Volume 147, Issue 6, Pages 1378-1392 (December 2014) Extrahepatic Platelet-Derived Growth Factor-β, Delivered by Platelets, Promotes Activation of Hepatic Stellate Cells and Biliary Fibrosis in Mice  Shuhei Yoshida, Naoki Ikenaga, Susan B. Liu, Zhen-Wei Peng, Jeanhee Chung, Deanna Y. Sverdlov, Makoto Miyamoto, Yong Ook Kim, Shinji Ogawa, Robert H. Arch, Detlef Schuppan, Yury Popov  Gastroenterology  Volume 147, Issue 6, Pages 1378-1392 (December 2014) DOI: 10.1053/j.gastro.2014.08.038 Copyright © 2014 AGA Institute Terms and Conditions

Figure 1 Hepatic and systemic PDGF-B expression patterns suggest an extrahepatic supply at the protein level. Platelets are homing to the liver endothelium in close proximity to hepatic stellate cells in fibrotic Mdr2-/- mice. (A) Although hepatic mRNA levels of PDGF-B (reverse-transcription polymerase chain reaction) are not increased at any time point, protein levels (ELISA) of hepatic and plasma PDGF-B increased several-fold in fibrotic Mdr2-/- mice. (B) CD41 immunofluorescence (upper panel, 200×) and Western blotting (lower panel) of liver lysates show significant accumulation of platelets in fibrotic Mdr2-/- livers (platelet aggregates are indicated by arrows). *Portal vein. (C) PDGF-B protein levels are increased in whole blood and platelet-free serum in Mdr2-/- mice (closed bars) compared with wild-type (WT) mice (open bars). (D) Triple immunofluorescent staining for the platelet marker CD41 (green), the HSC marker desmin (red), and the endothelial marker LYVE1 (blue). High-resolution confocal microscopy shows that CD41+ platelet clusters are immediately adjacent to LYVE1+ endothelial cells and in close proximity to desmin+ HSC (arrows). Original magnification: 630× (upper panels) and 1000× (lower panels). Supplementary Figure 2 shows the corresponding Z-stack analysis. arb, arbitrary; PLT, platelet. Gastroenterology 2014 147, 1378-1392DOI: (10.1053/j.gastro.2014.08.038) Copyright © 2014 AGA Institute Terms and Conditions

Figure 2 Serum PDGF-BB levels and platelet counts in patients with liver diseases at various stages of liver fibrosis progression. (A) Characteristics of the patients with liver diseases. Patients were subjected to a diagnostic biopsy and subdivided into 5 groups according to fibrosis stage (Metavir). (B) A significant increase of serum PDGF-BB levels in patients with liver fibrosis was observed at early histologic stages (F1–2), and decreased at an advanced stage (F4). (C) Circulating platelet (PLT) counts significantly decreased in patients with advanced liver fibrosis (F4). Data are expressed as means ± SEM. Multiple comparisons were performed by one-way analysis of variance followed by the Dunnett post-test. *P < .05 compared with F0. AIH, autoimmune hepatitis; ALB, albumin; HBV, hepatitis B virus; HCV, hepatitis C virus; NAFLD, nonalcoholic fatty liver disease; NASH, nonalcoholic steatohepatitis; PBC, primary biliary cirrhosis. Gastroenterology 2014 147, 1378-1392DOI: (10.1053/j.gastro.2014.08.038) Copyright © 2014 AGA Institute Terms and Conditions

Figure 3 Platelet depletion reduces hepatic PDGF-B levels and diminishes hepatic fibrogenesis markers in vivo. Platelet depletion was achieved by a single injection of α-CD41 mAb (1 mg/kg IP) in 8-week-old female Mdr2-/- mice and perfused livers were analyzed 24 and 48 hours later. (A) Platelet depletion leads to a more than 3-fold reduction in hepatic PDGF-B protein levels in Mdr2-/- mice, and (B) significantly diminished HSC activation marker α-SMA at 48 hours, as measured by Western blotting. Platelet depletion was confirmed by hepatic CD41 with β-actin as a loading control (lower row). (C) Platelet depletion significantly reduced (profibrogenic) TIMP-1 and procollagen-α1(I), and significantly increased (profibrolytic) collagenase expression MMP-8 and MMP-13 (24 and 48 h, respectively). Data are means ± SEM (n = 5). *P < .05 compared with the isotype-treated group (Iso). arb, arbitrary; CTRL, control; PLT, platelet; WT, wild type. Gastroenterology 2014 147, 1378-1392DOI: (10.1053/j.gastro.2014.08.038) Copyright © 2014 AGA Institute Terms and Conditions

Figure 4 Antiplatelet therapy with low-dose dietary aspirin reduces long-term fibrosis outcomes in Mdr2-/- mice. Mdr2-/- mice were fed control and aspirin-supplemented diet (150 mg/kg of diet) from 6 weeks of age and analyzed at 12 weeks (12w) and 12 months (12m) of age. (A) Aspirin supplementation reduces hepatic and circulating PDGF-B levels in Mdr2-/- mice. (B) Hepatic collagen content (via hydroxyproline) and (C) serum ALT levels in Mdr2-/- in young (12w) and aged (12m) Mdr2-/- mice. (D) Connective tissue staining shows that severe sinusoidal fibrosis is prevented in aged aspirin-treated Mdr2-/- mice (12m). Representative Sirius Red images (magnification: upper panel, 50×; lower panel, 400×). (E) α-SMA expression in liver lysates of 1-year-old Mdr2-/- mice (Western blotting and densitometry). (F) Primary liver cancer (hepatocellular carcinoma [HCC]) incidence and burden in Mdr2-/- mice treated with aspirin (Ctrl, n = 8; ASP, n = 9). Tumor size and counts were recorded at macroscopic liver examination at 12 months of age. *P < .05, **P < .01 compared with control diet-fed Mdr2-/- mice (Student t test). Ctrl, control; WT, wild type. Gastroenterology 2014 147, 1378-1392DOI: (10.1053/j.gastro.2014.08.038) Copyright © 2014 AGA Institute Terms and Conditions

Figure 5 Selective targeting of PDGF-B with monoclonal antibody MOR8457 strongly and dose-dependently inhibits biliary fibrosis progression in the Mdr2-/- model. (A) Experimental design: anti–PDGF-B antibody (MOR8457, IP weekly at 1, 10, and 100 mg/kg) or isotype control mAb (100 mg/kg) was administered for 6 weeks in Mdr2-/- mice of both sexes. Nonselective tyrosine kinase inhibition with imatinib (50 mg/kg/day, oral gavage) was used for comparison. (B) MOR8457 dramatically reduced circulating levels of PDGF-B (serum ELISA). (C) Hepatic collagen content was reduced significantly in Mdr2-/- mice treated with increasing doses of MOR8457. (D) Thickness of periductular onion-skin scarring is reduced dose-dependently in Mdr2-/- mice receiving anti–PDGF-B mAb, but not in the Iso- or imatinib-treated group (Sirius Red, 200×). *Portal vein. (E) Fibrosis-related gene mRNA expression (procollagen α1(I), TIMP-1, and TGFβ2) quantified by quantitative reverse-transcription polymerase chain reaction. Data are means ± SEM. *P < .05 compared with the isotype control-treated group (Iso) of respective sex. arb, arbitrary; WT, wild type. Gastroenterology 2014 147, 1378-1392DOI: (10.1053/j.gastro.2014.08.038) Copyright © 2014 AGA Institute Terms and Conditions

Figure 6 Platelets induce PDGF-dependent fibrogenic activation of freshly isolated hepatic stellate cells in vitro. (A) Addition of isolated platelets (platelet [PLT], 5 × 104/μL in murine plasma) or recombinant PDGF-BB (PDGF, 25 ng/mL) to freshly isolated HSC cultures for 48 hours induces loss of lipid droplets, spreading HSC phenotype (upper panel: phase-contrast at magnification 100×) and (B) robust induction of α-SMA (red immunofluorescence; lower left panel: magnification, 200×; right panel: Western blotting of cell lysates with densitometry) compared with controls (CTRL, treated with an equal volume of murine platelet-free plasma). (C) Transcript levels of procollagen α1(I), TIMP-1, and TGFβ1 in HSC cultures incubated with increasing concentrations of platelets (1× and 5 × 104/uL). (D) Neutralization of PDGF-B inhibits platelet-induced α-SMA induction in primary HSCs as assessed by immunofluorescence and Western blotting with densitometry (right panel). Cells were incubated with PLT (5 × 104 /μL) or PDGF-BB (25 ng/mL) in the presence of anti–PDGF-B mAb (MOR, 0.25 umol/L) or isotype control antibody (Iso) for 24 hours. All experiments were performed in triplicate, and replicated in 3 separate cell isolations. Data are means ± SEM. *P < .05. Gastroenterology 2014 147, 1378-1392DOI: (10.1053/j.gastro.2014.08.038) Copyright © 2014 AGA Institute Terms and Conditions

Figure 7 Antibody targeting of PDGF-B achieves robust antifibrotic efficacy in a DDC-feeding model of cholangitis and biliary fibrosis. MOR8457 mAb (10 and 100 mg/kg/wk) or isotype control antibody (100 mg/kg) were administered IP weekly during fibrosis induction by DDC. (A) Connective tissue staining shows a dose-dependent inhibition of both onion-skin fibrosis (original magnification, 200×) (*portal vein) and (B) a quantitative reduction in collagen deposition (via hydroxyproline) in DDC-fed mice treated with anti–PDGF-B mAb. (C) Decrease in ductular reaction marker (K19, upper panel) is accompanied by a dramatic down-regulation in hepatic stellate cell activation marker α-SMA (middle panel) by Western blot analysis. *P < .05 compared with the isotype control-treated group. (D) Scheme of proposed pathophysiological mechanisms of the platelet role in PDGF-B–driven HSC activation and hepatic fibrogenesis. Circulating platelets are activated in chronic biliary liver disease and are home to the liver endothelium, supplying and releasing PDGF-B protein that binds PDGFR-β on stellate cells and promotes their fibrogenic activation. Activated stellate cells and reactive cholangiocytes provide survival and profibrogenic stimuli for each other, perpetuating the fibrogenic response. This pathway can be targeted therapeutically at the platelet activation level by antiplatelet therapy such as low-dose aspirin, thereby preventing platelet activation and homing to the liver, or by neutralization of PDGF-B protein released from platelets using selective anti–PDGF-B antibody, disrupting the HSC activation and profibrogenic signaling downstream of platelet activation. CTRL, control. Gastroenterology 2014 147, 1378-1392DOI: (10.1053/j.gastro.2014.08.038) Copyright © 2014 AGA Institute Terms and Conditions