DNAM-1 Immunoreceptor Protects Mice from Concanavalin A–Induced Acute Liver Injury by Reducing Neutrophil Infiltration

Acute liver injury is a life-threatening disease with complications, such as multiple organ failure (1, 2). However, effective therapies for acute liver injury remain to be developed. Immune responses are known to be involved in the pathogenesis of acute liver injury: immune cells in the liver invoke excess inflammation, consequently leading to hepatocyte cell death in autoimmune liver injury, drug-induced liver injury, and viral hepatitis (3). Con A–induced acute liver injury is widely used as an animal model of immune response–mediated acute liver injury. Con A is a plant lectin that binds to glycosylated immunoreceptors, including the TCRs, and induces strong T cell activation (4–7). Previous studies have demonstrated that administration of Con A causes acute liver injury in a T cell–dependent manner, as demonstrated by the findings that T cell–depleted or T cell–deficient mice do not develop acute liver injury after Con A injection (7, 8). In Con A–induced acute liver injury, activated T cells efficiently produce IFN-γ and TNF-α, exacerbating the acute liver injury. Further, myeloid cell subsets, such as macrophages (Mϕs), neutrophils, and inflammatory monocytes, infiltrate the liver and exaggerate acute liver injury by producing inflammatory cytokines, including IL-6 and TNF-α, in Con A–induced acute liver injury (9–13). However, the molecular mechanisms by which immune responses exacerbate acute liver injury are not fully understood.

DNAX accessory molecule-1 (DNAM-1; also as known as CD226) is an activating immunoreceptor of the Ig superfamily and is expressed in several types of immune cell, including T cells (14). DNAM-1 is involved in the cytotoxicity of NK cells and cytotoxic T lymphocytes (14). DNAM-1 acts as a costimulatory molecule to promote proliferation and IFN-γ production by NK cells, CD8⁺ T cells, and type 1 innate lymphoid cells (ILC1s) (14–22). The ligands for DNAM-1 are CD155 and CD112, both of which are broadly expressed on hematopoietic cells and nonhematopoietic cells (23–26). CD155 is highly expressed in the liver and upregulated during acute liver injury (15, 27). Although CD155 is well known as a ligand shared by DNAM-1, TIGIT, and CD96, the function of CD155 itself is still largely unknown. CD155 acts as a costimulatory molecule in CD4⁺ T cells (28), whereas CD155 on nonhematopoietic cells transmits an inhibitory signal through an ITIM in its cytoplasmic portion (29–32) and suppresses cell adhesion in fibroblasts and p38 MAPK-mediated apoptosis in hepatocellular carcinoma. However, the role of a DNAM-1–CD155 axis in the pathogenesis of acute liver injury remains elusive.

In this study, we aimed to elucidate the role of DNAM-1 and CD155 in the pathogenesis of Con A–induced acute liver injury.

Wild-type (WT) C57BL/6 (B6) mice were purchased from CLEA Japan (Tokyo, Japan). Rag-1–deficient (Rag1^−/−) B6 mice were purchased from Jackson Laboratory (Bar Harbor, ME). DNAM-1–deficient (Cd226^−/−) mice on the B6 background were generated as described previously (18). Rag-1 and DNAM-1 double-deficient (Rag1^−/−Cd226^−/−) B6 mice were generated by crossing Rag1^−/− B6 mice with Cd226^−/− B6 mice. TIGIT-deficient (Tigit^−/−) mice were generated as described previously (33). CD155-deficient (Pvr^−/−) mice were kindly provided by Prof. G. Bernhardt (Hannover Medical School, Hannover, Germany) (34). Male mice at 8–11 wk of age were used for experiments. All mice were housed and maintained under specific pathogen-free conditions, and all procedures were approved by the University of Tsukuba animal ethics committee (approval number 22-153) and performed at the laboratory animal resource centers of the University of Tsukuba in accordance with the guidelines of the institutional animal ethics committees.

WT and Cd226^−/− mice were i.v. injected with 200 μl of 7.5 mg/kg body weight Con A (Sigma-Aldrich, St. Louis, MO) in PBS. Rag1^−/− and Rag1^−/−Cd226^−/− mice were i.v. injected with 200 μl of 15 mg/kg body weight Con A. Blood was collected 12 h after Con A injection, and plasma was prepared by centrifugation at 3000 rpm for 15 min at 4°C. Concentrations of alanine aminotransferase (ALT) in the plasma were measured using DRI-CHEM 7000V and GPT/ALT-PIII slides (Fujifilm Wako Pure Chemical Corporation, Tokyo, Japan).

