Heat shock proteins in cancer: diagnostic, prognostic, predictive, and treatment implications

Abstract

Heat shock proteins (Hsps) are overexpressed in a wide range of human cancers and are implicated in tumor cell proliferation, differentiation, invasion, metastasis, death, and recognition by the immune system. We review the current status of the role of Hsp expression in cancer with special emphasis on the clinical setting. Although Hsp levels are not informative at the diagnostic level, they are useful biomarkers for carcinogenesis in some tissues and signal the degree of differentiation and the aggressiveness of some cancers. In addition, the circulating levels of Hsp and anti-Hsp antibodies in cancer patients may be useful in tumor diagnosis. Furthermore, several Hsp are implicated with the prognosis of specific cancers, most notably Hsp27, whose expression is associated with poor prognosis in gastric, liver, and prostate carcinoma, and osteosarcomas, and Hsp70, which is correlated with poor prognosis in breast, endometrial, uterine cervical, and bladder carcinomas. Increased Hsp expression may also predict the response to some anticancer treatments. For example, Hsp27 and Hsp70 are implicated in resistance to chemotherapy in breast cancer, Hsp27 predicts a poor response to chemotherapy in leukemia patients, whereas Hsp70 expression predicts a better response to chemotherapy in osteosarcomas. Implication of Hsp in tumor progression and response to therapy has led to its successful targeting in therapy by 2 main strategies, including: (1) pharmacological modification of Hsp expression or molecular chaperone activity and (2) use of Hsps in anticancer vaccines, exploiting their ability to act as immunological adjuvants. In conclusion, the present times are of importance for the field of Hsps in cancer, with great contributions to both basic and clinical cancer research.

INTRODUCTION

Levels of the heat Hsp molecular chaperones are elevated in many cancers, and Hsp overexpression signals a poor prognosis in terms of survival and response to therapy in specific cancer types (Ciocca et al 1993; Cornford et al 2000; Blagosklonny 2001; van de Vijver et al 2002; van 't Veer et al 2002). Elevated Hsp expression in malignant cells plays a key role in protection from spontaneous apoptosis associated with malignancy as well as the apoptosis generated by therapy, mechanisms which may underlie the role of Hsp in tumor progression and resistance to treatment (Volloch and Sherman 1999; Nylandsted et al 2000a, 2000b; Ciocca et al 2003; Gyrd-Hansen et al 2004). Hsp transcription requires activated heat shock transcription factor 1 (hsf1), which is itself overexpressed in cancer and plays a role in invasion and metastasis (Wu 1995; McMillan et al 1998; Hoang et al 2000; Wang et al 2004). However, the molecular mechanisms linking increased Hsp and HSF1 expression to tumor progression are not currently understood. Because Hsps are induced only under stress conditions in normal cells, some aspects of the malignant phenotype apparently cause Hsp dysregulation (Lindquist and Craig 1988).

Heat shock proteins

Hsps were first discovered as a cohort of proteins that are powerfully induced by heat shock and other chemical and physical stresses in a wide range of species (Lindquist and Craig 1988; Georgopolis and Welch 1993). The Hsps have been subsequently characterized as molecular chaperones, proteins, which have in common the property of modifying the structures and interactions of other proteins (Beckmann et al 1990; Gething and Sambrook 1992; Netzer and Hartl 1998; Freeman and Yamamoto 2002). Molecular chaperone function dictates that the Hsps often interact in a stoichiometric manner with their substrates, necessitating high intracellular concentrations of the proteins (Lindquist and Craig 1988; Georgopolis and Welch 1993). As proteins that shift the balance from denatured, aggregated protein conformation toward ordered, functional conformation, Hsps are particularly in demand when proteins are disordered by heat shock, oxidative stress, or other protein-damaging events (Lindquist and Craig 1988; Hightower 1991; Gething and Sambrook 1992; Georgopolis and Welch 1993). The Hsp28, 40, 70, and 110 genes have therefore evolved a highly efficient mechanism for mass synthesis during stress, with powerful transcriptional activation, efficient messenger RNA (mRNA) stabilization, and selective mRNA translation (Voellmy 1994). Hsp27, 70, 90, and 110 increase to become the dominantly expressed proteins after stress (Hickey and Weber 1982; Landry et al 1982; Li and Werb 1982; Subjeck et al 1982; Henics et al 1999; Zhao et al 2002). Hsp gene transcription is regulated by transcription factors belonging to the heat shock factor (HSF) family that ensure prompt transcriptional activation in stress and equally precipitous switch-off after recovery (Sorger and Pelham 1988; Wu 1995). The hsf gene family includes heat shock transcription factor 1 (hsf1), the molecular coordinator of the heat shock response, as well as 2 less well-characterized genes, heat shock transcription factor 2 (hsf2) and heat shock transcription factor 4 (hsf4) (Rabindran et al 1991; Schuetz et al 1991; Nakai et al 1997). In addition to the Hsps induced by heat, cells also contain a large number of constitutively expressed Hsps (Tang et al 2005). Recent studies have shown that the constitutive Hsps are found in a variety of multiprotein complexes containing both Hsps and cofactors (Buchner 1999). These include Hsp10 and Hsp60 complexes that mediate protein folding and Hsp70- and Hsp90-containing complexes that are involved in both generic protein-folding pathways and in specific association with key regulatory proteins within the cell (Netzer and Hartl 1998; Pratt and Toft 2003). Hsp90 plays a particularly versatile role in cell regulation, forming complexes with a large number of cellular kinases, transcription factors, and other molecules.

Mechanisms for Hsp elevation

Hsp expression is tailored for induction by the stress response, and the proximal signal for Hsp induction is apparently the accumulation of denatured proteins (Voellmy 2004). It is thus rather a mystery as to how the Hsps become overexpressed in cancer. One hypothesis is that the physiopathological features of the tumor microenvironment (low glucose, pH, and oxygen) tend toward Hsp induction. Whether this is true or not, we do not know. However, one recent study indicates that when cells are transferred from tissue culture to growth as xenografts in vivo, Hsp expression declines markedly (Tang et al 2005). Because the elevated Hsp levels associated with malignancy tend to persist when cells are grown in tissue culture, they may well be related to the genetic changes associated with tumor progression (Tang et al 2005). Oncoproteins may appear during carcinogenesis (eg, mutated p53), and these mutated and conformationally altered proteins may elicit an Hsp response. However, the exact mechanisms are yet to be determined although they likely involve molecular changes common to a wide range of cancer cells with the potential to feed into the signaling mechanisms that lead to HSF1 activation.

In this short review, we will provide an overview of the current status of the Hsps in cancer with special emphasis on the clinical setting. This is not an easy task because there are more than 200 types of cancers and there are several Hsp families, each with multiple members (Tang et al 2005). Finally, the process of carcinogenesis involves a complex array of genetic and epigenetic alterations, which contribute to cancer pathogenesis (Hahn and Weinberg 2002), leading ultimately to a unique cancer tissue (often with mixed cancer clones) within a unique molecular milieu and in this may change dramatically the Hsp context and behavior. Therefore, we need to define whether altered expression of the Hsps at genomic or proteomic level is of importance to cancer prevention, diagnosis, prognosis, prediction, and treatment.

IMPLICATIONS IN DIAGNOSIS

In these cases, Hsp expression has been analyzed in relation to the histopathological characteristics of the tumor tissues (eg, tumor type, grade of differentiation), with the expression of other molecules (eg, estrogen receptors, c-myc, mutated p53), and with patient parameters like sex and age. In addition, levels of circulating Hsps and anti-HSP antibodies have been correlated with patient and tumor characteristics. Table 1 summarizes the publications regarding the diagnostic implications of the Hsps in cancer.

Table 1.

