HSF1 (heat-shock factor 1) is the master regulator of the heat-shock response; however, it is also activated by cancer-associated stresses and supports cellular transformation and cancer progression. We examined the role of HSF1 in relation to cancer cell clonogenicity, an important attribute of cancer cells. Ectopic expression or HSF1 knockdown demonstrated that HSF1 positively regulated cancer cell clonogenic growth. Furthermore, knockdown of mutant p53 indicated that HSF1 actions were mediated via a mutant p53-dependent mechanism. To examine this relationship more specifically, we ectopically co-expressed mutant p53R273H and HSF1 in the human mammary epithelial cell line MCF10A. Surprisingly, within this cellular context, HSF1 inhibited clonogenicity. However, upon specific knockdown of endogenous wild-type p53, leaving mutant p53R273H expression intact, HSF1 was observed to greatly enhance clonogenic growth of the cells, indicating that HSF1 suppressed clonogenicity via wild-type p53. To confirm this we ectopically expressed HSF1 in non-transformed and H-RasV12-transformed MCF10A cells. As expected, HSF1 significantly reduced clonogenicity, altering wild-type p53 target gene expression levels consistent with a role of HSF1 increasing wild-type p53 activity. In support of this finding, knockdown of wild-type p53 negated the inhibitory effects of HSF1 expression. We thus show that HSF1 can affect clonogenic growth in a p53 context-dependent manner, and can act via both mutant and wild-type p53 to bring about divergent effects upon clonogenicity. These findings have important implications for our understanding of HSF1's divergent roles in cancer cell growth and survival as well as its disparate effect on mutant and wild-type p53.
HSF1 (heat-shock factor 1) is transcriptionally activated by cells in response to a variety of extrinsic and intrinsic stresses, including heat shock, oxidative stress, nutrient deprivation and oncogene activation . Its activation results in the expression of the highly conserved family of HSPs (heat-shock proteins), which, upon acute and chronic forms of stress, function as molecular chaperones, maintaining intracellular protein homoeostasis, as well as providing cytoprotection to limit stress-induced cell death. Consistent with this role, the action of HSF1 in malignancy has long been seen as indirect, via its transcriptional regulation of HSPs and its provision of cytoprotection; however, it has recently emerged that HSF1 can directly co-ordinate a vast number of transcriptional networks that are unique to the malignant state and are distinct from the heat-shock response . Although the exact mechanisms by which HSF1 may achieve this control are still to be fully elucidated, it is thought that the cellular context, and the unique interactions of HSF1 therein, may be an important determinant in eliciting the unique transcriptional networks.
The actions of HSF1 in regulating these networks results in the support of fundamental processes within the cancer cell that maintains its ‘fitness’, such as protein translation, glucose metabolism, cell-cycle control and ribosome biogenesis [2,3]. Consistent with this, HSF1 has been shown to support and promote the oncogenic activity of a number of oncogenes, such as Ras, ErbB2, heregulin β1 and PDGF (platelet-derived growth factor)-B [3–5]. Previous studies have also demonstrated that HSF1 is required for lymphoma development in p53-knockout mice and protects mice from tumours induced by oncogenic p53R172H .
Although decreased levels of HSF1 are implicated in aging and protein-folding diseases, such as neurodegenerative diseases , consistent with a role in tumorigenesis and cancer progression, HSF1 expression has been shown to be increased in a number of cancer types and has been strongly associated with cancer progression and poor prognosis [2,7,8].
A feature of the malignant cancer cell is its ability to survive and grow in isolation, or its clonogenicity, marking the cancer cell's ability for unlimited proliferation [9,10]. This feature has been associated with cancer ‘stem-like’ properties that allow for increased tumour- and metastasis-initiating capacities [10–13]. It has also long been recognized that many factors can positively or negatively affect cancer cell clonogenicity. Among these, WT (wild-type) p53 has been shown to negatively affect clonogenicity, whereas mutated forms of the tumour suppressor are known to increase the clonogenic capacity of cancer cells [14–16].
