HIF2α cooperates with RAS to promote lung tumorigenesis in mice

Members of the hypoxia-inducible factor (HIF) family of transcription factors regulate the cellular response to hypoxia. In non–small cell lung cancer (NSCLC), high HIF2α levels correlate with decreased overall survival, and inhibition of either the protein encoded by the canonical HIF target gene VEGF or VEGFR2 improves clinical outcomes. However, whether HIF2α is causal in imparting this poor prognosis is unknown. Here, we generated mice that conditionally express both a nondegradable variant of HIF2α and a mutant form of Kras (Kras^G12D) that induces lung tumors. Mice expressing both Hif2a and Kras^G12D in the lungs developed larger tumors and had an increased tumor burden and decreased survival compared with mice expressing only Kras^G12D. Additionally, tumors expressing both Kras^G12D and Hif2a were more invasive, demonstrated features of epithelial-mesenchymal transition (EMT), and exhibited increased angiogenesis associated with mobilization of circulating endothelial progenitor cells. These results implicate HIF2α causally in the pathogenesis of lung cancer in mice, demonstrate in vivo that HIF2α can promote expression of markers of EMT, and define HIF2α as a promoter of tumor growth and progression in a solid tumor other than renal cell carcinoma. They further suggest a possible causal relationship between HIF2α and prognosis in patients with NSCLC.

HIF is a heterodimeric transcription factor composed of a labile α subunit (HIFα) and a stable β subunit (HIFβ) (1). The human genome encodes 3 HIFα subunit genes (HIF1A, HIF2A, HIF3A) and 3 HIFB genes (1). Under normal oxygen conditions, HIFα subunits bind with the VHL gene product, pVHL, which acts as the substrate recognition component of an E3 ubiquitin ligase complex. pVHL, along with other members of the pVHL ubiquitin ligase complex, polyubiquitinates and destroys HIFα subunits in a proteasome-dependent manner. The binding of HIFα by pVHL is dependent upon the oxygen-dependent, posttranslational hydroxylation of HIF subunits on either of two conserved prolyl residues (1). This prolyl hydroxylation is catalyzed by the oxygen-dependent EGLN (also known as PHD) family of prolyl hydroxylases (2). Thus, HIFα subunits are normally degraded by pVHL in the presence of oxygen but stabilized under conditions of hypoxia or in the setting of pVHL inactivation. HIFα subunits have a very short half-life (measured in minutes during normoxia) and therefore are sensitive to changes in protein translation, such as those mediated by the mammalian target of rapamycin (mTOR) (3–5).

While HIF protein stability is modulated primarily by oxygen levels and the pVHL status of cells, HIF mRNA levels and protein synthesis are regulated primarily in an oxygen-independent manner. For example, activation of the PI3K and MAPK pathways via growth factor receptor tyrosine kinases (RTKs), tumor suppressor loss, or oncogene activation have all been shown to result in variable increases in both HIF mRNA and protein levels (1, 6–10). HIF accumulation leads to the direct activation of more than 100 target genes (1). While the majority of these genes are involved in the cellular and organismal adaptation to hypoxia, a substantial proportion of these genes are potentially capable of promoting tumor progression. For example, genes involved in angiogenesis (VEGF, PDGF, PIGF), degradation of the extracellular matrix (MMPs), metastasis (LOX, CXCR4, SDF1), epithelial to mesenchymal transition (SNAIL, SIP, ZEB), and immortalization (hTERT) are known HIF target genes (1, 11).

Despite sharing similar functional domains and substantial sequence homology, HIF1α and HIF2α do not appear to be fully redundant in function. First, while knockout of either Hif1a or Hif2a in mice leads to embryonic lethality, both the timing and cause of death appear to differ. Hif1a^–/– mice die in mid-gestation and exhibit severe blood vessel defects, whereas Hif2a^–/– mice exhibit late embryonic or early postnatal lethality as a consequence of catecholamine deficiency or lung maturation defects (12–14). Second, the functional differences between HIF1α and HIF2α do not appear to be limited to their roles in development, as a recent report also demonstrates that postnatal deletion of Hif1a or Hif2a results in differing phenotypes (15). Finally, studies from our laboratory and others have now demonstrated that HIF1α and HIF2α produce overlapping yet distinct gene expression profiles and that certain target genes are uniquely regulated by either HIF1α or HIF2α (16–20).

HIF’s role in tumorigenesis is best characterized in the context of renal cell carcinoma (RCC), where HIF2α appears to play a special role (21). VHL^–/– renal carcinomas demonstrate obligate expression of HIF2α, but not HIF1α, suggesting a selective pressure to maintain HIF2α expression (22, 23). An apparent switch from HIF1α to HIF2α expression occurs in preneoplastic lesions arising in human VHL^+/– kidneys in association with increasing dysplasia and cellular atypia (24). Finally, in VHL^–/– RCCs, HIF2α, rather than HIF1α, appears to be necessary and sufficient for tumor growth (25–28). Thus, at least in VHL-deficient RCC, HIF2α is the relevant oncogenic HIF family member (29).

Evidence also suggests that hypoxia, HIF, and in particular HIF2α may play an important role in the pathogenesis and prognosis of lung cancer. Prior work has indicated that intratumoral hypoxia is clearly present, even in some early-stage lung cancers, and correlates with a decreased overall survival (30, 31). Furthermore, HIF1A mRNA is one of 3 transcripts that have been validated to stratify early-stage non–small cell lung cancer (NSCLC) patients into groups with different prognoses (32). Interestingly, while multiple studies have shown that HIF1α is overexpressed at the protein level in a substantial proportion (33%–62%) of NSCLCs, patients with elevated HIF1α protein levels do not have a shortened overall survival (33, 34).

HIF2α, like HIF1α, is frequently overexpressed at the protein level in NSCLC (50%) (33). However, overexpression of HIF2α, in contrast to HIF1α, correlates with decreased survival, suggesting that HIF2α either causes, or is a surrogate marker for, poor clinical outcome (33). Finally, it is notable that bevacizumab, the monoclonal antibody targeting VEGF (a canonical HIF target gene) and sunitinib, the small molecule tyrosine kinase inhibitor that targets VEGFR2 (among other tyrosine kinases), have shown clinical benefit for patients with metastatic NSCLC (35, 36).

To evaluate the role of HIF2α in lung tumorigenesis, we generated cohorts of mice that conditionally expressed mutant Kras^G12D with or without a conditionally expressed variant of stabilized HIF2α (HIF2dPA). Our data reveal that lung-specific expression of HIF2dPA on a background of mutant Kras leads to the development of lung cancers with highly pleomorphic histology and features of epithelial-mesenchymal transition (EMT), increased vascularization and blood flow, as well as elevated levels of circulating endothelial cells (CECs) and circulating endothelial progenitor cells (CEPs). Furthermore, mice harboring HIF2dPA-expressing lung cancers exhibited an increased tumor burden, more invasive tumors, and a shortened median survival. These genetic results complement existing human clinical data to implicate HIF2α as a negative prognostic factor in human NSCLC and demonstrate that HIF2α can promote tumor growth in tumors other than renal cell carcinoma and specifically in Kras mutant lung carcinoma.