 Heat shock proteins in cancer: diagnostic implications

Because of space limitations, we cannot perform a detailed analysis of each study (eg, adequacy of the number and homogeneity of samples incorporated). However, the following main conclusions can be made. (a) Diagnostic implications—Hsps are overexpressed in a wide range of malignant cells and tissues. Therefore, Hsp detection is not useful in diagnostic immunopathology (there are other more restricted molecular markers to identify the lineage of origin of cancer tissues: carcinoma, sarcoma, lymphoma, etc). However, it might be useful to apply in a panel of immunopathology antibodies, the detection of αBcrystallin for identification of renal cell carcinomas (Pinder et al 1994), and the detection of Hsp27 or Hsp90 for identification of Reed-Sternberg cells (Hsu and Hsu 1998). The presence of Hsps and antibodies to the Hsps in the serum of cancer patients is still a new research area. Although it seems that autoantibodies to certain Hsps are of significance as tumor markers in osteosarcomas, ovarian cancer and others, at present, we need more studies to draw a clear conclusion on this important subject. The same applies to the study of the polymorphism of the Hsp70-2 gene. (b) Carcinogenesis—Hsp expression levels can help indicate the presence of abnormal changes during the process of carcinogenesis (in certain tissues). For example, Hsp27 is overexpressed in hyperplastic endometrium, and this protein appears as a marker of squamous metaplasia in the uterine cervix; Hsp10 and Hsp60 are related with the process of carcinogenesis of the uterine cervix and colon; and Hsp70 is associated with carcinogenesis of the oral epithelium and as a marker of early hepatocellular carcinoma. In oesophageal carcinomas, Hsp27 decreases during the carcinogenesis that ends in adenocarcinomas but increases during the carcinogenesis that ends in squamous carcinomas. Then, Hsps can be used as subrogate biomarkers of certain cancers. (c) Differentiation—Hsp expression correlates with the degree of differentiation in certain tissues. Hsps associated with higher differentiation are: Hsp27 and Hsp90 in endometrial carcinomas, Hsp27 in squamous carcinomas (uterine cervix, oral epithelium), and Hsp27 as a marker of keratinocyte differentiation in the skin. In contrast, Hsps associated with poor differentiation are Hsp70 in cancers of the breast, ovary, and oral epithelium, Grp78 in lung carcinomas, and Hsp27 in astrocytomas. At present, we do not have a clear explanation for these disparities and associations. Hsp70 has been involved not only with poor tumor differentiation but also with increased cell proliferation (breast, uterine cervix, lung), lymph node metastasis (breast, colon), increased tumor size (uterine cervix), presence of mutated p53 (breast, endometrium), and higher clinical stage (oral, colon, melanoma). (d) Associations with other molecules—In general, several Hsps are coexpressed in cancer tissues; in addition, certain Hsps can be significantly associated with other molecules. For example, Hsp27 has been associated with ERα in female breast carcinomas and endometrial carcinomas, but this protein did not appear associated with ERα in male breast carcinomas, cervical uterine carcinomas, hepatocellular carcinomas, and meningiomas (tissues that may express ERα). It is interesting that Hsp27, which was first described as an estrogen-regulated protein, is significantly associated with ERα in the female breast and endometrium (Table 1). These 2 organs are under strong estrogen and progesterone regulation. On the other hand, Hsp70 has been described as an important molecule in the assembly and trafficking of steroid receptors, and in breast cancer, Hsp70 has been found associated with ERα (Takahashi et al 1994). It is of interest to mention that Hsp70 can increase ERα transcriptional activity and growth in MCF-7 breast cancer cells (Spears and Barnes 2003), which in turn may explain the increased cell proliferation found in breast tumor biopsy samples that express Hsp70 (Vargas-Roig et al 1997). In addition, Hsp70 has been associated and complexed with mutant p53 in cancer cell lines (Lehmann et al 1991). This association has been studied in several cancer tissues, and the results have shown this association in certain cases only.

