S100A8 Production in CXCR2-Expressing CD11b⁺Gr-1^high Cells Aggravates Hepatitis in Mice Fed a High-Fat and High-Cholesterol Diet Free

Nonalcoholic fatty liver disease (NAFLD) comprises a spectrum of liver diseases ranging from nonalcoholic fatty liver (NAFL) to nonalcoholic steatohepatitis (NASH) (1). This disease has become one of the most common chronic liver diseases worldwide. In the United States and Western Europe, as many as 20–30% of adults have NAFLD, and ∼10% of patients with NAFLD are diagnosed with NASH (2). NASH is a progressive form of NAFLD, and patients with NASH have a potential to develop liver cirrhosis and hepatocellular carcinoma, which are often indications for liver transplantation (3). Several factors, such as oxidative stress, mitochondrial dysfunction, endoplasmic stress, and apoptosis, have been reported to play significant roles in disease progression to NASH (4). However, the immunological mechanism that promotes progression of NAFLD remains poorly understood.

S100A8 belongs to the calcium-binding S100 protein family and is the most abundant cytoplasmic proteins in myeloid lineage cells (5). Upon secretion, S100A8 acts as an endogenous ligand for TLR4 (6), and it can play a crucial role in the pathogenesis of inflammatory diseases, such as rheumatoid arthritis, inflammatory bowel disease, and sepsis (5). Recent studies have reported that S100A8 is associated with the pathological conditions underlying metabolic syndrome, such as atherosclerosis (7), visceral fat adiposity (8), and insulin resistance (9). NAFLD is considered a hepatic manifestation of metabolic syndrome and also a chronic inflammatory disease. However, it remains unclear whether S100A8 is associated with the pathology of NAFLD.

In the current study, we investigated the involvement of S100A8 in the development of NAFLD. We used a high-fat and high-cholesterol diet (HFHCD) model of NAFLD (10, 11), in which mice were fed either HFHCD or a normal diet (ND) as a control. HFHCD-fed mice, but not ND-fed mice, displayed hepatitis with steatosis and liver fibrosis. S100A8 expression was remarkably elevated in the livers of HFHCD-fed mice compared with ND-fed mice. S100A8-expressing cells were mostly CD11b⁺Gr-1^high myeloid lineage cells, which were significantly increased in the livers of HFHCD-fed mice and, unlike myeloid-derived suppressor cells (MDSCs), did not suppress the proliferation of ConA-activated splenocytes in vitro. S100A8-positive cells also expressed CXCR2, a chemokine receptor on myeloid lineage cells. The expression of CXCL1, a ligand for CXCR2, was significantly elevated in the livers of HFHCD-fed mice compared with ND-fed mice. S100A8 was found to significantly upregulate CXCL1 expression in vitro. TNF-α expression was also remarkably elevated in the livers of HFHCD-fed mice compared with ND-fed mice, and the deficiency of TNF-α ameliorated the liver inflammation in HFHCD-fed mice. S100A8 significantly upregulated TNF-α expression in vitro. The production of TNF-α in the liver leukocyte population in response to S100A8 stimulation mostly occurred in CD11b⁺F4/80⁺ cells, which were significantly increased in the livers of HFHCD-fed mice compared with ND-fed mice. S100A8 upregulated M1 gene expression and downregulated M2 gene expression in bone marrow–derived CD11b⁺F4/80⁺ macrophages in vitro. Furthermore, in liver biopsy samples from patients with NAFLD, significantly higher S100A8 expression was observed in patients with NASH than in patients with NAFL. Collectively, these results suggest that S100A8 plays an important role in the development of NAFLD.

Specific pathogen-free C57BL/6 male mice were purchased from Charles River Japan (Kanagawa, Japan). TNF-α^−/− male mice on a C57BL/6 background were purchased from The Jackson Laboratory (Bar Harbor, ME). These mice were maintained in isolation facilities at the Institute of Experimental Animal Science, Osaka University, and received humane care. All experimental animal protocols were approved by the Institutional Animal Care and Use Committee of Osaka University Graduate School of Medicine (approval number J004913-011). All possible efforts were made to minimize animal suffering. The animals were maintained in a temperature-controlled, specific pathogen-free room on 12-h light and dark cycles with ad libitum access to water and diet as indicated.

At 6 wk of age, the mice used in the experiments were fed either an HFHCD consisting of 38.5% of the calories from fat, which contain 15% cocoa butter fat, 1.25% cholesterol, and 0.5% cholic acid (Oriental Yeast, Suita, Osaka, Japan) (11), or ND consisting of 13% of the calories from fat (MF-1; Oriental Yeast) for 3 wk. Mouse sample preparations were performed as previously described (12). Serum alanine aminotransferase (ALT) activities were measured as previously described (12, 13).

For histological analyses, removed livers from mice were fixed in 10% formalin, embedded in paraffin, and sectioned. The liver sections were stained with H&E to evaluate the level of steatohepatitis. The liver sections were also stained with Sirius Red to assess liver fibrosis, which was quantified based on the size of the positive area using image-analysis software WinROOF (Mitani Corporation, Fukui, Japan), as previously described (12).

Liver leukocyte and splenocyte populations were prepared as previously described (14).

