Structure and Function of the BAH Domain in Chromatin Biology

http://informahealthcare.com/bmg ISSN: 1040-9238 (print), 1549-7798 (electronic) Editor: Michael M. Cox Crit Rev Biochem Mol Biol, 2013; 48(3): 211–221 ! 2013 Informa Healthcare USA, Inc. DOI: 10.3109/10409238.2012.742035

REVIEW ARTICLE Structure and function of the BAH domain in chromatin biology

Na Yang and Rui-Ming Xu

National Laboratory of Biomacromolecules, Institute of Biophysics, Chinese Academy of Sciences, Beijing, China

Abstract Keywords Proteins containing Bromo Adjacent Homology (BAH) domain are often associated with DNA replication, histone modification, biological processes involving chromatin, and mutations in BAH domains have been found in nucleosome remodeling, protein–protein human diseases. A number of structural and functional studies have revealed that the BAH interaction, structure, transcriptional domain plays diverse and versatile roles in chromatin biology, including protein–protein silencing interactions, recognition of methylated histones and nucleosome binding. Here we review recent developments in structural studies of the BAH domain, and intend to place the structural History results in the context of biological functions of the BAH domain-containing proteins. A converging theme from the structural studies appears that the predominantly b-sheet fold Received 28 August 2012 of the BAH domain serves as a scaffold, and function-specific structural features are Revised 17 October 2012 incorporated at the loops connecting the b-strands and surface-exposed areas. The structures Accepted 17 October 2012 clearly specified regions critical for protein–protein interactions, located the position of Published online 27 November 2012 methyllysine-binding site and implicated areas important for nucleosome binding. The structural results provided valuable insights into the molecular mechanisms of BAH domains in molecular recognitions, and the information should greatly facilitate mechanistic under- standing of BAH domain proteins in chromatin biology.

Introduction SNF-like chromatin-remodeling complex (Wang et al., 1996a, 1996b; Thompson, 2009). The yeast RSC chromatin-remo- The term Bromo Adjacent Homology (BAH) domain was deling complex has two BAH domain-containing subunits, coined in 1996 by an analysis of domain structures of the RSC1 and RSC2 (Cairns et al., 1996, 1999). Many BAH chicken polybromo (gPB) protein (Nicolas & Goodwin, domains are present together with domains frequently 1996). Polybromo contains six tandem bromo domains at associated with nuclear functions, such as Bromo, SANT the N-terminus half, followed by two repeated sequence and PHD domains, which are known to bind histones or DNA. motifs of unknown functions, thus called BAH domains, and a A schematic diagram showing major classes of BAH domain HMG DNA-binding box near the C-terminal end (Figure 1). proteins are shown in Figure 1. Further sequence analyses extended the original definition of Early analyses suggested that the BAH domain might be a the BAH domain from an 85 residue motif to one with protein–protein interaction module, but little was known 120–140 residues, and revealed the broad presence of the about its interaction partners and mode of interactions BAH domain in proteins associated with chromatin processes (Callebaut et al., 1999; Goodwin & Nicolas, 2001; Zhang in species ranging from yeasts to mammals (Callebaut et al., et al., 2002). Recent developments, notably from structural 1999; Goodwin & Nicolas, 2001). At present, SMART 2013 studies, have significantly advanced understandings of the protein domain database tabulated 1082 BAH domains in molecular mechanisms of BAH domains, particularly in areas 903 proteins, which include five proteins in Saccharomyces of epigenetic inheritance and gene regulation. This review is cerevisiae, four in Caenorhabditis elegans, eight in intended to provide an overview of the structural findings and Drosophila,26inArabidopsis, 21 in mouse and 19 proteins to place the structural results in the context of biological in human (http://smart.embl-heidelberg.de/). Most of the functions of the BAH domain-containing proteins. characterized BAH domain proteins have clear connections to chromatin processes, such as nucleosome-remodeling, histone and DNA modifications. The human ortholog of Overall structure polybromo, BAF180/PB1, is a component of PBAF, a SWI/ To date, six BAH domain structures have been determined (Figure 2). They are the BAH domains of yeast and mouse Orc1 (Zhang et al., 2002; Hou et al., 2005; Hsu et al., 2005; Address for Correspondence: Rui-Ming Xu or Na Yang, National Kuo et al., 2012), yeast Sir3 (Connelly et al., 2006; Hou Laboratory of Biomacromolecules, Institute of Biophysics, Chinese Academy of Sciences, Beijing 100101, China. E-mail: rmxu@sun5. et al., 2006), the first BAH domain of gPB (Oliver et al., ibp.ac.cn; [email protected] 2005) and the tandem BAH domains (BAH1 and BAH2) 212 N. Yang & R.-M. Xu Crit Rev Biochem Mol Biol, 2013; 48(3): 211–221

Figure 1. Schematic diagram showing domain structures of representative BAH domain-containing proteins. Shapes and colors indicating different types of domains are annotated at the bottom of the figure. Protein names are indicated on the left, and the drawings are approximately in scale with the length of the protein. (see colour version of this figure online at www.informahealthcare/bbm).

