Reduced plasma levels of Copine 5 correlate with sepsis-induced vascular leakage and mortality in human patients and a murine sepsis model

Abstract

Background

Vascular leakage is a major cause of multiple organ failure and mortality in sepsis, and factors that regulate endothelial integrity could serve as promising biomarkers of septic shock development. Copine family members (CPNEs) are well-characterized as soluble membrane-binding proteins, whether CPNEs play a critical role in maintaining vascular integrity during sepsis, however, remains unclear.

Methods

Human aorta single-nucleus RNA-sequencing data were analyzed for the expression profile of all Copine family members (CPNE1-9). Plasma levels of CPNE5, Ang-II, sICAM-1, and SDC-1 were measured in human sepsis patients at admission and healthy donors as well as in septic mice induced by injection i.p. with cecal slurry. The correlation of CPNE5 levels to other three factors (Ang-II, sICAM-1, SDC-1) were analyzed. CPNE5-knockdown endothelial cells (ECs) and global CPNE5-knockout (KO) mice were utilized to determine the critical role of CPNE5 in sepsis-triggered vascular leakage, organ damage and mortality.

Results

Among nine CPNEs, only CPNE5 is predominantly expressed in human aorta endothelial cells. In sepsis patients (n = 77), plasma levels of CPNE5 were significantly reduced, whereas plasma levels of Ang-II, sICAM-1, and SDC-1 were markedly elevated, compared to healthy donors (n = 44; p < 0.01). Similar findings were also observed in a murine sepsis model induced by cecal slurry (CS)-injection intraperitoneally. Furthermore, in the supernatants of cultured ECs treated with LPS or pro-inflammatory cytokine mixture (Cytomix: TNFα/IL-1β/IFNγ, each 10 ng/mL), the concentrations of CPNE5 were significantly lower, which was negatively correlated with the higher EC permeability, compared to the control group. Accordingly, siRNA-mediated knockdown of CPNE5 in ECs caused hyperpermeability upon stimulation with LPS or Cytomix. In vivo, we observed that loss of CPNE5 increased vascular leakage, leading to severe organ injury and higher mortality, compared to WT mice upon septic conditions. The initial mechanistic analysis showed that the reduction of CPNE5 in cardiac and pulmonary ECs was linked to the increased cleavage of membrane tight junctions and adherens junctions.

Conclusions

These observations from human sepsis patients and a murine sepsis model suggest that reduced plasma levels of CPNE5 may contribute to sepsis-induced vascular leakage and mortality.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13054-025-05711-1.

Introduction

Sepsis is characterized as life-threatening organ dysfunction caused by a dysregulated host response to infection with concurrent hyper-inflammation and immunosuppression [1]. Although the clinical managements with early adequate antimicrobial medication to control infection and the immediate hemodynamic supportive treatment to prevent organ injury are effective, sepsis remains a leading cause of death in intensive care units (ICU) with mortality rate approaching 30–50% [1]. Such an unacceptable high mortality could be ascribed to the following facts: (1) sepsis can be presented differently with the heterogeneous nature of the host response, making it challenging to recognize and diagnose early [2–4]; (2) both population ageing and prevalence of chronic disease are increased [5]; and (3) the clinical application of immunomodulatory agents to septic patients have failed to improve the outcome [6]. Hence, these sobering facts reinforce the need to understand the molecular/cellular mechanism that contributes to the pathogenesis of sepsis.

Over the past decades, it has become evident that loss of endothelial barrier integrity and increased vascular leakage are fundamental pathophysiological events that occurs early in sepsis pertaining to organ failure and death [7–9]. Accordingly, specific indicators of endothelial cell injury can facilitate early prediction of the progress and prognosis of sepsis and thereby, improving survival outcomes. At present, dozens of molecules/factors related to vascular leakage have been identified as potential diagnostic and prognostic biomarkers for evaluating sepsis outcomes [10–18]. For example, syndecan-1 (SDC-1), a transmembrane protein, is critical for maintaining the integrity of the endothelial glycocalyx, which degradation contributes to increased vascular permeability [10]. In this context, several studies have shown that high plasma levels of SDC-1 are associated with the severity of sepsis and mortality [11–13]. The Angiopoietin proteins (i.e., Ang-I and Ang-II) are well-characterized endothelium-derived factors that competitively interact with Tie-2 kinase receptor to regulate vascular integrity and permeability [14]. In this regard, multiple clinical reports have shown that higher circulating Ang-II levels and lower Ang-I levels in adult patients with sepsis admitted to the ICU are predictive of sepsis severity including organ dysfunction and death [15–17]. Similarly, intercellular adhesion molecule-1 (ICAM-1) is well-recognized to modulate endothelial permeability and herein, higher circulating levels of soluble ICAM-1 (sICAM-1), particularly during the early stage of sepsis, has been closely associated with endothelial damage and increased mortality in sepsis patients [11, 13, 18]. Nevertheless, while these vascular-associated molecules/factors have been assessed for their diagnostic potential in sepsis, none of these has shown satisfactory performance to accurately predict the outcome of a patient with sepsis due to the complex and diverse nature of the host response [1–4]. Thus, there is an urgent need to discover more new reliable biomarkers for improving prognostic sensitivity and specificity of sepsis-related vascular leakage and mortality among heterogeneous patients.

Recently, Copine family proteins (CPNEs) have garnered attention in human diseases because of their unique architecture and a multitude of biological functions [19, 20]. To date, nine members of the CPNE family (CPNE1-9) have been identified in the human genome with distinct patterns of tissue and cell distribution [20]. Notably, all CPNE proteins share a similar structure, featuring a highly conserved two tandem C2 domains at the N-terminus and one C-terminal von Willebrand factor (vWF)-A domain [20]. The C2 domain acts as phospholipid-binding motif and thereby regulate membrane dynamics and vesicle trafficking [20]. The vWF-A (vWA) domain is known to interact with multiple plasma and membrane proteins and mediate cell adhesion [20]. Therefore, considering such a strong ability to associate with membranes, these CPNEs should be fundamentally important for maintaining cell homeostasis. Indeed, accumulating evidence has indicated that CPNEs play critical roles in the regulation of metabolic disorders, nervous system diseases, tumorigenesis and progression [19–24]. For example, the elevated levels of CPNE1, 3, or 8 are linked to the prognosis of poor survival in patients with various types of cancer [19, 21]. Low expression of CPNE3 is associated with the risk of acute myocardial infarction in patients with stable coronary artery disease [22] and CPNE3 is also reported to be essential for maintaining normal β-cell function [23]. Genetic variants of CPNE5 are associated with abnormalities in fatty acid and lipid metabolism [24]. What’s more, CPNE4-8 are highly expressed in central nervous system and involved in neuron synaptic function and plasticity [20]. However, knowledge of CPNE’s function in the pathogenesis of sepsis is currently very limited, and whether CPNEs affect endothelial barrier integrity during sepsis remains unknown.

