Post-translational modifications drive the effects of HMGB1 in alcohol-associated liver disease

Abstract

Background:

We previously identified that high-mobility group box-1 (HMGB1) is increased and undergoes post-translational modifications (PTMs) in response to alcohol consumption. Here, we hypothesized that specific PTMs, occurring mostly in hepatocytes and myeloid cells, could contribute to the pathogenesis of alcohol-associated liver disease (AALD).

Methods:

We used the Lieber-DeCarli (LD) model of early alcohol-induced liver injury, combined with engineered viral vectors and genetic approaches to regulate the expression of HMGB1, its PTMs (reduced [H], oxidized [O], acetylated [Ac], both [O + Ac]), and its receptors (RAGE, TLR4) in a cell-specific manner (hepatocytes and/or myeloid cells).

Results:

Hmgb1 ablation in hepatocytes or myeloid cells partially protected, while ablation in both prevented steatosis, inflammation, IL1B production, and alcohol-induced liver injury. Hepatocytes were a major source of [H], [O], and [Ac] HMGB1, whereas myeloid cells produced only [H] and [Ac] HMGB1. Neutralization of HMGB1 prevented, whereas injection of [H] HMGB1 increased AALD, which was worsened by injection of [O] HMGB1. While [O] HMGB1 induced liver injury, [Ac] HMGB1 protected and counteracted the effects of [O] HMGB1 in AALD. [O] HMGB1 stimulated macrophage (MF) migration, activation, IL1B production, and secretion. Ethanol-fed Rage ^ΔMye but not Tlr4 ^ΔMye, Rage ^ΔHep, or Tlr4 ^ΔHep mice were protected from AALD, indicating a crucial role of RAGE in myeloid cells for AALD. [O] HMGB1 recruited and activated myeloid cells through RAGE and contributed to steatosis, inflammation, and IL1B production in AALD.

Conclusions:

These results provide evidence for targeting [O] HMGB1 of hepatocyte origin as a ligand for RAGE signaling in myeloid cells and a driver of steatosis, inflammatory cell infiltration, and IL1B production in AALD. Importantly, we reveal that [Ac] HMGB1 offsets the noxious effects of [O] HMGB1 in AALD.

INTRODUCTION

Alcohol-associated liver disease (AALD) is a significant clinical problem and a major cause of morbidity and mortality worldwide. To date, the most effective therapy is alcohol abstinence. The guidelines from the AASLD and the EASL recommend corticosteroids as the first-line treatment in severe alcoholic hepatitis (AH) and pentoxifylline as the alternative in patients with ongoing infections or acute renal failure¹; still, the current treatment options for patients with severe AH and for those who do not achieve abstinence are suboptimal.

Along with steatosis, inflammation is a hallmark of AALD and is in the spotlight of recent NIAAA-sponsored clinical trials.²,³,⁴,⁵ Indeed, the presence of inflammation in early biopsies from patients with alcoholism is of predictive value for disease progression.⁶ Patients with AALD show neutrophilia, infiltration of macrophages (MFs), activation of KCs and/or MFs, and increased production of IL1B.¹,⁷,⁸ Reducing these events has been a primary focus of ongoing efforts to alleviate AALD.⁹

In addition to pathogen-associated molecular patterns, recognizing endogenous molecules of hepatic origin as danger signals is an emerging concept in the activation of innate immunity.¹⁰ Damage-associated molecular patterns (DAMPs) are released by cells upon infection or injury, alerting the immune system to evoke inflammation. Identifying key alarmins and better understanding the molecular mechanisms that drive hepatic inflammation remains challenging in AALD.

HMGB1 is a critical bona fide DAMP upregulated in response to sterile and nonsterile tissue injury.¹¹,¹²,¹³ It became the focus of attention in immunity after its discovery as a potent endogenous instigator of inflammation.¹⁴,¹⁵,¹⁶,¹⁷,¹⁸ While initially believed to be simply an architectural protein,¹⁹,²⁰,²¹,²²,²³ we identified that posttranslational modifications (PTMs) drive HMGB1 nucleocytoplasmic shuttling and secretion in AALD.¹¹,¹² Thus, the HMGB1 isoforms resulting from PTMs could elicit hepatic infiltration of immune cells and proinflammatory cytokines such as IL1B and communicate injury to neighboring cells to drive AALD progression. HMGB1 binds multiple receptors, such as the receptor for advanced-glycation end-products (RAGE), toll-like receptors (TLRs) 2/3/4/9, Mac-1, syndecan-1, phosphacan protein-tyrosine phosphatase-ζ/β, and CD24,²⁴,²⁵,²⁶,²⁷,²⁸ of which RAGE and TLR4 have been implicated in chronic liver disease.²⁹,³⁰,³¹,³²

So far, the specific HMGB1 isoforms, cellular sources, and receptors involved in the pathogenesis of AALD have not been identified. In this study, our central hypothesis was that the increase in HMGB1 production, secretion, and specific PTMs occurring mostly in hepatocytes and myeloid cells could drive steatosis, inflammatory cell infiltration, and IL1B production, contributing to the pathogenesis of AALD. To demonstrate this, we used the Lieber-DeCarli (LDC) model of early alcohol-induced liver injury³³ combined with engineered viral vectors and genetic approaches to regulate the expression of HMGB1, its PTMs (reduced [H], oxidized [O], acetylated [Ac], both [O + Ac]), and its receptors (RAGE, TLR4) in a cell-specific manner (hepatocytes and/or myeloid cells). Our results provide evidence for targeting [O] HMGB1 of hepatocyte origin as a ligand for RAGE signaling in myeloid cells as the primary driver of steatosis, inflammatory cell infiltration, and IL1B production in AALD. Further, we reveal that [Ac] HMGB1 offsets the noxious effects of [O] HMGB1.

METHODS

Ethical guidelines

This project was approved by the appropriate ethics and/or institutional review committee(s) of the University of Illinois Chicago. All animals received humane care according to the Guide for Care and Use of Laboratory Animals guidelines. Housing, breeding, and husbandry conditions were approved by the University of Illinois Chicago IACUC office before initiating the studies. All in vivo experiments were carried out according to the Animal Research: Reporting of In Vivo Experiments guidelines.³⁴

Models of early alcohol-induced liver injury

The LDC model was used to provoke early AALD in mice.³³ An equal number of 10-week-old male and female mice were randomly assigned to either the control or the ethanol group, with 8 mice in each group. The control and ethanol LDC diets (Bio-Serv Inc.) are equicaloric and have the same composition of fat (42% of calories) and protein (16% of calories). The content of carbohydrates is 42% of total calories (dextrin-maltose) in the control diet and 12% of total calories in the ethanol diet, where up to 30% of carbohydrate calories are replaced by ethanol.³³ To determine the contribution of the cellular source of HMGB1 and the involvement of RAGE and/or TLR4 signaling in AALD, an equal number of 10-week-old male and female Hmgb1 ^ΔHep, Hmgb1 ^ΔMye, Hmgb1 ^ΔHepΔMye, Rage ^ΔHep, Rage ^ΔMye, Tlr4 ^ΔHep, Tlr4 ^ΔMye, Hmgb1&Rage ^ΔHepΔMye, and their corresponding control littermates were subjected to the LDC model of AALD. Upon acclimatization by feeding control LDC diet for 3 days, mice were fed the ethanol LDC diet with a progressive increase in the percentage of ethanol-derived calories from 10% (1 wk) to 20% (1 wk), 25% (2 wk), and 30% (2 wk). Mice pair-fed with similar calories derived from maltose-dextrin served as control. To further validate the role of HMGB1 in AALD, a separate cohort of WT mice and Rage ^ΔMye mice was injected i.p., a pan-HMGB1 neutralizing Ab (ST326052233, Shino-Test) at a dose of 0.3 µg/g, [H] HMGB1 (230609, HMGBiotech) or [O] HMGB1 (230817, HMGBiotech) at a dose of 0-0.1 µg/g³⁵ throughout the ethanol feeding protocol.

To explore the role of the HMGB1 PTMs in AALD, Hmgb1 ^ΔHep and Hmgb1 ^ΔHepΔMye mice were transduced with the following adeno-associated viruses serotype-8 (AAV8) under the thyroxine-binding globulin (Tbg) promoter to target hepatocytes specifically: (1) AAV8.Tbg.WT.Hmgb1.Gfp to overexpress native or WT HMGB1 that can undergo all PTMs³⁶; (2) AAV8.Tbg.NLS1(2C→2S).Gfp that cannot generate [O] HMGB1 (Δ[O] HMGB1)³⁶; (3) AAV8.Tbg.NLS1(8K→8A).Gfp that cannot generate [Ac] HMGB1 (Δ[Ac] HMGB1); and (4) AAV8.Tbg.NLS1(2C→2S;8K→8A).Gfp that cannot generate both [O] and [Ac] HMGB1 (Δ[O + Ac] HMGB1). Two weeks later, mice were fed the LDC diets. At the end of each experiment, mice were fasted for 4 hours, blood was drawn by submandibular bleeding under anesthesia, and they were euthanized.

Statistical analysis

Data are expressed as mean ± SEM. Data were analyzed using ANOVA, and n = 8 mice per group were used in all the in vivo experiments.

RESULTS

Hmgb1 ablation in hepatocytes or myeloid cells partially protects, while ablation in both prevents steatosis, inflammation, IL1B production, and alcohol-induced liver injury

Previous work from our laboratory demonstrated that hepatocytes and, to some extent, KCs and infiltrating MFs are a major source of the HMGB1 isoforms in AALD.¹¹ To further define the involvement of these cell-specific HMGB1 isoforms in early alcohol-induced liver injury, WT, Hmgb1 ^ΔHep, Hmgb1 ^ΔMye, and Hmgb1 ^ΔHepΔMye mice were fed the LDC diets.

Hematoxylin and eosin (H&E) staining, the liver-to-body weight ratio, serum ALT activity, liver triglycerides (TGs), and the histopathological scores revealed minimal effect in all mice fed control diet; however, in mice fed the ethanol diet, there was partial protection from liver injury in Hmgb1 ^ΔHep and Hmgb1 ^ΔMye, and significant protection in Hmgb1 ^ΔHepΔMye compared to WT mice (Figures 1A, B). Since HMGB1 signals through RAGE and TLR4, and both receptors are implicated in chronic liver disease,²⁹,³⁰,³¹,³² we evaluated their expression by qPCR with the primers listed in Supplemental Table 1, http://links.lww.com/HC9/B52 and found it similar in all ethanol-fed mice (Supplemental Figure S1, http://links.lww.com/HC9/B52 and not shown, respectively). The HMGB1 immunohistochemistry (IHC) and morphometry analysis proved the specificity of the targeting strategy used to ablate Hmgb1 in hepatocytes, myeloid cells, or both in these mice (Figures 1C, first row and D, first column). The remaining HMGB1 staining observed in Hmgb1 ^ΔHepΔMye mice was from liver sinusoidal endothelial cells, lymphoid cells, and stellate cells. The serum HMGB1 decreased more in Hmgb1 ^ΔHep and Hmgb1 ^ΔHepΔMye than in Hmgb1 ^ΔMye mice, suggesting that hepatocytes are the main cellular source (Supplemental Figure S2, left column, http://links.lww.com/HC9/B52).

