In vitro and In vivo Models Analyzing von Hippel-Lindau Disease-Specific Mutations

Transcription

1 [CANCER RESEARCH 64, , December 1, 2004] In vitro and In vivo Models Analyzing von Hippel-Lindau Disease-Specific Mutations W. Kimryn Rathmell, 1,3 Michele M. Hickey, 1,2 Natalie A. Bezman, 1 Christie A. Chmielecki, 3 Natalie C. Carraway, 3 and M. Celeste Simon 1,2 1 Abramson Family Cancer Research Institute and the Department of Cell and Molecular Biology, and 2 Howard Hughes Medical Institute, University of Pennsylvania, Philadelphia, Pennsylvania; and 3 Lineberger Comprehensive Cancer Center, University of North Carolina, Chapel Hill, North Carolina ABSTRACT Mutations in the von Hippel-Lindau (VHL) tumor suppressor gene cause tissue-specific tumors, with a striking genotype-phenotype correlation. Loss of VHL expression predisposes to hemangioblastoma and clear cell renal cell carcinoma, whereas specific point mutations predispose to pheochromocytoma, polycythemia, or combinations of hemangioblastoma, renal cell carcinoma, and/or pheochromocytoma. The VHL protein (pvhl) has been implicated in many cellular activities including the hypoxia response, cell cycle arrest, apoptosis, and extracellular matrix remodeling. We have expressed missense pvhl mutations in Vhl / murine embryonic stem cells to test genotype-phenotype correlations in euploid cells. We first examined the ability of mutant pvhl to direct degradation of the hypoxia inducible factor (HIF) subunits HIF1 and HIF2. All mutant pvhl proteins restored proper hypoxic regulation of HIF1, although one VHL mutation (VHL R167Q ) displayed impaired binding to Elongin C. This mutation also failed to restore HIF2 regulation. In separate assays, these embryonic stem cells were used to generate teratomas in immunocompromised mice, allowing independent assessment of the effects of specific VHL mutations on tumor growth. Surprisingly, teratomas expressing the VHL Y112H mutant protein displayed a growth disadvantage, despite restoring HIF regulation. Finally, we observed increased microvessel density in teratomas derived from Vhl / as well as VHL Y112H, VHL R167Q, and VHL R200W embryonic stem cells. Together, these observations support the hypothesis that pvhl plays multiple roles in the cell, and that these activities can be separated via discrete VHL point mutations. The ability to dissect specific VHL functions with missense mutations in a euploid model offers a novel opportunity to elucidate the activities of VHL as a tumor suppressor. INTRODUCTION Mutations in the von Hippel-Lindau (VHL) tumor suppressor gene give rise to a variety of tumors and other abnormalities in affected individuals including renal clear cell carcinoma, pheochromocytoma, hemangioblastoma, and polycythemia (1). Some of these disease manifestations derive from the unregulated expression of the hypoxia inducible factor (HIF) transcription complex. HIF is a heterodimer composed of a highly regulated subunit (either HIF1 or HIF2 / EPAS1) and a ubiquitously expressed subunit (HIF1 /ARNT; refs. 2 4). This complex, in cooperation with the coactivators campresponsive element binding protein and p300, activates transcription via a hypoxia response element in the promoter or enhancer regions of many hypoxia-responsive genes. Genes transcriptionally activated by HIF include factors involved in angiogenesis, glucose transport, glycolysis, and erythropoiesis as recently reviewed (5 7), and lipid Received 4/22/04; revised 9/8/04; accepted 10/1/04. Grant support: NIH Grant K08-CA and a V Foundation Scholar Award (W. Rathmell), Howard Hughes Medical Institute Predoctoral Fellowships and the Ruth L. Kirschstein National Research Service Award (M. Hickey and N. Bezman), and NIH Grant HL66130 and the Howard Hughes Medical Institute (M. Simon). 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. Requests for reprints: M. Celeste Simon, University of Pennsylvania, Abramson Family Cancer Research Institute, BRB II/III, Room 450, Philadelphia, PA Phone: (215) ; Fax: (215) ; celeste2@mail.med.upenn.edu American Association for Cancer 8595 transport and droplet formation (8, 9). Unregulated expression of the full complement of hypoxia response genes is surmised to contribute much of the clinical and pathological phenotype of renal cell carcinoma, which is characterized as a highly vascular, glycolytic, lipidrich tumor that can be associated with polycythemia (10, 11). Detailed studies of the effects of VHL loss have identified the regulation of the hypoxia response pathway via the proteasomal degradation of HIF1 and HIF2 as a major activity of the VHL protein (pvhl; refs ). The pvhl acts as the substrate receptor for an E3 ubiquitin ligase complex and targets HIF1 and HIF2 for degradation under normal oxygen (O 2 ) tension (16). The O 2 -dependent regulation of HIF (referring to either HIF1 or HIF2 ) is imparted by posttranslational hydroxylation of specific prolines located in the HIF O 2 - dependent degradation domain by HIF prolyl hydroxylases (17 19). When O 2 levels drop, HIF1 and HIF2 accumulate (5). In the absence of pvhl, HIF1 and HIF2 are maximally stabilized in the presence of atmospheric O 2 levels and show no additional O 2 - dependent accumulation (13). We previously reported that Vhl / murine embryonic stem cells predictably disrupt the hypoxia response pathway (20). In Vhl / embryonic stem cells, HIF1 and HIF2 are highly expressed under normoxic conditions, and protein levels are not additionally stabilized in response to O 2 deprivation. Furthermore, Vhl deletion gives rise to highly glycolytic cells, as would be expected in cells with elevated levels of glycolytic enzymes. When Vhl / embryonic stem cells are grown in a teratoma model system the tumors display a highly vascular phenotype, although with a surprising reduction in tumor volume (20). The spectrum of tumors arising in individual VHL patients strongly correlates with specific VHL mutations (21, 22), and the syndrome has been divided into type 1 and type 2 disease based on predisposition for pheochromocytoma (Fig. 1A). Type 1 disease is characterized by mutations, which delete, silence, or destabilize the VHL gene, and this patient group has a low frequency of pheochromocytoma (23). Type 2 disease is characterized by a strong predisposition for pheochromocytoma, and it is additionally subdivided into 2A (pheochromocytoma, hemangioblastoma, and low risk for renal cell carcinoma), 2B (pheochromocytoma, hemangioblastoma, and high risk for renal cell carcinoma), and 2C (pheochromocytoma only). Molecular analysis of these mutations shows that type 2A mutants and type 2C mutants maintain interaction with the ubiquitin ligase complex, whereas 2B mutants are deficient in this interaction (23). In addition, homozygosity for a VHL mutation in the extreme COOH-terminal domain has been associated with a highly penetrant congenital form of polycythemia termed Chuvash Polycythemia (24, 25). No tumor predisposition has yet been identified with this mutation. In addition to VHL disease, mutations in VHL have been identified in the majority of patients with sporadic clear cell renal cell carcinoma (26 28), which constitutes the majority of carcinomas of the kidney. This form of renal cell carcinoma is characterized by a highly vascular cystic or solid mass comprised of nests of tumor cells containing abundant clear cytoplasm consisting of glycogen and neutral lipid. Although loss of VHL is not a transforming event in such cells, VHL

