Organizational constraints on Ste12 cis-elements for a pheromone response in Saccharomyces cerevisiae


  • Ting-Cheng Su,

    1.  Department of Biochemistry and Molecular Biology, Molecular Epigenetics, LSI, University of British Columbia, Vancouver, Canada
    2.  Graduate Program in Genetics, University of British Columbia, Vancouver, Canada
    Search for more papers by this author
  • Elena Tamarkina,

    1.  Department of Biochemistry and Molecular Biology, Molecular Epigenetics, LSI, University of British Columbia, Vancouver, Canada
    Search for more papers by this author
  • Ivan Sadowski

    1.  Department of Biochemistry and Molecular Biology, Molecular Epigenetics, LSI, University of British Columbia, Vancouver, Canada
    Search for more papers by this author

I. Sadowski, Department of Biochemistry and Molecular Biology, University of British Columbia, 2350 Health Sciences Mall, Vancouver, BC, V6T 1Z3, Canada
Fax: +1 604 822 9311
Tel: +1 604 822 4524


Ste12 of Saccharomyces cerevisiae binds to pheromone-response cis-elements (PREs) to regulate several classes of genes. Genes induced by pheromones require multimerization of Ste12 for binding of at least two PREs on responsive promoters. We have systematically examined nucleotides of the consensus PRE for binding of wild-type Ste12 to DNA in vitro, as well as the organizational requirements of PREs to produce a pheromone response in vivo. Ste12 binds as a monomer to a single PRE in vitro, and two PREs upstream of a minimal core promoter cause induction that is proportional to their relative affinity for Ste12 in vitro. Although consensus PREs are arranged in a variety of configurations in the promoters of responsive genes, we find that there are severe constraints with respect to how they can be positioned in an artificial promoter to cause induction. Two closely-spaced PREs can induce transcription in a directly-repeated or tail-to-tail orientation, although PREs separated by at least 40 nucleotides are capable of inducing transcription when oriented in a head-to-head or tail-to-tail configuration. We characterize several examples of promoters that bear multiple consensus PREs or a single PRE in combination with a PRE-like sequence that match these requirements. A significant number of responsive genes appear to possess only a single PRE, or PREs in configurations that would not be expected to enable induction, and we suggest that, for many pheromone-responsive genes, Ste12 must activate transcription by binding to cryptic or sub-optimal sites on DNA, or may require interaction with additional uncharacterized DNA bound factors.


electrophoretic mobility shift assay


filamentous response element


mitogen-activated protein kinase


pheromone response element


relative competition strength


Tec1 binding site


Ste12 protein of the budding yeast Saccharomyces cerevisiae has attracted considerable interest as a model eukaryotic transcription factor because, much like metazoan factors with a similar function, it regulates multiple distinct classes of genes in response to combinations of signal transduction pathways. In haploid yeast, Ste12 activates genes required for mating between MATa and MATα cells to form diploids, in response to peptide pheromones produced by the opposite mating type [1]. Ste12 also activates genes necessary for filamentous growth in response to nutrient limitation in a process known as invasive or pseudohyphal growth. In both cases, Ste12 activity is regulated by two inhibitor proteins, Dig1 and Dig2 [2], whose functions are considered to be antagonized by a prototypical mitogen-activated protein kinase (MAPK) signaling cascade [3–5]. Genes induced by pheromones include those that encode many of the mating pheromone response pathway components, proteins that cause G1 cell cycle arrest along with the morphological alterations necessary for mating, and gene products that eventually contribute to down-regulation of the pheromone response, allowing re-entry into the cell cycle following mating [6,7]. Nutrient limitation induces filamentous growth through up-regulation of genes that alter cell cycle progression, budding pattern, formation of an elongated cellular morphology, increased agar invasiveness and enhanced cellular adhesion [8,9]. The regulation of this response involves Ste12 in combination with a host of additional DNA bound factors, including Tec1, Phd1, Flo8 and Sok2 [10], through signals transmitted by the pheromone response MAPK, Ras-cAMP-protein kinase A and Snf1/AMP-activated protein kinase pathways [11,12].

The capacity of Ste12 to activate these multiple distinct classes of genes in response to pheromone and nutrient signals is considered to involve the binding to DNA at pheromone-response elements (PREs), with the consensus 5′-ATGAAACA-3′ [13] in combination with additional factors bound to adjacent sites [7,14–16]. For example, the function of Ste12 with respect to the activation of genes involved in filamentous growth requires interaction with another transcription factor, Tec1 [17]. Some filamentous response genes have a PRE adjacent to a Tec1 binding site (TCS) element, and this combination of cis-elements is designated a filamentous response element (FRE) [15]. Ste12 and Tec1 bind cooperatively to FREs from the TEC1, FLO11 and TY1 promoters in vitro [15]. Several different classes of genes can also be distinguished amongst the pheromone-responsive genes. MATa and MATα-specific pheromone-inducible genes, including those encoding the peptide-mating pheromones and their receptors, appear to be regulated by Ste12 bound to DNA in combination with Mcm1 and α1 protein, respectively [14,16]. By contrast, pheromone-responsive genes common to both MATa and MATα haploids are considered to require multimerization of Ste12 for binding to multiple adjacent PREs. Additionally, genes that become activated later during the pheromone response, such as KAR3 and PRM2 involved in karyogamy, may be regulated by Ste12 in combination with Kar4, whose expression is itself induced by a pheromone [18].

Despite having served as an important model for eukaryotic signal-responsive transcription factors for several decades, there is presently little mechanistic or structural information available regarding how Ste12 forms multimers and interacts with additional factors for the regulation of these different classes of genes. Global localization of Ste12 indicates that there are more than 800 target genes in untreated cells [7,19,20], presumably representing those involved in both pheromone and filamentous responses. It is generally accepted that Ste12 activates genes for the filamentous response when bound cooperatively to DNA at PREs closely positioned to a binding site for Tec1 [15,21]. However, an examination of the arrangement of Ste12 and Tec1 binding sites in promoters of this class reveals a variety of spacing and orientations between PREs and TCS elements, and the FRE-like orientation as characterized from the TY1 and TEC1 promoters is quite rare. An implication of this observation is that cooperative interaction between Ste12 and Tec1 must be accommodated by a variety of orientations between their sites. Similarly, haploid-specific pheromone-responsive genes, common to both MATa and MATα haploid cells, are presumed to be solely activated by Ste12 multimers bound to adjacent PREs [2]. Global expression analysis indicates that more than 200 genes become induced within 30 min of treatment with mating pheromone [6,7]. Examination of the promoters of a group of the most strongly induced pheromone-responsive genes does not reveal a simple correlation between either the number or arrangement of predicted consensus pheromone response elements (PREs) and the relative level of inducibility (Fig. 1), and there are also a significant number of pheromone-induced genes that appear to completely lack PREs (not shown in Fig. 1) [6,7]. It might be concluded that there are few restrictions on the arrangement of multiple PREs to enable cooperative interaction for DNA binding of Ste12 for activation of pheromone response. Most analyses of Ste12- and pheromone-responsive transcription have been performed in the context of the FUS1 promoter, which contains four PREs within a 100 nucleotide upstream sequence (Fig. 1), and whose expression is strongly induced in both MATa and MATα haploid cells in response to α- and a-factor, respectively [13,22]. Within the FUS1 promoter, a single PRE was found to confer some responsiveness to pheromone, although a minimum of two were shown to be necessary for a significant response. Deletion of all four PREs eliminated the response to pheromone, and the response could be restored by insertion of oligonucleotides bearing the PRE consensus [13]. The contribution of spatial and orientation differences between multiple PREs to produce pheromone response was not examined in this previous study and, in any case, the experiments were performed using high copy reporter genes, making it difficult to compare requirements for the expression of chromosomal genes.

Figure 1.

 Organization of a selection of strongly inducible pheromone-responsive promoters. Schematic representation of the organization of consensus PREs within nine of the 35 most strongly induced pheromone response genes (excluding pseudogenes and genes without obvious PREs), as identified by global expression analysis (30 min of α-factor treatment) [6,7]. Numbers between any two PREs indicate the spacing in nucleotides, whereas the number furthest to the right indicates the distance to the translation start site. The promoters are arranged in the relative order of inducibility (top to bottom). STE12 is within the top 100 pheromone-inducible genes, and was included here because we have examined this promoter in some detail.

Given the apparently relaxed organizational requirements for PREs on pheromone-responsive genes, we expected that it should be relatively trivial to produce artificial pheromone-responsive promoters. Instead, in the present study, we find that there are rather stringent constraints on how two consensus PREs can be positioned within a minimal artificial promoter to enable a response to pheromone. Wild-type Ste12 binds to a single PRE as a monomer in vitro, and a minimum of two PREs positioned in specific orientations are necessary to cause induction in vivo. We find that there is a direct linear relationship between the response to pheromone and the combined strength of the two PREs positioned in an optimal orientation. Many natural pheromone-responsive promoters do not possess PREs in optimal orientations [7] and, for these genes, we propose that Ste12 must activate transcription when bound to cryptic or sub-optimal sites, or in cooperation with additional uncharacterized transcription factors.


Recombinant wild-type Ste12 binds as a monomer to a single PRE in vitro

Several previous studies have examined the binding of recombinant maltose-binding domain-Ste12 fusions or Ste12 DNA-binding domain fragments to an FRE [15], or the FUS1 promoter in vitro [23]. We have expressed 6-His-Ste12 in insect cells using baculovirus, and found that the protein is capable of forming complexes in vitro with an oligonucleotide (S26D) containing two directly-repeated PREs from the FUS1 promoter, previously shown to be capable of conferring pheromone-responsiveness in vivo (Fig. 2A, lane 2). Antibodies recognizing various Ste12 regions inhibit the formation of the complex (Fig. 2A, lanes 8–10) but not control antibodies (Fig. 2A, lane 11). Additionally, competition with unlabeled wild-type S26D oligo inhibits complex formation (Fig. 2A, lane 3) but not competition with an oligonucleotide bearing a cis-element for an unrelated transcription factor (Fig. 2A, lane 6), demonstrating that recombinant wild-type Ste12 protein produced in insect cells forms a sequence-specific interaction with a PRE-containing oligonucleotide in vitro. The complex that we observed in an electrophoretic mobility shift assay (EMSA) likely represents the binding of Ste12 to a single PRE on the oligo because competition with an unlabeled competitor bearing a mutation of only one of the PREs does not prevent its formation (Fig. 2A, lane 4), although a competitor bearing mutations of both PREs does not compete for binding of Ste12 (Fig. 2A, lane 5). Furthermore, oligonucleotide probes containing only a single PRE produce a complex with identical mobility to that produced by oligos with two PREs (not shown; Figs 3 and 4). Recombinant full-length Ste12 appears to have an autoinhibitory effect because the addition of greater concentrations of protein causes the loss of DNA binding activity altogether (not shown), rather than producing multiple complexes. This effect appears to require the C-terminus because a truncated derivative lacking the C-terminal 73 amino acids is able to form multiple complexes on this same probe (not shown). By contrast, recombinant Ste12 and Tec1, both produced in insect cells, are capable of binding individually to an FRE-containing oligonucleotide in vitro (Fig. 2C, lanes 1 and 2), and form a higher-order complex when added together in binding reactions (Fig. 2C, lane 3). This indicates that recombinant Ste12, although capable of forming terniary complexes with Tec1 in vitro, is excluded from forming multimerized complexes with two closely-spaced PREs in vitro, which indicates that the binding of wild-type Ste12 to multiple PREs in vivo may require additional factors or post-translational modifications. We are currently investigating the significance of this feature with respect to the pheromone response, and we discuss the implications of these observations below.

