Analysis of Mitogen-Activated Protein Kinase Signaling Specificity in Response to Hyperosmotic Stress: Use of an Analog-Sensitive HOG1 Allele

Strains, media, and genetic techniques.Yeast strains (Table 1) were cultivated in standard rich medium (YPD) or in defined synthetic complete medium (49) lacking appropriate nutrients and, where necessary, containing 5-fluoroorotic acid (9) to select for plasmids or integrative transformants. When used, d-sorbitol (Sigma Chemical Co., St. Louis, MO) was present at a 1 M final concentration. DNA-mediated transformation was conducted by the Li acetate procedure (51). All gene replacements and disruptions were confirmed both by phenotypic analysis and via PCR with appropriate gene-specific primers.

Routine recombinant DNA procedures were used for the construction and propagation of plasmid DNA in Escherichia coli strain DH5α. The HOG1 gene was amplified by PCR with primers with XbaI linkers from genomic DNA of strain SO329 and cloned into vector pCR2.1 (Invitrogen Corp., Carlsbad, CA), yielding pPW01. Site-directed mutagenesis of pPW01 was performed with the QuikChange kit (Stratagene Inc., La Jolla, CA) according to the manufacturer's recommendations to generate pPW02 (hog1-as), pPW28 [hog1(K52R)], and pPW51 [hog1(D144A)]. The mutagenic primers used were 5′-CCATTG GAA GAT ATC TAT TTT GTC GCG GAA TTA CAA GGA ACA G (pPW02), 5′-CAG CCA GTT GCC ATT AGA AAA ATC ATG AAA CCT (pPW28), and 5′-GCG TCA TTC ATA GAG CTT TGA AAC CGA GCA A (pPW51). All mutagenesis products were confirmed by DNA sequence analysis. The resulting hog1-as, hog1(K52R), and hog1(D144A) DNAs were excised and integrated into the genome under control of the native HOG1 promoter by selecting for gene transplacement of the hog1::URA3 locus in strain PW014 with 5-fluoroorotic acid. All integrants were confirmed by PCR and DNA sequencing.

The FUS1::lacZ::LEU2 and FUS1::lacZ::leu2 reporter genes have been described previously (34). To generate the integrating TEC1::lacZ::URA3 plasmid, the 5′ untranslated region of TEC1 was amplified by PCR from genomic DNA of strain Σ1278b DNA with primers with XhoI sites. The resulting 1-kb product was inserted into the polylinker in pJM133 (a plasmid carrying CYC1::lacZ::URA3 from the Herskowitz lab plasmid collection), yielding pPW45. The appropriate fragment was excised from pPW45 with HindIII and used for DNA-mediated transformation of appropriate yeast strains.

Integrated versions of HOG1, hog1-as, hog1(K52R), and hog1(D144A) tagged with Aequorea victoria green fluorescent protein (GFP) were created with the GFP-marked modules described by Longtine et al. (26) along with HOG1-specific primers. To install a C-terminal three-hemagglutinin [(HA)₃] tag on Rck2, the HA-marked modules described by Longtine et al. (26) were used along with appropriate RCK2-specific primers. To tag a nucleoporin, Nup1, and the karyopherin Nmd5 with mCherry, a derivative of monomeric Discosoma sp. red fluorescent protein (52), the coding sequence in pRSET-mCherry (gift of R. Y. Tsien, University of California, San Diego) was amplified with primers with PacI and AscI sites, respectively, and the resulting product was inserted into PacI- and AscI-cleaved pKT209 (gift of K. Thorn, Harvard University), thereby replacing the GFP module in this vector (53), yielding pPW58. Nup1-mCherry and Nmd5-mCherry fusions were then generated with the mCherry-containing module in pPW58 and integrated into the desired yeast strains as previously described (53).

To generate pPW86 expressing Ssk2ΔN (28), the sequence of SSK2 encoding Met1173 to the COOH terminus was amplified by PCR with primers containing XbaI and HindIII sites, and the resulting product was inserted into the polylinker in YEp352-GAL (5).

Measurement of reporter gene expression.Expression of β-galactosidase from lacZ-marked reporter genes was assayed with a colorimetric substrate as described previously (56). Values presented (A₄₂₀) represent the average of at least three independent trials, each conducted in triplicate. The enzyme sources were samples of cultures of exponentially growing cells that were collected by centrifugation and resuspended in fresh YPD medium in the absence or presence of 4-amino-1-tert-butyl-3-(1-naphthylmethyl)phenylpyrazolo[3,4-d]pyrimidine (1-NM-PP1) (8) lacking or containing 1 M sorbitol, as indicated, for various times (up to 5 h). To select the inhibitor concentration used, the growth of hog1-as cells and that of HOG1⁺ cells were compared on YPD without and with 1 M sorbitol over a range (5 to 20 μM) of 1-NM-PP1 concentrations. The 1-NM-PP1 was added from a stock dissolved in dimethyl sulfoxide (DMSO), and controls always received the appropriate volume of DMSO alone. We found that 8 μM 1-NM-PP1 was the minimum concentration that completely blocked the growth of hog1-as cells on 1 M sorbitol but not on YPD alone, yet even 20 μM 1-NM-PP1 had no discernible effect on wild-type cells on either medium (data not shown). On this basis, and to ensure complete inhibition, we routinely used a final concentration of 12 μM 1-NM-PP1 in both liquid and solid media.

To determine how fast 1-NM-PP1 acts to block the kinase activity of Hog1-as, we conducted the following comparison on the basis of our observation (see Results) that 1-NM-PP1 blocks the hyperosmotic stress-induced nuclear import of Hog1-as. One sample of an exponentially growing culture of cells expressing Hog1-as-GFP was pretreated for 10 min with 12 μM 1-NM-PP1, collected rapidly by centrifugation, and then exposed to 1 M sorbitol in the continuing presence of 12 μM 1-NM-PP1, and another sample of the same cells was treated simultaneously with 1 M sorbitol and 12 μM 1-NM-PP1. The two samples were then examined immediately and over the course of the next 10 min, by which time Hog1-as-GFP shows pronounced nuclear accumulation in >90% of the cells challenged with 1 M sorbitol in the absence of inhibitor. Whether inhibitor was added 10 min before 1 M sorbitol or added simultaneously, the fraction of the cells in which nuclear import was blocked was the same within sampling error (95% versus 93%). Hence, by this criterion, 1-NM-PP1 blocks kinase activity in <1 min.

Microscopy.Exponentially growing cells were resuspended in fresh YPD medium in the absence or presence of 12 μM (final concentration) I-NM-PP1 for 10 min and then exposed or not exposed to 1 M sorbitol for various times (up to 20 min) without or with 1-NM-PP1, as indicated, before viewing. For routine observation, yeast cells were visualized under the 100× objective of an epifluorescence microscope (BH2; Olympus Microscopes, GmbH, Hamburg, Germany) and images were captured digitally with a charge-coupled device camera and MagnaFire SP software (Olympus). For certain experiments, optical sections of cells were viewed by confocal fluorescence microscopy on a DeltaVision Spectris system (Applied Precision LLC, Issaquah, WA) with a 100× objective lens, and Applied Precision SoftWoRx software was used for image capture, deconvolution, and adjustments to brightness and contrast. To visualize DNA, cells were treated prior to examination with 2 μg/ml (final concentration) 4′,6′-diamidino-2-phenylindole (DAPI). For image clarity in data presentation, cropping and equivalent adjustments to brightness and contrast were made with Photoshop CS (Adobe Systems, Inc., San Jose, CA).

