Oral Methylthioadenosine Administration Attenuates Fibrosis and Chronic Liver Disease Progression in Mdr2−/− Mice

Background Inflammation and fibrogenesis are directly related to chronic liver disease progression, including hepatocellular carcinoma (HCC) development. Currently there are few therapeutic options available to inhibit liver fibrosis. We have evaluated the hepatoprotective and anti-fibrotic potential of orally-administered 5′-methylthioadenosine (MTA) in Mdr2−/− mice, a clinically relevant model of sclerosing cholangitis and spontaneous biliary fibrosis, followed at later stages by HCC development. Methodology MTA was administered daily by gavage to wild type and Mdr2−/− mice for three weeks. MTA anti-inflammatory and anti-fibrotic effects and potential mechanisms of action were examined in the liver of Mdr2−/− mice with ongoing fibrogenesis and in cultured liver fibrogenic cells (myofibroblasts). Principal Findings MTA treatment reduced hepatomegaly and liver injury. α-Smooth muscle actin immunoreactivity and collagen deposition were also significantly decreased. Inflammatory infiltrate, the expression of the cytokines IL6 and Mcp-1, pro-fibrogenic factors like TGFβ2 and tenascin-C, as well as pro-fibrogenic intracellular signalling pathways were reduced by MTA in vivo. MTA inhibited the activation and proliferation of isolated myofibroblasts and down-regulated cyclin D1 gene expression at the transcriptional level. The expression of JunD, a key transcription factor in liver fibrogenesis, was also reduced by MTA in activated myofibroblasts. Conclusions/Significance Oral MTA administration was well tolerated and proved its efficacy in reducing liver inflammation and fibrosis. MTA may have multiple molecular and cellular targets. These include the inhibition of inflammatory and pro-fibrogenic cytokines, as well as the attenuation of myofibroblast activation and proliferation. Downregulation of JunD and cyclin D1 expression in myofibroblasts may be important regarding the mechanism of action of MTA. This compound could be a good candidate to be tested for the treatment of (biliary) liver fibrosis.


Introduction
Liver diseases are currently the fifth cause of mortality in the Western world, and as opposed to other major causes of mortality their incidence is increasing [1,2]. The principal causes of the most common liver ailments are well known, and include chronic alcohol consumption, infection by hepatitis viruses, genetic conditions such as hemochromatosis or a1-antitrypsin deficiency, autoimmune hepatitis and metabolic disorders [1,2]. Regardless of the etiology, chronic tissue injury and inflammation are increasingly recognized as driver mechanisms in liver disease progression from fibrosis, characterized by the excessive accumulation of extracellular matrix (ECM), to hepatocellular carcinoma (HCC) [1][2][3]. The molecular links in the so-called inflammation-fibrosis-cancer axis in the liver are currently being elucidated in experimental models of acute and chronic injury. These links include a variety of intracellular pathways triggered by extracellular mediators like the cytokines interleukin 1 (IL1), IL6 and tumour necrosis factor alpha (TNFa), platelet-derived growth factors and transforming growth factor beta (TGFb), among others [3][4][5]. These growth factors and cytokines stimulate ECM synthesis by the activation of hepatic stellate cells and myofibroblasts in the hepatic parenchyma [3][4][5]. Targeted interference with these and other mediators has clearly shown a beneficial effect on the course of the experimental disease, lending support to their pathological role [6,7]. However, in spite of these remarkable advances the availability of safe and efficacious therapies to halt the progression of fibrosis and liver disease in humans is still limited [8,9]. The use of experimental models of chronic liver injury resembling the human pathology is important to evaluate candidate drugs with chances of succeeding in the clinical setting. One such experimental model are the Mdr2/Abcb4-deficient mice (Mdr2 2/2 ), which lack the canalicular phosphatidylcholine flippase [10]. The absence of phosphatidylcholine from bile occurring in these mice leads to bile regurgitation into the portal tracts, causing periportal inflammation and injury early in life (2-3 weeks), periportal fibrosis (4 weeks), and the appearance of preneoplastic lesions (at 4-6 months) [10][11][12]. These pathogenic characteristics resemble what occurs in human primary sclerosing cholangitis and biliary fibrosis, making these mice an excellent model to study disease mechanisms, and a test ground for hepatoprotective and antifibrotic therapies [10][11][12][13]. In analogy to the situation found in humans, protracted injury and inflammation lead to HCC development in virtually 100% of Mdr2 2/2 mice by 16 months [14].
