Identification of a Multifunctional Domain in Autonomously Replicating Sequence-Binding Factor 1 Required for Transcriptional Activation, DNA Replication, and Gene Silencing

Abstract

Autonomously replicating sequence-binding factor 1 (ABF1) is a multifunctional, site-specific DNA binding protein that is essential for cell viability in Saccharomyces cerevisiae. ABF1 plays a direct role in transcriptional activation, stimulation of DNA replication, and gene silencing at the mating-type loci. Here we demonstrate that all three activities of ABF1 are conferred by the C terminus of the protein (amino acids [aa] 604 to 731). Furthermore, a detailed mutational analysis has revealed two important clusters of amino acid residues in the C terminus (C-terminal sequence 1 [CS1], aa 624 to 628; and CS2, aa 639 to 662). While both regions play a pivotal role in supporting cell viability, they make distinct contributions to ABF1 functions in various nuclear processes. CS1 specifically participates in transcriptional silencing and/or repression in a context-dependent manner, whereas CS2 is universally required for all three functions of ABF1. When tethered to specific regions of the genome, a 30-aa fragment that contains CS2 alone is sufficient for activation of transcription and chromosomal replication. In addition, CS2 is responsible for ABF1-mediated chromatin remodeling. Based on these results, we suggest that ABF1 may function as a chromatin-reorganizing factor to increase accessibility of the local chromatin structure, which in turn facilitates the action of additional factors to establish either an active or repressed chromatin state.

The temporal and spatial regulation of chromatin structure is a critical determinant of every aspect of nuclear functions in eukaryotes. Intense research in recent years has shed much mechanistic insight into the trans-acting factors that participate in chromatin reorganization (66). Furthermore, there has been a wealth of information collected concerning the molecular and biochemical nature of changes in chromatin structure, resulting from this regulation, that ultimately lead to activation or repression of various nuclear functions. For example, it is now well accepted that covalent modifications of histone tails, including acetylation, phosphorylation, methylation, and ubiquitination, constitute an important code system that dictates the state of competence of a particular region of the genome for the initiation of a nuclear event (56). This modification process requires the complex action of numerous histone-modifying enzymes, many of which exist in multiprotein complexes (3, 52). In addition, other chromatin-remodeling complexes (e.g., the ATP-dependent SWI/SNF complex) work in concert with the histone-modifying enzymes to control the degree of compactness of chromatin structure and accessibility of the DNA therein (15, 64).

The ability of chromatin-modifying complexes to locate and function on their coordinate cis elements is critical to the biology of the cell and our understanding of it. Mounting evidence has suggested that many of these complexes are recruited to specific regions of the genome by DNA binding proteins that recognize specific sequences and/or structures. For example, both in vivo and in vitro studies demonstrate that site-specific transcription factors target histone acetyltransferases and/or chromatin-remodeling complexes to a particular transcriptional promoter, which then triggers changes in chromatin structure and activation of gene expression (9, 68). While much has been learned about the process of recruitment of chromatin-modifying complexes during activation of gene expression, less is known on the role of chromatin remodeling in regulation of other nuclear events. However, emerging evidence suggests that changes in chromatin structure associated with other nuclear processes also involve the same sequence-specific transcription factors that activate gene expression. For instance, it is well established that transcription factors bind to auxiliary sequences adjacent to many viral and cellular replication origins, induce changes in chromatin structure, and enhance replication origin usage (23, 55, 61). Likewise, transcription factors have also been implicated in gene silencing (29), DNA repair, and recombination (42). Present data support the notion that the transcription factors that are involved in these various nuclear events may use a similar strategy to facilitate changes in chromatin structure and to help assemble an initiation complex for these processes (6). However, it is not clear whether one transcription factor uses different functional domains and/or recruits distinct chromatin-modifying complexes to fulfill its functions in multiple and diverse chromosomal events.

The yeast Saccharomyces cerevisiae has served as an excellent model for studying the mechanism by which transcription factors facilitate multiple nuclear processes. For example, the repressor/activator protein 1 (RAP1) is an extensively characterized, multifunctional transcription factor. In addition to its role in transcriptional activation, RAP1 also plays an important role in transcriptional repression, gene silencing, recombination, and maintenance of telomere length (41, 53). The entire 827-amino-acid (aa) protein can be divided into N- and C-terminal domains of approximately equal size, separated by a central DNA binding domain (DBD). RAP1’s functions in these processes are mediated by distinct domains in the C-terminal region of the protein (41, 53).

Another yeast multifunctional protein, autonomously replicating sequence (ARS)-binding factor 1 (ABF1), shares much functional similarity with RAP1. Like RAP1, ABF1 is abundant and essential for cell growth (13, 20, 51). ABF1 binds to several ARSs and stimulates initiation of DNA replication from these yeast origins of replication. Both ABF1 and RAP1 consensus binding sites have been found in the promoter regions of numerous yeast genes (11, 41), and Abf1 has been shown to activate or repress transcription of many genes involved in disparate processes, including carbon source regulation (16, 54, 63), sporulation (12, 44), and ribosomal functions (40, 45). Furthermore, ABF1 and RAP1 work in concert to prevent gene expression at the silenced mating-type loci (29). Abf1 has also been implicated in gene silencing within the subtelomeric regions (46) and nucleotide excision repair of silenced chromosomal regions (48). The C-terminal regions of Abf1 and Rap1p share limited sequence homology (13), and indeed they are functionally interchangeable in cell viability assays (19).

The multifunctional nature of ABF1 and RAP1 suggests that they may employ similar mechanisms, such as chromatin remodeling, to stimulate various nuclear processes. In support of this notion, numerous studies have shown that ABF1 binding to ARS1 is responsible for preventing nucleosomal encroachment into the core of the replication origin. Venditti et al. (62) have shown that the presence of ABF1 reduces the number of nucleosomes positioned at ARS1 and specifically prevents nucleosomal invasion of the ARS consensus sequence (A box). Further work by Hu and colleagues (25) has shown that the ABF1 binding site at ARS1 (B3 box) is responsible for an alteration of the local chromatin structure. A recent in vitro biochemical study has also demonstrated that the ABF1-dependent nucleosome reconfiguration at ARS1 is required for efficient assembly of the prereplicative complex and subsequent origin firing (34).

ABF1 may function in a similar manner to activate transcription. At the promoter of the RPS28A ribosomal protein gene, both the ABF1 binding site and a T-rich element are required to generate a nucleosome-free region over the promoter and loss of ABF1 binding results in a loss of ordered nucleosome positioning (28). Likewise, RAP1 has also been shown to induce changes in chromatin structure at certain transcriptional promoters (67). Recent work also suggests that ABF1 and RAP1 that are bound at the promoters of many ribosomal genes act together to recruit Esa1, an essential histone acetyltransferase in S. cerevisiae (49). Taken together, these studies support the notion that ABF1- and RAP1-mediated reconfiguration of chromatin structure may be a common step in activation of multiple nuclear processes.

Compared with our knowledge of RAP1, much less is understood about the relationship between ABF1 structure and its multiple functions. Previous work has localized the DBD to the first 500 aa of the protein, which can be further divided into an essential, novel zinc finger domain (aa 40 to 91) and a unique region that determines the sequence specificity of binding (aa 323 to 496) (8). The region downstream of the DBD contains a trans-activation domain (aa 608 to 731) for transcription and replication (32). However, a systematic dissection of the ABF1 domains responsible for its multiple nuclear functions is lacking. To address this important issue, we carried out a structural and functional analysis of ABF1. Our work led us to the identification of two critical regions in the C terminus of the protein (C-terminal sequence 1 [CS1] and CS2) that are required for ABF1 functions. CS2 corresponds to one of the two highly conserved regions at the C termini of various Abf1 homologues. While the first region (CS1) is specialized in transcriptional silencing and/or repression, the second region (CS2) is universally involved in all three aspects of ABF1 nuclear functions. Finally, we present evidence suggesting that CS2 plays an important role in reorganizing the local chromatin structure.

MATERIALS AND METHODS

Yeast media and genetic analysis.

Standard techniques were used for preparation of yeast media, genetic analysis, and yeast transformation (1).

Strains and plasmids.

Strains used in this study are listed in Table 1. The strain TMY71 was made by transforming strain W303a with a PCR-amplified fragment containing the sequence of HIS3 from pFA6a-His3MX6 (35). The 60-mer primers used in the PCR contain 40 bases corresponding to the 5′ or 3′ region of the ABF1 open reading frame. Replacement of the ABF1 gene with His3MX6 was confirmed by PCR. TMY86(MATa) and TMY87(MATα) were obtained by tetrad dissection of the TMY71 strain carrying pRS416-ABF1.

TABLE 1.

Strains in this study

Strain	Description	Reference or source
RL1	MATaura3-52 leu2-3,112 his4-519 ade1gal4::HIS4 ars1/-B123/G5Δtrp1	32
JRY4997	mata1 abf1-102 ste14 ade2 leu2-3,112 lys2-801 ura3	36
JRY2726	MATahis4	36
YSB35	MATα ade2-1 his3-11,15 leu2-3,112 trp1-1 ura3-1 can1-100 gal4::LEU2 ΔEΔB hmr3xUASG::TRP1	5
YSB67	MATα ade2-1 his3-11,15 leu2-3,112 trp1-1 ura3-1 can1-100 gal4::LEU2 ΔEΔB hmr::TRP1	5
W303a	MATa/α ade2-1/ade2-1 his3-11,15/his3-11,15 leu2-3,112/leu2-3,112 trp1-1/trp1-1 ura3-1/ura3-1 can1-100/can1-100	59
TMY71	Same as W303a except Δabf1::HIS3MX6/ABF1	This study
TMY86	MATaade2-1 his3-11,15 leu2-3,112 trp1-1 ura3-1 can1-100 Δabf1::HIS3MX6 pRS416-ABF1	This study
TMY87	MATα ade2-1 his3-11,15 leu2-3,112 trp1-1 ura3-1 can1-100 Δabf1::HIS3MX6 pRS416-ABF1	This study

The plasmid pRS416-ABF1 was derived from pRS315-ABF1 (50). The ABF1 C-terminal deletion constructs were made by either PCR or the mutagenesis method described previously (27). All alanine-scanning mutation constructs were made by the mutagenesis method as described earlier (27). Plasmid YCAL is a pRS315 derivative containing the ADH1 promoter that drives the expression of the hemagglutinin (HA)-tagged GAL4 DBD (aa 1 to 94). This plasmid was used to construct the GAL4 fusion genes used in the β-galactosidase and plasmid stability assays. PCR fragments were cloned into the XbaI and BamHI sites immediately downstream of HA-GAL4 (1-94). Plasmids pSB362 and pSB136 were used in the HMR-E targeting assay: the former for constructing the GAL4-ABF1 fusion genes and the latter for the GAL4-RAP1 (aa 653 to 827) positive control (5). Both pSB362 and pSB136 contain the DNA sequence that encodes the GAL4 DBD (aa 1 to 147), which is driven by the native RAP1 promoter in the pRS313 backbone. GAL4-ABF1 fusion genes in the pSB362 background were constructed by cloning ABF1 PCR fragments into the SmaI and BamHI sites immediately downstream of the GAL4 DBD. The copper-inducible GAL4 expression vectors used in the two-dimensional (2-D) gel assay were described previously (32).

