Previous studies in Saccharomyces cerevisiae established that depletion of histone H4 results in the genome-wide transcriptional de-repression of hundreds of genes. To probe the mechanism of this transcriptional de-repression, we depleted nucleosomes in vivo by conditional repression of histone H3 transcription. We then measured the resulting changes in transcription by RNA–seq and in chromatin organization by MNase–seq. This experiment also bears on the degree to which trans-acting factors and DNA–encoded elements affect nucleosome position and occupancy in vivo. We identified ∼60,000 nucleosomes genome wide, and we classified ∼2,000 as having preferentially reduced occupancy following H3 depletion and ∼350 as being preferentially retained. We found that the in vivo influence of DNA sequences that favor or disfavor nucleosome occupancy increases following histone H3 depletion, demonstrating that nucleosome density contributes to moderating the influence of DNA sequence on nucleosome formation in vivo. To identify factors important for influencing nucleosome occupancy and position, we compared our data to 40 existing whole-genome data sets. Factors associated with promoters, such as histone acetylation and H2A.z incorporation, were enriched at sites of nucleosome loss. Nucleosome retention was linked to stabilizing marks such as H3K36me2. Notably, the chromatin remodeler Isw2 was uniquely associated with retained occupancy and altered positioning, consistent with Isw2 stabilizing histone–DNA contacts and centering nucleosomes on available DNA in vivo. RNA–seq revealed a greater number of de-repressed genes (∼2,500) than previous studies, and these genes exhibited reduced nucleosome occupancy in their promoters. In summary, we identify factors likely to influence nucleosome stability under normal growth conditions and the specific genomic locations at which they act. We find that DNA–encoded nucleosome stability and chromatin composition dictate which nucleosomes will be lost under conditions of limiting histone protein and that this, in turn, governs which genes are susceptible to a loss of regulatory fidelity.
Twenty-five years ago, Michael Grunstein's laboratory began a series of experiments in which histones H2B or H4 were depleted in vivo in S. cerevisiae (yeast). After histone gene silencing, the yeast cells complete a single round of DNA replication, reducing their histone-DNA ratio by a factor of approximately two, and enter cell-cycle arrest. A number of conditionally expressed genes, including PHO5, GAL1, CYC1, CUP1, and HIS3, were reported to be transcriptionally de-repressed following histone depletion –. In 1999, Richard Young's laboratory revisited the transcriptional effects of H4 depletion, this time on a genomic scale using microarrays. Roughly 15% (888) of yeast genes were de-repressed 3-fold or greater, and another 10% of genes (569) were repressed at least 3-fold following H4 depletion . Southern blots revealed changes in chromatin structure upstream of the PHO5 promoter during H4 depletion, but at the time these experiments were performed, high-resolution genomic methods were not available to measure the widespread chromatin changes hypothesized to underpin the observed transcriptional changes. We therefore revisited these classic experiments using RNA-seq and MNase digestion coupled with next-generation sequencing (MNase-seq) to monitor RNA and chromatin changes after histone depletion. These experiments also bear on the recent debate regarding the degree to which DNA-encoded elements or other cellular factors control nucleosome position and occupancy –. The raw RNA-seq and MNase-seq data are available at GEO accession number GSE29294.
H3 depletion results in changes in nucleosome occupancy throughout the genome
Two genes encode histone H3 in wildtype S. cerevisiae, HHT1 and HHT2. We obtained a strain in which HHT1 had been deleted and HHT2 had been placed under control of the GAL1 promoter (Figure 1A) . When grown in galactose, this “H3 shutoff” strain (DCB200.1) grows similarly to wildtype yeast (YEF473A), whereas cultivation in dextrose results in growth arrest in the G2/M phase of the cell cycle after a single round of DNA synthesis as large-budded cells (Figure S1A, S1B) . After 3 hours in dextrose, RNA-seq shows that HHT2 transcription is reduced to near zero, and Western blot analysis shows that histone H3 protein levels have dropped by a factor of two, as expected (Figure S1C, S1D). This evidence, coupled with previous cytological evidence using this strain , shows that the switch to dextrose successfully depleted histone H3.
We mapped nucleosome position and relative nucleosome occupancy using MNase-seq in the wildtype and H3 shutoff strains at 0 and 3 hours after transitioning the cells from galactose media to dextrose media (Figure S2A–S2C). While previous studies examined transcriptional changes 6 hours after the shift , we chose 3 hours for our study based on our characterization of the rate of histone depletion (Figure S1A–S1D) and to decrease the time the cells were under stress. Consistent with previous studies , , the chromatin was generally more sensitive to MNase digestion following H3 depletion, resulting in increased background smearing (Table S1, p<0.05; Figure S2D–S2G). Mono-nucleosome bands were extracted from the gel and sequenced, and the resulting reads were mapped back to the genome (Materials and Methods).
For the four possible strain and growth combinations, we aligned all of the genes by their +1 nucleosome relative to the transcription start site (TSS)  and calculated the smoothed dyad density across the gene body (Materials and Methods). The genes were then sorted by length from shortest to longest (Figure 1B). This plot reveals several key features of our dataset. First, the nucleosome organization of the wildtype strain in galactose (0 hours) is very similar to the nucleosome organization of the H3 shutoff strain in galactose (0 hours). Second, the nucleosome organization of the wildtype strain in galactose is very similar to the nucleosome organization of the wildtype strain in dextrose (3 hours). Third and most importantly, there are dramatic changes in nucleosome organization in the H3 shutoff strain between galactose (normal histone levels) and dextrose (depleted levels of H3). As histone availability decreases, the overall nucleosome positioning at any given position is weaker. Nucleosome positioning is still evident at the +1 and +2 nucleosomes but then decays rapidly, leading to a nearly complete loss of visible positioning until the gene's transcription termination site (TTS) is reached. There, one can still see a nucleosome at the −1 position relative to the TTS, although it is much more weakly positioned than in the wildtype strain.