We and others have previously demonstrated the in vivo antiinflammatory effects of 59-methylthioadenosine (MTA), a sulphurcontaining adenine nucleoside produced from S-adenosylmethionine (SAMe) during polyamine biosynthesis [15]. Parenteral MTA administration reduces the production of inflammatory mediators triggered by bacterial lipopolysaccharide or pro-inflammatory peptides in mice [16][17][18], and protects from liver injury in rats treated with CCl 4 or with the carcinogen diethylnitrosamine [19,20]. MTA is therefore a compound with an attractive pharmacological profile for the treatment of chronic liver disease. Using the Mdr2 2/2 mice as a model, here we show for the first time that orally administered MTA attenuates liver injury and inflammation, and inhibits the progression of hepatic fibrosis. The mechanism of action of MTA is likely to be multifaceted. MTA reduced the production of inflammatory and pro-fibrogenic mediators and inhibited the proliferation and activation of fibrogenic cells. These observations gathered in the Mdr2 2/2 mouse model indicate that MTA could be useful for the treatment of biliary fibrosis and conditions such as primary sclerosing cholangitis, a severe disease frequently associated with hepatobiliary cancer and with limited therapeutic options.

Ethics statement
Animals received humane care and study protocols complied with our institution's guidelines and the recommendations of the European Accreditation of Laboratory Animal Care. The protocol was approved by the Committee on the Ethics of Animal Experiments of the University of Navarra (protocol number: 025-10).

Animal studies
Male Mdr2 2/2 and Mdr2 +/+ mice (The Jackson Laboratory, Bar Harbor, ME) were fed standard laboratory diet. Two groups of mice per genotype were established (2 months old, n = 8 per group). One group received a daily oral dose of 30 mg/Kg body weight of MTA dissolved in 0.2% dimethylsulfoxide by gavage for three weeks, the control group received the same volume of vehicle (0.2% dimethylsulfoxide). MTA was from Enantia S.L. (Barcelona, Spain). This dose of MTA has been used previously, but was administered intraperitoneally [16,17]. Liver enzymes and bilirrubin levels were analyzed in serum (CobasH analyzer, Roche). Livers were excised, weighted, and were either snap frozen or fixed in formalin and paraffin-embedded.

Metabolite measurement in liver tissues
Glutathione (GSH) was measured using the Gluthatione Assay Kit (Sigma, St. Louis, USA). Tissue samples were processed as recommended by the manufacturer. SAMe was measured as previously described [21].

Tissue staining and immunohistochemistry
Tissue sections were stained with Picro-Sirius Red to visualize collagen deposition as reported [22]. For morphometric analysis images were captured at 100X magnification (Nikon Eclipse 1000), and percentage of collagen cover per field was estimated with an Arkon software (Arkon Resources Inc. Arcadia, CA) as described [22]. Values are the means of five fields taken from different tissue sections per mouse. Immunodetection of a-smooth muscle actin (aSMA) on liver sections was carried out as described using a monoclonal anti-aSMA antibody (#A2547) from Sigma (St. Louis, MO) [22]. Tenascin-C and Ki-67 immunostainings were performed using rabbit polyclonal antibodies from Millipore (#AB1951) (Millipore Iberica, Madrid, Spain) and Abcam (#ab833) (Cambridge, UK) respectively. Immunostaining for CD45 was performed with a rat-anti-mouse CD45 antibody (clone 30-F11, #103101) from BioLegend (San Diego, CA).

Cell culture and treatments
Liver myofibroblasts were isolated from wild type and Mdr22/2 mice by collagenase perfusion and density-gradient purification in Nycodenz, and cultured in Dulbecco's Modified Eagle's medium (DMEM) containing 10% fetal calf serum (FCS) as described [23]. For [ 3 H]thymidine incorporation myofibroblasts were incubated under the specified conditions for 24 h with 2 mCi/ml [ 3 H]thymidine. Cells were lysed and radioactivity measured as described [22]. Human recombinant platelet derived growth factor BB (PDGF) was from Calbiochem (#521225) (Merck KGaA, Darmstadt, Germany). MTA showed no cytotoxic effects, as determined by measuring lactate-dehydrogenase in culture media using the Cytotox assay (Promega, Madison, WI). Apoptosis was measured in cultured myofibroblasts using the Cell Death Detection Assay from Roche (Barcelona, Spain) as previously described [22].

