- Email: [email protected]
The von Hippel–Lindau tumour suppressor protein: new perspectives
Reviews
MOLECULAR MEDICINE TODAY, JUNE 1999 (VOL. 5)
The von Hippel–Lindau tumour suppressor protein: new perspectives Michael Ohh and William G. Kaelin Jr
von HippelÐLindau (VHL) disease is a hereditary cancer syndrome caused by germline mutations of the VHL tumour suppressor gene. The VHL gene product, pVHL, forms multiprotein complexes that contain elongin B, elongin C and Cul-2, and negatively regulates hypoxia-inducible mRNAs. pVHL is suspected to play a role in ubiquitination given the similarity of elongin C and Cul-2 with Skp1 and Cdc53, respectively. pVHL can also interact with fibronectin and is required for the assembly of a fibronectin matrix. Finally, pVHL, at least indirectly, plays a role in the ability of cells to exit the cell cycle. Thus, pVHL is a tumour suppressor protein that regulates angiogenesis, extracellular matrix formation and the cell cycle. INHERITANCE of a mutant allele of the von Hippel–Lindau (VHL) tumour suppressor gene causes VHL disease, which is characterized by a range of cancers, including retinal angiomas, cerebellar and spinal haemangioblastomas, renal cell carcinomas, phaeochromocytomas, pancreatic adenomas and islet cell tumours, epididymal cystadenomas and endolymphatic sac tumours of the inner ear1. Renal carcinomas and central nervous system (CNS) haemangioblastomas remain the major causes of morbidity and mortality for VHL patients1. VHL disease affects 1 in 36 000 individuals and clinically displays an autosomal dominant pattern of inheritance with high penetrance1. However, at the molecular level, VHL disease is autosomal recessive insofar as tumours arise only after the remaining wild-type VHL allele is somatically mutated or silenced1.
Picture kindly provided by Dr Alain Gaudric, Hopital Lariboisiere, Paris, France.
Cloning of the VHL gene
Michael Ohh PhD Research Associate William G. Kaelin Jr* MD Associate Professor Howard Hughes Medical Institute, Division of Adult Oncology, Dana-Farber Cancer Institute and Dept of Medicine, Harvard Medical School, Boston, MA 02115, USA. *e-mail:[email protected]
The VHL gene was cloned in 1993 (Ref. 2), and intragenic germline mutations in members of VHL kindreds have confirmed the authenticity of the gene1. The VHL gene is widely expressed in both fetal and adult tissues, demonstrating that expression of the VHL transcript is not restricted to organs affected by VHL disease1. Nonetheless, the pattern of expression in the fetal kidney is consistent with a role in normal renal tubular development and differentiation3,4. The VHL coding sequence is contained within three exons, and two alternatively spliced mRNAs that reflect the presence (isoform I) or absence (isoform II) of exon 2 have been detected5. Some VHL patients harbour VHL mutations that lead to the exclusive production of isoform II. This suggests that the VHL mRNA isoform II is not sufficient to inhibit tumorigenesis5. In keeping with Knudson’s two-hit model, functional inactivation or loss of both VHL alleles has been demonstrated in the majority of sporadic clear cell renal carcinomas and cerebellar haemangioblastomas that have so far been examined1. Moreover, reports that de-
1357-4310/99/$ - see front matter © 1999 Elsevier Science. All rights reserved. PII: S1357-4310(99)01481-1
257
Reviews
MOLECULAR MEDICINE TODAY, JUNE 1999 (VOL. 5)
scribe loss of heterozygosity at the VHL locus in early premalignant lesions of the kidney, such as atypical cysts6, suggest that inactivation of the VHL tumour suppressor protein (pVHL) is an early step in the pathogenesis of clear cell renal carcinoma. In this regard, VHL might play a ‘gatekeeper’ role with respect to renal carcinogenesis, similar to that of APC (adenomatous polyposis coli) in colorectal carcinogenesis, as proposed by Kinzler and Vogelstein7. Moreover, VHL is required for embryogenesis, as demonstrated by murine gene knockout experiments8. VHL mutations can be detected in ~100% of patients who carry a clinical diagnosis of VHL disease9. VHL mutations are extremely heterogeneous and are widely distributed throughout the open reading frame10. Specific genotype–phenotype correlations are emerging in VHL disease. Thus, although retinal angiomas and CNS haemangioblastomas are hallmarks of the disease, some families display a high risk of concurrent phaeochromocytoma (type 2 VHL) but others do not (type 1 VHL)10. Type 2 VHL families can be further subdivided into type 2A (low risk of renal cell carcinoma), type 2B (high risk of renal cell carcinoma) and type 2C (exclusively phaeochromocytoma without other characteristics of VHL disease)1. In general, missense mutations are associated with the development of type 2 disease, whereas mutations that lead to truncated versions of pVHL and gross deletions are primarily associated with type 1 disease1. Hence, although speculative, the development of phaeochromocytoma might be linked to a partial, rather than a complete, loss of pVHL function(s) or a gain of an unknown function. With the increasing emergence of these genotype–phenotype correlations, it seems plausible that pVHL has multiple functions, some of which might be tissue-specific.
that pVHL can shuttle back and forth between the nucleus and cytoplasm in response to certain signals. pVHL19 appears to be equally distributed between the nucleus and cytoplasm and does not associate with membranes12. In this regard, it is noteworthy that pVHL19 lacks an N-terminal acid pentameric repeat that resembles a sequence found in Trypanosoma brucei membrane protein2. Whether pVHL19 shuttles in response to particular cellular stimuli is not known.
The evolving model of tumour suppression by pVHL The primary sequence of pVHL does not resemble that of any protein of known function. With the hope of discerning the function(s) of pVHL, several groups have sought to identify cellular proteins that are capable of binding to pVHL. The experiments carried out to date have identified several pVHL-associated proteins. These are discussed below (Fig. 1).
Elongins B and C Elongins B and C bind stably to each other to form a binary complex that was first identified by virtue of its ability to stimulate the transcriptional elongation activity of elongin A (Ref. 20). This trimeric elongin–SIII complex enhances the rate of elongation by RNA polymerase II by suppressing the basal transcription machinery from pausing or stalling along the DNA template20. The minimal region of pVHL required for binding elongin B–C complex in vitro, corresponding to C-terminal residues 157–172, is the only region of pVHL that is highly similar to elongin A (Refs 20,21). As expected, the homologous region of elongin A likewise interacts with elongin B–C complex22, thereby establishing this stretch of amino acid residues as an elongin B–C-binding motif. Furthermore, binding of the elongin B–C complex to pVHL and elongin A is, at least in vitro,
A single VHL gene and its two products The VHL mRNA encodes two protein products (pVHL) with apparent molecular weights of ~30 and 19 kDa (Refs 11,12). The former contains 213 amino acid residues and the latter contains 160 amino acid residues as a result of internal translation from the second methionine at residue 54 within the VHL open reading frame. The first and second methionine codons are flanked by residues that conform roughly to Kozak’s rules for translation initiation sites13,14. Reintroduction of either 30-kDa (pVHL30) or 19-kDa (pVHL19) pVHL into VHL–/– renal carcinoma cells suppresses their ability to form tumours in nude mice11,15,16. This observation might account for the finding that VHL mutations almost invariably map C-terminal to residue 54 (Ref. 1). Specifically, mutations affecting the first 54 amino acids would, in principle, still allow the production of wild-type pVHL19 (Refs 11,12). Biochemical fractionation and immunohistochemical studies suggest that pVHL30 is predominantly cytoplasmic1. However, a significant fraction of pVHL30 can also be found in the nucleus and in association with the endoplasmic reticulum (ER)12,17,18. When artificially overproduced in mammalian cells, the subcellular localization of pVHL appears to vary in a cell-density-dependent manner19. This observation raises the possibility 258
Cul-2 B
C
PKCd pVHL
FN
PKCz
Sp1
VBP-1
Figure 1. von HippelÐLindau tumour suppressor protein (pVHL)-associated proteins. Several cellular proteins have been shown to bind, directly or indirectly, to pVHL, including elongin B (B), elongin C (C), Cul-2, fibronectin (FN), VHL binding protein 1 (VBP-1), the transcription factor Sp1 and protein kinase C (PKC) isoforms d and z. The biochemical interaction of pVHL with FN, PKC and Sp1 requires independent confirmation.
Reviews
MOLECULAR MEDICINE TODAY, JUNE 1999 (VOL. 5)
mutually exclusive23. This almost certainly accounts for the ability of pVHL to inhibit elongin–SIII activity in vitro23. The elongin-binding domain of pVHL is a frequent site of mutations in VHL kindreds, suggesting that elongin binding contributes to tumour suppression by pVHL (Ref. 1). The above considerations led to a model in which tumour suppression by pVHL would result, at least in part, from inhibition of elongin–SIII in vivo21,23. Although not formally disproven, several lines of evidence suggest that this model is incorrect. Firstly, there is no evidence that elongin–SIII regulates the transcription of specific genes in vivo. Secondly, and perhaps more importantly, elongins B and C are in vast molar excess with respect to both elongin A and pVHL (Ref. 24). Indeed, recent work suggests that pVHL and elongin A are but two examples of a growing family of proteins that contain similar, colinear, elongin B–C-binding domains24.
Cullin 2 Two groups have independently shown that pVHL binds to cullin 2 (Cul-2), a member of the cullin family18,25. Cullins were first identified in Caenorhabditis elegans during a screen for mutants that displayed excess postembryonic cell divisions26. Loss of Cul-1 function resulted in hyperplasia of diverse tissue types owing to a failure to generate or to respond to the normal developmental signals to properly exit the cell cycle26. This suggests that Cul-1 functions as a negative regulator of the cell cycle and is required for cells to enter either G0 or the apoptotic pathway. There are now known to be at least three cullins in budding yeast, five in nematodes and six in humans27. Notably, Cul-1 and Cul-2 bear striking similarity to yeast Cdc53 (Ref. 27), which plays a role in the ubiquitination, and hence degradation, of certain proteins27. Cdc53 function is intricately tied to its ability to bind Cdc34, which functions as an E2 ubiquitin-conjugating enzyme; to Skp1 (S-phase kinase-associated protein 1, suppressor of Cdk inhibitor proteolysis and of kinetochore protein), which probably acts as an assembly factor; and an F-box protein (so called because of apparent homology to cyclin F), such as Grr1 or Cdc4, which probably confers substrate/target specificity27. A complex that consists of Skp1, Cdc53 and an F-box protein has been dubbed the SCF (Ref. 28). For example, SCFCdc4 specifically targets phosphorylated Sic1 for ubiquitination, whereas SCFGrr1 promotes phosphorylated cyclin 1 (Cln1) and cyclin 2 (Cln2) for ubiquitin-mediated proteolysis, both of which are required for G1–S transition27. Recently, Cul-1 has been demonstrated, in humans, to bind Skp1, Cdc34 and an F-box-containing protein, Skp2. This SCFSkp2 complex interacts with cyclin A/Cdk2 and is thought to play a role in S-phase entry and traversal29. Intriguingly, the elongin–SIII complex is remarkably similar to the SCF complex. Elongins A, B and C contain regions of similarity to an F-box protein, ubiquitin and Skp1, respectively25,30. Nonetheless, there is currently no evidence that elongin A interacts with a cullin. In contrast, pVHL binds to Cul-2 via elongins B and C (Ref. 25). Thus, the elongin B–C complex bridges pVHL to Cul-2 in a manner analogous to Skp1 of the SCF complex joining Cdc53 to an F-box protein. Although pVHL does not have a recognizable F-box, it might still play a role similar to that of F-box-containing proteins – that is, it might function to confer substrate specificity. A potential precedent for F-box-independent substrate recognition by ubiquitin-conjugating enzymes is provided by examination of APC (the anaphase-promoting complex or cyclosome)31. APC, like SCF, contains a cullin homologue called Apc2 (Ref. 27). In budding yeast, Cdc20 and Cdh1/Hct1,
Glossary Allele Ð One of two or more forms of a gene that exist at a single locus. Angiogenesis Ð The formation of new blood vessels. Cell cycle Ð The progression through all the stages that are necessary for DNA replication and cell division. Cullin Ð Member of a family of proteins with sequence homology to yeast Cdc53. Elongin Ð A trimeric complex, containing elongin A, elongin B and elongin C, that enhances the rate of transcriptional elongation by RNA polymerase II in vitro by suppressing the basal transcription machinery from pausing or stalling along the DNA template. KnudsonÕs two-hit model Ð A model that suggests that the inactivation of both the maternal and paternal alleles of a tumour suppressor gene are necessary for tumour development. Loss of heterozygosity Ð The condition that results from loss of maternal or paternal DNA at a particular genetic locus. Transcriptional elongation Ð The process by which ribonucleotides are processively added to a nascent RNA transcript. Tumour suppressor gene Ð A gene the inactivation of which contributes to carcinogenesis. Ubiquitin-mediated proteolysis Ð A mechanism for the removal of specific proteins in which the protein to be eliminated is covalently linked to ubiquitin and subsequently degraded by the proteosome.
two components of the APC, lack F-boxes and yet target proteins such as Pds1 and Ase1, respectively, for ubiquitination27. In light of this knowledge, pVHL might play a role in the ubiquitination of certain proteins (Fig. 2). For example, the pVHL–elongin–Cul-2 (VEC) complex might degrade certain proteins involved in the stabilization of hypoxia-inducible mRNAs (see below) or in the degradation of malfolded proteins present in the ER. In this regard, Gorospe et al. reported that cells that lack pVHL are more readily killed by certain forms of ER stress than their wild-type counterparts32. Furthermore, they documented the abnormal accumulation of ubiquitinated proteins in pVHL-defective cells. These latter observations are consistent with a role of pVHL, whether direct or indirect, in ubiquitination. Based on the association of pVHL with Cul-2, Pause et al. examined whether pVHL plays a role in cell-cycle control33. They found that pVHL was required for cells to exit the cell cycle in the absence of serum and correlated this requirement with changes in p27Kip1 levels in VHL+/+ versus VHL–/– cells under these conditions33. However, there is no evidence that p27 is a direct target of the VEC complex. Furthermore, the effect of pVHL in these assays was influenced by various cell culture parameters, including cell density and the timing of cell plating and media changes (A. Pause and S. Lee, unpublished). This observation is intriguing given the fact that several of the pVHL target genes identified to date encode secreted proteins, including mitogens, or proteins that play a role in extracellular pH (see below; Box 1). 259
Reviews
MOLECULAR MEDICINE TODAY, JUNE 1999 (VOL. 5)
a SCF
Cdc4
Sic1 Skp1
Cdc4
Cdc34 Ubq Cdc53
b VEC X C
pVHL
?E2
B
Ubq Cul-2
c APC
Pds1 Ase1 Ubq Clb2
Cdh1Ð Hct1
Cdc20 Apc2
Figure 2. Comparison between SCF (a complex that comprises Skp1, Cdc53 and an F-box protein), VEC [von HippelÐLindau tumour suppressor protein (pVHL)ÐelonginÐCul-2 complex] and APC (the anaphase-promoting complex) complexes. (a) In budding yeast, Cdc53 binds to Skp1 and an F-box protein, such as Cdc4 (SCFCdc4 complex). This complex, in association with the E2 ubiquitin-conjugating protein Cdc34, selectively targets phosphorylated Sic1 for ubiquitination. Subsequent destruction of the polyubiquitinated Sic1 by the 26S proteosome allows G1ÐS transition. (b) pVHL binds to elongin C (C) and Cul-2, which are similar to Skp1 and Cdc53, respectively. A current model suggests that the VEC complex interacts with a cognate E2 to ubiquitinate an undefined protein (ÔXÕ) for degradation. (c) APC, like SCF and VEC, contains a Cdc53 homologue (Apc2). Cdc20, which lacks an F-box, when complexed to APC triggers the destruction of Pds1. This promotes the segregation of chromosomes at the metaphaseÐanaphase transition. APC then forms a second complex with Cdh1ÐHct1, which also lacks an F-box, to direct the ubiquitination of the spindle protein Ase1 and the mitotic cyclin Clb2.
Fibronectin Recently, pVHL has been demonstrated to bind specifically, at least indirectly, to fibronectin in association with the ER (Ref. 17). This interaction is abrogated by every disease-causing mutation that has been examined to date, including those shown or predicted not to influence the binding of pVHL to elongin B–C–Cul-2 complex, emphasizing the potential significance of fibronectin in the ability of pVHL to suppress neoplastic growth17. Moreover, VHL null tumours, as well as cells derived from VHL knockout mouse embryo fibroblasts, fail to assemble a proper fibronectin matrix17. Loss of extra260
cellular fibronectin matrix is a hallmark of many transformed cells34. Conversely, fibronectin or overproduction of a fibronectin receptor, such as a5b1 integrin, can, in certain experimental settings, promote differentiation and revert some aspects of malignant behaviour35. Thus promotion of fibronectin matrix assembly might contribute to tumour suppression by pVHL. This notion is supported by a recent study in which VHL+/+ and VHL–/– renal carcinoma cells were grown as three-dimensional spheroids in vitro36. Cells that lacked pVHL grew as tightly packed, amorphous spheroids. In contrast, cells that produced pVHL formed loose aggregates which, on microscopic and ultrastructural examination, exhibited evidence of epithelial differentiation, including the formation of trabecular and tubular structures36. These differences correlated with differences in fibronectin matrix deposition, raising the possibility that fibronectin played an active role in the differentiation of the VHL+/+ cells under these conditions36. How might loss of pVHL lead to failure to assemble a fibronectin matrix? Preliminary data suggest that pVHL-defective cells are capable of secreting fibronectin but that the fibronectin, once secreted, fails to polymerize properly17. One tentative model, based on analogy with other secreted proteins37, is that pVHL plays a role in the elimination of malfolded or malprocessed fibronectin. Specifically, malfolded or malprocessed fibronectin might undergo retrograde transport to the cytoplasmic surface of the ER membrane whereupon it would be acted upon by the VEC complex. If this were a coupled process, loss of pVHL might lead to the secretion of malfolded/malprocessed fibronectin that could then interfere with the polymerization of fibronectin into a recognizable fibrillar array.
VBP-1
Hino and co-workers identified VHL-binding protein 1 (VBP-1) in a yeast two-hybrid screen, using pVHL as bait38. Furthermore, truncation of the Cterminal 26 amino acids of pVHL abolished the binding of VBP-1 (Ref. 38). In the absence of pVHL, VBP-1 is exclusively localized in the cytosol38. However, when co-expressed with an epitope-tagged pVHL, VBP-1 can be detected in the nucleus, indicating that VBP-1, when complexed to pVHL, can translocate to the nucleus38. It is not known whether pVHL binds to VBP-1 under physiological conditions, and no functional significance has been ascribed to the interaction of pVHL with VBP-1.
Sp1 Sukhatme and co-workers reported that pVHL can inhibit transcription from the vascular endothelial growth factor/vascular permeability factor (VEGF/VPF) promoter39. The minimal region of the VEGF promoter necessary for this effect was found to contain binding sites for Sp1 (Ref. 39). Furthermore, pVHL was demonstrated to bind
Reviews
MOLECULAR MEDICINE TODAY, JUNE 1999 (VOL. 5)
Box 1. The downstream targets of the von HippelÐLindau tumour suppressor protein (pVHL)
• • • • • •
Vascular endothelial growth factor (VEGF) Glucose transporter 1 (GLUT-1) Platelet-derived growth factor B (PDGF-B) Transforming growth factor a (TGF-a) Carbonic anhydrase 9 (CA9) Carbonic anhydrase 12 (CA12)
directly to Sp1, at least in vitro, suggesting a model in which pVHL inhibits Sp1-mediated transcription by sequestering Sp1 (Ref. 39). Thus, in VHL disease, a loss of functional pVHL might lead to a loss of Sp1 inhibition, which might in turn lead to an inappropriate expression of VEGF mRNA (see also below).
PKC Mukhopadhyay and co-workers showed that the binding of pVHL to two protein kinase C (PKC) isoforms, PKCd and PKCz, inhibited the transcriptional activation of VEGF (Refs 40,41). Furthermore, PKCz was demonstrated to directly bind and phosphorylate the zinc finger domain of Sp1 (Ref. 41). Overexpression of PKCz led to increased transcription from the VEGF promoter41. More importantly, pVHL blocked the interaction between Sp1 and PKCz, thereby suppressing the phosphorylation of Sp1 and hence downregulating the Sp1dependent transcriptional activation of VEGF (Ref. 41) (see below).
Downstream targets of pVHL pVHL regulates, at least indirectly, the abundance of a variety of mRNAs (see Box 1). These include the mRNAs that encode VEGF/VPF, platelet-derived growth factor B (PDGF-B), glucose transporter 1 (GLUT-1), transforming growth factor a (TGF-a) and carbonic anhydrases (CA) 9 and 12 (Refs 1,42). The search for pVHL target genes was initially aided by the knowledge that VHLassociated neoplasms are typically hypervascular and, on occasion, produce a syndrome that is characterized by the excessive production of red blood cells (secondary polycythaemia or paraneoplastic erythrocytosis)1. The former has been linked to the overproduction of angiogenic peptides, such as VEGF/VPF, whereas the latter has been attributed to the excessive production of erythropoietin (EPO) by the tumour cells1. The production of VEGF and EPO are normally triggered by hypoxia, presumably to augment the delivery of oxyhaemoglobin to cells1. This raised the possibility that pVHL played a role in certain stress responses, such as changes in ambient oxygen tension. Indeed, VHL–/– tumour cells produce high levels of hypoxia-inducible mRNAs, such as the VEGF and GLUT-1 mRNAs, under both hypoxic and normoxic conditions1. Reintroduction of wild-type, but not mutant, pVHL into these cells specifically inhibited the production of hypoxia-inducible mRNAs under normoxic conditions, thus restoring their hypoxia-inducible profile1. Thus, the hypervascular nature of VHL-associated tumours is probably linked to the overproduction of hypoxia-inducible mRNAs, including those that encode angiogenic peptides. Moreover, the inhibition of VEGF has been demonstrated to inhibit tumorigenesis in vivo and thus might contribute to the ability of pVHL to suppress tumour formation1.
The outstanding questions Does the von Hippel–Lindau tumour suppressor protein •(pVHL)–elongin–Cul-2 complex play a role in ubiquitination and, if so, what are its targets?
negative regulation of hypoxia-inducible genes by pVHL •dueIsprimarily to transcriptional effects, post-transcriptional effects, or both? Might the relative contributions of these two mechanisms differ in different cell types? Why is pVHL required for assembly of a fibronectin matrix? What physiological relevance can be ascribed to the ability of pVHL to shuttle between the cytoplasm and nucleus? What is the biochemical basis of the genotype–phenotype correlations observed in pVHL disease?
• • •
There are conflicting data as to whether the effect of pVHL on VEGF is primarily transcriptional or post-transcriptional. As noted above, two reports suggested that pVHL, directly or indirectly, inhibits Sp1 and hence transcription of the VEGF promoter39,41. In contrast, two groups have reported that the effect of pVHL on VEGF mRNA abundance was largely due to changes in VEGF mRNA stability16,43. In particular, pVHL was required for the destabilization of the VEGF mRNA under well-oxygenated conditions16,43. Intriguingly, the 39-untranscribed regions of hypoxia-inducible mRNAs, such as VEGF mRNA and GLUT-1 mRNA, contain an AUUUA-rich element, which can bind to as yet unidentified proteins in RNA gel-shift assays44. The binding of these proteins correlates with mRNA stabilization44. In normal cells, these RNA-binding factors are only detected under hypoxic conditions. In contrast, these factors are detected under both hypoxic and normoxic conditions in cells that lack pVHL (Ref. 44). Thus, a tentative model for pVHL action would suggest that the VEC complex regulates the stability and/or function of these, as yet, unidentified RNA-binding proteins (Fig. 3). Perhaps in keeping with this view, Lonergan et al. showed that binding to elongin B, elongin C and Cul-2 was necessary, but not sufficient, for pVHL to regulate hypoxia-inducible mRNAs (Ref. 25). Lerman and co-workers, using differential display, found that carbonic anhydrase 9 (CA9) and carbonic anhydrase 12 (CA12) were upregulated in VHL-defective renal carcinoma cells42. This finding is intriguing given earlier studies that showed that these proteins are frequently overproduced in renal cell carcinomas and might serve as markers for this disease42. Furthermore, CA9 and CA12 might affect peritumoral pH and thus might indirectly affect tumour cell growth, based on earlier observations that the extracellular pH of tumours is generally more acidic than normal cells and that, at least in vitro, acidic pH favours the growth, spreading and invasive properties of tumour cells45.
Concluding remarks The von Hippel–Lindau tumour suppressor protein (pVHL) forms specific complexes with several cellular proteins. Some of these interactions, such as binding to fibronectin, Sp1 and certain PKC isoforms, require independent confirmation. In contrast, there is general agreement that pVHL binds to elongin B, elongin C and Cul-2. These complexes are thought to play a role in ubiquitination, based largely 261
Reviews
MOLECULAR MEDICINE TODAY, JUNE 1999 (VOL. 5)
supported by the Medical Research Council of Canada and W.G.K. is a Howard Hughes Medical Institute assistant investigator.
References
Hypoxia
B pVHL
C
Cul-2
X
Hypoxia-inducible mRNAs
Figure 3. Regulation of hypoxia-inducible mRNAs by the von HippelÐLindau tumour suppressor protein (pVHL). In the current model, the VEC (pVHLÐelonginÐCul-2) complex ubiquitinates a protein (ÔXÕ). ÔXÕ is a protein that, directly or indirectly, promotes the accumulation of hypoxia-inducible mRNAs.