In this study, we first analyzed single-nucleus RNA-sequencing data in human aorta and found that CPNE5 is the only one among all CPNE members predominantly expressed in aortic endothelial cells. We then collected plasma samples from human patients with sepsis and healthy donors to measure CPNE5 levels together with Ang-II, sICAM-1 and SDC-1, three well-characterized prognostic biomarkers for vascular leakage [7–9]. Furthermore, both CPNE5-knockdown endothelial cells and global CPNE5-knockout mice were utilized to validate whether reduced levels of CPNE5, observed in human septic patients, were able to correlate with vascular hyperpermeability and thereby, serving as prognostic indicator for sepsis severity and mortality.

Results

Single-nucleus expression profiling of Copine family genes in human aorta

Endothelial cells (ECs) line the inner surface of blood vessels and thereby, being among the first responders to pathogens and systemic inflammatory mediators during sepsis [7]. Given that nine Copine members are featured with distinct patterns of tissue and cell distribution [20], we then questioned which member is highly expressed in vascular ECs. To this end, we analyzed the expression pattern of CPNE1-9 in single-nucleus RNA-sequencing database generated from adult human ascending aorta that comprises 14 clusters for 11 major cell-types [25] (Fig. 1A). As shown in Fig. 1B-J, CPNE1, 3, 4, and 8 are exhibited highly expression; CPNE2 and 9 are moderately expressed, whereas CPNE6 and 7 are displayed low expression in all clusters of human aortic cells. Interestingly, CPNE5 presents varied cell distribution with a predominant expression in all three EC clusters (Endothelial I, II, and Lymphatic endothelial cells, red circles in Fig. 1F). Furthermore, similar expression patterns of CPNE1-9 in human aortic cells were also observed when analyzing another single-nucleus RNA-sequencing data generated using paired samples from ascending and descending aorta [26], in which CPNE5 was the only member of CPNE family exclusively enriched in subpopulations of ECs (Fig. 1K and Supplemental Fig. S1A-J). Such a unique cellular distribution of CPNE5 in human aorta suggests its specific role in the regulation of vascular EC permeability during sepsis. Therefore, we selected CPNE5 for the following studies.

Fig. 1.

The distinct expression pattern of Copine family genes in human aortic cell clusters measured by single-cell RNA-sequencing. (A) different cell clusters in human aneurysmal aorta. (B-J) Human aneurysmal aortic cell clusters of CPNE1-9 expression. (K) Cell clusters of CPNE5 expression in human thoracic aorta show that CPNE5 is highly enriched in endothelial cell clusters

Circulating CPNE5 levels are reduced in patients with sepsis and reflect disease severity

To evaluate the clinical relevance of CPNE5 in sepsis, we enrolled a total of 121 participants, including 77 adult patients diagnosed with sepsis and 44 age- and sex-matched healthy controls. The information on patient cohort enrollment regarding inclusion/exclusion criteria and timing of sample collection is detailed in the Supplemental Fig. S2. The overall baseline demographic and clinical characteristics of sepsis patients are presented in Supplemental Table S1 and S2, which Table S2 also summarizes the characteristics stratified by 28-day mortality status. There were no significant differences in age, sex, ICU length of stay, diastolic blood pressure, procalcitonin (PCT), and white blood cell (WBC) counts between survivors and non-survivors. However, non-survivors exhibited significantly lower systolic blood pressure (p = 0.027), higher SOFA scores (p < 0.001), and elevated serum lactate levels (p < 0.001), indicating greater disease severity of sepsis.

Next, plasma levels of CPNE5 were measured upon ICU admission using an ELISA kit and revealed significantly lower in patients with sepsis than healthy controls (p < 0.0001, Fig. 2A). Receiver operating characteristic (ROC) analysis demonstrated that plasma levels of CPNE5 were effectively distinguished sepsis patients from healthy individuals with an area under the curve (AUC) of 0.738 (95% CI: 0.646–0.831) and an optimal cutoff concentration of 6.46 ng/mL, yielding a sensitivity of 97.4% and a specificity of 59% (Fig. 2B). Notably, plasma CPNE5 concentrations were significantly reduced in non-survivors compared to survivors (p = 0.0452), albeit both were markedly lower than healthy donors (p < 0.0001, Fig. 2D). Accordingly, CPNE5 levels were inversely correlated with SOFA scores (r = −0.3095, p = 0.0062, Fig. 2C), supporting its potential association with disease severity. To further explore whether plasma levels of CPNE5 reflect endothelial injury, we measured three established vascular leakage markers including angiopoietin II (Ang-II), soluble intercellular adhesion molecule-1 (sICAM-1), and syndecan-1 (SDC-1) [7–9]. All three markers were significantly elevated in the plasma of sepsis patients, particularly among non-survivors (Fig. 2E-G) which exhibited significant inverse correlations with plasma CPNE5 levels (Ang-II: r = −0.2836, p = 0.0124, Fig. 2H; sICAM-1: r = −0.3243, p = 0.0040, Fig. 2I; SDC-1: r = −0.3319, p = 0.0032, Fig. 2J). Lastly, we performed ROC curve analysis to compare the prognostic value of CPNE5 with classical (i.e., lactate, procalcitonin, and SOFA score). We found that CPNE5 demonstrated predictive value for 28-day mortality with an AUC of 0.625 (95% CI: 0.498–0.751) and an optimal cutoff concentration of 2.36 ng/mL, yielding a sensitivity of 64.6% and a specificity of 68.9%. Please note that the predictive value of plasma CPNE5 levels is not superior to the predictive value of SOFA score (AUC: 0.743; 95% CI: 0.629–0.857) but reveals similar as predictive values of lactate (Lac, AUC: 0.659; 95% CI: 0.529–0.789) and procalcitonin (PCT, AUC: 0.515; 95% CI: 0.385–0.645) (Fig. 2K). In addition, when combining the SOFA score with CPNE5 for 28-day mortality prediction, the improvement in AUC appeared negligible (SOFA alone: 0.743 vs. SOFA + CPNE5: 0.745), but the sensitivity remarkably increased from 58.6% to 72.4% (Supplemental Table S3). This suggests that CPNE5 may provide complementary prognostic value to the SOFA score by enhancing its sensitivity, thereby reducing missed high-risk cases. Therefore, plasma CPNE5 levels may serve as one of potential indicators for severity of sepsis and clinical outcomes.

Fig. 2.

