If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Translational Medicine Research Center, Fourth Medical Center and Medical Innovation Research Division of the Chinese PLA General Hospital, Beijing 100048, ChinaDepartment of Pulmonary and Critical Care Medicine, Beijing Chaoyang Hospital, Capital Medical University, Beijing 100020, China
Translational Medicine Research Center, Fourth Medical Center and Medical Innovation Research Division of the Chinese PLA General Hospital, Beijing 100048, China
Translational Medicine Research Center, Fourth Medical Center and Medical Innovation Research Division of the Chinese PLA General Hospital, Beijing 100048, China
scRNA-seq reveals the cellular composition and dynamic transcriptomic landscape of major liver nonparenchymal cells in sepsis-induced acute liver dysfunction.
2.
scRNA-seq analysis and immunofluorescence identify the endothelial cell (CD31+Sele+Glut1+) and neutrophil (Ly6G+Lta4h+Sort1+) subsets that were closely associated with acute liver dysfunction.
3.
The ATF4 pathway activation is a potential mechanism contributing to sepsis-induced acute liver dysfunction, and ATF4 inhibition alleviates sepsis-induced acute liver dysfunction and prolongs the survival of septic mice.
Abstract
Background & Aims
Sepsis-induced acute liver dysfunction often occurs early in sepsis and can exacerbate the pathology by triggering multiple organ dysfunction and increasing lethality. Nevertheless, our understanding of the cellular heterogeneity and dynamic regulation of major nonparenchymal cell lineages remains unclear.
Methods
Here, single-cell RNA sequencing (scRNA-seq) was used to profile multiple nonparenchymal cell subsets and dissect their crosstalk during sepsis-induced acute liver dysfunction in a clinically relevant polymicrobial sepsis model. The transcriptomes of major liver nonparenchymal cells from control and sepsis mice were analyzed. The alterations in the endothelial cell and neutrophil subsets that were closely associated with acute liver dysfunction were validated using multiplex immunofluorescence staining. Additionally, the therapeutic efficacy of inhibiting activating transcription factor 4 (ATF4) in sepsis and sepsis-induced acute liver dysfunction was explored.
Results
Our results present the dynamic transcriptomic landscape of major nonparenchymal cells at single-cell resolution. We observed significant alterations and heterogeneity in major hepatic nonparenchymal cell subsets during sepsis. Importantly, we identified endothelial cell (CD31+Sele+Glut1+) and neutrophil (Ly6G+Lta4h+Sort1+) subsets that were closely associated with acute liver dysfunction during sepsis progression. Furthermore, we found that ATF4 inhibition alleviated sepsis-induced acute liver dysfunction, prolonging the survival of septic mice.
Conclusion
These results elucidate the potential mechanisms and subsequent therapeutic targets for the prevention and treatment of sepsis-induced acute liver dysfunction and other liver-related diseases.
Acute liver dysfunction often occurs in the early stage of sepsis and has a significant effect on the severity and prognosis of sepsis. To date, specific therapeutic measures in current clinical practice, to resolve sepsis-induced acute liver dysfunction other than liver transplantation, is still lacking.
However, the shortage of liver sources as well as lifelong immunosuppression and a high medical cost associated with liver transplantation has resulted in an urgent need for alternative therapeutic interventions that is effective in treating sepsis-induced acute liver dysfunction.
Clarifying the pathogenesis of sepsis-induced acute liver dysfunction will provide opportunities to develop novel therapeutic interventions. Although previous studies have reported that acute liver dysfunction involves a complex interplay between multiple nonparenchymal cell lineages including neutrophils, endothelial and Kupffer cells,
their cellular heterogeneity and dynamic regulation leading to acute liver dysfunction remain poorly understood.
The rapid development of single-cell RNA sequencing (scRNA-seq) provides high-dimensional information about tissues and an unprecedented understanding of cellular composition, response, and crosstalk in a pathological state.
and not many have been conducted on acute liver diseases.
Therefore, in this study, we performed a scRNA-seq experiment on livers following sepsis-induced acute liver dysfunction in a clinically relevant polymicrobial sepsis model. Our analyses revealed the major cell types associated with acute liver dysfunction including neutrophils, endothelial and Kupffer cells, identified subclusters of each cell type, and uncovered their dynamic transformations and interactions. Notably, we reported the extensive activating transcription factor 4 (ATF4) activation in major hepatic cells, as well as the therapeutic efficacy of ATF4 inhibition for sepsis and sepsis-induced acute liver dysfunction.
Materials and methods
2.1 Animal care, sepsis model, and animal experiment
All experiments and procedures conformed to the National Institutes of Health guidelines, with the approval of the Institutional Animal Care and Use Committee of the AMMS. The male C57BL6/J mice (7–8 weeks) were obtained from Beijing Vital River Laboratories (Beijing, China), and housed according to the following standard laboratory procedures. After a minimum of 3 days of acclimatization, the cecal ligation and puncture (CLP) sepsis model was induced as described previously.
Briefly, mice were anesthetized with pentobarbital sodium (50 mg/kg). After sterilization, a midline laparotomy was conducted to expose the cecum, which was ligatured, and perforated with a sterile 21-gauge needle. Then a small droplet of feces was extruded from the puncture, and the cecum was returned into the peritoneal cavity. Subsequently, the mice were resuscitated via a subcutaneous administration of 1 mL of normal saline.
Nine mice were randomly allocated to control or sepsis groups; the normal group (CK, n = 3) acted as control group, whereas sepsis groups were subjected to CLP and sacrificed at 6 (n = 3) or 24 h (n = 3). The livers and blood from mice in the three groups were immediately collected for further processing.
2.2 Cell isolation
After being washed twice with cold normal saline, liver tissues from the same group were pooled and minced to < 0.5-mm cubic pieces, followed by enzymatic digestion (1 h at 37 °C) with manual shaking every 5 min. After filtering, dead cell removal, and lysis of red blood cells, the hepatic cells were counted using BD Rhapsody™ Scanner (BD Biosciences, CA, USA).
2.3 Single‐cell RNA-sequencing
Single-cell libraries were prepared using the BD Rhapsody Single-Cell Analysis System (BD Biosciences, CA, USA) following the manufacturer’s guidelines. Libraries were sequenced using multiple runs on an Illumina NovaSeq 6000 platform (Illumina, CA, USA) in a 2 × 150 bp paired-end mode.
2.4 Sequencing data analysis
The obtained RNA-sequencing data were processed into the expression matrix Fastq via the BD Rhapsody Analysis Pipeline (V1.9). After quality control, normalization and batch correction using R package “Seurat” and fastMNN, BD DataView software (BD Biosciences, CA, USA) and the R package “Seurat” (V3.1.1) was used to analyze the expression matrix.
To reduce the dimensionality of all data, principal component analysis was conducted in Seurat. For t-distributed stochastic neighbor embedding (tSNE) projection and clustering analysis, the cells were clustered based on a graph-based clustering approach and visualized in two dimensions using tSNE. The FindAllMarkers function in Seurat was used to identify the marker genes of each cluster, and the FindMarkers function was used to analyze the differentially expressed genes (DEGs). P value < 0.05 and |log2fold change| > 0.58 were set as thresholds for significantly differential expression.
The Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis was performed using the R package “enrichplot”. Cell–cell interaction analysis was performed using the CellChat R package (version 1.1.3).
Single-cell regulatory network inference and clustering (SCENIC) analysis was conducted using the motif database for RcisTarget and GRNboost (SCENIC version v1.1.2.2, which corresponds to AUCell v1.4.1 and RcisTarget v1.2.1). The “RcisTarget” package was used to identify transcription factor binding motifs and potential target genes (regulons). The activity of regulon group in each cell was classified using the “AUCell” package. The regulon specificity scores for each cell type were calculated using the “scFunctions” package.
2.5 Drugs administration
To evaluate the therapeutic effect of ISRIB, a total of 28 mice were randomly divided into control or experimental groups; the normal group (CK, n = 3) and the ISRIB-treated group (5 mg/kg, n = 3) acted as controls, whereas the experimental groups were intraperitoneally treated with ISRIB (5 mg/kg, n = 11) or an equal volume of vehicle (n = 11) 2 h before CLP. Mice were sacrificed 24 h after CLP, and the livers and plasma were collected and stored until assayed.
2.6 Blood biochemistry and bloodocyte analysis
Blood biochemistry was analyzed using a biochemical auto analyzer (Pointcare V2, MNCHIP, Tianjin, China). Complete blood counts were analyzed using a BC-5000 Vet auto hematology analyzer (Mindray, Shenzhen, China).
2.7 Lipid peroxidation, neutrophil infiltration, and interleukin (IL)-6 content in livers
The livers were homogenized in ice-cold saline (Shijiazhuang Siyao Ltd, Hebei, China) and centrifuged (1000 × g, 6 min, 4 °C). The obtained homogenates were then assayed for malondialdehyde (MDA) content, myeloperoxidase (MPO) activity, and IL-6 level, according to the manufacturer’s instructions, as described previously.
The paraformaldehyde-fixed livers were dehydrated, embedded in paraffin, and cut into 5–7-μm thick sections before being subjected to hematoxylin and eosin (HE) staining as described previously [12]. The severity level of liver damage was assessed for inflammatory infiltration, cell swelling, and tissue architecture disruption in a blinded fashion, and scored on a 4-point scale (0, none; 1, slight; 2, moderate; 3, severe).
For immunofluorescence assays, the liver sections were incubated with anti-CD31 (1:3000, Proteintech, IL, USA), anti-Glut1 (1:100, Proteintech), anti-Sele (1:2000, Proteintech), anti-Ly6G (1:3000, Servicebio, Wuhan, China), anti-Lta4h (1:1000, Proteintech), anti-Sort1 (1:400, Proteintech), anti-ATF4 (1:1000, ABclonal, Wuhan, China), anti-Fosl1 (1:1000, ABclonal), anti-NF-κB1(1:1000, ABclonal), and anti-F4/80 (1:500, Servicebio) primary antibodies, followed by washing and incubation with the fluorophore-labeled secondary antibody, and visualization using a confocal microscope (Nikon, Japan).
2.9 Survival experiments
A total of 100 mice were randomized into normal control or sepsis model groups. The sepsis model groups were intraperitoneally treated with ISRIB (8 mg/kg, n = 45) or an equivalent volume of vehicle (n = 45) 2 h before CLP. The normal mice acted as the control group.
2.10 Statistical analysis
Data are shown as mean ± standard error of mean. Statistical differences between groups were determined using one-way analysis of variance (ANOVA) or unpaired Student’s t-test. Survival data were analyzed using the log-rank test. *P＜0.05, **P＜0.01, and ***P＜0.001 were considered the thresholds for statistical significance of differences.
3. Results
3.1 Single-cell RNA-seq identified multiple cell populations in the liver from control and sepsis mice
To elucidate diverse hepatic nonparenchymal cell types and comprehensively characterize their dynamic changes during the pathological progression of sepsis-induced acute liver dysfunction, we performed scRNA-seq analysis of the livers at different time points (6 and 24 h) after sepsis using the BD Rhapsody platform (Figure 1A). These two time points reflected the major stages in the development of sepsis-induced acute liver dysfunction, corresponding to a progressive increase in liver injury. As shown in Figure 1B and C, the plasma levels of alanine aminotransferase (ALT) and aspartate aminotransferase (AST), which are used as biomarkers for acute liver dysfunction,
Neutrophil membrane-mimicking nanodecoys with intrinsic anti-inflammatory properties alleviate sepsis-induced acute liver injury and lethality in a mouse endotoxemia model.
gradually became elevated over time. Similarly, MDA content, MPO activity, and histological injury in the liver also increased gradually with time, indicating that sepsis could induce lipid peroxidation and neutrophil infiltration during its pathological progression (Figure 1D, E, and F). In addition, the sepsis model mice also exhibited acute kidney injury, had low level of glucose and total protein in blood (Figure S1A–D).
Figure 1Single-cell RNA sequencing (scRNA-seq) identified hepatic cell populations and specific markers in the mice liver. (A) Schematic diagram indicating the procedure for scRNA-seq. For each experimental condition, the hepatic cells were pooled from 3 mice in each group. (B) The levels of alanine aminotransferase (ALT) in plasma. Sham, CLP-6h, 6 h after cecal ligation and puncture (CLP); CLP-24h, 24 h after CLP. (C) The levels of aspartate aminotransferase (AST) in plasma. Myeloperoxidase (MPO) activity (D), and the malondialdehyde (MDA) content (E) in the liver. (F) Histological injury of liver (Scale bar, 100 μm). (G) UMAP plot of hepatic cells with colors denoting different cell type clusters. (H) UMAP plot of hepatic cells with colors denoting different clusters. (I) UMAP plot of cell clusters in hepatic cells across the indicated conditions. (J) The markers indicating group identities. (K) The proportion of hepatic cell populations in each sample. *p < 0.05; **p < 0.01; ***p < 0.001. Statistical differences between groups were assessed using one-way ANOVA for Fig. 1B, C, D, and E.
A total of 27,972 liver nonparenchymal cells and hepatocytes from control and sepsis mice were further analyzed after passing quality control metrics and having corrected for batch effect. Eight major cellular clusters consisting of 19 clusters (Marker genes shown in Figure S1F) were identified (Figure 1G–J), including endothelial cells (Ptprb, Aqp1, Igfbp7, Clec4g, Ehd3, Ushbp1, Oit3, Il1a, F8, Bmp2, C1qtnf1, Mmrn2, Pcdh12, Dpp4), Kupffer cells (C1qa, C1qb, C1qc, Clec4f, Csf1r, Adgre1, Clec4f, Irf7, Spic, Timd4, Marco), neutrophils (Csf3r, Retnlg, S100a8, S100a9, Slpi, Mmp9, Mmp8, Adam8), T/Natural killer (NK) cells (Nkg7), B cells (Igkc, Cd22, Cd79b, Cd19, Cd79a, Ebf1, Pax5), monocytes/monocyte-derived macrophages (Itgam, Ccr2), hepatocytes (Alb, Cyp2e1, Apob, Asgr1, Pck1, Hp, Ass1, Apoe), and hepatic stellate cells (DCN, Hgf, Col14a1, Col1a1, Col1a2, Col3a1, Colec11, Cxcl12, Cygb).
Additionally, the relative composition of different cell types across different stages of disease and healthy controls was investigated to uncover substantial changes during disease progression. A rapid increase of the neutrophil population was observed 6 h after CLP when comparing with controls. Subsequently, the proportion of neutrophils declined at 24 h after CLP compared to that at 6 h after CLP (Figure 1K). The relative proportion of the endothelial and B cells decreased with pathological progression, whereas that of monocytes/monocyte-derived macrophages increased with pathological progression. In addition, there was a slight decrease in the proportion of Kupffer cell population at 6 h after CLP compared with that of controls, but this proportion robustly increased at 24 h after CLP, which indicated a rapid supplement of Kupffer cells from blood circulation. These results suggested that different hepatic nonparenchymal cell types, especially immune populations changed dynamically at different time points and reacted collectively to acute liver dysfunction via distinct functions.
Pecam1(CD31), Adgre1(F4/80), and Ly6g, have been reported as cell makers of endothelial and Kupffer cells, and neutrophils, respectively.
The tSNE plots for Pecam1, Ly6g, and Adgre1 expressions in all cell clusters are shown in Figure S1E. Our results show that regions with high Ly6g, Pecam1, and Adgre1 expression are strongly correlated with the neutrophils, endothelial and Kupffer cells respectively.
Thus, further analysis on our scRNA-seq data was performed to investigate the transcriptional diversity of endothelial cells in liver.
The liver endothelial cells were grouped into four subclusters (Figure 2A and B); the ECs-1 and 2 accounted for up to 98.44% of liver endothelial cells under healthy conditions (Figure 2C), whereas endothelial cells in ECs-3 and 4 were nearly exclusively present in the livers of sepsis mice, and the proportion of endothelial cells in ECs-4 increased with the pathological progression of sepsis-induced acute liver dysfunction (Figure 2C), indicating that the ECs-4 subcluster represented a unique population of endothelial cells associated with the pathogenesis of sepsis-induced acute liver dysfunction. Based on a subsequent gene enrichment analysis, ECs-3 was found differentially expressing Zbp1, Batf2, and Isg15. Zbp1 can act as DNA sensor and mediate the recruitment of RIP1 and RIP3 to activate the nuclear factor-kappa B (NF-κB) and NLRP3 inflammasome pathways.
Figure 2Hepatic endothelial cell dynamics and functionality in healthy and septic livers. (A) Sub-clustering of endothelial cells in healthy and septic livers. (B) UMAP plot of endothelial cells across the indicated conditions. (C) The proportion of endothelial cell subclusters in each sample. (D) Heatmap plots demonstrating expression of specified genes among endothelial cell subclusters. (E-G) Heatmap showing gene expression by the indicated endothelial cell subclusters. (H) The enriched pathway of associated with endothelial cell subclusters 3 and 4. Colors indicate significance of enrichment and circle sizes indicate number of genes falling into respective categories. (J) Subpopulation-specific regulons of each subpopulation revealed by SCENIC analysis. (K) Immunofluorescence staining results showing the spatial distribution of the ECs-4 subpopulations in healthy and septic mouse livers. Scale bars, 20 μm.
