Abstract
Background: Inactivation of the Von Hippel-Lindau (VHL) tumour suppressor gene leading to overexpression of hypoxia-inducible transcription factors (HIF)-1α and -2α is a critical event in the pathogenesis of most clear cell renal cell carcinomas (RCC). HIF-1α and HIF-2α share significant homology and regulate overlapping repertoires of hypoxia-inducible target genes but may have differing effects on RCC cell growth. Loss of HIF-1α expression has been described in RCC cell lines and primary tumours. Whether mutations in the alpha-subunits of HIF-1α and HIF-2α contribute to renal tumourigenesis was investigated here. Materials and Methods: Mutation analysis of the complete coding sequence of HIF-1α and HIF-2α was carried out in primary RCC (n=40). Results: The analysis revealed a somatic HIF1A missense substitution, p.Val116Glu, in a single RCC. Functional studies demonstrated that p.Val116Glu impaired HIF-1α transcriptional activity. Genotyping of HIF1A variants p.Pro582Ser and p.Ala588Thr demonstrated no significant differences between RCC patients and controls. Conclusion: The detection of a loss-of-function HIF1A mutation in a primary RCC is consistent with HIF-1 and HIF-2 having different roles in renal tumourigenesis, However, somatic mutations of HIF1A are not frequently implicated in the pathogenesis of RCC.
Investigations of the VHL tumour suppressor gene have highlighted the role of the heterodimeric hypoxia-inducible transcription factors (HIF)-1α and -2α in the pathogenesis of clear cell renal cell carcinoma (ccRCC) (1, 2). Germline mutations in the VHL tumour suppressor gene cause von Hippel-Lindau (VHL) disease, a dominantly inherited familial cancer syndrome characterized by susceptibility to retinal and central nervous system haemangioblastomas, ccRCC, phaeochromocytoma, pancreatic islet cell tumours and renal, pancreatic and epididymal cysts (3, 4). In addition, VHL inactivation by somatic mutations and de novo promoter region methylation occurs in most sporadic ccRCC (5-8). Although multiple functions have been ascribed to the VHL gene product, pVHL, the best described function is its role in regulating the protein levels of HIF-1α and HIF-2α subunits (1, 9-11). Under normoxic conditions, pVHL binds to the HIF-α subunits, targeting them for ubiquitylation and proteasomal degradation.
Inactivation of pVHL in VHL disease and sporadic ccRCC leads to stabilization of the α-subunits and up-regulation of HIF-1α and -2α expression (1). HIF-1α and HIF-2α play a critical role in regulating cellular responses to hypoxia, including the regulation of a wide range of hypoxia-inducible genes involved in energy metabolism (e.g. glucose transporter 1, GLUT1), angiogenesis (e.g. vascular endothelial growth factor, VEGF), and apoptosis/autophagy (e.g. BCL2/ adenovirus E1B 19k Da interacting protein 3, BNIP3) (12). Accordingly, RCC with VHL inactivation demonstrate diffuse expression of hypoxia-inducible genes, although pVHL may also regulate target gene expression by HIF-independent mechanisms (13-19). Deregulation of HIF is also a feature of other familial RCC syndromes such as that associated with germline fumarate hydratase mutations (20, 21).
Previously, we investigated whether mutations in elongins B (TCEB2) and C (TCEB1P3), RBX1, and in the oxygen-dependent degradation domain (ODDD) of HIF-1A (codons 549-582) were implicated in the pathogenesis of RCC without VHL mutations. Although we did not identify any pathogenic sequence variants in elongin B and C or RBX1, two HIF-1α amino acid substitutions (p.Pro582Ser and p.Ala588Thr) were detected in the germline of RCC patients and controls (22). Subsequently, it was reported that the p.Pro582Ser was more frequent in RCC patients than in controls (23). Moreover, investigations of early neoplastic lesions in VHL disease patients demonstrated that whereas HIF-1α up-regulation could be detected in the earliest lesions with VHL inactivation, in more advanced lesions HIF-2α was up-regulated and HIF-1α expression might be lost (24). Recently Gordan et al. (25) reported that 61% of RCC expressed increased expression of both HIF-1α and HIF-2α and ~27% of HIF-2α alone; no tumours expressed HIF-1α only. These reports prompted us to investigate whether mutations in HIF1A and HIF2A occurred during renal tumourigenesis and whether common germline variants in HIF1A influenced RCC susceptibility.
Materials and Methods
Patients and samples. For single nucleotide polymorphism (SNP) association analysis, 332 consecutive cases of RCC collected between 1999 and 2006 in Szczecin, Poland, were analysed. Population-matched control samples were randomly selected from healthy individuals (with negative family history of cancer) matched to cases by sex, year of birth and from the same region of the country. Local ethics committees approved the collection of samples and informed consent was obtained from each patient. DNA was extracted from primary RCC by standard techniques.
Mutation analysis of HIF1A and HIF2A. Intron-exon boundaries were determined by matching the cDNA sequence for HIF1A and HIF2A with the working draft sequence of the human genome, the UCSC assembly 5 (http://genome.ucsc.edu). Mutation screening was performed by direct sequencing on an ABI 3730 automated sequencer and WAVE Analysis (Applied Biosystems Inc. CA, USA). The sequences of the primers used are listed in Table I: the annealing temperature for all PCRs was 58°C. In preliminary experiments, a heterozygous nonsense mutation (c.370 C>T p.Gln124Stop) in exon 3 of HIF1A was identified in the VHL-mutated RCC cell line KTCL26.
Genotype analysis of Polish sporadic and familial RCC patients. Genotyping of the p.Pro582Ser variant and the p.Ala588Thr variant was performed by KBioscience (Herts, UK) using a modified TaqMan-based assay (KASPar), details of which can be found at http://www.kbioscience.co.uk. The genotyping success rate was 99.4% for the 313 controls and 98.9% for the 332 sporadic cases.
p.Val116Glu mutant plasmids. These were generated in pcDNA3 using QuikChange® site-directed mutagenesis kit (Stratagen, TX, USA). The mutation and the integrity of the whole HIF1A coding region were confirmed by DNA sequencing. Two independent clones were used for experiments.
HRE-reporter assay. MCF7 cells were seeded in 12-well dishes at 2×105 cells per dish and incubated overnight. The cells were then co-transfected using Fugene6 transfection reagent (Roche Applied Science, West Sussex, UK) with plasmids expressing HRE-responsive luciferase reporter (0.1 mg) and beta-galactosidase (pCMV Gal, 0.12 μg) minus or plus wild-type or mutant HIF1A. After 24 h, cell extracts were prepared in passive lysate buffer (Promega, Southampton, UK) and analyzed for beta-galactosidase and luciferase activities. For Western blot showing HIF-1α expression levels, cell extracts were prepared in UREA/SDS buffer and mAb from BD-Transduction Laboratories was used.
Statistical analysis. Analysis of data was performed using 2×2 contingency tables and Fisher's exact test, calculated using GraphPad software (http://www.graphpad.com/quickcalcs). A p-value of <0.05 was considered statistically significance.
Results
Mutation analysis of HIF-1A and HIF-2A in primary RCC. To determine whether mutations in HIF1A and HIF2A might contribute to renal tumourigenesis, mutation analysis of the complete coding sequence of both genes was performed in 40 RCC (12 ccRCC with VHL mutations, 21 RCC without VHL mutations and 7 non-cc RCC). Two HIF1A missense substitutions were detected in RCC tumours but not in controls. A somatic missense mutation, p.Val116Glu, was detected in a ccRCC with VHL inactivation; this mutation was not present in 188 control chromosomes. In addition, a p.Ala475Ser missense substitution was present in the germline and in tumour DNA of 1/40 RCC and 0/184 control chromosomes. Bioinformatic analysis (PolyPhen: http://genetics.bwh.harvard.edu/pph) suggested that the p.Val116Glu substitution was likely to be pathogenic (PolyPhen score 2.37), p.Val116 is conserved in all known HIF1A orthologues. However, similar analysis suggested that p.Ala475Ser was likely to be a benign variant (PolyPhen score 0.157). Moreover, p.Ala475 is not conserved in chicken and Xenopus laevis orthologues.
Germline HIF1A variants detected in both patients and controls included p.Thr418Ile (1/40 RCC patients and 1/92 control chromosomes) and p.Pro582Ser (7/40 RCC patients and in 11/184 control chromosomes). To further investigate whether mutations within the HIF-1α ODDD (pVHL-binding region) might be implicated in RCC, mutation analysis of this region was performed in a further 51 RCC tumours but no novel variants were identified.
In order to investigate whether HIF1A variants might influence susceptibility to RCC, two variants, p.Pro582Ser and p.Ala588Thr, which we identified in our sequencing screen (Figure 1) and have previously been associated with RCC (23), were genotyped. Genotyping was carried out for 332 RCC patients and 313 ethnically matched control samples (see Table II). The p.Pro582Ser variant demonstrated a slight shift towards homozygosity of the major allele (CC) in the sporadic RCC patients and an increased frequency of the C allele, but neither were statistically significant (p=0.22, and p=0.18 respectively). The p.Ala588Thr variant showed no significant differences between patients and controls.
Functional analysis of the p.Val116Glu substitution in HIF-1α. The effect of the p.Val116Glu substitution on HIF1A transcriptional activity was tested using an HRE-linked reported gene. Compared to the wild-type HIF-1α expressing vector, the p.Val116Glu substitution significantly reduced expression of HIF-1α (Figure 2).
Discussion
Although the VHL tumour suppressor gene appears to have multiple functions, dysregulation of the hypoxic gene response appears to play a critical role in renal tumourigenesis associated with VHL inactivation (1). Thus VHL mutations that do not compromise the ability of pVHL to regulate HIF expression (Type 2C) are not associated with RCC susceptibility and the risk of RCC associated with VHL mutations that dysregulate HIF correlates with the degree of dysregulation (26-28). More direct evidence for a role of HIF overexpression in renal tumourigenesis associated with VHL inactivation was provided by studies in which mutant HIF-1α and -2α that were resistant to pVHL-mediated-degradation were overexpressed in an RCC cell line with wild-type pVHL. Interestingly HIF-2α, but not HIF-1α, overexpression was associated with increased tumour growth (29-31). In addition, Raval et al. (32) reported that whereas enhanced expression of wild-type HIF-2α promoted growth of 786-O RCC cells as subcutaneous tumour xenografts in nude mice, HIF-1α had a negative effect on tumour growth. Interestingly, immunohistochemical studies of kidneys from patients with VHL disease demonstrated that whereas the earliest neoplastic lesions exhibited HIF-1α expression, HIF-2α up-regulation was most prominent in larger and more dysplastic lesions and overt carcinoma (4, 32). These observations suggest that although HIF-1α and -2α have overlapping functions and many hypoxia-inducible genes are regulated by both transcription factors, some targets are preferentially regulated by one form of HIF (e.g. thymosin beta (TMSB), integrin beta 3-binding protein (ITGB3BP), BCL2/adenovirus E1B 19 kDa interacting protein 3 (BNIP3) by HIF-1α and cyclin D1 (CCND1), erythropoietin (EPO) and vascular endothelial growth factor (VEGF) by HIF-2α) (18, 19, 32, 33). Furthermore, a recent report described opposing effects of HIF-1 and HIF-2 on c-Myc (HIF-1 inhibited and HIF-2 potentiated c-Myc transcriptional activity and cellular proliferation) (25). Thus it might be predicted that there might be preferential growth of cells with VHL and HIF-1 inactivation during renal tumourigenesis.
Previously, we performed mutation analysis of the ODDD of HIF1A in 46 RCC without VHL inactivation but did not find evidence that HIF1A ODDD mutations were implicated in the pathogenesis of these tumours. However, in view of the increasing knowledge regarding the differential effects of HIF-1α and HIF-2α in renal oncogenesis, in this study, we extended our investigations to cover all of the HIF1A and HIF2A coding regions. However, we found that HIF1A mutations were rare in primary tumours, although, consistent with the hypothesis that HIFA inactivation might promote RCC tumourigenesis, a somatic p.Val116Glu missense mutation detected in an RCC with VHL inactivation was shown to impair HIF-1A transcriptional activity. Although genes may be silenced by promoter hypermethylation, we have found no evidence that loss of HIF-1α expression in RCC cell lines results from epigenetic silencing (unpublished data). Consistent with these observations, Gordan et al. (25) found that there was no clear difference in HIF1A mRNA levels between VHL-null RCC tumours expressing both HIF-1 and HIF-2 and those that only expressed HIF-2.
We investigated whether germline HIF1A variants might influence RCC susceptibility. The HIF1A p.Pro582Ser missense substitution was of particular interest as (a) it is within the ODDD domain (amino acids 549-582) and therefore might potentially influence pVHL HIF-1α interactions and (b) previously p.Pro582Ser was linked to HIF-1α protein expression in non-small cell lung carcinomas (34) and to susceptibility to androgen-independent prostate cancer (35). In non-small cell lung carcinomas, the T allele (coding for p.Ser582) was associated with higher expression of HIF-1α protein and it is this allele that was also reported to be associated with susceptibility to androgen-independent prostate cancer. Previously, in a smaller study (160 RCC patients and 162 controls), Ollerenshaw et al. (23) demonstrated statistically significant overrepresentation of the C1772 (p.Pro582) allele homozygotes in RCC patients. Although we observed a similar trend towards overrepresentation of homozygosity for the C (p.Pro582) allele in RCC patients, this failed to reach statistical significance (p=0.22). Hence further investigation of p.Pro582 is required to substantiate the previously reported association with RCC. Although Percy et al. (36) did not detect any effect of the p.Pro582Ser substitution on hydroxylation of the two prolines that determine pVHL-mediated HIF-1α degradation, it is intriguing that the C allele overrepresented in RCC patients was reported to be associated with lower HIF-1α expression in lung cancer (34) and that HIF1A inactivation was a feature of many RCC cell lines.
Despite evidence that HIF-2 promotes tumourigenesis we did not detect evidence of HIF2A activating mutations. Recently, a germline HIF2A mutation (c.1609C>T, p.Gly537Trp) that impairs hydroxylation of HIF-2α causing abnormal protein stabilization was reported to cause familial erythrocytosis (37). However, none of the three mutation carriers had a history of RCC. Although we cannot exclude that activating mutations of HIF2A could occur in RCC, our findings suggest that such mutations would be rare. Furthermore, it is unclear whether in the absence of VHL mutations activation mutations of HIF2A would be sufficient to promote tumourigenesis. At present, the mechanism by which HIF-1 expression is down-regulated in a subgroup of VHL-mutated RCC is unclear as somatic inactivation by HIF1A mutations appears infrequent. Nevertheless, further analysis of known regulators of HIF-1 translation or stability (e.g. the mammalian target of rapamycin (mTOR) pathway (38, 39)) might provide further insights into the role of HIF-related pathways in renal oncogenesis.
Acknowledgements
We thank Cancer Research UK for financial support. No funding body influenced the design or conduct of the studies described.
- Received July 28, 2009.
- Revision received October 12, 2009.
- Accepted October 15, 2009.
- Copyright© 2009 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved