Heat shock proteins in animal neoplasms and human tumours—a comparison

Abstract

Heat shock proteins (HSPs) are implicated in all phases of cancer from proliferation, impaired apoptosis and sustained angiogenesis to invasion and metastasis. The presence of abnormal HSP levels in several human tumours suggests that these proteins could be used as diagnostic and/or prognostic markers, whilst the direct correlation between HSP expression and drug resistance in neoplastic tissues means they could also be used to predict cancer response to specific treatment. HSPs have also been successfully targeted in clinical trials modifying their expression or chaperone activity. Preliminary studies in veterinary medicine have also demonstrated the presence of altered HSP expression in neoplasms, and the study of carcinogenesis and the role of HSPs in animal models will surely be an additional source of information for clinical cancer research.

Introduction

Heat shock proteins (HSPs), also known as “stress proteins”, are a large class of proteins that have been highly conserved throughout evolution and are expressed by prokaryote and eukaryote organisms. HSPs control protein biogenesis by assisting in the correct folding of newly formed polypeptides, oligomeric assembly and intracellular translocation (Mathew and Morimoto 1998; Nollen and Morimoto 2002) and are thus crucial in the maintenance of cellular homeostasis. HSPs also prevent inappropriate stress-induced protein aggregation by assisting in the repair of denatured proteins or by promoting their degradation. As a result of these roles, HSPs have also been referred to as molecular “chaperones” (Whitley et al. 1999). HSPs can be classified according to their molecular weight, expressed in kDa: HSP15–30, HSP40, HSP60, HSP70, HSP90 and HSP100. Each HSP family consists of several molecules, all sharing a similar primary structure and able to perform analogous functions in different subcellular compartments.

HSPs were so-called because their expression was induced by heat shock (Ritossa 1962; Tissieres et al. 1974). However, since then, a wide variety of environmental and metabolic factors including hypoxia, oxidative injury, glucose starvation, exposure to heavy metals or anti-cancer agents have been shown to elicit stress protein expression. Cellular stress response is a unique and important defence mechanism put into act by the cell to cope with a wide range of harmful conditions (Whitley et al. 1999). This response includes increased HSP synthesis, which has been detected in many pathophysiological conditions such as tissue injury and repair, hypertrophy, fever, inflammation, viral and bacterial infections (Morimoto 1998).

A growing body of evidence suggests that HSPs are also closely involved in a number of crucial processes in tumour development such as the regulation of cell cycle progression (Helmbrecht et al. 2000), control of apoptotic pathways (Didelot et al. 2006; Garrido et al. 2006; Schmitt et al. 2007) and immunosurveillance against cancer (Li 2001; Multhoff 2006). Indeed, studies are underway to determine whether these proteins could be used as diagnostic and/or prognostic markers or represent new targets for therapy. Altered HSP expression has been observed in preliminary studies on rodent and canine neoplasms suggesting a similar pattern of tumour development. These parallel findings underline the relevance of animal models in studies aimed at elucidating the multiple roles of HSPs in carcinogenesis both in animals and humans.

Comparative evaluation of altered HSPs expression in animal and human tumours

Since HSPs are overexpressed in many kinds of human malignant cells, from a diagnostic point of view, their immunodetection does not help in identifying the lineage of origin (Ciocca and Calderwood 2005). However, anti-αBcrystallin might be included in a panel of antibodies for the identification of renal cell carcinomas when a metastatic deposit or a small biopsy is evaluated (Pinder et al. 1994). Plasma levels of Hsp70, along with PSA, might also prove useful in the identification of patients with early-stage prostate cancer (Abe et al. 2004). In addition, serum levels of autoantibodies directed against HSPs in cancer patients could be of significance as tumour markers in different kinds of tumour (Korneeva et al. 2000; Trieb et al. 2000; Oka et al. 2001; Luo et al. 2002; Zhong et al. 2003).

In veterinary literature, high levels of Hsp60 and Hsp70 were reported in canine transmissible venereal tumour (CTVT), and it was thought that these HSPs could be considered potential markers for CTVT cells (Chu et al. 2001). However, more recent studies have shown high levels Hsp70 expression in canine mammary tumours (Kumaraguruparan et al. 2006; Romanucci et al. 2006), confirming that HSP expression cannot be relied upon for the recognition of a specific tumour histological type. Nevertheless, increased levels of Hsp60 have been linked to CTVT regression (Chu et al. 2001).

Many further studies have looked at the potential prognostic value of HSP expression; however, the data obtained so far are controversial and strictly linked to tumour type and organ. This is without doubt a reflection of the multiple and still unidentified roles exerted by HSPs both in different normal tissues and in cancer.

Hsp27 expression has been extensively studied in human breast cancer: Hsp27 overexpression has been correlated with oestrogen receptor levels (Thor et al. 1991; Hurlimann et al. 1993; Love and King 1994; Takahashi et al. 1995; O’Neill et al. 2004) and better differentiation of cancer cells (Love and King 1994; Têtu et al. 1995). However, other findings indicate that some but not all oestrogen receptor-positive breast tumours express Hsp27 (Ciocca et al. 1993b). In vitro data suggest that Hsp27 expression is associated with resistance to chemotherapeutic drugs (Oesterreich et al. 1993; Conroy and Latchman 1996; Hansen et al. 1999). In fact, despite the positive link with oestrogen receptors, suggesting a correlation between high amount of Hsp27 and better prognosis, an association between Hsp27 overexpression and more aggressive tumours has also been detected (Thor et al. 1991). Likewise, Hsp27 positivity in tumours from node-negative patients was correlated to lower overall survival and survival after first recurrence (Thanner et al. 2005). We have also observed a similar correlation between Hsp27 expression and tumour invasiveness in association with reduced overall survival in canine malignant mammary neoplasms (Romanucci et al. 2006). The detection of Hsp27, particularly in canine mammary infiltrating neoplastic cells, supports the theory that Hsp27 overexpression may influence the invasive and metastatic potential of human breast cancer cells (Lemieux et al. 1997) by controlling their migration on laminin-5 (Rust et al. 1999). In fact, treatment of tumour cells with a synthetic inhibitor of Hsp27 phosphorylation (Shin et al. 2005) and knockdown of such HSP using transfection with short interference RNA (Shin et al. 2005; Bausero et al. 2006) has been found to halt tumour cell migration. In Hsp27-overexpressing human breast cancer cells, an increased expression of matrix metalloproteinase 9 has also been observed and appears to be correlated with a down-regulated expression of the Src family tyrosine protein kinase Yes (Hansen et al. 2001). However, Hsp27 levels have also been correlated with different biological features in early and advanced human breast cancer such as short disease-free survival in node-negative patients but with prolonged survival from first recurrence. It is thought that the high levels of Hsp27 in advanced cancer are indicative of long survival because of the link to hormone response; however, the biological explanation for the switch from Hsp27 being a bad to a good prognostic factor in early and advanced breast cancer remains to be clarified (Love and King 1994). Moreover, Hsp27 seems to sort out cases with a better prognosis from the oestrogen receptor-negative group of patients with a poor prognosis (Hurlimann et al. 1993). Nevertheless, other findings reveal a lack of association between Hsp27 expression and the clinical outcome of this kind of neoplasm (Hurlimann et al. 1993; Têtu et al. 1995; Oesterreich et al. 1996; Ioachim et al. 2003) and its response to hormone therapy (Hurlimann et al. 1993; Ciocca et al. 1998). Antibodies to Hsp27, on the other hand, have been associated with improved survival in patients with breast cancer (Conroy et al. 1998a).

Overexpression of Hsp70 has been frequently observed in several kinds of human tumours and, in particular, breast cancer where Hsp70 expression has been correlated to adverse prognostic indicators, such as high tumour grade and presence of nodal metastasis (Lazaris et al. 1997), and appears to negatively influence overall survival and survival after recurrence (Thanner et al. 2003). It could well prove useful in sorting out node-negative patients at high risk of recurrence, thus influencing decisions regarding treatment (Ciocca et al. 1993a). A strict correlation between Hsp70 levels and oestrogen receptors has also been detected, which is in agreement with other research demonstrating the association of this protein with steroid hormone receptors (Takahashi et al. 1994).

