In the Eker rat model, inactivation of the Tuberous Sclerosis-2 (Tsc-2) tumor suppressor gene leads to high frequency of spontaneous renal cell carcinoma (RCC). By analogy to human RCC in which mutations in the von Hippel-Lindau (VHL) tumor suppressor gene result in accumulation of hypoxia-inducible factor α (HIFα) and up-regulation of vascular endothelial growth factor (VEGF), we investigated the regulation of HIF and its target gene VEGF in rat RCC resulting from Tsc-2 defects. To examine HIFα activity, a panel of rat renal epithelial cells were analyzed for expression of HIF1α and the homologous protein, HIF2α, under normoxic and hypoxic conditions. RCC-derived cell lines exhibited high basal levels of HIF activity as determined using hypoxia response element-luciferase reporter constructs. HIF2α was stabilized in RCC-derived cell lines and in five of six primary tumors compared with normal kidney, which was consistent with the high levels of hypoxia response element-reporter activity observed in the cell lines. Primary RCCs that developed in Eker rats were highly vascularized, which was similar to their human counterparts. Furthermore, reverse-transcriptase PCR and immunoblotting demonstrated that VEGF was abundantly expressed in both rat RCC cell lines and primary tumors. The 120-, 164-, and 188-amino-acid isoforms of VEGF were expressed at the RNA and protein levels in RCC-derived cell lines, although only a single band was observed in primary tumors. Taken together, these data suggest that RCC caused by loss of the Tsc-2 tumor suppressor gene (which retain wild-type Vhl) up-regulate VEGF via a HIF2α-mediated mechanism. Thus, loss of Tsc-2 and VHL tumor suppressor gene function appears to have similar consequences in Eker rats and humans respectively, identifying dysregulation of HIFα and VEGF expression as a common pathway for the development of RCC in different species and in tumors with different molecular etiologies.
The ability to adapt to changes in oxygen availability is critical for tumor angiogenesis, metastasis, and physiological and pathological processes such as development and wound healing. The basic-helix-loop-helix PAS3 domain transcription factor hypoxia-inducible factor 1 (HIF1) plays a critical role in oxygen homeostasis (1). It is composed of two subunits, HIF1α and HIF1β. HIF1α is the oxygen-regulated component that determines HIF activity (2). HIF1α accumulation occurs during hypoxia as a result of inhibition of its proteolytic degradation through the ubiquitin proteasome pathway (3, 4, 5). HIF regulates many genes involved in maintaining O2 homeostasis and the physiological response to O2 deprivation, such as erythropoietin, glucose transporters, glycolytic pathway enzymes, VEGF, heme oxygenase, and inducible nitric oxide synthase (1). Sharing 48% homology with HIF1α, HIF2α is also called endothelial PAS domain protein-1 (EPAS1; see Ref. 6), HIF1α-like factor (HLF; see Ref. 7), HIF-related factor (HRF; see Ref. 8), or MOP-2 (“member of PAS superfamily”; see Ref. 9) and is also present in a number of cell types and tissues (10). Although they differ in abundance and distribution, HIF-1α and HIF2α appear to function similarly. Both become stabilized in response to hypoxia and function as a heterodimer with HIF1β, transactivating the expression of reporter genes containing HRE in DNA (6).
Comprised of at least five known isoforms, VEGF is a potent endothelial cell-specific mitogen that promotes the growth and maintenance of vascular endothelial cells and is the major angiogenesis inducer in vivo (11). VEGF expression can be regulated at the transcriptional level or by stabilization of VEGF mRNA. Binding of HIF1 to the consensus HRE in the VEGF promoter promotes VEGF gene expression (12, 13). Studies on regulation of VEGF and erythropoietin by hypoxia (12, 14, 15) revealed that for both genes, hypoxia-inducibility is conferred by homologous enhancer sequences. A 28-bp region in the 5′ promoter of rat and human VEGF (15) has high homology and similar protein binding characteristics as the HIF1 binding sites (HRE sites) in EPO (16). Many studies have shown that high levels of VEGF are produced by various types of tumors [for review see Ref. 17)]. These and other studies have established the role of VEGF in tumor angiogenesis and underscore the importance of identifying the regulatory mechanisms of VEGF expression in tumors.