2 Fig. 1. Embryonic stem cell expression of representative VHL missense mutations. A. VHL disease subtypes, characteristic patterns of disease, and mutations used in this analysis to represent each subtype of VHL disease. B, schematic of VHL gene. The hemagglutinin tag is located at the NH 2 terminus as indicated. Important residues are identified, as are the known interaction domains. C, human pvhl immunoblot. Protein extracts from embryonic stem cells expressing the indicated version of pvhl, pvhl null, and endogenous mouse pvhl were compared with in vitro translated human pvhl protein. Note the lack of signal in the endogenous pvhl extract from Vhl / extracts. D, murine pvhl immunoblot. Protein extracts from pvhl wild-type, heterozygous, and null embryonic stem cells. (HA, hemagglutinin) 8596 mutations have been identified in renal cysts and dysplastic lesions, strongly suggesting they represent an early event in renal tumorigenesis (29). Furthermore, renal cell carcinoma cells regain O 2 -regulated gene expression (15, 30) and lose their transformed phenotype on reintroduction of wild-type pvhl activities (31, 32). Whereas VHL clearly regulates the hypoxia response pathway, the cohort of mutations giving rise to type 2C disease has not been found to affect HIF regulation (33). In fact, VHL disease-associated pheochromocytoma has been speculated to develop as a result of a gain of function of pvhl, as this tumor type is almost exclusively observed in the setting of missense mutations (34). Furthermore, several activities have been ascribed to pvhl in addition to the regulation of the hypoxia response pathway and were recently reviewed (35), including potential function in cell cycle regulation (36 38), cell growth (39, 40), cytoskeletal structure (41, 42), and extracellular matrix deposition (43). In this report we take advantage of genotype-phenotype correlation, and we have complemented Vhl / embryonic stem cells with a selected group of mutations representing the various subtypes of VHL disease, including Chuvash Polycythemia, to dissect potentially distinct activities of pvhl that are intrinsic to tissue-specific tumor types seen in the hereditary disease. Although this system is not an absolute recreation of a tumor environment, this model system offers several tangible advantages for the study of pvhl function. First, embryonic stem cells permit examination of pvhl activity in euploid cells free of the competing mutations intrinsic to tumor cell lines. Second, embryonic stem cells permit the examination of three-dimensional growth in a teratoma model system. Finally, embryonic stem cells have uses beyond the scope of this report, including multiple avenues of in vitro differentiation and the opportunity to generate chimeric animals to model VHL disease. Previous mouse models of VHL disease have been hampered by lethality or severe vascular phenotypes (44 47). Using missense mutations designed to represent distinct disease patterns and exploit the effects of limited disruption of pvhl activity, we propose to discriminate the activities of various VHL missense mutations in a variety of biological and functional assays. In the context of this unique model, we observed, surprisingly, that all of the mutant forms of pvhl restored HIF1 hypoxic regulation. Additionally, the mutation associated with VHL type 2B disease (VHL R167Q ) uniquely conferred impaired interaction with Elongin C as well as a disruption of HIF2 hypoxia responsive regulation, whereas HIF1 protein regulation remained intact. Additionally, a mutant associated with VHL type 2A disease (VHL Y112H ) showed a suppressive growth effect similar to that observed with Vhl loss in teratomas. Consistent with previous observations, we detected no impact on the hypoxia response pathway by a mutant associated with VHL type 2C disease (VHL L188V ), in which patients are only susceptible to pheochromocytoma (33). All of the mutants in this study showed defective deposition of fibronectin, extending previous observations that extracellular matrix remodeling is perturbed by most pathologically significant VHL mutations. Finally, the mutation associated with the newest subtype of VHL disease, Chuvash Polycythemia (VHL R200W ), conferred regulation of the hypoxia response pathway. These results show the use of a novel model system able to distinguish the effects of pathogenic mutations associated with different manifestations of VHL disease. These findings provide additional support for the hypothesis that pvhl has multiple cellular activities, and that activities in addition to the regulation of HIF1 are integral to VHL-mediated tumor cell growth and tumor vascularization. MATERIALS AND METHODS Cell Lines. All of the embryonic stem cells were cultured in DMEM-H (Invitrogen, Carlsbad, CA) supplemented with 15% FBS, 1% nonessential amino acids, 2% L-glutamine, 1% penicillin-streptomycin, leukemia inhibitory factor (ESGRO, Life Technologies, Inc., Rockville, MD), and 2-mercaptoethanol (Invitrogen). Embryonic stem cells were derived from a Vhl / clone described previously (20) by transfection of linearized vector containing an hemagglutinintagged human VHL cdna construct (a generous gift of Dr. William G. Kaelin, Harvard Medical School, Boston, MA) driven by an EF1 promoter containing either wild-type VHL sequence or the following point mutations: Tyr112His, Arg167Gln, Leu188Val, or Arg200Trp. Generation of missense mutations was done with a PCR-based method described previously (48) with the primer pairs (Y112Hf, CACGAGGTCACCTTTGGCTCTTCAGAG; Y112Hr: GCTGTG- GATGCGGCGGCCCGTG; R167Qf, AGAGCCTAGTCAAGC-CTGAGAA- TTA; R167Qr, GGACAACCTGGAGGCATCGCTCTTT; L188Vf, GTGGAA-

3 GACCACCCAAATGTGCAGA; L188Vr, ATCTTCGTAGAGCGACCTGA-C- GATG; R200Wf, TGGCTGACACAGCGCATTGCAC; and R200Wr, CTC- CAG-GTCTTTCTGCACATTTGGG). Clones were grown in hygromycin selection media and screened by Southern blotting for integration. The construct is depicted in Fig. 1B, which also shows the location of each point mutation. Immunoblot Analysis and Immunoprecipitation. Immunoblots for HIF1 and HIF2 were done as described previously (20). Whole cell extracts were prepared by lysis in 1% NP40 lysis buffer and quantified by BCA protein assay (Pierce, Rockford, IL) for the HIF1 analysis. Nuclear extracts were prepared by a 400 mmol/l NaCl extraction and quantitated by Bradford protein assay (Bio-Rad, Hercules, CA) for the HIF2 analysis. Twenty-five micrograms of protein were loaded for the VHL immunoblots, 50 g of protein was loaded for the HIF immunoblots. Coimmunoprecipitation was done with the ProFound Mammalian hemagglutinin tag IP/CoIP kit according to the manufacturer s specifications (Pierce). Separation was done on 8 to 10% SDS polyacrylamide gel and transferred to ECL nitrocellulose (Amersham, Piscataway, NJ). Equal loading and even transfer was confirmed by reversible staining with Ponceau S. Primary antibodies were used at a dilution of 1:1,000 for murine pvhl (Santa Cruz Biotechnology, Santa Cruz, CA), 1:2,000 for HIF1 (Cayman Chemical, Ann Arbor, MI), 1:2,000 for HIF2 (Novus Biologicals, Littleton, CO), 1:1,000 for human pvhl (BD PharMingen, San Diego, CA), and 1:500 for Elongin C (Santa Cruz Biotechnology). Secondary antibodies were horseradish peroxidase conjugated, detected by ECL (Amersham), and exposed to autoradiograph film (Eastman Kodak, Rochester, NY). Scanning densitometry was done with Scion software (Scion, Frederick, MD). Electrophoretic Mobility Shift Assay. Gel shift analysis was done for HIF1 on nuclear extracts with a radiolabeled probe incorporating the hypoxia response element of the erythropoietin promoter as described previously (20). Experiments were also done with vascular endothelial growth factor (VEGF) hypoxia response element sequence with identical results. HIF1 antibody (Novus Biologicals) was added to control lanes to show HIF1 participation in the shifted complex. The bands were detected with phosphoimager analysis (Molecular Dynamics, Sunnyvale, CA). ELISA. Secretion of soluble VEGF was detected by ELISA (R&D Systems, Minneapolis, MN), according to the manufacturer s instructions. Conditioned media was analyzed after 16-hour incubation of linearly growing cells at either 1.5% O 2 or ambient O 2 tension. Duplicate analysis was done on triplicate cultures for all of the sample conditions. Colorimetric change was measured at 495 nmol/l with a Becton-Dickinson ELISA plate reader. Northern Blot Analysis. Northern analysis of total RNA was done on samples prepared after 16 hours of either hypoxia (1.5% O 2 ) or normoxia treatment. RNA was prepared in TRIzol according to the manufacturer s recommendations (Life Technologies, Inc.). Samples were separated by 1% denaturing agarose gel electrophoresis and passively transferred to Hybond n nitrocellulose membrane (Amersham) in 10 SSC. Blots were probed with RNA specific oligos for VEGF, phosphofructokinase, aldolase A, and -actin labeled with 32 P by High Prime (Boehringer Mannheim, Mannheim, Germany). After UV fixation with 1200 Joules (Stratalinker; Stratagene, La Jolla, CA), the membrane was probed sequentially with radiolabeled probes after stripping in boiling 0.5% SDS. Signals were detected by phosphoimager analysis (Molecular Dynamics). Teratoma Analysis. For the generation of teratomas, cells in 100 L sterile PBS were injected s.c. between the shoulder blades of NIH III immunodeficient mice (Jackson Laboratories, Bar Harbor, ME, and Charles River Laboratories, Wilmington, MA). The tumors were permitted to grow for 21 days and were monitored for tumor growth. The tumors were then harvested, weighed, formalin fixed, and embedded in paraffin. Paraffin sections were stained by standard immunohistochemistry techniques with the following primary antibodies: Ki67 (Novocastra Laboratories, Ltd., Newcastle upon Tyne, UK), cleaved caspase 3 (Cell Signaling, Beverly, MA), Fibronectin (Cell Signaling), VEGF (Neomarkers, Fremont, CA), CD34 (Cal Biochemicals, Darmstadt, Germany), CD31 (Neomarkers), and VWF (Dako, Carpinteria, CA). Secondary detection was done with biotinylated mouse, rabbit, or goat specific antibodies, ABC enhancement kit (Vector, Burlingame, CA), and DAB detection reagent (Vector). Dilute hematoxylin, or cresol green in the case of CD31 stain, was used for counterstain (Vector) RESULTS The pvhl Expression in Vhl / Embryonic Stem Cells. Vhl / embryonic stem cells generated as described previously (20) were transfected with plasmid encoding a hemagglutinin-tagged human wild-type or mutant VHL cdna representing each subtype of VHL disease (Fig. 1, A and B). Integration was confirmed by Southern blotting (data not shown). Immunoblots were done to show that the selected clones expressed levels of human pvhl protein (Fig. 1C) comparable with that of endogenous expression of the murine protein in wild-type or heterozygous embryonic stem cells (Fig. 1D). Fig. 1D shows expression of endogenous pvhl in wild-type and heterozygous clones as detected by a murine pvhl-specific antibody. Fortyeight transgenic clones were isolated for each mutation, and clones with matched minimal levels of expression were selected to approximate the expression of endogenous murine pvhl and to avoid effects of overexpression (Fig. 1, C and D). Regulation of HIF1 Expression Is Restored by pvhl Mutant Proteins. Four representative clones of each VHL disease subtype were analyzed for expression of HIF1 protein after growth for 4 hours at normoxia (21% O 2 ) or hypoxia (1.5% O 2 ). Extracts from Vhl / cells had low normoxic HIF1 expression, which increased after hypoxia treatment. In contrast, Vhl / cells showed high normoxic levels of HIF1 without additional induction after hypoxia (Fig. 2A). Reconstitution of the Vhl / cells with wild-type human VHL cdna (VHL wt ) restored normoxic suppression and hypoxic induction of HIF1 expression. Of note, each of the mutants also restored normoxic regulation of HIF1. This observation was confirmed in multiple independently derived clones encompassing a wide range of expression with minor clonal variation (data not shown). This somewhat surprising result is in contrast to previous evaluations of the VHL Y112H and VHL R167Q mutations, which exhibited dysregulation of HIF1 in human renal cell carcinoma cell cultures (23). Furthermore, VHL R200W has been correlated with a subtle increase in normoxic HIF1 expression but preserved hypoxic induction in human cells (24). However, our results are consistent with previous observations that type 2C mutations (including VHL L188V ) had no effect on hypoxic regulation of HIF1 (33). To confirm the observation that HIF1 regulation was restored in the VHL mutant clones and assay levels of active HIF1, nuclear extracts from two independently derived clones of each mutant were tested for binding activity to an hypoxia response element derived from the erythropoietin promoter sequence. The wild-type complemented clones, and each of the mutants restored regulation of HIF1 DNA binding activity (Fig. 3C). These results were confirmed with a second radiolabeled probe containing the hypoxia response element derived from the VEGF promoter sequence (data not shown). Antibody directed against HIF1 was added to control lanes to show participation of HIF1 in hypoxia response element binding by supershifting the complex. Each of the oligonucleotides used for EMSA are specific for HIF1 binding, 4 and addition of an antibody to HIF2 did not result in a mobility shift of the DNA binding complex (data not shown). These results suggest that hypoxia response element binding assays exclusively detect HIF1 complex formation (and not HIF2 complexes). Although embryonic stem cells express adequate levels of HIF2 protein, it is nonfunctional as a DNA binding protein (8). HIF2 Expression Is Dysregulated in Type 2B Mutant Clones. Embryonic stem cells complemented with VHL wt or the VHL L188V and VHL R200W mutants suppressed HIF2 protein expression under normoxic conditions and produced increased levels of HIF2 on hypoxic 4 B. Keith and M. C. Simon, unpublished observations.

4 Fig. 2. Effects of VHL mutation on HIF1 and HIF2 expression and interactions with Elongin C. A, HIF1 immunoblot. The top of two bands illustrated shows HIF1 protein; the bottom is a nonspecific band of slightly smaller size. Paired lanes represent extracts prepared from embryonic stem cells expressing the indicated form of pvhl. B, HIF2 immunoblot. Each panel represents two independently derived embryonic stem cell clones expressing the form of pvhl indicated at the right. *, a nonspecific protein detected as a loading control. C, HIF1 electrophoretic mobility shift analysis. Embryonic stem cell clones expressing the form of pvhl at the tops of the panels are grouped in triplicates. First lane, normoxia treatment; second lane, hypoxia treatment; third lane, hypoxia treatment and HIF1 antibody supershift. *, a nonspecific DNA binding protein showing controlled loading. D, Elongin C coimmunoprecipitation. Hemagglutinin immunoprecipitates from each indicated cell line immunoblotted for VHL and Elongin C. Relative capture of Elongin C compared with pvhl. (H, hypoxia treatment; WT, wild-type) stimulation. Clones expressing the mutant protein representing type 2A VHL disease (VHL Y112H ) showed a variable pattern of HIF2 regulation. VHL Y112H clones exhibited at least partial restoration of hypoxic regulation. This mutation was unique among the panel in showing HIF2 regulation that seemed to be dose-dependent over a wide range of pvhl expression. Independently derived clones expressing VHL Y112H (Fig. 2B) showed elevated HIF2 levels under normoxia, which were additionally induced in response to hypoxia. Unique among the panel of mutants studied in this assay, dysregulation of HIF2 expression was consistently observed in extracts from four independently derived clones expressing the VHL disease type 2B mutant (VHL R167Q ), two of which are shown in Fig. 2B. These clones showed elevated expression of HIF2 under normoxia, with no additional induction in protein levels at 1.5% O 2. This effect was observed across many expression levels of pvhl. In contrast to the regulation of HIF1 observed in these mutant clones, the impaired Fig. 3. HIF target gene expression in cultured embryonic stem cells. A, VEGF ELISA., normoxia conditions; f, hypoxia conditions. Bars, SEM. B, Northern blot analysis of VEGF, phosphofructokinase, Aldolase A, and tubulin as a loading control. VHL genotype of each embryonic stem cell clone indicated at top. (WT, wild-type; H, hypoxia treatment; PFK, phosphofructokinase) 8598

5 Fig. 4. Teratoma growth effects of VHL missense mutations. A, teratoma mass. The tumor genotype is indicated at the bottom of each bar. Mass of the excised tumor is shown relative to the Vhl / standard. Number of tumors included in this analysis is shown above each bar. Histology of teratomas. B. H&E stained teratomas were visualized at 10 magnification, and the number of enlarged vascular lesions was counted on 10 to 15 random fields. C, H&E staining of teratomas. Vhl /, Vhl /, VHL Y112H, VHL R167Q, VHL L188V, and VHL R200W are indicated on each panel. All images are at 20 magnification. Arrows indicate the location of hemangiomas. Bars, SEM. (WT, wild-type) O 2 -dependent regulation of HIF2 is the expected result for this mutation based on previous investigations of the hypoxia response pathway in cells with mutations affecting the domain of pvhl (23). From these observations, we conclude that the forced expression of human VHL R167Q results in the exclusive disruption of HIF2 regulation. A nonspecific band shown in Fig. 2B corresponding to the lanes displayed for the VHL R167Q mutant showed equal loading. This control is representative of loading controls for each of the other mutants (data not shown). To determine whether the human pvhl mutants produce this effect on HIF regulation as a result of species-specific effects on the formation of the proteasomal complex, we subjected the cells to coimmunoprecipitation with the hemagglutinin tag on the mutant pvhl as an immunologic target. Immunoprecipitated proteins were electrophoresed and probed for presence of pvhl and Elongin C. Fig. 2D displays the pvhl present only in immunoprecipitates from complemented clones and shows that VHL R167Q is the sole mutant with reduced interaction with Elongin C. The relative amounts of Elongin C to VHL complex are displayed after scanning densitometry measurements. VHL Mutants Restore HIF1 Signaling Pathways. The hypoxia response pathway is regulated via rapid stabilization of the subunits of the HIF complex under conditions of low O 2. These transcription factors in turn activate the expression of a number of genes important in the cellular response to O 2 deprivation. Notably, VEGF expression is induced by the activity of these transcription factors. Secreted VEGF, measured by ELISA, correlated with HIF1 protein levels. Vhl / clones showed low-level secretion of VEGF under normoxic conditions, with a significant increase after 16-hour hypoxia treatment (1.5% O 2 ). Vhl / clones expressed high levels of VEGF at 21% O 2 and did not show additional induction on hypoxic stimulation. In contrast to the Vhl / parental clone, each mutant form of pvhl restored VEGF regulation. Clones expressing mutants VHL L188V and 8599 VHL R200W showed an intermediate level of normoxic VEGF expression but preserved hypoxic induction (Fig. 3A). VEGF protein expression determined by ELISA was confirmed by Northern analysis. Each of the mutants rescued VEGF transcriptional regulation similar to VHL wt. Additional targets of the HIF transactivator complex were also tested. Regulation of phosphofructokinase and aldolase A both showed dysregulated expression in Vhl / cells, but low normoxic expression and hypoxic induction were restored on expression of VHL wt and VHL mutants (Fig. 3B). These observations paralleled our findings of preserved regulation of HIF1 with each mutant and show an intact hypoxia response in this in vitro system. Activation of each of these HIF target genes was not impacted by elevated normoxic expression of HIF2 seen in mutants VHL Y112H and VHL R167Q. This in vitro finding is consistent with recent evidence that HIF2 is not functionally active in embryonic stem cells in culture (8). VHL Mutants Restore Tumor Growth in a Teratoma Assay. The teratoma model provides a differentiated in vivo system in which HIF2 has been observed to have activity. 5 Previously, we showed that loss of Vhl in embryonic stem cells results in decreased tumor size in a teratoma assay. This surprising effect on growth from loss of a tumor suppressor was reversed on VHL wt expression (20). Here we observed the effect of complementing Vhl / embryonic stem cells with mutated versions of VHL on teratoma growth. Data are shown for tumor mass after 3 weeks of growth (Fig. 4A). The teratomas generated from embryonic stem cells expressing VHL Y112H showed a general trend toward the development of smaller tumors as was observed for Vhl / cells. Increased numbers of tumors were generated with this mutation to enhance the significance of this trend and to make 5 K. Covello, M. C. Simon, and B. D. Keith, submitted for publication.

