Yeast heterochromatin regulators Sir2 and Sir3 act directly at euchromatic DNA replication origins

Most active DNA replication origins are found within euchromatin, while origins within heterochromatin are often inactive or inhibited. In yeast, origin activity within heterochromatin is negatively controlled by the histone H4K16 deacetylase, Sir2, and at some heterochromatic loci also by the nucleosome binding protein, Sir3. The prevailing view has been that direct functions of Sir2 and Sir3 are confined to heterochromatin. However, growth defects in yeast mutants compromised for loading the MCM helicase, such as cdc6-4, are suppressed by deletion of either SIR2 or SIR3. While these and other observations indicate that SIR2,3 can have a negative impact on at least some euchromatic origins, the genomic scale of this effect was unknown. It was also unknown whether this suppression resulted from direct functions of Sir2,3 within euchromatin, or was an indirect effect of their previously established roles within heterochromatin. Using MCM ChIP-Seq, we show that a SIR2 deletion rescued MCM complex loading at ~80% of euchromatic origins in cdc6-4 cells. Therefore, Sir2 exhibited a pervasive effect at the majority of euchromatic origins. Using MNase-H4K16ac ChIP-Seq, we show that origin-adjacent nucleosomes were depleted for H4K16 acetylation in a SIR2-dependent manner in wild type (i.e. CDC6) cells. In addition, we present evidence that both Sir2 and Sir3 bound to nucleosomes adjacent to euchromatic origins. The relative levels of each of these molecular hallmarks of yeast heterochromatin–SIR2-dependent H4K16 hypoacetylation, Sir2, and Sir3 –correlated with how strongly a SIR2 deletion suppressed the MCM loading defect in cdc6-4 cells. Finally, a screen for histone H3 and H4 mutants that could suppress the cdc6-4 growth defect identified amino acids that map to a surface of the nucleosome important for Sir3 binding. We conclude that heterochromatin proteins directly modify the local chromatin environment of euchromatic DNA replication origins.


Introduction
In eukaryotic cells, efficient genome duplication requires the function of multiple DNA sir2Δ or sir3Δ is not easily explained by defects in classic Sir-heterochromatic gene 119 silencing of known loci because Sir2, Sir3 and Sir4 are each equally essential for HM-120 and telomere silencing and only Sir2 is required for rDNA silencing. In addition, multiple 121 sir2 alleles specifically defective in rDNA silencing do not suppress cdc6-4 ts lethality,

122
indicating that a loss of rDNA silencing is not sufficient to explain SIR2's negative effect 123 on euchromatic origins [29]. These findings support a distinct, rDNA-and classic 124 heterochromatin-independent function of SIR2 on euchromatic origins. Importantly, a 125 sir2 catalytic mutant or a histone mutant that converts H4K16 into a residue that mimics 126 acetylated H4K16 (H4K16Q) suppresses cdc6-4, indicating that the Sir2 deacetylase cdc6-4 sir2Δ mutant cells showed MCM association, albeit to varying degrees, at most 167 origins. (Figure 1B, C, S1 Figure). Therefore, deletion of SIR2 rescued origin-specific 168 association of MCM at the majority of chromosomal origins in cdc6-4 mutant yeast, 169 consistent with SIR2 have a pervasive negative effect at the majority of euchromatic 170 origins.

171
A previous study screened for origins on chromosomes III and VI (Figure 1 ChIP-Seq signal in wild type cells, and because we eliminated all origins that exist within 178 heterochromatic domains for deeper analyses, as described below. The plasmid-based 179 study suggested that at least 20% of yeast origins were likely to be SIR2-responsive 180 [23]. In contrast, the MCM-ChIP-Seq revealed that ~80% of origins showed MCM 181 binding in cdc6-4 sir2Δ cells ( Figure 1B, 1C and S1 Figure). We note that the plasmid-182 based assay demanded that origin function was rescued to a level in cdc6-4 sir2Δ cells cells ( Figure 2D). Thus, reducing rDNA copy number was neither necessary nor 217 sufficient to explain SIR-mediated suppression of cdc6-4. Cells harboring a reduction in 218 rDNA copy number in cdc6-4 sir2Δ populations likely arise because loss of Sir2 219 increases recombination frequency within the rDNA array, and rDNA copy numbers are 220 reduced over time in many mutants compromised for replication initiation [35,41,42]. between origin and non-origin loci [43,44], nucleosome occupancy around the ORC site 243 was assessed using the MNase-ChIP-Seq data generated from this more recent study 244 [32]. The results from our analyses confirmed the conclusion that origin-adjacent 245 nucleosomes show high occupancy at more defined positions around the nucleosome-246 depleted ORC site compared to the control non-origin adjacent nucleosomes ( Figure   247 3C). Next, the H4K16ac status of nucleosomes surrounding the ORC site for both 248 groups was determined and normalized to the H4K16ac status from a collection of 249 nucleosomes present in a distinct collection of euchromatin intergenic regions that 250 contained neither origins nor matches to the ORC site (n=239). This analysis revealed 251 that nucleosomes adjacent to origins were depleted for H4K16ac but that nucleosomes 252 adjacent to the control non-origin nucleosomes were not ( Figure 3D). H3K9ac is also a 253 potential substrate for Sir2 but, in contrast to a H4K16Q substitution, an H3K9Q 254 substitution fails to suppress cdc6-4 [19,23]. Also in contrast to H4K16ac, H3K9ac was 255 depleted similarly from nucleosomes adjacent to origin and non-origin control loci 256 ( Figure 3D). Thus, H4K16ac was distinct among nucleosome acetylation marks in

258
If depletion of H4K16ac on origin-adjacent nucleosomes was relevant to SIR2-259 dependent inhibition of MCM loading in cdc6-4 cells, then the origins most responsive to 260 deletion of SIR2 might be expected to show the greatest depletion of H4K16ac. We 261 defined origin SIR2-responsiveness as the ratio of the MCM ChIP-Seq signal in cdc6-4 262 sir2Δ cells to that in sir2Δ cells; the most SIR2-responsive origins generated ratios near   Figure 3D, although we note the 278 relative level of origin-specific H4K16ac depletion was slightly greater in these 279 experiments, possibly due to differences in strain backgrounds or growth conditions.

280
Regardless, the key result was that, in contrast to wild type, depletion of H4K16ac was 281 lost from origin-adjacent nucleosomes in the sir2Δ cells, while the behavior of non-origin 282 ORC site control nucleosomes was unchanged. Importantly, these effects also 283 correlated with SIR2 responsiveness ( Figure 4B). Thus, origin-specific depletion of 284 H4K16ac nucleosomes at euchromatic origins required SIR2.

286
Sir2 and Sir3 were detected at origins. Sir2 and Sir3 are physical components of 287 yeast heterochromatin that have been detected at rDNA (Sir2) and telomeres and HM 288 loci (Sir2 and Sir3) by ChIP experiments in multiple studies (reviewed in [15]). However, 289 neither protein has been reported to associate with euchromatic origins [45]. The level of 290 H4K16 hypoacetylation at euchromatic origins was estimated to be between 25-40% of 291 that detected at heterochromatic origins for the +1 nucleosome (S4 Figure). This high-resolution ChIP-Seq Sir2 and Sir3 data sets and excluded all nucleosomes from 297 known heterochromatin domains for normalization, as above, so that the baseline would 298 not be affected by the extensive amount of Sir2 and Sir3 binding known to occur at these 299 domains [34,47]. This analysis detected Sir2 and Sir3 ChIP-Seq signals on nucleosomes 300 adjacent to origins but not to non-origin controls ( Figure 5A). Because the Sir3 data was 301 generated from a high-resolution MNase ChIP-Seq experiment, we also examined these 302 data at nucleotide resolution across the 1201 bp origin and non-origin fragments 303 described in Figure 3B, normalizing the number of reads that contained a given 304 nucleotide in the ChIP sample to the total number of reads that contained that nucleotide 305 in the starting material, as in [48] ( Figure 5B). These analyses revealed a Sir3 ChIP-Seq 306 signal that was strongest at the most proximal origin-adjacent nucleosomes (-1 and +1) 307 but also detectable above baseline at the -3, -2, +2 and +3 nucleosomes ( Figure 5C).

