TGF-β-SMAD3 signaling mediates hepatic bile acid and phospholipid metabolism following lithocholic acid-induced liver injury.

Transforming growth factor-β (TGFβ) is activated as a result of liver injury, such as cholestasis. However, its influence on endogenous metabolism is not known. This study demonstrated that TGFβ regulates hepatic phospholipid and bile acid homeostasis through MAD homolog 3 (SMAD3) activation as revealed by lithocholic acid-induced experimental intrahepatic cholestasis. Lithocholic acid (LCA) induced expression of TGFB1 and the receptors TGFBR1 and TGFBR2 in the liver. In addition, immunohistochemistry revealed higher TGFβ expression around the portal vein after LCA exposure and diminished SMAD3 phosphorylation in hepatocytes from Smad3-null mice. Serum metabolomics indicated increased bile acids and decreased lysophosphatidylcholine (LPC) after LCA exposure. Interestingly, in Smad3-null mice, the metabolic alteration was attenuated. LCA-induced lysophosphatidylcholine acyltransferase 4 (LPCAT4) and organic solute transporter β (OSTβ) expression were markedly decreased in Smad3-null mice, whereas TGFβ induced LPCAT4 and OSTβ expression in primary mouse hepatocytes. In addition, introduction of SMAD3 enhanced the TGFβ-induced LPCAT4 and OSTβ expression in the human hepatocellular carcinoma cell line HepG2. In conclusion, considering that Smad3-null mice showed attenuated serum ALP activity, a diagnostic indicator of cholangiocyte injury, these results strongly support the view that TGFβ-SMAD3 signaling mediates an alteration in phospholipid and bile acid metabolism following hepatic inflammation with the biliary injury.

LPCAT4 induction was observed in mouse primary hepatocytes ( 33 ). TGF ␤ is ubiquitously expressed as latent type, and it is transformed to active type by a variety of agents. Active TGF ␤ binds to the TGF ␤ receptors 1 and 2, resulting in enhanced SMAD3 phosphorylation ( 34 ). Phosphorylated SMAD3 can lead to dynamic alterations in gene expression patterns. Thus, although TGF ␤ -SMAD3 signaling is expected to mediate the metabolic alteration in cholestasis, in vivo studies are required to confi rm this hypothesis. In the current study, LCA-induced metabolic alterations in Smad3 -null mice compared with wild-type mice was investigated, revealing that TGF ␤ -SMAD3 signaling is involved in BA and LPC metabolism in LCA-induced liver injury.

Animals and diets
Male mice (C57BL/6), MAD homolog 3 ( Smad3 )-null mice ( 35 ), and background-matched wild-type mice were housed in temperature-and light-controlled rooms and given water and pelleted NIH-31 chow ad libitum. For the LCA studies, mice were given 0.6% LCA-supplement diet with the AIN93G diet as a control (Dyets, Bethlehem, PA). Three wild-type and three Smad3null mice were fed the control diet, and fi ve wild-type and four Smad3 -null mice were given the LCA diet. All animal studies were carried out in accordance with Institute of Laboratory Animal Resources (ILAR) guidelines and protocols approved by the National Cancer Institute Animal Care and Use Committee.

Serum chemistry
Serum was prepared using Serum Separator Tubes (Becton, Deckinson and Co., Franklin Lakes, NJ). The serum catalytic activity of alanine aminotransferase (ALT) and alkaline phosphatase (ALP) was measured with ALT and ALP assay kit, respectively (Catachem, Bridgeport, CT).

UPLC-ESI-QTOFMS analysis
Serum was prepared using Serum Separator Tubes (Becton, Dickinson and Co.). The serum was diluted with 19 vol of 66% acetonitrile and centrifuged twice at 18,000 g for 20 min to remove insoluble materials. UPLC-ESI-QTOFMS was preformed as previously reported ( 27 ). In brief, the aliquots (5 l) were injected into a reverse-phase 50 × 2.1 mm ACQUITY 1.7 m C18 column (Waters, Milford, MA) using an ACQUITY UPLC system (Waters) with a gradient mobile phase comprising 0.1% formic acid and acetonitrile containing 0.1% formic acid. The eluant was introduced by electrospray ionization into the mass spectrometer [Q-TOF Premier (Waters)] operating in negative electrospray ionization mode. The capillary and sampling cone voltages were set to 3000 and 30 V, respectively. The desolvation gas fl ow was set to 650 l/h, and the temperature was set to 350°C. The cone gas fl ow was 50 l/h, and the source temperature was 120°C. Data were acquired in centroid mode from m/z 50 to 800 in maintained by enterohepatic circulation in which greater than 90% are reabsorbed from the small intestine and transported to the liver.
Interruption of BA excretion from the liver can enhance hepatic BA levels, leading to cholestatic liver disease. Accumulation of hepatic BA results in activation of the farnesoid X receptor (FXR). A critical role for FXR in the regulation of BA homeostasis was established using Fxrnull mice ( 8 ). FXR binds to BA and directly or indirectly contributes to the downregulation of CYP7A1 (BA synthesis), SLC10A1 and SLCO family (BA uptake) expression (9)(10)(11), and upregulation of ABCB11 and ABCC2 (BA export) expression (12)(13)(14). In addition, FXR activation enhances expression of the organic solute transporter ␣ / ␤ (OST ␣ / ␤ ) genes in the liver ( 15 ) and accelerates BA excretion to the bloodstream. Thus, FXR functions as the chief sensor of intracellular BA levels and contributes to BA homeostasis ( 16 ).
When hepatic BA accumulate in liver, hepatic infl ammation is induced by tumor necrosis factor (TNF) ␣ and transforming growth factor (TGF) ␤ . TNF ␣ signaling is known to downregulate SLCO1A1 and SLC10A1 expression in the liver ( 17 ) and reduce bile canalicular contraction ( 18 ). In addition, TNF ␣ upregulates multidrug resistanceassociated protein 3 (MRP3; basolateral bile salt export transporter, also called ABCC3) and can protect from liver injury that results from obstructive cholestasis ( 19 ). Kupffer cell activation mediates induction of hepatic multidrug resistance-associated protein 4 (MRP4; basolateral bile salt export transporter, also called ABCC4) expression ( 20 ), which is involved in basolateral BA export. The TGF ␤ signal regulates expression of CYP7A1 gene ( 21 ), the ratelimiting enzyme of the BA synthesis. Proinfl ammatory cytokines can also infl uence BA homeostasis.
Among the BA, lithocholic acid (LCA) is the most potent chemical causing liver toxicity. LCA levels are elevated in patients with liver disease ( 22 ) and intrahepatic cholestasis ( 23 ). Experimental interventions to protect against LCA toxicity were investigated using animal models, revealing that the nuclear receptor pregnane X receptor (PXR) protects against LCA toxicity through upregulation of cytochrome P450 3A (CYP3A) and sulfotransferase 2A (SULT2A) that metabolize LCA for elimination (24)(25)(26). A variety of LCA metabolites were reported to be associated with this protection (27)(28)(29)(30)(31). Recently, endogenous bile acid metabolism associated with LCA toxicity was investigated ( 27 ), and LCA exposure was reported to change serum chemistry, such as phospholipids, cholesterol, free fatty acids, and triglycerides ( 32 ). Furthermore, a comprehensive view of LCA-induced alterations in endogenous metabolites was investigated by use of metabolomics for the detection and characterization of small organic chemicals in biological matrices. LCA exposure decreased serum lysophosphatidylcholine (LPC) levels, leading to cholestasis ( 33 ). The change in serum LPC was associated with increased serum alkaline phosphatase (ALP), a marker of cholangiocyte injury, and increased hepatic lysophosphatidylcholine acyltransferase (LPCAT)4 expression. In addition, the TGF ␤ -MAD homolog 3 (SMAD3)-dependent 2 h, the hepatocytes were exposed to 5 ng/ml of TGF ␤ (R and D Systems, Minneapolis, MA) for 12 h, and then collected and lysed for gene expression analysis by use of qPCR. HepG2 cells were seeded on a 12-well plate and infected with recombinant adenovirus. Two days later, the HepG2 cells with the starvation were exposed to 5 ng/ml of TGF ␤ . The cells were subjected to the Western blotting (treatment with TGF ␤ for 1 h) or the qPCR (treatment with TGF ␤ for 12 h).

Statistical analysis
Statistical analysis was performed using Prism version 5.0c (GraphPad Software, San Diego, CA). A P -value of less than 0.05 was considered as signifi cant difference.

Lithocholic acid exposure enhanced TGF ␤ and the receptors mRNA level in the liver
The infl uence of lithocholic acid (LCA) exposure on TGF ␤ signaling was investigated using C57BL/6 mice treated with the synthetic AIN93G diet (Cont) and 0.6% LCA-supplemented AIN93G diet (LCA) for 7 days. Hepatic TGFB1, TGFBR1, and TGFBR2 mRNA levels increased after LCA exposure, although TGFBR3 mRNA level did not changed in the livers ( Fig. 1 ). These results suggest that LCA exposure stimulates TGF ␤ signaling in the livers.

LCA-induced liver injury was alleviated in Smad3 -null mice
TGF ␤ activates SMAD3 via the TGF ␤ receptors. Thus, to investigate whether SMAD3 was involved in LCA-induced liver injury, Smad3 -null mice were treated with control diet and LCA diet for 6 days. After LCA exposure, the liver mass of Smad3 -null mice was smaller than that of LCAtreated wild-type mice ( Fig. 2A ). In addition, LCA-increased serum ALP activities were signifi cantly attenuated in the Smad3 -null mice, although serum ALT activities were not changed ( Fig. 2B, C ). Furthermore, liver histology showed mild features of infl ammatory cell infi ltration around the portal vein in Smad3 -null mice, which was not observed in similarly treated wild-type mice ( Fig. 2D ). Immunohistochemistry revealed TGF ␤ protein around the portal vein with lower expression of the TGF ␤ in Smad3 -null mice compared with wild-type mice ( Fig. 2E ), suggesting lower TGF ␤ stimulation of the SMAD3 activation in the liver MS scanning. Tandem MS collision energy was scanned from 5 to 35 V.

Data processing and multivariate data analysis
Data processing and multivariate data analysis were conducted as previously reported ( 27 ). Partial least squares (PLS) and contribution analyses were performed using SIMCA-P+12 (Umetrics, Kinnelon, NJ).

Quantifi cation of serum bile acid and lysophosphatidylcholine
Serum bile acid levels were determined with standard curves using authentic metabolites. Quantifi cation of serum lysophosphatidylcholine was performed according to a previously reported method ( 33 ).

RNA analysis
RNA was extracted using TRIzol reagent (Invitrogen, Carlsbad, CA), and qPCR was performed using cDNA generated from 1 µg total RNA with a SuperScript II Reverse Transcriptase kit and random oligonucleotides (Invitrogen). Primers were designed using qPrimerDepot. All sequences are listed in supplementary  Table I. Quantitative PCR reactions were carried out using SYBR green PCR master mix (Applied Biosystems, Foster City, CA) in an ABI Prism 7900HT Sequence Detection System. Values were quantifi ed using comparative CT method, and samples were normalized to 18S rRNA.

Western blotting
For whole-cell extracts, a culture cell was lysed with sample buffer and subjected to Western blotting. The samples were boiled for 5 min and then separated and transferred to PVDF membranes using standard Western blotting techniques. The membranes were incubated with an antibody against phospho-SMAD3 at a dilution of 1:1,000, ab52903 (Abcam, Cambridge, UK) or SMAD3 at a dilution of 1:1,000 (ab28379, Abcam). The signals were normalized to signals obtained with a GAPDH Ab used at a dilution of 1:10,000 (MAB374, Millipore, Billerica, MA).

Histological analysis
Small blocks of liver tissue were immediately fi xed in 10% neutral formalin and embedded in paraffi n. Sections (4 µm thick) were stained with hematoxylin and eosin. At least three discontinuous liver sections were evaluated for each mouse. Immunolocalization of TGF ␤ 1 and phospho-Smad3 were performed as described ( 36 ) with the addition of antigen retrieval in 1 mM EDTA (pH 8) at 95°C for 10 min for sections stained with the pSmad3 antibody. TGF ␤ 1 was detected with the primary antibody LC-1-30-1 at 3 µg/ml, and pSmad3 was detected with a rabbit monoclonal antibody from Epitomics (Cat. No. 1880-1) at a dilution of 1:500.

Generation of Smad3-expressing adenovirus with Cre/LoxP system
A Smad3-expressing adenovirus with Cre/LoxP system (Ad-L-Smad3) was generated by using the Adenovirus Cre/LoxP-Regulated Expression Vector Set (Takara, Tokyo, Japan). A Creexpressing adenovirus (Ad-Cre) and a GFP-expressing adenovirus (Ad-L-GFP) with Cre/LoxP system were previously generated ( 37 ). The titer of these adenoviruses was measured by the 50% tissue culture infectious dose method.

Culture of primary hepatocytes and HepG2 cells
Primary hepatocytes were prepared as previously reported ( 33 ). After starvation with FBS-negative Williams' Medium E for

Role of TGF ␤ -SMAD3 signal in bile acid homeostasis
Quantifi cation of serum bile salts was performed with authentic compounds. All of the tested bile acid levels were lower in Smad3 -null mice after LCA exposure than those in the wild-type mice ( Fig. 4A ). As bile acid metabolites can be produced from LCA (supplementary Fig. II) in the liver, the expression of the major metabolic enzymes CYP3A11 and sulfotransferase 2A (SULT2A) was investigated. CYP3A11 and SULT2A expression was not, however, different between the wild-type and the Smad3 -null mice after LCA feeding ( Fig. 4B ). Thus, hepatic expression of bile acid-related transporter genes was investigated, including mRNAs encoding ABCB11, ABCC1, ABCC2, ABCC3, ABCC4, ABCC5, solute carrier (SLC)10A1, solute carrier organic anion transporter (SLCO)1A1, SLCO1A4, SLCO1B2, organic solute transporter (OST) ␣ , and OST ␤ . Differences in expression of the bile salt uptake transporters SLCO1A1, SLCO1A4, and SLCO1B2 (intake to hepatocyte, Fig. 4C ), major bile salt exporters ABCC2 and ABCB11 (export to bile duct, Fig. 4D ), and bile acid synthesis enzyme CYP7A1 ( Fig. 4E ) were not observed in the liver. However, expression of the basolateral exporting transporter OST ␤ and ABCC4 was much lower in the Smad3null liver than that in the wild-type mice ( Fig. 4F ). SMAD3 (supplementary Fig. I). In addition, a dramatic attenuation of the SMAD3 phosphorylation signal was observed in the hepatocyte nuclei of Smad3 -null mice ( Fig. 2E ). These results suggest that TGF ␤ -SMAD3 signaling is associated with the LCA-induced biliary injury and raise the possibility that TGF ␤ -SMAD3 signaling alters hepatic metabolism.

Difference between wild-type and Smad3 -null mice in the serum metabolome after LCA exposure
To examine serum metabolites, PLS and contribution analyses were performed with UPLC-ESI-QTOFMS negative mode data derived from serum of mice fed LCA or control diet. PLS analysis showed a separation between the LCA-treated wild-type ( Fig. 3A ) and the LCA-treated Smad3null group that was further examined with a loadings plot ( Fig. 3B ). Contribution analysis revealed 10 enhanced and 10 attenuated ions as the top-ranking ions giving rise to the separation. Lysophosphatidylcholine (LPC) and fatty acid fragments were determined as raised ions in Smad3null mice compared with the wild-type mice ( Table 1 ). The most lowered ions were derived from bile salts ( Table 2 ). After LCA feeding, the serum metabolome of Smad3 -null mice was much different from that of the wild-type mice in LPC and bile salts. with isolated and cultured mouse primary hepatocytes. TGF ␤ -elevated OST ␤ and LPCAT4 mRNA levels were observed in hepatocytes, while ABCC4 expression was decreased ( Fig. 6A ). In addition, with Cre recombinasedependent human SMAD3 expression ( Fig. 6B ), the induction was also observed in the human hepatocarcinoma cell line HepG2 ( Fig. 6C ). These results strongly indicate that TGF ␤ -SMAD3 signaling mediates induction of Ost ␤ and Lpcat4 gene expression.

DISCUSSION
The current study demonstrated that TGF ␤ -SMAD3 signaling was involved in bile acid and phospholipid homeostasis in part through inducing hepatic expression of Ost ␤ and Lpcat4 genes ( Fig. 7 ). UPLC-ESI-QTOFMS, in conjunction with a PLS analysis, illustrated a clear difference between wild-type and Smad3 -null mice in serum metabolites after LCA exposure, with BA metabolites being markedly elevated while LPC were decreased. Alteration in the may be involved in the induction of the Ost ␤ and Abcc4 genes, as LCA exposure decreased hepatic expression of FXR that upregulates the genes (supplementary Fig. III). In addition to the bile acids, quantifi cation of serum palmitoyl-LPC (16:0 LPC) and setearoyl-LPC (18:0 LPC) was performed with use of the authentic compounds. Wild-type mice showed attenuated serum 16:0 LPC and 18:0 LPC after LCA feeding, but Smad3 -null mice did not ( Fig. 5A ). Induction of the related gene expression, as previously reported, was attenuated in Smad3 -null mice ( Fig. 5B-D ). Notably, LPCAT4 expression was lower in the Smad3 -null mice compared with the wild-type mice. Thus, Ost ␤ , Abcc4 , and Lpcat4 were expected to be TGF ␤ -SMAD3 signal-related genes in the liver.

TGF ␤ -SMAD3 signal regulated the genes encoding OST ␤ and LPCAT4 in mouse hepatocytes and human cells
To investigate whether TGF ␤ signaling directly induced the expression of Ost ␤ , Abcc4 , and Lpcat4 genes in the hepatocyte, the respective mRNAs were quantitated  The ion ranking, based on the PLS-DA analysis, indicates the highest confi dence and greatest contribution to separation between the wild-type and the Smad3 -null mice after LCA exposure. RT, retention time.
LCA induction of OST ␤ expression was observed in the liver. Considering the enhancement of hepatocyte SMAD3 activation and SMAD3-dependent elevation of serum BA levels, these observations may suggest that TGF ␤ -SMAD3-OST ␤ signaling is an important factor for controlling the hepatocyte BA levels under conditions of reduced hepatic FXR, whereas TNF ␣ and nuclear factor (erythroid-derived 2)-like 2 (NRF2) mediate elevation of serum BA levels via hepatic ABCC3 and ABCC4 induction ( 19,45 ). Thus, proinfl ammatory cytokine signaling (with oxidative stress) was considered to turn the direction of hepatic bile acid fl ow (from vein to hepatocyte, and bile duct). Hepatic BA accumulation can stimulate enhanced oxidative stress and accelerate the proinfl ammatory response in the liver. Proinfl ammatory cytokines may play an important role in the bile acid-induced acute liver injury. Future studies will be required to establish the pathophysiological significance of these fi ndings.
Cholic acid (CA) challenge is frequently used as a cholestasis model, as well as LCA. Interestingly peroxisome proliferator-activated receptor (PPAR) ␣ -defi cient mice showed liver injury after cholic acid CA challenge, whereas the corresponding wild-type mice did not ( 38 ). As PPAR ␣ is activated by fatty acid metabolites and controls lipid metabolism, this observation may suggest that lipid is associated with BA homeostasis. Synthetic PPAR ␣ activators, such as [4-chloro-6-(2,3-xylido)-2-pyrimidinylthio)] acetic acid (also called Wy-14,643), ciprofi brate, and bezafi brate, are known to downregulate the human CYP7A1 and rodent Cyp7a1 gene promoters (46)(47)(48). In addition, PPAR ␣ was reported to mediate clofi brate-induced hepatic MRP3 (ABCC3) and MRP4 (ABCC4) expression ( 49 ), and ciprofi brate reduces hepatic NTCP (SLC10A1), OATP1 (SLCO1A1), and BSEP (ABCB11) expression in mice ( 50 ). However, Ppara -null mice did not diminish altered hepatic expression of these genes after CA challenge ( 38 ), and thus, the pathophysiological signifi cance of PPAR ␣ on bile acid homeostasis remains unclear. Therefore, the PPAR ␣mediated protection against CA toxicity may be accomplished through the other gene function. The hydrophobic BA, such as LCA, deoxycholic acid, and chenodeoxycholic acid, can lead to mitochondria dysfunction and induce hepatic apoptosis and necrosis. CA challenge elevates the saturated LPC tightly associates not only with LCA-induced liver injury but also with the other cholestasis model ( 38 ), which is related to the increased serum BA. Interestingly, Smad3 -null mice showed lowered serum ALP activity (a cholangiocyte injury marker) associated with acute liver injury compared with wild-type mice, but they did not have altered serum ALT activity (a hepatocyte injury marker). In chronic liver disease, such as cholestasis and primary biliary cirrhosis, chronic autoimmune-mediated damage to bile duct epithelial cell (BDEC) can cause infl ammation with activation of wound healing in the liver ( 39 ). In bile duct ligation (BDL) model of obstructive cholestasis, the BDEC-expressed ␤ 6 integrin activates the latent TGF ␤ ( 40,41 ). In another biliary injury model, the BDEC-selective toxicant ␣ -naphthylisothiocyanate-exposed mouse, ␤ 6 integrin-mediated TGF ␤ activation was observed ( 42 ). In addition, TGF ␤ expression was observed in metaplastic bile duct epithelium ( 43 ), as was found in the present study . These observations strongly support the view that the serum metabolic alteration refl ects the severity of biliary injury. This fi nding, together with ALT, ALP, and ␥ -glutamyltransferase, may contribute to an accurate diagnosis of liver injury.
TGF ␤ was reported to be a controller of BA synthesis through regulation of the human CYP7A1 gene ( 21 ), whereas no difference in hepatic CYP7A1 mRNA level was observed between wild-type and Smad3 -null mice in the current study, thus suggesting that other factors were important for regulating mouse Cyp7a1 gene transcription. On the other hand, the present study revealed that TGF ␤ -SMAD3 was a mediator of basolateral export transporter OST ␤ expression. OST ␤ together with OST ␣ are expressed in hepatocytes and cholangiocytes, and they mediate BA movement from hepatocyte and cholangiocyte to the circulation. In addition, decreased BA pool size and attenuated serum BA levels were observed in mice lacking OST ␣ / ␤ ( 44 ). FXR, which is a chief sensor of intracellular BA levels, is believed to control the OST ␣ / ␤ gene transcription in human and rodents that regulates hepatic BA levels via controlling the enterohepatic system. LCA attenuated hepatic FXR expression, but not PXR expression, that regulates expression of the BA metabolic enzymes CYP3A and SULT2A. Despite hepatic FXR attenuation, The ion ranking, based on the PLS analysis, indicates the highest confi dence and greatest contribution to separation between the wild-type and the Smad3 -null mice after LCA exposure. RT, retention time.
a No clear separation between tauro-␣ -muricholate and tauro-␤ -muricholate. b No clear separation between tauroursodeoxycholate and taurohyodeoxycholate. and development of cirrhosis and cancer. Some patients with NASH were reported to show increased plasma bile acid levels ( 55 ). In a previous report, methionine-and choline-defi cient diet-fed mice (experimental NASH model) also showed increased serum bile acids ( 56 ). In addition, experimental NASH was associated with decreased serum LPC, similar to the LCA-exposed mice ( 56 ). Furthermore, elevated hepatic TGF ␤ levels were observed in the NASH model ( 56 ). These observations suggest a metabolic link, in which TGF ␤ is involved in lipid and BA metabolism following a variety of liver diseases. Understanding the molecular mechanism of bile acid-induced liver injury (cholestasis) that is common with fat accumulation-related liver disease (NASH) would contribute to the development of clinical therapies.
hydrophobic BA levels in the liver. PPAR ␣ activation protects from liver injury due to mitochondria dysfunction, such as acetaminophen-induced liver injury ( 51 ), and the PPAR ␣ protection against mitochondria damage may contribute to alleviation of the CA-induced cholestasis. However, a detailed understanding requires further study. As we observed that Fxr -null mice showed increased hepatic lipid levels with hepatic accumulation of bile acid ( 52 ), lipid metabolism may be associated with bile acid-induced liver injury or bile acid homeostasis. Recently, the number of patients with fat accumulationrelated liver disease (including hepatic steatosis and nonalcoholic steatohepatitis; NASH) is increasing not only in middle agers but also in teenagers ( 53,54 ). NASH, with chronic hepatic infl ammation , can lead to fi brogenesis signaling may be involved in biliary injury. Proinfl ammatory cytokines, such as TGF ␤ , are mainly secreted from nonparenchymal cells. Thus, a further understanding of liver diseases requires additional studies on the role of In conclusion, the present study revealed that TGF ␤ -SMAD3 signaling was involved in hepatic bile acid and phospholipid homeostasis regulating LPCAT4 and OST ␤ gene expression. Serum LPC levels following TGF ␤ -SMAD3  1-4 (B), phospholipase D (C), and de novo synthesis of phosphatidylcholine (D). Signifi cance was determined by one way-ANOVA with Bonferroni's test (* P < 0.05; ** P < 0.01; *** P < 0.001). Fig. 6. TGF ␤ stimulates both mouse and human OST ␤ and LPCAT4 gene transcription. (A) qPCR analysis for OST ␤ , ABCC4, and LPCAT4 mRNA expression in mouse primary hepatocytes after TGF ␤ exposure. Signifi cance was determined by unpaired t -test (** P < 0.001; *** P < 0.001). (B) Western blotting for human SMAD3 expression and TGF ␤stimulated SMAD3 phosphorylation in HepG2 cells infected with SMAD3-expressing adenovirus. Two days after the infection (Ad-L-GFP, Ad-Cre, and/or Ad-L-SMAD3), the HepG2 cells (seeded on a 12-well plate) were subjected to Western blotting. Adenovirus (13 MOI) was adjusted with Ad-L-GFP. Ϫ , +, and ++ is presented as 0, 3, and 10 MOI, respectively. TGF ␤ (2.5 ng/ml) exposure was performed with 2 h starvation (FBS negative) after the SMAD3 introduction with Ad-Cre. One hour later, the TGF ␤ -exposed HepG2 cells were subjected to Western blotting. (C) qPCR analysis for OST ␤ and LPCAT4 gene expression in the SMAD3-introduced HepG2 cells after TGF ␤ exposure. Signifi cance was determined by one way-ANOVA with Bonferroni's test (*** P < 0.001). NS, no signifi cance. hepatocyte-nonparenchymal cell interactions in hepatic metabolism. Knowledge of metabolic linkage between proinfl ammatory signal and hepatic lipid and bile acid homeostasis may contribute to further understanding of chronic liver diseases, such as steatohepatitis, liver fi brosis and cirrhosis, as well as cholestasis .