6 Fig. 5. Ki67 staining for cellular proliferation. A, teratoma sections from Vhl /, Vhl /, VHL Y112H, VHL R167Q, VHL L188V, and VHL R200W. B, area of 20 field with positive Ki67 nuclear stain. VHL genotype of each teratoma is indicated below the bar. Bars, SEM. available more tumor tissue for subsequent analysis. Additionally, the histology of the teratomas displayed a striking correlation between pvhl mutation and the formation of large blood-filled cavities lined by epithelial cells characteristic of hemangiomas (Fig. 4, C H). The appearance of these lesions was quantified by counting their number in 10 random low-power fields from three independent teratomas for each mutation (Fig. 4B). Teratomas derived from Vhl /, VHL Y112H, Vhl R167Q, and VHL R200W embryonic stem cells showed a high incidence of these lesions. It should be noted, however, that Vhl / and VHL Y112H teratomas seemed to have a higher proportion of very large hemangiomas, and this was not accounted for in the quantitation. To address the disparate growth trends for Vhl / and VHL Y112H teratomas, we analyzed teratomas for proliferative activity by immunostain for Ki67 expression. A small but reproducible proliferative advantage was noted for genotypes associated with larger tumors (Fig. 5, A G). Evaluation of activated caspase 3 expression revealed no striking difference among any teratomas, ruling out a significant apoptotic phenotype in the smaller tumors (data not shown). We have previously reported on both activated caspase 3 and terminal deoxynucleotidyl transferase-mediated nick end labeling assay for the Vhl / and Vhl / tumors and showed no significant difference between the two types of teratomas (20). These results corroborate our previous observations in Vhl / teratomas that loss of a putative tumor suppressor gene causes a proliferative defect. Additionally, a mutant that does not cause the same HIF dysregulation, VHL Y112H, also caused diminished proliferation. Finally, hemangiomas were observed in all of the mutants except the type 2C mutant VHL L188V. VHL Mutant Teratomas Display Marked Vascularity Not Correlated with VEGF Expression. One remarkable histologic observation was that the Vhl /, VHL Y112H, and VHL R167Q teratomas seemed to be more hemorrhagic, with large blood-filled cavities, as compared with the other teratomas (Fig. 4, B D). Immunostains for CD31/platelet/endothelial cell adhesion molecule (PECAM), CD34, 8600 and von Willebrand Factor were positive for the cells lining the blood-filled cavities, although high-power examination suggested a morphologically abnormal endothelial layer (Fig. 6, A F; data not shown). These immunostains detected both large and small vessels throughout the masses. Quantification of the microvessel density revealed increased microvessel density for Vhl / tumors and the mutants VHL Y112H, VHL R167Q, and VHL R200W (Fig. 6G). This pattern of increased microvessel density is consistent with the numbers of hemangiomas observed in the teratomas and correlates well with VHL mutations associated with development of renal cell carcinoma or polycythemia. Interestingly, this pattern of vascularity does not correlate with dysregulated HIF expression, suggesting a potential alternate mechanism for angiogenesis in this model. VEGF Protein Expression in Teratomas. VEGF immunohistochemical staining of the teratomas revealed dense VEGF staining of Vhl / sections but low level staining for Vhl /, VHL Y112H, VHL R167Q,orVHL L188V teratomas (Fig. 7). This result interestingly correlates directly with the in vitro embryonic stem cell findings regarding HIF1 dysregulation. Parallels to HIF2 could not be drawn in this differentiated system. Furthermore, the VEGF expression pattern observed by immunostaining does not correlate with observed differences in tumor growth characteristics, microvessel density, or the hemangiomas noted in a subset of the teratomas, suggesting that alternate activities of pvhl are important for the observed phenotypes. From this data we can conclude that VEGF is highly expressed in Vhl / teratomas, which may account for the vascular nature of these teratomas. However, the vascularity of the other mutant clones cannot be directly correlated to tissue expression of VEGF-165, although expression of other isoforms of VEGF or other angiogenic factors may play a role in the vascular phenotype. Fibronectin Deposition Is Decreased in All VHL Mutants. Previous analysis of pvhl activity in renal cell carcinoma cell lines with loss or mutation of VHL has consistently showed a defect in extracellular

7 Fig. 6. Endothelial cell staining of teratomas. A, PECAM/CD31 staining on teratomas of the indicated VHL genotype. B, CD34 staining on teratomas of the indicated VHL genotype. All immunohistochemical analyses are shown at 20 magnification. C, quantitative analysis of microvessel density measured in vessels per field. Bars, SEM. (WT, wild-type) matrix remodeling as a result of perturbation of direct or indirect interactions with fibronectin (42, 43). We have previously reported a defect in fibronectin deposition in teratomas derived from Vhl / cells (20), which is restored on introduction of wild-type human VHL cdna. Analysis of these teratomas confirmed this observation and extended the observed defect in fibronectin deposition to teratomas expressing VHL Y112H, VHL R167Q, VHLl L188V, and VHL R200W (Fig. 8). This result suggests that perturbations in any pvhl domain can cause defective extracellular matrix deposition and may impact on each of the various phenotypes observed in VHL disease. DISCUSSION To better understand the molecular mechanisms underlying VHL-mediated tumorigenesis, we have generated a panel of pathologically significant human VHL mutations in a primary culture system. This model system provides a unique opportunity to observe the effects of wild-type and mutant forms of pvhl protein on cellular and metabolic pathways in the absence of competing tumor-promoting genetic lesions. Additionally, these primary cells are amenable to three-dimensional growth as grafted teratomas, and thus provide a platform for the study of in vivo growth control, vascularization, and extracellular matrix deposition. The regulation of the hypoxia response by manipulation of pvhl activity has undergone evaluation in both in vitro studies as well as in cultured human tumor cells. These results have showed that in renal tumor cells (13) and murine embryonic stem cells (20), loss of VHL results in constitutive stabilization of the HIF1 and HIF2 transcription factors and activation of the HIF-mediated Fig. 7. VEGF expression in teratomas. Panels show representative expression patterns of VEGF by immunohistochemical analysis in teratomas derived from the indicated embryonic stem cell clone. All images are at 20 magnification. (WT, wild-type) 8601

