Variation in SNX29 and Acute Vasodilator Response in Pulmonary Arterial Hypertension.
May 18, 2026DOI: 10.1161/circresaha.125.327117Journal ArticleAHA
Written by: Jason H KarnesChanglei BaoShuxin LiangTimothy ThayerJohn ZagorskiSamisubbu R NaiduJames D WestTae-Hwi Schwantes-AnEhsan KhajoueiKasturi BanerjeeHemant K TiwariJuvie FarolKiana L MartinezDingyuan LiuYao NingYanan SunYangfan JiaJian WangAmit AroraKen BataiStephen J HallidayKatie LutzAnna WalsworthAndrea L FrumpOlga RafikovaZhiyu DaiJoe G N GarciaIl-Man KimAnna R HemnesEvan BrittainRaymond L BenzaWilliam C NicholsAiai ChuHaiyang TangAnkit A Desai
Pulmonary arterial hypertension (PAH) is an uncommon but devastating disorder characterized by progressive vascular remodeling leading to increased resistance and right heart failure 1. Despite a low prevalence of PAH 2,3, PAH is considered fatal with a 40-60% 5-year survival rate when untreated 4,5. Even with the use of modern treatments, the annual mortality rate for idiopathic PAH remains around 10% 6. Due to lack of availability of clinical tools that distinguish responders from non-responders, treatment response to PAH drug therapy is characterized by substantial variability 7. The substantial variation in treatment response between patients underscores the inadequately characterized heterogeneity in PAH etiology and the need to study the underlying pathology that differentiates responders and non-responders.A notable source of treatment response heterogeneity is vasodilator-responsive PAH (VR-PAH), which characterizes up to 5-10% of PAH patients 8. This subset stratifies patients who qualify for therapy with calcium (Ca+) channel blockers in idiopathic, heritable and drug-induced PAH 8–10. In other disease-associated Group 1 PAH cases broadly, patients with VR-PAH also display markedly improved functional status and/or survival 11–14. VR-PAH patients are currently diagnosed using vasodilator testing during right heart catheterization (RHC), defined by a significant and acute drop in mean PA pressures (mPAP) with a pulmonary vasodilator which typically initiates Ca2+ channel blockers. With repeat testing over 6 months to a year, a sustained decrease in mPAP with a pulmonary vasodilator identifies VR-PAH subjects who are eligible for long-term calcium channel blocker treatment. This diagnostic pathway requires serial invasive catheterizations and biomarkers are not available to differentiate VR-PAH from non-vasoresponsive PAH (NR-PAH) patients 15. Elucidating genetic mechanisms of acute or sustained VR-PAH may generate insights into the pathophysiology of PAH and of the therapeutic benefits of calcium channel blockers in PAH.Although several studies have identified genomic influences on PAH using genome-wide association study (GWAS) and sequencing approaches 16–18, less data is available on the genomic determinants of VR-PAH 19. The rarity of this phenotype, being a subset of an already uncommon disease, and the additional stringent criteria of evidence for sustainable vasodilator responsiveness (e.g., repeat invasive vasodilator testing over 6 months to 1 year) have precluded a full-scale GWAS of this phenotype or studies on genetic mechanisms. To fill this gap, the objectives of this study were to: 1) compile all available sequenced cohorts nationally; 2) identify genetic predictors of acute vasodilator response in PAH, the first criteria for VR-PAH, using a GWAS approach; and 3) functionally validate replicated associations.
MATERIALS AND METHODS
All clinical studies were IRB approved at participating institutions and were conducted in accordance with the Declaration of Helsinki. Written informed consent was obtained from all patients. Detailed descriptions of cohorts, genotyping, quality control, statistical analyses, and experimental methodology are available in the Supplemental Materials.
