Peroxisome Proliferator-Activated Receptor Gamma Negatively Regulates the Differentiation of Bone Marrow-Derived Mesenchymal Stem Cells Toward Myofibroblasts in Liver Fibrogenesis

Liver fibrosis is characterized by overproduction of collagen, especially collagen α1 (I) (Col α1 (I)) and collagen α1 (III) (Col α1 (III)), and other extracellular matrix (ECM) components [1,2]. Myofibroblasts, characterized by the cytoskeletal protein α-smooth muscle actin (α-SMA), are known to be the predominant cells to generate Col α1 (I)and Col α1 (III). In addition to hepatic stellate cells (HSCs), portal fibroblasts, circulating fibrocytes and epithelial-mesenchymal transition (EMT), bone marrow (BM)-derived cells have recently been considered as another origin of hepatic myofibroblasts [2,3,4]. In particular, by using lethally irradiated female mice which receive male BM transplants, and tracking BM-derived cells through in situ hybridization for the Y chromosome, Russo et al. [5] found that 70% of hepatic myofibroblast populations are from BM-derived mesenchymal stem cells (BMSCs). In keeping with this finding, we also proved that a significant proportion of hepatic myofibroblasts is of BMSC-differentiation origin with a model of lethally irradiated mice receiving BM transplants from enhanced green fluorescent protein (EGFP) transgenic mice [6]. Considering the importance of BMSCs in liver fibrosis, identification of the molecular mechanisms underlying BMSC differentiation toward myofibroblast may represent an effective strategy for the treatment of fibrotic liver disease.

Transforming growth factor β1 (TGF-β1) is a key profibrotic cytokine in the fibrogenesis [7,8,9,10], and can induce the differentiation of several sorts of fibroblasts toward myofibroblasts in vitro, such as fibroblasts from skin, lung, kidney and heart origin [11,12,13,14]. Our previous study has demonstrated that sphingosine kinase/sphingosine 1-phosphate (S1P)/S1P receptor axis is involved in the differentiation of BMSCs toward myofibroblasts induced by TGF-β1 during liver fibrosis [9]. Nevertheless, it is still not clearly illustrated whether other molecules implicated in this process and the underlying mechanisms.

Peroxisome proliferator-activated receptors (PPARs) are ligand-activated transcription factors belonging to the nuclear hormone receptor superfamily [15], with three subtypes, PPARα, PPARβ/δ and PPARγ, relate to multiple functions, including metabolism, immune responses and cellular proliferation [16]. PPARγ is mainly known to regulate adipocyte differentiation and fatty-acid uptake and storage [17,18,19]. Some clinical studies reported that synthetic PPARγ agonists, such as thiazolidinedione, have beneficial effects in patients with type 2 diabetes [20,21]. Besides its function in metabolism, a new action of PPARγ in fibrosis has been revealed. It has been reported that activation of PPARγ by specific agonists could ameliorate fibrosis in several organs, such as lung, kidney and heart [22]. In particular, several recent studies have illustrated that PPARγ participates in the myofibroblast differentiation [11,13,23,24]. However, whether PPARγ is link to regulate the differentiation of BMSCs toward hepatic myofibroblasts and its molecular mechanisms remain obscure.

Based on the scientific background, the aim of this study is to elucidate the function of PPARγ in BMSCs differentiation toward myofibroblasts in liver fibrogenesis. Here, we showed that PPARγ expression was reduced in mouse liver fibrogenesis, as well as in the differentiation of primary mouse BMSCs toward myofibroblasts induced by TGF-β1 in vitro. In addition, knock down of PPARγ by siRNA contributed to the differentiation of BMSCs. Furthermore, activation of PPARγ by ligands inhibited BMSCs differentiation toward myofibroblasts in vitro and in vivo, and ameliorated liver fibrosis. These observations strongly indicate that PPARγ negatively governs the differentiation of BMSCs toward myofibroblasts, suggesting that PPARγ-targeted therapy could be a potential strategy in the treatment of liver fibrosis.

