Yeast PKA represses Msn2p/Msn4p‐dependent gene expression to regulate growth, stress response and glycogen accumulation

Angela Smith, Mary P. Ward, Stephen Garrett

Author Affiliations

  1. Angela Smith1,
  2. Mary P. Ward2 and
  3. Stephen Garrett*,3
  1. 1 Department of Biochemistry, Duke University Medical Center, Durham, NC, 27710, USA
  2. 2 Department of Biochemistry and Biophysics, School of Medicine at UNC, Chapel Hill, NC, 27599‐7260, USA
  3. 3 Department of Microbiology and Molecular Genetics, UMDNJ, New Jersey Medical School, Newark, NJ, 07103‐2714, USA
  1. *Corresponding author. E-mail: garretst{at}
View Full Text


Yeast cAMP‐dependent protein kinase (PKA) activity is essential for growth and antagonizes induction of the general stress response as well as accumulation of glycogen stores. Previous studies have suggested that the PKA effects on the two latter processes result in part from transcription repression. Here we show that transcription derepression that accompanies PKA depletion is dependent upon the presence of two redundant Zn2+‐finger transcription factors, Msn2p and Msn4p. The Msn2p and Msn4p proteins were shown previously to act as positive transcriptional factors in the stress response pathway, and our results suggest that Msn2p and Msn4p also mediate PKA‐dependent effects on stress response as well as glycogen accumulation genes. Interestingly, PKA activity is dispensable in a strain lacking Msn2p and Msn4p activity. Thus, Msn2p and Msn4p may antagonize PKAdependent growth by stimulating expression of genes that inhibit growth. In agreement with this model, Msn2p/Msn4p function is required for expression of a gene, YAK1, previously shown to antagonize PKA‐dependent growth. These results suggest that Msn2p/Msn4p‐dependent gene expression may account for all, or at least most, of the pleiotropic effects of yeast PKA, including growth regulation, response to stress and carbohydrate store accumulation.


All living cells employ mechanisms that allow survival under conditions of environmental stress. The best studied of these has been termed the heat shock response and involves the induction of heat shock gene transcription in response to temperature shift. In eukaryotes, temperature shift results in activation of the ubiquitous heat shock transcription factor (HSF) which recognizes specific elements (HSEs) in HSF‐dependent promoters (Wu et al., 1994; Mager and De Kruijff, 1995). The yeast Saccharomyces cerevisiae also employs a general, or global, stress response pathway that functions independently of HSF (and HSE) and is stimulated by a wide variety of stresses. Such stresses include temperature shift, osmotic shock, nutrient starvation, as well as DNA and oxidative damage (Mager and De Kruijff, 1995; Ruis and Schuller, 1995). The general stress response also differs from the heat shock response in that the spectrum of genes induced by these environmental stimuli is broad. Genes considered part of the general stress response include the cytosolic catalase T gene CTT1, the DNA damage‐responsive gene DDR2, the heat shock gene HSP12 and genes (TSP2 and GLC3) involved in carbon store accumulation (Mager and De Kruijff, 1995; Ruis and Schuller, 1995).

Although the precise mechanism of the general stress response pathway has not been elucidated, recent studies have implicated the related Zn2+‐finger transcription factors Msn2p and Msn4p (Martinez‐Pastor et al., 1996; Schmitt and McEntee, 1996) in this process. Strains lacking MSN2 and MSN4 are sensitive to various forms of stress and fail to accumulate stress‐regulated messages following heat and osmotic stress, as well as nutrient starvation and DNA damage (Martinez‐Pastor et al., 1996; Schmitt and McEntee, 1996). Moreover, Msn2p binds specifically to DNA sequences, referred to as the stress response element or STRE (Martinez‐Pastor et al., 1996; Schmitt and McEntee, 1996), previously shown to be necessary and sufficient to impart stress‐induced expression on a heterologous reporter gene (Marchler et al., 1993). One attractive hypothesis is that Msn2p and the STREs represent the ‘integration point’ of multiple signaling pathways that respond to diverse stresses and induce the same set of genes (Martinez‐Pastor et al., 1996). In support of this hypothesis, osmotic induction of the stress response genes is impaired by mutations that inactivate the high osmolarity glycerol response (HOG) pathway or Msn2p/Msn4p function. By contrast, heat shock‐ and nitrogen starvation‐dependent expression of the same genes is independent of the HOG pathway (Schuller et al., 1994; Martinez‐Pastor et al., 1996). However, the HOG pathway has not been linked directly to Msn2p‐dependent gene expression, and substantial Msn2p‐dependent osmotic induction remains in a hog1Δ background (Martinez‐Pastor et al., 1996). Moreover, the stress response signaling mechanism may be more complicated than first appreciated because STRE‐dependent transcription, as well as Msn2p nuclear localization, is negatively regulated by yeast cAMP‐dependent protein kinase (PKA) activity (Marchler et al., 1993; Gorner et al., 1998).

Yeast PKA activity has been implicated in numerous cellular processes, including growth, carbon storage, response to stress and differentiation (Cameron et al., 1988; Broach and Deschenes, 1990; Gimeno et al., 1992). PKA‐deficient cells arrest in G1, accumulate storage carbohydrates (glycogen and trehalose) and become resistant to heat and oxidative stress. These phenotypes are similar to the phenotypes displayed by wild‐type cells deprived of nutrients (Johnston et al., 1977). In contrast, cells with elevated PKA activity fail to store carbohydrate reserves and are exquisitely sensitive to various forms of stress (Cannon and Tatchell, 1987; Cameron et al., 1988). These phenotypes have been taken as evidence that the yeast PKA pathway plays a central role in coordinating cell growth and metabolism in response to environmental stimuli (Broach and Deschenes, 1990; Markwardt et al., 1995).

At least some of the physiological effects of PKA can be accounted for by changes in transcription. For example, PKA activity inhibits both STRE‐dependent expression of stress response genes (Belazzi et al., 1991; Engleberg et al., 1994; Varela et al., 1995) and expression of genes (GAC1, GSY2) essential to glycogen synthesis (Francois et al., 1992; Hardy et al., 1994). By contrast, the mechanism by which PKA affects yeast growth is unknown. Although ribosomal protein synthesis is positively regulated by PKA (Klein and Struhl, 1994; Neuman‐Silberberg et al., 1995), PKA effects on ribosomal protein synthesis cannot account for the PKA growth requirement (Neuman‐Silberberg et al., 1995).

One hint that PKA might regulate growth, stress response and glycogen accumulation by a common mechanism came from studies of the high‐copy suppressor SOK2 (Ward et al., 1995). Like PKA, Sok2p affected growth, sensitivity to stress and glycogen stores. Moreover, the effects of Sok2p on the latter two processes were associated with transcriptional repression. Thus, we suggested that Sok2p was a PKA‐dependent repressor that regulated expression of one or more genes antagonistic to cell growth and division (Ward et al., 1995). Indeed, one candidate for such a growth‐inhibitory function had been identified previously as a recessive suppressor of a PKA‐deficient strain (Garrett and Broach, 1989). Strains lacking the three, redundant PKA catalytic subunit genes (TPK1, TPK2 and TPK3) grew slowly if the YAK1 gene was also deleted (Garrett and Broach, 1989). However, Yak1p has not been assigned definitively to the PKA pathway (Hartley et al., 1994), and PKA effects on growth, stress and glycogen accumulation are much greater than can be accounted for by disruption or overexpression of SOK2 (Ward et al., 1995).

Here we describe the characterization of three, allelic suppressors of the tpk(Ts) growth defect. Although these suppressors are semi‐dominant in diploid analysis, they result from the loss of function of the stress‐related transcriptional activator, Msn2p (Martinez‐Pastor et al., 1996; Schmitt and McEntee, 1996). Interestingly, the growth defect of a tpkΔ deletion strain is alleviated by the inactivation of Msn2p and its functional homolog, Msn4p. This is the first indication that Msn2p and Msn4p proteins play a role in PKA‐dependent growth. Interestingly, YAK1 may be one of the growth‐related genes under Msn2p and Msn4p control because YAK1 expression is induced by PKA depletion only in strains containing Msn2p (and Msn4p) activity.


Dominant suppressors of the tpk(Ts) growth defect

Sixteen temperature‐resistant revertants of haploid tpk2‐63(Ts) strain SGY398 contained dominant (or semi‐dominant) suppressors as judged by diploid analysis (unpublished results). Three of the 16 suppressors (SOK3‐218, SOK3‐234 and SOK3‐244; suppressor of kinase) were weak, semi‐dominant, tightly linked (no recombinants in 10 tetrads each) and mapped within 20 cM of the tpk2‐63(Ts) high‐copy suppressor gene SOK2 (data not shown). Three point crosses placed the SOK3 alleles within 2.2 cM of the chromosome XIII gene MIH1, in the order MIH1SOK3MCM1 (Materials and methods; Figure 1). This corresponds to a physical distance of ∼6.6 kb (Olson, 1991).

Figure 1.

Genetic mapping of SOK3 (msn2). A bold line denotes the physical region surrounding the SOK3‐234 suppressor. Open boxes above the bold line depict labeled genes. Genetic distances measured in three point crosses between mih1::LEU2, SOK3‐234 (msn2‐234) and MCM1‐URA3 are indicated by lines with arrows at either end.

The dominant SOK3 suppressor results from the loss of MSN2 function

Despite varied approaches (random genomic libraries, PCR and gap repair of chromosomal DNA from the region predicted by genetic mapping to contain the SOK3 alleles), we were unable to clone any of the SOK3 alleles on the basis of their dominant suppression of the tpk2‐63(Ts) growth defect. One explanation for this result was that tpk2(Ts) suppression resulted from the loss of SOK3 function. In that scenario, the semi‐dominance of the spontaneous SOK3 suppressors could reflect the limiting nature of the Sok3 protein in diploids (haploinsufficiency). Genetic mapping data suggested that SOK3 might be identical to MSN2, a gene previously shown to encode a non‐essential protein involved in the general stress response (Estruch and Carlson, 1991; Martinez‐Pastor et al., 1996; Schmitt and McEntee, 1996). Interestingly, msn2Δ::HIS3 transformants of tpk2‐63(Ts) strain SGY446 grew at 33 and 35°C (Figure 2; see below). Growth was the result of the msn2Δ deletion because suppression was complemented by the low‐copy MSN2 plasmid pAS15, but not by either its gapped version (pAS16) or a derivative containing chromosomal DNA from the SOK3‐234 strain (pAS24) (data not shown). Thus, the tpk2‐63(Ts) growth defect is alleviated by loss of Msn2p function. Moreover, a tpk2‐63(Ts)/tpk2‐63(Ts) msn2Δ::HIS3/MSN2 diploid strain exhibited poor, but demonstrable, growth at the non‐permissive temperature of 35°C (data not shown). Thus, inactivation of a single allele of MSN2 is sufficient to partially alleviate the conditional growth of a tpk(Ts)/tpk(Ts) diploid. Finally, temperature‐resistant growth of the three dominant tpk2‐63(Ts) revertants (SOK3‐218, SOK3‐234 and SOK3‐244) was reversed by the MSN2 plasmid pAS15, consistent with the notion that, at least by plasmid complementation, the three suppressors are recessive. By these criteria, SOK3 is allelic with MSN2 and we have renamed the three alleles msn2‐218, msn2‐234 and msn2‐244 to reflect their recessive nature in the ‘plasmid’ complementation assay.

Figure 2.

Suppression of the tpk(Ts) growth defect by deletion of MSN2 and MSN4. Isogenic derivatives of the haploid tpk2‐63(Ts) strain were streaked onto rich medium agar and incubated at 23, 33 and 35°C. The strains were tpk2(Ts) [tpk2‐63(Ts), SGY446], tpk2(Ts) msn2Δ [tpk2‐63(Ts) msn2Δ::HIS3, ASY21], tpk2(Ts) msn2Δ msn4Δ [tpk2‐63(Ts) msn2Δ::HIS3 msn4Δ::LEU2, ASY29] and tpk2(Ts) msn4Δ [tpk2‐63(Ts) msn4Δ::LEU2, ASY30].

PKA activity is dispensable in a msn2Δ msn4Δ mutant

Although MSN2 accounts for the majority of STRE‐dependent transcription, the structural homolog, MSN4, also plays a significant role (Martinez‐Pastor et al., 1996; Schmitt and McEntee, 1996). To determine the contribution of both genes to PKA‐dependent growth, the tpk2‐63(Ts) strain SGY446 was transformed with msn2Δ::HIS3 and msn4Δ::LEU2 deletions (Materials and methods). As shown in Figure 2, the conditional growth defect was partially alleviated by either mutation alone (the msn4Δ mutation suppressed weakly even at 33°C); however, the tpk2‐63(Ts) strain lacking all Msn2p and Msn4p activity exhibited robust growth at the elevated temperature (Figure 2). Thus, both Msn2p and Msn4p antagonize PKAdependent growth.

To test whether deletion of MSN2, MSN4 or both would suppress the growth defect of a strain lacking all PKA activity, haploid strains ASY29 [MATα tpk1::ADE8 tpk2‐63(Ts) tpk3::TRP1 msn2Δ::HIS3 msn4Δ::LEU2] and MWY155 (MATa tpk1::URA3 tpk2::HIS3 TPK3 MSN2 MSN4) were mated and sporulated at 23°C. Because the msn2Δ and tpk2Δ deletions were both marked with HIS3, we focused on tetrads that segregated two His+ colonies and two His colonies (the His+ colonies in these tetrads could be assumed to contain both tpk2Δ::HIS3 and msn2Δ::HIS3). Eight of 37 tetrads formed two His+ colonies and two His colonies, and nine of the His+ colonies within this group were Trp+ (tpk3::TRP1). Interestingly, five of these His+ Trp+ colonies were Leu+ (tpk1 tpk2Δ::HIS3 tpk3::TRP1 msn2Δ::HIS3 msn4Δ::LEU2) (Figure 3) and formed colonies that were similar in size to the TPK3 strains (TPK3 MSN2 MSN4 and TPK3 msn2Δ msn4Δ) from the same cross (Figure 3). The other His+ colonies were Leu (tpk1 tpk2Δ::HIS3 tpk3::TRP1 msn2Δ::HIS3 MSN4) and formed slow‐growing colonies that were smaller than colonies of the tpkΔ yak1Δ strain, SGP406 (Figure 3). Thus, the tpkΔ growth defect is weakly suppressed by Msn2p inactivation and completely alleviated by loss of Msn2p and Msn4p function (the doubling times of strains ASY62 and ASY63 were <5% longer than the doubling times of ASY58, data not shown). Finally, a strain lacking PKA activity as a result of the deletion of the adenylate cyclase activator genes RAS1 and RAS2 (Broach and Deschenes, 1990) exhibited robust growth if MSN2 and MSN4 were deleted (data not shown).

Figure 3.

PKA is dispensable in a strain lacking Msn2p/Msn4p function. Strains were incubated on rich medium agar for 3 days at 30°C. All strains (with the exception of SGP406) were haploid segregants of the heterozygous diploid ASY48 [tpk1::URA3. tpk1::ADE8 tpk2‐63(Ts)/tpk2::HIS3 tpk3::TRP1. TPK3 msn2Δ::HIS3. MSN2 msn4Δ::LEU2. MSN4]. The strains were TPK3 MSN2 MSN4 [TPK3 MSN2 MSN4, ASY59], TPK3 msn2Δ msn4Δ [TPK3 msn2Δ::HIS3 msn4Δ::LEU2, ASY58], tpkΔ yak1Δ [tpkΔ yak1Δ::LEU2, SGP406], tpkΔ msn2Δ msn4Δ [tpkΔ msn2Δ::HIS3 msn4Δ::LEU2, ASY62] and tpkΔ msn2Δ MSN4 [tpkΔ msn2Δ::HIS3 MSN4, ASY61].

MSN2 overproduction exacerbates growth of a tpk(Ts) strain

Because the tpk(Ts) growth defect was suppressed by Msn2p inactivation, we determined if the growth defect was exacerbated by MSN2 overexpression. The tpk2‐63(Ts) strain SGY448 was transformed with the high‐copy MSN2 plasmid (pAS27), as well as the pRS305‐2μ vector control, and purified on selective agar at 23, 30 and 33°C. Even at the ‘permissive temperature’ of 23°C, the high‐copy MSN2 transformants formed much smaller colonies than the vector transformants (Figure 4). In contrast, the high‐copy MSN2 plasmid had little effect on the growth of the isogenic wild‐type TPK strain (Figure 7), consistent with previous observations (Estruch and Carlson, 1991). Thus, Msn2p overproduction exacerbates the growth of a PKA‐compromised strain. Similar, although less dramatic, results were observed when a high‐copy MSN4 plasmid was transformed into the tpk2‐63(Ts) strain (data not shown), consistent with the notion that MSN2 and MSN4 carry out similar functions.

Figure 4.

MSN2 overproduction exacerbates growth of a tpk2(Ts) strain. Strains containing the high‐copy MSN2 plasmid pAS27 (↑MSN2) or the control vector pRS305‐2μ were streaked onto selective minimal medium agar and incubated at 23, 30 or 33°C. The strains were tpk2(Ts) yak1Δ [tpk2‐63(Ts) yak1Δ::ADE8, SGY470] and tpk2(Ts) [tpk2‐63(Ts), SGY448].

The YAK1 gene previously was identified as a loss‐of‐function suppressor of defects in the PKA pathway (Garrett and Broach, 1989; Hartley et al., 1994). The epistasis relationship between Msn2p and Yak1p was tested by asking if the effect of MSN2 overexpression on tpk2‐63(Ts) growth required Yak1p function. As shown in Figure 4, deletion of YAK1 relieved the debilitating effect of the high‐copy MSN2 plasmid at 23°C, but was not completely epistatic to MSN2 overexpression at 30 or 33°C. A model in which Yak1p is on a pathway parallel with Msn2p most easily explains this result. However, it can be reconciled with a model in which Yak1p functions downstream of Msn2p as long as Yak1p is not the only Msn2p target affecting growth. This second model is at least consistent with the different strengths of tpkΔ suppression by mutations in MSN2 and MSN4 versus mutations in YAK1 (Figure 3).

Msn2p/Msn4p activity is necessary for expression of YAK1

The pleiotropic effects of mutations in MSN2 and MSN4 (Figure 3; Martinez‐Pastor et al., 1996; Schmitt and McEntee, 1996) can be contrasted with the more specific effects of YAK1 inactivation (Garrett and Broach, 1989; Garrett et al., 1991; Hartley et al., 1994). This difference is consistent with a model in which Msn2p and Msn4p are required for expression of genes involved in growth inhibition (YAK1) as well as sensitivity to stress (HSP12, CTT1 and DDR2). In such a model, modulating Msn2p and Msn4p activity would alter YAK1 expression. To determine if Msn2p and Msn4p regulate YAK1 expression, we measured β‐galactosidase activities of strains containing two different YAK1lacZ promoter fusion plasmids. The YAK1lacZ promoter fusions (UASYAK1–455 and UASYAK1–595) contain the translation start and putative TATA box of YAK1, as well as 134 and 274 bp upstream of the putative TATA box, respectively. Whereas the promoter‐less control plasmid (CYC1lacZ) was expressed inefficiently in all strains tested, both YAK1lacZ fusions were expressed efficiently in the tpk2(Ts) strain as judged by high β‐galactosidase activities (Figure 5). Interestingly, neither YAK1 fusion was expressed in the tpk2(Ts) strain lacking Msn2p and Msn4p function (Figure 5). This difference was not due simply to better growth of the tpk2(Ts) msn2Δ msn4Δ strain because both fusions were expressed in an isogenic TPK2 strain (Figure 5). Moreover, introduction of the bcy1Δ::LEU2 deletion into the TPK2 strain reduced YAK1lacZ expression, even though strains with elevated PKA activity grow more slowly than wild‐type strains. Finally, an unrelated promoter fusion RCElacZ (Cuevo et al., 1997) was not affected by PKA or Msn2p/Msn4p activity (data not shown).

Figure 5.

YAK1 regulation by PKA and Msn2p/Msn4p. β‐Galactosidase activities of strains containing the indicated promoter fusions on high‐copy plasmids. Each number represents the average of at least four independent measurements. Plasmids are described in Materials and methods and are based on the promoter‐less CYC1lacZ vector pJLB. The strains were tpk2(Ts) [tpk2‐63(Ts), SGY446], tpk2(Ts) msn2Δ msn4Δ [tpk2‐63(Ts) msn2Δ::HIS3 msn4Δ::LEU2, ASY29], TPK2 [TPK2, ASY18] and TPK2 bcy1Δ [TPK2 bcy1Δ::LEU2, ASY54].

To confirm that the effects of Msn2p and Msn4p on YAK1 expression were a reflection of in vivo function rather than an artifact of the YAK1lacZ promoter fusions, we monitored mRNA levels of YAK1, the glycogen metabolism gene GLC3 and the heat shock gene HSP12. As judged by Northern analyses, the mRNA levels of all three genes were greater in the tpk2(Ts) strain than in the TPK2 strain, and this increase was completely dependent upon Msn2p and Msn4p function (Figure 6A). mRNA levels of the control gene, ACT1, are unaffected by either PKA or Msn2p/Msn4p function (Martinez‐Pastor et al., 1996) and serve as loading controls. Quantification showed that YAK1, GLC3 and HSP12 expression was >3‐, 6‐ and 11‐fold greater, respectively, in the tpk2‐63(Ts) strain, SGY446, than in the isogenic TPK2 strain, ASY18 (Figure 6B). Moreover, deletion of MSN2 and MSN4 reduced expression of all three genes to levels that were below (YAK1) or similar to (GLC3 and HSP12) levels of the wild‐type TPK2 strain, ASY18 (Figure 6B). Thus, YAK1 expression is repressed by PKA and is dependent upon Msn2p/Msn4p activity, similarly to expression of the stress‐responsive genes HSP12 and CTT1 (Figure 6B; Martinez‐Pastor et al., 1996).

Figure 6.

Transcription regulation by PKA and Msn2p/Msn4p. (A) Total RNAs from strains tpk2(Ts) [tpk2‐63(Ts), SGY446], tpk2(Ts) msn2Δ msn4Δ [tpk2‐63(Ts) msn2Δ::HIS3 msn4Δ::LEU2, ASY29] and TPK2 [TPK2, ASY18] were probed sequentially with the indicated DNAs. (B) Quantitation of YAK1, GLC3 and HSP12 mRNA levels from (A).

Figure 7.

Msn2p control of glycogen accumulation. Strains were incubated on minimal medium agar for 4 days at 23°C and then exposed to iodine vapors as described (Garrett and Broach, 1989). The strains were tpk2(Ts) [tpk2‐63(Ts), SGY446], TPK3 bcy1Δ [TPK3 bcy1Δ::LEU2, MWY161 and MWY162], TPK2MSN2 [TPK2 pRS305‐2μ‐MSN2, ASY67], TPK2 [TPK2, ASY18] AND tpk2(Ts) msn2Δ msn4Δ [tpk2‐63(Ts) msn2Δ::HIS3 msn4Δ::LEU2, ASY29].

Msn2p and Msn4p regulate glycogen accumulation

PKA‐depleted strains accumulate elevated levels of glycogen, due in part to an increase in expression of genes involved in glycogen synthesis (Figure 6; Francois et al., 1992; Ward et al., 1995). Because induction of at least one (GLC3) of those genes was dependent upon Msn2p/Msn4p function (Figure 6), we measured glycogen accumulation by inverting colonies over iodine vapors. Colonies of the tpk2(Ts) strain SGY446 stained much darker than colonies of the tpk2(Ts) msn2Δ::HIS3 msn4Δ::LEU2 strain ASY29, indicating that glycogen accumulation was dependent upon Msn2p/Msn4p function (Figure 7). Indeed, by this measure, the tpk2(Ts) msn2Δ::HIS3 msn4Δ::LEU2 strain accumulated as little glycogen as the strain (TPK3 bcy1Δ::LEU2) with elevated PKA activity (Figure 7), suggesting that Msn2p and Msn4p account for all of the effects of PKA on glycogen accumulation.

Diploid strains containing elevated PKA activity fail to sporulate (Broach and Deschenes, 1990). To examine if this effect can also be accounted for by PKA regulation of Msn2p/Msn4p‐dependent gene expression, we determined the sporulation frequencies of diploid strains containing different levels of Msn2p activity. The homozygous msn2Δ/msn2Δ msn4Δ/msn4Δ diploid strain SGY584 was transformed with the low‐copy MSN2 plasmid pAS15 or the vector control (pRS316) and placed in sporulation medium at 30°C. After 3 days, four‐spored asci formed 16 and 15% of the population of cells from the pAS15‐ and pRS316‐containing strains, respectively. Thus, neither Msn2p nor Msn4p are required for sporulation, suggesting that PKA does not seem to inhibit sporulation by inactivating Msn2p/Msn4p‐dependent transcription.

Msn2p and Sok2p both contribute to PKA‐dependent growth regulation

The SOK2 gene was isolated previously as a pleiotropic high‐copy suppressor of the tpk2‐63(Ts) growth defect (Ward et al., 1995). To determine the relative contribution and order of function of Sok2p and Msn2p in PKA‐dependent growth, we asked if SOK2 overexpression or deletion affected the growth of tpk2(Ts) strains containing various levels of Msn2p/Msn4p activity. SOK2 overexpression enhanced growth of the tpk2(Ts) msn2Δ MSN4 strain (Table I); however, this could be explained by an effect on Msn4p function, rather than an effect on a separate pathway. Because the tpk2(Ts) msn2Δ msn4Δ strain did not exhibit a growth defect (Figures 2 and 3), it was not surprising that SOK2 overexpression had no effect on growth of a strain lacking all Msn2p/Msn4p activity (Table I). Nevertheless, deletion of SOK2 compromised growth of the tpk2(Ts) msn2Δ msn4Δ strain (Table I), consistent with the notion that Sok2p and Msn2p/Msn4p affect PKA‐dependent growth by separate, although perhaps related, mechanisms.

View this table:
Table 1. Tests of epistasis with SOK2 and MSN2


Previous reports suggested that the PKA effects on glycogen accumulation and stress response resulted from transcription repression (Broach and Deschenes, 1990; Ruis and Schuller, 1995), although the precise mechanism was not known. This report shows that transcription repression can also explain the growth requirement for PKA. PKA activity is largely dispensable in a strain lacking the transcription activators Msn2p and Msn4p (Figure 4), suggesting that the single essential function of PKA is to inhibit expression or function of growth‐inhibitory genes under Msn2p/Msn4p control (Figure 8). Msn2p and Msn4p were identified recently as the major transcription factors of the yeast multistress response (Martinez‐Pastor et al., 1996; Schmitt and McEntee, 1996), and our results show that they play a significant role in the expression of genes involved in glycogen metabolism (Figure 7). Because both PKA and Msn2p/Msn4p regulate these processes at the level of transcription, it seems likely that their opposing effects on growth also reflect antagonistic roles in gene expression. If that model is correct, it suggests that PKA is an integral part of the stress response, which normally involves transient arrest, increased resistance to stress and accumulation of carbon stores (Werner‐Washburne et al., 1993; Ruis and Schuller, 1995). Such a model would also fit well with the exquisite stress sensitivity displayed by strains with elevated PKA activity (Broach and Deschenes, 1990).

Figure 8.

Model of PKA regulation of genes involved in growth, stress response and glycogen accumulation.

One prediction of the model shown in Figure 8 is that genes that inhibit PKA‐dependent growth should be repressed by PKA and induced by Msn2p/Msn4p. The YAK1 gene was identified originally as a recessive suppressor of the PKA‐deficient growth defect (Garrett and Broach, 1989) and thus behaves as one would expect for an inhibitor of PKA‐dependent growth (Hartley et al., 1994). Consistent with the model, YAK1 expression is inversely correlated with PKA activity (Figures 5 and 6). This result provides a molecular explanation for the observation that Yak1p protein (and activity) is elevated in PKA‐depleted strains (Garrett et al., 1991). In addition, YAK1 expression is absolutely dependent upon Msn2p/Msn4p function, consistent with the presence of three, closely spaced STRE sites within the 150 bp region upstream of the putative YAK1 TATA box (Garrett and Broach, 1989). Finally, the model is consistent with the observation that an increase in YAK1 expression restricts growth of a PKA‐compromised strain (Garrett et al., 1991; Hartley et al., 1994). Thus, it seems likely that Yak1p activation, by PKA depletion and Msn2p/Msn4p activation, inhibits a cellular process that is essential for growth. Although the nature of that process is unknown, these results should provide additional impetus for further investigation into the physiological and biochemical role of the Yak1p protein kinase.

Msn2p/Msn4p‐dependent YAK1 expression cannot explain all of the growth defects of a PKA‐deficient strain. In contrast to the robust growth of a tpkΔ msn2Δ msn4Δ strain (Figure 3), the tpkΔ yak1Δ strain grows slowly (Figure 3; Garrett and Broach, 1989; Thompson‐Jaeger et al., 1992). Moreover, an increase in Yak1p activity is not sufficient to inhibit growth of a wild‐type strain (Garrett et al., 1991; Hartley et al., 1994), suggesting that one or more additional genes must be derepressed to restrict growth. Finally, YAK1 function contributes to, but is not essential for, the toxic effect of MSN2 overexpression on tpk(Ts) growth (Figure 4). Together, these results suggest that YAK1 is not the only PKA‐regulated gene that can inhibit growth. One attractive candidate gene encodes the stress‐induced transcriptional repressor, Xbp1p (Mai and Breeden, 1997). The XBP1 gene contains several STRE consensus sites and is expressed in a manner expected of Msn2p/Msn4p‐regulated genes (Mai and Breeden, 1997). Moreover, XBP1 overexpression results in a slow‐growth phenotype, lengthening of G1 and repression of G1 cyclin expression (Mai and Breeden, 1997), phenotypes consistent with a factor involved in PKA‐dependent growth (Broach and Deschenes, 1990). It remains to be seen if XBP1 and YAK1 can together account for the growth defect of a tpkΔ strain.

The model presented in Figure 8 suggests that PKA blocks Msn2p‐dependent expression through a direct effect on Msn2p function. In support of this interpretation, Gorner and collegues recently have shown that the nuclear localization of Msn2p is correlated inversely with PKA activity (Gorner et al., 1998). Although there is as yet no evidence that PKA phosphorylates Msn2p directly, these results suggest that PKA exerts its effect on Msn2p‐dependent gene expression by directly inhibiting Msn2p function. Interestingly, nuclear localization of Msn2p is also induced rapidly by conditions of stress, suggesting that PKA activity is modulated by exposure to stress or that other signaling pathways also impinge upon Msn2p function. It is interesting to note that other protein kinase activities have been implicated in the stress response, including Hog1p (Schuller et al., 1995; Varela et al., 1995; although see Martinez‐Pastor et al., 1996), Pho85p (Timblin and Bergman, 1997) and Snf1p (Thompson‐Jaeger et al., 1991; Hubbard et al., 1992). In certain respects, Snf1p is a prime candidate for a regulator of Msn2p/Msn4p function because it affects glycogen accumulation (Thompson‐Jaeger et al., 1992) and MSN2 was first identified as a high‐copy suppressor of the carbohydrate utilization defect of an snf1(Ts) strain (Estruch and Carlson, 1992). However, Snf1p is not required for stress induction of STRE‐dependent genes (Martinez‐Pastor et al., 1996). Thus, it will be interesting to determine the contribution of each of these pathways to Msn2p translocation and function. Finally, it is important to note that the Sok2p repressor protein also regulates at least a subset of the genes under PKA and Msn2p/Msn4p control (Ward et al., 1995). Thus, PKA might regulate gene expression through its actions on the repressor Sok2p as well as by its modulation of Msn2p/Msn4p function (Figure 8).

Msn2p and Msn4p originally were identified as transcriptional activators of stress response genes; however, Ruis and Schuller (1995) pointed out the presence of STREs within regions upstream of a number of genes involved in the synthesis of glycogen (and trehalose). Several of those genes were known, or suspected, to be repressed by PKA activity (Francois et al., 1992; Hardy et al., 1994; Ward et al., 1995), so it was assumed that those genes, and as a result glycogen synthesis, would also be regulated by Msn2p and Msn4p. Consistent with that hypothesis, inactivation of MSN2 and MSN4 restricts the expression (Figure 6) and function (Figure 7) of at least one gene (GLC3) involved in the synthesis of glycogen. Schmitt and McEntee (1996) have observed similar effects of Msn2p inactivation on stress‐induced expression of the trehalose synthesis gene TPS2.

Materials and methods

Media and growth conditions

Media used, including yeast rich and minimal media, as well as bacterial media, were prepared as described previously (Ward et al., 1995). Glycogen accumulation and sporulation frequency were monitored by methods described previously (Garrett and Broach, 1989).

Strains and plasmids

Yeast strains are listed in Table II. Bacterial strains MC1066 [Δ(lac. X74 galU galK strAr hsdR trpC9830 leuB6 pyrF::Tn5] and DH5α {F′/endA hsdR17(rK mK) supE44 thi‐1 recA1 gyrA relA1 Δ(lacZYA‐argF. U169 [φ80dlacΔ(lacZ. M15]} have been described (Casadaban et al., 1983; Woodcock et al., 1989). The bacterial vector pBSKII+ has been described (Stratagene Product catalog). Plasmid pRS316 is a low‐copy URA3 yeast vector and has been described (Sikorski and Hieter, 1989). The high‐copy LEU2 yeast vector pRS305‐2μ has been described (Ward et al., 1995). The high‐copy, URA3‐based plasmid (pJLB; Finley et al., 1990) containing a UAS‐less cytochrome c (CYC1) promoter upstream of the lacZ gene was a kind gift from Jon Horowitz (North Carolina State University). The pNUT1a derivative of pJLB1 contains an octameric p180‐binding site upstream of the CYC1 promoter and has been described (Cuevo et al., 1997). The yak1::ADE8 disruption has been described (Ward and Garrett, 1994). The Yip5 (URA3)‐based MCM1 plasmid pGC155‐2 (Christ and Tye, 1991) was digested with SphI to mark MCM1. The MIH1 gene was disrupted with mih1::LEU2 by digesting plasmid pDLB337 with NheI as described (Russell et al., 1989). Plasmid pGS127 contains the YAK1 coding region on the high‐copy URA3 plasmid YEp24 (Garrett and Broach, 1989), and the high‐copy SOK2 plasmid pMW61 (YEp‐SOK2) has been described (Ward et al., 1995).

View this table:
Table 2. List of yeast strains

DNA manipulations

All DNA manipulations were done according to published procedures (Kaiser et al., 1994) or according to the recommendations of the manufacturers. The low‐copy MSN2 plasmid (pAS15) was constructed by inserting the 4.4 kb HindIII fragment from pGS222 (YCp50‐MIH1 MSN2 LYS7 SUB1) into the HindIII site of pRS316. The related msn2Δ and msn2Δ::HIS3 deletion plasmids, pAS16 and pAS19, were constructed by digesting pAS15 with BglII and religating or inserting the 1.7 kb BamHI HIS3 fragment, respectively. Chromosomal integration of the msn2Δ::HIS3 deletion was achieved by transforming the indicated strains with pAS19 that had been digested with EcoRV and ClaI. A gap‐repaired derivative of pAS16 was constructed by transforming sok3‐234 strain SGY455 with BglII‐digested pAS16. The gap‐repaired plasmid, pAS24, contained the MSN2 region as judged by size and restriction pattern. The high‐copy MSN2 plasmid, pAS27, consists of the 2.9 kb HindIII–EagI fragment MSN2 inserted into the corresponding sites of pRS305‐2μ. The MSN4 plasmid pAS25 was generated by cloning the 3.2 kb BamHI–EcoRI MSN4 fragment (from PCR‐amplified chromosomal DNA) into the corresponding sites of pBSKII+. The high‐copy MSN4 plasmid (pAS28) was constructed by placing the 3.3 kb BamHI–XhoI fragment of pAS25 into the same sites of pRS305‐2μ. An msn4Δ::LEU2 disruption was created by inserting the 2.8 kb LEU2 fragment of YEp13 into the PstI–XmnI sites of pAS25 to create plasmid pAS26. Plasmid pAS26 was digested with BamHI and HindIII to liberate the msn4Δ::LEU2 deletion for transplacement.

The YAK1lacZ fusion plasmids (pDS455 and pDS595) were constructed in two steps. Plasmid pDS1 is a derivative of pJLB1 in which the 300 bp BamHI fragment (containing the translation start site and TATA sequence of CYC1) was removed and replaced with an 8 bp XhoI linker after blunting with Klenow fragment. Plasmids pDS455 and pDS595 were constructed by ligating XhoI linkers to the blunt ends of 455 bp SnaBI–StuI and 595 bp SnaBI–SnaBI fragments of YAK1 and inserting those fragments in the proper orientation into the XhoI site of pDS1.

Genetic mapping

The SOK3‐234 mutation was genetically mapped by crossing a mih1::LEU2 derivative of SGY455 [tpk2‐63(Ts) SOK3‐234] with a MCM1‐URA3‐marked derivative of strain SGY446 [tpk2‐63(Ts)]. The results of those crosses (SOK3‐234×MIH1‐mih1::LEU2 = 109 PD, 0 NPD, 5 T), (SOK3‐234×MCM1‐URA3 = 96 PD, 0 NPD, 18 T), and (MIH1‐mih1::LEU2×MCM1‐URA3 = 90 PD, 0 NPD, 24 T) were used to determine the map distance based on the formula of Perkins (1949).

β‐Galactosidase assays

β‐Galactosidase activity was measured according to a published procedure (Kaiser et al., 1994).

Northern (RNA) analyses

Total RNA was isolated by the LETS/phenol/chloroform extraction method (Kaiser et al., 1994) from cells grown at 30°C to an OD600 of between 0.5 and 1.0, separated by electrophoresis and transferred to nylon membranes. DNA fragments to be used as probes (YAK1, 548 bp XbaI–EcoRV fragment from pGS127; HSP12, a 358 bp PCR fragment; GLC3, a 612 bp PCR fragment; and ACT1, a 320 bp BglII fragment from pSP72‐ACT1) were gel purified and then radiolabeled by random priming. Radioactivity was quantified on a Molecular Dynamics PhosphorImager.


The authors thank J.Posey, D.Seo and D.Cortez for preliminary observations, as well as M.Brandriss for critical comments on an early version of the manuscript. D.Mitchell, G.Sprague, J.Fassler and D.Lew generously provided plasmids. This work was supported by grant GM44666 from the National Institutes of Health.


View Abstract