FIGURE 1.

Hmgb1 ablation in hepatocytes or myeloid cells partially protects, while ablation in both prevents steatosis, inflammation, IL1B production, and alcohol-induced liver injury. WT, Hmgb1 ^ΔHep, Hmgb1 ^ΔMye, and Hmgb1 ^ΔHepΔMye mice were fed the LDC diets for 6 weeks. H&E staining showing steatosis (black arrows) and inflammation (yellow arrows) (A). The liver-to-body weight ratio, serum ALT activity (U/L), liver TG (µg/mg), and the histopathological scores (steatosis, hepatocyte ballooning degeneration, inflammation) (B). IHC for HMGB1, where the orange arrows indicate ablation of HMGB1 in either hepatocytes, myeloid cells, or both (insets: ×630). IHC shows NASDCA⁺ (green arrows), F4/80⁺ (blue arrows), or IL1B⁺ (pink arrows) cells (C). Quantitative HMGB1 morphometry analysis (left). NASDCA, F4/80, and IL1B indexes (number of positive cells in 10 fields at ×200, right) (D). Quantification of liver IL1B (pg/mg) by ELISA (E). Heatmaps of IL1R and NFκB pathways (F). Results are expressed as mean ± SEM. n = 8/group. *p < 0.05, **p < 0.01, and ***p < 0.001 for Hmgb1 ^ΔHep, Hmgb1 ^ΔMye, and Hmgb1 ^ΔHepΔMye versus WT; ^•p < 0.05 and ^••p < 0.01 for ethanol-fed versus control diet-fed. Abbreviations: H&E, hematoxylin and eosin; HMGB1, high-mobility group box-1; Hmgb1 ^ΔHep, conditional Hmgb1 knockout mice in hepatocytes; Hmgb1 ^ΔMye, conditional Hmgb1 knockout mice in myeloid cells; Hmgb1 ^ΔHepΔMye, conditional Hmgb1 knockout mice in hepatocytes and myeloid cells; IHC, immunohistochemistry; LDC, Lieber-DeCarli; NASDCA, naphthol AS-D chloroacetate; TG, triglyceride.

FIGURE 1.

(Continued)

To determine the effect of reducing HMGB1 in alcohol-induced inflammation, we performed an IHC analysis. There was a decrease in the number of cells positive for naphthol AS-D chloroacetate (NASDCA) that labels neutrophils, F4/80 that labels MFs and KCs, and IL1B in ethanol-fed Hmgb1 ^ΔHep and Hmgb1 ^ΔMye, which were lowest in Hmgb1 ^ΔHepΔMye compared to WT mice (Figures 1C, second to fourth rows and D, right). Notably, IL1B expression was limited to inflammatory cells and nearly absent in hepatocytes, confirming that IL1B production by immune cells plays a major role in AALD.³⁷ The serum and liver IL1B protein decreased in all ethanol-fed mice with Hmgb1 ablation compared to WT mice (Figure 1E and Supplemental Figure S2, right column, http://links.lww.com/HC9/B52). To identify potential signals altered by Hmgb1 ablation, we performed microarrays. The Biocarta leading-edge heat map showed downregulation of the IL1R pathway in all ethanol-fed mice, especially in Hmgb1 ^ΔHepΔMye mice, likely due to lower NFκB signaling upstream of the Il1b gene, as well as in IL1B production (Figure 1F). Therefore, these findings suggest that Hmgb1 ablation in hepatocytes or myeloid cells partially protects while ablation in both almost prevents the ethanol-mediated increase in steatosis, inflammation, IL1B production, and liver injury; yet, if specific HMGB1 PTMs are involved in these effects, the receptor they interact with, in which cell type, and the downstream signals they convey remained unknown.

Neutralization of HMGB1 prevents, whereas injection of [H] or [O] HMGB1 worsens AALD

We previously detected [H], [O], and [Ac] HMGB1 in liver and serum from patients and mice with AALD, whose key residues are color-coded as depicted in Figure 2A, and identified that hepatocytes were their major source.¹¹ To dissect if blocking or increasing HMGB1 and whether any of these isoforms conditioned the response to alcohol, WT mice were fed the ethanol LDC diet and injected either a pan-HMGB1 neutralizing Ab, [H] HMGB1, or [O] HMGB1 throughout the ethanol feeding regimen.³⁸ Since [Ac] HMGB1 is not commercially available, it could not be injected; nevertheless, an alternative strategy to determine its contribution to AALD was used as described below.

FIGURE 2.

Neutralization of HMGB1 prevents, whereas injection of [H] or [O] HMGB1 worsens AALD. Schematic representation of the HMGB1 protein structure where cysteines (C) can undergo oxidation are written in yellow, and lysines (K) that can undergo acetylation are written in blue. These residues were mutated in the constructs to engineer the AAV8 vectors or in the conditional knockin mice (A). WT mice were injected with a pan-HMGB1 neutralizing Ab, [H] HMGB1, or [O] HMGB1 throughout the ethanol feeding regimen of 6 weeks. IHC shows NASDCA⁺ (green arrows), F4/80⁺ (blue arrows), or IL1β⁺ (pink arrows) cells in these mice (B). NASDCA, F4/80, and IL1β indexes (number of positive cells in 10 fields at ×200) (left) and concentration of HMGB1 in serum (ng/dL) (right) (C). Results are expressed as mean ± SEM. n = 8/group. *p < 0.05, **p < 0.01 and ***p < 0.001 for cotreated versus ethanol + BSA. Abbreviations: AAV8, adeno-associated viruses serotype-8; BSA, bovine serum albumin; HMGB1, high-mobility group box-1; [H] HMGB1, native or fully-reduced HMGB1; [O] HMGB1, disulfide HMGB1; NASDCA, naphthol AS-D chloroacetate; WT, wild-type.

Liver injury assessed by H&E staining, the liver-to-body weight ratio, serum ALT activity, liver TG and the histopathological scores was significantly reduced in ethanol-fed WT mice injected with the HMGB1 neutralizing Ab while these parameters were increased in mice injected with [H] HMGB1; yet, far more when injected with [O] HMGB1 (all shown in Ge et al³⁸), suggesting a major role for [O] HMGB1 in the pathogenesis of AALD. We then evaluated their effect on inflammation. IHC analysis revealed an increase in NADSCA ⁺ , F4/80 ⁺ , and IL1B ⁺ cells in WT mice injected with [H] or [O] HMGB1, whereas HMGB1 neutralization significantly reduced inflammation (Figures 2B, C, left). As anticipated, serum HMGB1 protein levels decreased by neutralizing it but increased by injecting the isoforms (Figure 2C, fourth column), especially by [O] HMGB1 as it enhanced liver injury and thus its secretion. Consequently, the HMGB1 isoforms and specifically [O] HMGB1 could be a potential target to prevent alcohol-induced liver injury; yet the specific receptor they interact with and the signals conveyed to upregulate IL1B production in AALD remain obscure.

[O] HMGB1 induces liver injury, whereas [Ac] HMGB1 protects from AALD

To further dissect the role of each isoform in AALD, Hmgb1 ^ΔHepΔMye and Hmgb1 ^ΔHep mice were transduced with AAV8 vectors designed to overexpress WT HMGB1, Δ[O] HMGB1, Δ[Ac] HMGB1, and Δ[O+Ac] HMGB1. Two weeks later, mice were fed with the ethanol LDC diet. Liver injury, assessed by H&E staining, the liver-to-body weight ratio, serum ALT activity, liver TG, and the histopathological scores, was significantly reduced in ethanol-fed Hmgb1 ^ΔHepΔMye mice overexpressing Δ[O] and Δ[O + Ac] HMGB1 as opposed to WT and Δ[Ac] HMGB1 (Figures 3A, B), highlighting the dominant role of [O] HMGB1 inducing liver injury. Notably, mice overexpressing Δ[Ac] HMGB1 exhibited a much worse phenotype with major lobular and centrilobular inflammation, microvesicular and macrovesicular steatosis, Mallory-Denk bodies, and hepatocyte ballooning degeneration, a predominant form of cellular injury in AH. A remarkable finding was that mice unable to oxidize and acetylate HMGB1 were significantly protected from liver injury compared to mice unable to acetylate the protein, suggesting that [Ac] HMGB1 may offset the noxious effects of [O] HMGB1. Since Hmgb1 ^ΔHep showed similar injury to Hmgb1 ^ΔHepΔMye mice transduced with the same AAV8 vectors, this suggested that hepatocyte-derived HMGB1 is more important than myeloid cell–derived HMGB1 for AALD (Figures 3A, B).

FIGURE 3.

[O] HMGB1 induces liver injury with higher inflammation, whereas [Ac] HMGB1 protects from AALD. Hmgb1 ^ΔHepΔMye and Hmgb1 ^ΔHep mice were transduced with AAV8 vectors containing the cDNA encoding for WT or for mutations of HMGB1 to prevent oxidation (Δ[O] HMGB1), acetylation (Δ[Ac] HMGB1), or both (Δ[O + Ac] HMGB1) of the protein. Two weeks later, mice were fed with the ethanol LDC diet for 6 weeks. H&E staining shows steatosis (black arrows), hepatocyte ballooning degeneration (open black arrow), and inflammation (yellow arrows) (A). The liver-to-body weight ratio, serum ALT activity (U/L), liver TG (µg/mg), and the histopathological scores (steatosis, hepatocyte ballooning degeneration, inflammation) (B). IHC shows HMGB1 expression (orange arrows), NASDCA⁺ (green arrows), F4/80⁺ (blue arrows), or IL1B⁺ (pink arrows) cells in these mice (C, D). Quantitative HMGB1 morphometric analysis (left). NASDCA, F4/80, and IL1β indexes (number of positive cells in 10 fields at ×200, right) (E). Results are expressed as mean ± SEM. n = 8/group. *p < 0.05, **p < 0.01, and ***p < 0.001 for Δ[O], Δ[Ac], or Δ[O + Ac] HMGB1 versus WT HMGB1; ^•p < 0.05 for Hmgb1 ^ΔHep versus Hmgb1 ^ΔHepΔMye. Abbreviations: AALD, alcohol-associated liver disease; AAV8, adeno-associated viruses serotype-8; H&E, hematoxylin and eosin; HMGB1, high-mobility group box-1; [Ac] HMGB1, acetylated HMGB1; [O] HMGB1, disulfide HMGB1; Hmgb1 ^ΔHep, conditional Hmgb1 knockout mice in hepatocytes; Hmgb1 ^ΔMye, conditional Hmgb1 knockout mice in myeloid cells; Hmgb1 ^ΔHepΔMye, conditional Hmgb1 knockout mice in hepatocytes and myeloid cells; IHC, immunohistochemistry; NASDCA, naphthol AS-D chloroacetate; TG, triglyceride.