of mouse DNMT1 (Song et al., 2011, 2012; Takeshita silencing functions, demonstrating that the BAH domain et al., 2011) (Figure 2). The yeast Orc1 (yOrc1) BAH alone is sufficient for Sir1-dependent silencing (Triolo & structure was the first to be solved (Zhang et al., 2002). Orc1 Sternglanz, 1996). is the largest subunit of the six-subunit origin recognition The structure of the BAH domain of yOrc1 shows that the complex (ORC), which functions in the initiation of DNA structurally defined region encompasses residues 1–215, with replication and transcriptional silencing (Bell & Stillman, a disordered internal loop spanning residues 23–36 (Zhang 1992; Bell et al., 1993; Fox et al., 1995, 1997; Loo et al., et al., 2002). The BAH domain has a predominantly b-sheet 1995). The BAH domain of yOrc1 is not required for DNA core composed of 10 b-strands, b1–b10 (Figure 2A). Two replication, but later studies show that it is important for the long antiparallel strands, b5 and b6, together with b4 and b7 association of the ORC complex with chromatin and form a central b-sheet. b7 is highly twisted, and in some BAH regulation of the origin activity (Bell et al., 1995; Lipford domains it breaks into two short strands (Figure 2D and E). & Bell, 2001; Noguchi et al., 2006; Eaton et al., 2010; Muller b1–b3 form a separate sheet and pack against the central et al., 2010). The BAH domain of yOrc1 functions in b-sheet at one end of the central b-sheet near the b5-turn-b6 transcriptional silencing of the mating-type (HM) loci, which region, with the two sets of b-strands are approximately are flanked by DNA elements, are known as silencers. perpendicular to each other. Parallel strands b8 and b9 form Silencer-bound ORC complexes interact with Sir1, which a sheet with a short strand, b10, and together with a small a recruits the Sir2-Sir4 histone deacetylase complex via inter- helix (aE), pack against the central b-sheet on the same action with Sir4 (Chien et al., 1993; Triolo & Sternglanz, side of the b-sheet surface, but at the end distal to the one 1996; Gardner et al., 1999; Bose et al., 2004). Sir4 can occupied by the b1–b3 sheet (Figure 2A). A pair of helices simultaneously interact with Sir3, leading to the assembly of at the C terminus pack against the middle portion of the the Sir2–Sir3–Sir4 complex required for silencing the HM central b-sheet. Finally, a small, independently folded, loci (Rusche et al., 2003). Hence, the silencer-bound ORC non-conserved helical domain (H-domain) is inserted complex confers Sir1-dependent silencing via the BAH between b6 and b7. domain of yOrc1. In fact, an isolated BAH domain of Only the central b-sheet is common among all BAH yOrc1 targeted to an artificial silencer via the Gal4 domains, while N- and C-terminal extensions, and the DNA-binding domain is capable of achieving similar H-domain are distinct (Figure 2). The conserved residues DOI: 10.3109/10409238.2012.742035 BAH domains 213

Figure 2. 3D structures of BAH domains. All six BAH domain structures known to date are shown in (A)–(F). The conserved b-sheet core is colored green, the common insertion region (H-domain) between b6 and b7 is colored magenta and auxiliary regions are colored grey. (see colour version of this figure online at www.informahealthcare/bbm). among the BAH domains are mostly involved in protein experiments indicate that the non-conserved H-domain within folding, suggesting that the BAH domain serves as a scaffold the BAH domain is important for Sir1 binding (Zhang et al., for harboring structural features with specific functions. 2002). The cocrystal structure shows that Sir1 OID is bound in a shallow groove formed between the b-sheet core and the non-conserved H-domain of yOrc1 BAH domain (Hou et al., BAH domain as a scaffold for protein–protein 2005; Hsu et al., 2005) (Figure 3). The three yOrc1 regions interaction contacting Sir1 OID are: (I) the loop connecting b6 and the Sir1 is the first BAH domain-binding partner identified. The first helix (aB) of the H-domain, (II) the third helix (aD) of Orc1-binding domain (OID) of Sir1 maps to an approximately the H-domain and (III) the b4–b5 turn region of the b-sheet 130-residue region near the C-terminus (Triolo & Sternglanz, core. Sir1 OID is composed of two juxtaposed subdomains 1996; Gardner et al., 1999; Bose et al., 2004). Structure- rich in b-strands. The N-terminal subdomain is involved in guided mutagenesis, GST-pull down and yeast two hybrid contacting the BAH domain of yOrc1. Two small residues, 214 N. Yang & R.-M. Xu Crit Rev Biochem Mol Biol, 2013; 48(3): 211–221 Orc1 N-terminal regions encompassing the BAH domain were found to interact with heterochromatin protein 1 (HP1), a major component of constitutive heterochromatin (Pak et al., 1997). Whether the BAH domain is directly involved in HP1 binding is not known, but a C-terminal extension beyond the predicted BAH domain has been shown to be required for HP1 binding. Further studies are needed to determine the exact roles of the Orc1 BAH domain in HP1 binding.

BAH domain as a nucleosome-binding unit Initial indications that the BAH domain may directly interact with the nucleosome came from the observation that a bacterially expressed Sir3 BAH domain hypermorphic mutant, D205N, binds oligonucleosomes and nucleosomal DNA (Connelly et al., 2006). Thorough investigations show that the Sir3 BAH domain produced in yeast specifically interact with the nucleosome, and the binding depends on acetylation of the N-terminal amino group of Sir3 by the Figure 3. Structure of the BAH domain of yeast Orc1 in complex with Ard1-Nat1 acetyltransferase complex (Onishi et al., 2007; Sir1. Orc1 is colored as in Figure 2, and Sir1 is colored cyan. Three Sampath et al., 2009). The same is true for the BAH domain regions of intermolecular contacts are labeled with Roman numerals. of yOrc1. Furthermore, the Sir3 BAH domain does not bind Main interaction residues between Orc1 and Sir1 are shown in stick models (Orc1: carbon, yellow. Sir1: carbon, cyan. All: oxygen, red; nucleosomes acetylated at H3K16, or with H3K16 or H3K79 nitrogen, blue). (see colour version of this figure online at www. mutations. Genetic screens revealed that nucleosome- informahealthcare/bbm). interacting residues of the Sir3 BAH domain are clustered in three regions: the most C-terminal helix (aG) outside of the conserved b-sheet core, the junction between the b-sheet core Pro492 and Ala505, of Sir1, located at the tips of b1–b2 and and the H-domain and the first residue, Ala2, of the b3–b4 turns, respectively, are positioned in shallow hydro- BAH domain (Sampath et al., 2009). In the budding yeast, phobic pockets of the BAH domain, and the small size of co-translational processing removes the initiating methionine these residues allows ideal surface complementarity between and acetylates the amino group of the following alanine the binding partners (Figure 3). Sir1 OID interacts with of Sir3 and yOrc1 (Wang et al., 2004). It is interesting to the b4–b5 turn region of the BAH domain of yOrc1 mainly note that N-terminal unacetylated Sir3 BAH domain can via hydrogen bonds and charge interactions involving a pair bind nucleosome in vitro, but requires a higher protein of arginine residues, Arg493 and Arg516, of Sir1 OID. concentration. The determinants for binding specificity towards Sir1 OID The crystal structure of the bacterially expressed D205 mainly reside in the non-conserved H-domain, as replacing mutant of the Sir3 BAH domain in complex with the the homologous region of the Sir3 N-terminal domain, which nucleosome core particle (NCP) has been determined does not bind Sir1 OID, with the yOrc1 H-domain confering (Armache et al., 2011). The structure shows that two Sir3 the Sir1 OID-binding ability (Zhang et al., 2002). The BAH BAH domain molecules are bound to NCP, one on each side domains of Sir3 and yOrc1 share 50% sequence identity and of the disk-like pseudo-symmetric NCP surface (Figure 4A). 70% sequence similarity, and the two domains have very The Sir3 BAH domain contacts all four histones, burying an similar 3D structures (Zhang et al., 2002; Connelly et al., approximately 1750 A˚ 2 of the intermolecular contact area 2006; Hou et al., 2006) (Figure 2A and B). The Sir3 BAH (Figure 4A and B). Nearly 30 residues of the Sir3 BAH domain has been shown to genetically interact with the domain are involved in interaction with NCP, and they form N-terminal region of Sir1. This region shares weak but a contiguous NCP interacting surface. Consistent with the identifiable sequence similarities with the C-terminal OID. findings from the genetic screens, a number of residues on the However, evidence for a physical interaction between the Sir3 C-terminal helix aG and the spatially adjacent short strand b8 BAH domain and the N-terminal OID-like region of Sir1 is of Sir3 interact with histones H3 and H4. The interactions are still lacking. Significantly, overexpression of the BAH domain mainly with residues located on the N-terminal tail of histone of Sir3 alone can achieve silencing of the HM loci when Sir1 H4 and the loop region between a1 and a2 of the histone fold is also overexpressed, suggesting that Sir3 BAH domain may domain of H3. Notably, Lys79 of H3 is located on this loop be able to interact with Sir1, possibly though its N-terminal and it interacts with Sir3’s Glu140 located on b8 and Glu84 region, under appropriate conditions (Connelly et al., 2006). on loop-3 (Figure 4B). Interestingly, many NCP-interacting All eukaryotic Orc1 proteins contain a BAH domain, residues are located on the loop regions of the BAH domain, but higher eukaryotes lack Sir1-like proteins. Hence, the including those connecting b4 and b5 (loop-2), b5 and b6 Orc1 BAH domain must interact with another partner if a (loop-3), b6 and the H-domain, which adopts a short helix similar mechanism of heterochromatin formation is operating (A1) conformation when NCP is bound, and the last strand of in higher eukaryotes. Indeed, Drosophila and Xenopus the b-sheet core and aF (loop-4) (Figure 4A). These loop DOI: 10.3109/10409238.2012.742035 BAH domains 215

Figure 4. Structure of the BAH domain of Sir3 in complex with the nucleosome core particle (NCP). (A) Two Sir3 BAH domains bind one NCP, one BAH domain on each side of the pseudo-symmetric, disc-like structure of NCP. The BAH domain is colored as in Figure 2, and the color code for the nucleosomal histone and DNA molecules is indicated in the colored boxes at the bottom right corner. (B) An enlarged view of the interaction between the BAH domain of Sir3 and NCP. (see colour version of this figure online at www.informahealthcare/bbm). regions primarily interact with the N-terminal tail of histone in the crystal structure is near Asp77 of histone H3 but too H4 (residues 13–24), which becomes ordered upon the far (4.1 A˚ ) to make hydrogen bonds (Figure 4B). A reasonable binding of the Sir3 BAH domain and adopts a conformation explanation is that a neutral residue at this position makes the different from that of free NCP (Luger et al., 1997). Two Sir3 binding of the Sir3 BAH domain to the negatively charged BAH domain loops, one connecting b5 and b6 (loop-3), and surface region of NCP energetically more favorable. Another another connecting b1 and b2 (loop-1 and A0), were important residue for NCP binding is Ala2 of Sir3, which is disordered in the NCP-free structure and became ordered unacetylated in the structure and is near the minor grove of upon NCP binding. However, residues on loop-1 and A0 have DNA (Figure 4). Possibly, acetylation of Ala2 increases the high-temperature factors, indicating that they are highly binding affinity through interaction with DNA, but this mobile. Nevertheless, their general location implies involve- interpretation remains to be experimentally verified. The ment in interactions with histones H2A and H2B, as loop-1 BAH domains of yOrc1 and Sir3 share a similar structure and and A0 are near a wedge formed between a2 of H2A and aC some of the nucleosome-binding properties, such as depen- of H2B (Figure 4B). Loop-3 contacts a concave surface area dence on Ala2 acetylation. However, the binding of the yOrc1 of NCP formed by the C-terminal region of a2 of histone H4 BAH domain is insensitive to H4K16 acetylation, although and a2 and aC of histone H2B, and this area is adjacent to Ser67 is conserved in yOrc1. This difference suggests an H3K79 (Figure 4B). altered binding mode between NCP and the BAH domain of Polar interactions are the main binding force between NCP yOrc1. It would be very interesting to compare common and the Sir3 BAH domain. In particular, the N-terminal tail of features and differences of NCP binding by the BAH domains histone H4 traverses a negatively charged surface area of the of yOrc1 and Sir3. Sir3 BAH domain (Armache et al., 2011). Two histone H4 residues, Lys16 and His18, are critical for Sir3-NCP BAH domain as a histone methyllysine recognition interaction. They are situated in a BAH domain pocket module formed by Ser67, Glu95 and Glu137 of Sir3 (Figures 4B and 5D). The amino group of H4K16 makes a hydrogen bond with The BAH domain of mammalian ORC1 has been shown to be Ser67, and the imidazole ring of H4H18 hydrogen bonds to important for chromatin association of the ORC complex, but Glu95 and Glu137. This binding mode is consistent with the direct binding of the BAH domain of ORC1 to nucleosome observation that the acetylation of H4K16 interferes with remains to be demonstrated (Noguchi et al., 2006; Muller the binding of the Sir3 BAH domain, as the introduction of an et al., 2010). A surprising recent discovery is that the BAH acetyl group to H4K16 is likely to cause steric clashes. domains of metazoan ORC1 preferentially recognizes The structure also provides mechanistic understandings of the dimethylated H4K20 (H4K20me2) (Kuo et al., 2012). An D205N mutation of Sir3, which is known to suppress the important difference between S. cerevisiae and metazoans is phenotype caused by mutations of positively charged residues that H4K20 is not methylated in the budding yeast, and yOrc1 at the N-terminal tail of histone H4. Biochemically, the does not bind methylated H4K20. Also, replication origins in D205N mutant binds the nucleosome more tightly. Asn205 S. cerevisiae are primarily specified by DNA sequences, 216 N. Yang & R.-M. Xu Crit Rev Biochem Mol Biol, 2013; 48(3): 211–221

Figure 5. Structural basis for recognition of H4K20me2 by the BAH domain of mORC1. (A) Structure of the BAH domain of mORC1 in complex with an H4K20me2 peptide. The BAH domain is colored the same as in previous figures, and the H4 peptide is shown as a stick model superimposed onto a coil representation (yellow). (B) Four aromatic residues (Y63, W87, Y114 and W119), a glutamate (E93) and a valine (V89) form the H4K20me2- binding pocket. The corresponding residues in yOrc1 are superimposed (cyan), and they do not form an aromatic cage suitable for binding methyllysine. (C) Structure-guided sequence alignment of the BAH domain regions involved in methyllysine recognition. The residues highlighted yellow correspond to those involved in H4K20me2 binding in mORC1 and hORC1, and the magenta box indicates the H-domain region. The alignment indicates that the BAH domain of the first BAH domain of polybromo, and possibly also the first BAH domain of DNMT1, may be able to bind methyllysine. (D) Superposition of the BAH domains of ySir3 and mORC1 in their complexes with NCP and the H4K20me2 peptide, respectively, shows that the H4 tail of NCP (cyan) and the H4K20me2 peptide (yellow) are bound in a similar N- to C-terminal orientation, but differs in register by approximately two residues, as indicated by the H4K16 positions. The H4 tail and peptide are roughly more than 5 A˚ apart. The aromatic cage residues of mORC1 (green) for H4K20me2 binding and the Sir3 residues (magenta) important for H4K16 binding are shown as references. (see colour version of this figure online at www.informahealthcare/bbm). whereas no clear metazoan replication origin sequences are origins (Noguchi et al., 2006; Eaton et al., 2010; Muller et al., known. It is believed that local chromatin environment plays 2010). It is an attractive idea that H4K20 dimethylation may an important role in the specification of metazoan replication participate in the specification of origin and origin activity DOI: 10.3109/10409238.2012.742035 BAH domains 217 regulation, as the H4K20me2 level (through the availability of between the BAH domain of mORC1 and the H4K20me2 H4K20me1) may be regulated in a cell cycle-dependent peptide. manner (Kuo et al., 2012). It should be noted that a proteomic Six mORC1 residues form the H4K20me2-binding pocket: study identified the ORC complex as interactors of the Tyr63 on b5, Trp87 and Val89 on b6, Glu93 on the H-domain, H3K9me3, H3K27me3 and H4K20me3 repressive marks. Tyr114 and Trp119 on the loop between b7a and b7b (Figure Based on the work of Kuo et al., the BAH domain of ORC1 5B). The aromatic residues and Glu93 circumscribe the should preferentially bind H4K20me2, but pull-downs with negatively charged H4K20me2-binding pocket, whereas the dimethylated peptide were not tested, and the interaction Val89 lies at the bottom of the pocket. The aromatic residues between the ORC complex and the trimethylated peptides interact with dimethylated H4K20 via van der Waals, may or may not be through the BAH domain of ORC1 hydrophobic and cation- interactions, and the hydroxyl (Vermeulen et al., 2010). Interestingly, human ORC1 has groups of Glu93 make hydrogen bonds with the nitrogen atom recently been shown to interact with the Cyclin A-CDK2 and of the dimethylammonium group of H4K20me2 (Figure 5B). Cyclin E-CDK2 complexes and inhibit their kinase activities. The binding mode is reminiscent of the binding of The BAH domain of hORC1 is responsible for binding and H4K20me2 to the Tudor domain of 53BP1 protein (Botuyan inhibiting the kinase activity of the Cyclin E-CDK2 complex, et al., 2006). Dynamic movements of Trp119 and Glu93 are and amino acids located on the long loop connecting the observed in the presence and the absence of bound H-domain and b7a opposite to the H4K40me2-binding H4K20me2 (Kuo et al., 2012). Trp119 is the only residue aromatic cage are involved in the inhibition of the kinase in the binding pocket that is not conserved in hORC1. The activity (Hossain & Stillman, 2012). Finally, certain ORC1 corresponding hORC1 residue is a cysteine (Figure 5C). BAH domain mutations are found in human Meier–Gorlin Changing Trp119 to a cysteine in the BAH domain of syndrome (MGS), which is a form of primordial dwarfism mORC1 resulted in a stronger preference for dimethyl-H4K20 (Bicknell et al., 2011a, 2011b). The BAH domain mutants over mono- and tri-methylated H4K20 (Kuo et al., 2012). show reduced H4K20me2-binding affinities and Cyclin ITC measurements show that the aromatic residue mutants E-CDK2 inhibition activity, suggesting a potential link of the BAH domain of mORC1 are defective in H4K20me2 between ORC1’s H4K20me2-binding function and the binding. Cells transfected with two hORC1 mutants, Y64A pathogenesis of MGS (Hossain & Stillman, 2012; Kuo and W88A, showed reduced population of cells in S-phase et al., 2012). and origin-associated hORC1. Interestingly, mice lacking The BAH domain structure of mouse ORC1 (mORC1) in enzymes generating H4K20me2 displayed DNA replication complex with a histone H4K20me2 peptide shows that the defects reminiscent of those with hORC1 mutants in the cell- b-sheet core is conserved (Kuo et al., 2012) (Figures 2D and based assay, suggesting that the recognition of H4K20me2 by 5A). Some new features include the absence of an a-helix the BAH domain is an integral part of the ORC1 function in immediately N-terminal to b9 and a significantly shorter DNA replication (Kuo et al., 2012). H-domain insertion. The crystallized mORC1 BAH domain misses 12 residues at the N-terminus, and there are no BAH domains in chromatin modifiers auxiliary a-helices C-terminal to the b-sheet core (Figure 2). These regions are not included in the crystallization construct Eukaryotic chromatin can adopt compact or relaxed of mORC1 BAH domain. Also, the long, highly twisted b7, structure in response to biological activities such as replica- similar to b8 and b9, is broken into two shorter strands, b7a tion, transcription and DNA repair. The dynamic behavior and b7b, in mORC1 (Figure 5A). of chromatin is highly regulated by many factors including The N-terminal portion of the H4 peptide is bound in the DNA methylation, histone modifications and nucleosome region formed by a short-turn C-terminal to b7b and the loop remodeling (Jenuwein & Allis, 2001; Li & Reinberg, 2011). connecting b4 and b5 (corresponding to loop-2 in the Sir3 Several proteins important for DNA and histone modi- BAH domain), and the C-terminal portion of the H4 peptide is fications contain BAH domains, including mammalian bound to the middle part of the loop between b7a and b7b maintenance DNA methylase DNMT1, histone H3K36 (Figure 5A). The N- to C-terminus-binding orientation of the methylase ASH1, and subunits of histone deacetylase H4 peptide in the mORC1 BAH domain complex is similar to and remodeling complexes MTA1 and MTA2 (Callebaut that of the N-terminal tail of histone H4 in the NCP-Sir3 BAH et al., 1999). domain complex. However, the H4 peptide is more than 5 A˚ The structure of DNMT1 shows that the two DNMT1 BAH apart from the histone H4 tail when the two BAH domains are domains have a conserved b-sheet core (Song et al., 2011, aligned (Figure 5D). Also, with respect to the histone H4 tail 2012; Takeshita et al., 2011) (Figures 2 and 6). The two BAH in the NCP complex, the H4K20me2 peptide is shifted domains are oriented similarly, and are connected by an laterally by two–three residues towards the N-terminal a-helix, forming a dumbbell-shaped configuration (Figure 6). direction, e.g. H4K20me2 binds in a BAH domain region Like the BAH domain of mORC1, an a-helix immediately near where H4H18 would occupy in the NCP-Sir3 complex N-terminal to b9 is missing in the first BAH domain of (Figure 5D). The H4K20me2 peptide complex shows that DNMT1, BAH1. In place of the a-helix is a long, solvent H4K16 is not bound in a spatially confined region as in the exposed and partially disordered loop. The N-terminal end of Sir3 BAH domain. This observation may explain that the BAH1 is connected to the DNA-binding CXXC domain via BAH domain of yOrc1 binds NCP regardless of H4K16 a long loop that traverses the entire span of the protein. b10 acetylation, assuming that the BAH domain of yOrc1 binds borders the catalytic domain, and together with b9 and b8of NCP’s histone H4 tail in a fashion similar to that observed the BAH domain forms a contiguous b-sheet with the 218 N. Yang & R.-M. Xu Crit Rev Biochem Mol Biol, 2013; 48(3): 211–221

