SWI/SNF is recruited to DNA byactivators

Next we tested directly whether SWI/SNF was recruited to the templates by activator in the yeast NEs. Assays were performed either without activator, with the Gal4(1–94) DNA-binding domain, or with Gal4–AH or Gal4–VP16, with both wild-type and TBPI143N NEs. In the wild-type extract, Gal4–AH stimulated recruitment of Rpb3p, Srb4p, TFIIB, and Toa2p 4- to 10-fold, as compared with recruitment in the absence of activator (Fig. ​(Fig.2,2, lane 2 vs. lane 6). Gal4–AH did not significantly recruit TBP (Ranish et al. 1999). Gal4–VP16, however, stimulated recruitment of all these components 2- to 20-fold (Fig. ​(Fig.2,2, lane 2 vs. lane 8). Recruitment of Swi3p and Snf5p by these activators was stronger than the recruitment of holoenzyme components. Gal4–AH increased Swi3p and Snf5p levels >13-fold, whereas Gal4–VP16 increased Swi3p and Snf5p levels >21-fold; both Swi3p and Snf5p were barely detectable in the absence of activator (Fig. ​(Fig.2,2, lanes 2,6,8). In contrast, the binding of subunits of two other SWI/SNF-like complexes, Isw1p, a component of ISW1 (Tsukiyama et al. 1999), and Sth1p, a component of RSC (Cairns et al. 1996), was constitutive and was not influenced by the presence or absence of activators, promoter sequences, or Pol II holoenzyme (Fig. ​(Fig.2;2; data not shown). The decrease in Sth1p levels seen on the ΔPromoter template is most likely due to the fact that this template is ~150 bp shorter than the wild-type template. To show that the recruitment of SWI/SNF was dependent on the activation domains, we used Gal4(1–94) as a control (Fig. ​(Fig.2,2, lane 4). Although it weakly stimulated binding of Rpb3p, Srb4p, TFIIB, and Toa2p levels, we saw no stimulation of either Swi3p or Snf5p recruitment. Both Gal4–AH and Gal4–VP16 stimulated Swi3p and Snf5p recruitment to the same extent in TBPI143N NE as in wild type (Fig. ​(Fig.2,2, lanes 10,12,14). Importantly, Swi3p and Snf5p recruitment was unaffected by the addition of rTBP to restore PIC assembly (Fig. ​(Fig.2,2, lanes 16,18,20). In the case of Gal4–AH, SWI/SNF recruitment clearly occurred in the absence of recruitment of other PIC components. However, Gal4–VP16 was also able to recruit nearly wild-type levels of Rpb3p, Srb4p, and TFIIB in the absence of rTBP, as noted previously (Fig. ​(Fig.2,2, lane 14; Ranish et al. 1999). This is because Gal4–VP16, unlike Gal4–AH, can recruit holoenzyme in the absence of TBP. By using GST–Swi3p to create a standard curve for Western blots, we determined that ~5 fmoles of SWI/SNF were recruited to the promoter by Gal4–AH (data not shown). This is significantly lower than the ~40 fmoles of TFIIA, TFIIB, and TFIIE and ~200 fmoles of TBP recruited by Gal4–AH in this assay (Ranish 1999). Thus, it is possible that SWI/SNF is not recruited to promoters as part of the PIC, because only ~10% of the templates are occupied by PICs in this assay (Ranish 1999). Together, these data are consistent with the idea that SWI/SNF is recruited to promoters independent of holoenzyme.

Open in a separate window

Figure 2

SWI/SNF is recruited to DNA by activators independently of promoter sequences. PICs were assembled on both wild-type (W) and ΔPromoter (Δ) templates as stated in Fig. ​Fig.1,1, with the nuclear extracts (NE) indicated. Reactions were performed either with no activator (−), the GAL4 (1–94) DNA-binding domain (G), or Gal4–AH (A) and Gal4–VP16 (V), as indicated. rTBP (400 ng) was added where indicated. PICs were analyzed by Western blotting with antibodies against the components indicated at right. (Lane 1) Dynabeads without templates were used in PIC assembly as a control for nonspecific binding to beads.

