Structure of a Sir2 substrate, Alba, reveals a mechanism for deacetylation-induced enhancement of DNA binding.

The targeted acetylation status of histones and several other transcriptional regulatory proteins plays an important role in gene expression, although the mechanism for this is not well understood. As a model to understand how targeted acetylation may effect transcription, we determined the x-ray crystal structure of the chromatin protein Alba from Archaeoglobus fulgidus, a substrate for the Sir2 protein that deacetylates it at lysine 11 to promote DNA binding by Alba. The structure reveals a dimer of dimers in which the dimer-dimer interface is stabilized by several conserved hydrophobic residues as well as the lysine 11 target of Sir2. We show that, in solution, the mutation of these hydrophobic residues or lysine 11 disrupts dimer-dimer formation and decreases DNA-binding affinity. We propose that the in vivo deacetylation of lysine 11 of archaeal Alba by Sir2 promotes protein oligomerization for optimal DNA binding. Implications for the mechanism by which histone acetylation modulates gene expression are discussed.

The targeted acetylation status of histones and several other transcriptional regulatory proteins plays an important role in gene expression, although the mechanism for this is not well understood. As a model to understand how targeted acetylation may effect transcription, we determined the x-ray crystal structure of the chromatin protein Alba from Archaeoglobus fulgidus, a substrate for the Sir2 protein that deacetylates it at lysine 11 to promote DNA binding by Alba. The structure reveals a dimer of dimers in which the dimer-dimer interface is stabilized by several conserved hydrophobic residues as well as the lysine 11 target of Sir2. We show that, in solution, the mutation of these hydrophobic residues or lysine 11 disrupts dimer-dimer formation and decreases DNA-binding affinity. We propose that the in vivo deacetylation of lysine 11 of archaeal Alba by Sir2 promotes protein oligomerization for optimal DNA binding. Implications for the mechanism by which histone acetylation modulates gene expression are discussed.
The histone protein building blocks that package DNA into chromatin play key roles in gene regulation. An octameric core of histone proteins containing two copies each of histones H2A, H2B, H3, and H4, together with ϳ146 base pairs of DNA wrapped around these proteins, make up a nucleosome particle (1). These nucleosome particles form the higher order chromatin structure that is the scaffold from which gene regulation must occur. The histone proteins themselves each contain a globular, highly helical core region that forms the interior of the nucleosome particle and a highly conserved and considerably more flexible N-terminal tail region. The histone tails play a very dynamic role in gene regulation by serving as sites of targeted and extensive post-translation modification, including acetylation, phosphorylation, and methylation (2,3). These histone modifications also appear to be binding sites for other protein domains (3). It appears that a specific histone modification or constellation of modifications leads to distinct transcriptional events (4 -6), although the mechanism by which this occurs is still poorly understood.
Histone H3 methylation at lysine 9 forms one example of how a histone modification can lead to a downstream transcriptional regulatory event. The methylation is carried out, at least in part, by the Su(var)3-9 protein, and the methylated lysine specifically recruits heterochromatin protein 1 (HP1) though its chromodomain to silence transcription from heterochromatic DNA (7)(8)(9). The mechanism by which other histone modifications, including histone acetylation, modulate transcription is less well understood. Surprisingly, many proteins initially identified as histone acetyltransferases (HATs), such as p300/CBP and PCAF, have also been shown to acetylate non-histone transcription factors, including many transcriptional activators, to promote gene activation (10). In addition, histone deacetylases (HDACs) of the Sir2 family have been shown to deacetylate non-histone transcription factors such as the p53 tumor suppressor protein (11,12). Moreover, a Sir2 homologue from the archaea Sulfolobus solfataricus has been reported to deacetylate a chromatin, non-sequence-specific DNA-binding protein called Alba (formally known as Sso10b) to promote Alba/DNA association and transcriptional repression (13). These additional acetylation and deacetylation substrates provide convenient models for understanding the mechanism of acetylation-dependent modulation of protein activity. For several reasons, the archaeal Alba protein provides a particularly interesting model for understanding how the acetylation status of eukaryotic histones may affect its biological activity. Like the eukaryotic histones, Alba binds DNA nonspecifically, is acetylated in its N-terminal terminus to affect its DNA binding and transcriptional regulatory properties, and also forms higher order structures to bind DNA (14 -16). Interestingly, Alba homologues are also found in eukaryotes (16), although their greater sequence divergence than that of the archaeal proteins suggests that they may have evolved a new function in eukaryotes.
