letters

The SIN domain of thehistone octamer is essentialfor intramolecular foldingof nucleosomal arraysPeter J. Horn1,2, Kimberly A. Crowley1,2, Lenny M. Carruthers1, Jeffrey C. Hansen3

and Craig L. Peterson1

1Program in Molecular Medicine, University of Massachusetts MedicalSchool, Worcester, Massachusetts 01605, USA. 2Both authors contributedequally to this work. 3Department of Biochemistry, University of TexasHealth Sciences Center, San Antonio, Texas 77303, USA

Published online: 11 February 2002, DOI: 10.1038/nsb762

The SIN domain within histones H3 and H4 is defined by a setof single amino acid substitutions that were initially identi-fied as mutations that alleviate the transcriptional defectsassociated with inactivation of the SWI/SNF chromatinremodeling complex. Here we use recombinant histones toinvestigate how Sin– versions of H4 alter the structure ofnucleosomal arrays. We find that an R45C substitution with-in the SIN domain of H4 does not disrupt nucleosome posi-tioning nor does this Sin– version alter the accessibility ofnucleosomal DNA. In contrast, we find that the R45C substi-tution eliminates Mg2+-dependent, intramolecular folding ofthe nucleosomal arrays. Our results suggest that Sin– ver-sions of histones may alleviate the need for SWI/SNF in vivoby disrupting higher-order chromatin folding.

A connection between the function of the SWI/SNF remodel-ing complex and chromatin was first established through geneticstudies in Saccharomyces cerevisiae in which mutations affectingseveral chromatin components alleviated the defects in growthand transcription due to swi– or snf – mutations1. For example, agroup of mutations in genes encoding histones H3 or H4 wereisolated that restored activator function in the absence of SWI orSNF products2. These Sin– (switch independent) mutations weresubsequently found to cause single amino acid substitutions inthe central histone-fold domains of H3 and H4 that disrupt thesame histone–DNA contact site near the nucleosomal dyad axis(the SIN domain)2,3. One of the residues altered in Sin– histonesis Arg 45 of histone H4 (H4R45), which is one of the 10 Argresidues within the histone octamer that protrudes into theminor groove of DNA. Because disruption of this DNA–histonecontact may weaken the central wrap of DNA, one attractivepossibility is that Sin– versions of histones may create an alterednucleosome structure that mimics SWI/SNF-dependent nucleo-some disruption. Similarities between SWI/SNF remodelingin vitro and the in vivo effects of Sin– versions of histone H4 havesupported this view4.

Assembly of recombinant nucleosomal arraysTo determine how Sin– versions of histone H4 disrupt nucleo-some structure and function, recombinant histone octamersthat harbor wild type or a Sin– version of histone H4 were usedto assemble model nucleosomal arrays. A DNA template com-posed of 11 or 12 head-to-tail repeats of a 208 bp 5S rRNA genefrom Lytechinus variegatus (the 208-11 or 208-12 template,respectively) was used (Fig. 1). Each 5S repeat can rotationally

and translationally position a nucleosome after in vitro salt dial-ysis reconstitution, yielding a positioned array of nucleo-somes5–7. Our analyses have focused on Sin– versions of histoneH4 in which Arg 45 is replaced by either a Cys (H4R45C) or Hisresidue (H4R45H), and the majority of our studies use theH4R45C version because it yields the strongest Sin– phenotypein vivo4. Wild type and mutant histones were expressed individ-ually in bacteria, and each was purified from the insoluble frac-tion of bacterial lysates. Denatured histones were mixed in equalratios, and the renatured histone octamers were isolated by gelfiltration as described8. No significant differences in the effi-ciency of octamer reconstitution were detected when H4R45Cor H4R45H were substituted for wild type H4. Nucleosomalarrays were then assembled by salt gradient dialysis6, using vary-ing ratios (r) of histone octamers to 5S DNA positioningsequences in order to generate a fully saturated array — forexample, r = 0.9–1.6.

