Two Homologous Domains of Similar Structure butDifferent Stability in the Yeast Linker Histone, Hho1p

Tariq Ali†, Patrick Coles†, Timothy J. Stevens†, Katherine Stott andJean O. Thomas*

Cambridge Centre forMolecular Recognition andDepartment of BiochemistryUniversity of Cambridge80 Tennis Court RoadCambridge CB2 1GA, UK

The Saccharomyces cerevisiae homologue of the linker histone H1, Hho1p,has two domains that are similar in sequence to the globular domain ofH1 (and variants such as H5). It is an open question whether bothdomains are functional and whether they play similar structural roles.Preliminary structural studies showed that the two isolated domains, GIand GII, differ significantly in stability. In 10 mM sodium phosphate(pH 7), the GI domain, like the globular domains of H1 and H5, GH1and GH5, was stably folded, whereas GII was largely unstructured. How-ever, at high concentrations of large tetrahedral anions (phosphate,sulphate, perchlorate), which might mimic the charge-screening effects ofDNA phosphate groups, GII was folded. In view of the potential signifi-cance of these observations in relation to the role of Hho1p, we havenow determined the structures of its GI and GII domains by NMR spec-troscopy under conditions in which GII (like GI) is folded. The backboner.m.s.d. over the ordered residues is 0.43 A for GI and 0.97 A for GII.Both structures show the “winged-helix” fold typical of GH1 and GH5and are very similar to each other, with an r.m.s.d. over the structuredregions of 1.3 A, although there are distinct differences. The potential forGII to adopt a structure similar to that of GI when Hho1p is bound tochromatin in vivo suggests that both globular domains might befunctional. Whether Hho1p performs a structural role by bridging twonucleosomes remains to be determined.

q 2004 Elsevier Ltd. All rights reserved.

Keywords: linker histone; NMR spectroscopy; winged helix; DNA-bindingdomain; intrinsically unstructured domain*Corresponding author

Introduction

Hho1p, the putative linker histone in Saccharo-myces cerevisiae,1 unusually, has two domains2 (GIand GII; 43% identity) that are homologous to thesingle “winged-helix”3 globular domain of histone

H14 and its variant H5.5 The inter-domain linker(,40 residues) is similar in sequence and compo-sition to the long C-terminal tail of histone H1and, like H1, Hho1p has a shorter basic N-terminaltail. When reconstituted with H1-depleted chroma-tin, the globular domains of H1 and H5,6 like theintact proteins,7 are able to protect about 166 bp(“chromatosome-length”) of DNA from nucleasedigestion, also a property of Hho1p when boundto HeLa dinucleosomes.8 Critical to an understand-ing of whether Hho1p is a “bifunctional” equival-ent of higher eukaryote H1 is an evaluation of thestructure and properties of its GI and GII domains.Hho1p might, in principle, be able to bridge adja-cent nucleosomes; alternatively, the two domainsmight bind within a nucleosome. Normally, only asingle H1 (and thus a single globular domain) isbound per nucleosome.9 However, since binding isasymmetric on the symmetric nucleosome core,10

there is formally a second, symmetry-related site

0022-2836/$ - see front matter q 2004 Elsevier Ltd. All rights reserved.

Supplementary data associated with this article can befound at doi: 10.1016/j.jmb.2004.02.046

Present addresses: T. Ali, Department of Biochemistry,South Parks Road, Oxford OX1 3QU, UK; P. Coles,Department of Chemical Engineering, University ofCalifornia, Berkeley, 201 Gilman Hall, Berkeley, CA94720, USA.

Abbreviations used: NOE, nuclear Overhauser effect;NOESY, NOE spectroscopy; TOCSY, total correlationspectroscopy; HSQC, heteronuclear single quantumcoherence.

† T.A., P.C. & T.J.S. contributed equally to this work.

E-mail address of the corresponding author:jot1@bioc.cam.ac.uk

doi:10.1016/j.jmb.2004.02.046 J. Mol. Biol. (2004) 338, 139–148

and presumably this is the source of lower-affinitybinding of a second H1 molecule to H1-depletedchromatin.11

In biochemical studies of the isolated GI and GIIdomains, expressed as recombinant proteins inEscherichia coli, we found that when added to linkerhistone-depleted chromatin they differed in theirability to protect about 166 bp of DNA from diges-tion by micrococcal nuclease.12 GI afforded protec-tion under the conditions tested, whereas GII didnot. Chromatosome protection by GH5 and GH1requires two functional DNA-binding sites onopposite faces of the globular domain;5,13 GII, incontrast to GI, lacks two of the four conservedbasic residues that would constitute site II,14

assuming that the three-dimensional structures aresimilar to those of the homologous GH5 and GH1.We also found that GI and GII differed significantlyin structural stability.12 The GI domain was foldedin 10 mM sodium phosphate, as were the isolatedglobular domains of H1 and H5,15 whereas GIIwas largely unfolded, but could be induced tofold at a higher concentration of phosphate or inthe presence of other large tetrahedral anions(perchlorate, sulphate), as we had found earlierfor the stabilization of a-helix in the C-terminaltails of H1 and several variants.16,17 GII is alsounstructured in intact Hho1p,12 and it appears tobe one of the growing number of proteins/domainsto be identified as “intrinsically unfolded”.18

Since the anions might mimic the effect on theprotein structure of interaction with DNA phos-phate groups, we have determined the structuresof GI and GII under ionic conditions in whichthey are both folded. The structures are very simi-lar, reflecting their high level of sequence identity,2

and provide a basis for further investigation of thestructural instability of GII.