Next, we determined whether SWI/SNF recruitment by activators required the presence of promoter sequences. The HIS4 promoter sequences from the wild-type template were deleted and replaced by nonpromoter DNA to form the ΔPromoter template (Fig. ​(Fig.1A).1A). We then tested this template in immobilized template assays using a wild-type NE. Swi3p and Snf5p recruitment by Gal4–AH and Gal4–VP16 was completely unaffected by deletion of the promoter sequences (Fig. ​(Fig.2,2, lanes 3,7,9). As expected, only small-to-undetectable amounts of the other general factors probed for were present when Gal4–AH was used with this template, whereas nearly wild-type amounts of holoenzyme components were recruited by Gal4–VP16. Identical results were seen when the TBPI143N mutant extract was used (Fig. ​(Fig.2,2, lanes 11,13,15). As before, the recruitment of Swi3p and Snf5p was unaffected by the addition of rTBP (Fig. ​(Fig.2,2, lanes 17,19,21).

We also performed an immobilized template assay using a template that terminated 14-bp downstream of the Gal4 site. Digestion of this template with PstI after incubation with NE produced a 53-bp fragment. Although approximately twofold less Gal4–AH bound to this template, we saw no recruitment of SWI/SNF (data not shown). These data, together with evidence that SWI/SNF can only bind nonspecifically to pieces of DNA >80 bp (Quinn et al. 1996), suggest that the formation of a stable complex between SWI/SNF and activator requires DNA, but not specific promoter sequences.

Transcriptional activators can target SWI/SNF remodelingactivity

Next, we wished to determine whether a transcriptional activator could target the chromatin remodeling activity of purified SWI/SNF complex. To investigate this possibility, we exploited a sensitive nucleosomal array remodeling assay in which the activity of a nucleosome remodeling complex is coupled to restriction enzyme activity (Logie and Peterson 1997, 1999). The DNA template that we used is composed of 11 head-to-tail repeats of a L. variegatus 5S rRNA gene (208-11S; Fig. ​Fig.3A).3A). Each repeat can position a nucleosome after in vitro reconstitution with purified histone octamers, yielding a homogeneous array of positioned nucleosomes. The sixth repeat of this DNA template contains a 5S sequence that bears a unique SalI–HincII restriction site close to the dyad axis of symmetry of the reconstituted nucleosome (Polach and Widom 1995), thereby allowing for a quantitative kinetic assay for nucleosome remodeling within nucleosomal arrays (Logie and Peterson 1999). To investigate whether a GAL4 activator can target SWI/SNF remodeling activity, we modified the DNA template so that it contains five high affinity GAL4-binding sites adjacent to the 5S repeat that harbors the SalI–HincII restriction site (208-11S–GAL4; Fig. ​Fig.3A).3A). Reconstitution of the 208-11S–GAL4 DNA template into a nucleosomal array positions the GAL4-binding sites in the linker region between two positioned nucleosomes (Fig. ​(Fig.3A).3A).

Open in a separate window

Open in a separate window

Figure 3

SWI/SNF remodeling activity can be targeted by a transcriptional activator. (A) Schematic of nucleosomal array templates. (B) The ³²P-labeled arrays (0.2 nm final) were mixed with unlabeled arrays (3 nm final) in the four possible combinations and then either incubated with HincII alone (○) or exposed to 0.8 nm SWI/SNF and HincII in the absence (crossed circles), or presence of 10 nm GAL4–VP16 (●). The digestion of the labeled array with time was quantified and graphed at left.