FIGURE 3.

(Continued)

HMGB1 expression and localization were confirmed by GFP fluorescence and HMGB1 IHC, along with quantitative morphometry analysis (Supplemental Figures S3A, B, http://links.lww.com/HC9/B52, Figures 3C, D, first row and Figure 3E, middle column). We then evaluated the effect of the HMGB1 isoforms on inflammation. IHC analysis revealed fewer lobular and pericentral NASDCA ⁺ , F4/80 ⁺ , and IL1B ⁺ cells in ethanol-fed mice unable to oxidize HMGB1 (Δ[O] HMGB1 and Δ[O + Ac] HMGB1) compared to mice overexpressing WT HMGB1; however, the number of positive cells was significantly higher in mice unable to acetylate HMGB1 (Figures 3C, D, second to fourth rows and E, right). Notably, the concentration of serum HMGB1 protein decreased exclusively in mice unable to oxidize HMGB1 (Figure 3E, second column) due to reduced cell injury and inflammation, and consequently, less HMGB1 release and oxidation.

To further prove the role of the HMGB1 isoforms in AALD, we generated [WT] Hmgb1 ^{KI Hep}, Δ[O] Hmgb1 ^{KI Hep}, Δ[Ac] Hmgb1 ^{KI Hep}, and Δ[O+Ac] Hmgb1 ^{KI Hep} mice and characterized their phenotype under a chow diet. Based on the histopathological sores, at 6 months of age, [WT] Hmgb1 ^{KI Hep} and Δ[Ac] Hmgb1 ^{KI Hep} mice showed spontaneous liver injury, which was minimal in Δ[O] Hmgb1 ^{KI Hep} and Δ[O+Ac] Hmgb1 ^{KI Hep} mice (Supplemental Figure S4A, http://links.lww.com/HC9/B52). To further characterize the phenotype of these mice, we performed flow cytometry analysis of the liver immune cells. The percentage of neutrophils (NFs), B, CD4 ⁺ T, CD8 ⁺ T, and natural killer (NKs) cells decreased at 6 months of age in Δ[O] Hmgb1 ^{KI Hep} compared to control mice, [WT] Hmgb1 ^{KI Hep}, and Δ[Ac] Hmgb1 ^{KI Hep} mice, yet plasmacytoid and conventional dendritic cells (pDCs, cDCs) increased in Δ[O] Hmgb1 ^{KI Hep} mice (Supplemental Figure S4B, http://links.lww.com/HC9/B52). At 12 months of age, Δ[O] Hmgb1 ^{KI Hep} mice had minimal liver injury (not shown), less inflammatory cells in the liver (B, CD4 ⁺ T, NK) but more pDCs (Supplemental Figure S4C, http://links.lww.com/HC9/B52). We then fed these mice the control and ethanol LDC diets. Liver injury, assessed by H&E staining, serum ALT activity, liver TG, and the histopathological scores, was significantly reduced in ethanol-fed Δ[O] Hmgb1 ^{KI Hep} and Δ[O+Ac] Hmgb1 ^{KI Hep} mice, but it was increased in Δ[Ac] Hmgb1 ^{KI Hep} mice (Figures 4A, B). Overall, these data suggest that [O] HMGB1 induces liver injury, whereas [Ac] HMGB1 protects from AALD; yet, if the effects are receptor-mediated, and the identity of the receptor involved remains to be determined.

FIGURE 4.

[O] HMGB1 induces liver injury, whereas [Ac] HMGB1 protects from AALD. [WT] Hmgb1 ^{KI Hep} and mice with conditional overexpression of different HMGB1 isoforms (Δ[O] Hmgb1 ^{KI Hep}, Δ[Ac] Hmgb1 ^{KI Hep}, Δ[O + Ac] Hmgb1 ^{KI Hep}) were fed the ethanol LDC diet for 6 weeks. H&E staining shows steatosis (black arrows) and inflammation (yellow arrows) (A). The liver-to-body weight ratio, serum ALT activity (U/L), liver TG (µg/mg), and histopathological scores (steatosis, hepatocyte ballooning degeneration, and inflammation) are shown (B). Results are expressed as mean ± SEM. n = 8/group. *p < 0.05 for Δ[O], Δ[Ac], or Δ[O + Ac] HMGB1 versus [WT] HMGB1; ^•p<0.05 and ^••p<0.01 for ethanol versus control. Abbreviations: AALD, alcohol-associated liver disease; H&E, hematoxylin and eosin; HMGB1, high-mobility group box-1; [Ac] HMGB1, acetylated HMGB1; [O] HMGB1, disulfide HMGB1; [WT] Hmgb1 ^{KI Hep}, conditional WT Hmgb1 knockin mice in hepatocytes; Δ[Ac] Hmgb1 ^{KI Hep}, conditional Δ[Ac] Hmgb1 knockin mice in hepatocytes; Δ[O] Hmgb1 ^{KI Hep}, conditional Δ[O] Hmgb1 knockin mice in hepatocytes; Δ[O + Ac] Hmgb1 ^{KI Hep}, conditional Δ[O + Ac] Hmgb1 knockin mice in hepatocytes; LDC, Lieber-DeCarli; TG, triglyceride.

[O] HMGB1 stimulates MF migration and activation

To dissect whether [H] or [O] HMGB1 chemoattract and activate MFs, RAW 264.7 MFs were plated on cell culture inserts and treated with 0–1 nM [H] or [O] HMGB1 for 24 hours in the presence or absence of mitomycin to inhibit mitosis.³² [O] HMGB1, but not [H] HMGB1, increased cell migration, which was unaffected by cotreatment with mitomycin but was abolished when [O] HMGB1 was heat-denatured (Figure 5A). Under the same conditions, [O] HMGB1, but not [H] HMGB1 or bovine serum albumin, stimulated IL1B production and secretion (Figure 5B).

FIGURE 5.

[O] HMGB1 stimulates MF migration and activation. RAW 264.7 cells were plated on cell culture inserts, each HMGB1 isoform or BSA (control) was added at a concentration of 1 nM, and cells were cultured for 24 hours. Hematoxylin staining of cells that migrated to the bottom side of the cell culture insert membrane (A). Western blot analysis of intracellular and secreted IL1B under the same culture conditions as in panel (A, B). The experiment was repeated 3 times. ***p < 0.001 for [H] or [O] HMGB1 versus BSA. Abbreviations: BSA, bovine serum albumin; [H] HMGB1, native or fully-reduced HMGB1; [O] HMGB1, disulfide HMGB1; MF, macrophage.

Ethanol-fed Rage ^ΔMye but not Tlr4 ^ΔMye, Rage ^ΔHep, or Tlr4 ^ΔHep mice are protected from AALD

The binding of HMGB1 to RAGE and TLR4 has been implicated in chronic liver disease²⁹,³⁰,³²,³⁸; therefore, their expression was evaluated by western blot and qPCR. No significant changes were found in these or any HMGB1 receptor protein or mRNA expression in the livers of all ethanol-fed mice (not shown). We then asked if steatosis, inflammation, IL1B production, and liver injury were RAGE-dependent and/or TLR4-dependent and cell-specific in AALD. To answer this, WT, Rage ^ΔHep, Tlr4 ^ΔHep, Rage ^ΔMye, and Tlr4 ^ΔMye mice were generated and fed the ethanol LDC diet. Liver injury assessed by H&E staining, the liver-to-body weight ratio, serum ALT activity, liver TG as well as the histopathological scores, was reduced in ethanol-fed Rage ^ΔMye but not in Rage ^ΔHep, Tlr4 ^ΔHep, and Tlr4 ^ΔMye mice compared to WT mice (Figures 6A, B), suggesting a major role for RAGE signaling in myeloid cells in driving liver injury in AALD. Notably, there were fewer NADSCA ⁺ , F4/80 ⁺ , and IL1B ⁺ cells in Rage ^ΔMye compared to WT but not in the other groups of mice (Figures 6C, second to fourth rows and D, right). We also observed a reduction in serum HMGB1 in Rage ^ΔHep and Rage ^ΔMye mice (Figure 6D, second column), likely due to decreased liver injury and, thus, less secretion of HMGB1.

FIGURE 6.

Ethanol-fed Rage ^ΔMye but not Tlr4 ^ΔMye, Rage ^ΔHep, or Tlr4 ^ΔHep mice are protected from AALD. WT, Rage ^ΔHep, Tlr4 ^ΔHep, Rage ^ΔMye, and Tlr4 ^ΔMye mice were fed the ethanol LDC diet for 6 weeks. H&E staining shows steatosis (black arrows) and inflammation (yellow arrows) (A). The liver-to-body weight ratio, serum ALT activity (U/L), liver TG (µg/mg), and the histopathological scores (steatosis, hepatocyte ballooning degeneration, and inflammation) (B). IHC shows HMGB1 expression (orange arrows), NASDCA⁺ (green arrows), F4/80⁺ (blue arrows), or IL1B⁺ (pink arrows) cells in these mice (C). Quantitative HMGB1 morphometry analysis and concentration of HMGB1 in serum (ng/dL) (left). NASDCA, F4/80, and IL1β indexes (number of positive cells in 10 fields at ×200, right) (D). In the HMGB1 morphometry, the quantification was expressed as fold-change of the WT, which was assigned a value of 1; the rest of the results are given as mean ± SEM. n = 8/group. *p < 0.05, **p < 0.01, and ***p < 0.001 for ethanol-fed Rage ^ΔHep, Tlr4 ^ΔHep, Rage ^ΔMye, and Tlr4 ^ΔMye versus WT. Abbreviations: AALD, alcohol-associated liver disease; H&E, hematoxylin and eosin; HMGB1, high-mobility group box-1; IHC, immunohistochemistry; LDC, Lieber-DeCarli; NASDCA, naphthol AS-D chloroacetate; Rage ^ΔHep, conditional Rage knockout mice in hepatocytes; Rage ^ΔMye, conditional Rage knockout mice in myeloid cells; WT, wild-type; Tlr4 ^ΔHep, conditional Tlr4 knockout mice in hepatocytes; Tlr4 ^ΔMye, conditional Tlr4 knockout mice in myeloid cells; TG, triglyceride.