Expression of HIF2dPA decreases survival of mice with Kras-induced lung tumors. We previously generated knock-in mice that expressed a conditionally activatable form of HA-tagged HIF2α (HIF2dPA) that escapes recognition by pVHL by virtue of proline-to-alanine substitutions (hereafter called LSL-HIF2 for simplicity) (16). Specifically, the ROSA26 locus of these mice harbors a Lox-Stop-Lox cassette interposed between the endogenous ROSA26 promoter and the HIF2dPA cDNA, and the levels of HIF2α expressed are relatively physiologic (16). Therefore, transcription is under control of the ROSA26 promoter but dependent upon Cre-mediated excision of the floxed stop element.

Adenovirus encoding Cre recombinase (Ad-Cre) can be delivered to the murine lung by intranasal administration. Mice harboring a conditionally activatable allele of mutant Kras^G12D, Lox-Stop-Lox Kras^G12D (hereafter called LSL-Kras) develop multifocal tumors with a relatively long latency and low invasive potential (37–40). To determine the role of HIF2α in lung tumorigenesis, we intercrossed the LSL-HIF2 mice to LSL-Kras mice to generate cohorts of LSL-Kras;LSL-HIF2 and littermate control LSL-Kras mice. These cohorts were then treated with a low dose of intranasal Ad-Cre (2.5 × 10⁶ PFU) (38). LSL-Kras mice treated with intranasal Ad-Cre developed multiple tumors with a moderate multiplicity and had a median survival of 57 weeks (Figure 1A and Table 1). LSL-Kras;LSL-HIF2 mice also developed lung tumors with approximately the same multiplicity (P = 0.69; Table 1) but with a shortened median survival of 42 weeks (P = 0.036; Figure 1A and Table 1). Therefore, the data indicate that expression of HIF2α within Kras-induced lung carcinomas does not increase tumor multiplicity but confers a shortened survival.

Figure 1

Mice harboring HIF2α-producing lung tumors have a shortened survival and larger tumors. (A) Kaplan-Meier survival curve of cohorts of LSL-Kras (n = 9) and LSL-Kras;LSL-HIF2 (n = 12) mice (P = 0.036). (B) Photomicrographs of representative lung tumors from LSL-Kras and LSL-Kras;LSL-HIF2 mice. Arrows indicate visible subpleural tumors. (C) Quantification of tumor area as a percentage of lung area. Error bars indicate 1 SEM (P = 0.031).

Table 1

Comparison of lung cancer cohorts

HIF2α-producing Kras-induced lung tumors are larger and result in an increased lung tumor burden. Gross examination of the visceral pleura of inflated lungs taken from 40-week-old LSL-Kras mice showed numerous tumors up to 3 mm in size. Age-matched LSL-Kras;LSL-HIF2 mice harbored significantly larger tumors, which occasionally reached 10 mm in diameter (P < 0.001; Figure 1B and Table 1). These data show that expression of HIF2α promotes the formation of larger tumors.

To determine whether the shortened survival of LSL-Kras;LSL-HIF2 mice resulted from an increase in lung tumor burden, we compared the total tumor area in LSL-Kras and LSL-Kras;LSL-HIF2 mice. To this end, we quantified low-power magnifications of a random section chosen from the right upper (apical) lobe (RUL) of mice for tumor area as well as total lung area. These measurements were used to determine the percentage of the RUL affected by tumor. At 40 weeks, LSL-Kras;LSL-HIF2 mice had a statistically significantly increased tumor burden relative to LSL-Kras mice (P = 0.031; Figure 1C). Despite these differences in tumor involvement, there was no significant difference in either proliferation, as measured by Ki67 staining (P = 0.45; Supplemental Figure 1A; supplemental material available online with this article; doi: 10.1172/JCI38443DS1), or apoptosis, as measured by TUNEL staining (P = 0.85; Supplemental Figure 1B), between the tumors of LSL-Kras and LSL-Kras;LSL-HIF2 mice. This may reflect that fact that these measurements were taken at a late time point in the growth of these tumors. Alternatively, it is also possible that the increased tumor burden found in the LSL-Kras;LSL-HIF2 mice is a reflection of alterations in tumor cell size, tumor edema or vascularity, or other increases in the non-neoplastic population of tumor-associated stromal cells (such as seen in Figure 2B). These data suggest that expression of HIF2dPA in Kras-induced lung carcinomas results in larger tumors and increased total lung tumor burden.

Figure 2

HIF2α-producing tumors are pleomorphic and invasive. (A) Representative low- and high-power photomicrographs of H&E-stained lung tumors of 40-week-old LSL-Kras and LSL-Kras;LSL-HIF2 mice. (B–G) H&E staining showing increased fibrous stroma (B, asterisk) as well as necrosis (C, asterisk) in a lung tumor from a 40-week-old LSL-Kras;LSL-HIF2 mouse. Tumors from 40-week-old LSL-Kras;LSL-HIF2 mice also showed features of invasiveness, such as invasion through the visceral pleura (D), myocardial invasion (E), direct invasion of mediastinal fat (F), and mediastinal lymph node metastases (G). Original magnification, ×100 (B, C, F, and G); ×40 (D and E). The dashed yellow line indicates the division between normal lymph node and tumor; yellow asterisk indicates tumor.

Kras-induced lung tumors expressing HIF2dPA have distinct gene expression changes. Numerous studies have used gene expression profiling to assess the transcriptional effects of VHL loss or HIF overexpression on cell lines (17, 19, 41, 42). We had previously examined the gene expression changes induced by Vhl loss or expression of stabilized forms of HIF1α and HIF2α within hepatocytes of mice (16). To validate that expression of stabilized HIF2α resulted in significant changes in gene expression, we extracted total mRNA from 5 macrodissected tumors of both LSL-Kras and LSL-Kras;LSL-HIF2 mice and performed gene expression analysis using high-density oligonucleotide arrays. The gene expression patterns of LSL-Kras and LSL-Kras;LSL-HIF2 tumors were then examined by unsupervised hierarchical gene clustering. The transcriptional effects of HIF2α were readily demonstrable by this approach; unsupervised analysis showed that the tumors clustered into two separate groups based on the presence or absence of HIF2dPA expression (Supplemental Figure 2) and were notable for the presence of several canonical HIF target genes such as Vegf, Tgfb1, Egln3, and Cxcl12 (Supplemental Figure 2). Therefore, our data indicate that expression of HIF2dPA within lung tumors results in activation of canonical HIF target genes and significant changes in gene expression.

HIF2α-producing lung tumors are poorly differentiated, with focal squamous differentiation, and display increased invasiveness. Histological examination of tumors from 40-week-old LSL-Kras mice, as expected, showed adenomatous alveolar hyperplasia and moderately to well-differentiated adenocarcinomas with bronchoalveolar carcinoma (BAC), acinar, and papillary features (Figure 2A), as described by others (37, 38, 43). Lungs from age-matched LSL-Kras;LSL-HIF2 mice harbored two distinct subsets of tumors. Tumors of the first subset were reminiscent of those found in LSL-Kras mice. However, tumors of the second subset were poorly differentiated and displayed more cytologic atypia with pleomorphic histology (Figure 2A and Table 1) as well as increased fibrous stroma (Figure 2B and Table 1); occasional necrosis (Figure 2C); rare invasion of local structures (Table 1) such as the pleura (Figure 2D), myocardium (Figure 2E), and mediastinal adipose tissue (Figure 2F); or metastases to regional lymph nodes (Figure 2G).

Poorly differentiated tumors were statistically significantly more frequent in LSL-Kras;LSL-HIF2 mice than their LSL-Kras counterparts (P = 0.027; Table 1). Examination of these tumors revealed occasional areas consistent with squamous differentiation. These cells had abundant eosinophilic cytoplasm, more centrally placed nuclei with prominent nucleoli, and less apparent polarity, features suggestive of squamous differentiation (Figure 3A). In addition, the areas that appeared morphologically squamous had high nuclear expression of the squamous cell marker p63 according to immunohistochemistry (Figure 3B), in a background of p63-negative adenocarcinoma cells (43).

Figure 3

Squamous differentiation of HIF2α-producing lung tumors. (A) An H&E-stained lung tumor from a 40-week-old LSL-Kras;LSL-HIF2 mouse shows squamous features, including abundant eosinophilic cytoplasm, centrally placed nuclei with prominent nucleoli, and apparent loss of polarity (asterisk). (B) High-power photomicrograph of high nuclear p63 expression in a background of p63-negative adenocarcinoma as detected by immunohistochemistry of a LSL-Kras;LSL-HIF2 tumor. Original magnification, ×400.

The presence of squamous differentiation was interesting, as Kras-induced lung tumors are uniformly adenocarcinomas (37–40, 43). Even concurrent deletion of p53 or Pten, while promoting the development of higher-grade tumors, does not alter their histologic spectrum (44, 45). In contrast, we had previously described that inactivation of the Lkb1 tumor suppressor gene within Kras-induced lung tumors results in an expanded histologic spectrum including adeno-, squamous, and large cell carcinomas (43). We therefore queried the DNA microarray dataset generated on these Lkb1-deficient mouse lung tumors to see whether Hif2a gene expression correlated with histologic subtype. In keeping with the notion that HIF2α promotes squamous differentiation, we found that Kras-induced, Lkb1-deficient squamous cell or mixed adeno-squamous tumors had significantly higher levels of HIF2α gene expression relative to Kras-induced, Lkb1-deficient adenocarcinomas (Supplemental Figure 3, A and B). Therefore, the data indicate that HIF2α gene expression is upregulated in mouse squamous cell lung carcinomas and that production of HIF2α in Kras-induced lung tumors promotes squamous differentiation but not frank squamous cell carcinoma.

Kras-induced lung tumors expressing HIF2dPA have increased vascularity and blood flow. We next compared the vascularity of the autochthonous lung tumors by CD34 staining and quantification of microvessel density (MVD). Lungs from Ad-Cre–treated LSL-Kras and LSL-Kras;LSL-HIF2 mice were harvested, formalin fixed, and stained with antibodies against the endothelial cell marker CD34. The number of microvessels per high-power field was quantified by a reviewer blinded to the research protocol. As expected, the lung tumors from LSL-Kras;LSL-HIF2 mice displayed a significantly increased number of microvessels per high-power field relative to lung tumors from age-matched, littermate LSL-Kras control mice (P < 0.001; Figure 4, A and B).

Figure 4

HIF2α-expressing tumors have increased vascularity and blood flow. (A) Quantification of microvessels per high-power field. Error bars indicate 1 SEM. P < 0.001. (B) Immunohistochemistry for CD34 from lung tumors of LSL-Kras and LSL-Kras;LSL-HIF2 mice. Original magnification, ×100. (C) Quantification of percent vascularity of subcutaneously growing LSL-Kras and LSL-Kras;LSL-HIF2 tumors. Lines indicate mean (P = 0.028). (D and E) Representative pseudocolored Doppler ultrasound images of subcutaneously growing LSL-Kras and LSL-Kras;LSL-HIF2 tumors (D) and exponential decay curves of percent vascularity in relation to tumor volume (E). LSL-Kras: r² = 0.87; LSL-Kras;LSL-HIF2: r² = 0.21.

Increases in MVD do not necessarily translate into augmentation of blood flow, in part because of alterations in vascular permeability and heterogeneous blood flow within tumors (46). We and others have previously shown that the vessels formed by expression of stabilized forms of HIF in the skin are not conspicuously hyperpermeable (16, 47). However, these studies did not directly assess blood flow or the effects of HIF on tumor-associated blood vessels.

We asked whether the increase in vascular density seen by CD34 staining of HIF2α-producing lung tumors correlated with actual increases in blood flow. Because ultrasound waves are disrupted by pulmonary air, we were not able to image the primary tumors in situ in the lung, and therefore single-cell suspensions of primary tumors from LSL-Kras and LSL-Kras;LSL-HIF2 mice were injected subcutaneously into the flanks of nude mice. These tumors were then serially imaged with ultrasound and power Doppler sonography. Regions of blood flow were determined by power Doppler, while tumor volume was determined by 3D reconstruction of ultrasound images. The percent vascularity was determined as the ratio of blood flow to tumor volume. When matched for tumor volume, LSL-Kras;LSL-HIF2 tumors had a statistically significantly higher percent vascularity than LSL-Kras tumors (P = 0.028; Figure 4, C and D). Additionally, while the percent vascularity of LSL-Kras tumors appeared to decline as the tumors enlarged (fitted exponential decay curve, r² = 0.87) the LSL-Kras;LSL-HIF2 tumors were more vascular at baseline, and this high percent vascularity persisted despite tumor growth (fitted exponential decay curve, r² = 0.21) (Figure 4E). Therefore, the increased MVD seen in tumors expressing stabilized HIF2α appears to be paralleled by an increase in blood flow in tumor explants derived from these tumors.

Mice harboring HIF2α-producing lung tumors have elevated levels of CECs and CEPs. Increased numbers of CECs and CEPs have been detected in the peripheral blood of patients with solid tumors, including those with lung cancer (48–52). Several chemokines and cytokines known to be upregulated by HIF, such as VEGF and CXCL12, have been shown to be involved in the mobilization and homing of CECs and CEPs (50, 53–55). We therefore asked whether production of stabilized HIF2α within lung tumors could stimulate the numbers of CECs and CEPs. To this end, we enumerated the numbers of CECs and CEPs by flow cytometry in the peripheral blood of LSL-Kras and LSL-Kras;LSL-HIF2 mice, 12 weeks after treatment with Ad-Cre. CECs were defined as CD45^–CD117^–CD31⁺VEGFR2⁺, while CEPs were defined as CD45^dimCD117⁺CD31⁺VEGFR2⁺. The lungs from these mice grossly appeared to have approximately the same tumor burden. Interestingly, the number of CECs and CEPs present in the peripheral blood of LSL-Kras;LSL-HIF2 mice was statistically significantly increased relative to that of LSL-Kras mice (P = 0.027 and P = 0.014, respectively) (Figure 5, A and B). This increase in CEPs, which are thought to be bone marrow derived, suggest that expression of stabilized HIF2α within tumors promotes the mobilization of CEPs in an endocrine fashion.

Figure 5

Mice harboring HIF2α-producing tumors have increased numbers of CECs and CEPs. Quantification of CECs (A) and CEPs (B) in the peripheral blood by flow cytometry expressed as a percentage of viable mononuclear cells. Error bars indicate 1 SEM. A: P = 0.027; B: P = 0.014.

HIF2α-producing tumors express markers of EMT in vivo. To better understand how HIF2α activation within Kras-induced lung tumors might impart the phenotypes described, we performed gene set enrichment analysis (GSEA) using GSEA 2.0 (http://www.broad.mit.edu/gsea) on our dataset of mouse lung tumors. Specifically, an established collection of more than 100 cancer-related, curated gene sets provided by the Molecular Signatures Database (MSigDB, broad.mit.edu/gsea/msigdb) as well as those collected from the literature (56–59) were used to interrogate the gene expression dataset in order to compare gene set enrichment in LSL-Kras and LSL-Kras;LSL-HIF2 tumors (Supplemental Table 2). We found 14 gene signatures to be enriched in HIF2α-producing tumors relative to their mutant Kras controls using a false discovery rate (FDR) calculation (P < 0.05) to account for multiple hypothesis testing (Supplemental Figure 4). Notably, a significant number of the enriched gene sets were classifiers for either EMT or pathways known to induce EMT, such as the WNT/β-catenin and TGF-β pathways (60).

HIF has been implicated in promoting EMT (1, 61–64). Given our GSEA results and the invasive and metastatic nature of tumors derived from LSL-Kras;LSL-HIF2 mice, we hypothesized that they may express markers associated with EMT. We first asked whether these tumors had increased expression of genes implicated in inducing EMT by directly querying our gene expression dataset of LSL-Kras and LSL-Kras;LSL-HIF2 tumors. Interestingly, tumors expressing HIF2dPA had increased levels of several mRNA transcripts implicated causally or associated with EMT including Sip, Snail, Zeb1, and vimentin (all P < 0.01) (Figure 6A). In addition, LSL-Kras;LSL-HIF2 tumors also stained positively for vimentin by immunohistochemistry (Figure 6B). These data are consistent with the notion that HIF2α can induce a program of genes and markers involved in EMT in mouse lung carcinomas and suggest a possible reason why HIF2α-producing tumors are more invasive.

Figure 6

HIF2α-producing tumors display markers of EMT. (A) Fold change in gene expression of markers or inducers of EMT in LSL-Kras;LSL-HIF2 tumors relative to LSL-Kras tumors (Sip1, Snail, Zeb1, vimentin, and Twist). All P < 0.01. Error bars indicate 1 SEM. (B) Expression of vimentin by immunohistochemistry in LSL-Kras and LSL-Kras;LSL-HIF2 tumors.

HIF2α activation and EMT gene signatures are associated in human lung cancer. We asked whether the association between HIF2α and EMT that we saw in our autochthonous murine lung tumors was present in human lung cancers. To that end, we first derived an HIF2α gene signature from our mouse lung tumors. The gene expression profiles of the LSL-Kras and LSL-Kras;LSL-HIF2 tumors were compared using 2-class unpaired significance analysis of microarrays (SAM). The lowest FDR allowing the selection of 200 upregulated genes was chosen. Using an FDR of 0.71%, we were able to define 396 significantly upregulated and 171 significantly downregulated genes. Human orthologs of the top 200 upregulated genes were determined and used to define an HIF2α gene signature (Supplemental Table 1).

We next queried publicly available gene expression datasets of human lung cancer cell lines (65) or primary NSCLCs (66) for the presence or absence of our HIF2α gene signature as well as the EMT, β-catenin, and TGF-β gene signatures (67–69). To determine whether an association was present between HIF2α activation and upregulation of gene signatures of EMT, β-catenin, and TGF-β, Genomica software (genomica.weizmann.ac.il) was used to calculate HIF2α, EMT, β-catenin, and TGF-β enrichment scores for each tumor or cell line. The association between each of the gene signatures was then determined by calculating a Pearson correlation coefficient. Our HIF2α gene signature correlated with enrichment of EMT, β-catenin, and TGF-β gene signatures in both the lung cancer cell lines and the primary human NSCLCs (all P < 0.001; Figure 7, A and B). Therefore, our HIF2α gene signature appears to be applicable across species and is associated with EMT gene expression changes in human lung cancers.

Figure 7

Association between HIF2α activation and EMT gene signatures in human lung cancer. (A) Heat maps of relative enrichment scores of gene signatures of HIF2α, EMT, β-catenin, and TGF-β in human lung cancer cell lines and primary human NSCLC. Tumors or cell lines are represented in columns. Gene signatures are represented in rows. Red indicates high enrichment score. Green indicates low enrichment score. (B) Graphical representation of correlation between EMT, β-catenin, and TGF-β enrichment scores and HIF2α enrichment score. Dots represent a single cell line or tumor.

We found that production of stabilized HIF2α in Kras-induced lung carcinomas resulted in larger, more invasive lung tumors and was associated with a shortened survival of mice harboring these tumors. Additionally, production of stabilized HIF2α led to the development of tumors with a poorly differentiated and pleomorphic histology that expressed markers of EMT. Consistent with HIF’s role in promoting angiogenesis, the HIF2α-producing lung tumors were found to have increased vascularity, and importantly, these tumors, when explanted into nude mice, were found to have increased blood flow at all stages of tumor development. Moreover, we detected higher levels of CECs and CEPs in the peripheral blood of mice harboring HIF2α-producing lung cancers, suggesting that tumors that produce HIF2α can mobilize these cells from the tumor endothelium and bone marrow. Collectively, these results build upon the clinical data implicating HIF2α as a negative prognostic factor in human NSCLC and demonstrate in an autochthonous cancer model that HIF2α can promote tumor progression in cancers other than renal cell carcinoma.

The majority of studies reinforce the notion that HIF is a positive regulator of tumor growth. For example, cell lines having reduced HIF transcriptional activity secondary to knockout of Hif1a or deficiency of Arnt grow more slowly under the skin of immunocompromised mice than wild-type controls (70–72). In genetically engineered mouse models, transgenic expression of Hif1a in basal keratinocytes facilitates the growth of HPV-16–induced cervical cancer, and conditional inactivation of Hif1a in a mouse mammary tumor model restrains the growth of polyoma middle T–induced breast tumors as well as pulmonary metastases (73, 74). However, other studies suggest that HIF may have a tumor suppressor role though induction of tumor cell apoptosis or cell-cycle arrest (75–78). Moreover, HIF can play starkly contrasting roles in tumor progression depending on the microenvironment in which the tumor cells are implanted, underscoring the importance of studying HIF in autochthonous tumor models (77). Our studies in the autochthonous lung cancers of genetically engineered mice support the notion that HIF2α promotes solid tumor growth and aggressiveness.

In the setting of renal cell carcinoma, VHL loss and consequent upregulation of HIF is an early event (24). However, pVHL inactivation or HIF activation alone does not appear to be sufficient for the formation of tumors in humans or mice (16, 24, 47, 79–82). Similarly, in our studies, we did not see tumor formation or significant histologic changes upon activation of only HIF2α in the lungs of LSL-HIF2 mice (data not shown). Furthermore, the tumor multiplicity of lungs from LSL-Kras and LSL-Kras;LSL-HIF2 mice did not differ significantly, suggesting that even in the presence of an oncogenic mutation such as mutant Kras, HIF2α does not initiate tumor formation.

In humans, non–small cell lung carcinomas can be subdivided into adeno-, squamous, and large cell carcinomas based on their histologic appearance. Kras-induced lung cancers in mice are uniformly well- to moderately differentiated adenocarcinomas. While concurrent loss of p53 or Pten appears to promote the development of higher-grade tumors, inactivation of these genes does not alter their histologic spectrum (44, 45). In contrast, Lkb1 inactivation within Kras-induced lung tumors results in an expanded histologic spectrum including adeno-, squamous, and large cell carcinomas (43). We noted that Lkb1-deficient squamous or mixed adenosquamous cell carcinomas expressed higher levels of Hif2a than Lkb1-deficient adenocarcinomas. We also noted that focal areas of HIF2α-producing tumors appeared to have squamous differentiation and stained positively for the squamous cell marker p63. Thus, while HIF2α appears to modulate the histologic spectrum of lung cancers, the low penetrance of squamous differentiation seen in our studies suggests that HIF2α expression alone is not sufficient to induce frank squamous cell carcinomas. These data also suggest that some of the phenotypic and transcriptional effects of Lkb1 loss may also be secondary to increased expression of HIF2α.

The tumor microenvironment critically controls aspects of tumor growth and progression. In this regard, we present evidence that HIF2α can increase tumor vascularity and may also augment tumor blood flow, suggesting that the neoangiogenesis induced by HIF2α-producing tumors results in a functional vasculature, as has been seen previously in non-neoplastic settings (16, 47). While the assessment of vascularity was performed on the autochthonous murine lung tumors, for technical reasons we had to assess the blood flow of HIF2α-producing lung tumors as tumor explants in which tumor cells are grown and examined under the skin of immunocompromised mice. Therefore, whether these studies adequately reflect the effects of HIF2α within autochthonous lung tumors is uncertain.

While angiogenesis has been historically thought to be a local process, it has recently been appreciated that endothelium-derived endothelial cells (CECs) and their bone marrow–derived precursors (CEPs) can contribute to both physiologic and pathologic neoangiogenesis (83, 84). Interestingly, we detected increased numbers of both CECs and CEPs in the peripheral blood of mice harboring HIF2α-producing tumors relative to those expressing only mutant Kras. These results suggest that HIF2α production within tumors is able to mobilize CECs and CEPs from the mature endothelium or bone marrow. Although it is tempting to speculate that this results from an increased elaboration of proangiogenic factors in HIF2α-producing tumors, the precise mechanism whereby HIF2α production is signaled to the bone marrow and the importance of these cells to tumor neovascularization remains to be determined.

Finally, the histologic and gene expression changes seen in the HIF2dPA-expressing tumors suggest that HIF2α production on a backdrop of Kras activation results in more poorly differentiated tumors as well as induction of markers and mediators of EMT. These findings in autochthonous tumors extend a body of in vitro and xenograft data previously demonstrating that either HIF1α or HIF2α can induce EMT (61–64, 85). Moreover, cross-species analysis of our HIF2α gene signature shows that HIF2α activation correlates with gene signatures of EMT in human lung cancer, supporting the notion that HIF2α initiates or is part of a genetic program resulting in EMT in human NSCLC.

HIF is predicted to transactivate a program of target genes that favor the progression of tumors. We have shown that HIF2α activation results in the promotion of tumor growth but not tumor initiation and that HIF2α’s effects on tumor size and invasiveness correlate with increased angiogenesis and expression of markers of EMT. Our data establish a causal link between HIF2α and the shortened survival of patients, describe a mechanism whereby HIF2α negatively impacts the overall survival of patients with NSCLC, and validate in autochthonous tumors that HIF2α is a promoter of tumor growth.

View Supplemental data

We thank K. Rathmell and N. Bardeesy for useful discussions. This work was supported by the Cecily and Robert Harris Foundation (K.-K. Wong); the Joan Scarangello Foundation to Conquer Lung Cancer (K.-K. Wong and W.Y. Kim); the Flight Attendant Medical Research Institute (K.-K. Wong); the Melanoma Research Foundation (W.Y. Kim); the Department of Defense (W.Y. Kim); and NIH grants K08 AG024004 (to K.-K. Wong), R01 AG2400401 (to K.-K. Wong), R01 CA122794 (to K.-K. Wong), and K08 CA097203 (W.Y. Kim). We also acknowledge the assistance of the UNC Lineberger Genomics and Bioinformatics Core Facility.

Address correspondence to: William Y. Kim, M.D., Lineberger Comprehensive Cancer Center, University of North Carolina at Chapel Hill, 450 West Drive, CB# 7295, Chapel Hill, North Carolina 27599, USA. Phone: (919) 966-4765; Fax: (919) 966-8212; E-mail: wykim@med.unc.edu.

Conflict of interest: The authors have declared that no conflict of interest exists.

Nonstandard abbreviations used: Ad-Cre, adenovirus encoding Cre recombinase; CEC, circulating endothelial cell; CEP, circulating endothelial progenitor cell; EMT, epithelial-mesenchymal transition; FDR, false discovery rate; GSEA, gene set enrichment analysis; MVD, microvessel density; NSCLC, non–small cell lung cancer.

Reference information: J. Clin. Invest.119:2160–2170 (2009). doi:10.1172/JCI38443

1. Semenza, G.L. 2003. Targeting HIF-1 for cancer therapy. Nat. Rev. Cancer. 3:721-732.
   View this article via: CrossRef PubMed Google Scholar
2. Kaelin (Jr.), W.G., Ratcliffe, P.J. 2008. Oxygen sensing by metazoans: the central role of the HIF hydroxylase pathway. Mol. Cell. 30:393-402.
   View this article via: CrossRef PubMed Google Scholar
3. Hudson, C.C., et al. 2002. Regulation of hypoxia-inducible factor 1alpha expression and function by the mammalian target of rapamycin. Mol. Cell. Biol. 22:7004-7014.
   View this article via: CrossRef PubMed Google Scholar
4. Thomas, G.V., et al. 2006. Hypoxia-inducible factor determines sensitivity to inhibitors of mTOR in kidney cancer. Nat. Med. 12:122-127.
   View this article via: CrossRef PubMed Google Scholar
5. Brugarolas, J.B., Vazquez, F., Reddy, A., Sellers, W.R., Kaelin (Jr.), W.G. 2003. TSC2 regulates VEGF through mTOR-dependent and -independent pathways. Cancer Cell. 4:147-158.
   View this article via: CrossRef PubMed Google Scholar
6. Zhong, H., et al. 2000. Modulation of hypoxia-inducible factor 1alpha expression by the epidermal growth factor/phosphatidylinositol 3-kinase/PTEN/AKT/FRAP pathway in human prostate cancer cells: implications for tumor angiogenesis and therapeutics. Cancer Res. 60:1541-1545.
   View this article via: PubMed Google Scholar
7. Fukuda, R., Hirota, K., Fan, F., Jung, Y.D., Ellis, L.M., Semenza, G.L. 2002. Insulin-like growth factor 1 induces hypoxia-inducible factor 1-mediated vascular endothelial growth factor expression, which is dependent on MAP kinase and phosphatidylinositol 3-kinase signaling in colon cancer cells. J. Biol. Chem. 277:38205-38211.
   View this article via: CrossRef PubMed Google Scholar
8. Fukuda, R., Kelly, B., Semenza, G.L. 2003. Vascular endothelial growth factor gene expression in colon cancer cells exposed to prostaglandin E2 is mediated by hypoxia-inducible factor 1. Cancer Res. 63:2330-2334.
   View this article via: PubMed Google Scholar
9. Zundel, W., et al. 2000. Loss of PTEN facilitates HIF-1-mediated gene expression. Genes Dev. 14:391-396.
   View this article via: PubMed Google Scholar
10. Brugarolas, J., et al. 2004. Regulation of mTOR function in response to hypoxia by REDD1 and the TSC1/TSC2 tumor suppressor complex. Genes Dev. 18:2893-2904.
    View this article via: CrossRef PubMed Google Scholar
11. Bertout, J.A., Patel, S.A., Simon, M.C. 2008. The impact of O2 availability on human cancer. Nat. Rev. Cancer. 8:967-975.
    View this article via: CrossRef PubMed Google Scholar
12. Compernolle, V., et al. 2002. Loss of HIF-2alpha and inhibition of VEGF impair fetal lung maturation, whereas treatment with VEGF prevents fatal respiratory distress in premature mice. Nat. Med. 8:702-710.
    View this article via: PubMed Google Scholar
13. Tian, H., Hammer, R.E., Matsumoto, A.M., Russell, D.W., McKnight, S.L. 1998. The hypoxia-responsive transcription factor EPAS1 is essential for catecholamine homeostasis and protection against heart failure during embryonic development. Genes Dev. 12:3320-3324.
    View this article via: CrossRef PubMed Google Scholar
14. Iyer, N.V., et al. 1998. Cellular and developmental control of O2 homeostasis by hypoxia-inducible factor 1 alpha. Genes Dev. 12:149-162.
    View this article via: CrossRef PubMed Google Scholar
15. Gruber, M., et al. 2007. Acute postnatal ablation of Hif-2alpha results in anemia. Proc. Natl. Acad. Sci. U. S. A. 104:2301-2306.
    View this article via: CrossRef PubMed Google Scholar
16. Kim, W.Y., et al. 2006. Failure to prolyl hydroxylate hypoxia-inducible factor alpha phenocopies VHL inactivation in vivo. EMBO J. 25:4650-4662.
    View this article via: CrossRef PubMed Google Scholar
17. Hu, C.J., Wang, L.Y., Chodosh, L.A., Keith, B., Simon, M.C. 2003. Differential roles of hypoxia-inducible factor 1alpha (HIF-1alpha) and HIF-2alpha in hypoxic gene regulation. Mol. Cell. Biol. 23:9361-9374.
    View this article via: CrossRef PubMed Google Scholar
18. Hu, C.J., Sataur, A., Wang, L., Chen, H., Simon, M.C. 2007. The N-terminal transactivation domain confers target gene specificity of hypoxia-inducible factors HIF-1alpha and HIF-2alpha. Mol. Biol. Cell. 18:4528-4542.
    View this article via: CrossRef PubMed Google Scholar
19. Raval, R.R., et al. 2005. Contrasting properties of hypoxia-inducible factor 1 (HIF-1) and HIF-2 in von Hippel-Lindau-associated renal cell carcinoma. Mol. Cell. Biol. 25:5675-5686.
    View this article via: CrossRef PubMed Google Scholar
20. Wang, V., Davis, D.A., Haque, M., Huang, L.E., Yarchoan, R. 2005. Differential gene up-regulation by hypoxia-inducible factor-1alpha and hypoxia-inducible factor-2alpha in HEK293T cells. Cancer Res. 65:3299-3306.
    View this article via: PubMed Google Scholar
21. Kim, W.Y., Kaelin, W.G. 2004. Role of VHL gene mutation in human cancer. J. Clin. Oncol. 22:4991-5004.
    View this article via: CrossRef PubMed Google Scholar
22. Maxwell, P.H., et al. 1999. The tumour suppressor protein VHL targets hypoxia-inducible factors for oxygen-dependent proteolysis. Nature. 399:271-275.
    View this article via: CrossRef PubMed Google Scholar
23. Gordan, J.D., et al. 2008. HIF-alpha effects on c-Myc distinguish two subtypes of sporadic VHL-deficient clear cell renal carcinoma. Cancer Cell. 14:435-446.
    View this article via: CrossRef PubMed Google Scholar
24. Mandriota, K.T., et al. 2002. Hif activation identifies early lesion in vhl kidneys: evidence for site-specific tumor suppressor function in the nephron. Cancer Cell. 1:459-468.
    View this article via: CrossRef PubMed Google Scholar
25. Kondo, K., Kim, W.Y., Lechpammer, M., Kaelin (Jr.), W.G. 2003. Inhibition of HIF2alpha is sufficient to suppress pVHL-defective tumor growth. PLoS Biol. 1:E83.
    View this article via: PubMed Google Scholar
26. Kondo, K., Klco, J., Nakamura, E., Lechpammer, M., Kaelin (Jr.), W.G. 2002. Inhibition of HIF is necessary for tumor suppression by the von Hippel-Lindau protein. Cancer Cell. 1:237-246.
    View this article via: CrossRef PubMed Google Scholar
27. Zimmer, M., Doucette, D., Siddiqui, N., Iliopoulos, O. 2004. Inhibition of hypoxia-inducible factor is sufficient for growth suppression of VHL–/– tumors. Mol. Cancer Res. 2:89-95.
    View this article via: PubMed Google Scholar
28. Maranchie, J.K., et al. 2002. The contribution of VHL substrate binding and HIF1-alpha to the phenotype of VHL loss in renal cell carcinoma. Cancer Cell. 1:247-255.
    View this article via: CrossRef PubMed Google Scholar
29. Kaelin (Jr.), W.G. 2008. The von Hippel-Lindau tumour suppressor protein: O2 sensing and cancer. Nat. Rev. Cancer. 8:865-873.
    View this article via: CrossRef PubMed Google Scholar
30. Swinson, D.E., et al. 2003. Carbonic anhydrase IX expression, a novel surrogate marker of tumor hypoxia, is associated with a poor prognosis in non-small-cell lung cancer. J. Clin. Oncol. 21:473-482.
    View this article via: CrossRef PubMed Google Scholar
31. Le, Q.T., et al. 2006. An evaluation of tumor oxygenation and gene expression in patients with early stage non-small cell lung cancers. Clin. Cancer Res. 12:1507-1514.
    View this article via: CrossRef PubMed Google Scholar
32. Lau, S.K., et al. 2007. Three-gene prognostic classifier for early-stage non small-cell lung cancer. J. Clin. Oncol. 25:5562-5569.
    View this article via: CrossRef PubMed Google Scholar
33. Giatromanolaki, A., et al. 2001. Relation of hypoxia inducible factor 1 alpha and 2 alpha in operable non-small cell lung cancer to angiogenic/molecular profile of tumours and survival. Br. J. Cancer. 85:881-890.
    View this article via: CrossRef PubMed Google Scholar
34. Kim, S.J., et al. 2005. Expression of HIF-1alpha, CA IX, VEGF, and MMP-9 in surgically resected non-small cell lung cancer. Lung Cancer. 49:325-335.
    View this article via: CrossRef PubMed Google Scholar
35. Socinski, M.A., et al. 2008. Multicenter, phase II trial of sunitinib in previously treated, advanced non-small-cell lung cancer. J. Clin. Oncol. 26:650-656.
    View this article via: CrossRef PubMed Google Scholar
36. Sandler, A., et al. 2006. Paclitaxel-carboplatin alone or with bevacizumab for non-small-cell lung cancer. N. Engl. J. Med. 355:2542-2550.
    View this article via: CrossRef PubMed Google Scholar
37. Meuwissen, R., Linn, S.C., van der Valk, M., Mooi, W.J., Berns, A. 2001. Mouse model for lung tumorigenesis through Cre/lox controlled sporadic activation of the K-Ras oncogene. Oncogene. 20:6551-6558.
    View this article via: CrossRef PubMed Google Scholar
38. Jackson, E.L., et al. 2001. Analysis of lung tumor initiation and progression using conditional expression of oncogenic K-ras. Genes Dev. 15:3243-3248.
    View this article via: CrossRef PubMed Google Scholar
39. Johnson, L., et al. 2001. Somatic activation of the K-ras oncogene causes early onset lung cancer in mice. Nature. 410:1111-1116.
    View this article via: CrossRef PubMed Google Scholar
40. Fisher, G.H., et al. 2001. Induction and apoptotic regression of lung adenocarcinomas by regulation of a K-Ras transgene in the presence and absence of tumor suppressor genes. Genes Dev. 15:3249-3262.
    View this article via: CrossRef PubMed Google Scholar
41. Wykoff, C.C., et al. 2004. Gene array of VHL mutation and hypoxia shows novel hypoxia-induced genes and that cyclin D1 is a VHL target gene. Br. J. Cancer. 90:1235-1243.
    View this article via: CrossRef PubMed Google Scholar
42. Wykoff, C.C., Pugh, C.W., Maxwell, P.H., Harris, A.L., Ratcliffe, P.J. 2000. Identification of novel hypoxia dependent and independent target genes of the von Hippel-Lindau (VHL) tumour suppressor by mRNA differential expression profiling. Oncogene. 19:6297-6305.
    View this article via: CrossRef PubMed Google Scholar
43. Ji, H., et al. 2007. LKB1 modulates lung cancer differentiation and metastasis. Nature. 448:807-810.
    View this article via: CrossRef PubMed Google Scholar
44. Iwanaga, K., et al. 2008. Pten inactivation accelerates oncogenic K-ras-initiated tumorigenesis in a mouse model of lung cancer. Cancer Res. 68:1119-1127.
    View this article via: CrossRef PubMed Google Scholar
45. Jackson, E.L., et al. 2005. The differential effects of mutant p53 alleles on advanced murine lung cancer. Cancer Res. 65:10280-10288.
    View this article via: CrossRef PubMed Google Scholar
46. Jain, R.K., Munn, L.L., Fukumura, D. 2002. Dissecting tumour pathophysiology using intravital microscopy. Nat. Rev. Cancer. 2:266-276.
    View this article via: CrossRef PubMed Google Scholar
47. Elson, D.A., et al. 2001. Induction of hypervascularity without leakage or inflammation in transgenic mice overexpressing hypoxia-inducible factor-1alpha. Genes Dev. 15:2520-2532.
    View this article via: CrossRef PubMed Google Scholar
48. Dome, B., et al. 2006. Identification and clinical significance of circulating endothelial progenitor cells in human non-small cell lung cancer. Cancer Res. 66:7341-7347.
    View this article via: CrossRef PubMed Google Scholar
49. Ho, J.W., et al. 2006. Significance of circulating endothelial progenitor cells in hepatocellular carcinoma. Hepatology. 44:836-843.
    View this article via: CrossRef PubMed Google Scholar
50. Kollet, O., et al. 2001. Rapid and efficient homing of human CD34(+)CD38(-/low)CXCR4(+) stem and progenitor cells to the bone marrow and spleen of NOD/SCID and NOD/SCID/B2m(null) mice. Blood. 97:3283-3291.
    View this article via: CrossRef PubMed Google Scholar
51. Willett, C.G., et al. 2004. Direct evidence that the VEGF-specific antibody bevacizumab has antivascular effects in human rectal cancer. Nat. Med. 10:145-147.
    View this article via: CrossRef PubMed Google Scholar
52. Zheng, P.P., Hop, W.C., Luider, T.M., Sillevis Smitt, P.A., Kros, J.M. 2007. Increased levels of circulating endothelial progenitor cells and circulating endothelial nitric oxide synthase in patients with gliomas. Ann. Neurol. 62:40-48.
    View this article via: CrossRef PubMed Google Scholar
53. De Falco, E., et al. 2004. SDF-1 involvement in endothelial phenotype and ischemia-induced recruitment of bone marrow progenitor cells. Blood. 104:3472-3482.
    View this article via: CrossRef PubMed Google Scholar
54. Asahara, T., et al. 1999. VEGF contributes to postnatal neovascularization by mobilizing bone marrow-derived endothelial progenitor cells. EMBO J. 18:3964-3972.
    View this article via: CrossRef PubMed Google Scholar
55. Furstenberger, G., et al. 2006. Circulating endothelial cells and angiogenic serum factors during neoadjuvant chemotherapy of primary breast cancer. Br. J. Cancer. 94:524-531.
    View this article via: CrossRef PubMed Google Scholar
56. Onder, T.T., et al. 2008. Loss of E-cadherin promotes metastasis via multiple downstream transcriptional pathways. Cancer Res. 68:3645-3654.
    View this article via: CrossRef PubMed Google Scholar
57. Ben-Porath, I., et al. 2008. An embryonic stem cell-like gene expression signature in poorly differentiated aggressive human tumors. Nat. Genet. 40:499-507.
    View this article via: CrossRef PubMed Google Scholar
58. Cole, M.F., Johnstone, S.E., Newman, J.J., Kagey, M.H., Young, R.A. 2008. Tcf3 is an integral component of the core regulatory circuitry of embryonic stem cells. Genes Dev. 22:746-755.
    View this article via: CrossRef PubMed Google Scholar
59. Bild, A.H., et al. 2006. Oncogenic pathway signatures in human cancers as a guide to targeted therapies. Nature. 439:353-357.
    View this article via: CrossRef PubMed Google Scholar
60. Yang, J., Weinberg, R.A. 2008. Epithelial-mesenchymal transition: at the crossroads of development and tumor metastasis. Dev. Cell. 14:818-829.
    View this article via: CrossRef PubMed Google Scholar
61. Gort, E.H., et al. 2008. The TWIST1 oncogene is a direct target of hypoxia-inducible factor-2alpha. Oncogene. 27:1501-1510.
    View this article via: CrossRef PubMed Google Scholar
62. Yang, M.H., et al. 2008. Direct regulation of TWIST by HIF-1alpha promotes metastasis. Nat. Cell Biol. 10:295-305.
    View this article via: CrossRef PubMed Google Scholar
63. Esteban, M.A., et al. 2006. Regulation of E-cadherin expression by VHL and hypoxia-inducible factor. Cancer Res. 66:3567-3575.
    View this article via: CrossRef PubMed Google Scholar
64. Krishnamachary, B., et al. 2006. Hypoxia-inducible factor-1-dependent repression of E-cadherin in von Hippel-Lindau tumor suppressor-null renal cell carcinoma mediated by TCF3, ZFHX1A, and ZFHX1B. Cancer Res. 66:2725-2731.
    View this article via: CrossRef PubMed Google Scholar
65. Zhou, B.B., et al. 2006. Targeting ADAM-mediated ligand cleavage to inhibit HER3 and EGFR pathways in non-small cell lung cancer. Cancer Cell. 10:39-50.
    View this article via: CrossRef PubMed Google Scholar
66. Shedden, K., et al. 2008. Gene expression-based survival prediction in lung adenocarcinoma: a multi-site, blinded validation study. Nat. Med. 14:822-827.
    View this article via: CrossRef PubMed Google Scholar
67. Kenny, P.A., Enver, T., Ashworth, A. 2005. Receptor and secreted targets of Wnt-1/beta-catenin signalling in mouse mammary epithelial cells. BMC Cancer. 5:3.
    View this article via: PubMed Google Scholar
68. Verrecchia, F., Chu, M.L., Mauviel, A. 2001. Identification of novel TGF-beta /Smad gene targets in dermal fibroblasts using a combined cDNA microarray/promoter transactivation approach. J. Biol. Chem. 276:17058-17062.
    View this article via: CrossRef PubMed Google Scholar
69. Jechlinger, M., et al. 2003. Expression profiling of epithelial plasticity in tumor progression. Oncogene. 22:7155-7169.
    View this article via: CrossRef PubMed Google Scholar
70. Ryan, H.E., et al. 2000. Hypoxia-inducible factor-1alpha is a positive factor in solid tumor growth. Cancer Res. 60:4010-4015.
    View this article via: PubMed Google Scholar
71. Ryan, H.E., Lo, J., Johnson, R.S. 1998. HIF-1 alpha is required for solid tumor formation and embryonic vascularization. EMBO J. 17:3005-3015.
    View this article via: CrossRef PubMed Google Scholar
72. Maxwell, P.H., et al. 1997. Hypoxia-inducible factor-1 modulates gene expression in solid tumors and influences both angiogenesis and tumor growth. Proc. Natl. Acad. Sci. U. S. A. 94:8104-8109.
    View this article via: CrossRef PubMed Google Scholar
73. Liao, D., Corle, C., Seagroves, T.N., Johnson, R.S. 2007. Hypoxia-inducible factor-1alpha is a key regulator of metastasis in a transgenic model of cancer initiation and progression. Cancer Res. 67:563-572.
    View this article via: CrossRef PubMed Google Scholar
74. Lu, Z.H., Wright, J.D., Belt, B., Cardiff, R.D., Arbeit, J.M. 2007. Hypoxia-inducible factor-1 facilitates cervical cancer progression in human papillomavirus type 16 transgenic mice. Am. J. Pathol. 171:667-681.
    View this article via: CrossRef PubMed Google Scholar
75. Carmeliet, P., et al. 1998. Role of HIF-1alpha in hypoxia-mediated apoptosis, cell proliferation and tumour angiogenesis. Nature. 394:485-490.
    View this article via: CrossRef PubMed Google Scholar
76. Mack, F.A., et al. 2003. Loss of pVHL is sufficient to cause HIF dysregulation in primary cells but does not promote tumor growth. Cancer Cell. 3:75-88.
    View this article via: CrossRef PubMed Google Scholar
77. Blouw, B., et al. 2003. The hypoxic response of tumors is dependent on their microenvironment. Cancer Cell. 4:133-146.
    View this article via: CrossRef PubMed Google Scholar
78. Acker, T., et al. 2005. Genetic evidence for a tumor suppressor role of HIF-2alpha. Cancer Cell. 8:131-141.
    View this article via: CrossRef PubMed Google Scholar
79. Haase, V.H., Glickman, J.N., Socolovsky, M., Jaenisch, R. 2001. Vascular tumors in livers with targeted inactivation of the von Hippel-Lindau tumor suppressor. Proc. Natl. Acad. Sci. U. S. A. 98:1583-1588.
    View this article via: CrossRef PubMed Google Scholar
80. Ma, W., et al. 2003. Hepatic vascular tumors, angiectasis in multiple organs, and impaired spermatogenesis in mice with conditional inactivation of the VHL gene. Cancer Res. 63:5320-5328.
    View this article via: PubMed Google Scholar
81. Rankin, E.B., Tomaszewski, J.E., Haase, V.H. 2006. Renal cyst development in mice with conditional inactivation of the von Hippel-Lindau tumor suppressor. Cancer Res. 66:2576-2583.
    View this article via: CrossRef PubMed Google Scholar
82. Scortegagna, M., et al. 2008. HIF-1alpha regulates epithelial inflammation by cell autonomous NFkappaB activation and paracrine stromal remodeling. Blood. 111:3343-3354.
    View this article via: CrossRef PubMed Google Scholar
83. Bergers, G., Benjamin, L.E. 2003. Tumorigenesis and the angiogenic switch. Nat. Rev. Cancer. 3:401-410.
    View this article via: CrossRef PubMed Google Scholar
84. Bertolini, F., Mancuso, P., Kerbel, R.S. 2005. Circulating endothelial progenitor cells. N. Engl. J. Med. 353:2613-2616; author reply 2613–2616.
    View this article via: CrossRef PubMed Google Scholar
85. Evans, A.J., et al. 2007. VHL promotes E2 box-dependent E-cadherin transcription by HIF-mediated regulation of SIP1 and snail. Mol. Cell. Biol. 27:157-169.
    View this article via: CrossRef PubMed Google Scholar
86. Weidner, N., et al. 1992. Tumor angiogenesis: a new significant and independent prognostic indicator in early-stage breast carcinoma. J. Natl. Cancer Inst. 84:1875-1887.
    View this article via: CrossRef PubMed Google Scholar
87. Subramanian, A., et al. 2005. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc. Natl. Acad. Sci. U. S. A. 102:15545-15550.
    View this article via: CrossRef PubMed Google Scholar
88. Segal, E., Friedman, N., Kaminski, N., Regev, A., Koller, D. 2005. From signatures to models: understanding cancer using microarrays. Nat. Genet. 37(Suppl.):S38-S45.
    View this article via: PubMed Google Scholar