SWI/SNF Chromatin Remodeling Complex
There are several large multi-subunit complexes that couple ATP hydrolysis with regulation of the chromatin landscape and are referred to as ATP-dependent chromatin remodelers. These complexes are primarily divided into four major classes based on the domain organization of their catalytic subunit. In this entry we focus on one of those major classes called the SWI/SNF subfamily. The SWI/SNF subfamily of chromatin remodelers are well conserved throughout all eukaryotes and typically the catalytic subunit has at least four signature motifs that are the ATPase, bromo, AT-hook, and HSA domains. The ATPase domain has sequence homology with the two lobes of such ATP-dependent DNA translocases like RecA and Rad54. The DNA translocation activity of SWI/SNF is essential for these complexes to move nucleosomes along DNA. The bromo domain binds to acetylated lysines in histone tails which may modulate the activity of the remodeler. While the A/T hook domain in other complexes is used to bind to DNA, the role of the A/T hook in SWI/SNF is not well understood. The HSA domain is not unique to SWI/SNF complexes, but is found in other chromatin remodelers and is generally used to recruit actin or actin-related proteins (Arp) to the catalytic subunit. The Arps have been shown to stimulate the catalytic activity and to be an important part of a core complex. The subunit composition of SWI/SNF subfamily members ranges from 12 to 17 subunits and can be quite varied in humans. In yeast and Drosophila there are two SWI/SNF complexes present, while in humans the number is significantly higher. The two yeast complexes called SWI/SNF and RSC have different catalytic subunits, but share the same two Arps (Arp7 and 9). In Drosophila there is only one catalytic subunit called Brahma or Brm which is assembled into two different complexes that differ primarily by a subunit which contains multiple bromo domains and is called polybromo. In yeast the RSC complex also differs from SWI/SNF in that besides the catalytic subunit there are seven bromodomains spread throughout three different subunits. Human SWI/SNF in some ways is a combination of both yeast and Drosophila. There are two catalytic subunits in humans called BRG1 and hBRM, like yeast SWI/SNF, which are assembled into distinct complexes. However, like Drosophila, in humans there is also an ortholog of the polybromo subunit called BAF180 that is assembled with BRG1 or hBRM. In addition there are different forms of several of the other subunits which are respectively assembled into different SWI/SNF complexes which add to the diversity of human SWI/SNF. Many of the human SWI/SNF complexes are tissue or developmentally specific which lends itself well to the many different combinations of SWI/SNF that can be formed using different subunits. In this entry we focus on the various roles of SWI/SNF in regulating processes that involve DNA. Much of this work has come from studies in yeast where the functional role of SWI/SNF can more readily be studied.
Involvement in Transcription
Chromatin dynamics for transcription regulation involve the intricate interplay of ATP-dependent chromatin remodelers, ATP-independent histone modifiers, and trans-acting activator/coactivator or repressor/corepressor proteins. However, the temporal control of these factors presents a chicken-and-egg situation. Some promoters may be initially remodeled by chromatin remodelers to create accessible sites for activator binding which in turn can recruit other factors promoting transcription. In other promoters, activator proteins can first bind to promoter regions and recruit chromatin remodelers. Alternatively, activators may recruit histone modifiers like histone acetyl transferases (HATs) that put on active acetyl marks that are in turn recognized by remodeler subunits bearing special “modification-identifying” domains. In some other cases, activators may initially bind to exposed low-affinity sites, recruit remodelers which in turn expose more high-affinity sites. Repressors may function similarly, except that they interact with remodelers that evenly space nucleosomes to form a repressive chromatin structure. The repressed state of chromatin can also be achieved by recruitment of histone deacetylases (HDACs). Besides a normal recruitment establishment and maintenance function, activator/repressor proteins might directly modify the catalytic activities of the remodeler complexes.
Genome-wide gene expression studies with budding yeast SWI/SNF mutants Δswi2 and Δswi1 showed that they affected ∼6% of genes in Saccharomyces cerevisiae (Sudarsanam et al. 2000). In Schizosaccharomyces Pombe, however, Δsnf2 and Δsnf5 mutants showed that they affected 2.6% of the genome but a different set of genes compared to S. cerevisiae (Monahan et al. 2008). In budding yeast, about 50% of the affected genes were downregulated in the mutants suggesting that SWI/SNF is involved equally in transcription activation and repression. Δswi2 and Δswi1 affected an overlapping set of genes suggesting they function as a complex. Given the spatial location of genes that were affected, it was evident that SWI/SNF affects individual promoters rather than chromosomal domains. More direct evidence of SWI/SNF’s role in transcription activation was provided by biochemical studies on individual promoters like SUC2, PHO5, PHO8, ARG1, RNR3, SNZ1 and INO1 among others (Ford et al. 2007; Ransom et al. 2009; Tomar et al. 2009; Kim et al. 2005). Chromatin immunoprecipitation (ChIP) studies provided evidence of the direct binding of SWI/SNF to these promoters. Swi2, Snf5, and Swi1 were shown to directly interact with acidic activators like Gcn4, Hap4, and Gal4 that help in targeting of the complex to promoters (Neely et al. 2002). In humans, BRG/BRM complex can associate with type I and type II nuclear receptors such as glucocorticoid receptor (GR), estrogen receptor (ER), androgen receptor (AR), retinoic acid receptor (RAR), vitamin D3 receptor (VDR), or peroxisome proliferator-activated receptor δ (PPARδ) in a ligand-dependent or independent manner to activate hormone responsive genes (Hsiao et al. 2003). Nuclease sensitivity assays, primer extension, UV cross-linking, and indirect end labeling showed SWI/SNF dependent nucleosome repositioning concomitant with transcript appearance at several genes.
The direct role of SWI/SNF in repression of yeast genes has been less well characterized. SER3 a serine biosynthesis gene was initially implicated as a target of SWI/SNF mediated repression (Martens and Winston 2003). However, it was later found to be an indirect effect of an intergenic transcript from an adjacent SRG1 promoter (Martens and Winston 2003). SWI/SNF has also been shown to interact with Hir1/Hir2 repressors, but this interaction activates transcription at the HTA1-HTB1 genes (Prochasson et al. 2005). SWI/SNF represses heat shock genes under heat stress but its effect is less pronounced (Shivaswamy and Iyer 2008). In S. pombe, Snf22 physically associates with fip1+, frp1+, and str3+ promoters and mediates the repression of hexose transport and iron transport genes (Monahan et al. 2008).
In humans, involvement of BRG/BRM complexes in repression is more studied and has important implications in tumor suppression. Knock-down, deletion, or mutations in BRG1, Brm, or hSNF5/Ini1 components produce malignancy (Medina and Sanchez-Cespedes 2008). BRG1/Brm is recruited by tumor suppressors such as prohibitin and retinoblastoma (Rb) protein to mediate repression of E2F-responsive genes (Wang et al. 2004; Lorch et al. 2001). BRG1 forms a repressor complex with Rb to inhibit cyclin A, D1, and E (Giacinti and Giordano 2006; Zhang et al. 2000). BRG1 represses the expression of c-fos proto-oncogene in an Rb-dependent but E2F or GR independent manner (Simpson et al. 1993). BRG1 and Brm associate with Rb to repress ribonucleotide reductase, thymidylate synthase, and dihydrofolate reductase genes by the recruitment of histone deacetylase mSin3B (Gunawardena et al. 2007).
In addition to tumor-suppressor interaction, BRG1 also associates with several corepressor complexes. Interaction with CoREST recruits BRG1 to neuronal genes to assist in REST mediated repression in conjunction with other deacetylases (Battaglioli et al. 2002). BRG1 components were also shown to interact with NCoR corepressor complex (Underhill et al. 2000). Recently, BRG1 was shown to directly interact with HP1a suggesting the involvement of SWI/SNF components in heterochromatin formation (Lavigne et al. 2009; Nielsen et al. 2002).
RSC conditional mutants affected 12% of the yeast genome in contrast to 6% in case of SWI/SNF. Rsc4 bromodomain conditional mutants downregulated 47–111 genes and upregulated 123–194 genes suggesting an important role of RSC in transcription activation and repression respectively. Another rsc4-Δ4 mutant (four amino acids from the C-terminus deleted) downregulated 168 genes but upregulated 338 (Kasten et al. 2004; Soutourina et al. 2006). Microarray on Δrsc30 showed RSC’s role in ribosomal gene repression (Angus-Hill et al. 2001). In support, rsc4-2 bromodomain mutant affected a large number of ribosomal and histone genes (Kasten et al. 2004). More direct evidence of RSC involvement in transcription activation comes from DNaseI accessibility data suggesting a repressive chromatin structure at DUT1 and SMX3 genes (Soutourina et al. 2006). Direct evidence of RSC-mediated repression of CHA1 was shown by nucleosome positioning in indirect labeling experiments suggesting an open chromatin structure (Moreira and Holmberg 1999). At the HTA1-HTB1 genes, RSC plays an antagonistic role to SWI/SNF by repressing transcription even though both chromatin remodeling complexes are recruited by the same Hir1/Hir2 repressor (Prochasson et al. 2005; Ng et al. 2002).
Unlike SWI/SNF which affects only polII genes, loss of RSC reduces transcription from polI, polII, and polIII genes as shown by genome-wide Sth1 localization and nucleosome occupancy studies (Ng et al. 2002; Parnell et al. 2008). This could partly be due to the association of Rsc4 with Rpb5, a subunit shared by all three RNA polymerases (Soutourina et al. 2006). Importantly, loss of Sth1 increases nucleosome density at polIII promoters (particularly tRNA genes) while polII promoters show more subtle single nucleosome changes that include gain or movement. This study has important implications on promoter-specific activity of RSC – nucleosome eviction at polIII vs. sliding at polII (Ng et al. 2002).
Several subunits of the SWI/SNF complex in yeast show genetic interactions with proteins involved in transcription elongation. Prominent among these, Swp29 genetically interacts with TFIIS elongation factor while Swi2, Snf5, and Snf6 with Spt16, a subunit of the elongation complex FACT (Shogren-Knaak et al. 2006; Malone et al. 1991). Drosophila Brahma, Moira, and Osa suppress the rough eye phenotype of asf1 mutants and physically interact with ASF1, another elongation factor implicated in redeposition of histones in the coding regions of genes (Moshkin et al. 2002). Human Brm is found in CD44 coding regions where it regulates alternate splicing (Batsche et al. 2006). Similarly, BRG1 is associated with HSP70 promoter and coding regions following heat shock (Corey et al. 2003). Several yeast SWI/SNF subunits are also localized to coding regions of inducible genes such as GAL1/GAL10, HSP82, HSP104, MET2 and MET6 under conditions of active transcription, traveling with the elongating RNA polymerase (Schwabish and Struhl 2007). The RSC complex is recruited to acetylated nucleosomal templates and helps in elongation by promoting the passage of stalled PolII. PolII passage through chromatin involves loss of a H2A/H2B dimer forming hexasomes. Complexes promoting dimer loss like SWI/SNF and RSC therefore promote polII passage during elongation. However, in an in vitro transcription assay, RSC was more potent than SWI/SNF in promoting polII passage across nucleosomal barriers (Carey et al. 2006). It has been proposed that ATP-dependent chromatin remodelers may also interact with histone chaperones to facilitate polII passage (Park and Luger 2008).
Transcription memory is the ability of daughter cells to remember the transcription activity of certain genes in the mother. This promotes the efficient “switching on” or “switching off” of genes in the daughter to quickly establish a “mother-like transcription status.” In this regard, histone modifications (particularly methylation) of promoter regions have been shown to play a role in recruiting transcription factors that activate or repress transcription (Francis et al. 2004). Recently, SWI/SNF has been implicated as being involved in establishing transcription memory of GAL1 genes in daughter cells. Daughter cells show fast kinetics of galactose reinduction following up to 4 h of repression with glucose. Normally, cells that have never been exposed to galactose take ∼20 min to induce the GAL1 gene peaking at 1 h. Daughter cells however take <10 min and are able to show this fast kinetics for one round of cell division suggesting epigenetic memory is transient. Transcriptional memory was not dependent on histone modifications, but an ISWI deletion in a Δswi2 background reinforced the fast reinduction kinetics. This suggested that SWI/SNF establishes transcriptional memory at GAL1 by antagonizing ISWI action (Kundu et al. 2007).
SWI/SNF and Histone Acetylation Coordinate to Control Chromatin Structure and Function
Current studies indicate that chromatin is dynamic rather than a passive structural scaffold. The chromatin modification enzymes incorporate post-translational modifications on the histone proteins to create docking sites for specific chromatin associated proteins containing specialized domains. Some of these modifications are temporary while others may persist through cell divisions to confer stable epigenetic memory. This principle of histone marking by covalent modifications forms the basis of a “histone code” (Strahl and Allis 2000).
The yeast SWI/SNF and RSC complexes contain one or several bromodomain(s), which are also found in HAT complexes like SAGA and the general transcription factor TAFII250 (Jacobson et al. 2000). Bromodomains are 110 residue protein motifs that specifically interact with acetylated lysine residues in histones as well as in non-histone proteins (Dhalluin et al. 1999). While, SWI/SNF possesses a single bromodomain motif residing in the Swi2/Snf2 subunit, 8 out of the 15 bromodomains in yeast are found in RSC. The ATPase motor subunit, Sth1, has only one bromodomain right at its C-terminus much like Swi2/Snf2. The Rsc1, Rsc2, and Rsc4 subunits each have two bromodomains (Kasten et al. 2004; Cairns et al. 1999). The two bromodomains in RSC4 are essential for cell viability and are located adjacent to each other to form the Rsc4 tandem bromodomain. The eighth bromodomain of RSC resides in its RSC10 subunit.
The presence of bromodomain(s) in RSC and SWI/SNF suggests that recognition of histone acetylation might play a major role in chromatin targeting and/or in their chromatin remodeling function. The single bromodomain in Swi2/Snf2 is dispensable for SWI/SNF activity and has little or no phenotypic effect upon its own deletion (Hassan et al. 2002). Likewise, the bromodomain in Gcn5, the catalytic subunit of SAGA and ADA HAT complexes, also does not play a significant role as is evident from the modest phenotypes observed in Gcn5 bromodomain deletion mutants (Hassan et al. 2002). The bromodomain in Sth1, on the other hand, is required for wild-type function, and deletion mutants lacking different portions of the Sth1 bromodomain are thermosensitive and arrest with highly elongated buds (Du et al. 1998). However, in combination with Gcn5 bromodomain deletion or Tra1 (a targeting subunit in SAGA) ts mutation bromodomain deleted SWI/SNF showed strong phenotypes including a synthetic growth defect on raffinose suggesting complimentary functions of SAGA and SWI/SNF in vivo (Hassan et al. 2002). These results are indeed consistent with other evidence for a functional interplay between histone acetylation and SWI/SNF. For example, prior remodeling by SWI/SNF is required for the recruitment of Gcn5 during late mitosis when chromatin is condensed (Krebs et al. 2000); conversely, the Gcn5 bromodomain is required to stabilize SWI/SNF on promoter nucleosomes (Syntichaki et al. 2000). Studies of several promoters induced during differentiation and development revealed that SWI/SNF and SAGA complexes do function jointly, but the temporal order of these complexes can vary. For instance in the case of yeast HO gene, expression that occurs at the end of mitosis absolutely required SWI/SNF action for Gcn5 recruitment while activation of the human interferon beta promoter involves a very different order of events wherein histone acetylation by Gcn5 is followed by SWI/SNF recruitment (Cosma et al. 1999; Agalioti et al. 2002). Transactivation by RAR/RXR heterodimers is found to require histone acetylation prior to SWI/SNF action (Dilworth et al. 2000). Similarly, at the PHO8 promoter histone acetylation is a prerequisite for nucleosome remodeling by SWI/SNF (Reinke et al. 2001). The human a1 antitrypsin promoter is unique in the sense that PIC or pre-initiation complex assembly occurs long before transcription activation by hBrm and two HAT complexes CBP and P/CAF are recruited simultaneously after PIC assembly (Soutoglou and Talianidis 2002).
Analogous to SWI/SNF, a series of elaborate genetic experiments established cooperativity between RSC and SAGA complexes. Deletions of SAGA encoding genes like spt20Δ (believed to abolish all SAGA function), gcn5Δ (required for SAGA’s histone acetyl transferase activity), and spt3Δ (required for a HAT independent activity of SAGA) when combined with rsc1Δ or rsc2Δ mutations produced the same pattern of synthetic phenotypes as observed with Gcn5 and Swi2/Snf2 double bromodomain deletion mutants (Cairns et al. 1999). Genetic evidence for a functional link between RSC and histone acetylation was further provided when RSC4’s tandem bromodomain ts mutation in combination with Gcn5 deletion or with histone H3K14 mutation turned out to be lethal suggesting that histone H3K14 acetylation is critical for Rsc4 function (Kasten et al. 2004).
Recently, using immobilized nucleosome arrays containing a block of four Gal4 binding sites it is shown that prior acetylation of the array with SAGA and NuA4 HAT complexes does not enhance SWI/SNF binding per se, but does allow retention of the SWI/SNF complex even after dissociation of the Gal4-VP16 activator that recruited SWI/SNF to the acetylated nucleosomal arrays. In a subsequent publication, they showed that retention of SWI/SNF on acetylated chromatin templates following removal of the transcription activator required the Swi2/Snf2 bromodomain. They validated the role of Swi2/Snf2 bromodomain in anchoring SWI/SNF to target loci in vivo by showing that the Swi2/Snf2 bromodomain contributes to SWI/SNF occupancy at the SUC2 promoter (Hassan et al. 2002). In addition to histone tail acetylation, H3-K56 acetylation in the structured globular region of the core histone has been also shown to facilitate SWI/SNF recruitment most likely via the disruption of histone-DNA interactions near the entry-exit site of the nucleosome (Xu et al. 2005). An independent experiment by Carey et al. demonstrated that SAGA acetylated nucleosomes also stimulate binding of RSC under in vitro conditions (Carey et al. 2006).
Targeting of chromatin regulators to particular locations in the genome occurs either through interactions with sequence-specific DNA binding transcription factors or by using specialized protein motifs or domains that recognize specific post-translational modifications on histone tails. While, not yet demonstrated in vitro, since histone acetylation stabilizes SWI/SNF binding to nucleosomes, in principle SWI/SNF could be recruited to chromatin via bromodomain dependent interaction with acetylated histone tails (Carey et al. 2006; Hassan et al. 2001, 2002). Particularly in the case of RSC based on the preponderance of bromodomains contained within its subunits one might envision that the multiple bromodomains might cumulatively increase the affinity for acetylated histone tails and provide efficient targeting of RSC to its substrates. Alternatively, the individual bromodomains within the RSC complex might not be redundant but may interact with different targets, acetylated histones and non-histone proteins, to localize RSC to different nuclear compartments or to regulate its remodeling activity. This may also explain why some of the bromodomains in RSC are essential and others are not. Further supporting the notion that the bromodomains in RSC have distinct binding partners, a recent finding showed that while one (BD#2) of the two essential tandem bromodomains in RSC4 is involved in histone H3-K14 acetylation recognition, the other (BD#1) participates in the auto-regulation of RSC activity by specifically interacting with acetylated Rsc4-K25 (VanDemark et al. 2007). This mechanism might have evolved to regulate the residence time of RSC at sites of remodeling which is critical for RSC to cater to the needs of ∼700 physiological targets in the yeast genome. That individual bromodomains within the RSC complex might not be redundant but could have special functional features is further exemplified by the fact that only one of the two bromodomains (BD#2) of RSC1 and RSC2 is essential for cell viability (Cairns et al. 1999).
Does acetylation induced stabilization of SWI/SNF binding only facilitate recruitment of these complexes or such stabilization also contributes to their catalytic activity is another relevant question that has begun to be addressed recently. Biochemical data from Hassan et al. and Chandy et al. showing the requirement of the Swi2/Snf2 bromodomain for octamer transfer and remodeling of SAGA acetylated nucleosomes suggest that histone acetylation can affect the functional activity of SWI/SNF complexes by acting as binding epitopes for recruitment (Hassan et al. 2006). Thus in vivo we see that Pho5 and co-regulated Pho8 gene activation requires transient acetylation of promoter nucleosomes by SAGA and only the nucleosomes acetylated by SAGA become marked for displacement by SWI/SNF (Reinke and Horz 2003; Barbaric et al. 2003). Additional evidence in supporting that histone acetylation affects chromatin remodeling function of SWI/SNF complexes is provided by Ferreira et al. who demonstrated that RSC catalyzed remodeling is enhanced several fold upon histone H3 tail acetylation via lowering of Km for acetylated nucleosomes (Ferreira et al. 2007).
Involvement in Double Strand Break Repair and Genome Stability
The maintenance of a stable and unaltered genome is crucial to cell survival and faithful propagation of genetic information to its progeny. Genome stability is threatened by numerous exogenous (ionizing radiation, radiomimetic agents) and endogenous (recombination, transposition, mating type switching, chromosome segregation) factors that introduce mutations and chromosomal rearrangements. A cell bearing an unstable genome is prone to growth arrest, premature ageing, apoptosis, and even malignant transformation. Instability is initiated among others by double strand breaks (DSBs) which are by far the most deleterious outcomes of genome-damaging agents (Paques and Haber 1999). Suppressors of DSBs include a class of trans acting factors that encompass replication, repair, and checkpoint proteins. A new addition to this class of factors has been ATP-dependent chromatin remodelers like INO80, SWR1, RSC, and SWI/SNF (van Attikum et al. 2007; Chai et al. 2005). SWI/SNF and RSC mutants (Δsnf2, Δsnf5, sth1, sfh1) show sensitivity to DNA damaging agents like hydroxyurea, bleomycin, MMS, and Δ-irradiation (Chai et al. 2005; Koyama et al. 2002; Bennett et al. 2001). A more direct role has been established by their recruitment to DSBs by ChIP analysis albeit with different kinetics. RSC is bound to induced DSBs at the MAT locus in a localized region (0.2 Kb surrounding the break) within 10 min whereas SWI/SNF is recruited by 40 min with increased binding for over 4 h suggesting a role late in the repair process (Chai et al. 2005).
Cells have adopted two highly regulated damage repair pathways to counter DSBs. The homologous recombination repair (HRR) pathway makes use of long stretches of homologous regions of the sister chromatids that act as templates for faithful replication. Nonhomologous end joining (NHEJ) is more error-prone and joins together the broken ends of DNA by recognizing microhomologies of 1–5 bp. An investigation into the repair pathway affected in SWI/SNF mutants revealed that they were able to rejoin cut plasmids through NHEJ but were unable to perform HRR. HRR proceeds via five steps: (1) 5′-3′ resection, (2) strand invasion, (3) new DNA synthesis, (4) unwinding from template and annealing, and (5) ligation (Peterson and Cote 2004). In SWI/SNF mutants, specifically the strand invasion step during homology search was affected and mutants decreased Rad51 and Rad52 association at HMLa (homologous donor template for MAT locus repair) (Chai et al. 2005). Given the remodeling and nucleosome displacement properties of SWI/SNF, it can be speculated that SWI/SNF plays a late but essential role in remodeling the homologous donor template region for efficient synapsis during HRR.
In contrast to SWI/SNF, RSC has been speculated to have roles in both NHEJ and HRR. RSC recruitment to DSBs has been shown to be dependent on yKu70, Mre11, and Rsc30 subunit. In addition, pull-down assays and two-hybrid studies indicate Rsc1 and Rsc2 physically interact with yKu80 and Mre11 (Shim et al. 2005). The yKu70/80 heterodimer is the central component of NHEJ recognizing DNA ends and recruiting other NHEJ factors. Mre11 is a component of the MRX/MRN (Mre11, Rad50, Xrs2/Mre11, Rad50, Nbs1) trimeric complex that is one of the key sensors of damaged DNA. It exhibits 3′ to 5′ exonuclease and endonuclease activities that are stimulated in presence of Rad50 and aid in 3′ end resection during NHEJ and HRR. A Δrsc30Δrad52 and Δrsc8Δrad52 double mutant showed enhanced hypersensitivity to DNA damaging agents suggesting that these RSC subunits affect a pathway other than that affected by Rad52 (Shim et al. 2005). Rad52 is an essential component in HRR and acts as a cofactor for Rad51 recombinase helping it to overcome competition from Replication Protein A (RPA) to bring about efficient strand exchange. Thus Rsc30 and Rsc8 are required for NHEJ. In contrast to RSC30 and RSC8; STH1, SFH1, RSC1, and RSC2 mutants are all able to perform NHEJ as efficiently as WT but not HR suggesting differential requirement of RSC subunits in the two modes of DSBR (Chai et al. 2005). Despite its early recruitment to DSB sites, a Δrsc2 mutant is unable to perform the final ligation step in HRR pointing to an early and late role of RSC in DSBR (Chai et al. 2005).
Chromosomal segregation during mitosis and meiosis is yet another cellular event that maintains genomic integrity and stability. From the S phase to anaphase, sister chromatids are held together at the centromeres and arms by a four-subunit cohesin complex comprising two SCC (Sister Chromatid Cohesion) proteins Mcd1/Scc1 and Scc3 and two SMC (Structural Maintenance of Chromosomes) proteins Smc1 and Smc3. Cohesion at centromeres and chromosomal arms is temporally regulated and recently the RSC complex has been implicated in cohesin loading. RSC association with cohesin in the arms is cell-cycle regulated while its association at the centromere is constitutive (Huang et al. 2004). However, STH1 and RSC2 mutants impair RSC recruitment specifically at the arms and result in their precocious separation. There are two alternative models proposed to explain the role of RSC in chromosomal arm cohesion. One model suggests RSC-mediated cohesin loading is coordinated with DNA replication. In support, factors involved in replication, Ctf7/Eco1, Ctf18, and Ctf8, interact genetically with RSC (Baetz et al. 2004). In another model, chromosomal arm cohesin loading is coupled to DNA repair and recombination. Cohesin has been found to be involved in post-replicative repair and two subunits Smc1 and Smc3 are components of recombination and repair complex RC1 (Huang and Laurent 2004). It is postulated that RSC remodels chromatin at chromosomal arms and recruits cohesin to maintain sister chromatid cohesion. These sister chromatids on one hand serve as templates for proper recombination and repair events and on the other allow for processive DNA replication by assuming ring structures.
Involvement of RSC and SWI/SNF in Cell Cycle Control, Differentiation, and Development
The temporal sequence of events that takes place in a eukaryotic cell before it undergoes replication to form two daughter cells together comprises the cell cycle or cell-division cycle. A typical cell cycle can be divided into two brief periods: interphase and mitotic or M phase. The interphase can be again divided into three sub-phases: gap1 or G1, synthesis or S, and gap2 or G2 phases while the mitotic M phase can be subdivided into prophase, metaphase, anaphase, and telophase.
As cell cycle controls DNA synthesis and cell proliferation, so the progression of cell cycle is highly regulated with specific checkpoints at particular points in the cell cycle like at the G1/S and G2/M transition points guarded by cyclins and cyclin dependent kinases (CDKs) to monitor the progress of cell cycle. In budding yeast, the genes regulating cell cycle progression have been studied using temperature sensitive lethal mutants (cdc mutants) whose growth is arrested at specific points of the cell cycle at the restrictive temperature (Hartwell et al. 1974). These studies generated substantial evidence for an interconnection between chromatin remodeling and cell division cycle progression. Two such elegant genetic studies conducted in budding yeast demonstrated the requirement of two key subunits of RSC, Sth1 and Sfh1, for cell cycle progression through mitosis. Tsuchiya et al. showed that when Sth1, the Swi2/Snf2 counterpart in RSC, is depleted cells undergo an arrest and accumulate at the large bud stage with a single nucleus containing DNA of G2/M phase (Tsuchiya et al. 1992). Consistent with these results a temperature sensitive mutation in Sfh1, the Snf5 paralog in yeast, was found to induce cell cycle arrest at the G2/M phase (Cao et al. 1997). Both of these studies suggested a functional link between cell cycle control and RSC. Subsequent genetic studies conducted to investigate the actual role that RSC plays in cell cycle regulation isolated a temperature sensitive Sth1 mutant with impaired chromatin structure surrounding the centromeres (Tsuchiya et al. 1998). These results provide the molecular basis of the link between RSC and cell cycle control. The defective chromatin organization in absence of a functional RSC complex around centromeres might be abolishing kinetochore function which ultimately results in cell cycle arrest at G2/M transition. The yeast SWI/SNF complex on the other hand plays no role in the progression of cell division cycle through mitosis. In the contrary, SWI/SNF is required to efficiently exit from mitosis by mediating the expression of some late mitotic genes (Krebs et al. 2000). However, in flies, depletion of BAP (Brahma/BRM associated protein) complex, the Drosophila SWI/SNF complex, via RNAi mediated knockdown of its OSA subunit resulted in G2 to M arrest while cells lacking the RSC homolog PBAP (polybromo-containing BAP) complex showed no cell cycle defects (Moshkin et al. 2007). The Drosophila BAP complex regulates cell cycle progression by controlling the expression of STRING/ CDC25 which is required for entry into mitosis. This apparent contradiction from yeast to flies suggests for a functional shift in the course of evolution. In humans, both hBRM and BRG1 (BRM related gene 1) complexes participate in the regulation of cell cycle control by interacting with retinoblastoma proteins (Dunaief et al. 1994). These observations once more reinforces the notion that functional differences between yeast SWI/SNF complexes cannot be directly translated to higher eukaryotes.
The RSC complex also stimulates expression of early meiotic genes and sporulation and thereby facilitates developmental processes in budding yeast (Yukawa et al. 1999). There are numerous examples of involvement of the human SWI/SNF complexes in various developmental programs, such as muscle, heart, blood, skeletal, neuron, adipocite, liver, and T-cell development. The SWI/SNF class of chromatin remodeling complexes has instructive and reprogramming roles during development. Of particular importance are the findings that in vertebrates SWI/SNF complexes are combinatorially assembled for maintaining pluripotency and multipotency. Studies of the mammalian SWI/SNF complex, BAF (Brahma associated factor) indicate that a progressive change in its subunit composition underlies the transition from pluripotent embryonic stem cells (ESCs) to multipotent neuronal progenitors to post-mitotic differentiated neurons during nervous system development (Lessard et al. 2007). Mouse ESCs express a BRG1 containing esBAF complex which also contains BAF155 but not BAF170 (Ho et al. 2009). The esBAF complex is crucial for maintaining pluripotency and self-renewal of mouse ESCs (Ho et al. 2009). As ESCs differentiate into neuronal progenitors, the esBAF complex undergoes dynamic subunit exchanges incorporating BRM and BAF60C and eliminating BAF60B to form the neuronal-progenitor specific BAF (npBAF) complex (Lessard et al. 2007; Ho et al. 2009). BRG1 in npBAF complex ensures self-renewal of the neuronal progenitors and their differentiation into committed neurons (Lessard et al. 2007). The BAF45A subunit in npBAF induces proliferation of neuronal progenitors past their normal mitotic exit point (Lessard et al. 2007). As neuronal progenitors exit from mitosis, the BAF45A and BAF53A subunits in the npBAF complex are replaced by BAF45B and BAF53B subunits to form the neuron-specific BAF (nBAF) complex (Lessard et al. 2007). More recently, it has been shown that BAF45A depletion in post-mitotic neurons is microRNA (miR-9* and miR-124) mediated (Yoo et al. 2009). The nBAF complex is essential for post-mitotic neuronal function including activity-dependent dendritic outgrowth by interacting with CREST, a Ca2+ responsive dendritic regulator (Peterson et al. 2007).
The tissue specific BAF complexes have been also reported to play a role in cardiac fate determination during heart development in vertebrates (Lickert et al. 2004). BAF60C which is selectively expressed in regions of the mouse embryo that gives rise to a heart is required for heart morphogenesis, differentiation of cardiac and skeletal muscle cells (Lickert et al. 2004) and more importantly when injected into the non-cardiogenic regions of the developing embryo BAF60C in coordination with transcription factors GATA4 and TBX5 induces the development of beating cardiomycetes from non-cardiogenic mesoderm (Lickert et al. 2004). These data suggest the existence of a BAF60C containing specialized cBAF complex although till date such a cardiogenic complex could not be isolated. On the other hand, the highly homologous polybromo or BAF180 containing BAF (PBAF) complex which is expressed in the epicardium plays roles in mediating coronary development and cardiac chamber maturation (Huang et al. 2008).
The role of BAF complexes is also implicated in T-cell development in the thymus. However, there is no report of any specialized BAF complexes in the thymocytes. During T-cell differentiation two co-receptors of the T-cell antigen receptor, CD4 and CD8, are differentially expressed. Cd4 is developmentally repressed by a distantly located silencer ∼2 kb from the transcription start site of Cd4 (Sawada et al. 1994). At an early stage of development BAF complex binds to the distant silencer element via BRG1 and BAF57 and recruits a transcription factor RUNX1 which mediates Cd4 gene silencing (Wan et al. 2009). This data also suggests fundamental mechanistic differences between yeast SWI/SNF and mammalian BAF complexes as the former is known to regulate its target genes by binding to promoters. In mice, BRG1 is again required at a later stage to activate Cd4 expression indicating that BAF complexes can both activate and repress the same gene depending on the developmental context (Dey et al. 2003).
Although it remains poorly understood, BRG1 and BAF complexes might also be required for skeletal muscle differentiation as a dominant negative allele of Brg1 or RNAi mediated depletion of BAF60C appears to be detrimental for myogenic transcription factor (MYOD1 and MEF2D) action (Lickert et al. 2004; de la Serna et al. 2001).
Clearly, there are many ways that the SWI/SNF chromatin remodelers are involved in regulation of transcription, replication, recombination, and DNA repair. It is also evident that with the large number of SWI/SNF complexes that they are not particularly redundant in terms of function, but rather have very specialized roles in the cell. While much can be learned about the fundamental mechanistic properties of SWI/SNF by studying one form of these complexes in such organisms like yeast, it seems likely that there will be differences in their mode of operation which will be context dependent. Differences between various SWI/SNF complexes will be reflected in part how they are differentially recruited to specific genomic regions, but aside from this the data suggest that they will also probably remodel nucleosomes in distinctive ways. So far progress has been made on cataloging the SWI/SNF complexes involved in different developmental pathways, but still there is no data to indicate how the presence or absence of a particular subunit could alter the mode of remodeling that occurs.