α-MEM was from Invitrogen (Grand Island, NY). Fetal bovine serum was from Hyclone/Thermo Scientific (Victoria, Australia). TGF-β1 came from PeproTech (London, UK). 15d-PGJ₂ was from Cayman Chemical (Ann Arbor, MI). Troglitazone and ciglitazone were from Biomol (Tebu, France). Antibodies against PPARγ, Col a1 (I), Col a1 (III) and tubulin were from Santa Cruz Biotechnology (CA, USA). Anti-glyceral-dehyde-3-phosphate dehydrogenase (GAPDH) monoclonal antibody was from Cell Signalling (Beverly, MA). PCR reagents were from Applied Biosystems (Foster City, CA). Anti-α-SMA antibody, bovine serum albumin (BSA) and other common reagents were from Sigma-Aldrich (St. Louis, MO).

BM cells were isolated from bone marrow by flushing the tibias and femurs of ICR mice aged 3 weeks (Laboratory Animal Center, Capital Medical University) with culture medium using a 25-gauge needle. The cells were then passed through 70-mm nylon mesh and were washed three times with PBS containing 2% fetal bovine serum. BMSCs cultured as described previously [5]. In brief, BM cells were cultured with BMSCs culture medium (α-MEM) containing 20% fetal bovine serum for 1 week at 37°C in 5% CO₂. To remove the nonadherent cells, the culture medium was replaced twice a week. α-MEM containing 15% fetal bovine serum was used to culture BMSCs after the first subculture. BMSCs are characterized by the phenotypes as positive for CD44, CD105 and CD166, but negative for CD14, CD34 and CD45 in flow cytometry analysis [6]. BMSCs of passage 3 to passage 5 were used in the experiments. All animal work was performed under the ethical guidelines of the Ethics Committee of Capital Medical University

Mouse models of chronic liver fibrosis were performed by injection of carbon tetrachloride (CCl₄) or bile duct ligation (BDL). ICR mice aged 6 weeks received intraperitoneal injections of 1 µL/g body weight of a CCl₄/olive oil (OO) mixture, 1:9 v/v, twice per week. 15d-PGJ₂ (0.3mg/kg body weight) or saline was administrated the day before CCl₄ or OO treatment, and then twice per week before CCl₄ or OO treatment for indicated times. The mice were sacrificed at 1, 3 days or at 1, 2, 4, or 8 weeks of CCl₄ treatment (n = 7 per group).

Another group of adult ICR mice received BDL operation as described previously [25]. In brief, mice were anesthetized to receive a midline laparotomy to expose the common bile duct which was ligated three times. One ligature was located in the distal portion of the bile duct and another two were placed in the proximal portion. Then, the bile duct was cut between the ligatures followed by closing the abdomen in layers. Sham-operated mice underwent a laparotomy with exposure, but no performance of ligation of the common bile duct, which used as controls. Mice were sacrificed at 3 days or at 1, 2 weeks of BDL (n = 7 per group).

ICR mice received lethal irradiation (8 Grays), and then immediately received transplantation by a tail-vein injection of 1.5 × 10⁷ whole BM cells obtained from 3-week-old EGFP transgenic mice. Four weeks later, mice received intraperitoneal injections of CCl₄ twice per week for 4 weeks. 15d-PGJ₂ (0.3 mg/kg body weight) or saline was applied the day before CCl₄ or OO injection, and twice per week before CCl₄ or OO treatment for 4 weeks (n = 7 per group).

Liver tissues were fixed in 4% paraformaldehyde for 24 hours and embedded in paraffin. Samples were cut into 5 µm slices and sections were stained for collagen visualization with Sirius Red.