Hsp90 expression has also been extensively studied in tumours, predominantly in breast cancer where a positive relationship with oestrogen receptor levels has been found (Shyamala et al. 1993). Hsp90 is a fundamental component of the multi-molecular, steroid receptor complex (Cheung and Smith 2000). Similarly to Hsp70 (Vargas-Roig et al. 1997), this Hsp also seems to be involved in the proliferation of human breast cancer, as levels of Hsp90α, an isoform of the HSP90 family, appears positively correlated with cyclin D1 expression in this type of tumour (Yano et al. 1999). In addition, the presence of autoantibodies to Hsp90 in the sera of breast cancer patients has been associated with poor survival (Conroy et al. 1998b). Hsp90 overexpression has also been reported to indicate a poor prognosis in human breast cancer (Jameel et al. 1992), defining a population of patients with decreased survival (Pick et al. 2007).

In canine malignant mammary tumours, although Hsp70 and Hsp90 levels were not of significant prognostic value, the high Hsp90 expression levels detected in neoplastic tissues, independently of tumour histological type or aggressiveness (Romanucci et al. 2006), suggest that such proteins could play a fundamental role in the multiple processes leading to malignant transformation and tumour progression in the canine mammary gland. Many of the mutations in oncogenes and tumour suppressor genes commonly found in cancer result in the expression of defective proteins that display unusually stable physical association with molecular chaperones. These molecular chaperones, particularly Hsp90, seem to serve as biochemical buffers at the phenotypical level for the multiple genetic lesions which usually characterize tumours, thus permitting cells to tolerate the mutations of crucial signalling molecules that would otherwise be lethal (Whitesell and Lindquist 2005). Furthermore, in breast cancer cells, Hsp90 is essential for the stability and function of steroid hormone receptors (Pratt and Toft 1997), whose expression has been found in both normal and neoplastic canine mammary tissues (Donnay et al. 1993). Likewise, the membrane receptor tyrosine kinase ErbB2 is also a Hsp90 client protein (Xu et al. 2001), whose enhanced expression correlates with malignancy of breast cancer progression (Miyata 2005) and which might also exert an important role in carcinogenesis of canine mammary gland (Ahern et al. 1996; Matsuyama et al. 2001; Martin de las Mulas et al. 2003; Dutra et al. 2004).

The elevated expression of the HSP70 family members in both cytoplasm and nucleus of canine mammary tumour cells, characterised by intense proliferation activity and/or stromal invasion (Romanucci et al. 2006), could be correlated to the roles exerted by these chaperones in cell cycle control (Helmbrecht et al. 2000). In addition, several mammalian cells typically show an increase and a nuclear translocation of Hsp72/73 (respectively, the inducible and constitutive member of HSP70 family) during S-phase, which suggests an enhanced requirement for nuclear protein transport during this phase (Milarski and Morimoto 1986; Shi and Thomas 1992; Zeise et al. 1998). These cells could also be manifesting the symptoms of environmental stress, such as lack of nutrients or hypoxia (Kaur et al. 1998; Jolly and Morimoto 2000), particularly in the more aggressive tumour areas. However, little information is available to support this latter hypothesis and in contrast, HSP expression, induced through the stress protein response, appears to interfere with other gene expression programmes in the cell, such as mitogenic signal transduction pathways (Calderwood 2005).

Increased transcription of hsp genes, on the other hand, may be directly induced by basic oncogenic pathways (Calderwood et al. 2006), such as those involving the c-myc oncogene or p53 protein. Whereas c-myc does not appear to exert a prominent role in oncogenesis of canine mammary gland (Engstrom et al. 1987), the p53 tumour suppressor gene results to be involved in both human and canine mammary tumour development and progression (Van Leeuwen et al. 1996; Kumar et al. 2007). Regulation of HSP expression in normal cells also involves the tumour suppressor protein p53, which represses transcription of the hsp70 gene through the inhibition of CBF/HSP70, a transcription factor binding to the CCAAT box on the hsp70 promoter (Agoff et al. 1993; Chae et al. 2005). In fact, mutation of the p53 gene reverses this effect with consequent transactivation of the hsp70 promoter (Tsutsumi-Ishii et al. 1995). Furthermore, both Hsp72/73 and Hsp90 have been found to be associated with the conformational mutant form of p53 forming a multi-chaperone complex which mediates the stabilisation, cytoplasmic sequestration and accumulation of mutated p53 by masking the p53 nuclear localisation signal (Akakura et al. 2001) and preventing its MDM2-mediated ubiquitination (Peng et al. 2001).

