The Cysteine-Peptidase Bleomycin Hydrolase Is A Member of the Galactose Regulon in Yeast*

Bleomycin hydrolase is a cysteine peptidase discovered through its ability to detoxify the anti-cancer glycopeptide, bleomycin. Although found in all tissues in mammals and in both eukaryotes and prokaryotes, the normal cellular function of this peptidase is not known. We had previously reported the purification of bleomycin hydrolase from yeast based on its unexpected ability to bind DNA. Recently we collaborated in solving the crystal structure of this protein, revealing a hexameric ring organization. We now report that the molecular characterization of the gene encoding yeast bleomycin hydrolase is also surprising. The transcription of the gene is regulated by galactose. Furthermore, this regulation is conveyed by a binding site for the Gal4 regulatory protein in its promoter, prompting the designation of this gene as GAL6. Gal6p also appears to have a negative effect on the GALsystem as a deletion of the gene leads to a 2–5-fold higher expression of the GAL1, GAL2, GAL7, and MEL1 genes. TheGAL6 deletion does not affect the expression of another inducible gene, HSP26. Neither the peptidase nor the nucleic acid binding activity of Gal6p as assayed is apparently required to convey this regulation, implying yet another function for this new member of the GAL regulon.

Bleomycin hydrolase is a cysteine peptidase discovered through its ability to detoxify the anti-cancer glycopeptide, bleomycin. Although found in all tissues in mammals and in both eukaryotes and prokaryotes, the normal cellular function of this peptidase is not known. We had previously reported the purification of bleomycin hydrolase from yeast based on its unexpected ability to bind DNA. Recently we collaborated in solving the crystal structure of this protein, revealing a hexameric ring organization. We now report that the molecular characterization of the gene encoding yeast bleomycin hydrolase is also surprising. The transcription of the gene is regulated by galactose. Furthermore, this regulation is conveyed by a binding site for the Gal4 regulatory protein in its promoter, prompting the designation of this gene as GAL6. Gal6p also appears to have a negative effect on the GAL system as a deletion of the gene leads to a 2-5-fold higher expression of the GAL1, GAL2, GAL7, and MEL1 genes. The GAL6 deletion does not affect the expression of another inducible gene, HSP26. Neither the peptidase nor the nucleic acid binding activity of Gal6p as assayed is apparently required to convey this regulation, implying yet another function for this new member of the GAL regulon.
The bleomycins are a family of glycopeptides produced by Streptomyces verticillus that can bind to and cleave DNA (1,2). This property has been exploited widely for cancer chemotherapy (3). Most cell types express bleomycin hydrolase, an enzyme which renders bleomycin unable to cleave DNA and therefore nontoxic, whereas some tumors overexpress the enzyme (4). The level of expression of this hydrolase has been suggested to determine a cell's susceptibility to bleomycin. A current difficulty in bleomycin chemotherapy is that certain tumor types can become tolerant to the drug, whereas normal tissues, particularly lung and skin, remain susceptible (5,6).
The yeast bleomycin hydrolase gene was cloned by different groups as BLH1 or YCP1 by pursuing the peptidase activity of the protein (7,8) or as a cytosolic calcium-dependent phospholipid-binding protein (9). We first purified the protein unintentionally while characterizing the DNA binding activity of Gal4p (10). Gal4p is the positive regulator of the GAL regulon of yeast which consists of both structural and regulatory genes. Five structural genes are involved in galactose metabolism as follows: GAL2 encodes a permease for galactose uptake; GAL1, GAL7, and GAL10 encode the Leloir pathway enzymes that convert galactose to glucose-6-phosphate; and MEL1 encodes an ␣-galactosidase which converts melibiose to galactose and glucose. All of these genes are highly regulated by Gal4p through binding to the upstream activating sequence (UAS G ) 1 in the promoter of these genes. The activity of Gal4p is also controlled by Gal80p, a negative regulator, and Gal3p, which is responsible for the signal transduction by galactose (11)(12)(13).
Here we report that the level of yeast bleomycin hydrolase is low in yeast grown on glucose medium and induced severalfold on galactose medium. A Gal4p-responsive UAS G was identified in the promoter region of the gene. It was shown to be responsible for this galactose regulation, prompting its designation as GAL6. Surprisingly, the deletion of the gene increases the expression level of galactose regulated genes, indicating that yeast bleomycin hydrolase, GAL6, is a negative regulator of the galactose regulon. This is the first evidence that a bleomycin hydrolase protein has a cellular function beyond detoxification of bleomycin.
Protein Purification and Mass Spectrometry-Yeast bleomycin hydrolase was purified from yeast cells grown in galactose as described previously (10). The molecular weight of purified protein was measured by mass spectrometry. Electrospray mass spectra were acquired with a VG QUATTRO II triple quadrupole Mass Spectrometer (Micromass Instruments, Manchester, UK) equipped with an electrospray interface at Howard Hughes Medical Institute Biopolymer Core Facility at the University of Texas-Southwestern.
The primers 28cl (5Ј-CAACCAGAACACGCGAGGCAGTCTG-3Ј) from 780 bp upstream of the GAL6 ORF and Nsi (5Ј-AGCCAGGAAA-AGAACAGGCTTATGACAGAA-3Ј) from 300 bp downstream of the GAL6 stop codon were used to amplify a PCR fragment that contains the GAL6 ORF and intact promoter from pXU2. This PCR fragment was cloned into the multiple cloning sites of YEP352 and designated pXU-606. The plasmid pXU616 is the same construct as pXU606 except that the DNA binding site of Gal4p (UAS G6 ) in the GAL6 promoter was deleted by the following method. Oligonucleotide B3 (5Ј-GGCGGATC-CATTGAT TACCACAT-3Ј) which is upstream of UAS G6 was used in combination with 28c1 to amplify PCR the fragment upstream of UAS G6 from pXU2. Oligonucleotide B1 (5Ј-AGTCGCCGACGGATCCCATAAA-TAAACG-3Ј) which is just downstream of UAS G6 was used with Nsi to amplify a PCR fragment that contains the GAL6 ORF from pXU2. Both B1 and B3 have a BamHI site. These two PCR fragments were then joined at the BamHI site and introduced into YEP352 to form pXU616, which deletes a 30-bp region containing the UAS G6 . To assay the effect of UAS G6 on GAL6 expression, two constructs were made. The EcoRI fragments from pXU606 and pXU616 which contain the promoter and the first 95 amino acid coding region of GAL6 were introduced into the EcoRI site in YIP356R. The GAL6 ORF is in-frame with the lacZ reporter gene in YIP356R. The resulting reporters were then integrated into the URA3 locus in the yeast strain Sc317 to study the regulation of GAL6 by galactose.
The SacI/BamHI fragment of the GAL6 gene from pXU606 was cloned into pUC118. A 1.8-kb BamHI fragment that contains the yeast HIS3 gene was also cloned into the BamHI site. The construct is designated pWZ1-3. The XhoI site in the HIS3 3Ј-flanking sequence is a unique site in pWZ1-3 and was used to linearize the construct and integrate it into the HIS3 locus of yeast genome.
The GAL6 knock-out construct was made by replacing the pXU616 BamHI-EcoRV fragment, which contains the coding region for the first 420 amino acids of the Gal6p N terminus, with the 0.9-kb StuI-EcoRI fragment that contains TRP1 gene from YRP6. The primers 28cl and Nsi were used to amplify by PCR a fragment with ends identical to the GAL6 locus using the null construct as the template. The PCR fragment was then transferred into the yeast genome by one-step gene replacement. All the GAL6 null strains were verified by Southern blot.
Mutagenesis of GAL6 -The GAL6 gene in pWZ1-3 was mutated using a Sculptor TM in vitro mutagenesis system from Amersham Corp. The oligonucleotide 5Ј-AAACAACCAAGATCTACCAGA-3Ј was used to change Cys 73 to alanine. To change Lys 242 , Lys 244 , and Lys 245 to alanines, the oligonucleotide 5Ј-AGTGTGGATTGCCGCGTCTGCGTC-TACGTATTCCC-3Ј was used. These amino acid designations are based on the ATG start determined in this work, which is different from previously published data (see below). All the mutants were confirmed by sequencing. The resulting mutants in this construct were linearized by XhoI in the 3Ј-flanking region of the His3 gene and integrated into the HIS3 locus of Sc454.
The protein/DNA binding reaction mixtures were incubated for 20 min at room temperature in a final volume of 20 l, containing either purified protein or crude yeast cell extract, 1 g of sonicated salmon sperm DNA, and 1 ng of oligonucleotide end-labeled with [␥-32 P]ATP by T4 kinase. The protein-DNA complexes were resolved in 4% polyacrylamide gels in 0.5 ϫ TBE. The gels were dried and autoradiographed on either Kodak x-ray film or a Molecular Dynamics PhosphorImager screen. The quantitation of bands was done on the PhosphorImager image, and the figures are from these files.
To assay the DNA binding activity of Gal6p, a single-stranded DNA oligonucleotide UAS G1 (AGCTTAGCGGAAATTTGTGGTCCGAGC) that contains one strand of a Gal4p consensus binding site was used based on our previous observation that Gal6p has very high affinity for single-stranded DNA (10). Two double-stranded oligonucleotides, UAS G6 which has the Gal4p binding site from the GAL6 gene promoter region and UAS G1 , were used in gel mobility shift assays with the Gal4p (1-147 amino acids) DNA binding domain purified from E. coli.
Enzyme Assays-␤-Galactosidase activity was determined as described elsewhere (18). To assay the ␤-galactosidase enzyme level in the cell, yeast cells were grown to mid-log phase in the designated medium. Cells were harvested, and extracts were made in extract buffer (20 mM HEPES-KOH, pH 7.5, 1 mM dithiothreitol). Twenty to 80 g of protein were used for the enzyme assay.
For induction assays, cells were grown in selective media supplemented with 2% glucose to A 600 nm ϭ 0.85-1.0 and then washed once in H 2 O. The washed cells were transferred into selective medium plus 2% galactose. At different times the cells were sampled and stored at Ϫ80°C. The sample cells were assayed for ␤-galactosidase activity as described (18). To determine the effect of gal6 mutations on galactose induction, the cells that harbored different gal6 mutants were subject to the same induction protocol, except that cells were harvested after growth in galactose medium for 30 -36 h. All assays were performed with three independent transformants.
RNase Protection Assay-Yeast whole cell RNA was prepared by the hot phenol method (19). To measure GAL1 messenger RNA in GAL6 wild type and deletion strains, a probe for GAL1 (520 base transcript from GAL1 ORF) and ACT1 as a control (420 base transcript from its ORF) was used. To determine the half-life of GAL1 messenger RNA, GAL1 and ACT1 probes were transcribed from templates of PCR fragments from yeast genomic DNA. For the 150-bp GAL1 fragment, 5Ј-GGGGGGGGGGCCTGTTTCTTATTGGCGAGAGACTC-3Ј was used as the forward primer. 5Ј-GTCTGTTTGCGGTGAGGAAGATC-3Ј was used as the reverse primer. For the ACT1 probe, the primers 5Ј-GGGGG-GGGGGAAACGTAGAAGGCTGG-3Ј and 5Ј-TCCTACGTTGGTGAT-GAAGC-3Ј were used to amplify a 250-bp fragment from yeast ACT1 gene. The T7 promoter (5Ј-GGATCCTAATACGACTCACTATAGGGA-GGGGGGGGGG-3Ј) was added by PCR. The templates then were transcribed to produce an RNA probe using a MAXIscript TM kit from Ambion. The RNase protection assay was accomplished by using an RPA II TM kit (Ambion). After RNase treatment, the RNA samples were separated on a 5% polyacrylamide, 8 M urea gel and dried on Whatman paper. The dried gels were exposed to a PhosphorImager screen and quantitated by ImageQuant software from Molecular Dynamics. Due to partial denaturation of the double-stranded RNA fragment, in some RNA protection assays the signals appeared as doublets. It does not affect our interpretation of the results since the control that has only the RNA probe does not have detectable signal.
Northern Blot-Total yeast RNA was prepared by the hot phenol method, and blots were carried out by standard protocols (18). The amount of hybridization was quantitated with a Molecular Dynamics computing densitometer using ImageQuant software. The relative amounts of transcript from the GAL2, GAL7, and MEL1 genes were normalized to the messenger RNA level of the chromosomal cyclophilin gene (CYP1) as an internal control. The probes used in Northern blots were as follows: for GAL2, the 1.7-kb ORF; for GAL7, the 0.5-kb fragment of a portion of GAL7 ORF; for MEL1, a 1-kb PCR fragment from the ORF; for CYP1, the whole ORF cloned into pUC119. All probes used for Northern blots were labeled with 32 P by the random primer method (18).
Western Blot-Yeast cell crude extract was made as described under "Enzyme Assays." Five to 10 g of protein were run on 10% Tricine SDS-polyacrylamide gels and transferred to PVDF TM membrane using a liquid-blotting apparatus (Idea Scientific). Gal6p was detected by a standard Western blot method using a rabbit polyclonal antibody generated against Gal6 protein purified from yeast (10).

RESULTS
The Structure of the Gene Encoding GAL6 -We previously reported the purification of a 48-kDa protein from yeast based on its DNA binding activity (10). We first obtained amino acid sequence from four tryptic digest fragments of this protein.
Based on these sequences the corresponding gene was isolated and the predicted 48-kDa protein identified as the yeast homolog of the rabbit bleomycin hydrolase. Because we had shown that the levels of this protein were regulated by galactose, it was designated as GAL6 (see below).
The GAL6 ORF would encode a protein of 51.9 kDa if the ϩ1 ATG designated in Fig. 1A is used. There is another upstream ATG that would encode a protein of 55 kDa if it were the authentic start site. This upstream ATG was designated as the start in previous publications (7,8). However, the ATG at the downstream site matches the consensus translation initiation sequence better (9/10) than the upstream site ( Fig. 1A) (9,20). To determine which ATG is used as the translation start site, we purified Gal6p from a wild type yeast strain and measured its molecular weight by mass spectrometry. As shown in Fig.  1B, the majority of purified protein has the molecular mass of 51.8 kDa, which is consistent with the use of downstream ATG as the translational start codon of GAL6. Our proposed translation of Gal6p would put its N terminus in alignment with that of the recently cloned human homolog (4,21). The N terminus of Gal6p was not discernible in the crystal structure (22). Enenkel and Wolf (7), Magdolen et al. (8), and Pei and Sebti (23) assumed that the upstream ATG was the initiation site. We have now entered the corrected sequence into Gen-Bank with accession number U74299.
GAL6 Is Regulated by Galactose-Transcription of the galactose genes in yeast is regulated by the transcription activator Gal4p. All the known galactose-regulated genes have a well characterized 17-bp Gal4p binding site (UAS G ) in their promoters. In glucose medium, the expression of GAL4 is suppressed, resulting in most, but not all, GAL genes being completely off. In glycerol medium, Gal4p levels are elevated, facilitating binding to the UAS G . However, the negative regulator GAL80 sup-presses GAL4 activity and blocks the expression of the GAL regulon (24). When cells are grown in galactose, Gal4p binds to the UAS G of the galactose-regulated genes and activates transcription (13). By several criteria we find that GAL6 is also a member of the galactose regulon.
