Comments
Abstract
Rationale: Idiopathic pulmonary arterial hypertension (IPAH) is usually without an identified genetic cause, despite clinical and molecular similarity to bone morphogenetic protein receptor type 2 mutation–associated heritable pulmonary arterial hypertension (PAH). There is phenotypic heterogeneity in IPAH, with a minority of patients showing long-term improvement with calcium channel–blocker therapy.
Objectives: We sought to identify gene variants (GVs) underlying IPAH and determine whether GVs differ in vasodilator-responsive IPAH (VR-PAH) versus vasodilator-nonresponsive IPAH (VN-PAH).
Methods: We performed whole-exome sequencing (WES) on 36 patients with IPAH: 17 with VR-PAH and 19 with VN-PAH. Wnt pathway differences were explored in human lung fibroblasts.
Measurements and Main Results: We identified 1,369 genes with 1,580 variants unique to IPAH. We used a gene ontology approach to analyze variants and identified overrepresentation of several pathways, including cytoskeletal function and ion binding. By mapping WES data to prior genome-wide association study data, Wnt pathway genes were highlighted. Using the connectivity map to define genetic differences between VR-PAH and VN-PAH, we found enrichment in vascular smooth muscle cell contraction pathways and greater genetic variation in VR-PAH versus VN-PAH. Using human lung fibroblasts, we found increased stimulated Wnt activity in IPAH versus controls.
Conclusions: A pathway-based analysis of WES data in IPAH demonstrated multiple rare GVs that converge on key biological pathways, such as cytoskeletal function and Wnt signaling pathway. Vascular smooth muscle contraction–related genes were enriched in VR-PAH, suggesting a potentially different genetic predisposition for VR-PAH. This pathway-based approach may be applied to next-generation sequencing data in other diseases to uncover the contribution of unexpected or multiple GVs to a phenotype.
Although mutations in the bone morphogenetic protein receptor type 2 gene are known to be associated with heritable forms of pulmonary arterial hypertension and are occasionally found in patients with idiopathic pulmonary arterial hypertension (IPAH), little is known about the potential genetic underpinnings of IPAH. Moreover, there is phenotypic heterogeneity in IPAH, with a minority of patients showing long-term improvement with calcium channel–blocker therapy. We used whole-exome sequencing (WES) to identify gene variants that underlie IPAH and determine whether genetic variants differ in vasodilator-responsive IPAH (VR-PAH) versus nonresponsive IPAH (VN-PAH).
Using a pathway-based analysis of WES data in 36 IPAH samples, we demonstrated multiple rare genetic variants that converge on a few key biological pathways, such as cytoskeletal function and Wnt signaling pathway. Vascular smooth muscle contraction–related genes were enriched in VR-PAH and suggest a potentially different genetic predisposition for VR-PAH versus VN-PAH. Differential Wnt pathway responses to stimulation were detected between IPAH and control lung fibroblasts, suggesting functional differences in this pathway in human IPAH. This pathway-based analytic approach may be applied to other diseases to uncover the role of multiple genes in a disease phenotype.
Pulmonary arterial hypertension (PAH) is a highly mortal disease with limited therapies characterized by progressive vascular obliteration and right heart failure (1). Although the association of rare variants with heritable forms of PAH (HPAH) has contributed substantially to the understanding of PAH pathogenesis, whether rare genetic variants (GVs) contribute to the development of idiopathic PAH (IPAH) is presently understudied (2–4). Understanding the genetic determinants of IPAH may identify underlying disease mechanisms and new therapies. Moreover, there is substantial variability in responses to current PAH-directed therapy across all subtypes of PAH, and genetic predictors of response to therapy are presently unknown (5).
IPAH and HPAH share common pathology and are clinically indistinguishable (2). However, approximately 80% of patients with HPAH have mutations in the gene bone morphogenetic protein receptor type 2 (BMPR2) (6, 7), but only a minority of patients with IPAH have mutations in this receptor (8). Furthermore, in HPAH, pulmonary vascular resistance generally cannot be improved acutely by administration of vasodilators such as nitric oxide (9). However, in IPAH approximately 10% of patients normalize pulmonary artery pressure with administration of vasodilators, and these patients have markedly improved survival (2, 10, 11). We recently showed that patients with this vasodilator-responsive IPAH (VR-PAH) phenotype have different transcriptomic patterns in peripheral blood compared with those without a response (VN-PAH) (12), but whether GVs drive these RNA expression patterns is presently unknown.
Whole-exome sequencing (WES) has been used to identify Mendelian genes in familial cardiopulmonary and other diseases (13–16). More recently this technique has been used to identify de novo GVs associated with pulmonary vascular disease (15, 17–19). In addition, analytic techniques applied to WES data can identify the presence of polygenic disease (20, 21), which is particularly useful when conditions are not clinically known to be heritable. To date, there is a single report of WES in IPAH, which used a candidate gene approach for analysis (3). However, a discovery-based approach has not been used to analyze WES data in this disease, and, additionally, the overlay of WES data on prior genetic studies in IPAH may allow discovery of important, previously unrecognized GVs. Furthermore, techniques such as these may uncover the role of multiple genes or point to new, unexpected variants within a pathway that contribute to disease risk or pathogenesis.
We hypothesized that GVs may predispose to the development of IPAH and that GVs that underlie VR-PAH will differ from VN-PAH. We tested this hypothesis using WES in two cohorts of well-phenotyped patients with IPAH, both VR-PAH and VN-PAH, and further explored an identified pathway in human lung fibroblasts from IPAH and healthy control subjects.
Patients with IPAH were recruited from two centers: Vanderbilt University Center for Pulmonary Vascular Disease and the University of Chicago. All patients provided informed consent (institutional review board #9401 Vanderbilt University Medical Center) and peripheral blood was obtained for DNA. Patients were carefully phenotyped in accordance with published guidelines (2) as outlined in the Methods section and Figure E1 in the online supplement.
WES and genotyping were performed at the Center for Inherited Disease Research Core Facility, Johns Hopkins School of Medicine (Baltimore, MD). Please see the Methods section of the online supplement for further details.
We conducted two phases of sequencing analysis (Figure 1). Details of our variant calling and filtering strategy can be found in the Methods section of the online supplement.
