Identification of Novel Mutations and Expressions of EPAS1 in Phaeochromocytomas and Paragangliomas

Endothelial PAS domain-containing protein 1 (EPAS1) is an oxygen-sensitive component of the hypoxia-inducible factors (HIFs) having reported implications in many cancers by inducing a pseudo-hypoxic microenvironment. However, the molecular dysregulation and clinical significance of EPAS1 has never been investigated in depth in phaeochromocytomas/paragangliomas. This study aims to identify EPAS1 mutations and alterations in DNA copy number, mRNA and protein expression in patients with phaeochromocytomas/paragangliomas. The association of molecular dysregulations of EPAS1 with clinicopathological factors in phaeochromocytomas and paragangliomas were also analysed. High-resolution melt-curve analysis followed by Sanger sequencing was used to detect mutations in EPAS1. EPAS1 DNA number changes and mRNA expressions were examined by polymerase chain reaction (PCR). Immunofluorescence assay was used to study EPAS1 protein expression. In phaeochromocytomas, 12% (n = 7/57) of patients had mutations in the EPAS1 sequence, which includes two novel mutations (c.1091A>T; p.Lys364Met and c.1129A>T; p.Ser377Cys). Contrastingly, in paragangliomas, 7% (n = 1/14) of patients had EPAS1 mutations and only the c.1091A>T; p.Lys364Met mutation was detected. In silico analysis revealed that the p.Lys364Met mutation has pathological potential based on the functionality of the protein, whereas the p.Ser377Cys mutation was predicted to be neutral or tolerated. The majority of the patients had EPAS1 DNA amplification (79%; n = 56/71) and 53% (n = 24/45) patients shown mRNA overexpression. Most of the patients with EPAS1 mutations exhibited aberrant DNA changes, mRNA and protein overexpression. In addition, these alterations of EPAS1 were associated with tumour weight and location. Thus, the molecular dysregulation of EPAS1 could play crucial roles in the pathogenesis of phaeochromocytomas and paragangliomas.


Introduction
Phaeochromocytomas and paragangliomas are rare catecholamines producing neural crest tumours derived from neuroendocrine chromaffin cells [1]. Phaeochromocytoma arises in the adrenal gland and paraganglioma in extra-adrenal chromaffin cells outside the adrenal gland [2,3]. The most common extra-adrenal location of this tumour is in the carotid body [3,4]. As the behaviour of this group of tumours is difficult to predict, the World Health Organization now classified the tumours as metastasizing and non-metastasizing (instead of benign and malignant) [5].
Recent studies have demonstrated that mutations in endothelial PAS domain-containing protein 1 (EPAS1) is associated with manifestations of phaeochromocytomas and paragangliomas [1,9,[12][13][14]. EPAS1, also known as HIF-2α, encodes for one of the hypoxia inducible factor (HIF) family members, which is involved in the hypoxic response [15]. The identification of somatic mutations affecting EPAS1 in phaeochromocytomas and paragangliomas led to the hypothesis that the stabilization of HIF-α plays crucial roles in the development of chromaffin tumours [12,14] through the phenomenon known as pseudo-hypoxia. In a pseudo-hypoxic state, abnormal HIF-α function enhances cell proliferation and tumour growth [16,17]. So far, EPAS1 mutations have been reported in a few phaeochromocytomas and paragangliomas [1,12]. Therefore, further research on the effects of EPAS1 mutation on gene expression in a large cohort of patients with phaeochromocytomas and paragangliomas is critical to unveil its roles in the phaeochromocytomas/paragangliomas pathogenesis. Herein, we investigated mutations in EPAS1 in a cohort of phaeochromocytoma (n = 57) or paraganglioma (n = 14) and have described the clinical, molecular and genetic features of patients carrying somatic EPAS1 mutations. In addition, EPAS1 DNA number variation, mRNA, protein expressions and their correlation with clinical and pathological features of patients with phaeochromocytomas and paragangliomas were also investigated.

Recruitments of Tissues and Sample Selection
Tumour tissues from patients who underwent resection of phaeochromocytomas (n = 57) and paragangliomas (n = 14) were collected from hospitals in Australia and Hong Kong during the period of 1973 to 2015. These patients were recruited prospectively with no selection bias. They were excluded from the study in cases of lacking adequate tumour tissue sampled or missing clinical data. Ethical approval for this work has been obtained from the Griffith University Human Research Ethics Committee (GU Ref No: MED/19/08/HREC and MSC/17/10/HREC).
The resected phaeochromocytomas/paragangliomas tissues were fixed in 10% formalin and embedded in paraffin wax for downstream analysis. Histological sections from these paraffin blocks were cut using microtome (Leika Biosystem, Wetzlar, Germany). The sections were then stained with haematoxylin and eosin (H&E) and examined under a light microscope for identifying the histopathological features of patients with phaeochromocytomas/paragangliomas by the author (A.K.L.). After reviewing the histological sections, a block was chosen from each of the phaeochromocytomas and paragangliomas from 71 patients (34 women; 37 men) and 10 non-neoplastic adrenal tissues, which act as controls (adrenal glands with no cancer collected from patients resected for renal cell carcinomas during the operation as a part of the procedure), to be included in this study. The selection of block from each case was based on having adequate tumour tissue portion (>70% area occupied by the tumour). The adrenal medulla tissue from non-neoplastic adrenal glands were micro dissected out for RNA and DNA extractions.

Clinical Data of the Selected Cases
The study population includes 56 Chinese patients from Hong Kong and 15 patients from Australia of European descent. The mean age of the patients with phaeochromocytomas and paragangliomas used in this study was 45 years (range, 2 to 79). There were 57 patients with phaeochromocytomas and 14 with paragangliomas. All the 14 patients with paraganglioma (located at the carotid body) and 8 patients with phaeochromocytomas do not have catecholamines detected clinically. Moreover, six patients with phaeochromocytomas harbouring EPAS1 mutation had shown catecholamine secretion. In addition, 4 had multiple endocrine neoplasia 2 (MEN 2) and 2 had neurofibromatosis. Of the 4 patients with MEN 2, one had bilateral phaeochromocytomas and one had two tumours in the same adrenal gland. Apart from these, there are 4 patients with bilateral phaeochromocytomas but with no known genetic predispositions. In addition, three patients have bilateral carotid paragangliomas. Furthermore, one patient had breast carcinoma and one had myocardial infarction.
The mean follow-up and median follow-up of the patients with phaeochromocytomas/ paragangliomas were 70.5 months and 54 months, respectively. Of these patients, 8 has metastasizing diseases. The interval between the date of surgery for tumours and the date of death or the closing date of the study was defined as the follow-up period in this study. The patients' actuarial survival rate was calculated from the date of surgical resection of phaeochromocytomas/paragangliomas to the date of death or last follow-up. The endpoint in the statistical analysis is defied by tumour-related death. A schematic flow chart of the experiments is shown in Figure 1.

Extractions of DNA and RNA
The selected tissues were sectioned using microtome (Leica Biosystem, Wetzlar, Germany) into 7 µm slices for RNA and DNA extractions [18]. The Qiagen DNeasy Blood and Tissue kit (Qiagen Pty. Ltd., Hilden, Germany) was used to extract the DNA according to the manufacturer's guidelines. The miRNeasy Mini kit (Qiagen Pty. Ltd., Hilden, Germany) was used to extract RNA from the tissue sections according to the manufacturer's protocol. Measurement of optical density (OD) using a nanodrop spectrophotometer was used to check the purity of the extracted DNA and RNA. The concentrations of DNA and RNA were noted in ng/µL and used for further analysis.

High-Resolution Melt (HRM) Curve Analysis
HRM analysis of exon 1 and exon 9 of EPAS1 gene sequence were carried out to screen the possible mutations in genomic DNA extracted from 71 tumours and 10 control tissues. The Rotor-Gene Q detection system (Qiagen Pty. Ltd., Hilden, Germany) was used to perform HRM by amplifying target sequences. The Rotor-Gene ScreenClust HRM Software was used to analyse the HRM curve. Exons 1 and 9 of the EPAS1 sequence were amplified using standard PCR system (10 µL). The reaction mixture consisted of 5 µL 2× sensimix HRM master mix, 1 µL of genomic DNA (30 ng/µL), 2 µL RNase free water and each primer 1 µL. Thermal cycling profile of the PCR was previously described [19]. All PCR experiment were carried out in triplicate including a no template (negative) control. In all assays, by increasing the temperature from 65 • C to 85 • C with a temperature increase rate of 0.05 • C/s and recording fluorescence, the melt curve data were generated.

Confirmation of Mutations by Sanger Sequencing
Sanger sequencing was used to confirm the possible mutations detected by HRM analysis. In short, the susceptible mutated PCR products were purified using the NucleoSpin ® Gel and PCR Clean-up kit (Macherey-Nagel, Bethlehem, PA, USA) according to the manufacturer's protocols followed by HRM analysis. Big Dye Terminator (BDT) chemistry Version 3.1 (Applied Biosystems, Foster City, CA, USA) under standardised cycling PCR conditions was used for sequencing the purified PCR products. A 3730xl Capillary sequencer (Applied Biosystems) was used analyse the generated data at the Australian Genome Research Facility (AGRF). Finally, the sequences were then examined using Sequence Scanner 2 software (Applied Biosystems).

In Silico Analysis
For computational analysis, the Ensembl transcript ID ENST00000263734 (EPAS1) was used as input. The identified mutations were computationally analysed using free tools, i.e., Mutation Taster with NCBI 37 and Ensembl 69 database release [20], protein variation effect analyser (PROVEAN) and sorting intolerant from tolerant (SIFT) to predict the impacts of the detected mutations on the protein functionality. In addition, the predicted outcomes were compared with ExAc and 1000 Genomes mutation databases. The cut-off values, −2.5 for PROVEAN and 0.05 for SIFT were used to predict the pathogenic/non-pathogenic mutations in the present study.
The PCR amplification efficiencies were normalised using multiple housekeeping genes, including beta-actin, 18s and GAPDH (glyceraldehyde 3-phosphate dehydrogenase). Finally, GAPDH was selected based on consistent results. DNA fold changes and mRNA expression were calculated according to the previously published protocol [19,24]. In this study, a fold change of more than 2 was considered as DNA amplification or high EPAS1 expression, whereas fold change of less 0.5 was considered as DNA deletion or low EPAS1 expression.

Immunofluorescence
The possible impact of EPAS1 mutation on protein expression in mutated (n = 8) and nonmutated (n = 8) tissues samples of phaeochromocytomas and paragangliomas was investigated by immunofluorescence analysis. For this, the tissue sections (4 µm) were de-waxed (xylene) and re-hydrated (alcohols and water). The tissues sections were then blocked with peroxidase block solution (Dako Australia Pty. Ltd., Sydney, NSW, Australia) and after that the sections were incubated with mouse EPAS1 (1:150) monoclonal antibody (Santa Cruz, Dallas, TX, USA) overnight at 4 • C. Then, the tissue sections were incubated with secondary antibody labelled with fluorescein isothiocyanate (FITC) (Sigma-Aldrich) at room temperature for two hours. Subsequently, the tissue sections were mounted on glass slides. Finally, the slides were examined under a confocal microscope (Nikon A1R+, Nikon Inc., Tokyo, Japan). The generated signals from EPAS1 was categorised as "0" (0% to less than 10%), "+" (10% to <30%), "++" (30% to <50%) and "+++" (>50%) according to the percentage and intensity of EPAS1 protein staining. Tissues in the categories of "0" and "+" were classified as No change or low expression, whereas "++" and "+++" were considered as high EPAS1 expression.

Statistical Analysis
For statistical analysis, all the data were entered into a computer database and analysed using the Statistical Package for Social Sciences for Windows (version 25.0, IBM SPSS Inc., New York, NY, USA). Variable groups were compared and analysed using the chi-square test, student t-test and Fisher's exact test. The Kaplan-Meier method used analysed the survival rates of patients. For all the analysis, the significance level was taken at p < 0.05.

Identification of Novel EPAS1 Mutations in Phaeochromocytomas/Paragangliomas
In the present study, genetic alterations in the EPAS1 sequence were noted in 12% (7/57) of phaeochromocytoma tumours (Table 1). Two novel mutations c.1091A>T and c.1129A>T were identified in exon 9 ( Figure 2). Both mutations were somatic heterozygous missense mutations (p.Lys364Met and P.Ser377Cys). In paragangliomas, 7% (1/14) of patients had shown EPAS1 mutations. The only c.1091A>T (p.Lys364Met) was identified tumour tissues in paraganglioma (Table 1). However, no mutant mutation was detected in non-neoplastic adrenal tissues. In silico analysis predicted that the identified mutation c.1091A>T (p.Lys364Met) was disease causing as it could cause changes in protein structure and function with pathological consequences (Table 1). Contrastingly, the c.1129A>T (P.Ser377Cys) mutation was predicted to be a polymorphism/neutral and was identified in two cases. These identified mutations were not found in the ExAc, PubMed or 1000 Genomes mutation databases.
Patients with phaeochromocytomas bearing mutation were four males and four females with a mean age of 41 years (age range 22-58). The mutations noted in this study were from functioning tumours. Among the seven patients with phaeochromocytoma harbouring mutated EPAS1, two patients had clinical confirmation of neurofibromatosis 1 (case 94 and case 122). Table 2 revealed the associations of identified mutations with the clinical and pathological factors of patients with phaeochromocytomas/paragangliomas ( Table 2). EPAS1 mutations occur in tumours with higher tumour weight (>50 gm) ( Table 2; p = 0.0001). Moreover, EPAS1 mutations were associated with the large tumour size (≥50 mm). Most of the mutated samples (7 out of 8) had larger tumours ( Table 2; p = 0.044). Other than this, there was no association between the mutations in EPAS1 with patients' age, sex, race or the side, location and presence of tumour metastasis in patients with phaeochromocytomas/paragangliomas.

EPAS1 Protein Expression in Phaeochromocytomas/Paragangliomas
The immunofluorescence staining of EPAS1 in the representative phaeochromocytomas/ paragangliomas and non-neoplastic adrenal tissue samples showed a different degree of staining under confocal microscopy ( Figure 4). Among the EPAS1 mutation-positive tumour samples (phaeochromocytomas, n = 7; paragangliomas, n = 1), 37.5% showed no change or low expression and 62.5% of samples had high EPAS1 protein expression (Figure 1). On the contrary, in EPAS1 mutation-negative tumour tissues (n = 8), 75% of samples showed no change or low expression and 25% of patients had shown high EPAS1 protein expression (Figure 1). The median overall follow-up of patients with phaeochromocytomas/paragangliomas was 54 months. There is no significant difference in survival rates among EPAS1-mutated, DNA numberchanged, and mRNA-altered groups were analysed (p > 0.05).

Association of EPAS1 DNA Number Variation, mRNA Expression, Protein Expression and Mutations in Phaeochromocytomas
A statistically significant positive correlation of EPAS1 DNA copy number changes and mRNA expression were noted in the present study (r = +0.538; p = 0.009, Fisher exact test). As shown in Figure 5A, 70.2% (40/57) copy number amplified tumour samples had higher EPAS1 mRNA expression, whereas EPAS1 mRNA downregulation was only noted in 71.5% (5/7) of the EPAS1 DNA number deletion tumours ( Figure 5A). The associations of EPAS1 mRNA expressions with copy number variation are presented in Figure 5B. Patients with EPAS1 DNA copy number amplification exhibited significantly higher mRNA expression in comparison to no change or deletion groups ( Figure 5B). Most of the tumour samples with no changes or deletion of EPAS1 DNA copy number exhibited similar results in mRNA expressions. On the other hand, most of the tumour samples with EPAS1 copy number amplifications had increased mRNA and protein expression. DNA copy number variations and mRNA expressions among EPAS1-mutated and non-mutated cases were also analysed ( Figure 6). EPAS1-mutated cases showed significant copy number amplification ( Figure 6A) and higher level of mRNA expression ( Figure 6B) when compared to those of non-mutated cases (p < 0.05). In addition, the majority (5/8) of tumours with EPAS1 mutations showed higher expression of protein when compared with non-neoplastic adrenal gland or mutation-negative tissues (Figure 4).

Discussion
This study identified novel mutations and clinicopathological implications of EPAS1 dysregulation in the pathogenesis of phaeochromocytoma/paraganglioma. The mutation analysis of EPAS1 in 71 phaeochromocytomas and paragangliomas resulted in the identification of two heterozygous somatic mutations, which have not been previously reported in phaeochromocytomas and paragangliomas. The detected mutation p.Lys364Met in exon 9 was predicted to be pathogenic to the functionality of EPAS1 protein in computational analysis. Contrastingly, the other mutation p.Ser377Cys was identified as a polymorphism and could act as tolerated or neutral on the protein functionality (Table 1). Moreover, in the present study, we have noted that the patients with phaeochromocytoma/paraganglioma having EPAS1 mutations had no mutations in phaeochromocytoma/paraganglioma-susceptible gene panels except NF1 in two patients with Neurofibromatosis 1. In addition, most of the mutated samples had shown gain-of-function of EPAS1, i.e., EPAS1 DNA number amplification, high mRNA and protein expression in the present study. Similar results were demonstrated in previous studies [1,12]. Furthermore, gain-of-function mutations (p.Phe374Tyr and p.Met368Ile) in exon 9 EPAS1 in phaeochromocytomas/paragangliomas associated with the increased stability of HIF2α [12,25]. Higher EPAS1 protein expression in mutated samples implied that the results of the current study are consistent with the previous findings. However, further studies are essential to determine the functional pathogenicity of the identified mutations.
EPAS1 mutations have been identified in several pathological conditions in humans, including congenital heart disease [26], erythrocytosis [27], Lynch syndrome [28], polycythaemia [29] and in various tumours, e.g., in paraganglioma [30,31], phaeochromocytoma [12], pancreatic adenocarcinoma [32]. Somatic EPAS1 mutations in different cancers indicate that these mutations may occur in a cell during embryogenesis or later, which in turn predispose the affected tissues or organ to form tumours [14]. In addition to the somatic mutation of EPAS1, inherited and constitutional mutations were associated with the pathogenesis of phaeochromocytoma and paragangliomas [1,25]. Moreover, the type of mutations along with additional accumulative genetic, epigenetic and environmental factors are attributed to the pathogenesis of phaeochromocytoma and paraganglioma.
The current study reports EPAS1 mutations in patients with phaeochromocytomas and paragangliomas and their correlation with various clinicopathological factors. The association of EPAS1 mutations with high tumour weight (p = 0.001) and larger tumour size (p = 0.044) implied that mutations of EPAS1 may contribute to the progression of this group of tumours. The underlying mechanism of EPAS1-induced carcinogenesis is poorly understood; however, it is reported that EPAS1 promote angiogenesis by interacting with both vascular endothelial growth factor (VEGF) and its receptor Fms Related Tyrosine Kinase 1 (Flt1) [33]. Thus, the gain-of-function mutations of EPAS1 can lead to increased expression of VEGF and Flt1 in endothelial cells, which in turn promotes angiogenesis, thereby promoting tumour growth and progression [33]. It was noted that the suppression of EPAS1 via shRNA in breast carcinomas cells reduced the cellular response and inhibited angiogenesis significantly, resulting in reduced tumour growth and development [34]. In addition, mutations in EPAS1 increased the stability of HIF2α leading to pseudo-hypoxic response, thus allowing for the activation of target genes and hence contributing to chromaffin cells tumorigenesis [35]. Therefore, mutations of EPAS1 may reduce HIF2α breakdown, which in turn could promote carcinogenesis by inducing pseudo-hypoxic conditions and promoting angiogenesis.
DNA copy number aberrations, dysregulated mRNA and protein level expressions in genes are commonly acquired changes in the cancer cells, thus playing a key role in the initiation and progression of cancers [36,37]. In the current study, the aberrant EPAS1 DNA number in patients with phaeochromocytomas/paragangliomas implied its potential roles in carcinogenesis. Similarly, the dysregulated expression of EPAS1 mRNA in tumour samples indicated the tumour-associated functionality of EPAS1 in phaeochromocytomas and paragangliomas. Importantly, the association of EPAS1 DNA number amplification (p = 0.037) and EPAS1 mRNA expression (p = 0.001) with tumour location (adrenal gland versus carotid body) suggested the clinical significance of EPAS1 in the carcinogenesis of phaeochromocytomas and paragangliomas. The differential EPAS1 DNA number and mRNA expression in phaeochromocytomas and paragangliomas implied that the aberration of EPAS1 could affect these tumours in a different manner. The difference in the molecular makeup of adrenal gland and carotid body may have contributed to this statistical significance of EPAS1 mRNA expression [38]. However, many patients with the metastatic tumours had shown lower EPAS1 mRNA expression when compared to that of non-metastatic patients (Table 4; p = 0.057). Although the statistical difference is above the cut-off value (p < 0.05), low metastatic sample size (n = 9) could be the contributing factor of this relationship.
The statistical relationship of EPAS1 DNA amplification and increased mRNA expression in patients with phaeochromocytomas/paragangliomas in this study indicated that hypoxic tumour niche induces molecular alterations of EPAS1, which, in turn, can promote carcinogenesis. Moreover, the DNA amplification and mRNA overexpression in patients with phaeochromocytomas/paragangliomas bearing EPAS1 mutations is indicative of the concerted aberration of EPAS1 in the pathogenesis for this group of tumours. In addition, higher expressions of EPAS1 protein in mutated samples indicate tumour-supporting roles of EPAS1 in the pathogenesis of phaeochromocytoma/paraganglioma. Previous studies reported that dysregulation and gain-of-function mutation of EPAS1 associated with neuroendocrine tumours such as paraganglioma and phaeochromocytomas by inducing pseudo-hypoxic tumour microenvironment [9,13,39]. Thus, the findings of the current study agree with previous studies.

Conclusions
This study has reported multiple novel EPAS1 mutations in patients with phaeochromocytomas/ paragangliomas. These mutations were noted to be related with altered expression and/or structural and functional changes in EPAS1, which in turn could play an important role in the pathogenesis of phaeochromocytomas and paragangliomas. In addition, the association of EPAS1 DNA number changes and mRNA expression with clinical and pathological factors, including tumour weight, size, location and the type of tumour, denotes the potential clinical significance of EPAS1 in predicting disease progression.