Advertisement

Transcriptional switch of hepatocytes initiates macrophage recruitment and T-cell suppression in endotoxemia

Open AccessPublished:March 08, 2022DOI:https://doi.org/10.1016/j.jhep.2022.02.028

      Highlights

      • The periportal zone of the hepatic lobules plays an essential role in the emergence of endotoxemia-associated liver injury.
      • The reprogramming of hepatocytes into a proinflammatory subpopulation is key in the endotoxemia-induced immune response.
      • Proinflammatory hepatocytes recruit circulating monocytes to form RMs via CCL2-CCR2 during endotoxemia.
      • RMs induce inhibitory ligand expression by hepatocytes, leading to T lymphocyte suppression partly via PD-1/PD-L1 interactions.

      Background & Aims

      The liver plays crucial roles in the regulation of immune defense during acute systemic infections. However, the roles of liver cellular clusters and intercellular communication in the progression of endotoxemia have not been well-characterized.

      Methods

      Single-cell RNA sequencing analysis was performed, and the transcriptomes of 19,795 single liver cells from healthy and endotoxic mice were profiled. The spatial and temporal changes in hepatocytes and non-parenchymal cell types were validated by multiplex immunofluorescence staining, bulk transcriptomic sequencing, or flow cytometry. Furthermore, we used an adeno-associated virus delivery system to confirm the major mechanisms mediating myeloid cell infiltration and T-cell suppression in septic murine liver.

      Results

      We identified a proinflammatory hepatocyte (PIH) subpopulation that developed primarily from periportal hepatocytes and to a lesser extent from pericentral hepatocytes and played key immunoregulatory roles in endotoxemia. Multicellular cluster modeling of ligand-receptor interactions revealed that PIHs play a crucial role in the recruitment of macrophages via the CCL2-CCR2 interaction. Recruited macrophages (RMs) released cytokines (e.g., IL6, TNFα, and IL17) to induce the expression of inhibitory ligands, such as PD-L1, on hepatocytes. Subsequently, RM-stimulated hepatocytes led to the suppression of CD4+ and memory T-cell subsets partly via the PD-1/PD-L1 interaction in endotoxemia. Furthermore, sinusoidal endothelial cells expressed the highest levels of proapoptotic and inflammatory genes around the periportal zone. This pattern of gene expression facilitated increases in the number of fenestrations and infiltration of immune cells in the periportal zone.

      Conclusions

      Our study elucidates unanticipated aspects of the cellular and molecular effects of endotoxemia on liver cells at the single-cell level and provides a conceptual framework for the development of novel therapeutic approaches for acute infection.

      Lay summary

      The liver plays a crucial role in the regulation of immune defense during acute systemic infections. We identified a proinflammatory hepatocyte subpopulation and demonstrated that the interactions of this subpopulation with recruited macrophages are pivotal in the immune response during endotoxemia. These novel findings provide a conceptual framework for the discovery of rational therapeutic targets in acute infection.

      Graphical abstract

      Keywords

      Linked Article

      • Untangling the web: The complex parenchymal-immune interface in endotoxemia
        Journal of HepatologyVol. 77Issue 2
        • Preview
          Despite advances in antibiotics and critical care, sepsis remains a significant cause of mortality and morbidity worldwide, with an estimated 300 cases per 100,000 people in each year, 30% of whom will not survive the condition.1 Sepsis is the most costly condition in the USA, accounting for 5.2% of total hospital costs. Sepsis arises from systemic immune dysregulation in response to bacterial infection, or bacterial endotoxins, causing multiple organ-failure and death. One of the hallmarks of sepsis is a rapid decline in liver function.
        • Full-Text
        • PDF
      See Editorial, pages 296–298

      Introduction

      The liver plays a crucial role in the regulation of immune defense during acute systemic infections.
      • Guo J.
      • Li Y.
      • Shan Y.
      • Shu C.
      • Wang F.
      • Wang X.
      • et al.
      Humanized mice reveal an essential role for human hepatocytes in the development of the liver immune system.
      ,
      • Kubes P.
      • Jenne C.
      Immune responses in the liver.
      Under physiological conditions, the liver is more highly exposed to endotoxins and microorganisms than other organs in the body due to its unique anatomical location; thus, the liver generally stays in a tolerogenic immune state to avoid excessive immune activation and autoimmune responses.
      • Thomson A.W.
      • Knolle P.A.
      Antigen-presenting cell function in the tolerogenic liver environment.
      ,
      • Wang J.
      • Ma J.
      • Nie H.
      • Zhang X.J.
      • Zhang P.
      • She Z.G.
      • et al.
      Hepatic regulator of G protein signaling 5 ameliorates nonalcoholic fatty liver disease by suppressing transforming growth factor beta-activated Kinase 1-c-Jun-N-Terminal Kinase/p38 signaling.
      During endotoxemia, the liver transitions from tolerant to an active immune state and forms the first line of defense against invading microorganisms.
      • Strnad P.
      • Tacke F.
      • Koch A.
      • Trautwein C.
      Liver - guardian, modifier and target of sepsis.
      ,
      • Yan J.
      • Li S.
      • Li S.
      The role of the liver in sepsis.
      This immune state transition in the liver triggers efficient elimination and clearance of microbial products from the circulation; however, an overwhelming inflammatory response results in liver injury and immune suppression, which may predict a negative prognosis for endotoxemia.
      • Strnad P.
      • Tacke F.
      • Koch A.
      • Trautwein C.
      Liver - guardian, modifier and target of sepsis.
      ,
      • Zhang L.
      • Tian R.
      • Yao X.
      • Zhang X.J.
      • Zhang P.
      • Huang Y.
      • et al.
      Milk fat globule-epidermal growth factor-factor 8 improves hepatic steatosis and inflammation.
      The liver defense mechanism against invading microorganisms relies on a complex network of immune cells, such as Kupffer cells (KCs), dendritic cells (DCs), T and B lymphocytes, neutrophils, and natural killer (NK) cells.
      • Mao K.
      • Baptista A.P.
      • Tamoutounour S.
      • Zhuang L.
      • Bouladoux N.
      • Martins A.J.
      • et al.
      Innate and adaptive lymphocytes sequentially shape the gut microbiota and lipid metabolism.
      Moreover, non-immune cells, such as hepatocytes and liver sinusoidal endothelial cells (LSECs), recruit proinflammatory immune cells that exert immune functions by producing acute-phase proteins, complement factors, inflammatory cytokines, and cell adhesion molecules.
      • Davies S.P.
      • Terry L.V.
      • Wilkinson A.L.
      • Stamataki Z.
      Cell-in-Cell structures in the liver: a Tale of four E's.
      • Megahed F.A.K.
      • Zhou X.
      • Sun P.
      The interactions between HBV and the innate immunity of hepatocytes.
      • Halpern K.B.
      • Shenhav R.
      • Massalha H.
      • Toth B.
      • Egozi A.
      • Massasa E.E.
      • et al.
      Paired-cell sequencing enables spatial gene expression mapping of liver endothelial cells.
      Although innate immune responses have been suggested to be essential in liver defense mechanisms, the spatial and temporal responses of hepatic immune and non-immune cells are not fully understood. The burgeoning field of single-cell transcriptomics enables us to decipher cell heterogeneity, differentiation, and intercellular communication during the pathogenic process of endotoxemia. In addition, accumulating evidence has shown that a number of genes expressed in hepatocytes and endothelial cells (ECs) are zonated, with specific functional specializations matching the local environment of the hepatic lobule.
      • Braeuning A.
      • Ittrich C.
      • Kohle C.
      • Hailfinger S.
      • Bonin M.
      • Buchmann A.
      • et al.
      Differential gene expression in periportal and perivenous mouse hepatocytes.
      ,
      • Gebhardt R.
      • Matz-Soja M.
      Liver zonation: novel aspects of its regulation and its impact on homeostasis.
      These zonation characteristics of hepatocytes and ECs enable the identification of spatial features of gene transcription and expression during the progression of endotoxemia.
      • Halpern K.B.
      • Shenhav R.
      • Massalha H.
      • Toth B.
      • Egozi A.
      • Massasa E.E.
      • et al.
      Paired-cell sequencing enables spatial gene expression mapping of liver endothelial cells.
      ,
      • Halpern K.B.
      • Shenhav R.
      • Matcovitch-Natan O.
      • Toth B.
      • Lemze D.
      • Golan M.
      • et al.
      Single-cell spatial reconstruction reveals global division of labour in the mammalian liver.
      ,
      • Gola A.
      • Dorrington M.G.
      • Speranza E.
      • Sala C.
      • Shih R.M.
      • Radtke A.J.
      • et al.
      Commensal-driven immune zonation of the liver promotes host defence.
      In this study, we identified the cellular landscape of hepatic cells in response to a “double-hit” endotoxemia model at 3 different stages in mice. Single-cell transcriptomic analyses revealed that phenotypic transition of hepatocytes in response to endotoxemia plays central roles in the recruitment of circulating monocytes to establish a recruited macrophage (RM) population and the inhibition of T lymphocytes. Therefore, we identified the specific roles of liver cellular clusters and intercellular communication pathways in the progression of endotoxemia.

      Materials and methods

      For detailed information on animal models, reagents, methodology for scRNA-seq data processing and bioinformatics analysis, tissue harvesting, isolation and culture of primary cells used, cellular functional measurements, please refer to the CTAT method tables and the supplementary information.

      Results

      Single-cell RNA-seq defines hepatic cellular composition and alterations during endotoxemia

      To investigate changes in the cellular composition and heterogeneity in the liver during endotoxemia, we performed scRNA-seq analysis of liver hepatocytes and non-parenchymal cells from mice with “double-hit” endotoxemia as previously established (Fig. 1A).
      • Hagar J.A.
      • Powell D.A.
      • Aachoui Y.
      • Ernst R.K.
      • Miao E.A.
      Cytoplasmic LPS activates caspase-11: implications in TLR4-independent endotoxic shock.
      A total of 56,828 hepatocytes and non-parenchymal cells at 3 time points were further analyzed after passing quality control metrics. We identified 9 major cellular clusters in control, 3-hour lipopolysaccharide (LPS) and 12-hour LPS livers, including hepatocytes (Alb, Saa1, Apoc1, Mup20), T cells (Cd3d, Cd8b1, Gzmb, Gzma), NK cells (Nkg7, Klrk, Klre1, Xcl1), B cells (Cd79a, Fcmr, Cd79b, Ebf1), neutrophils (S100a8, Retnlg, S100a9, Mmp9), macrophages (Wfdc17, Csf1r, Sdc3, Ifitm2), KCs (Stmn1, Top2a, Ube2c, Ccnb2), ECs (Igfbp7, Ptprb, Clec4g, Kdr), and DCs (Ly6d, Siglech, Rnase6, H2-Ab1) (Fig. 1B-C, E-F; Table S1). Compared to control, endotoxemia induced significant proportional alterations in cell clusters (Fig. 1B-D). Among the 9 clusters, 3 clusters, namely, neutrophils, macrophages, and KCs, developed from myeloid cell lineage and were the most abundantly expanded during endotoxemia. The proportions of ECs, hepatocytes and T lymphocytes were decreased in the liver in an LPS exposure time-dependent manner (Fig. 1B-D). Deconvolution of LPS-treated bulk RNA-seq liver samples using the available references yielded consistent proportional changes in the immune cell populations in the liver during endotoxemia (Fig. 1G-H).
      • Mascaux C.
      • Angelova M.
      • Vasaturo A.
      • Beane J.
      • Hijazi K.
      • Anthoine G.
      • et al.
      Immune evasion before tumour invasion in early lung squamous carcinogenesis.
      Using flow cytometry (FCM), we also observed that the endotoxemia-induced proportional changes in neutrophils, macrophages, KCs, B cells, ECs, and hepatocytes were consistent with the trends that we observed by scRNA-seq (Fig. 1I-J). Representative FCM plots and gating strategies are shown in Fig. S11. However, under LPS challenge, a decrease in the proportion of T cells was not observed by FCM. This result might be explained by the significant reduction in the proportion of ECs causing an increase in T lymphocytes among the overall non-parenchymal cell populations. It is important to test the generality of this finding in other models of acute infection; thus, we examined changes in hepatic cellular composition in a cecal ligation and puncture (CLP)-induced sepsis model and in septic patients (Fig. S1).
      • Snell L.M.
      • Xu W.
      • Abd-Rabbo D.
      • Boukhaled G.
      • Guo M.
      • Macleod B.L.
      • et al.
      Dynamic CD4(+) T cell heterogeneity defines subset-specific suppression and PD-L1-blockade-driven functional restoration in chronic infection.
      In the CLP-induced septic model, the aforementioned changes in hepatic cellular composition were consistently observed using FCM (Fig. S1A-C). Deconvolution of bulk RNA-seq of liver from septic patients (GSE141864) by Brusletto et al.’s group also revealed similar discoveries, showing that the sum of cells derived from myeloid cell lineage among hepatic immune cells was increased by 150.14% in septic patients compared to non-infected patients. The percentage of hepatic T cells decreased by 42.0% among immune cells in patients with sepsis (Fig. S1D-F).
      Figure thumbnail gr1
      Fig. 1Single-cell RNA-seq defines the hepatic cellular composition and alterations in endotoxemia.
      (A) Experimental scheme depicts the murine endotoxemia model and tissue harvesting for RNA-seq, flow cytometric assay, and H&E and immunofluorescence staining. Liver hepatocytes and non-parenchymal cells were mixed for bulk RNA sequencing and were separately harvested, purified and subjected to scRNA-seq (n = 6 mice per group). (B) UMAP plot displays the major 9 color-coded cell clusters in the liver with or without endotoxemia. There were 7,173, 6,750, 5,872 hepatocytes and 11,753, 12,211, 13,069 non-parenchymal cells in the control, LPS 3-hour, LPS 12-hour group, respectively. (C) Pie charts display the proportion of each cell type in overall liver cells by scRNA-seq. (D) The percentage change ratio (percentage change over previous time point) of each cell type with LPS challenges. (E) Heatmap shows the average expression levels (color-scaled, row-wise Z scores) of the top 20 DEGs (rows) across the cell clusters (columns). Four signature genes of each cluster are displayed on the right. (F) Violin plots show the normalized log-transformed expression levels of representative genes (rows) used to define each cell type (columns). (G-H) Bar charts and quantification show the deconvolution analysis results at the indicated time point from liver bulk RNA-seq data. (I-J) Bar charts and quantification show the proportions of major cell types among all immune cells at the indicated time point by liver flow cytometry. DC, dendritic cell; DEGs, differentially expressed genes; EC, endothelial cell; KC, Kupffer cell; LPS, lipopolysaccharide; NK, natural killer; scRNA-seq, single-cell RNA sequencing; UMAP, uniform manifold approximation and projection. (This figure appears in color on the web.)

