| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
, HIF-2
(EPAS1), and Their Target Genes in Paraganglioma and Pheochromocytoma with VHL and SDH Mutations
Molecular and Population Genetics Laboratory (P.J.P., I.P.M.T.), Histopathology Service (G.E.), In Situ Hybridisation Service (R.P., T.H., R.J., P.S.), and Bioinformatics & Biostatistics Service (G.K.), London Research Institute, Cancer Research UK, 44 Lincolns Inn Fields, London WC2A 3PX, United Kingdom; Imperial College Department of Histopathology (M.E.-B., G.W.S.), Division of Investigative Science, Hammersmith Hospital, DuCane Road, London W12 0NN, United Kingdom; Department of Otolaryngology (M.J.G.), Guys Hospital, London SE1 9RT, United Kingdom; Department of Clinical Genetics (J.B., S.V.H.), St. Georges Hospital, London SW17 0RE, United Kingdom; and Cancer Research UK Renal Molecular Oncology Group (P.K., F.L., E.R.M.), Section of Medical and Molecular Genetics, University of Birmingham School of Medicine, Institute of Biomedical Research, Birmingham B15 2TT, United Kingdom
Address all correspondence and requests for reprints to: Professor Eamonn R. Maher, Cancer Research UK Renal Molecular Oncology Group, Section of Medical and Molecular Genetics, University of Birmingham School of Medicine, Institute of Biomedical Research, Birmingham B15 2TT, UK. E-mail: e.r.maher{at}bham.ac.uk.
| Abstract |
|---|
|
|
|---|
Objective: Our objective was to compare protein (HIF-1
, EPAS1, SDHB, JunB, CCND1, CD34, CLU) and gene (VEGF, BNIP3) expression patterns in VHL and SDHB/D associated tumors.
Results: Overexpression of HIF-2 was relatively more common in VHL than SDHB/D PCC (12 of 13 vs. 14 of 20, P = 0.02), whereas nuclear HIF-1 staining was relatively more frequent in SDHB/D PCC (19 of 20 vs. 13 of 16, P = 0.04). In addition, CCND1 and VEGF expression (HIF-2 target genes) was significantly higher in VHL than in SDHB/D PCC. These findings suggest that VHL inactivation leads to preferential HIF-2 activation and CCND1 expression as described previously in VHL-defective renal cell carcinoma cell lines but not in other cell types. These similarities between the downstream consequences of VHL inactivation and HIF dysregulation in renal cell carcinoma and PCC may explain how inactivation of the ubiquitously expressed VHL protein results in susceptibility to specific tumor types. Both VHL and SDHB/D PCC demonstrated reduced CLU and SDHB expression. SDHB PCC are associated with a high risk of malignancy, and expression of (proapototic) BNIP3 was significantly lower in SDHB than VHL PCC.
Conclusion: Although inactivation of VHL and SDHB/D may disrupt similar HIF-dependent and HIF-independent signaling pathways, their effects on target gene expression are not identical, and this may explain the observed clinical differences in PCC and associated tumors seen with germline VHL and SDHB/D mutations.
| Introduction |
|---|
|
|
|---|
PCC from individuals with germline SDHB and SDHD mutations demonstrate evidence of HIF dysregulation and activation of hypoxia-inducible target genes (10, 11). A major function of the VHL tumor suppressor gene product (pVHL) is the targeting of the
-subunits of the HIF-1 and HIF-2 transcription factors for ubiquitination and proteolytic degradation (12, 13). The ability of pVHL to bind to and promote ubiquitination of the HIF-
subunits depends on the hydroxylation status of two proline residues (14, 15). Hydroxylation of these residues by EglN/PHD enzymes is dependent on the availability of oxygen, and in hypoxic conditions, the lack of prolyl hydroxylation renders the
-subunits resistant to pVHL binding; so in hypoxia and in cells with VHL tumor suppressor gene inactivation, stabilization of the
-subunits enables HIF-1 and HIF-2 to activate hypoxia-inducible target genes. Furthermore, inactivation of succinate dehydrogenase increases intracellular succinate, which impairs HIF prolyl hydroxylases, leading to HIF dysregulation (7). Consistent with the hypothesis implicating HIF dysregulation in the pathogenesis of VHL and SDHB/D associated PCC, gene expression microarray studies have demonstrated overexpression of hypoxia-induced angiogenic pathway genes in both VHL and SDHB/D PCC (16, 17). However, genotype-phenotype correlations in VHL disease have suggested that HIF dysregulation alone is not sufficient to cause PCC susceptibility such that rare VHL missense mutations that are associated with PCC susceptibility (but not RCC or hemangioblastomas, Type 2C VHL disease) retain the ability to regulate HIF-1 in vitro (18, 19, 20, 21). In addition, recently it was suggested that a HIF-independent defect in developmental apoptosis of sympathetic neurone precursor cells might contribute to PCC susceptibility associated with VHL and SDH subunit mutations (9). To gain further insight into the role of HIF-dependent and HIF-independent pathways in VHL and SDHB/D PCC, we undertook protein and gene expression studies in a large set of VHL and SDHB/D PCC and HNPGL. To investigate the role of pseudo-hypoxic isoform-specific regulation of HIF target genes in the pathogenesis of PCC and HNPGL, we selected genes (e.g. CCND1, BNIP3, and VEGF) that have previously been shown to be preferentially regulated by specific HIF isoforms, HIF1
and HIF2
(EPAS1), in VHL-defective RCC lines (28).
| Patients and Methods |
|---|
|
|
|---|
We analyzed fixed, paraffin-embedded, archival specimens of PCC and HNPGL from 36 individuals with germline mutations (mutation details available on request).
Riboprobe in situ hybridization
VEGF and BNIP3 35S-UTP in situ hybridization was performed as described previously (10, 22). The VEGF probe is common to all four splice variants (23). Patterns of hybridization were studied under dark-field reflected light, and expression was scored semiquantitatively as absent (), weak (+), moderate (++), strong (+++), or very strong (++++) by three independent observers (P.J.P., R.P., and M.E.-B.). A ß-actin probe was used as a positive control for each specimen.
Immunohistochemistry
Immunohistochemistry was carried out using standard procedures. Sections (4 µm) were deparaffinized in xylene, rehydrated, and endogenous peroxidase blocked with 3% H2O2. Antigen retrieval was carried out in a pressure cooker for 3 min at 16 psi in 0.01 M (pH = 6.0) citrate buffer (HIF-1
, CD34, SDHB, JunB, CCND1) or overnight in 50 mM Tris-EDTA (pH = 9) (EPAS1). Primary antibodies were as follows: HIF-1
(Nordic Immunology, Tillburg, The Netherlands, 1/1000), CD34 (Dako, Glostrup, Denmark; 1/25), SDHB (Molecular Probes, Invitrogen, Paisley, Scotland, UK; 1/1000), JunB (AQI Santa Cruz, Santa Cruz, CA; 1/50), CCND1 (Zymed, Invitrogen; 1/50). EPAS1 was a kind gift from Helen Turley and was used undiluted. Antibody-antigen binding was detected using appropriate biotinylated secondary antibodies (Dako) followed by addition of peroxidase-conjugated streptavidin (Vector Laboratories, Peterborough, UK). Sites of peroxidase activity were demonstrated using 3,3'-diaminobenzidine (Sigma, Haverhill, UK). Slides were counterstained with hematoxylin for 30 sec. Omission of primary antibody was used for negative controls, and appropriate positive controls were used for each antibody [VHL: mutant clear cell RCC (HIF1
, EPAS1), adrenal medulla (SDHB, CLU), breast carcinoma (CCND1), CD30+ lymphoma (JunB)].
The cellular distribution of the proteins between cytoplasm and nucleus was assessed, and the intensity of staining semiquantitatively was scored as absent (), weak (+), moderate (++), strong (+++), or very strong (++++). HIF-1
and HIF-2
were scored for nuclear staining (although we observed cytoplasmic staining in a few cases) using the above scale for intensity. For most cases the intratumor variability of the expression of the studied proteins was negligible. If there was a small focus that showed different intensity of expression (<10%), the case was assessed according to the predominant pattern. If there were larger different components, both were recorded as, for example, moderate to strong. For CD34, each section was assessed for volume fraction by counting the number of hits in 20 high power (x400) fields as viewed through a calibrated graticule. All immunohistochemistry was scored by two independent observers (P.J.P., M.E.-B.).
Statistical analysis
Numbers were assigned for ranking (not as a numeric value), for example, absent () = 0, weak/moderate (+/++) = 1.5. For the correlation analysis, we used Spearmans rank order correlation coefficient because the scoring data were ordinal. To calculate the P values on the comparisons among the various groups of tumors and among the groups of genes, we used proportional-odds logistic regression on the ordinal score data, and the
2 statistics were computed on the fitted models to calculate the P value against the null hypotheses score that didnt depend on PGL PCC status or, separately, VHL, SDHB, or SDHD status. All computations were carried out in R (24), and the logisitic regression was specifically calculated using polr from the MASS package (25). Fisher exact testing was also performed, and statistical significance was taken as 5%.
| Results |
|---|
|
|
|---|
|
nuclear staining was detected in 39 of 46 tumors analyzed, including 13 of 17 VHL PCC (four negative had weak to moderate cytoplasmic staining); 16 of 16 SDHB PCC; three of three SDHD PCC; three of four SDHB HNPGL (one negative had moderate cytoplasmic staining); and four of six SDHD HNPGL (one was negative for cytoplasmic staining, and the other had weak cytoplasmic staining). Increased HIF-2
expression (compared with matched normal tissue (taken as a score of 1.5) was detected in 27 of 38 tumors, including 12 of 13 VHL PCC, six of 15 SDHB PCC, four of four SDHD PCC, two of two SDHB HNPGL, and three of four SDHD HNPGL. Thus, overexpression of HIF-2 was relatively more common in VHL than SDHB/D PCC (12 of 13 vs. 10 of 19; P = 0.02), whereas nuclear HIF-1 staining was relatively more frequent in SDHB/D PCC (19 of 19 vs. 13 of 17; P = 0.04). In PCC with VHL mutations, there was a negative correlation between HIF-1
and HIF-2
expression (r = 0.41; P < 0.02).
HIF-1
expression in SDHB PCC and HNPGL correlated with expression of the BNIP3 target gene (Spearman r = 0.80, P < 0.01; and r = 0.61, P < 0.01, respectively).
Comparison of CD34 staining patterns in VHL and SDHB/D PCC revealed significantly greater vessel density in the former (mean = 121.38 vs. 81.17;
2 P < 0.001). Similarly, mean VEGF expression was significantly higher in VHL than in SDHB/D PCC (VHL mean = 2.50, SDHB/D mean = 1.64; P < 0.01) (Fig. 1A
). HIF-2
(EPAS1) expression was significantly higher in VHL than in SDHB/D PCC (2.35 vs. 1.79;
2 P < 0.01) (Fig. 1B
), and a similar pattern was observed for the HIF-2
target CCND1 (1.53 vs. 0,61;
2 P < 0.02). BNIP3 expression was significantly higher in VHL than in SDHB/D PCC (2.18 vs. 0.96;
2 P < 0.01) (Fig. 1C
). SDHB expression was significantly reduced in PCC from individuals with germline SDHB mutations compared with VHL PCC (2.40 vs. 1.86; P < 0.02) (Fig. 2A
). However, both VHL- and SDH-mutant PCC showed decreased SDHB expression in comparison with normal adrenal tissue (which scored 3.0).
|
|
Comparison of expression patterns between SDHB/D PCC and HNPGL was limited by small group sizes, but HNPGL demonstrated higher nuclear expression of HIF-1
than PCC (2.13 vs. 1.53;
2 P < 0.02), and CCND1 expression was significantly higher in SDHB PCC than SDHB HNPGL.
| Discussion |
|---|
|
|
|---|
and HIF-2
are the best characterized HIF-
isoforms and are both up-regulated by hypoxia and genetic events such as VHL inactivation. However, it appears that HIF-1 and HIF-2 regulate overlapping but distinct repertoires of target genes and may have differing effects in tumorigenesis (28, 29, 30). Thus, overexpression of HIF-2, but not HIF-1, was reported to overcome pVHL-induced tumor suppression in a RCC cell line (28, 29, 30). Interestingly, we found that overexpression of HIF-2 was significantly relatively more common in VHL than SDHB/D PCC (P = 0.02), whereas nuclear HIF-1 staining was significantly relatively more frequent in SDHB/D PCC (P = 0.04). Recently, Raval et al. (28) reported an unexpected reciprocal interaction between HIF-1
and HIF-2
levels in RCC, such that, in this tissue, overexpression of HIF-2
down-regulated HIF-1
protein expression, and HIF-2
siRNA was associated with up-regulation of HIF-1
expression. Immunohistochemical staining of nephrectomy specimens from VHL patients has demonstrated that activation of the HIF system is an early feature of disease, such that up-regulation of HIF-1
occurs in very early lesions, whereas HIF-2
up-regulation occurs later and is most prominent in overt RCC (28, 31). We found an inverse relationship between HIF-1
and HIF-2
expression in VHL PCC. These findings suggest that the reciprocal relationship between HIF-1
and HIF-2
expression described previously for VHL RCC is also a feature of VHL PCC and that, as for RCC, HIF-2
and HIF-1 dysregulation may have differential effects on PCC growth in VHL disease.
We found that CCND1 expression was significantly higher in VHL PCC than SDHB/D PCC. This finding is consistent with relative expression patterns of HIF-1
and HIF-2
in VHL and SDHB/D PCC because CCND1 is preferentially regulated by HIF-2
(28). The effect of hypoxia on CCND1 expression is tissue specific. Thus, whereas in RCC cell lines, CCND1 expression is induced by hypoxia, in bladder, breast, and other cancer cell lines, there is no effect (32, 33). We note that Cyclin D expression is induced by RET mutations associated with MEN 2A and MEN 2B, further implicating CCND1 up-regulation in PCC tumorigenesis (34). An intriguing feature of VHL disease is the apparent tissue specificity of the pVHL gatekeeper role, despite ubiquitous tissue expression (35). That up-regulation of CCND1 in response to VHL inactivation/HIF activation is a feature of RCC and PCC but not other tumor types suggests that tissue-specific differences in HIF-target genes could determine the VHL disease tumor spectrum. Interestingly, in SDHB/D tumors, CCND1 expression was significantly higher in PCC than HNPGL, suggesting the sympathetic neuron-derived PCC cells and parasympathetic neuron-derived HNPGL cells may differ with respect to hypoxic regulation of CCND1 expression.
BNIP3 is a proapoptotic protein that is up-regulated by p53 and HIF-1 (36). In contrast, HIF-2 is a negative regulator of BNIP3 (28). In nephrectomy specimens from VHL patients, there was an inverse relationship between HIF-2/CCND1 expression and BNIP3 expression (with the former being expressed in larger later-stage lesions) (28). Interestingly, despite preferential overexpression of HIF-1
in SDHB/D PCC and HIF-2
in VHL PCC, we found significantly lower levels of BNIP3 expression in SDHB/D PCC. Thus, loss of BNIP3 expression in SDHB/D PCC may be related to HIF-independent pathways and, although none of the SDHB PCC that we analyzed was known to be malignant, differentiating malignant and benign PCC is difficult, and with prolonged follow-up, some tumors initially thought to be benign may prove to be malignant. Further investigations are required to elucidate whether loss of BNIP3 expression might be implicated in the increased frequency of malignant transformation reported in SDHB PCC (37). Recently, it has been suggested that VHL inactivation and HIF dysregulation lead to a secondary loss of SDHB expression (17). However, although we observed decreased expression of SDHB in both SDHB and VHL PCC (compared with normal adrenal), we found that SDHB protein expression was significantly higher in VHL PCC than in SDHB PCC.
Previous studies have emphasized the similarities in gene expression patterns between VHL and SDHB/D PCC (10, 11, 17). Nevertheless, although VHL and SDHB/D inactivation both result in HIF dysregulation and defects in developmental apoptosis (7, 8, 9), there are important clinical differences between the molecular subgroups. Thus, SDH subunit mutations are associated with a higher frequency of extra-adrenal PCC, and malignancy is frequent in SDHB PCC but uncommon in VHL PCC (37, 38). We have demonstrated significant differences in gene expression patterns between VHL and SDHB/D PCC. These differences may relate to differing mechanisms of HIF dysregulation and/or HIF-independent pathways of tumorigenesis. At present, the relative importance of HIF dysregulation and HIF-independent pathways in PCC tumorigenesis is unclear. Overexpression of HIFs and downstream target genes was detected in almost all PCC analyzed, but in vitro studies of RCC cell lines suggested that HIF dysregulation might not be an absolute requirement for PCC development. VHL Type 2C (PCC only) mutations retained the ability to regulate HIF (19, 21). Recently it was reported that VHL inactivation was associated with loss of CLU and up-regulation of JunB expression via a HIF-independent pathway (9, 39). We found that, as reported previously, there was a marked loss of CLU expression in VHL RCC and reduced expression in VHL PCC (39). However, CLU expression was similarly reduced in SDHB/D PCC. This evidence for shared HIF-independent mechanisms of tumorigenesis in two forms of familial PCC is consistent with the suggestion that loss of developmental culling involving a JunB/cJun/EglN3 is a common feature to all familial PCC syndromes (9). Because somatic inactivation of familial cancer genes is rare, further studies are indicated to determine if CLU immunohistochemistry might provide a biomarker for indicating which individuals with apparently sporadic PCC should be screened for germline mutations in VHL, SDHB, or SDHD.
| Acknowledgments |
|---|
| Footnotes |
|---|
Abbreviations: CLU, Clusterin; HNPGL, head and neck paragangliomas; MEN, multiple endocrine neoplasia; PCC, pheochromocytoma; pVHL, VHL tumor suppressor gene product; RCC, renal cell carcinoma; VHL, von Hippel-Lindau.
Received May 1, 2006.
Accepted August 29, 2006.
| References |
|---|
|
|
|---|
prolyl hydroxylase. Cancer Cell 7:7785[CrossRef][Medline]
in tumours which result from germline FH and SDH mutations. Hum Mol Genet 14:22312239
binding and ubiquitylation by the von Hippel-Lindau tumor suppressor protein. J Biol Chem 275:2573325741
to the von Hippel-Lindau ubiquitylation complex by O2-regulated prolyl hydroxylation. Science 292:468472
targeted for VHL-mediated destruction by proline hydroxylation: implications for O2 sensing. Science 292:464468
regulatory loop links hypoxia and mitochondrial signals in pheochromocytomas. PLoS Genet 1:7280
regulation by disease-causing pVHL mutations correlate with patterns of tumourigenesis in von Hippel-Lindau disease. Hum Mol Genet 10:10291038
to the phenotype of VHL loss in renal cell carcinoma. Cancer Cell 1:247255[CrossRef][Medline]This article has been cited by other articles:
![]() |
U. Srirangalingam, B. Khoo, L. Walker, F. MacDonald, R. H Skelly, E. George, D. Spooner, L. B Johnston, J. P Monson, A. B Grossman, et al. Contrasting clinical manifestations of SDHB and VHL associated chromaffin tumours Endocr. Relat. Cancer, June 1, 2009; 16(2): 515 - 525. [Abstract] [Full Text] [PDF] |
||||
![]() |
C. Jimenez, M. E. Cabanillas, L. Santarpia, E. Jonasch, K. L. Kyle, E. A. Lano, S. F. Matin, R. F. Nunez, N. D. Perrier, A. Phan, et al. Use of the Tyrosine Kinase Inhibitor Sunitinib in a Patient with von Hippel-Lindau Disease: Targeting Angiogenic Factors in Pheochromocytoma and Other von Hippel-Lindau Disease-Related Tumors J. Clin. Endocrinol. Metab., February 1, 2009; 94(2): 386 - 391. [Abstract] [Full Text] [PDF] |
||||
![]() |
B E Baysal Clinical and molecular progress in hereditary paraganglioma J. Med. Genet., November 1, 2008; 45(11): 689 - 694. [Abstract] [Full Text] [PDF] |
||||
![]() |
C. Ricketts, E. R. Woodward, P. Killick, M. R. Morris, D. Astuti, F. Latif, and E. R. Maher Germline SDHB Mutations and Familial Renal Cell Carcinoma J Natl Cancer Inst, September 3, 2008; 100(17): 1260 - 1262. [Abstract] [Full Text] [PDF] |
||||
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |