Novel mutations and decreased expression of the epigenetic regulator TET2 in pulmonary arterial hypertension

Abstract

Background:

Pulmonary arterial hypertension (PAH) is a lethal vasculopathy. Hereditary cases are associated with germline mutations in BMPR2 and 16 other genes. However, these mutations occur in under 25% of idiopathic PAH patients (IPAH) and are rare in PAH associated with connective tissue diseases (APAH). Preclinical studies suggest epigenetic dysregulation, including altered DNA methylation, promotes PAH. Somatic mutations of Tet-methylcytosine-dioxygenase-2 (TET2), a key enzyme in DNA demethylation, occur in cardiovascular disease and are associated with clonal hematopoiesis, inflammation and adverse vascular remodeling. The role of TET2 in PAH is unknown.

Methods:

To test for a role of TET2, we utilized a cohort of 2572 cases from the PAH Biobank. Within this cohort, gene-specific rare variant association tests were performed using 1832 unrelated European PAH patients and 7509 non-Finnish European gnomAD subjects as controls. In an independent cohort of 140 patients, we quantified TET2 expression in peripheral blood mononuclear cells. To assess causality, we investigated hemodynamic and histologic evidence of PAH in hematopoietic Tet2-knockout mice.

Results:

We observed an increased burden of rare, predicted deleterious, germline variants in TET2 in PAH patients of European ancestry (9/1832) compared to controls (6/7509; relative risk=6, p=0.00067). Assessing the whole cohort, 0.39% (10/2572) of patients had 12 TET2 mutations (75% predicted germline and 25% somatic). These patients had no mutations in other PAH-related genes. Patients with TET2 mutations were older (71±7 years versus 48±19 years, p<0.0001) unresponsive to vasodilator challenge (0/7 vs 140/1055 (13.2%)), had lower PVR (5.2±3.1 versus 10.5±7.0 Woods units, p=0.02) and had increased inflammation (including elevation of IL-1β). Circulating TET2 expression did not correlate with age and was decreased in >86% of PAH patients. Tet2-knockout mice spontaneously developed PAH, adverse pulmonary vascular remodeling and inflammation, with elevated levels of cytokines, including IL-1β. Chronic therapy with an antibody targeting IL-1β blockade regressed PAH.

Conclusions:

PAH is the first human disease related to potential TET2 germline mutations. Inherited and acquired abnormalities of TET2 occur in 0.39% of PAH cases. Decreased TET2 expression is ubiquitous and has potential as a PAH biomarker.

Introduction

Pulmonary arterial hypertension (PAH) is a lethal vasculopathy hemodynamically characterized by increased mean pulmonary arterial pressure (mPAP >20 mmHg) and pulmonary vascular resistance (PVR >3 Wood units)¹. Histologically PAH is characterized by obliterative pulmonary vascular remodeling. The current classification system divides Group 1 PAH into idiopathic (IPAH), hereditary (HPAH), and associated PAH (APAH; a category which includes patients with connective tissue diseases (CTD), such as scleroderma). 7-year survival rates of IPAH and CTD-PAH are 56%, and 35%, respectively². High mortality rates in PAH are a consequence of late diagnosis, due to the non-specific clinical manifestations of the disease, the lack of biomarkers, and the absence of a curative treatment³. Moreover, the fundamental cause(s) of PAH remain elusive in many patients.

The etiology of PAH is heterogeneous and remains imperfectly understood, although it is characterized by increased inflammation and fibrosis and impaired angiogenesis, mitochondrial metabolism and mitochondrial dynamics⁴. At the genetic level, germline pathologic variants have been reported in 17 genes, by far the most prevalent of which is BMPR2^55–7. Epigenetic dysregulation of genes is also important in PAH, and pathological activation of DNA methyltransferases (DNMT) has been shown to increase DNA methylation of specific genes associated with disease progression^8,9. DNA methylation is a dynamic process that reflects the balance between the activity of DNMT, which adds methyl groups, and ten-eleven translocation methylcytosine dioxygenase (TET), which removes methyl groups, from cytosine nucleotides in DNA. Excessive methylation generally inhibits gene transcription, although there are also interactions between methylation sites and methyl binding proteins, which can alter the expression of other genes¹⁰. Somatic inactivating TET2 mutations have recently been associated with development of inflammation^11,12 and atherosclerosis^13,14. In addition, acquired mutations in this gene underlie clonal hematopoiesis of indeterminate potential (CHIP), a precursor to myelodysplastic syndrome (MDS), myeloproliferative neoplasms (MPN) and even acute myeloid leukemia (AML)^15,16. There is no known germline TET2 mutation syndrome.

In this study we evaluated gene specific TET2 exome sequencing data from the largest PAH cohort assembled to date, including 2572 patients in the PAH Biobank. Unlike prior genetic studies, the biobank includes subjects with APAH and non-European ancestry. We performed gene-specific rare variant association analyses using up to 1832 cases of European origin from the PAH Biobank and transcriptomic analysis in an independent cohort to assess TET2 expression. In the entire cohort, we identified 12 predicted deleterious variants in TET2 (75% predicted germline and 25% somatic), novel to PAH. None of the variant carriers were responsive to acute vasodilator challenge. This is the first time that putative germline TET2 mutation has been associated with a human disease. We also identified ubiquitous downregulation of the expression of TET2 in the peripheral blood mononuclear cells (PBMC) of IPAH and APAH patients. Finally, we evaluated Tet2 depleted mice and demonstrated that they spontaneously develop inflammation, pulmonary vascular obliteration, and pulmonary hypertension, providing biological plausibility that disorders in this pathway can cause PAH.

Methods

The data, analytic methods, and study materials for the purposes of reproducing the results or replicating procedures can be made available on request to the corresponding author who manages the information.

Subject study

All patient samples were obtained following written informed consent and approval from local Institutional Review Boards. The study was approved by an institutional review committee and the subjects gave informed consent. Clinical characteristics of the PAH Biobank and John Hopkins cohorts are described in Table-S1. Additional methodologies are available in online supplements.

Exome sequencing, bioinformatics and statistical analyses for genetic studies

DNA was extracted from whole blood from participants in the National Biological Sample and Data Repository for PAH (PAH Biobank), and exome sequencing was carried out in collaboration with the Regeneron Genetics Center or at the Cincinnati Children’s Hospital Medical Center DNA Sequencing and Genotyping Core, as described previously^17,18. We used Burrows-Wheeler Aligner (BWA-MEM)¹⁹ to map and align paired-end reads to human reference genome (GRCh38), Picard MarkDuplicates to identify and flag PCR duplicates and GATK²⁰ HaplotypeCaller to call genetic variants. We used ANNOVAR²¹ to aggregate variant annotations, including population allele frequency (gnomAD²² exome/genome, ExAC, and 1000 genomes), predicted functional effects based on RefSeq, and predicted deleterious scores by methods such as REVEL²³.

For case-control comparisons, we performed principle components and relatedness analyses for the whole PAH cohort and identified 1832 unrelated European cases. We defined rare variants as variants with an allele frequency <0.01% across all gnomAD exome sequencing samples (European and non-European). We used heuristic filters to minimize technical artifacts between cases and controls, excluding variants that met any of the following criteria: missingness >10%, minimum alternate allele read depth ≤4 reads, alternative allele fraction ≤25%, or genotype quality <90. We used 7509 European gnomAD whole genome sequence (WGS) data as the control set. Only variants with FILTER “PASS” in gnomAD WGS (data release v 2.02) and located in the IDT xGen captured protein coding region were included in the analysis. Finally, we observed that the frequency of rare synonymous variants in cases and controls was virtually identical, indicating that the data sets are comparable (enrichment rate=1.0, p=0.11) (Table-S2). Assuming linkage disequilibrium is negligible among ultra-rare variants, we performed gene-wise burden tests of rare variants for TET2 using Fisher’s exact test to compare carrier frequency between cases and controls. We tested the association in three groups of variants between cases and controls: (a) rare deleterious missense variants (“D-MIS”, defined as REVEL score >0.5); (b) likely-gene-disrupting (LGD) variants; and (c) D-MIS+LGD. We tested 3 genes TET1, TET2, TET3. We also tested the association in IPAH alone and APAH alone. Thus, we performed 11 tests (3 genes × 3 sets of variants and 1 gene × 2 PAH subclasses) and set the Bonferroni-corrected threshold for significance at p<0.0045. We then assessed the entire PAH Biobank (n=2572 cases) for rare (AF<0.0001), deleterious (LGD or D-MIS) variants in TET2 using Integrative Genome Viewer (Illumina, San Diego, CA). Nearly 100% of both cases and controls had >10X sequencing coverage across the gene, and 99.6% of the targeted regions had >15X coverage in cases and 100% in controls (Figure-S1A,B), indicating that there was no systematic bias in the detection of TET2 variants in cases vs controls.

To detect likely somatic variants (mosaicism) in TET2, we first used SAMtools (version 1.3.1–42)²⁴ to improve calling of genetic variants with low allele fraction. We then processed SAMtools calls using a set of heuristic filters to remove variants located in repeat regions (mapability, segmental duplication) or showing evidence of strand bias, and screened all variants using Integrative Genome Viewer. We then took the union of variants called by GATK HaplotypeCaller and SAMtools and considered variants with alternative allele fraction less than 25% as likely somatic mutations. Mosaic mutations were confirmed by TA cloning (Thermo Fisher, Waltham, MA) of exonic PCR products followed by Sanger sequencing of individual clones (Figure-S1C). Details are available in the online supplements.

Microarray and gene expression

A microarray assay was performed on RNA extracted from peripheral blood mononuclear cells (PBMC) of 50 scleroderma-associated PAH (SScPAH), 30 IPAH, 19 scleroderma without PAH patients (SSc) and 41 healthy controls from Johns Hopkins PH Program and Johns Hopkins Scleroderma Center, as previously published²⁵ and described in online supplementals. The expression data have been deposited in NCBI’s Gene Expression Omnibus and are accessible through GEO Series accession number GSE 33463. Analytical methods were performed as previously published²⁵. Details are available in the online supplements.

Animal experiments

All experiments were performed in accordance with Queen’s University biosafety and ethical guidelines (ROMEO/TRAQ#6016826). The procedures followed were performed in accordance with institutional guidelines. Conditional, haematopoietic heterozygous (Tet2^+/−) and homozygous (Tet2^–/–) knockouts were generated by crossing parental floxed (Tet2^f/f, B6;129S-Tet2tm1.1Iaai/J) and Vav1-Cre (B6.Cg-Tg(Vav1-Cre)A2Kio/J) mice (Jackson Laboratory, Bar Harbor, ME) and validated, as previously described^12,26. IL-1β PAH regression experiments were performed on 5 Tet2^f/f and 10 Tet2^–/–. 10 Tet2^–/– mice (7 months old) were randomly distributed in 2 groups of 5 age/sex-matched mice and treated with anti-IL-1β antibody (a generous gift from Novartis Pharma AG) at a dose of 10 mg/kg/week IP for 6 weeks) or IgG2a (a generous gift from Novartis Germany) as a placebo. Treatments and data collection were performed by scientists blinded to treatment groups. PAH development was assessed by echocardiography, right heart catheterization, 2-photon, confocal microscopy lung perfusion and histology, as described in online supplements. Inflammation was assessed by fluorescence-activated cell sorting (FACS) and NanoString nCounter PanCancer Immune Panel Profiling (Seattle; WA), as previously published¹² (see online supplements).

Results

Exome sequencing identifies rare deleterious variants in TET2.

Detailed characterization of the PAH Biobank and gnomAD cohort are described in a separate report¹⁸ and in Table-S1. The cohort includes 2572 total cases: 43% IPAH, 48% APAH 4% FPAH and 5% other (Table S1). Among APAH patients, 58% (722/1239) had some form of connective tissue disease. The majority of cases are adult-onset, with a 3.7:1 ratio of females to males, typical of adult PAH. The genetically determined ancestries were 72% European, 12% Hispanic, 11% African, and smaller percentages of other ancestries.

