Fatty acid metabolism promotes TRPV4 activity in lung microvascular endothelial cells in pulmonary arterial hypertension

Abstract

Pulmonary arterial hypertension (PAH) is a morbid disease characterized by significant lung endothelial cell (EC) dysfunction. Prior work has shown that microvascular endothelial cells (MVECs) isolated from animals with experimental PAH and patients with PAH exhibit significant abnormalities in metabolism and calcium signaling. With regards to metabolism, we and others have shown evidence of increased aerobic glycolysis and evidence of increased utilization of alternate fuel sources (such as fatty acids) in PAH EC. In the realm of calcium signaling, our prior work linked increased activity of the transient receptor potential vanilloid-4 (TRPV4) channel to increased proliferation of MVECs isolated from the Sugen/Hypoxia rat model of PAH (SuHx-MVECs). However, the relationship between metabolic shifts and calcium abnormalities was not clear. Specifically, whether shifts in metabolism were responsible for increasing TRPV4 channel activity in SuHx-MVECs was not known. In this study, using human data, serum samples from SuHx rats, and SuHx-MVECs, we describe the consequences of increased MVEC fatty acid oxidation in PAH. In human samples, we observed an increase in long-chain fatty acid levels that was associated with PAH severity. Next, using SuHx rats and SuHx-MVECs, we observed increased intracellular levels of lipids. We also show that increasing intracellular lipid content increases TRPV4 activity, whereas inhibiting fatty acid oxidation normalizes basal calcium levels in SuHx-MVECs. By exploring the fate of fatty acid-derived carbons, we observed that the metabolite linking increased intracellular lipids to TRPV4 activity was β-hydroxybutyrate (BOHB), a product of fatty acid oxidation. Finally, we show that BOHB supplementation alone is sufficient to sensitize the TRPV4 channel in rat and mouse MVECs. Returning to humans, we observe a transpulmonary BOHB gradient in human patients with PAH. Thus, we establish a link between fatty acid oxidation, BOHB production, and TRPV4 activity in MVECs in PAH. These data provide new insight into metabolic regulation of calcium signaling in lung MVECs in PAH.

NEW & NOTEWORTHY In this paper, we explore the link between metabolism and intracellular calcium levels in microvascular endothelial cells (MVECs) in pulmonary arterial hypertension (PAH). We show that fatty acid oxidation promotes sensitivity of the transient receptor potential vanilloid-4 (TRPV4) calcium channel in MVECs isolated from a rodent model of PAH.

INTRODUCTION

Pulmonary arterial hypertension (PAH) is a highly morbid disease characterized by endothelial dysfunction in the pulmonary vasculature. In PAH, lung endothelial cell (EC) dysfunction is thought to contribute to formation of vaso-occlusive lesions and vascular remodeling. We have previously shown that lung microvascular endothelial cells (MVECs) isolated from the SU5416/Hypoxia experimental model of PAH (SuHx-MVECs) exhibit a quasi-oncogenic phenotype, characterized by increased proliferation and altered metabolism, similar to what has been observed in human PAH (1). These two phenomena, increased proliferation and altered metabolism, are linked; proliferating cells often undergo glycolytic shift (i.e., increased shunting of glucose carbons to alternate pathways, resulting in decreased availability of glucose-derived carbons for the TCA cycle). Consequently, proliferative cells often utilize alternative fuel sources to sustain levels of metabolites of the TCA cycle (i.e., anaplerosis). As detailed next, we previously observed this pair of metabolic abnormalities (glycolytic shift and anaplerosis) in SuHx-MVECs.

We previously reported that SuHx-MVECs were more glycolytic than normoxic controls (N-MVECs; 1, 2), mirroring what has been observed in human PAH EC samples. For instance, we observed increased lactate production and decreased respiration in SuHx-MVECs compared with controls. Yet, when we measured TCA metabolites, we noted that levels of these metabolites (such as citrate and succinate) were similar between N- and SuHx-MVECs (2), suggesting that an alternative nonglucose fuel (such as amino acids or fatty acids) was providing carbons to replenish TCA metabolites in the context of decreased availability of glucose-derived carbons. Indeed, prior work has implicated increased serum fatty acid levels in patients with PAH (3), suggesting that increased lipid uptake may represent a maladaptive metabolic shift in PAH (4). However, the extent to which FAs were contributing to other cellular signaling abnormalities (including anaplerosis) observed in SuHx-MVECs remained unclear.

In addition to increased proliferation and dysregulated metabolism, we previously reported increased basal intracellular calcium ([Ca²⁺]_i) in SuHx-MVECs secondary to increased activity of transient receptor potential vanilloid-4 (TRPV4) channels (1, 5). In these studies, inhibition of TRPV4 decreased [Ca²⁺]_i and abnormal proliferation in SuHx-MVECs (1). In particular, a striking abnormality in these experiments was an increased sensitivity, in SuHx-MVECs, of TRPV4 channels to chemical agonists [such as the specific TRPV4 agonist GSK1016790A (2)]. Such increased sensitivity occurred despite similar levels of total TRPV4 protein, suggesting that the mechanism may be related to posttranslational modification of the TRPV4 channel. In recent years, regulation of TRPV4 activity has been investigated, and it is now known that a variety of factors (e.g., phosphorylation, various intracellular signaling molecules such as cyclic nucleotides) can alter the sensitivity of the TRPV4 channel to external stimuli. However, whether the metabolic abnormalities observed in SuHx-MVECs contributed to increased TRPV4 sensitivity in these cells was not clear.

In this study, we tested the hypothesis that fatty acid oxidation may be a unifying mechanism linking metabolism and TRPV4 activity in SuHx-MVECs. Specifically, we hypothesized that serum FA levels were increased in humans and rodents with PAH and that one or more of the metabolic byproducts of fatty acid oxidation (FAO) may also sensitize the TRPV4 channel in SuHx-MVECs. In our experiments, we first sought to determine 1) whether long-chain fatty acids (LCFA) were increased in clinical samples from patients with PAH and correlated with clinical severity; 2) if byproducts of FAO were increased in SuHx-MVECs, and 3) the effect, in N- and SuHx-MVECs, of manipulating FAO (by either provisioning exogenous long chain fatty acids or inhibiting FAO by preventing β-oxidation of FA by the mitochondria) on TRPV4 activity

METHODS

Approvals

All animal procedures were approved by the Animal Care and Use Committee of the Johns Hopkins University School of Medicine. Institutional Review Board approval and written informed consent was obtained for all human studies at the respective institutions.

Human Samples

RNASeq analysis.

Microarray-based profiling of mRNA transcript levels in control and PAH lungs from data set GSE113439 (n = 22) was downloaded via the NIH Gene Expression Omnibus (GEO). Transcript levels were analyzed using GEO-R. Normalization was performed using limma/GEOquery. Adjustment for multiple comparisons was performed using the Benjamini–Hochberg method. Geneset enrichment analysis (GSEA) was conducted using clusterProfiler (5).

Metabolomics data set no. 1.

This is a single-center cohort of scleroderma patients with and without PAH. Oleic and palmitate levels were previously shown to be significantly different in PLS-DA analysis of the full data set (6). For the current analyses, levels of oleate and palmitate were extracted from the metabolomics data and matched with corresponding patient-level clinical data. The clinical characteristics of the patients in this cohort are included in Supplemental Table S1. A logistic regression model was constructed using the presence or absence of PAH as the categorical outcome. Oleate and palmitate values were log transformed before inclusion in the model. Diabetes, dyslipidemia, and prednisone use were coded as categorical variables (presence/absence). Univariate and multivariable analyses were conducted in R (7). Collinearity was assessed using variance inflation factors (VIF). Expectedly, significant multicollinearity was observed with dyslipidemia and diabetes covariates (VIF > 5). Additional models were run after exclusion of patients with a diagnosis of dyslipidemia or diabetes, and the same significant relationships between oleate and palmitate and presence of PAH were observed in multivariable analyses. A full markdown of the model code and results is provided in the Supplemental Data.

Metabolomics data set no. 2.

The Servetus study is a single institution cohort of PAH participants enrolled between 2014 and 2016 (8). The clinical details of this cohort were reported previously (8). For the current analysis, 100 participants had available transthoracic echocardiogram data with reported right ventricular systolic pressure (RVSP), age, sex, and physician-confirmed etiology of PAH. Metabolites were characterized using ultrahigh-performance liquid chromatography-tandem mass spectroscopy analysis on plasma samples (Metabolon, Inc.; Durham, NC). Metabolomics data were preprocessed and normalized in R (7). Metabolites were measured as area under the peak curve, normalized to run-day medians, and log2-transformed. For the current analyses, three metabolites were included based on Metabolon annotation to be oleate, palmitate, and 3-hydroxybutyrate. Linear regression was used to estimate relationships between metabolites and RVSP. Unadjusted analyses were considered. Adjusted models accounted for differences in age, sex at birth, body mass index (BMI), and PAH etiology. Metabolites correlated with RVSP with a P-value < 0.05 were considered as potentially significant.

Metabolomics data set no. 3.

Patients without PH, those with diagnosed pulmonary venous hypertension (PVH), and those with PAH were recruited for this single center study performed at Vanderbilt University. Blood samples were obtained at the time of right heart catheterization. H-NMR data were available for 71 paired patient samples (see LC/MS Metabolomics section for full details). The mean age for the cohort was 59 (±12) yr, and predominantly female (82%). The mean pulmonary artery (PA) pressure in control, PVH, and PAH groups was, respectively (in mmHg): 18 (± 4), 39 (±12), and 46 (±14).

Animal Studies

Rat SuHx model.

Male Wistar rats were treated with SU5416 (Sugen) or vehicle control and maintained in either normoxia (controls) or hypoxia (3 wk) followed by return to normoxia (2 wk). Hemodynamics and lung harvests were performed at 5 wk, as previously described (1, 9). Briefly, rats were anesthetized, and a midline abdominal incision was made to gain access to the diaphragm. Using the transdiaphragmatic approach, a 23-gauge heparinized needle was advanced into the RV for measurement of right ventricular systolic pressure. Next, the chest was quickly opened via sternotomy, and blood was drawn from the RV. The animal was then euthanized by exsanguination. The lung and heart were excised en bloc. The RV was dissected and weighed, along with the LV/septum. Meanwhile, the lungs were subjected to MVEC harvest, as detailed in Rat MVEC cell culture section.

Rat MVEC cell culture.

MVECs from normoxic (N-MVEC) and SuHx (SuHx-MVECs) rats were isolated at 5 wk. Peripheral strips of rat lung were digested using collagenase and subjected to dual-step magnetic bead selection (using CD31- and Griffonia simplicifolia lectin-conjugated beads), as described in detail by us previously (1, 2). Cells were frozen at passage 3. Before use at passage 4, N-MVECs were rephenotyped to ensure lack of transdifferentiation (i.e., endothelial mesenchymal transition/acquisition of smooth muscle cell markers). Cells were routinely tested for mycoplasma testing. The cell culture media was based on Lonza EGM2 but without hydrocortisone. Serum concentrations in the media were optimized as follows. Growth media (used during the isolation process during the initial growth to confluence, before freezing at passage 3) contained 10% FBS (Hyclone). The FBS concentration in the media was reduced to 5% for subsequent cell culture, including experiments. The other components of the media recipe were reported by us recently (6). For metabolomics/lipidomics experiments, pairs of N- and SuHx animals were anesthetized consecutively on the same day, and the ensuing lots of N- and SuHx-MVECs were subsequently used together, as a control/experimental cell lot pair. Thus, metabolomic/lipidomic harvests were performed on N- and SuHx-MVECs that had been under culture for identical periods of time (typically < 48 h).

[Ca²⁺]_i measurements.

Semiconfluent monolayers of MVECs were seeded on glass coverslips, treated with Fura-2 AM for 30–60 min, and mounted in a temperature-controlled chamber perfused with freshly prepared modified Krebs solution (118 mM NaCl, 4.7 mM KCl, 0.57 mM MgSO₄, 1.18 mM KH₂PO₄, 25 mM NaHCO₃, 2.5 mM CaCl₃, 10 mM glucose). The Krebs buffer was gassed (16% O₂, 5% CO₂) and warmed (37°) before being perfused into the cell chamber at a rate of 1 mL/min. [Ca²⁺]_i imaging was subsequently performed as described by us previously (10). Cells were perfused for 15 min without any imaging to allow for equilibration to flow, before measurement of baseline [Ca²⁺]_i (5 min). Then, cells were perfused with Krebs buffer or Krebs buffer containing 5 mM BOHB (Sigma; 166898) for 10 min. The cells were finally perfused with 1.5 µM GSK1016790A (GSK; SelleckChem S8107) for 5 min. BOHB treatment calcium experiments involved 15 min of baseline measurements (5 min of untreated baseline [Ca²⁺]_i measurement followed by 10 min of [Ca²⁺]_i measurement with BOHB perfusion) before GSK exposure. For experiments involving oleate treatment, cells were treated with oleate (0.5 mM, Sigma O3008) or diluent overnight before calcium imaging. For both BOHB- and oleate-treated cells, we measured basal [Ca²⁺]_i for 15 min before GSK exposure, to ensure that the time spent in the Ca²⁺ measurement chamber was identical in both treatment groups (and respective controls). Following acquisition of images, individual cells were outlined, and [Ca²⁺]_i was calculated as described previously (10). Baseline [Ca²⁺]_i and change in [Ca²⁺]_i were also calculated as described previously (10).

BODIPY intracellular lipid staining.

Passage 4 N- and SuHx-MVECs were incubated with BODIPY (Invitrogen, 2 µM) for 15 min before imaging in an epifluorescence microscope equipped with a heated, gassed chamber (as described previously) (1). Individual cells were chosen at random for each technical replicate, and lipid droplets were quantified using CellProfiler. The average number of droplets for all cells from technical replicates was averaged to yield a lipid droplet number for a biological replicate.

Nonesterified fatty acids assay.

Free fatty acids were measured in rat serum samples using a nonesterified fatty acid (NEFA) assay kit (Wako Diagnostics). About 15 µL of sample was incubated with assay reagents for 15 min before absorbance measurements at 550 nm. A calibration curve was used to determine free fatty acid (FFA) concentrations.

LC/MS Metabolomics

13C-labeling.

We initially constituted ¹³C-palmitate in methanol to achieve a final concentration of 0.3 mM and performed 13C-labeling experiments following incubation for 24 h. However, cells appeared to be morphologically unhealthy at this concentration of palmitate/diluent, and there was concern for cell viability. Therefore, we measured lipid droplets and BOHB concentrations in cells treated with half dose, 0.15 mM, for 48 h. We observed increased lipid uptake and BOHB concentrations, suggesting that sufficient palmitate uptake was occurring at this dose. Both palmitate and diluent treated cells also appeared healthy at this dose. Therefore, 13C-labeled experiments conducted at this dose/time point were analyzed.

Extraction.

Paired confluent flasks of N- and SuHx-MVECs were thawed and grown to confluence. Cells were washed three times with mass-spec grade ice-cold PBS. Metabolites were extracted with a 2:1.5:1 ratio (vol/vol/vol) solution of acetonitrile:MS-grade water:chloroform. The samples were then sonicated and cold-centrifuged to extract the metabolites and separate the aqueous-phase and organic-phase metabolites from the proteins. Next, a 2:1:1 mM mixture of chloroform:methanol:butylated hydroxytoluene (BHT) was used to further extract the metabolites from the cell samples. Both the aqueous and the organic samples were then separately subjected to a speed vacuum and a lyophilization process to remove the organic and aqueous solvents. Finally, the aqueous phase dried metabolites were reconstituted for metabolomic data acquisition in a solution of 50% (vol/vol) acetonitrile in MS-grade water, whereas the organic phase dried lipids were reconstituted in chloroform:methanol:BHT 2:1:1 mM mixture.

Mass spec approach.

The liquid chromatography-mass spectrometer used for data acquisition was the Q Exactive Plus Orbitrap Mass Spectrometer by Thermo Scientific coupled to a Vanquish UPHLC system. The autosampler chamber was maintained at 4°C and 15°C throughout to safeguard the metabolite and lipid sample’s integrity, respectively, during data acquisition. Reverse phase chromatography was carried out on a Discovery HSF5 column (Sigma) with a Supelco guard column, both held at a temperature of 35°C. The polar metabolomics method used 0.1% formic acid in MS-water (A) and 0.1% formic acid in acetonitrile (B) mobile phases and ran for a total of 15 min. The gradient was as follows: from 0 to 3 min, 98% A and 2% B at a flow rate of 0.3 mL/min; from 3 min to 5 min, increase from 2% to 30% B at a flow rate of 0.3 mL/min; from 5 to 8 min, increase from 30% to 100% B at a flow rate of 0.3 mL/min; from 8 to 10 min, 100% B at a flow rate of 0.3 mL/min; at 10 min, decrease from 100% to 2% B at a flow rate of 0.3 mL/min; from 10 to 11 min, increase flow rate from 0.3 mL/min to 0.6 mL/min at 98% A and 2% B; and from 11 min to end of run at 15 min, 98% A and 2% B at a flow rate of 0.6 mL/min. During this time, full MS scans and full MS/ddMS2 (top10) scans were acquired in the first 11 min, followed by column re-equilibration for the final 4 min. The lipidomics method used 10 mM ammonium formate solution in 60% acetonitrile and 40% mass spec water containing 0.1% formic acid as mobile phase A, whereas mobile phase B consisted of 10 mM ammonium formate solution in 90% isopropanol and 10% acetonitrile solution containing 0.1% formic acid. The gradient was as followed: from 0 to 2 min, increase from 30% to 43% B; from 2 to 2.1 min, increase from 43% to 55% B; from 2.1 to 12 min, increase from 55% to 65% B; from 12 to 18 min, increase from 65% to 85% B; from 18 min to 20 min, increase from 85% to 100% B; from 20 to 25 min, 100% B; from 25 to 25.1 min, decrease from 100% to 30% B; and from 25.1 to 28 min, 30% B. The flow rate is constant throughout the run at 0.26 mL/min. The lipidomics method ran for a total of 28 min, with both full MS scans and full MS/ddMS2 scans. To ensure the instrument’s sensitivity and data accuracy, routine evaluation and calibration were performed before acquisition.

Metabolomics data analysis.

The obtained data were subjected to analysis using a suite of software tools, including Compound Discoverer from Thermo Fisher for qualitative and quantitative analysis of the unlabeled compounds, as well as identification of compounds in the labeled samples. Identification was achieved through database of isotopic pattern, fragmentation pattern matching, and mass-to-charge ratio (m/z) accuracy matching. LipidSearch software was also used for the identification of lipids. FreeStyle and TraceFinder software were then utilized for further metabolite and lipid identification and quantification, respectively. Chromatographic peak integration was used to extract intensity data, which were subsequently normalized based on the weight of the sample. To determine fold change, the intensity data for each sample were divided by the intensity data for the control group for the unlabeled samples or intensity data of the M + 0 isotopologue of the control group for the labeled samples.

¹H-NMR metabolomics.

¹H NMR metabolomics data were acquired as previously described (11). Briefly, plasma was collected from patients during right heart catheterization. Plasma samples (150 µL) were extracted with methanol (450 µL), vortexed for 2 min on a multitube vortexer, and centrifuged at 16,000 rcf for 5 min. A 400 µL aliquot of the supernatant was dried and reconstituted with 250 µL of NMR master mix solution containing Chenomx ISTD: DSS-d6 and 0.20 M phosphate buffer at pH 7.4. Samples were vortexed, centrifuged, and a 200 µL aliquot of supernatants were transferred into NMR tubes. Pooled plasma samples were prepared by combining equal aliquots (15 µL) from each study sample, and aliquots of the pooled samples were processed identically to the study samples and were used for quality check purposes. ¹H NMR spectra of plasma extracts were acquired on a Bruker Avance III 700 MHz spectrometer using a 5 mm cryogenically cooled ATMA inverse probe and ambient temperature of 25°C. A noesypr-1d pulse sequence was used for data acquisition. For each sample, 64 transients were collected into 64 k data points using a spectral width of 12.0227 ppm, 2 s relaxation delay, 100 ms mixing time, and an acquisition time of 3.89 s per Free Induction Decay (FID). The water resonance was suppressed using resonance irradiation during the relaxation delay and mixing time. Spectra were zero filled, and Fourier transformed after exponential multiplication with line broadening factor of 0.5. Phase and baseline of the spectra were manually corrected for each spectrum. Spectra were referenced internally to the DSS signal. The quality of each NMR spectrum was assessed for the level of noise and alignment of identified markers. Spectra were assessed for missing data and underwent quality checks. BOHB was library matched, and the relative concentration was determined by using Chenomx NMR Suite 9.0 Professional software (Chenomx Inc., Edmonton, AB, Canada).

Targeted intracellular NADH measurements.

Confluent T75s of N- and SuHx-MVECs were harvested for metabolomics analyses using methanol as previously described (2) and stored immediately in −80°C. Samples were subsequently lyophilized, resuspended, and subjected to LC/MS using the parameters and settings as described previously (1, 12). Values were normalized to lysate protein concentration [bicinchoninic acid (BCA) assay].

BOHB assay.

A calorimetric BOHB assay (ab83390, Abcam) was used. N- and SuHx-MVECs cells were plated in MVEC media (see section on Rat MVEC cell culture) containing 5% FBS. Cells were trypsinized and pelleted before resuspension in assay buffer. Deproteinization was performed using perchloric acid as per kit instructions. Following incubation with the BOHB enzyme mix and the developer solution, absorbance was measured at 450 nm. To confirm these results, we also conducted the same experiment using a different calorimetric BOHB assay (Beta-Hydroxybutyrate kit, Biomedical Research Service, University of Buffalo). In this assay, samples were deproteinized using an alternate method (PEG precipitation). The sample was then incubated with BOHB assay solutions, and absorbance was measured at 492 nm, per assay instructions. Since both assays provided similar results in terms of differences in BOHB concentrations between N- and SuHx-MVEC lysates, the results were combined. Results were normalized to protein concentrations (BCA assay).

Proliferation.

MVECs were seeded in a 96-well plate at a density of 3,000 cells/well, and then incubated with BrDU overnight before fixation and BrDU incorporation measurement as previously described (1). FBS-treated wells were used as a positive control.

cAMP measurement in live cells.

Semiconfluent monolayers of N- and SuHx-MVECs were grown on glass bottomed plates and treated with the cADDIS downward sensor construct (Montana Molecular; 60 µL sensor per 1 mL of media) and incubated for 24 h before imaging. Imaging was performed using an epifluorescence microscope (see section on BODIPY intracellular lipid staining for details) outfitted with a temperature controlled, gassed chamber. Baseline fluorescence images for a field of cells were obtained before treatment with 25 µM of forskolin. The forskolin-induced cAMP levels were calculated by averaging the plateau response after drug addition. In this construct (downward sensing cADDIS), fluorescence values decrease when cAMP levels increase. Therefore, F values were transformed using F = 1/F so that the increasing cAMP levels corresponded with an increase in the graphed fluorescence value. The average F_forskolin value was calculated to determine F_max. The average fluorescence value for a 5-min period before forskolin treatment was used to determine the baseline fluorescence (F_baseline). Thus, % of forskolin max was calculated as F_baseline/F_max.

Statistics.

Differences between two groups were tested using a nonparametric (Mann–Whitney) test. Comparisons across multiple groups were formed using one- or two-way ANOVA, followed by post hoc individual comparisons (Sidak or Holm–Sidak method, respectively). P values < 0.05 were considered significant.

Raw data/code availability.

Mass spectrometry data for lipidomics and 13c-labeling experiments, along with the code used to perform the RNAseq/GSEA (Fig. 1A) and logistic regression analyses for cohort no. 1 (Fig. 1B), are provided online (github.com/suresh-laboratory/SuHx-FAO-Metabolomics). The raw ¹H-NMR data are available at the NIH Common Fund’s National Metabolomics Data Repository (NMDR) website, the Metabolomics Workbench, https://www.metabolomicsworkbench.org, where it has been assigned Project ID PR000356.

Figure 1.

Increased FA availability and utilization in human samples. A: volcano plot showing logFC in transcript levels in PAH samples (n = 11) compared with control lung samples (n = 10) in data set GSE113439—a RNAseq dataset of control and PAH lungs. Solid line indicates FDR adjusted P value cutoff (0.05). Inset: the role of highlighted transcripts in the fatty acid oxidation pathway. B: plot showing odds ratios (and 95% CI) for the association between the oleate/palmitate and the presence or absence of PAH in cohort no. 1 (scleroderma patients with and without PAH), assessed using logistic regression models (n = 112 patients). C: plot showing regression coefficients (and 95% CI) for the association between oleate/palmitate and RVSP in a cohort of patients with PAH (cohort no. 2; n = 100 patients). *P < 0.05 for the association in the specified linear regression model. FA, fatty acids; FDR, false discovery rate; PAH, pulmonary arterial hypertension; RVSP, right ventricular systolic pressure.

RESULTS

We began by querying for evidence of increased lipid metabolism in human PAH samples. Prior evidence has shown enrichment of plasma fatty acid metabolites in patient with PAH samples (3). We examined transcript levels of proteins involved in lipid metabolism in a RNA-seq data set of explanted PAH lungs (compared with control lungs). We noted increased transcript levels of a variety of genes implicated in the uptake, processing, and oxidation of long-chain fatty acids (Fig. 1A). Consistent with this finding, there was significant enrichment of genesets associated with fatty acid oxidation/lipid metabolism in PAH lung samples (Supplemental Fig. S1A). Next, we examined the levels of specific long-chain fatty acids (LCFA) in the serum of a cohort of scleroderma patients with and without PAH (n = 112 patients). Clinical details for this cohort are provided in Supplemental Table S1. We first noted that serum levels of oleate and palmitate were significantly higher in scleroderma patients with PAH compared with scleroderma patients without evidence of PH (Supplemental Fig. S1B). To more robustly examine the relationship between oleate and palmitate levels and the presence of PAH in this cohort, we constructed a logistic regression model to determine the association between oleate or palmitate levels and the presence or absence of PAH. As shown in Fig. 1B, in both univariate and multivariable models, oleate and palmitate levels were significantly associated with the presence of PAH. Since this cohort was restricted to patients with scleroderma, we also examined the relationship between oleate and palmitate and right ventricular systolic pressure in a separate cohort of 100 patients with PAH of various etiologies (Servetus cohort). Again, we examined this relationship using both univariate and multivariable models. As shown in Fig. 1C, oleate and palmitate were associated with higher RVSP in the multivariable model in this cohort (n = 100 patients). These findings in human samples demonstrated a positive association between concentrations of the LCFAs oleate and palmitate and PAH.

To explore possible mechanisms behind this relationship between LCFA and PAH and to ascertain whether increased serum LCFA levels were contributing to lung microvascular ECs (MVEC) dysfunction in PAH, we turned to the SuHx rat model of PAH. Given the data demonstrating increased fatty acid availability to the pulmonary circulation in PAH, we measured free fatty acid (FFA) levels in serum drawn from the RV of SuHx rats at the time of hemodynamic measurement (i.e., at 5 wk, the same time as MVEC isolation). As shown in Fig. 2A, we observed significantly higher FFA levels in SuHx rat sera. Next, we examined intracellular lipids in SuHx-MVECs. As shown in Fig. 2, B and C, we observed increased numbers of intracellular lipid droplets in SuHx-MVECs. To further assess intracellular lipid content in SuHx-MVECs, we performed lipidomic analyses in confluent flasks of N- and SuHx-MVECs. As shown in Fig. 2D, we observed significantly higher levels of triacylglycerides (TGs) containing a variety of saturated and unsaturated long-chain fatty acids, including palmitate (16:0) and oleate (18:2).

Figure 2.

Serum and intracellular fatty acids in SuHx-MVECs. A: scatter plot showing free fatty acid levels in the serum collected from the RV of normoxic and SuHx rats collected at 5 wk. B: representative photomicrographs of N- and SuHx-MVECs stained for intracellular lipid droplets (with BODIPY). C: scatter plot showing number of lipid droplets/cell in N- and SuHx-MVECs. Each dot represents a biological replicate, obtained by averaging cells obtained from 5 to 10 images. D: lipidomic analysis in unlabeled N- and SuHx-MVECs showing significant increased levels of triglycerides including those containing oleic acid (18:1) and palmitic acid (16:0). Horizontal line represents P value cutoff of 0.05. RV, right ventricular; SuHx-MVECs, Sugen/Hypoxia rat model of PAH.

As discussed earlier, we previously observed that TRPV4 sensitivity (i.e., GSK-induced Ca²⁺ influx) was significantly higher in SuHx-MVECs. Given prior work implicating a link between LCFA and TRPV4 activity in mesenteric arteries (13), we hypothesized that, in lung MVECs, increased intracellular fatty acids may be contributing to increased TRPV4 sensitivity. To test this hypothesis, we began with a gain-of-function experiment, where we artificially increased intracellular lipid levels in N-MVECs. Treatment of N-MVECs with oleate increased the number of lipid droplets (Fig. 3, A and B), suggesting that uptake of oleate was occurring in N-MVECs. Next, we observed the effect of oleate treatment on TRPV4 sensitivity. As shown in Fig. 3, C–E, basal [Ca²⁺]_i ^-was not significantly different in oleate-treated MVECs. However, GSK-induced Ca²⁺ influx was significantly higher in oleate-treated N-MVECs, suggesting that oleate treatment was sufficient to increase TRPV4 sensitivity in N-MVECs.

Figure 3.

Exogenous FA supplementation increases TRPV4 sensitivity in N-MVECs. A: representative photomicrographs showing lipid droplets in N-MVECs with and without treatment with oleate (0.05 mM; 16 h). B: scatterplot showing quantification of intracellular lipid droplets in N-MVECs with and without oleate treatment. C: representative trace showing [Ca²⁺]_i in untreated (Norm) and oleate-treated (N+Oleate) N-MVECs at baseline and following treatment with GSK1016790A (GSK; 1.5 µM). Scatterplots showing basal (D) and delta [Ca²⁺]_i following GSK treatment (E) in untreated (Norm) and Oleate-treated (N+Oleate) N-MVECs. FA, fatty acid; N-MVECs, MVECs from normoxic; TRPV4, transient receptor potential vanilloid-4.

To determine the functional consequence of fatty acid oxidation in disease, we examined the effect of inhibiting fatty acid oxidation in SuHx-MVECs. We have previously shown that increased TRPV4 activity contributes to increased basal [Ca²⁺]_i levels in SuHx-MVECs (1). Thus, we examined the effect of inhibiting FAO on basal [Ca²⁺]_i in SuHx-MVECs. As shown in Fig. 4A, basal [Ca²⁺]_i was higher in SuHx-MVECs, consistent with our prior work. Treatment with etomoxir (Eto), an inhibitor of fatty acid oxidation, significantly decreased [Ca²⁺]_i in SuHx-MVECs (Fig. 4A), similar to what we observed previously with TRPV4 inhibition. To determine whether global loss of FA availability would similarly attenuate basal [Ca²⁺]_i in SuHx-MVECs, we measured basal [Ca²⁺]_i in N- and SuHx-MVECs cultured in serum-free media for 4 h before measurement and observed that this treatment reduced basal [Ca²⁺]_i in SuHx-MVECs, similar to Eto treatment (Supplemental Fig. S2A).

Figure 4.

Inhibition of FAO decreases basal Ca²⁺ and proliferation in SuHx-MVECs. A: scatterplots showing basal [Ca²⁺]_i in N- and SuHx-MVECs with and without Etomoxir (Eto) treatment (50 µM; 16 h). B: scatterplots showing fold change in proliferation (BrDU) in N- and SuHx-MVECs with and without treatment with Etoximir (100 µM; 24 h and 48 h). For B, values above bars indicate P values from a post hoc multiple comparisons test following a two-way ANOVA. FAO, fatty acid oxidation; SuHx-MVECs, Sugen/Hypoxia rat model of PAH.

To determine whether inhibiting lipid transfer to the mitochondria would lead to increased lipid droplets (due to decreased consumption), we measured lipid droplets in untreated and Eto-treated N- and SuHx-MVECs. We observed that Eto treatment increased lipid droplet numbers in N-MVECs and SuHx-MVECs (Supplemental Fig. S2B). Next, we turned to another functional outcome, proliferation. We had also noted previously that increased proliferation in SuHx-MVECs was TRPV4-dependent (1). We reasoned that if inhibiting oxidation normalized TRPV4-dependent [Ca²⁺]_i levels, FAO inhibition may also attenuate proliferation in SuHx-MVECs. Indeed, we observed that proliferation in SuHx-MVECs was decreased in in MVECs with either 24 or 48 h of etomoxir (Eto) treatment (Fig. 4B).

Thus far, the data suggested that 1) fatty acid levels were increased in SuHx-MVECs; 2) provisioning FA was sufficient to accentuate TRPV4 activity in N-MVECs; and 3) inhibiting FAO was sufficient to attenuate TRPV4-dependent cell outcomes (basal [Ca²⁺]_i and proliferation) in SuHx-MVECs. However, these data did not shed light on how increasing intracellular fatty acid levels may be affecting TRPV4 activity. First, we examined whether oleate supplementation increased TRPV4 activity by increasing the amount of TRPV4 protein. To examine this, we measured TRPV4 protein levels by immunoblot in control and oleate-treated N-MVECs. We did not observe a significant difference in TRPV4 protein levels (Supplemental Fig. S3A), suggesting that the effect of exogenous FA supplementation on TRPV4 activity may be occurring posttranslationally. To explore how fatty acids may be regulating TRPV4 activity, we chose to focus on two possible theories. First, we have already shown that anaplerosis was present in SuHx-MVECs (1). Thus, one possibility was that fatty acid oxidation could be providing anaplerotic carbons to the TCA, and signaling via these TCA metabolites might be contributing to TRPV4 signaling. Another possibility was that a by-product of β-oxidation itself, such as β-hydoxybutyrate (BOHB), may be directly responsible for fine-tuning TRPV4 sensitivity in MVECs.

To test the first hypothesis, we performed ¹³C₁₆-palmitate-labeling experiments using stable isotope-resolved metabolomics. As shown in Supplemental Fig. S3B, treatment of N-MVECs with ¹³C₁₆-Palmitate was sufficient to increase intracellular lipid droplets and intracellular BOHB levels, suggesting that ¹³C-labeled FA uptake was occurring in these cells. We then performed LC/MS analyses on these cell lysates. We were surprised to note that there was very little labeling of TCA metabolites by ¹³C in either N- or SuHx-MVECs, as indicated by the low fraction of labeled isotopologues (M + 1 and beyond) as compared with the nonlabeled M + 0 isotopologue (Fig. 5). In contrast, we observed an active metabolism of ¹³C₁₆-Palmitate into other pathways including ribonucleotide synthesis and glutamine metabolism as shown by the high fraction or percentage of labeled isotopologues beyond natural abundant M + 1 (14) (for example, M + 2, M + 3, and M + 4 for UDP-N-acetylglucosamine, M + 3 for 2-HG, and M + 5 and M + 6 for NAAG). These data suggested that, at least in SuHx-MVECs, FA were not the source of anaplerotic carbons for the TCA cycle, thus arguing against this being a possible mechanism for FAO-mediated TRPV4 dysfunction in SuHx-MVECs.

Figure 5.

¹³C₁₆-palmitate labeling does not reveal divergent destinations for fatty acid oxidation derived carbons in SuHx-MVECs. Stacked bar graphs showing mole fractions of unlabeled (M0) and labeled (M1–M4) carbons in 13C-palmitate-treated N- and SuHx-MVECs after 48 h of incubation with 13C-palmitate. Metabolites from the TCA cycle as well as ribonucleotide and glutamine metabolism are shown, n = 5 cell lots/group, each lot isolated from a different animal. SuHx-MVECs, Sugen/Hypoxia rat model of PAH; TCA, trichloroacetic acid.

We turned to our alternate hypothesis that the BOHB may be the link between FAO and TRPV4 sensitivity. In fact, prior work has suggested that BOHB itself may signal via GPCRs to effect cAMP levels (15), and cAMP levels in turn are known to regulate TRPV4 activity (16). Thus, we began examining the breakdown of LCFA in ¹³C-labeled experiments.

We examined our ¹³C-metabolomics data to determine whether products of palmitate breakdown were enriched in ¹³C₁₆-palmitate-treated SuHx-MVECs, focusing on labeled compounds. As shown in Fig. 6, fully labeled ¹³C₁₆-palmitate is converted to oct-2-enoyl-CoA, carrying a maximum of eight labeled ¹³C (M + 8 isotopologue), and then to butanoyl-CoA, carrying a maximum of four labeled ¹³C (M + 4 isotopologue). The M + 4 isotopologue of butanoyl-CoA, which is an expected breakdown product from ¹³C₁₆-palmitate, was significantly increased in SuHx-MVECs. In addition, acetoacetate, a metabolite produced from butanoyl-CoA that is readily interconverted to BOHB, was also increased in SuHx-MVEC samples. However, we could not confidently extract peaks for BOHB in these experiments. Therefore, we measured BOHB directly, using a targeted metabolite assay. Indeed, we observed a significant increase in intracellular BOHB levels in SuHx-MVECs (Fig. 7A) We noted that the mechanism by which commercial assays measure metabolites like BOHB typically involves using a dehydrogenase enzyme (in this case, β-hydroxybutyrate dehydrogenase) to generate NADH and then measuring the resulting change in NADH levels in the second reaction. However, we reasoned that if NADH levels were increased at baseline in SuHx-MVECs, this may artificially affect our results. Thus, using targeted metabolomics, we specifically measured NADH levels, and noted no difference in NADH levels between N- and SuHx-MVECs (Fig. 7B). This provided us with more confidence in the results of our BOHB assay results. BOHB can also be transported across the cell membrane. To determine whether increased intracellular BOHB levels were a consequence of increased circulating BOHB levels, we measured BOHB levels in the RV serum of N- and SuHx rats and observed no difference (Fig. 7C). Next, since we observed increased TRPV4 sensitivity in our oleate-treated N-MVECs, we hypothesized that BOHB should also be elevated in oleate- or palmitate-treated N-MVECs. As shown in Fig. 7, D and E, we observed such an increase, although the magnitude of increase in BOHB concentration was less compared with differences between and N- and SuHx-MVECs. Thus, we concluded that intracellular production of BOHB was occurring as a consequence of increased FAO in SuHx-MVECs and in FA-treated N-MVECs. Finally, we hypothesized that if intracellular production (and not serum levels/increased import) of BOHB was responsible, then the association between serum oleate/palmitate and PAH severity observed in humans would not be present when serum BOHB levels (drawn from peripheral venous blood) were regressed against RVSP in the same cohort. Indeed, as shown in Fig. 7F, unlike oleate/palmitate, serum BOHB was not significantly associated with RVSP in the Servetus cohort.

Figure 6.

¹³C₁₆-palmitate labeling reveals breakdown of LCFA in SuHx-MVECs. Diagram showing palmitate metabolism and scatterplots (±SE) showing labeling of palmitate breakdown products Oct-2-enoyl-CoA, butanoyl-CoA, and acetoacetate in ¹³C₁₆-palmitate labeled N- and SuHx-MVECs. LCFA, long chain fatty acids; SuHx-MVECs, Sugen/Hypoxia rat model of PAH. *P < 0.05.

Figure 7.

BOHB levels in cells, animals, and humans with PAH. Scatterplot showing fold change in intracellular (i.e MVEC lysate) BOHB concentration (A) as well as NADH levels [measured using targeted metabolomics (B)] in SuHx-MVECs compared with N-MVEC controls. C: scatterplot showing fold change and mean (± SE) serum BOHB concentration in RV serum of normoxic (N) and SuHx rats. D and E: scatterplots showing fold change and mean in intracellular BOHB levels in oleate- and palmitate-treated N-MVECs. F: plot showing regression coefficients (and 95% CI) for the association between BOHB and RVSP in a cohort of patients with PAH (n = 100). Model 1 represents univariate analysis (RVSP ∼ metabolite) and Model 2 represents multivariable analysis (RVSP∼metabolite+age+sex+BMI+PAH etiology). BMI, body mass index; BOHB, β-hydroxybutyrate; MVECs, microvascular endothelial cells; N-MVECs, MVECs from normoxic; PAH, pulmonary arterial hypertension; RV, right ventricular; RVSP, right ventricular systolic pressure; SuHx-MVECs, Sugen/Hypoxia rat model of PAH.

We next sought to reconcile our in vitro findings in rat MVECs (increased BOHB levels at baseline, increased production of BOHB with LCFA treatment) with the serum findings in rats and humans (no difference in BOHB levels in the RV/venous circulation). We reasoned that if lung MVECs were a source of BOHB, then, in contrast to the RV findings, there may exist a transpulmonary gradient for BOHB across the pulmonary vasculature. Levels of many metabolites (including BOHB) can vary across vascular beds (17), and thus, if BOHB was consumed in various organ beds (e.g., muscle) in PAH, we may not see differences in RV serum even if regional gradients exist elsewhere in the circulation. To examine this, we turned to a data set of human H-NMR metabolomics performed in a cohort of patients without PH, pulmonary venous hypertension (PVH), and pulmonary arterial hypertension (PAH). Importantly, in this data set, patients underwent both PA and pulmonary capillary wedge (PCW) blood sampling. As observed in our other data sets (both rat and human), there was no difference in precapillary (i.e., PA) BOHB levels (Fig. 8A). However, the PCW/PA ratio for BOHB was significantly higher in patients with PAH compared with PVH/no PH controls (Fig. 8B). These data suggested that, although BOHB levels in blood returning to the RV were not different between control and PAH patients/animals, BOHB secretion may in fact be occurring at the level of the pulmonary vasculature in PAH. Next, we turned to how this secreted BOHB could, in autocrine/paracrine fashion, effect MVEC function with regards to TRPV4 sensitivity. For thus, we turned again to N- and SuHx-MVECs.

Figure 8.

Increased transpulmonary BOHB gradient in human PAH. Scatter plots showing PA BOHB levels (normalized to NMR standard DSS-d6) (A) and mean (±SE) transpulmonary BOHB ratios (B) in human PA samples in patients without PH, patients with pulmonary venous hypertension (PVH), and patients with PAH. BOHB, β-hydroxybutyrate; PAH, pulmonary arterial hypertension.

We reasoned that if BOHB was downstream of FAO signaling with regards to TRPV4 sensitization, exogenous BOHB treatment should mimic the effect of FA treatment in N-MVECs. Given our abovementioned transpulmonary metabolomics data, it seemed likely that MVECs in PAH may experience increased BOHB levels in part due to the production and secretion of this metabolite in the pulmonary microvasculature. To determine the dose of BOHB to use, we examined our intracellular BOHB data and chose a dose of BOHB that was within 2 SD of the mean BOHB levels observed in SuHx-MVECs, so as not to use concentrations dramatically higher than observed intracellular levels (Supplemental Fig. S3). When we now treated N-MVECs with BOHB (5 µM), we observed a significantly more robust response to the TRPV4 agonist GSK (Fig. 9, A–C) without an apparent change in basal Ca²⁺.

Figure 9.

Exogenous BOHB sensitizes TRPV4 channel activity in response to agonists in rat and mouse MVECs. A: representative [Ca²⁺]_i trace of control and BOHB-treated N-MVECs before and after treatment with TRPV4 agonist GSK. Scatterplots showing mean ± SE basal [Ca²⁺]_i (B) and change in GSK-induced change (C) in [Ca²⁺]_i in control (Ctl) and BOHB-treated N-MVECs. D: traces showing cAMP fluorescence as measured by the cADDIS cAMP sensor, and represented as a % of Forskolin-induced max cAMP fluorescence. Each trace represents the average of a field of cells (n = 5–10) imaged over time. E: scatter plot showing means± SE cAMP (% of forskolin max) in N- and SuHx-MVECs. BOHB, β-hydroxybutyrate; MVECs, microvascular endothelial cells. *P < 0.05.

BOHB has recently been reported to be an agonist for Gpr41, a G_i-coupled GPCR¹⁷. cAMP levels have previously been implicated as a regulator of TRPV4 activity (16). Thus, if BOHB was acting via Gpr41 to decrease cAMP and increase TRPV4 sensitivity, we hypothesized that cAMP levels would be lower in SuHx-MVECs. Thus, we performed live-cell imaging of intracellular cAMP concentrations (using the cADDIs sensor) in N- and SuHx-MVECs. As shown in Fig. 9D, basal cAMP levels (when expressed as a percentage of forskolin-induced max cAMP) were decreased in SuHx-MVECs compared with N-MVECs. These observations suggest a link between BOHB, cAMP, and TRPV4 activity in SuHx-MVECs.

DISCUSSION

Herein, we describe a relationship between fatty acid oxidation and TRPV4 activity in SuHx-MVECs. Using animal and human data, we provide evidence of increased long-chain fatty acid concentrations in human and rat serum samples and observe evidence of FAO in MVECs isolated from the SuHx rat. Furthermore, our data suggest that FAO-induced TRPV4 sensitivity occurs, in part, via BOHB. Our human data, from metabolomics and transcriptomic studies of explanted human lungs, suggest that oleate and palmitate levels are increased in the serum of patients with PAH, are associated with worsened hemodynamics, and PAH lung tissues have increased transcript levels of enzymes known to participate in FAO. That said, the two human cohorts in which we explored serum LCFA levels are not identical. One of the three cohorts (cohort no. 1) features patients with scleroderma with and without PAH, whereas cohorts no. 2 and 3 feature patients with PAH due to various etiologies. Thus, although similar associations are seen in both of these cohorts, differences related to the biology of scleroderma could be contributing to the differences in LCFA levels seen in cohort no. 1. Nevertheless, these patient data provide a rationale for our in vitro experiments, which, in turn, suggest that one possible destination for the increased FA observed in humans may be the lung microvasculature, where oxidation of these FA may effect hyperproliferative MVEC behavior in PAH.

TRPV4 has long known to be a multifunctional calcium channel in the pulmonary vasculature, known to be responsive to stimuli like heat and stretch (18, 19). However, considerably less is known about how channel activity is regulated. For instance, prior studies, including work by us, implicated phosphorylation in regulating channel sensitivity to stimuli (10, 20–22). However, various metabolites are also known to act as agonists and/or regulators of TRPV4 activity. Here, we implicate BOHB, a ketone body and byproduct of β-oxidation, as a sensitizer of TRPV4 activity to GSK, a chemical agonist. As noted earlier, the exact mechanism by which BOHB sensitizes TRPV4 remains unclear. However, our data suggest that although BOHB levels are increased, cAMP levels are decreased in SuHx-MVECs. Thus, an attractive hypothesis for future work is that BOHB acts via Gpr41 to decrease cAMP levels, which in turn promotes sensitization of TRPV4. Of note, although our data suggest that TRPV4 responsiveness to GSK is accentuated in the context of BOHB production, this may not be true of all other TRPV4 stimuli (such as heat or shear). Although our BOHB treatments were acute, our oleate treatment time was longer (overnight). We chose this longer time point reasoning that β-oxidation of LCFA may require a longer time than the acute effects of a byproduct of β-oxidation (BOHB). Our Western blot data showing lack of significant increase in TRPV4 expression with oleate treatment, combined with the observation that acute BOHB treatment achieves similar effects with regards to increasing TRPV4 sensitivity, argues that the mechanism may be related to posttranslation regulation of the channel rather than changes in TRPV4 protein production.

Our Etomoxir data suggest that blocking FAO in SuHx-MVECs is sufficient to decrease basal Ca²⁺ and proliferation—two processes that we have previously shown to be TRPV4-dependent (1). However, in our gain-of-function experiments, exogenous FA addition, while sensitizing TRPV4 to GSK, did not increase basal Ca²⁺. One possibility is that although FA-treated N-MVECs mimic SuHx-MVECs with regards to TRPV4 sensitivity, the magnitude of FAO in oleate-treated MVECs may not be sufficient to promote TRPV4 activation in the absence of chemical agonists (i.e., increase basal [Ca²⁺]_i). Indeed, the magnitude of increase in BOHB induced by oleate (mean FC: 1.19 ± 0.06) appears lower than the basal levels of BOHB found SuHx-MVEC lysates (mean FC: 1.77 ± 0.23). The same pattern holds true for number of lipid droplets; the difference in the number of lipid droplets (per cell per biological replicate) between oleate-treated N-MVECs and controls was 32 (±8), whereas the difference between SuHx and N-MVECs was 53 (±12). These comparisons suggest that the magnitude of lipid uptake and processing may be higher in SuHx-MVECs compared with FA-treated N-MVECs. Although the mechanisms underlying this difference are not known, this discrepancy may explain why oleate-treated N-MVECs fail to recapitulate all of the calcium abnormalities (such as increased basal Ca²⁺) observed in SuHx-MVECs.

Although we mechanistically focused on signaling induced by BOHB produced as a consequence of FAO, LCFA have been shown to affect other aspects of cellular signaling. For instance, desaturation of lipids has been shown to be a source of NAD⁺ (23), whereas membrane fluidity alterations (as a consequence of changing the saturation of fats in the phospholipid bilayer) have been suggested to be the mechanism underlying increased activity of two mechanosensitive channels, TRPV4 and PIEZO1, following exogenous FA supplementation (13, 24). However, we reason that these mechanisms would not be prevented by etomoxir, and thus suspect that they may not play as prominent a role in FAO-induced sensitivity of TRPV4 in the lung. However, we cannot fully exclude these additional mechanisms, including a component of cytoplasmic (i.e., peroxisomal rather than intra-mitochondrial) oxidation of LCFA, as possible contributors to FA-induced increases in [Ca²⁺]_i in SuHx-MVECs. When deciding on possible mechanisms of FA-induced TRPV4 sensitization, we initially chose to explore anaplerosis as a possible mechanism because this phenomenon has been previously shown to be important in endothelial-mesenchymal transition, another feature observed in PAH ECs in both patients and rodents (25, 26). However, our ¹³C labeling data suggest that although LCFAs are taken up in increased amounts and oxidized, the carbons from this increased FA uptake are not destined for the TCA cycle.

Our in vitro BOHB data suggest that BOHB sensitizes TRPV4 similar to LCFA, whereas our human data suggest that BOHB production is occurring across the pulmonary vasculature. It is possible that paracrine effects of BOHB may lead to TRPV4 sensitization of the microvasculature in PAH. Furthermore, the lack of differences in BOHB levels in the precapillary circulation in our data suggest that the BOHB produced by the pulmonary vasculature may in fact be consumed elsewhere in the circulation. In general, our data are consistent with those of Mey et al. (27) in implicating ketosis as a feature of PAH. However, our BOHB measurements in the RV are in contrast to the findings of Mey et al. (27), who observed a baseline increase in BOHB levels in the venous circulation in patients with PAH before a hyperglycemic clamp. One explanation for this difference could be that patients in the Mey study underwent dietary monitoring, restriction on vigorous physical activity, and 10 h of fasting, which may have influenced baseline BOHB measurements. Another explanation could be the site of blood draw. Recent literature has shed light on significant transorgan gradients for many metabolites, including BOHB, in mammals (17). Therefore, the site of blood draw may have also impacted BOHB levels. Despite these differences, both our study and the study by Mey et al. (27) are concordant in implicating dysregulated ketogenesis in PAH; our in vitro work extends these observations to suggest a role of increased serum ketones in sensitizing calcium channels in the lung microvasculature in PAH.

There are several limitations to this work. Although our in vitro data support the concept of MVEC β-oxidation as a regulator of TRPV4 activity, MVECs are likely not the only recipient of increased serum levels of FAs. Other cell types implicated in PAH, including macrophages and T cells, are known to alter phenotype based on the degree of FAO. Thus, MVECs are unlikely to be the sole source of FA consumption in the lung microvasculature. With regards to BOHB, we focused on this specific ketone body (as opposed to acetoacetate) in part because of prior literature implicating this molecule in regulating cAMP levels. One limitation to our lipid analyses is that we did not perform serial measurements of media/intracellular fatty acids to determine rates of fatty acid oxidation (i.e., metabolic flux analyses). Therefore, further work is needed to definitively demonstrate increased flux of LCFA oxidation in MVECs in PAH. Also, the lack of signals for short and medium-chain FA in our human data does not rule out the possibility that these FA metabolites may also be contributing to BOHB production in PAH. Another caveat to our data is that the effect of BOHB on GSK-induced Ca²⁺ influx, while significant, is also clearly bimodal. Why SuHx-MVECs isolated from some rats, but not others, respond more robustly to BOHB treatment requires further exploration. Given that our data suggest that MVECs produce BOHB, we attempted to measure BOHB levels in the media; however, the samples were too dilute and we were unable to obtain reliable BOHB measurements in deproteinated media samples. However, our human data showing a positive (i.e., PCW/PA > 1) gradient for BOHB in PAH samples suggest that BOHB secretion may in fact be occurring across the lung microvasculature.

In summary, our results suggest a novel link between fatty acid metabolism, ketone body formation, and TRPV4 sensitivity in lung microvascular endothelial cells. These data suggest that increased TRPV4 activity in SuHx-MVEC may be fueled by multiple factors, including the production of metabolites that sensitize the channel. Further exploration of these links between FAO and TRPV4 activity may provide novel therapeutic targets aimed at ameliorating MVEC dysfunction in PAH.

DATA AVAILABILITY

Mass spectrometry data are openly available at Github online (https://github.com/suresh-lab/SuHx-FAO-Metabolomics). ¹H-NMR data are openly available at the NIH Common Fund’s National Metabolomics Data Repository (NMDR) website under Project ID PR000356 (https://www.metabolomicsworkbench.org).

SUPPLEMENTAL DATA

Supplemental Table S1 and Supplemental Figs. S1–S3: https://doi.org/10.5281/zenodo.10212041.

GRANTS

Support for this study was provided by the National Heart, Lung, and Blood Institute Grants K08HL132055 and R01HL151530 (to K.S.), R01HL126514 (to L.A.S.), F32HL124727 and K08HL133475 (to J.C.H.), K23HL153781 (C.S.), and R01HL152724 (to P.J.L.). Additional support includes JHU Pulmonary/Critical Care Medicine INSPIRE award (to K.S.). The H-NMR metabolomics work is supported by NIH Grant U2C-DK119886, Common Fund Data Ecosystem (CFDE) Grant 3OT2OD030544, and Metabolomics Consortium Coordinating Center (M3C) Grant 1U2C-DK119889.

DISCLOSURES

Larissa Shimoda is an editor of American Journal of Physiology-Lung Cellular and Molecular Physiology and was not involved and did not have access to information regarding the peer-review process or final disposition of this article. An alternate editor oversaw the peer-review and decision-making process for this article. None of the other authors has any conflicts of interest, financial or otherwise, to disclose.

AUTHOR CONTRIBUTIONS

N.P., S.M., S.V., P.J.L., E.B., L.A.S., and K.S. conceived and designed research; N.P., X.Y., H.P., S.M., Z.H., J.F., P.P., C.Z., W.P. L.S., H.J., J.C.H., S.V., P.J.L., S.A.G., A.L., and K.S. performed experiments; N.P., H.P., S.M., Z.H., J.F., P.P., C.Z., W.P., S.S., L.S., H.J., J.C.H., W.M.O., S.V., P.J.L., S.A.G., C.E.S., A.L., and K.S. analyzed data; N.P., H.P., Z.H., J.F., C.Z., W.P., S.S., L.S., H.J., J.C.H., W.M.O., P.J.L., S.A.G., E.B., C.E.S., A.L., L.A.S., and K.S. interpreted results of experiments; K.S. prepared figures; K.S. drafted manuscript; W.M.O., S.A.G., A.L., L.A.S., and K.S. edited and revised manuscript; N.P., X.Y., H.P., S.M., Z.H., P.P., W.P. S.S., W.M.O., S.V., P.J.L., E.B., C.E.S., A.L., L.A.S., and K.S. approved final version of manuscript.