Hepatocyte-derived [O] HMGB1 signals through RAGE in myeloid cells to drive AALD

Since [O] HMGB1 and RAGE in myeloid cells appeared critical in AALD, we transduced Hmgb1&Rage ^ΔHepΔMye mice with each AAV8 vector to further investigate if [O] HMGB1 signaled through RAGE. Liver injury, assessed by H&E staining, the liver-to-body weight ratio, serum ALT activity, liver TG, and the histopathological scores, was reduced in all ethanol-fed Hmgb1&Rage ^ΔHepΔMye mice transduced with the AAV8 vectors targeting the HMGB1 PTMs compared to the Hmgb1 ^ΔHepΔMye mice shown in Figure 3 (Figures 7A, B). We previously showed that Rage ^ΔMye mice injected with [H] or [O] HMGB1 and fed the ethanol LDC diet were protected from AALD compared to WT mice injected with the same HMGB1 isoforms (Figure 3).³⁸ Overall, these results suggest a significant role for hepatocyte-derived [O] HMGB1 signaling through RAGE in myeloid cells in driving steatosis, inflammation, IL1B production, and liver injury in AALD. They also prove that [Ac] HMGB1 can protect from AALD by counteracting the noxious effects from [O] HMGB1.

FIGURE 7.

Hepatocyte-derived [O] HMGB1 signals through RAGE in myeloid cells to drive AALD. Hmgb1&Rage ^ΔHepΔMye mice were transduced with AAV8 vectors containing the cDNA encoding for WT or mutations of HMGB1 to prevent oxidation (Δ[O] HMGB1), acetylation (Δ[Ac] HMGB1), or both (Δ[O+Ac] HMGB1) of the protein. Two weeks later, mice were fed the ethanol LDC diet for 6 weeks. H&E staining showing steatosis (black arrows) and inflammation (yellow arrows) (A). The liver-to-body weight ratio, serum ALT activity (U/L), liver TG (µg/mg), and histopathological scores (steatosis, hepatocyte ballooning degeneration, inflammation) are shown (B), n = 16/group. Abbreviations: AALD, alcohol-associated liver disease; AAV8, adeno-associated virus serotype-8; H&E, hematoxylin and eosin; HMGB1, high-mobility group box-1; Δ[Ac] HMGB1, acetylation-incompetent HMGB1 mutant; Δ[O] HMGB1, oxidation-incompetent HMGB1 mutant; Hmgb1&Rage ^ΔHepΔMye, conditional Hmgb1 and Rage knockout mice in hepatocytes and myeloid cells; [O] HMGB1, disulfide HMGB1; LDC, Lieber-DeCarli; RAGE, receptor for advanced-glycation end-products; WT, wild-type; TG, triglyceride.

DISCUSSION

Previous work from our laboratory demonstrated a robust increase in HMGB1 expression in human liver biopsies and experimental AALD, correlating with disease stage.¹¹ In the present study, we aimed to dissect the potential role of these isoforms resulting from PTMs of HMGB1 in the pathogenesis of AALD.

We showed that Hmgb1 ablation in hepatocytes or myeloid cells partially protects, while ablation in both significantly prevents steatosis, inflammation, IL1B production, and alcohol-induced liver injury. In addition, we found that hepatocytes were the primary source of [H], [O], and [Ac] HMGB1, while myeloid cells produced [H] and [Ac] HMGB1.¹¹,³⁸ Since the ethanol LDC model induces early AALD, with oxidative stress increasing over time due to alcohol metabolism in the liver, mitochondrial injury, and changes in the antioxidant defense, the disease progresses, and [H] HMGB1 undergoes oxidation intracellularly and extracellularly. This could drive immune cell infiltration further enhancing or prolonging liver damage, but further investigation was required.

Next, we focused on the role of each HMGB1 isoform in AALD. First, we demonstrated that HMGB1 neutralization prevented, whereas injection of [H] HMGB1 enhanced alcohol-induced liver injury, which was remarkably worsened by injecting [O] HMGB1. Second, using AAV8 vectors, we overexpressed mutants of the HMGB1 isoforms in hepatocytes and demonstrated that [O] HMGB1 is the most critical isoform driving steatosis, inflammation, and IL1B production in AALD, while the lack of [Ac] HMGB1 exacerbated liver injury. Third, using mice with conditional ablation of each HMGB1 isoform in hepatocytes, we showed similar results to those in mice injected with the AVV8 vectors. Importantly, we demonstrated that [Ac] HMGB1 counteracts the noxious effects of [O] HMGB1; however, whether this is due simply to acetylation of HMGB1 or acetylation of [O] HMGB1 remains inconclusive and is a limitation of this study.

Importantly, we identified that RAGE signaling in myeloid cells played a significant role in the development of AALD as ethanol-fed Rage ^ΔMye but not Tlr4 ^ΔMye, Rage ^ΔHep, or Tlr4 ^ΔHep mice were protected from alcohol-induced liver injury. RAGE shares common ligands and signaling pathways in the host immune response.¹²,²⁹ Thus, it is conceivable that if HMGB1 is a key ligand for this receptor in AALD, its PTMs could affect receptor binding along with associated downstream events.³⁹,⁴⁰,⁴¹,⁴²,⁴³,⁴⁴ Virtually nothing was known about whether [H], [O], or [Ac] HMGB1 regulated IL1B production in AALD by engaging RAGE; consequently, additional experiments to validate this were performed.

