Mechanism of hsp70i Gene Bookmarking

See allHide authors and affiliations

Science  21 Jan 2005:
Vol. 307, Issue 5708, pp. 421-423
DOI: 10.1126/science.1106478


In contrast to most genomic DNA in mitotic cells, the promoter regions of some genes, such as the stress-inducible hsp70i gene that codes for a heat shock protein, remain uncompacted, a phenomenon called bookmarking. Here we show that hsp70i bookmarking is mediated by a transcription factor called HSF2, which binds this promoter in mitotic cells, recruits protein phosphatase 2A, and interacts with the CAP-G subunit of the condensin enzyme to promote efficient dephosphorylation and inactivation of condensin complexes in the vicinity, thereby preventing compaction at this site. Blocking HSF2-mediated bookmarking by HSF2 RNA interference decreases hsp70i induction and survival of stressed cells in the G1 phase, which demonstrates the biological importance of gene bookmarking.

During mitosis, the genome must be compacted in order for chromosomes to be segregated during cytokinesis. An enzyme called condensin, which is composed of five subunits, plays an important role in this compaction process and is activated at the onset of mitosis by phosphorylation of the CAP-G, CAP-H, and CAP-D2 subunits by the Cdc2–cyclin B kinase (15). However, a number of gene promoters, including those of the hsp70i and c-myc genes, do not appear to be tightly compacted in mitotic cells (69). The mechanism responsible for preventing compaction of specific gene regions during mitosis, called bookmarking, is unknown.

HSF2 is a transcription factor that can bind heat shock elements (HSEs) in the hsp70i promoter. But unlike the related HSF1 factor, which mediates stress-induced transcription of this gene, the function of HSF2 has remained unclear (1015). To better understand HSF2 function, we performed a yeast two-hybrid screen to identify HSF2-interacting proteins. One of the isolated clones contains a segment of the CAP-G protein, which is a subunit of the condensin enzyme (fig. S1A). The HSF2-interacting sequence is in the C-terminal region of CAP-G, and consists of amino acids 811 to 957 (fig. S1B). Immunoprecipitation analysis demonstrated that endogenous HSF2 and CAP-G interact and that more of the complex is detected in mitotic than asynchronous cells (Fig. 1A). Our previous work demonstrated that HSF2 is covalently modified at lysine 82 by the 97–amino acid small ubiquitin-related modifier (SUMO)–1 protein (16). Transfection of HeLa cells with myc-tagged wild-type HSF2 or a sumoylation-deficient K82R HSF2 mutant (in which Lys82 is replaced by Arg), followed by CAP-G immunoprecipitation analysis, revealed that sumoylation of HSF2 positively regulates its association with CAP-G (Fig. 1B). We postulated that sumoylation of HSF2 might be regulated in a cell-cycle–dependent manner to control HSF2 interaction with CAP-G. Consistent with this, Western blot analysis of extracts of cells that were obtained by fluorescence-activated cell sorting (FACS) indicated that G2/M-phase extracts displayed a band whose size was consistent with SUMO-modified HSF2, which was also detected in asynchronous cell extracts but displayed lower intensity in G0/G1- or S-phase cell extracts (Fig. 1C). Immunoprecipitation analysis of HSF2 from extracts of each cell population followed by SUMO-1 Western blot confirms that higher levels of sumoylated HSF2 are found in G2/M cells than in G0/G1 or S cells (Fig. 1D).

Fig. 1.

HSF2 interacts with the condensin subunit CAP-G in mitotic cells. (A) Extracts of asynchronous or mitotic (nocodazole-blocked) cells were immunoprecipitated with antibodies to CAP-G or nonspecific immunoglobulin (IgG), and the immunoprecipitates were subjected to Western blot with HSF2 antibodies. (B) HSF2 sumoylation stimulates CAP-G interaction. Extracts of HeLa cells transfected with myc-tagged wild-type HSF2 or nonsumoylatable K82R HSF2 were immunoprecipitated with CAP-G antibodies or nonspecific IgG followed by Western blot with myc antibodies. (C) HSF2 exhibits modification associated with G2/M cells. Extracts of equal numbers of G0/G1, S, and G2/M populations of human K562 cells obtained by FACS and asynchronous cells were subjected to HSF2 Western blot. (D) HSF2 exhibits increased sumoylation in G2/M cells. Extracts of equal numbers of FACS-sorted G0/G1, S, and G2/M populations of K562 cells were immunoprecipitated with HSF2 antibodies followed by Western blot with SUMO-1 antibodies.

The results above suggested that HSF2 may bind to the hsp70i promoter in a mitotic-dependent manner as part of a mechanism for preventing compaction of this promoter. This suggestion was tested by chromatin immunoprecipitation (ChIP) assay, which demonstrated that HSF2 does bind to the hsp70i promoter in mitotic cells, and more binding was observed in these cells as compared with asynchronous cells (Fig. 2A). HSF2 interaction was not observed for the histone H2B promoter gene in mitotic cells, demonstrating specificity (Fig. 2B).

Fig. 2.

HSF2 binds to the hsp70i promoter in mitotic cells. (A) ChIP assay was performed on asynchronous or mitotic (nocodazole-blocked) Jurkat cells by using HSF2 antibodies or control IgG antibodies and polymerase chain reaction (PCR) primers specific to the hsp70i gene. (B) Asynchronous and mitotic Jurkat cells were analyzed by ChIP assay by using HSF2 antibodies or control IgG, followed by PCR with primers specific to the histone H2B gene.

To assess directly the importance of HSF2 for hsp70i promoter bookmarking, we tested the effect of decreasing cellular HSF2 levels on the accessibility of this promoter in mitotic cells. HSF2 levels were reduced by using RNA interference (RNAi) methods, which decreased cellular HSF2 protein levels without affecting levels of the related HSF1 protein (Fig. 3A). Chromatin of the HSF2 RNAi-treated mitotic HeLa cells was subjected to a restriction enzyme accessibility assay, which measures the accessibility of a specific DNA region within intact chromatin (1720). In this assay, chromatin from HSF2 small interfering RNA (siRNA)–treated nocodazole-blocked (mitotic) cells was incubated with the restriction enzyme Sac I, which cleaves at nucleotide –76 near the proximal HSE of the hsp70i promoter (Fig. 3B, left side). Genomic DNA was then extracted, cut to completion with Nco I to yield a defined end beyond this Sac I site, and subjected to primer extension by using an hsp70i gene-specific primer. The intensity of the 123-nucleotide extension product was taken as a measure of the accessibility of the hsp70i promoter Sac I site within intact chromatin. The initial Sac I digestion of DNA within the chromatin was not anticipated to be completely efficient, even if the DNA target site was completely accessible. Therefore, some fragments that were not digested by Sac I (319-nucleotide products) were always expected. HSF2 siRNA treatment resulted in decreased accessibility of this Sac I site, as compared with the scrambled siRNA control (Fig. 3B, right side), which indicates a critical role for HSF2 in bookmarking the hsp70i promoter. Performing the same experiment with S cells suggests that the effect of lowered HSF2 levels on hsp70i promoter accessibility is more pronounced in G1 cells, which is consistent with the bookmarking hypothesis (fig. S2A). HSF2 reduction does not affect accessibility of another promoter accessible in mitotic cells, the histone H2B gene promoter (7), which indicates specificity (fig. S2B).

Fig. 3.

Reducing cellular HSF2 levels decreases hsp70i promoter bookmarking in mitotic cells and hsp70i inducibility and survival of stressed G1 cells. (A) Reduction of cellular HSF2 levels by RNAi. HeLa cells were transfected with no siRNA (mock), scrambled siRNA, or HSF2-specific siRNA, and extracts analyzed by HSF2 and HSF1 Western blot. (B) Lowering cellular HSF2 levels decreased accessibility of the hsp70i promoter in mitotic cells. (Left) An accessibility assay of hsp70i promoter in intact chromatin, indicating positions of the Sac I test site, the Nco I complete digestion site, the primer binding site, and the expected sizes of extension products. (Right) Chromatin of HeLa cells that were treated with scrambled siRNA or HSF2-specific siRNA, then blocked in mitosis by nocodazole treatment, was incubated with Sac I for 30 min at 37°C. Genomic DNA was extracted, digested to completion with Nco I, then subjected to primer extension analysis with an hsp70i-specific primer. (C) Reducing HSF2 levels decreased the ability of G1 cells to induce hsp70i expression in response to stress. HeLa cells were transfected with scrambled siRNA or HSF2 siRNA, then blocked with nocodazole before being released from blockage for 2 hours to allow entry into the G1 phase. The HeLa cells were next subjected to heat treatment at 43°C for 30 min (or kept at 37°C) followed by recovery at 37°C for 3 hours, then analyzed by Western blot with hsp70i antibody or β-actin antibody (loading control). (D) Reducing HSF2 levels increased stress-induced killing of G1 cells. HSF2 RNAi-treated HeLa cells that were nocodazole-blocked and released for 2 hours as in (C) were then subjected to thermo-tolerance–inducing treatment of 43°C for 30 min. Cells were then recovered at 37°C for 4 hours to allow hsp70i expression, followed by severe 45°C 30-min heat stress. After 40 hours back at 37°C, cell viability was measured by trypan blue exclusion.

It has been postulated that hsp70i bookmarking maintains the gene in a transcription-competent state so that induction could occur even in early G1 phase if stress were to arise. To test this hypothesis, HSF2 siRNA-treated cells blocked in mitosis with nocodazole but released from the block for 2 hours to permit entry into the G1 phase were subjected to 43°C treatment for 30 min followed by recovery at 37°C for 3 hours. Western blot indicated that hsp70i protein induction was significantly lower in HSF2 RNAi-treated cells compared with cells treated with scrambled siRNA (Fig. 3C). Furthermore, this reduced ability to induce hsp70i protein was correlated with a significant increase in cell death after stress treatment (Fig. 3D), which indicates the critical importance of HSF2-mediated hsp70i bookmarking for cell stress survival.

The results above show that in mitotic cells, HSF2 binds the hsp70i promoter, interacts with the CAP-G subunit of condensin, and is critical for the bookmarking of this promoter. But how does HSF2 mediate inhibition of condensin activity at this locus? Condensin activity requires phosphorylation of the CAP-G, CAP-D2, and CAP-H subunits by the mitotic kinase Cdc2–cyclin B (21, 22). We previously showed that HSF2 interacts with the serine-threonine protein phosphatase 2A (PP2A) (23), and other studies have suggested that a member of the PPP family of serine-threonine phosphatases, which includes PP2A, is involved in dephosphorylating the condensin subunits that are phosphorylated by Cdc2 (24). On the basis of these data, we hypothesized that HSF2 mediates hsp70i promoter bookmarking by recruiting PP2A to this promoter. When condensin interacts with HSF2 via CAP-G, it could be dephosphorylated and inactivated by the HSF2-associated PP2A to prevent compaction of this region of DNA. Supporting this hypothesis, immunoprecipitation experiments demonstrate that HSF2 does associate with PP2A in mitotic cells, and that more of the complex is observed in mitotic than in asynchronous cells, suggesting mitosis-dependent regulation of this interaction (Fig. 4A). Furthermore, ChIP assays show that PP2A exists within cross-linking distances of the hsp70i promoter in mitotic cells and that more association is observed in mitotic cells than in asynchronous cells (Fig. 4B). To confirm the function of PP2A as a condensin phosphatase, we isolated condensin by immunoprecipitation with CAP-G antibodies, phosphorylated it with purified Cdc2–cyclin B and [γ-32P]adenosine triphosphate (ATP) in vitro, and then incubated the phosphorylated condensin with purified PP2A. The results show that PP2A can dephosphorylate CAP-G, CAP-D2, and CAP-H (fig. S3A). Finally, we took advantage of the ability of the Mpm2 antibody to detect the phosphorylated forms of Cdc2–cyclin B substrates, including CAP-G (25), to show that HSF2 immunoprecipitates from mitotic cells contain an okadaic acid–inhibitable phosphatase activity that can efficiently dephosphorylate phosphorylated CAP-G that was immunoprecipitated from mitotic cells (Fig. 4C). This activity is not detected in HSF1 immunoprecipitates from mitotic cells (fig. S3B).

Fig. 4.

PP2A associates with HSF2 and the hsp70i promoter in mitotic cells and dephosphorylates mitotic CAP-G. (A) HSF2 complexes with PP2A in mitotic cells. HSF2 was immunoprecipitated from extracts of asynchronous or mitotic HeLa cells (with nonspecific IgG as negative control) followed by Western blot with PP2A catalytic subunit antibodies. (B) PP2A is associated with the hsp70i promoter in mitotic cells. ChIPassay was performed on asynchronous or mitotic (nocodazole-blocked) Jurkat cells by using PP2A catalytic subunit antibodies or control IgG antibodies and hsp70i promoter primers. (C) Endogenous mitotic HSF2 associates with okadaic acid–inhibitable phosphatase activity that dephosphorylates mitotic CAP-G. The first two lanes show immunoprecipitates pulled down by non-specific IgG or CAP-G antibody subjected to Western blot with either Mpm2 (detects phosphoCAP-G) or CAP-G antibodies (loading and degradation control). The next lanes show reactions containing CAP-G immunoprecipitates from mitotic cell extracts resuspended in phosphatase reaction buffer (NEB) and incubated with HSF2 immunoprecipitates (± 200 nM okadaic acid) or nonspecific IgG immunoprecipitates from mitotic cell extracts for 2 hours or no time.

The results in this paper suggest that HSF2 mediates hsp70i promoter bookmarking by binding to this promoter in mitotic cells, recruiting PP2A, and interacting with condensin enzyme by binding to the CAP-G subunit to allow the PP2A to dephosphorylate efficiently and to inactivate this condensin, thereby preventing compaction of this region of chromosomal DNA. The results also demonstrate the biological importance of bookmarking by showing that the loss of HSF2-mediated bookmarking significantly reduces the ability of cells to induce hsp70i gene expression and to survive stress. This knowledge opens the door to investigating the mechanisms and biological roles of other gene bookmarking events, toward understanding the full biological ramifications of bookmarking for cell function.

Supporting Online Material

Materials and Methods

Figs. S1 to S3

References and Notes

References and Notes

View Abstract

Navigate This Article