Abstract

   Endothelial barrier dysfunction and the resulting vascular injury are
   responsible for multiorgan failure in sepsis. Myeloid C-type lectin
   domain family 5 member A (CLEC5A) is a pattern recognition receptor
   involved in host defense against infection. Mice lacking CLEC5A were
   resistant to cecal ligation and puncture (CLP)–induced polymicrobial
   sepsis and lipopolysaccharide (LPS)–induced endotoxemia, as observed by
   decreased mortality. Single-cell RNA sequencing revealed transcriptomic
   heterogeneity of vascular endothelial cells in CLEC5A-deficient lungs
   following CLP. Endothelial-specific knockdown of CLEC5A improved
   survival of CLP-challenged mice, which was completely ineffective with
   reexpression of endothelial CLEC5A. The survival benefits were
   attributed to alleviated inflammatory storm and vascular leakage.
   Furthermore, endothelial CLEC5A deficiency protected mice against
   Escherichia coli–induced pneumonia. In vitro, CLEC5A deletion
   maintained trans-endothelial electrical resistance, and inhibited
   adhesion and trans-endothelial migration of monocytes/neutrophils under
   LPS stimulation. The study unveils the importance of CLEC5A in
   regulating endothelial barrier function and suggests endothelial CLEC5A
   as a therapeutic target for pneumonia or sepsis-causing bacterial
   infection.
     __________________________________________________________________

   Endothelial CLEC5A deficiency protects against inflammatory lung injury
   and provides survival benefits in sepsis.

INTRODUCTION

   Sepsis is a life-threatening condition with multiorgan failure caused
   by a dysregulated host response to infection ([48]1, [49]2). For many
   of the patients who survive, it is associated with a long-term
   morbidity. Although optimal management is fundamental in the intensive
   care setting, early recognition and rapid intervention during sepsis
   are required to improve outcomes ([50]3). Ongoing research continues to
   increase understanding of molecular and biochemical profiles of patient
   populations and improves disease definition, diagnosis, and
   pharmacological strategies to achieve more favorable outcomes in sepsis
   ([51]4).

   Pulmonary infection is a common cause of sepsis ([52]4). Patients with
   sepsis are at a high risk of developing acute respiratory distress
   syndrome (ARDS), which occurs in up to 50% of patients ([53]5). Lung
   injury is one of the major attributable cause of death in sepsis. It is
   characterized by alveolar-capillary barrier dysfunction and resulting
   vascular leakage and subsequent influx of albumin-rich edema fluid into
   the alveolar cavity, causing hypoxemia ([54]6). Over several decades,
   the incidence of sepsis has risen and, in sepsis-associated organ
   dysfunction, especially ARDS, morbidity and mortality remain high. The
   necessity for advances in our understanding of the pathophysiological
   mechanisms and development of effective therapeutic strategies cannot
   be overstated.

   Sepsis and pneumonia share several similar clinical presentations and
   disease characteristics, in which infections and inflammation are the
   key. In response to infection, immunity maintains homeostasis between
   immunopathology and immunosuppression. A failure of immune homeostasis
   is considered the key pathophysiology of sepsis ([55]2). Activation of
   the local innate immune response is important for host defense against
   pathogen spreading. This activation may become progressive during
   sepsis, causing tissue injury and leading to multiorgan failure and
   ultimately death ([56]7). Although the major cell sources for
   inflammatory cytokines in sepsis are myeloid cells, nonmyeloid cells
   such as endothelial cells are also involved and have been highly valued
   ([57]8). Endothelial cells are an unconventional type of immune cells
   that lines blood vessels and participates in immune and inflammatory
   responses upon activation ([58]9). In sepsis, endothelial cells exhibit
   phenotypic changes toward pro-inflammation, pro-adhesion,
   pro-coagulation, and pro-apoptosis, and these changes contribute to
   leukocyte recruitment and pathogen elimination. However, severe and/or
   persistent activation causes damage to the endothelium, resulting in
   barrier dysfunction and exacerbating local inflammation and tissue
   damage ([59]10). The endothelium is considered a target in the
   treatment of sepsis and limiting barrier dysfunction is a promising
   avenue for preventing the development of lung injury.

   Myeloid C-type lectin domain family 5 member A (CLEC5A) is a pattern
   recognition receptor (PRR) and plays a critical role in innate immunity
   against viral infection ([60]11, [61]12). Apart from being a
   promiscuous PRR to viruses, CLEC5A has recently been noted in host
   defense against bacterial infection ([62]13, [63]14). Sung et al.
   reported that deletion of CLEC5A or administration of anti-CLEC5A
   monoclonal antibodies alleviated lung injury and improved survival of
   Pseudomonas aeruginosa–challenged mice ([64]14). These observations
   imply the therapeutic potential of targeting CLEC5A in infectious
   disorders and the resulting lung injury; however, the underlying
   cellular and molecular mechanism has not been fully unveiled. Current
   studies mostly focus on CLEC5A in myeloid cells, and little is known
   about its function in other cell types. A comprehensive understanding
   of CLEC5A is needed to further its implications as a therapeutic target
   in the clinical setting.

   In the present study, we uncover the role of CLEC5A in endothelial
   cells regarding phenotypic modifications and functional alterations
   during inflammation and bacterial infection, as well as in cellular
   interaction with immune cells. The immune response in the local
   environment of sepsis was investigated by modeling the interaction of
   endothelial cells and immune cells under lipopolysaccharide (LPS)
   stimulation in vitro. Deficiency of endothelial CLEC5A prevented
   against barrier failure and lung injury and improved survival in murine
   sepsis models. This study broadens the profile of CLEC5A to the
   endothelium and improves the existing therapeutic algorithms of CLEC5A
   in lung injury. We suggest that CLEC5A could be a target in
   translational endothelial-driven therapeutic approaches to bacterial
   pneumonia and sepsis.

RESULTS

Global deletion of CLEC5A improves the survival of mice with endotoxemia or
polymicrobial sepsis

   To explore the potential targets of gene therapy for sepsis and
   sepsis-induced lung injury, we obtained and analyzed the gene
   expression profiles of four datasets ([65]GSE28750, [66]GSE206635,
   [67]GSE1871, and [68]GSE34901) from the NCBI Gene Expression Omnibus
   (GEO) database ([69]https://ncbi.nlm.nih.gov/geo/) ([70]Fig. 1A). The
   datasets [71]GSE28750 collect data from blood samples from patients
   with sepsis and healthy volunteers, and data of [72]GSE206635 are from
   the lungs of patients with COVID-19. [73]GSE1871 and [74]GSE34901
   include data from lungs from LPS-induced murine lung injury and
   Escherichia coli–induced murine pneumonia, respectively. The
   differentially expressed genes (DEGs) were derived from the four
   datasets with the linear models for microarray data (limma) following
   the screening criteria of fold change (FC) > 1.5 and P < 0.05. A total
   of 154 robust genes were screened out according to the robust rank
   aggregation with a score of < 0.05 and frequency ≥ 3 (number of
   datasets). Among genes in C-type lectin domain family, CLEC5A is top
   ranked and significantly up-regulated in all four datasets
   (Score < 0.0001 and frequency = 4) ([75]Fig. 1B). CLEC5A is widely
   expressed in human tissues and highly expressed in the lungs based on
   the Genotype-Tissues Expression (GTEx) dataset and the FANTOM5 dataset
   (fig. S1), which were obtained from an open source database, the Human
   Protein Atlas (HPA; [76]http://www.proteinatlas.org).

Fig. 1. Genetic knockout of CLEC5A improves the survival of mice with
endotoxemia or polymicrobial sepsis.

   [77]Fig. 1.
   [78]Open in a new tab

   (A) Analysis of the gene expression profiles of datasets ([79]GSE28750,
   [80]GSE206635, [81]GSE1871, and [82]GSE34901) downloaded from the NCBI
   GEO database and identification of CLEC5A as a potential gene related
   to sepsis and sepsis-induced acute lung injury. Top DEGs among C-type
   lectin domain family were displayed based on the limma with P < 0.05
   and FC > 1.5 and sorted by the robust rank aggregation. Genes on white
   panel meet score (P value) < 0.05 and frequency (Freq) ≥ 3. Blue font
   highlights the top-ranked gene. NA, not applicable. (B) mRNA expression
   of CLEC5A in blood samples from patients with sepsis ([83]GSE28750),
   lung tissues from patients with COVID-19 ([84]GSE206635), and lungs
   from mice with LPS-induced lung injury ([85]GSE1871) and E.
   coli–induced pneumonia ([86]GSE34901). (C) Two mouse models of sepsis
   were established in this study, including bacterial LPS-induced
   endotoxemia and CLP-induced polymicrobial sepsis. PMVECs were isolated
   from corresponding mice. (D) Representative images of
   immunofluorescence staining for CLEC5A with DAPI counterstaining of
   nuclei in lung tissues, 4 hours after LPS or 12 hours after CLP
   (biological replicates, n = 6 per group). Scale bars, 100 μm. (E)
   Western blotting analysis and representative blots showing the
   expression of CLEC5A in PMVECs from mice challenged by LPS or CLP
   (biological replicates, n = 6 per group). h, hours. (F) mRNA expression
   of CLEC5A in mouse PMVECs (biological replicates, n = 6 per group). (G)
   Schematic diagram of CLEC5A genetic knockout (KO) mice (CLEC5A^−/−)
   generation. CLEC5A^−/− mice were generated on a C57BL/6 background by
   the CRISPR-Cas9 system, and WT littermates were used as the control.
   (H) Survival rate of mice after LPS or CLP challenge (biological
   replicates, n = 10 per group). The statistical significance was
   determined by repeated measures two-way ANOVA analysis using the
   Geisser-Greenhouse correction. The statistical significance between
   survival curves was determined by P value using the log-rank
   (Mantel-Cox) test. n.s., not significant.

   By analyzing these public datasets, we hypothesize that high expression
   of CLEC5A might play a role in the pathophysiology of these lung
   diseases caused by bacterial or viral infection, in which endothelial
   barrier dysfunction is a key. The present study aims to investigate the
   role of CLEC5A in sepsis-associated lung injury and endothelial barrier
   dysfunction using two models of sepsis, intraperitoneal injection of
   LPS and cecal ligation and puncture (CLP) surgery ([87]Fig. 1C). The
   expression of CLEC5A in the lungs was up-regulated upon LPS or CLP
   challenge (fig. S2, A and B). During the induction of experimental
   sepsis, CLEC5A exhibited differential expression at different time
   points (4, 12, 24, and 72 hours) post-LPS or CLP but maintained an
   up-regulated trend in general. Immunofluorescence staining showed an
   increase in CLEC5A expression in lung tissues 4 hours after LPS or 12
   hours after CLP challenge ([88]Fig. 1D). Of note, a significant
   increase in the expression of CLEC5A was found in primary pulmonary
   microvascular endothelial cells (PMVECs) isolated from septic mice
   following LPS or CLP induction for 4 or 12 hours ([89]Fig. 1, C, E, and
   F).

