INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Eukaryotic cells react to their environment by regulating transcription factors bound to promoters of responsive genes (28). Cells grown in culture are typically provided with sufficient essential nutrients and factors to ensure unchecked propagation. However, in their natural environment, cell growth is ordinarily limited by the scarcity of one or more factors or nutrients. In such circumstances, there must be mechanisms to ensure that transcriptional responses to one signal do not surpass what the cell can accommodate with its limited growth potential. This issue has not yet been addressed in eukaryotes. In this report, we demonstrate that the prototypical transactivator Gal4p is regulated by two separate signals represented by the specific inducer galactose and the RNA polymerase II (Pol II) holoenzyme-associated cyclin-dependent kinase Srb10p/CDK8. These observations suggest a mechanism whereby responses to a specific inducer can be coordinated with the physiological environment.

Gal4p regulates expression of the yeast GAL genes in response to galactose. In noninducing conditions, Gal4p is bound to the upstream activating sequences for galactose (UAS_G) but is prevented from activating transcription by the inhibitor Gal80p (32, 43). Rapid induction by galactose requires the product of GAL3 (40, 52, 57), which is a regulatory protein with similarity to the galactokinase encoded by GAL1 (3, 9, 50), although Gal3p does not have galactokinase activity (9). Recent experiments demonstrate that Gal3p, when bound to galactose, directly interacts with Gal80p in the presence of ATP (42, 50, 59, 60). Gal3p-galactose is thought to cause induction of the GAL genes by producing a conformational change in the Gal4p-Gal80p complex that allows interaction of the Gal4p activating domains with the general transcription factors (42, 59). The induced conformation may involve a shift of Gal80p from the C terminus to the central region of Gal4p (48). Yeast bearing gal3 disruptions are still able to induce GAL transcription in response to galactose, but induction requires several days rather than the minutes to hours required in wild-type (WT) yeast (7, 57). The mechanism for the delayed induction in gal3 yeast, known as long term adaptation (LTA), has remained elusive despite the fact that it was observed in some of the earliest laboratory yeast strains (44, 57).

Gal4p becomes phosphorylated on multiple serines when it activates transcription (37, 38, 45, 46) (see Fig. 1A). We have shown that most of these phosphorylations are mediated by the cyclin-dependent protein kinases of the RNA Pol II holoenzyme (24). These results are consistent with earlier observations which suggested that Gal4p phosphorylation in vivo requires both its DNA-binding and transcriptional activation functions (46) and that phosphorylation is impaired by mutations in gal11 (33, 45), which encodes a component of the RNA Pol II holoenzyme mediator subcomplex (5). One phosphorylation at serine 699 is necessary for full GAL gene induction (24, 45). S699 phosphorylation was shown to be mediated by Srb10p/CDK8 of the mediator subcomplex. SRB10 is required for S699 phosphorylation in vivo, and purified Srb10p/Srb11p (cyclin C) complexes phosphorylate S699 in vitro. Furthermore, SRB10 and Gal4p S699 phosphorylation were shown genetically to represent a common mechanism for GAL gene induction (24).

Although these results confirmed our earlier prediction that phosphorylation occurs as a consequence of transcriptional activation, they are enigmatic because Gal4p does not become phosphorylated unless it activates transcription (46) and yet it is not fully active unless S699 is phosphorylated (24, 45). This suggests that full induction of Gal4p activity must involve at least two mechanisms distinguished by the occurrence of S699 phosphorylation. In this report, we examine whether GAL regulation by SRB10 requires the inducer protein Gal3p, in order to determine whether both mechanisms are galactose specific. We found that phosphorylation at S699 occurs independently of Gal3p but is required for sensitive response of the GAL genes to the presence of galactose. Furthermore, SRB10 and GAL3 are shown to act independently in GAL gene expression. These observations suggest a mechanism whereby GAL transcription can respond to the environment through regulation of Srb10p and its associated cyclin C subunit Srb11p.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Plasmids and yeast strains. Strains used for these experiments are listed in Table 1. Plasmid YCpG4 is a TRP1, ARS-CEN vector which expresses GAL4 from its own promoter (45). Plasmid pMH683 is TRP1, ARS-CEN and expresses the GAL4683 derivative from the ADH1 promoter (24, 45). Plasmid pKOG3 was created by cloning a BglII LEU2 fragment into the BglII sites within the GAL3 coding region. GAL3 disruptions were made by transforming yeast with an NcoI fragment from this plasmid. GAL4 was disrupted by using plasmid pKOG4 which has the BglII-BamHI hisG-URA3-hisG fragment from pNKY51 (1) inserted (with blunt ends) between the SphI and XhoI sites of YCpG4. Disruptions (gal4::hisG) were produced by transforming yeast with a HindIII-BamHI fragment from this construct. GAL1 disruptions were produced by using pIS120, a two-step URA3 disrupter that deletes nucleotides +34 to +1141 (gal1120). The gal2::his5 disruption, which substitutes the complete GAL2 open reading frame (nucleotides +1 to +1722) with Schizosaccharomyces pombe his5, was generated by transformation with a DNA fragment produced by in vitro amplification with oligonucleotides oIS664 (GAL2 nucleotides 61 to 1-F1) and oIS665 (complementary to GAL2 +1785 to +1723-R1) by using pFA6a-His3MX6 as a template (34). Strains YJT1 and YJT2 are derived from YJR10 and contain a GAL1-HIS3 reporter gene integrated at LYS2 with pBM1571 (20). Single-copy WT and S699A GAL4 integrants were constructed by using pJT001 and pJT002 which have BamHI-HindIII genomic WT and S699A GAL4 fragments, respectively, inserted into YIpade101, which is a URA3 plasmid for two-step disruption of ade8. Single-copy integrations of this plasmid in an ade2 background produce 5-fluoroorotic acid-resistant white colonies. The GAL4 S699E mutation was constructed by mutagenesis in plasmid pGSH with oligonucleotide oIS155 (GTTTCTCCTGGCGAAGTAGGGCCTTCAC) and then subcloned into YCpG4 (45).

-Galactosidase assays and growth media. -Galactosidase assays with W303-1A-derived strains were performed as described (22). Briefly, cells were grown in minimal selective medium containing 5% glycerol and 2% lactate and were induced by adding galactose from a 40% sterile stock. All results are an average of at least three independent determinations, with a 5 to 10% average deviation from the mean. -Galactosidase activity in S288C-derived strains was measured in cell extracts prepared by lysing with glass beads as described (23, 46). Ethidium bromide-galactose or -glucose (EB-gal and EB-glu, respectively) contained yeast extract-peptone-dextrose (YEPD) supplemented with 2% galactose or glucose and 20 mg of ethidium bromide per liter (18). EB-gal and EB-glu plates were inoculated with 5 µl of cell cultures grown to saturation in minimal medium containing glycerol and lactate as the carbon sources, and growth was allowed for 5 days at 30°C.

Antibodies, metabolic labeling, and tryptic phosphopeptide analysis. Rabbit anti-Gal4p DNA-binding domain polyclonal antibody was as described (46). [³²P]orthophosphate labeling of yeast, immunoprecipitations, and tryptic phosphopeptide analysis was performed as described (26). Cells bearing the GAL4683 expression plasmid were grown in minimal medium containing glycerol and lactate to an A₆₀₀ of 1.0, were collected by centrifugation, and were washed twice and resuspended in phosphate-depleted medium. After 2 h, galactose was added to 2% and the cells were then labeled for 90 min with 5 mCi of [³²P]orthophosphate per ml. The cells were then lysed, Gal4p was recovered by immunoprecipitation, and phosphopeptides were analyzed by electrophoresis at pH 2.1 in the horizontal dimension and by chromatography (butanol-acetic acid-dH₂O-pyridine [75/50/37.5/15.5]) in the vertical dimension. Phosphopeptides were visualized by exposure to Kodak Biomax film.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

The requirement for Gal4p S699 phosphorylation can be suppressed by high galactose concentrations. Gal4p is phosphorylated on at least five sites when it activates transcription, four of which have been identified (24, 45) (Fig. 1A). An additional phosphorylation site in the DNA-binding domain has not been precisely located (24). Changing Gal4p S699 to alanine or glutamate severely impairs induction of a GAL1-LacZ reporter in strains with the S288C genetic background (Fig. 1B) (24, 45). In contrast, phosphorylations at serines 691, 696, and 837 do not appear to be required for GAL induction under the conditions of our assay because Gal4p bearing alanine substitutions at all three of these residues (691, 696, and 837) activates transcription efficiently (Fig. 1B). This observation strongly suggests that the S699A mutation prevents GAL induction because of loss of that specific phosphorylation, rather than an effect on Gal4p structure, as was previously suggested (15).

