GCN5 Is a Positive Regulator of Origins of DNA Replication in Saccharomyces cerevisiae

GCN5 encodes one of the non-essential Histone Acetyl Transferases in Saccharomyces cerevisiae. Extensive evidence has indicated that GCN5 is a key regulator of gene expression and could also be involved in transcriptional elongation, DNA repair and centromere maintenance. Here we show that the deletion of GCN5 decreases the stability of mini-chromosomes; that the tethering of Gcn5p to a crippled origin of replication stimulates its activity; that high dosage of GCN5 suppresses conditional phenotypes caused by mutant alleles of bona fide replication factors, orc2-1, orc5-1 and mcm5-461. Furthermore, Gcn5p physically associates with origins of DNA replication, while its deletion leads to localized condensation of chromatin at origins. Finally, Δgcn5 cells display a deficiency in the assembly of pre-replicative complexes. We propose that GCN5 acts as a positive regulator of DNA replication by counteracting the inhibitory effect of Histone Deacetylases.


Introduction
DNA replication in Saccharomyces cerevisiae initiates at discrete origins referred to as Autonomously Replicating Sequences (ARS). Origins share an essential core element called ACS (ARS Consensus Sequence), which binds the Origin Recognition Complex (ORC). In early G1 phase Cdc6p and six MCM (MiniChromosome Maintenance) proteins (MCM2-7) join ORC to form prereplicative complexes [1]. In S phase these complexes are activated by protein kinases to initiate DNA replication [1].
It is granted that histone acetylation can play a key role in the regulation of origins of DNA replication [2,3]. Convincing evidence has shown that in S.cerevisiae two major Histone-DeACetylases (HDAC), RPD3 and SIR2, act as negative regulators of origins [4,5,6,7]. It has been postulated that their inhibitory role is mediated by global deacetylation rather than local recruitment and retention of HDACs to replication foci. However, a recent study suggests that RPD3 may be targeted to multiple non-telomeric origins but the mechanism of recruitment remains elusive [7]. A link between histone deacetylation by Rpd3p and negative impact on origin activity has also been proposed in Drosophila [8]. At the same time, another HDAC, HST1, surprisingly acted as a stimulator of certain origins [9] thus suggesting a more complex cross-talk between histone acetylation and replication. Nevertheless, the prevailing notion is that deacetylation of histones has a negative impact on origin activity [3].
It is conceivable that the opposing activity of Histone Acetyl Transferases (HAT) stimulates origins of DNA replication, but details about the HATs involved are sparse. For example, in S.cerevisiae the tethering of Gcn5p to the late origin ARS1412 causes it to fire earlier [4]. Similarly, the tethering of Chameau (a HAT) to a chromosomal origin in Drosophila stimulates the activity of this origin [8] thus reiterating the general link between histone acetylation and activity of origins. Several recent studies in human cells have demonstrated that the HAT HBO1 is involved in the loading of MCM proteins on chromatin in early G1 phase [10,11,12] and in the progression of replication forks in S-phase [13]. Furthermore, HBO1 seems to respond to p53 and FAD24 to halt cell cycle progression [14,15]. HBO1 has no obvious structural homologue in S.cerevisiae, however the Sas2p-containing HAT complex NuA3 has been proposed as a remote functional homologue [16,17]. Interestingly, at semi-permissive temperature the deletion of SAS2 exacerbates the low origin activity caused by a mutation in orc2 and delays progression through S-phase in conjunction with orc2 mutations [18]. Another recent study has documented an interaction between Hat1p and ORC and has shown preferential association of Hat1p with origins at the time of their activation [19].
In this study we compared the effects of the deletions or mutations in several HATs including SAS2 and HAT1 on DNA replication in S.cerevisiae. We obtained several lines of evidence indicating that GCN5 is a major positive regulator of origins of DNA replication.