We first limited the association analysis to 1832 unrelated European cases and 7509 non-Finnish, European controls. Using a REVEL score >0.5 to define D-MIS variants, we tested for association in three categories of variants: D-MIS, LGD or D-MIS+LGD. Among all PAH cases, we observed significant enrichment of LGD (8/1832 cases vs 4/7509 controls; RR=8.18, p=0.0005) as well as D-MIS+LGD variants for TET2 (9/1832 cases vs 6/7509 controls; RR=6.15, p=0.00068) (Table-1). The association was largely due to patients with IPAH (RR=10.79, p=8.483e-05) (Table-2).

Table 1.

Enrichment of rare predicted deleterious variants* in candidate PAH risk gene TET2† amongst European PAH cases.

TET2
Mutation type‡	PAH cases (n=1832)	gnomAD WGS controls (n=7509)	Relative risk	P-value
D-MIS	1	2	2.05	0.48
LGD	8	4	8.18	0.0005
D-MIS or LGD	9	6	6.15	0.00068

Table 2.

Enrichment of rare predicted deleterious variants^* in candidate PAH risk gene, TET2^†, among European APAH and IPAH cases.

TET2
PAH subclass	D-MIS or LGD‡	Relative risk	P-value
APAH (n=843)	2	3.00	0.19
IPAH (n=812)	7	10.79	8.483e-05

^*Variant filters: MAF <0.0001 and likely gene damaging or missense with REVEL score >0.5.

^†TET2 transcript: NM_001127208.2.

^‡D-MIS, missense with REVEL score >0.5; LGD, likely gene damaging

Abbreviations: APAH associated pulmonary arterial hypertension; MAF, minor allele frequency; IPAH, idiopathic pulmonary arterial hypertension; WGS, whole genome sequencing.

We then identified 1 additional mutation in the 740 non-European patients of the PAH Biobank (Table-3 and S3A). In the total cohort of 2572 cases, we identified 9 unique likely germline variants in 8 patients (Table-3 and S3A) and 3 somatic variants in 3 patients (Table-3 and S3B and Figure 1A). All mutations were unique, meaning the affected patients were not shown to carry variants in known PAH risk genes^7,27,28. In addition, none of the TET2 mutant patient carriers had hematologic malignancies at the time of enrollment in the PAH Biobank. Two-dimensional structure of the encoded protein showed that 6 deleterious variants (both germline and somatic) localized to the conserved catalytic (TET/J-binding protein methylcytosine dioxygenase activity) domain (Figure 1B).

Table 3.

Rare, predicted deleterious germline and somatic variants in candidate PAH risk gene TET2 amongst 2572 PAH Biobank cases.

TET2*
Patient ID	Mutation type	Nucleotide change	Amino acid change	MAF, gnomAD WES	VAF	CADD phred	REVEL score	PAH class	Gender	Genetic ancestry	Age (y) enrollment	Age (y) onset	Mean PAP, (mmHg)	Mean PCWP (mmHg)	CO (L/min)	PVR (WU)	Mean SAP (mmHg)	Mean SAP:PAP
03–057	Germ	c.1530_1531insCACCT	p.Lys513Thrfs22	.	31%	.	.	IPAH	F	EUR	74	65	38	14	4.2	5.71	NA	NA
03–100	Germ	c.2233C>T	p.Gln745✶	.	38%	37	.	IPAH	F	EUR	71	70	47	10	5.55	6.67	NA	NA
12–206	Germ	c.2872C>T	p.Gln958✶	1.22E-05	39%	41	.	IPAH	F	EUR	84	84	55	8	3.67	12.81	87	1.58
29–016†	Germ Soma	c.4081G>A c.5618T>C	p.Gly1361Ser p.Ile1873Thr	. 1.33E-05	49% 46%	34 26	0.57 0.61	IPAH	F	EUR	72	69	47	17	9.23	3.25	80	1.7
19–036	Germ Germ	c.4546C>T c.817C>T	p.Arg1516✶ p.Gln273✶	1.59E-05	54% 51%	40 36	.	IPAH	F	EUR	74	67	33	9	9.44	2.54	117	3.5
12–156	Germ	c.4893T>G	p.Tyr1631✶	.	36%	36	.	IPAH	M	EUR	84	83	40	12	3.3	8.48	88	2.2
13–025	Germ	c.3336delA	p.Asp1113Ilefs✶4	.	26%	.	.	APAH-Porto	M	EUR	67	66	36	13	4.8	4.79	NA	NA
08–055	Germ	c.4396C>T	p.Gln1466✶	.	47%	45	.	APAH-CTD	F	EUR	73	69	45	10	3.7	9.46	NA	NA
30–037	Soma	c.2862G>A	p.Trp954*	.	18%	39	.	APAH-CTD	M	Hispanic	48	47	46	5	6.53	6.28	116	2.52
23–014	Soma	c.5639T>C	p.Leu1880Pro	.	26%	26	0.62	APAH-CHD	F	EUR	86	79	46	4	5	8.4	NA	NA
Mean ± SD, TET2 carriers									3.5:1		73.3± 10.9	66.9 ± 10.7	43.3± 6.5	10.4 ± 3.6	5.5 ± 2.2	6.8 ± 3.1	97.6 ± 17.5	2.3 ± 0.8
n, TET2 carriers											10	10	10	10	10	10	5	5
Mean ± SD, cohort APAH+IPAH excluding TET2 carriers											52 ± 18	48 ± 19	50 ± 14	10 ± 4	4.5 ± 1.8	10.5 ± 7.0	90 ± 19	2.0 ± 0.7
n, cohort APAH+IPAH excluding TET2 carriers											2340	2340	2292	2232	1641	1595	1399	1397
p-Value											1.00E-05	1.00E-05	0.002	NS	NS	0.02	NS	NS

Variant filter: allele frequency <0.0001 and likely gene disrupting (stop/gain, frameshift or canonical splicing) or missense with REVEL score >0.5.

