Activation Mechanism of the Multifunctional Transcription ... - NCBI

Download PDF
3 downloads 0 Views 796KB Size Report
Comment
AND HENRY V. BAKER*. Department of Molecular Genetics and Microbiology, College of Medicine,. University of Florida, Gainesville, Florida 32610-0266.
MOLECULAR AND CELLULAR BIOLOGY, June 1996, p. 3187–3196 0270-7306/96/$04.0010 Copyright q 1996, American Society for Microbiology

Vol. 16, No. 6

Activation Mechanism of the Multifunctional Transcription Factor Repressor-Activator Protein 1 (Rap1p) ´ PEZ, CAROLYN M. DRAZINIC, JEFFREY B. SMERAGE, M. CECILIA LO AND HENRY V. BAKER* Department of Molecular Genetics and Microbiology, College of Medicine, University of Florida, Gainesville, Florida 32610-0266 Received 11 December 1995/Returned for modification 25 January 1996/Accepted 8 March 1996

Transcriptional activation in eukaryotic organisms normally requires combinatorial interactions of multiple transcription factors. In most cases, the precise role played by each transcription factor is not known. The upstream activating sequence (UAS) elements of glycolytic enzyme genes in Saccharomyces cerevisiae are excellent model systems for the study of combinatorial interactions. The yeast protein known as Rap1p acts as both a transcriptional repressor and an activator, depending on sequence context. Rap1p-binding sites are found adjacent to Gcr1p-binding sites in the UAS elements of glycolytic enzyme genes. These UAS elements constitute some of the strongest activating sequences known in S. cerevisiae. In this study, we have investigated the relationship between Rap1p- and Gcr1p-binding sites and the proteins that bind them. In vivo DNA-binding studies with rap1ts mutant strains demonstrated that the inability of Rap1p to bind at its site resulted in the inability of Gcr1p to bind at adjacent binding sites. Synthetic oligonucleotides, modeled on the UAS element of PYK1, in which the relative positions of the Rap1p- and Gcr1p-binding sites were varied were prepared and tested for their ability to function as UAS elements. The ability of the oligonucleotides to function as UAS elements was dependent not only on the presence of both binding sites but also on the relative distance between the binding sites. In vivo DNA-binding studies showed that the ability of Rap1p to bind its site was independent of Gcr1p but that the ability of Gcr1p to bind its site was dependent on the presence of an appropriately spaced and bound Rap1p-binding site. In vitro binding studies showed Rap1p-enhanced binding of Gcr1p on oligonucleotides modeled after the native PYK1 UAS element but not when the Rap1p- and Gcr1p-binding sites were displaced by 5 nucleotides. This work demonstrates that the role of Rap1p in the activation of glycolytic enzyme genes is to bind in their UAS elements and to facilitate the binding of Gcr1p at adjacent binding sites. glycolytic enzymes (20, 33, 47, 49, 50). While the occurrence and importance of CT boxes in the UAS elements of glycolytic enzyme genes were first noted (4, 10, 15, 45) several years after the isolation of gcr1 mutant strains (21) and the cloning of GCR1 (33, 37), the identification of CT boxes as Gcr1p-binding sites (2, 35) was slow in coming. The ability to identify Gcr1p as a DNA-binding protein was hampered by the relatively low degree of specificity that Gcr1p displays for its DNA-binding site in vitro (34). However, in vivo Gcr1p-binding sites adjacent to bound Rap1p-binding sites are nearly fully occupied (35, 51, 54). To date, genetic analysis has implicated CT boxes as important features of the UAS elements of six glycolytic enzyme genes: TPI1 (51), TDH3 (4), ENO1 (10), ENO2 (58), PGK1 (15), and PYK1 (10). In most cases, mutation of the CT boxes reduced expression .10-fold. Although CT boxes are essential features of glycolytic enzyme gene UAS elements, they are not functional as UAS elements by themselves (4, 10, 51, 53). However, LexA-Gcr1p hybrid polypeptides are able to activate reporter genes which contain lex operators in place of their native UAS elements (51, 55). The ability of hybrid LexAGcr1p proteins to activate is not dependent on the presence of Rap1p-binding sites adjacent to the lex operator. In glycolytic enzyme gene UAS elements, Rap1p-binding sites are located immediately adjacent to CT boxes (4, 10, 17, 35, 58). Rap1p is an abundant multifunctional protein which is encoded by an essential gene (8, 52). Rap1p carries out many diverse cellular functions, and depending on context, Rap1p can act as both an activator and a repressor of transcription mediated by RNA polymerase II (8, 10, 13, 52). Rap1p also binds to telomeres, and mutations in RAP1 have been shown to effect telomeric length (22, 40). One of the intriguing questions

In Saccharomyces cerevisiae, the genes encoding glycolytic enzymes are among the most highly expressed (25, 30, 32). Much effort has been expended in characterizing the regulatory sequences governing expression of these genes. Studies to date have revealed that the upstream activating sequence (UAS) elements of glycolytic enzyme genes tend to be complex in nature, having binding sites for several different transcription factors (4, 7, 15, 16, 44, 51, 58). In this respect, they resemble enhancers and proximal promoters of higher eukaryotes. The hallmark of a glycolytic enzyme gene UAS element is the occurrence of one or more CT boxes adjacent to a Rap1p-binding site (4, 35, 45, 58). In addition to CT boxes and Rap1p-binding sites, these UAS elements usually have binding sites for either Reb1p (14, 18, 51) or Abf1p (7, 16). CT boxes are now known to be binding sites for the transcription factor Gcr1p (2, 34). Gcr1p appears to function primarily in the expression of glycolytic enzyme genes (20); however, there are some indications that it may be required for the expression of some other genes (19). Gcr1p is encoded by a nonessential gene which is expressed at a low level (1, 33). The first indication that glycolytic gene expression was coordinately controlled in S. cerevisiae came with the isolation of gcr1 mutants (21). The levels of glycolytic enzymes are severely reduced in gcr1 mutant strains (1, 20, 21), and several lines of evidence indicate that gcr1 lesions result in reduced levels of the transcripts specifying * Corresponding author. Mailing address: Department of Molecular Genetics and Microbiology, Box 100266, University of Florida, Gainesville, FL 32610-0266. Phone: (352) 392-0680. Fax: (352) 392-3133. Electronic mail address: [email protected]. 3187

3188

DRAZINIC ET AL.

MOL. CELL. BIOL.

concerning Rap1p function is how it can carry out seemingly opposite roles in transcription, functioning as an activator in some contexts and a repressor in others. Rap1p’s role in expression of glycolytic enzyme genes was first suggested when putative Rap1p-binding sites were noted in the 59 noncoding region of several of these genes (13). Ensuing biochemical and genetic analyses of the putative binding sites confirmed the role of Rap1p as an activator in the expression of glycolytic enzyme genes (12, 17, 41, 42, 44, 50, 51, 58). In most cases, mutations of Rap1p-binding sites in the UAS elements of glycolytic enzyme genes result in a .10-fold reduction in expression of the cognate gene. Rap1p-binding sites by themselves either do not function or function as relatively weak UAS elements (10, 23, 28, 51, 52, 59). However, when Rap1p-binding sites are located adjacent to Gcr1p-binding sites, a strong synergism is observed between the binding sites and their ability to act as UAS elements (4, 10, 17, 45, 51). Synergism has also been observed between Reb1p- and Abf1p-binding sites and the other binding sites (Rap1p and Gcr1p) which make up glycolytic enzyme gene UAS elements (7, 14, 16, 18, 51). Remacle and Holmberg (46) have shown that an Abf1p-binding site can functionally replace a Reb1pbinding site. On the basis of the observations of others (9, 18, 46), it has been suggested that the role of Reb1p and Abf1p binding at these UAS elements may be to prevent the formation of nucleosomes over the adjacent Rap1p- and Gcr1pbinding sites, thereby allowing free access to those binding sites (51). Several models (49, 51, 55) have been advanced to explain the interdependence of the transcription factors Rap1p and Gcr1p. It was recently suggested that Gcr1p binding to CT boxes is required for activation of Gcr1p-dependent genes and that the primary role of Rap1p binding at the UAS element of these genes is to facilitate the binding of Gcr1p at adjacent Gcr1p-binding sites (51). Here, we provide three independent lines of evidence that the in vivo DNA-binding activity of Gcr1p is dependent on the presence of Rap1p bound at an adjacent site. We also show that subtle differences in the positions of Rap1p- and Gcr1p-binding sites relative to one another have a large effect on the ability of Gcr1p to bind at its site and mediate expression of the cognate gene.

was used to prepare the XhoI-NheI restriction fragments for use as probes containing the Rap1p- and Gcr1p-binding sites under study. Probes containing both binding sites were 71 bp long after radiolabeling, whereas DR and DG were 60 and 66 bp long, respectively. Rap1p was synthesized in vitro in a rabbit reticulocyte lysate (2). The DNA-binding domain of Gcr1p was synthesized in Escherichia coli as part of a fusion protein between the maltose-binding protein (MBP) of E. coli and the carboxyl-terminal 154 amino acid residues of Gcr1p [MBP-Gcr1p(BD)] (35). Integrative transformations. The lithium acetate transformation method of Ito et al. (36) was used. Plasmid integration in the genetic background of yeast strain S150-2B was directed to the URA3 locus by predigestion of plasmids with the restriction endonuclease StuI, which cuts the reporter plasmids at a unique site within the URA3 gene. Transformants were plated on synthetic medium with the appropriate supplements to promote growth and to select for plasmid integration. To ensure that single-copy integrants were used for further study, transformants were screened by Southern blot hybridization experiments as described previously (50). Enzyme assays. b-Galactosidase activities of glass bead extracts of yeast cultures grown in YPD medium were determined essentially by the method of Miller (43) as described previously (51). Glass bead extracts of yeast cultures were prepared by the method of Himmelfarb et al. (31). Protein concentrations of the extracts were determined by the method of Bradford (5). Reported values are averages for independent transformants of single-copy integrants from at least three cultures. In vivo methylation protection studies. In vivo footprinting of the TPI1 UAS element was done essentially as described by Huie et al. (35). In experiments with strains YDS485 and YDS487, cultures were grown at 248C with shaking to an optical density (A600) of ca. 1. Cultures were concentrated 100-fold in clarified medium prior to treatment with dimethyl sulfate (DMS) at either 24 or 378C, the permissive and nonpermissive temperatures, respectively. Cultures treated at 378C were shifted to the nonpermissive temperature, after concentration, 30 min prior to DMS treatment. Unless otherwise stated, when DMS treatment was carried out at 378C, the concentration of DMS was reduced fivefold to a final concentration of 10 mM to avoid overmethylation of the genomic DNA. Following DMS treatment, chromosomal DNA was prepared and processed as described previously (35), except that in some cases chromosomal DNA was isolated by utilizing Qiagen-tip 500 columns according to the manufacturer’s recommendations. Yeast strains with synthetic sequences governing expression of the TPI1::lacZ reporter gene integrated at the URA3 locus were grown in YPD medium at 308C and treated with DMS, and chromosomal DNA was prepared as described previously (35). The resulting DNA was digested to completion with HindIII and processed as described previously for in vivo footprinting (35). Aliquots (5 mg) were loaded onto a 6% polyacrylamide–8 M urea gel. In adjacent lanes, chemical cleavage reaction mixtures of naked DNA were run to identify the areas of protection. Following electrophoresis, the gel was vacuum blotted (39) to a Hybond N1 (Amersham) filter, and the filter was processed for hybridization as described by Huie et al. (35).

MATERIALS AND METHODS

Rap1p binding required for Gcr1p binding in vivo. In vitro binding studies with the DNA-binding domain of Gcr1p suggested that Gcr1p binds at its binding sites with a relatively high affinity but a low degree of specificity (34). The low degree of specificity displayed by Gcr1p for its binding site in vitro suggested to us that there are factors in the cell that function to facilitate the binding of Gcr1p at its specific binding sites. The presence of Rap1p-binding sites next to Gcr1p-binding sites makes Rap1p an obvious candidate to facilitate the binding of Gcr1p. If Rap1p facilitates the binding of Gcr1p, then the inability of Rap1p to bind its site should lead to the inability of Gcr1p to occupy its site. Since Rap1p is encoded by an essential gene, we were unable to carry out in vivo DMS protection analysis (in vivo footprinting) of Gcr1p-binding sites in a rap1 null mutant. Alternatively, we sought to conduct in vivo DMS footprinting experiments with conditional mutants of RAP1. We obtained rap1ts mutants which had been previously isolated and characterized by Kurtz and Shore (38). Those investigators conducted in vitro DNA-binding studies which indicated that the DNA-binding activity of mutant Rap1p was defective at a high temperature. To determine if Rap1p binding facilitated the binding of Gcr1p at adjacent sites, we carried out in vivo footprinting of the TPI1 locus in the rap1-2ts mutant at the permissive and

Strains and growth conditions. The yeast strains used throughout this study are YDS485 (MATa ade2-1 his3-11,13 trp1-1 ura3-1), YDS487 (MATa ade2-1 his3-11,13 trp1-1 ura3-1 rap1-2ts), and S150-2B (MATa leu2-3,112 his3D trp1-289 ura3-52) and derivatives of S150-2B [MATa leu2-3,112 his3D trp1-289 ura3-52:: YIpCDxx(URA3 TPI1::lacZ)] which harbored a TPI1::lacZ reporter gene integrated at the URA3 locus under the control of various synthetic oligonucleotides. Cultures were grown in yeast extract-peptone medium (48) supplemented with 2% glucose (YPD). Unless otherwise noted, cultures were grown at 308C with shaking. In experiments involving temperature-sensitive strains, cultures were grown at permissive (248C) temperatures in YPD medium with shaking. Plasmids and plasmid constructions. Integrative plasmids (YIpCDxx series) used in this study are derivatives of YIpES90 (51), which carries a TPI1::lacZ gene fusion. Synthetic oligonucleotides shown below (see Fig. 2) were cloned, in two steps using standard techniques, between the unique HindIII and PacI restriction endonuclease sites of YIpES90. Introduction of the test sequences between the indicated restriction sites resulted in the loss of material from the TPI1 noncoding region 59 to position 2287 with respect to the translational start of TPI1. Removal of this sequence eliminates the native TPI1 UAS element which resides between positions 2401 and 2326. Thus, the synthetic oligonucleotides were positioned adjacent to the 39 end of URA3 at the HindIII site and 59 to the TPI1::lacZ reporter gene at the PacI site. The DNA sequence of the cloned synthetic oligonucleotide test sequences was confirmed by DNA sequence analysis. DNA band shift assays. DNA band shift assays based on the procedures of Fried and Crothers (26) and Garner and Revzin (27) were carried out as described previously (50). The typical reaction was carried out in a 20-ml volume which contained 0.5 to 1 ng of radiolabeled probe DNA along with 0.2 mg of poly(dI-dC) per ml as a nonspecific competitor. The YIpCDxx series of plasmids

