Pheromone-regulated Sumoylation of Transcription Factors That Mediate the Invasive to Mating Developmental Switch in Yeast*

A fundamental question in biology is how different signaling pathways use common signaling proteins to attain different developmental outcomes. The yeast transcription factor Ste12 is required in at least two distinct signaling processes, each regulated by many of the same protein kinases. Whereas Ste12-Ste12 homodimers promote transcription of genes required for mating, Ste12-Tec1 heterodimers activate genes required for invasive growth. We report that Ste12 and Tec1 undergo covalent modification by the ubiquitin-related modifier SUMO. Stimulation by mating pheromone promotes sumoylation of Ste12 and diminishes the sumoylation of Tec1. In the absence of sumoylation Tec1 is more rapidly degraded. We propose that pheromone-regulated sumoylation of Ste12 and Tec1 promotes a developmental switch from the invasive to the mating differentiation program.

A fundamental question in biology is how different signaling pathways use common signaling proteins to attain different developmental outcomes. The yeast transcription factor Ste12 is required in at least two distinct signaling processes, each regulated by many of the same protein kinases. Whereas Ste12-Ste12 homodimers promote transcription of genes required for mating, Ste12-Tec1 heterodimers activate genes required for invasive growth. We report that Ste12 and Tec1 undergo covalent modification by the ubiquitin-related modifier SUMO. Stimulation by mating pheromone promotes sumoylation of Ste12 and diminishes the sumoylation of Tec1. In the absence of sumoylation Tec1 is more rapidly degraded. We propose that pheromone-regulated sumoylation of Ste12 and Tec1 promotes a developmental switch from the invasive to the mating differentiation program.
The budding yeast Saccharomyces cerevisiae can initiate distinct developmental programs depending on the presence or absence of specific external cues. Mating is initiated when a and ␣ haploid cell types secrete and respond to cell type-specific pheromones acting through G protein-coupled receptors; once activated, the a and ␣ cells fuse to form an a/␣ diploid cell. Invasive or filamentous growth occurs in nutrientpoor conditions and is manifested by altered budding and formation of long branching filaments, as well as increased adherence and invasion of the substratum. Both developmental outcomes require activation of a protein kinase cascade comprised of Ste20, Ste11, Ste7, and Fus3 or Kss1 (1,2).
Although the mechanisms of MAP kinase and transcription factor activation are well established, less understood is how signaling pathways that share the same components attain different developmental fates (21). Signal identity has been ascribed to differences in signal magnitude, duration, and frequency (22), as well as to the scaffolded association of protein kinase components (23). Even where such differences have been documented, the kinase signals must still be interpreted by nuclear transcription factors to initiate expression of a distinct set of genes (9). Here we have shown that Ste12 and Tec1 are covalently modified by the small ubiquitin-related protein SUMO (Smt3 in yeast). Although stimulation by mating pheromone promotes sumoylation of Ste12, the same treatment inhibits sumoylation of Tec1 and the protein is instead degraded. These findings suggest that pheromone-regulated sumoylation of transcription factors underlies the developmental switch from the invasive to the mating differentiation program.

EXPERIMENTAL PROCEDURES
Strains and Plasmids-Standard methods for the growth, maintenance, and transformation of bacteria and for the manipulation of DNA were used throughout. Details of plasmid construction are available from the authors. Except where stated otherwise, TEC1 and STE12 were expressed from single copy plasmids using the native promoter.
Growth,Transcription,andDegradationBioassays-Thepheromonedependent growth inhibition (halo) and reporter-transcription assays were conducted as described previously (25). Unless indicated otherwise, the concentration of ␣-factor was 3 M, which is 10-fold above the EC 50 for transcriptional induction. To monitor the loss of Tec1 over time, mid-log cell cultures were treated with ␣-factor for 60 min, followed by cycloheximide addition as described previously (26). Immunoprecipitations, 6.5% gel electrophoresis, and immunoblot analysis were carried out as described previously (26) using antibodies against SUMO (from C. Hoege and S. Jentsch), Protein A (Sigma), FLAG (Sigma), and ubiquitin (Sigma).

RESULTS
A long-standing question in cell regulation is how signaling pathways that share component proteins can attain different developmental fates. To address this question we investigated whether Ste12 or Tec1 is covalently modified by the small ubiquitin-related protein SUMO (Smt3 in yeast) and whether pheromone-regulated modification of these signaling proteins could underlie the switch from the invasive growth to mating differentiation transcription program. We focused initially on sumoylation because it is a reversible modification known to regulate the activity of nuclear transcription factors, co-activators, and co-repressors (27,28). In addition, mass spectrometry sequencing identified Ste12 as one of several hundred proteins that co-purify with Smt3, making it a likely substrate for sumoylation (29).
Ste12, fused to a TAP tag (24), was immunopurified, resolved by gel electrophoresis, and detected by immunoblotting. The antibodies strongly recognized the native form of the protein (Fig. 1, A and B), as well as additional species ϳ12 and 24 kDa larger than the native protein, a difference corresponding to the molecular mass of one or two copies of mature SUMO, respectively. The presence of multiple high molecular mass species suggested that Ste12 undergoes sumoylation at two or more sites. The larger forms of the protein were more abundant in cells treated with ␣-factor pheromone, were less abundant in a mutant strain lacking the E3 SUMO ligase SIZ1 (Fig. 1A), and were completely absent in a ubc9-1 mutant strain deficient in SUMO-conjugating enzyme activity ( Fig. 1B) (30). Ste12 also migrated more slowly in the ubc9 -1 mutant strain; however, this mobility shift did not appear to be pheromone regulated. Moreover, the shift is too small to arise from addition of ubiquitin or a ubiquitin-like modifier and may instead result from another known modification such as phosphorylation (3)(4)(5)(6)(31)(32)(33).
To confirm that Ste12 is sumoylated we immunopurified Ste12 tagged with the FLAG epitope. The higher molecular mass form of Ste12 was recognized by SUMO antibodies and was again enriched by FIGURE 1. Pheromone promotes sumoylation and activation of Ste12. A, whole cell lysates from wild-type BY4741 and isogenic siz1⌬ or siz2⌬ mutant cells containing integrated TAP-tagged Ste12 were immunoprecipitated with IgG-Sepharose and detected by immunoblotting using anti-SUMO antibodies as indicated and as detailed under "Experimental Procedures." Note that TAP is recognized by any IgG and is therefore recognized by anti-SUMO antibodies. Molecular mass standards (kDa) are provided to the right of the panel. B, immunoprecipitates were prepared as described in panel A except that a ubc9-1 mutant strain and the isogenic wild-type were used. The predominant band in each case is the unmodified protein, whereas the two slower migrating bands are sumoylated Ste12. C, whole cell lysates from wild-type (WT) and isogenic fus3⌬ or kss1⌬ mutant cells containing FLAG epitope-tagged Ste12 or the parent vector were immunoprecipitated with M2 anti-FLAG resin, and the modified form of Ste12 was detected using anti-SUMO antibodies (top panel). Unmodified Ste12 was detected by immunoblotting of whole cell lysates using anti-FLAG antibodies (bottom panel) to show relative expression. Numbers under each lane refer to the difference in band intensity relative to lane 1, as determined by scanning densitometry. D, transcription activity was measured in ste12⌬ cells transformed with a plasmid containing an invasive-specific PRE reporter (FUS1 promoter, lacZ reporter) and a plasmid expressing FLAG-tagged Ste12 or FLAG-Ste12 fused to SUMO (Smt3) and treated with the indicated concentrations of ␣-factor. E, the same cells as in panel D were assayed for growth arrest response following exposure to 45 g of ␣-factor for 48 h. The same cells were spotted onto solid YPD medium and after 2 days rubbed vigorously under a stream of water to detect invasive growth. F, the same cells as in panel D were analyzed by immunoblotting of whole cell lysates using anti-FLAG antibodies. Note that Ste12 tagged with the FLAG epitope (ϳ1 kDa) migrates faster than Ste12 tagged with TAP (ϳ20 kDa). Note also that the strains used in panel B are deficient in KSS1 (data not shown) and for this reason are used only in this experiment to demonstrate that sumoylation requires UBC9.
pheromone stimulation (Fig. 1C). The increase in sumoylation was proportionately greater than the induction of Ste12 expression that is typically observed in pheromone-treated cells. Moreover, deletion of either FUS3 or KSS1 dampened the effect of pheromone on Ste12 sumoylation, consistent with the partially overlapping function of the two kinases in this pathway (1,16,34). Kss1 and Fus3 phosphorylate many of the same proteins but with distinct substrate specificities, and this presumably accounts for the differences in Ste12 sumoylation in the two mutants (35). We conclude from these data that Ste12 is sumoylated and that sumoylation is enhanced in response to pheromone stimulation.
Ste12 binds to PREs present within the promoters of several genes involved in mating. The promoter of the FUS1 gene contains multiple PREs and has been widely used to monitor pheromone-stimulated transcription activity (25). Thus, a reporter consisting of the PRE promoter fused to lacZ (␤-galactosidase) was used to determine how sumoylation affects Ste12 function. Initially we attempted to block sumoylation in Ste12 by substituting Arg for Lys-174 and Lys-409, the two residues that most closely match the preferred sequence for sumoylation (⌿-K-X-D/E, where ⌿ is a hydrophobic residue and X is any residue) (36). However, neither mutation, tested alone or in combination, blocked the pheromone-stimulated mobility shift, suggesting that one or more of the other 42 Lys residues in Ste12 is modified (data not shown). We also investigated how the pheromone response is altered in the ubc9 mutant strain. In this mutant PRE-lacZ induction was substantially reduced relative to the wild-type, suggesting that sumoylation is required for full Ste12 activity (data not shown). However, even though the PRE reporter is highly specific for Ste12 activity, the ubc9 mutant could have multiple effects that indirectly affect reporter transcription activity. As an alternative approach we examined PRE-lacZ activity in cells expressing a SUMO-Ste12 fusion protein. Similar SUMO substrate fusions have been used previously to show that sumoylation can diminish transcrip-tion factor function (37)(38)(39). In the absence of pheromone, cells expressing SUMO-Ste12 exhibited no change in basal lacZ activity, suggesting that sumoylation of Ste12 is not sufficient to initiate new gene transcription. In the presence of pheromone, however, cells expressing SUMO-Ste12 exhibited a Ͼ2-fold increase in maximum transcription activity (Fig. 1D). The SUMO-Ste12 fusion also increased the pheromone-mediated growth arrest response, as indicated by a larger zone of growth inhibition surrounding a source of ␣-factor (the response was also more transient, as indicated by resumption of colony growth and the formation of a turbid halo) (Fig. 1E). Finally, activation of the mating response by Ste12-SUMO resulted in a concomitant decrease in the invasive growth response (Fig. 1E). Expression levels of the SUMO-Ste12 fusion were not elevated compared with Ste12 (Fig. 1F). However, whereas native Ste12 is only partially sumoylated after pheromone stimulation, the SUMO-Ste12 fusion resembles a protein that is fully sumoylated all the time, and this presumably accounts for the higher transcription and growth arrest response compared with the native partially sumoylated protein. Thus, sumoylation of Ste12 appears to confer a gain-of-function phenotype, but transcription activity is still contingent on stimulation by upstream components in the pathway.
We then considered whether Tec1 is also regulated by sumoylation. As before, a FLAG-Tec1 fusion was immunopurified and probed by immunoblotting with antibodies to FLAG as well as SUMO. Once again we could detect multiple bands with the FLAG antibody, one corresponding to native Tec1 and a second corresponding to Tec1 potentially modified by sumoylation. The higher molecular mass species was also detected using anti-SUMO-antibodies, and both bands were absent in cells containing only the parent vector ( Fig. 2A). Pheromone treatment resulted in diminished expression and substantially diminished sumoylation of Tec1 (Fig. 2B). The pheromone-dependent decrease in Tec1 sumoylation is in marked contrast to the pheromone-dependent increase observed for Ste12. Because increased sumoylation promotes Ste12 activity we anticipated that decreased sumoylation might result in diminished Tec1 activity. To test this model we replaced a consensus site Lys at position 54 and found that it successfully eliminated detectable sumoylation concomitant with higher overall expression levels (Fig. 2B). We then measured invasive growth and transcription induction using a FRE promoter fused to lacZ. Rather than decrease activity, the sumoylationdeficient Tec1 K54R mutant exhibited slightly elevated FRE-lacZ and invasive growth activities (Fig. 2, C and D) (8,40,41). Together these data reveal that sumoylation is not required for full Tec1 transcription activity or proper regulation of invasive growth behavior.
The higher abundance of Tec1 K54R could result from a loss of ubiquitination. Ubiquitin attachment typically leads to rapid capture and degradation of the substrate by the proteasome protease complex (27), and it was recently documented that Tec1 undergoes phosphorylationdependent ubiquitination and accelerated degradation in pheromonestimulated cells (42)(43)(44). Moreover, modification by SUMO and ubiquitin are often antagonistic and can sometimes occur on the same Lys residue (45,46). In agreement with this model, the unsumoylated Tec1 K54R mutant was ubiquitinated at a reduced level relative to the wild-type protein (data not shown); the residual ubiquitination was most likely due to modification of an alternate site when the primary site (Lys-54 in this case) becomes unavailable. Moreover, Tec1 K54R accumulates a prominent species with reduced mobility but that is not recognized by the anti-SUMO antibodies (Fig. 2B); we presume that this band is due to an accumulation of phosphorylated but poorly ubiquitinated protein.
Tec1 ubiquitination is regulated by the MAP kinases Fus3 and Kss1 (42)(43)(44). These MAP kinases phosphorylate many common substrates and have partially overlapping function in vivo. Expression of either kinase is sufficient to partially sustain the mating response (Fig. 1C) (1,16,34). However, significant differences have been reported for the invasive pathway. Deletion of KSS1 is sufficient to block invasive growth, whereas deletion of FUS3 promotes invasive behavior. Moreover, deletion of FUS3 (but not KSS1) results in an increase in FRE binding activity (9) and transcription of FRE-containing genes (8,10,16). To determine whether sumoylation might underlie these differences we compared Tec1 expression in fus3⌬ and kss1⌬ mutant strains. Deletion of FUS3 (but not KSS1) resulted in elevated expression of Tec1 and also of the sumoylated form of Tec1 (Fig. 3A). Moreover, pheromone treatment of the fus3⌬ mutant strain no longer diminished, and even increased slightly, the expression of Tec1. These differences in expression could be ascribed to differences in protein stability because Tec1 was degraded more slowly in the absence of FUS3 expression (Fig.  3B) (42)(43)(44). Thus, the negative regulation of Tec1 abundance by Fus3 (but not Kss1) is fully concordant with the previously reported role of Fus3 (but not Kss1) in FRE transcription and invasive growth behavior. Stated differently, there is a strong correlation between Tec1 sumoylation, stability, expression, and transcription activity.
Finally, the relationship between Tec1 sumoylation and abundance sug-

Pheromone-regulated Sumoylation in Yeast
gests that sumoylation might stabilize the protein. A prediction of this model is that sumoylated Tec1 should degrade more slowly than the unmodified protein.
To test this, we compared the abundance of sumoylated and non-sumoylated Tec1 over time, following cycloheximide treatment to block new protein synthesis. As predicted, the sumoylated form of the protein persisted much longer than the unmodified species, whether expressed under the control of the native promoter (Fig. 3C) or a heterologous inducible promoter (from GAL1, Fig. 3D).

DISCUSSION
Here we have demonstrated that the transcription factors Ste12 and Tec1 are sumoylated. These are the first components of the MAP kinase signaling cascade shown to undergo sumoylation in yeast. More significantly, sumoylation of both proteins is regulated by pheromone. Whereas pheromone promotes sumoylation of Ste12, the same treatment results in diminished sumoylation of Tec1.
With respect to Tec1, immunopurification experiments revealed a marked reduction in the proportion of sumoylated protein following treatment with ␣-factor pheromone (Fig. 2B). Immunoblotting of whole cell extracts also revealed a marked decrease in Tec1 abundance (Fig. 3A) and accelerated Tec1 degradation (Fig. 3B) after pheromone treatment. In contrast, cells lacking Fus3 exhibited no pheromone-dependent change in Tec1 expression, sumoylation (Fig. 3A), or stability (Fig. 3B). These data support a model in which Tec1 is de-sumoylated and consequently degraded in response to MAP kinase signaling.
Another function of sumoylation evidently is to regulate Tec1 ubiquitination. Sumoylated Tec1 is more stable than the non-sumoylated protein, and mutating the lysine residue used for sumoylation leads to an increase in Tec1 abundance. The simplest explanation is that the same lysine residue is used for both sumoylation and ubiquitination and that sumoylation enhances protein stability by competing with ubiquitin for a common target site. Tec1 K54R is still partially ubiquitinated, however, suggesting that other regulatory mechanisms could exist (data not shown). For instance, sumoylation might prevent recognition of the protein by the proteolytic machinery and thus protect the protein from degradation.
We also attempted to block Ste12 sumoylation by mutating the two best candidate sites, alone and in combination. Unfortunately, neither of those mutations blocked Ste12 sumoylation. Besides these two residues, there are another 42 lysines in Ste12 that could potentially serve as sites of sumoylation. Furthermore, it is possible that mutating a preferred site leads to sumoylation at another cryptic site. For these reasons we took an alternative approach and examined the behavior of a SUMO-Ste12 fusion. The gain-of-function phenotype exhibited by the fusion protein, together with the stimulus-dependent nature of Ste12 sumoylation, supports the suggestion that sumoylation promotes Ste12 activity. However, a rigorous test of this aspect of the model will require that the sumoylation sites in Ste12 be definitively mapped and mutated.
Tec1 is one of a small but growing list of signaling proteins that undergo ubiquitination (42,43), a list that in yeast includes pheromone receptors (48,49), the G protein ␣ subunit Gpa1 (50,51), the RGS protein Sst2 (26), and the effector kinase Ste7 (52,53). Ubiquitination of Ste2, Ste7, and Sst2 are stimulated by pheromone and may therefore contribute to feedback regulation of signaling. In contrast, pheromoneregulated sumoylation of Ste12 and Tec1 is less likely to regulate signal intensity but rather signal specificity by favoring activation of the mating pathway and inactivation of the invasive growth pathway. Given the prevalence of transcription factors that are both sumoylated and ubiq-uitinated, and of signaling pathways that use shared MAP kinase components, the mechanisms established here are likely to be recapitulated in other signaling systems.