Srb/mediator proteins interact functionally and physically with transcriptional repressor Sfl1

Wenjie Song, Marian Carlson

Author Affiliations

  1. Wenjie Song1 and
  2. Marian Carlson*,2
  1. 1 Departments of Genetics and Development, Columbia University, 701 W. 168th St., New York, NY, 10032, USA
  2. 2 Microbiology, Columbia University, 701 W. 168th St., New York, NY, 10032, USA
  1. *Corresponding author. E-mail: mbc1{at}


Srb/mediator proteins that are associated with RNA polymerase II holoenzyme have been implicated in transcriptional repression in Saccharomyces cerevisiae. We show here that the defect in repression of SUC2 caused by mutation of SRB8, SRB9, SRB11, SIN4 or ROX3 is suppressed by increased dosage of the SFL1 gene, and the genetic behavior of the sfl1Δ mutation provides further evidence for a functional relationship. Sfl1 acts on SUC2 through a repression site located immediately 5′ to the TATA box, and Sfl1 binds this DNA sequence in vitro. Moreover, LexA–Sfl1 represses transcription of a reporter, and repression is reduced in an srb9 mutant. Finally, we show that Sfl1 co‐immunoprecipitates from cell extracts with Srb9, Srb11, Sin4 and Rox3. We propose that Sfl1, when bound to its site, interacts with Srb/mediator proteins to inhibit transcription by RNA polymerase II holoenzyme.


Transcriptional regulation requires the interactions of specific regulatory proteins with components of the transcription machinery. Recent work has indicated that RNA polymerase II holoenzyme forms play an important role in transcriptional regulatory mechanisms (for a review, see Greenblatt, 1997). In particular, the Srb/mediator proteins that are associated with the holoenzyme have been implicated in both transcriptional activation and repression.

The SRB genes were identified in Saccharomyces cerevisiae by Young and colleagues as suppressors of a C‐terminal heptapeptide repeat domain (CTD) truncation of RNA polymerase II (Nonet and Young, 1989; Hengartner et al., 1995; Liao et al., 1995), and the Srb proteins were found associated with an RNA polymerase II holoenzyme that responds to transcriptional activators (Thompson et al., 1993; Koleske and Young, 1994; Hengartner et al., 1995). A holoenzyme form containing a mediator that confers responsiveness to activators was identified independently by Kornberg and colleagues (Kim et al., 1994). The mediator is associated with the CTD and also stimulates basal transcription and phosphorylation of the CTD. The mediator comprises Srb2, Srb4, Srb5, Srb6, Srb7, Gal11, Sin4, Rgr1, Rox3, Hrs1 and Med proteins, but not Srb8, Srb9, Srb10 or Srb11 (Kim et al., 1994; Li et al., 1995; Gustafsson et al., 1997; Myers et al., 1998). RNA polymerase II holoenzyme complexes have also been isolated from mammalian cells and found to contain Srb homologs, including Srb7 and cyclin C/cdk8 (Ossipow et al., 1995; Chao et al., 1996; Maldonado et al., 1996; Pan et al., 1997; Neish et al., 1998), which is a homolog of the Srb10/Srb11 kinase (Surosky et al., 1994; Kuchin et al., 1995; Liao et al., 1995).

Genetic analysis has revealed that some Srb/mediator proteins have roles in transcriptional repression. Mutations in SRB8, SRB9, SRB10, SRB11, SIN4, ROX3, GAL11, RGR1 and HRS1 affect the negative regulation of a diverse set of promoters and have been isolated in many different mutant searches (Sternberg et al., 1987; Sakai et al., 1990; Rosenblum‐Vos et al., 1991; Chen et al., 1993; Covitz et al., 1994; Stillman et al., 1994; Surosky et al., 1994; Balciunas and Ronne, 1995; Kuchin et al., 1995; Wahi and Johnson, 1995; Song et al., 1996; Piruat et al., 1997; for review, see Carlson, 1997). The mechanism by which these genes affect repression remains unclear, but evidence suggests a role in the response to DNA‐binding repressors. The mutations sin4 and rgr1 relieve repression of reporters by Rme1, a repressor of meiotic gene expression (Covitz et al., 1994; Shimizu et al., 1997); sin4 and srb10 relieve repression by α2‐Mcm1 (Chen et al., 1993; Wahi and Johnson, 1995); and srb10 and srb11 reduce repression by Mig1 (Kuchin and Carlson, 1998). α2‐Mcm1 and Mig1 function in concert with the Ssn6(Cyc8)–Tup1 corepressor (Keleher et al., 1992; Treitel and Carlson, 1995; Tzamarias and Struhl, 1995), and SRB10 and SRB11 are required for repression of reporters by LexA fusions to Ssn6 and Tup1 (Kuchin and Carlson, 1998). These data implicate Srb/mediator proteins in the response to Ssn6–Tup1; however, evidence indicates that Ssn6–Tup1 also represses transcription by mechanisms involving chromatin (Cooper et al., 1994; Roth, 1995; Edmondson et al., 1996). No direct physical interaction between Ssn6–Tup1 and Srb/mediator proteins has been reported.

Our laboratory has focused on the role of Srb/mediator proteins in glucose repression of SUC2 transcription. We previously identified alleles of SRB8, SRB9, SRB10, SRB11, SIN4 and ROX3 as ssn (suppressor of snf1) mutations that affect SUC2 repression (Carlson et al., 1984; Kuchin et al., 1995; Song et al., 1996); for simplicity, we will refer to these six collectively as srb/ssn mutations. Repression of SUC2 requires Ssn6–Tup1, which is recruited to upstream sites by Mig1 and a second DNA‐binding protein, Mig2 (Schultz and Carlson, 1987; Nehlin and Ronne, 1990; Williams et al., 1991; Treitel and Carlson, 1995; Tzamarias and Struhl, 1995; Lutfiyya and Johnston, 1996). The srb/ssn mutations synergize strongly with mig1 to relieve glucose repression of SUC2 (Vallier and Carlson, 1994).

In this work, we have identified a new mechanism for repression of SUC2 that directly involves Srb/mediator proteins. We recovered the SFL1 gene as a suppressor of srb9. We show that the Sfl1 protein functions as a repressor, binds to a repression site near the SUC2 TATA sequence, and interacts functionally and physically with Srb/mediator proteins.


Increased dosage of the SFL1 gene suppresses srb/ssn mutations

While cloning the SRB9 gene (Song et al., 1996), we recovered a low‐copy suppressor of the srb9 mutation. Our cloning strategy took advantage of the flocculent phenotype conferred by srb9 and the synergy between srb9 and mig1 in relieving glucose repression of SUC2 (see Materials and methods). We transformed an srb9 mig1 strain with a library in a centromere vector and recovered clone A45‐3, which suppressed both phenotypes (Figure 1). Subcloning and sequencing identified the gene as SFL1, which encodes a 767‐amino‐acid protein with homology (residues 65–142 and 182–205) to the conserved DNA‐binding domain of heat‐shock transcription factors (Fujita et al., 1989).

Figure 1.

Restriction maps of SFL1 plasmids. Only the inserted yeast DNA is shown. Plasmids were tested for complementation of the flocculent phenotype of strain MCY3304 (srb9 mig1 snf1). Clones A45‐3 and pWS6 were also tested for complementation of the SUC2 repression defect, and the invertase activities in glucose‐grown transformants were 4.4 and 16 U, respectively, compared with 53 U in cells carrying the vector YCp50. The invertase activity in a mig1 snf1 mutant was 4 U (Song et al., 1996). Restriction sites: B, Bsp120I; C, ClaI; P, PvuII; S, SalI; X, XhoI.

To test whether SFL1 suppresses defects associated with other srb/ssn mutations, we used pWS6 (Figure 1) to transform strains carrying each of the mutations srb8, srb10, srb11, sin4 and rox3 in a mig1 mutant background. pWS6 partially suppressed the flocculent phenotypes of all the mutants and, except in the case of srb10, their defects in glucose repression of SUC2 (Table I). A likely explanation for the lack of suppression of srb10 is that Sfl1 is unstable in this mutant background; tagged Sfl1 proteins were smaller than full size in the srb10 mutant (data not shown). Suppression was not dependent on the presence of mig1, as pWS6 also suppressed the SUC2 repression defect caused by a single srb11 mutation.

View this table:
Table 1. Suppression of srb/ssn mutations by low‐copy SFL1

Disruption of SFL1 confers phenotypes similar to those of srb/ssn mutations

To disrupt the SFL1 gene, we introduced deleted alleles (Figure 1) into wild‐type haploid strains; the gene is not essential for viability (Fujita et al., 1989). The sfl1Δ mutation caused flocculence and a slight defect in glucose repression of SUC2, and synergized with mig1Δ to relieve glucose repression (Figure 2). The sfl1Δ mutation also weakly suppressed the growth defect of a snf1Δ mutant on sucrose (data not shown) and can thus be categorized as an ssn suppressor. No temperature sensitivity, cold sensitivity, or defect in mating, sporulation or derepression of SUC2 was observed.

Figure 2.

Genetic interactions between sfl1Δ and the mutations mig1Δ srb9Δ, srb11Δ and sin4Δ. Bars represent the invertase activity in cells of the indicated genotype grown in SC + 3% glucose. Values are averages of two to four assays. Standard errors were <10%. All strains have the S288C genetic background, and the alleles were sfl1Δ1::HIS3, srb9/ssn2Δ1::URA3, srb11/ssn8Δ2::LEU2, ssn8Δ2::HIS3 and sin4Δ::TRP1. Wild type (WT) was FY250.

We also examined genetic interactions between sfl1 and srb/ssn mutations. sfl1Δ did not synergize with srb9Δ, srb11Δ or sin4Δ to release repression of SUC2 (Figure 2), whereas each of these mutations showed synergy with mig1 (Figure 2; Vallier and Carlson, 1994). In crosses of the sfl1Δ1 mutant to srb/ssn mutants, we observed partial non‐complementation between sfl1Δ1 and srb8 (ssn5‐4) for the flocculent phenotype.