      Endotoxemia induces macrophage recruitment and neutrophil accumulation in the liver

      Previous studies have highlighted liver macrophage populations that orchestrate the progression of endotoxemia or sepsis.
      • Knolle P.A.
      • Gerken G.
      Local control of the immune response in the liver.
      We further grouped liver macrophages into 3 clusters, which were annotated as RMs, bone marrow (BM)-derived KCs, and yolk sac (YS)-derived KCs (Fig. 2A). RMs were enriched in the expression of Cx3cr1, Fabp5, Ms4a7, and Clec4b1; BM-derived KCs were highly enriched in the expression of Vsig4, Folr2, and Clec4f; and YS-derived KCs were differentiated by high expression of Hmox1, Timd4, Ank2, and Cd163 (Fig. 2C; Fig. S2A; Table S2). BM-derived and YS-derived KCs accounted for up to 94.34% of liver macrophages under healthy conditions (Fig. 2B). These 2 KC subpopulations play important roles in host defense responses to bacteria. YS-derived KCs were specifically responsible for the response to interferon-β (INF-β) (Fig. 2D-E). BM-derived KCs were expanded during infection and liver injury and exhibited a more robust bacterial pathogen uptake and clearance capability than YS-derived KCs (Fig. 2D-E).
      • Beattie L.
      • Sawtell A.
      • Mann J.
      • Frame T.C.M.
      • Teal B.
      • de Labastida Rivera F.
      • et al.
      Bone marrow-derived and resident liver macrophages display unique transcriptomic signatures but similar biological functions.
      Endotoxemia induced significant enrichment in RMs in the liver in a time-dependent manner (Fig. 2B). The RMs cluster expressed Cx3cr1, Fabp5, and Clec4b1 and played essential roles in secreting cytokines and chemokines and recruiting lymphocytes and circulating myeloid cells (Fig. 2C-E; Fig. S2A). We established an endotoxemia model with CX3CR1gfp/gfp reporter mice (Fig. 2F). This reporter gene does not label hepatic tissue macrophages in adult mice and thus serves as a tool to distinguish RMs and KCs in the liver.
      • Yona S.
      • Kim K.W.
      • Wolf Y.
      • Mildner A.
      • Varol D.
      • Breker M.
      • et al.
      Fate mapping reveals origins and dynamics of monocytes and tissue macrophages under homeostasis.
      ,
      • Daemen S.
      • Gainullina A.
      • Kalugotla G.
      • He L.
      • Chan M.M.
      • Beals J.W.
      • et al.
      Dynamic shifts in the composition of resident and recruited macrophages influence tissue remodeling in NASH.
      The findings of this experiment confirmed the abundance of RM (gfp+) infiltration in endotoxic livers. More importantly, we discovered that the enriched RMs were distributed adjacent to the portal vein region in the liver after LPS challenge, as shown by immunofluorescence staining (Fig. 2G-I). Surprisingly, we also found that RMs were the major cellular subtype responsible for synthesizing an array of programmed cell death-ligand 1 (PD-L1)-inducible factors in the early stage of endotoxemia, such as IL-17 (Il17), IL-27 (Ebi3), and IL-6 (Il6) (Fig. 2J-N; Fig. S2B-G).
      Figure thumbnail gr2
      Fig. 2Endotoxemia induces macrophage recruitment and neutrophil accumulation in the liver.
      (A) t-SNE plots display color-coded cell clusters of myeloid cells in the liver with or without endotoxemia. There were 700, 1,023, 1,212 cells in the control, LPS 3-hour, LPS 12-hour group, respectively. (B) Bar charts show the proportion of major myeloid cell subtypes among all myeloid cells at different time points by scRNA-seq. (C) Violin plots show the normalized log-transformed expression levels of representative genes (rows) used to define RMs, BM-derived KCs, and YS-derived KCs. (D) The heatmap shows the correlation between WGCNA-identified 5-gene co-expression modules and the indicated cell clusters. (E) GO analysis shows the upregulated signaling pathways across the 5 modules. (F) Experimental schematic of the endotoxemia model established with CX3CR1gfp/gfp mice. (G) Representative images show GFP+ RMs and F4/80+CLEC4F+ KCs located in different zones of the hepatic lobule. White arrows indicate F4/80+CLEC4F+ KCs. Yellow arrows indicate GFP+ RMs. (H) Quantification of the proportion of GFP+ RMs in total cells. (I) Quantification of the ratio of the total number of GFP+ RMs in central vein to periportal ROIs for the periportal and central vein lobular areas. (J) The PD-L1 inducible factors score in the indicated groups. (K) The expression of PD-L1 inducible factors in myeloid cell subgroups in the indicated groups. (L) Quantification of the percentages of the IL-6+ (top) and IL-17+ (bottom) GFP+ RMs among total cells by immunofluorescence images. (M) Quantifications of the percentages of IL-6+ or IL-17+ F4/80+ macrophages (left) and IL-6+ or IL-17+ CLEC4F+ KCs (right) in total cells by immunofluorescence staining. (N) Quantification of the percentages of IL-6+ (top) and TNF-α+ (bottom) GFP+ RMs in NPC by flow cytometry (n = 6 mice per group). n≥20 fields per group from 4 mice, 5 fields per mouse in H-I, L-M. Data shown are mean ± SD (J-N). Data were first analyzed and passed the normality test (Shapiro-Wilk test in H-I, L-N). P values were shown and assessed by one-way ANOVA with Tukey’s test in H-I, L-N. BM, bone marrow; CV, pericentral vein; GO, gene ontology; LPS, lipopolysaccharide; NPC, non-parenchymal cells; PV, periportal vein; RMs, recruited macrophages; ROI, regions of interest; t-SNE, t-distributed stochastic neighbor embedding; WGCNA, weighted gene co-expression network analysis; YS, yolk sac. (This figure appears in color on the web.)
      As immune cells play a key role in infection, neutrophils are one of the main cell types that defend against and eliminate invading pathogens. Our results revealed that neutrophils accumulated abundantly in the early stage of endotoxemia in the liver and released various cytokines and chemokines (Fig. S8A, B, I). These chemokines then promoted the migration of neutrophils to specific liver sites to further exert anti-infection effects. Specific signaling pathways were activated in neutrophils, as shown by gene ontology (GO) analysis (Fig. S8B). Neutrophils from the control group were mainly grouped into 1 cluster (Neutrophil; Fig. S8C-D), while neutrophils were segregated into 2 additional clusters, namely, early neutrophil extracellular trap (NET)-associated neutrophils and late NET-associated neutrophils, in LPS 3-hour and LPS 12-hour groups (Fig. S8C-D). Early NET-associated neutrophils were enriched in genes involved in endoplasmic reticulum stress (Atf3, Atf4, and Bax), ERK1/2 signaling cascade (Cd74, C3, and Hmgb1), reactive oxygen species synthesis (Ptgs2, Arg2, and Il1b), and oxidative stress (Rps3, Prdx6, and Jun) (Fig. S8E, G-H). Late NET-associated neutrophils were enriched in processes, such as complement receptor-mediated signaling pathways (Fpr1 and Fpr2) and response to interferon (Ifitm1, Ifitm3, and Ifi47) (Fig. S8E, G-H).
      • Radermecker C.
      • Sabatel C.
      • Vanwinge C.
      • Ruscitti C.
      • Marechal P.
      • Perin F.
      • et al.
      Locally instructed CXCR4(hi) neutrophils trigger environment-driven allergic asthma through the release of neutrophil extracellular traps.
      The transcriptomic profiles related to NET formation were upregulated in a time-dependent manner across neutrophils (Fig. S8F-H) and NET formation was validated by SYTOX Green staining using FCM (Fig. S8I-J).

      Endotoxemia alters T-lymphocyte transcriptional profiles and suppresses T-cell functions

      Single-cell mapping of liver cells showed 6 distinct T-lymphocyte subsets in control, 3-hour LPS and 12-hour LPS livers, including CD3+ T cells, CD4+CD8+ T cells, CD4+ T cells, CD8+ T cells, CD4+ tissue-resident memory (Trm) T cells, and natural killer T (NKT) cells, and an NK-cell subpopulation (Fig. 3A, D; Fig. S3A). Endotoxemia induced changes in T-lymphocyte composition in mice, causing significant reductions in CD4+ T cells, Trm cells and NKT cells from the early phase (Fig. 3B-C). These cells play roles in regulating adaptive immunity and lymphocyte proliferation (Fig. S3B). There were slight increases in the numbers of CD3+ T cells, CD4+CD8+ T cells, and CD8+ T cells in the early phase, but the numbers were reduced in the later phase of endotoxemia (Fig. 3B-C). These cells mainly functioned as cytotoxic cells during endotoxemia (Fig. S3B). NK cells are derived from the lymphatic lineage; however, NK cells function as a component of the innate immune system,
      • Zhang Z.
      • Zhang S.
      • Zou Z.
      • Shi J.
      • Zhao J.
      • Fan R.
      • et al.
      Hypercytolytic activity of hepatic natural killer cells correlates with liver injury in chronic hepatitis B patients.
      and the number of NK cells increased in the liver after exposure to LPS (Fig. 3B-C). The proportional changes in CD3+ T cells, CD4+ T cells, CD8+ T cells, and NK cells, were confirmed by FCM (Fig. 3E; Fig. S3C). Consistently, decreased percentages of T cells among immune cells in the liver and among peripheral blood mononuclear cells were observed in patients with sepsis by RNA-seq (Fig. S1D-J). As shown by subtype analysis of NK cells, the proportion of resting NK cells decreased while that of cytotoxic NK cells increased significantly in the liver after exposure to LPS (Fig. S9A-D). The signaling pathway used by cytotoxic NK cells to produce tumor necrosis factors was significantly activated after LPS stimulation (Fig. S9E), as the prime for killing of bacteria.
      • Giannikopoulos G.
      • Antonopoulou A.
      • Kalpakou G.
      • Makaritsis K.
      • Panou C.
      • Papadomichelakis E.
      • et al.
      The functional role of natural killer cells early in clinical sepsis.
      The transcriptomic profiles related to cell proliferation, apoptosis, and effector production were upregulated in a time-dependent manner in NK-cell subsets (Fig. 3F; Fig. S9F).
      Figure thumbnail gr3
      Fig. 3Endotoxemia alters T-lymphocyte transcriptional profiles and suppresses T-cell functions.
      (A) t-SNE plots display color-coded cell clusters of NK and T cells in the liver with or without endotoxemia. There were 6,037, 6,333, 6,402 cells in the control, LPS 3-hour, LPS 12-hour group. (B) Bar charts show the number of T-cell subtypes and NK cells at different time points by scRNA-seq. (C) The percentage change ratio of each cell type. (D) Heatmap indicates the average expression (color-scaled, row-wise Z scores) of the top 10 DEGs (rows) across the cell clusters (columns). Four signature genes of each cluster were displayed on the right. (E) Quantification of the ratio of total T cell (CD3+), CD4+ T cell (CD3+CD4+), CD8+ T cell (CD3+CD8+) among total NPC by flow cytometry (n = 6 mice per group). (F) Violin plots show the proliferation, apoptosis, antigen-presenting, effect and suppression scores in the indicated groups. (G) Violin plots show the gene expression of immune-inhibitory genes in the indicated groups. (H) Cell viability of CD4+ T cells purified from the liver tissue of mice given control or LPS 3-hour treatment by CCK-8 assay. Purified CD4+ T cells were co-stimulated with anti-CD3, anti-CD28, and IL-2 for 24 hours before CCK-8 assay (n = 6 mice per group). (I) Quantification of the percentages of PD-1+, Ki67+, IL-2+, IL-4+, and TNF-α+ CD4+ cells in total CD4+ T cells by flow cytometry. (J-K) Feature plots and corresponding violin plots show the expression scores (z-scores) of Cd274 across different cell clusters. Data shown are mean ± SD (E, H-I). Data were first analyzed and passed the normality test (Shapiro-Wilk test in E, H-I). P values were shown and assessed by one-way ANOVA with Tukey’s test in E and unpaired 2-tailed t tests in H-I. DEGs, differential expressed genes; NPC, non-parenchymal cells; Trm, tissue-resident memory T cell; t-SNE, t-distributed stochastic neighbor embedding. (This figure appears in color on the web.)
      We further deciphered the functional changes in T-cell subsets. Transcriptomic profiles related to cell proliferation, apoptosis, antigen presentation, and effector production were upregulated in a time-dependent manner across all T-lymphocyte subsets (Fig. S3D; Fig. 3F). T-lymphocyte dysfunction and suppression are hallmarks of severe infections in animal models and humans.
      • Venet F.
      • Monneret G.
      Advances in the understanding and treatment of sepsis-induced immunosuppression.
      Notably, we observed that immunosuppressive genes were distinctively upregulated in CD4+ T cells and Trm cells at the 3-hour time point, which was associated with a dramatic reduction in the numbers of CD4+ T cells and Trm cells at this time point (Fig. 3F). Therefore, we hypothesized that the inhibition of CD4+ T cells and Trm cells is mediated by immunosuppressive genes. Reference immunosuppressive pathways can be grouped into 3 main categories: cell surface inhibitory receptors (such as Pdcd1), soluble inhibitory factors (such as Il10), and immunoregulatory cell types (such as regulatory T cells and other cells).
      • Wherry E.J.
      T cell exhaustion.
      We identified liver T lymphocytes, particularly CD4+ T cells and Trm cells, during endotoxemia that primarily express high levels of inhibitory receptor genes, including Pdcd1, Lag3, Sh2d1a, Cd160, and Ctla4, at the early time point (Fig. 3G).
      In the CLP-induced sepsis model, we observed upregulated Cd274 (encoding PD-L1) in hepatocytes by quantitative reverse-transcription PCR (RT-qPCR) (Fig. S4I) and FCM (Fig. S4J-K). The expression of CD274 was upregulated in liver tissue from septic patients (Fig. S3E). Consistent with the observation in endotoxemia, the mRNA expression of immune-inhibitory receptor genes, including PDCD1, LAG3, SH2D1A, CD160, and CTLA4, was found to be upregulated in liver tissue (GSE141864) by Brusletto et al.’s group and CD4+ T cells in peripheral blood mononuclear cells from septic patients (GSE167363 and GSE175453) (Fig. S3E-G).
      • Qiu X.
      • Li J.
      • Bonenfant J.
      • Jaroszewski L.
      • Mittal A.
      • Klein W.
      • et al.
      Dynamic changes in human single-cell transcriptional signatures during fatal sepsis.
      ,
      • Darden D.B.
      • Dong X.
      • Brusko M.A.
      • Kelly L.
      • Fenner B.
      • Rincon J.C.
      • et al.
      A novel single cell RNA-seq analysis of non-myeloid circulating cells in late sepsis.
      Co-expression of inhibitory receptors can substantially impair T-lymphocyte function, such as the production of IL2, IL4, and TNFα (Fig. 3I), and induce a reduction in cell number (Fig. 3B-C, E-G). Consistently, CD4+ T cells isolated from liver tissue in mice with LPS exposure expressed significantly higher levels of inhibitory receptors (such as PD-1) (Fig. 3I), and anti-CD3, anti-CD28 and IL-2 co-stimulated T-cell proliferation was significantly inhibited in CD4+ T cells from mice exposed to LPS (Fig. 3H).
      • Chang D.Y.
      • Song S.H.
      • You S.
      • Lee J.
      • Kim J.
      • Racanelli V.
      • et al.
      Programmed death-1 (PD-1)-dependent functional impairment of CD4(+) T cells in recurrent genital papilloma.
      Among the examined inhibitory receptors and their paired ligands, the PD-1/PD-L1 pair (encoded by the Pdcd1/Cd274 genes, respectively) is the most well-characterized and can be targeted by FDA-approved therapeutic reagents. Notably, single-cell mapping revealed that hepatocytes expressed the highest level of Cd274 among all cell types (encoded PD-1) at 3 hours after endotoxemia, which was consistent with the expression pattern of Pdcd1 on T lymphocytes and the severity of T-cell inhibition (Fig. 3J-K).

      Innate and adaptive B-cell immunity is activated in endotoxemia and accompanied by spatial adaptation

      To examine the transcriptional dynamics and function of liver B/plasma cells in endotoxemia, we next separated B/plasma cells into 5 subclusters (Fig. S5A): antigen-presenting B (APB) cell (Pdia4 and Myc); mature B cell (Cd19 and Cd38); plasma B cell (Ly6c2 and Jchain); memory B cells (Plp2 and Maml2); and pre/plasmablast B cells (Birc5 and Hmgb2) (Fig. S5D). Among B-cell subclusters, pre-plasmablasts were the only subpopulation with significant proportional alterations during endotoxemia (Fig. S5B-C).
      Each B-cell subset plays a regulatory role in the immune response in the septic liver, and pre/plasmablasts possess strong proliferative abilities (Fig. S5E). The total number of B cells (CD45+CD19+) gradually increased in the livers of septic mice, and this result was also confirmed by FCM (Fig. S5G). Furthermore, to understand the capacity of each B-cell subset to produce antibodies to resist infection, we compared immunoglobulin expression in our scRNA-seq data (Fig. S5F). APB cells and pre/plasmablasts secreted microscale antibodies. Mature B cells, memory B cells, and plasma B cells mainly secreted IgM and gradually increased IgM production during the process of endotoxemia (Fig. S5F). It has been reported that B cells are involved in protective early innate immune responses in an antibody-independent manner.
      • Kelly-Scumpia K.M.
      • Scumpia P.O.
      • Weinstein J.S.
      • Delano M.J.
      • Cuenca A.G.
      • Nacionales D.C.
      • et al.
      B cells enhance early innate immune responses during bacterial sepsis.
      Therefore, we next performed scoring for innate immunity, cytokine production, chemokine production, antigen presentation, antibody production, and proliferation and then compared different time points and B-cell subpopulations (Fig. S5H-I). APB cells mainly played roles in antigen presentation, while the ability of mature B cells and plasma cells to secrete cytokines and chemokines was enhanced at the 3-hour time point. The proliferative ability of each B-cell subgroup in the liver during endotoxemia was quite enhanced, but it was notable that pre-plasmablasts proliferated rapidly in the early stage of endotoxemia. This fast-proliferating B-cell subset generated early in infection might result from T cell-independent activation of B cells.
      • Herlands R.A.
      • Christensen S.R.
      • Sweet R.A.
      • Hershberg U.
      • Shlomchik M.J.
      T cell-independent and toll-like receptor-dependent antigen-driven activation of autoreactive B cells.
      To further confirm the tissue infiltration of B cells in the livers of septic mice, immunofluorescence staining was performed (Fig. S5J). Similar to that of macrophages, the spatial distribution of B cells was rearranged in the liver after LPS stimulation. An increased proportion of B cells accumulated around the periportal areas (Fig. S5J-M).

      Hepatocytes play a key immunoregulatory role in endotoxemia

      We further explored hepatocyte heterogeneity and functional changes during endotoxemia. We grouped hepatocytes into 3 subclusters along the portal-central lobule axis: periportal hepatocytes (PPHs, expressing Igf1, Ass1, and Pigr), sinusoidal hepatocytes (SHs, expressing Hamp, and Igfbp2), and pericentral hepatocytes (PCHs, expressing Cyp1a2, Gasta3, Glul, and Ugt1a1); the panels of hallmark genes have been summarized in previous studies (Fig. 4A-B, H).
      • Halpern K.B.
      • Shenhav R.
      • Matcovitch-Natan O.
      • Toth B.
      • Lemze D.
      • Golan M.
      • et al.
      Single-cell spatial reconstruction reveals global division of labour in the mammalian liver.
      Efficient metabolism is promoted by the topographical distribution of key enzymes and spatial expression patterns along the lobule axis.
      • Braeuning A.
      • Ittrich C.
      • Kohle C.
      • Hailfinger S.
      • Bonin M.
      • Buchmann A.
      • et al.
      Differential gene expression in periportal and perivenous mouse hepatocytes.
      GO annotation showed that PPHs had higher capacities for amino acid and fatty acid metabolism and that glycolysis was more active in PCHs, which was consistent with prior studies (Fig. 4E).
      • Burke Z.D.
      • Reed K.R.
      • Phesse T.J.
      • Sansom O.J.
      • Clarke A.R.
      • Tosh D.
      Liver zonation occurs through a beta-catenin-dependent, c-Myc-independent mechanism.
      ,
      • Jungermann K.
      • Kietzmann T.
      Zonation of parenchymal and nonparenchymal metabolism in liver.
      Under baseline conditions, PPHs and PCHs accounted for up to 75% of the total hepatocytes (Fig. 4C). Endotoxin induced qualitative transcriptional differences in cell clustering between control livers and insulted livers. The proportions of PPHs and PCHs were significantly reduced from 47% to 4% and from 28% to 4%, respectively, after endotoxin exposure (Fig. 4C-D). These significant proportional reductions in PPHs and PCHs were not consistent with the cellular apoptosis rate observed by FCM and immunofluorescence staining (Fig. S12A-D, F-G). Therefore, we hypothesized that instead of undergoing apoptosis, PPHs and PCHs gradually lose feature genes that identify their functions and zonation profiles during endotoxemia. To test this hypothesis, we performed a higher resolution clustering analysis of hepatocytes without significant expression of a panel of hallmark periportal and pericentral genes. Principal component analysis distinguished 2 subclusters within the hepatocytes without hallmark periportal or pericentral genes, one of which mainly demonstrated a metabolic capacity, while the other cluster had a strong proinflammatory profile and was hereinafter referred to as proinflammatory hepatocytes (PIHs) (Fig. 4F, I). Unsupervised hierarchical clustering of the top 50 differentially expressed genes in each subgroup of hepatocytes distinguished PPHs, PCHs, and SHs from PIHs (Fig. 4G-H).
      Figure thumbnail gr4
      Fig. 4Hepatocytes play a key immunoregulatory role in endotoxemia.
      (A) t-SNE plot displaying the major hepatocytes in different hepatic zones. There were 7,173, 6,750, 5,872 cells in the control, LPS 3-hour, LPS 12-hour group, respectively. (B) Violin plots show the normalized log-transformed expression levels of representative genes (columns) used to define PCH, SH, and PPH (rows). (C) Pie charts displaying the proportion of each cell type in hepatocytes in the indicated groups by scRNA-seq. (D) The percentage change ratio of each cell type. (E) Dot plots show the upregulated signaling pathways across different hepatocytes by GO analysis. (F) Principal component analysis of SH and PIHs. (G) Cartoon depicts the phenotypic transition of hepatocytes in response to endotoxemia. (H) Heatmaps show the average expression levels (color-scaled, row-wise Z scores) of top DEGs (rows) across hepatocyte subtypes (columns) at different time points. Signature genes of the hepatocyte subtypes are displayed on the right. (I) Dot plots showing the upregulated signaling pathways in SH and PIH determined by GO analysis. (J-K) Scatter plots show the DEGs of RNA-seq data between PIH and PPH (left), and PCH (middle), and SH (right). (L) RNA velocity analysis distinguished 5 sets of velocity vectors across the pseudotime axis. (M) Slingshot analysis revealed the sets of vectors across the pseudotime axis. (N) Heatmap shows the interaction scores between ligands and receptors in different cell types by Cellchat.DB. Rows represent the ligands (up) or receptors (down) of the indicated cell types, and columns represent cell types. (O) Heatmaps show the interaction scores of specified ligand-receptor pairs between the indicated cell types. (P) Volcano plots show the genes with different expression in PIH between the LPS 3-hour group and the control group. DEGs are highlighted in red or blue, which were determined by Wilcoxon Rank-Sum test. (Q) Volcano plots show the genes with different expression in RM between the LPS 3-hour group and the control group. (R) The gene expression of Cd274 in hepatocytes in the indicated groups. (S) The gene expression of Cd274 in hepatocytes subgroups at different time points. GO, gene ontology; PCH, pericentral hepatocytes; SH, sinusoidal hepatocytes; PPH, periportal hepatocytes; PIH, proinflammatory hepatocytes; t-SNE, t-distributed stochastic neighbor embedding. (This figure appears in color on the web.)
      To further investigate the hepatocyte origin of PIHs, pairwise comparisons were performed between periportal, pericentral, sinusoidal and proinflammatory hepatocytes. The fewest significantly altered genes were identified when comparing PIHs and PPHs from 3-hour and 12-hour endotoxin-exposed mice (Fig. 4J-K). We also explored the inferred relationship among these cells using RNA velocity and Slingshot analysis. RNA velocity analysis revealed that PIHs were mainly generated by transformation of PPHs (Fig. 4L). RNA velocity analysis distinguished 5 sets of vectors, defined as paths (Path1 – Path5). Path1 and Path2 vectors appeared to represent the differentiation of some PPHs and a fraction of PCHs, respectively, toward a mixed-origin cell population PIHs (Fig. 4L). Path3, Path4, and Path5 vectors appeared to represent slight chronological phenotypic changes in mainly SHs, PPHs, and PCHs, respectively (Fig. 4L). Slingshot was applied and yielded a consistent result with the Velocity analysis (Fig. 4M). These findings suggest that PIHs could be differentiated from both PPHs and PCHs, but mainly from PPHs.
      To further explore hepatocyte-mediated cellular communication during endotoxemia, we analyzed intercellular interactions within heterogeneous populations of liver cells at the single-cell level (Fig. 4N-O).
      • Gu W.
      • Ni Z.
      • Tan Y.Q.
      • Deng J.
      • Zhang S.J.
      • Lv Z.C.
      • et al.
      Adventitial cell atlas of wt (wild type) and ApoE (apolipoprotein E)-Deficient mice defined by single-cell RNA sequencing.
      Expression profiles and interaction scores of ligands and receptors suggested that cell-cell communication was highly activated at the 3-hour time point and that hepatocytes and macrophages were the most active cell types involved in cellular interactions (Fig. 4N; Fig. S13). Among all the interaction pathways, the CCL chemokine pathway ranked the highest in PIHs and RMs (Fig. 4O). Focused analysis showed that CCL2-CCR2 (ligand-receptor) had the highest interaction score, particularly in PIHs and RMs (Fig. 4O). CCL2 is also known as MCP-1. Consistently, we identified differentially expressed genes between the LPS 3-hour and control groups, and our data showed significantly increased levels of Ccl2 in PIHs (Fig. 4P) and Ccr2 in RMs (Fig. 4Q). These results suggested that, in the early stage of endotoxemia, hepatocytes could transform into PIHs to play an essential role in the recruitment of macrophages via CCL2-CCR2 interaction.
      It is worth noting that the expression of Cd274 in hepatocytes was highly upregulated at the 3-hour time point after endotoxin exposure (Fig. 4R). Among hepatocyte subclusters, PIHs exhibited the highest expression of Cd274, followed by PPHs in the early phase (Fig. 4S). The expression pattern of Cd274 (encoded PD-L1) in hepatocytes was consistent with the pattern of Cd279 (encoded PD-1) in T lymphocytes, which may suggest a novel mechanism of T-cell suppression in endotoxemia.
      Following a similar pattern, hepatocyte apoptosis started with a zonation profile in endotoxemia-induced liver injury, which was confirmed by karyolysis or karyopyknosis and Tunel staining of hepatocytes (Fig. S12A-D). Comparisons of endotoxemia model groups and control groups showed a larger number of apoptotic hepatocytes in proximity to the periportal zone than to the pericentral zone. FCM and immunofluorescence staining showed that the number of apoptotic hepatocytes was increased at the 12-hour time point after LPS stimulation (Fig. S12F-G). Accordingly, the levels of serum aminotransferases were increased in a time-dependent manner (Fig. S12H).

      Validation of hepatocyte recruitment of macrophages via a CCL2-mediated pathway

      We further performed in vitro and in vivo experiments to validate that hepatocytes induce macrophage recruitment to the liver via a CCL2-mediated pathway. Using RT-qPCR, we verified that the mRNA expression of Ccl2 in hepatocytes significantly increased as endotoxemia progressed (Fig. 5A). In vitro, the mRNA expression of Ccl2 was also markedly enhanced in cultured primary hepatocytes after 3 hours of poly(I:C) exposure and followed by 3 hours or 12 hours of LPS stimulation (Fig. 5B-C). Cell migration assays showed that primary hepatocytes from mice treated with poly(I:C) and LPS significantly increased the migration of BMDMs (Fig. 5F). Compared with control group, the adeno-associated virus (AAV)-8-NC injection had no significant influence on the mRNA expression of Ccl2 or Cd274, and the percentage of lymphocytes and myeloid cells in the liver (Fig. S14). Using AAV-8-shCCL2, we successfully inhibited LPS-induced Ccl2 expression in the liver tissue and hepatocytes by over 50%, which led to a significant reduction in hepatocyte-induced BMDM migration (Fig. 5C-F). This interaction between hepatocytes and macrophages was also observed with LO2 cells and THP-1 cells in a co-culture system (Fig. S4A-C). The inhibition of Ccl2 in hepatocytes led to prolonged survival times for LPS-induced endotoxic and CLP-induced septic mice compared with AAV-8-control (AAV-8-NC) treatment (Fig. 5G; Fig. S4E). Consistently, ELISA showed that the endotoxin-induced increases in the levels of the inflammatory factors CCL2, IL1β, and TNFα were significantly suppressed in mice treated with AAV-8-shCCL2 compared to AAV-8-NC-treated mice (Fig. 5H). Ccl2 suppression in hepatocytes mitigated endotoxin-induced liver injury, as evidenced by serum aminotransferase levels (Fig. 5I). Flow cytometric analysis, H&E and Tunel staining also confirmed that endotoxin-induced hepatocyte apoptosis was alleviated by CCL2 inhibition (Fig. 5J-L). The number of CCR2+ macrophages (CCR2+F4/80+), defined as inflammatory RMs,
      • Ambade A.
      • Lowe P.
      • Kodys K.
      • Catalano D.
      • Gyongyosi B.
      • Cho Y.
      • et al.
      Pharmacological inhibition of CCR2/5 signaling prevents and reverses alcohol-induced liver damage, steatosis, and inflammation in mice.
      was also significantly reduced by Ccl2 inhibition in hepatocytes (Fig. 5J; Fig. S4F-H). Additionally, the infiltration of F4/80+ macrophages was also suppressed by AAV-8-shCCL2 (Fig. 5J, L). Consistent with the observations in endotoxemia, the mRNA expression of CCL2 was significantly upregulated in isolated hepatocytes isolated from a polymicrobial murine sepsis model established with a CLP protocol (Fig. S4D) and in liver tissue from septic patients (Fig. S3E).
      • Brusletto B.S.
      • Loberg E.M.
      • Hellerud B.C.
      • Goverud I.L.
      • Berg J.P.
      • Olstad O.K.
      • et al.
      Extensive changes in transcriptomic "fingerprints" and immunological cells in the large organs of patients dying of acute septic shock and multiple organ failure caused by Neisseria meningitidis.
      In summary, this in vitro and in vivo evidence showed that hepatocytes recruited RMs to the liver via CCL2-mediated pathways during endotoxemia.
      Figure thumbnail gr5
      Fig. 5Validation of hepatocyte recruitment of macrophages via a CCL2-mediated pathway.
      (A) RT-qPCR analysis of the gene expression of Ccl2 relative to GAPDH in primary hepatocytes isolated from the endotoxemia mouse model (n = 6 mice per group). (B) RT-qPCR analysis of the gene expression of Ccl2 in isolated primary hepatocytes relative to GAPDH with or without stimulation by LPS in vitro. (C) Primary hepatocytes were isolated from mice with AAV-8-shCCL2 or AAV-8-NC infection. RT-qPCR analysis was performed after in vitro LPS stimulation and the expression of Ccl2 relative to GAPDH was quantified. (D) Quantification of Ccl2 mRNA expression in the liver tissue with or without in vivo LPS exposure and with AAV-8-shCCL2 or AAV-8-NC infection (n = 6 mice per group). (E) Primary hepatocytes were isolated from mice infected with AAV-8-shCCL2 or AAV-8-NC infection. RT-qPCR analysis was performed to evaluate isolated hepatocytes after in vivo LPS stimulation for the indicated time and the expression of Ccl2 relative to GAPDH was quantified (n = 6 mice per group). (F) Primary hepatocytes from mice were treated with poly I:C (1 ug/ml) for 3 hours followed by LPS (10 ug/ml) for 4 hours, transwell assays were performed to measure BMDM migration. Representative images (left) and quantitative plots (right) show the numbers of BMDM migration. (G) Kaplan-Meier plots show the survival in endotoxic mice after injection with AAV-8-shCCL2 or AAV-8-NC 3 weeks in advance (n = 12 in each group). (H) The expression of MCP-1 (left), IL-1β (middle), and TNF-α (right) release level in serum was measured via ELISA (n = 8 in each group). (I) The levels of serum ALT (top) and AST (bottom) in mice with different treatments. (J) Representative FAC plots (top) show the percentage of living hepatocytes (Annexin-PI-) in total hepatocytes. Quantitative plots (bottom) show the percentage of living hepatocytes (Annexin-PI-) in total hepatocytes, the percentage of macrophages (F4/80+) and CCR2+ macrophages (CCR2+F4/80+) in total NPC by flow cytometry (n = 7 mice per group). (K) Representative H&E images of liver show the distinct populations of hepatocytes with karyolysis or karyopyknosis of the indicated groups in the different hepatic zone (left). Arrows denote akaryolysis or bkaryopyknosis. Quantitative plots show the changes in hepatocytes with karyolysis (a, b), karyopyknosis (c, d) of total hepatocytes in the indicated groups and hepatic zone (right). (L) Representative images (left) and quantitative plots (right) show the distinct populations of F4/80+ macrophages (top and middle) and Tunel+ apoptotic cells (bottom). Higher magnifications are provided in the circles. n≥20 fields per group from 4 mice, 5 fields per mouse in K-L. Data shown are mean ± SD (A-F and H-L). Data were first analyzed and passed the normality test (Shapiro-Wilk test in A-F and H-L). P values are shown and assessed by one-way ANOVA with Tukey’s test (A-F, I, and K-L) and unpaired 2-tailed t tests (H and J). AAV, adeno-associated virus; ALT, alanine aminotransferase; AST, aspartate aminotransferase; LPS, lipopolysaccharide; NPC, non-parenchymal cell; PC, pericentral zone; PP, periportal zone; RT-qPCR, quantitative reverse-transcription PCR; S, sinusoidal zone. (This figure appears in color on the web.)

      Upregulation of PD-L1 in hepatocytes induces T-cell suppression during endotoxemia

      As mentioned previously, Cd274 was upregulated in PIHs during endotoxemia (Fig. 4R-S). We further confirmed this finding in primary hepatocytes isolated from mice with LPS-challenged BMDMs by RT-qPCR (Fig. 6A). Interestingly, in the absence of BMDMs or THP-1 cells, Cd274 expression was not significantly elevated in poly(I:C)- and LPS-challenged primary hepatocytes or LO2 hepatocytes (Fig. 6A; Fig. S6A-B). In vivo, the attenuation of macrophage recruitment by AAV-8-shCCL2 infection significantly inhibited Cd274 and encoded PD-L1 expression in liver tissue and isolated hepatocytes by RT-qPCR and FCM (Fig. 6B; Fig. S4I-J; Fig. S6C).
      Figure thumbnail gr6
      Fig. 6Upregulation of PD-L1 in hepatocytes induces T-cell suppression during endotoxemia.
      (A) Quantification of Cd274 gene expression relative to GAPDH in primary hepatocytes with or without the co-culture with LPS-pretreated BMDMs at different time points. (B) Quantification of the expression of Cd274 relative to GAPDH in the isolated hepatocytes of endotoxic mouse model with AAV-8-NC or AAV-8-shCCL2 3 weeks in advance (n = 6 mice per group). (C) Quantification of expression levels of Cd274 relative to GAPDH expression in the isolated hepatocytes of endotoxic mouse model with AAV-8-NC or AAV-8-shPD-L1 injection 3 weeks in advance (n = 6 mice per group). (D) Quantitative analysis shows the percentage of CD3+ (left), CD4+ (middle), and CD8+ (right) T cells in total NPC by flow cytometry (n = 7 mice per group). (E) Kaplan-Meier plots show the survival of endotoxic mice 3 weeks after injection of AAV-8-shPD-L1 or AAV-8-NC (n = 14 in each group). (F) The ex-vivo experiments designed to demonstrate that the proinflammatory RMs mediate T-cell suppression by reprogrammed hepatocytes. (G) CCK-8 assay was applied to determine CD4+ T-cell viability in the indicated groups. (H) CCK-8 assay was used to examine CD4+ T-cell viability when cultured with or without mouse IL-27 p28 neutralizing antibody. (I) The ratios of PD-1+ (left) and Ki67+ (right) CD4+ T cells in all CD4+ T cells by flow cytometry (n = 6 per group). (J) The ratios of PD-1+ (left) and Ki67+ (right) CD4+ T cells in all CD4+ T cells with or without mouse IL-27 p28 neutralizing antibody treatment by flow cytometer (n = 6 per group). Data shown are mean ± SD (A-D and G-J). Data were first analyzed and passed the normality test (Shapiro-Wilk test in A-D and G-J). P values are shown and assessed by one-way ANOVA with Tukey’s test in A-D, G, and I-J and unpaired 2-tailed t tests in H. BMDM, bone marrow-derived macrophage; B, BMDM; H, hepatocyte; Nab, neutralizing antibody; RM, recruited macrophage; sup., supernatant; T, CD4+ T cell. (This figure appears in color on the web.)
      T-cell immunosuppression was observed in our endotoxemia model, with the upregulation of an array of suppressive genes, such as Pdcd1, Lag3, Sh2d1a, Cd160 and Ctla4 (Fig. 3G). To examine whether hepatocytes induced T-cell suppression via the PD-1/PD-L1 axis, we conducted an in vivo study by introducing AAV-8-shPD-L1 to suppress PD-L1 expression in LPS-challenged mice. Compared with the control group, the AAV-8-NC injection had no significant influence on the mRNA expression of Ccl2 or Cd274, and the percentage of lymphocytes and myeloid cells in the liver (Fig. S14). Furthermore, we measured the expression level of Cd274 and encoded PD-L1 in hepatocytes during endotoxin exposure by RT-qPCR and FCM (Fig. 6C; Fig. S6D-E). Cd274 expression in hepatocytes was upregulated after LPS stimulation or CLP-induced sepsis but was significantly inhibited by AAV-8-shPD-L1 (Fig. 6C; Fig. S6D-E; Fig. S4K-L). Flow cytometric analysis revealed that PD-L1 inhibition in hepatocytes significantly enhanced the numbers of CD4+ and CD8+ T cells in the liver during endotoxemia and CLP-induced sepsis (Fig. S6F; Fig. 6D; Fig. S4N-P). In addition, Cd274 suppression by AAV-8-shPD-L1 infection in hepatocytes significantly extended the survival time in mice after endotoxin or sepsis exposure (Fig. 6E; Fig. S4M). This evidence showed that the upregulation of PD-L1 in hepatocytes induced T-cell suppression in the liver during endotoxemia. Importantly, the upregulation of PD-L1 in hepatocytes requires the presence of both LPS and macrophages.
      To validate that RMs mediate T-cell suppression via reprogrammed hepatocytes in vitro, we developed a co-culture system in which CD4+ T cells were exposed to the environment with or without BMDM-derived condition medium and hepatocyte contact (Fig. 6F). As Fig. 2K suggests, the expression of Il27 was significantly increased in RMs in endotoxemia. Therefore, we further assessed whether blocking the effect of IL27 p28 (also called IL30) using a neutralizing antibody could lessen T-cell suppression induced by hepatocytes and BMDMs. CCK-8 assay showed that the proliferation and viability of CD4+ T cells were markedly inhibited when contact co-cultured with hepatocytes in supernatant from LPS-challenged BMDMs (Fig. 6G). Consistently, FCM showed that intracellular proliferation marker Ki67 was also significantly attenuated in CD4+ T cells contact co-cultured with hepatocytes and simultaneously treated with supernatant from LPS-stimulated BMDMs (Fig. 6I). Furthermore, the inhibitory effect of hepatocytes and conditioned BMDM supernatant on CD4+ T-cell viability was ameliorated by a neutralizing antibody against IL27 p28 (also called IL30) (Fig. 6H, J). Quantitative RT-qPCR revealed that immune-inhibitory receptor gene (Cd279, encoded PD-1) and proapoptotic genes (Bax and Caspase3) were significantly upregulated in CD4+ T cells simultaneously treated with supernatant from LPS-challenged BMDMs and contact co-cultured with hepatocytes, while the immune effector genes (Il2, Il4, Ifng, and Tnfa) and apoptosis-suppressing gene (Bcl2) were significantly suppressed in CD4+ T cells under the aforementioned co-culture conditions (Fig. S7C). FCM assays further validated these findings (Fig. S7A), while the inhibitory effect of CD4+ T-cell immune function induced by hepatocytes and BMDMs could be reversed by IL27 p28 neutralizing antibody (Fig. S7B, D).

      Zonation patterns of liver endothelial cell injury in endotoxemia

      The graded lobule microenvironment gives rise to spatial division of labor among LSECs. Adapting a LEC hallmark gene panel discovered in a previous study, we clustered LSECs into 3 major subpopulations: periportal ECs (expressing Dll4 and Ltbp4), pericentral ECs (expressing Wnt2 and Rspo3), and sinusoidal ECs (SECs, expressing Stab2 and Lyve2) (Fig. 7A, D-E). Notably, our results showed that the number of periportal and pericentral ECs decreased in a time-dependent manner during the progression of endotoxemia by scRNA-seq (Fig. 7B-C). The sharp reduction in LSECs was also confirmed by FCM (Fig. 7F-G) and immunofluorescence staining (Fig. 7H-I). This zonated reduction in ECs might be a prerequisite for the infiltration of immune cells and effective protection against infection.
      Figure thumbnail gr7
      Fig. 7Zonation patterns of liver endothelial cell injury in endotoxemia.
      (A) t-SNE plots display the major endothelial cell types in different hepatic zones. There were 1,968, 1,308, 739 cells in the control, LPS 3-hour, LPS 12-hour group, respectively. (B) Bar charts show the number of major EC types at different time points. (C) The percentage change ratio of each EC subtype by scRNA-seq. (D) Heatmap shows the average expression (color-scaled, row-wise Z scores) of top 20 DEGs (rows) across EC subtypes (columns). Signature genes of each cluster displayed on the left. (E) Violin plots show the normalized log-transformed expression of selected genes (rows) to define PEC, SEC, and CEC (columns). (F-G) Representative FAC plots and quantitative plots show the percentage of hepatic ECs in total NPC by flow cytometry (n = 6 mice per group). (H) Representative images show distinct CD31+ ECs in different hepatic zones at different time points. E-cadherin+ area indicates the periportal regions. (I) Quantitative plots show the total number of CD31+ ECs per 100um
      • Kubes P.
      • Jenne C.
      Immune responses in the liver.
      (top) and fold of change compared to control of ECs (n ≥20 fields per group from 4 mice, 5 fields per mouse). (J-K) The gene set variation analysis scores reveal the top upregulated and downregulated signaling pathways in CEC (left), SEC (middle), and PEC (right) compared LPS 3-hour group and LPS 12-hour group with control group. (L) Line plots show the dynamic changes in inflammatory response, apoptotic process, cellular metabolism, and cell proliferation in CEC (blue), SEC (pink), and PEC (green) in the indicated groups. Data shown are mean ± SD (G and I). Data were first analyzed and passed the normality test (Shapiro-Wilk test in G and I). P values were shown and assessed by one-way ANOVA with Tukey’s test (G and I). CEC, pericentral endothelial cells; CON, control; CV, pericentral vein; LPS, lipopolysaccharide; PC, pericentral zone; PEC, periportal endothelial cells; PP, periportal zone; PV, periportal vein; S, sinusoidal zone; SEC, sinusoidal endothelial cells; t-SNE, t-distributed stochastic neighbor embedding. (This figure appears in color on the web.)
      GO annotation analysis and gene set variation analysis revealed that LSECs around large vascular areas exhibited progressively increased inflammatory and apoptotic responses and suppressed metabolic rates (Fig. 7J-L). In addition, similar to hepatocytes, LSECs around the portal zone exhibited further increased inflammatory and apoptotic responses compared to those of cells close to the central zone.

      Discussion

      Here, we investigated hepatic cellular transcriptomics and anatomic characteristics in a mouse model of endotoxemia. RMs were largely enriched during endotoxemia, particularly proximal to the portal vein.
      • Mossanen J.C.
      • Krenkel O.
      • Ergen C.
      • Govaere O.
      • Liepelt A.
      • Puengel T.
      • et al.
      Chemokine (C-C motif) receptor 2-positive monocytes aggravate the early phase of acetaminophen-induced acute liver injury.
      Hepatocytes around the perivascular zone switched to a proinflammatory phenotype, and these cells play a crucial role in the recruitment of RMs via the CCL2-CCR2 interaction. Subsequently, RMs release a wide array of cytokines and induced the expression of inhibitory ligands on hepatocytes, thereby leading to the inhibition of CD4+ T lymphocytes in the context of endotoxemia. Consistently, SECs expressed the highest levels of pro-apoptotic and inflammatory genes in the periportal zone, facilitating increases in the number of fenestrations and infiltration of immune cells from circulation. These key findings were also validated in a CLP-induced sepsis model and with RNA-seq data for septic patient livers. The findings provide evidence of the essential roles of phenotypic transitions of hepatocytes around the periportal zone in the initiation of immune cell infiltration, immune suppression, and liver injuries during acute systemic infection.
      Using a panel of zonated hallmark genes developed by Halpern et al., we grouped hepatocytes into 3 subclusters (PPHs, PCHs, and SHs).
      • Halpern K.B.
      • Shenhav R.
      • Matcovitch-Natan O.
      • Toth B.
      • Lemze D.
      • Golan M.
      • et al.
      Single-cell spatial reconstruction reveals global division of labour in the mammalian liver.
      Previous studies have examined genome-wide differences in gene expression between hepatocytes enriched in the periportal and pericentral zones.
      • Braeuning A.
      • Ittrich C.
      • Kohle C.
      • Hailfinger S.
      • Bonin M.
      • Buchmann A.
      • et al.
      Differential gene expression in periportal and perivenous mouse hepatocytes.
      ,
      • Gebhardt R.
      • Matz-Soja M.
      Liver zonation: novel aspects of its regulation and its impact on homeostasis.
      The expression of featured genes exhibited spatial expression patterns along the lobule axis.
      • Halpern K.B.
      • Shenhav R.
      • Matcovitch-Natan O.
      • Toth B.
      • Lemze D.
      • Golan M.
      • et al.
      Single-cell spatial reconstruction reveals global division of labour in the mammalian liver.
      For example, outer highly oxygenated PPHs expressed higher levels of enzymes involved in energy-demanding tasks such as gluconeogenesis and ureagenesis, whereas inner PCHs specialized in glycolysis and xenobiotic metabolism.
      • Braeuning A.
      • Ittrich C.
      • Kohle C.
      • Hailfinger S.
      • Bonin M.
      • Buchmann A.
      • et al.
      Differential gene expression in periportal and perivenous mouse hepatocytes.
      PPHs and PCHs gradually lost expression of feature genes that identified their function and promoted inflammation during endotoxemia, and we termed these hepatocytes “PIHs”. PIHs have the fewest gene alterations compared to PPHs, indicating that endotoxemia may primarily reprogram PPHs. PIHs demonstrated a strong proinflammatory profile and played a central role in immune cell recruitment. Moreover, the high expression of inhibitory ligands by PIHs suggested a novel mechanism mediating T-cell suppression in endotoxemia.
      Hepatic macrophages are the central cell type involved in eliminating invading microorganisms, and these cells make up the constitutive element of the host defense system. In our study, RMs had high gene expression of Il17a, Il1rn, S100a9, and Il17f, which represent the enhanced capabilities of these cells to secrete chemoattractant, present MHC-II antigens, and activate adaptive immune responses. RMs derived from circulating monocytes have been reported in many acute and chronic liver diseases and actively participate in disease progression. For example, LY6Chi and TREM2+CD9+ macrophages derived from circulating monocytes represent scar-associated macrophage subsets that play an essential role in the progression of non-alcoholic steatohepatitis.
      • Seidman J.S.
      • Troutman T.D.
      • Sakai M.
      • Gola A.
      • Spann N.J.
      • Bennett H.
      • et al.
      Niche-specific reprogramming of epigenetic landscapes drives myeloid cell diversity in nonalcoholic steatohepatitis.
      ,
      • Ramachandran P.
      • Dobie R.
      • Wilson-Kanamori J.R.
      • Dora E.F.
      • Henderson B.E.P.
      • Luu N.T.
      • et al.
      Resolving the fibrotic niche of human liver cirrhosis at single-cell level.
      We discovered that hepatocytes are the crucial cell type that provides chemoattractant signals (such as CCL2) to recruit and instruct monocyte switching to RM. Although resident macrophages and RMs contribute to host defense in endotoxemia, we revealed that RMs were the major cell type associated with subsequent immune responses and liver injury in this study. This finding is in contrast to the previous understanding of the pathogenesis of sepsis-associated liver injury, in which KCs were considered the predominant cell type. Moreover, we discovered that RMs produced the major cytokines responsible for the inducible expression of PD-L1 in hepatocytes. In addition, RMs were largely enriched in the area proximal to the portal vein, which strongly indicates that the distinct microenvironment in this region drives RM differentiation and cellular communication.
      Previous studies have shown that treatment with CCL2 inhibitors, including mNOX-E36, cenicriviroc, and CCX872, results in reduced liver injury and liver inflammation.
      • Ambade A.
      • Lowe P.
      • Kodys K.
      • Catalano D.
      • Gyongyosi B.
      • Cho Y.
      • et al.
      Pharmacological inhibition of CCR2/5 signaling prevents and reverses alcohol-induced liver damage, steatosis, and inflammation in mice.
      ,
      • Baeck C.
      • Wehr A.
      • Karlmark K.R.
      • Heymann F.
      • Vucur M.
      • Gassler N.
      • et al.
      Pharmacological inhibition of the chemokine CCL2 (MCP-1) diminishes liver macrophage infiltration and steatohepatitis in chronic hepatic injury.
      • Baeck C.
      • Wei X.
      • Bartneck M.
      • Fech V.
      • Heymann F.
      • Gassler N.
      • et al.
      Pharmacological inhibition of the chemokine C-C motif chemokine ligand 2 (monocyte chemoattractant protein 1) accelerates liver fibrosis regression by suppressing Ly-6C(+) macrophage infiltration in mice.
      • Parker R.
      • Weston C.J.
      • Miao Z.
      • Corbett C.
      • Armstrong M.J.
      • Ertl L.
      • et al.
      CC chemokine receptor 2 promotes recruitment of myeloid cells associated with insulin resistance in nonalcoholic fatty liver disease.
      However, few studies have focused on mechanisms of macrophage recruitment and chemotactic regions. Our research elucidates an important pathway by which hepatocytes recruit macrophages to the liver during septic inflammatory activation. Neutrophils are among the first immune cells to infiltrate the liver during acute liver injury or inflammation.
      • Marques P.E.
      • Oliveira A.G.
      • Pereira R.V.
      • David B.A.
      • Gomides L.F.
      • Saraiva A.M.
      • et al.
      Hepatic DNA deposition drives drug-induced liver injury and inflammation in mice.
      ,
      • Cai J.
      • Li H.
      • Zhang C.
      • Chen Z.
      • Liu H.
      • Lei F.
      • et al.
      The neutrophil-to-lymphocyte ratio determines clinical efficacy of corticosteroid Therapy in patients with COVID-19.
      We found that neutrophils infiltrated the liver within the first few hours of endotoxemia and initiated a rapid process of NET formation for host defense against pathogens. Thus, both neutrophils and macrophages are important players in driving inflammation during the early stages of endotoxemia.
      The inhibition of T-cell function and expansion was observed in our scRNA-seq analysis. Two major themes explain the mechanism of T-cell inhibition during severe infection. One is the extrinsic regulatory pathways that suppress T-cell functionality, and the other is the negative regulatory pathways mediated by immune-inhibitory receptors, including CTLA4, LAG3, PD1, and CD160.
      • Schnell A.
      • Bod L.
      • Madi A.
      • Kuchroo V.K.
      The yin and yang of co-inhibitory receptors: toward anti-tumor immunity without autoimmunity.
      Our data and those of previous studies revealed that blocking the PD-1 pathway altered CD4+ T-cell function more significantly than CD8+ T cell function.
      • Tang Z.S.
      • Hao Y.H.
      • Zhang E.J.
      • Xu C.L.
      • Zhou Y.
      • Zheng X.
      • et al.
      CD28 family of receptors on T cells in chronic HBV infection: expression characteristics, clinical significance and correlations with PD-1 blockade.
      More surprisingly, scRNA-seq and FCM revealed increased PD-L1 expression in hepatocytes, particularly in PIHs, after 3 hours of endotoxemia. Liver-targeted anti-PD-L1 treatment increased the number of local T cells and extended the survival time of mice with endotoxemia. This observation suggests that anti-PD-L1-targeted immunotherapy may serve as an efficient treatment for endotoxemia-associated immunosuppression and mortality. Similar to our animal research, PD-L1 blockade improved survival and alleviated disease morphology in septic mice with liver injury.
      • Zhang Y.
      • Zhou Y.
      • Lou J.
      • Li J.
      • Bo L.
      • Zhu K.
      • et al.
      PD-L1 blockade improves survival in experimental sepsis by inhibiting lymphocyte apoptosis and reversing monocyte dysfunction.
      PD-L1 was also expressed on KCs and LSECs.
      • Wang B.J.
      • Bao J.J.
      • Wang J.Z.
      • Wang Y.
      • Jiang M.
      • Xing M.Y.
      • et al.
      Immunostaining of PD-1/PD-Ls in liver tissues of patients with hepatitis and hepatocellular carcinoma.
      ,
      • Mueller S.N.
      • Vanguri V.K.
      • Ha S.J.
      • West E.E.
      • Keir M.E.
      • Glickman J.N.
      • et al.
      PD-L1 has distinct functions in hematopoietic and nonhematopoietic cells in regulating T cell responses during chronic infection in mice.
      Transcriptomic expression of PD-L1 was observed in ECs but an increase during endotoxemia was not observed, which might result from the sharp decrease in ECs. These results deepen our understanding of immunosuppression, especially T-cell exhaustion, in endotoxemia.
      Previous understanding that the immunosuppressive response occurs during the late phase following the early hyperinflammatory phase in acute infectious disease has now been challenged.
      • Niu B.
      • Zhou F.
      • Su Y.
      • Wang L.
      • Xu Y.
      • Yi Z.
      • et al.
      Different expression characteristics of LAG3 and PD-1 in sepsis and their synergistic effect on T cell exhaustion: a new strategy for immune checkpoint blockade.
      • Brahmamdam P.
      • Inoue S.
      • Unsinger J.
      • Chang K.C.
      • McDunn J.E.
      • Hotchkiss R.S.
      Delayed administration of anti-PD-1 antibody reverses immune dysfunction and improves survival during sepsis.
      • Guignant C.
      • Lepape A.
      • Huang X.
      • Kherouf H.
      • Denis L.
      • Poitevin F.
      • et al.
      Programmed death-1 levels correlate with increased mortality, nosocomial infection and immune dysfunctions in septic shock patients.
      Accumulating evidence shows that immunosuppression occurs simultaneously with inflammation during the earliest stages of severe infection.
      • Boomer J.S.
      • To K.
      • Chang K.C.
      • Takasu O.
      • Osborne D.F.
      • Walton A.H.
      • et al.
      Immunosuppression in patients who die of sepsis and multiple organ failure.
      Excessive inflammation and immunosuppression are responsible for treatment dilemmas in the clinic and for poor prognoses.
      • Boomer J.S.
      • To K.
      • Chang K.C.
      • Takasu O.
      • Osborne D.F.
      • Walton A.H.
      • et al.
      Immunosuppression in patients who die of sepsis and multiple organ failure.
      ,
      • Bone R.C.
      Immunologic dissonance: a continuing evolution in our understanding of the systemic inflammatory response syndrome (SIRS) and the multiple organ dysfunction syndrome (MODS).
      Our study showed that the early increased expression of PD-L1 on hepatocytes and its subsequent interaction with PD-1 on lymphocytes are important mechanisms involved in lymphocyte exhaustion in acute infectious diseases, such as endotoxemia and sepsis. Targeting CCL2 and PD-L1 signaling in hepatocytes may be a promising therapeutic approach for alleviating immunosuppression and disease severity. Combined with advanced organ delivery systems,
      • Salunkhe S.A.
      • Chitkara D.
      • Mahato R.I.
      • Mittal A.
      Lipid based nanocarriers for effective drug delivery and treatment of diabetes associated liver fibrosis.
      such as nanocarriers, this approach could achieve organ-specific therapeutic effects without causing undesirable systemic adverse effects.
      The production of antibodies by activated B lymphocytes is critical for the efficient eradication of invading bacteria. In addition, B cells are involved in protective early innate immune responses in an antibody-independent manner.
      • Kelly-Scumpia K.M.
      • Scumpia P.O.
      • Weinstein J.S.
      • Delano M.J.
      • Cuenca A.G.
      • Nacionales D.C.
      • et al.
      B cells enhance early innate immune responses during bacterial sepsis.
      The fast-proliferating B-cell population is generated early in infection as a result of T cell-independent activation of pre/plasmablasts B cells or the extrafollicular response of T cell-dependent activation of B cells.
      • Herlands R.A.
      • Christensen S.R.
      • Sweet R.A.
      • Hershberg U.
      • Shlomchik M.J.
      T cell-independent and toll-like receptor-dependent antigen-driven activation of autoreactive B cells.
      Moreover, our results show that B cells are able to link the innate and adaptive immunity and that innate immunity is activated in the early stage, but not in the late stage when adaptive immunity is activated. During endotoxemia, the increase in cytokines and chemokines may promote hepatic B cells to migrate to other sites through chemokine-chemokine receptor interactions after LPS stimulation.
      • Matsumoto A.
      • Tsujimoto H.
      • Ono S.
      • Kinoshita M.
      • Habu Y.
      • Kawabata T.
      • et al.
      Loss of hepatic B cells following lipopolysaccharide injection and polymicrobial sepsis.
      Under endotoxic conditions, LSECs exhibit progressively increased expression of proinflammatory and proapoptotic genes in the area proximal to hepatic vessels, particularly around the periportal zone. Accordingly, the progressive loss of LSECs followed a similar pattern. This distinct pattern of LSECs and hepatocytes coordinately provides strong signals and a microenvironment that recruits circulating immune cells into specific locations in liver sinusoids.
      We also acknowledge the limitations of our study. First, our scRNA-seq analysis did not include all cell types involved in liver pathophysiology during endotoxemia. For instance, hepatic stellate cells (HSCs) have been reported to be activated in an acute endotoxemia model and chronic endotoxemia model by triggering the Smad-dependent profibrogenic signaling pathway.
      • Kao Y.H.
      • Chen P.H.
      • Wu T.Y.
      • Lin Y.C.
      • Tsai M.S.
      • Lee P.H.
      • et al.
      Lipopolysaccharides induce Smad2 phosphorylation through PI3K/Akt and MAPK cascades in HSC-T6 hepatic stellate cells.
      ,
      • Gonnert F.A.
      • Kunisch E.
      • Gajda M.
      • Lambeck S.
      • Weber M.
      • Claus R.A.
      • et al.
      Hepatic fibrosis in a long-term murine model of sepsis.
      However, we did not detect a distinct cluster of stellate cells in our model. One possible explanation for the low detection of stellate cells might be related to the gentle tissue digestion approach, with the single-cell isolation procedure failing to release all stellate cells from liver sinusoids. The other reason is that the total number of HSCs is very small (approximately 1% of NPCs), and previous studies have enriched HSCs or used specific digestion schemes to isolate HSCs.
      • Zhang W.
      • Conway S.J.
      • Liu Y.
      • Snider P.
      • Chen H.
      • Gao H.
      • et al.
      Heterogeneity of hepatic stellate cells in fibrogenesis of the liver: insights from single-cell transcriptomic analysis in liver injury.
      ,
      • Mederacke I.
      • Dapito D.H.
      • Affo S.
      • Uchinami H.
      • Schwabe R.F.
      High-yield and high-purity isolation of hepatic stellate cells from normal and fibrotic mouse livers.
      It has been reported that activated HSCs specifically express a group of chemokines, such as CCL2, CXCL1, and CXCL2, in response to carbon tetrachloride-induced liver fibrosis, which may mediate macrophage recruitment and liver injury, but further verification is needed in an infectious model.
      • Zhang W.
      • Conway S.J.
      • Liu Y.
      • Snider P.
      • Chen H.
      • Gao H.
      • et al.
      Heterogeneity of hepatic stellate cells in fibrogenesis of the liver: insights from single-cell transcriptomic analysis in liver injury.
      Moreover, the findings of our current study are based on endotoxemia and sepsis models established with mice, which have not been sufficiently validated in human studies. Further evidence from human studies is needed to confirm the translational value of our findings.
      In summary, our study provides the first evidence from single-cell transcriptomic analyses that phenotypic transition of hepatocytes in response to endotoxemia plays central roles in the recruitment of circulating monocytes to generate an RM population and in the inhibition of T lymphocytes. This study advances the understanding of the pathophysiology of endotoxemia from the perspective of liver, and these findings provide pivotal foundations for the development of an effective therapeutic approach for acute infection.

      Abbreviations

      AAV, adeno-associated virus; APB, antigen-presenting B cell; BM, bone marrow; BMDMs, bone marrow-derived macrophages; CLP, cecal ligation and puncture; DCs, dendritic cells; ECs, endothelial cells; FCM, flow cytometry; GO, gene ontology; HSCs, hepatic stellate cells; KCs, Kupffer cells; LPS, lipopolysaccharide; LSECs, liver sinusoids endothelial cells; NETs, neutrophil extracellular traps; NK, natural killer; NKT, natural killer T; PIH, proinflammatory hepatocyte; PPH, periportal hepatocyte; RM, recruited macrophages; RT-qPCR, quantitative reverse-transcription PCR; scRNA-seq, single-cell RNA sequencing; SEC, sinusoidal endothelium cells; SH, sinusoidal hepatocytes; Trm, tissue-resident memory; YS, yolk sac.

      Financial support

      This study was supported by the Natural Science Foundation of China 82170436 , 81870171 (to J.J.C), 81770403 , 81974054 (to H.Y), the National Key Research and Development Projects 2019YFF0216305 (to J.J.C), and 2016YFC0900802 (to H.Y), the Hunan Distinguished Young Scholars 2018JJ1048 (to J.J.C), the Independent Exploration and Innovation Project for Graduate Students of Central South University 1053320210374 (to X.J.S).

      Authors' contributions

      X.J.S performed and validated the major experiments, analyzed and interpreted the single-cell RNA sequencing data, and wrote the initial draft. J.R.W and Y.Y.C made the murine model of endotoxemia and collected samples. J.R.W, Y.T. and S.Z.L performed staining experiments, bioinformatics and statistical analysis and critical revision. L.L. performed single-cell RNA sequencing analysis, analyzed and curated data, and organized figures. Y.X.J performed the in vitro experiments and validated animal experiments. Y.Y.L and H.C. collected all samples for single-cell RNA sequencing, provided the reporter mice propagation, and constructed adeno-associated virus delivery system in the animal models. Y.L and H.Y provided data interpretation, constructive suggestions, and technical and material support. Z.Y.C and J.J.C conceived the idea, designed projects, performed critical reviews of the manuscript, supervised the study and approved to publish, and provided funding for the study. All authors read and approved the manuscript.

      Data availability statement

      The data that support the findings of this study are available on reasonable request. Single-cell RNA sequencing data of our study are available in Gene Expression Omnibus (GSE186554 for sc-RNA sequencing of liver hepatocytes and GSE186558 for sc-RNA sequencing of liver non-parenchymal cells from the murine model of endotoxemia).

      Conflict of interest

      The authors declare no conflicts of interest that pertain to this work. Please refer to the accompanying ICMJE disclosure forms for further details.

      Supplementary data

      References

        • Guo J.
        • Li Y.
        • Shan Y.
        • Shu C.
        • Wang F.
        • Wang X.
        • et al.
        Humanized mice reveal an essential role for human hepatocytes in the development of the liver immune system.
        Cell Death Dis. 2018; 9: 667
        • Kubes P.
        • Jenne C.
        Immune responses in the liver.
        Annu Rev Immunol. 2018; 36: 247-277
        • Thomson A.W.
        • Knolle P.A.
        Antigen-presenting cell function in the tolerogenic liver environment.
        Nat Rev Immunol. 2010; 10: 753-766
        • Wang J.
        • Ma J.
        • Nie H.
        • Zhang X.J.
        • Zhang P.
        • She Z.G.
        • et al.
        Hepatic regulator of G protein signaling 5 ameliorates nonalcoholic fatty liver disease by suppressing transforming growth factor beta-activated Kinase 1-c-Jun-N-Terminal Kinase/p38 signaling.
        Hepatology. 2021; 73: 104-125
        • Strnad P.
        • Tacke F.
        • Koch A.
        • Trautwein C.
        Liver - guardian, modifier and target of sepsis.
        Nat Rev Gastroenterol Hepatol. 2017; 14: 55-66
        • Yan J.
        • Li S.
        • Li S.
        The role of the liver in sepsis.
        Int Rev Immunol. 2014; 33: 498-510
        • Zhang L.
        • Tian R.
        • Yao X.
        • Zhang X.J.
        • Zhang P.
        • Huang Y.
        • et al.
        Milk fat globule-epidermal growth factor-factor 8 improves hepatic steatosis and inflammation.
        Hepatology. 2021; 73: 586-605
        • Mao K.
        • Baptista A.P.
        • Tamoutounour S.
        • Zhuang L.
        • Bouladoux N.
        • Martins A.J.
        • et al.
        Innate and adaptive lymphocytes sequentially shape the gut microbiota and lipid metabolism.
        Nature. 2018; 554: 255-259
        • Davies S.P.
        • Terry L.V.
        • Wilkinson A.L.
        • Stamataki Z.
        Cell-in-Cell structures in the liver: a Tale of four E's.
        Front Immunol. 2020; 11: 650
        • Megahed F.A.K.
        • Zhou X.
        • Sun P.
        The interactions between HBV and the innate immunity of hepatocytes.
        Viruses. 2020; 12
        • Halpern K.B.
        • Shenhav R.
        • Massalha H.
        • Toth B.
        • Egozi A.
        • Massasa E.E.
        • et al.
        Paired-cell sequencing enables spatial gene expression mapping of liver endothelial cells.
        Nat Biotechnol. 2018; 36: 962-970
        • Braeuning A.
        • Ittrich C.
        • Kohle C.
        • Hailfinger S.
        • Bonin M.
        • Buchmann A.
        • et al.
        Differential gene expression in periportal and perivenous mouse hepatocytes.
        FEBS J. 2006; 273: 5051-5061
        • Gebhardt R.
        • Matz-Soja M.
        Liver zonation: novel aspects of its regulation and its impact on homeostasis.
        World J Gastroenterol. 2014; 20: 8491-8504
        • Halpern K.B.
        • Shenhav R.
        • Matcovitch-Natan O.
        • Toth B.
        • Lemze D.
        • Golan M.
        • et al.
        Single-cell spatial reconstruction reveals global division of labour in the mammalian liver.
        Nature. 2017; 542: 352-356
        • Gola A.
        • Dorrington M.G.
        • Speranza E.
        • Sala C.
        • Shih R.M.
        • Radtke A.J.
        • et al.
        Commensal-driven immune zonation of the liver promotes host defence.
        Nature. 2021; 589: 131-136
        • Hagar J.A.
        • Powell D.A.
        • Aachoui Y.
        • Ernst R.K.
        • Miao E.A.
        Cytoplasmic LPS activates caspase-11: implications in TLR4-independent endotoxic shock.
        Science. 2013; 341: 1250-1253
        • Mascaux C.
        • Angelova M.
        • Vasaturo A.
        • Beane J.
        • Hijazi K.
        • Anthoine G.
        • et al.
        Immune evasion before tumour invasion in early lung squamous carcinogenesis.
        Nature. 2019; 571: 570-575
        • Snell L.M.
        • Xu W.
        • Abd-Rabbo D.
        • Boukhaled G.
        • Guo M.
        • Macleod B.L.
        • et al.
        Dynamic CD4(+) T cell heterogeneity defines subset-specific suppression and PD-L1-blockade-driven functional restoration in chronic infection.
        Nat Immunol. 2021; 22: 1524-1537
        • Knolle P.A.
        • Gerken G.
        Local control of the immune response in the liver.
        Immunol Rev. 2000; 174: 21-34
        • Beattie L.
        • Sawtell A.
        • Mann J.
        • Frame T.C.M.
        • Teal B.
        • de Labastida Rivera F.
        • et al.
        Bone marrow-derived and resident liver macrophages display unique transcriptomic signatures but similar biological functions.
        J Hepatol. 2016; 65: 758-768
        • Yona S.
        • Kim K.W.
        • Wolf Y.
        • Mildner A.
        • Varol D.
        • Breker M.
        • et al.
        Fate mapping reveals origins and dynamics of monocytes and tissue macrophages under homeostasis.
        Immunity. 2013; 38: 79-91
        • Daemen S.
        • Gainullina A.
        • Kalugotla G.
        • He L.
        • Chan M.M.
        • Beals J.W.
        • et al.
        Dynamic shifts in the composition of resident and recruited macrophages influence tissue remodeling in NASH.
        Cell Rep. 2021; 34: 108626
        • Radermecker C.
        • Sabatel C.
        • Vanwinge C.
        • Ruscitti C.
        • Marechal P.
        • Perin F.
        • et al.
        Locally instructed CXCR4(hi) neutrophils trigger environment-driven allergic asthma through the release of neutrophil extracellular traps.
        Nat Immunol. 2019; 20: 1444-1455
        • Zhang Z.
        • Zhang S.
        • Zou Z.
        • Shi J.
        • Zhao J.
        • Fan R.
        • et al.
        Hypercytolytic activity of hepatic natural killer cells correlates with liver injury in chronic hepatitis B patients.
        Hepatology. 2011; 53: 73-85
        • Giannikopoulos G.
        • Antonopoulou A.
        • Kalpakou G.
        • Makaritsis K.
        • Panou C.
        • Papadomichelakis E.
        • et al.
        The functional role of natural killer cells early in clinical sepsis.
        APMIS. 2013; 121: 329-336
        • Venet F.
        • Monneret G.
        Advances in the understanding and treatment of sepsis-induced immunosuppression.
        Nat Rev Nephrol. 2018; 14: 121-137
        • Wherry E.J.
        T cell exhaustion.
        Nat Immunol. 2011; 12: 492-499
        • Qiu X.
        • Li J.
        • Bonenfant J.
        • Jaroszewski L.
        • Mittal A.
        • Klein W.
        • et al.
        Dynamic changes in human single-cell transcriptional signatures during fatal sepsis.
        J Leukoc Biol. 2021; 110: 1253-1268
        • Darden D.B.
        • Dong X.
        • Brusko M.A.
        • Kelly L.
        • Fenner B.
        • Rincon J.C.
        • et al.
        A novel single cell RNA-seq analysis of non-myeloid circulating cells in late sepsis.
        Front Immunol. 2021; 12: 696536
        • Chang D.Y.
        • Song S.H.
        • You S.
        • Lee J.
        • Kim J.
        • Racanelli V.
        • et al.
        Programmed death-1 (PD-1)-dependent functional impairment of CD4(+) T cells in recurrent genital papilloma.
        Clin Exp Med. 2014; 14: 305-313
        • Kelly-Scumpia K.M.
        • Scumpia P.O.
        • Weinstein J.S.
        • Delano M.J.
        • Cuenca A.G.
        • Nacionales D.C.
        • et al.
        B cells enhance early innate immune responses during bacterial sepsis.
        J Exp Med. 2011; 208: 1673-1682
        • Herlands R.A.
        • Christensen S.R.
        • Sweet R.A.
        • Hershberg U.
        • Shlomchik M.J.
        T cell-independent and toll-like receptor-dependent antigen-driven activation of autoreactive B cells.
        Immunity. 2008; 29: 249-260
        • Burke Z.D.
        • Reed K.R.
        • Phesse T.J.
        • Sansom O.J.
        • Clarke A.R.
        • Tosh D.
        Liver zonation occurs through a beta-catenin-dependent, c-Myc-independent mechanism.
        Gastroenterology. 2009; 136 (2313): 2316-2324 e2311
        • Jungermann K.
        • Kietzmann T.
        Zonation of parenchymal and nonparenchymal metabolism in liver.
        Annu Rev Nutr. 1996; 16: 179-203
        • Gu W.
        • Ni Z.
        • Tan Y.Q.
        • Deng J.
        • Zhang S.J.
        • Lv Z.C.
        • et al.
        Adventitial cell atlas of wt (wild type) and ApoE (apolipoprotein E)-Deficient mice defined by single-cell RNA sequencing.
        Arterioscler Thromb Vasc Biol. 2019; 39: 1055-1071
        • Ambade A.
        • Lowe P.
        • Kodys K.
        • Catalano D.
        • Gyongyosi B.
        • Cho Y.
        • et al.
        Pharmacological inhibition of CCR2/5 signaling prevents and reverses alcohol-induced liver damage, steatosis, and inflammation in mice.
        Hepatology. 2019; 69: 1105-1121
        • Brusletto B.S.
        • Loberg E.M.
        • Hellerud B.C.
        • Goverud I.L.
        • Berg J.P.
        • Olstad O.K.
        • et al.
        Extensive changes in transcriptomic "fingerprints" and immunological cells in the large organs of patients dying of acute septic shock and multiple organ failure caused by Neisseria meningitidis.
        Front Cel Infect Microbiol. 2020; 10: 42
        • Mossanen J.C.
        • Krenkel O.
        • Ergen C.
        • Govaere O.
        • Liepelt A.
        • Puengel T.
        • et al.
        Chemokine (C-C motif) receptor 2-positive monocytes aggravate the early phase of acetaminophen-induced acute liver injury.
        Hepatology. 2016; 64: 1667-1682
        • Seidman J.S.
        • Troutman T.D.
        • Sakai M.
        • Gola A.
        • Spann N.J.
        • Bennett H.
        • et al.
        Niche-specific reprogramming of epigenetic landscapes drives myeloid cell diversity in nonalcoholic steatohepatitis.
        Immunity. 2020; 52: 1057-1074 e1057
        • Ramachandran P.
        • Dobie R.
        • Wilson-Kanamori J.R.
        • Dora E.F.
        • Henderson B.E.P.
        • Luu N.T.
        • et al.
        Resolving the fibrotic niche of human liver cirrhosis at single-cell level.
        Nature. 2019; 575: 512-518
        • Baeck C.
        • Wehr A.
        • Karlmark K.R.
        • Heymann F.
        • Vucur M.
        • Gassler N.
        • et al.
        Pharmacological inhibition of the chemokine CCL2 (MCP-1) diminishes liver macrophage infiltration and steatohepatitis in chronic hepatic injury.
        Gut. 2012; 61: 416-426
        • Baeck C.
        • Wei X.
        • Bartneck M.
        • Fech V.
        • Heymann F.
        • Gassler N.
        • et al.
        Pharmacological inhibition of the chemokine C-C motif chemokine ligand 2 (monocyte chemoattractant protein 1) accelerates liver fibrosis regression by suppressing Ly-6C(+) macrophage infiltration in mice.
        Hepatology. 2014; 59: 1060-1072
        • Parker R.
        • Weston C.J.
        • Miao Z.
        • Corbett C.
        • Armstrong M.J.
        • Ertl L.
        • et al.
        CC chemokine receptor 2 promotes recruitment of myeloid cells associated with insulin resistance in nonalcoholic fatty liver disease.
        Am J Physiol Gastrointest Liver Physiol. 2018; 314: G483-G493
        • Marques P.E.
        • Oliveira A.G.
        • Pereira R.V.
        • David B.A.
        • Gomides L.F.
        • Saraiva A.M.
        • et al.
        Hepatic DNA deposition drives drug-induced liver injury and inflammation in mice.
        Hepatology. 2015; 61: 348-360
        • Cai J.
        • Li H.
        • Zhang C.
        • Chen Z.
        • Liu H.
        • Lei F.
        • et al.
        The neutrophil-to-lymphocyte ratio determines clinical efficacy of corticosteroid Therapy in patients with COVID-19.
        Cell Metab. 2021; 33: 258-269 e253
        • Schnell A.
        • Bod L.
        • Madi A.
        • Kuchroo V.K.
        The yin and yang of co-inhibitory receptors: toward anti-tumor immunity without autoimmunity.
        Cell Res. 2020; 30: 285-299
        • Tang Z.S.
        • Hao Y.H.
        • Zhang E.J.
        • Xu C.L.
        • Zhou Y.
        • Zheng X.
        • et al.
        CD28 family of receptors on T cells in chronic HBV infection: expression characteristics, clinical significance and correlations with PD-1 blockade.
        Mol Med Rep. 2016; 14: 1107-1116
        • Zhang Y.
        • Zhou Y.
        • Lou J.
        • Li J.
        • Bo L.
        • Zhu K.
        • et al.
        PD-L1 blockade improves survival in experimental sepsis by inhibiting lymphocyte apoptosis and reversing monocyte dysfunction.
        Crit Care. 2010; 14: R220
        • Wang B.J.
        • Bao J.J.
        • Wang J.Z.
        • Wang Y.
        • Jiang M.
        • Xing M.Y.
        • et al.
        Immunostaining of PD-1/PD-Ls in liver tissues of patients with hepatitis and hepatocellular carcinoma.
        World J Gastroenterol. 2011; 17: 3322-3329
        • Mueller S.N.
        • Vanguri V.K.
        • Ha S.J.
        • West E.E.
        • Keir M.E.
        • Glickman J.N.
        • et al.
        PD-L1 has distinct functions in hematopoietic and nonhematopoietic cells in regulating T cell responses during chronic infection in mice.
        J Clin Invest. 2010; 120: 2508-2515
        • Niu B.
        • Zhou F.
        • Su Y.
        • Wang L.
        • Xu Y.
        • Yi Z.
        • et al.
        Different expression characteristics of LAG3 and PD-1 in sepsis and their synergistic effect on T cell exhaustion: a new strategy for immune checkpoint blockade.
        Front Immunol. 2019; 10: 1888
        • Brahmamdam P.
        • Inoue S.
        • Unsinger J.
        • Chang K.C.
        • McDunn J.E.
        • Hotchkiss R.S.
        Delayed administration of anti-PD-1 antibody reverses immune dysfunction and improves survival during sepsis.
        J Leukoc Biol. 2010; 88: 233-240
        • Guignant C.
        • Lepape A.
        • Huang X.
        • Kherouf H.
        • Denis L.
        • Poitevin F.
        • et al.
        Programmed death-1 levels correlate with increased mortality, nosocomial infection and immune dysfunctions in septic shock patients.
        Crit Care. 2011; 15: R99
        • Boomer J.S.
        • To K.
        • Chang K.C.
        • Takasu O.
        • Osborne D.F.
        • Walton A.H.
        • et al.
        Immunosuppression in patients who die of sepsis and multiple organ failure.
        JAMA. 2011; 306: 2594-2605
        • Bone R.C.
        Immunologic dissonance: a continuing evolution in our understanding of the systemic inflammatory response syndrome (SIRS) and the multiple organ dysfunction syndrome (MODS).
        Ann Intern Med. 1996; 125: 680-687
        • Salunkhe S.A.
        • Chitkara D.
        • Mahato R.I.
        • Mittal A.
        Lipid based nanocarriers for effective drug delivery and treatment of diabetes associated liver fibrosis.
        Adv Drug Deliv Rev. 2021; 173: 394-415
        • Matsumoto A.
        • Tsujimoto H.
        • Ono S.
        • Kinoshita M.
        • Habu Y.
        • Kawabata T.
        • et al.
        Loss of hepatic B cells following lipopolysaccharide injection and polymicrobial sepsis.
        J Gastroenterol Hepatol. 2009; 24: 262-269
        • Kao Y.H.
        • Chen P.H.
        • Wu T.Y.
        • Lin Y.C.
        • Tsai M.S.
        • Lee P.H.
        • et al.
        Lipopolysaccharides induce Smad2 phosphorylation through PI3K/Akt and MAPK cascades in HSC-T6 hepatic stellate cells.
        Life Sci. 2017; 184: 37-46
        • Gonnert F.A.
        • Kunisch E.
        • Gajda M.
        • Lambeck S.
        • Weber M.
        • Claus R.A.
        • et al.
        Hepatic fibrosis in a long-term murine model of sepsis.
        Shock. 2012; 37: 399-407
        • Zhang W.
        • Conway S.J.
        • Liu Y.
        • Snider P.
        • Chen H.
        • Gao H.
        • et al.
        Heterogeneity of hepatic stellate cells in fibrogenesis of the liver: insights from single-cell transcriptomic analysis in liver injury.
        Cells. 2021; 10
        • Mederacke I.
        • Dapito D.H.
        • Affo S.
        • Uchinami H.
        • Schwabe R.F.
        High-yield and high-purity isolation of hepatic stellate cells from normal and fibrotic mouse livers.
        Nat Protoc. 2015; 10: 305-315