The purinergic receptor P2X4R impacts lipid droplet size in fatty liver by stimulating macroautophagy

Abstract

Background & Aims

P2X4R is a lysosomal ATP-gated cation channel receptor, playing roles in liver pathophysiology through poorly understood mechanisms. We aim at defining the role of P2X4R in macroautophagy and lipid droplet (LD) homeostasis.

Methods

We used P2X4R-KO, LC3-GFP transgenic and WT mice, a specific P2X4R antagonist (BAY1797), and three experimental models of steatosis (n = 5–10 mice per group). Autophagic flux was measured in human embryonic kidney reporter cell line, and Atg5-/- mouse embryonic fibroblasts. Oleic acid cell loading, Oil Red O and LipidTox cell and tissue staining, and LC3-II Western blotting were performed.

Results

The autophagic flux in the liver and in primary isolated hepatocytes was decreased in P2X4R-KO mice compared with WT mice, as well as in BAY1797 (compared with vehicle)-treated WT animals and hepatocytes. P2X4R overexpression stimulated the autophagic flux, indicating an impact of P2X4R on lysosome-autophagosome fusion. BAY1797 treatment in fasted mice as well as in both methionine- and choline-deficient diet and high-fat, high-sucrose, cholesterol-containing diet models resulted in small LD instead of large LD steatosis (vehicle-treated mice) and stimulated cytosolic lipolysis. Similarly, BAY1797 inhibited lipophagy and stimulated cytosolic lipolysis in vitro in oleic acid-loaded primary hepatocytes and in mouse embryonic fibroblasts, resulting in smaller LDs. Lysosome position analysis upon starvation revealed that P2X4R inhibition interfered with lysosomal trafficking in vitro. In patients with metabolic dysfunction-associated steatotic liver disease (n = 24), we found a correlation between LD size, autophagic flux, and P2X4R expression.

Conclusions

P2X4R inhibition decreased the autophagic flux in hepatocytes, resulting in LD size reduction, during fasting- as well as diet-induced steatosis in mice. P2X4R is proposed as a new therapeutic target to modulate lipophagy in the liver.

Impact and implications

We provide in vitro and in vivo evidence that P2X4R is a regulator of macroautophagy, and thereby affects LD homeostasis in the liver. Both genetic deletion and specific pharmacological inhibition of P2X4R results in LD size reduction in three murine models of fatty liver, in association with inhibition of lipophagy, stimulation of lipolysis, and reduction of liver inflammation. Finally, in a series of human patients with MASLD, we found that P2X4R expression correlated with autophagic flux and LD size. Thus, based on our data, we report that P2X4R inhibition is hepatoprotective in the context of MASLD.

Graphical abstract

In hepatocytes, large LDs are handled by cytosolic lipolysis resulting in small LDs which are degraded by lipophagy. P2X4R inhibition results in reduced lipophagy and increased lipolysis, leading to accumulation of small LDs, associated with hepatoprotection.

Highlights

• •We provide in vitro and in vivo evidence that P2X4R is a regulator of macroautophagy, and thereby affects lipid droplet homeostasis in the liver.
• •P2X4R inhibition reduces lipid droplet size and liver inflammation in murine fatty liver models.
• •In a series of patients with MASLD, P2X4R expression was correlated with autophagic flux and lipid droplet size.
• •P2X4R inhibition is hepatoprotective in the context of MASLD.

Introduction

Metabolic dysfunction-associated steatotic liver disease (MASLD), one of the most common etiologies of liver diseases in western countries, encompasses a broad clinical spectrum ranging from simple steatosis to metabolic dysfunction-associated steatohepatitis (MASH), advanced fibrosis, cirrhosis, and finally hepatocellular carcinoma. As the mechanisms of fibrogenesis and steatogenesis are not fully understood, there is currently only scarce and partially efficient specific treatment for these patients.^1,2 During liver injury and repair, the hepatic microenvironment is exposed to mechanically induced and inflammation-induced factors. Among these, purines, such as extracellular ATP and its breakdown products, are released by different types of liver cells. However, their role in liver pathophysiology remains largely unexplored. Purines and their purinoceptors (P1 and P2) constitute a powerful signaling network that has been relatively less explored in liver pathophysiology than in the nervous, immune, renal, and cardiovascular systems.³ We previously demonstrated that, in addition to P2Y receptors, P2X4R and P2X7R were highly expressed and functional in rat hepatocytes, representing the predominant P2X receptors in this cell type.⁴ Furthermore, we provided evidence as following: (1) ATP was released by the liver after partial hepatectomy, in blood and bile, via mechanisms depending on portal hyperpressure; (2) the released ATP contributed to push quiescent hepatocytes into the cell cycle; and (3) ATP was also released from human liver after partial hepatectomy in the context of living donor liver transplantation, suggesting that purinergic signaling may also have an impact on human liver repair response.⁵ We reported that P2X4R was highly concentrated in lysosomes and had a positive impact on liver regeneration in mice, through the regulation of biliary homeostasis, and that these observed effects were associated with lysosomal exocytosis in bile.⁶ We also found that P2X4R was the most abundant purinergic receptor in liver myofibroblasts, where it strongly mediates profibrogenic functions.⁷ Furthermore, the profibrogenic effect of P2X4R was at least in part related with lysosomal exocytosis. Importantly, in the absence of P2X4R, lysosomes were smaller and their distribution in hepatocytes was altered, switching from their typical pericanalicular position toward more diffuse cytosolic location.⁶

P2X4R belongs to the P2X receptor family of ATP-gated trimeric cation channels displaying a very high sensitivity for extracellular ATP, its most potent agonist (EC50, 1–10 μM), to induce both Na⁺ and Ca²⁺ entry while allowing a K⁺ efflux.[8], [9], [10] P2X4R is mainly implicated in diverse physiological functions: (1) chronic inflammatory pain processing through as it is expressed in tissue-resident macrophages including microglia; (2) vessel tone and remodeling, as well as in ischemia-reperfusion through its expression in vascular endothelial cells; and (3) immune cell responses through its expression in macrophages and lymphocyte subsets.⁸ Interestingly, P2X4R also plays a critical role in epithelial cells, where it participates in exocytosis¹¹ and ion secretion.¹² Importantly, although other P2XRs are predominantly expressed at the plasma membrane, P2X4R is mainly localized intracellularly along the endo-lysosomal compartment,¹³ as described in numerous cell types and cell lines (e.g. macrophages, astrocytes, and hepatocytes). From there, it can be targeted at the plasma membrane to fulfill exocytotic/secretory functions.^11,12 P2X4R is important for endo-lysosomal fusion with other membrane compartments such as endosomes or plasma membrane.^14,15 However, there is only scarce and contradictory literature suggesting that P2X4R would contribute to the autophagy process in non-hepatic cells,^16,17 during which autophagosome fusion with lysosomes is a fundamental step to terminate degradation of damaged organelles or other cellular components. Thus, the impact of P2X4R on liver pathophysiology through interference with autophagy remains unexplored.

Although conflicting reports have been published suggesting that mutual regulation between lipids and autophagy is more complex than expected, it is well established that autophagy is involved in lipid droplet (LD) catabolism^18,19 through different pathways, including macrolipophagy²⁰ and chaperone-mediated autophagy.²¹ Moreover, the role of autophagy in fatty liver diseases remains poorly defined. On the one hand, non-alcoholic fatty liver disease (NAFLD) can inhibit autophagy by interfering with lysosomal proteolytic activity, lysosome-autophagosome fusion, and/or autophagy gene expression. In line, stimulation of autophagy resulted in NASH resolution in mice.²² On the other hand, autophagy can affect the pathophysiological course of NAFLD at the level of macrophages (anti-inflammatory impact), hepatic stellate cells (HSCs), or hepatocytes, and rare mutations in the Atg7 gene reportedly predispose patients to severe steatosis.²³

In the present study, we provide data showing that P2X4R significantly contributes to the process of autophagy, and more specifically lipophagy, by facilitating the fusion between lysosomes and autophagosomes. Although exact mechanisms still remain to be deciphered, our data show that P2X4R interferes with lysosome trafficking. Surprisingly, P2X4R antagonism resulted in lipophagy inhibition, lipolysis activation, and LD size reduction during several experimental in vitro and in vivo settings of fatty liver, with global beneficial effects on liver inflammation and lipotoxicity. This study suggests that translational perspectives based on P2X4R targeting in MASH could be envisioned.

Materials and methods

Experimental procedures used on animals

All animals were treated in accordance with the 3R principle and the European Directive 2010/63/EU, under the supervision of our ethics committee CEEA-059. P2X4R-KO mice and C57BL/6J WT mice were initially obtained from the laboratory of François. Rassendren and Lauren Ullman (Montpellier, France). LC3-GFP mice were generously provided by the team of Guido Kroemer and Chiara Maiuri (Cordeliers Research Center, Paris, France). Male mice aged 16–20 weeks were used for the different series of experiments. For the study of fasting-induced steatosis, mice received (gavage) for 3 days in the morning and evening a solution of polyethylene glycol (PEG 400, 50%) with or without (vehicle group) 20 mg/ml of BAY1797 (Bayer Laboratories, Berlin, Germany). Mice were then either fasted or fed ad libitum during the last 48 h.

Diet-induced steatosis models

For high-fat diet-induced steatosis studies, mice were fed a high-fat, high-sucrose, cholesterol-containing (HFHSC) diet (S9400-E708; ssniff Spezialdiäten GmbH, Soest, Germany) for 2 months. Other series of mice were fed a methionine- and choline-deficient (MCD) diet for 2 weeks (E15653-93, E15654-04; ssniff Spezialdiäten GmbH). In these two models, mice were given two gavages per day (morning and evening) 3 days before sacrifice with a solution of PEG 400 (50%) with or without (vehicle group) 20 mg/ml of BAY1797. The administered volume varied according to the body mass of each mouse to achieve a dose of 50 mg/kg. Mice received an i.p. injection of leupeptin (Leu, 40 mg/kg, SP-04-2217; Euromedex, Souffelweyersheim, France) 4 h before sacrifice, for autophagic flux calculation as previously reported.24 Laparotomies were performed under 4.5% isoflurane anesthesia. Samples were either snap-frozen in liquid nitrogen-cooled isopentane or stored in RNA later (QIAGEN, Hilden, Germany) at -80 °C until use. Portions of liver were also placed in cassettes immersed in 4% formaldehyde before being paraffin embedded. All surgical procedures were performed on fed mice in the morning (08:00–11:00).

Hepatocyte isolation, Western blot analysis, cell and LD size measurements, as well as fatty acid loading and lysosome positioning experiments are detailed in Supplementary Materials and Methods.

Immunohistochemistry

Antibodies against Ki67 (# ab15580; Abcam), E-cadherin (# 13-1900; Thermo Fisher Scientific), DDP4 (# ab187048; Abcam), LAMP-1 (# 1D4B; Developmental Studies Hybridoma Bank, Iowa City, IA, USA), as well as phalloidin (P1951; Sigma-Aldrich) were used on ethanol/acetone-fixed 10 μm liver cryosections. Images were acquired using epifluorescence microscopy (Microscopes Nikon TE300 and ZEISS Axioskop and confocal EZ-C2; Nikon), and analyzed with the ImageJ software. H&E and Oil Red O staining on liver sections were performed as described.