RESULTS

VOL. 16, 1996

FIG. 1. In vivo guanine methylation protection footprints of the TPI1 UAS in wild-type and rap1-2ts mutant strains at permissive and nonpermissive temperatures. Lanes 1, 2, 8, and 13, guanine-specific chemical sequencing reactions of the TPI1 UAS; lanes 3 to 5, 5-min in vivo DMS treatment of the rap1-2ts mutant strain, YDS487, at the permissive temperature, 248C, and 30 and 60 min after a shift to the nonpermissive temperature, 378C, respectively; lanes 6 and 7, 2-min in vivo DMS treatment of strain YDS487 at the permissive temperature, 248C, and 30 min after a shift to the nonpermissive temperature, 378C; lanes 9 and 10, 5-min in vivo DMS treatment of the isogeneic wild-type RAP1 strain, YDS485, 30 min after a shift to 378C; lanes 11 and 12, 5-min in vivo DMS treatment of strain YDS485 after 30-min exposure to cycloheximide. (Lanes 9 through 13 were run on a different gel from the other lanes on the figure.) Sequencing reactions and in vivo footprints were resolved by electrophoresis on denaturing polyacrylamide gels, transferred to Hybond N1, and visualized after indirect end labeling by autoradiography. The positions of Reb1p-binding sites (stippled ovals), Gcr1pbinding sites (open ovals), and Rap1p-binding sites (closed ovals) are indicated.

nonpermissive temperatures. Figure 1 shows a series of in vivo footprints of the TPI1 locus when cultures were treated with DMS at the permissive and nonpermissive temperatures. At the permissive temperature, the four known binding sites in the TPI1 UAS element were nearly fully occupied. This pattern of protection was identical to that observed with a wild-type strain. At the nonpermissive temperature, the pattern of protection differed markedly from that observed at the permissive temperature. After as little as 30 min at the nonpermissive temperature, partial deprotection at the Rap1p-binding site

ACTIVATION MECHANISM OF Rap1p

3189

was detected, with corresponding deprotections of the adjacent Gcr1p-binding sites. In contrast to the marked deprotections noted at the Rap1p- and Gcr1p-binding sites, there was little, if any, change in the occupancy state of the Reb1p-binding site. Two sets of experiments were done to address the question of whether the deprotections observed at Gcr1p-binding sites were directly related to the dissociation of Rap1p from its binding site or if the dissociation of Gcr1p from its binding sites was due to trivial reasons. First, we addressed the ability of temperature per se to alter the occupancy state of the Gcr1p-binding sites. Accordingly, we conducted in vivo footprinting analyses of the TPI1 UAS element in a strain carrying a wild-type RAP1 allele at the temperatures considered permissive and nonpermissive for the rap1ts mutant strain. The patterns of protection in each case were identical at the two temperatures; the four known binding sites in the TPI1 UAS element were protected. Second, we considered the possibility that protein synthesis was inhibited in the rap1ts mutant strain at the nonpermissive temperature. If protein synthesis was inhibited, we reasoned that turnover of Gcr1p in the absence of synthesis may reduce the concentration of Gcr1p in the cell. According to the laws of mass action, such a reduction in Gcr1p concentration may result in reduced occupancy of the Gcr1p-binding sites. Since the expression level of Gcr1p in wild-type cells is below the level of detection of our anti-Gcr1p antibody in Western blot (immunoblot) analysis, we sought to inhibit protein synthesis and then to assess the in vivo occupancy of the Gcr1p-binding sites. To accomplish this task, we added the protein synthesis inhibitor cycloheximide, at a concentration (10 mg/ml) sufficient to arrest growth, to a yeast culture containing the wild-type RAP1 allele and carried out an in vivo footprint of the TPI1 UAS element. Again, the pattern of protection over the TPI1 UAS element was identical to the pattern observed with the wild-type strain treated under standard conditions. Namely, the four known binding sites in the TPI1 UAS element were nearly fully protected. These control experiments show that the deprotections observed at the Gcr1pbinding sites in the rap1ts mutant strain at the nonpermissive temperature are most likely the result of Rap1p dissociating from the neighboring binding site and not due to any effect which elevated temperature or decreased protein synthesis might have had on the ability of Gcr1p to bind its site. Importance of spacing between Rap1p- and Gcr1p-binding sites for transcriptional activation. The close association between Rap1p- and Gcr1p-binding sites is a distinctive feature of glycolytic enzyme gene UAS elements. The results presented above showed that Rap1p binding was required for Gcr1p binding in vivo. To investigate the importance of spacing between Rap1p- and Gcr1p-binding sites, we prepared a series of synthetic oligonucleotides in which spacing between a single Rap1p-binding site and a single Gcr1p-binding site was systematically varied. The oligonucleotides designed were modeled on the sequence of the UAS element of the PYK1 gene (11, 44), which in addition to binding sites for Rap1p and Gcr1p contains binding sites for Reb1p and Abf1p. Figure 2 shows the oligonucleotides used in this study. Starting with native (N) spacing between the binding sites, we systematically moved their relative positions by 5-nucleotide intervals. In the case of the first interval, 15, since the consensus binding sites overlap by 3 nucleotides, the three nucleotides of the overlap were duplicated and two residues from the 39 side of the Gcr1p-binding site were inserted between the binding sites. The net effect of this manipulation was to position the center of the Gcr1p-binding site 5 nucleotides further from the center of the Rap1p-binding site proximal to the TATA box. For the other oligonucleotides in which the spacing between the bind-

3190

DRAZINIC ET AL.

MOL. CELL. BIOL.

FIG. 2. Synthetic oligonucleotides used to test the importance of spacing between Rap1p- and Gcr1p-binding sites for UAS activity. (A) 59 extension to oligonucleotide sequences shown in panel B. (B) Oligonucleotides with varied spacing between Rap1p- and Gcr1p-binding sites. Unless stated otherwise, oligonucleotides used in this study extend directly from panel A to panel B (e.g., . . .TAAATCTCGAG. . .). The HindIII and PacI restriction endonuclease sites used for cloning into YIpES90 (underlined in panels A and B, respectively) are indicated.

ing sites was varied, blocks of 5 nucleotide residues were successively moved from the 39 side of the Gcr1p-binding site to the 59 side of the site. Two oligonucleotides (N and N15) in which the native spacing between the Rap1p- and Gcr1p-binding sites was maintained were prepared. These oligonucleotides, after cloning, differ in the relative position of the binding sites with respect to the TATA box. These oligonucleotides serve as controls for the oligonucleotides in which spacing between the binding sites is varied and in addition control for possible effects of moving the UAS with respect to the TATA box or other as yet unrecognized regulatory sequences. We also prepared two additional control oligonucleotides, DR and DG, in which the Rap1p-binding site and the core of the Gcr1p-binding site, respectively, were deleted. Once prepared, the oligonucleotides were cloned and sequenced. The ability of the synthetic oligonucleotides to bind Rap1p and Gcr1p in vitro was established by carrying out a series of DNA band shift experiments. Radiolabeled XhoI-NheI restriction fragments carrying the binding sites under study were prepared and used as probes with a rabbit reticulocyte lysate containing Rap1p and E. coli extracts containing the DNAbinding domain of Gcr1p. The results of the DNA band shift experiments are shown in Fig. 3. As expected, DR and DG were unable to bind Rap1p and Gcr1p, respectively. All of the re-

maining oligonucleotides were capable of binding both Rap1p and Gcr1p in vitro. The ability of the synthetic oligonucleotides to function in vivo as UAS elements was then tested. YIp reporter plasmids, harboring the TPI1::lacZ gene fusion under control of the various test sequences, were integrated into the yeast genome at the URA3 locus. Transformants were screened by Southern analysis to identify single-copy integrants (data not shown). Once single-copy integrants were in hand, we tested the ability of the synthetic oligonucleotides to function as UAS elements by measuring the levels of b-galactosidase expressed from the TPI1::lacZ reporter gene. Figure 4 shows the results of b-galactosidase assays of the various strains. Neither Rap1p- nor Gcr1p-binding sites alone functioned as a UAS element. The two oligonucleotides in which native spacing between the Rap1p- and Gcr1p-binding sites was maintained (N and N15) were fully functional as UAS elements and directed high-level expression of the reporter gene. When five nucleotide residues (oligonucleotide 15) were inserted between the binding sites, the ability of the binding sites to work together to form a functional UAS element was abolished. Insertion of 10 nucleotides (oligonucleotide 110) between the binding sites partially restored UAS activity, with the level of b-galactosidase activity being approximately 50% of that observed when native