   CLEC5A genetic knockout (CLEC5A^−/−) mice were used in this study
   ([90]Fig. 1G), and the efficacy of CLEC5A deletion was validated in
   different tissues, including the lung, brain, heart, kidney and liver,
   from both male and female CLEC5A^−/− mice (fig. S3). Global deletion of
   CLEC5A significantly increased survival of mice in the setting of
   LPS-induced endotoxemia and CLP-induced polymicrobial sepsis ([91]Fig.
   1H). All wild-type (WT) littermates died within 108 hours after LPS
   challenge, whereas only 50% of CLEC5A^−/− mice died within the same
   time period and 40% survived until 156 hours post-LPS. In the model of
   polymicrobial sepsis, 80% of CLEC5A^−/− mice survived until 72 hours
   post-CLP compared to 100% death during the same timeframe in WT
   littermates. Depletion of CLEC5A reduced mortality from 100 to 60% in
   mice within 156 hours following CLP or LPS challenge. Improvement in
   the survival of CLEC5A^−/− mice challenged by CLP or LPS suggests that
   CLEC5A is a crucial factor in the development or progression of sepsis.

scRNA-seq profiles of major cell types in the lungs following CLP: Focusing
on endothelial cells

   To further explore the role of CLEC5A in sepsis-induced lung injury, a
   comprehensive single-cell atlas of the lungs was generated by
   single-cell RNA sequencing (scRNA-seq) ([92]Fig. 2A). Mice were
   subjected to CLP, and the lungs were collected 12 hours postsurgery.
   Single-cell suspensions were prepared from lung tissues of three
   CLEC5A^−/− mice and three WT littermates. Pooled cells from these six
   mice were subjected to scRNA-seq processing using the 10×Genomics
   Chromium system. A total of 55,177 single cells, including 29,306 cells
   from WT lungs and 25,871 cells from CLEC5A^−/− lungs, were obtained
   after removal of multiplets and filtration of low-quality cells. The
   software programs Cell Ranger and Seurat were used to analyze the
   sequencing data and generate the single-cell transcriptome. According
   to principal components analysis (PCA) and clustering analysis, total
   cells were grouped into 34 distinct clusters by uniform manifold
   approximation and projection (UMAP), and these cell clusters were
   further identified as different cell types based on the expression of
   their marker genes using the FindAllMarkers of Seurat ([93]Fig. 2B).
   UMAP analysis showed a total of 14 distinct cell types comprising the
   mouse lungs, including granulocytes, T cells, B cells,
   alveolar-macrophages (AM), monocytes, macrophages, natural killer T
   cells (NKT), mesothelium, natural killer cells (NK), endothelial cells,
   fibroblasts, dendritic cells (DC), epithelial cells, and smooth muscle
   cells (SMC). They could be grouped into immune cells (including
   granulocyte, T cell, B cell, AM, monocyte, macrophage, NKT, NK, and DC)
   and nonimmune cells (including mesothelium, endothelial, fibroblast,
   epithelial, and SMC). In general, the septic lungs were largely
   populated by the immune cells, in which granulocytes, T cells and
   monocytes were the most prevalent types. Overall, there was an increase
   in the relative percentage of immune cells in lung tissues of
   CLEC5A^−/− mice compared to WT lungs following CLP. Among immune cells,
   the proportion of most cell types increased in CLEC5A-deficient lungs.
   Notably, the proportion of monocytes was largely decreased in the lungs
   of CLEC5A-deficient mice in the setting of CLP-induced lung injury.
   Among nonimmune cells, endothelial and epithelial cells were relatively
   dominant in the septic lungs of CLEC5A^−/− mice.

Fig. 2. scRNA-seq analysis of endothelial cells in the lungs after CLP.

   [94]Fig. 2.
   [95]Open in a new tab

   (A) Overview of the study design for scRNA-seq. (B) UMAP plot of 14
   cell types identified from the lungs of CLEC5A^−/− mice and WT
   littermates (biological replicates, n = 3 per group), including
   granulocyte, T cell, B cell, AM, monocyte, macrophage, NKT,
   mesothelium, NK, endothelial, fibroblast, DC, epithelial, and SMC. They
   were grouped into immune cells and nonimmune cells. (C) UMAP plot of
   endothelial features showing the distribution of established marker
   gene PECAM1. Gene expression level is shown by purple color grading.
   (D) Violin plot depicting the expression level of endothelial cell-type
   marker gene PECAM1. (E) Representative immunofluorescence images
   showing coexpression of CLEC5A and CD31 (encoded by gene PECAM1) in
   lung tissues (biological replicates, n = 6 per group). Scale bars, 50
   μm. Percentage of CLEC5A^+CD31^+ in total CD31^+ cells was quantified.
   (F) Relative expression of PECAM1 mRNA and CD31 protein levels in
   PMVECs. The statistical significance between two groups was determined
   by an unpaired two-tailed t test and comparisons among more than two
   groups by one-way ANOVA analysis. (G) UMAP plot of endothelial cell
   clusters in CLEC5A^−/− mice (red dots) and WT littermates (blue dots).
   (H) UMAP plot of endothelial cell subtypes and cell percentage,
   including endothelial cells of lymphatic, aerocyte, gCap, arterial, and
   venous types. (I) GO and KEGG enrichment analysis for DEGs in
   endothelial cells between CLEC5A^−/− mice and WT littermates (FC > 1.5
   or FC < 0.67 and P < 0.05). Bar plots of top 10 enriched GO-BP and KEGG
   pathways. The significance of enrichment was determined by P value <
   0.05.

   Although much is known about CLEC5A in immune cells, there is little
   information about the role of CLEC5A in endothelial cells. The
   distribution and expression level of the marker gene PECAM1, which has
   been used previously for identification of endothelial cells in whole
   lung tissues, were displayed by UMAP analysis ([96]Fig. 2C). Mice
   lacking CLEC5A had a slightly decreased trend in the number of
   endothelial cells ([97]Fig. 2B). Changes in the proportion of
   endothelial cells were further examined by specific marker gene
   expression. We found the expression level of PECAM1 was significantly
   up-regulated in CLEC5A^−/− lungs compared to control lungs ([98]Fig.
   2D). As the gene PECAM1 encodes for the endothelial surface marker
   CD31, we used CD31 staining to identify endothelial cells.
   Immunofluorescence staining confirmed the expression of CLEC5A in
   CD31-positive endothelial cells in mouse lung tissues ([99]Fig. 2E).
   There was a decreased trend of CD31 staining in septic lungs, whereas
   coexpression of CLEC5A and CD31 was increased after CLP. The mRNA and
   protein expression of PECAM1/CD31 was significantly down-regulated in
   the lungs of CLP mice but up-regulated in CLEC5A^−/− mice compared to
   WT mice after CLP ([100]Fig. 2F). CLEC5A was significantly up-regulated
   in CD31-positive endothelial cells in the lungs upon septic infection.

Characterization of endothelial cell subtypes and transcriptome analysis by
scRNA-seq in CLEC5A^−/− mice

   An in-depth analysis was performed examining aspects of cell subtype
   characterization, cell proportion, and transcriptomics in the
   endothelium ([101]Fig. 2G). Within the endothelial cell clusters (a
   total of 1621 single cells), five subtypes were identified, including
   lymphatic, aerocyte, general capillary (gCap), arterial, and venous
   types ([102]Fig. 2H). The cellular composition was further analyzed by
   calculating the relative percentage of each subtype. The gCap subtype
   was the most populous of the endothelial cell subtypes, and there was
   no significant difference in the proportion of each subtype between two
   genotypes. Generally, endothelial cells belong to two main subsets:
   lymphatic and vascular. The vascular endothelial cells contain the
   subtypes arterial, venous, and capillary (aerocyte and gCap). Capillary
   endothelial cells were the most prevalent type in the endothelium of
   septic lungs ([103]Fig. 2H). We further examined the expression pattern
   of CLEC5A in theses endothelial subtypes in the lungs of WT mice after
   CLP challenge (fig. S4). Immunostaining showed that CLEC5A was highly
   expressed in all endothelial subtypes, with the use of cell marker Gap
   Junction Protein Alpha 5 (GJA5) for arterial type, Von Willebrand
   Factor (VWF) for venous type, Endothelin Receptor Type B (EDNRB) for
   capillary type, Protein Tyrosine Phosphatase Receptor Type B (PTPRB)
   for gCap type, and Prospero Homeobox 1 (PROX1) for lymphatic type.

   The DEGs between the endothelial cells comprising CLEC5A^−/− and WT
   lungs that meet the criteria of FC >1.5 or <0.67 and P < 0.05 were
   subjected to Gene Ontology (GO) enrichment analysis and Kyoto
   Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis
   ([104]Fig. 2I). Among the top 10 enriched GO and KEGG pathways, tumor
   necrosis factor (TNF) signaling pathway, Staphylococcus aureus
   infection, interleukin-17 (IL-17) signaling pathway, inflammatory
   response to wounding, and activation of membrane attack complex are
   strongly associated with the immune/inflammatory response and the
   development of sepsis and lung injury ([105]15–[106]17). These DEGs
   were also significantly enriched in genes involved in negative
   regulation of cell migration and negative regulation of cell motility,
   which are involved in immune cell recruitment and thus
   immune/inflammation regulation ([107]18).

CLEC5A^−/− transgenic mice are resistant to sepsis-induced lung injury and
inflammation

   There were significant abnormalities in the lungs following CLP
   ([108]Fig. 3). The bronchoalveolar lavage fluid (BALF) and lungs were
   harvested 12 hours post-CLP. The morphological changes in the lungs
   were attenuated in CLEC5A^−/− mice following CLP, as evidenced by less
   pulmonary hemorrhage and edema ([109]Fig. 3A), as well as decreased
   lung index (lung-to-body weight ratio) and lung wet-to-dry weight ratio
   ([110]Fig. 3, B and C). Histopathological analysis was carried out with
   hematoxylin and eosin (H&E) staining ([111]Fig. 3D) and lung injury
   score ([112]Fig. 3E). H&E staining displayed alveolar hemorrhage,
   alveolar space enlargement, and inflammatory cell infiltration in the
   lungs of mice subjected to CLP-induced polymicrobial sepsis, which
   correspond to a higher lung injury score. The severity of CLP-induced
   lung injury was significantly lessened in CLEC5A^−/− mice compared to
   WT littermates. CLP induced pulmonary fluid collections, and the
   transudate sampled by BALF was enriched in infiltrated cells, including
   leukocytes, neutrophils, and monocytes, in comparison with
   infiltrate-poor cells in mice that received sham operation ([113]Fig.
   3F). Along with inflammatory cell infiltration, the levels of
   pro-inflammatory cytokines including TNF-α and IL-6, and chemokines
   including monocyte chemotactic protein 1 (MCP-1) and C-X-C motif
   chemokine ligand 5 (CXCL5), were increased in the lungs from mice
   challenged by CLP ([114]Fig. 3G), as well as in the BALF ([115]Fig.
   3H). Deletion of CLEC5A inhibited inflammatory cell influx into BALF
   and decreased levels of pro-inflammatory cytokines and chemokines in
   the BALF and lungs of CLP-challenged mice ([116]Fig. 3, G and H). No
   significant difference was observed between CLEC5A^−/− and WT mice
   under sham conditions, indicating that dysregulated CLEC5A might
   contribute to lung injury only under pathological conditions.

Fig. 3. Deletion of CLEC5A protects against CLP-induced lung injury in mice.

   [117]Fig. 3.
   [118]Open in a new tab

   CLEC5A^−/− transgenic mice and WT littermates were subjected to CLP to
   induce polymicrobial sepsis, and mice received sham operation as the
   control. The lungs were collected at 12 hours post-CLP. (A) Gross
   morphology of the lung at 12 hours post-CLP operation. (B) Lung index
   shown as the percentage of lung weight to body weight. (C) Ratio of
   lung wet-to-dry weight. (D) Representative images of H&E staining of
   lung tissues in sham and CLP mice. In the CLP group, images on the
   right panel represent the region marked by a black square from the left
   panel. Scale bars, 100 μm. (E) Severity of CLP-induced lung injury
   determined by inflammation, edema, hemorrhage, and alveolar septal
   thickening, corresponding to H&E staining. (F) Inflammatory cell counts
   in BALF, including total leukocytes and differential neutrophils and
   monocytes. (G and H) Levels of inflammatory cytokines and chemokines in
   BALF and lung tissues, including TNF-α, IL-6, MCP-1, and CXCL5. The
   statistical significance for lung wet/dry weight ratio and lung injury
   score was determined by the Kruskal-Wallis test, and others were
   determined by one-way ANOVA analysis (biological replicates, n = 6 per
   group for each experiment).

Endothelial knockdown of CLEC5A improves survival of mice with polymicrobial
sepsis

   CLP-induced lung injury is primarily due to alveolar-capillary barrier
   dysfunction. The results of scRNA-seq transcriptome analysis in
   CLEC5A-deficient endothelium suggest the importance of endothelial
   CLEC5A in the development of lung injury and therefore mortality in
   sepsis. Endothelial specific expression or knockdown of CLEC5A was used
   in CLEC5A^−/− and WT mice under the endothelial-specific promoter Tie1
   by adeno-associated virus type 9 (AAV9) system ([119]Fig. 4A). We
   confirmed the cellular target of in vivo AAV9-Tie1 infection in the
   lungs of CLEC5A^−/− mice. Five subtypes (arterial, venous, gCap,
   aerocyte, and lymphatic) of endothelial cells were identified from
   scRNA-seq. Immunostaining displayed that the expression of CLEC5A was
   notably increased in each subtype of endothelial cells via the
   AAV9-Tie1 system in CLEC5A^−/− mice (fig. S5, A and B). Primary PMVECs
   lining the lung’s small blood vessels, including capillaries,
   arterioles and venules, were isolated from WT mice. The efficiency of
   AAV9 carrying Tie1 promoter approach to endothelial-specific CLEC5A
   overexpression (CLEC5A^Tie1-oe) or knockdown (CLEC5A^Tie1-sh) was
   validated in PMVECs (fig. S6A). Alveolar epithelial cells (AECs) are an
   important cell type in the lung and responsible for alveolar-capillary
   barrier function, which are closely associated with pulmonary edema and
   inflammation. In nonendothelial AECs, CLEC5A expression was not
   affected by the AAV9-Tie1 system (fig. S6B). These data indicated the
   efficacy of promoter Tie1 to selectively target CLEC5A in pulmonary
   endothelial cells via AAV9 in vivo infection, whereas no specificity in
   endothelial cell subtypes. The restoration of CLEC5A expression level
   was examined in PMVECs of CLEC5A^−/− mice after AAV9-Tie1 injection
   ([120]Fig. 4B). We found that the level of endothelial CLEC5A in
   CLEC5A^−/− mice rescued by AAV9-Tie1 (CLEC5A^Tie1-oe) infection was
   still lower than its endogenous level in WT mice. Mice with endothelial
   CLEC5A overexpression showed no significant difference in survival rate
   after CLP challenge compared with WT mice, 100% of which died within 72
   hours postoperation ([121]Fig. 4C). In contrast, the survival after CLP
   was significantly improved in mice lacking endothelial CLEC5A. Thirty
   percent of CLEC5A^Tie1-sh mice survived until the end of study. In the
   setting of global deletion of CLEC5A, 50% of mice survived until 156
   hours post-CLP, whereas restoration of endothelial CLEC5A expression
   reduced the survival rate to 20% in CLEC5A^−/− mice ([122]Fig. 4D),
   which is still higher than WT mice. This survival difference could be
   attributed to the level of endothelial CLEC5A, which was not fully
   restored by the AAV9-Tie1 system. We suggest endothelial CLEC5A as a
   potential pathological factor in CLP-induced polymicorbial sepsis, and
   its level may be correlated with the survival. Whether the magnitude of
   CLEC5A expression within different subtypes of endothelial cell
   influences the survival outcomes and the influence of each subtype
   needs more studies.

Fig. 4. Specific deletion of endothelial CLEC5A improves the survival after
CLP and decreases pulmonary endothelial barrier permeability.

   [123]Fig. 4.
   [124]Open in a new tab

   (A) In vivo endothelial-specific expression or knockdown of CLEC5A was
   carried out via tail vein injection of AAV9 under promoter Tie1. (B)
   Expression of CLEC5A in PMVECs isolated from WT and CLEC5A^−/− mice
   (biological replicates, n = 4 per group). (C) Survival rate of WT mice
   with endothelial-specific CLEC5A overexpression or knockdown
   (CLEC5A^Tie1-oe or CLEC5A^Tie1-sh) (biological replicates, n = 10 per
   group). (D) Survival rate of CLEC5A^−/− mice with CLEC5A^Tie1-oe
   (biological replicates, n = 10 per group). The statistical significance
   between survival curves was determined by P value using the log-rank
   test for trend. (E) Experimental design for in vivo and in vitro
   studies of pulmonary endothelial barrier function. (F) Pulmonary
   microvascular albumin leakage was determined as the content of EB
   dye-labeled albumin in the lung (biological replicates, n = 6 per
   group). (G and H) Expression of VE-cadherin in lung tissues and PMVECs
   isolated from CLEC5A^−/− mice and WT littermates after CLP challenge
   (biological replicates, n = 6 per group). (I) The permeability of the
   PMVEC monolayer was measured using FITC-labeled dextran by a transwell
   assay under LPS (10 μg/ml, 24 hours) (biological replicates, n = 4 per
   group). (J) TEER was assessed in LPS-challenged PMVECs (10 μg/ml, 4
   hours) (biological replicates, n = 4 per group). (K) Quantification and
   representative images of immunofluorescence staining for VE-cadherin in
   PMVECs. Scale bar, 50 μm. (L) Expression of VE-cadherin in PMVECs with
   lentivirus-mediated CLEC5A overexpression (CLEC5A^oe) or knockdown
   (CLEC5A^sh) under LPS treatment (10 μg/ml, 24 hours) (biological
   replicates, n = 4 per group). The statistical significance between two
   groups was determined by an unpaired two-tailed t test and comparisons
   among more than two groups by one-way ANOVA analysis.

   Having observed the survival advantages of CLEC5A genetic deletion or
   endothelial-specific knockdown in CLP mice, we further assessed the
   bacterial loads in this model. At 24 hours after CLP, CLEC5A^−/− mice
   had an increase in the number of colony-forming units (CFUs) in the
   blood but decreases in the BALF and lungs, compared to WT littermates
   (fig. S7A). There was no significant difference in the blood, BALF, and
   lung bacterial contents from mice deficient in endothelial CLEC5A,
   compared to the controls (fig. S7B). These results indicate that the
   favorable effect of endothelial CLEC5A deficiency on the survival is
   likely not associated with bacterial clearance or spreading in the
   setting of CLP-induced abdominal sepsis. The mechanism underlying the
   therapeutic potential of targeting endothelial CLEC5A in sepsis needs
   to be unveiled and is the focus of this study.

CLEC5A deficiency ameliorates pulmonary endothelial barrier dysfunction
during inflammatory stimulation

   Pulmonary microvascular albumin leakage was determined by using Evans
   blue (EB) dye after CLP or sham operation ([125]Fig. 4, E and F). CLP
   induced microvascular leakage of EB-labeled albumin in the lungs of
   both CLEC5A^−/− transgenic mice and WT littermates. Mice lacking CLEC5A
   had less pulmonary albumin leakage than WT littermates after CLP
   challenge. CLP induces endothelial barrier dysfunction and resulting
   hyperpermeability and vascular leakage, which leads to albumin leakage
   and inflammatory cytokine and cell infiltration into the endothelium,
   thus causing tissue damage ([126]Fig. 3). The expression of vascular
   endothelial (VE)–cadherin, an endothelial cell junctional protein, was
   examined to assess endothelial integrity. In line with the in vivo
   data, the expression of VE-cadherin in lung tissues and primary PMVECs
   was increased in CLEC5A^−/− mice ([127]Fig. 4, G and H). We next
   investigated the dysfunctional cellular mechanisms underlying the
   therapeutic potential of CLEC5A in sepsis-associated lung injury. Under
   pathological conditions, dysfunction of vascular endothelial cells
   causes barrier failure ([128]10). Whether CLEC5A contributes to PMVEC
   dysfunction was explored using inflammatory stimulation with LPS.
   PMVECs were isolated from WT littermates and identified by
   CD31-positive immunofluorescence staining (fig. S2, C and D). The
   expression of CLEC5A was significantly increased in PMVECs upon LPS (10
   μg/ml) stimulation (fig. S2, E and F). The barrier integrity and
   function of the PMVEC monolayer were determined during LPS-induced
   inflammation, and vehicle treatment served as the negative control. LPS
   stimulation impaired PMVEC barrier function, as shown by increased cell
   permeability of 1.75 fold in WT littermates ([129]Fig. 4I). Deletion of
   CLEC5A protected PMVECs against LPS-induced hyperpermeability. The
   trans-endothelial electrical resistance (TEER) of PMVECs was lowered in
   the presence of LPS, and CLEC5A-deficient PMVECs were resistant to
   LPS-induced TEER loss ([130]Fig. 4J). PMVECs challenged by LPS
   expressed less immunofluorescence for VE-cadherin, and genetic deletion
   of CLEC5A restored VE-cadherin level to a certain extent ([131]Fig.
   4K). Lentivirus-mediated gene delivery and knockdown was applied in
   vitro with RNA interference (RNAi) technology, and the efficiency of
   targeting CLEC5A was validated in PMVECs (fig. S6C). In WT PMVECs,
   VE-cadherin expression was reduced by overexpression of CLEC5A and
   increased by knockdown of CLEC5A under LPS stimulation ([132]Fig. 4L).

   In consistent with the in vivo data, deletion of CLEC5A had no effect
   on the barrier integrity and function of PMVECs under sham or vehicle
   treatment ([133]Figs. 3 and [134]4). CLEC5A was lowly expressed in
   normal lung tissues and PMVECs ([135]Fig. 1, D to F). In comparison,
   CLEC5A was highly expressed during CLP or LPS-induced inflammation and
   closely associated with endothelial barrier dysfunction and lung
   injury. We suggest a pathological role of endothelial CLEC5A in
   sepsis-associated lung injury.

CLEC5A promotes endothelial adhesion and transmigration of monocytes

   From the scRNA-seq analysis of mouse lungs, we observed that the number
   of monocytes was significantly lower in the septic lungs of CLEC5A^−/−
   mice compared to those of WT littermates ([136]Fig. 5A). UMAP analysis
   visualized the difference in monocyte clusters from CLEC5A^−/− and WT
   lungs. A total of 5786 single cells were identified as monocytes based
   on clustering analysis and cell type characterization, 3581 of which
   were obtained from WT littermates (n = 3) and 2205 of which were from
   CLEC5A^−/− mice (n = 3). Consistently, the proportion of monocytes in
   total immune cells was significantly decreased in CLEC5A-deficient
   lungs. Pulmonary infiltration of inflammatory cells, including
   monocytes, was increased in the setting of CLP-induced lung injury,
   which was ameliorated by genetic deletion of CLEC5A in mice ([137]Fig.
   2B). Moreover, the expression levels of adhesion molecules were
   generally down-regulated in CLEC5A-deficient lung-derived endothelial
   cells, specifically in the capillary subtype of endothelial cells
   ([138]Fig. 5B). Among these adhesion molecules, down-regulation of
   intercellular adhesion molecule 1 (ICAM-1) and vascular cell adhesion
   molecule 1 (VCAM-1) was the most significantly affected by deletion of
   CLEC5A. According to gene set enrichment analysis (GSEA), we found
   multiple GO–Biological Process (BP) that are associated with cell
   adhesion were significantly enriched in the endothelial cells
   ([139]Fig. 5C), such as the regulation of cell-matrix adhesion
   (P = 0.0036) and the negative regulation of leukocyte adhesion to
   vascular endothelial cell (P = 0.0021). These cell adhesion–related GO
   pathways were mainly enriched by the DEGs that were up-regulated in
   CLEC5A^−/− endothelial cells. These findings suggest that endothelial
   cells may contribute to prevention of monocyte influx into the lung,
   thus protecting the lung from septic infection.

Fig. 5. CLEC5A promotes endothelial adhesion and transmigration of monocytes
in vitro.

   [140]Fig. 5.
   [141]Open in a new tab

   (A) UMAP plot of monocyte clusters in WT lungs (n = 3) and CLEC5A^−/−
   lungs (n = 3) and percentage in immune cells. (B) Dot plot showing the
   expression of adhesion molecules in each subtype of endothelial cells.
   (C) GSEA of the DEGs in the endothelial cells between CLEC5A^−/− and WT
   lungs showing significant enrichment of cell-adhesion related-GO
   pathways, in which the up-regulated DEGs were mainly enriched. The
   significance was determined by P value with false discovery rate (FDR)
   < 0.25. (D) Mouse PMVECs were isolated and challenged by LPS (10 μg/ml)
   for 24 hours. (E) Levels of MCP-1 and VCAM-1 in the supernatant by
   ELISA. (F) Relative mRNA expression of MCP-1 and VCAM-1 in PMVECs. (G)
   Overview of in vitro adhesion and trans-endothelial migration assays in
   mouse PMVECs and HUVECs. (H) Mouse PBMCs were labeled by BCECF-AM, and
   the adherent monocytes to PMVECs were observed under a fluorescence
   microscopy. Scale bar, 50 μm. Monocyte-endothelial adhesion was
   quantified as the percentage of fluorescence intensity relative to the
   WT group. (I) Transmigration of mouse PBMCs across the PMVEC monolayer
   was shown as the percentage of transmigrated cells relative to the WT
   group by MTT assay. Lentivirus-mediated gene delivery for CLEC5A
   (CLEC5A^oe) or shRNA targeting CLEC5A (CLEC5A^sh) were carried out in
   HUVECs. (J) Adhesion of THP-1 to HUVECs was shown as fluorescence and
   quantified as the percentage of fluorescence intensity relative to the
   control group. Scale bar, 50 μm. (K) Migration of THP-1 through the
   HUVEC monolayer shown as the percentage of transmigrated cells relative
   to the control group. The statistical significance was determined by an
   unpaired two-tailed t test (or with Welch’s correction) for comparison
   between two groups and by one-way ANOVA test for comparisons among four
   groups (biological replicates, n = 4 per group for each experiment).

   We here explored the role of CLEC5A in leukocyte-endothelial
   interactions and assessed the interaction of vascular endothelial cells
   and monocytes in vitro. PMVECs isolated from CLEC5A^−/− mice and WT
   littermates were stimulated by LPS for modeling sepsis-induced
   immune/inflammatory dysfunction in the endothelium ([142]Fig. 5D). The
   extravasation of immune cells is partially attributed to functional
   alterations in vascular endothelial cells, such as phenotypic changes
   toward pro-adhesion and production of adhesion molecules. The cellular
   mRNA expression and protein secretion of PMVEC-derived chemokine MCP-1
   and adhesion molecule VCAM-1 were increased with LPS stimulation
   ([143]Fig. 5, E and F). The production of these LPS-induced molecules
   was lowered in PMVECs lacking CLEC5A. Loss of CLEC5A had no effect on
   the expression and secretion of MCP-1 and VCAM-1 by PMVECs in the
   absence of LPS. After exposure to LPS (10 μg/ml) for 5 hours, PMVECs
   were cocultured with fluorescence probe (BCECF-AM)–labeled mouse PBMCs
   for 1 hour, and decreased adherent monocytes were observed in PMVECs
   lacking CLEC5A with fluorescence microscopy ([144]Fig. 5, G and H).
   Deletion of CLEC5A caused more than a 40% reduction in
   monocyte-endothelial adhesion after LPS stimulation. Similarly, the
   transmigration of PBMCs across LPS-challenged PMVECs had a 50% decrease
   in the absence of CLEC5A ([145]Fig. 5I).

   We further performed gain-of-function and loss-of-function experiments
   for CLEC5A in PMVECs with the use of lentiviral vectors (fig. S8A).
   Consistent with the above results, knockdown of CLEC5A in PMVECs
   decreased the production of MCP-1 and VCAM-1 (fig. S8, B and C) and
   inhibited monocyte adhesion (fig. S8D) and trans-endothelial migration
   (fig. S8E). Conversely, overexpression of CLEC5A promoted LPS-induced
   transmigration of PBMCs across PMVECs.

   Next, the regulation of CLEC5A on human monocyte-endothelial cell
   interactions was investigated in vitro. The mRNA and protein expression
   of CLEC5A was increased in both human umbilical cord endothelial cells
   (HUVECs) and human pulmonary microvascular endothelial cells (HMVECs)
   under LPS stimulation (fig. S2, G and H). Knockdown of CLEC5A in HUVECs
   suppressed LPS-induced endothelial adhesion and transmigration of human
   monocytes THP-1 ([146]Fig. 5, J to K, and fig. S6D). In contrast,
   forced overexpression of CLEC5A increased THP-1 adhesion to HUVECs and
   trans-endothelial migration.

CLEC5A contributes neutrophil-endothelial interaction

   Damage to endothelial cells and resulting barrier dysfunction during
   sepsis stimulate more immune cells, such as neutrophils, to cross the
   endothelium, leading to local inflammation and tissue damage ([147]9).
   These in vivo experiments show that genetic deletion of CLEC5A
   inhibited the extravasation of neutrophils and influx into the lungs in
   mice with CLP-induced polymicrobial sepsis. LPS stimulation induced
   expression and release of CXCL5 and ICAM-1 that contribute to
   neutrophil recruitment and adhesion ([148]Fig. 6, A to D). Knockdown of
   CLEC5A significantly lowered the production of CXCL5 and ICAM-1 in
   LPS-challenged PMVECs, whereas it had no effect in the absence of LPS.
   Down-regulation of these factors expressed by CLEC5A-deficient PMVECs
   might account for reduced transmigration of neutrophils. By comparing
   PMVECs isolated from CLEC5A^−/− transgenic mice and WT littermates, we
   found CLEC5A deficiency caused a 40% reduction in the transmigration of
   neutrophils across the PMVEC monolayer under LPS-stimulated
   inflammation ([149]Fig. 6, E and F). The involvement of CLEC5A in
   neutrophil-endothelial cell interaction was further validated by gene
   knockdown and overexpression experiments using lentiviral transduction.
   PMVECs with lentivirus-mediated CLEC5A knockdown induced decreases in
   CXCL5 and ICAM-1 expression and neutrophil transmigration under LPS
   stimulation ([150]Fig. 6, B, D, and G). In contrast, viral delivery of
   transgene CLEC5A increased trans-endothelial migration of neutrophils
   and PMVEC-derived CXCL5 and ICAM-1 levels. In the absence of LPS, loss
   of CLEC5A did not induce damage to endothelial cells either in terms of
   permeability or barrier function ([151]Figs. 4 to [152]6).

Fig. 6. CLEC5A promotes trans-endothelial migration of neutrophils in vitro.

   [153]Fig. 6.
   [154]Open in a new tab

   Mouse PMVECs were exposed to LPS (10 μg/ml) for 24 hours, and the
   production of CXCL5 and ICAM-1 was detected in the presence or absence
   of CLEC5A. Expression of CXCL5 and ICAM-1 by real-time PCR in (A)
   PMVECs from CLEC5A^−/− transgenic mice and (B) in PMVECs with
   lentivirus-mediated overexpression (CLEC5A^oe) or knockdown of CLEC5A
   (CLEC5A^sh). (C and D) Respective levels in the supernatant by ELISA.
   (E) Migration of neutrophils across LPS-challenged PMVECs was
   determined by a transmigration assay. Peripheral blood granulocytes
   were obtained from normal C57BL/6J mice, and granulocytes that
   transmigrated across the PMVEC monolayer were identified as neutrophils
   under phase contrast microscopy. Representative images of transmigrated
   neutrophils to cross (F) PMVECs from CLEC5A^−/− mice and (G) PMVECs
   infected with lentivirus-expressing CLEC5A (CLEC5A^oe) or shRNA
   targeting CLEC5A (CLEC5A^sh). Scale bars, 50 μm. Trans-endothelial
   migration of neutrophils was quantified as the percentage of
   transmigrated cell number relative to WT or control group. Cells with
   LPS stimulation but without infection were served as the controls. The
   statistical significance was determined by an unpaired two-tailed t
   test for comparison between two groups and by one-way ANOVA test for
   comparisons among four groups (biological replicates, n = 4 per group
   for each experiment).

CLEC5A is critical in bacterial pneumonia induced by E. coli intratracheal
instillation

   To further determine whether the role of CLEC5A in lung inflammation
   was restricted to abdominal sepsis, we established a mouse model of
   bacterial pneumonia. CLEC5A^−/− mice and WT littermates received an
   intratracheal instillation of Gram-negative bacterium E. coli. Similar
   to the results in the CLP model, E. coli–induced pulmonary edema and
   pathological lesions in lung tissues by H&E staining were alleviated in
   CLEC5A^−/− mice ([155]Fig. 7, A and B) and endothelial CLEC5A-deficient
   mice ([156]Fig. 7, C and D). Global or endothelial-specific deletion of
   CLEC5A increased the expression of VE-cadherin in lung tissues after E.
   coli infection ([157]Fig. 7, E and F). Consistently, pulmonary
   microvascular albumin leakage ([158]Fig. 7, G and H) as well as
   inflammatory cell infiltration ([159]Fig. 7, I and J) and cytokine
   levels in BALF ([160]Fig. 7, K and L) were reduced in the absence of
   endothelial CLEC5A. Either global deletion or endothelial-specific
   knockdown of CLEC5A had advantages in host defense against E. coli
   infection and the resulting lung inflammation and vascular injury.
   Furthermore, the bacterial loads were determined in E. coli–induced
   pneumonia model. The bacterial CFUs were increased in the blood but
   decreased in the BALF and lungs in CLEC5A^−/− mice after E. coli
   intratracheal inoculation (fig. S7C), whereas there was no significant
   difference in mice with endothelial CLEC5A knockdown (fig. S7D).