The pARS1 and pARS1-B3 plasmids used in the plasmid stability assays were described previously (38), except that the sup4-o gene was cloned at the SphI restriction site. The pARS1/-B23/G5 plasmid used in the plasmid stability assay was described previously (32). The lacZ reporter plasmid was derived from pJL638 (30). The GAL1 promoter in pJL638 was replaced by the TUB2 core promoter (21) and a fragment containing five GAL4 DNA binding sites from G5BCAT (33). The resulting plasmid, pTM592G5, was linearized, integrated at the URA3 locus, and used as the lacZ reporter in the transcription activation assay. Plasmid pJR1425 is a derivative of pRS316 containing the α mating information downstream of hmrΔE (14).

Plasmid stability and β-galactosidase assays.

The plasmid stability and β-galactosidase assays for GAL4 DBD fusion proteins were described previously (32). For the full-length ABF1 plasmid loss assay shown in Fig. 3, the plasmid loss rate was measured and calculated as described previously (10) with the following minor changes. Strains were grown in synthetic complete medium (SCM) lacking uracil, inoculated to YPAD medium (1), and grown for an additional 8 to 12 generations. Dilutions of cells before and after incubation in the YPAD medium were plated both on SCM lacking uracil and on YPAD. The whole procedure was carried out at 30°C.

FIG. 3.

CS2 plays an important role in ABF1-dependent activation of plasmid DNA replication. Plasmid loss rate was measured in cells that expressed either wild-type or deletion mutants of ABF1 at 30°C. Two test plasmids, pARS1 (black bar) and pARS1-B3 (grey bar), were used in this assay. pARS1 contains the wild-type ARS1, whereas pARS1-B3 contains two point mutations in the B3 element which abolish ABF1 binding to ARS1. Also shown on the top is an illustration of the cis- and trans-acting elements involved in ARS1 function. Shown on the bottom is the presence (+) or absence (−) of CS1 and CS2 in the corresponding ABF1 constructs. ORC, origin recognition complex.

MNase assay.

The micrococcal nuclease (MNase) assay was performed as previously described (25) with the following modification: the lysed spheroplasts were treated with 25 U of MNase/ml for 2.5 and 8 min, and the probe used was a 420-bp EcoRI-NheI fragment 3′ to the ARS1 sequence.

2-D gel electrophoresis.

This assay was performed as described previously (32), except that yeast cultures were grown in SCM lacking leucine, with 50 μM copper sulfate.

Quantitative mating assay.

The assay was conducted essentially as previously described (36), with the following minor changes: strain JRY4997 was sequentially transformed with plasmids pJR1425 (36) and pRS315-ABF1 (or its derivatives); cells were grown to log phase (∼2.0 × 10⁷ cells/ml) in SCM lacking uracil and leucine; and following mating, dilutions of cell suspensions were plated on synthetic dextrose to select for diploids and dilutions of the tested strain were plated on SCM lacking uracil and leucine. Mating efficiency was calculated as described earlier (36).

HMR targeting assay.

The pSB362-based GAL4 DBD-ABF1 fusion constructs were introduced into strains YSB35 and YSB67 (Table 1). Cells were grown overnight in SCM lacking histidine. Starting with approximately 2.0 × 10⁷ cells/ml, 10-fold serial dilutions were spotted onto SCM lacking histidine, with or without tryptophan. Plate contents were incubated at 30°C. Pictures were taken 24 to 48 h after plating.

RESULTS

The carboxyl terminus of ABF1 contains two clusters of amino acids that are important for cell viability.

Previous studies have indicated that the ABF1 C-terminal trans-activation domain is essential for cell growth (Fig. 1A) (8, 32, 50, 65). However, the exact boundary and amino acid residues for supporting cell viability remain to be defined. Even less is known about the possible partitioning of this region into multiple functional domains. As an initial step to address these issues, we used a plasmid-shuffling method to assess the ability of a series of deletion mutants to support cell viability (Fig. 1). The expression levels of the mutant proteins were comparable to that of the full-length protein (data not shown). Based upon their behaviors in the viability assay, these mutants are categorized into three groups (Fig. 1A).

FIG. 1.

Ability of various ABF1 mutants to support cell viability. (A) The upper panel is a schematic diagram of the ABF1 protein. The lower panel is a summary of the growth profile of the cells harboring various ABF1 deletion mutants and two previously reported mutants abf1-1(C49Y) and abf1-5(P357L) (50). At each temperature tested, the growth rates were compared with those of the wild-type strain. The mutants scored as +++ grew at the same rate as the wild type. Those scored as ++ formed smaller-than-wild-type colonies that appeared on the same day as the wild type. Those scored as + gave rise to colonies that appeared 1 day later than the wild type. The mutants scored as +/− grew very slowly and appeared only after prolonged incubation. The minus sign indicates no detectable growth even after long incubation. The mutant strain labeled lethal could not be retrieved from the 5-fluoroorotic acid plate at the plasmid-shuffling step. (B) Representative plates of the wild-type and several mutant abf1 yeast cells upon incubation for 2 or 3 days at the indicated temperatures.

Cells expressing ABF1 (1-662) displayed the same growth rate as those expressing the full-length protein (aa 1 to 731) at all temperatures tested (group 1 in Fig. 1A; also see Fig. 1B). This suggests that the last 69 aa of the protein are dispensable for cell growth, even though they include one of the highly conserved regions (aa 717 to 731) among ABF1 homologues from three yeast species (18, 43). Further deletion of the region between aa 592 and 643 resulted in a temperature-sensitive phenotype (group 2 in Fig. 1A; also see 1-643 and 1-607 in Fig. 1B). The temperature-sensitive growth phenotype of group 2 is similar to that of abf1-1(C49Y) and abf1-5(P357L), two previously characterized mutants that affect ABF1 DNA binding (50) (also see Fig. 1A and B). It is also worth noting that the sequences truncated in the second group contain another highly conserved region of ABF1 (aa 644 to 662). The third group of mutants, which had a deletion of additional regions upstream of aa 592, either failed to support cell growth (1-587, 1-571, 1-563, and 1-521) or exhibited a very severe growth defect (1-582 and 1-577). In fact, cells expressing ABF1 (1-582) or ABF1 (1-577) only grew into barely visible colonies after 4 days of incubation at room temperature. Taken together, this result suggests that the region between aa 571 and 662 includes amino acids that are critical for ABF1 function in vivo.

To determine the critical amino acid residues in this part of ABF1, we performed a saturated alanine-scanning mutagenesis study of the region between aa 568 and 662, in which case five consecutive amino acids were replaced at a time with alanine residues (Fig. 2). The effect of these mutations on cell viability was tested in the context of ABF1 (1-662). All mutants were expressed at levels similar to the wild-type protein (data not shown). Consistent with the deletional study, several alanine substitution mutations in this region resulted in severe growth defects (Fig. 2). All growth-defective mutations were clustered in two confined areas downstream of aa 623 (aa 624 to 628 and 639 to 662), which were designated CS1 and CS2, respectively. Intriguingly, two mutants within CS1 and CS2 regions failed to support cell viability at all (A624-628 and A644-648). The possible causes for the lethal phenotype are discussed later.

FIG. 2.

Growth profile of the cells carrying alanine substitution mutations across the C-terminal ABF1 (aa 568 to 662). All alanine mutations were tested in the context of ABF1 (1-662). Two regions that are critical for ABF1 function in supporting cell growth are designated CS1 and CS2. Growth rates were scored in the same way as for Fig. 1A.

Role of CS1 and CS2 in activation of DNA replication in a plasmid context.

Previous work has shown that the region between aa 608 and 731 can act as a trans-activation domain in replication and transcription assays (32). Coupled with the results from the viability assay as shown above, we reasoned that CS1 and CS2 might play important roles in ABF1-facilitated nuclear functions. To test this possibility, we first examined the impact of these two regions on ABF1 function in activation of DNA replication. As illustrated in the diagram in Fig. 3, ABF1 binds to the B3 element of ARS1 and stimulates initiation of DNA replication in both plasmid and chromosomal contexts (37, 38). Using an established plasmid stability assay, we tested the ability of various ABF1 constructs to stimulate DNA replication of an ARS1-containing test plasmid (Fig. 3, pARS1). To distinguish a direct effect on DNA replication from an indirect one, we also included a test plasmid in which the ABF1 binding site (B3) was abolished (pARS1-B3) (38). An ABF1 mutation that directly affects its role in DNA replication is expected to behave in a B3-dependent manner.

As shown in Fig. 3, the plasmid loss rates for pARS1 and pARS1-B3 in the wild-type strain (1-731) were 7.9 and 20% per generation, respectively (Fig. 3, columns 1 and 2), consistent with the known function of B3 in enhancing ARS1 activity. ABF1 (1-662) was still fully capable of supporting ARS function (Fig. 3, columns 3 and 4), whereas further deletion of sequences upstream of aa 662 significantly increased the plasmid loss rate for pARS1 (Fig. 3, columns 5, 7, and 9). Furthermore, the plasmid loss rates of pARS1 and pARS1-B3 in the ABF1 (1-607) background were very similar (Fig. 3, compare columns 9 and 10). These data suggest that the most critical amino acid residues for stimulating DNA replication reside between aa 607 and 662. Partial or complete deletion of CS2, as in the case of ABF1 (1-643) and (1-638), doubled the plasmid loss rate of pARS1 (compare column 1 with columns 5 and 7 in Fig. 3). However, further deletion of CS1 did not exacerbate the plasmid loss (compare columns 7 and 9). This result indicates that CS2 but not CS1 is required for ABF1 function in activation of DNA replication. Compared to that for pARS1, the plasmid loss rate of pARS1-B3 was not significantly affected by any of the deletion mutants (Fig. 3, compare column 2 with columns 6, 8, and 10). This implies that the deletional effect as detected in this assay is largely caused by abrogation of a direct role of ABF1 at the replication origin.

The lethal phenotype associated with the CS1 and CS2 alanine substitution mutations (Fig. 2) precluded us from testing their effects on function of the native ARS1. To circumvent this problem, we fused the GAL4 DBD with the C-terminal region of ABF1. All chimeric proteins were expressed at a similar level (Fig. 4A; data not shown). The ability of these GAL4 fusions in stimulating DNA replication was assessed in a plasmid stability assay using a modified ARS1-containing plasmid, which contains five GAL4 binding sites and crippled B2 and B3 boxes. In this assay, plasmid stability was defined as the percentage of cells that still retained the tester plasmid following 14 generations of growth in nonselective growth (32, 38). Previous work has demonstrated that replication efficiency of this test plasmid is stimulated by a variety of GAL4-derived transcriptional activators (31, 32). As shown in Fig. 4B, ABF1 (604-662) conferred the same degree of stimulation of plasmid stability as the entire C terminus (aa 604 to 731; compare the solid bars in Fig. 4B, columns 2 and 3). This observation corroborates the finding shown in Fig. 3 that the last 69 aa residues of the protein are dispensable for activation of DNA replication.

FIG. 4.