VOL. 16, 1996

ACTIVATION MECHANISM OF Rap1p

3191

the Reb1p-binding site in our constructs. We deleted the Reb1pbinding site and surrounding sequence (see Fig. 2A for sequence deleted) and tested the ability of the oligonucleotide with native spacing between the Rap1p-, Gcr1p-, and Abf1pbinding sites (see N in Fig. 2B for nucleotide sequence remaining) to function as a UAS element. The pattern of expression of this construct was indistinguishable from that of the construct which contained the Reb1p-binding site (data not shown). Thus, the Abf1p- and Reb1p-binding sites do not appear to be playing a major role in the expression of the TPI1::lacZ reporter gene used in this study. Importance of spacing between Rap1p- and Gcr1p-binding sites for Gcr1p binding in vivo. The aforementioned results clearly demonstrate the importance of spatial relationships between Rap1p- and Gcr1p-binding sites for their ability to form functional UAS elements. The results are consistent with two models for combinatorial interactions between Rap1p and Gcr1p. In the first model, Gcr1p binding is required for activation, and Rap1p facilitates the binding of Gcr1p at the adjacent binding site. As the relative positions of the binding sites are altered, the ability of Rap1p to mediate binding of Gcr1p is impaired, resulting in the inability of the oligonucleotide to function as a UAS element. In the second model, Rap1p and Gcr1p bind to their sites independently of one another. The importance of spacing between the sites is related to the ability of Rap1p and Gcr1p to functionally interact with each other or another component of the transcriptional complex. The relative positions of the binding sites would serve to position Rap1p and Gcr1p in either an appropriate or inappropriate context for transcriptional activation to occur. The first model predicts that the in vivo occupancy state of the Gcr1p-binding site should be positively correlated with the ability of a given oligonucleotide to function as a UAS element, whereas the second model predicts that the occupancy state of the Gcr1p-binding site should be unaffected by changes in the relative distance between the Rap1p- and Gcr1p-binding sites.

FIG. 3. In vitro binding of Gcr1p and Rap1p to synthetic oligonucleotides used in this study. (A) Gcr1p-binding assays. DNA band shift analysis was carried out with the DNA-binding domain of Gcr1p synthesized in E. coli as part of a fusion protein between MBP and the carboxyl-terminal 154 amino acid residues of Gcr1p [MBP-Gcr1p(BD)]. DNA probes were prepared from the YIpCDxx series of plasmids as described in Materials and Methods. The oligonucleotides used in each lane are indicated. Standard DNA band shift reaction mixtures with (even-numbered lanes) and without (odd-numbered lanes) E. coli extract containing MBP-Gcr1p(BD) were used. (B) Rap1p-binding assays. DNA band shift analysis was carried out with Rap1p synthesized in a rabbit reticulocyte lysate system. Standard DNA band shift reaction mixtures with (even-numbered lanes) and without (odd-numbered lanes) rabbit reticulocyte lysate containing Rap1p were used.

spacing between the binding sites was maintained. When the binding sites were further separated (oligonucleotides 115, 120, 125, and 130), there was again a diminished ability of the oligonucleotides to function as UAS elements; however, in each of these cases, the level of b-galactosidase produced was significantly higher than that produced when only five nucleotide residues separated the binding sites. We made a similar series of constructs in which the spacing between the Rap1p- and Gcr1p-binding sites was varied but which did not contain the Abf1p-binding site. The pattern and the level of expression obtained with these constructs were essentially identical to the pattern of expression described above (data not shown). Likewise, we examined the importance of

FIG. 4. Ability of synthetic oligonucleotides with varied spacing between Rap1p- and Gcr1p-binding sites to function as UAS elements. b-Galactosidase activities were expressed from a TPI1::lacZ reporter gene integrated in a single copy at the URA3 locus under control of the various oligonucleotides shown in Fig. 2. Reported values are the means and standard deviations of at least three cultures of each indicated construct.

3192

DRAZINIC ET AL.

MOL. CELL. BIOL.

FIG. 5. In vivo occupancy state of Rap1p- and Gcr1p-binding sites as a function of distance between the sites. Odd-numbered lanes, guanine-specific chemical sequencing reactions of the synthetic oligonucleotides; even-numbered lanes, in vivo guanine methylation protection footprints of the synthetic oligonucleotides, integrated at the URA3 locus driving expression of the TPI1::lacZ reporter gene. Sequencing reactions and in vivo footprints were resolved by electrophoresis on denaturing polyacrylamide gels, transferred to Hybond N1, and visualized after indirect end labeling by autoradiography. The positions of Gcr1p-binding sites of the indicated constructs (open symbols) and Rap1p-binding sites (closed symbols) are shown (the rightmost set of closed symbols marks the position of the Rap1p-binding site in the N15 construct).

To discriminate between these two models, we used in vivo DMS methylation protection assays to determine the in vivo occupancy state of the Rap1p- and Gcr1p-binding sites under investigation in the various strains used in this study. Inspection of Fig. 5 and 6 shows that the occupancy state of the Gcr1p-binding site corresponds well with the level of expression of the reporter gene. No protection over the Gcr1p-binding site in the construct in which the Rap1p-binding site was deleted was observed (Fig. 5, lane 2). The Rap1p-binding site was nearly fully protected when the adjacent core Gcr1p-binding site was deleted (Fig. 5, lane 4). Both Rap1p- and Gcr1pbinding sites were protected in constructs in which native spacing between the binding sites was maintained (N and N15). When the distance between the centers of the binding sites was increased by 5 nucleotides (oligonucleotide 15), the Rap1pbinding site remained nearly fully protected, whereas the Gcr1p-binding site became susceptible to methylation, indicating that the Gcr1p-binding site was not occupied in vivo (Fig. 5, lane 10, and Fig. 6, lane 4). Further inspection of Fig. 5 shows the Gcr1p-binding site partially protected from methylation when 10 nucleotide residues (110) were inserted between the centers of the binding sites; again, the Rap1p-binding site remained nearly fully protected. Similar patterns of occupancy of the binding sites were noted when the sites were placed farther apart; the Rap1p-binding site remained pro-

tected, while the Gcr1p-binding site remained susceptible to methylation. The patterns of expression of the reporter gene and the patterns of protection observed in the in vivo footprinting experiments described above are consistent with the hypothesis that in vivo Rap1p facilitates the binding of Gcr1p at the UAS element and that Gcr1p binding is required for activation. Importance of spacing for Rap1p-facilitated binding of Gcr1p in vitro. The most striking differences observed with the synthetic oligonucleotides, in terms of both expression and in vivo binding, were between oligonucleotides N and 15. The in vivo methylation protection studies suggested that Rap1p facilitated the binding of Gcr1p at an appropriately spaced binding site but not when the binding sites were displaced by 5 nucleotides. To determine if Rap1p directly facilitated the binding of Gcr1p, we carried out a series of DNA band shift experiments with oligonucleotides N and 15. We hypothesized that if Rap1p directly facilitated the binding of Gcr1p at an appropriately spaced site, then we should observe cooperative binding of Rap1p and Gcr1p on oligonucleotide N, whereas we should observe independent binding of the proteins on oligonucleotide 15. Figure 7 shows the results of in vitro binding studies in which the abilities of Rap1p and Gcr1p to bind oligonucleotides N and 15 were determined both separately and together. Inspection of Fig. 7 reveals the appearance of a

VOL. 16, 1996

ACTIVATION MECHANISM OF Rap1p

FIG. 6. Importance of spacing between Rap1p- and Gcr1p-binding sites for in vivo occupancy of Gcr1p-binding sites. Lanes 1 and 3, guanine-specific chemical sequencing reactions of the synthetic oligonucleotides N and 15, respectively; lanes 2 and 4, in vivo guanine methylation protection footprints of oligonucleotides N and 15, respectively, integrated at the URA3 locus driving expression of the TPI1::lacZ reporter gene. Sequencing reactions and in vivo footprints were resolved by electrophoresis on denaturing polyacrylamide gels, transferred to Hybond N1, and visualized after indirect end labeling by autoradiography. The positions of the Gcr1p-binding site of the indicated constructs (open symbols) and of Rap1p-binding sites (closed symbols) are shown.

