Glucose represses the lactose-galactose regulon in Kluyveromyces lactis through a SNF1 and MIG1- dependent pathway that modulates galactokinase (GAL1) gene expression
Glucose represses the lactose-galactose regulon in Kluyveromyces lactis through a SNF1 and MIG1 - dependent pathway that modulates galactokinase ( GAL1 ) gene expression Jinsheng Dong+ and Robert C. Dickson*
Department of Biochemistry and the L. P. Markey Cancer Center, University of Kentucky College of Medicine, Lexington, KY 40536-0084, USA
Received May 21, 1997;Revised and Accepted July 17, 1997
ABSTRACT
Expression of the lactose-galactose regulon in Kluyveromyces lactis is induced by lactose or galactose and repressed by glucose. Some components of the induction and glucose repression pathways have been identified but many remain unknown. We examined the role of the SNF1 (KlSNF1) and MIG1 (KlMIG1) genes in the induction and repression pathways. Our data show that full induction of the regulon requires SNF1; partial induction occurs in a Klsnf1-deleted strain, indicating that a KlSNF1-independent pathway(s) also regulates induction. MIG1 is required for full glucose repression of the regulon, but there must be a KlMIG1-independent repression pathway also. The KlMig1 protein appears to act downstream of the KlSnf1 protein in the glucose repression pathway. Most importantly, the KlSnf1-KIMig repression pathway operates by modulating KlGAL1 expression. Regulating KlGAL1 expression in this manner enables the cell to switch the regulon off in the presence of glucose. Overall, our data show that, while the Snf1 and Mig1 proteins play similar roles in regulating the galactose regulon in Saccharomyces cerevisiae and K.lactis, the way in which these proteins are integrated into the regulatory circuits are unique to each regulon, as is the degree to which each regulon is controlled by the two proteins.
INTRODUCTION
Kluyveromyces lactis is one of the few yeasts that can use the milk sugar lactose as a carbon and energy source, which suggests that this yeast may have evolved under different and unique selection pressures, particularly for carbon sources, than have many other yeasts including Saccharomyces cerevisiae (reviewed in 1 ). Kluyveromyces lactis grows slightly more rapidly with lactose as a carbon source than with glucose (2 ), but at least in some strains, glucose is the preferred carbon source since it represses expression of the genes necessary for utilization of lactose or galactose (3 ). Few components of the glucose repression pathway have been identified and we have only a rudimentary outline of the way in which the pathway represses expression of the genes necessary for lactose and galactose utilization-the lactose- galactose regulon (4 -7 ). To further our understanding of the repression pathway, we examined the role of the SNF1 (KlSNF1) and the MIG1 (KlMIG1) genes in glucose repression and induction. We show here that at least one glucose repression pathway contains both the KlSNF1 and the KlMIG1 gene products and we identify a way in which this pathway modulates expression of genes in the lactose-galactose regulon.
Utilization of lactose or galactose requires induction of transcription of KlLAC4 ([beta]-galactosidase; 8 ) and KlLAC12 (lactose permease), which are transcribed in opposite directions from a common promoter (9 ), and KlGAL1, KlGAL7 and KlGAL10 [coding for galactokinase (EC 2.7.1.6), galactose-1-phosphate uridylyltransferase (EC 2.7.7.12, transferase) and uridine diphosphoglucose 4-epimerase (EC 5.1.3.2, epimerase), respectively], which are tightly linked, with KlGAL1 and KlGAL10 transcribed in opposite directions from a common promoter (reviewed in 1 ).
The transcription induction pathway centers around the DNA-binding protein KlGal4p (10 ,11 ) whose concentration is tightly regulated by an autoregulatory loop that produces a 2-3-fold increase in its concentration, an essential ingredient in the induction pathway (4 ,7 ). Activation of transcription is also controlled by the negative regulator KlGal80p which is bound to and modulates the transcription activator activity of KlGal4p (12 ). Induction of the regulon also requires an uncharacterized activity of the KlGal1 protein that is independent of its galactokinase activity (13 ). This uncharacterized activity may be responsible for the galactose and ATP-dependent binding of KlGal1p to KlGal80p, an interaction that permits KlGal4p to activate transcription (12 ).
Glucose represses expression of the lactose-galactose regulon in some but not all strains of K.lactis (3 ). Repressing and non-repressing strains differ by two bases in the KlGAL4 promoter (4 ). How this region of the promoter modulates glucose repression is unknown. The KlGal80 protein is another known component of the glucose repression pathway. Expression of the lactose-galactose regulon is only slightly (10-20%) repressed in a Klgal80 deletion strain (14 ). Finally, the FOG1/GAL83 gene may be necessary for the glucose repression pathway(15 ), but this inference relies heavily upon what we know about GAL83 in S.cerevisiae.
