Neutrophil extracellular trap-borne C3-driven endothelial dysfunction in Klebsiella pneumoniae liver abscess

ABSTRACT

Klebsiella pneumoniae (K. pneumoniae) liver abscess (KPLA) is a severe bacterial infection that is frequently complicated by intrahepatic thrombophlebitis and extrahepatic metastatic infections leading to high mortality rates. This study investigates the role of neutrophil extracellular traps (NETs) in endothelial injury and disease progression in KPLA. Our findings reveal that KPLA patients with intrahepatic thrombophlebitis are more prone to developing sepsis and extrahepatic migratory infections. K. pneumoniae induces NET formation via the TLR4-PI3Kα-AKT signaling pathway, and C3 deposition on NETs significantly contributes to endothelial injury. In a KPLA mouse model, increased C3 levels were observed in the liver, with NETs carrying substantial amounts of C3, disrupting the endothelial barrier and exacerbating liver injury. Treatment with the C3 inhibitor AMY-101 reduced C3 deposition on NETs, alleviated endothelial damage, significantly improved survival, and reduced extrahepatic dissemination, inflammatory infiltration, and lung injury while also suppressing systemic inflammation. Our findings underscore the pivotal role of C3 in NET-mediated endothelial damage and the pathogenesis of KPLA. Thus, targeting C3 deposition on NETs may be a promising therapeutic strategy to reduce endothelial injury, thrombosis, and extrahepatic infections in KPLA without compromising neutrophil antimicrobial function.

Introduction

Klebsiella pneumoniae is a gram-negative opportunistic pathogen that is the leading cause of monomicrobial liver abscesses (LAs) in Asia [1]. K. pneumoniae liver abscess (KPLA) often presents with thrombophlebitis, an independent risk factor for extrahepatic metastatic infections (EMIs) [2,3]. EMI is a severe complication of KPLA frequently manifested as septic pulmonary embolism, endophthalmitis, and infections of the nervous system and soft tissues, with a mortality rate of 2.8–10.8% [4,5]. These findings underscore the urgent need for novel therapeutic strategies to manage KPLA and prevent its progression to EMI.

The pathological mechanism of thrombophlebitis is primarily characterized by endothelial cell injury and inflammatory cell infiltration. Liver sinusoidal endothelial cells (LSECs) play an essential role in the immune response by forming a barrier within the liver sinusoids and clearing bacterial products, such as lipopolysaccharide (LPS), from portal venous blood while avoiding inflammation. However, sustained inflammation recruits neutrophils, which disrupt the LSEC barrier, promoting localized inflammation and thrombosis [6–8]. Neutrophils are the first line of defense against infections, utilizing phagocytosis, degranulation, and the production of reactive oxygen species (ROS) to combat pathogens. Additionally, neutrophils can undergo NETosis, a process wherein they release neutrophil extracellular traps (NETs) as an antimicrobial mechanism. While K. pneumoniae triggers NETosis, the precise mechanisms underlying this process remain unclear [9,10]. Although NETs are essential in K. pneumoniae infections, their components, such as histones and myeloperoxidase (MPO), damage endothelial cells and promote thrombosis [11–15].

The PI3K/AKT signaling pathway is crucial for regulating neutrophil immune functions, including NET formation in response to various stimuli [16,17]. AKT activation promotes the shift from neutrophil apoptosis to NETosis [18]. The complement system is an essential component of innate immunity that defends against infections and clears immune complexes and damaged cells [19]. Neutrophils have key complement components, including complement factor P (CFP), complement factor B (CFB), and C3. Studies have shown that NETosis induced by stimuli like phorbol myristate acetate (PMA) or Pseudomonas aeruginosa activates the complement, leading to the deposition of CFP and C5b-9 on NETs [20]. However, the complement components carried by KP-NETs and their functions remain to be fully elucidated, as they may vary depending on the inducing stimuli [21,22].

In this study, we demonstrate that K. pneumoniae induces NET formation through the TLR4-PI3Kα-AKT signaling pathway. C3-loaded KP-NETs contribute to endothelial cell damage and dysfunction of the liver sinusoidal barrier, leading to thrombophlebitis and exacerbating KPLA progression. Moreover, the C3 inhibitor AMY-101 effectively reduces C3 deposition on NETs, restoring liver sinusoidal barrier integrity. These findings suggest that KP-NETs and their associated C3 May serve as potential biomarkers for early detection and therapeutic targets to prevent EMI in patients with KPLA.

