Differential Effects of Selective- and Pan-PPAR Agonists on Experimental Steatohepatitis and Hepatic Macrophages

Research Article NAFLD and Alcohol-Related Liver Diseases

Differential effects of selective- and pan-PPAR agonists on experimental steatohepatitis and hepatic macrophages

Graphical abstract Authors

Sander Lefere, Tobias Puengel, PPARα agonism PPARδ agonism PPARγ agonism . Lanifibranor Lanifibranor Lanifibranor Jana Hundertmark, , Fat-laden Infiltrating Activated hepatic Lindsey Devisscher, hepatocytes macrophages stellate cells Guillaume Wettstein, Frank Tacke PPARα PPARδ PPARγ

α-SMA Correspondence PDK4 Collagen 3 CPT1b IL-6 [email protected] CPT2 CCL2 (F. Tacke). CD36 PLIN2 TNF-α TGF-β iNOS Collagen 1 Lay summary CXCL2 Peroxisome proliferated- Steatosis Inflammation Fibrosis activated receptors (PPARs) are essential regulators of metabolism and inflammation. Highlights We demonstrated that the pan-PPAR agonist lanifibranor PPARs are beneficial regulators of metabolism, making them promising drug ameliorated all aspects of non- candidates in NAFLD. alcoholic fatty liver disease in independent experimental The pan-PPAR agonist lanifibranor reduces steatosis, inflammation and mouse models. Non-alcoholic fibrosis in two NAFLD mouse models. fatty liver disease and fatty acids induce a specific polari- Pan-PPAR agonism indirectly inhibits hepatic macrophage infiltration. zation status in macrophages, fi Human and murine macrophages display a metabolically activated pheno- which was altered by lani - type in NAFLD. branor to increase expression of lipid handling genes, Lanifibranor decreases pro-inflammatory activation of macrophages via thereby decreasing inflamma- PPARd agonism. tion. PPAR isoforms have differential therapeutic effects on fat-laden hepatocytes, activated hepatic stellate cells and inflammatory macrophages, supporting the clinical development of pan-PPAR agonists.

https://doi.org/10.1016/j.jhep.2020.04.025 © 2020 European Association for the Study of the Liver. Published by Elsevier B.V. All rights reserved. J. Hepatol. 2020, -,1–14 Research Article NAFLD and Alcohol-Related Liver Diseases

Differential effects of selective- and pan-PPAR agonists on experimental steatohepatitis and hepatic macrophagesq

Sander Lefere1,2,†, Tobias Puengel2,3,†, Jana Hundertmark2,3, Christian Penners2, Anna Katharina Frank2, Adrien Guillot3, Kevin de Muynck4, Felix Heymann2,3, Vanessa Adarbes5, Evelyne Defrêne5, Céline Estivalet5, Anja Geerts1, Lindsey Devisscher4, Guillaume Wettstein5, Frank Tacke2,3,*

1Department of Gastroenterology and Hepatology, Hepatology Research Unit, Ghent University, Ghent, Belgium; 2Department of Medicine III, University Hospital Aachen, Aachen, Germany; 3Department of Hepatology and Gastroenterology, Charité University Medicine Berlin, Berlin, Germany; 4Department of Basic and Applied Medical Sciences, Gut-Liver Immunopharmacology Unit, Ghent University, Ghent, Belgium; 5Inventiva, Daix, France

Background & Aims: Peroxisome proliferator-activated re- macrophages, as well as patient-derived circulating monocytes, ceptors (PPARs) are essential regulators of whole-body meta- in a PPARd-dependent fashion. bolism, but also modulate inflammation in immune cells, notably Conclusion: Pan-PPAR agonists combine the beneficial effects of macrophages. We compared the effects of selective PPAR ago- selective PPAR agonists and may counteract inflammation and nists to those of the pan-PPAR agonist lanifibranor in disease progression more potently. PPARd agonism and lanifi- non-alcoholic fatty liver disease (NAFLD), and studied isoform- branor directly modulate macrophage activation, but not infil- specific effects on hepatic macrophage biology. tration, thereby synergizing with beneficial metabolic effects of Methods: Lanifibranor or selective PPARa (fenofibrate), PPARc PPARa/c agonists. (pioglitazone) and PPARd (GW501516) agonists were therapeu- Lay summary: Peroxisome proliferated-activated receptors tically administered in choline-deficient, amino acid-defined (PPARs) are essential regulators of metabolism and inflamma- high-fat diet (CDAA-HFD)- and Western diet (WD)-fed mouse tion. We demonstrated that the pan-PPAR agonist lanifibranor models of NAFLD. Acute liver injury was induced by carbon tet- ameliorated all aspects of non-alcoholic fatty liver disease in rachloride (CCl4). The role of PPARs on macrophage functionality independent experimental mouse models. Non-alcoholic fatty was studied in isolated hepatic macrophages, bone marrow- liver disease and fatty acids induce a specific polarization status derived macrophages stimulated with palmitic acid, and circu- in macrophages, which was altered by lanifibranor to increase lating monocytes from patients with NAFLD. expression of lipid handling genes, thereby decreasing inflam- Results: Lanifibranor improved all histological features of stea- mation. PPAR isoforms have differential therapeutic effects on tohepatitis in CDAA-HFD-fed mice, including liver fibrosis, fat-laden hepatocytes, activated hepatic stellate cells and in- thereby combining and exceeding specific effects of the single flammatory macrophages, supporting the clinical development PPAR agonists. Its potent anti-steatotic efficacy was confirmed in of pan-PPAR agonists. a 3D liver biochip model with primary cells. Infiltrating hepatic © 2020 European Association for the Study of the Liver. Published by monocyte-derived macrophages were reduced following PPAR Elsevier B.V. All rights reserved. agonist administration, especially with lanifibranor, even after short-term treatment, paralleling improved steatosis and hepa- Introduction titis. Lanifibranor similarly decreased steatosis, liver injury and Non-alcoholic fatty liver disease (NAFLD) has become the most monocyte infiltration in the WD model. In the acute CCl4 model, common chronic liver disease worldwide. NAFLD, particularly its neither single nor pan-PPAR agonists directly affected monocyte inflammatory form non-alcoholic steatohepatitis (NASH), can recruitment. Hepatic macrophages isolated from WD-fed mice progress to fibrosis, cirrhosis and hepatocellular carcinoma.1 In- displayed a metabolically activated phenotype. Lanifibranor sulin resistance is a central pathophysiological mechanism and attenuated the accompanying inflammatory activation in both interconnects NAFLD with its comorbidities such as visceral murine palmitic acid-stimulated bone marrow-derived obesity, type 2 diabetes and atherosclerosis.2 At present, lifestyle modification is difficult to achieve and sustain, and approved pharmacological therapy is lacking.3 Keywords: NAFLD; Fibrosis; Lanifibranor; Biochip; Therapy. Peroxisome proliferator-activated receptors (PPARs) are nu- Received 11 September 2019; received in revised form 30 March 2020; accepted 13 April clear receptors that bind fatty acids and their derivatives, and 2020; available online xxx integrate metabolic and inflammatory signalling pathways, * Corresponding author. Address: Department of Hepatology and Gastroenterology, Charité University Medicine Berlin, Augustenburger Platz 1, D-13353 Berlin, making them attractive therapeutic targets for NAFLD. Germany. Tel.: +49 (30) 450 553 022; fax: +49 (30) 450 553 902. The 3 isoforms, PPARa,PPARc and PPARd(b), have different E-mail address: [email protected] (F. Tacke). q tissue distributions and functions. PPARa exerts its main actions in Guest Editor: Robert Schwabe. † Both authors contributed equally to this manuscript. the liver, where it transcriptionally drives genes regulating gluco- 4 https://doi.org/10.1016/j.jhep.2020.04.025 neogenesis, b-oxidation, ketogenesis and lipid transport. The

