Research ArticleLIVER INJURY

LC3-associated phagocytosis protects against inflammation and liver fibrosis via immunoreceptor inhibitory signaling

See allHide authors and affiliations

Science Translational Medicine  15 Apr 2020:
Vol. 12, Issue 539, eaaw8523
DOI: 10.1126/scitranslmed.aaw8523

Phagocytic liver protection

Chronic liver injury is characterized by sustained inflammation that can lead to organ failure. Reprogramming the immune response toward an anti-inflammatory phenotype could help limit the inflammation during liver injury. Now, Wan et al. showed that a noncanonical form of autophagy called LC3-associated phagocytosis (LAP) is a protective mechanism activated in monocytes during liver injury that inhibit proinflammatory and profibrogenic pathways in human samples and in a mouse model of liver injury. LAP blockade in monocytes increased liver damage, whereas increasing LAP reduced inflammation and fibrosis in a mouse model of liver injury. The results suggest that LAP might be targeted for treating liver injury.

Abstract

Sustained hepatic and systemic inflammation, particularly originating from monocytes/macrophages, is a driving force for fibrosis progression to end-stage cirrhosis and underlies the development of multiorgan failure. Reprogramming monocyte/macrophage phenotype has emerged as a strategy to limit inflammation during chronic liver injury. Here, we report that LC3-associated phagocytosis (LAP), a noncanonical form of autophagy, protects against hepatic and systemic inflammation during chronic liver injury in rodents, with beneficial antifibrogenic effects. LAP is enhanced in blood and liver monocytes from patients with fibrosis and cirrhosis. Pharmacological inhibition of LAP components in human monocytes from patients with cirrhosis or genetic disruption of LAP in mice with chronic liver injury exacerbates both the inflammatory signature in isolated human monocytes and the hepatic inflammatory profile in mice, resulting in enhanced liver fibrosis. Mechanistically, patients with cirrhosis showed increased monocyte expression of Fc fragment of IgG receptor IIA (FcγRIIA) and enhanced engulfment of immunoglobulin G in LC3+ phagosomes that triggers an FcγRIIA/Src homology region 2 domain–containing phosphatase-1 (SHP-1) inhibitory immunoreceptor tyrosine-based activation motif (ITAMi) anti-inflammatory pathway. Mice overexpressing human FcγRIIA in myeloid cells show enhanced LAP in response to chronic liver injury and resistance to inflammation and liver fibrosis. Activation of LAP is lost in monocytes from patients with multiorgan failure and restored by specifically targeting ITAMi signaling with anti-FcγRIIA F(ab′)2 fragments, or with intravenous immunoglobulin (IVIg). These data suggest the existence of an ITAMi-mediated mechanism by which LAP might protect against inflammation. Sustaining LAP may open therapeutic perspectives for patients with chronic liver disease.

INTRODUCTION

Chronic liver injury develops in response to alcohol, nonalcoholic steatohepatitis (NASH), or viral hepatitis, and exposes to fibrosis and end-stage cirrhosis (1, 2). Sustained hepatic inflammation originating from monocytes/macrophages is crucial for progression of chronic liver diseases (1, 3). In response to hepatocyte stress, damage and death, and pathogen-associated molecular patterns (PAMPs), Kupffer cells (the resident liver macrophages) acquire a proinflammatory phenotype. The release of proinflammatory cytokines and chemokines leads to the recruitment of blood monocytes that infiltrate the liver and perpetuate the inflammatory response, resulting in the development of fibrosis (13). At advanced stages of the disease, cirrhosis is characterized by severe immune dysfunction and sustained systemic inflammation, arising from increased number of immune cells, particularly inflammatory monocytes that drive patients to organ failure, a syndrome referred to as acute-on-chronic liver failure (ACLF) (4, 5), and ultimately to death. Therefore, there is growing interest in the identification of mechanisms that reprogram monocyte/macrophage phenotype to limit inflammation during chronic liver disease.

LC3-associated phagocytosis (LAP) is a noncanonical form of autophagy that has recently emerged as a major mechanism for monocyte/macrophage phenotype reprogramming to an anti-inflammatory phenotype (6). LAP controls the autoimmune response by preventing auto-antigen presentation, allowing dead cell clearance and dampening of proinflammatory signals. LAP is initiated upon phagocytosis of particles that engage innate immune receptors, such as pattern recognition receptors (toll-like receptors and C-type lectin receptors such as Dectin-1), phosphatidylserine receptor T cell immunoglobulin and mucin domain containing 4 (TIM4) receptors TIM4, or Fcγ receptors, particularly the FcγRIIA [Fc fragment of immunoglobulin G (IgG) receptor IIA] (6, 7). This phagocytic process results in the recruitment of some, but not all, members of the autophagic machinery to the stimulus-containing phagosome, facilitating rapid phagosome maturation, degradation of engulfed pathogens, and modulation of the immune response (6, 7). LAP and autophagy are functionally and mechanistically distinct. In contrast to autophagy, LAP proceeds independently of the preinitiation AMP-activated protein kinase–mammalian target of rapamycin–Unc-51 like autophagy activating kinase (AMPK/mTOR/ULK1) autophagy activating kinase (ULK1) autophagy complex but nevertheless requires some autophagic components, such as the class III Phosphatidylinositol-3-kinase (PI3KCIII) complex and elements of the ubiquitinylation-like, protein conjugation system [autophagy-related gene 5 and 7 (ATG5 and ATG7)] (8). In addition, at early stages of LAPosome formation, the RUN domain–containing cysteine-rich protein Rubicon is needed for producing phosphatidylinositol 3-phosphate from a complex that contains Beclin 1, ultraviolet radiation resistance-associated gene protein, and Vps34 (vacuolar protein sorting). Rubicon is also a positive regulator of the phagocytic reduced form of nicotinamide adenine dinucleotide phosphate (NADPH) oxidase 2 (NOX2) complex, both by binding and stabilizing the phagocytic oxidase p22phox and by promoting assembly of the NOX2 complex (6, 7, 9, 10). In contrast, Rubicon inhibits autophagy by preventing maturation of autophagosome (1113). The need for Rubicon has therefore allowed researchers to experimentally distinguish the respective roles of LAP and autophagy in the control of the inflammatory response (9).