Activation of transcription and replication by GAL4-ABF1 fusion proteins. (A) HA-tagged GAL4 fusion proteins were detected by immunoblotting using an anti-HA monoclonal antibody (12CA5). Equal amounts of cell extract were loaded. (B) Plasmid stability (solid bar) and β-galactosidase activity (hatched bar) were measured in the strains that expressed various GAL4-ABF1 fusions. The tester plasmid used in the plasmid stability assay contains a centromere (CEN), the URA3 marker, and the modified GAL4-responsive ARS1. Cells harboring the tester plasmid were inoculated into uracil-containing medium and grown for 14 generations. The plasmid stability value is defined as the percentage of cells that retain the tester plasmid following nonselective growth. For the transcription assay, five GAL4 binding sites with the TUB2 core promoter were placed in front of the lacZ gene and the entire cassette was integrated at the URA3 locus. Also shown are the diagrams of the modified origin of replication and the transcription promoter. All alanine-scanning mutations were tested in the context of GAL4-ABF1 (604-662).

The effect of the alanine-scanning mutations on DNA replication was examined in the context of GAL4-ABF1 (604-662). As shown in Fig. 4B, the alanine mutation at the center of CS2 (A644-648 in column 13) completely abolished the ability of the fusion protein to enhance origin function. On the other hand, the CS1 mutation (A624-628) did not display any significant effects on plasmid stability. Furthermore, a small fragment containing CS2 alone (633-662) was capable of activating plasmid replication (Fig. 4B, column 5), albeit to a lesser extent than the entire ABF1 C terminus. In contrast, a fragment that contains CS1 alone fails to confer any stimulation of ARS1 function (Fig. 4B, column 4). Thus, these results strongly suggest that CS2 is both sufficient and necessary for stimulating ARS function in a plasmid context. CS1, on the other hand, is not required for ABF1 function in DNA replication.

CS2 is sufficient for activation of chromosomal replication.

To confirm the role of CS2 in stimulating initiation of DNA replication, we made use of a yeast strain in which the native ARS1 on chromosome IV was replaced with a modified, GAL4-responsive ARS1. By use of a neutral/neutral 2-D gel electrophoresis method (2), it has previously been shown that initiation of chromosomal replication from this modified ARS1 can be stimulated by a number of GAL4-derived activators (32). As indicated by the arrows in Fig. 5, the “bubble” arcs represent the replication intermediates initiated from the modified replication origin. On the other hand, the “fork” arcs shown at the lower part of each panel mainly arise from initiation at replication origins outside the genomic fragment. The intensity of the bubble arcs relative to the fork arcs was enhanced by GAL4-ABF1 (604-662) (Fig. 5, compare panels 1 and 2). A shorter GAL4 fusion that contained only the CS2 region (633-662) was also capable of enhancing the origin function (Fig. 5, panel 4), albeit somewhat less robustly than the longer one (Fig. 5, panel 2). In contrast, GAL4 DBD fused with CS1 alone (604-633) did not exhibit any stimulatory effects on initiation of replication. As a control, none of the GAL4 derivatives affect the origin usage of ARS501, an origin of replication that does not contain any GAL4 binding sites (data not shown). Thus, CS2 is capable of stimulating DNA replication from both plasmid-borne and chromosome-embedded origins of replication.

FIG. 5.

ABF1 (633-662) contains a minimal activation domain for chromosomal replication. The chromosome-embedded ARS1 at the native locus contains five GAL4 binding sites and crippled B1, B2, and B3 elements. Genomic DNA isolated from exponentially growing cells was digested with NcoI and subjected to 2-D gel electrophoresis. Arrows mark the “bubble” arcs, which represent replication intermediates initiated from the modified ARS1. The fusion proteins are indicated on the top of the panels.

Role of CS1 and CS2 in transcriptional activation.

Previous studies suggest that the ABF1 C terminus also contains a transcriptional activation domain (32, 60). However, the lethality associated with the CS1 and CS2 alanine-scanning mutations made it technically difficult to assess their effects on the expression of endogenous ABF1-regulated genes. To circumvent this problem, we examined the ability of the GAL4-ABF1 fusion proteins carrying the same mutations to activate transcription from a synthetic GAL4-responsive promoter. The hatched bars in Fig. 4B represent the β-galactosidase activity as a result of activation of transcription by the GAL4 derivatives. The transcriptional activity of most of the GAL4-ABF1 fusion proteins mirrors that in DNA replication. For instance, ABF1 (604-662) and ABF1 (633-662) are capable of stimulating both processes (Fig. 4B, columns 3 and 5). Moreover, the alanine substitution mutation in CS2 (A644-648) abrogated ABF1 activity in both assays (Fig. 4B, column 13), suggesting that the same amino acid residues in CS2 are involved in both ABF1-mediated activation of transcription and DNA replication.

While activation of transcription and replication correlated well for most of the GAL4-ABF1 fusion proteins, the alanine substitution mutant at CS1 (A624-628) exhibited differential activities in the two functional assays. This mutation had no effect on ABF1 function in DNA replication, but it consistently resulted in an elevated level of activation of transcription (compare columns 3 and 9 in Fig. 4B). This result raises the possibility that CS1 may be involved in negatively regulating ABF1 function in transcriptional activation.

Localization of the silencing domain of ABF1 to the C-terminal region of the protein.

In addition to its function in activation of replication and transcription, ABF1 is also involved in gene silencing at both mating-type loci (HML and HMR) (29, 36) and the subtelomeric regions (46). However, the structural requirement for ABF1 function in silencing remains to be determined. To address this important issue, we focused our effort on the extensively characterized silencer HMR-E. HMR-E consists of three cis-acting elements, A, E, and B, which serve as the binding sites for the origin recognition complex, RAP1, and ABF1, respectively. The three elements are functionally redundant, in that two out of three are sufficient for the full silencing activity.