3193

binding sites. We interpret this result in the following manner. As the mutant is shifted to the nonpermissive temperature, the elevated temperature causes a conformational change in Rap1p that reduces the affinity of Rap1p for its binding site, leading to disassociation of Rap1p from the DNA. Once Rap1p disassociates from its binding site, the stabilizing effect of Rap1p on Gcr1p binding at adjacent Gcr1p-binding sites is lost, resulting in disassociation of Gcr1p from its binding sites. In a complementary set of experiments, we used synthetic oligonucleotides to explore the nature of the combinatorial interactions between Rap1p and Gcr1p. In these studies, the ability of synthetic oligonucleotides to function as UAS elements to direct the expression of a TPI1::lacZ reporter gene integrated at the URA3 locus was investigated. Neither a Rap1p-binding site nor a Gcr1p-binding site alone was sufficient to function as a UAS element. However, when the two sites were immediately adjacent to one another, as found in the PYK1 UAS element, they were able to mediate high-level expression of the reporter gene. In vivo footprint analysis of the synthetic oligonucleotides integrated at the URA3 locus showed that the Rap1p-binding site was occupied and that its occupancy was not dependent on the presence of Gcr1p bound at an adjacent Gcr1p-binding site. This finding is consistent with the observation that Rap1p-binding sites are occupied in gcr1 mutant strains (35, 54). Unlike Rap1p-binding sites, Gcr1p-binding sites were occupied only when immediately adjacent to a bound Rap1p-binding site. This observation complements those obtained from the in vivo DMS methylation protection experiments with the rap1ts mutant at the nonper-

single band of lower mobility when each oligonucleotide reacted with either Rap1p or Gcr1p alone. When both Rap1p and Gcr1p were present in the binding reaction mixture, three bands of altered mobility were observed, one band each corresponding to the oligonucleotide bound by either Gcr1p or Rap1p and a supershifted band corresponding to the oligonucleotide bound by both Rap1p and Gcr1p. A comparison of lanes 4 and 5 of Fig. 7 shows enhanced binding of both Rap1p and Gcr1p on oligonucleotide N compared with that on oligonucleotide 15. This observation is consistent with cooperative binding between Rap1p and Gcr1p on oligonucleotide N but not on oligonucleotide 15. DISCUSSION Rap1p-facilitated binding of Gcr1p in vivo. In S. cerevisiae, the UAS elements of glycolytic enzyme genes are composed of multiple binding sites for several different transcription factors. In combination, the binding sites form powerful promoter elements, yet alone they display little, if any, UAS activity. Key to the function of glycolytic enzyme gene UAS elements are binding sites for Rap1p and Gcr1p. Here, we provide three lines of experimental evidence which demonstrate the dependence of Gcr1p binding on the binding of Rap1p at an adjacent site. First, we studied the dependence of Gcr1p binding on Rap1p binding at the UAS element of the TPI1 gene in its natural genomic context on chromosome IV at the TPI1 locus. Utilizing a rap1ts mutant, we showed that at the nonpermissive temperature guanosine residues in both Rap1p-binding sites and Gcr1p-binding sites became susceptible to methylation by DMS, indicating that the proteins no longer occupied their

FIG. 7. Importance of spacing between Rap1p- and Gcr1p-binding sites on Rap1p-facilitated in vitro binding of Gcr1p. DNA band shift assays were carried out with DNA probes prepared as described in Materials and Methods. The sources of Rap1p and Gcr1p were as described in the legend to Fig. 3. In lanes 1 to 4, oligonucleotide N was used as a probe; in lanes 5 to 8, oligonucleotide 15 was used as a probe. Lanes 1 and 8, standard DNA binding mixture without added protein; lanes 2 and 7, standard binding mixture with E. coli extract containing MBP-Gcr1p(BD); lanes 3 and 6, standard binding mixture with rabbit reticulocyte lysate containing Rap1p; lanes 4 and 5, standard binding mixture containing both Gcr1p and Rap1p. Free indicates the position of the unbound probe.

3194

DRAZINIC ET AL.

missive temperature; namely, Gcr1p binding requires Rap1p binding at an adjacent site. The ability of the synthetic oligonucleotides to function as UAS elements correlated with the occupancy state of the Gcr1pbinding site in vivo. The in vivo occupancy of the Gcr1p-binding site was dependent on the presence of an appropriately spaced and bound Rap1p-binding site. The combinatorial interactions between Rap1p and Gcr1p can be interpreted in the following way: Rap1p’s function at glycolytic enzyme gene UAS elements is to facilitate the binding of Gcr1p, which plays a more direct role in the activation process. Uemura and Jigami (56) have provided data to suggest that the function of the activation activity of Gcr1p is to recruit Gcr2p to the activation complex. The observation that LexA-Gcr1p hybrid proteins can activate a lexAop::GAL1::lacZ reporter gene indicates that Gcr1p can activate transcription in the absence of Rap1p (51, 55). Thus, in the wild-type situation at glycolytic enzyme genes, Rap1p’s role in the activation process is to facilitate the binding of Gcr1p. The molecular mechanisms by which Gcr1p and the proteins that interact with it mediate high-level expression remain to be elucidated. From an engineering standpoint, a bipartite element in which spacing is an important variable allows for the formation of regulatory sequences of various strengths from a limited number of components. Inspection of the UAS elements of glycolytic enzyme genes reveals a rather large degree of heterogeneity between the relative positions of Rap1p-binding sites and Gcr1p-binding sites. In addition, many UAS elements appear to have several Gcr1p-binding sites. Gcr1p-binding sites can surround a Rap1p-binding site, as for the TPI1 UAS element (35), or several Gcr1p-binding sites can reside on one side of a Rap1p-binding site, as found in the PGK1 UAS element (29, 54). The spacing experiments we reported here involved moving a TATA-proximal Gcr1p-binding site. Whereas our results allow prediction of the relative strengths of UAS elements composed in a similar manner, at present, the strengths of UAS elements in which the relative orientations of individual sites are reversed cannot be inferred because of uncertainties as to the appropriate indexing point between the sites. In general, however, we propose that the activating potential of these UAS elements is a function of the amount of Gcr1p bound to the element over time. We further suggest that elements with optimal spacing between Rap1p- and Gcr1p-binding sites will be stronger than elements which have suboptimal spacing between the sites. It should be possible, however, to compensate for suboptimal spacing by introducing additional Gcr1p-binding sites, thereby increasing the probability that Gcr1p will be bound at the element at any given time. The further variable of number of binding sites in addition to spacing between sites may serve to increase the dynamic range of UAS elements composed of multiple binding sites. Models for Rap1p-facilitated binding of Gcr1p in vivo. There are several potential mechanisms by which Rap1p may facilitate the binding of Gcr1p. Rap1p may play a passive role in which, upon binding, Rap1p alters the local chromatin structure, thereby allowing Gcr1p access to its binding site. Such a mechanism has been proposed by Devlin et al. (23) for the role of Rap1p in the regulatory region of HIS4. At the HIS4 locus, Rap1p binding is thought to allow the binding of Bas1p/Bas2p and Gcn4p at the UAS element, although there is no direct evidence at this point. Alternatively, Rap1p may play an active role in the recruitment of Gcr1p. The observation of Rap1pfacilitated binding of Gcr1p in vitro argues strongly in favor of an active role for Rap1p in the binding of Gcr1p. There are two potential mechanisms by which cooperative binding reactions may occur. Rap1p may stabilize Gcr1p at neighboring binding

MOL. CELL. BIOL.

sites by interactions between the proteins (protein-protein interaction model), or Rap1p-induced bending of DNA may alter the topology of the adjacent Gcr1p-binding site (DNAbending model), thereby forming a Gcr1p-binding site of higher affinity. In both of these cooperative-binding models, the true in vivo binding substrate for Gcr1p is not just a naked DNA sequence but, rather, a nucleoprotein complex. The protein-protein interaction and DNA-bending models are not mutually exclusive, and both models may contribute in part to the overall binding reaction. Three observations argue against a model of passive Rap1pfacilitated binding of Gcr1p. First, the rapidity with which protection of the Gcr1p-binding site was lost in the rap1ts mutant strain after exposure to the nonpermissive temperature suggests that Rap1p binding is required to stabilize Gcr1p on its binding site in vivo (Fig. 1). Second, we note an oscillation and dampening in the level of expression of the reporter gene as the Rap1p- and Gcr1p-binding sites are displaced from one another (Fig. 3, N, 15, 110, and 115). Third, we cite the enhanced binding in vitro of Rap1p and Gcr1p to a DNA fragment containing appropriately spaced binding sites (Fig. 7) as evidence against a passive role for Rap1p in the binding of Gcr1p to its site. These three observations are consistent with both the protein-protein interaction and the DNA-bending models for the cooperative-binding interaction. If Rap1p stabilizes Gcr1p at its binding site by protein-protein interactions, then the spatial geometry of the binding sites would be expected to be important in positioning the proteins so that appropriate contacts between the proteins and their respective binding sites could be established. Inserting five nucleotides between the binding sites would displace the binding sites by one-half turn of the helix. It is not hard to envision how such a manipulation may prevent Gcr1p from making appropriate contacts with its binding site and Rap1p simultaneously. Insertion of 10 nucleotides results in the binding sites being displaced farther from one another; however, the binding sites remain on the same side of the helix with respect to their original positions. Both expression and binding studies showed that insertion of 10 nucleotides had a less severe effect than did the insertion of 5 nucleotides. Our interpretation of this result, according to the protein-protein interaction model, is that when the binding sites return to their original side of the helix, Gcr1p regains the ability to make simultaneous contacts with its binding site and Rap1p, albeit at a reduced frequency compared with that of elements with native spacing between the binding sites. Continuing this line of reasoning, as greater distances between the binding sites are introduced, the ability of Gcr1p to make the necessary appropriate contacts is greatly diminished. The bending model states that Rap1p-induced DNA bending changes the topology of the Gcr1p-binding site, resulting in the formation of high-affinity Gcr1p-binding sites. In the bending model, as with the previous model, the spatial geometry of the binding sites is critical for cooperative interactions between the binding sites to occur. Our results are also consistent with the bending model for the interdependence of Rap1p and Gcr1p. According to this model, when 5 nucleotides are inserted between the binding sites, the Gcr1p-binding site is displaced to the opposite side of the helix and is placed in an unfavorable conformation for Gcr1p binding, whereas insertion of 10 nucleotides results in a conformation somewhat less favorable than the native conformation but still allows a sufficient degree of Gcr1p binding to activate expression to approximately half that of the native spacing. As the sites are displaced to a greater extent, the ability of Rap1p to facilitate the binding of Gcr1p is further reduced.

VOL. 16, 1996

The data we presented here do not allow us to distinguish between the protein-protein interaction and bending models; however, there are reports which can be used to support the protein-protein interaction model. Tornow et al. (55) reported that antibody against epitope-tagged Rap1p can be used to immunoprecipitate Gcr1p in an epitope-tagged Rap1p-dependent manner. If the protein-protein interaction model is solely responsible for Rap1p-facilitated binding of Gcr1p, then the in vitro binding study results we presented in Fig. 7 map a Rap1p interaction domain of Gcr1p to the carboxyl-terminal 154 amino acid residues of the polypeptide. Gcr1p may also have an additional Rap1p interaction domain in the amino-terminal half of the protein. Both Tornow et al. (55) and Uemura and Jigami (57) have provided evidence that certain gcr1 mutant constructs, carried on high-copy-number plasmids which express truncated forms of Gcr1p lacking its DNA-binding domain, are able to complement gcr1 mutants. The apparent complementation of gcr1 mutants by truncated forms of Gcr1p may be explained by mass action. If Rap1p facilitates the binding of Gcr1p to adjacent Gcr1p-binding sites by virtue of proteinprotein interactions, then overexpression of Gcr1p or truncated forms of it may drive an association between Rap1p and Gcr1p in the absence of DNA contacts. The observation (10, 23, 28, 51, 52, 59) that Rap1p-binding sites by themselves are not sufficient to mediate Gcr1p-dependent gene expression indicates that in wild-type cells expressing normal levels of Rap1p and Gcr1p, the association between the proteins in the absence of Gcr1p-binding sites is not sufficient to mediate Gcr1p-dependent gene expression. Further experimentation will be required to elucidate the mechanism(s) by which Rap1p facilitates the binding of Gcr1p. We emphasize that the protein-protein interaction model and the bending model are not mutually exclusive. We are not aware of any experimental results related to the binding of Gcr1p which can be used to exclude either model. The arrangement of tandem Gcr1p-binding sites adjacent to a Rap1p-binding site, as is the case at the PGK1 UAS element (54), raises the possibility that once bound, Gcr1p itself may promote the binding of other molecules of Gcr1p to neighboring sites. Recently, Uemura and Jigami (57), using the yeast two-hybrid system, have provided data showing that Gcr1p has the capacity to interact with itself. In vivo footprint analysis of the PGK1 UAS element revealed that tandem Gcr1p-binding sites are occupied adjacent to the Rap1p-binding site (29). Here, we have described the combinatorial interactions that occur between Rap1p and Gcr1p to mediate high-level glycolytic enzyme gene expression. Rap1p-binding sites are also found in the UAS elements of numerous genes whose expression is not dependent on Gcr1p. Recently, Gonc¸alves et al. (28) suggested that Rap1p potentiated transcriptional activation through the T-rich element found in the UAS elements of many ribosomal protein genes. We propose that the general role of Rap1p in the activation process is to act as an architectural protein to facilitate the binding of the transcriptional activator at an adjacent binding site, be it Gcr1p or another, as yet uncharacterized activator protein. Function of Reb1p and Abf1p at glycolytic enzyme gene UAS elements. The presence or absence of Reb1p- or Abf1p-binding sites on the synthetic oligonucleotides used in this study did not affect the ability of the oligonucleotides to drive expression of the TPI1::lacZ reporter gene integrated at the URA3 locus. Several groups have proposed that Reb1p’s function in the activation process is to create a nucleosome-free area in the vicinity of its binding site, thereby allowing other activator proteins access to binding sites (6, 24). The ability of an Abf1pbinding site to functionally replace a Reb1p-binding site (46)

ACTIVATION MECHANISM OF Rap1p

3195

suggests that Abf1p may play the same role as Reb1p. The dispensability of the Reb1p-binding site in the TPI1 UAS element when the element was located adjacent to the HindIII restriction site at the 39 end of URA3 (50), yet a requirement for the Reb1p-binding site when sequence 59 to the site was included in the reporter gene construct (51), was previously noted. The work of Bernardi et al. (3) on nucleosome positioning elements may provide an answer to the apparent paradox as to the necessity of Reb1p-binding sites and, by extension, the Abf1p-binding sites. Bernardi et al. (3) have reported the presence of sequence, most likely a protein-binding site, at the 39 end of URA3 which inhibits the formation of nucleosomes. Studies with plasmids indicated that the nucleosomefree region originating at the 39 end of URA3 extended past the HindIII site into plasmid sequence (3). In the studies we have reported, we have used the HindIII site at the 39 end of URA3 for cloning the synthetic oligonucleotides under study 59 to the TPI1::lacZ reporter gene. Thus, we may have eliminated the need for the Reb1p-binding site by cloning the oligonucleotides downstream from URA3 adjacent to a sequence which performs a similar function. We are in the process of studying the function of Reb1p and Abf1p and their binding sites away from the end of URA3 in a more natural chromatin context. ACKNOWLEDGMENTS We thank David Shore for the generous gift of rap1ts mutant yeast strains and the reviewers for their insightful suggestions. This work was supported by grant MCB-9404721 from the National Science Foundation. REFERENCES 1. Baker, H. V. 1986. Glycolytic gene expression in Saccharomyces cerevisiae: nucleotide sequence of GCR1, null mutations, and evidence for expression. Mol. Cell. Biol. 6:3774–3784. 2. Baker, H. V. 1991. GCR1 of Saccharomyces cerevisiae encodes a DNA binding protein whose binding is abolished by mutations in the CTTCC sequence motif. Proc. Natl. Acad. Sci. USA 88:9443–9447. 3. Bernardi, F., M. Zatchej, and F. Thoma. 1992. Species specific protein-DNA interactions may determine the chromatin units of genes in S. cerevisiae and in S. pombe. EMBO J. 11:1177–1185. 4. Bitter, G. A., K. K. H. Chang, and K. M. Egan. 1991. A multi-component upstream activation sequence of the Saccharomyces cerevisiae glyceraldehyde-3-phosphate dehydrogenase gene promoter. Mol. Gen. Genet. 231:22– 32. 5. Bradford, M. M. 1976. A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding. Anal. Biochem. 72:248–254. 6. Brandl, C. J., and K. Struhl. 1990. A nucleosome-positioning sequence is required for GCN4 to activate transcription in the absence of a TATA element. Mol. Cell. Biol. 10:4256–4265. 7. Brindle, P. K., J. P. Holland, C. E. Willett, M. A. Innis, and M. J. Holland. 1990. Multiple factors bind the upstream activation sites of the yeast enolase genes ENO1 and ENO2: ABF1 protein, like repressor-activator protein RAP1, binds cis-acting sequences which modulate repression or activation of transcription. Mol. Cell. Biol. 10:4872–4895. 8. Buchman, A. R., W. J. Kimmerly, J. Rine, and R. D. Kornberg. 1988. Two DNA-binding factors recognize specific sequences at silencers, upstream activating sequences, autonomously replicating sequences, and telomeres in Saccharomyces cerevisiae. Mol. Cell. Biol. 8:210–225. 9. Buchman, A. R., and R. D. Kornberg. 1990. A yeast ARS-binding protein activates transcription synergistically in combination with other weak activating factors. Mol. Cell. Biol. 10:887–897. 10. Buchman, A. R., N. F. Lue, and R. D. Kornberg. 1988. Connections between transcriptional activators, silencers, and telomeres as revealed by functional analysis of a yeast DNA-binding protein. Mol. Cell. Biol. 8:5086–5099. 11. Burke, R. L., P. Tekamp-Olson, and R. Najarian. 1983. The isolation, characterization, and sequence of the pyruvate kinase gene of Saccharomyces cerevisiae. J. Biol. Chem. 258:2193–2201. 12. Butler, G., I. W. Dawes, and D. J. McConnell. 1990. TUF factor binds to the upstream region of the pyruvate decarboxylase structural gene (PDC1) of Saccharomyces cerevisiae. Mol. Gen. Genet. 223:449–456. 13. Capieaux, E., M.-L. Vignais, A. Sentenac, and A. Goffeau. 1989. The yeast H1-ATPase gene is controlled by the promoter binding factor TUF. J. Biol. Chem. 264:7437–7446.

3196

DRAZINIC ET AL.

14. Carmen, A. A., and M. J. Holland. 1994. The upstream repression sequence from the yeast enolase gene ENO1 is a complex regulatory element that binds multiple trans-acting factors including REB1. J. Biol. Chem. 269:9790– 9797. 15. Chambers, A., C. Stanway, A. J. Kingsman, and S. M. Kingsman. 1988. The UAS of the yeast PGK gene is composed of multiple functional elements. Nucleic Acids Res. 16:8245–8260. 16. Chambers, A., C. Stanway, J. S. H. Tsang, Y. Henry, A. J. Kingsman, and S. M. Kingsman. 1990. ARS binding factor 1 binds adjacent to RAP1 at the UASs of the yeast glycolytic genes PGK and PYK. Nucleic Acids Res. 18: 5393–5399. 17. Chambers, A., J. S. H. Tsang, C. Stanway, A. J. Kingsman, and S. M. Kingsman. 1989. Transcriptional control of the Saccharomyces cerevisiae PGK gene by RAP1. Mol. Cell. Biol. 9:5516–5524. 18. Chasman, D. I., N. F. Lue, A. R. Buchman, J. W. LaPointe, Y. Lorch, and R. D. Kornberg. 1990. A yeast protein that influences the chromatin structure of UASG and functions as a powerful auxiliary gene activator. Genes Dev. 4:503–514. 19. Ciriacy, M., K. Freidel, and C. Lo ¨hning. 1991. Characterization of transacting mutations affecting Ty and Ty-mediated transcription in Saccharomyces cerevisiae. Curr. Genet. 20:441–448. 20. Clifton, D., and D. G. Fraenkel. 1981. The gcr1 (glycolysis regulation) mutation of Saccharomyces cerevisiae. J. Biol. Chem. 256:13074–13078. 21. Clifton, D., S. B. Weinstock, and D. G. Fraenkel. 1978. Glycolysis mutants of Saccharomyces cerevisiae. Genetics 88:1–11. 22. Conrad, M. N., J. H. Wright, A. J. Wolf, and V. A. Zakian. 1990. RAP1 protein interacts with yeast telomeres in vivo: overproduction alters telomere structure and decreases chromosome stability. Cell 63:739–750. 23. Devlin, C., K. Tice-Baldwin, D. Shore, and K. T. Arndt. 1991. RAP1 is required for BAS1/BAS2- and GCR4-dependent transcription of the yeast HIS4 gene. Mol. Cell. Biol. 11:3642–3651. 24. Fedor, M. J., N. F. Lue, and R. D. Kornberg. 1988. Statistical positioning of nucleosomes by specific protein-binding to an upstream activating sequence in yeast. J. Mol. Biol. 204:109–127. 25. Fraenkel, D. G. 1982. Carbohydrate metabolism, p. 1–37. In J. N. Strathern, E. W. Jones, and J. R. Borach (ed.), The molecular biology of the yeast Saccharomyces: metabolism and gene expression. Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y. 26. Fried, M., and D. M. Crothers. 1981. Equilibria and kinetics of lac repressoroperator interactions by polyacrylamide gel electrophoresis. Nucleic Acids Res. 9:6505–6525. 27. Garner, M. M., and A. Revzin. 1981. A gel electrophoresis method for quantifying the binding of proteins to specific DNA regions: applications to components of Escherichia coli lactose operon regulatory systems. Nucleic Acids Res. 9:3037–3060. 28. Gonc¸alves, P. M., G. Griffioen, R. Minnee, M. Bosma, L. S. Kraakman, W. H. Mager, and R. J. Planta. 1995. Transcriptional activation of yeast ribosomal protein genes requires additional elements apart from binding sites for Abf1p or Rap1p. Nucleic Acids Res. 23:1475–1480. 29. Henry, Y. A. L., M. C. Lo´pez, J. M. Gibbs, A. Chambers, S. M. Kingsman, H. V. Baker, and C. A. Stanway. 1994. The yeast protein Gcr1p binds to the PGK UAS and contributes to the activation of transcription of the PGK gene. Mol. Gen. Genet. 245:506–511. 30. Hess, B., A. Boiteux, and J. Kruger. 1969. Cooperation of glycolytic enzymes. Adv. Enzyme Regul. 7:149–169. 31. Himmelfarb, H. J., J. Pearlberg, D. H. Last, and M. Ptashne. 1990. GAL11P: a yeast mutation that potentiates the effect of weak GAL4-derived activators. Cell 63:1299–1309. 32. Holland, M. J., and J. P. Holland. 1978. Isolation and identification of yeast messenger ribonucleic acids coding for enolase, glyceraldehyde-3-phosphate dehydrogenase, and phosphoglycerate kinase. Biochemistry 17:4900–4907. 33. Holland, M. J., T. Yokoi, J. P. Holland, K. Myambo, and M. A. Innis. 1987. The GCR1 gene encodes a positive transcriptional regulator of the enolase and glyceraldehyde-3-phosphate dehydrogenase gene families in Saccharomyces cerevisiae. Mol. Cell. Biol. 7:813–820. 34. Huie, M. A., and H. V. Baker. DNA-binding properties of the yeast transcriptional activator, Gcr1p. Yeast, in press. 35. Huie, M. A., E. W. Scott, C. M. Drazinic, M. C. Lopez, I. K. Hornstra, T. P. Yang, and H. V. Baker. 1992. Characterization of the DNA-binding activity of GCR1: in vivo evidence for two GCR1-binding sites in the upstream activating sequence of TPI of Saccharomyces cerevisiae. Mol. Cell. Biol. 12: 2690–2700. 36. Ito, H., K. Fukuda, K. Murata, and A. Kimura. 1983. Transformation of

MOL. CELL. BIOL. intact yeast cells treated with alkali cations. J. Bacteriol. 153:163–168. 37. Kawasaki, G., and D. G. Fraenkel. 1982. Cloning of yeast glycolysis genes by complementation. Biochem. Biophys. Res. Commun. 108:1107–1112. 38. Kurtz, S., and D. Shore. 1991. RAP1 protein activates and silences transcription of mating-type genes in yeast. Genes Dev. 5:616–628. 39. Lo ´pez, M. C., J. B. Smerage, and H. V. Baker. 1993. A simplified vacuum blotting method for genomic sequencing and in vivo footprinting. BioTechniques 15:362–363. 40. Lustig, A. J., S. Kurtz, and D. Shore. 1990. Involvement of the silencer and UAS binding protein RAP1 in regulation of telomere length. Science 250: 549–553. 41. Machida, M., Y. Jigami, and H. Tanaka. 1989. Purification and characterization of a nuclear factor which binds specifically to the upstream activation sequence of Saccharomyces cerevisiae enolase 1 gene. Eur. J. Biochem. 184: 305–311. 42. McNeil, J. B., P. Dykshoorn, J. N. Huy, and S. Small. 1990. The DNAbinding protein RAP1 is required for efficient transcriptional activation of the yeast PYK glycolytic gene. Curr. Genet. 18:405–412. 43. Miller, J. H. 1972. Experiments in molecular genetics. Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y. 44. Nishizawa, M., R. Araki, and Y. Teranishi. 1989. Identification of an upstream activating sequence and an upstream repressible sequence of the pyruvate kinase gene of the yeast Saccharomyces cerevisiae. Mol. Cell. Biol. 9:442–451. 45. Ogden, J. E., C. A. Stanway, S. Kim, J. Mellor, A. J. Kingsman, and S. M. Kingsman. 1986. Efficient expression of the Saccharomyces cerevisiae PGK gene depends upon an upstream activation sequence but does not require a TATA sequence. Mol. Cell. Biol. 6:4335–4343. 46. Remacle, J. E., and S. Holmberg. 1992. A REB1-binding site is required for GCN4-independent ILV1 basal-level transcription and can be functionally replaced by an ABF1-binding site. Mol. Cell. Biol. 12:5516–5526. 47. Rodicio, R., J. J. Heinisch, and C. P. Hollenberg. 1993. Transcriptional control of yeast phosphoglycerate mutase-encoding gene. Gene 125:125– 133. 48. Rose, M. D., F. Winston, and P. Hieter. 1990. Methods in yeast genetics. Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y. 49. Santangelo, G. M., and J. Tornow. 1990. Efficient transcription of the glycolytic gene ADH1 and three translational component genes requires the GCR1 product, which can act through TUF/GRF/RAP binding sites. Mol. Cell. Biol. 10:859–862. 50. Scott, E. W., H. E. Allison, and H. V. Baker. 1990. Characterization of TPI gene expression in isogeneic wild-type and gcr1-deletion mutant strains of Saccharomyces cerevisiae. Nucleic Acids Res. 18:7099–7107. 51. Scott, E. W., and H. V. Baker. 1993. Concerted action of the transcriptional activators REB1, RAP1, and GCR1 in the high-level expression of the glycolytic gene TPI. Mol. Cell. Biol. 13:543–550. 52. Shore, D., and K. Nasmyth. 1987. Purification and cloning of a DNA binding protein from yeast that binds to both silencers and activator elements. Cell 51:721–732. 53. Stanway, C. A., A. Chambers, A. J. Kingsman, and S. M. Kingsman. 1989. Characterization of the transcriptional potency of sub-elements of the UAS of the yeast PGK gene in a PGK mini-promoter. Nucleic Acids Res. 17:9205– 9218. 54. Stanway, C. A., J. M. Gibbs, S. E. Kearsey, M. C. Lo´pez, and H. V. Baker. 1994. The yeast co-activator GAL11 positively influences transcription of the phosphoglycerate kinase gene, but only when RAP1 is bound to its upstream activation sequence. Mol. Gen. Genet. 243:207–214. 55. Tornow, J., X. Zeng, W. Gao, and G. M. Santangelo. 1993. GCR1, a transcriptional activator in Saccharomyces cerevisiae, complexes with RAP1 and can function without its DNA binding domain. EMBO J. 12:2431– 2437. 56. Uemura, H., and Y. Jigami. 1992. Role of GCR2 in transcriptional activation of yeast glycolytic genes. Mol. Cell. Biol. 12:3834–3842. 57. Uemura, H., and Y. Jigami. 1995. Mutations in GCR1, a transcriptional activator of Saccharomyces cerevisiae glycolytic genes, function as suppressors of gcr2 mutations. Genetics 139:511–521. 58. Willett, C. E., C. M. Gelfman, and M. J. Holland. 1993. A complex regulatory element from the yeast ENO2 gene modulates GCR1-dependent transcriptional activation. Mol. Cell. Biol. 13:2623–2633. 59. Woudt, L. P., A. B. Smit, W. H. Mager, and R. J. Planta. 1986. Conserved sequence elements upstream of the gene encoding yeast ribosomal protein L25 are involved in transcription activation. EMBO J. 5:1037–1040.