Research ArticleCell Biology

Single-Cell Analysis Reveals That Insulation Maintains Signaling Specificity Between Two Yeast MAPK Pathways with Common Components

See allHide authors and affiliations

Science Signaling  19 Oct 2010:
Vol. 3, Issue 144, pp. ra75
DOI: 10.1126/scisignal.2001275


Eukaryotic cells use multiple mitogen-activated protein kinase (MAPK) cascades to evoke appropriate responses to external stimuli. In Saccharomyces cerevisiae, the MAPK Fus3 is activated by pheromone-binding heterotrimeric guanosine triphosphate–binding protein (G protein)–coupled receptors to promote mating, whereas the MAPK Hog1 is activated by hyperosmotic stress to elicit the high-osmolarity glycerol (HOG) response. Although these MAPK pathways share several upstream components, exposure to either pheromone or osmolyte alone triggers only the appropriate response. We used fluorescence localization– and transcription-specific reporters to assess activation of these pathways in individual cells on the minute and hour time scale, respectively. Dual activation of these two MAPK pathways occurred over a broad range of stimulant concentrations and temporal regimes in wild-type cells subjected to costimulation. Thus, signaling specificity is achieved through an “insulation” mechanism, not a “cross-inhibition” mechanism. Furthermore, we showed that there was a critical period during which Hog1 activity had to occur for proper insulation of the HOG pathway.


Budding yeast (Saccharomyces cerevisiae) contains multiple mitogen-activated protein kinase (MAPK) pathways (1). The mating pheromone response pathway and the Sho1 branch of the high-osmolarity glycerol (HOG) pathway both use the MAPK kinase kinase (MAPKKK) Ste11 and its cofactors and activators (Fig. 1, highlighted in red). In the Sho1 branch of the HOG pathway (2), Ste11 activates the MAPKK Pbs2 (Fig. 1A); Pbs2 serves as a scaffold to hold these components in a complex (3) and activates Hog1 MAPK. Activated Hog1 confers hyperosmotic stress resistance by modifying cytoplasmic substrates that increase glycerol production (4, 5) and promotes adaptation to hyperosmotic conditions by entering the nucleus and regulating gene expression (6). The mating pathway (7) is initiated by G protein–coupled receptor (GPCR)–mediated activation of Ste11 (Fig. 1B), which activates the MAPKK Ste7, which activates Fus3 MAPK. This MAPK signaling cascade is facilitated by the scaffold protein Ste5 (8, 9). Pheromone also stimulates a second MAPK (Kss1), but only transiently (10) and in a manner that does not require its direct binding to Ste5 (9, 11). Fus3 promotes polarized growth (shmoo formation) and enters the nucleus, inducing expression of genes for mating (7, 12). The primary role of Kss1 in haploids is to promote expression of the genes required for invasive growth (IG) when glucose becomes limiting (13, 14).

Fig. 1

The HOG and mating MAPK pathways in S. cerevisiae. Physical interactions are denoted by overlap of the proteins depicted. (A) The HOG pathway responds to an increase in external osmolarity and induces the STL1prom-td-Tomato reporter. In isotonic medium, the Sln1 osmosensor phosphorylates Ypd1, which in turn transfers phosphate to Ssk1. Hyperosmotic conditions reduce Sln1 activity, resulting in unphosphorylated Ssk1, which binds to redundant MAPKKKs Ssk2 and Ssk22, alleviating their autoinhibition (3). Also, upon hyperosmotic shock, an osmosensor comprising a tetraspanin (Sho1) and its associated factors, Opy2 and two transmembrane mucins (Msb2 and Hkr1), recruits guanosine triphosphatase Cdc42 in its GTP-bound state, leading to activation of p21-activated kinase (PAK) Ste20 (3, 55). Ste20 in turn activates MAPKKK Ste11 brought to the membrane through interactions with Sho1, Pbs2, and Cdc42 (indirectly through Ste50) (56). Both the Sln1 and Sho1 arms converge on Pbs2 (MAPKK), leading to phosphorylation and activation of Hog1 (MAPK). (B) The mating pathway (1, 7) is stimulated by pheromone and induces the FUS1prom-eGFP reporter and a pronounced change in morphology (shmoo formation). In MATa cells, binding of α-factor to its GPCR (Ste2) causes release of the Gβγ complex from its inhibitory Gα subunit. Liberated Gβγ binds Ste20, Far1, and Ste5 to initiate signaling. Far1 recruits the guanine exchange factor Cdc24 (57), facilitating formation of GTP-bound Cdc42, activating Ste20. Ste20 phosphorylates MAPKKK Ste11, which is delivered to the membrane by association with the Ste5 scaffold protein and interaction (mediated by Ste50) with Cdc42. The MAPKK Ste7 and MAPK Fus3 bind Ste5, whereas the MAPK Kss1 does not. MATα cells respond to a different pheromone, a-factor, through a different GPCR (Ste3), but activate the same downstream pathway. Components used by both pathways (red) raise the potential for cross talk.

When Hog1 is absent (15) or inhibited (16), hyperosmotic stress inappropriately evokes mating (and IG) pathway gene expression. Hence, another role of Hog1 is to block this improper “cross talk.” Hog1 could prevent activation of the other pathways by two general mechanisms: direct inactivation of one or more of the pathway components (inhibition) or confinement of one or more shared signaling proteins to the HOG pathway (insulation). To distinguish these alternatives, we examined the behavior of individual cells exposed to both stimuli simultaneously. If mutual inhibition occurs, stimulation of one pathway should prevent activation of the other; cells should display one or the other (or neither) response, but not both. Although a previous study indicated that activation of the HOG and pheromone response pathways was mutually exclusive (17), we found, using reporters that monitor minute and hour time scale responses, that costimulation activates both pathways concurrently in every cell. Our findings suggest that the maintenance of signaling specificity observed in response to a single stimulus arises from insulation, not from mutual cross-inhibition. In addition, our kinetic analysis defined a temporal requirement for Hog1 activity in pathway insulation. Whereas osmotic stress can stimulate the mating (and IG) pathway inappropriately if Hog1 activity is compromised, we also found that the mating pathway failed to inappropriately activate Hog1 even when Fus3, Kss1, or both were inactivated. Thus, our studies reveal previously unappreciated interrelationships among these MAPK pathways.


Analysis of the short-term response with pathway-specific localization reporters

To examine HOG and mating pathway response in the short term, we monitored cells expressing a red fluorescent protein (RFP)–coupled Hog1 fusion protein (Hog1-td-Tomato) and a green fluorescent protein (GFP)–coupled Ste5 fusion protein [Ste5-(GFP)3] (Table 1). Cells expressed both integrated Hog1-td-Tomato, which translocates into the nucleus within 5 to 10 min after hyperosmotic challenge (16, 18), and plasmid-borne Ste5-(GFP)3, which accumulates as discrete puncta at the plasma membrane within 15 min after pheromone treatment (19, 20). Treatment of cells with α-factor alone (1 μM) relocalized Ste5 to plasma membrane puncta but did not cause nuclear entry of Hog1 (Fig. 2A, second column), and exposure to sorbitol alone (1 M) caused nuclear accumulation of Hog1 but did not cause membrane recruitment of Ste5 (Fig. 2A, third column). In contrast, upon costimulation for the same time period (15 min), Hog1-td-Tomato was nuclear, and Ste5-(GFP)3 was recruited to the plasma membrane in most (80%) of the cells in which both probes could be readily visualized (250 total cells scored) (Fig. 2A, fourth column). Thus, no mutual cross-inhibition occurred between the HOG and mating pathways on the minute time scale.

Table 1

Yeast strains used in this study.

View this table:
Fig. 2

The HOG and mating pathways do not display cross-inhibition upon concomitant stimulation. (A) Representative images of the localization of Hog1-td-Tomato and Ste5-(GFP)3 in cell cultures before and after stimulation for 15 to 16 min with the indicated concentrations of sorbitol, α-factor, or both. Arrows, Hog1-td-Tomato nuclear localization (upper right) and Ste5-(GFP)3 accumulation at the plasma membrane (middle right). Scale bar, 1 μm. (B) A strain (YJP212) harboring integrated transcriptional reporters for the mating (FUS1prom-eGFP) and HOG (STL1prom-td-Tomato) pathways was grown to mid-exponential phase and then subjected to either no stimulus or the indicated concentrations of sorbitol, α-factor, or both and then examined after 2 hours by fluorescence microscopy. Scale bar, 10 μm. (C) Scatterplot of the images obtained in (B), in which each dot depicts the average pixel intensities of the eGFP and td-Tomato fluorescence in an individual cell under each of the conditions tested (n ≈ 400 cells per condition). The plot is divided into four quadrants (1, 2, 3, and 4) to quantify the proportion of the cells that induced one, the other, or both reporters under the four conditions indicated, which are color coded.

Analysis of the longer-term response with pathway-specific transcriptional reporters

To assess pathway activation in individual cells on the hour time scale, we used transcriptional reporters that each express a different fluorescent protein: STL1prom-td-Tomato (Fig. 1A, bottom), which in wild-type cells responds only to hyperosmotic stress, producing a red signal, and FUS1prom-eGFP (enhanced GFP) (Fig. 1B, bottom), which in wild-type cells responds only to pheromone, producing a green signal. These reporters (Table 1) were integrated at their respective chromosomal loci without disrupting the native STL1 or FUS1 coding sequences or promoters, as described in Materials and Methods. An important aspect of the creation of these constructs is that we replaced the LEU2 marker in each construct with a wild-type version of the ADE2 gene because ade2 mutants accumulate a purine pathway intermediate, which can polymerize to form a red pigment that interferes with fluorescence microscopy.

In the absence of any stimulus, 99% of the cells displayed neither a red nor a green signal (Fig. 2, B and C, quadrant 3). When the reporter strain was exposed to 0.5 M sorbitol alone, 96% of the cells produced a readily detectable signal in the red channel, but not in the green channel (Fig. 2, B and C, quadrant 4); when treated with 5 nM α-factor alone, 99% of the cells exhibited a readily detectable green signal, but not a red signal (Fig. 2, B and C, quadrant 1). Costimulation with osmolyte and pheromone elicited both red and green signals in 93% of the cells (Fig. 2, B and C, quadrant 2), and we observed this dual response over a range of sorbitol (0.25 to 1.0 M) and α-factor (5 to 30 nM) concentrations, after 1 or 2 hours of stimulation (figs. S1 and S2), and using a completely independent analysis method, flow cytometry (fig. S3). Thus, both pathways were activated in nearly every cell, and activation of one pathway did not prevent response from the other pathway.

Toxicity associated with simultaneous exposure to high osmolarity and pheromone

Because our results conflicted with a previous report that claimed that co-stimulation of another strain containing similar transcriptional reporters evoked expression of one or the other reporter in any given cell, but never both (17), we sought to understand the basis for the difference (see Supplementary Materials for additional explanation). Long-term (1 to 2 days) exposure of MATa ssk1Δ cells to both strong hyperosmotic stress and high pheromone concentration causes cell death (21). One mechanism proposed for this toxicity was that, in ssk1Δ cells with only the Sho1 arm of the HOG pathway, Fus1 produced in response to mating pathway activation bound to Sho1, which uncoupled the Sho1-dependent branch of the HOG pathway (21), which would make the cells intolerant of hyperosmotic stress. Despite the presence of both branches of the HOG pathway in our dual reporter strain, we observed loss of viability (33% dead cells of 474 total cells scored) within 2 hours upon costimulation with 30 nM α-factor and 1 M sorbitol (Fig. 3). This cell death required the presence of Fus1, a transmembrane protein involved in the cell fusion step of mating (22). Removal of Fus1 rescued viability (only 2% dead cells of 531 scored), but did not alter the concomitant appearance of the HOG pathway and mating pathway outputs (Fig. 3).

Fig. 3

Concurrent stimulation with high concentrations of sorbitol and α-factor can cause cell death. Strains YJP213 (FUS1+) and YJP385 (fus1Δ) were grown to mid-exponential phase, stimulated with 30 nM α-factor and 1 M sorbitol for 2 hours, and then examined by fluorescence microscopy. Arrows, dark nonrefractile cell profiles in bright-field images (left panels) are dead cells. Merged td-Tomato and eGFP images showing coexpression of the HOG and mating pathway reporters (right panels). Scale bars, 10 μm.

To determine whether either arm of the HOG pathway was responsible for the Fus1-dependent toxicity associated with dual exposure of yeast to pheromone and hyperosmotic stress, we analyzed pathway activation in Sho1- and Ssk1-deficient cells. Under our standard conditions (5 nM α-factor and 0.5 M sorbitol), both reporter genes were expressed in individual cells exposed to both stimuli regardless of whether cells had only the Sln1 branch (sho1Δ cells) or only the Sho1 branch (ssk1Δ cells) (Fig. 4, A and B); however, the absence of Sho1 attenuated the response to α-factor somewhat. We then examined the response in the same cells over a range of stimulant concentrations. We always observed dual pathway activation in either sho1Δ (fig. S4) or ssk1Δ cells (fig. S5), although activation through the Sho1 branch (ssk1Δ cells) was less responsive to the lowest concentration (0.25 M) of sorbitol, in agreement with other evidence indicating that the Sho1 branch responds best to severe hyperosmotic shock (23). Moreover, after 2 hours of exposure to the highest concentrations of stimuli (30 nM α-factor and 1 M sorbitol), the fraction of the sho1Δ cells that were dead (23% of 532 scored) was not substantially lower than the fraction of the ssk1Δ cells that were dead (39% of 450 scored).

Fig. 4

No cross-inhibition is exerted on either MAPK of the mating pathway or on either branch of the HOG pathway. (A) Derivatives of the dual transcriptional reporter strain expressing Fus3 alone (YJP313), Kss1 alone (YJP336), the Sln1 branch of the HOG pathway only (YJP407), or the Sho1 branch of the HOG pathway only (YJP406) were analyzed as in Fig. 2A. Scale bar, 10 μm. (B) The data obtained in (A) are plotted as described in Fig. 2C. (C) Dependency of cross talk from the HOG pathway on Ste5 and Kss1. Strains YJP131 (hog1Δ), YJP301 (hog1Δ kss1Δ), YJP571 (hog1Δ ste5Δ), and YJP573 (hog1Δ kss1Δ ste5Δ) were grown to mid-exponential phase and then treated with 1 M sorbitol for 2 hours. Bars show average intensity of the eGFP signal, and SD for the cell population is shown. For merged bright-field and fluorescence images of both reporters and scatterplots of the average pixel intensity for individual cells, see fig. S10. WT, wild type.

Because we detected dual pathway activation in cells lacking the Sln1 arm (ssk1Δ), this suggests that, if Fus1 blocks the Sho1-dependent branch, 2 hours was insufficient time for this to occur. Additionally, in contrast to the previously suggested mechanism of toxicity, which was analyzed after 1 to 2 days of exposure in Sln1-deficient cells (1), our findings indicate that the toxicity caused by acute stimulation (≤2 hours) with both sorbitol and pheromone cannot simply be attributed to Fus1 inhibition of the Sho1-dependent branch of the HOG pathway.

Given the loss of viability observed in our experimental paradigm, we performed a detailed analysis of the paradigm used by McClean et al. (17) (see Supplementary Materials for additional explanation) and determined that the combination of sonication and hyperosmotic shock that they used greatly increased the amount of cell death (as judged by the dark nonrefractile cell profiles in bright-field images; fig. S6, left). These dead cells displayed increased fluorescence, mainly in the red channel (fig. S6, right). We attributed this signal to autofluorescence because it displayed a broad excitation and emission spectrum that was brightest in red wavelengths and did not resemble spectrally either eGFP or RFP. Our investigation indicates that McClean et al. (17) failed to monitor monomeric RFP (mRFP) signals (indicative of HOG pathway activity) because they were relatively dim and instead counted the autofluorescent dead cells as those that had activated the HOG pathway.

Effect of pathway preactivation

We tested whether the ability of one pathway to block the other required previous exposure to one stimulus before challenge with the other. Simultaneous costimulation (fig. S7A), as well as pretreatment of cells with 15 nM α-factor for various periods (up to 45 min) before challenge with sorbitol, did not prevent subsequent HOG pathway activation (fig. S7B). Likewise, preexposure of cells to 0.5 M sorbitol for various periods (up to 45 min) did not prevent subsequent mating pathway activation by pheromone (fig. S7C).

Redundancies in each pathway might permit reporter transcription, even if mutual cross-inhibition occurs between distinct branches in each pathway. However, regardless of whether cells had only Fus3 MAPK or only Kss1 MAPK for mating pathway signaling, or only the Sln1 branch or only the Sho1 branch for HOG pathway signaling, both reporter genes were expressed in individual cells exposed to both stimuli (Fig. 4, A and B, and figs. S4, S5, S8, and S9). FUS1prom-eGFP expression stimulated by Kss1 alone (fus3Δ cells) was less robust than that stimulated by Fus3 alone (kss1Δ cells), as expected (24) because of the transient activation of Kss1 by pheromone.

When cells lacking both Hog1 and Kss1 were subjected to hyperosmotic stress, there was only a modest decrease in the amount of FUS1-eGFP reporter gene expression evoked by cross talk relative to cells lacking Hog1 alone (Fig. 4C and fig. S10). There was a more substantial reduction in the transcriptional output elicited by cross talk in cells lacking both Hog1 and Ste5 (Fig. 4C and fig. S10), indicating that Fus3 activated in a Ste5-dependent manner is an important contributor to the observed response. In cells lacking Hog1, Ste5, and Kss1, reporter gene expression was greatly reduced, indicating that direct activation of Fus3 by Ste7 is inefficient (Fig. 4C).

Hog1 and establishment of pathway insulation

Our results indicate that signaling through the mating pathway and the HOG pathway occurred independently without any detectable cross-inhibition; therefore, an insulation mechanism appears to maintain pathway independence. Because the absence of functional Hog1 results in activation of the mating pathway inappropriately in response to hyperosmotic challenge (15, 16), Hog1 activity is required to maintain insulation, implying that Hog1-mediated phosphorylation of one or more targets establishes conditions that limit signal propagation solely to HOG pathway constituents.

To determine the amount of Hog1 catalytic activity that was required for HOG pathway insulation, we replaced the HOG1 locus in the dual reporter strain with Hog1(T100A), a mutant susceptible to rapid inhibition by the cell-permeable competitive inhibitor 1-NM-PP1 (16) (Table 1). Exposure of cells to 15 μM inhibitor is sufficient to completely block Hog1 activity (16). The presence of 15 μM inhibitor combined with exposure of the cells to 1 M sorbitol induced activation of the mating pathway reporter but failed to induce the HOG pathway reporter (Fig. 5A and fig. S11), consistent with the requirement for Hog1 activity for pathway insulation and activation. Conversely, at low inhibitor concentration (0.015 μM), which should allow near-full activity of Hog1, addition of osmolyte only induced expression of the HOG pathway reporter (Fig. 5A). At an intermediate inhibitor concentration (0.15 mM), expression of both the HOG and the mating reporters was partial (roughly 50% of the maximum in each case), as expected if the ability to suppress cross talk is proportional to Hog1 activity (Fig. 5A). By adding 1-NM-PP1 at various times after challenge with 1 M sorbitol, we found that Hog1 activity was required for 20 to 30 min to establish the conditions to insulate the HOG pathway (Fig. 5B and fig. S11).

Fig. 5

Sustained Hog1 catalytic activity is required to prevent cross talk to the mating pathway. The cells were a derivative (YJP123) of the dual transcriptional fluorescent reporter strain, harboring as the sole source of Hog1 an analog-sensitive mutant, Hog1(T100A) (16). (A) The yeast was grown to mid-exponential phase, exposed to 1 M sorbitol at the indicated concentrations of the inhibitor (1-NM-PP1) for 2 hours, and then examined by fluorescence microscopy. Average intensity and SD for the cell population of eGFP (green squares) and td-Tomato (red triangles) reporters are shown. For fluorescence images and scatterplots of the average pixel intensity for individual cells, see fig. S11. (B) Yeast was grown to mid-exponential phase, the inhibitor (15 μM 1-NM-PP1) was added at the indicated times after cells were subjected to 1 M sorbitol, and the outputs of the HOG and mating pathways were then assessed 2 hours later by fluorescence microscopy. Average intensity and SD for the cell population of eGFP (green squares) and td-Tomato (red triangles) reporters are shown. For fluorescence images and scatterplots of the average pixel intensity for individual cells, see fig. S11.

Insulation of the pheromone response pathway from the HOG pathway

Our work provides further evidence for the well-established role for Hog1 in preventing activation of the mating pathway under conditions of hyperosmotic stress (15, 16). However, little is known about the insulation of the mating pathway from the HOG pathway. To examine mating pathway insulation, specifically whether Fus3 or Kss1 action, or the activity of both, prevented HOG pathway activation in response to pheromone, we replaced the FUS3 and KSS1 loci in the dual reporter strain with 1-NM-PP1–sensitive alleles (Table 1): Fus3(Q93G), which has been described previously (25), and Kss1(E94A), which we constructed for this work (fig. S12). At the highest inhibitor concentration (15 μM), which should completely block Fus3 and Kss1 activity, pheromone failed to induce expression of the mating pathway reporter; however, we also did not detect expression of the HOG pathway reporter (Fig. 6A). Decreased inhibitor concentration resulted in greater mating pathway reporter expression without detectable HOG pathway reporter expression. A MATa fus3Δ kss1Δ double-null mutant stimulated with even a high pheromone concentration (1 μM) also exhibited no detectable expression of the HOG pathway reporter (Fig. 6B, left). Thus, pheromone did not activate the HOG pathway, regardless of the activity state of the Fus3 or Kss1 MAPKs.

Fig. 6

Inhibition of Fus3 and Kss1 during pheromone stimulation does not cause cross talk to the HOG pathway. (A) A derivative (YJP334) of the dual transcriptional reporter strain, harboring as the sole source of both Fus3 and Kss1 analog-sensitive mutants Fus3(Q93G) and Kss1(E94A), was grown to mid-exponential phase, treated with 30 nM α-factor at the indicated concentrations of the inhibitor (1-NM-PP1) for 2 hours, and examined by bright-field and fluorescence microscopy; the corresponding images were overlaid (upper panels), and the output in single cells (average pixel intensities for the td-Tomato and eGFP signals) is shown as scatterplots (lower panels). The table indicates the percentages of the population that were unbudded, had formed a shmoo, and had both formed a shmoo and displayed prominent reporter gene expression at each concentration of inhibitor tested. (B) Several derivatives of the dual reporter strain (YJP308 fus3Δ kss1Δ, YJP387 fus3Δ kss1Δ fus1Δ, and YJP390 fus3Δ kss1Δ fus1Δ ssk1Δ) were grown to mid-exponential phase, treated with 1 μM α-factor for 2 hours, and examined as in (A). Scale bars, 10 μm.

To determine whether the two arms of the HOG pathway exerted any inhibitory effect on each other in response to pheromone signaling, we analyzed HOG pathway reporter expression in cells lacking both of the mating pathway MAPKs, as well as lacking either Fus1 (to potentially enhance Sho1 signaling) or both Fus1 and Sln1 signaling (Table 1). Pheromone failed to stimulate the HOG pathway reporter in MATa fus3Δ kss1Δ fus1Δ cells (Fig. 6B, middle) or in MATa fus3Δ kss1Δ fus1Δ ssk1Δ cells (Fig. 6B, right). Thus, these data indicate that pheromone-activated Ste11 does not phosphorylate Pbs2, which suggests that the Ste5-bound pool of Ste11 does not exchange with the Sho1- and Pbs2-bound pool. Activation of the HOG pathway by a pheromone stimulus requires “artificial” conditions to purposefully divert the signal, such as engineered synthetic scaffold proteins (26) or chimeric MAPKs (27).

MAPK activity and mating pheromone responses

Under conditions of no extracellular pheromone proteolysis, the dose of α-factor required for cells to exhibit half-maximal cell division arrest and half-maximal induction of a pheromone-induced gene (cell surface agglutinin) was two orders of magnitude lower than the dose of α-factor required for half-maximal shmoo formation (28). One interpretation of these findings is that different amounts of Fus3 activity evoke different aspects of pheromone response. To test this proposition, in our experiments where we inhibited our Fus3(Q93G) Kss1(E94A) strain with different concentrations of 1-NM-PP1 (Fig. 6A), we quantified the number of unbudded cells (exhibited G1 cell cycle arrest), activity through the mating pathway (reporter gene expression), and shmoo formation in response to 30 nM α-factor after 2 hours.

This analysis showed that shmoo-forming cells were almost invariably also bright green; thus, if a cell had sufficient Fus3 activity to arrest the cell cycle and induce shmoo morphology, it also showed robust FUS1 expression (Fig. 6A). In this regard, one function of Fus3 is phosphorylation of Far1, an inhibitor of the cyclin-dependent kinase Cdc28, thereby causing cell cycle arrest in G1 (29), the stage at which mating pathway activation normally occurs (19). Thus, our observations support the conclusion that the amount of Fus3 activity required for shmoo formation is indeed greater than that needed to impose G1 arrest and induce detectable FUS1 transcription.


Inadvertent stimulation of Fus3 in the pheromone pathway prevented by Hog1 through an insulation mechanism

We found that yeast can simultaneously respond to hyperosmotic stress and pheromone by showing that reporters for each pathway were activated upon costimulation and even when both stimuli were applied sequentially. Hence, our data rule out that signaling specificity is imposed by the mechanism of “mutual cross-inhibition” as suggested (17) and support an insulation mechanism in which Hog1 action imposes signaling specificity by preventing the HOG pathway signal from reaching or initiating the Fus3 MAPK (or Kss1 MAPK) cascades. McClean et al. (17) reported that cells exposed to pheromone and osmolyte at the same time activated either the pheromone-responsive gene reporter or the HOG pathway gene reporter, but never both. Upon detailed analysis of the yeast exposed to both stimuli, we found that toxicity associated with the methods they used may have obfuscated the proper interpretation (see Supplementary Materials). Although the specific substrates on which Hog1 acts to prevent cross talk remain unknown, there is already support in the literature consistent with an insulation mechanism (30).

Although our insulation model may appear to conflict with studies showing that phosphorylation of Kss1 MAPK (and its cognate MAPKK Ste7) occurs even in HOG1+ cells within 15 min after cells are challenged with hyperosmotic KCl (30, 31), we believe that our model is compatible with these data. Kss1 phosphorylation occurs slowly, with low abundance, and is transient (detected only at 15 min) (30). In contrast, when Hog1 was absent, Kss1 activation upon exposure to KCl was markedly more robust and sustained (continuously increasing for at least 90 min) (30). Moreover, no activated Fus3 (the primary MAPK of the pheromone response pathway) was observed after KCl treatment in HOG1+ cells, whereas activation of Fus3 was detectable in the absence of Hog1 (30). Thus, the presence of functional Hog1 impeded activation of both the Fus3 and the Kss1 MAPKs in response to KCl, which is consistent with our insulation model.

Because detectable Kss1 activation, but not detectable Fus3 activation, occurs in HOG1+ cells challenged with KCl, it is possible that this response may occur through a nonspecific effect of high salt, which may activate the plasma membrane tetraspanin Sho1 (32, 33) as an IG sensor, rather than through Sho1 molecules that are serving as osmosensors per se. When hyperosmotic stress activates Sho1, it acts in conjunction with a small, single-pass transmembrane protein (Opy2) (34), whereas for initiation of the IG response under conditions of nutrient limitation, Sho1 mainly acts in conjunction with two larger, heavily O-glycosylated, mucin-like single-pass transmembrane proteins (Msb2 and Hkr1) (35, 36). Indeed, other conditions that abrogate the ability of the HOG pathway osmosensors to activate Hog1 (deficient N- and O-linked glycosylation) can still lead to activation of Kss1 (37, 38), perhaps because under these circumstances the complexes that contain Ste11 bound to Sho1 and Pbs2 are disrupted, allowing this MAPKKK to escape, thereby permitting activation of Ste7.

To prevent nonspecific effects due to high salt, we subjected the cells to hyperosmotic stress with a nonpermeant uncharged osmolyte (sorbitol). Even if the observed transient Kss1 phosphorylation seen in experiments with KCl as the hyperosmotic stress arises through cross talk from Sho1 acting as an osmosensor, the fact that this weak Kss1 activation disappears within 15 min after KCl treatment is consistent with our finding (using the analog-inhibitable Hog1 allele) that Hog1 must be active for at least 20 to 30 min to impose its inhibition of cross talk (Fig. 5B).

We observed a concurrent decrease in FUS1prom-driven eGFP expression when cells were stimulated with pheromone and concomitantly with increasing concentrations of sorbitol (figs. S1 and S2), which may appear at odds with our insulation model. However, in this experimental paradigm, the readout for stimulation of each pathway is production of the cognate FP reporter, which requires protein synthesis. Hyperosmotic conditions cause a modest and progressive drop in overall translation rate as the osmolyte concentration increases (3941). Thus, it is not unexpected that the eGFP signal produced at a given dose of pheromone might be decreased as the concentration of sorbitol was increased, which is what we observed. Moreover, at a given sorbitol concentration, the eventual amount of eGFP produced was the same regardless of whether pheromone was added at the same time as the cells were challenged with osmolyte or whether Hog1 was preactivated by the osmolyte treatment before challenge with pheromone (fig. S7). Hence, the observed diminution in FP signal was independent of the amount of time activated Hog1 was present and thus due to the osmolyte per se. Therefore, the observed modest decrease in eGFP signal as the sorbitol concentration increased is likely due to less efficient translation of the messenger RNA encoding the FP reporter, rather than any specific inhibitory effect that Hog1 exerts. Consistent with this view, as we demonstrated with the pathway-specific relocalization reporters (Fig. 2A), hyperosmotic conditions did not inhibit mating signaling by preexisting proteins, because hyperosmotic conditions did not detectably reduce the efficiency of Ste5 recruitment to the plasma membrane in response to a pheromone stimulus.

Potential insulation mechanisms

In wild-type cells, hyperosmotic challenge does not cross-activate pheromone response. By contrast, in Hog1-deficient cells, cross-activation of pheromone (or IG) response upon hyperosmotic stress occurs after a delay on the hour time scale (15, 16, 23). Thus, Hog1-mediated phosphorylation might confine Ste11 to the HOG pathway if such modifications cause conformational changes or create interfaces that stabilize the multivalent contacts in Sho1-Opy2-Ste50-Ste11-Pbs2 complexes, preventing “escape” of activated Ste11.

The Ste11-containing ensembles for HOG response are tethered to the plasma membrane. Only Ste11 associated with the scaffold protein Ste5 can activate pheromone response. Ste5 undergoes constitutive nucleocytoplasmic shuttling but is located primarily in the nucleus when cells are not exposed to pheromone (19, 42). This compartmentalization may also limit stimulation of pheromone signaling by Ste11 activated through hyperosmotic stress, especially if, after its stimulus-induced translocation into the nucleus, Hog1 phosphorylates and modifies the properties of Ste5, preventing its association with Ste11. However, our own findings make this an unlikely scenario for how Hog1 catalytic activity prevents cross talk to the pheromone response pathway. Indeed, given our finding that hyperosmotic stress does not interfere with pheromone signaling, it seems clear that the target upon which Hog1 acts to prevent cross talk cannot be any component unique to the mating pathway.

Ste11 is situated relatively proximal to the initial stimuli in both the HOG and the pheromone pathways (Fig. 1); hence, how Hog1 action might affect Ste11 is a straightforward way to view how cross talk may be prevented. However, there are other shared proteins operating at or near this same node, including Cdc42, Ste50, and Ste20. Indeed, one study suggested that Hog1 phosphorylation of Ste50 prevents cross talk (43), but we and others (31) were unable to confirm this. In cells expressing a Ste50 mutant lacking all five of its consensus MAPK phosphorylation sites (all -SP- and -TP- converted to -AP-), normal induction of the HOG transcriptional reporter was observed, but induction of the mating pathway reporter could not be detected (fig. S13A). We have noted that Hog1 feedback phosphorylates Pbs2 in vitro (fig. S13B), which theoretically might help to sequester Ste11 in vivo. However, in cells expressing a Pbs2 mutant lacking all six of its consensus MAPK phosphorylation sites, or in cells expressing both this Pbs2 mutant and the Ste50 mutant, normal induction of the HOG transcriptional reporter was observed, but induction of the mating pathway reporter could not be detected (fig. S13A). Hence, the targets of Hog1 that prevent hyperosmotic stress from cross-activating the pheromone and IG responses remain unidentified.

Hierarchies in MAPK response networks

Activation of the mating pheromone response pathway obviates subsequent transcriptional output from the IG pathway because Fus3-mediated phosphorylation initiates ubiquitin- and proteasome-dependent destruction of the transcription factor (Tec1) required for expression of IG genes (44, 45). Thus, once stimulated, the pheromone response pathway establishes primacy over the IG pathway, imposing a specific decision tree with respect to these two developmental options. Do similar hierarchies exist among the other MAPK pathways in S. cerevisiae? First, our results show that a function of Hog1 MAPK is to establish conditions that prevent inadvertent activation of the pheromone response MAPK Fus3 in response to hyperosmotic stress alone. Second, and in contrast to the HOG pathway, our findings demonstrate that pheromone action alone does not improperly activate the HOG pathway, even when the Fus3 and Kss1 MAPKs are inactive. Third, we found that, upon costimulation, neither pathway overrides the transcriptional output of the other. A physiological rationale for independent function of the pheromone and HOG responses may be the need to maintain osmotic balance during the cell fusion step of mating. That these two pathways act independently despite involving some of the same gene products illustrates the combinatorial diversity with which MAPK signaling modules can be assembled. Cdc42, Ste20, Ste50, and Ste11 are capable of multitasking, in the sense that they incorporate into complexes that propagate discrete signals without degradation or inappropriate cross talk, provided that active Hog1 is present. Elucidation of the molecular mechanism by which Hog1 action prevents activated HOG components from inadvertently stimulating the mating and IG pathways may reveal some new principles underlying MAPK signaling specificity.

Materials and Methods

Yeast strains and plasmids

All plasmid and strain constructions were carried out with standard genetic and molecular biology methods (46, 47). Yeast strains generated and used in this study are listed in Table 1. Plasmids constructed and used in this study are listed in Table 2. The transcriptional reporter constructs were created by simple single-crossover integration (“loop-in”) events, yielding a tandem promoter-FP-vector-promoter-ORF (open reading frame) arrangement. To do so, we fused the promoters of the genes of interest to the sequence encoding the desired FP and cloned them into a yeast integration vector (YIp) containing LEU2 as the selectable marker. The resulting YIp plasmid was digested with a restriction enzyme that has a single unique cleavage site within the promoter of interest, and the resulting linearized DNA was used for transformation of the appropriate recipient leu2 ade2 yeast strain, selecting for Leu+ transformants [by growth on SC-Leu medium (synthetic complete medium lacking leucine)]. Specifically, the FUS1prom-eGFP reporter consisted of 993 base pairs (bp) 5′ of the FUS1 start codon fused to eGFP that was hemagglutinin (HA)–tagged at its N terminus, cloned into the Bam HI and Eco RI sites of the LEU2-marked integration vector, YIplac128. The resulting plasmid was then linearized with Bsa AI and used to transform YPH499. The STL1prom-td-Tomato reporter contained 972 bp 5′ of the STL1 start codon fused to td-Tomato that was also HA-tagged at its N terminus, cloned into the Bam HI and Eco RI sites of YIplac128. The resulting plasmid was linearized with Nru I and used to transform YPH500. Creation of the expected insertions at the FUS1 locus [FUS1prom-eGFP::(vector)-LEU2-(vector)::FUS1prom-FUS1] and at the STL1 locus [STL1prom-td-Tomato::(vector)-LEU2-(vector)::STL1prom-STL1], respectively, was confirmed by polymerase chain reaction (PCR) analysis with appropriate primers. Both YPH499 and YPH500 are ade2 strains, which accumulate a pigment that can cause significant red autofluorescence. To eliminate this background, we generated derivatives of both transformants in which the integrated LEU2 marker was replaced with ADE2, using standard one-step (double crossover) gene replacement, as described elsewhere (48). The resulting now-ADE2+ single integrants were then mated, the resulting diploid was sporulated, and tetrads were dissected to obtain a MATa spore containing both reporters (YJP73). In experiments where cells were treated with mating pheromone, to avoid changes in pheromone concentration, bar1Δ derivatives of the dual reporter strain were used, which lack the secreted endoprotease responsible for α-factor degradation (49, 50).

Table 2

Plasmids used in this study.

View this table:

To construct YJP490, we replaced the STE50 locus in YJP213 with the C.g TRP1 gene, and plasmid pRS306-STE50-5A was linearized with Age I and then integrated into the promoter of the disrupted ste50::C.g.TRP1 locus. To construct YJP455, we replaced the PBS2 locus in YJP213 with the C.g HIS3 gene, and a URA3-marked CEN plasmid YCplac33-PBS2-6A expressing Pbs2(6S,T→A) was introduced by DNA-mediated transformation. PBS2 in strain YJP490 was replaced with the C.g HIS3 gene, and then the STE50-5A::URA3 gene in this strain was converted to STE50-5A::LYS2; the resulting derivative was transformed with plasmid YCplac33-PBS2-6A, yielding strain YJP495. A PCR product containing the PBS2 coding sequence, as well as 545 bp upstream of the start codon and 65 bp downstream of the stop codon, was cloned into the Bam HI–Pst I sites of YCplac33, and the six consensus MAPK phosphorylation sites were converted to Ala by site-directed mutagenesis, yielding YCplac33-PBS2-6A.

Growth conditions

The transcriptional reporter strain to be examined was grown overnight in 5 ml of synthetic complete dextrose (SCD) medium, reinoculated into fresh medium at an absorbance at 600 nm (A600 nm) of 0.05, and grown for 5 hours to mid-exponential phase. Stimulation was achieved by diluting the culture with an equal volume of fresh medium containing twice the desired final concentrations of α-factor or d-sorbitol (Sigma-Aldrich), or both, at 30°C in glass test tubes pretreated with 1% bovine serum albumin (Sigma-Aldrich) and rinsed before use, and gently shaken in a water bath at 250 rpm. Precoating of the test tubes with 1% bovine serum albumin was required to prevent variable nonspecific adsorption of α-factor to the glass walls. When multiple concentrations and/or different conditions were examined, the timing of culture growth and treatment was staggered appropriately to permit sufficient time for manipulation and analysis, such that the total elapsed time of stimulation was identical for all samples. Experiments with the localization reporters were conducted in a similar manner, except that to maintain selection for the plasmid expressing Ste5-(GFP)3 the growth medium was SCD-Ura and cultures were reinoculated at A600 nm of 0.1.