Prepared liver leukocytes were suspended in PBS/0.3% w/v BSA supplemented with 0.1% w/v sodium azide. To avoid nonspecific binding of Abs to FcγR, the cells were preincubated with an anti-mouse CD16/32 mAb (2.4G2; BD Biosciences, Franklin Lakes, NJ). To analyze intracellular S100A8 expression, the cells were subsequently stained with a saturating amount (1 μg/10⁶ cells) of biotin-conjugated anti-mouse CD11b (M1/70; BD Biosciences), Pacific Blue–conjugated anti-mouse Gr-1 (RB6-8C5; BD Biosciences), and allophycocyanin-conjugated anti-mouse CXCR2 (clone 242216; R&D Systems, Minneapolis, MN) mAbs for 15 min at 4°C. After cell-surface marker staining, the cells were fixed and permeabilized with Cytofix/Cytoperm solution (BD Biosciences) for 20 min at 4°C, followed by washing with Perm Wash Buffer (BD Biosciences). The cells were subsequently stained with a rabbit anti-mouse/human S100A8 mAb (Abcam, Cambridge, MA) for 20 min at room temperature, followed by staining with a secondary Alexa Fluor 488–conjugated goat anti-rabbit IgG Ab (Cell Signaling Technology, Danvers, MA), a PE-conjugated anti-mouse F4/80 mAb (clone 521204; R&D Systems), and PerCP streptavidin (SAv-PerCP) (BD Biosciences) for 20 min at room temperature. The corresponding isotypes were used as controls. To analyze the expression of lymphoid cell-surface markers or myeloid cell surface markers on CD11b⁺Gr-1^high cells, the cells preincubated with an anti-mouse CD16/32 mAb (2.4G2; BD Biosciences) were subsequently stained for 15 min at 4°C with a saturating amount (1 μg/10⁶ cells) of the following combinations of Abs and a tetramer: 1) FITC-conjugated anti-mouse CD11b (M1/70; BD Biosciences), Pacific Blue–conjugated anti-mouse Gr-1 (RB6-8C5; BD Biosciences), biotin-conjugated anti-mouse TCRβ (H57-597; BD Biosciences) mAbs, and allophycocyanin-conjugated mouse CD1d tetramer preloaded with α-galactosylceramide (Proimmune, Oxford, U.K.), followed by staining with SAv-PerCP (BD Biosciences) for 15 min at 4°C; 2) biotin-conjugated anti-mouse CD11b (M1/70; BD Biosciences), PE-conjugated anti-mouse Gr-1 (RB6-8C5; BD Biosciences), and allophycocyanin-conjugated anti-mouse NK 1.1 (PK136; BD Biosciences) mAbs, followed by staining with SAv-PerCP (BD Biosciences) for 15 min at 4°C; 3) allophycocyanin-conjugated anti-mouse CD11b (M1/70; BD Biosciences), Pacific Blue–conjugated anti-mouse Gr-1 (RB6-8C5; BD Biosciences), FITC-conjugated anti-mouse Ly6C (AL-21; BD Biosciences), PE-conjugated anti-mouse CD115 (clone 460615; R&D Systems), and PerCP-conjugated anti-mouse Ly6G (1A8; BD Biosciences) mAbs; 4) allophycocyanin-conjugated anti-mouse CD11b (M1/70; BD Biosciences), Pacific Blue–conjugated anti-mouse Gr-1 (RB6-8C5; BD Biosciences), FITC-conjugated anti-mouse CD11c (HL3; BD Biosciences), and PE-conjugated anti-mouse MHC class II (M5/114.5.2, BD Biosciences) monoclonal Abs; 5) allophycocyanin-conjugated anti-mouse CD11b (M1/70; BD Biosciences), Pacific Blue–conjugated anti-mouse Gr-1 (RB6-8C5; BD Biosciences), PerCP-conjugated anti-mouse CD45R/B220 (RA3-6B2; BD Biosciences) mAbs, and PE-conjugated anti-mouse CD31 (R&D Systems) polyclonal Ab; and 6) allophycocyanin-conjugated anti-mouse CD11b (M1/70; BD Biosciences), biotin-conjugated anti-mouse Gr-1 (RB6-8C5; BD Biosciences), and FITC-conjugated anti-mouse F4/80 (BM8; eBioscience) mAbs, followed by staining with SAv-PerCP (BD Biosciences) for 15 min at 4°C. The corresponding isotypes were used as controls.

Prepared liver leukocytes from normal mice were incubated at 1 × 10⁶ cells/ml with S100A8 (Abcam) at 5 μg/ml for 5 h together with GolgiStop (BD Biosciences) at 1 μl/ml for the last 4 h in 2 ml of RPMI 1640 containing 10% FCS and antibiotics at 37°C in six-well flat-bottom plates. Prepared liver leukocytes from ND-fed mice or HFHCD-fed mice were also incubated at 4 × 10⁶ cells/ml with GolgiStop (BD Biosciences) at 1 μl/ml in 500 μl RPMI containing 10% FCS and antibiotics for 5 h at 37°C in 24-well flat-bottom plates. After the incubation, the cells were collected and preincubated with an anti-mouse CD16/32 mAb to avoid nonspecific binding of Abs to FcγR. The cells were subsequently stained with Pacific Blue–conjugated anti-mouse Gr-1 (RB6-8C5; BD Biosciences) and allophycocyanin-conjugated anti-mouse CD11b (M1/70; BD Biosciences) mAbs. The cells were then fixed and permeabilized with Cytofix/Cytoperm solution (BD Biosciences) at 4°C for 20 min and washed with Perm/Wash Buffer (BD Biosciences). Thereafter, the cells were stained with FITC-conjugated anti-mouse F4/80 (BM8; eBioscience) and PE-conjugated anti-mouse TNF (TN3-19; eBioscience) mAbs at room temperature for 20 min. The corresponding isotypes were used as controls.

