Hypoxia and Regulation of Cancer Cell Stemness

Abstract

Spontaneous tumors often contain heterogeneous populations of tumor cells with different tumor-initiating potentials or cancer cell “stemness.” Clonal heterogeneity can be traced to specific locations inside a tumor where clones with different metastatic capabilities are identified, suggesting that the tumor microenvironment can exert a significant effect on the evolution of different clonal populations. Hypoxia is a common feature of tumor microenvironments and has the potential to facilitate malignant progression. This chapter provides a synopsis of hypoxia-regulated pathways implicated in the maintenance of cancer stem cells.

2.1 Introduction

In primary tumors, there are often functional and phenotypical heterogeneities among tumor cell populations (Marusyk et al. 2012) although they share the same clonal origin. As was elegantly shown by Yachida et al. (2010), clonal populations with variable metastatic potentials are found in distinct regions within the primary carcinoma of patients with pancreatic cancer, although these clones are genetically evolved from the original parental, nonmetastatic clone. The mechanisms underlying clonal heterogeneity, however, remain to be investigated. Nonetheless, it is highly possible that the heterogeneous nature of tumor microenvironments plays a critical role in the evolution and selection of aggressive clones.

One of the most commonly recognized features of tumor microenvironment is hypoxia, that is, insufficient oxygenation to meet the metabolic demands of viable tumor cells (Vaupel and Mayer 2007). Hypoxic (oxygen partial pressure pO₂ < 10 mmHg) regions have been directly detected using a needle-based, polarographic pO₂ electrode in many types of human cancers (Vaupel et al. 2007). The presence of tumor hypoxia also has been indirectly analyzed by immunohistochemistry using hypoxia-activated compounds (Evans and Koch 2003), such as EF5 and pimonidazole, or endogenous hypoxia-induced molecules (Moon et al. 2007), such as hypoxia-inducible factor (HIF)-1α, glucose transporter 1, and carbonic anhydrase 9. However, the staining patterns of different hypoxia markers are not identical (Li et al. 2007b; Vukovic et al. 2001), likely because of their different modes of activation and/or regulation by hypoxia. Nevertheless, it has been shown that hypoxic regions are randomly distributed within the tumor proper (Horsman et al. 2012), suggesting that hypoxia may play a critical role in the clonal evolution of tumor cells in different tumor microenvironments.

Tumor hypoxia is, clinically, an independent prognostic factor for poor patient survival (Nordsmark and Overgaard 2004; Brizel et al. 1996, 1999; Hockel et al. 1996 ; Young et al. 1988). Hypoxic tumor cells seem to be more aggressive, with reduced apoptosis (Graeber et al. 1996), increased drug-resistance (Wartenberg et al. 2003; Comerford et al. 2002), and increased metastatic potential (Rofstad 2000; Subarsky and Hill 2003). Hypoxia can also increase genomic instability by down-regulating the expression of DNA repair genes (Koshiji et al. 2005; Mihaylova et al. 2003; Bindra et al. 2004, 2005). These observations strongly suggest that hypoxia exerts a powerful selection pressure for the emergence of aggressive tumor clones.

The malignant progression of a benign growth is a slow process. It often takes more than a decade for metastatic clones to emerge in spontaneous human tumors (Luebeck 2010; Yachida et al. 2010; Jones et al. 2008; Beerenwinkel et al. 2007). Malignant progression often results from cumulative genetic mutations in oncogenes and tumor suppressor genes, as well as epigenetic changes (Hanahan and Weinberg 2000; Vogelstein and Kinzler 2004). It is imperative to note that these seemingly random and independent mutational events must take place in a single tumor cell originating from the initial oncogenic transformation and that this tumor cell must be able to copy itself so that previously acquired mutations can be inherited in the subsequent daughter cell stages. Therefore, only a stem cell–like cancer cell can complete this protracted journey of change from a benign cell to a metastatic tumor cell.

A number of recent studies have shown that hypoxia can inhibit differentiation of embryonic stem cells and progenitor cells (Ezashi et al. 2005; Gustafsson et al. 2005; Lin et al. 2006; Yun et al. 2002). Hypoxic tumor cells seem to be poorly differentiated and express stem cell markers (Das et al. 2008; Jogi et al. 2002). Under hypoxic conditions, tumor cells show increased clonogenic potential (Desplat et al. 2002; Kim et al. 2009; Schmaltz et al. 1998). Exposure to hypoxia in vitro also results in enhanced tumorigenic potential in vivo (Jogi et al. 2002). These interesting observations lead to a new paradigm that tumor hypoxia may facilitate the emergence of malignant clones by maintaining cancer stem cells in their undifferentiated stem cell state, which permits self-renewal and uninterrupted accumulation of genetic and epigenetic changes over a protracted period of time. This chapter briefly reviews the current advances in the understanding of hypoxia and its role in stem cell maintenance. The field of cancer stem cell research is witnessing an explosive expansion. We apologize to those authors whose work is not cited because of space limitations.

2.2 Mechanisms of Hypoxia-Dependent Stemness Regulation

2.2.1 Hypoxia-Inducible Factors and Cancer Cell Stemness

In addition to being an essential molecule for oxidative phosphorylation in mitochondria, oxygen (O₂) also functions as an important signaling molecule and regulates a wide range of biological processes, including erythropoiesis, angiogenesis, and cellular differentiation. O₂ cannot be stored in cells and needs to be constantly supplied to support cellular functions and maintain cell viability. Because of the limited supply, specific O₂-sensing pathways have evolved in higher-order organisms, especially mammals, to deal with potential O₂ deficiency.

The most prominent and best understood hypoxia-induced signaling pathways currently are anchored by HIF-1 and HIF-2, heterodimeric transcription factors consisting of an O₂-regulated alpha subunit (HIF-1α or -2α), and the O₂-insensitive HIF-1β (Semenza 2003). Although they share similar structures and functions, HIF-1α is ubiquitously expressed, whereas HIF-2α has relatively limited tissue distribution; they also each have nonoverlapping functions (Hu et al. 2006). Furthermore, the expression of HIF-1α and HIF-2α is differentially regulated under conditions of acute and chronic hypoxia, respectively (Lin and Yun 2010).

The O₂-sensing ability of HIF-α subunits is realized via O₂-dependent hydroxylation of two proline residues located in the O₂-dependent degradation domain (Ivan et al. 2001; Jaakkola et al. 2001). The hydroxylated HIF-α interacts with the von Hippel-Lindau (VHL) protein in the E3 ligase complex for ubiquitination and proteasome-mediated degradation (Maxwell et al. 1999; Ohh et al. 2000). Under hypoxia (generally at pO₂ levels < 2 %), proline hydroxylation is impaired and the unhydroxylated HIF-α translocates into the nucleus, where it dimerizes with the O₂-insensitive HIF-1β. The enhancer regions of hypoxia-induced genes typically contain one or more of the consensus sequence 5′-ACGTG-3′, dubbed the hypoxia-responsive enhancer element (HRE), which is directly bound by HIF-1 or -2 (Semenza 2000; Harris 2002).

In general, increased HIF accumulation and activity facilitate tumor development, which is perhaps best illustrated by renal cell carcinomas (RCCs). Genetic mutations of the VHL tumor suppressor gene result in loss of function of the VHL tumor suppressor protein and consequent activation of the HIF pathway under normoxia, which promotes RCC development (Ohh et al. 2000; Maxwell et al. 1999). However, it should be noted that Vhl deletion in murine renal proximal tubule cells does not lead to the development of renal cancers (Rankin et al. 2006), suggesting that additional pathways also are required for RCC development. Nevertheless, solid tumors often show elevated levels of the HIF-1α protein compared to adjacent normal tissues (Harris 2002; Semenza 2003; Vaupel and Mayer 2007). Elevated levels of HIF-1α protein (Aebersold et al. 2001; Burri et al. 2003) or HIF-2α protein (Holmquist-Mengelbier et al. 2006) are significantly correlated with poor patient survival. Furthermore, studies have shown that HIF-1α and HIF-2α can synergize with different protooncogenes, such as Akt and c-Myc, to facilitate tumor cell survival and growth (Bedogni et al. 2005; Gordan et al. 2007).

However, it is worth noting that, under certain circumstances, increased HIF expression or activity seems to have a negative effect on tumor growth. Teratomas derived from HIF-1α-deficient murine embryonic stem (ES) cells grow faster than those from the wild-type ES cells, in part because apoptosis occurs less often in HIF-1α-deficient, ES-derived tumors (Carmeliet et al. 1998). On the other hand, overexpression of HIF-2α in rat glioma tumors increases tumor cell apoptosis and reduces the growth of these tumors, despite enhanced angiogenesis (Acker et al. 2005). These inconsistent observations suggest that the dichotomous functions of HIFs may depend on interactions between the HIF and other pathways in different tumor cell types, their microenvironments, or both.

Several lines of evidence have demonstrated that HIF activation is associated with an undifferentiated phenotype. In primary pancreatic cancers, nuclear accumulation of the HIF-1α protein is primarily found in poorly differentiated tumor cells (Couvelard et al. 2005). Increased levels of HIF-1α and HIF-2α have been found in the stem cell–like populations of neuroblastomas (Pietras et al. 2008, 2009) and gliomas (Li et al. 2009). Downregulation of HIF-1α or HIF-2α by RNA interference results in reduced growth of the tumor sphere, an in vitro assay of self-renewal, and survival of glioma stem cells (Li et al. 2009). It is interesting to note that HIF-2α expression can be easily detected in glioma stem cells under hypoxic conditions (Li et al. 2009). In a similar way, HIF-2α is also preferentially expressed in immature neural crest-like neuroblastoma cells in vivo and seems to be required for maintenance of the undifferentiated neuroblastoma cells (Pietras et al. 2008, 2009). These data strongly support a role of HIF-1 and/or HIF-2 in the maintenance of cancer stem cells. These studies also point out the functional differences between HIF-1 and HIF-2 in the maintenance of undifferentiated cancer stem cell phenotypes.

2.2.2 Hypoxia-Inducible Factors and Stem Cell Gene Expression

Studies have shown that the HIF pathway is involved in upregulating the expression of several stem cell genes. The pluripotency gene POU5F1 (Oct3/4) is one of the four or five critical genes that collectively transform adult somatic cells into pluripotent stem cells (Meissner et al. 2007; Takahashi et al. 2007; Yu et al. 2007). In transgenic mice with doxycycline-inducible expression of POU5F1, induced POU5F1 expression results in inhibition of cellular differentiation and dysplastic growths in epithelial tissues (Hochedlinger et al. 2005), thus demonstrating a direct role of POU5F1 in tumorigenesis. Consistent with this notion, it has been found that germ cell cancers and several types of somatic cancers – including human cervical carcinomas, breast carcinomas, and pancreatic cancers – express elevated levels of POU5F1 (Cheng 2004; Gidekel et al. 2003; Jones et al. 2004; Tai et al. 2005).

Using a genetic “knock-in” mouse model, Covello et al. (2006) replaced the endogenous Hif1a gene locus with the Hif2a locus. The increased Hif2a gene dosage and the absence of Hif1a resulted in increased expression of HIF-2α-specific genes including POU5F1 in mouse embryonic tissues (Covello et al. 2006). HIF-2α, but not HIF-1α, directly binds to the POU5F1 promoter/enhancer. Loss of HIF-2α reduces the number of embryonic primordial germ cells that require POU5F1 for survival and maintenance. Furthermore, the loss of POU5F1 results in decreased growth of mouse ES cell–derived teratomas (Covello et al. 2006). Reduced HIF-2α expression similarly results in decreased expression of POU5F1 and other stem cell genes in human ES cells cultured at 5 % O₂ (Forristal et al. 2010). These observations strongly suggest that HIF-2α plays a significant role in stem cell maintenance. It will be interesting to see whether hypoxia increases POU5F1 expression in common types of tumors.

Delta-like 1 homolog (Drosophila), or DLK1, is a type I transmembrane protein with abundant expression in embryonic tissues and immature cells, but not in differentiated adult tissues (Floridon et al. 2000), suggesting a role for DLK1 in the regulation of stem cells and progenitor cells. Elevated expression of DLK1 has been reported in several tumor types (Jensen et al. 1994; Tornehave et al. 1996; Yin et al. 2006; Sakajiri et al. 2005; Van Limpt et al. 2003; Li et al. 2005). Studies have shown that DLK1 is robustly expressed in undifferentiated, but not differentiated, neuroblastoma cells (Begum et al. 2012; Kim et al. 2009). Downregulation of DLK1 by RNA interference sensitizes neuroblastoma cells to spontaneous neuronal differentiation, decreases clonogenicity or colony-forming potential, and suppresses tumorigenicity (Begum et al. 2012; Kim et al. 2009). Overexpression of DLK1, on the other hand, inhibits differentiation, enhances clonogenicity, and increases tumorigenicity (Kim et al. 2009). The DLK1 cytoplasmic domain, especially tyrosine-339 and serine-355, is required for maintaining both clonogenicity and tumorigenicity (Kim et al. 2009). The HIF pathway directly regulates DLK1 transcription as both HIF-1α and HIF-2α can bind to the HRE in the upstream DLK1 promoter/enhancer region under hypoxic conditions (Kim et al. 2009). In neuroblastoma xenografts, the DLK1-positve neuroblastoma cells seem to be preferentially localized in the pimonidazole-positive hypoxic region (Begum et al. 2012). These observations demonstrate that the HIF-DLK1 pathway has the potential to maintain cancer stem cells in the hypoxic tumor microenvironment.

The pentaspan transmembrane glycoprotein prominin-1 (CD133), a widely used marker for isolating perspective cancer stem cells from a variety of tumors (Visvader and Lindeman 2008), experiences increased expression in hypoxia -treated (1 % O₂) human glioma cells and can promote the expansion of the CD133 ⁺ tumor cell population (Griguer et al. 2008; Seidel et al. 2010; Soeda et al. 2009). Both HIF-1α and HIF-2α seem to be involved in the hypoxia-dependent induction of CD133 expression because knocking down either HIF-1α (Soeda et al. 2009) or HIF-2α (Seidel et al. 2010) reduces the hypoxia-induced CD133 expression in glioma cells. However, it remains to be determined how HIF enhances CD133 transcription. On the other hand, severe hypoxia (0.1 % O₂) seems to downregulate CD133 expression in several gastric, colorectal, and lung cancer cell lines (Matsumoto et al. 2009). These seemingly contradictory findings nonetheless suggest that CD133 may be more involved in cancer stem cell maintenance under moderate (1 % O₂) rather than severe (0.1 % O₂) hypoxia. Investigation of the transcriptional regulation of CD133 expression by HIF at different pO₂ levels may provide mechanistic insights into the O₂ concentration–dependent regulation of CD133 expression.

The CD44⁺/CD24^−/low signature has been used to identify breast cancer stem cells (Al-Hajj et al. 2003). As shown by global gene expression and genetic profiles, CD24 ⁺ and CD44 ⁺ breast cancer cells from the same tumor are clonally related but genetically different (Shipitsin et al. 2007). Elevated CD24 levels have been found to significantly – but counterintuitively – correlate with advanced disease stages in several types of human epithelial cancers, including breast cancer, ovarian cancer, and prostate cancer (Kristiansen et al. 2004). Large-scale immunohistochemical analyses of CD24 and CD44 protein levels in human breast cancer tumor samples have found that the combined CD44 ⁺/CD24⁻ phenotype is associated with the most favorable prognosis, whereas the CD44 ⁻/CD24⁺ phenotype predicts the worst outcome (Mylona et al. 2008 ; Ahmed et al. 2012). In addition, CD24 ⁺ tumor-initiating populations also have been found in pancreatic cancers (Ishizawa et al. 2010; Li et al. 2007a), liver cancers (Lee et al. 2011), and colorectal cancers (Vermeulen et al. 2008; Ke et al. 2012). An interesting recent study has shown that CD24 expression is strongly induced by hypoxia in a human bladder cancer cell line (Thomas et al. 2012). Promoter analysis has demonstrated that an HRE in the upstream promoter/enhancer region is required for both hypoxia-induced and HIF-1α-dependent CD24 expression (Thomas et al. 2012). Combined HIF-1α ⁺ and CD24 ⁺ immunostaining in a cohort of 101 human urothelial cancer samples showed a statistically significant association with reduced overall survival (Thomas et al. 2012). These data suggest that HIF and/or hypoxia may play an important role in the clonal maintenance or evolution of the aggressive CD24 ⁺ tumor stem populations in the tumor microenvironment.

2.2.3 Other Hypoxia-Regulated Genes and Cancer Stemness

Structures of chromosomes dynamically change during DNA replication and gene transcription and are accompanied by posttranslational modifications of histones, including acetylation of lysine residues and methylation of lysine or arginine residues. Histone demethylases are members of the JmjC domain-containing 2-oxoglutarate oxygenases and catalyze the removal of Nε-methyl groups from lysine residues via O₂-dependent hydroxylation (Loenarz and Schofield 2011). They play an important role in both normal embryonal development and cancer (Yamane et al. 2007; Klose et al. 2007; Lan et al. 2007; Iwase et al. 2007). Using the histone 3 lysine 4 (H3K4) demethylase JARID1B (KDM5B/PLU-1/RBP2-H1) as a biomarker, a small subpopulation of slow-cycling melanoma cells that are essential for continuous tumor growth has been identified in patients with advanced tumors (Roesch et al. 2008, 2010). It is interesting that JARID1B expression in melanoma cells increases rapidly under hypoxia (1 % pO₂) and gradually returns to normal levels after extended culture under atmospheric conditions (Roesch et al. 2010). However, it is not yet clear how JARID1B expression and its enzymatic activity are regulated under hypoxic conditions. Nonetheless, because melanoma cells can easily transition between JARID1B⁺ and JARID1B⁻ states, these data suggest that the hypoxic microenvironment may play a significant role in maintaining a population of melanoma cells with long-term repopulating potential, at least in part by augmenting JARID1B expression.

Krieg et al. (2010) also have reported that histone demethylase genes JMJD1A, JMJD2B, and JARID1B are induced by hypoxia in human RCCs. Interestingly, their hypoxia-dependent expression is abolished in HIF-1α-knockout mouse embryonic fibroblasts, suggesting that HIF-1 is necessary for hypoxic induction. Furthermore, downregulation of JMJD1A reduces xenograft tumor growth in vivo (Krieg et al. 2010). These data indicate that hypoxia can facilitate tumor growth via histone demethylase–mediated chromatin remodeling.

The histone methyltransferase mixed-lineage leukemia 1 (MLL1), also known as human trithorax or acute lymphocytic leukemia-1, is a member of the trithorax family of global transcription activators. MLL1 is preferentially expressed in glioma stem cells and is necessary for maintaining their self-renewal (Heddleston et al. 2012). Hypoxia significantly increases MLL1 expression in both stem and non–stem cell populations. Both HIF-1α and HIF-2α seem to be involved in the regulation of MLL1 expression (Heddleston et al. 2012), although the mechanism of regulation remains to be determined. It is interesting to note that inhibition of MLL1 expression decreases the expression of HIF-2α as well as that of hypoxia-induced genes (Heddleston et al. 2012). These data suggest a positive feedback between HIF-2α and MLL1 for the induction and maintenance of glioma stem cells.

2.3 Summary

As discussed earlier, hypoxia clearly has the potential to exert significant effect on the maintenance and evolution of cancer stem cells via both HIF-dependent transcriptions and chromatin remodeling in cancer cells (Fig. 2.1). Hypoxia also inhibits differentiation of mesenchymal stem/progenitor cells (Lin et al. 2006, 2008, 2009; Yun et al. 2002, 2005), thus creating a niche wherein cancer stem cells could be arrested in an undifferentiated state via interactions with their surrounding immature stromal cells (Lin and Yun 2010). However, it is worth noting that because of the plasticity of stemness and the expression of a heterogeneous array of stem cell markers (Magee et al. 2012 ; Shipitsin et al. 2007 ; Visvader and Lindeman 2012), cancer stem cells that are localized in or emerge from a hypoxic microenvironment may exist in a different stem cell state or express different sets of stem cell markers compared to developmentally similar cancer stem cells localized in nonhypoxic regions. Nonetheless, the hypoxia-stemness paradigm offers a new perspective on the role of hypoxia in facilitating malignant progression and therapy resistance. Since hypoxic regions are heterogeneously located throughout the tumor proper (Horsman et al. 2012), it is highly probable that hypoxic cancer stem cell niches may contribute to the microenvironment-specific emergence of metastatic clones (Yachida et al. 2010). Therefore, targeting the hypoxic cancer stem cell niche would be highly effective for controlling tumor growth, as well as for preventing metastasis.

Fig. 2.1.

Hypoxia-Activated Pathways Leading to Cancer Stem Cell Maintenance. Multiple stem cell-related genes encoding cell surface proteins, transcription factors, or chromatin modifying enzymes are up-regulated under hypoxic conditions either directly by the HIF transcription factor pathway or by other yet unknown mechanisms. These different pathways may function either synergistically or additively to maintain cancer stem cells by enhancing their self-renewal and blocking their differentiation. Increased lifespan of cancer stem cells allows inheritable accumulation of multiple genetic mutations and epigenetic changes that are crucial for clonal evolution and malignant progression.