In experiments in which the degree of expression of the fluorescent transcriptional reporters was assessed, the cells in a sample (1.5 ml) of the stimulated culture were collected by brief centrifugation in a microfuge, rapidly resuspended in 15 μl of SCD, immediately spread on a 0.75% agarose pad, and examined under an epifluorescence microscope [model BH-2, equipped with a 60× 1.4 numerical aperture (NA) objective lens; Olympus America]. For each of four fields, the GFP fluorescence was assessed with a 470-nm (40-nm bandwidth) excitation filter and a 525-nm (50-nm bandwidth) emission filter (Endow GFP 47001; Chroma Technology Corp.), and the red td-Tomato fluorescence was imaged with a 560-nm (40-nm bandwidth) excitation filter and a 630-nm (60-nm bandwidth) emission filter (31004 Texas red; Chroma). For automated unbiased cell identification in images, total autofluorescence was monitored with a 330- to 385-nm band-pass excitation filter (UG1; Olympus America). Images were captured with a charge-coupled device camera (Magnafire SP CCD; Olympus America), and for any given filter set, the exact same exposure times were used throughout the entire study. For imaging Hog1-td-Tomato and Ste5-(GFP)3 localization, cultures were stimulated for 10 min, collected by brief centrifugation in a microfuge, resuspended in 5 μl of media with or without stimuli, and then viewed after spreading on a poly-l-lysine–coated glass slide. Images were captured as above, except that a 100× 1.4 NA objective and a neutral density filter (Olympus America) were used.

Image processing

The eight-bit TIFF images were processed with automated image analysis software (CellProfiler, version 1.0.5122) (51). For quantifying transcriptional reporter outputs, appropriate corrections for variable background and uneven illumination were applied where necessary, cell margins were identified with autofluorescence and used to align with the fluorescent images, and then the average pixel intensity was calculated for each cell identified. Units are arbitrary but internally consistent for each fluorophore in each independent experiment presented. Higher pixel intensity represents more fluorescent protein; however, because of the convolution of light under the optical conditions with which these data were acquired, the intensity signals are not strictly correlated linearly with the amount of eGFP or td-Tomato present in the cells (that is, double the fluorophore concentration does not yield an exact doubling of the pixel intensity). For this reason, flow cytometry was also used to validate the observations made by fluorescence microscopy. The scatterplots of average pixel intensity provide a convenient graphical means to represent the fluorescence status of individual cells in the population in the fluorescence microscope images. Images of the Hog1-td-Tomato and Ste5-(GFP)3 fluorescence localization reporter images were also corrected for uneven illumination and variable background with CellProfiler, and the absolute intensities of the images were adjusted to a uniform brightness and contrast with Photoshop CS2 (Adobe Systems Inc.).


Acknowledgments: We thank K. Shokat for the gift of 1-NM-PP1, P. Pryciak for providing the Ste5-(GFP)3 construct, F. Posas for the gift of the Pbs2-EE plasmid, S. Ramanathan for supplying YSR31 strain, and H. Nolla of the University of California, Berkeley, Cancer Research Laboratory for assistance with flow cytometry. Funding: This work was supported by NIH Predoctoral Traineeship grant GM07232 and a Genentech Predoctoral Fellowship (to J.C.P.), by NIH Predoctoral Traineeship grant GM07232 (to E.S.K.), and by NIH R01 Research Grant GM21841 (to J.T.). Author contributions: J.C.P. and J.T. designed the experiments; J.C.P. conducted the experiments; E.S.K. constructed the analog-sensitive Kss1 allele; J.C.P. and J.T. analyzed the results; and J.C.P. and J.T. wrote and revised the manuscript. Competing interests: The authors declare that they have no competing interests.

Supplementary Materials

Materials and Methods


Fig. S1. Co-stimulation for 1 hour results in coactivation of the HOG and mating pathway transcriptional reporters in single cells.

Fig. S2. Co-stimulation for 2 hours results in further coactivation of the HOG and mating pathway transcriptional reporters in single cells.

Fig. S3. Flow cytometry confirms that dual stimulation leads to dual pathway activation in individual cells.

Fig. S4. Restricting MAPK activation to the Sln1 branch of the HOG pathway does not alter HOG and mating pathway reporter coexpression.

Fig. S5. Restricting MAPK activation to the Sho1 branch of the HOG pathway does not alter HOG and mating pathway reporter coexpression.

Fig. S6. Sonication followed by concurrent stimulation with high concentrations of sorbitol and α-factor causes cell death.

Fig. S7. Sequential stimulation of cells does not result in mutually exclusive, or any other history-dependent change in, HOG and mating pathway reporter expression.

Fig. S8. Restricting MAPK activation by the mating pathway to Fus3 does not alter HOG and mating pathway reporter coexpression.

Fig. S9. Restricting MAPK activation by the mating pathway to Kss1 does not alter HOG and mating pathway reporter coexpression.

Fig. S10. Both Kss1- and Ste5-dependent Fus3 activation contribute to cross talk in Hog1-deficient cells.

Fig. S11. Sustained Hog1 catalytic activity is required to prevent cross talk to the mating pathway.

Fig. S12. Validation of analog-sensitive KSS1 alleles.

Fig. S13. Lack of Hog1 phosphorylation of Ste50 or Pbs2 does not permit cross talk.


References and Notes

View Abstract

Stay Connected to Science Signaling

Navigate This Article