There are three HIFα family members called HIF1α, HIF2α, and HIF3α. Deregulation of HIFα, in particular HIF2α, appears to be a driving force in pVHL-defective kidney cancer. For example, the risk of renal carcinoma linked to different VHL mutations correlates with the degree to which those mutations deregulate HIF [55–57]. VHL−/− renal carcinoma cells frequently silence the expression of FBP1, which is an other endogenous inhibitor of HIF activity [58].

pVHL-defective clear cell renal carcinomas overproduce HIF2α but, in some cases, fail to produce HIF1α [28, 42, 59, 60]. Production of a non-hydroxylatable version of HIF2α, but not HIF1α, can override the tumor suppressor activity of pVHL in preclinical models [61, 62]. Similarly, exogenous overexpression of HIF2α, but not HIF1α, promotes tumor formation by VHL−/− renal cancer cells [63, 64]. Moreover, downregulation of HIF2α, but not HIF1α, is sufficient to suppress tumor formation by pVHL-defective clear cell renal carcinomas [65, 66]. The appearance of HIF2α in premalignant renal lesions in patients with VHL disease heralds malignant transformation [67, 68], and a human single nucleotide polymorphism (SNP) linked to HIF2α on chromosome 2p21 has been associated with the risk of developing clear cell renal carcinomas [69]. Finally, much of the pathology observed after VHL inactivation in genetically engineered mouse models can be linked to the inappropriate accumulation of HIF2α [68, 70–75]. It should be noted that VHL inactivation, but not bona fide hypoxia, is sufficient to induce HIF2α in mouse renal tubular epithelial cells and cause renal cyst formation [68, 72, 76]. Neither VHL inactivation nor increased HIF2α activity, however, is sufficient to cause clear cell renal carcinoma in genetically engineered mouse models [68, 72, 76, 77]. This presumably reflects the need for cooperating genetic events (see below) and perhaps species differences.

As noted above, some clear cell renal carcinoma cell lines and tumors produce low, or undetectable, amounts of HIF1α. Indeed, some VHL−/− clear cell renal carcinoma lines harbor homozygous mutations of the HIF1α locus [60]. Reintroduction of wild-type HIF1α into such lines suppresses their proliferation in cell culture and in nude mice xenograft studies [60, 63, 64]. Conversely, downregulation of HIF1α in HIF1α-proficient VHL−/− clear cell renal carcinoma lines enhances their proliferation in cell culture and in xenograft assays [59, 60]. Interestingly, HIF1α resides on chromosome 14q, which is frequently deleted in clear cell renal carcinomas (together with chromosome 3p loss and chromosome 5q amplification) [60]. Clear cell renal carcinomas with chromosome 14q deletions have gene expression signatures consistent with decreased HIF1α activity [60, 78]. In some VHL−/− clear cell carcinomas that express both HIF1α and HIF2α, the ratio of HIF2α to HIF1α is enhanced by loss of specific microRNAs miR-30c-2-3p and miR-30A-3p that normally serve to repress HIF2α [79]. Finally, loss-of-function intragenic HIF1α mutations have occasionally been identified in VHL−/− clear cell renal carcinomas [60, 80–82]. Collectively, these findings suggest that HIF1α, in contrast to HIF2α, acts as a tumor suppressor in VHL−/− clear cell renal carcinoma.

In apparent disagreement with this contention, expression of a stabilized version of HIF1α, but not a stabilized version of HIF2α, in the proximal renal tubular epithelial cells of mice caused renal cell dysplasia, including evidence of increased proliferation, increased DNA damage, and clear cell histological changes [83, 84]. Similarly, ablation of VHL in primarily mouse collecting ducts caused hyperplastic changes that could be reversed by simultaneous inactivation of HIF1α [85]. Finally, it has also been shown that silencing HIF1α inhibits, rather than augments, tumor growth by human VHL+/+ renal carcinoma growth [86].

There are a number of caveats to these studies, however. For example, the cell of origin for VHL−/− clear cell renal carcinoma is still debated but likely involves a distal tubular epithelial cell that is permissive for HIF2α accumulation and the expression of specific HIF2α target genes (e.g. cyclin D1) following pVHL loss [67, 68, 87]. In this regard, forced expression of a stabilized version of HIF2α in the murine proximal renal tubule did not recapitulate the induction of HIF targets seen in VHL−/− clear cell renal carcinoma [83], perhaps because the wrong cell type was targeted. The genetically engineered mouse studies might also be confounded by biological differences between mice and men, as has been observed with many other cancer genes. Finally, the apparent dependence of human VHL+/+ renal carcinomas on HIF1α for tumor growth does not preclude a tumor suppressor role for HIF1α in VHL−/− renal carcinomas, especially bearing in mind potential differences in cell of origin and cooperating genetic events.

There are a number of quantitative and qualitative differences between HIF1α and HIF2α that could account for their seemingly antagonistic effects in VHL−/− clear cell renal carcinoma. These differences likely reflect the fact that some HIF target genes are preferentially activated by specific HIFα family members as well as by the existence of non-canonical HIF functions that are unique to specific HIFα proteins. HIF2α cooperates with c-Myc to promote the proliferation of VHL−/− clear cell renal carcinoma cells, while HIF1α is capable of inhibiting c-Myc [88–91]. Both HIF1α and HIF2α can induce REDD1 and thereby suppress the activity of the TORC1 complex, which contains mTOR, and Cap-dependent translation [92–95]. HIF2α, however, and not HIF1α, can also stimulate translation. HIF2α transcriptionally induces the amino acid transporter SLC7A5 and thereby increases intracellular amino acid availability, which activates TORC1 [96]. In addition, HIF2α forms a complex with RBM4 and eIF4E that promotes Cap-dependent translation in cells with depressed TORC1 activity [97]. HIF1α and HIF2α also appear to differentially regulate p53 and the DNA damage response [59, 63, 98, 99].

pVHL has a number of other functions that, although incompletely understood biochemically, appear to be a least partly HIF-independent. These include a role in the maintenance of a specialized structure called the primary cilium on the cell surface that serves as a mechanosensor [76, 100–103], possibly by virtue of pVHL’s role in stabilization of microtubules [104–106]. Interestingly, a number of diseases characterized by visceral cyst formation, including VHL disease, are caused by mutations that disrupt the primary cilium [107, 108]. pVHL also suppresses autophagy via both HIF-independent and HIF-dependent pathways, perhaps contributing to the increased autophagy seen in clear cell renal carcinomas [109, 110]. In addition, pVHL plays roles in extracellular matrix formation by fibronectin [111–114], epithelial-epithelial contacts [115, 116], NFκB signaling [117–120], control of atypical PKC activity [121–125], Rpb1 expression and activity [126–128], receptor internalization [129–131], and mRNA turnover [20, 26, 132–135]. It is possible that these other functions also contribute to tumor suppression by pVHL.

It is clear that pVHL loss is an important, but not sufficient, step in renal carcinogenesis. This is most clearly demonstrated by studies of the natural history of von Hippel-Lindau disease. Patients with von Hippel-Lindau disease can develop hundreds of premalignant renal cysts, very few of which will go on to become clear cell renal carcinomas [67, 136] (Fig. 3.2). This bottleneck presumably reflects the requirement for additional genetic events, occurring stochastically, to fully transform renal epithelial cells. Indeed, a number of nonrandom genomic abnormalities have been described in clear cell renal carcinoma including, most notably, 5q amplification and 14q loss [6, 137–143] (Fig. 3.2). The triad of 3p loss, 14q loss, and 5q gain is a signature of clear cell renal carcinoma, and some clear cell renal carcinomas have unbalanced translocations involving 3p and 5q that result in loss of 3p and gain of 5q sequences [60, 144–150].