Many genes necessary for glucose repression of the galactose regulon in S.cerevisiae have been identified and their role in the pathway is becoming clearer. A central component is ScSNF1, encoding a serine/threonine protein kinase (16 ). The ScSnf1 protein regulates many cellular functions (17 ) and is particularly critical for governing carbon metabolism (reviewed in 18 ). The protein has been conserved in organisms ranging from yeasts to plants to man where the ScSnf1 homolog, termed the AMP- dependent protein kinase, plays roles in cellular stress responses (19 ) and regulation of cholesterol and fatty acid biosynthesis (20 ). Much of what is known about ScSnf1p function has come from studying its role in glucose starvation. These studies have shown that the protein kinase activity of ScSnf1p is regulated in response to glucose by ScSnf4p and by other proteins (17 ).
Another key component necessary for glucose repression of the galactose regulon in S.cerevisiae is the Mig1 protein (reviewed in 18 ). Mig1p acts to repress transcription of the galactose regulon by binding to GC-boxes (21 ) present in the ScGAL1 and ScGAL4 promoters (22 ,23 ). Cells lacking Mig1p show partial derepression of the galactose regulon and this phenotype is epistatic to loss of ScSnf1p, suggesting that ScMig1p acts downstream of ScSnf1p (reviewed in 18 ). DNA-bound Mig1p represses transcription by forming a complex with Tup1p and Ssn6p (24 ,25 ). It is not yet known how Snf1p communicates with the Mig1p-Tup1p-Ssn6p complex.
While a great deal is known in S.cerevisiae about the mechanisms Snf1p and Mig1p use to regulate galactose and other gene expression, it remains to be determined if these proteins function in similar signal transduction pathways and similar mechanistic modes in other fungi and in more complex eucaryotes. Data from mammals demonstrate that the Snf1p homolog, AMP-activated protein kinase, performs unique functions and regulates isoprenoid and fatty acid biosynthesis (19 ). Although K.lactis is closely related to S.cerevisiae on an evolutionary time scale (26 ), the two organisms have experienced different selective pressures and are not likely to use Snf1p and Mig1p in identical ways, particularly to regulate galactose metabolic genes, since K.lactis but not S.cerevisiae evolved to utilize lactose as a carbon source.
MATERIALS AND METHODS
Yeast strains and growth media
The S.cerevisiae (MCY1845) and K.lactis strains used in these studies are listed in Table 1 . Strain JSD1 was derived from strain JA6 by one step gene replacement (4 ) of the wild-type KlSNF1 chromosomal allele with the klsnf1-[Delta]1 deletion allele, which has nucleotides -143 to +1629 replaced with a 1.1 kb DNA fragment carrying the S.cerevisiaeURA3 (ScURA3) gene. The klsnf1-[Delta]1 allele, released from pBDsnf1 by cleaving the EcoRI sites, was transformed into strain JA6 followed by selection for Ura+ transformants. Because homologous recombination is less frequent in K.lactis than in S.cerevisiae, Ura+ transformants were screened for Lac- and Gal- cells by replica plating and cells with these phenotypes were then analyzed by Southern blotting to confirm that the SNF1 locus had been replaced by the klsnf1-[Delta]1 deletion allele. Strain JSD1/R is a Ura- derivative of strain JSD1 isolated for resistance to 5"-fluoroorotic acid (27 ).
