2925 Journal of Cell Science 110, 2925-2934 (1997) Printed in Great Britain © The Company of Biologists Limited 1997 JCS9684 HSF1 granules: a novel stress-induced nuclear compartment of human cells José J. Cotto, Susan G. Fox and Richard I. Morimoto* Department of Biochemistry, Molecular Biology, and Cell Biology, Rice Institute for Biomedical Research, Northwestern University, Evanston, IL 60208, USA *Author for correspondence (e-mail: r-morimoto@nwu.edu) SUMMARY Heat shock factor 1 (HSF1) is the ubiquitous stress-responsive transcriptional activator which is essential for the inducible transcription of genes encoding heat shock proteins and molecular chaperones. HSF1 localizes within the nucleus of cells exposed to heat shock, heavy metals, and amino acid analogues, to form large, irregularly shaped, brightly staining granules which are not detected during attenuation of the heat shock response or when cells are returned to their normal growth conditions. The kinetics of detection of HSF1 granules parallels the transient induction of heat shock gene transcription. HSF1 granules are also detected using an HSF1-Flag epitope tagged protein or a chimeric HSF1-green fluorescent protein which reveals that these nuclear structures are stress-induced and can be detected in living cells. The spatial organization of HSF1 granules in nuclei of stressed cells reveals that they are novel nuclear structures which are stress-dependent and provides evidence that the nucleus undergoes dynamic reorganization in response to stress. INTRODUCTION of the inert monomer remains unresolved, however, HSF1 trimers appear within minutes of activation and can be detected bound to DNA shortly thereafter. The rapid induction of the heat shock response suggests that the relocalization of HSF1 must involve a dynamic process. Yet, unlike most proteins which either translocate constitutively to the nucleus or exhibit a constant pattern of localization, the mechanisms involving HSF1 are likely to be distinct. Previous studies on the immunolocalization of the HSFs have detected the appearance of granules which seem to correlate with activation (Sarge et al., 1993; Nakai et al., 1993, 1995; Nakai et al., 1997). In this study we examine the cell biological properties of HSF1 using a collection of monoclonal antibodies which specifically detect the appearance of HSF1 granules within the nucleus of stressed cells. The kinetics of induction of HSF1 granules by heat shock and other stresses which lead to heat shock gene transcription strongly indicates a role in the stress response. The cellular response to adverse environmental and physiological conditions such as heat shock, exposure to amino acid analogs, heavy metals, oxidative stress, anti-inflammatory drugs, and arachidonic acid leads to the rapid and transient activation of genes encoding heat shock proteins (hsps) and molecular chaperones (Lindquist, 1986; Morimoto et al., 1990, 1994; Jurivich et al., 1992). Stress-induced transcription is regulated by a family of heat shock transcription factors (HSF). In vertebrates, four members of the HSF gene family (HSFs 14) have been characterized (Rabindran et al., 1991; Sarge et al., 1991; Schuetz et al., 1991; Nakai and Morimoto, 1993; Nakai et al., 1997). The co-expression of multiple HSFs and characterization of regulatory conditions has revealed that different members of the HSF family mediate the response to distinct forms of cellular stress. Consistent with this, HSF1 responds to the classical inducer of the heat shock response, HSF2 is activated during embryogenesis, spermatogenesis and erythroid differentiation, HSF3 functions as a high temperature activator, and HSF4 has properties of a negative regulator of heat shock gene expression (Sistonen et al., 1992, 1994; Sarge et al., 1993, 1994; Nakai et al., 1995, 1997). Under normal conditions of cell growth, HSF1 is maintained in an inert non-DNA binding state which undergoes reversible oligomerization to a DNA binding competent trimer in stressed cells (reviewed by Morimoto et al., 1994 and Wu, 1995). Two distinct mechanisms involving negative regulatory domains and constitutive phosphorylation at serine residues participate to maintain HSF1 in its inert state (Green et al., 1995; Shi et al., 1995; Kline and Morimoto, 1997; Knauf et al., 1996). How the stress signal is transduced and results in the de-repression Key words: Subnuclear structure, Heat shock, Transcription, Stress response MATERIALS AND METHODS Cell culture Human HeLa, A431 and HOS cells were grown in Dulbecco’s modified Eagle’s medium (DMEM) with 5% fetal calf serum (FCS). Primary epithelial and fibroblast cells were grown in DMEM with 10% FCS supplemented with essential and non-essential amino acids, vitamins, and buffered with 1 M Hepes, pH 7.4. HeLa S3 cells were grown in Joklik’s medium with 5% calf serum. Cell growth, heat shock conditions and exposure to heavy metals, amino acid analogs and anti-inflammatory drugs were as described before (Mosser et al., 1988; Jurivich et al., 1992; Sarge et al., 1993). Antibodies and indirect immunofluorescence The subcellular localization of HSF1 was determined using a panel of 2926 J. J. Cotto, S. G. Fox and R. I. Morimoto anti-HSF1 rat monoclonal antibodies (10H8, 10H4 and 4B4) generated from rat hybridoma cell lines using purified recombinant mouse HSF1 as the antigen. The monoclonal antibodies were characterized for specificity to HSF1 by ELISA and western blot analysis using purified recombinant mouse HSF1 and total cell extracts from mouse and other vertebrates (see Table 1). The epitope recognized by each monoclonal antibody was determined by western blot analysis of a collection of mouse HSF1 deletion mutants (Shi et al., 1995; Kline and Morimoto, 1997, see Fig. 1). The subcellular localization of the antigens recognized by each antibody was determined by indirect immunofluorescence. The specificity of each antibody for the control monomeric non-DNA binding form of HSF1 and the active trimer was determined by immunoprecipitation from extracts of control and heat shocked cells and by the use of the antibodies