Unique mechanistic insights into the beneficial effectsof soluble epoxide hydrolase inhibitors in theprevention of cardiac fibrosisPadmini Sirisha,1, Ning Lia,1, Jun-Yan Liub, Kin Sing Stephen Leeb, Sung Hee Hwangb, Hong Qiua, Cuifen Zhaoc,Siu Mei Maa, Javier E. Lópeza, Bruce D. Hammockb, and Nipavan Chiamvimonvata,d,2

aDivision of Cardiovascular Medicine, University of California, Davis, CA 95616; cDepartment of Pediatrics, Qilu Hospital, Shandong University, Jinan 250012,China; dDepartment of Veterans Affairs, Northern California Health Care System, Mather, CA 95655; and bDepartment of Entomology and UCDComprehensive Cancer Research Center, University of California, Davis, CA 95616

Edited by Eric N. Olson, University of Texas Southwestern Medical Center, Dallas, TX, and approved January 29, 2013 (received for review December 31, 2012)

Tissue fibrosis represents one of the largest groups of diseases forwhich there are very few effective therapies. In the heart, myocar-dial infarction (MI) resulting in the loss of cardiac myocytes canculminate in adverse cardiac remodeling leading to eventual heartfailure. Adverse cardiac remodeling includes myocyte hypertrophy,fibrosis, and electrical remodeling.We have previously demonstratedthe beneficial effects of several potent soluble epoxide hydrolaseinhibitors (sEHIs) in different models of cardiac hypertrophy and fail-ure.Here,wedirectlydetermine themolecularmechanismsunderlyingthe beneficial effects of sEHIs in cardiac remodeling post-MI. Treat-ment with a potent sEHI, 1-trifluoromethoxyphenyl-3-(1-propionylpi-peridine-4-yl)urea (TPPU), which was started 1 wk post-MI in amurinemodel, results in a significant improvement in cardiac function.Importantly, treatment with TPPU results in a decrease in cardiacfibrosis as quantified using histological and immunostaining techni-ques. Moreover, single-cell–based assays demonstrate that treatmentwith TPPU results in a significant decrease not only in the percentagesbut also the proliferative capacity of different populations of cardiacfibroblasts as well as a reduction in the migration of fibroblasts intothe heart from the bone marrow. Our study provides evidence fora possible unique therapeutic strategy to reduce cardiac fibrosis andimprove cardiac function post-MI.

epoxyeicosatrienoic acids | dihydroxyeicosatrienoic acids |thymocyte differentiation antigen | Cluster of Differentiation 90 |fibroblast specific protein 1

Tissue fibrosis represents one of the largest groups of diseasesfor which there are very few effective therapies. In the heart,

adverse cardiac remodeling consists of cardiac myocyte hyper-trophy, fibrosis, and electrical perturbations and represents theprevailing response of the heart to extrinsic and intrinsic stimuliincluding myocardial infarction (MI) as well as pressure or vol-ume overload. Cardiac fibrosis is characterized by the accumu-lation of extracellular matrix in the myocardium and is associatedwith increased stiffness that contributes significantly to diastolicdysfunction (1, 2). Even though important progress has beenmade in the treatment of systolic dysfunction, at the presenttime, there is no effective therapy for diastolic dysfunction (3, 4).Tissue injury results in acute and robust inflammatory respon-

ses, involving the synthesis and release of chemokines and cyto-kines and the recruitment of leukocytes and fibroblasts. One ofthe important inflammatory responses involves the inflammatorylipid mediators with the activation of phospholipase A2 and therelease of arachidonic acid (AA). Eicosanoids are oxylipids thatare potent modulators of immune responses and are derived fromAA or similar fatty acids. AA is metabolized through three enzy-matic pathways, namely, cyclooxygenase (COX), lipoxygenase(LOX), and cytochrome P450 (CYP450) pathways. Whereas theCOX and LOX pathways have been studied in detail, underpinnedby the translation of inhibitors of these enzymatic pathways exem-plified by aspirin and zileuton in the treatment of inflammatorydiseases, the translational manipulation of the CYP450 pathway

remains unexplored mechanistically and underused clinically. Wehypothesize that theCYP450smay represent the last robust frontierin the inflammatory pathway that we may be able to manipulate toreduce inflammatory responses in ischemic cardiomyopathy andpressure overload hypertrophy.AA when metabolized through the CYP450 epoxygenase path-

way generates epoxyeicosatrienoic acids (EETs). EETs function asautocrine and paracrine effectors in the cardiovascular system (5, 6)and are shown to have cardioprotective properties. However, EETsare furthermetabolized by soluble epoxide hydrolase (sEH) to formthe corresponding dihydroxyeicosatrienoic acids (DHETs) (7).We have previously revealed the beneficial effects of several sEH

inhibitors (sEHIs) in the prevention of hypertrophy and electricalremodeling in cardiac myocytes from animal models of myocardialinfarction and pressure-overload hypertrophy (8–14). In the pres-ent study, we directly demonstrate that treatment with sEHIscan prevent cardiac fibrosis by inhibiting cardiac fibroblast prolif-eration. Single-cell–based phenotyping was used to quantify thepercentages and the proliferative response in different populationsof cardiac fibroblasts (CFs) in twomodels of cardiac injury, namelyischemic cardiomyopathy and pressure overload hypertrophy.Our study demonstrates the molecular mechanisms of selectiveand potent inhibitors of inflammation that reduce fibrosis andadverse cardiac remodeling. Treatment with sEHIs results in theimprovement of cardiac function by preventing the developmentof cardiac fibrosis.

Results1-Trifluoromethoxyphenyl-3-(1-Propionylpiperidine-4-yl)Urea (TPPU)Prevents the Development of Cardiac Hypertrophy in a Murine MIModel. After determining the potency, pharmacokinetics, andphysiochemical properties of 11 different sEHIs, TPPU (containinga piperidine ring; Fig. 1A) was found to have high inhibitory po-tency, drug-like physicochemical properties, pharmacokinetics withhigh area under the curve values, a relatively longer half-life, andlower plasma protein binding properties than many previous com-pounds (15).MI was generated in 8- to 10-wk-old male C57BL/6J mice

(Charles River) using previously described techniques (16). Oneweek after the surgery, mice were randomized to receive eitherdrinking water containing TPPU (Fig. 1A, 15 mg/L) or vehicle

Author contributions: P.S., N.L., J.E.L., B.D.H., and N.C. designed research; P.S., N.L., J.-Y.L.,K.S.S.L., S.H.H., H.Q., and S.M.M. performed research; J.-Y.L., K.S.S.L., S.H.H., and B.D.H.contributed new reagents/analytic tools; P.S., N.L., J.-Y.L., K.S.S.L., S.H.H., H.Q., C.Z., J.E.L.,and N.C. analyzed data; and P.S., B.D.H., and N.C. wrote the paper.

Conflict of interest statement: B.D.H. and N.C. have filed patents with the University ofCalifornia for sEH inhibitors and cardiac hypertrophy therapy.

This article is a PNAS Direct Submission.1P.S. and N.L. contributed equally to this work.2To whom correspondence should be addressed. E-mail: nchiamvimonvat@ucdavis.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1221972110/-/DCSupplemental.

5618–5623 | PNAS | April 2, 2013 | vol. 110 | no. 14 www.pnas.org/cgi/doi/10.1073/pnas.1221972110

alone for 3 wk (Fig. 1B) (17). Sham-operated mice were alsorandomized to receive either TPPU or vehicle alone at week onefor 3 wk. The investigators were blinded to the treatment groups.Whole hearts from the MI mice after 3 wk of follow up

exhibited cardiac dilatation, which was prevented in the MI ani-mals treated with TPPU (Fig. 1C). As expected, whole heartsfrom the sham-operated and sham-operated-TPPU–treatedgroups showed no significant hypertrophy or dilatation. Summarydata in Fig. 1D andE illustrate the significant increase in the heartweight and the ratio of heart weight/body weight in the MI groupcompared with sham-operated hearts. Treatment with TPPUresulted in a significant decrease in the heart weight and the heartweight/body weight ratio in the MI animals. There were no sig-nificant changes in the sham-operated mice treated with TPPU.

Treatment with TPPU Results in a Significant Improvement in CardiacFunction as Assessed by Echocardiography. The chamber size andsystolic function were assessed in the four groups of animals usingechocardiography. Two-dimensional and motion-mode (M-mode)echocardiography showed evidence of cardiac chamber dilatationin the MI mice that was prevented in TPPU-treated animals (Fig.

2A and Table S1). However, there were no significant differencesbetween the two sham-operated groups. Fig. 2B and Table S1summarize the percentages of the fractional shortening (FS) be-fore and 3 wk after treatment with TPPU. Indeed, treatment withTPPU in the MI mice resulted in a significant improvement in theFS compared with the MI alone. In contrast, there were no sig-nificant differences in FS between TPPU-treated sham-operatedmice compared with sham alone. Taken together, these data sug-gest that the treatment with TPPU prevented adverse cardiacremodeling and improved cardiac function in the MI model.

Beneficial Effect of TPPU Treatment on Cardiac Fibrosis in the InfarctZone. Here, we specifically sought to determine the effect of asEHI on cardiac fibrosis within the infarct zone as well as theremote zone. To this end, cardiac sections (100 μm) fromcorresponding areas in the four groups of animals were stainedusing Picrosirius Red to quantify the amount of collagen (18, 19).Histological analysis demonstrated that treatment with TPPUresulted in a marked decrease in the infarct size and preventedthe development of cardiac dilatation post-MI (Fig. 2C). Directquantification of the infarcted area from the four groups showed

Fig. 1. TPPU prevents the development of cardiac hypertro-phy in amurinemyocardial infarction (MI)model. (A) Structureof the sEHI, 1-trifluoromethoxyphenyl-3-(1-propionylpiper-idine-4-yl) urea (TPPU) used in our studies. (B) Schematic rep-resentation of the experimental protocol. (C) Examples ofwhole hearts from sham-operated, TPPU-treated–sham-oper-ated, MI, and TPPU-treated MI mice. Mice were killed after3 wk of follow-up. (Scale bar, 2 mm.) (D) Summary data forheart weight (in milligrams), and (E) heart weight/bodyweight ratio (percentage) from MI mice compared with un-treatedMI mice. Error bars represent SE, n = 12 per group, and§P < 0.05 by Student t test.

Fig. 2. Noninvasive echocardiographic assessment of the effectof TPPU on cardiac function and immunohistochemistry. (A)Examples of 2D and M-mode echocardiography in sham-oper-ated, MI, and MI treated with TPPU after 3 wk of treatmentshowing evidence of chamber dilatation in MI mice. TPPU pre-vented the development of chamber dilatation in MI mice. (B)Summary data for percentage of fractional shortening (FS). (C)Cardiac sections stained with Sirius Red to demonstrate theamount of collagen deposition. (D) Quantification of the per-centage of infarct zone. (E) Confocal images of wheat germagglutinin (WGA) stains showing a significant decrease in col-lagen deposition in the MI model treated with TPPU comparedwithMI alone. (Scale bar, 50mm.) Error bars represent SE, n = 12per group. §P < 0.05 by Student t test and *P < 0.05 by ANOVA.

Sirish et al. PNAS | April 2, 2013 | vol. 110 | no. 14 | 5619

MED

ICALSC

IENCE

S

a significant reduction in the amount of collagen in the treatedMI hearts compared with the MI alone (Fig. 2D). There were nosignificant differences between the two sham-operated groups.Examination of collagen in the fibrotic areas using wheat germagglutinin showed a marked decrease in the amount of collagendeposition in the treated MI hearts compared with the MI alone(Fig. 2E). The data suggest that treatment with TPPU post-MIprevents adverse cardiac remodeling at least in part by reducinginfarct size and cardiac fibrosis.

Effects of TPPU on Cardiac Fibrosis in the Remote Zone in the MIModel. Because the heart contains a mixture of heterogenouspopulations of cells, to directly quantify the percentages of CFs,single-cell–based assays using flow cytometric analyses wereperformed as we have previously described (20, 21). The nucle-ated cells from the fresh myocardial preparation were enumer-ated based on the incorporation of 7-aminoactinomycin D (7-AAD) (Fig. 3A). A cardiac-specific Troponin T (cTnT) antibodylabeled the myocytes (MCs), which show a higher fluorescencelevel and gate them separately from the nonmuscle cells (NMCs)to increase the cell specificity of this analysis. Controls werestained with isotype-matched IgG antibodies to set gates thatidentified positive cells.Two populations of CFs were identified in the remote zone

away from the infarct zone. CFs were defined by Thy1.2 (22) andfibroblast-specific protein 1 (FSP-1) expression (23–25) and the

lack of other lineage markers (Lin). Thy1 [thymocyte differenti-ation antigen or Cluster of Differentiation 90 (CD90)] is a smallglycoprotein localized at the surface of several cell types includingCF (22). Further characterization of Thy1.2+ cells using fluores-cence-activated cell sorting (FACS) and PCR revealed the ex-pression of collagen Ia and IIIa. The Thy1.2+ cells lacked theexpression of platelet endothelial cell adhesion molecule (PECAM)and Von Willebrand factor (vWF) for endothelial cells andNkx2.5 (Fig. S1). FSP-1, also known as S100A4 is a member ofthe S100 superfamily of EF-hand calcium-binding proteins, hasbeen shown to be specific for CFs (23–25). In addition, a popula-tion of CD34+CD45+ fibroblasts has previously been shown tobe derived from bone marrow and contributes to cardiac fibrosisin angiotensin II (AngII)-induced cardiac hypertrophy (18). Thispopulation was also analyzed separately in our study.For flow cytometric analysis, Thypos CFs were identified in our

study as Thy1.2+/Lin−/CD31−/CD34−/CD45− cells (22) (Fig. 3A).Flow cytometric analyses demonstrated that there was a significantincrease in Thypos cells in the remote area in MI mice comparedwith the two groups of sham animals (Fig. 3 B and C). Moreover,treatment with TPPU in the MI animals resulted in a significantdecrease in Thypos cells compared with MI alone.Next, Ki67, a nuclear antigen that is expressed in actively cy-

cling cells (26, 27), was used to directly test the hypothesis thatthere is a significant increase in the proliferative capacity amongCFs isolated from the MI model. There was a significant increasein the percentage of Ki67 in the Thypos population in MI micecompared with sham animals. Treatment with TPPU in the MIanimals resulted in a significant decrease in Ki67 positivity in Thypos

cells (Fig. 3D).Using FSP-1 as a second marker for CFs, we also documented

the same beneficial effects of TPPU on the percentages andproliferative capacity of FSP-1. Specifically, there was a signifi-cant increase in FSPpos cells (FSP-1+/Lin−/CD31−/CD34−/CD45−cells) in the remote area in MI mice compared with sham animals.Treatment with TPPU in MI animals resulted in a significant de-crease in the percentages and Ki67 positivity of FSPpos cells com-pared with MI alone (Fig. 3 E–G). Finally, treatment with TPPUin MI animals significantly decreased the percentages of CD34+CD45+ fibroblasts compared with MI alone (Fig. S2). Taken to-gether, our data suggest that treatment with sEHI prevents theproliferation of the resident CFs as well as the migration of CFsinto the heart post-MI, leading to a significant decrease in car-diac fibrosis both in the infarct zone and the remote area fromthe infarct.

Mechanistic Insights into the Effect of TPPU on the Activation of CFs.CFs have been shown to play critical roles in cardiac remodeling (1,23). In response to injury, CFs differentiate intomyofibroblasts thatexpress contractile proteins including α-smooth muscle actin andexhibit increased proliferative, migratory, and secretory properties.CFs respond to proinflammatory cytokines, vasoactive peptides(e.g., AngII and endothelin-1), and hormones (e.g., norepineph-rine). These factors play critical roles in cardiac fibrosis (1, 23–25,28). Chemokines such as monocyte chemoattractant protein-1(MCP-1) also play important roles in cardiac fibrosis (2, 29).Here, we demonstrate that treatment with TPPU resulted in

a significant decrease in inflammatory cytokines and chemokinesincluding interleukin-12 (IL-12), tumor necrosis factor-α (TNF-α),and MCP-1 (Fig. S3). The results are consistent with our previousfindings in an acute treatment in the MI model (12). Moreover,sEHIs have been shown to block AngII-induced cardiac hyper-trophy (10) and AngII is a well-characterized profibrotic moleculewhose downstream mediators include mitogen-activated proteinkinases (MAPKs) (1, 23). An unbiased approach of metabolicprofiling of oxylipids was also performed to document the targetengagement by TPPU, demonstrating significant increases in EETs/DHETs ratios in the sEHI-treated groups compared with no-treatment groups (Fig. S4).To investigate the mechanisms underlying the observed ben-

eficial effects of TPPU on cardiac fibrosis, the activation of the

Fig. 3. Flow cytometric analysis of cardiac fibroblasts from the remote zonein the in vivo MI model. (A) Selection of nucleated cells (Nucl cells) from themixed population based on the incorporation of 7-AAD and the separation ofthe myocytes (MCs) from the nonmuscle cells (NMCs) using cardiac troponinT(cTnT)-specific antibody. All of the analyses were conducted on the Lin−/CD31−/CD45−/CD34− cells (pink box). X and Y axes represent arbitrary units.(B) Flow cytometric analysis of Thy1.2+/Lin−/CD31−/CD45−/CD34− (Thypos) cellsfrom sham-operated, TPPU-treated–sham-operated, MI and TPPU-treated MImice. (C) Summary data from B (n = 3 per group). (D) Summary data showingthe proliferative Thypos cells using Ki67 proliferative marker. (E) Flow cyto-metric analysis of FSP+/Lin−/CD31−/CD45−/CD34− (FSPpos) cells from sham-operated, TPPU-treated–sham-operated, MI and TPPU-treated MI mice. (F)Summary data from E (n = 3 per group). (G) Summary data showing theproliferative FSPpos cells using Ki67 proliferative marker. Representativeresults are shown. Error bars represent SE and *P < 0.05.

5620 | www.pnas.org/cgi/doi/10.1073/pnas.1221972110 Sirish et al.

major downstream signaling molecules in the MAPK pathway, ex-tracellular signal-regulated kinases 1 and 2 (ERK1 and 2), wasexamined in the Thypos subpopulation of the CFs. Our analysisshowed a significant increase in the levels of phosphorylatedERK1/2 (pERK1/2) in the MI mice compared with the sham-operated animals (3 ± 0.4% and 1.1 ± 0.06%, respectively). Indeed,there was a significant decrease in phosphorylated ERK1/2 bytreatment with TPPU in the MI model (1.3 ± 0.3%, Fig. 4 A and B).Moreover, in vitro treatment of cultured mouse cardiac fibroblastswith AngII plus TPPU showed a significant decrease in the levelsof pERK1/2 compared with AngII treated Thypos CFs (Fig. S5).The Thypos subpopulation of the CFs was then sorted using

FACS from the four groups of animals (Fig. 4C). Consistently,Western blot analyses from the sorted cells showed an increasein pERK1/2 levels in MI mice compared with sham animals.Treatment with TPPU for 3 wk resulted in a significant decreasein the pERK1/2 toward the control level (Fig. 4 D and E, n = 3,P < 0.05). Our data suggest that sEHIs are beneficial in preventingadverse ventricular remodeling at least in part, by decreasing theproduction of proinflammatory cytokines and chemokines leadingto a reduction in the proliferation and activation of resident CFsand/or recruitment of fibroblast progenitors.

Treatment with TPPU in the MI Model Prevents Cardiac MyocyteHypertrophy as Assessed by Cell Volume and Hypertrophic Markers.To further elucidate how TPPU antagonizes the initiation ofhypertrophic response, we examined the effect of TPPU on thesize of the cardiac myocytes isolated from the remote zone usingmyocyte volume measurements by a Coulter multisizer assay (21).Our data show that the myocyte volume significantly increased inthe MI mice, which was restored to the sham-operated controllevels by the TPPU treatment (Fig. S6 A and B).Beta-myosin heavy chain (β-MyHC) is a well known myosin

fetal isoform that is reexpressed in the adult heart during path-ological states in minor subpopulations of myocytes (21). Weused the validated NOQ7.5.4D antibody to assess the effect ofTPPU on the reexpression of β-MyHC in single myocytes. Ourresults demonstrated that MI resulted in significant up-regulation

of percentages of myocytes expressing β-MyHC (Fig. S6 C andD). Treatment with TPPU in the MI animals resulted in a sig-nificant reduction in the percentages of myocytes expressingβ-MyHC to similar levels as the control groups.We also analyzed the induction of another hypertrophic

marker, atrial natriuretic factor (ANF) in the MI and MI-treatedanimals using whole-heart lysates by Western blot analysis. MIshowed an up-regulation of ANF, which was reduced with thetreatment of TPPU (Fig. S6 E and F).

TPPU Reduces Cardiac Fibrosis in a Murine Thoracic Aortic ConstrictionModel. To directly demonstrate that the observed beneficialeffects of TPPU in the prevention of cardiac fibrosis were modelindependent, a thoracic aortic constriction (TAC) model wasused, resulting in a chronic pressure overload-induced hypertro-phy. One week after the surgery, the TAC mice were randomizedto receive TPPU for 3 wk. The effect of TPPU on cardiac functionwas assessed by 2D andM-mode echocardiography. There was anincrease in chamber dilation associated with a significant decreasein FS in the TAC mice that was prevented in the TPPU-treatedgroup (Fig. 5 A and B). There were no significant differences inthe two sham-operated groups. Cardiac sections (100 μm) fromfour groups of animals stained using Picrosirius Red demonstratedthat treatment with sEHI resulted in a decrease in chamber fibrosisas assessed using Picrosirius Red (Fig. S7).The analysis of Thypos and FSPpos CFs isolated from the left

ventricular freewall and the septum showed a significant increase inThypos and FSPpos CFs in the TAC mice compared with the sham-operated mice. Treatment with TPPU in TAC animals resulted ina significant decrease in Thypos and FSPpos cells in the TAC mice(Fig. 5C,D, andF). The proliferating Thypos cells as assessed by thepresence of Ki67 showed a similar trend in the TAC-treated andthe MI-treated mice (Fig. 5E), suggesting a model independentaction of TPPU in the prevention of cardiac fibrosis.

DiscussionIn the present study, we provide mechanistic insights into thebeneficial effects of sEHIs on adverse cardiac remodeling in twoanimal models with distinct pathophysiology, namely, ischemiccardiomyopathy and pressure-overload hypertrophy. We dem-onstrate the beneficial effect of sEHI on the reduction in thepercentages and the proliferative capacity of the different pop-ulations of CFs resulting in a significant decrease in cardiac fi-brosis and adverse cardiac remodeling. Treatment with sEHIsignificantly decreases the systemic levels of inflammatory cyto-kines and chemokines. Moreover, treatment with sEHI results ina significant decline in the phosphorylation of one of the keysignaling molecules, ERK1/2, in the MAPK pathway in CFpopulation. Our data provide evidence that an increase in thebiological activities of EETs prevents adverse cardiac remodelingby reducing the percentages, proliferative capacity, and activa-tion of the different populations of CFs. Moreover, treatmentwith sEHI leads to a significant decrease in chemokines, result-ing in a decrease in the migration and recruitment of a pop-ulation of fibroblasts derived from circulating bone marrow cells.Taken together, treatment with sEHI not only prevents the de-velopment of pathologic cardiac myocyte hypertrophy, sEHI mayalso result in an improvement in diastolic dysfunction by de-creasing CF proliferation and cardiac fibrosis.

Effects of sEHI on the Subpopulations of Cardiac Fibroblasts. Fibroticscar tissue accumulation leads to adverse remodeling and a sig-nificant loss of cardiac function (2). In the pathologic myocar-dium, CFs are considered to be the most important contributorsof the collagen matrix deposition (23, 24). In the past, fibroblastswere defined based on morphological characteristics or collagensynthesis (24, 30). However, CFs represent a heterogeneouspopulation that are derived from various distinct tissue niches in-cluding resident fibroblasts, endothelial cells, bone marrow sources(23–25, 31), circulating progenitors, as well as progenitors residingwithin the vascular walls (32, 33). In addition, a truly definitive cell-

Fig. 4. Activation of CFs in cardiac fibrosis through the pERK1/2 pathway. (A)Flow cytometric analysis of pERK1/2+nonmuscle cells from sham-operated, TPPU-treated–sham-operated, MI and TPPU-treated MI mice. (B) Summary data fromA (n = 3 per group). (C) Florescent activated cell sorting (FACS) of Thy1.2+/Lin−/CD31−/CD45−/CD34− cells (Thypos). Representative results are shown. (D) Rep-resentative lanes of Western blot assay and (E) summary data for pERK1/2 andERK1/2 (loading control) performed from FACS sorted Thypos cells from sham-operated, TPPU-treated–sham-operated, MI and TPPU-treated MI mice (n = 3).Error bars represent SE and *P < 0.05.

Sirish et al. PNAS | April 2, 2013 | vol. 110 | no. 14 | 5621

MED

ICALSC

IENCE

S

specific marker has yet to be defined. The origin of proliferatingCFs in pathological conditions is not completely defined. It alsobecomes clear that a particular marker only labels a subset of CFs(23, 24, 31).To overcome some of these challenges, we have used single-

cell phenotyping to evaluate and quantify the percentages and theproliferative capacity of the different subpopulations of CFs. TheCFs were identified by a combination of markers. We demonstratethat treatment with sEHI prevents the proliferation and migrationof different populations of fibroblasts (Fig. 3 and Fig. S2) post-MI,leading to a significant decrease in cardiac fibrosis. However,additional markers may need to be explored to define subpopula-tions of CFs and their responses to treatment with sEHI.

Mechanistic Insights into the Signaling Pathways Involved in CardiacFibrosis.CFsmaintain the homeostasis of the extracellular matrixby modulating the expression of proteins such as collagens (I,III, IV, V, and VI) and matrix metalloproteinases (MMPs). CFsalso secrete growth factors and cytokines that exert autocrineand paracrine effects on cellular processses such as proliferationand apoptosis. However, under pathological conditions, in re-sponse to the increase in proinflammatory cytokines (e.g., TNF-α, IL-1, IL-6, and TGFβ), CFs exhibit up-regulated proliferation,migration, differentiation, secretion of collagens, cytokines, andactivation MMPs. All these factors contribute to the developmentof perivascular and interstitial fibrosis, which ultimately leads todiastolic and systolic ventricular dysfunction (1, 23–25, 28).In addition, studies have provided compelling evidence to

support the roles of chemokines in cardiac fibrosis (2, 29). Oneof the best-studied chemokines is MCP-1 (also known as CCL2).Effects of MCP-1 in cardiac fibrosis are highly complex including(i) recruitment and activation of mononuclear cell subsets andfibroblast progenitors, and (ii) possible direct effects on residentfibroblasts (2, 29).Our previous and current data support the notion that sEHIs

are potent anti-inflammatory agents that significantly decreasethe systemic levels of cytokines and chemokines (Fig. S3) (8–14,34, 35). Of considerable relevance are our findings that there isa significant decrease in several inflammatory cytokines andchemokines including TNF-α, IL-12, and MCP-1 levels by treat-ment with sEHIs. IL-6 has been shown to be involved in the in-creased proliferation of fibroblasts via the MAPK pathway (36).TNF-α and MCP-1 also contribute to cardiac fibrosis via the ex-tracellular signal regulated kinase (ERK) signaling cascade (2,37). Our flow cytometric and immunoblot data provide evidencethat sEHI results in a significant decrease in the activation of

ERK1/2 in CFs in the MI mice (Fig. 4). Taken together, our datasuggest that treatment with sEHI decreases CF proliferation atleast in part, via the inhibitory effects on the MAPK pathway.

Future Directions. CFs are pleiomorphic and pleiotropic cells thatcan respond to multiple profibrotic factors that are highly com-plex in nature, many of which exert synergistic effects with oneanother with cross-talk at multiple levels. AngII is a well-char-acterized profibrotic molecule. Prominent downstream media-tors from AngII include TGFβ and MAPKs (1, 23). AngII up-regulates the three inactive isoforms of TGFβ (TGFβ1, TGFβ2,and TGFβ3), which are complexed with latent TGFβ bindingproteins (LTBPs). Upon activation, the proteolytic cleavage ofLTBP dissociates active TGF by molecules such as thrombo-spondin 1 (tsp-1), plasmin proteases, and integrin αvβ6 (38, 39).AngII induces another profibrotic factor, connective tissue

growth factor (CTGF), which promotes cardiac fibrosis throughthe protein kinase-C pathway. CTGF has also been shown topromote fibroblast proliferation (38).Cardiac fibrosis also triggers endothelial–mesenchymal tran-

sition (EndMT), which contributes to about 30% of the activatedfibroblasts. TGFβ1 has been shown to induce endothelial cells toundergo EndMT to acquire a more fibroblast-like phenotype andenter the interstitium where they contribute to cardiac fibrosis(25). Another downstream mediator of TGFβ and AngII in CFsis endothelin 1 (ET-1), which acts as a growth factor to inducecardiac fibroblast proliferation and increase collagen synthesisvia the activation of cell surface ETA and ETB receptors (1, 39).ET-1 can also produce a profibrotic effect by reducing the col-lagenase activity of MMPs via the ETA receptors (1). It has alsobeen shown that ET-1 can induce CTGF expression (39).Here, we have demonstrated that treatment with TPPU sig-

nificantly decreases the activation of the MAPK pathway in CFs.However, CFs contribute to cardiac fibrosis through severalsynergistic factors including AngII, ERK, TGFβ, ET-1, andCTFG. Hence, other signaling cascades have to be investigatedto test whether these factors are also down-regulated by sEHIs.

Materials and MethodsA detailed description of the methods is provided in SI Methods.

MI Model in Mice. The MI model in mice was created using procedure aspreviously described (16). Echocardiograms were performed 1 wk aftersurgery after which mice were randomized to receive TPPU (15 mg/L) (17) inthe drinking water or water alone for a period of 3 wk (Fig. 1B), at whichtime repeat echocardiograms were performed.

Fig. 5. Beneficial effect of TPPU on cardiac functionin TAC animals. (A) Examples of 2D and M-modeechocardiography in mouse models with sham op-eration, TAC, and TAC treated with TPPU after 3 wkof treatment showing evidence of cardiac failurewith chamber dilatation in TAC mice. (B) Summarydata for percentage of fractional shortening (FS). (C)Flow cytometric analysis of Thy1.2+/Lin−/CD31−/CD45−/CD34− (Thypos) cells from sham-operated,TPPU-treated–sham-operated, TAC and TPPU-trea-ted TAC mice. (D) Summary data from C (n = 3 pergroup). (E) Summary data showing the proliferativeThypos cells using Ki67 proliferative marker. (F)Summary data of FSP+/Lin−/CD31−/CD45−/CD34−

(FSPpos) cells from sham-operated, TPPU-treated–sham-operated, TAC and TPPU-treated TAC mice(n = 3 per group). Representative results are shown.Error bars represent SE and *P < 0.05.

5622 | www.pnas.org/cgi/doi/10.1073/pnas.1221972110 Sirish et al.

Metabolomic Profiling of Oxylipins. Oxylipin profiling was performed usinga modification of a previously published method (12, 40).

Measurement of Plasma Cytokine Levels. Plasma cytokine levels were analyzedusing a Cytometric Bead Array kit (BD Biosciences) (12).

Flow Cytometric Analysis of Mouse Fibroblasts. Isolated cells were fixed andtreated with anti-cTnT, anti-CD31, anti-CD45, anti-CD34, anti-Thy1.2, anti-FSP-1, anti-Ki67, anti–β-MyHC, anti-pERK1/2, and anti-ERK1/2 antibodies.

ACKNOWLEDGMENTS. This work was supported by National Institutes ofHealth (NIH) HL85727 and HL85844 and Veterans Administration MeritReview Grant I01 BX000576 (to N.C.); Howard Hughes Medical Institute

Med-into-Grad Training Program to University of California Davis (UCD)(P.S.); American Heart Association (AHA) Western States Affiliate Predoc-toral Fellowship Award (to P.S.); P.S. is supported by NIH T32 TrainingGrant in Basic and Translational Cardiovascular Science (T32 HL86350).Fellow-to-Faculty Award from Sarnoff Cardiovascular Research Founda-tion (to J.E.L.); an AHA Western States Affiliate Beginning Grant-in-Aid(to J.E.L.); and Harold Amos Medical Faculty Development Award fromRobert Wood Johnson Foundation (to J.E.L.). Partial supports were pro-vided by the National Institute of Environmental Health Sciences (NIEHS)Grant R37 ES02710, the NIEHS Superfund Basic Research Program (P42ES04699), the NIEHS Center for Children’s Environmental Health and DiseasePrevention (P01 ES11269), and a Technology Translational Grant from UCDHealth System. B.D.H. is a George and Judy Marcus Senior Fellow of theAmerican Asthma Society.

1. Porter KE, Turner NA (2009) Cardiac fibroblasts: At the heart of myocardial remod-eling. Pharmacol Ther 123(2):255–278.

2. Dobaczewski M, Frangogiannis NG (2009) Chemokines and cardiac fibrosis. FrontBiosci (Schol Ed) 1:391–405.

3. Kass DA, Bronzwaer JG, Paulus WJ (2004) What mechanisms underlie diastolic dys-function in heart failure? Circ Res 94(12):1533–1542.

4. Ouzounian M, Lee DS, Liu PP (2008) Diastolic heart failure: Mechanisms and contro-versies. Nat Clin Pract Cardiovasc Med 5(7):375–386.

5. Roman RJ, Maier KG, Sun CW, Harder DR, Alonso-Galicia M (2000) Renal and car-diovascular actions of 20-hydroxyeicosatetraenoic acid and epoxyeicosatrienoic acids.Clin Exp Pharmacol Physiol 27(11):855–865.

6. Schmelzer KR, et al. (2005) Soluble epoxide hydrolase is a therapeutic target for acuteinflammation. Proc Natl Acad Sci USA 102(28):9772–9777.

7. Yu Z, et al. (2000) Soluble epoxide hydrolase regulates hydrolysis of vasoactive ep-oxyeicosatrienoic acids. Circ Res 87(11):992–998.

8. Xu D, et al. (2006) Prevention and reversal of cardiac hypertrophy by soluble epoxidehydrolase inhibitors. Proc Natl Acad Sci USA 103(49):18733–18738.

9. Chiamvimonvat N, Ho CM, Tsai HJ, Hammock BD (2007) The soluble epoxide hydrolaseas a pharmaceutical target for hypertension. J Cardiovasc Pharmacol 50(3):225–237.

10. Ai D, et al. (2009) Soluble epoxide hydrolase plays an essential role in angiotensinII-induced cardiac hypertrophy. Proc Natl Acad Sci USA 106(2):564–569.

11. Harris TR, Li N, Chiamvimonvat N, Hammock BD (2008) The potential of soluble ep-oxide hydrolase inhibition in the treatment of cardiac hypertrophy. Congest HeartFail 14(4):219–224.

12. Li N, et al. (2009) Beneficial effects of soluble epoxide hydrolase inhibitors in myo-cardial infarction model: Insight gained using metabolomic approaches. J Mol CellCardiol 47(6):835–845.

13. Li N, et al. (2011) Use of metabolomic profiling in the study of arachidonic acid me-tabolism in cardiovascular disease. Congest Heart Fail 17(1):42–46.

14. Qiu H, et al. (2011) Soluble epoxide hydrolase inhibitors and heart failure. CardiovascTher 29(2):99–111.

15. Ulu A, et al. (2012) Pharmacokinetics and in vivo potency of soluble epoxide hydrolaseinhibitors in cynomolgus monkeys. Br J Pharmacol 165(5):1401–1412.

16. Tarnavski O, et al. (2004) Mouse cardiac surgery: Comprehensive techniques for thegeneration of mouse models of human diseases and their application for genomicstudies. Physiol Genomics 16(3):349–360.

17. Hwang SH, Tsai HJ, Liu JY, Morisseau C, Hammock BD (2007) Orally bioavailable po-tent soluble epoxide hydrolase inhibitors. J Med Chem 50(16):3825–3840.

18. Haudek SB, et al. (2010) Monocytic fibroblast precursors mediate fibrosis in angio-tensin-II-induced cardiac hypertrophy. J Mol Cell Cardiol 49(3):499–507.

19. Matkovich SJ, et al. (2010) MicroRNA-133a protects against myocardial fibrosis andmodulates electrical repolarization without affecting hypertrophy in pressure-overloaded adult hearts. Circ Res 106(1):166–175.

20. Anversa P, et al. (2005) Myocardial aging—a stem cell problem. Basic Res Cardiol100(6):482–493.

21. López JE, et al. (2011) β-myosin heavy chain is induced by pressure overload in a minorsubpopulation of smaller mouse cardiac myocytes. Circ Res 109(6):629–638.

22. Hudon-David F, Bouzeghrane F, Couture P, Thibault G (2007) Thy-1 expression bycardiac fibroblasts: Lack of association with myofibroblast contractile markers. J MolCell Cardiol 42(5):991–1000.

23. Krenning G, Zeisberg EM, Kalluri R (2010) The origin of fibroblasts and mechanism ofcardiac fibrosis. J Cell Physiol 225(3):631–637.

24. Zeisberg EM, Kalluri R (2010) Origins of cardiac fibroblasts. Circ Res 107(11):1304–1312.

25. Zeisberg EM, et al. (2007) Endothelial-to-mesenchymal transition contributes to car-diac fibrosis. Nat Med 13(8):952–961.

26. Epting CL, et al. (2008) Stem cell antigen-1 regulates the tempo of muscle repairthrough effects on proliferation of alpha7 integrin-expressing myoblasts. Exp Cell Res314(5):1125–1135.

27. Landberg G, Tan EM, Roos G (1990) Flow cytometric multiparameter analysis ofproliferating cell nuclear antigen/cyclin and Ki-67 antigen: A new view of the cellcycle. Exp Cell Res 187(1):111–118.

28. Leask A (2010) Potential therapeutic targets for cardiac fibrosis: TGFbeta, angioten-sin, endothelin, CCN2, and PDGF, partners in fibroblast activation. Circ Res 106(11):1675–1680.

29. Dobaczewski M, Frangogiannis NG (2008) Chemokines in myocardial infarction:Translating basic research into clinical medicine. Future Cardiol 4(4):347–351.

30. Kompa AR, et al. (2012) Soluble epoxide hydrolase inhibition exerts beneficial anti-re-modeling actions post-myocardial infarction. Int J Cardiol, 10.1016/j.ijcard.2011.12.062.

31. Souders CA, Bowers SL, Baudino TA (2009) Cardiac fibroblast: The renaissance cell.Circ Res 105(12):1164–1176.

32. Camelliti P, Borg TK, Kohl P (2005) Structural and functional characterisation of car-diac fibroblasts. Cardiovasc Res 65(1):40–51.

33. Sartore S, et al. (2001) Contribution of adventitial fibroblasts to neointima formationand vascular remodeling: From innocent bystander to active participant. Circ Res89(12):1111–1121.

34. Liu JY, et al. (2010) Metabolic profiling of murine plasma reveals an unexpectedbiomarker in rofecoxib-mediated cardiovascular events. Proc Natl Acad Sci USA107(39):17017–17022.

35. Liu JY, et al. (2010) Inhibition of soluble epoxide hydrolase enhances the anti-in-flammatory effects of aspirin and 5-lipoxygenase activation protein inhibitor ina murine model. Biochem Pharmacol 79(6):880–887.

36. Leicht M, Briest W, Zimmer HG (2003) Regulation of norepinephrine-induced pro-liferation in cardiac fibroblasts by interleukin-6 and p42/p44 mitogen activated pro-tein kinase. Mol Cell Biochem 243(1-2):65–72.

37. Westermann D, et al. (2007) Tumor necrosis factor-alpha antagonism protects frommyocardial inflammation and fibrosis in experimental diabetic cardiomyopathy. BasicRes Cardiol 102(6):500–507.

38. Leask A, Abraham DJ (2004) TGF-beta signaling and the fibrotic response. FASEB J18(7):816–827.

39. Leask A (2007) TGFbeta, cardiac fibroblasts, and the fibrotic response. Cardiovasc Res74(2):207–212.

40. Morisseau C, et al. (2002) Structural refinement of inhibitors of urea-based solubleepoxide hydrolases. Biochem Pharmacol 63(9):1599–1608.

Sirish et al. PNAS | April 2, 2013 | vol. 110 | no. 14 | 5623

MED

ICALSC

IENCE

S

Supporting InformationSirish et al. 10.1073/pnas.1221972110SI Materials and MethodsThe investigation conforms to the Guide for the Care and Use ofLaboratory Animals published by the National Institutes of Healthand was approved by the University of California, Davis In-stitutional Animal Care and Use Committee.

Soluble Epoxide Hydrolase Inhibitor. The soluble epoxide hydrolaseinhibitor (sEHI), 1-trifluoromethoxyphenyl-3-(1-propionylpiper-idine-4-yl)urea (TPPU) was used in the study (Fig. 1A). Thesynthesis, physical properties, and spectral characteristics ofTPPU were performed as previously described (1, 2).

Myocardial Infarction Model in Mice. All animal care and proce-dures were approved by the University of California, Davis In-stitutional Animal Care and Use Committee. The myocardialinfarction (MI) model in mice was created using procedures aspreviously described (3). Briefly, 10-wk-old male C57BL/6J mice(Charles River) were anesthetized with i.p. ketamine 80 mg/kgand xylazine 6 mg/kg. Intubation was performed perorally andmechanical ventilation was initiated. An oblique 4-mm incisionwas made 4 mm away from the left sternal border in the third–fourth intercostal space. The chest retractor was inserted and theheart, which was partially covered by the lung, was then visual-ized. The pericardium was gently picked up and pulled apart.The left anterior decending (LAD) coronary artery was thenvisualized and ligated 1–2 mm below the tip of the left auricle inits normal position. Occlusion was confirmed by the change ofcolor of the anterior wall of the left ventricle (LV) and main-tained for 45 min after which the occlusion was released. Thesham-operated mice underwent the same procedure without ty-ing the suture but moving it behind the LAD artery.Echocardiograms were performed 1 wk after surgery after which

micewere randomized to receiveTPPU(15mg/L) (2) in thedrinkingwater or water alone for a period of 3 wk (Fig. 1B). Four groups ofanimals including sham, sham treated with TPPU, MI, and MItreated with TPPUwere followed for a period of 3 wk at which timerepeat echocardiograms were performed. The investigators wereblinded to the treatment groups. A total of 104male C57BL/6Jmicewere used. Eight animals died in the perioperative period. Sevenanimals were excluded from the study because of failure of thecoronary artery occlusion, leaving a total of 89 mice in the study.

Thoracic Aortic Constriction Model in Mice. Thoracic aortic con-strictions (TACs) were preformed in 8-wk-old male C57BL/6J(Charles River) mice as previously described (1). Briefly, animalswere anesthetized with i.p. ketamine 50 mg/kg and xylazine2.5 mg/kg. Intubation was performed perorally and mechanicalventilation was initiated. Aortic constriction was created viaa left thoracotomy by placing a ligature securely around theascending aorta and a 26-gauge needle and then removing theneedle. The chest was closed with 3-0 dexon rib sutures, 5-0 dexon IImuscle sutures, and buried skin sutures. Negative plural pressure wasreestablished via a temporary chest tube until spontaneous breathingoccurs. Sham-operated animals underwent the same procedurewithout tying the suture. Echocardiograms were performed 1 wkafter surgery after whichmice were randomized to receive TPPU (15mg/L) (2) in the drinking water or water alone for a period of 3 wk.Four groups of animals including sham, sham treated with TPPU,MI, andMI treated with TPPUwere followed for a period of 3 wk atwhich time repeat echocardiograms were performed.

Analysis of Cardiac Function by Echocardiography. Echocardiogramsto assess systolic function were performed by using motion-mode(M-mode) and 2D measurements as described previously (4). Themeasurements represented the average of six selected cardiac cyclesfrom at least two separate scans performed in random-blind fashionwith papillarymuscles used as a point of reference for consistency inthe level of scan. Fractional shortening (FS), a surrogate of systolicfunction, was calculated from left ventricle dimensions as follows:FS= ((EDD-ESD)/EDD)× 100%, whereEDDandESD representend-diastolic and end-systolic dimensions, respectively.

Histological Analyses. Hearts were excised and retrogradely per-fusedwith phosphate-buffered solution towash out blood andfixedin 10% (vol/vol) formalin overnight. Hearts were then embeddedin paraffin, serial cardiac sections of 100 μm in thickness weretaken along the longitudinal axis, and stained with Picrosirius Redto assess for collagen content. The percentage of infarcted arearepresents the ratio of connective tissues to total left ventricle(LV) area and was calculated by computerized planimetry (Na-tional Institutes of Health Image J software) (5). An observerblinded to the treatment groups performed measurements.

Immunofluorescence Confocal Laser Scanning Microscopy. Addi-tional cardiac sections were stained with wheat germ agglutinin.The cardiac sections were deparaffinized with xylene before re-hydrating with serial dilution of ethanol. The sections wereblocked with donkey serum and stained with wheat germ agglutinin(10 μg·mL−1; Molecular Probes). Secondary antibodies conjugatedto Alexa Fluor 488 were used. Cardiac sections from correspondingarea from the four groups were scanned. Identical settings wereused for all of the specimens.FAC sorted Lin−/Thy1.2+ cells were plated on tissue culture

dishes, fixed with 1% paraformaldehyde (PFA), and stained withanticollagen1a and antiprocollagen3a antibodies (Santa CruzBiotechnology). Immunofluorescence-labeled and differentialinterference contrast (DIC) images were obtained using a ZeissLSM700 confocal laser-scanning microscope.

Metabolomic Profiling of Oxylipins. Plasma samples stored at−80 °Cwere thawed at room temperature. Aliquots of plasma (200 μL)were spiked with a set of odd chain length analogs and deuteratedisomers of several target analytes including hydroxyeicosatetraenoicacids (HETEs), prostaglandins, thromboxanes, epoxyoctadecenoicacids and epoxyeicosatrienoic acids (EpOMEs and EETs), anddihydroxyoctadecenoic acids and dihydroxyeicosatrienoic acids(DHOMEs and DHETs) contained in 10 μL of methanol, and thenwere extracted by solid phase extraction usingOasis HLB cartridges(Waters). The HLB columns (1 cc, 60 mg) were washed with 2 mLmethanol and preconditionedwith 2mLwater/methanol/acetic acid[95/5/0.1 (vol/vol)]. Samples were then mixed with 200 μL of thepreconditioning solution and loaded onto the column. The loadedcolumn was then washed with 2 mL of the preconditioning solutionand then dried for 5 min in vacuo. Target analytes were then elutedwith 2mLof ethyl acetate. The collected eluents were evaporated todryness using a centrifugal vacuum concentrator and redissolved in40 μL of methanol. The spiked samples were vortexed for 1 min,centrifuged at 10,956 × g for 5 min, and then transferred to ana-lytical vials containing 150 μL inserts for analysis.The oxylipin profiling was performed using a modification of

a previously published method (6). The separation of plasma oxy-lipins was conducted in a Shimadzu LC-10ADVP instrument (Shi-madzu) equipped with a 2.1 mm × 150 mm Pursuit XRs-C18 5-μm

Sirish et al. www.pnas.org/cgi/content/short/1221972110 1 of 6

column (Varian) held at 40 °C. A gradient of water containing 0.1%acetic acid [(vol/vol), solvent A] and acetonitrile/methanol/aceticacid [800/150/1 (vol/vol); solventB]was used to elute the columnwiththe flow rate of 0.4 mL·min−1 (Table S1). The injection volume was10 μL and the samples were kept at 10 °C in the auto sampler.Analytes were detected on a 4000 QTRAP (Applied Biosystems)hybrid, triple-quadrupole, and linear ion trap mass spectrometerequipped with a Turbo V ion source and operated in negativemultiple reaction monitoring mode. The source was operated innegative electrospray mode and the QTRAP was set as follows:CUR= 20 psi, GS1 = 50 psi, GS2 = 30 psi, IS = −4,500 V, CAD =high, TEM= 400 °C, ihe= on,DP=−60V, where CUR,GS1,GS2,IS, CAD,TEM, ihe, andDP refer to CurtainGas, Ion SourceGas 1,Ion Source Gas 2, IonSpray Voltage, CADGas, Temperature, ihe:1=ONand 0=OFF, andDeclustering Potential, respectively. Thecollision energies used for CAD ranged from −18 to −38 eV.

Measurement of Plasma Cytokine Levels. Plasma samples were col-lected 3 wk after sham or MI operation and stored at −70 °C untilassayed. Plasma cytokine levels were analyzed using a CytometricBead Array kit (CBA mouse inflammation kit; BD Biosciences).Briefly, thawed plasma samples were mixed for 2 h at room temper-ature with florescence-labeled capture beads with the PE detectionreagents to measure the concentrations of interleukin-6 (IL-6), in-terleukine-1β (IL-1β), interleukin-10 (IL-10), monocyte chemo-attractant protein-1 (MCP-1), tumor necrosis factor-α (TNF-α) andinterleukin-12p70 (IL-12 p70). Samples were then washed withwashing buffer and analyzed on a FACScan flow cytometer (BD Im-munocytometry Systems). Data were analyzed using BD CytometricBead Array Analysis software (BD Immunocytometry Systems).

Flow Cytometric Analysis of Nonmuscle Cells. Single-cell suspensionwas obtained from 8- to 12-wk-old male C57BL/6 mice as previouslydescribed (7). The procedure was performed according to the ap-provedUniversity ofCaliforniaDavisAnimalCareandUseprotocol.Briefly,micewere injectedwith 0.1mLheparin (1,000 units·mL−1) 10min before heart excision and then anesthetized with pentobarbitalintraperitoneally (80 mg·kg−1). Hearts were removed and placed inTyrode’s solution (mmol·L−1: NaCl 140, KCl 5.4, MgCl2 1.2, N-2-hydroxyethylpiperazine-N′-2-ethanesulphonic acid (Hepes) 5 andglucose 5, pH 7.4). All chemicals were obtained from Sigma Chem-icals unless stated otherwise. The aorta was cannulated under a dis-sectingmicroscope andmounted on the Langendorff apparatus. Thecoronary arteries were retrogradely perfused with Tyrode’s solutiongassed with O2 at 37 °C for 3 min at a flow rate of ∼3 mL·min−1. Thesolution was switched to Tyrode’s solution containing collagenasetype 2 (1mg·mL−1, 330 units·mg−1;WorthingtonBiochemical). After∼12 min of enzyme perfusion, hearts were removed from the perfu-sion apparatus and gently teased in high-K+ solution (mmol·L−1:potassiumglutamate120,KCl20,MgCl2 1,EGTA0.3, glucose10andHepes 10, pH 7.4 with KOH). The cells were filtered through a 200-μmcell strainer, resuspended inCa2+andMg2+ freePBS, treatedwithphytoerythrin-conjugated anti-Thy1.2 (BD Bioscience), anti–FSP-1(Millipore), lineage antibodymixture (CD3e,CD11b,CD45R,Ly-6C(lymphocyte antigen), Ly-6G, and Ly76, 1:100 dilution; BD Bio-science), anti-CD34, anti-CD45 (BD Bioscience), anti-troponin Tantibody (Thermo Scientific), anti-CD31 (BD Bioscience), AlexaFluor 488 conjugated anti-phosphorylated extracellular signal-regu-lated kinase 1/2 (pERK1/2) (Cell Signaling), anti-extracellular signal-regulated kinase 1/2 (ERK1/2) (Cell Signaling), anti–beta-myosinheavy chain (-β-MyHC) (clone NOQ7.5.4D; Sigma), and pro-liferation-specificKi67 antibody (15μg·mL−1; BDBioscience) inPBSwith 5% donkey serum and 20 μg·mL−1 DNase-free RNase (Sigma)overnight at 4 °C (8). Cells were also stained with 40 μg·mL−1

7-amino-actinomycin D (7AAD; BD Bioscience) to measure theDNA content. Data were collected using a standard FACScan cy-tometer (BD Biosciences) upgraded to a dual laser system with theaddition of a blue laser (15mWat 488 nm) and a red laser (25mWat

637 nm; Cytek Development) or Becton Dickinson LSR-II Flow cy-tometer.Datawere acquired usingCellQuest andDIVA6.2 software(BD Bioscience). Cells stained with isotype-matched IgG antibodieswere used as controls to determine the positive cell population. Datawere analyzed using FlowJo software (version 9.4 Treestar).

In Vitro Cardiac Fibroblast Culture. After obtaining the single cellsuspension as described above, the cells were plated on a 100-mmculture dish, incubated at 37 °C for 45 min. The culture dish waswashed two times withmedium and cultured until the cells reachedconfluency. Cells were treated with angiotensin II (AngII) (1 μM)alone or with TPPU (1 μM) for 30 min before flow cytometricanalysis of Thypos (Thy1.2+/Lin−/CD31−/CD34−/CD45−) cells.

Western Blot Analysis. Immunoblots were performed as previouslydescribed (9). The following primary antibodies were used: (i)Rabbit monoclonal antiphospho-p44/42 mitogen-activated proteinkinase (MAPK) antibody (1:2,000 dilution; Cell Signaling), (ii)rabbit monoclonal anti-p44/42 MAPK antibody (1:1,000 dilution;Cell Signaling), (iii) polyclonal antiatrial natriuretic polypeptide(1:2,000; Millipore), and (iv) monoclonal anti-glyceraldehyde 3-phosphate dehydrogenase (GAPDH) antibody (1:10,000 dilution;Fitzgerald Industries) was used as an internal loading control.

Myocyte Volume Measurements by Coulter Multisizer. The isolatedcells were fixed with 0.4% (wt/vol) PFA for 10 min and wereanalyzed using Coulter Multisizer 4 as previously described (10).At least 10,000 myocytes were assayed and the mean volumedistribution was calculated from the distribution curves gener-ated by the Coulter Multisizer 4 software.

Semiquantitative RT-PCR Analysis. Total RNA was isolated fromFAC-sorted Lin− (using lineage antibodies mixture directed againstCD3e, CD11b, Cd45R, Ly-6C, Ly-6G, andLy76)/Thy1.2+ cells usingRNA isolation kit (Qiagen) according to the manufacturer’s pro-tocol. A total of 1 mg of total RNA was reverse transcribed withSuperScript III reverse transcriptase (Invitrogen) with Oligo dTprimers (Invitrogen) according to the manufacturer’s protocol.cDNA was diluted with water (1:1), used for PCR amplification for35 cycles with gene-specific primers and separated on a 1.5% aga-rose gel. Photography of the gel and quantification of the intensity ofthe ethidium bromide-stained bands were performed by using theFluorchem 8000 imaging system (Alpha Innotech). GAPDH wasused as an internal standard to compare the expression levels of thetested genes using the following primers.

Statistical Analysis. Statistical comparisons were analyzed byStudent t tests or one-way ANOVA followed by Bonferroni testsfor post hoc comparison. Statistical significance was consideredto be achieved when P < 0.05.

Gene Primers

Collagen 1a F: CCCACCCCAGCCGCAAAGAG

R: GGGGCCAGGCACGGAAACTC

Collagen 3a F: TGCAGGGTCCCCTGGCTCAA

R: GGAACCAGCTTCGCCCCGTT

Nkx2.5 transcriptionfactor

F: GAGCCTACGGTGACCCTGACCCAG

R: TGACCTGCGTGGACGTGAGCTTCA

Von Willebrand factor F: GCCTGGGGCCTCCAAAGCAG

R: CCCGTGCACACACAAGGGCA

Platelet endothelialcell adhesion molecule

F: AGCCTCACCAAGCTCTGGGAAC

R: TGGGCCTTCGGCATGGAACG

GAPDH F: ACCACAGTCCATGCCATCAC

R: TCCACCACCCTGTTGCTGTA

Sirish et al. www.pnas.org/cgi/content/short/1221972110 2 of 6

1. Xu D, et al. (2006) Prevention and reversal of cardiac hypertrophy by soluble epoxidehydrolase inhibitors. Proc Natl Acad Sci USA 103(49):18733–18738.

2. Hwang SH, Tsai HJ, Liu JY, Morisseau C, Hammock BD (2007) Orally bioavailable potentsoluble epoxide hydrolase inhibitors. J Med Chem 50(16):3825–3840.

3. Tarnavski O, et al. (2004) Mouse cardiac surgery: Comprehensive techniques for thegeneration of mouse models of human diseases and their application for genomicstudies. Physiol Genomics 16(3):349–360.

4. Li N, et al. (2009) Beneficial effects of soluble epoxide hydrolase inhibitors inmyocardial infarction model: Insight gained using metabolomic approaches. J Mol CellCardiol 47(6):835–845.

5. Dawn B, et al. (2006) Postinfarct cytokine therapy regenerates cardiac tissue andimproves left ventricular function. Circ Res 98(8):1098–1105.

6. Morisseau C, et al. (2002) Structural refinement of inhibitors of urea-based solubleepoxide hydrolases. Biochem Pharmacol 63(9):1599–1608.

7. Sirish P, et al. (2012) MicroRNA profiling predicts a variance in the proliferativepotential of cardiac progenitor cells derived from neonatal and adult murine hearts.J Mol Cell Cardiol 52(1):264–272.

8. López JE, et al. (2011) β-myosin heavy chain is induced by pressure overload in a minorsubpopulation of smaller mouse cardiac myocytes. Circ Res 109(6):629–638.

9. Xu Y, et al. (2005) The effects of intracellular Ca2+ on cardiac K+ channel expressionand activity: novel insights from genetically altered mice. J Physiol 562(Pt 3):745–758.

Fig. S1. Characterization of fluorescence-activated cell sorting of Thy1.2+ cells. (A) Semiquantitation RT-PCR analysis. Lane 1, positive control; lane 2, negativecontrol (HL-1 cells and brain); lane 3, Thy1.2+ cells. Col1a, collagen 1a; Col3a, collagen 3a; and vWF, von Willebrand factor. (B) Photomicrographs of FAC-sortedThy1.2+ cells at 10× and 20× magnification. Below, immunofluorescence confocal images of Thy1.2+ cells showing collagen 3a (red) and procollagen 1a (green)staining.

Fig. S2. (A) Flow cytometric analysis of CD34+CD45+ single and double positive cells from the four groups of animals including sham ± TPPU, and MI ± TPPU.(B) Summary data from A (n = 3 per group). (C) Summary data showing the proliferative CD34+CD45+ cells using Ki67 proliferative marker. Representativeresults are shown. Data shown represent mean ± SE and *P < 0.05.

Sirish et al. www.pnas.org/cgi/content/short/1221972110 3 of 6

Fig. S3. Serum concentration (picograms per milliliter) of cytokines from the four groups of animals including sham ± TPPU, and MI ± TPPU. Data shownrepresent mean ± SE and *P < 0.05.

Fig. S4. Oxylipin profiling from the four groups of animals including sham ± TPPU, and MI ± TPPU at 3 wk of follow-up. Data shown represent mean ± SE and*P < 0.05 comparing MI + TPPU from sham ± TPPU or MI groups, §P < 0.05 comparing sham + TPPU from sham or MI ± TPPU groups.

Sirish et al. www.pnas.org/cgi/content/short/1221972110 4 of 6

Fig. S5. (A) Flow cytometric analysis showing pERK1/2 signal in cultured cardiac mouse Thypos fibroblasts treated with angiotensin II (ANGII, 1 μM, green)alone and with TPPU (1 μM, purple) for 30 min. (B) Summary data from A showing a significant increase in pERK1/2 levels in fibroblasts treated with AngII(green) compared with control (yellow) and a significant decrease in the pERK1/2 levels in the TPPU-treated cells (purple).

Fig. S6. TPPU in the MI model prevents cardiac myocyte hypertrophy as assessed by cell volume and hypertrophic markers. (A) Graph showing the myocytevolume measurements by the Coulter Multisizer assay. Green curve (sham operated), pink curve (TPPU treated, sham operated), blue curve (MI), and red curve(TPPU-treated MI). (B) Summary data from A (n = 3 per group). (C) Flow cytometric analysis of β-MyHC expression in myocytes in four groups. (D) Summary datafrom C. (E) Representative lanes of Western blot assay and (F) summary data for atrial natriuretic factor (ANF) and GAPDH (loading control) from whole celllysates from the four groups (n = 3 per group).

Sirish et al. www.pnas.org/cgi/content/short/1221972110 5 of 6

Fig. S7. Examples of cardiac sections stained with Sirius Red from sham-operated, sEHI-treated–sham-operated, MI and sEHI-treated MI mice. The mice werekilled after 3 wk of follow-up. (Scale bars, 200 μm.)

Table S1. Summary of echocardiographic data at 3 wk after MI

Treatment N EDD, cm ESD, cm LVPW (D), cm LVPW (S), cm FS, %

Sham 15 0.33 ± 0.01 0.15 ± 0.01 0.11 ± 0.01 0.15 ± 0.01 54.9 ± 1.8Sham + TPPU 10 0.34 ± 0.01 0.16 ± 0.01 0.11 ± 0.005 0.15 ± 0.01 53.6 ± 2.6MI alone 19 0.36 ± 0.01 0.20 ± 0.01* 0.09 ± 0.005 0.12 ± 0.005* 44.0 ± 1.3*MI + TPPU 20 0.36 ± 0.01 0.18 ± 0.01† 0.09 ± 0.005 0.13 ± 0.005† 50.0 ± 1.4†

EDD, end diastolic dimension; ESD, end systolic dimension; LVPW (D), left ventricular posterior wall thicknessin diastole; LVPW (S), left ventricular posterior wall thickness in systole. Data are mean ± SEM (one-way ANOVAfollowed by Bonferroni tests for post hoc comparison, *P < 0.05 comparing MI alone with sham, †P < 0.05comparing treated vs. untreated MI animals).

Sirish et al. www.pnas.org/cgi/content/short/1221972110 6 of 6