Fig. 7. Endothelial CLEC5A deficiency protects mice against E. coli–induced
pneumonia.

   [161]Fig. 7.
   [162]Open in a new tab

   Mice were intratracheally inoculated with 1.5 × 10^7 CFUs of E. coli
   and euthanized after 16 hours. The lungs were collected from CLEC5A^−/−
   mice and endothelial-deficient mice (CLEC5A^Tie1-sh). (A and C) Ratio
   of lung wet-to-dry weight. (B and D) Representative images of H&E
   staining of lung tissues and quantitative analysis of lung injury
   severity. Scale bars, 100 μm. (E and F) Relative expression of
   VE-cadherin in lung tissues. (G and H) Mice received EB at 16 hours
   after bacterial inoculation, and pulmonary microvascular albumin
   leakage (μg EB/g lung per minute) was determined after 30 min. (I and
   J) BALF cell counts of leukocyte, neutrophil, and monocyte. (K and L)
   BALF levels of TNF-α and IL-6. The statistical significance between two
   groups was determined by an unpaired two-tailed t test, and the
   Mann-Whitney test was used for lung injury score (biological
   replicates, n = 6 per group for each experiment).

Exploration of the downstream targets of CLEC5A responsible for endothelial
dysfunction in inflammatory lung injury

   Having observed the role of CLEC5A in the protection of endothelium
   against barrier failure during inflammation as well as the contribution
   of endothelial CLEC5A to sepsis-induced mortality, we further explored
   the downstream targets of CLEC5A to uncover potential molecular
   mechanisms in the endothelium. An in-depth scRNA-seq transcriptomic
   analysis was then carried out in endothelial cells. Heatmaps reveal the
   total or specific DEGs in vascular and capillary endothelial cells from
   CLEC5A^−/− and WT lungs challenged by CLP ([163]Fig. 8A). Capillary
   endothelial cells were found to have the largest number of DEGs between
   CLEC5A^−/− and WT lungs ([164]Fig. 8B), implying that capillary
   endothelial cells play an important role in sepsis-associated lung
   injury. Analysis of scRNA-seq showed many overlapping DEGs in the
   vascular subtype of endothelial cells, and some specifically
   dysregulated in the capillary type of endothelial cells. These
   single-cell transcriptomic data demonstrate cell heterogeneity of the
   lung endothelium and endothelial cell hierarchy in CLEC5A-deficient
   mice following CLP challenge. Specifically, capillary endothelial cells
   isolated from septic lungs of CLEC5A^−/− mice display overlapping and
   unique features in aspects of endothelial dysfunction, sepsis
   pathophysiology, and the development of sepsis-associated lung injury.

Fig. 8. Transcriptome changes in vascular/capillary endothelial cells by
scRNA-seq analysis.

   [165]Fig. 8.
   [166]Open in a new tab

   An in-depth analysis of the DEGs was carried out in specific
   endothelial subtypes. (A) Heatmaps of the DEGs among five subtypes of
   endothelial cells between WT and CLEC5A^−/− lungs. The column is
   hierarchically ordered by the subtypes of endothelial cells (aerocyte,
   arterial, gCap, lymphatic, and venous) and then groups of WT and
   CLEC5A^−/−, with single-cell gene expression represented by each
   column. The vascular endothelial cells include subtypes of aerocyte,
   arterial, gCap and venous. The capillary endothelial cells include
   aerocyte and gCap of endothelial cells. (B) Number of the DEGs in
   capillary (gCap and aerocyte), arterial, venous, and lymphatic types of
   endothelial cells. (C) Violin plots showing the expression of SCARB1,
   PODXL, RAMP2, and FABP4 in capillary endothelial cells. The expression
   was further validated in primary PMVECs isolated from CLEC5A^−/− mice
   or WT littermates 12 hours post-CLP. (D) Representative blots showing
   SCARB1, PODXL, FABP4, and RAMP2 expression. (E and F) The relative
   protein and mRNA expression levels were analyzed (biological
   replicates, n = 4 per group). The statistical significance was
   determined by an unpaired two-tailed t test for comparison between two
   groups.

   We observed that some DEGs were significantly and specifically
   dysregulated in the vascular or capillary subtype of endothelial cells
   ([167]Fig. 8, A and B). Among them, we noticed a PRR, scavenger
   receptor class B membrane 1 (SCARB1), and an anti-adhesive factor,
   podocalyxin like (PODXL), the expression of which was significantly
   up-regulated ([168]Fig. 8C). Of note, fatty acid–binding protein 4
   (FABP4) and receptor activity modifying protein 2 (RAMP2) were found to
   be significantly down-regulated in capillary endothelial cells from
   CLEC5A-deficient lung ([169]Fig. 8C). These potential targets
   downstream of CLEC5A were further confirmed in CLEC5A-deficient PMVECs.
   We found that the expression of SCARB1, PODXL, FABP4, and RAMP2 was not
   affected in PMVECs from CLEC5A^−/− mice under physiological conditions
   (fig. S9, A and B), whereas they were specifically targeted by CLEC5A
   upon CLP challenge. Deletion of CLEC5A increased endothelial expression
   of SCARB1 and PODXL and decreased RAMP2 and FABP4 at both protein and
   mRNA levels in mice subjected to CLP ([170]Fig. 8, D to F). Next, we
   examined whether these DEGs are involved in the therapeutic effect of
   CLEC5A deficiency in CLP-induced sepsis and lung injury. Endothelial
   specific knockdown of SCARB1 or PODXL and re-enforced expression of
   FABP4 or RAMP2 was carried out under the endothelial-specific promoter
   Tie1 by AAV9 in CLEC5A^−/− mice as previously described.

Potential molecular mechanisms underlying endothelial CLEC5A-mediated lung
inflammation and vascular injury in sepsis

   SCARB1 as a PRR is able to sense PAMPs and DAMPs, involves in host
   defense against microbial infection ([171]19), and has been reported to
   protect against polymicrobial sepsis and endotoxemia in mice ([172]20,
   [173]21). Inhibition of endothelial SCARB1 completely abolished the
   improvement in the survival rate observed in CLP-challenged CLEC5A^−/−
   mice ([174]Fig. 9A). All mice that lacked CLEC5A and endothelial SCARB1
   died within 72 hours after CLP. Loss of endothelial SCARB1 aggravated
   CLP-induced lung injury ([175]Fig. 9E), along with increases in
   inflammatory cell (leukocyte, neutrophil, and monocyte) infiltration
   and inflammatory cytokine (TNF-α and IL-6) and chemokine (MCP-1)
   production in the BALF and lungs of CLEC5A^−/− mice ([176]Fig. 9, J, L,
   and N). These results suggest a key contribution of SCARB1 to
   CLEC5A-mediated barrier function in the endothelium and lung protection
   during septic infection.

Fig. 9. Potential downstream targets of CLEC5A in CLP-induced lung injury.

   [177]Fig. 9.
   [178]Open in a new tab

   To validate SCARB1, PODXL, RAMP2, and FABP4 as potential targets of
   CLEC5A in vivo, endothelial-specific gene knockdown/overexpression was
   carried out under Tie1 by AAV9 through tail vein injection. (A to D)
   Survival rate of CLEC5A^−/− mice with endothelial
   knockdown/overexpression of target gene after CLP (biological
   replicates, n = 10 per group). The statistical significance between
   survival curves was determined by P value using the log-rank
   (Mantel-Cox) test. (E to G) H&E staining showing histological changes
   in lung tissues and the corresponding lung injury score (biological
   replicates, n = 6 per group). Scale bars, 100 μm. (H and I) Pulmonary
   microvascular albumin leakage represented as μg EB/g lung per minute
   (biological replicates, n = 6 per group). (J and K) Levels of TNF-α,
   IL-6, and MCP-1 in the BALF (biological replicates, n = 6 per group).
   (L and M) Levels of TNF-α, IL-6, and MCP-1 in the lungs (biological
   replicates, n = 6 per group). (N and O) Quantification of inflammatory
   cells in the BALF, including leukocyte, neutrophil, and monocyte
   (biological replicates, n = 6 per group). (P) HUVECs were coinfected
   with LV expressing CLEC5A^sh and PODXL^sh. The monocyte-endothelial
   adhesion and trans-endothelial migration assays were carried out under
   LPS stimulation. (Q) Adhesion of THP-1 to HUVECs (biological
   replicates, n = 4 per group). (R) Transmigration of THP-1 across HUVECs
   (biological replicates, n = 4 per group). The statistical significance
   was determined by an unpaired two-tailed t test (or with Welch’s
   correction) for comparison between two groups.

   FABP4 is a cytosolic lipid binding protein that plays a role in
   regulating the function of capillary endothelium and
   immune/inflammatory responses ([179]22, [180]23), and its inhibitor had
   therapeutic effects in experimental sepsis-induced ALI ([181]24).
   Genetic overexpression of endothelial FABP4 not only abolished the
   therapeutic effect of CLEC5A depletion but also aggravated CLP-induced
   mortality and lung injury in mice, as evidenced by 100% mortality
   within 60 hours after CLP ([182]Fig. 9B) and higher lung injury score
   at 12 hours after CLP ([183]Fig. 9F). Up-regulation of FABP4 increased
   inflammatory cytokine levels and infiltrated cells in CLP-challenged
   lungs of CLEC5A^−/− mice ([184]Fig. 9, K, M, and O). Investigation of
   FABP4 as a downstream effector of CLEC5A in the endothelium or a
   potential regulator in CLEC5A-mediated inflammatory signaling pathways
   is an area that requires further study.

   PODXL is widely expressed in vascular endothelium and is required for
   functional barrier formation and maintenance under resting and
   inflammatory conditions ([185]25). Following CLP, we found that
   deletion of endothelial PODXL increased CLP-induced mortality to a
   certain extent ([186]Fig. 9C) and significantly increased lung injury
   and vascular leakage in CLEC5A^−/− mice ([187]Fig. 9, G and H).
   Knockdown of PODXL enhanced leukocyte-endothelial adhesion and
   trans-endothelial migration of THP-1 in CLEC5A-deficient HUVECs
   ([188]Fig. 9, P to R). Endothelial specific knockdown of PODXL worsened
   lung injury in CLEC5A^−/− mice by inducing endothelial cell dysfunction
   and its pro-adhesive phenotype.

   Adrenomedullin is a potent vasodilator and contributes to
   anti-inflammation and antimicrobial infection in sepsis, although its
   vasodilatory effect could be detrimental ([189]26). RAMP2 acts as
   adrenomedullin-receptor accessory protein, and adrenomedullin-RAMP2
   system is crucial in maintaining vascular integrity, endothelial
   barrier function and organ homeostasis ([190]27). Although there was a
   down-regulation of RAMP2 in the absence of CLEC5A, no effect on
   survival after CLP or pulmonary vascular leakage was observed in
   CLEC5A^−/− mice with overexpression of endothelial RAMP2 ([191]Fig. 9,
   D and I). Therefore, CLEC5A might not function via endothelial RAMP2 in
   the setting of CLP-induced lung injury and barrier dysfunction.