To assess the overall similarity of our experimental replicates, we compared the overall nucleosome occupancy for each base pair among the replicates and found the experiments to be highly correlated (Figure S2H–S2K). As expected from the positioning data, the wildtype and H3 shutoff strains had very highly correlated nucleosome occupancy profiles when grown in galactose. When both strains were grown in dextrose, which results in histone H3 depletion in the shutoff strain, the correlation between the nucleosome occupancy profiles of the wildtype and H3 shutoff strains was lower, as expected (Figure 1C).
Defined subsets of nucleosomes are susceptible to changes in occupancy and/or position following H3 depletion
To identify changes in specific nucleosomes, we determined the position and occupancy of individual nucleosomes in each replicate at each time point (Materials and Methods). We used a slightly modified version of previously proposed definitions for measuring nucleosome positions and occupancies  that relies upon the midpoint of the paired-end reads or the center of the extended single-end reads (the “nucleosome dyad”) when examining nucleosome position and occupancy (Materials and Methods; Figure 2A). This method reduces positional noise that might be introduced from MNase producing fragments of varying lengths.
Changes in each nucleosome's position between the 0- and 3-hour time points were assessed using the t-test on the distribution of the read centers. The nucleosome position was then classified as being “altered” or “unchanged” based on the resulting p-value. Nucleosomes with borderline p-values were placed in a “no call” category and not used for the downstream analyses (Materials and Methods; Figure S3). The use of the t-test allowed us to account statistically for the “fuzziness” of each individual nucleosome. We note that p-values represent the confidence that an observation differs from the null expectation, not the degree to which an observation is biologically relevant.
Changes in each nucleosome's occupancy were assayed using the binomial distribution test on the number of read centers that fell within a 100-bp window centered on the nucleosome's consensus dyad. We stress that our experimental approach did not allow absolute nucleosome occupancy calculations, so the occupancy at a given location is in effect measured relative to all other nucleosomes in the same experiment. Nucleosomes were classified as “preferentially reduced” (decreased occupancy relative to other nucleosomes), “unchanged” (similar occupancy relative to others), or “preferentially retained” (increased relative occupancy). Nucleosomes with borderline p-values were placed in a separate “no call” category, similar to the position classification (Figure S3).
Nucleosomes that were classified identically in two or more replicates were selected as being “reliably characterized” (Materials and Methods; Figure S3). By comparing nucleosomes that were reliably characterized in both the wildtype and H3 shutoff experiment, we were able to remove nucleosomes for which the behavior could be attributed to the change in carbon source (Figure 2B and Figure S3). All further analysis was performed only with the “H3-depletion dependent” set of nucleosomes. The source code for all of the custom nucleosome calling tools can be found at http://sourceforge.net/p/callnucleosomes.
As an illustration of our nucleosome calls, we produced a graphical representation of the nucleosome structure surrounding PHO5, which had been previously shown to have altered chromatin and increased transcription following H3 depletion , and IDP2, a gene with increased transcription following H3 depletion (Figure S4 and Figure S5). After H3 shutoff, a new nucleosome configuration was observed upstream of both genes.
Preferentially lost nucleosomes tend to occur at gene promoters and to be acetylated
We sought to understand the factors that were responsible for the nucleosomes that were preferentially lost or retained or that responded to histone depletion by altering their positions. We compared each of the five classes of nucleosome changes (“position altered”, “position unchanged”, “occupancy reduced”, “occupancy unchanged”, and “occupancy retained”) to genome annotations and to previously published ChIP-chip or ChIP-seq experiments, including in vivo maps of histones, histone modifications, histone variants, and DNA-associated proteins (Figure 3 and Figure S6 and Table S1). Nearly all of the published data we used was obtained under standard growth conditions in dextrose, which matches the growth conditions of our strains upon the initiation of the H3 transcriptional shutoff (Materials and Methods). Therefore, the published datasets offer a reasonable representation of the chromatin landscape during histone depletion. We note that while our data has single-base pair technical resolution, the resolution of the published datasets varies (see Figure 3). Therefore statements about nucleosomes lacking or harboring a specific histone modification or other property should be interpreted with this in mind.
Upon histone H3 depletion, reduced-occupancy nucleosomes tended to occur in promoter regions, while nucleosomes in the gene bodies were associated with unchanged occupancy (Figure 3A). Consistent with this, a mark associated with transcriptional elongation, H3K36me2, was strongly enriched on nucleosomes classified as “position unchanged” or “occupancy unchanged” (Figure 3B). H3K36me2 has been proposed to contribute to increased nucleosome stability in the wake of RNA Polymerase II (RNA Pol II) transcription via the recruitment of the histone deacetylase (HDAC) Rpd3 , .
Consistent with nucleosome destabilization due to histone acetylation, “altered position” or “reduced occupancy” nucleosomes were enriched for 14 of the 16 of the histone acetylation marks we examined (p<0.01) (Figure 3B). Acetylated nucleosomes are generally less stable than those that are not acetylated –, so it follows that acetylated nucleosomes would be more susceptible to changes following H3 depletion. We also found that nucleosomes with reduced occupancy or altered position tended to exhibit high replication-independent turnover rates (Figure 3E and Figure S7) , suggesting that the more dynamic histone-DNA interactions at these locations result in a relative loss under histone-limiting conditions.
The conserved co-repressor Tup1 is associated with stable nucleosome positions and occupancy, while Bdf1, Rap1, and Reb1 binding are associated with nucleosome loss
We compared the H3-depletion responsive nucleosomes to previously published genome-wide data for several transcription factors, including Bdf1, Rap1, Reb1, and Tup1 , . Sites of Bdf1, Rap1, and Reb1 binding were all associated with reduced nucleosome occupancy following H3 depletion, which is consistent with the known ability of these proteins to displace nucleosomes (Figure 3C).
In contrast, Tup1 is a known transcriptional repressor that can recruit histone deacetylases (HDACs) –. Sites of Tup1 binding were associated with “position unchanged” and “occupancy retained” or “occupancy unchanged” nucleosomes (Figure 3C). These results support the hypothesis that Tup1 represses transcription by stabilizing nucleosome position and occupancy , .
Isw2 is associated with the unique combination of retained occupancy and altered position, providing evidence for its biochemical activity in vivo
Isw2 is an ATP-dependent chromatin remodeler that positions nucleosomes at the 5′ and 3′ ends of genes by binding to both the histone octamer and DNA , . ISW2 is known to catalyze the centering of a nucleosome on a DNA substrate in vitro, . Based on the distribution of catalytically inactive Isw2 enzyme (Isw2K215R) , Isw2 is associated with the unique combination of “occupancy retained” and “position altered” nucleosomes in our experiment. This is in contrast to most other factors associated with “occupancy retained” nucleosomes, which are typically classified as “position unchanged” nucleosomes (Figure 3D).
Based on its biochemical nucleosome-centering activity , , we hypothesized that Isw2 may help retain nucleosome occupancy by stabilizing histone-DNA interactions, while at the same time causing the position of bound nucleosomes to be especially sensitive to the loss of an adjacent nucleosome. Adjacent nucleosome loss could provide free DNA for the nucleosome centering activity of ISW2. Consistent with this hypothesis, of the 653 nucleosomes bound by Isw2K215R and classified as “position altered”, 429 move in the same direction in all 4 replicates, while another 142 shift in the same direction in 3 of the 4 replicates, indicating that for over 85% of the affected nucleosomes there is a clear directionality to the Isw2-associated shift in vivo (Figure S8A). The shift in position of these Isw2-bound nucleosomes was strongly associated with a decrease in nucleosome occupancy within 600 bp of the direction of the positional shift (p = 6.9E−11). In other words, the Isw2-bound nucleosomes consistently shifted specifically in the direction of a nearby, lost nucleosome. In contrast, Isw2-bound nucleosomes that do not change position are associated with adjacent “occupancy retained” nucleosomes (p = 0.059) and are not associated with a reduced occupancy of adjacent nucleosomes (p = 1), suggesting that a loss of adjacent nucleosomes is required for ISW2-mediated positional changes (Figure S8B, S8C). Fully consistent with the known in vitro activity of ISW2, the “linker” length (distance to the next mapped nucleosome) increased both upstream and downstream of Isw2-bound nucleosomes following nucleosome depletion such that the Isw2-bound nucleosomes became centered on the new local DNA substrate. While the ability of Isw2 to slide nucleosomes in vivo has been demonstrated previously , our result suggests that under conditions of lowered nucleosome density, Isw2 acts to create regularly-spaced nucleosomal arrays in vivo. We note that position-altered nucleosomes not bound by the mutant Isw2K215R as described in  were also centered, suggesting that wildtype Isw2 or another remodeling enzyme may center nucleosomes on newly-created gaps during genome-wide histone depletion (Figure S8D).
“Position altered,” Isw2-bound nucleosomes tended to contain the histone variant H2A.z (p<0.01) relative to all Isw2-bound nucleosomes , while Isw2-bound nucleosomes that were stably positioned were enriched for Tup1 (p<0.001) and Rsc8 (p<0.0001) binding relative to all Isw2-bound nucleosomes. Tup1 has previously been shown to localize independently to Isw2-bound regions, suggesting that the two proteins may work independently to maintain nucleosome position . Taken together, these patterns suggest that Isw2's centering function in vivo may be aided by incorporation of H2A.z and restricted by Tup1 and Rsc8.
Nhp6a is associated with preferential loss of nucleosomes specifically in transcribed regions
Nhp6a is an HMG-group protein known to associate with chromatin (reviewed in ). Loss of Nhp6a interferes with conditional gene activation  and has been reported to stabilize nucleosomes at promoters . Human HMGB1 aids in depositing nucleosomes on a DNA template in vitro, and deletion of both Nhp6a and Nhp6b in S. cerevisiae results in a 20–30% decrease in histone levels in vivo, suggesting that Nhp6a/b functions in regulating nucleosome stability or deposition.
We found that previously measured Nhp6a binding was associated with nucleosomes that were preferentially lost in our experiments following H3 depletion (Figure 3D) . This is generally consistent with a role for Nhp6 in nucleosome destabilization or deposition. To investigate if Nhp6a has different functions in promoters and gene bodies, we divided nucleosomes affected by H3 depletion into nucleosomes that fell into intergenic or genic regions using annotations from a recent study on the transcribed portion of the yeast genome . For 23 of the 40 data sets shown in Figure 3, the pattern of association with nucleosome behavior was similar between the intergenic and genic regions. That is, factors that were enriched or depleted in a given classification category were enriched or depleted, respectively, in both the intergenic and genic regions. However, the chromatin remodelers Rsc8, Isw2, and Nhp6a were among those that that showed the most striking differences in nucleosome behavior in intergenic regions versus gene bodies (Figure S9 and Table S2). Nhp6a binding was weakly associated with the “occupancy reduced” class in intergenic regions but was strongly associated with the “position altered” and the “occupancy reduced” classes in transcribed regions. This is consistent with a function for Nhp6a in nucleosome incorporation  and suggests that areas to which Nhp6a is recruited in transcribed regions may be especially sensitive to a reduction in the available histone pool.