Results

Intrahepatic thrombophlebitis is associated with adverse outcomes in KPLA

This study analyzed 357 patients with clinically diagnosed KPLA and confirmed to have K. pneumoniae infections based on blood or abscess fluid cultures. Among these, 96 patients had intrahepatic thrombophlebitis (Supplementary Table S1) and experienced significantly worse clinical outcomes, including more pronounced inflammatory responses (Figure 1(A)), higher rates of sepsis (p = 0.011), and a greater prevalence of extrahepatic metastatic infections (p < 0.001) than the others (Figure 1(B)). Common complications included lung abscesses, renal abscesses, and soft tissue infections (Figure 1(C)). Intrahepatic thrombophlebitis is a risk factor for extrahepatic infections in KPLA. Our findings suggest that it may also accelerate disease progression. Infections often lead to endothelial damage and dysfunction, which are key drivers of thrombosis [23]. To explore this relationship, we assessed markers of endothelial injury in patients with KPLA. Patients with intrahepatic thrombophlebitis had significantly higher levels of endothelial injury markers than those without thrombophlebitis (Figure 1(D)).

Figure 1.

Thrombophlebitis exacerbates the disease progression of KPLA. (A) cytokine levels in the circulation of patients with KPLA and thrombophlebitis vs. non-thrombophlebitis (n = 10). (B) incidence of adverse events in patients with KPLA with thrombophlebitis vs. non-thrombophlebitis. (C) black arrows indicate liver abscess, lung abscess, kidney abscess, and neck muscle tissue infection; white arrows indicate intrahepatic thrombophlebitis, manifested as low-density filling defects in the right branch of the portal vein. (D) plasma levels of endothelial injury markers (vWF, ICAM-1, VCAM-1) in healthy controls (hc), patients with KPLA with non-thrombophlebitis (non-throm), and patients with KPLA with thrombophlebitis (throm). hc: healthy controls; non-throm: patients with KPLA with non-thrombophlebitis; throm: patients with KPLA with thrombophlebitis. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

These results indicate that endothelial injury – associated thrombophlebitis is likely an important contributor to disease severity and progression in patients with KPLA.

Intestinal K. pneumoniae induces bacterial liver injury in mice

Excessive bacterial growth is a critical factor in bacterial translocation and liver injury, often leading to severe infectious conditions, such as spontaneous bacterial peritonitis, pyogenic liver abscess, and sepsis [24–26]. To investigate how K. pneumoniae compromises the liver sinusoidal barrier and induces bacterial liver injury, we established a KPLA mouse model by administering a high dose of K. pneumoniae orally over a short period. This approach favors intestinal colonization by K. pneumoniae. It is associated with small-intestinal injury – widened subepithelial spaces, villous misalignment with edema and increased inflammatory cells – and reduced expression of E-cadherin and ZO-1, supporting the plausibility of bacterial translocation to the liver via the gut – liver axis (Figure 2(A)). We observed that K. pneumoniae significantly disrupted the liver sinusoidal barrier in mice using Evans blue dye injected via the retro-orbital route (Figure 2(B,C)). Histopathological analysis revealed that KPLA mice with intrahepatic thrombophlebitis had more extensive liver damage, greater inflammatory cell infiltration (Figure 2(D,E)), and higher circulating levels of IL-6, TNF-α, and IL-1β compared to KPLA mice without intrahepatic thrombophlebitis (Figure 2(F,H)). Immunohistochemical staining further confirmed severe neutrophil infiltration and endothelial damage in regions of thrombophlebitis (Figure 2(I)).

Figure 2.

Thrombophlebitis exacerbates K. pneumoniae-induced bacterial liver injury. (A) Representative images of intestinal morphology and permeability assessment in mice orally gavaged with pbs or K. pneumoniae. scale bar = 100 μm (H&E), 50 μm (immunohistochemical images). (B, C) mice were infected with 2 × 10^7 cfu of K. pneumoniae via the esophagus, and 72 h later, Evans blue was injected into the orbital vein to assess the extent of liver sinusoidal barrier dysfunction (n = 5). (D, E) Representative histological images of liver tissue from mice stained with hematoxylin and eosin and quantification of bacterial liver injury area (n = 5); scale bar = 1 mm (whole liver lobe), 100 μm (magnified image). (F-H) plasma cytokine levels 72 h post-infection (n = 8). (I) Representative images showing neutrophil infiltration (Ly6G) and endothelial injury (VCAM-1) by immunohistochemical staining; scale bar = 200 μm. Non-throm: KPLA model mice with non-thrombophlebitis; throm: KPLA model mice with thrombophlebitis. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

These findings indicate that K. pneumoniae compromises the liver sinusoidal barrier, promotes the formation of thrombophlebitis, and exacerbates bacterial liver injury.

K. pneumoniae induces net formation in neutrophils, disrupting the hepatic sinusoidal barrier and causing thrombophlebitis

To further investigate the molecular mechanisms underlying K. pneumoniae-induced bacterial liver injury, RNA sequencing was performed to analyze differential gene expression in KPLA model mice. Compared with PBS-treated controls, 97 and 21 genes were significantly upregulated and downregulated, respectively, in KPLA model mice (Figure 3(A)). KEGG pathway analysis revealed that these differentially expressed genes (DEGs) were primarily involved in complement and coagulation cascades and neutrophil extracellular trap formation pathways (Figure 3(B)). Pathway enrichment analysis further mapped the interaction network, highlighting associations between NETs and the PI3K-AKT and Toll-like receptor signaling pathways (Figure 3(C)).

Figure 3.

Net formation in KPLA mice leads to endothelial injury. (A) differential gene expression between PBS-treated and KPLA model mice (n = 3). (B) kegg pathway enrichment analysis for differentially expressed genes. (C) network diagram depicting interactions between enriched pathways; ko04613: neutrophil extracellular trap formation; ko04151: PI3K-Akt signaling pathway; ko04620: toll-like receptor signaling pathway. (D) circulating cfDnA and Mpo levels in healthy adult volunteers, patients with KPLA with non-thrombophlebitis, and patients with KPLA with thrombophlebitis (n = 5). (E, F) Western blot analysis of PAD4 and CD31 in liver lysates from KPLA mice, with β-actin as a loading control, and quantification of relative protein levels. (G, H) correlation analysis between endothelial injury markers (ICAM-1, VCAM-1) and net markers (mpo) in the circulation of healthy volunteers and patients with KPLA (n = 15). Pearson correlation method was used for analysis. (I) immunofluorescence staining of liver sinusoidal sections from KPLA mice showing mpo (green), VCAM-1 (red), and dapi (blue); scale bar = 50 μm. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

To validate the role of NETs in endothelial injury among patients with KPLA, we analyzed plasma samples. Patients with intrahepatic thrombophlebitis exhibited significantly higher levels of NET (cfDNA and MPO) and endothelial injury (VCAM-1 and ICAM-1) markers (Figure 3(D)) than those without. In KPLA model mice, hepatic expression of protein-arginine deiminase type 4 (PAD4), a nuclear enzyme critical for citrullination during NET formation, was markedly increased, while that of CD31, an endothelial marker, was significantly reduced (Figure 3(E,F)). These findings underscore the importance of understanding the role of NETs in KPLA-related thrombophlebitis.

Correlation analysis of clinical samples further revealed a positive association between NET formation and endothelial injury in patients with KPLA (Figure 3(G,H)). Immunofluorescence staining for MPO and VCAM-1 demonstrated extensive NET deposition and significant endothelial damage in regions of thrombophlebitis (Figure 3(I)). Collectively, these results suggest that NET formation is associated with KPLA severity, potentially exacerbating disease progression through thrombosis and endothelial injury.

K. pneumoniae induces net formation in neutrophils in vitro

To investigate whether K. pneumoniae induces NET formation, we isolated highly active and pure neutrophils from the anticoagulated whole blood of healthy volunteers. Our experiments revealed that K. pneumoniae induced the formation of extracellular web-like structures in a time-dependent manner (Figure 4(A)). After 1 h of stimulation, immunofluorescence staining showed significant release of MPO and histone H3 from neutrophils (Figure 4(B)). Transmission electron microscopy further revealed marked morphological changes in neutrophil nuclei, including chromatin decondensation, separation of the inner and outer nuclear membranes, formation of DNA-containing vesicles, and vesicle rupture to release NETs (Figure 4(C)). These findings confirm that K. pneumoniae effectively induces NET formation in vitro.

Figure 4.

K. pneumoniae promotes net formation in neutrophils via the TLR/PI3K p110α/AKT pathway. (A) neutrophils were co-cultured with K. pneumoniae at different time points, and extracellular dna was detected using SYTOX green dye (scale bar = 100 μm). (B) Representative immunofluorescence images showing untreated neutrophils and neutrophils infected with K. pneumoniae; dna (blue), mpo (red), and histone H3 (green); scale bar = 50 μm. (C) transmission electron microscopy showing the process of net release in neutrophils infected with K. pneumoniae. a. Untreated neutrophil with a clear lobulated nucleus (scale bar = 2 μm); b. Chromatin decondensation in the nucleus (scale bar = 2 μm); c. Vesicles containing dna chains fusing with the cell membrane and releasing into the extracellular space (scale bar = 1 μm); d. Vesicle rupture releasing NETs (scale bar = 2 μm). (D-E) Western blot analysis of PI3K-AKT signaling pathway proteins in untreated and K. pneumoniae-infected neutrophils, with β-actin as loading control; n = 3. (F-G) Western blot analysis of PI3K-AKT pathway proteins in neutrophils pretreated with TLR4 inhibitor before infection with K. pneumoniae, with β-actin as loading control; n = 3. (H) immunofluorescence staining of neutrophils pretreated with TLR4 inhibitor and PI3K p110α inhibitor after K. pneumoniae stimulation, showing dna (blue), mpo (red), and histone H3 (green); scale bar = 50 μm. (I-J). Western blot analysis and quantification of PAD4 protein in neutrophils pretreated with TLR4 inhibitor and PI3K p110α inhibitor after K. pneumoniae stimulation, with β-actin as loading control; n = 3. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