IMPLICATIONS IN THE PROGNOSIS

We have seen that expression of certain Hsps can be correlated with the carcinogenic process as well as with the degree of differentiation and cell proliferation, and moreover, they have been implicated in the regulation of apoptosis. Therefore, it was reasonable to study the prognostic implications of Hsps, and they emerged as useful in certain cancer types. The prognosis of a particular cancer patient is very important in the clinic to individualize cancer treatments, to plan the patient's follow-up, and to answer questions from the patient or relatives. Overtherapy with cytotoxic drugs can be avoided in cancer patients if they are correctly identified as having good prognosis and vice versa (Table 2). Again, because of space limitations, we cannot perform a detailed analysis of each study; however, the following conclusions can be made about the Hsps that have been studied most. (a) Hsp27— breast cancer is one of the sites where numerous studies on prognostic factors have been reported. To date, at the proteomic level, we can establish a categorization of several prognostic factors that, integrated with the traditional clinicopathological factors, provide a very good idea of the disease outcome (Gago et al 1998). Hsp27 is not among the list of useful prognostic factors in breast cancer (Oesterreich et al 1996). Hsp27 expression has been associated with poor prognosis in ovarian, gastric, liver and prostate cancer, and osteosarcomas. In contrast, Hsp27 expression has been associated with good prognosis in endometrial adenocarcinomas, oesophageal cancer, and in malignant fibrous histiocytomas. Although there are fewer studies in other cancers, the data suggest that Hsp27 has no prognostic value in head and neck squamous cancer, bladder and renal cancer, and leukemia (except when associated with other markers). There are contradictory data in oral cancer and ovarian cancer. (b) Hsp70—Hsp70 expression is correlated with poor prognosis in breast cancer, endometrial cancer, uterine cervical cancer, and transitional cell carcinoma of the bladder. This is consistent with the Hsp70 associations with poor differentiation, lymph node metastasis, increased cell proliferation, block of apoptosis, and higher clinical stage, which are markers of poor clinical outcome. In contrast, high Hsp70 expression was correlated with good prognosis in oesophageal cancer, pancreatic cancer, renal cancer, and melanoma. Hsp70 expression showed no correlation with prognosis in ovarian cancer, oral cancer, head and neck squamous cancer, gastric and prostate cancer, and leukemia. (c) Hsp90—Hsp90 expression in cancer tissues and presence of autoantibodies to Hsp90 have been correlated with poor prognosis in breast cancer. In contrast, Hsp90 expression is associated with good prognosis in endometrial cancer. Loss of Hsp90 (and Hsp60) expression has been associated in bladder carcinoma with invasive recurrence risk. Hsp90 expression was of no prognostic value in ovarian and oral cancer.

Table 2.

 Heat shock proteins in cancer: prognostic implications

It is evident that we need more studies on Hsps to confirm whether they have prognostic value in specific cancers. At this point, it is tempting to speculate that the unique molecular context or milieu present in each cancer type drives the correlations of Hsps with the disease prognosis. For example, in breast cancers, Hsp90 can be chaperoning the oncoprotein HER-2/neu and the mutated p53 protecting these molecules from degradation by the proteasome, which is good for the tumor but bad for the patient. In contrast, in endometrial cancers Hsp90 may be chaperoning progesterone receptors contributing to its maturation, maintaining a more differentiated and less aggressive tumor phenotype with a better response to synthetic progestational agents. Therefore, depending of the cancer type, each Hsp has unique associations with the prognosis of the disease.

PREDICTIVE IMPLICATIONS

There are studies exploring the use of the Hsps to predict the response (or lack of response) of a set of cancer patients to a specific treatment(s) (Table 3). These studies are very important because they may tailor the treatment strategy to individual cancer patients. These studies may be aided by a clearer understanding of the molecular link between the malignant phenotype and Hsp expression although such studies are currently at a very early stage. The following conclusions can be deduced from the published data (considering only those papers that presented a relatively large number of patients, with an homogeneous treatment, and with a clinical follow-up). (a) Hsp27—Although the expression of Hsp27 correlated with that of ERα in breast cancer, detection of Hsp27 does not predict the response to tamoxifen. Hsp27 expression was found in 31% of prostate cancer patients refractory to hormone therapy (Bubendorf et al 1999), although we need more studies on this important subject. Regarding chemotherapy, Hsp27 overexpression has been correlated with a shorter disease-free survival in advanced breast cancer patients who received neoadjuvant chemotherapy (Vargas-Roig et al 1998). This clinical implication of Hsp27 expression with resistance to chemotherapy is in agreement with studies performed in ovarian cancer, in head and neck cancer, esophageal squamous cell carcinoma, and leukemia (in association with other molecular markers). Hsp27 has shown no predictive value to chemotherapy in rectal cancer, malignant fibrous histiocytoma, and central nervous system tumors (induction radiochemotherapy). Regarding the brain tumors, we should point out that glioblastoma multiforme is rather resistant to radiochemotherapy and that these tumors already show a high expression of Hsp27 (as well as other Hsps) and that a further elevation in Hsp content may not correlate with a more resistant phenotype. We need more studies to know whether Hsp27 is related to radioresistance or radiosensitivity in cancer. The molecular mechanisms involving Hsp27 and other Hsps in resistance to cancer therapies can be explained in several ways: (1) as molecular chaperones they can confer cytoprotection by repairing more efficiently the damaged proteins resulting from cytotoxic drug administration, (2) protecting cancer cells from apoptosis (Arrigo et al 2002), (3) protecting the microvasculature inside tumors, because Hsp27 is found in endothelial cells (Ciocca et al 2003), and (4) enhancing DNA repair (Mendez et al 2003; Nadin et al 2003). (b) Hsp70—Although Hsp70 expression is associated with ERα expression in breast cancer, Hsp70 (like Hsp27) did not show predictive value for tamoxifen administration. In contrast, Hsp70 is emerging as a predictor of resistance to chemotherapy in breast cancer. Moreover, high Hsp70 levels predicted lower response of breast cancers to radiation and hyperthermia. In a recent study in multiple myeloma cells using oligonucleotide arrays, Chauhan et al (2003) identified several members of the Hsp family (including Hsp70) among the molecules conferring resistance to the conventional treatment with dexamethasone. Interestingly, they reported a new compound that overcomes dexamethasone resistance, which decreased the levels of Hsp27, Hsp70, and Hsp90 in the myeloma cells. Cancer cells have several defense mechanisms against cytotoxic drugs, which may be redundant, and in order to predict more accurately the resistance of cancer cells to certain therapies, it will be necessary to examine alternative pathways. Moreover, at the proteomic level, we will need to examine not only the expression but also the localization of the Hsps because this seems to be an important factor in their predictive value (Vargas-Roig et al 1998). Depending on the type of cancer (and their molecular profile and interactions), the Hsps can have a more marginal predictive role; for instance, in lung cancer, Hsp70 showed a weak predictive value compared with other molecules (Volm and Rittgen 2000). On the other hand, in osteosarcomas, Hsp70 predicted a better response to neoadjuvant chemotherapy (Trieb et al 1998), which may be explained by the different molecular context of these tumors. However, we need more studies regarding the predictive value of Hsps in cancer to deduce the precise significance of HSP expression.

Table 3.

 Heat shock proteins in cancer: predictive implications

IMPLICATIONS IN THE TREATMENT

This is an exciting new door for the field of Hsps in cancer. Hsp and the HSF family could provide a true Achilles heel for cancer therapy because they seem to be required for cell survival during tumor progression and metastasis (Volloch and Sherman 1999; Hoang et al 2000; Nylandsted et al 2000a, 2000b; Jones et al 2004). Hsps or HSFs may be targeted by drugs and new classes of drugs targeting Hsps are beginning to be deployed, most notably at this time targeting Hsp90 (Neckers and Ivy 2003). The elevated Hsps may also provide a tempting target for immunotherapy protocols because they are able to chaperone tumor antigens and act as biological adjuvants to break tolerance to tumor antigens and cause immune killing by cytotoxic CTL and tumor regression (Arnold-Schild et al 1999; Belli et al 2002; Manjili et al 2002; Noessner et al 2002; Srivastava 2002; Todryk et al 2003; Castelli et al 2004; Daniels et al 2004). Dependence on the selective advantages for growth offered by the protective effects of Hsps may thus render tumor cells vulnerable to detection through immunosurveillance and killing by chaperone-based immunotherapy. Table 4 shows the preclinical and the few clinical studies implicating Hsps in cancer treatment.

Table 4.