RCC arises from the epithelial cells of the renal nephron and is characterized by its many different variants (18). Alterations in the von Hippel-Lindau (VHL) tumor suppressor gene are associated with the clear-cell variant of this disease (19, 20), which accounts for ∼75% of all RCC (21). Both HIF1α and HIF2α are regulated by VHL-mediated ubiquitination and degradation and are short-lived under normoxic conditions. Under normoxic conditions, HIF1α/2α is a major target for the specific E3 ubiquitin ligase activity of pVHL (1). Proline hydroxylation of HIF in the presence of oxygen targets this protein for recognition by VHL, ubiquitination, and proteolysis (22, 23, 24, 25, 26, 27, 28). Mutations in VHL in human RCC result in accumulation of HIF1α/2α and further up-regulation of VEGF mRNA expression (23, 29, 30, 31, 32, 33).
In contrast to human RCC, the Tsc-2 tumor suppressor gene is the primary target for RCC in rodents. A hereditary form of RCC occurs in a line of Long-Evans rats carrying a mutation in the Tsc-2 gene (Tsc-2Ek/+; see Ref. 34). The Eker mutation, a result of a retroviral insertion event in the Tsc-2 gene, predisposes these animals to a high frequency of spontaneous and carcinogen-induced RCC (35, 36, 37, 38), which develop in these animals subsequent to loss of the wild-type Tsc-2 allele (39, 40). Tsc-2 knockout mice similarly develop spontaneous RCC (41, 42) in contrast to Vhl knockout mice, which do not develop these tumors (43, 44). In addition, carcinogen-induced rat RCC also exhibits mutations in the Tsc-2 gene (40, 45, 46). Phenotypically, murine RCC differs from its human counterpart in that these tumors are predominantly chromophilic and, although predominantly solid, often have a prominent cystic component (47, 48). RCC that develops in the Eker rat expresses wild-type Vhl (49, 50), and Vhl mutations in rat RCC are uncommon, although when they occur, they are associated with rare RCC with clear-cell cytology (51).
To determine whether dysregulation of HIF1α/2α and VEGF expression occurs in RCC resulting from inactivation of the Tsc-2 tumor suppressor gene, we examined HIF1α/2α expression and activity in normal and neoplastic rat renal cells under normoxic and hypoxic conditions. Cells derived from Tsc-2 null rat RCC showed constitutively high expression of HIF2α and HIF transcriptional activity. Furthermore, HIF2α accumulation in primary tumors correlated with up-regulation of VEGF and tumor vascularization.
MATERIALS AND METHODS
The hypoxia mimetics DFX and CoCl2 and proteasomal protease inhibitor ALLN were from Sigma (St. Louis, MO). Mouse monoclonal anti-HIF1α (NB100-105, clone H1α67), rabbit polyclonal anti-HIF2α (NB100-122) antibody (Novus, Littleton, CO), and rabbit polyclonal anti-VEGF (Santa Cruz, Santa Cruz, CA) were used for Western analysis. For reporter assays, luciferase reporter constructs PL949 (a generous gift from Dr. Chris Bradfield, Madison, WI) and HRE-Luc with six or three tandem copies of HRE of the erythropoietin gene were used. pGL3 promoter vector and pRL-TK Renila constructs were obtained from Promega (Madison, WI).
Cells Lines and Culture Conditions.
The ERC15 and ERC18 cell lines were derived from Eker rat RCC (38). As described previously, neither of them expresses the Tsc-2 gene product tuberin (52). The TRKE2 cell line was established by in vitro transformation of kidney epithelial cells with N-methyl-N′-nitro-N-nitrosoguanidine (53) and has no mutations within the Tsc-2 coding region (unpublished data). EKT2 carries two copies of the mutant Tsc-2 allele (Tsc-2EK/EK) and does not express tuberin (52). ERC15, ERC18, EKT2, and TRKE2 were maintained as described previously (52). The human renal tumor cell lines A498, 786-O derived from clear-cell RCCs, and 112, ACHN derived from non-clear-cell RCC were maintained as previously described (54).
