In vivo binding of active heat shock transcription factor 1 to human chromosome 9 heterochromatin during stress

Abstract

Activation of the mammalian heat shock transcription factor (HSF)1 by stress is a multistep process resulting in the transcription of heat shock genes. Coincident with these events is the rapid and reversible redistribution of HSF1 to discrete nuclear structures termed HSF1 granules, whose function is still unknown. Key features are that the number of granules correlates with cell ploidy, suggesting the existence of a chromosomal target. Here we show that in humans, HSF1 granules localize to the 9q11-q12 heterochromatic region. Within this locus, HSF1 binds through direct DNA-protein interaction with a nucleosome-containing subclass of satellite III repeats. HSF1 granule formation only requires the DNA binding competence and the trimerization of the factor. This is the first example of a transcriptional activator that accumulates transiently and reversibly on a chromosome-specific heterochromatic locus.

Figure 1.

Mapping of HSF1 granule chromosomal target. HSF1 (green) was detected by immunofluorescence on metaphase spreads prepared from normal fibroblasts exposed for 1 h at 45°C together with either chromosome 9 centromeres (A) or pHuR98 probe (B) detected by FISH (red). DNA was counterstained with DAPI. (C) Ideogram of human chromosome 9 showing the precise location of HSF1 granules (arrow). (D) HSF1 granules (red) were codetected with RNA polymerase II transcription sites revealed by BrUTP incorporation (green) in fibroblasts exposed for 1 h at 45°C. Bar, 5 μm.

Figure 2.

Mapping of the HSF1 protein domains required for the targeting to the granules. Images of the distribution at 37°C or after 30 min at 42°C of the different HSF1–GFP mutants transiently expressed in HeLa cells are shown with the corresponding DAPI image. Bar, 5 μm.

Figure 3.

In vitro reconstitution of HSF1 granules on metaphase chromosome spreads. (A) Cytogenetic preparations of chromosomes were incubated with recombinant human HSF1 protein, processed for immunofluorescence for HSF1 (green), and subsequently processed for chromosome 9 centromeres detection by FISH (red). (B) The same experiment was performed on chromosome spreads treated with 100 μg/ml proteinase K for 15 min at 37°C prior to incubation with HSF1 (green). DNA was counterstained with DAPI. Bar, 5 μm.

Figure 4.

In vitro binding of HSF1 to chromosome 9 satellite III repeats and to HSP70 heat shock element. (A) Sequence comparison of the proximal HSE of HSP70 promoter (nt-130 to -81) and pHuR98. The plain grey boxes indicate the individual HSF1 binding sites (Kroeger et al., 1993). The empty grey boxes indicate the possible HSF1 binding sites. (B) EMSA was performed by incubating a labeled HSE (lanes 1–9) or pHuR98 probe (lanes 10–19) with various amounts of recombinant HSF1. For supershift assay, 1 μl of polyclonal anti-HSF1 antibody was added to the reaction (lane 19). The arrowhead points to a high mobility form of pHuR98 free probe which is not shifted by HSF1. (C) HSF1 (12.5 ng) was incubated with various excesses of unlabeled HSE competitor probe prior to the addition of a fixed amount of labeled pHuR98 probe.