Materials and Methods
The yeast strains used in this study are listed in Table 1. Yeast cells were routinely grown in YPD or SC media at 23uC except in Table 2 as indicated.
Mini-chromosome stability assays with pARS1wt and pRS305 were performed as in [23,24]. Suppression of plasmid loss in W303, mcm5-461, orc2-1 and orc5-1 cells was performed by double transformation with pARS1wt and pGBKT7 or pGBKT7-GCN5, respectively. Cells were selected on SC-trp-ura plates. Three individual colonies were inoculated in SC-trp to allow for retention of the pGBKT7 episomes and for loss of pARS1wt and grown for 20 generations. Each culture was serially diluted 1/10, and 5 ul of each dilution were spotted on both SC-trp-ura and on SC-trp plates. Assays with pARS1/-B23/GAL4 were performed by co-transforming in DF5gal4D cells with pARS1/-B23/GAL4 together with pGAL4DBD-GCN5, pGAL4DBD-GCN5(KQL), pGAL4DBD-SAS2, pGAL4DBD-SAS3, pGAL4DBD-ESA1 and pGAL4DBD-HAT1, respectively. Cells were selected on SC-his-ura plates, grown for 20 generations in SC-his and then plated on SC-his and SC-hisura. In all cases plasmid loss was calculated as in [23].

Chromatin Immuno-Precipitation (ChIP)
We used a protocol, which favors weaker protein interactions with chromatin. It includes a second cross-linking agent with a longer arm in the cross-linking step [25]. Briefly, GCN5-MYC or ADA3-MYC cells at OD 600 = 1.0 in YPD were cross-linked with 1.6% formaldehyde for 30 minutes at 25uC and then suspended in 10 ml PBS/10 mM DMA(Pierce)/0.2%DMSO for another 30 minutes at 25uC. The cross-linking was quenched with 2.5 ml of 2.5 M glycine, cells were collected and crushed by glass beads in LB/IP buffer (50 mM Tris.HCl, 140 mM NaCl 2 , 5 mM EDTA, 1x PI cocktail (Sigma)) and sonicated by 12 bursts of 30 seconds at maximum output of Micronix cuphorn to produce fragments of about 500 bp. Chromatin was extracted in LP/IP plus 0.3% TX100 and 0.05% Na deoxycholate and immuno-precipitated overnight with anti-MYC beads (Sigma). The beads were washed three times with 1 ml LB/IP/0.5% TX100, 0.05% Na deoxycholate, one time with 1 ml LB/IP plus 350 mM NaCl, one time with 150 ul TE plus 10 mM NaCl (Final Wash) and eluted/uncrosslinked overnight in 150 ul TE/ 0.2% SDS at 65uC. After treatment with Proteinase K, DNA was extracted by Phenol/Chloroform and precipitated. The fragments of interest were quantified by PCR (after confirming linear range of amplification with crude genomic DNA) and ImageQuant software. Primer sequences are provided (Suppl .  Table S1).
For the analysis of MCM loading on origins, BY4247 and Dgcn5 cells were grown at 23uC in YPD to OD 600 = 1.0, arrested in Mitosis with 12 mg/ml Nocodazole for 3.5 hours, then released from mitotic arrest in fresh YPD. Aliquots of 100 ml were crosslinked and processed as above. Immunoprecipitation was conducted with a mixture of 2 ug each of anti-yMCM2 and anti-yMCM5 antibody (Santa Cruz) and Protein G beads (Sigma). Washing, elution, reversal of cross-linking and PCR were as described above.

Restriction Nuclease Accessibility Assay
We used the protocol for targeted analysis of loci [26] with slight modifications. BY4247 and Dgcn5 cells (100 ml) were grown in YPD to OD 600 = 1.0 and treated with Zymolase and pelleted. Spheroplasts were re-suspended in 5 pellet volumes of nuclease buffer (30 mM Tris.HCl pH 7.5, 10 mM MgCl 2 , 5 mM CaCl 2 , 50 mM NaCl, 1/100 volume of PI cocktail (Sigma)), split into 0.7 ml aliquots and incubated for 30 min at 37uC with 50 U of BglII, 50 U of DraI, 200 U of Micrococcal Nuclease (MNase) or no