^*TET2 transcript: NM_001127208.2

^†Exceptions to the mosaic pipeline variant filter (MAF and alternate allele fraction); detected as known mutation hotspot in cancer.

Abbreviations: APAH-CHD, pulmonary arterial hypertension associated with congenital heart disease; APAH-CTD, pulmonary arterial hypertension associated with connective tissue diseases; APAH-Porto, pulmonary arterial hypertension associated with porto-pulmonary hypertension; FPAH, familial pulmonary arterial hypertension; Germ, germline; MAF, minor allele frequency; IPAH, idiopathic pulmonary arterial hypertension; PAP, pulmonary artery pressure; PCWP, pulmonary capillary wedge pressure; PVR, pulmonary vascular resistance; SAP, systemic arterial pressure; Soma, somatic; WES, whole exome sequencing; VAF, variant allele frequency; WU, Wood unit. Unpaired t-test

Figure 1. Increased mutation and decreased TET2 gene expression in blood from PAH patients.

Topologic analysis and gene expression of the Human DNA tet methylcytosine dioxygenase 2 (TET2) in pulmonary arterial hypertension (PAH). A) Distribution of TET2 somatic and germline deleterious variants. B) Locations of rare deleterious PAH patient-derived TET2 variants within the two-dimensional protein structures. The numbers of variants at each amino acid position are indicated on the y-axis. Germline variants are shown above the protein schematic; mosaic variants are below. D-MIS, predicted damaging missense; LGD, likely-gene-disrupting (stopgain, frameshift). TET/JBP, TET/J-binding protein catalytic domain. Protein domain coordinates were modified according to UniProtKB). C) TET2 gene expression in peripheral blood mononuclear cells (PBMCs) from 41 healthy control, 30 idiopathic PAH (IPAH), 50 scleroderma associated PAH (SSc-PAH) and 19 scleroderma without PAH (SSc). Violin plot was used to visualise data distribution; black line is the median; red lines are the upper and lower quartile. one-way ANOVA. **P<0.01; ****P<0.0001; n.s. non-significant.

TET2 variant carriers exhibited increased overall inflammation compared to age/sex-matched PAH non-mutated patients, and age/sex-match controls, as assessed by measurement of 42 circulating inflammatory markers (area under the curve AUC 266.9 ±81.74 for TET2 carriers; 117.1 ±58.9 TET2 non-carriers; 41±5 for healthy control p<0.05) (Figure S2A–B). We focused on the expression of pro-inflammatory cytokines/chemokines and reported that TET2 variant carriers had increased expression of 30 pro-inflammatory cytokines (AUC 204.3 ±72.14 versus 75.26 ±17.92 p<0.001) (Figure 2A–B and S2A–B and Table-S4). IL-1β expression was increased in 70% of TET2-mutated patients compared to their matched non-mutated PAH patients (Figure S2C).

Figure 2. TET2 mutation is associated with a pro-inflammatory phenotype in blood of PAH patients.

Expression of 30 pro-inflammatory cytokines were assessed in blood of 9 healthy patients, 10 age/sex-matched PAH TET2 variant carriers and 10 matched PAH TET2 non-carriers patients. A) TET2 variant carriers show increased levels of 28 cytokines (IFNa2; IP-10; IL-12p40; IL12p70; IL-6; IL-1β; IL-2; Fractalkine; IFNy; IL-15; IL-1a; IL-18; IL-3; G-CSF; IL-7; TNFa; IL-17A; MIP-1a; MIP-1B; MDC; TNFB; IL-5; Flt-3L; MCP-3; IL-8; RANTES; GRO alpha) and decreased levels of 2 cytokines (MCP-1; Eotaxin) compared to matched non-carriers patients. Values are expressed as fold change compared to healthy patients. B) TET2 carriers display overall increased levels of pro-inflammatory markers measured by an increase of the area under the curve. One-way ANOVA. Values are expressed as mean±SEM. ***P<0.001

Clinical phenotypes, including demographics, PAH medication and right heart catheterization (RHC) data, are provided in Tables 3 and 4. None of the TET2 variant carriers (0/7) vs 140/1055 (13.2%) in the remainder of the cohort, were vasodilator responders, as assessed by standard criteria (Table 4)²⁹. Consistent with the severity associated with lack of vasodilator responsiveness, the use of endothelin receptor inhibitors and soluble guanylate cyclase stimulators was increased in TET2 variant carriers (Table 4). Two-thirds (6/10) of the patients with rare deleterious germline or somatic TET2 variants had IPAH (Table 4 and S3) Female:male ratio and genetic ancestry of TET2 germline and somatic carriers were similar to the overall cohort whilst the mean age-of-onset (66.9 ± 10.7years) was significantly older than that of all APAH+IPAH (48±19 years, p<0.0001) (Table 3). In addition the age of TET2 subjects in our study was older than in previous reports for patients with PAH related to mutations in BMPR2^5,6,17 and TBX4¹⁷. TET2 variant carriers had lower mPAP (43.3±6.5 mmHg) and PVR (6.8±3.1 Woods units) compared to all the remaining APAH+IPAH subjects in our cohort (50±14 mmHg, p<0.0001 and 10.5±7.0 Woods units, p=0.02) (Table 3).

Table 4.

Vasodilator response^* and PAH medication among TET2 deleterious variant carrier

Vasodilator used	Cohort APAH+IPAH excluding TET2 carriers	APAH+IPAH germline TET2 carriers	APAH+IPAH somatic TET2 carriers	APAH+IPAH TET2 carriers†
Oxygen + Nitric Oxide	50/286 (17.5%)	0/3 (0%)	0/1 (0%)‡	0/3 (0%)
IV Epoprostenol	13/110 (11.8%)	-	-	-
Inhaled Nitric Oxide	60/564 (10.6%)	0/2 (0%)	0/1 (0%)	0/3 (0%)
Oxygen	5/16 (31.2%)	-	-	-
IV Nitroprusside	1/7 (14.3%)	-	-	-
IV Adenosine	5/37 (13.5%)	-	0/1 (0%)	0/1 (0%)
Inhaled Iloprost	1/2 (50%)	-	-	-
Inhaled Epoprostenol	1/6 (16.7%)	-	-	-
Alprostadil	2/8 (25%)	-	-	-
Calcium Channel Blocker	0/3 (0%)	-	-	-
Unknown	2/16 (12.5%)	-	-	-
Total	140/1055 (13.2%)	0/5 (0%)	0/3 (0%)	0/7 (0%)
PAH medication	Cohort APAH+IPAH excluding TET2 carriers			APAH+IPAH TET2 carriers†
PDE5 inhibitor	2014/2674 (75.3%)			14/18 (77.8%)
Prostacyclin analog	1356/2674 (50.7%)			10/18 (55.6%)
Endothelin receptor inhibitor	1036/2674 (38.7%)			11/18 (61.1%)‡
Stimulator of soluble guanylate cyclase	44/2674 (1.6%)			1/18 (5.6%)
Calcium channel blocker	239/2674 (8.9 %)			0/18 (0%)
RTK inhibitor	5/2674 (0.2%)			0/18 (0%)
Unknown	29/2674 (1.1%)			1/18 (5.6%)

^*Positive vasodilator responders were defined according to standard criteria as a drop of mean pulmonary arterial pressure (mPAP) >10mmHg to mPAP at rest <40mmHg with preserved or improved cardiac output.

^†Somatic + germline carrier (patient 29–016)

^‡Difference statistically significant compared to cohort APAH+IPAH excluding DNMT3A and TET2 carriers; p< 0.05, z-test.

APAH, associated PAH; IPAH, idiopathic PAH; IV, intra-venous; PDE5, phosphodiesterase 5; RTK, receptor tyrosine kinase

Decreased TET2 gene expression in PAH

Having established the occurrence of TET2 rare deleterious variants in PAH patients, we next investigated TET2 expression in PBMCs. Gene expression omnibus (GEO) analysis was acquired from 50 SSc-PAH patients, 30 IPAH patients, 19 scleroderma without PAH patients (SSc) and 41 healthy controls. All the cases had adult-onset PAH, with genetically determined ancestries as follows: 82.1% Caucasian, 14.3% African, with smaller percentages of other ancestries. The sex ratios (female:male) were control 4.9:1, IPAH 5:1, and SSc-PAH 3.7:1 whilst all SSc patients were female (Table-S1). TET2 gene expression, relative to healthy subjects, was decreased in 86% of SSc-PAH patients (relative expression; 0.75; p<0.0001); 86.7% of IPAH patients (relative expression; 0.74; p<0.0001) and 68% of scleroderma patients without PAH (SSc; relative expression; 0.79; p<0.01) (expression normalized to 1) (Figure-1C). Receiver operating characteristic curve (ROC) analysis was performed for 41 healthy control and 80 PAH patients (IPAH/SSc-PAH) and revealed a potential biomarker value for TET2 expression (AUC:0.78; p<0.0001) (Figure S3A–G). When assessed using multiple logistic regression analysis, including both age and TET2 expression, ROC curve analysis reveals an AUC of 0.86, p<0.0001 [specificity 81%, sensitivity 80% (Figure S3–H)]. The AUC is significantly higher in the model including age and TET2 compared to a model including age alone (Z-score =2.47), as analysed following the method described by Hanley and McNeil³⁰. This observation confirms that TET2 expression has a potential value as biomarker for PAH. Gene expression of other TET paralogs, TET1 and TET3, is not affected in PAH. Moerover,TET2 expression does not correlate with the age (Table-S5A–B, Figure S3I,J)

Experimental Tet2 depletion is associated with spontaneous development of PH in mice

We next assessed the biologic plausibility that hematopoietic TET2 depletion could result in PAH. We performed hemodynamic measurement in 15 conditional Tet2 KO mice (Tet2^−/−) and 15 sex and age-matched control mice (Tet2^f/f) (9 males; 6 females; age 7 to 10 months). Mice with Vav-Cre-mediated Tet2 depletion spontaneously developed pulmonary hypertension, evident as a significant decrease in PAAT (p<0.01), an increase in RVSP (p<0.001), TPR (p<0.01), arterial elastance (p<0.01), adverse pulmonary vascular remodeling (p<0.05) and decreased perfusion of distal pulmonary arteries (p<0.01), compatible with obliteration of the microvasculature (Figure-3A–H and Figure-S4A–F; S5A–D; S6A). We observed no change in heart rate, stroke volume, cardiac output, RV-dP/dT_max, RV-dP/dT_min or LV function (LVS weight, E/E’, MAPSE, arterial systolic and diastolic pressure, LV-EF, LVSP, LVEDP, Tau Mirsky, dP/dt_max, dP/dt_min) (Figure-S6B–F and S7A–K). Heterozygous animals (Tet2^+/−) also had increased RVSP (p<0.05) and TPR (p<0.05), with a trend toward increased arterial elastance but no changes in CO (Figure-S8A–D). We assessed PH development in 2-month-old Tet2^−/− mice (5 males) and showed that, compared to age-matched Tet2^f/f mice, Tet2 depletion in young animals was not associated with PH phenotype (no significant differences in RVSP, TPR, CO) (Figure-S9A–C).

Figure 3. TET2 depletion in hematopoietic cells and pulmonary hypertension in a murine model.