for antibody upshift assays. The results of the characterization of each antibody are summarized in Table 1. Antibodies that recognized other nuclear structures included mouse monoclonal anti-splicing factor SC-35 antibody, from Sigma (Catalog # S-4045), human autoimmune anti-kinetochore antibodies and antiNuMA, monoclonal anti-nuclear lamin A and B antibody (provided by Dr Robert Goldman, Northwestern University Medical School), antip80-coilin rabbit polyclonal antibody (provided by Dr Angus Lamond, University of Dundee and Dr Edward Chen, The Scripps Research Institute), anti-fibrillarin (provided by Dr David Spector, Cold Spring Harbor) and anti-PML monoclonal antibody (provided by Dr Luitzen de Jong, University of Amsterdam). The antiserum titer was established by sequential dilution to a range of 1:100 to 1:300 before use. Horseradish peroxidase (HRP)-conjugated goat anti-rat IgG was obtained from Pierce (catalog #31475G) and goat anti-rabbit IgG antibody was obtained from Promega (catalog # W4011). Texas Red and fluorescein (FITC)-conjugated goat anti-rat IgG (catalog # 712-095-153, 712-075153), goat anti-rabbit IgG (catalog # 711-075-152) and goat anti-mouse IgG antibodies (catalog # 715-095-151) were obtained from Jackson Immunoresearch and mouse anti-bromo-uridine triphosphate (BrUTP) monoclonal antibody was obtained from Sigma (catalog # B2531). For immunofluorescence analysis, adherent cells on coverslips were washed in 1× PBS, fixed for 10 minutes with 2% paraformaldehyde in 1× PBS at room temperature, washed twice with 1× PBS, and permeabilized with 0.1% Triton X-100. The permeabilized cells were washed twice with PBS and incubated for 1 hour with a blocking solution consisting of 1% bovine serum albumin (BSA) in PBS at room temperature prior to incubation with antibodies. Fixed cells were incubated for 1 hour at 37°C with either rat monoclonal or rabbit polyclonal anti-HSF1 antibody at dilutions of 1:100 and 1:300, respectively. To study the co-localization of HSF1 granules with other known nuclear structures, a mix of rat monoclonal anti-HSF1(1:100) with either anti-Brdu (1:100), anti-SC-35 (1:100), anti-kinetochore (1:200), anti-p80 coilin (1:100), anti-fibrillarin (1:1), anti-lamin A and B (1:00), anti-NuMA (1:100) and anti-PML (1:5) was prepared and cells were incubated for 1 hour at 37°C prior to detection. The antibodies were detected using Texas Red or FITC-conjugated goat anti-rat and goat anti-mouse and the staining pattern was analyzed by conventional epifluorescence microscopy on a Zeiss Axiophot microscope or by laser confocal microscopy. To establish the spatial organization of HSF1 granules and the relationship of HSF1 granules with other nuclear structures, Z-sections of the stained cells were collected in a confocal laser scanning microscope (Zeiss) and the data were analyzed with the program NIH Image, Vers. 1.60 for Macintosh. Nuclease treatment Cells permeabilized with 0.05% Triton X-100 were treated with either RNAse A (100 mg/ml) or DNAse I (5 units/30 ml of RNAse free DNAse) from Boehringer Mannheim at 37°C for pre-determined times and washed with PBS prior to fixation and immunostaining. Cell extraction and biochemical analysis The procedure for in situ sequential fractionation of heat shocked cells was performed as described (He et al., 1990; Bissoto et al., 1995) with some variations. Briefly, heat shocked (42°C) HeLa S3 cells were extracted in suspension with centrifugation steps (600 g, 3 minutes) between treatments. Supernatants were collected for SDS-PAGE and glycerol gradient analysis. Cells were first extracted in low ionic buffer (20 mM Hepes, pH 7.9, 25% glycerol, 0.42 M NaCl, 1.5 mM MgCl2, 0.2 mM EDTA, 0.5 mM PMSF, 0.5 mM DTT) for 5 minutes at 4°C. After a brief wash with 1× PBS, cells were extracted with cytoskeleton buffer (10 mM Pipes, pH 6.8, 100 mM NaCl, 300 mM sucrose, 3 mM MgCl2, 1 mM EGTA, 0.5% Triton X-100, 4 mM vanayl riboside complex, 1 mM PMSF) for 3 minutes at 4°C. After washing with 1× PBS, DNA was digested with 25 units/ml DNAse I for 30 minutes at 25°C in digestion buffer (essentially the same as cytoskeleton buffer but 50 mM NaCl and 0.05% Triton X-100). Chromatin was then removed by three 10 minute washes with 0.25 mM ammonium sulfate in digestion buffer to yield the nuclear matrix intermediate filament structure (He et al., 1990). An additional high salt treatment was applied by washing the nuclear matrix with 2 M NaCl in digestion buffer. As a last step the pellets were boiled in SDS sample buffer for SDS-PAGE analysis. Construction and expression of HSF1-Flag and HSF1-GFP fusion proteins To generate the carboxy-terminal epitope tagged mHSF1-Flag fusion protein, an oligonucleotide encoding an eight amino acid peptide (nDYKDDDDK-c) recognized by the monoclonal antibody Anti•Flag M2 (IBI Flag System, Kodak) was cloned at the 3′ end of the mouse HSF1 cDNA (Sarge et al., 1991) in the eukaryotic expression vector pCDNA3 (InvitroGen), and verified by sequencing across the region. The construct corresponding to HSF1-green fluorescent protein (HSF1-GFP) fusion protein was made using the eukaryotic expression vector pEGFP-N1 (Clontech) to clone the green fluorescent protein (GFP) coding sequence to the 3′ end of mouse HSF1 cDNA. For the expression of the fusion proteins, HeLa cells at a density of 40 to 50% confluence were transfected with 20 mg of DNA per 10 cm plate. Plasmid DNA was combined with 250 mM CaCl2 in a 500 ml final volume. After chilling on ice, the DNA-Ca2+ solution was added dropwise to 500 ml of 2× Hepes (N-2-hydroxyethylpiperazine- Table 1. Summary of the imunological characteristic of HSF1 specific monoclonal antibodies Western blot Immunoprecipitation Supershift assay Immunofluorescence Antibody Epitope* 37°C 42°C 37°C 42°C 37°C 42°C 37°C 42°C 10H8 10H4 378-395 295-378 +†(1,2,3,4) +†(1,3,4) +†(1,2,3,4) +†(1,3,4) +†(1,4) − +†(1,4) − N/A‡ N/A +†(1,4) +†(1,4) +†(1,2,3,4) +†(1,2,3,4) +†(1,2,3,4) +†(1,2,3,4) 4B4 425-439 +†(1,2,3,4) +†(1,2,3,4) +†(1,4) +†(1,4) N/A +†(1,4) +†(1,2,3,4) +†(1,2,3,4) *Numbers correspond to the minimal amino acid region containing the