Deletion of GCN5 Impairs Mini-Chromosome Stability
Mini-chromosomes are stably inherited (3-5% loss per generation) plasmids, which contain an ARS, a centromere (CEN) and a selection marker gene. Increased loss rate in non-selective medium is most frequently associated with flawed replication of the mini-chromosome independently of effects on other processes [24]. We used the mini-chromosome stability assay to test the involvement of several HATs in DNA replication. Strains with disrupted non-essential HAT genes (Dgcn5, Dhat1, Dsas2, Dsas3 and Drtt109) or mutations in the essential ESA1 gene (esa1-414, esa1L245P [27]) were transformed with two different minichromosomes carrying ARS1 (pARS1wt) or ARS305 (pRS315). In Fig. 1A we show that the disruption of GCN5 dramatically decreased the stability of both mini-chromosomes, while the deletion of the other HATs or the mutations in esa1 had marginal effects. The level of mini-chromosome instability in Dgcn5 cells was comparable to that of the mcm5-461 strain, which harbors a mutation in one of the essential origin licensing factors, MCM5 [28,29].

Tethering of Gcn5p Stimulates the Activity of an Impaired ARS1
The tethering of Gcn5p to a late origin (ARS1412) causes it to fire earlier [4]. We designed a similar tethering assay to directly compare the overall activity of several HATs on a minichromosomal origin. The mini-chromosome pARS1/-B23/G24 (Fig. 1B) carries ARS1, in which two of the auxiliary elements, B2 and B3, are destroyed and a GAL4 binding site is inserted instead of B3 [20]. These ablations render pARS1/-B23/G24 unstable to a point where its maintenance is critically dependent on the tethering of stimulators via the GAL4 binding site [20,30]. To test the roles of different HATs on this crippled origin, we cotransformed DFDgal4 cells with pARS1/-B23/G24 plus a set of constructs expressing GAL4-fusions of Gcn5p, gcn5p(KQL) (a GCN5 mutant lacking HAT activity), Sas2p, Sas3p, Esa1p and Hat1p, respectively, and measured the stability of pARS1/-B23/ G24. In Fig. 1B we show that Gcn5p and Hat1p reduced the loss rate of pARS1/-B23/G24 from 25% to 10 and 12%, respectively. Since the theoretical maximum loss measurable by this assay is 50% for non-replicating plasmids and is in the range of 30% for very poorly replicating plasmids [24], these effects were considered significant. In contrast, Gcn5p(KQL), Sas2p, Sas3p and Esa1p did not reduce the loss rate of pARS1/-B23/G24 (Fig. 1B). We note that all GAL4-fusion constructs express active HAT proteins in vivo as demonstrated elsewhere [22].

High Dosage of GCN5 Suppresses Mini-Chromosome Instability in Replication Factor Mutants
In an earlier study we have shown that high dosage of TRA1 (a component of the HAT complexes SAGA and NuA4) can suppress the Ts 2 phenotype of a strain with profound replication deficiency, mcm5-461 [29]. We reasoned that if GCN5 works in DNA replication, it may have a similar effect in this and other replication factor mutants. We supplemented the mcm5-461 [29], orc2-1 [31] and orc5-1 [32] strains with stable high copy 2 mm episomes, which carry GCN5 or a non-HAT component of the SAGA/ADA complex, SPT8. As controls, we used two genes, which encode components of the FACT complex, SPT16 and POB3. FACT has been implicated in pol II transcription and in the elongation step of DNA replication, but not in origin activation [33,34]. In Table 2 we show that high copy of GCN5 and SPT8 suppressed the conditional Ts 2 phenotypes of mcm5-461, orc2-1 and orc5-1. In contrast, SPT16 and POB3 had no effect. Next, we tested if the high dosage of GCN5 in these mutants also suppressed the severe instability of mini-chromosomes. We co-transformed these strains with a test mini-chromosome (pARS1wt) and the GCN5carrying episome and assessed the level of loss of pARS1wt. The high dosage of GCN5 significantly improved the stability of pARS1wt relative to the control (Fig. 2). Hence, high dosage of GCN5 specifically suppressed the replication deficiencies of mcm5-461, orc2-1 and orc5-1.

Gcn5p Associates with Origins of DNA Replication
Next, we used ChIP assays to test if Gcn5p directly associates with origins of DNA replication. In the first set of experiments we used a modified pARS1wt/link mini-chromosome, which has been designed to minimize overlap of ChIP signals from URA3 and ARS1 ( [21], see Fig. 3). It contains a 2.2 kb fragment of nonyeast DNA between ARS1 and the URA3 reporter. pARS1wt/link Table 2. High dosage of GCN5 and SPT8 partially suppress the Ts 2 phenotypes of replication factor mutants.

High dosage gene
Growth at 23uC Growth at 37uC orc2-1, orc5-1 and mcm5-461 cell were transformed with high copy episomes (2 mm) carrying the shown genes and then grown at 23uC or 37uC. Growth was assessed by size of colony after 3 days at 23uC or after 5 days at 37uC. doi:10.1371/journal.pone.0008964.t002 was introduced in the ADA3-MYC strain [35] and ChIP was performed with anti-MYC antibodies. Ada3p, Ada2p and Gcn5p build up a stable complex, which forms the catalytic core of all Gcn5p-containing HATs [35]. The amounts of precipitated minichromosomal ARS1 only (one of the primers anneals to the plasmid backbone) and two flanking fragments that are positioned about 1 kb away from it were quantified by triplicate PCR. The data showed that about 4% of the mini-chromosomal ARS1, 0.6% of the fragment between ARS1 and URA3 and 0.1% of the fragment next to CEN4 were immunoprecipitated by anti-MYC antibodies (Fig. 3, lane 4). In comparison, the anti-MYC antibodies precipitated none of the genomic Ty1b transposon (negative control) (Fig. 3, lower panel) and about 5% of the genomic ACT1 (positive control). The non-crosslinked samples (Fig. 3, lane 2) and the samples without antibody (Fig. 3, lane 3) uniformly produced negligible signals with all loci tested. We concluded that Ada2p/Ada3p/Gcn5p specifically associates with the mini-chromosomal ARS1.
In another set of experiments we analyzed the association of Gcn5p with chromosomal origins. We must note that the high gene density in the yeast genome does not provide the comfortable distances between genes and origins that were available on pARSwt/link. Because of this concern we focused on two extensively characterized origins with finely mapped ACS and auxiliary elements, ARS1 and ARS305, in a GCN5-MYC18 strain [36]. The ACS elements of these origins are located more than 500 base pairs (the average resolution of ChIP) away from the nearest promoter. In these assays we compared the signal from the core ACS to the signals from two flanking elements positioned between 200 and 500 bases away from the ACSs (see maps in Fig. 4). The relative amounts of each of the analyzed fragments in the Load, Wash and anti-MYC-Eluate fractions were quantified by PCR (triplicate). The per cent of signal in Eluate versus Load was calculated and plotted next the maps of ARS1 and ARS305 (Fig. 4). In addition, we tested association of Gcn5-MYCp to two control loci: the ACT1 gene and the repressed Tyb1 transposon (Fig. 4).
The ChIP experiments consistently produced higher signals from the ACS elements of ARS1 and ARS305 as compared to the flanking fragments. The moderately high signal of 1.3% at the ARS1 flanking element towards TRP1 probably reflects its proximity to the origin (196 bp) or strong association of Gcn5p to the upstream TRP1 gene. The recovery of DNAs in the immunoprecipitates represented 2.9% for ARS1 and 1.8% for ARS305 of the total DNA analyzed, while the recovery of ACT1 (positive control) DNA was 4.3% (Fig. 4). As expected, no association of Gcn5p to the repressed Ty1b element was observed (Fig. 4). These weaker but still noticeable chromatin immunioprecipitation signals from origins relative to promoters are in tune with the observations in [36]. In this comprehensive genome-wide  mapping of Gcn5p association (data available on http://frodo.wi. mit.edu/cgi-bin/yeast_histones/gbrowse.cgi/yeast_histones) moderately higher signals of Gcn5p binding at ARS1 and ARS305 (and also at ARS501, ARS606, ARS1021, ARS1414, ARS1618) relative to the immediate neighboring regions can be observed. However, these peaks were below the preset threshold of the computational analysis and were not discussed in [36].We should also mention that many ARS reside within short distance from gene promoters thus adding to the problematic interpretation of such weak signals. We completely agree that the genome-wide approach in [36] has not identified association of Gcn5p to ARSs. Nevertheless, their data are in tune with our focused approach and support the notion of weak association of Gcn5p to origins. We also need to note that we used a protocol with two cross-linkers (formaldehyde and DMA, [25]) that enhances weak ChIP signals as compared to a single cross-linker (formaldehyde) in [36]. In summary, the results in Figure 4 showed specific, but weaker association of Gcn5p to chromosomal origins relative to the better isolated mini-chromosomal ARS1 origin (Fig. 3). This outcome could be a consequence of the expected higher background signal on chromosomal origins. The alternatively explanation is that for some unknown reason the association of Gcn5p to the mini-chromosomal ARS1 origin is intrinsically stronger.

