CD11b⁺CD43^hiLy6C^lo splenocyte‐derived macrophages exacerbate liver fibrosis via spleen–liver axis

Background and Aims:

Monocyte‐derived macrophages (MoMFs), a dominant population of hepatic macrophages under inflammation, play a crucial role in liver fibrosis progression. The spleen serves as an extra monocyte reservoir in inflammatory conditions; however, the precise mechanisms of involvement of the spleen in the pathogenesis of liver fibrosis remain unclear.

Approach and Results:

By splenectomy and splenocyte transfusion, it was observed that splenic CD11b⁺ cells accumulated intrahepatically as Ly6C^lo MoMFs to exacerbate CCl₄‐induced liver fibrosis. The splenocyte migration into the fibrotic liver was further directly visualized by spleen‐specific photoconversion with KikGR mice and confirmed by CD45.1⁺/CD45.2⁺ spleen transplantation. Spleen‐derived CD11b⁺ cells purified from fibrotic livers were then annotated by single‐cell RNA sequencing, and a subtype of CD11b⁺CD43^hiLy6C^lo splenic monocytes (sM‐1s) was identified, which was markedly expanded in both spleens and livers of mice with liver fibrosis. sM‐1s exhibited mature feature with high expressions of F4/80, produced much ROS, and manifested preferential migration into livers. Once recruited, sM‐1s underwent sequential transformation to sM‐2s (highly expressed Mif, Msr1, Clec4d, and Cstb) and then to spleen‐derived macrophages (sMφs) with macrophage features of higher expressions of CX₃CR1, F4/80, MHC class II, and CD64 in the fibrotic hepatic milieu. Furthermore, sM‐2s and sMφs were demonstrated capable of activating hepatic stellate cells and thus exacerbating liver fibrosis.

Conclusions:

CD11b⁺CD43^hiLy6C^lo splenic monocytes migrate into the liver and shift to macrophages, which account for the exacerbation of liver fibrosis. These findings reveal precise mechanisms of spleen–liver axis in hepatic pathogenesis and shed light on the potential of sM‐1 as candidate target for controlling liver diseases.

INTRODUCTION

Liver fibrosis is an outcome of liver injury hallmarked by excessive accumulation of extracellular matrix (ECM) produced by activated and transdifferentiated hepatic stellate cells (HSCs). Fibrogenesis is often preceded by chronic inflammation,1,2 and durable inflammation further triggers the development of cirrhosis and even hepatocellular carcinoma.3 In response to the inflammatory signals, activated immune cells are recruited to the injured liver, helping maintain the inflammatory hepatic environment and, thus, facilitate the progression of liver diseases.4,5 Emerging evidence suggests the spleen as an extra inflammatory cell resource beyond bone marrow in certain liver diseases, based on the results that the spleen affects the composition of hepatic immune cells.6–9 However, the precise mechanisms of involvement of the spleen in the pathogenesis of liver fibrosis have not been well understood.

The spleen and liver are both involved in immune response and connected anatomically and physiologically. All splenic blood flows into the liver through the portal vein. Pathologically, liver cirrhosis with portal hypertension often manifests with splenomegaly and hypersplenism. Changes in splenic blood microcirculation and spleen stiffness are capable of evaluating portal hypertension.10,11 Clinically, splenectomy ameliorates patients' condition.12–14 Moreover, a spleen–liver axis has been defined to describe the crosstalk between the spleen and liver.15–17 Spleen‐derived cytokines including transforming growth factor‐β1 (TGF‐β1), IL‐10, and TNF,18–20 chemokine like CCL2,21 and factors involved in metabolic activities16,22 have been shown to play essential roles in linking the spleen and liver.23 Yet, little evidence is provided for splenocytes acting as mediators of the spleen–liver axis, although a few studies in liver diseases have demonstrated altered composition of hepatic immune cells upon splenectomy.6–9,24 Therefore, the response of splenocytes to liver inflammatory signals and the use of cell tracing technology to study splenocyte migration deserve further investigation to understand the pathogenesis of the spleen–liver axis in liver diseases.

Macrophages are the most abundant immune cells in the liver, playing critical roles in liver inflammation. Liver macrophages include liver‐resident macrophages (Kupffer cells) and monocyte‐derived macrophages (MoMFs),25–27 and the latter are the major players during liver inflammation. MoMFs play a proinflammatory and profibrotic role by activating themselves,28–30 helping recruit leukocytes and HSCs, and establishing crosstalk with other cells.31–34 Bone marrow has long been considered the only source of circulating monocytes35; however, the spleen has lately been proven to serve as a monocyte reservoir to accommodate the demands of rapid‐onset inflammation in conditions such as ischemic myocardial injury, ischemic brain injury, atherosclerotic lesions, and hepatopulmonary syndrome.36–40 Therefore, exploration of the response of splenic monocytes in inflammatory liver diseases has predictable significance.

In this study, by using mouse models of liver fibrosis induced by both CCl₄ and a Western high‐fat diet (WD‐HFD), a subtype of spleen‐derived monocytes identified as CD11b⁺CD43^hiLy6C^lo cells are demonstrated to preferentially recruit into fibrotic liver and acquire macrophage features to exacerbate fibrogenesis. These findings provide strong evidence for the critical role of the spleen–liver axis in the progression of chronic liver disease.

MATERIALS AND METHODS

Mice

Adult male C57BL/6 mice aged 6–8 weeks were purchased from GemPharmatech Co., Ltd. B6.Cg‐Gt(ROSA)26Sor<tm1.1(CAG‐kikGR)Kgwa>(Riken BRC 09256, referred to as KikGR) mice were purchased from Riken BRC Experimental Animal Division. B6.SJL‐Ptprc ^a Pepc ^b /BoyJ (no. 002014, referred to as CD45.1⁺) mice were purchased from the Jackson Laboratory. All mice were maintained under specific pathogen‐free controlled conditions and received humane care according to the criteria outlined in the NIH Guide for the Care and Use of Laboratory Animals.

Mouse models

Liver fibrosis was induced by intraperitoneal injection of 0.4 μl g⁻¹ body weight CCl₄ (Sigma‐Aldrich) diluted 1:3 in olive oil twice a week for 6 weeks, and control mice were given olive oil. Otherwise, mice were fed with a WD‐HFD (with 60 kcal% fat) to induce NAFLD or a corresponding control diet for 10 weeks.

Patients

Liver samples were obtained from patients with hepatitis B cirrhosis‐related portal hypertension complicated with hypersplenism. These patients underwent a splenectomy and liver biopsy at The Second Affiliated Hospital of Xi'an Jiaotong University. Details on patient information are provided in Table S3.

Cell

The mouse HSC line JS1 was a gift from Jinsheng Guo of Department of Gastroenterology and Hepatology, Zhongshan Hospital (Fudan University, China) and cultured as described previously.41,42

KikGR photoconversion and intravital microscopy

KikGR transgenic mice were anesthetized with isoflurane (Sigma‐Aldrich) supplemented with oxygen (1%–3% vol, by inhalation), and the spleen was externalized gently, with two pieces of sterile aluminum foil arranged on either side of the spleen, shielding the skin and abdominal cavity. Then the spleen was irradiated 5 min/field with short wavelength laser light (405 nm) laser under a compound microscope (Zeiss, Germany) equipped with 5× objective. After photoconversion, the spleen was replaced, and the abdominal cavity and skin were closed. Twenty‐four hours after photoconversion, the mice were anesthetized and placed within a custom‐designed box, with the liver exposed for imaging. Imaging was performed on an inverted TCS SP8 laser scanning confocal microscope (Leica).

Spleen transplantation

Spleen donor mice (CD45.1⁺ or KikGR) were anesthetized and intravenously injected with heparin. The thorax was opened, and the right atrium was incised to allow blood to flow out during perfusion. Then, the donor spleen containing the splenic artery and splenic vein was removed and stored in ice cold heparin‐saline. The spleen of the anesthetized recipient (CD45.2⁺) was removed after ligation of the splenic vein and artery at the splenic hilar. After that, the donor spleen was placed at the right flank of recipient abdomen without twists. End‐to‐side anastomosis was performed on the donor's portal vein to the recipient's portal vein, as well as donor's aortic cuff to the recipient's infrarenal aorta. The abdominal cavity and skin were closed.

Single‐cell RNA sequencing

The viable CD11b⁺ cells were isolated and subjected to single‐cell RNA sequencing (scRNA‐seq) analysis using the Chromium microfluidic chips with 30 v3 Chemistry and barcoded with a 10× Chromium Controller (10X Genomics). RNA from the barcoded cells was subsequently reverse‐transcribed and sequencing libraries were constructed with reagents from a Chromium Single Cell 30 v3 reagent kit (10X Genomics) according to the manufacturer's instructions. Sequencing was performed with Illumina NovaSeq according to the manufacturer's instructions (Illumina).

scRNA‐seq data analysis

Raw fastqs were mapped to genome reference consortium mouse build 38 organism to generate count matrix by the Cell Ranger software (v4.0). Downstream analysis was performed by R package Seurat (v3.0.2), including quality control, principal component analysis, t‐distributed stochastic neighbor embedding, or Uniform Manifold Approximation and Projection dimension reduction and clustering. Cells with more than 500 genes and less than 8000 genes, and with less than 10% unique molecular identifiers (UMIs) derived from the mitochondrial genes, were retained for downstream analysis. Thirty components and 0.5 resolution were used in reduction and clustering, respectively; other parameters in each step were ran with the default value. Candidate cluster markers were calculated by function Find All Markers in Seurat. Genes from each cluster with average log fold change ≥0.5 and adjusted p value ≤ 0.05 were used as input gene sets for kyoto encyclopedia of genes and genomes (KEGG) and gene ontology (GO) enrichment analysis in R package Cluster Profiler (v3.12.0). RNA velocity was estimated by R package velocyto.R (v0.6) to infer the cell trajectory.

Statistical analysis

All data are shown as mean ± SEM. Student's t test, one‐way analysis of variance (ANOVA) and further least significant difference (LSD) for multiple comparisons, and Mann–Whitney rank sum test were performed for the statistical significance among groups. Regressive analysis was run for significance of correlation. p < 0.05 is considered to be statistically significant.

Statement

All animal experiments were ethically approved by the Animal Use and Management Committee of the Xi'an Jiaotong University of China. All research involving human samples was conducted in accordance with both the Declarations of Helsinki and Istanbul. All participants provided written informed consent. All procedures were approved by the Ethics Committee of The Second Affiliated Hospital of Xi'an Jiaotong University of China (no. 2018479).