To localize the functional domain(s) of ABF1 that are involved in gene silencing, we first made use of a yeast strain (JRY4997) that bears a particular mutation in ABF1 (abf1-102) and the MATa locus (mata1) (36). This strain also carries a plasmid (pJR1425) that contains the α genes placed downstream of a crippled HMR-E element (HMRα-E-RAP1-10) (diagram in Fig. 6A). In the abf1-102 background, function of this particular silencer was compromised and the plasmid-borne α genes were expressed, enabling the parental strain to mate a-type cells. In the present study, various ABF1 expression vectors were introduced into JRY4997 with pJR1425 and their ability to complement the silencing defect of abf1-102 was assessed. An ABF1 construct that rescued the silencing defect of the parental strain would render it less competent for mating with a-type cells, due to lower expression of the α genes.

FIG. 6.

ABF1 aa 604 to 731 are necessary and sufficient for silencing function at HMR-E. (A) Complementation of the abf1 silencing-deficient mutant (abf1-102) by the wild-type ABF1 and various mutants. The host strain contains the mata1 and abf1-102 mutations and a plasmid (pJR1425) that has the α locus information downstream of HMR-E-RAP1-10. A schematic diagram of the silencer is presented on the top. ORC, origin recognition complex. (B) Targeted-silencing assay. The ABF1 C-terminal regions are tethered by GAL4-DBD to the modified HMR-E element, which lacks both E and B elements (panels a and b). As a control, the same GAL4-ABF1 fusions were also tested for their effects on a silencer that did not contain GAL4 binding sites (panels c and d). Panels a and c show cell growth in the absence of tryptophan (TRP), whereas the control panels (panels b and d) are from tryptophan-containing minimal medium.

As expected, the full-length, wild-type ABF1 expression vector (aa 1 to 731) complemented the silencing defect of the abf1-102 mutant and thus suppressed the mating capability of the parental strain (Fig. 6A, compare columns 1 and 2). ABF1 (1-662) but not ABF1 (1-607) was capable of rescuing the silencing function of HMR-E (columns 3 and 4), suggesting that the critical region for mating-type silencing resides between aa 608 and 662 of the protein. Interestingly, mutations at CS1 (A624-628) and those at CS2 (A644-648), to a lesser extent, impaired the silencing function of the plasmid-borne HMR-E, as indicated by the relatively high mating efficiency associated with these mutants (Fig. 6A, columns 5 and 6).

The complementation assay described above provides important clues into the role of the C terminus of ABF1 in gene silencing. However, interpretation of the results was complicated by the ascription of multiple functions to the same region of the protein. In particular, the involvement of CS1 and CS2 in ABF1-mediated gene activation made it difficult to distinguish a direct from an indirect effect of these regions on gene silencing in the context of the native HMR-E. As an alternative approach, we employed the targeted-silencing assay developed by Buck and Shore (5). In this assay, the chromosomal HMR-E underwent deletion of both E and B elements and three GAL4 binding sites were engineered adjacent to the A element (Fig. 6B, diagram). In addition, the TRP1 gene was integrated at the HMR locus in place of the normal a1 and a2 genes. Direct recruitment of known silencing proteins (i.e., RAP1 and SIR1) to the GAL4 binding sites via the GAL4 DBD resulted in silencing of the TRP1 gene, as indicated by impaired cell growth in synthetic medium lacking tryptophan (5, 7). Consistent with previous work, we showed that a GAL4 fusion protein containing the C terminus of RAP1 strongly silenced gene expression in this assay (Fig. 6B, compare rows 1 and 2 in panel a).

As shown in Fig. 6B, cells that expressed GAL4-ABF1 (604-731) grew much more slowly in the absence of tryptophan than did those with the GAL4 DBD (compare rows 1 and 3 in Fig. 6B, panel a), yet the same yeast cells grew at a comparable rate in the tryptophan-containing medium (Fig. 6B, panel b). Furthermore, the ABF1-mediated gene silencing occurred in a GAL4 binding-dependent manner, as the same fusion proteins did not affect expression of the TRP1 gene in the absence of the GAL4 binding sites (Fig. 6B, panels c and d). GAL4-ABF1 (604-662) still retained some silencing activity (Fig. 6B, panel a, row 6), whereas GAL4-ABF1 (663-731) failed to confer any silencing function (Fig. 6B, panel a, row 7). These data suggest that the region between aa 604 and 662 contains a minimal silencing domain, the function of which may be further aided by additional sequences in aa 663 to 731.

Consistent with the result from the silencing complementation assay (Fig. 6A), the alanine substitution mutation in CS2 (A644-648) also compromised the silencing function of the GAL4 derivative (Fig. 6B, panel a, row 5). This strongly suggests that CS2 is not only required for ABF1’s function in activation of transcription and replication but also for its activity in gene silencing. Interestingly, the CS1 mutation (A624-628), which abolished ABF1 silencing activity in the complementation assay, did not affect the ability of the GAL4-ABF1 fusion protein to silence gene expression. We interpret this result as an indication that the role of CS1 in silencing may be indirect or context dependent.

CS2 is directly involved in ABF1-mediated chromatin remodeling in vivo.

Previous work (25, 28, 62) has demonstrated that ABF1 induces changes in chromatin structure upon binding to yeast chromosomes. To establish a firm link between chromatin remodeling and the ABF1-mediated regulation of multiple chromosomal events, we employed an indirect end-labeling MNase digestion assay (26) to examine the impact of GAL4 derivatives on the chromatin structure around the GAL4-responsive chromosomal ARS1. Previous studies have shown that although all GAL4 derivatives can get access to the chromosome-embedded GAL4 binding sites, only those that activate transcription and chromosomal replication induce distinct changes in the pattern of nuclease digestion (25, 31). Furthermore, GAL4 derivative-induced chromatin remodeling occurs even in the absence of DNA replication, suggesting that chromatin remodeling is a cause, rather than an effect of DNA replication (25).

Compared with the GAL4 DBD alone, GAL4-ABF1 (604-662) induced distinct changes in the nuclease digestion pattern around the GAL4 binding sites (Fig. 7A, compare lanes 1 and 2 with lanes 3 and 4). In particular, the intensity of a hypersensitive site (A) was attenuated, whereas that of a neighboring site (B) was enhanced in the presence of the GAL4-derived activator (also see quantitation in Fig. 7B). This result is consistent with the behavior of other GAL4 derivatives that activate transcription and DNA replication (25, 31, 32). The alanine substitution mutant at CS1 (A624-628) was still capable of chromatin remodeling, as indicated by the B-to-A ratio similar to that for the wild-type fusion protein (Fig. 7A and B, compare lanes 3 and 4 with lanes 5 and 6). On the other hand, the mutant at CS2 (A644-648) abolished the ability of the GAL4-ABF1 fusion protein to induce chromatin remodeling, as indicated by a significantly reduced B-to-A ratio (Fig. 7A and B, lanes 7 and 8 ). These results suggest that the two functionally important regions of ABF1 have different effects on chromatin structure. In light of the essential role of CS2 in activation of transcription, replication, and gene silencing, CS2-mediated chromatin remodeling may be an important mechanism used by ABF1 to stimulate multiple nuclear processes.

FIG. 7.

CS2 is responsible for chromatin remodeling in vivo. (A) An indirect end-labeling MNase assay was carried out to assess the nucleosome organization around the modified chromosomal ARS1. Isolated nuclei were treated with MNase for various lengths of time and were digested with EcoRI to completion. The arrows (labeled A and B) indicate the two nuclease-cutting sites, the intensity of which was most significantly affected by the GAL4-ABF1 fusion proteins. The organization of the modified ARS and the probe used in the indirect end-labeling assay are indicated on the left. (B) The intensity of bands B and A was measured by densitometry. The ratio of B to A is presented. The number of each column corresponds to that in panel A. The data shown here are representative of at least four independent experiments.

DISCUSSION

In addition to their role in transcriptional activation, many site-specific transcription factors also have the potential to facilitate other nuclear processes, including DNA replication, repair, recombination, and gene silencing. However, the molecular basis for the multifunctional property of transcription factors remains to be understood. ABF1 represents an excellent model for studying the common underlying mechanisms involved in regulation of diverse chromosomal events. Previous work has demonstrated that the C-terminal region of ABF1 contains the activation domain for both transcription and DNA replication (32). In the present study, we performed a comparative study of the critical amino acid residues required for ABF1 functions in supporting cell growth, facilitating various nuclear functions, and remodeling chromatin. Our work led to the localization of the silencing domain to the C terminus of the protein. In addition, the same amino acid residues that are required for activation of transcription and chromosomal DNA replication are also critical for the gene silencing function. Lastly, the data also suggests that ABF1’s function in multiple nuclear processes is intimately linked to its impact on chromatin structure.

Several previous studies have assessed the functional importance of the C-terminal region of ABF1 (19, 32, 50, 60, 65). Our data, for the most part, are consistent with previous findings concerning the role of this region in ABF1-mediated activation of transcription and replication. However, in a previous study by Wiltshire et al. (65), it was found that a C-terminal fragment of ABF1 (aa 635 to 684) that activated DNA replication lacked the ability to stimulate transcription from a GAL1-derived promoter. This led to the conclusion that the transcription function of ABF1 could be separated from its role in DNA replication. In contrast, our study shows that ABF1 (633-662), which is included in the original fragment used by Wiltshire et al., is capable of stimulating both ARS1 function and transcription from a TUB2-derived core promoter (Fig. 4B and 5). The discrepancy between the two studies is most likely due to the promoter-dependent nature of ABF1 activity in transcription (39, 60). Indeed, we found that the TUB2-derived core promoter was much more responsive to GAL4-ABF1 than was a GAL1-derived core promoter (T. Miyake, C. M. Loch, and R. Li, unpublished data). It is also worth noting that unlike the native GAL1 promoter, the native TUB2 promoter contains an ABF1 binding site and its transcriptional activity is regulated by ABF1 (4, 21).

It is puzzling that strains that carry several of the C-terminal deletion mutants are sick but still viable (Fig. 1), whereas two of the alanine substitution mutations in CS1 and CS2 result in lethality (Fig. 2). The lethal phenotype is unlikely due to gross misfolding of the mutant proteins, as the CS1 mutant (A624-628) is still capable of activating transcription and replication and remodeling chromatin. To reconcile the data from the deletional and alanine substitution studies, we postulate that the region upstream of CS1 may contain a cryptic domain for cell viability, the function of which may only manifest in the absence of the entire C-terminal region. Alternatively, one could imagine that a proper distribution of ABF1 and its coactivators among the various ABF1-mediated nuclear events is critical for cell survival. In the case of the alanine mutations at CS1 or CS2, perhaps ABF1 and its cofactors are misallocated away from those events that are essential for viability, leading to the lethality observed. A third possibility is that the alanine substitution mutants of CS1 or CS2 may fortuitously sequester a factor(s) that is essential for cell growth. However, a shuffling strain (TMY86) that contains one wild-type ABF1 allele and one CS1 or CS2 mutant allele grows normally, arguing that the mutants do not act in a dominant-negative fashion (Loch, Miyake, and Li, unpublished data). Lastly, given our finding of the enhanced ability of the CS1 mutant in transcriptional activation, it also seems possible that the CS1 mutation may result in misexpression of certain ABF1-regulated genes, which may be more detrimental to cell growth than deletion of the entire C-terminal region. Consistent with the last possibility, it is known that overexpression of ABF1 in yeast cells strongly inhibits normal cell growth (47, 50) (Loch et al., unpublished). Additional work is needed to determine the exact cause(s) for the growth phenotype associated with the different types of mutations.

