ABSTRACT
Although it has been demonstrated that eukaryotic cellular origins of DNA replication may harbor stimulatory elements that bind transcription factors, how these factors stimulate origin function is unknown. In Saccharomyces cerevisiae, the transcription factor Abf1p stimulates origin function of ARS121 and ARS1. In the results presented here, an analysis of Abf1p function has been carried out utilizing LexA(BD)-Abf1p fusion proteins and an ARS121 derivative harboring LexA DNA-binding sites. A minimal region which stimulates origin function mapped to 50 amino acids within the C-terminus of Abf1p. When tested for transcriptional activation of a LacZ reporter gene, the same LexA(BD)-Abf1p fusion protein had negligible transcriptional activation potential. Therefore, stimulation of ARS121 may occur independently of a transcriptional activation domain. It has been previously observed that the Gal4p, Rap1p DNA-binding sites and the LexA-Gal4p fusion protein can replace the role of Abf1p in stimulating ARS1. Here we show that the stimulatory function of Abf1p at ARS121 cannot be replaced by these alternative DNA-binding sites and the potent chimeric transcriptional activator LexA(BD)-Gal4(AD)p. Hence, these results strongly suggest that the Abf1p stimulation of replication may differ for ARS121 and ARS1, and imply specificity in the Abf1p/ARS121 relationship.
Analyses of eukaryotic nuclear origins of DNA replication have been greatly facilitated by the biochemically and genetically tractable yeast, Saccharomyces cerevisiae. DNA sequences of S.cerevisiae were identified which enabled the extrachromosomal replication of plasmids (1 ). These sequences, known as ARSs (autonomously replicating sequences), act as genuine origins of replication (2 ,3 ). While a subset of ARSs have been demonstrated to function as origins in their native chromosomal location, all chromosomal origins identified to date correspond to ARSs (for a review, see ref. 4 ).
Detailed characterization of a number of ARSs has identified individual elements which together elicit full activity (4 ). The only highly conserved sequence that is essential for function in ARSs is the 11 bp [(A/T)TTTAT(A/G)TTT(A/T)] ARS consensus sequence (ACS) (5 ,6 ). The ACS is part of a larger essential element, which varies in size depending on the ARS. Auxiliary elements are found on both sides of the essential element. Elements are often systematically referred to as residing in domain A, B or C (7 ). Domain A currently refers to the entire essential ACS-containing region as defined by linker substitutions (4 ). Domains B and C currently refer to the regions outside of domain A, 3' and 5' to the T-rich strand of the ACS, respectively (4 ).
It has been shown that the ARS essential sequences are specifically recognized by ATP-dependent DNA-binding proteins (8 ,9 ). Genetic evidence has implicated one of these, the ORC, in the initiation of DNA replication (10 ,11 ) as well as in transcriptional silencing (12 ,13 ). Genetic and/or physical interactions with ORC have been identified for a number of factors which have also been implicated in the initiation of DNA replication (14 -19 ).
An auxiliary element is located at several ARSs is the DNA-binding site for Abf1p. Abf1p is an essential, multifunctional protein, with roles in DNA replication and in the transcriptional activation and repression of a number of diverse genes (for a review, see ref. 20 ). The function of Abf1p in replication has been examined in greatest detail at ARS121 and ARS1, each of which contain auxiliary Abf1p DNA-binding sites. At ARS121, a series of linker substitutions through the Abf1p DNA-binding sites were utilized to demonstrate a direct, strong correlation between in vitro binding of Abf1p to the ARS and in vivo ARS function (21 ). In addition, at the semi-permissive temperature in abf1(ts) strains, ARS121 function was severely reduced (22 ). Linker substitution analyses at ARS1 also demonstrated that elimination of Abf1p binding in vitro abolished stimulation of ARS1 by the Abf1p DNA-binding site (23 ).