While finding that Tlr4 ^ΔMye mice were not protected from AALD may be a paradigm shifting in the field of AALD, this still opened the possibility that other receptors, such as RAGE, could play a more dominant role. We previously established that [O] HMGB1 forms a protein complex with IL1B that binds with RAGE in MFs, activates them, and enhances the production of proinflammatory mediators,³⁸ indicating a proinflammatory feedback loop.

In this study, we demonstrated that hepatocyte-derived [O] HMGB1 signaled through RAGE in myeloid cells to trigger alcohol-induced liver injury. The effect was myeloid cell-specific, as Rage ^ΔHep mice were unprotected from AALD. A possible limitation of this study is that other ligands (i.e., free fatty acids, advanced-glycation end-products, S100 proteins, lysophosphatidic acid, amyloid-β-peptide), if present, may compete with [O] HMGB1 for binding to RAGE. Similarly, soluble RAGE may act as a decoy because it binds to and sequesters RAGE, thereby competing with [O] HMGB1 for receptor binding and signal transduction.

The discovery that [O] HMGB1 stimulates IL1B production is clinically relevant for AALD. In addition to promoting the recruitment of inflammatory cells to the liver,⁴⁵ increasing TG accumulation in hepatocytes,⁴⁶ and triggering hepatocyte dysfunction, necrosis, and apoptosis,⁴⁷ IL1B increases gut permeability, exacerbating alcohol-induced gut leakiness and liver damage.⁴⁸ Moreover, IL1B stimulates the production of proinflammatory mediators, magnifying the inflammatory response. Since it is clearly linked to key clinical symptoms of AALD, such as fever, inflammation, and muscle waste, averting this cytokine early might be an attractive treatment strategy. The close link between the alcohol-driven IL1B production and injury strongly supports efforts to target this cytokine for AALD intervention; yet, new upstream anti-inflammatory therapies are urgently needed, and as shown here, [O] HMGB1 may be one of them. Although one possibility is that the reduced intrahepatic IL1B in our model may result from decreased infiltration of inflammatory cells, the intensity of IL1B staining was much lower in ethanol-fed mice unable to oxidize HMGB1 and in Rage ^ΔMye mice. Another possibility is that the downregulation of NFκB restricted inflammasome activation by eliminating damaged mitochondria, both central players in AALD.⁴⁹

Overall, how does this project challenge and shift existing paradigms? While great emphasis has been placed on the effects of nonsterile inflammation triggered by pathogen-associated molecular patterns such as LPS, growing clinical and experimental evidence suggests that sterile inflammation significantly contributes to AALD.⁵⁰ Endogenous DAMPs released during cellular stress or injury have the potential to drive sterile inflammatory responses to a self-perpetuating noxious loop that results in chronic liver injury. The balance between [O] and [Ac] HMGB1 may regulate this process and alter RAGE signaling in immune cells. Thus, this study challenges our current knowledge and drives the field of AALD forward: first, by highlighting the role of [O] and [Ac] HMGB1 as major targetable upstream regulators of inflammation; second, by underscoring the critical role of RAGE in myeloid cells for inflammation and IL1B production; and third, by emphasizing that specific HMGB1 isoforms control IL1B production. Of note, HMGB1, unlike other proinflammatory cytokines, such as TNFA, provides a much wider time frame for clinical intervention against progressive inflammation due to its remarkable cellular mobility, prolonged half-life, and continued receptor activation.¹³

A potential challenge is how to integrate that mice lacking Hmgb1 in hepatocytes and myeloid cells were significantly protected from AALD while acetylation counteracted the noxious effects of [O] HMGB1. Since both hepatocytes and myeloid cells produce [Ac] HMGB1, the involvement of both cell types perhaps is key for this effect, as demonstrated by most of the protection from AALD found in Hmgb1 ^ΔHepΔMye compared to Hmgb1 ^ΔHep and Hmgb1 ^ΔMye mice. We posit that inflammatory cells are recruited to the liver to acetylate HMGB1 but that acetylation of HMGB1 perhaps limits their extravasation.

In conclusion, we propose that hepatocyte-derived [O] HMGB1 is a crucial mediator whose presence correlates with steatosis, increased immune cells, IL1B production, and alcohol-induced liver injury. These effects are mediated through RAGE signaling in myeloid cells. However, [Ac] HMGB1 may be protective by counteracting the harmful effects of [O] HMGB1 and inflammatory cell infiltration. Thus, developing therapeutics such as antibodies or small molecules to block [O] HMGB1 or RAGE signaling will certainly ameliorate AALD, benefiting many patients worldwide.

Supplementary Material

hc9-8-e0549-s001.docx ^{(1.5MB, docx)}

FUNDING INFORMATION

This work is supported by the US Public Health Service Grant R01AA025907 from the National Institute on Alcohol Abuse and Alcoholism (Natalia Nieto) and the US Veterans Administration Grant I01BX005093 from the Biomedical Laboratory Research & Development (Natalia Nieto). The views expressed in this article are those of the authors and do not necessarily reflect the position or policy of the Department of Veterans Affairs or the US government.

CONFLICTS OF INTEREST

Yujin Hoshida advises and owns stock in Alentis Therapeutics and Espervita Therapeutics. He consults for Roche and advises Helio Genomics and Elevar Therapeutics. The remaining authors have no conflicts to report.