The Metabolic and Proliferative State of Vascular Adventitial Fibroblasts in Pulmonary Hypertension is Regulated through a MiR-124/PTBP1/PKM Axis

Abstract

Background

An emerging “metabolic theory” of pulmonary hypertension (PH) suggests that cellular and mitochondrial metabolic dysfunction underlies the pathology of this disease. We and others have previously demonstrated the existence of hyper-proliferative, apoptosis-resistant, pro-inflammatory adventitial fibroblasts from human and bovine hypertensive pulmonary arterial walls (PH-Fibs) exhibit constitutive reprogramming of glycolytic and mitochondrial metabolism, accompanied by an increased ratio of glucose catabolism through glycolysis versus the TCA cycle. However, the mechanisms responsible for these metabolic alterations in PH-Fibs remain unknown. We hypothesized that, in PH-Fibs, miR-124 regulates polypyrimidine tract binding protein 1 (PTBP1) expression to control alternative splicing of pyruvate kinase muscle isoforms 1 and 2 (PKM1 and PKM2) resulting in an increased PKM2/PKM1 ratio which promotes glycolysis and proliferation even in aerobic environments.

Methods

Pulmonary adventitial fibroblasts were isolated from calves and humans with severe PH (PH-Fibs) and from normal subjects (CO-Fibs). PTBP1 gene knockdown was achieved via PTBP1-siRNA, restoration of miR-124 was performed with miR-124 mimic. TEPP-46 and Shikonin were utilized to manipulate PKM2 glycolytic function. HDACi were used to treat cells. Metabolic products were determined by Mass spectrometry-based metabolomics analyses (UHPLC-MS), and mitochondrial function was analyzed by confocal microscopy and spectrofluorometry.

Results

We detected an increased PKM2/PKM1 ratio in PH-Fibs compared to CO-Fibs. PKM2 inhibition reversed the glycolytic status of PH-Fibs, decreased their cell proliferation and attenuated macrophage IL-1β expression. Further, normalizing the PKM2/PKM1 ratio in PH-Fibs by miR-124 overexpression or PTBP1 knockdown reversed the glycolytic phenotype (decreased the production of glycolytic intermediates and byproducts, i.e. lactate), rescued mitochondrial reprogramming and decreased cell proliferation. Pharmacological manipulation of PKM2 activity with TEPP-46 and Shikonin, or treatment with histone deacetylase inhibitors (HDACi), produced similar results.

Conclusions

In PH, miR-124, through the alternative splicing factor PTBP1, regulates the PKM2/PKM1 ratio, the overall metabolic, proliferative and inflammatory state of cells. This PH phenotype can be rescued with interventions at various levels of the metabolic cascade. These findings suggest a more integrated view of vascular cell metabolism, which may open unique therapeutic prospects in targeting the dynamic glycolytic and mitochondrial interactions and between mesenchymal inflammatory cells in PH.

INTRODUCTION

Pulmonary hypertension (PH) is a prevalent co-morbid condition that significantly worsens morbidity and mortality in patients with a wide variety of disorders¹. In all these circumstances PH is consistently associated with early and persistent perivascular inflammation, fibroproliferative changes and pulmonary arterial remodeling^{2, 3}. This remodeling involves an imbalance of cell proliferation vs. cell death, which, taken in conjunction with inflammation, has led to the hypothesis that the cellular and molecular features of PH resemble hallmark characteristics of cancer^{4, 5}. The fact that changes in cell metabolism in cancer cells, as well as in cells in the surrounding stroma, are essential for cancer cells to proliferate, migrate, and exhibit pro-inflammatory characteristics is now well recognized. As such, there is an intense effort in the cancer field to define the mechanisms regulating the links among changes in metabolism, growth, and inflammation, as they may offer new opportunities for therapy. Strikingly, metabolic changes, resembling those observed in cancer, have also recently been reported in PH^{4, 6}. These changes have been described to occur in smooth muscle cells⁶ endothelial cells⁷, and fibroblasts^8–10 and perhaps precede the development of pulmonary hypertension¹¹. These observations have led to a “metabolic theory” of PH whereby mitochondrial and cytosolic defects drive a “Warburg-like” cell phenotype (described originally for cancer cells, which reprogram metabolic pathways toward aerobic glycolysis to support high proliferation) that can explain the molecular and functional abnormalities seen in PH cells, including excessive proliferation, apoptosis resistance, and inflammation^{6, 8, 9, 12}.

Recently, however, it has become increasingly clear that there is a tremendous heterogeneity in how metabolic needs are met in different environmental/micro-environmental conditions^{13, 14}. This dynamic character of the cellular metabolic network is commonly known as “metabolic plasticity” and is now recognized as an important feature of cell physiology. Thus, it is possible that the different cells that comprise the pulmonary vascular wall might utilize distinct mechanisms to modulate metabolic and functional responses to environmental stress or injury.

Fibroblasts comprise the principal cell population of the adventitia. In both the normal pulmonary and systemic vasculature they are known to regulate vascular function through continuous production of extracellular matrix and matricellular proteins^{3, 15}. Because of these functions, fibroblasts, exhibit a distinct metabolic phenotype compared to many other non-transformed cells even when not dividing¹⁶. Additionally, these cells are often the first to become activated, proliferate, and differentiate in response to injury^{3, 15}. We have recently shown that pulmonary artery adventitial fibroblasts from calves with severe PH and humans with IPAH (PH-Fibs) exhibit a distinct metabolic phenotype that is directly responsible for many aspects of their functional phenotype and could be a potential biochemical feature for cytotoxic drug selection. However, at present, little is known about the mechanisms that contribute to the generation and maintenance of the metabolic abnormalities observed in PH-Fibs. Deciphering the mechanisms that underline this distinct metabolic reprogramming in PH-Fibs as well as other cells in the hypertensive vessel wall and identification of a novel, selective, and promising molecular target to alleviate PH is the over-arching goal of our work.

In many rapidly proliferating cells and most cancer cells, pyruvate kinase, the enzyme that produces pyruvate and ATP, has been implicated as a critical determinant of the metabolically altered cell phenotype^{17, 18}. It is argued that pyruvate is at the center of the entire metabolic network¹⁴. Indeed pyruvate occupies the junction between cytosolic and mitochondrial metabolism and the pattern of pyruvate generation and disposition dictates in part the metabolic status of the cell. Pyruvate is synthesized by pyruvate kinase (PK), which exists as either or both the M1 or M2 isoform in most cell types. PKM is alternatively spliced to produce the PKM1 or PKM2 isoforms, which contain exon 9 or exon 10, respectively and the regulation of the balance between these isoforms in cells is complex^{14, 19}. PKM1 and PKM2 are functionally different. PKM1 exists constitutively in a high-activity tetrameric form while the PKM2 exists in both a low-activity dimeric form and a high-activity tetrameric form²⁰. The low activity dimeric PKM2 is a very important driver for glycolysis by promoting conversion of pyruvate to lactate, whose formation is promoted by tyrosine kinases-mediated phosphorylation that can be stimulated by environmental cues and growth factor signaling. On the other hand, high activity PKM2 and PKM1 tetramers drive the tricarboxylic acid (TCA) cycle.^{17, 18}. Recent studies suggest that heightened expression of PKM2 or an elevated PKM2/PKM1 ratio is critical for the maintenance of cancer cell growth¹⁷. Importantly, cells with an elevated PKM2/PKM1 ratio slow the production of pyruvate in response to pro-proliferative signaling enabling utilization of glycolytic intermediates for biosynthesis of cellular building blocks such as nucleotides, amino acids, and one-carbon donors²¹. Neither the state of PKM isoform expression nor its function has been examined in PH-fibroblasts or other pulmonary vascular cells to our knowledge.

The state of PKM isoform expression is controlled by three heterogeneous nuclear ribonucleoproteins (hnRNPs): polypyrimidine tract-binding protein 1 (PTBP1, also known as hnRNPI), hnRNPA1, and hnRNPA2, that bind repressively to sequences flanking exon 9. In the presence of these PKM alternative splicing proteins, exon 10 is included in the mature PKM transcript while exon 9 is excluded, resulting in a high PKM2/PKM1 ratio^{19, 22}. Recent studies conducted in fibroblasts as well as different cell types, including ours, have implicated hnRNPs as downstream targets of miRNAs, including miR-124^23–26. Importantly, we proved that miR-124 regulated PH-Fibs proliferation through its direct target PTBP1²⁴. A recent publication further showed that microRNA-124 inhibits cancer cell growth through a PTBP1/PKM1/PKM2 feedback cascade in colorectal cancer²⁵. However, the function and relationships between miR-124, PTBP1, PKM1, and PKM2 in vascular cells have not been elucidated. We thus sought to test the hypothesis that, in PH-Fibs, high PTBP1 expression controlled by silenced miR-124 mechanistically underlies the elevated PKM2/PKM1 ratio, metabolic reprogramming and heightened proliferative capacity in PH-Fibs.

MATERIALS AND METHODS

A detailed Methods section is provided in the Supplement Materials.

Animals

Neonatal calves and C57BL/6 mice exposed to chronic hypoxia were used in this study. All animal procedures were performed under the Guidelines for Animal Experimentation established and approved by the University of Colorado Anschutz Medical Campus and Colorado State University IRBs.

Cell Isolation and Culture

Bovine pulmonary artery adventitial fibroblasts were isolated from normal control calves or calves with severe experimental hypoxic pulmonary hypertension as previously described^{8, 24} (Supplemental Table 1). Human pulmonary artery fibroblasts were derived from patients with idiopathic undergoing lung or from control donors undergoing lobectomy or pneumonectomy as previously described^{8, 9} (Supplemental Table 2). Cells were cultured under normoxic conditions. All experiments were performed on cells at passages 5–8.

Proliferation assay

Cell proliferation was evaluated by using CyQUANT proliferation analysis Kit (Thermo Fisher Scientific, REF# C7026).

Real-time RT-PCR

Total RNA from fibroblasts was extracted and reverse-transcribed to cDNA as described previously²⁴. All primer sets for real-time RT-PCR are listed in Supplemental Table 3. Real-time RT-PCR product analysis using iQ SYBR Green Supermix (Bio-Rad) in triplicate on the CFX96 Real-Time System (BioRad). Results were normalized to control groups using the delta-delta CT method.

Western blot

Western blot was performed as previously described²⁴. The specific antibodies for PTBP1 (NOVUS #H00005725-M01), PKM1 (NOVUS #NBP2-14833), PKM2 (NOVUS #NBP1-48308), β -actin (SIGMA #A5316) were used. Primary antibodies of phosphorylated PDH, PDH, NDUFS4, and Tim23 were purchased from Abcam (Cambridge, MA). Signal was detected using ECL (Thermo Scientific). Quantitative analysis was performed by Image J software densitometry. Native gel electrophoresis assays were performed to evaluate the formation of tetrameric and dimeric PKM2.

Cell transfection

Transfection was performed with 50 nmol/L of miR-124 mimic, scramble, or siRNA targeting PTBP1 or PKM2 (Supplemental Table 3) using DharmaFECT Transfection Reagents as previously described²⁴. The cells were harvested for mRNA expression after 48hrs transfection and for protein expression after 72hrs transfection. The cells were harvested for UHPLC-MS analysis and mitochondria metabolism analysis after 72hrs transfection.

PTBP1 Plasmid construction

The plasmids overexpressing PTBP1 from the CMV promoter, which contained the full-length PTBP1 cDNA, but not its authentic 3’UTR that is resistant to miR-124-mediated inhibition were constructed as previous described²⁴.

PKM splicing reporter

The PKM splicing reporter was constructed by PCR-mediated amplification of the human PKM gene from exon 8 to exon 11 including introns from human genomic DNA and inserted downstream of the CMV promoter of the pcDNA3.1 vector (Invitrogene). The full-length PKM splicing reporter was sequenced.

Pharmacologic inhibition of PKM2 glycolytic function with TEPP-46 and Shikonin

The cultured cells were treated with a tetrameric PKM2 activator, TEPP-46 (30, 50, 75,100 µM, EMD Millipore Corp, Billerica, MA) or PKM2 glycolytic activity inhibitor, Shikonin (1.0 and/or 0.5 µM, EMD Millipore Corp, Billerica, MA). The cells were harvested for protein expression after 72 hrs treatment. The cells were harvested for UHPLC-MS analysis and mitochondria metabolism analysis after 72 hrs treatment. Following 48 hrs of treatment, condition medium was collected for experiments examining the effect of PH-Fibs on macrophages activation (see supplemental methods).

HDAC inhibitor treatment

The cultured cells were treated with pan-HDAC inhibitor, suberoylanilide hydroxamic acid (SAHA; 10 µM, ChemieTek, Indianapolis, IN) or class I HDAC inhibitor apicidin (3 µM) (Enzo, New York, NY) as previously described²⁴. The cells were harvested for mRNA expression after 48 hrs treatment and for protein expression after 72 hrs treatment. The cells were harvested for UHPLC-MS analysis and mitochondria metabolism analysis after 72 hrs treatment.

RNA Sequencing (RNA-Seq)

As described previously⁸, RNAs from bovine CO-Fibs and PH-Fibs (n=6, each group) were used to perform RNA sequencing analysis.

Ultra-high Pressure Liquid Chromatography-Mass Spectrometry-Based Metabolomics Analyses

As described previously⁸, briefly, CO-Fibs and PH-Fibs 72 hrs post-transfection with siPKM2 or miR-124 mimic or siPTBP1, or post-treatment with TEPP-46, Shikonin, or HDACis (n=3, each group) were extracted (2×10^6 cells/ml buffer) for Metabolomics Analyses through an ultra-high performance chromatographic system (UHPLC - Vanquish, Thermo Fisher).

Semiquantification of matrix superoxide release rate (Jm) using confocal microscopy

MitoSOX Red (Life Technologies) was used to detect in situ surplus O₂^•− release to the matrix (Jm) as described previously⁹.

Semiquantification of cytosolic reactive oxygen species (ROS) levels using confocal microscopy and spectrofluorometry

Carboxy-H2DCFDA (LifeTechnologies) was used for microscopic ROS detection as described previously⁹

Semiquantification of mitochondrial membrane potential using confocal microscopy

As described previously⁹, semi-quantification of mitochondrial membrane potential was performed using JC1 probe (Life Technologies)

High-resolution respirometry

As described previously⁹ O₂ consumption was measured using an Oxygraph 2k (Oroboros, Innsbruck, Austria).

Statistical analysis

Values were expressed as mean ± SEM. For basic comparisons of two gaussian distributed sample sets, we used student t-tests, and for unequal variance sample sets we used Welch's t-test. GraphPad Prism software version 7.03 was used to perform two-way repeated measures ANOVA followed by Sidak's multiple comparisons test and one-way ANOVA with Tukey multiple column comparison test when comparing means amongst a set of means resulting from different conditions. Differences with p values <0.05 were considered statistically significant.

RESULTS

The level of PKM2 relative to PKM1 is high in PH-Fibs, accompanied by decreased mitochondrial pyruvate carrier (MPC) and SIRT3 expression

To address the possible role of PKM in the distinct metabolic phenotype of PH-Fibs^{8, 9}, PKM1 and PKM2 mRNA and protein levels in CO- and PH-Fibs were quantified. Compared to CO-Fibs, the hyper-proliferative and pro-glycolytic bovine and human PH-Fibs exhibited significantly decreased levels of PKM1 mRNA (Fig. 1A, I) and protein (Fig. 1C, K, E, M) expression. Bovine PH-Fibs exhibited increased PKM2 protein expression (Suppl. Fig. 1B), though no significant increase in PKM2 mRNA levels was observed (Suppl. Fig. 1A). In humans, there was no significant difference between CO-Fibs and PH-Fibs’ levels of PKM2 mRNA and protein (Suppl. Fig. 1C, D). Overall the PKM2/PKM1 ratio (RNA and protein) in both bovine (Fig. 1B, D) and human (Fig. 1J, L) PH-Fibs was dramatically increased, which is primarily due to the reduced PKM1 levels (Fig. 1 A, C, I, and K). Consistent with in vitro data, immunofluorescent staining of PKM1 and PKM2 in CO and PH calf lungs demonstrates that PKM1 expression is very low (if not absent) in most cells of the remodeled adventitia, whereas PKM2 is expressed in mostly all cells in the remodeled adventitia (Suppl. Fig. 1E).

Fig. 1. The levels of PKM2 relative to PKM1 is increased in PH-Fibs compare to CO-Fibs, accompanied by decreased MPC and SIRT3 expression.

Real-time RT-PCR analysis of PKM1 mRNA levels in CO- and PH-Fibs (A, bovine; I, human). PKM2/PKM1 mRNA ratio in CO- and PH-Fibs (B, bovine; J, human). Quantification of PKM1 protein levels (C, bovine; K, human) and PKM2/PKM1 protein ratio in Fibs (D, bovine; L, human) are presented. (Based on multiple WBs from several Fibs). Representative Western-blot of PKM1 and PKM2 protein expression in CO- and PH-Fibs (E, bovine; M, human). Real-time RT-PCR showed mRNA levels of MPC1 (F, bovine; N, human), MPC2 (G, bovine; O, human) and SIRT3 (H, bovine; P, human) in CO- and PH-Fibs. (Data are presented as mean ±S.E.M, n=4–8, *P<0.05 vs. CO-Fibs).

Cytosolic pyruvate synthesized by PK can either be transported into mitochondria or be further processed to lactate under hypoxic conditions or as in PH-Fibs, through aerobic glycolysis in cells expressing the glycolytic (Warburg) program. In previous studies⁸, we performed NMR and MS-based tracing experiments with heavy labeled substrates ¹³C₁- and U-¹³C-glucose to conclude that glucose catabolism through mitochondria is decreased in PH-Fibs, suggesting decreased shuttling of glycolytic pyruvate into mitochondria. Pyruvate is transported into mitochondria by the MPC, which is composed of the MPC1 and MPC2 subunits both of which exhibit decreased expression in cancer cells, with MPC1 being the most strongly and consistently affected²⁷. Thus we sought to investigate the expression of MPC as another indicator of pyruvate subcellular compartmentalization and utilization. Importantly, we found decreased MPC1 expression in both bovine and human PH-Fibs (Fig. 1F, N) and decreased MPC2 expression in human PH-Fib (Fig. 1O). This result was confirmed by RNA-seq in which the MPC1 transcript was reduced in bovine PH-Fibs (Suppl. Fig. 1F). It is interesting to note that decreased activity of the mitochondrial deacetylase sirtuin-3 (SIRT3), which is known to modulate MPC expression, is associated with the development of PH²⁸. Thus we sought to determine whether the expression of SIRT3 would be altered in PH-Fibs. We found that SIRT3 mRNA expression is decreased in both bovine and human PH-Fibs (Fig. 1H, P). These observations are consistent with the decreased shuttling of glycolytic pyruvate into mitochondria in PH-Fibs.

Collectively, these findings, taken in the context of our previous report of increased pyruvate dehydrogenase kinase-1 (PDK1) activity⁹, begin to explain the control of pyruvate and the metabolic phenotype of PH cells.

PKM2 knockdown by siRNA reverse the metabolic and proliferative phenotypes of human PH-Fibs

The observation of PKM2/PKM1 ratios increasing in PH-Fibs in comparison to CO-Fibs may either be correlative or causative with respect to the glycolytic metabolic reprogramming observed in PH-Fibs^{9, 29}. To test the latter hypothesis and directly address the role of altered PKM2/PKM1 ratio in the metabolic phenotype of PH-Fibs, we decreased the PKM2/PKM1 ratio by knocking down PKM2 using specific PKM2 siRNA in human PH-Fibs. As a result, PKM2 mRNA (Fig. 2A) and PKM2/PKM1 mRNA ratio (Fig. 2B) were decreased dramatically in PH-Fibs treated with PKM2 siRNA, which was further supported by reduced PKM2 protein expression (Fig. 2C). As expected, PKM1 protein levels were not decreased in PKM2 siRNA treated cells, indicating the specificity of the PKM2 siRNA (Fig. 2C). Steady state UHPLC-MS demonstrated decreased lactate production (Fig. 2D), as well as decreased levels of other metabolites involved in glycolysis and mitochondrial oxidative phosphorylation in human PH-Fibs post-transfection with PKM2 siRNA, as gleaned with multivariate analyses (a detail of hierarchical clustering analysis) is shown in Fig. 2E. Partial least-square discriminant analysis (Fig. 2F) showed siPKM2 altered the overall metabolic status of human PH-Fibs towards to CO-Fibs. We also observed that the proliferation of PH-Fibs was significantly decreased by PKM2 knockdown using CyQUANT assay (Fig. 2G). To confirm the results we evaluated the effects of PKM2 shRNA and verified significant decreases in proliferation (Suppl. Fig. 2). Collectively, these results demonstrated an important role of the increased PKM2/PKM1 ratio in PH-Fibs in controlling the metabolic “switch” from glucose oxidation to aerobic glycolysis. Normalization of the PKM2/PKM1 ratio through PKM2 silencing abrogates metabolic dysfunction and decreases cell proliferation.

Fig. 2. PKM2 siRNA reduces PKM2 levels, PKM2/PKM1 ratio, lactate production and cell proliferation in human PH-Fibs.

(A). Real-time RT-PCR analysis of PKM2 mRNA in human PH-Fibs targeted with PKM2 specific siRNA (siPKM2) or scrambled siRNA (SCR) (n=3, *P<0.05, compared to scrambled siRNA treated PH-Fibs) (B). PKM2/PKM1 mRNA ratio in human PH-Fibs targeted with SCR or siPKM2 (n=3, *P<0.05, compared to scrambled siRNA treated PH-Fibs). (C). Representative of Western-blot analysis of PKM2 and PKM1 protein expression in PH-Fibs targeted with SCR or PKM2 siRNAs (n=3). (D). UHPLC-MS analysis of lactate production in PH-Fibs targeted with SCR or siPKM2 after transfection (n=3, *P<0.05, compared to scrambled siRNA treated PH-Fibs). (E). Heat-map generated from MS (a detail of hierarchical clustering analysis) showed the levels of metabolites involved in glycolysis and mitochondrial oxidative phosphorylation post-siRNA transfection. (F) Data generated from MS (partial least-square discriminant analysis) showed siPKM2 altered the overall metabolic status of human PH-Fibs. (G) CyQUANT assay showed cell proliferation decreased by siPKM2. (n=3, *P<0.05, compared to scrambled siRNA treated PH-Fibs-repeated measures ANOVA).

Pharmacologic inhibition of PKM2 glycolytic function by TTEP-46 and Shikonin in PH-Fibs rescues glycolytic reprogramming, decreases cell proliferation and inflammatory activation of macrophages

Pharmacological intervention with the small molecule (TEPP-46) to promote PKM2 tetramer formation has been shown to reverse metabolic abnormalities and mitochondrial dysfunction resulting in decreased cancer cell proliferation³⁰ and kidney pathology reversion³¹. Consistent with these observations, we found that TEPP-46 treatment of human PH-Fibs stabilized the PKM2 tetramer (Suppl. Fig. 3A), with no effect on PKM1 expression (Suppl. Fig. 3B). This resulted in decreased lactate generation in TEPP-46-treated PH-Fibs (Fig. 3A), as well as a general correction of PH-Fibs metabolism. The levels of additional metabolites that accumulate in PH Fibs were normalized upon TEPP-46 treatment, including glycolytic intermediates (glucose, glyceraldehyde 3-phosphate, fructose 1-6-bisphosphate and lactate), metabolites associates with redox homeostasis (glutathione), mitochondrial function (α-Ketoglutarate, succinate and citrate) and pentose phosphate pathway (alpha-D-Ribose 1-phosphate; Fig. 3B). The beneficial effect of TEPP-46 on PH-Fibs metabolism was paralleled by a decrease in cell proliferation (Fig. 3C). Importantly, we also observed that TEPP-46 rescued the mitochondrial bioenergetics by increasing ATP synthesis in bovine PH-Fibs, illustrated by the significantly increased respiratory control ratio (Fig. 3D) and significant elevation of maximal respiratory capacity (Fig. 3E). These results indicate that allosterically regulated PKM2 activity underlies metabolic reprogramming of PH-Fibs, and that a tetrameric activator of PKM2 is sufficient to inhibit PKM2 glycolytic function, and thus reverse the glycolytic and proliferative phenotypes of human PH-Fibs.

Fig. 3. Treatment with TTEP-46 and Shikonin decreases lactate production and cell proliferation, and rescues mitochondrial bioenergetics in PH-Fibs.

(A) Lactate production in human PH-Fibs after treatment by TEPP-46 (100 µM). (F) Glucose and lactate in human CO and PH-Fibs after treatment by Shikonin (0.5 µM and 1.0 µM) (n=3, *P<0.05, compared to DMSO treated PH-Fibs). (B, G) Heat-map generated from MS showed the levels of metabolites involved in glycolysis, serine biosynthesis and mitochondrial oxidative phosphorylation in human Fibs with treatment of TEPP-46 and Shikonin. (H) Overall metabolic phenotype of CO and PH-Fibs after DMSO or Shikonin treatment as gleaned by Partial Least Square-Discriminant analysis. (C, I) Proliferation assay showed human PH-Fibs proliferation after treatment by TEPP-46 and Shikonin. (n=3, *P<0.05, compared to DMSO treated PH-Fibs-repeated measures ANOVA). Mitochondrial respiratory control ratio (D) and maximal respiratory capacity (E) are presented in bovine PH-Fibs after TEPP-46 (100µM) treatment. (n=3, *P<0.05, compared to DMSO treated PH-Fibs).

To further evaluate the role of PKM2 in controlling PH-Fib phenotype, we examined the effects of Shikonin on PH-Fibs (bovine and human) based on recent reports indicating that it is a potent PKM2 glycolytic function inhibitor in cancer cells and macrophages³². Metabolomics analyses revealed that treatment of PH-Fibs with Shikonin was sufficient to normalize the metabolic phenotypes back to control values, as gleaned by multivariate analysis of bovine (Suppl. Fig. 4A) and human (Fig. 3H). In human PH-Fibs, both 0.5 µM and 1.0 µM of Shikonin were sufficient (1.0 µM of Shikonin resulted in more significant effects) to rescue the hyperglycolytic phenotype (as determined by higher steady state levels of glucose and lactate in comparison to CO-Fibs) (Fig. 3F). In bovine PH-Fibs, a concentration of 0.5 µM of Shikonin elicited a similar response (Suppl. Fig. 4B). Notably, Shikonin treatment decreased additional glycolytic intermediates (D-Fructose 1-6-bisphosphate, phosphoenolpyruvate and glyceraldehyde 3-phosphate) and metabolites associates with redox homeostasis (cysteine and dehydroascorbate), mitochondrial function (succinate and fumarate) and pentose phosphate pathway (alpha-D-Ribose 1-phosphate) in bovine (Suppl. Fig. 4C) and/or human PH-Fibs (Fig. 3G). Consistent with this metabolic rescue, we observed that Shikonin significantly reduced proliferation of bovine (Suppl. Fig. 4D) and human PH-Fibs (Fig. 3I).

We previously established that fibroblasts play an essential role in initiating and perpetuating the perivascular inflammation that characterizes PH^{8, 29, 33}. They induce a distinct alternative macrophage phenotype, which includes expression of mRNA for the pro-inflammatory cytokine IL-1β^{8, 29, 33}. To further determine the role of PKM2 in fibroblast mediated activation of macrophages, we exposed mouse bone marrow derived macrophages (BMDM) to conditioned media (CM) from Shikonin treated PH-Fibs. We found expression of IL-1β was significantly reduced in BMDMs exposed to CM from Shikonin treated PH-Fibs relative to DMSO treated PH-Fibs (Suppl. Fig. 5A). We also treated mouse BMDMs with Shikonin (Suppl. Fig. 5B) or TEPP-46 (Suppl. Fig. 5C) and observed reduced expression of IL-1 β mRNA relative to DMSO treated BMDMs. We also found TEPP-46 inhibited LPS induced IL-1β expression (Suppl. Fig. 5D), which was previously shown to be PKM2 dependent³⁴.

MiR-124 regulates PKM2/PKM1 expression through PTBP1, which regulates PKM isoform expression by alternative splicing

Having confirmed the direct role of excess PKM2 in the metabolic reprogramming of PH-Fibs, we sought to determine the molecular mechanisms underlying altered PKM2/PKM1 ratios in PH-Fibs. PTBP1, an alternative splicing factor, promotes generation of the PKM2 isoform by regulating PKM RNA splicing in tumor cells^{19, 22}. We reported previously that PTBP1 is highly expressed in PH-Fibs and that PTBP1 is a direct target of miR-124 whose expression is significantly reduced in PH-Fibs²⁴. We observed that either overexpression of miR-124 or silencing of PTBP1 in bovine PH-Fibs reduced the PKM2/PKM1 ratio (Fig. 4 A and B), indicating that miR-124 reduction and/or PTBP1 overexpression are critically important for the altered PKM2/PKM1 ratio observed in PH-Fibs. To determine whether miR-124 controls PKM2/PKM1 ratio by controlling PTBP1 levels, we co-transfected miR-124 mimics with a PTBP1 overexpression vector, pcDNA3.1-CMV-PTBP1, which contained the full-length PTBP1 cDNA, but not its authentic 3’UTR that is resistant to miR-124-mediated inhibition. Importantly, transfection of miR-124 resistant pcDNA PTBP1 abrogated miR-124-mediated PKM2/PKM1 ratio reduction in bovine PH-Fibs (Fig. 4 A, B). This data supported a conclusion that miR-124 regulates PKM2/PKM1 ratio by controlling PTBP1 levels.

Fig. 4. MiR-124 regulates PKM2/PKM1 expression through PTBP1 by alternative splicing.

(A) Real-time RT-PCR showed overexpression of PTBP1 overrode the decreased PKM2/PKM1 ratio induced by miR-124 mimic in bovine PH-Fibs. (B) Representative western-blot of PTBP1, PKM1, and PKM2 in bovine PH-Fibs targeted with indicated treatments. (C) Schematic representation of a PKM splicing reporter driven by CMV promoter. Dashed lines indicate that primers are designed at the junctions of the exons. Arrows represent primers used for real-time RT-PCR. E9/E11 or E10/11, with Vector-R (located in vector sequence) for PKM1 or PKM2 transcripts expressed from the splicing reporter while Vector F with PKM total was used to quantify the total transcripts produced from the splicing reporter. The total transcript results were used to control transfection/expression efficiency of the splicing reporter in CO and PH-Fibs. (D) Relative PKM2 transcripts expressed from the splicing reporter in human CO-Fibs and PH-Fibs (with or without transfection of miR-124 mimic and siPTBP1) were presented. PKM2 levels were presented after calibrated with total PKM transcripts expressed from the splicing reporter to control for transfection and expression variation between CO and PH-Fibs. (n=3; *P<0.05).

To further examine the role of PTBP1 in regulating PKM RNA splicing, we generated a human PKM splicing reporter that contains exon 8 to exon 11 of human PKM gene (Fig. 4C). PKM splicing reporter DNA was transfected into human CO and PH-Fibs and levels of total PKM (by Vector-F/PKM total), PKM1 (by E9/E11/vector-R) and PKM2 (by E10/E11/vector-R) transcripts produced from the splicing reporter were evaluated by qRT-PCR. PKM2 transcripts were found to be highly expressed in PH-Fibs, much higher than that in the CO-Fibs (Fig. 4D). Further, PKM2 transcript levels were reduced by siPTBP1 or miR-124 mimic in PH-Fibs (Fig. 4D). However, in contrast to the endogenous PKM gene, we were not able to detect PKM1 transcripts expressed from the splicing reporter. Nevertheless, the significant reduction of PKM2 by siPTBP1 is sufficient to support a role of PTBP1 in regulating the PKM alternative splicing process in fibroblasts.

To provide a possible explanation as to why PKM1 was dramatically decreased, while PKM2 didn’t increase significantly, we assessed the levels of the total PKM transcripts in CO and PH-Fibs. We observed that compared with CO-Fibs, total PKM expression was decreased in both bovine and human PH-Fibs (Suppl. Fig. 6A, B) by qRT-PCR, a result that was confirmed by RNA-seq in which PKM transcripts were reduced in bovine PH-Fibs by 1.69 fold (Suppl. Fig. 6C). Interestingly, despite the reduction of PKM gene transcription, the levels of PKM2 transcripts and proteins are still maintained or even slightly increased (Suppl. Fig. 1), supporting our hypothesis that PKM2 splicing is favored in PH-Fibs.

It is possible that two other members of the hnRNP family of splicing factors hnRNPA1 and hnRNPA2, also play a role in PKM alternative splicing in PH-Fibs because we found hnRNA2 was increased in bovine PH-Fibs (Suppl. Fig. 7B), and both hnRNPA1 and hnRNPA2 were increased in human PH-Fibs (Suppl. Fig. 7F, G). Furthermore, we observed that knockdown of hnRNPA1 and hnRNPA2 decreased PKM2/PKM1 ratio in both bovine and human PH-Fibs (Suppl. Fig. 7E, J). These complementary findings merit further investigation.

MiR-124 and PTBP1 regulate the relative levels of PKM2 to PKM1 and their restoration reverses the glycolytic switch in PH-Fibs

Having established altered expression and a functional role of increases in the PKM2/PKM1 ratio in PH-Fibs and the role of miR-124/PTBP1 in controlling PKM splicing axis, we wanted to determine whether miR-124 and/or PTBP1 regulates PKM2/PKM1 ratio and thus the metabolic status in both bovine and human PH-Fibs. Using a well-established transfection system described previously²⁴, we transiently transfected mature miR-124 mimic, PTBP1 siRNA or scrambled siRNA sequence (SCR, as control) into the cells. Overexpression of miR-124 or silencing of PTBP1 in both bovine and human PH-Fibs caused a significant decrease in the PKM2/PKM1 mRNA ratio, compared with PH-Fibs treated with SCR, which were also corroborated by protein expression (Fig. 5A, E). These data supported an important role of miR-124 and PTBP1 in regulating PKM2/PKM1 ratio in PH-Fibs.

Fig. 5. MiR-124 overexpression or PTBP1 silencing decreases the PKM2/PKM1 ratio and regulates the metabolic phenotype in PH-Fibs.

Real-time RT-PCR showed PKM2/PKM1 mRNA ratio in CO-Fibs treated with scrambled siRNA (SCR) and PH-Fibs targeted with scrambled siRNA, miR-124 mimics or siPTBP1. (A, bovine; E, human, up panel). Representative Western blots of PKM1 and PKM2 are shown (A, bovine; E, human, low panel). Steady state UHPLC-MS demonstrated decreased lactate production in PH-Fibs post-transfection of miR-124 mimic or siPTBP1 (B, bovine; F, human). MS metabolomic analysis revealed that miR-124 mimic or siPTBP1 altered the overall metabolic status of both bovine (C) and human (G) PH-Fibs toward CO-Fib-like status. Heat-map generated from MS showed the levels of metabolites involved in glycolysis and mitochondrial oxidative phosphorylation post-transfection with miR-124 mimic or siPTBP1. Each column represents an individual sample from each group (D, bovine; H, human). (n=3, *P<0.05, compared to scrambled siRNA treated PH-Fibs).

To determine if reduced miR-124 and/or increased PTBP1 expression is responsible for the metabolic phenotype observed in PH-Fibs, we performed extensive metabolic studies in PA adventitial fibroblasts using UHPLC-MS. We found that miR-124 add back or PTBP1 reduction in PH-Fibs caused a significant decrease in lactate production, compared with PH-Fibs treated with the corresponding scrambles (Fig. 5B, F). This is relevant in the light of the hyperglycolytic phenotype of PH-Fibs⁸ and the role of a Warburg-like phenotype in sustaining proliferation in a cancer-like fashion³⁵. Multivariate analysis of MS metabolomics data revealed that miR-124 mimic and siPTBP1 altered the overall metabolic phenotypes of both bovine and human PH-Fibs toward CO (Principal Component 1, which explains the highest percentage of the metabolic variance across samples, could discriminate PH-Fibs from the other groups for bovine and human cell experiments - Fig. 5C, G). Heat-maps generated from metabolomics analyses showed a reversion back to CO-Fibs levels of the metabolites involved in glycolysis and mitochondrial oxidative phosphorylation post-transfection of PH-Fibs with miR-124 mimic or siPTBP1 (p<0.05 ANOVA) (Fig. 5D, H). Normalization of mitochondrial metabolism is relevant in the light of the previously appreciated role of altered mitochondrial function in sustaining PH Fibs proliferation⁹, a phenomenon accompanied by the accumulation of carboxylic acids to promote lipid anabolism and anaplerosis^{9, 35}. In this view, particularly intriguing was the restoration of serine, glutamine, succinate, α-Ketoglutarate and ribose phosphate to normal following treatment with miR-124 mimic or siPTBP1 (Suppl. Fig. 8). Even though only steady state measurements were performed, these results suggest a role of miR-124 in controlling glutaminolysis/glutathione homeostasis, pentose phosphate pathway and one-carbon metabolism, key regulatory pathways for the biosynthesis of nucleosides and thus contributing to the hyperproliferative phenotype of PH-Fibs^{8, 35}. Taken together, these data demonstrated the functional role of miR-124 and PTBP1 on the PKM2/PKM1 ratio axis, which in turn controls the metabolic “switch” from glucose oxidation to aerobic glycolysis in PH-Fibs.

Normalization of miR-124 or PTBP1 levels reverses mitochondrial abnormalities in PH-Fibs

Increased glycolysis is reported to provide the cell with sufficient building blocks for cell proliferation. Although most tumor cells have functional mitochondria, their activity is altered³⁶. Our recent work suggests that mitochondrial abnormalities may play a central role in PH-Fibs activation⁹. Thus, after having established the role of miR-124 and PTBP1 on aerobic glycolysis in PH-Fibs, we sought to identify miR-124 and PTBP1’s effects on mitochondrial activity. To determine if altered miR-124 and PTBP1 expression is responsible for constitutive reprogramming of mitochondrial metabolism in PH-Fibs, we evaluated respiratory parameters of bovine fibroblasts focusing on intensity and capacity of oxidative phosphorylation along with the activity of respiratory chain complexes with emphasis on Complex I.

Employing a variety of redox-sensitive probes we determined redox status in both the mitochondria and cytosol in bovine PH-Fibs in response to changes in the level of miR-124 or PTBP1 within the cell. We found a significantly decreased phosphorylated PDH/total PDH ratio in PH-Fibs transfected with miR-124 mimic or siPTBP1 (Fig. 6A), which facilitates pyruvate oxidation in the mitochondria of cells. We found that either miR-124 add back or silencing PTBP1 rescued the mitochondrial bioenergetics by increasing ATP synthesis in PH-Fibs, illustrated by increased respiratory control ratio and elevation of maximal respiratory capacity (Fig. 6B, C). Furthermore, we observed rescue of mitochondrial complex I activity (Fig. 6D) which was accompanied by increased expression of NDUFS4 assembly subunit of complex I (Fig. 6E) following either overexpression of miR-124 or silencing of PTBP1. Increased mitochondrial bioenergetics in PH-Fibs with miR-124 overexpression or PTBP1 silencing led to a decrease of mitochondrial membrane potential (Fig. 6F) corresponding to a drop in mitochondrial superoxide production (Fig. 6G), as well as cytosolic ROS (Fig. 6H). Together, these studies demonstrated that restoration of miR-124 or reduction of PTBP1 is sufficient to rescue mitochondrial reprogramming in PH-Fibs.

Fig. 6. MiR-124 overexpression or PTBP1 silencing rescues mitochondrial alterations in bovine PH-Fibs.

(A) Quantification of phosphorylation of PDH expressed as phosphorylated (P-PDH)/total PDH protein ratio in PH-Fibs transfected with miR-124 mimic or siPTBP1 (up panel). Representative western blots for phosphorylated PDH and total PDH in PH-Fibs are shown (low panel). Respiratory control ratio expressed as endogenous (state3)/ATP synthase inhibited (oligomycin, state4) respiratory ratio (B), maximal respiratory capacity expressed as uncoupled (FCCP)/ ATP synthase inhibited (oligomycin, state4) respiratory ratio (C), and in situ quantification of Complex I activity determined by rotenone-inhibited decrease of respiration (100% represents non-treated samples) (D) in CO-Fibs (CO-SCR) and PH-Fibs targeted with miR-124 mimics, or siPTBP1 are shown. (E) Quantification of NDUFS4 subunit of Complex I in PH-Fibs post-transfection by western blot analysis using specific NDUFS4 antibodies (up panel). Representative western blot is shown (low panel), including control of protein loading (Tim23 quantification). Mitochondrial membrane potential expressed by JC1 ratio (F), mitochondrial superoxide expressed by MitoSOX fluorescence increase rate Jm (G), and cytosolic ROS production detected using DCF probe (H) were revealed in CO-Fibs and PH-Fibs after transfection with miR-124 mimic or siPTBP1. (n=3–4, *P<0.05, compared to scrambled siRNA treated PH-Fibs).

HDAC inhibitors normalize the PKM2/PKM1 ratio and reverse the metabolic reprogramming of PH-Fibs

Reversible lysine acetylation has emerged as a critical mechanism for controlling the function of nucleosomal histones as well as diverse non-histone proteins. Roles for histone deacetylases as therapeutic targets in pulmonary hypertension have been increasingly recognized^{37, 38}. We have previously shown that PH-Fibs have increased class I histone deacetylase (HDAC) expression²⁹ and that HDAC inhibitors (HDACi) restore miR-124 expression and decrease PTBP1 expression²⁴. Thus, we were interested in determining whether HDACi would also reverse the metabolic phenotype of PH-Fibs. We observed that incubation of PH-Fibs with the pan-HDAC inhibitor SAHA or with the selective class I HDAC inhibitor Apicidin reduced the PKM2/PKM1 ratio (Fig. 7A, D), decreased glucose uptake and lactate production (Fig. 7B, E), and decreased levels of metabolites involved in glycolysis and mitochondrial oxidative phosphorylation (Fig. 7C, F). These findings provide additional mechanistic underpinnings for targeting histone deacetylases as a novel therapeutic strategy in PH.

Fig. 7. HDAC Inhibitors normalize the PKM2/PKM1 ratio and reverse the glycolytic metabolic phenotype of PH-Fibs.

Real-time RT-PCR showed HDAC inhibitors decreased PKM2/PKM1 mRNA ratio (up panel; A, bovine; D, human). Representative western blot of PKM1 and PKM2 protein levels in these PH-Fibs are shown (low panel; A, bovine; D, human). Steady state UHPLC-MS demonstrated decreased glucose uptake and lactate production in PH-Fibs post-treatment of HDACi (B, bovine; E, human). Heat-map generated from MS showed the levels of metabolites involved in glycolysis, serine biosynthesis and mitochondrial oxidative phosphorylation in PH-Fibs post-treatment of HDACi. Each column represents an individual sample from each group (C, bovine; F, human). (n=3–4, *P<0.05, compared to DMSO treated PH-Fibs).

To investigate the effects of HDACi treatment on mitochondrial function, we performed extensive mitochondrial function studies in bovine PH-Fibs. We observed that HDACi treatment (SAHA and Apicidin) significantly increased respiratory control ratio (Fig. 8A), maximal respiratory capacity (Fig. 8B), Complex I activity (Fig. 8C) and NDUFS4 subunit of Complex I expression (Fig. 8D) in bovine PH-Fibs. We also found mitochondrial superoxide production (Fig. 8E) and cytosolic ROS production (Fig. 8F) decreased after HDACi treatment, which is consistent with recent work by Chen et al.³⁷.

Fig. 8. HDAC Inhibitors rescue mitochondrial alterations in bovine PH-Fibs.

HDACi treatment significantly increased respiratory control ratio (A), maximal respiratory capacity (B), Complex I activity (C) and NDUFS4 subunit of Complex I (D) in bovine PH-Fibs. Mitochondrial superoxide production (E) and cytosolic ROS production (F) decreased after HDACi treatment. (n=3–4, *P<0.05, compared to DMSO treated PH-Fibs). (G) Proposed mechanism contributing to metabolic reprogramming in pulmonary hypertensive fibroblasts. In PH-Fibs, the state of PKM isoform expression is controlled by an alternative splicing complex, composed of PTBP1, hnRNPA1 and hnRNPA2. In the presence of these PKM alternative splicing proteins, especially PTBP1 (a direct target of miR-124), exon 10 is included in the mature PKM transcript while exon 9 is excluded, resulting in a high PKM2/PKM1 ratio. The elevated PKM2/PKM1 ratio is critical for the constitutive reprogramming of glycolytic and mitochondrial metabolism, accompanied by an increased ratio of glucose catabolism through glycolysis versus the TCA cycle and increased ROS levels. PH-Fibs with an elevated PKM2/PKM1 ratio slow the production of pyruvate in response to pro-proliferative signaling enabling utilization of glycolytic intermediates for biosynthesis of cellular building blocks, which promote cell proliferation. Restoration of the PKM2/PKM1 ratio toward normal with miR-124 mimic and siPTBP1, or pharmacologic inhibition of PKM2 glycolytic function with TEPP-46 and Shikonin, or treatment with HDACi reverse glycolytic and mitochondrial reprograming toward normal. These findings provide additional mechanistic underpinnings for potentially targeting the miR-124-PTBP1-PKM axis as a novel therapeutic strategy in PH.

Shikonin treatment attenuates the metabolic, proliferative and inflammatory changes in lungs of hypoxic mice

Because of the striking effects of PKM2 inhibition on PH-Fibs metabolism, proliferation and macrophage interactions, we performed preliminary experiments examining the effects of Shikonin in hypoxic mice (Suppl. Fig. 9A). We found despite the limited number of biological replicates, using Principal Component Analysis of metabolomics, that Shikonin treatment at a dose of 8.0 mg/kg rescued (at least partially) the metabolic alterations observed in the hypoxic lung (Suppl. Fig. 9B). In particular, significant improvements were observed with respect to lung amino acid levels, nitrogen metabolism and glycolysis (Suppl. Fig. 9C). We also examined the effect of Shikonin on cell proliferation, inflammatory cell recruitment and provisional extracellular matrix accumulation, using lungs immune-fluorescently stained for Ki67, CD68 and tenascin C respectively. We observed significant reductions in proliferation and pro-inflammation phenotypes in Shikonin treated lung (Suppl. Fig. 9E). We observed a not significant trend in RV systolic pressure with Shikonin treatment (Suppl. Fig. 9D) in hypoxic mice. Based on these preliminary results, further extensive evaluation of the effects of manipulating PKM2 in various models of pulmonary hypertension and vascular remodeling would be indicated.

DISCUSSION

An emerging “metabolic theory” of pulmonary hypertension (PH) suggests that common metabolic and mitochondria-based cell dysfunction may underlie the disease’ pathology^{4, 39}. Thus, numerous molecular abnormalities described in pulmonary vascular cells may have a common denominator, i.e. mitochondrial suppression of glucose oxidation with upregulation of cytoplasmic glycolysis, - similar to cancer cells. We showed previously that PH-Fibs exhibit profound metabolic changes^{8, 9, 24, 29} that may begin to explain functional abnormalities, including hyper-proliferation, apoptosis resistance, and a pro-inflammatory state. These observations emphasize that the metabolic reprogramming of PH-Fibs (and/or other vascular wall cells) present a potential therapeutic strategy. We now show that similar to cancer cells, high PTBP1 expression, controlled by decreased expression of miR-124, mechanistically underlies the elevated PKM2/PKM1 ratio and subsequent metabolic reprograming in PH-Fibs. In an accompanying paper we demonstrate that a very similar miR-124-PTBP1-PKM-signaling axis also controls the metabolic and functional state of pulmonary artery endothelial cells and blood outgrowth endothelial cells in human iPAH and fPAH (Caruso et al). The present study also demonstrates a distinct PKM profile of pulmonary artery adventitial fibroblasts, which is different from that observed in many cancer cells. Most cancer cells assessed in cell culture preferentially express PKM2 over PKM1, and PKM2 is important for cancer metabolism and tumor growth^{17, 19}. Interestingly, in PH-Fibs, PKM1 expression is dramatically decreased, but PKM2 is not remarkably increased. In other words, decreased PKM1 is the main contributor to the increased PKM2/PKM1 ratio and the low PK activity in PH-Fibs and not elevated PKM2 per se. These results support the idea that the absolute elevations in PKM2 are not required for cell metabolic reprogramming. Rather the low PK activity, due primarily to low PKM1 expression, is the major contributor to the PH-Fibs metabolic switch. Further, as we show in this study, in PH-Fibs, overexpression of miR-124, knock down of PTBP1 and HDACi treatment all restore expression of PKM1, normalize PKM2/PKM1 ratio, and restore the metabolic phenotype toward control cells (Fig. 8G). These results, including upregulation of hnRNPA1/A2 in human PH-Fibs, are consistent with observations in certain cancer cells where downregulation of the splicing factors hnRNPA1/A2 and PTBP1 rescued PKM1 expression, and decrease aerobic glycolysis¹⁹. Collectively these findings provide additional mechanistic underpinnings for potentially targeting the miR-124-PTBP1-PKM axis as a novel therapeutic strategy in PH.

Our findings of what appears to be a dynamic and flexible metabolic network controlled principally at the level of pyruvate metabolism through changes in the pyruvate kinase isoform expression are compatible with the emerging principle of metabolic plasticity in cells throughout the body. Recent studies support the idea that, to a greater or lesser extent, most cells carry out metabolism in very distinct ways¹⁴. Intriguing observations from Lemmons et al. suggest that fibroblasts exhibit a distinct metabolic phenotype under conditions of both quiescence and proliferation¹⁶. Even under quiescent conditions, fibroblasts exhibit a high metabolic activity compared to other cells, which is thought to be due in part to their role in the synthesis, breakdown, re-synthesis of proteins and lipids and the secretion of extracellular matrix proteins. Because they are often the first to become activated in response to vascular injury, it is clear that fibroblasts must have a capability of rapidly adapting to their microenvironment and, as we know, may accomplish this by rapidly altering their metabolic network in response to environmental needs. Numerous studies support the idea that the selection for absolute or relative increases of PKM2 in proliferating cells promotes a unique metabolic program through reduced PK enzymatic activity. In particular, low PK enzymatic activity, as would occur when PKM2 is preponderant, has been proposed to divert glycolytic intermediates toward biosynthetic pathways such as the pentose phosphate pathway and serine biosynthesis^{13, 40}. Consistent with studies in cancer, our current and previous tracing experiments in PH-Fibs clearly indicated increased anabolic fluxes for nucleoside synthesis, serine biosynthesis, and pentose phosphate pathway activation⁸. Here, small molecule regulation of PKM2 by stabilization of its tetrameric form with TEPP-46, pharmacologic inhibition of PKM2 glycolytic activity with Shikonin, and/or restoration of the PKM2/PKM1 ratio toward normal in PH-Fibs with miR-124 mimic, siPTBP1 and especially by HDACi led to decreased levels of ribose phosphate and serine, intermediates required for the biosynthesis of nucleotides and proteins, or the generation of reducing equivalents necessary to fuel lipid synthesis and keep in check the redox poise (e.g. NADPH)¹³. It is also important to note that PKM2 enzymatic activity can be inhibited through endogenous mechanisms, which support the idea that selection for PKM2 in certain highly synthetic or proliferating cells can confer metabolic flexibility by allowing PK enzymatic activity to be turned on or off in response to the environment. Expression of PKM1, on the other hand, likely renders cells less responsive to intracellular cues on energy state, nutrient availability, and growth.

Our observations suggest that different from the metabolic inflexibility observed in certain cancer cells, these hypertensive fibroblasts are still able to change rapidly, especially in response to environment or pharmacologic interventions. Our findings support the idea that pyruvate metabolism can be exquisitely controlled by miR-124 as well as PTBP1. It is known that in oxidative cells, pyruvate is transported into the mitochondria by MPC. Decreased expression and activity of MPC seems to be an essential feature of the metabolic program in many cancer cells. Our observations of decreased MPC1 expression in PH-Fibs are consistent with these observations. It is interesting to note that our observations of decreased SIRT3 expression are also compatible with the reports demonstrating that decreased activity of SIRT3, which is known to modulate MPC activity, are associated with the development of PH²⁸. Relevant also are observations that SIRT3 is downregulated in constitutively activated fibroblasts from systemic sclerosis patients and that its pharmacologic activation mitigates fibrosis⁴¹. It is known that SIRT3 responds directly to mitochondrial NAD⁺ levels. Thus it is likely that the increased NADH/ NAD⁺ ratio in bovine and human PH-Fibs together with intensive glycolysis represses SIRT3 activity. The next step in the carbohydrate oxidation pathway is the conversion of pyruvate to acetyl Co-A in the mitochondrial matrix by pyruvate dehydrogenase (PDH). We recently showed that in PH-Fibs, PDH is markedly inhibited due to activation of PDK1⁹. Our data clearly demonstrate that restoration of miR-124 or reduction of PTBP1 toward normal restored PDH activity and led to the restoration of mitochondrial respiratory capacity, complex I activity, and reductions in mitochondrial membrane potential, superoxide production, and cytosolic ROS production. These results strongly support the idea that miR-124 and PTBP1 play an important role in mitochondrial abnormalities in PH-Fibs likely through control of pyruvate metabolism. Further studies of these metabolic mitochondrial alterations in PH may, therefore, open new therapeutic perspectives in this disease.

MiR-124 may have targets other than PTBP1, which are important in controlling the metabolic, proliferative, and pro-inflammatory phenotype of PH-Fibs. It has been shown that hnRNPA2, as a component of PKM splicing complex, is also a miR-124 target gene²³. This may explain that in our experiments restoration of miR-124 using miR-124 mimic resulted in more significant effects on the metabolic profile than siPTBP1 (Fig. 5). It has also been shown that miR-124 selectively targets both human and mouse MCT1 (Monocarboxylate transporter 1, Lactate/pyruvate transporter) 3'-UTRs, thus potentially regulating the ability of the tumor cells to secrete lactate⁴². MCT-1 is part of a core Wnt signaling gene program for glycolysis in colon cancer, MCT-1 can export lactate, the byproduct of Warburg metabolism, and it is the essential transporter of pyruvate as well as a glycolysis-targeting cancer drug, 3-bromopyruvate (3-BP)⁴³. Consistent with this, we have observed that PH-Fibs activate naive macrophage via secreting IL-6³³ and lactate (unpublished). Results shown here indicated that inhibiting PKM2 in fibroblasts and/or macrophages can abrogate this pro-inflammatory fibroblast-macrophage interaction. Thus, considering the prominent role of miR-124 in lactate production in fibroblasts in present study, miR-124 may underlie the cell-cell interaction between fibroblasts and macrophages, thus merit further investigation to seek glycolysis-targeting drugs for PH.

We investigated the possibility that HDAC inhibitors (HDACi) could reverse the metabolic abnormalities observed in PH-Fibs based on previous observations, including ours, that: Class I HDAC (1,2,3) activity is increased in PH-Fibs³⁸, HDACi decrease proliferation in PH-Fibs³⁸, increase miR-124 expression²⁴ and prevent and reverse hypoxic-induced PH^{37, 38, 44}. The present work aimed at determining the mechanisms involved in these responses shows that HDACi (SAHA and apicidin) can normalize the PKM2/PKM1 ratio, shift the metabolic state toward oxidative phosphorylation, and rescue mitochondria abnormalities in PH-Fibs. Since we previously established that the catalytic activity of Class I HDACs are increased in PH-Fibs, and HDACi can restore miR-124 expression and decrease PTBP1 expression²⁹, it is possible that the mechanism of HDACi on reversing the metabolic abnormalities of PH-Fibs is at least partly through the miR-124/PTBP1/PKM axis. Our results are consistent with those in several cancers where it has been shown that genetic knockdown of individual HDACs, most notably HDAC1, 2, 3, and 6, in a variety of tumor types induced apoptosis and cell cycle arrest, implicating HDAC activity as a key mediator of survival⁴⁵. Indeed HDAC inhibitors can induce tumor cell apoptosis, growth arrest, senescence, differentiation, and immunogenicity, and inhibit angiogenesis at least in certain cancer cell types. Collectively, these findings suggest that it would be nearly impossible to conceive of a single mechanism of action that would be responsible for these effects of HDAC inhibitors in different cell types. In cancer, it is accepted that therapeutic outcomes will likely depend on the genetic lesions driving the tumor of interest and the HDAC inhibitors under investigation. Our studies demonstrate that PH-Fibs are a cell in which HDAC inhibition can significantly abrogate the miR-124-PTBP1-PKM2 axis, there by having or resulting in significant effects on metabolism and cell function.

Conversely, it has also been shown that metabolic intermediates may control HDAC activity in cancer cells⁴⁶. This may represent a cross-talk connecting cell metabolism, transcription, and other HDACs-controlled processes in physiological and pathological conditions. This provides direction for future research as it will be important to evaluate the potential existence of a “positive feedback” for PH in which metabolism and HDACs cross-talk to drive vascular remodeling.

The treatment of PH has historically been restricted by a limited number of therapeutic options. Endothelial dysfunction and abnormal proliferation and contraction of smooth-muscle cells represent logical pharmacologic targets^{47, 48}. Despite recent major improvements in symptomatic treatments, no current treatment cures this devastating condition⁴⁹. Recent advances in our understanding of the pathophysiological and molecular mechanisms, the new paradigm of an “outside-in” hypothesis, in which the adventitial compartment is viewed as a critical regulator of vessel wall function,⁵⁰ and the emergence of adventitial fibroblasts with constitutive pro-inflammatory, glycolytic and mitochondrial reprogramming phenotypes may orchestrate recruitment, retention, and activation of circulating inflammatory cells^{8, 9, 29}. Perhaps inside-out and outside-in hypothesis are not mutually exclusive as the work by Caruso et al., in this journal show very similar signaling pathways exist in resident endothelial cells and endothelial progenitors. These authors also show a link between the most common genetic causes of PAH, mutations in the bone morphogenetic protein type II receptor, and the miR-124/PTBP1/PKM axis. Thus common metabolic abnormalities may underlie pulmonary arterial hypertension and may lead to the development of new pharmacologic therapies, providing renewed hope for both patients and their physicians. Studies presented here using genetic, epigenetic and small molecule approaches to reverse metabolic reprogramming and proliferative status in adventitia fibroblasts may offer new therapeutic opportunities in this disease.

CLINICAL PERSPECTIVE.

What is New?

• In pulmonary hypertension (PH), little is known regarding the mechanisms that contribute to the generation and maintenance of the metabolic abnormalities in pulmonary arterial cells, especially in fibroblasts.

• Herein we establish a clear link between MiR-124 and its downstream target polypyrimidine tract binding protein 1 (PTBP1) to alterations in pyruvate kinase muscle (PKM) alternative splicing and thus alterations in the overall metabolic, proliferative and inflammatory state of hypertensive adventitial fibroblasts.

• These findings support the novel hypothesis that pyruvate is at the center of the entire metabolic network, which confers metabolic flexibility of and between the cells in different micro-environmental conditions.

Clinical Implications

• We found in fibroblasts from the pulmonary hypertensive vessel that restoration of the PKM2/PKM1 ratio toward normal with miR-124 mimic and siPTBP1, or pharmacologic inhibition of PKM2 glycolytic function with TEPP-46 and Shikonin, or treatment with HDACi reversed glycolytic and mitochondrial reprograming toward normal, and decreased proliferative capacities.

• Inhibition of PKM2 glycolytic function in fibroblasts also abrogated their pro-inflammatory effects on macrophages.

• These findings provide additional mechanistic underpinnings for potentially targeting the miR-124-PTBP1-PKM axis as a novel therapeutic strategy in PH.