Figure 2.

 Recombinant Ste12 produced in insect cells binds to a single PRE in vitro. (A) EMSA reactions were performed with extracts of Sf21 insect cells producing recombinant Ste12 protein (lanes 2–11) or uninfected cells (lane 1) using an oligonucleotide probe containing two directly-repeated PREs (sites II and III from the FUS1 promoter, S26D). Unlabeled oligonucleotide competitor oligos were added at ten-fold molar excess (lanes 3–5), as indicated in (B). The binding reaction in lane 6 contained a ten-fold molar excess of an RBEIII oligonucleotide [37]. Antibodies against Ste12 (lanes 8–10) or preimmune serum (lane 11) were added to the binding reactions. (C) Full-length recombinant Ste12 and Tec1 form a complex on an FRE in vitro. EMSA reactions using a labeled FRE probe (CN140/141) derived from the TY1 LTR were performed with Ste12 (lane 1), Tec1-flag (lane 2) or both Ste12 and Tec-1 flag (lanes 3 and 4). Anti-flag sera were added to the binding reaction in lane 4.

Figure 3.

 Ste12 binds to a PRE as a monomer. (A) EMSA reactions were performed with a labeled oligo containing a single PRE (IS1430/1431) and full-length Ste12 (lane 1), Ste12 1–476 (lane 2), Ste12 1–350 (lane 3) and Ste12 1–215 (lane 4). Full-length Ste12 was mixed with 1–476 (lane 5), 1–350 (lane 6) or 1–215 (lane 7) prior to adding the labeled oligo and performing the binding reaction. (B) Reactions were performed with in vitro translated Ste12 1–476 (lanes 1, 3 and 4), 1–350 (lanes 2, 3, 4, 5, 7 and 8) or 1–215 (lanes 6–8). The Ste12 derivatives were synthesized separately in vitro and then mixed prior to EMSA (lanes 3 and 7) or were co-translated (lanes 4 and 8).

Figure 4.

 Nucleotides required for binding of full-length Ste12 to the consensus PRE in vitro. (A) EMSA reactions were performed with recombinant wild-type Ste12 and a labeled oligonucleotide bearing a single consensus PRE (RS010/011). Binding reactions contained no competitor (lane 1), or a 0.625- (lanes 2 and 7), 1.25- (lanes 3 and 8), 2.5- (lanes 4 and 9), 5- (lanes 5 and 10) or 10- (lanes 6 and 11) fold molar excess of unlabeled consensus oligo (lanes 2–6) or the indicated mutant oligos (lanes 7–11). Mutant oligos (lines 1–7) contained a single nucleotide substitution from the consensus PRE (Table S1). (B) The sequence of the FUS1 promoter indicating the position of four PREs (designated sites I, II, III and IV, 5′–3′). EMSA reactions were performed as in (A) but using a labeled oligonucleotide bearing PRE IV (IS1428/1429), and with the unlabeled competitors as indicated. (C) The RCS was calculated for each mutant oligo (Table 1). The effect that mutation of each nucleotide of the consensus PRE has on the binding of Ste12 in vitro is indicated proportional to the font size for each residue.

To determine the stoichiometry of Ste12 bound to a single PRE in vitro, we expressed a series of C-terminal truncations for use in the analysis of hetero-complex formation. Wild-type Ste12 protein produced in insect cells (Fig. 3A, lane 1) or truncated versions of Ste12 containing residues 1–476 (Fig. 3A, lane 2), 1–350 (Fig. 3A, lane 3) or 1–215 (Fig. 3A, lane 4), produced by in vitro transcription and translation, each were capable of forming complexes with a single PRE-containing oligo in EMSA. We then mixed the full-length protein together with the truncated forms in vitro prior to adding the labeled oligonucleotide probe and performing EMSA. In these experiments, none of the truncated species caused the production of an intermediate complex in combination with wild-type Ste12 (Fig. 3A, lanes 5–7), which would be expected if there were multiple protein molecules bound to a single PRE. Because it is possible that co-translation of Ste12 may be necessary for hetero-complex formation, as is the case with proteins such as GCN4 and GAL4 [24,25], we also performed this experiment using co-translation of the truncated Ste12 derivatives (Fig. 3B). We found that when the 1–476 and 1–350 or 1–350 and 1–215 derivatives are produced by co-translation (Fig. 3B, lanes 4 and 8, respectively), we also do not observe intermediate-sized complexes that would indicate formation of hetero-multimers. From these results, we argue that Ste12 protein likely binds to a single PRE as a monomer.

Sequence requirement of the PRE for binding Ste12 in vitro

The sequence requirements for binding of Ste12 to DNA have largely been inferred from a comparative analysis of pheromone-responsive promoters and genomic localization of Ste12 protein in vivo [7,10,19]. To characterize residues of the PRE that are necessary for affinity of Ste12 in vitro, we performed a systematic analysis using competitions with mutant oligonucleotides in EMSA (Fig. 4A). Within the eight nucleotide consensus (ATGAAACA), we found that mutation of each of the residues impairs the ability to compete for binding to the wild-type oligo (Fig. 4A, PRE mutants). In particular, mutations of residues A5 and A6 of the central AAA trinucleotide to G significantly impair competition (Fig. 4A, lines 5 and 6), as does substitution of G3 with a pyrimidine (C or T) (Fig. 4A, line 2; Table 1). We also compared the relative affinities of the four PREs within the FUS1 promoter (Fig. 4B, designated I, II, III and IV, 5′–3′, top). Amongst these, site II is identical to the eight nucleotide consensus, sites III and IV have substitutions of the outer 3′ and 5′ nucleotides, respectively, and site I has a substitution of A5 within the AAA trinucleotide. Using competition experiments, we were able to rank the relative strengths of PREs within the FUS1 promoter as sites II, IV, III and I (Fig. 4B, strongest to weakest; Table 1).

Table 1.   RCS of mutant PREs for binding of wild-type Ste12 to a PRE consensus (ATGAAACA) in vitro.
FUS1 PREaSequenceRCSb
  1. a PREs represented in the FUS1 promoter (Fig. 4B). b RCS for each oligo was calculated from the concentration of unlabeled competitor oligonucleotide required to compete 50% of total Ste12 protein bound to the consensus PRE, relative to competition in the same experiment with a wild-type PRE (Fig. S1). c Concentrations of oligo required for 50% competition was calculated by extrapolation.


Because higher concentrations of recombinant Ste12 produce an autoinhibitory effect, we were unable to determine affinity constants using EMSA with this reagent. However, for each mutant oligonucleotide, we calculated a relative competition strength (RCS) value, which represents the ratio of competitor oligonucleotide required to compete for 50% binding of total Ste12 relative to the consensus oligonucleotide within the same experiment (Fig. S1 and Table 1). From the RCS values, we predict the relative contribution of each nucleotide within the consensus PRE for binding of wild-type Ste12 in vitro, as shown in Fig. 4C.

Relative affinity of Ste12 for PREs in vitro correlates directly with the pheromone response in vivo

To determine by how much the relative affinity of Ste12 for PREs in vitro contributes to the pheromone response in vivo, we inserted oligonucleotides bearing the consensus or mutant PREs into a reporter with a minimal GAL1 core promoter upstream of LacZ, which were integrated in single copy at a lys2 disruption. We found that none of the PREs inserted individually upstream of the GAL1 core element were capable of inducing a response to pheromone, even with the strongest of the PREs from the FUS1 promoter (not shown). By contrast, reporters with an insertion of two identical directly-repeated PREs, in either orientation relative to the transcriptional start site (not shown), and arranged in the same context as FUS1 PREs II and III (Fig. 4B), all produced a response to pheromone and, interestingly, the level of inducibility correlated with the RCS values for the PREs as determined in vitro (Fig. 5A). Accordingly, a duplicated PRE with a substitution of residue A5 of the central AAA trinucleotide to G, which seriously inhibits binding of Ste12 in vitro, produces a small but detectable level of inducibility (Fig. 5A, line 4), whereas the duplicated consensus PRE causes a level of pheromone response comparable to the full FUS1 promoter (Fig. 5A, lines 1 and 5).

Figure 5.

 The pheromone response conferred by two directly-repeated PREs in vivo is proportional to their relative affinity for Ste12 in vitro. (A) Strains bearing single-copy integrations of a minimal GAL1-LacZ reporter bearing two copies of the indicated PRE (lines 1–4) were left untreated (red bars) or treated with α-factor for 60 min (blue bars) prior to harvesting the cells and assaying β-galactosidase activity. The shading of the boxes containing the PRE sequence indicates the relative competition strength for Ste12 in vitro, with the stronger PREs being shaded darker and the weaker PREs shaded lighter. Line 5 shows results from a strain bearing the full FUS1-LacZ promoter. (B) Reporter genes bearing a consensus PRE and PREs containing substitutions of the central AAA trinucleotide were assayed as in (A). (C) Combinations of consensus PREs and PREs bearing the indicated mutations were assayed in the same context as described above.

Because the inducibility of reporters bearing two directly-repeated PREs appeared to be approximately proportional to the relative affinity for Ste12 in vitro, we were interested in determining the extent that mutations of one PRE would have in combination with a strong consensus element. To address this, we introduced mutations of the central AAA trinucleotide into the 3′ PRE of the artificial reporter constructs. Mutation of the central A5 residue of the trinucleotide, causes an approximately three-fold reduction in pheromone inducibility in combination with a consensus PRE (compare Fig. 5B, line 1, with Fig. 5A, line 1). Mutation of two of the central A residues compromises the response by approximately ten-fold (Fig. 5B, line 2), and a PRE bearing substitution of all three A residues completely prevents the response to pheromone (lines 3–5). The latter mutation also completely prevents binding of Ste12 in vitro (not shown) and, in effect, the reporters indicated in lines 4 and 5 of Fig. 5B possess only a single functional PRE. We also examined the effect that mutations in both directly-repeated PREs have on pheromone response, and observed that inducibility was reduced significantly when both elements have mutations that limit binding of Ste12 in vitro. For example, directly-repeated PREs with substitutions of residues A1 and A8, respectively, comprising mutations that have a relatively minor effect on binding Ste12 in vitro, cause an approximately four-fold defect in inducibility relative to two consensus PREs (Fig. 5C, line 5). Combinations of PREs that have more serious defects in binding Ste12 produce proportionally less response (Fig. 5C, lines 6 and 7), although even two quite weak directly-repeated PREs retain a detectable level of inducibility (Fig. 5C, line 8). These results demonstrate that a significant response to pheromone can be conferred by a single strong consensus PRE in combination with much weaker adjacent PREs, with a level of inducibility proportional to the relative strength of the second PRE. Additionally, duplicated PREs with substitutions that inhibit Ste12 binding are capable of inducing a response to pheromone, but at significantly lower levels. Interestingly, when we examined the effect of the combined RCS of two directly-repeated PREs on the response to pheromone, we observed a direct and simple linear relationship between the product of the RCS values and pheromone responsiveness (Fig. 6). This analysis indicates that, in the context of the minimal GAL1 promoter, the limiting factor for transcriptional activation in pheromone-treated cells appears to be binding of Ste12 multimers to DNA.

Figure 6.

 The combined relative strength of two directly-repeated PREs produces a proportionally linear response to pheromone. A combined relative PRE strength for each of the reporter genes described in Fig. 5 was calculated as log(RCSPRE1 × RCSPRE2) and plotted against the respective pheromone responsiveness for each reporter (β-galactosidase activity (× 10−3).

Organizational constraints on multiple PREs for a pheromone response

When examining the promoters of some of the most strongly induced pheromone response genes (Fig. 1), we noted that PREs are arranged in a variety of configurations. Most promoters have PREs in a directly-repeated orientation, although there are many instances of PREs arranged in a tail-to-tail configuration (PRM6, FUS1, AGA1 and STE12). Also, there is considerable variability in spacing between multiple PREs (Fig. 1). To examine the significance that these differences in configuration have for pheromone response, we compared the responses of a GAL1 minimal promoter bearing two consensus PREs positioned at different orientations with respect to each other (Fig. 7). In the FUS1 promoter, two PREs (sites II and III) are positioned in a directly-repeated orientation separated by three nucleotides (Fig. 7A, line 2) (i.e. the same context as the experiments described above). We found that inverting one of the PREs such that they are positioned in a head-to-head orientation completely prevented the response to pheromone (Fig. 7A, line 3). By contrast, two consensus PREs from the STE12 promoter positioned in a tail-to-tail configuration, separated by a single nucleotide, caused considerably greater induction compared to the directly-repeated PREs from FUS1 (Fig. 7A, line 1). This indicates that there are severe organizational constraints for closely-positioned PREs that must limit binding and activation by Ste12. We then examined how the spacing between two directly-repeated consensus PREs affects the observed response, and found that they could not be moved apart without seriously compromising induction (Fig. 7B). Separation of PREs by even one nucleotide completely prevented induction, as did separation by three, five, seven (not shown), 10 or 20 nucleotides (Fig. 7B, lines 4–9). Curiously, however, two PREs spaced 40 nucleotides apart in either a head-to-head or tail-to-tail orientation produced a significant level of pheromone response (Fig. 7B, lines 2 and 3, respectively). Taken together, these results indicate that there must be structural constraints on Ste12 that allow binding to closely-spaced PREs in several different configurations. Additionally, the fact that head-to-head and tail-to-tail PREs separated by 40 nucleotides allow induction implies that a sufficient length of intervening DNA is required to bend or twist into a conformation enabling an interaction between Ste12 proteins bound to these PREs. We discuss the possible implications of these results further below.

Figure 7.

 Organizational constraints on closely-spaced PREs for pheromone response in vivo. (A) Pheromone responsiveness of minimal promoters containing PREs II and III from the STE12 promoter in a tail-to-tail orientation (line 1), directly-repeated consensus PREs from the FUS1 promoter (PRE II, line 2) or with the second consensus PRE inverted into a head-to-head orientation (line 3). (B) The consensus PREs from the FUS1 promoter were moved apart to produce an intervening spacing of ten (lines 7–9), 20 (lines 4–6) or 40 (lines 1–3) nucleotides, with the orientation of the PREs as indicated.

PREs from the STE12 promoter demonstrate organizational constraints

To examine whether the organizational constraints that we observe on artificially produced arrangements of PREs are representative of pheromone-responsive promoters in vivo, we examined the contribution of PREs within the STE12 promoter, which contains four PREs: three in the forward orientation and one in the reverse orientation (Fig. 1, bottom). We found that a sub-fragment bearing only the three 5′ elements (sites I, II and III) caused an elevated level of basal expression, which is dependent upon STE12 (Fig. 8, basal expression, compare lines 1 and 2) and, furthermore, that a sub-fragment bearing only the inverted PREs II and III could account for almost all pheromone inducibility of the STE12 promoter (Fig. 8, pheromone induction, line 1, compare lines 1 and 4). Similarly, mutation of site I had only a small negative effect on the response (Fig. 8, line 3), whereas mutation of either sites II or III completely prevented induction (Fig. 8, lines 5 and 6). These observations indicate that, although PREs may be scattered throughout the promoters of pheromone-responsive genes, in some cases, the majority of pheromone response may involve only two properly spaced and oriented binding sites for Ste12.

Figure 8.

 Orientation and spacing of PREs contributing to response of the STE12 promoter. The sequence of the STE12 promoter region containing the three most distal PREs (designated I, II, and III, 5′–3′) is indicated. An oligonucleotide representing this sequence, or bearing mutations or deletions as indicated, was inserted upstream of the minimal GAL1 core promoter-LacZ reporter gene. The expression of the reporter was measured in untreated cells (basal expression, left) or cells treated with α-factor for 60 min (pheromone induction).

