Center for Human and Clinical Genetics (F.J.H.), Leiden University Medical Center, Leiden, The Netherlands; Department of Clinical Endocrinology and Laboratory of Endocrinology (J.W.M.H.), and Department of Clinical Endocrinology (C.J.M.L.), University Medical Center Utrecht, Utrecht, The Netherlands
Address all correspondence and requests for reprints to: Dr. C. J. M. Lips, University Medical Center Utrecht, Department of Clinical Endocrinology, P.O. Box 85500, 3508 GA Utrecht, The Netherlands. E-mail: c.j.m.lips{at}digd.azu.nl.
A germline mutation in the Von Hippel-Lindau (VHL) gene predisposes carriers to development of predominantly abundant vascularized tumors in multiple organs. The VHL gene is a tumor suppressor gene and is involved in blood vessel formation by regulation of the activity of hypoxia-inducible factor (HIF)-1. On the basis of its clinical expression, VHL disease has been divided into four subtypes with a central role for pheochromocytoma. In general, germline mutations predicting a loss of function of VHL protein(s) (pVHL) seem to be associated with VHL type 1 (involving both excessively vascularized hemangioblastoma and renal cell carcinoma, but not pheochromocytoma). In contrast, putative gain of function mutations is associated with VHL type 2C, involving nonexcessively vascularized pheochromocytoma only. Phenotypes 2A and 2B have, next to a low or high risk of renal cell carcinoma, respectively, development of both hemangioblastoma and pheochromocytoma. The co-occurrence of these two kinds of tumors is a consequence of a single amino acid substitution in pVHL but is putatively caused by opposing tumorigenic mechanisms (i.e. gain of function and loss of function). It is hypothesized that this apparent discrepancy might be explained by tissue-specific pVHL dosage effects, possibly in combination with tissue-specific involvement of other proteins mediating HIF-1
degradation.
Definition of VHL
Description of the syndrome.
VHL disease is an autosomal, dominant inherited tumor syndrome with an estimated prevalence of 23 per 100,000 persons (OMIM no. 193300). A germline mutation in the VHL gene predisposes carriers to tumors in multiple organs (1). These tumors may include hemangioblastoma in the retina (also referred to as retinal angioma), cerebellum, and spine; renal cell carcinoma (clear cell type); pheochromocytoma; islet cell tumors of the pancreas; endolymphatic sac tumors; and cysts and cystadenoma in the kidney, pancreas, epididymis, and broad ligament (2, 3). At present, metastases from renal cell carcinoma and neurological complications from cerebellar hemangioblastoma are the most common causes of death. However, periodic clinical examination and advanced operation techniques will reduce both morbidity and mortality in patients with VHL disease (2).
Clinical diagnosis.
In the presence of a positive family history, VHL disease can be diagnosed clinically in a patient with at least one typical VHL tumor (2, 3). Typical VHL tumors are retinal, spinal, and cerebellar hemangioblastoma; renal cell carcinoma; and pheochromocytoma (2). Endolymphatic sac tumors and multiple pancreatic cysts suggest a positive carriership (in the presence of a positive VHL family history), because they are uncommon in the general population. In contrast, renal and epididymal cysts occur very frequently in the general population and are, as sole manifestation, not reliable indicators for VHL disease (4). In patients with a negative family history of VHL-associated tumors, a diagnosis of VHL disease can also be made when they exhibit two or more hemangioblastoma or a single hemangioblastoma in association with a visceral manifestation (e.g. pheochromocytoma or renal cell carcinoma; Ref. 2).
Description of the VHL gene and expression of the gene
The VHL gene is a relatively small gene that covers approximately 14,500 basepairs (bp) of genomic DNA (1). The VHL gene encodes a ubiquitously expressed messenger RNA of 4,700 nucleotides, and the protein-coding region is contained in three exons.
The VHL gene is a tumor suppressor gene according to Knudsons two-hit hypothesis: for a normal cell, inactivation of both copies of the VHL gene is required to develop into a tumor cell. In carriers of a VHL gene germline mutation, tumors tend to occur multicentric and bilateral, and they manifest at a younger age than in patients without a VHL gene germline mutation, where each of the two VHL gene alleles in a cell has to become affected by an independent hit at the somatic level.
The VHL protein
Expression.
The VHL gene encodes a 213-amino acid protein with a molecular weight of about 2830 kDa (pVHL30) that does not closely resemble any other protein. A second VHL gene product of 160 amino acids and with a molecular weight of 18 kDa (pVHL18) arises from alternate translation initiation at an internal methionine codon (Met54) within the VHL gene open reading frame (5). In many cells, the shorter form is the more abundant of the two protein isoforms. Both VHL gene products behave similarly in the biochemical and functional studies performed to date. Therefore, both VHL proteins are referred to as pVHL in this review.
pVHL is widely expressed in normal human tissues and is even expressed in organs not at risk for the disease, suggesting a role for pVHL that goes beyond the organs involved in the disease (6). In human embryos, pVHL was expressed in all three germ layers, with strong expression noted in the central nervous system, kidneys, testes, and lungs (7).
Function.
In VHL patients, not only hemangioblastoma but also renal cell carcinoma show an abundance of blood vessels, which is driven, at least in part, by overproduction of vascular endothelial growth factor (VEGF; Ref. 8). Overproduction of VEGF mRNA is a feature of many human cancers and has been primarily linked to hypoxia. In 1999, it was demonstrated that pVHL plays a role in the degradation of the transcription factor HIF-1 (9). pVHL fulfills its function by binding to Elongin C, which is connected with Elongin B in a VCB-complex (Fig. 1
). Initially, a function of pVHL in transcription elongation via interaction with Elongins B and C was postulated. However, experimental data supporting such a function for pVHL have not been described. Instead, the VCB-complex targets substrates for intracellular degradation via a process called ubiquitination (10). In this process, the target is bound by a protein called ubiquitin that labels a protein for degradation (11). The destruction itself takes place inside a proteasome, a protein-digesting complex. Among the possible target genes regulated by HIF is the VEGF gene (9). Consequently, excessive blood vessel formation may occur when hypoxia-inducible proteins are not properly degraded. In other words, in cells lacking wild-type pVHL, degradation of HIF is impaired, and thus those cells behave as if they were deprived of oxygen.
|
VHL gene mutations: type and frequency
The VHL gene as a tumor suppressor gene.
Although VHL disease manifests itself as an autosomal dominant trait, it can be considered as a recessive trait at the cellular level. Similar to other hereditary tumors caused by germline inactivation of a tumor suppressor gene, the genotype of each typical VHL tumor in a VHL patient arises by a loss of the wild-type VHL allele, while maintaining the mutated allele. Such loss of heterozygosity (LOH; caused by deletion, nondisjunction, somatic recombination, etc.) at the VHL locus has been demonstrated in most VHL disease-associated tumors (13). Apart from LOH, other mechanisms of somatic inactivation of the VHL gene (point mutations, promoter hypermethylation) have also been observed in these tumors.
Germline mutations.
VHL gene germline mutations are consistently detected in 100% of classic families with more than one affected family member, or classic sporadic patients with multiple VHL-related tumors (14, 15). The mutation spectrum is heterogeneous, with mutations scattered throughout most of the VHL gene (Fig. 2). Missense mutations (leading to an amino acid substitution in pVHL) are found in 40% of the families with an identified VHL gene germline mutation (15). Microdeletions (118 bp), insertions (18 bp), splice site, and nonsense mutations, all predicted to lead to a truncated protein, are found in 30% of the families. Large deletions (including deletions encompassing the entire gene) account for the remaining 30% of the VHL gene germline mutations (15).
|
In nonfamilial (VHL-related) tumors, tumorigenesis is thought to be initiated by independent somatic alteration of both alleles of the VHL tumor suppressor gene. Somatic VHL gene mutations and allele loss are frequent events in sporadic clear cell renal cell carcinoma and sporadic central nervous system hemangioblastoma, suggesting that the VHL tumor suppressor gene plays a role in their tumorigenesis (13). However, somatic VHL gene mutations are uncommon in sporadic (i.e. non-tumor syndrome associated) pheochromocytoma (16).
Pathophysiology
Genotype-phenotype correlations.
On the basis of its clinical expression, VHL disease has been divided in four subtypes, with a central role for pheochromocytoma (Table 1). Patients with VHL type 1 have no pheochromocytomas. Mutations in patients with VHL type 1 are mostly loss of function mutations, leading to a truncated pVHL or no pVHL at all. Type 2 families have pheochromocytomas and are divided into subtypes with a low (type 2A) or high (type 2B) risk of renal cell carcinoma, whereas type 2C families present with pheochromocytoma only. In VHL type 1, missense mutations are commonly found in the ß-domain of the pVHL and are also predicted to lead to a loss of function of pVHL (10). In type 2A and type 2B, mutations hamper either the binding to Elongin C by the
-domain of pVHL or the capture of target proteins by the ß-domain of pVHL. Consequently, such target proteins (e.g. HIF) cannot be properly degraded in the proteasome (8, 9, 10). In contrast, VHL type 2C families are predominantly associated with specific missense mutations, notably in the
-domain. This will be discussed in more detail below.
|
The well vascularized phenotype of these VHL tumors suggests that inactivation of the VHL gene induces either up-regulation of an angiogenic factor or down-regulation of an inhibitor of angiogenesis. The pathophysiology of the richly vascularized hemangioblastoma can be readily explained with the HIF and ubiquitination theory as described above. Loss of pVHL function reduces HIF degradation and increases VEGF expression that leads in turn to angiogenesis (8, 9, 10). So far, there is no evidence for down-regulation of an inhibitor of angiogenesis in the tumorigenesis of hemangioblastoma.
