Genome-Wide Mapping of Yeast Histone Chaperone Anti-Silencing Function 1 Reveals Its Role in Condensin Binding with Chromatin

Genome-wide participation and importance of the histone chaperone Asf1 (Anti-Silencing Function 1) in diverse DNA transactions like replication, repair, heterochromatic silencing and transcription are well documented. Yet its genome-wide targets have not been reported. Using ChIP-seq method, we found that yeast Asf1 associates with 590 unique targets including centromeres, telomeres and condensin-binding sites. It is found selectively on highly transcribed regions, which include replication fork pause sites. Asf1 preferentially associates with the genes transcribed by RNA polymerase (pol) III where its presence affects RNA production and replication-independent histone exchange. On pol II-transcribed genes, a negative correlation is found between Asf1 and nucleosome occupancy. It is not enriched on most of the reported sites of histone exchange or on the genes, which are misregulated in the asf1Δ cells. Interestingly, chromosome-wide distributions of Asf1 and one of the condensin subunits, Brn1 show a nearly identical pattern. Moreover, Brn1 shows reduced occupancy at various condensin-binding sites in asf1Δ cells. These results along with high association of Asf1 with heterochromatic centromeres and telomeres ascribe novel roles to Asf1 in condensin loading and chromatin dynamics.


Introduction
Organization of the eukaryotic genome into chromatin enables its compaction inside the cell nucleus and concomitant regulation of DNA-related processes [1]. Several mechanisms including histone modifications and ATP-dependent chromatin-remodeling, culminate into an altered chromatin structure [2,3], which renders the cis-acting sites on the DNA accessible to the trans-acting factors. They often involve localized chromatin assembly/disassembly via eviction/deposition of the histones by specific histone chaperones, which bind dimers of canonical or variant histones H2A/H2B or H3/H4 [4,5]. Histone chaperones also assist DNA transactions by exchanging old histones with new ones [6] and play important roles in replication and repair processes [7]. Anti-Silencing Function 1 (Asf1) is a highly conserved histone chaperone, which assists essentially all aspects of chromatin biology, including genome silencing by evicting/depositing H3/ H4 dimers [8]. It participates in regulating histone synthesis [9], maintains supply of histones, interacts with DNA replication machinery at active replication forks, helps progression of replication fork and maintains replisome integrity [10][11][12]. It cooperates with other chaperones like CAF1, HIR, FACT and several other histone-binding factors in replication-coupled or replication-independent chromatin assembly [13][14][15][16].
Yeast cells lacking ASF1 are sensitive to DNA damaging agents, as it plays important role in checkpoint signaling and genomic stability [13,17,18]. In addition to its role as chromatin assembly and disassembly factor, Asf1 is essentially required, in collaboration with histone acetyl-transferases (HATs), for the acetylation of lysine residues of histone H3 at positions 9 and 56 [19,20]. Several studies have shown that chromatin reassembly and deactivation of damage checkpoint after double-strand break (DSB) repair require cooperative acetylation of H3K56 by the histone acetylase Rtt109 and Asf1 [21][22][23][24]. Deletion of Asf1 causes up-regulation as well as down-regulation of genes suggesting its role in transcriptional repression as well as activation in a context-dependent manner [25][26][27][28]. Association of Asf1 with active genes promotes histone eviction and deposition, thus facilitating transcription elongation by the RNA polymerase (pol) II [29,30]. Studies in the budding yeast have shown replication-independent exchange of histone H3 throughout the genome [31][32][33][34]. Outside of the S phase, H3 exchange at promoters is higher than coding regions of the actively transcribed genes and requires Asf1 [32][33][34].
As evident from the above account, significant advances have been made in understanding the functional diversity of Asf1. However, despite genome-wide functional studies, its genomewide association map in the budding yeast is not reported yet. Our high-resolution mapping of Asf1 along the 16 chromosomes of the budding yeast revealed that Asf1 specifically associates with highly active pol II and pol III-transcribed loci, which are also known sites of replication fork arrest [35,36]. Its absence on most of the known sites of Asf1-dependent histone exchange or transcription, suggests its functions at a genomic locus do not depend on stable association with the site. The highest levels at centromeres as compared to telomeres and co-localization with condensin subunit Brn1 ascribe novel roles to Asf1.

Materials and Methods
Yeast strains, plasmids, growth conditions and antibodies List of the yeast strains and sequences of the primers used in this study are given under the Tables S1 and S2 respectively. Unless otherwise stated, yeast cells were grown in YEP+2% glucose (YPD) medium to an OD 600 of 0.7. Chromatin Immunoprecipitation (ChIP)-grade anti-HA was from Millipore; anti-Myc, anti-GFP and anti-HA antibodies were from Santa Cruz Biotechnology.

ChIP, Real Time PCR and ChIP-sequencing
Binding of a histone chaperones to chromatin can be ubiquitous. Difficulty in cross-linking Asf1 to chromatin [37] has been one of the reasons that its genome-wide map has been elusive. We used a yeast strain expressing ASF1-18myc from its genomic locus (gift from Kevin Struhl, [30]) and prepared ChIP DNA samples as previously described [38,39], except that cells were cross-linked with 1% formaldehyde for six hours [40] and chromatin was fragmented to mean size of 200-400 bp, using Bioruptor (Diagenode). The quality and size of the sheared DNA was checked by gel electrophoresis ( Figure S1A). Purified DNA was either used for quantifying the amplicons ( Figure S1B) by Real Time quantitative PCR or library preparation for sequencing with Illumina Genome Analyzer II, as per manufacturer's instructions. High-quality sequencing reads (phred score .20) were aligned, reporting unique and best alignment for each read, to budding yeast genome version sacCer 3. ChIP-seq data analysis was made according to our previously published method [41], as detailed further in the Supporting Information (Methods S1). HOMER package was used to make ready-to-visualize bedgraph files and peak calling. For each chromosomal feature, Asf1 tag count was calculated 1000 bp both upstream and downstream of genes and aligned according to TSS or TTS, using a 10 bp window and an overlap of 2 bp. Heat-maps were created using MultiExperiment Viewer (MeV).

Histone exchange assay
As detailed in the Supplementary Information, yeast cells carrying a plasmid that contains GAL1-driven H3-(HA) 3 gene were used for histone exchange assay [34]. Though the method used does not involve balanced expression of H3/H4, it does not influence the results obtained [33]. To calculate H3 exchange at Asf1-targeted loci, genome-wide histone H3 exchange data were extracted from a previous report [33] and analysed.

Asf1 associates with the highly transcribed genes and the heterochromatic regions
Since Asf1 is a general H3/H4 chaperone, we used peak calling criteria similar to nucleosome peak calling (peak size 150 bp and spacing of peaks by 300 bp), with a FDR of 0.1%. Analysis of the ChIP-sequencing data revealed only 590 chromosomal locations (Table S3) associated with Asf1 throughout the 16 chromosomes. The number of assigned Asf1 peaks on a chromosome shows a linear relationship with the chromosome length ( Figure 1A), indicating that Asf1 is uniformly distributed on all chromosomes. Interestingly, ,57% of all Asf1 targets are found in the inter-genic (between protein-coding genes) regions of diverse categories viz., transfer RNA (tRNA), small nuclear RNA (snRNA) and small nucleolar RNA (snoRNA) genes, autonomously replicating sequences (ARS) as well as centromeres ( Figure 1B).
Asf1 occupies the gene body of almost all (262 out of 275, 95%) tRNA genes ( Figure 1B) and other non-coding RNA genes (SNR6, SNR52, RPR1 and SCR1) transcribed by pol III ( Figure  S2) as well as 242 ORFs (open reading frames) and more than 50% of the snoRNA genes, transcribed by pol II. Majority of these ORFs (Table S3) belong to the highly expressed category in the genome-wide gene expression data [42]. Using MIPS Functional Catalogue annotation to find whether these ORFs represent genes of a special functional category, we found that the genes for ribogenesis contribute the major fraction of the ORFs targeted by Asf1 (Table S4). Pol III targets, which are known sites of replication fork stalling [36], are highly transcribed. Even other highly transcribed genes and centromeres targeted by Asf1 are known sites of replication fork stalling [35,43], where Asf1 is reported to help maintain replisome integrity [10,11].
The presence of Asf1 on the telomeres agrees with its previously established role in telomeric silencing [44]. Since we used only unique reads to map Asf1 targets, peak finding program did not pick all telomeres ( Figure 1B). In comparison, all the 16 centromeres in the budding yeast show a strong peak of Asf1 symmetrically centered at the middle of centromeric feature ( Figure 1C). A comparison of the average occupancy of Asf1 on highly occupied groups of chromatin features revealed the highest levels on the centromeres followed by tRNAs while the lower Asf1 levels on the ORFs and snoRNA genes show a broad distribution along the length of gene ( Figure 1D). As compared to mcm mutants, Asf1 has a minor role in chromosome maintenance [44], while no requirement of Asf1 for any centromeric function is reported in the budding yeast.

Asf1 association at the pol II ORFs matches their transcription activity
Our Asf1 enrichment data have under-representation of ORFs (p value,2.2e-16). We divided ,4000 pol-II transcribed ORFs into five categories, based on their transcriptional activity [42] and compared Asf1 occupancy with reported pol II profiles [33] on them. Pol II ORFs falling within 1 kb window of inter-genic (tRNA genes, snoRNAs genes, centromeres) and telomeric Asf1 targets were eliminated from this analysis to avoid contribution from these features. A heat map comparison of the occupancy data on 500 genes expressed at highest or lowest levels ( Figure S3) shows that Asf1 association increases with an increase in transcriptional activity. We aligned the Asf1 occupancy data on every ORF in each category according to its transcription start site (TSS) or transcription termination site (TTS). Average occupancies in a window of 1 kb upstream and downstream of TSS or TTS on genes belonging to a category are plotted together in the Figure 2A. Asf1 occupancy shows higher levels at highly transcribed ORFs and vice versa (Figure 2A), indicating a positive correlation with the transcriptional status of a given ORF. Accordingly, we did not find Asf1 on ARG1, a gene reported as repressed under growth on enriched media [24]. A comparison of the Asf1 occupancy profile with the average pol II occupancy for the same set of five categories of ORFs ( Figure 2B) revealed a high similarity between the two. Further, the peak heights of Asf1 and pol II at the 59 and 39 ends of genes match better for highly expressed genes (.16 mRNA copies/cell, Figures 2A and B).
Transcription-dependent binding of Asf1 to the coding region of some inducible genes was shown earlier to facilitate pol II elongation [30]. Above results find that even on the constitutive pol II-transcribed ORFs, higher the Asf1 association, higher is the transcription activity.

Asf1 and Nucleosome occupancies are inversely related on pol II-transcribed genes
Yeast pol II-transcribed genes generally have a nucleosome-free region (NFR) at the 59 end of genes, flanked by two positioned nucleosomes, 21 and +1 [45]. Another NFR at 39-end of genes coincides with the TTS of the genes. A comparison with the data on nucleosome occupancy data [46] revealed that relatively constant levels of Asf1 are found at the coding region while the position of the dip near TSS for ORFs expressed at moderate or low levels matches with the NFR in nucleosome profile (cf. Figures 2A and C). Importantly, a prominent peak of Asf1 is seen at the nucleosome preceding the TTS but not the downstream nucleosome on highly transcribed genes. Similarly, Asf1 occupancy towards the 59 half of the genes shows negative correlation with nucleosome occupancy (cf. Figures 2A and C). Our data agree   [42]. Legends box shows the color code, numbers indicate mRNA molecules/cell in the category. Vertical pink bar on the graphs represents the break in the middle. Panel A shows that the ChIP signals are well above the mock on highly transcribed genes. (D) Heat map depicting the Asf1 ChIP-Seq signal 500 bp upstream and downstream of the TSS (bent arrow) at snoRNA genes. Color gradient code is shown at the bottom. (E) Asf1 association at pol II-transcribed snoRNA genes, 1 kb upstream and 1 kb downstream of the TSS is compared with pol II and nucleosome occupancy profiles at the gene loci. Averages of Asf1 ChIP-and mock-Seq signals for 39 out of 77 genes are plotted. The vertical arrow denotes the gene 39-end. (F) Venn intersections of Asf1-occupied (this study) genes with those misregulated in asf1D cells [28]. P value of the overlap is 3.9610 211 (significant overlap). (G) Venn intersections of Asf1-occupied (this study) genes with those misregulated in asf1D cells [25] or after Asf1 depletion [25], with P values 0.118 and 0.42 respectively (insignificant overlaps). doi:10.1371/journal.pone.0108652.g002 with previous findings that histones and pol II occupancies are inversely related and Asf1 assists pol II elongation by evicting histones from gene body and ends [30,[32][33][34]. Similar correlation of the Asf1, pol II and nucleosome occupancy is found on another pol II-transcribed gene class, the snoRNA genes. Asf1 shows a broad peak covering these genes ( Figure 2D). Its distribution matches well with the pol II profile, while showing a negative correlation with nucleosome occupancy profile ( Figure 2E).

Asf1 binding is not required for its influence on the global gene expression
Global gene expression studies in the budding yeast have revealed transcriptional misregulation of many genes upon deletion of Asf1 [25,28]. We compared the list of Asf1-occupied ORFs from our ChIP-Seq data with the list of the genes misregulated upon Asf1 deletion in both the reports (Figures 2F  and G). In the two lists of misregulated genes in asf1D cells (color coded pink and green), only 89 genes were common, while only 26 genes were found misregulated ( Figure 2G) under the two conditions of Asf1 absence; Asf1 deletion or conditional depletion [25]. In all, we found only a small number (,15%) of misregulated genes occupied by Asf1. Asf1 occupies 43 out of 324 [28] and only 16 out of 524 [25] such genes ( Figures 2F and 2G). Out of the 95 genes misregulated when Asf1 is depleted [25], only 5 show binding of Asf1 in this study ( Figure 2G). Asf1-occupied genes (Table S5) and the misregulated genes in asf1D cells [25] do not show any functional overlap also. This analysis shows that Asf1 association with most of the genes misregulated in its absence is not stable. Accordingly, we did not find Asf1 on HUG1 and RNR3 loci, earlier reported to show an increase in the asf1DZ cells and a requirement of Asf1-dependent H3K56 acetylation for their transcriptional regulation [28].

Asf1 deletion increases RNA levels from the pol III-transcribed genes
Other than centromeres, Asf1 occupancy on tRNA genes (268 out of 590, ,50%) is highest of all targets ( Figures 1C, 1D, 3A). Using previously published data [41], we compared average occupancy profile of Asf1, pol III and nucleosomes on 274 tRNA genes of yeast ( Figure 3B). Unlike nucleosomes, Asf1 occupies the entire coding region of tRNA genes ( Figure 3B). It shows a distribution nearly identical to that of pol III, with a peak centered at TSS and positive correlation coefficient 0.54 (R 2 0.29) between the two. Validation of the ChIP-Seq data for Asf1 occupancy at different pol III-transcribed genes by ChIP-qPCR shows high occupancy of Asf1 ( Figures 3C,D) on tRNA genes and centromeric DNA, as compared to other ORF-free intergenic regions ( Figure 3C). On non-tRNA genes, Asf1 levels at the 59 ends are found higher than at the 39 ends ( Figure 3D).
Asf1 occupies the nucleosome-free bodies of the pol IIItranscribed genes, which show transcribed regions are nucleosome-free while nucleosomes are found on both 59-and 39flanking regions of the genes ( Figure S4A, [41]). Its association with pol III-transcribed genes and an increase in levels of SNR52 and some tRNAs in asf1D cells have been suggested as related to transcription activity [30] and RNA stability [28]. Using Real Time qPCR method, we found 1.5-2.5 fold increase in RNA levels from other tRNA (representing five different tRNA families) as well as non-tRNA (SNR6 and SNR52) pol III-transcribed genes, in asf1D cells as compared to the wild-type cells, although RPR1 shows only marginal increase ( Figure S4B).
A defective nucleosome assembly but no change in the nucleosome density or the core histone levels in asf1D cells were reported earlier [14,18]. The increase of RNA and a lower nucleosome occupancy on all (except SNR6) Asf1 targets ( Figure 3) in asf1D cells (Figures S4C and D) in this study, imply that the higher RNA levels in asf1D cells could be related to the lower nucleosome occupancy on transcribed regions. Therefore, the relationship of Asf1 with transcription and nucleosome occupancy on pol III and pol II-transcribed genes may be different.

Asf1 influences histone exchange at the 39-end of pol III-transcribed genes
Asf1 has been implicated in genome-wide replication-independent H3 exchange at ,2000 pol II ORFs, tRNA genes and snoRNAs genes, although the direct binding of Asf1 to these features has not been shown [33,34]. A comparison of genomewide exchange data [33] with our Asf1 occupancy data revealed that despite being occupied by Asf1, one-fourth of Asf1-occupied pol II genes and one-third of the tRNA genes do not show transcription-dependent H3/H4 exchange. Our measurements at some of these genes showed an Asf1-dependent H3/H4 exchange on the promoter and 39 end of the positive control gene, GAL7 (Figure 4), but not on any of the SNR6 gene regions ( Figure S4E). The level of H3/H4 exchange at the RPR1 downstream region, where Asf1 and nucleosome occupancies are very low, shows a very small (,1.6 fold) decrease in asf1D cells (cf. Figures 3D, S4C,  4A and B). Out of the six selected regions on SCR1 ( Figure S1B), high exchange rate is seen towards 39end and on the gene body ( Figure S4E). Similarly, high exchange rate on the gene bodies of SNR52, RPR1 and two of tested tRNA genes could be seen ( Figures 4A and S4E). In contrast, despite the presence of Asf1 at the 59 end and upstream regions of SCR1, RPR1 and whole of SNR6 gene ( Figure 3D), little exchange of H3/H4 is seen on these regions ( Figures 4A and S4E). Thus, pol III transcription on these genes is accompanied by rapid histone exchange at the 39 end and transcribed gene regions. This exchange is Asf1-dependent as it is abolished in asf1D cells (cf. Figures 4A and B) and Asf1 is found on all these locations (Figures 3C and D). Nevertheless, the correlation between Asf1 occupancy and histone exchange on different pol III-transcribed genes is not direct (cf. Figures 3C, 3D, 4A and S4E), similar to a lack of correlation between global H3 exchange [33] and our Asf1 occupancy data.

Asf1 targets are co-occupied by the condensin complex
We noticed that Asf1 targets found in this study matched with the reported genome-wide targets of condensin [47] including tRNA, snoRNA and ribosomal protein encoding genes. The linear distribution of assigned Asf1 peaks along the length of all chromosomes ( Figure 1A) is also similar to that for condensin in the earlier report [47]. Therefore, we calculated the normalized occupancy [47] of Brn1, a condensin subunit, at the assigned Asf1 peaks and found that the majority of Asf1 targets (493 out of 607, ,81%), are co-occupied by Brn1 with a positive correlation coefficient 0.52 (R 2 = 0.27). A comparison of the peak distributions and a close up view of both the arms of chromosome V show that Asf1 and Brn1 peak distributions are nearly identical ( Figure 5A, cf. two panels). ChIP-qPCR measurements show reduced Brn1 levels on some of its tested target sites in asf1D cells ( Figure 5B) although Brn1 levels do not change in the deletion mutant ( Figure 5C). Similar to condensin mutants, asf1D yeast cells, though viable, accumulate in G2/M phase of cell cycle [13,17]. In the context of the known functions of both in the chromatin structure and higher order chromatin organization, the highly similar Asf1 and Brn1 occupancy profiles along with decreased Brn1 occupancy upon Asf1 deletion imply a facilitatory role of Asf1 in condensin loading on the chromosomes.

Discussion
This study brings forth a functional overlap between transcription, replication, and chromosome architecture mediated by Asf1. Asf1 binding to chromatin has been envisaged as non-specific [24,28]. Its binding to only a fraction of the pol II-transcribed genes suggests that its reported roles in genome-wide histone exchange and transcription regulation may be indirect or redundant with other chaperones. However, the absence of physical association with most of the functional targets may not be unusual. It has also been observed in earlier genome-wide studies on some other proteins involved in regulating chromatin structure and dynamics, suggesting a transient association with chromatin [48][49][50][51].
Highly transcribed regions are reported ''hyper-ChIPable'' where several unrelated and biologically non-meaningful proteins also show enrichment [52]. However, literature has ample research evidence for functional relevance of Asf1 binding to highly transcribed loci [30,32,33]. Accordingly, high association of Asf1 with such regions, found in our ChIP-seq data may not be  artefactual. Asf1-associated genomic regions include both highly transcribed features and heterochromatic regions. Asf1 role in maintaining replisome integrity [11] suggests an increased dwelling time at highly transcribed genes, which are also reported to be the replication fork pause sites. This comparatively nontransient association at certain sites may also be the reason that higher Asf1 levels are found at replication fork pause sites like tRNA genes (highly transcribed) and centromeres (non-transcribed).
It was earlier demonstrated that high histone exchange seen at NFRs represents a continuous process that disrupts nucleosomes and maintains DNA accessibility [53]. This may be the reason that NFR on tRNA genes is actively maintained by participation of several ATP-dependent chromatin remodelers [41]. The Asf1dependent histone exchange at the 39 end of the pol IIItranscribed genes may also be related to the higher nucleosome dynamics at the downstream end for these genes, reported earlier [41], probably to keep the gene terminator accessible. However, further experiments are required to understand how Asf1 regulates pol III transcription.
Asf1 is required for H3K56 acetylation by Rtt109 [20] and in its absence, eviction of unacetylated nucleosomes is reduced [54]. On the pol II-transcribed PHO5 gene, higher H3 levels were found in asf1D cells [37] and similar to HUG1 and RNR3 loci [28], its activation required H3K56 acetylation [54], again suggesting that Asf1 regulates genes due to its requirement for H3K56 acetylation, which in turn, promotes nucleosome eviction by Asf1. A strong correlation has been reported between H3 exchange and H3K56 acetylation, genome-wide (r = 0.865) as well as on tRNA genes (correlation coefficient 0.844; R 2 = 0.712) in the same data [33]. Therefore, role of Asf1 in histone-exchange may be facilitatory, locus-specific and indirect. In agreement, a recent study has computed that the histone H3 exchange pattern in yeast is specific for each gene and does not depend on transcription rate of the gene [55].
Further probing of the functional and physiological implications of its high association with centromeres and genome-wide colocalization with condensin may reveal novel functions of Asf1. An earlier study [47] had reported a partial correlation of condensin occupancy with its other possible loaders on the chromatin. In this study, loss of Brn1 from its binding sites in Asf1 deletion cells suggests a requirement for Asf1 in condensin loading. A physical association between the two is a possibility, which has not been tested or reported yet. Cohesin loader Scc2/4 is also required for loading condensin but does not interact with it [47] while transcription factor of RNA polymerase III, TFIIIC and condensin co-localize to tRNA genes but not to centromeres [47]. It was also found in the same study that though TFIIIC binding sites may be the minimum condensin-binding regions, deletion of tRNA gene sequence does not abolish condensin binding [47], suggesting involvement of other facilitatory factors for condensin loading. In light of these observations, Asf1 could be this missing factor, as a dramatic fall in Brn1 occupancy is seen on all tested loci in asf1D cells. Since Asf1 is a highly conserved H3/ H4 chaperone, its targets in higher eukaryotes may be similar. The outcomes of this study also propose that identifying genome-wide targets of other histone chaperones can resolve known redundancy and specific or novel functions of different histone chaperones.   [41] were used to generate the heat map of nucleosome occupancy 1 kb upstream and 1 kb downstream of the TSS (bent, green arrow) of the four non-tRNA pol III-transcribed genes. Genes reside in the NFR on the right side of their TSS. Gene body is denoted by grey bars. Color code for the occupancy gradient is shown at the bottom of the panel. (B) RNA levels of some of the pol-III transcribed genes in the asf1D cells, as compared to RNA levels in the wild-type cells, set to 1. SCR1 was used as an internal control. The averages of five independent experiments with error bars are shown. (C) and (D) show results of nucleosome occupancy measured on various pol III-transcribed genes in wild-type and asf1D cells. Mononucleosomal DNA isolated from chromatin digested with micrococcal nuclease in situ was used for qPCR. Averages from three independent estimations with error bars are shown.

Supporting Information
Amplicons A-F for SCR1 regions (panel D) are described in the Figure S1. (E) Replication-independent histone H3 exchange occurs in the coding region of Pol III-transcribed genes. Timecourse analysis of replication-independent H3 exchange [34] on some of the pol III-transcribed genes is shown. ChIP-qPCR analysis was used to follow the H3 exchange at various genes. (PDF) Methods S1 This file contains details of the methods followed and the supplementary references cited in the Supporting Information files. (PDF)