Journal of Hepatology 2020 vol. - j 1–14 Research Article NAFLD and Alcohol-Related Liver Diseases hepatic expression of PPARa, but not PPARc or d, correlates with the free environment at the animal facilities of the University Hos- presence of NASH and its histological features.5 In animal models, pitals of Aachen and Ghent. Mice were given free access to food PPARa deletion is associated with a worsening of hepatic steatosis and water and housed in a 12 h light/dark cycle. whereas the selective PPARa agonist Wy-14,643 reversed NASH All in vivo experiments were approved by and conducted in and fibrosis.6,7 PPARc is the predominant isoform in adipose tissue; agreement with the appropriate institutional and governmental it controls glucose metabolism, lipogenesis and adipocyte differ- authorities. Reporting was conforming to the ARRIVE guidelines entiation, and upregulates adiponection.8,9 By promoting the for animal experiments. storage of fatty acids as triglycerides, PPARc acts as an insulin Steatohepatitis was induced by feeding 8-week old male mice sensitizer and prevents ectopic fat accumulation. Indeed, although either a choline-deficient, L-amino acid-defined, high-fat diet the hepatocyte-specific deletion of PPARc decreased hepatic stea- enriched with 2% cholesterol (CDAA-HFD) (E15673-940, Ssniff, tosis in genetically obese mice, whole-body insulin resistance was Soest, Germany) for up to 12 weeks, or a WD rich in saturated fat, aggravated.10 The role of PPARd is less clear, although it promotes sucrose and cholesterol (TD.08811 + 1% cholesterol, Ssniff) for 16 hepatic fatty acid oxidation and limits inflammation.8 weeks. Macrophages have emerged as key mediators of After 6 weeks of CDAA-HFD and 10 weeks of WD feeding, inflammation-mediated insulin resistance.2 During liver injury, mice were randomized to receive either vehicle (methylcellulose monocytes massively infiltrate the liver and differentiate into 1% + poloxamer 0.1%), the PPARa agonist fenofibrate (100 mg/kg/ proinflammatory monocyte-derived macrophages (MoMFs), day), the PPARc agonist pioglitazone (30 mg/kg/day), the PPARd which replace the resident Kupffer cells (KCs) as the dominant agonist GW501516 (10 mg/kg/day) or the pan-PPAR agonist macrophage population.11,12 Macrophages are able to respond to lanifibranor (30 mg/kg/day) once daily via oral gavage for a a variety of stimuli, including metabolic ones, which can induce period of up to 6 weeks, while diet feeding was continued. specific polarization states.13,14 PPARc negatively regulates this Acute liver injury was induced by a single intraperitoneal proinflammatory polarization, while both PPARc and PPARd are injection of carbon tetrachloride (CCl4) (Merck, Germany), dis- involved in anti-inflammatory polarization. Interestingly, dele- solved in corn oil, at 0.6 ml/kg body weight, as previously tion of either PPAR isoform in myeloid cells exacerbated insulin described.22 PPAR agonists or vehicle were given by oral gavage resistance and hepatic steatosis.15–17 directly after induction of liver injury and 24 h later. Mice were

Their multiple beneficial actions in metabolism and inflam- sacrificed 36 h after the CCl4 injection. mation indicate that pharmacological agonists of PPAR(s) represent attractive therapeutic approaches in NAFLD. Fibrates, Statistical analysis synthetic agonists of PPARa, have not shown a consistent bene- Statistical analysis was performed using Graphpad Prism 6 ficial effect in NAFLD, although large trials are lacking. In the (Graphpad Software Inc., La Jolla, CA, USA) and SPSS 25.0 (SPSS PIVENS trial, the PPARc agonist pioglitazone improved hepatic Software, IBM Corp., NY, USA). Differences between groups were steatosis, lobular inflammation and ballooning, but not fibrosis.18 compared using a one-way ANOVA with post hoc testing. A two- However, a meta-analysis on the use of pioglitazone in NASH sided p value <0.05 was considered statistically significant. indicated beneficial effects on advanced fibrosis.19 Single PPARd Continuous variables are presented as mean ± SD. agonists have faced safety concerns,8 and in a recent phase II trial, the PPARd agonist seladelpar failed to reduce liver fat as Results quantified by MR imaging (NCT03551522). Thus, the focus has Pan-PPAR agonism combines the differential effects of shifted towards the development of dual and pan-PPAR agonists. selective PPAR agonists on steatohepatitis and fibrosis The dual PPARa/d agonist elafibranor has demonstrated benefi- progression cial effects on NASH resolution in patients with highly active To investigate the efficacy of single and pan-PPAR agonists in the disease without a clear antifibrotic signal in a phase II clinical treatment of progressive steatohepatitis, we employed the trial.20 The pan-PPAR agonist lanifibranor reduced disease CDAA-HFD model, which induces severe inflammation and liver severity in 2 preclinical NAFLD models, the methionine/choline- fibrosis. Mice were fed the CDAA-HFD for 12 weeks, and PPAR deficient diet and in foz/foz mice fed a high-fat diet.21 agonist treatment was administered during the last 6 weeks of Despite advances in our understanding of the beneficial ac- diet feeding (Fig. 1A). Adequate dosing was confirmed by dif- tions of PPARs on insulin sensitivity and NAFLD, the relative ferential PPAR target engagement. Specifically, pioglitazone and potency of the different PPAR agonists in the treatment of NAFLD lanifibranor increased serum levels of adiponectin (Fig. 1B), a and their effect on macrophages have not been elucidated. main PPARc target, whereas fenofibrate and lanifibranor In this study, we assessed the therapeutic potential of lanifi- increased the hepatic expression of the PPARa target genes py- branor in comparison with single PPAR agonists in two murine ruvate dehydrogenase kinase (Pdk) 4, carnitine palmitoyl- models of NASH and fibrosis. Furthermore, we examined the func- transferase (Cpt)1b and Cpt2(Fig. S1A). Fibrates are indicated for tional consequences of PPAR agonism on macrophage biology as a the treatment of hypertriglyceridemia, whereas thiazolidine- factorcontributing to PPAR-mediated attenuation of steatohepatitis. diones have indirect effects on circulating lipids.23 In our study, lanifibranor decreased serum triglyceride levels, which also Materials and methods tended to decrease upon fenofibrate and pioglitazone treatment Part of the materials and methods are described in a (Fig. 1B). The PPAR agonists were well tolerated, and no signifi- supplementary file. cant effects on body weight or adipose tissue weight were observed, while the liver-to-body weight ratio was lower in mice Liver injury models and pharmacological treatment treated with lanifibranor and pioglitazone (Fig. S1B). Seven-week old C57BL/6J wild-type mice (Janvier Labs, Le Importantly, treatment with lanifibranor reversed steatohe- Genest-Saint-Isle, France) were housed in a specific-pathogen- patitis, as evidenced by highly significant reductions in the

2 Journal of Hepatology 2020 vol. - j 1–14 p 150 2.0 = A wk 0 wk 6 wk 12 B **** **** **** **** 0.06 *** 1.5 CDAA-HFD 100

g/ml) 1.0

Lanifibranor (g/L) (μ 50 Selective PPAR 0.5

agonists Serum adiponectin 0 Serum triglycerides 0.0 Vehicle C C+V V F P G L C C+V V F P G L CDAA-HFD CDAA-HFD C H&E Sirius red D **** **** ****# #### #### **** **** **** ### ### *** 3 8 6 2 4 1 2 per 20x field Control diet Steatosis grade

0 Inflammation foci 0 C C+V V F P G L C C+V V F P G L CDAA-HFD CDAA-HFD

** ** ##** **** **** ****# #### #### **** 2 8 6 + vehicle

CDAA-HFD 1 4 score 2 NAFLD activity

Ballooning grade 0 0 C C+V V F P G L C C+V V F P G L CDAA-HFD CDAA-HFD

EF200 n.s. n.s. 500 n.s. ****** ## #### **** ******** ####** #### ****

CDAA-HFD 400 + Fenofibrate 150 300 100 200

50 (μg/g liver ) Hydroxyproline Triglycerides 100

(mg/g liver tissue) 0 0 C C+V V F P G L C C+V V F P G L CDAA-HFD CDAA-HFD CDAA-HFD + Pioglitazone 10 n.s. GH8 = 0.06 ****** * * ** ******** p p = 0.07 ** 6 8 6

4 mRNA 4 2 fold change 2 α-SMA Collagen area (%)

CDAA-HFD 0 0 + GW501516 CVFPGL C C+V V F P G L CDAA-HFD CDAA-HFD

Control (C) Control + vehicle (C + V) Vehicle (V) Fenofibrate (F) Pioglitazone (P) CDAA-HFD

CDAA-HFD GW501516 (G) + Lanifibranor Lanifibranor (L)

Fig. 1. Therapeutic administration of PPAR agonists ameliorates steatohepatitis and fibrosis. (A) Starting after 6 weeks of experimental steatohepatitis diet, CDAA-HFD diet-fed mice were treated once daily in a therapeutic setting for 6 weeks with single PPARa, c and d agonists and the pan-PPAR agonist lanifibranor. (B) Serum adiponectin and triglyceride levels. (C) Representative H&E and Sirius red staining (magnification 100×). (D) Scoring of histological features of steatosis, lobular inflammation and ballooning, and NAFLD activity score. (E) Quantification of liver triglyceride content. (F–G) Liver fibrosis was assessed by quantification of hepatic hydroxyproline content (F) and the fraction of Sirius red positive area (G). (H) Expression of Acta2 (aSMA). Data are presented as mean ± SD (n = 6 for the control group and n = 8 for the other groups). *p <0.05; **p <0.01; ***p <0.001; ****p <0.0001 (one-way ANOVA with post hoc test). # denotes the level of significance vs. lanifibranor-treated mice. Data are pooled from 2 independent experiments. CDAA-HFD, choline-deficient, amino acid-defined high-fat diet; NAFLD, non-alcoholic fatty liver disease.