View larger version (25K): [in this window] [in a new window]   FIG. 1.   Phosphorylation of Gal4p S699 is required for efficient GAL gene induction. (A) Schematic representation of Gal4p and the location of identified and predicted phosphorylations. Abbreviations: DNA, DNA-binding domain; AR1, activating region 1; AR2, activating region 2; GRD, glucose response domain. (B) Yeast strain YT6G80 bearing a control plasmid () or expressing WT GAL4 (), GAL4 S699A (), GAL4 S699E (), or GAL4 bearing alanine substitutions of serines 691, 696, and 837 () from plasmid YCpG4 were induced with 2% galactose for the indicated times. GAL expression was measured by assaying -galactosidase activity produced by the GAL1-lacZ reporter gene. (C) Yeast strain YJR10::131 bearing a vector control (), expressing WT GAL4 ( and ), or GAL4 S699A ( and ) were induced with either 2% galactose (closed symbols) or 0.02% galactose (open symbols), and GAL1-lacZ activity was measured at the indicated times.

Gal4p S699 phosphorylation is necessary for GAL induction in a strain bearing a weak GAL3 allele. Because GAL induction is more severely impaired by the GAL4 S699A mutation in YT6G80 than in the gal2 W303-1A background (compare Fig. 1B with Fig. 2A), we suspected that an additional GAL defect must contribute to this difference. YT6G80 bears the trp1-901 allele (Table 1), which has a deletion of DNA to 240 of the GAL3 open reading frame, causing weaker galactose induction, presumably because of reduced Gal3p levels (47). We examined the relative contribution of the weak GAL2 and GAL3 alleles on Gal4p S699-dependence in diploids produced by mating YT6G80 with W303-1A-derived strains bearing gal2 or gal3 disruptions. The GAL4 S699A mutation did not affect GAL induction by 2% galactose in a diploid strain produced by mating YT6G80 with a WT W303-1A-derived haploid (YJR53 YT6/WT) (Fig. 2B), demonstrating that the phenotype with respect to GAL4 S699A in YT6G80 is recessive to W303-1A. In the YT6G80/gal2 W303-1A diploid (ISY46 YT6/gal2) (Fig. 2B), GAL1-LacZ induction by WT GAL4 was slightly impaired relative to YJR53 (YT6/WT), and the GAL4 S699A mutation further impaired induction. However, induction by WT GAL4 was reduced by approximately 50% in a YT6G80/gal3 W303-1A diploid (YJR54 YT6/gal3) (Fig. 2B) relative to the comparable WT diploid strain (YJR53). Furthermore, the GAL4 S699A mutation significantly impaired GAL induction relative to WT GAL4 in the YT6G80/gal3 (YJR54) diploid strain (Fig. 2B). These results indicate that the trp1-901-linked GAL3 allele in YT6G80 contributes significantly to the dependence on Gal4p S699 phosphorylation for induction, more so than the weak GAL2 allele. Combined with the results shown above, these observations demonstrate that the S699 phosphorylation is necessary for full Gal4p activity in response to limiting galactose or when the inducer signaling mechanism (Gal2p and/or Gal3p) is weak. This suggests that WT levels of Gal3p, in the presence of high galactose concentrations, can maintain Gal4p in an active form without the S699 phosphorylation. We discuss the implication of this below.

View larger version (17K): [in this window] [in a new window]   FIG. 2.   Weak S288C GAL2 and GAL3 alleles contribute to dependence on the Gal4p S699 phosphorylation. (A) Yeast strain ISY45 (gal2 W303-1A) bearing a vector control () and expressing WT GAL4 () or GAL4 S699A () was induced with 2% galactose, and GAL1-lacZ activity was measured at the indicated times. (B) Diploid strains produced by crosses between YT6G80 and a WT W303-1A derivative (YJR53 YT6/WT), gal2 W303-1A (ISY46 YT6/gal2), or gal3 W303-1A (YJR54 YT6/gal3) were transformed with a vector control () or plasmids expressing WT GAL4 or GAL4 S699A (S669A). Cultures were induced with 2% galactose for 2.5 h before GAL1-lacZ expression was measured.

