Neutrophil extracellular traps induce intrahepatic thrombotic tendency and liver damage in cholestatic liver disease

Abstract

Background:

Cholestatic liver diseases induce local and systemic hypercoagulation, with neutrophil extracellular traps (NETs) serving as major drivers. These NETs have been linked to decreased liver function in patients with obstructive jaundice. However, the impact of NETs on liver hypercoagulation in cholestatic liver disease remains unknown.

Methods:

We utilized bile duct ligation to create experimental mice and analyzed NETs formation in the liver. Fibrin deposition, tissue factor expression, and inflammation in the liver were visualized through western blot and immunohistochemical techniques. LSECs were incubated with isolated NETs, and we detected endothelial procoagulant activity using coagulation protein production assays and measuring endothelial permeability. In both in vivo and in vitro settings, DNase I was applied to clarify the effect of NETs on intrahepatic hypercoagulability, hepatotoxicity, LSEC, and macrophage activation or injury.

Results:

Bile duct ligation mice exhibited significantly increased levels of NETs in liver tissue, accompanied by neutrophil infiltration, tissue necrosis, fibrin deposition, and thrombophilia compared to sham mice. Notably, NETs resulted in phosphatidylserine and tissue factor exposure on LSEC, enhancing coagulation Factor Xa and thrombin production. The enhanced procoagulant activity could be reversed by degrading NETs with DNase I. Additionally, NETs-induced permeability changes in LSECs, characterized by increased VE-cadherin expression and F-actin retraction, which could be rescued by DNase I. Meanwhile, NET formation is associated with KC activation and the formation of inflammatory factors.

Conclusions:

NETs promote intrahepatic activation of coagulation and inflammation, leading to liver tissue injury. Strategies targeting NET formation may offer a potential therapeutic approach for treating cholestatic liver disease.

INTRODUCTION

Cholestasis, a complex pathological state, is characterized by compromised bile flow, leading to the accumulation of bile acids and harmful substances within the liver, thereby initiating liver damage.¹ Numerous studies indicate an association between cholestatic liver diseases (CLD) and an overall procoagulant state^2,3 coupled with impaired fibrinolysis,^4,5 resulting in poor clinical outcomes, such as disseminated intravascular coagulation and multiple organ dysfunction.^6,7 A recent study demonstrated that administering anticoagulant treatment to cholestatic mice can mitigate liver damage.⁸ However, routine anticoagulant treatment carries a bleeding risk, particularly in liver disease.⁹ Consequently, there is an increasing urgency to elucidate the underlying mechanisms of hypercoagulation in CLD.

CLD may facilitate bacterial translocation into the portal circulation, contributing to immune dysfunction and chronic activation of liver immune cells in response to endotoxin exposure.^10,11 The excessive activation of neutrophils, accompanied by delayed clearance and infiltration into the liver, is thought to play essential roles in the pathogenesis of CLD.^12,13 DNA scaffolds-composed neutrophil extracellular traps (NETs) that also contain histones and granular proteases are released during this process.^14,15 Although NETs play a crucial role in defending the host against pathogens, their excessive formation has been linked to coagulopathy and thrombogenicity in various disease models, including sepsis,¹⁶ deep venous thrombosis,¹⁷ and cancers.¹⁸ Elevated levels of NETs have been correlated with poor prognosis in liver diseases, such as NASH¹⁹ and HCC.²⁰ Our previous study indicated increased NET levels in the circulation of patients with CLD due to heightened bilirubin, with these NETs amplifying circulation coagulation activity in patients with CLD.²¹ In CLD, accumulating large bilirubin in the liver prompts activated KCs to release chemotactic factors, recruiting neutrophils. However, the presence of NETs in the liver in CLD and their role in promoting liver coagulation activity remain undetermined.

Thrombosis and inflammatory mechanisms are critical for perpetuating cholestatic liver injury.^22,23 In addition to hepatocytes, nonparenchymal liver cells, such as endothelial cells, KCs, and infiltrating immune cells, are also involved in this process. Studies have demonstrated that elevated tissue factor (TF) expression on hepatocytes in bile duct ligation (BDL) mice is a significant cause of intrahepatic coagulation.²⁴ However, the increased coagulation activity of hepatocytes alone cannot explain the predominant fibrin deposition within hepatic sinusoids in CLD. Furthermore, CLD is associated with endothelial cell damage and increased procoagulant activity of damaged endothelium.²⁵ However, the impact of the interaction between NETs and LSECs on the development of coagulopathy in CLD remains unclear. An increase in VE-cadherin and intercellular adhesion molecule-1 (ICAM-1) levels was reported in LSECs under chronic inflammation,^26,27 coinciding with impaired transport function between the blood and liver parenchyma. However, whether NETs play a role in the permeability dysfunction of LSECs remains uncertain. CLD patients and experimental cholestatic models are characterized by macrophage infiltration and activation.^28,29 Increased levels of inflammatory cytokines and macrophage activation triggered by NETs have been observed in atherosclerosis and systemic lupus erythematosus.^30,31 These findings prompted us to investigate the interaction between NETs and macrophages, exploring their influence on the cytokine burden, thereby mediating both thrombogenesis and the pathogenesis of CLD.