To understand the mechanistic basis for how the acetylation status of Alba may affect its DNA binding properties, we have determined the high resolution crystal structure of Alba from the archaea Archaeoglobus fulgidus (Af-Alba), 1 and have carried out a series of functional studies for correlation with our structural findings. We find that Af-Alba forms a dimeric structure that is similar to the dimeric Alba from the archaea S. solfataricus (15). More interestingly, both crystal forms of Af-Alba characterized in this study show homologous dimer-dimer contacts that involve several conserved hydrophobic residues as well as the lysine 11 substrate for acetylation. Correlating with the importance of this dimer-dimer interface for DNA binding by Af-Alba and acetyl-lysine regulation, cross-linking studies reveal the presence of tetrameric forms of Af-Alba, and mutations of lysine 11 or other hydrophobic residues at the dimer-dimer interface reduce tetramer formation and DNA binding activity by Af-Alba. The implications of these studies for understanding how the acetylation status of eukaryotic histones may affect its biological function are discussed.

MATERIALS AND METHODS
Protein Expression and Purification-The full-length Alba gene from A. fulgidus (Af-Alba) was PCR amplified from A. fulgidus genomic DNA and cloned into the pGEX4T-1 vector to express the N-terminal glutathione S-transferase (GST) fusion protein. The plasmid was transformed into the Escherichia coli BL21 (DE3) strain, grown at 37°C to an absorbance of 0.5-0.7 at 590 nm, and induced at 30°C for 3-4 h by adding 0.5 mM isopropyl-1-thio-␤-D-galactopyranoside (IPTG). For protein purification, a cell pellet from a 3-liter growth culture was suspended and sonicated in a solution containing 20 mM Hepes, pH 7.5, 200 mM NaCl, 10 mM dithiothreitol, and 1 mM PMSF (buffer A). The supernatant was mixed with about 3 ml of glutathione resin for 2 h at 4°C, packed into a column, and washed with Ͼ40-column volume buffer A without PMSF. GST-Alba-bound glutathione resin was removed from the column and treated with thrombin protease at 4°C for Ͼ20 h. The intact Af-Alba protein containing three additional N-terminal residues remaining from the thrombin protease cleavage was released from GST in a buffer containing 20 mM Na-citrate, pH 6.0, 100 mM NaCl, 2 mM dithiothreitol, and 1 mM PMSF (buffer B). The supernatant was recovered by centrifugation, concentrated to about 2 ml using Millipore (Amicon) filtration units, and heat treated at 75-80°C for 10 -20 min to precipitate contaminating proteins. The supernatant was then recovered by centrifugation and chromatographed on a Superdex-75 gel filtration column run with buffer B without PMSF. Peak fractions containing Af-Alba protein eluted in a position close to the 158-kDa globular protein standard, suggesting an oligomeric state and/or elongated shape for the 10-kDa Af-Alba protein. Af-Alba was judged to be Ͼ95% pure by SDS-PAGE and concentrated to 20 -30 mg/ml for storage at Ϫ70°C. Selenomethionine-derivatized Af-Alba protein was overexpressed from pGEX4T-1/GST-Alba-transformed bacterial strain B834 (DE3) (Novagen) and grown in MOPS-based minimal medium as described elsewhere (17). Substitution mutations in Af-Alba were gener-ated from the pGEX4T-1 plasmid overexpressing GST-Alba with sitedirected mutagenesis (QuikChange, Stratagene) (18), and the respective mutations were confirmed by DNA sequencing. Selenomethionine-derivatized and mutant Af-Alba proteins were purified and stored essentially as described for the unmodified protein.
Crystallization and Structure Determination-Crystals of native Af-Alba and selenomethionine-derivatized Af-Alba were grown at room temperature using the hanging drop vapor diffusion method. Crystals were obtained in two different forms in the space groups P4 3 2 1 2 with two molecules per asymmetric unit cell and I2 1 2 1 2 1 with one molecule per asymmetric unit cell. The P4 3 2 1 2 crystal form was obtained by mixing 10 mg/ml protein, 1:1 l, with a reservoir solution containing 15% isopropanol, 50 mM sodium cacodylate, pH 6.0, 100 mM KCl, and 25 mM MgCl 2 and equilibrating over 0.5 ml of reservoir solution. The I2 1 2 1 2 1 crystal form was obtained by mixing 6 mg/ml protein, 1:1 l, with a reservoir solution containing 30% polyethylene glycol (PEG) 400, 100 mM Tris-HCl, pH 8.5, and 200 mM sodium citrate. Both crystal forms grew to a typical size of 100 ϫ 100 ϫ 30 m over 6 days. The P4 3 2 1 2 and I2 1 2 1 2 1 crystal forms were transferred to a reservoir solution supplemented with 15% 2-methyl-2,4-pentanediol (MPD) and 40% polyethylene glycol 400 cryoprotecting reagent, respectively, prior to flash freezing and storage in solid propane prior to data collection. Native data from both crystal forms and a three-wavelength (peak, inflection, and remote) multiwavelength anomalous diffraction (MAD) data set from selenomethionine-derivatized P4 3 2 1 2 crystals were collected on beamline BM19B at the Advanced Photon Source using an ADSC Quantum-4 CCD detector at 100 K. All data was processed with the HKL-2000 suite (HKL Research Inc.). Data collection statistics are presented in Table I. Four selenium sites were identified in the P4 3 2 1 2 crystal form using CNS (19) and SOLVE (20) and confirmed with cross-difference Fourier maps. Those sites were further refined by CNS and SOLVE. The map was improved by solvent flattening with the program RESOLVE (20), and the program O (21) was used to build residues 4 -89 of the protein using the selenomethionine sites as guides. Refinement was carried out using simulated annealing (22) and torsion angle dynamic (23) refinement protocols in CNS with iterative manual adjustments of the model using the program O with reference to 2F o Ϫ F c and F o Ϫ F c electron density maps. At 2.65 Å resolution, a bulk solvent correction (24) was applied to the data, individual atomic Bfactors were adjusted, and solvent molecules were modeled into the where T is a test data set of 10% of the total reflections randomly chosen and set aside prior to refinement).
electron density map. The final model was checked for errors with composite-simulated annealing omit maps (25). The I2 1 2 1 2 1 crystal form was determined by molecular replacement using the program CNS, with the refined P4 3 2 1 2 crystal form as a search model, and refined to a resolution of 2.0 Å essentially as described for the P4 3 2 1 2 crystal form. Both structures were refined with excellent residuals and stereochemistry (Table I). The P4 3 2 1 2 crystal form was refined to 2.65 Å resolution with an R working and R free of 23.7% and 28.5%, respectively, and the I2 1 2 1 2 1 form was refined to 2.0 Å resolution with an R working and R free of 22.6% and 25.6%, respectively (Table I).
Cross-linking and Gel Retardation Assays-The reagent ethylene glycol bis[succinimidylsuccinate] (EGS; Pierce) was used for native and mutant Af-Alba cross-linking studies. Different concentrations of EGS were mixed with 20 g of protein in 10 l of a solution containing 20 mM Hepes, pH 7.5, and 150 mM NaCl and reacted at room temperature for 30 min. The reaction was stopped by adding 0.35 l of 1 M Tris-HCl, pH 7.5, and an equal volume of SDS-loading dye and analyzed on a 4 -20% gradient SDS-PAGE gel. For gel retardation assays with the native and mutant Af-Alba proteins, 250 ng of pRSET plasmid DNA was mixed with different concentrations of protein in a solution containing 20 mM MES, pH 6.5, 100 mM NaCl, 1 mM MgCl 2 , and 0.1 mg/ml bovine serum albumin and equilibrated at room temperature for 15 min. The samples were run on 0.7% agarose gels and stained with ethidium bromide for DNA visualization.

RESULTS AND DISCUSSION
Overall Structure of A. fulgidus Alba-The Af-Alba protein monomer adopts an elongated shape with dimensions of roughly 23 ϫ 26 ϫ 50 Å with a topology of ␤1-␣1-␤2-␣2-␤3-␤4 with each successive secondary structural element running in opposite directions roughly parallel to the long dimension of the molecule (Fig 1a). The ␤3 and ␤4 strands are particularly long, extending almost the entire length of the molecule. Both the P4 3 2 1 2 and I2 1 2 1 2 1 crystal forms reveal a 2-fold symmetric Af-Alba dimer burying about 1500 Å 2 of solvent-excluded surface in which the molecules interact side-by side along the long dimension of the monomer and with the 2-fold axis aligned perpendicular to the long dimension of dimer (Fig. 1b). Residues from the ␣2 helix and ␤3 and ␤4 strands mediate all of the dimer contacts (Fig 2a). Not surprisingly, the dimers in both crystal forms reported here are essentially superimposable, with an r.m.s.d. of 1.3 Å for C␣ atoms, and they are also very similar to two different crystal forms of a previously reported Alba dimer from the archaea S. solfataricus (15) with an r.m.s.d. of 3.8 and 3.7 Å for the two crystal forms, respectively. Notably, nearly all of the residues involved in dimer interactions are conserved within the Alba proteins, including the Alba homologues from eukaryotic organisms (Fig. 3), suggesting that this dimer is a conserved feature of these proteins.
Despite the excellent superposition of the four Alba protein dimers, the tip of the extended ␤3-␤4 hairpin superimposes less well, suggesting that this region of the protein may be inherently more flexible. Interestingly, the high mobility group protein IHF uses a similarly extended ␤-hairpin to bind DNA (26), suggesting that this region of Alba may be used for DNA binding.
Higher Order Oligomerization of Af-Alba-Interestingly, both crystal forms of Af-Alba show very similar dimer-dimer contacts in the crystal lattice with an r.m.s.d. of 4.3 Å. for the C␣ atoms between both tetramers (Fig. 2, b and d). Like the dimerization interaction, the dimer-dimer interactions are 2-fold symmetric and, in this case, the interface is formed primarily by antiparallel interactions between hydrophobic residues in the ␣1 helices of one subunit of each of the dimers but also involves the C-terminal tip of the ␣2 helix. In the P4 3 2 1 2 crystal form, the dimer-dimer interface is formed by van der Waals interactions between Met 14 and Met 55 Ј (where the prime designates a residue from the opposing subunit), Leu 18 and Leu 18 Ј-Leu 21 Ј, Phe 54 and Phe 54 Ј, and an H-bond between Asn 25 and Asn 15 Ј with a overall solvent-excluded surface of 4,620 Å 2 (Fig. 2c). The dimer-dimer interface of the I2 1 2 1 2 1 crystal form is somewhat more intimate, with a solvent-excluded surface of 4,665 Å 2 . This dimer-dimer interface shows van der Waals interactions between Leu 18 , Leu 18 Ј-Leu 21 Ј, and the aliphatic region of Asn 25 Ј and betweenPhe 54 and Phe 54 Ј. In addition, this interface shows a direct hydrogen bond between Lys 11 (in a loop just N-terminal to the ␣1 helix) and Glu 26 Ј (Fig.  2b). Strikingly similar dimer interfaces are observed in two different crystal lattices of the Alba structures from S. solfataricus (15), although the dimer-dimer interfaces in these structures are more similar to the P4 3 2 1 2 crystal form of Af-Alba (Fig. 2c). Also significantly, most of the residues mediating dimer-dimer contacts in Af-Alba are conserved among the archaeal proteins (Fig. 3). In particular, Phe 54 , Met 14 , and Lys 11 are strictly conserved, and positions 18 and 21 are invariably hydrophobic. Taken together, these structural features suggested that the dimer-dimer interactions observed in the crystals may be physiologically relevant. Noticeably, the conservation of archaeal residues at the dimer-dimer interface did not extend to the eukaryotic Alba homologues (Fig. 3), suggesting that the eukaryotic proteins may not form similar dimer-dimer contacts.
Solution Oligomerization and DNA Binding Properties of Native and Mutant Af-Alba Proteins-To test directly the physiological relevance of the observed dimer-dimer contacts in the crystals, we carried out mutagenesis of residues that were implicated in playing an important role at this interface. We chose to mutate Leu 18 and Phe 54 because they participated in dimer-dimer contacts in both crystal forms; we also chose to mutate Lys 11 because it is the site of deacetylation by the Sir2 protein, and we entertained the possibility that deacetylation of this residue may play a modulatory role in dimer-dimer formation. For residues 18 and 54, we chose mutations that would be predicted to be disruptive to the dimer-dimer interface; therefore, we prepared leucine to arginine (L18R) and phenylalanine to arginine (F54R) substitutions at residues 18 and 54, respectively. For lysine 11, we chose both conservative (arginine, K11R) and non-conservative substitutions for mutagenesis (glutamine, K11Q; and methionine, K11M). Modeling studies suggested that each of these mutations would disrupt the Lys 11 -Glu 26 Ј hydrogen bond revealed to stabilize the hydrogen bond in the I2 1 2 1 2 1 crystal form. As a control, we also FIG. 2. Oligomerization contacts in Af-Alba. a, the interface of the Alba dimer showing side chains that participate in dimer contacts. b, the dimer-dimer Alba interface is shown highlighting the side chain residues that stabilize the interface with a color-coded van der Waals surface representation. I2 1 2 1 2 1 and P4 3 2 1 2 indicate the tetramer in the two different crystal forms, respectively. The right panels are close-up views of the dimer-dimer interface. c, superposition of the Alba dimer-dimer (tetramer) complex in four different crystal lattices; the two reported here, in space groups P4 3 2 1 2 (cyan) and I2 1 2 1 2 1 (green), and the previously reported Alba structures from S. solfataricus, 1HOX (red) and 1HOY (yellow) are shown.

Structure of the Sir2 Substrate Alba 26074
prepared an asparagine to alanine substitution at residue 10 (N10A). Residue 10 was located at the border of the dimerdimer interface; however, it did not appear to play a direct role in dimer formation in either of the two crystal forms of Af-Alba reported here. We therefore expected that the N10A substitution would behave similarly to the native protein in its ability to form dimer-dimer contacts. We prepared proteins containing substitution mutations by site-directed mutagenesis and purified each of the mutants to homogeneity for analysis in vitro.
To determine the oligomerization states that Af-Alba and each of the mutants formed in solution, we subjected each of the proteins to cross-linking with increasing concentrations of the cross-linking reagent EGS. As can be seen from Fig. 4a, the native and N10A control proteins form both dimers and tetramers, and, at higher EGS concentrations, the relative proportions of dimer and tetramer are comparable. Each of the Af-Alba mutants at the dimer-dimer interface also show cross-linked dimers and tetramers; but in contrast to the native protein and the N10A mutant, the amount of cross-linked tetramer is significantly reduced relative to the amount of respective dimer for each of the mutants. In addition, the hydrophobic substitution mutants L18R and F18R also form high order oligomers. Together, the cross-linking data demonstrates that the Af-Alba mutants at the dimer-dimer interface form less stable tetramers than the native protein, consistent with the physiological relevance of the Af-Alba tetramer observed in the crystal lattice.
A prediction from the cross-linking studies is that dimerdimer Af-Alba contacts play a positive regulatory role in DNA binding by Af-Alba, and, therefore, one would expect that the mutations at the dimer-dimer interface would decrease DNA binding. To test this directly, we assayed the ability of Af-Alba and the mutants to band-shift plasmid DNA in agarose gels. We and others have shown that Alba does not have appreciable affinity for oligonucleotides (Ref. 15 and data not shown). Fig  4b shows that 250 ng of plasmid DNA is visibly shifted at Af-Alba concentrations above 62 ng. At higher concentrations of Af-Alba the plasmid DNA becomes more retarded until it apparently saturates the plasmid DNA at about 250 ng of Alba protein. In contrast, each of the proteins with mutations at the dimer-dimer interface visibly shifts the plasmid DNA at reproducibly higher protein concentrations of roughly 6 -8-fold. The N10A mutation, located outside of the dimer-dimer interface and still showing robust tetramer cross-linking activity (Fig.  4a), binds DNA only slightly less well (ϳ2-fold) than the native protein, suggesting that asparagine 10 of Af-Alba may be in-volved in direct DNA contact. Taken together with the structural and cross-linking results, these DNA binding studies imply that dimer-dimer formation in Af-Alba plays a stimulatory role on DNA binding by Alba. Moreover, the structural and solution data supports the proposal that lysine 11 of Alba, the target of deacetylation by the Sir2 protein, plays an important role in modulating this activity. We presume that the slight variability of specific dimer-dimer interactions in the various structures (Fig. 2) and the relatively low solvent-excluded surface at the dimer-dimer interface relative to the monomermonomer interface may further facilitate acetylation-dependent regulation of Alba oligomerization for DNA binding.
Model for DNA Binding by Alba and Its Regulation by Acetylation-The structural and functional data described above strongly implies that Af-Alba binds DNA as a dimer of dimers. The high conservation of residues at the dimer-dimer interface among other archaeal Alba proteins also implies that all members of the entire family of archaeal Alba proteins share the same feature (Fig. 3). Notably, many of these residues are not conserved within the eukaryotic Alba proteins, suggesting that they do not form similar oligomers for DNA binding (Fig. 3). As noted previously (15), a DALI search using the Alba monomer reveals structural similarity to two other known nucleic acid-binding proteins, translation initiation factor 1 (r.m.s.d. ϭ 2.2 Å) and DNase I (r.m.s.d. ϭ 2.7 Å). The structure of the DNase I in complex with an 8-base pair DNA duplex suggests where Alba may bind DNA. A superposition of the protein from the DNase I/DNA complex with one subunit of the Af-Alba dimer places the DNA on the surface of the protein formed by the ␤1-␣1 and ␤2-␣3 loops as well as the tip of the ␤3-␤4 hairpin (Fig. 5a). This superposition also places the Af-Alba subunit on the minor groove side of the DNA. Consistent with this general mode of DNA binding, this surface of the Alba protein contains the most electropositive charged surface (Fig. 5b), which would nicely complement the electronegative minor groove of the DNA, and the ␤1-␣1 and ␤2-␣3 loops contain two of the most highly conserved sequence patches within the archaeal Alba proteins (Fig. 3). Also consistent with DNA binding in this mode, only a relatively minor reorientation of the DNA helical axis of the model would allow each subunit of the Alba dimer to make symmetric interactions with a B-form DNA duplex (Fig. 5). Taken together, we propose that the Alba dimer binds along the minor groove side of DNA, in contrast to a previous model that places it across the ends of the dimer along the minor groove with the center of the molecule across the DNA major groove (15). an open circle, and residues involved in dimer-dimer contacts are highlighted with a closed circle. The lysine target for the Sir2 protein is indicated with a cross (ϩ), and the residues that were mutated for the current study are indicated with an arrow, which points to the mutated residue.
How may the Af-Alba tetramer bind DNA? Upon extending our model for DNA binding by the Af-Alba dimer to DNA binding by the Af-Alba tetramer, we find that a sharp DNA bend of about 60°is required to allow each Af-Alba dimer of the tetramer to bind DNA in the same way for each dimer (Fig. 5). This proposal is consistent with the structure of the high mobility group protein IHF bound to DNA, which also binds essentially B-form DNA as a dimer along the minor groove and uses extended antiparallel ␤-hairpin arms, similar to the extended ␤3-␤4 hairpin arms of the Af-Alba dimer, to sharply bend DNA just outside of the arm regions (26). The proposal that Alba may bend DNA is also consistent with its presumed biological role in facilitating DNA packaging in Archaea (15) by inducing negative DNA super coiling (27) without significant DNA compaction (14). Interestingly, this modeling exercise with DNA also places lysine 11, at the edge of the dimer-dimer interface, in position to also interact with DNA directly. This is consistent with the slightly greater mutational sensitivity to DNA binding of Lys 11 mutations over the Leu 18 and Phe 54 mutations (Fig. 4c). Lysine 11 may therefore play a duel role in facilitating dimer-dimer contacts as well as direct protein-DNA contacts to enhance Af-Alba affinity for DNA. The importance of lysine 11 in Af-Alba function is consistent with the important regulatory role of the deacetylation of this residue by the Sir2 protein. It is possible that Af-Alba forms even higher order structures on DNA, as suggested by recent studies showing that S. solfataricus Alba saturates plasmid DNA with a stoichiometry of about 10 bp per protein dimer (15), a stoichiometry that would be higher than that predicted from our model for DNA binding. Without further structural information, it is unclear how this may occur, but it could involve the incorporation of Alba dimers to bridge contacts between two Alba tetramers bound to DNA.
Like the archaeal Alba protein, eukaryotic histone proteins are also acetylated to modulate gene expression. In the case of the histone proteins, hyperacetylation is generally correlated with gene activation, whereas hypoacetylation is correlated with repression. The mechanism by which the acetylation status of histones regulates gene expression is still unclear; however, several models have been proposed. In one model, acetylation of lysine residues neutralizes the lysine charge of the lysine residues, thereby loosing the interaction of the histone tails with DNA and thus increasing DNA access for the binding of transcription factors to promote transcriptional activation. A second model argues that the unacetylated histone tails interact with neighboring nucleosomes in chromatin and that acetylation relieves these interactions to destabilize repressive higher order chromatin. A third model argues that histone acetylation is one histone modification among many that establish a "histone code" to mark histone binding sites for other transcriptional regulatory proteins to elicit modification-dependant transcriptional activities (4 -6). There have been several studies that support each of these models, and it is likely that they are not mutually exclusive. The studies we present here suggest that acetylation-induced regulation of archaeal Alba proteins involve the modulation of a higher order Alba structure. This would be consistent with a model in which histone tail acetylation would disrupt higher order chromatin structure. Confirmation of this model would certainly require further direct studies involving histone proteins. Nonetheless, the structural and functional studies presented here suggest that the acetylation status of archaeal Alba effects its oligomerization status, and it is possible that the evolution to eukaryotic systems and histone proteins may have maintained this mechanism of transcriptional regulation. FIG. 4. Solution oligomerization and DNA binding properties of Af-Alba. a, cross-linking of native Af-Alba, a control N10A mutant, and mutants across the dimer-dimer interface (K11Q, K11M, K11R, L18R, and F54R). Cross-linking was carried out with 20 g of protein and increasing concentrations of the cross-linking reagent EGS and reacted for 30 min. Lane 2 is Af-Alba without EGS. Reaction products are analyzed on 4 -20% SDS-PAGE gels. b, gel retardation analysis of native Af-Alba and mutants. 250 ng of plasmid DNA was incubated with increasing concentration of protein for 15 min followed by electrophoresis on agarose gels. Lanes 1 and 2 are plasmid and Af-Alba protein alone, respectively.