Arjunolic acid, a peroxisome proliferator-activated receptor α agonist, regresses cardiac fibrosis by inhibiting non-canonical TGF-β signaling

Cardiac hypertrophy and associated heart fibrosis remain a major cause of death worldwide. Phytochemicals have gained attention as alternative therapeutics for managing cardiovascular diseases. These include the extract from the plant Terminalia arjuna, which is a popular cardioprotectant and may prevent or slow progression of pathological hypertrophy to heart failure. Here, we investigated the mode of action of a principal bioactive T. arjuna compound, arjunolic acid (AA), in ameliorating hemodynamic load-induced cardiac fibrosis and identified its intracellular target. Our data revealed that AA significantly represses collagen expression and improves cardiac function during hypertrophy. We found that AA binds to and stabilizes the ligand-binding domain of peroxisome proliferator-activated receptor α (PPARα) and increases its expression during cardiac hypertrophy. PPARα knockdown during AA treatment in hypertrophy samples, including angiotensin II-treated adult cardiac fibroblasts and renal artery-ligated rat heart, suggests that AA-driven cardioprotection primarily arises from PPARα agonism. Moreover, AA-induced PPARα up-regulation leads to repression of TGF-β signaling, specifically by inhibiting TGF-β-activated kinase1 (TAK1) phosphorylation. We observed that PPARα directly interacts with TAK1, predominantly via PPARα N-terminal transactivation domain (AF-1) thereby masking the TAK1 kinase domain. The AA-induced PPARα-bound TAK1 level thereby shows inverse correlation with the phosphorylation level of TAK1 and subsequent reduction in p38 MAPK and NF-κBp65 activation, ultimately culminating in amelioration of excess collagen synthesis in cardiac hypertrophy. In conclusion, our findings unravel the mechanism of AA action in regressing hypertrophy-associated cardiac fibrosis by assigning a role of AA as a PPARα agonist that inactivates non-canonical TGF-β signaling.

Cardiac hypertrophy and associated heart fibrosis remain a major cause of death worldwide. Phytochemicals have gained attention as alternative therapeutics for managing cardiovascular diseases. These include the extract from the plant Terminalia arjuna, which is a popular cardioprotectant and may prevent or slow progression of pathological hypertrophy to heart failure. Here, we investigated the mode of action of a principal bioactive T. arjuna compound, arjunolic acid (AA), in ameliorating hemodynamic load-induced cardiac fibrosis and identified its intracellular target. Our data revealed that AA significantly represses collagen expression and improves cardiac function during hypertrophy. We found that AA binds to and stabilizes the ligand-binding domain of peroxisome proliferator-activated receptor ␣ (PPAR␣) and increases its expression during cardiac hypertrophy. PPAR␣ knockdown during AA treatment in hypertrophy samples, including angiotensin II-treated adult cardiac fibroblasts and renal artery-ligated rat heart, suggests that AA-driven cardioprotection primarily arises from PPAR␣ agonism. Moreover, AA-induced PPAR␣ up-regulation leads to repression of TGF-␤ signaling, specifically by inhibiting TGF-␤-activated kinase1 (TAK1) phosphorylation. We observed that PPAR␣ directly interacts with TAK1, predominantly via PPAR␣ N-terminal transactivation domain (AF-1) thereby masking the TAK1 kinase domain. The AA-induced PPAR␣bound TAK1 level thereby shows inverse correlation with the phosphorylation level of TAK1 and subsequent reduction in p38 MAPK and NF-Bp65 activation, ultimately culminating in amelioration of excess collagen synthesis in cardiac hypertrophy. In conclusion, our findings unravel the mechanism of AA action in regressing hypertrophy-associated cardiac fibrosis by assigning a role of AA as a PPAR␣ agonist that inactivates non-canonical TGF-␤ signaling.
Cardiac hypertrophy is accompanied by excess deposition of collagen and other extracellular matrix (ECM) 4 proteins in the heart, and this leads to cardiac stiffness and eventual heart failure (1,2).
Increased hemodynamic load in heart leads to activation of the renin-angiotensin system resulting in heightened local concentration of angiotensinII (AngII), which has been established as a principal causal factor for cardiac hypertrophy and associated fibrosis (3)(4)(5). Therefore, ameliorating AngII-induced cardiac fibrosis might be one of the possible measures for prevention of such disease progression with improved efficacy of cardiac performance.
Phytomedicines have been greatly appreciated in recent times as promising candidates toward alternative therapeutic regime in several fibrotic diseases (6,7). Terminalia arjuna (arjuna), considered as one of the most accepted beneficial plants from ancient times, has also widely been known to have cardiotonic functions (8). Preclinical studies with arjuna extracts as well as clinical trials in the form of combination therapy with standard drugs have shown to be protective in several models of cardiac ailments (9,10).
Arjunolic acid (AA), a chiral triterpenoid, is one of the principal bioactive components of arjuna extracts. Aqueous extract of arjuna bark containing AA significantly inhibited isoprenaline-induced increase in oxidative stress and also prevented fibrosis without regression of hypertrophy or improvement of cardiac function (11). Purified AA resulted in anti-oxidant, anti-platelet, anti-coagulant, anti-necrotic, anti-apoptotic, free radical-scavenging, anti-inflammatory, hypolipidemic, or hypotensive effects in various cardiac disease models (9,10,12,13). However, the precise molecular mechanism of cardioprotection by AA has still remained elusive. Furthermore, pleiotropic effects of AA make it even harder to annotate specific molecular targets within the cellular milieu.
Thus, the objective of our study was to look into the effect of AA, which is purified and crystallized from ethyl acetate and methanol extracts of the corewood of arjuna plant (14) on cardiac hypertrophy-associated fibrosis, and to seek out its plausible mechanism of action on fibrotic signaling through identification of its specific molecular target.
Our study for the first time establishes PPAR␣ as a molecular target of AA. It is a ligand-dependent transcription factor, and the most abundant PPAR isoform in heart (15). PPAR␣ primarily acts as a modulator of energy metabolism but also acts as an anti-inflammatory agent (16,17). Growing evidence has shown that PPAR␣ activators are involved in integrating inflammatory and hypertrophic pathways thereby governing the pathological outcome of hypertrophy-associated fibrosis (18 -20) in hypertensive rats in vivo. Moreover, PPAR␣ double knock-out mice subjected to transverse aortic constriction (TAC) showed heightened fibrotic and inflammatory gene expressions along with severely aggravated hemodynamic function compared with wild-type mice that has undergone TAC (21).
During pressure overload hypertrophy, autocrine and paracrine actions of locally expressed AngII induce TGF-␤ signaling in cardiac fibroblasts (2,3). Upon ligand binding, TGF-␤ receptor II (T␤RII) transphosphorylates TGF-␤ receptor I (T␤RI). In the canonical branch, active T␤RI phosphorylates the "suppressor of mothers against decapentaplegic" (SMAD 2/3), which combines with co-SMADs (SMAD 4) for nuclear translocation. Active nuclear SMADs up-regulate transcription of several genes involved in ECM synthesis (3,22). SMAD3 Ϫ/Ϫ mice have shown reduced cardiac fibrosis after TAC (23). Independent of the SMAD kinase activity, the non-canonical pathway is principally governed by MAPKs, and TAK1 is a principal MAP3K. Phosphorylation-dependent activation of TAK1 requires interaction of its kinase domain with the C terminus of TAK1-binding protein (TAB1) (24,25). Once activated, TAK1 phosphorylates the inhibitor of B kinase ␤ (IKK␤) (but not IKK␣), MKK4/7, and MKK3/6, which in turn phosphorylate NF-B, JNK, and p38 MAPK, respectively (25,26). Ultimately, NF-B and other transcription factors downstream of JNK and p38 MAPK are activated resulting in transcription of the target genes (27). The importance of the non-canonical branch via TAK1 has also been well-documented in cardiac hypertrophy and fibrosis (28). Dominant-negative TAK1 has been found to inhibit TGF-␤-induced hypertrophic events in mouse cardiomyocytes and fibroblasts, including ECM production (29).
The inhibitory effect of PPAR activation upon TGF-␤ signaling has been extensively documented (30 -32). Our work thus aims to study the role of PPAR␣ in AA-driven modulation of the TGF-␤ axes during cardiac hypertrophy and associated fibrosis. This work identifies the PPAR␣-TAK1 interaction as a causal factor preventing TAK1 phosphorylation, and it further analyzes the relative strength of different PPAR␣ domains contributing toward inhibition of non-canonical TGF-␤ axes and regression of subsequent collagen synthesis during cardiac hypertrophy.