While the mutational study indicates that CS1 is important for ABF1 function, the exact role of this region remains to be elucidated. Unlike CS2, CS1 does not appear to play a universal role in ABF1-mediated nuclear processes. For example, the CS1 mutation did not affect ABF1 function in DNA replication, and unexpectedly, the same mutation enhanced the ability of the C terminus of ABF1 to stimulate transcription. The contribution of this region to ABF1 function in silencing also seems to be less certain than that of CS2. On one hand, ABF1 (1-662) with the alanine substitution mutation at CS1 (A624-628) was unable to complement the silencing defect associated with abf1-102 (Fig. 6A). However, in the GAL4 targeted-silencing assay (Fig. 6B), the same mutation did not significantly reduce the silencing efficiency of GAL4-ABF1. Thus, CS1’s role in gene silencing at the mating-type locus appears to be more indirect or context-dependent than CS2’s. It is possible that the effect of CS1 on silencing may be due to its impact on gene expression. Alternatively, CS1 could be required for the chromatin association of the full-length Abf1 at the silencer locus. In such a case, direct tethering of the C-terminal domain of Abf1 via the GAL4 DBD to the silencer may bypass the need for CS1 in gene silencing.

Several lines of evidence suggest that ABF1 (633-662) contains a functional domain (CS2) that is pivotal to all three ABF1 functions examined. First, both deletion and alanine-scanning mutations that affected the integrity of this region severely impaired ABF1 functions in cell viability, gene activation, gene silencing, and DNA replication. Second, the same mutations also abolished ABF1-mediated changes in chromatin structure. Third, when tethered to the appropriate chromosomal sites, this region of the protein was sufficient to activate transcription and DNA replication. Localization of the essential ABF1 functional domain to the relatively small region of the protein will facilitate future work to isolate the putative partner(s) that interacts with CS2 and mediates its activity.

Consistent with its importance to ABF1 functions, the amino acid sequence encompassing CS2 is highly conserved among ABF1 proteins from three different yeast species (Fig. 8A, Sc-, Kl-, and KmABF1). Interestingly, this region of ABF1 also shares limited sequence homology with the C-terminal region of RAP1 (RAP1, aa 666 to 678) (13). In particular, the alanine substitution mutation in ABF1 with the most severe phenotype (A644-648) changed the largest block of conserved residues between ABF1 and RAP1 (amino acids with asterisks in Fig. 8A), further underscoring the functional significance of the conserved amino acid sequence. Previous mutational studies of RAP1 have defined the silencing domain (aa 667 to 827) (5) and trans-activation domain (aa 630 to 692) (22). Interestingly, the conserved amino acid residues between ABF1 and RAP1 (aa 666 to 678) fall into the overlapping region between the trans-activation and silencing domains of RAP1. These findings raise the intriguing possibility that this conserved region in both proteins may employ a similar mechanism to facilitate silencing and transcriptional activation.

FIG. 8.

Structural feature of CS2. (A) Alignment of the homologous sequences of the ABF1 CS2 regions from S. cerevisiae (Sc), Kluyveromyces lactis (Kl), Kluyveromyces marxianus (Km), and ScRAP1. The numbers shown at the top indicate the amino acid positions of ScABF1. Asterisks represent those conserved amino acids that were mutated in the alanine substitution mutant A644-648. (B) Helix wheel model of ABF1 (644-661). This region is predicted to form an α-helix by several secondary structure prediction programs. The helix wheel model of this region reveals the tripartite α-helix marked as hydrophobic (black), acidic (hatched), and basic (grey) regions. The homologous amino acids between ScABF1 and ScRAP1 are shaded. The acidic and basic residues are marked by single and double underlines, respectively.

It is interesting that the same amino acid residues in CS2 are important for gene activation and gene silencing, two opposing events that presumably invoke opposite changes in chromatin structure. The CS2 region (aa 644 to 662) is predicted to form an α-helical structure. As illustrated in Fig. 8B, the helical wheel of this region reveals distinct patches of hydrophobic, acidic, and basic amino acid residues that reside on different surfaces of the wheel, which is similar, but not identical to the amino acid arrangement seen in the amphipathic α-helical structure of a typical acidic trans-activation domain (17). The homologous region in RAP1 shares the same structural feature with ABF1, and furthermore, most of the amino acid residues in the three patches are conserved between the two proteins (highlighted in Fig. 8B).

CS2 may interact with distinct partners to facilitate two nuclear processes as diverse as gene activation and gene silencing. Alternatively, CS2 may recruit one universal partner to facilitate multiple chromosomal events. In this scenario, CS2 may not act as a bona fide activation or repression domain. Rather, by recruiting a putative chromatin-remodeling machine, CS2 could simply act as an inducer of chromatin reorganization. An increase in the fluidity of the nucleosome structure may create a window of opportunity for other trans-acting factors to reset the chromatin state. According to this model, ABF1 at the silenced mating-type loci could help establish the silenced chromatin state by allowing easier loading of the silencing proteins (e.g., SIR1 to -4) to the silencer. On the other hand, ABF1 bound at a transcriptional promoter or an origin of replication may increase chromatin accessibility to the transcription or replication machineries. In other words, the outcome of the ABF1-mediated chromatin remodeling, i.e., repressive versus active chromatin state, may rest upon the additional factors that act in concert with ABF1 at the same chromosomal locus. Consistent with this notion, the SWI/SNF chromatin remodeling complexes have been implicated in both activation and repression of gene expression (24, 57, 58). Future work will be aimed at identifying the exact chromatin remodeling complex(es) that is recruited by CS2 to facilitate multiple nuclear functions.