INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

In the yeast Saccharomyces cerevisiae, many pathways are controlled by the action of a large family of Cys₆ Zn(II)₂ binuclear proteins that share a similar domain organization. Each family member has a well-characterized DNA binding domain containing six cysteine residues that combine with two Zn²⁺ ions, forming the Cys₆ zinc cluster. Adjacent to the DNA binding domain is a dimerization domain, followed by a central domain of unknown function and an acidic activation domain, frequently located in the carboxy terminus (33). Among the best-characterized members of this class are Gal4p, Put3p, Leu3p, Ppr1p, Hap1p, and Cha4p. These proteins regulate galactose metabolism, proline utilization, leucine biosynthesis, pyrimidine biosynthesis, oxygen-responsive genes, and the catabolism of hydroxylated amino acids, respectively (12, 16, 19, 23, 25, 45).

Of the 58 members of this family, Gal4p has been the most thoroughly studied. Regulated production of the galactose-utilizing enzymes results from an interplay between Gal4p, the repressor Gal80p, and a third protein, Gal3p, in the presence of the inducer galactose (5, 18, 24, 30). Gal4p binds as a dimer to a 17-bp upstream activation sequence (UAS_GAL) that is characterized by the sequence 5'-CGG-N₁₁-CCG-3' (15), and the specificity of that binding is found in the linker region between the DNA binding domain and the dimerization domain (11, 31).

The Put3p transcriptional regulator is required to activate genes that allow the growth of S. cerevisiae on proline as the sole nitrogen source (reviewed in reference 28). Put3p binds as a dimer to a 16-bp promoter sequence called UAS_PUT with the structure 5'-CGG-N₁₀-CCG-3' (13, 35). Although Put3p binds constitutively to its target sequences in vivo (2), the PUT genes are maximally expressed only in the presence of proline and in the absence of preferred sources of nitrogen (8, 9, 21, 42, 43). Unlike Gal4p, Put3p does not associate with a system-specific repressor and appears to regulate transcription by posttranslational modifications and conformational changes (17; S. G. des Etages et al., unpublished data).

Reece and Ptashne (31) showed that bacterially synthesized amino-terminal fragments of Gal4p could bind in vitro to sequences containing CGG triplets separated by 10 and 12 bp with an affinity about 1/10 that of the natural Gal4p binding site with the 11-bp spacer. These authors also reported that an amino-terminal fragment of Ppr1p showed some flexibility in its recognition site (5'-CGG-N₆-CCG-3') binding to CGG triplets with 7- and 8-bp spacers. However, amino-terminal Put3p fragments could only bind in vitro to sequences containing CGG separated by 10 bp, which is the spacing found in nature. Vashee et al. (40) compared the affinity of a purified Gal4p fragment to bind a variety of UASs in vitro to the ability of full-length Gal4p to activate reporter genes with those same UASs in vivo. They observed that many sites to which a fragment of Gal4p was able to bind with substantial affinity in vitro did not work in vivo to permit transcription by full-length Gal4p.

Given these findings, the aim of this study was to test whether Gal4p and Put3p can functionally substitute for one another in vivo. In the in vivo studies cited above, the activator proteins being compared were present in the cells and were competent to bind the test sequences in competition with one another. In this report, we provide evidence that under the appropriate conditions, Gal4p can activate the PUT genes, but only in a strain lacking Put3p. Put3p, however, could not under any of our test conditions activate the GAL genes. Our results confirm the in vitro data showing that Gal4p is more promiscuous than Put3p in its DNA binding ability. This suggests that under appropriate circumstances, Gal4p may be able to function in place of a related family member to activate expression.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Plasmids. The plasmids used in this study are listed in Table 1. Plasmid pABC4 (35) contains a PUT2-lacZ reporter gene with the wild-type PUT2 promoter extending to bp 890 upstream of the first codon of the open reading frame. Plasmid pABC18 (35) is identical to plasmid pABC4, except that it carries a 35-bp deletion of the Put3p binding site, UAS_PUT, from bp 164 to 128. Plasmid pMDB2 contains the GAL4 DNA binding domain (GBD, codons 1 to 147) fused to the activation domain of PUT3 (codons 890 to 979) and was constructed in two steps, as follows. Plasmid pSDB4 (13), carrying GBD-PUT3(890-979) under the control of the ADH1 promoter, was digested with ClaI and religated to remove the 450-bp CEN-ARS region, yielding plasmid pMDB1. Plasmid pMDB1 was digested with BglII and ligated to the 2.8-kb BglII LEU2 fragment from plasmid YEp13 (6) to form pMDB2, a yeast integrating plasmid carrying ADH1p-GBD-PUT3(890-979) and LEU2.

Yeast strains and growth media. The S. cerevisiae strains used in this study are listed in Table 2. The gal4 strain MB1478 was constructed as follows. Strain DB27-7C was transformed with a SacI-KpnI fragment carrying GAL4 flanking sequences ligated to hisG-URA3-hisG from plasmid pBM2387. Ura⁺ transformants were selected; those that had secondarily acquired a Gal phenotype were plated on minimal plates containing 5-fluoroorotic acid to select for those that had popped out the URA3 gene. Strain MB1478 was shown to carry a deletion of GAL4 by DNA hybridization analysis. It was crossed to strain DB27-5B to form diploid strain MB838, which was dissected to yield gal4, put3, and gal4 put3 mutant strains.

Genetic analysis. Mating, sporulation, and tetrad analysis were carried out using standard procedures (34). Yeast transformation was performed using the method of Gietz and Schiestl (14). Plasmid DNA was prepared from Escherichia coli by the method of Birnboim and Doly (4). E. coli transformation was performed by the CaCl₂ method (10).

Growth of yeast strains, extract preparation, and -galactosidase assays. The cultivation of yeast strains, the preparation of extracts, and -galactosidase assays have been described previously (25). The units of specific activity are nanomoles of o-nitrophenol formed per minute per milligram of protein. The numbers are the averages of two or three determinations; variation was usually 20% or as listed in the table footnotes. Protein concentrations of crude extracts were determined by the method of Bradford (7), using crystalline bovine serum albumin as the standard.

DNA hybridization. Yeast genomic DNA was isolated (29), separated on 0.7% agarose gels, transferred to nylon membranes (Schleicher and Schuell), and blotted according to the method of Southern (36). For strains containing the GBD-PUT3(890-979) DNA, the membrane was probed with a 0.4-kb fragment containing codons 890 to 979 of PUT3. For strains with disruptions at the GAL4 locus, the probe was a 0.8-kb fragment corresponding to the 5' untranslated region of GAL4. Both probes were radioactively labeled using the Multiprime labeling kit (Amersham) and [-³²P]dATP.

Immunoblotting. Proteins from extracts of yeast were separated on sodium dodecyl sulfate-polyacrylamide gel electrophoresis gels according to the method of Laemmli (22) and transferred to polyvinylidene difluoride membranes according to methods described by Clontech. Anti-Put3p antiserum (44) was diluted 1:2,000. The ECL chemiluminescence protocol (Amersham) was used to detect the proteins, following the instructions of the manufacturer.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Behavior of a Gal4p-Put3p hybrid protein. S. cerevisiae strains lacking Put3p, the proline utilization pathway-specific activator, fail to grow on proline as the sole source of nitrogen because the PUT1 and PUT2 genes, which encode the enzymes of the pathway, are not expressed. In the absence of Put3p, strains growing on alternative nitrogen sources (e.g., ammonium sulfate or GABA) produce a level of expression of PUT1 and PUT2 that is approximately half the level observed in the wild-type strain, depending on growth conditions and strain background. This low-level expression may be the result of activation by other regulators.

View larger version (14K): [in this window] [in a new window]   FIG. 1.   Growth rates of strains carrying Put3p or GBD-Put3p(890-979). Three independent transformants were grown at 30°C in minimal medium containing glucose and proline. Growth was monitored with a Klett-Summerson colorimeter (blue filter). Similar results were obtained with three independent transformants; one curve for each is shown. Each strain also carried plasmid YEp13 to complement the leucine auxotrophy. , strain DB27-3A (put3) transformed with high-copy-number PUT3 plasmid pDB37; , strain DB27-3A transformed with low-copy-number PUT3 plasmid pDB107; , strain DA1000 [GBD-PUT3(890-979)] transformed with plasmid YEp24.

Effect of carbon sources on the expression of PUT2. If Gal4p were involved in regulating expression of the PUT genes, we would expect to see an effect of different carbon sources on the expression of the PUT2 gene. Strains were grown on minimal media containing glucose, glycerol plus ethanol, raffinose, or galactose as the sole carbon source, with ammonia as the sole nitrogen source. Expression was measured using a genomic PUT2-lacZ reporter gene. In a wild-type strain, the expression of PUT2 did not vary significantly as a function of carbon source (Table 3). When a put3 strain was grown with glucose, glycerol plus ethanol, or raffinose, PUT2-lacZ expression dropped by about 50%, in agreement with previous studies of strains in which the system-specific regulator was missing (13, 25). In the put3 strain grown on galactose, PUT2-lacZ expression increased fourfold over that observed on glucose and to a level somewhat higher than that of the wild-type strain grown under the same conditions (Table 3). This result suggested that when Put3p is not present, Gal4p can bind upstream of PUT2-lacZ and turn on gene expression.

Analysis of PUT2 expression in galactose induction. To determine if Gal4p was responsible for the increase in expression described above, strains carrying the four combinations of deletion alleles of PUT3 and GAL4 were examined. Cultures were grown in media containing a low glucose concentration (0.05%) with or without galactose (2%). Under these conditions, glucose repression is minimized, and an increase in expression is due to galactose induction. In the absence of both Put3p and Gal4p, there is a background level of expression of -galactosidase that may be due to the contribution of other yet-unrecognized factors. The background level (22 U in the absence of galactose and 35 U in the presence of galactose) was subtracted in order to compare the specific contribution of each activator. In the absence of galactose, Put3p is responsible for all of the expression observed, in the presence or absence of Gal4p (Table 4, Gal). In the presence of galactose, both Gal4p and Put3p contribute to the expression of PUT2 (Table 4, +Gal).

put3

put3 gal4

Put3p cannot replace Gal4p in activation of the GAL genes. To determine if Put3p was able to bind UAS_GAL elements in vivo to activate the GAL genes, GAL1-lacZ expression was measured in strains carrying a deletion of GAL4. Strains MB838-2C (PUT3 gal4) and MB838-3A (put3 gal4) were transformed with the low-copy-number GAL1-lacZ plasmid pBM746 or pMDB8 and grown on media containing GABA or GABA plus Pro as nitrogen sources with glucose as the carbon source. The -galactosidase specific activity was 1 under either condition measured and was the same whether PUT3 was present in the genome (single copy) or on a high-copy-number plasmid. The control, the congenic strain MB838-1A (put3 GAL4) carrying the same GAL1-lacZ plasmid, had a -galactosidase specific activity of 407 on a medium containing galactose and GABA.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The Cys₆ Zn(II)₂ binuclear cluster family of transcription factors controls a variety of pathways in S. cerevisiae. Pathway-specific regulation by these factors is maintained using a combination of differential affinity for DNA binding to particular UASs, interaction with small-molecule inducers, binding to repressor proteins, and posttranslational modifications. However, some of these fungal activators may be sufficiently similar to substitute for one another in vivo to a limited extent and under the appropriate conditions. In this report, we have demonstrated the ability of Gal4p to regulate the proline utilization pathway but only when Put3p, the system-specific regulator, has been removed.

Activation of PUT genes by Gal4p requires induction by galactose and the Put3p binding site, can be accomplished with the low levels of Gal4p made from a single GAL4 gene in the genome, and is not observed in the presence of Put3p. However, there is no reciprocity: Put3p did not activate the GAL genes under any of our test conditions, including overproduction of Put3p in the presence of the inducer proline. These proteins are closely related and share an amino-terminal DNA binding motif, a homologous central domain, and a carboxy-terminal acidic activation domain. Both recognize a binding site with two CGG triplets, separated by 11 bp (Gal4p) or 10 bp (Put3p).

Bacterially synthesized fragments of the DNA binding domains of Gal4p and Put3p fragments have been purified and analyzed by X-ray crystallography (27, 38) and nuclear magnetic resonance spectroscopy (3, 20, 41). Both proteins rely on conserved regions in the Zn₂ Cys₆ domain to contact the two inverted CGG triplets and a region downstream of the zinc-binding domain containing a linker and part of the dimerization domain to specify the number of base pairs in the spacer region. In vitro DNA binding experiments that measured the affinities of Gal4p, Put3p, and another family member, Ppr1p, for a variety of sites demonstrated that the Gal4p linker-dimerization element region permitted binding with a 10-fold-reduced affinity to sites with spacers that were 10 or 12 bp, rather than the naturally occurring 11 bp, but that Put3p was unable to bind any site having a spacing different from 10 bp (31).

Vashee et al. (40) compared the DNA binding affinity of Gal4p for a variety of GAL4 sites, both naturally occurring and mutated, in vivo and in vitro. They found that many sites to which Gal4p could bind with moderate or high affinity in vitro did not support Gal4p-activated transcription in vivo. In agreement with the results reported by Reece and Ptashne (31), Vashee et al. (40) found a 25-fold decrease in in vitro binding by a Gal4p fragment to a site with a 10-bp spacer, as compared to the 11-bp natural spacer. However, there was no detectable activation of transcription of a reporter gene carrying this 10-bp site by full-length Gal4p in vivo. These authors concluded that in many cases there was not a quantitative correlation between in vitro DNA binding and in vivo activation of transcription. The discrepancy might reflect differences in behavior between a recombinant fragment of Gal4p and the entire full-length protein, the use of unnatural sites, or the existence of another protein that can bind UAS_GAL in vivo.

Vashee et al. (40) also reported that Gal4p did not activate transcription in vivo when the 11-bp GAL UAS was replaced with a 10-bp site resembling the PUT UAS, although they did observe in vitro binding of DNA containing this 10-bp site by a Gal4p fragment. We were able to see Gal4p activation of the PUT genes in vivo, but only when there was no Put3p present. In the presence of Put3p, we saw no effect of activation by Gal4p (Table 4). In the experiments described by Vashee et al., it is likely that Put3p was present and binding its own site, protecting it from binding by Gal4p.

An unusually high level of background activity in the expression of certain reporter constructs led Vashee et al. (40) to suggest the existence of one or more other activators that could recognize some of their consensus GAL4 binding sites under noninducing conditions. We also have observed a higher level of background PUT2 expression in our put3 strains than in some other put3 mutants known to carry point mutations (13). This observation led to the suggestion that other regulators could activate the PUT genes only in the absence of Put3p; the mutant Put3 proteins that could still bind DNA were perhaps blocking the UAS from activation by other regulators, leading to a lower background level of expression.

In addition to its ability to activate the PUT genes, Gal4p can also cooperate with Put3p in activating gene expression. In a more recent study, Vashee et al. (39) used synthetic reporter constructs that contained binding sites for both Gal4p and Put3p spaced about 26 bp apart to show that these activators interacted synergistically at this artificial promoter. The authors argue that Gal4p and Put3p can bind to nearby sites in a nontraditional cooperative fashion that does not rely on their interacting with one another but perhaps by contacting different proteins of the transcriptional apparatus.