BMSCs with or without treatments were fixed in 4% paraformaldehyde in PBS for 30 minutes. Then cells were washed twice with PBS, permeabilized in 0.5% Triton X-100 for 15 minutes, blocked with 2% BSA for 1 hour, and incubated with antibodies to α-SMA (1:1000), Col a1(I) (1:100), Col a1(III) (1:200) and PPARγ (1:100) overnight, followed by secondary antibody conjugated with Cy3 (1:500, Jackson Immunoresearch Laboratories, West Grove, PA) for 1 hour. Cells were washed twice with PBS after incubation with PBS containing 10µg/mL DAPI for 5 minutes, and 200 µL PBS were left in each well. For negative controls, cells were processed the same way, except the primary antibody was omitted. The plates were imaged on a Thermo Scientific Cell Insight personal cell imaging (PCI) platform (Cellomics, Inc., Thermo Fisher Scientific Inc., Waltham, MA), with a ×10 objective using the Thermo Scientific Cellomics iDEV Software. High content analysis was performed as described previously [26]. In brief, at least 3,000 cells in thirty-six fields were automatically acquired by the software. The fluorescence intensity in each well was analyzed using Cellomics Cell Health Profiling BioApplication software.

Immunofluorescent detection for α-SMA in liver tissue was performed with a Vector M.O.M. (mouse-on-mouse) immunodetection kit (Vector Laboratories) [9]. Prepared liver sections were 5 µm and antibody to α-SMA was a 1:1000 dilution.

Western blot analysis of α-SMA, Col α1(I), Col α1(III) and PPARγ in BMSCs or liver tissue were performed with 50 µg of protein extract using primary antibodies of α-SMA (1:1000), Col α1(I) (1:100), Col α1(III) (1:200) and PPARγ (1:500). The bands were displayed using ODYSSEY and quantified by Odyssey v3.0 software. Results were normalized relative to the GAPDH (1:1000) or tubulin (1:1000) expression to correct for variations in protein loading and transfer.

The siRNA sequence specifically targeting mouse PPARγ was synthesized by Dharmacon: L-040712-00-0005 (Thermo Scientific, Lafayette, CO). 40% to 50% confluent BMSCs were prepared. Transient transfection of siRNA (40 nmol/L) was performed by using Invitrogen Lipofectamine RNAiMAX, as recommended by the manufacturer. Control cells were treated with 40 nmol/L RNAi Negative Control Duplexes (scramble siRNA). Cells were collected to perform real-time RT-PCR and Western blot analysis after 48 hours.

Total RNA was extracted from frozen liver specimens or cultured BMSCs with or without treatments, using an RNeasy kit (Qiagen, Hilden, Germany). Real-time RT-PCR was performed with an ABI Prism 7300 sequence-detecting system (Life Technologies, Foster City, CA), as described previously [6]. Primers (MWG Biotech, Ebersberg, Germany) used for real-time RT-PCR were as follows: 18s rRNA, sense, 5'-GTA ACC CGT TGA ACC CCA TT-3', and anti-sense, 5'-CCA TCC AAT CGG TAG TAG CG-3'; α-SMA: sense, 5'-ATG CTC CCA GGG CTG TTT T-3', and anti-sense, 5'-TTC CAA CCATTA CTC CCT GAT GT-3'; Col α1(I): sense, 5'-AGG GCG AGT GCT GTG CTT T-3', and anti-sense, 5'-CCC TCG ACT CCT ACA TCT TCT GA-3'; Col α1(III): sense, 5'-TGA AAC CCC AGC AAA ACA AAA-3', and anti-sense, 5'-TCA CTT GCA CTG GTT GAT AAG ATT AA-3'; TGF-β1: sense, 5'-TGC GCT TGC AGA GAT TAA AA-3', and anti-sense, 5'-TCA CTG GAG TTG TAC GGC AG-3'; PPARγ: sense, 5'-GCC CAC CAA CTT CGG AAT C-3', and anti-sense, 5'-TGC GAG TGG TCT TCC ATC AC-3'.

All results were confirmed in three independent experiments. Data are expressed as mean ± SEM and were analyzed by Student's t-test or ANOVA for analysis of variance. Statistical significance was defined as P < 0.05.

There are recent reports that expression of PPARγ is decreased in fibrogenesis [13,27,28]. We assessed PPARγ content in liver tissue of mouse chronic liver fibrosis models induced by CCl₄ administration or BDL operation. The results showed that PPARγ expression in both mRNA and protein levels were markedly reduced in CCl₄ administration group compared with OO-treated group (Fig. 1A and B). These findings were consistent with the results in BDL-induced liver fibrogenesis (Fig. 1C and D).

The expressions of main ECM, Col α1 (I) and Col α1 (III), were up-regulated in mouse liver fibrogenesis [29]. Here, we undertook correlation analysis between PPARγ and the two molecules in liver tissue. Results presented in Fig. 1E and F demonstrated that the mRNA level of PPARγ had negative correlations with Col α1 (I) and Col α1 (III), respectively (correlation coefficient r = -0.503, P =0.024; r = -0.525, P =0.025, respectively). In addition, the expression of α-SMA, a marker of myofibroblasts, which produces ECM predominantly, was obviously increased as well after liver injury [29]. We also found a negative correlation between the mRNA levels of PPARγ and α-SMA (correlation coefficient r = -0.505, P = 0.039) (Fig. 1G). Furthermore, there was a significant negative correlation between mRNA levels of PPARγ and TGF-β1, a key profibrotic cytokine and contributing to BMSC differentiation toward myofibroblasts in fibrotic liver [9] (correlation coefficient r = -0.610, P = 0.009) (Fig. 1H). Collectively, these results suggested PPARγ might play an essential role in liver fibrogenesis.

In addition to representing the extent of liver fibrosis, presence of α-SMA and increased levels of Col α1 (I) and Col α1 (III) also denote the differentiation of BMSCs toward myofibroblasts [9]. Given that PPARγ was negatively correlated with these molecules and TGF-β1in liver tissue, we then sought to determine the effects of PPARγ on the differentiation of BMSCs. In this regard, we explored the abundance of PPARγ in the context of BMSC differentiation toward myofibroblasts in vitro. Immunofluorescent staining showed that the degree of positive staining for PPARγ was much weakened when BMSCs were induced differentiation toward myofibroblasts by TGF-β1 (Fig. 2A). Furthermore, the mRNA level of PPARγ was strikingly decreased during differentiation of BMSCs trigged by TGF-β1 in a dose-dependent manner (Fig. 2B). In parallel, PPARγ expression in protein level was attenuated, too (Fig. 2C). It was noteworthy that there were also inverse correlations between mRNA expressions of PPARγ and α-SMA, Col α1 (I) or Col α1 (III) in BMSCs (Fig. 2D-F) (correlation coefficient r = -0.691, P = 0.000046; r = -0.792, P =0.000001; r = -0.737, P = 0.000008, respectively). These observations highlight the link between PPARγ and the differentiation of BMSCs toward myofibroblasts.

To further characterize the function of PPARγ in BMSC differentiation, we used an endogenous ligand (15d-PGJ₂) and two synthetic ligands (troglitazone and ciglitazone) of PPARγ. As illustrated in Fig. 3A, α-SMA mRNA in differentiated BMSCs triggered by TGF-β1 was obviously diminished when pretreatment with either natural or synthetic ligands of PPARγ. In accord, α-SMA protein expression induced by TGF-β1 was down-regulated by 15d-PGJ₂, troglitazone or ciglitazone (Fig. 3B and C). Furthermore, siRNA technology was performed to knock-down of PPARγ. Transfection of BMSCs with PPARγ-siRNA decreased the mRNA expression of PPARγ by 70% (Fig. 3D). Meanwhile, Western blot analysis indicated PPARγ protein expression was also significantly reduced after knock-down of PPARγ (Fig. 3E). The results showed that PPARγ knock-down enhanced α-SMA mRNA expression, whereas scrambled siRNA (SCR siRNA) had no effects (Fig. 3F).

To further confirm the function of PPARγ in BMSC differentiation, we performed immunofluorescent staining for α-SMA to reveal morphological changes in BMSCs. Isolated BMSCs showed very weak expression of α-SMA under normal condition. TGF-β1 treatment resulted in the appearance of numerous bundles of actin microfilaments and dramatically increased the expression of α-SMA in BMSCs (Fig. 3H). However, when BMSCs were pretreated with ligands of PPARγ, the ability of TGF-β1 to enhance the expression of α-SMA was markedly blunted (Fig. 3H). Additionally, the total fluorescence intensity of α-SMA examined by high content analysis demonstrated that exposure to 15d-PGJ₂, troglitazone or ciglitazone resulted in reducing α-SMA protein level in differentiated BMSCs initiated by TGF-β1 (Fig. 3G).