epitopes for anti-HSF1 monoclonal antibodies as detected by western blot analysis of recombinant mouse HSF1 deletion mutants (Shi et al., 1995; Kline and Morimoto, 1997). †HSF1 species reactivity: 1, human; 2, monkey; 3, rat; 4, mouse. ‡Not applicable. HSF1 granules 2927 N′-2-ethanesulfonic acid)-buffered saline (pH 7.06; 2× HeBes = 280 mM NaCl, 1.5 mM Na2HPO4, 50 mM Hepes). After 20 minutes at room temperature the precipitate was added to the cells dropwise and allowed to settle on the cells for 6 to 8 hours at 37°C. The plates were then removed from the incubator and washed twice with 10 ml of 1× phosphate-buffered saline (PBS)-1 mM ethylene glycol-bis(Baminoethyl ether)N,N,N′,N′-tetraacetic acid (EGTA) and replaced with fresh medium for 48 hours before analysis. Gel mobility shift assay and western blot analysis of HSF1 HSF1 DNA-binding activity was analyzed using the gel mobility shift assay as described previously (Mosser et al., 1988) using a 32P-labeled double-stranded synthetic HSE containing four inverted nGAAn repeats (Sarge et al., 1991). Western blot analysis was performed using whole cell extracts and rat monoclonal anti-HSF1 antibody (10H8). The immune complexes were analyzed using the ECL detection system (Amersham). RESULTS HR C 212 4B4 HR A,B 120 132 10H4 DNB 1 16 10H8 HSF1 localizes to discrete nuclear granules upon heat shock Upon exposure of HeLa cells to heat shock, HSF1 relocalizes to form brightly staining nuclear foci or granules which were detected using polyclonal antisera (Sarge et al., 1993). To further characterize these granules, we prepared a collection of rat monoclonal antibodies (10H8, 10H4 and 4B4) that specifically recognized HSF1 in cells from mouse, human and other vertebrate species. Although the epitopes for each antibody mapped to different regions of HSF1 (see Fig. 1 and Table 1), both the inactive and active trimeric form of HSF1 can be detected by various immunological assays including, immunoprecipitation and gel mobility shift-supershift assays (data not shown, see Table 1). Indirect immunofluorescence analysis using any of these monoclonal antibodies in heat shocked cells + + + + + + + + - + + + - - + - - + + - - - - - 503 378 407 mHSF1 227 451 I 288 425 II 295 498 III 395 503 IV 425 503 V 378 407 VI 439 503 4B4 10H8 VII 10H4 revealed the presence of HSF1 granules for which antibody 10H8 gave typical results (Fig. 2B). To examine whether the immunofluorescence staining pattern detected by these monoclonal antibodies was either the consequence of how the cells were prepared for indirect immunofluorescence or an unusual feature of these antibody reagents, we constructed a Flag epitope-tagged mouse HSF1 gene (mHSF1-Flag) and a green fluorescent protein (GFP)HSF1 chimeric gene for transient expression studies in HeLa cells. The endogenous and heterologous HSF1 proteins were detected by double-label immunofluorescence using rat monoclonal antibody 10H8 and the anti-Flag antiserum or the intrinsic fluorescence of GFP, respectively. The mHSF1-Flag protein (Fig. 3B) co-localizes with human HSF1 (Fig. 3A) as a general diffuse nuclear staining pattern under control conditions (Fig. 3C). Upon heat shock, mHSF1-Flag (Fig. 3E) localizes to the same granules detected with the monoclonal antibodies which recognize human HSF1 (Fig. 3D and F). Similar results of co-localization were observed using HSF1GFP (Fig. 4). Whereas GFP alone is distributed in a diffuse pattern throughout the cell under control (Fig. 4A) and heat shocked conditions (Fig. 4B), mHSF1-GFP is primarily nuclear in control cells (Fig. 4C) and upon heat shock relocalizes to form HSF1 granules (Fig. 4D). These results establish that the heat shock induced formation of HSF1 granules are a specific property of HSF1, they can be detected with an epitope tag and in living stressed cells. We next examined whether HSF1 granules are found in other human tissue culture cell lines. Primary fibroblasts and epithelial cells were exposed to various heat shock temperatures and examined by indirect immunofluorescence. Brightly staining HSF1 granules were readily detected upon exposure of either primary cell to heat shock with the majority of cells containing two large foci and occasional smaller speckles (Fig. 5B and D). The optimal temperatures required to detect HSF1 granules in primary human cells was higher (43-45°C) than required in transformed human cells (see companion paper in this issue: Jolly et al., 1997). Examination of the HSF1 staining pattern in two other human transformed cell lines, HOS (hyperdiploid osteosarcoma cell line) and A431 (hypotetraploid epidermal carcinoma cell line) revealed that HOS cells have an average of 4 to 5 granules per nucleus (Fig. 5F) and HeLa (hypotetraploid cervical carcinoma cell line) and A-431 cells contain Fig. 1. Schematic representation of various mouse HSF1 deletion mutants and a summary of the pattern of recognition detected by anti-HSF1 monoclonal antibodies 10H8, 10H4 and 4B4. E. coli whole cell extracts expressing the recombinant mouse HSF1 deletion mutants were analyzed by western blot and the minimal peptide regions recognized by the different antibodies are indicated. Conserved structural motifs corresponding to DNA binding domain (DNB) and heptad repeats A,B and C (HR A,B and C) are shown. Fig. 2. HSF1 nuclear granules are formed upon heat shock in HeLa cells. Cells cultured at 37°C or heat shock at 42°C were subjected to immunofluorescence analysis using anti-HSF1 rat monoclonal antibody 10H8 (A and B). Bar, 5 µm. 2928 J. J. Cotto, S. G. Fox and R. I. Morimoto HSF1-Flag Fig. 3. Immunofluorescence analysis of HeLa cells transfected with mouse HSF1Flag expression vector. Transfected HeLa cells were cultured at 37°C and stained with anti-HSF1 (A) or anti-FLAG (B) antibodies. (C) Co-localization of HSF1 and mHSF1FLAG. (D-F) Cells incubated at 42°C heat shock and treated as in A-C. Bar, 5 µm. Heat Shock Control HSF1 an average of 7 granules per nucleus (Fig. 5H and J). The number of HSF1 granules per cell is relatively constant in any particular cell line with fewer granules being detected in primary cells