How Abf1p stimulates ARS function is unknown. It has been shown that transcriptional activators can substitute for Abf1p function at the ARS1 origin (23 ). Transcriptional activators are known to stimulate replication for a number of eukaryotic viruses. Studies in viral systems have indicated that auxiliary factors which stimulate DNA replication can do so in a number of ways (for reviews, see refs 24 ,25 ). For example, stimulation of bovine papilloma virus (BPV) DNA replication in vitro was mediated by the interaction of a transcriptional activation domain with the replication protein A (RPA) (26 ). Another mode of action, the alleviation of chromatin repression, required that the DNA-binding site of the auxiliary factor be located adjacent to essential core sequences (24 ,25 ,27 ). However, Abf1p DNA-binding sites can function in an orientation and distance independent manner as replication enhancers (28 ). Therefore, a role for Abf1p in overcoming chromatin repression may be less likely.
In this paper we report the results of experiments carried out to address Abf1p's role in replication. We have assessed the ability of LexA(BD)-Abf1p fusion proteins to stimulate ARS121. We show that a 50 amino acid region within the C-terminus of Abf1p is capable of stimulating ARS121. This LexA(BD)-Abf1p fusion protein lacks significant transcriptional activation potential. In addition, alternative factors, including known transcriptional activators, are unable to substitute for Abf1p at ARS121. These results indicate that ARS stimulation may occur independently of a transcriptional activation domain, and that ARS stimulation exhibits specificity at the ARS121 replication origin.
Saccharomyces cerevisiae strain FY251 (MATa, his3[Delta]200, leu2[Delta]1, trp1[Delta]63, ura3-52; a gift of F. Winston, Harvard) was used. Chromosomal integrations at the BAR1 and ura3 loci of FY251 were carried out as described below. Selective media was 0.17% yeast nitrogen base/0.5% ammonium sulfate, supplemented with uracil, histidine, leucine and tryptophan as required (29 ). Non-selective media was 1% yeast extract/2% bactopeptone containing galactose or glucose at 2% final concentration (29 ). Escherichia coli HB101 was used for plasmid preparations.
Plasmid pHLO (constructed in this lab) is a derivative of pOBF1 (30 ) and encodes a 6His-LexA(BD)-Abf1p fusion. The 6His is present to facilitate purification of certain LexA(BD)-Abf1p fusion proteins for studies not described in this manuscript. LexA(BD) refers to LexA residues 1-87 (31 ). To facilitate construction of deletions, an NdeI site is present at the LexA(BD)-Abf1 junction. In addition, an XbaI site and stop codons in all three reading frames immediately follow the Abf1 gene stop codon. LexA(BD)-Abf1 fusions were generated with convenient restriction enzyme sites or by cloning a PCR product generated with appropriate oligonucleotides. The Abf1p residues present in each fusion are indicated. The construct encoding LexA(BD)-Gal4(AD)p (activation domain) was made by first cloning the KpnI-Sau3AI fragment from pGAD424 (Clonetech) into the KpnI-BamHI sites of pUC19, and then moving an EcoRI(filled)-XbaI fragment from the resulting plasmid into the NdeI(filled) plus XbaI sites of pHLO. For the LexA(BD) construct, pHLO was digested with NdeI plus XbaI, end filled and re-circularized. Constructs were diagnosed by restriction analysis and verified by DNA sequencing.
All fusion proteins were expressed from a disrupted non-essential chromosomal locus. For chromosomal integration we utilized the plasmid pbar1::GAL1P (constructed in this lab), which is a derivative of pRS405 (32 ). This plasmid harbors (i) LEU2 for selection, (ii) a SalI cassette for directing gamma integration/disruption at the BAR1 locus, and (iii) an EcoRI-BamHI GAL1 promoter from pMH101 (30 ). When the LexA(BD)-Abf1 BglII fragment from pHLO is inserted in the correct orientation into the BamHI site of pbar1::GAL1P, the encoded protein is placed under the control of the GAL1 promoter, as previously described for ABF1 (30 ). For each pHLO derivative, the appropriate BglII fragment was moved into the BamHI site of pbar1::GAL1P. Plasmid with a monomer insert in the correct orientation was then prepared, and gamma integrations were carried out as previously described (32 ). Plasmid was linearized with HindIII, and the linear DNA was used to transform the yeast FY251. This transformation process targets integrative transformation at the BAR1 locus of FY251 with high efficiency. The BAR1 gene is non-essential and encodes a protease that degrades [alpha] factor (33 ). Purified transformants were screened by Southern analyses for correct integration. Westerns, with an anti-LexA serum (kindly provided by Erica Golemis, Fox Chase Cancer Center), indicated that each fusion was appropriately produced at comparable levels. The size of the fusion proteins produced was within the range known to diffuse into the nucleus (34 ,35 ).