For the discovery cohort, participants represented all adult cases (age greater than or equal to 18 years) defined as World Symposium of Pulmonary Hypertension Group 1 PAH and were extracted from the National Biological Sample and Data Repository for Pulmonary Arterial Hypertension (PAH Biobank). Eligible cases were prospectively recruited from PH centers across North America including biobanking between October 2012 to February 2018. Baseline demographic, clinical, and hemodynamic data were extracted from available clinical data as previously described 20. Testing for VR-PAH was clinically indicated and performed during RHC as previously described 7. Differences between hemodynamics at rest (e.g. baseline mPAP) and after administration without major changes in cardiac output were determined to define acute vasodilator drug response as a continuous measure.
The replication cohort consisted of combined patient data from BioVU, the Vanderbilt DNA databank that links DNA extracted from discarded blood samples to de-identified electronic medical records, and the Allegheny Health Network (AHN) Cohort 21,22. In BioVU and AHN, testing for VR-PAH was done similar to the PAH Biobank cohort and consistent definitions were used for the difference between hemodynamics at rest and after vasodilator administration 7. Analysis was restricted to Group 1 PAH in both cohorts. In BioVU, detailed pre- and post -vasodilator response data was collected for RHC-confirmed PAH patients as described 21,22. Patients with RHC hemodynamic variables consistent with PAH were electronically screened as described. Demographic data were extracted from the clinic visit closest to the date of the RHC. Details of the AHN cohort have been described 23–25.
In the discovery cohort, differences in characteristics at rest and after vasodilator testing were determined using Wilcoxon Signed Rank tests. The primary analysis analyzed the effect of genetic variation on change in mPAP. Differences between hemodynamics at rest (baseline mPAP) and after vasodilator administration without a major change in cardiac output (<0.3 liters/minute) were determined to define acute vasodilator drug response as a continuous measure. In GWAS analysis, single-locus tests of association were performed using linear regression in an additive model adjusted for sex and PCs 1, 2, and 3 for all single nucleotide polymorphisms (SNP). After genomic imputation and exclusion of samples with high degree of relatedness (PI_HAT>0.2) and sex and race/ethnicity discordance, we excluded SNPs with poor imputation quality (Rsq<0.3), minor allele frequency (MAF)≤3%, and Hardy Weinberg Equilibrium P-value <0.001. In sensitivity analyses, covariate adjustment for sex, age, age2, and PCs 1-10 was used after elimination of outliers (n=2) and standardization of phenotype data as described in the Supplemental Materials. The primary analysis in the discovery cohort was restricted to patients with European ancestry with α=5.0x10−8. In the replication cohort, significant SNPs after assessment of linkage disequilibrium were tested for association with change in mPAP similar to the discovery cohort. Linear regressions were adjusted for sex and self-reported race/ethnicity. Random effects meta-analysis was employed to combine associations from both BioVU and AHN and a meta-analysis alpha level of 0.05 divided the number of SNPs tested was used for replication with an alpha level of 0.05 for nominal replication. All statistical analyses were performed using PLINK26 and SAS 9.4 (SAS, Cary, NC).
All animal experiments were approved by the Animal Research Ethics Committee of the First Affiliated Hospital of Guangzhou Medical University and were conducted in accordance with University Guidelines for the Care and Use of Animals. Adeno-associated virus 9 (AAV9)-Snx29-3×FLAG (2×1011 genomic copies/mouse) were administered to 6-week-old wild-type male mice (C57BL/6J, Vital River, Beijing, China) via an intra-tracheal route to overexpress SNX29 (Snx29-OE) in the lungs. A control group of mice were treated with saline plus AAV9 alone at the same time. We used male mice only because they exhibit worse severity of disease, hence, allowing for greater sensitivity for observations. After AAV9 administration, exposed mice were monitored for 14 additional weeks. At 20 weeks, a subset of mice either underwent measurements using the isolated perfused/ventilated lung system as described previously 27 or were subjected to 4 weeks of chronic hypoxia (10% FiO2) to induce pulmonary hypertension (n=6-10 per group) 28,29. In parallel, AAV vectors expressing shRNA targeting Snx29 (AAV-shSnx29) were delivered via the same intra-tracheal route to knock down Snx29 (Snx29-KD) in a separate group of mice. All histological images represent selections based on what is either quantified for most illustrative for that context.
