Physical interaction of ORC with Hat1p and Hat2p
To search for additional proteins that interact with Hat1p, we used the tandem affinity purification (TAP) technique, which has proven to be an effective method for detecting protein complexes in yeast [37, 38]. TAP-tagged versions of Hat1p were purified 106-fold to virtual homogeneity from soluble cell extracts derived from large-scale log-phase cultures [39]. As expected, Hat1p co-purified with Hat2p and Hif1p (Figure 1). In addition, and unexpectedly, we detected each of the six subunits of ORC in independent purifications of TAP-tagged Hat1p. The association of ORC was sub-stoichiometric relative to Hat1p, Hat2p and Hif1p, but reproducible and highly specific in that ORC was detected in only a few other successful purifications of over 2500 different yeast proteins [37].
As ORC has not been found previously to interact with any member of the Hat1p complex, we performed stringent reciprocal affinity tagging and purification experiments using C-terminally TAP-tagged versions of ORC subunits (Orc1p-6p). In addition to all known subunits of the ORC complex, Hat1p and Hat2p were often detected, albeit again sub-stoichiometrically and not easily visible by silver stain (Figure 1A,B). We independently tested the ORC-Hat1p interaction by immunoprecipitation of an endogenously myc-tagged version of Hat1p using a monoclonal antibody specific for Orc3p (Figure 2). Co-immunoprecipitation with this antibody led to the detection of myc-tagged Hat1p as seen with myc-tagged Orc5p (Figure 2A), whereas these proteins were not detected in immunoprecipitations with an anti-GFP control antibody. When we tested for the association of Hat2p and Hif1p with ORC using an ORC3-specific antibody, we found that ORC interacts with Hat2p but not Hif1p (Figure 2B). Importantly, co-immunoprecipitation of Hat1p with ORC was also observed when precipitated proteins were treated with the DNA-degrading enzyme benzonase (Additional file 1).
To further define Hat1p complex architecture, mainly with respect to its association with ORC, we subsequently performed purifications of TAP-tagged Hat1p and Hif1p with other subunits deleted (Figure 1C, Additional file 1). Notably, Hat1p was observed in sub-stoichiometric amounts when Hif1-TAP was purified whereas the Hat2p/Hif1p association was unaffected by HAT1 deletion. Given that the association of Hat1p/Hat2p with ORC was only weakly visible by SDS-PAGE/silver staining gel, concentrated purification products were analyzed by Western blot using antibodies against the Orc2p, Orc3p, and Orc5p subunits (Figure 1D). Upon repeated purification of Hif1-TAP, the Hat2p and Hat1p subunits were identified, but ORC was conspicuously absent (Figure 1C,D). Importantly, we also found that ORC remains associated with Hat1p/Hat2p even in a strain where HIF1 is deleted, whereas loss of Hat2p leads to disintegration of the entire complex (Figure 1D). Collectively, our analysis confirms a specific, and previously overlooked, physical interaction between ORC and a Hat1p/Hat2p sub-complex that is distinct from the established nuclear Hat1p/Hat2p-Hif1p.
To assess the enzymatic functions of the purified complexes, we assayed the transfer of 14C-labeled acetyl-CoA by the Hat1p sub-complexes onto purified core histones from chicken erythrocytes (Figure 1E, Additional file 1). In agreement with previous findings [30, 31], we found that histone acetyltransferase activity by Hat1p was enhanced approximately 10-fold in the presence of Hat2p. Substrate specificity and activity on histones, however, seemed not to be altered by the presence of Hif1p and ORC. When we tested for the in vivo association of Lys12-acetylated histone H4 by Western blot, we found that it is largely dependent on intact Hat1p complex (Figure 1F). The absence of Hat1p or Hat2p leads to a complete loss of acetylated histone H4 from the residual complex components, implying both of these factors in the interaction with acetylated H4.
Considering the role of ORC in the formation of the pre-RC during G1 phase, we next examined if the association of Hat1p with ORC was cell-cycle dependent. We performed co-immunoprecipitations using whole cell extracts prepared from yeast cultures arrested either in G1 phase with α-factor, in S-phase with hydroxyurea, or in G2/M with nocodazole, as well as from cells synchronously released into S-phase after an α-factor imposed G1 block (Figure 2C). Surprisingly, a comparable level of myc-Hat1p co-precipitated with ORC throughout these cellular treatments, suggesting that Hat1p, and by inference the Hat1p/Hat2p sub-complex, remained stably associated with ORC throughout the cell cycle. Consistent with this finding, large-scale purifications and mass spectrometric analysis of TAP-tagged Orc2p from α-factor arrested cells confirmed the association of Hat1p/Hat2p with ORC in the G1 phase (data not shown). By contrast, we failed to detect components of the pre-RC (such as Cdc6p) under these same purification conditions (BS, unpublished observations). The pre-RC could be readily disrupted, whereas the Hat1p/Hat2p-ORC interaction may be physically more robust under the chosen experimental conditions.
Genetic interaction between orc-ts and hatΔ/hat2Δmutants
Previously, genome-scale screens using the synthetic genetic array (SGA) methodology were performed to identify novel functional partners of ORC [11]. An array of about 4600 haploid deletions was combined with conditional temperature-sensitive (ts) orc2-1 and orc5-1 alleles which are hypomorphic for initiation of DNA replication, and scored for slow growth or lethality of the double-mutant meiotic progeny. We repeatedly detected genetic interactions (enhanced slow growth or synthetic sickness), leading to a reduced maximum permissive temperature, between either orc2-1 or orc5-1 and hat1 or hat2 deletions. These functional associations were highly specific as few other interactions with hat1 or hat2 were identified in SGA screens of over 200 unrelated query genes [40].
The synthetic composite phenotype of hat1 and hat2 in combination with orc5-1 and orc2-1 was independently confirmed by tetrad analysis in the W303 genetic background (Figure 3). Deletions of either HAT1 or HAT2 markedly reduced the growth rate of orc5-1 mutant cells at the semi-permissive temperature of 31°C. The double mutants formed small, slower growing colonies (Figure 3A). To exclude the possibility that the reduced colony size of the double mutants was caused by a delay in germination, the synthetic growth defect of orc5-1 hat1Δ and orc5-1 hat2Δ mutants was further confirmed by a serial dilution assay of mitotically dividing cells. While the double mutants were fully viable at 23°C, a loss of viability was observed at 31°C (Figure 3B), which is the maximal permissive temperature for the orc5-1 allele. Combination of orc2-1 or orc1-161 with either hat1Δ or hat2Δ also reduced growth at the maximum permissive temperatures of 26°C and 28°C (Figure 3C,D). The hat2Δ mutation caused a slightly greater slow-growth phenotype than hat1Δ when combined with either the orc2-1 and orc5-1 alleles. The reason for this subtle distinction is presently unclear, but could be an indication of redundant HAT activities. However, no further reduction of viability was observed in orc5-1 hat1Δhat2Δ triple mutants, consistent with a virtually complete functional (i.e. epistatic) overlap between Hat1p (catalytic core) and Hat2p (histone targeting co-factor). By contrast, deletion of HIF1 did not lead to a reduction of the permissive temperature for orc5-1 mutations (Figure 3E), pointing again to the absence of a functional interaction between Hif1p and ORC. As synthetic genetic interactions with conditionally lethal mutations are often indicative of a shared or overlapping biological function, these data suggest that Hat1p/Hat2p likely shares a function with ORC, presumably in DNA replication, separate from its function with Hif1p in chromatin assembly [28]. However, we observed only a very subtle reduction of colony size when hat1 was combined with the replication defective mutant mcm2-1 (Additional file 2), while virtually no interaction was detectable between hat1 or hat2 null alleles in combination with ts-alleles in other essential initiation factors examined (e.g. cdc7-1 or cdc6-1; Additional file 2 and data not shown). Hence, our data suggest that the genetic interaction observed between orc-ts mutations and hat1Δ/hat2Δ is specific and closely mirrors the physical interaction.
HAT1 mutations affect cell viability and cell-cycle progression in orc5-1mutants
To quantitatively monitor the loss of viability, asynchronous cultures of hat1Δ, orc5-1 single, and orc5-1 hat1Δ double mutants were shifted to the restrictive temperature for orc5-1 (36°C) for up to 5 h (Figure 4A). Samples were taken at distinct timepoints and the fraction of cells capable of forming viable microcolonies upon rescue to room temperature was measured. In this manner, we confirmed that hat1Δ indeed enhanced the loss-of-viability of orc5-1 alleles held at 36°C. By contrast, hat1Δ single mutant cells retained full viability as compared to the wild-type control, indicating that the hat1Δ loss-of-function mutation specifically amplifies the viability defects associated with impaired ORC. When synchronized cells were shifted to the restrictive conditions, orc5-1 cells were distinctly more sensitive to inactivation in G1 relative to S-phase (Figure 4B), reflecting the major role of ORC in pre-RC formation in G1 and being consistent with previous findings for orc2-1 [41]. Deletion of HAT1 enhanced this effect approximately twofold.
Growth arrest of orc2-1 and orc5-1 strains occurs predominantly at the G2/M transition [7, 9], and is dependent on activation of the DNA damage and spindle surveillance checkpoints in response to impaired or incomplete DNA replication [42]. As a measure of any further compromise in DNA replication efficiency, we examined whether the G2/M transition was additionally delayed in orc2-1 hat1Δ strains using flow cytometry (Figure 4C). Log-phase cultures of the single and double mutants and the wild-type controls were arrested with α-factor, and then synchronously released into the cell cycle at about 26°C, which is semi-permissive for orc2-1. Partial accumulation of orc2-1 cells in G2/M was readily observed following completion of the bulk of DNA synthesis (around 130 min), as expected from previous studies [43]. However, the fraction of orc2-1 hat1Δ double mutants that remained permanently arrested in G2/M at 26°C was substantially larger than that seen with the orc2-1 single mutant alone (Figure 4C). A similar enhancement of the orc-ts phenotype was observed in combination with the orc5-1 allele (see Additional file 3). By contrast, the overall cell-cycle progression profiles of the hat1Δ and hat2Δ single mutant strains were unperturbed, and comparable to a wild-type control (Figure 4C, Additional file 3). Consistent with inactivation of origins by orc-ts mutations resulting in a DNA damage response, we detected faster cell-cycle progression of orc5-1 single and orc5-1 hat1Δ double mutants after inactivating the RAD9 and RAD24-dependent DNA damage checkpoint pathways (see Additional file 3). In addition to the impaired G2/M transition, the consequences of inactivation of origins in orc5-1 strains were manifested by impaired S-phase progression at the fully restrictive temperature of 36°C (Figure 4D). Flow cytometric analysis showed that completion of DNA synthesis took longer in orc5-1 mutants as compared to wild-type cells. Importantly, this delay in S-phase progression became even more pronounced in the orc5-1 hat1Δ double mutants (see profiles at 25 and 30 min), consistent with the loss of HAT1Δfurther compromising the DNA replication and cell-cycle defects stemming from the orc5-1 conditional allele. Hence, the shared Hat1-ORC function appeared to reflect some core aspect of ORC-driven DNA replication.
Hat1p localizes to firing origins and persists with replicating DNA during S-phase
As ORC recruits replication initiation factors to origins [44], we examined if Hat1p associates with origin DNA, either throughout the cell cycle (such as ORC, and as might be expected for a stable Hat1p/Hat2p-ORC complex) or specifically during S-phase (such as Cdc45p). To this end, we performed chromatin immunoprecipitation (ChIP) assays to measure direct binding of TAP-tagged Hat1p to selected ARS sequences and control DNA at various timepoints after release into S-phase from α-factor-mediated G1 arrest (Figure 5). Parallel analysis was performed with strains containing TAP-tagged versions of Cdc45p as positive controls. ORC is bound exclusively to origin DNA, whereas Cdc45p, after being recruited to origins, associates with the advancing replication fork upon origin firing [5, 6]. We measured binding to two early origins, ARS305 and ARS1, and to late origin ARS1412, as well as to the control sequence R11 that is not known to be immediately adjacent to any active origin. To improve overall resolution, S-phase progression was slowed by release of the synchronized cell population into growth medium at 16°C or 20°C.
While no or only minor binding of Hat1p-TAP to ARS sequences was observed immediately at α-factor arrest (0 min timepoint), Hat1p-TAP bound specifically to ARS305 after release into S-phase. (Figure 5A). Peak ARS binding of Hat1p coincided with that of Cdc45p, and by inference with origin activation, and was followed by diminished binding at later timepoints, concurrent with the completion of initiation. An increase of immunoprecipitate was detected at the R11 control sequence with a lag relative to ARS305. An enrichment of Hat1p during S-phase, similar to that of Cdc45p, was also evident with ARS1 (Figure 5B). The association of Hat1p-TAP with the late firing ARS1412 was virtually indistinguishable from the binding to R11, consistent with late activation and replication timing of this origin (Figure 5C). Thus, our data indicate that Hat1p becomes associated with origins of replication around the time of origin firing and later with non-origin sequences, such as Cdc45p, suggesting that Hat1p also associates with advancing replication forks. No recruitment of Hat1p to ARS1 and R11 was observed when DNA replication was compromised by shifting the orc2-1 allele to the restrictive temperature (36°C). However, association with ARS1 is similarly abolished or lowered by the cdc7-1 mutation at restrictive (36°C) and at permissive (23°C) temperature (Figure 5D). As cdc7-1 did not show a genetic interaction with hat1Δ, the recruitment of Hat1p depends on DNA replication in general but not specifically on intact ORC. Importantly, the dynamic recruitment of Hat1p to origins during S-phase contrasts also with the constitutive association of Hat1p/Hat2p with ORC as seen by co-immunoprecipitation (Figure 2C). Hence, the discrepancies in our data can at least partly be explained by the existence of two separate complexes, Hat1p/Hat2p-ORC and the previously reported Hat1/Hat2-Hif1p. The S-phase specific recruitment of Hat1p could reflect the documented role in chromatin assembly during fork progression mediated by the Hif1p-containing complex [28].
A replication defect in the absence of Hat1p may be expected in case Hat1p/Hat2p-ORC contributes to the function of an origin. To understand whether there was a contribution of Hat1p/Hat2p to the function of a specific origin (ARS1), we made use of a plasmid-based minichromosome maintenance assay, which is often used to characterize mutations that affect the efficiency of initiation of DNA replication [45]. Maintenance of a single ARS1-containing plasmid was largely perturbed at the restrictive temperature in orc5-1 mutants (Figure 6A), but rescued by the presence of multiple ARSs. Multiple ARSs also suppressed the plasmid loss defect seen with the orc5-1 hat1 double mutants; however, this strain was not any more defective in plasmid maintenance in comparison with the orc5-1 single mutant. Moreover, the rate of plasmid loss in hat1Δ single mutants was not elevated relative to wild-type cells. Thus, Hat1p seems not to have an effect on initiation frequency and DNA replication of the widely studied origin ARS1.
To allow for detection of incomplete or defective replication intermediates, or of altered timing of chromosomal origin function, we investigated the effect of deleting HAT1 on the activity of specific chromosomal origins (ARS305 and ARS1) using neutral two-dimensional gel electrophoresis. For DNA extraction, we performed the cetyltrimethylammonium bromide (CTAB) method, developed by Lopes and colleagues [46], The level of active origin firing was assessed by comparing active replication intermediates (bubble arcs) to the amount of monomeric DNA, whereas passive replication was reflected by the relative amounts of detectable small Y-form molecules (Figure 6D, calculations not shown). However, no distinct alterations in any replicative intermediates were detectable at either of several early (ARS1, ARS305) or late (ARS603, ARS1412) origins tested in a hat1Δ single mutant background (Figure 6B,C, and data not shown). Similarly, no obvious effect on replication initiation (ARS1, ARS305) was observed when replicative intermediates are compared between orc5-1 and orc5-1 hat1 (Additional file 4). The absence of a direct role of Hat1p in the efficiency and timing of these origins could be explained by the possibility that Hat1p/Hat2p does not interact with ORC at these origins and that the S-phase recruitment of Hat1p is solely mediated by Hat1p/Hat2p-Hif1p. However, crosstalk between the two complexes may also occur and other plausible explanations for these observations cannot be dismissed (see Discussion).
Is it possible that Hat1p/Hat2p-ORC highlights a distinct function of ORC that is different from its role in DNA replication? A pertinent function in chromosome metabolism may be related to the role of ORC in transcriptional silencing that is genetically separable from its role in DNA replication [47]. The N-terminal portion of Orc1p is nonessential and therefore dispensable for DNA replication, but is required for the ORC silencing function [48]. This defect can be functionally substituted by the N-terminus of Sir3p, which shares sequence similarity with the Orc1p N-terminus. However, we observed that myc-tagged Hat1p still co-precipitated with ORC in a strain bearing a N-terminal truncation of Orc1p or the N-terminus of Sir3 (Additional file 5). From this, the Hat1p/Hat2p-ORC association seems not to be dependent on the silencing function by the Orc1p N-terminus.
Histone H4 acetylation is associated with yeast ORC
As Hat1p/Hat2p has been implicated in a specific pattern of histone acetylation [30, 31], we next determined if the relevant role of Hat1p with regard to ORC function lies in its targeted acetylation of lysine residues 5 and 12 on histone H4. Specifically, we tested if non-acetylatable versions of histone H4 that mimic the deacetylated state would exhibit an enhanced phenotypic effect in conjunction with impaired ORC function (Figure 7). To this end, we created a strain expressing a conditional orc5-1 allele into cells in which the only source of histone H4 is a modified form of the histone H4 gene encoding residues 5 and 12 mutated to non-acetylable arginines. This variant was introduced along with other selected control histone mutants into either orc5-1 or orc5-1 hat1Δ strain backgrounds. When combined with orc5-1, non-acetylable H4 lysines 5 and 12 resulted in a noticeable reduction in growth (Figure 7A, Additional file 6), indicating that acetylation of these residues is indeed functionally relevant with respect to ORC activity. The lack of a complete epistatic effect (i.e. lysines 5 and 12 mutations conferred a stronger effect than hat1Δ) could reflect that histone H4 is a target of additional HATs such as Esa1p and Gcn5p [49, 50].
The N-terminal tails of histones H3 and H4 have been shown to be functionally interchangeable [51, 52]. To determine if a partly redundant HAT complements Hat1p with regard to ORC function, we first examined if the viability of orc-ts hat1Δ double mutant cells was dependent on native (i.e. fully functional) histone H3. Using a plasmid shuffle assay, we generated various combinations of N-terminal lysine substitution mutations in histones H3 in the orc5-1 and orc5-1 hat1Δ mutant backgrounds bearing an ADE2-marked wild-type histone construct (Figure 7B). The ability of the strains to tolerate the mutant histone derivatives was then scored visually based on the rate of loss of the ADE2 marked plasmid (sectoring) when no selection was applied. While a control TRP1-marked wild-type histone H3 plasmid readily supported growth in orc5-1 and orc5-1 hat1Δ strains, a histone H3 multi-point mutant lacking acetylatable N-terminal lysines (K9, 14, 18, 23, 27R) conferred a severe loss of viability as indicated by the exclusively white colony color (Figure 7C). Likewise, orc5-1 H3 (K9, 14, 18, 23, 27R) mutants were completely inviable when shifted at 30°C (see Additional file 6). Interestingly, mutations of lysine residues 14 and 23 in histone H3 enhanced the growth defects of orc5-1, and were incompatible with orc5-1 hat1Δ double mutants (Figure 7C, Additional file 6). Taken together, these results indicate that the functional overlap between H3 acetylation and Hat1p-mediated acetylation of H4 likely extends to the ORC-dependent process.
We next examined if hat1Δ combined with a non-acetylatable histone H3 mutant tail leads to a plasmid loss phenotype. Plasmid loss rates were measured in wild-type and hat1Δ strains using both a plasmid bearing the late replicating telomeric origin ARS120 (YCp120) or the early replicating origin ARSH4 (pRS316) in combination with either wild-type histone H3 or one of two multi-point mutants H3 (K9, 14, 18, 23, 27R) and H3 (K14, 23R). Elevated loss rates were observed for both plasmids in strains expressing a fully non-acetylatable H3 variant (Figure 7D), a situation that compromised growth of the cells (Additional file 6). However, no additional or synergistic defect was detected when the histone mutants were further combined with the hat1Δ mutation. Although this experiment did not establish a replication function for Hat1p at these two origins, it nevertheless showed that plasmid loss occurs below a threshold level of core histone acetylation. While this effect may be an indirect consequence of the histone H3 mutations on transcription or other pathways, it is also consistent with the view that multiple HAT activities converge to maintain genomic integrity.
The functional connection between ORC and histone H4 acetylation was further corroborated by examining the acetyltransferase activity of affinity purified ORC towards histone H4 in vitro (Figure 8). Comparison of the relative activity of ORC5-TAP purified from a wild-type strain with that obtained from a hat1Δ mutant demonstrated that nearly all histone H4 HAT activity (> 90%) was dependent on Hat1p (Figure 8A,B). Nevertheless, the residual H4 and some possible H3 acetylation in the absence of Hat1p suggests that at least one alternate HAT associates with ORC, albeit more weakly. Thus, whereas the H4-specific HAT activity by Hat1p associated with ORC established a direct connection between ORC function and histone acetylation, additional evidence also suggested the existence of one or more partly redundant, yet unidentified, HAT enzymes that seemingly function in conjunction with ORC.