To deplete neutrophils, we i.v. injected mice with 200 μg neutrophil-depleting mAb Gr-1 (clone RB6-8C5, kindly provided by Dr. M. Kondo, Toho University School of Medicine) or control rat IgG (BioXCell, Lebanon, NH) 24 h before Con A injection. To deplete CD4⁺ T cells, we i.p. injected mice with 200 μg CD4⁺ cell–depleting mAb (clone GK1.5; ATCC, Manassas, VA) (35) or control rat IgG 48 h before Con A injection. To neutralize TIGIT, we i.p. injected mice with 200 μg anti-TIGIT neutralizing mAb (clone TX99) (36) or control rat IgG 24 h before Con A injection.

Blood was collected 12 h after Con A injection, and plasma was prepared by centrifugation at 3000 rpm for 15 min at 4°C. Concentrations of IFN-γ, IL-6, IL-10, IL-12p70, CXCL1, and TNF-α in the plasma were analyzed by using BD Cytometric Bead Array Mouse Flex Sets for mouse IFN-γ, mouse IL-6, mouse IL-10, mouse IL-12p70, mouse KC, and mouse TNF (BD Biosciences, Franklin Lakes, NJ) and BD LSRFortessa (BD Biosciences). These concentrations were calculated using FCAP Array v3 software (BD Biosciences).

T cells, NKT cells, NK cells, ILC1, and non–platelet-adherent B cells in the liver were identified as TCRβ⁺NK1.1⁻ or CD41⁻CD19⁻TCRβ⁺NK1.1⁻ cells, TCRβ⁺NK1.1⁺DX5⁻ or CD41⁻CD19⁻TCRβ⁺NK1.1⁺ cells, TCRβ⁻NK1.1⁺DX5⁺CD49a⁻ cells, TCRβ⁻NK1.1⁺DX5⁻CD49a⁺ cells, and CD41⁻CD19⁺ cells, respectively, as described previously (15). To evaluate the percentages and number of T cell subsets in the liver, we identified conventional CD4⁺ T cells (Tconvs), regulatory T cells (Tregs), and CD8⁺ T cells as CD3ε⁺NK1.1⁻CD4⁺Foxp3⁻, CD3ε⁺NK1.1⁻CD4⁺Foxp3⁺, and CD3ε⁺NK1.1⁻CD8⁺ lymphocytes, respectively, as described previously (15).

The liver myeloid cell subsets were isolated as described previously (15). Myeloid cell subsets were gated on leukocytes with large forward scatter (FSC)-side scatter (SSC) values followed by gating CD45.2⁺TCRβ⁻B220⁻NK1.1⁻ cells, conventional type 1 dendritic cells (cDC1s; F4/80⁻CD11c⁺I-A^b+CD11b⁻), conventional type 2 dendritic cells (F4/80⁻CD11c⁺I-A^b+CD11b⁺), Kupffer cells (F4/80^highCD11c⁻CD11b⁻), Mϕs (F4/80⁺CD11c⁻CD11b^high), neutrophils (CD11b⁺Ly6G⁺Ly6C^{− to low}), and inflammatory monocytes (CD11b⁺Ly6G⁻Ly6C^high).

Hepatocytes were isolated as described previously with some modifications (15). In brief, the liver of naive and Con A–injected mice was perfused at 37°C with 20 ml Ca²⁺ chelator buffer followed by 20 ml collagenase liver perfusion buffer containing 40 mg collagenase type IV per mouse by using a 25G needle (Terumo Corporation, Tokyo, Japan) and a Pharmacia Biotech Pump P-1 (GE Healthcare, Chicago, IL) at a flow rate of 20 (100 ml/h) via the portal vein. The liver was transferred into a 10-cm dish with 10 ml PBS supplemented with 10% FCS, shaken gently, and repeatedly pinched with forceps to release hepatocytes. The liver cell suspension was passed through a ϕ100-μm cell strainer. The remaining liver tissues were ground on the ϕ100-μm cell strainer using the plunger of a 10-ml syringe, resuspended in PBS supplemented with 10% FCS, and collected by being passed through the ϕ100-μm cell strainer. To collect hepatocytes, we combined both cell suspensions and washed twice by centrifugation at 50 × g for 1 min. These cell pellets were resuspended and used as crude hepatocytes for quantitative reverse transcription–mediated PCR. For the primary cell culture, crude hepatocytes were resuspended with 4 ml PBS, gently mixed with 5 ml 90% Percoll PLUS medium (GE Healthcare), and centrifuged at 800 rpm for 10 min at 20°C. The cell pellet was washed three times by centrifugation at 600–700 rpm for 5 min with PBS supplemented with 10% FCS and gently resuspended with the culture medium for primary hepatocytes.

