The oxidised histone octamer does not form a H3 disulphide bond

Biochimica et Biophysica Acta 1764 (2006) 1356 – 1362 www.elsevier.com/locate/bbapap

The oxidised histone octamer does not form a H3 disulphide bond Christopher M. Wood a,⁎, Sirirath Sodngam a , James M. Nicholson b , Stanley J. Lambert a , Colin D. Reynolds a , John P. Baldwin b a

School of Biomolecular Sciences, Liverpool John Moores University, Byrom Street, Liverpool, L3 3AF, UK College of Biology and Medicine, CCLRC Daresbury Laboratory, Warrington, Cheshire, WA4 4AD, UK

b

Received 5 March 2006; received in revised form 12 June 2006; accepted 13 June 2006 Available online 21 July 2006

Abstract A H3 dimer band is produced when purified native histone octamers are run on an SDS-PAGE gel in a β-mercaptoethanol-free environment. To investigate this, native histone octamer crystals, derived from chicken erythrocytes, and of structure (H2A–H2B)–(H4–H3)– (H3′–H4′)–(H2B′–H2A′), were grown in 2 M KCl, 1.35 M potassium phosphates and 250–350 μM of the oxidising agent S-nitrosoglutathione, pH 6.9. X-ray diffraction data were acquired to 2.10 Å resolution, yielding a structure with an Rwork value of 18.6% and an Rfree of 22.5%. The space group is P65, the asymmetric unit of which contains one complete octamer. Compared to the 1.90 Å resolution, unoxidised native histone octamer structure, the crystals show a reduction of 2.5% in the c-axis of the unit cell, and freeenergy calculations reveal that the H3–H3′ dimer interface in the latter has become thermodynamically stable, in contrast to the former. Although the inter-sulphur distance of the two H3 cysteines in the oxidised native histone octamer has reduced to 6 Å from the 7 Å of the unoxidised form, analysis of the hydrogen bonds that constitute the (H4–H3)–(H3′–H4′) tetramer indicates that the formation of a disulphide bond in the H3–H3′ dimer interface is incompatible with stable tetramer formation. The biochemical and biophysical evidence, taken as a whole, is indicative of crystals that have a stable H3–H3′ dimer interface, possibly extending to the interface within an isolated H3–H3′ dimer, observed in SDS-PAGE gels. © 2006 Elsevier B.V. All rights reserved. Keywords: H3 dimer; Histone octamer; Hydrogen bond modulation; Oxidation; Reduction

1. Introduction The histone octamer (HO) forms the foundation upon which is built chromatin, the hierarchical DNA structure of eukaryotic-cell nuclei. Around the HO are wrapped 147 base-pairs (bp) of superhelical DNA, that are organised into 1.65 turns to form the nucleosome-core particle (NCP) [1–3]. Attached to the NCP is a species-dependent length of DNA, of Abbreviations: CCLRC, Council for the Central Laboratory of the Research Councils; HO, histone octamer; GHO, GSNO-exposed histone octamer—the 2.10 Å histone octamer model (PDB code 2ARO) described herein; GSNO, S-nitrosoglutathione; NCP, nucleosome-core particle; OD278, UV spectrographic measurement at 278 nm; PDB, protein database; RSH, any thiol; RSNO, any S-nitrosothiol; SM, starting model—the 1.90 Å histone octamer model (PDB code 1TZY) ⁎ Corresponding author. Tel.: +44 151 231 2565; fax: +44 151 298 2624. E-mail address: [email protected] (C.M. Wood). 1570-9639/$ - see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.bbapap.2006.06.014

the order of 50 bp [3], which is associated with H1 linker histone. Together, the H1 linker histone and the NCP form the nucleosome, and it is this unit that is repeated to form the continuous structure that is chromatin. The HO itself is comprised of a set of eight proteins known as core histones, there being two each of H2A, H2B, H3 and H4. The eight histones are arranged as a central tetramer, of form (H4–H3)– (H3′–H4′), adjoined on each side by a (H2A–H2B) dimer, that results in the octamer structure of (H2A–H2B)–(H4–H3)– (H3′–H4′)–(H2B′–H2A′). Each histone consists of three αhelices: α1, α2 and α3, interconnected by loops L1 and L2, respectively [4], which also serve to bind the DNA of the NCP to the HO. These elements form a structure known as the histone fold, pairs of which interlock to form histone-fold pairs. The integrity of both these structures is maintained by numerous hydrogen bonds and hydrophobic interactions. The histone-fold pair is the mechanism by which the (H2A–H2B)

C.M. Wood et al. / Biochimica et Biophysica Acta 1764 (2006) 1356–1362

and (H3–H4) dimers are held together. Finally, the four dimers are held together as the HO by three four-helix bundle interactions (Fig. 2 of reference [5]). Whilst these four-helix bundles are fundamental to the HO structure, there are also other key binding regions, the importance of which has been discussed elsewhere [6,7]. When purified HOs were run on a non-reducing SDS-PAGE gel (Fig. 1), a band corresponding to H3 dimers (twice the molecular weight of a H3 monomer) was observed, and since each H3 contains one cysteine, those dimers could be stabilised by a disulphide bridge. On reducing gels, only H3 monomers were observed. Neither the HO nor the NCP have a disulphide bond between H3 C110 and H3′ C110, and the role of these residues in the histone octamer has long been the subject of discussion [8]. Therefore it was decided to crystallise HOs in the presence of the oxidising agent S-nitrosoglutathione (GSNO) to see if such a bond could be induced and, if successful, to assess its functional significance. GSNO was chosen for its ability to induce disulphide bond formation by S-nitrosylation or S-transnitrosylation. We present herein results that clearly show that, although no disulphide bond is formed, GSNO nevertheless induces changes in the HO (GHO) that involve a reduction in the c-axis of the GHO, brought about by alterations to the hydrogen bond infrastructure. Together, these factors have created in the GHO a stable H3–H3′ dimer, as part of the tetramer.

Fig. 1. In non-reducing SDS-PAGE gels, H3 runs as a dimer. Lane B shows purified histones running under reducing conditions, in the presence of beta mercaptoethanol. It can be seen that the normal distribution for the four core histones is obtained, with H3 in the heaviest position and H4 in the lightest. However, when the purified histones are run under non-reducing conditions, as in lane A, the H3 monomer band almost disappears, and a new band is created in a position that corresponds to a molecule having twice the molecular mass of a H3 monomer. Since both H3 histones in the SM and GHO contain a single cysteine, it is quite likely the new band in lane A comprises of H3 dimers, possibly with a disulphide bond interconnecting the monomers.

1357

2. Materials and methods The purification and preparation of chicken HOs utilised the same process used in earlier studies [5,7], however the crystallisation conditions were modified with GSNO. Chicken HOs were initially prepared in 2 M KCl, 0.2 M K2HPO4, 0.2 M KH2PO4 and then dialysed against 2 M KCl, 0.475 M K2HPO4, 0.475 M KH2PO4. Dialysis was carried out overnight at 277 K, with two changes of buffer and centrifuged at 13,000×g for 1 h. The supernatant containing pure HOs was concentrated to 20 mg ml− 1 (OD278 = 9). Some of the supernatant (Sample A) was taken to run a SDS-PAGE gel, while the remainder (Sample B) was used for crystallisation experiments. Sample A was dialysed against distilled water for 24 h, with four changes of solution. The pure, salt-free Sample A was then run in two lanes on a SDS-PAGE gel. The first lane (lane A of Fig. 1) was run under non-reducing conditions (without beta mercaptoethanol), while the second lane (lane B of Fig. 1) was run as normal, under reducing conditions (with beta mercaptoethanol). Sample B was partitioned into 50 μl crystallisation cells, which were divided into three groups and then each group was dialysed against 10 ml of 2 M KCl, 0.675 M K2HPO4, 0.675 M KH2PO4, GSNO to crystallise. The three values of GSNO concentration were 250 μM, 350 μM and 750 μM. These levels of GSNO concentration were chosen as they are the levels that have been determined as necessary to stimulate enzyme activation by S-nitrosylation (Fig. 3A of reference [9]). Below about 150 μM GSNO, there is a sudden drop in enzyme activity, but above 200 μM, up to 1.0 mM GSNO, there is linear growth in activity [9,10]. To give the best conditions for crystal growth, anhydrous KCl and phosphates were used in the buffers. A range of crystallisation-buffer pH values were prepared by changing the ratios of monobasic to dibasic phosphate ions by small variations around the two 0.675 M values, keeping the total phosphate at 1.35 M. This gave a range of samples covering a pH window centred on 6.9. After a few weeks, 2–3 mm long crystal needles were found in the buffer conditions containing 350 μM and 250 μM GSNO, but not in the 750 μM GSNO solution. The resulting GHO crystals were hexagonal, space group P65 [5,11], with a unit cell of a = b = 158.08, c = 101.04 Å, with one octamer in the asymmetric unit [Protein Database Code (PDB) code: 2ARO]. One day prior to data collection, the crystal-containing cells were removed from the dialysis solutions and placed in a new dialysis solution of 2 M KCl, 0.675 M K2HPO4, 0.675 M KH2PO4, 20% glycerol, to dialyse overnight. Prior to data collection, the crystals were removed from the dialysis cells on Hampton cryo-loops and then placed into liquid nitrogen and finally placed in the 100 K cryo-jet of the X-ray diffractometer for the data acquisition. Data were collected on beamline 10.1 of the CCLRC (Council for the Central Laboratory of the Research Councils) Daresbury Laboratory Synchrotron Radiation Source using a MAR DTB and 225 CCD detector. Initially MOSFLM [12] was used to process the data, and later the Collaborative Computational Project, Number 4 (CCP4) crystallographic package [13], specifically SCALA, yielding an overall merging-R value of 8.0% and 36.9% in the 2.21–2.10 Å highest-resolution shell. In all, 81294 unique reflections were recorded, corresponding to a completeness of 97.2% overall and 97.1% in the highest-resolution shell, with a multiplicity of 3.5 (3.4 in the highest-resolution shell) and a mosaicity of 0.37°. The average I/σ(I) was 14.4 (2.3 in the highestresolution shell). The structure was solved by using the 1.90 Å model of the HO (PDB code 1TZY) as a search model [7]. The GHO data were collected as an independent data set, with none of the 1.90 Å model structure factors being used. The model development was carried out using the AMoRe program [13,14] and refinement used the program REFMAC5 [13,15]. Model building and refinement initially involved just the residues of the search model. Rigid-body refinement for five cycles resulted in a value for Rwork of 27.6% and for Rfree of 27.5%. Five cycles of translation/libration/screw (TLS) refinement, followed by five cycles of restrained refinement brought these values down to 23.1% for Rwork and 24.9% for Rfree. The structure was then observed in Coot [16] on a residue-by-residue basis using 2Fo–Fc and Fo–Fc maps. Positions of missing phosphate and chloride ions were clearly visible, and these were manually inserted into the appropriate positions. After further refinement, waters were added using the ARP/wARP program [17]. Waters were added in cycles of 25 at a time, with a refinement at the end of each cycle-27 cycles in total. A water was kept if above 3.2σ of the Fo–Fc map and rejected if below 1σ of the 2Fo–Fc map. A total of 494 waters were included

1358

C.M. Wood et al. / Biochimica et Biophysica Acta 1764 (2006) 1356–1362

in the final model, which resulted in a value for Rwork of 18.2% and for Rfree of 22.1%. A final round of refinement was carried out, generating final values of 18.6 for Rwork and 22.5 for Rfree. The solvent content of the final model is 69.18%, giving a Matthew's coefficient of 4.02 Å3 Da− 1. The majority of residues either fell into the most-favoured (94.8%) or the allowed (5.1%) areas of the Ramachandran plot [18]. Table 1 gives a summary of the data collection and final refinement statistics.

3. Results 3.1. H3 runs as a dimer on non-reducing SDS-PAGE gels Sample A was dialysed against distilled water for 24 h and then run in two lanes of a SDS-PAGE gel. One lane was run under non-reducing conditions, the other under reducing conditions (Fig. 1). In lane B of Fig. 1, it can be seen that under reducing conditions (with beta mercaptoethanol), there is the expected distribution of histone protein bands, with H3 running at the highest molecular mass and H4 at the lightest. All the histones in lane B are running as monomers. Lane A is run under non-reducing conditions, without beta mercaptoethanol. In the H3 monomer position in lane A it can be seen that the band

3.2. GSNO induces changes in the GHO structure

Table 1 Data collection and refinement statistics Crystal dimensions (mm) X-ray source Detector Wavelength (Å) Temperature (K) Rotation range (°) Rotation step (°) Space group Unit-cell parameters (Å) Resolution (Å) Mosaicity (°) Total reflections Unique reflections Merging R Factor (%) Completeness (%) Mean I/σ(I) Multiplicity Rwork (Rfree) (%) No. of protein atoms No. of solvent molecules No. of heteroatoms R.m.s. deviations from ideal geometry Bonds (Å) Angles (°) H bonds (Å) Torsion angles (°) Estimated coordinate error Maximum likelihood (Å) Free R value Ramachandran plot Most favoured (%) Allowed (%) Average B factors (Å2) Overall protein atoms Main-chain protein Side-chain protein Water Phosphates Chlorides

intensity is reduced to almost zero. There is still some sign of H3 monomers, but there are very few and they are probably monomers that have become degraded. The top band of lane A resides at a position that is exactly twice the molecular mass of H3. It is postulated that this is a stable H3–H3′ dimer formed by a disulphide bond. The reason why a disulphide bond is postulated is that both the H3 histones contain a cysteine amino-acid residue at position 110. Since lane A is run under non-reducing conditions, the disulphide bond assumption is a reasonable one. However, to be of physiological significance, it would be necessary to find either a disulphide bond in the histone octamer, or that conditions had become more favourable for disulphide bond formation. Since it is known that neither the 1.90 Å histone octamer (SM), nor the NCP have a disulphide bond, it was necessary to determine whether another conformational state of the HO would contain such a bond. For this reason it was decided to attempt the crystallisation of histone octamers exposed to the oxidising agent GSNO. Crystals were successfully grown at GSNO concentrations of 250 μM and 350 μM, but not at 750 μM.

1.5 × 0.25 × 0.25 Station 10.1, SRS Daresbury UK MAR 225 DTB/CCD 0.98 100 75 0.25 P65 a = b = 158.08, c = 101.04 10.0–2.10 0.37 276,338 (40,534) 81,294 (11,801) 8.0 (36.9) 97.2 (97.1) 14.4 (2.3) 3.5 (3.4) 18.4 (22.5) 5965 458 48 0.02 1.67 0.19 6.03 0.10 0.14 94.8 5.1 49.76 48.16 54.49 46.43 81.68 50.86

To establish what, if any, changes were apparent in the GHO, the SM was aligned to the GHO using LSQKAB [19] and used as the reference set of coordinates. On observing the accessibility of surface amino-acids to solvent for the two models in Deepview [20], it was apparent that many residues in the GHO had become less solvent accessible. In previous work we identified crystalpacking surfaces of the HO (Table 2 and Fig. 3 of reference [5]). There are four regions: two acidic and two basic that interact to form two crystal-packing interfaces (designated A–B and C–D in reference [5]), one large and one small. In that work it was postulated that the larger interface in particular may be involved in higher-order chromatin function, and so these regions were examined to see if there were changes in the solvent accessibility of their amino-acid residues. In the SM the A–B region had six solvent-accessible amino-acid residues, and the C–D region fifteen. In the GHO, these had fallen to just one for the A–B region and to eight for the C–D region. This reduction in the solvent accessibility is due to the fact that the c-axis parameter of the GHO was 2.5% shorter than in the SM, which gives rise to a unit cell volume reduction of 3.2% compared to the SM. Since the GHO unit cell dimensions and volume had changed, it was likely the volume of the GHO had a concomitant reduction. Indeed, it was found that the GHO volume had reduced to 29742 Å3 compared to the 30073 Å3 of the SM. Oxidation of a protein can result in the formation of disulphide bonds between cysteines or the conversion of methionines into methionine sulphoxide [21]. Upon examination of the 2Fo–Fc electron density map for the GHO, the sulphur of C110 in H3 and the sulphur of C110 in H3′ were found to be only slightly closer (6 Å) than in the SM (7 Å). There was also no connecting electron density in the Fo–Fc map (data not shown). To double check, an omit map was generated by removing amino-acid residues 109–111 of both H3 and H3′,

C.M. Wood et al. / Biochimica et Biophysica Acta 1764 (2006) 1356–1362 Table 2 Comparison of the number of hydrogen bonds in the SM and GHO

SM GHO SM GHO SM GHO

H2A

H2B

H3

H4

H2A′

H2B′

H3′

H4′

11 9 36 35 5 4

14 13 47 40 13 10

19 15 43 35 8 7

16 12 41 39 16 12

11 10 35 35 7 8

14 12 44 34 10 12

20 16 41 34 9 10

15 12 41 39 14 13

α1 α2 α3

The table shows the number of hydrogen bonds within the α-helices of the different histones. Notice that H3 and H2B have the largest changes between the SM and GHO. H2B and H3 are having their redox potentials modulated [29]. However, as H2A and H4 make up key binding areas in the histone octamer [7], it is necessary that their association constants remain unaltered from one conformational change to the next [29]. References to the SM occur before a set of parentheses, references to the GHO are within parentheses. In the Composition column the letters are chain labels used within the PDB models 1TZY and 2ARO. Both models use the same chain letters, defined as: A-H2A, B-H2B, C-H3, D-H4, E-H2A′, F-H2B′, GH3′ and H-H4′. Together, a group of letters refers to a structure within the SM and GHO, with, for example, CGDH referring to the (H4–H3)–(H3′–H4′) tetramer. A ‘✗’ after a group of letters indicates that the structure has dissociated in the GHO and a ‘✓’ means the structure has not dissociated in the GHO. A ′−′ anywhere in the table simply means not applicable. The Stable column indicates whether the structure is thermodynamically stable, and is determined by considering ΔGDISS. The Surface Area column is the solvent-accessible surface area of the structure. The Buried Area column is the solvent-accessible area buried upon formation of the structure. ΔGINT is the solvation free energy gain upon formation of the structure, given by ΔGS(A1, A2… A3) − ∑i 1, nΔGS(Ai) − EHBNHB − ESBNSB, where ΔGS is the solvation = energy of a protein complex, ∑ΔGS is the solvation free energy of dissociated sub-units, EHB is the free energy of hydrogen bond formation, NHB is the number of hydrogen bonds between sub-units, ESB is the free energy of salt bridge formation and NSB is the number of salt bridges between sub-units. ΔGDISS is the free energy barrier of structure dissociation. Positive values of ΔGDISS indicate that an external driving force should be applied in order to dissociate the structure and, therefore, structures with ΔGDISS > 0 are thermodynamically stable. Calculations obtained using the Pisa system [22].

and then regenerating structure factors and difference maps. However, once again, in both sets of maps there was no bridging electron density. Whilst cysteines are the most readily oxidised amino-acids, methionines are the second most easily oxidised and have been identified as being involved in a protective, anti-oxidant role by sweeping up free radicals. Examination of both the 2Fo–Fc and Fo–Fc electron density maps again revealed that there were no methionine sulphoxides in the GHO.