Cell extracts and immunoblotting.To estimate the cellular content of both total Hog1-GFP protein and the fraction of activated (dually phosphorylated) Hog1-GFP (or of the corresponding Hog1-as-GFP, Hog1(K52R)-GFP, and Hog1(D144A)-GFP variants), a sample (A₆₀₀ of 0.5) of exponentially growing cells in 5 ml of YPD (without or with 1 M sorbitol and without or with 12 μM 1-NM-PP1, as indicated for the microscopy experiments) was concentrated in a microcentrifuge. After removal of culture medium by aspiration, cell pellets were resuspended in 1× loading buffer (5% sodium dodecyl sulfate [SDS], 0.1 M Tris-HCl [pH 7.5], 5% glycerol, 0.07 M 2-mercaptoethanol, 0.02 mM bromophenol blue) and boiled for 5 min. After clarification in a Microfuge, a sample of the supernatant solution (10 μl) was resolved by SDS-polyacrylamide gel electrophoresis (PAGE) on an 8% gel and transferred to a sheet of nitrocellulose filter paper. The filter was incubated in Odyssey blocking buffer (Li-Cor Biosciences, Inc., Lincoln, NE) for 1 h at room temperature and then overnight with a mouse monoclonal anti-GFP antibody (Roche Diagnostics, Inc., Basel, Switzerland) and rabbit monoclonal anti-phospho-p38 antibody (Cell Signaling Technology, Inc., Danvers, MA) in the same buffer. Filter-bound antibodies were detected with Alexa Fluor-conjugated secondary anti-mouse and anti-rabbit anti-immunoglobulin G antibodies (Molecular Probes, Inc., Eugene, OR) and visualized with an Odyssey infrared imaging system (Li-Cor). To detect HA epitope-tagged Rck2, mouse monoclonal anti-HA epitope antibody 6E2 (Cell Signaling Technology) was used.

Construction and characterization of a functional hog1-as allele.We generated a conditional allele (dubbed hog1-as) susceptible to inhibition by the cell-permeable adenine analog 1-NM-PP1 on the basis of principles developed by Shokat and coworkers (4, 7, 8, 55). By using sequence alignments and modeling of the back of the ATP-binding pocket of Hog1 on the crystal structure of its closest mammalian ortholog, p38 (21), we determined that a T100A substitution in Hog1 should yield the desired allele (alteration of the equivalent residue to Ala in other protein kinases has sometimes been referred to as an “-as2” allele). The corresponding mutant DNA was inserted into the chromosome in place of the normal locus under the control of its native promoter. Indeed, the hog1-as allele behaved exactly as anticipated. Like HOG1⁺ cells, but unlike hog1Δ mutant cells, hog1-as mutant cells grew under conditions of hyperosmotic stress (1 M sorbitol) (Fig. 1A). Unlike wild-type cells, hog1-as mutant cells failed to grow on 1 M sorbitol and 12 μM 1-NM-PP1, indistinguishable from the inability of hog1Δ mutant cells to grow on 1 M sorbitol (Fig. 1A).

As an alternative and inverse means of demonstrating that 12 μM 1-NM-PP1 efficaciously inhibits hog1-as but otherwise has no deleterious effect on the cells, we took advantage of another known property of the HOG pathway, namely, that constitutive hyperactivation of Hog1 prevents cell growth (50). One way to achieve strong hyperactivation is to overexpress (from the GAL1 promoter) a derivative of the upstream MAPKKK Ssk2 which lacks its N-terminal negative regulatory domain (28). Control cells carrying the GAL expression vector alone grew well on glucose medium, on galactose medium, or on galactose medium containing 12 μM 1-NM-PP1, regardless of whether the cells were HOG1⁺ or hog1Δ mutant or carried the hog1-as allele (Fig. 1B). However, when the cells were HOG1⁺ or contained the hog1-as allele, galactose-induced expression of Ssk2ΔN blocked their growth, whereas hog1Δ mutant cells continued to grow, as expected. Strikingly, on galactose medium in the presence of 12 μM 1-NM-PP1, cells carrying the hog1-as allele grew as well as the control cells or the hog1Δ mutant cells whereas HOG1⁺ cells still grew very poorly (Fig. 1B). The fact that 1-NM-PP1-imposed inhibition of Hog1-as promotes cell growth under conditions where Hog1 activity would be deleterious further confirms that other aspects of cell physiology are not perturbed at the concentration of the compound used.

To determine if inhibition of Hog1-as by 12 μM 1-NM-PP1 was as severe as the loss of catalytic activity caused by point mutations in conserved residues known to be critical for protein kinase activity (23), we used site-directed mutagenesis to generate first a previously described kinase-dead mutant, Hog1(K52R) (20, 46), which alters a Lys required for proper tertiary structure and binding of the γ-phosphate of ATP, and a novel kinase-dead mutant, Hog1(D144A), which eliminates the putative general base catalyst for the phosphotransferase reaction. Indeed, the Hog1(D144A) mutant was unable to grow on 1 M sorbitol, just like hog1Δ mutant cells and hog1-as-carrying cells in the presence of 1-NM-PP1 (Fig. 1A). By contrast, Hog1(K52R) retained residual activity that permitted slow (but readily detectable) growth on 1 M sorbitol (Fig. 1A). This residual growth on 1 M sorbitol was eliminated in the presence of 1-NM-PP1. We presume that the K52R mutation, because of its effects on kinase conformation, makes this already crippled enzyme somewhat susceptible to the inhibitor.

As yet another independent measurement of the catalytic competence of Hog1-as and its inhibition by 1-NM-PP1 in vivo, we examined the Hog1-dependent phosphorylation of a known downstream target, the MAPK-activated protein kinase (MAPKAP kinase) Rck2 (6, 57). In normal cells under isosmotic conditions, the bulk of Rck2-(HA)₃ migrated as a hypophosphorylated form (band 1, Fig. 2A), with a trace of a slightly slower-mobility species (band 2, Fig. 2A). Upon exposure to hyperosmotic stress, no detectable band 1 remained, most of the Rck2-(HA)₃ migrated as band 2, and a third, even slower-mobility form appeared (band 3, Fig. 2A). In hog1Δ mutant cells, the bulk of the Rck2-(HA)₃ remained hypomodified (band 1), even in the presence of 1 M sorbitol. By contrast, in hog1-as cells in the absence of 1-NM-PP1, Rck2-(HA)₃ underwent conversion to bands 2 and 3 under hyperosmotic challenge, just as it did in HOG1⁺ cells, whereas in the presence of inhibitor, Rck2-(HA)₃ modification was completely abrogated (Fig. 2A). It should be noted that, even when bound to 1-NM-PP1, Hog1-as was still an efficient substrate for the upstream kinase Pbs2 (Fig. 2A, bottom).

Collectively, these findings demonstrate first that Hog1-as is a fully functional enzyme in the absence of 1-NM-PP1. Second, the catalytic activity of Hog1-as is acutely and fully inhibited upon addition of 1-NM-PP1 because this treatment rapidly abrogates all known phenotypic and biochemical hallmarks of Hog1 function in response to a hyperosmotic challenge.

Catalytic activity of Hog1 is needed to prevent cross talk.Inappropriate activation of the mating pheromone response pathway does not occur when Hog1 is present but does occur when Hog1 is absent (34). Loss of signaling specificity upon hyperosmotic stress can be measured semiquantitatively by monitoring the induction of a diagnostic transcriptional reporter of pheromone response, FUS1^prom-lacZ. In HOG1⁺ cells, little or no expression of β-galactosidase was observed when the cells were grown on rich medium (Fig. 2B, lane 1), on the same medium containing 1 M sorbitol (Fig. 2B, lane 2), or on l M sorbitol and 1-NM-PP1 (Fig. 2B, lane 3), whereas a hog1Δ mutant displayed robust reporter expression when subjected to a hyperosmotic challenge (Fig. 2B, lanes 2 and 3). The hog1-as-carrying cells showed no induction of the reporter under normal growth conditions or under conditions of hyperosmotic stress, like wild-type cells, but did so to the same extent as hog1Δ mutant cells when challenged with 1 M sorbitol plus 1-NM-PP1 (Fig. 2B, lane 3). This latter result provides yet another independent demonstration that the inhibitor concentration used completely abrogated Hog1-as kinase activity whereas wild-type Hog1 was unaffected. The cells expressing one of the catalytically inactive point mutants, Hog1(D144A), behaved just like hog1Δ mutant cells. By contrast, the other putative kinase-dead allele, Hog1(K52R), seemed to retain partial activity (and thus the ability to prevent the full appearance of cross talk) because the degree of reporter expression upon hyperosmotic challenge was reproducibly and significantly less than that displayed by hog1Δ mutant cells or cells expressing either Hog1(D144A) or the hog1-as allele in the presence of 1-NM-PP1 (Fig. 2B, lane 2). Moreover, the Hog1(K52R) mutant displayed further elevation in reporter expression in the presence of 1-NM-PP1, equivalent to that displayed by hog1Δ mutant cells (Fig. 2B, lane 3). These observations provide independent evidence that Hog1(K52R) retains significant residual activity (as shown in Fig. 1A) and that conformational flexibility caused by the K52R mutation makes the enzyme somewhat susceptible to the inhibitor because this mutant always displayed a greater degree of cross talk in the presence of 1-NM-PP1 than in its absence.

Efficient dephosphorylation of activated Hog1 requires its catalytic function.Upon hyperosmotic stress, Hog1 is activated by phosphorylation on T174 and Y176 in its activation loop by the MAPKK Pbs2. Dual phosphorylation of Hog1 was readily detected by immunoblotting with a commercially available rabbit monoclonal antibody raised against activated (dually phosphorylated) mammalian p38 (22, 54). To permit convenient quantitation of total Hog1 (or its variants) and its visualization in live cells in real time, it was fused to GFP (65) and substituted for the normal chromosomal HOG1 locus by a PCR-based method (26).

Upon exposure to 1 M sorbitol, Hog1-GFP was phosphorylated rapidly, with modification peaking at 10 to 20 min and then declining as cells adapted, as shown previously for epitope-tagged Hog1 (10) (Fig. 3A, left). By 90 min, the amount of activated Hog1 declined to ≤10% of the maximum, as quantified from the relative band intensities measured with an infrared detection system (Odyssey; Li-Cor, Inc.). Likewise, the hyperosmotic stress-induced phosphorylation pattern for Hog1-as-GFP was virtually indistinguishable (Fig. 3B, left). However, under the same conditions but in the presence of 1-NM-PP1, phosphorylation of Hog1-as-GFP was sustained throughout the time course of the experiment (Fig. 3B, right) whereas the kinetic behavior of wild-type Hog1-GFP phosphorylation was unaffected (Fig. 3A, right). In agreement with the conclusion that the kinase activity of Hog1 is required to promote its own dephosphorylation after hyperosmotic-stress-induced activation, both kinase-dead point mutants Hog1(K52R)-GFP and Hog1(D144A)-GFP displayed a pattern of persistent phosphorylation just like Hog1-as-GFP in the presence of inhibitor (Fig. 3C and D).

Nuclear import of activated Hog1 requires its catalytic function.All three GFP fusions in which Hog1 catalytic activity was ablated (two kinase-dead point mutants and 1-NM-PP1-treated Hog1-as) displayed sustained phosphorylation. This finding is consistent with previous work showing that Hog1 function is required for stimulating the activities of two phosphotyrosine-specific phosphoprotein phosphatases, Ptp2 and Ptp3, that dephosphorylate Hog1 (63) and for inducing elevated expression of the corresponding genes (24), especially PTP3 (42). Ptp2 appears to be the phosphatase primarily responsible for deactivation of Hog1 and is located predominantly in the nucleus (31, 68). Hence, the observed inefficient dephosphorylation of catalytically inactive Hog1 could arise from lack of import of the enzyme into the nucleus. To determine if lack of nuclear import is indeed a contributing factor, cells expressing the Hog1-GFP chimeras were examined by fluorescence microscopy.

During growth under isosmotic conditions, Hog1-GFP, the Hog1-as-GFP mutant, and both kinase-dead point mutants Hog1(K52R)-GFP and Hog1(D144A)-GFP were all found primarily in the cytosol (Fig. 4, left). Within 10 min after exposure to 1 M sorbitol, Hog1-GFP and Hog1-as-GFP were found almost exclusively in the nucleus (Fig. 4, right), confirming the rapid hyperosmotic stress-induced translocation observed before (20, 46). However, in contrast to previous claims that nuclear import of Hog1(K52R)-GFP occurs to the same extent and at the same rate as that of wild-type Hog1-GFP in response to exposure to 1 M sorbitol (20, 46), we found that this catalytically compromised variant expressed at its endogenous level did not enter the nucleus efficiently (Fig. 4, right). In agreement with this conclusion, our novel, completely nonfunctional point mutant Hog1(D144A)-GFP displayed even weaker nuclear accumulation. Close inspection indicated that Hog1(D144A)-GFP did not actually enter the nucleus but remained at the nuclear periphery (Fig. 4, right). This impression was confirmed by examination of the cells by confocal microscopy (data not shown; see also below and Fig. 6). Thus, contrary to previous studies, we found that when the catalytic activity of Hog1 is compromised, nuclear import is blocked.

In further confirmation of this conclusion, we found that upon exposure to 1 M sorbitol in the presence of 1-NM-PP1, Hog1-as-GFP did not enter the nucleus in at least 95% of the cells (Fig. 5A), in contrast to the behavior of normal Hog1-GFP (Fig. 5A) and in contrast to the behavior of Hog1-as-GFP in the absence of inhibitor (Fig. 4). The apparent lack of nuclear accumulation of catalytically inactive Hog1 upon hyperosmotic stress could be due to a failure of import or to a lack of retention (or an elevated rate of export) following import. To rule out these latter possibilities, cells expressing either wild-type Hog1-GFP or Hog1-as-GFP were exposed to 1 M sorbitol for 10 min to induce nuclear accumulation, treated with 1-NM-PP1 (which acts within <1 min to fully inhibit Hog1-as; see Materials and Methods), and then examined. Two minutes after addition of the inhibitor (Fig. 5B), and even as late as 30 min after addition of the inhibitor (data not shown), the levels of wild-type Hog1-GFP and Hog1-as-GFP retained in the nucleus were indistinguishable. By later time points (45 to 120 min), the nuclear content of Hog1-GFP, uninhibited Hog1-as-GFP, and inhibited Hog1-as-GFP had declined significantly but equivalently (data not shown), presumably because of their eventual dephosphorylation and export. Thus, once Hog1 has entered the nucleus, inhibition of kinase activity does not promote its rapid efflux. Taken together, our findings demonstrate that the catalytic function of Hog1 is required for its nuclear import after a hyperosmotic challenge.

Lack of importin, Nmd5, and inhibition of Hog1 activity both block nuclear entry at the nuclear periphery.Prior work implicated a specific karyopherin, Nmd5, in nuclear import of Hog1 in response to hyperosmotic stress (20). Correspondingly, we found by conventional epifluorescence microscopy that Hog1-GFP did not accumulate in the nucleus in an nmd5Δ mutant, whereas it did in wild-type (NMD5⁺) cells. However, upon close inspection by confocal microscopy, we discovered that Hog1-GFP decorates the nuclear rim in a distinct pattern of puncta in nmd5Δ mutant cells but not in wild-type cells (Fig. 6A). To confirm that the perinuclear puncta where Hog1-GFP accumulates when cells lacking Nmd5 are challenged with 1 M sorbitol represent the nuclear pore complexes (NPCs), we exploited a nup133Δ mutant which others have amply demonstrated to cause clustering of NPCs at one side of the nuclear envelope (37). We found that in a nup133Δ nmd5Δ double mutant, Hog1-GFP accumulates in the expected clusters and colocalizes there with a bona fide NPC marker, Nup1 (17), tagged with mCherry, a color variant of monomeric Discosoma sp. red fluorescent protein (52) (Fig. 7A). Thus, in the absence of Nmd5, Hog1-GFP still translocates from the cytosol to the nuclear periphery under hyperosmotic stress but remains trapped at the NPC and does not enter the nucleoplasm.

This pattern was strikingly similar to that seen in a high proportion of the cells expressing the kinase-dead point mutant Hog1(D144A)-GFP (Fig. 4), suggesting that lack of the cognate importin and lack of kinase activity both block Hog1 nuclear import at similar stages. In further confirmation of this conclusion, we found that exposure to 1 M sorbitol induced translocation of Nmd5-mCherry to the nuclear envelope only whereas Hog1-as-GFP was imported concomitantly into the nucleus (Fig. 6B). However, in the presence of 1-NM-PP1, inhibited Hog1-as and its import receptor colocalized and accumulated at the nuclear periphery (Fig. 6B). The speckled pattern seen for Hog1-as-GFP in the cells treated with 1-NM-PP1 and challenged with 1 M sorbitol corresponded to NPCs because it was largely congruent with the staining observed for Nup1-mCherry in the same cells (Fig. 7B). Thus, under conditions of hyperosmotic stress, the inhibited enzyme is still able to translocate to the NPC but the catalytic activity of Hog1 is required for its subsequent passage through the channel of the NPC, even when Nmd5 is present.

Hog1 kinase activity is needed continuously for maintenance of the adaptive response to hyperosmotic stress.Our findings with the Hog1-as and Hog1(D144A) alleles demonstrated, as had not been shown before, that Hog1 catalytic activity, and not the mere presence of the Hog1 polypeptide, is required both for mounting an initial response to a hyperosmotic challenge (10) and for preventing inadvertent activation of the pheromone response pathway (34). The hog1-as allele also provided a unique tool with which to explore another feature of HOG pathway regulation, namely, whether Hog1 kinase activity is required for maintenance of pathway signaling specificity, even after relatively long-term adaptive changes have occurred.

Exposure to a high external dissolved-solute concentration causes an ∼1-h delay in cell cycle progression during which the osmotic balance is restored and reassembly of the actin cytoskeleton occurs, permitting resumption of active growth (11, 13, 64). To determine if Hog1 catalytic activity is still needed to prevent cross talk even after growth resumes, cells expressing Hog1-as-GFP and containing the FUS1^prom-lacZ reporter were exposed to 1 M sorbitol in the absence and presence of 1-NM-PP1 and both cell density and reporter gene expression were monitored. As expected, the onset of cross talk was readily detected within the first hour after hyperosmotic stress in the culture that received 1-NM-PP1 initially but not in the control (DMSO-only) culture (Fig. 8A). By 3 h after exposure to 1 M sorbitol, the culture lacking inhibitor had resumed growth, as judged by a significant increase in its A₆₀₀ (data not shown), whereas the culture that received inhibitor had not, also as expected. At that point (Fig. 8A, arrow), 1-NM-PP1 was added to the control culture that had not previously received the inhibitor. Expression of the cross talk reporter commenced almost immediately, and the onset of its expression was only slightly slower than when the Hog1-as mutant was treated with 1 M sorbitol and the analog at the same time. Thus, the catalytic activity of Hog1 is required persistently to prevent inappropriate cross talk between the HOG and pheromone response pathways.

Another measurement of cell adaptation following hyperosmotic stress is that the activated (dually phosphorylated) state of Hog1 wanes over time (Fig. 3). Hence, as an independent measurement of whether cells had achieved the adapted state, the fraction of phospho-Hog1 was followed after exposure to 1 M sorbitol. The fraction of activated (dually phosphorylated) Hog1-as-GFP fell precipitously by 90 min after exposure to 1 M sorbitol, behavior indistinguishable from that of normal Hog1-GFP (Fig. 8B). At that point, 1-NM-PP1 was added to each culture, additional samples were withdrawn at various times thereafter, and the total amounts of Hog1 and phospho-Hog1 in those samples were analyzed. Addition of inhibitor did not significantly increase the fraction of phosphorylated normal Hog1. By contrast, elevation of phospho-Hog1-as was detectable within 10 min after addition of inhibitor and continued to increase over time (Fig. 8B). Thus, maintenance of the adapted (largely dephosphorylated) state of the population of Hog1 molecules requires its persistent catalytic activity, at least at some basal level.

Hog1 catalytic activity is necessary to prevent cross talk between the HOG pathway and filamentous-growth-signaling pathways.Connections between the HOG pathway and the filamentous-growth-signaling pathway have been noted. For example, in diploid strains of the Σ1278b lineage (which display robust filamentation in response to nitrogen deprivation), cells lacking the Sho1 osmosensor (34) or the mucin-like protein (Msb2) with which it interacts (15) display markedly reduced pseudohyphal growth. Conversely, it has been reported that, when subjected to hyperosmotic stress, hog1Δ (or pbs2Δ) mutant cells express genes normally expressed only under conditions that promote filamentous growth and require components of the filamentous-growth response pathway (e.g., Ste7 and Kss1) to do so (16). Indeed, in our experiments, haploid hog1Δ and pbs2Δ mutants of non-Σ1278b lineage displayed robust invasive growth on 1 M sorbitol but HOG1⁺ cells did not (P. J. Westfall, unpublished results). Thus, as observed for the pheromone response, Hog1 seems to be required to prevent inadvertent activation of the filamentous-growth pathway by hyperosmotic stress.

The hog1-as allele allowed us to assess whether this cross talk behavior was manifested because of alterations of cell physiology that arise in hog1Δ mutants because of the chronic absence of Hog1 or is also observed in otherwise wild-type cells upon acute inhibition of Hog1 catalytic activity. To assess the degree of this kind of cross talk, we attempted to use a well-accepted reporter of the pseudohyphal growth pathway, FRE(Ty1)-lacZ (27). However, as we have observed before (14), this reporter displays a rather high basal level of expression in the Σ1278b background. Therefore, we sought a different reporter that would give very low basal expression when introduced into the Σ1278b background. The TEC1 gene is specifically induced under filamentous-growth conditions and has been shown to be a reliable and specific indicator of this signaling pathway (15, 16, 25). Hence, we constructed a TEC1^prom-lacZ::URA3 fusion which was targeted for integration in single copy just upstream of the resident TEC1 gene at its normal chromosomal locus. In this way, Σ1278b cells containing this reporter remained TEC1⁺ and thus capable of mounting a normal filamentous-growth response. We then prepared an isogenic set of haploid and MATa/MATα diploid cells of the Σ1278b lineage containing this reporter and either wild-type HOG1 or the four hog1 mutants we used throughout this study. For all intents and purposes, the haploids and diploids behaved essentially identically (Fig. 9A and B). Wild-type (HOG1⁺) cells did not express TEC1^prom-lacZ when grown in YPD before or after exposure to 1 M sorbitol. In contrast, in hog1Δ mutant cells, basal expression of the reporter was somewhat elevated on YPD alone and its expression was robustly induced (∼8- to 10-fold) in medium containing 1 M sorbitol (Fig. 8). Likewise, and in agreement with the other evidence that the Hog1(D144A) allele is a complete loss-of-function mutation, it behaved like hog1Δ mutant cells. Correspondingly, and in keeping with the other evidence that Hog1(K52R) retains partial function, the degree of cross talk activation of the TEC1^prom-lacZ reporter was reduced reproducibly. Finally, Hog1-as allele-carrying cells behaved like wild-type (HOG1⁺) cells in the absence of 1-NM-PP1 but displayed readily detectable expression of the cross talk reporter when exposed to 1 M sorbitol in the presence of 1-NM-PP1. We noted, however, that, at least by the criterion of the results obtained in this assay format, 1-NM-PP1 was not causing complete inhibition of Hog1-as. We presume that 1-NM-PP1 was less efficacious because cells of the Σ1278b lineage are less permeable to the analog, have larger competing intracellular adenine pools, or have a different spectrum of ABC transporters that extrude the compound more effectively (or some combination of the three), compared to the other strains we used (all derivatives of EG123).