# paper.pdf ## Metadata - PDFFormatVersion=1.7 - Language=de-DE - IsLinearized=true - IsAcroFormPresent=false - IsXFAPresent=false - IsCollectionPresent=false - IsSignaturesPresent=false - Author=Yongneng Zhang - CreationDate=D:20260504184855+05'30' - Creator=Adobe InDesign 15.1 (Windows) - ModDate=D:20260505122631+05'30' - Producer=Adobe PDF Library 15.0 - Subject=DOI: 10.1161/CIRCULATIONAHA.125.078674 Circulation, Volume. 153, Issue. 19, BACKGROUND: Right ventricular failure drives both morbidity and mortality in pulmonary arterial hypertension (PAH), but the mechanism of transition from compensated right ventricle (cRV) to decompensated RV (dRV) is unknown, and some PAH patients develop dRV faster than others. METHODS: We explored the role of RV cardiac myofibroblasts (cMFBs) on this transition utilizing 2 rat models: monocrotaline (MCT) versus the less inflammatory pulmonary artery banding (PAB); and 3 patient cohorts (n=81) of PAH versus group 2 pulmonary hypertension (PHT-2). We hypothesized that loss of UCP2 (uncoupling protein 2), critical for mitochondrial calcium (mCa++) regulation and cardiac fibroblasts (cFBs) differentiation to cMFBs, is associated with dRV; and that a loss-of-function UCP2 single nucleotide polymorphism (SNP; rs659366) may predict dRV in human PAH. RESULTS: We separated rat cRV from dRV based on catheterization and echocardiographic criteria and found a significant increase in cMFBs from MCT but not PAB RV (which decompensated much later than MCT RV). In isolated hearts, MCT RV contractility was lower in dRV but not in isolated cardiomyocytes (CMs), pointing to a non-CM cause. Mitochondrial respiration was lower in MCT dRV cMFBs than in control and cRV cFBs but increased in MCT–RV CMs. mCa++ was progressively decreased from normal to MCT–cRV to dRV cMFBs, and the same was true for c(M)FBs (but not CMs) UCP2. Tumor necrosis factor alpha (TNF-α) decreased UCP2 mRNA/protein levels in RV cFBs but not RV CMs. In human PAH, but not PHT-2, dRVs had more cMFBs and less UCP2 than control and cRVs. Decreased UCP2 levels and rs659366 were associated with worse RV performance (tricuspid annular plane systolic excursion and cardiac index), even among patients with similar mean pulmonary arterial pressure. CONCLUSIONS: Our data suggest that a change of cell identity (cFBs to cMFBs) in the RV is a driver of RV decompensation. TNF-α or rs659366 is a potential biomarker for early RV decompensation in PAH. - Title=Critical Contribution of Cardiac Myofibroblasts in Right Ventricular Failure and the Role of UCP2 SNPs in the Predisposition to RV Decompensation in Pulmonary Arterial Hypertension - Trapped=False - dc:format=application/pdf - dc:title=Critical Contribution of Cardiac Myofibroblasts in Right Ventricular Failure and the Role of UCP2 SNPs in the Predisposition to RV Decompensation in Pulmonary Arterial Hypertension - dc:creator=Yongneng Zhang,Sébastien Bonnet,Steeve Provencher,Jiyuan Piao,Alois Haromy,Yongsheng Liu,Yuan-Yuan Zhao,Sandra Breuils-Bonnet,Sarah-Eve Lemay,Dawn E. Bowles,Michelle Mendiola Pla,Gopinath Sutendra,Evangelos D. Michelakis - dc:subject=calcium ◼,heart failure ◼,hypertension pulmonary ◼,hypertrophy right ventricular ◼,myofibroblasts ◼,tumor necrosis factor-alpha ◼,uncoupling protein 2 - dc:description=DOI: 10.1161/CIRCULATIONAHA.125.078674 Circulation, Volume. 153, Issue. 19, BACKGROUND: Right ventricular failure drives both morbidity and mortality in pulmonary arterial hypertension (PAH), but the mechanism of transition from compensated right ventricle (cRV) to decompensated RV (dRV) is unknown, and some PAH patients develop dRV faster than others. METHODS: We explored the role of RV cardiac myofibroblasts (cMFBs) on this transition utilizing 2 rat models: monocrotaline (MCT) versus the less inflammatory pulmonary artery banding (PAB); and 3 patient cohorts (n=81) of PAH versus group 2 pulmonary hypertension (PHT-2). We hypothesized that loss of UCP2 (uncoupling protein 2), critical for mitochondrial calcium (mCa++) regulation and cardiac fibroblasts (cFBs) differentiation to cMFBs, is associated with dRV; and that a loss-of-function UCP2 single nucleotide polymorphism (SNP; rs659366) may predict dRV in human PAH. RESULTS: We separated rat cRV from dRV based on catheterization and echocardiographic criteria and found a significant increase in cMFBs from MCT but not PAB RV (which decompensated much later than MCT RV). In isolated hearts, MCT RV contractility was lower in dRV but not in isolated cardiomyocytes (CMs), pointing to a non-CM cause. Mitochondrial respiration was lower in MCT dRV cMFBs than in control and cRV cFBs but increased in MCT–RV CMs. mCa++ was progressively decreased from normal to MCT–cRV to dRV cMFBs, and the same was true for c(M)FBs (but not CMs) UCP2. Tumor necrosis factor alpha (TNF-α) decreased UCP2 mRNA/protein levels in RV cFBs but not RV CMs. In human PAH, but not PHT-2, dRVs had more cMFBs and less UCP2 than control and cRVs. Decreased UCP2 levels and rs659366 were associated with worse RV performance (tricuspid annular plane systolic excursion and cardiac index), even among patients with similar mean pulmonary arterial pressure. CONCLUSIONS: Our data suggest that a change of cell identity (cFBs to cMFBs) in the RV is a driver of RV decompensation. TNF-α or rs659366 is a potential biomarker for early RV decompensation in PAH. - dc:rights=© 2026 The Authors. - xmp:createdate=2026-05-04T18:48:55+05:30 - xmp:creatortool=Adobe InDesign 15.1 (Windows) - xmp:modifydate=2026-05-05T12:26:31+05:30 - xmp:metadatadate=2026-05-05T12:26:31+05:30 - xmpmm:instanceid=uuid:707bb721-a830-49bf-87f4-22f8b506f31e - xmpmm:originaldocumentid=xmp.did:DE4D607E8A24681180838231FB1F3433 - xmpmm:documentid=xmp.id:afdfb10c-83b6-4948-9ad6-394bd9b513c1 - xmpmm:renditionclass=proof:pdf - xmpmm:versionid=1 - xmpmm:derivedfrom=xmp.iid:78e60842-4746-a84b-a0f9-13feeaeeafdfxmp.did:00cb8259-66fe-fc45-941f-e2bad84730eedefault - xmpmm:history=convertedfrom application/x-indesign to application/pdfAdobe InDesign 15.1 (Windows)2026-05-04T18:48:55+05:30converteduuid:987fe904-242c-4e5a-bad9-9ee496511fefconverted to PDF/A-3aPreflight2026-05-05T12:26:30+05:30 - xmprights:marked=True - pdf:producer=Adobe PDF Library 15.0 - pdf:trapped=False - pdfuaid:part=1 - pdfaid:part=3 - pdfaid:conformance=A - pdfaextension:schemas=http://ns.adobe.com/pdf/1.3/pdfAdobe PDF SchemainternalA name object indicating whether the document has been modified to include trapping informationTrappedTexthttp://ns.adobe.com/xap/1.0/mm/xmpMMXMP Media Management SchemainternalUUID based identifier for specific incarnation of a documentInstanceIDURIinternalThe common identifier for all versions and renditions of a document.OriginalDocumentIDURIhttp://www.aiim.org/pdfa/ns/id/pdfaidPDF/A ID SchemainternalPart of PDF/A standardpartIntegerinternalAmendment of PDF/A standardamdTextinternalConformance level of PDF/A standardconformanceTexthttp://www.aiim.org/pdfua/ns/id/pdfuaidPDF/UA ID SchemainternalPart of PDF/UA standardpartInteger ## Contents ### Page 1 ORIGINAL RESEARCH ARTICLE # Critical Contribution of Cardiac Myofibroblasts inRight Ventricular Failure and the Role of UCP2SNPs in the Predisposition to RV Decompensationin Pulmonary Arterial Hypertension Yongneng Zhang[ ![Equation 1]() ](https://orcid.org/0000-0002-0791-9381) , MD, PhD; Sébastien Bonnet[ ![Equation 1]() ](https://orcid.org/0000-0001-5753-1900) , PhD; Steeve Provencher[ ![Equation 1]() ](https://orcid.org/0000-0001-7535-9972) , MD, PhD; Jiyuan Piao[ ![Equation 1]() ](https://orcid.org/0000-0002-4435-7285) , MD, PhD;Alois Haromy[ ![Equation 1]() ](https://orcid.org/0000-0003-4744-9879) , MS; Yongsheng Liu, MS; Yuan-Yuan Zhao, PhD; Sandra Breuils-Bonnet, MS; Sarah-Eve Lemay[ ![Equation 1]() ](https://orcid.org/0000-0002-6625-9918) , BS;Dawn E. Bowles, PhD; Michelle Mendiola Pla[ ![Equation 1]() ](https://orcid.org/0000-0001-9348-7708) , MD, MHS; Gopinath Sutendra[ ![Equation 1]() ](https://orcid.org/0000-0002-8181-1483) , PhD; Evangelos D. Michelakis[ ![Equation 1]() ](https://orcid.org/0000-0003-0787-7721) , MD Correspondence to: Evangelos D. Michelakis, MD, Department of Medicine, University of Alberta, 13-103 Clinical Sciences Building, 11304-83 Ave NW, Edmonton,Alberta, Canada T6G 2B7. Email Supplemental Material is available at. © 2026 The Authors. Circulation is published on behalf of the American Heart Association, Inc., by Wolters Kluwer Health, Inc. This is an open access article underthe terms of the License, which permits use, distribution, and reproduction in any medium, provided that theoriginal work is properly cited, the use is noncommercial, and no modifications or adaptations are made. BACKGROUND: Right ventricular failure drives both morbidity and mortality in pulmonary arterial hypertension (PAH), but themechanism of transition from compensated right ventricle (cRV) to decompensated RV (dRV) is unknown, and some PAHpatients develop dRV faster than others. METHODS: We explored the role of RV cardiac myofibroblasts (cMFBs) on this transition utilizing 2 rat models: monocrotaline(MCT) versus the less inflammatory pulmonary artery banding (PAB); and 3 patient cohorts (n=81) of PAH versus group2 pulmonary hypertension (PHT-2). We hypothesized that loss of UCP2 (uncoupling protein 2), critical for mitochondrialcalcium (mCa++) regulation and cardiac fibroblasts (cFBs) differentiation to cMFBs, is associated with dRV; and that a loss-of-function UCP2 single nucleotide polymorphism (SNP; rs659366) may predict dRV in human PAH. RESULTS: We separated rat cRV from dRV based on catheterization and echocardiographic criteria and found a significantincrease in cMFBs from MCT but not PAB RV (which decompensated much later than MCT RV). In isolated hearts, MCTRV contractility was lower in dRV but not in isolated cardiomyocytes (CMs), pointing to a non-CM cause. Mitochondrialrespiration was lower in MCT dRV cMFBs than in control and cRV cFBs but increased in MCT–RV CMs. mCa++ wasprogressively decreased from normal to MCT–cRV to dRV cMFBs, and the same was true for c(M)FBs (but not CMs) UCP2.Tumor necrosis factor alpha (TNF-α) decreased UCP2 mRNA/protein levels in RV cFBs but not RV CMs. In human PAH,but not PHT-2, dRVs had more cMFBs and less UCP2 than control and cRVs. Decreased UCP2 levels and rs659366 wereassociated with worse RV performance (tricuspid annular plane systolic excursion and cardiac index), even among patientswith similar mean pulmonary arterial pressure. CONCLUSIONS: Our data suggest that a change of cell identity (cFBs to cMFBs) in the RV is a driver of RV decompensation.TNF-α or rs659366 is a potential biomarker for early RV decompensation in PAH. Key Words: calcium ◼ heart failure ◼ hypertension, pulmonary ◼ hypertrophy, right ventricular ◼ myofibroblasts ◼ tumor necrosis factor-alpha◼ uncoupling protein 2 The main driver of mortality and morbidity in pul-monary arterial hypertension (PAH) is the functionof the right ventricle (RV) rather than the degreeof rise in mean pulmonary artery pressure (mPAP).1,2 InPAH, the compensating RV stage (cRV) is quite short,with decompensation occurring often in months,3 com-pared with left ventricular hypertrophy, in which the com-pensation can last decades.4 Between 2 patients with ### Page 2 the same age/sex and mPAP, one may have a shortercRV and decompensate faster than the other,5,6 compli-cating PAH clinical management and timing of referralfor transplantation. What makes the transition from cRVto decompensated RV (dRV) and why one RV is moreprone to failure than another is unknown. The presump-tion is that there is an energetic failure of cardiomyocytes(CMs) combined with inadequate angiogenesis in thehypertrophied RV (ie, relative ischemia), resulting in celldeath, collagen deposition, and structural remodeling.7,8However, this theory could not explain why RV functionoften, but not always, recovers after lung transplantationin patients with RV failure in PAH,9 as CMs cannot regen-erate. One possibility is that the RV failure is inducedby cell types other than CMs in the right ventricle, andthe pathology is potentially reversible. In the left ventricle(LV), cardiac myofibroblasts (cMFBs) increase quicklyafter myocardial infarction, mostly by differentiation fromcardiac fibroblasts (cFBs) in order to limit the myocardialwound via secretion of collagen and scar formation.10 Incontrast to cFBs, cMFBs express high levels of alphasmooth muscle actin (α-SMA) that allows them to con-tract and migrate to the wound site, but also increase thestiffness of the ventricle and compromise its function.11cMFBs can reach >30% to 40% of the cell populationin the LV and also increase in other cardiac pathologiesas well, such as hypertrophy or diabetes,12,13 but this hasnot been studied in the RV. There is evidence that differ-entiation and proliferation of cMFBs can both be haltedand even reversed,14 providing a rationale for therapeutictargeting of cMFBs. Although this has not been testedin the RV, it is very important because after dRV ensues,any improvement in mPAP through PAH therapies would ## Clinical Perspective ## What Is New? - • Although right ventricular failure drives morbidityand mortality in pulmonary arterial hypertension(PAH), the mechanism of transition from compen-sated right ventricle (RV) hypertrophy to RV decom-pensation and failure is unknown, and its onsetvaries among PAH patients. - • We found that this transition is driven by RV myo-fibroblast activation rather than cardiomyocyte ino-tropic dysfunction in rats and humans with PAH. - • This is attributable to RV myofibroblast mitochon-drial uncoupling protein 2 downregulation by thepresence of loss-of-function single nucleotidepolymorphisms (such as rs659366) or inflamma-tory cytokines such as tumor necrosis factor alpha(TNF-α). - • Germline rs659366 and TNF-α are predictive bio-markers of early RV decompensation in PAH (butnot group 2 pulmonary hypertension). ## What Are the Clinical Implications? - • Precision health in PAH may identify patients pre-disposed to RV failure based on presence ofrs659366 or high TNF-α levels, requiring moreintense PAH/supportive therapies, frequent follow-ups, earlier referral for transplantation, and antici-pating slower recovery after transplantation. - • Inotropic agents may not be helpful in decompen-sated RV patients. - • TNF-α inhibitors may have direct benefits on the RVof PAH patients over and above any direct benefitson the lung vasculature. - • Functional RV imaging like fluorodeoxyglucose-positron emission tomography needs to considerthat much of the signal comes from myofibroblastsand not just cardiomyocytes, cell types with oppos-ing but “symbiotic” mitochondrial metabolismaccording to our findings. ## Nonstandard Abbreviations and Acronyms α-SMA alpha smooth muscle actin ACLY ATP-citrate ligase ADMA asymmetric dimethylarginine cFB cardiac fibroblast CI cardiac index CM cardiomyocyte cMFB cardiac myofibroblast cRV compensated right ventricle ddPCR droplet digital PCR dRV decompensated right ventricle GO glucose oxidation HET heterozygote HOM homozygote mCa++ mitochondrial calcium MCT monocrotaline MCT1/4 monocarboxylate transporter¼ MCU mitochondrial calcium uniporter MICU1 mitochondrial calcium uptake 1 mPAP mean pulmonary artery pressure OCR oxygen consumption rate PAB pulmonary artery banding PAH pulmonary arterial hypertension PDH pyruvate dehydrogenase PHT-2 group 2 pulmonary hypertension PRMT1 protein arginine methyltransferase 1 RV right ventricle RVEDD right ventricular end-diastolic diameter SNP single nucleotide polymorphism TAPSE tricuspid annular plane systolicexcursion UCP2 uncoupling protein 2 WT wild-type ### Page 3 be obsolete if the RV cannot recover. This suggests aneed for studies to understand the mechanisms promot-ing dRV and cMFB regulation in the RV, which may bedifferent than the left ventricle, as the 2 ventricles havedifferent embryological origin (different heart fields),metabolism, and biochemistry in the adult heart.15,16 Recent studies have identified a central role of mito-chondrial calcium (mCa++) in the differentiation and acti-vation of myofibroblasts.17,18 UCP2 (uncoupling protein2) has been shown to regulate the mitochondrial cal-cium uniporter (MCU) function and mitochondrial calcium(mCa++) uptake when the mitochondrial calcium uptake 1(MICU1), a component of the MCU, is methylated.19,20 Areduction in UCP2, whether because of genetic knock-out in mice or germ line loss-of-function single nucleo-tide polymorphisms (SNPs) in humans, combined withincreased MICU1 methylation, can result in decreasedmCa++ levels. The resulting inhibition of mCa++-sensitiveenzymes, such as pyruvate dehydrogenase (PDH) acti-vated via Ca2+-dependent PDH phosphatases, can alterthe production of diffusible mitochondrial metabolites thatcan regulate epigenetic mechanisms in the nucleus, lead-ing to a change of cell identity in cFBs, toward cMFBs.Our group has previously shown that genetic loss ofUcp2 in mice promotes PAH and RV failure, and a UCP2loss-of-function SNP (rs659366) is prevalent in PAHpatients, occurring in ≈40% of individuals heterozygouslyand ≈20% homozygously, and is associated with worseoutcomes.21,22 Furthermore, inflammatory cytokines liketumor necrosis factor alpha (TNF-α) increase from normalto cRV to dRV in PAH rats and are increased in the serumof PAH patients.23,24 TNF-α inhibits UCP2 expression anddecreases mCa++,25,26 which could promote cMFB differ-entiation. Based on the inflammatory phenotype of PAHand the positive association between the presence of theUCP2 loss-of-function SNP (rs659366) and worse PAHoutcomes, we aimed to determine the role of UCP2 in thetransition of fibroblasts to myofibroblasts in the rat RV, aswell as the relationship between this UCP2 SNP and RVfunction in PAH patients. We hypothesized that reducedUCP2 levels would activate myofibroblasts, thereby accel-erating RV decompensation, and that the UCP2 SNPwould correlate with worse RV function in human PAHcohorts. We show that a transition from cFBs to cMFBs isassociated with RV decompensation in rodent and humanPAH, driven by genetic and TNF-α–induced UCP2 loss. ## METHODS Data supporting the findings of this study are available from thecorresponding author upon reasonable request. Detailed meth-ods are available in the. ### Human Tissue Samples Cohort 1: All experimental procedures using human tissueswere performed with the approval of Laval University and theInstitute Universitaire de Cardiologie et Pneumologie biosafetyand ethics committees (CER20773). Cohort 2: All work with human clinical data and blood sam-ples was performed with approval from the University of Albertahuman research ethics board (Pro00057828). Cohort 3: Tissue samples used for this study were obtainedfrom the Duke Human Heart Repository, a Duke UniversityHospital System institutional review board–approved tissuerepository (Pro00005621). All samples for all cohorts werecollected with written informed consent. ### Animal Studies Our work with rats and mice was performed with permissionfrom animal care and use committees from the Universityof Alberta (AUP00000298) and from Laval University(2022-1049). ### Statistical Analysis All statistical analyses performed on GraphPad Prism 10(GraphPad Software, La Jolla, CA). Values are expressed asmean±SEM. The use of parametric or nonparametric testswas decided after assessment of normal distribution of the val-ues by the Shapiro-Wilk normality test. For parametric tests,we used 2-tailed, unpaired Student t test to assess statisticalsignificance between 2 groups, and multiple groups were com-pared using one-way analysis of variance (ANOVA) followed bya Bonferroni post hoc analysis. For nonparametric tests, Mann-Whitney U test was used for comparisons between 2 groups,and Kruskal-Wallis test was used to compare multiple groups.In cases of multiple measurements in the same subject, a linearmixed-effects model was used to show the differences werefrom groups (fixed effect) rather than random effects frommeasurements within each animal or among different animalswithin the same group. Nonparametric correlation analysis ofthe clinical data was performed using the Spearman correlationcoefficient. Significance for all statistical testing was P<0.05. ## RESULTS ### cMFBs and the Transition From cRV to dRV inRats Contractility decreases in dRV in isolated hearts, but notin isolated RV CMs. We used an established rat model ofPAH using monocrotaline (MCT) injection to study thehemodynamic transition from normal RV to cRV and ul-timately to dRV. MCT is a plant (crotalaria genus) toxinthat injures pulmonary artery endothelial cells and elicitsa strong inflammatory response. Although this inflam-matory model does not replicate the lung histology ofhuman PAH (ie, it does not replicate the plexiform le-sions), it shares the inflammation (particularly in patientswith idiopathic PAH or scleroderma-associated PAH),the very high elevation of pulmonary pressures, and therapid transition from cRV to dRV observed in patients.We felt that the lack of plexiform lesions in the lungs wasnot a mechanistic concern as we were not interested inthe lung vasculature, but rather, the RV biology, which is ### Page 4 agnostic to lung histology. We studied rats after saline(control) versus MCT injection at 3 or 4–5 weeks. Wechose these time points because in this model, it is knownthat clinical signs of pulmonary hypertension appear af-ter 3 weeks and peak between the 4th and 5th weekafter MCT, followed by 100% mortality soon thereafter.We used right heart catheterization (internal jugular can-ulation and advance of a Millar catheter tip into the mainpulmonary artery), as well as echocardiography, to recordmean PA pressure (mPAP; ie, the severity of PAH), RVsystolic pressure (RVSP), right atrial (RA) pressure andcardiac output (CO; ie, the stage of heart failure), RV freewall thickness, right ventricular end-diastolic diameter(RVEDD), and tricuspid annular plane systolic excursion(TAPSE; ie, the structural and functional RV remodel-ing).27 We defined cRV as a state of increased mPAP andRVSP but preserved CO and mildly increased RA pres-sure; along with increased RV free wall thickness butnot increased RVEDD. In contrast, we defined the dRVas the stage characterized by high mPAP but a drop inRVSP (attributable to RV failure), rate of contraction (dp/dt max), rate of relaxation (ie, dp/dt min), CO, and TAPSE,as well as an increase in RA pressure and RVEDD (seeMethods for criteria; Figure 1A through 1C). All the RVsand tissues labeled as cRV or dRV thereafter, came fromrats that strictly fulfilled these hemodynamic criteria. Thisoffered an accurate distinction between the 3 RV func-tional stages. All the rats studied in 3 weeks had cRV,and all rats studied in 4–5 weeks had dRV. Evaluating RV contractility in vivo is complicated bymany factors, for example, afterload (eg, mPAP) vari-ability, anesthesia level, or hormone effects. Therefore,we also measured the RV contractility ex vivo using theLangendorff perfusion system in which a balloon wasinserted into the RV to measure contractility under astandardized afterload (gravity). Although we measuredcontractility without a pacemaker for the first 2 minutes,we also used a pacemaker beyond that point, as arrhyth-mias commonly occur compromising the recordings, andalso to standardize measurements under the same heartrate (which tends to vary among different heart prepara-tions). As expected, RV contractility decreased from cRVto dRV when studied under the same afterload and heartrate (Figure 1D). However, when we isolated RV CMs tostudy contractility in vitro (measured by sarcomere short-ening), we found no significant differences among the 3RV functional stage groups (Figure 1E). This suggeststhat the differences of contractility ex vivo may be attrib-utable to factors external to CMs. dRVs have increased cMFBs, which exhibit a Warburgeffect with decreased glucose oxidation and increasedaerobic glycolysis, in contrast to RV CMs. To investigatea potential role of cMFBs, we first measured their num-ber based on the fact that they express both vimentin(like cFBs) but also α-SMA. We found that the number ofcMFBs is not different between normal RV and cRV, butit is dramatically increased in dRV (Figure 2A). We thenisolated and cultured RV c(M)FBs in short culture fromthe three groups and found that α-SMA is not present innormal RV and cRV cFBs in contrast to dRV ().Furthermore, we also found that cMFBs in dRV, in addi-tion to an obvious change in cell shape compared withcFBs, express high Ki67 levels compared with cFBs, inkeeping with their high proliferative nature (Figure 2B). We then studied whether there is a difference in mito-chondrial function among the 3 RV stages using theSeahorse analyzer. We found that the dRV cMFBs haddecreased basal and maximal oxygen consumption rates(OCR) compared with normal RV and cRV cFBs (Fig-ure 2C and 2D). However, an opposite pattern was foundin dRV CMs, which had increased basal and maximalOCR (Figure 2E). Compared with normal and cRV cFBs,dRV cMFBs also had decreased TCA cycle diffusiblemetabolites, including citrate and fumarate, measuredby mass spectrometry (Figure 2F), which is in keepingwith decreased glucose oxidation (GO). We then mea-sured lactate, the terminal product of glycolysis, by usingC-13–labeled glucose and found that dRV cMFBs haveincreased levels (both intracellularly and extracellularly)of C-13–labeled lactate compared with normal and cRVcFBs (Figure 2G). This “Warburg effect” in cMFBs, char-acterized by decreased glucose oxidation and increasedaerobic glycolysis is similar to what has been reportedin pulmonary arterial smooth muscle cells in PAH and incancer cells.28 #### A Metabolic Cross-Talk Between cMFBs and CMs Intrigued by the disassociation between the mitochon-drial function of the c(M)FB versus the CM, we specu-lated that the two cells may communicate in a paracrinemanner, in an attempt to harmonize energy homeosta-sis, similarly to what has been described between neu-rons and glial cells.29,30 It is possible that the RV CMscan use the glycolysis byproduct lactate produced by theneighboring RV cMFBs as a fuel to further support ATPsynthesis, which could explain the “paradoxical” increasein OCR that we found in dRV CMs. Therefore, we mea-sured the levels of 2 enzymes: lactate dehydrogenaseA (LDHA), which preferentially converts pyruvate to lac-tate, and lactate dehydrogenase B (LDHB), which pref-erentially converts lactate back to pyruvate; as well as2 transporters of lactate, monocarboxylate transporter 4(MCT4) for export, and monocarboxylate transporter 1(MCT1) for import of lactate. We found the LDHA pro-gressively increased from normal RV to cRV to dRV c(M)FBs, and the ratio of LDHA/LDHB sharply increasedin dRV cMFBs. Moreover, dRV cMFBs had decreasedMCT1 but increased MCT4 compared with normal RVand cRV cFBs (Figure 2H). In contrast, dRV CMs hadincreased MCT1 compared with normal CMs (Figure 2I).To prove that lactate can be used as a fuel to generateATP in CMs, we treated CMs from normal RV and dRV ### Page 5 ![Figure 1]() ### Page 6 Figure 1 Continued. B, Representative tracings of RV systolic pressure (RVSP) and pulmonary artery pressures (PAPs) in CTRL, cRV, and dRVof MCT rats (left) performed by close-chest right heart catheterization. Parameters measured (mPAP, RVSP, and RAP) are shown on the rightand expressed as mean±SEM; n=6 animals for each group. *P<0.05; **P<0.01. C, Representative tracings of dp/dt of RV pressures in CTRL,cRV, and dRV of MCT rats (left), with values (right) expressed as mean±SEM; n=6 animals for each group. *P<0.05; **P<0.01. D, Representativetracings (left) of RV pressures in ex vivo isolated hearts from CTRL, cRV, and dRV of MCT rats, with values (middle) expressed as mean±SEM;n=7 animals for each group. The picture (right) shows one isolated rat heart setup. A balloon was placed into RV (pointed by the yellow arrow),and pacemaker wires (pointed by the red arrow) were inserted into the myocardium. ***P<0.001; ****P<0.0001. E, Representative tracingsof sarcomere length (left) of RV CM contractility (sarcomere shortening) from normal RV, cRV, and dRV of MCT rats with values (middle)expressed as mean±SEM; n= ≈50 cells from 7 animals for each group. The picture (right) shows where the sarcomere shortening was measured(yellow rectangle). Comparisons were made using one-way ANOVA followed by Bonferroni post hoc analysis. CM indicates cardiomyocyte; cRV,compensated right ventricle; CTRL, control (normal right ventricle, without monocrotaline injection); dRV, decompensated right ventricle; mPAP,mean PA pressure; PAP, pulmonary artery pressure; RAP, right atrium pressure; RV, right ventricle; RVEDD, right ventricular end-diastolic diameter;RVP, right ventricle pressure; RVSP, right ventricular systolic pressure; and TAPSE, tricuspid annular plane systolic excursion. with C-13–labeled lactate and then measured C-13–la-beled TCA cycle intermediates. We found that the dRVCMs had increased C-13–labeled metabolites comparedwith normal CMs (). The proposed mechanismof the metabolic cross-talk between cMFBs and CMs isshown in Figure 2J. #### Decreased Mitochondrial Calcium and IncreasedMICU1 Methylation Is Found in dRV cMFBsCompared With Normal or cRV cFBs The OCR is a reflection of the overall mitochondrial func-tion, particularly the production of nicotinamide adeninedinucleotide (reduced form) (NADH) and flavin adeninedinucleotide (reduced form) FADH2 from the TCA cycle,which are used in electron transport chain along withoxygen to generate ATP. The production of NADH andFADH2 depends on the activity of many enzymes thatare mCa++-dependent (eg, PDH). Thus, we measuredmCa++ in c(M)FBs and found decreased mCa++ fromnormal to cRV and a further decrease to dRV cMFBs(Figure 3A). We then measured PDH activity and foundthat dRV cMFBs had decreased PDH activity comparedwith those from normal and cRV (Figure 3B). To determine the basis of this difference, we first mea-sure the levels of the mCa++ uniporter MCU and its regu-latory subunit MICU1. It has been shown that MICU1negatively regulate MCU function through its methylation,with protein arginine methyltransferase 1 (PRMT1) servingas the major protein arginine methyltransferase catalyz-ing asymmetric dimethylarginine (ADMA) on MICU1. Wetherefore measured the PRMT1 expression and the meth-ylation status of MICU1 using methylated (ADMA)/totalMICU1 ratio, assessed by both immunoblotting and immu-noprecipitation. First, with RV tissue and c(M)FB staining,we found that the cytoplasmic PRMT1 and ADMA levelsare increased in dRV cMFBs compared with normal andcRV cFBs (Figure 3C through 3F). Similar results werefound in immunoblots, in which the PRMT1 and ADMA lev-els are increased in dRV cMFBs compared with normal RVcFBs (Figure 3G). However, the PRMT1 and ADMA signal,although they had a significant trend, they did not reachsignificance. Therefore, we proceeded with immunopre-cipitation of MICU1 from isolated mitochondrial, in orderto precisely measure its methylation (ADMA signal). Wefound that MICU1 methylation is increased in dRV cMFBscompared with normal RV and cRV cFBs (Figure 3H). Inaddition, because there is no difference in MCU protein lev-els among the 3 groups, we concluded that the decreasedmCa++ was not attributable to MCU levels change, butrather, to an increase in the MICU1 methylation (). In contrast, we found no difference in PRMT1 andADMA in CMs among the 3 RV stages (), whichmeans this change is fibroblast-specific. #### UCP2 levels Are Decreased From Normal to cRV todRV c(M)FBs UCP2 is essential for mCa++ uptake when MICU1 ismethylated, and its downregulation or loss of function canreduce mCa++.20 Accordingly, we measured UCP2 levelsin whole RV tissue, as well as in RV CMs and RV cMFBsseparately. UCP2 levels were exclusively decreased indRV tissue compared with cRV and normal RV, but thisdecrease was not found in the corresponding LVs, sug-gesting that is RV-specific (Figure 4A). The UCP2 levelwas decreased in dRV c(M)FB compared with normal RVand cRV (Figure 4B), but not in CMs (), meaning that fibroblasts are the main source ofthe reduced UCP2 levels observed in the RV. To determine the role of UCP2 in the regulation ofmCa++ independent of M1CU1 methylation, we mea-sured mCa++ level in cFBs from wild-type (WT), Ucp2 het-erozygote, and homozygote KO mice (KO was confirmedin both RNA and protein levels in cMFBs in). We found a gene dose-dependentdecrease in mCa++ as UCP2 levels decreased (Fig-ure 4C). These data together suggest that the reasonfor the progressive loss of mCa++ from normal to cRVto dRV may be an increase of MICU1 methylation com-bined with a loss of UCP2 in cMFBs. This can explain theincreased number of cMFBs from cRV to dRV. #### TNF-α Increases α-SMA in cFBs and SuppressesUCP2 Expression Patients with group-1 PAH have a highly inflammatoryphenotype compared with other types of pulmonary hy-pertension. We have previously shown that during thetransition from cRV to dRV in MCT rats, serum levels ofthe cytokine TNF-α are increased.23 Because TNF-α candecrease the expression of UCP2,31 we measured UCP2 ### Page 7 ![Figure 2]() ### Page 8 Figure 2 Continued. We studied 3 rats for each group (shown as 1, 2, and 3 on x axis), and for each rat, we measured 3 fields. The percentpositive ratio for each field is shown in circles along with the median of all rats (right). n=9 fields (≈200 (myo)fibroblasts from 3 rats for eachgroup). We initially used the The Mann-Whitney test to show a significant difference between the 2 groups (P=0.0002). To address potentialrandom effects from variability within different fields from each rat, or among different rats within each group, we also performed a linear mixed-effects model, which showed that the only difference was between groups (fixed effects): (F(1, 12)=43.48; P<0.0001) and not among differentanimals within the same group (F(2, 12)=1.359; P=0.2937). ****P< 0.0001 when comparing these 2 groups. C, A representative Seahorseexperiment in (myo)fibroblasts from normal RV, cRV, and dRV of MCT rats. Oligomycin inhibits ATP synthase and thus oxidative phosphorylation,whereas the ionophore FCCP uncouples the electron transport chain, exposing maximal respiration. The ETC complex I and III inhibitors rotenoneand antimycin are used to uncover the mitochondrial component of respiration. The basal mitochondrial oxygen consumption rate (OCR) is therate before any drugs, minus the residual rate after rotenone/antimycin. D, Seahorse values of mitochondrial basal and maximal OCR in cardiac(myo)fibroblasts from normal RV, cRV, and dRV of MCT rats expressed as mean±SEM; n=6–8 measurements from 3 animals for each group.**P<0.01; ***P<0.001. E, Seahorse values of mitochondrial basal and maximal OCR in CMs from normal RV, cRV, and dRV of MCT rats expressedas mean±SEM; n=5 measurements from 3 animals for each group. **P<0.01. F, Mass spectrometry shows the intracellular citrate (left) andfumarate (right) levels in (myo)fibroblasts from normal RV, cRV, and dRV of MCT rats with values expressed as mean±SEM; n=4 animals for eachgroup. *P<0.05. G, Mass spectrometry shows the intracellular (left) and supernatant (right) lactate levels in (myo)fibroblasts from normal RV,cRV, and dRV of MCT rats after treatment with C-13 glucose (17.5 mM) for 12 hours with values expressed as mean±SEM; n=4 animals for eachgroup. *P<0.05. H, Immunoblot (left) shows the level of LDHA, LDHB, MCT1, MCT4, and β-actin in (myo)fibroblasts from normal RV, cRV, anddRV of MCT rats. Quantification values (right) are expressed as mean±SEM; n=4–5 animals for each group. *P<0.05; **P<0.01. I, Immunoblot(left) shows the levels of LDHA, LDHB, MCT1, MCT4, and β-actin in CMs from normal RV, cRV, and dRV of MCT rats. Quantification values(right) are expressed as mean±SEM; n=4 animals for each group. *P<0.05. J, A schematic of the metabolic cross-talk between myofibroblastsand CMs: cell-type–specific differences in the levels of pyruvate/lactate metabolizing enzymes and lactate importers/exporters allow dRV CMs touptake the lactate produced by cMFB (as a result of their decreased mitochondrial respiration) and feed their TCA cycle to produce ATP, perhapsexplaining their increased respiration compared with normal CMs. Comparisons were made using Kruskal–Wallis tests followed by pairwiseMann-Whitney U tests. 3-C13 indicates 3 carbon-13 labeled lactate; α-SMA, alpha smooth muscle actin; CM, cardiomyocyte; c(M)FB, cardiac(myo)fibroblast; cRV, compensated right ventricle; CTRL, control (normal right ventricle, without monocrotaline injection); DAPI, 4′,6-diamidino-2-phenylindole (labels the nucleus); dRV, decompensated right ventricle; FCCP, carbonyl cyanide-p-trifluoromethoxyphenylhydrazone; Highmag, high magnification; LDHA, lactate dehydrogenase A; LDHB, lactate dehydrogenase B; Low mag, low magnification; MCT (in immunoblots),monocarboxylate transporter; and OCR, oxygen consumption rate. levels in cFBs treated with different doses of TNF-α. Wefound that UCP2 mRNA and protein levels decreasedin a dose-dependent manner (Figure 4D and 4E). Notsurprisingly, the mCa++ level also decreased after TNF-αtreatment (Figure 4F). To determine whether TNF-α in-creases α-SMA over and above the classic inducer ofcMFBs (ie, transforming growth factor-beta [TGF-β]), westudied the effect of TGF-β alone and TNF-α in combi-nation with TGF-β on UCP2, mCa++ and α-SMA levelsin RV cFBs. TGF-β alone does not affect UCP2 levelin cFBs, whereas TNF-α in combination with TGF-β de-creases UCP2 mRNA and protein levels compared withTGF-β alone (). RV cFBs treat-ed with TGF-β alone have increased α-SMA and slightlydecreased mCa++; whereas cFBs treated with bothTGF-β and TNF-α show further increases in α-SMA andadditional reduction in mCa++ (Figure 4G and 4H). Wespeculated that TGF-β activates RV cMFBs via a UCP2-independent mechanism, whereas TNF-α promotescMFB activation by suppressing UCP2. To test this, weperformed the combination experiment in WT or Ucp2KO cFBs. We found that TGF-β increases α-SMA in bothWT and Ucp2 KO cFBs, whereas the additional stimu-latory effect of TNF-α was observed only in WT cFBs(). These results indicate that TGF-β activatesRV cMFBs independently of UCP2, whereas TNF-α in-duces the transition from cFBs to cMFBs through UCP2.Because the TGF-β axis is disturbed in PAH because ofgermline mutations or other factors,32–37 our data sug-gest that, in addition to the pulmonary vascular cells, theRV fibroblasts may also be affected, contributing to dRV.Inflammatory phenotypes, particularly in patients withiPAH or scleroderma-associated PAH, or in the presenceof a UCP2 loss-of-function SNP would intensify this ef-fect. In contrast, TGF-β and TNF-α have no effect onUCP2 levels in RV-CM, indicating that the suppressiveeffect of TNF-α on UCP2 expression is specific to fibro-blasts (). #### Number of Myofibroblasts, MICU1 Methylation andUCP2 Level Show No Difference Between cRV anddRV in Pulmonary Artery Banding (PAB) Rats To further support the hypothesis that inflammation andUCP2 are critical for the accelerated progression of RVfailure, we compared the pulmonary artery banding (PAB)rat model with the MCT rat model. The main differencebetween these models (which otherwise have similar in-creases in mPAP) is inflammation, particularly involvingcytokines such as TNF-α. Our group and others havepreviously shown that during the transition from cRV todRV in MCT rats, plasma TNF-α levels are increased.23,38Although TNF-α mRNA expression is elevated in the RVsof PAB mice,39 TNF-α protein levels are not increased ineither plasma or RV myocardium in animals with PAB.40It has also been reported that in mice with transverseconstriction of the aorta, another mechanical obstruc-tion model, TNF-α protein levels peak at 6 hours and re-turned to baseline levels within 72 hours.41 This suggeststhat mechanical obstruction models, including PAB, mayexhibit only acute and short-lasting inflammation relatedto the surgery procedure. To directly compare the survivalbetween the MCT and PAB model, we also used MCT ### Page 9 ![Figure 3]() ### Page 10 Figure 3 Continued. We initially used the Kruskal-Wallis test to show significant differences among groups (P<0.0001). To address potentialrandom effects from variability from cells in each rat, or among different rats in each group, we also performed a linear mixed-effects model, whichshowed that the only difference was between groups (fixed effects): (F(1.413, 93.27)=35.95; P<0.0001) and not among different animals withinthe same group (F(2, 132)=0.3773; P=0.6865). Post hoc comparisons between groups were performed using Tukey’s multiple-comparisons test.*P<0.05; **** P<0.0001. B, PDH activity was measured using an nicotinamide adenine dinucleotide (reduced form) (NADH)-based enzymaticactivity dipstick assay kit, in (myo)fibroblasts from normal RV (CTRL), cRV, and dRV of MCT rats. Quantification values (right) are expressed asmean±SEM; n=4–5 animals for each group. *P<0.05; **P< 0.01 when comparing the 2 groups. The linear mixed-effects model was used, whichshowed that the only significant difference (fixed effects) was between comparing the 2 groups, whereas the random effects were not significant(P=0.69). C, Confocal microscope (left) shows the PRMT1 (ie, arginine methyltransferase) levels in (myo)fibroblasts (vimentin-positive cells) fromnormal RV, cRV, and dRV of MCT rats. We studied 3 rats for each group (shown as 1, 2, and 3 in I axis), and for each rat, we measured 15–20cells. The level of PRMT1 in each cell is shown in circles along with mean±SEM (right). n=50 cells from 3 animals for each group. We initiallyused the Kruskal-Wallis test to show significant differences among groups (P<0.0001). To address potential random effects from variabilitydifferent cells from each rat, or among different rats in each group, we also performed a linear mixed-effects model, which showed that the onlydifference was between groups (fixed effects): (F(1.599, 95.15)=86.38; P<0.0001) and not among different animals within the same group(F(2,119)=0.4995; P=0.6081). Post hoc comparisons between groups were performed using Tukey’s multiple-comparisons test. ***P<0.001;****P<0.0001. D, Confocal microscopy (left) shows the ADMA (ie, arginine dimethylation) levels in (myo)fibroblasts (vimentin-positive cells) fromnormal RV, cRV, and dRV of MCT rats. We studied 3 rats for each group (shown as 1, 2, and 3 on x axis), and for each rat, we measured 15–20cells. The level of ADMA in each cell is shown in circles along with mean±SEM (right). n=50 cells from 3 animals for each group. We initiallyused the Kruskal-Wallis analysis to show significant differences among groups (P<0.0001). To address potential random effects from variabilityamong different cells from each rat, or among different rats in each group, we also performed a linear mixed-effects model, which showed thatthe only difference was between groups (fixed effects): (F(1.707, 134.9)=15.06; P<0.0001) and not among different animals within the samegroup (F(2,158)=0.1966; P=0.8218). Post hoc comparisons between groups were performed using Tukey’s multiple-comparisons test. *P<0.05;****P<0.0001. E, Confocal microscopy (left) shows the cytoplasmic PRMT1 levels in cultured (myo)fibroblasts from normal RV, cRV, and dRV ofMCT rats. We studied 3 rats for each group (shown as 1, 2, and 3 on x axis), and for each rat, we measured 15–20 cells. The cytoplasmic level ofPRMT1 in each cell is shown in circles along with mean±SEM (right). n=50 cells from 3 animals for each group. We initially used the Kruskal-Wallis test to show significant differences among groups (P=0.0003). To address potential random effects from variability among different cellsin each rat, or different rats in each group, we also performed a linear mixed-effects model, which showed that the only difference was betweengroups (fixed effects): (F(1.59, 64.4)=12.08; P=0.0001) and not among different animals within the same group (F(2, 50)=0.101; P=0.9041).Post hoc comparisons between groups were performed using Tukey’s multiple comparisons test. *P<0.05; ****P<0.0001. F, Confocal microscopy(left) shows the cytoplasmic ADMA levels in cultured (myo)fibroblasts from normal RV, cRV, and dRV of MCT rats. We studied 3 rats for eachgroup (shown as 1, 2, and 3 on x axis), and for each rat, we measured 15–20 cells. The cytoplasmic level of ADMA in each cell is shown incircles along with mean±SEM (right). n=50 cells from 3 animals for each group. We initially used the Kruskal-Wallis test to show significantdifferences among groups (P=0.0012). To address potential random effects from variability among different cells in each rat and different rats ineach group, we also performed a linear mixed-effects model, which showed that the only difference was between groups (fixed effects): (F(1.953,125)=9.617; P=0.0001) and not among different animals within the same group (F(2, 128)=1.782; P=0.1725). Post hoc comparisons betweengroups were performed using Tukey’s multiple-comparisons test. **P<0.01; ***P<0.001. G, Immunoblot (left) shows the levels of PRMT1, ADMA,MICU1, VDAC1/2, and β-actin in (myo)fibroblasts from normal RV, cRV, and dRV of MCT rats. Quantification values (right) are expressed asmean±SEM; n=4–5 animals for each group. H, MICU1 immunoprecipitation (top) on isolated mitochondria from (myo)fibroblasts from normalRV, cRV, and dRV of MCT rats shows the levels of MICU1 and ADMA. MICU1 methylation is shown as ADMA/MICU1. Quantification values(bottom) are expressed as mean±SEM from 4 experiments. *P<0.05. B, G, and H were made using Kruskal–Wallis tests followed by pairwiseMann-Whitney U tests. ADMA indicates asymmetric dimethylarginine; cFB, cardiac fibroblast; cMFB, cardiac myofibroblast; cRV, compensatedright ventricle; CTRL, control (normal right ventricle, with vehicle instead of MCT injection); DAPI, 4′,6-diamidino-2-phenylindole (labels thenucleus); dRV, decompensated right ventricle c(M)FB, cardiac (myo)fibroblasts; IP, immunoprecipitation; ; MICU1, mitochondrial calcium uptake 1;PDH, pyruvate dehydrogenase; PRMT1, protein arginine methyltransferase 1; and VDAC, voltage-dependent anion channel. rats that we did not study at 3 or 4–5 weeks (n=56). Wefound that MCT rats had 100% mortality shortly after the5th week, whereas the PAB rats (n=76) had <10% mor-tality by 4–5 weeks, suggesting a much-delayed tran-sition between cRV and dRV. Their mortality increasedsignificantly only after 7–8 weeks (). Thus, westudied PAB rats at that point, when the transition fromcRV to dRV is happening, and found that PAB rats, de-spite having somewhat higher mPAP in cRV, had smallerdecreases in CO and smaller increases in RVEDD (ie,less RV decompensation), even though they had higherRV afterload for a longer period compared with MCT rats(7–8 versus 4–5 weeks), explaining their higher surviv-al (). We also found that the number of RVcMFBs increased only mildly from control to cRV, withno further increase in dRV (). In contrast toMCT rats, in which UCP2 levels decreased progres-sively from control to cRV to dRV (Figure 4A and 4B),UCP2 levels in PAB rats (as well as PRMT1 and MICU1methylation) were increased in both cRV and dRV com-pared with control, without a difference between cRVand dRV (). Thus, we conclude that it is theinflammation-driven loss of UCP2 that drives the accel-erated cRV to dRV transition. #### Translation in Human: RV cMFBs and UCP2 Levelsin Patients With PAH and Group 2 PHT We studied 2 cohorts of patients with PAH (cohort 1from Laval University and cohort 2 from the Universityof Alberta) and one cohort of patients with group 2 pul-monary hypertension (PHT-2; cohort 3) from Duke Uni-versity. Cohort 1 included RV tissues (from heart/lungtransplant recipients or autopsies) from 30 patients with-out (sudden cardiac death with no known heart diseaseand macroscopically normal RV controls) or with group1 PAH (), in which the PAH diagnosis had been ### Page 11 ![Figure 4]() ### Page 12 Figure 4 Continued. Quantification values (right) normalized by β-actin are expressed as mean±SEM; n=5 animals for each group. *P<0.05;**P<0.01. C, Confocal microscopy (left) shows the mitochondrial calcium levels (Rhod-2 fluorescence) in cultured right ventricular (myo)fibroblasts from WT, Ucp2 heterozygous, and Ucp2 knockout mice. We studied 3 mice from each group (shown as 1, 2, and 3 on x axis), andfor each mouse, we measured 15–20 cells. The Rhod-2 fluorescence level of each cell is shown in circles along with mean±SEM (right).n=50 cells from 3 animals for each group. We initially used the Kruskal-Wallis test to show significant differences among groups (P<0.0001).To address potential random effects from variability among different cells in each mouse, or different mice in each group, we also performeda linear mixed-effects model, which showed that the only difference was between groups (fixed effects): (F(1.53, 69.63)=71.93; P<0.0001)and not among different animals within the same group (F(2, 49)=2.318; P=0.1092). Post hoc comparisons between groups were performedusing Tukey’s multiple-comparisons test. ***P<0.001; ****P<0.0001. D, Immunoblot shows the UCP2 protein levels in normal rat right ventricularfibroblasts treated with or without TNF-α at doses of 10, 20, or 40 ng/mL. E, q-PCR shows the Ucp2 mRNA levels in normal rat right ventricularfibroblasts treated with or without TNF-α at 10, 20, or 40 ng/mL. Quantification values are expressed as mean±SEM from 4 experiments.*P<0.05. F, Confocal microscopy (left) shows the mitochondrial calcium levels (Rhod-2 fluorescence) in cultured normal rat right ventricularfibroblasts treated with or without TNF-α at 20 ng/mL. We performed 3 independent experiments from 3 rats (shown as 1, 2, and 3 on x axis),and for each experiment, we measured 10–15 cells for each treatment group. The Rhod-2 fluorescence level of each cell is shown in circlesalong with mean±SEM (right). n=40 cells from 3 independent experiments for each treatment group. We initially used the Mann-Whitneytest to show a significant difference between the 2 treatment groups (P<0.0001). To address potential random effects from variability amongdifferent cells in each rat, or among different rats, we also performed a linear mixed-effects model, which showed that the only difference wasbetween treatment groups (fixed effects): F(1, 34)=92.5; P<0.0001) and not among different rats within the same group (F(2, 38)= 0.539;P=0.5877). ****P<0.0001 when comparing these 2 groups. G, Confocal microscopy (left) shows the α-SMA levels (far red) in cultured normalrat right ventricular fibroblasts treated with vehicle, TGF-β (10 ng/mL), or a combination of TGF-β (10 ng/mL) and TNF-α (20 ng/mL). Weperformed 3 independent experiments from 3 rats (shown as 1, 2, and 3 on x axis), and for each experiment, we measured 15–20 cells for eachtreatment group. The α-SMA level of each cell is shown in circles along with mean±SEM (right). n=50 cells from 3 independent experiments foreach treatment group. We initially used the Kruskal-Wallis test to show significant differences among groups (P<0.0001). To address potentialrandom effects from variability among different cells in each rat, or among different rats, we also performed a linear mixed-effects model, whichshowed that the only difference was between groups (fixed effects): F(1.946, 157.6)=120.1; P<0.0001) and not among different rats withinthe same group (F(2, 162)=1.288; P=0.2785). Post hoc comparisons between groups were performed using Tukey’s multiple-comparisons test.****P<0.0001. H, Confocal microscopy (left) shows the mitochondrial calcium levels (Rhod-2 fluorescence) in cultured normal rat right ventricularfibroblasts treated with vehicle, TGF-β (10 ng/mL) or a combination of TGF-β (10 ng/mL) and TNF-α (20 ng/mL). We performed 3 independentexperiments from 3 rats (shown as 1, 2, and 3 on x axis), and for each experiment, we measured 15–20 cells for each group. The Rhod-2fluorescence level of each cell is shown in circles along with mean±SEM (right). n=50 cells from 3 independent experiments for each treatmentgroup. We initially used the Kruskal-Wallis test to show significant differences among treatment groups (P<0.0001). To address potential randomeffects from variability among different cells in each rat, or different rats, we also performed a linear mixed-effects model, which showed thatthe only difference was between treatment groups (fixed effects): (F(1.166, 58.88)=24.18; P<0.0001) and not among different rats within thesame group (F(2,54)=0.6307; P=0.5361). Post hoc comparisons between groups were performed using Tukey’s multiple-comparisons test. **P<0.01; **** P<0.0001. B and E, Comparisons were made using Kruskal-Wallis tests followed by pairwise Mann-Whitney U tests. α-SMA indicatesalpha smooth muscle actin; cFB, cardiac fibroblast; cRV, compensated right ventricle; CTRL, control (normal right ventricle, without monocrotalineinjection); dRV, decompensated right ventricle; HET, heterozygous; KO, knockout; LV, left ventricle; RV, right ventricle; TGF-β, transforming growthfactor beta; TNF-α, tumor necrosis factor alpha; UCP2, uncoupling protein 2; and WT, wild-type. confirmed with right heart catheterization and echocar-diography. The majority of PAH patients had availablemPAP, cardiac index (CI) and TAPSE values. Cohort 2(University of Alberta) included 21 patients with PAH() who underwent catheterization and echocar-diography within 2 days from each other, approximately1 year after the establishment of diagnosis and initia-tion of double oral PAH therapy with phosphodiesterasetype 5 inhibitors and endothelin receptor antagonists.Cohort 2 did not have RV tissues, as they were all aliveand not transplanted, but offered blood. The same treat-ment protocol across this cohort, allowed us to comparecRV and dRV independent of therapy. The diagnosis ofPAH in both cohorts was based on mPAP >20 mm Hgand PAWP <13 mmHg, along with structurally and func-tionally normal LV on echocardiography and the absenceof thromboembolic and parenchymal lung disease (chestCT angiogram and pulmonary function tests). Cohort 3included 30 patients whose RVs were obtained fromheart transplant recipients for ischemic or nonischemiccardiomyopathy (). 21 of them had group 2 PHT(PHT-2; mPAP >20 mm Hg) with available TAPSE or CIvalues for most patients. We identified dRV as CI <2.2 L/min/m2 or TAPSE<1.5 cm for all cohorts. We studied RVs using immunohis-tochemistry, droplet digital PCR (ddPCR), immunoblots,and the presence of a heterozygous and homozygousloss-of-function UCP2 SNP (rs659366) by ddPCR ineither RV tissues or blood cells. We correlated the num-ber of RV cMFBs or the presence of the SNP with RVUCP2 mRNA and protein levels as well as with CI andTAPSE. For PAH cohorts 1 and 2, we found that the num-ber of cMFBs did not increase from normal RV to cRVbut increased sharply in dRV in cohort 1 (Figure 5A) andcorrelated negatively with TAPSE and CI (Figure 5B and5C). We also found the cRV cFBs have increased cyto-plasmic PRMT1 and ADMA levels compared with normalRV fibroblasts but not significantly different comparedwith dRV cMFBs (Figure 5D and 5E); and no correla-tion was found with TAPSE or CI (). In contrast,the cMFB UCP2 levels are decreased in dRV comparedto normal and cRV fibroblasts (Figure 6A). The UCP2protein and mRNA levels correlated negatively with thenumber of RV cMFBs and positively with TAPSE or CI(Figure 6B through 6G). These human data suggested ### Page 13 ![Figure 1]() ### Page 14 ![Figure 2]() ### Page 15 that that the main difference from cRV to dRV is the pro-gressive decrease in UCP2 levels, similarly to the MCTrat data. The UCP2 SNP (rs659366) affects the promoter ofthe UCP2 gene, resulting in decreased UCP2 mRNAlevels in carriers of the -866 variant allele (G to A), com-pared with the WT genotype.42 The presence of thisloss-of-function UCP2 SNP is associated with worseoutcomes in PAH patients.19 We first measured the pres-ence of UCP2 SNP from all human RV tissues (cohorts 1and 3) and found that overall the presence of the SNPsis associated with decreased UCP2 protein and mRNAlevels in a gene dose-dependent manner, ie, progres-sively decreasing from WT to heterozygotes (HET) tohomozygotes (HOM; Figure 7A and 7B). This suggeststhat, overall, the presence of the germline SNP reflectsthe levels of UCP2 in the RV and thus could be used asa surrogate of RV UCP2 levels without the need of abiopsy. In cohort 1 patients with PAH, decreased TAPSEwas associated with the presence of UCP2 SNP, andthere was a trend for CI (Figure 7C and 7D). This wastrue even for patients who had similar mPAP. Patientswith WT UCP2 (white dots in Figure 7C) had higherTAPSE than patients with heterozygous SNP (grey dots)and homozygous UCP2 SNP (red dots), even though theyhad similar mPAPs (Figure 7C). In cohort 2 PAH patients,the same associations were found but with significancein CI (Figure 7E and 7F). When we pulled cohort 1 and 2patients with PAH, achieving sample sizes of 36 and 39for TAPSE and CI respectively, we found that the pres-ence of the SNP clearly separated dRV from cRV basedon both TAPSE and CI (Figure 7G and 7H). In thesepatients, the proportion of patients with heterozygous orhomozygous SNP increased from 39.1% of cRV patientsto 81.3% of patients with dRV (Figure 7I). For group 2 PHT (PHT-2; cohort 3), in contrast to thePAH cohorts, no differences in the number of RV cMFBswere observed between normal, cRV, and dRV (). However, RVs had higher numbers of cMFBscompared with control patients in cohort 1 under identi-cal immunohistochemistry protocols, potentially becausein these cardiomyopathy RVs, their biology is affectedby the primary disease as well (ie, ischemia and dilatedcardiomyopathy). No significant differences were foundin cytoplasmic PRMT1 or UCP2 levels in RV c(M)FBs inPHT 2 RVs (). In cohort 3 patientswith pulmonary hypertension (ie, mPAP >20 mm Hg), weobserved no association between TAPSE or CI and thepresence of the UCP2 SNP or with the RV UCP2 mRNAlevels (Figure 7J through 7L). There are several potentialexplanations for this difference with PAH, in addition tothe fact that the PHT-2 mPAPs were lower than in thePAH cohorts. First, PHT-2 is generally associated with alower inflammatory burden compared with PAH, and ourrat data showed that TNF-α decreases the expressionof UCP2. Second, in the cardiomyopathy cases of cohort3, the RV is affected by the primary disease as much asthe LV, over and above the effects of the RV afterload,unlike PAH. These findings suggest that UCP2 levels arespecifically associated with dRV in PAH, but not in PHT-2, although it is possible that a similar role for UCP2 maybe present in PHT-2 cases with severe rise in mPAP andperhaps significant systemic inflammation, as can be seenin obese patients with metabolic syndrome and HEpEF. ## DISCUSSION Our work challenges the concept that RV failure resultsfrom energetic failure and subsequent CM death, andthat it occurs solely and predictably because of the in-crease in its afterload (ie, mPAP) in PAH. Instead, weshowed that genetic (ie, loss-of-function UCP2 SNPs)and extramyocardial factors (ie, inflammation), whichlead to a loss of RV cFB UCP2 levels promoting dif-ferentiation to cMFB, are associated with acceleratedtransition of cRV to dRV in both rodents and patientswith PAH. This suggests that perhaps the focus for thetreatment of PAH, that currently is restricted to target-ing pulmonary vascular tone and remodeling, should ex-pand and include treatments that may prevent or evenreverse RV cMFB activation. Importantly, we suggest thatthe presence of UCP2 SNPs may identify patients witha predisposition to RV decompensation early after PAHdiagnosis, currently not possible, facilitating optimal clini-cal management (eg, therapy intensification or earlierconsideration for transplantation). Although this needsto be further prospectively confirmed in multicenter co-horts, we believe our work opens a new window to muchneeded precision health approaches to PAH, a relativelycommon disease (in contrast to the rare iPAH) that re-mains deadly despite very expensive treatment options. ### Animal Models and Mechanistic Studies An MCT injection leads to severe pulmonary hyperten-sion with an inflammatory phenotype in several organsand, after a short phase of cRV hypertrophy, causes dRVand death in ≈4 to 5 weeks (). A completeright heart catheterization with an advancement of a Mil-lar catheter to the main pulmonary artery to measuremPAP (as in humans), as well as simultaneous echo-cardiography, allowed us to define hemodynamic andstructural criteria (which included, among several others,CO and TAPSE) to define the compensatory RV hyper-trophy, in which CO and TAPSE are preserved, versusthe decompensated RV, in which, in addition to a dropin CO and TAPSE, there is a “paradoxical” decrease inRV systolic pressure attributable to RV contractile failure,an increase in RA pressure, and an increase in RVEDD.Thus, all MCT-treated rat RV tissues were characterizedas cRV or dRV, seldom done in the field, where thereis often no distinction between the 2. This allowed us ### Page 16 ![Figure 5]() ### Page 17 Figure 7 Continued. Scatter dot plots represent individual values. n=11 for WT, n=10 for UCP2 HET SNP, and n=4 for UCP2 HOM SNP.*P<0.05; **P<0.01. C, UCP2 genotyping and echocardiography (left) show the UCP2 genotype and TAPSE in patients with PAH (mPAP >20mm Hg from cohort 1). Patients were genotyped for UCP2 and categorized into WT (n=6), UCP2 HET SNP (n=5), and UCP2 HOM SNP (n=4).Quantification values (right) are expressed as mean±SEM. Scatter dot plots represent individual values. *P<0.05; **P<0.01. D, UCP2 genotypingand echocardiography (left) show the UCP2 genotype and cardiac index (CI) in patients with PAH (mPAP>20mmHg from cohort 1). Patientswere genotyped for UCP2 and categorized into WT (n=8), UCP2 HET SNP (n=6), and UCP2 HOM SNP (n=4). Quantification values (right)are expressed as mean±SEM; Scatter dot plots represent individual values. E, UCP2 genotyping and echocardiography (left) show the UCP2genotype (from blood) and TAPSE in patients with PAH (mPAP>20mmHg from cohort 2). Patients were genotyped for UCP2 and categorizedinto WT (n=10), UCP2 HET SNP (n=6), and UCP2 HOM SNP (n=5). Quantification values (right) are expressed as mean±SEM. Scatter dotplots represent individual values. *P<0.05; **P<0.01. F, UCP2 genotyping and echocardiography (left) show the UCP2 genotype and CI inpatients with PAH (mPAP >20 mm Hg from cohort 2). Patients were genotyped for UCP2 and categorized into WT (n=10), UCP2 HET SNP(n=6), and UCP2 HOM SNP (n=5). Quantification values (right) are expressed as mean±SEM. Scatter dot plots represent individual values.***P<0.001. G, UCP2 genotyping and echocardiography (left) show the UCP2 genotype and TAPSE in patients with PAH (mPAP >20 mm Hgfrom cohort 1 and 2). Patients were genotyped for UCP2 and categorized into WT (n=16), UCP2 HET SNP (n=11), and UCP2 HOM SNP (n=9).Quantification values (right) are expressed as mean±SEM. Scatter dot plots represent individual values. *P<0.05; **P<0.01; ****P<0.0001. H,UCP2 genotyping and echocardiography (left) show the UCP2 genotype and CI in patients with PAH (mPAP >20 mm Hg from cohort 1 and 2).Patients were genotyped for UCP2 and categorized into WT (n=18), UCP2 HET SNP (n=12), and UCP2 HOM SNP (n=9). Quantification values(right) are expressed as mean±SEM. Scatter dot plots represent individual values. *P<0.05. I, UCP2 genotyping shows the ratio of patientswith WT, UCP2-HET SNP, and UCP2-HOM SNP in cRV (n=23) vs dRV (n=16) in patients with PAH (cohort 1 and 2). J, UCP2 genotyping andechocardiography (left) show the UCP2 genotype and TAPSE in patients with group 2 PHT (PHT-2; mPAP >20 mm Hg from cohort 3). Patientswere genotyped for UCP2 and categorized into WT (n=6), UCP2 HET SNP (n=8), and UCP2 HOM SNP (n=3). Quantification values (right)are expressed as mean±SEM. Scatter dot plots represent individual values. K, UCP2 genotyping and echocardiography (left) show the UCP2genotype and CI in patients with group 2 PHT (mPAP >20 mm Hg from cohort 3). Patients were genotyped for UCP2 and categorized intoWT (n=8), UCP2 HET SNP (n=9), and UCP2 HOM SNP (n=4). Quantification values (right) are expressed as mean±SEM. Scatter dot plotsrepresent individual values. L, Spearman correlation test shows no significant correlation between UCP2 mRNA levels and TAPSE in patients withgroup 2 PHT (mPAP >20 mm Hg from cohort 3). Scatter dot plots show individual values. These comparisons were made using Kruskal-Wallistests followed by pairwise Mann-Whitney U tests. cFB indicates cardiac fibroblast; cMFB, cardiac myofibroblast; cRV, compensated right ventricle;dRV, decompensated right ventricle; mPAP, mean PA pressure; nRV, normal right ventricle; PAH, pulmonary arterial hypertension; PHT-2, group 2pulmonary hypertension; SNP, single nucleotide polymorphism; UCP2, uncoupling protein 2; UCP2 HET, heterozygous UCP2 SNP; UCP2 HOM,homozygous UCP2 SNP; TAPSE, tricuspid annular plane systolic excursion; VDAC, voltage-dependent anion channel; and WT, wild-type. not only to observe a sharp increase in the number ofcMFBs in dRV, but also an increase in PRMT1 levels andMICU1 methylation in both the cRV and dRV comparedwith control hearts; but importantly, that there was alsoan ongoing decrease in UCP2 levels that specificallyseparated and defined the transition from cRV to dRV. Mechanistically, we found that a central driver of fibro-blast differentiation is the reduction in mCa++ levels,driven by the loss of UCP2 in the setting of the MICU1hypermethylation (Figure 8). Decreased mCa++ levelssuppress the activity of Ca++-sensitive mitochondrialenzymes (as we confirmed with PDH) and the produc-tion of diffusible metabolites that regulate the epigeneticchanges involved in the change of cFB cell identity. Ourwork showed that dRV cMFB have decreased citrate andfumarate (Figure 2F), which are involved in histone acet-ylation via the production of acetyl-CoA in the nucleusby the ATP-citrate ligase (ACLY).43 In addition, we havepreviously shown that translocation of mitochondrialPDH into the nucleus directly affects histone acetylationvia production of acetyl-CoA in the nucleus.44 The PDHinhibition is in keeping with work showing that myofibro-blast activation requires histone deacetylation-mediatedgene repression.45,46 To confirm the direct primary role ofcMFBs UCP2 loss on mCa++ regulation independent ofMICU1 methylation, we showed that cFBs from homozy-gote or heterozygote Ucp2 KO mice have a gene dose–dependent decreased mCa++. Moreover, we showedthat UCP2 expression levels decreased in response toTNF-α alone, or TNF-α in combination with TGF-β ([-](https://www.ahajournals.org/doi/suppl/10.1161/CIRCULATIONAHA.125.078674)), consistent with the finding that TNF-α alonecan activate myofibroblasts, and that TNF-α together withTGF-β further enhances cMFB activation ().Disturbances in the TGF-β axis are common in PAH, inboth familiar PAH attributable to BMPR mutations andsporadic cases not related to known mutations32–37; andinflammation and cytokines like TNF-α are upregulatedboth in rodent and human PAH.47 Interestingly, it has beenshown by others that loss of UCP2 is associated withactivation of tissue macrophages,48 and in our Ucp2 KOmice lungs we observed a significant increase in the lunginflammatory cells infiltration (also a feature of humanPAH21). Thus, a germline UCP2 loss-of-function SNPcould affect UCP2 levels in both bone marrow–derivedinflammatory cells in the lungs and RV cFBs, produc-ing a mutually reinforcing feedback loop between proin-flammatory cMFBs and circulating inflammatory cells topotentiate cMFB activation and acceleration of dRV. This important role of inflammation was supported bythe fact that in the less inflammatory PAB model (com-pared with MCT) there was no progressive loss of UCP2(if anything, there was some increase), although, similarto MCT, there was an increase in MICU1 methylation incRV and dRV compared with control. This explains whythe PAB rats, despite an even higher increase in RVafterload, have fewer RV cMFBs, much delayed cRV-to-dRV transition, and much decreased mortality comparedwith MCT rats (). ### Page 18 ![Figure 3]() Our human cohorts included cohort 1 (Laval Univer-sity): RV tissues and clinical data including CI or TAPSEfor the majority of patients who had been formally diag-nosed with group 1 PAH (mPAP >20 mm Hg, pulmonaryartery wedge pressure (PAWP) <13 mm Hg, normal leftventricular function, no thromboembolic lung or paren-chymal lung disease). Cohort 2 (University of Alberta)included blood and clinical data of formally diagnosedgroup 1 PAH patients who had catheterization and echo-cardiography (performed within 2 days from each other)1 year after the initiation of oral PAH therapy (PDE5inhibitors and endothelin receptor antagonists). Cohort3 (Duke University) included RV tissues and clinicaldata from heart transplant recipients with group 2 PHT(mPAP >20 mm Hg) attributable to ischemic or dilatedcardiomyopathy. We were able to detect the presence ofthe UCP2 SNP (rs659366) in either RV tissues or bloodsamples from a total of 81 patients and show that it cor-relates negatively with UCP2 levels. In the patients fromcohort 1 and 3, we also measured RV UCP2 mRNA andprotein levels and correlate them with CO and TAPSE.First, the negative correlation between the presence ofa homozygous or heterozygous SNP with UCP2 levelsconfirmed that this germline SNP reflects a decrease inRV tissue UCP2 levels across all cohorts. Second, thenegative correlation between UCP2 SNP and RV func-tion (CI and TAPSE) in PAH but not in PHT-2 patients,suggests that the UCP2 SNP is associated more with ### Page 19 dRV than cRV in PAH and this is true even for patientsthat have similar mPAP. Overall, these data suggest thatthe UCP2 SNP can serve as a surrogate for RV UCP2levels without the need of a biopsy, as well as a predic-tor or earlier cRV-to-dRV transition and worse progno-sis (independent of the decree of mPAP rise) in livingPAH patients. Our work may provide explanation on whysome patients with PAH decompensate faster than oth-ers independently of the rise in mPAP. Similarly, our workmay explain why some patients after lung transplantationrecover their RVs later or less completely than others. ### Clinical Implications Our work suggests that biomarker-based (ie, UCP2SNPs and TNF-α serum levels), precision health ap-proaches could be applied in PAH patients, early afterthe establishment of the diagnosis, when it comes toanticipating, preventing or potentially reversing dRV inthe future. A patient that carries the UCP2 SNP (par-ticularly in a homozygote manner) can be followed moreclosely, with intense PAH and supportive therapy. Suchpatients should perhaps have a lower threshold for trans-plantation. Clinical trials with precision health design mayassess whether such patients may benefit from anti-inflammatory strategies (eg, anti–TNF-α) more thanpatients without the SNP, when it comes to RV decom-pensation. At this point, anti-inflammatory strategies areonly attempted with a goal to improve pulmonary vascu-lar remodeling. The same is true with therapies targetingthe TGF-β/BMPR axis (like sotatercept)32,33 the targetsof which may include RV fibroblasts as much or evenmore than pulmonary artery vascular cells. Our work sug-gests that both of these classes of therapeutics couldalso be applied with a goal of preventing or reversingRV decompensation. However, in such studies it may bedifficult to separate vascular from cardiac effects, unlessspecifically designed and interpreted with this in mind. At this point, there is no RV-specific therapy for dRVin PAH, with the exception of perhaps mild selectiveRV inotropic effects for PDE5 inhibitors.49,50 It is inter-esting that our work suggests that the answer to RVfailure may not be increased RV inotropy, a reminderof the challenges of inotropes in the treatment of LVfailure. Instead, we suggest that perhaps better targetsmay be the mitochondria (ie, metabolic modulatorslike dichloroacetate51,52) or epigenetic pathways thatregulate RV fibroblast differentiation,53 along with anti-inflammatory therapies. It has been shown that TNF-αdecreases UCP2 expression via inducible nitric oxidesynthase (iNOS).25 Other pathways like NF-kB, MAPKor PI3K could also be involved. UCP2 loss-of-function SNPs are relatively common.For example, the rs659366 G-to-A SNP that we studied,has been reported in ≈16% of patients with obesity anddiabetes in the homozygous form and in up to 40% inthe heterozygous form.54,55 Because it is so common, thisSNP may not appear in genetic databases comparingPAH with control patients, which typically focus on thediscovery of rare mutations. For example, the rs659366SNP may appear in PAH but filtered out because it is socommon to non-PAH controls that may have obesity ordiabetes and carrying the SNP, suggesting that this maybe a feature of the non-PAH controls rather than PAH.When we specifically looked for this SNP using PCR in2 of our previous studies in the Journal of the AmericanHeart Association and Science Translational Medicine, wefound that its prevalence in PAH patients is indeed simi-lar to that reported for the obese and diabetes patients,and furthermore, associated with worse outcomes andresistance to therapy compared with noncarriers.21,51 Finally, we found that, unexpectedly, the dRV CMshad increased mitochondrial respiration in contrast to thedRV cMFBs that had decreased oxidative phosphoryla-tion and increased glycolysis and lactate production. Infact, using labeled lactate we found that the CMs canuse the lactate produced by neighboring cMFBs as fuelto perhaps compensate for the potentially decreasedATP production. This suggests that the interpretationof functional RV metabolic imaging (eg, by FDG-PET56)needs to be interpreted carefully and differently in theclinic, considering that a large part of the signal maycome from non-CM cells and may not directly reflect theenergetic or metabolic status of the CMs as it may do innormal RVs. ## LIMITATIONS A mouse model with germline expression of a loss-offunction gene variant like the UCP2 rs659366, wouldbe ideally needed to study the response of the RV underincreased afterload, in either PAH or PA banding mod-els, better recapitulating the condition of PAH patientswith the UCP2 SNP. Alternatively, a conditional cardiacfibroblast specific Ucp2 KO could reveal a direct causalrole of UCP2 loss to RV decompensation. However, inthis first implication of UCP2 in RV failure, we provideddata that: (1) show that both UCP2 mRNA and proteinlevels decrease in human RVs in a gene dose-dependentmanner by the presence of a loss-of-function SNP; (2)TNF-α alone or in combination with TGF-β decreasesRV UCP2 levels in fibroblasts but not in CMs; (3) in amouse Ucp2 KO model (which we previously showed isassociated with pulmonary hypertension), we found thatRV cMFBs have decreased mCa++ levels, the basis ofthe cMFB activation shown in our and in other studies(at least on the left ventricle); and (4) in the absence ofUCP2, TNF-α does not promote cMFB differentiation.We believe that although our study did not prove a directcausal role of UCP2 loss in RV failure to identify it as atherapeutic target, it provides enough support to identifyUCP2 SNP as a much-needed biomarker predicting RV ### Page 20 failure in humans, with an easy test (eg, blood or buccalmucosa detection of SNPs with a simple PCR). Our provocative finding that in rat RV failure thecontractility of isolated CMs is not decreased as com-monly assumed, may be biased by the technique of invitro sarcomere length measurements, which is select-ing mostly morphologically normal and beating CMs,rather than less healthy-looking ones. In addition, ourdata were not supported by similar studies in humanCMs isolated from failing RVs, a very challenging task.Also, the fact that TNF-α decreases UCP2 mRNA andprotein levels in RV cFBs but not in RV CMs suggestscell type–specific signaling or post-translational modi-fication, which we did not explore further. We hope thatour findings will inspire such important future studiesin centers that can accomplish this. Nevertheless, ourstudy is a reminder that perhaps targeting CM inotropymay not necessarily be an attractive option in RV fail-ure, as shown by the many failed attempts to do this inLV failure. ## ARTICLE INFORMATION Received December 4, 2025; accepted February 10, 2026. ### Affiliations Department of Medicine (Cardiology), Faculty of Medicine and Dentistry, Univer-sity of Alberta, Edmonton, Alberta, Canada (Y.Z., J.P., A.H., Y.L., Y.-Y.Z., G.S., E.D.M.).Pulmonary Hypertension and Vascular Biology Research Group CRIUCPQ,Department of Medicine, Laval University, Quebec, Canada (S.B., S.P., S.B.-B.,S.-E.L.). Institute for Lung Health, German Center for Lung Research (DZL),Cardio-Pulmonary Institute, Justus Liebig University Giessen, Giessen, Ger-many (S.B.). Department of Surgery, Duke University Medical Center, Durham,NC(D.E.B., M.M.P.). ### Author Contributions Y.Z. and E.D.M. conceived and designed the study. Y.Z., S.B., S.P., J.P., A.H., Y.L., Y.-Y.Z., S.B.-B, S.-E.L., D.E.B., M.M.P., and G.S. performed experiments and analyzedresults. Y.Z. and E.D.M. wrote the article. All authors revised and approved thearticle. ### Sources of Funding Y. Zhang is supported by Paroian Family PH Research Scholarship. E. Michelakisis supported by Heart and Stroke Foundation of Canada, Canadian Institutes ofHealth Research and Canada Research Chair (tier 1). G. Sutendra is supportedby an Alberta Innovates translational health chair and a Canada research chair(tier 2), along with grants from Canadian Institutes of Health Research. S. Bonnetis supported by a Canada research chair (tier 1), and this work was supported bythe Heart and Stroke Foundation of Canada and Canadian Institutes of HealthResearch. ### Disclosures None. ### Supplemental Material Checklist Expanded Methods Figures S1–S10 Tables S1–S3 References 57–61 ## REFERENCES 1. Michelakis ED, Wilkins MR, Rabinovitch M. Emerging concepts andtranslational priorities in pulmonary arterial hypertension. Circulation.2008;118:1486–1495. doi: 10.1161/CIRCULATIONAHA.106.673988 2. Sutendra G, Michelakis ED. Pulmonary arterial hypertension: chal-lenges in translational research and a vision for change. Sci Transl Med.2013;5:208sr205. doi: 10.1126/scitranslmed.3005428 3. McLaughlin VV, Archer SL, Badesch DB, Barst RJ, Farber HW, Lindner JR,Mathier MA, McGoon MD, Park MH, Rosenson RS, et al; ACCF/AHA. ACCF/AHA 2009 expert consensus document on pulmonary hypertension: a re-port of the American College of Cardiology Foundation Task Force on ExpertConsensus Documents and the American Heart Association: developed incollaboration with the American College of Chest Physicians, American Tho-racic Society, Inc., and the Pulmonary Hypertension Association. Circulation.2009;119:2250–2294. doi: 10.1161/CIRCULATIONAHA.109.192230 4. Lorell BH, Carabello BA. Left ventricular hypertrophy: pathogen-esis, detection, and prognosis. Circulation. 2000;102:470–479. doi:10.1161/01.cir.102.4.470 5. Zelt JGE, Chaudhary KR, Cadete VJ, Mielniczuk LM, Stewart DJ. Medicaltherapy for heart failure associated with pulmonary hypertension. Circ Res.2019;124:1551–1567. doi: 10.1161/CIRCRESAHA.118.313650 6. Rich S, Pogoriler J, Husain AN, Toth PT, Gomberg-Maitland M, Archer SL.Long-term effects of epoprostenol on the pulmonary vasculature in idio-pathic pulmonary arterial hypertension. Chest. 2010;138:1234–1239. doi:10.1378/chest.09-2815 7. Zhou B, Tian R. Mitochondrial dysfunction in pathophysiology of heart fail-ure. J Clin Invest. 2018;128:3716–3726. doi: 10.1172/JCI120849 8. Iacobazzi D, Suleiman MS, Ghorbel M, George SJ, Caputo M, Tulloh RM.Cellular and molecular basis of RV hypertrophy in congenital heart disease.Heart. 2016;102:12–17. doi: 10.1136/heartjnl-2015-308348 9. Hansmann G, Diekmann F, Chouvarine P, Ius F, Carlens J, Schwerk N,Warnecke G, Vogel-Claussen J, Hohmann D, Alten T, et al. Full recovery ofright ventricular systolic function in children undergoing bilateral lung trans-plantation for severe PAH. J Heart Lung Transplant. 2022;41:187–198. doi:10.1016/j.healun.2021.10.014 10. Venugopal H, Hanna A, Humeres C, Frangogiannis NG. Properties andfunctions of fibroblasts and myofibroblasts in myocardial infarction. Cells.2022;11:1386. doi: 10.3390/cells11091386 11. Humeres C, Frangogiannis NG. Fibroblasts in the infarcted, remodel-ing, and failing heart. JACC Basic Transl Sci. 2019;4:449–467. doi:10.1016/j.jacbts.2019.02.006 12. Fowlkes V, Clark J, Fix C, Law BA, Morales MO, Qiao X, Ako-Asare K,Goldsmith JG, Carver W, Murray DB, et al. Type II diabetes promotes a myo-fibroblast phenotype in cardiac fibroblasts. Life Sci. 2013;92:669–676. doi:10.1016/j.lfs.2013.01.003 13. Kamo T, Akazawa H, Komuro I. Cardiac nonmyocytes in thehub of cardiac hypertrophy. Circ Res. 2015;117:89–98. doi:10.1161/CIRCRESAHA.117.305349 14. Nagaraju CK, Robinson EL, Abdesselem M, Trenson S, Dries E, Gilbert G,Janssens S, Van Cleemput J, Rega F, Meyns B, et al. Myofibroblast pheno-type and reversibility of fibrosis in patients with end-stage heart failure. J AmColl Cardiol. 2019;73:2267–2282. doi: 10.1016/j.jacc.2019.02.049 15. Reddy S, Bernstein D. Molecular mechanisms of right ven-tricular failure. Circulation. 2015;132:1734–1742. doi:10.1161/CIRCULATIONAHA.114.012975 16. Lippmann MR, Maron BA. The right ventricle: from embryologic de-velopment to RV failure. Curr Heart Fail Rep. 2022;19:325–333. doi:10.1007/s11897-022-00572-z 17. Lombardi AA, Gibb AA, Arif E, Kolmetzky DW, Tomar D, Luongo TS, JadiyaP, Murray EK, Lorkiewicz PK, Hajnoczky G, et al. Mitochondrial calcium ex-change links metabolism with the epigenome to control cellular differentia-tion. Nat Commun. 2019;10:4509. doi: 10.1038/s41467-019-12103-x 18. Gibb AA, Lazaropoulos MP, Elrod JW. Myofibroblasts and fibrosis: mi-tochondrial and metabolic control of cellular differentiation. Circ Res.2020;127:427–447. doi: 10.1161/CIRCRESAHA.120.316958 19. Koshenov Z, Oflaz FE, Hirtl M, Bachkoenig OA, Rost R, Osibow K, GottschalkB, Madreiter-Sokolowski CT, Waldeck-Weiermair M, Malli R, et al. The con-tribution of uncoupling protein 2 to mitochondrial Ca(2+) homeostasis inhealth and disease - a short revisit. Mitochondrion. 2020;55:164–173. doi:10.1016/j.mito.2020.10.003 20. Madreiter-Sokolowski CT, Klec C, Parichatikanond W, Stryeck S, GottschalkB, Pulido S, Rost R, Eroglu E, Hofmann NA, Bondarenko AI, et al. PRMT1-mediated methylation of MICU1 determines the UCP2/3 dependencyof mitochondrial Ca(2+) uptake in immortalized cells. Nat Commun.2016;7:12897. doi: 10.1038/ncomms12897 21. Zhang Y, Zervopoulos SD, Boukouris AE, Lorenzana-Carrillo MA, SalemeB, Webster L, Liu Y, Haromy A, Tabatabaei Dakhili SA, Ussher JR, etal. SNPs for genes encoding the mitochondrial proteins sirtuin3 and ### Page 21 uncoupling protein 2 are associated with disease severity, type 2 diabetes,and outcomes in patients with pulmonary arterial hypertension and this isrecapitulated in a new mouse model lacking both genes. J Am Heart Assoc.2021;10:e020451. doi: 10.1161/JAHA.120.020451 22. Dromparis P, Paulin R, Sutendra G, Qi AC, Bonnet S, Michelakis ED. Uncou-pling protein 2 deficiency mimics the effects of hypoxia and endoplasmicreticulum stress on mitochondria and triggers pseudohypoxic pulmonaryvascular remodeling and pulmonary hypertension. Circ Res. 2013;113:126–136. doi: 10.1161/CIRCRESAHA.112.300699 23. Paulin R, Sutendra G, Gurtu V, Dromparis P, Haromy A, Provencher S, BonnetS, Michelakis ED. A miR-208-Mef2 axis drives the decompensation of rightventricular function in pulmonary hypertension. Circ Res. 2015;116:56–69.doi: 10.1161/CIRCRESAHA.115.303910 24. Soon E, Holmes AM, Treacy CM, Doughty NJ, Southgate L, MachadoRD, Trembath RC, Jennings S, Barker L, Nicklin P, et al. Elevated lev-els of inflammatory cytokines predict survival in idiopathic and familialpulmonary arterial hypertension. Circulation. 2010;122:920–927. doi:10.1161/CIRCULATIONAHA.109.933762 25. Gomez-Hernandez A, Perdomo L, de las Heras N, Beneit N, EscribanoO, Otero YF, Guillen C, Diaz-Castroverde S, Gozalbo-Lopez B, CachofeiroV, et al. Antagonistic effect of TNF-alpha and insulin on uncoupling pro-tein 2 (UCP-2) expression and vascular damage. Cardiovasc Diabetol.2014;13:108. doi: 10.1186/s12933-014-0108-9 26. Bonora M, De Marchi E, Patergnani S, Suski JM, Celsi F, Bononi A, Giorgi C,Marchi S, Rimessi A, Duszynski J, et al. Tumor necrosis factor-alpha impairsoligodendroglial differentiation through a mitochondria-dependent process.Cell Death Differ. 2014;21:1198–1208. doi: 10.1038/cdd.2014.35 27. Zhu Z, Godana D, Li A, Rodriguez B, Gu C, Tang H, Minshall RD, Huang W,Chen J. Echocardiographic assessment of right ventricular function in exper-imental pulmonary hypertension. Pulm Circ. 2019;9:2045894019841987.doi: 10.1177/2045894019841987 28. Sutendra G, Michelakis ED. The metabolic basis of pulmonary arterial hyper-tension. Cell Metab. 2014;19:558–573. doi: 10.1016/j.cmet.2014.01.004 29. Jha MK, Morrison BM. Lactate transporters mediate glia-neuron metaboliccrosstalk in homeostasis and disease. Front Cell Neurosci. 2020;14:589582.doi: 10.3389/fncel.2020.589582 30. Liu L, MacKenzie KR, Putluri N, Maletic-Savatic M, Bellen HJ. The glia-neuron lactate shuttle and elevated ROS Promote lipid synthesis in neu-rons and lipid droplet accumulation in glia via APOE/D. Cell Metab.2017;26:719–737.e6. doi: 10.1016/j.cmet.2017.08.024 31. Merial C, Bouloumie A, Trocheris V, Lafontan M, Galitzky J. Nitric oxide-dependent downregulation of adipocyte UCP-2 expression by tumor necro-sis factor-alpha. Am J Physiol Cell Physiol. 2000;279:C1100–C1106. doi:10.1152/ajpcell.2000.279.4.C1100 32. Guignabert C, Humbert M. Targeting transforming growth factor-beta re-ceptors in pulmonary hypertension. Eur Respir J. 2021;57:2002341. doi:10.1183/13993003.02341-2020 33. Long L, Crosby A, Yang X, Southwood M, Upton PD, Kim DK, Morrell NW.Altered bone morphogenetic protein and transforming growth factor-betasignaling in rat models of pulmonary hypertension: potential for activin receptor-like kinase-5 inhibition in prevention and progression of disease. Circulation.2009;119:566–576. doi: 10.1161/CIRCULATIONAHA.108.821504 34. Sanada TJ, Sun XQ, Happe C, Guignabert C, Tu L, Schalij I, Bogaard HJ,Goumans MJ, Kurakula K. Altered TGFbeta/SMAD signaling in human andrat models of pulmonary hypertension: an old target needs attention. Cells.2021;10:84. doi: 10.3390/cells10010084 35. Thomson JR, Machado RD, Pauciulo MW, Morgan NV, Humbert M, ElliottGC, Ward K, Yacoub M, Mikhail G, Rogers P, et al. Sporadic primary pulmo-nary hypertension is associated with germline mutations of the gene en-coding BMPR-II, a receptor member of the TGF-beta family. J Med Genet.2000;37:741–745. doi: 10.1136/jmg.37.10.741 36. Koehler R, Grunig E, Pauciulo MW, Hoeper MM, Olschewski H, WilkensH, Halank M, Winkler J, Ewert R, Bremer H, et al. Low frequency ofBMPR2 mutations in a German cohort of patients with sporadic idio-pathic pulmonary arterial hypertension. J Med Genet. 2004;41:e127. doi:10.1136/jmg.2004.023101 37. Long L, Ormiston ML, Yang X, Southwood M, Graf S, Machado RD, MuellerM, Kinzel B, Yung LM, Wilkinson JM, et al. Selective enhancement of endo-thelial BMPR-II with BMP9 reverses pulmonary arterial hypertension. NatMed. 2015;21:777–785. doi: 10.1038/nm.3877 38. Feng S, Chen S, Yu W, Zhang D, Zhang C, Tang C, Du J, Jin H. H(2)Sinhibits pulmonary arterial endothelial cell inflammation in rats withmonocrotaline-induced pulmonary hypertension. Lab Invest. 2017;97:268–278. doi: 10.1038/labinvest.2016.129 39. Luitel H, Sydykov A, Schymura Y, Mamazhakypov A, Janssen W, PradhanK, Wietelmann A, Kosanovic D, Dahal BK, Weissmann N, et al. Pressureoverload leads to an increased accumulation and activity of mast cells in theright ventricle. Physiol Rep. 2017;5:e13146. doi: 10.14814/phy2.13146 40. Pischke SE, Hestenes S, Johansen HT, Fure H, Bugge JF, Espinoza A,Skulstad H, Edvardsen T, Fosse E, Mollnes TE, et al. Sepsis causes rightventricular myocardial inflammation independent of pulmonary hyper-tension in a porcine sepsis model. PLoS One. 2019;14:e0218624. doi:10.1371/journal.pone.0218624 41. Baumgarten G, Knuefermann P, Kalra D, Gao F, Taffet GE, MichaelL, Blackshear PJ, Carballo E, Sivasubramanian N, Mann DL. Load-dependent and -independent regulation of proinflammatory cytokine andcytokine receptor gene expression in the adult mammalian heart. Circulation.2002;105:2192–2197. doi: 10.1161/01.cir.0000015608.37608.18 42. Wang H, Chu WS, Lu T, Hasstedt SJ, Kern PA, Elbein SC. Uncoupling protein-2polymorphisms in type 2 diabetes, obesity, and insulin secretion. Am J PhysiolEndocrinol Metab. 2004;286:E1–E7. doi: 10.1152/ajpendo.00231.2003 43. Wellen KE, Hatzivassiliou G, Sachdeva UM, Bui TV, Cross JR, ThompsonCB. ATP-citrate lyase links cellular metabolism to histone acetylation. Sci-ence. 2009;324:1076–1080. doi: 10.1126/science.1164097 44. Sutendra G, Kinnaird A, Dromparis P, Paulin R, Stenson TH, Haromy A,Hashimoto K, Zhang N, Flaim E, Michelakis ED. A nuclear pyruvate dehydro-genase complex is important for the generation of acetyl-CoA and histoneacetylation. Cell. 2014;158:84–97. doi: 10.1016/j.cell.2014.04.046 45. Jones DL, Haak AJ, Caporarello N, Choi KM, Ye Z, Yan H, Varelas X, OrdogT, Ligresti G, Tschumperlin DJ. TGFbeta-induced fibroblast activation re-quires persistent and targeted HDAC-mediated gene repression. J Cell Sci.2019;132:jcs233486. doi: 10.1242/jcs.233486 46. Smith ER, Wigg B, Holt S, Hewitson TD. TGF-beta1 modifies histone acetyla-tion and acetyl-coenzyme A metabolism in renal myofibroblasts. Am J Physi-ol Renal Physiol. 2019;316:F517-F529. doi: 10.1152/ajprenal.00513.2018 47. Sutendra G, Dromparis P, Bonnet S, Haromy A, McMurtry MS, BleackleyRC, Michelakis ED. Pyruvate dehydrogenase inhibition by the inflam-matory cytokine TNFalpha contributes to the pathogenesis of pulmo-nary arterial hypertension. J Mol Med (Berl). 2011;89:771–783. doi:10.1007/s00109-011-0762-2 48. Emre Y, Hurtaud C, Karaca M, Nubel T, Zavala F, Ricquier D. Role of uncou-pling protein UCP2 in cell-mediated immunity: how macrophage-mediatedinsulitis is accelerated in a model of autoimmune diabetes. Proc Natl AcadSci U S A. 2007;104:19085–19090. doi: 10.1073/pnas.0709557104 49. Nagendran J, Archer SL, Soliman D, Gurtu V, Moudgil R, Haromy A, St AubinC, Webster L, Rebeyka IM, Ross DB, et al. Phosphodiesterase type 5 ishighly expressed in the hypertrophied human right ventricle, and acuteinhibition of phosphodiesterase type 5 improves contractility. Circulation.2007;116:238–248. doi: 10.1161/CIRCULATIONAHA.106.655266 50. Archer SL, Michelakis ED. Phosphodiesterase type 5 inhibitors for pul-monary arterial hypertension. N Engl J Med. 2009;361:1864–1871. doi:10.1056/NEJMct0904473 51. Michelakis ED, Gurtu V, Webster L, Barnes G, Watson G, Howard L,Cupitt J, Paterson I, Thompson RB, Chow K, et al. Inhibition of pyruvatedehydrogenase kinase improves pulmonary arterial hypertension in ge-netically susceptible patients. Sci Transl Med. 2017;9:eaao4583. doi:10.1126/scitranslmed.aao4583 52. Sutendra G, Michelakis ED. Pyruvate dehydrogenase kinase as a novel therapeu-tic target in oncology. Front Oncol. 2013;3:38. doi: 10.3389/fonc.2013.00038 53. Zhao L, Chen CN, Hajji N, Oliver E, Cotroneo E, Wharton J, Wang D, LiM, McKinsey TA, Stenmark KR, et al. Histone deacetylation inhibitionin pulmonary hypertension: therapeutic potential of valproic acid andsuberoylanilide hydroxamic acid. Circulation. 2012;126:455–467. doi:10.1161/CIRCULATIONAHA.112.103176 54. Dalgaard LT, Andersen G, Larsen LH, Sorensen TI, Andersen T, DrivsholmT, Borch-Johnsen K, Fleckner J, Hansen T, Din N, et al. Mutational analysisof the UCP2 core promoter and relationships of variants with obesity. ObesRes. 2003;11:1420–1427. doi: 10.1038/oby.2003.191 55. D’Adamo M, Perego L, Cardellini M, Marini MA, Frontoni S, Andreozzi F,Sciacqua A, Lauro D, Sbraccia P, Federici M, et al. The -866A/A geno-type in the promoter of the human uncoupling protein 2 gene is associ-ated with insulin resistance and increased risk of type 2 diabetes. Diabetes.2004;53:1905–1910. doi: 10.2337/diabetes.53.7.1905 56. Fang W, Zhao L, Xiong CM, Ni XH, He ZX, He JG, Wilkins MR. Com-parison of 18F-FDG uptake by right ventricular myocardium in idiopathicpulmonary arterial hypertension and pulmonary arterial hypertension as-sociated with congenital heart disease. Pulm Circ. 2012;2:365–372. doi:10.4103/2045-8932.101651 ### Page 22 57. Suen CM, Chaudhary KR, Deng Y, Jiang B, Stewart DJ. Fischer rats exhibitmaladaptive structural and molecular right ventricular remodelling in severepulmonary hypertension: a genetically prone model for right heart failure.Cardiovasc Res. 2019;115:788–799. doi: 10.1093/cvr/cvy258 58. Lorenzana-Carrillo MA, Gopal K, Byrne NJ, Tejay S, Saleme B, Das SK,Zhang Y, Haromy A, Eaton F, Mendiola Pla M, et al. TRIM35-mediated deg-radation of nuclear PKM2 destabilizes GATA4/6 and induces P53 in car-diomyocytes to promote heart failure. Sci Transl Med. 2022;14:eabm3565.doi: 10.1126/scitranslmed.abm3565 59. Paulin R, Dromparis P, Sutendra G, Gurtu V, Zervopoulos S, Bowers L,Haromy A, Webster L, Provencher S, Bonnet S, et al. Sirtuin 3 deficiencyis associated with inhibited mitochondrial function and pulmonary arterialhypertension in rodents and humans. Cell Metab. 2014;20:827–839. doi:10.1016/j.cmet.2014.08.011 60. Boukouris AE, Zhang Y, Saleme B, Kinnaird A, Zhao YY, Liu Y, ZervopoulosSD, Das SK, Mittal RD, Haromy A, et al. A reversible metabolic stress-sensitive regulation of CRMP2A orchestrates EMT/stemness and in-creases metastatic potential in cancer. Cell Rep. 2022;38:110511. doi:10.1016/j.celrep.2022.110511 61. Saleme B, Gurtu V, Zhang Y, Kinnaird A, Boukouris AE, Gopal K,Ussher JR, Sutendra G. Tissue-specific regulation of p53 by PKM2 isredox dependent and provides a therapeutic target for anthracycline-induced cardiotoxicity. Sci Transl Med. 2019;11:eaau8866. doi:10.1126/scitranslmed.aau8866