Molecular Cell, Vol. 12, 983–990, October, 2003, Copyright 2003 by Cell Press

Targeting Activity Is Required for SWI/SNF Function In Vivo and Is Accomplished through Two Partially Redundant Activator-Interaction Domains Philippe Prochasson,1,4 Kristen E. Neely,3 Ahmed H. Hassan,2 Bing Li,1,4 and Jerry L. Workman1,4,* 1 Howard Hughes Medical Institute and Department of Biochemistry and Molecular Biology 306 Althouse Laboratory The Pennsylvania State University University Park, Pennsylvania 16802 2 Faculty of Medicine and Health Sciences Department of Biochemistry UAE University Al-Ain United Arab Emirates

Summary The SWI/SNF complex is required for the expression of many yeast genes. Previous studies have implicated DNA binding transcription activators in targeting SWI/ SNF to UASs and promoters. To determine how activators interact with the complex and to examine the importance of these interactions, relative to other potential targeting mechanisms, for SWI/SNF function, we sought to identify and mutate the activator-interaction domains in the complex. Here we show that the N-terminal domain of Snf5 and the second quarter of Swi1 are sites of activation domain contact. Deletion of both of these domains left the SWI/SNF complex intact but impaired its ability to bind activation domains. Importantly, while deletion of either domain alone had minor phenotypic effect, deletion of both resulted in strong SWI/SNF related phenotypes. Thus, two distinct activator-interaction domains play overlapping roles in the targeting activity of SWI/SNF, which is essential for its function in vivo. Introduction The yeast SWI/SNF complex is an ATP-dependent chromatin-remodeling complex that can mobilize nucleosomes for the activation or repression of a subset of yeast genes (Wang, 2003). SWI/SNF is able to bind to DNA and nucleosomes with high affinity but without DNA sequence specificity (Coˆte´ et al., 1998; Quinn et al., 1996). Thus, mechanisms by which SWI/SNF is targeted to specific genes have been a topic of considerable interest. Experiments thus far suggest three different mechanisms that might contribute to promoter recruitment of SWI/SNF. First, both yeast and human SWI/SNF have been found to associate with the RNA polymerase II holoenzyme, suggesting that SWI/SNF *Correspondence: [email protected] 3 Present address: Dana-Farber Cancer Institute, Harvard Medical School, Boston, Massachusetts 02115. 4 Present address: Stowers Institute of Medical Research, 1000 E. 50th street, Kansas City, Missouri 64110.

might be directly recruited to promoters via this interaction (Cho et al., 1998; Wilson et al., 1996). A second mechanism of SWI/SNF recruitment comes from several different studies that indicate the SWI/SNF complex can be recruited to target genes through direct interactions with sequence-specific transcription activators. Yeast SWI/SNF and its human homologs interact with sequence-specific transcription factors. The human SWI/SNF has been found to interact with C/EBP␤, the human heat shock factor 1 (hHSF1), the lymphoidspecific Ikaros activator, EKLF, p53, c-Myc, MyoD, and nuclear hormone receptors (i.e., glucocorticoid receptor, estrogen receptor) (Armstrong et al., 1998; Fryer and Archer, 1998; Sullivan et al., 2001; for review, see Hassan et al., 2001). Studies on the yeast SWI/SNF have shown that it can directly interact with acidic activators such as the herpes virus VP16 protein, the yeast Gcn4, Hap4, Gal4, Pho4, and Swi5 activators (Neely et al., 1999, 2002; Yudkovsky et al., 1999). Yeast and human SWI/SNF enhance transcription from nucleosomal templates in vitro upon targeting by an activator (Kadam et al., 2000; Neely et al., 1999; Wallberg et al., 2000). Thus, a growing body of evidence implicates sequence-specific transcription factors in targeting SWI/SNF to promoters. A third potential mechanism that could contribute to SWI/SNF recruitment is based on the fact that the Swi2/ Snf2 subunit contains a bromodomain (Haynes et al., 1992). Bromodomains bind acetyl-lysine within the aminoterminal tails of core histones (Dhalluin et al., 1999; Jacobson et al., 2000). The Swi2/Snf2 bromodomain contributes to stable binding of SWI/SNF to acetylated nucleosome arrays in vitro (Hassan et al., 2002). While not yet demonstrated in vitro, SWI/SNF could in principle be recruited by acetylated histones at promoter sequences. Thus, targeting HAT complexes like SAGA to promoters could lead to SWI/SNF targeting through histone acetylation. Such a mechanism is consistent with observed connections between the Gcn5 acetyltransferase and chromatin remodeling by SWI/SNF (Barbaric et al., 2001). To determine the role of different potential targeting mechanisms in vivo, it is necessary to generate mutants that affect each mechanism independently. The interactions of SWI/SNF with holo RNA polymerase II have not been characterized enough for such mutations to be designed. By contrast, potential targeting by the bromodomain interactions with acetylated histones can be addressed by deleting the Swi2/Snf2 bromodomain. While the Swi2/Snf2 bromodomain deletion on its own had little phenotypic effect, when combined with a bromodomain deletion in GCN5 or a temperature sensitive mutation in Tra1 (a targeting subunit in SAGA), SWI/SNFrelated phenotypes were uncovered (Hassan et al., 2002). This included the SNF (Sucrose Non Fermentor) phenotype of slow growth on raffinose (Hassan et al., 2002). These data support a role of the Swi2/Snf2 bromodomain in the recruitment and/or stable retention of SWI/SNF at promoters. The fact that the Swi2/Snf2 bromodomain participates in binding SWI/SNF to promoter nucleosomes raises

Molecular Cell 984

the possibility that binding to acetylated histones is the primary mechanism of targeting SWI/SNF. Both the SAGA and NuA4 HAT complexes have been shown to be recruited by transcription activators (Utley et al., 1998) and can lead to acetylation of promoter nucleosomes (Vignali et al., 2000). These acetylated nucleosomes could then serve to target SWI/SNF. In this scenario transcription activators would recruit HAT complexes, which acetylate promoter nucleosomes, which then recruit SWI/SNF. Thus, activator recruitment of SWI/SNF would be indirect, and the direct interactions of activators with SWI/SNF detected in vitro might be superfluous in vivo. To address the in vivo relevance of direct transcription activator-SWI/SNF interactions, we sought to create mutations within the complex that would perturb activator-SWI/SNF interactions. Toward this end we initially sought to identify the subunits of SWI/SNF that mediated these interactions. Surprisingly, multiple subunits of the complex displayed independent and specific interactions with activation domains. We found that acidic activation domains directly contact the Snf5, Swi1, and Swi2/Snf2 subunits within the yeast SWI/SNF complex (Neely et al., 2002). Similarly, human SWI/SNF has been shown to interact with many gene-specific factors (see above), and the human homologs of Swi2/Snf2, Swi1, and Snf5, which are BRG1 or hBRM, BAF250, and hSnf5/ INI1, respectively, have been implicated in these interactions (Cheng et al., 1999; Kowenz-Leutz and Leutz, 1999; Lee et al., 1999; Wu et al., 1996). In this study, we sought to identify and mutate the activator-interacting domains of Swi1, Snf5, and Swi2/ Snf2 subunits. While we were unable to map a discreet domain of interaction within Swi2/Snf2, we found that the N-terminal domain of Snf5 and the second quarter of Swi1 directly contact acidic activators. Deletion of both of these regions left the complex intact but impaired in its binding to activation domains. While deletion of either the Snf5 or Swi1 activator-interaction domains had modest phenotypic effect, deletion of both showed strong SWI/SNF-related phenotypes comparable to that of complete deletion of SWI1 and SNF5 genes, which disrupts the complex. Thus, these data demonstrate a crucial role of activator targeting in SWI/SNF function in vivo and, importantly, demonstrate that partially redundant activator interactions with the Swi1 and Snf5 subunits mediate targeting. Results and Discussion Identification of the Activator-Interacting Domains within SWI/SNF In order to understand the basis of multiple activator targets within the yeast SWI/SNF complex and their functional and biological relevance, we sought to determine the activator-interaction domains of the three subunits, Swi1, Snf5, and Swi2/Snf2, which are contacted by the activators (Neely et al., 2002). Our goal was to make SWI/SNF mutant complexes that are intact but defective in activator interactions. These subunits cannot simply be deleted, because their deletions are detrimental to the integrity of the complex in vivo (Peterson et al., 1994; Peterson and Herskowitz, 1992). To identify activator-

interaction domains, the Swi1, Snf5, and Swi2/Snf2 genes were divided into smaller pieces, translated into 35 S-labeled proteins in vitro, and tested for interaction with acidic activators, VP16 and Gcn4, by GST pulldown analysis. A schematic of the Snf5 protein and the regions that were translated is shown in Figure 1A. The N terminus domain of Snf5 (Snf5N [B] that includes amino acids 1 through 334) interacted strongly with VP16 and, albeit slightly more weakly, with Gcn4 (Figure 1A, lanes 5 and 7, row B). Consistent with this finding, the Snf5 protein deleted of the N terminus domain (Snf5MC1 [E], amino acids 335 through 905) did not interact with either activator (Figure 1A, lanes 5 and 7, row E). The Swi1 gene was divided into four smaller pieces and similarly tested for interaction with acidic activators VP16 and Gcn4 (Figure 1B). The second quarter of Swi1 (Swi1 329-657 [B]) was found to interact strongly with both activators (Figure 1B, lanes 5 and 9, row B). The other three pieces, Swi1 1-328 (A), Swi1 658-985 (C), and Swi1 985-1314 (D) did not interact with VP16 or Gcn4 (Figure 1B, lanes 5 and 9, rows A, C, and D). Moreover, a Swi1 mutant protein lacking amino acids 329–657 was found not to bind acidic activators in this assay (data not shown). These results led us to conclude that the second quarter of Swi1 (B) (Swi1 329-657) contains the activator-interacting domain. We similarly attempted to identify a specific activatorinteraction domain within the Swi2 protein. However, we were unsuccessful in identifying a specific domain in the Swi2 protein. Generation of an Activator-Interaction-Deficient SWI/SNF Complex The experiments showed in Figure 1 indicate that transcription activation domains bind the N-terminal third of the Snf5 protein and the second quarter B domain of the Swi1 protein. We also tested for the minimal domains of these proteins required to maintain the integrity of the SWI/SNF complex and found that only the middle third of Snf5 (Snf5M) was required for the integrity of SWI/SNF. Moreover, the complex also remained intact when the second quarter of Swi1 was deleted (data not shown). Since these mutant forms of Snf5 and Swi1 lacked the regions containing the activator-interaction domains, they were used in an attempt to form a mutant complex with decreased activator interactions. It was our hope that even though Swi2/Snf2 subunit would remain intact, perturbation of activator interactions with Swi1 and Snf5 would suffice to reduce activator interactions with the complex. A double deletion swi1⌬ snf5⌬ strain was generated and transformed with plasmids encoding either Swi1 wild-type or Swi1⌬B (Swi1 deleted of the amino acids 329 through 657) and with either Snf5 wild-type or Snf5M (amino acids 335 through 700). A schematic of the Swi1⌬B and Snf5M mutants is shown in Figure 2. Wholecell extracts were made from the cells of the resulting strains and tested for SWI/SNF interaction with VP16 and Gcn4, by GST pull-down analysis (Figure 2). The presence of SWI/SNF in these extracts and immunoprecipitates was monitored with antibody against the Swi3 subunit. In the deletion strain lacking Swi1 and Snf5 there was little Swi3 present in the extract (lane 1), illus-

SWI/SNF Activator Interactions 985

Figure 2. Deletion of Residues 329–657 of Swi1 and the Snf5 N-Terminal Domain Impairs SWI/SNF Complex Activator Interaction Schematic of the Swi1 and Snf5 mutants used. Swi1⌬B corresponds to the Swi1 protein deleted of amino acids 329–657; Snf5M corresponds to amino acids 335–700 of Snf5 protein. Dashed lines represent the region removed from the Swi1 and the Snf5 proteins. Yeast whole-cell extracts from strains expressing different combination of Swi1 and Snf5 wild-type (WT), mutant (⌬B, M), or no protein (⫺), as indicated, were incubated with either GST-Vp16, GST-Gcn4 fusion protein, or GST alone bound to glutathione Sepharose beads. The input whole-cell extracts and the bead fractions were assayed for the presence of Swi3 by Western blotting.

Figure 1. The Snf5 N-Terminal Domain and Residues 329–657 of Swi1 Interact with Activator (A) Schematic of the Snf5 protein and the different constructs used. The two highly glutamine-rich domains (Q), the three proline-rich regions (P), a highly charged middle region (hatched), and the two conserved repeats Rep1 and Rep2 domains (solid line) are indicated. Solid lines below the Snf5 protein scheme represent the smaller constructs that were expressed for the GST pull-down experiments. The letters indicated on the right correspond to the following pieces of Snf5 used: A, full-length Snf5 (1–905 aa); B, Snf5N 1–334 aa; C, Snf5M 335–700 aa; D, Snf5C 701–905 aa; E, Snf5MC1 335–905 aa; F, Snf5NM 1–700 aa. GST pull-down assays were performed using the GST fusion proteins indicated and individually expressed the 35 S-labeled Snf5 proteins indicated on the right. Fifty percent of each Input and supernatant (S) and 100% of the proteins associated with the beads (B) were loaded on a 10% SDS-PAGE. The radiolabeled proteins were visualized by autoradiography. (B) Schematic of the Swi1 protein and the different constructs used. The asparagine-rich N terminus domain (N), the glutamine-rich domain (Q), the ARID (AT rich interacting domain, conserved among Swi1 homologs proteins in Drosophila and humans) domain, and the two highly conserved domains C1 and C2 among Swi1 homologs proteins are indicated. Solid lines below the Swi1 protein scheme represent the smaller constructs used for the GST pull-down analysis. The letters indicated on the right correspond to the following pieces of Swi1 used: A, Swi1 1–328 aa; B, Swi1 329–657 aa; C, Swi1 658–985 aa; D, Swi1 985–1314 aa. GST pull-down assays were performed as describe above. The GST fusion proteins and the individually expressed 35S-labeled Swi1 proteins are indicated.

trating the dependence of the stability of this subunit and the complex in general on the presence of Swi1 and Snf5 (Peterson et al., 1994; Peterson and Herskowitz, 1992). When wild-type (WT) Swi1 and Snf5 were expressed, the levels of Swi3 were restored and pulled down with the Vp16 or Gcn4 activation domains (lane 2). In the absence of only the Swi1 activator-interacting domain (Swi1⌬B), only a slight reduction of SWI/SNF interaction with GST-Gcn4 was observed compared to the wild-type complex, while no reduction was noticeable with GST-VP16 (Figure 2, lane 3, and compare to lane 2). On the other hand, the absence of the Snf5 activator-interacting domain (Snf5M) showed a small reduction of the interaction of SWI/SNF with VP16 and a greater decreased interaction with Gcn4 (Figure 2, compare lane 4 with lane 2). Strikingly, when the two mutant subunits, Swi1⌬B and Snf5M, were combined together, there was a substantial reduction in interaction of the resulting SWI/SNF mutant complex with VP16 and no detectable interaction with Gcn4 (Figure 2, lane 5). These results strongly suggest that the Swi1 and Snf5 activator-interacting domains provide alternate surfaces to mediate the interaction of SWI/SNF with acidic activators. By removing both the Swi1 and Snf5 activator-interaction domains, we thus generated an activatorinteraction-deficient SWI/SNF complex. In order to demonstrate that effects on activator interactions in vitro and phenotypes in vivo were not due to the destabilization of the complex, we wanted to further verify that the mutant complex was intact. FLAG-IPs via a FLAG-tag on the Snf6 subunit in these different strains were carried out and were able to pull down other subunits of these mutant complexes including Swi3, Swi2 and Swp61/Arp7, as well as the Swi1 and the Snf5 mutant proteins, strongly suggesting that the complex was intact even in the absence of the activator-interacting domains (data not shown). In a second type of experi-

Molecular Cell 986

Figure 3. The Acidic Activator-Interacting-Deficient SWI/SNF Complex Is Intact The SWI/SNF complexes were purified as described in the Experimental Procedures from the following yeast cells containing the wild-type SWI/SNF complex (WT) (Swi1- and Snf5-wild-type), the SWI/SNF activator-interacting mutant complex (M) (Swi1⌬B, Snf5M), and the SWI/SNF complex lacking Swi1p and Snf5p (⫺) (swi1⌬, snf5⌬). The fractions peak of the MonoQ purification were loaded onto a Superose 6 column as described, and the Superose 6 fractions were assayed for the presence of Swi3 by Western blotting.

ment, the whole-cell extract was fractionated over a cation-exchange column (SP-sepharose) followed by an anion-exchange chromatography (Mono Q), and peak fractions containing SWI/SNF were pooled and then further fractionated on a gel filtration column (Superose 6). As can be seen in Figure 3, the double mutant complex (M) fractionated at the same exclusion size as the wildtype complex (WT) (fractions 8–10), indicating again that the activator-interaction deficient complex was intact. As a negative control, the SWI/SNF complex completely lacking the Swi1 and Snf5 subunits (⫺) was also fractionated and, as expected, presented an exclusion size greatly reduced to only approximately 700 kDa. Thus, deletion of these two genes severely destabilized SWI/ SNF while the partial deletions removing the activatorinteracting domains did not. Even though the mutant complex is intact, a second very important matter concerning this complex was to confirm that its remodeling activity remains the same as the wild-type SWI/SNF complex. Indeed, the removal of the activator-interacting domains could in some unexpected way impair its remodeling activity and therefore lead to swi/snf phenotypes without being related to the loss of activator interactions. In order to address this issue, we used two different in vitro remodeling assays, the restriction enzyme accessibility and the DNase I digestion assays on a 183 bp nucleosome core. These two assays are complementary, since the first one is quantitative and the second one is more qualitative. For these two assays, we used highly purified wild-type and mutant SWI/SNF complex. The whole-cell extract was fractionated over a cation-exchange column (SP-sepharose) followed by an anion-exchange chromatography (Mono Q), and peak fractions containing SWI/SNF were pooled and then immunoprecipitated using a FLAG-tag on the Snf6 subunit. The SWI/SNF complex was eluted off the beads using FLAG peptides and used for the remodeling assays. The same amounts of purified wildtype and mutant SWI/SNF complexes were used is these two assays (Figure 4B). The 183 bp DNA template generated by PCR using a radiolabeled 5⬘ primer is derived from plasmid pGALUSFBEND and was reconstituted into a nucleosome template by octamers transfer (see Experimental Procedures). We analyzed the abilities of

Figure 4. The Wild-Type and the Activator Mutant SWI/SNF Complexes Show Identical Remodeling Activity Using Restriction Enzyme Accessibility Assays (A) 5⬘ end-labeled GUB template was mock reconstituted (Naked DNA) or reconstituted into mononucleosome cores (Mononucleosome) and incubated with the same amount of indicated SWI/SNF complex (WT, wild-type; or Mut, activator mutant) in the presence or absence of ATP for 1 hr at 30⬚C. Then, SalI, XhoI, or PvuII restriction enzymes were added to the sample and incubated 30 more minutes at 30⬚C. After ethanol precipitation, DNA was dissolved in formamide loading buffer, heat denatured, and resolved on an 8% acrylamide-8 M urea sequencing gel. The gel was dried and exposed to Phosphorimager to be quantified. The quantifications of three independent experiments are indicated as percentage of cleavage corrected for the background cleavage obtained in the absence of ATP. The SalI, XhoI, and PvuII generate labeled cut fragments of, respectively, 70, 112, and 130 bp after cleavage of the 183 bp template. (B) Western blotting probed for the presence of the Swp61 SWI/ SNF subunit showing that equal amounts of wild-type and activator mutant purified SWI/SNF complexes were used for the experiments shown in Figures 4A and 5. The wild-type and activator mutant SWI/ SNF complexes were purified over a cation exchange column (SP sepharose) followed by an anion exchange (Mono Q HR5/5 column), and then were Flag-immunoprecipitated.

three different restriction enzymes (SalI, XhoI, and PvuII) to cleave their respective sites located in the 3⬘ half of the 183 bp nucleosome-reconstituted DNA fragment in the presence of the wild-type or the mutant SWI/SNF complexes in presence or absence of ATP. As shown in Figure 4A, the increased accessibility of these restriction sites was the same in the presence of either the wildtype or mutant SWI/SNF complex and was ATP dependant. Quantification of the experiment shown in Figure 4A and two repeats using phosphorimager analysis illustrated that a similar percentage of the nucleosome template was cleaved in the presence of either wildtype or mutant complex, indicating that the remodeling activity of the mutant SWI/SNF complex was not significantly reduced. Using the same nucleosome template, we compared the profile of DNase I digestion of the nucleosome template after being remodeled by the wild-type or the mutant SWI/SNF complexes. As shown in Figure 5, a hypersensitivity to DNase I was observed in the presence of wild-type or mutant SWI/SNF complex that was ATP dependant. This is characteristic of the remodeling activity of SWI/SNF. The profile and the intensity of the

SWI/SNF Activator Interactions 987

Figure 6. The Acidic Activator-Interacting-Deficient SWI/SNF Mutant Shows Reduced Growth under Conditions Requiring Activated Transcription Yeast strain YPP33 (swi1::HIS3 snf5::HIS3) was transformed with vector alone (⫺) or plasmids encoding either wild-type (⫹) or activator mutant (⌬B or M) Swi1p or Snf5p. The resulting transformants were plated on either CSM media (⫺Trp⫺Ura) with dextrose (Control), or with raffinose (⫹Raffinose), or with galactose (⫹Galactose) or on CSM dextrose media (⫺Trp⫺Ura) with 1 ␮g/ml of sulfometuron methyl (⫹SMM) or without inositol (⫺Inositol). Scans of plates are representative of at least three individual experiments.

Figure 5. Equal Remodeling Activity of Wild-Type and Activator Mutant SWI/SNF Complex Using DNase I Digestion Assay 5⬘ end-labeled GUB template was mock reconstituted (Naked DNA) or reconstituted into mononucleosome cores (Mononucleosome), and incubated with the same amount of indicated SWI/SNF complex (WT, wild-type; or Mut, activator mutant) in the presence or absence of ATP for 1 hr at 30⬚C. Then, the remodeling reaction mixtures were treated with 0.4 or 0.04 unit of DNase I for 1 min at RT, for nucleosome template or naked DNA, respectively. After stopping the digestion, the DNA was ethanol precipitated, then dissolved in formamide loading buffer, heat denatured, and resolved on an 8% acrylamide-8 M urea sequencing gel.

bands generated by the DNase I digestion are identical in both cases, showing that the mutant complex retained a qualitatively similar remodeling activity to that of the wild-type complex. These data demonstrate that the activator-interaction mutant SWI/SNF complex has the same quantitative and qualitative remodeling activity as the wild-type SWI/SNF complex. Therefore, the removal of the two activator-interacting domains did not disrupt the integrity or the remodeling activity of the mutant SWI/SNF complex. The Activator-Interacting Domains of Swi1 or Snf5 Are Required for the Function of the SWI/SNF Complex In Vivo Having created a SWI/SNF mutant complex that is no longer able to interact with activators, we sought to determine the impact of the loss of activator recruitment

on SWI/SNF function in vivo. Therefore, we tested cell growth under a variety of conditions in order to determine the phenotypes of this mutant strain. As can be seen in Figure 6, deletion of either the Swi1 or Snf5 activator-interaction domains had limited effect under several growth conditions. However, when both the Swi1 and Snf5 activator-interaction domains were deleted, a very strong effect on cell growth was observed in the presence of sulfometuron methyl (SMM) or in the absence of inositol, and a slightly more moderate effect was observed in the presence of raffinose or galactose. Growth in the presence of SMM requires activation of Gcn4 dependent genes (Jia et al., 2000). The growth in the absence of inositol requires the activation and expression of the INO1 gene. SWI/SNF is required for INO1 transcription (Peterson et al., 1991). Growth on raffinose requires the product of the SUC2 gene for invertase, and its transcription is affected by SWI/SNF (Neigeborn and Carlson, 1984). The growth on galactose induces activation of Gal4-driven genes that requires SWI/SNF complex (Peterson and Herskowitz, 1992), and the Gal4 activation domain interacts with SWI/SNF in vitro (Carrozza et al., 2002). We examined the level of the SUC2 gene expression in the wild-type and activator-interaction mutant SWI/ SNF strains in order to confirm that the phenotype observed in the presence of low dextrose (i.e., in the presence of raffinose) represented a transcriptional defect of the SUC2 gene. As can be seen in Figure 7, when plasmids encoding wild-type SWI1 and SNF5 genes (WT) were transformed into the double deletion swi1⌬ snf5⌬ strain, SUC2 expression was rescued under derepression conditions (low dextrose), albeit not quite to levels of the parental strain containing the endogenous genes (W303). By contrast, plasmids containing the activation-interaction-deficient mutant SWI1 and SNF5 genes

Molecular Cell 988

activation domain interactions with the complex and bore strong swi/snf phenotypes. Thus, the results presented here illustrate the importance of transcription activator recruitment of SWI/SNF in the function of the complex and, importantly, illustrate a functional redundancy of the Snf5 and Swi1 activator-interaction domains in facilitating SWI/SNF recruitment. Figure 7. Expression of SUC2 mRNA in Derepression Condition Is Strongly Affected in the SWI/SNF Activator Mutant Strain Compared to the Wild-Type Strain Total RNA (16 ␮g per lane) from W303 (wild-type control strain), WT (corresponding to the swi1::HIS3 snf5::HIS3 strain bearing centromeric expression plasmids for wild-type Swi1 and Snf5), and Mut strain (swi1::HIS3 snf5::HIS3 bearing centromeric expression plasmids for Swi1⌬B and Snf5M activator mutant) was electrophoresed through a 1.2% agarose gel and transferred to a charged Nylon membrane. Northern blots were hybridized with full-length SUC2 and SCR1 probes and exposed to autoradiographic film.

(MUT) were unable to rescue SUC2 expression. These results show that the activator-interacting mutant SWI/ SNF complex is unable to correctly regulate the SUC2 gene expression and therefore indicate that the acidic activator recruitment of the SWI/SNF complex to the SUC2 promoter is essential for its transcriptional regulation. Taken together, these data suggest that the Swi1 and Snf5 acidic-activator-interacting domains are important in the activation of Gcn4 and Gal4-driven genes as well as for the SUC2 and the INO1 gene expression. Indeed, the activator-interaction domain double deletion has growth defects under these conditions that are similar to deletion of the entire swi1 and snf5 genes, which disrupts SWI/SNF complex. Thus, these data indicate that recruitment by activators is a crucial aspect of SWI/ SNF function in vivo. A long-standing question regarding the function of SWI/SNF and other chromatin remodeling complexes has been the mechanisms that localize them to promoters for transcription activation. Previous studies by us, and others, demonstrated that the yeast SWI/SNF complex could be recruited to promoters by binding directly to transcription activation domains (Natarajan et al., 1999; Neely et al., 1999; Yudkovsky et al., 1999). Surprisingly, the interaction of the complex with transcription activators could not be attributed to a single subunit. Indeed, a number of different biochemical assays indicated that multiple subunits could bind transcription activators including the Swi1, Snf5, and Swi2/Snf2 subunits (Neely et al., 2002). This observation raises the obvious question as to which, if any, of these interactions are relevant in vivo. While we have not identified a specific activator-interaction domain within the Swi2/ Snf2 subunit, it does seem that potential interactions of activators with Swi2/Snf2 are not sufficient to mediate activator interactions with the complex in vitro or to prevent the appearance of swi/snf phenotypes in vivo. Deletion of either the Snf5 or the Swi1 activator-interaction domains alone had only a small effect on activation domain interactions with the complex or on the appearance of swi/snf phenotypes. By contrast, deletion of both activator-interaction domains markedly crippled

Experimental Procedures SWI/SNF Subunit Constructs and GST Pull-Down Assays The ORF of SWI/SNF genes and various subdomains were cloned into the vector pRSETA (Invitrogen, Carlsbad, CA), which contains a T7 promoter. Plasmids and sequences used in this study are available upon request. The proteins were expressed with a T7-TNT coupled reticulocyte lysate system (Promega) that generated 35SMethionine labeled product. The GST fusion proteins were expressed in bacteria and purified, and GST pull-down assays were done according to the method described previously (Neely et al., 2002). SWI/SNF Purification Yeast SWI/SNF complexes were purified from the strains used in this study that derived from the YPP33 strain. Whole-cell extracts were prepared from 6 liters of yeast cells grown in CSM media (lacking tryptophane and uracil) as described previously (Grant et al., 1999). The extract was diluted down to 100 mM NaCl and bound with 5 ml of SP-Sepharose Fast Flow resin (Amersham). The proteins were step-eluted from the resin with extract buffer containing 100– 600 mM NaCl by 100 mM steps. Fractions were monitored for the presence of SWI/SNF by immunoblotting. Peak fractions (400 and 500 mM NaCl) were pooled and dialyzed against Mono Q buffer (Grant et al., 1999) and loaded onto an FPLC MonoQ HR5/5 column (Amersham). Bound proteins were eluted with a 25 ml salt gradient of 100–500 mM NaCl in MonoQ buffer. Fractions containing SWI/ SNF (ⵑ300 mM NaCl) were pooled and concentrated to 0.3 ml on a Centricon-30 (Amicon). 0.05 ml was loaded on an FPLC Superose 6 PC 3.2/30 column (Amersham) equilibrated with extract buffer (350 mM NaCl) and run at 20 ␮l/min. Fractions were monitored for the presence of SWI/SNF complex by immunoblotting. DNase I Digestion Assay The single Gal4-site probe (GUB) was generated by PCR as described (Juan et al., 1993) and used as naked DNA or a reconstituted mononucleosome in this assay. Then, WT or Mut SWI/SNF was added to approximately 10 ng of the GUB template in a buffer that contains 10 mM HEPES (pH 7.8), 50 mM KCl, 5 mM DTT, 5 mM PMSF, 5% glycerol, 0.25 mg/ml BSA, and 2 mM MgCl2 in the presence or absence of 2 mM ATP. After incubation for 1 hr at 30⬚C, the binding reactions were then treated with DNase I (0.04 U for naked DNA templates and 0.4 U for nucleosomal templates) (Roche) in 50 mM MgCl2 for 1 min at room temperature (Coˆte´ et al., 1994). An equal volume of stop buffer (20 mM Tris-HCl [pH 7.5], 50 mM EDTA, 2% SDS, 0.2 mg/ml proteinase K, 1 mg/ml glycogen) was added to the DNase I reactions and incubated at 50⬚C for 1 hr. Deproteinized samples were precipitated with 200 mM NaCl and 3 volumes of ethanol, and the pellet resuspended in 5 ␮l of the formamide dye (95% formamide, 10 mM EDTA, 0.1% xylene cyanol, 0.1% bromophenol blue). After 5 min incubation at 95⬚C, the samples were run on an 8% acrylamide (19:1 acrylamide to bis-acrylamide), 8 M urea sequencing gel at 60 watts for 2.5 hr and visualized by autoradiography. Restriction Enzyme Accessibility Assay The same template (GUB) utilized for the DNase I assay was used in this assay. The reaction conditions for the remodeling are the same as the DNase I assay. Briefly, after 1 hr of incubation with WT or Mut SWI/SNF at 30⬚C, 10 units of SalI or XhoI or 2.5 units of PvuII were added to the reaction for 30 min at 30⬚C. The digestion was stopped and treated as described above. The samples were run on an 8% acrylamide (19:1), 8 M urea gel at 10 watts for 1 hr and

SWI/SNF Activator Interactions 989

visualized by autoradiography. The gels were exposed to Phosphorimager and quantified.

Hassan, A.H., Neely, K.E., Vignali, M., Reese, J.C., and Workman, J.L. (2001). Promoter targeting of chromatin-modifying complexes. Front. Biosci. 6, D1054–D1064.

Growth Analysis and Strains The Swi1⌬ Snf5⌬ double deletion strain (YPP33 [MAT␣ ura3-1 trp1-1 his3-11,15 leu2-3,112 can1-100, ade2-1, swi1::HIS3 snf5::HIS3]) used for the study of Swi1 and Snf5 acidic-activator mutants was created for this study and is derived from W303 background. This strain was transformed with either pSWI1 wild-type-TRP1-CEN or pSWI1 ⌬B-TRP1-CEN and with either pSNF5 wild-type URA3-CEN or pSNF5 M TRP1-CEN, and were recovered by growth on CSMTrp-Ura plates. Plasmids and sequences are available upon request. Growth of strains obtained were compared to each other and to the double deletion strain transformed with vector alone. Cells were collected in stationary phase, washed with water, and compared by colony dilution assay (4-fold dilutions from a starting OD600nm of 1.0) after 2 or 3 days at 30⬚C.

Hassan, A.H., Prochasson, P., Neely, K.E., Galasinski, S.C., Chandy, M., Carrozza, M.J., and Workman, J.L. (2002). Function and selectivity of bromodomains in anchoring chromatin-modifying complexes to promoter nucleosomes. Cell 111, 369–379.

Acknowledgments We thank E.T. Young, B. Cairns, and B. Laurent for antisera against SWI/SNF subunits. A special thanks goes to Michael Carrozza for his help with the FPLC systems. Finally, we thank the members of the Workman, Simpson, Reese, and Tan labs at Penn State for valuable discussions. P.P. is a Postdoctoral Research Fellow of the Leukemia & Lymphoma Society, and J.L.W. is an Associate Investigator of the Howard Hughes Medical Institute. This work was supported by NIGMS grant R37 GM47867 to J.L.W. Received: April 1, 2003 Revised: July 16, 2003 Accepted: August 8, 2003 Published: October 23, 2003 References Armstrong, J.A., Bieker, J.J., and Emerson, B.M. (1998). A SWI/ SNF-related chromatin remodeling complex, E-RC1, is required for tissue-specific transcriptional regulation by EKLF in vitro. Cell 95, 93–104. Barbaric, S., Walker, J., Schmid, A., Svejstrup, J.Q., and Horz, W. (2001). Increasing the rate of chromatin remodeling and gene activation–a novel role for the histone acetyltransferase Gcn5. EMBO J. 20, 4944–4951. Carrozza, M.J., John, S., Sil, A.K., Hopper, J.E., and Workman, J.L. (2002). Gal80 confers specificity on HAT complex interactions with activators. J. Biol. Chem. 277, 24648–24652. Cheng, S.W., Davies, K.P., Yung, E., Beltran, R.J., Yu, J., and Kalpana, G.V. (1999). c-MYC interacts with INI1/hSNF5 and requires the SWI/SNF complex for transactivation function. Nat. Genet. 22, 102–105. Cho, H., Orphanides, G., Sun, X., Yang, X.-J., Ogryzko, V., Lees, E., Nakatani, Y., and Reinberg, D. (1998). A human RNA polymerase II complex containing factors that modify chromatin structure. Mol. Cell. Biol. 18, 5355–5363. Coˆte´, J., Quinn, J., Workman, J.L., and Peterson, C.L. (1994). Stimulation of GAL4 derivative binding to nucleosomal DNA by the yeast SWI/SNF complex. Science 265, 53–60. Coˆte´, J., Peterson, C.L., and Workman, J.L. (1998). Perturbation of nucleosome core structure by the SWI/SNF complex persists after its detachment, enhancing subsequent transcription factor binding. Proc. Natl. Acad. Sci. USA 95, 4947–4952. Dhalluin, C., Carlson, J.E., Zeng, L., He, C., Aggarwal, A.K., and Zhou, M.M. (1999). Structure and ligand of a histone acetyltransferase bromodomain. Nature 399, 491–496.

Haynes, S.R., Dollard, C., Winston, F., Beck, S., Trowsdale, J., and Dawid, I.B. (1992). The bromodomain: a conserved sequence found in human, Drosophila and yeast proteins. Nucleic Acids Res. 20, 2603. Jacobson, R.H., Ladurner, A.G., King, D.S., and Tjian, R. (2000). Structure and function of a human TAFII250 double bromodomain module. Science 288, 1422–1425. Jia, M.H., Larossa, R.A., Lee, J.M., Rafalski, A., Derose, E., Gonye, G., and Xue, Z. (2000). Global expression profiling of yeast treated with an inhibitor of amino acid biosynthesis, sulfometuron methyl. Physiol. Genomics 3, 83–92. Juan, L.-J., Walter, P.P., Taylor, I.C., Kingston, R.E., and Workman, J.L. (1993). Nucleosome cores and histone H1 in the binding of GAL4 derivatives and the reactivation of transcription from nucleosome templates in vitro. Cold Spring Harb. Symp Quant Biol 58, 213–223. Kadam, S., McAlpine, G.S., Phelan, M.L., Kingston, R.E., Jones, K.A., and Emerson, B.M. (2000). Functional selectivity of recombinant mammalian SWI/SNF subunits. Genes Dev. 14, 2441–2451. Kowenz-Leutz, E., and Leutz, A. (1999). A C/EBP␤ isoform recruits the SWI/SNF complex to activate myeloid genes. Mol. Cell 4, 735–743. Lee, C.H., Murphy, M.R., Lee, J.S., and Chung, J.H. (1999). Targeting a SWI/SNF-related chromatin remodeling complex to the beta-globin promoter in erythroid cells. Proc. Natl. Acad. Sci. USA 96, 12311– 12315. Natarajan, K., Jackson, B.M., Zhou, H., Winston, F., and Hinnebusch, A.G. (1999). Transcriptional activation by Gcn4p involves independent interactions with the SWI/SNF complex and the SRB/mediator. Mol. Cell 4, 657–664. Neely, K.E., Hassan, A.H., Wallberg, A.E., Steger, D.J., Cairns, B.R., Wright, A.P.H., and Workman, J.L. (1999). Activation domain-mediated targeting of the SWI/SNF complex to promoters stimulates transcription from nucleosome arrays. Mol. Cell 4, 649–655. Neely, K.E., Hassan, A.H., Brown, C.E., Howe, L., and Workman, J.L. (2002). Transcription activator interactions with multiple SWI/ SNF subunits. Mol. Cell. Biol. 22, 1615–1625. Neigeborn, L., and Carlson, M. (1984). Genes affecting the regulation of SUC2 gene expression by glucose repression in Saccharomyces cerevisiae. Genetics 108, 845–858. Peterson, C.L., and Herskowitz, I. (1992). Characterization of the yeast SWI1, SWI2, and SWI3 genes, which encode a global activator of transcription. Cell 68, 573–583. Peterson, C.L., Kruger, W., and Herskowitz, I. (1991). A functional interaction between the C-terminal domain of RNA polymerase II and the negative regulator SIN1. Cell 64. Peterson, C.L., Dingwall, A., and Scott, M.P. (1994). Five SWI/SNF gene products are components of a large multisubunit complex required for transcriptional enhancement. Proc. Natl. Acad. Sci. USA 91, 2905–2908. Quinn, J., Fyrberg, A.M., Ganster, R.W., Schmidt, M.C., and Peterson, C.L. (1996). DNA-binding properties of the yeast SWI/SNF complex. Nature 379, 844–847. Sullivan, E.K., Weirich, C.S., Guyon, J.R., Sif, S., and Kingston, R.E. (2001). Transcriptional activation domains of human heat shock factor 1 recruit human SWI/SNF. Mol. Cell. Biol. 21, 5826–5837.

Fryer, C.J., and Archer, T.K. (1998). Chromatin remodeling by the glucocorticoid receptor requires the BRG1 complex. Nature 393, 88–91.

Utley, R.T., Ikeda, K., Grant, P.A., Coˆte´, J., Steger, D.J., Eberharter, A., John, S., and Workman, J.L. (1998). Transcriptional activators direct histone acetyltransferase complexes to nucleosomes. Nature 394, 498–502.

Grant, P.A., Berger, S.L., and Workman, J.L. (1999). Identification and analysis of native nucleosomal histone acetyltransferase complexes. Methods Mol. Biol. 119, 311–317.

Vignali, M., Steger, D.J., Neely, K.E., and Workman, J.L. (2000). Distribution of acetylated histones resulting from Gal4–VP16 recruitment of SAGA and NuA4 complexes. EMBO J. 19, 2629–2640.

Molecular Cell 990

Wallberg, A.E., Neely, K.E., Hassan, A.H., Gustafsson, J.-A˚., Workman, J.L., and Wright, A.P.H. (2000). Recruitment of the SWI-SNF chromatin remodeling complex as a mechanism of gene activation by the glucocorticoid receptor ␶1 activation domain. Mol. Cell. Biol. 20, 2004–2013. Wang, W. (2003). The SWI/SNF family of ATP-dependent chromatin remodelers: similar mechanisms for diverse functions. In Protein Complexes that Modify Chromatin, J.L. Workman, ed. (Berlin; New York: Springer), pp. 144–169. Wilson, C.J., Chao, D.M., Imbalzano, A.N., Schnitzler, G.R., Kingston, R.E., and Young, R.A. (1996). RNA polymerase II holoenzyme contains SWI/SNF regulators involved in chromatin remodeling. Cell 84, 235–244. Wu, D.Y., Kalpana, G.V., Goff, S.P., and Schubach, W.H. (1996). Epstein-Barr virus nuclear protein 2 (EBNA2) binds to a component of the human SNF-SWI complex, hSNF5/Ini1. J. Virol. 70, 6020–6028. Yudkovsky, N., Logie, C., Hahn, S., and Peterson, C.L. (1999). Recruitment of the SWI/SNF chromatin remodeling complex by transcriptional activators. Genes Dev. 13, 2369–2374.