The liver cell suspension, containing hepatocytes, endothelial cells, hepatic stellate cells, and hematopoietic cells, was prepared as described earlier. The cell suspensions were centrifuged at 50 × g for 1 min, and the supernatants were collected for sorting endothelial cells, hepatic stellate cells, and hematopoietic cells. Endothelial cells (CD45.2⁻CD31⁺Vitamin A/retinol⁻), hepatic stellate cells (CD45.2⁻CD31⁻Vitamin A/retinol⁺), and hematopoietic cells (CD45.2⁺) were stained for cell-surface molecules and sorted by BD FACSAria III (BD Biosciences). Vitamin A and retinol were excited using a 405-nm violet laser, and the emission was detected at 450/50-nm filter.

Ex vivo culture of liver lymphocytes was performed as described previously (15). Liver lymphocytes were enriched from the liver of naive and Con A–injected mice 12 h after the injection, and 1 million cells were cultured in the presence of BD GolgiStop (BD Biosciences) for 5 h. The cells were then collected, washed, stained for surface molecules and intracellular IFN-γ and TNF-α, and analyzed by flow cytometry.

Twenty-five thousand hepatocytes were cultured at 37°C, and 10% CO₂ in DMEM/F12 culture medium (Sigma-Aldrich) was supplemented with sodium bicarbonate, 10% FCS, 50 μM 2-ME, 2 mM l-glutamine, 100 U penicillin, 0.1 mg/ml streptomycin, 20 mM HEPES, 100 nM dexamethasone, 0.2 mM l-ascorbic acid 2-phosphate sesquimagnesium salt hydrate, and 2 μg/ml insulin from bovine pancreas (Sigma-Aldrich) in the wells of 48-well cell culture plates coated with 5 μg/cm² mouse natural collagen type IV (Corning, NY). At 12 h after plating, floating dead hepatocytes were removed, and the medium was replaced with a fresh culture medium. At 24 h after initial plating, 250,000 liver lymphocytes prepared as described earlier were added and cocultured in the presence of 3 μg/ml Con A for 12 h at 37°C and 10% CO₂.

Fc receptors (CD16 and CD32) were blocked with mAb clone 2.4G2 (BD Biosciences and Tonbo Biosciences, San Diego, CA) before surface and intracellular staining with the indicated fluorochrome-conjugated mAbs or isotype-matched control Abs (Tonbo Biosciences; BD Biosciences; BioLegend, San Diego, CA; and Thermo Fisher Scientific, Waltham, MA). mAbs used were FITC-conjugated anti-mouse B220 (clone RA3-6B2), FITC-conjugated anti-mouse CD19 (clone 1D3), FITC-conjugated anti-mouse CD31 (clone 390), FITC-conjugated anti-mouse CD45.2 (clone 104), FITC-conjugated anti-mouse CD49b (clone DX5), FITC-conjugated anti–mouse I Ab (clone AF6-120.1), FITC-conjugated anti-mouse NK1.1 (clone PK136), FITC-conjugated anti-mouse TCRβ (clone H57-597), PE-conjugated anti-mouse B220 (clone RA3-6B2), PE-conjugated anti-mouse CD3ε (145-2C11), PE-conjugated anti-mouse CD8α (clone 53-6.7), PE-conjugated anti-mouse CD96 (clone 3.3), PE-conjugated anti-mouse NK1.1 (clone PK136), PE-conjugated anti-mouse Ly-6G (clone 1A8), PE-Cy5–conjugated anti-mouse CD11b (clone M1/70), PerCP-Cy5.5–conjugated anti-mouse NK1.1 (clone PK136), Alexa Fluor 647–conjugated anti-mouse CD3ε (clone 145-2C11), Alexa Fluor 647–conjugated anti-mouse CD31 (clone 390), allophycocyanin-conjugated anti-mouse CD11c (clone HL3), allophycocyanin-conjugated anti-mouse CD45.2 (clone 104), allophycocyanin-conjugated anti-mouse CD49a (clone HMa1), allophycocyanin-conjugated anti-mouse TIGIT (clone 1G9), allophycocyanin-conjugated anti-mouse Siglec-H (clone 551.3D3), Alexa Fluor 700–conjugated anti-mouse CD4 (clone RM4-5), Alexa Fluor 700–conjugated anti-mouse CD8α (clone 53-6.7), Alexa Fluor 700–conjugated anti-mouse Ly-6C (clone HK1.4), allophycocyanin-Cy7–conjugated anti-mouse CD41 (clone MWReg30), PE-Cy7–conjugated anti-mouse B220 (clone RA3-6B2), PE-Cy7–conjugated anti-mouse CD25 (clone PC61), PE-Cy7–conjugated anti-mouse F4/80 (clone BM8), PE-Cy7–conjugated anti-mouse TCRβ (clone H57-597), BD Horizon V450-conjugated anti-mouse Ly-6G and Ly-6C (clone: RB6-8C5), Pacific Blue–conjugated anti-mouse CD4 (clone RM4-5), Pacific Blue–conjugated anti-mouse CD69 (clone H1.2F3), Brilliant Violet 421–conjugated anti-mouse CD45.2 (clone 104), Brilliant Violet 421–conjugated anti-mouse CD112 (clone 829038), Brilliant Violet 421–conjugated streptavidin, biotinylated anti-mouse CD155 (clone TX56, homemade) (20), biotinylated anti-mouse DNAM-1 (clone TX42.1) (16), biotinylated anti-mouse TIGIT (clone TX99) (36), Brilliant Violet 421–conjugated streptavidin, Brilliant Violet 605–conjugated streptavidin, and Brilliant Violet 711–conjugated anti-mouse CD8 (clone 53-6.7).

For staining of intracellular Foxp3, cells were fixed and permeabilized with a Foxp3 Transcription Factor Staining Buffer Set (eBioscience, San Diego, CA), washed with the wash buffer (eBioscience), and stained with Alexa Fluor 488–conjugated anti-mouse Foxp3 mAb (clone 150D; BioLegend).

For staining of intracellular IFN-γ and TNF-α, liver lymphocytes were fixed and permeabilized with BD Cytofix/Cytoperm solution (BD Biosciences), washed with Intracellular Staining Perm Wash Buffer (BioLegend), and stained with PE-conjugated anti-mouse IFN-γ (clone XMG1.2; BioLegend) and PE-conjugated anti-mouse TNF-α (clone MP6-XT22; BD Biosciences).

For staining of intracellular albumin in the crude hepatocytes, cells were fixed with 1% paraformaldehyde (Fujifilm Wako Pure Chemical Corporation), permeabilized with BD Phosflow Perm Buffer III (BD Biosciences), washed with 10% PBS, and stained with rabbit anti-mouse albumin (clone EPR20195; Abcam, Cambridge, U.K.), followed by PE-conjugated donkey anti-rabbit polyclonal IgG (BioLegend).

In all experiments, doublet cells were excluded by FSC-A and FSC-H gating, followed by SSC-A and SSC-W gating, and dead cells and cells with bright autofluorescence were excluded by using AmCyan (525/50-nm filter) as a dump channel of BD LSRFortessa and BD FACSAria III (BD Biosciences). In some experiments, cells were stained with Zombie NIR Fixable Viability Kit (BioLegend) or propidium iodide (Sigma-Aldrich), and living cells were defined as Zombie NIR- or propidium iodine–negative cells. Samples were run on BD LSRFortessa or BD FACSAria III (BD Biosciences), and the data were analyzed with FlowJo v10 software (FlowJo, Ashland, OR).

The liver of naive and Con A–injected mice (12 h after Con A injection) was fixed with 4% paraformaldehyde overnight at 4°C. Paraffin-embedded specimens were sliced into 5-µm cross sections and stained with H&E, or incubated with rabbit anti-mouse Ly6G mAb (clone E6Z1T; Cell Signaling Technology, Danvers, MA), followed by HRP-conjugated donkey anti-rabbit IgG (BioLegend), and then stained with Liquid DAB+ Chromogen Solution (Agilent, Santa Clara, CA). Images were acquired under a BZ-X710 microscope (Keyence, Osaka, Japan). The damaged area and infiltration of neutrophils in the damaged areas were quantified at five random images in one lobe of the liver of individual mice by using Keyence BZ-X analyzer and ImageJ software, respectively. The congestion, vacuolization, and necrosis in each sample were graded as 0 (no symptoms), 1 (minimal), 2 (0–30%), 3 (31–60%), and 4 (>60%) for a sign of acute liver injury defined by Suzuki’s scoring criteria (37).

The liver of naive and Con A–injected mice 6 h after the injection was isolated, and 5-mm³ pieces of the liver were excised and homogenized in Isogen reagent (Nippon Gene, Tokyo, Japan) by using frosted slide glasses (Matsunami Glass Industry, Osaka, Japan). In some experiments, hepatocytes, endothelial cells, hepatic stellate cells, and hematopoietic cells were sorted from the liver of naive and Con A–injected mice 3 h after the injection and resuspended with Isogen. First-strand DNA was synthesized by using a High-Capacity cDNA Reverse Transcription Kit (Thermo Fisher Scientific). Quantitative reverse transcription–mediated PCR was performed on an ABI 7500 Fast real-time PCR system with ABI Power SYBR Green PCR Master Mix (Thermo Fisher Scientific) and 50 nM (Cd112d), 100 nM (Actb, Cd112a, and Pvr), and 500 nM (Cxcl1 and Cxcl2) primers. The primers were as follows: Actb forward, 5′-ACTG TCGAGTCGCGTCCA-3′; Actb reverse, 5′-GCAGCGATATCGTCATCCAT-3′; Cd112a forward, 5′-GCATCATTGGAGGTATTATCGCT-3′; Cd112a reverse, 5′-GAGGGAGGTCCTTCCAGTTC-3′; Cd112d forward, 5′-CTCTGTGGATCGAATGGTCA-3′; Cd112d reverse, 5′-GGCAGCGATAATACCTCCAA-3′; Cxcl1 forward, 5′-CCGAAGTCATAGCCACACTCAA-3′; Cxcl1 reverse, 5′-GCAGTCTGTCTTCTTTCTCCGTTA-3′; Cxcl2 forward, 5′-GAAGTCATAGCCACTCTCAAGG-3′; Cxcl2 reverse, 5′-CCTCCTTTCCAGGTCAGTTAGC-3′; Pvr forward, 5′-CAACTGGTATGTTGGCCTCA-3′; Pvr reverse, 5′-ATTGGTGACTTCGCACACAA-3′. Relative quantification of gene expression was performed using the comparative cycle threshold method. The amount of target gene, normalized to β-actin and relative to a calibrator (naive WT), was determined by the arithmetic Equation 2^−ΔΔCt.

