Targeting therapy-resistant cancer stem cells by hyperthermia

Abstract

Eradication of all malignant cells is the ultimate but challenging goal of anti-cancer treatment; most traditional clinically-available approaches fail because there are cells in a tumour that either escape therapy or become therapy-resistant. A subpopulation of cancer cells, the cancer stem cells (CSCs), is considered to be of particular significance for tumour initiation, progression and metastasis. CSCs are considered in particular to be therapy-resistant and may drive disease recurrence, which positions CSCs in the focus of anti-cancer research, but successful CSC-targeting therapies are limited. Here, we argue that hyperthermia – a therapeutic approach based on local heating of a tumour – is potentially beneficial for targeting CSCs in solid tumours. First, hyperthermia has been described to target cells in hypoxic and nutrient-deprived tumour areas where CSCs reside and ionising radiation and chemotherapy are least effective. Second, hyperthermia can modify factors that are essential for tumour survival and growth, such as the microenvironment, immune responses, vascularisation and oxygen supply. Third, hyperthermia targets multiple DNA repair pathways, which are generally upregulated in CSCs and protect them from DNA-damaging agents. Addition of hyperthermia to the therapeutic armamentarium of oncologists may thus be a promising strategy to eliminate therapy-escaping and -resistant CSCs.

Keywords:

• Hyperthermia
• cancer stem cells
• microenvironmental niche
• hypoxia
• radiation and chemotherapy resistance

Introduction

Despite advances in cancer treatment, overall survival of cancer patients remains limited and relapse occurs frequently, indicating that there are cancer cells in a tumour that escape from therapy or become therapy-resistant [1]. It is now generally assumed that at least a fraction of these cells are tumour-initiating cells, or so-called cancer stem cells (CSCs) [2]. CSCs are characterised by reduced or even arrested cell cycle progression and by increased DNA repair capacity, which render CSCs more resistant to therapy than the bulk of the tumour [3,4]. CSCs are found in oxygen-deprived (hypoxic) tumour areas, where they are surrounded by a specific microenvironment, the CSC niche, that supports and controls their growth [5,6]. CSCs and their niches have recently been visualised histochemically in primary brain tumours in hypoxic areas around arterioles, which are transport vessel and not oxygen- and carbon dioxide-exchanging vessels [7,8]. As CSCs are associated with aggressiveness of tumours and are negatively prognostic for overall survival [9,10], there is an urgent need for therapies targeting this CSC population, but so far only few successful approaches have been developed, such as in leukaemia where clinical trials are presently running that are focussed on forcing CSCs out of the haematopoietic stem cell niches [11]. A successful anti-CSC therapy should be characterised by the ability to (i) target cells in areas populated by CSCs which are often difficult to reach by conventional therapies; (ii) destroy or modify the tumour microenvironment that sustains CSCs; (iii) kill cycling and non-cycling cells; and (iv) overcome the efficient DNA repair pathways that protect CSCs’ genomes. Here we argue that hyperthermia, defined as local heating of the tumour to 40–43 °C, is one anti-cancer treatment that may fit this challenging bill.

Cancer stem cells

The idea that a unique subset of cancer cells drives tumour (re)growth crystallised only a decade ago [12–14], although the first hints of a unique population was found a decade earlier in acute myelogenous leukaemia [15]. It was shown that CD34⁺/CD38⁻ leukemic cells have the capacity of tumour-initiation and expansion, even though they only represent a small fraction of transformed cells that are hiding in hypoxic haematopoietic stem cell niches. All other leukemic cells that were not CD34⁺/CD38⁻ lacked the capacity to re-establish the disease [15,16]. The study of Lapidot et al. [15] described typical CSC features that are now routinely used to identify this population. CSCs share some hallmarks of normal tissue stem cells, like the potential of self-renewal, unlimited cell proliferation and the ability to re-populate and re-establish entire tissue or organ (or, in this case, tumour) structures, even from a single cell [13,17,18]. They may propagate by cell proliferation or originate from non-CSCs by de-differentiation [19–21] and their plasticity is stimulated by the microenvironment [21]. Microenvironmental factors contributing to plasticity include low pH [22], hypoxia [6] and inflammation [23]. Besides microenvironmental factors, de-differentiation of non-CSCs into CSCs is also induced by anti-cancer treatments, such as radiation therapy [24–28] and chemotherapy [29]. Some CSCs are characterised by reduced or arrested cell cycle progression and by increased DNA repair capacity, which makes them more resistant to therapies than the bulk of the tumour [3,4]. Glioblastoma CSCs are found in hypoxic tumour areas, supported by the vascular stem cell niche, containing endothelial cells [5,6]. Several markers have been shown to identify CSCs in various tumour types [30]. Some of these markers seem to characterise CSCs in multiple cancer types, while some tumours have their own stem cell signature which correlates with aggressiveness of the disease and can predict the survival of patients [9]. CSCs are considered to drive tumour recurrence and therapy resistance and some of the CSC markers, like aldehyde dehydrogenase (ALDH) activity, are indicative of radioresistance [31] and chemoresistance [32], hinting that this population may be difficult to eradicate by conventional anticancer therapies.

Hyperthermia

Hyperthermia, defined here as local heating of the tumour to relatively moderate temperatures of 40–43 °C for approximately one hour, is an excellent radiosensitiser and chemosensitiser effective in multiple tumour types [33–41]. Clinically, hyperthermia is always combined with either radiotherapy or chemotherapy [35,42–44]. When applied prior to conventional therapy, hyperthermia can increase blood flow, improve oxygenation and enhance the therapy effects by stimulating production of oxygen radicals [45–48]. An increased blood flow creates higher oxygen levels, changes the pH in the tumour, influencing hypoxic, malnourished and acidosis [49–51]. When hyperthermia is applied simultaneously with or shortly after radiotherapy, hyperthermia can interfere with the repair of therapy-induced DNA damage, thereby indirectly contributing to tumour cell killing [52,53]. Hyperthermia can also activate heat shock proteins (HSPs), triggering protein unfolding, thereby causing loss of functionality and cell death [54]. In all cases, hyperthermia can additionally enhance therapy responses by directly targeting the resistant hypoxic population [55]. Whatever the mechanisms, the treatment results in significantly improved local tumour control, reduction of metastasis and improved overall survival. For instance, in sarcoma patients the addition of hyperthermia to radiotherapy resulted in a 50% lower probability to develop distant metastases [56] and in cervical cancer patients it caused fewer recurrences, lower metastasis formation and higher overall survival than radiotherapy alone [57]. A phase III clinical trial demonstrated a 15% increase in local progression-free survival for soft-tissue sarcoma patients when hyperthermia was added to chemotherapy [37]. It is tempting to hypothesise that at least some of these effects are mediated by hyperthermia-mediated suppression of the CSC population via multiple avenues [58], which will be discussed in the next paragraphs. An indication is that CSCs appear to be more sensitive to hyperthermia than their counterpart normal stem cells [59].

Targeting the pro-CSC factors by hyperthermia

Quiescence and resistance to therapies targeting cycling cells

Like normal stem cells, CSCs exploit protective mechanisms that contribute to their survival and therapy resistance. Among these mechanisms is the maintenance of a quiescent state. Quiescent stem cells are characterised by their low RNA content and their lack of proliferation markers [60]. Quiescence may be a protective response triggered, at least in part, by hypoxia (see also section “Hypoxia”) and poor nutrient supply [61] or a high cell density [62]. Since non-cycling cells are generally less susceptible to radiation and chemotherapeutic DNA-damaging agents [63–65], quiescence may be an important contributor to CSC therapy resistance [3]. Paradoxically, eradication of the cycling tumour cells arrests disease progression but it can also stimulate proliferation of quiescent cells that evaded therapy, which can then repopulate the tumour [66–68].

Subsequently, cells enter S-phase, which is associated with increased (radio) resistance [69]. A combination of irradiation with hyperthermia activates cells from quiescence to S-phase [70]. Proliferating cells have been found to be sensitive to hyperthermia [71] and these cells are also sensitive to chemotherapeutic agents and radiation. Although not all effects of hyperthermia are completely clear, previous studies have demonstrated that hyperthermia prolongs the cell cycle at G1/S transition [72–74] and block cells in G2 [75] Moreover, hyperthermia has been found to decrease the hypoxic fraction of quiescent cells [76,77]. Therefore, hyperthermia in combination with conventional therapy may push CSCs out of quiescent state, after which they become more vulnerable to anti-cancer therapies.

Hypoxia

Oxygen supply is a key factor that affects tumour homeostasis and microenvironment. Rapidly dividing solid malignancies struggle to maintain sufficient oxygenation due to poor vascularisation, increasing the likelihood of hypoxic regions [78–82]. Tumour types in which hypoxia is common include malignant brain tumours [83], melanomas [84], cervical carcinomas [85] and soft tissue sarcomas [86]. Hypoxia is also correlated with poor prognosis and metastasis [87–89]. For instance, patients with hypoxic soft tissue sarcomas have a significantly higher incidence of lung metastasis and shorter disease-free survival than those with less hypoxic tumours [56,86]. Similarly, patients with hypoxic lymph node-negative cervical cancer treated with radiotherapy have a significantly higher risk of distant metastasis than patients with better oxygenated tumours [85]. Since oxygen stimulates induction of DNA damage by ionising radiation and some chemotherapeutic agents [90], hypoxia can decrease the efficacy of treatment and thus protect tumour cells from therapy [91–93].

Hypoxic areas are indispensable in any type of stem cell niche – either embryonic, adult or cancer. Hypoxia preserves the undifferentiated state of stem cells, reduces oxidative DNA damage [5] and stimulates upregulation of markers associated with a stem-like phenotype [94]. Importantly, the self-renewal and differentiation capacity of CSCs has been found to increase in regions deprived of oxygen [5]. In addition, hypoxia-inducible factor 1α (HIF1α), a protein abundantly expressed after radiation exposure [95] and under hypoxic conditions, contributes to CSC proliferation and survival [96]. HIF1α can directly upregulate the activity of Notch signalling, a pathway promoting survival and self-renewal of CSCs [97], and CSCs with a reduced HIF1 activity were unable to form tumours in vitro [98]. As a consequence, hypoxia has been suggested to promote the phenotype of CSCs, to enhance their tumorigenicity and to be a biomarker of radioresistance [99,100].

Hyperthermia appears to counteract some of the effects of hypoxia, primarily by stimulating the blood flow and increasing the leakiness of tumour blood vessels [101]. Hyperthermia has been shown to improve tumour oxygen supply in several clinical studies [102–104]. This was not only evident in chronically hypoxic areas, where hyperthermia decreased the amount of quiescent cells considerably, but was also detectable in the total tumour cell population [76]. In addition, re-oxygenation induces microenvironmental changes, such as decreased acidosis and induces cell cycle arrest, resulting in autophagy and eventually apoptosis [105].

Immune responses

Tumorigenesis can only occur when cancer cells evade the immune system. The immune response is critical for eliminating cancer cells based on the expression of tumour-associated antigens [106]. Additionally, inflammatory responses of the immune system can increase the burden of DNA damage and, as a consequence, induce tumorigenesis [107]. On the other hand, inflammation can cause an increase in body temperature, which may activate specific HSPs, such as HSP70 [108]. HSPs are potent immune modulators and can lead to stimulation of both the innate and adaptive immune responses against transformed cells. However, once a malignancy develops, immune cells, such as tumour-associated macrophages (TAMs), can also promote malignant cell growth by stimulating angiogenesis, invasion and metastasis [109], and high levels of TAMs are correlated with poor prognosis [110]. CD8+ and IFNɣ-producing helper T cells are also present frequently in the tumour microenvironment and they exert tumour and metastasis: promoting effects [111–114]. A recent study demonstrated that tumour-bearing mice had higher CSC-specific IgG levels in serum after vaccination with CSC lysate-activated dendritic cells (DCs). When administered complement, these mice eradicated CSCs more efficiently than mice vaccinated with whole tumour-activated DCs [115]. Furthermore, CSCs were targeted more efficiently in vitro by cytotoxic T cells from these CSC-vaccinated mice. CSC-derived antigens can thus directly stimulate anti-CSC immune responses. On the other hand, CSCs have been reported to suppress immune system functionality [116–118].

Anti-tumour immune responses can be stimulated by radiotherapy or chemotherapy [119] and enhanced by mild hyperthermia. Hyperthermia triggers the immune system by stimulating the production of HSPs [120] and their release by necrotic tumour cells [121,122]. HSPs, in turn, can trigger maturation of DCs, resulting in systemic anti-tumour response [123–125]. Furthermore, hyperthermia can enhance antigen display, facilitating recognition of malignant cells by the immune system [126,127]. In animal experiments, hyperthermia also increased levels of natural killer cells, resulting in significant tumour growth inhibition [128]. In conclusion, these results suggest that hyperthermia can be exploited to potentiate anti-tumour and anti-CSC immune responses.

Microenvironment

The CSC population co-exists in a symbiotic relationship with the tumour microenvironment, which preserves and regulates their plasticity, but is also affected by their presence [21,129]. Many mechanisms are involved in this bi-directional communication, as summarised by Plaks et al. [130]. Endothelial cells play an essential role in CSC proliferation [131]. They secrete nitric oxide, activating the Notch signalling pathway [132] which controls cell fate, self-renewal, angiogenesis and endothelium interactions in the CSC microenvironment. In glioblastoma, the Notch signalling causes radioresistance in CSCs [133] and inhibition of this pathway or angiogenesis can deplete the CSC population [134,135]. Furthermore, fibroblasts that inhabit the tumour microenvironment are essential for cell proliferation and for promotion of angiogenesis [136].

Both endothelial cells and fibroblasts have been found to be extremely sensitive to hyperthermia [137]. Hyperthermia can additionally inhibit endothelial cell adhesion [138] and cell proliferation and promote apoptosis [139]. The resulting reorganisation of CSC microenvironment can deprive CSCs of the factors that are essential for the maintenance of their stem-like potential and that promote their resistance to therapy.

Viral infections

Approximately 15% of human cancers are caused by infections with oncoviruses [140]. Infections with hepatitis B virus [141], Epstein–Barr virus [142], papilloma virus [143] or herpes virus-8 [144] can initiate cancer. Fortunately, in most cases a healthy immune system is capable of eliminating these viruses. However, in a small number of cases, viral infections can slowly progress into cancer, which can take years or even decades [145].

Human papilloma virus (HPV) is a pathogen that has been shown to affect CSCs homeostasis. Recent studies suggest that HPV plays an important role in stimulating CSCs. This may be caused by production of the viral oncoproteins E6 and E7 that interfere with tumour suppressor proteins p53 and retinoblastoma protein (Rb), abolishing cell cycle regulation and apoptosis [146–148]. Overexpression of E6 was found to promote stemness and self-renewal in HPV⁺ cervical CSCs [149] and the ALDH1⁺ CSC population in HPV⁺ tumours was considerably higher, than in HPV⁻ tumours [150]. Furthermore, various transcription factors that are overexpressed in non-CSCs and CSCs, such as Oct4, Nanog and Notch – essential for self-renewal and for re-establishing pluripotency – are altered by HPV infections [21]. HPV⁺ and HPV⁻ tumours differ in therapy sensitivity, which may be at least partly due to CSC fractions in these tumours [151–153]. Our recent study showed that hyperthermia temporary inactivates the viral oncoprotein E6 [75]. Since E6 normally suppresses p53-mediated cell death in cervical cancer, it is plausible that hyperthermia may also block HPV-induced immortality.

Hyperthermia stimulates immune function and tumour immunogenicity which may affect the tumour microenvironment, and other viruses may be tackled successfully as well [154]. However, further investigations are needed to elucidate the anti-viral effects and eradication of viral related cancer cells.

DNA repair

CSCs are often characterised by enhanced activity of DNA repair pathways [1,155]. Increased expression of major DNA repair proteins, including BRCA1, ATR and ATM, has been observed in pancreatic, breast and prostate CSCs [156–158]. Reduced DNA repair capacity in CSCs has also been reported (reviewed by Wang [159]). In any case, DNA-damaging therapies targeting CSCs certainly benefit from inhibition of DNA repair as well as from increasing the total DNA damage burden. Hyperthermia can likely perform both. First, elevated temperatures have been shown to inhibit pathways that are responsible for repairing DNA lesions that are relevant in clinical cancer treatment [160]. More specifically, hyperthermia induces degradation of the BRCA2 protein and thereby inactivates homologous recombination, one of the major pathways responsible for repairing DNA double-strand breaks [161,162]. Furthermore, hyperthermia (above 43 °C) has been found to interfere with base excision repair [163,164], leading to more severe breaks. Interference with DNA repair pathways is important to increase levels of DNA breaks, causing accumulation of unrepaired DNA lesions, resulting in cell death. Hyperthermia has been found to decrease the CSC population due to inhibition of DNA repair [28,58,165,166]. Hyperthermia not only inhibits repair of DNA damage caused by radiation but also suppresses AKT signalling, a radiation-induced survival mechanism preferentially utilised by the CSC population [167]. Second, it is becoming clear that hyperthermia can induce DNA damage by causing protein denaturation and interfering with DNA replication (reviewed by Oei et al. [168]). There has been a long debate whether hyperthermia induces DNA damage or not, as increased (unrepaired) DNA damage is observed after combinational therapies that include hyperthermia in nearly all studies that tested the effectiveness of hyperthermia. Direct methods to detect immediate induction of DNA breaks failed in most of the studies as mild hyperthermia interferes predominantly with DNA repair pathways. Therefore, the effects of hyperthermia on (unrepaired) DNA damages may therefore be caused by indirect effects ([169] and reviewed by Oei et al. [168]).

Summary and future perspectives

Conventional radiotherapy and chemotherapy target the bulk of the tumour but not the therapy-resistant CSCs, often resulting in tumour relapse (Figure 1). After tumour relapse, conventional therapies may be combined with hyperthermia, thereby increasing effectiveness of therapy. We argue that first-line conventional treatment combined with hyperthermia may directly sensitise CSCs by inducing leaky vessels that allow chemotherapeutics to reach deeper regions of the tumour; oxygen levels will rise (re-oxygenation) thereby increasing sensitivity to radiotherapy; increased number of DNA breaks and decrease of DNA repair cause accumulation of DNA damages; quiescent cells may repopulate after first doses of therapy and the immune system may be triggered to interfere with the CSCs microenvironment. All these factors may also prevent CSCs from acquiring resistance to therapy.

Figure 1. Schematic overview of targeting CSCs by hyperthermia. The tumour consists of many different cell types including CSCs. Conventional first-line radiotherapy and chemotherapy are able to destroy all tumour cells and especially CSCs that often reside in hypoxic areas. These cells are considered to drive tumour recurrence and metastasis formation. Combined with conventional first-line treatment, hyperthermia can target CSCs via multiple avenues, including stimulation of blood flow and re-oxygenation of hypoxic areas, causing blood vessel leakage, triggering the immune response and inhibiting DNA repair.

Hyperthermia is successful in sensitising tumours with hypoxic areas to radiotherapy and chemotherapy. The question is whether the effectiveness of hyperthermia on hypoxic tumour tissue is mediated through a direct effect of hyperthermia on CSCs or by a change in the tumour microenvironment after which the CSCs become more sensitive to radiotherapy or chemotherapy treatment, or both. One would expect that in case of a direct effect, hyperthermia affects the CSC population even without re-oxygenation of the CSC niche. However, clinical data seem to suggest that this re-oxygenation is needed for the effectiveness of hyperthermia [104]. But we cannot be certain that this precludes a direct effect as radiotherapy is less effective in the absence of oxygen radicals and chemotherapy may not be delivered to poorly-perfused hypoxic areas. In other words, even when hyperthermia directly eliminates CSCs, the traditional modalities remain ineffective without reperfusion and re-oxygenation, and these modalities are needed to eliminate the non-CSC tumour cell population.

This suggests that hyperthermia can sensitise CSCs directly to traditional cancer therapy and that this sensitisation may be the underlying mechanism responsible for at least part of the clinical success of hyperthermia. However, a thorough understanding of the mechanisms driving this sensitisation is lacking. It is difficult to derive conclusions from the presently available data because any effects of hyperthermia on the CSCs are difficult to distinguish from other suggested mechanisms that can account for the established effects of hyperthermia on hypoxic tumour areas. This distinction requires further research, that focuses on determining reliable and specific CSC markers in preclinical in vitro and in vivo tumour models under various microenvironmental conditions.

Isolation of CSCs on the basis of expression of specific CSCs markers, can help to dissect how hyperthermia affects CSCs. However, it is essential to study the microenvironmental factors to eliminate and target the complete tumour. In vivo experiments should be focussed on comparison of the effectiveness of radiotherapy and hyperthermia in aerobic and hypoxic tumour regions. The first important step to accomplish this is to visualise these hypoxic areas [170] to facilitate investigations whether hyperthermia indeed targets the cells in these areas. Imaging the dynamics of tumour responses in different parts of the tumour is also important. Furthermore, specific markers of CSCs are necessary to detect in order to observe whether CSCs are preferentially located in the tumour areas targeted by hyperthermia and to monitor whether the frequencies of cells that show stemness characteristics decrease sufficiently in numbers that they lose their ability to repopulate tumours.

Unravelling this mechanism may have major clinical impact as it will permit more effective multi-modality strategies against therapy-resistant tumour types.

Novelty and impact

Successful cancer therapy requires effective targeting of cancer stem cells. Cancer stem cells are resistant to conventional radiation and chemotherapy. Hyperthermia is an alternative therapeutic strategy to target this resistant cancer cell population.

Acknowledgements

The authors wish to thank Prof. Dr. C.J.F. van Noorden for discussion and useful suggestions on the manuscript. This work was supported by the Dutch Cancer Society (grants #UVA 2008-4019, #UVA 2012-5540 and #UVA 2011-4962), NWO Medium grant (#40-00506-98-16015) and the Maurits en Anna de Kock foundation.

Disclosure statement

No potential conflict of interests were disclosed.

Additional information

Funding

This work was supported by the KWF Kankerbestrijding (Dutch Cancer Society): [grant number UvA 2008-4019], [grant number UvA 2012-5540], [grant number UvA 2011-4962], de Nederlandse Organisatie voor Wetenschappelijk Onderzoek (NWO) [grant number 40-00506-98-16015]. The Maurits and Anne de Kock sponsored laboratory equipment.