nature structural biology • volume 9 number 3 • march 2002 167

Fig. 1 Sin– versions of histone H4 are competent to form nucleosomalarrays. a, Schematic of the array templates. The templates used for thesestudies are composed of head-to-tail repeats of a 208 bp 5S rDNAsequence. The central repeat of the 208-11 template has been altered tocontain a unique SalI restriction site (shown in parentheses). Otherrestriction sites used in these studies are present in each repeat of the208-11 and 208-12 templates. The major and minor nucleosome transla-tional positions are indicated by arrows. b, SDS-PAGE of recombinanthistones. The recombinant Xenopus histones used were separated on18% SDS-PAGE gels and stained with Coomassie blue. c, R45C Sin– nucle-osomes are indistinguishable by native PAGE. Arrays harboring eitherwild type or the R45C version of histone H4 were cleaved with EcoRI,electrophoresed on a native 4% PAGE and stained with ethidium bro-mide. The ‘Nuc’ and ‘Naked’ labels indicate mononucleosomal andnaked DNA, respectively. d, Sin– and wild type nucleosomes have identi-cal translational positioning. Assembled arrays were digested with a col-lection of restriction enzymes that recognize sites throughout the 5Srepeat to measure the relative occlusion of these sites by nucleosomeoccupancy. Following cleavage, arrays were deproteinized to removehistones, run on a 1% agarose TAE gel and Southern blotted with a ran-dom primed probe to 208-11.

a

b

c

d

©20

02 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://s

tru

ctb

io.n

atu

re.c

om

letters

As an initial means to monitor the efficiency of nucleosomeassembly, we digested the reconstituted nucleosomal arrays withEcoRI. Because each 5S rDNA repeat in the 208-11 and 208-12array templates is bordered by EcoRI restriction sites (Fig. 1a),EcoRI cleavage releases either a 208 bp free-DNA fragment or amononucleosome that can be identified by its slower mobilityafter native gel electrophoresis. Typically, a fully assembled (sat-urated) nucleosomal array that is reconstituted with chickenerythrocyte histone octamers yields ∼ 2–5% free DNA in theEcoRI assay9–11. Recombinant wild type and Sin– histoneoctamers yielded similar levels of nucleosome density at nearlyidentical ratios of octamers to 5S DNA repeat (r = 1.2–1.3), indi-cating that the recombinant octamers readily formed nucleo-somes and that the Sin– version of H4 does not disruptnucleosome assembly (Fig. 1c; data not shown). Similarly, diges-tion with a collection of restriction enzymes that recognize sitesboth internal and external to the major 5S nucleosome position-ing frame yielded an identical cleavage pattern for both wild typeand Sin– nucleosomal arrays, indicating that the wild type andSin– nucleosomes are similarly positioned on the DNA templates(Fig. 1d).

A second technique, multigel analysis, was used to verify thatthe wild type and Sin– nucleosomal arrays were of comparable

saturation extent11–13. This technique allows determination ofthe gel-free electrophoretic mobility (µo) of a nucleosomal arrayby extrapolation of a plot of its migration in different agaroseconcentrations. The µo value is proportional to the average sur-face charge density of the array; thus, µo monitors the degree ofnucleosome assembly. This gel analysis also allows the measure-ment of the effective radius (Re) of the array at different effectivegel pore sizes (Pe). Multigel analysis of the wild type recombi-nant nucleosomal arrays displayed an effective radius (Re) that isessentially identical to that of saturated nucleosomal arraysreconstituted with chicken erythrocyte histone octamers11–13

(Table 1). The gel-free mobility (µo) of the wild type recombi-nant arrays was slightly larger than the µo of chicken arrays,which is expected due to the complete lack of posttranslationalmodifications for the recombinant arrays. These recombinantnucleosomal arrays sediment as a fairly homogeneous popula-tion of 29–30S species in the analytical ultracentrifuge, similarto the behavior of saturated chicken erythrocyte 208-12 nucleo-somal arrays in low salt buffer6,10,11,14 (Fig. 2a). The Sin– nucleo-somal arrays showed a µo value identical to that of the wild typerecombinant arrays, confirming that they are of equal saturationextent (Table 1). However, the Sin– arrays had an increased effec-tive radius (Re), suggesting a more extended conformation inthis low salt TAE buffer (Table 1). Consistent with this observa-tion, Sin– arrays sedimented at smaller S values in the analyticalultracentrifuge, with a distribution range spanning 24–27S(Fig. 2a). Similar results were obtained for arrays reconstitutedwith octamers harboring the R45H version of H4 (data notshown). The more extended conformation of the Sin– arraycould result from a slight unwrapping of nucleosomal DNA inlow salt buffer, similar to previous studies using nucleosomesthat lack the histone N-terminal domains10,15,16. Together, these

168 nature structural biology • volume 9 number 3 • march 2002

Fig. 2 Sin– versions of histone H4 abolish intramolecular folding ofarrays. a, Intramolecular folding is abolished by the R45C version of his-tone H4. This is shown by sedimentation velocity analysis of 208-12arrays in the presence or absence of Mg2+. The G(s) distributions (as cal-culated by the method of van Holde and Weischet)28 are depicted for theindicated arrays sedimented in either TE (10 mM Tris 8.0 and 0.25 mMEDTA) or TE with 1.75 mM MgCl2. S20,w is the sedimentation corrected towater at 20 °C. b, Intermolecular association is not altered by the R45Cversion of histone H4. Nucleosomal arrays were incubated in TE withvarying concentrations of MgCl2 at room temperature for 15 min, fol-lowed by centrifugation in a microfuge at 14,000g for 10 min. The per-centage of array remaining in the supernatant is plotted as a function ofMgCl2 concentration.

a

b

Fig. 3 SWI/SNF efficiently remodels both wild type and Sin– arrays.Radiolabeled wild type or Sin– 208-11 nucleosomal arrays (1 nM) wereassayed for SWI/SNF remodeling. Following addition of SalI and yeastSWI/SNF (1 nM), reactions were initiated with ATP (closed symbols) andcompared to control reactions lacking ATP (open symbols). At the timepoints indicated, aliquots of the remodeling reactions were removed,deproteinized and run on a 1% agarose gel. The percent DNA cleaved,reflecting remodeling by SWI/SNF, was determined by PhosphorImagerquantification of the dried gel. Squares are wild type arrays; triangles,H4R45C Sin– arrays; and circles, H4R45H Sin– arrays. Inset: a shorter time-course of remodeling with 1 nM of each array and 3 nM SWI/SNF.

©20

02 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://s

tru

ctb

io.n

atu

re.c

om

letters

data confirm that Sin– versions of histone H4 do not significantlyimpair either octamer or nucleosome assembly. However, theSin– nucleosomal arrays do adopt a slightly more extended con-formation in low salt buffers.

Remodeling of Sin– arrays by SWI/SNFThe molecular basis for the Sin– phenotype of H4 mutantscould be explained by the simple hypothesis that the require-ment for SWI/SNF might be bypassed if a Sin– version of H4creates a chromatin structure that mimics the remodeled state.To test this hypothesis, we used a quantitative remodeling assayto monitor SWI/SNF activity on arrays reconstituted with Sin–

histone octamers (H4R45C or H4R45H). This remodeling assayuses a modified version of the 208-11 template that contains aunique SalI restriction enzyme site in the central repeat of thearray17 (208-11-SalI; Fig. 1a). In the reconstituted complex withhistone octamers, the SalI site in the template is located near thenucleosomal dyad axis of the central nucleosome. Determiningthe rate of SalI digestion of these 208-11–SalI arrays yields aquantitative measurement of the accessibility of nucleosomalDNA in the presence or absence of remodeling activity17–19. IfSin– versions of H4 create a nucleosome structure that mimics a

SWI/SNF-remodeled state, then the Sin– nucleosomal arraysshould harbor accessible SalI restriction sites. Contrary toexpectations, Sin– nucleosomal arrays exhibited nearly identicalocclusion of the SalI site in the absence of ATP (or in theabsence of SWI/SNF) (Fig. 3; data not shown). Thus, the DNAwrapped onto Sin– histone octamers does not appear to beinherently more accessible to restriction enzymes. Further-more, when ATP was added to the remodeling assays, we foundthat the Sin– arrays were excellent substrates for SWI/SNFaction (Fig. 3). Notably, measurements of the initial velocity ofthe remodeling reaction demonstrated that the Sin– arrays wereactually better substrates for SWI/SNF than the wild type arrays(Fig. 3, inset). Because this remodeling assay may measure therate of nucleosome movements by ATP-dependent remodelingenzymes20, these data suggest that Sin– nucleosomes are inher-ently easier to mobilize.

Folding of Sin– nucleosomal arraysIf Sin– nucleosomes are not equivalent to the remodeled state (asdefined by in vitro assays), then how do Sin– histones alleviatethe need for SWI/SNF action in vivo? Recently, we have shownthat SWI/SNF plays a key role in the transcription of a set of

Fig. 4 The SIN domain plays a key role in the organization ofboth the central and peripheral wraps of nucleosomal DNA. a, Location of the ‘SIN domain’ on the nucleosome. The ribbondiagram of the nucleosome, with the dyad oriented to the topof the panel, shows the H3 (blue)–H4 (green) tetramer associ-ated with the dyad and entry/exit face of the particle. Theentry/exit DNA segments are highlighted as yellow duplexes.The region of the nucleosome enlarged in (b) is boxed in (a).For clarity, only one H2A (orange)–H2B (red) dimer is depicted.b, Interactions of Arg 45 of H4 at the nucleosomal dyad (SHL0.5). The ribbon diagram of histones H3 (blue) and H4 (green)shows the L1L2 region (encompassing the SIN domain) of his-tones H3 and H4 associating with the central wrap of nucleoso-mal DNA near the dyad (SHL 0.5). Arg 45 of histone H4 isshown inserting into the minor groove at SHL 0.5. Predictedhydrogen bonds between amino acids Thr 118 of H3, Arg 45 ofH4, and DNA are shown as dashed green lines. The DNAstrands are shown in orange and yellow, with the DNA base(A229 as denoted in the crystal coordinates3) contacted by theN-helix of histone H3 in red. Note the compression of theminor grove and bending of the DNA due to the interaction ofArg 45 with DNA. c, The H3 N-helix helps organize both thecentral and peripheral wraps of nucleosomal DNA. The rotatedview of the nucleosome shows the N-helix of histone H3 (bluehelix at center), including the region (light blue Cα trace)involved in contacting the phosphoribose of A229 (red DNAnucleoside in b,c). This helix bridges SHL 0.5 (two base pairsshown at left) and SHL 6.5 (the multicolored helix at right). Aportion of the N-terminal tail of histone H3 is shown extendingbetween the two gyres (blue coil extending from the N-helix).Structural coordinates3 were obtained from the BrookhavenPDB database (1AOI). For a detailed discussion of histone–DNAinteractions, see ref. 29. Images were produced usingMOLSCRIPT30.

a b

c

Table 1 Low salt properties of 208-12 nucleosomal arrays

Histone octamer source R-value1 Save (S)2 –µo cm2 (V s 10–4)–1 Re (nm)3 Intramolecular folding Self-association

Chicken erythrocyte4 1.2 29 1.95 26.7 ± 0.7 Yes YesWild type recombinant Xenopus 1.3 29 1.88 27.5 ± 1.2 Yes YesH4 R45C recombinant Xenopus 1.2 27 1.89 29.0 ± 1.8 No Yes

1Values indicate molar ratio of core histone octamer:208 bp repeat used for array reconstitution2Values represent the S20,w obtained at the boundary fraction midpoint in TEN Buffer (10 mM Tris-HCl, 0.25 mM Na2EDTA and 2.5 mM NaCl, pH 7.8)3Values represent the mean standard deviation of 7–9 determinations at Pe ≥ 200 nm in E Buffer (40 mM Tris-HCl and 0.25 mM Na2EDTA, pH 7.8).4Values shown are from ref. 13.

nature structural biology • volume 9 number 3 • march 2002 169

©20

02 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://s

tru

ctb

io.n

atu

re.c

om

letters

genes expressed at the end of mitosis21. These studies suggested amodel in which SWI/SNF counteracts chromatin condensationrather than disrupting individual nucleosomes. This model raises an alternative view for why transcription of Sin– chro-matin might be less dependent on SWI/SNF action — perhapsSin– nucleosomal arrays are less competent to fold into higherorder, condensed states. To investigate this possibility, we tookadvantage of the observation that model 208-12 arrays aredynamic macromolecular assemblies that undergo reversiblefolding events, much like physiological chromatin fibers22. Inlow ionic strength buffers, such as TE, 208-12 arrays exist as anextended, flexible fiber that sediments in the analytical ultracen-trifuge as a nearly homogenous distribution of ∼ 29S species(Fig. 2a). When divalent cations (Mg2+) are introduced, thearrays form a heterogeneous, faster-sedimenting species with a30–55S distribution; formation of the 55S species is consistentwith formation of a compact, 30 nM-like chromatin fiber6,23.These same folding equilibria are established with our wild typerecombinant nucleosomal arrays, which sediment at 30–55S inMg2+-containing buffers (Fig. 2a). However, the Sin– arrays sedi-ment in Mg2+ buffer as a population of 27–32S species, indicat-ing a strong defect in array folding and a complete inability toform the fully compacted 55S state (Fig. 2a). This striking defectin intramolecular folding is similar to the disruption of arrayfolding due to removal of the histone N-terminal ‘tail’domains10,15,16. In addition, Sin– and ‘tailless’ arrays both show asmall Mg2+-dependent increase in S20,w (sedimentation correctedto water at 20 °C)15. This behavior has been attributed to acation-dependent stabilization of the peripheral wrapping ofDNA on each individual nucleosome within the array15. Thesedata are fully consistent with the idea that Sin– versions of his-tones alleviate the need for SWI/SNF in transcription by destabi-lizing the folding of chromatin fibers. Likewise, these resultssuggest that in vivo SWI/SNF may counteract chromatin-mediated transcriptional repression by disrupting the folding ofchromatin fibers.

In addition to these intramolecular folding reactions, divalentcations (>2 mM Mg2+) or low concentrations of polyamines(∼ 200 µM) can induce nucleosomal arrays to reversiblyoligomerize24,25. Intermolecular oligomerization generatesdefined structures that sediment >1,000S and are believed tomimic the fiber–fiber interactions that stabilize higher orderchromosomal domains, such as chromonema fibers24,26. Despitethe defect in intramolecular folding, Sin– and wild type nucleo-somal arrays displayed identical oligomerization profiles whenincubated in varying MgCl2 concentrations (Fig. 2b). Thus, Sin–

versions of histone H4 selectively disrupt intramolecular foldingand do not perturb fiber oligomerization. This is consistent withprevious results indicating that these two transitions are notobligatorily coupled24,25.

How do Sin– versions of H4 disrupt array folding?Previous studies have demonstrated that the histone N-terminaltail domains are also required for intramolecular folding ofmodel 208-11 and 208-12 arrays10,15,16. However, histone tailsalso play an essential role in the oligomerization of 208-11 and208-12 arrays10,24. Thus, Sin– versions of H4 are the first exam-ples of a histone alteration that selectively disrupts intramolec-ular folding. Although the basis for the defect in intramolecularfolding is not clear, examination of the crystal structure of thenucleosome provides a possible explanation3 (Fig. 4). Arg 45 ofhistone H4 and several other residues altered in Sin– versions ofhistone H3 are involved in a collection of critical histone–DNA

contacts within the DNA minor groove near the nucleosomaldyad (designated SHL0.5)3. Insertion of Arg 45 into the DNAminor groove leads to a significant distortion and compressionof the minor groove. These changes in DNA structure are neces-sary to allow the N-terminal helix (‘N-helix’) of histone H3 tomake a number of hydrogen bonding contacts with the centralwrap of DNA (Fig. 4b,c). In addition to contacting the centralwrap of DNA, the N-helix of H3 also helps to organize theperipheral DNA at the edge of the nucleosome (Fig. 4c). Thus,Arg 45 of H4 plays a unique role in organizing a complex hydrogen-bonding network between histones and DNA at boththe nucleosomal dyad and the periphery of the nucleosome(Fig. 4c). Disruption of this network by Sin– versions of H3 orH4 could, therefore, disrupt DNA–histone interactions at boththe central wrap and at the nucleosomal edge. Destabilization ofDNA at the nucleosomal periphery is consistent with our obser-vation that Sin– nucleosomal arrays appear to be more extendedin low salt buffers — that is, nucleosomal DNA is released intothe linker — and also with a previous study of Sin– versions ofhistone H3 (ref. 27). Disruption of DNA conformation could byitself be responsible for the defects in folding, perhaps through adirect steric clash with the geometrical constraints of the con-densed state. Alternatively, the disruption of contacts betweenthe H3 N-helix and DNA may influence the conformation ororientation of the histone H3 N-terminal tail as it exits thenucleosome core particle. We favor the latter model in whichthese indirect effects propagated through the H3 N-terminaldomain may disrupt nucleosome–nucleosome interactionsrequired for fiber condensation.

MethodsRecombinant histone and histone octamer preparation.Recombinant Xenopus laevis histones were purified from inclusionbodies of BL21 (DE3) cells using published methods8. Proteins weretypically >95% pure after the first chromatography step and >99%pure after two steps. Following SDS-PAGE, histone-containing frac-tions were immediately dialyzed against water to remove urea andthen lyophilized.

For octamer reconstitution, lyophilized proteins (0.2 mmol eachof histones H3 and H4, and 0.21 mmol each of histones H2A andH2B) were denatured in guanidine HCl (7 M guanidine-HCl, 20 mMTris-HCl, pH 7.5, and 10 mM dithiothreitol (DTT)) and renatured bydialysis into TE (10 mM Tris 8.0 and 0.25 mM EDTA) with 2 M NaCl.Histone octamers were fractionated by gel filtration on a Superose200 gel filtration column (Amersham), and identified by SDS-PAGEand Coomassie blue staining.

Nucleosomal array characterization. Histones were depositedonto DNA templates (Fig. 1) by salt dialysis6. Deposition was verifiedby EcoRI cleavage (10–20 units) of nucleosomal arrays (250–500 ng)in 10 mM Tris-HCl, pH 8.0, 125 mM NaCl, 2.5 mM MgCl2 and 1 mMEDTA for 30 min to 1 h at 37 °C. Quantitative agarose gel analysiswas performed as described12,13, and data were analyzed exactly asdescribed11. Data are reported as the averages and standard devia-tions of Re from the entire data set.

SWI/SNF remodeling. Nucleosomal arrays (1 nM) were assayed forSWI/SNF (1 or 3 nM) remodeling in buffer containing 20 mM Tris-HCl, pH 8.0, 125 mM NaCl, 5 mM MgCl2, 100 µg ml–1 BSA and1 mM DTT essentially as described17.

Sedimentation velocity analyses. Sedimentation velocity exper-iments were performed in a Beckman Optima XL-I analytical ultra-centrifuge using scanner optics at 260 nm. Initial absorbances forthe experiments shown were >0.6 A260. Samples were equilibratedat 20 °C under vacuum for at least 1 h prior to sedimentation at25,000g in an An 60 Ti rotor. Boundaries were analyzed by the

170 nature structural biology • volume 9 number 3 • march 2002

©20

02 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://s

tru

ctb

io.n

atu

re.c

om

letters

method of van Holde and Weischet28 using UltraScan version 4.0 forUnix. Data were plotted as boundary fraction (y-axis) versus S20,w

(sedimentation corrected to water at 20 °C) to yield the G(s) distrib-utions (Fig. 3a).

AcknowledgmentsWe would like to thank K. Luger for assistance in preparation of recombinanthistones and helpful advice throughout the course of these studies, and M.Shogren-Knaak and E. Merithew for help with the preparation of Fig. 4. Thesestudies were supported by grants from the NIH to C.L.P. and J.C.H., and a NIHNRSA to P.J.H.