The involvement of autophagy in liver inflammation and fibrosis is only emerging (14, 15). We have shown that mice lacking Atg5 in myeloid cells are more susceptible to liver inflammation and fibrosis when exposed to chronic toxic injury (16, 17). Combining human studies in patients with cirrhosis and mice models, together with pharmacological and genetic approaches, we identify LAP as a protective mechanism against hepatic and systemic inflammation during chronic liver injury, with antifibrogenic consequences. We also highlight a mechanism of LAP action, via stimulation of inhibitory immunoreceptor tyrosine-based activation motif (ITAMi) pathway, after FcγRIIA/Src homology region 2 domain–containing phosphatase-1 (SHP-1) activation. Our results suggest that LAP might constitute an interesting target for the management of chronic liver disease.

RESULTS

LAP is enhanced in blood monocytes from patients with cirrhosis and restrains inflammation

We performed studies in healthy donors and in a cohort of patients with fibrosis or cirrhosis from different etiologies (alcoholic or nonalcoholic fatty liver disease and viral hepatitis C; tables S1 and S2). We first evaluated the status of LC3-II, a marker for autophagy and LAP, by Western blot analysis of protein extracts from peripheral blood mononuclear cells (PBMCs). We observed an increase in LC3-II lipidation in patients, as compared to healthy donors, that was observed whatever the etiology (Fig. 1A). In addition, monocytes from patients showed higher endogenous LC3 fluorescence intensity, as compared to healthy counterparts (Fig. 1B). Increased LC3-II lipidation was further enhanced by the lysosomal pH inhibitor chloroquine in PBMCs and monocytes from patients with cirrhosis and affected more marginally cells from healthy donors (Fig. 1A, bottom, and fig. S1A).

Fig. 1 LAP is enhanced in blood monocytes from patients with liver fibrosis and cirrhosis.

(A) Representative Western blot analysis (n = 5) of LC3 and quantification of LC3II/β-actin in PBMCs from healthy donors, NASH-, alcohol-, and hepatitis C virus (HCV)–related cirrhosis exposed to 30 μM chloroquine (CQ) or vehicle. Results are expressed as fold over healthy. **P < 0.01 for patients with cirrhosis (n = 16) versus healthy donors (n = 12) (top) and *P < 0.05 for chloroquine versus vehicle (n = 8 for healthy and n = 9 for patients). (B) Typical images of LC3 immunofluorescence and quantification of LC3B integrated density. *P < 0.05 for cirrhosis (n = 6) versus healthy (n = 4). AU, arbitrary units. DAPI, 4′,6-diamidino-2-phenylindole. (C) Representative confocal image and mean number of LAPosome-containing IgG latex beads per cell per field. ****P < 0.0001 for cirrhosis (n = 3) versus healthy (n = 3). (D) Representative confocal and quantification of the number of LAPosome containing E. coli bioparticles per individual. Quantification was performed on at least 30 cells from 10 fields per mouse liver and normalized to 100. *P < 0.05 or **P < 0.01 for fibrosis (n = 10) and cirrhosis (n = 15) versus healthy (n = 13). (E) Representative confocal images from cirrhotic monocytes (n = 3), incubated with pHrodo Red E. coli bioparticles and immunostained with either Vps34 or p22phox. (F) Mean number of LAPosome per cell per field quantified as in (C) in monocytes exposed to either vehicle, 3-MA (10 mM), or N8 (30 μM). *P < 0.05 for cirrhosis (n = 3) versus healthy (n = 3) (×63 magnification; scale bars, 5 μm). Statistical analysis was performed using two-way analysis of variance (ANOVA) (A), Kruskal-Wallis (C), or Mann-Whitney U test (B to D and F).

Because both LAP and autophagy control inflammation in monocytes, we investigated whether enhanced LC3-II lipidation in patients with cirrhosis is related to LAP or to canonical autophagy. We investigated the recruitment of LC3 to phagosomes in blood monocytes from patients and healthy donors, after uptake of either IgG-coated beads (Fig. 1C) or pHrodo Red Escherichia coli bioparticles (Fig. 1D). Recruitment of LC3 to pHrodo+ phagosomes (LAPosome) was minimal in monocytes from healthy donors but strongly increased in cells from patients (Fig. 1, C and D). The number of LAPosomes was also increased in patients with noncirrhotic liver fibrosis (Fig. 1D and table S2). The association between cirrhosis and monocyte LAPosome number was confirmed using linear regression (table S3). Although gender was a parameter associated with LAPosome number, bivariate analysis adjusted on gender showed that cirrhosis was still an independent predictor of higher LAPosome number (table S3).

Recruitment of LC3 to LAPosomes paralleled that of PI3KCIII, Vps34, and the p22phox subunit of the NADPH complex (Fig. 1E). Concordantly, disrupting LAP components either by the PI3KCIII inhibitor 3-methyladenine (3-MA) (18) or with an N-terminal 8–amino acid peptide derived from p22phox coupled to HIV-tat protein (N8 peptide), which uncouples p22phox-Rubicon interaction (19), led to a reduction in LAPosome number in monocytes from patients with cirrhosis (Fig. 1F). In contrast, the amount of the autophagy cargo protein p62 (20) did not vary between PBMCs from both groups (fig. S1B). Moreover, activation of the early macroautophagic steps was not observed in PBMCs from patients with cirrhosis. In particular, there was no modification of pThr172-AMPKα as compared to healthy donors (fig. S1B), and cells from patients showed enhanced phosphorylation of ULK1 on Ser757, a phosphorylation site that prevents ULK1 activation and its interaction with AMPK (fig. S1B) (21). These data suggested that LC3-II lipidation is mostly related to LAP rather than canonical autophagy in circulating monocytes from patients with cirrhosis.

We therefore focused on LAP and further investigated the link between LAP activation and inflammation in monocytes from patients with cirrhosis. We found an inverse correlation between the number of LAPosomes and blood quantity of C-reactive protein, an inflammatory biomarker routinely used in clinics (Fig. 2A). There was no correlation between LAPosome number and serum alanine transaminase (ALT) or aspartate transaminase (AST) in patients with cirrhosis (fig. S2A), but as usually observed, serum transaminase amount was low in these patients. The frequency of CD14+IL8+ monocytes was higher in patients with cirrhosis than in healthy participants and was further increased by 3-MA or the N8 peptide (Fig. 2B and fig. S2B). Moreover, gene expression of cytokines and chemokines was also higher in monocytes from patients than in treated cells from healthy donors exposed to either 3-MA, N8 peptide, or the nonselective NOX inhibitor diphenyleneiodonium (DPI; Fig. 2C). Together, these data indicate that LAP is activated in blood monocytes from patients with cirrhosis and restrains systemic inflammation.