Arjunolic acid regresses collagen expression and improves cardiac function during cardiac hypertrophy
AngII treatment in cardiac fibroblasts in vitro (Fig. 1A) and renal artery ligation in vivo (Fig. 1B) confirmed a significant increase in collagen-1 (col-1; 2.29 Ϯ 0.032-fold in vitro and 1.71 Ϯ 0.017-fold in vivo) and collagen-3 (col-3; 1.96 Ϯ 0.035fold in vitro and 1.76 Ϯ 0.015-fold in vivo) gene expressions compared with respective DMSO-treated control samples as revealed by RT-PCR analyses (Fig. 1, A and B). Hydroxyproline assay also showed significant up-regulation in total secreted collagen content in the fibroblast culture supernatant in vitro (471.99 Ϯ 20.188 ng/ml vis à vis 220.57 Ϯ 10.289 ng/ml culture supernatant; Fig. 1C) as well as significantly increased total left ventricular collagen content in vivo (446.63 Ϯ 12.808 g/g vis à vis 214.91 Ϯ 18.828 g/g of wet tissue; Fig. 1D) in hypertrophy samples compared with respective control samples in vitro and in vivo (Fig. 1, C and D).
AA treatment in AngII-pretreated cardiac fibroblasts (Fig.  1A) and renal artery-ligated rat heart (Fig. 1B) showed significant down-regulation of col-1 (2.51 Ϯ 0.025-fold in vitro and 2.20 Ϯ 0.022-fold in vivo) and col-3 (1.77 Ϯ 0.050-fold in vitro and 1.86 Ϯ 0.016-fold in vivo) gene expressions compared with respective DMSO-treated hypertrophy samples in vitro and in vivo (Fig. 1, A and B). Hydroxyproline assay further revealed that AA treatment during hypertrophy leads to significant reduction in total secreted collagen content in fibroblast culture supernatant (301.49 Ϯ 22.191 ng/ml vis à vis 471.99 Ϯ 20.188 ng/ml of culture supernatant; Fig. 1C) as well as total left ventricular collagen content (315.57 Ϯ 28.761 g/g vis à vis 446.63 Ϯ 12.808 g/g of wet tissue; Fig. 1D) compared with respective hypertrophy samples in vitro and in vivo (Fig. 1, C  and D).
In vitro adult cardiac fibroblasts were also treated with equivalent amounts of AA. There was no significant difference in col-1 and col-3 gene expressions (Fig. 1A) or total secreted collagen content in the culture supernatant (Fig. 1C) of AA-treated control fibroblasts compared with fibroblasts treated with equivalent amounts of DMSO.
The viability measured was more than 90% after AA treatment in all hypertrophied fibroblasts as checked by cell viability assay (data not shown). The effective doses of AA for both in vitro and in vivo experiments were determined by maximal regression of collagen during hypertrophy via dose-dependent study (data not shown).

AA directly interacts with PPAR␣
Fluorescence and circular dichroism (CD) titrations of AA with PPAR␣-Steady-state fluorescence titrations of PPAR␣ with increasing concentrations of AA showed no apparent shifts around 330 nm ( max ), indicative of unaltered tertiary architecture. However, the corresponding hypochromic shifts observed indicated binding of AA to PPAR␣ ( Fig. 2A, panel i). The binding was also apparent when monitored through CD analyses where similar titrations induced marginal secondary structure perturbations (near 55 M) in this predominantly ␣-helical protein. (Fig. 2A, panel ii).
Fourier transform infrared spectroscopy (FTIR) analysis-FTIR spectrum of AA supported previous reports (14) showing characteristic -OH, C-H, and CϭO stretches at 3467, 2927, and 1708 cm Ϫ1 positions, respectively. Characteristic C-O stretching vibrations were also obtained at 475, 598, 1042, and 1456 cm Ϫ1 . In vitro translated and purified PPAR␣ revealed intense absorption characteristic of amide-I and amide A bands of the secondary structure of a protein (33,34). The peak at 1642 cm Ϫ1 position represented the amide-I bond. The amide-A band (at 3300 -3600 cm Ϫ1 ), representing the N-H stretching vibrations, was also obtained. Successful interaction between PPAR␣ and AA by in vitro interaction assay was shown AngII-treated cells showed significant decrease in col-1 and col-3 gene expressions compared with hypertrophied fibroblasts. Rpl-32 was used as internal loading control. Control cells treated with either DMSO or AA yielded similar results. AngII-treated cells were also treated with equivalent amounts of DMSO. n ϭ 10 for each experimental group. All the results were expressed as ϮS.E. of three independent experiments. Representative graphs showing relative alterations in collagen gene expressions among different experimental groups. ***, p Ͻ 0.001 with respect to DMSO-treated control cells; ###, p Ͻ 0.001 with respect to AngII-treated cells. B, RT-PCR analyses showing significant increase in col-1 and col-3 gene expressions in right renal artery-ligated rat heart compared with sham-operated control rat group. AA treatment in ligated animals showed significant decrease in col-1 and col-3 gene expressions compared with ligated animals. Rpl-32 was used as internal loading control. Sham-operated control animals and renal artery-ligated animals were also treated with equivalent amounts of DMSO. n ϭ 7 for each experimental group. All the results were expressed as ϮS.E. of three independent experiments. Representative graphs showing relative alterations in collagen gene expressions in different in vivo experimental groups. **, p Ͻ 0.01 with respect to sham-operated control rat group; ***, p Ͻ 0.001 with respect to sham-operated control rat group; ##, p Ͻ 0.01 with respect to renal artery-ligated rat group; ###, p Ͻ 0.001 with respect to renal artery-ligated rat group. C, graphical representation of in vitro hydroxyproline assay showing significantly increased collagen content in AngII-treated fibroblast culture supernatant compared with control cells. AA treatment in AngII-treated fibroblasts showing significant decrease in the collagen content in the culture supernatant compared with AngII-treated cells. Control cells treated with either DMSO or AA yielded similar results. AngII-treated cells were also treated with equivalent amounts of DMSO. n ϭ 10 for each experimental group. Results were expressed as ϮS.E. of three independent experiments. *, p Ͻ 0.05 with respect to control cells; #, p Ͻ 0.05 with respect to Ang II-treated cells. D, graphical representation of in vivo hydroxyproline assay showing significantly increased total left ventricular collagen content in ligated rat hearts compared with sham-operated control hearts. AA treatment in ligated rat hearts showed decreased total left ventricular collagen content compared with ligated rat heart. Equivalent amounts of DMSO were administered into sham-operated and renal artery-ligated rat group. n ϭ 7 for each experimental group. Results were expressed as ϮS.E. of three independent experiments. **, p Ͻ 0.01 with respect to sham-operated control rat group; #, p Ͻ 0.05 with respect to renal artery-ligated rat group. E, graphical representation of HW/BW ratio (in milligrams/gram) ratio showing significant increase in renal artery-ligated rat group compared with sham-operated control rat group. AA treatment in ligated animals showed significant down-regulation of HW/BW ratio compared with hypertrophied rat group. Sham-operated and renal artery-ligated rat group were also treated with equivalent amounts of DMSO. n ϭ 7 for each experimental group. Results were expressed as ϮS.E. of three independent experiments. **, p Ͻ 0.01 with respect to sham-operated control rat group; ##, p Ͻ 0.01 with respect to renal artery-ligated rat group. F, graphical representation of CSA measured from H&E-stained heart tissue sections showing significant up-regulation in renal artery-ligated rat heart compared with sham-operated control rat group. AA treatment in ligated animals showed significant down-regulation of CSA compared with ligated rat group. Sham-operated and ligated rats were also treated with equivalent amounts DMSO. n ϭ 7 for each experimental group. Results were expressed as ϮS.E. of three independent experiments. ***, p Ͻ 0.001 with respect to sham-operated control rat group; ###, p Ͻ 0.001 with respect to renal artery-ligated rat group. G, RT-PCR analyses showing significant increase in hypertrophy marker gene expressions, namely anf, ␤-mhc, and acta1 in renal artery-ligated rat hearts compared with sham-operated control hearts. AA treatment in ligated animals showed significant down-regulation of all these gene expressions compared with ligated rat hearts. Sham-operated control and renal artery-ligated animals were also treated with an equivalent amounts of DMSO. Rpl-32 was used as internal loading control. n ϭ 7 for each experimental group. Results were expressed as ϮS.E. of three independent experiments. Representative graphs showing relative alterations in the hypertrophy marker gene expressions among different experimental groups. ***, p Ͻ 0.001 with respect to sham-operated control rat group; **, p Ͻ 0.01 with respect to sham-operated control rat group; ###, p Ͻ 0.001 with respect to renal artery-ligated rat group; ##, p Ͻ 0.01 with respect to renal artery-ligated rat group. H, M-mode echocardiographic analyses from parasternal short axis view at papillary muscle level graphically representing decreased %FS and increased LvIDd in renal artery-ligated rat group compared with sham-operated control rat group. AA treatment in ligated animals showed restored %FS and lowered LvIDd in ligated rats compared with the hypertrophied rat group. Sham-operated control and renal artery-ligated animals were also treated with equivalent amounts of DMSO. n ϭ 7 for each experimental group. Results were expressed as ϮS.E. of three independent experiments. **, p Ͻ 0.01 with respect to sham-operated control rat group; ##, p Ͻ 0.01 with respect to renal artery-ligated rat group; #, p Ͻ 0.05 with respect to renal artery-ligated rat group.

