Hepatocyte-independent PAR1-biased signaling controls liver pathology in experimental obesity

Abstract

Background and objective:

Protease-activated receptor-1 (PAR1) has emerged as an important link between coagulation and the complications of obesity including metabolic dysfunction-associated steatotic liver disease (MASLD). PAR1 is expressed by various cells and cleaved by different proteases to generate unique tethered agonists that activate distinct signaling pathways. Mice expressing PAR1 with an R41Q mutation have disabled canonical thrombin-mediated signaling, whereas R46Q mice express PAR1 resistant to non-canonical signaling by activated protein C (APC).

Methods:

Mice with whole body and hepatocyte-selective PAR1 deficiency, as well as PAR1 R41Q and R46Q mice were fed a high fat diet to induce MASLD.

Results and conclusions:

High fat diet (HFD)-fed R41Q mice displayed reduced hepatic steatosis and liver/body weight ratio. In contrast, HFD-fed R46Q mice displayed increased relative liver weight and hepatic steatosis alongside increased serum ALT activity. Surprisingly, despite the distinct impact of PAR1 mutations on steatosis, selective deletion of PAR1 in hepatocytes had no impact. To evaluate a viable PAR1-targeted approach, mice with HFD-induced obesity were treated with the allosteric PAR1 modulator NRD-21, which inhibits canonical PAR1 inflammatory signaling but promotes PAR1 protective, non-canonical anti-inflammatory signaling. NRD-21 treatment reduced plasma TNFα, serum ALT activity, hepatic steatosis, and insulin resistance (HOMA-IR), but increased plasma active GLP-1. The results suggest non-hepatocellular canonical PAR1 cleavage drives MASLD in obese mice and provide translational proof-of-concept that selective pharmacological modulation of PAR1 yields multiple metabolic benefits in experimental obesity.

Introduction

Metabolic dysfunction-associated steatotic liver disease (MASLD, formerly NAFLD), is defined by the presence of hepatic steatosis and at least one of several cardiometabolic risk factors ¹. MASLD and its progression to metabolic dysfunction-associated steatohepatitis (MASH) and cirrhosis represent a major clinical burden, particularly given the prevalence of obesity ². MASLD and MASH patients display maintained capacity for ex vivo thrombin generation and increased plasma biomarkers of coagulation activity. Likewise, coagulation activation is evident in experimental settings of MASLD/MASH ^3–6. Strong experimental evidence from both genetic and pharmacologic approaches indicates that tissue factor-dependent thrombin generation drives multiple features of MASLD/MASH ^{3, 7}. Thrombin-catalyzed fibrin clot formation and activation of protease activated receptor (PAR) signaling have been implicated as drivers of experimental MASLD pathogenesis ^6–9. Complete PAR1 deficiency or administration of the PAR1 antagonist vorapaxar reduced hepatic steatosis and limited advanced pathologies including inflammation, hepatic injury, and fibrosis in various experimental settings of MASLD ^{7, 8}. PAR1 likely couples coagulation to MASLD pathogenesis, but beyond germline PAR1 deficiency and global PAR1 inhibition, detailed mechanistic connections are lacking. Toward the goal of defining the precise mechanisms linking PAR1 to MASLD pathogenesis and advancing PAR1 as a druggable target in MASLD, we sought i) to define the PAR1 expressing cell type driving MASLD pathogenesis, ii) to determine the role of PAR1 biased agonism in obesity-associated MASLD, and iii) to provide the proof-of-concept that PAR1 modulation can be used to treat established MASLD.