Haemodynamic assessment of pulmonary hypertension using right heart catheterization and echocardiography of 6–15 Tet2^−/− and age-sex matched Tet2^f/f mice show A) increased right ventricular systolic pressure (RVSP), B) decreased pulmonary artery acceleration time (PAAT) and, C) increased total pulmonary resistance (TPR) in Tet2^−/− mice compared to Tet2^f/f animals. D) Pulmonary arterial vascular remodelling was blindly quantified by the percent of wall thickness: (total diameter−internal diameter)/total diameter by immunofluorescence (smooth muscle actin); 10 arteries per mice on 5 mice per group. Pulmonary vascular wall thickness is increase in Tet2^−/− mice compared to Tet2^f/f. Perfusion of pulmonary vessels has been assessed in Tet2^−/− and Tet2^f/f mice (FITC albumin perfusion, 2 photon microscopy). Perfused vessels have been clustered according to their volumes in 3 categories: small (15–225μm³), intermediate (225–3347μm³) and big (3347–50000μm³). Tet2^−/− mice display decreased numbers of E) small and F) intermediate (P=0.08) perfused vessels whilst G) number of large vessels perfused remains the same compared to Tet2^f/f (number of vessels/1e⁶ μm3). H) Representative pictures of vessels perfused by FITC-albumin in the lung of Tet2^−/− and Tet2^f/f (500X550μm). I) Compared to Tet2^f/f animals, lungs from mutated mice (Tet2^−/−) show an elevated macrophages population measured by fluorescence activated cell sorting (FACS; F4/80; CD11b; n=6 Tet2^−/− and age-sex matched Tet2^f/f). J) Quantification of inflammatory cytokines and chemokines in total lung of 3 Tet2^−/− and age-sex matched Tet2^f/f mice displays up-regulation of Il1b, Cxcr2, Csf3r, C5ar1, Fpr2, Amica1, Ccr1, Mmp9, Cd33, Itgam, H2-Q10, Arg2, Clec4n, Il1rn, Rsad2, Il1r2 and downregulation of Ccr6 gene expression in mutated animals. Results show change >1.8 fold Log2-transformed normalized NanoString mRNA counts. Unpaired t- test. Values are expressed as mean±SEM. n.s.: non-significant; *P<0.05; **P<0.01; ***P<0.001.

PH development was associated with increased inflammation in the lungs of Tet2^−/− mice, evident as macrophage accumulation (p<0.01; Figure-I) and dysregulation of 61 inflammatory markers (59 upregulated and 2 downregulated, Table-S6). Seventeen inflammatory markers had a change >1.8-fold (up-regulation of Il1b, Cxcr2, Csf3r, C5ar1, Fpr2, Amica1, Ccr1, Mmp9, Cd33, Itgam, H2-Q10, Arg2, Clec4n, Il1rn, Rsad2, Il1r2 and downregulation of Ccr6) (Figure-3J). Tet2 is a critical regulator of DNA methylation. It is not surprising that we observed an increased DNA methylation (5mC) level in bone marrow and lung tissue of Tet2^−/− mice (Figure-S10A–C). Tet2 depletion was confirmed by PCR and by decreased Tet2 protein in the lung tissue of Tet2^−/− mice (Figure-S11A–B). Thus, hemopoietic Tet2 depletion is sufficient to increase DNA methylation and exacerbate inflammation and induce PH in mice, providing biological plausibility for the importance of the mutations and pathway perturbations we found in patients.

IL1-β blockade reverses the PH phenotype in Tet2 depleted mice.

Having established that PH development in Tet2 mutated mice is associated with increased inflammation, we assessed the potential therapeutic effect of anti-inflammatory therapy. We treated 5 Tet2^−/− mice (3 males and 2 females) with anti-IL-1β antibody (IP; 10 mg/kg/week; 6weeks) and 5 age/sex-matched Tet2^−/− mice with IgG2a isotype control (IP; 10 mg/kg/week; 6weeks) at the age of PH onset (7 months) assessed by decreased PAAT (Figure-S12A). After 6 weeks of treatments, we first confirmed that animals treated with placebo developed PH evident as increased Fulton index, decreased PAAT, increased RVSP, mPAP, and TPR index (Figure-4A–E and Figure-S12A). Anti-IL-1β antibody treatment prevented weight loss observed in Tet2^−/− mice and improved PH hemodynamics and phenotype parameters to healthy Tet2^f/f, levels (PAAT, RVSP, mPAP, TPRI) (Figure-4A–E). Note that IL1-β antibody treatment induced no kidney or liver toxicity (Table-S7). Our results suggest that targeting inflammation through IL1-β blockade improves PH in Tet2^−/− mice and confirms the contribution of inflammation (especially IL1-β) in the etiology of PH related to Tet2 depletion.

Figure 4. IL-1β blockade reverses PH in Tet2 mutated mice.

10, 7-monthTet2^−/− mice were treated with an antibody against IL-1β (Tet2^−/− + IL-1β ab; IP; 10mg/kg/week; 6weeks) or IgG2a (Tet2^−/− + IL-1β ab IgG2a; IP; 10mg/kg/week; 6 weeks). Tet2^−/− + IL-1β ab mice showed A) significant increased pulmonary artery acceleration time (PAAT); B) decreased right ventricular systolic pressure (RVSP); C) decreased mean pulmonary arterial hypertension (mPAP); D) reduced Total pulmonary resistance (TPR) normalized by body weight associated. E) IL-1β antibody treatment significantly prevents weight loss observed in Tet2^−/− mice treated with IgG2a. n=5 per group. One-way ANOVA; mixed effect ANOVA. Values are expressed as mean±SEM. *P<0.05, **P<0.01, ***P<0.001.

Discussion

We used the largest cohort of PAH patients available to identify TET2 as a novel gene that is mutated in PAH. TET2 mutation, observed in 0.39% of PAH cases (40% APAH, 60% IPAH), is associated with a 6.15-fold increased risk of PAH, relative to the gnomAD control database. TET2 mutations occurred in patients who were confirmed to be free of mutations in established PAH genes. Based on their allele fraction in peripheral blood, 75% of TET2 mutations are predicted to be germline versus 25% somatic. In an independent cohort, a decrease of TET2 expression was found in >86% of APAH and IPAH patients. Finally, we reported that conditional hematopoietic Tet2 knockout is sufficient to induce PH in mice.