Results and Discussion

Multiple-sequence alignment (Figure 1) illus-trates the presence of the two regions within theHho1p sequence that are homologous to theglobular domain of histone H1.2 According to the

alignment, the structured core of the globulardomain of GH5 determined by X-raycrystallography5 (residues 24–97) is homologousto residues 43–116 and residues 176–250 ofHho1p, which are contained within the recombi-nant constructs GI (residues 38–130) and GII(171–258), respectively.

Stability of the folded domains

The conditions for NMR were optimized using15N-heteronuclear single quantum coherence(HSQC) spectra, the results of which are summar-ized in Figure 2. NMR measurements for GI weremade on samples containing 2.7 mM protein, 90%H2O/10% 2H2O in 100 mM sodium phosphate (pH7.0), 1 mM EDTA. Under these conditions, GIadopted a single, stably folded conformation, asjudged by the 15N-HSQC spectrum (Figure 2a).Under the same conditions, GII exhibited two con-formations, one folded and the other unfolded, inslow exchange on the NMR chemical-shift time-scale (Figure 2b). The population of the foldedcomponent increased to .95% on raising the con-centration of sodium phosphate to 250 mM (Figure2c). The final conditions for NMR measurementson GII were thus 1 mM protein, 90% H2O/10%2H2O in 250 mM sodium phosphate (pH 7.0),1 mM dithiothreitol (DTT), 1 mM EDTA. For bothGI and GII, some peaks broadened at temperaturesabove 288 K. The GI sample was stable for manymonths, whereas GII samples had a useful lifetimeof only one to two weeks, during which time sig-nals from the folded component decreased steadilyin intensity and the viscosity increased, the sampleeventually taking on a gel-like consistency. It seemslikely that irreversible, non-native interactionsbetween the unfolded form(s) removed this com-ponent from solution, and thus reduced theamount of folded form by displacement of theunfolded $ folded equilibrium.

Extent of assignment and secondary structure

All backbone resonances were assigned in bothdomains, with the exception of Ser218 HN in GII.

Figure 1. A multiple-sequence alignment of the globular domains of several histone H1 homologues and the GI andGII regions of Hho1p. Positively charged residues equivalent to the GH5 DNA-binding sites I and II are highlighted inbold. The secondary structure indicated above the alignment is based on the X-ray crystal structure of GH5.5

140 Structures of the GI and GII Domains of Hho1p

For GI, all side-chain 13C and 15N resonances wereassigned except Lys117 Cd and Lys117 C1 and allside-chain protons except those in Lys38 andLys62 Hd, Lys117 Hd and H1. For GII, all side-chain 13C and 15N resonances were assigned andall side-chain protons except those in Lys171 andSer218.

Sequential and medium-range NOEs, backbonetorsion angles and chemical shift index show thatGI has a structured core, comprising residues47–117. GII shows a similar picture with anordered core comprising residues 181–251.

Backbone dynamics of GI

Sample instability precluded the acquisition ofhigh-quality relaxation data for GII. For GI, R1

and R2 relaxation rates, {1H}15N heteronuclearNOE values and generalized order parameter

values ðS2Þ are illustrated in Figure 3. R1 and R2

were obtained by a two-parameter fit to a single-exponential decay. The rotational diffusion tensorwas obtained as follows. Residues with largeamplitude internal motions on a time-scale longerthan a few hundred picoseconds were first ident-ified as having a value for the heteronuclearnuclear Overhauser effect (NOE) ,0.65. Anadditional residue, Lys107, was identified as hav-ing a significant contribution to R2 arising fromchemical exchange. The remaining residues wereused in combination with the lowest-energy struc-ture of GI to calculate the rotational diffusion ten-sor, which was found, to a good approximation, tobe axially symmetric with correlation time 11.5 nsand Dk=D’ 1.21. The entire data set was then ana-lyzed according to the model-free formalism toobtain values for the generalized order parameter,S2, and the correlation time for internal motion,

Figure 2. 15N-HSQC spectra at 288 K showing the GI domain and the effect of the concentration of sodium phos-phate on the GII domain. (a) GI in 100 mM sodium phosphate; (b) GII in 100 mM sodium phosphate; (c) GII in250 mM sodium phosphate. An unfolded conformation of GII in (b) is revealed by the presence of many small peaksin the range 8.0–8.5 1H ppm that disappear in (c). GII requires a minimum of 250 mM sodium phosphate in order toattain a single, stably folded conformation.

Figure 3. {1H}15N heteronuclearNOE values, R1 and R2 relaxationrates and generalized order para-meters, S2, for the GI domain,plotted versus residue number. Therotational diffusion tensor wasfound, to a good approximation, tobe axially symmetric with corre-lation time 11.5 ns and Dk=D’ 1.21.

Structures of the GI and GII Domains of Hho1p 141

te.19 – 21 An additional parameter, Rex, describing the

contribution to R2 arising from chemical exchange,was included in the fit for Lys107. Random errorswere estimated from the pairwise average r.m.s.d.of fitted values obtained from two complete datasets recorded separately.

The data show that the structured core of GIdetermined from the relaxation measurementscomprises residues 46–116. The core is mostlyrigid, with the exception of the loop regions,which show suppressed values of both the hetero-nuclear NOE and S2: Interestingly, the b-sheet

Figure 4. Ensembles of backbonestructures for 25 energy-minimizedconformations of the GI and GIIdomains. Ordered regions of back-bone (GI: r.m.s.d. ,0.6 A; GII:r.m.s.d. ,1.0 A) are illustrated inblack, flexible loops in blue andflexible termini in red.

Table 1. Summary of the restraints used in the calculation of the domain structures and characterization of the energy-minimized structures

GI GII

NOE upper distance limitsTotal 2771 1509Intra-residue 871 786Sequential ðli 2 jl ¼ 1Þ 665 302Medium range ðli 2 jl , 5Þ 581 156Long range ðli 2 jl . 4Þ 654 251Dihedral angle constraintsTotal 42 102Residual NOE violations (A)Number $0.5 0.2 ^ 0.6 0.5 ^ 0.5Number $0.1 32.4 ^ 4.0 17.6 ^ 2.5Maximum 0.85 1.34Residual angle violations (deg.)Number $2.0 0 0Energies (kcal mol21)Total 23509 ^ 155 23150 ^ 79van der Waals 2695 ^ 6 2647 ^ 8Electrostatic 23463 ^ 146 23051 ^ 86Ramachandran statistics (% residues)Core regions 85.1 83.7Allowed regions 10.2 13.7Generously allowed regions 2.0 1.5Disallowed regions 2.7 1.1r.m.s.d. from ideal geometryBond lengths (A) 0.0038 ^ 0.0001 0.0035 ^ 0.0002Bond angles (deg.) 0.516 ^ 0.026 0.467 ^ 0.023r.m.s.d. to mean coordinates (A)Backbone (GI:47–117; GII:181–251) 0.427 0.974Heavy atoms (GI:47–117; GII:181–251) 0.837 1.357All atoms (GI:47–117; GII:181–251) 1.051 1.602

For GI, dihedral constraints (backbone f) were obtained from analysis of a 3D HNHA experiment. For GII, dihedral constraints(both backbone f and w) were derived using the program TALOS.35 For the analysis of each globular domain, ten water-refined,energy-minimized structural conformations were used. Ramachandran statistics were calculated using PROCHECK.38

142 Structures of the GI and GII Domains of Hho1p

region also appears to show some flexibility, whichis reflected in suppressed heteronuclear NOE andS 2 values for residues 102–105 and 112–116relative to the rest of the structured regions. Inaddition, Lys107 in the intervening turn appearsto display conformational mobility on the micro-to millisecond time-scale, as indicated by itsrelatively much increased R2 relaxation rate. Theanisotropy in the diffusion tensor is reflected inthe values of R1 and R2 for residues in helix III,which lies parallel with the long axis of the protein,and therefore shows lower R1 and higher R2 valuesthan the residues in helices I and II.

Structure calculation

Overall, the structure calculations for the GI andGII domains converged readily upon consistentstructural models. The higher quality NMR spectrafor the GI domain yielded more structuralrestraints than for GII (Table 1) because the GIIsample deteriorated during the NMR experiments(see above). The GI structure is therefore much bet-ter defined (0.43 A backbone r.m.s.d. over theordered residues compared with 0.97 A for GII).As expected, in both cases the a-helices and b-strands are better defined than the chain terminiand many of the loops (Figure 4), where there is alack of structural constraints due to the flexibilityshown by the relaxation data (Figure 3).

Description of the structure

The GI and GII domains are very similar (Figure5) and, like GH14 and GH5,5 show the “winged-helix” fold characteristic of the forkhead family of

transcription factors such as HNF3g,3 with a single“wing” as in GH5 and GH1. GI and GII both com-prise three major a-helices (I–III) that are approxi-mately orthogonal to one another, with a three-strand b-sheet capping one end of the structurebetween three helix termini. Between helices IIand III there is a further small helical region thatis not present in GH1 and GH5. In GI, this is struc-tured and comprises a full a-helical turn, whereasin GII, which has one more residue in this region,it is defined more loosely, although it followsapproximately the same path as in GI. The b-sheethas one strand that lies between helices I and II inthe sequence and two strands to the C-terminalside of helix III. There are flexible loops at eitherend of the b-sheet and immediately before thethird major helix. The most striking structuraldifference within the ordered regions of the twodomains is in the b-sheet region. The b-sheet ofGII is relatively flat compared with the twistedsheet of GI, resulting in the flanking loops (whichincludes the wing) being at different angles relativeto the helix axes in GI and GII (Figure 6).

Folding and stability of GI and GII

Despite the sequence similarities between GI andGII, and the structural similarities under the con-ditions of the NMR experiments, the structuralstability of the two domains differs dramatically.Preliminary experiments using CD and 1D 1H-NMR had suggested that in 10 mM sodium phos-phate GII was relatively unstructured, whereas GIwas folded stably.12 Moreover, as described above,15N-HSQC experiments clearly show that the GIIsample contains a significant unfolded componenteven in 100 mM sodium phosphate (Figure 2b), inwhich only a folded component was observed forGI (Figure 2a). The basis for the instability of GIIis not clear from the final structures, determinedunder stabilizing conditions; both have well-packed hydrophobic cores and the structured GIIis not lacking in any obvious interactions com-pared with GI. The reasons for the differences in

Figure 5. Ribbon diagrams of (a) the top and (b) thefront aspects of the generated cores of the GI (residues47–117) and GII (residues 181–251) backbones, con-structed using MOLMOL.39

Figure 6. Comparison of GI (yellow), GII (red) andGH5 (blue) structures. The structures were super-imposed by minimizing the r.m.s.d. between a-helixbackbone atoms. The diagrams were generated usingMOLMOL.39

Structures of the GI and GII Domains of Hho1p 143

stability would not be expected to be obvious fromsequence comparisons. However, we note thatthere are two prominent regions of sequence differ-ence between GI and GII that might perhaps con-tribute to the differences in stability, andspecifically might contribute to the instability ofGII. The first conspicuous difference (Figure 1) liesbetween helices II and III in the structures. The rel-evant sequences are: GI (residues 78–83), PIVGSA;GII (residues 211–217), SSKLKTS; on the GESscale22 these have hydrophobicity values of26.9 kcal mol21 and þ7.9 kcal mol21, respectively.This is the region in both GI and GII that containsan additional short helical segment (about oneturn), which is less well defined in GII than in GI.It may be significant that in GI there is a hydro-phobic interface between Ile and Val in this seg-ment and major helix II, whereas in GII thecorresponding residues are hydrophilic. The lossof hydrophobic contacts might lead to a loosertether between helix II and helix III in GII, whichcould be a contributor to folding instability, butthere is no salt-bridge or side-chain hydrogenbond in GI that is absent from GII. Another poten-tial contribution to the instability of GII arisesfrom the lysine residues in the loop, one of which(Lys215) might pack against Lys207 in helix II ifthe loops in GII were organized in the same wayas those in GI.

The second region of conspicuous sequencedifference (Figure 1) occurs in the loop just C-term-inal to helix I (GI residues 56–61, TALKER; GIIresidues 190–195, PQLNDG). This corresponds tothe region of GH5 that contains two of the fourbasic residues in the “secondary” DNA-bindingsite, site II. GII, in contrast to GI, lacks basicresidues at the two corresponding positions, andhas a proline residue at the C-terminal end ofhelix I that is not present in GI or GH5. Thesesequence differences may contribute to thedifferent pitch of the b-strands in GII comparedwith GI, and could have implications for folding.Introduction of the two basic residues into the GIIloop was not sufficient to confer on GII the ability

to protect chromatosome-length DNA fromdigestion (S. Cooper & J.O.T., unpublished results).Surprisingly, the two additional positive chargesslightly reduced the structural stability of GII (asassessed by the ionic concentration needed torestore a-helix, judged by CD), despite the factthat the mutations were in an exposed surface loop.This may indicate that a key contributor to theinstability of GII in the absence of counterions maybe a particularly unfavorable distribution of surfacepositive charge (hence the need for large anions toallow folding), which is made worse in the mutantby the addition of two further positive charges.

Overall, the lack of hydrophobic contacts (in theloop between helices II and III), like-chargerepulsions (between Lys207 and Lys215), and anunfavorable surface charge distribution may allcontribute to the instability of the isolated GIIdomain in vitro.23 If GII is indeed folded in Hho1pbound to chromatin, the stability due to DNAbinding presumably compensates for these desta-bilizing features, which are not present in GI.

Comparison of GI and GII with the globulardomains of H1 and H5

Comparison of the GI and GII structures withthe other previously determined H1 and H5 globu-lar domain structures, GH1 and GH5 (PDB codes1GHC and 1HST, respectively) (Figure 6), showsthat the similarity is greatest in the helical regionsof the winged-helix fold, with most variationoccurring in the loop regions, including “wing 1”where the number of residues varies. In GI andGII the extra residues in the loop between helicesII and III, as compared to GH5 and GH1, areaccommodated by a helical turn. Over the regionsof common secondary structure, GI (of Hho1p)and GH5 are the most similar pair, with a back-bone r.m.s.d. of 1.1 A; the values for GII/GI andGII/GH5 are 1.3 A and 1.5 A, respectively,although the GII structure is of lower quality thanthat of GI and GH5 because of sample deterio-ration (see above). The a-helical regions of GII are

Figure 7. Ribbon diagrams of GH5, GI and GII structures with basic site I (red) and site II (blue) DNA-binding resi-dues of GH5 (see the text) and the corresponding residues in GI and GII highlighted. The diagrams were generatedusing MOLMOL.39

144 Structures of the GI and GII Domains of Hho1p

more similar to those of GH5 than GI. However,the b-sheet of GII has a different pitch from that ofGH5 as well as that of GI, as discussed above. TheGH1 structure is about 2.5 A r.m.s.d. from theother structures, but is not sufficiently well defined(backbone resolution 2.51 A r.m.s.d.) to allow pre-cise comparison with the other globular domainstructures.

Surface charge distribution and DNA binding

Basic residues in GI and GII corresponding tothe seven basic residues in site I (three residues)and site II (four residues) of GH5 are highlightedin the sequence alignment shown in Figure 1.These residues occupy essentially the samepositions relative to the winged-helix fold as inGH5 (Figure 7). In both GI and GII there is nobasic residue corresponding to Lys69 in site I ofGH5, and indeed a basic residue at this position isnot conserved across species. However, basicresidues are present at the other two positions,corresponding to site I residues Arg73 and Lys85in GH5. The electrostatic surface potential maps

illustrated in Figure 8 show that GI, GII and GH5all have an extensive tract of positive charge overtheir “front” aspects, which includes the site Iresidues. Site I is therefore expected to interactnormally with DNA, in the same way as site I inGH5. The “back” aspects of the structures(Figure 8) differ more than the front aspects. Atthe back, GII and GH5 both have significant posi-tive charge, albeit in different areas, whereas GIhas a negative charge.

Significant structural differences between GI andGII occur in the region corresponding to site II inGH5, which contains four basic residues. GI alsohas four basic residues and is structurally similarto GH5 in this region. In contrast, GII is lackingbasic residues at two of the four positions (residues193 and 195), which affects the charge of the loopin this region, which also contains an acidic resi-due, Asp194 (Figure 8). GII differs from both GIand GH5 in lacking a positively charged bulge (asfound at Arg61 in site II of GI), and in its placethere is a flatter region of b-sheet from whichAsp194 protrudes (Figure 8). The correspondingresidue in GI, Glu60, lies under the positively

Figure 8. Electrostatic surfacepotential plots for the front andback aspects of GI, GII and GH5.Regions of positive charge are indi-cated in blue and those of negativecharge in red. For GI and GII,only the structured cores (residues47–117 and 181–251, respectively)were used. For the significance ofthe labelled residues, see the text.The diagrams were generatedusing MOLMOL.39

Structures of the GI and GII Domains of Hho1p 145

charged site II residues (Lys59 and Arg61), whichare pushed up by the twist of the b-sheet (apparentin Figure 7). The corresponding residue in GH5(residue 41) is not acidic (Figure 1); another acidicresidue nearby, Glu39, does not protrude and theoverall distribution of charge in this region ismore similar to that in GI than GII. If GII is foldedin Hho1p bound to chromatin in the same awayas it is in the presence of high concentrations ofstabilizing anions, then the differences arisingfrom the nature and orientation of residues in theputative site II are likely to contribute to the dif-ferences in the chromatosome protection propertiesof GII compared with GI and GH5.12

Concluding Remarks

We have shown that the isolated GI and GIIdomains of the yeast linker histone Hho1p adoptbroadly similar winged-helix structural folds inthe presence of relatively high concentrations oftetrahedral anions (phosphate, sulphate, perchlor-ate). The folds are closely similar to the globulardomains of the canonical histone H1 of highereukaryotes and its variants, such as H5. This struc-tural similarity between GI and GII reflects the 43%sequence identity2 between the two domains and isnot surprising. However, the difference in thestructural stability between the two domains12 wasunexpected. GII, unlike GI or the globular domainsof histones H1 and H5, is largely unstructured atlow concentrations of tetrahedral anions (e.g.10 mM sodium phosphate) and at even high con-centrations (e.g. 0.5 M) of sodium chloride, mostprobably due to ineffective screening of the surfacepositive charge.12 GII is thus also likely to besignificantly unfolded at “physiological” ionicstrength (generally taken to correspond to,150 mM monovalent cations) and to be an“intrinsically unstructured” or “natively unfolded”domain.18,23,24 It is unfolded as an isolated domainand evidently also within Hho1p12 where it istethered to the folded homologous GI domain. Inview of the effects of large anions on the structure,it seems likely that GII will fold upon interactionwith the DNA phosphate groups in chromatin,possibly after recruitment by the folded GIdomain, which probably binds asymmetrically inthe vicinity of the nucleosome dyad, like GH5, con-tacting two duplexes.10 Such folding on binding tochromatin is difficult to demonstrate directly butis under investigation, as is the issue of whetherGI and GII bind to adjacent nucleosomes or withinthe same nucleosome.

The high-resolution structures have given cluesto possible causes of the instability of GII relativeto GI. Both a lack of critical hydrophobic contactsand an unstable distribution of the high surfacepositive charge might be contributing factors, assuggested for a number of other proteins that are“natively unfolded”.23 In the case of GII, appro-priate negatively charged ligands (high concen-

trations of large anions in vitro, DNA phosphategroups in vivo) appear to tip the balance by offset-ting the large surface positive charge, allowingfolding to occur; specific ion binding may alsocontribute. The biological advantage (if any) of anunfolded but foldable domain in Hho1p is unclear.One possibility is that one of its functions is to bindother protein partners, which might “select” differ-ent conformations. The advantage of versatilitycould offset the unfavorable loss of entropy thataccompanies folding.18 This and other possibilitiesare under investigation.

While this manuscript was under review,another NMR structure of the GI domain ofHho1p was reported25 (PDB code 1UHM). Thefragment (residues 41–118) was shorter than thatused here (residues 38–130) but the two GI struc-tures are very similar, the largest differences occur-ring in the loop regions. The GI structure presentedhere was calculated with more NOE distance con-straints (2771 versus 1481) and is thus definedmore precisely.

Materials and Methods

Recombinant GI and GII

The recombinant “globular” domains of Hho1p, GI(defined by tryptic digestion as residues 38–130), andGII (residues 171–258), were expressed in E. coliBL21(DE3) cells and purified by cation-exchangechromatography on a SP Sepharose Fast Flow FPLCcolumn (Pharmacia), hydrophobic chromatography onPhenylSepharose (Pharmacia) and cation-exchangechromatography on a MonoS (Pharmacia) column.12 Toproduce uniformly 15N (.95%) and 13C (.95%) labelledGI and GII, growth and expression were carried out in6 mM 15NH4Cl and 0.5 g l21 of [13C]glucose in Mopsminimal medium.26 The proteins were concentrated andbuffer-exchanged into 100 mM sodium phosphate (pH7.0), 1 mM EDTA (for GI) or 250 mM sodium phosphate(pH 7.0), 1 mM EDTA, 1 mM DTT (for GII) using aCentricon concentrator (Amicon; 5 kDa cut-off).

NMR sample preparation

NMR measurements for GI were made on samplescontaining 2.7 mM protein, 90% H2O/10% 2H2O in100 mM sodium phosphate (pH 7.0), 1 mM EDTA, andfor GII on samples containing 1 mM protein, 90% H2O/10% 2H2O in 250 mM sodium phosphate (pH 7.0), 1 mMDTT and 1 mM EDTA.

Data collection, processing and assignment

All NMR experiments were carried out at 288 K on aBruker DRX500 spectrometer equipped with a triple-resonance HCN probe head and actively-shieldedz-gradients. HN, N, Ca, Cb and C0 assignments werederived from standard versions of HN(CO)CA, HNCA,CBCA(CO)NH, HNCACB and HNCO experiments. Theremaining backbone and side-chain resonances wereassigned using 3D (H)CC(CO)NH-total correlation spec-troscopy (TOCSY), 15N-HSQC-TOCSY, NOESY-15N-HSQC,HC(C)H-TOCSY and NOESY-13C-HSQC experiments.27

146 Structures of the GI and GII Domains of Hho1p

HN exchange rates were assessed qualitatively from aCLEANEX-PM experiment.28 For GI, 3JHNHa scalarcoupling constants were derived from a 3D 15N-HNHAexperiment.29 R1 and R2 relaxation rates and {1H}15Nheteronuclear NOE values were measured for GI at500 MHz30 and analyzed using ModelFree v4.15.31,32

Sample instability precluded the acquisition of high-quality relaxation data for GII. In all experiments, watersuppression was achieved using established flip-backmethods, using shaped selective pulses to return watermagnetization to the z-axis, followed by WATERGATEto purge any remaining transverse magnetization priorto acquisition.33 States/TPPI was used for quadraturedetection in all indirect dimensions. Data were processedusing the AZARA suite of programs (q1993–2002Wayne Boucher and Department of Biochemistry, Uni-versity of Cambridge, unpublished) and assignmentswere made using ANSIG v3.3.34

Structure calculations

For the non-prolyl residues in the GI domain, back-bone f torsion-angle restraints were derived from 3JHNHa

scalar coupling constants.29 3JHNHa scalar coupling con-stants were not available for the GII domain due to bothsample instability and the presence of peaks arisingfrom the unfolded component (,5%; see Results andDiscussion), making the measurement of peak volumesunreliable. For the GII domain, backbone f and ctorsion-angle restraints were derived using TALOS.35

Distance restraints were derived from the heightsof crosspeaks assigned in the NOESY-15N-HSQC andNOESY-13C-HSQC experiments. These, together with tor-sion angle restraints, provided the input for the iterativeassignment protocol ARIA.36 For the GI domain, 1955unambiguous and 816 ambiguous NOE restraints wereused, and for GII 1058 unambiguous and 451 ambiguousNOE restraints.

For both structures, the ARIA protocol was performedwith eight iterations. For each iteration, ambiguity in thedistance restraints was reduced in successive iterations,based upon the ten lowest-energy structures from atotal of 30 calculated structures. For the final iteration,100 structures were calculated with the violation toler-ance set at 0.1. After the last iteration, the 25 lowest-energy structures were subjected to a final water refine-ment to give an ensemble of structures. Torsion-anglemolecular dynamics simulations for the refinement ofstructures were performed using CNS.37 The simulatedannealing protocol to calculate each structure used38,000 molecular dynamics steps, including those forrefinement, and two cooling stages (to 1000 K and 50 K).