A recent study has also demonstrated a similar pattern of change in Hsp70, Hsp90 and apoptosis-associated proteins, such as Bcl-2, Bcl-X_L, Bax, Caspases 3 and 8, in both human and canine mammary tumours. The resulting shift of balance towards expression of HSPs and anti-apoptotic proteins suggests the existence of similar mechanisms to evade apoptosis in both humans and canines (Kumaraguruparan et al. 2006). In this connection, an increasing number of studies have greatly contributed in defining the anti-apoptotic activity of several HSPs, including Hsp70 and Hsp90, which can interfere with both the mitochondrial (“intrinsic”) and death receptor-mediated (“extrinsic”) apoptotic pathways (Didelot et al. 2006; Garrido et al. 2006; Schmitt et al. 2007). It seems to be conceivable that HSPs might play analogous functions both in humans and animals, as the amino acid sequence of the canine hsp70 gene shares 90–95% sequence similarity to the bovine, human and mouse Hsp70 proteins (Kano et al. 2004).

HSPs expression has also been investigated in human, mouse and canine cutaneous squamous cell carcinoma (SCC). The data obtained from these studies indicate that Hsp27 expression is strictly correlated to keratinocyte differentiation, suggesting that the absence of this protein in epidermal cells could be regarded as a marker of epidermal malignancy in all the species so far investigated (Trautinger et al. 1995; Kiriyama et al. 2001; Romanucci et al. 2005). As a matter of fact, canine Hsp27 is also very similar to the human form with its primary structure deduced from nucleotide sequence revealing a 209 amino acid protein sharing 86–89% homology with human, mouse, rat and hamster small Hsp (Larsen et al. 1995). In human epidermis, Hsp27 appears to operate as a chaperone of cornification, as it colocalises with keratins and proteins of the cornified cell envelope (Jonak et al. 2002), whilst another study suggests that Hsp27 could be also involved in the regulation of differentiation-associated gene expression (Hell-Pourmojib et al. 2002).

Finally, in both canine mammary tumours and cutaneous SCC, Hsp90 and Hsp73 exhibit a clear-cut expression in mitotic cells (Romanucci et al. 2005, 2006), lending further support to the role of HSPs in regulating the assembly of mitotic apparatus. In fact, Hsp73 has been found to localise on centrosomes where it probably assists the centrosomal chaperonine tailless complex protein-1 (TCP-1) in tubulin folding (Brown et al. 1996a, b). It has also been found on the fibres of spindles and asters during metaphase (Agueli et al. 2001). Hsp90 is also a core centrosomal component, and it has been found that HeLa cells treated with a specific competitive inhibitor of Hsp90, geldanamycin, tend to stop at metaphase (Lange et al. 2000). Hsp90 seems to regulate metaphase–anaphase transition (de Carcer 2004) by promoting the stabilisation of the Polo kinase, an essential centrosomal protein which regulates several aspects of cell division including centrosome maturation and function (de Carcer et al. 2001).

Animal models in HSP-based cancer therapy

Although potentially dangerous, the ability of HSPs to stabilise altered conformations of signal transduction molecules and to impair apoptotic pathways also represents a weakness as far as tumour cells are concerned, as the inhibition of their chaperone function can be expected to affect the survival of such cells, independently of the alteration responsible for the oncogenic phenotype (Mosser and Morimoto 2004). Therefore, HSPs have become targets for anti-cancer drug design: in particular, Hsp90 has emerged as an especially promising molecular target, given its interaction with over 100 client proteins, many of which are involved in cancer-associated signalling pathways. Consequently, inhibition of Hsp90 functions can affect multiple oncogenic substrates simultaneously, thus helping to circumvent the genetic plasticity that may allow cancer cells to escape the toxic effects of most molecularly targeted agents which attack on a single signalling node (Neckers 2006). Although the combinatorial action of Hsp90 inhibitors is a major advantage of this class of anti-cancer drugs, this does not exclude a role for their action against a specific oncogene product in particular tumours (Sharp and Workman 2006), such as ErbB2 in breast cancer or B-Raf in melanoma. Furthermore, as Hsp90 has been demonstrated to play crucial roles in regulating angiogenic responses, evidence suggests that Hsp90 inhibitors may provide therapeutic benefit not only via direct effects on tumour cells but also by interfering with several steps of tumour angiogenesis (Kaur et al. 2004; Sanderson et al. 2006).

