|Home | About | Journals | Submit | Contact Us | Français|
The MSI1 (CAC3) gene of Saccharomyces cerevisiae has been implicated in diverse cellular functions including suppression of the RAS/cAMP/Protein Kinase A signaling pathway, chromatin assembly and transcriptional co-repression. Seeking to identify the molecular mechanisms by which Msi1p carries out these distinct activities, a novel genetic interaction was uncovered with YAK1, which encodes a kinase that antagonizes the RAS/cAMP pathway. MSI1 was capable of efficiently suppressing the heat shock sensitivity caused by deletion of yak1. Surprisingly, the YAK1 gene is required for Msi1p to associate with Cac1p in the yeast two-hybrid system. A new activity of Msi1p was identified: the ability to activate transcription of a reporter gene when tethered near the promoter, but only in the absence of fermentable carbon sources. This transcriptional activation function was substantially diminished by the loss of YAK1. Furthermore, MSI1 influences YAK1 function; overexpression of YAK1 decreased the growth rate, but only in the presence of a functional MSI1 gene. Finally, it is shown that YAK1 antagonizes nuclear accumulation of Msi1p in nonfermenting cells. Taken together, these data demonstrate a novel interaction between Msi1p and Yak1p in which each protein influences the activity of the other.
Growth of the budding yeast Saccharomyces cerevisiae is coordinated by multiple signal transduction pathways that are sensitive to nutrient availability. Among these, the RAS/cAMP pathway is a major mechanism by which yeast respond to environmental glucose. Glucose-dependent activation of a Ras protein stimulates adenylyl cyclase, leading to the accumulation of cAMP and activation of Protein Kinase A (PKA).1 Activation of the RAS/cAMP pathway leads to a broad suite of responses that includes increases in glycolytic and growth rates and a concomitant decrease in storage carbohydrate concentrations and resistance to environmental stresses. This pathway is essential as these yeasts must retain at least partial activity for the survival and growth. Despite considerable investigation, the molecular mechanisms by which the RAS/cAMP pathway regulates growth in response to carbon source remain to be elucidated. In this paper, we present new data that expands our knowledge of the interdependent regulation of this pathway by the products of the MSI1 and YAK1 genes.
YAK1 was initially identified as a gene whose disruption suppressed the lethal phenotype associated with either the loss of RAS function or the loss of the catalytic subunit of PKA.2 Yak1p is a kinase, which acts as a negative regulator of overall growth and activates many of the same processes that are affected by PKA, including heat shock resistance 3 and pseudohyphal growth.4 Yak1p can be phosphorylated by PKA directly in vitro5 and is transcriptionally activated indirectly by PKA6indicating that Yak1p functions downstream of PKA. In cells with decreased RAS/cAMP pathway activity, excess copies of YAK1 leads to growth arrest5 confirming the role of this gene as a growth inhibitor.
In addition to being affected by PKA, Yak1p has also been implicated as a regulator of PKA function in what is thought to be an autoregulatory loop. The presence of glucose is sensed by yeast cells and leads to activation of PKA, whereas growth on nonfermentable carbon sources such as glycerol or ethanol do not activate the kinase.1 Griffioen et al. demonstrated that the phosphorylation and subcellular localization of Bcy1p (the regulatory subunit of PKA) is regulated by the carbon source available to the cell. Bcy1p is mostly nuclear in glucose-grown cells but found in both the nucleus and cytoplasm in ethanol-grown cells in a YAK1- and ZDS1-dependent process. The localization is a direct result of Yak1p-dependent phosphorylation of Bcy1p, where the phosphoprotein is either actively exported from the nucleus or retained in the cytoplasm by associating with Zds1p.7 Surprisingly, phosphorylation by Yak1p and relocalization of the regulatory subunit does not seem to have a substantial effect on many PKA functions but is hypothesized to redirect PKA to a different set of substrates.8 Furthermore, the subcellular localization of the Yak1 protein itself is carbon source-dependent, as it accumulates in the nucleus in the absence of glucose.9 Yak1p is an essential participant in at least one other nutritional signaling pathway beyond the RAS/cAMP pathway. The TOR pathway is conserved from yeast to humans and is active in well-fed cells but becomes blocked by either amino-acid starvation or the drug rapamycin. Inhibition of this pathway in yeast leads to a decrease in translation initiation and cell-cycle arrest10. Blocking the TOR pathway leads to the nuclear accumulation of Yak1p11 and activates Yak1p to phosphorylate the Crf1p co-repressor allowing its nuclear retention as well. The Crf1 protein has recently been shown to interact with the Fhl1p transcription factor and inhibits the transcription of ribosomal protein genes.12 Viewed together, these data clearly demonstrate that the Yak1p kinase plays an important role in nutritional signaling through both the RAS/cAMP and TOR pathways.