Cells were harvested in TEN (40 mm Tris-HCl, 10 mm EDTA, 150 mm NaCl) buffer and then lysated in cell extraction WCE buffer (10 mm HEPES, 400 mm NaCl, 0.1 mm EDTA, 5% glycerol) with protease inhibitors. After centrifugation, the supernatant containing the total cell lysate was quantitated. Proteins separated in SDS-polyacrylamide gels were transferred onto polyvinylidene difluoride membranes, and blocked for 1 h in 5% milk with washing buffer (Tris 5 mm, NaCl 35 mm, 0.1% Tween 20). Primary antibodies were hybridized for 2 h in 2% milk for 2 h at room temperature. A secondary antibody conjugated to horseradish peroxidase (1:2000 dilution) was hybridized for 1 h in 2% milk with washing buffer. After extensive rinses with washing buffer, the complexes were visualized using KPL LumiGLO (KPL, Gaithersburg, MD). Tumors and normal tissue samples were pulverized under liquid N2 using mortar and pestle and processed as above.
Luciferase Reporter Assays.
Twenty-four-well plates were seeded with 10,000 cells/well and grown overnight in media containing 10% serum. For dual-reporter assay, a Fugene 6 Transfectant Reagent Kit (Roche, Indianapolis, IN) was used to transfect cells with pGL3 promoter vector or PL949 construct together with pRL-TK Renila construct, according to the manufacturer’s instructions. Alternatively, cells were cotransfected with HRE-Luc and pCMV-β-galactosidase construct as a control. After 24 h of transfection, cells were treated with 250 μm DFX for 6 h. A Microtiter Plate Luminometer (Thermo Lab Systems, Helsinki, Finland) was used to determine the Firefly and Renila luciferase activity using a Dual-Luciferase Reporter Assay System (Promega, Madison, WI), or Firefly and β-galactosidase activity with a luciferase assay kit by Tropix (Bedford, MA).
Primers U-14 (5′ CTGCTCTCTTGGGTGCACTGG 3′) and L-556 (5′ CACCGCCTTGGCTTGTCACAT 3′) were designed to discriminate the various isoforms of rat VEGF. cDNA from renal cell carcinomas and normal kidney were synthesized by reverse transcription of total RNA. Reverse transcription was performed at 42°C for 60 min. VEGF cDNA fragments were amplified by 30 rounds of PCR at 1 min at 95°C, 1 min at 58°C, 1 min at 72°C with a Perkin-Elmer Thermal Cycler (Perkin-Elmer, Boston, MA), and Taq DNA polymerase (Applied Biosystems, Foster City, CA).
Rat RCC and normal kidney samples were extracted with WCE buffer (10 mm HEPES, 400 mm NaCl, 9.1mMEDIA, 5% glycerol, protease inhibitors) and protein concentration quantified using BCA Protein Assay kit (Pierce, Rockford, IL). The VEGF content per 10 μg of total protein was determined using the VEGF ELISA assay kit (R&D Systems, Minneapolis, MN) according to manufacturer’s instructions.
RCC specimens were fixed with 10% formalin and embedded in paraffin according to routine procedures. Formalin-fixed, paraffin-embedded sections of the tumor tissue were examined immunohistochemically with goat polyclonal anti-CD-31 antibody (Santa Cruz, Santa Cruz, CA). Immunoreactive products were visualized with 3,3′-diaminobenzidine (Sigma, St. Louis, MO). Light microscopy was used to visual vessels and micrographs were taken at ×100 magnification (×10 objective and ×10 ocular).
HIF1α and HIF2α Expression.