8 Fig. 8. Fibronectin deposition around small vessels in embryonic stem cell-derived teratomas. Arrows indicate the location of small vessels, which most clearly show the abnormality. All images are at 40 magnification. hypoxia response pathway. VHL mutations linked to type 2C VHL disease have been previously shown to preserve the ability to negatively regulate HIF1 and HIF2 (33) comparable with the ectopic expression of wild-type VHL. However, previous work in human renal carcinoma cells has showed that VHL disease type 2A and 2B mutant proteins fail to restore normoxic ubiquitylation of HIF1 and fail to fully repress normoxic expression of both HIF1 and HIF2 (23). However, in this primary cell culture system, all of the pvhl mutants fully restored the wild-type pattern of HIF1 expression. A VHL type 2B disease mutant, VHL R167Q, was the sole mutant in which an impaired interaction with Elongin C was observed, corroborating previous studies (23). Despite this apparent deficit in Elongin C binding, however, this mutant displayed oxygen-dependent regulation of HIF1 protein levels, binding activity, and target gene activation. VHL R167Q expression, however, resulted in dysregulated HIF2 levels, suggesting either that pvhl to Elongin C interaction is only necessary for regulation of HIF2, or that even a small interaction with the ubiquitin ligase complex is sufficient to retain oxygen-dependent regulation of HIF1. Evaluation of an embryonic stem line bearing targeted arginine to glutamine mutation at the equivalent site in both murine Vhl alleles confirms elevated levels of HIF2 in this setting. 6 This model system represents the first example of a primary cell culture system in which the regulation of HIF1 and HIF2 are disconnected and provides a model system in which the distinct roles of each of these factors can be additionally elucidated. A limitation of this system is the potential for different interactions between human pvhl and the murine milieu. Although human and mouse pvhl share considerable homology, differences do exist 6 W. K. Rathmell and M. C. Simon, unpublished data which may impart the differential regulation of HIF1 and HIF2 we have observed in our model system. Although such an analysis is beyond the scope of this report, the mechanism that contributes to the independent regulation of HIF2 is certainly an interesting avenue of investigation. The isolated dysregulation of HIF2 observed in the VHL R167Q mutant lines provides a unique opportunity to study the effect of high normoxic expression of this protein on tumor growth. Although investigations of the hypoxia response pathway have largely focused on the regulation of HIF1, perturbations of HIF2 regulation have been observed independently of HIF1 dysregulation in renal cell carcinomas. As an example, the intensively studied renal carcinoma cell line, 786-0, which lacks pvhl expression, has been found to have a high normoxic expression of HIF2, whereas HIF1 is not expressed at any detectable level in those cells, even with hypoxic stimulation. In the cell line, inhibition of HIF2 results in a restoration of wild-type growth in xenograft tumors (49), suggesting that HIF2 may actually be the more critical mediator of clear cell renal cell carcinoma. We have additionally observed that despite stabilized HIF2 protein in VHL R167Q embryonic stem cells, a panel of HIF targets were not transcriptionally activated unless stimulated by hypoxia. In embryonic stem cells, we have previously observed the failure of HIF2 to transcriptionally activate target genes, even when present in high levels (8). Other reports have observed a localization defect of HIF2 in mouse embryo fibroblasts (50). However, HIF2 was present in nuclear extracts in this study. The inactivity of HIF2 may be because of a protein modification or transcriptional inhibitor present in this particular cell type. Additionally, high-level expression of HIF2 may, in fact, play an important role in the phenotype of the cells exclusive of the typical cohort of HIF-responsive genes. Alternate transcriptional targets or alternate activities altogether may account for the phenotype observed in teratomas derived from cells with HIF2 overexpression. Our analysis, however, corroborates previous evidence that VHL L188V, associated with type 2C disease, must map to a region outside the HIF regulatory domains. In teratomas we observed that VHL Y112H tumors behaved like the Vhl / tumors with regard to reduced tumor growth, whereas the other mutant tumors restored teratoma growth similar to VHL wt. This experiment strongly supports the hypothesis that reduced tumor growth is a VHL-specific event. Furthermore, we observed that the VHL mutant VHL Y112H, which has a markedly different signature on the regulation of the hypoxia response compared with Vhl / cells, has a similar growth defect. This may provide insight into the potentially distinct roles of HIF1 and HIF2 accumulation in tumor growth. An interesting effect of Vhl loss in various tissues in mouse models of VHL disease is the development of extensive vascular malformations in multiple VHL disease target organs (45, 46). In our panel of VHL mutants, we observed markedly increased vascularity in VHL / and VHL Y112H teratomas and moderately increased vascularity in VHL R167Q and VHL R200W teratomas. These VHL mutants are associated with phenotypes of hemangioblastoma and erythropoietin overexpression, which are hallmarks of activation of the hypoxia response pathway. Finally, multiple lines of evidence have shown a role for pvhl in the maintenance of the extracellular matrix. We have observed a defect in fibronectin deposition in each of our mutant embryonic stem cell teratomas, although exogenous expression of VHL wt does restore this phenotype (20). This unique and interesting effect of VHL perturbation may be an integral contributor to the growth characteristics of VHL disease-associated tumors, possibly by threatening the structural integrity of the extracellular matrix and thus permitting the formation of large vascular lesions characteristic of the teratomas.

9 Taken together, these results suggest a complex pathway of regulation of tumor angiogenesis and growth mediated by subtle alterations in the pvhl protein. It is likely that pvhl roles outside of the regulation of the hypoxia response pathway are impacting cellular growth, and that cell-specific factors directing the transcriptional hypoxia response are an important determinant of downstream effects of VHL mutation. Although the specific mechanisms mediating tumor growth and development in this model system remain to be identified, this analysis provides additional evidence of the tissue-specific regulation of HIF2, and the potential importance of this member of the HIF transcription factor family in the development of clear cell renal cell carcinoma. REFERENCES 1. Maher ER, Kaelin WG Jr. von Hippel-Lindau disease. Medicine (Baltim) 1997;76: Wang GL, Jiang BH, Rue EA, Semenza GL. Hypoxia-inducible factor 1 is a basic-helix-loop-helix-pas heterodimer regulated by cellular O2 tension. Proc Natl Acad Sci USA 1995;92: Wang GL, Semenza GL. Purification and characterization of hypoxia-inducible factor 1. J Biol Chem 1995;270: Semenza GL. Hypoxia-inducible factor 1: master regulator of O2 homeostasis. Curr Opin Genet Dev 1998;8: Wenger RH. Cellular adaptation to hypoxia: O2-sensing protein hydroxylases, hypoxia-inducible transcription factors, and O2-regulated gene expression. FASEB J 2002;16: Semenza GL, Roth PH, Fang HM, Wang GL. Transcriptional regulation of genes encoding glycolytic enzymes by hypoxia-inducible factor 1. J Biol Chem 1994;269: Ratcliffe PJ, O Rourke JF, Maxwell PH, Pugh CW. Oxygen sensing, hypoxiainducible factor-1 and the regulation of mammalian gene expression. J Exp Biol 1998;201(Pt 8): Hu CJ, Wang LY, Chodosh LA, Keith B, Simon MC. Differential roles of hypoxiainducible factor 1alpha (HIF-1alpha) and HIF-2alpha in hypoxic gene regulation. Mol Cell Biol 2003;23: Saarikoski ST, Rivera SP, Hankinson O. Mitogen-inducible gene 6 (MIG-6), adipophilin and tuftelin are inducible by hypoxia. FEBS Lett 2002;530: Wykoff CC, Pugh CW, Harris AL, Maxwell PH, Ratcliffe PJ. The HIF pathway: implications for patterns of gene expression in cancer. Novartis Found Symp 2001; 240:212 25; discussion Wiesener MS, Munchenhagen PM, Berger I, et al. Constitutive activation of hypoxiainducible genes related to overexpression of hypoxia-inducible factor-1alpha in clear cell renal carcinomas. Cancer Res 2001;61: Ohh M, Park CW, Ivan M, et al. Ubiquitination of hypoxia-inducible factor requires direct binding to the beta-domain of the von Hippel-Lindau protein. Nat Cell Biol 2000;2: Maxwell PH, Wiesener MS, Chang GW, et al. The tumour suppressor protein VHL targets hypoxia-inducible factors for oxygen-dependent proteolysis. Nature (Lond) 1999;399: Cockman ME, Masson N, Mole DR, et al. Hypoxia inducible factor-alpha binding and ubiquitylation by the von Hippel-Lindau tumor suppressor protein. J Biol Chem 2000;275: Iliopoulos O, Levy AP, Jiang C, Kaelin WG Jr, Goldberg MA. Negative regulation of hypoxia-inducible genes by the von Hippel-Lindau protein. Proc Natl Acad Sci USA 1996;93: Pugh CW, Ratcliffe PJ. The von Hippel-Lindau tumor suppressor, hypoxia-inducible factor-1 (HIF-1) degradation, and cancer pathogenesis. Semin Cancer Biol 2003;13: Ivan M, Haberberger T, Gervasi DC, et al. Biochemical purification and pharmacological inhibition of a mammalian prolyl hydroxylase acting on hypoxia-inducible factor. Proc Natl Acad Sci USA 2002;99: Huang J, Zhao Q, Mooney SM, Lee FS. Sequence determinants in hypoxia-inducible factor-1alpha for hydroxylation by the prolyl hydroxylases PHD1, PHD2, and PHD3. J Biol Chem 2002;277: Jaakkola P, Mole DR, Tian YM, et al. Targeting of HIF-alpha to the von Hippel- Lindau ubiquitylation complex by O2-regulated prolyl hydroxylation. Science (Wash DC) 2001;292: Mack FA, Rathmell WK, Arsham AM, et al. Loss of pvhl is sufficient to cause HIF dysregulation in primary cells but does not promote tumor growth. Cancer Cell 2003;3: Zbar B, Kishida T, Chen F, et al. Germline mutations in the Von Hippel-Lindau disease (VHL) gene in families from North America, Europe, and Japan. Hum Mutat 1996;8: Crossey PA, Richards FM, Foster K, et al. Identification of intragenic mutations in the von Hippel-Lindau disease tumour suppressor gene and correlation with disease phenotype. Hum Mol Genet 1994;3: Clifford SC, Cockman ME, Smallwood AC, et al. Contrasting effects on HIF-1alpha regulation by disease-causing pvhl mutations correlate with patterns of