Genotype and origin of yeast strains used in these studies
Name
Genotype
Source
JA6
MAT[alpha] ade trp1 ura3
(3)
SD12
MAT[alpha] trp1 ura3 lac4
(3)
JSD1
Derivative of JA6 carrying snf1-[Delta]1
this work
JSD1/R
Ura- derivative of JSD1
this work
JSD2
Derivative of JA6 carrying snf1-[Delta]1 mig1-[Delta]1
this work
JSD2/R
Ura- derivative of JSD2
this work
JSD3
MATa ade ura3 mig1-[Delta]1
this work
JSD4
Derivative of JA6carrying snf1-[Delta]1 gal1-10
this work
JSD5
Derivative of JA6carrying gal1-10
this work
JSD6
JA6 carrying GAL1-11
this work
JSD7
JA6 carrying snf1-[Delta]1 GAL1-11
this work
MCY1845
MATa snf1-[Delta]10 ade2-101 ura3-52 SUC2
(44)
The snf1 mig1 double deletion strain JSD2 was derived from strain JSD1. The KlMIG1 gene was obtained by amplification of JA6 chromosomal DNA using the PCR and two primers, 5"-CGGAATTCCGTGCGATTAGGTCAGTTCA and 5"-CGGA- ATTCCGGTGTTCATCGATAGTCGT, which have an EcoRI site added to their 5"-end to facilitate cloning. The sequence of the primers was based on the published KlMIG1 DNA sequence (28 ). The amplified KlMIG1 gene was cloned into the EcoRI site of pBLUESCRIPT (In Vitrogen, San Diego, CA) to give pBDMIG1. The region between the two NdeI sites within KlMIG1, nucleotides +87 to +1109, was replaced with a 1.4 kb DNA fragment carrying the ScTRP1 geneto yield the klmig1-[Delta]1 allele which is carried in pBDmig1. pBDmig1 DNA was digested with EcoRI and transformed into strain JSD1 with selection for Trp+ transformants. Southern blot analysis confirmed that the KlMIG1 locus had been replaced by the klmig1-[Delta]1 allele. A Ura- derivative of strain JSD2, JSD2/R, was isolated by resistance to 5"-fluoroorotic acid.
The mig1 mutant strain JSD3 was constructed by crossing strains SD12 and JSD2, selecting diploids on medium lacking uracil and adenine (29 ), sporulating diploids and dissecting tetrads. Haploid Trp+ offspring were identified and the presence of the ScTRP1-marked klmig1-[Delta]1 allele was verified by Southern blot analysis.
Strains JSD6 and JSD7 carry the KlGAL1-11 allele in which the putative Mig1 binding site, the GC-AT-BOX, of the GAL1 promoter are inactivated by multiple mutations (Fig. 1 ). These strains were constructed in several steps. First, intermediate strains JSD4 and JSD5 were made by replacing the wild-type KlGAL1 promoter with the klgal1-10 allele, having nucleotides -488 to +119 replaced with the ScURA3 gene. This allele was made by cloning a 1.4 kb XbaI-BamHI DNA fragment of KlGAL1 into the cognate sites of pBLUESCRIPT, yielding pBSKlgal, which was cut at the unique BglII site (+119 relative to the GAL1 start codon) and the BspMI site (-488 relative to the GAL1 start codon), treated with Klenow DNA polymerase I to make the ends blunt, and ligated to a 1.1 kb DNA fragment carrying ScURA3 to give pBSKlgalURA3. The klgal1-10 allele, released from pBSKlgalURA3 as a SmaI and SacII DNA fragment, was transformed into strain JSD1/R with selection for Ura+ cells, followed by screening for Gal- Lac- cells. Only two out of nearly 2000 Ura+ transformants were Gal- Lac-, and one of these was designated strain JSD4. Strain JSD5 was made by crossing strains SD12 and JSD4, sporulating diploids, and identifying Ura+ Gal- Lac- offspring. The presence of the klgal1-10 allele in strains JSD4 and JSD5 was verified by Southern blotting (data not shown).
Gene isolation and reporter plasmids
The KlSNF1 gene was selected from a K.lactis genomic library carried on the multi-copy vector pAB24 (31 ). Portions of the original plasmid carrying KlSNF1 were subcloned into YEp352 (32 ) and tested for complementation of the Suc- phenotype of strain MCY1845. pBSNF1 carries the KlSNF1 gene on a 3.1 kb EcoRI DNA fragment inserted into the EcoRI site of pBLUESCRIPT. The nucleotide sequence of both strands of the 3.1 kb fragment was determined using a commercial DNA sequencing kit (United States Biochemical Corp., Cleveland, OH).
The reporter plasmid pKlgal4CAT contains the KlGAL4 promoter fused to the coding region of the chloramphenicol acetyl transferase (CAT) gene (6 ). The reporter plasmid pC80GUS contains the KlGAL80 promoter fused to the [beta]-glucuronidase (GUS) gene (14 ).
Enzyme assays and miscellaneous procedures
For assaying CAT activity, transformed yeast cells were pre-grown overnight to saturation in defined medium supplemented with the carbon sources indicated in the text. Saturated overnight cultures were diluted into 10 ml of fresh medium to an optical density at 600 nm (OD600) of 0.25-0.30 and grown to an OD600 of 0.7-0.8. Cells were centrifuged for 5 min at 5000 g at 4oC, and suspended in 300 [mu]l of ice-chilled breaking buffer (50 mM sodium phosphate, pH 7.5, 5% glycerol, 1 mM EDTA and 1 mM PMSF; this buffer was found to give more reproducible assays than a previously described buffer (6 ). An equal volume of 0.5 mm diameter acid-washed glass beads was added and the cells were disrupted by vortexing at 4oC for 10-15 min. Samples were centrifuged at 4oC for 5 min, and the supernatant fluid was used immediately for enzyme assay (6 ). A unit of CAT activity is defined in Table 5 .