NAFLD activity score as well as the subcomponent scoring of Fenofibrate improved these scores, especially steatosis, to a steatosis (validated by reduced hepatic triglyceride content), lesser extent, whereas pioglitazone and GW501516 did not lobular inflammation and hepatocellular ballooning (Fig. 1C–E). significantly impact the histological disease severity. Alanine

Journal of Hepatology 2020 vol. - j 1–14 3 Research Article NAFLD and Alcohol-Related Liver Diseases aminotransferase (ALT) levels also tended to be lower in Lanifibranor decreased the number of hepatic macrophages lanifibranor-treated mice (Fig. S1C). Lanifibranor ameliorated after just 2 weeks of treatment, as demonstrated on F4/80 liver fibrosis, with significant reductions in collagen area, liver immunohistochemistry and hepatic Ccr2 gene expression hydroxyproline, and reduced expression of fibrogenic mediators (Fig. 3B,E). Strikingly, the proportion and absolute number of (Fig. 1C, F–H and Fig. S1D). The single PPAR agonists improved hepatic MoMFs were reduced specifically by lanifibranor fibrosis to a lesser extent, with fenofibrate exerting a stronger compared to the single PPAR agonists, which was accompanied effect in mice than the PPARc and d agonists. by a significant decrease in the overall number of leukocytes (Fig. 3F–G; Fig. S3D). In agreement with the long-term treatment, Pan-PPAR agonism inhibits macrophage accumulation in the hepatic monocytes were decreased as well, to the greatest extent steatohepatitis/fibrosis model by lanifibranor, whereas the liver lymphocyte compartment As liver MoMF are key drivers of NASH and fibrosis progres- remained unaffected and the number of blood leukocytes was sion,2,12 we studied the impact of PPAR agonism on the hepatic decreased by all PPAR agonists (Fig. 3F–G; Fig. S3E). immune cell composition. PPAR agonists, especially lanifibranor, reduced the number of intrahepatic macrophages, as assessed by PPAR isoforms combine to improve NAFLD in an obese mouse F4/80 immunohistochemistry as well as F4/80 and Ccr2 mRNA model expression (Fig. 2A). We validated these findings using flow Although the CDAA-HFD diet induces severe inflammation and cytometry and observed a marked decrease in the proportion of progression to fibrosis, obesity and insulin resistance do not infiltrating MoMFs upon lanifibranor treatment, which was develop in this model.26 We therefore investigated the PPAR significantly more pronounced than upon treatment with either agonists in a NAFLD mouse model with concomitant obesity and single PPAR agonist (Fig. 2B). Lanifibranor similarly reduced metabolic syndrome. Mice were fed a WD rich in fat, sucrose and infiltrating monocytes, whereas KCs, which are depleted in cholesterol for 10 weeks, after which daily treatment was experimental NASH,24,25 were not affected (Fig. 2B). Notably, the administered for 6 weeks (Fig. 4A). Fenofibrate, GW501516 and intrahepatic lymphocyte populations remained unaffected by lanifibranor caused mild weight loss, which was not observed treatment, or were only relatively increased due to the sharp following pioglitazone administration (Fig. 4B). This was asso- decrease in MoMFs (Fig. S2A). ciated with a significant decrease in adipose tissue weight Treatment with all PPAR agonists decreased the proportion of (Fig. 4C). In accordance with known beneficial metabolic effects, circulating blood monocytes to levels comparable with the fenofibrate and lanifibranor decreased serum triglycerides controls, in part through a reduction in immature Ly6C+ mono- (Fig. 4C). Lanifibranor improved liver injury, as evidenced by cytes (Fig. 2C). Blood granulocytes and lymphocyte subsets were decreased serum ALT levels (Fig. 4D). unaltered (Fig. S2B). Histologically, WD feeding caused a less severe NAFLD The reduction in liver MoMFs was accompanied by reduced phenotype compared to CDAA-HFD feeding, with marked he- expression of the proinflammatory mediators Tnfa, iNos, Il6 and patic steatosis, mild to moderate inflammation yet absent Cxcl2, which was most pronounced following lanifibranor ballooning and fibrosis. Lanifibranor decreased both steatosis treatment (Fig. 2D). and inflammation. The former was mostly related to PPARa These data indicate that pan-PPAR agonism inhibits MoMF (although PPARd agonism had a partial effect), while the latter infiltration to a significantly larger extent than obtained by could be linked to a PPARd effect. As such, the composite NAFLD stimulation with single PPAR isoforms, which might contribute activity score was most impacted by lanifibranor (Fig. 4E). to reducing the hallmarks of disease, such as triglyceride accu- Both PPARa and PPARd agonists reduced the infiltration with mulation, steatohepatitis and fibrosis. monocytes and MoMF, with lanifibranor having the most significant effects. Moreover, the absolute number of leukocytes in Reduced NASH activity and macrophage infiltration are early the liver were decreased following PPARa,PPARd and lanifi- consequences of lanifibranor treatment in experimental branor treatment, to a similar level as chow diet-fed control mice steatohepatitis (Fig. 4F). To examine the pathophysiological sequence of steatohepatitis amelioration upon treatment with PPAR agonists, we analysed Lanifibranor improves steatosis in a 3D liver biochip but not the effects of short-term treatment, in which PPAR agonists were 2D primary hepatocyte culture administered for 2 weeks in mice that had been subjected to 6 Hepatic triglycerides rapidly decreased after lanifibranor treat- weeks of CDAA-HFD (Fig. 3A). Adequate dosing was again ment in experimental steatohepatitis in mice (Fig. 3D), in line confirmed by PPAR isoform-specific target engagement with the well-documented beneficial metabolic effects of (Fig. S3A). Similar to the long-term treatment, lanifibranor and, different PPAR agonists.8,27 To confirm this mode of action, we to a lesser degree, fenofibrate, ameliorated NASH as shown by a tested the anti-steatotic efficacy of lanifibranor in vitro. Cultured reduced grading of steatosis, lobular inflammation and primary murine hepatocytes were stimulated for 24 h or 48 h ballooning, and thus the overall NAFLD activity score (Fig. 3B–C; with a mixture of oleic acid (OA) and palmitic acid (PA) in the Fig. S3B). This was accompanied by a significant reduction in presence or absence of lanifibranor. Unexpectedly, lanifibranor hepatic lipid content and expression of inflammatory cytokines treatment did not attenuate hepatocyte fatty acid accumulation. in lanifibranor-treated mice, whereas fenofibrate did not Indeed, treatment moderately elevated intracellular lipids in improve these variables. Pioglitazone and GW501516 did not control hepatocytes (Fig. 5A). have significant effects on either of these disease markers nor We then examined hepatocyte steatosis in a 3D liver biochip, liver histology. Not surprisingly for short-term treatment, liver which more closely mimics the in vivo liver anatomy by assem- fibrosis was only modestly, yet significantly, improved by lani- bling the different parenchymal and non-parenchymal cell types. fibranor and not by single PPAR agonists (Fig. 3B–D; Fig. S3C). We advanced a previously reported culture system that had been

4 Journal of Hepatology 2020 vol. - j 1–14 A F4/80 10 ** * 8 Control (C) Control diet CDAA-HFD CDAA-HFD + vehicle + Fenofibrate Control + vehicle (C + V) 6 Vehicle (V) Fenofibrate (F) 4 Pioglitazone (P) CDAA-HFD 2 GW501516 (G) F4/80 staining Lanfibranor (L) surface area (%) 0 CVFPGL CDAA-HFD 8 n.s. 20 CDAA-HFD CDAA-HFD CDAA-HFD ******** **** * **** ******** **** *** **** **** + Pioglitazone + GW501516 + Lanfibranor 6 15

4 10 (F4/80) Ccr2 2 5 Emr1 mRNA fold change mRNA mRNA fold change mRNA 0 0 C C+V V F P G L C C+V V F P G L CDAA-HFD CDAA-HFD CDAA-HFD CDAA-HFD B Control diet 30 n.s. + vehicle + Lanfibranor ******** ***## ####* #### **** 20

MoMF 10 Mo Mo Mo 0 (% of live leukocytes) C C+V V F P G L

CDAA-HFD Ly6C 10 n.s. n.s. 15 **** **** ******** *** ### ### **** KC KC KC 8 10 6 4 5 Monocytes 2 cells Kupffer MoMF MoMF MoMF 0 0 (% of live leukocytes) (% of live leukocytes) Tim4 CD11b C C+V V F P G L C C+V V F P G L F4/80 CDAA-HFD CDAA-HFD CDAA-HFD CDAA-HFD 30 15 C Control diet + vehicle + Lanfibranor ******** *** ** ** ** ** *