Deletion of GCN5 Perturbs Chromatin Structure of a Chromosomal Origin
We reasoned that if the weak association of Gcn5p to chromosomal origins is functionally relevant, the deletion of GCN5 would affect their chromatin structure. Initial analyses by limited MNase digestion showed the existence of hypersensitivity sites in the vicinity of ARS1, ARS305 and ARS501 (not shown). However, the differences between Dgcn5 and wt cells were hard to assess because this enzyme also cleaves in the broader area of the analyzed loci. It is noteworthy that a similar DNAase accessibility assay has failed to detect variations in the structure of ARS1 between wt and orc2-1 cells [18], yet it is known that ARS1 activity is severely impaired in orc2-1 cells [31]. We concluded that the MNase or DNAase sensitivity assays are inadequate for analysis of ARS. Therefore, we tailored a focused restriction enzyme (BglII) accessibility assay [26] for the analysis of ARS1 on chromosome IV. BglII cleaves ARS1 between the B2 and B3 elements [20]. Briefly, equal amount of spheroplasts from exponentially growing wt or Dgcn5 cells, respectively, were lysed and exposed to excess of BglII. Under these conditions the cleavage of sites depends on site accessibility thus representing the level of chromatin compactness of the tested locus. Control samples were incubated without BglII to monitor for endogenous nuclease activity. In addition, to distinguish specific signals during the indirect endlabeling, excess of MNase was added in parallel samples of lysed spheroplasts. Subsequently, total DNA from all samples was purified and a second complete digestion was performed with EcoRI and EcoRV, which cut on either side of ARS1. Finally, the produced DNA fragments were analyzed by indirect end-labeling with a probe, which anneals 475 bp away from the BglII site. The ratio of un-cleaved EcoRI-EcoRV versus BglII-cleaved fragment was used as a measure for the accessibility to ARS1 and therefore for chromatin compactness of the analyzed locus.
Using this assay we reproducibly observed high accessibility to the BglII site in ARS1 in wt cells (Fig. 5A, lane 2), while the parallel samples from Dgcn5 cells (Fig. 5A, lane 5) consistently showed moderate accessibility. The bands corresponding the BglII-cleaved (475 bp) and un-cleaved (1066 bp) fragment were abolished by

Deletion of GCN5 Has a Relatively Minor Effect on the Structure of CEN4
An earlier study has implicated GCN5 in the regulation of centromeres [37]. More specifically, the study has shown increased accessibility of DraI sites in centromeres in Dgcn5 cells. We attempted to evaluate the extent of chromatin modifications at origins and centromeres in Dgcn5 cells using a similar DraIaccessibility assay for the analysis of CEN4. Samples from wt and Dgcn5 cells prepared in parallel with the samples for Fig. 5A were digested with DraI and processed accordingly with KspI and BglII. Next, indirect end-labeling with a probe annealing 600 bp from the three juxtaposed DraI sites in CEN4 was performed (Fig. 5B). In these experiments we consistently observed fairly similar levels of DraI cleavage in both wt and Dgcn5 cells (Fig. 5B, lanes 2 and 5). Calculation of the intensity of the cleaved (600 bp) and uncleaved (878 bp) CEN4 fragments in Fig. 5B showed that 14% and 18% of the DraI sites were digested in wt cells or Dgcn5 cells, respectively. In agreement, the accessibility to DraI in [37] was similarly estimated at 9% in wt cells and 13% in Dgcn5 cells.
In summary, our assays have demonstrated that the deletion of GCN5 produces minor alterations in the chromatin structure of CEN4 and major alterations of ARS1.

Deletion of GCN5 Impairs the Assembly of Pre-Replicative Complexes on Origins in G1
Earlier studies have shown that the lack of GCN5 delays progression through G 2 and Mitosis after synchronization with a-factor [37,38]. When cells are synchronized in Mitosis, they also progress slower through G1/early S phase and display later appearance of buds as compared to wild type cells ( Figure S1). We considered that an event in Mitosis or G1 that is important for regulation of origins could be impaired in Dgcn5 cells.
Shortly after exit from Mitosis the origins assemble and ''license'' pre-replicative complexes. This ''licensing'' is highlighted by the loading of the MCM proteins [1]. Consequently, the binding of MCM proteins to origins is used as an assessment for the efficiency of assembly of pre-replicative complexes [39]. We tested the hypothesis that the negative effect on DNA replication in Dgcn5 cells is linked to impaired loading of MCM proteins. We synchronized wt and  and Suppl. Fig. S1). Another sample was collected and cross-linked 60 or 80 min past release for wild type or Dgcn5 cells, respectively. Cross-linked chromatin was sheared, extracted and immunoprecipitated with a mixture of anti-MCM2 and anti-MCM5 antibodies (Santa Cruz). We have found that this mixture of anti-peptide antibodies is more efficient in recovery of origin DNA as compared to the individual antibodies used alone [40,41]. Finally, the immunoprecipitated DNA was analyzed by quantitative PCR with primers for ARS1 and ARS305. Consistent with many previous reports (reviewed in [1]), in wt cells we saw significant enrichment of anti-MCM immunoprecipitated ARS1 and ARS305 DNA at the 20 th min past release (G1) with significant decrease of the signals at the 60 th min (S) (Fig. 6). In comparison, ACT1 showed very little association with MCM proteins at the time points of our experiment. A similar pattern of MCM binding to origins was observed in Dgcn5 cells except that the S phase sample was collected at the 80 th minute to compensate for the slower progression through the cell cycle. Importantly, in Dgcn5 cells the recovery of ARS1 and ARS305 DNA in G1 was two-three fold lower as compared to wt cells (Fig. 6). These results demonstrate significant deficiency in the assembly of prereplicative complexes in Dgcn5 cells.