The stained cells were analyzed using an FACSCanto II flow cytometer (BD Biosciences), and the data were processed using FlowJo software (Tree Star, Ashland, OR). The number of cells in a cell subset was determined using the following calculation: total liver leukocytes number × the corresponding cell subset proportion to the total cells.

CD11b⁺Gr-1^high cells and CD11b⁺Gr-1^dim cells were purified from prepared liver leukocytes by MACS using a cell isolation kit (Miltenyi Biotec, Bergisch Gladbach, Germany) according to the manufacturer’s protocol. The purities of the CD11b⁺Gr-1^high cell and CD11b⁺Gr-1^dim cells populations were >80 and 90%, respectively. Both the isolated cell populations and the remaining cells, which were not CD11b⁺Gr-1^high or CD11b⁺Gr-1^dim cells, were washed and resuspended in medium for further analysis.

CD11b⁺Gr-1^high cells were isolated from the liver leukocytes obtained from HFHCD-fed mice. Splenocytes were prepared from normal mice. The following groups were incubated with ConA (Sigma-Aldrich, St. Louis, MO) at 2 μg/ml for 72 h together with BrdU (Roche Applied Science, Mannheim, Germany) at 10 μmol for the final 3 h in 100 μl RPMI 1640 containing 10% FCS, antibiotics, 1% l-glutamine (Thermo Fisher Scientific, Yokohama, Japan), 1 mmol nonessential amino acids (Thermo Fisher Scientific), 1 mmol sodium pyruvate (Thermo Fisher Scientific), and 55 μmol 2-ME (Thermo Fisher Scientific) at 37°C in 96-well flat bottom plates: 1) 1 × 10⁵ prepared splenocytes; 2) 1 × 10⁵ prepared splenocytes together with isolated CD11b⁺Gr-1^high cells at the indicated ratios; 3) 1 × 10⁵ isolated CD11b⁺Gr-1^high cells; and 4) a cell-free control. Subsequently, cell proliferation was quantified using a Cell Proliferation ELISA (Roche Applied Science) according to the manufacturer’s protocol.

Prepared liver leukocytes or the cells isolated from liver leukocytes were respectively incubated at concentrations of 1 × 10⁶ or 6 × 10⁵ cells/ml in 150 μl RPMI containing 10% FCS for 24 h at 37°C in 96-well flat-bottom plates, and subsequently, the supernatants were collected. S100A8 levels in the supernatants were determined using a commercial ELISA kit (USCN Life Science, Hubei, China) according to the manufacturer’s instructions.

Prepared liver leukocytes at 1 × 10⁶ cells/ml were preincubated with or without TLR4 antagonist (LPS-RS Ultrapure; InvivoGen, San Diego, CA) at 100 μg/ml in 150 μl RPMI 1640 containing 10% FCS and antibiotics for 90 min at 37°C in 96-well flat-bottom plates, and then cultured for 12 h with or without rS100A8 (Abcam) at 1 or 5 μg/ml or LPS (Sigma Aldrich) at 1 μg/ml. The supernatants were then collected. BNL CL.2, a murine normal hepatocyte cell line, was obtained from the American Type Culture Collection (Manassas, VA). The cells were maintained in a humidified 5% CO₂ atmosphere at 37°C. Culture medium consisted of DMEM supplemented with 10% FCS and antibiotics. Prepared BNL CL.2 cells at 1 × 10⁵ cells/ml were incubated with or without rS100A8 (Abcam) at 1 μg/ml or palmitic acid (Sigma-Aldrich) at 300 μmol in 1 ml culture medium for 6 h at 37°C in 24-well flat-bottom plates. Subsequently, the supernatants were collected, and the cells were harvested for further analysis.

The levels of TNF-α or CXCL1 in the mice sera or cell supernatants were determined using commercial ELISA kits according to the manufacturer’s instructions. Kits for TNF-α and CXCL1 were obtained from R&D Systems.

To obtain bone marrow–derived macrophages (BMDMs), femurs and tibias were harvested bilaterally from normal mice, and marrow cores were lysed using RBC lysis buffer (Sigma-Aldrich). Collected cells were washed, plated, and cultured in 10 ml RPMI 1640 containing 10% FCS, antibiotics, and 20% supernatant from M-CSF–secreting L929 cells (a gift from Dr. Miyoshi) at 4 × 10⁶ cells/ml at 37°C in 10-cm dishes. After 7 d, the cells were harvested, and >95% of cells had differentiated into macrophages (assessed as CD11b⁺ F4/80⁺ cells by flow cytometry). To test the direct effect of S100A8 on macrophages, the BMDMs were treated with S100A8 (Abcam) alone at 1 or 5 μg/ml, IFN-γ (R&D Systems) at 20 ng/ml together with LPS (Sigma-Aldrich) at 100 ng/ml, or IL-4 (R&D Systems) alone at 20 ng/ml in 1 ml RPMI containing 10% FCS and antibiotics at 1 × 10⁶ cells/ml at 37°C in six-well flat-bottom plates for 6 h. Subsequently, the cells were harvested for further analysis.

Total RNA was isolated from liver tissues or cultured cells using the RNeasy kit (Qiagen, Hilden, Germany) according to the manufacturer’s instructions. cDNA was synthesized from isolated RNA using ReverTra Ace (Toyobo Life Science, Tokyo, Japan). Real-time RT-PCR analysis was performed using TaqMan Gene Expression Assays (Applied Biosystems, Foster City, CA) to examine the expression of the following genes: mouse-s100a8 (Mm00496696_g1), mouse-col1a1 (Mm00801666_g1), mouse-col1a2 (Mm00483888_m1), mouse-cxcl1 (Mm04207460_m1), mouse-tnf (Mm00443260_g1), mouse–β-actin (Mm00607939_s1), mouse–il1-β (Mm00434228_m1), mouse-mrc1 (Mm00485148_m1), and mouse-arg1 (Mm00475988_m1). All expression levels were corrected with the quantified expression level of β-actin. Data are shown as a fold increase compared with the control.

The protocol was approved by the institutional review boards at Osaka University Hospital, Osaka Rosai Hospital, and Ikeda Municipal Hospital. At the time of liver biopsy, written informed consent was obtained from all patients, and this study was conducted in accordance with the Helsinki Declaration.

A total of 54 patients with NAFLD were confirmed by liver biopsy and enrolled in this study. The profile of these subjects is shown in Table I. None of the patients had viral hepatitis, autoimmune disease, or excessive alcohol consumption (>20 g/d). The biopsy samples were embedded in paraffin and stained with H&E and Azan. According to the classification of Matteoni et al. (15), the subjects were divided into two groups: NAFL and NASH.

Immunohistochemical analyses of S100A8, CXCL1, and neutrophil elastase protein were performed on formalin-fixed paraffin-embedded liver sections using standard techniques. The sections were deparaffinized and rehydrated in PBS. Ag retrieval was performed following S100A8 and CXCL1 staining using a pressure cooker. For Ag retrieval following neutrophil elastase staining, liver sections were covered with 0.1% trypsin in PBS and incubated for 10 min at 37°C in a humidified chamber. Endogenous peroxidases were neutralized with 3% hydrogen peroxide and blocked with 5% goat serum in PBS. Subsequently, the samples were immunostained using an anti-mouse/human S100A8 mAb (Abcam), anti-mouse CXCL1 polyclonal Ab (Abcam), or an anti-mouse/human Neutrophil elastase polyclonal Ab (Abcam), and the Dako Envision system-HRP (DakoCytomation, Tokyo, Japan). Diaminobenzidine tetrahydrochloride (Vector Laboratories, Burlingame, CA) was used as a chromogen. The tissue sections were counterstained with Mayer’s hematoxylin. In the analyses for S100A8, the stained cell forms were homogeneous or heterogeneous in the mouse or human liver tissues, respectively. Thus, the stained cell number or stained cell area was evaluated in the mouse or human liver tissues, respectively. The stained cell number was counted. The stained cell area was quantified using image analysis software WinROOF (Mitani Corporation). The average stained cell number in a liver tissue section from an individual mouse or the average stained cell area in a liver tissue section from an individual subject was analyzed in three to four random fields of microscopic view (×200 original magnification) and determined as the S100A8-positive cell number of an individual mouse or the S100A8-positive area of an individual human subject, respectively.

Statistical analysis was performed using JMP 10.0 software (SAS Institute, Cary, NC). Data obtained from the mice or human experiments are presented as the means ± SD or as individual values, respectively. To assess the significant differences between groups as indicated, Student t test or Wilcoxon test was performed in mice or human experiments, respectively. A p value <0.05 was considered statistically significant.

Previous studies have shown that HFHCD-fed mice develop steatohepatitis (10). We investigated whether C57BL/6 male mice fed HFHCD for 3 wk developed steatohepatitis. Histological analyses by H&E staining demonstrated that steatosis and inflammation developed in the livers of HFHCD-fed, but not ND-fed, mice (Fig. 1A). The serum ALT level in HFHCD-fed mice was significantly greater than that in ND-fed mice (Fig. 1B). The total number of liver leukocytes in HFHCD-fed mice increased compared with that in ND-fed mice (Fig. 1C). These results indicated that the mice fed HFHCD for 3 wk developed steatohepatitis. Moreover, histological analyses by Sirius Red staining demonstrated that the stained areas in the liver sections of HFHCD-fed mice were significantly greater than those of ND-fed mice (Fig. 1A, 1D). The collagen 1a1 or collagen 1a2 gene expression level in the livers of HFHCD-fed mice significantly increased compared with that of ND-fed mice (Fig. 1E). These results indicated that the mice fed HFHCD for 3 wk developed steatohepatitis accompanied with liver fibrosis.