Western blot analysis
Tissues and cells were homogenized and analyzed by Western blotting as described [22]. The antibodies used were: anti-cyclin-

Statistical analysis
Data are means 6SEM. Analyses were performed using GraphPad Prism version 5.00 (GraphPad Software, San Diego, USA). Data were compared among groups using the Student t test. A P value of ,0.05 was considered significant.

Results
MTA treatment reduces liver injury in Mdr2 2/2 mice By three months of age Mdr2 2/2 mice typically display hepatomegaly and elevated serum levels of hepatic enzymes  [10,11,14]. MTA administration was well tolerated, and treated animals gained similar weight as controls. MTA reduced the liver to body weight ratio (Fig. 1A), and improved aspartate aminotransferase (AST) and alanino aminotransferase (ALT) serum levels (Fig. 1B). Circulating levels of alkaline phosphatase (AP) and bilirubin were also lowered by MTA ( Fig. 1C and 1D).
MTA treatment attenuates liver fibrosis in Mdr2 2/2 mice MTA significantly reduced the periductal fibrosis that spontaneously develops in these mice [11], as determined by Sirius Red staining of crosslinked collagen in liver sections and morphometric quantitation of stained areas ( Fig. 2A and B). MTA administration to Mdr2 +/+ mice had no effect on collagen levels (not shown). As previously described [11], we found an increase in periductal aSMA-positive cells and aSMA mRNA levels, which were also reduced by MTA treatment (Fig. 3A and B). In accordance with the histology, the expression of a1(I)procollagen mRNA was reduced in MTA treated mice (Fig. 4A). As reported [11], the expression of the interstitial collagenase MMP13, and that of its inhibitor TIMP1, was significantly increased in Mdr2 2/2 mice ( Fig. 4B and C). MTA reduced MMP13 expression, and although not statistically significant there was a tendency towards decreased TIMP1 mRNA levels ( Fig. 4B and C).
MTA reduces hepatic levels of inflammatory and profibrogenic mediators in Mdr2 2/2 mice Although the expression of inflammatory cytokines in the liver of Mdr2 2/2 mice is maximal between 2-8 weeks [10,11], we still detected elevated TNFa and IL6 mRNA levels ( Fig. 5A and B). MTA tended to reduce TNFa mRNA, and significantly lowered IL6 expression ( Fig. 5A and B). Consistently, the expression of iNOS gene, a transcriptional target of pro-inflammatory cytokines, was decreased by MTA (Fig. 5C). Importantly, upon MTA administration we observed reduced Mcp-1 expression (Fig. 5D), a pro-fibrogenic chemokine up-regulated in Mdr2 2/2 mice [12]. TGFb1 and TGFb2 expression is elevated in the liver of these animals, particularly during the first 8 weeks of life [10][11][12]. Accordingly we found that by 11 weeks TGFb1 expression in Mdr2 2/2 mice was similar to wild types, while TGFb2 levels were still high, and were attenuated by MTA ( Fig. 5E and F). These findings were accompanied by a significant decrease in inflammatory infiltrate, as assessed by staining for the leukocyte marker CD45 (Fig. 6).
Tenascin-C is a TGFb-inducible extracellular protein expressed in liver fibrogenic cells that participates in inflammation and fibrogenesis [24,26]. As described in other models of chronic liver injury [24], we found increased tenascin-C staining around portal areas in Mdr2 2/2 mice (Fig. 7A). MTA treatment reduced tenascin-C immunostaining and mRNA levels ( Fig. 7A and B). However, the expression of avb6 integrin, which is induced in Mdr2 2/2 mice biliary epithelia and binds tenascin-C [13], was unaffected by MTA (not shown).