The yeast centromeric plasmid YCpA121 harbors a truncated version of ARS121 known as A121 (21 ). The truncation removes the Abf1p DNA-binding sites present at ARS121, but leaves the essential core and the auxiliary A+T rich element (21 ,36 ). To test LexA(BD)-Abf1p fusions for their ability to stimulate origin function, we utilized YCpLEX/A121 (constructed in this lab) in which eight tandem LexA DNA-binding sites (37 ) are inserted into the KpnI site of YCpA121. YCpABF1/A121 [previously called YCpA121c(-), (28 )] harbors a synthetic Abf1p DNA-binding site inserted into the KpnI site of YCpA121. For substitution experiments, appropriate oligonucleotides were kinased, annealed and inserted into the KpnI site of YCpA121. YCpGAL4/A121, YCpRAP1/A121 and YCpREN1501/A121 harbor a Gal4p DNA-binding site (31 ), Rap1p DNA-binding site (38 ) and the REN1501 element (Raychaudhuri and Eisenberg, unpublished), respectively. Since these elements stimulate replication in an orientation independent manner (28 ,23 ; Raychaudhuri and Eisenberg, unpublished), each was analysed in a single orientation. The inserts, 5' to 3' along the same strand as the T-rich strand of the ARS121 essential core element, are as follows: GAL4 GGTACCTACGTACGGAAGACTCTCC-TCCGAGATCTGGT-ACC, RAP1 GGTACCTTTTTTCAAGGTTGATGGGTTTT-TGCAATATCGTACC, and REN1501 GGTACCTAGTATATTGCTATGTATGGTTTGCGTACC. The structure of ARS121 and the ARS121 derivatives described above are shown in Figure 1 .
For transcriptional activation assays of LexA fusions, we utilized the LacZ reporter borne on a pSH18-34 plasmid (39 ), or a derivative thereof. For qualitative assays we utilized pSH18-34, which harbors sequences from the 2µ plasmid for maintenance in yeast. For quantitative assays, we used a chromosomally integrated derivative of pSH18-34. The regulatory LexA DNA-binding sites present in this case are those of YCpLEX/A121. To allow for chromosomal integration, we removed the 2µ sequences, replacing the pSH18-34 SpeI fragment with an XbaI HIS3 cassette. Plasmid DNA was linearized with ApaI, and the linear DNA was used to transform the appropriate strains to Ura+, His+ autotrophy (40 ). Purified transformants were screened by Southern analyses to identify those with correct integration.
Mitotic stability assays (four to five independent transformants each) were performed as described elsewhere (21 ). With all fusion proteins, or when testing Gal4p, galactose was the carbon source instead of glucose. Mitotic stabilities are reported as the percentage of cells harboring the plasmid following ~14 generations of non-selective growth. Quantitative [beta]-galactosidase assays were carried out as described (39 ,41 ) utilizing extracts prepared from cultures grown in galactose-containing media. Extracts were prepared by breakage with glass beads. Time-course experiments verified linear [beta]-galactosidase activity beyond the time-point utilized for the assays. Activities are reported in Miller units (41 ) standardized to micrograms of protein and represent the average of three or four assays. Protein concentrations were determined by the Lowry method.
To address Abf1p stimulation of origins of DNA replication, we tested the ability of LexA(BD)-Abf1p fusions to stimulate an origin harboring LexA DNA-binding sites. Origin activity was tested by plasmid mitotic stability measurements. Yeast centromeric plasmids (YCp) contain an ARS element for replication and a CEN element for segregation. Since the CEN element promotes efficient plasmid segregation, the mitotic stability of a YCp is a commonly used indicator of the efficiency of the ARS element, the yeast origin of replication borne on the plasmid.