The Student t test, one-way ANOVA, and two-way ANOVA were used to compare the data by using GraphPad Prism software (GraphPad Software, San Diego, CA). A p value <0.05 was considered statistically significant. Each data point represents the average of samples. Error bars show the SEM.

To investigate the role of the DNAM-1–CD155 axis in the pathogenesis of Con A–induced liver injury, we first examined the expression of DNAM-1 on lymphocyte subsets in the liver. Liver T cells, NKT cells, NK cells, and ILC1 significantly expressed DNAM-1 (Fig. 1A, Supplemental Fig. 1A). These cell subsets expressed negligible or small amounts of TIGIT and CD96, which share the ligand CD155, in naive state (Supplemental Fig. 1B). Twelve hours after Con A injection, liver T cells, NKT cells, and NK cells/ILC1 maintained the expression of DNAM-1, whereas part of these cell subsets slightly upregulated the TIGIT expression and downregulated the CD96 expression (Supplemental Fig. 1C, 1D). Liver T cells did not show any skewed expression of TIGIT on either DNAM-1^high or DNAM-1^low T cells (Supplemental Fig. 1C, 1D). Liver B cells did not express DNAM-1, TIGIT, and CD96 in naive state and after Con A injection (Supplemental Fig. 1C, 1E). We injected Con A into WT and DNAM-1–deficient (Cd226^−/−) mice and evaluated plasma concentrations of ALT as a marker of acute liver injury 12 h after Con A injection when Con A–induced acute liver injury peaks (38, 39). Because Con A–induced acute liver injury depends on T cells (7, 8), in which DNAM-1 mediates a costimulatory signal, we hypothesized that Cd226^−/− mice would exhibit ameliorated liver injury compared with WT mice after Con A injection. Unexpectedly, however, Cd226^−/− mice rather showed exacerbated liver injury compared with WT mice 12 h after Con A injection (Fig. 1B). In contrast, neutralization of TIGIT had little impact on liver injury in WT mice 12 h after Con A injection (Supplemental Fig. 1F). To examine whether DNAM-1 on T cells is required for the milder liver injury in WT mice compared with Cd226^−/− mice, we injected Con A into Rag-1–deficient (Rag1^−/−) and Rag-1 and DNAM-1 double-deficient (Rag1^−/−Cd226^−/−) mice lacking T and B cells. Both Rag1^−/− and Rag1^−/−Cd226^−/− mice showed comparable levels of plasma ALT after Con A injection (Fig. 1C). Because B cells did not express DNAM-1, these results suggested that DNAM-1 on T cells was responsible for the milder liver injury after Con A injection. Cd226^−/− mice also exhibited higher plasma concentrations of inflammatory cytokines, such as IL-6 and TNF-α, but not IFN-γ, IL-10, and IL-12, compared with WT mice 12 h after Con A injection (Fig. 1D). These results suggest that DNAM-1 deficiency exacerbates Con A–induced acute liver injury.

To analyze the function of DNAM-1 in T cell activation after Con A injection, we evaluated the number and activation of Tconvs, Tregs, and CD8⁺ T cells in the liver of naive and Con A–injected mice. A similar number of these T cell subsets were present in the liver of naive and Con A–injected WT and Cd226^−/− mice 6 and 12 h after the injection (Fig. 2A). WT and Cd226^−/− liver T cells equivalently upregulated T cell activation markers, CD69 and CD25, 6 and 12 h after Con A injection (Fig. 2B). Further, WT and Cd226^−/− liver T cells equivalently produced IFN-γ and TNF-α ex vivo after Con A injection (Fig. 2C). These results suggest that DNAM-1 is not critical for the activation of CD4⁺ and CD8⁺ T cells in the liver after Con A injection.

Previous studies have demonstrated that myeloid cell subsets are involved in the exacerbation of Con A–induced acute liver injury (10–13). We hypothesized that myeloid cells are involved in the severe liver injury of Cd226^−/− mice after Con A injection. When we evaluated the infiltration of myeloid cell subsets into the liver after Con A injection (Supplemental Fig. 2A), a significantly larger number of neutrophils infiltrated into the liver of Cd226^−/− mice 12 h after Con A injection than into the liver of WT mice (Fig. 3A). However, neutrophils in the liver of naive and Con A–injected mice did not express DNAM-1 (Supplemental Fig. 2B). Nevertheless, to examine whether a larger number of neutrophils in the liver exacerbated acute liver injury in Cd226^−/− mice, we depleted neutrophils by injection of neutrophil-depleting mAb Gr-1 (Supplemental Fig. 2C) and then evaluated the severity of the acute liver injury. Depletion of neutrophils ameliorated Con A–induced liver injury, as demonstrated by low concentrations of ALT and decreased IL-6 and TNF-α in the plasma in Cd226^−/− mice to levels comparable with WT mice after Con A injection (Fig. 3B, 3C). These results suggest that neutrophils exacerbate Con A–induced acute liver injury in Cd226^−/− mice and neutrophils are the major sources of IL-6 and TNF-α, both of which are known to exacerbate Con A–induced acute liver injury (7, 40). Further, Cd226^−/− mice exhibited larger damaged areas and higher histological liver injury scores after Con A injection than WT mice, and depletion of neutrophils decreased the damaged areas and liver injury scores in the histology in Cd226^−/− mice to levels comparable with those in WT mice (Fig. 3D). Consistent with increased infiltration of neutrophils in the liver of Cd226^−/− mice after Con A injection (Fig. 3A), a larger number of neutrophils were infiltrated into the damaged areas of the liver of Cd226^−/− mice after Con A injection than into those of WT mice (Fig. 3E). Taken together, these results demonstrate that DNAM-1 deficiency results in increased infiltration of neutrophils into the liver and exacerbation of acute liver injury after Con A injection.

To investigate the mechanisms of increased neutrophil infiltration into the liver of Cd226^−/− mice, we quantified transcripts of neutrophil-recruiting chemokines, Cxcl1 and Cxcl2, in the liver of naive and Con A–injected mice 3–6 h after Con A injection, which were time points before acute liver injury culminates 12 h after Con A administration. Cxcl1, but not Cxcl2, was highly expressed in the liver, and larger amounts of Cxcl1 were expressed in the liver of Cd226^−/− mice than those of WT mice (Fig. 4A). Consistent with these findings, Cd226^−/− mice showed higher plasma concentration of CXCL1 than WT mice 12 h after Con A injection (Fig. 4B), suggesting that increased CXCL1 production induces the infiltration of a large number of neutrophils in the liver of Cd226^−/− mice. Next, we investigated the source of CXCL1 in the liver. Previous studies have demonstrated that hepatocytes, endothelial cells, hepatic stellate cells, and Kupffer cells in the liver produce CXCL1 during liver injury (41–44). We prepared the crude hepatocytes and sorted endothelial cells, hepatic stellate cells, and hematopoietic cells in the liver of WT mice (Supplemental Fig. 3A). Consistent with previous studies (45, 46), the large and small cell populations were present in the crude hepatocytes, and both populations expressed a hepatocyte marker albumin (Supplemental Fig. 3B). The crude hepatocytes contained 2–8% nonparenchymal cells, including endothelial cells, hepatic stellate cells, and hematopoietic cells (Supplemental Fig. 3B), indicating that the purity of hepatocytes in the crude hepatocytes was not <92% in this study. The expression of Cxcl1 was higher in the crude hepatocytes than that in purified endothelial cells, hepatic stellate cells, and hematopoietic cells in the liver of WT mice particularly after Con A injection (Fig. 4C). These results demonstrate that hepatocytes are the major source of CXCL1 during Con A–induced acute liver injury. Consistent with higher expression of Cxcl1 in the liver and high concentration of CXCL1 in the plasma of Cd226^−/− mice after Con A injection (Fig. 4A, 4B), hepatocytes in Cd226^−/− mice expressed larger amounts of Cxcl1 than those in WT mice after Con A injection (Fig. 4D). These results indicate that DNAM-1 indirectly regulates CXCL1 production by hepatocytes after Con A injection.

Previous studies have demonstrated that CD4⁺ T cells are critical for the development of Con A–induced acute liver injury (7). To examine whether DNAM-1 on CD4⁺ T cells is involved in milder acute liver injury and decreased neutrophil infiltration in the liver, we depleted CD4⁺ T cells by injection of a CD4⁺ cell–depleting mAb into both genotypes of mice before injection of Con A. Depletion of CD4⁺ T cells ameliorated liver injury and decreased plasma concentrations of IL-6 and CXCL1 in Cd226^−/− mice to levels comparable with those in WT mice after Con A injection (Fig. 5A, 5B). These results indicate that DNAM-1 on CD4⁺ T cells is involved in the amelioration of acute liver injury and decreased plasma concentrations of IL-6 and CXCL1 after Con A injection.

Hepatocytes did not express DNAM-1 in naive and Con A–injected mice (Supplemental Fig. 4A). By contrast, they expressed a DNAM-1 ligand CD155 and upregulated the expression of its Pvr transcript after Con A injection (Fig. 6A, 6B). In contrast, another DNAM-1 ligand CD112 was expressed on neither hepatocytes nor hematopoietic cells before and after Con A injection (Supplemental Fig. 4B–4D). Previous studies have demonstrated that CD155 acts as an inhibitory receptor in nonhematopoietic cells (30, 31). Thus, we hypothesized that CD155 on hepatocytes inhibits CXCL1 production by hepatocytes after Con A injection. To examine this hypothesis, we established a coculture system of primary hepatocytes and liver lymphocytes in the presence of Con A. To investigate the role of the DNAM-1–CD155 axis in CXCL1 production by hepatocytes, we cocultured WT or Cd226^−/− liver lymphocytes and WT hepatocytes in the presence of Con A and evaluated the concentration of CXCL1 in the culture supernatant. When hepatocytes were cocultured with Cd226^−/− liver lymphocytes, hepatocytes produced a large amount of CXCL1, as compared with those cocultured with WT liver lymphocytes (Fig. 6C). In contrast, TIGIT deficiency of liver lymphocytes had little impact on CXCL1 production by hepatocytes (Supplemental Fig. 4E). To directly address the inhibitory effect of CD155 on CXCL1 production by hepatocytes, we cocultured WT or CD155-deficient (Pvr^−/−) hepatocytes with WT liver lymphocytes. When hepatocytes were cocultured with WT liver lymphocytes, Pvr^−/− hepatocytes produced a larger amount of CXCL1, as compared with WT hepatocytes (Fig. 6D). These results strongly suggest that CD155 on hepatocytes inhibits CXCL1 production through the interaction with DNAM-1 on CD4⁺ T cells in the liver, leading to ameliorating Con A–induced acute liver injury by reducing neutrophil infiltration and proinflammatory cytokines (Fig. 6E).

In this study, we investigated the role of DNAM-1 and the ligand CD155 in the pathogenesis of Con A–induced acute liver injury. We found that the DNAM-1–CD155 axis ameliorates acute liver injury by reducing CXCL1 production by hepatocytes, resulting in reduced neutrophil infiltration into the liver. Because Con A–induced liver injury is dependent on T cells (7, 8), in which DNAM-1 mediates a costimulatory signal (20), we originally speculated that Cd226^−/− mice exhibited impaired activation of T cells and milder liver injury than did WT mice. However, rather, Cd226^−/− mice exhibited more severe liver injury than did WT mice. Because Con A induces rapid and strong T cell activation through strongly cross-linking the TCR (4–7), it is possible that Con A–mediated T cell full activation could overcome the lack of DNAM-1–mediated costimulation in Cd226^−/− T cells in vivo after Con A injection; consequently, both CD4⁺ and CD8⁺ T cells in the liver showed equivalent activation and proliferation and production of IFN-γ and TNF-α between WT and Cd226^−/− mice after Con A injection.

We demonstrated that increased neutrophil infiltration exacerbates Con A–induced acute liver injury in Cd226^−/− mice compared with WT mice. Importantly, the depletion of neutrophils ameliorated liver injury and decreased plasma concentrations of proinflammatory cytokines IL-6 and TNF-α in Cd226^−/− mice to levels comparable with those in WT mice. It may suggest that neutrophils could be the major effector immune cells on liver injury possibly through IL-6 and TNF-α in Cd226^−/− mice. Indeed, previous studies demonstrated that neutrophils exacerbate acute liver injury through the production of inflammatory cytokines, generation of reactive oxygen species, and release of detrimental enzymes in acute liver injury induced by acetaminophen, ischemia and reperfusion, and Con A in mice and ischemia and reperfusion in humans (13, 47–50). Taken together, our findings suggest that infiltrating neutrophil-derived IL-6 and TNF-α might exacerbate the acute liver injury in Cd226^−/− mice, although the involvement of other functions of neutrophils, such as reactive oxygen species generation, NETosis, and digestive enzyme release, was not addressed in this study.

Intriguingly, a prior study has demonstrated that CXCL1 contributes to the aggravation of mouse acetaminophen-induced acute liver injury and recruits neutrophils in the necrotic areas of the liver (51). An elevation of the systemic concentration of CXCL1 has been reported to be positively correlated with the severity of acute liver injury after HBV infection (52). However, little is known about the regulation of CXCL1 production during acute liver injury. We demonstrated that hepatocytes are the major source of Cxcl1 in the early course of Con A–induced acute liver injury and, more importantly, CD155 inhibits CXCL1 production by hepatocytes. Previous studies have demonstrated that nonparenchymal cells in the liver, such as endothelial cells, hepatic stellate cells, and Kupffer cells, also produce CXCL1 during liver injury (41–44). CD155 is broadly expressed on hematopoietic cells and nonhematopoietic cells (23–26). Given that CD155 inhibits CXCL1 production by not only hepatocytes but also nonparenchymal cells, the DNAM-1–CD155 axis might have a protective role in other types of acute liver injury or the late phase of liver injury through reducing neutrophil infiltration.

Although CD155 is well known as a ligand of DNAM-1 (23, 24), the function of CD155 is still largely unknown. Although we have previously reported that CD155 mediates an activating signal in CD4⁺ T cells as a costimulatory molecule (28), CD155 has been reported to mediate an inhibitory signal through the ITIM in the cytoplasmic portion in nonhematopoietic cells: CD155 suppresses cell adhesion in fibroblasts and p38 MAPK-mediated apoptosis in hepatocellular carcinoma and breast cancer cells (30–32). Our findings that CD155 inhibits CXCL1 production by hepatocytes during acute liver injury strongly support an aspect of CD155 as an inhibitory receptor. However, the downstream signaling pathway has not been addressed in this study. Previous studies have demonstrated that CD155 transmits an inhibitory signal through the recruitment of SHP-2 in fibroblasts and inhibits the p38 MAPK signaling pathway in hepatocellular carcinoma (30, 31). Moreover, an activating signal for CXCL1 production that is inhibited by CD155 still remains elusive. TNF-α promotes transcription of Cxcl1 gene via transcription factors NF-κB in hepatocytes in an acute liver injury model by injecting necrotic cells (41). TNF-α also activates the transcription factor Egr-1, which promotes the transcription of Cxcl1 (53). Moreover, the CD147-mediated FAK/PI3K signaling pathway, MMP12-mediated ERK/p38 MAPK signaling pathway, and IL-17– or TNF-α–mediated AP-1 also promote expression of Cxcl1 (54–57). In this study, we could not identify an activating signal to stimulate CXCL1 production by hepatocytes after Con A injection. An identification of an activating signaling pathway to promote CXCL1 production in hepatocytes, whose signal is counteracted by CD155-mediated inhibitory signaling, might be critical in the pathogenesis of acute liver injury and the consequent severity of the acute liver injury.

CD155 is shared as a ligand for DNAM-1, TIGIT, and CD96 (14, 23, 24, 58, 59). Liver T cells, NKT cells, and NK cells/ILC1 stably expressed DNAM-1 in a naive state and after Con A injection. By contrast, these cell subsets expressed small amounts of TIGIT and CD96 in a naive state, although they slightly upregulated TIGIT after Con A injection. Moreover, neutralization of TIGIT in vivo had little impact on acute liver injury after Con A injection, and coculture of Tigit^−/− liver lymphocytes with hepatocytes did not promote CXCL1 production by hepatocytes, suggesting that TIGIT has little impact on Con A–induced acute liver injury. Taken together, an interaction of DNAM-1, but neither TIGIT nor CD96, on liver lymphocytes with CD155 on hepatocytes would play a pivotal role in the inhibition of CXCL1 production during the acute liver injury. DNAM-1 is physically associated with LFA-1 on the cell surface of NK cells and CD4⁺ T cells (19, 60). A ligand for LFA-1, ICAM-1 is highly expressed on hepatocytes (61). ICAM-1 transmits an activating signal by activating not only Grb2 followed by MEK-1 and ERK but also Rho GTPase and subsequently JNK (62–64). Thus, a possible scenario is that immune cells expressing DNAM-1 and LFA-1 encounter with hepatocytes, and these molecules interact with inhibitory CD155 and activating ICAM-1 on hepatocytes, respectively. However, mechanistically whether ICAM-1 stimulates CXCL1 production and CD155 inhibits the production remain unaddressed. Although this study implies that CD155 on hepatocytes could be a therapeutic target for acute liver injury, further studies on the regulation of CXCL1 production and CD155-mediated inhibitory signals are needed.

The authors have no financial conflicts of interest.

We thank S. Tochihara, W. Saito, and H. Furugen for secretarial assistance. We are grateful to Günter Bernhardt (Hannover Medical School) for providing CD155-deficient mice.