To explore the molecular mechanisms underlying NET formation induced by K. pneumoniae, we conducted RNA sequencing analysis of liver tissues from KPLA model mice. LPS, a major component of the outer membrane of gram-negative bacteria, activates innate immune responses by binding to TLR4, a process facilitated by LBP and CD14 [27,28]. In KPLA mice, we observed significantly increased expression of LBP and CD14 genes, suggesting enhanced LPS-TLR4 interaction and activation of downstream signaling pathways. Pathway enrichment analysis and network mapping of DEGs highlighted a close association between NET formation and the PI3K-AKT signaling pathway. Previous studies have demonstrated that AKT is essential for shifting neutrophil fate from apoptosis to NETosis [18]. To pinpoint the upstream signaling pathways involved in K. pneumoniae-induced NET formation, we examined the expression of PI3K catalytic subunits (α, γ, δ, and σ) and the regulatory subunit (p85) in neutrophils stimulated with K. pneumoniae. Among these, PI3Kα expression was significantly increased (Figure 4(D,E)). Treatment of neutrophils with TLR4 and PI3Kα inhibitors markedly suppressed NET formation (Figure 4(F,J)). These findings suggest that K. pneumoniae promotes NET formation via the TLR4-PI3Kα-AKT signaling axis.

C3-dependent endothelial Cell injury mediated by KP-NETs

Transcriptomic analysis of liver tissues from KPLA mice revealed significant enrichment in complement and coagulation cascades, identifying C3 as a key DEG (Figure 3(A)). To investigate the role of C3 in KP-NETs, we isolated neutrophils from healthy donors and induced KP-NET formation. Immunofluorescence analysis demonstrated a significant deposition of C3 and its activation fragments C3b and C3c on KP-NETs (Figure 5(A)). Even after NET degradation by DNase I, western blot analysis confirmed high levels of C3 within the KP-NETs. However, treatment with the C3 inhibitor AMY-101 effectively reduced C3 deposition on KP-NETs (Figure 5(B,C)). HUVECs were treated with KP-NETs produced under various conditions. Results showed that KP-NETs with reduced C3 deposition due to AMY-101 treatment significantly attenuated HUVEC apoptosis and necrosis (Figure 5(D–F)), as well as the expression of VCAM-1 and ICAM-1 (Figure 5(G–J)). Conversely, while DNase I degradation of NETs also reduced VCAM-1 and ICAM-1 expression, it simultaneously impaired neutrophil bactericidal activity, leading to a marked increase in bacterial load (p < 0.05) (Figure 5(K)). This highlights a trade-off where NET degradation mitigates endothelial damage but compromises antimicrobial defense. Taken together, these findings underscore the critical role of C3 in KP-NET-induced endothelial cell injury and point to its dual impact on host defense and tissue damage.

Figure 5.

KP-NETs induce endothelial injury in a C3-dependent manner. (A) immunofluorescence images of human peripheral blood neutrophils infected with K. pneumoniae and pretreated with C3 inhibitor and DNase I; NETs (dna [blue] + histone H3 [green]), C3 (red). (B-C) Western blot analysis and quantification of C3 expression in NETs isolated from human peripheral blood neutrophils after a 2-h K. pneumoniae infection, pretreated with C3 inhibitor or DNase I, using mpo as a control. (D-F) treatment of human umbilical vein endothelial cells with NETs (1.5 μg/mL) for 24 h and analysis of cell apoptosis and necrosis rate using flow cytometry (n = 3). (G-J) treatment of human umbilical vein endothelial cells with NETs (1.5 μg/mL) for 24 h and analysis of endothelial injury markers (VCAM-1, ICAM-1) using flow cytometry (n = 3). (K) bacterial load in the supernatant of human peripheral blood neutrophils pretreated with C3 inhibitor and DNase I after K. pneumoniae infection (n = 3). *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

AMY-101 protects liver sinusoidal barrier function and improves outcomes in KPLA mice

To further validate the endothelial cytotoxicity of KP-NETs, we employed the KPLA mouse model. Among untreated KPLA mice, approximately 50% developed KPLA, with 50% of these cases exhibiting extrahepatic metastatic infections. AMY-101 treatment reduced liver abscess formation (from 50 to 25%) and completely prevented extrahepatic infections (from 50 to 0%), significantly improving survival rates (Figure 6(A,B)). Histological analysis revealed that untreated KPLA mice exhibited severe sinusoidal endothelial barrier dysfunction, intrahepatic thrombophlebitis, and extensive liver damage. In contrast, AMY-101-treated mice showed marked alleviation of sinusoidal endothelial injury, with no evidence of thrombophlebitis, effectively mitigating K. pneumoniae-induced bacterial liver damage (Figure 6(C,D)). Triple immunostaining for C3, VCAM-1, and MPO revealed significant C3 deposition on NETs in untreated mice, accompanied by endothelial injury (Figure 6(E)). This finding indicates that NET-borne C3 damages endothelial cells, promoting thrombophlebitis. In AMY-101-treated mice, NET-borne C3 deposition was significantly reduced, ameliorating endothelial damage. Importantly, AMY-101 significantly reduced extrahepatic dissemination and lung inflammatory infiltration (Figure 7(A)), alleviated lung tissue injury (Figure 7(B)), and markedly mitigated systemic inflammatory responses in the KPLA mouse model (Figure 7(C,D)).