on the similarity of elongin C and Cul-2 with Skp1 and Cdc53. Formation of these complexes is tightly linked to the ability of pVHL to negatively regulate hypoxia-inducible mRNAs, such as VEGF. Deregulated VEGF almost certainly contributes to the vascular nature of VHL disease-associated neoplasms. In addition to serving as a negative regulator of angiogenesis, pVHL is also required for the formation of fibronectin matrix and for cell-cycle exit. Thus, pVHL is a multifunctional tumour suppressor. Hopefully, continued insights into the functions of pVHL will provide new opportunities for the treatment of VHL disease, sporadic renal cell carcinoma and, perhaps, other tumours. In this regard, several clinical studies that are on-going or planned will test whether drugs that block VEGF signalling can cause regression of haemangioblastomas and renal cell carcinomas. Acknowledgements. We thank the members of the Kaelin laboratory for critical reading of this manuscript and many useful discussions. We apologize to our colleagues whose work may not have been cited as a result of our oversight or because of space limitations. M.O. is
262
1 Maher, E. and Kaelin, W.G. (1997) von Hippel–Lindau disease, Medicine 76, 381–391 2 Latif, F. et al. (1993) Identification of the von Hippel–Lindau disease tumor suppressor gene, Science 260, 1317–1320 3 Kessler, P. et al. (1995) Expression of the von Hippel–Lindau tumor-suppressor gene, VHL, in human fetal kidney and during mouse embryogenesis, Mol. Med. 1, 457–466 4 Richards, F., Schofield, P., Fleming, S. and Maher, E. (1996) Expression of the von Hippel–Lindau disease tumour suppressor gene during human embryogenesis, Hum. Mol. Genet. 5, 639–644 5 Gnarra, J.R. et al. (1994) Mutations of the VHL tumour suppressor gene in renal carcinoma, Nat. Genet. 7, 85–90 6 Lubensky, I.A. et al. (1996) Allelic deletions of the VHL gene detected in multiple microscopic clear cell renal lesions in von Hippel–Lindau disease patients, Am. J. Pathol. 149, 2089–2094 7 Kinzler, K. and Vogelstein, B. (1996) Lessons from hereditary colorectal cancer, Cell 87, 159–170 8 Gnarra, J. et al. (1997) Defective placental vasculogenesis causes embryonic lethality in VHL-deficient mice, Proc. Natl. Acad. Sci. U. S. A. 94, 9102–9107 9 Stolle, C. et al. (1998) Improved detection of germline mutations in the von Hippel–Lindau disease tumor suppressor gene, Hum. Mutat. 12, 417–423 10 Zbar, B. et al. (1996) Germline mutations in the von Hippel–Lindau (VHL) gene in families from North America, Europe, and Japan, Hum. Mutat. 8, 348–357 11 Schoenfeld, A., Davidowitz, E. and Burk, R. (1998) A second major native von Hippel–Lindau gene product, initiated from an internal translation start site, functions as a tumor suppressor, Proc. Natl. Acad. Sci. U. S. A. 95, 8817–8822 12 Iliopoulos, O., Ohh, M. and Kaelin, W. (1998) pVHL19 is a biologically active product of the von Hippel–Lindau gene arising from internal translation initiation, Proc. Natl. Acad. Sci. U. S. A. 95, 11661–11666 13 Kozak, M. (1987) At least six nucleotides preceding the AUG initiator codon enhance translation in mammalian cells, J. Mol. Biol. 196, 947–950 14 Kozak, M. (1989) The scanning model for translation: an update, J. Cell Biol. 108, 229–241 15 Iliopoulos, O., Kibel, A., Gray, S. and Kaelin, W.G. (1995) Tumor suppression by the human von Hippel–Lindau gene product, Nat. Med. 1, 822–826 16 Gnarra, J.R. et al. (1996) Post-transcriptional regulation of vascular endothelial growth factor mRNA by the VHL tumor suppressor gene product, Proc. Natl. Acad. Sci. U. S. A. 93, 10589–10594 17 Ohh, M. et al. (1998) The von Hippel–Lindau tumor suppressor protein is required for proper assembly of an extracellular fibronectin matrix, Mol. Cell 1, 959–968 18 Pause, A., et al. (1997) The von Hippel–Lindau tumor-suppressor gene product forms a stable complex with human CUL-2, a member of the Cdc53 family of proteins, Proc. Natl. Acad. Sci. U. S. A. 94, 2156–2161 19 Lee, S. et al. (1996) Nuclear/cytoplasmic localization of the von Hippel–Lindau tumor suppressor gene product is determined by cell density, Proc. Natl. Acad. Sci. U. S. A. 93, 1770–1775 20 Aso, T., Lane, W.S., Conaway, J.W. and Conaway, R.C. (1995) Elongin (SIII): a multisubunit regulator of elongation by RNA polymerase II, Science 269, 1439–1443 21 Kibel, A. et al. (1995) Binding of the von Hippel–Lindau tumor suppressor protein to elongin B and C, Science 269, 1444–1446 22 Aso, T. et al. (1996) The inducible elongin A elongation activation domain: structure, function and interaction with elongin BC complex, EMBO J. 15, 101–110 23 Duan, D.R. et al. (1995) Inhibition of transcriptional elongation by the VHL tumor suppressor protein, Science 269, 1402–1406
Reviews
MOLECULAR MEDICINE TODAY, JUNE 1999 (VOL. 5)
24 Kamura, T. et al. (1998) The elongin BC complex interacts with the conserved SOCS-box motif present in members of the SOCS, ras, WD-40 repeat, and ankyrin repeat families, Genes Dev. 12, 3872–3881 25 Lonergan, K.M. et al. (1998) Regulation of hypoxia-inducible mRNAs by the von Hippel–Lindau protein requires binding to complexes containing elongins B/C and Cul2, Mol. Cell Biol. 18, 732–741 26 Kipreos, E.T. et al. (1996) Cul-1 is required for cell cycle exit in C. elegans and identifies a novel gene family, Cell 85, 829–839 27 Peters, J-M. (1998) SCF and APC: the Yin and Yang of cell cycle regulated proteolysis, Curr. Opin. Cell Biol. 10, 759–768 28 Skowyra, D. et al. (1997) F-box proteins are receptors that recruit phosphorylated substrates to the SCF ubiquitin–ligase complex, Cell 91, 209–219 29 Lisztwan, J. et al. (1998) Association of human CUL-1 and ubiquitin-conjugating enzyme CDC34 with the F-box protein p45 (SKP2): evidence for evolutionary conservation in the subunit composition of the CDC34–SCF pathway, EMBO J. 17, 368–383 30 Bai, C. et al. (1996) Skp1 connects cell cycle regulators to the ubiquitin proteolysis machinery through a novel motif, the F-box, Cell 86, 263–274 31 King, R., Deshaies, R., Peters, J-M. and Kirschner, M.W. (1996) How proteolysis drives the cell cycle, Science 274, 1652–1659 32 Gorospe, M. et al. (1999) Protective function of pVHL against impaired protein processing in renal carcinoma cells, Mol. Cell Biol. 19, 1289–1300 33 Pause, A., Lee, S., Lonergan, K.M. and Klausner, R.D. (1998) The von Hippel–Lindau tumor suppressor gene is required for cell cycle exit upon serum withdrawal, Proc. Natl. Acad. Sci. U. S. A. 95, 993–998 34 Hynes, R. (1992) Integrins: versatility, modulation, and signalling in cell adhesion, Cell 69, 11–25 35 Pasqualini, R. et al. (1996) A polymeric form of fibronectin has antimetastatic effects against multiple tumor types, Nat. Med. 2, 1197–1203
36 Lieubeau-Teillet, B. et al. (1998) von Hippel–Lindau gene-mediated growth suppression and induction of differentiation in renal cell carcinoma cells grown as multicellular tumor spheroids, Cancer Res. 58, 4957–4962 37 Kopito, R.R. (1997) ER quality control: the cytoplasmic connection, Cell 88, 427–430 38 Tsuchiya, H., Iseda, T. and Hino, O. (1996) Identification of a novel protein (VBP-1) binding to the von Hippel–Lindau (VHL) tumor suppressor gene product, Cancer Res. 56, 2881–2885 39 Mukhopadhyay, D. et al. (1997) The von Hippel–Lindau tumor suppressor gene product interacts with Sp1 to repress vascular endothelial growth factor promoter activity, Mol. Cell Biol. 17, 5629–5639 40 Pal, S., Claffey, K., Dvorak, H. and Mukhopadhyay, D. (1997) The von Hippel–Lindau gene product inhibits vascular permeability factor/vascular endothelial growth factor expression in renal cell carcinoma by blocking protein kinase C pathways, J. Biol. Chem. 272, 27509–27512 41 Pal, S., Claffey, K., Cohen, H. and Mukhopadhyay, D. (1998) Activation of Sp1mediated vascular permeability factor/vascular endothelial growth factor transcription requires specific interaction with protein kinase Cz, J. Biol. Chem. 273, 26277–26280 42 Ivanov, S. et al. (1998) Down-regulation of transmembrane carbonic anhydrase in renal cell carcinoma cell lines by wild-type von Hippel–Lindau transgenes, Proc. Natl. Acad. Sci. U. S. A. 95, 12596–12601 43 Iliiopoulos, O. et al. (1996) Negative regulation of hypoxia-inducible genes by the von Hippel–Lindau protein, Proc. Natl. Acad. Sci. U. S. A. 93, 10595–10599 44 Levy, A.P., Levy, N.S. and Goldberg, M.A. (1996) Hypoxia-inducible protein binding to vascular endothelial growth factor mRNA and its modulation by the von Hippel–Lindau protein, J. Biol. Chem. 271, 25492–25497 45 Martinez-Zaguilan, R. et al. (1996) Acidic pH enhances the invasive behavior of human melanoma cells, Clin. Exp. Metastasis 14, 176–186
Late-breaking news Several reports appeared after this review was accepted for publication that are of direct relevance to the models shown in Figs 2 and 3. The crystal structure of pVHL bound to elongins B and C has similarities to SCF complexes1. In this structure, pVHL forms two subdomains, called a and b, both of which are frequently mutated in VHL disease. The a domain comprises three helices that bind to a helix in elongin C to form a four-helix cluster. One of the pVHL helices in this arrangement corresponds to pVHL residues 157–172 that, as noted earlier in this review, is sufficient to bind to elongins B and C in vitro. The b domain represents a putative substrate-docking site. Elongin C, as predicted, resembles Skp1 and the elongin C binding domain of pVHL loosely resembles an F-box. Together, the structure of pVHL bound to elongin B and C is consistent with the notion that pVHL, when bound to elongin B, elongin C and Cul2, targets proteins for covalent linkage to ubiquitin or ubiquitin-like molecules. This notion is further supported by the recent demonstration that pVHL co-purifies with a protein called Rbx1 (Ref. 2). This protein, which is also called ROC1, interacts with Cdc53/Cullinlike molecules and facilitates the ubiquitination of substrates by
SCF complexes by recruiting a ubiquitin-conjugating enzyme (E2; Refs 2–5). These recent discoveries will, no doubt, intensify and accelerate efforts aimed at determining whether pVHL plays a direct role in ubiquitination. 1 Stebbins, C.E., Kaelin, W.G. and Pavletich, N.P. (1999) Structure of the VHLElonginC-elonginB complex: implications for VHL tumor suppressor function, Science 284, 455–461 2 Kamura, T. et al. (1999) Rbx1, a component of the VHL tumor suppressor complex and SCF ubiquitin ligase, Science 284, 657–661 3 Ohta, T. et al. (1999) ROC1, a homolog of APC11, represents a family of cullin partners with an associated ubiquitin ligase activity, Mol. Cell 3, 535–541 4 Tan, P. et al. (1999) Recruitment of a ROC1-CUL1 ubiquitin ligase by Skp1 and HOS to catalyse the ubiquitination of lkBa, Mol. Cell 3, 527–533 5 Skowyra, D. et al. (1999) Reconstitution of G1 cyclin ubiquitination with complexes containing SCFGrr1 and rbx1, Science 284, 662–665 William Kaelin
263
MOLECULAR MEDICINE TODAY, JUNE 1999 (VOL. 5)
The von Hippel–Lindau tumour suppressor protein: new perspectives Michael Ohh and William G. Kaelin Jr
von HippelÐLindau (VHL) disease is a hereditary cancer syndrome caused by germline mutations of the VHL tumour suppressor gene. The VHL gene product, pVHL, forms multiprotein complexes that contain elongin B, elongin C and Cul-2, and negatively regulates hypoxia-inducible mRNAs. pVHL is suspected to play a role in ubiquitination given the similarity of elongin C and Cul-2 with Skp1 and Cdc53, respectively. pVHL can also interact with fibronectin and is required for the assembly of a fibronectin matrix. Finally, pVHL, at least indirectly, plays a role in the ability of cells to exit the cell cycle. Thus, pVHL is a tumour suppressor protein that regulates angiogenesis, extracellular matrix formation and the cell cycle. INHERITANCE of a mutant allele of the von Hippel–Lindau (VHL) tumour suppressor gene causes VHL disease, which is characterized by a range of cancers, including retinal angiomas, cerebellar and spinal haemangioblastomas, renal cell carcinomas, phaeochromocytomas, pancreatic adenomas and islet cell tumours, epididymal cystadenomas and endolymphatic sac tumours of the inner ear1. Renal carcinomas and central nervous system (CNS) haemangioblastomas remain the major causes of morbidity and mortality for VHL patients1. VHL disease affects 1 in 36 000 individuals and clinically displays an autosomal dominant pattern of inheritance with high penetrance1. However, at the molecular level, VHL disease is autosomal recessive insofar as tumours arise only after the remaining wild-type VHL allele is somatically mutated or silenced1.
Picture kindly provided by Dr Alain Gaudric, Hopital Lariboisiere, Paris, France.
Cloning of the VHL gene
Michael Ohh PhD Research Associate William G. Kaelin Jr* MD Associate Professor Howard Hughes Medical Institute, Division of Adult Oncology, Dana-Farber Cancer Institute and Dept of Medicine, Harvard Medical School, Boston, MA 02115, USA. *e-mail:[email protected]
The VHL gene was cloned in 1993 (Ref. 2), and intragenic germline mutations in members of VHL kindreds have confirmed the authenticity of the gene1. The VHL gene is widely expressed in both fetal and adult tissues, demonstrating that expression of the VHL transcript is not restricted to organs affected by VHL disease1. Nonetheless, the pattern of expression in the fetal kidney is consistent with a role in normal renal tubular development and differentiation3,4. The VHL coding sequence is contained within three exons, and two alternatively spliced mRNAs that reflect the presence (isoform I) or absence (isoform II) of exon 2 have been detected5. Some VHL patients harbour VHL mutations that lead to the exclusive production of isoform II. This suggests that the VHL mRNA isoform II is not sufficient to inhibit tumorigenesis5. In keeping with Knudson’s two-hit model, functional inactivation or loss of both VHL alleles has been demonstrated in the majority of sporadic clear cell renal carcinomas and cerebellar haemangioblastomas that have so far been examined1. Moreover, reports that de-
1357-4310/99/$ - see front matter © 1999 Elsevier Science. All rights reserved. PII: S1357-4310(99)01481-1
257
Reviews
MOLECULAR MEDICINE TODAY, JUNE 1999 (VOL. 5)
scribe loss of heterozygosity at the VHL locus in early premalignant lesions of the kidney, such as atypical cysts6, suggest that inactivation of the VHL tumour suppressor protein (pVHL) is an early step in the pathogenesis of clear cell renal carcinoma. In this regard, VHL might play a ‘gatekeeper’ role with respect to renal carcinogenesis, similar to that of APC (adenomatous polyposis coli) in colorectal carcinogenesis, as proposed by Kinzler and Vogelstein7. Moreover, VHL is required for embryogenesis, as demonstrated by murine gene knockout experiments8. VHL mutations can be detected in ~100% of patients who carry a clinical diagnosis of VHL disease9. VHL mutations are extremely heterogeneous and are widely distributed throughout the open reading frame10. Specific genotype–phenotype correlations are emerging in VHL disease. Thus, although retinal angiomas and CNS haemangioblastomas are hallmarks of the disease, some families display a high risk of concurrent phaeochromocytoma (type 2 VHL) but others do not (type 1 VHL)10. Type 2 VHL families can be further subdivided into type 2A (low risk of renal cell carcinoma), type 2B (high risk of renal cell carcinoma) and type 2C (exclusively phaeochromocytoma without other characteristics of VHL disease)1. In general, missense mutations are associated with the development of type 2 disease, whereas mutations that lead to truncated versions of pVHL and gross deletions are primarily associated with type 1 disease1. Hence, although speculative, the development of phaeochromocytoma might be linked to a partial, rather than a complete, loss of pVHL function(s) or a gain of an unknown function. With the increasing emergence of these genotype–phenotype correlations, it seems plausible that pVHL has multiple functions, some of which might be tissue-specific.
that pVHL can shuttle back and forth between the nucleus and cytoplasm in response to certain signals. pVHL19 appears to be equally distributed between the nucleus and cytoplasm and does not associate with membranes12. In this regard, it is noteworthy that pVHL19 lacks an N-terminal acid pentameric repeat that resembles a sequence found in Trypanosoma brucei membrane protein2. Whether pVHL19 shuttles in response to particular cellular stimuli is not known.
The evolving model of tumour suppression by pVHL The primary sequence of pVHL does not resemble that of any protein of known function. With the hope of discerning the function(s) of pVHL, several groups have sought to identify cellular proteins that are capable of binding to pVHL. The experiments carried out to date have identified several pVHL-associated proteins. These are discussed below (Fig. 1).
Elongins B and C Elongins B and C bind stably to each other to form a binary complex that was first identified by virtue of its ability to stimulate the transcriptional elongation activity of elongin A (Ref. 20). This trimeric elongin–SIII complex enhances the rate of elongation by RNA polymerase II by suppressing the basal transcription machinery from pausing or stalling along the DNA template20. The minimal region of pVHL required for binding elongin B–C complex in vitro, corresponding to C-terminal residues 157–172, is the only region of pVHL that is highly similar to elongin A (Refs 20,21). As expected, the homologous region of elongin A likewise interacts with elongin B–C complex22, thereby establishing this stretch of amino acid residues as an elongin B–C-binding motif. Furthermore, binding of the elongin B–C complex to pVHL and elongin A is, at least in vitro,
A single VHL gene and its two products The VHL mRNA encodes two protein products (pVHL) with apparent molecular weights of ~30 and 19 kDa (Refs 11,12). The former contains 213 amino acid residues and the latter contains 160 amino acid residues as a result of internal translation from the second methionine at residue 54 within the VHL open reading frame. The first and second methionine codons are flanked by residues that conform roughly to Kozak’s rules for translation initiation sites13,14. Reintroduction of either 30-kDa (pVHL30) or 19-kDa (pVHL19) pVHL into VHL–/– renal carcinoma cells suppresses their ability to form tumours in nude mice11,15,16. This observation might account for the finding that VHL mutations almost invariably map C-terminal to residue 54 (Ref. 1). Specifically, mutations affecting the first 54 amino acids would, in principle, still allow the production of wild-type pVHL19 (Refs 11,12). Biochemical fractionation and immunohistochemical studies suggest that pVHL30 is predominantly cytoplasmic1. However, a significant fraction of pVHL30 can also be found in the nucleus and in association with the endoplasmic reticulum (ER)12,17,18. When artificially overproduced in mammalian cells, the subcellular localization of pVHL appears to vary in a cell-density-dependent manner19. This observation raises the possibility 258
Cul-2 B
C
PKCd pVHL
FN
PKCz
Sp1
VBP-1
Figure 1. von HippelÐLindau tumour suppressor protein (pVHL)-associated proteins. Several cellular proteins have been shown to bind, directly or indirectly, to pVHL, including elongin B (B), elongin C (C), Cul-2, fibronectin (FN), VHL binding protein 1 (VBP-1), the transcription factor Sp1 and protein kinase C (PKC) isoforms d and z. The biochemical interaction of pVHL with FN, PKC and Sp1 requires independent confirmation.
Reviews
MOLECULAR MEDICINE TODAY, JUNE 1999 (VOL. 5)
mutually exclusive23. This almost certainly accounts for the ability of pVHL to inhibit elongin–SIII activity in vitro23. The elongin-binding domain of pVHL is a frequent site of mutations in VHL kindreds, suggesting that elongin binding contributes to tumour suppression by pVHL (Ref. 1). The above considerations led to a model in which tumour suppression by pVHL would result, at least in part, from inhibition of elongin–SIII in vivo21,23. Although not formally disproven, several lines of evidence suggest that this model is incorrect. Firstly, there is no evidence that elongin–SIII regulates the transcription of specific genes in vivo. Secondly, and perhaps more importantly, elongins B and C are in vast molar excess with respect to both elongin A and pVHL (Ref. 24). Indeed, recent work suggests that pVHL and elongin A are but two examples of a growing family of proteins that contain similar, colinear, elongin B–C-binding domains24.
Cullin 2 Two groups have independently shown that pVHL binds to cullin 2 (Cul-2), a member of the cullin family18,25. Cullins were first identified in Caenorhabditis elegans during a screen for mutants that displayed excess postembryonic cell divisions26. Loss of Cul-1 function resulted in hyperplasia of diverse tissue types owing to a failure to generate or to respond to the normal developmental signals to properly exit the cell cycle26. This suggests that Cul-1 functions as a negative regulator of the cell cycle and is required for cells to enter either G0 or the apoptotic pathway. There are now known to be at least three cullins in budding yeast, five in nematodes and six in humans27. Notably, Cul-1 and Cul-2 bear striking similarity to yeast Cdc53 (Ref. 27), which plays a role in the ubiquitination, and hence degradation, of certain proteins27. Cdc53 function is intricately tied to its ability to bind Cdc34, which functions as an E2 ubiquitin-conjugating enzyme; to Skp1 (S-phase kinase-associated protein 1, suppressor of Cdk inhibitor proteolysis and of kinetochore protein), which probably acts as an assembly factor; and an F-box protein (so called because of apparent homology to cyclin F), such as Grr1 or Cdc4, which probably confers substrate/target specificity27. A complex that consists of Skp1, Cdc53 and an F-box protein has been dubbed the SCF (Ref. 28). For example, SCFCdc4 specifically targets phosphorylated Sic1 for ubiquitination, whereas SCFGrr1 promotes phosphorylated cyclin 1 (Cln1) and cyclin 2 (Cln2) for ubiquitin-mediated proteolysis, both of which are required for G1–S transition27. Recently, Cul-1 has been demonstrated, in humans, to bind Skp1, Cdc34 and an F-box-containing protein, Skp2. This SCFSkp2 complex interacts with cyclin A/Cdk2 and is thought to play a role in S-phase entry and traversal29. Intriguingly, the elongin–SIII complex is remarkably similar to the SCF complex. Elongins A, B and C contain regions of similarity to an F-box protein, ubiquitin and Skp1, respectively25,30. Nonetheless, there is currently no evidence that elongin A interacts with a cullin. In contrast, pVHL binds to Cul-2 via elongins B and C (Ref. 25). Thus, the elongin B–C complex bridges pVHL to Cul-2 in a manner analogous to Skp1 of the SCF complex joining Cdc53 to an F-box protein. Although pVHL does not have a recognizable F-box, it might still play a role similar to that of F-box-containing proteins – that is, it might function to confer substrate specificity. A potential precedent for F-box-independent substrate recognition by ubiquitin-conjugating enzymes is provided by examination of APC (the anaphase-promoting complex or cyclosome)31. APC, like SCF, contains a cullin homologue called Apc2 (Ref. 27). In budding yeast, Cdc20 and Cdh1/Hct1,
Glossary Allele Ð One of two or more forms of a gene that exist at a single locus. Angiogenesis Ð The formation of new blood vessels. Cell cycle Ð The progression through all the stages that are necessary for DNA replication and cell division. Cullin Ð Member of a family of proteins with sequence homology to yeast Cdc53. Elongin Ð A trimeric complex, containing elongin A, elongin B and elongin C, that enhances the rate of transcriptional elongation by RNA polymerase II in vitro by suppressing the basal transcription machinery from pausing or stalling along the DNA template. KnudsonÕs two-hit model Ð A model that suggests that the inactivation of both the maternal and paternal alleles of a tumour suppressor gene are necessary for tumour development. Loss of heterozygosity Ð The condition that results from loss of maternal or paternal DNA at a particular genetic locus. Transcriptional elongation Ð The process by which ribonucleotides are processively added to a nascent RNA transcript. Tumour suppressor gene Ð A gene the inactivation of which contributes to carcinogenesis. Ubiquitin-mediated proteolysis Ð A mechanism for the removal of specific proteins in which the protein to be eliminated is covalently linked to ubiquitin and subsequently degraded by the proteosome.
two components of the APC, lack F-boxes and yet target proteins such as Pds1 and Ase1, respectively, for ubiquitination27. In light of this knowledge, pVHL might play a role in the ubiquitination of certain proteins (Fig. 2). For example, the pVHL–elongin–Cul-2 (VEC) complex might degrade certain proteins involved in the stabilization of hypoxia-inducible mRNAs (see below) or in the degradation of malfolded proteins present in the ER. In this regard, Gorospe et al. reported that cells that lack pVHL are more readily killed by certain forms of ER stress than their wild-type counterparts32. Furthermore, they documented the abnormal accumulation of ubiquitinated proteins in pVHL-defective cells. These latter observations are consistent with a role of pVHL, whether direct or indirect, in ubiquitination. Based on the association of pVHL with Cul-2, Pause et al. examined whether pVHL plays a role in cell-cycle control33. They found that pVHL was required for cells to exit the cell cycle in the absence of serum and correlated this requirement with changes in p27Kip1 levels in VHL+/+ versus VHL–/– cells under these conditions33. However, there is no evidence that p27 is a direct target of the VEC complex. Furthermore, the effect of pVHL in these assays was influenced by various cell culture parameters, including cell density and the timing of cell plating and media changes (A. Pause and S. Lee, unpublished). This observation is intriguing given the fact that several of the pVHL target genes identified to date encode secreted proteins, including mitogens, or proteins that play a role in extracellular pH (see below; Box 1). 259
Reviews
MOLECULAR MEDICINE TODAY, JUNE 1999 (VOL. 5)
a SCF
Cdc4
Sic1 Skp1
Cdc4
Cdc34 Ubq Cdc53
b VEC X C
pVHL
?E2
B
Ubq Cul-2
c APC
Pds1 Ase1 Ubq Clb2
Cdh1Ð Hct1
Cdc20 Apc2
Figure 2. Comparison between SCF (a complex that comprises Skp1, Cdc53 and an F-box protein), VEC [von HippelÐLindau tumour suppressor protein (pVHL)ÐelonginÐCul-2 complex] and APC (the anaphase-promoting complex) complexes. (a) In budding yeast, Cdc53 binds to Skp1 and an F-box protein, such as Cdc4 (SCFCdc4 complex). This complex, in association with the E2 ubiquitin-conjugating protein Cdc34, selectively targets phosphorylated Sic1 for ubiquitination. Subsequent destruction of the polyubiquitinated Sic1 by the 26S proteosome allows G1ÐS transition. (b) pVHL binds to elongin C (C) and Cul-2, which are similar to Skp1 and Cdc53, respectively. A current model suggests that the VEC complex interacts with a cognate E2 to ubiquitinate an undefined protein (ÔXÕ) for degradation. (c) APC, like SCF and VEC, contains a Cdc53 homologue (Apc2). Cdc20, which lacks an F-box, when complexed to APC triggers the destruction of Pds1. This promotes the segregation of chromosomes at the metaphaseÐanaphase transition. APC then forms a second complex with Cdh1ÐHct1, which also lacks an F-box, to direct the ubiquitination of the spindle protein Ase1 and the mitotic cyclin Clb2.
Fibronectin Recently, pVHL has been demonstrated to bind specifically, at least indirectly, to fibronectin in association with the ER (Ref. 17). This interaction is abrogated by every disease-causing mutation that has been examined to date, including those shown or predicted not to influence the binding of pVHL to elongin B–C–Cul-2 complex, emphasizing the potential significance of fibronectin in the ability of pVHL to suppress neoplastic growth17. Moreover, VHL null tumours, as well as cells derived from VHL knockout mouse embryo fibroblasts, fail to assemble a proper fibronectin matrix17. Loss of extra260
cellular fibronectin matrix is a hallmark of many transformed cells34. Conversely, fibronectin or overproduction of a fibronectin receptor, such as a5b1 integrin, can, in certain experimental settings, promote differentiation and revert some aspects of malignant behaviour35. Thus promotion of fibronectin matrix assembly might contribute to tumour suppression by pVHL. This notion is supported by a recent study in which VHL+/+ and VHL–/– renal carcinoma cells were grown as three-dimensional spheroids in vitro36. Cells that lacked pVHL grew as tightly packed, amorphous spheroids. In contrast, cells that produced pVHL formed loose aggregates which, on microscopic and ultrastructural examination, exhibited evidence of epithelial differentiation, including the formation of trabecular and tubular structures36. These differences correlated with differences in fibronectin matrix deposition, raising the possibility that fibronectin played an active role in the differentiation of the VHL+/+ cells under these conditions36. How might loss of pVHL lead to failure to assemble a fibronectin matrix? Preliminary data suggest that pVHL-defective cells are capable of secreting fibronectin but that the fibronectin, once secreted, fails to polymerize properly17. One tentative model, based on analogy with other secreted proteins37, is that pVHL plays a role in the elimination of malfolded or malprocessed fibronectin. Specifically, malfolded or malprocessed fibronectin might undergo retrograde transport to the cytoplasmic surface of the ER membrane whereupon it would be acted upon by the VEC complex. If this were a coupled process, loss of pVHL might lead to the secretion of malfolded/malprocessed fibronectin that could then interfere with the polymerization of fibronectin into a recognizable fibrillar array.
VBP-1
Hino and co-workers identified VHL-binding protein 1 (VBP-1) in a yeast two-hybrid screen, using pVHL as bait38. Furthermore, truncation of the Cterminal 26 amino acids of pVHL abolished the binding of VBP-1 (Ref. 38). In the absence of pVHL, VBP-1 is exclusively localized in the cytosol38. However, when co-expressed with an epitope-tagged pVHL, VBP-1 can be detected in the nucleus, indicating that VBP-1, when complexed to pVHL, can translocate to the nucleus38. It is not known whether pVHL binds to VBP-1 under physiological conditions, and no functional significance has been ascribed to the interaction of pVHL with VBP-1.
Sp1 Sukhatme and co-workers reported that pVHL can inhibit transcription from the vascular endothelial growth factor/vascular permeability factor (VEGF/VPF) promoter39. The minimal region of the VEGF promoter necessary for this effect was found to contain binding sites for Sp1 (Ref. 39). Furthermore, pVHL was demonstrated to bind
Reviews
MOLECULAR MEDICINE TODAY, JUNE 1999 (VOL. 5)
Box 1. The downstream targets of the von HippelÐLindau tumour suppressor protein (pVHL)
• • • • • •
Vascular endothelial growth factor (VEGF) Glucose transporter 1 (GLUT-1) Platelet-derived growth factor B (PDGF-B) Transforming growth factor a (TGF-a) Carbonic anhydrase 9 (CA9) Carbonic anhydrase 12 (CA12)
directly to Sp1, at least in vitro, suggesting a model in which pVHL inhibits Sp1-mediated transcription by sequestering Sp1 (Ref. 39). Thus, in VHL disease, a loss of functional pVHL might lead to a loss of Sp1 inhibition, which might in turn lead to an inappropriate expression of VEGF mRNA (see also below).
PKC Mukhopadhyay and co-workers showed that the binding of pVHL to two protein kinase C (PKC) isoforms, PKCd and PKCz, inhibited the transcriptional activation of VEGF (Refs 40,41). Furthermore, PKCz was demonstrated to directly bind and phosphorylate the zinc finger domain of Sp1 (Ref. 41). Overexpression of PKCz led to increased transcription from the VEGF promoter41. More importantly, pVHL blocked the interaction between Sp1 and PKCz, thereby suppressing the phosphorylation of Sp1 and hence downregulating the Sp1dependent transcriptional activation of VEGF (Ref. 41) (see below).
Downstream targets of pVHL pVHL regulates, at least indirectly, the abundance of a variety of mRNAs (see Box 1). These include the mRNAs that encode VEGF/VPF, platelet-derived growth factor B (PDGF-B), glucose transporter 1 (GLUT-1), transforming growth factor a (TGF-a) and carbonic anhydrases (CA) 9 and 12 (Refs 1,42). The search for pVHL target genes was initially aided by the knowledge that VHLassociated neoplasms are typically hypervascular and, on occasion, produce a syndrome that is characterized by the excessive production of red blood cells (secondary polycythaemia or paraneoplastic erythrocytosis)1. The former has been linked to the overproduction of angiogenic peptides, such as VEGF/VPF, whereas the latter has been attributed to the excessive production of erythropoietin (EPO) by the tumour cells1. The production of VEGF and EPO are normally triggered by hypoxia, presumably to augment the delivery of oxyhaemoglobin to cells1. This raised the possibility that pVHL played a role in certain stress responses, such as changes in ambient oxygen tension. Indeed, VHL–/– tumour cells produce high levels of hypoxia-inducible mRNAs, such as the VEGF and GLUT-1 mRNAs, under both hypoxic and normoxic conditions1. Reintroduction of wild-type, but not mutant, pVHL into these cells specifically inhibited the production of hypoxia-inducible mRNAs under normoxic conditions, thus restoring their hypoxia-inducible profile1. Thus, the hypervascular nature of VHL-associated tumours is probably linked to the overproduction of hypoxia-inducible mRNAs, including those that encode angiogenic peptides. Moreover, the inhibition of VEGF has been demonstrated to inhibit tumorigenesis in vivo and thus might contribute to the ability of pVHL to suppress tumour formation1.
The outstanding questions Does the von Hippel–Lindau tumour suppressor protein •(pVHL)–elongin–Cul-2 complex play a role in ubiquitination and, if so, what are its targets?
negative regulation of hypoxia-inducible genes by pVHL •dueIsprimarily to transcriptional effects, post-transcriptional effects, or both? Might the relative contributions of these two mechanisms differ in different cell types? Why is pVHL required for assembly of a fibronectin matrix? What physiological relevance can be ascribed to the ability of pVHL to shuttle between the cytoplasm and nucleus? What is the biochemical basis of the genotype–phenotype correlations observed in pVHL disease?
• • •
There are conflicting data as to whether the effect of pVHL on VEGF is primarily transcriptional or post-transcriptional. As noted above, two reports suggested that pVHL, directly or indirectly, inhibits Sp1 and hence transcription of the VEGF promoter39,41. In contrast, two groups have reported that the effect of pVHL on VEGF mRNA abundance was largely due to changes in VEGF mRNA stability16,43. In particular, pVHL was required for the destabilization of the VEGF mRNA under well-oxygenated conditions16,43. Intriguingly, the 39-untranscribed regions of hypoxia-inducible mRNAs, such as VEGF mRNA and GLUT-1 mRNA, contain an AUUUA-rich element, which can bind to as yet unidentified proteins in RNA gel-shift assays44. The binding of these proteins correlates with mRNA stabilization44. In normal cells, these RNA-binding factors are only detected under hypoxic conditions. In contrast, these factors are detected under both hypoxic and normoxic conditions in cells that lack pVHL (Ref. 44). Thus, a tentative model for pVHL action would suggest that the VEC complex regulates the stability and/or function of these, as yet, unidentified RNA-binding proteins (Fig. 3). Perhaps in keeping with this view, Lonergan et al. showed that binding to elongin B, elongin C and Cul-2 was necessary, but not sufficient, for pVHL to regulate hypoxia-inducible mRNAs (Ref. 25). Lerman and co-workers, using differential display, found that carbonic anhydrase 9 (CA9) and carbonic anhydrase 12 (CA12) were upregulated in VHL-defective renal carcinoma cells42. This finding is intriguing given earlier studies that showed that these proteins are frequently overproduced in renal cell carcinomas and might serve as markers for this disease42. Furthermore, CA9 and CA12 might affect peritumoral pH and thus might indirectly affect tumour cell growth, based on earlier observations that the extracellular pH of tumours is generally more acidic than normal cells and that, at least in vitro, acidic pH favours the growth, spreading and invasive properties of tumour cells45.
Concluding remarks The von Hippel–Lindau tumour suppressor protein (pVHL) forms specific complexes with several cellular proteins. Some of these interactions, such as binding to fibronectin, Sp1 and certain PKC isoforms, require independent confirmation. In contrast, there is general agreement that pVHL binds to elongin B, elongin C and Cul-2. These complexes are thought to play a role in ubiquitination, based largely 261
Reviews
MOLECULAR MEDICINE TODAY, JUNE 1999 (VOL. 5)
supported by the Medical Research Council of Canada and W.G.K. is a Howard Hughes Medical Institute assistant investigator.
References
Hypoxia
B pVHL
C
Cul-2
X
Hypoxia-inducible mRNAs
Figure 3. Regulation of hypoxia-inducible mRNAs by the von HippelÐLindau tumour suppressor protein (pVHL). In the current model, the VEC (pVHLÐelonginÐCul-2) complex ubiquitinates a protein (ÔXÕ). ÔXÕ is a protein that, directly or indirectly, promotes the accumulation of hypoxia-inducible mRNAs.
on the similarity of elongin C and Cul-2 with Skp1 and Cdc53. Formation of these complexes is tightly linked to the ability of pVHL to negatively regulate hypoxia-inducible mRNAs, such as VEGF. Deregulated VEGF almost certainly contributes to the vascular nature of VHL disease-associated neoplasms. In addition to serving as a negative regulator of angiogenesis, pVHL is also required for the formation of fibronectin matrix and for cell-cycle exit. Thus, pVHL is a multifunctional tumour suppressor. Hopefully, continued insights into the functions of pVHL will provide new opportunities for the treatment of VHL disease, sporadic renal cell carcinoma and, perhaps, other tumours. In this regard, several clinical studies that are on-going or planned will test whether drugs that block VEGF signalling can cause regression of haemangioblastomas and renal cell carcinomas. Acknowledgements. We thank the members of the Kaelin laboratory for critical reading of this manuscript and many useful discussions. We apologize to our colleagues whose work may not have been cited as a result of our oversight or because of space limitations. M.O. is
262
1 Maher, E. and Kaelin, W.G. (1997) von Hippel–Lindau disease, Medicine 76, 381–391 2 Latif, F. et al. (1993) Identification of the von Hippel–Lindau disease tumor suppressor gene, Science 260, 1317–1320 3 Kessler, P. et al. (1995) Expression of the von Hippel–Lindau tumor-suppressor gene, VHL, in human fetal kidney and during mouse embryogenesis, Mol. Med. 1, 457–466 4 Richards, F., Schofield, P., Fleming, S. and Maher, E. (1996) Expression of the von Hippel–Lindau disease tumour suppressor gene during human embryogenesis, Hum. Mol. Genet. 5, 639–644 5 Gnarra, J.R. et al. (1994) Mutations of the VHL tumour suppressor gene in renal carcinoma, Nat. Genet. 7, 85–90 6 Lubensky, I.A. et al. (1996) Allelic deletions of the VHL gene detected in multiple microscopic clear cell renal lesions in von Hippel–Lindau disease patients, Am. J. Pathol. 149, 2089–2094 7 Kinzler, K. and Vogelstein, B. (1996) Lessons from hereditary colorectal cancer, Cell 87, 159–170 8 Gnarra, J. et al. (1997) Defective placental vasculogenesis causes embryonic lethality in VHL-deficient mice, Proc. Natl. Acad. Sci. U. S. A. 94, 9102–9107 9 Stolle, C. et al. (1998) Improved detection of germline mutations in the von Hippel–Lindau disease tumor suppressor gene, Hum. Mutat. 12, 417–423 10 Zbar, B. et al. (1996) Germline mutations in the von Hippel–Lindau (VHL) gene in families from North America, Europe, and Japan, Hum. Mutat. 8, 348–357 11 Schoenfeld, A., Davidowitz, E. and Burk, R. (1998) A second major native von Hippel–Lindau gene product, initiated from an internal translation start site, functions as a tumor suppressor, Proc. Natl. Acad. Sci. U. S. A. 95, 8817–8822 12 Iliopoulos, O., Ohh, M. and Kaelin, W. (1998) pVHL19 is a biologically active product of the von Hippel–Lindau gene arising from internal translation initiation, Proc. Natl. Acad. Sci. U. S. A. 95, 11661–11666 13 Kozak, M. (1987) At least six nucleotides preceding the AUG initiator codon enhance translation in mammalian cells, J. Mol. Biol. 196, 947–950 14 Kozak, M. (1989) The scanning model for translation: an update, J. Cell Biol. 108, 229–241 15 Iliopoulos, O., Kibel, A., Gray, S. and Kaelin, W.G. (1995) Tumor suppression by the human von Hippel–Lindau gene product, Nat. Med. 1, 822–826 16 Gnarra, J.R. et al. (1996) Post-transcriptional regulation of vascular endothelial growth factor mRNA by the VHL tumor suppressor gene product, Proc. Natl. Acad. Sci. U. S. A. 93, 10589–10594 17 Ohh, M. et al. (1998) The von Hippel–Lindau tumor suppressor protein is required for proper assembly of an extracellular fibronectin matrix, Mol. Cell 1, 959–968 18 Pause, A., et al. (1997) The von Hippel–Lindau tumor-suppressor gene product forms a stable complex with human CUL-2, a member of the Cdc53 family of proteins, Proc. Natl. Acad. Sci. U. S. A. 94, 2156–2161 19 Lee, S. et al. (1996) Nuclear/cytoplasmic localization of the von Hippel–Lindau tumor suppressor gene product is determined by cell density, Proc. Natl. Acad. Sci. U. S. A. 93, 1770–1775 20 Aso, T., Lane, W.S., Conaway, J.W. and Conaway, R.C. (1995) Elongin (SIII): a multisubunit regulator of elongation by RNA polymerase II, Science 269, 1439–1443 21 Kibel, A. et al. (1995) Binding of the von Hippel–Lindau tumor suppressor protein to elongin B and C, Science 269, 1444–1446 22 Aso, T. et al. (1996) The inducible elongin A elongation activation domain: structure, function and interaction with elongin BC complex, EMBO J. 15, 101–110 23 Duan, D.R. et al. (1995) Inhibition of transcriptional elongation by the VHL tumor suppressor protein, Science 269, 1402–1406
Reviews
MOLECULAR MEDICINE TODAY, JUNE 1999 (VOL. 5)
24 Kamura, T. et al. (1998) The elongin BC complex interacts with the conserved SOCS-box motif present in members of the SOCS, ras, WD-40 repeat, and ankyrin repeat families, Genes Dev. 12, 3872–3881 25 Lonergan, K.M. et al. (1998) Regulation of hypoxia-inducible mRNAs by the von Hippel–Lindau protein requires binding to complexes containing elongins B/C and Cul2, Mol. Cell Biol. 18, 732–741 26 Kipreos, E.T. et al. (1996) Cul-1 is required for cell cycle exit in C. elegans and identifies a novel gene family, Cell 85, 829–839 27 Peters, J-M. (1998) SCF and APC: the Yin and Yang of cell cycle regulated proteolysis, Curr. Opin. Cell Biol. 10, 759–768 28 Skowyra, D. et al. (1997) F-box proteins are receptors that recruit phosphorylated substrates to the SCF ubiquitin–ligase complex, Cell 91, 209–219 29 Lisztwan, J. et al. (1998) Association of human CUL-1 and ubiquitin-conjugating enzyme CDC34 with the F-box protein p45 (SKP2): evidence for evolutionary conservation in the subunit composition of the CDC34–SCF pathway, EMBO J. 17, 368–383 30 Bai, C. et al. (1996) Skp1 connects cell cycle regulators to the ubiquitin proteolysis machinery through a novel motif, the F-box, Cell 86, 263–274 31 King, R., Deshaies, R., Peters, J-M. and Kirschner, M.W. (1996) How proteolysis drives the cell cycle, Science 274, 1652–1659 32 Gorospe, M. et al. (1999) Protective function of pVHL against impaired protein processing in renal carcinoma cells, Mol. Cell Biol. 19, 1289–1300 33 Pause, A., Lee, S., Lonergan, K.M. and Klausner, R.D. (1998) The von Hippel–Lindau tumor suppressor gene is required for cell cycle exit upon serum withdrawal, Proc. Natl. Acad. Sci. U. S. A. 95, 993–998 34 Hynes, R. (1992) Integrins: versatility, modulation, and signalling in cell adhesion, Cell 69, 11–25 35 Pasqualini, R. et al. (1996) A polymeric form of fibronectin has antimetastatic effects against multiple tumor types, Nat. Med. 2, 1197–1203
36 Lieubeau-Teillet, B. et al. (1998) von Hippel–Lindau gene-mediated growth suppression and induction of differentiation in renal cell carcinoma cells grown as multicellular tumor spheroids, Cancer Res. 58, 4957–4962 37 Kopito, R.R. (1997) ER quality control: the cytoplasmic connection, Cell 88, 427–430 38 Tsuchiya, H., Iseda, T. and Hino, O. (1996) Identification of a novel protein (VBP-1) binding to the von Hippel–Lindau (VHL) tumor suppressor gene product, Cancer Res. 56, 2881–2885 39 Mukhopadhyay, D. et al. (1997) The von Hippel–Lindau tumor suppressor gene product interacts with Sp1 to repress vascular endothelial growth factor promoter activity, Mol. Cell Biol. 17, 5629–5639 40 Pal, S., Claffey, K., Dvorak, H. and Mukhopadhyay, D. (1997) The von Hippel–Lindau gene product inhibits vascular permeability factor/vascular endothelial growth factor expression in renal cell carcinoma by blocking protein kinase C pathways, J. Biol. Chem. 272, 27509–27512 41 Pal, S., Claffey, K., Cohen, H. and Mukhopadhyay, D. (1998) Activation of Sp1mediated vascular permeability factor/vascular endothelial growth factor transcription requires specific interaction with protein kinase Cz, J. Biol. Chem. 273, 26277–26280 42 Ivanov, S. et al. (1998) Down-regulation of transmembrane carbonic anhydrase in renal cell carcinoma cell lines by wild-type von Hippel–Lindau transgenes, Proc. Natl. Acad. Sci. U. S. A. 95, 12596–12601 43 Iliiopoulos, O. et al. (1996) Negative regulation of hypoxia-inducible genes by the von Hippel–Lindau protein, Proc. Natl. Acad. Sci. U. S. A. 93, 10595–10599 44 Levy, A.P., Levy, N.S. and Goldberg, M.A. (1996) Hypoxia-inducible protein binding to vascular endothelial growth factor mRNA and its modulation by the von Hippel–Lindau protein, J. Biol. Chem. 271, 25492–25497 45 Martinez-Zaguilan, R. et al. (1996) Acidic pH enhances the invasive behavior of human melanoma cells, Clin. Exp. Metastasis 14, 176–186
Late-breaking news Several reports appeared after this review was accepted for publication that are of direct relevance to the models shown in Figs 2 and 3. The crystal structure of pVHL bound to elongins B and C has similarities to SCF complexes1. In this structure, pVHL forms two subdomains, called a and b, both of which are frequently mutated in VHL disease. The a domain comprises three helices that bind to a helix in elongin C to form a four-helix cluster. One of the pVHL helices in this arrangement corresponds to pVHL residues 157–172 that, as noted earlier in this review, is sufficient to bind to elongins B and C in vitro. The b domain represents a putative substrate-docking site. Elongin C, as predicted, resembles Skp1 and the elongin C binding domain of pVHL loosely resembles an F-box. Together, the structure of pVHL bound to elongin B and C is consistent with the notion that pVHL, when bound to elongin B, elongin C and Cul2, targets proteins for covalent linkage to ubiquitin or ubiquitin-like molecules. This notion is further supported by the recent demonstration that pVHL co-purifies with a protein called Rbx1 (Ref. 2). This protein, which is also called ROC1, interacts with Cdc53/Cullinlike molecules and facilitates the ubiquitination of substrates by
SCF complexes by recruiting a ubiquitin-conjugating enzyme (E2; Refs 2–5). These recent discoveries will, no doubt, intensify and accelerate efforts aimed at determining whether pVHL plays a direct role in ubiquitination. 1 Stebbins, C.E., Kaelin, W.G. and Pavletich, N.P. (1999) Structure of the VHLElonginC-elonginB complex: implications for VHL tumor suppressor function, Science 284, 455–461 2 Kamura, T. et al. (1999) Rbx1, a component of the VHL tumor suppressor complex and SCF ubiquitin ligase, Science 284, 657–661 3 Ohta, T. et al. (1999) ROC1, a homolog of APC11, represents a family of cullin partners with an associated ubiquitin ligase activity, Mol. Cell 3, 535–541 4 Tan, P. et al. (1999) Recruitment of a ROC1-CUL1 ubiquitin ligase by Skp1 and HOS to catalyse the ubiquitination of lkBa, Mol. Cell 3, 527–533 5 Skowyra, D. et al. (1999) Reconstitution of G1 cyclin ubiquitination with complexes containing SCFGrr1 and rbx1, Science 284, 662–665 William Kaelin
263