20 10 NK NK NK 10 5 Monocytes (% of leukocytes) (% of leukocytes) 0 + monocytes Ly6C 0 Mo Mo Mo C C+V V F P G L C C+V V F P G L NK1.1 CD115 CDAA-HFD CDAA-HFD D Tnf-α Inos Il-6 Cxcl-2 3 20 n.s. 10 ******** **** ** **** 8 ******** **** *** *** **** ******* **** ** * **** ******** **** * **** 8 6 15 2 6 4 10 4 1 5 2 2

mRNA fold change mRNA 0 0 0 0 C C+V V F P G L C C+V V F P G L C C+V V F P G L C C+V V F P G L CDAA-HFD CDAA-HFD CDAA-HFD CDAA-HFD

Fig. 2. Lanifibranor treatment decreases macrophage infiltration and inflammation in the CDAA-HFD model. (A) F4/80 immunohistochemistry with representative sections (magnification 100×) and quantification. Expression of Ccr2 and Emr1 (F4/80) in liver tissue. (B) Representative flow cytometric plots and quantification of liver MoMFs, monocytes and Kupffer cells. (C) Representative flow cytometric plots and quantification of blood monocytes and Ly6C+ blood monocytes. (D) Hepatic expression of proinflammatory markers. Data are presented as mean ± SD (n = 6 for the control group and n = 8 for the other groups). *p <0.05; **p <0.01; ***p <0.001; ****p <0.0001 (one-way ANOVA with post hoc test). # denotes the level of significance vs. lanifibranor-treated mice. Data are pooled from 2 independent experiments. CDAA-HFD, choline-deficient, amino acid-defined high-fat diet; MoMFs, monocyte-derived macrophages.

Journal of Hepatology 2020 vol. - j 1–14 5 Research Article NAFLD and Alcohol-Related Liver Diseases

p = wk 0 wk 6 wk 8 0.14 A C 4 n.s. n.s. n.s. **** **** ****#### #### **** ******** ** 8 CDAA-HFD 3 6 Lanifibranor 2 4 Selective PPAR agonists score 2 1 NAFLD activity Vehicle 0 Collagen area (%) 0 C C+V V F P G L C C+V V F P G L CDAA-HFD CDAA-HFD ******** n.s. n.s. n.s. * H&E Sirius red F4/80 500 ******** n.s. * n.s. **** 600 ## # B D #### #### #### 400 300 400 200 200

Triglycerides 100 Control diet (mg/g liver tissue) 0 0

C C+V V F P G L Hydroxyproline (μg/g liver) C C+V V F P G L CDAA-HFD CDAA-HFD 6 n.s. n.s. n.s. E ******** ** 20 ******** * * n.s. **** 15 4 + vehicle CDAA-HFD 10 2 5 F4/80 staining (surface area %) 0 0

C C+V V F P G L fold change Ccr2 mRNA C C+V V F P G L CDAA-HFD CDAA-HFD ******** ** p = n.s. CDAA-HFD 25 ## ### ## 10 + Fenofibrate ******** 0.06 * *** F 20 8 15 6 10 4 MoMF

5 Monocytes 2 0 0 (% of live leukocytes) (% of live leukocytes) CDAA-HFD

+ Pioglitazone C C+V V F P G L C C+V V F P G L CDAA-HFD CDAA-HFD n.s. n.s. 6 G 8.0x10 ******** * ## **

6.0x106

4.0x106 CDAA-HFD + GW501516 2.0x106 Leukocytes (/g liver tissue) 0 C C+V V F P G L Control (C) Control + vehicle (C + V) CDAA-HFD Vehicle (V) Fenofibrate (F) 7 2.0x10 ******** * *** ** ** Pioglitazone (P) CDAA-HFD CDAA-HFD

+ Lanifibranor GW501516 (G) 1.5x107 Lanifibranor (L)

1.0x107

5.0x106 Leukocytes (number/ml) 0 C C+V V F P G L CDAA-HFD

Fig. 3. Short-term pan-PPAR agonist treatment reduces steatohepatitis and macrophage accumulation. (A) Starting after 6 weeks of experimental steatohepatitis diet, CDAA-HFD diet-fed mice were treated once daily in a therapeutic setting for 2 weeks with single PPAR agonists or the pan-PPAR agonist lanifibranor. (B) Representative H&E, Sirius red (magnification 100×) and F4/80 (magnification 200×) immunohistochemistry stainings. (C) NAFLD activity score and quantification of liver fibrosis as the fraction of Sirius red positive area. (D) Quantification of liver triglyceride and hydroxyproline content. (E) F4/80 positive area fraction and Ccr2 gene expression. (F) Liver MoMFs and monocytes, determined by flow cytometry with gating described as in Fig. 2. (G) Absolute numbers of liver leukocytes per gram of liver tissue and blood leukocytes per ml of blood. Data are presented as mean ± SD (n = 6 for both control groups and n = 7–8 for the CDAA-HFD other groups). *p <0.05; **p <0.01; ***p <0.001; ****p <0.0001 (one-way ANOVA with post hoc test). # denotes the level of significance vs. lanifibranor- treated mice. Data are pooled from 2 independent experiments. CDAA-HFD, choline-deficient, amino acid-defined high-fat diet; MoMFs, monocyte-derived macrophages.

6 Journal of Hepatology 2020 vol. - j 1–14 established with cell lines28 by seeding primary murine hepa- reduced expression of the latter while further enhancing tocytes and stellate cells into a “hepatic chamber”, whereas expression of lipid metabolism-related genes (Fig. 7A). primary endothelial and KCs were seeded into the “portal It was previously reported that PPARc was able to counter chamber”, with both chambers being separated by a porous MMe inflammatory activation, if induced in vitro through stim- membrane (Fig. 5B). The addition of a PA/OA mixture led to fatty ulation with FFAs.30 Notably, both isolated MoMFs and cultured acid accumulation in hepatocytes, which was reduced by bone marrow-derived macrophages (BMDMs) expressed high approximately 33% upon simultaneous treatment with lanifi- levels of PPARc and d, and low levels of PPARa, which remained branor (Fig. 5C). unaltered after WD feeding or PA stimulation, respectively The discrepancy between the 2D and 3D culture may indicate (Fig. 7A, Fig. S4A–B). Stimulation with the saturated fatty acid PA that PPAR activation in non-parenchymal cells, which interact induced the expression of inflammatory genes as well as genes with and influence hepatocytes, is required to exert its anti- characteristic of MMe polarization, such as Plin2 and Cd36, but steatotic actions. not lysosomal-associated membrane protein 2 (Lamp2)(Fig. 7B; Fig. S4C). Treatment with lanifibranor, but not single PPAR ago- Lanifibranor counteracts stellate cell activation nists, further increased the expression of plin2 and cd36 (Fig. 7B), 31 fi Since lanifibranor ameliorated fibrosis in the CDAA-HFD model, which enable macrophages to handle excess fat. Only lani - fl we examined its effects on primary human stellate cells in vitro. branor reduced the in ammatory gene expression induced by PA Lanifibranor reduced the production of a-SMA by stellate cells stimulation (Fig. 7B), indicating that the synergistic involvement fl activated either by TGF-b stimulation or by the stiffness of the of multiple PPAR isoforms is required to decrease in ammation d culture plastic (Fig. 5D). We obtained similar dose-dependent and improve lipid handling in macrophages. Of these, PPAR results in a therapeutic setting, when stellate cells had been seems the major isoform involved, as the administration of a d activated on stiff plastic by 7 days of culture before the addition PPAR antagonist (GSK0660) either in isolation or combined fi of lanifibranor (Fig. 5E). with lani branor increased the expression of Il6 and impaired that of lipid metabolism genes (Fig. 7C). PPARa antagonism pri- marily affected lipid metabolic genes, whereas the effects of fi PPAR agonists do not directly inhibit monocyte in ltration PPARc antagonism were minor (Fig. S4D). Importantly, multiplex Although PPARs have been studied in the context of NAFLD, their immunohistochemistry staining confirmed the expression of effects on hepatic macrophages are less clear. As we observed PPARd in hepatic macrophages in experimental NAFLD in vivo – strong effects on hepatic macrophage accumulation (Figs. 2 3), (Fig. 7D; Fig. S4E). we focused on the impact of PPAR agonism on macrophage biology. Circulating monocytes are metabolically activated in patients fi We rst investigated whether PPAR agonists directly inhibit with fibrosing NAFLD fi leukocyte in ltration into the liver, by means of a single injection The MMe phenotype is highly conserved between animal models of CCl4. The acute hepatocyte damage following CCl4 provokes a and humans. Adipose tissue macrophages isolated from the 22 massive macrophage recruitment to the injured liver, thereby omental and subcutaneous adipose tissue from obese individuals serving as an in vivo model to assess monocyte chemotaxis to the exhibited a similar upregulation of inflammatory and metabolic liver (Fig. 6A). Liver injury was assessed by serum ALT levels genes as macrophages from HFD-fed mice, whereas markers of fi (Fig. 6B) and was accompanied by massive leukocyte in ltration, ‘classical’ M1 activation were not induced.30 To ascertain if the which was unaffected by treatment with either PPAR agonist reversal of this phenotype by lanifibranor in vitro is relevant to – fi (Fig. 6C D). Importantly, the increased in ltration of monocytes human NAFLD, we investigated whether circulating monocytes fi – and MoMFs was not attenuated by lani branor (Fig. 6C F). in human patients are comparably polarized and whether this To validate this conclusion, we performed an in vitro correlated to the severity of NAFLD. chemotaxis assay in which bone marrow leukocytes were stim- The clinical and biochemical patient characteristics (n = 26) ulated to migrate in a transwell chamber by the chemokine CCL2. are documented in Table S1. Lean healthy volunteers (n = 6; fi In accordance with the CCl4 experiment, the addition of lani - median Fibroscan value 5.0 kPa, median controlled attenuation branor did not change spontaneous or CCL2- induced migration parameter 199 dB/m) served as controls. We performed flow of monocytes (Fig. 6G). cytometric and RNA expression analysis of classical monocytes Thus, in contrast to chemokine and chemokine receptor an- (CD14+ CD16−) as the most abundant subset, which remained tagonists undergoing evaluation in NAFLD, for instance CCR2/5 proportionally stable between controls and increasing stages of 29 fi inhibitors, lani branor does not directly inhibit hepatic MoMF NAFLD (Fig. 8A). PPARd was the most abundant isoform, and its recruitment. expression was not influenced by the presence of NAFLD (Fig. 8B). PPARd activation inhibits macrophage fatty acid-induced Monocytes isolated from patients with NAFLD expressed proinflammatory activation elevated levels of the inflammatory cytokines IL-6, TNF-a and IL- Using single-cell RNA sequencing techniques, we recently iden- 1b, which reached statistical significance in patients with fibrosis tified the distinct ‘metabolically activated’ macrophage (also (Fig. 8C). Notably, these genes tended to be downregulated in termed MMe) phenotype in the hepatic and bone marrow patients with decompensated cirrhosis. Closely mirroring PA- compartments in mice fed a Western-style diet.13 Upon isolation stimulated BMDMs, as well as MoMF isolation from livers of from livers of mice fed the WD, MoMF displayed this particular mice fed the WD, CD36 and PLIN2 were upregulated in these gene expression profile, characterized by an increased expression monocytes as well (Fig. 8D). Next, we isolated monocytes from of lipid metabolism genes (Cd36, perilipin-2 (Plin2)) and in- healthy controls and patients with NAFLD without fibrosis and flammatory markers (Ccl2, Trem2). Lanifibranor treatment cultured these for 24 h in the presence or absence of lanifibranor