1359

Since the majority of the secondary structure in the GHO is in the form of α-helices, it is these elements that were next examined. The internal hydrogen-bonding arrangements of all α-helices were generated [13] for a range of 2.7 to 3.3 Å (Table 2). This analysis revealed that, whilst there were few changes in the number of hydrogen bonds associated with the α-helices of H2A, H4, H2A′ and H4′ of the GHO, as compared to the SM, there were significant changes involving H2B, H3, H2B′ and H3′. This shortfall of hydrogen bonds in the latter group of histones causes a slight destabilisation of α-helices that results in the shrinkage of the GHO's c-axis parameter and volume. Thus, the conclusion is that exposure of native histone octamers at up to 350 μM concentration of GSNO does not involve the creation of disulphide bonds, or the creation of methionine sulphoxides, but a change in the hydrogen bond distribution. 3.3. Energy calculations reveal thermodynamic stability in the GHO H3–H3′ interface region There are clear qualitative changes in the GHO as compared to the SM, those being a change in unit cell dimensions and volume, a reduction in the volume of the GHO and, thereby, a change in key areas of the molecular surface's acid–base profile and significant changes in the hydrogen-bonding pattern within certain areas of the GHO. However, to establish quantitative data from which sound conclusions can be made, energy calculations were made. It was decided to look at the energy contributions of key structures in the GHO and any interfaces therein. Since the biochemical evidence of Fig. 1 might suggest the formation of a disulphide bond between H3 and H3′, it was this interface that was considered, as part of the (H4–H3)–(H3′–H4′) tetramer structure. Structure and energy analysis was carried out using the Pisa system (a project funded by the UK Biotechnology and Biological Sciences Research Council (BBSRC) and associated with the CCP4 program suite) which uses graph theory techniques to analyse PDB models, by identifying structures therein and analysing the energy of those structures and their interfaces, to establish stability [22]. The results of this analysis are shown in Table 3. The key structure on which the analysis was carried out is the (H4–H3)–(H3′–H4′) tetramer, shown as CGDH in Table 3 (hereafter referred to as tetramer). The interface within the tetramer to which energy calculations were applied is the H3–

Table 3 Energy and area calculations for identical HO and GHO structures Composition AFGH (✗) EBCD (✓) CGDH (✓) EF (✓) AB (✓) AECG (✗) FH (✗) BD (✗)

Stable NO (−) NO (−) NO (YES) YES (YES) YES (YES) NO (−) NO (−) NO (−)

Surface Area (Å2) 23,646 (−) 22,650 (−) 19,835 (19,747 ) 11,020 (10,950) 11,368 (11,277) 29,009 (−) 13,995 (−) 13,777 (−)

Buried Area (Å2) 8633 (−) 8523 (−) 11,224 (11,256) 5039 (5073) 4965 (5058) 4370 (−) 1116 (−) 1182 (−)

ΔGINT (kcal/mol)

ΔGDISS (kcal/mol)

−59.5 (−) −59.3 (−) −91.2 (−92.4) −45.5 (−44.4) −44.0 (−42.8) −27.3 (−) −4.9 (−) −5.4 (−)

−1.4 (−) −1.9 (−) −0.4 (0.3) 43.4 (41.2) 42.5 (42.8) −1.3 (−) −1.1 (−) −1.1 (−)

1360

C.M. Wood et al. / Biochimica et Biophysica Acta 1764 (2006) 1356–1362

H3′ contact region. Compared to the SM, it can be seen that the surface area for the GHO tetramer has reduced, as would be expected, given the unit cell volume change. Since the interface region has moved closer, the buried surface area of that interface has increased. This has had the effect of changing ΔGINT, which is the solvation free energy gain upon creation of the tetramer. Significantly though, it can be seen that the tetramer component is not a stable structure for the SM, but for the GHO it becomes stable. This stability has been brought about by the positive value of ΔGDISS, which is the free energy barrier for dissociation of the H3–H3′ interface, and thus the tetramer. The positive value of ΔGDISS implies that an input of energy is required to disrupt the tetramer's interface. In other words, the tetramer in the GHO has become more thermodynamically stable, supporting the biochemical evidence that H3–H3′ association involves the formation of a stable dimer. Though we can say the interface is stable, there is insufficient data from the energy calculations to categorically state that this involves the creation of a disulphide bond, in agreement with the structural studies. Since the SM and GHO have been crystallised under the exact same conditions (save for GSNO), the energy differences are representative of the structures themselves and independent of any buffer variations. 4. Discussion The occurrence of H3 dimer bands on electrophoretic gels has been known about for a while [8], and has previously been seen by our group [11]. This observation has prompted researchers to address the question of whether a disulphide bond occurs between the two H3 C110 amino acid residues in vivo [8]. Subsequently, attempts to create nucleosomes that contain a disulphide bond have been successful [8], and, also, nucleosomes have been generated that have a H3–H3′ dimer, made by cross-linking the two H3 cysteines using dimethyl suberimidate [23]. In the former case, nucleosomes were reconstituted around a disulphide-bonded H3–H3′ dimer. The work herein seeks to build on these successes, and endeavours to establish if a disulphide bond is formed in the GHO.

GSNO was used as the oxidising agent, as it is known that this S-nitrosothiol can induce the formation of disulphide bonds. Two of the ways in which S-nitrosothiols such as GSNO achieve this bond creation are by S-nitrosylation and S-transnitrosylation. These mechanisms can be utilised in the redox signalling pathway for the activation of certain types of gene, as in the case of the mitogen-activated kinase (MAP) pathway [24]. In both S-nitrosylation and S-transnitrosylation, the thiols of cysteines are modified [24–27]. S-transnitrosylation can be described by Reaction Eq. (1): RSNO þ R VSHYRSH þ RVSNO

ð1Þ

where a S-nitrosothiol (RSNO) interacts with a thiol (R′SH), the reaction being reversible [27]. Both cases can result in the creation of a disulphide bond between cysteines, which is linked to transcriptional activity [28]. Since the range of GSNO concentrations for enzyme transcriptional activation (see Section 2 for more detail) is 200 μM to 1 mM [9,10], GSNO concentrations for crystallisation were selected as 250 μM, 350 μM and 750 μM. Of these values, only the first two produced GHO crystals. The reason for the failure of crystals at the latter concentration is that the formation of a stable H3–H3′ dimer necessitates that there be a certain amount of instability in the rest of the GHO, thereby rendering conditions unfavourable to crystal contact formation. Different buffer conditions are currently being assessed for their ability to foster crystallisation at higher concentrations of GSNO. Previously, the authors of the work herein have produced HO crystals at a resolution of 1.90 Å (PDB code 1TZY) [7]. In that structure, as for the current model, the crystals grown have been in the same P65 space group, with a unit cell c-axis parameter of 103.58 Å. With the current model, the c-axis parameter would be expected to be approximately the same, but has actually reduced to 101.04 Å. As all sets of crystals were produced in the same buffer conditions, variations in the crystallisation environment can be ruled out as accounting for changes in the unit cell c-axis parameter and, therefore, the reduction in the unit cell c-axis of the GHO is due solely to the effects of GSNO.

Fig. 2. The H3–H3′ interface regions. Stereo image of the H3–H3′ structure in the GHO, and showing the interface region. The colour coding is H3-blue and red, H3′ orange and yellow. The red and orange regions are the actual interface areas. The amino-acid residues (for each H3) that form the interface region are: Asp 106, Leu 109, Cys 110, His 113, Ala 114, Lys 115, Arg 116, Lys 122, Asp 123, Leu 126, Ala 127, Arg 129, Ile 130 and Arg 131. The image was generated in Deepview [20], from a Pisa-generated PDB file [22]. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

C.M. Wood et al. / Biochimica et Biophysica Acta 1764 (2006) 1356–1362

1361

Fig. 3. GSNO modifies the hydrogen bond distribution. This stereo image shows a small section of the interface region shown in Fig. 2, with some amino-acid residues that are close to, though not in the region. The image shows amino-acid residues Cys 110, Ala 111, Ile 112, His 113 and Ala 114 (not labelled) in H3 and His 113 (not labelled) in H3′. In the results it was shown how the H3–H3′ dimer had, in the GHO, become thermodynamically stable. The explanation for that stability can be seen (for part of the interface) in this image which shows the same set of amino-acid residues for both the SM (the 1.90 Å starting model-PDB code 1TZY) [7] and the GHO (the GSNO-exposed histone octamer, described herein-PDB code 2ARO). The colour code is blue for PDB model 1TZY, red for PDB model 2ARO and light/dark green for hydrogen bonds, with lighter green for the stronger, dark green for the weaker bonds. In the 1TZY model, the oxygen of Cys 110 has two hydrogen bonds going to a histidine and an alanine. Notice, though, in the 2ARO model, how these two hydrogen bonds have been replaced by a single bond going to a different aminoacid residue. Although there are fewer hydrogen bonds in the 2ARO interface region, there is a greater predominance of these longitudinal hydrogen bonds. Their addition to the hydrogen bond distribution is that, in the 2ARO model, the H3–H3′ dimer is pulled closer together when compared to the 1TZY H3–H3′ dimer, and is an example of redox modulation [29]. The result is an increase in thermodynamic stability. The image was generated in Deepview [20]. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

An analysis of the hydrogen bonding pattern of the GHO (Table 2) has shown that there are major changes to H2B (H2B′) and H3 (H3′), whilst H2A (H2A′) and H4 (H4′) are essentially unchanged. The explanation of this is that the hydrogen bonds in H2B and H3 are being modulated in order that the redox potential of the GHO may be altered [29], as would be required in the redox-signalling pathway. Although the redox potential is altered in the redox-signalling pathway, it is necessary to maintain the association constants of any docking regions in the GHO at a constant value [29]. Since the H2A and H4 docking domains have been identified as an octamer binding site [7], their association constants should not vary and they therefore experience little in the way of hydrogen bond changes. Although no disulphide bond was found in the GHO, there is a marked change in the hydrogen bond distribution. This redistribution increases the buried surface area between the two H3s and this changes the energy profile of the H3–H3′ interface. The H3–H3′ interface for the GHO is shown in Fig. 2, and in greater detail in Fig. 3. Fig. 3 shows, at the amino-acid residue level, how hydrogen bonds have been altered by the GSNO. These hydrogen-bond changes – occurring across most of the GHO – alter the free energy status quo. The energy calculations shown in Table 3 clearly indicate that the H3–H3′ interface in the GHO has become thermodynamically stable, compared to the SM. This increase in free energy is almost precisely what would be expected for the amount of movement seen in the GHO [30]. The increase in free energy of the H3– H3′ interface, with its associated stability, is brought about only by the disruption of the hydrogen bonding pattern across the GHO. Higher concentrations of GSNO may, as has been suggested [31], or may not result in the formation of a

disulphide bond between the H3 and H3′ cysteines—our results allow no conclusions to be drawn on this matter. However, it is clear from the results that higher levels of GSNO will result in further losses of hydrogen bonds within secondary structure elements and, therefore, loss of tertiary structure. This work, then, clearly and conclusively shows that disulphide bond formation is incompatible with a stable (H4–H3)–(H3′– H4′) tetramer within the GHO. Acknowledgments This work was supported by John Moores University and forms part of the Daresbury Collaborative Research Programme. Harrisons Poultry, Antrobus, Cheshire, UK kindly supplied the chicken blood. Structural data were collected on multiple-wavelength anomalous (MAD) diffraction beamline 10.1 at the CCLRC Daresbury Laboratory, Warrington, UK. Beamline 10.1 is run by the North-West Structural Genomics Centre (NWSGC) and is funded by BBSRC grants 719/B15474 and 719/REI20571 and North-West Development Agency (NWDA) project award N0002170 [32]. The authors thank Michele Cianci and Mark Ellis for their assistance on beamline 10.1 and Steven Buffey for help with earlier, automated GHO crystal screening on SRS station 14.2.

References [1] C.W. Akey, K. Luger, Histone chaperones and nucleosome assembly, Curr. Opin. Struct. Biol. 13 (2003) 6–14. [2] J.M. Harp, B.L. Hanson, D.E. Timm, G.J. Bunick, Asymmetries in the nucleosome core particle at 2.5 Å resolution, Acta Crystallogr. D56 (2000) 1513–1534.

1362

C.M. Wood et al. / Biochimica et Biophysica Acta 1764 (2006) 1356–1362

[3] K. Luger, A.W. Maeder, R.K. Richmond, D.F. Sargent, T.J. Richmond, Structure of the nucleosome core particle at 2.8 Å resolution, Nature 389 (1997) 251–260. [4] G. Arents, R.W. Burlingame, B.C. Wang, W.E. Love, E.N. Moudrianakis, The nucleosomal core octamer at 3.1 Å resolution: a tripartite protein assembly and a left-handed superhelix, Proc. Natl. Acad. Sci. U. S. A. 88 (1991) 10148–10152. [5] L. Chantalat, J.M. Nicholson, S.J. Lambert, A.J. Reid, M.J. Donovan, C.D. Reynolds, C.M. Wood, J.P. Baldwin, Structure of the histone core octamer in KCl/phosphate crystals at 2.15 Å resolution, Acta Crystallogr. D59 (2003) 1395–1407. [6] J.M. Nicholson, C.M. Wood, C.D. Reynolds, S.J. Lambert, L. Chantalat, J. P. Baldwin, Structural molecular biology of the histone octamer, in: S.G. Pandalai (Ed.), Recent Research Developments in Molecular Biology 1, Research Signpost, Trivandum, India, 2003, pp. 165–187. [7] C.M. Wood, J.M. Nicholson, S.J. Lambert, L. Chantalat, C.D. Reynolds, J.P. Baldwin, High-resolution structure of the native histone octamer, Acta Crystallogr. F61 (2005) 541–545. [8] R.D. Camerini-Otero, G. Felsenfeld, Histone H3 disulfide dimers and nucleosome structure, Proc. Natl. Acad. Sci. U. S. A. 74 (1977) 5519–5523. [9] Y. Ji, V. Toader, B.M. Bennett, Regulation of microsomal and cytosolic glutathione S-transferase activities by S-nitrosylation, Biochem. Pharmacol. 63 (2002) 1397–1404. [10] Q. Shi, Y.J. Lou, Microsomal glutathione transferase 1 is not S-nitrosylated in rat liver microsomes or in endotoxin challenged rats, Pharmacol. Res. 51 (2005) 303–310. [11] S.J. Lambert, J.M. Nicholson, L. Chantalat, A.J. Reid, M.J. Donovan, J.P. Baldwin, Purification of histone core octamers and 2.15 Å X-ray analysis of crystals in Kcl/phosphate, Acta Crystallogr. D55 (1999) 1048–1051. [12] A.G.W. Leslie, Integration of macromolecular diffraction data, Acta Crystallogr. D55 (1999) 1696–1702. [13] CCP4, The CCP4 suite: programs for protein crystallography, Acta Crystallogr. D50 (1994) 760–763. [14] J. Navaza, AMoRe: an automated package for molecular replacement, Acta Crystallogr. D50 (1994) 157–163. [15] G.N. Murshudov, A.A. Vagin, E.J. Dobson, Refinement of macromolecular structures by the maximum-likelihood method, Acta Crystallogr. D50 (1997) 240–255. [16] P. Emsley, K. Cowtan, Coot: model building tools for molecular graphics, Acta Crystallogr. D60 (2004) 2126–2132. [17] V.S. Lamzin, K.S. Wilson, Automated refinement of protein models, Acta Crystallogr. D49 (1994) 129–149.

