Smooth Muscle Contact Drives Endothelial Regeneration by BMPR2-Notch1 Mediated Metabolic and Epigenetic Changes

Abstract

Rationale:

Maintaining endothelial cells (EC) as a monolayer in the vessel wall depends on their metabolic state and gene expression profile, features influenced by contact with neighboring cells such as pericytes and smooth muscle cells (SMC). Failure to regenerate a normal EC monolayer in response to injury can result in occlusive neointima formation in diseases such as atherosclerosis and pulmonary arterial hypertension.

Objective:

We investigated the nature and functional importance of contact-dependent communication between SMC and EC to maintain EC integrity.

Methods and Results:

We found that in SMC and EC contact co-cultures, bone morphogenetic protein receptor 2 (BMPR2) is required by both cell types to produce collagen IV to activate integrin-linked kinase. This enzyme directs phospho c-Jun N-terminal kinase (p-JNK) to the EC membrane, where it stabilizes presenilin1 and releases Notch1 intracellular domain (N1ICD) to promote EC proliferation. This response is necessary for EC regeneration following carotid artery injury. It is deficient in EC-SMC Bmpr2 double heterozygous mice in association with reduced collagen IV production, decreased N1ICD and attenuated EC proliferation, but can be rescued by targeting N1ICD to EC. Deletion of EC-Notch1 in transgenic mice worsens hypoxia-induced pulmonary hypertension, in association with impaired EC regenerative function associated with loss of pre-capillary arteries. We further determined that N1ICD maintains EC proliferative capacity by increasing mitochondrial mass and by inducing the phosphofructokinase PFKFB3. ChIP-seq analyses showed that PFKFB3 is required for citrate-dependent histone acetylation (H3K27) at enhancer sites of genes regulated by the acetyl transferase p300, and by N1ICD or the N1ICD target MYC and necessary for EC proliferation and homeostasis.

Conclusions:

Thus, SMC-EC contact is required for activation of Notch1 by BMPR2, to coordinate metabolism with chromatin remodeling of genes that enable EC regeneration, to maintain monolayer integrity and vascular homeostasis in response to injury.

INTRODUCTION

In response to an angiogenic stimulus, Notch1 signaling is required to coordinate metabolism and gene regulation that is critical in the proliferation of stalk cells in contact with the migrating tip cell^{1, 2}. In mature vessels, such as carotid³ and pulmonary⁴ arteries, smooth muscle cells (SMC) can interact with endothelial cells (EC) through the porous internal elastic lamina via myoendothelial junctions. These junctions regulate blood flow and maintain vessel integrity by juxtacrine mechanisms that include transfer of calcium and other ions and induction of Notch signaling that is critical for EC proliferation and monolayer regeneration^{2, 5}. In angiogenesis, Notch1 signaling is essential in regulating the metabolism of stalk cells¹, but the metabolic function of Notch1 in EC of mature vessels is largely unknown.

A variety of stimuli alter cellular metabolism to remodel chromatin and influence expression of genes that determine cell fate and function⁶. This metabolo-epigenetic interaction is established in the pluripotency of stem cells, where glycolysis increases acetyl-CoA required for histone acetylation⁷. How these events are coordinated to regulate a specific gene expression program relevant to vascular homeostasis has not been investigated.

Failure to regenerate a functionally normal EC monolayer can lead to the formation of an occlusive neointima in diseases such as atherosclerosis and pulmonary arterial hypertension (PAH)^{8, 9}. The latter is a life-threatening disease often associated with a mutation or dysfunction of bone morphogenetic protein receptor 2 (BMPR2) that can alter EC metabolism¹⁰, causing both a propensity to apoptosis and loss of factors that control SMC proliferation¹¹. BMPR2 and Notch1 are both required for the formation of normal vascular structures, but little is known about specific mechanisms related to their interaction¹². We therefore hypothesized that SMC-EC contact promotes EC regeneration by activating Notch1 in a BMPR2-dependent manner to coordinate metabolism with chromatin remodeling and gene regulation.

We now report, in SMC-EC contact co-cultures that BMPR2 in both cell types is required for the activation of Notch1, to prime EC for regeneration. Consistent with this, a transgenic mouse heterozygous for Bmpr2 in EC and SMC has impaired Notch1 activation in EC resulting in reduced EC regeneration after carotid artery injury that is rescued by expressing N1ICD in EC. Conditional deletion of Notch1 in EC worsens hypoxia-induced pulmonary hypertension in association with more severe loss of distal precapillary arteries attributed to impaired EC regenerative capacity by Ki67 staining. Notch1 enhances glycolysis and mitochondrial metabolism, and promotes histone acetylation at enhancer binding sites for Notch1 and its target MYC. This results in the expression of genes critical in EC regeneration in response to injury.

METHODS

The authors declare that all supporting data are available within the article and its online supplementary files. RNA-seq and ChIP-seq data have been deposited to GEO and can be accessed at: https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?token=cpohswmsprcblex&acc=GSE89788.

Expanded Methods are available in the Online Supplement.

Cell culture.

Primary human pulmonary artery (PA) EC and SMC were harvested from lungs explanted from hereditary PAH patients with a BMPR2 mutation, or from unused donor lungs. All co-cultures were performed using 1μm pore cell culture inserts for 48h. Monoculture was established by seeding EC on the bottom or on both sides of the insert, whereas in contact co-cultures EC and SMC were seeded on each side of the insert as previously described¹³. Non-contact co-cultures were established by seeding EC on the bottom of an insert after seeding of SMC on the bottom of the cell culture plate (Supplemental Fig. IA shows a schema).

Immunofluorescence.

Cells on culture inserts were fixed by paraformaldehyde after 48h in culture, blocked and incubated with primary antibodies against Ki67, phospho c-Jun N-terminal kinase (p-JNK), caveolin1 (CAV1) and N1ICD.

Immunoblotting.

Western Immunoblotting was performed as previously described¹⁰.

siRNA transfection.

siRNA specific for human BMPR2, fructose-2,6-bisphosphatase 3 (PFKFB3), SLC25A1 (Citrate transporter, CTP) and ColIagen IVA1 and A2 were transfected into EC or SMC by nucleofection or lipofectamine. Knockdown efficacy was determined by immunoblotting or reverse transcription quantitative polymerase chain reaction (RT-qPCR).

ILK activity assay.

Integrin linked kinase (ILK) activity assay was performed using Akt kinase kit as previously described¹⁴.

