In order to induce chronic liver disease, mice underwent bile duct ligation surgery for 28 days. In a subgroup of mice, a single 5 g/kg alcohol binge was given to induce ACLF (Figure 1A). ACLF mice showed increased ALT levels in the circulation (Figure 1B), altered liver histology showing ductular reaction in portal triads (arrowheads) and necrotic areas (dashed lines) (Figure 1C), decreased hepatic albumin mRNA expression (Figure 1D), increased expression of fibrosis markers (Figure 1E), and increased expression of S100 calcium-binding protein A8 (S100a8/9 heterodimer), also known as calprotectin (Figure 1F), in the liver. ACLF mice also showed increased neutrophil count in the blood (Figure 1G), increased circulating levels of calprotectin (Figure 1H), and pro-inflammatory cytokines including interleukin (IL) 6 and IL-18 (Figure 1I) indicating systemic inflammation.

G-CSF was administered to a subgroup of ACLF mice by intraperitoneal injection on day 25 and 27 after BDL (Figure 2A). G-CSF treatment increased gene expression of regenerative markers in the liver including hepatocyte growth factor (Hgf) and nanog homeobox (Nanog) (Figure 2B), and expression of proliferation Ki-67 assessed by immunohistochemistry staining in the liver (Figure 2C). Moreover, flow cytometry analysis of liver-infiltrated cells revealed an increase in neutrophils in the liver without significant changes in the monocyte/macrophage populations (Figure 2D). At the mRNA level, hepatic expression of the pan-leukocyte marker, Cd11b, and the neutrophil marker, Ly6g, were significantly increased after G-CSF treatment without changes in the pro-inflammatory/-resolutive macrophage markers, Cd68 and Cd206 (Figure 2E). This suggested that G-CSF treatment in our ACLF model increases liver regeneration and neutrophil recruitment in the liver.

In-depth characterization of neutrophil phenotype after G-CSF treatment revealed an increased expression of the chemokine receptor CXC motif chemokine receptor (CXCR) 2 and decreased expression of the “do not eat me” signal CD47 in neutrophils. Moreover, there was no significant change in the neutrophil expression of the adhesion molecules, CD54 or CEA, cell adhesion molecule (CEACAM) 1 after the G-CSF treatment (Figure 3A). In addition, hepatic expression of the neutrophil activation marker, leukotriene 4 receptor (Ltb4r1), was increased after G-CSF treatment (Figure 3B). Expression of calprotectin, both at the mRNA (Figure 3C) and protein (Figure 3D) level, was greatly and significantly increased in the liver of ACLF mice that received G-CSF compared to the vehicle-treated group. This data suggests that upon G-CSF treatment, activated neutrophils are recruited to the injured liver in ACLF.

Calprotectin is known to account for around 50% of the cytosolic proteins in neutrophils, serving as an activation and NETosis marker for neutrophils (Pruenster et al., 2016). Calprotectin has been shown to induce cytokine release and inflammasome activation via ROS production in autoimmune inflammatory diseases (Simard et al., 2013), myocardial infarction (Sreejit et al., 2020), and acute kidney injury (Tan et al., 2017); nevertheless their role in ACLF has been unknown to date.

Given the observation that G-CSF induces an increase in the expression of calprotectin, we next explored liver damage pathways in our ACLF model. We observed an activation of oxidative stress pathways as shown by increased mRNA expression of the NADPH oxidase subunits p40phox, p47phox and gp91phox (Figure 4A) and genes involved in type I interferon response including interferon beta (Ifnβ) and interferon-stimulated gene 15 (Isg15) in ACLF mice that received G-CSF compared to saline treated ACLF controls (Figure 4B). Moreover, liver fibrosis markers including Sirius Red staining (Figure 4C) and mRNA expression of collagen 1α1, 1α2 and 4α1 and the extracellular matrix remodeling enzymes Timp1 and Mmp9 were increased in G-CSF treated ACLF mice (Figure 4D). Additionally, inflammasome activation genes including Nlrp3, Asc, Il18 and Il1β were upregulated in the liver of ACLF mice treated with G-CSF (Figure 4E). Increased calprotectin and inflammasome activation in G-GSF treated ACLF mice was also indicated by significantly higher circulating levels of IL-1β (Figure 4F) and calprotectin (Figure 4G) in the serum of ACLF mice treated with G-CSF compared to vehicle.

Given that hepatic encephalopathy is a common extrahepatic organ dysfunction in ACLF, observation of increased systemic IL-1β prompted us to explore a potential effect of systemic inflammation in the brain of the ACLF mice. The cerebellum was analyzed because it is the most sensitive brain region to ammonia and inflammatory mediators, a known trigger of encephalopathy during ACLF (Parvez and Ohtsuki, 2022). Similarly to findings in the liver, cerebellum from ACLF mice treated with G-CSF exhibited increased neutrophil infiltration and activation compared to ACLF alone as shown by the significant increase in the expression of Ly6g (Figure 5A) and the calprotectin components S100a8 and S100a9 (Figure 5B) respectively. Moreover, G-CSF treatment resulted in increased p40phox (Figure 5C), Asc, Il18 and Il1β (Figure 5D) mRNA expression in the cerebellum, suggesting the induction of oxidative stress and inflammasome activation. No significant change was observed in the type I interferon response upon G-CSF treatment in the brain (Figure 5E).

This highlights the activation of common molecular and cellular pathways including neutrophil infiltration, calprotectin expression and activation of oxidative stress and inflammasome both in the liver and brain in G-CSF-treated ACLF mice.

We next assessed the response of glial and neuronal cells in the cerebellum to G-CSF treatment. Microglia, resident brain macrophages, become activated and proliferate following brain damage or stimulation (Mander et al., 2006). After ACLF, mice treated with G-CSF showed an increased expression of Aif1 and microglia-specific markers Hexb and Sall1, indicating microglia proliferation. We also observed increased expression of Lgals3 which plays a key role in microglia-mediated neuroinflammation (Ge et al., 2022). There was no change in Cd68 expression but we found an increase in Cd206 expression in the G-CSF-treated group (Figure 6A). Furthermore, expression of reactive astrocytes markers AldoC and Glt1, but not Gfap, were increased after G-CSF treatment (Figure 6B). Moreover, we observed increased expression of neuronal markers Neun and neurotrophic factor Bdnf. There was no significant change in synaptophysin (Syp) and myelin binding protein (Mbp) after G-CSF treatment when compared to vehicle-treated ACLF mice (Figure 6C). Moreover, the death receptor, Fas, and its downstream apoptotic markers, Bcl2 and Bax, were upregulated after G-CSF suggesting an induction of apoptosis in the cerebellum of G-CSF-treated ACLF mice (Figure 6D). No significant change was observed in expression of additional death markers such as Bak, Trailr or Tnfr1 (Figure 6D).

C57BL/6J mice were purchased from The Jackson Laboratory (Bar Harbor, ME). Mice were housed in a specific pathogen-free mouse facility at Beth Israel Deaconess Medical Center (BIDMC), and all animal handling was performed in compliance with institutional guidelines. Additional procedures were approved by the BIDMC Institutional Animal Care and Use Committee (protocol #010-2019 and #030-2022).

Bile duct ligation (BDL) surgery, was performed as previously described (Zhuang et al., 2023). Briefly, 10-14- week-old male C57BL/6J mice were anesthetized and placed on an operating pad. Mice were shaved and the skin was disinfected with 70% ethanol. Through an abdominal incision, the common bile duct was identified and ligated. The abdomen and peritoneum were closed with a running silk suture. For sham control mice, the same surgical procedure was performed without the bile duct ligation. Animals were monitored during recovery and treated with buprenorphine (0.1 mg/kg) to avoid pain-induced stress after surgical intervention.

Acute administration of ethanol (5 g/kg) by oral gavage was adapted from (Bertola et al., 2013). Briefly, a 47.3% (vol/vol) ethanol solution was prepared from pure alcohol (1000002000, Pharmco) in water. Volume administration to each mouse was calculated as follows: gavage volume (µL) of 47.3% (vol/vol) solution for each mouse = mouse body weight in grams × 15. Control group received water binge.

In order to develop an alcohol-induced ACLF model, advanced cholestatic liver fibrosis was induced by 28-day BDL surgery followed by a single alcohol binge to trigger acute liver damage (Figure 1A). BDL mice were randomly assigned to receive either alcohol or water as vehicle, those receiving water binge were used as chronic liver disease controls. Consistent with advanced liver disease, 10% of mice developed visible ascites during the last week of BDL and, according to IACUC guidelines, were euthanized and not considered in the study. Nine hours after the alcohol binge, animals were euthanized for sample collection. Specimens were snap frozen in liquid nitrogen and stored at −80°C for molecular analysis or processed in 10% formalin for histology.

ACLF mice were randomized to receive either 200 µL of saline or G‐CSF (Neupogen; 200 μg/kg body weight) through intraperitoneal injection. G‐CSF injection was performed on days 25 and 27 after BDL surgery.

Mice serum or tissue lysate levels of IL-6 (DY406, R&D Systems), IL-18 (DY7625, R&D Systems), S100a8/9 heterodimer (DY8596, R&D Systems) and IL-1β (MHSLB00, R&D Systems) were quantified by using commercially available kits according to the manufacturer’s instructions. ALT levels in serum were measured following manufacturer’s instructions (A525-240, Teco Diagnostics).

Whole blood was collected in Eppendorf tubes containing 4% sodium citrate and analyzed in Hemavet (M-950HV, Drew Scientific) for quantification of neutrophil population.

Liver histology and liver fibrosis were assessed on paraffin-embedded liver sections by Hematoxylin/Eosin and Sirius Red staining respectively. Ki67 was assessed by immunohistochemistry (IHC) on paraffin-embedded liver sections. Briefly, sections were deparaffinated and antigens were exposed by 30min steamer incubation in Tris-EDTA (pH = 9) buffer. After antigen retrieval procedure, endogenous peroxidase activity was inhibited for 10 min with 3% hydrogen peroxide. Sections were blocked for 1 h with normal goat serum and incubated with anti-Ki67 (1:300; 12202S, Cell Signaling Technology) primary antibody O/N at 4°C. HRP-Rabbit (7074S, Cell Signaling Technology) secondary antibody was added. Color development was induced by incubation with a DAB kit (8059S, Cell Signaling Technology) and the sections were counterstained with hematoxylin. Sections were dehydrated and mounted. Positive cell count was performed using QuPath software.

Total RNA was extracted and purified with RNeasy^® Mini Kit (74106, Qiagen) according to the manufacturer instructions. RNA yield was quantified by Nanodrop ND-1000 spectrophotometer (NanoDrop Technologies, Wilmington, DE, United States). A maximum amount of 1 µg RNA was reverse transcribed using a High-Capacity cDNA Reverse Transcription Kit (1708891BUN, Bio-Rad) in a cDNA Master cycler X50s (Eppendorf), and qPCR was performed in a CFX96 Real-Time PCR System (Bio-Rad), using Sybr Green primers and PCR Master Mix (1725124, Applied Biosystems). RNA expression levels were normalized following the 2^−ΔΔCT method, with gapdh as housekeeping genes. Detailed primer sequences used for this study can be found in Table 1.

Fifty μL of peripheral blood was collected in Eppendorf tubes containing 0.5M EDTA to prevent coagulation. Blood samples were incubated with Fc block for 10 m on ice (TruStain FcX™, Biolegend, Cat# 101320) followed by a cocktail of antibodies (1:100 dilution) for 30 min on ice; APC/Cyanine7 anti-mouse CD45 Antibody (Biolegend, 103116, clone 30-F11), Brilliant Violet^® 711 Anti-CD11b Rat Monoclonal Antibody (Biolegend, Cat#101242, clone: M1/70), Alexa Fluor^® 700Anti-Ly-6C Rat Monoclonal Antibody (Biolegend, Cat#128024, clone: HK1.4), PerCP/Cyanine5.5 anti-mouse CXCR2 Antibody (Biolegend, Cat#149307), PE anti-mouse CD47 Antibody (Biolegend, Cat#127507), Alexa Fluor^® 594 anti-mouse CEACAM1a Antibody (Biolegend, Cat#134522), Zombie NIR™ Fixable Viability Kit (Biolegend, Cat#423106), and APC anti-mouse CD54 Antibody (Biolegend, Cat#116119). After 30 min, all samples were incubated with BD Phosflow™ Lyse/Fix Buffer (BD Biosciences, 558049) following manufacturer’s protocol. Following lyse/Fix, cells were washed twice with 1× PBS containing 2% FBS, and resuspended in FACS buffer containing 2% FBS in PBS. Samples were run in Aurora Spectral Flow Cytometer (Cytek) and data analysis was done using Flowjo version 8.8.7 software.

Statistical analyses were performed using the Graphpad Prism 9 statistics software for Windows. Normality of sample distribution was assessed using the Kolmogorov-Smirnov test. For samples fitting a normal distribution, means were compared by the Student’s t-test (2 groups) or ANOVA (>2 groups) followed by the Tukey post hoc analysis. Otherwise, means were compared using the non-parametric Kruskal-Wallis test followed by the Mann-Whitney U test. Differences were considered significant at p < 0.05. Statistical analyses were performed using GraphPad Prism 8.0.2 software (GraphPad Software, Inc.; San Diego, CA, United States).