[18] G.N. Ramachandran, G. Ramakrishnan, V. Sasisekhran, Stereochemistry of polypeptide chain configurations, J. Mol. Biol. 7 (1963) 95–99. [19] W. Kabsch, A solution for the best rotation to relate two sets of vectors, Acta Crystallogr. A32 (1976) 922–923. [20] N. Guex, M.C. Peitsh, SWISS-MODEL and Swiss-PDBViewer: an environment for comparative protein modeling, Electrophoresis 18 (1997) 2714–2723. [21] J. Moskovitz, S. Bar-Noy, W.M. Williams, J. Requena, B.S. Berlett, E.R. Stadtman, Methionine sulfoxide reductase (MsrA) is a regulator of antioxidant defense and lifespan in mammals, Proc. Natl. Acad. Sci. U. S. A. 98 (2001) 12920–12925. [22] E. Krissinel, K. Henrick, Detection of protein assemblies in crystals, in: M.R. Berthold, et al., (Eds.), CompLife 2005, LNBI 3695, SpringerVerlag, Berlin, 2005, pp. 163–174. [23] L. Wyns, I. Lasters, R. Hamers, Cross-linking of nucleosomal histones with monofunctional imodoesters, Nucleic Acids Res. 5 (1978) 2345–2348. [24] V.V. Sumbayev, I.M. Yasinska, Regulation of MAP kinase-dependent apoptotic pathway: implication of reactive oxygen and nitrogen species, Arch. Biochem. Biophys. 436 (2005) 406–412. [25] H. Choi, S. Kim, P. Mukhopadhyay, S. Cho, J. Woo, G. Storz, S. Ryu, Structural basis of the redox switch in the OxyR transcription factor, Cell 105 (2001) 103–113. [26] H.F. Gilbert, Molecular and cellular aspects of thiol-disulfide exchange, Adv. Enzymol. Relat. Areas Mol. Biol. 63 (1990) 69–172. [27] E.R. Stadtman, Role of oxidant species in aging, Curr. Med. Chem. 11 (2004) 1105–1112. [28] S.O. Kim, K. Merchant, R. Nudelman, W.F. Beyer, OxyR: a molecular code for redox-related signaling, Cell 109 (2002) 383–396. [29] M. Gray, A.O. Cuello, G. Cooke, V.O. Rotello, Hydrogen bonding in redox-modulated molecular recognition. An experimental and theoretical investigation, J. Am. Chem. Soc. 125 (2003) 7882–7888. [30] D. Kern, E.R.P. Zuiderweg, The role of dynamics in allosteric regulation, Curr. Opin. Struct. Biol. 13 (2003) 748–757. [31] S.B. Hake, C. David Allis, Histone H3 variants and their potential role in indexing mammalian genomes: the “H3 barcode hypothesis”, Proc. Natl. Acad. Sci. 103 (2006) 6428–6435. [32] M. Cianci, S. Antonyk, N. Bliss, M.W. Bailey, S.G. Buffey, K.C. Cheung, J.A. Clarke, G.E. Derbyshire, M.J. Ellis, M.J. Enderby, A.F. Grant, M.P. Holbourn, D. Laundy, C. Nave, R. Ryder, P. Stephenson, J.R. Helliwell, S.S. Hasnain, A high-throughput structural biology/proteomics beamline at the SRS on a new multi-pole wiggler, J. Synchrotron Radiat. 12 (2005) 455–466.