Figure 6. Structure of DNMT1 in complex with DNA. The b-sheet cores of the two DNMT1 BAH domains are shown in green ?and cyan, respectively, and their H-domain insertions are colored magenta. A schematic diagram showing the domain organization and the color code of the structure diagram is shown at the top. (see colour version of this figure online at www.informahealthcare/ bbm).

catalytic domain b-strands (Figure 6). The H-domain of Gregory et al., 2007; Tanaka et al., 2007; An et al., 2011). BAH1 latches onto the methylase domain though an edge C-terminal to their catalytic SET domains have a Bromo helix of the quasi-Rossmann fold catalytic domain. A novel domain, a PHD domain and a BAH domain (Figure 1). At feature of BAH2 is a long insertion loop between b4 and b5 present, the molecular functions of the BAH domain of Ash1 that projects out from the body of the BAH domain, and its proteins remain unknown. distal end interact with the target recognition domain (TRD), which is involved in DNA binding. A loop connecting the C- BAH domains in chromatin remodelers terminal end of BAH2 to the methylase domain is disordered. The precise roles of the BAH domains cannot be inferred Eukaryotic chromatin is dynamically remodeled by ATP- from the structure of the DNMT1-DNA complex. dependent nucleosome-remodeling complexes in fundamental Nevertheless, the two domains form large accessible surface biological processes such as gene transcription and DNA area of DNMT1; it is conceivable that they are involved in repair (Clapier & Cairns, 2009; Morrison & Shen, 2009; protein–protein interactions with partners of DNMT1. BAH1 Hargreaves & Crabtree, 2011). In the budding yeast, an is closely juxtaposed with the methylase domain and a abundant multisubunit protein complex, the RSC complex, is stretch of its surface area is positively charged and borders the essential for growth and cell-cycle progression (Cairns et al., DNA-binding region, making it possible for BAH1 to contact 1996, 1999). It shares many subunits with the SWI/SNIF DNA or nucleosome in a chromatin setting. It is interesting nucleosome-remodeling complex important for transcrip- to note that four of the six residues forming the methyllysine- tional regulation. The RSC complex has been shown to play binding pocket in mORC1 are conserved in BAH1 of an important role in DNA repair, besides being implicated in DNMT1, leaving the possibility that BAH1 may be able to transcriptional regulation (Chai et al., 2005; Shim et al., bind a methyllysine ligand (Figures 5C and 7C). 2005). Further analyses showed that two BAH domain- Small and homeotic disks protein 1 (Ash1) are absent in containing subunits of the RSC complex, RSC1 and RSC2, Drosophila and its mammalian homolog ASH1L are H3K36 are part of two distinct RSC complexes that share rest of the methylases (KMT2H), although other lysine targets have also protein subunits and have similar but distinguishable func- been reported (Beisel et al., 2002; Byrd & Shearn, 2003; tions (Cairns et al., 1999; Chambers et al., 2012). Both DOI: 10.3109/10409238.2012.742035 BAH domains 219