Figure 6.

KP-NETs induce liver sinusoidal endothelial injury in a C3-dependent manner. (A) mice were infected with 2 × 10^7 cfu K. pneumoniae via the esophagus and treated with or without C3 inhibitor. KPLA: KPLA model mice with bacterial liver injury; non-KPLA: KPLA model mice without bacterial liver injury; KPLA with lung abscess: KPLA model mice exhibit bacterial liver injury accompanied by lung abscesses. (B) survival rates of KPLA mice treated with or without C3 inhibitor (n = 8 per group). (C) liver pathology scores of untreated and treated KPLA model mice (n = 8 per group). (D) Representative H&E staining and immunohistochemical images of Ly6G and VCAM-1 in KPLA mouse livers, comparing PBS-treated and C3 inhibitor-treated groups (whole liver lobe scale bar = 1 mm, magnified image scale bar = 100 μm). (E) Representative immunofluorescence images of KPLA mouse liver sinusoidal tissue showing dapi (blue), C3 (green), mpo (pink), and VCAM-1 (red); scale bar = 20 μm. *p < 0.05, **p < 0.01.

Figure 7.

C3 inhibition reduces extrahepatic infection and alleviates systemic inflammatory response in KPLA mice. (A) Representative H&E and Ly6G immunohistochemical staining of lung tissue from KPLA mice treated with or without C3 inhibitor following K. pneumoniae infection (whole lung lobe scale bar = 1 mm, magnified image scale bar = 50 μm). (B) lung pathology scores (n = 8 per group). (C-D) cytokine levels in the circulation of KPLA model mice 72 h post-infection, with or without C3 inhibitor treatment (n = 8 per group). *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

In summary, despite the absence of antibiotics to control K. pneumoniae, reducing C3 deposition on NETs effectively alleviated KP-NET-induced endothelial cytotoxicity, reduced thrombosis, and prevented extrahepatic infections, ultimately improving survival rates in KPLA mice. These findings highlight that targeting NET-borne C3 is a potential therapeutic strategy that does not compromise the neutrophil antimicrobial activity. This discovery offers a promising direction for the treatment of infectious diseases.

Discussion

Researchers initially focused on how NETs capture microorganisms, but their potentially harmful effects have increasingly garnered attention. When NETs form in the circulatory system, they can promote blood clotting, block blood vessels, and trigger thrombosis. Our findings show that patients with KPLA frequently develop blood clots in the liver, which are linked with severe symptoms and a high risk of sepsis and extrahepatic infections. In this study, we demonstrate that K. pneumoniae induces NET formation by activating the TLR4-PI3Kα-AKT signaling pathway. Additionally, we found that C3 deposited on these NETs damages the inner lining of blood vessels and disrupts the liver sinusoidal barrier, further exacerbating liver injury in KPLA. Furthermore, the C3 inhibitor AMY-101 reduces C3 deposition on NETs and significantly protects the vascular structure of the liver, without compromising the infection-fighting capability of neutrophils. These results suggest that NETs and their associated C3 could serve as valuable biomarkers for the early detection and treatment of KPLA-related metastatic infections.

In addition to promoting blood clotting, endothelial damage is closely linked to systemic inflammation and multiple organ failure [29,30]. Preserving the integrity of blood vessel walls is essential for preventing sepsis progression, with recent studies focusing on ways to prevent or reverse vascular dysfunction [31,32]. Our findings underscore that patients with KPLA who develop liver clots often experience severe endothelial damage, highlighting the need to understand the underlying mechanisms to slow disease progression.

Having confirmed the critical role of liver thrombophlebitis in KPLA, we investigated how K. pneumoniae induces endothelial damage. Our analyses revealed that genes associated with NET formation were significantly upregulated in a mouse model of KPLA. Moreover, K. pneumoniae induced NET formation by neutrophils, which helps contain pathogens and control infection, consistent with prior research [33–40]. In our study, disrupting NETs during infection weakened the bacteria-killing function of neutrophils. However, NETs and their components also have considerable toxic effects on cells, particularly in damaging the endothelium and promoting clot formation [12,41,42]. Furthermore, in patients with KPLA, a positive correlation exists between NET formation and endothelial injury. Accordingly, clarifying how K. pneumoniae induces NET formation may help refine our understanding of KPLA pathogenesis.

