Table of contents

Protein components of the SWI/SNF complex

Purification and peptide sequencing of the subunits of the complex reveal beta-actin as well as a novel actin-related protein, BAF53. beta-actin and BAF53 are required for maximal ATPase activity of BRG1 and are also required with BRG1 for association of the complex with chromatin/matrix. The many cellular functions of actin are regulated by small actin-binding proteins such as profilin and cofilin that are in turn regulated by PIP2. If actin were a functional component of the BAF complex, one would predict that the BAF complex would interact with PIP2-responsive actin-regulatory proteins, such as profilin and cofilin. BAF complexes are shown to bind to profilin and cofilin. The BAF complex also specifically binds to DNAse I, which binds actin. Since profilin and cofilin bind to domains 1 and 3 on the barbed end of actin, these results indicate that these domains would be available for the BAF-associated actin to bind to actin-regulatory molecules. The observation that BRG1 is required for actin and BAF53 to associate with the BAF complex led immuniprecipitation experiments indicating that actin and BRG-1 are likely to directly interact. Both actin and BAF53 were found to coimmunoprecipitate with BRG-1. These experiments demonstrate that actin and BAF53 directly interact with BRG1 in the BAF complex in vitro (Zhao, 1998).

The yeast Saccharomyces cerevisiae contains two related chromatin-remodeling complexes, RSC and SWI/SNF, which are shown to share the actin-related proteins Arp7 and Arp9. Depending on the genetic background tested, arp7 delta and arp9 delta mutants are either inviable or show greatly impaired growth and Swi-/Snf- mutant phenotypes. Unlike swi/snf mutants, viable arp7 delta or arp9 delta mutants have an Spt- phenotype, suggesting that RSC affects transcription. Temperature-sensitive mutations in ARP7 and ARP9 were isolated, and the amino acid changes support the structural relationship of Arp7 and Arp9 to actin. However, site-directed mutations predicted to impair ATP binding or hydrolysis do not detectably affect Arp7 or Arp9 function. These results suggest that actin-related proteins perform important roles in chromatin-remodeling complexes by virtue of structural rather than enzymatic similarities to actin (Cairns, 1998).

Nuclear fractionation experiments reveal that the BAF complex is stably associated with the nuclear matrix in HeLa cells, which contain the BRG1-, BAF53-, and beta-actin-containing "complete" complex, but not in SW-13 cells, that contain a complex lacking BAF53, actin, and BRG1. Furthermore, in an in vitro assay, the "complete" complex from Jurkat cells binds stably to the nuclei of SW-13 cells, raising the possibility that stable association of the complex with the nucleus requires BRG1, BAF53, and beta-actin. To test this possibility, nuclear matrix/chromatin was prepared from SW-13 cells transiently transfected with HA-tagged BRG1. Double immunostaining determined whether the complex would be retained in the nuclear matrix/chromatin in the absence or presence of BRG1. Whole-cell staining demonstrates that BRG1 is detected only in transfected cells, while BAF57 (BAF57 is shared by all mammalian complexes and contains a high-mobility-group domain adjacent to a kinesin-like region), which is in the SW-13 complex, and BAF53, which is not in the SW-13 complex, are detected in the nucleus of every cell. However, the BAF57 and BAF53 signals are only detected in the nuclear matrix preparations where BRG1 is detected. It has been concluded that the assembly and stable association of the complex with the nuclear matrix/chromatin requires BRG1, BAF53, and beta-actin (Zhao, 1998)

A model is proposed for the regulation of the BAF complex by membrane signaling. Signals from the lymphocyte antigen receptor regulate chromatin remodeling by regulating PIP2 levels, which in turn control the association of the BAF complex with chromatin or some component of nuclear matrix. A likely role for PIP2 is to control the actions of actin in the BAF complex by regulating the function of a nuclear actin-binding protein. PIP2 is known to directly regulate actin function by displacing actin-binding proteins from actin. For most actin-binding proteins this would require that domains 1 and 3 of actin be exposed in the BAF complex, which is the case since both profilin and cofilin bind to actin within the complex. The role of actin in targeting the complex is supported by the observation that BRG1, BAF53, and actin are required for association of the BAF complex with the nuclear matrix/chromatin in SW-13 cells, which have a complex that lacks these three subunits. This model requires that PIP2 and PI4-phosphate 5-kinase, which is the enzyme that responds to rho family GTPases and regulates PIP2 synthesis, be present within the nucleus. This is in fact the case, and a large proportion of both PIP2 and PI 4-phosphate-5 kinase are found to be present in the nucleus. Furthermore, the beta-isoenzyme of phospholipase C, which hydrolyzes PIP2, is found on the nuclear scaffold or matrix. Although changes in PIP2 in response to lymphocyte signaling have not been studied, when Friend cells are induced to differentiate, nuclear PLCbeta activity is reduced, accompanied by a significant increase in nuclear PIP2. The increase in nuclear PIP2 would be expected to displace actin-binding proteins from actin in the BAF complex, freeing the complex and allowing its association with chromatin. While substantial data support each of these steps, definitive evidence for this model will require purification of the proposed nuclear PIP2-responsive nuclear actin regulatory protein(s) and biochemical add back experiments using purified BAF complexes. The observation that the mSWI/SNF or BAF complex becomes stably associated with chromatin within 10 min of antigen receptor stimulation strongly indicates that neither protein synthesis nor new gene activation is required for targeting the complexes, since no detectable increase in transcription or protein synthesis is seen before this time. At present, little is known of how external stimuli lead to rapid chromatin decondensation. The finding that BAF complexes are capable of chromatin remodeling in vitro, contain actin and actin-related proteins, and are rapidly targeted to chromatin and the nuclear scaffold or matrix by a PIP2-dependent mechanism, defines a direct interface between chromatin regulation and signal transduction. While actin is a likely candidate to mediate the targeting of BAF complexes to chromatin/nuclear matrix, additional studies will be essential to demonstrate this directly (Zhao, 1998).

Lymphocyte activation is accompanied by visible changes in chromatin structure. A group of proteins tightly associated with BRG1 and hBRM called BAFs (BRG or Brm associated factors) are related to subunits of the yeast SWI/SNF complex. Antigen receptor signaling induces the rapid association of the BAF complex with chromatin. Phosphatidyl inositol 4,5-bisphosphate (PIP2), which is regulated by activation stimuli, is sufficient in vitro to target the BAF complex to chromatin, but it has no effect on related chromatin remodeling complexes containing SNF2L or hISWI. Addition of PIP2, one of the major mediators of signaling in lymphocytes, induces the nuclear association of BAF complexes. PIP2 is more effective than phosphatidylinositol 4-phosphate (PIP). Thus, the ability of phosphoinositols to control the association of BAFs with the nucleus parallels the activity of these signaling intermediates to regulate actin-binding proteins (Zhao, 1998).

p300 and the closely related CREB binding protein (CBP) are transcriptional adaptors that are present in intracellular complexes with TATA binding protein (TBP) and bind to upstream activators including p53 and nuclear hormone receptors. They have intrinsic and associated histone acetyltransferase activity, suggesting that chromatin modification is an essential part of their role in regulating transcription. Detailed characterization of a panel of antibodies raised against p300/CBP has revealed the existence of a 270-kDa cellular protein, p270, distinct from p300 and CBP but sharing at least two independent epitopes with p300. The subset of p300/CBP-derived antibodies that cross-reacts with p270 consistently coprecipitates a series of cellular proteins with relative molecular masses ranging from 44 to 190 kDa. Purification and analysis of various proteins in this group reveals that they are components of the human SWI/SNF complex and that p270 is an integral member of this complex (Dallas, 1998).

Chromatin structure plays a crucial regulatory role in the control of gene expression. In eukaryotic nuclei, enzymatic complexes can alter this structure by both targeted covalent modification and ATP-dependent chromatin remodeling. Modification of histone amino termini by acetyltransferases and deacetylases correlates with transcriptional activation and repression, cell growth, and tumorigenesis. Chromatin-remodeling enzymes of the Snf2 superfamily use ATP hydrolysis to restructure nucleosomes and chromatin, events that correlate with activation of transcription. A multi-subunit complex was purified from Xenopus laevis eggs that contain six putative subunits, including the known deacetylase subunits Rpd3 (see Drosophila Rpd3) and RbAp48/p46, as well as substoichiometric quantities of the deacetylase-associated protein Sin3. In addition, one of the other components of the complex was identified as Mi-2, a Snf2 superfamily member previously identified as an autoantigen in the human connective tissue disease dermatomyositis. Mi-2's Drosophila homolog physically interacts with Drosophila Hunchback and is involved in Hunchback mediated repression (Kehle, 1998). Nucleosome-stimulated ATPase activity precisely copurifies with both histone deacetylase activity and the deacetylase enzyme complex. This association of a histone deacetylase with a Snf2 superfamily ATPase suggests a functional link between these two disparate classes of chromatin regulators (Wade, 1998).

SAGA, a recently described protein complex in Saccharomyces cerevisiae, is important for transcription in vivo and possesses histone acetylation function. Both biochemical and genetic analyses are reported for members of three classes of transcription regulatory factors contained within the SAGA complex. A correlation exists between the phenotypic severity of SAGA mutants and SAGA structural integrity. Specifically, null mutations in the Gcn5/Ada2/Ada3 or Spt3/Spt8 classes cause moderate phenotypes and subtle structural alterations, while mutations in a third subgroup, Spt7/Spt20, as well as Ada1, disrupt the complex and cause severe phenotypes. Interestingly, double mutants (gcn5Delta;spt3Delta and gcn5Delta;spt8Delta), which cause the loss of a member of each of the moderate classes, have severe phenotypes, similar to spt7Delta, spt20Delta, or ada1Delta mutants. In addition, biochemical functions suggested by the moderate phenotypic classes have been investigated. Normal nucleosomal acetylation by SAGA requires a specific Gcn5 domain, termed the bromodomain. Deletion of this domain also causes specific transcriptional defects at the HIS3 promoter in vivo. SAGA interacts with TBP, the TATA-binding protein, and this interaction requires Spt8 in vitro. Overall, these data demonstrate that SAGA harbors multiple, distinct transcription-related functions, including direct TBP interaction and nucleosomal histone acetylation. Loss of either of these causes slight impairment in vivo, but loss of both is highly detrimental to growth and transcription (Sterner, 1999).

Protein complexes of the SWI/SNF family remodel nucleosome structure in an ATP-dependent manner. Each complex contains between 8 and 15 subunits, several of which are highly conserved between yeast, Drosophila, and humans. An ATP-dependent chromatin remodeling complex has been reconstituted using a subset of conserved subunits. Unexpectedly, both BRG1 and hBRM, the ATPase subunits of human SWI/SNF complexes, are capable of remodeling mono-nucleosomes and nucleosomal arrays as purified proteins. The addition of INI1, BAF155, and BAF170 to BRG1 increases remodeling activity to a level comparable to that of the whole hSWI/SNF complex. These data define the functional core of the hSWI/SNF complex (Phelan, 1999).

Recruitment of SWI/SNF to promoters

Evidence has been provided for a functional interaction between HATs and SWI/SNF. Using immobilized template nucleosome assays, it has been shown that acetylation of nucleosomal array templates by either the SAGA or NuA4 HAT complex stabilizes SWI/SNF binding to promoter nucleosomes after the dissociation of the activator that recruited SWI/SNF to the template (Hassan, 2001). Thus, stable binding of SWI/SNF has been recapitulated on nucleosome templates in vitro and found to require acetylated histones (Hassan, 2001). These results are consistent with other evidence for functional interactions between acetylation and SWI/SNF. Gcn5, the catalytic subunit of SAGA, participates in the stabilization of SWI/SNF binding to promoters in vivo. Histone acetylation by Gcn5 has been shown to be followed by SWI/SNF recruitment during activation of the interferon-ß (INF-ß) promoter in vitro. Transactivation by RAR/RXR was found to require histone acetylation prior to SWI/SNF action. A transient histone hyperacetylation at the PHO8 promoter is required for nucleosome remodeling by SWI/SNF, and the CBP and P/CAF histone acetyltransferases as well as hBrm remodeling complex are recruited to stimulate transcription of the human alpha1 antitrypsin promoter (Hassan, 2002).

The role of acetylated histones in SWI/SNF promoter occupancy suggest a molecular basis for the functional links between SWI/SNF and SAGA and raises the possibility that the bromodomain of the Swi2/Snf2 subunit might play an important role in anchoring SWI/SNF to promoters. A number of chromatin-modifying complexes, including SWI/SNF and SAGA, contain highly conserved bromodomain(s). Bromodomains have been implicated as acetyl-lysine recognition modules. The phenotypic affect of the bromodomain varies in different proteins. For example, no phenotype for the bromodomain deletion in the Swi2/Snf2, Spt7, or Drosophila Brahma has been observed. In contrast, a bromodomain deletion in one of the human homologs of Swi2/Snf2, hBrm, results in the loss of nuclear localization and a decrease in protein stability. In addition, deletion of the Gcn5 bromodomain affects transcription to some extent. Yeast contains a SWI/SNF-related complex, RSC, which appears to play a more global role in the yeast life cycle, is more abundant than SWI/SNF, and contains subunits with multiple bromodomain motifs. Some of the bromodomains in the Sth1, Rsc1, or Rsc2 subunits in the yeast RSC complex are essential for cell viability. Deletion of the Sth1 bromodomain results in a conditional phenotype, whereas only one of the bromodomains in the other related RSC subunits (Rsc1 and Rsc2; the two forms of the RSC complex that contain either of these two different gene products) is essential for function (Hassan, 2002 and references therein).

The bromodomains of Swi2/Snf2 and Gcn5 were necessary for the stable occupancy of the SWI/SNF and the SAGA complexes, respectively, on acetylated promoter nucleosomes. This acetylation can be brought about by either the SAGA or NuA4 HAT complexes. In contrast, the Spt7 bromodomain in the SAGA complex is not necessary for the retention of SAGA on acetylated nucleosome arrays, but will anchor SAGA if it is swapped into Gcn5, indicating that specificity of bromodomain function is determined in part by the subunit it occupies. These data illustrate the selectivity and specificity of bromodomain-containing subunits in anchoring chromatin-modifying complexes on promoter nucleosomes. Synthetic phenotypes observed upon combining deletions of the Swi2/Snf2 bromodomain with deletions of the Gcn5 bromodomain or temperature-sensitive mutants in Tra1 suggest related roles of these bromodomains in recruitment and retention of these complexes in chromatin (Hassan, 2002).

One of the predictions of the histone code hypothesis is the existence of functional interactions between chromatin remodeling complexes, such as SWI/SNF, and histone acetylase complexes, such as GCN5. Recent studies have elucidated the temporal sequence in which these coactivators of transcription are recruited to promoters in vivo and how their enzymatic properties contribute to gene activation. The best-characterized example in mammals is provided by the human IFN-ß gene. The gene is switched on by three transcription factors (NF-kappaB, IRFs, and ATF-2/c-Jun), and an architectural protein [HMG I(Y)], all of which bind cooperatively to the nucleosome-free enhancer DNA to form an enhanceosome. The enhanceosome targets the modification and repositioning of a nucleosome that blocks the formation of a transcriptional preinitiation complex on the IFN-ß promoter. This is accomplished by the ordered recruitment of HATs, SWI/SNF, and basal transcription factors. Initially, the GCN5 HAT-containing complex is recruited, and it acetylates the nucleosome. This is followed by the recruitment of the CBP-PolII holoenzyme complex. Next, the SWI/SNF remodeling machine arrives at the promoter via bivalent interactions with CBP and the acetylated histone N tails. SWI/SNF alters the structure of the nucleosome via an unknown mechanism, thus allowing recruitment and DNA binding of TFIID to the TATA box. The DNA bending induced upon TFIID binding to the promoter causes sliding of the SWI/SNF-modified nucleosome to a new position 36 bp downstream, thus allowing the initiation of transcription. This ordered recruitment and nucleosome sliding is consistent with the view that histone acetylation sets the stage for ATP-dependent remodeling by establishing a recognition surface for the bromodomains present in SWI/SNF-like remodeling machines. Furthermore, since histone acetylation precedes the recruitment of additional complexes bearing bromodomains, such as TFIID, it is possible that this modification also regulates recruitment of TFIID to promoters (Agalioti, 2002).

Experiments were carried out to test the histone code hypothesis. Only a small subset of lysines in histones H4 and H3 are acetylated in vivo during viral infection, and this modification is carried out by the GCN5 transcriptional coactivator complex. Reconstitution of recombinant nucleosomes bearing mutations in these lysine residues reveals a biochemical cascade for gene activation via a point-by-point interpretation of the histone code through the ordered recruitment of bromodomain transcription complexes. More specifically, acetylation of H4 lysine 8 is required for recruitment of the SWI/SNF complex, whereas acetylation of lysines 9 and 14 in histone H3 is critical for the recruitment of the general transcription factor TFIID. Thus, the information contained in the DNA address of the enhancer is extended (transferred) to the histone N termini by generating novel adhesive surfaces that participate in the recruitment of transcription complexes (Agalioti, 2002).

The precision by which the histone code is decoded is remarkable. Most likely, the code is translated via specific interactions of bromodomains with the acetylated histone N termini. The bromodomain in BRG1 associates with the H4 tail acetylated at K8, whereas the double bromodomain in TAFII250 binds to the doubly acetylated (at K9 and K14) H3 tail. The competition assays using either acetylated histone N termini peptides or bromodomain polypeptides as competitors revealed an unprecedented level of specificity for the interactions between bromodomains and acetyl-lysine histone N termini. Again, this remarkable degree of specificity may be dictated by the conformational changes forced upon these complexes by their interactions with other proteins in the complex. Thus, the point-by-point interpretation of the histone acetylation code may rely on the precise allosteric changes induced in many proteins upon their association with transcription factor complexes. For example, the initial recruitment of SWI/SNF via its association with CBP is stabilized on the promoter through the association of the BRG1 bromodomain with the H4 K8 acetylated tail. Although, BRG1's bromodomain could interact at least in principle with other acetylated lysine residues on H3, these interactions may not be of sufficient strength to ensure stable binding of the SWI/SNF complex to the promoter. Similarly, recruitment of TFIID to the SWI/SNF-modified promoter is stabilized via two types of interactions. The first with various enhanceosome components and the second with the interaction between the two bromodomains of TAFII250 and the two acetyl groups on residues K9 and K14 of histone H3. Several observations support this notion: (1) both TBP and TAFII250 are simultaneously recruited to the promoter with almost identical kinetics in vivo; (2) recruitment of TFIID in vivo occurs only when both H3 K9 and K14 are acetylated; (3) mutations in either K9 or K14 abrogate TFIID recruitment. The data show that the TAFII250 double bromodomain, when recruited to the natural IFN-ß promoter, interacts specifically with the H3 K9 and K14 acetylated residues and not with the acetylated H4 tails. However, when tested in isolation and out of the promoter/chromatin natural context, the double TAFII250 bromodomain interacts with similar affinities to both H4 and H3 acetylated tails, a result consistent with in vitro observations. It is proposed that the inordinate set of interactions occurring with purified bromodomains and acetylated histone tails is 'fixed' when present in natural promoter/chromatin contexts. In the latter case, it is possible that the competing interactions cannot take place either because the alternative target is occupied (e.g., the H4 tail is already bound by SWI/SNF) or the specific three-dimensional conformation of the transcription complex does not permit these interactions to occur (Agalioti, 2002).

A model is presented that depicts the ordered biochemical cascade decoding the DNA and histone acetylation code during activation of human IFN-ß gene transcription following Sendai virus infection. It is thought that the promoter DNA code contains all the information for the assembly of the enhanceosome in response to virus infection. The enhanceosome that assembles at the promoter element recruits the GCN5 histone acetyltransferase. Subsequently GCN5 acetylates initially H4K8 and H3K9. An unknown kinase recruited by the enhanceosome phosphorylates H3 Ser 10, a prerequisite for H3K14 acetylation by GCN5. The histone code is subsequently translated by recruiting of additional components required for transcription. The bromodomain containing transcription complexes SWI/SNF and TFIID are recruited to the promoter via bivalent interactions between the enhanceosome and specifically acetylated histone N termini, and this subsequently stimulates transcription of the IFN-ß gene (Agalioti, 2002).

Eukaryotic transcription initiation requires the complex dynamics of hundreds of proteins, many of which are found in large multisubunit complexes. Recent experiments have suggested stepwise recruitment of preassembled complexes, including chromatin remodeling, general transcription factor, mediator, and polymerase complexes, in which the actual order of recruitment may vary for different promoters. How do these complexes access target sequences contained within tightly condensed chromatin? While chromatin remodeling activities may facilitate the accessibility of large transcription and polymerase complexes to promoters, it is not known how they themselves are targeted within condensed chromatin. Gene activation in the context of condensed chromatin does occur. A yeast acidic activator, Gal4, can overcome heterochromatin gene silencing in Drosophila (Ahmad, 2001), and the addition of LCRs (locus control regions) to transgenes overcomes position effect silencing, even within centromeric chromatin. In this study, the recruitment of HAT and SWI/SNF components was directly visualized after tethering the VP16 acidic activation domain within condensed chromatin. A recruitment delay of tens to hundreds of minutes for catalytic HAT subunits and SWI/SNF subunits, relative to other HAT and SWI/SNF components, suggests sequential recruitment/assembly of chromatin remodeling complexes within condensed chromatin (Memedula, 2003).

A simplified system was used to visualize in vivo recruitment of cofactors to specific transcription factors. A lac operator 256 copy direct repeat was inserted in an expression vector for DHFR (dihydrofolate reductase). Transfection into DHFR- cells allowed selection of stable transformants, which were then used for methotrexate gene amplification. The A03_1 stable clone contains an ~90 Mb, condensed, late-replicating amplified chromosome region, organized as transgene repeats that are ~400 kb in length and are spaced by flanking, coamplified genomic DNA that is estimated to be ~1000 kb in length. During most of interphase, this amplified chromosome region forms a compact mass that is ~1 üm in diameter. Targeting the strong transcriptional activator, VP16 acidic activation domain (AAD), to this condensed chromosome region via a lac repressor-VP16 AAD fusion leads to histone hyperacetylation, recruitment of histone acetyltransferases, and transcriptional activation, accompanied by a dramatic, large-scale chromatin decondensation (Memedula, 2003).

The results demonstrate varying kinetics for the accumulation of several different components of SWI/SNF remodeling and SAGA-containing HAT complexes at a condensed, interphase chromosome region. These results cannot be attributed to trivia peculiarities related to restricted reactivity of the particular antibodies used against the endogenous proteins for these studies for two reasons. (1) Despite independently raised antibodies, the results show significant internal consistency. Functionally similar proteins such as Brm and Brg1, whose yeast homolog Swi2 is known to bind directly to acidic activators, are recruited with similar early kinetics, while BAF155 and BAF170, known to interact directly in a complex when coexpressed, show strikingly similar but significantly later recruitment kinetics. Moreover, the catalytic HAT subunits GCN5, PCAF, and p300/CBP show quite similar recruitment kinetics, peaking very close to the peak of histone H3 acetylation. (2) Early recruitment of endogenous TRRAP, whose yeast homolog Tra1 has been shown to directly bind acidic activators, was verified independently by using transient transfection of a FLAG-tagged TRRAP construct, as was the late recruitment of TIP60 by using transient transfection of a FLAG-tagged TIP60 construct (Memedula, 2003).

The results imply that the targeting of protein components of these complexes can occur without them being in a single, obligatory large complex. Moreover, critical subunits, such as Brm and Brg1 for SWI/SNF and TRRAP for SAGA-like complexes, are recruited within several minutes in large amounts to condensed chromatin before significant recruitment of other complex components. Both Brm and Brg1 are known to be active as monomers in defined biochemical assays. It is speculated, therefore, that recruitment of either individual components or partial subcomplexes, which themselves produce some level of chromatin-modifying activity, provides a mechanism by which certain transcriptional activators can activate target genes within the context of condensed chromatin. Partial chromatin opening by these subunits might then allow accessibility and stable binding and/or localized assembly of full-size complexes, which provide a wider range of chromatin modifying and/or regulatory responses (Memedula, 2003).

TopBP1 (DNA topoisomerase IIbeta binding protein I) contains multiple BRCT domains and is involved in replication and the DNA damage checkpoint. Through its BRCT domain, TopBP1 interacts with and represses exclusively E2F1 but not other E2F factors. This regulation of E2F1 transcriptional activity is mediated by a pRb-independent, but Brg1/Brm-dependent mechanism. TopBP1 recruits Brg1/Brm, a central component of the SWI/SNF chromatin-remodeling complex, to E2F1-responsive promoters and represses the activities of E2F1, but not E2F2 or E2F3. This regulation is crucial in the control of E2F1-dependent apoptosis during normal cell growth and DNA damage. Interestingly, TopBP1 is induced by E2F and interacts with E2F1 during G1/S transition. Thus, TopBP1 functions as a critical modulator and serves as a negative feedback regulator of E2F1 by inhibiting E2F1-dependent apoptosis during G1/S transition as well as DNA damage to promote cell survival (Liu, 2004).

To identify E2F1-specific regulators, an E2F1-specific fragment (N terminus) was used as a bait and TopBP1 was isolated in a yeast two-hybrid screen. TopBP1 contains eight BRCA1 C-terminal (BRCT) motifs and interacts with several other proteins, including human papilloma virus type 16 (HPV16) transcription/replication factor E2, DNA polymerase epsilon, checkpoint protein hRad9, and Miz-1. It appears to be involved in DNA replication because incubation of an antibody against the sixth BRCT motif of TopBP1 inhibits DNA replication in an in vitro HeLa nuclei replication assay. TopBP1 is induced during DNA damage and is also involved in DNA damage checkpoint. Upon gamma-irradiation, TopBP1 colocalizes with Nbs1, BRCA1, and 53BP1 in the ionizing radiation-induced foci representing stalled replication forks. In addition to the control of DNA replication, TopBP1 is also required for cell survival. Inhibition of TopBP1 expression by antisense Morpholino oligomers induces apoptosis. Thus, TopBP1 is involved in several important aspects of growth control. So far, the detailed mechanism by which TopBP1 regulates these signaling events remains poorly understood (Liu, 2004 and references therein).

TopBP1 interacts with E2F1 through the sixth BRCT motif of TopBP1 and the N terminus of E2F1. This interaction is induced by ATM-mediated phosphorylation of E2F1 at Ser 31 during DNA damage. The interaction between BRCT domains and phosphopeptides is a general phenomenon. Through this interaction, the transcriptional and apoptotic activities of E2F1 are repressed, and E2F1 is recruited to DNA damage-induced foci. Moreover, the interaction between TopBP1 and E2F as well as the repression of E2F activity are specific to E2F1, but not seen in E2F2, E2F3, and E2F4, suggesting that TopBP1 is an E2F1-specific regulator (Liu, 2004).

E2F1 is shown to be regulated by a novel Retinoblastoma protein (pRb)-independent mechanism, in which TopBP1 recruits Brg1/Brm, a central subunit of the SWI/SNF chromatin-remodeling complex, to inhibit E2F1 transcriptional activity. This regulation is specific for E2F1 and is critical for the control of E2F1-dependent apoptosis during S phase and DNA damage. TopBP1 is induced by E2F and interacts with E2F1 during G1/S transition. Thus, E2F1 and TopBP1 form a feedback regulation to prevent apoptosis during DNA replication (Liu, 2004).

The mechanism by which TopBP1 represses E2F1 is through recruiting Brg1/Brm. Evidence to support this assertion: (1) TopBP1-mediated repression of E2F1 is defective in a Brg1/Brm-deficient cell line, C33A. The repression is restored by reconstitution with Brg1/Brm. (2) Dominant-negative mutants of Brg1 or Brm inhibit TopBP1 to repress E2F1. (3) TopBP1 interacts with Brg1/Brm and facilitates the interaction between E2F1 and Brg1/Brm. (4) TopBP1 recruits Brg1/Brm to E2F1-responsive promoters. (5) Dominant-negative mutants of Brg1/Brm derepress E2F1 activity during DNA damage. (6) Whereas TopBP1 siRNA induces E2F1-dependent apoptosis in HEK293 cells and wild-type MEFs, it fails to induce apoptosis in Brg1/Brm-deficient C33A cells (Liu, 2004).

SWI/SNF complex and DNA topology

Yeast SWI/SNF complex belongs to a family of enzymes that use the energy of ATP hydrolysis to remodel chromatin structure. The role of DNA topology in the mechanism of yeast SWI/SNF remodeling has been examined. The ability of yeast SWI/SNF to enhance accessibility of nucleosomal DNA is nearly eliminated when DNA topology is constrained in small circular nucleosomal arrays and this inhibition can be alleviated by topoisomerases. Remodeling of these substrates does not require dramatic histone octamer movements or displacement. These results suggest a model in which yeast SWI/SNF remodels nucleosomes by using the energy of ATP hydrolysis to drive local changes in DNA twist (Gavin, 2001).

Two distinct types of remodeling assays have demonstrated that chromatin remodeling by SWI/SNF-like enzymes involves changes in DNA topology. First, the incubation of circular chromatin templates with SWI/SNF and a topoisomerase decreases the number of negative supercoils constrained by nucleosomes. This reaction is not the result of nucleosome loss but must reflect either an unwrapping of nucleosomal DNA or a major change in the path of DNA around the histone octamer (a change in writhe). An ATP-dependent unwrapping of nucleosomal DNA is consistent with EM studies that indicate a loss of ~40 bp of DNA from remodeled nucleosomes. SWI/SNF remodeling may disrupt the 20 base pairs of DNA at both the exit and entry termini of nucleosomal DNA, since these DNA regions are less tightly bound to the histone octamer and are released first during low-energy transitions. Importantly, these putative 'unwrapped' nucleosomes are not equivalent to a fully remodeled nucleosome species, since they still harbor DNA that is inaccessible to restriction enzyme digestion. Thus, such species may represent either an intermediate or a stable byproduct of nucleosome remodeling (Gavin, 2001).

Whereas previous studies have shown that SWI/SNF action can lead to changes in DNA topology, current studies indicate that the ability of SWI/SNF to enhance nucleosomal DNA accessibility requires changes in DNA topology. Since SWI/SNF is not a topoisomerase and, thus, cannot directly change the linking number of DNA, SWI/SNF action must involve changes in the writhe or twist of nucleosomal DNA. One possibility is that SWI/SNF changes the path of DNA on the surface of the histone octamer or alters the orientation of the entry and exit angles of nucleosomal DNA. Alternatively, a model is favored in which SWI/SNF uses the energy of ATP hydrolysis to drive local changes in DNA twist that diffuse throughout the nucleosome and weaken both wraps of histone-DNA interactions. This reaction would lead to a nucleosome in which the histone octamer is inherently more mobile and DNA more accessible to nucleases and transcription factors. Since this accessible nucleosomal state requires continuous ATP hydrolysis, these data suggests that yeast SWI/SNF must continuously alter DNA twist in order to generate a remodeled state. Such continuous changes in DNA topology might have a high energy cost on topologically constrained minicircles, and, thus, remodeling would not be proficient on these chromatin templates. This type of model is supported by recent studies which demonstrate that SWI/SNF-like enzymes can use the energy of ATP hydrolysis to generate unconstrained superhelical torsion in DNA and nucleosomal substrates (Gavin, 2001).

This proposed model for ATP-dependent remodeling may be mechanistically related to how the bacterial DNA helicase, PcrA, couples ATP binding and hydrolysis to localized DNA helix destabilization. PcrA is a member of helicase superfamily I, and it contains the seven DNA helicase motifs that are also found in SWI2/SNF2 family members. Recent functional and structural studies of PcrA have suggested that this helicase has two complementary but distinct ATP-dependent functions: ATP-dependent ssDNA tracking activity and ATP-dependent DNA duplex destabilization. In the latter case, the binding of ATP induces PcrA to bind double-stranded DNA in such a way as to destabilize the DNA helix. Subsequent ATP hydrolysis leads to the translocation of the enzyme one nucleotide along the adjacent ssDNA tail and the release of the distorted DNA duplex. It is proposed that SWI/SNF may harbor only one of the two ATP-dependent activities of PcrA: ATP-dependent DNA duplex destabilization. Consistent with this view, the ATPase subunit of SWI/SNF, SWI2/SNF2, shares significant structural homology to the 2B domain of PcrA (crucial for ATP-dependent DNA duplex destabilization) but lacks homology to the domain of PcrA involved in ssDNA translocation (Gavin, 2001 and references therein).

SWI/SNF complex and the beta-globin gene

Erythroid Kruppel-like factor (EKLF) is necessary for stage-specific expression of the human beta-globin gene. EKLF has been shown to require a SWI/SNF-related chromatin remodeling complex, EKLF coactivator-remodeling complex 1 (E-RC1), to generate a DNase I hypersensitive, transcriptionally active beta-globin promoter on chromatin templates in vitro. To examine the nucleosomal structure of in vitro-assembled ß-globin promoters, indirect end-labeling analysis of DNase I digested chromatin was performed. Importantly, these digestions were carried out prior to transcription and thus reflect potentially active, rather than transcribing, promoters. A DNase I hypersensitive region from approximately -120 to +10 in the beta-globin promoter is observed in the presence of both EKLF and E-RC1. This pattern resembles the hypersensitive structure of the active beta-globin promoter in vivo, which maps from 100-150 bp 5' of the cap site. Neither EKLF nor E-RC1 are sufficient alone to generate an open chromatin configuration. Since most chromatin remodeling events are energy-dependent, it was asked whether hypersensitive site formation in the beta-globin promoter is dependent on ATP. Chromatin templates were incubated with the ATP-hydrolyzing reagent apyrase prior to addition of EKLF and E-RC1. Indeed, formation of the open beta-globin promoter structure by these proteins requires ATP. E-RC1 contains BRG1, BAF170, BAF155, and INI1 (BAF47), all homologs of yeast SWI/SNF subunits, as well as a subunit unique to higher eukaryotes, BAF57, which is critical for chromatin remodeling and transcription with EKLF. E-RC1 displays functional selectivity toward transcription factors, since it cannot activate expression of chromatin-assembled HIV-1 templates with the E box-binding protein TFE-3. Thus, a member of the SWI/SNF family acts directly in transcriptional activation and may regulate subsets of genes by selectively interacting with specific DNA-binding proteins (Armstrong, 1998).

SWI/SNF complexes in yeast and higher eukaryotes are thought to facilitate gene activation and transcription factor binding by disrupting repressive chromatin structures. Little is known, however, about how these complexes target specific genes for activation. A specialized SWI/SNF-related complex (PYR complex) has been purified from murine erythroleukemia (MEL) cell nuclear extract. PYR complex binds pyrimidine-rich elements at the human and murine beta-globin loci. PYR complex DNA-binding activity is restricted to definitive hematopoietic cells and is both DNA sequence- and length-dependent. Mass spectrometric identification of purified peptides and antibody supershift assays indicate that PYR complex contains at least four known mammalian SWI/SNF subunits: BAF57 (a high-mobility-group [HMG] domain protein with an adjacent kinesin-like region); INI1 (a SNF5 homolog); BAF60a (a novel protein), and BAF170 (a SWI3 homolog). PYR complex broadly footprints a 250-bp pyrimidine-rich element between the human fetal and adult beta-globin genes. A short intergenic deletion that removes this element from a human globin locus cosmid construct results in delayed human fetal-to-adult globin gene switching, presumably by opening the locus in the region of the adult genes to permit the binding of beta-globin transcriptional activators (O'Neill, 1999).

The SWI/SNF family of chromatin-remodeling complexes plays a key role in facilitating the binding of specific transcription factors to nucleosomal DNA in diverse organisms from yeast to man. Yet the process by which SWI/SNF and other chromatin-remodeling complexes activate specific subsets of genes is poorly understood. Mammalian SWI/SNF regulates transcription from chromatin-assembled genes in a factor-specific manner in vitro. The DNA-binding domains (DBDs) of several zinc finger proteins, including EKLF, interact directly with SWI/SNF to generate DNase I hypersensitivity within the chromatin-assembled ß-globin promoter. Interestingly, two SWI/SNF subunits (BRG1 and BAF155) are necessary and sufficient for targeted chromatin remodeling and transcriptional activation by EKLF in vitro. Remodeling is achieved with only the BRG1-BAF155 minimal complex and the EKLF zinc finger DBD, whereas transcription requires, in addition, an activation domain. In contrast, the BRG1-BAF155 complex does not interact or function with two unrelated transcription factors, TFE3 and NF-kappaB. It is concluded that specific domains of certain transcription factors differentially target SWI/SNF complexes to chromatin in a gene-selective manner and that individual SWI/SNF subunits play unique roles in transcription factor-directed nucleosome remodeling (Kadam, 2000).

Interaction of HP1 with Brahma homologs

Mammalian heterochromatin protein 1 (HP1) alpha, HP1ß and HP1gamma are closely related non-histone chromosomal proteins that function in gene silencing, presumably by organizing higher order chromatin structures. It has been shown by co-immunoprecipitation that HP1alpha, but neither HP1ß nor HP1gamma, forms a complex with the BRG1 chromatin-remodeling factor in HeLa cells. In vitro, BRG1 interacts directly and preferentially with HP1alpha. The region conferring this preferential binding has been mapped to residues 106-180 of the HP1alpha C-terminal chromoshadow domain. Using site-directed mutagenesis, three amino acid residues I113, A114 and C133 have been identified in HP1alpha (K, P and S in HP1ß and HP1gamma) that are essential for the selective interaction of HP1alpha with BRG1. Interestingly, these residues were also shown to be critical for the silencing activity of HP1alpha. Taken together, these results demonstrate that mammalian HP1 proteins are biochemically distinct and suggest an entirely novel function for BRG1 in modulating HP1alpha-containing heterochromatic structures (Nielsen, 2002).

There are several mechanisms by which BRG1 could contribute to HP1alpha-dependent silencing. Reconstitution studies have shown that, on its own, BRG1 can remodel nucleosomes in an ATP-dependent manner, indicating that the presence of this ATPase subunit within a complex is sufficient for chromatin remodeling. HP1alpha can bind to nucleosomes in vitro and associates with chromatin in vivo through a direct interaction with the histone fold domain of histone H3. Thus, the chromatin-remodeling activity of BRG1 may facilitate the binding of HP1alpha to its nucleosomal sites. Recent studies have also described a specific binding of HP1alpha to the methylated tail domain of H3, which is critical for its targeting to centromeric heterochromatin. Similarly, the underacetylated state of the histone tails is an important determinant for the maintenance of the HP1alpha protein at heterochromatic sites and for full silencing. Thus, the BRG1 activity may enhance access of the histone tails to HP1alpha-associated deacetylases and methyltransferases, which may, in turn, promote the formation of a local, heterochromatin environment that results in effective gene silencing. Another way in which BRG1 might participate in the assembly of HP1alpha-dependent heterochromatic structures is to alter nucleosome spacing. In general, sequences with repressive chromatin domains are packaged into highly regular nucleosome arrays, the regularity of which correlates with gene silencing. Thus, it is tempting to speculate that BRG1 may contribute to HP1alpha-mediated silencing by manipulating nucleosomal spacing (Nielsen, 2002).

Interaction of zinc finger transcription factors with Brahma homologs

Mammalian SWI/SNF chromatin remodeling complexes are involved in critical aspects of cellular growth and genomic stability. Each complex contains one of two highly homologous ATPases, BRG1 and BRM, yet little is known about their specialized functions. BRG1 and BRM are shown to associate with different promoters during cellular proliferation and differentiation, and in response to specific signaling pathways by preferential interaction with certain classes of transcription factors. BRG1 binds to zinc finger proteins through a unique N-terminal domain that is not present in BRM. BRM interacts with two ankyrin repeat proteins that are critical components of Notch signal transduction. Thus, BRG1 and BRM complexes may direct distinct cellular processes by recruitment to specific promoters through protein-protein interactions that are unique to each ATPase (Kadam, 2003).

SWI/SNF interacts with zinc finger proteins (ZFP)s through the ZF DNA-binding domain (DBD) and the BRG1 ATPase. The basis for the observed specificity between ZFPs and BRG1 complexes is that interaction occurs within a domain of BRG1 that is nonhomologous with BRM. The role of individual ZFs within two structural motifs, C2H2 and C4, was investigated in mediating BRG1 SWI/SNF function. Using the erythroid factors EKLF and GATA-1 as representative proteins that contain C2H2 and C4 domains, respectively, these studies demonstrate that BRG1 binds to individual ZFs that are the most critical for DNA binding. This may seem paradoxical; however, ZF DBDs have been shown to associate with both RNA and protein. The EKLF and GATA-1 DBDs interact with a variety of cofactors, often through specific ZFs. The significance of such critical protein-protein interactions, including that of SWI/SNF, occurring through domains that must also bind DNA has yet to be elucidated. The functional relationship between BRG1-containing SWI/SNF and the ZF DBDs of EKLF or GATA-1 may pertain to other members of these transcription factor families which contain conserved DBDs but also contain highly divergent activation domains which contribute to their specialized functions in gene regulation (Kadam, 2003).

The BRM ATPase is expressed at high levels in differentiating cells, yet the functional role of this protein and the identity of the genes it regulates are poorly understood. In this regard, the observation that two components of the Notch signaling pathway, CBF-1 and ICD22 (the intracellular domain of Notch), strongly associate with BRM but not BRG1 is especially intriguing. This pathway controls cell fate commitment in a broad range of developmental processes. CBF-1 recruits BRM to two natural target genes, Hes1 and Hes5, in myoblasts before Notch induction. This indicates that these promoters are already in a remodeled configuration and accessible to bind the activator, Notch2, upon signaling (Kadam, 2003).

Table of contents

brahma: Biological Overview | Regulation | Developmental Biology | Effects of Mutation | References

Home page: The Interactive Fly © 1995, 1996 Thomas B. Brody, Ph.D.

The Interactive Fly resides on the
Society for Developmental Biology's Web server.