Figure 7. Potential methyllysine pockets in the BAH domains of Polybromo and DNMT1. (A) The H4K20me2-binding pocket of mORC1. (B) and (C) The H4K20me2 peptide in the mORC1 complex is superimposed onto the first BAH domain of chicken gPB (gPB–BAH1) and the first BAH domain of DNMT1, respectively. gPB and DNMT1 residues correspond to the H4K20me2-binding residues of mORC1 are shown in a stick representation. (see colour version of this figure online at www.informahealthcare/bbm). complexes participate in DNA repair, but only RSC1 is et al., 1998; Xue et al., 1998; Zhang et al., 1998). However, implicated in nucleosome sliding at DNA double-strand nothing is known about the functions of the BAH domains break sites. The BAH domain of RSC1 appears to be of MTA1 and MTA2. Given the known examples of BAH associated with the nucleosome-sliding activity, as a partial domains in protein–protein and protein–nucleosome interac- nucleosome-remodeling activity of RSC2 is detected tions, it would not be surprising if the BAH domains of MTA1 when its BAH domain is replaced with that of RSC1 and MTA2 are found to be involved in the assembly of (Chambers et al., 2012). However, the molecular bases for the NURD complex or contacting the nucleosome. the functions of RSC1 and RSC2 BAH domains, particularly Clearly, much effort is needed to further understand the those concerning whether and how they interact with the functions of MTA proteins, their BAH domains, in particular, nucleosome, remain elusive. in the protein complex that couples histone deacetylation with The PBAF (Polybromo, Brg1-Associated Factors) com- nucleosome remodeling. plex, originally known as SWI/SNF-B, shares core compo- nents with the mammalian SWI/SNF homolog, the BAF Conclusions complex, but differs by the presence of polybromo-1 (PB1), and the more recently discovered BAF200 subunit (Yan et al., Over the last decade or so, the BAH domain has risen from 2005; Thompson, 2009; Hargreaves & Crabtree, 2011). Like an obscure amino acid sequence motif to a structurally chicken polybromo, PB1 contains six tandem bromo domains, defined protein domain that features prominent roles in two adjacent BAH domains and an HMG box. Based on the chromatin biology. Structural studies have played important presence of BAH domains in PB1, the PBAF complex has roles in the advancement of mechanistic understandings of been suggested to be the mammalian homolog of RSC the molecular functions of the BAH domains. A converging complexes. The precise functions of the PB1 BAH domains theme appears to be that the conserved b-sheet core serves a are presently unknown, although it has been speculated that scaffold of the protein domain, while the loops connecting the BAH domains might mediate interactions with BAF200, the b-strands and the H-domain insertion endow specific which is required to anchor PB1 within the PBAF complex protein–protein interaction properties. It is to be expected (Yan et al., 2005; Thompson, 2009). An alignment of the that more structural and functional analyses of BAH BAH domain structures of gPB and mORC1 shows that domains will be forthcoming, and the development should the methyllysine-binding aromatic cage found in the BAH further enlighten the functions of BAH domain-containing domain of mORC1 is preserved in BAH1 of gPB, and the proteins in chromatin biology. corresponding residues are conserved in mammalian PB1 proteins (Figure 7). This observation raises a further Declaration of interest speculation that the BAH1 domain of PB1 may be involved in the recognition of methylated histone tails. This is an This work was supported by grants from the Ministry attractive scenario if true, as the Bromo and BAH domains of Science and Technology (2009CB825501 and of PB1 in combination may sense both histone acetylation 2010CB944903 to R.M. Xu and 2012CB910702 to and methylation marks. N. Yang), the Natural Science Foundation of China Finally, BAH domain-containing proteins MTA1 and (90919029 and 3 098 801) and Chinese Academy of MTA2 are subunits of NURD, a histone deacetylase and Sciences (CAS). R.M. Xu holds a CAS-Novo Nordisk Great nucleosome-remodeling complex (Tong et al., 1998; Wade Wall Professorship. 220 N. Yang & R.-M. Xu Crit Rev Biochem Mol Biol, 2013; 48(3): 211–221 References Goodwin GH, Nicolas RH. (2001). The BAH domain, polybromo and the RSC chromatin remodelling complex. Gene 268:1–7. An S, Yeo KJ, Jeon YH, Song JJ. (2011). Crystal structure of the human Gregory GD, Vakoc CR, Rozovskaia T, et al. (2007). Mammalian histone methyltransferase ASH1L catalytic domain and its implica- ASH1L is a histone methyltransferase that occupies the transcribed tions for the regulatory mechanism. J Biol Chem 286:8369–74. region of active genes. Mol Cell Biol 27:8466–79. et al Armache KJ, Garlick JD, Canzio D, . (2011). Structural basis of Hargreaves DC, Crabtree GR. (2011). ATP-dependent chromatin silencing: Sir3 BAH domain in complex with a nucleosome at 3.0 A remodeling: Genetics, genomics and mechanisms. Cell Res 21:396– resolution. Science 334:977–82. 420. Beisel C, Imhof A, Greene J, et al. (2002). Histone methylation by the Hossain M, Stillman B. (2012). Meier–Gorlin syndrome mutations Drosophila epigenetic transcriptional regulator Ash1. Nature 419:857–62. disrupt an Orc1 CDK inhibitory domain and cause centrosome reduplication. Genes Dev 26:1797–810. Bell SP, Kobayashi R, Stillman B. (1993). Yeast origin recognition complex functions in transcription silencing and DNA replication. Hou Z, Bernstein DA, Fox CA, Keck JL. (2005). Structural basis of the Science 262:1844–9. Sir1-origin recognition complex interaction in transcriptional silencing. Proc Natl Acad Sci USA 102:8489–94. Bell SP, Mitchell J, Leber J, et al. (1995). The multidomain structure of Orc1p reveals similarity to regulators of DNA replication and Hou Z, Danzer JR, Fox CA, Keck JL. (2006). Structure of the Sir3 transcriptional silencing. Cell 83:563–8. protein bromo adjacent homology (BAH) domain from S. cerevisiae at 1.95 A resolution. Protein Sci 15:1182–6. Bell SP, Stillman B. (1992). ATP-dependent recognition of eukaryotic origins of DNA replication by a multiprotein complex. Nature Hsu HC, Stillman B, Xu RM. (2005). Structural basis for origin 357:128–34. recognition complex 1 protein-silence information regulator 1 protein Bicknell LS, Bongers EM, Leitch A, et al. (2011a). Mutations in the interaction in epigenetic silencing. Proc Natl Acad Sci USA pre-replication complex cause Meier-Gorlin syndrome. Nat Genet 102:8519–24. 43:356–9. Jenuwein T, Allis CD. (2001). Translating the histone code. Science Bicknell LS, Walker S, Klingseisen A, et al. (2011b). Mutations in 293:1074–80. ORC1, encoding the largest subunit of the origin recognition complex, Kuo AJ, Song J, Cheung P, et al. (2012). The BAH domain of ORC1 cause microcephalic primordial dwarfism resembling Meier-Gorlin links H4K20me2 to DNA replication licensing and Meier–Gorlin syndrome. Nat Genet 43:350–5. syndrome. Nature 484:115–19. Bose ME, McConnell KH, Gardner-Aukema KA, et al. (2004). The Li G, Reinberg D. (2011). Chromatin higher-order structures and gene origin recognition complex and Sir4 protein recruit Sir1p to yeast regulation. Curr Opin Genet Dev 21:175–86. silent chromatin through independent interactions requiring a common Lipford JR, Bell SP. (2001). Nucleosomes positioned by ORC facilitate Sir1p domain. Mol Cell Biol 24:774–86. the initiation of DNA replication. Mol Cell 7:21–30. Botuyan MV, Lee J, Ward IM, et al. (2006). Structural basis for the Loo S, Fox CA, Rine J, et al. (1995). The origin recognition complex in methylation state-specific recognition of histone H4-K20 by 53BP1 silencing, cell cycle progression, and DNA replication. Mol Biol Cell and Crb2 in DNA repair. Cell 127:1361–73. 6:741–56. Byrd KN, Shearn A. (2003). ASH1, a Drosophila trithorax group Luger K, Mader AW, Richmond RK, et al. (1997). Crystal structure of protein, is required for methylation of lysine 4 residues on histone H3. the nucleosome core particle at 2.8 A resolution. Nature 389:251–60. Proc Natl Acad Sci USA 100:11535–40. Morrison AJ, Shen X. (2009). Chromatin remodelling beyond transcrip- Cairns BR, Lorch Y, Li Y, et al. (1996). RSC, an essential, abundant tion: the INO80 and SWR1 complexes. Nat Rev Mol Cell Biol chromatin-remodeling complex. Cell 87:1249–60. 10:373–84. et al Cairns BR, Schlichter A, Erdjument-Bromage H, . (1999). Two Muller P, Park S, Shor E, et al. (2010). The conserved bromo-adjacent functionally distinct forms of the RSC nucleosome-remodeling homology domain of yeast Orc1 functions in the selection of DNA complex, containing essential AT hook, BAH, and bromodomains. replication origins within chromatin. Genes Dev 24:1418–33. Mol Cell 4:715–23. Nicolas RH, Goodwin GH. (1996). Molecular cloning of polybromo, a Callebaut I, Courvalin JC, Mornon JP. (1999). The BAH (bromo- adjacent homology) domain: A link between DNA methylation, nuclear protein containing multiple domains including five bromodo- replication and transcriptional regulation. FEBS Lett 446:189–93. mains, a truncated HMG-box, and two repeats of a novel domain. Gene 175:233–40. Chai B, Huang J, Cairns BR, Laurent BC. (2005). Distinct roles for the RSC and Swi/Snf ATP-dependent chromatin remodelers in DNA Noguchi K, Vassilev A, Ghosh S, et al. (2006). The BAH domain double-strand break repair. Genes Dev 19:1656–61. facilitates the ability of human Orc1 protein to activate replication origins in vivo. EMBO J 25:5372–82. Chambers AL, Brownlee