Anti-fibrotic role of arjunolic acid as a PPAR␣ agonist
by the absence of peaks at 3467, 2927, and 1708 cm Ϫ1 in the complex compared with free AA. Peaks characteristic of free AA at 952, 1018, 1317, 1412, and 1456 cm Ϫ1 and those characteristic of free in vitro translated PPAR␣ at 1642 cm Ϫ1 and in the range of 3300 -3600 cm Ϫ1 were maintained in the complex when compared with free AA or free in vitro translated protein, respectively (Fig. 2B).

AA stabilizes tertiary and secondary architecture of PPAR␣: Thermal melting studies
The stabilization of the tertiary and secondary architecture of PPAR␣ by AA was assessed by monitoring changes in melting temperature (T m ) (Fig. 2C, panels i and ii).

AA binds to the ligand-binding domain (LBD) of PPAR␣: Molecular modeling and molecular docking studies
Rattus norvegicus PPAR␣ protein model was built, and the integrity of the model was confirmed through all atomistic molecular dynamics simulation. To gain atomistic details, binding mode of AA with rat-PPAR␣ (UniProtKB-P37230) was compared with the other known agonists, namely iloprost, TIPP-703, and fenofibrate.

Anti-fibrotic role of arjunolic acid as a PPAR␣ agonist
other PPAR␣ agonists have been reported, including the predocked complex with fenofibrate. With root mean square deviation (RMSD) of ϳ0.62 Å among all aligned structures, it showed structural superposition of AA with other agonists, confined to the same ligand-binding site. This was further confirmed through tertiary structure alignment of modeled PPAR␣ (bound to AA and fenofibrate) with known crystal structures of PPAR␣ in complex with agonists such as iloprost and TIPP-703 ( Fig. 2D and supplemental Table 1).

AA stabilizes the LBD of PPAR␣
For all atomistic molecular dynamics (MD) simulations, the molecular dynamics simulation trajectories of both free and AA-bound PPAR␣ were analyzed for changes in backbone RMSD, gyration radius (R g ), root mean square fluctuations in C␣ atoms (RMSF), as well as protein-ligand H-bond network occupancy. In simulations of PPAR␣ alone, projection of variations in backbone RMSD versus R g showed dual distribution of species along both components ( Fig. 2E, panel i). The projections were, however, dominated by the species with high RMSD and R g compared with initial model. However, I the presence of AA, only moderate variations were observed along both components ( Fig. 2E, panel ii). This system showed single distribution with dominant population close to initial starting conformation (low RMSD and marginal variations in R g ). These variations were identically reflected in RMSF analysis of both systems (

AA-induced PPAR␣ transcriptional activation results in increased ppar␣ gene expression in an autoregulatory loop
Effect of AA upon PPAR␣-mediated autoregulation of PPAR␣ transcription was analyzed by chromatin immunoprecipitation (ChIP) using anti-PPAR␣ antibody coupled to qRT-PCR analyses of PPAR-response element (PPRE) within the PPAR␣ promoter, both in vitro and in vivo. DMSO and nonspecific siRNA (NS siRNA)-treated hypertrophy samples showed significant down-regulation in binding of PPAR␣ upon PPRE sequence within the PPAR␣ promoter compared with DMSO and NS siRNA-treated control samples (1.99 Ϯ 0.078fold in vitro and 2.75 Ϯ 0.199-fold in vivo). However, AA treatment in NS siRNA-treated hypertrophied groups showed significant fold enrichment in binding of PPAR␣ to the PPRE (3.00 Ϯ 0.075-fold in vitro and 3.27 Ϯ 0.237-fold in vivo) compared with DMSO and NS siRNA-treated hypertrophy samples. PPAR␣ knockdown in AA-treated hypertrophied groups showing significantly lowered binding (6.49 Ϯ 0.520-fold in vitro and 8.17 Ϯ 0.749-fold in vivo) compared with AA-treated hypertrophy samples pretreated with NS siRNA were used as negative controls (Fig. 3A). Also, qRT-PCR analyses both in vitro and in vivo indicated significant down-regulation in ppar␣ gene expression in the hypertrophy samples compared with the respective control samples (1.81 Ϯ 0.026-fold in vitro and 3.37 Ϯ 0.154-fold in vivo) and a significant increase in ppar␣ gene expression in AA-treated hypertrophy samples compared with the respective hypertrophied groups (3.45 Ϯ 0.131-fold in vitro and 3.94 Ϯ 0.179-fold in vivo). PPAR␣ knockdown in AAtreated hypertrophied groups showing significantly down-regulated ppar␣ gene expression (5.49 Ϯ 0.208-fold in vitro and 7.04 Ϯ 0.609-fold in vivo) compared with the respective AAtreated hypertrophy samples were used as negative controls (Fig. 3A).

AA increases PPAR␣ protein expression during cardiac hypertrophy
Significant reduction of PPAR␣ protein expression was observed in AngII-treated fibroblasts (2.05 Ϯ 0.088-fold in vitro) and renal artery-ligated rat heart (1.98 Ϯ 0.059-fold in vivo) compared with the respective control groups as analyzed by Western blot. The AA-treated hypertrophy samples showed significant up-regulation in PPAR␣ expression (2.12 Ϯ 0.127fold in vitro and 2.05 Ϯ 0.031-fold in vivo) compared with the hypertrophy groups (Fig. 3B). Successful knockdown of PPAR␣ expression by PPAR␣ siRNA treatment in AA-treated hypertrophy samples was also confirmed by Western blot analyses (3.18 Ϯ 0.191-fold down-regulation in vitro and 1.67 Ϯ 0.025fold down-regulation in vivo compared with respective AAtreated hypertrophy samples) (Fig. 3B). Moreover, time-point analyses showed significant up-regulation of PPAR␣ expression in AngII-treated fibroblasts when treated along with AA compared with only AngII treatment at the respective time points under study (Fig. 3C).

AA-mediated up-regulation of PPAR␣ results in regression of cardiac hypertrophy-associated fibrosis
AA-treated hypertrophy samples pretreated with PPAR␣ siRNA resulted in significantly increased expression of col-1 (2.48 Ϯ 0.137-fold in vitro and 2.58 Ϯ 0.058-fold in vivo) and col-3 (3.19 Ϯ 0.06-fold in vitro and 3.42 Ϯ 0.103-fold in vivo) genes compared with respective AA-treated hypertrophy groups as revealed by qRT-PCR (Fig. 4A).
Hydroxyproline assay also revealed significant recovery of secreted collagen content in fibroblast culture supernatant However, PPAR␣ knockdown in TGF-␤-treated fibroblasts during AA treatment showed a significant increase in phospho/ total TAK1 level (1.75 Ϯ 0.069-fold) with no significant change in T␤RI, T␤RII phospho/total SMAD 2, and phospho/total SMAD 3 expression levels compared with TGF-␤-infused cells treated along with AA ( Fig. 5B and supplemental Fig. S2B).

Inhibition of either TAK1 or its downstream signaling intermediates results in regression of collagen gene expression in AngII-treated cardiac fibroblasts
Significant down-regulation of collagen gene expressions was observed due to siRNA-mediated knockdown of TAK1 (3.71 Ϯ 0.141-fold for col-1 and 2.24 Ϯ 0.072-fold for col-3), Figure 3. AA treatment during hypertrophy increases PPAR␣ expression in an autoregulatory loop. A, ChIP assay with anti-PPAR␣ antibody followed by qRT-PCR analyses of the PPRE showing relative binding of PPAR␣ to the PPRE within PPAR␣ promoter among different experimental groups both in vitro and in vivo are represented graphically on logarithmic scale. Hypertrophy samples showed significantly down-regulated-fold enrichment in binding of PPAR␣ to PPRE within the PPAR␣ promoter compared with respective controls. AA treatment in hypertrophy samples further showed significantly increased-fold enrichment of the same, compared with hypertrophy samples. PPAR␣ siRNA treatment in AA-treated hypertrophy groups showing down-regulated-fold enrichment in binding of PPAR␣ to the PPRE compared with the AA-treated hypertrophy samples were used as negative controls. Control and hypertrophy samples in vitro and in vivo were treated with equivalent amounts of DMSO and non-specific (NS) siRNA. AA-treated hypertrophy samples were treated with equivalent amounts of NS siRNA. Chromatin from each experimental group immunoprecipitated with anti-IgG antibody were used for normalization. n ϭ 3 both in vitro and in vivo. Results were analyzed by one-way ANOVA followed by Tukey's post hoc test and expressed as ϮS.E. of three independent experiments. **, p Ͻ 0.01 with respect to control samples; ###, p Ͻ 0.001 with respect to hypertrophy samples; 11, p Ͻ 0.01 with respect to AA-treated hypertrophy samples in vitro and/or in vivo. Corresponding changes in ppar␣ mRNA expressions between different experimental groups as observed by qRT-PCR are represented graphically on logarithmic scale. Rpl-32 was used as internal loading control. n ϭ 3 both in vitro and in vivo. Results were analyzed by ANOVA followed by Tukey's post hoc test and expressed as ϮS.E. of three independent experiments. **, p Ͻ 0.01 with respect to control samples; #, p Ͻ 0.05 with respect to hypertrophy samples; ##, p Ͻ 0.01 with respect to hypertrophy samples; 1, p Ͻ 0.05 with respect to AA-treated hypertrophy samples; 11, p Ͻ 0.01 with respect to AA-treated hypertrophy samples in vitro and/or in vivo. B, Western blot analyses revealed significant decrease in PPAR␣ protein expression during hypertrophy compared with respective control groups that again showed significant recovery in AA-treated hypertrophy samples compared with respective hypertrophy groups in vitro and in vivo. Successful knockdown of PPAR␣ protein expression was also confirmed by PPAR␣ siRNA pretreatment in AA-treated hypertrophy samples. GAPDH was used as internal loading control. Control and hypertrophy samples were treated with equivalent amounts of DMSO and NS siRNA. AA-treated hypertrophy samples were also treated with NS siRNA. Results were analyzed by ANOVA followed by Tukey's post hoc test and expressed as ϮS.E. of three independent experiments. n ϭ 10 in vitro, n ϭ 7 in vivo for each experimental group. Representative graphs showing relative changes in expression of PPAR␣ among different experimental groups on logarithmic scale. **, p Ͻ 0.01 with respect to control samples; #, p Ͻ 0.05 with respect to hypertrophy samples; ##, p Ͻ 0.01 with respect to hypertrophy samples; 11, p Ͻ 0.01 with respect to AA-treated hypertrophy samples in vitro and/or in vivo. C, Western blot analyses showing time-dependent increase in PPAR␣ expression in AA-treated hypertrophied fibroblasts compared with AngII-treated cells at respective time points under study. GAPDH was used as internal loading control. Results were analyzed by ANOVA followed by Tukey's post hoc test and expressed as ϮS.E. of three independent experiments. n ϭ 5 for each experimental group. Representative graphs showing relative changes in PPAR␣ expressions at respective time points among different groups under study. C, control fibroblasts; A, AngII-treated fibroblasts at different time points; A ϩ AA, AngII-treated fibroblasts at different time points, treated along with AA. C and A cells were also treated with equivalent concentration of DMSO. , Ͻ0.01 with respect to A samples at the 3-h time point; øø, Ͻ0.01 with respect to A ϩ AA samples at the 3-h time point.
NF-Bp65 activity on the collagen1 (Col1a1) promoter was analyzed by dual-luciferase assay showing a significant increase in luciferase activity (2.08 Ϯ 0.120-fold) of the Col1a1 promoter in AngII-treated cardiac fibroblasts compared with respective control cells. Furthermore, NF-Bp65-driven Col1a1 promoter activity was found to be significantly down-regulated (2.03 Ϯ 0.032-fold) by AA treatment in AngII-treated cells compared with hypertrophied fibroblasts. PPAR␣ knockdown in AA-infused AngII-treated fibroblasts significantly restored (1.92 Ϯ 0.058-fold) such promoter activity compared with AA-treated hypertrophied cells. NF-Bp65 knockdown in AngII-treated cells showing down-regulated NF-Bp65 activity (2.14 Ϯ 0.081-fold) on the Col1a1 promoter compared with AngIItreated cells pretreated with NS siRNA was used as negative control (Fig. 6C).

Analyses of interaction between PPAR␣ and TAK1
Molecular modeling study-R. norvegicus TAK1 protein model was built, and integrity of the model was confirmed through all atomistic molecular dynamics simulations (supplemental Fig. S4).
Molecular docking study-For docking between PPAR␣ and TAK1, the interfacial interacting residues between these two proteins were identified using CPORT to limit the plausible protein-protein docking landscape. Using HADDOCK for docking, 23 clusters of the 296 predicted structures were found that represented 70.4% of the water-refined models. The best conformation in the top cluster according to best fit HAD-DOCK score revealed not only that PPAR␣ and TAK1 extensively interact via inter-protein hydrogen bonds, but that it also gives an insight into the possible binding interface. The model with the best HADDOCK score revealed maximum contribution of N-terminal transactivation domain (AF-1) of PPAR␣ for hydrogen bonding with TAK1 with added contribution of a few hydrogen bonds through the hinge region and LBD. Only one residue at position 107 from DBD has been shown to interact with TAK1 (Fig. 7A). Status of interaction between PPAR␣ and TAK1 is summarized in supplemental Table 2, A and B.
Co-immunoprecipitation (Co-IP) assay-AA treatment in hypertrophied fibroblasts resulted in significant up-regulation (3.63 Ϯ 0.334-fold) of PPAR␣-bound TAK1 expression compared with respective hypertrophy samples in vitro as revealed by immunoprecipitation with anti-PPAR␣ antibody followed by Western blot with anti-TAK1 antibody. PPAR␣ overexpression in AngII-treated fibroblasts showing significantly increased binding between PPAR␣ and TAK1 (2.73 Ϯ 0.177-fold) compared with hypertrophied cells pretreated with empty pCDNA6/V5 His B mammalian expression vector were used as positive control (Fig. 7C).

Study of the effect of PPAR␣ domains upon PPAR␣-TAK1 interaction
For His-tagged plasmids with inserts of full-length PPAR␣ (F), N-terminal transactivation domain (AF-1), DNA-binding domain (DBD), or hinge region along with the ligand-binding Control and AngII-treated cells were treated with equivalent amounts of DMSO and NS siRNA. AA-treated hypertrophied cells were also treated with equivalent amounts of NS siRNA. Results were analyzed by ANOVA followed by Tukey's post hoc test and expressed as ϮS.E. of three independent experiments. n ϭ 10 for each experimental group. **, p Ͻ 0.01 with respect to control cells; #, p Ͻ 0.05 with respect to AngII-treated cells; 1, p Ͻ 0.05 with respect to AngII-treated cells treated along with AA. C, graphical representation of in vivo hydroxyproline assay showing significant recovery in total left ventricular collagen content in PPAR␣ siRNA-pretreated AA-infused hypertrophied heart compared with AA-treated renal artery-ligated rat heart in vivo. Sham-operated control rats and renal artery-ligated rats were also treated with equivalent amounts of DMSO and NS siRNA. AA-treated ligated rats were also treated with equivalent amounts of NS siRNA. Results were analyzed by ANOVA followed by Tukey's post hoc test and expressed as ϮS.E. of three independent experiments. n ϭ 7 for each experimental group. **, p Ͻ 0.01 with respect to sham-operated control rat group; ##, p Ͻ 0.01 with respect to renal artery-ligated rat group; 1, p Ͻ 0.05 with respect to AA-treated renal artery-ligated rat group. D, micrographs of Massons' trichrome staining showing significantly decreased %CVF in AA-treated hypertrophy samples compared with renal artery-ligated rat heart, which again showed significant recovery in AA-treated ligated samples pretreated with PPAR siRNA. S, sham-operated control group; L, ligated rat group; L ϩ AA, AA-treated ligated rat group; L ϩ AA ϩ PPAR␣ si, PPAR␣ siRNA-infused AA-treated ligated rat group. S and L groups were treated with equivalent amounts of DMSO and NS siRNA. L ϩ AA group animals were also treated with equivalent amounts of NS siRNA. Results were analyzed by ANOVA followed by Tukey's post hoc test and expressed as ϮS.E. of three independent experiments. n ϭ 7 for each group. (Scale bar, 40 M, magnification ϭ ϫ60). **, p Ͻ 0.01 with respect to sham-operated control samples; ##, p Ͻ 0.01 with respect to renal artery-ligated hypertrophy samples; 11, p Ͻ 0.01 with respect to renal artery-ligated samples treated with AA. E, M-mode echocardiographic analyses of PPAR␣ siRNA-pretreated ligated rats treated with AA showing significant decrease in %FS and significant increase in LvIDd compared with AA-treated ligated rats. Sham controls and renal artery-ligated rats were treated with equivalent amounts of DMSO and NS siRNA. AA-treated ligated rats were also treated with equivalent amounts of NS siRNA. n ϭ 7 for each group. Results were analyzed by ANOVA followed by Tukey's post hoc test and expressed as ϮS.E. of three independent experiments. **, p Ͻ 0.01 with respect to sham-operated control rat group; ##, p Ͻ 0.01 with respect to renal artery-ligated rat group; 11, p Ͻ 0.01 with respect to AA-treated renal artery-ligated rat group.

Study of the effect of PPAR␣ domains in AngII-treated fibroblasts upon modulation of non-canonical TGF-␤ pathway and collagen synthesis
The F plasmid overexpressed AngII-treated cardiac fibroblasts showed significant reduction in the level of phospho/ total TAK1 (3.68 Ϯ 0.065-fold), phospho/total NF-Bp65 The qRT-PCR analyses showed significant down-regulation in col-1 and col-3 gene expressions (3.48 Ϯ 0.17-fold for col-1 and 3.47 Ϯ 0.249-fold for col-3) in F-overexpressed AngIItreated fibroblasts compared with AngII-treated cells pretreated with empty plasmid. A similar trend of regression was Figure 5. AA-mediated up-regulation of PPAR␣ specifically targets TAK1 for regression of TGF-␤ signaling during cardiac hypertrophy. A, Western blot analyses showing significantly reduced T␤RI, T␤RII, phospho/total levels of SMAD 2, SMAD 3, and TAK1 in AA-treated hypertrophy samples compared with hypertrophy groups both in vitro and in vivo. Significantly restored levels of all these proteins were observed in AA-treated hypertrophy groups pretreated with PPAR␣ siRNA compared with AA-treated hypertrophy samples. Control and hypertrophy samples were treated with equivalent amounts of DMSO and NS siRNA. AA-treated hypertrophy samples were also treated with equivalent amounts of NS siRNA. GAPDH was used as internal loading control. n ϭ 10 in vitro and n ϭ 7 in vivo for each experimental group. B, Western blot analyses showing significantly decreased phospho/total TAK1 level during AA treatment in TGF-␤-treated cardiac fibroblasts compared with only TGF-␤-treated cells. PPAR␣ knockdown in TGF-␤-and AA-infused cells showed significant recovery in phospho/total TAK1 level compared with AA-treated fibroblasts pretreated with TGF-␤. Expressions of T␤RI, T␤RII, and phospho/ total levels of SMAD 2 and SMAD 3 remained unaltered in these experimental groups. Control and TGF-␤-treated fibroblasts were also treated with equivalent amounts of DMSO and NS siRNA. AA-treated fibroblasts pretreated with TGF-␤ were also treated with equivalent amounts of NS siRNA. GAPDH was used as internal loading control. n ϭ 5 for each group.

Discussion
Phytomedicines are promising candidates for unraveling novel strategies to combat maladaptive cardiac remodeling. This study focuses on the precise mechanism of protection conferred by the phytochemical AA during pressure overload cardiac hypertrophy and fibrosis. Our investigation showed that AA significantly inhibits excess synthesis of collagen and its subsequent deposition in ECM during cardiac hypertrophy both in vitro and in vivo. Improvement in cardiac function was also observed with AA treatment in renal artery-ligated rats along with significant regression of collagen synthesis and redemption from hypertrophic load (Fig.  1, A-H, and supplemental Fig. S1, A and B). This encouraged us to probe into the specific molecular mechanism involved in AA-mediated amelioration of fibrosis during cardiac hypertrophy.
Arjunolic acid (2,3,23-trihydroxyolean-12-en-28-oic acid) is a pentacyclic triterpenoid monocarboxylic acid substituted by three hydroxy groups at positions 2, 3, and 23 (CHEBI: 68381). Two hydroxyl groups and one hydroxymethyl group are attached to the "A" ring. The carboxyl group is attached at the ring junction of the cis-fused "D" and "E" rings (35,36). Triterpenoids have long been known for their ability to suppress inflammatory pathways (37) and have been recognized to activate different PPARs (38,39). As PPAR␣ is the predominant cardiac isoform of the PPAR family (15), interaction of PPAR␣ with AA was studied in detail.
Stabilizing interaction between AA and PPAR␣ was confirmed by fluorescence, CD titrations, and thermal shift assays (Fig. 2, A and C). FTIR spectrum of the conjugate in a cell-free system also indicated successful docking via masking of the hydrogen bond-forming groups of AA. However, amide-I and amide-A peaks of PPAR␣ remained intact in the conjugate indicating maintenance of protein architecture (Fig. 2B). Further evidence for AA as a potential agonist of PPAR␣ came through molecular docking studies where the former occupied an identical binding pocket as reported for established agonists such as iloprost, TIPP-703 (40,41), and fenofibrate (42), which is a wellknown PPAR␣ activator exerting its effect in cardiac fibrosis (Fig. 2D) (18, 19). Moreover, higher affinity of PPAR␣ to AA compared with the other agonists as estimated by their free energy of binding suggests that AA might act as an even better agonist (supplemental Table 1). Simulation outcomes further confirmed AA-driven stabilization of PPAR␣ via persistence of strong hydrogen bonding network between AA and the LBD of PPAR␣ (Fig. 2E). In many previous reports, the agonistic potential had been attributed to stabilization of helices in this region (31,43,44), which is also observed in our case through RMSF analysis (Fig. 2E, panel iii). Altogether, these data suggested that AA-driven stabilization of PPAR␣ could shift the equilibrium of PPAR␣ toward the active configurations resulting in higher interactions with different co-regulators, thus augmenting activation of PPAR␣-driven transcriptional machinery (45,46).
The presence of PPRE, the putative PPAR␣-binding site on the PPAR␣ promoter, suggests that PPAR␣ might regulate its own transcription (47). ChIP analyses coupled to qRT-PCR revealed AA induced increased binding of PPAR␣ upon PPRE within the PPAR␣ promoter during hypertrophy, and the resultant PPAR␣ transcriptional activation increased PPAR␣ mRNA and protein expression during AA treatment in hypertrophy in an autoregulatory loop (Fig. 3, A and B). Time-dependent rise in PPAR␣ expression in AA-treated hypertrophied cells further confirmed AA-mediated PPAR␣ agonism during hypertrophy (Fig. 3C).
A significant increase in collagen transcription and its extracellular accumulation during PPAR␣ knockdown in AA-infused AngII-treated fibroblasts compared with AA-treated hypertrophied cells implied the importance of PPAR␣ in modulation of collagen transcription in cardiac fibroblasts. Moreover, PPAR␣ knockdown in AA-treated ligated rats resulted in Significantly restored levels of all these proteins were observed during PPAR␣ knockdown in AA-treated hypertrophied groups compared with AA-treated hypertrophy samples. Control and hypertrophy groups were also treated with equivalent amounts of DMSO and NS siRNA. AA-treated hypertrophy samples were treated with equivalent amounts of NS siRNA. GAPDH was used as internal loading control. n ϭ 10 in vitro, n ϭ 7 in vivo for each experimental group. B, graphical representation of qRT-PCR analyses showing significant down-regulation of col-1 and col-3 gene expressions in AngII-induced fibroblasts during knockdown of either TAK1 or NF-Bp65 or p38 MAPK via specific siRNA treatments compared with hypertrophied fibroblasts. JNK-specific siRNA treatment in AngII-treated fibroblasts showed no significant regression of collagen gene expression compared with AngII-treated cells. Control and AngII-treated cells were also treated with equivalent amounts of NS siRNA. Rpl32 was used as internal reference control. Results were analyzed by ANOVA followed by Tukey's post hoc test and expressed as ϮS.E. of three independent experiments. n ϭ 5 for each group. **, p Ͻ 0.01 with respect to control cells; ***, p Ͻ 0.001 with respect to control cells; #, p Ͻ 0.05 with respect to AngII-treated cells; ##, p Ͻ 0.01 with respect to AngII-treated cells. C, dual-luciferase assay showing significant increase in the Col1a1 promoter activity in AngII-treated fibroblasts compared with control fibroblasts. AA treatment in hypertrophied fibroblasts showed significant reduction in Col1a1 promoter activity compared with AngII-treated fibroblasts that again showed significant restoration in AA-treated hypertrophied fibroblasts pretreated with PPAR␣ siRNA. Reduced Col1a1 promoter activity shown in NF-Bp65 siRNA pretreated hypertrophied cells compared with AngII-treated cells was used as negative control. Control and AngII-treated cells were also treated with NS siRNA with or without DMSO. AA-treated hypertrophy samples were also treated with equivalent amounts of NS siRNA. Results were normalized by Renilla luciferase activity in all the treatment groups. Results were analyzed by ANOVA followed by Tukey's post hoc test and expressed as ϮS.E. of three independent experiments. n ϭ 5 for each group. **, p Ͻ 0.01 compared with DMSO and NS siRNA-treated control cells; , p Ͻ 0.01 compared with NS siRNA-treated control cells; ##, p Ͻ 0.01 compared with DMSO and NS siRNA-infused AngII-treated cells; 11, p Ͻ 0.01 with respect to AA-treated hypertrophied cells pretreated with NS siRNA; , p Ͻ 0.01 with respect to AngII-treated cells pretreated with NS siRNA.

Anti-fibrotic role of arjunolic acid as a PPAR␣ agonist
increased synthesis and accumulation of collagen with subsequent deterioration of cardiac function that altogether confirms the influence of PPAR␣ in AA-mediated cardioprotection during hypertrophy and associated fibrosis (Fig. 4, A-E, and  supplemental Fig. S1, A and B). PPAR␣ activation during AA treatment caused inhibition of AngII-induced TGF-␤ signaling by affecting both canonical and non-canonical branches ( Fig.  5A and supplemental Fig. S2A). However, this effect might result from reduction in the TGF-␤ level as a secondary response to minimized hypertrophic load. Therefore, fibroblasts were treated with TGF-␤ instead of AngII in the presence or absence of AA, and the results proclaimed that AA predominantly affects the non-canonical branch of the TGF-␤ signaling pathway via selectively inactivating TAK1 and not T␤RI, T␤RII, or SMADs. PPAR␣ knockdown in TGF-␤-treated fibroblasts during AA treatment restored TAK1 phosphorylation compared with AA-infused TGF-␤-treated fibroblasts further confirming the specificity of AA toward PPAR␣-mediated Results were analyzed by ANOVA followed by Tukey's post hoc test and expressed as ϮS.E. of three independent experiments. Graph showing relative changes in the level of interaction between PPAR␣ and TAK1 between different experimental groups. C, control fibroblasts; A, AngII-treated fibroblasts; A ϩ AA, AA co-treated AngII infused fibroblasts; A ϩ pOV, PPAR␣ overexpressed AngII-treated fibroblasts. *, p Ͻ 0.01 with respect to control fibroblasts; ##, p Ͻ 0.01 with respect to AngII-treated fibroblasts.
AA-mediated inactivation of TAK1 and downstream proteins, namely IKK␤/NF-Bp65, p38 MAPK, and JNK, also suggested significant involvement of PPAR␣ during AA treatment in hypertrophy groups ( Fig. 6A and supplemental Fig. S2C). Knockdown experiments revealed remarkable contribution of TAK1 as well as NF-Bp65 and p38 MAPK in promoting collagen transcription in AngII-treated fibroblasts in vitro. In contrast, JNK showed no effect upon collagen synthesis at hypertrophic stimulus to cardiac fibroblasts in vitro (Fig. 6B). Our group and others have previously shown the involvement of p38 MAPK (48,49) and NF-Bp65 (50,51) in cardiac collagen biosynthesis. In addition to that, analysis of luciferase activity of the Col1a1 promoter containing the NF-Bp65-binding site also revealed the role of NF-Bp65 as a transcriptional activator of collagen promoter during hypertrophy, which could be targeted for inhibition by AA-driven PPAR␣ activation (Fig. 6C). Interestingly, PPAR␣-mediated down-regulation of non-canonical TGF-␤ axes during AA treatment in hypertrophy samples was associated with decreased phosphorylation of TAK1, without any alteration in its total level (Fig. 5, A and B). This clearly indicates that changes in TAK1 activation during AA treatment in hypertrophy samples result from post-translational modifications and are not related to PPAR␣-mediated regulation of gene transcription. Reports stating the roles of PPARs in modulating signaling pathways by interacting with other signal intermediates (52-54) prompted us to check whether PPAR␣ directly interacts with TAK1. Successful docking between PPAR␣ and TAK1 as revealed by in silico analyses (Fig. 7A) was further validated by FRET-FRAP assay in fibro-

Anti-fibrotic role of arjunolic acid as a PPAR␣ agonist
blasts (Fig. 7B). Additionally, co-IP data showing significantly higher interaction between PPAR␣ and TAK1 in AA-treated hypertrophied cells as well as in PPAR␣-overexpressed hypertrophied cells compared with AngII-treated cells also suggested that the level of PPAR␣-bound TAK1 is directly proportional to the availability of PPAR␣ (Fig. 7C).
Analyses of domain specificity for PPAR␣ and TAK1 interaction in AngII-treated fibroblasts revealed that the AF-1 domain of PPAR␣ contributes maximally to the binding strength of PPAR␣ with TAK1 when compared with that of full-length PPAR␣ (F) (Fig. 8A and supplemental Fig. S5A). The best fit cluster of PPAR␣-TAK1 interaction model in our bioinformatic analyses also identified the highest number of interacting amino acids from the AF-1 domain of PPAR␣ (supplemental Table 2A). Interestingly, in silico data suggest that TAK1 interacts with PPAR␣ via amino acids belonging to its kinase domain (36 -291 amino acids in rat-TAK1), which also contains TAK1 phosphorylation sites (Thr-184/187) (Fig. 7A and supplemental Table 2A). This further indicates that PPAR␣-TAK1 interaction inactivates TAK1 possibly by preventing its phosphorylation.
Wet lab experiments confirmed that the higher the binding between different PPAR␣ domains and TAK1 in AngII-treated fibroblasts, the greater is the amount of regression in TAK1 phosphorylation with subsequent down-regulation of non-canonical TGF-␤ axes ( Fig. 8B and supplemental Fig. S5B). This finding also corroborates the inactivation of the non-canonical branch of TGF-␤ signaling (Figs. 5A and 6A) during AA treatment in hypertrophy samples resulting from AA-induced PPAR␣ up-regulation (Fig. 3, A and B) and increased interaction between PPAR␣ and TAK1 (Fig. 7C). The degree of regression of collagen synthesis as observed by overexpression of different PPAR␣ domains into hypertrophied fibroblasts also followed the similar trend revealing maximum regression by AF-1 overexpression (Fig. 8C).
An earlier report suggested that deletion of the AF-1 region of PPAR␣ did not affect its function keeping its DNA-binding activity unaltered (55). However, our study shows that the AF-1 region of PPAR␣ promotes PPAR␣-TAK1 protein-protein interaction resulting in suppression of the non-canonical TGF-␤ pathway and reduced collagen gene transcription during cardiac hypertrophy. The H region (helix-1) of PPAR-␥ has been reported to bind with cytosolic kinases such as ERK5 (52) and PKC-␣ (53) in other cell types. Although significantly lower than AF-1, interaction of H ϩ LBD of PPAR␣ with TAK1 also had a significant role in regression of non-canonical TGF-␤ pathway-induced collagen synthesis during hypertrophy.
In summary, our work for the first time shows a precise mechanism of AA action in regression of cardiac hypertrophyassociated fibrosis and subsequent improvement of cardiac function. This work assigns a new role of AA as a PPAR␣ agonist, and such transcriptional activation and increased expression of PPAR␣ during the abovementioned pathophysiological condition inactivates the non-canonical TGF-␤ axes (Fig. 9A). Specific molecular mechanism involves direct interaction of PPAR␣, predominantly by its N-terminal AF-1 domain with TAK1 kinase domain, leading to decreased phosphorylation of TAK1 and its downstream signal intermediates ultimately inhibiting excess collagen transcription during hypertrophy (Fig. 9B).
TGF-␤ signaling is one of the key mediators for maintenance of normal cellular function and homeostasis of cardiac fibroblasts (56). As TAK1 resides at a moderately downstream posi- Figure 9. Schematic representation of the molecular mechanism of AA action upon cardiac hypertrophy-associated fibrosis. A, during cardiac hypertrophy TGF-␤ signaling pathway action is promoted leading to excess collagen synthesis with down-regulated PPAR␣ expression. Treatment with AA in hypertrophy samples increases PPAR␣ expression in an autoregulatory loop leading to increased binding of PPAR␣ to TAK1 thereby ameliorating TAK1-driven non-canonical TGF-␤ axes with subsequent regression of collagen synthesis. B, schematic representation of different PPAR␣ domains interacting with TAK1 and the role of PPAR␣-TAK1 interaction in prevention of phosphorylation-dependent activation of TAK1 for subsequent regression of collagen synthesis in AngII-treated adult cardiac fibroblasts.

Anti-fibrotic role of arjunolic acid as a PPAR␣ agonist
tion in TGF-␤ axes, inhibiting TAK1 by using AA as an antifibrotic agent might be therapeutically advantageous. Further detailing of PPAR␣-binding surface in TAK1 or the influence of TAK-binding proteins in such interaction would be necessary for better understanding of the regulation of TAK1 autoactivation and its role in ECM turnover. Not only that, but transition of such a natural compound from initial screening through preclinical and clinical trials, and finally, to a marketable drug form is associated with challenging demands for the plant source. Therefore, organic synthesis of AA bio-mimetics might be beneficial considering the wide range of protective effects exerted by AA with enhanced clinical significance against such deadly diseases.

Isolation and culture of adult cardiac fibroblasts
Adult cardiac fibroblast cells were isolated from 28-week-old male Wistar rat hearts by the collagenase dispersion method. Briefly, animals were euthanized in a prefilled carbon dioxide (CO 2 ) chamber with 100% concentration of CO 2 , and the hearts were chopped and digested by collagenase (80 units/ml DMEM; Worthington). The cells were pelleted by centrifugation and resuspended in fresh complete DMEM supplemented with 10% fetal bovine serum and plated in cell culture flask. Cells were maintained at 37°C with 5% CO 2 and were subsequently passaged (49).

Generation of cardiac hypertrophy in vitro and in vivo
75-80% confluent serum-starved adult cardiac fibroblasts were treated with 10 Ϫ8 mol/liter (Sar1)-AngII (Bachem, Torrance, CA) for 24 h to generate cardiac hypertrophy model in vitro. AngII was replenished every 6 h throughout the incubation period (49).
In vivo cardiac hypertrophy model was generated by ligating the right renal artery of 28-week-old male Wistar rats (250 -300 g) as described earlier (49). Rats were anesthetized with an intraperitoneal injection of a mixture of ketamine (80 mg/kg) and xylazine (12 mg/kg). Sham-operated control group underwent a similar procedure without renal artery ligation. Animals were maintained in optimum condition for 14 days and were sacrificed on the 15th day after surgery. Hearts were removed, collected in liquid N 2 , and stored in Ϫ80°C for future use.
Hypertrophy was measured by the HW (in milligrams) to BW (in grams) ratio, and activation of hypertrophy marker genes (anf, ␤-mhc, and acta1) was measured by RT-PCR analyses (57).

Measurement of cardiomyocyte CSA
To analyze cardiomyocyte cross-sectional area as a marker of hypertrophy, hearts were excised, washed in 1ϫ PBS, and fixed in Karnovsky's fixative. The samples were then paraffin-embedded and cut into 4 -5-m sections. Tissue sections were processed and stained with hematoxylin and eosin (H&E), and the cardiomyocyte CSA in each sample was quantified (57) using a computer morphometric program (ImageJ, National Institutes of Health).

Evaluation of collagen deposition
Hydroxyproline assay-Hydroxyproline assay was performed to measure total secreted collagen content in fibroblast culture supernatant in vitro and total left ventricular collagen content from rat heart tissues. With the help of standard curve, hydroxyproline content in the unknown samples was calculated. The amount of collagen was estimated by multiplying hydroxyproline content by a factor of 8.2 (49).
Estimation of %CVF-Heart tissues from each group of animals were taken for the analysis of %CVF. Coronal sections obtained from the equator of the left ventricle were fixed and stained with Massons' trichrome (Sigma) to analyze collagen deposition under microscope (Nikon NIS BR) (Nikon, Shinagawa, Tokyo, Japan) following standard protocol. Fifty sections were scanned, and at least 10 images were captured from each section. Images were digitized and processed by ImageJ, a computer morphometric program. CVF was calculated as the sum of all collagen-stained tissue areas of the coronal sections represented as percentage (%) of the total surface of the section. Color segmentation was applied to separate the stained tissues from other nonspecific objects (58).

Determination of cardiac function
Cardiac function of lightly sedated animals from shamoperated control group, renal artery-ligated group, AA-treated hypertrophied rat group, and PPAR␣ siRNA-infused AAtreated ligated rat group was measured by M-mode echocardiographic analysis by a transthoracic study in a short axis view at a papillary muscle level on the 15th day before euthanization. In vivo ventricular PPAR␣ siRNA injection was also guided by echocardiography. Digitized images were obtained using an ultrasound system (Vivid S5 system, GE Healthcare) for the calculation of %FS, LvIDd, LvAWDd. LvAWDs, LvPWDd, LvP-WDs, SV, and EF following standard guidelines (58).

Treatment of fibroblasts with TGF-␤
Another set of adult cardiac fibroblasts was treated with active TGF-␤-2 (Life Technologies, Inc.) at a dose of 8 ng/ml serum-free media for 16 h, as described previously (59), to induce hypertrophic effect. TGF-␤-treated cells were pretreated with AA with or without simultaneous treatment of PPAR␣ siRNA following standard protocol.

Treatment with AA in vitro and in vivo
AA was extracted and purified from core wood of the plant T. arjuna (14). In vitro hypertrophied fibroblasts were treated Anti-fibrotic role of arjunolic acid as a PPAR␣ agonist with 20 M AA (dissolved in DMSO) 6 h after first AngII treatment. Effective dose was chosen from a dose-dependent study (5-100 M AA) based on optimal results with minimum mortality. A group of control fibroblasts was also treated with 20 M AA. All other control and AngII-treated cells were also treated with equivalent amounts of DMSO. Time-dependent changes upon AA treatment were analyzed in AngII-treated cells at 3 and 6 h after induction with AA. A set of TGF-␤-treated cells were also treated with 20 M AA 6 h after TGF-␤ treatment.
In vivo renal artery-ligated rats were treated with AA and dissolved in DMSO by intraperitoneal injection on every alternate day from day 6 to day 14 at a dose of 10 mg/kg/day after ligation. Effective dose was chosen after a dose-dependent study (5 mg to 50 mg/kg/day as described above) based on optimal results. Untreated sham-operated rats and renal arteryligated rats were also treated with equivalent amounts of DMSO. Animals were sacrificed after 14 days. Commercially available arjunolic acid (SMB00119; Sigma) was also used at similar dosage showing uniform results (data not shown).

Cell viability assay
Cell viability and cytotoxicity in the presence or absence of AA in AngII-treated fibroblasts were determined colorimetrically by Cell titer 96 AQ ueous one solution cell proliferation assay (Promega, Madison, WI) following the manufacturer's protocol.
For in vivo experiments, PPAR␣ siRNA (S130650, Ambion In vivo siRNA, Thermo Fisher Scientific) in RNase-free sterile PBS was injected directly into ventricles of ligated animals (49) that are to be treated with AA with simultaneous echocardiographic guidance, following the manufacturer's instructions at a concentration of 10 nmol on every alternate day from the 5th day of ligation before sacrifice. Sham-operated and renal arteryligated rats were also treated with equivalent amounts of DMSO along with in vivo NS siRNA (4457287, Ambion In vivo Negative Control siRNA, Thermo Fisher Scientific). Ligated rats to be treated with AA were also pretreated with equivalent amounts of NS siRNA.

Structural modeling of PPAR␣ and TAK1
The amino acid sequence of R. norvegicus PPAR␣ (P37230) and TAK1 (P0C8E4) were retrieved from the UniProtKB data-base (http://www.uniprot.org/). 5 The retrieved amino acid sequences of PPAR␣ and TAK1 were subjected to a proteinprotein BLAST (BLASTp) search against the PDB to identify a suitable template structure for comparative modeling. The 3D structure predictions of these two proteins were then performed using I-TASSER (60). For rat-PPAR␣, PDB code 4BCR, chain A, was selected as the best suited template with 99% identity and 45% sequence coverage (E value 0.0). For rat-TAK1, PDB code 5JGA, chain A, was selected as a suitable template with 93% identity and 58% sequence coverage (E value 5e Ϫ170 ). The best I-TASSER outputs were refined using ModRefiner (61). The I-TASSER Z-scores were analyzed to check the likelihood that the domains had correct folds and correct overall tertiary structure. The final model was also checked for its quality using ProQ2 (62). Furthermore, refined structures were subjected for MD simulation.
All atomistic MD simulation-To check for conformational stability, MD simulation was done using GROMACS 4.6.1 (63). All atomistic simulations were carried out using the CHARMM36 all-atom force field (64 -66) using periodic boundary condition. The starting model was solvated in a periodic box with TIP3 water model. Nine Na ϩ ions were added to the solvent to neutralize electrical net charge of the protein.
The system was then minimized for 50,000 steps using a steepest descent algorithm with emtol of 200 kJ/mol after a minimization with emtol of 100 kJ/mol. This was followed by an equilibration run of 100 ps in NVT ensemble with restraints on the protein atoms. The NPT ensemble was used for production simulation. Systems were simulated at 310 K and maintained by a Berendsen thermostat with a time constant of 1 ps with the protein and non-protein molecules coupled separately. Pressure coupling was done employing a Parrinello-Rahman barostat using a 1 bar reference pressure and a time constant of 2.0 ps with compressibility of 4.5e Ϫ5 bar using isotropic scaling scheme. Electrostatic interactions were calculated using the Particle Mesh Ewald (PME) summation with a 2-fs time step for each run. For simulations of PPAR␣-AA complex, identical parameters were employed. Topology for AA (PubChem ID: CID73641) and other parameters were assigned using SwissParam (67).
Three-dimensional structural modeling of the chemical compounds-The 2D structures of AA (PubChem ID: CID73641) and fenofibrate (PubChem ID: CID3339) were retrieved from the NCBI PubChem database. The compounds were drawn using ChemDraw 8 software. The two-dimensional molecules were then converted to three-dimensional structures using OpenBabel (68). The energy minimization was performed using Maestro 9.8 (Schrödinger, LLC, New York).
Molecular docking-Docking of AA to PPAR␣ was performed using the BSP-SLIM tool (69). Briefly, the holo-structures with similar global topology to PPAR␣ were identified, and the geometric centers of bound ligands in the holo-forms were clustered to identify putative AA-binding sites. Subsequently, shape and chemical feature comparison of multiple target ligand conformers with all negative images of binding pockets was done employing the OMEGA program. Best overlay for each ligand conformer with the negative images was carried out using OEChem toolkit version 1.7, whereas the chemical features of AA were assigned using the Implicit-Mills-Dean color force field (70). Finally, all ligand conformations were sorted by docking scores with a RMSD tolerance limit of 4 Å.
The top lowest energy binding complex was re-ranked by performing flexible docking using Molecular Operating Environment (version 2009.10). The triangular matching placement method with the Affinity dG scoring function was employed to generate most favorable poses of ligand conformations by aligning ligand triplets of atoms with triplets of receptor site points. AA-PPAR␣ docking scores were compared with the score obtained for known agonists of PPAR␣, i.e. iloprost (PDB code 3SP6) and TIPP-703 (PDB code 2ZNN) and fenofibrate (obtained through pre-docking by BSP-SLIM). In contrast, in silico binding affinities were calculated using AutoDock Vina 1.1.2 (71) extension of UCSF chimera (72).
For protein-protein docking, residues forming the interface surfaces were predicted. CPORT program drew a consensus for the interface residues using PIER, cons-PPISP, SPPIDER, and PINUP. The consensus residues were mapped onto the model. The predicted interface residues were submitted to the PPAR␣ and TAK1 monomer model for protein-protein docking to build the heterodimer model, using HADDOCK web server (73,74). Molecular graphics and analyses were performed with the UCSF Chimera (72).

Fluorescence and CD spectroscopy
Fluorometric titrations were carried out in LS55 fluorescent spectrometer (PerkinElmer Life Sciences). At constant PPAR␣ (in vitro translated and purified) protein concentration of 8 M, titrations were carried with increasing concentrations of AA up to 100 M with a pre-scan delay of 300 s. Incubated samples were excited at 274 nm due to absence of tryptophan, and emission spectra were recorded from 290 to 400 nm with excitation/ emission slits set 10/10.
CD titrations were carried out in J-1500 circular dichroism spectrophotometer (Jasco, Easton, MD). Far-UV CD titrations were carried out following similar protocol in a 2-mm path length quartz cuvette and scanned from 250 to 200 nm at scan rate of 200 nm per min.

Fluorescence-based thermal shift (FTS) assay
FTS assay for PPAR␣ was done to assess tertiary structure stabilization by monitoring change in Tyr emission at 330 nm over the temperature range from 25 to 90°C with step size of 2°C. Excitation and emission slits were kept at 5/10 with an equilibration time of 1 min at each step.
Thermal denaturation of PPAR␣ (8 M) was done both in the absence or presence of AA (100 M). AA incubation with PPAR␣ was done 30 min prior to the experiment. To assess secondary structure stabilization of PPAR␣ by AA, changes in both 218 and 222 were monitored against temperature range of 25 to 90°C at ramping rate of 1°C per min.

In vitro coupled transcription-translation and in vitro interaction assay
Full-length PPAR␣, cloned in pCDNA6/V5 His B mammalian expression plasmid (Invitrogen), was subjected to in vitro coupled transcription-translation using TNT Quick-coupled Transcription/Translation system (Promega, Madison, WI) according to the manufacturer's specifications. Briefly, 2 g circular plasmids were added to the rabbit reticulocyte lysates in 50-l reactions for 60 -90 min at 30°C. These His 6 -tagged products were then purified using nickel-nitrilotriacetic acid spin kit (Qiagen, Hilden, Germany) in native condition and analyzed by SDS-PAGE followed by Coomassie Blue staining to check the purity (data not shown). Purified products were then subjected to in vitro interaction in binding buffer overnight on a rocker at 4°C with AA (59). This complex along with the purified protein was used for further assay.

FTIR spectroscopy
FTIR spectra were obtained with a Jasco FTIR 6300 spectrometer (Jasco, Easton, MD). The spectra presented were obtained from an average of 64 scans. The resolution of the spectra was 4 cm Ϫ1 . The spectra were acquired at room temperature. All data analyses were performed with the data analysis software Origin (OriginLab, Northampton, MA) provided with the FTIR instrument.

RNA isolation
Total RNA from cells (24-h treatment) and tissues (on the 15th day after surgery) was isolated using TRIzol reagent (Invitrogen) following the manufacturer's instructions.

RT-PCR and qRT-PCR
1 g of total RNA was used to make cDNA using the cloned avian myeloblastosis virus first-strand cDNA synthesis kit (Invitrogen). RT-PCR was done for col-1, col-3, anf, ␤-mhc, acta1, and ribosomal protein L32 (rpl 32). rpl32 was used as loading control.
Relative quantification of PCR-amplified products was also done by qRT-PCR with Power SYBR Green TM PCR master mix using Applied Biosystems StepOne real-time PCR system (4369074, Applied Biosystems) for col-1, col-3, ppar␣, and rpl32, where rpl32 was used as a reference gene to normalize expression of other genes. Relative gene expression was quantified by comparative "Ct" (2 Ϫ⌬⌬Ct ) method (49). Both RT-PCR and qRT-PCR primer sequences are summarized in supplemental Table 3, A and B.

ChIP assay
ChIP analysis was performed with modifications from the methodology described by Kouskouti and Talianidis (75). Briefly, cardiac fibroblasts and cardiac tissue sections were treated with 37% formaldehyde (final concentration 1%) for cross-linking proteins to DNA. Cross-linking was stopped by adding 10ϫ glycine (final concentration 0.125 M) after 10 min and mixed at room temperature. Cardiac fibroblasts were now scraped out in ice-cold 1ϫ PBS with protease inhibitor cocktail (PIC; P8340; Sigma). Cardiac tissue samples with cross-linked Anti-fibrotic role of arjunolic acid as a PPAR␣ agonist chromatin were also taken in ice-cold 1ϫ PBS with PIC and were thoroughly minced with a Dounce homogenizer. Both in vitro and in vivo samples were now centrifuged at 1500 rpm for 5 min at 4°C, and the pellets were resuspended in sonication buffer (50 mM HEPES, 140 mM NaCl, 1 mM EDTA, 1% Triton X-100, 0.1% sodium deoxycholate, and 0.1% SDS, pH 7.9). The samples were now sonicated and centrifuged to obtain supernatants containing 500 -800-bp chromatin fragments. For each reaction, 400 g of purified chromatin was precleared with protein G-Sepharose bead (P3296; Sigma) and immunoprecipitated overnight at 4°C either by anti-PPAR␣ antibody or anti-IgG antibody (used as a negative control) (antibodies listed in supplemental Table 4). Furthermore, antibody-bound chromatin complexes were washed using low-and high-salt wash buffers followed by Tris/EDTA (TE) buffer wash, all at 4°C. Protein-DNA complexes were eluted from beads by elution buffer (50 mM NaHCO 3 , 50 mM Tris, 1 mM EDTA, and 1% SDS) for 10 min at 65°C. To separate protein from DNA, samples were treated with 21 l of 4 M NaCl at 65°C overnight. Further purification of DNA was done by RNase A (R4875, Sigma) incubation at 37°C for 1 h followed by addition of proteinase K (P6556, Sigma), EDTA, and Tris, pH 6.5, for 2 h at 42°C. DNA extraction was done following phenol/chloroform isoamyl alcohol method, and the purity and concentration were analyzed. 2 l of DNA from each sample were used for ChIP-coupled qRT-PCR (primers listed in supplemental Table 3B). Data analysis was done by the "Fold Enrichment Method," where ChIP signals are normalized by the IgG signals.

Dual-luciferase assay
2.4-kb collagen 1a1-luc plasmid with insert sequence ranging between Ϫ2300 and 100 bp of col1a1 gene (UniProt Q63121) with putative NF-Bp65-binding site was constructed in PGL-3 vector following standard protocol. Primers are listed in supplemental Table 3D. Clones were confirmed by sequencing (Applied Biosystems TM , 3730 DNA analyzer). Col1a1-luc and pRL-TK (Promega) were used for dual-luciferase assay as per the manufacturer's instruction. Firefly, luciferase activity of Col1a1 promoter containing PGL-3 vector, and Renilla luciferase activity of PRL-TK were measured in 20 l of cell lysate using the Dual-Luciferase Reporter Assay System (Promega) using a luminometer (Varioskan TM LUX multimode microplate reader, VL0LA0D0, Thermo Fisher Scientific). Relative activity was defined as the ratio of firefly luciferase activity to Renilla luciferase activity (76).

Protein isolation
Protein from adult cardiac fibroblasts was isolated using M-PER mammalian protein extraction reagent (Thermo Fisher Scientific) according to the manufacturer's protocol. Protein extracts were prepared from ventricular tissue following the procedure described previously (77).

Co-IP assay
Co-IP was done following the manufacturer's protocol (Pierce Co-IP kit, Thermo Fisher Scientific). Briefly, proteins were incubated with fast-flow protein G-or protein A-Sepharose beads and centrifuged to eliminate non-specifically bound proteins (59). 200 g of precleared protein were immunoprecipitated using primary antibody. Immunoprotein complex was again incubated with protein G (P3296, Sigma) or protein A (P3391; Sigma)-Sepharose beads. Attached proteins were then eluted from beads in 1% SDS buffer followed by immunoblotting. In this study, immunoprecipitation was done using primary antibodies for anti-PPAR␣ or anti-His (antibodies are listed in supplemental Table 4). For immunoprecipitation with anti-His, proteins were purified in nickel column, as described previously for in vitro protein purification above. Western blot analysis was done using monoclonal antibody against TAK1. Normalization was done by Western blot analysis using the same antibodies used to immunoprecipitate the proteins.

Anti-fibrotic role of arjunolic acid as a PPAR␣ agonist FRAP
For endogenous PPAR␣ and TAK1 interaction, cells were treated with PPAR␣-TRITC and TAK1-FITC. As TRITC and FITC work as a donor-acceptor in FRAP assay, TRITC was subjected to 50% bleaching, and FITC intensity in a single cell was measured. FRET efficiency was calculated in confocal microscope (Olympus FV1200, Olympus, Singapore) (76).

Statistical analysis
All the results were expressed as means Ϯ S.E. of three independent experiments. Data were analyzed by independent sample t test by SPSS (version 14.0). For multiple comparison experiments, one-way analysis of variance (ANOVA) was conducted followed by Tukey's post hoc test. Results with a p value Ͻ0.05 were considered significant.
Author contributions-S. Sarkar conceived the idea of the project. T. B. designed and conducted most of the experiments, analyzed the results, and wrote the paper with S. Sarkar. E. C. assisted in most experiments with animals. J. S. and B. K. did most of the bioinformatic and biophysical studies and analyzed the data. A. R. conducted protein modeling and protein-protein docking, and analyzed the data with S. Sengupta. V. T. did extraction and purification of arjunolic acid used in the study.