Plasma CPNE5 levels are reduced in septic patients and inversely correlate with vascular injury biomarkers. (A) Plasma CPNE5 levels in healthy controls (n = 44) and septic patients (n = 77). (B) ROC curve analysis of CPNE5 for discriminating sepsis from healthy donors. (C) Correlation between plasma CPNE5 levels and SOFA score. (D) Plasma CPNE5 levels in survivors (n = 48), non-survivors (n = 29), and healthy controls. (E-G) Plasma levels of Ang-II, sICAM-1, and SDC-1 in survivors and non-survivors. (H-J) Correlations of plasma CPNE5 levels with Ang-II, sICAM-1, and SDC-1. (K) AUROC predicting 28-day mortality. CPNE5 (AUC 0.625), SOFA score (AUC 0.743), procalcitonin (PCT) (AUC 0.515), and lactate (Lac) (AUC 0.659). Data are presented as mean ± SD and were analyzed with Student’s t-test (A, D, E, F and G) or Spearman’s correlation coefficient test (C, H, I and J). AUROC analysis (K) was conducted in R (version 4.4.2) using the pROC package (version 1.19.0.1), detailed in the Supplementary material online, Methods

Plasma levels of CPNE5 negatively correlate with vascular injury in septic mice

Building on our clinical findings above, we next asked whether the alterations of plasma CPNE5 observed in sepsis patients could be recapitulated in an experimental animal sepsis model. To this end, we employed a murine model of polymicrobial sepsis induced by injection (i.p.) of cecal slurry (CS) in wild-type C57BL/6 mice (Fig. 3A). As expected, CS-injected mice developed progressive sepsis symptoms (Fig. 3A), assessed by the murine sepsis scoring systems [27] (Supplemental Table S4), confirming successful establishment of CS-induced polymicrobial sepsis model. Subsequently, we measured plasma levels of CPNE5 in these CS-mice and controls and interestingly, observed that plasma CPNE5 levels exhibited a biphasic pattern, with a transient increase at 24 h post-CS followed by a significant decline at 48 h post-CS (p < 0.05, Fig. 3B). This dynamic profile suggests that CPNE5 may be involved in the early compensatory protective response and late injurious phases of sepsis progression.

Fig. 3.

Dynamic changes in plasma levels of CPNE5 and vascular injury markers in CS-septic mice. (A) Schematic illustration of the experimental protocol for CS-induced polymicrobial sepsis in wild-type (WT) mice and sepsis clinical scores. Mice with the same volume of PBS injections were used as sham group. (B) plasma CPNE5 levels in CS-septic mice. (C-E) plasma levels of vascular leakage markers (Ang-II, sICAM-1, and SDC-1) in mice following CS injection. (F-H) Correlations between plasma CPNE5 levels and each vascular leakage marker at 48 h post-CS. Data are presented as mean ± SD and were analyzed with one-way ANOVA (B, C, D and E, vs. 0 h) or Spearman’s correlation coefficient test (F, G and H). *, p < 0.05 vs. 0 h; n = 18

In parallel, we assessed plasma levels of typical endothelial injury biomarkers including Ang-II, sICAM-1, and SDC-1, which exhibited progressively elevation following CS-injection (Fig. 3C-E). Next, we utilized plasma levels at the time point of 48 h post-CS to conduct correlation analysis and found that plasma CPNE5 levels had strong inverse correlations with each of the vascular leakage markers: Ang-II (r = −0.5748, p = 0.0126, Fig. 3F), sICAM-1 (r = −0.5315, p = 0.0232, Fig. 3G), and SDC-1 (r = −0.5369, p = 0.0216, Fig. 3H). Collectively, these data generated from CS-induced septic mice are closely mirrored those findings in human patients with sepsis (Fig. 2H-J), attesting a conserved association between plasma CPNE5 and endothelial integrity across species.

The expression of CPNE5 is dynamically altered in endothelial cells during sepsis

Given the inverse correlation of circulating CPNE5 levels with vascular injury markers during sepsis, we next determined the alterations of CPNE5 expression in endothelial cells upon systemic inflammatory challenge in vivo and in vitro. First, aorta and hearts were collected at different time-points from mice following CS-injection to isolate total RNA for measuring CPNE5 expression. Our RT-PCR results showed that the expression of CPNE5 was significantly elevated in both aorta and hearts at 12 h post-CS, but was markedly declined at 24 h and 48 h post-CS (Fig. 4A/B, p < 0.05), suggesting its transient compensatory protective upregulation followed by an injurious downregulation in response to systemic inflammation. Considering that there are more endothelial cells (ECs) in mouse hearts than aorta, we next isolated ECs, fibroblasts, and cardiomyocytes from the heart of mice at 12 h and 24 h post-CS to determine CPNE5 expression levels. As shown in Fig. 4C, the expression levels of CPNE5 were remarkably increased in cardiac ECs at 12 h post-CS, whereas being significantly decreased at 24 h post-CS. A similar but less pronounced pattern of CPNE5 alterations was observed in cardiomyocytes at 12 h and 24 h post-CS (Fig. 4C). However, the expression of CPNE5 was unchanged in cardiac fibroblasts following CS-injection (Fig. 4C), possibly due to the relatively lower baseline expression of CPNE5 in cardiac fibroblasts [28], as shown in Supplemental Figure S3. Nonetheless, these findings suggest that endothelial cells could be the predominant source to the circulating CPNE5 during sepsis, albeit cardiomyocytes may also contribute to a lesser extent.

Fig. 4.

CPNE5 expression is dynamically altered in vascular endothelial cells upon inflammatory conditions in vivo and in vitro. (A/B) CPNE5 mRNA levels in mouse aorta and hearts collected at the indicated time points after CS injection. (C) mRNA levels of CPNE5 were measured in endothelial cells (ECs), fibroblasts, and cardiomyocytes isolated from mouse hearts at the indicated time points after CS injection. (D/E) mRNA levels of CPNE5 were measured in mouse cardiac endothelial cells (MCECs) treated with Cytomix (30 ng/mL) or LPS (10 µg/mL) for the indicated time points. (F/G) CPNE5 concentrations were measured in culture supernatants of MCECs upon treatment with Cytomix or LPS. All data are presented as mean ± SD and were analyzed with Student’s t-test. *, p < 0.05 vs. 0 h; n = 4

Lastly, we utilized a mouse cardiac endothelial cell line (MCEC) to further determine the expression of CPNE5 in response to inflammatory stimuli. To mimic the in vivo septic conditions, MCECs were cultured and incubated with either pro-inflammatory cytokine mixture (Cytomix: TNF-α, IL-1β, IFN-γ, 30 ng/mL [29]) or LPS (10 µg/mL). We then collected MCECs at different time points following treatment for RT-qPCR analysis. Our results showed that mRNA levels of CPNE5 were significantly reduced in MCECs at 1.5, 3 and 6 h following either Cytomix-exposure (Fig. 4D) or LPS-challenge (Fig. 4E), compared to basal conditions. In parallel, we collected cell culture supernatants to measure CPNE5 concentration and observed the similar decline pattern in the secretion of CPNE5 from MCECs upon either Cytomix (Fig. 4F) or LPS exposure (Fig. 4G). Together, these in vitro data suggest that inflammatory conditions could inhibit CPNE5 expression in ECs.

CPNE5 knockdown disrupts endothelial barrier integrity under inflammation

To clarify whether reduction of CPNE5 in ECs directly impairs endothelial barrier integrity, we transfected MCECs with siRNA targeting specifically to CPNE5 mRNA (siCPNE5), followed by stimulation with either LPS or Cytomix to determine endothelial permeability. Non-targeting siRNA (siCtrl) was used as controls. RT-qPCR analysis confirmed that mRNA expression of CPNE5 was successfully knocked down by siCPNE5 in ECs with a 76% reduction, compared to control-cells (Fig. 5A). In addition, Western-blotting results validated such an efficient knockdown of endogenous CPNE5 at its protein level in siCPNE5-transfected cells (Fig. 5B/C). We next assessed endothelial permeability in siCPNE5- and siCtrl-transfected cells upon treatment with either LPS or Cytomix, using three methods including FITC-dextran transwell flux, Evans Blue-labeled albumin (EB-BSA) leakage, and trans-endothelial electrical resistance (TEER). As shown in Fig. 5D-F, siCPNE5-ECs exhibited significantly higher leakage of FITC-dextran (Fig. 5D) and EB-BSA (Fig. 5E), accompanied by a sustained lower in TEER (Fig. 5F) upon LPS challenge (10 µg/mL), compared to control cells. Similar results were also observed in siCPNE5-ECs following treatment with Cytomix, in comparison with controls (Fig. 5D-F). Together, these in vitro results demonstrate that CPNE5 may act as an essential factor in maintaining endothelial barrier integrity.

Fig. 5.

Knockdown of CPNE5 in ECs compromises endothelial barrier integrity in vitro. (A-C) Mouse cardiac endothelial cells (MCECs) were transfected with control siRNA (siCtrl) or CPNE5-targeting siRNA (siCPNE5). Knockdown efficiency was confirmed by RT-qPCR (A) and Western blotting (B/C) with quantification normalized to GAPDH (*, p < 0.05 vs. siCtrl; n = 3). (D-F) Endothelial monolayer permeability was assessed by (D) FITC-dextran flux, (E) EB-albumin flux, and (F) trans-endothelial electrical resistance (TEER) measurement (*, p < 0.05 vs. siCtrl; n = 6). All data are presented as mean ± SD and were analyzed with Student’s t-test (A, C) or with two-way ANOVA (D-F)

Loss of CPNE5 aggravates vascular leakage in septic mice

To assess the in vivo role of CPNE5 in sepsis-triggered vascular leakage, we generated CPNE5-knockout (KO) mice using CRISPR/Cas9-mediated deletion. Two sgRNAs targeting exons 4 and 7 resulted in a genomic deletion in the intervening region (Fig. 6A). Genotyping with primers flanking the deleted sequence (P1, P2, and P4) yielded a 426-bp fragment specifically in KO mice, and a 357-bp fragment specifically in WT mice (Fig. 6B). The absence of CPNE5 expression was further confirmed by RT-qPCR (Fig. 6C) and Western-blotting in heart tissues (Fig. 6D). It is important to note that CPNE5-KO mice are viable, normal breeding, and exhibit no overt abnormalities under physiological conditions, consistent with the findings reported by Ding et al. [30]. Next, we injected i.p. cecal slurry (CS) into WT and CPNE5-KO mice to induce sepsis. Twenty hours later, these mice received EB dye via tail-vein injection. Subsequently, we observed that KO mice exhibited visibly less blue in the limbs/footpads/toes and mouth/lips than WT controls (Fig. 6E). Such less peripheral circulation of EB dye in CS-KO mice could be interpreted by a significant portion of the EB dye leaked out into the internal organs.

Fig. 6.

CPNE5 deletion exacerbates vascular leakage in septic mice. (A) Schematic diagram of CRISPR/Cas9 strategy targeting exons 4 and 7 to generate CPNE5-knockout (KO) mice. (B) Genotyping using primers flanking the deleted region yielded a 426-bp fragment specific to KO mice. (C/D) CPNE5-KO mouse model was further validated at (C) mRNA levels by RT-PCR and (D) protein levels by Western blotting using heart tissue samples. (E) Representative images showing less EB dye circulated in mouth lips and limbs of CPNE5-KO septic mice compared to WT-septic mice at 1 h after EB injection. (F-J) Gross images of aorta, heart, lung, stomach, and small intestine collected from CPNE5-KO/CS mice and WT/CS mice at 1 h after EB injection. (K-O) Quantification of EB dye retained within organs by formamide elution and measurement of absorbance at 620 nm. (P) Wet-to-dry (W/D) ratios of the aorta, heart, lung, stomach, intestine, spleen, liver, and kidney collected at 24 h post-CS in CPNE5-KO mice versus WT mice. All data are presented as mean ± SD and analyzed with Student’s t-test. *, p < 0.05 vs. WT/CS; n = 8

To test the above interpretation, we harvested major organs at 1 h post-EB injection for visual and quantitative analysis. Overall visual examination revealed markedly stronger blue coloration in the aorta, heart, lung, stomach, and small intestine of CS-KO mice than those organs collected from CS-WT mice (Fig. 6F-J), whereas the liver, spleen, and kidney showed no visible difference between two groups (Supplemental Fig. S4A-C). Accordingly, quantification of tissue-extravasated EB by formamide elution measured absorbance at 620 nm revealed significantly greater elution of EB dye from these major organs of CS-KO mice compared to WT-controls: approximately 1.5-fold in the aorta, 1.2-fold in the heart, 1.4-fold in the lung, 1.8-fold in the stomach, and 1.8-fold in the small intestine (Fig. 6K-O). No significant differences were observed in the liver, spleen, or kidney between two groups (Supplemental Fig. 4D-F). In addition, we assessed organ-level fluid retention by calculating the wet-to-dry (W/D) weight ratios. Consistent with the EB-elution results, CS-KO mice showed significantly higher W/D ratios in the aorta, heart, lung, stomach, and intestine than WT controls, whereas no significant differences were observed in the liver, spleen, or kidney (Fig. 6P). Furthermore, to gain a closer look at EB leakage within tissues, we examined frozen sections of major organs using confocal fluorescence microscopy. Please note that EB dye is red under fluorescence microscopy. We observed that the intensity of red fluorescence in the interstitial space was stronger in the aorta, heart, lung, stomach, and small intestine of CS-KO mice than CS-WT controls, indicating more severe EB extravasation into the tissue interstitium (Supplemental Fig. S4G-K). ImageJ-based quantification results showed that the mean fluorescence intensity (MFI) were markedly elevated across all examined organs of CS-KO mice versus controls: the aorta showed the most striking increase with approximately 5.6-fold higher than CS-WT controls (Supplemental Fig. S4L), the heart with 1.8-fold increase (Supplemental Fig. S4M), the lung with 2.1-fold (Supplemental Fig. S4N), the stomach with 1.9-fold (Supplemental Fig. S4O), and the small intestine with 2.9-fold increase (Supplemental Fig. S4P). Together, multiple lines of evidence consistently indicate that CPNE5 is an essential factor for endothelial cells to maintain vascular barrier integrity during sepsis. Its deficiency augments sepsis-caused vascular leakage.