We investigated whether S100A8 was associated with the steatohepatitis in HFHCD-fed mice. Quantitative RT-PCR analyses demonstrated that the S100A8 gene expression levels in the livers of HFHCD-fed mice were significantly higher than those in ND-fed mice (Fig. 2A). To examine the distribution of S100A8 protein in the liver, we performed immunohistochemical analyses. The analyses demonstrated that the stained area was localized to nonparenchymal cells, but not parenchymal cells, in the livers of both HFHCD-fed and ND-fed mice (Fig. 2B). The number of S100A8-positive cells was significantly higher in the liver sections from HFHCD-fed mice than in those from ND-fed mice (Fig. 2B). Flow cytometric analyses also demonstrated the presence of S100A8-positive cells in the liver leukocyte population (Fig. 2C), and the numbers and proportions of S100A8-positive cells in the livers of HFHCD-fed mice were significantly higher compared with those of ND-fed mice (Fig. 2C). Collectively, these results suggested that S100A8-expressing cells were increased in the livers of HFHCD-fed mice compared with ND-fed mice. Moreover, ELISA analyses demonstrated that the liver leukocytes isolated from HFHCD-fed mice secreted significantly higher levels of S100A8 protein than those from ND-fed mice (Fig. 2D), suggesting that S100A8 was not only intracellularly expressed but also extracellularly secreted.

Next, we investigated which cells in the liver leukocyte population expressed and secreted S100A8. S100A8 has been described as a myeloid cell–related protein (5). We examined the expression of myeloid cell surface markers, including CD11b and Gr-1, on the liver leukocytes obtained from ND-fed or HFHCD-fed mice. Flow cytometric analyses revealed that ∼80 or 94% of the S100A8 positive cells were CD11b⁺Gr-1⁺ cells in the liver leukocytes obtained from ND-fed mice or HFHCD-fed mice, respectively (Fig. 3A). We also examined the expression of macrophage surface marker F4/80 together with the expression of CD11b on the S100A8-positive cells. More than 90% of the S100A8-positive cells in the liver leukocytes obtained from ND-fed mice or HFHCD-fed mice were CD/11b⁺F4/80⁻ cells, suggesting that the majority of S100A8-positive cells were not CD/11b⁺F4/80⁺ macrophages (Fig. 3A). The CD11b⁺Gr-1⁺ cells in the liver leukocytes from either ND-fed mice or HFHCD-fed mice comprised CD11b⁺Gr-1^high cells and CD11b⁺Gr-1^dim cells (Fig. 3B). More than 80 or 70% of CD11b⁺Gr-1^high cells in the livers of ND-fed mice or HFHCD-fed mice expressed intracellular S100A8 (Fig. 3B), respectively, whereas almost none of the CD11b⁺Gr-1^dim cells expressed this protein. Therefore, CD11b⁺Gr-1^high cells were a major cell subset that expressed intracellular S100A8. Moreover, the number of CD11b⁺Gr-1^high cells in the livers of HFHCD-fed mice was significantly higher compared with ND-fed mice (Fig. 3C). We subsequently examined the extracellular secretion of S100A8 from these cells. We isolated CD11b⁺Gr-1^high cells and CD11b⁺Gr-1^dim cells from the liver leukocytes of HFHCD-fed mice to evaluate the levels of extracellular S100A8 secretion using ELISA. Whereas CD11b⁺Gr-1^dim cells, as well as cells other than CD11b⁺Gr-1⁺ cells, did not secrete S100A8 at detectable levels, CD11b⁺Gr-1^high cells released a substantial amount of S100A8 (Fig. 3D). These results suggested that the CD11b⁺Gr-1^high cells in liver leukocytes predominantly expressed and secreted S100A8.

We further characterized the CD11b⁺Gr-1^high cells in the livers from ND-fed mice or HFHCD-fed mice. We examined the expression of myeloid cell-surface markers, such as Ly6G, Ly6C, CD11c, MHC-II, F4/80, and CD115, as well as lymphoid cell surface markers, such as NK1.1, TCRβ, CD45R, and Vα14, and an endothelial cell surface marker, CD31. The majority of the CD11b⁺Gr-1^high cells from the livers of both ND-fed mice and HFHCD-fed mice highly expressed Ly6G and Ly6C and did not express lymphoid cell-surface markers or the endothelial cell-surface marker (Fig. 4A, 4B). It has been reported that MDSCs express both CD11b and Gr-1and are comprised of two cell subsets: CD11b⁺Gr-1^high cells and CD11b⁺Gr-1^dim cells (16). We examined whether the CD11b⁺Gr-1^high cells in the current study possessed a similar suppressor function to MDSCs. A cell proliferation assay demonstrated that the CD11b⁺Gr-1^high cells isolated from the liver leukocytes of HFHCD-fed mice did not suppress the proliferation of ConA-activated splenocytes (Fig. 4C), suggesting that these CD11b⁺Gr-1^high cells in the livers from HFHCD-fed mice did not possess a suppressor function.

We investigated the accumulation of CD11b⁺Gr-1^high cells, a primary source of S100A8, in the livers of HFHCD-fed mice. CXCL1 is a chemokine responsible for the recruitment of granulocytic cells. We examined whether CXCL1 expression was elevated in the livers of HFHCD-fed mice. The serum CXCL1 levels in HFHCD-fed mice were significantly higher than in ND-fed mice (Fig. 5A). The levels of CXCL1 gene expression in the livers of HFHCD-fed mice were significantly higher compared with those of ND-fed mice (Fig. 5B). Immunohistochemical analyses for CXCL1 protein demonstrated that the stained areas in the liver sections of HFHCD-fed mice were clearly increased compared with those in ND-fed mice (Fig. 5C). Although the stained areas included both nonparenchymal and parenchymal cells in the livers of HFHCD-fed mice, the majority of these areas corresponded to parenchymal cells (Fig. 5C). We subsequently examined whether S100A8 was associated with the elevation of CXCL1 expression in the livers of HFHCD-fed mice. Stimulation with S100A8 in vitro significantly enhanced CXCL1 production from liver leukocytes, which was comparable to LPS-stimulated CXCL1 production in these cells (Fig. 5D). Stimulation with S100A8 in vitro also dramatically induced CXCL1 production in the normal murine hepatocyte cell line, BNL CL.2 (Fig. 5E). We also examined whether the lipid accumulation in the hepatocytes of HFHCD-fed mice was associated with the elevation of CXCL1 expression observed in the livers of these mice. In vitro stimulation with palmitic acid significantly enhanced CXCL1 gene expression in the normal murine hepatocyte cell line, BNL CL.2 (Fig. 5F).

We investigated whether increased CXCL1 expression led to the recruitment of CD11b⁺Gr-1^high cells into the livers of HFHCD-fed mice. The chemokine receptor CXCR2 is the counterpart of CXCL1. We examined the expression of CXCR2 on liver leukocytes. Flow cytometric analyses demonstrated that CXCR2-positive cells existed in the liver leukocyte population, and the majority of CXCR2-expressing cells were CD11b⁺Gr-1⁺ cells in the livers of HFHCD-fed mice and ND-fed mice (Fig. 6A). We subsequently examined CXCR2 expression on CD11b⁺Gr-1^high or CD11b⁺Gr-1^dim cells in the liver leukocytes of HFHCD-fed or ND-fed mice and assessed the relationship between S100A8 expression in and CXCR2 expression on these cells. The majority of CD11b⁺Gr-1^high cells simultaneously expressed both CXCR2 and S100A8 in the livers of HFHCD-fed and ND-fed mice (Fig. 6B). However, most of the CD11b⁺Gr-1^dim cells expressed neither CXCR2 nor S100A8 in the livers of HFHCD-fed and ND-fed mice (Fig. 6B). The proportion and number of CD11b⁺Gr-1^high cells expressing both CXCR2 and S100A8 in the livers of HFHCD-fed mice were significantly increased compared with those of ND-fed mice (Fig. 6C).

We investigated whether S100A8 was involved in the development of steatohepatitis in HFHCD-fed mice. TNF-α is a key factor in the development of steatohepatitis. Therefore, we first examined whether TNF-α was associated with the development of steatohepatitis in HFHCD-fed mice. The levels of TNF-α gene expression were significantly higher in the livers of HFHCD-fed mice compared with ND-fed mice (Fig. 7A). Liver leukocytes from HFHCD-fed mice produced remarkably higher levels of TNF-α compared with those from ND-fed mice (Fig. 7B). We next examined the role of TNF-α in the development of steatohepatitis in HFHCD-fed mice using TNF-α–deficient mice. Histological analyses by H&E staining demonstrated steatosis, but not inflammation, in the livers of TNF-α–deficient mice fed HFHCD (Fig. 7C). The total number of liver leukocytes in TNF-α–deficient mice fed HFHCD was remarkably decreased compared with wild-type mice fed HFHCD (Fig. 7C). These results suggested that TNF-α played a crucial role in the liver inflammation developed in mice fed HFHCD. We further examined whether S100A8 was associated with TNF-α expression. Liver leukocytes stimulated in vitro with S100A8 showed significantly enhanced TNF-α production, which was comparable to the levels of LPS-stimulated TNF-α production in these cells (Fig. 7D). Similar to the effect of LPS, the effect of S100A8 on TNF-α production in liver leukocytes was significantly reduced using a TLR4-specific antagonist (Fig. 7D). S100A8 also significantly upregulated TNF-α gene expression in the normal murine hepatocyte cell line, BNL CL.2 (Fig. 7E). Taken together, these results suggested that S100A8 was involved in the development of steatohepatitis in HFHCD-fed mice through the upregulation of TNF-α.

We further investigated the link between S100A8 and TNF-α production in liver leukocytes. We determined what cell types primarily generated TNF-α in response to S100A8 stimulation. Flow cytometric analyses revealed that the majority of TNF-α–positive cells in the liver leukocyte from normal mice following S100A8 stimulation in vitro were Gr-1^dim cells, and not Gr-1^high cells and that ∼80% of the TNF-α–producing cells were CD11b⁺F4/80⁺ cells (Fig. 8A). We also examined the TNF-α–producing cells in vivo in the liver leukocytes from ND-fed mice or HFHCD-fed mice. Flow cytometric analyses demonstrated that the majority of TNF-α–positive cells in the liver leukocytes from ND-fed mice or HFHCD-fed mice were Gr-1^dim cells, and not Gr-1^high cells, and that ∼75% of the TNF-α–producing cells were CD11b⁺F4/80⁺ cells (Fig. 8B). Moreover, CD11b⁺F4/80⁺ cell number in the liver leukocytes from HFHCD-fed mice was significantly greater than that from ND-fed mice (Fig. 8B). These results demonstrated that CD11b⁺F4/80⁺ cells are the main target of S100A8 with respect to TNF-α production both in mice and in primary liver leukocytes. Considering that CD11b⁺F4/80⁺ cells phenotypically represent macrophages (17), we next investigated whether S100A8 was associated with M1/M2 polarization in macrophages. IFN-γ with LPS or IL-4 alone have been reported to induce BMDMs to become M1- or M2-activated macrophages, respectively (17, 18). We examined the gene expression levels of M1/M2 markers in BMDMs in response to stimulation with S100A8 and compared these levels with those produced by stimulation with IFN-γ and LPS or with IL-4 alone as controls. Similar to the effect produced by IFN-γ with LPS and contrary to the effect produced by IL-4, S100A8 significantly upregulated the gene expression of M1 markers, such as IL-1β and TNF-α and significantly downregulated the gene expression of M2 markers, such as Arg1 and Mrc1, in BMDMs (Fig. 8C). Taken together, these results suggested that S100A8 induced M1 activation in CD11b⁺F4/80⁺ macrophages in the liver leukocytes from HFHCD-fed mice.

To investigate whether S100A8 is associated with the pathology of NAFLD, we performed immunohistochemical analyses on liver biopsy tissues from patients with NAFLD. The S100A-stained area was localized to nonparenchymal cells, but not parenchymal cells, in patients with NAFL and NASH (Fig. 9A). The S100A8-positive areas in the livers of patients with NASH were significantly increased compared with those in the livers of patients with NAFL (Fig. 9B). These results suggested that S100A8 is associated with the progression of NAFLD.

It has been suggested that S100A8 acts as an endogenous activator of the immune response (6). S100A8 expression is strongly upregulated in various inflammatory diseases, such as sepsis, rheumatoid arthritis, inflammatory bowel disease, vasculitis, and cancer (5). NAFLD has been increasingly recognized as an inflammatory disease as well as a metabolic disease (4). However, the involvement of S100A8 in the pathogenesis of NAFLD remains unclear. In the current study, we demonstrated that S100A8 expression was remarkably increased in the livers of HFHCD-fed mice that developed steatohepatitis. We also demonstrated that S100A8 protein was significantly more expressed in the liver tissues of patients with NASH than in those of patients with NAFL. These results suggested that S100A8 was associated with the development of NAFLD. Furthermore, we demonstrated that S100A8 was primarily produced in the myeloid lineage cell population, CD11b⁺Gr-1^high cells in the liver leukocytes of both ND- and HFHCD-fed mice and that S100A8 enhanced TNF-α expression in hepatic leukocytes as well as in a hepatocyte cell line. In vitro study, we found that CD11b⁺F4/80⁺ cells were the main target of S100A8 to produce TNF-α in primary liver leukocytes. The absence of TNF-α in mice ameliorated the liver inflammation induced through HFHCD feeding. Collectively, these results suggested that the cellular cross-talk between CD11b⁺Gr-1^high cells and CD11b⁺F4/80⁺ cells involved TNF-α production in response to S100A8 and that S100A8 is involved in the development of steatohepatitis, particularly the development of hepatitis, in HFHCD-fed mice through the upregulation of TNF-α expression.

In the current study, we observed that CD11b⁺Gr-1^high cells, in which S100A8 was primarily produced as described above, were significantly increased in the livers of HFHCD-fed mice. Ryckman et al. (19) demonstrated that S100A8 was a potent inducer of neutrophil chemotaxis and adhesion. Wiechert et al. (20) also reported that the hepatocyte-specific transgene expression of S100A8 in the livers of mice induced the specific transcriptional activation of CXCL1, a chemokine for myeloid lineage cell recruitment, and the systemic mobilization of neutrophils, suggesting a significant role for the S100A8–CXCL1 axis in myeloid lineage cell mobilization. In human samples, Bertola et al. (21) demonstrated that CXCL1 gene expression was upregulated in the livers of patients with NASH. Thus, we hypothesized that CXCL1 would be upregulated in the livers of HFHCD-fed mice, as we observed the development of steatohepatitis, the elevated expression of S100A8, and the recruitment of CD11b⁺Gr-1^high cells, a myeloid lineage cell, in the livers of HFHCD-fed mice. We observed that CXCL1 expression was significantly increased in the livers of HFHCD-fed mice. We also observed that S100A8 enhanced CXCL1 expression in hepatic leukocytes and hepatocytes. Palmitic acid, which induces lipid accumulation in hepatocytes (22), also upregulated CXCL1 expression in hepatocytes. Thus, it is possible that S100A8, produced from CD11b⁺Gr-1^high cells, and the lipid accumulation in hepatocytes through palmitic acid from HFHCD enhanced CXCL1 expression in the livers of HFHCD-fed mice.

CXCR2, the agonistic receptor for CXCL1, is expressed on neutrophils and is crucial for the recruitment of neutrophils to sites of inflammation (23). Recently, CXCR2 expression has been reported on other types of myeloid lineage cells, such as monocytes, mast cells, and macrophages, and on endothelial and epithelial cells, and this receptor plays an important role in the pathological conditions of lung disease, sepsis, atherosclerosis, and cancer (24–27). In the current study, we demonstrated that CXCR2, together with S100A8, was predominantly expressed on CD11b⁺Gr-1^high cells, and the number of these cells increased in the livers of HFHCD-fed mice, in which the expression of CXCL1, a ligand for CXCR2, was upregulated. S100A8 enhanced CXCL1 expression in hepatic leukocytes and hepatocytes. Thus, these results suggest that CD11b⁺Gr-1^high cells accumulated in the livers of HFHCD-fed mice through CXCR2 in response to increased CXCL1 expression in the liver, and the accumulated CD11b⁺Gr-1^high cells further augmented CXCR2-mediated recruitment through the production of S100A8 to upregulate CXCL1 expression in the liver. Recruited CD11b⁺Gr-1^high cells would cointeract with resident CD11b⁺F4/80⁺ cells, leading to production of TNF-α by S100A8. We propose that the S100A8–CXCL1 amplification loop, via CXCR2, substantially affected the development of steatohepatitis in HFHCD-fed mice. We performed CXCR2 blocking experiments using CXCR2 antagonists and an anti-CXCR2 Ab. However, we could not effectively prevent the recruitment of CXCR2-positive cells into the livers of HFHCD-fed mice (data not shown). Further studies are needed to determine whether CXCR2 is crucial for the development of steatohepatitis in HFHCD-fed mice.

Myeloid-derived CD11b⁺Gr-1⁺ cells have been grouped into two cell subsets: CD11b⁺Gr-1^high and CD11b⁺Gr-1^dim cells. The former corresponds to CD11b⁺Ly6G⁺ cells, and the latter corresponds to CD11b⁺Ly6C⁺ cells (16, 28). Deng et al. (29) reported that, among myeloid-derived cells, CD11b⁺Ly6C⁺ cells played a crucial role in the liver inflammation observed in mice fed a high-fat diet. However, the role of CD11b⁺Gr-1^high cells in the development of steatohepatitis remains unclear. In the current study, we demonstrated that CD11b⁺Gr-1^high cells predominantly expressed and secreted S100A8 in murine livers, and the number of these cells increased in the livers of HFHCD-fed mice. Considering that S100A8 would potentially deteriorate liver inflammation through the upregulation of TNF-α expression, the CD11b⁺Gr-1^high cells in the current study would act as proinflammatory cells in the development of steatohepatitis in HFHCD-fed mice.

Recent studies have reported that MDSCs act as anti-inflammatory cells in malignant diseases though the suppression of CD4⁺ T and/or CD8⁺ T cell activation against tumor cells (16). MDSCs comprise heterogeneous myeloid lineage cell populations. In mice, two major populations of MDSCs have been identified: monocytic MDSCs (CD11b⁺Gr-1^dim) and granulocytic MDSCs (CD11b⁺Gr-1^high) (16). Indeed, the CD11b⁺Gr-1^high cells in the current study possessed phenotypically the same surface markers as granulocytic MDSCs. However, the cells in the current study were functionally different from granulocytic MDSCs because they did not suppress the proliferation of ConA-activated splenocytes. The CD11b⁺Gr-1^high cells in the current study could act as proinflammatory cells by producing S100A8 to upregulate TNF-α and CXCL1 expression.

Neutrophils also express the same surface markers as the CD11b⁺ Gr-1^high cells examined in the current study (30). Moreover, neutrophils are reported to possess intracellular S100A8 (31). In addition to killing micro-organisms, neutrophils play a proinflammatory role in some immunological conditions and an anti-inflammatory role in other immunological conditions (32, 33). Neutrophils have been suggested to be associated with the development of NASH (34, 35). Thus, it is possible that the CD11b⁺Gr-1^high cells in the current study might be associated with a proinflammatory neutrophil population. Our preliminary immunohistochemical analysis of neutrophil elastase staining together with S100A8 staining on serial liver sections of HFHCD-fed mice demonstrated that some S100A8-positive cells also exhibited positive Neutrophil elastase staining (Supplemental Fig. 1). Considering that a majority of S100A8-positive cells were CD11b⁺Gr-1^high cells, which highly expressed Ly6G in the current study, a subset of the CD11b⁺Gr-1^high cells could be a neutrophil population. Further studies are needed to characterize these cells in more detail.

In conclusion, the results obtained in the current study suggest that S100A8, produced from CXCR2-expressing CD11b⁺Gr-1^high cells, upregulates CXCL1 and TNF-α expression and plays an important role in the development of NAFLD. To the best of our knowledge, this study is the first to suggest the involvement of S100A8 in the development of NAFLD. As the current study raises the possibility that S100A8 might be a therapeutic target for NAFLD, further investigation of the roles of this protein in the pathogenesis of NAFLD is warranted.

We thank Dr. Yoshihiro Kamada and Dr. Eiichi Morii for evaluation of liver histology and Dr. Eiji Miyoshi for providing L929 cells.

The authors have no financial conflicts of interest.