Biochemical assays

Alanine aminotransferase (ALT), triglyceride (TG), and free fatty acid (FFA) measurements in the plasma, as well as glycogen and TG quantification in the liver, were performed as detailed in Supplementary Materials and Methods.

Quantitative RT-PCR and Western blotting experiments are further described in Supplementary Materials and Methods and Table S1.

Statistical analysis

The Student's t test was used to compare sample means with controls. Results are expressed as mean ± SEM. The Spearman correlation was used to measure the degree of association between two variables in human studies. All statistical analysis were performed in GraphPad Prism 7.0 software (GraphPad Software, San Diego, CA, USA). The p values ≤0.05 (∗), ≤0.01 (∗∗), and ≤0.001 (∗∗∗) were considered statistically significant.

Results

P2X4R antagonism or gene deletion inhibits macroautophagy

Compared with WT mice, total P2X4R-KO mice had a significantly lower autophagic flux in the liver, especially upon fasting. This was well highlighted by Western blot LC3 analysis of liver protein extracts from fed or fasted mice injected or not with leupeptin before harvesting, as explained in the Materials and methods section (Fig. 1A). In the same line, WT mice (but not P2X4R-KO mice; Fig. S1A) treated with the specific P2X4R antagonist BAY1797²⁵ exhibited a slower autophagic flux in the liver in fasted conditions compared with vehicle-treated mice (Fig. 1B). Importantly, these P2X4R-related effects were not due to an impact on autophagosome production, as shown by the lack of Atg13 protein expression decrease upon BAY1797 treatment (Fig. S1B). Finally, BAY1797 treatment significantly inhibited the autophagic flux in primary hepatocytes isolated from WT mice but not from P2X4R-KO mice (Fig. 1C). Further strengthening these data, we found that the accumulation of autophagosomes in the liver from fasted LC3-GFP transgenic mice²⁶ was exacerbated upon BAY1797 treatment (Fig. 1D). These effects were mostly observed in leupeptin-injected mice (Materials and methods).

Fig. 1.

P2X4R antagonism or gene deletion inhibits macroautophagy.

Representative images of LC3 Western blots and semi-quantitative analysis. (A) WT and P2X4R-KO mice livers in fed and starved conditions, 4 h after i.p. injection with vehicle (Veh) or leupeptin (Leu) for autophagic flux analysis (Materials and methods section); n = 5 mice per group. (B) WT mice fed or fasted, treated with or without vehicle (Veh) or BAY1797 (see experimental protocol in the Materials and methods section and in Fig. 4A), 4 h after i.p. injection with Veh or Leu; n = 5 mice per group. (C) Autophagic flux analysis based on LC3 Western blot on extracted proteins from WT and P2X4R-KO primary hepatocytes, cultured in complete medium with or without chloroquine; n = 5 mice per group. (D) BAY1797-induced autophagosome accumulation visualized on liver sections from fasted LC3-GFP transgenic mice injected with Leu. Representative confocal microscope images (each image is the mean of 16 acquisitions) and quantitative analysis from n = 5 mice per group are shown. E-cadherin immunostaining (red) identifies periportal (PP) area of the hepatic lobule. LC3-GFP (green) dots (autophagosomes) accumulate in the centrolobular (CL) (E-cadherin negative) zone. Scale bar: 50 μm. For A–D, ∗p ≤0.05; ∗∗p ≤0.01; ∗∗∗p ≤0.001 (Student’s t test). Data are mean ± SEM. CLV, centrolobular vein; PV, portal vein.

Interestingly, the previously reported fasting-induced reduction in the liver/body weight ratio, which is associated with hepatocyte size reduction,²⁷ was completely abolished upon P2X4R blockade in WT mice (Fig. 2A,B) but not in P2X4R-KO mice (Fig. S1C). As a control, hepatocyte proliferation, recently reported to be enhanced upon fasting,²⁸ remained unchanged and negligible in the different groups (Fig. S1D). These data suggest that when inhibiting P2X4R, autophagic flux is slowed down and the autophagy-related reduction in cell size is dampened, as described in other settings of deficient autophagy.²⁹ Notably, this impact on cell size reduction was mostly observed in the pericentral (PC) area of the lobule (Fig. 2C–F), a region where autophagy might be differentially regulated compared with the periportal (PP) zone.³⁰

Fig. 2.

P2X4R contributes to hepatocyte cell size regulation during starvation in mice.

WT fed and fasted mice treated or not (vehicle) with BAY1797 (BAY) (n = 15 mice per group) were analyzed for (A) liver/body weight ratio, (B) body weight loss (%), (C–E) hepatocyte surface, and (F) lobular area, measured on phalloidin-stained liver sections. Representative images are shown in C. Scale bar: 100 μm. For A–C: ∗p ≤0.05; ∗∗p ≤0.01; ∗∗∗p ≤0.001 (Student’s t test). Data are mean ± SEM.

Based on the above-described data, we investigated the fate of P2X4R during fasting-induced autophagy. As per our in vivo studies using liver homogenates and lysosomal fractions, we found that lysosomal fractions from fed mice were highly enriched in P2X4R and exhibited two specific bands (70 and 55 kDa) compared with liver homogenates that exhibited a single band (70 kDa) (Fig. S2A). During fasting (12 and 36 h), although P2X4R remained detectable in the lysosomal fraction, the 70 kDa band progressively faded, accompanied by an increase in the 55 kDa band and appearance of an additional lighter band (50 kDa); conversely, P2X4R expression in total homogenates remained the same. This shift in molecular weight did not result from deglycosylation and did not occur with boiled samples (Fig. S2B), whereas we provided evidence that lysosomal P2X4R protein changing during fasting could be mimicked by decreasing the incubation pH from 8 to 4 (Fig. S2C). Finally, protease inhibition experiments in lysosomal extracts or in vivo (Fig. S2D and E) suggested that fasting-induced P2X4R expression pattern in lysosomes resulted from an acidic proteolytic activity. Based on these data, reminiscent of reported post-translational processing of proteins involved in lysosome-autophagosome fusion,³¹ we reinforce that P2X4R would be involved in the regulation of autophagy. These experiments also showed similar expression of LAMP2A in lysosome fractions from WT and P2X4R-KO mice in fed and fasted conditions, suggesting that P2X4R did not impact chaperone-mediated autophagy (Fig. S2A).

P2X4R facilitates the fusion between lysosomes and autophagosomes

Consistent with these in vivo data, we observed an elevated autophagic flux in P2X4R overexpressing human embryonic kidney (HEK) cells, which was subsequently inhibited by BAY1797 treatment, as shown in LC3 Western blots (Fig. 3A). We confirmed these data after further transfection with LC3-GFP-RFP reporter plasmid, as explained in the Materials and methods section (Fig. 3B). In those experiments, P2X4R overexpression increased, whereas BAY1797 treatment significantly reduced the number of autolysosomes (red dots). Together with the fact that P2X4R inhibition did not impact autophagosome formation (Fig. S1B), these data indicated that P2X4R promoted the autophagic flux in hepatocytes by facilitating the fusion between lysosomes and autophagosomes. Building on our previous observations that lysosome positioning was altered in hepatocytes when P2X4R was lacking,⁶ we further found that lysosomes exhibited P2X4R-dependent zonated location at the lobular scale. In Fig. S3A, LAMP-1 staining in the PC area was predominantly pericanalicular (co-localizing with dipeptidyl peptidase 4 [DPP4] staining, Fig. S3B), whereas it was more cytosolic in PP hepatocytes (more distant from DPP4 staining, Fig. S3A and C). This pattern was found in both fed and fasted WT mice; conversely, this zonated distribution was altered in P2X4R-KO mice, especially the predominant pericanalicular position of lysosomes was less obvious in the PC area (Fig. S3A and B). Notably, upon leupeptin treatment, pan-lobular pericanalicular LAMP-1 staining was observed in WT mice but not in P2X4R-KO mice (Fig. S3A). Interestingly, leupeptin-induced pericanalicular lysosome redistribution was altered upon P2X4R inhibition (Fig. S3A–C). Based on these observations, we hypothesized that P2X4R could regulate lysosome trafficking and position along cytoskeleton networks, and thereby their fusion with autophagosomes during fasting.³² Lysotracker-stained mouse embryonic fibroblasts (MEFs) revealed that lysosomes exhibited a clear biphasic movement upon fasting. In complete medium, they concentrated in a perinuclear ring; in Earle's balanced salt solution (EBSS) medium (nutrient-deprived medium), they progressively moved toward the cell periphery (during the first 2 h), and then they came back to their perinuclear location (at 5 h, EBSS medium) (Fig. 3C). This positioning pattern had previously been reported in MEFs.³³ Interestingly, P2X4R inhibition with BAY1797 significantly altered the first phase of lysosome movement toward the periphery, retaining lysosomes in the perinuclear region. These data suggest that P2X4R interferes with lysosome movement during fasting and thereby may contribute to autophagosome-lysosome meeting in the late phase of autophagy.

Fig. 3.

P2X4R facilitates the fusion between lysosomes and autophagosomes.

(A) Autophagic flux analyzed in HEK-mCherry cells (control cells) and mCherry-P2X4R-overexpressing HEK cells, based on LC3 Western blot analysis as in Fig. 1C. Representative images and quantitative analysis of n = 4–6 experiments per condition. (B) Autophagic flux analyzed in HEK cells expressing the tandem plasmid LC3-RFP-GFP, either on a “WT” or “P2X4R-overexpressing” (P2X4R) background. Yellow and red dots were analyzed in cells cultured in complete (CM) or nutrient-deprived (EBSS) medium, either treated or not (vehicle) with BAY1797 (BAY). Representative images and analysis from five independent experiments. Scale bar: 20 μm. (C) Lysosome positioning analysis in lysotracker-stained MEFs (Mouse Embryonic Fibroblasts), in complete (CM) or nutrient-deprived (EBSS) medium, either treated or not (vehicle) with BAY1797 (BAY). Representative images and analysis from n = 40–75 cells in each condition, in three to four independent experiments, as explained in Supplementary Materials and Methods. Scale bar: 20 μm. For A–C: ∗p ≤0.05; ∗∗p ≤0.01; ∗∗∗p ≤0.001 (Student’s t test). Data are mean ± SEM. CQ, chloroquine; EBSS, Earle's balanced salt solution.