RESULTS

Splenic CD11b⁺ cells accumulate intrahepatically as Ly6C^lo MoMFs to facilitate liver fibrosis progression

Compared with Oil‐treated mice, mice treated with CCl₄ showed obvious liver fibrosis with increased levels of collagen III and α‐smooth muscle actin (α‐SMA) deposition, whereas splenectomy markedly reduced the deposition (Figure 1A). Along with fibrogenesis, MoMFs marked by CD11b^hiF4/80^int accumulated in livers (Figure 1B). The frequency of MoMFs showed close correlation with the fibrosis area (Figure 1C), and accumulation of MoMFs was impaired upon spleen removal (Figure 1D). Moreover, the accumulated MoMFs in livers of CCl₄‐treated mice were mainly composed of the Ly6C^lo group (Figure 1E), which also significantly decreased in splenectomy mice (Figure 1F). These findings suggest the spleen be responsible for the accumulated MoMFs, especially the Ly6C^lo ones in fibrotic liver, and thus exacerbate the progression of liver fibrosis.

FIGURE 1.

Spleen facilitates intrahepatic accumulation of Ly6C^lo monocyte‐derived macrophages (MoMFs) to exacerbate liver fibrosis. (A) Representative histological images of hematoxylin and eosin (H&E), Sirius Red, collagen III, and α‐smooth muscle actin (α‐SMA) staining in formalin‐fixed and paraffin‐embedded liver tissue from mice treated with oil or CCl₄, and quantification of positive area fraction. Scale bar, 100 μm (H&E and α‐SMA) or 200 μm (Sirius Red and collagen III). (B) Representative cytometry for gating MoMFs (CD45⁺Ly6G⁻CD11b^hiF4/80^int) from livers of mice treated with oil or CCl₄. Frequencies and absolute numbers of MoMFs were then calculated. (C) Correlation analysis of hepatic MoMF frequency and fibrotic area. (D) Representative cytometry for gating MoMFs from livers of sham‐operated (Sham) or splenectomy (Spx) mice treated with CCl₄. Frequencies and absolute numbers of MoMFs were then calculated. (E) Representative cytometry of Ly6C expression on hepatic MoMFs of mice treated with oil or CCl₄. Absolute numbers of Ly6C^hi and Ly6C^lo MoMFs were then calculated. (F) Representative cytometry of Ly6C expression on hepatic MoMFs of Sham or Spx mice. Absolute numbers of Ly6C^hi and Ly6C^lo MoMFs were then calculated. Data are shown as the mean ± SEM. Regression analysis and the Student's t test are performed for statistical significance. *, **, and *** stand for p < 0.05, <0.01, and <0.001, respectively.

To verify the candidate cells that account for the role of spleen in liver fibrosis, splenic CD11b⁺ or CD11b⁻ cells were transfused to splenectomy mice. Compared with the splenectomy mice, mice receiving the transfusion CD11b⁺ cells, but not CD11b⁻ cells, showed exacerbated liver fibrosis (Figure 2A, Figure S1A). Also, CD11b⁺ cell transfusion facilitated the accumulation of mainly Ly6C^lo MoMFs and a few Ly6C^hi ones (Figure 2B,C), and this was independent of the cell sourcing from Oil‐ or CCl₄‐treated mice (Figure S1B.

FIGURE 2.

Splenic CD11b+ cells accumulate Ly6C^lo monocyte‐derived macrophages (MoMFs) to exacerbate liver fibrosis. (A) Representative histological images of Sirius Red, collagen III, and α‐SMA staining in formalin‐fixed and paraffin‐embedded liver tissues, and quantification of positive area fraction. Scale bar, 100 μm (collagen III and α‐SMA) or 200 μm (Sirius Red). (B) Representative cytometry for gating MoMFs from livers of mice treated with CCl₄ that underwent different operations. Absolute numbers of MoMFs were then calculated. (C) Representative cytometry of Ly6C expression on hepatic MoMFs. Absolute numbers of Ly6C^hi and Ly6C^lo MoMFs were then calculated. Male C57BL/6 mice intraperitoneally injected with CCl₄ for 6 weeks were divided into four groups and underwent sham operation (Sham), splenectomy (Spx), or adoptive transfusion (AT) with splenic CD11b⁺ or CD11b cells 2 weeks after Spx (Spx+CD11b⁺AT or Spx+CD11b⁻AT), respectively. Data are shown as the mean ± SEM. One‐way analysis of variance (ANOVA) and multiple comparisons (LSD) after ANOVA were performed for statistical significance. *, ** and, *** stand for p < 0.05, <0.01, and <0.001, respectively.

Migration of splenic CD11b⁺ cells to the liver was confirmed by cell transfusion and tracking (Figure 3A). While being transfused to Oil‐treated recipients, splenic CD11b⁺ cells from CCl₄‐treated mice showed significant migration advantage to livers than cells from Oil‐treated mice, but the advantage was faded in CCl₄‐treated recipients (Figure 3B). In consideration of the possible competitive migration of splenic cells from CCl₄‐treated recipients with the donor cells, cell transfusion was further performed in recipients with spleen (SP⁺) or without spleen (SP⁻). In Oil‐treated recipients, splenic CD11b⁺ cells from both donors showed advantageous migration to SP⁻ recipients, but only cells from CCl₄‐treated donor mice showed advantageous migration to CCl₄‐treated SP⁻ recipients (Figure 3C). Migration of splenic CD11b⁺ cells to other organs showed no significant difference (Figure 3D). These data demonstrate that splenic CD11b⁺ cell migration to livers account for the role of the spleen in facilitating intrahepatic Ly6C^lo MoMF accumulation and exacerbating liver fibrosis progression.

FIGURE 3.

Splenic CD11b⁺ cells from mice with liver fibrosis manifest advantageous migration into liver. (A) Scheme of splenic cell adoptive transfusion assays. A total of 1 × 10⁶ splenic CD11b⁺ cells sorted from either Oil‐ or CCl₄‐treated KikGR mice were adoptively transferred to Oil‐ or CCl₄‐treated wild‐type (WT) mice, repeatedly. KikGreen cells (KikG+) migrated to the recipient livers were detected by flow cytometry 24 h later. In some experiments (D), spleens of the recipient mice were removed 2 weeks before cell transfusion. (B) Representative cytometry for gating transfused splenic KikG⁺CD11b⁺ cells in livers of recipient mice. Recruitment index of transfused KikG⁺ cells in liver single cells and CD11b⁺ cells in KikG⁺ cells. (C) Recruitment index of transfused KikG⁺ cells in livers of recipients with (SP+) or without spleen (SP−). (D) Recruitment index of transfused KikG+ cells in lymph node (LN), lung, bone marrow (BM), and peripheral blood (PB). Recruitment index: frequency of target cells in the recipient liver / frequency of target cells in transfused splenic cells. Data are shown as the mean ± SEM. Mann–Whitney rank sum test and the Student's t test were performed for statistical significance. *, **, and *** stand for p < 0.05, <0.01, and <0.001, respectively. FCM, flow cytometry; i.v., intravenous injection; NS, no significance.

Splenic CD11b⁺ cells are visualized to accumulate intrahepatically in mice with liver fibrosis

To obtain in vivo evidence for splenic cells migration to liver, we used photoconvertible KikGR transgenic mice.43 Upon site‐specific irradiation at 405 nm, KikGreen splenic cells were converted into KikRed (Figure 4A) and discovered in the liver within 24 h (Figure 4B). Increased migration of KikRed splenic cells into livers was observed in mice treated with CCl₄ (Figure 4D). Furthermore, in the short‐term spleen transplantation, there were also more CD45.1⁺ splenic cells, especially CD45.1⁺CD11b⁺ cells, detected in livers of CD45.2⁺ mice that received spleens from a CCl₄‐treated CD45.1⁺ donor (Figure 4D).

FIGURE 4.

Splenic CD11b⁺ cells are visualized to accumulate intrahepatically in mice with liver fibrosis. (A) Imaging of photoconverted spleen, and representative cytometry for photoconversion efficiency of splenocytes. (B) Representative two‐photon in vivo imaging of livers of mice received spleen‐specific photoconversion. (C) Representative imaging of livers of mice undergoing different treatments followed by spleen‐specific photoconversion. KikRed⁺ splenocytes in livers were then counted. (D) Short‐term (ST) spleen transplantation experiment. Representative cytometry for gating donor spleen‐derived CD45.1⁺Ly6G⁻CD11b⁺ cells in livers of ST spleen transplantation mice. Frequencies of the cells in hepatic CD45⁺ cells and absolute numbers of the cells in livers were then calculated. (E) Long‐term (LT) spleen transplantation experiment. Representative cytometry for gating donor spleen‐derived CD45.1⁺Ly6G⁻CD11b⁺ cells in livers of LT spleen transplantation mice. Absolute numbers of CD45.1⁺ and CD45.1⁺Ly6G⁻CD11b⁺ cells in livers were then calculated. (F) Frequencies of donor spleen‐derived CD45.1⁺ cells and CD45.1⁺Ly6G⁻CD11b⁺ cells in single cells of lung, lymph node (LN), bone marrow (BM), and peripheral blood (PB). Data are shown as the mean ± SEM. Student's t test was performed for statistical significance. *, **. and *** stand for p < 0.05, <0.01, and <0.001, respectively. CCl4‐1w, mice treated with CCl4 for one week; FCM, flow cytometry; i.p., intraperitoneal; Oil‐1w, mice treated with oil for one week.

To observe long‐term (LT) maintenance of spleen‐derived CD11b⁺ cells in the liver, we traced splenic cell distribution in vivo for a 6‐week duration. The results also showed more abundant CD45.1⁺CD11b⁺ cells in livers of recipient mice treated with CCl₄ (Figure 4E). The results were further confirmed in KikGR/wild‐type LT spleen transplantation (Figure S2A). Besides, we observed an increased distribution of CD45.1⁺CD11b⁺ cells in peripheral blood (Figure 4F), suggesting the active mobilization of splenic cells upon liver fibrosis. These data provide solid evidence for the migration and accumulation of spleen‐derived CD11b⁺ cells in fibrotic livers.

Spleen‐derived monocytes‐1 as a specific subtype from splenic CD11b⁺ cells are identified in fibrotic liver

To depict the unique characteristics of these specially accumulated splenic CD11b⁺ cells, CD45.1⁺CD11b⁺ cells were isolated from livers of CCl₄‐treated CD45.1⁺/CD45.2⁺ spleen transplantation mice (fibrotic spleen‐derived myeloid cells; FSDMs) and profiled with scRNA‐seq. Similarly, CD11b⁺ cells from spleens or livers of Oil‐ or CCl₄‐treated mice, respectively, were also profiled (Figure 5A). Based on graph‐based clustering and known marker expression, 27 immune cell clusters across nine lineages were identified, among which monocytes/macrophages (Monos) and neutrophils accounted for the major groups (Figure S3A, B). Each lineage was denoted by their unique signature genes (Figure S3C). Quality control metrics were highly reproducible among individual samples and conditions (Figure S4 and Supporting Materials S1 and S2).

FIGURE 5.

Specific monocyte/macrophage subtypes derived from splenic CD11b⁺ cells are identified in fibrotic livers (A) Scheme of cell isolation, and single‐cell RNA sequencing (scRNA‐seq) of CD11b⁺ myeloid cells from indicated samples. Five cell samples were prepared for scRNA‐seq as follows: spleen and liver CD11b⁺ cells were sorted from pools of three Oil‐treated or three CCl₄‐treated mice, and corresponding spleen and liver cells were derived from the same mice; the fibrotic spleen‐derived myeloid cells (FSDMs) were CD45.1⁺CD11b⁺ cells isolated from a mixture of liver cells from nine CCl₄‐treated CD45.1⁺/CD45.2⁺ spleen transplantation mice. (B) Clustering 8064 monocytes/macrophages (Monos) from different samples. Uniform Manifold Approximation and Projection (UMAP) plots display single‐cell RNA sequencing data with annotated clusters. (C) Cell count and cell percentage of each Mono cell cluster of the five samples. (D) Annotation of pseudotemporal dynamics among Mono clusters (left). Pseudotemporal trajectory and directionality among Mono clusters revealed by RNA velocity (right). (E) Gene expression patterns of Spn2, Treml4, Cx3cr1, Ly6c2, and Ccr2 among Mono clusters. (F) Violin plots of relative marker gene expressions of Mono‐3 (sM‐2) and Mono‐5 (sMφ). i.p., intraperitoneal; SP, spleen.

Among the sequenced cells, Monos showed an obvious enrichment in FSDMs and a striking heterogeneity in livers (Figure 5B, Figure S3). The 8064 Monos from the five samples revealed five clusters. Mono‐2 showed expansion in both splenic and hepatic Monos of CCl₄‐treated mice and was the dominant population of Monos in FSDMs. Mono‐3 and Mono‐4 were enriched in FSDMs and CCl₄‐treated livers but not present in spleens. Mono‐5 was abundant only in FSDMs but slightly present in other samples. Mono‐1 was also a large Mono subpopulation but showed no dominance in the liver and spleen of CCl₄‐treated mice (Figure 5B). These differences were validated by quantification of cell counts and cell percentages (Figure 5C).

After that, the sequential relationship of the Mono clusters was realized by mapping them along a pseudotemporal trajectory, and their directionality was interrogated via spliced and unspliced mRNA ratios.44 Both results suggested the sequential relationship of “Mono‐2–Mono‐4–Mono‐3–Mono‐5” (Figure 5D). We assumed Mono‐2 to be the initial spleen‐derived population that migrated to fibrotic livers and named it as spleen‐derived monocyte (sM)‐1, named Mono‐3 as sM‐2, and considered Mono‐4 as a neglected cluster for its small count. Specific signatures for Mono‐2 were identified, among which Spn (encoding CD43 protein), Treml4, and Cx3cr1 were highly expressed, and Ly6c2 and Ccr2 were lowly expressed (Figure 5E). Therefore, we defined sM‐1s as CD11b⁺ Ly6G⁻CD43^hiLy6C^lo cells. Genes highly expressed in Mono‐3 (Msr1, Clec4d, Mif, Cstb) also showed impressive expression levels in Mono‐5; however, Mono‐5 specifically expressed the macrophage signature molecules such as the Mertk, Adgre1, Fcgr1, and H2 family (Figure 5F). We named Mono‐5 as spleen‐derived macrophage (sMφ).

sM‐1s marked with CD11b⁺CD43^hiLy6C^lo function in much mature status and prefer migrating into livers of mice with liver fibrosis

Flow cytometry analysis confirmed the expansion of CD11b⁺Ly6G⁻CD43^hiLy6C^lo‐gated sM‐1s in spleens of CCl₄‐treated (Figure 6A) and WD‐HFD mice (Figure 6B). Consistent with the scRNA‐seq data, the CD43^hiLy6C^lo subpopulation of CD11b⁺Ly6G⁻ cells had the highest frequencies of both CX₃CR1⁺ and TREML4⁺ cells (Figure S5A, B), and these cells had a similar mRNA profile and pathway enrichment to Mono‐2 in the scRNA‐seq data (Figure S5C, D). Besides, CD43^hiLy6C^lo cells had higher level of reactive oxygen species (ROS) production and lower levels of major histocompatibility complex II (MHC II) and CD11c expression, which was distinguished from the CD43^loLy6C^lo subpopulation but similar to the CD43^intLy6C^hi subset. Additionally, CD43^hiLy6C^lo cells expressed a high level of F4/80, indicating a much mature status (Figure 6C).

FIGURE 6.

Splenic monocyte (sM)‐1s marked with CD11b⁺CD43^hiLy6C^lo function in much mature status and prefer migrating into livers of fibrosis mice. (A) Representative cytometry for gating splenic CD11b⁺CD43^hiLy6C^lo, CD11b⁺CD43^intLy6C^hi, and CD11b⁺CD43^loLy6C^lo cells of mice treated with oil or CCl₄. Frequencies and absolute numbers of splenic CD11b⁺CD43^hiLy6C^lo cells were then calculated. (B) Representative cytometry for gating splenic CD11b⁺CD43^hiLy6C^lo of mice fed with normal diet or Western high‐fat diet (WD‐HFD). Frequencies of splenic CD11b⁺CD43^hiLy6C^lo cells were then calculated. (C) Representative cytometry of ROS, F4/80, MHC II, and CD11c expression on different subtypes of splenic CD11b⁺ cells. Percentages of ROS⁺ and MHC II⁺ cells and MFI of F4/80 and CD11c were then calculated. (D) Representative cytometry for gating hepatic CD11b⁺CD43^hiLy6C^lo cells of mice treated with oil or CCl₄. Frequencies and absolute numbers of the cells were then calculated. (E) Representative cytometry for gating hepatic CD11b⁺CD43^hiLy6C^lo cells of mice fed with normal diet or WD‐HFD. Frequencies of hepatic CD11b⁺CD43^hiLy6C^lo cells were then calculated. (F) Frequencies of donor spleen‐derived CD45.1⁺ cells in hepatic CD11b⁺CD43^hiLy6C^lo or CD11b⁺ cells, and CD11b⁺CD43^hiLy6C^lo cells in hepatic CD45.1⁺ or CD45.2⁺ cells of short‐term and long‐term CD45.1⁺/CD45.2⁺ spleen transplantation mice. Data are shown as the mean ± SEM. Mann–Whitney rank sum test and the Student's t test were performed for statistical significance. *, **, and *** stand for p < 0.05, <0.01, and <0.001, respectively. MFI, median fluorescence intensity; MHC II, major histocompatibility complex II; ROS, reactive oxygen species.

Expansion of sM‐1s was also confirmed in livers of CCl₄‐treated (Figure 6D) and WD‐HFD mice (Figure 6E). The intrahepatic sM‐1s had similar CX₃CR1, higher F4/80, and lower MHC II expression compared with their counterparts in the spleen (Figure S6). Within livers of the CCl₄‐treated CD45.1⁺/CD45.2⁺ spleen transplantation mice, the frequency of CD45.1⁺ spleen‐derived cells in sM‐1s was higher than that in CD11b⁺ cells, and the proportion of sM‐1s in CD45.1⁺ cells was higher than that in CD45.2⁺ cells (Figure 6F). These findings demonstrate preferential migration of splenic sM‐1s into fibrotic livers.

sM‐1s are trained to be pathogenic sM‐2s and sMφs in fibrotic hepatic milieu

To study the intrahepatic development of the recruited sM‐1s, splenic sM‐1s were treated with fibrotic hepatocyte conditioned medium (liver‐CM) to mimic their transformation in the hepatic environment. Compared with freshly sorted sM‐1s, liver‐CM–treated sM‐1s had increased expressions of Cstb, Mif, Msr1, and Clec4d (Figure 7A), which were highly expressed in both sM‐2 and sMφ. Liver‐CM–treated sM‐1s also had decreased ROS production (Figure 7B), decreased MHC II, and slightly increased F4/80 (encoded by Adgre1) and Ly6C expressions (Figure 7C), and the changes in expressions of F4/80 and MHC II were similar with their differential expressions between liver and spleen sM‐1s (Figure S6). Moreover, liver‐CM–treated sM‐1s had a more similar mRNA profile to sM‐2 (Figure S7A) than that of sMφ (Figure S7B), indicating that liver‐CM treatment has led to the transformation of sM‐1 to sM‐2, which might be a transitional state before transforming into sMφ. Cytokines highly expressed in the liver‐CM of CCl₄‐treated mice (IL‐1α, IL‐12, IL‐13, and IL‐6) may play important roles in supporting cell survival and promoting differentiation of monocyte to macrophage (Figure 7D, Figure S8).

FIGURE 7.

Spleen‐derived monocyte (sM)‐1 trained in fibrotic hepatic milieu gain sM‐2 and spleen‐derived macrophage (sMφ) features. (A) Indicated gene expression of freshly sorted sM‐1s and fibrotic hepatocyte conditioned medium (liver‐CM)–treated sM‐1s analyzed by real time quantitative PCR (RT‐qPCR). sM1, freshly sorted sM‐1s; Liver‐CM, liver‐CM–treated sM‐1s. (B) Representative cytometry of ROS expression on sM‐1s treated with liver‐CM or control complete medium (Control) for 24 h. Percentages of ROS⁺ cells were then calculated. (C) Representative cytometry of MHC II, F4/80, and Ly6C expression on differentially treated sM‐1s. Percentages of MHC II⁺ cells and MFI of F4/80 and Ly6C on differentially treated cells was then calculated. (D) Cytokine expression profiles of liver‐CM of Oil‐treated mice (Oil) and CCl₄‐treated mice (CCl₄) detected by Luminex. Class A, proinflammatory; Class B, anti‐inflammatory; Class C, cell survival and macrophage differentiation. (E) Scheme of the in vivo training experiment. Primary sM‐1s were sorted from spleens of CCl₄‐treated CD45.2⁺ mice and intrahepatically injected to CCl₄‐treated CD45.1⁺ mice. Features of freshly isolated sM1s and cells trained by the fibrotic hepatic milieu for 5 days were compared. (F) Representative cytometry of CX3CR1, F4/80, MHC II, and CD64 expression on freshly isolated sM‐1s and intrahepatically trained cells. Percentages of CX3CR1⁺, F4/80⁺, MHC II⁺, and CD64⁺ cells were then calculated. Data are shown as the mean ± SEM. Student's t test was performed for the statistical significance. *, **, and *** stand for p < 0.05, <0.01, and <0.001, respectively. i.p., intraperitoneal; FCM, flow cytometry; MFI, median fluorescence intensity; MHC II, major histocompatibility complex II; ROS, reactive oxygen species.

We also determined the phenotypic transition of sM‐1s in vivo. Freshly isolated splenic sM‐1s were intrahepatically injected to Oil‐ or CCl₄‐treated recipient mice (Figure 7E). There were more transferred cells observed in CCl₄‐treated recipients (data not shown), and these cells gained a higher expression of CX₃CR1, F4/80, MHC II, and CD64 within livers of both recipients (Figure 7F, Figure S9). These alterations were consistent with the changes among sM‐1 and sMφ (Figure 5F), further confirming the sequential development from sM‐1 to sM‐2 and then to sMφ.

To reveal the role of spleen‐derived Monos in fibrotic liver, several genes related to macrophage‐mediated HSC recruitment, activation, or survival were combined as a module for expression score analysis. The sM‐2 possessed the highest expression of Il1b, Tgfb1, and Vegfa, sMφ possessed the highest expression of Tnfa and Ccl3 (Figure S10A), and sM‐2 and sMφ had a significantly higher score in this module (Figure S10B), prompting them to play crucial roles in activating HSCs. Accordingly, the HSC line JS1 cocultured with liver‐CM–treated sM‐1s, but not freshly sorted ones, showed increased expression of ECMs (Figure 8A). Transfusion of splenic sM‐1s to CCl₄‐treated splenectomy mice resulted in exacerbated liver fibrosis (Figure 8B,C) and increased Ly6C^lo MoMF accumulation (Figure 8D). We also stained the human counterparts of sM‐1 as CD11b⁺CD16⁺CD43⁺ cells45,46 and observed localization of these cells in close vicinity to the fibrotic region with high expression of collagen I (Figure 8E, Figure S11) in cirrhotic liver tissues.

FIGURE 8.

Spleen‐derived monocyte (sM)‐2s and spleen‐derived macrophages (sMφs) derived from sM‐1s promote liver fibrosis. (A) Expression of hallmark genes for hepatic stellate cell (HSC) activation analyzed by real time quantitative PCR (RT‐qPCR) after coculture with freshly sorted or fibrotic hepatocyte conditioned medium (liver‐CM)–treated sM‐1s. sM‐1, freshly sorted sM‐1s; Liver‐CM sM‐1, liver‐CM–treated sM‐1s. (B) Representative histological images of hematoxylin and eosin (H&E), Sirius Red, collagen III, and α‐SMA staining in formalin‐fixed and paraffin‐embedded liver tissue, and quantification of positive area fraction. Scale bar, 100 μm (H&E, collagen III, and α‐SMA) or 200 μm (Sirius Red). (C) Expression of hallmarker genes for HSC activation analyzed by RT‐qPCR among mice that underwent different operations. (D) Absolute numbers of hepatic Ly6C^hi and Ly6C^lo monocyte‐derived macrophages (MoMFs). (C–D) Male C57BL/6 mice intraperitoneally injected with CCl₄ for 6 weeks were splenectomized (Spx) and divided into three groups and received adoptive transfusion (AT) of splenic sM‐1s or CD11b⁻ cells 2 weeks after Spx (Spx+sM‐1 AT or Spx+CD11b⁻AT), respectively. (E) Representative immunofluorescence images of CD11b⁺CD16⁺CD43⁺ cell and collagen I staining in optimal cutting temperature compound (O.C.T.)‐embedded frozen liver tissue. The fibrotic livers were representative images of tissues from three patients with hepatitis B cirrhosis‐related portal hypertension complicated with hypersplenism. Scale bar, 100 μm. Data are shown as the mean ± SEM. One‐way analysis of variance (ANOVA) and multiple comparisons (LSD) after ANOVA were performed for the statistical significance. *, **, and *** stand for p < 0.05, <0.01, and <0.001, respectively.

DISCUSSION

The present study found that a unique subtype of CD11b⁺CD43^hiLy6C^lo monocytes expands in spleens of mice with liver fibrosis. These cells are actively recruited to fibrotic liver and shift to macrophage phenotype, facilitate intrahepatic Ly6C^lo MoMFs accumulation, and exacerbate liver fibrogenesis by activating HSCs. Our findings provide direct evidence that splenocytes migrate into the fibrotic liver and account for the pathogenic population, illustrating a partial sketch of the spleen–liver axis involvement in liver diseases.

The spleen and liver are both important components of portal circulation; influx of splenic blood into liver through the portal vein enables substance exchange and cell interflow between these two organs.47 The spleen–liver axis has been proposed as playing roles in conditions including chronic liver disease15,48 and metabolism.16 However, the knowledge about the properties of the spleen–liver axis and the mechanism of the spleen–liver axis influencing liver pathological condition is very limited. Although the spleen has been suggested as acting as a source of profibrotic cytokines18,49 or immune cells6,7,24,50 to participate in the progression of chronic liver diseases, little direct evidence is provided. Our present study introduced diverse in vivo splenic cell‐tracing models of spleen‐specific photoconversion43 and spleen transplantation. By virtue of these models, increased migration of splenic CD11b⁺ cells to fibrotic livers was observed (Figure 4), and the intrahepatically located spleen‐derived CD11b⁺ cells were successfully isolated for further characterization

Generally, two different subsets of monocytes exist in mice.51,52 Classical monocytes marked with Ly6C^hiCCR2^hiCX₃CR1^lo are “vanguards” that originate from bone marrow and rapidly respond to inflammatory signals.53 Nonclassical monocytes marked with Ly6C^loCCR2^loCX₃CR1^hi with ambiguous sources rarely immigrate into peripheral tissues but abundantly patrol along blood vessel endothelia and extravasate into tissues during an inflammatory response.54,55 In our study, spleen‐derived CD11b⁺ cells isolated from fibrotic livers were annotated by scRNA‐seq, and a subtype of monocytes (sM‐1s) marked with CD11b⁺CD43^hiLy6C^lo was identified. The sM‐1s were characterized with “nonclassical”‐like phenotype of high CX₃CR1 expression but low CCR2 and Ly6C expression (Figure 5). Generally, Ly6C^hi monocytes as vital effector cells are actively mobilized to peripheral tissues where they downregulate Ly6C expression and differentiate into macrophages.9,56,57 Ly6C^lo monocytes are supposed to be a mature monocyte subtype and have longer half‐life.58,59 Consistently, in our study, sM‐1s present a much mature status with higher expression of F4/80 than the Ly6C^hi subpopulation. However, the cells also have a similar high production of ROS with Ly6C^hi subtype, indicating that they might be in the effector state. Although there have been reports of Ly6C^lo monocytes differentiating into inflammatory macrophages,60 they are more likely to function by inducing the differentiation of Ly6C^hi monocytes,61 helping recruit active immune cells,62 or secreting growth factors.63,64

The sequential relationship analysis in the present study revealed a sequential development of sM‐1 to sM‐2, which gained higher expressions of Cstb, Mif, Msr1, and Clec4d and enhanced cytokine activity, and then to sMφ, which gained expressions of macrophage function–associated molecules like MERTK,65 F4/80, and CD6427,66 (Figure 7, Figures S7 and S10). Nonetheless, the in vitro induced “sM‐2” in our system also had some characteristics of sMφ (e.g., increased F4/80), so we thought that sM‐2 might be a transition state of sMφ. Unexpectedly, the similar transition was observed in sM‐1s transfused to Oil‐treated recipients, and this might be attributed to fewer sM‐1s recruited to liver. After that, these transformed cells of both sM‐2s and sMφs were verified to exacerbate liver fibrogenesis by activating HSCs. These results indicate that differentiation of sM‐1s themselves into profibrotic macrophages might be a crucial means by which sM‐1s participate in liver fibrosis. However, the profibrotic function of sM‐1s and their “descendants” needs to be further confirmed by intervention like cell‐specific depletion. The potential factors for inducing splenic sM‐1s expansion or mediating their intrahepatic transformation have been preliminarily explored, but more details remain to be elucidated. Moreover, the existence of sM‐1s in NAFLD mice and in patients with cirrhotic portal hypertension has been verified, and it would be interesting to investigate the contribution of these cells to fibrogenesis in clinical liver diseases and rodent models closer to human liver diseases.

Identification of the CD11b⁺CD43^hiLy6C^lo monocyte subset might provide a therapeutic strategy for chronic liver disease by targeting unique cells in the spleen rather than splenectomy. Antagonists against the chemokine receptor CCR2 have shown promising effects in ameliorating liver injury through inhibiting Ly6C^hi monocyte recruitment.53,54,67 CD43 appears to function by antiadhesion68,69 and suppressing activation70 in immune cells. Thus, we make the hypothesis that the high expression level of CD43 on sM‐1s might facilitate their emigration from spleen to liver, and the following downregulation of CD43 in liver might promote their maturation. Therefore, our present study highlights CD43 as a potential therapeutic target. CX₃CR1 is highly expressed on sM‐1s, and the CX₃CR1 antagonist has been proven a potent inhibitor of atherosclerotic lesions progression.71 Targeted interruption of the spleen–liver axis while preserving other physiological functions of the spleen, for example, by blocking CD43 and CX₃CR1, might be effective for clinical practice.

In summary, our work describes a subtype of splenic monocytes that preferentially migrate into fibrotic liver, shift to macrophage phenotype, and function as the pathogenic population in fibrogenesis by activating HSCs. Our findings propose an attractive mode of the spleen–liver axis involved in liver disease and indicate a particular process of spleen‐derived monocytes developing into pathogenic macrophages in the liver. Identifying the features of the distinct spleen‐derived Monos and signals that instructing the recruitment and transition of these cells may provide strategies for chronic liver disease therapy by targeting the source.