Cdk8 Regulates Stability of the Transcription Factor Phd1 To Control Pseudohyphal Differentiation of Saccharomyces cerevisiae

Abstract

INTRODUCTION

Differentiation of eukaryotic cells requires global reprogramming of gene expression in response to environmental and physiological cues in a process that typically involves multiple gene-specific transcriptional regulators controlled by a network of signaling pathways. This is evident in the budding yeast Saccharomyces cerevisiae, which differentiates into filamentous forms in response to limiting nutrients. Diploid cells differentiate from their normal round growth pattern to an elongated cellular morphology in branched filaments of pseudohyphae that extend outwards from the colony and invade the agar, a response that is thought to facilitate foraging for nitrogen (17). Haploid cells produce a similar response, thought to occur in response to limiting carbon, by forming elongated filamentous cells that invade the agar immediately below the colony (36, 40). These differentiation processes, known as pseudohyphal and haploid invasive growth, or collectively as filamentous growth, have similar genetic requirements, and the presence of different colony morphologies between haploids and diploids is primarily related to their distinct budding patterns (36). Differentiation is regulated by a set of sequence-specific DNA binding factors, including Ste12, Tec1, Flo8, Phd1, Mga1, and Sok2 (3, 29), among others, whose activities are responsive to signals transmitted by the pheromone response Kss1–mitogen-activated protein kinase (MAPK) (2, 9), RAS-cyclic AMP (cAMP)-protein kinase A (PKA) (5, 37), and Snf1–AMP-dependent protein kinase (AMPK) (24) signaling pathways for activation of genes required to drive differentiation to a filamentous morphology, known as filamentous response genes. Global genomic localization indicates a complex network of interactions among these factors that include multiple autofeedback and cross-regulatory circuits. Phd1 and Mga1 in particular appear to be master regulators of filamentation as their promoters are bound by all six factors, including themselves (3), and overexpression on their own can induce differentiation (16, 27).

Despite years of intense investigation, few mechanistic details of how these pathways regulate downstream transcription factors have been described. Ste12 and Tec1 bind cooperatively to the promoters of some filamentous response genes, and this combination of factors are regulated by components of the pheromone response pathway through the MAPK Kss1 (28). Ste12 and Tec1 complexes are inhibited by Dig1 and Dig2, and presumably these regulators are antagonized by Kss1 (2, 8, 9). Specificity of pheromone signaling was shown to involve phosphorylation of Tec1 by Fus3, thereby causing degradation of this factor in pheromone-treated cells to prevent inappropriate expression of filamentous response genes (1, 7). Genetic evidence suggests that Sok2 (44) and Flo8 (34) are regulated by cAMP-PKA signaling, and consistent with this view, Flo8 DNA binding activity is stimulated by PKA-dependent phosphorylation in vitro (34). Similarly the PKA pathway was shown to regulate multiple transcription factors involved in filamentous growth, including Sok2, through the function of an additional protein kinase, Yak1 (29). Less is known regarding the role of the AMPK-Snf1 pathway in controlling differentiation, although importantly, Snf1 becomes hyperphosphorylated in response to nitrogen limitation, and snf1 diploids do not differentiate (10, 24, 33).

An additional protein kinase that is regulated in response to nitrogen limitation is Cdk8, a component of the RNA polymerase-associated mediator complex. Cdk8 activity is rapidly lost in cells switched to limiting nitrogen, an effect that is produced by degradation of the kinase itself (32). Additionally, Cdk8 expression is dramatically reduced in logarithmically growing cultures as cells approach the diauxic shift, consistent with a postulated role for this enzyme as a negative regulator of stress-responsive genes (21). Cdk8 negatively regulates both pseudohyphal differentiation and haploid invasive growth; cdk8 null diploids produce dramatically enhanced filaments on low-nitrogen agar, and otherwise nonfilamenting stains derived from W303 or S288C grow invasively as cdk8 haploids (32). Part of this effect involves phosphorylation of the transcription factor Ste12 to promote its degradation (32). Ste12 is destabilized by Cdk8 under ideal growth conditions, but it can accumulate for activation of filamentous response genes during nitrogen limitation because of loss of Cdk8. Mutation of the Cdk8-dependent phosphorylation sites on Ste12 causes enhanced pseudohyphal response, although this response is not as robust as that caused by a cdk8 null mutation, and consequently, we reasoned it likely that additional factors involved in pseudohyphal differentiation must also be regulated by Cdk8.

We show here that stability of Phd1, defined as a master regulator of filamentous growth, is also regulated by Cdk8-dependent phosphorylation. Furthermore, the Σ1278b strain, capable of differentiating to pseudohyphae, was found to have a natural sequence polymorphism that eliminates a site of Cdk8-dependent phosphorylation required for degradation of Phd1, relative to the S288C and W303 backgrounds, which have lost differentiation capability. These results support the view that Cdk8 is a primary target for signaling mechanisms that induce differentiation and that loss of Cdk8 expression causes upregulation of PHD1 expression through stabilization of the activator Ste12, and stabilization of Phd1 protein itself, which can autoactivate its expression. This work illustrates how a protein kinase, responsive to global signaling pathways, can drive a differentiation program by regulating the stability of multiple “hub” transcription factors within complex regulatory networks.

MATERIALS AND METHODS

RESULTS

Phd1 is destabilized by Cdk8.

Cdk8 is known to regulate yeast filamentous growth in part by phosphorylating the transcription factor Ste12. Ste12 is destabilized in cells growing in rich medium by phosphorylation, but a switch to nitrogen limitation eliminates Cdk8 activity and stabilizes Ste12 for activation of filamentous response genes (32). Because mutation of the Cdk8-dependent phosphorylation sites on Ste12 produces a less dramatic enhancement of pseudohyphal growth than does a cdk8 null allele (32), we suspected that additional transcription factors required for differentiation might also be Cdk8 substrates. To examine this, we compared expression of epitope-tagged forms of Phd1 and Mga1, which were defined as master regulators that can enhance pseudohyphal growth when overproduced (3), and Tec1, which is known to activate filamentous response genes cooperatively with Ste12 (1, 7) in wild-type (WT) and cdk8 null strains. We observed that steady-state levels of both Phd1 and Tec1 were increased by ∼20-fold and 7-fold, respectively, in the cdk8 null strain relative to the wild type (Fig. 1A, lanes 4 to 7). In contrast, steady-state levels of Mga1 were roughly equivalent in both strains (Fig. 1A, lanes 2 to 3). Because the epitope-tagged proteins were produced from the TEF1 promoter, these results suggest that the Phd1 and Tec1 proteins, but not Mga1, may be destabilized in cells expressing Cdk8. We found that Cdk8 dramatically increases the turnover rate of Phd1 protein: in the S288C and W303 wild-type strain backgrounds, Phd1 has a half-life of less than 10 min (Fig. 1B, lanes 5 to 7), whereas this is extended beyond 45 min in cdk8 null strains (Fig. 1B, lanes 2 to 4) (see below). This indicates a significantly greater effect of Cdk8 on Phd1 stability than was previously observed for Ste12 (32). These observations implicate Phd1 as a major substrate for regulation of filamentous growth by Cdk8. We have not examined the role of Cdk8 in Tec1 stability further, and it is possible that Cdk8 directly regulates Tec1 stability by phosphorylation. Alternatively, because Tec1 is stabilized by interaction with Ste12 (19), it may become destabilized in cdk8 strains as an indirect consequence of decreased Ste12 abundance.

Open in a separate window

Fig 1

Phd1 is destabilized by Cdk8. (A) Flag-epitope-tagged Mga1 (lanes 2 and 3), Phd1 (lanes 4 and 5), and Tec1 (lanes 6 and 7) were expressed in wild-type (lanes 1, 2, 4, and 6) or cdk8 (lanes 3, 5, and 7) W303 strains, and steady-state levels of protein were detected by immunoblotting with anti-Flag (α-flag) (upper) or antitubulin (α-tubulin) antibodies (lower panel). Relative levels of the Flag fusion proteins were quantified and are indicated as normalized to tubulin below the lanes ([Fusion]). Lane 1 contains an extract from the strain bearing an empty vector. (B) Flag-tagged Phd1 (cloned from S288C) was expressed in wild-type (lanes 5 to 7) or cdk8 (lanes 2 to 4) W303 yeast grown to mid-log phase. Cycloheximide was added to the cultures, and protein extracts were prepared for immunoblotting at the times indicated above (minutes).

Phd1 from the Σ1278b strain background has a natural polymorphism that eliminates a potential Cdk8 phosphorylation site and stabilizes the protein.

Specific phosphorylations for Cdk8 in vivo have been described on Ste12 (32), Gal4 (20), Med2 (18), and Gcn4 (6). All of the Cdk8 sites identified to date are serines or threonines, flanked by a proline 1 or 2 residues toward the C terminus, and several Cdk8-dependent sites also have a proline 2 to 4 residues toward the N terminus. In examining the Phd1 sequence, we observed two potential candidate sites with similar properties at serines 92 (PQSP) and 305 (PKSSP) and additional possible sites with C-terminal flanking prolines at T23, S66, and S165. To localize phosphorylated residues on the Phd1 protein, we examined phosphorylation of a series of N-terminal truncation derivatives (Fig. 2A) and determined that a cluster of phosphorylations in vivo must exist within a serine-rich region between residues 73 and 124, immediately N terminal to the MluI-box DNA binding domain. Specifically, a deletion mutant, Phd1-D1, bearing residues 73 to 366 produces multiple species in SDS-PAGE and is expressed at slightly lower abundance in wild-type (WT) cells relative to a cdk8 null strain (Fig. 2B, lanes 2 and 6), but it is not as significantly affected by CDK8 as full-length Phd1 (Fig. 2B, lanes 1 and 5) (Fig. 1). This indicates that truncation of the N terminus must eliminate residues necessary for efficient CDK8-dependent degradation. Mutants with larger N-terminal truncations were expressed at much higher levels than full-length Phd1, were not affected by CDK8, and did not produce obvious multiple species in SDS-PAGE (lanes 3 and 4 and 7 and 8). Additionally, we found that full-length Phd1 and the Phd1-D1 deletion mutant were labeled in vivo with ³²P_i, but not the smaller deletion mutants (Fig. 2C) (data not shown). Within the serine-rich region between residues 73 and 124, S92 in particular seemed a strong candidate site for phosphorylation by Cdk8, as the flanking sequences share similarity to both of the Cdk8-dependent phosphorylation sites on Ste12 at S261 and S451, and S92 resembles a hybrid of these, flanked on both sides by conserved proline and glutamine residues (Fig. 2D).

Open in a separate window

Fig 2

Localization of Phd1 phosphorylations. (A) Schematic representation of the Phd1 protein indicating the MluI-box DNA binding domain and serine 92 within a serine/threonine-rich region (SSSS). D1, D2, and D3 indicate the span of residues contained within deletion mutants used to localize N-terminal phosphorylations. (B) Full-length Phd1 (lanes 1 and 5) or the D1 (lanes 2 and 6), D2 (lanes 3 and 7), and D3 (lanes 4 and 8) N-terminal deletion mutants, fused to a C-terminal 3×Flag epitope, were expressed in wild-type (lanes 5 to 8) or cdk8 null (lanes 1 to 4) W303 haploid yeast strains. Steady-state levels of Flag-tagged protein were detected by immunoblotting. (C) W303 strains expressing full-length Phd1 (lane 2) or the D1 (lane 3), or D2 (lane 4) deletion-3×Flag fusions or bearing a vector control (lane 1) were labeled with ³²P_i. Phd-1–Flag fusion proteins were recovered by immunoprecipitation, resolved on SDS-PAGE, and visualized by autoradiography. (D) Comparison of the sequence surrounding Phd1 S92 (top line) with the S261 (middle) and S451 (bottom) sites on Ste12 phosphorylated by Cdk8. Conserved proline (purple) and glutamine (green) residues flanking the serine (pink) are indicated.

In the course of constructing epitope-tagged fusions, we analyzed clones of the PHD1 open reading frame from various strain backgrounds, including the naturally filamenting Σ1278b strain, and found four sequence polymorphisms relative to S288C, three of which are silent and are also observed in several other laboratory strains (Table 3). Interestingly, the fourth polymorphism is unique to the Σ1278b background and, to our surprise, happens to produce a missense mutation of serine 92 to phenylalanine (S92F). Because this residue seemed a good candidate for phosphorylation by Cdk8, we compared stabilities of Phd1 produced by the S288C (S92) and Σ1278b (S92F) alleles in various strain backgrounds. Similar to the results shown above, we found that S92 Phd1 was quite unstable in both haploid and diploid Σ1278b-derived strains, with a half-life of 10 to 15 min (Fig. 3A, lanes 1 to 4, and B) (data not shown). In contrast, Phd1 S92F protein was found to have a much longer half-life of over 40 min in wild-type strains (Fig. 3A, lanes 5 to 8, and B). Consistent with the results shown above, disruption of cdk8 caused significant stabilization of both S92 and S92F Phd1 in the sigma background, to produce an estimated half-life of ∼80 min (Fig. 3A, lanes 9 to 12 and 13 to 16, respectively) (data not shown). In combination with the results shown above, these observations demonstrate that Phd1 stability is sensitive to Cdk8 activity, and that serine 92 is particularly important for Cdk8-dependent degradation.

Open in a separate window

Fig 3

Serine 92 on Phd1 is necessary for Cdk8-dependent destabilization. (A) 3×Flag epitope-tagged S288C-derived S92 Phd1 (lanes 1 to 4 and 9 to 12) or Σ1278b-derived S92F Phd1 (lanes 5 to 8 and 13 to 16) were expressed in wild-type (lanes 1 to 8) or cdk8 (lanes 9 to 16) Σ1278b diploid strains and grown to mid-log phase. Cycloheximide was added to the cultures, and protein extracts were prepared for immunoblotting at the times indicated above (minutes). (B) Flag-tagged Phd1 proteins were quantified, and the relative abundance, normalized to tubulin, was plotted as a function of time post-cycloheximide addition.

Cdk8 phosphorylates Phd1 on serine 92.

To determine whether Cdk8 is capable of phosphorylating Phd1, we performed in vitro kinase reactions with hemagglutinin (HA)-tagged Cdk8, recovered from yeast by immunoprecipitation, and recombinant Phd1 protein as substrates (Fig. 4B). In these experiments, we found that Phd1 was phosphorylated as efficiently as Gal4, a previously characterized substrate for Cdk8 (20) (Fig. 4A, lanes 2 and 4). S92F Phd1 was also phosphorylated by Cdk8-HA (lane 3), although reproducibly slightly less efficiently than S92 Phd1 (Fig. 4D). Cdk8 immunoprecipitates also contain small amounts of Cdk7 (20), and thus importantly, we also found that Phd1 could be phosphorylated in vitro by purified Cdk8/cyclin C complexes (Fig. 4B, lane 3), demonstrating that Phd1 is a direct substrate for phosphorylation by Cdk8 in vitro. Phd1 serine 92 resides on a very large potential tryptic peptide of 81 residues (73 to 154), bearing a total of 10 S/T residues, and consequently we were not able to detect phosphopeptides derived by phosphorylation at this specific residue using mass spectrometry (MS) (data not shown). Nevertheless, we analyzed phosphopeptides from the in vitro kinase reactions by 2-dimensional fingerprinting and found that S92 Phd1 phosphorylated by Cdk8 in vitro produces a very large phosphorylated tryptic peptide that generates multiple spots representing differentially phosphorylated species (Fig. 5A, arrows). Double digestion with trypsin and chymotrypsin produces additional smaller peptides, one of which is observed with S92 substrate protein but not S92F Phd1 (arrow; compare Fig. 5B and C). Furthermore, we found that Cdk8-HA was able to phosphorylate a synthetic peptide bearing S92 in vitro, but not an equivalent peptide representing the S92F polymorphism (Fig. 4E). These results demonstrate that Cdk8 directly phosphorylates Phd1 on multiple residues between residues 73 and 154, one of which is serine 92.

Open in a separate window

Fig 4

Cdk8 phosphorylates serine 92 on Phd1 in vitro. (A) In vitro kinase reactions were performed with HA-tagged Cdk8 recovered from yeast by immunoprecipitation. Reaction mixtures contained equivalent amounts of recombinant S92 Phd1 (lane 2), S92F Phd1 (lane 3), Gal4 (lane 4), GST-CTD (lane 5), GST (lane 6), or no substrate (lane 1). Samples were resolved by SDS-PAGE and analyzed by autoradiography. RNA polymerase II coimmunoprecipitates with Cdk8 at low stoichiometry, and the largest subunit becomes phosphorylated on the CTD in these reactions (indicated by *, left). (B) Recombinant S92 Phd1 (lane 1), S92F Phd1 (lane 2), Gal4 (lane 3), and GST-CTD (lane 4) substrate proteins, detected by Coomassie blue stain. In vitro kinase reaction mixtures contained 3 (lanes 1, 2, and 4) or 2 (lane 3) μg protein. (C) Recombinant wild-type Cdk8 (lane 3) or Cdk8 bearing a mutation at the ATP-binding site (D290A; lane 2) complexes copurified with Flag-tagged cyclin C, produced by expression in insect cells using baculovirus, were added to in vitro kinase reaction mixtures containing recombinant S92 Phd1. Lane 1 contained no added kinase. (D) Total phosphate incorporation into recombinant S92 and S92F Phd1 substrates was determined from in vitro kinase assays, the results of which are shown in panel A. Results are from five independent assays, and differences between the values are statistically significant with P = 0.0011. (E) In vitro kinase reactions were performed with HA-Cdk8 and peptide substrates representing S92 and S92F Phd1. Total phosphate incorporation was determined from three independent assays, and the values were corrected relative to reaction mixtures containing no peptide.

Open in a separate window

Fig 5

Cdk8 phosphorylates multiple sites on Phd1 in vitro, including serine 92. Phd1 substrates from in vitro kinase reactions (Fig. 4A) were digested with trypsin (A) or trypsin and chymotrypsin (B and C); phosphopeptides were resolved in 2 dimensions by electrophoresis (horizontal) and then chromatography (vertical) and visualized by autoradiography. The substrate proteins were S92 Phd1 (A and B) and S92F Phd1 (C). In panel A, arrows indicate phosphopeptides produced by multiple phosphorylations of a single large peptide spanning serine 92. In panel B, the arrow indicates a phosphopeptide unique to S92 relative to S92F in panel C.

The S92F PHD1 allele causes enhanced pseudohyphal growth.

We compared the effects that the S92 and S92F Phd1 forms have on pseudohyphal differentiation in the Σ1278b strain background. PHD1 is essential for formation of pseudohyphae in response to limiting nitrogen, as diploid strains with a phd1 null allele form smooth colonies that are devoid of filaments on synthetic low-ammonium dextrose (SLAD) medium (Fig. 6A, vector, top). Expression of either the S92 or S92F PHD1 alleles cause enhanced pseudohyphal growth, but the stabilized S92F form produces significantly enhanced filamentation relative to S92 Phd1 (Fig. 6A, compare S92 middle panels with S92F right panels). These results reinforce the notion that Phd1 is a master regulator of differentiation and that the Cdk8-dependent phosphorylation at S92 has a negative regulatory function for this factor. Importantly however, the enhanced pseudohyphal effect caused by S92F Phd1 is also apparent in a cdk8 null strain (Fig. 6A, lower right panel), which illustrates the possibility that Cdk8 must negatively regulate filamentous growth through phosphorylation of additional factors and further raises the possibility that the S92F substitution may enhance Phd1 activity independently of its effect on protein stability. We discuss the implications of this observation below.

Open in a separate window

Fig 6

S92F Phd1 produced by Σ1278b causes enhanced pseudohyphal growth. (A) Σ1278b diploid yeast strains with phd1 (top) or both phd1 and cdk8 (bottom) null alleles were transformed with a vector control (vector, left) or plasmids expressing S92 Phd1 (center) or S92F Phd1 (right) from the TEF1 promoter. The strains were plated on limiting nitrogen agar plates (SLAD), and colonies were photographed after 3 days. (B) S288C produces pseudohyphae when expressing Σ1278b FLO8 and PHD1. Combinations of the Σ1278b and S288C FLO8 and PHD1 alleles were expressed in an S288C diploid. Yeast strains were cotransformed with a plasmid expressing Σ1278b FLO8 (top panels) or a vector control (bottom panels), plasmids expressing the Σ1278b S92F Phd1 (right panels) or S288C S92 Phd1 (center panels) from the TEF1 promoter or a vector control (left). Strains were plated on SLAD, and colonies were photographed after 3 days.

Strains derived from the S288C background, used for the systematic gene deletion set, are incapable of forming pseudohyphae in response to limiting nitrogen. It was previously shown that the FLO8 gene of S288C has a premature stop codon (flo8-1), relative to the Σ1278b strain background, presumably causing the production of a truncated and inactive protein (26). Expression of the Σ1278b FLO8 allele in S288C conferred the ability to differentiate into pseudohyphae, demonstrating the essential role for Flo8 in this process. Considering that PHD1 also has a naturally occurring polymorphism between these strains at codon S92, we wondered whether this difference may also contribute to the robust pseudohyphal growth phenotype of Σ1278b relative to S288C. To investigate this, we expressed the S92 and S92F PHD1 alleles from the TEF1 promoter in combination with FLO8 from the Σ1278b background in an S288C diploid. As previously observed, S288C expressing its own alleles of PHD1 (S92) and FLO8 (flo8-1) forms smooth round colonies on limiting nitrogen agar (Fig. 6B, bottom left, vector), whereas expression of Σ1278b FLO8 in this background causes the appearance of occasional pseudohyphae (Fig. 6B, top left, vector). Interestingly, expression of S92 Phd1 from the TEF1 promoter in combination with flo8-1 (Fig. 6B, center bottom) has no effect, but coexpression with Σ1278b FLO8 causes filamentous growth (Fig. 6B, center top). Furthermore, coexpression of Σ1278b FLO8 along with S92F Phd1 causes significantly enhanced filamentous growth of S288C diploids (Fig. 6B, top right) that are as extensive, if not more so, than is observed with Σ1278b (compare Fig. 6A and B, top right panels). Importantly, neither PHD1 allele is capable of causing filamentation in combination with flo8-1 (Fig. 6B, lower panels), indicating that Σ1278b FLO8 and PHD1 contribute synergistically to differentiation, consistent with the finding that many target genes required for this process bind both factors (3). Furthermore, the finding that selection of S288C for its nonflocculating and smooth colony morphology resulted in a defective flo8 allele (26), while the dimorphic Σ1278b strain expresses a stabilized S92F Phd1, which disrupts a Cdk8 phosphorylation site, indicates the central importance of these factors and Cdk8 for regulation of differentiation.

S92 Phd1 protein is stabilized in nitrogen-starved cells.

Given the results shown above and our previous observations indicating that Cdk8 activity is lost in cells switched to a limiting source of nitrogen (32), we expected that Phd1 should become stabilized in cells starved for nitrogen. We examined this by comparing stability of Phd1 in Σ1278b-derived diploid cells growing in rich medium (SD) and in cultures switched to SLAD for 2 or 4 h. In these experiments, we observed a half-life of ∼15 min for Phd1 expressed in cells growing in SD (Fig. 7A, lanes 1 to 4, and B), but the half-life was increased to ∼40 min in cells switched to SLAD for 2 h (Fig. 7A, lanes 5 to 8, and B), and by 4 h post-SLAD, the half-life had increased to greater than 45 min (Fig. 7A, lanes 9 to 12). Similar results were observed in a Σ1278b-derived haploid strain and in W303 (not shown). These observations confirm that Phd1 protein becomes stabilized in response to nitrogen starvation, presumably because Cdk8 activity is lost under these conditions.

Open in a separate window

Fig 7

Phd1 becomes stabilized in nitrogen-starved cells. (A) A Σ1278b diploid strain transformed with a plasmid expressing S92 Phd1 from the TEF1 promoter was grown to mid-log phase in SD medium (Rich; lanes 1 to 4), or switched to SLAD for 2 h (lanes 5 to 8) or 4 h (lanes 9 to 12). Protein extracts were prepared from cultures treated with cycloheximide for the times indicated above (minutes) and immunoblotted with anti-Flag (top panels) or antitubulin (bottom panels) antibodies. (B) Flag-tagged Phd1 proteins were quantified, and the relative abundance, normalized to tubulin, was plotted as a function of time post-cycloheximide addition.

PHD1 expression is partially regulated by Ste12 in response to limiting nitrogen.

Filamentous growth is regulated by multiple signaling mechanisms, including the Kss1-depenent pheromone response pathway, which controls activity of Ste12 in association with Tec1 (28). Given the central role of Phd1 in this process, we wondered how Ste12 relates to expression of Phd1. Genome-wide chromatin immunoprecipitation (IP) experiments indicate that the PHD1 promoter may be bound by many factors known to be involved in filamentous growth, including Ste12, Tec1, Flo8, Sok2, and Phd1 itself (3). We analyzed the PHD1 promoter sequence for consensus elements and found multiple potential sites for both Ste12 (PRE) and Tec1 (TCS), as well as a potential site for Phd1 (Fig. 8A). Indeed, we can detect binding of recombinant Ste12 protein to each of these five potential sites by DNase I footprinting, with the strongest binding occurring over two closely spaced PREs in a tail-to-tail orientation at −485 from the translational start site (Fig. 8B). We also find that PHD1 mRNA expression is upregulated in wild-type cells switched from rich medium to nitrogen starvation conditions (SLAD); after 2 h in SLAD, expression was increased by 2-fold (Fig. 8C, lanes 3 and 6, and D) and by 6-fold after 4 h (Fig. 8D). This effect is only partially dependent upon Ste12; induction is reduced somewhat in cells bearing an ste12 null allele and is enhanced in a strain expressing a stabilized allele of STE12 bearing a mutation of one of the Cdk8-dependent phosphorylation sites at S261 (32) (Fig. 8C and D). These results demonstrate that PHD1 represents one target of Ste12 for induction of filamentous growth, and this effect is influenced by Cdk8.

Open in a separate window

Fig 8

Ste12 contributes to induction of PHD1 in response to nitrogen limitation. (A) Schematic representation of the PHD1 promoter. Location of consensus binding sites for Ste12 (PRE), Tec1 (TCS), and Phd1 are indicated. The numbering indicates the location relative to the translational start site. Recombinant Ste12 binds to each of the PREs in vitro as measured by footprinting. Electrophoretic mobility shift assay (EMSA) with Phd1 also produces a complex with the consensus site at −348 (data not shown). Flo8, Mga1, and Sok2 also bind the PHD1 promoter in vivo (3), but these factors either do not have defined consensus sequences, or we were unable to detect their cis-elements within the promoter sequence. (B) Recombinant Ste12 protein was used in footprinting reactions with PHD1 promoter template DNA. Reaction mixtures contained no protein (lane 1) or increasing amounts of recombinant Ste12 (lanes 2 to 7). Lane 8 contains an M&G G+A sequencing ladder. Shown is a region of the footprint spanning two PREs between −485 and −465 from the translational start site. (C) RNA was extracted from Σ1278b ste12 diploid yeast bearing plasmids expressing wild-type STE12 (lanes 3 and 6) or STE12 bearing a mutation of serine 261 to alanine (lanes 2 and 5), grown in rich medium (lanes 1 to 3) or switched to SLAD (lanes 4 to 6) for 2 h, and analyzed by Northern blotting with probes for PHD1 (top) or PDA1 (bottom). Levels of PHD1 expression proportional to cells growing in rich medium expressing WT Ste12 (lane 3) and normalized to PDA1 are indicated below the lanes. (D) Relative levels of PHD1 expression were quantified from Northern blots, as in panel A, from cells growing in rich medium (open bars) or from cells switched to SLAD for 2 h (hatched bars) or 4 h (solid bars).

Phd1 accumulates in nitrogen-starved cells.

Because Phd1 expressed from the TEF1 promoter is stabilized in cells switched to limiting nitrogen and its expression is also upregulated at the mRNA level, we expected that the chromosomally expressed protein should accumulate in cells growing in SLAD. To examine this, we tagged chromosomal copies of PHD1 with in-frame 3×HA epitopes in both the Σ1278b (S92F PHD1) and W303 (S92 PHD1) strain backgrounds. Consistent with our finding above that S92 Phd1 in W303 has a half-life of ∼10 min, we found it impossible to detect Phd1 protein in this strain background growing in rich medium (Fig. 9A, lane 1, bottom panels). However, we did observe Phd1–3×HA in W303 cells switched to SLAD for 4 h (lane 2, bottom), and the protein continued to accumulate an additional 14-fold over the following 24 h (lanes 3 to 5). In contrast, the stabilized S92F Phd1 protein expressed in Σ1278b is produced at significant levels in cells growing in rich medium but still accumulates ∼3-fold over 24 h of nitrogen starvation (Fig. 9A, top panels). In combination with the above findings, these results indicate that Phd1 protein accumulates in nitrogen-starved cells primarily by a mechanism involving its stabilization and to a lesser extent by transcription induction.

Open in a separate window

Fig 9

S92 Phd1 protein accumulates in nitrogen-starved cells. (A) PHD1 was tagged with an in-frame 3×HA epitope in the Σ1278b (top panels) or W303 (bottom panels) backgrounds. Proteins from strains growing in rich medium (lanes 1) or switched to nitrogen-limiting conditions (SLAD) for the indicated times in hours (lanes 2 to 5) were immunoblotted with anti-HA antibodies (Phd1-3HA) or anti-histone H3 (H3). Levels of Phd1-3HA expression, relative to zero time (Σ1278b) or 4 h poststarvation (W303) and normalized to histone H3, are indicated. (B) PHD1 overexpression can induce pseudohyphae in an ste12 strain. Wild-type (top panels) and ste12 (bottom) Σ1278b yeast strains were transformed with the 2μm plasmid (2μ) expressing PHD1 (S92) (right panels) or a vector control (left). Colonies were grown on SLAD for 3 days.

PHD1 was initially identified in a screen for high-copy effectors of enhanced pseudohyphal growth (16), and in accord with this result, we find that expression of Phd1 from a high-copy-number plasmid caused enhanced pseudohyphal growth in a wild-type Σ1278b diploid strain (Fig. 9B, top panels). Consistent with previous results (31), we find that PHD1 overexpression also induces pseudohyphae in an ste12 null strain (Fig. 9B, lower). This illustrates the central role of Phd1 in differentiation and indicates that Ste12 function and the Kss1-MAPK pathway are not required for differentiation, provided sufficient Phd1 can accumulate to activate downstream target genes in combination with Flo8.

DISCUSSION

Differentiation of eukaryotic cells toward specialized cell types is typically driven by a limited number of gene regulatory factors which define particular lineages. Even the simple unicellular eukaryote Saccharomyces cerevisiae exists as a variety of different cell types during its life cycle, whose identities are produced by the expression of several key regulators. Yeast pseudohyphal growth, in which branched chains of cells with enhanced cellular adhesion and substrate invasiveness grow out from the colony, is induced in response to limited nutrients (17). Transcription of ∼600 genes is altered during pseudohyphal differentiation, and at least a dozen transcription factors have been implicated in the process, including Ste12 (28, 32), Tec1 (28), Flo8 (34), Phd1 (16), Mga1 (27, 37), Sok2 (29, 44), Sfl1 (15, 37), Ash1 (4), Nrg1/2 (24), and Mss10/11 (45). Many of these factors regulate additional cellular responses; for example Ste12 is necessary for mating, while Nrg1/2 (43), Mss10/11 (35, 45), and Sok2 (29) regulate response to a variety of stress conditions. Among these factors, Ste12, Tec1, Flo8, and Phd1 are essential for either haploid invasive (Fig. 10) or diploid pseudohyphal growth, and only PHD1 and MGA1 overexpression can induce filamentous growth on their own (3). Genetic observations indicate that these factors induce filamentous growth in response to nutrient limitation through multiple signaling mechanisms, including the pheromone response Kss1-MAPK, cAMP-PKA, and Snf1-AMPK pathways. In particular, the Kss1-MAPK pathway regulates Ste12 and Tec1 (2, 9), while the PKA pathway is thought to regulate Sok2 and Flo8 (29, 34, 37). The Snf1-AMPK pathway was shown to antagonize the transcriptional repressors Nrg1 and 2 (24).

Open in a separate window

Fig 10

STE12 and PHD1 are essential for haploid invasive growth. Haploid (MATa) Σ1278b-derived yeast strains bearing the indicated gene disruptions were spotted onto YPD plates and allowed to grow for 3 days at 30°C (Total; left panel). Yeast cells growing above the surface of the agar were gently washed away under a stream of water to reveal the extent of invasive growth (right panel).

Cdk8, and its subunit cyclin C, along with the cofactors Srb8/Med12 and Srb9/Med13, form the kinase submodule of the RNA polymerase-associated mediator complex (25). This complex has both positive and negative roles for gene regulation, and the finding that CDK8 mutations are frequently observed in colorectal cancers (13) suggests that it must have important roles in growth control. An expanding number of gene-specific transcriptional regulators have been defined as Cdk8 substrates, which produce positive or negative effects on specific genes. For example phosphorylation of Gal4 by Cdk8 is required for full induction of the GAL genes (20, 38), and Cdk8 positively regulates p53 (11). Conversely, significantly more transcription factors are negatively regulated by Cdk8. Phosphorylation of Msn2 and -4 causes their export to the cytoplasm, while Gcn4 and the Notch intracellular domain (NCID) are targeted for degradation (6, 14).

This report adds to a growing list of factors whose degradation is induced upon phosphorylation by Cdk8. We find that Phd1 is an unstable protein with a half-life of ∼10 to 15 min but is stabilized in strains bearing a cdk8 null allele or when the Cdk8-dependent phosphorylation site is eliminated by the natural S92F polymorphism. Furthermore, Phd1 is stabilized in cells switched to limiting nitrogen, where Cdk8 activity is lost (32). Ste12 was previously shown to be degraded by Cdk8-dependent phosphorylation, but by comparison to Phd1, Ste12 is a much more stable protein. In rich medium, it has a half-life of ∼30 min and is stabilized to a half-life of an hour or more upon the switch to nitrogen-limiting conditions or when cdk8 is disrupted (32). The finding that Cdk8 controls the stability of (at least) two factors that play essential roles in differentiation underscores the central role of this enzyme in this process.

Our results indicate that Phd1 expression is regulated at both the transcriptional and posttranscriptional levels in response to nitrogen limitation. Ste12, Tec1, Mga1, Flo8, Sok2, and Phd1 itself bind the PHD1 promoter in vivo, as determined by global localization (3), and accordingly we observe consensus binding sites for Tec1 and Phd1 in addition to sites for Ste12 on this promoter (Fig. 8A). Ste12 is capable of binding these sites in vitro, and induction of PHD1 in response to nitrogen starvation is partially dependent on Ste12. We propose that PHD1 expression is limited in cells growing in rich medium because Cdk8 destabilizes Ste12 protein and, indirectly, Tec1, along with Phd1 itself. PHD1 also appears to be directly repressed by Sok2 (Fig. 11A). When cells encounter a nitrogen-limiting environment, Cdk8 protein is degraded, through signaling mechanisms that may involve Snf1 and/or PKA, to allow accumulation of its target factors. Stabilized Ste12 and Tec1, along with other factors, cause induction of PHD1. This induction together with the stabilization of Phd1 protein due to the absence of Cdk8 initiates a positive-feedback loop for Phd1 production (Fig. 11B) (3). Additionally, the repressive effect of Sok2 is relieved by a signaling mechanism involving the PKA isoform Tpk1 and Yak1 (29). Because downstream filamentous response genes (e.g., FRG in Fig. 11) are regulated by combinations of these factors (3), accumulation of Ste12, Tec1, Phd1, and other transcription factors can drive initiation of differentiation through induction of this large spectrum of coregulated genes (Fig. 11B). For example, FLO11, which encodes a cell surface-associated glycoprotein that contributes to pseudohyphal formation, flocculation, and invasive growth, is controlled in response to both the Kss1-MAPK (Ste12/Tec1) and the cAMP-PKA (Flo8, Sok2, and Phd1) pathways, and its promoter is bound by all six target hub factors (3, 39). Cdk8 likely regulates additional factors involved in filamentous growth, besides Ste12 and Phd1. This is illustrated by the observation that diploid cells bearing phd1 null alleles form smooth colonies on limiting nitrogen (Fig. 6A, top left), but deletion of cdk8 in a phd1 mutant strain causes infrequent filaments (lower left). We observe a similar result with filamentous growth of haploid cells where phd1 and ste12 haploids produce a noninvasive phenotype, but filamentous growth is at least partially restored in these backgrounds by disruption of cdk8 (Fig. 10).

Open in a separate window

Fig 11

Regulation of pseudohyphal differentiation by Cdk8. (A) PHD1 expression is regulated by multiple factors, including Ste12 (blue), Tec1 (yellow), Phd1 (green), and Sok2 (red). In rich medium, PHD1 expression is limited because Cdk8 inhibits the activity of Ste12, Phd1, and likely additional factors (purple) by direct phosphorylation and is repressed by Sok2. (B) Nitrogen limitation inhibits Cdk8 by signaling that may involve Snf1 and/or PKA, which allows accumulation of Ste12 (and Tec1) to cause upregulation of PHD1 expression. PKA signaling through Yak1 also inhibits the repressor Sok2. Phd1 accumulates in nitrogen-starved cells and autoactivates its own expression in a positive-feedback loop. Accumulation of combinations of these factors causes activation of downstream target filamentous response genes (FRG), in combination with Flo8, whose activity is positively regulated by PKA signaling.

The Σ1278b strain background provides an important simple model for a complex differentiation process involving multiple signaling pathways and transcriptional regulators. Recent sequence analysis of the Σ1278b genome and comparison to that of S288C reveals that these strains have single nucleotide polymorphisms in 54% of 6,800 open reading frames and have 44 and 13 genes, respectively, that are uniquely required for viability (12). In addition to differences in essential genes, these genetic backgrounds have two genetic alterations affecting key transcription factors that govern filamentous growth. The S288C background, and related strains, including W303, have a premature stop codon in the FLO8 gene resulting in the production of a truncated and presumably inactive product (26). Additionally, we show here that the Σ1278b background has a natural polymorphism that eliminates a Cdk8-dependent phosphorylation site (S92F) on Phd1. It is interesting that expression of S92F Phd1 from the TEF1 promoter enhances pseudohyphal growth in the Σ1278b background more than does S92 Phd1, even in a cdk8 null strain (Fig. 6A, right panels). This implies either that S92 may be phosphorylated by kinases other than Cdk8, albeit less efficiently, or that the S92F substitution activates Phd1 function independently of the effect on stability. If the former were true, Phd1 S92 should be less stable in cdk8 cells than S92F. We find that both have half-lives of ∼80 min in cdk8 null strains; in most experiments, S92 appears to be slightly less stable than S92F, but with a half-life of only 2 to 3 min shorter at most (Fig. 3B). It may be unlikely that this small difference could account for the larger hyperfilamentous effect of S92F, and therefore this polymorphism may affect another aspect of Phd1 function. Unfortunately, apart from the predicted MluI-box DNA binding domain (Fig. 2A), there is no information on mechanistic details of its function. We have not found differences in DNA binding capability of the S92F variant. Perhaps S92F affects the transcriptional activation function or interaction of Phd1 with other factors. It is likely that Phd1 interacts with additional DNA binding protein(s), as we find it binds very weakly to DNA on its own in vitro (not shown). It is also interesting to note that Σ1278b strains express significant levels of Phd1 S92F even in rich medium (Fig. 9A). This indicates that differentiation requires activation of additional factors in response to nitrogen starvation and loss of Cdk8 activity, but the stabilized Phd1 S92F may produce an environment that hypersensitizes cells for this process.

Cdk8 is ideally positioned to regulate multiple transcription factors in response to global signals that require a response as drastic as pseudohyphal differentiation. Transcriptional activators become exposed to Cdk8 as potential substrates during recruitment of the mediator, and in the case of Ste12 and Phd1, this interaction with Cdk8 targets them for degradation. These observations illustrate how a developmental program can be driven by a single central protein kinase whose activity is regulated by global signaling and controls multiple hub transcriptional regulators to drive a differentiation process.

ACKNOWLEDGMENTS

Footnotes

REFERENCES