GAL 4 Mutations That Separate the Transcriptional Activation and GALSO - Interactive Functions of the Yeast GAL 4 Protein

The carboxy-terminal28 amino acids of the Saccharomyces cermisiae transcriptional activator protein GAL4 execute two functions-transcriptional activation and interaction with the negative regulatory protein, GALSO. Here we demonstrate that these two functions are separable by single amino acid changes within this region. We determined the sequences of four GAL4"-mutations, and characterized the abilities of the encoded GAL4" proteins to activate transcription of the galactose/melibiose regulon in the presence of GAL80 and superrepressible GAL80" alleles. One of the GAL4" mutations can be compensated by a specific GAL80" mutation, resulting in a wild-type phenotype. These results support the idea that while the GAL4 activation function tolerates at least minor alterations in the GAL4 carboxyl terminus, the GALSO-interactive function is highly sequence-specific and sensitive even to single amino acid alterations. They also argue that the GAL80" mutations affect the affinity of GAL80 for GAL4, and not the ability of GAL80 to bind inducer. T HE GAL4 gene of Saccharomyces cerevisiae encodes a protein that activates transcription of galactose and melibiose catabolic enzyme genes (DOUGLAS and HAWTHORNE 1966; JOHNSTON and HOPPER 1982; LAUGHON and GESTELAND 1982; HASHIMOTO et al. 1983). Activation is achieved primarily through two functions of the GAL4 protein; an amino-terminal DNA binding function which positions GAL4 upstream of regulated structural genes (GUARENTE, YOCUM and GIFFORD 1982; JOHNSTON and DAVIS 1984; WEST, YOCUM and PTASHNE 1984; YOCUM et al. 1984; BRAM and KORNBERG 1985; GINIGER, VARNUM and PTASHNE 1985; BRAM, LUE and KORNBERG 1986; KEEGAN, GILL and PTASHNE 1986), and a carboxy-terminal "activation" function which interacts with cellular transcription factors (BRENT and PTASHNE 1985; JOHNSTON et al. 1986; JOHNSTON, SALMERON and DINCHER 1987; MA and PTASHNE 1987a). Expression of the wild-type galactose/melibiose regulon is not constitutive, due to the activity of the negative regulatory gene GAL80 (DOUGLAS and HAWTHORNE 1966; NOGI et al. 1984; YOCUM and JOHNSTON 1984). In the absence of the regulon inducer (galactose or a metabolite thereof), the GAL4 protein binds target DNA sequences but does not activate transcription, due to the binding of the GAL80-encoded protein to GAL4 (JOHNSTON and HOPPER 1982; GINIGER, VARNUM and PTASHNE 1985; SELLECK and MAJORS 1987; LUE et al. 1987; MA and PTASHNE (1988). Inducer frees GAL4 of GAL80mediated negative regulation (DOUGLAS and HAWGenetics 125: 21-27 (May, 1990) THORNE 1966). Mutations in the GAL80 gene, termed GAL80S, result in proteins that block activation by GAL4 under both noninducing and inducing conditions (DOUGLAS and HAWTHORNE 1972; NOGI and FUKASAWA 1984). GAL80S proteins are hypothesized either to have lost inducer binding function (DOUGLAS and HAWTHORNE 1972; NOGI and FUKASAWA 1984), or to possess a high affinity for GAL4 (SALMERON, LANGDON and JOHNSTON, 1989). The protein encoded by the gu14-62 nonsense allele is defective because it lacks the carboxy-terminal 28 amino acids of wild-type GAL4, which contain the activation function (JOHNSTON, SALMERON and DINCHER 1987; MA and PTASHNE 1987a; Figure 1) . We previously characterized a Gal+ revertant of gal462, GAL4'-62. GAL4' alleles encode proteins that escape regulation by GAL80 (DOUGLAS and HAWTHORNE 1966). The GAL4"-62 mutation causes a frameshift that replaces the last four amino acids of the ga14-62 protein with three new amino acids, but does not restore the wild-type GAL4 carboxyl terminus (JOHNSTON, SALMERON and DINCHER 1987; also see Figure 1). Its constitutive phenotype indicates that the carboxy-terminal28 amino acids of GAL4 contain a GAL80-interactive function as well as an activation function. Similar results are obtained by MA and PTASHNE (1987a, b), who analyzed the effects of in vitro generated deletions of the GAL4 gene on GAL4 protein activity. Together these findings prompted a model whereby the transcriptional activity or inactivity of the GAL/MEL regulon is determined by the binding 22 J. M. Salmeron, K. K. Leuther and S. A. Johnston of either GAL80 or transcription factors to the carboxyl terminus of GAL4 (JOHNSTON, SALMERON and DINCHER 1987; MA and PTASHNE 1987b). Although these results provide information critical to the understanding of GAL4-GAL80 interaction, many questions remain. Three of the most important are: (1) are the transcriptional ctivation and GAL80interactive functions of the GAL4 carboxyl terminus separable?, (2) what are the specific amino acid interactions that occur between GAL4 and GAL80? and (3) what is the nature of GAL80S mutations? To address these questions, we characterized and sequenced four additional GAL4" mutations. As concerns the first question it appears there are minimal primary sequence requirements for the GAL4 transcriptional activation. JOHNSTON and DOVER (1 988) characterized 41 gal4 missense mutations and found none that affected the GAL4 carboxyl terminus, indicating that most single amino acid changes do not eliminate the GAL4 activation function. GINICER and PTASHNE (1 987) described a synthetic 15-amino-acid peptide that bore little similarity to the GAL4 carboxyl terminus yet activated transcription when fused to the GAL4 DNA binding region. Most strikingly, MA and PTASHNE ( 1 9 8 7 ~ ) identified multiple peptides, encoded by random Escherichia coli DNA fragments, that activate transcription in yeast when fused to the GAL4 DNA binding region. Thus, the GAL4 activation region appears to have minimal primary sequence requirements, at least when the gene is highly over-