 Heat shock proteins in cancer: treatment implications

Agents that modify the molecular levels or molecular capabilities of the Hsps—this is achieved by the inhibition of Hsp90 by the natural product geldanamycin or the geldanamycin analog 17AAG. These drugs target the nucleotide-binding site in the N-terminal domain of Hsp90, the same as the adenosine triphosphate–binding site (inhibition of the adenosine triphosphatase activity) causing inhibition of the binding of Hsp90 to the client proteins (Workman 2002). These proteins are stress response, or survival-related, or mutated proteins that without the binding to Hsp90 are not properly folded (less stable) and destroyed by the proteasome. For instance, Hsp90 has been shown to bind mutant p53 more avidly than wild-type p53 (Blagosklonny et al 1996). In normal cells most Hsp90 is not bound to other proteins. Therefore, the effect on normal cells of the Hsp90-inhibitory drugs is not so drastic as in tumor cells. In fact, these drugs are toxic, but the toxicity of 17AAG is manageable. This strategy is interesting because of its ability to affect multiple oncogenic substrates simultaneously; in the phase I clinical trials on cancer patients, 17AAG produced in some patients stable disease, higher apoptosis, and less proliferation of the tumors but with lower potency than radiotherapy or chemotherapy. Moreover, 17AAG can be used in combination with radiotherapy or chemotherapy (enhancement of sensitization). 17AAG has some limitations such as limited solubility with low oral bioavailability, complex formulation, and modest potency on targets, and it is a substrate for P-glycoprotein. This has increased the interest in the search for other Hps90 inhibitors. Finally, HSF-1, Hsp27, Hsp70, and grp78 are also targets of antisense oligonucleotide therapies or other manipulations with possibilities for anticancer therapies. These interesting approaches are still at the preclinical level.

Use of Hsps as carriers/adjuvant to present tumor molecules to the immune system—the objective is to elicit in a cancer patient a specific and active immune response against its own tumor using the Hsps as natural adjuvants that present to the immune system the molecules (usually protein fragments, polypeptides but also relatively large molecules) that have shielded the potential epitopes from immune recognition. The immunization is carried out with tumor-derived Hps (gp96, Hsp70, and others), which bring attached the specific tumor peptides. When injected as a therapeutic vaccine, the Hsps interact with receptors on the professional antigen presenting cells (dendritic cells, macrophages). These cells introduce the antigen(s) into the MHC class I and II pathways, inducing a specific cytotoxic T lymphocyte response and the production of proinflammatory cytokines (Table 4, Srivastava et al 1998). Another approach is to use recombinant Hsps with oncoproteins such as HER-2/neu or proteins from oncogenic viruses such as E7 of HPV. The tumor-derived auto-vaccines based on Hsp or the recombinant Hsp fusion proteins induce cytokine and costimulatory molecules with activation of CD4⁺ and CD8⁺ T cells and increases in CD11c⁺ cells and NK cells that kill tumor cells (Table 4, Rivoltini et al 2003). So far the most promising effects are being obtained in renal cancer and melanoma patients, but several other cancer patients are being treated with the vaccines based on Hsps including patients with colorectal, gastric and pancreatic cancers, leukemia, and lymphoma. These Hsp-based vaccines exhibit minimal toxicity, and if they continue to show good results, they may ultimately be incorporated into the armamentarium against patients with limited or minimal cancer disease (the immune system has a relatively limited capacity to kill large tumor burdens).

CONCLUSIONS

Our studies indicate a profound role for Hsp in many aspects of tumor progression and response to therapy. Although at the diagnostic level Hsps are not informative, they are effective biomarkers for carcinogenesis in some tissues and signal the degree of differentiation and aggressiveness of certain cancers. In addition, the levels of Hsp and anti-Hsp antibodies in the serum of cancer patients are useful in tumor diagnosis. Furthermore, several Hsps are implicated with the prognosis of specific cancers, including Hsp27 expression, which is associated with poor prognosis in gastric, liver, and prostate carcinoma and osteosarcomas, and Hsp70, which is correlated with poor prognosis in breast, endometrial, uterine cervical, and bladder carcinomas. Hsp may also predict the response to some anticancer treatments. Implication of Hsp in tumor progression and response to therapy has led to its successful targeting in therapy by two main strategies, including: (1) pharmacological modification of Hsp expression or molecular chaperone activity and (2) use of Hsps as adjuvants to present tumor antigens to the immune system. Study of Hsp in cancer at the cell and molecular level, although promising, is still in its infancy, and we currently have little information on how Hsp regulation is subverted in cancer and how Hsp dysregulation affects the molecular events involved with tumor growth, invasiveness, and metastasis. Such studies will be essential in interpreting and directing the studies aimed at targeting Hsps in cancer therapy.

Note: we apologize to the colleagues whose works have not been cited in the present review because of space limitations and our limitations in finding their work in the literature search.

Table 1.

 Continued

Table 1.

 Continued

Table 2.

 Continued