Renal cell carcinoma.
In VHL patients, not only hemangioblastoma but also renal cell carcinoma show an abundance of blood vessels, which also suggests a central role for the HIF and ubiquitination theory. Recently, it was demonstrated that HIF activation (induced by pVHL inactivation) is an early event occurring in morphologically normal cells within renal tubules of patients with a VHL gene germline mutation (17). It has been shown that loss of the wild-type allele and retention of the inherited, mutated VHL allele occurs in both cystic lesions and renal cell carcinoma from VHL patients.
Pheochromocytoma.
The role of pVHL in pheochromocytoma is still under discussion. Pheochromocytoma differ from hemangioblastoma and clear cell renal cell carcinoma in that they are not richly vascularized and they are involved in several other tumor syndromes. The group of susceptibility genes for pheochromocytoma includes, next to the tumor-suppressor gene VHL, the proto-oncogene RET (associated with multiple endocrine neoplasia type 2), the NF-1 gene (associated with neurofibromatosis type 1), and the newly identified pheochromocytoma genes encoding succinate dehydrogenase subunit D and succinate dehydrogenase subunit B (18).
Most mutations of patients with VHL type 2 (predominantly specific missense mutations) are located in the -domain. These mutations allow a residual ability to bind Elongin C (and thereby form the VCB complex), as demonstrated for the L188V mutation that retains the ability to bind to and ubiquitinate HIF (19). Clearly, this does not explain why these missense mutations induce tumorigenesis. It may be that the change of one amino acid leads to a pVHL that binds other proteins. In other words, pVHL affected by a type 2 missense mutation would exert a gain of function (10). If one would imagine pVHL as a key, the change of one amino acid could lead to a key fitting in another lock that opens other doors (functional pathways). Alternatively, this type of mutation may display a dominant negative effect; i.e. the mutant protein may negatively influence activity of the wild-type protein encoded by the nonmutated allele. A third mechanism of action of VHL type 2 missense mutations, which would not require inactivation of the second allele, is via a gene dosage effect. In this model, different cell types would have individual thresholds for protein levels that determine cell proliferation, differentiation, and apoptosis (20). Thus, tumorigenesis of pheochromocytoma could be induced by a reduction of 50% in the level of normal pVHL, whereas only 50% of the normal level of pVHL would still be sufficient to prevent tumorigenesis in other cell types. In all three cases, i.e. gain of function, dominant negative effect, and gene dosage effect, one hit would be enough to induce tumorigenesis of chromaffin cells. In accordance with this notion, in VHL patients LOH is observed less frequently in pheochromocytoma compared with renal cell carcinoma and hemangioblastoma (16).
Another mechanism putatively involved in the initiation of pheochromocytoma is the binding of pVHL with fibronectin. It has been suggested that loss or change of fibronectin matrix assembly may be responsible for the development of pheochromocytoma (19). Hoffman et al. (19) demonstrated that products of VHL type 2C mutations of the VHL gene retain the ability to down-regulate HIF but are defective for the promotion of fibronectin matrix assembly. pVHL fulfills this function by direct binding to fibronectin (12).
Co-occurrence of loss and gain of pVHL function in VHL types 2A and 2B.
VHL type 1 (with hemangioblastoma and renal cell carcinoma) is thought to be due to loss of pVHL function, whereas VHL type 2C (pheochromocytoma only) is thought to be due to gain of pVHL function. On the basis of the available data, i.e. nonsense vs. missense mutations, frequent vs. less frequent occurrence of LOH, and excessive vs. nonexcessive tumor vascularization, one generally assumes a different tumorigenic mechanism for development of VHL type 1 and VHL type 2C. If this is correct, then one has to explain the co-occurrence of both phenotypes, putatively caused by opposing mechanisms, in VHL types 2A/2B, as a consequence of one and the same single amino acid change. A clue might be to postulate tissue-specific effects of loss of pVHL-mediated HIF degradation in VHL types 2A and 2B, possibly in combination with tissue-specific involvement of other proteins mediating HIF degradation. Consequently, a dominantly-acting missense mutation would cause pheochromocytoma by a gain of function effect (oncogene effect: not requiring a second hit), while not initiating development of hemangioblastoma or renal cell carcinoma, because partial pVHL-mediated HIF degradation is retained by the mutant allele. However, if the remaining normal allele is inactivated (LOH), near complete loss of pVHL-mediated HIF degradation would cause the hemangioblastoma and renal cell carcinoma, but not excessive angiogenesis in pheochromocytoma, due to a tissue-specific pVHL dosage effect and/or specific involvement of another protein. Also, in VHL type 1, 50% loss of pVHL-mediated HIF degradation by the germline nonsense mutation will not lead to hemangioblastoma and renal cell carcinoma until the second allele is lost by a somatic mutation. In this case, complete loss of pVHL-mediated HIF degradation does not cause pheochromocytoma, which might suggest that a residual HIF degradation by other proteins than pVHL is sufficient to prevent tumor formation in the adrenal medulla. In VHL type 2C, the missense mutation has the gain of function effect causing pheochromocytoma, but these particular missense mutations do not substantially affect pVHL-mediated HIF degradation. Thus, even loss of the second VHL gene allele would not lead to development of hemangioblastoma or renal cell carcinoma in patients with a type 2C mutation.