Intracellular free calcium concentration ([Ca2+]cyt) in human pulmonary artery smooth muscle cells (hPASMCs) was measured according to prior work30. Briefly, PASMCs transfected, as described above, were grown to 60-70% confluence on 25-mm-diameter circular glass coverslips. The cells were treated with a Ca2+ indicator, Fura-2 (4 μM, Invitrogen-Molecular Probes, Eugene, OR) for 60 mins at 37°C with 95% air and 5% CO2. Cyclepiazonic acid (Cpa, a specific Ca2+-ATPase inhibitor) was used to induce store-operated calcium entry (SOCE). The fluorescence intensity emitted at 520 nm, excited at 340 and 380 nm, was recorded at a frequency of 1 Hz throughout the experiment and background subtracted. 340/380 emission ratios were calculated with IonOptix Ion Wizard software to determine [Ca2+]cyt in hPASMCs.
Statistical analyses were performed using GraphPad Prism 10.6 (GraphPad Software Inc, San Diego, CA). Data are presented as mean ± SEM (in bar graphs) or as box plots showing the median, interquartile range, and minimum to maximum values (whiskers). Normality was assessed using the Shapiro-Wilk test for sample sizes ≥ 6. For comparisons between two groups, a two-tailed unpaired Student’s t-test was applied, or Welch’s t-test if variances were unequal. One-way ANOVA followed by Tukey’s post hoc test was used for multiple group comparisons. In cases of non-normal distribution or sample sizes smaller than 6, group differences were evaluated using the Mann-Whitney U test (for two groups) or the Kruskal-Wallis test followed by Dunn’s post hoc test (for multiple groups). Statistical significance was defined as P < 0.05. Representative curves, Western blots, and immunofluorescence images were selected to reflect the mean/average trends of each experimental group across biological replicates, rather than choosing the best or most extreme examples. All images are illustrative of the overall experimental outcomes and consistent with corresponding quantitative statistical results. Detailed statistical information for each experiment is provided in the corresponding figure legends.
Additional materials and methods including details on differential gene expression, lymphocytes, immunostaining, real - time polymerase chain reaction, western blot, co - immunoprecipitation and other related assays, are provided in the Supplement.
RESULTS
A total of 467 individuals were identified in the discovery cohort with change in mPAP data after acute vasodilator testing (AVT) and after genotyping quality control. A total of 415 individuals were of European ancestry and included in the primary analysis. The clinical characteristics are presented in Table 1 (and Table S1), which was primarily female (n=356 [76.2%]) with an average age of 53.1 (standard deviation [SD] 15.6) years. Differences in hemodynamic characteristics before and after AVT are in Table 2.
A total of four SNPs reached statistical significance in the analysis with change in mPAP (lambda=1.000) (Figure 1). Two SNPs in a 13.5 kbp region on chromosome 2 within the death associated protein like 1 gene (DAPL1) were the most strongly associated with change in mPAP (top SNP rs79362544 MAF=0.04; beta=9.31 mmHg [standard error (SE) 1.53], p=3.04x10−9) and were in high LD (r2=0.92). A significantly associated SNP was observed in the mitotic arrest deficient 1 like 1 gene with the A allele for rs57871170 (MAF=0.04; beta=7.45 mmHg [SE 1.30], p=2.00x10−8). A significantly associated SNP was also observed in the sorting nexin 29 gene (SNX29) with the T allele for rs8057488 (MAF=0.05; beta=7.62 mmHg [SE 1.28], p=4.95x10−9). In the discovery cohort, the rs8057488 genotype distribution was 89.9% C/C (n=374), 9.6% C/T (n=40), and 0.5% T/T (n=2). The SNP rs8057488 is located in the last intron of SNX29 and no other SNPs from this gene were in LD with this signal (maximum r2=0.21) (Figure S1).