The upstream signaling pathways regulating NETosis are not entirely understood. Under stimulation by phorbol myristate acetate (PMA), various microbes, immune complexes, and multiple kinases – including MEK (MAPK/ERK kinase), extracellular signal-regulated kinase (ERK), IL-1 receptor-associated kinase (IRAK), phosphatidylinositol 3-kinase (PI3K), and AKT – participate in NET formation [43–45]. In bacterial infections, sterile inflammation, and cancer, TLR4 promotes neutrophil NET formation and contributes to disease progression [46–48]. As a gram-negative bacterium, K. pneumoniae possesses LPS on its outer membrane, which can provoke inflammation by interacting with TLR4 [49,50]. In our study, NET formation in liver tissues of a KPLA mouse model was closely linked to TLR signaling and the PI3K/AKT pathway. Furthermore, LBP and CD14 were strongly expressed in the liver, suggesting that LPS more readily binds TLR4 during K. pneumoniae infection. A key signal for shifting neutrophils from apoptosis to NET formation is AKT activation [18]. Further analysis of PI3K, which acts upstream of AKT, revealed a significant upregulation of the catalytic subunit p110α. Targeting TLR4 or PI3K p110α reduced NET formation. Because NET formation is associated with thrombosis and endothelial dysfunction, inhibiting NET formation or blocking its upstream pathways may provide a promising therapeutic strategy for patients with KPLA [51,52].

We aimed to develop an approach that would preserve pathogen-killing ability of NETs while mitigating their toxic effects on the endothelium because they have a critical role in combating infections. By examining genetic changes in our KPLA mouse model, we found that complement C3 was significantly overexpressed. Previous research highlighted the central role of the complement system, particularly C3, in driving inflammation during infections [53–55]. Large amounts of complement can accumulate on NETs in conditions such as sterile inflammation and autoimmune disorders, worsening tissue damage [56]. We confirmed that C3 and its activated fragments accumulate extensively on NETs induced by K. pneumoniae, and this C3 accumulation closely correlates with endothelial damage. Blocking C3 deposition on NETs reduced endothelial damage, alleviated clotting in the liver, mitigated liver injury, decreased the risk of lung abscesses, and improved the disease course in our mouse model. This approach preserved the infection-fighting function of NETs while reducing their harmful effects on the endothelium, offering promising therapeutic potential.

This study has several limitations. First, although our mouse model provides insight into how NETs and NET-bound C3 May contribute to intrahepatic thrombophlebitis in KPLA, species differences limit direct extrapolation, and additional clinical studies are needed. Second, the C3 inhibitor used here is not neutrophil-specific; potential immunologic off-target effects require further assessment. Finally, antibiotics were not used in our animal studies, which differs from clinical practice and may affect assessment of NETs and NET-bound C3. Future work will incorporate standard antibiotic regimens to test whether NET-driven endothelial injury and thrombophlebitis remain significant under clinically relevant treatment.

In conclusion, our results show how K. pneumoniae triggers NET formation and how C3 leads to endothelial injury and thrombosis once deposited on NETs. Thus, we propose a potential strategy to prevent C3 accumulation on NETs by targeting C3, possibly halting disease progression while preserving the pathogen-fighting properties of NETs, thereby providing a foundation for developing targeted therapies.

Materials and methods

Patient and sample collection

This retrospective study collected data from 357 consecutive patients diagnosed with liver abscesses at Shengjing Hospital of China Medical University from September 2012 to September 2022. Inclusion criteria were as follows: (1) diagnosis of pyogenic liver abscess (PLA) based on contrastenhanced computed tomography scan upon admission; and (2) isolation of K. pneumoniae strains from either pus or blood culture. Patients were excluded if they: (1) immunodeficiency or long-term immunosuppressive therapy, (2) concomitant severe liver diseases (e.g. decompensated cirrhosis, HCC), and (3) incomplete clinical data. Among them, 261 patients were diagnosed with KPLA without thrombophlebitis, and 96 exhibited intrahepatic thrombophlebitis. Peripheral blood samples were collected between January 2023 and June 2023 from five healthy volunteers, five patients with KPLA without thrombophlebitis, and five patients with intrahepatic thrombophlebitis. Clinical data and peripheral blood samples from patients with KPLA were approved by the Ethics Committee of Shengjing Hospital of China Medical University (No. 2022PS146K). It was carried out by the ethical standards laid down in the 1964 Declaration of Helsinki and its later amendments.

Mice

Male C57BL/6J wild-type mice (8 weeks old) were purchased from Beijing Huafukang Biotechnology Co., Ltd. The mice were housed at 23 ± 3 °C with 35 ± 5% humidity under a 12-h light/dark cycle and had free access to food and water. All animal procedures were approved by the Ethics Committee of Shengjing Hospital of China Medical University (No. 2022PS434K), and complied strictly with the ARRIVE guidelines. The ARRIVE checklist has been deposited in a public repository (https://doi.org/10.6084/m9.figshare.28639523.v7).

HUVEC Cell line

Human umbilical vein endothelial cells (HUVECs; Cat: CL-0675, Pricella, China) were purchased from Wuhan Pricella Biotechnology Co., Ltd., and cultured in a specialized medium at 37 °C and 5% CO₂.

K. pneumoniae infection

K. pneumoniae strains used in this study were isolated from a patient with KPLA who had intrahepatic thrombophlebitis and extrahepatic dissemination, including pulmonary abscess. For the KPLA mouse model, mice were infected with 2 × 10^7 CFU K. pneumoniae via the esophagus. In the treatment group, mice were subcutaneously injected with AMY-101 acetate at 4 mg/kg (Cat: HY-P1717B, MedChemExpress, USA) 30 min before infection. Mice that lost > 20% body weight or had a core body temperature below 32 °C were considered moribund. Mice surviving after 72 h were euthanized for further experiments. All mice undergoing surgery or invasive procedures were anesthetized with isoflurane (CAS 26,675–46-7; Sigma-Aldrich, Germany) delivered in oxygen via a precision vaporizer; anesthetic depth was continuously monitored (loss of pedal/palpebral reflexes) to ensure animals remained pain-free. At prespecified or humane endpoints, animals were euthanized under deep anesthesia using CO₂ inhalation with a gradual fill rate (≈10–30% chamber volume/min) or by barbiturate overdose (CAS 4390–16-3; Sigma-Aldrich, Germany), in accordance with the AVMA Guidelines for the Euthanasia of Animals.

RNA-seq analysis

Total RNA was extracted from fresh frozen liver tissues of mice using TRIzol reagent (n = 3 mice per group). Eukaryotic mRNA was enriched using magnetic beads with Oligo(dT), fragmented by ultrasound, and used as a template for cDNA synthesis. cDNA libraries were prepared, and sequencing was performed on the Illumina Novaseq 6000 platform. Library quality was assessed by agarose gel electrophoresis to analyze RNA integrity and DNA contamination and NanoPhotometer spectrophotometry for RNA purity (OD260/280 and OD260/230 ratios) and Qubit2.0 Fluorometry for RNA concentration. RNA integrity was assessed using an Agilent 2100 bioanalyzer.

Neutrophil isolation and net stimulation

Neutrophils were isolated from healthy volunteers using a Neutrophil Isolation Kit (Cat: LZS11131, Tianjin Haoyang Bio, China). The purity of the neutrophils ( > 95%) was confirmed by an automatic blood cell analyzer. K. pneumoniae and neutrophils were incubated in a 10:1 ratio to assess NETosis induced by K. pneumoniae. NETs were stained with 100 nM SYTOX Green (Cat: KGE2503-500, KeyGEN BioTECH, China), and the time-dependent induction of NETs was assessed from 0 to 60 min. To investigate the molecular mechanisms, neutrophils were co-incubated with K. pneumoniae and inhibitors, including TLR4 inhibitor (100 nM TAK-242, Cat: HY-11109, MedChemExpress, USA), PI3K p110α inhibitor (1 μM BYL-719, Cat: HY-15244, MedChemExpress, USA), DNase I (40 U, Cat: D8071, Solarbio, China), and C3 peptide inhibitor AMY-101 acetate (5 μM, Cat: HY-P1717B, MedChemExpress, USA). Immunofluorescence (IF) staining was used to detect characteristic NET markers, such as myeloperoxidase (MPO) and histone H3.

Net purification

K. pneumoniae and neutrophils were incubated for 2 h at a 10:1 ratio, with AMY-101 and DNase I added to the different groups. Supernatants were collected, and NETs were digested with 10 U/mL Micrococcal nuclease (Cat: EN0181, Thermo Scientific, USA) at 37 °C for 15 min. The resulting NETs were collected by centrifugation at 3,000 g for 5 min at 4 °C and stored at −20 °C for future use. NET concentration was determined by measuring free DNA.

Elisa

Plasma concentrations of NET biomarkers, endothelial injury markers, and circulating cytokines were measured using commercial ELISA kits for MPO (Cat: SEKH-0262, Solarbio, China), ICAM-1 (Cat: SEKH-0053, Solarbio, China), VCAM-1 (Cat: SEKH-0055, Solarbio, China), vWF (Cat: SEKH-0343, Solarbio, China), IL-6 (Cat: RK00008, ABclonal, China), IL-1β (Cat: RK04878, ABclonal, China), and TNFα (Cat: RK00027, ABclonal, China), according to the manufacturer’s instructions.

Transmission electron microscopy

Neutrophils were incubated with K. pneumoniae at a 10:1 ratio for 1 h and then fixed with an electron microscope fixative for 2 h. After dehydration, infiltration, embedding, and sectioning, samples were stained with uranyl acetate and lead citrate for 15 min each. Finally, the samples were observed under a JEM-F200 transmission electron microscope.

Histological staining

Liver and lung tissues were fixed in formalin for 24 h, embedded in paraffin, sectioned to 4 μm thickness, and stained with hematoxylin and eosin (H&E). Tissue injury in the KPLA model mice was scored from 0 to 3 based on the extent of thrombosis, inflammation, and necrosis [57].

Immunofluorescence and immunohistochemistry

Neutrophils were incubated with K. pneumoniae for 2 h and fixed with 4% formaldehyde. Permeabilization was performed with 0.5% Triton X-100, followed by blocking with 5% goat serum. Primary antibodies, including anti-Histone H3 (Cat: 14,269, Cell Signaling Technology, USA), anti-MPO (Cat: 15,178, Cell Signaling Technology, USA), anti-ZO-1 (Cat: 21,773–1-AP, Proteintech, China), anti-E-cadherin (Cat: 20,874–1-AP, Proteintech, China), and anti-C3 (Cat: 66,157–1-Ig, Proteintech, China), were applied overnight at 4 °C. Alexa Fluor-conjugated secondary antibodies were incubated at room temperature for 2 h, and cells were stained with DAPI for nuclear visualization. For tissue immunofluorescence, liver tissue was fixed, dehydrated, and embedded in paraffin, and antigen retrieval was performed using citrate buffer.

Flow cytometry

HUVECs were treated with 1.5 μg/mL purified NETs for 24 h to model endothelial injury. Cell apoptosis and necrosis were assessed by Annexin V-FITC/PI staining (Cat: KGA1102-100, KeyGEN BioTECH, China). HUVEC injury was measured by staining ICAM-1 (Cat: 65,567–1-MR, Proteintech, China), VCAM-1 (Cat: 39,036, Cell Signaling Technology, USA), and mouse IgG (H+L) Alexa Fluor 488 (F(ab’)2 Fragment). Flow cytometry was performed using a BD Flow Cytometer, and data were analyzed with FlowJo V.10 software (TreeStar, Ashland, USA).

Western blotting

Total protein was extracted using RIPA buffer (Cat: R0010, Solarbio, China) supplemented with protease and phosphatase inhibitors (Cat: P0100, P8990, Solarbio, China). Proteins were separated using 8 or 12% polyacrylamide gel electrophoresis (PAGE) and transferred onto PVDF membranes. The membranes were blocked with 5% nonfat milk before being probed with primary antibodies. Primary antibodies included anti-PI3 Kinase p110α (Cat: 4249, Cell Signaling Technology, USA), anti-PI3 Kinase p110δ (Cat: 34,050, Cell Signaling Technology, USA), anti-PI3 Kinase p110γ (Cat: 5405, Cell Signaling Technology, USA), anti-PI3 Kinase p110β (Cat: 3011, Cell Signaling Technology, USA), anti-PI3 Kinase p85 (Cat: 4257, Cell Signaling Technology, USA), anti-phospho-AKT (Cat: 4060, Cell Signaling Technology, USA), anti-AKT (Cat: 4691, Cell Signaling Technology, USA), anti-MPO (Cat: 15,178, Cell Signaling Technology, USA), anti-C3 (Cat: 66,157–1-Ig, Proteintech, China), anti-CD31 (Cat: abs119772, Absin, China), anti-PAD4 (Cat: ab214810, ab315269, Abcam, UK), and anti-β-actin (Cat: ABL1010, Abbkine, China). The secondary antibodies included goat anti-mouse and anti-rabbit IgG HRP (Cell Signaling Technology, USA). Protein bands were quantified using ImageJ (NIH Image, Bethesda, MD).

Statistical analysis

Statistical analysis was performed using GraphPad Prism 9 (USA; San Diego). Unpaired Student’s t-test was used to compare means between two independent groups when data followed a normal distribution. If normality assumptions were not met, a Mann – Whitney U test was applied. All t-tests shown in the manuscript were two-sided. One-way ANOVA was applied for comparisons between more than two groups when data followed a normal distribution, with Tukey’s or Dunnett’s multiple comparison correction. For non-parametric data, a Kruskal – Wallis test with Dunn’s multiple comparison test was used. Correlations were assessed using Pearson’s correlation analysis. A p-value < 0.05 was considered statistically significant.

Funding Statement

This study was supported by the National Natural Science Foundation of China [82272097].

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The data that support the findings of this study are openly available in figshare at https://doi.org/10.6084/m9.figshare.28639523.v7, reference number [58].

Supplemental data

Supplemental data for this article can be accessed online at https://doi.org/10.1080/21505594.2025.2580104