Fig. 2 Inhibition of LAP components enhances the inflammatory signature in blood monocytes from patients with cirrhosis.

(A) Inverse correlation between circulating C-reactive protein amount and the number of LAPosome containing E. coli bioparticles in blood monocytes from patients with cirrhosis (n = 14). (B) Percentage of IL-8 production in CD14+ cells from patients with cirrhosis (n = 8) and healthy donors (n = 6). PBMCs were incubated with either vehicle, 10 mM 3-MA, 10 μM DPI, 30 μM N8, or control mutated N8W9A peptide. (C) Reverse transcription polymerase chain reaction (RT-PCR) analysis of inflammatory gene expression in blood monocytes from healthy donors (n = 6 to 8) and patients with cirrhosis (n = 6 to 11). Results are expressed as fold over respective control. *P < 0.05, **P < 0.01, and ***P < 0.001 for cirrhosis versus healthy or inhibitors versus respective control. Statistical analysis was performed using Spearman correlation test (A), two-way ANOVA (B), and Mann-Whitney U test (C).

LAP is enhanced in monocytes/macrophages from livers of patients with cirrhosis and in animal models

Recruitment of LC3 to LAPosomes was enhanced in monocytes/macrophages isolated from livers of patients with cirrhosis, as compared to cells isolated from livers of control patients (Fig. 3A, left). Increase in LAPosome number was also observed in intrahepatic monocytes/macrophages isolated from mice with liver fibrosis, induced by chronic administration of carbon tetrachloride (CCl4) (Fig. 3A, right) and bile duct ligation (BDL), and from mice fed a Western diet (fig. S3A).

Fig. 3 LAP is increased in monocytes from human or mice cirrhotic livers and protects against hepatic inflammation and fibrosis.

(A) Human intrahepatic monocytes were isolated from liver explants of control nontumor or cirrhotic livers. Representative images and mean number of LAPosome per cell per field in intrahepatic monocytes isolated from human livers (left) of control patients (n = 5) and patients with cirrhosis (n = 4) or livers from C57BL6/J mice (right) exposed to vehicle (n = 2) or CCl4 (n = 2) for 7 weeks. Quantification was performed on 30 to 60 cells. **P < 0.01 for cirrhotic versus control liver (human) or CCl4-exposed versus vehicle-exposed liver (mice). (B) Representative liver tissue sections stained by hematoxylin and eosin and IBA1 immunostaining with quantification from CCl4-exposed RubiconMye−/− mice and WT littermates. (C) Hepatic inflammatory gene expression and hepatic cytokine production were respectively assayed by RT-PCR and enzyme-linked immunosorbent assay (ELISA). (D) Representative images of Sirius Red staining, liver hydroxyproline content, and α-SMA immunostaining quantification. *P < 0.05 for RubiconMye−/− (n = 3 to 4) versus Rubiconflox/flox (n = 5 to 6) (B to D). Statistical analysis was performed using Mann-Whitney U test.

LAP in myeloid cells protects against hepatic inflammation and fibrosis in mice models

To investigate the consequences of LAP blockade on inflammation and chronic liver injury, we developed mice lacking Rubicon in the myeloid lineage (RubiconMye−/−) by crossing Rubiconflox/flox (22) with transgenic mice expressing the recombinant Lysozyme-M Cre recombinase. The efficiency of the deletion was confirmed by the decreased Rubicon protein expression in bone marrow–derived macrophages from RubiconMye−/− mice (fig. S3B). In addition, the number of LAPosomes in bone marrow monocytes (BMMs) from CCl4-exposed RubiconMye−/− mice was similar to that of vehicle counterparts and decreased as compared to that of CCl4-exposed Rubiconflox/flox mice (fig. S3C). Deletion of Rubicon in myeloid cells had no effect on CCl4- or BDL-induced liver injury, as reflected by similar amounts of serum ALT and AST in RubiconMye−/− mice and wild-type (WT) counterparts (fig. S3D). However, CCl4-exposed RubiconMye−/− mice showed enhanced inflammatory infiltrate in the liver compared to CCl4-exposed Rubiconflox/flox (control) mice (Fig. 3B, left), without modification of the density of ionized calcium–binding adapter molecule 1 (IBA1) positive macrophages (Fig. 3B, right). In addition, CCl4-exposed RubiconMye−/− mice displayed increased hepatic interleukin-6 (Il6) and tumor necrosis factor (Tnf) expression compared to control mice, and not of Il1β, but hepatic IL1β secretion was increased in CCl4-exposed RubiconMye−/− mice (Fig. 3C). Moreover, CCl4-exposed RubiconMye−/− mice showed exacerbated fibrosis compared to control mice, as evidenced by enhanced Sirius Red staining area, as well as increased number of fibrogenic cells, evaluated by α–smooth muscle actin (α-SMA) staining (Fig. 3D), although liver hydroxyproline content did not statistically differ between groups. Similar results were obtained in RubiconMye−/− mice after BDL (fig. S3E). These data suggest that in response to chronic liver injury, mice deficient in LAP in myeloid cells might develop increased hepatic inflammation and might be more prone to develop liver fibrosis.

IgG-mediated activation of FcγRIIA–SHP-1–ITAMi pathway underlies LAP-mediated anti-inflammatory signaling in patients with cirrhosis

We next characterized the mechanisms underlying endogenous activation of LAP in cirrhotic monocytes and first evaluated the expression of several plasma membrane LAP-triggering receptors, using available microarray results comparing baseline PBMC transcriptome between healthy donors and patients with cirrhosis (5). We found that gene expression of the Fcγ immunoreceptor FCGR2A was higher in patients’ PBMCs as compared with healthy donors, whereas neither the expression of CLEC7A nor that of the anti-inflammatory FCGR2B differed between the groups, and that of TIM4 was decreased (Fig. 4A). Fluorescence-activated cell sorting analysis confirmed that the expression of FcγRIIA in cirrhotic monocytes was increased both at the cell surface and intracellularly, whereas there was no difference in Dectin-1, TIM4, or FcγRIIB (Fig. 4A and fig. S4A). FcγRIIA is a low-affinity IgG receptor that binds all human and mouse IgG (23). It is known for decades that patients with chronic liver diseases show hyperimmunoglobulinemia (24), but the consequences on the inflammatory features of monocytes during cirrhosis remain unclear. A strong increase in IgG immunostaining was observed in monocytes from patients with cirrhosis compared to healthy donors (Fig. 4B and fig. S4B). IgG immunostaining was mainly colocalized in LC3+ and lysosomal-associated membrane protein (LAMP1+) compartments in monocytes from patients but not from healthy donors (Fig. 4B and fig. S4B). LAMP1 immunostaining, commonly used as a phagosome maturation marker, increased in parallel to that of LC3 in patient monocytes, suggesting that, as previously reported, LC3 recruitment allows phagosome maturation (6, 7). These data indicate that during cirrhosis, IgG internalization by monocytes leads to the recruitment of the LAP machinery and to the initiation of an anti-inflammatory response.

Fig. 4 IgG-mediated activation of FcγRIIA underlies LAP-mediated anti-inflammatory signaling in monocytes from patients with cirrhosis.

(A) Left: Gene expression of LAP receptors in PBMCs from patients and healthy donors by microarray analysis. *P < 0.05 for cirrhosis (n = 4) versus healthy (n = 4). Right: Flow cytometry analysis of surface staining of FcγRIIA, FcγRIIB, TIM4, and Dectin-1, and intracellular staining of FcγRIIA expression in whole blood. Data are expressed as geometric mean fluorescence intensity (MFI) of indicated antibody normalized to control isotype. *P < 0.05 and **P < 0.01 for cirrhosis (n = 6 to 25) versus healthy (n = 5 to 13). (B) Representative images of IgG colocalization with LC3+ and LAMP1+ vesicles in blood monocytes from patients with cirrhosis (n = 3) and healthy donors (n = 3). (C) IgG was detected by Western blot in high-performance liquid chromatography (HPLC) fractions of serum proteins (equally loaded, 10 μg per well) from patients with cirrhosis (n = 3), healthy donors (n = 3), and a patient with lupus nephritis. F1, F2, F3, and F4 represent the peak of each HPLC fraction according to their molecular weight. (D) Quantification of the number of LAPosomes per individual in blood monocytes from patients with cirrhosis (n = 7) and percentage of IL-8 or IL6 production in CD14+ cells from PBMCs of patients (n = 7), after cross-linking with a preformed complex [anti-hFcγRII F(ab′)2 fragments + anti-kappa chain]. *P < 0.05 for cross-linked versus control. (E) Mean number of LAPosome per cell per field in THP-1–FcγRIIA+–CD14+ and control THP-1–CD14+ cells exposed to IVIg (10 mg/ml) for 3 hours at 37°C. ****P < 0.0001 for THP-1–FcγRIIA+–CD14+ versus control THP-1–CD14+ cells. Statistical analysis using Mann-Whitney U test (A and E) or Wilcoxon matched-pairs signed rank tests (D).

FcγRIIA can function as a bi-functional receptor to trigger either anti- or proinflammatory signals depending on the type of ligand, a property that can be exploited to modulate inflammatory disease development (25, 26). Thus, cross-linking of FcγRIIA by multimeric IgG immune complexes results in the phosphorylation of immunoreceptor tyrosine–based activation motif (ITAMa) tyrosine residues, triggering proinflammatory signals (2729). In contrast, upon interaction with uncomplexed IgG (either in monomeric or dimeric forms) or with specific antibody F(ab′)2 fragments, the receptor generates ITAMi-dependent signals (30) with anti-inflammatory effects, after stable recruitment and activation of the phosphatase SHP-1 (25, 26). To determine whether serum IgG from patients with cirrhosis was under complexed or uncomplexed form, we analyzed serum IgG fractions from size chromatography column by SDS–polyacrylamide gel electrophoresis on nonreducing conditions. We found that serum IgGs from patients with cirrhosis were essentially in uncomplexed forms but did not contain high–molecular weight IgG complexes, whereas as previously reported (31), serum IgG from a patient with lupus nephritis displayed high–molecular weight IgG complexes (Fig. 4C). Moreover, patients with cirrhosis showed increased amounts of dimeric IgG and not of monomeric IgG, as compared to healthy donors (Fig. 4C and fig. S4C). We therefore hypothesized that interaction of uncomplexed IgG with FcγRIIA may drive LAP to inhibit inflammation via ITAMi in cirrhosis. We first explored the consequences of either multivalent cross-linking or mono-/divalent targeting of the receptor in patient monocytes, thereby mimicking interactions with complexed or uncomplexed forms of IgG, respectively. The induction of ITAMa that blunts ITAMi signaling (25, 26) was promoted by cross-linking of FcγRIIA with a preformed complex [anti-FcγRII F(ab′)2 fragments + anti-kappa chain] and decreased the number of LC3+ phagosomes in monocytes from patients (Fig. 4D and fig. S4D). This effect was associated with an increased capacity of patient monocytes to produce proinflammatory cytokines (Fig. 4D). In contrast, stimulation of ITAMi signaling upon exposure to intravenous immunoglobulin (IVIg) of human acute monocytic leukemia cell line (THP-1) monocytes transfected with the human FcγRIIA resulted in an increase in the number of LC3+ phagosomes (Fig. 4E and fig. S4E) and LC3-II lipidation (Fig. 5A and fig. S5A), as compared to control. Furthermore, immunoprecipitation of FcγRIIA revealed not only its association with SHP-1 in IVIg-stimulated THP-1 cells but also its interaction with LC3 (Fig. 5A and fig. S5A). Association of FcγRIIA/SHP-1 with LC3 was also observed in FcγRIIA immunoprecipitates from monocytes of patients with cirrhosis but not of healthy donors (Fig. 5B and fig. S5B). Moreover, the activated phosphorylation form of SHP-1 (p-SHP-1Y536) colocalized with pHrodo+ phagosomes in patients’ monocytes (Fig. 5C and fig. S5C). In keeping with these results, small interfering RNA (siRNA) silencing of SHP-1 blocked both the increase in LC3 expression and its interaction with FcγRIIA in immunoprecipitates from THP-1 monocytes transfected with the human FcγRIIA, as compared to control siRNA (Fig. 5D and fig. S5D). Together, these data indicate that during cirrhosis, engulfment of IgG by monocytes promotes FcγRIIA-mediated ITAMi signaling, which recruits LC3 and promotes LAP.

