HSF1 has been involved in the etiology of cancer by its multiple effects in facilitating transformation and tumor invasiveness in response to diverse oncogenic stimuli [10]. In this respect, Hsf1^−/− mice show lower incidence of tumors induced by mutations of the RAS oncogene or a hot spot mutation in the tumor suppressor p53, and they show improved survival [10]. HSF1 is also crucial for cell transformation and tumorigenesis induced by the human epidermal growth factor receptor-2 (HER2), an oncogene responsible for breast tumors aggressiveness. Knockdown of HSF1 leads to growth arrest and senescence of HER-2-expressing cells [63]. Moreover, HSF1 is required for the maintenance of the transformed phenotype in: (i) established oncogenic cell lines; (ii) breast cells lines with progressive oncogenic states (i.e., primary human mammary epithelial (PHME) cells; immortalized human mammary epithelial (HME) cells; fully transformed and tumorigenic HME cells (HMLER); [64]); (iii) a collection of breast cell lines derived from spontaneous human tumors, with various p53 status (MCF-7 or various mutant alleles BT-474, MDAMB-231, and T47D); (iv) malignant cells of diverse histological origins either derived from human tumors (HeLa (cervix), PC-3 (prostate), and S462 and 90-8 (peripheral nerve sheath) or derived by in vitro transformation (293T; kidney). Furthermore, in immortalized MEFs, HSF1 is essential for basal and EGF-induced migration, a process crucial for tumor invasion and metastasis [65].

HSF1 is often elevated in human cancer cells compare to normal cells. However, somatic mutations in Hsf1 have not yet been identified in human cancers, and overexpression of HSF1 does not lead to transformation, as mutant RAS does. Here, HSF1 does not display typical oncogenic features per se but represents a mechanistic and therapeutic challenging target. It is possible that the core function pathways regulated by HSF1 in normal cells might be massively mobilized in cancer cells, like signal transduction, ribosome biogenesis, translation and glucose metabolism [10]. Indeed, cancer cells growth and development highly depend on normal cellular functions governed by genes, which are not typical cancer related genes, and through which HSF1 can promote tumorigenesis.

The unique HSF present in fission yeast cells is essential for normal growth and drives the transcription of target genes that encode proteins with a broad range of biological functions, including protein folding and degradation, energy generation, protein trafficking, maintenance of cell integrity, small molecules transport, cell signaling and transcription [66,67]. Conversely, in murine cells, HSF1 is more dispensable for cell growth and survival, although Hsf1^−/− mice show defects in postnatal growth and placenta formation, suggesting that in normal cells HSF1, while a major actor in the heat shock response, is also important for other cellular process [68].

In normal mammalian cells, HSF1 is known to regulate Hsp genes, but also a wide range of non-Hsp genes upon stress or differentiation [2,3]. In aggressive cancer cells, HSF1 is expressed at high levels, which could amplify its activity and broaden the spectrum of its targets. By analogy to what happens in yeast, HSF1 could regulate central functions of the cell biology, in addition to the already known HSP expression. However, although the levels of HSPs (especially HSP27, HSP70 and HSP90) are elevated in different types of cancers, there is not always a correlation between HSF1 constitutive activation and HSP expression. In this way, genetic knockdown of HSF1 fails to induce a decrease in the levels of HSPs in some cancer cell lines [69]. It is believed that the elevated expression of HSF1 in cancer cells does not lead to a kind of global “heat shock response”, but only specific members of the HSP family are induced in a tumor-dependent manner. That explains why HSP expression in cancer cells displays considerable variation, depending on the HSP member and the tumor characteristics. For instance, in HER-2 (human epidermal growth factor receptor-2 oncogene) expressing cells, in aggressive breast cancer models, HSF1 knockdown is accompanied by a specific down-regulation of HSP27 and HSP70 expression [63].

HSF1 expression has been shown to be elevated in prostate carcinoma compared to its normal counterpart [70]. An elevated protein expression of HSF1 and some HSPs was reported in the aggressively malignant cell lines DU145 and CA-HPV-10. However, no difference in the RNA level was detected suggesting that HSP induction was not mediated by HSF1-dependent transcription. In addition, growth of PC3 cells in vivo, as tumor xenografts, was accompanied by a marked decrease in HSPs expression (HSP27, HSP70, HSP60, HSP90) whereas HSF level was not modulated. In a PC3 metastatic variant (PC3M), a strong increase in HSF1 mRNA was observed, as well as an increase in HSF1 protein and nuclear localization [71]. Overexpression of HSF1 in non-metastatic PC-3 cells was accompanied by an upregulation of HSP27 at the protein level, but HSP70 and HSP90 were not affected, once again pointing out that the elevated expression of HSPs in cancer cells does not always depend on HSF1. Interestingly, in prostate cancer tissue, HSP27 upregulation, but not HSP70 or HSP90, was significantly associated with clinicopathological factors [72]. In prostate cancer cells, HSF1 also influences the development of an aneuploid state and mitotic progression [73].

HSF1 may partly participate in breast cancer progression by inducing specific HSPs, mainly HSP27 [74]. In breast cancer, the highly malignant factor heregulin beta-1 (HRGβ1) is a secreted factor, which binds to c-ErbB-3 and -4 receptors and provokes the recruitment of c-erbB-2 and receptors heterodimerization. The binding of HRGβ1 to the cell surface induces an increase in HSF1 levels that results in anchorage-independent growth and protection from cisplatin-induced apoptosis mediated by HRGβ1 [74]. One aspect of this process relies on the inhibition of GSK3, a kinase that antagonized HSF1 activation. Part of the anti-apoptotic effects of HSF1 seems to operate via activation of the Hsp70 promoter and therefore likely involves HSPs. Remarkably a crosstalk between the beta-catenin/Wnt pathway and the heat shock cascade has been identified in breast cancer tumors with high metastatic potential. Indeed, the formation of a complex between beta-catenin, HSP27 and HSF1 has been detected in breast cancer biopsies. The striking presence of beta-catenin in the cytoplasm (and not only at the membrane) co-expressed with HSP27 and with HSF1 being also nuclear could have some clinical relevance in terms of prognosis [75]. Notably, HSF1 plays an additional role in the etiology of breast cancer. Unlike normal cells, cancer cells preferentially catabolize glucose by glycolysis, thereby producing high levels of lactic acid. For instance, Zhao et al. have shown that ErbB2 promotes glycolysis in a HSF1-dependent manner. Overexpression of ErbB2 increases the expression of HSF1, which binds to lactate dehydrogenase A (LDH-A) promoter and activates its transcription [76], allowing the metabolization of pyruvate to lactate.

The search for activation of signal transduction pathways in sporadic colorectal cancers (CRC) revealed an increase in the mRNA levels for heat shock and NFκB pathway genes [77]. Namely, an increase of Hsf1 mRNA in 86% of patients was reported in sporadic colorectal cancer. In that context, HSF1 upregulation in patients was accompanied by an elevated expression of HSP27, HSP90 but also of iNOS (NO synthase; 63%). In addition, HSF1, through the induction of the HSP70 co-chaperone BAG-3, which stabilizes the level of anti-apoptotic Bcl-2 family members, facilitates colon cancer cell survival during pro-apoptotic stress [78]. The oncogenesis of hereditary CRC cancers is believed to involve four signal transduction pathways: (a) the APC-β-catenin-TCF-myc (Wnt) pathway; (b) the microsatellite unstable pathway; (c) the p53 pathway; and (d) the estrogen receptor hypermethylation pathway [79,80]. Although no data are currently available on the potential role of HSF1 in hereditary CRC, at least two of these pathways (Wnt and p53) are known to be affected by HSF1 [77,79,80].

In gastro-intestinal cancer, the level of HSPs is often upregulated with a high variability and without evident correlation with HSF1 level [81]. Instead, the increase observed in HSF1 expression in gastro-intestinal cancer tissues, drives the repression of the pro-apoptotic protein, X-linked inhibitor of apoptosis protein (XIAP)-associated factor-1 (XAF-1) [82]. Thus, HSF1 can impair the apoptotic pathway in cancer cells, at least partly, in a HSP independent manner.

Interestingly, the protein TC1 (a positive regulator of the signaling Wnt/beta-catenin via inhibition of Chibby), which is upregulated in an aggressive subtype of gastric cancers, correlates with poor prognosis and has been reported to induce a heat shock response in cancer cells by favoring HSF1 expression. Moreover, HSF1 and TC1 mutually activate each other [83,84].

The first direct in vivo evidence of HSF1 implication in the development of spontaneous tumors arises from p53^−/− mice. The rapid tumor evolution observed in mice lacking p53 mainly results in lymphoma, while p53^−/−Hsf1^−/− mice rarely develop lymphomas, but rather succumb to other types of cancers, like testicular carcinoma [89]. Tumor suppressor p53 is frequently inactivated by genetic mutations in different cancers [90]. In this context, Dai et al. have shown that HSF1 deficiency dramatically reduces spontaneous tumor formation in mice carrying a common, dominant-negative mutation of the p53 gene, whereas Hsf1^+/+ and Hsf1^+/− mice bearing dominant-negative mutation of p53 develop a broad spectrum of tumor types (sarcomas, lymphoma, carcinomas; [10]). This was confirmed by Hsf1 knock down, using specific siRNAs, in different mouse and human cell systems [10]. These different results observed in these complementary works [10,89] may be explained by differences in the models used, involving distinct mouse genetic background, and the use of clinically relevant p53 mutation versus p53 null strategy.

Dysfunction in the proteasome pathway can lead to many disorders including cancers [91], and proteasome inhibitors are currently used in therapeutical approaches [92]. Recently, Lecomte et al. have reported that some proteasome subunits, such as Psmb5 and Gankyrin, were significantly downregulated in Hsf2^−/− MEFs [6]. Interestingly, Gankyrin is an oncoprotein involved in p53 degradation, thus suggesting a potential role of HSF2 in the regulation of the tumor suppressor p53.

Compared with wild-type MEFs, Hsf1^−/− MEFs are refractory to proliferation and resistant to focus formation, driven by oncogenic H-RASV12D or by the proto-oncogene PDGF-B, which are both mitogenic signal transducers [10]. In contrast, expression of c-MYC and LTA (regulator of cell-cycle progression, which are not expected to increase proliferation in already-immortalized cells, but predisposing cells to apoptosis) does not induce proliferation in immortalized Hsf1^+/+ MEFs. Accordingly, Hsf1^−/− MEFs show no enhanced survival in response to RAS and PDGF/B expression, but reduced survival in response to c-MYC and LTA expression [10].

HSF1 is also required for cell transformation and tumorigenesis induced by the oncogene HER2 (Human Epidermal growth factor Receptor-2), which is responsible for aggressive breast tumors. Indeed, knockdown of Hsf1 prevents neoplastic transformation (foci formation or tumor growth in xenografts) induced by HER2 expression in untransformed human mammary epithelial MCF-10A cells. This suggests that the proliferation of HER2-expressing cells is critically dependent on HSF1 [63]. Strikingly, this anti-tumorigenic effect of HSF1 downregulation was associated with HER2-induced upregulation of the cyclin-dependent kinase inhibitor p21, a major regulator of senescence in cancer cells [93], and accompanied by a decrease in the mitotic regulator survivin, which is also involved in growth arrest and senescence. Survivin is a member of the inhibitors of apoptosis (IAPs) family and a critical regulator of mitosis by modulation of aurora B kinase [94]. Its decrease is partially controlled by both HSF1 and p21 [63]. In addition, HSF1-mediated control of breast cancer cell senescence was due, at least partly, to the regulation of HSP70 and HSP27 expression.

The potential role of HSF1 in cancer metastasis was first reported in prostate cancer where HSF1 overexpression was correlated with aggressiveness of the tumors [70,71]. Since then, studies have shown that HSF1 is able to mediate tumorigenic effects through the recruitment of the prometastatic co-repressor gene MTA1 (a component of the NuRD complex containing the histone deacetylases HDAC1 and HDAC2) [95]. MTA1 is expressed in numerous human cancers, including breast, prostate and gastro-intestinal cancers and its expression correlates with tumor aggressiveness and metastasis [96–98]. Herein, MTA1 protein expression is higher in hormone-refractory metastatic prostate cancer compared to clinically localized disease and benign prostatic tissues [97].

Breast cancer cell can also escape from hormonal control, leading to higher metastatic potential. Khaleque et al., have reported that HSF1 binds to the corepressor MTA1 in human cultured breast cancer cells and carcinoma samples. The HSF1-MTA1 complex, strongly induced by the transforming ligand heregulin (a ligand for erbB receptor from the EGF family), assembles on the chromatin of breast cancer cells and mainly binds to the promoter of estrogen responsive genes [95]. Thus, HSF1 induces repression of estrogen-dependent gene transcription, an effect linked to cancer invasiveness.

Migration properties in correlation with metastasis potential were also shown to be controlled by HSF1. Mouse embryonic cells derived from Hsf1^−/− mice are deficient in both basal and EGF-mediated migration. This default in migration, which is partly due to the downregulation of EGF receptor and likely involves Ras, is associated with an impairment of the MAPK signaling pathway [65].

As described above, HSFs have been described to modulate the structure of chromatin and to establish some epigenetic marks (see introduction). In particular HSF1, in association with HSF2, was shown to steer the transcription of the human DNA satellite III in the centromeric regions, establishing an epigenetic status resembling euchromatin [14,99,100]. The involvement of HSF2 in the expression of satellite III DNA was further confirmed [27]. Remarkably, transcription of satellite III DNA is associated with aneuploidy in several cancers such as ovarian cancers, melanoma or myeloma [101–103] and thus could be related to HSF status.

Accordingly, Lee et al. have recently shown that HSF1, via its regulatory domain, when overexpressed in radiation induced-fibrosarcoma cells, interacts with cdc20, the co-activator of anaphase-promoting complex (APC), and thus, inhibits the mitotic exit and the ubiquitination activity of APC on two key anaphase inhibitors, Cyclin B1 and securin. As a consequence, this non-transcriptional HSF1 effect leads to aneuploidy and multinucleated cells associated with micronuclei and genomic alteration [104]. In prostate carcinoma cells, a dominant negative construct of HSF1 dramatically alters the DNA content of PC3 cells and inhibits aneuploidy and Cyclin B1 stabilization [73]. Further, HSF1-mediated aneuploidy was facilitated in p53-defective cells. This phenomenon was associated to an increase in HSF1 activity through phosphorylation by a Polo-like kinase, as well as cdc20/HSF1 interaction [105].

Altogether, HSF1 activity is hijacked in a pleiotropic manner by a large diversity of cancer cells and oncogenes to favor tumor initiation and progression. In sharp contrast, it is noteworthy that HSF1 knockout has a minimal effect on the proliferation/survival of normal primary human cells.