than in transformed cell lines. These results and those of Jolly et al. (1997) suggest a possible relationship between the number of granules and chromosomal ploidy. The kinetics of HSF1 granule formation parallels the activation of HSF1 in stressed cells The activation of HSF1 is associated with a series of rapidly occurring events including oligomerization of the non-DNA binding monomer to the DNA binding trimer, inducible serine phosphorylation, and transcriptional induction of heat shock genes (Sorger and Pelham, 1988; Baler et al., 1993; Sarge et al., 1993; Cotto et al., 1996). During continuous exposure to heat shock, HSF1 activity attenuates as reflected by the loss of transcriptional activity, dephosphorylation, and conversion of trimers to monomers (Abravaya et al., 1991a,b; Sarge et al., 1993). Therefore, we examined whether the appearance of HSF1 granules is temporally associated with its activity as a transcriptional activator. Within 30 minutes of heat shock, HSF1 granules ranging in size from 0.5 to 1.5 µm were detected in 80-90% of HeLa cells (Fig. 6A), and by 60 minutes of heat shock, 95% of the cells exhibited brightly staining HSF1 granules. Up through two hours of heat shock, HSF1 granules were ubiquitous in all cells. Analysis of the size distribution of the granules in HeLa cells heat shocked at 42°C for 2 hours reveals that they can be described as two populations of which 60% of the foci correspond to smaller (0.5 to 1.5 µm) brightly staining granules or speckles and the remaining are larger (1.5 to 2.5 µm) clustered or ring-like granular structures (Fig. 7A). The majority (55%) of the cells contained an average of 7 HSF1 granules, although there was substantial cell-to-cell variation (Fig. 7B). During continuous exposure to heat shock, both the fluorescence intensity and the numbers of HSF1 foci increased. Comparison to the level of HSF1 DNA binding activity (Fig. 6B) reveals that the appearance of HSF1 granules correlates closely with both the acquisition of HSF1-DNA binding activity and the inducibly phosphorylated state of HSF1 (Fig. 6B and C). After 2 hours of continuous heat shock, the fraction of cells which exhibit HSF1 granules rapidly declines to 5% of the population; likewise HSF1 DNA binding attenuates and is dephosphorylated to the control state (Fig. 6B and C). The activation of HSF1 is a multi-step process which involves the stable appearance of intermediate states (Jurivich et al., 1992; Lee et al., 1995; Cotto et al., 1996). To examine whether the appearance of HSF1 granules reflects the formation of HSF1 DNA binding trimers alone or requires complete activation to Fig. 4. mHSF1-GFP fusion protein localizes to nuclear granules in heat shocked HeLa cells. HeLa cells at 37°C or heat shocked at 42°C were transfected with the GFP vector (A and B) or the mHSF1-GFP vector (C,D) and visualized by confocal microscopy. Bar, 5 µm. HSF1 granules 2929 Fig. 5. Subcellular localization of HSF1 in various primary and transformed human cells. Cells either at control or 42°C heat shock conditions were analyzed by immunofluorescence using rat monoclonal antibody 10H8. Human primary fibroblasts (A,B), epithelial cells (C,D), human HOS cells (E,F), HeLa (G,H), and A431 (I,J) cells were examined. Bar, 5 µm. Fig. 6. Kinetics of HSF1 granule formation and comparison with DNA binding activity and inducible phosphorylation during heat shock. (A) Immunofluorescence analysis in HeLa cells prior to heat shock (0 minutes) or after incubation at 42°C for 30, 60, 120, 180 and 240 minutes using HSF1 rat monoclonal antibody 10H8. (B) Kinetics of HSF1 DNA binding activity measured using gel mobility shift analysis, and (C) western blot analysis of whole cell extracts from samples indicated in B. The slower mobility of HSF1 in heat shocked cells is due to inducible serine phosphorylation. Bar, 5 µm. 2930 J. J. Cotto, S. G. Fox and R. I. Morimoto ditionally, these results reveal that HSF1 granules are not the consequence of heat shock-induced aggregation of HSF1. Fig. 7. Quantitative analysis of the number and size of HSF1 granules in HeLa cells. (A) The average diameter was calculated for 200 granules using the program NIH image. In HeLa cells, HSF1 granules can be divided in two sub-populations based on their average size; small granules (~0.5 to 1.6 µm) and large granules (~1.6 to 3 µm). (B) The number of HSF1 granules was established by analysis of 100 nuclei. The mean number of HSF1 granules/cell is 6.8±2.4. the transcriptionally competent trimer state, HeLa cells were exposed to three inducers of HSF1 activity. Sodium salicylate induces HSF1 trimers which are nuclear-localized, transcriptionally inert, and not-inducibly phosphorylated (Jurivich et al., 1992; Cotto et al., 1996), the amino acid analogue azetidine induces a transcriptionally active form of HSF1 which is not inducibly phosphorylated, and the heavy metal cadmium induces a transcriptionally active and inducibly phosphorylated form of HSF1 (Sarge et al., 1993) (Fig. 8). Although each of these conditions activate equivalent levels of HSF1 DNA binding activity (Fig. 8A), HSF1 granules were only detected in azetidine or cadmium treated cells and not in salicylate treated cells (Fig. 8B). These results reveal, that the appearance of HSF1 granules is a reliable visual indicator of the transcriptional activity of HSF1 and that HSF1 granules are induced by other stresses which activate the heat shock response. Ad- HSF1 granules are novel sub-nuclear structures To determine whether HSF1 granules represent a novel nuclear compartment or correspond to previously characterized subnuclear structures, we used double immunofluorescence with a number of antibodies to nuclear antigens and analysis by laser confocal microscopy. The fluorescence labeling patterns of each antibody and 8 to 10 horizontal optical sections of each field were scanned from top to bottom of the cell and the results of stacked images are presented. The data in Fig. 9 represent the superimposed images of fluorescein (FITC) and Texas Red-coupled secondary antibody labeling of different primary antibodies. In Fig. 9A, the sites of DNA replication were marked by incorporation of BrdU prior to heat shock and detected with anti-BrdU antibody. No co-localization of HSF1 granules with sites of DNA replication were detected. Likewise, the immunofluorescence pattern of mitotic cells stained with anti-HSF1 and anti-kinetochore antibody (Fig. 9B), the anti-splicing factor SC-35 antibody (Fig. 9C), and the coiled body marker anti-p80-coilin antibody (Fig. 9D) did not reveal co-localization with HSF1 granules. In this study, no heterogeneity was noticed in the appearance or morphology of the HSF1 granules during the cell cycle. As previous studies have shown that the nucleolar morphology is affected by heat shock (Welch and Suhan, 1985), we investigated whether the HSF1 granules would correspond to an accumulation of the factor into nucleoli. This assumption was strongly supported by the granular morphology of HSF1 foci which is similar to that of nucleoli detected with a marker of the dense fibrillar center (Roussel et al., 1993). HSF1 was detected by immunofluorescence together with a marker of the dense fibrillar center, the UBF cofactor (upstream binding factor) using an anti-UBF antibody (Roussel et al., 1993), however, no co-distribution between HSF1 foci and nucleoli was observed (Fig. 9E). Likewise, HSF1 granules do not colocalize with PML, a nuclear localized protein involved in promyelocytic leukemia (Koken et al., 1994; Weis et al., 1994) or with nuclear matrix proteins, such as nuclear lamins (A or B) and NuMA (nuclear matrix mitotic apparatus protein) which have been shown to accumulate into nuclear domains at specific points in the cell cycle (Moir et al., 1994) (data not shown). Overall, these results clearly demonstrated that HSF1 accumulates into discrete nuclear substructures which are distinct from other previously described nuclear granules. Co-localization experiments carried out in our laboratory have clearly shown that HSF1 granules do not contain other heat shock transcription factors (data not shown). Indeed, in earlier studies, HSF2, HSF3, and HSF4 have been visualized as nuclear speckles, not granules (Sarge et al., 1993; Sheldon and Kingston, 1993; Nakai et al., 1995, 1997). Furthermore, these speckles are detected constitutively and do not correlate with gene activation and heat shock response. In contrast, the appearance of HSF1 granules correlates with transformation of the factor from the inert to the active state and with gene activation. Recently, an increasing number of transcription factors, including the mineralocorticoid and glucocorticoid receptors, the haemopoietic factors GATA-1 and -3, and p53, have been shown to accumulate into nuclear domains (Jackson et al., 1994; van Steensel et al., 1995, 1996; Elefanty et al., 1996). Although we have not performed a comprehensive comparison HSF1 granules 2931 Fig. 8. Effects of different stress conditions on the activation of HSF1 and the formation of HSF1 granules. (A) Gel mobility shift analysis of HSF1 DNA binding activity in whole cell extracts from HeLa cells at control (37°C) conditions or treated with 20 mM salicylate, 5 mM azetidine, 30 µM CdSO4, or 42°C heat shock. (B) Intracellular localization of HSF1 in HeLa cells exposed to conditions indicated in A and stained with monoclonal anti-HSF1 antibody. Bar, 5 µm. Control 20 mM salicylate between HSF1 and each one of these transcription factors, HSF1 does not co-localize with members of the GATA family (data not shown) thus ruling out a common mechanism for compartmentalization of transcription factors. To assess whether nucleic acids are a component of HSF1 granules, heat shocked cells were permeabilized and incubated with either DNAse I or RNAse A (Fig. 10). Interestingly, the number, size, or distribution of HSF1 granules were not affected despite the substantial reduction in nuclear DNA observed in DNase I treated cells as detected by staining with Hoechst dye (Fig. 10A). Likewise, RNAse A treatment did not have an effect on HSF1 granules (Fig. 10B, a-c). These results reveal that the general features of the granules are not altered by depolymerization of nucleic acids. However, given the relatively large size of the HSF1 granules, it is also possible that their location within the nucleus may be unaffected even if nucleic acids are an important structural component. The possible association of HSF1 granules with the nuclear matrix was examined using biochemical fractionation conditions known to extract the nuclear matrix (He et al., 1990; Bissoto et al., 1995). The majority of the HSF1 which was removed by extraction with a low ionic strength buffer (Fig. 11, compare lanes 1 and 2) corresponds to trimeric HSF1; the remaining HSF1 can be extracted with the non-ionic detergent Triton X-100 (lane 4). HSF1 was not detected in the insoluble pellet corresponding to the nuclear matrix (lane 7). The ease of extraction of HSF1 by low ionic buffer and relatively mild non-ionic detergents reveal that HSF1 is neither in an insoluble fraction nor associated specifically with the nuclear matrix. 5 mM azetidine 30 µM CdSO4 Heat shock (42°C) These results also suggest that HSF1 granules are labile and readily disrupted even upon gentle lysis of the nucleus. DISCUSSION HSF1 granules represent a unique class of subcellular structures which appear transiently in the nucleus of human cells when heat shock genes are transcriptionally induced and disappear rapidly during attenuation of the heat shock response as the transcription of heat shock genes diminishes to control levels. These results establish HSF1 granules as a novel dynamic feature of the heat shock response and underscores the potential for new information on the effects of stress on nuclear structure. The detection of HSF1 granules is likely to reflect the general response to stress as exposure of human cells in culture to heat shock, cadmium and azetidine gave indistinguishable results. Furthermore, these results rule out the possibility that HSF1 granules result from the aggregation of HSF1 at elevated temperatures, as other stresses which induce these granules are effective at 37°C. Additionally, the effects of elevated temperatures and other stresses on the biochemical properties of HSF1 result in the oligomerization to a trimeric DNA binding state rather than association of HSF1 with the nuclear matrix or other high molecular sized nuclear structures. Another observation presented here which links the appearance of HSF1 granules with the transcriptional activity of heat shock genes is the absence of HSF1 granules in salicylate-treated 2932 J. J. Cotto, S. G. Fox and R. I. Morimoto Fig. 9. HSF1 does not co-localize with other known subnuclear structures. Immunofluorescence analysis of double-stained HeLa cells. Green channel represents HSF1 staining and red channel represents (A) DNA replication sites (anti-Brdu), (B) kinetochores, (C) splicing factor SC35, (D) coiled body and (E) nucleolar dense fibrillar center staining. Bar, 5 µm. cells. Although salicylate treatment induces HSF1 trimers which exhibit complete DNA binding activity, this form of HSF1 is transcriptionally inert. In contrast, stress-inducers such as heavy metals, amino acid analogues, and heat shock results in the fully active form of HSF1 which corresponds with stress-induced granules. Finally, the temporal link in the well established kinetics of the heat shock response and the rapid recovery during attenuation parallels precisely the transient appearance and disappearance of HSF1 granules. Taken Fig. 10. Analysis of the effects of nuclease treatment on HSF1 granules. (A) Heat shocked HeLa cells grown on coverslips were permeabilized in 0.05% Triton X100 and incubated in DNAse I for 0, 15 or 30 minutes prior to immunofluorescence analysis with HSF1 monoclonal antibody 10H8 (a-c). Reduction of Hoechst staining (d-f) indicates digestion of DNA in these cells. (B) Same as A, but cells were treated with RNAse A prior to immunofluorescence analysis (a-c). Treatment with RNAse did not affect Hoechst staining (d-f). together, these results reveal that HSF1 granules represent a new component of the heat shock response in human cells. Our results and those presented in the accompanying paper reveal that HSF1 granules are in all human cells. The relationship between the number of granules per nucleus and chromosomal ploidy noted here and by the accompanying paper (Jolly et al., 1997) is intriguing and suggests the existence of specific chromosomal targets for HSF1 foci. The many unique features of HSF1 granules encouraged us HSF1 granules 2933 Fraction no. 1 2 3 4 5 6 7 69 kDa Fig. 11. Biochemical fractionation of heat shocked cells. (A) Heat shocked (42°C) HeLa S3 cell pellets (approximately 2×107 cells/pellet) were either directly solubilized with SDS-sample buffer to detect total amount of HSF1 in the cell (lane 1) or extracted sequentially with Buffer C (lane 2), a 1× PBS wash (lane 3), 0.5% Triton X-100 in cytoskeleton buffer (lane 4), DNAse (25 units/ml) and 0.25 M ammonium sulfate to remove chromatin (lane 5), a high salt wash in 2 M NaCl (lane 6) and the insoluble material was solubilized directly in SDS-sample buffer (lane 7). Extracted material was dialyzed and lyophilized before solubilization in sample buffer (lanes 2-6). Extracts were analyzed directly by western blot analysis using monoclonal antibody 10H8. to redouble our efforts to ensure that the granules could be detected by other reagents or methodologies which did not depend on indirect immunofluorescence or traditional forms of cell fixation. Two complementary approaches were described in which an epitope tagged HSF1 allowed the detection of HSF1 granules via an antibody to a heterologous epitope and by direct visualization of HSF1 granules in living cells using an HSF1GFP fusion protein. Thus, the detection of HSF1 granules is not an artifact of either the polyclonal or the newly characterized monoclonal antibodies. The ability to detect HSF1 granules in living heat shocked cells by direct visualization of HSF1-GFP chimeric protein also rules out possible artifacts inherent in the procedures employed for visualization and detection of the antiHSF1 antibodies. Additionally, the latter result reveals that other proteins (e.g. GFP) can be translocated into HSF1 granules. Our knowledge on nuclear structure, while still limited, has grown rapidly in recent years. Most nuclear functions including replication, transcription, RNA splicing and RNA transport are localized within the nucleus rather than distributed diffusely (reviewed by de Jong et al., 1990; van Driel et al., 1991, 1995; Moen et al., 1995). Among the most prominent sub-nuclear structures is the nucleolus which organizes events associated with the transcription and processing of ribosomal RNA (Scheer and Benavente, 1990; Hernandez-Verdun, 1991). The structure and function of other nuclear domains are less well characterized, in part because functional compartmentalization is less apparent. Some nuclear domains for which a function has been suggested include the small nuclear RNP clusters involved in pre-mRNA processing (Piñol-Roma et al., 1989; Fu and Maniatis, 1990; Spector, 1990; Lamond and Carmo-Fonseca, 1993b) and coiled bodies, which may play a role in spliceosome assembly or recycling and in the metabolism of small nuclear RNAs (Carmo-Fonseca et al., 1991; Lamond and CarmoFonseca, 1993a; Bohmann et al., 1995). Analysis of the subcellular distribution of certain transcription factors also reveals a non-homogeneous distribution within the nucleus (Jackson et al., 1994; van Steensel et al., 1995, 1996; Elefanty et al., 1996). Yet, for each of these factors, granules or speckles are constitutive and do not appear to correspond with other functionally characterized sub-nuclear structures. Indeed, other heat shock factors (HSF2, 3, 4) have been visualized as nuclear speckles (Sarge et al., 1993; Sheldon and Kingston, 1993; Nakai et al., 1995, 1997). However, these speckles are present constitutively and do not correlate with gene activation and heat shock response, and they are distinct from the HSF1 granules. Only in the case of HSF1 does the appearance of these granules correlate with the activation of the factor from the inert to active states. Despite the cell biological analysis and biochemical characterization presented in this study and the accompanying paper (Jolly et al., 1997), the role of HSF1 granules in the heat shock response remains enigmatic. Perhaps this is because the analysis only encompasses our current expectations of a traditional role for HSF1 in the heat shock response. While there has been substantial progress to understand the organization and regulation of HSFs, only a few molecular targets for HSF binding, corresponding to the Hsp90 and Hsp70 genes, have been characterized. Others have shown that HSFs are associated with other chromosomal loci and suggested that HSFs may have roles in the repression of non heat-shock genes during heat shock (Westwood et al., 1991; Cahill et al., 1996). Indeed, heat shock causes a complete arrest in the transcription of non-heat shock genes by mechanisms which have yet to be addressed. We thank Dr Robert Goldman for the anti-lamin A and B monoclonal antibody and the NuMA autoantibody, Dr Angus Lamond for the anti-coilin polyclonal antibody, Dr Edward Chen for the anti-p80coilin polyclonal antibody R288, Dr David Spector for anti-fibrillarin autoantibodies, and Dr Luitzen de Jong for the anti-PML monoclonal antibody 5E10. We thank Dr Claire Vourc’h and Caroline Jolly for their valuable discussion during the course of this work and their comments on the manuscript and Dr Robert Holmgren for advice in microscopy. These studies were supported by a grant to R.I.M. from the National Institutes of General Medical Sciences. REFERENCES Abravaya, K., Phillips, B. and Morimoto, R. I. (1991a). Heat shock-induced interactions of heat shock transcription factor and the human hsp70 promoter examined by in vivo footprinting. Mol. Cell. Biol. 11, 586-592. Abravaya, K., Phillips, B. and Morimoto, R. I. (1991b). Attenuation of the heat shock response in HeLa cells is mediated by the release of bound heat shock transcription factor and is modulated by changes in growth and heat shock temperatures. Genes Dev. 5, 2117-2127. Baler, R., Dahl, G. and Voellmy, R. (1993). Activation of human heat shock genes is accompanied by oligomerization, modification, and rapid translocation of heat shock transcription factor HSF1. Mol. Cell. Biol. 13, 2486-2496. Bisotto, S., Lauriault, P., Duval, M. and Vincent, M. (1995). Colocalization of a high molecular mass phosphoprotein of the nuclear matrix (p255) with spliceosome. J. Cell. Sci. 108, 1873-1882. Bohmann, K., Ferreira, J., Santama, N., Weis, K. and Lamond, A. I. (1995). Molecular analysis of the coiled body. J. Cell Sci. 19, 107-113. Cahill, C. M., Waterman, W. R., Xie, Y., Auron, P. E. and Calderwood, S. K. (1996). Transcriptional repression of the prointerleukin 1beta gene by heat shock factor 1. J. Biol. Chem. 271, 24874-24879 Carmo-Fonseca, M., R. Pepperkok, B. S. Sproat W. Ansorge, M. S. Swanson and Lamond, A. I. (1991). In vivo detection of snRNP-rich organelles in the nuclei of mammalian cells. EMBO J. 10, 1863-1873. Cotto, J. J., Kline, M. and Morimoto, R. I. (1996). Activation of heat shock factor 1 DNA binding precedes stress induced serine phophorylation: Evidence for a multistep pathway of regulation. J. Biol. Chem. 271, 3355-3358. de Jong, L., vanDriel, R., Stuurman, N., Meejne, A. M. L. and van Renswoude, J. (1990). Principles of nuclear organization. Cell Biol. Intl. Rep. 14, 1051-1074. Elefanty, A. G., Antoniou, M., Custodio, N., Carmo-Fonseca, M. and 2934 J. J. Cotto, S. G. Fox and R. I. Morimoto Grosveld, F. G. (1996). GATA transcription factors assoicate with a novel class of nuclear bodies in erytoblasts and megakaryocytes. EMBO J. 15, 319-333. Fu, X. D. and Maniatis, T. (1990). Factor required for mammalian spliceosome assembly is localized to discrete regions in the nucleus. Nature 343, 437-441. Green, M., Schuetz, T. J., Sullivan, E. K. and Kingston, R. E. (1995). A heat shock-responsive domain of human HSF1 that regulates transcription activation domain function. Mol. Cell. Biol. 15, 3354-3362. He, D., Nickerson, J. A. and Penman, S. (1990). Core filaments of the nuclear matrix. J. Cell. Biol. 110, 569-580. Hernandez-Verdun, D. (1991). The nucleolus today. J. Cell Sci. 99, 465-471. Jackson, D. A., Hassan, A. B., Errington, R. J. and Cook, P. R. (1994). Sites in human nuclei where damage induced by ultraviolet light is repaired: localization relative to transcription sites and concentrations of proliferating cell nuclear antigen and the tumour suppressor protein, p53. J. Cell Sci. 107, 1753-1760. Jolly, C., Morimoto, R. I., Robert-Nicoud, M. and Vourc’h, C. (1997). HSF1 transcription factor concentrates in nuclear foci during heat shock: II. Relationship with transcription sites. J. Cell Sci. 110, 2935-2941. Jurivich, D. A., Sistonen L., Kroes R. A. and Morimoto, R. I. (1992). Effect of sodium salicylate on the human heat shock response. Science 255, 1243-1245. Kline, M. P. and Morimoto, R. I. (1997). Repression of the heat shock factor 1 transcriptional activation domain is modulated by constitutive phophorylation. Mol. Cell. Biol. 17, 2107-2115. Knauf, U., Newton, E. M., Kyriakis, J. and Kingston, R. E. (1996). Repression of human heat shock factor 1 activity at control temperature by phosphorylation. Genes Dev. 10, 2782-2793. Koken, M. H., Puvion-Dutilleul, F., Guillemin, M. C., Viron, A., LinaresCruz, G., Stuurman, N., de Jong, L., Szostecki, C., Calvo, F. and Chomienne, C. (1994). The t(15;17) translocation alters a nuclear body in a retinoic acid-reversible fashion. EMBO J. 13, 1073-1083. Lamond, A. I. and Carmo-Fonseca, M. (1993a). The coiled body. Trends Cell Biol. 3, 198-204. Lamond, A. I. and Carmo-Fonseca, M. (1993b). Localization of splicing snRNPs in mammalian cells. Mol. Biol. Rep. 18, 127-133. Lee, B. S., Chen J., Angelidis C., Jurivich D. A. and Morimoto, R. I. (1995). Pharmacological modulation of Heat Shock Factor 1 by anti-inflammatory drugs results in protection against stress-induced cellular damage. Proc. Nat. Acad. Sci. USA 92, 7207-7211. Lindquist, S. (1986). The heat shock response. Ann. Rev. Biochem. 55, 11511191. Moen, P. T., Smith, K. P. and Lawrence, J. B. (1995). Compartmentalization of specific pre-mRNA metabolism: an emerging view. Hum. Mol. Genet. 4, 1779-1789. Moir R. D., Montag-Lowy M. and Goldman R. D. (1994). Dynamic properties of nuclear lamin: Lamin B is associated with sites of DNA replication. J. Cell Biol. 125, 1201-1212. Morimoto, R. I., Tissieres, A. and Georgopoulos, C. (1990). Stress Proteins in Biology and Medicine. (Ed. R. I. Morimoto, A. Tissieres and C. Georgopoulos). Cold Spring Harbor, N. Y. Cold Spring Harbor Laboratory Press, 1-36. Morimoto, R. I., Jurivich, D. A., Kroeger, P. E., Mathur, S. K., Murphy, S. P., Nakai, A., Sarge, K. D., Abravaya, K. and Sistonen, L. T. (1994). The Biology of Heat Shock Proteins and Molecular Chaperones. (Ed. R. I. Morimoto, A. Tissieres and C. Georgopoulos). Cold Spring Harbor, N. Y. Cold Spring Harbor Laboratory Press, 417-456. Mosser, D. D., Theodorakis, N. G. and Morimoto, R. I. (1988). Coordinate changes in heat shock element-binding activity and hsp70 gene transcription rates in human cells. Mol. Cell. Biol. 8, 4736-4744. Nakai, A. and Morimoto, R. I. (1993). Characterization of a novel chicken heat shock transcription factor, HSF3, suggests a new regulatory pathway. Mol. Cell. Biol. 13, 1983-1997. Nakai, A., Y. Kawazoe, M. Tanabe, K. Nagata and Morimoto, R. I. (1995). The DNA-binding properties of two heat shock factors, HSF1 and HSF3, are induced in the avian erythroblast cell line HD6. Mol. Cell. Biol. 15 (10), 5268-5278. Nakai, A., Tanabe, M., Kawazoe, Y., Inazawa, J., Morimoto, R. I. and Nagata, K. (1997). HSF4 a New Member of the human heat shock factor family which lacks properties of a transcriptional activator. Mol. Cell. Biol. 17, 469-481. Piñol-Roma, S., Swanson M. S., Gall J. G. and Dreyfuss, G. (1989). A novel heterogeneous nuclear RNP protein with a unique distribution on nascent transcripts. J. Cell Biol. 109, 2575-2587. Rabindran, S. K., Giorgi, G., Clos, J. and Wu, C. (1991). Molecular cloning and expression of a human heat shock factor, HSF1. Proc. Nat. Acad. Sci. USA 88, 6906-6910. Roussel, P., André, C., Masson, C., Géraud, G. and Hernandez-Verdun, D. (1993). Localization of the RNA polymerase I transcription factor hUBF during the cell cycle. J. Cell Sci. 104, 327-337. Sarge, K. D., Zimarino, V., Holm, K., Wu, C. and Morimoto, R. I. (1991). Cloning and characterization of two mouse heat shock factors with distinct inducible and constitutive DNA-binding ability. Genes Dev. 5, 1902-1911. Sarge, K. D., Murphy, S. P. and Morimoto, R. I. (1993). Activation of heat shock gene transcription by HSF1 involves oligomerization, acquisition of DNA binding activity, and nuclear localization and can occur in the absence of stress. Mol. Cell. Biol. 13, 1392-1407. Sarge, K. D., Park-Sarge, O.-K., Kirby, J. D., Mayo, K. E. and Morimoto, R. I. (1994). Expression of heat shock factor 2 in muse testis: potential roleaas a regulator of heat shock protein gene expression during spermatogenesis. Biol. Reprod. 50, 1334-1343. Scheer, U. and Benavente, R. (1990). Functional and dynamic aspects of the mammalian nucleolus. BioEssays. 12, 14-21. Sheldon, L. A. and Kingston, R. E. (1993). Hydrophobic coiled-coil domains regulate the subcellular localization of human heat shock factor 2. Genes Dev. 7, 1549-1558. Shi, Y., Kroeger, P. E. and Morimoto, R. I. (1995). The carboxyl-terminal transactivation domain of heat shock factor 1 is negatively regulated and stress responsive. Mol. Cell Biol. 15, 4309-4318. Schuetz, T. J., Gallo, G. J., Sheldon, L., Tempst, P. and Kingston, R. E. (1991). Isolation of a cDNA for HSF2: evidence for two heat shock factors genes in humans. Proc. Nat. Acad. Sci. USA. 88, 6911-6915. Sistonen, L., Sarge, K. D., Phillips, B., Abravaya, K. and Morimoto, R. I. (1992). Activation of heat shock factor 2 during hemin-induced differentiation of human erythroleukemia cells. Mol. Cell. Biol. 12, 4104-4111. Sistonen, L., Sarge, K. D. and Morimoto, R. I. (1994). Human heat shock factors 1 and 2 are differentially activated and can synergistically induce hsp70 gene transcription. Mol. Cell. Biol. 14, 2087-2099. Sorger, P. K. and Pelham, H. R. B. (1988). Yeast heat shock factor is an essential DNA-binding protein that exhibits temperature-dependent phosphorylation. Cell 54, 855-864. Spector, D. L. (1990). Higher order nuclear organization: three dimensional distribution of small nuclear ribonucleoprotein particles. Proc. Nat. Acd. Sci. USA. 87, 147-151. van Driel, R., Humbel B. and de Jong, L. (1991). The nucleus: a black box being opened. J. Cell Biochem. 47, 1-6. van Driel, R., Wansink, D. G., van Steensel, B., Grande, M. A., Schul, W. and de Jong, L. (1995). Nuclear domains and the nuclear matrix. Int. Rev. Cyt. 162A, 151-189. van Steensel, B., Brink, M., van der Meulen, K., van Binnendijk, E. P., Wansink, D. G., de Jong, L., de Kloet, E. R. and van Driel, R. (1995). Localization of the glucocorticoid receptor in discrete clusters in the cell nucleus. J. Cell Sci. 108, 3003-3011. van Steensel, B., van Binnendijk, E. P., Hornsby, C. D., van der Voort, H. T. M., Krozowski, Z. S., de Kloet, E. R. and van Driel, R. (1996). Partial colocalization of glucocorticoid and mineralocorticoid receptors in discrete compartments in nuclei of rat hippocampus neurons. J. Cell Sci. 109, 787-792. Weis, K., Rambaud, S., Lavau, C., Jansen, J., Carvalho, T., CarmoFonseca, M., Lamond, A. and Dejean, A. (1994). Retinoic acid regulates aberrant nuclear localization of PML-RAR alpha in acute promyelocytic leukemia cells. Cell 76, 345-356. Welch, W. J. and Suhan, J. P. (1985). Morphological study of the mammalian stress response: characterization of changes in cytoplasmic organelles, cytoskeleton, and nucleoli, and appearance of intranuclear actin filaments in rat fibroblasts after heat-shock treatment. J. Cell Biol. 101, 1198-1211. Westwood, J. T., Clos, J. and Wu, C. (1991). Stress-induced oligomerization and chromosomal relocalization of heat-shock factor. Nature 353, 822-827. Wu, C. (1995). Heat shock transcription factors: structure and regulation. Annu. Rev. Cell Dev. Biol. 11, 441-469. (Accepted 1 October 1997)