Discussion
In S.cerevisiae, ten HATs have been described to date [42,43,44]. Several genome-wide studies on gene expression [45,46,47,48] and chromatin association [36,49] have provided significant details on the engagements of the HAT complexes SAGA, TFIID, Mediator and NuA4 at gene promoters. However, these papers have not suggested the participation of any of these HATs in DNA replication. On the other hand, focused studies of origins or studies on protein interactions have tentatively linked HAT1 [19] and SAS2 [18] to DNA replication. In the present study we have compared the effects on DNA replication of deletions of HAT1, SAS2, SAS3, GCN5 and RTT109 and mutations in the essential ESA1 gene. These assays clearly outlined a major replication deficiency in the Dgcn5 strain with minor effects in the others (Figures 1 and 2). All subsequent analyses supported the notion that GCN5 is a positive regulator of origins. We postulate that GCN5 counteracts the negative effects of HDACs such as SIR2 and RPD3 [4,5,6]. Our observations expand the functional repertoire of GCN5, which has already been implicated in gene expression [43] and also in centromere function [37].
GCN5 has been extensively characterized as a global regulator of gene expression [36,47,50,51]. Is it then possible that all effects we have observed are indirect via deregulated gene expression? Several arguments reduce the likelihood of this possibility. Firstly, we have detected credible signals of direct association of Gcn5p with a mini-chromosomal ARS1 (Fig. 3). Whereas the chromosomal ARS1 and ARS305 do not provide the luxury arrangement and distances between the genes and origins on the minichromosome, signals from direct Gcn5p association at these loci have been detected by us (Fig. 4) and others [36]. Finally, transcriptome analysis in Dgcn5 cells has revealed a bias towards the regulation of stress induced genes, but not towards genes responsible for DNA replication factors [47]. We therefore favor the idea that GCN5 directly regulates DNA replication.
The fact that Gcn5p weakly and possibly only transiently associates with origins ( Fig. 4) suggests that its effects on origins are produced by global activity rather than targeted recruitment as is the case on promoters [36]. This idea is in tune with the notion that the negative origin regulators SIR2 and RPD3 are likely to act by global deacetylation [4,5,6]. A recent study suggests that at least in the case of RPD3 some targeted recruitment to origins is possible, but the mechanism remains unknown [7]. Regardless of whether the mode of action of Gcn5p is targeted or global, the effect of GCN5 on the chromatin structure of origins is substantial as shown in Fig. 5A. Another point of consideration regarding the mechanism of action of GCN5 is the observed deficiency of assembly of pre-replicative complexes in Dgcn5 cells (Fig. 6). This result suggests that GCN5 could have a major role in chromatin de-condensation, which in turn affects the licensing of origins. Studies in human cells have allocated a similar role in G1 for HBO1 [10,11,12]. At this point it is not clear if in human cells HBO1 is the only HAT engaged in G1 or if GCN5 also participates. As for S.cerevisiae, the remote functional homologue of HBO1, SAS2, shows little activity consistent with a major role in replication (Fig. 1). On the other hand, the deletion of HAT1 seems to have only minor effect on mini-chromosome stability (Fig. 1A), but strongly stimulates an origin when tethered to it (Fig. 1B). This observation and the data in [19] indicate that HAT1 could be involved in replication but the assays employed by us did not properly reveal such function. Finally, the proposed role of GCN5 in chromatin de-condensation and facilitation of pre-replicative complexes in G1 does not at all exclude additional roles in S phase as activator of origins or during elongation.
A very recent study has shown that in human cells GCN5 acetylates the key component of pre-replicative complexes CDC6 thus modulating its subsequent phosphorylation by Cyclin A-CDKs [52]. It is not clear if similar acetylation of Cdc6p by Gcn5p takes place in S. cerevisiae. It is tempting to speculate that at least some of the effects we observe here are consequence of not only histone acetylation, but of acetylation of Cdc6p as well. This question remains open until clear link between GCN5 and CDC6 is established.
Our data suggests that GCN5 has a major role in DNA replication. Nevertheless, GCN5 is not an essential gene so it is obviously not the only HAT involved. In support, the deletion of sas2, sas3 and hat1 and the esa1 mutants displayed slightly higher mini-chromosome instability as compared to wild type cells, but significantly lower as compared to gcn5 (Fig. 1A). It is conceivable that these HATs also play roles in the same steps, which are regulated by GCN5. In particular, it is not clear to what extent the mutations in ESA1 impair its function thus rendering its role in DNA replication far for from being excluded. It is noteworthy that ESA1 and GCN5 show a significant level of functional overlap on gene promoters [36].
Another issue stemming from our study is the possible role of GCN5 on centromeres [37]. As mentioned earlier, GCN5 has been implicated as a regulator of chromatin structure and in the function of centromeres [37]. Indeed, double deficiency of the ARS and CEN elements can explain the magnitude of Dgcn5 effect in Fig. 1. Whereas such scenario is not unlikely, our data on two chromosomal origins and a centromere shows that the alteration in chromatin structure is significant at origins and moderate to low on centromeres (Fig. 5). Even more, GCN5 strongly suppresses the replication deficiency phenotype of the mcm5-461, orc2-1 and orc5-1strains (Fig. 2). Hence, we suspect that GCN5 is a major positive replication factor independently of its activity on centromeres.
In summary, we have assigned a new role for GCN5. We have also introduced a new HAT in the regulation of DNA replication. Future studies should focus on the interplay between GCN5 and other HATs in modulating the activity of origins and the subsequent steps in DNA replication.