Diagnosing VHL disease
DNA analysis.
Clinical diagnosis of VHL disease can be confirmed by DNA analysis, i.e. mutation analysis of the VHL gene. Criteria for DNA analysis are presented in Table 2 (Refs. 21 and 22). Using direct sequencing of the protein-encoding region and quantitative Southern blot analysis, VHL gene germline mutations are identified in virtually all families with classic VHL disease (families with multiple tumors) and in the majority of isolated patients with multiple VHL-associated tumors (15). Techniques like single strand conformation polymorphism have been shown to miss certain VHL gene mutations (23). Because the size of the protein-encoding region of the VHL gene is small, it seems justified to perform screening for point mutations for VHL always by direct nucleotide sequence analysis of the entire protein-encoding region and splice junctions.
|
Periodic clinical surveillance.
Although DNA analysis means a psychological burden to the patient and closely related family members and the costs are high, tested individuals are no longer uncertain regarding their risk for developing the disease, and family members who are noncarriers are relieved of the burden of periodic examination. Clinical monitoring should be primarily organized around those VHL patients who have tested positive for a VHL gene germline mutation. In addition, the following persons should be monitored: first- and second-degree family members in a VHL family without an identified germline mutation; first-degree family members that decline a DNA test; patients (and first-degree family members) with a typical VHL tumor and features that suggest the presence of a germline mutation (i.e. the presence of multicentric or bilateral tumors, involvement in more than one organ, a suspected family history, and young age at diagnosis).
Methods for patient detection.
Because early detection, periodic clinical surveillance, and timely treatment of VHL patients lead to a better prognosis (2), information on the possibility of DNA analysis has to be provided to all family members. For privacy reasons, it is not allowed for health care workers to contact family members of VHL patients directly. With an information brochure (spread via the patient or informed family members), persons at risk for VHL disease can be informed and advised to seek genetic counseling themselves. In this way, they have a free choice for themselves, whether they (or their offspring) want to be tested. Patients can turn to VHL support groups (www.vhl.org) for support with emotional distress, information, and advice on social issues.
Concluding remarks
The VHL protein is small, but it is complicated by the various functions in distinct organ systems. On the basis of its clinical expression, VHL disease has been divided into four subtypes. In general, loss of pVHL function seems to be associated with VHL type 1 (involving hemangioblastoma and renal cell carcinoma), and gain of function with VHL type 2C (pheochromocytoma only). Tissue-specific pVHL dosage effects and/or tissue-specific expression of other genes might explain the involvement of both kinds of tumors in VHL types 2A and 2B, as a consequence of one and the same amino acid substitution in pVHL.
In addition to evidence for interfamilial variability, intrafamilial variability is a well observed characteristic in VHL disease. Apparently, there is no simple relationship between a germline mutation in the VHL gene (genotype) and the manifestation (age of onset and type) of VHL-related tumors (phenotype). It cannot be excluded that, apart from the germline VHL gene mutation, the sharing of (linked) modifier genes and/or environmental factors by the members of particular VHL families might be responsible for the VHL phenotype observed within that family. There are relatively too small and too few well studied VHL families to establish a definitive genotype-phenotype correlation. Therefore, current knowledge doesnt allow the implementation of genotype-phenotype correlations in individualized clinical monitoring of VHL disease gene carriers. Nevertheless, DNA diagnosis can greatly reduce VHL-associated mortality and morbidity by enabling early detection of disease gene carriers, periodic clinical surveillance, and timely treatment.
Footnotes
Abbreviations: bp, Base pair(s); HIF, hypoxia-inducible factor; LOH, loss of heterozygosity; pVHL, VHL proteins; VEGF, vascular endothelial growth factor; VHL, Von Hippel-Lindau.
Received October 14, 2002.
Accepted December 10, 2002.
References