|Home | About | Journals | Submit | Contact Us | Français|
Hypoxia inducible factors (HIFs) are transcription factors controlling energy, iron metabolism, erythropoiesis, and development, and, when dysregulated, contribute to tumorigenesis, cancer progression, and invasion. However, HIFα mutations have not previously been identified in any cancer. Here we report two novel somatic gain-of-function HIF2α mutations in two patients, one presenting with a paraganglioma and a second with both paraganglioma and somatostatinoma. Both mutations were shown to confer increased HIF2α activity and protein half-life. While germline mutations of regulators of HIFα, including VHL and EGLN1, have been reported in pheochromocytomas/paragangliomas, this is the first report of a somatic gain-of-function mutation in HIF.
Hypoxia-inducible factors (HIFs), originally described by Semenza et al., are transcription factors that respond to changes in tissue oxygen concentration.1 HIFs are highly conserved proteins composed of α and β subunits. The HIFβ subunit is constitutively expressed, whereas the α subunit is inducible by hypoxia and associated with the aggressive, treatment-refractory behavior of tumors.2, 3 Under normoxic conditions, HIF1α, HIF2α, and HIF3α are hydroxylated on specific prolyl residues, allowing for recognition by the von Hippel-Lindau (VHL) tumor suppressor protein and rapid degradation via the proteasome.4 Under hypoxic conditions, prolyl hydroxylation of HIFα proteins is reduced, resulting in their stabilization and in turn transcription of genes involved in the hypoxia response, including angiogenesis, glycolysis, apoptosis, proliferation, and growth.5, 6 Paragangliomas/pheochromocytomas are catecholamine-producing tumors derived from the chromaffin cells of extra-adrenal paraganglia.7 The pioneering work of Neumann et al. showed that about one fourth of these tumors were hereditary, a fraction now closer to one third since the identification of mutations in several additional genes, including SDHA, SDHAF2, TMEM127, and MAX.8, 9 Importantly, the identification of these genetic drivers led to the demonstration of HIFα stabilization and dysregulation in pheochromocytoma and paraganglioma tumors harboring mutations in VHL and succinate dehydrogenase (SDH).10, 11
Somatostatinomas are endocrine tumors producing somatostatin that are also believed to originate from neural crest cell origin. They are occasionally diagnosed in patients with VHL syndrome, multiple endocrine neoplasias type 1 or 2, or neurofibromatosis type 1.12
Polycythemia, a disease state in which the proportion of blood volume that is occupied by red blood cells increases, can be acquired or congenital; both forms can be primary (abnormality in erythroid progenitors causing increased erythroid proliferation), secondary (mediated by circulating erythropoietin), or related to abnormalities in hypoxia-sensing pathways (increased erythropoietin production and increased sensitivity of progenitors to erythropoietin).13 Germline mutations in VHL, EGLN1, or HIF2α result in increased erythropoietin and congenital polycythemia 14.
While mutations of VHL and EGLN1 have been associated with tumorigenesis of neural crest tumors due at least in part to a dysregulation of HIFα, until now no HIF mutations have been identified in these tumors. In this study, we have identified two novel somatic gain-of-function mutations in the HIF2α gene in multiple tumors from patients presenting with paraganglioma and somatostatinoma associated with polycythemia. These particular mutations disrupt prolyl hydroxylation of HIF2α, and in turn recognition by the VHL protein, resulting in a failure of HIF2α ubiquitination and degradation. Because these mutant HIF2α proteins have a longer half-life, targets downstream of HIF2α, including erythropoietin, are upregulated, leading to polycythemia. Our findings suggest the multiple phenotypes in these patients may represent an acquired disease syndrome initiated by gain-of-function mutations in HIF2α.
A thirty-year-old female was referred to our institution for evaluation of metastatic paraganglioma. From infancy the patient had polycythemia. At age 14 she was found to have a left periaortic paraganglioma, and at age 15 an aortocaval additional paraganglioma was detected. At age 23 an abdominal MRI showed multiple masses that were resected with diagnoses of paragangliomas (Table 1, Figure 1A). One year later she presented with a recurrent paraganglioma and 3 years later with at least 5 abdominal paragangliomas. At the time of her referral, CT and MRI demonstrated masses in the duodenum and pancreatic head, which were found to be a paraganglioma and somatostatinoma at the time of surgical resection (Figure 1A).
An eighteen-year-old female was referred to our institution for resection of a right “adrenal pheochromocytoma” and an aortic bifurcation paraganglioma. From birth she had cyanosis and was then diagnosed with polycythemia with normal sensitivity of erythroid progenitors to erythropoietin (Table 1). At age 18, after having episodes of hypertensive crises, she was found to have multiple paragangliomas at the time of surgical exploration. Two years later she again presented with another 3 cm periaortic paraganglioma that was successfully removed.
This study was approved by the IRB of the NICHD/NIH, and patients gave written informed consent.
Genomic DNA was extracted from either tumor tissue or white blood cells with the NucleoSpin Tissue Kit (Macherey-Nagel). HIF2α exons were amplified from genomic DNA by polymerase chain reaction (PCR). The primer sets for exon amplification were as previously reported.15 The DNA sequence of each exon was identified by forward and reverse sequencing.
Peptides corresponding to wild type HIF2α (ELDLETLAPYIPMDGEDFQ), ΔHIF2α-A530T HIF2α (ELDLETLTPYIPMDGEDFQ), and ΔHIF2α-A530V HIF2α (ELDLETLVPYIPMDGEDFQ) were synthesized and N-terminal biotinylated (Peptide 2.0).
The HA-HIF2α-pcDNA3 plasmid containing the human HIF2α gene coding sequence was obtained from Dr. William Kaelin (Addgene plasmid 1895016). The ΔHIF2α-A530T and ΔHIF2α-A530V mutations in the HIF2α gene were introduced into this plasmid using the Quikchange Lightning Site-Directed Mutagenesis Kit (Agilent). DNA sequences of the plasmids were verified by sequencing the entire coding region of the HIF2α gene. The coding sequence of the human VHL gene was inserted into a pCMV6-Entry vector (Origene). The HA-EGLN1 plasmid containing the human PHD2 gene was a generous gift from Dr. Richard Bruick.17
A HIF2α hydroxylation assay was performed as previously described.18 Peptide hydroxylation was analyzed by matrix-assisted laser desorption ionization time-offlight (MALDI-TOF) mass spectrometry as previously described.19 For detection of VHL binding after HIF2α hydroxylation, a radioactively labeled VHL protein was synthesized using the TNT Quick Coupled Transcription/Translation System (Promega) in the presence of [35S]-methionine (Perkin Elmer). EGLN1-treated beads were washed and incubated with [35S]-VHL for 2 hr at 4°C. Bound VHL was eluted by heating in an SDS-containing sample buffer and resolved on NuPAGE Bis-Tris Gels (Invitrogen).
Total RNA was extracted from microdissected tumor samples using the RNeasy Extraction Kit (Qiagen). The mRNA expression of hypoxia-related genes was examined through real-time PCR on an ABI 7500 real-time PCR system. The primer sets used were EDN1 (Origene HP205717), EPO (Origene HP200740), GLUT1 (Origene HP209446), VEGFA (Qiagen QT01682072), and β-Actin (Promega G5740).
We analyzed canonical hypoxia-related pathways by measuring gene expression in tumors from the index patients. A normal adrenal medulla specimen was used as a control. We found a marked up-regulation of HIF2 downstream-regulated genes, including EDN1, EPO, GLUT1, and VEGFA, in tumor specimens from both patients (Figure 1B), suggesting the clinical presentation of polycythemia was a consequence of hypoxia-related signaling.
To determine the causative gene responsible for polycythemia, we sequenced both germline DNA from leukocytes and DNA from the tumors. We identified two novel heterozygous mutations in the tumor specimens in both patients (Figure 1C). In patient 1, both the paraganglioma and somatostatinoma were found to have an identical G to A substitution at base 1588 in exon 12 of the HIF2α gene. In patient 2, the paraganglioma was found to have a C to T substitution at base 1589 in exon 12 of HIF2α. Neither of the mutations, nor other HIF2α mutations, was found in the germline DNA. Both the G1588A and C1589T mutations in HIF2α resulted in a substitution of alanine 530 in the HIF2α protein, with threonine in patient 1 and valine in patient 2. Analysis of tumor cDNA from both patients found equal proportions of the mutant and wild-type HIF2α alleles and no evidence of aberrant mRNA splicing.
Alignments of multiple HIF2α peptide sequences indicated that Ala530 is located in the vicinity of the primary hydroxylation site of the HIF2α protein and is conserved across different species and other HIF family members (Figure 1D). Amino acid substitution of Ala530 is likely to affect the conformation of the HIF2α hydroxylation domain and possibly interfere with prolyl hydroxylase recognition and VHL binding. To test this hypothesis, we performed in vitro hydroxylation assays, as previously described.18 Synthetic HIF2α peptides were incubated in the presence of PHD2. Prolyl hydroxylation of the HIF2α peptides was investigated by MALDI-TOF mass spectrometry (Figure 2A). Consistent with previous reports, 60.6% of wild type HIF2α peptides were hydroxylated. In contrast, only 9.1% and 16.7% of the ΔHIF2α-A530T and ΔHIF2α-A530V peptides were hydroxylated in the same assay. To further investigate the impact of the mutations on HIF2α signaling, we analyzed the binding of VHL to the HIF2α peptide. Radioactively labeled VHL protein showed strong binding to wild type HIF2α peptide (Figure 2B), consistent with efficient E3 ligase binding and the possibility of proteasomal degradation (Figure 2C). By comparison, the VHL protein had much less affinity for the ΔHIF2α-A530T and ΔHIF2α-A530V peptides, with only 18.4% and 24.0% of the VHL protein bound to the mutant peptides, respectively.
Finally, we determined the stability of mutant HIF2α using cycloheximide. The wild type HIF2α protein had a 14.4 min lifespan, in agreement with the rapid turnover of HIF2α in normoxic conditions. In contrast, ΔHIF2α-A530T and ΔHIF2α-A530V were more stable, with half-lives of 57.6 and 79.8 min, respectively (Figures 2D and 2E).
Previously, HIFα germline mutations have been reported in patients with familial polycythemia, with mutations at hot spots in exon 12 of the HIF2α gene.20, 21 Members of the HIFα family have also been demonstrated to play an important role in tumorigenesis.22 The interplay between HIF1α and HIF2α in tumorigenesis is unclear, but recent evidence suggests these two transcription factors may promote, initiate, or maintain different cell functions, resulting in varied tumor development and behavior, including prognosis.3, 23 Dysfunction of the VHL and SDH proteins as a result of gene mutations leads to HIFα dysregulation and overexpression and induces pseudo-hypoxia, a mechanism described in the pathogenesis of hereditary pheochromocytomas and paragangliomas.14 Furthermore, altered hydroxylation of proline residues on HIFα due to mutations of the prolyl hydroxylase gene can also cause downregulation of HIFα, which may also play a role in tumorigenesis.24 In aggregate, these studies have shown changes in the regulation of, but not mutations in, HIFα subunits in various tumors.
We now report somatic gain-of-function HIF2α mutations in paragangliomas and somatostatinoma associated clinically with polycythemia. These mutations were identified in the vicinity of the primary hydroxylation site of the HIF2α protein. They affected prolyl hydroxylation and VHL protein binding, resulting in reduced HIF2α degradation. We found increased erythropoietin mRNA and protein levels in both tumors that may result in a high serum erythropoietin level and polycythemia. However, we cannot rule out the possibility that HIF2α somatic mutations were present in precursor cells that secret excessive erythropoietin resulting in the early onset of polycythemia before gross tumor formation.
The identical somatic mutation detected in two separate and functionally distinct tumors, paraganglioma and somatostatinoma, in two young patients indicates the mutation may have occurred in a cell at early embryogenesis. The HIF2α mutant cells may have then been distributed throughout multiple organs during development. A similar genetic mechanism can be found in McCune-Albright syndrome, in which a post-zygotic mutation in GNAS1, a protein involved in G protein signaling, presents as a mosaic in only a fraction of cells.
The patients described herein presented with a rare combination of disease phenotypes, yet shared mutations of the same gene at a similar domain. Our findings raise several important questions, including (a) whether the HIF2α mutation is unique to paraganglioma and somatostinoma associated with polycythemia, and (b) whether patients presenting with paragangliomas or somatostatinomas need to be screened for HIF2α mutations.
We would like to acknowledge James W. Nagle for help with DNA sequencing and the expert assistance of Karen T. Adams, Thanh-Truc Hyunh, and Victoria Martucci in the production of this manuscript. We thank Dr. Anne-Paule Gimenez-Roqueplo and her staff at the Service de Génétique, Hôpital Européen Georges Pompidou, Paris, France for providing sequencing information for the PHD gene for one of the two patients. We also thank both patients and their families for their participation and assistance. This research was supported, in part, by the Intramural Research Program of the NIH, NICHD and NINDS.