The MSI1 gene from yeast was originally identified as a high-copy suppressor of activated RAS/cAMP signaling pathway.13 Overexpression of this gene was found to efficiently suppress heat shock sensitivity and defective sporulation phenotypes associated with the constitutively activated RAS2G19V allele.13 All eukaryotic genomes examined to date contain at least one MSI1 homolog, but to date no homolog has been identified in prokaryotes.14 Most genes in this family encode proteins with seven WD40 repeat domains that are thought to form a flat disk consisting of seven β-propeller folds which have been proposed to mediate a variety of protein-protein interactions.15 The yeast Msi1p also contains a 57-residue amino-terminal extension not found in most other members of the family. The human genome encodes two Msi1p homologs, RbAp46 and RbAp48, and the overexpression of either of these proteins can substitute for Msi1p and suppress an activated RAS/cAMP pathway in yeast.16 More recently, it has been demonstrated that overexpression of Msi1p does not alter the quantity, subcellular localization or phosphorylation state of Ras2p.17 The ability of Msi1p to suppress the RAS/cAMP pathway is fully dependent on the presence of Bcy1p, yet surprisingly does not change cAMP concentrations or PKA activity.18 Furthermore, this suppression is dependent upon the physical interaction of Msi1p with the TOR pathway-regulated kinase Npr1p.17 The precise mechanism(s) by which Msi1p suppresses the RAS/cAMP pathway in yeast remains unclear.
Considering that the human Ras proteins are key players in oncogenesis,19 proteins that suppress their activity are of great interest. In yeast, Ras proteins activate PKA by stimulating cAMP synthesis.1 However, in higher eukaryotes, Ras proteins do not signal through cAMP but rather work through multiple effectors including serine/threonine kinases, phosphoinositide 3-kinases and phospholipases.19 Thus, Msi1p homologs would not necessarily be expected to suppress Ras signal transduction in animals. However, the C. elegans homolog of Msi1p (lin-53) has been shown to functionally antagonize the Ras signaling pathway in vivo.20, 21 Overexpression of RbAp48 in human cells correlates with the dephosphorylation of Akt, which is downstream of a phosphoinositide 3-kinase and Ras.22 Furthermore, overexpression of RbAp48 has recently been shown to induce p53-mediated apoptosis in mammalian exocrine cells.23 Therefore, Msi1p homologs may indeed down-regulate Ras signaling in higher organisms as well as yeast.
Surprisingly, Msi1p has activities beyond the suppression of RAS/cAMP signaling. Msi1p/Cac3p was isolated as one of three subunits of the Chromatin Assembly Factor-I (CAF-I), along with Cac1p and Cac2p.24 The CAF-I complex is conserved from yeast to humans and assembles histones H3 and H4 onto newly replicated DNA.24 In vertebrates, lack of CAF-I activity causes checkpoint activation, S phase arrest and eventually leads to programmed cell death.25-27 In contrast, CAF-I function is not essential for viability in S. cerevisiae 24, 28 but its absence leads to significantly abnormal chromatin and the misregulation of many functions. In yeast, the change in chromatin structure leads to a variety of phenotypes, including increased sensitivity to UV radiation,24 defects in transcriptional silencing, 24, 28 retrotransposition,29 and kinetochore function.30 Many of the phenotypes associated with the loss of CAF-I function are relatively mild. The additional mutation of the HIR genes, ASF1, MEC1, APC5 or the SAS genes lead to a synergistic enhancement of many of these phenotypes. 29-34 Importantly, deletion of either CAC1 or CAC2 does not impair the ability of Msi1p to suppress the activated RAS/cAMP pathway,17, 18 indicating that Msi1p has at least two separable functions.
Yet a third function was demonstrated by Kennedy et al. who showed that Msi1p and the histone deacetylase Rpd3p are essential for the ability of the retinoblastoma susceptibility gene product (pRB) to repress transcription in yeast.35 In higher eukaryotes, pRB is a well-characterized tumor suppressor protein that helps maintain the quiescent cell state. pRB physically interacts with the E2F family of transcription factors and represses transcription from the associated promoters in G0 phase.36 The requirement of Msi1p as a co-repressor is clearly conserved in Drosophila, as the Msi1p homolog p55 binds to pRB homologs, localizes to regulated promoters and is needed for the correct regulation of a subset of pRB-regulated genes.37 Similarly, RbAp48 has been shown to bind pRB in vitro and in vivo 16 and to co-purify with multiple histone deacetylases.38, 39 Microinjection of anti-RbAp48 antibodies into murine cells increases the activity of an E2F-driven reporter gene, consistent with the idea that RbAp48 is acting as a co-repressor with pRB.39 Taken together, these data suggest a model in which pRB acts by binding to Msi1p/RbAp48, which in turn associates with one or more histone deacetylases to repress transcription.35 Significantly, the ability of Msi1p to act as a co-repressor for pRB is independent of Cac2p,35 suggesting that this is likely a third function for Msi1p.
The mechanism(s) by which Msi1p suppresses RAS/cAMP signaling and the molecular basis for the multiple functions of this protein remain obscure. To better understand Msi1p function, we have investigated the interaction of MSI1 with YAK1. In this paper, we present new data that shows a mutual dependence between these two genes for many, but not all, of their phenotypes.
Overexpression of MSI1 suppresses dominant RAS2 mutations that constitutively activate the RAS/cAMP pathway but does not suppress activation of the pathway by deletion of BCY1. This observation has been interpreted to mean that Msi1p acts between Ras and PKA.13 However, the more recent observation by Zhu et al. that MSI1 overexpression does not change cAMP concentration or PKA activity and that the ability of MSI1 to suppress this pathway is dependent upon an intact BCY1 gene18 indicates that the effect of MSI1 upon the RAS/cAMP pathway is more complex. To better understand how MSI1 affects the RAS/cAMP pathway, we sought to determine how MSI1 and YAK1 interact. Deletion of the yak1 gene causes a variety of phenotypes that are highly similar to an activated RAS/cAMP pathway, including sensitivity to transient heat shock.3 Additionally, loss of yak1 synergizes with the constitutively activated RAS2G19V allele, causing enhanced sensitivity.3 We measured the ability of MSI1 to suppress the heat shock sensitivity of yak1Δ mutants alone or in combination with an activated RAS2 allele. We found that excess copies of Msi1p efficiently suppressed the heat shock sensitivity of either RAS2G19V or yak1Δ mutants, as well as the hypersensitivity of cells carrying both of these genetic alterations (Fig. 1). Overexpression of Msi1p has been proposed to suppress the RAS/cAMP pathway by sequestering the Npr1p kinase.17 In agreement with that model, deletion of NPR1 is also able to suppress the heat shock sensitivey caused by an activated RAS2 allele or deletion of YAK1 (Fig. 1). Considering that YAK1 has been shown to act downstream of PKA, 3, 6 our data indicate that Msi1p can suppress the RAS/cAMP pathway at points beyond PKA.
Having established this connection between MSI1 and YAK1 in the regulation of the RAS/cAMP pathway, we investigated whether YAK1 is important for other known functions of Msi1p, beginning with its ability to physically associate with Cac1p and Cac2p to form CAF-I.24 To investigate this function, we utilized the previous observation that Msi1p and Cac1p interact in the yeast two-hybrid system,17 enabling efficient transcription of the HIS3 reporter gene and allowing the cells to grow in the absence of histidine and in the presence of the His3p competitive inhibitor 3-aminotriazole (3-AT). Considering that neither CAC1 nor CAC2 is needed for MSI1 to suppress heat shock sensitivity caused by RAS2G19V, 17, 18 we hypothesized that loss of yak1 or activation of RAS2 would not affect the formation of CAF-I. Surprisingly, we found that an intact YAK1 gene was essential for Msi1p/Cac1p two-hybrid association (Fig. 2). Loss of YAK1 does not simply interfere with the two-hybrid reporter system as a fragment of the murine p53 protein and SV40 large T antigen still interacted efficiently in the yak1Δ strain. Additionally, the yak1Δ modification does not interfere with all Msi1p associations by simply destabilizing the protein, as Msi1p was still able to bind efficiently to Npr1p (Fig. 2). Furthermore, Cac1p and Cac2p bind to each other in the two-hybrid system and this binding was independent of both the MSI1 and YAK1 genes. A robust two-hybrid interaction was not detected between Msi1p and Cac2p (data not shown) which is consistent with the idea that the two smaller subunits of CAF-I (i.e., Cac2p and Msi1p) bind only to the largest subunit (Cac1p) and do not directly bind to each other. Since loss of YAK1 leads to the activation of the RAS/cAMP pathway, we next examined if the loss of Msi1p/Cac1p was due to the activation of this pathway or was specific to yak1Δ. The activated RAS2G19V allele had no effect on the association of Msi1p and Cac1p. Overexpression of the PDE2 gene encoding the high-affinity phosphodiesterase capable of degrading cAMP suppresses the RAS/cAMP pathway40 but also had no effect on the Msi1p/Cac1p interaction (Fig. 2). Therefore, loss of the YAK1 gene can prevent only Msi1p from associating with the CAF-I complex and need not lead to the general dissociation of the complex. Although the effect that MSI1 has on the RAS/cAMP pathway has been established for more than fifteen years,13 this is the first indication that the Yak1 kinase from this pathway can have an effect on the chromatin assembly complex.
Knowing that the RAS/cAMP pathway is principally activated by the presence of glucose,1 we attempted to extend these observations by altering the carbon source present and assessing its effect on the two-hybrid interaction. To our surprise, we found that Msi1p alone activated transcription of the HIS3 reporter gene (and thus resistance to 3-AT) when the cells were grown using glycerol as a carbon source (Fig. 3a). We refer to this phenomenon as the carbon-source regulated one-hybrid activity of Msi1p. Other proteins fused to the Gal4p DNA binding domain (DB), including Act1p from S. cerevisiae, lamin C from Homo sapiens, a p53 fragment from Mus musculus or RscS from Vibrio fischeri did not show a one-hybrid activity under identical conditions (data not shown). Neither DB-Cac1p nor DB-Cac2p showed a one-hybrid activity (data not shown) indicating that the transcriptional activation is not due to the recruitment of CAF-I to the promoter. Msi1p showed a regulated one-hybrid activity in either the PJ694 or HF7c reporter strain (data not shown).
To characterize this carbon-source regulated one-hybrid activity, fifteen different carbon sources were tested for their ability to induce one-hybrid activity when they were the sole carbon source in the media. All carbon sources either fully induced the one-hybrid activity or were fully noninducing; no intermediate inductions were observed. We found that the one-hybrid activity of Msi1p was robustly induced by ten different carbon sources, all of which are metabolized nonfermentatively (Fig. 3b). They include the two- and three-carbon molecules glycerol, acetate, ethanol, pyruvate and lactate, which are generally catabolized via the Krebs cycle and thus consumed nonfermentatively. The monosaccharide ribose and the disaccharides cellobiose and trehalose are also inducing carbon sources. Ribose cannot be catabolized by fermentation.41 Both trehalose and cellobiose are glucose dimers, joined by different linkages. Despite the fact that the constituent monosaccharide supports fermentative growth, trehalose42 and cellobiose43 are consumed nonfermentatively by several yeast species, a phenomenon known as the Kluyver effect. We confirmed that this specific strain of S. cerevisiae metabolizes ribose, trehalose and cellobiose strictly nonfermentatively by showing that respiration-deficient (ρ0) mutants44 failed to grow when these sugars were the sole carbon source (data not shown). Finally, both myristate (a saturated fatty acid) and oleate (an unsaturated fatty acid) induced a one-hybrid response.
Noninducing carbon sources include the monosaccharides glucose and fructose, the disaccharides maltose and sucrose, and the trisaccharide raffinose. All of these carbon sources support alcoholic fermentation in Saccharomyces cerevisiae. Exposing cells to both 2% glucose and 2% glycerol simultaneously did not induce one-hybrid activity (data not shown) presumably because the cells attempt to ferment the glucose first but can’t grow on this media without activation of the reporter gene. Providing a sub-optimal amount of glucose (0.2% instead of the usual 2%) also did not induce a one-hybrid activity (data not shown), suggesting that the one-hybrid activity did not result from simple nutrient limitation or slower growth. Other stressful conditions, including elevated temperature, high osmolarity, limiting concentrations of nitrogen or the inclusion of the TOR-inactivating drug rapamycin, or histone deacetylase inhibitors trichostatin A or sodium butyrate did not induce the one-hybrid activity in glucose nor block it in glycerol (data not shown). Under the conditions tested thus far, the correlation between one-hybrid activity of Msi1p and nonfermentative growth is perfect (Fig. 3b).
The carbon-source regulated one-hybrid activity was unaffected by the deletion of CAC1, NPR1 or ZDS1, the RAS2G19V mutation or overexpression of either PDE2, NPR1 or YAK1 (Fig. 3c and data not shown). However, when the YAK1 gene was deleted, we observed a reproducible diminishment of the regulated one-hybrid activity (Fig. 3a). The strength of the one-hybrid activity was reduced approximately 100-fold and the resulting colonies were smaller than the wildtype controls. The npr1Δ yak1Δ double mutant had no additional effect on the one-hybrid activity (data not shown). The effect of YAK1 is specific to Msi1p, because the deletion of YAK1 had no effect on activation of the reporter gene by an intact Gal4p (Fig. 3a). Therefore, the presence of the Yak1 kinase is essential for full induction of the regulated one-hybrid activity of Msi1p.
The human tumor suppressor protein pRB is one of the best-understood negative regulators of cell cycle progression in mammals. pRB indirectly associates with DNA via additional proteins and acts as a transcriptional repressor during the G0/G1 phases of the cell cycle. At least three distinct mechanisms of repression have been demonstrated, including the recruitment of histone deacetylases.36 Some reporter strains of Saccharomyces cerevisiae show leaky transcription of the HIS3 reporter leading to a baseline level of 3-AT resistance; pRB fused to the Gal4 DB (DB-pRB) can represses this leaky transcription.35 Significantly, full repression by pRB requires an intact MSI1 gene.35 Presumably, this reflects the demonstrated ability of pRB to associate with the human Msi1p homologs RbAp46 and RbAp48,16 which in turn associate with at least one histone deacetylase.45 The co-repressor phenotype seems to be less dramatic in our hands than the one published by Kennedy et al., but we did find a reproducible difference in resistance that is dependent upon MSI1. We then sought to determine if YAK1 influenced the ability of Msi1p to act as a co-repressor for pRB. Deletion of the YAK1 gene had no effect on transcriptional repression by pRB either in the absence or presence of MSI1 (Fig. 4). Additionally, growing these cells on glycerol instead of glucose had no effect on the ability of Msi1p to act as a co-repressor of transcription with pRB (data not shown). In sharp contrast to other activities of Msi1p, the corepressor activity is not affected by YAK1.
YAK1 has been described as a negative regulator of growth in Saccharomyces cerevisiae3 and overexpression of the kinase has been shown to cause full growth arrest if signaling through the RAS/cAMP pathway is diminished.5 We sought evidence that excess copies of YAK1 might lead to more subtle and quantifiable growth defects in the presence of wildtype RAS/cAMP signaling. Yeast cells carrying the plasmid pRS426-ADH1-YAK1 (a high-copy plasmid that has the YAK1coding region under the control of the strong ADH1 promoter4) had an exponential phase doubling time in glucose that was nearly twice as long as cells carrying a control plasmid (Fig. 5). When grown using glycerol as the carbon source, overexpression of YAK1 still caused the maximal growth rate to slow nearly two-fold (data not shown). However, in the absence of the MSI1 gene, the overexpression of YAK1 had no effect on the doubling time of the culture compared to the wildtype yeast cells. Deletion of the CAC1 gene did not affect the ability of YAK1 to decrease the growth rate (Fig. 5), indicating that this phenotype is specific to MSI1 and not due to general chromatin defects that can lead to a modest activation of checkpoints during the cell cycle.46 Similarly, deletion of NPR1 had no effect on the ability of YAK1 to increase the doubling time. The lag times for these strains showed no reproducible differences (data not shown). This is the first reported indication that the product of the MSI1 gene is required for a function of Yak1p.
We sought mutations in the MSI1 gene that might be able to genetically separate some of these functions. We began by taking advantage of an existing restriction site to generate msi1-8, a mutant that truncates the protein after amino acid 281. This mutation removes three of the seven WD40 domains present in Msi1p and is similar to the naturally occurring cac3-1/msi1-1 allele that truncates the protein after amino acid 292, which appears to have fully lost the ability function in chromatin assembly.29 The ability of msi1-1 to support other functions is unknown. Immunoblotting indicated that DB-msi1-8 fusion protein was approximately as abundant as the DB-Msi1 fusion protein (data not shown), indicating that this variant is reasonably stable in vivo. As predicted, Msi1-8p failed to productively interact with Cac1p in the two-hybrid system (Fig. 6a). In contrast, both Msi1-8p and full-length Msi1p bound Npr1p equally well (Fig. 6a) demonstrating that the N-terminal extension and the first four WD40 repeats are sufficient for Npr1p binding. Considering that suppression of the RAS/cAMP pathway is believed to require the binding of Npr1p and is independent of Cac1p, we hypothesized that overexpression of the msi1-8 allele would restore heat-shock resistance to cells with an activated RAS/cAMP pathway. Indeed, overexpression of this allele suppressed an activated RAS/cAMP pathway (Fig. 6b, compare with Fig. 1). Thus, this allele effectively separates the ability of Msi1p to bind Cac1p and assemble chromatin from its ability to bind Npr1p and suppress the RAS/cAMP pathway.
We next investigated the ability of the Msi1-8p truncated protein to support the carbon-source regulated one-hybrid activity. This protein had completely lost the ability to activate transcription when the cells were grown on a nonfermentable carbon source and had a much more severe phenotype than deletion of yak1 (Fig. 6c). These data indicate that the carboxy-terminal 130 residues are required both for binding Cac1p to associate with the CAF-I complex and for a second function that allows the carbon source regulated one-hybrid activity. Furthermore, the phenotypes resulting from the truncated msi1 allele are qualitatively similar to those of yak1Δ; that is, both mutations block Cac1p binding and decrease the regulated one-hybrid activity without affecting the binding of Npr1p or suppression of the RAS/cAMP pathway. The most likely explanation for this convergence is that Yak1p affects Msi1p via the last three WD40 repeats of Msi1p.
We hypothesized that the regulated one-hybrid activity may be simply explained by a quantitative change in Msi1p levels in response to nonfermentative conditions. We tagged the endogenous MSI1 gene with the myc13 epitope47 in strains with and without an intact YAK1 gene. These strains were grown in either glucose or glycerol, lysed and the fusion protein was separated by SDS-PAGE and detected by immunoblotting. The immunoblots revealed that the Msi1 protein isolated from respiring cells consistently accumulated to a higher steady-state level compared to fermenting cells (Fig. 7). Quantitation of the immunoblots showed approximately 3-fold more Msi1 protein in cells grown in the nonfermentable carbon sources of glycerol, acetate or ethanol than cells grown in either glucose or sucrose (Fig. 7 and data not shown). The elevated accumulation of Msi1p caused by nonfermentable carbon sources was independent of either YAK1 or NPR1. Considering that the levels of MSI1 mRNA do not appreciably change as yeast cells shift from fermentative to aerobic growth,48, 49 it appears that the Msi1 protein is regulated translationally or post-translationally based on the carbon-source that the cells are utilizing. Additionally, no clear change in mobility that could be caused by a phosphorylation of Msi1-myc13p was identified (Fig. 7). Treatment of the protein lysate with either potato acid phosphatase or lambda phosphatase had no effect on the mobility of Msi1-myc13p (data not shown). Considering that at least one of the mammalian homologs of Msi1p is phosphorylated on a tyrosine residue in response to a mitogenic signal50 and that Yak1p has been reported to be an active tyrosine kinase,51 we also sought evidence that Msi1p might be phosphorylated on a tyrosine residue. Immunoprecipitated Msi1-myc13p was analyzed by western blotting with an anti-phosphotyrosine serum. Regardless of growth conditions or YAK1 status, no evidence of tyrosine phosphorylation on Msi1p could be detected (data not shown). These data are consistent with the idea that Yak1 does not directly phosphorylate Msi1p in response to metabolic changes.
