The Saccharomyces cerevisiae Sln1 protein is a ‘two‐component’ regulator involved in osmotolerance. Two‐component regulators are a family of signal‐transduction molecules with histidine kinase activity common in prokaryotes and recently identified in eukaryotes. Phosphorylation of Sln1p inhibits the HOG1 MAP kinase osmosensing pathway via a phosphorelay mechanism including Ypd1p and the response regulator, Ssk1p. SLN1 also activates an MCM1‐dependent reporter gene, P‐lacZ, but this function is independent of Ssk1p. We present genetic and biochemical evidence that Skn7p is the response regulator for this alternative Sln1p signaling pathway. Thus, the yeast Sln1 phosphorelay is actually more complex than appreciated previously; the Sln1 kinase and Ypd1 phosphorelay intermediate regulate the activity of two distinct response regulators, Ssk1p and Skn7p. The established role of Skn7p in oxidative stress is independent of the conserved receiver domain aspartate, D427. In contrast, we show that Sln1p activation of Skn7p requires phosphorylation of D427. The expression of TRX2, previously shown to exhibit Skn7p‐dependent oxidative‐stress activation, is also regulated by the SLN1 phosphorelay functions of Skn7p. The identification of genes responsive to both classes of Skn7p function suggests a central role for Skn7p and the SLN1‐SKN7 pathway in integrating and coordinating cellular response to various types of environmental stress.
Saccharomyces cerevisiae Sln1p is a histidine kinase belonging to the family of ‘two‐component’ regulators which play a prominent role in the response of prokaryotic organisms to the extracellular environment (Ota and Varshavsky, 1993; Alex and Simon, 1994). The term ‘two‐component’ refers to the prototype in which component one is a sensor/kinase, consisting of an extracellular domain and a histidine autokinase activity (transmitter), and component two is a response regulator, consisting of a receiver domain with a conserved aspartate residue. In many cases the response regulator also carries a domain (e.g. DNA‐binding domain) responsible for pathway output. Sln1p itself contains both kinase and receiver domains and is thus classified as a hybrid kinase.
One function of Sln1p is to regulate the osmotic response MAP kinase pathway (Maeda et al., 1994, 1995). It does so via a phosphorelay system involving Ypd1p and the response regulator, Ssk1p (Posas et al., 1996). The initial event in SLN1 signaling is autophosphorylation of histidine 576 (Posas et al., 1996). This phosphate is transferred to a conserved aspartate (D1144) in the Sln1p‐associated receiver domain. Sln1p D‐P is subsequently transferred to a histidine on the small phosphorelay molecule, Ypd1p. Finally Ypd1p‐P is transferred to an aspartate in the receiver domain of the response regulator, Ssk1p. Unphosphorylated Ssk1p interacts physically with Ssk2p and Ssk22p (Posas and Saito, 1998), two MEK kinases whose activation ultimately leads to Hog1p phosphorylation and the changes in gene expression required for survival in a hyperosmotic environment (Brewster et al., 1993). Under normal conditions, phosphorylation of Sln1p and the downstream (effector) proteins, Ypd1p and Ssk1p, prevent activation of the HOG1 MAP kinase pathway. In contrast, hyperosmotic conditions promote dephosphorylation of Sln1p and consequently activate the HOG1 MAP kinase pathway (Posas et al., 1996).
In addition to regulating the HOG1 osmotic‐response pathway, a separate role for SLN1 was revealed using the MCM1‐dependent reporter gene, P‐lacZ. Deletion of the SLN1 gene reduces reporter gene expression 10‐fold, while activating alleles of SLN1 (sln1*) increase reporter gene expression between three‐ and fivefold (Yu et al., 1995; Fassler et al., 1997; Tao et al., 1998). SLN1 regulation of the HOG1 and P‐lacZ pathways appears to be reciprocal since sln1* mutations not only increase the activity of the P‐lacZ reporter but also decrease Hog1p phosphorylation and enhance cellular sensitivity to hyperosmotic conditions (Fassler et al., 1997). Neither Hog1p nor Ssk1p is directly involved in SLN1‐mediated P‐lacZ regulation since HOG1 and SSK1 deletions do not affect reporter activity (Yu et al., 1995; Fassler et al., 1997). These observations, together with the mapping of sln1* mutations to residues in SLN1 that possibly affect rates of phosphotransfer (Fassler et al., 1997), led to a model for Sln1p function in which the HOG1 MAP kinase cascade represents one branch of a bifurcated pathway and the P‐lacZ reporter represents a second branch. The HOG1 branch is activated by the accumulation of unphosphorylated Sln1p while the P‐lacZ branch is postulated to be activated by the accumulation of phosphorylated Sln1p. The effect of sln1* mutations was hypothesized to shift the equilibrium between unphosphorylated (Sln1p‐0) and phosphorylated Sln1p (Sln1p‐P) to favor Sln1p‐P. In the context of this hypothesis, Sln1p activation of the P‐lacZ reporter was presumed to involve the transfer of phosphate from Sln1p to intermediate signaling molecules. Proteins with homology to two‐component regulators, including the phosphorelay protein, Ypd1p required for SLN1‐HOG1 signaling (Posas et al., 1996) and the response regulator homolog, Skn7p (Brown et al., 1993; Krems et al., 1996), were obvious candidates.
Although SKN7 was previously ruled out as a potential intermediate in the SLN1‐HOG1 pathway (Brown et al., 1994), our current studies indicate that SKN7 plays an important role in SLN1 signaling. Specifically, SKN7 was uncovered in our genetic screen to define genes which, when overexpressed, result in activation of the P‐lacZ reporter gene, as seen in sln1* mutants (Yu et al., 1995). Involvement of the Skn7p response regulator in a SLN1‐regulated phosphotransfer pathway is supported by two additional lines of evidence presented in this paper: (i) deletion of the SKN7 gene or mutation of the conserved receiver domain aspartate blocked the activating phenotype of sln1* mutations; and (ii) in vitro phosphotransfer experiments demonstrating Ypd1p‐dependent phosphorylation of Skn7p by Sln1 kinase using recombinant proteins. The existence of a Sln1‐Skn7 pathway was also the recent conclusion of Ketela et al. (1998) to explain genetic data which suggest that Skn7p functions downstream of Sln1p.