In contrast, ECs-4 showed the highest expression of Glut1, Serpine1, Sele, Fgf23, Car4, Rnd1, and Tnfrsf9 (Figure 2D and Figure S2C). Among these highly expressed genes of subclusters 4, the expression of Serpine1, Sele, and Fgf23 are associated with endothelial dysfunction.
Similarly, the increase of Rnd1 expression could induce actin cytoskeletal rearrangement of endothelium, thereby aggravating the inflammatory response.
This evidence further indicates that the dysfunction of ECs-4 leads to the acceleration of sepsis-induced acute liver dysfunction.
Additionally, ECs-3 showed an upregulation of inflammation-associated genes (Il18, Nlrc4, Txn1, Nod1) (Figure 2F and G). As for cell adhesion molecules, ECs-3 exhibited relatively high expression of Vcam-1 and a low expression of Tjp1 and Cldn5 (Figure 2E), indicating poor cellular adhesion of endothelial cells in ECs-3. The ECs-4 subcluster also showed an upregulation of genes related to inflammation (Nlrp3, Tnfaip3, Ube2n, Cxcl2, Ccl2, Ccl5, Csf3, Nfkb1, Nod2) (Figure 2F and G), cell adhesion (Cdh5, Selplg, Icam1, Selp) (Figure 2E), and apoptosis (Apaf1, Bax, Trp53, Bcl2l11) (Figure 2E). Interestingly, ECs-4 exhibited a low level of Hmgb1 (Figure 2F). Hmgb1 is considered an important mediator in the pathogenesis of sepsis-induced acute liver dysfunction.
KEGG pathway analyses revealed that the genes that were upregulated in ECs-3 belonged to the Toll-like receptor, antigen processing and presentation, and NOD-like receptor signaling pathways (Figure 2H); whereas the genes that were upregulated in ECs-4 belong to the TNF/NF-κB, IL-17, and NOD-like receptor signaling pathways (Figure 2H, Figure S2A and B).
Furthermore, SCENIC analysis was used to assess the expression status of transcription factors (TFs) in different endothelial cell subpopulations. As shown in Figure 2J, the genes regulated by the Yy1, Fosl1, Hif1α, and Atf4 transcription factors were upregulated in subclusters 4. Notably, Atf4 is involved in the complex process of cellular stress response,
Small-Molecule Integrated Stress Response Inhibitor Reduces Susceptibility to Postinfarct Atrial Fibrillation in Rats via the Inhibition of Integrated Stress Responses.
Journal of Pharmacology Experimental Therapeutics.2021; 378: 197
and Yy1 expression was correlated with the inflammatory NK-κB activity and neutrophil infiltration, and subsequently resulted in enhanced inflammatory effects.
Insulin-like growth factor-1 enhances neuroprotective effects of neural stem cell exosomes after spinal cord injury via an miR-219a-2-3p/YY1 mechanism.
Fosl1 is part of the activator protein complex, which makes up the transcription factor activator protein 1 (AP-1). AP-1 has been suggested to be responsible for activating inflammation.
To confirm the emergence of ECs-4 during the pathogenesis of sepsis-induced acute liver dysfunction, the immunofluorescence staining was performed using CD31, Sele, and Glut1 antibodies to detect this endothelial cell subpopulation (Figure 2K). Consistent with single-cell analysis, the proportion of CD31+Sele+Glut1+ endothelial cells increased gradually, concomitant with the development of acute liver dysfunction. In addition, total endothelial cells were detected using CD31 immunofluorescence staining. Our results indicate that the number of total endothelial cells decrease with pathological progression, being consistent with the single-cell RNA-seq result (Figure S2D).
Altogether, the analysis of endothelial cells indicated that hepatic endothelial cells exhibited heterogeneity, and differences among subtypes with distinct gene expression patterns represented their unique functionality during sepsis. Here, the unique subpopulation of endothelial cells with endothelial dysfunction-related genetic characteristics was also identified and is speculated to be closely related to the pathology of sepsis-induced acute liver dysfunction.
3.3 Single-cell RNA-seq revealed heterogeneity of neutrophils in liver during sepsis
Neutrophils, which are well characterized to be recruited rapidly at sites of infection, function as first-line responders specialized in elimination of invading pathogens. However, simultaneously exaggerated activation and uncontrolled tissue infiltration of neutrophils cause excess oxidative stress, inflammation, and subsequent tissue injury,
Neutrophil membrane-mimicking nanodecoys with intrinsic anti-inflammatory properties alleviate sepsis-induced acute liver injury and lethality in a mouse endotoxemia model.
However, neutrophil populations in sepsis are not homogenous; therefore, exploring neutrophil heterogeneity will help regulate neutrophils accurately to obtain a balance between protective immunity and tissue injury.
The hepatic neutrophils were further grouped into three subclusters: neutrophils-1 (mainly expressing Gadd45b, Icam1, Gbp5, Tifa), neutrophils-2 (mainly expressing Myadm, Cd300ld, Ly6g), and neutrophils-3 (mainly expressing Lta4h, Sort1, Rgs18) (Figure 3A, B, D, and Figure S3A). The neutrophils-2 was the major subcluster of neutrophils and accounted for up to 71.54% of liver neutrophils in healthy mice (Figure 3C). The neutrophils-1 was the major subcluster (60.61%) of neutrophils at the early stage of CLP (6h), whereas the neutrophils in neutrophils-3 constituted the major subcluster (48.16%) of neutrophils at 24 h after CLP. Impressively, the proportion of neutrophils-3 in liver exhibited a continuous increase in a time-dependent manner during sepsis, which were verified via immunofluorescence staining of Ly6G, Lta4h, and Sort1 in healthy and sepsis mice (Figure 3G). In addition, our results indicate that the changes in the number of total neutrophils in different groups were consistent with the single-cell RNA-seq result (Figure S3B).
Figure 3Hepatic neutrophil dynamics and functionality in healthy and septic livers. (A) Sub-clustering of neutrophils in healthy and septic livers. (B) UMAP plot of neutrophils across the indicated conditions. (C) The proportion of neutrophil subclusters in each sample. (D) Heatmap plots demonstrating expression of specified genes among neutrophil subclusters. (E-F) Heatmap showing gene expression by the indicated neutrophil subclusters. (G) Immunofluorescence staining results showing the spatial distribution of the Neu-3 subpopulations in healthy and septic mouse livers. Scale bars, 20 μm. (H) The enriched pathway of associated with neutrophil subclusters 1 and 3. Colors indicate significance of enrichment and circle sizes indicate number of genes falling into respective categories. (I) Subpopulation-specific regulons of each subpopulation revealed by SCENIC analysis.
Further analysis revealed that neutrophils subclusters exhibited different transcriptional characteristics (Figure 3E and F). Both neutrophils-1 and neutrophils-3 were inflammatory neutrophils that bore a proinflammatory-gene signature. Neutrophils-1 showed a high expression of proinflammatory cytokines (Il1b, Tnf, Mif), neutrophil-attracting chemokines (Cxcl1, Cxcl2, Cxcl3), and adhesion molecule (Icam1) (Figure 3E). Notably, neutrophils-1 also exhibited a high expression of immune regulatory genes (Cd274 and Socs3) (Figure 3F). Neutrophils-3 showed a hyperactivated phenotype with a high expression of calgranulins (S100a9 and S100a8), neutrophil activation markers (Mpo, Ltf, Cd69), neutrophil infiltration-related genes (Cd47, Spn/Cd43, Mmp8, Mmp25), and proinflammatory cytokines (Il-6) (Figure 3E and F). Additionally, neutrophils-3 showed apoptotic properties via the upregulation of the pro-apoptosis-related genes (Fadd, Casp6, Bad, Bclaf1) (Figure 3F). Collectively, it is suspected that the neutrophils-1 responded to the microbial invasion in liver, and became activated to eliminate microbial organisms, while concomitantly releasing inflammatory cytokines and chemokines to recruit more immune cells, including neutrophils-3, to the liver. Consequently, a significant number of activated neutrophils-3 with robust pathogen removal capability could have infiltrated the liver and eliminated the invading microbial organisms via phagocytosis and the release of cytotoxic antimicrobial molecules (reactive oxygen species, lactoferrin), ultimately leading to liver injury.
KEGG pathway analyses revealed that the genes that were upregulated in neutrophils-1 belonged to the NOD-like receptor, TNF/NF-κB, C-type lectin receptor, and Toll-like receptor signaling pathways (Figure 3H). Meanwhile, the genes that were upregulated in neutrophils-3 belonged to the leukocyte transendothelial migration, endocytosis, and phagosome (Figure 3H).
SCENIC analysis was performed to assess which TFs were responsible for the differences in gene expression between different cell clusters. Hif1a, Atf4, Atf6, Irf1, Nfkb1, and Irf5 were identified as candidate transcription factors underlying the gene signature of neutrophils-1 (Figure 3I), whereas Xbp1, Fosl1, and Ltf were the candidate transcription factors underlying the differential gene expression in neutrophils-3.
3.4 Single-cell RNA-seq revealed heterogeneity of Kupffer cells in liver during sepsis
Kupffer cells represent the major fraction of liver macrophages,
In this study, the Kupffer cells were grouped into five clusters (Figure 4A, B, D, and F), which were annotated as KCs-1 (enriched in the expression of Kcna2, Abcg3, Slc1a2), KCs-2 (enriched in the expression of Slamf1, Edn1, Adora2b, Adam8), KCs-3 (enriched in the expression of Enpp2, Cxcl9, Tmc3, Cnn2), KCs-4 (enriched in the expression of Il1r2, Vdr, Tgfb3, Cdh1), and KCs-5 (enriched in the expression of Cdca3, Stil, Rad51). KCs-1 constituted up to 89.48% of Kupffer cells under healthy conditions, whereas KCs-3 only constituted 7.29% of the total Kupffer cells (Figure 4C). During the early stage of sepsis (6 h), the proportions of KCs-2 and KCs-3 were increased and accounted for 92.21% of all Kupffer cells. As liver injury progressed in severity, the proportion of KCs-2 increased slightly (39.20% to 40.93%), and the proportion of KCs-3 decreased considerably (53.01% to 23.61%) at 24 h after CLP. Importantly, the KCs-4 increased from less than 2% of total Kupffer cells in sham and CLP-6h groups, to 32.57% in the CLP-24h group. Based on the dynamic changes in the proportion of Kupffer cell subclusters, it is suspected that KCs-2, KCs-3, and KCs-4 play important roles in the development of sepsis-induced acute liver dysfunction. Additionally, total Kupffer cells were detected using F4/80 immunofluorescence staining. Our results indicate that the changes in the number of F4/80 positive cells in different groups were consistent with the single-cell RNA-seq result (Figure S4).
Figure 4Hepatic Kupffer cell dynamics and functionality in healthy and septic livers. (A) Sub-clustering of Kupffer cells in healthy and septic livers. (B) UMAP plot of Kupffer cells across the indicated conditions. (C) The proportion of Kupffer cell subclusters in each sample. (D) Heatmap plots demonstrating expression of specified genes among Kupffer cell subclusters. (E) Heatmap showing gene expression by the indicated Kupffer cell subclusters. (F) UAMP plot showing subcluster-specific marker genes of Kupffer cell.
Further analysis revealed that KCs-1, KCs-2, KCs-3, and KCs-4 exhibited different expression profiles of inflammation-related genes (Figure 4E). KCs-2 showed a high expression of pro-inflammatory genes / M1 markers (eg, Il-6, Ccl2, Ccl3, Ccl4, Nos2, Cxcl1, Cxcl2, Cxcl3, Tlr2, Ptgs2, Fpr2, Cd80, Cd38) but also a high expression of several feedback inhibitors of activation (Il10, Il13, Arg1, Tgm2). KCs-3 also exhibited a high expression of pro-inflammatory genes, specifically Cxcl9, Fpr2, and Gpr18. Similarly, KCs-4 showed a relatively high expression of pro-inflammatory genes (Il-6, Ccl2, Ccl3, Ccl5, Il-1b, Cd38, Cxcl2, Fpr2), as well as genes associated with M2-like Kupffer cells (Il10, Mrc1, Socs2,
). In contrast, KCs-1 exhibited a very low expression of pro-inflammatory genes and a high expression level in genes associated with M2-like Kupffer cells (Myc and Egr2). These results indicated that KCs-2, KCs-3, and KCs-4 were activated and played a vital role in releasing cytokines and chemokines, which most likely activated and recruited circulating macrophages and neutrophils to liver tissue, subsequently leading to liver injury.
3.5 Cell-to-cell communication in CLP-induced liver dysfunction: receptor-ligand analysis
Elucidating the explicit interaction among liver cells during sepsis will shed light on the pathogenesis of sepsis-induced acute liver dysfunction. The ligand-receptor pairs among the major cell types are shown in Figure S5. Notably, the Kupffer cells showed the most interactions with other cell types during sepsis, in particular with endothelial cells and neutrophils, at 24h after sepsis. To further explore the detailed cellular communication during CLP-induced liver injury, the intercellular interactions within heterogeneous populations of neutrophils, Kupffer and endothelial cells were analyzed. As shown in Figure 5A, ECs-4, the major subcluster of endothelial cells at 24 h after CLP, showed enhanced interactions with Kupffer cells via Csf3r/Csf3, Csf1r/Csf3, CD44/Sele, and Ccr1/Ccl2 axes (Figure 5A). Furthermore, endothelial–neutrophil interactions are crucial for neutrophil infiltration to liver.
Neutrophil membrane-mimicking nanodecoys with intrinsic anti-inflammatory properties alleviate sepsis-induced acute liver injury and lethality in a mouse endotoxemia model.
Notably, ECs-4, the major subcluster of endothelial cells at 24 h after CLP, expressed relatively high levels of chemokines (Ccl2, Ccl5), cytokines (Csf3/G-CSF), and adhesion molecules (Sele and Selp), whereas the corresponding receptors were widely expressed in neutrophils-3, suggesting that functional interactions between neutrophils-3 and ECs-4 may play significant roles in enhancing neutrophil infiltration into liver tissues (Figure 5B). Additionally, the enhanced interactions between Kupffer cells (KCs-2, KCs-3, and KCs-4) and neutrophils, such as Il6 receptor/Il6 and Cd80/Cd274, were the most noticeable (Figure 5C).
Figure 5Molecular interactions of neutrophils, endothelial and Kupffer cells. (A) Interactions between four endothelial cell subclusters and four Kupffer cell subclusters. (B) Interactions between three neutrophil subclusters and four endothelial cell subclusters. (C) Interactions between three neutrophil subclusters and four Kupffer cell subclusters. Dot sizes and colors represent logarithmic-transformed P values and mean expression of interacting molecules in corresponding cells. (D) Predicted main regulatory network among endothelial cells, neutrophils, and Kupffer cells during sepsis.
Our results found that the activated endothelial cells secrete colony stimulating factors (Csf3) and chemokines (Ccl2) to interact with immune cells including Kupffer cells and neutrophils, and facilitate the immune cell adhesion via elevating the adhesion molecule expression (Sele and Selp) during the sepsis (Figure 5D), which can provide precise drug targets for early treatment of sepsis.
3.6 ATF4 inhibition alleviated CLP-induced acute liver dysfunction and prolonged the survival of septic mice
In the present study, the TFs underlying the differential gene expression in endothelial and Kupffer cells, neutrophils, and monocytes/monocyte-derived macrophages during the sepsis-induced acute liver dysfunction were identified (Figure S6A–D), and the representative results were verified via immunofluorescence. As shown in Figure S5E and F, a rapid increase of ATF4 and NF-κB1 expression in endothelial cells were observed 6h after CLP when compared with that of healthy controls. Subsequently, the ATF4 and NF-κB1 expression declined at 24 h after CLP compared to that at 6 h after CLP. In addition, the Fosl1 expression in Kupffer cells increased with pathological progression (Figure S6G).
Among these TFs, ATF4 was identified as the hallmark TF in endothelial cells, Kupffer cells, neutrophils, and monocytes/monocyte-derived macrophages at 6 h after sepsis (Figure S6A–D). These results suggested that ATF4 activated the endothelial and immune cells associated with the hyperinflammatory response, which contributed to the pathogenesis of CLP-induced acute liver dysfunction in early sepsis stage, thus we hypothesized that the inhibition of ATF4 would attenuate the CLP-induced acute liver dysfunction.
ISRIB, a specific small-molecule inhibitor, can suppress the upregulation of ATF4 and decrease nuclear translocation of ATF4.
As shown in Figure 6A and B, the plasma ALT and AST activities were used as biomarkers of liver dysfunction. CLP significantly elevated the plasma AST and ALT activities compared to that in the normal controls, which indicated liver dysfunction in the CLP group. ISRIB administration significantly reduced the other liver dysfunction biomarkers, including direct bilirubin, glutamyl-transpeptidase, globulin, total bile acid, and albumin/globulin ratio, in plasma compared to those of the CLP group (Figure 6D–G, Figure S6H). Notably, ISRIB administration alone showed no significant differences in the liver dysfunction biomarkers compared to that in the normal controls (P > 0.05), which suggested that ISRIB exerted no acute toxicity at the administered dosage.