Loss of VHL in RCC has been reported to lead to accumulation of HIF1α/2α because of the abrogated ubiquitin mediated degradation of HIF1α/2α. To confirm these data and establish conditions for using hypoxia mimetics DFX and CoCl2, a panel of RCC cell lines derived from clear-cell (mutant VHL) and non-clear-cell (wild-type VHL) tumors was examined for expression of HIF1α/2α under normoxic or hypoxic conditions (Fig. 1 A). In VHL-null 786-O and A498 cells, HIF1α could not be detected, but HIF2α was constitutively stabilized under both normoxic and hypoxic conditions. VHL wild-type cells, ACHN and 112, did not express HIF1α or HIF2α under normoxia, but accumulated both HIF1α and HIF2α in response to either 250 μm DFX or 250 μm CoCl2 treatment. Treatment with the proteasome inhibitor ALLN did not result in stabilization of HIF2α in ACHN or 112 cells, although a slight stabilization of HIF1α was observed in these cells, primarily in ACHN cells, which was consistent with a previous report that HIF1α is regulated under normoxic conditions through the proteasome mediated pathway (5).
We next analyzed HIF1α and HIF2α expression in a series of rat renal epithelial cell lines containing wild type Vhl (49) under normoxic and chemically induced hypoxic conditions. Under normoxic conditions, Eker rat tumor-derived cell lines (Tsc-2Ek/−) ERC15, ERC18, homozygous mutant (Tsc-2Ek/Ek) EKT2 and TRKE2 (Tsc-2+/+) cell lines demonstrated undetectable or low levels endogenous HIF1α expression (Fig. 1,B). After treatment with 250 μm DFX or 250 μm CoCl2, HIF1α accumulation increased dramatically in ERC18, TRKE2, and EKT2, whereas HIF1α expression in ERC15 remained undetectable. Treatment with 200 μm ALLN for 6 h also stabilized HIF1α protein in ERC18, TRKE2, and EKT2 cell lines, suggesting ubiquitin-mediated degradation of HIF1α occurred under normoxic conditions in these cells (Fig. 1,C). Primary RCC from Eker rats and normal kidney were also analyzed for HIF1α expression, and as shown in Fig. 1 D, HIF1α expression was undetectable in all six tumors and in normal kidney.
In contrast to HIF1α, under normoxic conditions, HIF2α was stabilized in ERC15, ERC18, EKT2, and TRKE2 cells as shown in Fig. 1,B. Treatment with 250 μm DFX or 250 μm CoCl2 did not significantly increase levels of HIF2α compared with normoxia, although ALLN treatment for 6 h did increase levels of HIF2α moderately in these cells, suggesting that further HIF2α stabilization could be achieved by blocking proteasome-mediated degradation (data not shown). In addition to these cell lines, five of six primary Eker rat RCCs also exhibited stabilization of HIF2α, whereas none of the six normal kidney samples examined showed detectable HIF2α expression (Fig. 1 D).
HIF Transcriptional Activity.
To confirm the transcriptional activity of stabilized HIF2α in rat RCC cells, luciferase reporter assays were performed under normoxic and hypoxic conditions. Constitutively high levels of HIF activity were observed in ERC15,TRKE2, and A498 (VHL−) cell lines, which was consistent with the endogenous high level of HIF2α expression detected by Western analysis (Fig. 1, A and B). As shown in Fig. 2,A, under normoxic conditions, high levels of HRE-driven luciferase activity were observed with PL949 in ERC15, TRKE2, and A498 lines ranging from 3.2–4.5-fold relative to control cells transfected with pGL3, in contrast to ACHN cells (wild-type VHL), which exhibited no significant increase relative to controls under normoxic conditions. Similarly, high levels of HRE-Luc activity were observed in ERC15, TRKE2, and A498 cells ranging from 5–22-fold relative to control under normoxic conditions with HRE-Luc construct (data not shown). However, as expected, after treatment with 250 μm DFX for 6 h, ACHN cells exhibited a >5.5-fold increase in HIF-mediated luciferase activity with PL949 relative to normoxic conditions as shown in Fig. 2,B. In contrast, ERC-15, TRKE2, and A498 exhibited no significant increase in HIF luciferase activity in response to treatment with 250 μm DFX relative to high normoxic levels of HIF activity. A slight increase in HIF activity was observed in TRKE2 cells under hypoxic condition, probably as a result of the accumulation of HIF1α that occurs in these cells (Fig. 1 B).
Expression of VEGF in RCC.
To determine whether the stabilization of HIF2α in Tsc-2-null RCC was associated with up-regulation of VEGF and angiogenesis, we next characterized VEGF expression in RCC-derived cell lines and primary tumors. Three isoforms of VEGF were detected in RCC and normal kidney by RT-PCR using isoform-specific primers as shown in Fig. 3,A. RCC-derived ERC cell lines expressed the 188-, 164-, and 120-aa-specific isoforms of VEGF, although the 188-aa isoform was barely detectable in these cells as shown in Fig. 3,B. By Western analysis, three VEGF isoforms could be detected in the total cell lysates of ERC15, ERC18, TRKE2, and EKT2 cells (Fig. 3,C). In seven of seven primary RCCs examined, VEGF was also expressed, although only a single VEGF isoform was present, which was highly expressed relative to normal kidney (Fig. 3,D). VEGF protein expression was confirmed by ELISA assay to quantitate VEGF production in tumors and normal kidney. Primary rat RCC (n = 6) contained 10.2 ± 3.1 pg of VEGF/μg of protein, whereas normal kidney (n = 6) contained 0.78 ± 0.34 pg of VEGF/μg of protein (data not shown). Consistent with VEGF overexpression, Eker rat RCC were also highly vascularized, exhibiting larger and more irregular vascular area as compared with the normal kidney (Fig. 4).
In this study, we have determined that the transcription factor HIF2α is stabilized in rat RCC expressing the wild-type VHL tumor suppressor gene. Primary RCC arising in Eker rats and RCC-derived cell lines expressed HIF2α had constitutively high HIF activity, exhibited VEGF accumulation, and were highly vascularized.
Angiogenesis and extensive tumor vascularization are characteristic of clear-cell RCC, a fact that has been attributed to stabilization of HIF1α/2α and expression of the proangiogenic growth factor VEGF resulting from loss of VHL function (55). In the present study, using the Eker rat model, Tsc-2 tumor suppressor gene defects and development of RCC also resulted in HIF stabilization and up-regulation of VEGF. VEGF is a specific mitogen and survival factor for endothelial cells and a vital promoter of angiogenesis physiologically and pathologically. Increased expression of VEGF in RCC has been demonstrated previously (56). Interestingly, VEGF189 isoform in human RCC has been shown to be tightly correlated with the stage and vascularization status of this disease (57). Tumor size is also significantly correlated with VEGF189 isoform expression, with tumors expressing the VEGF189 isoform being more vascularized than those lacking this isoform (57). Rodent VEGF isoforms are one aa shorter than their human homologs (58), and we identified three isoforms of VEGF at the mRNA level in Eker rat RCC-derived cell lines corresponding to the 188-, 164-, and 120-aa isoforms, which were also expressed in primary tumors. In all cases, RNA expression in tumors and cell lines correlated with the up-regulation of VEGF protein as determined by ELISA assay and Western analysis. These data indicate that in RCC associated with loss of Tsc-2 tumor suppressor gene function, stabilization of HIF2α correlates with tumor angiogenesis.
Rat RCC differs histologically from the human disease, having primarily a chromophilic rather than clear-cell cytology (47, 48). These tumors often have a prominent cystic component and can also have a tubulopapillary appearance. Expression of HIF1α and -2α in human RCC is tightly correlated with the clear-cell variant, particularly HIF2α expression (31). Whereas 17 of 17 clear-cell tumors exhibited stabilization of HIF2α, none of the papillary, chromophobe, or collecting duct tumors examined expressed this gene (31). Whereas human RCC of the clear-cell type results primarily from alterations in the VHL tumor suppressor gene, murine RCC results from loss of function of the Tsc-2 tumor suppressor gene. However, murine and human RCC have the same cell of origin, the proximal tubule epithelial cell of the renal nephron (47, 48). Therefore, expression of HIF2α in both VHL− human and Tsc-2− rat tumors represents a point of etiologic convergence between tumor development via these two pathways. That TSC-2- and VHL-mediated tumorigenic pathways intersect is further supported by the finding that clear-cell RCC in tuberous sclerosis patients have defects in TSC-2 but retain wild-type VHL (59). Dissecting pathways involved in TSC-2- and VHL-related RCC might provide important information linking the molecular events that participate in the development of different subsets of RCC.
The VHL tumor suppressor protein targets HIFα subunits for ubiquitin-mediated proteolysis, and in RCC bearing inactivating mutations in both VHL alleles, HIFα subunits are stabilized and accumulate at high levels irrespective of oxygen levels. Tsc-2-associated tumors and tumor-derived cell lines express wild-type VHL (49, 50), however, suggesting that mechanism(s) other than loss of Vhl function are responsible for HIF stabilization. Interestingly, the TRKE-2 cell line, which expresses wild-type tuberin, the product of the Tsc-2 gene, also exhibited HIF2α stabilization, suggesting either that loss of tuberin is not directly responsible for HIF stabilization in Tsc-2-null cells, or that in the TRKE cell line, which was transformed in vitro by MNNG (53), contains other alterations in the cellular pathway(s) in which tuberin participates. Alternate pathways for HIFα stabilization other than loss of VHL function include loss of p53 function and cell-signaling via PI3K (60). Interestingly, although tuberin-null tumors and cell lines contain wild-type p53 (61, 62), tuberin has recently been shown to be a critical down-stream regulator of PI3K signaling (63). This suggests the possibility that stabilization of HIFα in tuberin-null cells may be a result of disrupted PI3K signaling downstream from AKT.
VHL-null clear-cell RCC and cell lines exhibit stabilization of HIF2α. Recent data have suggested functional differences in the role of HIF1α and -2α in RCC. These two proteins share 48% homology and a common VHL-mediated degradation pathway. Both of these proteins form heterodimers with HIF1β and bind to the same HRE in the promoter region of target genes to transactivate target gene expression. However, differences do exist in their regulation and function. Structurally, the two proteins differ at a critical cysteine residue, which determines HIF2α susceptibility to redox regulation (64). HIF1α and -2α also exhibit different tissue distributions. HIF1α mRNA is ubiquitously expressed, whereas HIF2α message is expressed most abundantly in the highly vascularized adult tissues, such as lung, heart, liver, and placenta and endothelial cells of the embryonic and adult mouse (6, 7). Targeted disruption of these genes in mice also leads to different phenotypes (65, 66). In our experiments, both HIF2α (ERC-15, ERC-18) and HIF1α (ERC-18) accumulated in response to the proteasome inhibitor ALLN, indicating the existence of an intact proteolysis pathway in Tsc-2-null RCC derived cells, although cell line specificity was observed. Stabilization specifically of HIF2α, rather than HIF1α, under normoxic conditions in these cells suggests that different regulatory mechanisms may exist for the stabilization of HIF1α and HIF2α. Recently, HIF2α has been shown to promote tumor growth, whereas overexpression of HIF1α alone in VHL wild-type RCC-derived cells does not reproduce the tumorigenic phenotype (67, 68), suggesting that of the two, HIF2α is the more relevant target for tumorigenesis. Furthermore, the phenotype of tumors in which HIF stabilization was induced by expression of a peptide containing the oxygen-dependent degradation domain of HIF exhibited a poorly differentiated histology, distinct from that of clear-cell type (67), suggesting that VHL function(s) other than stabilization of HIF are responsible for the clear-cell histological phenotype. Specific stabilization of HIF2α, but not HIF1α in spontaneous RCC exhibiting loss of the Tsc-2 tumor suppressor gene with a chromophilic, rather than clear-cell histology, would support both these hypotheses. The identification of these two different pathways for regulation of HIF in RCC should facilitate future studies aimed at understanding the role of this transcription factor in tumorigenesis and mechanism(s) for selective stabilization of HIF2α versus HIF1α.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported in parts by NIH Grants CA63613 and ES08263 (to C. L. W.) and ES07784.
The abbreviations used are: PAS, Per-Arnt-Sim; VEGF, vascular endothelial growth factor; HRE, hypoxia response element; VHL, von Hippel-Lindau; Tsc-2, tuberous sclerosis complex-2; DFX, desferrioxamine; ALLN, Ac-LLnL-CHO; RT, reverse transcriptase; aa, amino acid(s).
We thank Dr. C. Bradfield for the PL949 luciferase construct, Dr. J. Gnarra for helpful discussions, and B. Brooks and S. Henninger for manuscript preparation.