308
As was true for the depletion of H4K16ac, the Sir3 signals correlated with the degree of 309 SIR2-responsiveness as defined in Figure  313 hypoacetylation -could be detected at euchromatin origins but not at euchromatic non-314 origin controls. In addition, the relative levels of each of these molecular hallmarks 315 correlated with how strongly a SIR2 deletion restored an MCM ChIP signal to origins in identified residues important for Sir3 binding. Histones H3 and H4 form a tetramer 320 that binds both to dsDNA and to two dimers of histones H2A and H2B to form the 321 nucleosome. Previous work showed that many conserved residues on the histone H3 322 and H4 N-terminal tails as well as within the globular core region ("core-modifiable" 323 residues) can be post-translationally modified. As discussed above, an H4K16Q 324 substitution that should mimic the acetylated form of this residue (i.e. the form that could 325 theoretically mimic the effect of a sir2Δ), suppressed the temperature-sensitive growth 326 defect of cdc6-4 [23]. To test whether we could identify additional alleles with similar

350
The four residues whose substitutions suppressed cdc6-4 ts lethality (H3E73, H3K79, 351 H3T80, and H4K79) clustered to a small patch on the nucleosome surface critical for

374
The ability of sir2Δ to enhance MCM complex loading at most euchromatic origins in 375 cdc6-4 cells was striking, but it did not indicate whether this effect was due to Sir2 acting 376 directly within euchromatin in general or at euchromatic origins in particular. Indeed, 377 recent reports establish that SIR2 can alter origin function within euchromatin in yeast 378 indirectly because of its function in rDNA heterochromatin formation that suppresses 379 many of the rDNA repeat origins [22,25]. However, analyses of the rDNA locus 380 presented here suggested that a similar rDNA-mediated mechanism could not explain 381 why sir2Δ or sir3Δ suppressed cdc6-4 so robustly. Instead, the data provided compelling 382 evidence that Sir2 functioned directly at euchromatic origins, and that it was this direct 383 function of Sir2 that made yeast so vulnerable to defects in the MCM complex loading 384 reaction caused by cdc6-4. First, relative hypoacetylation of H4K16 was observed for 385 nucleosomes immediately adjacent to euchromatic origins but not for nucleosomes 386 adjacent to non-origin control loci. Importantly, these nucleosome states were observed

430
sensitivity and origin-specific MCM loading, indicated that H4K16 acetylation might be 431 unique among histone tail lysines in being particularly relevant to origin licensing [23,29].

432
This study revealed that H4K16 was also unique among histone H3 and H4 tail lysines

528
For the downstream analyses (Figure 1 and S1), all data for a given strain were 529 combined as each replicate generated virtually identical reads. However, for the cdc6-4 530 strain, which produced extremely low signal-to-noise data, only two technical replicates 531 from a single biological sample were used. For downstream analyses using MochiView 532 the CG1 format data was collated into 25 bp bins (Figure 1). The combined data from 533 each strain was run through CisGenome to identify the top 1000 peaks, divided into 100 534 peak bins and the percent of peaks that overlapped with at least one ARS (ARS 535 coordinates used from oriDB) determined. The oriDB lists 410 yeast origins as 536 confirmed. Based on these data and the above analyses we chose to examine and 537 compare the top 400 peaks from each sample for our initial analyses ( Figure S1). The 538 raw data for CDC6 SIR2, sir2Δ and cdc6-4 sir2Δ were normalized and scaled to the 539 same cdc6-4 data for to generate the final MCM signal values for downstream analyses.