All fusions were constructed, integrated and expressed from the BAR1 chromosomal locus as described in Materials and Methods. All fusion proteins were expressed to comparable levels as demonstrated by western analysis using anti LexA antibodies (unpublished). The effect of these fusion proteins on replication was measured by the mitotic stabilities of YCpLEX/A121 and YCpA121 plasmids. These plasmids harbor A121, a version of ARS121 deleted from Abf1p DNA-binding sites (Fig. 1 ). In LEX/A121, eight tandem LexA DNA-binding sites were inserted adjacent to A121, as shown in Figure 1 . The ability of LexA-Abf1p fusions to stimulate ARS function resulted in elevated mitotic stability of YCpLEX/A121 relative to YCpA121.
As shown in Figure 2 , three LexA(BD)-Abf1p fusion proteins, bearing residues 1-323, 326-589 or 482-731, were used to survey the length of Abf1p for replication function. Of these three fusions, only LexA(BD)-Abf1(482-731)p, harboring the C-terminal 250 residues of Abf1p, stimulated the YCpLEX/A121 origin. The stimulatory effect was expressed as a ratio of the mean mitotic stabilities of the plasmids YCpLEX/A121 and YCpA121 (Fig. 2 ). Further analysis was then used to delimit the active region(s) more precisely (Fig. 2 , bottom). Truncations from the N-terminus beyond amino acid residue 635 abolished activity, as did truncations from the C-terminus beyond residue 684. Analysis of additional fusions indicated that LexA(BD)-Abf1(635-684)p was sufficient to stimulate ARS121. All fusions harboring residues 635-684 provided origin stimulation, and all truncations which removed this region abolished the stimulation of replication. Furthermore, an internal deletion of residues 635-684 from the LexA(BD)-Abf1(482-731)p fusion also abolished stimulation of replication. While not excluding the possible contribution of other regions of the protein, these results indicate that a 50 amino acid domain of Abf1p, residues 635-684, is capable of stimulating ARS function when tethered via a heterologous DNA-binding domain.
Abf1p is also a transcription factor, functioning at promoters of a diverse number of genes (for a review, see ref. 42 ). Abf1p DNA-binding sites act as moderately potent transcriptional activators (38 ,43 ) and the same is true for LexA(BD)-Abf1p fusion proteins harboring C-terminal regions of Abf1p (44 ; Wiltshire and Eisenberg, unpublished). Substitution experiments of alternative transcription factors for Abf1p at ARS1 suggested that origin stimulation by factors such as Abf1p required a transcriptional activation domain (23 ).
To determine if the LexA(BD)-Abf1(635-684)p fusion protein constituted a transcriptional activator, we carried out transcriptional activator assays with a LacZ reporter gene. Yeast harboring the reporter plasmid pSH18-34 (39 ) and expressing LexA(BD)-Abf1(635-684)p produced a white colony with only a hint of extremely pale blue color following extended growth on X-gal containing plates (unpublished observations). This analysis was followed up by quantitative assays performed with a chromosomally integrated derivative of pSH18-34, harboring the reporter LacZ gene, in which the regulatory LexA DNA-binding sites present are those of YCpLEX/A121 (Fig. 1 ). As shown in Figure 3 , the quantitative analysis confirmed the qualitative observations, namely LexA(BD)-Abf1(635-684)p as well as LexA(BD)p had negligible transcriptional activation potential. Under the same conditions the LexA(BD)-Gal4(AD)p fusion protein resulted in approximately three orders of magnitude higher levels of [beta]-galactosidase activity (Fig. 3 ).
Although LexA(BD)-Abf1(635-684)p lacked transcriptional activation potential, we nonetheless considered it possible that a transcriptional activation domain may be capable of ARS121 stimulation. We therefore tested LexA(BD)-Gal4(AD)p, harboring the Gal4p activation domain, for its ability to stimulate ARS121. The results of these mitotic stability assays are presented in Figure 4 A, along with the results for LexA(BD)-Abf1(635-684)p and LexA(BD)p. While LexA(BD)-Gal4(AD)p served as a potent transcriptional activator (Fig. 3 ), it provided negligible ARS stimulation (Fig. 4 A). In contrast, LexA(BD)-Abf1(635-684)p provided negligible transcriptional activation (Fig. 3 ), but stimulated the ARS (Fig. 4 A). Hence, it appears that there is no direct correlation between the ability of a protein to activate transcription and its ability to stimulate the ARS121 replication origin.
The results presented here indicate that a 50 amino acid region within the C-terminus of Abf1p is capable of stimulating the ARS121 origin. Origin stimulation by LexA(BD)-Abf1(635-684)p and by Abf1p require tethering of the protein to the ARS. Obviously, a functionally important region of Abf1p is the DNA-binding domain, even if only to serve a tethering role. It has been shown before that the DNA-binding domain of Abf1p spans an extended region, from residue 1 to ~530 (46 ), including a possible zinc finger structure located at the N-terminal portion of the protein (46 ,47 ).
Three LexA(BD)-Abf1p fusions, harboring Abf1p residues 1-323, 329-589 and 482-731, were utilized to survey the length of Abf1p for DNA replication activity. Of these three fusion proteins, only LexA(BD)-Abf1(482-731)p stimulated ARS121. Subsequent analysis delimited the replication stimulatory region to residues 635-684 of Abf1p. While it is possible that Abf1p residues 635-684 serve as the sole ARS-stimulatory region of Abf1p, the experiments presented here do not rule out the existence of additional domains that may function in DNA replication. For example, (i) placement of the LexA(BD)p at the N-terminus of Abf1p(1-323) or Abf1p(329-589) might have obscured another stimulatory region; (ii) a stimulatory region might have been disrupted by the breakpoints at Abf1p residues 323 and 329; (iii) exposure of a stimulatory region may require the binding of Abf1p to its cognate DNA-binding site.
Functional redundancies have often been observed within a given factor which activates transcription (48 ). It is unknown if this is true for factors which stimulate replication. We have recently replaced the ABF1 gene with a version that encodes Abf1p deleted of residues 635-684. The resulting strain appears to have wild type growth rates and exhibits wild type stability levels for plasmids harboring ARS121 (Wiltshire and Eisenberg, unpublished). Thus, if the 635-684 region of Abf1p stimulates ARS function as part of Abf1p, this region is likely to be one of at least two functionally redundant domains. It is interesting to note that Abf1p harbors two overlapping, functionally redundant domains for viability (44 ). The Abf1p (635-684) region, which stimulates ARS121 function as a fusion protein, overlaps with one of the two redundant domains.
Previous studies indicated that Abf1p DNA-binding sites and LexA(BD)-Abf1p fusion proteins are capable of acting as transcriptional activators (38 ,43 ,44 ). In fact it has been shown that the C-terminal 132 amino acids (residues 600-731) exhibit transcriptional activation potential when fused to the LexA(BD) (44 ). Using our experimental system we have observed that the LexA(BD)-Abf1(635-684)p fusion protein had very little transcriptional activity. However, larger fusions, in agreement with other studies, exhibited more significant transcriptional activation potential (unpublished). Hence, the region encompassing amino acids 635-684 may be a part of a larger C-terminal Abf1p domain involved in transcription.
Clearly, the C-terminal region of the Abf1p protein harbors the ability to activate transcription and stimulate the function of a replication origin. In our functional analysis we observe a lack of a direct correlation between the ability of a fusion protein to stimulate replication and its ability to activate transcription (unpublished observations). Namely, fusion proteins that were more active in transcription were not necessarily more active in replication, suggesting that the function of the two activities may be separable. The resolution of this question, however, may require further functional analysis by point mutations within the C-terminal region of Abf1p. A separation of function has been noted for a viral DNA replication system. The BPV E2 protein stimulates both replication and transcription. Detailed analysis of the E2 protein indicated that a replication-stimulatory domain coincided with, but was genetically separable from, a transcriptional activation domain (49 ).
While a region of Abf1p which stimulates replication may genetically be separable from a region of Abf1p that can act as a transcriptional activator, it is nonetheless possible that other transcriptional activation domains might stimulate ARS121. However, transcriptional activators LexA(BD)-GAL4(AD)p and Gal4p appeared unable to substitute for Abf1p. Moreover, the Rap1p DNA-binding site and REN1501 element were also unable to substitute for the Abf1p DNA-binding site, again indicating specificity of the Abf1p stimulatory effect on ARS121 activity. Therefore, even if Abf1p (635-684) is part of a larger domain involved in transcriptional activation, this domain harbors a capacity apparently lacking from other transcription factors.
Although it could be argued that the inability of alternative transcription factors to functionally replace Abf1p at the ARS121 origin is a result of inaccessibility of these factors to their cognate DNA-binding sites, we view this possibility as unlikely. All heterologous binding elements were inserted into a position at which the Abf1p DNA-binding site is functional.
In contrast to ARS121, ARS1 exhibits promiscuous stimulation by a variety of transcriptional activators, such as Gal4p, Rap1p and a LexA(BD)-Gal4(AD)p fusion (23 ). These results strongly suggest that the same functional element, an Abf1p DNA-binding site, may function differently at different ARSs. Specifically, it is possible that regions of Abf1p required for stimulation of ARS121 may be genetically separable from those required for stimulation of ARS1.
The results obtained strongly suggest a specific requirement for Abf1p at ARS121. This apparent specificity of the Abf1p/ARS121 interaction is not an exception. Recent studies from this laboratory have demonstrated the presence of a replication enhancer at ARS1501 (Raychaudhuri and Eisenberg, unpublished). This apparently novel enhancer, termed REN1501, does not appear to bind Abf1p, Rap1p or other known transcription factors. Substitution experiments demonstrated that only the Abf1p DNA-binding site, but not the sites that bind Rap1p or Gal4p, could partially replace the REN1501 element in stimulating ARS1501 (Raychaudhuri and Eisenberg, unpublished). Thus, among the three ARSs (ARS1, ARS1501 and ARS121) at which substitution experiments have been carried out, a spectrum of specificity in the action of DNA replication enhancers has been observed. Such specificity may reflect the diversity of ARS composition. Domains A of ARSs are non-identical, and ARSs harbor a differing nature, number and arrangement of auxiliary elements.
It appears that ORC/ARS1 interactions are Abf1p independent (50 ,51 ). However, this situation may be different at ARS121. We previously demonstrated that Abf1p is one of at least three factors required in vitro for the formation of a stable multiprotein complex at essential sequences of the ARS121 origin of replication (9 ,52 ). The binding properties and certain physical properties of one of the activities, core binding factor (9 ,52 ), indicated that it could be ORC or ORC-related. The results presented in our current analyses indicate that Abf1p may stimulate ARS121 and ARS1 activity by different mechanisms. Taking into consideration all available data, a simple model emerges. Abf1p at ARS121 may stimulate origin function by acting to stabilize a complex at the origin (9 ,52 ). In contrast, Abf1p at ARS1 may stimulate the origin by interacting with other factors, perhaps RPA (23 ,51 ).
A differential requirement for Abf1p stabilization of an origin complex could arise from divergent ORC DNA-binding sites. For example, such divergence is apparent in a comparison of the essential sequences of ARS121 and ARS1. The ARS121 domain A is 18-23 bp in length and its ACS is a 9/11 match to the consensus ACS (36 ), whereas the ARS1 domain A is only 15 bp in length and its ACS is an 11/11 match (23 ). Studies on the strength of ORC/ARS interactions for different ARSs have not been reported.
How Abf1p and/or LexA(BD)-Abf1(635-684)p carry out their stimulation of ARS121 is unknown. In simple terms, these factors could work by directly altering DNA structure or by protein-protein interactions. As Abf1p potently stimulates ARS function even at distances >1 kb (28 ), models invoking direct alterations of DNA structure seem less likely. The fact that LexA(BD)-Abf1(635-684)p harbors a region of Abf1p outside of the Abf1p DNA-binding domain, lends further support to the notion that Abf1p functions at ARS121 through contact(s) with other proteins. Therefore, identification and characterization of factors which interact with Abf1p should provide further insight into the mechanism of origin stimulation by Abf1p.
We are grateful to Nader Shakibai for a careful reading of the manuscript. This work was supported by the National Institute of Health Grant GM46760-03 (awarded to SE).
Bell,S.P., Kobayashi,R. and Stillman,B. (1993) Science 262, 1844-1849.
REFERENCES