Top associated SNPs from each gene (n=3) were tested in the replication cohort comprised of both BioVU and AHN (details of clinical characteristics are presented in Tables S1 and S2 and hemodynamics in Table S3). No association was observed with change in mPAP for DAPL1 or MAD1L1 SNPs. The SNX29 SNP, rs8057488, was nominally (alpha=0.05) but not significantly (alpha=0.05/3=0.0167) associated with change in mPAP in the same direction in the replication cohort (n=107; MAF=0.10; meta-analysis beta=4.03 [SE 1.83], p=0.027) with no evidence of heterogeneity (p=0.70). In the replication cohort, the rs8057488 genotype distribution was 80.4% C/C (n=86), 17.8% C/T (n=19), and 1.9% T/T (n=2).
Additional results are presented in the Supplementary Materials including various sensitivity analyses incorporating performance of the top SNX29 SNP rs8057488 in a GWAS with additional covariates (standardized beta=0.68 [SE 0.15], p=8.61x10−6) including type of Group 1 PAH (standardized beta=6.45 [SE 1.18], p=7.22x10−8) and a GWAS of clinically-confirmed VR-PAH (odds ratio 2.72 [95%CI 1.37-5.43], p=0.004, Figures S2–S7 and Tables S4–S6). In contrast to the primary GWAS analysis of acute vaso-responsiveness, the latter are those patients who met the clinical definition as a difference in mPAP at rest and mPAP after vasodilator of greater than or equal to 10 mmHg with mPAP after vasodilator less than 40 mmHg without significant reductions in CO/CI. The distribution of SNX29 rs8057488 genotypes by VR-PAH and NR-PAH status is presented for the discovery and replication cohorts in Table 3. While limited by sample size as we stratify by SNP and then live/dead status, we next assessed associations with mortality in the discovery cohort which was sufficiently large to explore. Using a Cox proportional hazard ratio analysis, the SNP is not significantly associated with survival (HR=0.74 [0.39-1.41] for T allele, N=409, p=0.36) after adjusting for age, sex, and two PCs. Subsequent experiments sought to determine the role of the SNX29 gene and variation at rs8057488 in PAH.
From Pulmonary Hypertension Breakthrough Initiative (PHBI)-derived RNA-seq profiles of explanted lungs including 52 failed donor (FD) controls and 78 patients with PAH, SNX29 levels were lower in PAH versus FD lungs (Figure 2), suggesting a pathogenic role for reduced SNX29 in PAH. There was no statistically significant difference in mean expression levels of SNX29 between rs8057488 T allele carriers and non-carriers (P=0.085) among FD group (Figure 2A). In PAH, the difference in SNX29 expression was significant (2.3-fold higher in the T allele carriers, P=0.036; Figure 2A) indicating T allele carriers have higher SNX29 expression than non-carriers. These data suggest that rs8057488 is an eQTL in PAH lungs.
While the role of circulating lymphocytes in VR-PAH is unknown, their potential as a genomic biomarker of VR-PAH has been described 19. Based on a published cohort of VR-PAH with available microarray data, SNX29 expression was significantly higher in circulating lymphocytes isolated from subjects with clinically defined VR-PAH versus NR-PAH (combination of heritable and idiopathic PAH subjects) and non-PAH control subjects (Kruskal Wallis P value=5.6E-04) (Figure 3).
Expression of SNX29 in PASMCs were examined by RT-qPCR, western blot, and immunofluorescence staining. While trends were reduced in PAH tissues, no significant differences in SNX29 expression were observed in PASMCs between idiopathic PAH (IPAH) and control patients (Figure S8A–C). Additionally, RT-qPCR indicates no significant differences in SNX29 expression in pulmonary arteries (PAs) isolated from monocrotaline (MCT)-exposed male and female rats (Figure S8D). To determine the location of SNX29 in cells, an eGFP-SNX29 fusion protein was expressed in PASMCs. Figure S9A shows the sub-cellular distribution of fluorescence of this fusion protein verses empty vector (eGFP). eGFP fluorescence was evenly distributed within the cytoplasm of cells with little sub-cellular accumulation. In contrast, eGFP-SNX29 fluorescence was punctate, indicating localization in small compartments. An examination of possible SNX29 association with other SNX proteins was also evaluated (Figures S9–11). No co-localization was observed between SNX29 and SNX1, 2, 3, 4 or 5 (data not shown), but very strong co-localization was observed with SNX27 (Figure 4A), a component of the retromer complex 31. Counter-staining eGFP-SNX29 expressing cells with the early and late endosome markers, Rab5 or Rab7, revealed substantial co-localization of eGFP-SNX29 with these endosome markers (Figures 4B and 4C, respectively). Furthermore, we observed that endogenous SNX29 similarly co-localizes with Rab5 and Rab7 in PASMCs isolated from patients without PAH (Controls) or from those with PAH (Figures 4D). These data demonstrate that SNX29 is indeed associated with endosomes and may in fact function throughout the early-late endosome pathway.
We next tested whether reduced SNX29 induces Ca2+ signaling, known to increase vasomotor tone and pulmonary vascular remodeling 32. Levels of ORAI1 and STIM1, both well-established SOCE and Ca2+ signaling mediators 33,34 were determined by western blots after depletion of SNX29 by siSNX29 in PASMCs. As shown in Figure 5A and B, SNX29 mRNA (P = 0.0079), and SNX29 protein (P = 0.0022) levels were significantly decreased in siSNX29-transfected PASMCs compared to the control group (scramble RNA, siNC). Levels of STIM1 and ORAI1 proteins increased significantly in the siSNX29 group (Figure 5B). Fluorescence intensity of ORAI1 and STIM1 were increased in PASMCs with SNX29 knockdown versus control (Figure 5C, D and Figure S12), corroborating the expression data. Notably, a change in localization of STIM and ORAI1 to cell and nuclear membrane regions were evident from the merged immunofluorescence images after transfection of siSNX29. Fluorescent tagged STIM1-mCherry and ORAI1-CFP transfected into HEK293 cells revealed translocation of STIM1 towards the endoplasmic reticulum (ER)-plasma membrane (PM) junctions after transfection with SNX29 siRNA (Figure S13). Similarly, total internal reflection fluorescence (TIRF) microscopy showed that SNX29 knockdown facilitated the translocation of STIM1 to the cell surface. Conversely, SNX29 overexpression reduced this translocation following treatment with thapsigargin, a SERCA pump inhibitor (Figure S14A).
To further dissect the molecular mechanism, we investigated whether SNX29 could interact with STIM1 to regulate its function. Coimmunoprecipitation results showed that SNX29 complexes with STIM1 in HEK293T cells co-transfected with GFP-SNX29 and STIM1-FLAG (Figure 5E). In addition, in HEK293T cells co-expressing STIM1-FLAG and ORAI1-HA, knockdown of SNX29 enhanced the STIM1-ORAI1 complex formation. Conversely, overexpression of SNX29 reduced the interaction between STIM1 and ORAI1 (Figure 5F). Consistent with its potential role in destabilizing the STIM1-ORAI1 complex, SNX29 did not localize within the STIM1-ORAI1 puncta at the plasma membrane based on TIRF imaging. Specifically, following ER Ca2+ depletion, SNX29 failed to co-localize with STIM1 clusters at the cell surface (Figure S14B). Since both STIM1 and ORAI1 were induced with reduced SNX29, we next hypothesized that reduced SNX29 may enhance SOCE based on their known roles 35. In PASMCs, [Ca2+]cyt was measured after loading with fura-2 acetoxymethyl ester (4 μM), a Ca2+ indicator dye, using a fluorescence microscope. Cyclopiazonic acid (CPA) was used to induce SOCE. Intracellular Ca2+ store depletion with subsequent restoration of extracellular Ca2+ resulted in a significant increase in extracellular Ca2+ influx, reflecting intracellular Ca2+ store depletion-activated SOCE. As a negative control, silencing STIM1 significantly abolished CPA-mediated SOCE. In contrast, reducing SNX29 mildly but significantly enhanced CPA-mediated SOCE in PASMC (Figure 5G). Unexpectedly, SNX29 overexpression in PASMCs did not significantly diminish SOCE compared to control cells (Figure S15).
To investigate whether knockdown of SNX29 could enhance vasoconstriction, mice received orotracheal administration of an AAV-shSnx29. Isolated perfused/ventilated lung assays showed that, compared with the control group, knockdown of Snx29 in mice significantly exacerbated acute hypoxia (1% FiO2)-induced pulmonary vasoconstriction (P=0.011), while there was no significant difference in high potassium (40 mM K+)-induced pulmonary vasoconstriction (P=0.058) (Figure 5H).
To confirm the role of SNX29 in vaso-responsiveness in PAH, we next evaluated a model of acute alveolar hypoxia-induced pulmonary vasoconstriction in isolated perfused/ventilated murine lungs along with chronic hypoxia-induced pulmonary hypertension using both control and SNX29-overexpressed mice. The AAV9 vectors encoding FLAG-tagged Snx29 were administered intra-tracheally to mice based on the depicted timeline (Figure 6A). Western blots and RT-qPCR demonstrated an increased expression of FLAG-tagged Snx29 in mice lungs, suggesting successful delivery (Figure 6B, 6C, and S16). Acute alveolar hypoxia (by ventilation of 1% FiO2) or intrapulmonary perfusion of 40 mM K+ reversibly increased PAP and reached plateau pressures within 5 min in both, control and Snx29-overexpressing mice (Figure 6D). Compared to control mice, acute hypoxic-induced increases in PA pressures were significantly attenuated in isolated perfused/ventilated lungs from Snx29-overexpressing mice (P=0.029), while high K+-induced increases in PAP showed no significant difference (P=0.20) (Figure 6D), indicating that Snx29 overexpression specifically reduced hypoxic pulmonary vasoconstriction. However, overexpression of Snx29 in mice did not alter the development of chronic hypoxia-induced PH. After four weeks of hypoxic exposure, right ventricular systolic pressure (RVSP) and right ventricular hypertrophy (RVH) were not significantly different between control and Snx29-overexpressing mice (Figure 6E and F). Western blot analysis shows STIM1, but not ORAI1, levels are lower in normoxic as well as hypoxic whole lungs of mice with over-expression of Snx29 compared to control mice (Figure 6G). Immunofluorescence imaging with α-SMA and STIM1 staining in mouse lung tissue confirmed over-expression of Snx29 in SMCs (Figure 6H and S16E).
To confirm the role of SNX29 in the contractility of PASMCs in vitro, a collagen gel contraction assay was performed in PASMCs isolated from patients with IPAH that were transfected with either siNC or siSNX29. As shown in Figures S17, knockdown of SNX29 increased contraction capacity in PASMCs.
We hypothesized that immortalized lymphocytes from patients with clinically-confirmed VR-PAH may mirror what is happening in the vascular cells and PASMCs of patients with VR-PAH as has been previously done 19. We specifically evaluated for expression levels of SNX29 and its target proteins including STIM1 in these cells derived from patients with and without VR-PAH as well as those carrying the validated SNX29 variant. As shown in Figures 6I and 6J, in immortalized lymphocytes from patients with VR-PAH carrying the variant, SNX29 was up-regulated and STIM1 was down-regulated compared to NR-PAH subjects, which further bolsters the in vitro findings.
DISCUSSION
The importance of vascular tone serves as the basis for the development of acute vasoreactive testing (AVT) as well as FDA-approved therapies for PAH. In a small subset of patients termed sustained VR-PAH, calcium channel blockers were found to be therapeutic long-term, deriving a survival benefit 8–10. Independent of this sustained response, AVT also appears to stratify, patient prognosis across multiple types of PAH as well as post-operative risk in neonatal PH 11–14,36–39. Despite its clinical significance, mechanisms for AVT response and, in particular, for acute VR-PAH remain unknown. We performed the first GWAS for AVT response in PAH and nominally replicated the SNX29 signal in an independent cohort. We observed reductions in mPAP during AVT with the T allele of the intronic SNP rs8057488 in SNX29. In subset analyses beyond AVT associations, the T allele remained nominally associated with sustained VR-PAH in the discovery cohort. Furthermore, the T allele was also associated with increased SNX29 expression in PAH lung tissue, suggesting increases in SNX29 may mediate acute AVT. To understand the role of SNX29 in acute vasodilator responsiveness (AVR), we studied PASMCs, the major cellular determinant of vascular tone. SNX29 demonstrates endosomal distribution, suggesting it functions in cellular trafficking. Based on the canonical role of Ca2+ signaling in vascular tone 16,17, SNX29 reduction also enhances SOCE, complexes with and regulates localization of SOCE proteins including STIM1. In reciprocal experiments, we confirmed that increased SNX29 mediates reduced hypoxic constriction in isolated perfused mouse lung experiments, suggesting that SNX29 mediates vasodilation. Finally, we found increased mRNA and protein levels of SNX29 and reduced levels of STIM1 in circulating cells from patients with VR-PAH and notably, in patients with the SNX29 SNP. These data cumulatively suggest that SNX29 represents a novel candidate gene of AVR and VR-PAH, in part, via regulation of calcium signaling and trafficking of SOCE proteins.
There is a distinct need for genetic biomarkers that differentiate VR-PAH and NR-PAH patients 40. RNA signatures from blood samples 19 and, separately, plasma levels of provirus integration sire for Moloney murine leukemia virus 41 have correctly identified VR-PAH from NR-PAH. In a smaller published cohort, SNPs in cytoskeletal function pathways were implicated in VR-PAH 42. Our study adds to this literature by providing a comprehensive evaluation of genetic influences for this phenotype. Since genetic risk factors for PAH (e.g., BMPR2, SOX17, etc) were not represented among significant signals in our GWAS for AVT, genetic determinants for AVT are likely distinct from known genetic risk factors for PAH 43,44. The newly identified variants may point to novel mechanisms for VR-PAH pathophysiology. Although limited data is available for the biological function of SNX29, the gene has a biologically plausible role in VR-PAH pathophysiology 45. No functional consequences for the associated variant were identified in silico but our expression analyses in lung tissues of patients with PAH suggested the variant may reflect a cis-eQTL. Further study of functional effects and linkage with other functional variants is warranted.
While limited in published data, SNX29 demonstrates an endosomal distribution in PASMCs, consistent with the function of other SNX family of proteins. The data also suggest that Snx29 over-expression in murine PH lungs attenuates vasoconstriction. However, Snx29 does not appear to influence pulmonary vascular remodeling and the development of hypoxic PH. Whether the latter observation is due, in part, to model-specific effects and limitations is unknown. For example, hypoxic murine PH reflects more Group 3 rather than Group 1 PAH and the role of Snx29 in rodent MCT or hypoxia-sugen models remains to be investigated.
Our data suggests that SNX29 mediates calcium signaling and SOCE, possibly via trafficking of STIM1, and rearrangement of STIM to the plasma membrane. Our data demonstrates that SNX29 and STIm1 form a complex but whether they directly bind is unclear. Gene Ontology annotations for SNX29, also known as RUN Domain-Containing Protein 2A (RUNDC2A), include microtubule motor activity and phosphatidylinositol binding 46. We speculate whether SNX29 may indirectly regulate SOCE via its ability to bind to phosphatidylinositol (PI). SOCE mediates Ca2+ signaling to regulate a variety of critical cell functions across a myriad of tissues after the generation of inositol 1,4,5-triphosphate (IP3) in response to both, receptor stimulation and the release of Ca2+ from the ER. The evolutionarily canonical 100-130-amino PX domain in the broader SNX family facilitates the interaction of SNXs with PIs, most commonly PI-3-phosphate (PI3P). SNX11 and SNX16 also belong to this group and promote the trafficking of proteins between cellular compartments 45. Mutation of the PX domain previously disrupted the association between SNX16 and PI3P 47. Like other SNX members, whether the PX domain of SNX29 interacts with PI3P during cellular trafficking of SOCE proteins remains unknown.