tumourigenesis in von Hippel-Lindau disease. Hum Mol Genet 2001;10: Ang SO, Chen H, Hirota K, et al. Disruption of oxygen homeostasis underlies congenital Chuvash polycythemia. Nat Genet 2002;32: Pastore YD, Jelinek J, Ang S, et al. Mutations in the VHL gene in sporadic apparently congenital polycythemia. Blood 2003;101: Foster K, Prowse A, van den Berg A, et al. Somatic mutations of the von Hippel- Lindau disease tumour suppressor gene in non-familial clear cell renal carcinoma. Hum Mol Genet 1994;3: Kondo K, Yao M, Yoshida M, et al. Comprehensive mutational analysis of the VHL gene in sporadic renal cell carcinoma: relationship to clinicopathological parameters. Genes Chromosomes Cancer 2002;34: Shuin T, Kondo K, Torigoe S, et al. Frequent somatic mutations and loss of heterozygosity of the von Hippel-Lindau tumor suppressor gene in primary human renal cell carcinomas. Cancer Res 1994;54: Mandriota SJ, Turner KJ, Davies DR, et al. HIF activation identifies early lesions in VHL kidneys: evidence for site-specific tumor suppressor function in the nephron. Cancer Cell 2002;1: Jiang Y, Zhang W, Kondo K, et al. Gene expression profiling in a renal cell carcinoma cell line: dissecting VHL and hypoxia-dependent pathways. Mol Cancer Res 2003; 1: Chen F, Kishida T, Duh FM, et al. Suppression of growth of renal carcinoma cells by the von Hippel-Lindau tumor suppressor gene. Cancer Res 1995;55: Iliopoulos O, Kibel A, Gray S, Kaelin WG Jr. Tumour suppression by the human von Hippel-Lindau gene product. Nat Med 1995;1: Hoffman MA, Ohh M, Yang H, et al. von Hippel-Lindau protein mutants linked to type 2C VHL disease preserve the ability to downregulate HIF. Hum Mol Genet 2001;10: Ohh M, Kaelin WG Jr. VHL and kidney cancer. Methods Mol Biol 2003;222: Kaelin WG, Iliopoulos O, Lonergan KM, Ohh M. Functions of the von Hippel-Lindau tumour suppressor protein. J Intern Med 1998;243: Pause A, Lee S, Lonergan KM, Klausner RD. The von Hippel-Lindau tumor suppressor gene is required for cell cycle exit upon serum withdrawal. Proc Natl Acad Sci USA 1998;95: Bindra RS, Vasselli JR, Stearman R, Linehan WM, Klausner RD. VHL-mediated hypoxia regulation of cyclin D1 in renal carcinoma cells. Cancer Res 2002;62: Baba M, Hirai S, Yamada-Okabe H, et al. Loss of von Hippel-Lindau protein causes cell density dependent deregulation of CyclinD1 expression through hypoxia-inducible factor. Oncogene 2003;22: Lieubeau-Teillet B, Rak J, Jothy S, et al. von Hippel-Lindau gene-mediated growth suppression and induction of differentiation in renal cell carcinoma cells grown as multicellular tumor spheroids. Cancer Res 1998;58: Baba M, Hirai S, Kawakami S, et al. Tumor suppressor protein VHL is induced at high cell density and mediates contact inhibition of cell growth. Oncogene 2001;20: Hergovich A, Lisztwan J, Barry R, Ballschmieter P, Krek W. Regulation of microtubule stability by the von Hippel-Lindau tumour suppressor protein pvhl. Nat Cell Biol 2003;5: Kamada M, Suzuki K, Kato Y, Okuda H, Shuin T. von Hippel-Lindau protein promotes the assembly of actin and vinculin and inhibits cell motility. Cancer Res 2001;61: Ohh M, Yauch RL, Lonergan KM, et al. The von Hippel-Lindau tumor suppressor protein is required for proper assembly of an extracellular fibronectin matrix. Mol Cell 1998;1: Gnarra JR, Ward JM, Porter FD, et al. Defective placental vasculogenesis causes embryonic lethality in VHL-deficient mice. Proc Natl Acad Sci USA 1997;94: Haase VH, Glickman JN, Socolovsky M, Jaenisch R. Vascular tumors in livers with targeted inactivation of the von Hippel-Lindau tumor suppressor. Proc Natl Acad Sci USA 2001;98: Ma W, Tessarollo L, Hong SB, et al. Hepatic vascular tumors, angiectasis in multiple organs, and impaired spermatogenesis in mice with conditional inactivation of the VHL gene. Cancer Res 2003;63: Kleymenova E, Everitt JI, Pluta L, et al. Susceptibility to vascular neoplasms but no increased susceptibility to renal carcinogenesis in Vhl knockout mice. Carcinogenesis (Lond) 2004;25: Byrappa S, Gavin DK, Gupta KC. A highly efficient procedure for site-specific mutagenesis of full-length plasmids using Vent DNA polymerase. Genome Res 1995;5: Kondo K, Kim WY, Lechpammer M, Kaelin WG Jr. Inhibition of HIF2alpha is sufficient to suppress pvhl-defective tumor growth. PLoS Biol 2003;1:E Park SK, Dadak AM, Haase VH, et al. Hypoxia-induced gene expression occurs solely through the action of hypoxia-inducible factor 1alpha (HIF-1alpha): role of cytoplasmic trapping of HIF-2alpha. Mol Cell Biol 2003;23: