Down-regulation of TORC2-Ypk1 signaling promotes MAPK-independent survival under hyperosmotic stress

In eukaryotes, exposure to hypertonic conditions activates a MAPK (Hog1 in Saccharomyces cerevisiae and ortholog p38 in human cells). In yeast, intracellular glycerol accumulates to counterbalance the high external osmolarity. To prevent glycerol efflux, Hog1 action impedes the function of the aquaglyceroporin Fps1, in part, by displacing channel co-activators (Rgc1/2). However, Fps1 closes upon hyperosmotic shock even in hog1∆ cells, indicating another mechanism to prevent Fps1-mediated glycerol efflux. In our prior proteome-wide screen, Fps1 was identified as a target of TORC2-dependent protein kinase Ypk1 (Muir et al., 2014). We show here that Fps1 is an authentic Ypk1 substrate and that the open channel state of Fps1 requires phosphorylation by Ypk1. Moreover, hyperosmotic conditions block TORC2-dependent Ypk1-mediated Fps1 phosphorylation, causing channel closure, glycerol accumulation, and enhanced survival under hyperosmotic stress. These events are all Hog1-independent. Our findings define the underlying molecular basis of a new mechanism for responding to hypertonic conditions. DOI: http://dx.doi.org/10.7554/eLife.09336.001

Hyperosmotic conditions evoke two known signaling modalities. Pathways coupled to alternative osmosensors (Sln1 and Sho1) activate MAPK Hog1, which drives both transcription-independent and -dependent responses that markedly increase both production and intracellular retention of glycerol (Westfall et al., 2008;Saito and Posas, 2012;Hohmann, 2015). Hyperosmotic shock also increases cytosolic [Ca 2+ ] thereby activating calcineurin (CN) (Denis and Cyert, 2002), promoting processes that stimulate retrieval of excess plasma membrane (Guiney et al., 2015). Although both CN-deficient and hog1Δ cells are quite sensitive to the ionic imbalances caused by high salt (e.g., 1 M NaCl), hog1Δ cells are significantly more sensitive to hypertonic stress per se, such as a high concentration of an uncharged impermeant osmolyte (e.g., 1 M sorbitol).
Our understanding of the response to high osmolarity remains incomplete, however. Although it is well documented that preventing glycerol efflux through the aquaglyceroporin Fps1 is essential for yeast to survive hyperosmolarity (Luyten et al., 1995;Tamás et al., 1999;Duskova et al., 2015), and that activated Hog1 can negatively regulate this channel by displacing the Fps1-activating proteins Rgc1/2 (Lee et al., 2013), Fps1 still closes in response to hyperosmotic shock in hog1Δ cells (Tamás et al., 1999;Babazadeh et al., 2014). Therefore, we explored the possibility, as suggested by our screen, that Fps1 is an authentic target of TORC2-dependent Ypk1-mediated phosphorylation, that this modification is important for Fps1 function, and that it is under regulation by hyperosmotic conditions.

Ypk1 phosphorylates Fps1 and hyperosmotic shock inhibits this phosphorylation
The 743-residue enzyme Gpt2 contains one Ypk1 phospho-acceptor motif (646RSRSSSI652). At such sites, Ser residues just penultimate to the canonical one (in red) can be phosphorylated in a Ypk1-dependent manner (Roelants et al., 2011). Therefore, we generated a Gpt2(S649A S650A S651A) mutant. One or more of these three Ser residues is phosphorylated in vivo because, compared to wild-type, Gpt2 3A exhibited a distinctly faster mobility upon SDS-PAGE, a hallmark of decreased phosphorylation ( Figure 1A), just like wild-type Gpt2 treated with phosphatase ( Figure 1-figure supplement 1). However, this phosphorylation did not appear to be dependent on Ypk1 because little change occurred in Gpt2 mobility when an analog-sensitive ypk1-as ypk2Δ strain was treated with the cognate inhibitor (3-MB-PP1) ( Figure 1A).
In marked contrast, three of four predicted Ypk1 sites in the 669-residue Fps1 channel (176RRRSRSR182, , and their phosphorylation requires Ypk1 activity, because, in inhibitor-treated ypk1-as ypk2Δ cells, the mobility of wild-type Fps1 was indistinguishable from that of Fps1(S181A S185A S570A) ( Figure 1B), just like wild-type Fps1 treated with phosphatase ( Figure 1-figure supplement 2C).
Moreover, a C-terminal fragment of Fps1 containing Ser570, one of the apparent Ypk1 phosphorylation sites delineated in vivo, is phosphorylated by purified Ypk1 in vitro and solely at the Ypk1 site (S570) (Figure 1-figure supplement 3). Furthermore, as for other Ypk1-dependent modifications (Muir et al., 2014), phosphorylation of these same sites in Fps1 in vivo was also TORC2-dependent, because treatment with a TORC2 inhibitor (NVP-BEZ235) (Kliegman et al., 2013) also reduced Fps1 phosphorylation ( Figure 1C). Thus, Fps1 is a bona fide Ypk1 substrate. We documented elsewhere using Phos-tag gel mobility shift that Ypk1 phosphorylation at T662, one of its well-characterized TORC2 sites, is eliminated when cells are subjected to hyperosmotic shock for 10 min (Lee et al., 2012), and the same effect is observed using a specific antibody (Niles et al., 2012) that monitors phosphorylation of Ypk1 at the same site ( Figure 1-figure supplement 4A). Using Ypk1 7A , which also permits facile detection of mobility shifts arising from TORC2-specific phosphorylation (K Leskoske and FM Roelants, unpublished results) (Figure 1-figure supplement 4B), we followed the kinetics of this change. Loss of TORC2-mediated Ypk1 phosphorylation upon hyperosmotic shock occurs very rapidly (within 1 min) and persists for about 15 min ( Figure 1D), but is transient. By 20 min after hyperosmotic shock, TORC2-mediated Ypk1 phosphorylation is again detectable and is nearly back to the pre-stress level by 75 min (Figure 1-figure supplement 5A). Rapid reduction in TORC2-mediated Ypk1 phosphorylation under hypertonic stress was still observed in mutants lacking the Sho1-or Sln1-dependent pathways that converge on Hog1 or Hog1   or Gpt2 3A -3xFLAG (pAX244) were grown to mid-exponential phase and then treated with vehicle (−) or 10 μM 3-MB-PP1 (+) for 90 min. Cells were harvested, extracts prepared, resolved by SDS-PAGE, and blotted as in 'Materials and methods'. (B) Wild-type cells expressing either Fps1-3xFLAG (yGT21) or Fps1 3A -3xFLAG (yGT22) from the FPS1 promoter at the normal chromosomal locus, or ypk1-as ypk2Δ cells expressing either Fps1-3xFLAG (yAM281) or Fps1 3A -3xFLAG (yAM284-A) from the FPS1 promoter at the normal chromosomal locus, were grown to mid-exponential phase and treated as in (A) with vehicle or 3-MB-PP1 for 60 min. Cells were harvested, extracts prepared, resolved by Phos-tag SDS-PAGE, and blotted as in 'Materials and methods'. Unphosphorylated Fps1 (red asterisk). (C) A tor2-as strain (yKL5) expressing Fps1-3xFLAG (pAX274) or Fps1 3A -3xFLAG (pAX275) was grown to mid-exponential phase and then treated with vehicle (−) or 2 μM BEZ-235 (+) for 30 min. Cells were harvested, extracts prepared, resolved and analyzed as in (B). (D) Wild-type (BY4741) or tor2-29 ts (JTY5468) cells expressing Ypk1 7A -myc (pFR252) were grown at 30˚C (left panel) or 26˚C (right panel) to mid-exponential phase, then diluted into fresh YPD in the absence (−) or presence of 1 M sorbitol (final concentration). After the indicated times (1-15 min), culture samples were collected, lysed and the resulting extracts resolved by Phos-tag SDS-PAGE and analyzed by immunoblotting with anti-myc mAb 9E10, as described in 'Materials and methods'. (E) As in (D), except for the genotype (strain) expressing Ypk1 7A -myc (pFR252), which were, aside from the wild-type control, hog1Δ (YJP544), sho1Δ (JTY5540), ssk1Δ (JTY5541), ssk22Δ (JTY5539), ssk2Δ (JTY5538) or pbs2Δ (JTY5537), and the treatment with 1 M sorbitol was for 1 min. (F) Wild-type (BY4741) or otherwise isogenic cna1Δ cna2Δ (JTY5574) cells expressing Ypk1 7A -myc (pFR252) were grown to mid-exponential phase then diluted into fresh YPD in the absence (−) or presence (+) of 1 M sorbitol (final concentration). After 1 min, the cells were collected, lysed and the resulting extracts resolved by Phos-tag SDS-PAGE and analyzed by immunoblotting with anti-myc mAb 9E10, as described in 'Materials and methods'. (G) Wild-type cells expressing either Fps1-3xFLAG (yGT21) or Fps1 3A -3xFLAG (yGT22) from the chromosomal FPS1 locus, were diluted into fresh YPD in the absence (−) or presence of 1 M sorbitol (final concentration) for the indicated times and then extracts of the cells prepared and analyzed as in (B). DOI: 10.7554/eLife.09336.002 The following figure supplements are available for figure 1:  Figure 1E) or CN ( Figure 1F). Thus, loss of TORC2-mediated Ypk1 phosphorylation upon hyperosmotic shock occurs independently of other known response pathways.
Given that Ypk1 phosphorylates Fps1 and that hyperosmotic stress rapidly abrogates TORC2dependent phosphorylation and activation of Ypk1, Ypk1 modification of Fps1 should be prevented under hyperosmotic stress. As expected, Ypk1 phosphorylation of Fps1 is rapidly lost upon hyperosmotic shock ( Figure 1G), yielding a species with mobility indistinguishable from Fps1 3A , remains low for at least 20 min, but returns by 75 min (Figure 1-figure supplement 5B), mirroring the kinetics of loss and return of both TORC2-mediated Ypk1 phosphorylation ( Figure 1D and Figure 1-figure supplement 5A) and Ypk1-dependent phosphorylation of Gpd1 that we observed before (Lee et al., 2012). Thus, hyperosmotic stress dramatically down-modulates Ypk1-mediated phosphorylation of Fps1.

Ypk1 phosphorylation of Fps1 promotes channel opening and glycerol efflux
In its open state, the Fps1 channel permits entry of toxic metalloid, arsenite, which inhibits growth (Thorsen et al., 2006), whereas lack of Fps1 (fps1Δ) or the lack of channel activators (rgc1Δ rgc2Δ) (Beese et al., 2009) or an Fps1 mutant that cannot open because it cannot bind the activators (Fps1 ΔPHD ) (Lee et al., 2013) are arsenite resistant. We found that Fps1 3A was at least as arsenite resistant as any other mutant that abrogates Fps1 function ( Figure 2A). Thus, Fps1 3A acts like a closed channel, suggesting that Ypk1-mediated phosphorylation promotes channel opening. Loss of individual phosphorylation sites led to intermediate levels of arsenite resistance ( Figure 2B). Thus, modification at these sites contributes additively to channel opening.
Others have shown that intracellular glycerol is elevated in fps1Δ cells in the absence of hyperosmotic stress (Tamás et al., 1999). If Fps1 3A favors the closed-channel state, then it should also cause constitutive elevation of intracellular glycerol concentration. Indeed, in the absence of any osmotic perturbation, Fps1 3A mutant cells accumulated ∼twofold as much glycerol as otherwise isogenic FPS1 + strains ( Figure 2C). Consistent with this result, we observed before that loss of Ypk1 (and Ypk2) activity caused an increase in glycerol level compared to control cells (Lee et al., 2012).
Consistent with Ypk1-dependent phosphorylation affecting Fps1 channel function per se, immunoblotting ( Figure 2D) and fluorescence microscopy ( Figure 2E) showed that the steady-state level and localization of Fps1 are unaffected by the presence or absence of these modifications.

Discussion
Aside from further validating the utility of our screen for identifying new Ypk1 substrates (Muir et al., 2014), our current findings demonstrate that TORC2-dependent Ypk1-catalyzed phosphorylation of Fps1 opens this channel and, conversely, that loss of Ypk1-dependent Fps1 phosphorylation upon hypertonic shock is sufficient to close the channel, prevent glycerol efflux, and promote cell survival. In agreement with our observations, in a detailed kinetic analysis of global changes in the S. cerevisiae phosphoproteome upon hyperosmotic stress (Kanshin et al., 2015), it was noted that two sites in Fps1 (S181 and T185), which we showed here are modified by Ypk1, become dephosphorylated.
We previously showed that Gpd1, the rate-limiting enzyme for glycerol production under hyperosmotic conditions (Remize et al., 2001), is negatively regulated by Ypk1 phosphorylation (Lee et al., 2012). Thus, inactivation of TORC2-Ypk1 signaling upon hyperosmotic shock has at least two coordinated consequences that work synergistically to cause glycerol accumulation and promote cell survival, a similar outcome but mechanistically distinct from the processes evoked by Hog1 activation (Figure 4). First, loss of TORC2-Ypk1 signaling alleviates inhibition of Gpd1, which, combined with transcriptional induction of GPD1 by hyperosmotic stress, greatly increases glycerol production. Second, loss of TORC2-Ypk1 signaling closes the Fps1 channel, thereby retaining the glycerol produced.
Presence of two systems (TORC2-Ypk1 and Hog1) might allow cells to adjust optimally to stresses occurring with different intensity, duration, or frequency. Reportedly, Hog1 responds to stresses occurring no more frequently than every 200 s (Hersen et al., 2008;McClean et al., 2009), whereas we found TORC2-Ypk1 signaling responded to hypertonic stress in ≤60 s. Also, the Sln1 and Sho1 sensors that lead to Hog1 activation likely can respond to stimuli that do not affect the TORC2-Ypk1 axis, and vice-versa.
A remaining question is how hyperosmotic stress causes such a rapid and profound reduction in phosphorylation of Ypk1 at its TORC2 sites. This outcome could arise from activation of a phosphatase (other than CN), inhibition of TORC2 catalytic activity, or both. Despite a recent report that Tor2 (the catalytic component of TORC2) interacts physically with Sho1 (Lam et al., 2015), raising the possibility that a Hog1 pathway sensor directly modulates TORC2 activity, we found that hyperosmolarity inactivates TORC2 just as robustly in sho1Δ cells as in wild-type cells. Alternatively, given the role ascribed to the ancillary TORC2 subunits Slm1 and Slm2 (Gaubitz et al., 2015) in delivering Ypk1 to the TORC2 complex (Berchtold et al., 2012;Niles et al., 2012), response to hyperosmotic shock might be mediated by some influence on Slm1 and Slm2. Thus, although the mechanism that abrogates TORC2 phosphorylation of Ypk1 upon hypertonic stress remains to be delineated, this effect and its consequences represent a novel mechanism for sensing and responding to hyperosmolarity.

Materials and methods
Construction of yeast strains and growth conditions S. cerevisiae strains used in this study (Supplementary file 1) were constructed using standard yeast genetic manipulations (Amberg et al., 2005). For all strains constructed, integration of each DNA fragment of interest into the correct genomic locus was assessed using genomic DNA from isolated colonies of corresponding transformants as the template and PCR amplification with an oligonucleotide primer complementary to the integrated DNA and a reverse oligonucleotide primer complementary to chromosomal DNA at least 150 bp away from the integration site, thereby confirming that the DNA fragment was integrated at the correct locus. Finally, the nucleotide sequence of each resulting reaction product was determined to confirm that it had the correct Figure 4. Saccharomyces cerevisiae has two independent sensing systems to rapidly increase intracellular glycerol upon hyperosmotic stress. (A) Hog1 MAPK-mediated response to acute hyperosmotic stress (adapted from Hohmann, 2015). Unstressed condition (top), Hog1 is inactive and glycerol generated as a minor side product of glycolysis under fermentation conditions can escape to the medium through the Fps1 channel maintained in its open state by bound Rgc1 and Rgc2. Upon hyperosmotic shock (bottom), pathways coupled to the Sho1 and Sln1 osmosensors lead to Hog1 activation. Activated Hog1 increases glycolytic flux via phosphorylation of Pkf26 in the cytosol and, on a longer time scale, also enters the nucleus (not depicted) where it transcriptionally upregulates GPD1 (de Nadal et al., 2011;Saito and Posas, 2012), the enzyme rate-limiting for glycerol formation, thereby increasing glycerol production. Activated Hog1 also prevents glycerol efflux by phosphorylating and displacing the Fps1 activators Rgc1 and Rgc2 (Lee et al., 2013). These processes act synergistically to elevate the intracellular glycerol concentration providing an osmolyte to counterbalance the external high osmolarity. (B) Unstressed condition (top), active TORC2-Ypk1 keeps intracellular glycerol level low by inhibition of Gpd1 (Lee et al., 2012) and Figure 4. continued on next page sequence. Yeast cultures were grown in rich medium (YPD; 1% yeast extract, 2% peptone, 2% glucose) or in defined minimal medium (SCD; 0.67% yeast nitrogen base, 2% glucose) supplemented with the appropriate nutrients to permit growth of auxotrophs and/or to select for plasmids.

Plasmids and recombinant DNA methods
All plasmids used in this study (Supplementary file 2) were constructed using standard laboratory methods (Green and Sambrook, 2012) or by Gibson assembly (Gibson et al., 2009) using the Gibson Assembly Master Mix Kit according to the manufacturer's specifications (New England Biolabs, Ipswich, Massachusetts, United States). All constructs generated in this study were confirmed by nucleotide sequence analysis covering all promoter and coding regions in the construct.

Protein purification and in vitro kinase assay
Ypk1 and GST-Fps1(531-0669) proteins were purified as previously described (Muir et al., 2014). Following protein purification, Ypk1 in vitro kinase assays were performed as previously described (Muir et al., 2014).

Measurement of intracellular glycerol accumulation
Measurement of intracellular glycerol was conducted as described (Albertyn et al., 1994a). Briefly, samples (∼40 ml) of exponentially-growing cultures were harvested by centrifugation, washed with 1 ml of medium, recollected and the resulting cell pellets frozen in liquid N 2 and stored at −80˚C prior Upon hyperosmotic shock (bottom), TORC2-dependent phosphorylation of Ypk1 is rapidly down-regulated. In the absence of Ypk1-mediated phosphorylation, inhibition of Gpd1 is alleviated, thereby increasing glycerol production. Concomitantly, loss of Ypk1-mediated phosphorylation closes the Fps1 channel, even in the presence of Rgc1 and Rgc2, thereby promoting glycerol accumulation to counterbalance the external high osmolarity. Schematic depiction of TORC2 based on data from Wullschleger et al. (2005); Liao and Chen (2012); Gaubitz et al. (2015). DOI: 10.7554/eLife.09336.010 to analysis. Each cell pellet was boiled for 10 min in 1 ml of 50 mM Tris-Cl (pH 7.0). This eluate was clarified by centrifugation for 15 min at 13,200 rpm (16,100×g) in a microfuge (Eppendorf 5415D). Glycerol concentration in the resulting supernatant fraction was measured using a commercial enzymic assay kit (Sigma Aldrich) and normalized to the protein concentration of the same initial extract as measured by the Bradford method (Bradford, 1976).

Fluorescence microscopy of Fps1-GFP
An fps1Δ strain was transformed with plasmids expressing wild-type Fps1-GFP or the mutant Fps1-GFP derivatives and grown in selective medium to mid-exponential phase. Samples of the resulting cultures were viewed directly under an epifluorescence microscope (model BH-2; Olympus America, Inc.) using a 100× objective fitted with appropriate band-pass filters (Chroma Technology Corp.). Images were collected using a CoolSNAP MYO charge-coupled device camera (Photometrics, Tucson, Arizona, United States).

Co-immunoprecipitation of Fps1 and Rgc2
Co-immunoprecipitation experiments were performed with minor modifications as previously described (Lee et al., 2013). Cells expressing Fps1-3xFLAG (yAM271-A), Fps1 3A -3xFLAG (yAM272-A) or untagged Fps1 (BY4742) were transformed with empty vector or the same vector expressing Fps1-3xFLAG (pAX302) or Fps1 3A -3xFLAG (pAX303) under control of the MET25 promoter. These transformants were then cotransformed with a plasmid expressing Rgc2-3xHA under control of the MET25 promoter (Lee et al., 2013). Cultures of each were grown to mid-exponential phase in SCD-Ura-Leu. Cultures were then diluted to A 600 nm = 0.2 in 1 l of SCD-Ura-Leu-Met to induce expression of Rgc2-3xHA and Fps1-3xFLAG and grown at 30˚C for 4 hr. Cells were harvested by centrifugation and resuspended in 5 ml of TNE+Triton+NP-40 (50 mM Tris-Cl [pH 7.5], 150 mM NaCl, 4 mM NaVO 4 , 50 mM NaF, 20 mM Na-PPi, 5 mM EDTA, 5 mM EGTA, 0.5% Triton-X100, 1.0% NP-40, 1× cOmplete protease inhibitor [Roche, Pleasanton, California, United States]). The cells were then lysed cryogenically using Mixer Mill MM301 (Retsch GmbH, Haan, Germany). The lysate was thawed on ice and then clarified by centrifugation for 20 min at 10,500 rpm (13,000×g) in the SS34 rotor of a refrigerated centrifuge (Sorvall RC-5B). Protein concentration of the clarified lysate was measured using BCA reagent (Thermo Fisher Scientific, Waltham, Massachusetts, United States) and then Fps1-3xFLAG was immunoprecipitated from a volume of extract containing a total of 10 mg protein using 50 μl of mouse anti-FLAG antibody coupled-agarose resin (Sigma Aldrich) equilibrated in TNE+Triton+NP-40. Binding was allowed to occur for 2 hr at 4˚C. The resin was then washed extensively with TNE+Triton+ NP-40 and the proteins remaining bound were then resolved by SDS-PAGE and analyzed by immunoblotting with appropriate antibodies to detect both Fps1-3xFLAG and Rgc2-3xHA.