Ssa1p chaperone interacts with the guanine nucleotide exchange factor of ras Cdc25p and controls the cAMP pathway in Saccharomyces cerevisiae


  • Marco Geymonat,

    1. Laboratoire Information Génétique et Développement, Institut de Génétique et Microbiologie, UMR CNRS Université 2225, Bât. 400, Université Paris-Sud, F-91405 Orsay Cedex, France.
    Search for more papers by this author
    • Present address: International Institute of Genetics and Biophysics, Via Marconi, 12. 80125 Naples, Italy

  • Lili Wang,

    1. Laboratoire Information Génétique et Développement, Institut de Génétique et Microbiologie, UMR CNRS Université 2225, Bât. 400, Université Paris-Sud, F-91405 Orsay Cedex, France.
    Search for more papers by this author
  • Hervé Garreau,

    1. Laboratoire Information Génétique et Développement, Institut de Génétique et Microbiologie, UMR CNRS Université 2225, Bât. 400, Université Paris-Sud, F-91405 Orsay Cedex, France.
    Search for more papers by this author
  • Michel Jacquet

    1. Laboratoire Information Génétique et Développement, Institut de Génétique et Microbiologie, UMR CNRS Université 2225, Bât. 400, Université Paris-Sud, F-91405 Orsay Cedex, France.
    Search for more papers by this author

Michel Jacquet E-mail; Tel. (1) 6915 7963; Fax (1) 6915 7296.


We have found that the guanine nucleotide exchange factor for ras, Cdc25p, interacts with Ssa1p in Saccharomyces cerevisiae. This interaction was observed with GST-fused Cdc25p polypeptides and confirmed by co-immunoprecipitation with the endogenous Cdc25p. Hsp82 appeared also to be co-immunoprecipitated with Cdc25p, albeit to a lower level than Hsp70. In a strain deleted for SSA1 and SSA2, we observed a reduced cellular content of Cdc25p. Consistent with a reduced activity of the cAMP-dependent PKA pathway, the rate of accumulation of both trehalose and glycogen was stimulated in the ssa-deleted strain. Expression of SSA1 reversed these effects, whereas co-expression of SSA1 and PDE2 restored high accumulation. The expression of genes repressed by cAMP, GAC1 and TPS1, fused to β-galactosidase, was also stimulated by deletion of SSA genes. The effect of ssa deletion on glycogen accumulation was lost in a strain deleted for CDC25 rescued by the RAS2ile152 allele. Altogether, these results lead to the conclusion that Ssa1p positively controls the cAMP pathway through Cdc25p. We propose that this connection plays a critical role in the adaptation of cells to stress conditions.


Ras proteins are molecular switches involved in signal transduction pathways. They are activated by guanine nucleotide exchange factors (GEF) that transduce various signals. In cells of higher eukaryotes, tyrosine kinase receptors, when phosphorylated in response to extracellular signals, recruit the GEF of the SOS family linked to Grb2, or grb2-like, adapters (Schlessinger, 1994). In unicellular lower eukaryotes such as yeast, in the absence of tyrosine kinase receptors and grb2-like adapters, the mode of control of GEF is poorly understood. In Saccharomyces cerevisiae, Cdc25p, the GEF for ras proteins, is essential for cAMP production and PKA activation. As it is an unstable protein containing a cyclin destruction box (Kaplon and Jacquet, 1995), its cellular content is likely to be a limiting factor potentially used for its regulation. This protein is tightly associated with the membrane (Garreau et al., 1996), and the control of the ras pathway by its recruitment to the membrane (Gross et al., 1992a) remains to be further substantiated. This protein belongs to a multimolecular complex, it interacts with itself and potentially Sdc25p in vitro and forms a dimer when expressed in Escherichia coli (Camus et al., 1997) and it contains an SH3 domain that has been reported to interact with the adenylate cyclase (Freeman et al., 1996). Therefore, the assembly process of a multimolecular structure at the membrane level could also be part of a control of ras by its GEF.

In the search for interacting proteins expected to be involved in the control of the Cdc25p function, we have found Ssa1p, a chaperone of the HSP70 superfamily. The SSA subfamily, is composed in yeast of four members (SSA1–4). During growth, only SSA1 and SSA2 are expressed at high level, whereas SSA3 and SSA4 are not. After heat shock or during a stress response, Ssa1p is overexpressed and SSA3 and 4 are induced (Werner-Washburne et al., 1987). These chaperones have multiple effects and control different functions in the cell. Ssa1/2p have been shown, in addition to their role in protein translocation in vivo (for review see Craig et al., 1994), to participate in vitro in the prevention of protein aggregation and in the refolding of denatured proteins (Bush and Meyer, 1996). Hsp70 are often found in association with Hsp90 chaperones that are encoded in yeast by two genes, HSP82 and HSC82 (Chang and Lindquist, 1994).

In this report we describe the identification of Ssa1p as a protein tightly bound to Cdc25p. We present evidence for a role of Ssa in the control of the Cdc25–Ras–cAMP pathway. We then conclude with the proposal that, in response to stress, the recruitment of chaperones to unfolded proteins reduces the amount of functional Cdc25p and then the activity of the PKA pathway.


Ssa1p co-purifies with a GST–Cdc25 fusion protein

We have previously demonstrated that a Cdc25p region of about 100 aa located in the C-terminal part of the protein is responsible for the membrane localization of Cdc25p (Garreau et al., 1996). To search for protein(s) that could interact with this region, we have constructed a plasmid in which the coding sequence of the C-terminal fragment of Cdc25p (from amino acids 1441–1589) has been fused to that of GST. The fused gene expressed in yeast produces the GST–Cdc25Cter chimera with the predicted molecular weight. It was localized in the same insoluble fraction as the complete Cdc25p or the C-terminal domain alone. The GST–Cdc25Cter chimera was solubilized by 2 mM EDTA pH 12, which is the only relatively mild treatment able to solubilize Cdc25p (Gross et al., 1992b). Indeed this treatment allows the detection of Cdc25–Ras interaction (Gross et al., 1992a) and Cdc25p dimerization (Camus et al., 1997). As shown in 1Fig. 1A a major band of 70 kDa was co-purified with the GST–Cdc25Cter chimera but not with the GST alone used as a control. Although the relative amount of this protein was lower than that of the GST chimera, as judged from staining, this ratio was highly reproducible, indicating a specific interaction. Furthermore, treatments with ATP (up to 400 mM) or high salt concentration (up to 1 M KCl) were not able to release the 70 kDa associated protein (data not shown).

Figure 1.

. Ssa1p associates with a Cdc25p C-terminal fragment fused with GST. A. C13ABYS86 yeast cells transformed with pEG(KG) (GST alone) or pEG-FS4 (GST–Cter–Cdc25 fusion) were grown in a galactose-based medium. Glutathione–agarose-bound proteins derived from cells harbouring pEG(KG) (GST) and pEG-FS4 (GST–Cter–Cdc25 fusion) were eluted, subjected to SDS–PAGE and stained with Coomassie blue. The 70 kDa specifically co-purified polypeptide is indicated as well as GST and GST-Cdc25 polypeptides. B. The sequences of the analysed peptides are compared with the sequences of the corresponding regions of Ssa1p, Ssa2p, Ssa3p and Ssa4p. Asterisks represent conserved amino acids (SWISSPROT references: P10591, P10592, P09435, P22202).

The eluted 70 kDa material was analysed by microsequencing of peptides produced with endoprotease Lys C (Boehringer). Although the different members of the Hsp70 family in yeast, namely Ssa1p, Ssa2p, Ssa3p and Ssa4p, present strong sequence similarities, the sequences of five peptides obtained in stoichiometric amounts correspond to the product of SSA1 (Fig. 1B). In addition the N-terminus of the 70 kDa protein is blocked as reported for the Ssa subfamily of Hsp70 proteins in yeast.

Hsp70 and possibly Hsp82 are co-immunoprecipitated with endogenous Cdc25p

To characterize further the interaction between Cdc25p and Ssa1p, we have performed immunoprecipitation experiments. In a first experiment we have used epitope-tagged (HA) full-length Cdc25 protein, tagged at the C-terminus (pTK3). To assess the presence of Ssa1p in the immunoprecipitate, we have taken advantage of the polyclonal antibody Ab119 raised against the last 13 aa of the human Hsp90β (Lees-Miller and Anderson, 1989), which recognizes α- and β-Hsp90 vertebrate forms and cross-reacts with vertebrate Hsp70 (Jerome et al., 1993). 2Figure 2A shows that Ab119 recognizes the co-purified Ssa1p previously characterized (lanes a and b) as well as the yeast Hsp82 (lanes c and d). The recognition of the Hsp82 was confirmed by the use of a strain deleted for HSP82 and HSC82 complemented either by the yeast HSP82 (lane c) or by chicken HSP90α (lane d). The difference in migration of the immunodetected bands corresponds to the difference between the sizes of chicken HSP90α and yeast Hsp82p. The recognition of Hsp70p and Hsp82 in yeast is likely to be the result of the conservation of the last four C-terminal aa between Hsp90 family and Hsp70 of the Ssa subfamily. In 2Fig. 2B the epitope-tagged Cdc25p is co-immunoprecipitated with Hsp70. The interaction of the endogenous Cdc25p with Hsp was also tested by immunoprecipitation using the polyclonal antibody A323 directed against the C-terminal portion of Cdc25p (Garreau et al., 1996). As shown in 2Fig. 2C, Hsp70 was also co-immunoprecipitated from a membrane extract of a CDC25 wild type but not from a strain deleted for CDC25 rescued by RAS2Ile152, a CDC25-independent ras allele (Camonis and Jacquet, 1988), although Hsp70 and Hsp82 were present at the same level in the two strains (Fig. 2C, left). It is interesting to note that a small amount of Hsp82 was co-immunoprecipitated together with the Cdc25–Ssa1p complex in the case of both the HA-tagged version and the endogenous Cdc25p. This result was reproducible and could suggest the existence of a complex with a member of the Hsp82 family.

Figure 2.

. Co-immunoprecipitation of Cdc25p and chaperones. A. Specificity of the polyclonal antibody Ab119. Western blot analysis of proteins extracted from C13ABYS86 transformed with pEG-FS4 (GST–Cter–Cdc25 fusion) bound to gluthatione–agarose beads (lanes a and b) and of membrane proteins extracted from ΔECU82a (hsp82, hsc82) transformed either with pKAT6, expressing the yeast HSP82 (lane c), or with pTGpd90α, expressing the chicken HSP90α (lane d). Lane a was stained with Ponceau S, and lanes b, c and d were labelled with Ab119. B. Immunoprecipitation of a Cdc25-tagged protein. Membrane proteins immunoprecipitated with 12CA5 from C13ABYS86 transformed with the empty vector (vec) or expressing the complete tagged Cdc25p in pKT3 (Ha-Cdc25) were blotted onto nitrocellulose and labelled with Ab119. C. Immunoprecipitation of the endogenous Cdc25p. Crude membrane extract (Extract) or proteins immunoprecipitated with A323 (anti-Cdc25p) (Ipp) from OL550–11 A (Δcdc) or from C13ABYS86 (wt) were blotted onto nitrocellulose and labelled with Ab119. Size markers are on the left.

To determine whether the interaction between Cdc25p and the Hsp70 is specific for Ssa1p or can also occur with the related Hsp70 encoded by SSA2, SSA3 and SSA4, similar immunoprecipitation experiments were performed in strains deleted for ssa1, and for ssa1 and ssa2, both of them transformed with pTK3. In Fig. 3, it can be seen that Ab119 recognizes a 70 kDa protein in both strains. This protein could be Ssa2p in the Δssa1 strain because SSA3 and SSA4 are not expressed at 26°C in this strain and Ssa3/4p in the Δssa1, Δssa2 mutant, because Ssa3/4p migrate at a slightly slower rate than Ssa1/2p in SDS–PAGE and are overexpressed in this strain (see Fig. 2) (Werner-Washburne et al., 1987). From this result it can be concluded that the interaction between Cdc25p and Ssa1p can be extended to Ssa2p, Ssa3p and/or Ssa4p, at least in the absence of Ssa1p.

Figure 3.

. Cdc25p also interacts with Ssa2p and Ssa3/4p. Crude membrane extract (Extract) or proteins immunoprecipitated with 12CA5 (Ipp) from T141 lacking SSA1(Δ1) or from OL565 lacking SSA1 and SSA2(Δ1,Δ2) transformed with pTK3 were blotted onto nitrocellulose and labelled with Ab119. Size markers are on the left.

The relative abundance of Cdc25p is reduced in aΔssa1, Δssa2 mutant

The biochemical function of chaperones is thought to assist protein folding, thus it can be hypothesized that the interaction with Cdc25p plays a role in the correct folding of this protein or in its assembly with membrane or other protein components. If this is the case, a defect in chaperones could lead to a lower amount of functional Cdc25p. To test this hypothesis, we have compared the amount of Cdc25p between a Δssa1, Δssa2 mutant and the isogenic strain complemented with SSA1. As the amount of Cdc25p is very low and barely detectable with Cdc25 antibodies, we have replaced the CDC25 gene at its locus with a three HA epitope tagged version. The tagged Cdc25p (T-Cdc25p) is functional, allowing normal growth. As seen in Fig. 4, T-Cdc25p migrates in gel electrophoresis at the expected position from its size. In contrast to other cellular proteins, the amount of T-Cdc25p found in the Δssa1, Δssa2 mutant (V) was lower than in the SSA1 complemented strain (SSA1). To establish further that the reduction of Cdc25p was due to an effect at the protein level rather than on its expression, we have measured the relative amount of mRNA in the two strains. We have used a quantitative RT-PCR assay using GPH1 and BCY1 as internal controls. From the experiment illustrated in 4Fig. 4B, no significant variation was observed at the mRNA level contrasting with the reduction seen at the protein level in the absence of Ssa1p. Therefore the limitation of Hsp70 because of the double deletion, although partially compensated by overexpression of SSA3 and SSA4, reduces the cellular content of Cdc25p but not of its mRNA relative to other cellular proteins.

Figure 4.

. Relative abundance of the Cdc25 protein and its mRNA in ssa1 and ssa2 mutant. The ssa1, ssa2, mutant strain (-4C) transformed either by the control plasmid pRS314 (V) or by pGRSSSA1 (SSA1) was grown at 30°C in minimal medium and harvested in log growth phase (OD700 = 0.5) for proteins and RNA measurements. A. Protein content. A 70 μg sample (a and c) and 35 μg sample (b and d) of total proteins from each culture were separated on 7.5% acrylamide gels as previously described (Garreau et al., 1996). After transfer, the upper part of the nitrocellulose sheet was processed for immunostaining of T-Cdc25p, the lower part was stained with Ponceau S. Densitometric analysis showed a threefold reduction in Cdc25p content in the ssa1, ssa2 mutant, when compared with the SSA1-complemented strain. B. mRNA content. A 5 μg sample of total RNA from each culture (SSA1, and V) was used in reverse transcription performed with the Superscript II kit (Gibco-BRL). Then PCR was primed with oligonucleotides specific for CDC25, GPH1 and BCY1. The reaction was run for 23 (a, d), 26 (b, e) and 29 (c, f) cycles. Densitometric quantification as well as independent radioactive measurements during kinetic gave a ratio of CDC25 mRNA found between the SSA1-containing cell and the double mutant in the range of 0.8 after normalization with either GPH1 or BCY1.

Ssa chaperones affect the cAMP-dependent rate of production of trehalose, glycogen and gene expression

If Hsp70 proteins control the cellular content of Cdc25p, responses to cAMP are expected to be modified by deletion of SSA1 and SSA2. To test this hypothesis, we have compared the behaviour of the Δssa1, Δssa2 mutant containing or not the SSA1 gene on a plasmid. We have followed the rate of production of glycogen and trehalose as markers of the efficiency of the cAMP pathway. The cellular content in glycogen is controlled by the cAMP pathway at the enzymatic level on both glycogen synthetase and glycogen phosphorylase (Rothman-Denes and Cabib, 1970; Huang and Cabib, 1972) and at the level of gene expression (François et al., 1992). Several mutants defective in glycogen accumulation were found to be affected on genes of the cAMP pathway [glc1 = ira1, glc4 = ira2, glc5 = ras2 (Cannon et al., 1994)]. Trehalase activation and trehalose degradation are also mediated by cAMP-dependent PKA (van Solingen and van der Plaat, 1975).

Glycogen is not accumulated constitutively during the culture but appears later before glucose exhaustion. As the cellular contents of the cell vary during the culture, we used the ‘Monod plot’ to compare the different strains. This representation (specific compound plotted versus total protein) allows the rate of accumulation (slope of the curve) to be followed and to be compared under different situations and even at different rates of growth. We found that the Δssa1, Δssa2 mutant complemented by SSA1 on a plasmid, which does not exhibit an altered phenotype, has a normal rate of glycogen accumulation (Fig. 5A). In contrast, in the absence of SSA1 and SSA2 the rate of accumulation of glycogen was dramatically increased. A similar increase was seen when the high-affinity cAMP phosphodiesterase, encoded by the PDE2 gene, was overexpressed from a plasmid in the strain containing SSA1. We have checked in the SSA1-containing strain that the rate of glycogen accumulation relative to protein was independent from the growth rate. Indeed we have obtained superimposable curves at different temperatures for different growth rates (data not shown).

Figure 5.

. Accumulation of intracellular glycogen and trehalose depends on the presence of the SSA1 gene product. Cells grown on minimal medium (YNB) were harvested during exponential growth between OD700 0.4 and 1.6. The amount of glycogen (A) or trehalose (B) is plotted versus. the amount of proteins (Monod plot) measured in a 1 ml culture. OL565 (ssa1, ssa2) transformed with pRS314 and YEp352 (▪), pGRSSA1 (SSA1) and YEp352 (•) or pGRSSA1 and pGR103 (PDE2) (▴).

We have also measured trehalose accumulation in the Δssa1, Δssa2 mutant (Fig. 5B). Trehalose is accumulated after the glycogen during batch cultures on glucose, most likely as the result of a different regulatory circuit but also as a function of the cAMP pathway. In contrast to the situation found for glycogen, the overexpression of phosphodiesterase did not result in a linear relationship between trehalose and protein, but a variation in the differential rate of accumulation was still observed, suggesting the existence of a cAMP-independent control on trehalose accumulation. Interestingly, the curve obtained in the non-complemented double mutant is very close to that obtained with the complemented mutant harbouring the PDE2 plasmid. The differential rate of trehalose accumulation was increased earlier in the non-complemented mutant and reached higher values, as in cells containing low cAMP levels.

It is also known that the cAMP pathway negatively regulates genes that contain stress-responsive elements (STRE) in their promoter (Ruis and Schüller, 1995). In the Δssa1, Δssa2 mutant, we have analysed the expression of β-galactosidase controlled by the promoters of two genes repressed by cAMP and containing STRE sequences in their promoter region, GAC1 (François et al., 1992) and TPS1 (Winderickx et al., 1996). In the early exponential growth phase, the transcription of GAC1 and TPS1 was four-fold higher in a strain deleted for both SSA1 and SSA2 than in the same strain complemented with SSA1 (Table 1). By contrast, the level of transcription of the CYC1LacZ fusion, used as a control, was independent from the presence of Ssa1/2p. Therefore, the loss of SSA1 and SSA2 mimics a low intracellular level of cAMP for glycogen and trehalose accumulation as well as cAMP-dependent gene expression.

Table 1. . GAC1 and TPS1 promoters are regulated by SSA1 gene product. The specific activity of the β-galactosidase (arbitrary units) was measured in early exponential phase (OD700 = 0.3) of OL565 (ssa1, ssa2) co-transformed with pJF169 (GAC1–LacZ), pTPS1 (TPS1–LacZ) or pLG669Z (CYC1–LacZ) and pGRSSA1or pRS314.Thumbnail image of

A CDC25 deletion rescued by RAS2ile152is epistatic on SSA deletions for the control of glycogen accumulation

The experiments presented above illustrate the effect of ssa chaperones on cAMP targets. To analyse further whether or not these effects are mediated by the cAMP pathway and more specifically Cdc25p we have used a cell in which CDC25 was completely deleted. As CDC25 is an essential gene, this experiment was performed in a strain rescued by the RAS2ile152 allele (Camonis and Jacquet, 1988). We chose this suppressor rather than others such as RAS2val19 to avoid any bias due to overactivity of the pathway. Therefore we have constructed the strain OL576-3C containing four mutations: (i) the deletion of the two genes ssa1 and ssa2 inherited from the strain MW123; and (ii) the complete deletion of the CDC25 open reading frame (ORF) with the integrated copy of RAS2ile152. This strain still exhibits cAMP-independent phenotypes for the defect of chaperones, reduced growth and thermosensitivity. As shown in Fig. 6, the relative rate of glycogen accumulation in the strain containing the four mutations was almost identical to that of the strain with the Δcdc25 RAS2ile152 mutations, whereas it was lower than the ssa-deleted strain. This difference was also observed in synthetic medium rather than the rich medium used here. Thus, it can be concluded that the lack of CDC25 and the activity of the product of RAS2ile152 is epistatic upon the deletion of ssa1 and ssa2 on the stimulation of glycogen accumulation.

Figure 6.

. Epistatic effect of Δcdc25, RAS2ile152 on Δssa1, Δssa2 for glycogen accumulation. Cells grown in rich medium (YEPD) were harvested during exponential growth between OD700 1.5 and OD700 5.5. The amount of glycogen is plotted versus the amount of proteins measured in 1 ml of culture for the strain OL565 (ssa1, ssa2 mutant) (▴), OL550–11A (Δcdc25,RAS2ile152) (▪) and OL576–3C (ssa1, ssa2, cdc25, RAS2ile152) (♦). The standard deviation on glycogen measurement is less than 10%.


Cdc25p interacts with chaperones

Searching for proteins interacting with the S. cerevisiae Ras GEFCdc25p, we have found that the SSA1 gene product co-purified with a GST–Cdc25p fusion protein containing a C-terminal fragment of Cdc25p. The reproducibility of this interaction has been assessed by immunoprecipitation experiments of either an HA-tagged Cdc25 polypeptide or the endogenous Cdc25p. In addition, using an antibody directed against Hsp70p of the Ssa subfamily and Hsp82, we have shown that both the HA-tagged and the endogenous Cdc25p were co-immunoprecipitated with Hsp70p chaperones. Hsp82 was also found associated with Cdc25p in immunoprecipitation experiments. As this interaction was weaker than with HSP70, we have not investigated it further in this report, although it might be important because interactions between members of the Hsp70 and Hsp80 families have already been described in yeast (Chang and Lindquist, 1994).

Hsp70 and other members of the cellular chaperones family have a general capability to bind unfolded proteins (Palleros et al., 1991). Overexpressed proteins are potential targets for such interactions, nevertheless we have shown that the interaction also occurs using the functional endogenous Cdc25p. It could also be argued that the solubilization procedure of Cdc25p is responsible for the binding to chaperone, but if this were the case the more abundant chaperones such as the Ssb proteins (Nelson et al., 1992) or even Ssa2p, which is more abundant than Ssa1p in growing cells (Boucherie et al., 1996), would have been preferentially found. Indeed, Cdc25p is able to bind to other Ssa proteins such as Ssa2p in the ssa1 mutant, but Ssa1p was the only Hsp70 protein identified in microsequencing analysis, suggesting a higher affinity for this chaperone, which does not preclude a partially redundant role of the other members of the family. As discussed below, genetic and physiological interactions between the cAMP pathway and SSA1 further establish the relevance of this interaction in vivo.

Ssa chaperones control the cAMP signalling pathway

The biological significance of this interaction was first assessed by comparison of the level of Cdc25p between the Δssa1, Δssa2 mutant and the SSA1-complemented strain. The amount of Cdc25p relative to other proteins was significantly reduced in this mutant. This reduction is not the result of a lower expression because the level of mRNA was found to be quite similar between the two strains. This result suggests a direct effect of the chaperones on Cdc25p. As Cdc25p, an unstable protein, is present in very low amounts in the cell (Kaplon and Jacquet, 1995), a regulation on its cellular content will be readily transmitted to the signalling pathway. Indeed, the activity of the cAMP pathway depends upon the amount of Cdc25p as deduced from the slow growth observed in cells where the GEF domain of CDC25 is expressed from a poor endogenous promoter (M. Jacquet, unpublished; Petitjean et al., 1990). Therefore, a reduction in activity of the Ras–cAMP–PKA pathway can be predicted in cells having a reduced amount of Ssa protein such as in the double ssa-deleted strain. Members of the SSA family are essential for growth and partially redundant; the double deletion of SSA1 and SSA2 is still viable because of the overexpression of SSA3 and SSA4, but this is not sufficient to maintain a normal growth rate. Among the pleiotropic phenotypes associated with the defect of these chaperones, several can be explained by the activation of Hsf1p (Craig and Jacobsen, 1984; Werner-Washburne et al., 1987; Boorstein and Craig, 1990) and are not suppressed by suppressors of the cAMP pathway, but some could be related to the cAMP pathway. Constitutive thermotolerance, which is a paradoxical phenotype for a defect of chaperone, could be related to the cAMP pathway because it was also found associated with mutations reducing the activity of the cAMP pathway (De Virgilio et al., 1993; Shin et al., 1987). In addition, in a strain depleted for Ssa chaperones, an increased level of trehalose after heat shock was found to result from a decrease in trehalase activity that is directly controlled by PKA (Hottiger et al., 1992). In the present report we have confirmed this observation on trehalose and further extended the conclusion to other targets of the cAMP pathway. We have shown that the rate of accumulation of trehalose depends upon the presence of SSA1 when SSA2 is lacking. The SSA1 effect requires the cAMP pathway function because it is lost when the pathway is blocked by reducing the level of cAMP with excess phosphodiesterase. The control of the glycogen content of the cell by SSA1 is a newly described phenotype for these chaperones. As for trehalose, the limitation of cellular chaperones mimics a reduced activity of the cAMP pathway; a modulation of the cAMP pathway by overexpression of the phosphodiesterase is again epistatic on the effect of SSA1 on glycogen accumulation. In addition, looking at transcriptional targets of the cAMP pathway, we have extended the effect of the Ssa chaperones to the expression of TPS1 and GAC1 promoters. The promoters of both TPS1 and GAC1 lacking canonical HSE, contain STRE sequences that mediate negative response to the cAMP/PKA pathway; they have been shown to respond to cAMP (D. Tadi and E. Boy Marcotte, unpublished). Therefore, their Ssap-dependent expression is not likely to be mediated by HSF but rather by the cAMP pathway, most likely through STRE.

The diversity of the cAMP-dependent phenotypes controlled by Ssa strongly argues for an effect of these chaperones on the control of the cAMP pathway. The evidence that this control acts upstream of the cascade comes from our last experiment, in which we have analysed the effect of the SSA deletions in a strain with a normal cAMP pathway and a strain where this pathway was defective in CDC25 and bypassed at the level of ras. Although the difference between the curves is small because Ras2pile152 is less effective in activating the cAMP pathway than a normal protein, this difference was reproducible and leads us to conclude that Ssa proteins should control the pathway upstream of ras. Altogether, the result of physical interaction of Cdc25p with the Ssa proteins and the physiological effect of Ssap on the cAMP–PKA pathway led us to propose a model in which the interaction of Hsp with the GEF of ras: Cdc25p, is one of the regulatory systems acting on the cAMP signalling pathway.

Why are chaperones modulating the cAMP signalling pathway?

A lowered activity of the PKA pathway appears to be required for an adaptation of the cell to stress conditions, as deduced from the inhibitory effect of a high PKA activity (Ruis and Schüller, 1995) (Boy-Marcotte et al., 1998), or from the thermotolerance acquisition due to deregulating mutations affecting this pathway (Shin et al., 1987). During heat shock, several genes containing STRE elements in their promoter are also induced (Lahskari et al., 1997). This element is controlled by Msn2p and Msn4p, which have been shown to have their nuclear localization controlled by the cAMP pathway (Görner et al., 1998). The interaction between Cdc25p and Hsp70 and Hsp90 could then play a critical role in this model to connect the activity of the cAMP pathway to the state of the cell. Hsp70 has been proposed to play the role of a cellular thermometer, and it would be recruited by unfolded proteins produced after heat shock and thus would be less available to repress Hsfp (Craig and Gross, 1991) (Shi et al., 1998). A similar model could be proposed for the control of the cAMP pathway: a transient accumulation of unfolded proteins after a stress would reduce the availability of Hsp70 for Cdc25p as for Hsf1p and then reduce the activity of the cAMP pathway. This reduction would allow the expression of the regulon controlled by STRE that is induced in response to several conditions of stress (Ruis and Schüller, 1995) or metabolic changes (Boy-Marcotte et al., 1998) when many proteins have to be removed from the cell.

Experimental procedures

Yeast strains

The previously described strains we used are W303-1B, W3031B/D (Nasmyth, 1985), C13ABYS86 (Heinemeyer et al., 1991), T141 (MATa, ura3-52, his3-11,15, leu2-3,112, lys2, ssa1::LEU2), MW123 (MAT a, ura3, his3, leu2, lys2, Δtrp1, ssa1::HIS3, ssa2::LEU2 (Werner-Washburne et al., 1987), ΔECU82a (MATa, ade2-1, ura3-1, his3-11,15, leu2-3,112, trp1-1, hsc82::LEU2, hsp82::LEU2).

The following strains have been constructed for the present work. OL565 (MATa, ura3, his3, leu2, lys2, trp1, ssa1::HIS3, ssa2::LEU2) was derived from a cross between MW123 and W303-1B/D. To construct the CDC25 deleted strains: (i) the RAS2ile152 allele (Camonis and Jacquet, 1988) contained within a Cla1–BamH1 fragment was cloned into the integration plasmid pRS304 (Sikorski and Hieter, 1989) to produce pGR113, this plasmid linearized with Pst1 was introduced into the diploid yeast strain FY1679 (Baudin et al., 1993) to give the trp+ OL549 strain; (ii) the CDC25 gene was replaced by the HIS3 gene amplified with two oligonucleotides corresponding to the 35 nucleotides upstream of the CDC25 ATG (at −26) and downstream of the stop codon (+48), respectively, extended by the HIS3 primers (Baudin et al., 1993); (iii) the resulting strain OL550 was sporulated and asci dissected. Among the progeny His+,Trp+ cells were recovered and analysed by PCR to confirm gene replacement. Two of these strains were used in this report: OL550–11A (MATα, ura3-52, his3-Δ200, leu2-Δ1, trp1-Δ63, cdc25::HIS3, RAS2-RAS2ile152::TRP1) and OL550-11B (MATa, ura3-52, his3-Δ200, leu2-Δ1, trp1-Δ63, cdc25::HIS3 RAS2-RAS2ile152:: TRP1).

The Δssa1, Δssa2 mutant containing the cdc25::HIS3, RAS2ile152 was obtained from a cross between the strain OL550-11B and OL565. One strain among the progeny, OL576-3C, was found to contain (i) the double ssa1 and ssa2 deletion, as judged from HIS3 and LEU2 segregation in the tetrad and the associated phenotypes: slow growth and thermosensitivity; (ii) the CDC25::HIS3 replacement from PCR analysis; and (iii) the presence of RAS2ile152 from TRP1 segregation and growth rescue of the cdc25 deletion.

The OL568-1C strain was derived from the diploid strain W303–1B by insertion of three tandem repeats encoding the influenza virus haemagglutinin epitope (HA) at the 5′ end of the CDC25 gene, between codons 4 and 5, using the method of Schneider et al. (1995). The 3xHA-URA3-3xHA cassette of the pMPy-3xHA plasmid was amplified by PCR, using ol25NHA up and ol25NHA down and was transferred to the W303-1B strain. Ura+ transformants with the CDC25 gene replaced were checked by PCR and plated onto 5-fluoro-orotic acid to select ura cells. After sporulation and tetrad analysis, the haploid OL568-1C strain (ade2-1, ura3-1, his3-11, leu2-3, 112, trp1-1, CDC25:3xHA) was recovered. As checked by PCR, it contained three tandem repeats of the HA epitope. The MGL20-4C strain (ura3, lys2, trp1, ssa1::HIS3, ssa2::LEU2, CDC25:x3HA) was obtained from a cross between OL568-1C and OL565.

Plasmid constructions

Plasmids used in this study are listed in Table 2. pTK3 was constructed by insertion of a three HA tag coding sequence within the Nhe1 site of pTK2, allowing expression of the full CDC25 gene under the GAL1-10 promoter (Kaplon and Jacquet, 1995). pEG-FS4 was obtained by insertion of a PCR fragment containing the last 148 codons of CDC25 in the SmaI–SalI sites of the pEG(KG) (Mitchell et al., 1993); this fragment was amplified with olYF1 and olYF2 on the plasmid pYFS4 with ol1690 and ol2218 and inserted into YeF1 (Cullin and Minvielle-Sebastia, 1994) by Not1 and EcoR1.

Table 2. . Plasmids and oligonucleotides used.Thumbnail image of

The SSA1 gene was cloned by gap repair into the plasmid pGRSSA1 using the intergenic flip-flop procedure (Mallet and Jacquet, 1996). The SSA1 promoter sequence (from −579 to −108 of ATG) and terminator sequence (from +50 to +585 of the stop codon) were amplified using oligonucleotides ol600, ol1053 and ol3155, ol3657 respectively. The second PCR has been primed with ol600 and ol3657 and the two amplified fragments. The resulting fragment was cut with XhoI, and NheI was cloned into XhoI–SpeI sites of pRS314 (Sikorski and Hieter, 1989). The Sma1 linearized plasmid was transformed into W303 on medium lacking tryptophan. The plasmid pGRSSA1 complements the thermosensitivity and the reduced rate of growth of a Δssa1, Δssa2 mutant.

pGR103, expressing the PDE2 gene, was obtained by cloning a 3 kb BamHI–BglII fragment obtained from pW4 (Wilson and Tatchell, 1988) into the BamHI site of YEp352 (Hill et al., 1986).

Whole-cell protein extract, subcellular fractionation, GST fusion proteins purification and immunoprecipitation

Crude cell extracts and membrane fractions were prepared as previously described (Garreau et al., 1990) with the addition of 20 mM sodium molybdate in the lysis buffer. Solubilization of complete Cdc25p or fused protein was performed as described by Gross et al. (1992b) with 2 mM EDTA pH 12. GST fusion proteins were further purified by incubation of the neutralized supernatant with glutathione–agarose beads for 1 h at 4°C. After extensive washing in washing buffer (PBS, 0.5% Triton X-100, 0.25 M KCl), bound proteins were recovered by boiling the beads in 2× Laemmli sample buffer and subjected to SDS–PAGE. For GST control, the supernatant of the first centrifugation was directly incubated with glutathione–agarose beads.

Immunoprecipitation was performed by incubation of the neutralized supernatant from 50 ml of culture, at 4°C, 1 h with the antibody and then 1 h in the presence of protein A–Sepharose. Bound proteins were recovered after washing by boiling in 2× Laemmli sample buffer and subjected to SDS–PAGE. Immunoreactive bands were detected by goat anti-rabbit or anti-mouse IgG serum, alkaline phosphatase conjugate (ProMega) or peroxidase conjugate (Amersham).

Glycogen, trehalose, β-galactosidase and protein determinations

Glycogen and trehalose were measured as described (Parrou and François, 1997). β-Galactosidase was measured from cell extracts with the Ozyme Galacto-Light plus chemiluminescent reporter assay kit as recommended by the manufacturer. Protein concentrations of extracts were measured according to the method of Bradford.

Peptide analysis

Digestion and separation of peptides were performed by the Laboratoire de Microséquençage des Protéines of the Institut Pasteur (Paris) as described in Kawasaki and Suzuki (1990) with slight modifications: proteins separated on SDS polyacrylamide gel were stained with 0.03% Amido Black; the buffer used for digestion was 0.1 M Tris-HCl pH = 8.8, 0.03% SDS, an Aquapore AX300 (Brownlee) column was used to remove SDS and the peptides were separated on a C18 Vydac 218TP52 column. Amino acid sequences were determined with an Applied Biosystems Procise ABI 498 (PROCISE-HT).


  1. Present address: International Institute of Genetics and Biophysics, Via Marconi, 12. 80125 Naples, Italy


We are very grateful to G. Renault for plasmids and yeast strains constructions and to J. d'Alayer for peptide sequencing. We thank S. Lindquist, E. A. Craig for providing yeast strains; C. Conessa and M. G. Catelli for the gift of 12CA5 and Ab119 antibodies; C. Cullin, R. J. Deschenes, J. M. François and J. M. Thevelein for the gift of plasmids. We are also indebted to E. Boy-Marcotte, M. G. Catelli and C. Mann for fruitful discussions. This work was supported by le Centre National de la Recherche Scientifique, le Ministère de l'Enseignement Supérieur et de la Recherche and by grants from l'Association pour la Recherche sur le Cancer and la Ligue Nationale contre le Cancer. M.G. was supported by a fellowship from l'Association pour la Recherche sur le Cancer. L.W. was supported by grant from le Ministère des Affaires Etrangères.