Fig. 5 SHP-1 is recruited by FcγRIIA and transmits an anti-inflammatory signal.

(A) Lysates from THP-1–FcγRIIA+–CD14+ cells exposed to IVIg (10 mg/ml) were immunoprecipitated with an anti-hFcγRII monoclonal antibody (mAb) F(ab′)2 fragment and immunoblotted with antibodies to FcγRIIA, SHP-1, and LC3. Representative immunoblot from duplicate experiments is shown. (B) Monocyte lysates from healthy donors (n = 4) and patients with cirrhosis (n = 4) were immunoprecipitated with anti-hFcγRII mAb F(ab′)2 fragment and immunoblotted as in (A). (C) Representative confocal images from monocytes of patients with cirrhosis (n = 4), incubated with pHrodo Red E. coli bioparticles and immunostained with p-SHP-1Y536 (×63 magnification; scale bars, 5 μm). (D) THP-1–FcγRIIA+–CD14+ cells transfected with SHP-1 or control siRNAs were exposed to IVIg (10 mg/ml). Eluates and cell lysates were immunoprecipitated and immunoblotted as in (A). Representative immunoblot from duplicate experiments is shown. (E) Representative confocal images and mean number of LAPosome per cell per field (up) and p-SHP-1Y536+ phagosomes per cell per field (down) in mouse BMMs, exposed to either control or 30 μM chloroquine (×63 magnification; scale bars, 5 μm). **P < 0.01 for hFcγRIIA-Tg versus WT littermates and for chloroquine versus control vehicle. Statistical analysis using Mann-Whitney U test (A to D) or two-way ANOVA (E).

Mice overexpressing FcγRIIA in myeloid cells show increased LAP and are resistant to inflammation and fibrosis

We next examined the consequences of FcγRIIA overexpression in a mice model of advanced chronic liver disease, after 10-week administration of CCl4, using animals overexpressing human FcγRIIA in myeloid cells [hFcγRIIA-Tg (transgenic) mice] and their WT counterparts. The number of LAPosomes was higher and further enhanced by chloroquine in BMM from hFcγRIIA-Tg as compared to WT animals (Fig. 5E); the number of p-SHP-1Y536 phagosomes was also increased in BMM from hFcγRIIA-Tg mice (Fig. 5E) and was not modified upon exposure to chloroquine. When exposed to CCl4, hFcγRIIA-Tg mice showed enhanced number of LAPosomes, as compared to CCl4-exposed WT counterparts (Fig. 6A and fig. S6A). Moreover, as compared to WT counterparts, CCl4-exposed hFcγRIIA-Tg animals showed reduced inflammatory infiltrate in the liver and lower production of hepatic inflammatory cytokines (Fig. 6B), without modification of the number of IBA+ macrophages (fig. S6B). Hepatic immunophenotyping of hFcγRIIA-Tg animals showed no difference in the frequency of CD8+ T and B cells, dendritic cells, and macrophages; increased neutrophil frequency; and decreased CD4+ T cell frequency in animals exposed to CCl4 as compared to vehicle counterparts (fig. S6C). CCl4-exposed hFcγRIIA-Tg mice were also more resistant to liver fibrosis than their WT counterparts, as shown by lower Sirius Red staining as well as reduced number of α-SMA–positive cells, but liver hydroxyproline content did not differ between groups (Fig. 6C).

Fig. 6 FcγRIIA-mediated activation of LAP as a potential therapeutic target.

(A) LAPosome number per individual in intrahepatic monocytes from hFcγRIIA-Tg mice (n = 7) and WT littermates (n = 6) exposed to CCl4 for 10 weeks. Quantification was performed on at least 30 cells from 10 fields per mouse liver and normalized to 100. **P < 0.01 for hFcγRIIA-Tg versus WT littermates. (B) Representative images and quantification of inflammatory infiltrate in hematoxylin and eosin–stained liver sections. Hepatic IL6 and IL1β cytokine production was measured by ELISA. *P < 0.05 and **P < 0.01 for hFcγRIIA-Tg versus WT littermates. (C) Representative images and quantification of Sirius Red staining, liver hydroxyproline content, and α-SMA immunostaining. *P < 0.05 for hFcγRIIA-Tg versus WT littermates. (D) MFI of surface FcγRIIA expression in CD14+ PBMCs from patients with ACLF (n = 7) and healthy donors (n = 5) by flow cytometry. Results from patients with cirrhosis (dashed lines) from Fig. 4A were included. *P < 0.05 for ACLF versus healthy or versus cirrhosis. (E) Number of LAPosomes per individual in blood monocytes from patients with ACLF (n = 7) and healthy donors (n = 6), quantified as in (A). Results from patients (dashed lines), taken from Fig. 1D, were included. **P < 0.01 for ACLF versus cirrhosis. (F) Representative images and quantification of the number of LAPosomes per individual from blood monocytes exposed to either vehicle, anti-hFcγRIIA F(ab′)2 antibody (n = 7), or IVIg (10 mg/ml; n = 6) for 3 hours. *P < 0.05 for anti-FcγRIIA F(ab′)2 antibody or IVIg versus control. Statistical analysis by Mann-Whitney U test (A to E), Kruskal-Wallis (D and E), or Wilcoxon matched-pairs signed rank (F) tests.

As hFcγRIIA direct targeting by anti-FcγRIIA F(ab′)2 fragments has anti-inflammatory effects, we injected anti-hFcγRIIA F(ab′)2 fragments [anti-human CD32 supplied as F(ab′)2, clone AT10; see Materials and Methods for details] to hFcγRIIA-Tg mice during CCl4 administration. The use of specific F(ab′)2 fragments was chosen to avoid nonselective effects of IVIg on other mouse FcγR receptors. There was no difference either in the area of Sirius Red staining, α-SMA staining, or in liver hydroxyproline content between the two groups (fig. S6D).

LAP is lost in patients with cirrhosis with organ failure and restored by targeting ITAMi signaling with anti-FcγRIIA F(ab′)2 fragments or IVIg

Patients with cirrhosis are prone to develop organ failure, ACLF (the most severe form of cirrhosis), and the severity of organ failure and mortality rate in ACLF patients is linked to the intensity of systemic inflammation (4, 5). We therefore hypothesized that FcγRIIA-mediated activation of LAP observed in circulating monocytes from patients with cirrhosis may be switched off in those with ACLF. As compared to blood monocytes from patients with cirrhosis, the increase in FcγRIIA expression was lower in ACLF patients (Fig. 6D and table S4). In addition, there was no increase in the number of LAPosomes in monocytes from patients with ACLF, as compared to healthy donors (Fig. 6E and fig. S6E). Incubation of ACLF monocytes with anti-FcγRIIA F(ab′)2 fragments was able to increase the number of LAPosomes (Fig. 6F). Moreover, LAPosome number increase was also observed upon exposure of ACLF monocytes to IVIg (Fig. 6F), which are already used in the treatment of inflammatory diseases (32).

DISCUSSION

In the present study, combining studies in human samples and mice models, we identify LAP as an anti-inflammatory pathway in monocytes that constrains both systemic and hepatic inflammation during cirrhosis, with potent antifibrogenic effects. We also unravel ITAMi as a LAP-mediated anti-inflammatory pathway. This mechanism is already increased during fibrosis and persists in end-stage cirrhosis but is lost when cirrhosis deteriorates to ACLF, a severe form of the disease characterized by a burst in systemic inflammation and a high risk of mortality (4, 5). LAP can be restored upon exposure of monocytes to monomeric IgG in patients with ACLF.

Our data support a role for LAP, rather than canonical autophagy, in the anti-inflammatory signal carried by monocytes from patients with cirrhosis, based on the lack of variations in AMPK phosphorylation, the increase in the inhibitory phosphorylation of ULK1, and the absence of p62 modulation (6, 7). In this context, the stimulatory effect of chloroquine on LC3-II was in line with the previously reported increase of LC3-II formation on single membrane–bound compartments in chloroquine-treated cells (33), and recruitment of the LAP machinery proteins was observed in LC3-decorated phagosomes, including Vps34 and p22phox. Concordantly, blockade of LAP by blunting either Vps34, NOX, or Rubicon-p22phox interaction exacerbated the inflammatory signature of cirrhotic blood monocytes, suggesting that LAP prevents reprogramming of monocyte/macrophage to a proinflammatory phenotype in patients with cirrhosis. These data suggest that LAP is a compensatory mechanism that ultimately protects against a further burst in systemic and hepatic inflammation and the evolution toward a more severe stage. The role of LAP was corroborated by studies in the liver, showing that this pathway is also increased in hepatic monocytes upon chronic liver injury in patients with cirrhosis and in animal models of chronic liver injury. Moreover, mice deficient for LAP in myeloid cells showed pronounced hepatic inflammatory cell infiltration and cytokine production in response to chronic liver injury, a step required for the development and progression of liver fibrosis (2, 3, 17, 34, 35). Accordingly, LAP-deficient mice in myeloid cells were more prone to develop fibrosis in response to chronic liver injury, highlighting a possible role for LAP in the regulation of inflammation-driven fibrogenesis. These data are in line with our recent results showing a similar phenotype in mice bearing deletion of Atg5 in myeloid cells that develop exaggerated hepatic inflammation and fibrosis when chronically exposed to CCl4 (17). However, because autophagy and LAP share similar and overlapping sets of proteins, a role for both pathways in the regulation of inflammation during cirrhosis cannot be entirely ruled out.

Several signals elevated during cirrhosis (1, 2) could mediate LAP activation, including apoptotic cells via TIM4, PAMPs via pattern recognition receptors, or Igs via Fcγ receptors. Hyperimmunoglobulinemia has been reported as a characteristic feature of cirrhosis (24). Our data further demonstrate that the serum of patients with cirrhosis had increased amounts of uncomplexed IgG, without high–molecular weight complexes, a characteristic feature of autoimmune diseases such as lupus nephritis (31). Moreover, the increased expression of FCGR2A/FcγRIIA in blood monocytes from patients with cirrhosis, associated with an increase in IgG complexes engulfment, suggests that IgG-FcγRIIA signaling is a likely candidate for LAP activation. FcγRIIA has double-edged sword functions, with anti- and proinflammatory properties depending on specific core signaling pathways (25, 26). Our results also unravel a beneficial role for FcγRIIA during chronic liver injury, since mice overexpressing FcγRIIA in myeloid cells are protected against hepatic inflammation with a resulting resistance to liver fibrosis, when exposed to chronic toxic insult. It is likely that this protection is mediated by endogenous mouse IgG binding to human FcγRIIA, as it has been previously shown that all subclasses of mouse IgG can bind to human FcγRIIA (36) and that mice with chronic liver injury, including chronic CCl4 administration, display hyperimmunoglobulinemia (37). The failure of AT10 F(ab′)2 anti-FcγRIIA to enhance FcγRIIA-ITAMi effects indicate that receptors might be occupied by mouse IgG during CCl4 exposure, therefore achieving maximal antifibrogenic effects. Because LAP is already increased in response to chronic liver injury but further augmented in CCl4-exposed mice overexpressing FcγRIIA, it is likely that enhanced LAP results both from the inflammatory environment and FcγRIIA overexpression.

Mechanistically, we have previously described that the inhibitory signal transmitted by FcγRIIA is generated via recruitment of SHP-1 to the receptor after its activation by monomeric IgG (25, 26). Immunoprecipitation studies of monocyte lysates from patients with an anti-FcγRIIA antibody revealed association of the receptor with LC3, an interaction that was blunted upon SHP-1 silencing. These data provide a link between LAP and FcγRIIA–SHP-1–ITAMi signaling to convey an anti-inflammatory response. Previous studies have shown that immune receptors associated with an ITAM motif, such as Dectin-1, are able to enhance LAP after recruitment of LC3 to the phagosome via the spleen tyrosine kinase (Syk) (25, 30). This kinase is required for initiation of signaling by receptors that use ITAM domains and is negatively regulated by SHP-1 (25, 26). However, we have also previously shown that transient activation of Syk by monomeric IgG is needed for recruiting SHP-1 by FcγRIIA and transmitting an anti-inflammatory signal (25, 26). In addition, our data are in line with previous reports showing that SHP-1 associates with phagosomes and promotes their biogenesis (38). Therefore, whether other ITAM-coupled receptors promote LAP via an ITAMi–SHP-1–dependent pathway and potentiate FcγRIIA signaling to promote LAP merits further investigation. In this context, Dectin is of particular interest, since it displays anti-inflammatory and antifibrogenic properties in the liver (39). Furthermore, it is tempting to hypothesize that the anti-inflammatory effects promoted by FcγRIIA-ITAMi signaling in experimental arthritis or lupus nephritis (25, 26) involves LAP. Last, the beneficial role of LAP in other human inflammatory disorders also merits further investigation, in light of our results in patients with cirrhosis, and those reported in mice for lupus erythematosus (40).

There are some limitations to our study. First, rodent models are incomplete representations of human disease, and in particular, there are no models that fully recapitulate features of cirrhosis and ACLF in mice. Second, although our data demonstrate that LAP can be restored in monocytes from patients with ACLF upon exposure to FcγRIIA F(ab′)2 fragments or IVIg, in vivo experiments could not be performed in a relevant model of ACLF. Last, human and mice studies were of limited size, which may preclude revealing differences between groups for some parameters analyzed. Future studies are, in particular, required to investigate whether the response in humans varies according to disease severity and among etiologies, such as alcohol versus NASH versus hepatitis C virus.

In conclusion, our data highlight a LAP-mediated ITAMi pathway, which broadens our knowledge on the general properties of LAP in the regulation of inflammation. They also add to our understanding of the mechanisms underlying inflammation-driven fibrogenesis and systemic inflammation in the context of cirrhosis. From a clinical perspective, our study also has potential major implications. Progression of cirrhosis to ACLF, the most severe form of cirrhosis, is a consequence of an acute accentuation of systemic inflammation over the chronic systemic inflammation already present in cirrhosis (4, 5). However, the mechanisms underlying the burst in inflammation are not fully understood. Our data showed that activation of LAP observed in blood monocytes from patients with cirrhosis is lost in patients progressing to ACLF. LAP could be restored by targeting ITAMi signaling with antibodies to FcγRIIA F(ab′)2 fragments or upon exposure to IVIg, which is already approved by the U.S. Food and Drug Administration. These data suggest that sustaining LAP in patients with cirrhosis might prevent progression to a syndrome that is associated with a high mortality rate.

MATERIALS AND METHODS

Study design

The objective of this study was to determine whether autophagy, particularly LAP, affects hepatic and systemic inflammation during cirrhosis and the resulting effects on liver fibrosis. We combined studies in human blood and liver samples from patients with cirrhosis and control subjects, with mice models, each of them integrating both pharmacological and genetic approaches. LAP was characterized in blood monocytes and the impact of its blockade on the inflammatory signature was studied using pharmacological inhibitors. In the liver, LAP activation status was also investigated, combining studies in liver samples from patients with cirrhosis and from mice exposed to chronic CCl4 or BDL to induce fibrosis. To characterize the role of LAP in hepatic inflammation and fibrosis, we compared the inflammatory and fibrogenic response of mice lacking Rubicon in the myeloid lineage exposed to CCl4 or BDL to their WT littermates. We studied the mechanism involved and focused on the immunoreceptor FcγRIIA, which was selectively induced in monocytes from patients with cirrhosis. We analyzed FcγRIIA and its downstream targets, particularly the anti-inflammatory ITAMi pathway, in human monocytes and in animal models of FcγRIIA overexpression in myeloid cells. Last, we characterized LAP in more severe patients with cirrhosis, progressing to organ failure. The absence of LAP in monocytes from these patients prompted us to investigate whether LAP could be restored by targeting the ITAMi signaling with anti-FcγRIIA F(ab′)2 fragments, or with IVIg.

Human blood samples were obtained from 73 patients with histologically proven stable cirrhosis admitted to the liver unit (table S1), 10 consulting patients with histologically proven liver fibrosis (table S2), and 21 patients with ACLF hospitalized in the intensive care unit (table S4) from Beaujon University Hospital, after approval by the local Ethics Committee [Comité de protection des personnes Ile de France III (no. 3194) and Comité d’Evaluation de l’Ethique des projets de Recherche Biomédicale Paris-Nord (IRB 00006477 no. 13-043)] and after obtaining written informed consent. Noninclusion criteria of patients with stable cirrhosis were as follows: (i) an acute event (hepatorenal syndrome, bacterial infection, and variceal bleeding) within 2 weeks before inclusion, (ii) current treatment with immunosuppressive drugs, (iii) current active alcohol consumption, (iv) presence of human immunodeficiency virus infection, and (v) hepatocellular carcinoma (HCC) outside Milan criteria or active extrahepatic cancer. In patients with ACLF, extensive bacteriological exams were performed to search for bacterial infection as precipitating event. Blood from healthy volunteers (n = 79) was obtained by Etablissement Français du Sang (agreement no. 2015012778). The size of the sample to be studied was determined on the basis of our previous studies investigating inflammatory response in monocytes from patients with cirrhosis, to reach a test power of 0.80 and a significance of 0.05. Processing of cells was performed simultaneously and in parallel for all conditions within each experiment. All data presented have been replicated in at least three biological replicates for in vitro experiments.

Human liver samples were obtained from surgical samples (resection or liver transplantation). In patients with cirrhosis, samples were obtained from liver explant during liver transplantation or from nontumoral liver during HCC resection. Control liver samples were taken from patients with normal liver biological tests who underwent resection surgery for nonhepatocellular primary tumor or colorectal cancer liver metastasis. Fibrosis quantification of all liver specimens was performed by a pathologist expert for liver diseases.

For mouse experiments, study size was selected on the basis of our past experience with models of chronic liver injury. We systematically used littermates for all studies, which facilitates appropriate randomization. BDL surgery was performed by an investigator who was not aware of the genotype or experimental assays to be performed after the surgery. Experiments were performed in accordance with protocols approved by the French Council of Animal Care guidelines and national ethical guidelines of Institut National de la Santé et de la Recherche Médicale (INSERM) Animal Care Committee (authorization no. 02529.02).

Data collection of each experiment was detailed in the respective figures, figure legends, and methods. No data were excluded from studies in this manuscript. Statistical tests were chosen on the basis of the nature of variables, assumption of data distribution, and effect size. In general, continuous variables were analyzed by Mann-Whitney U test or by Wilcoxon matched-pairs signed rank test as indicated, and in case of multiple variables, two-way analysis of variance (ANOVA) was used. Nonparametric Spearman test was used as correlation test.

Mice

Human FcγRIIA-Tg mice expressing the human FcγRIIAR131 in CD11b+ cells were from the Jackson Laboratory (JAX). All mice were of C57BL/6 strain and mice carrying the FcγRIIA transgene were used as heterozygous animals. WT littermates were used as controls. Myeloid cell-specific Rubicon-deficient mice were generated by crossing RBCN-loxP/loxP (22), provided by T. Yoshimori (Osaka University, Japan) to LysM-Cre mice (JAX, Charles River), and backcrossing the resulting double heterozygotes (LysM-Cre+/−, RBCN+/loxP) with RBCN-loxP/loxP mice to produce myeloid-specific Rubicon knockout mice (LysM-Cre+/−, RBCN-loxP/loxP, RubiconMye−/−) and WT littermates (LysM-Cre−/−, RBCN-loxP/loxP, Rubiconflox/flox mice). Rubiconflox/floxLysM-Cre−/− littermates were used as controls.

Statistical analysis

Results are expressed as mean ± SEM or median (interquartile range), as indicated. Comparison between groups was performed using appropriate nonparametric tests, Mann-Whitney U test, or Wilcoxon matched-pairs signed rank tests for continuous variables and chi-square test for categorical variables. Correlations were performed using nonparametric Spearman test. In case of multiple variables, the effect of each variable and their interaction was analyzed using two-way ANOVA. All P values are two-sided, and P values less than 0.05 were considered to be statistically significant. The potential relationship between patient characteristics and LAPosome number was analyzed by linear regression univariate analysis. Each variable achieving P < 0.05 was then introduced into a bivariate model. Analyses were performed using GraphPad Prism version 8 and SPSS 22.0 (SPSS Inc., Chicago, IL, USA).

SUPPLEMENTARY MATERIALS

stm.sciencemag.org/cgi/content/full/12/539/eaaw8523/DC1

Supplementary Methods

Fig. S1. Representative Western blot and quantification of LC3 in monocytes and of p62, pThr172-AMPKα, and pSer757-ULK1, in PBMCs from healthy donors and patients with cirrhosis.

Fig. S2. Lack of correlation between LAPosome number and ALT/AST in blood monocytes from patients with cirrhosis and representative profile of IL-8 expression in CD14+ monocytes from patients with cirrhosis by flow cytometry.

Fig. S3. LAP is increased in monocytes from mice with bile-duct ligation and Western diet and protects against hepatic inflammation and fibrosis.

Fig. S4. IgG-mediated activation of FcγRIIA underlies LAP-mediated anti-inflammatory signaling in monocytes from patients with cirrhosis: quantitative analysis and representative profiles.

Fig. S5. SHP-1 is recruited by FcγRIIA and transmits an anti-inflammatory signal: quantitative analysis.

Fig. S6. FcγRIIA-mediated activation of LAP as a potential therapeutic target: representative images and immunophenotyping.

Table S1. Characteristics of healthy donors and patients with cirrhosis used for studies in blood samples.

Table S2. Characteristics of patients with fibrosis used for studies in blood samples.

Table S3. Factors associated with LAPosome number.

Table S4. Characteristics of patients with ACLF.

Table S5. Primer sequences.

Data file S1. Raw data.

References (4145)

REFERENCES AND NOTES

Acknowledgments: Rubiconflox/flox mice are available from T. Yoshimori (Osaka University, Japan) under a material transfer agreement with the INSERM. We are grateful to INSERM U1149, Paris; M. Siebert for performing BDL in mice; G. Gauthier and V. Gratio from the flow cytometry platform, S. Benadda from the imaging platform, and N. Sorhaindo from the biochemistry platform for help in the histology, flow cytomery, confocal microscopy, and liver function tests, respectively; and J. El Benna for helpful suggestions. We thank O. Thibaudeau (Plateau de Morphologie, INSERM UMR 1152, Paris) and M. Albuquerque (Pathology Department, Hopital Beaujon, Clichy) for providing liver samples. Funding: This work was supported by grants from INSERM (all authors), the Université de Paris (all authors), Labex Inflamex (to S.L., R.C.M., and S.B.M.), the Fondation pour la Recherche Médicale (FRM; DEQ20150331726 to S.L.), the National Research Agency (ANR-18-CE14-0006 to S.L., P.C., R.C.M., and L.S.), and the Societé Française d’Anesthésie-Réanimation (SFAR; to E.W. and T.T.S.). Author contributions: J.W. and S.B.M. designed, conducted, and analyzed data from patients, mice, and cell experiments. E.W. designed and interpreted human blood studies and contributed to analysis and interpretation of human data, clinical insights, and discussion. M.M., P.-M.C., O.P., T.T.-S., P.H., and L.B. contributed to acquisition of data and analysis of in vitro experiments. D.P. provided human blood samples and M.B. developed the RubiconMye−/− mice; J.W., E.W., S.B.M., R.M., and H.G. contributed to writing and provided feedback. L.S., P.C., and R.C.M. contributed to supervision, conception, and design of studies; interpretation of the results; and writing. S.L. conceived, designed, and supervised the study and wrote the manuscript. Competing interests: J.W., E.W., S.B.M., L.S., P.C., R.C.M., and S.L. are inventors on a European patent application EP19305050, submitted by Inserm Transfert on 16 January 2019 that covers the use of agents capable of inducing LAP for treating sustained inflammation in patients suffering from chronic liver disease. Data and materials availability: All data associated with this study are present in the paper or the Supplementary Materials.

Stay Connected to Science Translational Medicine

Navigate This Article