retains H3T nucleo-somes, the robust expression of H3T in the testis suggests itspresence in the sperm. The specific biochemical and structuralproperties of the H3T nucleosome, like those containing theH2B variants, could also play an important epigenetic role inthe regulation of genes containing nucleosomes in the sperm.

Materials and MethodsPurification of Human Histones. Human H2A, H2B, H3.1, H3T, and H4 wereoverexpressed in Escherichia coli cells, and were purified by a methodaccording to previous papers (6, 28) with modifications. Details are describedin SI Materials and Methods.

Preparation of the H3T and H3.1 Nucleosomes. Details are described inSI Materials and Methods. The 146 base-pair DNA (13, 17) was preparedas described previously (11). Purified H2A/H2B (1.7 mg), H3T/H4 (1.7 mg)or H3.1/H4 (1.7 mg), and the DNA (2 mg) were mixed in a solution containing2 M KCl. The H3T and H3.1 nucleosomes were reconstituted by the salt-dialysis method and were purified from the free DNA and histones by non-denaturing polyacrylamide gel electrophoresis, using a Prepcell apparatus(Bio-Rad).

Salt Resistance Assay. The nucleosomes (240 ng∕μL) were incubated in thepresence of 0.4, 0.6, 0.7, and 0.8 M NaCl at 42 °C for 2 h. After incubation,NaCl concentration of the samples were adjusted to 0.4 M, and the sampleswere analyzed by nondenaturing 6% PAGE.

Crystallization and Structure Determination. Details are described inSI Materials and Methods. Briefly, crystals of the purified H3T and H3.1 nu-cleosomes were obtained by the hanging drop method. The H3T-nucleosomecrystals belonged to the orthorhombic space group P212121, with unit cellconstants of a ¼ 105.5 Å, b ¼ 109.5 Å, and c ¼ 181.1 Å, and contained onenucleosome per asymmetric unit. The H3.1-nucleosome crystals also belongedto the same space group, with unit cell constants of a ¼ 105.8 Å, b ¼ 109.5 Å,and c ¼ 180.9 Å, and contained one nucleosome per asymmetric unit. High-resolution diffraction data were obtained using the synchrotron radiationsource at the beamline BL41XU station of SPring-8, Harima, Japan.

The structures of the H3Tand H3.1 nucleosomes were initially solved to 2.7and 2.5 Å resolutions, respectively, by the molecular replacement method,using the MOLREP program (29) and the human nucleosome structure(Protein Data Bank ID code 2CV5) as a guide (13). For H3T nucleosome,the Ramachandran plot of the final structure showed 93.3% of the residuesin the most favorable regions and no residues in the disallowed region. ForH3.1 nucleosome, the Ramachandran plot of the final structure showed94.9% of the residues in the most favorable regions and no residues inthe disallowed region. Summary of the data collection and refinementstatistics is provided in Tables S2 and S3. All structure figures were createdusing the PyMOL program (30). The atomic coordinates of the H3T nucleo-some and the H3.1 nucleosome have been deposited, with the RCSB ID codes,3A6N and 3AFA, respectively.

Fig. 6. Mutational analyses of interactions of H2A/H2B with H3T/H4 or H3.1/H4. Gel filtration analyses were performed as described in Fig. 2 D and E,except that a HiLoad 16∕60 Superdex 200 prep grade column was used.Fractions indicated with dots are analyzed by 18% SDS-PAGE with Coomassiebrilliant blue staining. The peak fractions denoted as a, b, and c correspond toH2A/H2B/H3/H4 octamer, H3/H4 tetramer, and H2A/H2B dimer, respectively.(A) H3.1-V71M. (B) H3.1-A98S. (C) H3.1-A111V. (D) H3T-M71V. (E) H3T-S98A.(F) H3T-V111A.

10458 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1003064107 Tachiwana et al.

Dow

nloa

ded

by g

uest

on

June

10,

202

0

hNap1-Mediated H2A/H2B Disassembly Assay. The nucleosomes (0.8 μM) wereincubated with hNap1 (6.5 μM), and were treated with PstI (30 unit) or EcoRI(36 unit) in 10 μL of 50 mM Tris · HCl buffer (pH 7.5), containing 10 mMMgCl2and 100 mM NaCl. After a 120 min incubation at 37 °C, the DNA was ex-tracted with phenol/chloroform. The DNA fragments were then analyzedby 10% PAGE in 0.5 × TBE buffer (45 mM Tris base, 45 mM Boric acid, and1 mM EDTA) (21 V∕cm for 1 h) and ethidium bromide staining.

Superdex 200 Gel Filtration Chromatography. Freeze-dried H3T (2.1 mg), H3.1(2.1 mg), or H3 mutants (2.1 mg) was mixed with H2A (1.9 mg), H2B (1.9 mg),and H4 (1.5 mg) in 5 mL of 20 mM Tris · HCl buffer (pH 7.5), containing 7 Mguanidine hydrochloride, and 20 mM 2-mercaptoethanol. The samples weredialyzed against 10 mM Tris · HCl buffer (pH 7.5), containing 2 M NaCl, and2 mM 2-mercaptoethanol, and were analyzed by Superdex 200 gel filtrationchromatography (GE Healthcare). The peak fractions were analyzed by 18%SDS-PAGE.

Fluorescence Recovery After Photobleaching. HeLa cells stably expressingGFP-H3.1 or GFP-H3T were grown on a glass-bottom dish (Mat-tek). FRAPwas performed using a confocal microscope (FV-1000; Olympus) with a60× UPlanSApo N:A: ¼ 1.35 lens, as described previously (12). Three confocalimages of a field containing 4–10 nuclei were collected (800 × 800 pixels,zoom 3, scan speed 2 μs∕pixel, pinhole 800 μm, Kalman filtration for fourscans, LP505 emission filter, and 0.1% transmission of 488-nm Ar laser).One-half of each nucleus was bleached using 75% transmission of 488 nmand 100% of 514 nm (two iterations), and images were collected usingthe original setting every 1 min. The fluorescence intensity of the bleachedarea was measured using Image J 1.39u. After subtracting the background,the intensity was normalized to the initial intensity before bleaching.

Analysis of the GFP-H3T Incorporation into Chromatin in HeLa Cells. Details aredescribed in SI Materials and Methods. Nucleosomes were prepared essen-tially according to previous paper (31) with slight modifications. Briefly, HeLacells, in which GFP-H3T was exogenously expressed, were collected andresuspended in 1 mL ice-cold 10 mM Hepes-NaOH (pH 7.4), 15 mM NaCl,

and 1.5 mMMgCl2 containing 1% Triton X100 and protease inhibitor cocktail(Nacalai Tesque). After disrupting the cells using Dounce homogenizer (tightpestle; five times), nuclei were collected and treated with micrococcal nucle-ase (2 × 106 Gel units∕mL; New England Biolabs) at 30 °C for 1 h (mixing byinverting every 15 min). After adding 10 mM EDTA (pH 8.0) and centrifuga-tion (10;000 × g; 10 min; 4 °C), the pellet was suspended in 540 μL 10 mMEDTA (pH 8.0). After the addition of 5 M NaCl solution (final 0.3 M), thesample was centrifuged (20;000 × g; 10 min; 4 °C), and the supernatantwas collected. The supernatant was then incubated at 55 °C for 2 h to dena-ture nonnucleosomal proteins. After the centrifugation (10;000 × g; 5 min; 4°C) to remove denatured proteins, the supernatant (nucleosome sample) wascollected. The nucleosome sample (0.6 mL) was fractionated in a 12 mL su-crose gradient (10–30%) by centrifugation (209;541 × g; 21 h; 4 °C) using aBeckman SW41Ti rotor. Fractions (0.6 mL each) were collected from the top.

For DNA analysis, fractions were mixed with SDS (0.2%) and analyzed by2% agarose gel electrophoresis in 1 × TAE (40 mM Tris base, 20 mM aceticacid, and 1 mM EDTA) (13.3 V∕cm for 1 h) followed by ethidium bromidestaining. For protein analysis, fractions containing mono-, di-, and trinucleo-somes were separated by 16% SDS-PAGE, and either stained with Coomassiebrilliant blue or transferred to Hybond-P PVDF membrane (GE Healthcare)using a semidry blotting system (BIO CRAFT). The GFP-H3T signals weredetected with anti-GFP antibody (Nacalai Tesque), peroxidase-conjugatedanti-mouse Ig, and ECLWestern Blotting Detection Reagents (GE Healthcare)using a LAS-4000 (Fujifilm).

ACKNOWLEDGMENTS. We thank Dr. K. Luger (Colorado State University) forproviding the plasmid to prepare the 146 base-pair DNA used in this study.We also thank Drs. N. Fujikawa and S. Yokoyama (RIKEN) for technicalassistance on nucleosome preparation, Dr. T. Fukagawa (National Instituteof Genetics) for general discussion, and Dr. S.-Y. Park (Yokohama City Univer-sity) for assistance on dynamic light scattering experiments. We are alsograteful to the beamline scientists, Drs. Y. Kawano and N. Shimizu for theirassistance in data collection at the BL41XU beamline of SPring-8. This workwas supported in part by Grants-in-Aid from the Ministry of Education,Culture, Sports, Science, and Technology, Japan. H. Kurumizaka is a researchfellow in the Waseda Research Institute for Science and Engineering.

1. Govin J, Caron C, Lestrat C, Rousseaux S, Khochbin S (2004) The role of histonesin chromatin remodelling during mammalian spermiogenesis. Eur J Biochem271:3459–3469.

2. Hammoud SS, et al. (2009) Distinctive chromatin in human sperm packages genes forembryo development. Nature 460:473–478.

3. Franklin SG, Zweidler A (1977) Non-allelic variants of histones 2a, 2b and 3 inmammals. Nature 266:273–275.

4. Albig W, Ebentheuer J, Klobeck G, Kunz J, Doenecke D (1996) A solitary human H3histone gene on chromosome 1. Hum Genet 97:486–491.