P2X4R modulates LD size

As a crucial function of autophagy process in the liver is to handle LDs, we analyzed fasting-induced steatosis in WT mice treated with or without BAY1797 (Fig. 4A). Analysis of H&E and Oil Red O-stained liver sections demonstrated that significantly smaller LDs accumulated upon BAY1797 treatment compared with vehicle treatment (Fig. 4B–D). Similarly, LDs were smaller in P2X4R-KO than in WT livers, and most importantly the effect of BAY1797 on LD size was not observed in KO mice, confirming BAY1797 specificity on mouse P2X4R (Fig. 4B,C and Fig. S4). We also provided in vitro data for BAY1797 specificity, as shown by ATP-induced Ca²⁺ signals analyzed in HEK cells overexpressing P2X4R or not, treated with or without BAY1797 (Fig. S5). Interestingly, the change in LD size was not associated with any significant difference in hepatic TG content between the different mice groups (Fig. 4E). As a control, we also confirmed that fasting-induced peripheral release of FFAs did not vary between vehicle- and BAY1797-treated mice (Fig. 4E). Assessment of the hepatic expression of adipose tissue triglyceride lipase (ATGL, the limiting enzyme for TG degradation) revealed that BAY1797 (compared with vehicle) treatment induced both ATGL protein and mRNA expression in fasting conditions (Fig. 4F). Importantly, these in vivo data were also obtained in vitro in MEFs after confirming that the cells expressed P2X4R (Fig. S6A). In those cells, BAY1797 treatment significantly induced ATGL protein expression (Fig. S6B). These data suggested that, although autophagy was inhibited, cytosolic lipolysis would be stimulated upon P2X4R inhibition in the fasting-induced steatosis model.

Fig. 4.

P2X4R inhibition reduces LD size during fasting-induced steatosis.

(A) Experimental design. WT and P2X4R-KO mice were treated with vehicle (PEG 400) or BAY1797 (BAY) for 3 days and fed or fasted during the last 48 h. n = 10 mice per group. (B) Representative images of H&E-stained liver sections from WT and P2X4R-KO mice, treated with vehicle (Veh) or BAY1797 (n = 10 mice per group). Scale bar: 50 μm. (C) Diagrams of hepatic LD size analysis in the different WT and P2X4R-KO mice groups (n = 10 mice per group). (D) Representative images of Oil Red O-stained liver sections from fasted WT mice treated or not with BAY1797. Upper panels: optic microscopy, Obj. ×10, scale bar: 50 μm. Lower panels: confocal microscopy, Obj ×60 (zoom 2), scale bar: 10 μm. (E) Liver triglyceride (TG, mg/g liver) and plasma free fatty acids (FFA, mmol/L) concentration measurements in the different mice groups. (F) Adipose tissue triglyceride lipase (ATGL) protein and mRNA expression in the different mice groups. For A–C: ∗p ≤0.05; ∗∗p ≤0.01; ∗∗∗p ≤0.001 (Student’s t test). Data are mean ± SEM. LD, lipid droplet; PEG, polyethylene glycol; H&E: Hematein & Eosin.

Importantly, our in vivo data on P2X4R-related LD size modifications were further confirmed by in vitro analysis of cultured cells loaded with oleic acid (OA) (Fig. 5A). We used Tet-off Atg5-/- MEFs²⁹ cultured with or without doxycycline (see Supplementary Materials and Methods and Fig. S6C), then loaded with OA during 2 h in the presence of vehicle or BAY1797, and analyzed their LD content after staining with LipidTox Green. OA-loaded MEFs (Dox-, Atg5 expressing cells) exhibited LD accumulation, whereas BAY1797 treatment significantly reduced LD size in these cells (Fig. 5A). When autophagy was dampened (Dox+, Atg5 extinction), we observed that BAY1797 treatment did not result in significant LD size reduction (Fig. 5A). Interestingly, experiments in which OA was removed from the medium after loading confirmed that BAY1797 treatment significantly slowed down LD clearance (Fig. 5B). We further reinforced these data by experiments on OA-loaded primary WT and P2X4R-KO hepatocytes. Compared with vehicle treatment, BAY1797 treatment resulted in the disappearance or significant decrease of large LDs and increase of smaller LDs in WT hepatocytes but not in P2X4R-KO hepatocytes.

Fig. 5.

P2X4R inhibition reduces LD size in vitro through inhibition of lipophagy.

(A) LD size analysis in Atg5-/- MEF (Tet-off), after oleic acid (OA) loading, without (vehicle, Veh) or with BAY1797 (BAY) treatment. Small LDs were defined as ranging from 0.01 to 0.2 μm². Scale bar: 25 μm. Obj. ×60 (zoom ×2), representative confocal microscopy images of five experiments. (B) LD clearance analysis after OA removal from the culture medium, in Veh and BAY-treated MEFs, at zero time point (H0), 3 h (H3), and 6 h (H6). H0 images in Veh and BAY1797-treated cells were similar, only one image is shown. Scale bar: 20 μm. Obj ×60 (zoom ×1), representative confocal microscopy images of six experiments. (C) LD size analysis in OA-loaded primary mouse hepatocytes upon Veh or BAY1797 treatment. Scale bar: 20 μm. Obj ×60 (zoom ×1), representative confocal microscopy images of four to six experiments. For A–C: ∗p ≤0.05; ∗∗p ≤0.01; ∗∗∗p ≤0.001 (Student’s t test). Data are mean ± SEM. LD, lipid droplet.

Together, these data suggested that inhibiting P2X4R reduced LD clearance by inhibiting lipophagy, and stimulated cytosolic lipolysis, resulting in the reduction of LD size.