Data Bank accession numbers

Chemical shift assignments have been deposited withBioMagResBank; the BMRB accession numbers for GIand GII are 10118 and 10120, respectively. The coordi-nates of the NMR structure ensembles have beendeposited in the RCSB Protein Data Bank under thePBD accession codes 1UST (for GI) and 1USS (GII).

Acknowledgements

This work was supported by the Biotechnology

and Biological Sciences Research Council (BBSRC)(grant to J.O.T.) T.A. thanks the EPSRC for aStudentship and P.C. thanks the Winston ChurchillFoundation for a Scholarship. The CambridgeCentre for Molecular Recognition was supportedby the BBSRC and The Wellcome Trust.

References

1. Ushinsky, S. C., Bussey, H., Ahmed, A. A., Wang, Y.,Friesen, J., Williams, B. A. & Storms, R. K. (1997).Histone H1 in Saccharomyces cerevisiae. Yeast, 13,151–161.

2. Landsman, D. (1996). Histone H1 in Saccharomycescerevisiae—a double mystery solved. Trends Biochem.Sci. 21, 287–288.

3. Clark, K. L., Halay, E. D., Lai, E. & Burley, S. K.(1993). Co-crystal structure of the HNF-3g fork headDNA-recognition motif resembles histone H5.Nature, 364, 412–420.

4. Cerf, C., Lippens, G., Ramakrishnan, V., Muyldermans,S., Segers, A., Wyns, L. et al. (1994). Homo- andheteronuclear two-dimensional NMR studies of theglobular domain of histone H1: full assignment, ter-tiary structure, and comparison with the globulardomain of H5. Biochemistry, 33, 11079–11086.

5. Ramakrishnan, V., Finch, J. T., Graziano, V., Lee, P. L.& Sweet, R. M. (1993). Crystal structure of globulardomain of histone H5 and its implications fornucleosome binding. Nature, 362, 219–223.

6. Allan, J., Hartman, P. G., Crane-Robinson, C. &Aviles, F. X. (1980). The structure of histone H1 andits location in chromatin. Nature, 288, 675–679.

7. Simpson, R. T. (1978). Structure of the chromato-some, a chromatin particle containing 160 base pairsof DNA and all the histones. Biochemistry, 17,5524–5531.

8. Patterton, H. G., Landel, C. C., Landsman, D.,Peterson, C. L. & Simpson, R. T. (1998). The bio-chemical and phenotypic characterization of Hho1p,the putative linker histone H1 of Saccharomycescerevisiae. J. Biol. Chem. 273, 7268–7276.

9. Bates, D. L. & Thomas, J. O. (1981). Histone H1 andH5: one or two molecules per nucleosome? Nucl.Acids Res. 9, 5883–5893.

10. Zhou, Y.-B., Gerchmann, S. E., Ramakrishnan, V.,Travers, A. & Muyldermans, S. (1998). Position andorientation of the globular domain of linker histoneH5 on the nucleosome. Nature, 395, 402–405.

11. Nelson, P. P., Albright, S. C., Wiseman, J. M. &Garrard, W. T. (1979). Reassociation of histone H1with nucleosomes. J. Biol. Chem. 254, 11751–11760.

12. Ali, T. & Thomas, J. O. (2004). Distinct properties ofthe two putative “globular domains” of the yeastlinker histone Hho1p. J. Mol. Biol. In press.

13. Goytisolo, F. A., Gerchman, S. E., Yu, X., Rees, C.,Graziano, V., Ramakrishnan, V. & Thomas, J. O.(1996). Identification of two DNA-binding sites onthe globular domain of histone H5. EMBO J. 15,3421–3429.

14. Thomas, J. O. (1999). Histone H1: location and role.Curr. Opin. Cell Biol. 11, 312–317.

15. De Petrocellis, L., Quagliarotti, G., Tomei, L. &Geraci, G. (1986). Structuring of H1 histone. Evi-dence of high affinity binding sites for phosphateions. Eur. J. Biochem. 156, 143–148.

16. Clark, D. J., Hill, C. S., Martin, S. R. & Thomas, J. O.

Structures of the GI and GII Domains of Hho1p 147

(1988). a-Helix in the C-terminal domain of histoneH1. EMBO J. 7, 69–75.

17. Hill, C. S., Martin, S. R. & Thomas, J. O. (1989). Astable a-helical element in the carboxy-terminaldomain of free and chromatin-bound histone H1from sea urchin sperm. EMBO J. 8, 2591–2599.

18. Dyson, H. J. & Wright, P. E. (2002). Coupling of fold-ing and binding for unstructured proteins. Curr.Opin. Struct. Biol. 12, 54–60.

19. Lipari, G. & Szabo, A. (1982). Model-free approachto the interpretation of nuclear magnetic resonancerelaxation in macromolecules. 1. Theory and rangeof validity. J. Am. Chem. Soc. 104, 4556–4559.

20. Lipari, G. & Szabo, A. (1982). Model-free approach tothe interpretation of nuclear magnetic resonancerelaxation in macromolecules. 2. Analysis of experi-mental results. J. Am. Chem. Soc. 104, 4559–4570.

21. Kay, L. E., Torchia, D. A. & Bax, A. (1989). Backbonedynamics as studied by 15N inverse detected hetero-nuclear NMR spectroscopy: application to staphylo-coccal nuclease. Biochemistry, 28, 8972–8979.

22. Engelman, D. M., Steitz, T. A. & Goldman, A. (1986).Identifying nonpolar transbilayer helices in aminoacid sequences of membrane proteins. Annu. Rev.Biophys. Biophys. Chem. 5, 321–353.

23. Uversky, V. N., Gillespie, J. R. & Fink, A. L. (2000).Why are “natively unfolded” proteins unstructuresunder physiologic conditions? Proteins: Struct. Funct.Genet. 41, 415–427.

24. Tsai, C.-J., Ma, B., Sham, Y. Y., Kumar, S. & Nussinov,R. (2002). Structured disorder and conformationalselection. Proteins: Struct. Funct. Genet. 44, 418–427.

25. Ono, K., Kusano, O., Shimotakahara, S., Shimizu, M.,Yamazaki, T. & Shindo, H. (2003). The linker histonehomolog Hho1p from Saccharomyces cerevisiae repre-sents a winged-helix-turn-helix fold as determinedby NMR spectroscopy. Nucl. Acids Res. 312,7199–7207.

26. Neidhardt, F. C., Bloch, P. L. & Smith, D. F. (1974).Culture medium for enterobacteria. J. Bacteriol. 119,736–747.

27. Cavanagh, J., Fairbrother, W. J., Palmer, A. G. &Skelton, N. J. (1996). Protein NMR Spectroscopy: Prin-ciples and Practice, Academic Press, San Diego, CA.

28. Hwang, T. L., van Zijl, P. C. M. & Mori, S. (1998).Accurate quantitation of water–amide protonexchange rates using the phase-modulated CLEANchemical EXchange (CLEANEX-PM) approach witha fast-HSQC (FHSQC) detection scheme. J. Biomol.NMR, 11, 221–226.

29. Vuister, G. W. & Bax, A. (1993). Quantitative J corre-lation—a new approach for measuring homonuclear3-bond JH(N)H(a) coupling-constants in 15N-enrichedproteins. J. Am. Chem. Soc. 115, 7772–7777.

30. Farrow, N. A., Muhandiram, R., Singer, A. U., Pascal,S. M., Kay, C. M., Gish, G. et al. (1994). Backbonedynamics of a free and a phosphopeptide-complexedSRC homology-2 domain studied by 15N NMR relax-ation. Biochemistry, 33, 5984–6003.

31. Palmer, A. G., Rance, M. & Wright, P. E. (1991). Intra-molecular motions of a zinc finger DNA-bindingdomain from Xfin characterized by proton-detectednatural abundance carbon-13 heteronuclear NMRspectroscopy. J. Am. Chem. Soc. 113, 4371–4380.

32. Mandel, A. M., Akke, M. & Palmer, A. G. (1995).Backbone dynamics of Escherichia coli RibonucleaseH1: correlations with structure and function in anactive enzyme. J. Mol. Biol. 246, 144–163.

33. Sklenar, V., Piotto, M., Leppik, R. & Saudek, V. (1993).Gradient-tailored water suppression for 1H 15NHSQC experiments optimized to retain full sensi-tivity. J. Magn. Reson. ser. A, 102, 241–245.

34. Kraulis, P. J. (1989). ANSIG—a program for theassignment of protein 1H 2D NMR spectra by inter-active computer graphics. J. Magn. Reson. 84,627–633.

35. Cornilescu, G., Delaglio, F. & Bax, A. (1999). Proteinbackbone angle restraints from searching a databasefor chemical shift and sequence homology. J. Biomol.NMR, 13, 289–302.

36. Linge, J. P., Habeck, M., Rieping, W. & Nilges, M.(2003). ARIA: automated NOE assignment andNMR structure calculation. Bioinformatics, 19,315–316.

37. Brunger, A. T., Adams, P. D., Clore, G. M., DeLano,W. L., Gros, P., Grosse-Kunstleve, R. W. et al. (1998).Crystallography and NMR system: a new softwaresuite for macromolecular structure determination.Acta Crystallog. sect. D, 54, 905–921.

38. Laskowski, R. A., MacArthur, M. W., Moss, D. S. &Thornton, J. M. (1993). PROCHECK: a program tocheck the stereochemical quality of protein struc-tures. J. Appl. Crystallog. 26, 283–291.

39. Koradi, R., Billeter, M. & Wuthrich, K. (1996). MOL-MOL: a program for display and analysis of macro-molecular structures. J. Mol. Graph. 14, 51–55; seealso pp. 29–32.

Edited by P. Wright

(Received 5 December 2003; received in revised form10 February 2004; accepted 13 February 2004)

Supplementary Material comprising one Figureshowing a secondary structure overview of GI andGII, and one Table showing a comparison of ther.m.s.d. values for the structures of GI and GII,and GH1 and GH5 is available on Science Direct

148 Structures of the GI and GII Domains of Hho1p