Loss of CPNE5 aggravates multi-organ injury and increases mortality in septic mice

Given that vascular hyperpermeability is a key driver of multiorgan dysfunction in sepsis [11–13], we next examined whether loss of CPNE5 exacerbates tissue injury in major organs during sepsis. Histological analysis results showed that the aorta collected from CS-treated CPNE5-KO mice had severe jagged clefts and disorganized smooth muscle layers, whereas the moderate surface irregularities were observed in CS-WT aorta (Fig. 7A). In the lungs of CS-KO mice showed prominent alveolar collapse, interstitial edema, and dense inflammatory cell infiltration, whereas CS-WT lungs displayed fewer pathological changes (Fig. 7B). Intestinal sections from CS-treated CPNE5-KO mice also demonstrated more pronounced villus atrophy, epithelial shedding, submucosal edema, and greater visible increase in goblet cells, compared to CS-WT controls (Fig. 7C). These histological differences were further evaluated by their respective injury scores which showed significantly higher in the aorta, lung, and intestine collected from CPNE5-KO mice upon the i.p. injection of cecal slurry, in comparison with WT-controls (Fig. 7D). Lastly, we assessed whether CPNE5-deficiency impacts survival outcomes under septic stress and observed that CPNE5-KO mice exhibited significantly lower survival rate (17.6%, n = 17) than WT controls (38.5%, n = 13, p < 0.001) at 96-h post-CS sepsis (Fig. 7E).

Fig. 7.

Loss of CPNE5 aggravates multi-organ injury and mortality during sepsis. (A-D) Representative H&E-stained sections of the aorta (A), lung (B), and intestine (C) from WT and CPNE5-KO mice at 24 h after CS injection, and (D) their quantification of histological injury scores. Data are presented as mean ± SD and analyzed by Student’s t-test (*, p < 0.05 vs. WT/CS; n = 5). (E) Kaplan-Meier survival curves were generated to compare mortality between two groups, significance was determined by log-rank (Mantel-Cox) test (*, p < 0.05 vs. WT/CS; n = 13 for WT/CS, n = 17 for CPNE5-KO/CS)

CPNE5-deficiency causes reduction of EC junctional protein levels upon septic conditions

To explore the possible mechanism by which CPNE5 modulates endothelial barrier function, we first assessed whether CPNE5-knockdown affects cell viability or cytotoxicity under inflammatory conditions. MCECs transfected with either siCtrl or siCPNE5 were stimulated with PBS, LPS, or Cytomix for 3 h. Neither MTS assay nor LDH release showed significant differences between groups, indicating that CPNE5-knockdown does not influence cell survival or death under short-term inflammatory stress (Fig. 8A/B). We next examined whether CPNE5 affects the expression of major vascular leakage mediators. RT-qPCR analysis results indicated that mRNA levels of Ang-II, ICAM-1, and SDC-1 were upregulated in MCECs upon stimulation with Cytomix compared to PBS, but no significant differences were observed between siCtrl and siCPNE5 groups (Supplemental Fig. S5A-C). Similar results were also observed for the secretion of these proteins from MCECs in the absence or presence of Cytomix between two groups (Fig. 8C-E). These findings suggest that CPNE5 does not alter the expression of these classical vascular leakage modulators.

Fig. 8.

CPNE5 deficiency reduces membrane levels of junctional proteins in ECs upon inflammatory insults. (A/B) Cell viability and death were assessed in MCECs transfected with siCtrl or siCPNE5 and treated with PBS, LPS, or Cytomix for 3 h. MCECs treated with Triton (0.5%) were used as positive controls of cell death. (C-E) ELISA analysis showed that knockdown of CPNE5 did not alter the supernatant levels of Ang-II (C), sICAM-1 (D), or SDC-1 (E) after Cytomix stimulation, compared to siCtrl-cells. (F/G) Representative flow cytometry histograms and (H) their quantification results showed lower membrane levels of VE-cadherin (VE-Cad) and Occludin in siCPNE5-transfected MCECs versus siCtrl-cells after incubation with Cytomix for 3 h. (I/J) Representative flow cytometry histograms and (K) their quantification results for membrane levels of VE-cadherin and Occludin in cardiac ECs isolated from CPNE5-KO and WT mice at 24 h post-CS. (L/M) Representative flow cytometry histograms and (N) their quantification results for membrane levels of VE-cadherin and Occludin in pulmonary ECs isolated from CPNE5-KO and WT mice at 24 h post-CS. (O) RT-qPCR analysis to determine expression levels of MMP2, MMP3, MMP9, ADAM10 and ADAM17 in MCECs transfected with siCtrl or siCPNE5 and treated with PBS or Cytomix (30 ng/mL) for 3 h. Gene expression was normalized to GAPDH and presented as relative fold change. All data are presented as mean ± SD and analyzed by Student’s t-test. *, p < 0.05; n = 4–8. n.s. stands for not significant

Given the central role of VE-cadherin and Occludin in maintaining endothelial junction integrity [31], we then investigated whether CPNE5-deficiency affects their expression at the transcriptional and translational levels. RT-qPCR results showed that Cytomix stimulation did downregulate mRNA expression of VE-cadherin and Occludin in both groups, but there was no difference between siCtrl- and siCPNE5-cells (Supplemental Fig. S5D/E). In addition, flow cytometry analysis revealed no significant difference in the EC membrane levels of VE-cadherin and Occludin under PBS basal conditions (Supplemental Fig. S5F-H). However, following treatment with Cytomix, knockdown of CPNE5 significantly reduced the membrane levels of VE-cadherin and Occludin, compared to siCtrl-cells (Fig. 8F-H). This suggests that CPNE5 may maintain the membrane retention of junctional proteins rather than their transcriptional regulation in ECs upon septic conditions. To validate these findings in vivo, CD31⁺-endothelial cells were sorted out from mouse hearts and lungs at 24 h post-CS or -PBS injection. We observed that, in PBS-treated WT and CPNE5-KO mice, membrane levels of VE-cadherin and Occludin were no difference in either cardiac or pulmonary ECs (Supplemental Fig. S5I-N); however, upon septic conditions, both cardiac and pulmonary ECs isolated from CPNE5-KO mice exhibited greatly reductions in the membrane levels of VE-cadherin and Occludin, compared to CS-WT controls (Fig. 8I-N).

Currently, it is well known that sepsis can stimulate the expression of multiple extracellular matrix (ECM) proteinase genes including MMP2, MMP3, MMP9, ADAM10, and ADAM17 [32–34]. To test whether knockdown of CPNE5 in ECs could affect expression of the above ECM proteinase genes, we next performed RT-qPCR and observed that siRNA-mediated knockdown of CPNE5 could significantly elevate expression levels of MMP2 and ADAM10, but not MMP3, 9 and ADAM17, compared to siCtrl-ECs upon basal and inflammatory conditions (Fig. 8O). These results suggest that reduced membrane levels of VE-cadherin and Occludin in CPNE5-knockdown ECs could be due to the increased cleavage by MMP2 and ADAM10.

Exogenous addition of Recombinant CPNE5 protein to ECs decreases inflammation-caused monolayer leakage

To test whether elevation of CPNE5 levels could reduce vascular permeability upon inflammatory conditions in vitro, we treated mouse cardiac endothelial cells (MCECs) with recombinant CPNE5 protein (rCPNE5, 100 ng/mL) or control BSA for 24 h, followed by the incubation with LPS (10 µg/mL) or Cytomix for barrier integrity and permeability assays. Please note that endotoxin levels in rCPNE5 and BSA proteins are < 0.01EU/µg, low enough to exclude its effects. We observed that treatment with rCPNE5 protein significantly mitigated LPS- or Cytomix-induced leakage of FITC-dextran (Fig. 9A) and EB-albumin (Fig. 9B), compared to control BSA-treated cells. Similarly, LPS- or Cytomix-caused decreases in TEER values over time were greatly attenuated in rCPNE5-ECs, compared to BSA-cells (Supplemental Fig. S6).

Fig. 9.

Exogenous additions of recombinant CPNE5 protein (rCPNE5) to ECs decreases inflammation-caused monolayer leakage. (A/B) MCECs were pretreated with rCPNE5 protein (100 ng/mL) or BSA for 24 h, followed by stimulation with LPS (10 µg/mL), or Cytomix (30 ng/mL) for 3 h. Endothelial permeability was assessed by (A) FITC-dextran flux assay and (B) Evans blue (EB)-albumin flux assay. (C) RT-qPCR analysis to measure expression levels of MMP2, MMP3, MMP9, ADAM10 and ADAM17 in rCPNE5-treated ECs versus BSA-treated cells upon exposure to LPS (10 µg/mL) and Cytomix (30 ng/mL) for 3 h. (D/E) Representative flow cytometry histograms and (F) their quantification results for membrane levels of VE-cadherin in rCPNE5-treated ECs versus BSA-treated cells upon exposure to LPS (10 µg/mL) and Cytomix (30 ng/mL) for 3 h. (G/H) Representative flow cytometry histograms and (I) their quantification results for membrane levels of Occludin in rCPNE5-treated ECs versus BSA-treated cells upon exposure to LPS (10 µg/mL) and Cytomix (30 ng/mL) for 3 h. Data are shown as mean ± SD and analyzed by Student’s t-test. *p < 0.05; n = 4–6

To further determine whether rCPNE5-elicited protection against EC monolayer leakage is associated with proteinase-mediated cleavage of junctional proteins, we first performed RT-qPCR to measure expression levels of major proteinase genes in rCPNE5-treated ECs and control cells upon basal and inflammatory conditions. We found that treatment of ECs with rCPNE5 protein significantly inhibited LPS- and Cytomix-caused upregulation of MMP2 and ADAM10, but not MMP3, 9 and ADAM17, compared to control BSA-cells (Fig. 9C). Accordingly, flow cytometry analysis showed that membrane levels of VE-Cadherin (Fig. 9D-F) and Occludin (Fig. 9G-I) were significantly higher in rCPNE5-treated ECs than control BSA-cells upon LPS or Cytomix insults.

Discussion

To our knowledge, this is the first study to test the impact of Copine family members in sepsis-associated vascular leakage. In this study, we identified that CPNE5 is predominantly expressed in vascular endothelial cells, compared to other members of Copine family. Furthermore, we observed that plasma levels of CPNE5 were remarkably lower in human patients with sepsis at admission than healthy donors, which were negatively correlated with the higher levels of Ang-II, sICAM-1, and SDC-1, three well-characterized biomarkers for vascular leakage. Consequently, septic patients with lower plasma levels of CPNE5 had worse overall survival. In addition, these findings were validated in the septic mouse model, which clearly showed a negative correlation of CPNE5 levels with vascular leakage biomarkers and severity of sepsis. Importantly, utilizing both in vitro endothelial cells and in vivo knockout mouse model, we demonstrated that CPNE5 deficiency augmented endothelial permeability upon inflammatory conditions and increased vascular leakage in septic mice, leading to multiple organ damage and death. Therefore, multiple lines of evidence clearly indicate that low levels of plasma CPNE5 could serve as a novel biomarker for the diagnosis of vascular leakage during sepsis.

Currently, abnormal expressions of CPNEs have been identified as potential effective molecular markers for the early diagnosis, progression and prognosis of various tumor types in human patients [19]. Interestingly, some CPNE members including CPNE1, 3, 6, and 8 are characterized as oncogenic genes [19, 21]; others such as CPNE2, 7, and 9 are considered as tumor-suppressor genes [19, 35]; whereas CPNE5 exhibits dual roles in cancers which downregulation is associated with a poor prognosis in esophageal squamous cell carcinoma (ESCC) [36] and oppositely, higher expression of CPNE5 is linked to a positive prognosis in multiple myeloma [37]. These prior studies on tumor cells suggest that each CPNE member is critical regulator of cellular activity in response to stress or disease conditions. Nonetheless, whether the expression of CPNEs is altered in vascular endothelial cells during sepsis remains totally unknown. Hence, our study presented here provides the first evidence to show that the expression of CPNE5 is downregulated in ECs upon in vitro inflammatory conditions and in vivo sepsis. Regarding the alterations of CPNE5 expression in ECs upon inflammatory stimulation, there are some differences between in vitro and in vivo. We observed that the expression levels of CPNE5 in inflammation-stimulated ECs were continuously decreased in vitro, whereas being acutely increased in murine ECs at 12 h post-sepsis and then remarkably decreased at time points of 24 h and 48 h post-sepsis. This could be interpreted by an in vivo (within a living mouse) response integrating multiple systems, including other cell types, soluble factors, and systemic feedback loops, that are absent from the in vitro culture dish for cellular experiments. Hence, such biphasic response of CPNE5 expression in the murine blood samples and organs (e.g., heart, aorta) is not detectable in cultured cellular experiments upon inflammatory conditions. Indeed, these results also suggest that the in vivo transient elevation of CPNE5 levels could be a compensatory mechanism to maintain vascular integrity and rescue mice from septic injury. Accordingly, the eventual reduction of CPNE5 in the blood and tissues of septic mice could be a major contributor to organ failure and mortality during sepsis.

Along this line, we observed that under basal conditions, CPNE5 deficiency did not affect endothelial barrier integrity, which seems inconsistent with its role in septic conditions. This could be due to the following facts that: (1) vascular permeability is tightly controlled and maintained under normal conditions through layers of redundant, post-transcriptional, and mechanical regulatory mechanisms; and (2) endothelial cells actively work together to preserve a stable barrier and consequently, various factors, such as blood flow, neighboring pericytes, and cell-junction complexes, can compensate for some genetic alterations to uphold this precise control of vascular integrity under normal conditions [38, 39]. Therefore, altered expressions of CPNE5 may not affect baseline vascular permeability and accordingly, CPNE5-KO mice do not show any obvious pathophysiological abnormality in the absence of disease or stress conditions which is consistent with previous report by Ding et al. [30]. However, upon septic conditions, uncontrolled inflammation could change such instinctive compensatory balance, leading to the increased vascular permeability in CPNE5-deficiency mice.

At present, among 9 members of CPNE family, only CPNE7 is reported to be secreted from ameloblasts [40]. While CPNE5 is detectable in the blood and cell culture supernatants, how plasma CPNE5 levels decline in human septic patients remains unclear. Given that CPNE5 is also expressed in other cell types (i.e., cardiomyocytes) in addition to vascular ECs [28] (Supplemental Fig. S3) and sepsis stimulates the expression of ECM proteinases, therefore, reduced plasma levels of CPNE5 during acute sepsis could be ascribed to multiple origin sources and other possible approaches such as altered vesicular release, proteolytic degradation, or accelerated clearance. What’s more, we performed additional analysis of bulk-RNA sequencing data generated by Tsalik et al.., who used whole blood cells from a cohort of sepsis patients with 78 survivors and 28 non-survivors (GEO dataset accession #GSE63042) [41]. Interestingly, we found that the mRNA expression levels of CPNE5 in peripheral blood cells were significantly decreased in sepsis non-survivors, compared to sepsis survivors (Supplemental Fig. S7), which is consistent with the plasma data in our cohort of septic patients. Hence, it is plausible that some fractions of reduced plasma CPNE5 in septic patients could also be derived from blood cells. More importantly, these findings from two cohorts of septic patients recruited at different clinic centers firmly suggest that low plasma levels of CPNE5, regardless of its origin sources, could be a prognostic indicator of the mortality in human patients with sepsis.

There are several limitations to this study. Firstly, a global CPNE5-knockout mouse model was used in the present study, which prevents clear attribution of the observed vascular leakage and mortality to endothelial loss of CPNE5. As discussed above, CPNE5 is expressed in multiple tissues and cell types, and its deletion in non-endothelial compartments could also contribute to the phenotype observed in CPNE5-KO mice upon septic conditions. Thus, future investigations using endothelial-specific CPNE5-KO model will be warranted to clarify this issue. Secondly, the mechanisms by which CPNE5 deficiency augments vascular leakage during sepsis remain underexplored. While we did initial measurements of proteinase gene expression and membrane junctional protein levels, future studies will be needed to assess junctional protein trafficking dynamics, phosphorylation status, interactions with scaffolding partners, and how CPNE5 regulates expression of MMP2 and ADAM10 in ECs. Lastly, the current study enrolled 77 patients with sepsis which provided initial evidence supporting CPNE5 as a promising biomarker for vascular leakage and sepsis severity, our findings should be interpreted with caution due to the relatively small sample size. A limited number of subjects may increase the risks of random error and selection bias, consequently affecting the generalizability and robustness of our results. Moreover, this was a single-center study, which may further limit the representativeness and external validity of our findings. Therefore, future large-scale, multicenter studies are necessary to validate the diagnostic and prognostic utility of CPNE5 across diverse patient populations with sepsis, and to establish precise clinical reference thresholds.

Conclusion

In summary, our results demonstrate that reduced plasma levels of CPNE5 correlate with the increased vascular leakage, multiple organ injury, and death in the setting of sepsis. Mechanistically, decreased expressions of CPNE5 could promote the cleavage of EC junctional proteins (i.e., VE-cadherin and Occludin) through the increased expressions of MMP2 and ADAM10. These findings suggest that reduced plasma levels of CPNE5 may serve as one of potential indicators for sepsis-induced vascular leakage and mortality. Given that sepsis is a highly heterogeneous syndrome that leads to significant morbidity and mortality, this study may imply CPNE5 as a therapeutic potential of precision medicine for improving survival outcomes in a heterogeneous patient population with sepsis.

Methods and materials

Human study

This study included the collection and analysis of peripheral blood samples from patients with sepsis and healthy controls. A detailed flowchart of the patient enrollment and sample collection is presented in Supplemental Figure S2. During June 2024 and March 2025, 83 adult patients diagnosed with sepsis were prospectively enrolled from the Intensive Care Unit (ICU) at Renmin Hospital of Wuhan University. Sepsis was defined according to the Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3) [42]. Patients with underlying malignancies or aged over 80 years were excluded, yielding a final cohort of 77 eligible sepsis patients. All patients were followed up for 28 days post-admission to evaluate survival outcomes and assess the prognostic relevance of biomarker levels. In parallel, 44 age- and sex-matched healthy individuals without signs of infection were recruited as controls.

For all participants, 3–5 mL of peripheral blood was collected within the first 24 h of ICU admission. Serum was isolated by centrifugation and stored at −80 °C until further analysis. Clinical parameters, including Sequential Organ Failure Assessment (SOFA) scores and arterial blood gas measurements, were recorded at the time of blood collection. Serum concentrations of CPNE5, soluble intercellular adhesion molecule-1 (sICAM-1), angiopoietin-II (Ang-II), and syndecan-1 (SDC-1) were quantified using enzyme-linked immunosorbent assays (ELISAs) according to the manufacturer’s protocols.

Murine model of sepsis

CPNE5-knockout (KO) mice were generated using the CRISPR/Cas9 system in C57BL/6 background by the Division of Developmental Biology at Cincinnati Children’s Hospital Medical Center. All these mice were bred in the Division of Laboratory Animal Resources at the University of Cincinnati Medical Center. To induce polymicrobial sepsis, both wild-type (WT) and CPNE5-KO mice (8-weeks old, either sex) were subjected to cecal slurry (CS)-injection (i.p.), as described previously [43] and detailed in Supplementary material online, Methods.

Enzyme-linked immunosorbent assay (ELISA)

The concentrations of CPNE5, Ang-II, sICAM-1, and SDC-1 in human and mouse serum samples were determined using ELISAs according to the manufacturers’ protocols. Human ELISA kits for CPNE5, Ang-II, ICAM-1, and SDC-1 were purchased from Elabscience (Wuhan, China), mouse ELISA kits for the same targets were obtained from Abbexa (Cambridge, UK). All samples were measured in duplicate, and detailed in Supplementary material online, Methods.

Genotyping and quantitative RT-qPCR

Mouse genotyping was performed by PCR amplification with specific primers designed to distinguish wild-type and CPNE5-knockout alleles. Quantitative real-time PCR (RT-qPCR) was performed in triplicate using the RADIANT SYBR Green Master Mix qPCR Kit (Alkali Scientific) and the Azure Cielo Real-Time PCR System (Azure Biosystems). All primer sequences are listed in Supplemental Table S5. Gene expression was normalized to the housekeeping gene GAPDH, and relative expression levels were calculated using the 2⁻ΔΔCt method.

Isolation of adult mouse cardiac endothelial cells, cardiomyocytes, and fibroblasts

Adult mouse cardiac endothelial cells (ECs), cardiomyocytes, and fibroblasts were isolated from hearts of mice subjected to CS injection for 12–24 h, as previously reported by Zhao et al. [44]. and detailed in Supplementary material online, Methods.

SiRNA transfection of mouse cardiac endothelial cells

MCECs were seeded in 6-well plates at a density of 3 × 10⁵ cells per well and cultured until reaching approximately 80% confluency. siRNA targeting CPNE5 (siCPNE5) and a universal negative control siRNA (siCtrl) were purchased from the Horizon Discovery. Transfection was performed using the TransIT-X2^® Dynamic Delivery System (Mirus Bio), following the manufacturer’s instructions and detailed in Supplementary material online, Methods.

Western-blotting analysis

Proteins were extracted from mouse heart tissues (wild-type and CPNE5-KO) and MCECs transfected with siCtrl or siCPNE5 using 1× RIPA lysis buffer (Fivephoton Biochemicals) supplemented with 0.1 mM PMSF (Sigma-Aldrich), protease inhibitor cocktail (Roche), and 1% phosphatase inhibitor cocktail (Cell Signaling Technology). Primary antibodies against CPNE5 (rabbit polyclonal, Novus Biologicals, 1:1000) and GAPDH (mouse monoclonal, ABclonal, 1:5000) were used and whole procedure is detailed in Supplementary material online, Methods.

In vitro endothelial permeability assays

MCECs were transfected with either siCtrl or siCPNE5 for 60 h and then seeded onto 12-well transwell inserts (Corning, 12-mm diameter, 0.4 μm pore size, polyester membrane) at a density of 2 × 10⁵ cells per well and cultured to form an intact endothelial monolayer, followed by incubation with either lipopolysaccharide (LPS, Sigma, 10 µg/mL) or a pro-inflammatory cytokine mixture (Cytomix; IL-1β, IFN-γ, and TNF-α, 30 ng/mL) [29] to induce monolayer leakage. For the treatment of recombinant mouse CPNE5 protein (rCPNE5, purchased from LB Leading Biology), ECs were directly seeded onto 12-mm Transwell and treated with rCPNE5 protein (100 ng/mL) or control BSA for 24 h, followed by the addition of LPS (10 µg/mL) or Cytomix (30 ng/mL) for barrier integrity and permeability assays. The Cytomix used for the in vitro ECs is to mimic the in vivo septic inflammatory conditions and this concentration is established relevance to human sepsis [29]. Endothelial permeability was assessed at 1.5, 3- and 6-h post-treatment using three complementary approaches: trans-endothelial electrical resistance (TEER), FITC-dextran flux, and Evans Blue (EB)-BSA leakage assays, as previously described [44] and detailed in Supplementary material online, Methods.

In vivo vascular permeability assay

To evaluate in vivo vascular permeability, WT and CPNE5-KO mice (8–10 weeks old) were subjected to CS injection (i.p.) to induce sepsis. At 20 h post-CS, mice were anesthetized and intravenously injected via the tail vein with 0.5% EB dye (40 µg/g BW, dissolved in PBS). After 1 h, mice were deeply anesthetized with ketamine (100 mg/kg) and xylazine (5 mg/kg), intravascular EB dye was flushed out by trans-cardiac perfusion with ice-cold PBS containing 2 mM EDTA (Fisher BioReagents), delivered slowly using a syringe via ventricular needle puncture until the effluent was colorless (total ~ 10–20 mL). For lungs, a 10 mL syringe with a sterile hypodermic needle was inserted into the right ventricle (ventricular needle puncture) and the left atrium (LA) was incised for outflow. Perfusion proceeded slowly until the effluent was clear and the lungs became pale/gray (typically need ~ 10–20 mL total). For other organs (intestine, stomach, heart, aorta), a 10 mL syringe/needle was inserted into the left ventricle, and the right atrium (RA) was opened; ~10–20 mL was perfused slowly until the effluent became clear. After flushing out the intravascular EB dye, organs including the heart, aorta, lung, liver, spleen, kidney, stomach, and terminal ileum were carefully harvested and imaged for gross comparison of EB accumulation between groups. Subsequently, each organ was incubated in 500 µL of formamide (Sigma) at 55 °C for 48 h to extract tissue-bound Evans blue. For small tissues such as the aorta, 200 µL of formamide was used. After incubation, 100 µL of the dye-containing supernatant was transferred to a clear 96-well flat-bottom plate, and optical density was measured at 620 nm using a microplate reader (Tecan Infinite F50, Tecan, USA). The EB content was quantified based on a standard curve generated by serial dilution of EB in formamide and normalized to organ weight.

The histological assessment of vascular leakage at the tissue level and tissue edema evaluation by calculating wet-to-dry (W/D) weight ratios are detailed in Supplementary material online, Methods.

Histological evaluation and injury scoring

At 24 h post-CS sepsis, tissues including the lung, small intestine, and aorta were collected, fixed in 4% paraformaldehyde for 24–48 h, embedded in paraffin, and sectioned at 5 μm thickness. Lung injury was assessed using a modified Smith scoring system, as described previously [45]. Intestinal injury was evaluated according to Chiu’s Score, which reflect mucosal damage and epithelial loss on a 0–5 scale [46]. Aortic injury was scored semi-quantitatively across four parameters including endothelial integrity, smooth muscle alignment, medial disruption, and inflammatory infiltration with each scored 0–3. The total score ranged from 0 to 12 and is described in Supplemental Table S6.

Flow cytometry analysis of EC junctional proteins

Flow cytometry analysis of VE-cadherin and Occludin in the cultured ECs and in the cardiac as well as pulmonary ECs is performed as described previously [47] and detailed in Supplementary material online, Methods.

Statistical analysis

Statistical analyses were performed using GraphPad Prism 8 (GraphPad Software, San Diego, CA, USA). Unless otherwise noted, all data are presented as mean ± standard deviation (SD). Comparisons between two groups were analyzed using the unpaired Student’s t-test, while comparisons among multiple groups were conducted using one-way ANOVA or two-way ANOVA. Survival analysis was conducted using Kaplan-Meier curves, and differences between groups were assessed using the log-rank (Mantel-Cox) test. Correlation analyses between clinical and molecular parameters were conducted using Spearman’s rank correlation coefficient due to the non-normal distribution of the data. A p value < 0.05 was considered statistically significant. ROC curve analysis was conducted in R (version 4.4.2) using the pROC package (version 1.19.0.1), detailed in Supplementary material online, Methods.

Author contributions

MP designed and performed experiments, analyzed data, and wrote manuscript. ZL and XW performed experiments and analyzed data. MP, JL and LYZ collected human samples and performed experiments. TY analyzed single-nucleus and bulk RNA-sequencing data. NEK helped with the data analysis. WH, HF, TP, and LYZ helped with experimental design and data analysis. GCF designed this project, analyzed all data, edited manuscript, provided financial and administrative support, and gave final approval of the manuscript. All the authors read and approved the final manuscript.

Funding

R01 HL160811/HL/NHLBI NIH HHS/United States (Fan, G-C).

R01 HL163148/HL/NHLBI NIH HHS/United States (Huang, W).

Data availability

1. The datasets used and/or analyzed in the current study are available from the corresponding author on reasonable request. Single-nucleus RNA-sequencing data that support the findings of this study can be found in the following links: (https://singlecell.broadinstitute.org/single_cell/study/SCP1909/) (https://singlecell.broadinstitute.org/single_cell/study/SCP1265/). 2. Whole blood bulk RNA-sequencing data of human patients with survivors and non-survivors can be accessed via the GEO repository (GSE63042): (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE63042/) 3. The original Western-blots can be found in the **Supplemental Fig. S8**.

Declarations

Ethics approval and consent to participate

The human study was approved by the Ethics Committee of Renmin Hospital of Wuhan University [Approval No.2024-K106 (C01)], and written informed consent was obtained from all patients or their legally authorized representatives prior to enrollment, in accordance with the Declaration of Helsinki. All animal experiments were conducted in compliance with the NIH Guide for the Care and Use of Laboratory Animals and were approved by the Institutional Animal Care and Use Committee (IACUC) of the University of Cincinnati (Protocol No. 24-07-03-01).

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.