Figure 1. Filtering strategy to identify rare genetic variants in whole-exome sequencing data. A total of 11,981 variants were filtered to 1,580 gene variants. CIDR = Center for Integrated Disease Research; IPF = idiopathic pulmonary fibrosis; PolyPhen2 = Polymorphism Phenotyping version 2; SIFT = Sorting Intolerant from Tolerant.
[More] [Minimize]To assess the function of the genes harboring variants detected from this WES project, we conducted functional enrichment tests using the online tool KOBAS (22). Those gene ontology (GO) terms or pathways were considered significantly enriched if their adjusted P value was <0.05, which was calculated by the hypergeometric test followed by the Benjamini-Hochberg multiple testing correction (23), as implemented in the KOBAS tool.
Drugs included in the connectivity map (cMAP) analysis were either known calcium channel blockers or known to elicit vasodilation in VR-PAH. Details of cMAP analysis can be found in the online supplement.
To form a connected pathway for our potential gene set, we first constructed a nonredundant human pathway–based interactome based on the consecutive metabolic reactions and signaling transduction data from HumanCyc, NCI signaling pathway database, Reactome, and PANTHER pathway databases (downloaded from Pathway Commons database [24] on April 19, 2014 from http://www.pathwaycommons.org/pc2/downloads.html). Further details can be found in the online supplement.
See the Methods section of the online supplement.
See the Methods section of the online supplement.
We selected 36 patients with IPAH for WES. Demographic data are presented in Table E1. Of these, 12 were from University of Chicago and 24 from Vanderbilt University Medical Center; 17 met criteria for VR-PAH and 19 met criteria for VN-PAH.
In total, 215,406 variants were identified in 36 IPAH samples in the WES results. To narrow down the candidate variants, a strict filtering strategy was performed (Figure 1; see online supplement for details). After this initial variant filtering, we identified 1,580 variants in 1,369 genes in our 36 patients (Table E2). Among the detected 1,580 variants, 57 of these were small insertions or deletions, and the remaining 1,523 were single-nucleotide polymorphisms (SNPs). In the VR-PAH group we identified 789 variants in 727 genes, and in the VN-PAH group we identified 797 variants in 722 genes (several variants are present in both groups and thus counted twice).
To obtain a functional overview for our identified mutated genes, we performed GO term enrichment analysis of the 1,369 mutated genes. Using all the human protein-coding genes as background, the statistically overrepresented pathways in descending order of strength of association were as follows: (1) microtubule motor activity, (2) ion binding, (3) carbohydrate derivative binding, (4) motor activity, (5) anion binding, (6) adenyl ribonucleotide binding, (7) small molecule binding, (8) adenyl nucleotide binding, (9) cell adhesion, (10) biological adhesion, (11) extracellular matrix, (12) extracellular matrix part, and (13) ATP binding (Table 1). Although overall these results suggest similar GO term groups as in prior publications on transcriptomic changes in PAH (6), particularly in the cytoskeletal findings, they also point to differences in ion and small molecule binding, on the basis of anionic binding, small molecule binding, ATP binding, and adenyl ribonucleotide binding above, on a genomic level in this PAH population enriched for VR-PAH.
| GO Term | GO ID | P Value | Corrected P Value* |
|---|---|---|---|
| Microtubule motor activity | GO:0003777 | 9.94 × 10−7 | 3.73 × 10−3 |
| Ion binding | GO:0043167 | 1.67 × 10−6 | 3.73 × 10−3 |
| Carbohydrate derivative binding | GO:0097367 | 2.33 × 10−6 | 3.73 × 10−3 |
| Motor activity | GO:0003774 | 2.84 × 10−6 | 3.73 × 10−3 |
| Anion binding | GO:0043168 | 2.90 × 10−6 | 3.73 × 10−3 |
| Adenyl ribonucleotide binding | GO:0032559 | 4.37 × 10−6 | 3.91 × 10−3 |
| Small molecule binding | GO:0036094 | 4.46 × 10−6 | 3.91 × 10−3 |
| Adenyl nucleotide binding | GO:0030554 | 4.56 × 10−6 | 3.91 × 10−3 |
| Cell adhesion | GO:0007155 | 5.48 × 10−6 | 3.95 × 10−3 |
| Biological adhesion | GO:0022610 | 6.03 × 10−6 | 3.95 × 10−3 |
| Extracellular matrix organization | GO:0031012 | 6.14 × 10−6 | 3.95 × 10−3 |
| Extracellular matrix part | GO:0044420 | 7.57 × 10−6 | 4.36 × 10−3 |
| ATP binding | GO:0005524 | 7.89 × 10−6 | 4.36 × 10−3 |
We next sought to explore the intersection of common variants previously identified through genome-wide association studies (GWASs) in IPAH with our own identified rare variants (25). We mapped the 1,580 novel mutations to the genome and compared this map with 318 SNP peaks identified from a prior GWAS in PAH (18), as these peaks represent moderately associated genomic regions in patients with IPAH (Figure 2) (26–29). In total, we found eight regions that are statistically significant (P < 0.01, hypergeometric test), including chr2:32100000–36600000, chr3:79800000–83500000, chr3:87200000–87900000, chr5:66700000–68400000, chr5:139500000–144500000, chr10:12200000–17300000, chr19:0–6900000, and chr17:75300000–81195210. In examining all the regions together, there were 346 WES-identified gene variants that overlapped to 318 GWAS peaks. We again performed a pathway analysis on these 346 genes using REACTOME and found that the five most overrepresented pathways in decreasing order were: (1) WNT5A-dependent internalization of FZD2, FZD5, and ROR2; (2) HSF1-dependent transactivation; (3) attenuation phase; (4) formyl peptide receptors; and (5) PKA activation in glucagon signaling (Table E3).
Figure 2. Circos plot to facilitate the identification and analysis of similarities and differences arising from comparisons of the 36 idiopathic pulmonary arterial hypertension exomes. Variation in genome structure between chromosomes (1–22, X, and Y) is demonstrated in this plot, which allows for demonstration of the frequency of variations according to phenotype. From outside to inside, the outer rim demonstrates the frequency of variants in the vasodilator-nonresponsive pulmonary arterial hypertension (VN-PAH) group according to chromosomal location, and the second-most outer rim demonstrates the frequency of variants in the vasodilator-responsive pulmonary arterial hypertension (VR-PAH) group according to chromosomal location. Next the frequency of genome-wide association study (GWAS) common variations (single-nucleotide polymorphisms) are sandwiched between the abundance of novel VN-PAH variants (outer faint orange circle) and the abundance of novel VR-PAH variants (inner faint orange circle). Relevant identified genes are shown in the innermost circle (see Figure 3).
[More] [Minimize]Although our prior analyses were discovery based, we also sought to identify variants and pathways of interest in a hypothesis-driven approach. On the basis of prior publications suggesting a potential role for the Wnt pathway in PAH (30–32), we chose to focus on this pathway for more detailed study. First, we looked for Wnt variants in our WES variants that correlated to GWAS peaks as shown in Figure 2. Of the 436 genes, 20 were associated with Wnt signaling by REACTOME. In addition, we identified a region in chromosome 5 with a high degree of intersection between GWAS and WES that included a gene cluster with nine protocadherin genes related to Wnt signaling pathway (Figure 3), including PCDHA12, PCDHAC2, PCDHB5, PCDHB6, PCDHB12, PCDHGA6, PCDHGB7, PCDHGA11, and PCDH12.
Figure 3. Region of interest on chromosome 5. Magnification of a region of chromosome 5 on the Circos plot in Figure 2 for closer evaluation of a portion with a high degree of intersection between our whole-exome sequencing data and the published genome-wide association study (GWAS) dataset. This region included a gene cluster with nine protocadherin genes related to the Wnt signaling pathway. PAH = pulmonary arterial hypertension; VN-PAH = vasodilator-nonresponsive pulmonary arterial hypertension; VR-PAH = vasodilator-responsive pulmonary arterial hypertension.
[More] [Minimize]In a second approach, we next sought to narrow the list of candidate variants by examining GVs shared by two or more patients (33, 34). There were 37 variants from 37 genes meeting this criterion (ANKRD6, ANLN, AQPEP, CASP8AP2, CHD7, CRIPAK, CRYBB1, CSRP3, CTBP2, CUBN, DOPEY2, FCRL4, GNLY, GRID1, GTF2F1, GULP1, LRRC66, LTBP4, MROH2B, MTFR1, MUC5B, MYH13, NACC2, PRB3, REV1, RGMA, SNED1, SYTL5, TAS2R46, TPSG1, TTC31, TUBA3E, TYRP1, WIF1, ZAN, ZBTB41, and ZDHHC20). In this smaller list of shared GVs, three Wnt signaling genes were found: CTBP2, MYH13, and WIF1 (hypergeometric test, P = 0.01). Three of the Wnt genes were from regions other than the chromosome 5 variants identified above.
We next sought to define the frequency of Wnt variant genes in the 1,369 genes with identified variants in our cohort. There were 24 genes related to the Wnt signaling pathway detected in 24 patients (Table E4; hypergeometric test, P = 3.05 × 10−5). For example, different variants in MYH13 were present in six patients with IPAH. One variant in MYH13 (chr17:10216635, G->A) is shared between two samples. Taken together, these three separate hypothesis-driven analyses suggest a high degree of genetic variation in Wnt signaling pathway genes in the patients with IPAH studied.
Recent data have suggested that there may be molecular differences between VR-PAH and VN-PAH (12, 35). We thus sought to determine if different GVs or variant patterns underlie these two distinct clinical phenotypes. We tested the hypothesis that cMAP-derived expression patterns of vasodilator drugs could be used to prioritize the candidate variants in VR-PAH. Using our list of vasodilator medications (found in the Methods section of the online supplement), we retrieved the cMAP Instance ID and found 56 expression profiles for 11 drugs. We next overlapped the cMAP top- and bottom-ranked probes with VR-PAH and VN-PAH gene variant lists (Figure 4A) to identify genes found in common (Table E5). In total, we found 26 cMAP up-regulated genes also shared by the VR-PAH group. Although 24 of the 26 genes represent distinct molecular activities, 2 of the 26 genes are involved in vascular smooth muscle contraction (hypergeometric test, P = 3.90 × 10−5). Another four genes related to vascular smooth muscle contraction were found in the gene list only from VR-PAH (hypergeometric test, P = 0.018).
Figure 4. Connectivity map (cMAP) analysis was used to define expression arrays associated with vasodilator drugs and prioritize genetic variants associated with vasodilator-nonresponsive pulmonary arterial hypertension (VN-PAH) and vasodilator-responsive pulmonary arterial hypertension (VR-PAH). (A) The areas of intersection are shown on this Venn diagram. There was overrepresentation of vascular smooth muscle genes in the VR-PAH cohort using this analysis (P < 0.05). (B) Heat map showing frequency of variants in specific pathway and number of affected pathways per patient. Rows represent individual patients and columns are affected pathways. Color key represents number of affected genes. down exp = decreased expression; IPAH = idiopathic pulmonary arterial hypertension; up exp = increased expression.
[More] [Minimize]We next tested the hypothesis that VR-PAH is associated with different pathway variants from VN-PAH and compared the most enriched by GO or cMap analysis or biologically relevant pathways in each phenotype (Table 2). There was overrepresentation of vascular smooth muscle contraction GVs in the VR-PAH cohort compared with the patients with VN-PAH (seven affected patients with VR-PAH with six genes vs. one affected patient with VN-PAH and one gene; Fisher exact test on the number of mutated samples, P = 1.55 × 10−2). Variants were identified in the following genes in VR-PAH: ADCY4, ADCY8, GNAS, PLA2G4E, PPP1R12B, and RAF1. More variants were also identified in cardiac muscle contraction in VR-PAH (four patients, four genes vs. one patient with one gene). These affected genes are ATP1A4, CACNA2D3, RYR2, and UQCR10. In addition, although there were relatively similar numbers of individual variants presented in two or more patients in both VR-PAH and VN-PAH, the number of affected genes was different in each phenotype (Figure 4B, Table E5). There were 33 genes with more than one variant in VR-PAH and only 10 genes with more than one variant in VN-PAH (Fisher exact test on the mutated genes, two-tail P = 5.00 × 10−4). These data suggest a high frequency of GVs within both smooth muscle and cardiac contraction in VR-PAH and, more broadly, that VR-PAH may have a higher frequency of variants common in two or more patients, than does VN-PAH.