Pheromone response of promoters with a single consensus PRE

Considering the results reported above, we questioned how it is possible that a number of genes amongst those that are strongly induced by pheromone have only a single consensus PRE (Fig. 1) [7]. CIK1, for example, is one of the most strongly induced genes in pheromone-treated cells, and apparently has only a single consensus PRE. We examined the CIK1 promoter to determine whether there were potential weaker binding sites for Ste12 falling within the constraints that we observed on the artificial promoters described above. Accordingly, we noted that the CIK1 PRE is positioned only three nucleotides downstream of a PRE-like sequence with substitution at residues T1 and A6 of the consensus (Figs 4C and 9A, top). A portion of the CIK1 promoter bearing these elements inserted upstream of a minimal promoter was found to be strongly induced by pheromone, although deletion of the PRE-like sequence completely prevented the response (Fig. 9A), indicating that this element does contribute to induction by Ste12 multimers in vivo. Similarly, on the PRM3 promoter, we observed the PRE-like sequence 5′-ATAAAACA-3′ 36 nucleotides upstream of the consensus PRE, positioned in a head-to-head orientation (Fig. 9B). In vitro, we found that an oligonucleotide bearing this sequence competes for binding to Ste12 only slightly less efficiently than does a consensus PRE (Table 1). A region including these elements inserted upstream of the GAL1 core promoter was responsive to pheromone (Fig. 9B, line 1), although the response was reduced considerably when the PRE-like sequence was deleted (Fig. 9B, line 2). These results indicate that this PRE-like sequence can produce a pheromone response by Ste12 multimers oriented in a head-to-head conformation approximately 40 nucleotides away from a consensus PRE, and we had demonstrated this effect with the artificial promoters. Taken together, these results indicate that, for some pheromone-responsive genes, Ste12 must activate transcription from sub-optimal binding sites, in combination with a single consensus PRE whose arrangement falls within specific organizational constrains. We note, however, that we have only examined sub-fragments for both of these promoters, and there are likely to be additional factors that contribute to response. In this vein, it is important to note that both were shown to be Kar4-dependent [18].

Figure 9.

 A single consensus PRE can confer pheromone responsiveness in conjunction with PRE-like sequences. (A) Sequence of the CIK1 promoter region, indicting the consensus PRE and a PRE-like sequence. An oligonucleotide representing this sequence, or bearing a deletion of the PRE-sequence, was inserted upstream of the minimal GAL1 core promoter-LacZ reporter, and expression was measured in untreated and pheromone-treated cells. (B) Sequence of the PRM3 promoter indicating the location of a consensus PRE and PRE-like sequence. The pheromone responsiveness of the minimal promoter bearing oligonucleotides representing the wild-type or mutant promoter sequences was measured in untreated and pheromone-treated cells.


The pheromone response pathway of Saccharomyces has provided an important model for understanding how genes are regulated in response to signals transmitted through MAP kinase cascades. However, despite almost 20 years of intensive research, there remain many unanswered questions regarding the function of Ste12, including the molecular mechanisms that control its activity by upstream MAPKs, how it causes transcriptional activation, and the nature of its interaction with PREs on DNA. To begin addressing the latter issue, we have performed a systematic analysis of Ste12 binding to the PRE in vitro, and studied the relationship between binding affinity and spatial orientation between two PREs for pheromone responsiveness in vivo. Ste12 likely binds to a single PRE in vitro as a monomer, and therefore the protein must require multimerization in vivo to bind DNA and activate the haploid-specific pheromone response because a minimum of two PREs are required.

Surprisingly, based on analysis of artificial promoters containing two PREs, there appear to be serious constraints with respect to how these can be positioned relative to one another to enable pheromone response of an artificial promoter. Two directly-repeated PREs cause activation only when located within three nucleotides of each other. By contrast, PREs inverted in a tail-to-tail conformation separated by a single nucleotide produce a very strong response. Additionally, PREs oriented in head-to-head or tail-to-tail configurations are only able to cause a pheromone response when separated by approximately 40 nucleotides. Taken together, these observations indicate that Ste12 must have structural features that can accommodate multimerization for binding of closely-spaced sites oriented in several different conformations (Fig. 10), such that binding to closely-positioned PREs in either a directly-repeated (Fig. 10A) or tail-to-tail conformation (Fig. 10B) may form multimers through interaction between surfaces on the Ste12 protein that are separated from the DNA-binding domain by a flexible linker in order to accommodate different orientations. Because PREs oriented in a head-to-head manner do not produce a response, the flexibility of Ste12 may not be able to accommodate this particular orientation, or perhaps the N-terminal DNA binding domain is sterically precluded from such an interaction (Fig. 10C). PREs oriented in either a head-to-head or tail-to-tail conformation are capable of inducing a pheromone response if positioned 40 nucleotides apart (i.e. approximately four helical turns of DNA), suggesting that Ste12 is capable of forming multimers that can bind these configurations, provided that the intervening DNA is able to bend or twist into a conformation that can accommodate the interaction (Fig. 10D). An additional possibility is that Ste12 multimerization in vivo, enabling accommodation of various PRE arrangements, may require additional nuclear factors. Accordingly, Ste12 was shown to associate on pheromone response promoters in vivo with both inhibitor proteins Dig1 and Dig2 [2], and so it is possible these proteins facilitate the binding of Ste12 to PREs arranged in various configurations. However, we consider this to be unlikely considering that the activation of Ste12-dependent genes appears to be constitutive in dig1 dig2 null strain backgrounds [3–5], presumably including genes requiring a variety of PRE orientations for a pheromone response.

Figure 10.

 Structural constraints on Ste12 for binding closely-positioned PREs. Schematic representation of a possible mechanism for the recognition of closely-spaced PREs in different conformations by Ste12 multimers. Interaction with directly-repeated PREs, positioned three nucleotides apart (A) or in a tail-to-tail orientation (B) may involve an interaction with C-terminal sequences separated from the N-terminal DNA binding domain by a flexible linker region. Some closely-spaced configurations appear to be excluded from binding Ste12 multimers, as in a closely-spaced head-to-head orientation (C). Head-to-head and tail-to-tail orientations may be accommodated providing that the sites are separated sufficiently to allow bending or twisting of the intervening DNA to enable binding of Ste12 multimers (D).

Curiously, recombinant wild-type Ste12 produced in insect cells is incapable of forming multimers on oligos containing two PREs in vitro, despite the fact that the same arrangement of PREs confers a strong response to pheromone in vivo. Furthermore, full-length Ste12 appears to have an autoinhibitory function because high concentrations of protein completely prevent binding to DNA. Because the deletion of the C-terminus prevents these effects (not shown), we suggest that multimerization of Ste12 in vivo must be regulated through a mechanism involving the C-terminus. Ste12 produced in insect cells becomes phosphorylated on most of the same residues that we have observed in yeast [26,27], and we find that mild treatment with phosphatase in vitro produces slower migrating complexes with oligos containing two PREs (T.-C. Su and I. Sadowski, unpublished results), suggesting that phosphorylation may regulate the ability to bind multiple adjacent PREs. By contrast, recombinant wild-type Ste12 does produce terniary complexes with Tec1 on an FRE-containing oligo in vitro (Fig. 2C). These results suggest that activation of haploid-specific pheromone-responsive genes, but not Ste12/Tec1-responsive genes, may require additional regulation in vivo involving dephosphorylation. The results obtained in the present study also raise the important question of why two PREs are required for pheromone response if wild-type Ste12 is able to bind to a single PRE in vitro. This indicates that either the activation domain of Ste12 is incapable of activating transcription when bound to a single site, or that binding to a single site in vivo is limited by additional factors. Consistent with the latter possibility, it was shown that Ste12 does not interact with filamentous response promoters (containing a single PRE) in the absence of Tec1 [2], indicating that Ste12 is prevented from binding a single PRE in vivo on its own. This effect is likely mediated by the inhibitor proteins Dig1 and/or Dig2 [3,28] and, consistent with this, we found that binding of wild-type Ste12 to a single PRE in vitro is inhibited by the addition of recombinant Dig1 and Dig2 (T.-C. Su and I. Sadowski, unpublished results).

We have systematically examined nucleotides within the PRE by mutagenesis, and have compared the relative affinity of natural sites within the FUS1 promoter for binding of wild-type Ste12 in vitro. Using an artificial reporter bearing two PREs arranged in a directly-repeated orientation, we find that there is a significant and simple linear relationship between the combined relative strength of the two PREs in vitro and the level of pheromone responsiveness in vivo (Fig. 6). This suggests that, in pheromone-treated cells, using a concentration of pheromone where presumably Ste12 is free of inhibition by the regulatory proteins Dig1 and Dig2 [2], the association of Ste12 protein with cis-elements on DNA is probably the limiting interaction for induction, at least in the context of our artificial promoters. However, we envisage that many, if not most, natural promoters controlled by Ste12 will also be subject to the additional effects of nucleosome positioning, which likely would significantly alter the effects produced by combinations of PREs with different affinities for Ste12 protein, as previously shown for transcriptional activation by Pho4 [29,30].

Upon cursory examination of the most strongly induced pheromone-responsive promoters in vivo, it could not be predicted that there should be such severe constraints on the organization of PREs for induction by pheromone (Fig. 1). Most of these promoters appear to have PREs arranged without any particular defined conformation, some promoters appear to only have a single PRE, and other pheromone-responsive promoters have none (not shown in Fig. 1). On the basis of the results obtained in the present study, we expect that many pheromone-responsive genes must rely on nonconsensus weaker binding sites for Ste12, which are positioned adjacent to consensus PREs in a conformation that can accommodate the binding of Ste12 multimers. We have detailed such instances on sub-fragments of the CIK1 and PRM3 promoters (Fig. 9). Both of these genes are also regulated by Kar4 [18], and it will be interesting to determine how these factors interact within the context of their full promoters to promote induction during pheromone response. On several promoters, including FUS1 and STE12, we find that only two PREs oriented in an optimal configuration can account for the majority of pheromone response, and this suggests that many genes strongly induced by pheromone may only require two properly oriented PREs. Many pheromone-responsive promoters bear consensus PREs positioned some distance apart, and the results obtained in the present study indicate that two consensus elements oriented in a head-to-head or tail-to-tail orientation at least 40 nucleotides apart can confer a significant response. Such configurations are observed on many natural pheromone-responsive promoters, including FUS3 and PRM6 (Fig. 1). Furthermore, PRE I of the FUS1 promoter is oriented in a tail-to-tail conformation with respect to the three more proximal sites (II, III, and IV) and, consequently, this may allow activation by Ste12 multimers from any combination of these proximal three sites. Several other promoters, with either a single consensus PRE or with two PREs in orientations that should occlude a pheromone response based on our data, have potential weaker Ste12 binding sites positioned in a tail-to-tail orientation. We find such examples within the FIG1 and PRM4 promoters (Fig. 1).

There also genes that are strongly induced by pheromone but appear to lack a consensus PRE, including many of the PRMs, ASG7, FIG2, FIG3, ECM18 and MCH2 (not shown). In these cases, Ste12 must activate from multiple nonconsensus binding sites or through cooperative interaction on weaker elements with additional DNA binding proteins, such as Mcm1 [16,31] and Kar4 [18], and perhaps with previously unrecognized additional factors. Consistent with this possibility, it was recently shown that there is a strong correlation between the association of Ste12 on pheromone-responsive promoters with potential binding sites for Flo8, suggesting that pheromone response for many genes may involve an association between these factors [7]. Accordingly, it is interesting that the function of Ste12 with respect to activating transcription in response to the pheromone-response MAPK pathway is remarkably similar to TFII-I, which is a protein in mammalian cells that performs this function in response to MAPK signaling downstream of RAS through cooperative interactions on upstream elements with a number of factors, including serum response factor, PHOX1, nuclear factor-κB and upstream stimulatory factor [32,33].

The results reported in the present study demonstrate that many aspects of Ste12 regulation at the molecular level are still not well understood. This protein appears to bind as a monomer to a single PRE in vitro, although at least two properly configured PREs are necessary for a pheromone response in vivo. It will be important to elucidate the structural features of Ste12 that impose these restrictions, as well as the mechanisms controlling the interaction of this factor with multiple PREs in vivo to mediate pheromone response.

Materials and methods

Oligoucleotides, plasmids and yeast strains

Sequences of oligonucleotides for construction of minimal promoter reporters are detailed in Table S2. Oligonucleotides for construction of reporter genes were annealed and cloned into the XhoI/XbaI sites of pIS341, which is a lys2 disintegrator vector [34], bearing the GAL1 core promoter region upstream of LacZ and the ADH1 terminator. All experiments were performed in a W303-1A strain background (MATa ade2 leu2 trp1 ura3 can1). Reporter gene plasmids were linearized by digestion with NruI prior to transformation into yeast using the LiAc technique [35]. URA+ transformants were allowed to grow nonselectively on yeast extract peptone dextrose for 3 days to allow rearrangement of the disintegrator, prior to streaking for single colonies on 5-fluoroorotic acid. Strains bearing reporter gene integrants at the lys2 disruption were identified by replica plating, and single copy integration was verified by analysis of chromosomal DNA using PCR [34]. The pheromone responsiveness of strains bearing the reporter genes was assayed in cultures grown in yeast extract peptone dextrose until A600 of 0.6 was reached. Pheromone was added at a concentration of 2 μg·mL−1. The cells were collected and β-galactosidase activity was assayed as described previously [36]; the results represent an average of three independent experiments.

Recombinant proteins and EMSA

Full-length Ste12 protein was expressed as an N-terminal 6-His fusion in insect cells using baculovirus in the Sf21 insect cell line [26]. Tec1 was expressed with a 6-His-N-terminal and C-terminal flag epitope tag using the Bac-to-Bac system (Invitrogen, Carlsbad, CA, USA). Antibodies A3, B3 and F3 were raised against Escherichia coli TrpE fused to Ste12 residues (265–688), (314–688) and (1–215), respectively. Sf21 cells infected with Ste12 and Tec1 virus were collected and washed in ice-cold lysis buffer (20 mm Tris, pH 8.0, 40 mm NaCl, 1 mm dithiothreitol, 5% glycerol, 2.5 mm MgCl2, 1 mm Na3VO4, 5 mm EGTA, 50 mm NaF and 20 mmβ-glycerol phosphate). The cells were lysed by forcing through a 27-gauge needle ten times, and then sonicated for 10 s. A clarified supernatant was obtained by centrifugation at 10 000 g for 10 min. Ste12 proteins were produced by in vitro transcription and translation using the TNT T7 Quick Coupled Transcription/Translation System (Promega, Madison, WI, USA). Briefly, plasmid pSC4, which contains a full-length genomic clone of STE12, was used as template for amplification with oligonucleotide oIS1144, in combination with oVT2, oET30 and oIS1146 (Table S3), to produce fragments with a 5′ T7 RNA polymerase promoter and encoding Ste12 (1–215), Ste12 (1–350) and Ste12 (1–479), respectively. The Ste12 derivatives were produced individually or by co-translation in 50 μL reactions containing 1 μL of T7 RNA polymerase and 40 μL of rabbit reticulocyte lysate. The reactions were carried out at 30 °C for 90 min and then assayed immediately for DNA binding activity.

Oligonucleotides used for EMSA are detailed in Table S1, and were annealed and labeled using Klenow (New England Biolabs, Beverly, MA, USA) with [32P]αdATP and [32P]αdTTP, as described previously [37]. The 5′ overhangs of unlabeled competitor oligonucleotides were filled in using Klenow and an unlabeled dNTP mixture. EMSA reactions contained 1 μL of labeled oligonucleotide probe (2 pmol), 2 μg of poly(dI-dC) (Sigma, St Louis, MO, USA), 2.5 mm MgCl2, 1% glycerol, 20 mm Tris (pH 8.0), 40 mm NaCl and 1 μL of Sf21 extract or in vitro translation reaction in a total volume of 20 μL. Labeled oligonucleotide probes were added to the binding reactions after a 30 min pre-incubation on ice with unlabeled competitor oligos or specific antibodies. Binding reactions were performed at room temperature for 30 min and the reactions were resolved on nondenaturing polyacrylamide gels containing 0.5 × TBE (89 mm Tris, 89 mm Boric acid, 2 mm EDTA, pH 8.0) buffer and 1% glycerol at 200 V for 3 h. Signals produced in the EMSA reactions were quantitated using imagequant software (GE Healthcare, Milwaukee, WI, USA).


We thank LeAnn Howe, Mike Kobor, Viven Measday, Sheetal Raithatha and Kris Barretto for their helpful comments. This research was supported by funds from the Canadian Cancer Society Research Institute (grant 018436).