PM, Durley SC, et al. (2012). The two different isoforms of the RSC chromatin remodeling complex play distinct roles Oliver AW, Jones SA, Roe SM, et al. (2005). Crystal structure of in DNA damage responses. PLoS One 7:e32016. the proximal BAH domain of the polybromo protein. Biochem J 389:657–64. Chien CT, Buck S, Sternglanz R, Shore D. (1993). Targeting of SIR1 protein establishes transcriptional silencing at HM loci and telomeres Onishi M, Liou GG, Buchberger JR, et al. (2007). Role of the conserved in yeast. Cell 75:531–41. Sir3-BAH domain in nucleosome binding and silent chromatin Clapier CR, Cairns BR. (2009). The biology of chromatin remodeling assembly. Mol Cell 28:1015–28. complexes. Annu Rev Biochem 78:273–304. Pak DT, Pflumm M, Chesnokov I, et al. (1997). Association of the origin Connelly JJ, Yuan P, Hsu HC, et al. (2006). Structure and function recognition complex with heterochromatin and HP1 in higher of the Saccharomyces cerevisiae Sir3 BAH domain. Mol Cell Biol eukaryotes. Cell 91:311–23. 26:3256–65. Rusche LN, Kirchmaier AL, Rine J. (2003). The establishment, Eaton ML, Galani K, Kang S, et al. (2010). Conserved nucleosome inheritance, and function of silenced chromatin in Saccharomyces positioning defines replication origins. Genes Dev 24:748–53. cerevisiae. Annu Rev Biochem 72:481–516. Fox CA, Ehrenhofer-Murray AE, Loo S, Rine J. (1997). The origin Sampath V, Yuan P, Wang IX, et al. (2009). Mutational analysis of the recognition complex, SIR1, and the S phase requirement for silencing. Sir3 BAH domain reveals multiple points of interaction with Science 276:1547–51. nucleosomes. Mol Cell Biol 29:2532–45. Fox CA, Loo S, Dillin A, Rine J. (1995). The origin recognition complex Shim EY, Ma JL, Oum JH, et al. (2005). The yeast chromatin remodeler has essential functions in transcriptional silencing and chromosomal RSC complex facilitates end joining repair of DNA double-strand replication. Genes Dev 9:911–24. breaks. Mol Cell Biol 25:3934–44. Gardner KA, Rine J, Fox CA. (1999). A region of the Sir1 protein Song J, Rechkoblit O, Bestor TH, Patel DJ. (2011). Structure of dedicated to recognition of a silencer and required for interaction with DNMT1-DNA complex reveals a role for autoinhibition in main- the Orc1 protein in Saccharomyces cerevisiae. Genetics 151:31–44. tenance DNA methylation. Science 331:1036–40. DOI: 10.3109/10409238.2012.742035 BAH domains 221 Song J, Teplova M, Ishibe-Murakami S, Patel DJ. (2012). Structure- Wang W, Cote J, Xue Y, et al. (1996a). Purification and biochemical based mechanistic insights into DNMT1-mediated maintenance DNA heterogeneity of the mammalian SWI-SNF complex. EMBO J methylation. Science 335:709–12. 15:5370–82. Takeshita K, Suetake I, Yamashita E, et al. (2011). Structural insight into Wang W, Xue Y, Zhou S, et al. (1996b). Diversity and specialization of maintenance methylation by mouse DNA methyltransferase 1 mammalian SWI/SNF complexes. Genes Dev 10:2117–30. (Dnmt1). Proc Natl Acad Sci USA 108:9055–9. Wang X, Connelly JJ, Wang CL, Sternglanz R. (2004). Importance of the Tanaka Y, Katagiri Z, Kawahashi K, et al. (2007). Trithorax-group Sir3 N terminus and its acetylation for yeast transcriptional silencing. protein ASH1 methylates histone H3 lysine 36. Gene 397:161–8. Genetics 168:547–51. Thompson M. (2009). Polybromo-1: The chromatin targeting subunit of Xue Y, Wong J, Moreno GT, et al. (1998). NURD, a novel complex with the PBAF complex. Biochimie 91:309–19. both ATP-dependent chromatin-remodeling and histone deacetylase Tong JK, Hassig CA, Schnitzler GR, et al. (1998). Chromatin activities. Mol Cell 2:851–61. deacetylation by an ATP-dependent nucleosome remodelling com- Yan Z, Cui K, Murray DM, et al. (2005). PBAF chromatin-remodeling plex. Nature 395:917–21. complex requires a novel specificity subunit, BAF200, to regulate Triolo T, Sternglanz R. (1996). Role of interactions between the origin expression of selective interferon-responsive genes. Genes Dev recognition complex and SIR1 in transcriptional silencing. Nature 19:1662–7. 381:251–3. Zhang Y, LeRoy G, Seelig HP, et al. (1998). The dermatomyositis- Vermeulen M, Eberl HC, Matarese F, et al. (2010). Quantitative specific autoantigen Mi2 is a component of a complex containing interaction proteomics and genome-wide profiling of epigenetic histone deacetylase and nucleosome remodeling activities. Cell histone marks and their readers. Cell 142:967–80. 95:279–89. Wade PA, Jones PL, Vermaak D, Wolffe AP. (1998). A multiple subunit Zhang Z, Hayashi MK, Merkel O, et al. (2002). Structure and function Mi-2 histone deacetylase from Xenopus laevis cofractionates with an of the BAH-containing domain of Orc1p in epigenetic silencing. associated Snf2 superfamily ATPase. Curr Biol 8:843–6. EMBO J 21:4600–11.