Gal4p S699 phosphorylation is required for sensitive response to inducer. Most experiments involving GAL induction in yeast employ galactose at 2%, despite the fact that Gal4p activity can be efficiently induced with far lower concentrations in a strain with fully functional GAL2 and GAL3 alleles (Fig. 1C). Therefore, we wondered whether the differential effect of the GAL4 S699A mutation we observed in W303-1A yeast induced with high and low galactose concentrations reflected a requirement of this phosphorylation for sensitive response to inducer. To examine this, we used a W303-1A yeast strain with a GAL1-HIS3 reporter gene (Fig. 3A) (20). In this strain, expression of HIS3 is completely dependent upon Gal4p activity. Yeast bearing this reporter and expressing either WT GAL4 or GAL4 S699A was plated on histidine-deficient (His) medium containing glycerol as the sole source of carbon, and sterile filters containing 2% galactose were placed in the center of the plates. We found that cells expressing WT GAL4 grew significantly better than cells expressing GAL4 S699A, as indicated by the larger halo of growth around the filter (Fig. 3B). This indicates that WT Gal4p is able to activate transcription of the GAL1-HIS3 reporter gene in response to much lower concentrations of galactose than is Gal4p S699A, in an otherwise WT yeast strain. In combination with the results shown in Fig. 1 and 2, this observation indicates that sensitive response to galactose requires fully functional GAL2 and GAL3 alleles as well as the Gal4p S699 phosphorylation.

View larger version (66K): [in this window] [in a new window]   FIG. 3.   Gal4p S699 phosphorylation is required for sensitive response to inducer. The W303-1A-derived strains YJT1 (WT GAL4) and YJT2 (GAL4 S699A) bearing an integrated GAL1-HIS3 reporter gene (A) were grown in YEPD, were washed with sterile water, and were plated in equivalent numbers in top agar on His plates containing glycerol as the sole source of carbon. Sterile discs were placed in the centers of the plates onto which 5 µl of 2% galactose was spotted. The plates were photographed after 3 days growth at 30°C (B).

Gal4p S699 phosphorylation is required for the LTA response. Yeast lacking gal3 induces GAL gene transcription several days after galactose addition, compared to less than an hour in WT strains (7, 57). This LTA response to galactose was observed very early in the development of Saccharomyces as a model eukaryote (44, 57) and has previously been used to support the idea that the GAL genes are regulated by a second mechanism independently of Gal3p (7). Since mutations to GAL4 S699 have a severe effect in S288C-derived strains bearing the trp1-901-linked GAL3 allele (Fig. 1B and 2B), we examined whether S699 phosphorylation was necessary for LTA response in yeast completely lacking gal3. GAL induction occurred 24 h after galactose addition in gal3 W303-1A expressing WT GAL4 (Fig. 4A). In contrast, gal3 yeast expressing the GAL4 S699A mutant did not induce GAL gene expression, even 60 h after galactose was added (Fig. 4A). This result demonstrates that S699 phosphorylation is absolutely necessary for GAL induction in the absence of Gal3p and supports the argument that this modification represents a mechanism for regulation of Gal4p activity that functions independently of Gal3p.

View larger version (32K): [in this window] [in a new window]   FIG. 4.   Gal4p S699 phosphorylation occurs independently of the Gal3p-galactose signaling mechanism. (A) Yeast strain YJR14::131 (gal3) bearing a vector control (), YCpG4 expressing WT GAL4 ( and ), or GAL4 S699A () was grown in minimal medium containing glycerol and were induced with 2% galactose or left uninduced (). GAL1-lacZ reporter gene expression was measured at the indicated times postinduction. (B) W303-1A (WT) and YJR58 (gal1 gal3) yeast expressing GAL4683 from a plasmid were labeled with [32P]orthophosphate in the presence of galactose. Tryptic phosphopeptides from labeled Gal4p were resolved in 2 dimensions and were visualized by autoradiography. Phosphopeptides 1 and 5 represent S699 and S837 phosphorylation, respectively. The major phosphopeptide 2 is derived from the DNA-binding domain (Fig. 1A). We do not know the origin of phosphopeptide 8 nor if the apparent increase in the gal1 gal3 strain is significant.

Gal4p is phosphorylated at S699 independently of the Gal3p-galactose signaling pathway. Since the results described above suggest that Gal4p S699 phosphorylation regulates the GAL genes independently of Gal3p, we wished to know whether phosphorylation at this site can occur independently of GAL3 or GAL1. The galactokinase encoded by GAL1 shares extensive homology with Gal3p (3, 9, 50). Constitutive expression of GAL1 can cause induction in the absence of Gal3p or galactose (8), and Gal1p can interact with Gal80p in vitro, in a galactose- and ATP-dependent manner, although apparently less efficiently than Gal3p (42, 56, 60). To date, we have been unable to perform tryptic phosphopeptide analysis of full-length Gal4p in vivo because it is difficult to detect by labeling when produced at normal levels, and its overproduction causes phenotypes that interfere with recovery of ³²P-labeled protein (unpublished observations). Consequently, for analysis of phosphorylation in vivo, we used the GAL4683 derivative described previously (24, 45). This derivative has a deletion of residues 148 to 682 of Gal4p, which eliminates the large inhibitory segment in the central region (49) (Fig. 1A) but which has all of the sites of phosphorylation known to occur on Gal4p in vivo (24, 45). Deletion of the inhibitory region causes elevated basal transcriptional activation, which is greatly exaggerated when GAL4683 is expressed from the ADH1 promoter for the purposes of in vivo [³²P]orthophosphate labeling (data not shown) (15). This feature enables examination of Gal4p phosphorylation in a gal1 gal3 mutant strain where WT Gal4p activity would normally be uninducible (see below) (9, 10). In cells labeled in the presence of galactose, we found no difference in total GAL4683 phosphorylation in gal1 gal3 yeast compared to WT (data not shown), nor did we observe significant differences in the individual Gal4p phosphorylations, as indicated by the intensity of tryptic phosphopeptides (Fig. 4B). Importantly, phosphorylation of S699 occurs in both WT and gal1 gal3 yeast, as indicated by the appearance of phosphopeptide 1 (Fig. 4B, indicated by an arrow) (24). This demonstrates that S699 phosphorylation is not dependent upon Gal3p or Gal1p.

Srb10p and Gal3p define two independent regulatory mechanisms for Gal4p. SRB10 is required for full induction of GAL transcription (4, 29, 31), and we have previously shown that this effect is mediated by Gal4p S699 phosphorylation (24). Since Gal3p is not required for Gal4p S699 phosphorylation, it seemed likely that the regulatory effect of Srb10p on GAL transcription must also occur independently of Gal3p. To examine this possibility, we assayed GAL gene expression in strains bearing combinations of gal3 and srb10 disruptions by examining growth on EB-gal plates. We found that an srb10 disruption on its own in W303-1A causes slightly slower growth on EB-gal and the formation of smaller individual colonies than WT (Fig. 5B, srb10). Consistent with previous observations (4, 29, 31), we also observed slightly slower induction of a GAL1-LacZ reporter in srb10 W303-1A (data not shown). Disruption of gal3 in W303-1A does not completely prevent growth on EB-gal, but rather causes the formation of infrequent colonies, most of which are slower growing (Fig. 5B, gal3). These infrequent gal3 colonies do not represent GAL revertants because they grow identically when recovered and restreaked on EB-gal after growth to saturation in nonfermentable carbon (data not shown). In contrast to gal3 W303-1A, yeast bearing a gal3 srb10 double disruption were completely incapable of growth on EB-gal (Fig. 5B, gal3 and srb10), demonstrating that Gal3p and Srb10p are both required for GAL gene expression. These results are consistent with the fact that Gal4p S699 phosphorylation is required for GAL induction in a gal3 strain (Fig. 3) and with our previous observation that S699A is genetically epistatic to SRB10 (24). Taken together, these results demonstrate that Gal4p activity is regulated by two independent mechanisms, involving Gal3p-galactose and the Srb10p-dependent phosphorylation at S699.

View larger version (51K): [in this window] [in a new window]   FIG. 5.   Gal3p and Srb10p represent independent mechanisms for GAL induction. Yeast strains W303-1A (WT), H617 (srb10), YJR7 (gal3), and YJR47 (gal3 srb10) were grown to saturation in minimal medium containing glycerol and lactate as the sole sources of carbon, and 5 µl was used to inoculate YEPD-glucose (A) or YEPD-galactose (B) plates containing ethidium bromide. Plates were photographed after incubation at 30°C for 5 days.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

In this report, we demonstrate that Gal4p phosphorylation at S699 occurs independently of Gal3p, and, furthermore, that GAL3 and SRB10 define two genetically distinguishable mechanisms for GAL gene expression. The important conclusion that can be drawn from these observations is that factors which influence Srb10p activity will also modulate Gal4p function independently of its specific inducer. Thus, we propose that the inducer Gal3p-galactose complex (42, 50, 59, 60) causes a transient alteration in the interaction between Gal80p and Gal4p (48) that allows the activating regions to contact the general transcription factors (Fig. 6B). Under conditions where Srb10p is active, Gal4p can become phosphorylated at S699, which may function to retain the Gal4p-Gal80p complex in an active conformation (Fig. 6C) (59). In this view, Srb10p will accelerate GAL induction under conditions in which it can phosphorylate Gal4p.

View larger version (40K): [in this window] [in a new window]   FIG. 6.   Gal4p activity is regulated by two independent signals. Under noninducing conditions (A), Gal4p activity is inhibited by the negative regulator Gal80p. Upon galactose addition (B), Gal3p-galactose interacts with Gal80p to cause a transient conformational alteration that allows Gal4p to activate transcription. During interaction with the RNA Pol II holoenzyme (C), Gal4p is phosphorylated at S699 by Srb10p; this phosphorylation stabilizes the active Gal4p-Gal80p conformation induced by Gal3p-galactose. The ability of Srb10p to phosphorylate Gal4p is regulated by independent environmental signals, thus modulating GAL induction to levels appropriate for the cellular environment.

Recent observations indicate that Srb10p is regulated by the environment. The regulatory cyclin C subunit for Srb10p, Srb11p (also known as Ume3p), has been shown to become degraded in response to heat and hypoxic stress (13, 14). Srb11p is also degraded when cells are shifted from fermentable carbon to poor carbon-containing medium (14). These observations are consistent with our previous results, indicating that WT Gal4p is unphosphorylated in gal80 yeast growing in the absence of a fermentable carbon source (45), even though it can activate transcription constitutively under these conditions (51). However, we observe the rapid appearance of WT Gal4p phosphorylated species upon the addition of any fermentable carbon in gal80 yeast, including glucose when Gal4p is overproduced to alleviate glucose inhibition (45, 49) (data not shown). This suggests that activity of the RNA Pol II holoenzyme-associated protein kinases, at least towards Gal4p, is inhibited in the absence of a fermentable source of carbon and is not specifically regulated by galactose. The levels of Srb10p also decrease as cultures approach the diauxic shift, where nutrients become limiting and the growth rate decreases (25). It has been suggested that Srb10p may be generally required for the repression of genes involved in stress response or growth under conditions of nutrient limitation (25).

Considering these results, we propose that Srb10p/CDK8 functions as a throttle for the GAL genes to allow robust induction under circumstances where cell growth is not limited by an essential nutrient or stress (Fig. 6C). In contrast, under conditions where the cell is subject to physiological stress, Srb10p expression or activity is inhibited or its cyclin C (Srb11p) is degraded. This causes inhibition of S699 phosphorylation on Gal4p and thereby limits GAL gene activation. Our model suggests a function for the RNA Pol II-associated CDK8 that involves communication of general physiological signals to gene-specific transcription factors during transcriptional activation.

The mechanism by which S699 phosphorylation regulates Gal4p activity has yet to be determined. Since activation by Gal4p is unaffected by the S699A mutation in gal80 cells, the simplest model is that the phosphorylation controls the interaction of Gal80p (24, 45). Some experiments indicate that galactose does not cause dissociation of Gal80p (12, 30, 41, 42), but rather may result in a shift to a central region of Gal4p with an overall weaker interaction (48). The precise effect of S699 phosphorylation on Gal4p-Gal80p interaction may be difficult to establish, particularly because this modification represents a minor species relative to total Gal4p phosphorylation (24, 45). S699 phosphorylation could stabilize a transiently induced conformation shift caused by Gal3p-galactose or, alternatively, could produce an additional conformational change. We favor the former possibility, because we find that S699 phosphorylation is not required for full GAL induction when yeast bearing WT GAL2 and GAL3 alleles are induced with high concentrations of galactose. This suggests that high concentrations of Gal3p-galactose may be able to hold Gal4p-Gal80p in the active conformation such that S699 phosphorylation becomes redundant. In contrast, under limiting concentrations of galactose, or when some aspect of galactose signaling is impaired (as in the S288C background), the inducer concentration may not be high enough to maintain Gal4p-Gal80p in an active conformation. This view can explain why fully functional GAL2/GAL3 and Gal4p S699 phosphorylation are required for sensitive response to galactose.

SRB10 and phosphorylation of Gal4p S699 are required for the LTA response observed in gal3 yeast (Fig. 4A and 5B). The observations presented here suggest an explanation, at least in part, for this phenomenon. In the absence of Gal3p, it is possible that Gal80p might spontaneously slip into the active conformation (Fig. 6B) at a low frequency. If this should occur in yeast growing under conditions where Srb10p is active, Gal4p can become phosphorylated on S699, which may stabilize the active conformation (59) (Fig. 6C). The presence of galactose, as a fermentable sugar, can stimulate activity of Srb10p independently of Gal3p (Fig. 4B). However, unlike other fermentable sugars, galactose can bind Gal1p, which would eventually accumulate due to elevated basal transcription caused by the Gal4p S699 phosphorylation. This could result in the LTA effect, since Gal1p-galactose can cause induction of Gal4p-Gal80p, albeit somewhat less efficiently than Gal3p (42, 56, 60). This model might explain why gal3 W303-1A yeast grows on EB-gal as infrequent colonies (Fig. 5B) that might represent clones in which Gal80p has spontaneously shifted to allow accumulation of phosphorylated Gal4p, which presumably is passed to daughter cells. We are currently investigating this possibility further.

The LTA response in gal3 yeast has also been shown to require genes encoding enzymes for conversion of galactose into glucose-1-phosphate (GAL1, GAL7, GAL10) (10) and entry into glycolysis (GAL5 and PGI1) (9) as well as for respiratory function (19). Part of the requirement for GAL1 can be explained as above (7, 9). The additional requirement for competent galactose catabolism and respiratory function has been suggested to reflect a secondary signal generated by the cellular energy status (9). One possibility, in view of the results presented here, is that Srb10p activity may be modulated in response to such an energy signal. However, we believe the requirement for enzymes downstream of GAL1 in LTA needs reinvestigation due to complications with earlier experiments. First, many of these previous experiments investigating GAL gene regulation were performed with derivatives of S288C GAL2 revertants (9, 10, 55). This GAL2 allele causes significantly impaired galactose uptake (16, 55) and results in the imp1 phenotype in which growth on galactose is dependent on respiratory function (2). The effect of the weak GAL2 allele on LTA has not been clearly established. Also, our observation that gal3 W303-1A forms infrequent colonies on EB-gal contradicts earlier experiments which showed that gal3 yeast is completely incapable of growth on EB-gal, where respiration is inhibited (54). The difference appears to relate to the fact that yeast was shifted directly from glucose-containing medium to EB-gal in previous experiments (54), whereas we inoculate EB-gal plates with cells from cultures grown to saturation in nonfermentable carbon. Glucose inhibits GAL4 expression (21), and, therefore, extended growth in glucose is likely to cause Gal4p depletion. The effect of shifting directly from glucose into galactose medium may also have contributed to misinterpretation of imp1 as a separate gene from gal2 (55).

Our results suggest that the yeast GAL genes are regulated by two independent signals to allow regulation by a specific signal while coordinating induction with the physiological environment. In this respect, we believe the relationship of Srb10p/CDK8 to GAL gene regulation presents some conceptual similarities to the Escherichia coli lactose operon. Full induction of the lac operon requires a lactose-specific signal that causes induction by inactivation of the lac repressor. The lac genes are concurrently regulated by a global signal represented by cyclic AMP, which stimulates transcription through the function of catabolite gene activator protein (6). Based on our results, we propose that Srb10p/CDK8 functions in a similar manner to allow accelerated transcription under favorable conditions by communicating signals generated by the physiological environment to gene-specific activators during their initial interaction with the RNA Pol II holoenzyme.