AIH patients were diagnosed according to the criteria established by the International Autoimmune Hepatitis Group in 2008 (19). The clinical characteristics of AIH patients and healthy controls who provided peripheral blood samples are listed in Supplementary Table S1 .

All the AIH patients enrolled provided written informed consent, and the study was approved by the Ethics Committee of Renji Hospital.

Immunohistochemistry and immunofluorescence were performed using primary antibodies against LXRα (ab41902, Abcam, Cambridge, UK), CD11b (ab238794, Abcam), and CD33 (ab199432, Abcam), according to the procedures described previously (8). Specifically, liver frozen sections of AIH patients who received liver transplantation were used for immunostaining of LXRα. Redundant liver explants from healthy donors were used as controls.

For murine experiment, liver tissue samples were fixed in 10% neutral buffered formalin and embedded in paraffin wax. Liver sections (4 μm) were stained with hematoxylin and eosin (H&E) for histological evaluation.

Public single-cell RNA sequencing data of hepatic nonparenchymal cells from healthy controls (n = 5) and patients of liver cirrhosis (n = 5) were downloaded from GEO dataset (GSE136103) (20). Hepatic expression of NR1H3 (encoding LXRα) was analyzed using Seurat R package v3.2.1 (21). Different immune lineages were identified with CellMarker dataset (22).

Wild-type (WT) C57BL/6J were purchased from the Shanghai SLAC Laboratory Animal Co. Ltd. LXRα^−/− mice on a C57BL/6 background were kindly provided by professor Jun Pu (Division of Cardiology, Renji Hospital). All the mice were housed under specific pathogen-free (SPF) environment at the animal facility of Renji Hospital, School of Medicine, Shanghai Jiao Tong University. Female mice aged between 8 and 10 weeks were used. All animal experiments were reviewed and approved by the Institutional Animal Care and Use Committee of Shanghai Jiao Tong University.

To induce acute hepatitis, LXRα^−/− mice and WT controls were i.v. injected with PBS or 8–10 mg/kg ConA (Sigma-Aldrich, St. Louis, MO, USA), respectively. In an attempt to antagonize the activation of LXR, WT mice were given SR9243 (30 mg/kg, SelleckChem, Houston, TX, USA) intraperitoneally twice at 24 and 1 h before ConA treatment. Mice were killed 24 h following ConA challenge to examine tissue injury, serum alanine aminotransferase (ALT), and aspartate aminotransferase (AST).

Serum levels of interferon-γ (IFN-γ), tumor necrosis factor-α (TNF-α), and interleukin-6 (IL-6) were measured with Mouse Th1/Th2/Th17 Cytokine Kit (BD Bioscience, San Jose, CA, USA).

Hepatic mononuclear cells (HMNCs) were prepared as previously described (7). Briefly, the liver was diced and homogenized by passed through a 70-μm strainer (BD Bioscience, USA), and then resuspended in 33% Percoll (GE Healthcare, North Richland Hills, TX, USA). The suspension was centrifuged at 900×g for 30 min, and red blood cells (RBCs) were removed by RBC Lysing Buffer (Sigma-Aldrich, USA).

Single-cell suspension of HMNCs were isolated and freshly labeled with fluorochrome-conjugated antibodies, including antimouse CD45, CD11b, Gr-1, Ly6G, Ly6C, Ki67 (BD Bioscience), CD3, CD4, CD8, CD25, CD69, NK1.1, TCRβ (BioLegend, San Diego, CA, USA), and anti-IRF-8 (eBioscience, San Diego, CA, USA) antibodies. Antihuman LXR Alpha antibody used in flow cytometry was purchased from LSBio (LS-C223499, Seattle, WA, USA), and the rabbit IgG isotype control was from Novus (NBP2-36463APC, St. Louis, MO, USA). Cellular apoptosis was detected with FITC Annexin V Apoptosis Detection Kit (BD Bioscience). Intracellular staining was performed using the Fixation/Permeabilization Kit (BD Bioscience) and Transcription Factor Buffer Set (BioLegend) according to the manufacturer’s instructions. Flow cytometry was performed with LSR Fortessa (BD Bioscience), and data were analyzed using the FlowJo software version 10.0.2 (Three Star, San Carlos, CA, USA).

MDSCs were magnetic sorted from the liver of LXRα^−/− or WT mice following 16 h ConA injection with MDSC Isolation Kit (Miltenyi Biotec, Auburn, CA, USA). T cells were obtained from the spleen of WT mice using Pan T Cell Isolation Kit (Miltenyi Biotec, USA). T cells labeled with CFSE (Invitrogen, Waltham, MA, USA) were activated with anti-CD3/CD28 beads (Miltenyi Biotec, USA) and further cocultured with purified liver MDSCs. The proliferation of T cells was assessed after 72 h and then analyzed with Flowjo software.

MDSCs were purified from the bone marrow of LXRα^−/− mice and WT mice treated with ConA for 3 h. Subsequently, 5 × 10⁶ MDSCs/mouse were transferred through tail-vein injection, and recipient WT mice were treated with ConA 1 h later. Mice were killed 16 h following ConA challenge and assessed for liver histology and transaminase levels.

Bone marrow cells were cultured with recombinant murine GM-CSF (40 ng/ml, PeproTech, Rocky Hill, NJ, USA) and IL-6 (40 ng/ml, PeproTech) in RPMI 1640 supplemented with 10% heat-inactivated FBS, 10 mM HEPES, 1 mM penicillin-streptomycin, and 50 mM 2-mercaptoethanol for 4 days. For LXR activation, 1 μM GW3965 was added at days 0 and 3.

MDSCs were purified from bone marrow of WT mice (n = 3) and LXRα^−/− mice (n = 3) following 24 h injection of ConA with MDSC Isolation Kit (Miltenyi Biotec, USA). Total RNA was extracted from MDSCs using Trizol (Invitrogen). Transcriptome libraries were generated with the TruSeq RNA sample preparation kit (Illumina, San Diego, CA, USA), and sequencing was performed using the Illumina HiSeq X Ten instrument by the commercial service of Genergy Biotechnology Co. Ltd. (Shanghai, China).

Total RNA was extracted with TRIzol Reagent (Invitrogen, USA) and cDNA was synthesized with PrimeScript™ RT Reagent Kit (Takara, Japan). Real-time PCR was performed using TB Green^® Fast qPCR Mix (Takara, Japan) on a StepOnePlus™ (Applied Biosystems, Waltham, MA, USA). Gene expression was normalized to the level of β-actin mRNA. The sequences for the primers used are as follows: murine IRF-8, forward-5′-GATCGAACAGATCGACAGCA-3′, reverse-5′-GCTGGTTCAGCTTTGTCTCC-3′; and β-actin, forward -5′-CTAAGGCCAACCGTGAAAAG-3′, reverse-5′-GGTACGACCAGAGGCATACA-3′.

Primary antibodies applied in Western blot assay mainly include antibody against caspase-3 (#9662, Cell Signaling, Danvers, MA, USA), caspase-8 (#4790, Cell Signaling), PCNA (#13110, Cell Signaling), IRF-8 (#5628, Cell Signaling), and β-actin (#4970, Cell Signaling).

Briefly, 1 × 10⁴ HEK293T cells were transfected with 10 ng Renilla luciferase plasmid, 100 ng firefly luciferase plasmid pGL3-IRF-8, and 100 g pCAGPuroAS05-NR1H3 plasmid using FuGENE^® HD Transfection Reagent (Promega, Madison, WI, USA). After 48 h, cell luciferase was measured by Dual-Glo^® Luciferase Assay System (Promega). Firefly luciferase activity was normalized to Renilla luciferase.

All analyses were performed using GraphPad Prism 6.0 software. Data were presented as the mean ± standard error (SEM). Statistical differences were determined by unpaired two-tailed t-tests, and significance was defined as *p > 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001.

Human LXRα is known to upregulate its own expression upon activation. We explored the expression of this nuclear factor in liver tissue from AIH patients and healthy donors. Immunochemistry staining of the frozen sections revealed that expression of LXRα was substantially increased in AIH patients, compared with healthy controls ( Figure 1A ). We examined previously published single-cell RNA sequencing data of liver nonparenchymal cells (GSE136103) and found that LXRα tended to be highly expressed in the “myeloid cell” cluster ( Figure 1B ). Subsequently, we utilized confocal microscopy to investigate localization of LXRα and the surface markers of MDSCs in AIH patients, including CD11b and CD33. Indeed, LXRα was colocalized with CD11b and CD33 ( Figures 1C, D ). To further validate the cell source of LXRα, expression of LXRα in the peripheral blood from AIH patients and healthy donors was examined by flow cytometry. Consistently, LXRα was preferentially expressed in myeloid cells including MDSCs (HLA-DR^−/loCD11b⁺CD33⁺) ( Figure 1E ). Moreover, a higher expression of LXRα was observed in the circulating immune cells of AIH patients than healthy controls ( Figure 1F ).

To investigate the potential role of LXRα in MDSCs in vivo, wild-type (WT) and LXRα^−/− knockout (LXRα^−/−) mice were challenged with ConA, respectively. ConA-induced hepatitis has been widely used as murine model of AIH and elicits rapid recruitment of MDSCs to the liver (7). Interestingly, hepatic area of inflammation and necrosis were significantly attenuated in mice deficiency of LXRα following ConA injection ( Figures 2A, B ), along with decreased levels of ALT and AST ( Figure 2C ). Levels of peripheral inflammatory cytokines, including IFN-γ, TNF-α, and IL-6, increased after ConA induction but were markedly lower in LXRα^−/− mice than WT controls ( Figure 2D ), confirming an ameliorated immune response upon LXRα deletion. Notably, PMN-MDSCs (CD11b⁺Ly6G⁺Ly6C^lo cells) and M-MDSCs (CD11b⁺Ly6G⁻Ly6C^high cells) were substantially expanded in the liver of LXRα^−/− mice, compared with WT mice (27.3% vs. 9.42%, p < 0.01; 2.02% vs. 1.18%, p < 0.01; Figures 2E, F ). MDSCs were reported to have potent immunosuppressive capacity, which probably explain the mitigated inflammation and less tissue injury upon LXRα ablation. In line, a decreased activation of T lymphocytes was observed in LXRα^−/− mice ( Supplementary Figures S1A, B ). There was no significant difference with regard to the frequency of T-regulatory (Treg) cell or macrophage between WT and LXRα^−/− mice ( Supplementary Figures S1C, D ). Subsequent antagonizing LXRα with SR9243 also resulted in an increased accumulation of hepatic MDSCs and simultaneously ameliorated liver injury ( Supplementary Figure S2 ).

To further explore the mechanisms of MDSC accumulation in LXRα^−/− mice, we examined the effects of LXRα knockout on proliferation and apoptosis of MDSCs. Intriguingly, CD11b⁺Gr-1⁺ MDSCs in the liver of LXRα^−/− ConA group possessed significantly higher frequency of ki67-positive cells than WT controls (92.7% vs. 76.5%, p < 0.01, Figures 3A, B ), which was supported by elevated expression of PCNA in MDSCs of LXRα^−/− mice treated with ConA ( Figure 3C ). It is known that peripheral MDSCs are prone to programmed cell death. In concordance, MDSCs in the liver of WT mice group exhibited substantial apoptosis as early as 3 h following ConA challenge, which further upregulated at the time point of 6 h. Conversely, MDSCs in LXRα^−/− ConA group showed much lower frequency of cell apoptosis, both at 3 and 6 h (19.45% vs. 47.81%, p < 0.001; 38.82% vs. 88.75%, p < 0.001, Figures 3D, E ). Western blot assay confirmed an excessive activation of apoptosis signaling pathway in the MDSCs of WT group, as evidenced by the cleavage of caspase-8 and caspase-3 ( Figure 3F ). In parallel, MDSCs treated with LXR agonist (GW3965) in vitro were more susceptible to cell death induced by TNF-α (p < 0.01, Figure 3G ).

To confirm the immunosuppressive effects of MDSCs in ConA-induced hepatitis, we isolated the hepatic MDSCs from LXRα^−/− and WT mice respectively and cocultured the MDSCs with T cells activated by anti-CD3/CD28 beads at different effector-and-target ratios. As expected, MDSCs purified from both LXRα^−/− and WT groups effectively suppressed the proliferation of T cells at both 1:3 and 1:10 ratios (MDSC:T cell) ( Figures 4A, B ). More importantly, ConA-induced MDSCs in LXRα^−/− mice exhibited slightly higher immunosuppressive capacity than that of WT controls ( Figure 4B ). In the next adoptive transfer experiment, mice were protected from ConA-mediated hepatitis by prior transfer of MDSCs from both WT and LXRα^−/− mice ( Figures 4C–E ). All the above findings support that MDSCs exert potent immunoregulatory role under LXRα knockout background and thereby efficiently protect liver from immune-mediated tissue injury.

IRF-8 has been well characterized as a key factor during the differentiation and maturation of myeloid cells. Mice defect in IRF-8 generate massive amount of MDSCs, while overexpression of IRF-8 led to depletion of MDSCs in murine models of carcinoma, indicating IRF-8 as a negative regulator in MDSC biology (3, 23, 24). By transcriptome sequencing of MDSCs isolated following ConA treatment, we noticed that the expression of IRF-8 in LXRα^−/− mice was significantly lower than its WT counterparts ( Figure 5A ). Additionally, S100A8 and S100A9, transcription factors known to induce MDSC differentiation, were upregulated in LXRα^−/− group. Lower expression of IRF-8 mRNA in the MDSCs from LXRα^−/− mice was validated by quantitative PCR ( Figure 5B ). Flow cytometry confirmed that hepatic MDSCs in LXRα^−/± mice exhibited much lower level of IRF-8 than WT mice challenged with ConA. However, such difference was not observed in spleen MDSCs ( Figures 5C–E ).

We next use cytokines to induce MDSCs from bone marrow cells. LXR agonist resulted in an impairment of MDSC generation, particularly PMN-MDSCs ( Figures 5F, G ). As expected, expression of IRF-8 decreased markedly over the 4-day induction by GM-CSF and IL-6. Consistent with the in vivo data, lower expression of IRF-8 was detected in bone marrow (BM)-derived MDSCs from LXRα^−/− mice than WT counterpart, whereas WT MDSCs treated with LXR agonist showed upregulation of IRF-8 expression ( Figure 5H ). Next, dual-luciferase reporter assay was conducted to identify functional interactions between LXRα and the promoter of IRF-8. Overexpression of LXRα led to twofold increase of the luciferase activity in HEK293T cells 48 h after transfection ( Figure 5I ), further supporting the transcriptional regulation of IRF-8 by LXRα.