Journal of Hepatology 2020 vol. - j 1–14 7 Research Article NAFLD and Alcohol-Related Liver Diseases

A wk 0 wk 10 wk 16 B Start treatment Control (C) 45 Vehicle (V) p = Fenofibrate (F) 10 * 0.07 * Control diet 40 Pioglitazone (P) GW501516 (G) 5 Vehicle Lanifibranor (L) 35 0 Western diet 30 -5 Lanifibranor -10 since Rx start

Body weight (g) 25 Selective -15 Body weight evolution PPAR agonists 0 C+V V F P G L Vehicle 042 6 8 10 12 14 16 Western diet Diet (weeks) C EF 8 ******** * *** **** *** * **** 6 3 25 ****n.s. * ** 20 4 Control 2 15 2 10 (% body weight) 1 Gonadal AT weight Gonadal AT 0 Monocytes 5

C+V V F P G L Steatosis grade (%of leukocytes) 0 0 Western diet C+V V F P G L C+V V F P G L 80 n.s. * ** WD + vehicle Western diet Western diet 60 # 40 ** p 5 10 = n.s. ****0.06 * 20 4 8 WD + Fenofibrate Triglycerides (mg/dl) Triglycerides 0 3 6 C+V VFPGL 2 4 MoMF Western diet

per 20x field 1 2 Inflammatory foci (% live leukocytes) 0 D 500 n.s. n.s. 0 *** * C+V V F P G L C+V V F P G L

400 WD + Western diet Western diet 300 Pioglitazone 200 ALT (U/L) ALT 100 **** ** *** **** p = 6 2.0x106 ***** 0.08. ** 0 C+V V F P G L 1.5x106 WD + 4

Western diet GW501516 1.0x106 2 5

Leukocytes 5.0x10

Control + vehicle (C + V) (/ g liver tissue)

Vehicle (V) NAFLD activity score 0 0 Fenofibrate (F) Western C+V V F PGL C+V V F P G L Pioglitazone (P)

diet (WD) WD + GW501516 (G) Western diet Western diet

Lanifibranor (L) Lanifibranor

Fig. 4. PPAR agonists reverse NAFLD and macrophage infiltration in an obese mouse model. (A) After 10 weeks of WD feeding, mice were treated once daily in a therapeutic setting for 6 weeks with single PPARa, c and d agonists and the pan-PPAR agonist lanifibranor. (B) Body weight evolution (C) Gonadal adipose tissue weight and serum triglyceride levels. (D) Serum ALT levels. (E) Representative H&E staining (magnification 100×) and scoring of histological features of steatosis and lobular inflammation, and NAFLD activity score. (F) Quantification of infiltrating monocytes and MoMF, and absolute number of liver leukocytes. Data are presented as mean ± SD (n = 7–8 per group). *p <0.05; **p <0.01; ***p <0.001; ****p <0.0001 (one-way ANOVA with post hoc test). ALT, alanine aminotransferase; NAFLD, non-alcoholic fatty liver disease; MoMFs, monocyte-derived macrophages; WD, Western diet. or GW501516. Both compounds decreased the expression of CCL2 as well as systemically, but also modulate inflammatory signal- and increased that of PLIN2, the former more efficiently ling pathways.4 In this paper, we investigated the therapeutic following lanifibranor (Fig. 8E). potential of lanifibranor in NAFLD mouse models and explored the PPARd-dependent regulation of macrophage activation in Discussion NASH. PPAR agonists have long been interesting drug candidates for Macrophages are central regulators of inflammation-induced NAFLD given their multiple (beneficial) effects on metabolic insulin resistance in the liver, adipose tissue and sites of ectopic pathways. Indeed, the PPAR-a/d agonist elafibranor induced lipid accumulation.32 In the liver, self-sustaining, yolk sac- resolution of NASH without fibrosis worsening in a post hoc derived KCs can be distinguished from the more immunogenic analysis of a relatively large phase II clinical trial.20 PPARs not MoMFs, which derive from infiltrating monocytes during active only perform a plethora of metabolic functions, both in the liver liver injury. Apart from aggravating inflammation, MoMFs

8 Journal of Hepatology 2020 vol. - j 1–14 n.s. A Control LanifibranorFFA FFA + Lanfibranor 10 **** **** 8 n.s. 6 Control 24 h **** **** Lanifibranor (L) 4 FFA FFA + Lanifibranor (FFA + L) AdipoRed 2 **** **** AdipoRed (MFI fold increase) 0

48 h L L L L FFA FFA Control FFA + Control FFA + 24 h 48 h **** Ports Primary KCs and B Membranes C vascular layer endothelial cells AdipoRed 5 **** **** ControlFFA FFA + Lanifibranor 4 3 2

AdipoRed 1

(MFI fold increase) 0

L Ports Primary hepatocytes FFA hepatic layer Control and stellate cells FFA +

DETGF-β1 Lanifibranor Lanifibranor (L) Lanifibranor (L) DMSO 0.1% Lanifibranor (L) DMSO 0.1% (L) D10 (μM) D14 (μM) DMSO 0.1% D10 3 μM 10 μM 30 μM D4 D7 3 μM D4 D7 3 10 30 DMSO 0.1% D14 31030

Vinculin 124 kDa

α-SMA 42 kDa

*** ** 2.5 ** n.s. 2.0 ** 20 **** *** 2.0 *** ** ** 1.5 15 1.5 1.0 10 1.0 0.5 5 (fold change) (fold change) (fold change) 0.5 α-SMA concentration α-SMA α-SMA concentration α-SMA α-SMA concentration α-SMA 0.0 0.0 0 M M M M D4 D7 M D7 DMSO DMSOL 3 μ 10 μM 30 μ DMSOL 3 μ 10 μM 30 μ L 3 μM 10 μM L L L L DMSO L L 30 μM DMSO DMSOD4 LD7 3 μ + TGFβ D10 D14

Fig. 5. Lanifibranor attenuates lipid accumulation and stellate cell activation. (A) 2D single layer culture of primary mouse hepatocytes. AdipoRed assay fluorescence images and quantification of MFI in hepatocytes treated as indicated for 24 h or 48 h. (B) Schematic of the 3D liver biochip that consists of hepatic and vascular layers separated by a porous PET-membrane. The “hepatic chamber” contains primary hepatocytes and hepatic stellate cells from mice, the “portal chamber” contains primary KCs and liver sinusoidal endothelial cells from mice. (C) AdipoRed assay fluorescence images and quantification of MFI of hepatocytes in a biochip treated as indicated for 48 h in biochips. (D–E) Primary human hepatic stellate cells were activated by TGF-b or prolonged culture on plastic plates. a- SMA concentration was determined by Western blot and normalized to Vinculin. Data are presented as mean ± SD. **p <0.01; ***p <0.001; ****p <0.0001 (one-way ANOVA with post hoc test). KCs, Kupffer cells; MFI, mean fluorescence intensity. stimulate liver disease progression trough the secretion of affecting monocyte infiltration, for instance antagonists of the fibrotic and angiogenic factors, and promote stellate cell sur- CCL2/CCR2 chemokine axis.2,29 vival.2,29 Collectively, the rapid and specific inhibition of hepatic The activation pattern of hepatic macrophage subsets is sha- MoMF accumulation by lanifibranor, preceding the regression of ped by the integration of signals from overnutrition, the gut, liver fibrosis, suggests this cell type may be a major target metaflammation and from the local environment of a steatotic through which pan-PPAR agonism ameliorates NAFLD, in addi- liver. These cues act on the highly plastic macrophages to induce tion to or in synergy with beneficial metabolic effects. Interest- unique polarization states that extend beyond the classical M1- ingly, CCl4 experiments suggested that lanifibranor treatment M2 concept. Adipose tissue macrophages display an Mme reduced hepatic monocyte accumulation only indirectly. As such, phenotype in obesity, characterized by the increased expression PPAR agonists could potentially be combined with drugs of proinflammatory cytokines, albeit to a milder degree than in

Journal of Hepatology 2020 vol. - j 1–14 9 Research Article NAFLD and Alcohol-Related Liver Diseases

CCl IP A 4 B 10,000 * n.s. 0 h 24 h 36 h 8,000 Control (C) Vehicle (V) Lanifibranor Lanifibranor 6,000 Fenofibrate (F) Pioglitazone (P) CCl4 Selective Selective 4,000 GW501516 (G) PPAR agonists PPAR agonists (U/L) ALT 2,000 Lanifibranor (L) Vehicle Vehicle 0 CVFPGL

CCl4

C Control CCl4 + vehicle CCl4 + Fenofibrate CCl4 + Pioglitazone CCl4 + GW501516 CCl4 + Lanifibranor F4/80

MoMF MoMF MoMF MoMF MoMF MoMF

KC KC KC KC KC KC CD11b F4/80

D 6 E F 6x10 ** 25 * 25 5 *** 20 20 4 4x106 15 15 3

MoMF 10 2x106 10 2 Monocytes Leukocytes 5 5 F4/80 staining 1 (/g liver tissue) (surface area %) (% of live leukocytes) 0 (% of live leukocytes) 0 0 0 CVFPGL CVFPGL CVFPGL CVFPGL

CCl4 CCl4 CCl4 CCl4 G n.s. n.s. n.s. n.s. + CCL2 5 ** ** 25 + Lanifibranor ** ** (L) 4 20 monocytes 3 monocytes 15 + + 2 10 Ly6C Ly6C + Bone marrow 1 + 5 (% of living cells) leukocytes (% of living cells) 0 0 CD115 C L CD11b C L

CCL2 CCL2+ L CCL2 CCL2+ L

Fig. 6. PPAR agonists do not impact leukocyte infiltration into injured liver. (A) Mice received a single intraperitoneal injection with CCl4 and were treated at 0 h and 24 h with a vehicle, single or pan-PPAR agonist and sacrificed after 36 h. (B) Serum ALT levels. (C) Representative F4/80 immunohistochemistry staining (magnification 100×) and flow cytometric plots. (D) Absolute number of liver leukocytes, per gram of liver tissue. (E) Quantification of liver MoMFs and monocytes. (F) Quantification of F4/80 staining positive area fraction (n = 5 per group). (G) Bone marrow cells were harvested for a transwell migration assay. Quantification of migrated monocytes following stimulation with CCL2 and/or lanifibranor treatment. Data are presented as mean ± SD *p <0.05; **p <0.01; ***p

<0.001; ****p <0.0001 (one-way ANOVA with post hoc test). ALT, alanine aminotransferase; CCl4, carbon tetrachloride; MoMFs, monocyte-derived macrophages.

M1 macrophages, as well as genes involved in lipid metabolism reproduced in vitro by the saturated fatty acid PA. Here, we found and lysosome biogenesis.30,33 We now report that human that lanifibranor reduced the expression of proinflammatory circulating CD14+ CD16− classical monocytes exhibit a similar mediators while upregulating genes involved in lipid meta- polarization status in patients with NAFLD, especially in more bolism. Our data suggest the possibility of uncoupling these key advanced stages of fibrosis. Notably, this proinflammatory MMe functions as a strategy to reduce insulin resistance and pattern was absent in patients with decompensated cirrhosis, NAFLD. Follow-up experiments with isoform-specific PPAR in- possibly due to immune exhaustion at this stage.34 hibitors pointed to PPARd as the main, but not exclusive, medi- We recently identified a highly similar MMe phenotype in ator of these therapeutic effects. Stimulation of human hepatic macrophages as well as bone marrow myeloid cells monocytes in vitro remarkably mimicked the findings in PA- isolated from Western-style diet-fed mice,13 which could be stimulated BMDMs.

10 Journal of Hepatology 2020 vol. - j 1–14 wk 0 wk 10 wk 16 Control + Western diet + A 800 vehicle vehicle Control diet MoMF 600

Vehicle in controls) 400 Western diet

PPARα 200 Lanifibranor

Relative expression 0 Selective (fold to α γ δ rγ PPAR agonists CD11b a Ppar Ppar Ppar Pparα Pp Pparδ Vehicle F4/80 PA stimulation B PPAR agonists/antagonists 24 h High in control High in WD High in WD High in WD + vehicle + vehicle + vehicle + Lanifibranor + M-CSF 7 days Plin2 Plin2 Bone marrow BMDMs *** ** leukocytes Cd36 Cd36 *** Cd36 * Plin2 ** Trem2 Trem2 * ** 15 8 Cxcr4 ** Cxcr4 * ** **** *** 6 Ccl2 **** Ccl2 *** 10 4 Il1β Il1β * 5 2 Il10 Il10 mRNA fold change mRNA 0 0 -6 -4 -2 0 246 -4 -2 0 24 CPAF P G L CPAF P G L Log fold change Log fold change 2 2 + PA + PA Il6 Tnfα C Il6 Plin2 Cd36 **** * * n.s. n.s. *** *** p **** ** = 0.07 * **** **** **** **** **** 15 8 4 30 **** 4 *** * n.s. **** *** 6 3 3 10 20 4 2 2 5 10 2 1 1 mRNA fold change mRNA mRNA fold change mRNA 0 0 0 0 0 PA - ++++ - ++++ - ++++ CPAF P G L CPAF P G L Lanifibranor ---++ ---++ ---++ GSK0660 --+ - + --+ - + --+ - + + PA + PA Il1β D CDAA-HFD + vehicle * p = 0.07 PPARγ IBA1 (macrophages) DAPI Merge 6 * Control (C) 4 PA Fenofibrate (F) Pioglitazone (P) + PA 2 GW501516 (G) Lanifibranor (L)

mRNA fold change mRNA 0 CPAF P G L + PA

Fig. 7. Lanifibranor alters metabolic macrophage activation through PPARd. (A) MoMFs were isolated from mice fed the WD as in Fig. 4. Gene expression analysis of PPAR isoforms and macrophage activation markers. (B–C) BMDMs were stimulated with PA and PPAR agonists and/or the PPARd antagonist GSK0660. mRNA expression of lipid metabolism and proinflammatory genes following treatment with PPAR agonists (B) or lanifibranor and GSK0660 (C). Data are presented as mean ± SD (n = 3 per group). (D) Immunofluorescent staining for PPARd, the macrophage marker IBA1 and DAPI, with overlay image. *p <0.05; **p <0.01; ***p <0.001; ****p <0.0001 (one-way ANOVA with post hoc test). BMDMs, bone marrow-derived macrophages; M-CSF, monocyte-colony stimulating factor; MoMFs, monocyte-derived macrophages; WD, Western diet.

These data reveal the striking similarity between hepatic Here, we show that lanifibranor ameliorated all histological macrophages in vivo, fatty acid-stimulated macrophages in vitro features of NASH in mice. The pan-PPAR agonist lanifibranor and circulating monocytes in patients with NAFLD, reinforcing thereby synergistically combines differential effects of single the concept that metabolic programming in innate immune cells PPAR agonists in experimental steatohepatitis. Our data suggest regulates inflammation in obesity and NAFLD. that the amelioration of liver steatosis was mostly achieved via

Journal of Hepatology 2020 vol. - j 1–14 11 Research Article NAFLD and Alcohol-Related Liver Diseases

A Classical monocytes B **** **** 15 n.s. **** ** **** Classical Intermediate 100 80 Control 10 PPARα F0-F1 in controls) PPARγ 60 α PPARδ F2-F4 NAFLD 40 Decompensated 5

cirrhosis PPAR Non 20 % of monocytes classical Relative expression 0

CD14 0 (fold to CD16 Healthy NAFLD controls patients

IL-6 TNF-α IL-1β CCL2 Healthy control 1 C E Healthy control 2 NAFLD patient 1 ** NAFLD patient 2 n.s. **** * ** * * NAFLD patient 3 16 16 32 5 ******** C: Control 8 8 16 G: GW501516 8 4 L: Lanifibranor 4 4 2 4 3 2 2 * 1 1 2 * 1 0.5 **** *** 0.5 1 * **** 0.5 0.25 0.25

mRNA fold change mRNA 0.25 0.125 0.125 0 mRNA fold expression mRNA CLG CLG CLG CLG CLG D CD36 PLIN2 LAMP2 PLIN2 4 ** 4 * 2 6 *** 2 2 4 **** 1 ** 1 1 2 **

mRNA fold change mRNA 0.5 0.5 0.5 0 mRNA fold expression mRNA CL G CL G CL G CL G CL G

Fig. 8. Monocytes isolated from patients with NAFLD display the metabolic activation phenotype. Peripheral blood mononuclear cells were isolated from whole blood, obtained from healthy controls (n = 6) and patients with NAFLD (n = 26), and subsequently stained for monocyte-specific markers. (A) Repre- sentative flow cytometric plot and quantification of the classical monocyte subset (CD3−, CD20−, CD56−, CD14+, CD16− cells). (B–D) mRNA expression of PPAR isoforms (B), proinflammatory (C) and lipid metabolism (D) genes in classical monocytes. (E) Monocytes were isolated from whole blood and stimulated with lanifibranor. mRNA expression of PLIN-2 and CD36. Data are presented as geometric mean ± 95% CI. *p <0.05; **p <0.01; ***p <0.001; ****p <0.0001 (one-way ANOVA with post hoc test). NAFLD, non-alcoholic fatty liver disease.

PPARa activation, whereas PPARd controlled hepatic inflamma- model, which has the benefit of inducing severe inflammation tion and macrophage activation. Furthermore, lanifibranor and progression to fibrosis without the disadvantage of cachexia improved liver fibrosis through the combined results of de- as observed in the methionine/choline-deficient diet.26,38 creases in steatosis, liver damage and macrophage-mediated Although this model does not display some particular features inflammation, as well as a direct deactivation of stellate cells, of metabolic disease (obesity, insulin resistance), these results which is mainly driven by PPARc.4 are complemented by findings in the WD model, since lanifi- Importantly, the functions of human and mouse PPAR isoforms, branor decreased steatosis, liver injury and macrophage accu- in particular PPARa and PPARc, are species dependent. For instance, mulation in both models. the hepatic PPARa expression is lower in humans than mice, and In summary, we provide evidence for a strong therapeutic PPARa affects hepatic glycolysis-gluconeogenesis only in the response to the pan-PPAR agonist lanifibranor in a preclinical latter.35 These differences could explain why PPARa dispropor- model of steatohepatitis and fibrosis, and have identified in vivo tionally improves murine NAFLD compared to human NAFLD, while and in vitro effects on hepatic macrophage accumulation and its effects in clinical studies are minor.8 Conversely, whereas PPARc activation. While our work deepens the understanding of the agonists (pioglitazone) are potent in humans,19 we (and others) myriad roles that PPAR isoforms play in the coordination of discerned very limited to no effect on murine NAFLD. Nevertheless, metabolism and inflammation during NAFLD, the translation into in this study, the pan-PPAR agonist lanifibranor exceeded the a clinical benefit for patients with NASH requires further work. species-linked positive effects of PPARa/PPARc, which may be Lanifibranor is currently being evaluated in a phase IIb clinical trial attributed to a PPARd-dependent alteration of macrophage polari- in adults with NASH and high inflammatory activity (NATIVE trial, zation. In line with this observation, the PPARa/d agonist elafibranor NCT03008070), which will ultimately determine the therapeutic retained some of its effect in Ppara knockout mice.36 potential of pan-PPAR targeting on inflammation and fibrosis. Related to this, the current mouse models for NASH are un- fortunately not optimal, and the specific research question often Abbreviations dictates which model may be best suited to study certain aspects ALT, alanine aminotransferase; BMDM, bone marrow-derived of NASH.37 In this study, we first employed the CDAA-HFD NAFLD macrophages; CCl4, carbon tetrachloride; CDAA-HFD, choline-

12 Journal of Hepatology 2020 vol. - j 1–14 deficient, amino acid-defined high-fat diet; KCs, Kupffer cells; M- [6] Montagner A, Polizzi A, Fouche E, Ducheix S, Lippi Y, Lasserre F, et al. Liver CSF, monocyte-colony stimulating factor; MFI, mean fluores- PPARalpha is crucial for whole-body fatty acid homeostasis and is pro- tective against NAFLD. Gut 2016;65(7):1202–1214. cence intensity; Mme, ‘metabolically activated’ macrophage; [7] Ip E, Farrell G, Hall P, Robertson G, Leclercq I. Administration of the potent MoMFs, monocyte-derived macrophages; NAFLD, non-alcoholic PPARalpha agonist, Wy-14,643, reverses nutritional fibrosis and steato- fatty liver disease; NASH, non-alcoholic steatohepatitis; OA, hepatitis in mice. Hepatology 2004;39(5):1286–1296. oleic acid; PA, palmitic acid; PPAR, peroxisome proliferator- [8] Fuchs CD, Traussnigg SA, Trauner M. Nuclear receptor modulation for the activated receptor; WD, Western diet. treatment of nonalcoholic fatty liver disease. Semin Liver Dis 2016;36(1):69–86. [9] Maeda N, Takahashi M, Funahashi T, Kihara S, Nishizawa H, Kishida K, Financial support et al. PPARgamma ligands increase expression and plasma concentrations – SL and AG are supported by the Research Foundation – Flanders of adiponectin, an adipose-derived protein. Diabetes 2001;50(9):2094 2099. (SL: fellowship 11W4716N and grant for study abroad V426818N; [10] Moran-Salvador E, Titos E, Rius B, Gonzalez-Periz A, Garcia-Alonso V, AG: senior clinical investigator, 1805718N). CP and FT are sup- Lopez-Vicario C, et al. Cell-specific PPARgamma deficiency establishes ported by the German Research Foundation (DFG; CP: anti-inflammatory and anti-fibrogenic properties for this nuclear receptor 331065168; FT: Ta434/3-1, Ta434/5-1, CRC1382 and SFB/TRR57). in non-parenchymal liver cells. J Hepatol 2013;59(5):1045–1053. These funding agencies were not involved in study design, [11] Devisscher L, Verhelst X, Colle I, Van Vlierberghe H, Geerts A. The role of macrophages in obesity-driven chronic liver disease. J Leukoc Biol analysis or reporting. 2016;99(5):693–698. [12] Krenkel O, Tacke F. Liver macrophages in tissue homeostasis and disease. – Conflict of interest Nat Rev Immunol 2017;17(5):306 321. [13] Krenkel O, Hundertmark J, Abdallah AT, Kohlhepp M, Puengel T, Roth T, VA, ED, CE and GW are employees of Inventiva. Work in FT's et al. Myeloid cells in liver and bone marrow acquire a functionally laboratory has received funding by Allergan, Galapagos, Inven- distinct inflammatory phenotype during obesity-related steatohepatitis. tiva and Bristol Myers Squibb. Gut 2020;69(3):551–563. Please refer to the accompanying ICMJE disclosure forms for [14] Xue J, Schmidt SV, Sander J, Draffehn A, Krebs W, Quester I, et al. Tran- scriptome-based network analysis reveals a spectrum model of human further details. macrophage activation. Immunity 2014;40(2):274–288. [15] Kang K, Reilly SM, Karabacak V, Gangl MR, Fitzgerald K, Hatano B, et al. Authors' contributions Adipocyte-derived Th2 cytokines and myeloid PPARdelta regulate FT designed and supervised the research. SL and TP performed macrophage polarization and insulin sensitivity. Cell Metab 2008;7(6): 485–495. the animal experiments and analysed the data. CP, AF and FH [16] Odegaard JI, Ricardo-Gonzalez RR, Goforth MH, Morel CR, designed and conducted the biochip experiments. JH, CP, AG, VA, Subramanian V, Mukundan L, et al. Macrophage-specific PPARgamma ED and CE conducted experiments. SL, KDM and AG collected controls alternative activation and improves insulin resistance. Nature – and analysed human samples. GW and LD provided important 2007;447(7148):1116 1120. [17] Odegaard JI, Ricardo-Gonzalez RR, Red Eagle A, Vats D, Morel CR, technical support and intellectual input. SL, TP and FT wrote the Goforth MH, et al. Alternative M2 activation of Kupffer cells by PPARdelta manuscript. All authors reviewed and approved the manuscript. ameliorates obesity-induced insulin resistance. Cell Metab 2008;7(6): 496–507. [18] Sanyal AJ, Chalasani N, Kowdley KV, McCullough A, Diehl AM, Bass NM, Acknowledgments et al. Pioglitazone, vitamin E, or placebo for nonalcoholic steatohepatitis. The PPAR agonists were kindly provided by Inventiva. We N Engl J Med 2010;362(18):1675–1685. [19] Musso G, Cassader M, Paschetta E, Gambino R. Thiazolidinediones and cordially thank Petra Van Wassenhove, Inge Van Colen, Els Van advanced liver fibrosis in nonalcoholic steatohepatitis: a meta-analysis. Deynse and Carmen Gabrielle-Tag for the excellent technical JAMA Intern Med 2017;177(5):633–640. support, Alexander S. Mosig (University Hospital Jena) for [20] Ratziu V, Harrison SA, Francque S, Bedossa P, Lehert P, Serfaty L, et al. providing biochips, and all members of our labs and collabo- Elafibranor, an agonist of the peroxisome proliferator-activated receptor- rating scientists for helpful discussions. alpha and -delta, induces resolution of nonalcoholic steatohepatitis without fibrosis worsening. Gastroenterology 2016;150(5):1147–11459 e5. [21] Wettstein G, Luccarini JM, Poekes L, Faye P, Kupkowski F, Adarbes V, et al. Supplementary data The new-generation pan-peroxisome proliferator-activated receptor Supplementary data to this article can be found online at https:// agonist IVA337 protects the liver from metabolic disorders and fibrosis. Hepatol Commun 2017;1(6):524–537. doi.org/10.1016/j.jhep.2020.04.025. [22] Puengel T, Krenkel O, Kohlhepp M, Lefebvre E, Luedde T, Trautwein C, et al. Differential impact of the dual CCR2/CCR5 inhibitor cenicriviroc on References migration of monocyte and lymphocyte subsets in acute liver injury. PLoS Author names in bold designate shared first authorship one 2017;12(9):e0184694. [23] Spanheimer R, Betteridge DJ, Tan MH, Ferrannini E, Charbonnel B, Investigators PR. Long-term lipid effects of pioglitazone by baseline anti- [1] Younossi Z, Anstee QM, Marietti M, Hardy T, Henry L, Eslam M, et al. hyperglycemia medication therapy and statin use from the PROactive Global burden of NAFLD and NASH: trends, predictions, risk factors and experience (PROactive 14). Am J Cardiol 2009;104(2):234–239. prevention. Nat Rev Gastro Hepat 2018;15(1):11–20. [24] Lefere S, Degroote H, Van Vlierberghe H, Devisscher L. Unveiling the [2] Lefere S, Tacke F. Macrophages in obesity and non-alcoholic fatty liver depletion of Kupffer cells in experimental hepatocarcinogenesis through disease: crosstalk with metabolism. JHEP Rep 2019;1(1):30–43. liver macrophage subtype-specific markers. J Hepatol 2019;71(3):631– [3] EASL, EASD, EASO. EASL-EASD-EASO Clinical Practice Guidelines for the 633. management of non-alcoholic fatty liver disease. J Hepatol [25] Devisscher L, Scott CL, Lefere S, Raevens S, Bogaerts E, Paridaens A, et al. 2016;64(6):1388–1402. Non-alcoholic steatohepatitis induces transient changes within the liver [4] Gross B, Pawlak M, Lefebvre P, Staels B. PPARs in obesity-induced T2DM, macrophage pool. Cell Immunol 2017;322:74–83. dyslipidaemia and NAFLD. Nat Rev Endocrinol 2017;13(1):36–49. [26] Farrell G, Schattenberg JM, Leclercq I, Yeh MM, Goldin R, Teoh N, et al. [5] Francque S, Verrijken A, Caron S, Prawitt J, Paumelle R, Derudas B, et al. Mouse models of nonalcoholic steatohepatitis: toward optimization of PPARalpha gene expression correlates with severity and histological their relevance to human nonalcoholic steatohepatitis. Hepatology treatment response in patients with non-alcoholic steatohepatitis. 2019;69(5):2241–2257. J Hepatol 2015;63(1):164–173.

Journal of Hepatology 2020 vol. - j 1–14 13 Research Article NAFLD and Alcohol-Related Liver Diseases

[27] Dubois V, Eeckhoute J, Lefebvre P, Staels B. Distinct but complementary adipose tissue macrophages independently of classic activation. Cell contributions of PPAR isotypes to energy homeostasis. J Clin Invest Metab 2013;18(6):816–830. 2017;127(4):1202–1214. [34] Albillos A, Lario M, Alvarez-Mon M. Cirrhosis-associated immune [28] Rennert K, Steinborn S, Groger M, Ungerbock B, Jank AM, Ehgartner J, dysfunction: distinctive features and clinical relevance. J Hepatol et al. A microfluidically perfused three dimensional human liver model. 2014;61(6):1385–1396. Biomaterials 2015;71:119–131. [35] Pawlak M, Lefebvre P, Staels B. Molecular mechanism of PPARalpha action [29] Krenkel O, Puengel T, Govaere O, Abdallah AT, Mossanen JC, Kohlhepp M, and its impact on lipid metabolism, inflammation and fibrosis in non- et al. Therapeutic inhibition of inflammatory monocyte recruitment re- alcoholic fatty liver disease. J Hepatol 2015;62(3):720–733. duces steatohepatitis and liver fibrosis. Hepatology 2018;67(4):1270–1283. [36] Staels B, Rubenstrunk A, Noel B, Rigou G, Delataille P, Millatt LJ, et al. [30] Kratz M, Coats BR, Hisert KB, Hagman D, Mutskov V, Peris E, et al. Metabolic Hepatoprotective effects of the dual peroxisome proliferator-activated dysfunction drives a mechanistically distinct proinflammatory phenotype receptor alpha/delta agonist, GFT505, in rodent models of nonalcoholic in adipose tissue macrophages. Cell Metab 2014;20(4):614–625. fatty liver disease/nonalcoholic steatohepatitis. Hepatology 2013;58(6): [31] Coats BR, Schoenfelt KQ, Barbosa-Lorenzi VC, Peris E, Cui C, Hoffman A, 1941–1952. et al. Metabolically activated adipose tissue macrophages perform detri- [37] Rinella ME, Tacke F, Sanyal AJ, Anstee QM, participants of the AEW. mental and beneficial functions during diet-induced obesity. Cell Rep Report on the AASLD/EASL joint workshop on clinical trial endpoints in 2017;20(13):3149–3161. NAFLD. J Hepatol 2019;71(4):823–833. [32] Reilly SM, Saltiel AR. Adapting to obesity with adipose tissue inflamma- [38] Lefere S, Van de Velde F, Hoorens A, Raevens S, Van Campenhout S, tion. Nat Rev Endocrinol 2017;13(11):633–643. Vandierendonck A, et al. Angiopoietin-2 promotes pathological angio- [33] Xu X, Grijalva A, Skowronski A, van Eijk M, Serlie MJ, Ferrante Jr AW. genesis and is a therapeutic target in murine nonalcoholic fatty liver Obesity activates a program of lysosomal-dependent lipid metabolism in disease. Hepatology 2019;69(3):1087–1104.

14 Journal of Hepatology 2020 vol. - j 1–14