P2X4R inhibition reduced LD size in metabolic dysfunction-associated fatty liver disease experimental models in mice

Given the data described above, we aimed at studying the impact of P2X4R targeting with the specific antagonist BAY1797 on diet-induced steatosis in the setting of two experimental mouse models—MCD and HFHSC (Materials and methods; Fig. 6A). In the MCD model, liver sections after 15 days showed massive LD accumulation, either predominating in the PC area or occupying the entire lobule (Fig. 6B and Fig. S7). Interestingly, we found that BAY1797 (compared with vehicle) treatment during the last 3 days of the diet, resulted in a significant reduction in LD size in WT mice but not in P2X4R-KO mice, without significantly changing the hepatic TG content (Fig. 6C and Fig. S7A and B). In the HFHSC diet model, mice exhibited massive steatosis, with a clear predominance in the PC area (Fig. 6D and Fig. S7C). Strikingly in this model, lobular distribution of LDs was reportedly heterogeneous, with small LDs in the PC zone and large LDs in the PP area.³⁴ Treatment with BAY1797 but not with vehicle for the last 3 days similarly resulted in LD size reduction in WT mice but not in P2X4R-KO mice, with a clear disappearance of PP large LDs (Fig. 6D and Fig. S7C). As observed in the MCD model, BAY1797 treatment did not interfere with liver TG concentration in HFHSC-fed mice (Fig. 6E). Importantly, ATGL protein expression was significantly induced in BAY1797-treated mice compared with control mice (Fig. 6E). Together with the lack of liver TG increase, these data suggest that stimulated cytosolic lipolysis balances the reduction of lipophagy in the HFHSC model, consistent with our observation in the fasting-induced steatosis model (Fig. 4F). This hypothesis was further reinforced by the fact that BAY1797 treatment did not increase circulating plasma TG concentration (Fig. S8A) (as it would have been expected if TG export had been stimulated). In parallel, our data demonstrated that BAY1797 treatment was associated with reduced fatty acid beta-oxidation and de novo lipogenesis, as revealed by hepatic gene mRNA expression analysis in both fasting- and diet-induced steatosis models (Fig. S8B).

Fig. 6.

P2X4R inhibition reduces LD size in steatotic liver disease experimental models in mice.

(A) Experimental design of the methionine- and choline-deficient (MCD) diet and high-fat, high-sucrose, cholesterol-containing (HFHSC) diet models of hepatic lipid overload. WT and P2X4R-KO mice were fed MCD diet for 2 weeks or HFHSC diet for 2 months. During the last 2 days, mice were treated with vehicle (PEG 400) or BAY1797. (B) Representative images of Oil Red O-stained liver sections from MCD-fed mice (n = 9 mice per group), and LD size analysis. Confocal microscopy. Obj. ×60 (zoom 2). Scale bar: 10 μm. (C) Liver triglyceride (TG, mg/g liver) measurement in MCD-fed mice (n = 9 mice per group). (D) Representative images of H&E-stained liver sections from HFHSC-fed mice (n = 9 mice per group), and LD size (μm²) analysis. Obj. ×20. Scale bar: 50 μm. (E) Left: liver TG (mg/g liver) measurement in HFHSC-fed mice (n = 9 mice/group). Right: hepatic ATGL protein expression in HFHSC-fed WT and P2X4R-KO mice, treated or not (Veh) with BAY1797. For A, C, and E: ∗p ≤0.05; ∗∗p ≤0.01; ∗∗∗p ≤0.001 (Student’s t test). Data are mean ± SEM. LD, lipid droplet; PC, pericentral; PP, periportal.

In association with LD size reduction in the liver upon BAY1797 treatment (compared with vehicle) in the HFHSC model, we observed less hepatic inflammation and injury: (1) inflammatory cell clusters were significantly reduced in number and size on H&E-stained liver sections (Fig. 7A); (2) MCP1 mRNAs were reduced, whereas IL10 transcripts were induced (Fig. 7B); (3) less fibrogenic markers (TGFβ and Col1α mRNAs) (Fig. 7C); and (4) less plasma ALT elevation (Fig. 7D). In both MCD and HFHCS models, less plasma FFA concentration upon BAY1797 treatment compared with vehicle treatment may also reflect the reduced lipophagy (Fig. S8A). We also observed that P2X4R inhibition was associated with an enhanced adaptative response to endoplasmic reticulum stress compared with vehicle-treated livers (Fig. S8C). Importantly, none of the BAY1797 treatment effects were observed in the lack of P2X4R (Fig. 7).

Fig. 7.

P2X4R inhibition reduces inflammation and injury in steatotic liver disease experimental models in mice.

(A) Representative images of infiltrating inflammatory cell clusters on H&E-stained liver sections, and semiquantitative analysis from HFHSC-fed WT and P2X4R-KO mice, treated or not (Veh) with BAY1797 (n = 10 mice per group) (see Fig. 6A). Inflammatory cell clusters are indicated by arrow heads). Entire liver sections were analyzed. Obj. ×20. Scale bar: 100 μm. (B) Quantitative PCR analysis of IL10 and MCP1 mRNAs in livers from WT and P2X4R-KO mice, fed a control diet (CD) or HFHSC diet, and either or not (Veh) treated with BAY1797 (n = 10 mice per group). (C) Quantitative PCR analysis of TGFβ and Col1α mRNAs in livers from WT and P2X4R-KO mice, fed a control diet (CD) or HFHSC diet, and either or not (Veh) treated with BAY1797 (n = 10 mice per group). (D) Plasma transaminases in WT and P2X4R-KO mice, fed HFHSC diet, and either or not (Veh) treated with BAY1797 (n = 10 mice per group). For A–D: ∗p ≤0.05; ∗∗p ≤0.01; ∗∗∗p ≤0.001 (Students’ t test). Data are mean ± SEM. ALT, alanine aminotransferase; HFHSC, high-fat, high-sucrose, cholesterol-containing diet; TGFb: Transforming Growth Factor Beta; Col1a: Collagen 1 alpha; MCP1: Monocyte Chemoattractant Protein 1

Taken together, these data suggest that P2X4R antagonism, through an inhibition of lipophagy and an associated activation of lipolysis, reduced the LD size and had a hepatoprotective impact in murine MASH models.

Correlated P2X4R expression, autophagic flux, and LD size in human patients with MASLD

Patients with MASLD (n = 24) undergoing hepatectomy for liver metastasis and healthy control patients (n = 9) were selected by pathologists at the Centre de Ressources Biologiques Paris-Saclay (France). Patients matched for age, sex, and histological score (Table S2). Most patients with MASLD harbored mixed LD size distribution, with or without predominantly large or small LDs according to the pathologist report. We first observed that P2X4R mRNA was significantly reduced in livers of patients with MASLD compared with control healthy patients (Fig. 8A). Interestingly, we found that P2X4R expression appeared to be associated with LD size as estimated on the pathologist report in patients with predominantly “small” or “large” LDs; however, in a limited number of patients who did not exhibit hepatic inflammatory infiltration (Fig. 8B and Table S2), P2X4R expression was reduced in patients with predominantly small LDs. Importantly, we found that P2X4R expression was significantly correlated with the autophagic flux as estimated on LC3II and p62 relative expression (Fig. 8C, left panel),³⁵ the more P2X4R was expressed, the higher was the autophagic flux. This observation was performed in livers from MASLD patients with no or very few histological signs of inflammatory cells infiltrate. In contrast, in patients with inflammatory livers, P2X4R and autophagic flux were not correlated (Fig. 8C, right panel). This contrast was in line with the fact that P2X4R is highly expressed in inflammatory cells,[6], [7], [8] likely explaining why any P2X4R/autophagic flux correlation in hepatocytes would be biased in inflammatory conditions. Accordingly, further analysis of patient samples showed that hepatic P2X4R expression was positively correlated with histological inflammation (Fig. 8D).

Fig. 8.

P2X4R expression in patients with MAFLD is correlated with autophagic flux and LD size.

(A) Quantitative PCR mRNA analysis of P2X4R expression in MAFLD (n = 24) and control (n = 9) patient liver samples. (B) Quantitative PCR mRNA analysis of P2X4R expression in MAFLD patient liver samples with predominantly small (“micro”, n = 3) or large (“macro”, n = 6) LDs, among non-inflammatory liver samples. (C) Correlation between autophagic flux estimated based on LC3 and p62 protein expression and P2X4R mRNA expression (qPCR analysis). Left graph: patients without inflammation (n = 8); right graph: patients with inflammation (n = 6). Pearson’s correlation analysis. (D) Correlation between hepatic inflammation based on the pathologist report and scoring, and P2X4R mRNA expression (qPCR analysis). Pearson’s correlation analysis. For A and B: ∗p ≤0.05; ∗∗p ≤0.01; ∗∗∗p ≤0.001 (Student’s t test). Data are mean ± SEM. LD, lipid droplet; MAFLD, metabolic dysfunction-associated fatty liver disease.

Discussion

The dominant evidence-based message from the literature is that autophagy would be beneficial for LD elimination from hepatocytes and that strategies aiming at enhancing lipophagy are protective against fat accumulation in the liver.^18,23,36 In this study, we report that inhibition of the purinergic P2X4R during different fatty liver pathophysiological settings, although resulting in dampened macroautophagy, is associated with reduced LD size, hepatic inflammation, and injury. Together with data found in patients with MASLD, this study encourages therapeutic strategies targeting P2X4R for a finely tuned interference with LD metabolism, thereby reducing its toxicity for liver parenchyma.

Our results suggest that P2X4R is a positive regulator of autophagy, likely at the terminal (lysosomal) phase, as previously suggested in extrahepatic contexts.^16,17 Indeed, both genetic deletion and pharmacological inhibition of P2X4R consistently impaired autophagic flux, without altering autophagosome formation. Although autophagic activity is primarily regulated at its initiation step rather than the fusion step, our data suggest a P2X4R-mediated facilitation of the autophagosome-lysosome fusion.

Mechanistically, P2X4R localization on lysosomal membrane would thus somehow interfere with lysosomal function in the autophagy process. As P2X4R is a receptor channel inducing Ca²⁺ entry, meaning Ca²⁺ release from the lysosome toward the cytosol, several nonexclusive hypotheses may be proposed, including CamKinase-dependent activation of membrane fusion processes.¹⁴ Moreover, altered lysosome positioning in hepatocytes in the lack of P2X4R that we described in a previous study⁶ and in the present study suggest that P2X4R-dependent Ca²⁺ movements across the lysosome membrane may regulate important protein interactions for lysosome anchoring and/or transport on cytoskeleton networks, as previously reported for TRPML1, another lysosomal Ca²⁺ release channel.³⁷ Importantly, MASLD patient liver samples analysis suggested that P2X4R would also be correlated with autophagic flux and LD size in humans.