WT p53 mediates its tumour suppressor actions via transcriptional pathways that regulate the expression of genes involved in DNA damage repair, cell-cycle arrest and apoptosis [17,18]. Mutation of p53 is the most frequent genetic change identified in cancer, with more than 50% of all cancers exhibiting a loss or mutation of the gene. Expression of mutant p53 is not simply equivalent to p53 loss, but can exert ‘gain-of-function’ properties that have been shown to be important at key stages of metastatic progression, via the promotion of cancer cell migration, invasion, survival and chemoresistance .
Although it has been suggested that HSF1 is required for mutant p53 activity, during genotoxic stress HSF1 is known to mediate pro-apoptotic actions by the modulation of WT p53. However, previous studies have provided conflicting reports on the effect of HSF1 depletion on WT p53. Although some studies have demonstrated that HSF1 regulates WT p53 proteasomal degradation, leading to the increase in p53 levels and activity upon HSF1 depletion [19,20], other studies have found that HSF1 depletion abrogates WT p53 activity, as HSF1 is required for normal WT p53 transactivation activity and nuclear translocation [18,21,22]. Therefore the full molecular and biological consequences of HSF1 activity on WT p53 within cancer are still to be fully elucidated. Moreover, knowledge regarding the positive effect of HSF1 upon p53 mutant isoform action is limited.
With HSF1 emerging as an attractive therapeutic target in cancer , it is important to determine whether altered HSF1 activity can positively or negatively regulate clonogenicity of cancer cells, the direct or indirect downstream targets of HSF1 that mediate these actions and how these may relate to cellular context.
In the present study, using both knockdown and ectopic expression approaches, we demonstrate that HSF1 can positively regulate breast cancer cell line clonogenicity in vitro. Moreover, we demonstrate that this occurs via a mutant p53-dependent mechanism. Conversely, we show that HSF1 can also positively regulate WT p53, thereby inhibiting clonogenicity in both non-transformed and H-RasV12-transformed human mammary epithelial cells. The present study demonstrates that HSF1 can have divergent effects upon cell clonogenicity dependent upon the cellular status of p53 and has implications for the targeting of HSF1 in differing cellular contexts.
Generation and sources of plasmid constructs
HSF1WT cDNA was amplified from MCF10A cDNA by PCR using Flag_HSF1_Fwd (5′-AGCTTATGGACTACAAGGACGACGATGACAAGGATCTGCCCGTGGGCCCCGGC-3′) and EcoRI_HSF1_Rev (5-′AATGAATTCCTCGGAGACAGTGGGGTCCTT-3′) primers. HSF1ΔRDT cDNA was synthesized from HSF1WT cDNA using a PCR site-directed mutagenesis method as described previously . The HSF1WT and HSF1ΔRDT cDNAs were then cloned into the BamHI/EcoRI site of pBABEpuro-IRES-EGFP supplied by L. Miguel Martins (Plasmid #14430, Addgene). The pBABEpuro_IRES_mCherry was generated by the ligation of three fragments: pBABEpuro-IRES-EGFP vector digested with EcoRI and SalI, an IRES with EcoRI and BstX1 overhangs and an mCherry with BstXI and SalI overhangs. Mutant p53R273H gene was excised from the vector pSUPER-p53R273H provided by Ygal Haupt (Peter MacCallum Cancer Center, Victoria, Australia) by digestion with EcoRI and cloned into the EcoRI site of the pBABEpuro IRES mCherry vector. MSCV_mCherry and MSCV_H-RasV12_mCherry plasmids were provided by Patrick Humbert (Peter MacCallum Cancer Center, East Melbourne, Victoria, Australia) . All expression vector sequences were confirmed by DNA sequencing (Micromon DNA Sequencing Facility, Monash University). HSF1-targeted shRNAmir (microRNA-adapted short hairpin RNA) vectors were constructed as described previously . pGIPZ lentiviral vectors expressing shRNAmirs targeting p53 were purchased from Open Biosystems.
Cell lines and cell cultures
The MCF10A cell line was obtained from the A.T.C.C. (Manassas, VA, U.S.A.) and cultured as described previously . T47D cells were grown in RPMI 1640 medium, SkBr3 cells were grown in McCoy's 5A medium, and HEK (human embryonic kidney)-293T and Hs578T cells were grown in DMEM (Dulbecco's modified Eagle's medium). The medium was supplemented with 10% (v/v) FBS (fetal bovine serum) and 1% (w/v) penicillin/streptomycin. All stable cell lines were generated by retroviral or lentiviral transduction as described previously . Viral stocks were generated by transient transfection of appropriate viral packaging vectors into the HEK-293T cell line as described previously .
Two-dimensional standard growth assay
Cell proliferation was examined in 96-well plates using the SRB (sulforhodamine B; Sigma–Aldrich) colorimetric assay as described previously . Briefly, cells were seeded at 2×104–5×104 cells/well in 100 μl of culture medium in triplicate, grown and fixed each day for 5 days in 50% TCA (trichloroacetic acid) at 4°C for 1 h, followed by five washes in distilled water. Cells were stained with SRB and solubilized in 150 μl of 10 mM Tris/HCl, pH 10.5. Absorbance at 550 nm was measured by spectrophotometry using a Multiskan FC Absorbance Plate Reader (Thermo-LabSystems).
Two-dimensional clonogenic growth assay
A 2D (two-dimensional) clonogenic growth assay was performed as described previously by Kattan et al. . MCF10A cells were plated at 100 cells/well and grown for 8 days. T47D cells were plated at 500 cells/well and grown for 3 weeks and SkBr3 cells were plated at 2×103 cells/well and grown for 4 weeks.
Three-dimensional adhesion-independent clonogenic growth assay
Cells were suspended in 1.5 ml of growth medium containing Bacto agar (top agar) and added over a pre-hardened base agar layer (bottom agar) comprising Bactor agar and 2 ml of growth medium in six-well plates. MCF10A cells were grown in 0.4% top agar and 1% bottom agar, whereas T47D and SkBr3 cells were grown in 0.35% top agar and 0.5% bottom agar. A total of 1 ml of the appropriate cell growth medium for each cell line was added to the plates and was replenished every 4 days. Cultures were stained with 0.005% Crystal Violet and colonies were counted using ImageJ software (NIH).
Western blot analysis and antibodies
Generation of protein lysates from cells and subsequent Western blot analysis were performed as described previously . All blots were incubated with primary antibodies overnight at 4°C and with horseradish peroxidase-conjugated secondary antibodies for 1 h. Protein bands were visualized by chemiluminescence (GE Healthcare). All antibodies were purchased from commercial sources and included anti-HSF1 (catalogue number SPA-901), anti-HSP27 (catalogue number SPA-800) and anti-HSP90 (catalogue number SPA-835) antibodies from Enzo Life Sciences; anti-HSP105/110 (catalogue number Sc-6241) antibody from Santa Cruz Biotechnology; anti-Ha-Ras (catalogue number 05-775) antibody from Merck Millipore; anti-actin (catalogue number MS-1295-P0) and anti-HSP70 (catalogue number MS-482-P0) antibodies from Thermo Scientific; anti-p53 (catalogue number 51-9002046), anti-CDKN1A (cyclin-dependent kinase inhibitor 1A; p21; catalogue number 556430), anti-Bcl-2 (catalogue number 51-9001912), anti-BAX (catalogue number 51-9001914), anti-BAD (catalogue number 51-9001911), anti-Bcl-x (catalogue number 51-9001913) and anti-XIAP (X-linked inhibitor of apoptosis protein; catalogue number 51-9001990) antibodies from BD Pharmingen.
RT (reverse transcription)–qPCR (quantitative PCR) and primers
RT–qPCR was performed as described previously . Briefly, total RNA was isolated using the Qiagen RNeasy kit according to the manufacturer's instructions. A total of 1–2 μg of total RNA was used to synthesize cDNA using the superscript VILO cDNA synthesis kit (Invitrogen). The synthesized cDNAs underwent qPCR using the Perfecta SYBR Green SuperMix (Quanta Biosciences) and was performed in a Rotogene 3000 light cycler (Corbett Research). Raw data were exported to Microsoft Excel and then analysed by LinRegPCR software (HFRC) to determine PCR efficiency (E) and threshold cycle value (CT) . The level of expression of target genes was represented as relative to the expression of the housekeeping gene RPL32 (ribosomal protein L32). Differences in gene expression between samples were expressed as a ratio of the relative gene expression of the treated sample to that of the control sample. RT–qPCR primers were designed using the NCBI (National Center for Biotechnology Information) primer design website (http://www.ncbi.nlm.nih.gov/tools/primer-blast/) such that each amplicon was between 100 and 150 bp and spanned at least one intron/exon boundary. The primers used in the present study were: CDKN1A_Fwd: 5′-AGCAGAGGAAGACCATGTGGACCT-3′, CDKN1A_Rev: 5′-GGAGTGGTAGAAATCTGTCATGCTGG-3′, BAX_Fwd: 5′-CACAGTGGTGCCCTCTCCCCAT-3′, BAX_Rev: 5′-TCAAGGTCACAGTGAGGTCAGGGG-3′, TP53I3_Fwd: 5′-ACCCACCTCCAGGAGCCAGC-3′, TP53I3_Rev: 5′-TACTGAGCCTGGCCCCCACC-3′, Mdm2_Fwd: 5′-TGTTTGGCGTGCCAAGCTTCT-3′ Mdm2_Rev: 5′-GGTGACACCTGTTCTCACTCACAG-3′, TP53_Fwd: 5′-GCCAGACTGCCTTCCGGGTCACT-3′ TP53_Rev: 5′-CATCCATTGCTTGGGACGGCAAGGG-3′, RPL32_Fwd: 5′-CAGGGTTCGTAGAAGATTCAAGGG-3′ and RPL32_Rev: 5′-CTTGGAGGAAACATTGTCAGCGATC-3′.
All cell biology assays were performed at least three times and the results were combined. The results are means±S.D. Student's t tests were conducted to determine whether the treatment group was statistically significant compared with control. Significance is represented as *P<0.05, **P<0.01 and ***P<0.001.
Ectopic expression of HSF1 promotes the clonogenicity of breast cancer cell lines
As recent studies have demonstrated that HSF1 expression and activation are correlated with a more advanced cancer phenotype [2,7,8], we wanted to determine the effect of ectopic expression of HSF1 in less aggressive breast cancer cell lines that contained lower levels of active HSF1. To achieve this, we utilized retroviral constructs to express HSF1 (HSF1WT) or a constitutively activated HSF1 mutant (HSF1ΔRDT) in the T47D (Figure 1A) and SkBr3 (Figure 1F) cell lines to levels comparable with more aggressive breast cancer cell lines. HSF1ΔRDT was generated by deletion of the regulatory domain and substitution of Leu395 with glutamic acid (L395E) thereby facilitating active HSF1 trimer formation. Consistent with increased expression and activation of HSF1, expression of both HSF1WT and HSF1ΔRDT resulted in increased levels of HSP expression levels (Figures 1A and 1F). Within both T47D and SkBr3 cells the ectopically expressed HSF1WT was activated to levels similar to that of the constitutively activated mutant (Figures 1A and 1F). Although HSF1 has been shown to have a role in cell proliferation, we observed no effect of HSF1 either upon cell morphology (Figures 1B and 1G) or in standard 2D cell proliferation assays (Figures 1C and 1H). However, when the cells were examined for their ability to survive and grow to form colonies, ectopic expression of HSF1WT and HSF1ΔRDT significantly increased clonogenicity in both 2D (Figures 1D and 1I) and 3D (three-dimensional) (Figures 1E and 1J) assays in both cell types. This indicates that HSF1 may be more importantly required for mediating the establishment and growth of viable cell colonies under stringent and stressed growth conditions rather than in proliferation.
Ectopic expression of HSF1 promotes clonogenicity of breast cancer cells
Knockdown of HSF1 reduces clonogenicity of breast cancer cells
To investigate further the effect of HSF1 upon clonogenic growth, we examined the consequences of HSF1 knockdown in the triple-negative breast cancer cell line Hs578T, which expresses high levels of activated HSF1. Western blot analysis confirmed that knockdown of HSF1 in Hs578T cells reduced HSP27, HSP72 and HSP90 (Figure 2A). Consistent with the T47D and SkBr3 models, knockdown of HSF1 in Hs578T cells did not affect the morphology of the cells under 2D conditions (Figure 2B), but significantly reduced their clonogenicity (Figure 2C).
HSF1 knockdown reduces clonogenicity of the triple-negative breast cancer cell line Hs578T
HSF1 affects clonogenicity via mutant p53 activity
HSF1 is known to regulate the expression of genes beyond that of HSPs to promote cancer progression; however, the exact mechanism whereby HSF1 achieves this is relatively unknown. Previously, it has been shown that HSF1 knockdown can reduce mutant p53 levels due to the reduction of HSP90, a required molecular chaperone for mutant p53 stability. Furthermore, and as discussed above, mutant p53 has been shown to also have a role in cancer cell clonogenicity [14–16]. As the breast cancer cell lines being examined contained mutant p53, we wanted to test whether mutant p53 acts in conjunction with HSF1 to enhance clonogenicity by initially examining downstream targets of mutant p53. Consistent with previous studies, knockdown of HSF1 reduced protein levels of mutant p53V175F in the Hs578T cell line (Figure 3A). Moreover, knockdown of HSF1 increased levels of CDKN1A, a WT p53 target, thus suggesting that HSF1 knockdown relieved the suppressing effect of mutant p53 (Figure 3A).
HSF1 stimulates mutant p53 activities
We then examined the effect of ectopic expression of HSF1 upon protein level and downstream targets of mutant p53L194F in the T47D cells. This mutant isoform is known to suppress p53 targets such as CDKN1A, TP53I3 (tumour protein p53-inducible protein 3) and Gadd45, while enhancing Bcl-2 [33,34]; however, it also retains some WT p53 functions . Ectopic expression of HSF1 resulted in the reduction of CDKN1A and increased Bcl-2 protein levels, indicating that HSF1 supported mutant p53 activities (Figure 3B). Consistent with this, mRNA levels of CDKN1A were also decreased, whereas TP53I3 mRNA levels were increased (Figure 3C). Interestingly, HSF1 also enhanced the retained WT activity of mutant p53L194F such that BAX and BAD were increased upon HSF1 expression. Critically, HSF1 expression did not significantly increase the protein level of mutant p53L194F in T47D cells, or substantially increase either inducible or constitutive HSP90 isoforms determined by using an antibody detecting both HSP90α and HSP90β isoforms (Figure 3B), indicating that HSF1 supports mutant p53 activities beyond that of enhancing its stability. To mechanistically confirm a novel relationship between HSF1 and mutant p53 in relation to clonogenicity, mutant p53 was knocked down in the T47D model (Figure 3D). Once mutant p53 expression was reduced, HSF1 no longer possessed an enhancing effect upon clonogenecity (Figures 3E and 3F), thus demonstrating that HSF1 acted via a mutant p53-dependent pathway.
HSF1 divergently affects clonogenicity via WT and mutant p53
To confirm further the involvement of mutant p53 in mediating HSF1 effects upon enhancing clonogenicity, we stably expressed mutant p53R273H in GFP (green fluorescent protein) control and HSF1ΔRDT-expressing MCF10A cells. The MCF10A cell line expresses WT p53 endogenously (Figure 4A). Mutant p53R273H is one of the most common point mutants in breast cancer and can act in a dominant-negative manner in relation to WT p53 function. However, it also possesses ‘gain-of-function’ activities that are emerging as important contributors to the metastatic phenotype of cancers . The expression of p53R273H enabled MCF10A cells to grow in 3D clonogenic growth assays consistent with its known oncogenic activity  (Figure 4B). However, surprisingly, expression of HSF1ΔRDT suppressed 3D clonogenicity within the context of p53R273H expression and endogenous WT p53 (Figure 4B). To determine whether endogenous WT p53 was a cause of this effect, specific knockdown of WT p53 was performed through shRNAmir targeting of the 5′-UTR (5′-untranslated region) of p53, leaving ectopically expressed p53R273H mRNA intact. The clonogenicity defect mediated by HSF1ΔRDT expression was rescued and, consistent with previous findings, HSF1ΔRDT was then able to promote clonogenicity in this cellular context (Figure 4B). Furthermore, this result suggested that HSF1 could support the ‘gain-of-function’ activities of p53R273H in the absence of WT p53, yet inhibit clonogenicity via a WT p53-dependent mechanism (Figure 4B).
HSF1 stimulates both WT and mutant p53 activities
Ectopic expression of HSF1 reduces clonogenic survival and growth of cells via the actions of WT p53
To examine this latter point, that is the effect of HSF1 on the clonogenicity of cells with a WT p53 background, we ectopically expressed HSF1WT or HSF1ΔRDT in non-transformed MCF10A cells (Figure 5A). To determine whether the effects of HSF1 upon WT p53 actions were altered in a transformed cellular context, ectopic expression of HSF1WT or HSF1ΔRDT was performed in isogenic matched H-RasV12 transformed MCF10A cells (Figure 5A). Consistent with increased expression and activation of HSF1, expression of HSF1WT and HSF1ΔRDT resulted in increased levels of HSP expression (Figure 5A). The expression of H-RasV12 in the MCF10A cell line induced morphological changes consistent with an EMT (epithelial–mesenchymal transition) (Figure 5B), enhanced 2D growth under limiting medium conditions (Figure 5D) as well as increasing 2D clonogenicity (Figure 5E). H-RasV12 expression also enabled MCF10A cells to grow in the 3D clonogenic anchorage-independent soft agar growth assay (Figure 5C), consistent with their transformed phenotype. Consistent with the findings described above, HSF1WT or HSF1ΔRDT expression did not significantly affect the cell morphology (Figure 5B) or alter proliferation rates in 2D standard growth assays for both the non-transformed and H-RasV12-transformed MCF10A cells (Figures 5C and 5D). Ectopic expression of HSF1WT or HSF1ΔRDT was also not sufficient in supporting MCF10A growth in the 3D clonogenic growth assays (Figure 5F), providing evidence that HSF1 is not a ‘bona fide’ oncogene. Interestingly, consistent with the notion that HSF1 acts through WT p53 to inhibit clonogenicity, expression of HSF1 in both the non-transformed and H-RasV12-transformed MCF10A cells significantly reduced clonogenicity under both 2D (Figure 5E) and 3D conditions (Figure 5F).
HSF1 ectopic expression reduces clonogenicity of cells with WT p53
HSF1 affects clonogenicity through WT p53 activity
To determine whether HSF1 acts through WT p53 activity in the MCF10A cell line models, leading to reduced clonogenicity, we initially examined p53 target expression at the protein and mRNA levels in the MCF10A cell line models. Analysis of mRNA expression of p53 and its target genes by RT–qPCR demonstrated that HSF1 expression did not alter p53 mRNA levels (Figures 6A and 6B); however, despite this, its expression significantly increased the mRNA levels of a panel of p53 positively regulated transcriptional target genes, namely CDKN1A, Mdm2 (murine double minute 2), TP53I3 and BAX, in both non-transformed and H-RasV12-transformed MCF10A cells (Figures 6A and 6B). In agreement with previously published findings [19,20], cells expressing HSF1WT or HSF1ΔRDT had lower levels of p53 in comparison with GFP control cells, most notably in the non-transformed MCF10A cells (Figure 6C). However, despite this reduction, HSF1WT and HSF1ΔRDT expression still increased the levels of the p53 transcriptional target, CDKN1A (p21), in line with the RT–qPCR results, and reduced the levels of anti-apoptotic proteins, Bcl-2, XIAP and Bcl-xL (Figure 6A), which are suppressed by WT p53 activity . Expression of HSF1WT and HSF1ΔRDT in the H-RasV12-transformed cells produced similar effects upon p53 target protein levels (Figure 6C).
HSF1 mediates clonogenicity via WT p53 activities
To mechanistically determine whether p53 is a mediator of the HSF1 inhibitory effect upon clonogenicity, we knocked down p53 with two independent shRNAmirs in both the GFP control and HSF1ΔRDT H-RasV12-transformed cells (Figure 7A). Knockdown of p53 in the GFP control cells did not significantly increase the clonogenicity of MCF10A cells (Figures 7B and 7C); however, p53 knockdown negated the inhibitory effect of HSF1ΔRDT expression upon clonogenicity to levels similar to that of control MCF10A cells (Figures 7B and 7C). These results indicate that HSF1 acts via WT p53 to reduce clonogenicity in both non-transformed and H-RasV12-transformed human mammary epithelial cells, and also suggests that these cellular contexts require mutant p53 ‘gain-of-function’ activities to enhance clonogenicity.
Knockdown of WT p53 negates HSF1-mediated inhibition of clonogenicity
HSF1 acts as a master regulator of the heat-shock response; however, it also facilitates malignant transformation, cell survival and proliferation by mediating distinct transcriptional networks within cancer cells [1,3–5]. In addition, it is emerging that HSF1 also supports malignant progression [8,38]. Consistent with this, increased HSF1 expression, activation and its nuclear localization have been associated with more advanced disease, metastasis and poorer patient outcomes [2,5,7,8].
Within the present study we examined whether HSF1 affects an attribute of highly malignant cancer cells, that of clonogenic growth and survival. This feature is associated with cancer ‘stem-like’ properties allowing for increased tumour- and metastasis-initiating capacities [10–13]. In line with the hypothesis that HSF1 supports a more advanced cancer phenotype, we identified that in a number of breast cancer cell lines and the human mammary epithelial cell line MCF10A, within the cellular context of mutant p53, HSF1 positively regulated clonogenicity. However, interestingly, within the cellular context of WT p53, HSF1 actually inhibited clonogenicity.
Although HSF1 is known to have multifaceted roles in cancer and that HSF1, either directly or indirectly, regulates distinct transcriptional networks, many of the mediators required for HSF1's cell biological actions are not known. In seeking to identify mediators of HSF1 action upon clonogenicity, we identified that HSF1 affected the action of WT p53 and mutant p53 isoforms to negatively or positively regulate clonogenicity respectively.
These findings indicate that HSF1 may enhance tumour progression and metastasis by promoting mutant p53 isoform actions, especially with respect to their ‘gain-of-function’ attributes that are emerging as important contributors to the metastatic phenotype [17,36]. However, paradoxically, HSF1 may also promote the actions of WT p53 to inhibit tumour progression. In line with our findings, Logan et al.  reported previously that co-expression of HSF1 with WT p53 in cancer cell lines caused a significant increase in p53 activity upon genotoxic stress, affecting the efficacy of growth inhibition by genotoxic agents such as doxorubicin. Furthermore, heat shock and HSF1 activation have been shown to enhance the expression of DNA damage response and pro-apoptotic proteins upon doxorubicin treatment, as well as support p53-mediated apoptosis [39,40]. Therefore we have not only demonstrated the novel finding that HSF1 mediates its effect via a mutant p53-dependent pathway to promote clonogenicity, but also extended the understanding that, in addition to genotoxic stress, HSF1 can act to support WT p53 actions in also abrogating clonogenicity. The latter finding suggests that activation of HSF1 within a WT p53 cellular context may be beneficial in combination cancer treatment regimes. Moreover, it could be hypothesized that, owing to its action on WT p53, tumours with highly activated HSF1 may be associated with p53 mutation status.
The accumulating evidence of an important role of HSF1 in cancer growth and progression has seen it emerge as an attractive therapeutic target; however, intriguingly, both activators of HSF1, such as withaferin A and celastrol, as well as HSF1 inhibitors, such as KNK437 and Triptolide, exhibit anti-cancer effects [23,41]. Our results indicate that the p53 status of the tumour may directly affect the therapeutic efficacy of such HSF1 activators or inhibitors in cancer treatment. More importantly, this should be a consideration for the future testing and development of such agents. Moreover, our results point towards the potential inhibition of HSF1 as providing a way of indirectly therapeutically targeting the diverse range of mutant p53 proteins that exist by a single targeted approach.
Although we have shown a clear functional association of HSF1 with both WT and mutant p53 pathways, the precise mechanism by which HSF1 achieves this still requires elucidation. However, previous studies have indicated that HSF1 can affect WT p53 activity by enhancing its translocation to the nucleus , which may be achieved indirectly by FKBP52 (FK506-binding protein 52), a transcriptional target of HSF1, which links p53 to dynein and the microtubule network, leading to p53 nuclear transport . A direct interaction between HSF1 and WT p53 has also been shown during genotoxic stress. This complex is then co-operatively recruited to p53-responsive genes where HSF1 enhances p53-mediated transcription .
Contrasting with the increased activity of WT p53, we observed that expression of HSF1 reduced the steady-state levels of WT p53 in the non-transformed MCF10A cells. WT p53 is a very labile protein and its level within the cell is regulated by the rate of its proteasomal degradation. In support of our findings, knockdown of HSF1 has previously been shown to increase p53 protein levels [19,20] owing to a reduction in αB-crystallin, an HSF1 transcriptional target. αB-crystallin interacts with Fbx4 (F-box only protein 4) ubiquitin ligase, targeting p53 for degradation and thus a reduction in its steady-state levels . Moreover, HSF1 and HSF2 complexes have been shown to transcriptionally regulate proteasomal subunits, such as Psmb5 [proteasome (prosome, macropain) subunit β type 5] and gankyrin, which are also involved in p53 degradation . Thus, although HSF1 decreases WT p53 levels, it also increases its transcriptional activity, suggesting a complex interplay between the transcription factors.
With respect to the actions of HSF1 upon mutant p53, it has been shown that mutant p53 forms a complex with the HSF1 transcriptional target, HSP90, and this interaction stabilizes mutant p53, protecting it from Mdm2 and CHIP (C-terminus of HSP70-interacting protein) E3 ligase-mediated proteasomal degradation . Consistent with this, we found that HSF1 depletion in the Hs578T cells led to a significant reduction in mutant p53 levels. However, a concordance between the decrease in HSP90 levels and a concomitant decrease in mutant p53 levels with an increase in CDKN1A levels was not evident in all cell lines examined, suggesting that the role for HSF1 in mediating mutant p53 activity may extend beyond that of mutant p53 stabilization. For example, ectopic expression of WT HSF1 in the T47D cell line, which undergoes activation as demonstrated by increased HSP110 and HSP27 levels, does not substantially increase either HSP90 or mutant p53 levels. Whether alternative mechanisms by which HSF1 mediates mutant p53 activity, such as a direct interaction between mutant p53 isoforms and HSF1 as is the case with WT p53  is currently unknown.
In conclusion, the present study provides novel compelling evidence of an important interplay between HSF1 and mutant and WT p53 in mediating disparate clonogenicity, and highlights the importance of a cellular context for HSF1-mediated actions.
cyclin-dependent kinase inhibitor 1A
green fluorescent protein
human embryonic kidney
heat-shock factor 1
murine double minute 2
microRNA-adapted short hairpin RNA
tumour protein p53-inducible protein 3
X-linked inhibitor of apoptosis protein
Chau Nguyen and Benjamin Lang designed and performed the experiments and analysed the data. Ryan Chai, Michelle Kouspou and Jessica Vieusseux discussed the data, commented and edited the paper prior to submission and provided technical support. John Price performed experimental design, analysed the data and was the grant holder under which the work was performed. Chau Nguyen and John Price wrote the paper.
We thank Monash Micro Imaging for assistance with microscope imaging, Monash Flowcore for assistance with FACS and Monash Micromon for DNA sequencing.
This work was supported by Monash University IPRS (International Postgraduate Research Scholarship) and MRS (Monash Residential Services) Awards, Monash University (to C.H.N.) Cancer Council Victoria [grant-in-aid number 545969 (to J.T.P.)], National Health and Medical Research Council (NHMRC) Project Grant [number 606549 (to J.T.P.)], and an NHMRC RD Wright Fellowship [number 395525 (to J.T.P.)].