The similar mutant phenotypes, genetic interactions of the mutations and dosage suppression together provide strong genetic evidence for a functional connection between Sfl1 and the Srb/mediator proteins.

DNA‐bound LexA–Sfl1 represses transcription

The genetic evidence suggested that Sfl1 functions in transcriptional repression of SUC2. In addition, overexpression of Sfl1 from the ADH1 promoter reduced derepression of SUC2 by 40% relative to the vector control (Table II). We therefore assayed LexA–Sfl1, containing the LexA DNA‐binding domain fused to Sfl, for the ability to repress transcription of a CYC1–lacZ reporter with lexA operators 5′ to the UAS. LexA–Sfl1 repressed the expression of this reporter 29‐fold in glucose‐grown cells (Figure 3B). In raffinose‐grown cells, no significant repression was detected (Figure 3B) and the LexA–Sfl1 protein was undetectable (Figure 3C).

Figure 3.

LexA–Sfl1 represses transcription. (A) Target plasmids. JK1621 (Keleher et al., 1992) is derived from pLG312ΔS (Guarente and Hoar, 1984). (B) Repression of target gene expression by the indicated LexA protein in wild‐type (WT) and mutant strains. Strains were MCY3647, MCY3817 and MCY1974. Expression plasmids were pSH2‐1, pWS41, pACTII and pWS35. Transformants were grown in SC + 4% glucose (repressed), or SC + 2% raffinose + 0.05% glucose (derepressed). β‐galactosidase activity was assayed in permeabilized cells and expressed in Miller units. Values are average for three transformants. In each case, the fold repression was derived by comparing the β‐galactosidase activity obtained for the target with four lexA operators with the activity for the target with no lexA operator. Standard errors were <10%. (C) The level of LexA–Sfl1 was monitored by immunoblotting using anti‐LexA antibody. Cultures of two independent transformants were collected by centrifugation, and the pellet was resuspended in sample buffer containing 5 mM EDTA and immediately boiled for 5 min. The supernatant was collected after centrifugation and the equivalent of 3 ml of culture at OD600 = 0.5 was used for each lane.

View this table:
Table 2. Overexpression of GAD–Sfl1 and HA3–Sfl1 allows SUC2 expression in glucose‐grown cells

To determine whether repression by Sfl1 requires Srb9, we assayed an isogenic srb9Δ mutant. Repression by LexA–Sfl1 was reduced by a factor of 4 (from 29‐ to 7.6‐fold); immunoblot analysis showed that the level of LexA–Sfl1 protein was not reduced (Figure 3C).

We also examined the dependence on Ssn6–Tup1, which is required for repression of SUC2. Repression by LexA–Sfl1 was abolished in ssn6Δ and tup1Δ mutants (1.3‐ and 1.6‐fold repression, respectively; Figure 3B and data not shown); however, immunoblot analysis of the ssn6Δ strain showed that the level of LexA–Sfl1 protein was ∼4‐fold lower than that in wild type (Figure 3C). The loss of repression and the instability of Sfl1 in the absence of Ssn6 suggest a functional connection.

The ability of Sfl1 to repress transcription distinguishes Sfl1 from the Srb/mediator proteins. LexA fusions to Srb9, Srb10 and Srb11 do not repress this reporter (data not shown). Conversely, DNA‐bound LexA fusions to Srb9, Srb11, Sin4, Rox3 and Gal11 activate transcription of reporters (Himmelfarb et al., 1990; Jiang and Stillman, 1992; Kuchin et al., 1995; Song et al., 1996), whereas LexA–Sfl1 does not function as an activator in such assays (data not shown).

GAD–Sfl1 acts through the element for response to Sfl1 (ERS) site immediately 5′ to the SUC2 TATA box

The presence of a putative DNA‐binding domain in the Sfl1 protein suggested that Sfl1 contributes to repression of SUC2 by binding to the promoter. We reasoned that a Gal4 activation domain (GAD) fusion to Sfl1 might function as a transcriptional activator and thereby facilitate localization of the Sfl1 recognition site. Expression of GAD–Sfl1 from the ADH1 promoter strongly activated SUC2 expression in glucose‐grown cells (Table II) and also caused flocculence and slow growth. These effects of GAD–Sfl1 were also detected in an sfl1Δ mutant, indicating that the native Sfl1 protein is not required. In control experiments, expression of the unfused Sfl1 from the ADH1 promoter did not cause these phenotypes (Table II).

To localize the site of Sfl1 function, we first showed that in glucose‐grown cells GAD–Sfl1 activates a lacZ reporter containing the entire SUC2 upstream region and the HIS3 TATA sequence (pBM3068; Figure 4A). We then tested the effect of GAD–Sfl1 on expression of reporters with the SUC2 UAS and the HIS3 or LEU2 TATA sequence; SUC2 sequences between −650 and −418 are required for wild‐type levels of SUC2 derepression and are sufficient to confer glucose‐regulated expression to a heterologous promoter (Sarokin and Carlson, 1984; Sarokin and Carlson, 1986). GAD–Sfl1 did not significantly affect expression of either reporter (pLS11 and pBM3087; Figure 4A), indicating that the SUC2 UAS does not mediate the effect of GAD–Sfl1. The region critical for the action of GAD–Sfl1 was identified by comparison of pBM3082 and pBM3087, which differ only by the presence of the SUC2 sequence from −222 to −135 (Figure 4A).

Figure 4.

GAD–Sfl1 interferes with transcriptional repression through the ERS site 5′ to the SUC2 TATA box. The upstream region of SUC2 is drawn to scale at the top. In each case, the fold activation was derived by comparing the activity under glucose‐repressing conditions (R) of transformants expressing GAD–Sfl1 with that of transformants expressing GAD. (A) Effects of GAD–Sfl1 on the expression of lacZ reporters under the control of SUC2 sequences. Strain MCY3824, which carries an integrated copy of pLS11, a LEU2–lacZ reporter driven by the upstream region of SUC2 (Sarokin and Carlson, 1985), was transformed with pWS35 (GAD–Sfl1) or vector pACTII (GAD). Strain MCY3647 was co‐transformed with pBM3068, pBM3082 or pBM3087, which carry HIS3–lacZ reporters driven by SUC2 upstream sequences (Ozcan et al., 1997), and either pWS35 or pACTII. Transformants were grown in SC + 4% glucose (R, repressed) and shifted to SC + 0.05% glucose for 3 h (D, derepressed). β‐galactosidase activity was assayed in permeabilized cells and expressed in Miller units. (B) Effects of GAD–Sfl1 on the expression of invertase from a series of deletions at the genomic SUC2 locus. Values for invertase activity of these deletion mutants are taken from Sarokin and Carlson (1984). Mutants were transformed with either pWS53 (GAD–Sfl1) or vector pWS52 (GAD), grown in SC + 4% glucose and assayed for invertase activity. (C) Effects of GAD–Sfl1 and Sfl1–HA on the expression of CYC1–lacZ reporters containing the ERS sequence. Transformants of strain MCY3647 were grown in SC + 4% glucose and assayed as in (A). Expression plasmids were pWS64, its parent vector pSK37, pWS35 and pACTII. Reporters were pLG312ΔS (Guarente and Hoar, 1984) and the indicated derivatives. Values in this figure are averages for two to four transformants; standard errors were <10%.

The importance of this region was confirmed by analysis of upstream deletions in the genomic SUC2 locus. Activation by GAD–Sfl1 required the sequence from −222 to −140; GAD–Sfl1 activated expression of the Δ−403/−223 deletion but had no significant effect on the Δ−418/−140 locus (Figure 4B). The sequence from −222 to −135 is termed ERS. The ERS is located immediately 5′ to the SUC2 TATA box at −133.

Evidence that GAD–Sfl1 interferes with repression of SUC2

Analysis of the SUC2 deletions also revealed that activation by GAD–Sfl1 requires the function of the SUC2 UAS. The level of activation by GAD–Sfl1 in glucose‐grown cells correlated with the integrity of the UAS, as monitored by SUC2 expression in derepressed cells (Figure 4B). These findings indicated that the activation of SUC2 by GAD–Sfl1 does not reflect simple transcriptional activation by the GAD sequence. An alternative possibility was that GAD–Sfl1 acts as a dominant‐negative factor to disrupt a repression mechanism involving the native Sfl1 and its recognition site. The finding that GAD–Sfl1 confers flocculence, a phenotype characteristic of sfl1Δ, also supported this view.

To test this idea, we first determined whether GAD–Sfl1 interferes with repression by LexA–Sfl1. In the presence of GAD–Sfl1, LexA–Sfl1 did not repress much more effectively than LexA87, whereas in the control with GAD, LexA–Sfl1 repressed 8‐fold better than LexA87 (Figure 3B). Secondly, we showed that the GAD moiety is not specifically required for the observed effects; overexpression of HA3–Sfl1, with an N‐terminal triple hemagglutinin (HA) epitope, also activated expression of SUC2 in glucose‐grown cells (Table II) and conferred flocculence. Neither Sfl1–HA4 (tagged at the C terminus), Sfl1–HA nor LexA87–Sfl1 caused either phenotype. Together, these findings strongly suggest that certain N‐terminal modified derivatives of Sfl1, when overexpressed, function to relieve Sfl1‐mediated repression.

ERS site mediates repression by Sfl1

To test directly whether the SUC2 ERS confers repression, we inserted the ERS between the UAS and TATA sequence of CYC1lacZ in pLG312ΔS (pWS84‐13; Figure 4C). Insertion of the ERS sequence reduced expression of β‐galactosidase 24‐fold (from 120 to 5.0 U), and overexpression of Sfl1–HA increased the repression to 64‐fold (from 170 to 2.7 U). In contrast, GAD–Sfl1 alleviated repression by ERS; repression was 36‐fold in the presence of GAD (from 210 to 5.9 U) and only 1.7‐fold in the presence of GAD–Sfl1 (from 53 to 31 U). As observed for other reporters containing ERS, the effect of GAD–Sfl1 was also apparent as activation of expression of pWS84‐13 (5.3‐fold relative to GAD). In control experiments, GAD–Sfl1 did not activate a reporter containing the ERS 5′ to the CYC1 core promoter (pWS116‐2), confirming that the effects of GAD–Sfl1 on pWS84‐13 are not due to transcriptional activation by GAD–Sfl1 bound to ERS.

Analysis of the SUC2 deletions provided further evidence that ERS is a repression site. The deletion Δ−418/−140 partially relieves glucose repression of SUC2, allowing invertase expression in glucose‐grown cells (15 U; Figure 4B). Moreover, the Δ−418/−140 mutation exhibits synergy with mig1Δ; repressed invertase activity in the double mutant was 93 U. In this respect, the deletion behaves similarly to sfl1Δ and srb/ssn mutations (see Figure 2).

Sfl1 binds the SUC2 ERS

Genetic evidence that Sfl1 functions via the SUC2 ERS suggested that Sfl1 binds to this site. We therefore tested whether immobilized HA‐tagged Sfl1 protein from yeast protein extracts can specifically retain 32P‐labeled ERS DNA. Extracts were prepared from glucose‐grown cells expressing Sfl1–HA, Sfl1–HA4 or HA3‐Sfl1, and were incubated with monoclonal anti‐HA antibody. Immune complexes were immobilized onto rProtein A–Sepharose beads and assayed by a DNA‐binding reaction for ability to retain a 32P‐labeled ERS fragment. All three HA‐tagged Sfl1 proteins bound labeled ERS fragment (Figure 5A, lanes 4–6). Control experiments showed that binding requires antibody (lane 3) and HA‐tagged Sfl1; no retention was observed in experiments with HA3, Sfl1 or HA3–Srb10 protein (lanes 1, 2 and 7). Competition experiments showed that this binding was ERS‐specific (Figure 5B). Binding was effectively competed by addition of unlabeled ERS fragment (lanes 3–6) but not by a 50‐fold excess of an unrelated 88 bp fragment [non‐specific (NS)] with identical ends and similar G/C content (Figure 5B, lane 7). Moreover, labeled NS fragment was not retained by Sfl1–HA4 in a binding assay (lane 8). Thus, Sfl1 binds specifically to the ERS sequence in vitro. It is possible that other proteins co‐purify with Sfl1 and contribute to this binding.

Figure 5.

Sfl1 binds the ERS sequence. Protein extracts were prepared from glucose‐grown strain MCY3806 (sfl1Δ1::HIS3) expressing the indicated proteins from plasmids pWS93, pWS42, pWS64, pWS94, pWS96 and pSK84 (expresses HA3–Srb10 from vector pWS93; gift of S.Kuchin). HA‐tagged proteins were immunoprecipitated with anti(α)‐HA and then were assayed for their ability to retain 32P‐labeled ERS sequence in a DNA‐binding reaction, as described in Materials and methods. (A) Anti‐HA antibody was added to the binding reactions, except for lane 3, in which no antibody was added. (B) Probes were 32P‐labeled ERS sequence (lanes 1–7) and the non‐specific NS sequence (see Materials and methods; lanes 8 and 9); free probes are shown in lanes 1 and 9. DNA‐binding reactions were carried out using protein extracts prepared from MCY3806 expressing HA3–Srb10 (lane 2) or Sfl1–HA4 (lanes 3–8). Unlabeled ERS DNA (specific competitor) was added to the binding reaction in 4×, 15× or 50× molar excess relative to the labeled probe (lanes 4–6); unlabeled NS DNA (non‐specific competitor) was added in 50× excess (lane 7). Autoradiograms are shown.

Sfl1 co‐immunoprecipitates with Srb9, Srb11, Sin4 and Rox3

The genetic interactions between SFL1 and srb/ssn alleles, together with the binding of Sfl1 to a site adjacent to the TATA sequence, suggested the possibility of direct interaction between Sfl1 and Srb/mediator proteins that are associated with RNA polymerase II holoenzyme. To test for physical interaction, we carried out co‐immunoprecipitation experiments. Extracts were prepared from glucose‐grown cells expressing Sfl1–HA4 and a LexA fusion to Srb9, Srb11, Sin4 or Rox3. Sfl1–HA4 was immunoprecipitated with monoclonal HA antibody, and the precipitates were analyzed by immunoblotting with LexA antibody. All four LexA fusion proteins co‐immunoprecipitated with Sfl1–HA4 (Figure 6A–C). In control experiments, very little or no LexA fusion protein was precipitated when an untagged Sfl1 protein was expressed; moreover, the control protein LexA–Snf6 did not co‐immunoprecipitate with Sfl1–HA4 (although LexA–Snf6 was weakly detected after long exposure). Nor did we detect any co‐precipitation of Sfl1–HA4 and LexA–Srb9 if an unrelated mouse monoclonal antibody or no antibody was used (data not shown).

Figure 6.

Srb9, Srb11, Sin4 and Rox3 co‐immunoprecipitate with Sfl1. Protein extracts (250 μg, except 500 μg for LexA–Snf6 control) were prepared from a glucose‐grown wild‐type strain FY250 expressing the indicated fusion proteins. (A) HA‐tagged proteins were immunoprecipitated (IP) with α‐HA, separated by 7% SDS–PAGE and immunoblotted with α‐LexA. The LexA–Srb9 lanes are from an independent experiment; the predominant degradation product of LexA–Srb9, marked by an asterisk, co‐migrates with a cross‐reacting IgG band that is visible in the second lane and was also visible in the LexA–Snf6 control lane for this experiment (not shown). (B) Immunoblot analysis of the input proteins (25 μg, except for 50 μg for the LexA–Snf6 control). (C) The immunoblot shown in (A) was reprobed with α‐HA to confirm the precipitation of Sfl1–HA4. (D) HA‐tagged proteins were immunoprecipitated, separated by SDS–PAGE and immunoblotted with α‐LexA. The positions of LexA–Sfl1 polypeptides are marked by arrows. The immunoblot shown here was reprobed with α‐HA to confirm the precipitation of HA3–Srb9, HA3–Sin4 and HA3–Srb11 (not shown). (E) Immunoblot analysis of the input LexA–Sfl1 protein (25 μg) used in (D). Proteins were expressed from (A–C) pWS94, pWS42, pWS54, pWS125, pIT220 (LexA–Rox3; Song et al., 1996), pSK32 (LexA‐Srb11; Kuchin et al., 1995) and pLexA–SNF6 (Laurent and Carlson, 1992) and (D and E) pWS93, pWS121, pWS98 and pSK86 (expresses HA3–Srb11 from vector pWS93; gift of S.Kuchin). Positions of the size standards (kDa) are marked.

We further tested for co‐immunoprecipitation of LexA–Sfl1 with HA3–Srb9, HA3–Srb11 and HA3–Sin4. Extracts were prepared from cells expressing each pair of proteins, monoclonal HA antibody was used to immunoprecipitate the HA‐tagged protein, and the precipitates were analyzed by immunoblotting with LexA antibody. LexA–Sfl1 co‐immunoprecipitated with all three HA‐tagged proteins, but did not precipitate when only the triple HA tag was expressed (Figure 6D and E).


Previous studies have shown that Srb/mediator proteins contribute to transcriptional repression of SUC2. Here we present genetic and biochemical evidence that the Sfl1 protein is functionally related to Srb/mediator proteins and that Sfl1 represses transcription of SUC2 via the ERS site 5′ to the TATA sequence.

Several lines of evidence support the view that Sfl1 is functionally related to Srb/mediator proteins. First, we recovered the SFL1 gene as a low‐copy suppressor of srb9 and showed that it also suppresses srb8, srb10, srb11, sin4 and rox3 mutations for flocculence and/or SUC2 regulation. Secondly, the sfl1Δ deletion resembles srb/ssn mutations in causing similar phenotypes and showing synergy with mig1 for release of glucose repression of SUC2. In contrast, sfl1Δ shows no synergy with srb9, srb11 or sin4, consistent with a related function. Thirdly, transcriptional repression of a reporter by DNA‐bound LexA–Sfl1 was partly dependent on Srb9. Finally, Sfl1 co‐immunoprecipitated with tagged Srb9, Srb11, Sin4 and Rox3 proteins. These findings indicate that Sfl1 interacts with complexes containing Srb/mediator proteins. Sfl1 has not been reported as an integral component of such complexes.

The genetic effects of sfl1Δ on SUC2 expression, together with the ability of LexA–Sfl1 to repress a reporter, implicate Sfl1 in transcriptional repression of SUC2. We mapped the sequence that mediates Sfl1 function to the ERS 5′ to the SUC2 TATA box. Several lines of genetic evidence indicate that the ERS is a site for repression by Sfl1. Deletion of the ERS partially relieves glucose repression of SUC2, and this deletion, like sfl1Δ, acts synergistically with mig1Δ. Moreover, insertion of the ERS between the UAS and the TATA sequence confers repression to a CYC1–lacZ fusion, and this repression is relieved by the overexpression of GAD–Sfl1. The sequence homology of Sfl1 to the DNA‐binding domains of heat‐shock transcription factors suggested that Sfl1 binds to the SUC2 promoter, and we showed that HA‐tagged Sfl1, when purified from cell extracts, specifically binds the ERS DNA sequence in vitro. These studies support a model in which Sfl1 binds to the ERS, perhaps in conjunction with other DNA‐binding proteins, and functions to repress SUC2 transcription. This Sfl1‐dependent repression is complementary to other repression mechanisms that involve Mig1, Mig2 and the Ssn6–Tup1 complex. The regulation of Sfl1 function by the glucose signal remains to be examined; both LexA–Sfl1 and Sfl1–HA4 are difficult to detect in extracts from glucose‐limited cells (W.Song, unpublished results), which may reflect the operation of a regulatory mechanism.

Why does overexpression of GAD–Sfl1 or HA3–Sfl1 dramatically relieve repression of SUC2, whereas loss of Sfl1 causes only a minor effect? A possible explanation is that one or more proteins function redundantly with Sfl1, and overexpression of GAD–Sfl1 or HA3–Sfl1 has a dominant‐negative effect on their function. The yeast genome includes four genes encoding proteins with similar DNA‐binding domains: HSF1, MGA1, SKN7 and YJR147w. Alternatively, it is possible that GAD–Sfl1 and HA3–Sfl1 relieve repression by interacting with Srb/mediator proteins and interfering with a general repression mechanism.

We have shown that Sfl1 binds to a repression site near the SUC2 TATA sequence, that Sfl1 contributes to transcriptional repression of SUC2, and that Sfl1 interacts functionally and physically with Srb/mediator proteins. We propose that Sfl1, when bound to its site, interacts with Srb/mediator proteins to repress transcription (Figure 7). It is unlikely that Sfl1 serves primarily to recruit Srb/mediator proteins to the promoter because previous studies have implicated such recruitment in transcriptional activation (Barberis et al., 1995; Farrell et al., 1996); rather, the interaction of Sfl1 with these proteins must have a specific inhibitory effect. Many steps in the transcription process are possible targets for repression (Johnson, 1995; Hanna‐Rose and Hansen, 1996), and a variety of mechanisms can be envisioned. Sfl1 may play an active role in a mechanism by which certain Srb/mediator proteins inhibit transcription; for example, Sfl1 may modulate the activity of the Srb10–Srb11 kinase, which has a role in CTD phosphorylation (Liao et al., 1995). Alternatively, the physical interaction of Srb/mediator proteins with DNA‐bound Sfl1 may block interactions with other proteins or restrict conformational changes in the holoenzyme, thereby hindering a step in the transcription process such as assembly of a functional complex, initiation or promoter clearance. Another model is that Sfl1 binds tightly to Srb/mediator proteins and simply restrains RNA polymerase II holoenzyme from leaving the promoter.

Figure 7.

Model for interaction of Sfl1 with Srb/mediator proteins at the SUC2 promoter. The Sfl1 protein is shown bound to the ERS site, where it interacts with Srb/mediator proteins to repress transcription. Possible mechanisms are discussed in the text. Other proteins may bind to the ERS with Sfl1. The RNA polymerase II holoenzyme is depicted with the mediator associated with its CTD. The relationship of Srb8–Srb11 to the mediator is controversial (Hengartner et al., 1995; Myers et al., 1998), so these proteins are shown as a separate, associated complex. Other general transcription factors (GTFs) are not represented in detail. Not drawn to scale.

Materials and methods

Strains, plasmids and genetic methods

S.cerevisiae strains are listed in Table III. Standard methods for yeast genetic analysis and transformation were followed (Rose et al., 1990). Selective synthetic complete (SC) medium was used to maintain selection for plasmids. Plasmids are listed in Table IV. pWS35, 41, 42, 53 and 96 were constructed with a BamHI PCR fragment produced from template pWS6. For pWS64 and pWS94, a BamHI PCR product encoding Sfl1 with an added C‐terminal HA sequence was used. pWS93 is a derivative of pSH2‐1 in which the LexA87 coding sequence between the HindIII and EcoRI sites has been replaced by a BglII site followed by a sequence encoding HA3; in addition, the HIS3 marker was replaced with URA3. pWS98 and pWS125 contain a BamHI fragment encoding Sin4 (Song et al., 1996) cloned into the BamHI site of pWS93 and pSH2‐1, respectively. pWS121 was made by cloning a BamHI–SalI fragment containing SRB9 into the cognate sites of pWS93. pWS54 has been described previously as pLexA–SSN2 (Song et al., 1996). The Escherichia coli strain used was XL1‐Blue.

View this table:
Table 3. List of S.cerevisiae strains
View this table:
Table 4. Expression plasmids constructed for this study

Isolation of SFL1 as a suppressor of srb9

A genomic library in the centromere vector YCp50 (Rose et al., 1987) was used to transform the srb9/ssn2‐4 mig1 strain MCY3304 (the snf1 allele is irrelevant for this study). We enriched for non‐flocculent transformants by differential sedimentation (Song et al., 1996), and plated for single colonies. Non‐flocculent colonies were identified and tested for recovery of flocculence after selection on 5‐fluoroorotic acid for plasmid loss. Plasmids were isolated by passage through bacteria. When used to retransform MCY3304, clone A45‐3 complemented the defect in repression of SUC2.

Disruption of chromosomal SFL1 locus

pWS6 was made by deleting the XhoI fragment in clone A45‐3. pWS17‐4 is pWS6 with the SmaI–NruI fragment deleted from the YCp50 backbone. The Bsp120I fragments (1.6 kb) in pWS17‐4 were then replaced with a Bsp120I–Eag. HIS3 fragment or a Sma. URA3 fragment, generating pWS24‐2 or pWS34‐27, respectively. The PvuII fragments from these plasmids were used to disrupt the genomic locus, yielding the alleles sfl1Δ1::HIS3 and sfl1Δ2::URA3.

β‐galactosidase and invertase assays

Cultures were grown to mid‐log phase. β‐galactosidase activity was assayed in permeabilized cells and is expressed in Miller Units (Guarente, 1983). The invertase activity was assayed as described previously (Vallier and Carlson, 1994) and is expressed as μmol glucose released per min per 100 mg cells (dry weight).

Co‐immunoprecipitation assays

Preparation of protein extracts and immunoprecipitation were essentially as described previously (Yang et al., 1992). The extraction buffer was 50 mM HEPES pH 7.5, 150 mM NaCl, 0.1% Triton X‐100, 5 mM EDTA, 1 mM dithiothreitol, 10% glycerol, containing 2 mM phenylmethylsulfonyl fluoride (PMSF) and Complete protease inhibitor cocktail (Boehringer Mannheim). rProtein A immobilized on Sepharose beads (RepliGen) was added to protein lysates, which were rotated for 20 min and cleared by centrifugation at 12 000 r.p.m. for 10 min. Monoclonal mouse anti‐HA antibody (12CA5) was added, and samples were mixed for 30 min and cleared by centrifugation for 5 min at 10 000 r.p.m. The supernatant was mixed with immobilized rProtein A for 1.5 h. The beads were collected by brief centrifugation and washed four times in 1 ml extraction buffer containing 1 mM PMSF by rotating for 10–15 min. The procedure was done at 4°C or on ice. Proteins were separated by SDS–PAGE and blotted. Primary antibodies were anti‐LexA (gift of C.Denis) and were detected by enhanced chemiluminescence with ECL reagents (Amersham).

DNA‐binding assays

The ERS DNA probe, containing SUC2 nucleotides −221 to −135, was prepared by PCR with template pRB58 and primers U221T32, 5′‐GGAATTCTCGAGCTCTATAGTAAACCATTTGG‐3′ and U135B31, 5′‐GGAATTCTCGAGTTTCTTTTCAGGAGGAAGG‐3′ (added XhoI sites are underlined). The NS fragment contains nucleotides 1127–1214 from the SUC2 coding region, and was prepared by PCR with the same template and primers SUC1127T, 5′‐GGAATTCTCGAGTTTATTACAATGTCGATTTGAGCAAC‐3′, and SUC1214B, 5′‐GGAATTCTCGAGTTAAATATGGTTTGTGTGGTGTTAACAGC‐3′. Products were digested with XhoI, gel purified and labeled with Klenow fragment (New England Biolabs) to a specific activity of 4–5×104 c.p.m./ng. Protein extracts were prepared from transformants grown in selective SC+4% glucose. The extraction buffer was 50 mM HEPES pH 8.0, 100 mM NaCl, 5 mM EDTA, 10% glycerol, containing 2 mM PMSF and Complete protease inhibitor cocktail (Boehringer Mannheim). Monoclonal anti‐HA (0.5 μl per 50 μg of protein extract) was added to the protein extracts and mixed for 30 min. rProtein A immobilized on Sepharose beads was added and mixed for 2 h. Beads were collected by centrifugation at 3000 r.p.m. for 10 s and washed with 1 ml of extraction buffer lacking Complete protein inhibitor cocktail. For each assay, an aliquot of beads (8–10 μl) which had been incubated with 80 μg (Figure 5A) or 60 μg (Figure 5B) of protein was then incubated with 32P‐labeled ERS (1 ng) in 50 μl of DNA‐binding reaction buffer containing 50 mM HEPES pH 8.0, 100 mM NaCl, 1 mM EDTA, 1 mM PMSF, 1 mg/ml BSA, 10 μg/ml poly(dI–dC) · poly(dI–dC), and 10% glycerol (Sorger and Nelson, 1989). After mixing for 2 h at 4°C, the beads were collected by centrifugation at 3000 r.p.m. for 10 s. The beads were washed twice in 50 mM HEPES pH 8.0, 100 mM NaCl, 1 mM EDTA, 1 mM PMSF and 10% glycerol by mixing at 4°C for 15 min. The beads were collected and resuspended in sample buffer (50 μl). After extraction with phenol:chloroform:isopropanol (25:24:1), DNA was subjected to electrophoresis on a 5% native polyacrylamide gel in 89 mM Tris‐borate, 2 mM EDTA, pH 8.3. Gels were dried and autoradiographed.


We thank S.Kuchin, S.Ozcan and M.Johnston for providing strains and plasmids, and C.Denis for providing LexA antibody. This work was supported by NIH grant GM47259 to M.C.