Epigenetic mechanisms link genes and environment. They include changes in DNA methylation, histone acetylation and production of micro-RNAs. TET enzymes are key regulators of DNA demethylation, catalyzing the conversion of the methylated nucleotide 5-methylcytosine into 5-hydroxymethylcytosine. By subtraction of methyl groups on DNA and association with histone deacetylases³¹, TET enzymes contribute to the epigenetic regulation of gene expression. Our group and others showed that hyper-methylation of specific target genes contributes to the development of PAH^8,9,32. Supporting our observation, TET2 loss of function leads to increased DNA methylation whilst TET2 hematopoietic depletion results in inflammation and cardiac dysfunction^33,34. We observed increased adverse pulmonary vascular remodeling in the lungs of Tet2^−/− mice (Figure 3D), consistent with previous observations implicating Tet2 depletion in development of atherosclerotic lesions³⁵. We also confirmed that hematopoietic Tet2 depletion increases DNA methylation (Figure-S10) and inflammation¹² evident as macrophage accumulation, and increased expression of inflammatory mediators (Il1b, Cxcr2, Csf3r, Ccr1, Mmp9, Cd33, Itgam, Il1rn, Il1r2) in Tet2^−/− lungs (Figure-3I, J). These factors are known to be associated with the development of PAH³⁶. Validating our observations, we reported a global increase of pro-inflammatory markers in the blood of human PAH patients with TET2 deleterious variants compared to age/sex-matched non-carriers (Figure 2). The fact that this mouse spontaneously develops PAH experimentally links hematopoietic Tet2 inactivation to vascular remodeling and inflammation, two common features of the PAH phenotype³⁶.

We generated conditional hematopoietic Tet2 mutated mice to mimic TET2 deleterious variants in human patients and the ubiquitous decrease in TET2 expression observed in the peripheral blood cells of PAH patients. We showed that homozygous Tet2−/− mice spontaneously developed a PH phenotype and notably manifested a profound loss of the pulmonary microvasculature. Heterozygous Tet2+/− mice have a genotype closer to the human condition described in this article and have a less pronounced phenotype (p<0.05), suggesting gene dose-effect response. This observation experimentally shows that, in an animal model, a total or partial loss of Tet2 function can induce vascular remodeling secondary to increased inflammation.

Age of onset of PAH is significantly higher in patients harboring somatic and germline TET2 variants compared to the remainder of the cohort (66.9 ±10.7 VS 48 ±19 years) (Table 3). Consistent with this observation, older (7 month) Tet2^−/−mice spontaneously developed PH, whilst we did not detect significant PH in younger (2 month) Tet2^−/−mice (Figure 3 and Figure S9). This observation suggests that aging contributes to PH related to TET2 mutations in both humans and mice. We believe a second hit may be required to elicit clinically evident PAH, such as occurs in CHIP and myeloid cancer risk. Indeed, Tet2 normally restrains inflammation, but an initial trigger may be required to induce inflammation. Thus, TET2 depletion exacerbates inflammation when it occurs in a pro-inflammatory environment. Chronic inflammation associated with aging might potentially be such a second hit³⁷ and thereby elicit clinically evident PAH in TET2 deleterious variant carriers.

Mutations in 12 established risk genes and 5 recently-validated risk genes predispose to PAH, with BMPR2 mutations as the most common genetic cause of HPAH and IPAH^7,18. Here we identify an additional novel PAH gene and report that predicted deleterious germline and/or somatic variants of TET2 underlie 0.39% of PAH cases. Our limited read depth, using whole exome data, likely underestimates the prevalence of somatic mutations (i.e. clonal hematopoiesis of indeterminate potential, CHIP). These TET2 mutations are observed in patients who were shown to be free of the other known PAH risk genes variants. Interestingly, TET2 mutations also occur in APAH patients, who are known to have more severe inflammatory states and greater mortality rates than other PAH groups³⁸. Similarly, we reported that TET2 variant carriers exhibit a pro-inflammatory phenotype compared to age/sex-matched, non-carrier, PAH patients. It is possible that patients with APAH might also be inflamed as a result of their connective disease. However, the restricted sample size of this sub-study (n=10 patients with or without TET2 mutations) and the variability in blood cytokine levels amongst patients, required us to perform our analysis on all mutation carriers, not just those with IPAH. Thus, we cannot exclude some confounding effects related to the inclusion of 4 patients with APAH and TET2 mutations in this cohort.

Unlike our study, the NIHR BioResource-Rare Diseases PAH study, which investigated deleterious variation in 1048 IPAH/FPAH patients, did not enroll non-European subjects or APAH patients and did not report deleterious variants in TET2³⁹. Our study has a ~2.5-fold larger sample size, used exome rather than genome sequencing, specifically assessed somatic mutations, and included APAH patients (49% of the 2572 patients). This is the first gene associated with APAH aside from those associated with congenital heart disease.

Over 86% of PAH patients have decreased TET2 expression, suggesting an acquired mechanism of TET2 gene dysregulation in PAH. Consistent with this speculation, hypoxia, as well as metabolic disorders observed in PAH, are also associated with pathologic regulation of TET2 expression⁴⁰. The difference between the rarity of TET2 mutations versus the ubiquitous dysregulation of TET2 expression mirrors what is seen with BMPR2 in PAH. BMPR2 mutations in non-hereditary PAH occur in the minority of subjects whereas BMPR2 mRNA and protein expression is downregulated in most forms of human and experimental PAH^5,6.

TET2 is one of the two most commonly mutated genes in CHIP, which is associated with an increased risk of atherosclerosis and elevated inflammation⁴¹. CHIP mutations lead to elevation of interleukin 1β⁴². The Canakinumab Anti-inflammatory Thrombosis Outcome Study (CANTOS)⁴³, which investigated the effect of a monoclonal antibody targeting interleukin-1β (Canakinumab) in cardiovascular diseases, identified the presence of CHIP (and TET2 mutation) in 8.8% of the cohort. Patients with a somatic mutation in TET2 showed a greater magnitude of risk for major adverse cardiovascular events (MACE); however, they also displayed an improved therapeutic response to canakinumab. We showed that conditional hematopoietic Tet2 depletion is associated with accumulation of IL-1β in the lung of mice (Figure 3J) and reported that IL-1β blockade reverses PH in Tet2^−/− mice (Figure 4). In PAH patients, elevated serum levels of IL-1β correlate with a worse outcome and targeting this cytokine, in a preclinical model of PAH, improved the disease⁴⁴. An IL-1β receptor antagonist is currently in phase 1 clinical trial in PAH (NCT03057028). As observed for CANTOS and in Tet2^−/− mice, our results suggest that genetic investigation of TET2 variants and expression might be relevant to predict the likelihood of benefit from IL-1β based therapy. Thus, discovery of this new gene mutation in PAH may have therapeutic consequences.

CHIP is also a precursor to myeloid neoplasms such as MDS and AML^13,15. Intriguingly, previous studies have shown that PAH is observed in 15–48% of patients with MDS, where TET2 is one of the most common mutated genes⁴⁵. We demonstrated that hematopoietic Tet2 depletion is associated with PAH and pulmonary vascular remodeling in mice. The same mutant mice ultimately develop an MDS-like disease²⁶, which experimentally confirms the link between Tet2 depletion, PAH and MDS; also suggesting a relationship between CHIP and PAH. The moderate severity of pulmonary hypertension reported in our animals (RVSP 30.2 ± 1.3mmHg; Figure-3A) and the milder severity of the hemodynamic derangement in TET2 mutant patients (Table 3) suggests this gene mutation might result in a less symptomatic state in patients, which would be hard to detect without active surveillance. Borderline elevation of systolic PAP (29mmHg at rest) was recently reported in a cohort of 34 MDS patients, lending support to our experimental observations⁴⁶.

Our observations suggest that CHIP and PAH are two manifestations of similar mutations. Whether the manifestation of the mutations will be vascular or hematopoietic (or both), likely reflects a complex interplay of factors including: mutation burden, mutation-related inflammatory consequences, and cellular/tissue distribution of the mutations. The Tet2-knockout mouse is an accepted model of human myeloid neoplasia and our discovery that PAH develops spontaneously as these mice age, indicates that both age and TET2 deficiency may also interact to yield different disease manifestations.

Limitations

Using the largest PAH cohort to date, we reported enrichment of rare deleterious variants for TET2 among 2572 PAH patients. However, we did not investigate the occurrence of TET2 mutations in other forms of pulmonary hypertension. This limitation precludes any conclusion regarding the specificity of the mutation to group 1 PH versus patients with group 2–5 PH. Similarly, as observed for BMPR2 mutation, which is mainly reported in PAH but also observed in cancer⁴⁷ (e.g. gastric, colorectal, breast cancer), we acknowledge that TET2 mutation might not be specific for PAH. Indeed, it is known that somatic mutations in TET2 are common in MDS and AML and can portend poor prognosis⁴⁸. However, none of our patients had known active myeloid neoplasms, and we have provided strong evidence of the contributions of deleterious mutations in TET2 to PAH’s etiology.

Kaasinen and colleagues have recently reported germline TET2 frameshift mutation in a lymphoma family⁴⁹. They did not mark observation of unusual predisposition to atherosclerosis nor abnormal pro-inflammatory cytokine or chemokine expression in this family. However, their study was conducted in a Finnish population which was not investigated in our work. The mutation (c.4500delA) reported by Kaasinen and al, was not observed in the PAH cohort. In their report, TET2 mutation was not associated with significantly decreased TET2 expression in peripheral blood cells. Based on this observation and our results, we can speculate that decrease TET2 expression in peripheral blood cells is required to induce a pro-inflammatory phenotype and cardiovascular disorders.

The PAH Biobank used the REVEAL relaxed criteria for enrollment of patients deemed to have Group 1 disease by their PH specialist. This allowed enrollment of patients with a PCWP ≤18 mmHg, which is the case for patient 12–206⁵⁰. While patient 29–016 had a PCWP of 17mmHg, which is above the conventional cut-off of 15mmHg, as defined by the World Symposium on Pulmonary Hypertension¹, this likely reflects the technical heterogeneity of PCWP measurements, rather than clinical misclassification. This patient’s PAH physician integrated all data to make the best clinical diagnosis.

We provided the hemodynamic data obtained in closest temporal proximity to enrollment in the biobank. In most cases these were the diagnostic hemodynamics, obtained prior to treatment; however in one patient (19–036), the low PVR we reported reflected the fact they were already being treated with two PH-targeted medications. At the time of their diagnosis, prior to PH-targeted treatment, their hemodynamics were typical of PAH (mPAP of 67mmHg and a PVR of 9.96WU).

The use of gnomAD as a control cohort can be consider as a limitation of the study. GnomAD spans 15708 genomes from unrelated individuals sequenced as part of various disease-specific and population genetic studies. GnomAD excludes individuals known to be affected by any severe pediatric disease, as well as their first-degree relatives; however, some individuals with severe disease may still be included in the data set, albeit likely at a frequency equivalent to or lower than that seen in the general population. Thus, the 7509 non-Finnish European gnomAD cohort used in our study likely includes individuals with severe diseases, potentially including hematologic disorders or atherosclerosis. This limitation could explain the 6 patients with TET2 variants observed in the gnomAD cohort. Regardless of this limitation we observed a 5.5-fold enrichment of TET2 deleterious variants in the PAH cohort.

Conclusion.

We identify TET2 as a new PAH-associated gene and highlight its importance as a potential mechanism for the component of PAH pathogenesis resulting from inflammation. The TET2 pathway offers potential new biomarkers and therapeutic targets for PAH therapy, including canakinumab.

Clinical Perspective.

What is new:

• TET2, encoding an epigenetic regulator that demethylates cytosine, is found to be mutated in both idiopathic pulmonary hypertension (IPAH) and associated PAH (APAH)

• TET2 expression is ubiquitously decreased in peripheral blood cells of both IPAH and APAH patients

• TET2 depletion creates a pro-inflammatory phenotype

• We present a new preclinical model of PAH, the Tet2 ^−/− mouse

• Il-1β blockade improves PAH in mice with Tet2 depletion

What are the clinical implications:

• TET2 expression might represent a potential PAH biomarker

• TET2 mutation(s) are a risk factor for PAH.

• The role of TET2 mutations in clonal hematopoiesis of indeterminant potential (CHIP) and PAH suggest a relationship between PAH and hematopoietic disorders.

• TET2 mutation(s) are a potential target for personalized medicine and potential indicator of benefit from anti-inflammatory therapies in PAH.