Oncogene (2000) 19, 6297 ± 6305 ã 2000 Nature Publishing Group All rights reserved 0950 ± 9232/00 $15.00 www.nature.com/onc Identi®cation of novel hypoxia dependent and independent target genes of the von Hippel-Lindau (VHL) tumour suppressor by mRNA di€erential expression pro®ling

Charles C Wyko€1, Christopher W Pugh2, Patrick H Maxwell2, Adrian L Harris1 and Peter J Ratcli€e*,2

1Institute of Molecular Medicine, John Radcli€e Hospital, Oxford OX3 9DS, UK; 2Wellcome Trust Centre for Human Genetics, Oxford OX3 7BN, UK

The von Hippel-Lindau tumour suppressor gene (VHL) Germline mutations in the von Hippel-Lindau targets hypoxia inducible factor (HIF)-a subunits for (VHL) tumour suppressor gene are associated with a ubiquitin dependent proteolysis. To better understand the dominantly inherited cancer syndrome, in which role of this and other putative pathways of gene inactivation or loss of the remaining wild type allele regulation in VHL function we subjected mRNA from as a somatic event results in the formation of typically VHL defective renal carcinoma cells and transfectants hypervascularized neoplasms involving the kidney, re-expressing a wild type VHL allele to di€erential retina, pancreas, adrenal gland and central nervous expression pro®ling, and analysed VHL target genes for system (Kaelin and Maher, 1998). In addition, somatic oxygen regulated expression. Among a group of newly inactivation or mutation of both VHL alleles is identi®ed VHL target genes the majority but not all were observed in the majority of sporadic renal cancers regulated by oxygen, indicating that whilst dysregulation (Gnarra et al., 1994). Recent work has connected the of the HIF system makes a dominant contribution to function of the gene product, pVHL, with regulation of alterations in transcription, VHL has other in¯uences on the transcriptional response to hypoxia. pVHL forms patterns of gene expression. Genes newly de®ned as part of a ubiquitin ligase complex (Iwai et al., 1999; targets of the VHL/hypoxia pathway (conditionally Lisztwan et al., 1999) that targets the regulatory a downregulated by VHL in normoxic cells) include subunits of hypoxia inducible factor-1 (HIF-1) for aminopeptidase A, collagen type V, alpha 1, cyclin G2, oxygen dependent proteolysis (Maxwell et al., 1999; DEC1/Stra13, endothelin 1, low density lipoprotein Cockman et al., 2000). HIF-1 is a heterodimeric DNA receptor-related protein 1, MIC2/CD99, and transgluta- binding complex that directs the expression of a large minase 2. These genes have a variety of functions family of hypoxia inducible genes including members relevant to tumour biology. However, not all are involved in glucose transport, glycolysis, and angiogen- connected with the promotion of tumour growth, some esis (Shweiki et al., 1992; Semenza et al., 1994; Ebert et being pro-apoptotic or growth inhibitory. We postulate al., 1995). In VHL defective cells, HIF-a subunits are that co-ordinate regulation as part of the HIF pathway constitutively stabilized leading to activation of the may explain this paradox, and that evolution of anti- transcriptional complex and its target genes irrespective apoptotic pathways may be required for tumour growth of oxygen concentration (Maxwell et al., 1999). under VHL-dysregulation. Our results indicate that it The direct targeting of a hypoxia inducible pathway will be necessary to consider the e€ects of abnormal that is most clearly linked with the regulation of activity in integral regulatory pathways, as well as the metabolism and angiogenesis is unusual amongst e€ects of individual genes to understand the role of tumour supressor genes. Though upregulation of abnormal patterns of gene expression in cancer. Onco- angiogenic growth factors provides an explanation for gene (2000) 19, 6297 ± 6305. the angiogenic phenotype of VHL syndrome, the role of HIF activation in other aspects of oncogenesis is less Keywords: VHL; hypoxia; HIF-1a clear. HIF might have other targets that could illuminate such a role, or pVHL might have functions other than HIF regulation that are relevant to its Introduction tumour suppressor function (Ohh et al., 1998; Pause et al., 1998; Iwai et al., 1999). Tumour suppressor genes and oncogenes are often We therefore sought to expand our understanding of involved, directly or indirectly, in the regulation of the patterns of gene expression that are a€ected by transcription. New opportunities for large-scale analy- pVHL, and to determine the extent to which these sis of the downstream patterns of abnormal gene genes are also subject to regulation by oxygen. We expression (Schena et al., 1995; Iyer et al., 1999) may employed a glass-chip based cDNA array of 9182 therefore provide important insights into the under- unbiased genes and ESTs to compare the gene lying processes which promote the development of expression pro®le in stable transfectants of the renal cancer. cell carcinoma (RCC) line, RCC4, that were either defective or competent for pVHL. Regulation by pVHL was con®rmed by ribonuclease protection assay *Correspondence: PJ Ratcli€e both in the original pair of RCC4 cell lines, and a Received 7 August 2000; revised 10 October 2000; accepted 10 second pair of transfectants based on another VHL October 2000 defective RCC line, 786-0. We demonstrate the mRNA differential expression profiling to identify VHL target genes CC Wykoff et al 6298 reproducibility of the screen, and show that for the for 30 genes showing the greatest induction or large majority of genes identi®ed by the screen, repression by transfection of VHL in the ®rst screen di€erential expression could be con®rmed by ribonu- plotted against expression data for the same cDNAs in clease protection assay. For about 50% of the the second screen. There was a good correlation, identi®ed genes, a similar pattern of di€erential though quantitative di€erences between the screens in expression in relation to VHL status was apparent in the amplitude of expression were commonly observed. both cell backgrounds. The majority of these genes were also regulated by hypoxia in multiple non-renal RNAse protection analysis of VHL dependent gene cell types, therefore de®ning several new classes of expression in different renal cell carcinoma cell lines hypoxia/pVHL responsive genes. In addition, other genes were de®ned whose expression was responsive to GEM analysis identi®ed 22 genes that were consis- pVHL but not to hypoxia, indicating that pVHL may tently repressed by VHL in both screens by an in¯uence additional pathways of gene expression that average of twofold or greater. For these genes, are not regulated by hypoxia. riboprobe templates were constructed to permit accurate quanti®cation of mRNA regulation by ribonuclease protection assay (RPA). Two di€erent sets of renal cell carcinoma transfectants were studied, Results the original pair of RCC4 and RCC4/VHL transfec- tants, and a similar pair of 786-0 transfectants Analysis of VHL dependent expression by gene expression expressing vector backbone alone or wild type human microarray pVHL (786-0 and 786-0/VHL). Results are summar- To examine the e€ects of the tumour suppressor gene ized in Table 1a. Of the 22 genes identi®ed by GEM VHL on global patterns of gene expression we screening, 20 were con®rmed by RPA as showing a employed an mRNA di€erential screen of 9182 unique clear reduction in expression in RCC4/VHL cells clones using a Gene Expression Microarray (GEM) versus RCC4 cells, whereas two genes were not based on competitive hybridization expression pro®ling expressed at levels quanti®able by RPA. Interestingly, (Incyte Genomics). Of the 9182 cDNA clones, 8481 are for many genes the amplitude of di€erential mRNA unique annotated genes. We compared mRNA from expression demonstrated by RPA was substantially stable transfectants of the pVHL de®cient renal underestimated by GEM analysis. Of the 20 genes carcinoma cell line RCC4 expressing vector backbone showing repression by re-expression of VHL in RCC4, alone (RCC4) or wild type human pVHL (RCC4/ seven were also repressed by re-expression of VHL in VHL). To test the reproducibility of the screen, the 786-0, ®ve were not expressed at levels quanti®able by procedure was carried out on a second independently RPA in 786-0 cells, and eight were expressed but not derived pair of mRNA samples extracted from RCC4 in¯uenced by re-expression of VHL in this back- and RCC4/VHL cells. Figure 1 shows expression data ground.

Figure 1 Reproducibility of the GEM data sets. Di€erential expression ratios are shown for the 30 genes showing either the greatest induction or the greatest repression following re-expression of wild type VHL in RCC4 cells as de®ned by the ®rst GEM screen. Data from this screen (GEM1) are plotted against expression data for the same genes in the second GEM (GEM2). Positive numbers indicate induction of expression following re-expression of VHL. Negative numbers indicate repression of expression following re-expression of VHL

Oncogene mRNA differential expression profiling to identify VHL target genes CC Wykoff et al 6299 Table 1 a Analysis of candidate VHL-repressible genes Fold regulation Accession number Gene RCC4 array RCC4 RPA 786-0 RPA

Y00749 endothelin 1 2.7 47 21 AB004066 DEC1 2.2 5.7 8.7 AI061430 MIC2 2.6 2.5 3.2 NM004613 transglutaminase 2 2.7 12 35 X56134 vimentin 2.7 2 2 Y00757 secretory granule neuroendocrine prot 1 6.9 45 14 AL047358 spermidine N1-acetyltransferase 2.5 3.3 3.9 NM002332 LRP1 2.5 11 NE M76729 collagen type V, alpha 1 2.3 21 NE AI271688 cyclin G2 2.4 8.3 NE L12468 aminopeptidase A 2.6 22 NE X80197 keratin, hair, basic, 1 9.8 450 NE AW373887 collagen type 1, alpha 1 5.4 450 NR NM002205 integrin, alpha 5 2.2 3.1 NR AW162231 ferritin, light polypeptide 2 2.1 NR AI096619 JM4 protein 2.1 3.3 NR AB002346 synaptojanin 2 2.7 3.2 NR X83703 cardiac ankyrin repeat protein 2.1 6 NR AF013711 transgelin 3.4 10 NR Z29373 L1 cell adhesion molecule 7.5 450 NR AL021918 zinc finger protein 184 2.2 NE NE AW275525 EST (Incyte ID: 1909292) 2.1 NE NE b Analysis of candidate VHL-inducible genes Fold regulation Accession number Gene RCC4 array RCC4 RPA 786-0 RPA

W40312 EST (Incyte ID: 1797836) 3.9 5 3.2 Y00318 I factor 9.6 450 NE X87212 cathepsin C 4.4 6 NR NM000224 keratin 18 4.6 24 NR AA774661 ATPase Na+/K+ transporter, beta 3 2.4 NR NR

NE: gene is not expressed at levels quanti®able by ribonuclease protection assay (RPA); NR: gene is expressed although not regulated by VHL by at least twofold. Fold regulation as determined by array screening is the average of two independent arrays. Fold regulation as determined by RPA was calculated by quanti®cation of the protected species, and standardized to an internal control, the constitutively expressed U6 small nuclear RNA. Four genes were regulated by VHL to an excess of 50-fold, above which, quanti®cation is less reliable and therefore labelled as 450

In addition to the genes which were downregulated RCC4 cells but not expressed in 786-0 cells (n=12). by VHL, GEM analysis of the RCC4 transfectants also Results are illustrated in Figure 2a ± c, and summarized identi®ed a similar number of genes that were in Table 2a. In total, eight of these genes (endothelin 1 upregulated by re-expression of VHL (Figure 1). Those (EDN1), di€erentiated embryo chrondocyte 1 (DEC1), ®ve showing the largest amplitude of di€erential monoclonal, Imperial Cancer Research Fund 2 expression were selected for further analysis by RPA. (MIC2), transglutaminase 2 (TGM2), low density Results are summarized in Table 1b. Again the GEM lipoprotein receptor-related protein 1 (LRP1), collagen results were largely reproduced by RPA, with four of type V, alpha 1 (COL5A1), cyclin G2 (CCNG2), and the ®ve genes showing clearly increased levels of aminopeptidase A (APA)), were reproducibly induced expression in RCC4/VHL. When these four genes were at least twofold by hypoxia in one or both assayed in the 786-0 transfectants, one was clearly transfectants re-expressing VHL (Figure 2a,b). In each induced by re-expression of VHL, whereas one was not case upregulation in the corresponding VHL de®cient expressed at levels quanti®able by RPA, and two were control transfectant was associated with loss of a expressed but not in¯uenced by re-expression of VHL hypoxic response. For the remaining four genes in this background. (vimentin (VIM), secretory granule neuroendocrine protein 1 (SGNE1), spermidine N1-acetyltransferase (SAT), and keratin, hair, basic, 1 (KRTHB1)) Oxygen responsiveness of VHL regulated genes responses to hypoxia were small or absent (Figure 2c). The above assay de®ned several new genes that are All four genes that were upregulated by VHL re- regulated by VHL. We next wished to determine the expression in RCC4 cells were examined for responses to extent to which these newly de®ned VHL responsive hypoxia in a similar manner, and results are illustrated genes were regulated by the hypoxia pathway. The in Figure 2d and summarized in Table 2b. Two of these pairs of transfectants of RCC4 and 786-0 cells were genes (cathepsin C (CTSC) and keratin 18 (K18)) were therefore exposed in parallel to normoxia (20% O2)or repressed by hypoxia in RCC4/VHL. Interestingly these hypoxia (0.1% O2) for 16 h. responses were not seen in the 786-0 background where Ribonuclease protection assays were performed for the genes also appeared unresponsive to VHL. Two all genes that were downregulated by VHL in both genes (EST (Incyte ID: 1797836) and I factor (IF)) were RCC4 and 786-0 cells, or downregulated by VHL in unresponsive to hypoxia in either cellular background.

Oncogene mRNA differential expression profiling to identify VHL target genes CC Wykoff et al 6300 ab

c d

e

Figure 2 Ribonuclease protection assays of di€erentially expressed transcripts in two VHL defective renal carcinoma cell lines, RCC4 and 786-0, and the corresponding stable transfectants re-expressing wild type VHL (+VHL). Cells were exposed to either normoxia (N; 20% O2) or hypoxia (H; 0.1% O2) for 16 h. (a) Examples of genes that are repressed by VHL and show hypoxia inducible responses in both +VHL backgrounds. (b) Examples of genes that are repressed by VHL and show hypoxia inducible responses in RCC4/VHL. (c) Examples of genes that are repressed by VHL to varying extents in RCC4 and 786-0, but show no clear induction by hypoxia. (d) Examples of genes that are induced by VHL. (e) Expression of control genes: GLUT-1, a known VHL/hypoxia target gene employed as a positive control for hypoxic stimulation of the samples assayed; LC, internal control assay (constitutively expressed U6 small nuclear RNA)

above results strongly suggested that the eight genes Oxygen dependent expression in non-renal cell types that were identi®ed as being both downregulated by Of the eight genes de®ned above as showing hypoxia pVHL in normoxic cells and upregulated by hypoxia in inducible responses in VHL competent cells, endothelin 1 VHL-competent cells, and the two which are regulated is the only gene whose hypoxic response has been studied by both stimuli in the opposite direction, are extensively. To examine whether these newly described responding directly or indirectly to HIF. To con®rm responses to hypoxia were observed generally, we this for two widely expressed genes, DEC1 and examined their hypoxia inducible expression in cell lines CCNG2, we constructed riboprobes to the homologous derived from lung (A549), bladder (EJ-28) and breast Chinese hamster genes and tested expression in a wild (HBL-100) tissues. Results are illustrated in Figure 3. type Chinese hamster ovary subline (C4.5), and a Five of the genes (DEC1, MIC2, TGM2, LRP1, and mutant derivative (Ka13) which is functionally defec- CCNG2) were induced by hypoxia in each of the three tive in HIF-1a (Wood et al., 1998). Both genes showed cell lines. APA was not expressed in any of the three non- strong induction by hypoxia in C4.5 cells, which was renal cell lines, whilst COL5A1 was strongly induced by abolished in the HIFa de®cient Ka13 cells (Figure 4). hypoxia in HBL100 but not expressed in A549 or EJ-28.

Discussion Expression in HIFa deficient cells Taken together with the function of pVHL in the We have used a glass-chip array to examine the oxygen regulated proteolysis of HIF-a subunits, the di€erences in mRNA expression resulting from stable

Oncogene mRNA differential expression profiling to identify VHL target genes CC Wykoff et al 6301 Table 2 a Analysis of hypoxia regulation of confirmed VHL repressible targets Fold upregulation RCC4/VHL 786-0/VHL Gene hypoxia hypoxia

EDN1 9.7 9.2 DEC1 4.8 4.7 MIC2 2.3 2.3 TGM2 5.1 7.1 LRP1 3.1 NE COL5A1 4.1 NE CCNG2 3.7 NE APA 3.7 NE VIM NR NR SGNE1 NE NE SAT NR NR KRTHB1 NE NE b Analysis of hypoxia regulation of confirmed VHL-inducible targets Fold downregulation RCC4/VHL 786-0/VHL Gene hypoxia hypoxia

EST (Incyte ID: 1797836) NR NR IF NR NE CTSC 2.2 NR K18 16 NR

NE: gene is not expressed at levels quanti®able by ribonuclease protecion assay (RPA); NR: gene is expressed although not regulated by hypoxia by at least twofold. Fold regulation as determined by RPA was calculated by quanti®cation of the protected species, and standardized to an internal control, the constitutively expressed U6 small nuclear RNA transfection of wild type pVHL into a VHL defective renal carcinoma cell line. Using a signi®cance threshold of twofold di€erential expression we de®ned a number of new genes that are regulated either negatively or positively by VHL, and have shown that the majority, Figure 3 Induction by hypoxia in non-renal derived cell lines. but not all, are responsive to hypoxia. Cells were exposed to either normoxia (N; 20% O2) or hypoxia (H; 0.1% O2) for 16 h. Ribonuclease protection analysis of Based on veri®cation by ribonuclease protection DEC1, MIC2, LRP1, TGM2, COL5A1, CCNG2, and APA in assay, GEM screening identi®ed VHL regulated genes lung (A549), bladder (EJ-28), and breast (HBL-100) derived cell with a low false positive rate. However, several lines. LC, internal control assay (constitutively expressed U6 indicators suggested that the screens may have small nuclear RNA) considerably underestimated the number of genes which are regulated in this way. First, for several genes the quantitative mRNA analyses by ribonuclease limited overlap is observed. Thus only DEC1/Stra 13 protection demonstrated a substantially greater ampli- (Ivanov et al., 1998) and vimentin (Moch et al., 1999) tude of regulation than that indicated by the GEM have previously been identi®ed as upregulated in VHL screens. Second, several genes known to be strongly de®cient cells, whereas LRP1 was the only gene regulated by VHL, such as LDH-A (Maxwell et al., identi®ed in common with a recent array screening 1999) and VEGF (Illiopoulos et al., 1996), showed for hypoxia inducible genes in cervical and other regulation in the GEM screens below the arbitrary squamous carcinoma cells (Koong et al., 2000). signi®cance threshold of twofold di€erential expression Analysis of expression in two sets of renal cell that we used to select genes for further study. This carcinoma transfectants that were either VHL defec- occurred despite the demonstration of substantially tive, or competent was used to de®ne a core group of greater levels of regulation for these genes by VHL responsive genes which were regulated by VHL in ribonuclease protection assay in the mRNA samples both sets of transfectants. Surprisingly, a substantial used for GEM screening (data not shown). Thus, using number of genes that were identi®ed as regulated by this threshold it is clear that the GEM screening for VHL status in the RCC4 transfectants were either not VHL targets had a substantial false negative rate. a€ected by VHL or not expressed at all in the 786-0 Taken together with the number of con®rmed VHL transfectants. Such a result might arise if there were targets from a screen of 9182 genes, this suggests that clonal di€erences between the RCC4 and RCC4/VHL the total number of targets whose expression is cells that were unrelated to VHL status, or if the range regulated directly or indirectly by VHL is likely to be of VHL responsive genes were di€erent in the two large, perhaps ranging into hundreds of genes. This renal carcinoma lines. Support for the latter possibility possibility is also supported by comparison of our data was obtained for one subset of genes. Thus for some with other recently published screens for either VHL genes that were di€erentially expressed in accordance regulated, or hypoxia regulated genes, in that only with VHL status in the RCC4 transfectants, but not

Oncogene mRNA differential expression profiling to identify VHL target genes CC Wykoff et al 6302 despite regulation by VHL in both cell backgrounds. Two of these genes, secretory granule neuroendocrine protein 1 and spermidine N1-acetyltransferase, were downregulated by VHL, and one, an EST, was upregulated by VHL. Taken together these results indicate that the role of VHL in regulation of the HIF pathway makes a dominant contribution to the overall pattern of altered gene expression in VHL defective cells. However, the results also indicate that pVHL most likely has other functions which impinge on patterns of gene expression, and which will need to be considered in the analysis of its role as a tumor suppressor. Though the functions of several of the VHL regulated genes de®ned in this study are incompletely understood, a number have known functions which are of interest in relation to the biology of tumour formation. For instance endothelin 1 is a hypoxia inducible HIF-1 target gene (Hieda and Gomez- Sanchez, 1990; Hu et al., 1998) capable of stimulating proliferation and migration, inducing VEGF produc- tion (Okuda et al., 1998; Kozawa et al., 2000), and increasing the expression of various proto-oncogenes Figure 4 Induction of DEC1 and cyclin G2 (CCNG2) by (Yin et al., 1992; Herman and Simonson, 1995). hypoxia is dependent on HIF-1a. Cells were exposed to either Interestingly, several of the VHL regulated genes have normoxia (N; 20% O2) or hypoxia (H; 0.1% O2) for 16 h. potential roles in matrix metabolism. LRP1 is a large Ribonuclease protection analysis of DEC1, CCNG2 and GLUT-1 (positive control) in wild type CHO cells (C4.5) and HIF-1a endocytic receptor that mediates the catabolism of a de®cient CHO cells (Ka13). LC, internal control assay (constitu- number of molecules important in vascular biology tively expressed U6 small nuclear RNA) (Strickland et al., 1995), and appears to play a role in regulation of cell motility (Okada et al., 1996; Chazaud et al., 2000). MIC2/CD99 is a cell surface glycoprotein expressed at quanti®able levels in the 786-0 cells, involved in cell adhesion (Hahn et al., 1997) that is striking responses to hypoxia were observed in several commonly expressed in haematological neoplasms cell backgrounds. Since VHL has a critical function in (Soslow et al., 1997; Zhang et al., 2000) and the responses to hypoxia through regulation of HIF-a upregulated in Ewing family tumours (Zoubek et al., (Maxwell et al., 1999) this provides independent 1995). Transglutaminase 2, otherwise known as tissue evidence that these genes are likely to be genuine transglutaminase, is a multifunctional protein capable VHL targets in many cell types. Thus the results of activating phospholipase C and catalyzing the cross- suggest that inactivation of a tumour suppressor gene linking of extracellular proteins through the formation may have substantially di€erent e€ects on gene of e-(g-glutamyl)lysine cross-links (for review see Chen expression even in closely similar cell backgrounds. and Mehta, 1999). Di€erences between RCC4 and 786-0 cells in the In some cases the identi®ed gene forms part of a residual function of mutant VHL alleles, di€erential large family and it seems likely that several members expression of HIF-1a and HIF-2a subunits, or other may be co-ordinately regulated by the hypoxia path- genetic di€erences a€ecting transcriptional availability way. For instance, regulation of collagen type V, alpha of potentially responsive genes are all possible 1 may form part of a more general response of collagen explanations. metabolism to the hypoxia pathway. Prolyl 4-hydro- Following the recognition that pVHL has a critical xylase alpha 1, a key enzyme in collagen biosynthesis, function in the regulation of HIF-a, it has become has recently been identi®ed as a HIF-1 dependent clear that genes already de®ned as HIF targets will hypoxia inducible gene (Takahashi et al., 2000). commonly be upregulated in VHL defective cells. Aminopeptidase A is an ectopeptidase member of the However to understand more completely the role of peptidase family m1 proposed to play a role in the the HIF pathway in VHL disease it is necessary to progression of cervical neoplasms (Fujimura et al., consider the converse relationship; that is, to de®ne the 2000). Interestingly, its expression in tissue culture extent to which global e€ects of VHL status on gene correlates with resistance of renal carcinoma cells to expression patterns are mediated through activation of the antiproliferative e€ects of alpha-interferon (Nanus the hypoxia pathway. Therefore an important aim of et al., 1993). Aminopeptidase A shows a 54% identity the study was to test for regulation by oxygen among to aminopeptidase N, a gene recently found to promote genes that were newly identi®ed as responsive to VHL invasion and to be expressed on proliferating endothe- status. Of a total of 16 such genes tested for regulation lial cells undergoing angiogenesis (Pasqualini et al., by oxygen, the majority (10/16) showed greater than 2000). twofold regulation by hypoxia in VHL competent cells. For several genes de®ned in the screen and Of the remaining six genes, vimentin showed slight demonstrated to show a high level of induction by (1.9-fold) induction by hypoxia in RCC4/VHL cells, hypoxia, it is dicult to understand the role in tumour whereas the other ®ve genes showed little or no promotion. For instance cyclin G2 is a member of the regulation by oxygen. For three genes this occurred G class of cyclins (Horne et al., 1996), which are

Oncogene mRNA differential expression profiling to identify VHL target genes CC Wykoff et al 6303 unusual amongst cyclins so far described in having a with 10% foetal calf serum (Globepharm), L-glutamine cell cycle inhibitory e€ect. Its expression is increased in (2 mM), penicillin (50 IU/ml), and streptomycin sulphate response to DNA damaging agents, di€erentiation (50 mg/ml). Studies of inducible gene expression were signals and growth inhibitory drugs (Bates et al., performed on cells approaching con¯uence in normal growth 1996; Horne et al., 1997). DEC1/Stra13 (Shen et al., medium for 16 h. Parallel incubations were performed on aliquots of cells in normoxia (humidi®ed air with 5% CO2)or 1997), is a transcriptional repressor whose expression is hypoxia. Hypoxic conditions were generated in a Napco 7001 associated with induction of growth arrest and terminal incubator (Precision Scienti®c) with 0.1% O ,5%CO, and di€erentiation (Boudjelal et al., 1997; Sun and Taneja, 2 2 balance N2. 2000). Furthermore, in addition to its role in matrix cross linking, increased expression of transglutaminase Gene expression microarray (GEM) screening 2 characterizes cells undergoing apoptosis, and inhibi- tion protects against several apoptotic stimuli (Oliverio mRNA was extracted and puri®ed using oligo-dT (Dynal) et al., 1999 and references within). beads according to manufacturer's instructions. 600 ng of Thus for many genes, known functions are not easily each mRNA sample was reverse transcribed into cDNA and accommodated into a simple model of tumour labelled with either Cy3 or Cy5 ¯uorescent dye and applied to a competitive hybridization Human UniGEM V 1.0 by promotion by upregulated expression following VHL Incyte Genomics. Data were analysed using GEMToolsTM inactivation. Such genes may have other as yet software. unrecognized functions. However, the demonstration that many of the genes are upregulated in parallel as a consequence of dysregulation of the HIF pathway RNA analysis provides another potential explanation for this para- For ribonuclease protection assays (RPA), total RNA was dox. Such genes must normally form part of a co- extracted by a modi®ed acid/guanidinium thiocyanate/ ordinated physiological response to hypoxia that might phenol/chloroform method (RNAzol B, Cinna/Biotec La- well, in the context of tumour growth, have both pro- boratories), and dissolved in hybridization bu€er (80% and anti-tumorigenic e€ects. For instance the physio- formamide, 40 mM PIPES, 400 mM sodium chloride, and 1mM EDTA, pH8). To generate riboprobe templates, logical response to hypoxia might include limitation of cDNA fragments of human genes were cloned by reverse cellular proliferation under hypoxic conditions (Grae- transcription-PCR and ligated into pSP72 (Promega). For ber et al., 1994). In relation to this possibility it is of details of fragments cloned for each gene, see Appendix. The interest that di€erent studies of tumour xenografts of identities of the cloned fragments were con®rmed by HIF de®cient cells have demonstrated enhanced nucleotide sequence analysis. DNA templates for generating (Carmeliet et al., 1998), as well as reduced (Maxwell 32P-labelled RNA probes were linearized for 16 h with BglII et al., 1997; Ryan et al., 1998) tumour growth. and transcribed using SP6 RNA polymerase. Quanti®cation Balancing e€ects of di€erent genes upregulated as part of the protected species was performed using a phosphoi- of a pathway of co-ordinate physiological regulation mager (Molecular Dynamics), and related to an internal might also contribute to the tissue speci®city of the control assay for the constitutively expressed U6 small nuclear RNA (LC), performed for each assay as described VHL tumour syndrome. Furthermore, our results (Maxwell et al., 1999). Glucose transporter-1 mRNA assays suggest that selection of cells capable of evading were used as a positive control for the conditions of hypoxic growth-inhibition/apoptosis is likely to be an early stimulation. Independent aliquots of 30 mg were analysed for step in VHL-mutant associated tumour progression. each gene. In summary these studies have demonstrated, using an unbiased set of newly identi®ed genes that are responsive to VHL, that activation of the HIF system Abbreviations makes a quantitatively dominant contribution to the VHL, von Hippel-Lindau; HIF-1, hypoxia inducible factor- changes in gene expression associated with the VHL 1;RCC,renalcellcarcinoma;RPA,ribonucleaseprotec- status of renal carcinoma cells. Most probably, tion assay; GEM, gene expression microarray; LC, U6 however, VHL status exerts e€ects on gene expression small nuclear RNA; GLUT-1, glucose transporter 1; through other pathways. Further studies of the new VEGF, vascular endothelial growth factor; EDN1, en- classes of VHL regulated and HIF regulated genes will dothelin-1; DEC1, di€erentiated embryo chrondocyte 1; be of interest both in understanding VHL associated Stra13, stimulated with retinoic acid 13; MIC2, mono- tumours and the role of microenvironmental hypoxia clonal, Imperial Cancer Research Fund 2; TGM2, trans- in other types of cancer. glutaminase 2; LRP1, low density lipoprotein receptor- related protein 1; COL5A1, collagen type V, alpha 1; CCNG2, cyclin G2; APA, aminopeptidase A; VIM, vimentin; SGNE1, secretory granule neuroendocrine pro- Materials and methods tein 1; SAT, spermidine N1-acetyltransferase; KRTHB1, keratin, hair, basic, 1; CTSC, cathepsin C; K18, keratin 18; Cell lines EST, expressed sequence tag; IF, I factor; CHO, Chinese hamster ovary. Stable RCC4 transfectants expressing vector backbone alone or wild type pVHL were as described (Maxwell et al., 1999). Stable transfectants of 786-0 cells expressing vector backbone alone or wild type pVHL were a gift from WG Kaelin. A549, EJ-28, and HBL-100 cell lines were from ECACC. Chinese Acknowledgements hamster ovary parental cell line, C4.5, and the HIF-1a This work was supported by the Wellcome Trust, the de®cient derivative, Ka13, were as described (Wood et al., Medical Research Council, and the Imperial Cancer 1998). Cells were grown in DMEM (Sigma) supplemented Research Fund.

Oncogene mRNA differential expression profiling to identify VHL target genes CC Wykoff et al 6304 References

Bates S, Rowan S and Vousden KH. (1996). Oncogene, 13, Lisztwan J, Imbert G, Wirbelauer C, Gstaiger M and Krek 1103 ± 1109. W. (1999). Genes Dev., 13, 1822 ± 1833. Boudjelal M, Taneja R, Matsubara S, Bouillet P, Dolle P and Maxwell P, Wiesener M, Chang G-W, Cli€ord S, Vaux E, Chambon P. (1997). Genes Dev., 11, 2052 ± 2065. Cockman M, Wyko€ C, Pugh C, Maher E and Ratcli€e P. Carmeliet P, Dor Y, Herbert JM, Fukumura D, Brusselmans (1999). Nature, 399, 271 ± 275. K, Dewerchin M, Neeman M, Bono F, Abramovitch R, Maxwell PH, Dachs GU, Gleadle JM, Nicholls LG, Harris MaxwellP,KochCJ,Ratcli€eP,MoonsL,JainRK, AL, Stratford IJ, Hankinson O, Pugh CW and Ratcli€e Collen D and Keshet E. (1998). Nature, 394, 485 ± 490. PJ. (1997). Proc. Natl. Acad. Sci. USA, 94, 8104 ± 8109. Chazaud B, Bonavaud S, Plonquet A, Pouchelet M, Moch H, Schraml P, Bubendorf L, Mirlacher M, Kononen J, Gherardi RK and Barlovatz-Meimon G. (2000). Exp. Cell Gasser T, Mihatsch MJ, Kallioniemi OP and Sauter G. Res., 258, 237 ± 244. (1999). Am.J.Pathol.,154, 981 ± 986. Chen JS and Mehta K. (1999). Int. J. Biochem. Cell Biol., 31, Nanus DM, Engelstein D, Gastl GA, Gluck L, Vidal MJ, 817 ± 836. Morrison M, Finstad CL, Bander NH and Albino AP. Cockman ME, Masson N, Mole DR, Jaakkola P, Chang (1993). Proc. Natl. Acad. Sci. USA, 90, 7069 ± 7073. GW, Cli€ord SC, Maher ER, Pugh CW, Ratcli€e PJ and Ohh M, Yauch RL, Lonergan KM, Whaley JM, Stemmer- Maxwell PH. (2000). J. Biol. Chem., 275, 25733 ± 25741. Rachamimov AO, Louis DN, Gavin BJ, Kley N, Kaelin Jr Ebert BL, Firth JD and Ratcli€e PJ. (1995). J. Biol. Chem., WG and Iliopoulos O. (1998). Mol. Cell, 1, 959 ± 968. 270, 29083 ± 29089. Okada SS, Grobmyer SR and Barnathan ES. (1996). Fujimura H, Ino K, Nagasaka T, Nakashima N, Nakazato Arterioscler. Thromb. Vasc. Biol., 16, 1269 ± 1276. H, Kikkawa F and Mizutani S. (2000). Oncology, 58, 342 ± Okuda Y, Tsurumaru K, Suzuki S, Miyauchi T, Asano M, 352. Hong Y, Sone H, Fujita R, Mizutani M, Kawakami Y, GnarraJR,ToryK,WengY,SchmidtL,WeiMH,LiH, Nakajima T, Soma M, Matsuo K, Suzuki H and LatifF,LiuS,ChenF,DuhF-M,LubenskyI,DuanDR, Yamashita K. (1998). Life Sci., 63, 477 ± 484. Florence C, Pozzatti R, Walther MM, Bander NH, Oliverio S, Amendola A, Rodolfo C, Spinedi A and Grossman HB, Brauch H, Pomer S, Brooks JD, Isaacs Piacentini M. (1999). J. Biol. Chem., 274, 34123 ± 34128. WB, Lerman MI, Zbar B and Linehan WM. (1994). Pasqualini R, Koivunen E, Kain R, Lahdenranta J, Nature Genet., 7, 85 ± 90. Sakamoto M, Stryhn A, Ashmun RA, Shapiro LH, Arap Graeber TG, Peterson JF, Tsai M, Monica K, Fornace Jr AJ W and Ruoslahti E. (2000). Cancer Res., 60, 722 ± 727. and Giaccia AJ. (1994). Mol. Cell. Biol., 14, 6264 ± 6277. Pause A, Lee S, Lonergan KM and Klausner RD. (1998). Hahn JH, Kim MK, Choi EY, Kim SH, Sohn HW, Ham DI, Proc. Natl. Acad. Sci. USA, 95, 993 ± 998. Chung DH, Kim TJ, Lee WJ, Park CK, Ree HJ and Park Ryan HE, Lo J and Johnson RS. (1998). EMBO J., 17, SH. (1997). J. Immunol., 159, 2250 ± 2258. 3005 ± 3015. Herman WH and Simonson MS. (1995). J. Biol. Chem., 270, Schena M, Shalon D, Davis RW and Brown PO. (1995). 11654 ± 11661. Science, 270, 467 ± 470. Hieda HS and Gomez-Sanchez CE. (1990). Life Sci., 47, Semenza GL, Roth PH, Fang H-M and Wang GL. (1994). J. 247 ± 251. Biol. Chem., 269, 23757 ± 23763. Horne MC, Donaldson KL, Goolsby GL, Tran D, Shen M, Kawamoto T, Yan W, Nakamasu K, Tamagami M, Mulheisen M, Hell JW and Wahl AF. (1997). J. Biol. Koyano Y, Noshiro M and Kato Y. (1997). Biochem. Chem., 272, 12650 ± 12661. Biophys. Res. Commun., 236, 294 ± 298. Horne MC, Goolsby GL, Donaldson KL, Tran D, Neubauer ShweikiD,ItinA,So€erDandKeshetE.(1992).Nature, M and Wahl AF. (1996). J. Biol. Chem., 271, 6050 ± 6061. 359, 843 ± 845. Hu J, Discher DJ, Bishopric NH and Webster KA. (1998). Soslow RA, Bhargava V and Warnke RA. (1997). Hum. Biochem. Biophys. Res. Commun., 245, 894 ± 899. Pathol., 28, 1158 ± 1165. Iliopoulos O, Levy AP, Jiang C, Kaelin Jr WG and Goldberg Strickland DK, Kounnas MZ and Argraves WS. (1995). MA. (1996). Proc. Natl. Acad. Sci. USA, 93, 10595 ± FASEB J., 9, 890 ± 898. 10599. Sun H and Taneja R. (2000). Proc. Natl. Acad. Sci. USA, 97, Ivanov SV, Kuzmin I, Wei M-H, Pack S, Geil L, Johnson 4058 ± 4063. BE, Stanbridge EJ and Lerman MI. (1998). Proc. Natl. Takahashi Y, Takahashi S, Shiga Y, Yoshimi T and Miura Acad. Sci. USA, 95, 12596 ± 12601. T. (2000). J. Biol. Chem., 275, 14139 ± 14146. Iwai K, Yamanaka K, Kamura T, Minato N, Conaway RC, Wood SM, Wiesener MS, Yeates KM, Okada N, Pugh CW, Conaway JW, Klausner RD and Pause A. (1999). Proc. Maxwell PH and Ratcli€e PJ. (1998). J. Biol. Chem., 273, Natl. Acad. Sci. USA, 96, 12436 ± 12441. 8360 ± 8368. Iyer VR, Eisen MB, Ross DT, Schuler G, Moore T, Lee JCF, Yin J, Lee JA and Howells RD. (1992). Brain Res. Mol. Brain Trent JM, Staudt LM, Hudson Jr J, Boguski MS, Res., 14, 213 ± 220. Lashkari D, Shalon D, Botstein D and Brown PO. Zhang PJ, Barcos M, Stewart CC, Block AW, Sait S and (1999). Science, 283, 83 ± 87. Brooks JJ. (2000). Mod. Pathol., 13, 452 ± 458. Kaelin WG and Maher ER. (1998). Trends Genet., 14, 423 ± Zoubek A, P¯eiderer C, Ambros PF, Kronberger M, 426. Dworzak MN, Gruber B, Luegmayer A, Windhager R, Koong AC, Denko NC, Hudson KM, Schindler C, Swiersz Fink FM, Urban C, Koscielniak E, Gardner H and Kovar L, Koch C, Evans S, Ibrahim H, Le QT, Terris DJ and H. (1995). Klin. Padiatr., 207, 242 ± 247. Giaccia AJ. (2000). Cancer Res., 60, 883 ± 887. Kozawa O, Kawamura H, Hatakeyama D, Matsuno H and Uematsu T. (2000). Cell Signal, 12, 375 ± 380.

Oncogene mRNA differential expression profiling to identify VHL target genes CC Wykoff et al 6305 Appendix Sequence and length of riboprobe templates Accession Protected Abbreviation Gene number 5' end 3' end length

APA aminopeptidase A L12468 1256 1439 183 ATPase Na+/K+ transporter, beta 3 U51478 178 357 179 CINP cardiac ankyrin repeat protein X83703 893 1118 225 CTSC cathepsin C U79415 1301 1450 149 COLIA1 collagen type 1, alpha 1 Z74615 4055 4203 148 COL5A1 collagen type V, alpha 1 M76729 4783 4920 137 CCNG2 cyclin G2 U47414 1531 1658 127 DEC1 DEC1 AB004066 851 1050 199 EDN1 endothelin 1 Y00749 276 424 148 EST (Incyte ID: 1797836) AA234494 158 327 169 EST (Incyte ID: 1909292) R48254 65 230 165 FTL ferritin, light polypeptide M11147 429 590 161 IF I factor T28390 721 916 195 ITGA5 integrin, alpha 5 X06256 2407 2596 189 JM4 JM4 protein AJ005896 870 1051 181 K18 keratin 18 NM000224 392 594 202 KRTHB1 keratin, hair, basic, 1 X81420 512 654 142 L1CAM L1 cell adhesion molecule M77640 2166 2372 206 LRP1 LRP1 X13916 10741 10901 160 MIC2 MIC2 AI061430 202 369 167 SGNE1 secretory granule neuroendocrine protein 1 Y00757 489 657 168 SAT spermidine N1-acetyltransferase M55580 26 163 137 SYNJ2 synaptojanin 2 AB002346 1099 1280 181 TAGLN transgelin M95787 150 315 165 TGM2 transglutaminase 2 NM004613 1374 1552 178 VIM vimentin X56134 1102 1278 176 ZNF184 zinc finger protein 184 U66561 1097 1231 134

Oncogene