Myofibroblasts are the predominant collagen-producing cells in liver fibrosis. Up-regulation of Col α1 (I) and Col α1 (III) is an essential event during BMSC differentiate toward myofibroblasts. Therefore, we sought to characterize the effects of PPARγ on the expression of Col α1 (I) and Col α1 (III) in BMSCs. As expected, all three PPARγ agonists, 15d-PGJ₂, troglitazone or ciglitazone blunted the increases of Col α1(I) and Col α1(III) mRNA expression in BMSCs treated with TGF-β1 (Fig. 4A and 5A). Western blot analysis exhibited similar results in protein level (Fig. 4B and 5B). In accord with α-SMA, mRNA expressions of Col α1 (I) and Col α1 (III) were also increased in PPARγ-knock-down cells (Fig. 4C and 5C).

Furthermore, immunofluorescence was also performed to study the effects of PPARγ on the protein expressions of Col α1 (I) and Col α1 (III) in BMSCs. The results showed that there were strong immunoreactivities for Col α1 (I) and Col α1 (III) in BMSCs after treatment with TGF-β1 (Fig. 4E and 5E), and collagen that was secreted into the medium was also detected by immunofluorescence (Fig. 4E and 5E, arrow). Manipulation of PPARγ ligands reduced the degree of positive staining for Col α1 (I) and Col α1 (III) induced by TGF-β1 (Fig. 4E and 5E). High content analysis indicated that the increased total fluorescence intensities of Col α1 (I) and Col α1 (III) were attenuated by 15d-PGJ₂, roglitazone or ciglitazone (Fig. 4D and 5D). Altogether, these findings showed that PPARγ negatively governed BMSCs differentiate toward myofibroblasts.

During liver injury, myofibroblasts are increased and its marker, α-SMA, is up-regulated to contribute to liver fibrogenesis [29]. In the present study, we found that treatment with the natural ligand of PPARγ, 15d-PGJ₂, reduced α-SMA mRNA in liver fibrosis induced by different periods of CCl₄ injection (Fig. 6A). Given that myofibroblasts are mostly BMSC-derived, we further determinedthe effect of PPARγ on BMSC differentiation toward myofibroblasts in vivo BM in the irradiated mice was reconstituted by transplanting BM cells from EGFP transgenic mice, and then the proportion of BMSC-derived myofibroblasts was measured. As shown in Fig. 6C, after CCl₄ treatment, there was a strong expression of α-SMA colocalizing with EGFP in the fibrotic areas, which indicated they were BMSC origin. However, 15d-PGJ₂ significantly inhibited the differentiation of BMSCs toward myofibroblasts in liver fibrogenesis (Fig. 6D), and markedly decreased the proportion of BMSC-derived myofibroblasts compared with CCl₄ group without 15d-PGJ₂ treatment (Fig. 6B). These results indicated that PPARγ activation could suppress BMSCs differentiate toward myofibroblasts in vivo.

Col α1 (I) and Col α1 (III) are the main ECM in fibrotic liver tissue [1,2]. After injection of CCl₄, there were dynamic changes in mRNA expressions of Col α1 (I) and Col α1 (III) in liver tissue, which markedly up-regulated in a way correlated with the progression of liver fibrosis (Fig. 7A). As expected, PPARγ activation by 15d-PGJ₂ markedly decreased these mRNA levels (Fig. 7A). In addition, hepatic collagen deposition was further evaluated by morphometric analysis with Sirius red staining. As shown in Fig. 7B, collagen deposition was markedly attenuated after 15d-PGJ₂ administration. These data suggest activation of PPARγ had anti-fibrotic properties.

Liver fibrosis is a common pathological course of chronic hepatic diseases. So far, there is no effective and safe strategy for the prevention or inhibition of fibrogenesis in clinical practice. Differentiation of BMSCs toward myofibroblasts is an essential event for liver fibrogenesis, thus, elucidating the regulatory mechanisms in this process may shed light on the treatment of liver fibrosis. In the current study, we delineated the potential role of PPARγ in the differentiation of primary mouse BMSCs toward myofibroblasts. We found PPARγ negatively regulated BMSC differentiation toward myofibroblasts in liver fibrogenesis. Expression of PPARγ was reduced in fibrotic liver tissue. In addition, differentiation of BMSCs was accompanied by the decrease of PPARγ expression. Activated PPARγ by specific ligands could attenuate BMSC differentiation in vitro and in vivo, and alleviate liver fibrosis. These results critically revealed a mechanism underlying differentiation of BMSCs toward myofibroblasts.

Accumulating evidence suggests that levels of PPARγ were dramatically reduced in the process of fibrosis [13,27,28,30,31], whereas overexpression of PPARγ could ameliorate the degree of fibrosis [32]. In the present study, the abundance of PPARγ was evidently reduced in fibrogenesis, similar to the reports in rats that PPARγ expression was down-regulated in fibrotic liver tissue induced by 2 or 8 weeks of CCl₄ combined with ethyl alcohol treatment or caused by 10 days of BDL operation [28,33,34]. In addition, mRNA expressions of PPARγ and those fibrotic related molecules, such as TGF-β1, α-SMA, Col α1 (I) and Col α1 (III), were negative correlations. Intriguingly, PPARγ and plasminogen activator inhibitor-1 (PAI-1), associated with fibrosis and implicated in its pathogenesis, were inversely correlated in skin [13]. In heart, PPARγ was also negatively correlated with tenascin-x, which was exclusively expressed in fibroblasts that can mediate fibrosis [35]. These results emphasize the importance of PPARγ in the pathogenesis of fibrosis in different tissues.

Several studies reveal that PPARγ was down-regulated in the differentiation of sorts of fibroblasts and myocytes toward myofibroblasts [11,13,36]. In our study, expression of PPARγ was remarkably decreased during BMSC differentiation toward myofibroblasts induced by TGF-β1, similar to our previous report that expression of PPARγ almost couldn't be detected in human hepatic myofibroblasts [37]. In addition, the down-regulation of PPARγ by specific siRNA had favorable effect on the differentiation of BMSCs. In accordance, enhanced myofibroblast formation potential was observed in PPARγ-deficient mouse fibroblasts [38]. BMSCs are multipotent cells with the potential to differentiate toward a variety of mesenchymal cells, such as adipocytes [39]. The decreased expression of PPARγ, a transcription factor that related to fat-formation, might make BMSCs lost potential to differentiate toward adipocytes, but contribute to the differentiation toward myofibroblasts in the present study. This hypothesis is in agreement with the results in HSCs, another type fat-storing cell, that significant reduced expression of PPARγ was paralleled by activation of HSCs [32,40,41,42]. It has been reported that the necdin-wnt pathway and a combination of methyl-CpG binding protein 2 (MeCP2), enhancer of zeste homolog 2 (EZH2) and miR132 cause epigenetic PPARγ repression and result in HSC activation [43,44]. In addition, PPARγ expression could be repressed by TGF-β1 through increasing the binding of histone deacetylase 1 (HDAC1) and decreasing the levels of acetylated histone3 (AcH3) at the PPARγ promoter or via smad binding elements in the PPARγ gene promoter to negatively regulate the promoter activity of PPARγ gene in cardiac fibroblasts or HSCs [11,24]. However, the precise regulatory mechanism of PPARγ expression in BMSCs is scarce and should be valued in further research.

PPARγ needs to be activated by ligands to exert effects. Results presented here showed that either natural or synthetic ligands of PPARγ could suppress BMSC differentiate toward myofibroblasts induced by TGF-β1 in vitro, with alleviation of α-SMA, Col α1 (I) and Col α1 (III) expression. More importantly, we found activation of PPARγ by natural ligand could inhibit the differentiation of BMSCs toward myofibroblasts in liver fibrogenesis and decreased the proportion of BMSC-derived myofibroblasts. These results provide strong evidence for that activated PPARγ can negatively regulate BMSC differentiate toward myofibroblasts. In particular, serum concentrations of 15d-PGJ₂ in Caucasian patients with ongoing hepatic fibrogenesis and Chinese patients with hepatocellular carcinoma are elevated, which suggests that increased 15d-PGJ₂ might trigger PPARγ activity to attempt to execute its beneficial effects [45]. Additionally, PPARγ activation by agonists attenuates skin and renal fibroblast differentiation [12,46] and inhibits HSC activation [32,42]. It is noteworthy that effects of activated PPARγ by different agonists may have diverse and complex roles in regulating cell differentiation. For instance, Da et al. [47] reported that PPARγ ligand, pioglitazone, did not suppress cultured HSC activation by assessment expression of α-SMA and Col α1(I).

Our previous result has indicated that sphingosine kinase/S1P/S1P receptor axis participates in the differentiation of BMSCs toward myofibroblasts [9]. In the present study, we found PPARγ negatively regulates BMSC differentiation toward myofibroblasts. There are several reports have linked PPARγ with S1P signaling pathway. In renal mesangial cells, synthetic PPARγ agonists thiazolidinediones up-regulate sphingosine kinase and intracellular S1P resulting in decrease expression of CTGF to causes an anti-fibrotic effect [48]. In addition, thiazolidinediones could increase the abundance of S1P receptor [49]. In turn, PPARγ is an intracellular target for S1P and S1P could activate PPARγ coactivator 1α (PGC-1α) [50,51]. These findings might partially explain the effect of PPARγ on BMSC differentiation toward myofibroblasts and its inhibitory function in liver fibrosis. However, the relationship between PPARγ and S1P pathway remains controversial. In 3T3-L1 preadipocytes, S1P down-regulates the expression of PPARγ [52], which reflects the complexity of signaling cascades in a biological system.

An importantissue of PPARγ is that PPARγ carries outits function requiring translocation from cytoplasm to nucleus, such as in the study of PPARγ governs the differentiation of skin fibroblasts [46]. Besides regulating target gene expression directly, PPARγ could interact with other cell signaling pathways, including fibrotic-related events in cells. It has been reported that PPARγ is capable of binding to Smad3 and p-Smad3, and prevents nuclear accumulation of p-Smad3, resulting in disruption of TGF-β1 signaling [53]. Activated PPARγ by rosiglitazone reduced the induction of the early-immediate transcription factor Egr-1, a mediator of non-Smad TGF-β1 signaling, to exhibitits antifibrotic function [54]. Furthermore, albumin may be a downstream effector of PPARγ in phenotypic switch of pancreatic stellate cells (PSCs) [55], and PPARγ activation might ameliorate renal fibrotic lesions via hepatocyte growth factor (HGF) [56]. However, whether these molecular mechanisms mediate the effects of PPARγ on BMSC differentiation toward myofibroblasts need to be further explored.

In summary, our results demonstrate that PPARγ is a negative regulator in BMSC differentiation toward myofibroblasts in liver fibrogenesis. Decrease in PPARγ expression is accompanied by the differentiation of BMSCs toward myofibroblasts, and activated PPARγ by agonists inhibits BMSC differentiation in vitro and in vivo, and causes an attenuation of liver fibrosis. Consequently, these findings expand our knowledge on the regulatory mechanisms of BMSC differentiation, and make PPARγ to be a promising target for treatment of liver fibrotic diseases.

PPARγ (peroxisome proliferator-activated receptor gamma); BMSCs (bone marrow-derived mesenchymal stem cells); α-SMA (α-smooth muscle actin); 15d-PGJ₂ (15-Deoxy-Δ^12,14-prostaglandin J₂); TGF-β1 (transforming growth factor β1).

This work was supported by grants from the National Natural and Science Foundation of China (81430013, 81300335) and the Project of Construction of Innovative Teams and Teacher Career Development for Universities and Colleges under Beijing Municipality (IDHT2015502).

The authors declared that they have no conflicts of interest related to this work.