| VR-PAH (n = 17) | VN-PAH (n = 19) | |||
|---|---|---|---|---|
| No. of Patients with Variant | No. of Affected Genes | No. of Patients with Variant | No. of Affected Genes | |
| Wnt signaling | 13 | 14 | 11 | 12 |
| Vascular smooth muscle contraction | 7 | 6* | 1 | 1 |
| Mutated variants with 2+ IPAH | 12 | 33 | 10 | 10 |
| Extracellular matrix organization | 10 | 17 | 13 | 16 |
| TGF-β signaling | 5 | 7 | 3 | 3 |
Because we did not identify a single variant that was uniformly associated with IPAH, we hypothesized that there might be pathways or biological networks of affected genes that were central to development of PAH. To test this hypothesis, we performed a network analysis focusing on key pathways or gene ontology groups and high-variant gene clusters identified in our prior analyses (Table 2). We created a central mutation map for IPAH (Figure 5) in which there were 75 nodes and 107 links in total. Four of the five categories were fully connected. Only vascular smooth muscle contraction was disconnected from the other functional groups. This fully connected map showed the relevance of the Wnt signaling pathway, which potentially connected to the transforming growth factor (TGF)-β signaling pathway (of which BMPR2 is a member) via four steps starting at WIF1, CDC25C, SRC, PLA2G4A, to TGF-β1. Although most of the 37 mutated genes (shared in at least two samples) are not linked to known biological processes, they may partly bridge the signaling events among Wnt signaling pathway, extracellular matrix, and TGF-β signaling pathway.
Figure 5. Central mutation map for idiopathic pulmonary arterial hypertension (IPAH). We performed a network analysis focusing on key pathways or gene ontology groups and high-variant gene clusters. The central mutation map for IPAH shows 75 nodes and 107 total links. Implicated linker genes were not identified in the variant analysis but are required to connect other identified gene variants in this analysis. At a minimum, four steps linked the candidate pathways of Wnt and transforming growth factor (TGF)-β; starting from WIF1 to CDC25C to SRC to PLA2G4A to TGF-β1.
[More] [Minimize]To confirm the accuracy of our WES data, we selected GVs for validation using Sanger sequencing (Table 3). Rationale for selection of the variants and their frequency is in the online supplement (Table E6).
| Gene Name | Gene ID | Variant Locus | Sequencing Results |
|---|---|---|---|
| ADCY8 | 114 | chr8:131826409:131826409:A:G | Validated |
| COL15A1 | 1306 | chr9:101748162:101748162:C:T | Validated |
| TGFB1 | 7040 | chr19:41850733:41850733:G:A | Validated |
| LTBP4 | 8425 | chr19:41119074:41119074:G:T | Failed |
| CTBP2 | 1488 | chr10:126727616:126727616:A:- | Failed |
| MYH13 | 8735 | chr17:10216635:10216635:G:A | Validated |
| MTFR1 | 9650 | chr8:66620310:66620310:C:G | Validated |
| ANKRD6 | 22881 | chr6:90337349:90337349:G:T | Validated |
| DNAH1 | 25981 | chr3:52380743:52380743:G:A | Validated |
| FER1L5 | 90342 | chr2:97335910:97335910:G:C | Validated |
| RGMA | 56963 | chr15:93616963:93616963:C:T | Failed |
| ZDHHC20 | 253832 | chr13:21976996:21976996:T:C | Validated |
| CDH23 | 64072 | chr10:73571088:73571088:G:A | Validated |
| chr10:73563154:73563154:G:C | Validated | ||
| chr10:73490244:73490244:G:T | Validated | ||
| chr10:73468866:73468866:G:T | Validated | ||
| chr10:73406479:73406479:G:C | Validated | ||
| PIM1 | 5292 | chr6:37138142:37138142:G:T | Failed |
| CRCP | 27297 | chr7:65579834:65579834:G:T | Validated |
| HLA-DRB1 | 3123 | chr6:32549501:32549501:A:T | Failed |
| WNT5A | 7474 | chr3:55513419:55513419:T:A | Validated |
| WIF1 | 11197 | chr12:65445142:65445142:T:C | Validated |
| chr12:65445142:65445142:T:C | Validated |
We further sought to determine if any of the variant pathways identified through WES have functional consequences using skin fibroblasts cultured from patients with IPAH (32). Gene ontology analysis was performed via Webgestalt (36) using genes with altered regulation in skin fibroblasts in patients with IPAH compared with control subjects. The most highly significant gene ontology groups are listed in Table E7, and they include cell adhesion and developmental genes, including Wnt pathway members, similar to the patterns found in our WES analysis. Multiple Wnt-related genes have altered expression in sporadic PAH-derived skin fibroblasts, including DKK2, FZD7, SFRP2, WIF1, WNT2, and WNT5A, among others. These data demonstrate transcriptional patterns in human IPAH cells that reflect the pattern analysis of our WES data, suggesting biological function of the identified variant pathways.
To further validate the microarray and gene variant findings in the Wnt pathway, we undertook studies of lung fibroblasts isolated from patients with IPAH at the time of lung transplantation or healthy control failed lung donors (see Table E8 for demographic information). Fibroblasts have previously been shown to be important in PAH, are from a tissue of relevance in PAH, and can be isolated from patients with IPAH (37, 38). We first tested the hypothesis that GVs in Wnt genes may result in altered expression of Wnt genes. In our IPAH fibroblast lines, we found significantly increased expression of SFRP1 and SFRP2, Wnt antagonists, and increased expression of WISP1, a downstream target of Wnt pathway activation (P < 0.05 for all; Figure 6). Examining protein expression, we found that Wisp1 trended to be increased in IPAH compared with control fibroblasts but did not reach statistical significance (P = 0.05). Finally, we tested the hypothesis that Wnt pathway dysregulation via GVs could be detected in a functional assay using our previously described Wnt activity assay (32). We first measured baseline Wnt activity in our lung fibroblast lines indirectly via TCF/LEF luciferase activity and normalized to Renilla luciferase activity after transfection. We also stimulated our control and IPAH cell lines using Wnt3a and LiCl. We found no baseline differences in Wnt activity but significantly increased activity of this pathway when stimulated with LiCl. Our data suggest that there is broad dysregulation of the Wnt pathway in IPAH compared with control subjects and differential responses to stimulation of the Wnt pathway in IPAH compared with control subjects, suggesting a potential functional role for GVs in this pathway in IPAH pathogenesis.
Figure 6. Analysis of Wnt pathway in human lung fibroblasts. Lung fibroblasts were isolated from human patients with idiopathic pulmonary arterial hypertension and healthy, failed donor control subjects and used for these studies (n = 3–4 each). (A) Polymerase chain reaction of cell homogenates for SFRP1, SFRP2 (Wnt antagonists), and AXIN2 (canonical Wnt signaling) show increased expression of Wnt antagonists and the Wnt target gene, WISP1. (B) Western analysis for Wisp1 demonstrated a trend to increased Wisp1. (C) Wnt pathway activity was assessed using a dual luciferase assay at baseline and after 72-hour stimulation with Wnt3a and LiCl. Idiopathic pulmonary arterial hypertension cells demonstrated significantly greater stimulation of Wnt activity after LiCl stimulation. *P < 0.05. HPRT = hypoxanthine phosphoribosyltransferase; LiCl = lithium chloride; PAH = pulmonary arterial hypertension.
[More] [Minimize]We analyzed the genomic DNA of 36 subjects with IPAH using WES and enriched our cohort for the subphenotype of VR-PAH to identify genetic signatures of IPAH broadly and determine whether genetic differences underlie clinical observations in VR-PAH compared with VN-PAH. Although no single rare variant was detected to explain all IPAH cases, multiple analytic approaches consistently highlighted the association of variants in several key pathways with IPAH of all types. In particular, variation in the Wnt pathway was highly associated with IPAH. To demonstrate the capacity of our approach to understanding gene variant data, we performed functional studies of this pathway in human lung fibroblasts and were able to demonstrate differences in Wnt pathway stimulation in IPAH compared with control subjects. This validation enhances confidence in the analytic approach to identify other, potentially important and perhaps unexpected pathways or GVs. Furthermore, consistent with our hypothesis, we identified differences in variant frequency and affected pathways in the VR-PAH and VN-PAH, including enrichment of pathways relevant to pulmonary vascular smooth muscle contraction in the VR-PAH subphenotype. Our data support the concept that IPAH may be a disease with a genetic basis, or at least a genetic susceptibility; however, multiple variants within biologically important pathways may be required for the development of the IPAH phenotype.
Next-generation sequencing techniques such as WES are well known to identify hundreds of rare GVs in healthy individuals, complicating the identification of variants that directly contribute to the disease of interest (39). This challenge may be overcome through the use of an appropriate filtering strategy that includes potentially important rare variants but appropriately excludes variants that are unrelated to disease (40). Our project used two techniques to address this issue: first, we reduced variability by performing WES on a moderate number of well-phenotyped patients with IPAH, considering the frequency of the disease. Second, we used a filtering algorithm that used stringent criteria based on prediction of disease-causing variants. In particular we excluded GVs that have been identified in idiopathic pulmonary fibrosis (41), under the hypothesis that the two diseases have different molecular etiology. This strategy is grounded in basic science and clinical data, yet there remains a risk that pathogenic variants may be removed from the dataset and thus missed. Our strategy involved filtering common variants, which allowed focus but does inherently risk losing potentially important common SNPs. In addition, as with all WES, the data must be interpreted with some caution, as WES only evaluates coding regions; thus, potentially important variants in the noncoding genome will be excluded, epigenetic modifications are not assessed, and, moreover, not all variants identified were validated, although our validation rates are within published norms (42).
Our initial filtering strategy did not show a single genetic variant, although we and other groups have identified single variants in other forms of heritable pulmonary vascular disease using advanced genetic analytic techniques (15, 18, 19, 43). In this context, given the possibility that multiple genes are required to develop some disease phenotypes, we hypothesize that IPAH may develop due to either (1) multiple GVs in certain key pathways, or (2) multiple affected key pathways within an individual patient. Using a gene ontology analysis, we were able to identify several pathways that were affected by GVs at a higher degree than expected in the general population. Several of these pathways could be combined broadly into cytoskeletal genes or ion-binding genes. Prior transcriptomic analyses in heritable PAH and IPAH and in other reports have pointed to cytoskeletal dysfunction in PAH (6, 44, 45). Similarly, several animal models have found ion-channel dysfunction in pulmonary vascular disease (46–48), but the identification of these pathways as a potential genetic predisposition to IPAH is new. In addition, as our patient population was enriched in VR-PAH, the ion-binding genes were particularly promising as being important to this subset in particular.
Although just one of the many pathways highlighted by our variant analysis, we chose to focus on the Wnt pathway for further study as an example of further narrowing and understanding data from WES. Wnt pathway GVs were overrepresented within several chromosomal regions. Further strengthening this association are Wnt variants present in multiple patients. Finally, to strengthen the gene variant findings, we tested the hypothesis that this pathway was of relevance to the IPAH (37, 38). Using human lung fibroblast lines, we found evidence of altered expression of Wnt pathway–modulating genes; borderline increased expression of Wisp1, a Wnt target regulating proliferation; and significantly increased Wnt activity in response to stimulation by LiCl. These functional studies in human cell lines from lung fibroblasts suggest functional alteration in a pathway identified through our gene variant analysis. In the future, transgenic rodent models querying all the identified variants may help to lend further specificity as to which variants in particular are important in PAH pathogenesis, and our analytic techniques to next-generation sequencing data may facilitate confident identification of new GVs and pathways involved in pathogenesis in IPAH or other diseases. Taken together, these analyses suggest a potential role for this pathway as a basis for a genetic predisposition in IPAH and are congruent with other reports demonstrating a role for Wnt signaling in pulmonary vascular remodeling (30, 32).
Our population was intentionally enriched in patients with VR-PAH, as usually this subphenotype accounts for less than 10% of patients with IPAH. We have shown differences in blood transcriptome in VR-PAH compared with VN-PAH, but whether GVs are associated with these different phenotypes was previously unknown. We used the cMAP as a tool to point to important pathways in VR-PAH compared with VN-PAH and identified two important features of VR-PAH, acknowledging the limitation that this tool is not specific to lung expression. First, vascular smooth muscle contraction genes were more enriched in the patients with VR-PAH than in those with VN-PAH. This may explain why many patients with VR-PAH reduce pulmonary arterial pressure when treated with several vasodilators with different mechanisms of action, as the defect may be a shared gene regulating smooth muscle cell contraction. In addition, the number of variants found in pathways per patient was higher in VR-PAH than in VN-PAH, suggesting there may be either modification of important pathways in VN-PAH that result in the VR-PAH phenotype or, perhaps, a completely different genetic predisposition for VR-PAH than VN-PAH.