Skn7 was identified previously in two high‐copy genetic screens. One screen was for high‐copy suppressors of a cell‐wall defect caused by deletion of the KRE9 gene (Brown et al., 1993). The second screen was for genes that could bypass the usual requirement for the Swi6 protein in directing the G1‐specific cell‐cycle expression of a reporter gene normally driven by a protein complex including the Mbp1 and Swi6 proteins (Morgan et al., 1995). In addition, mutations in SKN7 were identified in a screen for mutants hypersensitive to oxidative stress (Krems et al., 1996).
The results in this paper establish that the Sln1‐regulated osmotic‐stress pathway is a bifurcated pathway in which Sln1 kinase signals via Ypd1p to two distinct response regulators, Ssk1p and Skn7p. Several lines of evidence indicate that activation of Skn7p by Sln1p is distinct from the previously described activation of Skn7p by oxidative stress (Morgan et al., 1997). Although the two pathways are distinct, we have found at least one gene, TRX2 that is responsive to both pathways. We therefore conclude that the SLN1‐SKN7 branch of the osmotic‐response pathway and the SKN7 oxidative‐stress pathway may converge on a common set of target genes. The participation of Skn7p in both the osmotic and oxidative‐stress pathways may coordinate patterns of gene expression in response to distinct types of environmental stress.
Multicopy genes stimulate expression of the MCM1‐dependent reporter gene, P‐lacZ
To identify potential signaling intermediates that function between Sln1p and Mcm1p, we introduced a high‐copy yeast genomic library into wild‐type strain JF1493 which carries a single integrated copy of the P‐lacZ reporter gene (Yu et al., 1995) in which a palindromic Mcm1‐binding site was substituted for the UAS of a CYC1‐lacZ fusion gene (see Materials and methods). The starting strain exhibited a white phenotype when plated on media containing 150 μg/ml X‐gal due to low expression of the reporter. Eleven thousand transformants were replica plated onto X‐gal media and colonies with increased blue color were selected for testing by liquid β‐galactosidase assays. Plasmids isolated from the best candidates were retransformed into the starting strain to verify that the X‐gal phenotype was plasmid associated.
Eight clones were examined by sequence analysis of the insert ends and found to represent six regions of the genome (Table I). Subcloning and deletion experiments have identified the genes responsible for P‐lacZ activation in several of the high‐copy clones, including BCK2, a gene encoding a serine/threonine‐rich protein with a role in the regulation of G1 cyclin genes (Lee et al., 1993; Epstein and Cross, 1994), MSS11, a gene involved in starch regulation (Webber et al., 1997), RCK2, encoding a Ca2+/calmodulin‐kinase‐like protein kinase (Dahlkvist and Sunnerhagen, 1994; Melcher and Thorner, 1996) and SKN7. Since the Skn7 protein includes a receiver domain homologous to those of two‐component response regulators (Brown et al., 1993), we sought to further examine its involvement in SLN1‐signal transduction. Both library plasmid, p49‐1 and an independent SKN7‐containing plasmid (gift of H.Bussey) were introduced into reporter strain, JF1493. Each caused a fivefold increase in P‐lacZ activity when compared with the same strain transformed with the corresponding empty vector (data not shown).
skn7 deletion suppresses the effect of sln1‐activating mutations
We reasoned that genes actually involved in SLN1 signaling should also exhibit a change in P‐lacZ activity when deleted. To examine the requirement for SKN7 in SLN1 signaling we deleted the SKN7 gene in strains bearing the 2 μ P‐lacZ reporter and containing either wild‐type or activated alleles of SLN1 (sln1*). Strains containing activated sln1 alleles (formerly designated nrp2) exhibit three to five times the P‐lacZ reporter activity of strains with the wild‐type SLN1 allele (Yu et al., 1995; Fassler et al., 1997; Tao et al., 1998). Deletion of SKN7 abolished the increase usually seen in sln1* mutants (Table II). To verify that the reduction in reporter activity was in fact due to the absence of SKN7, a CEN‐based SKN7 plasmid was introduced into each skn7‐Δ strain. Introduction of the SKN7 plasmid into a sln1* strain increased reporter activity 14‐fold, demonstrating that the reduction in reporter activity in the sln1* skn7‐Δ strain was due to the absence of SKN7. We postulate that the absence of an effect of the skn7‐Δ on P‐lacZ activity in the SLN1 background reflects the existence of SKN7‐independent activators of the P‐lacZ reporter gene. The change in the sln1* phenotype due to loss of SKN7 indicates that Skn7p may play an important role in signaling to P‐lacZ and that SKN7 is likely to function downstream of SLN1.
Interestingly, levels of reporter activity were significantly higher in strains carrying the centromere‐based SKN7 plasmid than in those carrying a single SKN7 gene (compare JF1787 with JF1331 and JF1795 with JF1359). Thus, even a small increase in SKN7 dosage increased activity of the P‐lacZ reporter. The threefold increase in P‐lacZ reporter expression in a sln1* strain carrying the CEN‐SKN7 plasmid versus the sln1* SKN7 strain suggests that Skn7p may normally be limiting in the sln1* mutant.
Skn7p D427 is required for SLN1 signaling to the P‐lacZ reporter
The Skn7 protein contains a typical receiver domain characterized by a conserved aspartate. Previous studies have shown that some, but not all, of the functions of Skn7p are dependent on this aspartate (Brown et al., 1994; Morgan et al., 1995, 1997). To examine the role of D427 in sln1* activation of the P‐lacZ reporter, we introduced a plasmid‐borne skn7 D427N mutant allele into the sln1* skn7‐Δ strain and measured β‐galactosidase activity from the P‐lacZ reporter gene. The skn7‐Δ P‐lacZ phenotype was unchanged in the presence of the skn7 D427N allele (Table III). Thus P‐lacZ activation requires residue D427 of Skn7p. The role of D427 in the SKN7 overexpression phenotype was also investigated. Activation of the P‐lacZ reporter by elevated SKN7 dosage also required the receiver domain aspartate (data not shown). Hence Sln1p activation requires the conserved aspartate in the Skn7p receiver domain. The D dependence of the deletion and overexpression phenotypes is consistent with a role for Skn7p as a response regulator protein in a two‐component signaling cascade involving the Sln1p kinase and Skn7p response regulator. To further investigate the importance of D427, a skn7 D427E mutant was introduced into SLN1 skn7‐Δ and sln1* skn7‐Δ strains. A change from aspartic acid to glutamic acid was shown previously to force the receiver domain of Skn7p into an active conformation (Brown et al., 1994). The presence of the skn7 D427E allele elevated P‐lacZ levels 3.7‐fold in the SLN1+ skn7‐Δ strain and this level was not stimulated further by the presence of the sln1* mutation (Table III). Thus, the D427E mutation partially mimics the effect of a sln1* mutation.
YPD1 is required for SLN1 signaling to P‐lacZ
In the SLN1‐HOG1 pathway, Sln1p histidine phosphate is transferred first to a conserved aspartate in the Sln1p‐associated receiver domain, then to a histidine in Ypd1p and finally to a conserved aspartate in the receiver domain of the response regulator, Ssk1p (Posas et al., 1996). If the SLN1 pathway that leads to P‐lacZ activation is organized analogously, phosphate from the Sln1p‐associated receiver domain would be transferred to the receiver domain of the response regulator, Skn7p via a phosphorelay intermediate. To evaluate the possibility that Ypd1p, the phosphorelay intermediate in the SLN1‐HOG1 pathway, might also be involved in the SLN1‐SKN7 signal transmission, we constructed a ypd1 deletion mutant. Like SLN1, YPD1 negatively regulates the activity of the HOG1 pathway by maintaining Ssk1p in the phosphorylated state. Since the presence of SSK1 (in the unphosphorylated form) is required to activate the MEK kinases in the HOG1 pathway, SSK1 deletion suppresses the inviability of sln1 and ypd1 deletions (Maeda et al., 1994; Posas et al., 1996). We have shown previously that the ssk1 mutation has no significant effect on P‐lacZ expression on its own (Fassler et al., 1997), and hence we were able to evaluate the effect of a YPD1 deletion on P‐lacZ expression in SLN1 and sln1* strains deleted for SSK1. The results (Table IV) show that the YPD1 deletion prevents the activation conferred by the sln1* mutation. Thus, the Ypd1p phosphorelay intermediate is shared between the HOG1 and P‐lacZ pathways.
Phosphotransfer from Sln1p to Ypd1p to Skn7p
Our genetic analysis suggested that activation of the P‐lacZ reporter depends on a phosphorelay involving Sln1p, Ypd1p and Skn7p. To gain biochemical support for this hypothesis we constructed and separately purified recombinant glutathione S‐transferase (GST)‐Sln1p kinase domain and GST‐Sln1p‐associated receiver domain, as well as GST‐Ypd1p and GST‐Skn7p with which to reconstitute the proposed phosphorelay. We have shown previously that the Sln1 kinase domain and Sln1 receiver domain can complement a sln1‐Δ mutation when expressed as separate molecules in the same yeast cell (Fassler et al., 1997). Sln1p kinase was first phosphorylated by incubation with [γ‐32P]ATP and then mixed with Sln1p receiver, Ypd1p and Skn7p. In mixtures containing all components, the labeled Sln1p phosphate was distributed rapidly between the Sln1p kinase, Sln1p receiver, Ypd1p and Skn7p (Figure 1, lane 4). The phosphorelay required the presence of all four components. Omission of either the Sln‐associated receiver (Figure 1, lane 6) or GST‐Ypd1p (lane 5) blocked phosphorylation of the Skn7p response regulator. Finally, substitution of the presumed phosphorylated aspartate of Skn7p with asparagine (D427N) completely abolished GST‐Skn7p phosphorylation (Figure 1, lane 7). Taken together, the genetic and biochemical data strongly suggest that Sln1p regulation of P‐lacZ reporter activity involves the histidine autokinase activity of Sln1p and phosphorelay to Ypd1p and Skn7p.
Sln1p, Ypd1p and D427 of Skn7p are dispensable for the oxidative‐stress response
In contrast to the proposed dependence of Skn7p phosphorylation for P‐lacZ regulation, the function of Skn7p in the oxidative‐stress response is independent of D427 since both peroxide resistance and inducibility of the oxidative‐stress gene, TRX2, are normal in a skn7 D427N mutant (Krems et al., 1996; Morgan et al., 1997). Therefore, we did not anticipate that the P‐lacZ reporter would respond to oxidative stress. Unexpectedly, however, control experiments in which the activity of the P‐lacZ reporter was assessed after exposure of cells to t‐butyl hydrogen peroxide revealed that expression of this reporter is activated by oxidative stress. The observation that the P‐lacZ reporter could be activated both by oxidative stress and by Sln1p prompted us to consider the possibility that in the context of this reporter gene the Sln1p and oxidative‐stress responses might be related. However, tests of the D‐dependence of the oxidative‐stress response of the P‐lacZ reporter ruled out this possibility. The oxidative‐stress responsiveness of the P‐lacZ reporter was independent of D427 (Table V, rows 1‐3). Hence the P‐lacZ reporter is regulated by both the D‐dependent and the D‐independent aspects of Skn7p function.
The lack of a requirement for D427 in oxidative stress implies that Sln1p and Ypd1p are unlikely to be involved in this pathway. Nevertheless, since the osmotic‐ and oxidative‐stress signals both apparently involve Skn7p, the possibility that certain Sln1p and Ypd1p activities might be important in the oxidative‐stress response was considered. To evaluate possible roles for Sln1p and Ypd1p in oxidative stress, the oxidative‐stress activation of the P‐lacZ reporter in double sln1 ssk1 and ypd1 ssk1 mutants was tested. In this experiment, the P‐lacZ reporter in the wild‐type strain was induced 5‐fold by treatment with t‐butyl hydrogen peroxide. This induction was eliminated completely in the skn7‐Δ strain. This indicates that the oxidative‐stress response of the P‐lacZ reporter is entirely Skn7p dependent. In contrast, neither the sln1 ssk1 nor the ypd1 ssk1 double mutant showed any defect in oxidative‐stress activation relative to the wild‐type strain (Table V). These data support the conclusion that Skn7p is a multifunctional protein with distinct roles in oxidative‐stress sensing and in SLN1 signaling.
SKN7‐dependent regulation of TRX2 by Sln1p
Earlier studies suggesting a role for the Mcm1p‐binding site in the P‐lacZ vector in sln1* activation were based on a comparison between the P‐lacZ reporter, pGY48 (Yu et al., 1995) which is a derivative of pLG670Z (Guarente and Ptashne, 1981) and reporters harboring alternate UAS elements (Yu et al., 1995). For example, in contrast to the 5‐fold effect of sln1* mutations on the P‐lacZ reporter, sln1* mutations had a negligible effect on a reporter containing the natural CYC1 UAS. To test further the role of the Mcm1‐binding site in sln1* activation, a pair of reporters was constructed consisting of four perfect Mcm1p‐binding sites or four mutant binding sites cloned in tandem into the unique XhoI site of pGY79 (Materials and methods). The binding‐site mutation TTTCCTAATTAGGAAA→TTTCCTAATTAATAAA is predicted to reduce DNA binding by Mcm1p by ∼90% relative to the wild‐type site (Acton et al., 1997). As expected, basal activity of pGY79 was increased significantly by the presence of the four Mcm1‐binding sites (Table VI) but not by the presence of four mutated Mcm1‐binding sites. Furthermore, activity of the 4x P‐lacZ reporter was increased by the sln1* mutation. Unexpectedly, however, the 4x mutP‐lacZ reporter was also stimulated by the sln1* mutation to an extent indistinguishable from the 4x P‐lacZ reporter (Table VI). These results introduce the possibility that the site of activation by sln1* may be distinct from the binding site for Mcm1p. However, an alternative explanation, that sln1* activation of the 4x mutP‐lacZ reporter might be due to residual binding of Mcm1 to the mutant P site, can not be ruled out.
Although the artificial reporter system allowed us to detect the effects of sln1* mutations and to analyze the SLN1‐SKN7 pathway, the possibility that the Mcm1 site may not be critical for sln1* activation led us to search for bona fide genes whose expression is increased in sln1* mutants. Since our analysis of the P‐lacZ reporter implied the presence of both D‐dependent (osmotic) and D‐independent (oxidative) regulatory elements, we tested whether known SKN7‐dependent oxidative‐stress responsive genes might also exhibit D‐dependent regulation. TRX2 appeared to be a good candidate, since a role for SKN7 in the oxidative‐stress induction of the TRX2 gene has already been demonstrated (Morgan et al., 1997) and electrophoretic mobility‐shift analysis revealed the presence of the Skn7 protein in a specific complex that forms at position −140 to −160 relative to the ATG in the TRX2 promoter (Morgan et al., 1997). We, therefore, tested whether TRX2 was responsive to SLN1 by measuring the activity of a TRX2‐lacZ reporter in wild‐type and sln1* strains. The sln1* mutation increased TRX2‐lacZ reporter activity 2.5‐fold, suggesting that SLN1 does in fact regulate an established oxidative‐stress response gene (Table VII). In addition, SKN7 is required for sln1* activation of TRX2. The 2.5‐fold increase in TRX2‐lacZ activity seen in sln1* mutants is eliminated in strains lacking SKN7. Likewise, substitution of the Skn7 D427N mutant protein for wild‐type Skn7p also eliminated the sln1* effect (Table VII). Interestingly, deletion of SKN7 caused a 10‐fold reduction in TRX2 expression in a SLN1 strain, thus suggesting an additional role (discussed below) for SKN7 in TRX2 expression.
A SLN1‐YPD1‐SKN7 phosphorelay
The activity of the P‐lacZ reporter gene was used as a tool to identify intermediates in a new SLN1‐signaling pathway. We found previously that a certain category of SLN1 mutations (sln1*) increased the activity of the P‐lacZ reporter (Yu et al., 1995). In this study we sought additional genes whose overexpression would cause an increase in the activity of this reporter. This screen led to the re‐identification of the SKN7 gene which encodes a protein containing a domain related to the receiver domain of bacterial two‐component proteins (Brown et al., 1993). Our genetic and biochemical analysis supports the idea that both Skn7 and the Ypd1 phosphorelay proteins operate downstream of Sln1p in a signaling pathway leading to activation of the P‐lacZ reporter and genes such as TRX2. Our studies also demonstrate that the conserved aspartate in the receiver domain of Skn7p is essential for this signaling pathway. These genetic data are supported by in vitro experiments demonstrating Ypd1p‐dependent phosphotransfer from Sln1p to Skn7p, and are consistent with a model in which Sln1p phosphorylation status regulates the activities of two distinct response regulators, Ssk1p and Skn7p. These studies provide direct experimental evidence supporting the model suggested previously by Morgan et al. (1997) to explain the apparently distinct roles of Skn7p in D‐dependent processes (regulation of cell wall and G1 cyclin gene expression) and the D‐independent oxidative‐stress response, and by Ketala et al. (1998) to explain genetic data showing that Skn7p functions downstream of Sln1p and exhibits genetic interactions with the HOG pathway.
Examples of pathways that bifurcate at the level of the response regulator protein are relatively rare. One organizational counterpart is represented by the bacterial chemotaxis system in which the kinase, CheA transduces signals to two response regulators, CheY and CheB. Phosphorylated CheY interacts with the flagellar switch to cause tumbling behavior while CheB‐P is a methylesterase that adjusts the signaling state of the receptor (Amsler and Matsumura, 1995). In contrast to the chemotaxis system in which both response regulators directly affect protein activities, the pair of Escherichia coli response regulators, NarL and NarP, function by altering patterns of gene expression (Stewart and Rabin, 1995). In the Nar system for the control of nitrate and nitrite metabolism under anaerobic conditions, the circuitry is further complicated by the presence of two sensor kinases. The physiological rationale for a system with two sensor kinases is possibly related to the differential phosphotransfer (kinase and phospho‐response regulator phosphatase) activities inherent in the two sensor proteins. Activation of the dual Nar response regulators, each of which also contains a DNA‐binding domain, results in more complex patterns of nitrate‐ and nitrite‐regulated gene expression than would be possible in a system with a single response regulator.
On the one hand, in the SLN1 signaling system, Ssk1p more closely resembles the CheY/B type of response regulator in that its phosphorylation state controls its interaction with and hence the activity of the MEK kinase, Ssk2p (Posas and Saito, 1998). Skn7p, on the other hand, possesses a (HSF‐like) domain which may mediate DNA interaction and the capability to activate transcription (Brown et al., 1994; Morgan et al., 1995). Whether these Skn7p activities are relevant to SLN1‐YPD1‐SKN7 signaling is currently under investigation. We postulate that the rationale for the complex organization of SLN1 signaling pathways is to coordinate metabolic and cell‐cycle adjustments to changes in environmental osmolarity.
The conserved aspartate of the Skn7p receiver is required for SLN1 signaling but not oxidative‐stress activation of the P‐lacZ reporter
The Skn7 protein has several recognizable domains in addition to the receiver domain. As first noted by Brown et al. (1993), its N‐terminus (aa 87‐150) is characterized by its similarity to the DNA‐binding domain of members of the eukaryotic heat‐shock transcription factor (HSF). Adjacent to the potential HSF DNA‐binding domain, Morgan et al. (1997) noted a coiled‐coil domain also found in HSF, probably involved in protein‐protein interactions and required for many aspects of Skn7 function (Alberts et al., 1998). The C‐terminus of the protein (497‐622 end) is glutamine rich. Whether these domains function together with or independent of the receiver is unclear. Although the presence of a receiver domain suggests that the activity of the Skn7 protein may be regulated by aspartyl phosphorylation, mutation of the critical aspartate to asparagine (D427N) abolishes some but not all Skn7p functions. For example, the D427N mutation eliminates suppression of both the kre9‐Δ cell‐wall defect and the swi6‐Δ cell‐cycle gene expression defect (Brown et al., 1994; Morgan et al., 1995), but does not confer sensitivity to oxidative stress (Morgan et al., 1997). These observations indicate that the various functions of Skn7p in the cell may involve distinct mechanisms.
Our studies on the effects of the skn7 D427N mutation on P‐lacZ expression are consistent with the conclusion that Skn7p has distinct roles to play in the cell. As expected from the previous observation that oxidative‐stress hypersensitivity of skn7 deletion mutants could be rescued by a SKN7 allele in which the conserved aspartate 427 had been mutated to asparagine (Morgan et al., 1997), P‐lacZ expression continued to be oxidative‐stress inducible in a skn7 D427N mutant (Table V). In contrast, the receiver domain aspartate in Skn7p is required for SLN1‐ and YPD1‐dependent signal transmission. Multiple copies of the gene increase expression of the downstream P‐lacZ reporter in a D‐dependent manner. Likewise, sln1* activation is eliminated in strains containing a D427N mutant allele of the SKN7 gene. These results support our hypothesis based on earlier data that phosphorylated Sln1p signals to target genes via a phosphotransfer mechanism (Fassler et al., 1997).
We further show that mutation of the conserved receiver domain aspartate to glutamate enhanced D‐dependent Skn7p activities, mimicking somewhat the effects of the sln1* mutation. The hyperactivity of Skn7p D/E was first noticed by Brown et al. (1994). Based on the absence of in vitro phosphorylation of D/E mutant alleles of several bacterial response regulators (Bourret et al., 1990; Green et al., 1991; Pazour et al., 1992), the hyperactivity of this allele was suggested to reflect constitutive phosphorylation‐independent activation (Brown et al., 1994). The observation that the sln1* mutation which causes accumulation of phosphorylated Sln1p did not further stimulate the elevated P‐lacZ activity in the skn7 D427E mutant is consistent with this hypothesis.
The role of SKN7 in oxidative stress is independent of SLN1 and YPD1
Oxidative‐stress induction of the P‐lacZ reporter gene did not require SLN1 or YPD1. The nature of the Sln1p‐independent effect of oxidative stress on Skn7p is unknown. However, previous studies showing that Skn7p phosphorylation is not limited to the receiver domain aspartate (Brown et al., 1994) are consistent with the possibility that the oxidative‐stress signal may affect the (serine/threonine) phosphorylation status of Skn7p.
Regulation of TRX2 by Skn7p and by the SLN1‐SKN7 pathway
The TRX2 gene was considered a candidate for regulation by the SLN1‐SKN7 pathway based on previous work establishing the presence of a Skn7p‐containing complex associated with sequences ∼150 bp upstream of the TRX2 ATG (Morgan et al., 1997). The 2.5‐fold increase in TRX2 expression due to sln1* mutation confirms that TRX2 is one target of the SLN1‐SKN7 pathway. In composite, three distinct Skn7p functions have been revealed with respect to TRX2 gene regulation. First, as shown by Morgan et al. (1997), Skn7p is involved in oxidative‐stress responsiveness of TRX2. Secondly, we show here that Skn7p is involved in receiving the SLN1 regulatory signal; and thirdly, the 10‐fold reduction in uninduced levels of TRX2‐lacZ gene expression shown in Table VII suggests a possible role for Skn7p in basal transcription of TRX2. We are currently investigating which of these Skn7p functions is mediated by the Skn7p‐binding site identified by Morgan et al. (1997).
Since the skn7 D427N allele elevates basal TRX2 activity back to near wild‐type levels (Table VII), the role of Skn7p in basal TRX2 transcription, like the role of Skn7 in oxidative‐stress induction of TRX2, is D independent. Thus, only Sln1‐mediated stimulation of TRX2 expression involves aspartate 427 of Skn7p.
Curiously, the reduction in basal activity of TRX2 we observe in skn7‐Δ mutants was not detected previously at the RNA level or in reporter assays (Morgan et al., 1997). The discrepancy between the reporter data presented here and previous reporter and RNA analysis may be due to strain background differences and is currently under investigation.
At this time, it is not clear whether the D‐phosphorylated form of Skn7p interacts directly with DNA or transmits a signal to the nucleus via additional intermediates. The genes identified in our high‐copy activation screen (Table I) are obvious candidate intermediates, and it is interesting to note that at least one of them, RCK2/CLK1, is known to have serine/threonine kinase activity (Melcher and Thorner, 1996).
Is the Mcm1p transcription factor downstream of the SLN1‐SKN7 signal‐transduction pathway?
Our initial studies on sln1* mutants employed an MCM1‐dependent reporter gene whose activity was responsive to activated alleles of SLN1 relative to reporters harboring non‐MCM1‐dependent UAS elements. Our early model held that the increased pool of phosphorylated Sln1p in a sln1* mutant strain might cause a change in Mcm1p activity, thereby changing the pattern of MCM1‐regulated genes in response to osmotic stress (Yu et al., 1995; Fassler et al., 1997). Our current data showing that mutation of the Mcm1p binding does not impair the sln1* response call into question the previous conclusion that Mcm1p might be a target of the SLN1‐SKN7 pathway. An alternative explanation for our results, which has not been ruled out, is that the mutant Mcm1p‐binding site may retain sufficient Mcm1‐binding activity to allow the stimulatory effect of sln1*. It should be noted that expression of the Mcm1‐regulated genes, Mfα1 and PMA1 was reduced significantly in sln1‐Δ mutants (Yu et al., 1995). However, no effect of the sln1*‐activating mutation could be found on expression of these genes (G.Yu and J.Fassler, unpublished data). Finally, since we find no recognizable Mcm1p‐binding site in the promoter of TRX2, there appear to be additional transcription factor targets of the SLN1‐SKN7 pathway. Further analysis of genes such as TRX2, which respond to SLN1 via phosphorylation of the Skn7p response regulator, will ultimately be required to reveal the nature of the trans‐acting factors and cis‐acting elements required for the sln1* response.
What is the role of SLN1‐YPD1‐SKN7 signaling in osmotic response?
Taken together, our data are consistent with the model depicted in Figure 2. SLN1 coordinately regulates expression of the osmotic‐response genes and a second set of genes by phosphorylation of the two response regulators, Ssk1p and Skn7p. Phosphorylation of Ssk1p maintains the HOG pathway in its inactive state, while phosphorylation of Skn7p promotes signaling to the P‐lacZ reporter and genes such as TRX2. Since expression of the P‐lacZ reporter and the TRX2 gene were responsive to both D‐dependent and D‐independent Skn7 regulation, we speculate that there may exist a family of target genes responsive to both the osmotic and the oxidative‐stress pathways. Earlier studies (Brown et al., 1993; Morgan et al., 1995) in which a high‐copy SKN7 plasmid was shown to suppress the effects of a mutation in the KRE9 gene, involved in cell‐wall biosynthesis, and separately to bypass the requirement for the known G1 transcription factors, SBF and MBF in a D‐dependent manner indicate that cell‐wall‐ and cell‐cycle‐regulated genes are among the targets of the SLN1‐SKN7 pathway. Whether the Skn7‐regulated cell‐cycle genes, CLN1 and CLN2, and unknown cell‐wall genes are also responsive to oxidative stress has not yet been examined.
In the SLN1 system the two response regulators have distinct roles to play. Under conditions that cause accumulation of phospho‐Sln1p, phosphorylated Ssk1p keeps the HOG1 MAP kinase pathway inactive, while Skn7p‐D‐P activates a set of genes including TRX2 and the P‐lacZ reporter. Conditions that cause the accumulation of dephospho‐Sln1p, however, activate the HOG1 MAP kinase pathway to allow survival under osmotic‐stress conditions. The reciprocal regulation of the two SLN1 branches in response to osmotic stress would indicate that the genes downstream of the SLN1‐SKN7 branch also play a role in the osmotic‐stress response. We have shown previously that the SLN1‐SKN7 branch is activated by a hypo‐osmotic stimulus (Tao et al., 1998). Hence, the bifurcated SLN1 pathway allows responsiveness to different aspects of osmotic stress.
The identification of Skn7 at the junction of the oxidative and osmotic‐stress pathways implies that this molecule has an important and general role to play in further integrating cellular stress responses. The multi‐functional nature of the Skn7 protein makes it uniquely suited for the integration of multiple signals and their relay to the nucleus. Future structure‐function studies on the Skn7 protein will reveal the domains and activities specifically required for each pathway.
Materials and methods
Strains, media and yeast techniques
The yeast strains used in this study are listed in Table VIII. skn7 disruption strains, JF1738 and JF1739 were generated by transformation of strains JF1331 and JF1359 with the SacI fragment from a skn7 disruption plasmid from H.Bussey (Brown et al., 1993) and selection on SC‐Trp plates containing 1.5 M sorbitol. Using this procedure, the fraction of transformants with true disruptions of the SKN7 gene was low. Transformants were therefore screened for hypersensitivity to hydrogen peroxide prior to Southern analysis to enrich for skn7 disruptions. SKN7 disruption was confirmed by Southern hybridization analysis of BglII‐ and XbaI‐digested DNA using a 0.9‐kb probe generated from the PCR fragment amplified using primers SKN7‐911F 5′‐GATGACGAATTTGTCAGTGG‐3′ and SKN7+6R 5′‐GCTCATATGGGATATCAAAAGTAAGC‐3′.
YPD1 disruption was achieved by digestion of plasmid pCLM656 with SpeI and SalI prior to transformation of ssk1‐Δ yeast strains. YPD1 disruption was confirmed by the reduced size of the 1.7‐kb EcoRI band in EcoRI‐digested DNA hybridized to a radiolabeled 1.0‐kb SpeI‐EcoRI YPD1 fragment.
Solid and liquid media were prepared as described by Sherman et al. (1986) and included synthetic complete medium (SC) lacking one or more specified amino acids (e.g. SC‐leucine) and rich medium (YPD). Plates for the detection of β‐galactosidase activity contained various amounts of 5‐bromo‐4‐chloro‐3‐indolyl β‐d‐galactopyranoside (X‐gal)/ml and were prepared as described by Larson et al. (1983). 5‐FOA plates used in recombinational loss of the hisG::URA3::hisG allele were prepared according to Boeke et al. (1987). The growth temperature for yeast cultures was 30°C. Responsiveness to oxidative stress was tested using t‐butyl hydrogen peroxide (Sigma) at a final concentration of 0.5 mM. Early log‐phase cultures were exposed to the reagent for 1 h.
The SKN7 CEN plasmid, pSL232, was constructed by cloning a 3.5‐kb SalI‐HindIII fragment from pSL49‐1 (genomic library plasmid; YEp13) (using the SalI and HindIII sites of YEp13) into the SalI and HindIII sites of the pRS315 (Sikorski and Hieter, 1989) polylinker. The D427N derivative was made by substituting the Nsi1‐MscI fragment from the SKN7 CEN plasmid, YCpASN (Morgan et al., 1995). The D427E mutation was introduced by in vitro mutagenesis using oligos OLI‐352 5′‐AAGTATAGGTATGATTTGGTTTTGATGGAAATTGTTATGCCAAACC‐3′ and OLI‐353 5′‐GGTTTGGCATAACAATTTCCATCAAAACCAAATCATACCTATACTT‐3′. These mutant derivatives were substituted for the wild‐type by swapping the NsiI‐MscI fragment in the CEN SKN7 vector to generate plasmids pCLM699 (D427N) and pCLM700 (D/E). Both alleles were tested for function by examining the oxidative‐stress phenotype of strains containing the mutant skn7 derivatives as their sole source of Skn7 protein. Since D427 is not required for oxidative stress, both mutant derivatives were expected (and found) to be resistant to media containing t‐bH2O2.
The GST‐Skn7 expression plasmid, pAA841, was constructed by cloning PCR amplified SKN7 [primers: OLI‐341 (5′‐TTGTTCGAATTCTTAGCTTTTCCACCATAAATAGCAACG‐3′ containing an EcoRI site and OLI‐340 (5′‐GATGTGCTCGAGTGATAGCTGGTTTTCTTGAAGTGTAG‐3′) containing an XhoI site into EcoRI and XhoI digested pGEX‐KG (Guan and Dixon, 1991)].
The YPD1 gene was PCR amplified from ATCC clone 71003 using primer OLI‐253 (5′‐GAACATTTAAACTAGTGTCATTCAG‐3′), and OLI‐255 (5′‐GAAGGATTCTGTCGACTTTGTTGGTAC‐3′) which contain novel SpeI and Sal1 sites respectively. The SpeI, SalI double‐digested PCR product was cloned into Litmus 38 (NEB) to create pCLM636. To construct the GST YPD fusion, pCLM655, YPD1 was amplified from the same template [primer: OLI‐254 (5′‐TCGATCGAGGATCCATGTCTACTATTCCCTCAGAAATC‐3′) containing a BamHI site and primer OLI‐255 described above]. The PCR product was digested with BamHI and SalI and cloned into pGEX‐KG (Guan and Dixon, 1991) to generate plasmid pCLM655. Plasmid pCLM659 was generated by linearizing pCLM636 with SnaBI, converting the ends to SalI using linkers, digesting with SalI to generate a 1.8‐kb fragment and cloning into the SalI site of pCLM654. pCLM654 is a modified YCp50 plasmid in which the URA3 gene was removed by SmaI‐SalI digest, LYS2 added as an EcoRI‐HindIII fragment and the HindIII site converted to a SalI site using SalI linkers.
The YPD1Δ::TRP1 disruption plasmid, pCLM656, was constructed in several steps. pCLM636 was digested with AgeI, filled in with Klenow, ligated in the presence of phosphorylated EcoRI linkers, digested with EcoRI and religated to create a 400‐bp deletion. TRP1 was amplified from pRS314 (Sikorski and Hieter, 1989) using primers complementary to TRP1 sequences and flanked by EcoRI sites. The PCR product was digested with EcoRI and ligated into the EcoRI site of the modified pCLM636 plasmid.
The GST‐Sln1p receiver domain plasmid (pAA689) was constructed by cloning the PCR‐amplified receiver domain [primers SLN1‐3200F (5′‐ACATCAAGTAGAAGAATTCCCACAGTCAAAGACG‐3′) containing an EcoRI site and OLI73573 (5′‐CGCGCAAGCTTTTGATTTCTC‐3′) containing a HindIII site] in frame into EcoRI‐HindIII‐cut digested pGEX‐KG (Guan and Dixon, 1991). The GST‐Sln1p kinase domain plasmid (pHL581) was constructed by cloning the PCR‐amplified kinase domain [primers OLI‐195 (5′‐ ATGACAGACGCATGAATTCAACATTATGCTCTTCTAG‐3′) containing an EcoRI site and OLI‐176 (5′‐CTTTCTACTCTCGAGGATTAAATTCGTC‐3′) containing a XhoI site into EcoRI‐XhoI‐digested pGEX‐KG (Guan and Dixon, 1991)].
The MCM1‐dependent reporter, P‐lacZ is carried on the previously described 2μ plasmid, pGY48 (Yu et al., 1995). pGY48 consists of a 16‐bp palindromic Mcm1‐binding site inserted into the XhoI site of the CYC1‐lacZ vector, pLG670Z (Guarente and Ptashne, 1981). For integration, plasmid pGY48 was converted to integrating plasmid, pGY109 by deletion of a 2.2‐kb fragment containing the 2μ replication origin by partial digestion of the plasmid with EcoRI. The resulting integrating plasmid was linearized in the URA3 gene with ApaI prior to transformation.
Reporters pCLM771 and pCLM772 were constructed in the pGY79 vector which is a derivative of pLGΔ312 (Guarente and Mason, 1983) from which sequences between the SmaI site at position −312 and the XhoI site at position −178 relative to the initiation site of CYC1 have been deleted. Because of the details of the construction, an XhoI site remains at the deletion junction. This vector therefore retains CYC1 sequences to −178 relative to the initiation site (−250 relative to the ATG), but has no CYC1 sequences upstream of the XhoI site. Phosphorylated oligonucleotides, P‐pal (5′‐TCGAGTTTCCTAATTAGGAAAC‐3′), or mutP1 (5′‐TCGAGTTTCCTAATTAATAAAC‐3′) and mutP2 (5′‐TCGAGTTTATTAATTAGGAAAC‐3′) all of which are flanked by XhoI sticky ends were allowed to anneal and then ligated separately into the XhoI site of pGY79. Transformants with four palindromic or four mutant oligonucleotide inserts were found by sequencing and designated pCLM771 (4x P‐lacZ) and pCLM772 (4x mutP‐lacZ).
The TRX2‐lacZ (pJF1078) plasmid was constructed by cloning the PCR‐amplified TRX2 promoter fragment (primers: TRX2 ‐393F, 5′‐TCACCCGGGCGAGGTCACCATTGCAAGCATTG‐3′ containing an added SmaI site, and TRX2+9R, 5′‐GTCAGGATCCAGTGACCATTATTGATGTGTTA‐3′ containing an added BamHI site) into SmaI‐ and BamHI‐digested lacZ vector, pLG670Z (Guarente and Ptashne, 1981). The resulting plasmid lacks all CYC1 regulatory sequences.
Yeast protein extracts were prepared by glass‐bead lysis from cultures grown at 30°C and harvested at a density of 107 cells/ml. Extracts were cleared by centrifugation in all cases. Activities were calculated in Miller units (Miller, 1972) and are the averages of at least three different transformants except where otherwise noted.
Dideoxy sequencing analysis
Sequencing reactions were performed using the Sequenase kit (version 2.0; United States Biochemical Corporation) and [32P]dATP (Amersham). Primers for sequencing the high‐copy clone inserts consisted of oligonucleotides complementary to sequences on either side of the BamHI‐cloning site in YEp13 (and pBR322) (YEp13B‐F, 5′‐GCTACTTGGAGCCACTATC‐3′; and YEp13B‐R, 5′‐GTGATGTCGGCGATATAG‐3′). Reaction products were resolved on 6% acrylamide (19:1 acrylamide/bisacrylamide) sequencing gels containing 7 M urea. Additional sequencing was performed by University of Iowa DNA Core Facility.
Expression and purification of Sln1, Ypd1 and Skn7 fusion proteins
E.coli cultures containing GST‐Sln1 kinase, GST‐Sln1‐associated receiver, GST‐YPD1 or GST‐SKN7 plasmids were grown to an OD600 of 0.6 at 37°C at which time 0.2 mM IPTG was added and the cultures incubated for an additional 2 h. Cell pellets were resuspended in 2 ml buffer B (Posas et al., 1996) per liter of culture. Extracts were prepared by lysis with a French press and removal of cell debris. For purification, aliquots corresponding to 1 l of culture were thawed and incubated with 1.0 ml washed 50% slurry of glutathione beads for 1 h at 4°C. Beads were removed by centrifugation and washed four times in an equal volume of buffer B, and four times in an equal volume of buffer C (Posas et al., 1996) modified to contain 5 mM MgCl2 and 0.12 mM β‐mercaptoethanol instead of dithiothreitol. Beads were resuspended as a 50% slurry in buffer C and stored at −20°C. Skn7p and Ypd1 proteins were eluted as GST fusion proteins by incubation in elution buffer (10 mM glutathione in buffer C) for 1 h at 4°C. The supernatant was collected by gravity flow from a chromatography column. Sln1p‐associated receiver was released from the glutathione beads by thrombin cleavage. Beads were washed twice in thrombin cleavage buffer (50 mM Tris, 50 mM NaCl, 2.5 mM CaCl2) and incubated with biotinylated thrombin for 1 h at 25°C. Thrombin was removed according to the protocol in the Thrombin Cleavage Capture Kit (Novagen). The supernatant was collected by gravity flow from a chromatography column.
Phosphotransfer assays were performed essentially as described previously (Posas et al., 1996) with the following modifications. GST‐tagged Sln1p kinase domain (1 μg) bound to glutathione‐agarose beads was incubated with [γ‐32P]ATP (30 μM, 25 Ci/mmol) for 10 min at 25°C, at which time 1 ml of buffer C (50 mM Tris pH 7.5, 50 mM KCl, 5 mM MgCl2, 0.1% β‐mercaptoethanol) was added, the beads washed four times and resuspended in 1.0 ml Buffer C. For each phosphotransfer reaction, an aliquot of 100 μl of bead suspension was centrifuged to remove the supernatant. Purified GST‐Ypd1p (0.5 μg) and GST‐Skn7p (0.25 μg) in buffer C and purified Sln1p‐associated receiver (0.5 μg) in thrombin‐cleavage buffer were added as indicated and the final volume adjusted to 45 μl. After 15 min, reactions were quenched by the addition of 15 μl loading buffer. Protein phosphorylation was analyzed after SDS‐PAGE (10%), transfer to nitrocellulose and autoradiography.
We thank Guoying Yu for constructing plasmid pGY48 and pGY109, and Howard Bussey for the SKN7 plasmid. We also thank Bob Malone and Scott Moye‐Rowley for critical review of the manuscript. This work was supported by awards to J.S.F. and R.J.D. from The American Cancer Society (RPG‐95‐097‐03‐MGO) and from the US National Institutes of Health (R01 GM56719).
- Copyright © 1998 European Molecular Biology Organization