Even if Hsp90 represents 1–2% of the total cellular protein content and chaperones several proteins that are essential for maintaining homeostasis of healthy cells, Hsp90 inhibitors have proven to be well tolerated. One possible explanation for their therapeutic selectivity against neoplastic cells is that oncogene-addicted tumour cells are far more sensitive that normal cells, which are responsive to a plethora of pathways and stimuli (Pearl 2005). Then, given the genetic instability which is a common event in tumour genesis, and the ability of Hsp90 to function as a biochemical buffer of the multiple genetic lesions, which usually characterise tumour cells, it is likely that Hsp90 in cancers could be far more involved in the constitution of multi-chaperone complexes, thus displaying a higher ATPase activity with an apparently higher affinity for Hsp90 inhibitors than does “free” Hsp90 in normal cells (Kamal et al. 2003; Pearl 2005). In fact, most Hsp90 inhibitors act by docking in the N-terminal nucleotide binding domain, thereby inhibiting intrinsic ATPase activity and thus blocking the formation of mature complexes. Such inhibitors include the benzoquinone ansamycin antibiotic geldanamycin and its derivatives, the macrocyclic antibiotic radicicol and its analogues, purine-scaffold derivatives and shepherdin (Sharp and Workman 2006; Xiao et al. 2006). The latter is specifically designed to block the interaction between Hsp90 and survivin (Plescia et al. 2005). 17-Allylamino, 17-demethoxygeldanamycin (17-AAG) has recently completed several phase I clinical trials (Banerji et al. 2005; Goetz et al. 2005; Grem et al. 2005; Ramanathan et al. 2005; Nowakowski et al. 2006) and entered phase II single agent therapy in various tumour types including melanoma, breast cancers and paediatric (Sharp and Workman 2006) and genitourinary (Lattouf et al. 2006) malignancies. There is also great interest in combining 17-AAG treatment with other cancer therapies, such as radiation (Bisht et al. 2003; Enmon et al. 2003; Russell et al. 2003; Machida et al. 2005; Shintani et al. 2006) or various cytotoxic agents (Nguyen et al. 2001; Rahmani et al. 2003; Solit et al. 2003; George et al. 2004, 2005; Mesa et al. 2005; Vasilevskaya and O’Dwyer 2005; Yao et al. 2005; Barker et al. 2006; Sain et al. 2006; Premkumar et al. 2006), as 17-AAG can sensitise tumour cells to the induction of apoptosis by other treatments.

The most limiting factor in clinical trials is that 17-AAG has poor solubility in water and lacks oral bioavailability. Thus, its highly soluble hydroquinone hydrochloride derivative IP-504 has been synthesised as an Hsp90 inhibitor and appears to be effective in cellular and mouse models of myeloma (Sydor et al. 2006). In addition, a second generation analogue of geldanamycin, 17-(dimetylaminoethylamino)-17-demethoxygeldanamycin (17-DMAG) has been developed, which is water soluble and orally bioavailable. A series of preclinical studies has been carried out to establish its in vitro and in vivo anti-tumour activity and spectrum of toxicity (Bull et al. 2004; Eiseman et al. 2005; Glaze et al. 2005; Hollingshead et al. 2005; Smith et al. 2005; Robles et al. 2006), which appears to be similar to 17-AAG. 17-DMAG is currently in phase I clinical trials (Shadad and Ramanathan 2006; Sharp and Workman 2006).

Despite their promising anti-cancer properties, one concern in the clinical application of Hsp90 inhibitors is that they may induce the expression of HSPs, including Hsp70, via the activation of HSF1 (Bagatell et al. 2000). The blocking of Hsp70 induction has been observed to significantly enhance the anti-leukaemia activity of 17-AAG (Guo et al. 2005). Hsp27 up-regulation could also play a significant role in 17-AAG resistance which may be mediated, in part, through glutathione regulation (McCollum et al. 2006). Further evaluation of Hsp90-targeted cancer therapy also appears to be essential, as a potential contraindication to this therapy has been found: 17-AAG appears to enhance bone metastasis of a human breast cancer cell line following intracardiac inoculation in the nude mouse (Price et al. 2005). Such findings underline the importance of in vivo models for further testing of Hsp90-targeted cancer treatments, and the abundant Hsp90 expression detected in canine malignant mammary tumours (Romanucci et al. 2006) suggests that the canine model may well prove useful in the testing of new breast cancer therapy.

In veterinary medicine, mammary tumours constitute the most common malignant neoplasms in the bitch (Misdorp 2002), showing wide pathological and clinical heterogeneity similar to the disease in humans. Similarities between human and canine mammary neoplasms on a molecular level allow more significant comparative evaluations of the molecular mechanisms involved in carcinogenesis with respect to the classical rodent model (Kumaraguruparan et al. 2006). As a matter of fact, translation of a therapeutic into the clinic requires the use of animal models that parallel the biological, genetic, etiological, immunological and therapeutic properties of human cancer (Talmadge et al. 2007). Several characteristics allow to consider spontaneously occurring tumours in dogs as an attractive model for human cancer (Vail and MacEwen 2000; Sutter and Ostrander 2004). In this respect, there is a greater genetic homology between dogs and humans than between either species and the mouse (Kirkness et al. 2003; Switonski et al. 2004). Furthermore, companion animals live in the same environment as humans and share similar environmental risk factors (Mueller et al. 2007). Naturally occurring canine neoplasms also represent autochthonous tumour models which are believed to reproduce human tumours more closely than transplanted tumours, as they show orthotopic growth, tumour histology devoid of transplantation induced changes, metastasis via lymphatic and vascular vessels surrounding and within the primary tumour (Talmadge et al. 2007). Since adjuvant treatments are mainly aimed at controlling micrometastases, the strong Hsp90 and Hsp73 immunolabelling detected in canine mammary neoplastic emboli (Romanucci et al. 2006) is significant as it suggests that these HSPs are necessary to cells with metastatic potential and that the inhibition of their functions could affect the survival of such cells, which does not always show the same pattern of expression respect to the primary tumour (Cardoso et al. 2001).

The anti-apoptotic relevance of Hsp70 in cancer cells has been confirmed, both in vitro and in vivo, by evaluating the effects of “antisense Hsp70 sequences” (AsHsp70) (Gibbons et al. 2000; Kaur et al. 2000; Nylandsted et al. 2002; He et al. 2005; Zhao and Shen 2005). The AsHsp70-induced apoptosis seems to be caspase-independent and not rescued by the Bcl-2 anti-apoptotic protein (Nylandsted et al. 2000). Thus, Hsp70 depletion may provide a new target for cancer therapy (Jones et al. 2004), especially when acquired chemoresistance occurs (Gabai et al. 2005). AsHsp70 could be particularly useful in the therapy of tumours characterised by local tissue infiltration and invasion without metastasis (Nylandsted et al. 2002). In addition, a peptide containing the AIF sequence involved in its interaction with Hsp70, called the AIF-derived decoy for Hsp70 (ADD70), has been shown to bind to and neutralise Hsp70 in the cytosol, thereby sensitising cancer cells to apoptosis induced by a variety of stimuli (Schmitt et al. 2003) and exerting anti-tumour effects in rodent models of colon cancer and melanoma (Schmitt et al. 2006). Notwithstanding this, drugs selectively inhibiting Hsp70 have not yet been identified.

Concluding remarks

Even if the roles of HSPs in cancer have not yet been completely clarified, the data so far obtained clearly indicate that they are involved in all the aspects of tumour biology. HSPs are essential for the survival and proliferation of neoplastic cells and represent targets for anti-cancer therapy. Preliminary studies carried out on animal tumours have identified similar changes in HSP expression with respect to their human counterparts, thus indicating similar roles/functions during human and animal carcinogenesis. Recent data suggests that the canine model would make a more suitable model with respect to the traditional rodent model to investigate the molecular mechanisms of tumour development and progression and to test the efficacy of new anti-cancer treatments.