Previous work has shown that Yak1p is either important or essential for the subcellular localization of several proteins involved in nutritional regulation 7, 9, 11, 12 so we investigated the possibility that Yak1p could influence the subcellular localization of Msi1p. Previous work has shown that GFP-tagged Msi1p expressed from a highly induced GAL promoter on a high-copy plasmid is located in both the nucleus and the cytoplasm of wildtype cells.17 To determine the subcellular localization of Msi1p under more physiologically relevant conditions, we used indirect immunofluorescence to examine yeast whose endogenous MSI1 gene is tagged with the myc epitope. When these cells are grown on the fermentable carbon source glucose, Msi1p is found throughout the cell, with no obvious accumulation of Msi1p in the nucleus of the cell. Under these physiological conditions, the subcellular location of Msi1p is independent of YAK1. However, when this strain was grown on the nonfermentable carbon source glycerol, a notable shift in localization occurred, with approximately 32% of the cells showing an accumulation of Msi1 in the nucleus (Fig. 8). In these nonfermenting conditions, YAK1 showed a substantial effect on Msi1p nuclear accumulation. When YAK1 was deleted, approximately 57% of the glycerol-grown cells showed a distinct nuclear localization of Msi1p. We found a striking reversal of this trend when Msi1p localization was assayed in a strain that overexpressed YAK1 - in fact, in this strain, only 13% of the cells exhibited a distinct nuclear accumulation of Msi1p, resulting in a localization pattern not substantially different from that observed when cells were grown on glucose. The percentage of glycerol-grown cells showing nuclear accumulation of Msi1p was significantly different (p<0.05) in all pairwise comparisons among the wildtype, yak1Δ and YAK1-overexpressed cells. Deletion of the npr1 gene did not affect the subcellular localization pattern of Msi1p in either carbon source (data not shown). Thus we conclude that the subcellular localization is a third characteristic of Msi1p that is significantly influenced by Yak1p and the carbon source available to the cells.
This work establishes a new phenotypic connection between the YAK1 and MSI1 genes of Saccharomyces cerevisiae. Although the association of each of these genes with the RAS/cAMP pathway has been established for some time,2, 13 our experiments revealed that their interaction is complex and extends well beyond this one pathway. Table 2 summarizes our finding that the Yak1 kinase is important for a subset of the multiple functions of Msi1p and that Msi1p is important for Yak1p function as well.
We have identified yak1Δ as a genetic condition that interrupts the two-hybrid interaction between Msi1p and Cac1p in the CAF-I complex without resulting in the apparent complete dissolution of the complex. This suggests a potential feedback loop to regulate RAS/cAMP signaling. The Yak1 kinase antagonizes RAS/cAMP signaling,3 thus the loss of Yak1p activity should have the effect of increasing RAS/cAMP signaling. We have shown that deletion of YAK1 releases Msi1p from CAF-I, perhaps facilitating the ability of Msi1p to associate with Npr1p and decrease RAS/cAMP signaling.17
In the course of this work, we have uncovered a surprising and previously unknown ability of Msi1p to activate transcription. It is not unusual to find that a “bait” fusion protein intended for use in the two-hybrid system shows direct activation of transcription instead.52 Msi1p, however, activates transcription of the reporter only in the absence of fermentation, and the YAK1 gene influences this activity. This regulatory pattern suggests that we have discovered a novel function of Msi1p that may be physiologically relevant. Thus far, we have only been able to measure the carbon source-regulated one-hybrid activity using the DB-Msi1p fusion protein and the artificial HIS3 reporter gene. Neither Msi1p nor other proteins in this family have been proposed to bind DNA directly, but are widely believed to associate with histones.14 The bone fida target(s) of the ability of Msi1p to activate transcription remain to be identified.
Yak1p has a substantial effect on the subcellular localization of Msi1p, promoting cytoplasmic accumulation under nonfermentative growth conditions. As such, Msi1p joins Crf1p12 and Bcy1p7 as proteins whose nuclear accumulation is influenced by Yak1p in response to nutrient availability. Strikingly, the means by which Yak1p influences the subcellular localization of these three proteins are distinctly different. For both Msi1p and Bcy1p, Yak1p antagonizes nuclear accumulation but it does so in differing physiological contexts: Msi1p accumulates in the nucleus when the cells are grown in glycerol, while Bcy1p accumulates in the nucleus when the cells are grown in glucose.7 Yak1p directly phosphorylates both Crf1p, increasing its nuclear accumulation,12 and Bcy1p, decreasing its nuclear accumulation in cooperation with Zds1p.7 However, Zds1p does not have an effect on the physical association of Msi1p and Cac1p (Fig. 2) or on the regulated one-hybrid activity, and Yak1p does not appear to phosphorylate Msi1p. Therefore, Yak1p appears to influence the localization of Msi1p by a mechanism different than those used for Bcy1p and Crf1p.
Growth on nonfermentable carbon sources has at least three effects on Msi1p: total protein accumulation increases, a cryptic ability to activate transcription when tethered to a promoter is revealed and nuclear localization of Msi1p increases. Only the last two of these effects are influenced by YAK1. It is tempting to speculate that the carbon-source regulated one-hybrid activity of Msi1p may be partially controlled by the protein’s quantity and/or subcellular localization. The fact that Msi1p accumulated in the nucleus in approximately three times as many cells when grown under nonfermentative conditions correlates with the protein’s activation of transcription under these same nutritional conditions. However, we discovered the transcriptional activation function using the Gal4 DB-Msi1p fusion protein, whose DNA-binding domain contains a strong, constitutive nuclear localization signal.53 Thus we expect that this fusion protein is likely found in the nucleus regardless of the carbon source. Furthermore, deletion of YAK1 modestly decreased the transcriptional activation function but increased the nuclear accumulation of Msi1p. Taken together, these findings indicate that Yak1p must have effects on Msi1p beyond influencing nuclear accumulation and that nuclear accumulation alone cannot explain the ability of Msi1p to activate transcription.
Importantly, Yak1p does not influence all of the functions of Msi1p. Loss of YAK1 had no effect on the increased accumulation of Msi1p in nonfermentable carbon sources, the ability of Msi1p to act as a transcriptional co-repressor for pRB or its ability to associate with Npr1p. Thus, the Yak1 kinase is essential for only a subset of the known functions of MSI1.
Additionally, we have established that Msi1p can have a sizeable influence on YAK1 phenotypes. Loss of YAK1 causes an increase in heat shock sensitivity of cells with or without an activated RAS allele3 which was fully suppressed by MSI1 overexpression. We have also found that elevated concentrations of Yak1p significantly increased the doubling time of a logarithmic yeast culture, but this phenotype is fully dependent on an intact MSI1 gene. The ability of excess copies of YAK1 to decrease growth rate can likely be accounted for by the intimate involvement of the RAS/cAMP pathway with cell cycle control1 and quiescence.54 What is surprising, however, is our finding that MSI1 is required for this function of YAK1 even though deletion of MSI1 alone has no effect on growth rate. It is possible that nuclear accumulation of Msi1p helps to decrease the cellular growth rate, either due to growth under nonfermentable conditions or due to overexpression of YAK1. The molecular basis of this requirement is not yet clear but is under active investigation.
Plasmids YEplac195,55 YEplac195-PDE2,56 pRS426-ADH-YAK1,4 pML9 (npr1::LEU2),57 pM2741 (zds1::URA3),58 pDB and pDB-RB35 have been previously described and were gifts from their respective laboratories. Plasmids pGAL4, pDBD-p53 and pAD-TAg were obtained from Stratagene and pGAD424 was obtained from Clontech. Plasmids YEp55-MSI1, pGBT9-MSI1, pGAD1-NPR1561-605 and pGAD1-CAC1 (which contains amino acids 119-606 of CAC1) have been described previously.17 The pGBT9-msi1-8 and YEp55-msi1-8 plasmids were made by cutting either pGBT9-MSI1 or YEp55-MSI1 with BglII and religating the large fragment, resulting in a deletion of the last 141 amino acids. The CAC2 gene was amplified by PCR from a W303a strain using primers that added an EcoRI and BamHI sites. After digestion with these two enzymes, this gene was ligated into pGBT9 to make pGBT9-CAC2.
Strains used in this study are listed in isogenic groups in Table 1. Strains were grown in standard laboratory complete media with the appropriate amino acid dropouts and the indicated carbon source.44 The npr1::TRP1 and yak1::TRP1 deletion alleles and the MSI1-myc13::HIS3MX6 allele were made by PCR epitope tagging as described47 and were verified by PCR analysis of the relevant locus. Auxotrophic markers were swapped as described.59 In some cases, the MSI1 or YAK1 gene was deleted by PCR amplifying the yak1::KanMX4 or msi1::KanMX4 allele from the deletion collection60 and transforming the PCR product into the appropriate strain. All genomic modifications were confirmed by PCR of the locus.
Determinations of heat shock survivability were performed essentially as described.17 Briefly, cells were grown in synthetic medium and 2% galactose for three days before being washed and incubated at 55°C for ninety minutes. A control aliquot of the same cells was not heat shocked to determine the number of cells initially present. After cooling to room temperature, serial dilutions were plated and fraction of cells surviving was determined. Averages of at least six trials for each strain are shown on the graphs with standard deviations indicated by the error bars.
One- and two-hybrid assays were conducted by growing the indicated yeast transformants or integrated strains on selective plates overnight followed by ten-fold serial dilutions to plates containing the specified media. Plates were incubated two to five days at 30°C before being photographed. For unknown reasons, the Msi1p/Npr1p was inefficient in the PJ69-4 background and thus the HF7c background was used to assay these interactions.
Growth dynamics were determined by monitoring the OD600 of the indicated yeast strains. Cultures were diluted into SC media with 2% glucose44 and grown at 30°C with shaking at 225 rpm. Doubling time and the length of lag phase were calculated by using the MicroFit program61 and were determined independently at least three times for each strain.
Actively growing cultures in rich media were collected by centrifugation, washed in water and resuspended in cold Lysis Buffer with protease and phosphatase inhibitors (10 mM Tris-Cl pH8.0, 150 mM NaCl, 10% glycerol, 0.05% Tween-20, 5 mM EDTA, 1 mM DTT, 1 mM PMSF, 50 mM NaF, 0.2 mM Na3VO4). The cells were lysed by vortexing with glass beads for five minutes at room temperature before unlysed cells were removed by centrifugation. Total protein concentration was estimated by measuring the absorbance at 280nm and equal quantities of protein were resolved by SDS-PAGE. The myc epitope was detected by incubating the blots with a 1:40,000 dilution of the 9E10 antibody (Santa Cruz Biotechnology) followed by a 1:40,000 dilution of anti-mouse conjugated to horseradish peroxidase and detected with SuperSignal substrate (Pierce).
Yeast strains were cultured in synthetic media lacking uracil to exponential phase before being harvested by filtration and fixed in buffer (0.1 M KH2PO4, 0.5 M MgCl2, 40 mM KOH) with formaldehyde (1:10 dilution) for 5 minutes. Cells were washed with phosphate buffered saline (PBS), briefly sonicated, washed again in PBS, and then washed in sorbitol citrate buffer (0.1 M K2HPO4, 33 mM citric acid, 1.2 M sorbitol, 2 mM DTT). Cells walls were removed by digestion with a 1:10 volume of glusalase (Perkin Elmer), a 1:100 volume of a 20 mg/ml concentration of zymolyase 20T (ICN), and 1 mM DTT for 2 hours at 30°C. Cells were washed four times with sorbitol citrate buffer. Cells were fixed to polylysine-coated slides by placing the cell suspension in wells for 3 minutes and aspirating off the suspension until they were dry. The slides were placed in 100% methanol for 3 minutes, 100% acetone for 30 seconds, and then air dried. Fixed cells were then rehydrated and blocked by incubation with 2% nonfat dry milk in PBS-Tween solution (1x PBS, 0.2% Tween 20) overnight at 4°C. Primary monoclonal antibody 9E10 was diluted 1:200 in blocking solution, cleared for 2 minutes by centrifugation in a microcentrifuge, and incubated on the cells for 2 hours. The wells were washed with 2% nonfat dry milk PBS-Tween solution quickly four times, followed by three 5 minute washes. Secondary antibody (Cy3-conjugated anti-mouse immunoglobulin G; Jackson Immunochemicals) was diluted 1:200 in blocking solution, cleared, and incubated on the wells for 2 hours. Cells were washed again and mounting solution (10% 1x PBS in glycerol with 0.02 μg/ml DAPI [4’, 6’-diamidino-2-phenylindole] and 1 mg/ml of phenylenediamine) was added to the wells. The negative control for integrated myc-tagged Msi1 protein was the wild-type strain yJB195. At least three independent samples were analyzed for each growth condition and more than 300 cells were counted per sample. Statistical significance of the data was determined by ANOVA analysis. Immunofluorescence was done using a BX51 Olympus Fluorescence Microscope (Olympus, Inc.) using an Apoplan 100X objective. Images were captured using a SPOT RX digital camera. All images presented were captured and processed in parallel by identical means.
The authors thank Sara Harley, Scott Halkyard, Alis Heidar, Jason Karpus, Devin Miller, Bethany Stark and Jon Visick for excellent technical assistance and helpful comments on the manuscript. We thank D. Kesler for statistical analysis of localization data. We thank Judy Berman, Joseph Heitman, Brian Kennedy, Joe Mymryk, David Stillman and Johan Thevelein for providing strains and plasmids. This work was supported by National Institutes of Health grant 1R15GM67262 to SDJ. MEM. was supported by grant number 1R15CA098257-01 from the National Cancer Institute. Its content are solely the responsibility of the authors and do not necessarily represent the official views of the National Cancer Institute.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.