Distinct histone methylation and transcription profiles are established during the development of cellular quiescence in yeast

Background Quiescent cells have a low level of gene activity compared to growing cells. Using a yeast model for cellular quiescence, we defined the genome-wide profiles of three species of histone methylation associated with active transcription between growing and quiescent cells, and correlated these profiles with the presence of RNA polymerase II and transcripts. Results Quiescent cells retained histone methylations normally associated with transcriptionally active chromatin and had many transcripts in common with growing cells. Quiescent cells also contained significant levels of RNA polymerase II, but only low levels of the canonical initiating and elongating forms of the polymerase. The RNA polymerase II associated with genes in quiescent cells displayed a distinct occupancy profile compared to its pattern of occupancy across genes in actively growing cells. Although transcription is generally repressed in quiescent cells, analysis of individual genes identified a period of active transcription during the development of quiescence. Conclusions The data suggest that the transcript profile and histone methylation marks in quiescent cells were established both in growing cells and during the development of quiescence and then retained in these cells. Together, this might ensure that quiescent cells can rapidly adapt to a changing environment to resume growth. Electronic supplementary material The online version of this article (doi:10.1186/s12864-017-3509-9) contains supplementary material, which is available to authorized users.


SUPPLEMENTAL TABLE LEGENDS
: Gene ontology (GO) categories for genes associated with Rpb3-Myc (RNAP II), H3K4me3, H3K36me, and H3K79me3 in growing and quiescent cells. The DAVID functional annotation tool was used to identify categories of genes enriched only in log cells, only in 7-day Q cells, or in both cell types (Common), P value ≥ 10 -4 .  information on the percent coverage of the IGR, GCR (gene coding region), or "other" region with the RNAP II CHER. The feature with the highest percentage of coverage is listed in the "dominant feature" column, followed by the % coverage of the IGR with CUTs, SUTs, and RNAs, and peak RNAP II enrichment in log and Q cells. Table S5: RNA-Seq data. Data were collected from 3 independent RNA-seq experiments conducted on total RNA isolated from growing (log) and 7-day quiescent cells (Q). The table contains a column with gene information followed by columns with conversions to TPM numbers of the merged values for the 3 log and Q replicates. The last column contains the log2 ratio of the merged log TPM values divided by the merged Q TPM values.

CHIP-CHIP DATA PROCESSING
An adapted version of the Model-based Analysis of Tiling Arrays (MAT) algorithm for the R programming environment (rMAT) was used to normalize probe intensities (Droit et al., 2010).
Each of the four ChIP DNAs (RNAP II, H3K36me3, H3K79me3 and H3K4me3) was normalized to input DNA, the resulting values were standardized, and biological replicates were averaged.
Pearson correlation coefficients of the log2-transformed probe intensities were consistently above 0.9 between replicates. For each probe, the MAT scores were calculated as in Schulze et al. (2011) with a sliding window of 300 bp using a customized Perl script (available on GitHub, https://github.com/khokamp/Perl-scripts). The resulting scores were visualized across the genome using the JBrowse genome browser. ChIP enriched regions (CHERs) were computed using a customized Perl script (available on GitHub, https://github.com/khokamp/Perl-scripts).
CHERs were computed using a 150 bp sliding window and were required to have a minimum average MAT score of 1.5.
Open reading frames (ORFs) were obtained from the Saccharomyces Genome Database (SGD) version R64.2.1. Only ORFs that were characterized as not 'Dubious' were used for the analysis.
For RNAP II, H3K36me3 and H3K79me3, an ORF was defined as ChIP enriched when a CHER overlapped at least 50% of the ORF. For H3K4me3, the region around the transcription start site (TSS) (+200/-100 bp) was analyzed for ChIP enrichment. The overlap between CHERs and ORFs/TSS regions was computed using BEDTools (Quinlan and Hall, 2010). Using these criteria, ORFs that were ChIP enriched in either log phase or quiescent cells, but where ChIP binding was entirely absent in the other, were defined as log-or quiescent-specific, respectively. ORFs common to both log and quiescent cells showed ChIP enrichment in both cell populations (Table1). Box plots, scatter plots and profile plots were all generated through the R statistical software (https://www.r-project.org/). For averaged gene profile plots the gene body and surrounding regions were organized into bins (40 for gene body, 20 each for upstream and downstream areas of 250 bp) to adjust for varying gene lengths. The Chromatra tool was utilized as described in Schulze et al. (2009).

DEFINITION OF IGRS
Intergenic regions (IGRs) that separate divergently transcribed ORFs were identified as described in Radonjic et al. (2005). Only ORFs with classification "verified" or "uncharacterized" were considered. GCRs represent ORFs. Any other regions in the genome not covered by IGRs or GCRS were classified as "other". The percentage coverage of Rpb3-Myc in IGRs, GCRs, and "other" regions was calculated from the Rpb3-Myc CHERs. IGR regions bound by Rpb3-Myc were based on 50% coverage of IGR and 50% coverage of adjacent GCR.

RNA-SEQ DATA PROCESSING
Reads were aligned and then assigned to genes in the yeast genome using a Perl script (available on GitHub, https://github.com/khokamp/Perl-scripts). The Pearson correlation coefficients of the log2-transformed read counts for the three log or Q cell replicates were consistently above 0.96. Read counts were transformed into TPMs (Transcript Per Million) following the algorithm described in Li et al. (2010). TPM values from replicates were merged using the mean, and the ratios between log and Q cells were calculated as the difference on the log2 scale of the merged values. Density plots of the log2-transformed TPM values showed a bimodal distribution, and the minimum between the two peaks in log cells was used to determine a threshold separating expressed genes from background noise (Fig. 1). Transcripts with TPM ≥2 were used for analysis.   MATa, hhf2-hht2::NAT, hta1-htb1::HPH, hht1-hhf1:: KAN,trp1∆2,; <pRS315 (HTA1-HTB1, HHT1-HHF1::LEU2)> YAF3