Widespread and precise reprogramming of yeast protein-genome interactions in response to heat shock

Abstract

Gene expression is controlled by a variety of proteins that interact with the genome. Their precise organization and mechanism of action at every promoter remains to be worked out. To better understand the physical interplay among genome-interacting proteins, we examined the temporal binding of a functionally diverse subset of these proteins: nucleosomes (H3), H2AZ (Htz1), SWR (Swr1), RSC (Rsc1, Rsc3, Rsc58, Rsc6, Rsc9, Sth1), SAGA (Spt3, Spt7, Ubp8, Sgf11), Hsf1, TFIID (Spt15/TBP and Taf1), TFIIB (Sua7), TFIIH (Ssl2), FACT (Spt16), Pol II (Rpb3), and Pol II carboxyl-terminal domain (CTD) phosphorylation at serines 2, 5, and 7. They were examined under normal and acute heat shock conditions, using the ultrahigh resolution genome-wide ChIP-exo assay in Saccharomyces cerevisiae Our findings reveal a precise positional organization of proteins bound at most genes, some of which rapidly reorganize within minutes of heat shock. This includes more precise positional transitions of Pol II CTD phosphorylation along the 5' ends of genes than previously seen. Reorganization upon heat shock includes colocalization of SAGA with promoter-bound Hsf1, a change in RSC subunit enrichment from gene bodies to promoters, and Pol II accumulation within promoter/+1 nucleosome regions. Most of these events are widespread and not necessarily coupled to changes in gene expression. Together, these findings reveal protein-genome interactions that are robustly reprogrammed in precise and uniform ways far beyond what is elicited by changes in gene expression.

Figure 1.

Factor occupancy at three genes before and after heat shock. Smoothed distribution of unshifted ChIP-exo tag 5′ ends (exonuclease stop sites) on forward and reverse strands (with the latter being inverted) on induced HSP42, repressed RPL3, and housekeeping REB1 genes upon mock or after 3 min of heat shock. Nucleosomes (Nuc) represent fragment midpoints of MNase-digested H3-immunoprecipitated chromatin. All TSSs are oriented such that the direction of transcription is to the right. The bottom panel of each column corresponds to melted PIC DNA derived from TFIIB PIP-seq. Foreground “T” nucleotides are in black and background “G” nucleotides are in green. Y-axes are scaled for each factor to fit the frame and are all on the same scale for a particular factor (row).

Figure 2.

Factor occupancy at all genes before and after heat shock. (A) Yeast cultures were either mock-treated (upper set of panels) or (B) heat shocked (lower) for 3 min at 37°C. ChIP-exo tag 5′ ends were plotted relative to transcription unit midpoints (start to end from left to right). Rows are sorted by unit length and grouped by class: ribosomal protein genes (RP, n = 130), SAGA-dominated genes (SAGA, n = 451), TFIID-dominated genes (TFIID, n = 4260), and noncoding SUTs (n = 846), CUTs (n = 924), and XUTs (n = 1720). All rows across data sets are linked.

Figure 3.

Segregated and sorted heat shock response. ChIP-exo tag 5′ ends were plotted from ±500 bp relative to +1 nucleosome dyads, panel-separated based on mock or 3 min of heat shock, segregated by gene class (RP, SAGA, TFIID) and then by increased/induced (I), decreased/repressed (R), or no changes (N) in TFIIH (Ssl2) occupancy upon heat shock: SAGA-dominated genes (I,R,N, n = 160, 55, 252, respectively), TFIID-dominated genes (I,R,N, n = 488, 353, 3358, respectively). Within each segregated group, rows were sorted by TFIIH occupancy in the region 100 bp upstream of to 100 bp downstream from the TSS. Also shown in the far upper right is a heat map of log₂ fold-changes in mRNA expression after 15 min of heat shock, using data from Yassour et al. (2009).

Figure 4.

Hsf1 and SAGA precisely colocalize between divergent PICs. (A) Log₂ changes in gene expression in an spt3Δ strain relative to wild type, under acute heat shock conditions (Huisinga and Pugh 2004). Gene classes are separated first by SAGA- vs. TFIID-dominated, then by heat shock-mediated changes in TFIIH occupancy (Induced, Repressed, No change), then by whether Spt3 was induced to bind upon heat shock (≥1.5 fold-change from mock at 3 min of heat shock). (B) Gene averaged plots for Hsf1 (red fill), SAGA (Spt3, blue trace), and TFIIB (Sua7, green filled) with respect to Hsf1 motif midpoint. Data were from 3-min heat shock time point. Genes were not oriented, which results in roughly equivalent levels of TFIIB on either side of Hsf1. (C) Gene-averaged plot of strand-separated ChIP-exo tag 5′ ends for SAGA (Spt3), distributed around −1 and TSS features. The direction of coding transcription is from left to right. Plots of tags mapping to the antisense strand are shown inverted. Data from 3 min of heat shock were used.

Figure 5.

RSC (Rsc9) relocates from gene bodies to promoters upon heat shock. (A) Averaged distribution of Rsc9 (normalized unshifted strand-separated tag 5′ ends) plotted with respect to +1 nucleosome dyads at all TFIID-dominated mRNA genes (n = 4260). Transcription is oriented to the right. ChIP-exo tags mapping to the antisense strand are inverted. Semitransparent purple filled plots represent mock heat shock, and dark purple traces represent acute heat shock (37°C for 3 min). Gray filled plots correspond to nucleosome dyads prior to heat shock which do not change in this bulk assessment upon heat shock. RSC peak distances (in bp) from the +1 dyad are indicated. (B) RSC (Rsc9) gene-averaged distribution of normalized shifted tags mapped with respect to +1 nucleosome dyad showing occupancy during mock heat shock (red trace), and a time course of acute heat shock (37°C) for 3, 6, 9, 12, and 15 min in green, blue, magenta, orange, and dark orange traces, respectively. Nucleosome dyads are shown in gray fill.

Figure 6.

Relocation of FACT upon heat shock. (A) Frequency distribution of gene-averaged ChIP-exo tag 5′ ends for PoI II (Rpb3, black trace) and FACT (Spt16, brown fill). Gray filled plots correspond to nucleosome dyads. Coding transcription is oriented to the right. Plots are separately scaled to 1 for each data set. (B) The absolute occupancy (normalized tag 5′ ends) of Pol II (Rpb3) (left panel) and FACT (Spt16) (right panel) were averaged at heat shock-induced genes (as defined by increased TFIIH occupancy).

Figure 7.

Melted PIC DNA is unaffected by Pol II accumulation in promoter regions. (A) Gene-averaged plot of TFIIB PIP-seq tag 5′ ends separated by gene class and heat shock response (i.e., changes in TFIIH occupancy). Mock, 3-, 6-min time points are indicated (gray fill, red trace, orange trace, respectively). Transcription is oriented to the right. Plots are scaled to 1. Transcribed regions have a yellow backdrop. Dashed line indicates position of transcription bubble upon heat shock. (B) Heat maps of TFIIB (Sua7) ChIP-exo (green), and PIP-seq (black) at 0, 3, and 6 min of heat shock. Gene classes are as in A. (C) Model for heat shock-induced reorganization of factors at most genes whose expression does not change, and genes whose expression is heat shock-induced.