Glycolysis and mitochondrial function assays.

EC were re-seeded onto an assay plate 48h after mono- or co-culture. Three hours after re-seeding, baseline mitochondrial function and mitochondrial stress response were measured by oxygen consumption rate (OCR) using the Cell Mito Stress Kit, and glycolytic capacity was measured by extracellular acidification rate (ECAR) using the XF Glycolysis Stress Test Kit, with an XF24 extracellular flux analyzer as previously described¹⁰ and the measurements were normalized to cell number.

RT-qPCR.

Total RNA was extracted and purified from trypsinized EC. The quantity and quality of RNA was determined by using a spectrophotometer. RNA was reverse transcribed to cDNA.

mitochondrial DNA (mtDNA) copy number.

DNA was isolated from EC using DNeasy Blood & Tissue Kit, and primers used to amplify and quantify mtDNA related to 18S (nuclear DNA)¹⁰.

Acetyl CoA quantification.

Acetyl CoA was measured by PicoProbe Acetyl CoA Assay Kit following the manufacturer’s instructions.

RNA-seq and analysis.

RNA was extracted from EC with RNeasy Mini Kit. Libraries were prepared with ScriptSeq Complete Gold kit and subjected to sequencing with an Illumina HiSeq 4000.

Chromatin Immunoprecipitation Sequencing (ChIP-seq) and analysis.

EC were cross-linked with 1% formaldehyde. Nuclear pellets were sonicated and immunoprecipitated with antibodies for NOTCH1, H3K27ac and p300. ChIP-DNA and input sample were reverse cross-linked and purified. The illumina TruSeq adapters were ligated and size-selected from the gel before doing PCR amplification. Purified samples were sequenced on HiSeq 4000.

Mouse carotid artery injury.

EC and SMC-specific Bmpr2^+/− mice were created by crossing inducible EC-Bmpr2^−/− and non-inducible SMC-Bmpr2^+/− mice. Tamoxifen was given intraperitoneally at 2mg per day for 10 days. Wild-type (WT) littermates were used as controls. The mice ranged in age from 8–10 weeks at the start of the experiment. The left common carotid artery was injured using a curved flexible wire (0.38mm). A single exposure of adeno associated virus (AAV)-TIE-flag-mN1ICD or AAV-TIE-GFP was delivered to the carotid artery segments. Re-endothelialization of the carotid artery was determined by Evans blue staining 7 days after the injury. To determine intimal area, carotid arteries were harvested 14 days after the injury. Carotid artery sections were stained by hematoxylin and eosin staining or Movat pentachrome staining, or for immunofluorescent imaging, tissue sections were blocked and incubated with primary antibodies against collagen IV, N1ICD, CD31, von Willebrand factor (vWF), flag and Ki67.

Hypoxia-induced pulmonary hypertension in mice.

EC-specific Notch1^−/− mice were created by crossing endothelial SCL-CRE ERT mice with Notch1^fl/fl mice. Tamoxifen was given intraperitoneally at 2mg per day for 10 days. WT littermates were used as controls. The mice were 8 weeks of age at the start of the experiment. EC Notch1^−/− mice were housed in normoxia (21% O₂) or hypoxia (10% O₂) for three weeks. Hemodynamic and lung vascular morphometric data were evaluated as previously described¹⁰.

Statistical analysis.

All data represent mean±standard error of the mean (SEM). Statistical significance was determined by two-way ANOVA or one-way ANOVA followed by Bonferroni’s test, or two-sided unpaired t-test analysis. A p-value of <0.05 was considered significant. For all cell culture experiments, n represents different combinations of EC and SMC from different patients or control subjects.

RESULTS

SMC-EC contact promotes EC proliferation in a BMPR2-dependent manner.

We used PA EC and SMC to investigate how EC function is regulated by contact with SMC, and whether the mechanism is dependent on BMPR2. We harvested EC and SMC from lungs removed at transplant from patients with PAH and a BMPR2 mutation, and from unused donor lungs as controls. Demographic and hemodynamic information is provided in Supplemental Table I. Contact with SMC enhanced EC proliferative capacity judged by Ki67 staining, except when both EC and SMC were mutant for BMPR2 (Fig. 1A) or when BMPR2 was reduced by RNA silencing in both EC and SMC (Fig. 1B). The increase in EC nuclear Ki67 mediated by SMC-EC contact was associated with a decrease in migratory function (Supplemental Fig. IB). EC from both contact and non-contact SMC-EC co-cultures were relatively resistant to apoptosis, independent of BMPR2 (Supplemental Fig. IC). SMC proliferation in response to EC contact was not altered nor BMPR2 dependent (Supplemental Fig. ID).

Figure 1: SMC-EC Contact Promotes EC Proliferation in a BMPR2-Dependent Manner.

(A) EC and SMC from lungs of PAH patients with a BMPR2 mutation (BMPR2mut, M) or from unused donor lungs (control, C), or (B) control cells transfected with non-targeting siRNA (Con) or BMPR2 siRNA (B2) were monocultured or co-cultured for 48h as described in “Methods”. Images of EC nuclei stained by Ki67 antibody (red), with quantification of percent of Ki67 positive nuclei, n=7 in A, and n=6 in B. (C) Images of EC stained with N1ICD antibody (red), with quantification by RFU over total number of nuclei, n=5. (D) Immunoblots of BMPR2, N1ICD and α-Tubulin, with quantification of EC-N1ICD, n=3. Data represent mean±SEM; **p<0.01, ***p<0.001 vs. control or BMPR2mut-EC in monoculture; ^#p<0.05, ^###p<0.001 vs. BMPR2mut- or B2-EC contact co-cultured with control SMC, by two-way ANOVA and Bonferroni’s post-test.

BMPR2-dependent Notch1 activation promotes the EC proliferative phenotype.

As EC from SMC-EC contact co-cultures exhibited a proliferative phenotype similar to those observed in Notch1-dependent regenerating EC⁵, we determined whether BMPR2 might be required for Notch1 activation in EC. Indeed, N1ICD was induced in EC in SMC-EC contact co-cultures except when both SMC and EC had a BMPR2 mutation or when BMPR2 was reduced in both cell types by RNA silencing (Fig. 1C). The N1ICD induction was independent of cell confluency (Supplemental Fig. IE). We also showed an induction of EC-N1ICD, by immunoblotting of EC lysates from SMC-EC co-cultures, that was reduced with BMPR2 silencing in both cell types (Fig. 1D), and present in the cytoplasm and in the nucleus (Supplemental Fig. IF).

Increased N1ICD was associated with an elevation in its targets of transcription, i.e., HEY1, HES1 and MYC¹⁵, and with a decrease in transcripts known to be repressed by Notch signaling, Neuropilin1 (NRP1) and Neuropilin2 (NRP2)^{16, 17} (Fig. 2A). NRP1 and NRP2 are predicted targets of HEY1 and HES1 by Contra software¹⁸. DAPT, a γ-secretase inhibitor, decreased N1ICD in the cytoplasm and nucleus in EC in contact with SMC (Fig. 2B, left), and this attenuated the heightened proliferative capacity of the EC as judged by nuclear Ki67 (Fig. 2B, right). Taken together, our data indicate that BMPR2 is required for SMC-EC contact-dependent activation of EC-Notch1 to promote an EC proliferative phenotype with reduced migratory function through regulation of Notch1 target genes.

Figure 2: BMPR2-Dependent Notch1 Activation Promotes EC Proliferative Phenotype.

Control cells with control non-targeting siRNA (Con) or BMPR2 siRNA (B2) were cultured as in Figure 1. (A) Notch1 target genes (HEY1, HES1 and MYC) and NRP1 and 2 in EC, relative to β-Actin by RT-qPCR, n=3. (B) Left, images of EC stained with N1ICD antibody (red) +DAPT (+) vs. DMSO (−), with quantification by RFU over total number of nuclei, n=3. Right, images of EC nuclei stained by Ki67 antibody (red), +DAPT (+) vs. vehicle (−) with quantification of percent of Ki67 positive nuclei, n=3. (C) Immunoblots for EC BMPR2, presenilin1, and α-Tubulin of EC, with quantification of presenilin1, n=3. (D) Immunoblots for BMPR2, p-JNK, total JNK and α-Tubulin of EC with quantification of p-JNK over total JNK, n=4. (E) Immunoblot for presenilin1, N1ICD and α-Tubulin, +SP600125 (+) vs. DMSO (−) with quantification, n=3. Data represent mean±SEM; *p<0.05, **p<0.01, ***p<0.001 vs. monoculture; ^#p 0.05, ^##p<0.01, ^###p<0.001 vs. B2-EC contact co-cultured with Con-SMC, or vs. contact co-cultured EC with vehicle, by two-way ANOVA and Bonferroni’s post-test.

SMC-EC contact activates Notch1 by compartmentalizing p-JNK to stabilize γ-Secretase.

We next determined whether BMPR2-dependent activation of Notch1 in SMC-EC contact co-cultures was related to an increase in the γ-secretase complex of rate-limiting proteolytic enzymes required for Notch1 activation¹⁹. The protein level of presenilin1, the main component of the γ-secretase complex, was elevated in SMC-EC contact co-cultures, except when BMPR2 was reduced in both cell types (Fig. 2C and Supplemental Fig. IIA). There was also a BMPR2-dependent increase in presenilin2 in SMC-EC contact co-cultures but no significant change in nicastrin or PEN2, the other components of γ-secretase complex (Supplemental Fig. IIB). We did not observe BMPR2-dependent expression of Notch1 ligands such as delta-like 4 (DLL4) or JAG1 in EC (Supplemental Fig. IIC), or in levels of the S2 cleavage enzyme, ADAM10¹⁹ in EC (Supplemental Fig. IID).

Because Presenilin1 mRNA was not altered under these culture conditions (Supplemental Fig. IIE), we determined whether BMPR2 signaling was required to stabilize presenilin1. We noted a striking increase in p-JNK, a known presenilin1 activator20, in SMC-EC contact and non-contact co-cultures compared to monocultures. This response was attenuated by BMPR2 silencing in both cell types in contact co-cultures and also with BMPR2 silencing in EC in non-contact co-cultures (Fig. 2D). However only in contact co-cultures was EC p-JNK associated with an increase in presenilin1 or in activation of Notch1 (Supplemental Fig. IE, IIA, IIIA). Moreover, a selective p-JNK inhibitor, SP600125, reduced EC presenilin1 protein not mRNA levels and decreased N1ICD levels in SMC-EC contact co-cultures (Fig. 2E and Supplemental Fig. IIE). There was no SMC-EC contact associated elevation in p-SMAD (Supplemental Fig. IIF), p-ERK, p-AKT and p-p38 (data not shown). We therefore determined whether the activation of presenilin1 was related to the cellular localization of p-JNK induced by SMC-EC contact.

Immunofluorescent imaging and immunoblotting (Fig. 3A) demonstrated that p-JNK localized mostly to the cytoplasm in SMC-EC contact co-cultures, whereas in non-contact co-cultures, it was mostly observed in the nucleus. CAV1 forms a complex with p-JNK and is a known regulator of the mitogen-activated protein kinase pathway²¹. Indeed, immunofluorescent imaging showed co-localization of CAV1 and p-JNK at the EC periphery (Fig. 3A), and co-immunoprecipitation demonstrated a complex between CAV1 and p-JNK that was increased under conditions of SMC-EC contact compared to non-contact co-cultures (Fig. 3A, right). p-JNK compartmentalization was downstream of CAV1 localization, as the p-JNK inhibitor SP600125 did not change CAV1 localization (Supplemental Fig. IIIB). Addition, of cycloheximide supported pJNK mediated stabilization of presenilin (Supplemental Fig. IIIC). Taken together, our data are consistent with previous studies showing that CAV1 compartmentalizes the γ-secretase complex and that p-JNK is required to increase presenilin1²⁰ inducing Notch1 activation²².

Figure 3: SMC-EC Contact Activates Notch1 by Compartmentalizing p-JNK to Stabilize γ-Secretase.

(A, B) Control EC and SMC were co-cultured ±contact or ±ILK inhibitor. (A) Images of EC stained by p-JNK (green) and CAV1 (red) antibodies. Low magnification (top), high magnification (middle) and expanded images (bottom). Immunoblots for p-JNK, total JNK, α-Tubulin and Lamin B1 of fractionated lysates of EC, with quantification of cytoplasmic p-JNK relative to α-Tubulin and nuclear p-JNK relative to Lamin B1 (center). On the right, immunoblots for total JNK, p-JNK, CAV1 and actin of EC input lysate and EC lysate immunoprecipitated with p-JNK antibody or normal IgG (far right), with quantification of co-immunoprecipitated CAV1 relative to p-JNK. (B) Immunoblots of ILK and ILK activity, measured by the ratio of p-GSK3α/β to immunoprecipitated ILK. Right: Control EC and SMC were co-cultured with DMSO or ILK inhibitor. Images of EC stained by p-JNK (green) and CAV1 (red) antibodies. Low magnification images (top), high magnification (bottom). Far right: Immunoblots for presenilin1 and α-Tubulin +ILK inhibitor (+) vs. DMSO (−) with quantification. Data represent mean±SEM, n=3; *p<0.05, **p<0.01, ***p<0.001 by t-test. (C) Control EC contact co-cultured with EC or SMC with non-targeting siRNA vs. BMPR2 siRNA (Con vs. B2). Immunoblots for Collagen IV and α-Tubulin with quantification. Data represent mean±SEM, n=3; ***p<0.001 vs. co-culture of EC with EC; ^#p<0.05 vs. contact co-culture of B2-EC with Con-SMC, by two-way ANOVA with Bonferroni’s post-test. (D) Immunoblots of EC-SMC contract co-culture input proteins in EC and SMC with non-targeting or Collagen IV siRNA (Con vs. Collagen IV). ILK activity (right) in EC contact co-cultured with SMC with quantification. Immunoblots of Collagen IV, ILK and α-Tubulin in the input lysate on the left with immunoblot for Collagen IV in SMC on the bottom. Data represent mean±SEM, n=3; *p<0.05 by t-test. (E) Proposed model: upon cellular contact, BMPR2 in SMC and EC contact through internal elastic lamina (IEL) mediates Collagen IV production. Collagen IV induces ILK activity and promotes p-JNK localization to the cytoplasmic membrane with CAV1, p-JNK compartmentalization to the caveolae, and stabilization of presenilin1 to enhance Notch1 activation upon ligand binding.

To investigate how BMPR2 mediated the p-JNK and CAV1 interaction, ILK was assessed as a potential modifier of CAV1 localization²². Indeed, ILK activity was elevated in EC in contact with SMC compared to non-contact co-cultures (Fig. 3B). Blocking ILK re-distributed p-JNK to the nucleus, resulting in reduced presenilin1 (Fig. 3B, right and far right). Collagen IV is an activator of ILK²³ and a BMPR2-dependent component of the extracellular matrix²⁴. We showed an increase in collagen IV deposition in SMC-EC contact co-cultures that was reduced by loss of BMPR2 in both cell types (Fig. 3C), and related to reduced expression of cytoplasmic CAV1 and p-JNK (Supplemental Fig. IIID and IIIE). Furthermore, when we reduced collagen IV using RNA silencing in both EC and SMC in contact co-cultures, ILK activity was attenuated (Fig. 3D).

Additional studies indicated that SMC-EC contact co-cultures establish myoendothelial junctions (Supplemental Fig. IIIF), and not only stimulate EC deposition of collagen IV but also allow EC with loss of BMPR2 to respond to the increase in SMC production of collagen IV (Supplemental Fig. IIIG). Collectively, our data as summarized in Fig. 3E, show that BMPR2 in SMC-EC contact co-cultures is responsible for the accumulation of collagen IV required for EC ILK activity; ILK activity promotes CAV1-p-JNK localization to the cytoplasmic membrane, stabilizing presenilin1 and resulting in activation of Notch1.

BMPR2 activation of Notch1 promotes EC regeneration following carotid artery injury.

Our findings suggested that SMC induction of the EC proliferative phenotype might be functionally important in EC regeneration. We confirmed, in contact co-cultures of aortic EC and SMC, that loss of BMPR2 in both cell types leads to reduced Notch1 activation (Supplemental Fig. IV) as was seen in PA cells. Others have shown that Notch activation is important in EC regeneration in systemic arteries in response to injury^{5, 25–27} although this can be context-dependent. It has also been shown that like the PA⁴, the carotid artery³ also has myoendothelial junctions. We therefore extended our findings to mice double heterozygous for a conditional Bmpr2 deletion in EC and SMC (EC-SMC Bmpr2^+/−) in which we induced de-endothelialization of the carotid artery by a wire-induced injury. In EC-SMC Bmpr2^+/− mice compared to WT littermates, we observed a decrease in collagen IV deposition in the sub-endothelium, associated with reduced Notch1 activation (Fig. 4A), recapitulating the cell culture studies. We then compared early re-endothelialization in WT mice and in EC-SMC Bmpr2^+/− mice, and determined whether re-endothelialization and vessel remodeling could be restored by delivering the N1ICD to the EC. We reconstituted N1ICD specifically in EC by applying an AAV-tagged N1ICD construct driven by the Tie1 promoter to the adventitial surface of the carotid artery at the time of injury, and confirmed EC-selective N1ICD expression (Fig. 4B). Reconstitution of N1ICD rescued the delayed re-endothelialization in EC-SMC Bmpr2^+/− mice (Fig. 4C), related to decreased Ki67 positive CD31⁺ cells at Day 7 (Fig. 4D). In association with impaired re-endothelialization of the carotid artery in the EC-SMC Bmpr2^+/− mice, we also observed abnormal arterial remodeling, namely neointimal thickening and medial hypertrophy (Fig. 4E). Thus, BMPR2-dependent Notch1 activation is required for EC proliferation and regeneration in response to EC injury, to prevent abnormal remodeling.

Figure 4: BMPR2 Activation of Notch1 Promotes EC Regeneration Following Carotid Artery Injury.

(A) Top: Images of left carotid artery from EC-SMC Bmpr2^+/− or WT mice stained by CD31 (red) and collagen IV antibody (green), with quantification of collagen IV by RFU in intima + media, n=4 mutant and 3 WT. Bottom: Images of left carotid artery from EC-SMC Bmpr2^+/− or WT mice stained by N1ICD (red) and CD31 antibody (green) with quantification of N1ICD by RFU in CD31 positive cells, n=3. Data represent mean±SEM; *p<0.05 by t-test. (B-E) Carotid arteries of EC-SMC Bmpr2^+/− mice were injured using a wire and exposed to AAV expressing GFP (AAV-GFP) or flag-N1ICD (AAV-N1ICD) directed to EC by Tie1 promoter, as described in “Methods”. (B) Images of carotid artery from EC-SMC Bmpr2^+/− or WT mice stained by flag antibody (red) and vWF antibody (green) 14 days after carotid artery injury and AAV-N1ICD or AAV-GFP exposure. (C) Images of Evans blue staining of the left carotid arteries 7 days after carotid artery injury and AAV-N1ICD or AAV-GFP exposure (WT; n=4, EC-SMC Bmpr2^+/−; n=6), with quantification of re-endothelialized area as % of unstained area. (D) Images of the left carotid arteries stained by Ki67 antibody (red) at 7 days after carotid artery injury and AAV-N1ICD or AAV-GFP exposure (WT; n=5, EC-SMC Bmpr2^+/−; n=6, respectively), with quantification of percent of Ki67 positive nuclei of total nuclei in CD31 positive cells. (E) Images of Hematoxylin and Eosin (H&E) and Movat pentachrome staining of left carotid arteries of mice treated with AAV-GFP or AAV-N1ICD at 14 days, with quantification of intimal area, medial area, external elastic lamina (EEL) length, and intimal/medial area, n=5. Data represent mean±SEM, *p<0.05, **p<0.01, ***p<0.001 vs. WT mice and ^#p<0.05, ^##p<0.01, ^###p<0.001 vs. AAV-GFP treated group, by two-way ANOVA and Bonferroni’s post-test.

Notch1 promotes EC proliferation and pulmonary arterial integrity in response to injury.

We extended our findings to experimental pulmonary hypertension where impaired EC regeneration in response to injury could lead to loss of precapillary arteries and enhanced muscularization of the remaining vessels. It was previously established that Bmpr2 heterozygous mice develop more severe pulmonary hypertension in response to inflammation²⁸ or hypoxia and serotonin²⁹. As proof of concept, we conditionally deleted Notch1 in EC (EC Notch1^−/−) that was normally observed in unmanipulated human or mouse artery EC (Supplemental Fig. VA). Indeed, in EC Notch1^−/− vs. WT mice, pulmonary hypertension, induced after three weeks of chronic hypoxia (10% O₂), was more severe, as judged by right ventricular systolic pressure (RVSP), right ventricular hypertrophy (Fig. 5A and Supplemental Fig. VB), and echocardiography (reduced PA acceleration time, PAAT, and dilated main PA) (Supplemental Fig. VC). Lung histology revealed a more severe reduction in distal PAs (Fig. 5B) and a greater increase in muscularized arteries in EC Notch1^−/− mice vs. WT mice (Fig. 5C) associated with decreased Ki67 positive CD31⁺ cells and increased Ki67 positive medial SMC (Fig. 5D). Heart rate, left ventricular systolic function and cardiac output by echocardiography were comparable in both genotypes (Supplemental Fig. VC). Taken together, our results in both animal models indicate that EC-Notch1 is important in preventing pathological remodeling.

Figure 5: Notch1 Promotes EC Proliferation and Reduced Hypoxic Pulmonary Hypertension.

Mice with Notch1 deleted in EC (EC Notch1^−/−) and WT littermate were exposed to three weeks of hypoxia (Hx) or normoxia (Nx). (A) RVSP (left) and the ratio of the weight of the right ventricle to the weight of the left ventricle and septum (RV/LV+S, right). Nx-WT (n=7), Nx-EC Notch1^−/− (n=6), Hx-WT (n=8), Hx-EC Notch1^−/− (n=6). (B) The number of distal pulmonary arteries (DPA) per 100 alveoli in sections. (C) Images stained by alpha smooth muscle actin (αSMA, red) and vWF (green) in sections from the EC Notch1^−/− and WT mice under Hx or Nx, with quantification of percent muscularized DPA of total DPA (Full, fully muscularized; Partial, partially muscularized; Non, non-muscularized). Nx-WT (n=7), Nx- EC Notch1^−/− (n=6), Hx-WT (n=9), Hx-EC Notch1^−/− (n=7). (D) Images of DPAs stained by Ki67 antibody (red) with quantification of percent of Ki67 positive nuclei relative to total nuclei in CD31 positive cells or medial smooth muscle cells. Nx-WT (n=7), Nx- EC Notch1^−/− (n=6), Hx-WT (n=8), Hx-EC Notch1^−/− (n=7). Data represent mean±SEM, *p<0.05, **p<0.01, ***p<0.001 vs. the same strain in Nx, and ^#p<0.05, ^###p<0.001 vs. WT mice, same condition (Nx or Hx), by two-way ANOVA and Bonferroni’s post-test.

BMPR2-dependent Notch1 mediates EC glycolysis and oxidative phosphorylation.

Since EC metabolism is essential in determining EC fate^{1, 30}, we further investigated the metabolic characteristics of the EC in the context of SMC-EC contact. SMC-EC contact co-cultures were characterized by BMPR2-dependent elevation in glycolysis, measured from the extracellular acidification rate (ECAR) associated with an increase in oxygen consumption rate (OCR) (Fig. 6A). We showed by qPCR and immunoblotting that an essential inducer of glycolysis in EC, PFKFB3¹, was elevated in EC from SMC-EC contact co-cultures in a BMPR2- (Fig. 6B) and Notch-dependent manner (Fig. 6C). Increased PFKFB3 was responsible for the increase in ECAR and maximum OCR as these were attenuated by PFKFB3 silencing (Supplemental Fig. VIA, B). Hexokinase II, another glycolysis related enzyme that could regulate EC function³¹, did not show a BMPR2 dependent increase (Supplemental Fig. VIC).

Figure 6: BMPR2 Mediating Notch1 Activation Promotes EC Glycolysis and Oxidative Phosphorylation.

(A, B) Control EC were monocultured or contact co-cultured with SMC ±BMPR2 silencing (Con vs. B2). ( A) ECAR under baseline conditions and in response to glucose (Glu), oligomycin (Oligo) and 2-deoxy glucose (2DG), n=3 (left) and OCR in response to oligomycin (Oligo), fluoro-carbonyl cyanide phenylhydrazone (FCCP) and rotenone with antimycin A (R/AA), n=4 (right). The measurements were normalized to cell number. (B) EC PFKFB3 mRNA relative to β-Actin by RT-qPCR, n=3 (left). Immunoblots for BMPR2, PFKFB3 and α-Tubulin of EC with quantification, n=4 (right). Data represent mean±SEM; *p<0.05, **p<0.01, ***p<0.001 vs. monoculture; ^#p<0.05, ^##p<0.01, ^###p<0.001 vs. contact co-culture with control EC and SMC, by two-way ANOVA and Bonferroni’s post-test. (C) EC ±SMC contact co-cultures were pre-treated +DAPT (+) or vehicle (−). PFKFB3 mRNA of EC relative to β-Actin by RT-qPCR, n=3 (left). Immunoblots for PFKFB3 and α-Tubulin of EC with quantification, n=3 (right). Data represent mean±SEM; *p<0.05, ***p<0.001 vs. monoculture; ^#p<0.05, ^###p<0.001 vs. vehicle by two-way ANOVA and Bonferroni’s post-test.

We also determined that SMC-EC contact induced an increase in EC mitochondrial mass assessed by mtDNA and Tom 20 (Supplemental Fig. VID, E) that was BMPR2 as well as Notch dependent (Supplemental Fig. VIF, G) but not PFKFB3-dependent (Supplemental Fig. VIH). We further showed co-distribution of mitochondria with N1ICD by confocal microscopy (Supplemental Fig. VII). ChIP-seq for EC-Notch1 showed a peak on the d-loop region of mtDNA (Supplemental Fig. VIJ), consistent with previous reports³², and with the transcription of mtDNA-encoding cytochrome c oxidase subunit 1 (COX1) (Supplemental Fig. VIK). Taken together, these results indicate that BMPR2-mediated activation of Notch1 in EC enhances EC glucose metabolism by transcription of PFKFB3 to induce glycolysis, accompanied by an N1ICD-dependent increase in mitochondrial replication and transcription to enhance mitochondrial function and ATP production.

PFKFB3 coordinates histone acetylation with transcription of Notch1 targets.

We posited that PFKFB3, by enhancing glycolysis, could induce acetylation of histones required for transcription of Notch1 target genes. Indeed, we found enhanced H3K9ac and H3K27ac in EC from SMC-EC contact co-cultures compared to monoculture, that was abrogated by PFKFB3 silencing (Fig. 7A). Moreover, the attenuation of histone acetylation that resulted from loss of PFKFB3 was also associated with reduced expression of Notch1 targets HES1, HEY1 and MYC, and reversed repression of NRP1 and 2 (Fig. 7B). Accordingly, the proliferative capacity of EC was decreased (Fig. 7C).

Figure 7: PFKFB3 Coordinates Histone Acetylation with Transcription of Notch1 Targets.

Control EC with non-targeting vs. PFKFB3 siRNA (Con vs. P3) or CTP siRNA (Con vs. CTP) were monocultured or co-cultured in contact with control SMC. (A) Immunoblots for PFKFB3, acetylation marks H3K27ac and H3K9ac, Histone3 and α-Tubulin of EC, with quantifications. n=3. (B) mRNA of PFKFB3, Notch1 targets (HEY1, HES1, MYC) and NRP1 and 2 of EC relative to β-Actin by RT-qPCR, n=3 for MYC and n=4 for other targets. (C) Images of EC nuclei stained by Ki67 antibody (red), with quantification of the percent of Ki67 positive nuclei, n=3. (D) Acetyl-CoA levels per 3 million ECs, n=4. (E) Immunoblots for CTP, PFKFB3, H3K9ac, H3K27ac, Histone3 and α-Tubulin of EC monocultured or contact co-cultured with SMC and non-targeting vs. CTP siRNA (Con vs. CTP), with quantification, n=4. Data represent mean±SEM; *p<0.05, **p<0.01, ***p<0.001 vs. monoculture; ^#p<0.05, ^##p<0.01, ^###p<0.001 vs. non-targeting siRNA, by two-way ANOVA and Bonferroni’s post-test (A-C and E) or t-test (D).

We posited that PFKFB3 might promote histone acetylation by increasing acetyl-CoA via mitochondria-mediated citrate transport, as previously described in the differentiation of embryonic stem cells or 293T cells^{7, 33}. Indeed, PFKFB3 silencing reduced the production of acetyl-CoA (Figure 7D), and in line with this finding, silencing the citrate transporter (CTP) with siRNA or with a CTP inhibitor (CTP-I) abrogated the increase in histone acetylation marks downstream of PFKFB3 (Fig. 7E and Supplemental Fig. VIIA). CTP levels were also PFKFB3-dependent, but inhibition of CTP did not impact PFKFB3 (Fig. 7E, Supplemental Fig. VIIB), suggesting that an increase in the citrate pool, downstream of PFKFB3, is necessary for acetylating histones. ChIP-seq for EC Notch1 and H3K27ac demonstrated PFKFB3 independent Notch1 and H3K27ac peaks at the PFKFB3 promoter region (Supplemental Fig. VIIC). Thus, Notch1 activation increases PFKFB3 to promote glycolysis, thereby increasing the citrate pool required for production of acetyl-CoA and acetylation of histones that could be required for regulation of Notch1 target genes.

PFKFB3-dependent histone acetylation of Notch1 and MYC targets.

We next investigated whether histone acetylation alters expression of genes associated with Notch1 via its DNA binding partner RBPJ³⁴. We performed H3K27ac ChIP-seq using EC derived from SMC-EC contact co-cultures with or without PFKFB3 silencing, and showed a reduction in H3K27ac density around Notch1 peaks (Fig. 8A) with loss of PFKFB3. There was no overall decrease in Notch1 peaks (i.e., RBPJ motif peaks) with PFKFB3 silencing (Supplemental Fig. VIIIA, B and Supplemental Table IIA). Most of these peaks were distributed on distal intergenic regions (Supplemental Fig. VIIIC), likely reflecting Notch1 function related to distal enhancers. However, there were fewer H3K27ac ChIP-seq peaks under PFKFB3 silencing (Supplemental Fig. VIIID and Supplemental Table IIB). Of the 148 Notch1 target genes that overlapped with H3K27ac peaks in a PFKFB3-dependent manner (Venn diagram in Fig. 8A, red-circle), the top downstream pathway was cellular growth and proliferation and included MYC (Fig. 8A, right, and Supplemental Table III).

Figure 8: EC Glucose Metabolism Impacts Epigenetics to Promote EC Proliferation.

Control EC were transfected with non-targeting or PFKFB3 siRNA (siControl vs. siPFKFB3) and co-cultured with control SMC, and ChIP-seq of EC for Notch1, H3K27ac and p300, and RNA-seq of EC were carried out. (A) Left, heatmaps of H3K27ac density ±1kb around the Notch1 peak. Center, Venn diagram with genes from Notch1 ChIP-seq peaks and H3K27ac ChIP-seq peaks and siControl vs. siPFKFB3. Red and blue-circled genes were picked for the analysis that follows. Right, pathway analysis of 148 genes from Notch1 and H3K27ac ChIP-seq under siControl vs. siPFKFB3 (red-circled in Venn diagram). Threshold of significance was set at p<0.001. (B) Venn diagram of genes from H3K27ac ChIP-seq independent of Notch1, differentially annotated vs. siPFKFB3 (blue; blue-circled in Figure 8A) and differentially expressed genes from RNA-seq with 100 FPKM change in control non-targeting vs. siPFKFB3 (red). Center, pathway analysis of 190 overlapping genes between H3K27ac ChIP-seq genes and RNA-seq genes under the control vs. siPFKFB3. Threshold of significance was set at p<0.001. On the right, validation of mRNA expression relative to 18S by RT-qPCR, of five genes in the pathway of cellular growth and proliferation. Bars represent mean±SEM, n=3; **p<0.01 by t-test. (C) Gene set enrichment analysis of hallmark MYC targets with differentially expressed genes derived from RNA-seq below FDR<0.05 in control non-targeting vs. siPFKFB3. (D) Peak plots of p300 ChIP-seq ±3kb beyond the p300 summit and heatmaps of H3K27ac density ±1 kb around the p300 Summit with loci of genes described in B. (E) Proposed model: Notch1 activation promotes PFKFB3 mediated glucose metabolism, and increases acetyl-CoA generated from citrate. p300 mediates transfer of acetylation from acetyl-CoA, to increase histone acetylation. This orchestrates expression of Notch1 targets including MYC as well as multiple MYC target genes, promoting the EC proliferative phenotype. This is coordinated with heightened energy production by the mitochondria.

We found 6,975 PFKFB3-dependent H3K27ac peaks that did not overlap with Notch1 binding sites (Fig. 8A Venn diagram, blue-circled). We overlaid these targets with 6,755 differentially expressed genes by RNA-seq under control vs. PFKFB3 silencing conditions (Supplemental Table IV). We then analyzed the overlapping genes most differentially expressed under control vs. PFKFB3 silencing conditions (>100-fold change) in fragments per kilobase of exon per million fragments mapped (FPKM) (Fig. 8B). Once again, we found cellular growth and proliferation as the top pathway represented (Fig. 8B, center, and Supplemental Table V). Also prominent by pathway analysis was the free radical scavenging pathway that is necessary to reduce oxidative stress that results from heightened mitochondrial respiration (Fig. 8B, center). Genes previously established as important in EC homeostasis and proliferation that included APLN, HMOX1, NOS3, ENG and ITGB5, were all validated by qPCR analysis (Fig. 8B, right). Interestingly, these genes are MYC targets of transcription. Gene set enrichment analysis of transcripts differentially expressed under control vs. PFKFB3 silencing conditions also demonstrated enrichment of MYC targets (Fig. 8C).

To better define the mechanism by which PFKFB3 promotes histone acetylation at specific gene targets, we assessed DNA binding of the acetyl transferase p300, a component of the Notch1 transcriptional complex and a main regulator of H3K27ac³⁴. ChIP-seq of p300 showed comparable peaks under control and PFKFB3 silencing conditions (Fig. 8D and Supplemental Table IIC), but H3K27ac density around p300 summits was largely abrogated by PFKFB3 silencing. Accordingly, the overlap between p300 and H3K27ac peaks under siControl conditions was considerably diminished by PFKFB3 silencing (97.4%, vs. 59.7%), (Fig. 8D and Supplemental Fig. VIIIE).

As examples, we show that Notch1 and p300 peaks related to HES1 transcription do not change with PFKFB3 silencing, but the H3K27ac marks are greatly diminished (Supplemental Fig. VIIIF). Notch1 binding to a known distal MYC enhancer³⁵, as well as H3K27ac marks, are both diminished with loss of PFKFB3, with little impact on p300 at this site. On the other hand, Notch1 binding to a known Hey1 enhancer was enhanced with loss of PFKFB3, but H3K27ac marks are greatly diminished with minimal changes in p300 peaks. These results are reflected in Hey1, HES1 and MYC gene expression (Fig. 7B), and MYC levels (Supplemental Fig. VIIIG). Thus Notch1-PFKFB3 mediated glycolysis increases the acetylation status of H3K27 by transfer of acetylation from acetyl CoA via p300 at Notch1 and MYC enhancers of genes that promote EC proliferation. This effect is co-regulated with Notch1-mediated heightened mitochondrial replication, transcription and energy production (Schema in Fig. 8E).

DISCUSSION

Here we show that contact-mediated communication between SMC and EC activates EC-Notch1, and alters the epigenome to regulate Notch1-dependent genes that maintain endothelial integrity and prevent adverse vascular remodeling in response to injury. BMPR2, a gene that protects against experimentally induced atherosclerosis³⁶ and is deficient or mutant in PAH³⁷, orchestrates this function. The mechanism involves BMPR2-dependent production of collagen IV, and subsequent activation in EC of ILK required to target the signaling molecule p-JNK to CAV1, thereby stabilizing presenilin1 and activating Notch1. These observations reinforce the functional importance of collagen IV in the preservation of EC integrity, and could explain why it is a locus for coronary artery disease as recognized by GWAS studies³⁸. Our results add to many others that show how the extracellular matrix regulates intracellular signaling, and how its disruption leads to vascular diseases^{39, 40}. We demonstrate that Notch1 expands mitochondrial mass and production of energy required by actively proliferating cells. Notch1 also induces PFKFB3 to produce acetyl-CoA, necessary to acetylate histones at p300-Notch1 and p300-MYC-sites to mediate transcription of genes related to EC proliferation and homeostasis.

In capillary angiogenesis, EC tip cells rely on glycolytic metabolism and minimal glucose oxidation to promote their sprouting migratory phenotype¹, whereas both enhanced glycolysis and oxidative phosphorylation are required for proliferation of stalk cells³⁰. While the sprouting of tip cells depends on PFKFB3 that is relatively decreased in stalk cells, PFKFB3 levels are still elevated when compared to quiescent EC¹. Angiogenesis studies indicate that FOXO1 regulates metabolic activity and gene expression and both are required for EC proliferation³⁰. Here we show that under conditions of SMC-EC contact, Notch1 mediates a profound increase in glycolysis and mitochondrial mass that accounts for the major increase in energy production.

Our work further demonstrates that the PFKFB3-mediated increase in the citrate pool provides acetyl-CoA for histone acetylation^{33, 41}. Although MYC in stalk cells is a regulator of glucose metabolism³⁰, our findings indicate that MYC gene upregulation is downstream of enhanced glucose metabolism, and promotes expression of genes necessary for EC proliferation, e.g., APLN, HMOX1 and NOS3. Upregulation of mitochondrial ROS scavenging genes also suggest resistance to oxidative damage.

It is interesting that in some cases Notch1-mediated gene regulation was unrelated to PFKFB3, while in other cases Notch1 interacted with other targets in response to loss of PFKFB3. The function of Notch1 in binding to these other targets remains to be understood. However, the increased co-distribution of Notch1, H3K27ac and p300 at active enhancer sites is consistent with a transcriptionally active complex³⁴ in the presence of elevated PFKFB3. Notch1 is a highly acetylated protein and acetylation can stabilize Notch1 to regulate its transcriptional activity⁴².

We show that endothelial regeneration following severe EC injury is impaired in the carotid artery of the EC-SMC Bmpr2^+/− mouse and results in adverse vessel remodeling, highlighting the importance of BMPR2 in preventing systemic arterial diseases³⁶. Furthermore, we show that deleting EC-Notch1 in mice causes more severe PH during hypoxia with loss of distal intra-acinar PAs and increased muscularity of remaining ones. We attribute this to EC monolayer vulnerability and to the loss of factors, such as APLN, that repress SMC proliferation¹¹.

While we describe a novel mechanism by which BMP receptor can activate Notch signaling, previous reports indicate different levels of co-regulation. For example, Smad 1/5, a canonical can form a transcriptionally active complex with the N1ICD and synergistically activate Notch target genes^{43, 44}. BMPR signaling can also upregulate Notch ligand gene expression⁴⁵.

A DLL4 neutralizing monoclonal antibody that blocks the Notch1 pathway induces dysregulation of tumor vasculature but also causes severe pulmonary hypertension as a major adverse effect⁴⁶. Notch3 has been implicated in the proliferation of PA SMC⁴⁷ and a γ-secretase inhibitor attenuated progressive pulmonary hypertension in a rat model⁴⁸, perhaps in part by suppressing Notch3 repression of target genes of N1ICD⁴⁹. Our study using human cells from PAH patients and animal models of disease indicates that, in the context of loss of BMPR2 function, a compound that activates Notch1 in EC may produce the best result and may also be beneficial in systemic vascular disease.

NOVELTY AND SIGNIFICANCE.

What Is Known?

• Atherosclerosis and pulmonary arterial hypertension are associated with alterations in bone morphogenetic protein receptor 2 (BMPR2) signaling that result in endothelial cell dysfunction and dysregulation of cellular metabolism.

• During angiogenesis, Notch1 signaling regulates EC metabolism, EC proliferation, and monolayer regeneration

• The functional importance of the interactions between BMPR2 and Notch1 in the regeneration of EC and in coordinating metabolism, chromatin remodeling and gene regulation, has not been investigated.

What New Information Does This Article Contribute?

• The contact between smooth muscle cell (SMC) and endothelial cells promotes BMPR2 mediated Notch1 activation by a mechanism that requires the production of collagen IV to compartmentalize phospho-c-Jun amino terminal kinase (pJNK).

• Notch1 activation increases mitochondrial mass and oxidative phosphorylation in endothelial cells, and, via fructose-2,6-bisphosphatase 3 (PFKFB3), glycolysis. This generates acetyl-CoA that is required for histone acetylation (H3K27ac) at enhancer sites of genes critical for proliferation.

• Transgenic mice heterozygous for Bmpr2 in EC and SMC have impaired Notch1 activation. These mice show a decrease in re-endothelialization after carotid artery injury, which is prevented by the reconstitution of Notch1 intracellular domain (N1ICD) in the endothelium. Transgenic mice lacking endothelial Notch1 have increased loss of precapillary arteries and pulmonary hypertension after exposure to hypoxia.

• In the context of loss of BMPR2 function, activation of Notch1 in endothelial cells can be of therapeutic importance in pulmonary and systemic vascular disease.

Notch-mediated endothelial metabolism and gene regulation play an important role in angiogenesis, and during development Notch interacts with BMPR2. However, it is unclear how these two factors affect homeostasis of the endothelial cells in mature blood vessels. We now report that contact between SMC and endothelial cells is required for BMPR2-mediated Notch1 activation in the endothelium. The mechanism involves BMPR2 dependent deposition of collagen IV and compartmentalization of pJNK to increase presenilin, required for Notch 1 activation. In endothelial cells, Notch1 enhances both oxidative phosphorylation and glycolysis by increasing mitochondrial mass and inducing PFKFB3 to maintain proliferative capacity. PFKFB3 increases acetyl-CoA to mediate histone acetylation at enhancer sites of proliferative genes regulated by Notch1. Following injury, the regeneration of the endothelial cells of the carotid artery is impaired in mice with compound heterozygosity for Bmpr2 in endothelial cells and SMC that exhibit attenuated EC-Notch1 activation and impaired EC proliferative capacity. This is prevented by endothelial reconstitution of the NOTCH1 intracellular domain. Deletion of endothelial Notch1 worsens hypoxia-induced loss of pre-capillary pulmonary arteries and pulmonary hypertension. These findings suggest that the contact between SMC and endothelial cells is required for BMPR2-mediated activation of Notch1 to coordinate metabolism with chromatin remodeling to enable EC regeneration in response to injury.

Nonstandard Abbreviations and Acronyms:

EC

Endothelial cells

SMC

Smooth muscle cells

PA

Pulmonary artery

PAH

Pulmonary arterial Hypertension

BMPR2

Bone morphogenetic protein receptor 2

N1ICD

Notch1 intracellular domain

DLL4

Delta like ligand 4

CAV1

Caveolin 1

PFKFB3

Fructose-2,6-bisphosphatase 3

RT-qPCR

Reverse transcription quantitative polymerase chain reaction

NRP

Neuropilin

ILK

Integrin linked kinase

ECAR

Extracellular acidification rate

OCR

Oxygen consumption rate

ChIP-seq

Chromatin immunoprecipitation sequencing

AAV

Adeno associated virus

vWF

von Willebrand factor

αSMA

Alpha smooth muscle actin

RVSP

Right ventricular systolic pressure

PAAT

Pulmonary artery acceleration time

COX1 or MT-CO1

Cytochrome c oxidase subunit I

CTP

Citrate transporter

FPKM

Fragments per kilobase of exon per million fragments mapped