SOCE refers to Ca2+ influx that is activated by depletion of sarco/ER (S/ER) Ca2+ stores 33,48. STIM proteins function as S/ER Ca2+ sensors, and upon S/ER Ca2+ depletion STIM rearranges to S/ER-plasma membrane junctions and activates Orai Ca2+ influx channels 34. In the current study, spatial distribution and cellular localization of ORAI1 and STIM1 displayed a similar change in their localization to the plasma membrane after SNX29 silencing, indicating that SNX29 may mediate their rearrangement to S/ER-plasma membranes. Whether this localization of STIM1 triggers activation of Orai Ca2+ influx channels, as further suggested by the increase in its expression at the plasma membrane is unknown. These spatial distribution changes may also reflect an indirect influence of SNX29 silencing. Our expression studies also indicate an inverse correlation between SNX29 and STIM1. While it is unclear how SNX29 regulates STIM1 expression, we speculate whether localization differences in STIM1 influence its overall stability and/or expression. Alternatively, SNX17 was previously shown to facilitate lysosomal degradation of SERCA2a 49 and whether SNX29 similarly regulates STIM1 degradation is unknown.
Several limitations are important to address. While the current study focuses on the role of SNX29 in AVR, its role in PAH development, VR-PAH, or calcium channel blocker response remains unknown, along with patients who meet the newer threshold of disease definition for PAH with a mean PA pressure (⩾_20mm Hg). While AVR is typically characterized in patients with Group 1 PAH, we utilized a hypoxic PH model to study vaso-responsiveness in vivo. This model has well-known limitations on its mimicry of Group 1 PAH histopathology; the role of SNX29 in PAH development in rat models remains unknown and an important area for consideration of future work. Similarly, the role of sex in Snx29 function remains unknown in murine models of PH. Moreover, the limits of interpretation of cell-specific roles of SNX29 in vivo remain uncertain given AAV administration is known to be associated with heterogenous uptake. Statistical power was limited for genome-wide signals for the dichotomous VR-PAH phenotype. To combat this effect, we leveraged the largest available PAH cohort to study the genetics of AVR and explored VR-PAH as part of our secondary analyses. The small absolute numbers constitute the largest reported population of patients with both genetic data and data from pre- and post-vasodilator testing in PAH. The rarity of this phenotype also explains the inclusion of non-European populations in the replication cohort. Multiple significant SNP effects were observed with change in mPAP, suggesting that the sample size was sufficient to identify the observed associations. Furthermore, change in mPAP (AVR) was used as a model to implicate relevance for the dichotomous VR-PAH phenotype, given the added power associated with more samples and the use of a continuous outcome measure. The low MAFs for associated SNPs might suggest a false positive association. However, these MAFs reflect the global frequency of VR-PAH among IPAH patients, estimated to be 5-10% 7. The SNX29 SNP neither passed a strict Bonferroni threshold in the sensitivity analysis with additional covariates, nor for three SNPs in replication, again suggesting the potential for a false positive signal. Concerns for spurious results are reduced by the nominal replication of our association in an independent replication cohort with similar phenotypic data and the extensive functional work described in this manuscript. The data as a whole indicate the likelihood of a true effect of variation in SNX29 on this unique and rare phenotype.
In summary, we describe the first GWAS for AVR in PAH followed by experimental characterization of the top locus. Our findings implicate genetic variation in SNX29 and differential responses to AVT in PAH. They advance our understanding of the molecular underpinnings of VR-PAH and suggests that increased SNX29 expression causes vasodilation through reduced SOCE, resulting from endosomal trafficking of SOCE proteins such as STIM1. Further, polymorphisms in SNX29 may increase SNX29 expression, partially explaining enhanced acute vaso-responsiveness in VR-PAH patients. Further study is warranted to assess the role of SNX29 in the pathophysiology of VR-PAH and as a druggable target in PAH.
AH
Dr. Anna Hemnes
Pulmonology & Critical Care · Pulmonary Vascular MedicineVanderbilt Healthannahemnes