Loss of chromosome 3p, which harbors the VHL tumor suppressor gene, is the most common genetic event in kidney cancer. Chromosome 3p has been suspected for many years, however, to contain at least one additional kidney cancer suppressor gene. Indeed, it is now clear that 3p harbors several renal cancer suppressor genes other than VHL including PBRM1, which encodes the BAF180 chromatin-associated protein; SETD2, which encodes a histone H3 lysine 36 methyltransferase; and BAP1, which encodes a ubiquitin hydrolase [82, 151–156] (Fig. 3.3). PBRM1 is, after VHL, the most frequently mutated gene in clear cell renal carcinoma. PBRM1 and BAP1 mutations are largely mutually exclusive and appear to define clinically distinct subgroups of renal cancers [152, 157, 158].

As described above, HIF1α is a likely target of the 14q deletions in VHL−/− clear cell renal carcinomas. These deletions are very large, however, suggesting there are additional renal cancer suppressor genes located at 14q. It should also be noted that most 14q deleted VHL−/− clear cell renal tumors (in contrast to cell lines) appear to retain a wild-type HIF1α allele [60]. This suggests that HIF1α is a haploinsufficient clear cell renal carcinoma suppressor and that loss of the remaining allele is associated with tumor progression in vivo or establishment of cell lines ex vivo.

SQSTM1, encoding p62, appears to be one of the renal carcinoma 5q oncogenes [159]. Increased expression of p62 promotes the growth of VHL−/− renal carcinoma cells in cell culture and tumor xenograft assays and increases their resistance to redox stress [159]. p62 plays important roles in autophagy and also signals to renal carcinoma relevant proteins including NRF2, NFκB, and mTOR [160–162].

Sequencing of kidney cancer genomes has identified additional genes that, when mutated, contribute to renal carcinogenesis including several more genes linked to chromatin regulation such as JARID1C (also known as KDM5C), which encodes a histone H3 lysine 4 demethylase; UTX (KMD6A), which encodes a histone H3 lysine 27 demethylase; and ARID1A, a component of a chromatin remodeling complex [82, 151–156, 163]. Notably, many histone demethylase genes are themselves transcriptionally induced by HIF [164–169]. It is possible that their inappropriate expression pursuant to VHL loss alters chromatin structure and creates the selection pressure to mutate specific chromatin regulators.

Genes linked to the mTOR pathway including PIK3CA, PTEN, TSC1, TSC2, and MTOR itself are occasionally mutated in clear cell renal carcinomas [7, 82, 152, 155, 156]. Preliminary data suggest that such mutations identify a subset of renal cell carcinoma patients more likely to derive significant benefit from TORC1 inhibitors [170].

The NFE2L gene, encoding NRF2, and the NRF2-negative regulator KEAP1 are occasionally mutated in clear cell renal carcinoma [82, 159]. Such mutations appear to be mutually exclusive with higher level SQSTM1 amplification [159]. Genes involved in the response to DNA damage, including p53, MDM4, and ATM, are also occasionally mutated in clear cell renal carcinoma [7, 82, 155, 156]. p53 loss cooperates with Vhl loss in mouse models to promote renal carcinogenesis [171].

The preclinical data outlined above suggest that drugs that inhibit HIF, and particularly HIF2α, would have antitumor activity in kidney cancer. Unfortunately, DNA-binding transcription factors, with the exception of the steroid hormone receptors, have historically been difficult to target with drug-like small molecules. Nonetheless, a number of approaches to targeting HIF are being explored in the laboratory, including the use of DNA-binding polyamides [172–174] and short interfering RNAs [175]. Moreover, HIF2α, but not HIF1α, has a potentially druggable pocket, and lead compounds have been identified that can inhibit HIF2α in biochemical, cell-based, and animal models [176–178].

mTOR participates in two complexes, called TORC1 and TORC2 [179]. The former can be inhibited with rapamycin-like drugs. Two such drugs, temsirolimus and everolimus, have been FDA approved for the treatment of renal cell carcinoma based on positive randomized clinical trial data [180, 181]. In theory the activity of these agents reflects direct effects on tumor cells, including modulation of HIF [182], and effects downstream of VEGF signaling in endothelial cells (see below). In preclinical models, VHL−/− renal carcinoma lines are more sensitive to rapamycin than are their pVHL-proficient counterparts [183]. As noted above, preliminary data suggest that concurrent mutations of the PI3K-MTOR pathway are enriched among renal carcinoma patients who exhibit the greatest clinical benefit from rapamycin-like drugs [170].

Two factors might, however, limit the effectiveness of rapamycin-like drugs in the treatment of renal cell carcinoma. First, the TORC1 complex feedback inhibits signaling by certain receptor tyrosine kinases [184, 185, 185a, 186–188]. As a result, treatment of tumor cells with rapamycin-like drugs can cause a paradoxical increase in receptor kinase activity leading to activation of TORC2, which is relatively rapamycin resistant, PI3K, and AKT, all of which might promote tumor growth [184, 185, 185a, 186–188]. Second, inhibition of TORC1 appears to preferentially inhibit HIF1α, which as argued above appears to act a renal cell carcinoma suppressor, rather than HIF2α [189]. In contrast, inhibition of TORC2 preferentially affects HIF2α [189]. Second-generation, ATP-like, mTOR inhibitors can inhibit both TORC1 and TORC2 and hence might be more active than rapamycin-like drugs in the treatment of clear cell renal carcinoma [190, 191]. Emerging preclinical data support such a view [192].

HIF1α upregulates a number of genes that promote glycolysis and lactate acid production. This potentially places a burden on pVHL-defective tumor cells to maintain pH homeostasis. Preclinical studies suggest that inhibition of lactate dehydrogenase A or carbonic anhydrase IX, both of which are HIF targets, would be a viable therapeutic strategy for treating pVHL-defective renal cell carcinomas [275–278].

Resequencing of renal cell carcinoma genomes has identified mutations affecting enzymes that regulate histone methylation, as described above. In addition, HIF transcriptionally activates a number of histone demethylases including JMJD1A, JMJD2B, and JARID1B [164–169]. In one study, inhibition of JMJD1A with a short hairpin RNA inhibited renal carcinoma growth [168]. Histone methylases and demethylases can, in principle, be inhibited with drug-like small molecules, and the identification of these enzymes as mutational targets in renal cell carcinoma and other neoplasms is motivating a deeper understanding of their biological functions as well as nascent drug discovery efforts.

Renal cell carcinoma is a common cancer that, historically, has been refractory to therapy with standard chemotherapeutic agents and radiation. High-dose interleukin-2 can induce durable remissions in a small subset of patients, but it is impossible to predict which patients will benefit from this toxic and expensive form of therapy. The von Hippel-Lindau tumor suppressor protein, pVHL, is frequently inactivated in clear cell renal carcinoma, which is the most common form of kidney cancer. The knowledge that pVHL inhibits the HIF transcription factor provided a conceptual framework for drugs that inhibit the HIF-responsive gene product VEGF. The clinical activity of mTOR inhibitors might also relate to HIF biology because mTOR regulates HIF synthesis and might also act downstream of VEGF. A number of other HIF-responsive gene products are also known or suspected of playing roles in tumorigenesis and are worthy of exploration as kidney cancer drug targets. Elucidation of the genetic events that cooperate with pVHL loss in clear cell carcinoma will hopefully yield additional targets. In this regard, it is anticipated that the frequent occurrence of mutations affecting chromatin regulatory proteins in clear cell renal carcinoma will create exploitable therapeutic vulnerabilities. Finally, there is a growing appreciation of how, mechanistically, clear cell renal carcinoma subverts immune recognition. The studies, in total, should provide a platform for the design and testing of effective therapeutic combinations for this disease.