Our study has limitations including a relatively small patient population. Although our analysis did not identify a monogenic variant underlying all IPAH, the gene ontology approach was informative, showing that patients with IPAH have a high degree of GVs within certain pathways that have biological plausibility as contributing to PAH. In addition, although the analysis of common variations in small numbers of subjects is prone to false discovery, the overlay of WES data with common variant information previously published allowed for the reduction in genetic pathways of interest to focus our discoveries. The frequency of multiple variants in each pathway in individual patients suggests a “dose effect,” such that multiple variants within at least one pathway may be required to develop the IPAH phenotype. This observation suggests that IPAH may be a polygenic disease and reinforces the role of these key pathways in the molecular etiology of IPAH. Given the limits of WES for detection of mitochondrial genomic variants (49) and its inability to detect nonexomic variants or epigenetic (50) or downstream mediators, these data do not exclude potentially important roles for these factors in PAH. The functional consequence of many variants identified in these analyses will need to be tested in animal models and in cell culture including pulmonary arterial and endothelial cells and also validated in other humans with IPAH. Finally, our samples did not all have matching expression data, which prevented quantitative trait locus analysis and analysis of genes previously associated with RNA expression differences in VR-PAH.
In conclusion, our systematic WES screening of GVs in 36 IPAH samples revealed that rare GVs may converge on a few key biological pathways, such as cytoskeletal function and Wnt signaling. Some of the mutated genes related to vascular smooth muscle contraction suggest a potentially different genetic predisposition for patients with VR-PAH and VN-PAH. The role of key gene nodes associated with IPAH needs to be confirmed in the future using patients with IPAH and preclinical models, but this pathway-based analytic approach that we have validated with functional studies may also be informative for other diseases to confidently uncover the role of multiple genes or unexpected GVs that may contribute to disease phenotype.
| 1. | Hoeper MM, Bogaard HJ, Condliffe R, Frantz R, Khanna D, Kurzyna M, Langleben D, Manes A, Satoh T, Torres F, et al. Definitions and diagnosis of pulmonary hypertension. J Am Coll Cardiol 2013;62:D42–D50. |
| 2. | Simonneau G, Gatzoulis MA, Adatia I, Celermajer D, Denton C, Ghofrani A, Gomez Sanchez MA, Krishna Kumar R, Landzberg M, Machado RF, et al. Updated clinical classification of pulmonary hypertension. J Am Coll Cardiol 2013;62:D34–D41. |
| 3. | de Jesus Perez VA, Yuan K, Lyuksyutova MA, Dewey F, Orcholski ME, Shuffle EM, Mathur M, Yancy L Jr, Rojas V, Li CG, et al. Whole-exome sequencing reveals TopBP1 as a novel gene in idiopathic pulmonary arterial hypertension. Am J Respir Crit Care Med 2014;189:1260–1272. |
| 4. | Drake KM, Dunmore BJ, McNelly LN, Morrell NW, Aldred MA. Correction of nonsense BMPR2 and SMAD9 mutations by ataluren in pulmonary arterial hypertension. Am J Respir Cell Mol Biol 2013;49:403–409. |
| 5. | Galiè N, Corris PA, Frost A, Girgis RE, Granton J, Jing ZC, Klepetko W, McGoon MD, McLaughlin VV, Preston IR, et al. Updated treatment algorithm of pulmonary arterial hypertension. J Am Coll Cardiol 2013;62:D60–D72. |
| 6. | Austin ED, Menon S, Hemnes AR, Robinson LR, Talati M, Fox KL, Cogan JD, Hamid R, Hedges LK, Robbins I, et al. Idiopathic and heritable PAH perturb common molecular pathways, correlated with increased MSX1 expression. Pulm Circ 2011;1:389–398. |
| 7. | Newman JH, Wheeler L, Lane KB, Loyd E, Gaddipati R, Phillips JA III, Loyd JE. Mutation in the gene for bone morphogenetic protein receptor II as a cause of primary pulmonary hypertension in a large kindred. N Engl J Med 2001;345:319–324. |
| 8. | Aldred MA, Vijayakrishnan J, James V, Soubrier F, Gomez-Sanchez MA, Martensson G, Galie N, Manes A, Corris P, Simonneau G, et al. BMPR2 gene rearrangements account for a significant proportion of mutations in familial and idiopathic pulmonary arterial hypertension. Hum Mutat 2006;27:212–213. |
| 9. | Elliott CG, Glissmeyer EW, Havlena GT, Carlquist J, McKinney JT, Rich S, McGoon MD, Scholand MB, Kim M, Jensen RL, et al. Relationship of BMPR2 mutations to vasoreactivity in pulmonary arterial hypertension. Circulation 2006;113:2509–2515. |
| 10. | Sitbon O, Humbert M, Jaïs X, Ioos V, Hamid AM, Provencher S, Garcia G, Parent F, Hervé P, Simonneau G. Long-term response to calcium channel blockers in idiopathic pulmonary arterial hypertension. Circulation 2005;111:3105–3111. |
| 11. | Rich S, Brundage BH. High-dose calcium channel-blocking therapy for primary pulmonary hypertension: evidence for long-term reduction in pulmonary arterial pressure and regression of right ventricular hypertrophy. Circulation 1987;76:135–141. |
| 12. | Hemnes AR, Trammell AW, Archer SL, Rich S, Yu C, Nian H, Penner N, Funke M, Wheeler L, Robbins IM, et al. Peripheral blood signature of vasodilator-responsive pulmonary arterial hypertension. Circulation 2015;131:401–409. [Discussion, p. 409.] |
| 13. | Austin ED, Ma L, LeDuc C, Berman Rosenzweig E, Borczuk A, Phillips JA III, Palomero T, Sumazin P, Kim HR, Talati MH, et al. Whole exome sequencing to identify a novel gene (caveolin-1) associated with human pulmonary arterial hypertension. Circ Cardiovasc Genet 2012;5:336–343. |
| 14. | Norton N, Li D, Rieder MJ, Siegfried JD, Rampersaud E, Züchner S, Mangos S, Gonzalez-Quintana J, Wang L, McGee S, et al. Genome-wide studies of copy number variation and exome sequencing identify rare variants in BAG3 as a cause of dilated cardiomyopathy. Am J Hum Genet 2011;88:273–282. |
| 15. | Best DH, Sumner KL, Austin ED, Chung WK, Brown LM, Borczuk AC, Rosenzweig EB, Bayrak-Toydemir P, Mao R, Cahill BC, et al. EIF2AK4 mutations in pulmonary capillary hemangiomatosis. Chest 2014;145:231–236. |
| 16. | Eyries M, Montani D, Girerd B, Perret C, Leroy A, Lonjou C, Chelghoum N, Coulet F, Bonnet D, Dorfmüller P, et al. EIF2AK4 mutations cause pulmonary veno-occlusive disease, a recessive form of pulmonary hypertension. Nat Genet 2014;46:65–69. |
| 17. | Sanders SJ, Murtha MT, Gupta AR, Murdoch JD, Raubeson MJ, Willsey AJ, Ercan-Sencicek AG, DiLullo NM, Parikshak NN, Stein JL, et al. De novo mutations revealed by whole-exome sequencing are strongly associated with autism. Nature 2012;485:237–241. |
| 18. | Germain M, Eyries M, Montani D, Poirier O, Girerd B, Dorfmüller P, Coulet F, Nadaud S, Maugenre S, Guignabert C, et al. Genome-wide association analysis identifies a susceptibility locus for pulmonary arterial hypertension. Nat Genet 2013;45:518–521. |
| 19. | Ma L, Roman-Campos D, Austin ED, Eyries M, Sampson KS, Soubrier F, Germain M, Trégouët DA, Borczuk A, Rosenzweig EB, et al. A novel channelopathy in pulmonary arterial hypertension. N Engl J Med 2013;369:351–361. |
| 20. | Klei L, Sanders SJ, Murtha MT, Hus V, Lowe JK, Willsey AJ, Moreno-De-Luca D, Yu TW, Fombonne E, Geschwind D, et al. Common genetic variants, acting additively, are a major source of risk for autism. Mol Autism 2012;3:9. |
| 21. | Kiezun A, Garimella K, Do R, Stitziel NO, Neale BM, McLaren PJ, Gupta N, Sklar P, Sullivan PF, Moran JL, et al. Exome sequencing and the genetic basis of complex traits. Nat Genet 2012;44:623–630. |
| 22. | Xie C, Mao X, Huang J, Ding Y, Wu J, Dong S, Kong L, Gao G, Li CY, Wei L. KOBAS 2.0: a web server for annotation and identification of enriched pathways and diseases. Nucleic Acids Res 2011;39:W316–W322. |
| 23. | Benjamini Y, Hochberg Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J R Stat Soc, B 1995;57:289–300. |
| 24. | Blinov ML, Schaff JC, Ruebenacker O, Wei X, Vasilescu D, Gao F, Morgan F, Ye L, Lakshminarayana A, Moraru II, et al. Pathway Commons at virtual cell: use of pathway data for mathematical modeling. Bioinformatics 2014;30:292–294. |
| 25. | Cirulli ET, Goldstein DB. Uncovering the roles of rare variants in common disease through whole-genome sequencing. Nat Rev Genet 2010;11:415–425. |
| 26. | Sun L, Dimitromanolakis A, Faye LL, Paterson AD, Waggott D, Bull SB; DCCT/EDIC Research Group. BR-squared: a practical solution to the winner’s curse in genome-wide scans. Hum Genet 2011;129:545–552. |
| 27. | Faye LL, Sun L, Dimitromanolakis A, Bull SB. A flexible genome-wide bootstrap method that accounts for ranking and threshold-selection bias in GWAS interpretation and replication study design. Stat Med 2011;30:1898–1912. |
| 28. | Bowden J, Dudbridge F. Unbiased estimation of odds ratios: combining genomewide association scans with replication studies. Genet Epidemiol 2009;33:406–418. |
| 29. | Ghosh A, Zou F, Wright FA. Estimating odds ratios in genome scans: an approximate conditional likelihood approach. Am J Hum Genet 2008;82:1064–1074. |
| 30. | de Jesus Perez VA, Alastalo TP, Wu JC, Axelrod JD, Cooke JP, Amieva M, Rabinovitch M. Bone morphogenetic protein 2 induces pulmonary angiogenesis via Wnt-beta-catenin and Wnt-RhoA-Rac1 pathways. J Cell Biol 2009;184:83–99. |
| 31. | Chow K, Fessel JP, Kaoriihida-Stansbury , Schmidt EP, Gaskill C, Alvarez D, Graham B, Harrison DG, Wagner DH Jr, Nozik-Grayck E, et al. Dysfunctional resident lung mesenchymal stem cells contribute to pulmonary microvascular remodeling. Pulm Circ 2013;3:31–49. |
| 32. | West JD, Austin ED, Gaskill C, Marriott S, Baskir R, Bilousova G, Jean JC, Hemnes AR, Menon S, Bloodworth NC, et al. Identification of a common Wnt-associated genetic signature across multiple cell types in pulmonary arterial hypertension. Am J Physiol Cell Physiol 2014;307:C415–C430. |
| 33. | Ng SB, Bigham AW, Buckingham KJ, Hannibal MC, McMillin MJ, Gildersleeve HI, Beck AE, Tabor HK, Cooper GM, Mefford HC, et al. Exome sequencing identifies MLL2 mutations as a cause of Kabuki syndrome. Nat Genet 2010;42:790–793. |
| 34. | Ng SB, Buckingham KJ, Lee C, Bigham AW, Tabor HK, Dent KM, Huff CD, Shannon PT, Jabs EW, Nickerson DA, et al. Exome sequencing identifies the cause of a mendelian disorder. Nat Genet 2010;42:30–35. |
| 35. | Langleben D, Orfanos SE, Giovinazzo M, Schlesinger RD, Hirsch AM, Blenkhorn F, Lesenko L, Armaganidis A, Catravas JD. Acute vasodilator responsiveness and microvascular recruitment in idiopathic pulmonary arterial hypertension. Ann Intern Med 2015;162:154–156. |
| 36. | Zhang B, Kirov S, Snoddy J. WebGestalt: an integrated system for exploring gene sets in various biological contexts. Nucleic Acids Res 2005;33:W741-8. |
| 37. | Das M, Dempsey EC, Reeves JT, Stenmark KR. Selective expansion of fibroblast subpopulations from pulmonary artery adventitia in response to hypoxia. Am J Physiol Lung Cell Mol Physiol 2002;282:L976–L986. |
| 38. | Plecitá-Hlavatá L, Tauber J, Li M, Zhang H, Flockton AR, Pullamsetti SS, Chelladurai P, D’alessandro A, El Kasmi KC, Ježek P, et al. Constitutive reprogramming of fibroblast mitochondrial metabolism in pulmonary hypertension. Am J Respir Cell Mol Biol [online ahead of print] 23 Dec 2015; DOI: 10.1165/rcmb.2015-0142OC. |
| 39. | Abecasis GR, Altshuler D, Auton A, Brooks LD, Durbin RM, Gibbs RA, Hurles ME, McVean GA; 1000 Genomes Project Consortium. A map of human genome variation from population-scale sequencing. Nature 2010;467:1061–1073. |
| 40. | DePristo MA, Banks E, Poplin R, Garimella KV, Maguire JR, Hartl C, Philippakis AA, del Angel G, Rivas MA, Hanna M, et al. A framework for variation discovery and genotyping using next-generation DNA sequencing data. Nat Genet 2011;43:491–498. |
| 41. | Cogan JD, Kropski JA, Zhao M, Mitchell DB, Rives L, Markin C, Garnett ET, Montgomery KH, Mason WR, McKean DF, et al. Rare variants in RTEL1 are associated with familial interstitial pneumonia. Am J Respir Crit Care Med 2015;191:646–655. |
| 42. | Lam HY, Clark MJ, Chen R, Chen R, Natsoulis G, O’Huallachain M, Dewey FE, Habegger L, Ashley EA, Gerstein MB, et al. Performance comparison of whole-genome sequencing platforms. Nat Biotechnol 2012;30:78–82. |
| 43. | Newman JH, Holt TN, Cogan JD, Womack B, Phillips JA III, Li C, Kendall Z, Stenmark KR, Thomas MG, Brown RD, et al. Increased prevalence of EPAS1 variant in cattle with high-altitude pulmonary hypertension. Nat Commun 2015;6:6863. |
| 44. | Johnson JA, Hemnes AR, Perrien DS, Schuster M, Robinson LJ, Gladson S, Loibner H, Bai S, Blackwell TR, Tada Y, et al. Cytoskeletal defects in Bmpr2-associated pulmonary arterial hypertension. Am J Physiol Lung Cell Mol Physiol 2012;302:L474–L484. |
| 45. | Wojciak-Stothard B, Tsang LY, Paleolog E, Hall SM, Haworth SG. Rac1 and RhoA as regulators of endothelial phenotype and barrier function in hypoxia-induced neonatal pulmonary hypertension. Am J Physiol Lung Cell Mol Physiol 2006;290:L1173–L1182. |
| 46. | Yamamura A, Yamamura H, Guo Q, Zimnicka AM, Wan J, Ko EA, Smith KA, Pohl NM, Song S, Zeifman A, et al. Dihydropyridine Ca(2+) channel blockers increase cytosolic [Ca(2+)] by activating Ca(2+)-sensing receptors in pulmonary arterial smooth muscle cells. Circ Res 2013;112:640–650. |
| 47. | Guo Q, Huang JA, Yamamura A, Yamamura H, Zimnicka AM, Fernandez R, Yuan JX. Inhibition of the Ca(2+)-sensing receptor rescues pulmonary hypertension in rats and mice. Hypertens Res 2014;37:116–124. |
| 48. | Archer SL, Souil E, Dinh-Xuan AT, Schremmer B, Mercier JC, El Yaagoubi A, Nguyen-Huu L, Reeve HL, Hampl V. Molecular identification of the role of voltage-gated K+ channels, Kv1.5 and Kv2.1, in hypoxic pulmonary vasoconstriction and control of resting membrane potential in rat pulmonary artery myocytes. J Clin Invest 1998;101:2319–2330. |
| 49. | Picardi E, Pesole G. Mitochondrial genomes gleaned from human whole-exome sequencing. Nat Methods 2012;9:523–524. |
| 50. | Archer SL, Marsboom G, Kim GH, Zhang HJ, Toth PT, Svensson EC, Dyck JR, Gomberg-Maitland M, Thébaud B, Husain AN, et al. Epigenetic attenuation of mitochondrial superoxide dismutase 2 in pulmonary arterial hypertension: a basis for excessive cell proliferation and a new therapeutic target. Circulation 2010;121:2661–2671. |
* Co–first authors.
Supported by Clinical and Translational Science Award UL1TR000445 from the National Center for Advancing Translational Sciences; American Thoracic Society/Pulmonary Hypertension Association Research grant (E.D.A.); NHLBI grant 1 PO1 HL 108800 (E.D.A., J.E.L., A.R.H., J.H.N., I.M.R., and J.W.); NHLBI grant 5 PO1 HL 092870-05 (J.E.L., M.Z., Z.Z., and J.C.); and Canada Research Chair in Mitochondrial Dynamics, William J. Henderson Foundation, and National Institutes of Health grants RO1-HL071115 and 1RC1HL099462-01 (S.L.A.).
Contents are solely the responsibility of the authors and do not necessarily represent official views of the National Center for Advancing Translational Sciences or the National Institutes of Health.
Author Contributions: A.R.H., M.Z., J.E.L., Z.Z., E.D.A., and S.M.M. designed the study, analyzed the data, and wrote the manuscript. J.W., J.H.N., S.R., S.L.A., J.E.L., and Z.Z. designed the study, analyzed the data, and provided critical edits of the manuscript. C.J. and C.G. performed the experiments and analyzed the data. J.C. and T.S.B. analyzed the data and provided critical edits of the manuscript. I.M.R. and J.A.K. recruited patients, analyzed the data, and provided critical edits of the manuscript.
This article has an online supplement, which is accessible from this issue’s table of contents at www.atsjournals.org
Originally Published in Press as DOI: 10.1164/rccm.201508-1678OC on February 29, 2016
Author disclosures are available with the text of this article at www.atsjournals.org.