GUS activity was measured as described by Jefferson (33 ) using cells and extracts prepared as for the CAT assay. Previously described assays were used to measure [beta]-galactosidase activity (4 ), transferase, epimerase and galactokinase activity (34 ) and lactose transport (35 ).
Yeast cells were transformed using the procedure of Gietz et al. (36 ).
RESULTS
Isolation of a SNF1 homolog from K.lactis
To isolate the KlSNF1 gene, S.cerevisiae strainMCY1845 (relevant features: snf1-[Delta]10, Suc-) was transformed with a K.lactis genomic DNAlibrary, Ura+ transformants were selected, pooled, and re-selected for Suc+ cells. To determine if a plasmid- borne gene was responsible for Suc+ colonies, plasmid DNA from 10 Suc+ transformants was recovered by transformation into and purification from Escherichia coli, followed by retransformation into strain MCY1845. All MCY1845 Ura+ transformants were Suc+, indicating that a plasmid-borne gene was responsible for the Suc+ phenotype. The plasmids carried the same 10 kb insert as determined from restriction endonuclease digestion. The complementing gene was localized within a 3.1 kb EcoRI restriction fragment by subcloning and complementation testing (data not shown).
The DNA sequence of the 3.1 kb fragment was determined and, when analyzed, showed one open-reading frame, predicted to encode a protein of 602 amino acids with a mass of 68 463 Da. This predicted protein is identical to one recently identified as the K.lactis Snf1 protein (KlSnf1p) (15 ). KlSnf1p shows 75% amino acid identity with the S.cerevisiae Snf1 protein (ScSnf1p), indicating that the two proteins are structural homologs.
There appears to be only one SNF1 coding sequence in K.lactis, since a Southern blot made using genomic DNA cut with SspI showed one band of hybridization with the KlSNF1-containing 3.1 kb EcoRI DNA fragment radiolabeled with 32P (data not shown).
Impaired carbon utilization in a Klsnf1 mutant strain
To determine if the KlSnf1 protein is necessary for expression of the lactose-galactose regulon, the growth rate of a Klsnf1 deletion strain (JSD1) was measured in a medium having lactose or galactose as the carbon source. The deleted strain grew much slower than the non-deleted strain on both sugars (Table 2 ), indicating that full expression of the lactose-galactose regulon requires KlSnf1p. Strain JSD1 grew, albeit slower than wild-type strain JA6, with sucrose as the carbon source. This result is in contrast to the situation in S.cerevisiae where SNF1 (sucrose non-fermenting) is required for utilization of sucrose (37 ). The Klsnf1 mutant strain JSD1, like a Scsnf1 mutant strain, grew slightly slower on glucose than did the wild-type strain JA6. Strain JSD1 failed to grow at all when sorbitol, raffinose, maltose, glycerol or ethanol were used as the carbon source (data not shown). These results demonstrate that Snf1p plays a central role in carbon metabolism in K.lactis, as it does in S.cerevisiae.
aCells were grown in defined medium supplemented with the indicated carbon source added to a final concentration of 2% (w/v), sonicated to dissociate clumped cells, and diluted into fresh medium to give a starting OD600 of 0.2-0.3. The doubling time is defined as the time in min for the OD600 to double. Mean values +- the standard deviation represent data from three independent determinations.
Deletion of Klsnf1 reduces expression of the linked GAL1, GAL7 and GAL10 genes
To begin to understand why the KlSNF1 gene is necessary for rapid growth on lactose and galactose, we determined which structural gene(s) in the lactose-galactose regulon requires the KlSnf1 protein for normal expression under uninduced (basal), induced and glucose-repressed conditions. The Klsnf1-deleted strain JSD1 had about the same uninduced level for the three enzymes as the wild-type strain JA6, but the induced level was reduced to 23-31% of that wild-type strain (Table 3 ). These results show that KlSNF1 is essential for full induction of expression of these three linked genes. In addition, the activities of the three enzymes are still repressed by glucose in the deletion strain, implying that KlSNF1 is not essential for maintaining glucose repression of these three genes.