Methods

Male mice on a C57Bl/6J background were fed HFD (60% calories from fat, D12492, Research Diets) or low fat (10%) sucrose-matched control diet (D12450J) beginning approximately 8 weeks of age for 12–14 weeks. R41Q and R46Q mice ¹⁰, wild-type mice (Jackson Laboratory), PAR-1^−/− mice ¹¹, and PAR1^flox/flox:AlbuminCre⁺ mice, generated by crossing PAR1^flox/flox mice ¹² with Albumin-Cre mice (Jackson Laboratory, Strain #003574) were used for these experiments. As a complementary approach to delete PAR1 in hepatocytes, PAR1^flox/flox mice were treated with an established adeno-associated viral vector that expresses Cre recombinase specifically in hepatocytes AAV8-TBG-Cre (2.5×10¹¹ GC/mouse; University of Pennsylvania Vector Core [now Addgene, Plasmid #107787) ¹³ or control (i.e., GFP expressing) vector. For parmodulin treatment studies, 6-week-old mice were fed HFD for 12 weeks (C57Bl/6J DIO, Jackson Laboratory, #380050), then treated with NRD21 (10 mg/kg) or its vehicle (20% DMSO, 20% PEG400, 60% PBS, 5 ml/kg, ip) every other day for 4 weeks. Hepatic steatosis in H&E-stained left lateral lobe liver sections was estimated using QuPath ¹⁴ by thresholding the eosin channel such that the area of tissue and lipid droplets (macro and microvesicular steatosis) could be distinguished. Serum ALT activity was measured using commercial reagents (Pointe Scientific). Hepatic levels of PAR1 mRNA were determined using SYBR green qRT-PCR as described previously ⁷. Blood glucose was measured using a commercial glucometer and related metabolic hormones and inflammatory mediators measured in citrated plasma collected with added protease inhibitors as per vendor specifications using a U-PLEX Metabolic Hormone panel (K-15306K-1; Meso-Scale Diagnostics) and Sector Imager. Statistical analyses were performed in GraphPad Prism (10.1.2) using one- or two-way analysis-of-variance or Student’s t-test were used, as appropriate. The criterion for statistical significance was P<0.05.

Results and Discussion:

Hepatic PAR1 mRNA expression was increased in HFD-fed wild-type mice (Fig. 1A). Prior studies found that PAR1 deficiency did not affect HFD-induced body weight gain ⁶. Using banked samples from these experiments, we found that hepatic steatosis was significantly reduced in PAR1^−/− mice fed a HFD (Fig. 1B–C). The precise PAR1 expressing cell type that drives steatosis in HFD-fed mice is not known. It has been posited that PAR1 expressed by hepatocytes directly promotes hepatic steatosis, although the role of hepatocyte PAR1 was inferred from in vitro studies and systemic PAR1 inhibition ⁸. To directly test the role of hepatocyte PAR1 in HFD-induced MASLD, we generated mice with hepatocyte-selective PAR1 deficiency (i.e, PAR1^flox/flox:AlbCre mice). PAR1 mRNA was dramatically reduced in isolated primary hepatocytes from PAR1^flox/flox:AlbCre⁺ mice (Fig. 1D) and this was reflected by reduced whole liver PAR1 mRNA (Fig. 1E). Hepatocyte PAR1 deficiency had no impact on HFD-induced body weight gain, relative liver or adipose weight, or serum ALT activity (Fig. 1F–I), and unlike global PAR1 deficiency, did not reduce hepatic steatosis (Fig. 1J–K). PAR1 mRNA levels were dramatically reduced in hepatocytes isolated from mice given AAV8-TBG-Cre, but this did not impact LW/BW, serum ALT activity or hepatic steatosis in mice fed HFD (Fig. 1M–O). Contrasting a recent suggestion ⁸, our results suggest hepatocyte PAR1 does not drive HFD-induced steatosis. Thus, PAR1 expressed by hepatic nonparenchymal cells, or even extrahepatic PAR1, appears to drive the pathogenesis of HFD-induced MASLD. In considering the cellular source of PAR-1, it is notable that unlike human platelets, mouse platelets do not express PAR-1 further restricting the causative cellular source on PAR-1 in this model. Indeed, another recent publication suggested a potential role of PAR1 expressed by brown adipose tissue (BAT) in obesity associated pathologies, although the impact of BAT PAR1 deficiency on liver pathology was not addressed ¹⁵.

Figure 1. Impact of hepatocyte PAR1 deficiency on hepatic steatosis in HFD-fed mice.

Wild-type, PAR1−/−, PAR1^flox/flox:AlbCre⁻, and PAR1^flox/flox:AlbCre⁺ mice were fed high fat diet (HFD) or control diet (CD) for 12 weeks (N=6–8 mice per group). (A) Hepatic PAR1 mRNA (F2r) levels were determined in HFD-fed wild-type mice using qRT-PCR and adjusted for three distinct housekeeper genes. (N=2 mice per group) (B) Representative photomicrographs of H&E-stained liver sections and (C) associated quantification of hepatic steatosis. PAR1 mRNA (F2r) levels were determined in (D) primary mouse hepatocytes isolated by collagenase digestion and (E) whole liver RNA isolates. (F-H) Body weight and liver weight (LW) and epididymal adipose (eWAT) to BW ratios were determined after HFD or CD feeding for 12 weeks. (I) Serum ALT activity was determined as biomarker of hepatocellular injury (normal ALT ~30 U/L). (J) Representative photomicrographs of H&E-stained liver sections and (K) associated quantification of hepatic steatosis. (L) PAR1 mRNA (F2r) levels were determined in primary mouse hepatocytes isolated from PAR1^flox/flox mice given AAV8 vectors driving either (AAV8-TBG-Cre or luciferase (Luc, control) expression under control of the thyroxine binding globulin (TBG) promoter (N=3–5 mice per group). (M-O) LW/BW, serum ALT activity and hepatic steatosis were measured after HFD for 12 weeks (N=8 mice per group). The results are plotted as individual mice (or cells isolated from individual mice [D, L]) and as mean ± SEM. Closed shapes are control diet-fed mice whereas open shapes are HFD-fed mice. *P<0.05, ****P<0.0001.

Although best known as a transducer of thrombin signaling, the N-terminal domain of PAR1 contains multiple distinct tethered ligands, each revealed by proteolytic cleavage at unique residues by different proteases. Specifically, biased PAR1 signaling can be evoked by cleavage of PAR1 at canonical (Arg41) and non-canonical (Arg46) N-terminal residues, by thrombin and activated protein C (APC), respectively, and potentially numerous other proteases ^{16, 17}. Whereas cleavage at Arg41 is proinflammatory, cleavage at Arg46 is cytoprotective and anti-inflammatory ^{16, 18}. In a recent study from Maeso-Diaz et al., a “thrombomodulin-PAR1 signaling axis” was proposed to promote steatosis by inducing hepatocellular senescence ⁸. However, impaired thrombomodulin-mediated generation of APC worsened steatosis in HFD-fed mice ⁶. Rather, pharmacologic inhibition of thrombin or genetically reducing prothrombin levels each significantly reduced HFD-induced steatosis ^{6, 19}. Despite these prior studies, the precise role of each PAR1 cleavage event in HFD-induced steatosis is not known.

To define the precise role of PAR1 biased signaling in HFD-induced MASLD, we used novel mice expressing mutant forms of PAR1 in which either canonical or noncanonical PAR1 cleavage has been selectively disabled ¹⁰. Canonical thrombin-mediated cleavage of PAR1 is disabled in R41Q mice, whereas non-canonical APC-mediated PAR1 cleavage is preserved. In contrast, canonical PAR1 cleavage is not ablated in R46Q mice, whereas APC-mediated noncanonical cleavage is disabled. HFD-induced body weight gain was equivalent in wild-type mice and R41Q mice, whereas body weight gain in HFD-fed R46Q mice tended to increase (P=0.06) (Fig. 2A). Liver weight/body weight ratio (LW/BW) was significantly reduced in HFD-fed R41Q mice compared with HFD-fed wild-type mice (Fig. 2B–C). This reduction in liver mass occurred alongside an increase in total adipose volume (subcutaneous + visceral), explaining in part the similar body weight gain in wild-type and R41Q mice. Notably, LW/BW increased dramatically in HFD-fed R46Q mice relative to wild-type mice (Fig. 2B–C). This as well as increased adipose tissue volume (Fig. 2D) likely account for the near significant (P=0.06) exacerbation of HFD-induced body weight gain in HFD-fed R46Q mice (Fig. 2B–D). The reduction in LW/BW in R41Q mice was coupled to a reduction in hepatic steatosis (Fig. 2E–F), analogous to PAR1^−/− mice. Periportal macrovesicular steatosis appeared to increase in HFD-fed R46Q mice, although this increase was not statistically significant (Fig. 2E–F). Increased ALT activity was also observed in HFD-fed R46Q mice (Fig. 2G). The results indicate that eliminating PAR1 canonical signaling reduces hepatic steatosis, potentially by promoting storage of lipids in peripheral adipose tissue. Moreover, disabling noncanonical PAR1 signaling appeared to worsen HFD-induced liver pathology, in direct opposition to the mechanism proposed by Maeso-Diaz and colleagues.

Figure 2: Role of PAR-1 biased agonism in high fat diet-induced steatosis.

Terminal body weight (A), liver weight (B) and liver weight/body weight ratio (C) were determined in wild-type, R41Q and R46Q mice after 14 weeks of feeding high fat diet (HFD) or control diet (CD). N=7–9 mice per group for control diet and 16–23 mice per group for HFD.(D) Representative μCT-derived images and quantification of total % adipose (visceral and subcutaneous) (n=4–8 mice per group). (E) Representative photomicrographs of H&E-stained liver sections and (F) associated quantification of hepatic steatosis. N=6–9 mice per group. (G) Serum ALT activity was determined as biomarker of hepatocellular injury. The results are plotted as individual mice and as mean ± SEM. Closed shapes are control diet-fed mice whereas open shapes are HFD-fed mice. *P<0.05, **P<0.01, ****P<0.0001.

The PAR1 antagonist vorapaxar reduced hepatic steatosis in obese db/db mice and in MASLD induced by combined western diet feeding and carbon tetrachloride challenge ⁸. Complete PAR1 inhibition may increase bleeding risk²⁰ and would also block beneficial noncanonical PAR1 signaling²¹. Inspired by our discovery that PAR1 biased signaling impacts the severity of HFD-induced MASLD in mice, we sought a pharmacologic approach that could dampen pathologic PAR1 effects while preserving beneficial PAR1 noncanonical activation. Small molecule PAR1 modulators with this potential have been discovered and are termed parmodulins. ²². The investigational compound NRD-21 inhibits canonical PAR1 inflammatory signaling and is thought to promote PAR1 protective, APC-like non-canonical anti-inflammatory signaling ²³. Indeed, the structurally related parmodulin ML161 (aka PM2) inhibited platelet activation, but did not increase clotting time (i.e., aPPT) ²⁴ or increase bleeding time in a mouse tail clip assay.²¹ Thus, parmodulins as a class of PAR-1 directed compounds may have decreased bleeding risks relative to orthosteric PAR1 antagonists, but this will need to be confirmed for NRD21. Prior studies have also documented improved endothelial barrier integrity and reduced cytotoxicity with ML161 relative to vorapaxar,²¹ consistent with the promotion of anti-inflammatory, non-canonical PAR1 signaling by parmodulins. We posited that NRD-21 treatment would reduce steatosis in obese mice. To test this hypothesis and explore the potential for small molecule PAR1 inhibition to impact obesity-related disease sequelae, mice with established HFD-induced obesity were purchased from the Jackson Laboratory, continuously fed HFD, and treated with NRD-21 (10 mg/kg) every other day for 4 weeks (Fig. 3A). NRD-21 treatment did not affect body weight but significantly reduced LW/BW, hepatic steatosis, and serum ALT activity (Fig. 3B–G). Fasting insulin and glucose levels were reduced by NRD-21 treatment (Fig. 3H–I), as were (Homeostatic Model Assessment for Insulin Resistance; HOMA-IR) values and plasma leptin (Fig. 3J–K). Plasma MCP-1 levels were unaffected, but plasma TNFα concentration was reduced by NRD-21 treatment (Fig. 3L–M). Particularly interesting given emerging benefits ascribed to glucagon-like peptide 1 mimetics in the setting of MASLD ²⁵, we were intrigued to discover that plasma active GLP-1 levels were significantly increased in NRD-21-treated mice, despite equivalent plasma glucagon concentration (Fig. 3N–O). These proof-of-concept studies document a therapeutic reduction in hepatic steatosis along with evidence of systemic metabolic benefit in mice treated with the parmodulin NRD-21.

Figure 3: Impact of parmodulin NRD-21 treatment on high fat diet-induced steatosis and metabolic hormones.

Wild-type mice with established HFD-induced obesity (18-week-old DIO mice, Jackson Laboratory) were treated (ip) every other day with NRD-21 (10 mg/kg) or vehicle (20% DMSO, 20% PEG400, 60% PBS) for 4 weeks (N=7–8 mice per group) (A) Body weight (B) and relative liver and adipose weights (C-D) were determined. (E) Representative photomicrographs of H&E-stained liver sections and (F) associated quantification of hepatic steatosis. Serum ALT activity (G) was determined as a biomarker of hepatocellular injury. Fasting (6 hr) glucose and insulin levels (H-I) were determined as described in methods and HOMA-IR (J) calculated. Additional metabolic hormones and cytokines were determined as described in methods (K-O). The results are plotted as individual mice and as mean ± SEM. Circles are vehicle-treated mice and squares are NRD21-treated mice. *P<0.05, **P<0.01.

In summary, the results herein identify a pathologic role for canonical non-hepatocellular PAR1 signaling in the development of HFD-induced steatosis. Moreover, the results suggest a beneficial effect of PAR1 noncanonical signaling in HFD-fed mice. Genetic and pharmacologic (i.e., parmodulin) approaches support these conclusions. The precise mechanisms linking each PAR1 cleavage site to MASLD, and how each PAR1 signaling pathway may counterbalance liver pathology in obesity are important topics for future studies. This report significantly advances our understanding of the links between PAR1 and MASLD and provides therapeutic proof-of-concept that biased agonism of PAR1 can be pharmacologically manipulated to treat obesity-associated sequelae including liver disease.