5. Witt O, Albig W, Doenecke D (1996) Testis-specific expression of a novel human H3histone gene. Exp Cell Res 229:301–306.

6. Tachiwana H, Osakabe A, Kimura H, Kurumizaka H (2008) Nucleosome formationwith the testis-specific histone H3 variant, H3t, by human nucleosome assemblyproteins in vitro. Nucleic Acids Res 36:2208–2218.

7. Hu RJ, Lee MP, Johnson LA, Feinberg AP (1996) A novel human homologue of yeastnucleosome assembly protein, 65 kb centromeric to the p57KIP2 gene, is biallelicallyexpressed in fetal and adult tissues. Hum Mol Genet 5:1743–1748.

8. Andersen JS, et al. (2005) Nucleolar proteome dynamics. Nature 433:77–83.9. Govin J, Caron C, Rousseaux S, Khochbin S (2005) Testis-specific histone H3 expression

in somatic cells. Trends Biochem Sci 30:357–359.10. OkuwakiM, Kato K, Nagata K (2010) Functional chracterization of human nucleosome

assembly protein 1-like proteins as histone chaperones. Genes Cells 15:13–27.11. Dyer PN, et al. (2004) Reconstitution of nucleosome core particles from recombinant

histones and DNA. Methods Enzymol 375:23–44.12. Kimura H, et al. (2006) A novel histone exchange factor, protein phosphatase

2Cgamma, mediates the exchange and dephosphorylation of H2A-H2B. J Cell Biol175:389–400.

13. Tsunaka Y, Kajimura N, Tate S, Morikawa K (2005) Alteration of the nucleosomalDNA path in the crystal structure of a human nucleosome core particle. Nucleic AcidsRes 33:3424–3434.

14. Kruger W, et al. (1995) Amino acid substitutions in the structured domains ofhistones H3 and H4 partially relieve the requirement of the yeast SWI/SNF complexfor transcription. Genes Dev 9:2770–2779.

15. Kurumizaka H, Wolffe AP (1997) Sin mutations of histone H3: Influence on nucleo-some core structure and function. Mol Cell Biol 17:6953–6969.

16. Wechser MA, Kladde MP, Alfieri JA, Peterson CL (1997) Effects of Sin- versions ofhistone H4 on yeast chromatin structure and function. EMBO J 16:2086–2095.

17. Luger K, Mäder AW, Richmond RK, Sargent DF, Richmond TJ (1997) Crystal structure ofthe nucleosome core particle at 2.8 A resolution. Nature 389:251–260.

18. Muthurajan UM, et al. (2004) Crystal structures of histone Sin mutant nucleosomesreveal altered protein-DNA interactions. EMBO J 23:260–271.

19. He Q, Yu C, Morse RH (2008) Dispersed mutations in histone H3 that affect transcrip-tional repression and chromatin structure of the CHA1 promoter in Saccharomycescerevisiae. Eukaryot Cell 7:1649–1660.

20. Ooi SL, Henikoff S (2007) Germline histone dynamics and epigenetics. Curr Opin CellBiol 19:257–265.

21. Syed SH, et al. (2009) The incorporation of the novel histone variant H2AL2 confersunusual structural and functional properties of the nucleosome. Nucleic Acids Res37:4684–4695.

22. Bao Y, et al. (2004) Nucleosomes containing the histone variant H2A.Bbd organize only118 base pairs of DNA. EMBO J 23:3314–3324.

23. Gautier T, et al. (2004) Histone variant H2ABbd confers lower stability to the nucleo-some. EMBO Rep 5:715–720.

24. Doyen CM, et al. (2006) Dissection of the unusual structural and functional propertiesof the variant H2A.Bbd nucleosome. EMBO J 25:4234–4244.

25. Li A, et al. (2005) Characterization of nucleosomes consisting of the human testis/sperm-specific histone H2B variant (hTSH2B). Biochemistry 44:2529–2535.

26. Churikov D, et al. (2004) Novel human testis-specific histone H2B encoded by theinterrupted gene on the X chromosome. Genomics 84:745–756.

27. Boulard M, et al. (2006) The NH2 tail of the novel histone variant H2BFWT exhibitsproperties distinct from conventional H2B with respect to the assembly of mitoticchromosomes. Mol Cell Biol 26:1518–1526.

28. Tanaka Y, et al. (2004) Expression and purification of recombinant human histones.Methods 33:3–11.

29. Collaborative Computational Project (1994) Number 4 The CCP4 suite: Programs forprotein crystallography. Acta Crystallogr D 50:760–763.

30. DeLanoWL (2008) The PyMOL Molecular Graphics System (DeLano Scientific LLC, PaloAlto, CA), www.pymol.org.

31. Kimura H, Hayashi-Takanaka Y, Goto Y, Takizawa N, Nozaki N (2008) The organizationof histone H3 modifications as revealed by a panel of specific monoclonal antibodies.Cell Struct Funct 33:61–73.

Tachiwana et al. PNAS ∣ June 8, 2010 ∣ vol. 107 ∣ no. 23 ∣ 10459

BIOCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

June

10,

202

0