Our data indicate that P2X4R inhibition is associated with dual, balanced effects on TG degradation; on the one hand lipophagy is inhibited, and on the other hand cytosolic lipolysis is activated. Although the induction of hepatic ATGL expression upon P2X4R inhibition does not warrant any corresponding enhancement of ATGL activity, the lack of TG increase (both in liver and plasma)—despite reduced lipophagy—indicates a parallel increase in TG degradation through cytosolic lipolysis. The two processes have already been reported to enhance each other;^21,38 however, the idea of reciprocal regulation of lipophagy and cytosolic lipolysis is not well established and should depend on the cellular and pathophysiological contexts. Analysis of lipid metabolism enzyme mRNA expression is also in agreement with the observed reduced lipophagy (Fig. S8B). Our data agree with the fact that cytosolic lipolysis degrades lipids from large LDs, resulting in small LDs that are subsequently degraded through lipophagy.³⁹ The striking disappearance of large LDs from PP areas upon BAY1797 treatment in the mice fed HFHSC diet is in line with this interpretation. Importantly, strategies aiming at overstimulating cytosolic lipolysis reportedly worsen liver injury because of the release of lipotoxic FFA,⁴⁰ suggesting that a more nuanced targeting of LD homeostasis should be envisioned. In this perspective, small LDs are reportedly associated with cell protection during lipolysis, less lipotoxicity-related endoplasmic reticulum stress,⁴¹ and less oxidative stress.⁴² Recent reports also highlighted that small LDs were associated with less mechanical stress in hepatocytes compared with large LDs.^43,44 As TG is a non-toxic lipid, incorporating FFA in TG into small LDs might reduce both lipotoxicity and mechanical stress, compared with large LDs and the related cytosolic release of FFA.^39,45 Thus, P2X4R inhibition-related beneficial effects on inflammatory processes observed in our study may be linked with these LD size reduction consequences. However, as P2X4R is expressed in inflammatory cells, including macrophages and lymphocytes, and influences their migration, adhesion, and activation,^46,47 the effect of systemic P2X4R inhibition on inflammatory liver infiltration in MASH models could also result from a direct impact on these cells, a possibility that this study cannot rule out.

The differential intracellular position of lysosomes across the lobule (Fig. S3) may reflect different levels of autophagic activity in hepatocytes, as lysosome positioning is a well-established readout of lysosomal functions.⁴⁸ To this respect, when inhibiting the terminal phase of autophagy in vivo, autophagosomes accumulated mainly in hepatocytes from the PC area of the lobule (Fig. 1D), suggesting that autophagy would be less active in this lobular zone. Pericanalicular lysosomes would thus be less active than more cytosolic ones, a hypothesis in line with pericanalicular accumulation of lysosomes upon leupeptin treatment (Fig. S3). We also provided evidence for a zonal effect of P2X4R-related autophagy inhibition on PC hepatocyte size (Fig. 2C,D), similarly indicating less autophagy in the PC area. Additionally, during MASLD, a condition in which autophagy is inhibited,^49,50 steatosis is predominant in the PC zone^51,52 and starts in this area.⁵³ Accordingly, our study further emphasizes a zonated LD size distribution in the lobule, with smaller LDs accumulating in the PC than in the PP area, suggesting a reciprocal zonation of lipophagy (which would be less active in PC hepatocytes) and cytosolic lipolysis (which would be less active in PP hepatocytes). This view is in agreement with recent spatial transcriptomic findings⁵⁴ and remains underexplored.

Apart from hepatocytes, macroautophagy reportedly governs phenotypic changes in HSCs via controversial mechanisms involving profibrogenic LD degradation⁵⁵ and antifibrogenic inhibition of extracellular vesicles release;⁵⁶ however, this field remains insufficiently explored. Because P2X4R had profibrogenic properties in liver myofibroblasts in mice and humans,⁷ we hypothesize that a slower autophagic/lipophagic flux in liver myofibroblasts could be responsible for the low fibrotic phenotype observed in P2X4R-KO mice. P2X4R antagonism effects on LD metabolism in HSCs, compared with those observed in hepatocytes, warrant dedicated experiments.

In this study, we provided in vivo and in vitro evidence that P2X4R-related signaling contributes to LD pathophysiology. We found that P2X4R contributes to LD size regulation in experimental settings of fatty liver in mice. Upon P2X4R antagonism, mechanisms involved include an inhibition of the late (lysosomal) step of autophagy through interference with lysosome-autophagosome fusion. Both molecular machineries of lysosome-autophagosome membrane fusion and of lysosome transport might be impacted by P2X4R-related calcium signals. Overall beneficial effects of P2X4R inhibition on liver inflammation and lipotoxicity in mice, as well as MASLD patient data, suggest that translational perspectives based on P2X4R targeting in MASLD could be envisioned.

Abbreviations

ALT, alanine aminotransferase; ATGL, adipose tissue triglyceride lipase; BAY, compound BAY1797 from Bayer Labs; CLV, centrolobular vein; CQ, chloroquine; DPP4, dipeptidyl peptidase 4; EBSS, Earle's balanced salt solution; FFA, free fatty acid; HEK cells, human embryonic kidney cells; HFHSC diet, high-fat, high-sucrose, cholesterol-containing diet; HSC, hepatic stellate cell; LD, lipid droplet; MAFLD, metabolic dysfunction-associated fatty liver disease; MASH, metabolic dysfunction-associated steatohepatitis; MASLD, metabolic dysfunction-associated steatotic liver disease; MCD diet, methionine- and choline-deficient diet; MEFs, mouse embryonic fibroblasts; NAFLD, non-alcoholic fatty liver disease; OA, oleic acid; PC, pericentral; PEG, polyethylene glycol; PP, periportal; PV, portal vein; P2X4R, P2X purinergic receptor 4; TG, triglyceride; Veh, vehicle.

Authors’ contributions

Concept and design: TT, GM, TP, IG.

Experiments and procedures: TP, IG, OF, MM, ID, OD, CLG, LB, JG, FC, CG.

Writing of the article: TT.

Data availability statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Financial support

This study was supported by Agence nationale de la recherche (ANR) (13-BSV1-0008-01).

Conflicts of interest

The authors declare no conflicts of interest that pertain to this work. Please refer to the accompanying ICMJE disclosure forms for further details.

Supplementary data

The following are the Supplementary data to this article: