Whole Exome Sequencing to Identify a Novel Gene (Caveolin-1) Associated with Human Pulmonary Arterial Hypertension

Abstract

Background

Heritable and idiopathic pulmonary arterial hypertension (PAH) are phenotypically identical and associated with mutations in several genes related to TGF beta signaling, including bone morphogenetic protein receptor type 2 (BMPR2), activin receptor-like kinase 1 (ALK1), endoglin (ENG), and mothers against decapentaplegic 9 (SMAD9). Approximately 25% of heritable cases lack identifiable mutations in any of these genes.

Methods and Results

We used whole exome sequencing to study a three generation family with multiple affected family members with PAH but no identifiable TGF beta mutation. We identified a frameshift mutation in Caveolin-1 (CAV1), which encodes a membrane protein of caveolae abundant in the endothelium and other cells of the lung. An independent de novo frameshift mutation was identified in a child with idiopathic PAH. Western blot analysis demonstrated a reduction in caveolin-1 protein, while lung tissue immunostaining studies demonstrated a reduction in normal caveolin-1 density within the endothelial cell layer of small arteries.

Conclusions

Our study represents successful elucidation of a dominant Mendelian disorder using whole exome sequencing. Mutations in CAV1 are associated in rare cases with PAH. This may have important implications for pulmonary vascular biology as well as PAH-directed therapeutic development.

Pulmonary arterial hypertension (PAH) is a highly fatal disease resulting in progressive right ventricular failure and death within 5 years of diagnosis for the majority of patients¹. PAH preferentially affects females and may occur as an idiopathic (IPAH) disease with sporadic presentation, or as heritable PAH (PAH) that occurs in families or those with a detectable mutation in a PAH-associated gene.^{2, 3} HPAH is a Mendelian disorder characterized by autosomal dominant transmission with reduced penetrance ⁴.

Mutations in bone morphogenetic protein receptor type 2 (BMPR2), a member of the transforming growth factor beta (TGF beta) superfamily of receptors^{5, 6} have been identified in ~75% of cases of HPAH, as well as ~15% of cases of IPAH⁷. Three additional genes have been identified that associate with PAH in a minority of cases, and all three are intimately related to TGF beta signaling: activin receptor-like kinase 1 (ALK1), endoglin (ENG), and mothers against decapentaplegic 9 (SMAD9).⁷ Despite advances in our understanding of the etiology of PAH, 25% of HPAH and 85% of IPAH cases have no identified genetic association and the inciting pathogenic events that promote PAH have not been precisely elucidated⁷.

The identification of new genes for PAH may provide novel insights into its pathogenesis and identify pathways which could facilitate new strategies for diagnosis, disease prevention, and therapy. Given advances in sequencing throughput and recent successful application for other genetic diseases,⁸ we used whole exome sequencing to evaluate a large family with HPAH with no identifiable mutation in the known genes for PAH. We discovered a new gene associated with HPAH that was confirmed by the independent finding of a de novo mutation in a second family.

Methods

Participants

We previously studied a family of European descent with 6 affecteds in three generations segregating PAH in a dominant manner (Figure 1). This family had no identifiable mutations in BMPR2 by dideoxy sequencing or multiple ligation-dependent probe amplification (MLPA) and no identifiable mutations in ALK1, ENG, or SMAD9 by exon sequencing.⁹ Patients in this family ranged in age from 4 to 67 years at the time of diagnosis.

Figure 1.

(A) Pedigree of family with pulmonary arterial hypertension. c.474delA, P158PfsX22 was identified in nine family members. In the four-generation family, 6 members were clinically affected (filled squares or circles). Subjects’ current ages and ages at diagnosis are labeled under each family member tested for the CAV1 mutation. CAV1 genotyping results are represented for PAH patients in this family; however, CAV1 genotyping results for healthy subjects are not displayed out of respect for the family represented. : male; : female; : affected male; : affected female; : affected female deceased. + indicates heterozygous for the c.474delA CAV1 mutation and − indicates normal CAV1 sequence. (B) Sequencing results of the CAV1 mutations. c.474delA was identified in a family with pulmonary arterial hypertension (label: II-4). c.463G>A V155I and c473delC, P158HfsX22 were identified in one patient previously diagnosed as IPAH (label: IPAH929).

DNA samples were available from 12 members across three generations (Figure 1). Four affected members of the family (II-4, III-2, III-5, IV-2) were analyzed by whole exome sequencing. The remaining 8 subjects were analyzed by Sanger sequencing. Two hundred and sixty unrelated Caucasian patients with HPAH (62) and IPAH (198) were subsequently studied in the replication arm of this study. Written informed consent for all studies was obtained in agreement with protocols approved by the institutional review boards (IRB) at Vanderbilt University and Columbia University Medical Centers. All patients met established international criteria for the diagnosis of PAH.²

Whole exome Sequencing and Sanger Sequencing

DNA was extracted from peripheral blood leukocytes using Puregene reagents (Gentra Systems Inc., Minnesota, USA). The four patients (II-4, III-2, III-5, IV-2; Figure 1) were analyzed by whole exome sequencing after capture and library preparation using the Agilent SureSelect Human All Exon Kit adapted for the SOLiD sequencing platform. Sequencing was performed on a SOLiD3 according to the manufacturer’s directions. Analysis was performed to identify the single nucleotide and indel variants shared among the four individuals using BWA (Burrows-Wheeler Aligner)¹⁰ and SAMtools¹¹ and screened using dbSNP build (129) on the human assembly hg18, SIFT, and the 1000 Genomes Project to remove polymorphisms. Novel variants present in all four patients were analyzed for effect on the amino acid sequence, assigned a coverage-dependent Phred-scaled mutation probability, and analyzed for predicted effect on the protein using SIFT and PolyPhen. Shared non-synonymous coding variants were confirmed in all 4 patients by Sanger sequencing. The full methods used for exome sequencing, Sanger sequencing, and sequence analysis are described in the Supplementary Appendix.

Skin Biopsy and Molecular Analysis of Cultured Cells

Skin biopsy specimens were obtained via a sterile 3mm punch skin biopsy technique from 3 of the PAH patients described above and 2 healthy controls.¹² Primary skin fibroblasts were cultured using standardized measures.¹³ All cell lines were grown in the same manner using DMEM (with 4.5 g/L glucose, L-glutamine, and sodium pyruvate) (Mediatech Inc., Manassas, VA) with 20% FBS (Invitrogen Co., Carlsbad, CA). Western blot analyses were performed using an anti–caveolin-1 antibody with immunogenicity to residues near the C-terminus of human caveolin-1 (Epitomics Inc., Burlingame, CA). Detection was performed using the Immobilon Chemiluminescent HRP substrate (Millipore, Billerica, MA). β-Actin (BD Biosciences Inc., Franklin Lakes, NJ) was used as a loading control. The highest density protein bands were considered 1 U, and the other lanes were normalized to this value. The experiments were performed in duplicate.

Lung Tissue Sampling

Lung tissue was obtained at open-lung biopsy for clinical diagnosis from one of the PAH patients studied, and that tissue was available for immunohistochemistry studies. While lung biopsy is not a routine component of PAH diagnosis, the patient (Table 1, subject IPAH929) was a very young child in whom the diagnostic process required a lung biopsy for confirmation of PAH in the opinion of the expert clinicians providing her care prior transfer of her care to us. This is not uncommon in pediatric PAH. That biopsy was consistent with a diagnosis of PAH, and supported to her diagnosis at that time of IPAH. The lung tissue was fixed in 10% formalin, processed, embedded in paraffin, sectioned and stained with hematoxylin-eosin, CD31, alpha smooth muscle actin (SMA), or von Willebrand factor (vWF), and Verhoeff-Van Gieson (VVG) Elastic Staining.

Table 1.

Clinical characteristics of CAV1 mutation carriers with pulmonary arterial hypertension.

	II-4	III-2	III-5	IV-1	IV-2	IV-3	IPAH929
Sample available for genetic testing	Yes	Yes	Yes	Yes	Yes	No	Yes
Diagnosis age, yrs	67	36	41	9.5	4	4	1
Death age, yrs	NA	NA	NA	NA	NA	4	NA
Current age, yrs	71	46	48	25	21	NA	5
Gender	M	F	F	M	F	F	F
Body Mass Index kg•m−2, at diagnosis	21.7	18.9	27.6	17.0	11	11	14.5
NYHA functional class, at diagnosis	2	1	2	1	NA	NA	NA
Baseline hemodynamic data, at diagnosis Right atrial pressure mmHg Mean pulmonary artery pressure mmHg Cardiac index L•min•m−2 Indexed PVR U•m2 Responsive to acute vasodilator challenge	11 53 NA NA NA	9 41 2.7 6.9 No	17 100 1.3 36 No	10 56 3.1 11 Yes	7 60 3.2 17 Yes	NA NA NA NA NA	10 53 3.2 12.6 Yes
PAH-specific therapies Calcium channel blocker Prostanoids Phosphodiesterase-5 inhibitors Endothelin receptor antagonists	Yes No No Yes	Yes Yes No Yes	No Yes Yes Yes	Yes No No No	Yes Yes Yes No	Yes Yes No No	Yes No Yes No

Immunohistochemical Analysis of caveolin-1 in the Small Arteries of Human Lungs

Immunolocalization of caveolin-1 within the small pulmonary arteries of human lungs was performed on archival paraffin-embedded human lung tissue obtained from controls and from one PAH patient (IPAH929) in whom a CAV1 mutation was detected and lung tissue available for analysis. Lung sections (5 µm thick) were deparaffinized and rehydrated. Heat mediated antigen retrieval was performed using 10mM Citrate buffer (pH 6.0) followed by blocking with 5% normal goat serum. The sections were incubated overnight with caveolin-1 antibody (BD Transduction Laboratories, San Jose, CA) (dilution 1:400) at 4°C. The sections were then incubated with Alexa Flour 488-labeled secondary antibody (Invitrogen, Carlsbad, CA). The slides were mounted using Vectashield mounting media containing DAPI (Vector Laboratories, Burlingame, CA) for confocal microscopy. The pictures were digitized. To preserve the relative fluorescence variation, the microscopy settings were the same for all the samples.

Results

Whole exome Sequencing of PAH Patients from One Family

We sequenced four samples with an average of 87.3M reads per sample. In total, we identified 54,540 genomic locations with variations, and 10,088 (18%) of these locations had shared variations across all 4 samples. The single nucleotide and indel variants were compared among the four individuals. There were 653 heterozygous variants shared among all four family members that were screened using dbSNP build on the human assembly hg18, SIFT, and the 1000 Genomes Project. Fifty two novel variants were present in all four patients, of which 31 were in coding regions and 16 were predicted to cause non-synonymous changes and analyzed for effect on the amino acid sequence. Non-synonymous variants, splice mutations, and coding insertions/deletions were also evaluated. These 16 variants were sequenced using dideoxy sequencing and 11 (see Appendix) were confirmed in all four samples.

Gene Mutation Confirmation in the Family

We used DNA from an additional family member with PAH (IV-1, who did not have whole exome sequencing) to attempt to narrow the possibilities among the 11 non-synonymous coding variants. Only 3 of the non-synonymous coding variants were also carried by this subject. These three variants were S36T in olfactory receptor family 1 subfamily Q member 1 (OR1Q1), H379Y coagulation factor II receptor like 1 (F2RL1), and c.474delA (P158P fsX22) in Caveolin-1 (CAV1). When we analyzed these variants using SNAP¹⁴, SIFT, and PolyPhen, the S36T OR1Q1 and the H379Y F2RL1 variants were predicted to have neutral effects with a reliability index of 2 and 4 and an expected accuracy of 69 and 85%, respectively. The c.474delA in CAV1 was predicted to cause a frameshift P158P fsX22 and add 21 novel amino acids at the C terminal domain of caveolin-1 protein.

All of the other available family members were then genotyped although results of the unaffected individuals are not displayed out of respect to the family’s privacy. All PAH patients carry the c.474delA mutation in CAV1. Several family members with normal physical exam and without detectable PAH also carry the c.474delA CAV1 mutation, which suggests that PAH associated with this CAV1 mutation demonstrates incomplete penetrance similar to that observed with other PAH associated genes (Table 1) ⁷.

Molecular Analysis: Reduced CAV1 Expression

Western blot analysis of cellular extracts derived from 3 of the HPAH patients’ (IV-1, II-4, III-III-2) primary fibroblasts and primary fibroblasts from 2 healthy control subjects was performed. This data supported the sequence data presented in Figure 1B. Specifically, while present, caveolin-1 protein was reduced in cellular extracts from all 3 HPAH patients (Figure 4, lanes 1–3) compared to controls. (Figure 2, lanes 4–5).

Figure 4.

Reduced caveolin-1 expression by lung small artery endothelial cells in CAV1-PAH. Caveolin-1 expression by CAV1-PAH lung endothelial cells was reduced along the plasma membrane of the vessel lumen compared to wild-type healthy control lung small arteries. Representative micrographs of immunostaining of lung sections with anti-caveolin-1 (green). Nuclei were counterstained with DAPI (blue). 100× magnification. Arrows represent the arterial lumen membrane of the endothelial cells.

Figure 2.

Reduced caveolin-1 expression by cells derived from patients with CAV1-PAH. Caveolin-1 expression by CAV1-PAH fibroblasts in culture compared to healthy wild-type controls. While present, CAV1-PAH cells exhibited reduced expression of Caveolin-1 compared to healthy wild type control fibroblasts cultured under identical conditions. β-actin was used as loading control. The highest density protein band (Lane 5) was considered 1.0 U, and the other lanes were normalized to this value.

Replication in a Cohort of Unrelated PAH Patients

We next sequenced the coding exons of CAV1 in 62 unrelated HPAH and 198 IPAH patients without detectable BMPR2 mutations. One IPAH case confirmed by cardiac catheterization (Table 1, case label IPAH929) and lung biopsy and had two variants in CAV1: c.463G>A (V155I) and c.473delC (P158H fsX22) (Fig. 1b). Diagnosed at age 15 months without a family history of PAH, the patient underwent lung biopsy as part of a routine clinical diagnostic evaluation by an outside center due to her young age. While rare in the diagnostic evaluation in adult patients, lung biopsy is more frequently required in the evaluation of the ill child with pulmonary hypertension. That patient’s lung biopsy was notable for medial thickening of the pulmonary arteries with persistent muscularization in the small peripheral arteries (Figure 3). Her healthy parents and sister were genotyped for the c.463G>A (V155I) and c.473delC (P158H fxX22) variants in CAV1. The c.463G>A (V155I) substitution is predicted to be tolerated for protein function, and was identified in her clinically unaffected father. However, the variant c.473delC (P158H fxX22) was de novo, as neither of her parents carried this deletion. Paternity and maternity testing was performed using a standard panel of 8 highly polymorphic microsatellite markers for identity testing and confirmed this to be a de novo variant.

Figure 3.

Lung biopsy from the patient previously diagnosed with IPAH. Medial thickening was in pulmonary arteries, most recognizable in small peripheral pulmonary artery branches (Panel A). Medial thickening was confirmed by immunohistochemistry for alpha smooth muscle actin (Panel B). Alveoli did not show hypercellularity or fibrosis (Panel C), and increase in endothelial cell number was not observed on CD31 immunohistochemistry (Panel D) (Hematoxylin and eosin, Original magnification, (A) × 100 and (C) × 150. DAB chromogen, Original magnification (B) × 100 and (D) × 150).

Given the association with PAH, CAV1 variants c.474delA (P158P fsX22) and c.473delC (P158H fxX22) were genotyped in 1000 ethnically-matched Caucasian, European controls and were not identified in any healthy individuals.

Lung Immunohistochemistry: Reduced Endothelial Cell Caveolin-1

Using an antibody to caveolin-1, we performed immunohistochemical analysis in paraffin embedded lung tissue from controls and one PAH patient with a CAV1 mutation (IPAH929) to determine cellular localization of caveolin-1. In control pulmonary arteries, caveolin-1 was predominantly localized on the endothelial cell surface with some staining in the cytoplasm of the endothelial cells. A striking decrease in localization of caveolin-1 was observed in the patient’s pulmonary arteries when compared to controls (Figure 4), consistent with previous reports.^15–17

Discussion

We used whole exome sequencing to evaluate a large family with autosomal dominant heritable PAH (HPAH), and identified the first example of a genetic mutation associated with human PAH not directly involved in TGF beta signaling. The association of CAV1 mutations with PAH was confirmed in an unrelated PAH patient found to have a de novo CAV1 mutation; while originally classified as having idiopathic PAH (IPAH), this patient thus has a heritable form of PAH as well. These findings demonstrate the power of whole exome sequencing when linked to precise phenotypic data to elucidate the etiology of a Mendelian disease. The association of PAH with CAV1 mutations affirms the importance of intact caveolar function to pulmonary vascular homoeostasis in humans, and may provide a novel target for therapeutic development, as has been previously suggested.¹⁶

Caveolin-1 is the predominant of three proteins (caveolin-1, caveolin-2, and caveolin-3) that coat the flask-like invaginations of the plasma membrane known as caveolae. Caveolin-1 is translated into the endoplasmic reticulum as a full-length protein of 178 amino acids in its α-isoform. The expression of CAV1 is necessary for the formation of caveolae, which are a subtype of specialized microdomains known as lipid rafts ¹⁸. Like other lipid rafts, caveolae are rich in cell surface receptors critical to initiation of a cellular signaling cascade ¹⁹. Receptor signaling cascades relevant to PAH such as the TGF beta superfamily, nitric oxide pathway, and G-protein coupled receptors rely heavily on proper caveolar function ²⁰. In fact, recent studies have demonstrated that caveolin-1 modifies TGF beta signaling at the plasma membrane, which may provide a mechanistic link between CAV1 and BMPR2 mutations in the pathogenesis of PAH ²¹.

The two frameshift mutations we identified are located in exon 3 of CAV1 and result in nearly the same amino acid sequence at the C terminus in a site highly conserved across species (Figure 5a and 5b). This may disrupt an adjacent site of cysteine palmitoylation of the C terminus, which is important to caveolin-1-mediated anchorage of caveolae to the plasma membrane. Disruption of the anchoring process could explain the reduced caveolin-1 staining along the endothelial cells of the small arteries of the lung (Figure 4). Heterozygous somatic mutations in CAV1 that occur adjacent to another palmitoylation site have been described in breast cancer ²². The multiple important roles of the C-terminus in proper caveolin-1 and caveolar function suggest that this mutation disrupts cell signaling.

Figure 5.

Conservation of amino acids of caveolin-1 and mutations identified in PAH patients. (A) Conservation of amino acids of CAV1 among human, mouse, chicken, xenopus and zebrafish are indicated. Variant identified is highlighted at amino acid 155 and mutated region is highlighted starting at amino acid 158. (B) Comparison of normal and mutated amino acids sequences as predicted by open reading frame finder are shown in Panel B. Mutation is highlighted.

The detection of CAV1 mutations in humans with PAH supports previous work in Cav1 knockout mice which have pulmonary hypertension with medial thickening, muscularization of distal pulmonary vessels and loss of total pulmonary artery surface area, as well as thickened alveolar septa, hypercellularity and an increase of extracellular fibrillar deposits ^23–25. Elegant genetic deletion studies with Nos3 suggested that the development of pulmonary vascular disease in the Cav1 null mice occurred due to hyperactive eNOS and subsequent tyrosine nitration–dependent impairment of protein kinase G (PKG) activity, and data in human IPAH supported this finding.¹⁷ Cav1 expression was also reduced in the pulmonary arteries of other established in vivo models of pulmonary hypertension, including the monocrotaline-induced forms of pulmonary hypertension.²⁰ Furthermore, mRNA and caveolin-1 protein expression was reduced in the lung endothelial cells of arterial plexiform lesions from patients with PAH, although protein expression was not clearly diminished in western blots of whole lung tissue extracts from those patients.^{15–17, 26} Our genetic and immunohistochemical findings are consistent with these findings, and further emphasize the important role of caveolin-1 in PAH pathogenesis, perhaps with an emphasis on its role in endothelial cell structure and/or function.

To date, HPAH in families is genetically inherited as an autosomal dominant condition characterized by reduced penetrance regardless of associated gene defect.⁴ BMPR2 is the gene most commonly associated with HPAH, although hereditary hemorrhagic telangiectasia (HHT) due to ALK1 and ENG mutations may associated with HPAH as well. In CAV1-associated HPAH, and in these other forms, it remains unclear why only autosomal dominant transmission with reduced penetrance has been identified. This transmission pattern suggests that additional genetic and/or environmental factors contribute to modify the genetic susceptibility and exceed a threshold necessary for development of clinically-detectable pulmonary vascular disease. Because caveolae are abundant in the plasma membrane of multiple cell types in the lung including endothelial cells, and are receptor-rich regions that include BMPR2 and other TGF beta receptors, it is tempting to speculate that defects in caveolae represent a common mechanism underlying the genetic basis of HPAH.²⁷ This question requires further investigation.

Limitations

While we identified mutations in a gene with firmly established biologic plausibility including a rodent model of PAH and previous implications in human PAH, further studies will be needed to precisely elucidate, and confirm the mechanism by which heterozygous mutations in CAV1 promote PAH in humans. We describe a novel association between a gene (CAV1) and human PAH using next generation sequencing methods, although definitive causation requires further study.

Western blot and lung tissue immunohistochemistry studies suggest insufficient caveolin-1 protein production; however, while insufficiency may be necessary for these subjects it may not be sufficient given the incomplete penetrance of the gene mutation in the family of study. The apparent incomplete penetrance of CAV1 mutations in the subjects evaluated, as with BMPR2-associated PAH, suggests that additional genetic and/or non-genetic modifiers influence the expression of disease. The progressive decline in age at diagnosis from generation II to III to IV (Figure 1), as well as the complete penetrance of disease among the progeny of subjects III-2 and III-3, could support a conclusion that additional modifying factors are present although this has not yet been determined. For example, it is possible that subject III-3 (Figure 1) carries and passed on an additional genetic variant which promoted heightened disease risk for the progeny. Similar to the pedigree demonstrated (Figure 1), BMPR2-associated PAH is associated with incomplete penetrance and variable age at diagnosis within and across families.²⁸ Of note, the pulmonary vascular changes seen in the current Cav1 rodent models of disease occur in null mice, not heterozygotes, although this difference in gene dosage effects across species is common. The mice exhibit alveolar abnormalities as well which were not seen in the patient lung tissue presented. The lack of alveolar changes in humans may suggest that the pulmonary vasculature is more susceptible to a single CAV1 mutation and perhaps the endothelial cell in particular; or other factors may be involved in pathogenesis that either protect the alveoli or increase the risk to the pulmonary vasculature.

Conclusion

In conclusion, we used whole exome sequencing to identify a novel genetic association with PAH, highlighting the utility of this technique when applied to the study of genetic diseases with a well-defined phenotype. The identification of CAV1 gene mutations may have important implications for our overall understanding of PAH pathogenesis, and is consistent with previous studies of caveolin-1 abnormalities in human and murine PAH. This finding highlights the importance of caveolae in the homeostasis of the pulmonary vasculature, which may be a mechanistic link with BMPR2 and the other previously established PAH genes. Further studies of the role of caveolin-1 and caveolae in the maintenance of normal pulmonary vascular biology should yield a clearer understanding of the converging signaling pathways that ultimately lead to PAH development.