We assembled four sets of remodeling reactions in which targeting of remodeling activity would occur on radiolabeled arrays competing against a 15-fold excess of unlabeled competitor arrays. (Fig. ​(Fig.3B).3B). HincII digestion of these four array mixtures was essentially equivalent in the absence of SWI/SNF [(Fig. 3B) first order rate=1×10⁻³/min; similar results are seen in the presence of SWI/SNF but in the absence of ATP (Logie and Peterson 1997; Logie et al. 1999)]. Then, we assessed the capacity of SWI/SNF to remodel the labeled nucleosomal arrays in the absence of a GAL4 derivative. Addition of a limiting concentration of SWI/SNF complex (0.8 nm) stimulated HincII kinetics three- to fivefold (first order rate=3×10³ to 5×10⁻³/min). Thus, SWI/SNF can remodel both types of nucleosomal arrays, and the presence or absence of GAL4-binding sites on the labeled or competitor array does not influence the kinetics (Fig. ​(Fig.3B;3B; data not shown). Next, we tested whether Gal4–VP16 could target SWI/SNF activity to an array bearing Gal4 sites. This was achieved by including 10 nm Gal4–VP16 in the four array mixtures. The presence of Gal4–VP16 did not affect remodeling kinetics when both the labeled and unlabeled arrays lacked Gal4 sites (first order cleavage rate=3×10⁻³/min; Fig. ​Fig.3B,3B, top). When remodeling of the labeled array lacking Gal4 sites (208-11S) was measured in the presence of a 15-fold excess of unlabeled array bearing Gal4 sites (208-11S–GAL4), we observed a ~3-fold reduction in the first order rate of restriction enzyme cleavage induced by Gal4–VP16 (1×10⁻³/min; Fig. ​Fig.3B,3B, second panel), indicating that the Gal4–VP16 that was bound to the competitor array sequestered SWI/SNF activity. When both the labeled and competitor array bore GAL4 sites, we observed an approximately two- to threefold increase in rate compared with the rate in the absence of Gal4–VP16 (5×10⁻³ vs. 14×10⁻³/min; Fig. ​Fig.3B,3B, third panel). Finally, when 10 nm Gal4–VP16 was added to reactions containing a labeled array bearing GAL4 sites (208-11S–GAL4) immersed in a 15-fold excess of unlabeled array lacking GAL4 sites (208-11S), we observed a dramatic enhancement of the cleavage rate when compared with SWI/SNF alone (23×10⁻³/min; Fig. ​Fig.3B,3B, bottom). This stimulation of restriction enzyme cleavage rate by activators was absolutely dependent on the continuous presence of ATP (data not shown; see also Logie and Peterson 1997) and on the presence of SWI/SNF complex (Fig. ​(Fig.4A).4A). Together, these experiments demonstrate that the Gal4–VP16 activator can recruit SWI/SNF remodeling activity in vitro to nucleosomal arrays bearing GAL4-binding sites.

Open in a separate window

Open in a separate window

Figure 4

SWI/SNF activity is recruited by acidic activation domains. HincII cleavage kinetics of reactions containing 0.2 nm labeled 208-11S–GAL4 array, 3 nm unlabeled 208-11S competitor array in the presence or absence of GAL4 derivatives, and in the absence (A) or presence (B) of 0.8 nm SWI/SNF. Note that the GAL4 derivatives did not affect HincII cleavage kinetics in the absence of SWI/SNF. (□) No Gal4/no SWI/SNF; (█) +SWI/SNF; () Gal4 (1–94); (●) Gal4–AH; () Gal4–Pro; (♦) Gal4–VP16.

Reconstitution of nucleosomalarrays

DNA templates were prepared by restriction enzyme digestion of pCL7c (208-11S) or pCL8b (208-11S–Gal4), purified by gel filtration, and labeled by Klenow fill-in reaction with [α-³²P]dATP (Logie and Peterson 1999). Reconstitution of histone octamers onto 208-11 array templates were performed by salt gradient dialysis with a ratio of histone octamer to 5S rRNA repeat of 1.0 (Logie and Peterson 1999). Coupled remodeling–restriction reactions were performed in 25 μl, with a final concentration of 5 mm MgCl₂, 125 mm NaCl, 10 mm Tris-HCL (pH 8.0), 1 mm DTT, 0.1 mg/ml BSA, and 1 mm ATP. HincII (New England Biolabs) was added to reactions at a final concentration of 500 units/ml. HincII cleavage was quantified by PhosphorImager analysis, and first-order rates were determined by curve fitting. In multiple independent experiments, the first-order rates of restriction enzyme cleavage for each particular combination of array, remodeler, and activator varied by ≤30%.

We thank members of the Hahn and Reeder laboratories for help and suggestions, and J. Ranish and T. Tsukiyama for thoughtful discussions. We also thank H. Sakurai for Gal11 antibodies, Nancy Thompson for Rpb3 antibodies, J. Movius for Srb4 antibodies, B. Laurent for Sth1 antibodies, and R. Young for Srb5 antibodies. We also thank J. Ranish for the TBP temperature-sensitive extract, J. Geiger for rTBP, and J. Flanagan for GST–SWI3 protein. We thank T. Imbalzano for the kind gift of GAL4(1–94) protein and N. Tanese for the GAL4–proline expression plasmid. This work was supported by grants from the National Institutes of Health to C.L.P. (GM496054) and to S.H. (GM42551), a grant from the National Cancer Institute to N.Y. (T32CA09657), and by a fellowship to C.L. from the Human Frontiers Science Organization. C.L.P. is a Scholar of the Leukemia Society of America, and S.H. is an Associate Investigator of the Howard Hughes Medical Institute.

The publication costs of this article were defrayed in part by payment of page charges. This article must therefore be hereby marked ‘advertisement’ in accordance with 18 USC section 1734 solely to indicate this fact.