Competing interests statementThe authors declare that they have no competing financial interests.

Correspondence should be addressed to C.L.P. email:Craig.Peterson@umassmed.edu

Received 24 September, 2001; accepted 28 December, 2001.

1. Winston, F. & Carlson, M. Trends Genet. 8, 387–391 (1992).2. Kruger, W. et al. Genes Dev. 9, 2770–2779 (1995).3. Luger, K., Mader, A.W., Richmond, R.K., Sargent, D.F. & Richmond, T.J. Nature

389, 251–260 (1997).4. Wechser, M.A., Kladde, M.P., Alfieri, J.A. & Peterson, C.L. EMBO J. 16, 2086–2095

(1997).

5. Simpson, R.T., Thoma, F. & Brubaker, J.M. Cell 42, 799–808 (1985).6. Hansen, J.C., Ausio, J., Stanik, V.H. & van Holde, K.E. Biochemistry 28, 9129–9136

(1989).7. Dong, F., Hansen, J.C. & van Holde, K.E. Proc. Natl. Acad. Sci. USA 87, 5724–5728

(1990).8. Luger, K., Rechsteiner, T.J. & Richmond, T.J. Methods Mol. Biol. 119, 1–16 (1999).9. Hansen, J.C. & Lohr, D. J. Biol. Chem. 268, 5840-5848 (1993).

10. Tse, C. & Hansen, J.C. Biochemistry 36, 11381–11388 (1997).11. Carruthers, L.M., Bednar, J., Woodcock, C.L. & Hansen, J.C. Biochemistry 37,

14776–14787 (1998).12. Fletcher, T.M., Serwer, P. & Hansen, J.C. Biochemistry 33, 10859–10863 (1994).13. Fletcher, T.M., Krishnan, U., Serwer, P. & Hansen, J.C. Biochemistry 33, 2226–2233

(1994).14. Hansen, J.C. & Lohr, D. J. Biol. Chem. 268, 5840–5848 (1993).15. Fletcher, T.M. & Hansen, J.C. J. Biol. Chem. 270, 25359–25362 (1995).16. Garcia-Ramirez, M., Dong, F. & Ausio, J. J. Biol. Chem. 267, 19587–19595 (1992).17. Logie, C. & Peterson, C.L. EMBO J. 16, 6772–6782 (1997).18. Logie, C., Tse, C., Hansen, J.C. & Peterson, C.L. Biochemistry 38, 2514–2522 (1999).19. Boyer, L.A. et al. J. Biol. Chem. 275, 18864–18870 (2000).20. Jaskelioff, M., Gavin, I., Peterson, C.L. & Logie, C. Mol. Cell. Biol. 20, 3058–3068

(2000).21. Krebs, J.E., Fry, C.J., Samuels, M. & Peterson, C.L. Cell 102, 587–598(2000).22. Fletcher, T.M. & Hansen, J.C. Crit. Rev. Eukaryot. Gene Expr. 6, 149–188 (1996).23. Schwarz, P.M. & Hansen, J.C. J. Biol. Chem. 269, 16284–16289 (1994).24. Schwarz, P.M., Felthauser, A., Fletcher, T.M. & Hansen, J.C. Biochemistry 35,

4009–4015 (1996).25. Pollard, K.J., Samuels, M.L., Crowley, K.A., Hansen, J.C. & Peterson, C.L. EMBO J.

18, 5622–5633 (1999).26. Belmont, A.S. & Bruce, K. J. Cell Biol. 127, 287–302 (1994).27. Kurumizaka, H. & Wolffe, A.P. Mol. Cell. Biol. 17, 6953–6969 (1997).28. van Holde, K.E. & Weischet, W.O. Biopolymers 17, 1387–1403 (1978).29. Luger, K. & Richmond, T.J. Curr. Opin. Struct. Biol. 8, 33–40 (1998).30. Kraulis, P.J. J. Appl. Crystallogr. 24, 946–950 (1991).

nature structural biology • volume 9 number 3 • march 2002 171

©20

02 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://s

tru

ctb

io.n

atu

re.c

om