We found that the nucleosomes with reduced occupancy following Nhp6a/b deletion according to  were not the same as the nucleosomes classified as “reduced occupancy” in our study after H3 depletion. Only 299 out of the ∼7000 nucleosomes with reduced occupancy following Nhp6a/b deletion in  were held in common with the ∼2000 nucleosomes classified as reduced occupancy in our study, suggesting that the mechanisms underlying nucleosome loss in the two experiments are distinct. To examine this more closely, we compared the average Nhp6a binding from  to changes in nucleosome occupancy due to H3 depletion (this study) or to changes in nucleosome occupancy due to the absence of Nhp6a/b . While there is a connection between Nhp6a binding and changes in occupancy following H3 depletion in our study (Figure S10A), a connection between Nhp6a binding and nucleosomes lost in in the Nhp6a/b deletion was not apparent (Figure S10C). Similarly, a significant enrichment for Nhp6a binding was found at nucleosomes that belonged to our “reduced occupancy” class (compared to all of the nucleosomes with classified occupancies following H3 depletion; p<1E−15; Figure S10B). However, there was only a weak association between Nhp6a binding and nucleosomes lost in in the Nhp6a/b deletion strain (p = 0.004; Figure S10D).
DNA sequence has a greater influence on nucleosome occupancy after histone H3 depletion
The influence of DNA sequence on nucleosome occupancy has been determined directly by in vitro reconstitution of nucleosomes using naked yeast DNA , . We compared genome-wide nucleosome occupancy in wildtype and H3 shutoff cells to nucleosome occupancy measured in nucleosome reconstitution experiments. At 0 hours, both wildtype and H3 shutoff cells showed similar correlations to in vitro nucleosome reconstitution (r = 0.69 and 0.67, respectively). After 3 hours in dextrose, the wildtype correlation to the in vitro data decreased (r = 0.59), but the H3 shutoff strain's correlation increased (r = 0.75; Figure 4A). To confirm that the decreased correlation in dextrose-grown wildtype cells was not specific to our experiment, we compared the correlation between cells grown in galactose and dextrose to the in vitro data from an independent strain and data set . A similar decrease in correlation with in vitro data was observed between galactose-grown (r = 0.774) and dextrose-grown (r = 0.716) yeast (Figure S11). This suggests that nucleosome occupancy during growth in galactose (as opposed to dextrose) is more similar to the organization observed in vitro. More relevant to our study, H3 depletion in vivo (which in this case occurs in dextrose) results in a chromatin organization that is much more similar to the in vitro configuration than cells with the normal complement of nucleosomes grown in either dextrose or galactose.
We next used a previously published DNA sequence-based model of nucleosome occupancy  to determine if the changes in a given nucleosome's position or occupancy after H3 depletion were influenced by the underlying DNA sequence. By definition, occupancy predictions based on DNA-sequence can change only if the nucleosome's position changes. Therefore, we calculated the change in a nucleosome's predicted occupancy based on its position before and after H3 depletion. As expected, the DNA-predicted occupancy value did not change at nucleosomes classified as having unchanged positions because the coordinates for the before and after locations were virtually identical. However, among the “position altered” nucleosomes, there was an average increase in the predicted occupancy at the position following histone depletion relative to the starting position (Figure 4B).
Approximately 25% of the occupancy-altered nucleosomes (reduced or retained) also showed a significant position change following H3 depletion. At these nucleosomes, the average predicted occupancy for the “before” and “after” positions corresponds with the actual observed change in nucleosome occupancy (Figure 4B). To see if this trend extended to all of the nucleosomes with altered occupancy, we examined the change in the actual and predicted occupancies for all of the nucleosomes classified as having altered occupancy and having any degree of shift in position (this group included all position classifications other than “position unchanged”, including “no call”). For both reduced and retained occupancy nucleosomes, the degree of nucleosome reduction or retention in vivo correlated with the degree of change in the DNA sequence-predicted nucleosome affinity (Figure 4C). In a recent study that examined nucleosome loss due to Nhp6a/b depletion, the underlying nucleosome-affinity of the DNA also corresponded with changes in nucleosome occupancy . Thus, there is strong evidence that under conditions of limiting histone concentrations, DNA sequence contributes directly to changes in nucleosome occupancy in vivo.
Decreased nucleosome occupancy in promoters following H3 depletion results in increased gene expression
We next revisited the hypothesis that gene expression changes in response to histone depletion are rooted in changes in chromatin organization , –. We used RNA-seq to quantify the relative abundance of transcripts from wildtype and H3 shutoff cells after 3 hours in dextrose. Our RNA-seq measurements of relative expression correlated well with previously published microarray experiments from H4-depleted cells, despite differences in histone-depletion methodology and RNA detection methods (r = 0.66 for expression arrays vs. RNA-seq RPMK). We used a BioConductor package, EdgeR, to analyze the RNA-seq data . We detected 2453 de-repressed genes following histone depletion, compared to the 888 that were previously identified . The number of genes with significantly decreased transcript levels following H3 depletion was 753 in this study, compared to the 569 reported previously for H4 depletion (Table S3) . Thus, the increased sensitivity of RNA-seq identified nearly half of yeast genes as having higher expression due to H3 depletion, three times the previous estimate.
We divided the yeast genes into three groups based on expression changes: increased, normal, and decreased. In the promoters of genes with increased or normal expression following H3 depletion, we found an over-representation of “occupancy reduced” nucleosomes (Figure 5 and Figure S12). In contrast, nucleosomes in the promoters of genes with decreased or normal expression were not significantly associated with any class of H3-depletion response. In the gene body, genes with decreased expression tended to harbor “occupancy retained” nucleosomes, while those with increased expression harbored significantly fewer “occupancy retained” nucleosomes than expected (Figure 5 and Figure S12).
The experiments described above support the following main conclusions: (1) Depletion of histone H3 levels causes a defined subset of nucleosomes to alter their position and/or occupancy in vivo. (2) Nucleosomes that are preferentially lost tend to be located at promoters, and this, in turn, leads to de-repression of downstream genes. (3) Isw2, an important ATP-dependent chromatin remodeler, is associated with stable nucleosome occupancy but altered position, especially when an adjacent nucleosome is destabilized. Such nucleosomal positioning shifts in the direction of the adjacent loss event are consistent with a nucleosome-centering activity for Isw2 in vivo, which to this point has been observed only in vitro. (4) Following nucleosome loss, the intrinsic DNA sequence preferences of nucleosomes have a greater influence on occupancy profiles, presumably due to reduced steric hindrance from adjacent nucleosomes. (5) Nhp6a is associated with preferentially lost nucleosomes in gene bodies. Deletion of Nhp6a/b causes a 20–30% reduction in histone abundance and altered transcription of approximately 10% of the yeast genome (fold-change >1.5 and p<0.05) . However, the sets of nucleosomes that we identify as being sensitive to H3 depletion are largely separate from the set identified as being destabilized by the loss of Nhp6a/b. This implies that the nucleosome loss events observed in the two studies may occur by independent mechanisms.
Our approach used MNase-seq to map nucleosome position and relative occupancy before and after H3 depletion. One concern with using MNase to measure nucleosome position and occupancy is that the enzyme's ability to digest DNA can be influenced by the local histone occupancy, with regions of lower histone occupancy being more susceptible to MNase digestion. Following H3 depletion, we observed increased smearing in the MNase-digested DNA fragments but saw no change in the average fragment length or the number of nucleosomes called by our algorithm. Thus, despite the increased sensitivity to MNase, the mononucleosome properties were equivalent to wildtype cells after adjusting the MNase concentration. The strong correlations to biologically relevant annotations provide additional evidence that the position and occupancy comparisons we make based on the data are informative.
We conclude that a combination of DNA-encoded nucleosome preference and chromatin composition regulate nucleosome occupancy and positional stability under conditions of limited histone protein availability. This, in turn, dictates which genes are most susceptible to a loss of regulatory fidelity. Most importantly, our data point to the factors likely to influence nucleosome stability under normal growth conditions, and to the specific genomic locations at which they are likely to act. This information serves as a platform for more detailed investigations into the mechanisms of nucleosome regulation.
Materials and Methods
Strains and growth conditions
Wildtype (YEF473A) and H3 shutoff (DCB200.1) cells were maintained on agar plates containing 2% galactose . For experiments, the cells were grown in liquid media with the indicated carbon source. Logarithmically growing cells were diluted to an OD600 of 0.0375 in 1.25 L of fresh YPGal media (1% yeast extract, 2% peptone, 2% galactose) and grown for 16 hours. Cells were collected via suction filtration with a 0.2 µm filter and washed with approximately 100 mL YPD (1% yeast extract, 2% peptone, 2% dextrose) before being resuspended in 1.25 L of YPD media. After switching the carbon source, 1 L of cells was immediately transferred to a fresh flask containing formaldehyde for our “0 hour” MNase digested time point (next section). We added 750 mL of fresh YPD to the remaining 250 mL, and the cells were grown at 30°C with shaking for 3 hours prior to collection for the final time point.
After collection, the cells were crosslinked with formaldehyde, the cell wall was digested with lyticase, and the DNA was digested with a titration of MNase. The resulting DNA was electrophoresed on an agarose gel, and the mono-nucleosome band was excised for sequencing. See Text S1 for additional details. The raw sequencing data is available at GEO accession number GSE29291 (single-end sequences) and GSE29292 (paired-end sequences).
RNA isolation and cDNA synthesis
RNA was isolated using the hot acidic phenol method  from wildtype and H3 shutoff cells grown as described above and transitioned to dextrose-containing media for 3 hours. RNA was further purified using the RNEasy Mini Kit (Qiagen 74104) to remove trace amounts of phenol. Ribosomal RNA was removed using the RiboMinus system (Invitrogen K155003). The quality of the RNA and the absence of ribosomal RNA were confirmed by gel electrophoresis. We then fragmented 4.5 µg of RNA using Ambion RNA fragmentation reagent (Ambion AM8740) at 70°C for 5 minutes and used the resulting RNA fragments as input for double-strand cDNA synthesis using a double-stranded cDNA synthesis kit (Invitrogen 11917-010) with random priming (Invitrogen 48190-011). The resulting cDNA was then prepared for Illumina sequencing. The raw RNA-seq data is available at GEO accession number GSE29293.
Illumina GAIIx sequencing
Samples were prepared for either single-end (two H3 shutoff, one wildtype MNase-seq replicates, and three replicates each of H3 shutoff and wildtype RNA data) or paired-end (two H3 shutoff and two wildtype replicates) sequencing using established protocols (Text S1). All libraries were sequenced using an Illumina Genome Analyzer IIx.
Mapping MNase reads
Reads were aligned to the sacCer1 build of the yeast genome using Bowtie v. 0.12.6 with default settings. Single-end reads were extended to the average fragment size in each experiment, which was calculated based on the distribution of reads on the Watson and Crick strands. Paired-end reads were required to have matching ends within 100–200 bases. Reads aligning to the rDNA locus (chr12 450,000–472,500) were removed. For analysis, all replicates were normalized with regard to sequencing depth by randomly selecting 5 million aligned reads per sample.
Calling nucleosome position and occupancy
A read center density map was created for each experiment using the center point of each extended (single-end) or paired (paired-end) read. The read-center density map was Gaussian smoothed with a standard deviation (s.d.) of 10 bp and a window of 3 s.d. The overall dyad density was visualized for Figure 1 using Matrix2png . To identify discrete nucleosomes, we repeated the following process: 1) The smoothed read-center density maximum was set as the center of a nucleosome. 2) The size of the region protected by the nucleosome was calculated as the average length of all extended/paired reads that covered the center base. 3) The s.d. of the nucleosome center in bases (the “fuzziness”) of the nucleosome was calculated using the number and locations of read centers that fell within the nucleosome's protected region. 4) The smoothed read-center density for all bases within one protected region of the nucleosome center was set to zero to prevent calling overlapping nucleosomes in successive rounds of nucleosome calling. 5) The nucleosome's occupancy was defined as the number of read centers falling within 50 bp on either side of the nucleosome center. This process was repeated until no additional nucleosomes could be called.
Classifying nucleosome behavior
Nucleosome positions at 0 and 3 hours in each replicate were required to overlap by 30 bp, and the coverage in each case was required to be greater than 5 reads. The nucleosome center positions at 0 and 3 hours were subjected to a t-test, and a p-value was calculated. The occupancy was compared using the binomial distribution test on the number of read centers in the 100-bp window centered on the nucleosomes' respective centers. In both cases, the resulting p-values were Bonferroni corrected based on the total number of nucleosomes compared in that replicate. Nucleosomes for each replicate were classified as having either unchanged (p>0.2) or altered (p<0.01) position and reduced (p<0.01), unchanged (p>0.2 and <0.8), or retained (p>0.99) occupancy relative to the 0 hour position and occupancy. Nucleosomes with p-values between 0.01 and 0.2 or 0.8 and 0.99 were left unclassified (“no call”) and were not used in downstream analyses. Nucleosome classification was compared between replicates, and only nucleosomes that were similarly classified in two or more replicates were used for further analyses. In addition, occupancy categories were restricted to nucleosomes that were not oppositely classified in any replicates (i.e., no nucleosomes were considered that were classified as occupancy reduced and occupancy retained in different replicates). Nucleosomes exhibiting behavior that was dependent on histone H3 depletion were identified by comparing the wildtype and H3 shutoff strains and removing any nucleosomes that were similarly classified in the wildtype experiments.
The majority of data sets used for comparative analysis were also generated from cells grown in glucose at an OD600 of ∼1. If the growth condition in the study was substantially different, it is noted in parentheses. We compared the nucleosome classes to available genome-wide data sets for the following marks that were downloaded from ChromatinDB (www.bioinformatics2.wsu.edu/cgi-bin/ChromatinDB/cgi/downloader_select.pl) as bulk histone occupancy normalized data: H2AK7ac, H2BK11ac, H2BK16ac, H3K9ac, H3K14ac, H3K18ac, H3K23ac, H3K27ac, H4K8ac, H4K12ac and H4K16ac ; H3 N-terminal ac and H4 N-terminal acetylation ; H3K36me2 ; H2A.zK14ac ; H3K56ac , ; and H3 and H4 occupancy . We also used H3K36ac  and H3K4me3  data normalized to H3 occupancy from . H3K4me2, H3K4me4, and H3R2me2a H3-normalized data is from . H3 turnover data is from  (grown in raffinose and galactose). Sir2 data is from . Tup1 data is from  (OD600∼0.6–0.8). Rpo21, Bdf1, Rap1, Reb1, Vps72, and Srm1 data is from . Rsc8 data is from . Isw2K215R data is from  (OD600∼0.7). Nhp6a data is from . Rpd3 data is from . High resolution H2A.z data is from . For Figure 3 and Figure S7, the average enrichment of the comparison data for nucleosomes in each class was calculated. To calculate significance, the average enrichment for an identical number of nucleosomes selected from the entire set of H3 depletion-dependent nucleosomes was calculated 100 times, and the standard deviation of the random average was used to assign a z-score for the actual value. The data is reported as the −log10 of the p-value of the z-score (raw z-scores are available in Table S2 and Table S3).
Comparison to Nhp6a/b depletion
H3-depletion affected histones identified in this study were compared to nucleosomes identified in a previous study in which Nhp6a and Nhp6b were deleted in a yeast strain . To compare changes in occupancy between the two experiment sets, we used the log2 ratio of the Nhp6a/b deletion strain's score to the wildtype score for the Celona et al. data  and the previously described classification of H3-depletion dependent occupancy.
RNA-seq reads from three biological replicates each of H3 shutoff and wildtype cells were mapped to the S. cerevisiae sacCer1 genome using Bowtie (v. 0.12.6.0) and TopHat (v. 1.1.0) with a maximum intron size of 1 kb. Samtools (v. 0.1.8.0) was used to determine read pileups at each base in the genome. The total coverage in bases reported as transcribed  was divided by the read length (28 bp) and was used as the coverage for determining differential gene expression using EdgeR . Genes with a reported p-value<0.001 were considered to be differentially expressed.
Histone depletion–dependent changes versus expression changes
The number of nucleosomes in each class that overlapped with each gene and upstream, non-transcribed region was determined. To calculate significance, a number of nucleosomes equal to those in the class were randomly chosen from all classified nucleosomes, and this subset was tested for overlap with the region. This was repeated 100 times, and the −log10 p-value of the z-score is reported.
1. HanMKimUKaynePGrunsteinM 1988 Depletion of histone H4 and nucleosomes activates the PHO5 gene in Saccharomyces cerevisiae. The EMBO journal 7 2221 2228
2. KimUHanMKaynePGrunsteinM 1988 Effects of histone H4 depletion on the cell cycle and transcription of Saccharomyces cerevisiae. The EMBO journal 7 2211 2219
3. HanMGrunsteinM 1988 Nucleosome loss activates yeast downstream promoters in vivo. Cell 55 1137 1145
4. DurrinLKMannRKGrunsteinM 1992 Nucleosome loss activates CUP1 and HIS3 promoters to fully induced levels in the yeast Saccharomyces cerevisiae. Mol Cell Biol 12 1621 1629
5. WyrickJJHolstegeFCPJenningsEGCaustonHCShoreD 1999 Chromosomal landscape of nucleosome-dependent gene expression and silencing in yeast. Nature 402 418 421
6. CelonaBWeinerADi FeliceFMancusoFMCesariniE 2011 Substantial histone reduction modulates genomewide nucleosomal occupancy and global transcriptional output. PLoS Biol 9 e1001086 doi:10.1371/journal.pgen.1001086
7. GkikopoulosTSchofieldPSinghVPinskayaMMellorJ 2011 A role for Snf2-related nucleosome-spacing enzymes in genome-wide nucleosome organization. Science 333 1758 1760
8. KaplanNMooreIKFondufe-MittendorfYGossettAJTilloD 2009 The DNA-encoded nucleosome organization of a eukaryotic genome. Nature 458 362 366
9. ZhangYMoqtaderiZRattnerBPEuskirchenGSnyderM 2009 Intrinsic histone-DNA interactions are not the major determinant of nucleosome positions in vivo. Nature structural & molecular biology 16 847 852
10. ZhangZWippoCJWalMWardEKorberP 2011 A packing mechanism for nucleosome organization reconstituted across a eukaryotic genome. Science 332 977 980
11. BouckDCBloomK 2007 Pericentric Chromatin Is an Elastic Component of the Mitotic Spindle. Current Biology 17 741 748
12. HanMChangMKimU-JGrunsteinM 1987 Histone H2B repression causes cell-cycle-specific arrest in yeast: Effects on chromosomal segregation, replication, and transcription. Cell 48 589 597
13. NagalakshmiUWangZWaernKShouCRahaD 2008 The Transcriptional Landscape of the Yeast Genome Defined by RNA Sequencing. Science 320 1344 1349
14. KaplanNHughesTRLiebJDWidomJSegalE 2010 Contribution of histone sequence preferences to nucleosome organization: proposed definitions and methodology. Genome Biology 11 140 152
15. KeoghM-CKurdistaniSKMorrisSAAhnSHPodolnyV 2005 Cotranscriptional Set2 Methylation of Histone H3 Lysine 36 Recruits a Repressive Rpd3 Complex. Cell 123 593 605
16. CarrozzaMJLiBFlorensLSuganumaTSwansonSK 2005 Histone H3 Methylation by Set2 Directs Deacetylation of Coding Regions by Rpd3S to Suppress Spurious Intragenic Transcription. Cell 123 581 592
17. NeumannHHancockSMBuningRRouthAChapmanL 2009 A Method for Genetically Installing Site-Specific Acetylation in Recombinant Histones Defines the Effects of H3 K56 Acetylation. Molecular Cell 36 153 163
18. LiWNagarajaSDelcuveGHendzelMDavieJ 1993 Effects of histone acetylation, ubiquitination and variants on nucleosome stability. Biochem J 296 737 744
19. Brower-TolandBWackerDAFulbrightRMLisJTKrausWL 2005 Specific Contributions of Histone Tails and their Acetylation to the Mechanical Stability of Nucleosomes. Journal of Molecular Biology 346 135 146
20. DionMFKaplanTKimMBuratowskiSFriedmanN 2007 Dynamics of Replication-Independent Histone Turnover in Budding Yeast. Science 315 1405 1408
21. BuckMJLiebJD 2006 A chromatin-mediated mechanism for specification of conditional transcription factor targets. Nat Genet 38 1446 1451
22. KoerberRTRheeHSJiangCPughBF 2009 Interaction of Transcriptional Regulators with Specific Nucleosomes across the Saccharomyces Genome. Molecular Cell 35 889 902
23. WuJSukaNCarlsonMGrunsteinM 2001 TUP1 Utilizes Histone H3/H2B-Specific HDA1 Deacetylase to Repress Gene Activity in Yeast. Molecular Cell 7 117 126
24. SmithRLJohnsonAD 2000 Turning genes off by Ssn6-Tup1: a conserved system of transcriptional repression in eukaryotes. Trends in Biochemical Sciences 25 325 330
25. CoureyAJJiaS 2001 Transcriptional repression: the long and the short of it. Genes & Development 15 2786 2796
26. WatsonADEdmondsonDGBoneJRMukaiYYuY 2000 Ssn6-Tup1 interacts with class I histone deacetylases required for repression. Genes & Development 14 2737 2744
27. RizzoJMMieczkowskiPABuckMJ 2011 Tup1 stabilizes promoter nucleosome positioning and occupancy at transcriptionally plastic genes. Nucleic Acids Research
28. GelbartMERechsteinerTRichmondTJTsukiyamaT 2001 Interactions of Isw2 Chromatin Remodeling Complex with Nucleosomal Arrays: Analyses Using Recombinant Yeast Histones and Immobilized Templates. Mol Cell Biol 21 2098 2106
29. FazzioTGGelbartMETsukiyamaT 2005 Two Distinct Mechanisms of Chromatin Interaction by the Isw2 Chromatin Remodeling Complex In Vivo. Mol Cell Biol 25 9165 9174
30. KassabovSRHenryNMZofallMTsukiyamaTBartholomewB 2002 High-resolution mapping of changes in histone-DNA contacts of nucleosomes remodeled by ISW2. Molecular and cellular biology 22 7524 7534
31. LangstGBonteEJCoronaDFVBeckerPB 1999 Nucleosome Movement by CHRAC and ISWI without Disruption or trans-Displacement of the Histone Octamer. Cell 97 843 852
33. FazzioTGTsukiyamaT 2003 Chromatin Remodeling In Vivo: Evidence for a Nucleosome Sliding Mechanism. Molecular Cell 12 1333 1340
34. ZhangHRobertsDNCairnsBR 2005 Genome-Wide Dynamics of Htz1, a Histone H2A Variant that Poises Repressed/Basal Promoters for Activation through Histone Loss. Cell 123 219 231
35. ZhangZReeseJC 2004 Ssn6-Tup1 requires the ISW2 complex to position nucleosomes in Saccharomyces cerevisiae. The EMBO journal 23 2246 2257
36. ThomasJOTraversAA 2001 HMG1 and 2, and related ‘architectural’ DNA-binding proteins. Trends in Biochemical Sciences 26 167 174
37. StillmanDJ 2010 Nhp6: A small but powerful effector of chromatin structure in Saccharomyces cerevisiae. Biochimica et Biophysica Acta (BBA) - Gene Regulatory Mechanisms 1799 175 180
38. DowellNLSperlingASMasonMJJohnsonRC 2010 Chromatin-dependent binding of the S. cerevisiae HMGB protein Nhp6A affects nucleosome dynamics and transcription. Genes & Development 24 2031 2042
39. ZhangYMoqtaderiZRattnerBPEuskirchenGSnyderM 2009 Intrinsic histone-DNA interactions are not the major determinant of nucleosome positions in vivo. Nat Struct Mol Biol 16 847 852
40. RobinsonMDMcCarthyDJSmythGK 2010 edgeR: a Bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics 26 139 140
41. XiaoTHallHKizerKOShibataYHallMC 2003 Phosphorylation of RNA polymerase II CTD regulates H3 methylation in yeast. Genes & Development 17 654 663
42. PavlidisPNobleWS 2003 Matrix2png: a utility for visualizing matrix data. Bioinformatics 19 295 296
43. KurdistaniSKTavazoieSGrunsteinM 2004 Mapping Global Histone Acetylation Patterns to Gene Expression. Cell 117 721 733
44. BernsteinBEHumphreyELErlichRLSchneiderRBoumanP 2002 Methylation of histone H3 Lys 4 in coding regions of active genes. Proceedings of the National Academy of Sciences of the United States of America 99 8695 8700
45. RaoBShibataYStrahlBDLiebJD 2005 Dimethylation of Histone H3 at Lysine 36 Demarcates Regulatory and Nonregulatory Chromatin Genome-Wide. Mol Cell Biol 25 9447 9459
46. MillarCBXuFZhangKGrunsteinM 2006 Acetylation of H2AZ Lys 14 is associated with genome-wide gene activity in yeast. Genes & Development 20 711 722
47. XuFZhangKGrunsteinM 2005 Acetylation in Histone H3 Globular Domain Regulates Gene Expression in Yeast. Cell 121 375 385
48. LeeC-KShibataYRaoBStrahlBDLiebJD 2004 Evidence for nucleosome depletion at active regulatory regions genome-wide. Nat Genet 36 900 905
49. MorrisSARaoBGarciaBAHakeSBDiazRL 2007 Identification of Histone H3 Lysine 36 Acetylation as a Highly Conserved Histone Modification. Journal of Biological Chemistry 282 7632 7640
50. XiaoTShibataYRaoBLaribeeRNO'RourkeR 2007 The RNA Polymerase II Kinase Ctk1 Regulates Positioning of a 5′ Histone Methylation Boundary along Genes. Mol Cell Biol 27 721 731
52. KirmizisASantos-RosaHPenkettCJSingerMAVermeulenM 2007 Arginine methylation at histone H3R2 controls deposition of H3K4 trimethylation. Nature 449 928 932
53. BadisGChanETvan BakelHPena-CastilloLTilloD 2008 A Library of Yeast Transcription Factor Motifs Reveals a Widespread Function for Rsc3 in Targeting Nucleosome Exclusion at Promoters. Molecular Cell 32 878 887
54. KurdistaniSKGrunsteinM 2003 Histone acetylation and deacetylation in yeast. Nat Rev Mol Cell Biol 4 276 284
55. AlbertIMavrichTNTomshoLPQiJZantonSJ 2007 Translational and rotational settings of H2A.Z nucleosomes across the Saccharomycescerevisiae genome. Nature 446 572 576
56. MavrichTNIoshikhesIPVentersBJJiangCTomshoLP 2008 A barrier nucleosome model for statistical positioning of nucleosomes throughout the yeast genome. Genome Research 18 1073 1083