Figure 6Therapeutic effects of ATF4 inhibition on sepsis-induced acute liver dysfunction and survival in CLP-induced sepsis model. (A) Experimental timeline of ATF4 inhibition in CLP-induced sepsis model (in vivo). The levels of alanine aminotransferase (ALT) (B), aspartate aminotransferase (AST) (C), direct bilirubin (DBIL) (D), glutamyl-transpeptidase (GGT), globulin (GLO), total bile acid (TBA) (G) in plasma 24 h after CLP. The platelet counts in blood (H), neutrophil-to-lymphocyte ratio (NLR) (I), and monocyte-to-lymphocyte ratio (MLR) (J). The malondialdehyde (MDA) content (K), myeloperoxidase (MPO) activity (L), and interleukin (IL)-6 (M) in liver 24 h after CLP. Representative images show the effect of ISRIB administration on histological injury (N) (Scale bar, 50 μm). (O) Survival rates of sepsis mice after 96 h following ISRIB administration. ISR: ISRIB, *p < 0.05; **p < 0.01; ***p < 0.001. Statistical differences between groups were assessed using one-way ANOVA for Fig. 6 B–G. Statistical differences between groups were assessed using t-test for Fig. 6 K–M. Statistical differences between groups were analyzed using the log-rank test for Fig. 6 O.
Neutrophil-lymphocyte ratio, monocyte-lymphocyte ratio and platelet-lymphocyte ratio in non-affective psychosis: A meta-analysis and systematic review.
The World Journal of Biol Psychiatry.2020; 21: 326-338
Neutrophil-lymphocyte ratio, monocyte-lymphocyte ratio and platelet-lymphocyte ratio in non-affective psychosis: A meta-analysis and systematic review.
The World Journal of Biol Psychiatry.2020; 21: 326-338
Blood analysis demonstrated that CLP significantly reduced the platelet count and increased the neutrophil-to-lymphocyte ratio and monocyte-to-lymphocyte ratio, which were markedly reversed by ISRIB treatment (Figure 6H–J). These results indicated that ISRIB treatment could decrease the systemic inflammatory response during the early stage of sepsis. Additionally, the ISRIB treatment significantly elevated the number of lymphocytes and lymphocyte percentage compared with that of the CLP group (Figure S6J and I), which suggests that ISRIB treatment promote the immune cell survival and maintain the immune function during sepsis.
Hepatic lipid peroxidation and neutrophil infiltration are detected by quantifying the MDA content and MPO activity.
Neutrophil membrane-mimicking nanodecoys with intrinsic anti-inflammatory properties alleviate sepsis-induced acute liver injury and lethality in a mouse endotoxemia model.
As shown in Figure 6K, a significantly higher hepatic MDA content was observed in septic model mice compared with normal control, and the mice administered with ISRIB exhibited a significantly decreased MDA content compared to that in the sepsis model mice. ISRIB administration significantly inhibited neutrophil infiltration in treated mice compared with untreated sepsis model mice (Figure 6L). In comparison to the normal control, significantly higher hepatic IL-6 level was indicated in the sepsis model mice treated with vehicle. ISRIB treatment significantly diminished the hepatic IL-6 level compared to that in the sepsis control (Figure 6M).
As shown in Figure 6N and Figure S6K, we observed a marked liver injury including cell swelling, inflammatory cell infiltration, and tissue architecture disruption in sepsis model mice, whereas ISRIB administration exhibited a significant improvement in the pathological injury. To further explore the therapeutic potential of ISRIB, the effect of ISRIB administration on the survival of CLP-induced septic mice was examined. The CLP induced the death in mice, and a single-dose administration of ISRIB significantly improved the survival of CLP mice from 52% to 75% (P < 0.05, Figure 6O).
4. Discussion
Liver tissue is composed of multiple nonparenchymal cell lineages including neutrophils, endothelial cells, and Kupffer cells, all of which are essential contributors to uncontrolled local inflammation during sepsis exposure, causing intractable liver injury. The characteristics of gene heterogeneity differentiates these major cell types and defines multiple subpopulations with distinct functions. Therefore, resolving the dynamic changes of the transcriptome at the single-cell level during disease progression is critical to understanding the intrinsic mechanisms of sepsis-induced acute liver dysfunction, which is of great significance for improving disease diagnosis and intervention. In the current study, single-cell RNA sequencing was used to characterize the dynamic cellular and molecular signatures along disease course, and the key findings were validated in a CLP-induced sepsis model.
This study provides new insights into the role of endothelial cell states in the pathogenesis of sepsis-induced acute liver dysfunction. The number of hepatic endothelial cells was observed to be decreasing in a time-dependent manner during the progression of sepsis, which is consistent with a recent study in a endotoxemia model.
We speculated that the weak cellular adhesion and high apoptosis revealed by scRNA-seq might be the main cause of the sharp reduction in the number of endothelial cells, which then triggers interstitial edema and consecutively worsens liver hypoperfusion, accelerating liver injury.
Circulating Endothelial Cells From Septic Shock Patients Convert to Fibroblasts Are Associated With the Resuscitation Fluid Dose and Are Biomarkers for Survival Prediction.
We found that the hepatic endothelial cells could get reprogrammed into cells displaying proinflammatory phenotype during sepsis, as ECs-3 and ECs-4 were observed to be in different proinflammatory states reflected by their inflammation-related gene expression profiles. Particularly, this result indicated that the ECs-3 may play an important role in the adaptive immune response to infection as antigen-presenting cells, as reported in other studies as well.
NOD1 is an important and well-characterized member of the NLR family, which was demonstrated to enhance antigen-presenting ability of liver sinusoidal endothelial cells during viral infection.
We also found Nod1 to be highly expressed in ECs-3. Thus, NOD1 is likely to be involved in the antigen-presenting activity of ECs-3 during sepsis. Additionally, ECs-4 was elicited to be in an endothelial activation/dysfunction state due to a high expression of endothelial activation/dysfunction markers, including Vcam-1, Icam-1, Selp, Sele, Serpine1, and Fgf23 18, 21, 22, 56. The activated ECs-4 could interact with neutrophils and exacerbate sepsis-induced acute liver dysfunction by cell-to-cell communication,
Neutrophil membrane-mimicking nanodecoys with intrinsic anti-inflammatory properties alleviate sepsis-induced acute liver injury and lethality in a mouse endotoxemia model.
we found that the Csf3r/Csf3 axis was only presented in the interactions between neutrophils and activated ECs-4, suggesting that blocking Csf3r/Csf3 axis may be a potential therapeutic avenue in treating sepsis-induced acute liver dysfunction, but this still requires further research.
Kupffer cells are a critical component of the mononuclear phagocytic system and are central to both the hepatic and systemic response to pathogens [50]. Previous studies have indicated that the polarization of Kupffer cells is recognized as a critical mediator of liver injury.
Traditionally, Kupffer cells are polarized into distinct phenotypes, M1 (inflammatory) and M2 (anti-inflammatory), depending on the local microenvironment.
In this study, beyond the quiescent KCs-1, the inflammatory Kupffer cells including KCs-2, KCs-3, and KCs-4 were also identified, and the expression of M1 and M2 Kupffer cells markers were subsequently determined in all five Kupffer cell subpopulations. We found that all the subpopulations showed dual expression of pro-inflammatory genes and anti-inflammatory genes. Specifically, although IL-10, Arg1, and Chil3 were reported as classical M2 Kupffer cell markers,
the inflammatory Kupffer cell subpopulations, namely KCs-2, KCs-3, and KCs-4 with their high expression of pro-inflammatory genes, also exhibited a high level of expression of IL-10, Arg1, and Chil3. Similar phenomena have also been reported in other studies.
Altogether, these results indicated that Kupffer cells are highly plastic and could undergo a broad spectrum of transcriptomic activation states under complex hepatic microenvironment, and cannot be simply classified into either M1 or M2 polarization. A more comprehensive classification system based on recent research is required to describe the activation states of Kupffer cells in vivo.
Since the solution injected into the abdominal cavity could flow out though the wound, we do not have the data to evaluate the therapeutic effect if ISRIB was to be given after CLP. Our results indicate that the ATF-4 activation occurs at 6h after CLP, and returns to normal at 24h after CLP. We speculate that ISRIB exhibits therapeutic effect if ISRIB is given early (<6h), and that the therapeutic effect might decrease with a prolonged administration time.
In summary, our study presents the dynamic transcriptomic landscape of major nonparenchymal cells at single-cell resolution, in which we identified the significant alterations and heterogeneity of hepatic nonparenchymal cell subsets during sepsis. Importantly, we identified the endothelial cell and neutrophil subsets that are associated with acute liver dysfunction during sepsis progression and explored the therapeutic effect of ATF4 inhibition. Overall, these results uncovered potential mechanisms and promising therapeutic targets for the prevention and treatment of sepsis-induced acute liver dysfunction and other liver-related diseases.
Conflict of interest
The authors declare no conflict of interest.
Lay summary
Sepsis-induced acute liver dysfunction often occurs early in sepsis and can lead to the death of the patient. Nevertheless, the pathogenesis of sepsis-induced acute liver dysfunction is not yet clear. We identified the major cell types associated with acute liver dysfunction and explored their interactions during sepsis. In addition, we also found that ATF-4 inhibition could be invoked as a potential therapeutic for sepsis-induced acute liver dysfunction.
Data availability
The data that support the findings of this study are available from the corresponding authors upon reasonable request.
Author contributions
The contributions of each author made to the study are specified as follows: Chen Gan, Yao Yongming, and Zhou Hong designed the study; Chen Gan, Ren Chao, and Xiao Yao performed the experiments and collected the data; Chen Gan, Xiao Yao, Ren Chao, Wang Yujing, Yao Renqi, Wang Quan, You Guoxing, Lu Mingzi, Zhang Xiaoyong, Yan Shaoduo, and Zhang Jun analyzed the data; Lu Mingzi, Chen Gan, Ren Chao, and Xiao Yao prepared the manuscript.
This work was supported by National Natural Science Foundation of China (82170229) and Foundation Strengthening Program Technology Fund Project (2019-JCJQ-JJ-164). The authors would like to thank Xiaohua Yao in OE Biotech Co., Ltd (Shanghai, China) for her assistance with the data analysis.
Appendix A. Supplementary data
The following is/are the supplementary data to this article:
Neutrophil membrane-mimicking nanodecoys with intrinsic anti-inflammatory properties alleviate sepsis-induced acute liver injury and lethality in a mouse endotoxemia model.
Small-Molecule Integrated Stress Response Inhibitor Reduces Susceptibility to Postinfarct Atrial Fibrillation in Rats via the Inhibition of Integrated Stress Responses.
Journal of Pharmacology Experimental Therapeutics.2021; 378: 197
Insulin-like growth factor-1 enhances neuroprotective effects of neural stem cell exosomes after spinal cord injury via an miR-219a-2-3p/YY1 mechanism.
Neutrophil-lymphocyte ratio, monocyte-lymphocyte ratio and platelet-lymphocyte ratio in non-affective psychosis: A meta-analysis and systematic review.
The World Journal of Biol Psychiatry.2020; 21: 326-338
Circulating Endothelial Cells From Septic Shock Patients Convert to Fibroblasts Are Associated With the Resuscitation Fluid Dose and Are Biomarkers for Survival Prediction.
00049-6&id=ga1.jpg){kind=link}
00049-6&id=gr1.jpg){kind=link}
00049-6&id=gr2.jpg){kind=link}
00049-6&id=gr3.jpg){kind=link}
00049-6&id=gr4.jpg){kind=link}
00049-6&id=gr5.jpg){kind=link}
00049-6&id=gr6.jpg){kind=link}