Gal6p was originally discovered in our lab by its ability to bind to DNA. It is the most significant activity in yeast cell extract that binds single-stranded DNA oligonucleotides in a gel mobility shift assay. We therefore used a gel shift as a sensitive assay to quantitate the amount of active Gal6p in cells expressing various amounts of Gal4p. As shown in Fig. 2A, the amount of Gal6p in the cell is proportional to the Gal4p levels. Deleting GAL4 (lanes 1 and 4) decreases the level of Gal6p, whereas increasing Gal4p levels with a multicopy plasmid increases Gal6p levels (lanes 3 and 6). As expected for a FIG. 1. A, structure of the GAL6 gene promoter. There is a predicted UAS G (Ϫ150 to Ϫ134, small box) in the promoter region, and there are two potential ATG start codons in-frame. The sequence in the large box was deleted to study Gal4p-dependent transcription regulation. B, the molecular weight (determined by electrospray mass spectrometry) of Gal6p purified from yeast indicates that native Gal6p is translated from the second translation start codon. The molecular weights of Gal6p predicted from the sequence that starts from the first ATG (55 kDa) and second ATG (51.9 kDa) are indicated by arrows with 1st ATG and 2nd ATG. 2. GAL6 is regulated by galactose. A, a gel mobility shift assay shows that the Gal6 protein levels in yeast cells correspond to changes in Gal4p levels (⌬gal4, GAL4 deletion strain; Sc, yeast strain that has a single copy of GAL4 gene; Mc, strain with multicopy GAL4 gene). A single-stranded oligonucleotide of UAS G1 (1 ng) was used along with 30 g of yeast protein for the gel shift assay for each sample. Extract was from yeast grown on raffinose or galactose. In lane 7, the Gal6p/DNA was supershifted by 12 g of rabbit IgG against Gal6p (ϩAB lane). Below the gel shift is an immunoblot for Gal6p of the same protein preparations used in the gel shift. Using a nonspecific crossreacting band as a standard, the relative protein levels in lanes 4 -6 of the immunoblot are 0.5:1:4. B, the amount of GAL6 mRNA is regulated by galactose as measured by a Northern blot. The 32 P-labeled probes used were generated by random primer labeling using either GAL6 or ACT1 (actin) ORF DNA as template. For each sample, 5 g of yeast total RNA was used. C, a gel mobility shift assay shows that Gal4 protein can bind to the UAS G1 from the GAL1 promoter and to the UAS G6 from the GAL6 promoter. The purified Gal4p (N-terminal 1-147 amino acids, 20 ng) was mixed with 1 ng of radiolabeled, doublestranded UAS G1 or UAS G6 oligonucleotide for 10 min. The binding mixtures were then separated on a 4% native polyacrylamide gel. D, the UAS G6 in the GAL6 promoter is required for the galactose regulation of the gene. The galactose regulation of GAL6 expression can be abolished by disrupting the UAS G6 in the GAL6 promoter. Yeast cells were grown to A 600 nm ϭ 0.8 in the media indicated and then assayed for ␤-galactosidase reporter activity.
Gal4p/galactose-regulated gene, when cells are grown in raffinose medium, Gal6p is expressed only at a low level. However, when cells are grown in galactose, the expression level of Gal6p increases approximately 5-fold over the glucose or glycerol levels ( Fig. 2A, compare lanes 2 and 5). An immunoblot for Gal6p levels reveals essentially the same relative effects of galactose induction and Gal4p response as the gel shift ( Fig. 2A, bottom). These differences in Gal6 protein levels are paralleled by the regulation of GAL6 mRNA levels, as shown by a Northern blot (Fig. 2B). This type of regulation is qualitatively the criteria for a GAL-regulated gene and quantitatively very similar to that of GAL80 and GAL3 genes (see "Discussion").
The regulation of GAL6 mRNA levels implies a Gal4p-mediated control. Consistent with this deduction, we found a single Gal4p binding site (UAS G6 ) in the promoter region of the GAL6 gene (Fig. 1A). As shown in Fig. 2C, Gal4p can bind to the UAS G6 in vitro. To test whether the UAS G6 is responsible for the galactose regulation of GAL6 expression, we fused the GAL6 promoter, with or without the UAS G6 , to a lacZ reporter gene. As shown in Fig. 2D, the ␤-galactosidase activity under the control of the wild type GAL6 promoter is regulated by galactose. A deletion of a 30-bp fragment containing the UAS G6 from the promoter eliminates this galactose regulation, indicating that the UAS G6 is responsible for the galactose regulation of GAL6 gene. The difference in reporter levels under repressing conditions (glucose) (60 versus 12 units) could be due to the 30-bp deletion disrupting binding of some other protein or that the Gal4-Gal80 complex contributes to the activation of GAL6 on glucose. Regardless, we conclude that GAL6 expression depends on Gal4p activity and that GAL6 is a newly identified member of the GAL regulon.
GAL6 Is a Negative Regulator of GAL Regulon Expression-Since GAL6 is regulated by galactose, we addressed the question of what role it might play in galactose regulation by deleting it. The deletion strain is viable with no notable effect on growth rate of glucose-grown cells (data not shown), which is consistent with previous observations (7)(8)(9). However, when we compared the induction of a galactose-regulated gene in the wild type and gal6 deletion strain, we found that GAL6 has a negative effect. In the gal6 deletion strain, the induction level of a ␤-galactosidase reporter under control of the GAL1/GAL10 promoter is higher than that in the wild type cells (Fig. 3A). The cells in this assay were transferred from repressing glucose medium to galactose medium at late mig-log phase, so they only grew a few generations before reaching stationary phase (final A 600 nm ϭ 4), accounting for the relatively low ␤-galactosidase activity. This assay was also performed for the enzyme output of the MEL1 gene. The ␣-galactosidase activity was ϳ2.2-fold higher in the deletion strain (data not shown). This protocol emphasizes the effects on the kinetics of derepression from glucose in the GAL6 and ⌬gal6 strains. When the same GAL6 and ⌬gal6 strains were grown continuously in galactose medium there was also more ␤-galactosidase expression in the deletion stain (Fig. 3B). This difference in ␤-galactosidase reporter activity is mirrored in an RNA protection assay to detect GAL1 mRNA (Fig. 3C) and Northern blots to detect GAL2, FIG. 3. GAL6 has a negative effect on the galactose gene induction. A, a gal6 deletion strain has a higher induced ␤-galactosidase level than the GAL6 wild type strain. Yeast cells (Sc454 with or without GAL6) were grown in 200 ml of glucose media to A 600 nm ϭ 1.0 and rinsed once in water. Cells were transferred into the same volume of galactose medium to continue growth at 30°C. Every 4 h, 10 ml of cells were harvested and assayed for ␤-galactosidase activity. B, a ␤-galactosidase assay shows that GAL6 has a negative effect on galactose gene expression. Yeast strains were grown continuously in glycerol/lactic acid/galactose medium to A 600 nm ϭ 0.4. Cells were harvested and subjected to ␤-galactosidase assay (shaded column) and protease assay (solid column). C, an RNase protection assay shows that the gal6 deletion strain has higher levels of GAL1 mRNA than in the GAL6 wild type strain. Yeast RNA (10 g) was isolated from the wild type or ⌬gal6 strain grown in galactose to A 600 nm ϭ 0.4 and subjected to an RNase protection assay with single-stranded RNA probes. The sizes of the probes are as follows: GAL1, 520 bp; actin (ACT), 420 bp. Lane 1, RNA from GAL6 wild type strain. D, Northern blots show that GAL2, GAL7, and MEL1 mRNA levels are negatively regulated by GAL6. RNA from GAL6 or ⌬gal6 strains were used for the assay. In each lane 10 g of total cellular RNA was used. The cyclophilin-normalized ratio of the different mRNAs from GAL6 or ⌬gal6 strains are as follows: GAL2, 1:2.6; GAL7, 1:2.4; MEL1, 1:1.7. E, an RNA protection assay shows that deleting GAL6 has no effect on the induction of the Hsp26 gene. Lane 1, actin probe. Lane 2, Hsp26 probe. Lanes 3-6 and lanes 7-10, RNA from GAL6 and ⌬gal6 strains after being heat shocked for 0, 5, 10, or 20 min, respectively. Lane 11, probe alone.
GAL7, and MEL1 mRNA levels (Fig. 3D). The deletion strain has approximately 2-5-fold more of these mRNAs. On a practical note for two hybrid and related assays, strains deleted for GAL6 show the blue color of a ␤-galactosidase reporter gene in plate assays 1-2 days earlier than the wild type strain (data not shown).
The GAL1, GAL2, GAL7, and MEL1 genes are highly inducible. We were curious as to whether GAL6 might affect other highly inducible genes. As a representative of such genes, we assessed the induction of the yeast heat shock gene, HSP26, in GAL6 wild type and gal6 deletion backgrounds. As shown in Fig. 3E, GAL6 has no apparent effect on the induction of HSP26, indicating that the effect of GAL6 deletion is not general for all highly inducible genes.
As shown above, the negative effect of GAL6 on the GAL gene expression is at the mRNA level. This effect could be through alterations in mRNA production or stability. To distinguish between these two possibilities, we tested the stability of the GAL1 mRNA in a GAL6 wild type and a gal6 deletion strain. Yeast cells were grown in galactose medium to mid-log phase. Glucose was then added to stop the transcription of the galactose-regulated genes. Whole cell RNA was prepared at different times after glucose was added. The RNA samples were subjected to an RNA protection assay to measure the amount of GAL1 mRNA. As shown in Fig. 4, the starting GAL1 mRNA level in a ⌬gal6 strain is higher than that of a wild type strain as expected from the results reported above, and the rate of decrease of GAL1 mRNA is essentially the same after adding glucose. This indicates that deleting GAL6 has no obvious effect on the stability of GAL1 mRNA at least in this experimental regime. By elimination, it appears that the effect of the GAL6 deletion on GAL1 mRNA levels is on the rate of production rather than the turnover rate of mRNA.
Either of the two known activities of Gal6p, peptidase or nucleic acid-binding, could be responsible for the repression phenotype. To investigate the basis of GAL6-mediated repression, we first created variant forms of the protein by site-directed mutagenesis which had lost either DNA binding or peptidase activity. The peptidase mutant, gal6C73A, was made by replacing the active site Cys 73 with alanine. This variant does not have measurable peptidase activity, as assayed against the substrate, N-Arg-7-amido-4-methylcoumarin (Fig. 5B). Disruption of the DNA binding activity of Gal6 protein was accomplished by changing Lys 242, Lys 244 , and Lys 245 to alanines. This alteration was guided by the crystal structure that showed these lysines to be near the surface of the hole in the hexamer (22). As shown in Fig. 5, A and B, these two mutations are independent, that is the DNA-binding variant (gal6db) retains normal peptidase activity and vice versa. When the wild type GAL6 was replaced with the protease-deficient form, the induction of ␤-galactosidase activity was unaltered. This indicates that the peptidase activity of Gal6p is not required in the negative regulation of GAL gene expression (Fig.  5C). We also tested whether the nucleic acid binding activity has any effect on GAL gene induction. The result shows that like the peptidase mutant, the strain bearing the Gal6p defective in nucleic acid binding has the same phenotype as the wild type (Fig.  5D). This indicates that the DNA binding activity of Gal6p may not be required for the negative regulation of the galactose genes. The double mutant has the same phenotype as the single mutants (data not shown). It appears that the negative regulatory activity of Gal6p does not require either the peptidase or nucleic acid binding functions, at least as measured by conventional in vitro assays.
Interestingly, when GAL6 was overexpressed by use of multicopy plasmids or an ADH1 promoter fusion there was no detectable effect on galactose growth or GAL1 promoter-LacZ reporter FIG. 5. Neither the protease activity nor the DNA binding activity of Gal6p is required for its negative regulation of the galactose genes. A, top: a gel mobility shift assay shows that changing Cys 73 to alanine (C73A) has no effect on DNA binding activity, but changing Lys 242, Lys 244 , and Lys 245 to alanines (gal6db) disrupts the DNA binding activity of Gal6p. Radiolabeled single-stranded oligonucleotide UAS G1 (1 ng) was used with 20 g of yeast protein for each assay. Bottom, a Western blot shows that neither mutation has an effect on the level of the protein in the cell. B, a protease assay shows that C73A has no protease activity, but the DNA binding mutant protein (gal6db) maintains its protease activity toward a synthetic substrate Arg-7-amido-4-methylcoumarin. C, the protease activity of Gal6p is not required for the negative regulation of the GAL regulon. The gal6deleted (⌬gal6), the wild type (GAL6), and the protease-deficient strain (C73A) were grown in glucose medium to A 600 nm ϭ 1.0 and then transferred to galactose medium and grown for 30 h. Then cells were harvested and subjected to a ␤-galactosidase assay. D, the DNA binding activity of Gal6p is not required for the negative regulation of the GAL regulon. Cells were assayed the same way as in C except they were grown for 36 h before being harvested. This may account for the higher activities than in C. The DNA-binding defective Gal6p is designated gal6db.
expression (data not shown). This implies that wild type levels of Gal6p are sufficient to convey full negative regulation. This may not be too surprising considering that Gal6 protein is an abundant protein. We estimate from quantitative Western blots using purified protein as a control that there are ϳ18,000 Gal6p molecules per cell under glucose growth conditions and 68,000 molecules under inducing (galactose) conditions. DISCUSSION The yeast form of bleomycin hydrolase has been discovered in several different, apparently unrelated purifications and selections. We now find that it is a galactose-regulated gene. Furthermore, deletion of this gene, GAL6, leads to higher expression of GAL-regulated genes. This implies that GAL6 is a newly defined negative regulator of the GAL system and part of an autoregulatory circuit.
There is no obvious link between bleomycin hydrolase activity and galactose metabolism. However, GAL6 is not the first example of a non-galactose metabolism gene under GAL4 control. The GCY1 gene is also regulated by GAL4. It is thought to encode a carbonyl reductase by sequence similarity to animal genes (25). Unlike GAL6, however, deletion of GCY1 has no effect on the cells' ability to grow on galactose or on GAL gene regulation. We propose that GAL6 and GCY1 may be part of the "environmental" galactose regulon. That is their gene products are not directly involved in the metabolism of galactose, but the presence of galactose generally correlates with some environmental condition that calls for GAL6 or GCY1. This condition may only be evident with yeast in their natural environment.
We report here that GAL6 functions as a negative regulator of the galactose system. Cells transferred from glucose to galactose medium show a faster induction of GAL1 and a higher level at stationary phase. This implies that deletion of GAL6 allows faster relief of glucose repression and higher total output of GAL1. This up-regulation is also evident for GAL1 when the strains are grown continuously in galactose (Fig. 3C) and for the other structural members of the GAL regulon, GAL2, GAL7, and MEL1 (Fig. 3D). A trivial explanation for this increased expression in the ⌬gal6 background is that the deletion of the GAL6 gene included the UAS G in the promoter leading to higher Gal4p occupancy of other GAL genes. However, a strain in which only the GAL6 open reading frame was deleted had the same phenotype (data not shown).
Besides GAL4, two other regulatory proteins of the system are known, Gal3p and Gal80p. GAL3 is defined as a positive element as its deletion results in a long delay in induction of the GAL genes (26 -28). GAL80, as with GAL6, is a negative regulator as its deletion leads to high level, constitutive expression of GAL4regulated genes (29 -32). Also like GAL6, GAL80 and GAL3 are themselves regulated by GAL4, each containing one Gal4p binding site in their promoters (33,34). All these genes have a significant GAL4-independent expression under non-inducing conditions (Fig. 2D) (33,34). It is interesting that there is measurable Gal6p expression on glucose medium even when GAL4 is deleted ( Fig. 2A), implying that there may be other regulator signals for GAL6 expression. Unlike the structural genes that are induced 50 -1000-fold, GAL3 and GAL80 genes, like GAL6, are only induced ϳ5-10-fold by galactose (13). Taking these comparisons together, GAL6 regulation is strikingly similar to that of the other negative regulator, GAL80.
Gal80p mediates negative regulation through binding the activation domain of Gal4p (24). How GAL6 conveys negative regulation is unknown. Since the GAL regulon is sensitive to Gal4p levels, one possibility is that GAL6 affects the steady state level of Gal4p through its protease activity. This seems unlikely for the following three reasons. 1) We do not observe any changes in Gal4p level in the gal6-deleted strain (data not shown); 2) Gal6p is a peptidase and there is no evidence it is a protease; and 3) most convincingly, the Cys 3 Ala mutation of the active site does not affect the GAL phenotype. Another possible explanation is that GAL6 enhances the activity of Gal80p, but this could not be through the protease activity of Gal6p.
The GAL phenotype could be mediated through the nucleic acid binding activity of Gal6p. However, the mutant defective in this activity behaves the same as the wild type strain in GAL induction, indicating that this activity may not be involved in the GAL phenotype. One possibility we cannot rule out is that the loss of nucleic acid binding activity as measured by the in vitro assay may not reflect the in vivo activity of the protein. However, it is also possible that there is an as yet unidentified protein that interacts with Gal6p which is involved in the GAL regulation or that Gal6p itself has another undiscovered activity.
We conclude that GAL6 is a newly recognized member of the GAL system, apparently acting as a negative regulatory protein. How it conveys its negative regulation is not evident from mutations in its known activities, implying a yet to be discovered function of this protein. This molecular analysis of GAL6 provides the first evidence that this highly conserved and ubiquitous protein has a cellular function other than hydrolyzing bleomycin.