DISCUSSION

   The immune profile of sepsis includes dysregulated immune responses to
   infection, resulting in a systemic inflammatory response ([192]28). In
   the initial phase of sepsis, the innate immunity is activated and
   initiates inflammation. Persistence of the inflammatory storm causes
   tissue damage, organ dysfunction, and subsequent death. Previous
   studies have shown that abdominal infection-induced sepsis in vivo,
   including intraperitoneal injection of LPS-induced endotoxemia and
   CLP-induced polymicrobial sepsis, results in severe peritoneal
   inflammation and multiple organ injury ([193]29, [194]30). The
   hyperinflammatory response is closely correlated with early death
   following CLP or LPS. Abdominal sepsis often results in lung injury and
   ARDS and ultimately death ([195]31). It has been well known that C-type
   lectins function in host defense against pathogens and immune
   homeostasis. CLEC5A, a C-type lectin, is identified as a PRR for
   several types of viruses and bacteria. Studies have shown that
   glycan-lectin interactions activate immune cells and trigger
   inflammatory responses ([196]32). Bacterial LPSs directly interact with
   cells via PRRs, including C-type lectins, and mediate intracellular
   signalosome formation and activation. C-type lectins, as glycan-binding
   receptors, recognize glycan structures on viruses, bacteria, and other
   microorganisms and initiate immune and inflammatory responses ([197]33,
   [198]34). It has been reported that CLEC5A interacts with the LPS of P.
   aeruginosa and mediates pneumonia in mice ([199]14). Blockade of CLEC5A
   or CLEC5A-mediated signaling inhibited pro-inflammatory responses and
   increased survival of mice exposed to P. aeruginosa. Deletion of CLEC5A
   in neutrophils inhibited pro-inflammatory cytokine production, but mice
   lacking CLEC5A had a higher susceptibility to Listeria infection
   ([200]13), which might be associated with failed homeostatic immunity.
   These observations suggest that CLEC5A acts as an inflammatory
   signaling receptor/sensor, and genetic deletion of CLEC5A interferes
   with sensing the danger signals or the signal transduction of LPS or
   CLP responsiveness. In our study, deletion of CLEC5A greatly improved
   survival of mice challenged by LPS or CLP, as well as prevented the
   development of lung inflammation and injury. These benefits might be
   attributed to CLEC5A’s role in modulating dysregulated immune and
   inflammatory responses.

   CLEC5A-deficient mice had increased number of bacterial CFUs in the
   blood but decreased number in the BALF and lungs. CLEC5A is critical
   for neutrophil extracellular trap formation and controlling bacterial
   spreading ([201]13, [202]14). The innate immune response to bacterial
   infection could be impaired in the CLEC5A^−/− mice and thus the failure
   of bacterial clearance, resulting in higher bacterial contents in the
   blood. Decreased bacterial loads in the BALF and lungs might be
   associated with less recruitment and infiltration of
   immune/inflammatory cells to the lungs, whereas this could be
   attributed to reduced production and secretion of inflammatory
   cytokines and chemokines in the lung. The inability of CLEC5A^−/−
   immune cells in controlling the spread of bacteria in the lungs might
   facilitate the spreading to the blood. These observations suggest the
   importance of CLEC5A in host defense against microbial infection,
   whereas the survival benefits of CLEC5A deficiency in sepsis models
   maybe not due to control of bacterial infection and spreading. Further
   exploration of the underlying mechanism and the relevance to mortality
   in sepsis is needed.

   There is currently a lack of evidence to support the immunopathologic
   mechanism of sepsis. A hypothesis proposes that barrier failure
   contributes to lethal organ dysfunction in sepsis ([203]2). Homeostatic
   organ function is dependent on barriers between tissue compartments.
   These barriers, maintained by endothelial and epithelial cells, support
   organ function and are responsible for regulating the movement of
   molecules across tissues. Impaired homeostatic regulation of barrier
   function is evident in clinical and experimental sepsis and associated
   with the development of lung injury ([204]35). Abnormalities in the
   lungs following CLP, including edema formation, vascular leakage,
   inflammatory cell infiltration, and pro-inflammatory cytokine
   overproduction, are likely attributed to a failure of the
   alveolar-capillary barrier, which is responsible for triggering ARDS,
   and its subsequent high mortality. Along with failed pulmonary
   homeostasis and lethal lung injury is an up-regulation of CLEC5A in the
   lungs. Current research focuses mostly on CLEC5A produced by myeloid
   cells and has shown that knockdown of CLEC5A inhibits the production of
   pro-inflammatory cytokines and attenuates lung injury under infectious
   conditions ([205]14, [206]36). Consistently, we showed that mice that
   lack CLEC5A were resistant to lung injury and inflammation, and thus
   mortality in the setting of CLP-induced polymicrobial sepsis. We
   further extend the study of CLEC5A to pneumonia and inflammatory lung
   injury. In the E. coli–induced pneumonia model, there were less lung
   inflammation and vascular leakage in CLEC5A^−/− mice, indicating
   deletion of CLEC5A protected the lung against E. coli infection. These
   observations were similar to that in the CLP model. Here, we asked
   whether and how CLEC5A affects cellular and organ dysfunction in the
   progression of microbial infection to lung injury and mortality.

   The endothelium has been well characterized in the control of local
   infection through releasing inflammatory mediators and recruiting
   leukocytes. During sepsis, the endothelium is activated by exogenous
   pathogen-associated molecular patterns (PAMPs) and endogenous
   damage-associated molecular patterns (DAMPs) via PRRs ([207]37). Its
   activation is beneficial for the local immune response to localized
   infection but detrimental under systemic activation in sepsis
   ([208]38). Dysfunction of the endothelium is responsible for lung
   injury, and targeting the endothelium is a promising approach to limit
   sepsis-associated organ failure and, consequently, death ([209]39).
   Selective delivery of therapeutic molecules can be achieved using
   liposomes that are designed to encapsulate pharmacologically active
   entities. Recent studies have shown endothelial cell specific delivery
   by liposomes modified with antibodies against VCAM-1 ([210]40,
   [211]41). Dysregulation of CLEC5A was found in pulmonary endothelial
   cells during sepsis and inflammation. CLEC5A deficiency prevented the
   triggering of lung injury in response to microbial infection and
   decreased mortality after CLP, and notably, specific deletion of
   endothelial CLEC5A had a similar outcome. The contribution of
   endothelial CLEC5A to poor outcomes in mice with polymicrobial sepsis
   suggests the pathogenic role of CLEC5A in the endothelium and the lungs
   in the defense against septic infection. CLEC5A^−/− transgenic mice
   seemed to have a better survival than endothelial-specific knockdown in
   mice following CLP. The role of CLEC5A, as a PRR, has been well
   described in myeloid cells and involvement in immune and inflammatory
   responses ([212]42). Previous studies have suggested the importance of
   CLEC5A in innate immunity against bacterial infection in vivo ([213]13,
   [214]14). The effect of global knockout (CLEC5A^−/−) in sepsis-induced
   lung injury and mortality might be due to CLEC5A from myeloid cells.

   In both CLP-induced abdominal sepsis and E. coli–induced pneumonia
   models, genetic deletion of CLEC5A might disturb innate immunity in
   response to bacterial infection and thus the ability of bacterial
   clearance and the control of bacterial spreading. In contrast,
   endothelial CLEC5A deficiency in mice had no effect on bacterial loads
   in the blood, BALF, and lung but inhibited the development of
   sepsis-associated lung injury and E. coli pneumonia. The lung
   protection of CLEC5A deletion may be due to inhibition of inflammatory
   storm in the lung and maintenance of endothelial barrier function. Our
   study identifies an immunological mechanism of CLEC5A in the
   endothelium by which CLEC5A mediates local inflammatory responses to
   microbial infection and contributes to the progression of lung injury.
   CLEC5A-deficient endothelium was resistant to sepsis-induced barrier
   dysfunction and thus inhibited immune cell influx into the lungs.
   Endothelial CLEC5A might be a target for therapeutic approaches to
   sepsis and its associated lung injury. Targeting endothelial CLEC5A
   with the use of Tie1 promoter could be a limitation of the study
   because Tie 1 is also expressed by immature hematopoietic cells.
   Endothelial-specific interference of CLEC5A expression/function is an
   issue need to be solved in the further research, as well as the
   specificity of AAV9-mediated deficiency or overexpression approach to
   the lungs. An alternative to selectively target CLEC5A in endothelial
   cells would provide a better understanding of CLEC5A’s role in the
   endothelium. The cellular function and mechanisms of CLEC5A in the
   setting of bacterial infection and the resulting inflammatory lung
   injury, and the possibility of other tissue or cell types that may be
   involved, deserve an in-depth investigation.

   Sepsis-induced damage to endothelial cells leads to impaired structure
   and function of the endothelium. PMVEC permeability impairment
   underlies pulmonary microvascular leakage. This leakage induces
   extravasation of albumin into the lungs, where it accumulates and
   produces edema ([215]10). Consistent with in vivo findings, deletion of
   CLEC5A protected PMVECs against LPS-induced barrier failure by
   maintaining endothelial integrity and hyperselective function. This
   endothelial protection accounted for decreased exudation and vascular
   leakage in CLEC5A-deficent mice following CLP and prevented lung
   injury. Leakage across the alveolar-capillary barrier allows the
   passage of pro-inflammatory cytokines (such as TNF-α and IL-6) that are
   released into the bloodstream from other inflammation sites to the
   pulmonary endothelium. It has been established by Tracey et al. that
   TNF produced by activated myeloid cells, mainly neutrophils and
   macrophages, is crucial in this immunopathology and sufficient to
   mediate lethal tissue injury ([216]43). TNF-α induces capillary leakage
   and abnormal vasodilation, and contributes to vascular changes and the
   progression of sepsis ([217]42). IL-6 released by leukocytes impairs
   intercellular junctions by decreasing VE-cadherin, leading to
   endothelial barrier disruption ([218]44). Loss of CLEC5A maintained
   cohesion and organization of PMVECs and inhibited LPS-induced
   intercellular gaps by increasing the endothelial junction protein
   VE-cadherin. These inflammatory mediators cause damage to endothelial
   cells, such as inducing cell apoptosis ([219]45). Apoptosis of
   endothelial cells leads to barrier failure and interstitial leakage,
   and we found that CD31-positive endothelial cells were decreased in the
   lungs of mice subjected to CLP. These adverse consequences in the
   endothelium were diminished by deletion of CLEC5A, and its contribution
   to maintaining barrier function could be due to the improvement of
   endothelial cell survival. On the other hand, apoptotic cells favor the
   expression of cytokines, chemokines and adhesion molecules and thus
   recruitment of leukocytes, which, in turn, amplifies local immune
   responses and propagates endothelial dysfunction ([220]46). Along this
   line, CLEC5A deficiency prevented against pulmonary inflammation and
   cell infiltration in the setting of CLP-induced endothelial
   dysfunction. These findings suggest that the lung protection observed
   from CLEC5A deficiency may be achieved through preventing CLP-induced
   endothelium failure, thereby improving survival.

   Under severe conditions, defects of the endothelial barrier are
   exacerbated and leaks are enlarged to enable infiltration of blood
   immune cells, including leukocytes, monocytes, and neutrophils, into
   the tissues. Neutrophils account for 50 to 70% of white blood cells in
   humans and act as the first line against infection in the innate immune
   system ([221]47). Excessive accumulation of neutrophils and recruitment
   of leukocytes to the tissues exacerbates local inflammation and
   vascular leakage ([222]8). CLEC5A-deficient PMVECs and HUVECs reduced
   trans-endothelial migration of monocytes and neutrophils across the
   endothelium during inflammation. Mice that lacked CLEC5A consistently
   had less pulmonary infiltration of immune cells and decreased
   inflammation following CLP. Regulation of immune cell recruitment and
   subsequent transmigration across the endothelium is important in the
   control of local inflammation and tissue injury during sepsis
   ([223]48). We demonstrate that CLEC5A is crucial for barrier function
   of pulmonary endothelium in cross-talk with immune cells, and targeting
   CLEC5A is promising as protection against immune cell influx and
   immune/inflammatory dysregulation in sepsis-induced lung injury. Sepsis
   results in a “gain-of-function” in endothelial cells, including
   pro-inflammatory and pro-adhesive properties. Upon engagement of PRRs,
   endothelial intracellular signaling is activated, followed by
   consecutive activation of transcription factors such as nuclear factor
   κB subunit 1 (NFKB1/NF-κB) ([224]49). The transcription of downstream
   genes consequently increases, including pro-inflammatory
   cytokines/chemokines and cell adhesion molecules. Expression of these
   mediators regulates leukocyte trafficking to the surrounding tissues
   and interaction with endothelial cells before transmigration ([225]50).
   Under inflammatory stimulation, PMVECs release MCP-1 and VCAM-1, which
   favor monocyte recruitment and adherence, and produce CXCL5 and ICAM-1
   that contribute to neutrophil influx into the lungs. CLEC5A-deficient
   endothelial cells negatively regulated adhesion of monocytes and
   neutrophils by inhibiting gene transcription and secretion of these
   mediators. Our findings support the notion that CLEC5A mediates
   endothelial activation and intracellular cytokine production, resulting
   in functional alterations in endothelial cells during sepsis. Improved
   understanding of the signaling pathways downstream of CLEC5A-mediated
   endothelial cell alterations would provide insights into therapeutic
   strategies targeting CLEC5A in sepsis.

   Together with these observations, CLEC5A is involved in endothelial
   phenotypic modification and pro-inflammatory responses upon septic
   infection. Sepsis-related endothelial change is a double-edged sword,
   and it is challenging to distinguish between its adaptive response and
   dysfunction. More in vivo and in vitro studies to understand the
   underlying mechanisms by which CLEC5A mediates endothelial activation
   and/or dysfunction in response to infection are still necessary, and
   differences in organ-specific endothelial cell responses need to be
   considered. A promising therapeutic strategy for sepsis-associated
   organ failure is the targeting of the pathways involved in endothelial
   dysfunction without affecting appropriate activation. Limiting vascular
   leakage protects the endothelium from inflammatory stimuli and is
   considered a therapeutic approach to experimental sepsis. However,
   these effects may also impair leukocyte recruitment and affect the
   local immune response, which is harmful in defense against pathogens
   ([226]51). Filewod and Lee and recent studies have reported that
   vascular leakage can be regulated without affecting immune responses
   and pathogen clearance ([227]52). In-depth exploration is needed to
   distinguish the pathways by which CLEC5A regulates leukocyte
   trafficking and vascular leakage/permeability.

   In the present study, we demonstrated that knockdown of endothelial
   CLEC5A increased survival of mice with polymicrobial sepsis or
   endotoxemia and alleviated lung injury and inflammation caused by
   bacterial infection. Deletion of CLEC5A protected against pulmonary
   endothelial cell dysfunction and barrier failure during inflammation,
   and inhibited leukocyte-endothelial adhesion and transmigration, thus
   maintaining pulmonary homeostasis. We suggest endothelial CLEC5A is a
   key protein responding to sepsis-causing bacterial infection, via
   mediating barrier function and endothelial-immune cell cross-talk. This
   work extends the potential of targeting CLEC5A in the treatment of
   inflammatory lung injury due to an infection such as bacterial
   pneumonia and sepsis and opens the field for mechanistic and
   therapeutic implications of CLEC5A in the progression of sepsis in
   humans exposed to a variety of infectious conditions.

MATERIALS AND METHODS

Ethics approval

   All animal experiments were approved by the Institutional Ethics of
   Animal Care and Use Committee at Shengjing Hospital of China Medical
   University (2021PS637K).

Animals

   C57BL/6 WT mice were purchased from Charles River (Wilmington,
   Massachusetts, USA). By using the online random number generator
   ([228]https://iikx.com/tool/radom.html), mice were randomly assigned
   into groups with no blinding based on study design. A priori power
   analysis was performed to determine the group size using an online
   calculator PowerAndSampleSize ([229]http://powerandsamplesize.com/),
   based on a two-sample and two-sided equality to compare two means with
   a power (1-β, Type II error) of 0.8 and an α (Type I error) of 0.05.
   Equal numbers of male and female mice were included in each animal
   experiment. All mice were used and euthanized at the age of 8 to 12
   weeks. All animal experiments were approved by the Institutional Ethics
   of Animal Care and Use Committee at Shengjing Hospital of China Medical
   University (2021PS637K).

   The CLEC5A gene (NCBI Gene ID: 23845) is located on mouse chromosome 6
   (Chr: 40551832-40562739) with a full length of 10,908 base pairs (bp).
   CLEC5A genetic knockout mice were generated on a C57BL/6 background and
   obtained from Wanlei Life Sciences Co. Ltd. (Shenyang, China). Briefly,
   two single guide RNAs (sgRNAs) were designed for targeting the region
   from exon 2 to exon 5 of CLEC5A (NCBI Reference Sequence:
   [230]NM_021364, knockout region size: 8501 bp). The target sequences
   are 5′-TGGTTCCGTAGCTCTCCGTG-3′ (#1) and 5′-AATGAGCCTTTGATCTACGG-3′
   (#2). A mixture of CRISPR-Cas9 recombinant protein and sgRNAs was
   coinjected into zygotes for generating CLEC5A^−/− mice. WT littermates
   were used as the control.

Model of experimental sepsis

   Mice of a C57BL/6 background, 8 to 10 weeks old, were used for the
   modeling. Two mouse models of sepsis were established based on modified
   previous methods. For induction of polymicrobial sepsis, mice were
   subjected to CLP with two punctures using a 25-gauge needle as
   previously described ([231]53, [232]54). Sham mice underwent the same
   operation except for cecal ligation or puncture. For endotoxemia, mice
   received a single intraperitoneal injection of LPS (10 mg/kg; E. coli
   055:B5; L2880#, Sigma-Aldrich, St. Louis, MO, USA). In vehicle group,
   mice received same amount of saline. For survival analysis, N = 10 mice
   were included in each group and the survival was recorded every 12
   hours for 1 week. N = 6 mice per group were applied in all other
   experiments.

Lung histological assessment and immunofluorescence staining

   The lungs were isolated, photographed, and weighted (wet and dry).
   Paraffin-embedded sections with a thickness of 5 μm were prepared and
   used for H&E staining. Severity of lung injury was blindly assessed
   based on inflammation, edema, hemorrhage, and alveolar septal
   thickening using a 0-4 point scoring system, individually ([233]55).
   Lung injury score was determined as the sum of each score from above
   four criteria. N = 6 mice per group were for histological analysis,
   including three randomly selected slides per mouse with three random
   fields for each slide. Immunofluorescence for CLEC5A (AF1639#, R&D
   system, Minneapolis, MN, USA) and CD31 (ab222783#, Abcam, Cambridge,
   UK) was performed according to a standard protocol with the use of
   4′,6-diamidino-2-phenylindole (DAPI) for nuclear counterstaining. For
   studying the expression pattern of CLEC5A in specific endothelial cell
   subtypes, we conducted double immunofluorescent staining for CLEC5A and
   GJA5 (680144#, Zenbio, Chengdu, China), VWF (AF3000#, Affinity
   Biosciences, China), EDNRB (DF7104#, Affinity Biosciences), PTPRB
   (AP58204#, Abcepta, Jiangsu, China), or PROX1 (bs-2774R#, Bioss,
   Beijing, China) in lung tissues.

BALF collection

   The left bronchus was ligated, and BALF was collected by flushing the
   lung with 0.4 ml of phosphate-buffered saline (PBS) (B548117#, Sangon
   Biotech, Shanghai, China) via the tracheal cannula, three times per
   mouse. Inflammatory cells in the BALF were collected, and total
   leukocyte counts were determined using a hemocytometer. Differential
   cell counts for neutrophils and monocytes were performed with
   Wright-Giemsa stain (D010#, Nanjing Jiancheng Bioengineering Institute,
   Nanjing, China) following morphological criteria.

Microvascular albumin leak

   Pulmonary microvascular permeability was evaluated by intravenous
   injection of EB dye as previously described ([234]56). Briefly, mice
   received EB dye (50 μg/g) via tail vein 30 min prior to euthanasia.
   PBS-perfused lung was collected and the EB dye was extracted by
   formamide incubation. The EB content was measured, and the leakage of
   EB-labeled albumin was calculated as mg EB/g lung per minute.

Model of pneumonia

   C57BL/6 mice were challenged by 1.5 × 10^7 CFUs of E. coli (American
   Type Culture Collection, strain 25922) in 50 μl of PBS through
   intratracheal instillation. The control mice received the same amount
   of PBS. Mice were euthanized 16 hours after bacterial infection. The
   blood, BALF, and lungs were collected, and N = 6 mice per group were
   included in each experiments.

Determination of CFUs

   The bacterial loads were assessed in diluted samples of the peripheral
   blood, BALF, and lungs. In the CLP model, the samples were collected 24
   hours after surgery. In the E. coli–induced pneumonia model, the
   samples were collected 16 hours after bacterial inoculation. For
   visceral CFU determination, the lungs were weighted and homogenized
   (0.1 g/ml in sterile PBS). After serial dilutions in sterile PBS (1:10,
   1:100, or 1:1000), 100 μl of the diluent was plated on tryptoic soy
   agar plates (T8650#, Solarbio). Bacterial CFUs were counted after
   incubation at 37°C for 24 hours. The results for the blood and BALF
   bacterial counts are presented as CFUs/ml, and for the lung is CFUs per
   gram tissue.

Single-cell RNA sequencing

   Lung tissues were obtained from CLEC5A^−/− and WT mice subjected to
   CLP. Single-cell preparation, scRNA-seq, and bioinformatics analysis
   were conducted by LC-Bio Technology (Hangzhou, China)
   ([235]http://lc-bio.com/tech-service-detail?cid=18&id=48). Briefly, the
   single-cell suspension was prepared from lung tissues according to the
   standard protocol. Cell viability (above 85%) was determined by Trypan
   blue exclusion, and samples were prepared at the concentration of
   0.7 × 10^6~1.2 × 10^6 cells/ml. Single-cell suspensions were processed
   using Chromium Single Cell Solution Kit (10×Genomics) according to the
   manufacturer’s instructions. It was followed by cDNA amplification, and
   library construction and running on the NovaSeq 6000 system for
   Illumina sequencing with 2 × 10^4~4 × 10^4 mean reads per cell. The
   sequencing data were analyzed using the Cell Ranger software (V7.0.0)
   ([236]https://support.10xgenomics.com/single-cell-gene-expression/softw
   are/overview/welcome) and aligned to Ensembl Genome GRCm39
   ([237]ftp://ftp.ensembl.org/pub/release-105/fasta/mus_musculus/dna).
   The output data were further analyzed by the Seurat software (V4.1.0)
   ([238]https://satijalab.org/seurat/). Briefly, a total of 64,581 single
   cells were captured from six mouse lungs following 10×Genomics Chromium
   Single Cell processing. The threshold for data quality control was set
   as follows: genes expressed per cell > 500 (gene number), mitochondrial
   genome content rate < 25%, and removal of multiplet cells by
   DoubleFinder. Overall, 55,177 single cells were obtained after
   filtration. For clustering and visualization, the data were normalized
   using the LogNormalize method by Seurat, followed by PCA. The top 20
   PCs were used for subsequent dimensional reduction and clustering
   analysis. A UMAP analysis was carried out to visualize the identified
   cell clusters. The top marker genes that highly expressed in each
   cluster were identified using the FindAllMarkers function of Seurat.
   Within a default process, the marker gene was selected according to the
   following thresholds: genes expressed in >10% cells in a cluster, P
   value ≤ 0.01, and log[2]FC ≥ 0.26.

Viral vector and gene transduction

   The loss-of-function study for CLEC5A was carried out by RNAi.
   Gene-specific short hairpin RNA (shRNA) was synthesized by General
   Biosystem (Anhui, China), and the target mRNA sequences are
   5′-CTTCGACTGTGTCACTATAGG-3′ (Mus musculus CLEC5A, [239]NM_021364) and
   5′-GAATCATCTTGGAATGAAAGC-3′ (Homo sapiens CLEC5A, [240]NM_001301167).
   AAV serotype 9 carrying the endothelial-specific promoter Tie1
   (AAV9-Tie1) and lentivirus (LV)–mediated gene transfer was used for in
   vivo and in vitro studies of CLEC5A, respectively. CLEC5A gene-specific
   shRNA or coding sequence (General Biosystem) was ligated into
   pAAV-CMV-U6 (Wanleibio, Shengyang, China), pLVX-shRNA1, or
   pLVX-IRES-puro (BR004# and BR025#, Fenghui Biotechnology, Hunan,
   China). For in vitro experiments, endothelial cells were infected with
   lentivirus expressing CLEC5A (CLEC5A^oe) or shRNA against CLEC5A
   (CLEC5A^sh). For in vivo experiments, AAV9 expressing CLEC5A or
   gene-specific shRNA under the promoter of Tie1 (CLEC5A^Tie1-oe or
   CLEC5A^Tie1-sh) was delivered into mice through tail vein injection
   ([241]57). Virus carrying empty vector or expressing shRNA that targets
   a nonspecific sequence was used as the negative control. For in vivo
   studies of CLEC5A downstream targets, CLEC5A^−/− mice were injected
   with AAV9-Tie1 expressing specific shRNA targeting SCARB1 or PODXL, or
   coding sequence of RAMP2 or FABP4 under Tie1 via tail vein. Fourteen
   days after injection of 100 μl of virus containing 1 × 10^10 AAV9
   vector genomes, mice received CLP surgery as described above. The
   efficiency of AAV/LV-mediated gene transfer in vivo and in vitro was
   validated by real-time polymerase chain reaction (PCR).

Cells and treatment

   Mouse PMVECs were isolated as previously described ([242]58). Cells
   were identified as positive immunostaining for CD31 and cultured in
   primary endothelial cell culture system (PriMed-iCell-002#, iCell,
   Shanghai, China). Cells between passages 3 and 6 were used for the
   experiments. Mouse alveolar type II epithelial cells were isolated by
   magnetic sorting ([243]59) and identified by staining for SFTPC
   (A23181#, ABclonal, Shanghai, China). Mouse peripheral blood monocytes
   (PBMCs) were isolated and purified using the commercial kit (P5230#,
   Solarbio, Beijing, China). Mouse peripheral blood granulocytes were
   isolated according to a previously described method ([244]60). HUVECs,
   HMVECs, and THP-1 cells were purchased from iCell Bioscience Inc.
   (Shanghai, China). HUVECs were cultured in HUVEC culture medium
   (iCell-h110-001b#, iCell). THP-1 cells were cultured in RPMI 1640
   medium (31800#, Solarbio) supplemented with 10% fetal bovine serum and
   0.05 mM β-mercaptoethanol. For inflammatory stimulation, mouse PMVECs
   or HUVECs were exposed to LPS (10 μg/ml) and the duration depended on
   experimental design. All in vitro experiments were conducted with N = 4
   biological repetitions per group.

Endothelial cell permeability assay

   Permeability assays for mouse PMVECs monolayers were carried out
   following a previous report ([245]61). Cells were seeded onto transwell
   inserts with 0.4-μm pore size (14312#, LABSELECT, Hefei, Anhui, China)
   in 24-well plates at a density of 4 × 10^4 per well. The cells, after
   monolayers were formed, were treated with LPS (10 μg/ml) for 24 hours,
   followed by incubation with fluorescein isothiocyanate (FITC)–labeled
   dextran (1:50) for 45 min. The permeability of monolayers was assessed
   by detecting the fluorescence of the lower chamber medium ([246]62).

TEER measurement

   The barrier integrity and function of mouse PMVECs and HUVECs was
   assessed in vitro by measuring TEER (Millicell ERS-2#, Millipore,
   Darmstadt, Germany). Cells were seeded onto 0.1% gelatin-coated
   transwell inserts (14312#, LABSELECT) in 24-well plates and grown to
   confluence. After the basal TEER was stabilized, cells were treated
   with LPS (10 μg/ml) for 4 hours, and the resistance was measured and
   normalized to the TEER value across an empty insert.

Monocyte-endothelial adhesion assay

   The adhesion of mouse peripheral blood monocytes to PMVECs and human
   monocytes THP-1 to HUVECs were assessed. Endothelial cells were seeded
   into 96-well plates and treated with LPS (10 μg/ml) for 5 hours. After
   removal of LPS, the cells were cocultured with BCECF-AM–labeled
   monocytes for 1 hour. The adherent cells were harvested and lysed in
   0.2% Triton X-100 buffer. The fluorescence intensity was measured, and
   the data are expressed as the percentage relative to the control group.

Trans-endothelial migration assay

   Trans-endothelial migration of monocytes was performed as previously
   described ([247]61). Mouse PMVECs were seeded onto collagen I (10
   μg/cm^2; BS929-1g#, Biosharp, Hefei, China)–coated transwell inserts
   (3-μm pore size; 14321#, LABSELECT) in 24-well plates. Cells were
   cultured until monolayer formation, and then exposed to LPS (10 μg/ml)
   for 5 hours. After coculturing with mouse peripheral blood monocytes or
   granulocytes (2 × 10^5 per well) for 4 hours, cells in the lower
   chamber were collected and cell morphology was examined under a phase
   contrast microscopy (IX53#, Olympus, Tokyo, Japan). Granulocytes across
   the endothelial monolayer were identified as 100% neutrophils. The
   number of transmigrated cells was determined using
   3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT)
   assay. THP-1 trans-endothelial migration by coculturing with HUVECs was
   assessed as described above.

Western blotting

   Western blotting was performed according to a standard protocol.
   Briefly, total protein was extracted with the use of
   radioimmunoprecipitation assay buffer (R0010#, Solarbio), and an equal
   amount of protein was loaded and separated by SDS–polyacrylamide gel
   electrophoresis and transferred onto a polyvinylidene difluoride
   membrane (IPVH00010#, Millipore). The blot was probed with anti-CLEC5A
   (PA5-34928#, Thermo Fisher Scientific, Pittsburgh, PA, USA), anti-CD31
   (ab222783#, Abcam), anti-VE-cadherin (ab205336#, Abcam), anti-SCARB1
   (A1584#, ABclonal Technology, Wuhan, China), PODXL (A10200#), RAMP2
   (A3075#), or FABP4 (A11481#) at a dilution of 1:500 or 1:1000, followed
   by anti-rabbit IgG, horseradish peroxidase–conjugated secondary
   antibody (SE134#, Solarbio). The target bands were developed using an
   enhanced chemiluminescent reagent (PE0010#, Solarbio), and analyzed by
   Gel-Pro-Analyzer software. The immunoblot was stripped and reprobed
   with glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (60004-1-Ig#,
   Proteintech Group Inc., Rosemont, IL, USA), which serves as an internal
   reference.

Quantitative real-time PCR

   Total RNA was extracted using TRIpure kit (RP1001#, BioTeke, Beijing,
   China) according to the manufacturer’s instructions. RNA concentration
   was determined, and reverse transcription was performed using the
   BeyoRT II M-MLV reverse transcriptase (D7160L#, Beyotime, Shanghai,
   China). The quantitative PCR was carried out with 2×Taq PCR MasterMix
   (PC1150#, Solarbio) and SYBR Green (SY1020#, Solarbio) in Exicycler 96
   (Bioneer, Daejeon, Korea). Sequences of specific PCR primers were
   included in table S1. The relative quantification was calculated using
   the 2^−ΔΔCT method, and the data are displayed relative to the control
   group that arbitrarily set as 100%.

Enzyme-linked immunosorbent assay

   Mouse PMVECs were exposed to LPS (10 μg/ml) for 24 hours, and the
   supernatant was collected for cytokine detection. Levels of TNF-α,
   IL-6, MCP-1, ICAM-1, and VCAM-1 were determined by commercial
   enzyme-linked immunosorbent assay (ELISA) kits from Lianke Technology
   (EK282#, EK206#, EK287#, EK289#, and EK290#, Hangzhou, China) and CXCL5
   by FineTest (EM1004#, Wuhan, China).

Statistical analysis

   All quantification in this study was carried out in a blinded manner.
   No sample or data were excluded from analysis. All figures shown are
   representative images that reflected the average level of at least
   three independent experiments. All data are analyzed using GraphPad
   Prism 8.0 software, and the graphs are shown as mean (SD) values of six
   biological repetitions for in vivo experiments and four biological
   repetitions for in vitro experiments. Data that meet the normal
   distribution by the Shapiro-Wilk test were subjected to variance
   analysis. For comparisons among three or more groups, the homogeneity
   of variance was confirmed by the Brown-Forsythe test or Bartlett’s
   test, and one-way analysis of variance (ANOVA) analysis or
   Brown-Forsythe and Welch ANOVA test was performed. Comparisons between
   groups were carried out by Tukey’s multiple comparisons test or
   Games-Howell’s multiple comparisons test. Lung injury score was
   analyzed by the Kruskal-Wallis test followed by Dunn’s multiple
   comparisons test. An unpaired two-tailed t test was used for
   comparisons between two groups with F test to compare variances or
   Mann-Whitney test. Data of CLEC5A expression at different time points
   following LPS or CLP challenge were performed with repeated measures
   ANOVA analysis with the Geisser-Greenhouse correction, followed by
   Dunnett’s or Sidak’s multiple comparisons test. For survival analysis,
   the log-rank (Mantel-Cox) test, log-rank for trend, or
   Gehan-Breslow-Wilcoxon test was used for comparisons of survival curves
   between groups. A P value less than 0.05 was considered statistically
   significant.

Acknowledgments