Deletion of Klsnf1 reduces expression of LAC4 and LAC12
We next measured expression of the LAC4 ([beta]-galactosidase) and the LAC12 (lactose permease) genes which are transcribed in opposite direction from the same promoter (9 ). Deletion of Klsnf1 greatly reduced [beta]-galactosidase activity under both uninduced and induced conditions (Table 4 ), but the induction mechanism was still operating on the LAC4 gene, although only about half as effectively as in wild-type cells (60-fold induction in wild-type JA6 cells compared with 24-fold in JSD1 cells).
aStrainsJA6, JSD1, JSD2 and JSD3 were transformed with the reporter plasmid pKlgal4CAT containing the full length KlGAL4 promoter fused to the CAT coding region. The transformants were grown in the selective medium with the indicated carbon source: Uninduced: 2% sucrose; Induced: 2% sucrose + 2% galactose; Repressed: 2% sucrose + 2% galactose + 2% glucose. Units of CAT activity are % conversion of substrate (c.p.m. measured in the organic phase expressed as a percentage of total c.p.m.) per mg protein per 45 min. Numbers in parenthesis represent the percentage of enzyme activity relative to the value for strain JA6. Mean values " standard deviation represent at least three independent determinations.
aStrains were transformed with pC80GUS which carries the KlGAL80 promoter fused to the GUS coding region. Transformants were grown in selective medium lacking Trp or Ura and supplemented with the carbon sources as described in the legend to Table 2. Numbers in parenthesis represent the percentage of enzyme activity relative to the value for JA6. Mean values " standard deviation represent at least three independent determinations.
The Klsnf1 mutation had its greatest effect on lactose transport activity; basal activity was reduced and no induction occurred (Table 4 ). To verify that lactose transport was not induced in JSD1 mutant cells, the kinetics of lactose uptake were followed over a 2 h period. During this time JSD1 cells failed to accumulate lactose (Fig. 2 ), indicating greatly reduced expression of LAC12.
We conclude from the data presented in Table 4 and Figure 2 that the KlSNF1 gene is required for a normal basal level of LAC4 and LAC12 expression. Full induction of LAC4 expression requires KlSNF1 but ~10% of the inducible expression is independent of KlSNF1. Induction of LAC12 expression is entirely dependent upon KlSNF1.
KlSNF1 is not necessary for glucose repression of LAC4 expression since [beta]-galactosidase activity was nearly the same under uninduced and glucose repressed conditions in the Klsnf1 mutant JSD1 (Table 4 ). Because LAC12 expression was not induced in strain JSD1, glucose repression could not be evaluated (Table 4 ).
Effect of the Klsnf1 deletion on expression of KlGAL4
Full induction of the lactose-galactose regulon requires autoactivation of KlGAL4 expression (6 ). To determine if KlSNF1 is necessary for autoactivation we measured KlGAL4 expression using a reporter gene in which the KlGAL4 promoter is fused to the coding region of the CAT gene. This reporter gene, when carried on a single-copy CEN vector, has been shown to be a very sensitive way to measure small changes in KlGAL4 expression (6 ). Induction of KlGAL4 expression in mutant strain JSD1 was reduced to 22% of the level seen in wild-type strain JA6 (Table 5 ).
Viewed another way, the level of KlGAL4 expression in the mutant strain under inducing conditions (15.6 CAT units) was only slightly above the uninduced level of the wild-type strain (11.4 CAT units). We conclude from the data shown in Table 5 that KlSNF1 is required for activation of KlGAL4 expression during induction of the lactose-galactose regulon.
Effect of the Klsnf1 deletion on expression of KlGAL80
KlGal4p binds to two UAS sequences in the KlGAL80 promoter and regulates its expression (14 ,38 ). Thus, we expected a klsnf1 deletion strain to show impaired KlGAL80 expression. A reporter plasmid, pC80GUS, containing the KlGAL80 promoter fused to the [beta]-glucuronidase (GUS) coding region was used to measure the effect of KlSNF1 on KlGAL80 expression. The klsnf1-deleted strain JSD1 had about the same GUS activity as the wild-type strain under the uninduced condition, indicating that mutation of klsnf1 had no effect on basal expression of KlGAL80 (Table 6 ). In contrast, GUS activity was induced only 13-fold in mutant strain JSD1 compared with the 48-fold induction seen in wild-type strain JA6. We conclude from these data that full induction of KlGAL80 expression requires the KlSNF1 gene. Glucose repressed GUS activity, indicating that KlSNF1 plays no role in maintaining repression of KlGAL80 expression (Table 6 ).
Role of the KlMig1 protein in expression of the lactose-galactose regulon
The data presented thus far show that KlSnf1p is needed for full induction of the lactose-galactose regulon but they do not indicate how the protein is working in the induction pathway. The ScSnf1 protein is known to exert some of its effects on transcription through the ScMig1 protein, thought to act by repressing transcription (reviewed in 18 ). The ScMig1 protein is known to bind the ScGAL4 and the ScGAL1 promoters, thereby repressing expression of the galactose regulon (23 ).
We first determined if expression of the lactose-galactose regulon is regulated by KlMig1p. This was done by measuring expression of the lactose-galactose genes in a klmig1-deleted strain, JSD3. The klmig1 mutation had the same general effect on expression of the GAL1, GAL7, GAL10, LAC4 and LAC12 genes; expression increased under uninduced and induced conditions and glucose did not repress expression as well as in the wild-type strain JA6 (Tables 3 and 3 ). These data indicate that KlMig1p normally acts to repress expression of these genes under uninduced, induced and glucose-repressed growth conditions.
We next determined if KlMig1 acts downstream of KlSnf1, as does the ScMig1 protein when it regulates the galactose regulon of S.cerevisiae, or whether it acts upstream. Action downstream of KlSnf1 would be indicated if a klmig1 mutation restored induction of LAC-GAL gene expression in a klsnf1 strain (23 ). The same trends were observed for expression of the GAL1, GAL7, GAL10, LAC4 and LAC12 genes and we will focus on GAL1, since as we show below, its expression appears to be of central importance to the regulon. The klsnf1 mutant strain JSD1 showed a 4.3-fold induction of GAL1 expression (Table 3 ), much less than the 19-fold induction seen in wild-type JA6 cells. The klsnf1 klmig1 double mutant strain JSD2 gave an 11.6-fold induction, showing that the klmig1 mutation can partially reverse the effect of the klsnf1 mutation. Thus, KlMig1p acts downstream of KlSnf1p in the signaling pathway for induction of the LAC-GAL genes.
The klmig1 mutation (strain JSD3) had no effect on expression of KlGAL4 in the uninduced and induced states but it caused a complete loss of glucose repression (Table 5 ). Similar trends were seen for KlGAL80 expression (Table 6 ). The implications of these results will be considered in the Discussion. Lastly, uninduced or basal expression of both KlGAL4 and KlGAL80 was not changed significantly by deletion of either snf1 or mig1 or both genes (Tables 5 and 3 ), indicating that basal expression of KlGAL4 and KlGAL80 is regulated in a manner independent of SNF1 and MIG1.
KlMig1p acts through the GAL1 promoter to govern expression of the lactose-galactose regulon
KlGAL1 encodes the Leloir pathway enzyme galactokinase, necessary for phosphorylation of galactose (34 ). In addition, the protein has a second, independent activity that is necessary for induction of the regulon (13 ). This second activity probably enables KlGal1p to bind KlGal80p, a reaction requiring both galactose and ATP (12 ). One model that explains these data envisages KlGal1p acting as a molecular sensor of galactose that switches KlGal4p between transcriptionally inactive and active forms. In the uninduced state, KlGal80p would complex with KlGal4 (39 ), thereby preventing transcription activation. During induction of the lactose-galactose regulon the inducer galactose would bind to KlGal1p and this complex would then bind to KlGal80p thereby switching KlGal4p from an inactive to an active form capable of turning on transcription of genes in the lactose-galactose regulon (12 ).
If this model is correct, it provides an explanation for our observation (strain JSD3, Tables 3 and 3 ) that deletion of klmig1 increases the basal and induced level of LAC-GAL gene expression and partially abrogates glucose repression. We imagine that in the uninduced state KlMig1p binding to the KlGAL1 promoter prevents expression. Early during induction, the repressive effect of KlMig1p is switched off so that transcription of KlGAL1 begins, followed by production of KlGal1p. KlGal1p in conjunction with galactose and ATP then complexes with KlGal80p, thereby enabling KlGal4p to activate expression of the other genes in the regulon.
As first pointed out by Cassart et al. (28 ), the KlGAL1 promoter contains a potential Mig1 binding site consisting of a GC box (G/C C/T G G G/A G) preceded on the 5" side by an A-rich region (21 ); we found no other promoters in the regulon with a Mig1p binding site. If this model is correct, it predicts that mutation of the KlMig1p binding site in the KlGAL1 promoter (Fig. 1 ) should partially abrogate glucose repression and cause a small increase in basal and induced expression of the regulon. As predicted by this hypothesis, we found that glucose repression of [beta]-galactosidease and galactokinase activity was partially abolished and both basal and induced expression were slightly increased in mutant strain JSD6 compared with wild-type strain JA6 (Tables 3 and 3 ).
We also determined if mutation of the GC-AT box region of the KlGAL1 promoter abrogated glucose repression of KlGAL4 expression as was seen in the klmig1 deletion strain JSD3 (Table 5 ). Glucose repression of KlGAL4 expression was abrogated in strain JSD6 compared with wild-type strain JA6 but not to the same extent as in strain JSD3 (Table 5 ). The difference between strains JSD3 and JSD6 could result from low affinity binding of KlMig1p to the mutant KlGAL1 promoter sequence in strain JSD6, whereas deletion of klmig1 would completely abolish promoter binding (strain JSD3).
DISCUSSION
The S.cerevisiaeSNF1 gene plays a global role in regulating carbon utilization (18 ,40 ). One aim of our research was to determine if SNF1 plays a similar role in K.lactis and, in addition, if it plays specific roles in induction and glucose repression of the lactose-galactose regulon. Based upon the inability of the Klsnf1-deleted strain JSD1 to utilize a variety of fermentable and non-fermentable carbon sources (Table 2 and data not shown) we conclude that SNF1 is a global regulator of carbon utilization in K.lactis. One difference between S.cerevisiae and K.lactis is that utilization of sucrose requires SNF1 in S.cerevisiae whereas this is not the case in K.lactis (Table 2 ). The physiological reason for this difference is not apparent. Goffrini et al. (15 ) also noted that a Klsnf1 (fog2) mutant strain fails to utilize numerous carbon sources including galactose. Our data agree with Goffrini et al. except that our klsnf1 mutant grew slowly on galactose. This difference may be due to the higher concentration of galactose (2%) we used compared with the lower concentration (0.5%) used by Goffrini et al.
Snf1p is necessary for full induction of the regulon
Our data show that KlSnf1p is essential for full induction of the lactose-galactose regulon. This conclusion is based both upon the slow growth rate of a klsnf1-deleted strain when lactose or galactose are the carbon source (Table 2 ) and upon analysis of the expression of the structural genes in the regulon including the GAL1, GAL7 and GAL10 gene cluster (strain JSD1 compared with JA6, Table 3 ) and the divergently transcribed LAC4 and LAC12 genes (Table 4 ), plus the positive regulator GAL4 (Table 5 ) and the negative regulator GAL80 (Table 6 ).
Since cells deleted for klsnf1 grow when lactose or galactose are the only carbon source and partially induce most genes in the regulon (Tables 3 -6 ), there must be a SNF1-independent mechanism that can partially activate expression of the regulon. In contrast, snf1-deleted S.cerevisiae cells do not grow on galactose (40 ) indicating that expression of the regulon is completely dependent upon SNF1.
Mig1p is necessary for full repression by glucose
The conclusion that KlMig1p is necessary for full glucose repression of the lactose-galactose regulon is based upon the inability of the klmig1 deletion strain JSD3 to repress expression of the structural and regulatory genes as well as the wild-type strain JA6 under glucose repressing conditions (Table 3 ). Because the expression level of GAL1, GAL7, LAC4, LAC12 and GAL80 (Tables 3 -5 ) under glucose repressing conditions (glucose plus galactose) is still below the level seen under inducing conditions, there must be a MIG1-independent mechanism for glucose repression. This mechanism does not affect expression of GAL10 (Table 3 ) or GAL4 (Table 5 ). A MIG1-independent mechanism for glucose repression of SUC2 expression has also been seen in S.cerevisiae (41 ).
Data for the behavior of the snf1 mig1 double mutant strain in comparison with the single mutant strains (Tables 3 and 3 ) argue that KlMig1p acts downstream of KlSnf1p in the induction (derepression) pathway. A similar epistatic relationship has been found for the two proteins in the pathway for derepressing expression of the galactose regulon in S.cerevisiae (22 ,23 ) and many other experiments argue that ScMig1p acts downstream of ScSnf1p (reviewed in 18 ). However, the situtation in K.lactis is probably not this simple, because the Klmig1 mutant strain (JSD3) does not fully repress galactose gene expression under glucose repressing conditions, while the Klsnf1 Klmig1 double mutant does fully repress (Tables 3 and 3 ). These data indicate that, with respect to glucose repression, KlSnf1 is epistatic to KlMig1. One interpretation of the glucose repression data is that there is a Mig1p-independent, KlSnf1p-dependent glucose repression pathway operating on the GAL genes in K.lactis.
If KlSnf1p worked solely through KlMig1p we would expect that under inducing conditions the snf1 mig1 double mutant strain would have gene expression levels that are similar to the wild-type values, but this is not the case for any of the genes in the regulon (compare strains JSD2 and JA6 in Table 3 ). These data add further support to the hypothesis that KlSnf1p has a second, KlMig1p-independent pathway, for activating expression of the regulon or that there is a KlSnf1p-independent pathway. Of these two hypotheses, the KlMig1p-independent pathway is supported by the data for the klmig1 mutant strain JSD3. Expression of the structural genes in strain JSD3 under inducing conditions is above the wild-type level (Tables 3 and 3 ), indicating that when Klmig1p is removed, expression of the structural genes can be fully induced. Recent data identify an ScSnf1p pathway that does not require ScMig1p. In this pathway ScSnf1p modulates the activity of the ScSip4 transcription activator (42 ). Genetic evidence suggests that ScSnf1p interacts with two other transcription activators, Msn2p and Msn4p (43 ), so there may be homologs of one or more of these proteins in K.lactis which might be necessary for full induction of the lactose-galactose regulon.
KlMig1p acts at the KlGAL1 promoter
In S.cerevisiae, Mig1p confers glucose repression on the galactose regulon by binding to the ScGAL4 and ScGAL1 promoters (22 ,23 ). A search of the known promoters in the lactose-galactose regulon of K.lactis identified only one putative Mig1p binding site located in the divergently transcribed KlGAL1 and KlGAL10 promoter (Fig. 1 ). Mutation of this site resulted in a strain, JSD6, that behaved qualitatively like the klmig1 deletion strain JSD3 as measured by expression of KlGAL1 (galactokinase activity, Table 3 ) and KlGAL4 ([beta]-galactosidase activity, Table 4 ). The value for these two enzymes was derepressed almost as much in strain JSD6 as in strain JSD3. The difference between the two strains could reflect low affinity binding of KlMig1 to the mutated promoter site in strain JSD6. Thus, these data support the hypothesis that KlMig1p regulates expression of the lactose-galactose regulon primarily by binding to the KlGAL1 promoter. Although it seems unlikely that KlMig1p regulates the LAC-GAL genes in some additional way, our data do not eliminate this possibility.
A model for regulation of the lactose-galactose regulon
Based upon the data presented here and upon data derived from S.cerevisiae, we propose (Fig. 3 ) that KlSnf1p acts in a signaling pathway that terminates with the KlMig1 repressor protein bound to the divergently transcribed KlGAL1-10 promoter (Fig. 1 ). When glucose is present in the culture medium KlMig1p is bound to the KlGAL1 promoter and transcription is repressed, even if inducer is present also in the culture medium (glucose repressing conditions), by a pathway requiring KlSnf1p. In the absence of glucose and the presence of inducer, KlMig1 repression is switched off by a KlSnf1p-dependent pathway and transcription of KlGAL1 and perhaps KlGAL10 is increased to produce klGal1p. Regulating the concentration of KlGal1p in this manner provides a mechanism for switching expression of the rest of the genes in the regulon on and off.
ACKNOWLEDGEMENTS
This work was support by grant MCB-9219839 from the National Science Foundation. We thank Drs Karin Breunig and Wolfgang Zachariae for strains and plasmids.
REFERENCES
1 Dickson,R.C. and Riley,M.I. (1989) In Barr,P.J., Brake,A.J. and Valenzuela,P. (eds), Yeast Genetic Engineering. Butterworth, Boston, pp. 19-40.
2 Dickson,R.C. and Markin,J.S. (1980) J. Bacteriol. 142, 777-785.MEDLINE Abstract
26 Hendriks,L., Goris,A., van de Peer,Y., Neefs,J.-M., Vancanneyt,M., Kersters,K., Berny,J.-F., Hennebert,G.L. and de Wachter,R. (1992) System. Appl. Microbiol. 15, 98-104.
27 Boeke,J.D., LaCroute,F. and Fink,G.R. (1984) MGG197, 345-346.
28 Cassart,J.P., Georis,I., Oestling,J., Ronne,H. and Vandenhaute,J. (1995) FEBS Lett. 371, 191-194.MEDLINE Abstract
29 Sheetz,R.M. and Dickson,R.C. (1980) Genetics 95, 877-890.MEDLINE Abstract