Effect of MTA on signalling pathways involved in fibrosis and hepatocellular proliferation in the Mdr2 2/2 mouse liver To further explore the mechanism of action of MTA we examined the activation of signalling pathways associated with fibrogenesis and disease progression, including the Smad and mitogen-activated protein kinase (MAPK) pathways [3,6,9,[27][28][29][30]. We found that total and phosphorylated Smad1/5/8 and Smad2 levels were increased in Mdr2 2/2 mouse liver extracts, and that MTA treatment reduced Smad2 and p-Smad2 contents ( Fig. 8A and B). We also observed increased MAPK pathway activity in Mdr2 2/2 mice, as evidenced by elevated levels of p-Erk1/2, p-JNK and p-c-Jun. These pathways were also inhibited by MTA ( Fig. 8C and D). No changes were observed in MTA-treated Mdr2 +/+ mice (not shown).
In agreement with the hyperactivation of pro-mitogenic signalling pathways, and confirming previous reports [6,30], we  found enhanced hepatocellular proliferation in Mdr2 2/2 mice, as evidenced by increased hepatocyte staining with the proliferation marker Ki-67 ( Fig. 9A and B). MTA reduced hepatocellular proliferation ( Fig. 9A and B) and cyclin D1 protein levels (Fig. 9C), markedly elevated in the liver parenchymal cells of these animals [14].

MTA effects on liver SAMe and GSH levels in Mdr2 2/2 mice
In vivo MTA is produced from SAMe during polyamine metabolism, and subsequently MTA can be metabolically recycled back to methionine [15,31]. Methionine metabolism in turn is necessary for the production of SAMe and GSH, and it is impaired in liver injury [31]. The expression of key genes in methionine metabolism is down-regulated in Mdr2 2/2 mice [32]. In view of this we measured MTA effects on liver SAMe levels. We observed that SAMe was not reduced in Mdr2 2/2 mice, and was unaffected by MTA treatment (82.9610.6 in Mdr2 +/+ ; 110.669.7 in Mdr2 2/2 and 103.2612.3 in MTA-treated Mdr2 2/2 , values are nmol/g tissue). GSH levels were not altered either in Mdr2 2/2 mice compared to Mdr2 +/+ , and were not affected by MTA (6.4960.57 in Mdr2 +/+ ; 5.9260.68 in Mdr2 2/2 and 5.6160.51 in MTA-treated Mdr2 2/2 , values are mmol/g tissue).

Effects of MTA on inflammatory cells
Our current in vivo observations indicate that MTA administration significantly attenuated the inflammatory response found in the liver of Mdr22/2 mice. MTA is structurally very close to adenosine, a potent endogenous immune regulatory molecule produced mainly by cells of the immune system during tissue injury and inflammation [33]. Adenosine mediates its cellular effects through the interaction with cell surface receptors such as A2A and A2B, which are expressed in monocytes/macrophages and play a major role in the anti-inflammatory and immunomodulatory actions of this nucleoside [33,34]. In view of this and considering the ability of MTA to interact with adenosine receptors [35], we decided to explore the potential implication of A2A and A2B receptors in MTA anti-inflammatory effects. To directly address this issue we took advantage of the murine monocyte/macrophage cell line RAW 264.7, which has been widely used in mechanistic studies on the anti-inflammatory effects of adenosine upon bacterial lipopolysaccharide (LPS) challenge [36,37]. First we evaluated the expression and release of TNFa in LPS-stimulated RAW 264.7 cells in the presence of MTA. As shown in Fig. 10A, and in agreement with our previous findings [16], we found a potent inhibitory effect of MTA on TNFa expression. We also observed that the inhibitory effect of MTA was preserved when RAW 264.7 cells were incubated in the presence of the A2A or A2B specific antagonists ZM 241385 and MRS 1754 respectively (Fig. 10A and B). On the other hand, and in accordance with the literature [38,39], the inhibition of LPSelicited TNFa production by the stable adenosine analogue NECA was attenuated by the A2A antagonist ZM 241385 (Fig. 10A). These observations suggested that the effect of MTA on the activation of inflammatory cells was not dependent on its potential direct interaction with adenosine receptors. However, although apparently MTA was not acting as an A2B receptor agonist we found that this molecule dramatically synergized with LPS in the up-regulation of A2B mRNA levels (Fig. 10C). Increased A2B expression has been linked to the activation of anti-inflammatory responses [37,39,40]. Interestingly this effect of MTA was not reproduced by the adenosine analogue NECA (Fig. 10C).

Effect of MTA on cultured liver myofibroblasts
Next we tested whether MTA had a direct effect on liver myofibroblasts, which participate in biliary fibrosis and represent a relevant source of extracellular matrix in Mdr2 2/2 mice [9,30,41]. Given that isolated myofibroblasts display high proliferation and activity upon in vitro culture, we pre-incubated the cells in serum free medium for 12 h to attenuate activation and proliferation, and subsequently cultures were stimulated with 10% FCS to promote myofibroblast activation in the presence or absence of MTA. We observed that the expression of the fibrogenic markers a1(I)procollagen and aSMA elicited by FCS treatment was inhibited by MTA ( Fig. 11A and B). Similar observations were made when myofibroblasts were challenged with TGFb (not shown). Additionally, TGFb1 and TGFb2 expression triggered by FCS was also attenuated by MTA ( Fig. 11C and D). In agreement with previous reports [30] we found that FCS treatment increased Mcp-1 expression, and this response was also reduced by MTA (Fig. 11E), while IL6 expression was unaltered (Fig. 11F).
We also examined whether MTA could affect myofibroblast proliferation. Figure 12A shows that MTA reduced FCStriggered DNA synthesis, without causing any toxicity (determined by lactate dehydrogenase release, not shown). Additionally we observed that MTA did not induce apoptosis in cultured myofibroblasts (data not shown). Our [ 3 H]thymidine incorporation experiments suggested that MTA inhibited cell cycle progression through the S phase (DNA synthesis), therefore we examined the expression of cyclin D1, a key cyclin in the transition from G 0 /G 1 to S phase. We found that cyclin D1 expression was reduced by MTA treatment both at the mRNA and protein levels ( Figure 12B). The expression of cyclin D1 can be regulated by AP-1 transcription factors, and among the components of the AP-1 complex JunD has been identified as a critical transcription factor in liver fibrogenesis [27,42]. Therefore we examined the binding of JunD to a region of the cyclin D1 promoter that contains a key AP-1 binding site [25] by ChIP analysis. As depicted in Figure 12C, MTA treatment almost completely blunted JunD association with cyclin D1 promoter in this regulatory region. In view of this response we examined the expression of JunD in cultured myofibroblasts treated with MTA, and observed that MTA significantly reduced JunD protein levels (Fig. 12D). These data suggested that the reduction of cyclin D1 expression by MTA could be mediated in part through the downregulation of JunD levels. Nevertheless, in spite of this effect of MTA on JunD cellular contents this compound almost totally inhibited cyclin D1 expression, and our ChIP experiment showed a near complete inhibition of JunD binding to the cyclin D1 promoter. This suggested that MTA could also affect the interaction of the remaining JunD protein with this cyclin D1 regulatory region. MTA is a well known inhibitor of methylation reactions [15], and it has been shown to inhibit H3K4 trimethylation, a hallmark of open chromatin conformation and transcriptional activation [43], in different cell types [18,44]. In view of this we examined the levels of tri-methylated-H3K4 in chromatin associated with this region of the cyclin D1 promoter. We detected increased levels of tri-methylated-H3K4 upon FCS stimulation, and observed that this modification was lowered by MTA treatment (Fig. 12E).
The relevance of these in vitro effects of MTA on the expression of fibrogenic and cell proliferation-related genes was supported by the observation of reduced a1(I)procollagen, aSMA, tenascin-C, TGFb1, TGFb2, JunD and cyclin D1 expression in myofibroblasts isolated from Mdr22/2 mice treated for 3 weeks with MTA when compared to cells isolated from untreated mice (Fig.13).
Finally, to gain additional insight into the mechanisms of the antifibrogenic effects of MTA we explored the direct effects of this molecule on the activity of intracellular signalling pathways related to the proliferation and fibrogenic activation of myofibroblasts [27][28][29][30]. As depicted in Fig. 14A, while MTA treatment was not able to directly modify the phosphorylation of Erk1/2 elicited by FCS, it reduced the levels of phospho-c-Jun, a target of the profibrogenic JNK signalling pathway in liver extracellular matrix producing cells [29]. Interestingly, MTA significantly attenuated the phosphorylation of the mTOR and p70 S6 kinase downstream target S6 ribosomal protein triggered by the pro-fibrogenic growth factor PDGF (Fig. 14B). p70 S6 kinase activity is required for G 1 cell cycle progression, and it has been also identified as a signal transducer for liver fibrogenic cell growth [45].

Discussion
We have shown that oral administration of the nucleoside MTA significantly ameliorates liver injury in Mdr2 2/2 mice. These mice represent a clinically relevant model of spontaneous chronic liver disease, spanning from peribiliary inflammation and fibrosis to HCC development [9][10][11][12][13][14]. MTA was well tolerated and a significant improvement in serum markers of hepatic injury and function was observed. This was accompanied by reduced hepatic fibrosis and fibrogenic cell abundance. Accordingly, a1(I)procollagen mRNA levels were lowered, and TIMP1 expression tended to decrease. The fact that MTA also reduced the expression of MMP13, the interstitial collagenase in rodents, suggests that MTA down-regulates overall tissue remodelling activity, inhibiting fibrosis progression rather than stimulating fibrosis resolution [3].
Previous studies have shown that MTA exerts remarkable biological effects, its pharmacologic administration modulates different pathways and cellular responses, including gene expression, cell cycle progression and inflammation [15][16][17][18][19][20]. Therefore, the mechanisms behind MTA therapeutic action on a complex model such as Mdr2 2/2 mice are likely multifaceted. Our current findings agree with those reported by Simile et al., who showed that parenteral MTA administration to rats undergoing chronic CCl 4 -induced liver injury reduced fibrosis [19]. This effect was mechanistically related to MTA antioxidant properties, that counteracted the oxidative stress and subsequent injury generated by CCl 4 metabolism [19]. However, oxidative stress is not a prominent feature in Mdr2 2/2 mice [14]. Accordingly, we found that GSH levels were unaltered in Mdr2 2/2 mouse liver, and were not affected by MTA, suggesting that MTA effects were not attributable to its antioxidant potential. In vivo MTA is metabolized to methionine in the so called methionine salvage pathway, and in the liver methionine is efficiently metabolized into SAMe [15]. SAMe is a key methyl donor with significant hepatoprotective properties, including antifibrotic activity in experimental cholestasis [31,46]. In view of the metabolic links between MTA and SAMe, we measured SAMe levels in control and MTA-treated Mdr2 2/2 mouse liver but did not observe major differences, indicating that MTA effects were not likely mediated through its potential contribution to the SAMe pool. Moreover, this may suggest that the antifibrotic effects of SAMe in experimental cholestasis [46] could be mediated in part by the metabolic or nonenzymatic conversion of SAMe into MTA [15].
Interestingly, MTA reduced the expression of the extracellular matrix protein tenascin-C in the liver of Mdr2 2/2 animals. This may be directly related to MTA antifibrotic effects, since tenascin-C is required for the development of experimental liver fibrosis [24]. Furthermore, tenascin-C is up-regulated in chronic hepatitis C, and its expression in peritumoral activated stellate cells in HCC associates with poor clinical prognosis [26,38,47]. In view of this, tenascin-C may be a previously unrecognized player in fibrosis and HCC development in Mdr2 2/2 mice, and its attenuation by MTA a key event in the anti-fibrotic and anti-neoplastic properties of this molecule [20]. Moreover, tenascin-C binds and activates avb6 integrin, a recently identified pharmacological target in biliary fibrosis [13]. As occurs in human chronic liver disease, progression of liver injury in Mdr2 2/2 mice is associated with persistent inflammation. The observation that anti-inflammatory compounds like ibuprofen or curcumin ameliorate injury and fibrosis in Mdr2 2/2 mice [6,30], and that TNFa blockade prevents hepatocarcinogenesis in these animals [6], underscore the role played by inflammation in this model. We and others have previously shown that MTA has a significant immunomodulatory potential. MTA inhibits NF-kB mediated signalling, reduces inflammatory cytokine production by macrophages and lymphocytes, and attenuates the cellular effects of these cytokines [16][17][18]. We now observed that in Mdr2 2/2 mice MTA reduced the expression of IL6, Mcp-1, the NOgenerating enzyme iNOS, and the profibrogenic factor TGFb2. Accordingly MTA attenuated the activation of Erk1/2, JNK and Smad2, intracellular pathways triggered by these cytokines and which are directly implicated in fibrogenesis [3,5,[27][28][29][30]. Therefore, interference with inflammatory cytokine and NO generation likely contributes to MTA hepatoprotective and anti-fibrogenic effects in this model [10,48].
From a mechanistic perspective, and in view of the remarkable anti-inflammatory effects of MTA, we cannot overlook the fact that this molecule is structurally very close to adenosine, a purine nucleoside released form cells with notorious anti-inflammatory properties [15,33]. Adenosine interacts with different G-proteincoupled cell surface receptors including the A1, A2A, A2B and A3 receptors. Until recently adenosine anti-inflammatory effects were mainly attributed to its interaction with the high affinity A2A receptor subtype [49], however an important role for the low affinity A2B receptor in the anti-inflammatory response to adenosine is currently emerging [34,38,40]. Using specific pharmacological antagonists of the A2A and A2B receptors we evaluated their potential implication in the immunomodulatory effects of MTA on mouse RAW 264.7 monocytes/macrophages challenged with bacterial LPS, a prototypic model of the inflammatory response. We found that the inhibition of LPSelicited TNFa production by MTA was not impaired in the presence of these drugs, suggesting that MTA effects were not mediated through these receptors. Nevertheless, the involvement of the adenosine signalling system in the biological effects of MTA cannot be completely disregarded. Interestingly we observed that MTA treatment resulted in a remarkable potentiation of the expression of A2B receptors triggered by LPS, and this effect was not shared by the adenosine receptor agonist NECA. As previously mentioned A2B receptors display low affinity for adenosine, but their expression is up-regulated during tissue injury and inflammation, conditions in which interstitial adenosine concentrations raise. The expression of A2B receptors seems to be very important for the attenuation of the systemic inflammatory response, as illustrated by the fact that A2B knockout mice show enhanced cytokine production and overall mortality during sepsis [39,50]. Therefore the stimulatory effect of MTA on A2B receptor expression could also contribute to dampen inflammation in the chronically injured Mdr22/2 mouse liver. In addition to these responses, a direct cytoprotective effect of MTA on hepatocytes may also be involved, as we previously observed that MTA can prevent hepatocyte death in vitro [51,52]. Mechanistically, the attenuation of JNK activity found in MTAtreated Mdr2 2/2 mice could be related to the hepatoprotective effects of this molecule, given that inhibition of JNK has been demonstrated to prevent liver damage in different in vitro and in vivo experimental models of hepatocellular injury [53][54][55]. In this context the reduced hepatocyte proliferation found in MTAtreated mice may be related in part to the attenuation of liver damage, since liver regeneration occurs in response to injury.
We also observed that MTA has direct effects on liver fibrogenic cells. A previous study had reported that MTA could reduce collagen gene expression in primary hepatic stellate cells [56].
Here we confirmed this observation in liver myofibroblasts, and we also found that MTA down-regulated also the expression of aSMA, TGFb1 and TGFb2, and Mcp-1, all markers of activated fibrogenic cells [3,6,57]. However, MTA did not modify IL6 mRNA levels in myofibroblasts. This supports the specificity of MTA effects on gene expression, and indicates that the downregulation of IL6 observed in vivo could be attributed to the interaction of MTA with other cell types, likely inflammatory cells.
The reduction of TGFb isoforms and Mcp-1 expression by MTA found in vivo as well as in cultured myofibroblasts may be important effects regarding the mechanisms of action of this molecule. TGFb1 and TGFb2 are key cytokines in the development of liver fibrosis that can be expressed in a variety of cells including activated stellate cells, myofibroblats, macrophages and activated cholangiocytes [58,59]. The inhibition of TGFb1 and TGFb2 expression and activation has been demonstrated to attenuate liver fibrosis, including biliary fibrosis [59]. On the other hand, Mcp-1 participates in the recruitment of inflammatory cells during liver injury, and has been cogently demonstrated to contribute to hepatic parenchymal damage [60]. Moreover, besides expressing Mcp-1 liver fibrogenic cells can be activated and induced to proliferate by this chemokine [61,62]. Together these findings suggest that attenuation of TGFb1/2 and Mcp-1 expression could have important mechanistic roles in the therapeutic action of MTA.
Increased proliferation and survival of ECM-producing cells are central events in the development of liver fibrosis [5,58]. MTA has been shown to induce apoptosis of different cell types, including HCC cells [51]. However, we did not observe any pro-apoptotic or cytotoxic effect of MTA in liver myofibroblasts. What we found was a significant inhibition by MTA of FCS-stimulated DNA synthesis. In line with this response we observed that MTA reduced the expression of cyclin D1, a key G 1 cyclin associated with liver fibrogenic cell proliferation [63]. AP-1 transcription factors can stimulate the expression of genes involved in the activation and proliferation of liver fibrogenic cells, and among them JunD is known to play an important role in liver fibrogenesis [27,42]. To gain further insight into the mechanism of action of MTA, we examined the binding of JunD to cyclin D1 promoter and found that it was markedly reduced in myofibroblasts treated with this compound. This can be due in part to the observed downregulation of JunD levels in MTA-treated cells, and therefore reduction of JunD expression may be a relevant mechanism in the antifibrogenic action of MTA.
In order to better understand the mechanisms behind the antiproliferative effects of MTA, we examined the activity of three signalling pathways that are involved in fibrogenic cell proliferation. We found that while MTA did not affect FCS-triggered Erk1/2 phosphorylation, it reduced phopho-c-Jun levels and significantly attenuated PDGF elicited ribosomal S6 protein phosphorylation. Both, the JNK and the mTOR and p70 S6 kinase pathways have been demonstrated as key regulators of hepatic fibrogenic cell proliferation [29,45], and therefore their inhibition by MTA could be relevant in the mechanism of action of this compound. The selective interaction of MTA with protein kinasemediated intracellular signalling pathways as described here is likely to have profound effects on cellular behaviour and thus deserves further examination.
We also noticed that MTA could attenuate the increase in H3K4 trimethylation elicited by FCS stimulation in the cyclin D1 promoter. H3K4 trimethylation is a marker of transcriptional active chromatin, and this modification contributes to gene activation [64]. MTA is an inhibitor of methylation reactions [15], including histone methylation [18], our observations suggest  that interference with this epigenetic mechanism could account in part for the effects of MTA on cyclin D1 gene expression. Moreover, this finding makes it tempting to speculate that MTA could exert its antifibrogenic effects also through an overall action on the epigenetic programming. In this regard 3 deazaneplanocin, another methyltransferase inhibitor, has been recently reported to prevent the activation of liver fibrogenic cells, however its efficacy and tolerability in chronic in vivo models of liver fibrosis have not been tested yet [65]. Together these findings suggest that the antifibrogenic effects of MTA observed in vivo may be mediated in part through the direct inhibition of myofibroblast activation and proliferation. Nevertheless, a direct effect of MTA on bile duct epithelial cell damage and cholangiocyte activation, important events in the pathogenesis of liver injury in Mdr2 2/2 mice [30], cannot be discarded and merit future evaluation. Figure 15 summarizes the proposed mechanisms of action of MTA in Mdr2 2/2 mice mainly based on the observations reported in this study.
In summary, here we have demonstrated that oral MTA administration is an effective therapeutic strategy to attenuate disease progression in a relevant model of liver injury and fibrosis.
In addition to our current study, MTA has been previously tested in in vivo experimental models never showing untoward effects [16,17,66,67], and an ID 50 of 2.960.4 g/kg was estimated when administered intramuscularly to rats [19]. In humans MTA administration is also well tolerated [68,69]. These considerations, together with the efficacy described herein, suggest that MTA could be a good candidate to be clinically tested in the context of liver injury and fibrosis. Figure 15. Potential mechanisms of action of MTA in Mdr2 2/2 mice. Our in vivo and in vitro observations indicate that MTA displays different actions that may underlie its beneficial effects on the course of liver injury and fibrosis in Mdr2 2/2 mice. MTA may exert a direct cytoprotective effect on hepatocytes by reducing JNK activity, preventing hepatocellular death and further inflammation. MTA also inhibits the production of cytokines by inflammatory cells, likely through interference with NFkB activity as described before [16][17][18]. In addition, enhanced expression of adenosine A2B receptors may contribute to the anti-inflammatory pharmacological profile of MTA. This compound can also exert direct effects on extracellular matrix (ECM) producing cells (myofibroblasts), reducing their activation, proliferation and the production of pro-fibrogenic factors. Inhibition of JunD expression may be an important event in the antifibrogenic action of MTA. doi:10.1371/journal.pone.0015690.g015