In this study, we demonstrate that mice with experimental cholestasis exhibit heightened formation of NETs within the liver. Furthermore, NETs contribute to LSEC injury and KC activation, fostering a thrombotic tendency and inflammation in the liver. This exacerbates liver fibrosis and damage in cholestatic mice. Our findings contribute to identifying potential mechanisms underlying CLD-associated coagulopathy and liver injury, offering insights into novel therapeutic targets for the intervention and prevention of CLD.

METHODS

Animals

Wild-type (C57BL/6) male mice (8–12 wk old) were procured from Zhejiang Animal Technology Corporation and housed in ventilated cages under a 12-hour light/dark cycle. The mice had unrestricted access to environmental enrichment, water, and food. All mice experiments were approved by the Animal Experiment Administration Committee of Jiaxing University.

BDL surgery

The mouse model for BDL-induced cholestatic liver fibrosis followed a previously described protocol.³² The common bile duct was isolated and ligated during laparotomy under ketamine/xylazine anesthesia. Sham mice underwent the same surgical incision but without ligation. For in vivo neutrophil depletion, mice received either an anti-neutrophil antibody (anti-Ly6G, 1A8clone, Biolegend, cat. no. 127632) or isotype antibody with a 5 μg/g dose 1 day before BDL. Intraperitoneal injections were administered every 2 days after the initial injection at a 2.5 μg/g dose until sacrifice. To study the impact of the degradation of NETs on the progression of CLD, mice were i.p. injected with DNase I (diluted in saline; New England Biolabs; 5 μg/g) every 2 days from the start of BDL until the end, with saline serving as the positive control.

Sample collection

At 3, 7, and 14 days after bile duct ligation, mice were anesthetized with isoflurane, and then, blood and liver samples were collected. For detailed information, please refer to SDC, Materials and Methods, Supplemental Materials, http://links.lww.com/SLA/F197.

Neutrophils and myeloperoxidase activity measurement

For detailed information on neutrophil measurement in blood and myeloperoxidase (MPO) activity measurement in the liver, refer to SDC, Materials and Methods, Supplemental Materials, http://links.lww.com/SLA/F197.

Biochemical assays of blood samples

Serum activities of aspartate aminotransferase (AST) and alanine aminotransferase (ALT) were quantified using commercially available assay kits (C009-2-1 and C010-2-1for ALT and AST, respectively, Nanjing Jiancheng Bioengineering Institute).

Liver histopathology

A portion of the liver samples was utilized for hematoxylin and eosin staining and Sirius red solutions. For detailed information, please refer to SDC, Materials and Methods, Supplemental Materials, http://links.lww.com/SLA/F197.

Measurement of NET-DNA complexes

Quantification of NET-DNA complexes was performed using modified ELISA kits. For detailed information, refer to SDC, Materials and Methods, Supplemental Materials, http://links.lww.com/SLA/F197.

Detection of NETs in the liver

Liver tissue sections were deparaffinized and then stained using antibodies against MPO and citrullinated histone H3 (CitH3) to visualize NETs. Immunofluorescent images were observed and recorded using confocal microscopy. For detailed information, refer to SDC, Materials and Methods, Supplemental Materials, http://links.lww.com/SLA/F197.

Western blot analysis

Western blotting was performed using 10–40 μg of liver extract protein and primary antibodies. For detailed information, refer to SDC, Materials and Methods, Supplemental Materials, http://links.lww.com/SLA/F197.

Quantitative real-time PCR

Total RNA from liver tissues was used for PCR assay. For detailed information, refer to SDC, Materials and Methods, Supplemental Materials, http://links.lww.com/SLA/F197.

Thrombin/antithrombin complex formation

The levels of thrombin/antithrombin (TAT) complexes in plasma samples were measured using the TAT Complexes ELISA kit (abcam, ab137994) according to the provided instructions.

Immunohistochemistry

Formalin-fixed tissue was analyzed after being embedded and then sectioned at 5 μm, and performed using the antibodies. For detailed information, refer to SDC, Materials and Methods, Supplemental Materials, http://links.lww.com/SLA/F197.

Immunofluorescent staining

To visualize TF in the liver, the sections were stained with anti-TF (Bioss, bs-4690R) and anti-CD31 antibodies (Abcam, ab7388). LSECs cultured on fibronectin-coated slides were stained with anti-VE-cadherin (Invitrogen, 14-1449-82), anti-TF (Bioss, bs-4690R), FITC-conjugated lactadherin (Hematologic Technologies), Propidium iodide (BD Biosciences), and TRITC-conjugated phalloidin (Yeasen). The secondary antibodies were goat anti-rabbit conjugated to Alexa Fluor 488 (Invitrogen, A-11008), donkey anti-rat conjugated to Alexa Fluor 594 (Invitrogen, A-21209), Donkey anti-mouse conjugated to Alexa Fluor 488 (Invitrogen, A-21202), and Goat anti-mouse Alexa Fluor 594 (Invitrogen, A-11005). All samples were analyzed by confocal microscopy (Olympus).

NETs generation, isolation, and quantification

NETs were generated, isolated, and quantified as before, with minor adjustments. For detailed information, refer to SDC, Materials and Methods, Supplemental Materials, http://links.lww.com/SLA/F197.

LSEC stimulation assays

The LSECs were seeded into a 12-well plate. To detect the cytotoxicity of NETs on LSECs (CP-M040, Procell, Wuhan, China), LSECs (1 × 10⁵ cells/mL) were incubated for 4 hours with or without isolated NETs (0.5 μg DNA/mL)³³. For inhibition assays, isolated NETs were pretreated with DNase I (100 U/mL, Termo Fisher) and then added to the supernatant of LSECs. At indicated time points, LSECs were centrifuged and then washed twice with PBS before the start of the experiments. Then, we detected TF and phosphatidylserine (PS) exposure by flow cytometer and performed intrinsic coagulation Factor Xa, extrinsic coagulation Factor Xa, and prothrombinase formation assays as described.¹⁸ Vascular cell adhesion molecule-1 (VCAM-1), ICAM-1, and VE-cadherin expression were analyzed using western blotting.

Endothelium permeability assay

Endothelium permeability assays are described in detail in the SDC, Materials and Methods, Supplemental Materials, http://links.lww.com/SLA/F197.

Macrophages stimulation assay

Macrophage stimulation assays are described in detail in the SDC, Materials and Methods, Supplemental Materials, http://links.lww.com/SLA/F197.

Statistical analysis

Quantitative data are presented as mean±SEM. Data normality was assessed using the Kolmogorov-Smirnov test. Statistical differences between the 2 groups were analyzed using either an unpaired 2-tailed Student’s t test or the Mann-Whitney U test, depending on whether the variables exhibited a normal distribution or not. Group differences were analyzed using one-way ANOVA with Tukey multiple-comparison test or Kruskal-Wallis with Dunn’s multiple-comparison test. Pearson or Spearman correlation analyses were employed to assess the relationships between NETs and ALT or AST, as well as between NETs and TAT. GraphPad Prism 9.0 was utilized for all analyses. A significance level of P < 0.05 was applied to all statistical comparisons.

RESULTS

Neutrophils induce liver damage through increased NET formation in cholestatic mice

We constructed a cholestasis model using BDL and demonstrated that BDL led to a significant increase in total bilirubin and liver injury in BDL mice on postoperative day (POD)-14. Additionally, there was an elevated expression of neutrophils in circulation on POD-7 and in liver tissue on POD-3 in BDL mice (Table 1). Immunohistochemical images and quantified results showed that the number of liver-infiltrating neutrophils was significantly increased in BDL mice compared to sham livers on POD-14 (Fig. 1A, C). Sham-operated controls exhibited a normal hepatic architecture in HE-stained liver sections. In contrast, cholestatic mice showed marked destruction of liver tissue with extensive necrosis on POD-14 (Fig. 1B). BDL was also associated with severe hepatocellular damage, as evidenced by a notable elevation of ALT and AST activities (190.6 ± 26.4 and 109.1 ± 14.2 IU/L, respectively) relative to those observed in the sham mice (28.2 ± 6 IU/L and 31.5 ± 8 IU/L, respectively) (Fig. 1E). Considering the exhaustion of neutrophils might be effective in alleviating liver damage, we performed neutrophil depletion by injecting anti-Ly6G antibody into BDL mice. The anti-Ly6G antibody administration significantly decreased neutrophil counts in the liver on POD-14 (Fig. 1A, C). However, unexpectedly, pretreatment with the anti-ly6G can only reduce liver damage by about 35%~50%, as indicated by the serum levels of ALT and AST (Fig. 1E), as well as liver histology (Fig. 1B, D). Therefore, we hypothesized that neutrophils may participate in cholestatic liver damage through other forms. To identify mechanisms of neutrophil-mediated liver injury, we evaluated NET formation. First, we assayed the concentrations of CitH3-DNA, MPO-DNA, neutrophil elastase–DNA, and cell free-DNA in plasma. The 4 markers of NETs were markedly elevated in cholestatic mice (Supplemental Figure S1, http://links.lww.com/SLA/F198). Western blot analysis showed a significant increase in CitH3 levels in liver tissues from BDL mice on POD-14, which was almost absent in BDL mice on POD-3 or sham mice (Fig. 1F). Then, we observed NETs formation in BDL mice on POD-14 as evidenced by increased colocalization of CitH3 and MPO (Fig. 1G). Notably, we observed a significant positive correlation between the levels of NETs and liver injury (Fig. 1H).

TABLE 1.

Characteristics of mice after sham and BDL operation

	Sham mice			BDL mice
	POD-3	POD-7	POD-14	POD-3	POD-7	POD-14
Total bilirubin (mg/dL)	1.0 ± 0.2	1.0 ± 0.2	1.1 ± 0.3	5.8 ± 1	15.4 ± 2.0a	25.4 ± 2.9b
ALT (IU/L)	9.8 ± 1.2	9.6 ± 1.2	9.8 ± 0.8	116.4 ± 14.1c	157.2 ± 19.4c	190.6 ± 26.4c
Neutrophils in blood (%)	18.4 ± 2.1	18.6 ± 1.3	19.8 ± 1.9	28.8 ± 2.4	36.2 ± 6.1a	51.6 ± 11.1d
MPO activity in liver (fold change)	0.046±0.004	0.045±0.005	0.047±0.004	0.21±0.034b	0.34±0.03c	0.43±0.05c

Note: Values are expressed as mean±SEM (n = 5).

^a P < 0.05.

^b P < 0.001.

^c P < 0.0001 versus POD-14 (Sham mice) by 1-way ANOVA or Kruskal-Wallis test.

^d P < 0.01.

Abbreviations: ALT, alanine aminotransferase; BDL, bile duct ligation; MPO, myeloperoxidase; POD, postoperative day.

FIGURE 1.

Neutrophils induce liver damage through increased NET formation in cholestatic liver disease. (A) Liver-infiltrating neutrophils on POD-14 with or without anti-Ly6G by immunohistochemical staining with MPO (original magnification, ×200). Black arrows indicate MPO+ neutrophils. (B) Liver sections on POD-14 were stained with hematoxylin to assess liver necrosis (original magnification, ×100). (C) Neutrophil count per field was quantified. (D) Necrotic areas (%) were quantified by Image J. (E) Concentrations of ALT and AST were measured in serum samples on POD-14. (F) Western blot analysis revealed the levels of CitH3 in liver tissues. (G) Representative immunofluorescence images showing the structure of NETs in liver tissue on POD-14 (original magnification, ×40). (H) ALT or AST were strongly correlated with CitH3-DNA or MPO-DNA in cholestatic mice on POD-14. Data are presented as the mean±SEM. ^*** P < 0.001; ^**** P < 0.0001 by 1-way ANOVA. r values were determined using Pearson or Spearman rank correlation. Abbreviations: ALT, alanine aminotransferase; AST, aspartate aminotransferase; BDL, bile duct ligation; CitH3-DNA, citrullinated histone H3-DNA; MPO, myeloperoxidase; POD, postoperative day.

NET degradation protects cholestatic mice against intrahepatic neutrophils infiltration and intrahepatic hypercoagulation

Since previous studies showed that DNase I could inhibit NET formation.³⁴ It is worth mentioning that the neutrophil number was also reduced in BDL mice upon DNase I treatment on POD-14 (Fig. 2A), suggesting that a decrease of the hepatic NETs level might be owing in part to altered neutrophil numbers in BDL mice. In addition, we also assessed neutrophil-recruiting chemokines in the liver on POD-14. The expression of neutrophil-attracting chemokines, such as chemokine (C-X-C motif) ligand 1, chemokine (C-X-C motif) ligand 2, and chemokine (C-X-C motif) ligand 5, were markedly increased in BDL mice, whereas the expression of these chemokines was reduced in DNase I treatment mice (Supplemental Figure S2, http://links.lww.com/SLA/F198). To investigate whether NETs contribute to BDL-induced liver injury through promoting coagulation, we examined plasma TAT complex levels, liver TF, and fibrin deposition on POD-14. Elevated plasma TAT levels were observed in BDL mice compared to the sham group, suggesting activation of the coagulation cascade (Fig. 2B). Moreover, the levels of TAT were significantly positively correlated with the levels of NETs (Fig. 2C, D). Minimal TF and fibrin deposition were observed in sham mice. In contrast, a significant elevation of sinusoidal TF and fibrin deposits was observed in BDL mice (Fig. 2E-I). Consistent with this observation, serum ALT and AST levels, as well as liver necrotic lesions area, were significantly elevated in BDL mice on POD-14 compared to sham mice (Fig. 2J–M). To further confirm NET-mediated liver injury, we inhibited NETs by injecting DNase I into mice after BDL. Pretreatment with DNase I decreased serum ALT activity to 30.6 ± 4.2 IU/L and AST levels to 41.8 ± 5.7 IU/L, corresponding to an 84% and 59% reduction, respectively (Fig. 2J, K). Meanwhile, inhibition of NETs through DNase I reduced BDL-induced liver necrosis by 78%, as indicated by HE (Fig. 2L, M). DNase I administration significantly decreased NET formation in the liver and attenuated NET-induced coagulation and liver damage. These results collectively indicate that NETs are critically involved in intrahepatic coagulation and liver injury in cholestatic mice.

FIGURE 2.

NET degradation protects cholestatic mice against intrahepatic neutrophils infiltration and intrahepatic hypercoagulation. (A) Liver-infiltrating neutrophils on POD-14 with or without DNase I by immunohistochemical staining with MPO (original magnification, 200×). Black arrows indicate MPO+ neutrophils. (B) Plasma levels of TAT on POD-14 were determined by ELISA. TAT complexes were strongly correlated with MPO-DNA (C) or CitH3-DNA (D) in BDL mice. (E) Levels of liver tissue factor (TF) and fibrin were determined by western blotting. Immunohistochemistry staining for TF (F) and fibrin (G) on POD-14 were performed (original magnification, 100×) and quantified (H, I). Black arrows indicate TF and fibrin deposition, and dilated hepatic sinusoid. Black star indicate necrosis area. (J, K) Serum ALT and AST activity were determined on POD-14. (M) Representative photomicrographs showing necrosis of liver sections (original magnification, 100×) and quantified (L). Data are presented as the mean±SEM. ^**** P < 0.0001 was determined using 1-way ANOVA. r values were determined using Spearman rank correlation. Abbreviations: ALT, alanine aminotransferase; AST, aspartate aminotransferase; BDL, bile duct ligation; CitH3-DNA, citrullinated histone H3-DNA; MPO, myeloperoxidase; POD, postoperative day; TAT, thrombin-antithrombin.

NETs convert LSECs to a procoagulant phenotype and decrease the permeability of LSECs

Immunohistochemical staining of liver sections from cholestatic mice revealed TF staining localized to the hepatic sinusoids (Fig. 2F). Co-staining of liver tissue on POD-14 revealed colocalization of TF with CD31 (a marker of LSEC activation) (Fig. 3B), supporting in vivo TF expression by LSECs.³⁵ This finding was further confirmed by TF staining observed in NETs-treated LSECs in vitro (Fig. 3C, D). Additionally, we detected PS expression on LSECs, a molecule known to enhance TF procoagulant activity.³⁶ Consistent with this, NET-treated LSECs exhibited higher PS expression compared to controls (Fig. 3E, F). We further investigated the procoagulant activity of LSECs and found it to be elevated after the NET treatment (Fig. 3G). Notably, DNase I treatment of NET-cultured LSECs significantly reduced the exposure of TF and PS (Fig. 3C–F) and protein production (Fig. 3H), suggesting its ability to reverse NETs-induced procoagulant activity. Furthermore, blocking PS with lactadherin and blocking TF with an anti-TF antibody significantly decreased LSEC procoagulant activity, with lactadherin showing a greater effect of over 62%, while anti-TF showed a lower effect (Fig. 3H). Our findings indicate that both PS and TF play a pivotal role in mediating the interactions between coagulation and NETs.

FIGURE 3.

NETs induce thrombogenicity in LSECs. (A, B) Liver sections on POD-14 were stained for tissue factor (TF) and CD31. Images were merged to determine the colocalization of TF and CD31 staining (yellow). (C) Representative images showing TF expression on LSECs. (E) The membrane of LSECs showed green fluorescence after being labeled with FITC-lactadherin, and the cell nuclei showed red fluorescence after being labeled with propidium iodide (PI). The proportion of TF (D) and PS (F) on LSEC incubated with NETs in the presence of DNase I. (G) In-Xa, Ex-Xa, and thrombin were evaluated. (H) For the inhibition assays, NET-treated LSECs were preincubated with DNase I (100 U/mL), lactadherin (128 nM), or an anti-TF antibody (40 mg/mL). Data are presented as the mean±SEM. ^* P < 0.05; ^** P<0.01; ^**** P < 0.0001 was determined using Kruskal-Wallis test (D) or 1-way ANOVA (F, H) or unpaired t-test (G). Abbreviations: BDL, bile duct ligation; CTL, control; Ex-Xa, extrinsic Fxa; In-Xa, intrinsic Fxa; NETs, neutrophil extracellular traps; POD, postoperative day; PS, phosphatidylserine.

We observed increased CD31 expression in the hepatic lobule of BDL mice on POD-14 compared to controls (Fig. 3A, B). The enhanced expression of CD31 in the hepatic lobule indicated that differentiated LSECs undergo capillarization³⁵ and are associated with increased expression of adhesion molecules. We then examined if NETs could induce adhesion molecule expression on LSECs. Since EC adherence relies on junctional proteins, we evaluated the expression levels of the intercellular protein VE-cadherin employing confocal microscopy and western blotting. We also examined the effect of NETs on the endothelial actin cytoskeleton. Stimulation of LSECs by NETs resulted in an increased number of VE-cadherin and F-actin stress fibers relative to the control group (Fig. 4A–C). Compared with the unstimulated cells, cells treated with NETs exhibited increased VCAM-1 and ICAM-1 expression, and the addition of DNase I abrogated this function (Fig. 4D). As the permeability of LSECs influences protein transfer in the liver, we investigated whether NETs could modulate LSEC permeability. Next, the transport of FITC-Dextran across a monolayer of LSECs was performed to analyze endothelial permeability in vitro. Over some time, the FITC-Dextran exhibited a quicker accumulation rate in the bottom well via the LSEC monolayer. Additionally, our study revealed a reduced permeability of the LSEC monolayer following treatment with NETs. Pretreatment with DNase I significantly attenuated the reduced permeability induced by NETs (Fig. 4E, F). These findings indicate that NETs function to hinder protein transport across LSECs by enhancing the tightness of cell-cell junctions. Crucially, engagement of DNase I prevents VE-cadherin and F-actin upregulation and improves endothelial permeability.

FIGURE 4.

Effect of NETs on LSEC structure and permeability. (A, B) Representative images showing VE-cadherin and F-actin in LSECs. The expression of (C) VE-cadherin, (D) ICAM-1, and VCAM-1 were measured by western blotting. (E) Scheme of the LSEC permeability experimental model. (F) LSECs treated with NETs (0.5 μg DNA/mL) with or without DNase I, values reported as the quantified amount of FITC in lower chamber per time point. The experiment was repeated four times with 3–4 replicates per group with similar results. Data are presented as the mean±SEM. ^* P < 0.05; ^*** P < 0.001; ^**** P < 0.0001 was determined using 1-way ANOVA (A, B) or Kruskal-Wallis test (F). Abbreviations: ICAM-1, intercellular adhesion molecule-1; NETs, neutrophil extracellular traps; VM-1, vascular cell adhesion molecule-1.

NETs exacerbate the inflammatory load in cholestatic mice and directly modulate macrophage activation

Sirius red staining (Fig. 5A) and alpha-smooth muscle actin immunohistochemical (Fig. 5B) showed a significant increase in fibrosis in the BDL mice as compared with sham mice on POD-14. Since fibrosis is often linked to inflammation, we investigated the expression of inflammatory factors in the livers of BDL mice on POD-14. Compared to sham controls, BDL mice exhibited elevated levels of TNF-α and IL-6 (Fig. 5C, D). M1 macrophages highly express TNF-α, which is associated with inflammation.³⁷ An immunohistochemical analysis of liver tissue using F4/80 (a marker of macrophages including M0) and CD86 (a marker of M1 macrophages)³⁷ demonstrated that BDL significantly increased both M0 and M1 macrophages in the liver on cholestatic mice on POD-14 (Fig. 5E, F). Notably, DNase I significantly reduced macrophage infiltration and activation (Fig. 5E, F), supporting the notion that infiltrating and activated macrophages are partially a consequence of NET formation. Importantly, DNase I blockade of NETs led to a reduction in the expression of TNF-α, IL-6, and fibrosis in the liver tissue (Fig. 5A–D), which mirrored the decrease in infiltrating macrophages. These findings highlight the significance of NETs in both inflammatory and fibrotic processes in the liver. We subsequently investigated whether NETs exert a direct regulatory effect on cytokine production by macrophages in vitro conditions. Our findings indicate that NETs significantly enhance macrophage activation, as evidenced by increased release of TNF-α and IL-6. Conversely, DNase I treatment effectively attenuates macrophage activation, demonstrated by the decreased levels of these proinflammatory cytokines (Supplemental Figure S3, http://links.lww.com/SLA/F198). Collectively, we hypothesize that heightened NET levels may contribute to liver inflammation and fibrosis progression in cholestatic mice through macrophage activation taken together, the above ex vivo and in vitro findings further support the pathogenic role of neutrophils in CLD through NETs formation (Fig. 6).

FIGURE 5.

NETs are linked to increased inflammatory cytokines and fibrosis in cholestatic liver diseases. (A) Sirius Red staining of the liver on POD-14 was performed (original magnification, ×100) and quantified. (B) Immunohistochemistry staining with α-SMA (original magnification, ×100) in mouse liver on POD-14 and the positive area were quantified. (C-F) The expression of TNF-α, IL-6, F4/80, and CD86 in the liver on POD-14 was detected by immunohistochemistry (original magnification, ×200) and quantified by Image J. Data represent the mean±SEM. ^**** P < 0.0001 was determined using 1-way ANOVA. Abbreviations: α-SMA, alpha-smooth muscle Actin; BDL, bile duct ligation; POD, postoperative day.

FIGURE 6.

NETs exacerbated intrahepatic thrombotic tendency and liver damage in cholestatic liver diseases. (A) Bilirubin induced a massive formation of NETs. NETs bind to the ICAM-1 and VCAM-1 on the surface of LSEC, exacerbating LSEC damage and further increasing the expression of adhesion molecules, which in turn enhances the adhesion of NETs to LSEC. (B) NETs simultaneously increased the procoagulant activity of LSECs by inducing the expression of TF and PS, promoting fibrin formation and facilitating thrombosis. (C) Furthermore, NETs reduced the permeability of LSEC and may impair transport function between the blood and liver parenchyma. (D) NETs indued an inflammation environment by activating macrophages to release IL-6 and TNF-a, and ultimately, the hepatocytes are damaged. However, DNase I reversed the hypercoagulable state, permeability dysfunction, and inflammation. Abbreviations: ICAM-1, intercellular adhesion molecule-1; NETs, neutrophil extracellular traps; PS, phosphatidylserine; TF, tissue factor; VCAM-1, vascular cell adhesion molecule-1.

DISCUSSION

This study identified 5 key observations. First, we observed a marked accumulation of neutrophils in the livers of cholestatic mice, which coincided with the formation of NETs. This suggested a potential link between neutrophil infiltration and NET generation in CLD. Second, our findings indicated that the abnormal generation of NETs contributed to the activation of the coagulation cascade. This was evidenced by elevated plasma TAT levels and significant deposition of TF and fibrin within the liver. Third, we demonstrated that NETs could induce LSEC dysfunction, which manifested as compromised endothelial permeability and increased procoagulant activity of LSECs. Fourth, this study suggested a pivotal role for NETs in initiating and driving the progression of CLD. NETs appeared to facilitate macrophage activation, leading to subsequent cytokine release and potentially worsening liver damage. Finally, inhibiting NET formation with DNase I treatment improved intrahepatic hypercoagulability, liver injury, and fibrosis in cholestatic mice. This finding highlighted the therapeutic potential of targeting NETs in CLD treatment.

Neutrophil infiltration into the liver is a hallmark of mechanical obstruction of the biliary tract and may significantly contribute to tissue damage.³⁸ In the current study, we showed that increased liver-infiltrating neutrophils as early as 3 days after BDL. This aligns with previous findings indicating increased ICAM-1 expression in the liver after 3 days of BDL, promoting leukocyte recruitment.³⁹ Blocking neutrophil infiltration has been shown to attenuate hepatocyte injury in mice. However, our results revealed that blocking neutrophils can only reduce liver damage by about 35%~50%. While DNase I could reduce liver damage by about 84%. These results might be owing to the paradoxical effects of neutrophils. Existing data have shown that neutrophils play a crucial role in host defense, inflammation resolution, and tissue healing, which might counteract the hepatotoxicity effect of neutrophils.⁴⁰ Studies suggest that elevated neutrophil counts enhance survival rates in patients with severe alcohol-associated hepatitis and acute liver failure⁴¹ and contribute to tissue repair in acetaminophen-induced liver injury models.⁴² Therefore, the overall effect of neutrophil depletion was quite limited. Based on these findings, we hypothesize that neutrophils may also contribute to liver injury through mechanisms beyond infiltration, such as NETs. Moreover, our results, coupled with evidence of inhibited neutrophil apoptosis in BDL rats,¹² suggest that this impaired apoptosis of neutrophils may partially contribute to enhanced NET formation in the liver. Primarily, DNase I treatment, which degrades NETs, exerted a protective effect on the liver in BDL mice. In addition to regulating the release of NETs, we found that DNase I also decreased neutrophil infiltration because DNase I reduced the expression of chemokines in the liver. This suggested that DNase I decreased the level of NETs in the liver not only by directly inhibiting NET generation but also by reducing neutrophil recruitment to the liver.

Our findings demonstrated that suppressing NET formation dramatically reduced circulating TAT complex levels and liver fibrin deposition in cholestatic mice. This suggested that NETs played crucial roles in the upregulation of intrahepatic coagulation. Consistent with our study, previous research has shown that the TF: FVIIa complex initiates intrahepatic coagulation during cholestasis.²⁴ Endothelial injury serves as a critical pathophysiologic mechanism underlying thrombosis formation. This present study showed that TF colocalized with LSECs, which indicated that LSECs are the primary source of TF. Furthermore, co-culture experiments indicated that NETs induce damage to LSECs by initiating PS exposure and TF expression, ultimately leading to a procoagulant phenotype. Based on these findings, we hypothesize that the overproduction of NETs in CLD plays a crucial role in LSEC injury within the liver sinusoids. Consequently, it can be speculated that procoagulant LSECs may predispose the liver to intrahepatic micro-thrombosis, potentially impairing microvascular perfusion and leading to local ischemia, disrupted blood flow, and ultimately, tissue injury.

Aside from the increased coagulation activity, NET treatment also alters LSEC permeability. Our results suggest that NETs decrease LSEC monolayer permeability, possibly by enhancing VE-cadherin-based intracellular junctions and promoting the formation of dense F-actin stress fibers. Studies have shown that increased VE-cadherin-dependent adhesion strengthens endothelial barrier integrity,⁴³ supporting our observations. Moreover, increases in F-actin stress fiber are concomitant with sinusoidal endothelial fenestrae contraction and defenestration.⁴⁴ Defenestration of LSECs showed impaired transfer of protein-bound substrates through the fenestrations,⁴⁵ which supported our results. More importantly, our in vivo study showed that BDL mice are associated with increased expression of CD31 along with the hepatic sinusoid, which further indicates LSEC capillarization.³⁶ Based on the above results, we speculated that decreased permeability of the liver sinusoidal endothelium impairs filtration between the liver parenchyma and sinusoids, thus lending credence to the existence of another critical underlying mechanism for the progression of CLD. However, further research is required to explore the direct impact of NETs on the defenestration of LSEC.

Accumulating inflammatory cells and releasing potentially cytotoxic cytokines is a well-characterized event involved in the mechanism of liver injury in cholestasis.²³ Our present results showed increased macrophage, IL-6, and TNF-α expression in the liver of BDL mice. Additionally, we discovered that NETs imposed a stimulating effect on the secretion of TNF-α and IL-6 from macrophage in vitro, and DNase I treatment could decrease the release of TNF-a and IL-6, which emphasized that NETs are the major culprits of macrophage activation and cholestasis-induced liver injury. A previous study showed that crosstalk between macrophages and HSCs is mediated by IL-1 and TNF-α, thereby inducing fibrosis.⁴⁶ Therefore, we speculated that NETs stimulate macrophages in the liver to release inflammatory factors; the proinflammatory milieu is harmful to both parenchymal and nonparenchymal liver cells, leading to fibrosis, and ultimately, loss of liver function. In addition, macrophage infiltration into the parenchyma, which occurs through transmigration from both sinusoids and portal venules, plays a role in cholestatic liver injury.¹³ The expression of VCAM-1 on the vascular endothelium of inflamed tissues plays a pivotal role in facilitating macrophage infiltration.⁴⁷ Our study demonstrated a significant downregulation of VCAM-1 expression on LSECs following DNase I treatment. This suggests that NETs regulate the expression of adhesion molecules, thereby mediating macrophage infiltration into the liver parenchyma. Consequently, targeting NETs with DNase I treatment could serve as a promising strategy to reduce excessive macrophage activation and migration, potentially mitigating the fibrotic tendencies associated with CLD.

Through our experimental investigation, we demonstrate evidence supporting the clinical repurposing of DNase I, an enzyme that effectively dissolves NETs due to their extracellular DNA composition, to treat CLD. FDA-approved recombinant human DNase I, boasting an impressive safety profile in patients, has been utilized for over 3 decades to diminish mucus viscosity and enhance airway functionality in cystic fibrosis patients.⁴⁸ Thus, DNase I may offer a promising treatment option that is both safe and feasible, clinically demonstrating its efficacy in improving liver function during cholestasis. NETs contain not only extracellular DNA but also histones and granular proteins, such as MPO and neutrophil elastase, which may directly contribute to tissue damage.¹⁴ While not explored in this study, Yehudit Shabat et al showed that treatment with alpha-1 anti-trypsin, a serine protease inhibitor that limits neutrophil elastase activity, exerts a hepatoprotective effect on animal models of immune-mediated hepatitis and acetaminophen-induced liver damage.⁴⁹ Therefore, adjunct therapies aimed at other facets of NETs could offer additional protection following cholestasis.

Recent advancements in the understanding of coagulopathy in chronic liver disease have offered strong evidence for anticoagulation as a novel therapeutic approach for patients with liver disease.⁹ Our present study showed that NETs may increase coagulation factors and fibrin formation and that these markers correlate with liver injury. A recent study showed that pharmacological targeting of the coagulation signaling with hirudin, an anticoagulatory antagonist of thrombin, improved obstructive cholestasis and attenuated liver fibrosis symptoms.⁸ In addition, clinical evidence suggests that enoxaparin delays hepatic decompensation in patients with advanced liver cirrhosis.⁵⁰ As such, it could be interesting to observe whether FDA-approved oral anticoagulants (eg, rivaroxaban, apixaban, dabigatran), which restrict thrombin proteolytic activity, can also be used as coagulation-directed therapies for CLD pathologies.

In conclusion, our study demonstrated that NETs played a role in the advancement of CLD and served as potential triggers for CLD-related intrahepatic hypercoagulation and liver injury. Furthermore, a better understanding of the interactions among NETs, LSECs, and macrophages would be crucial for elucidating the mechanisms of liver injury in CLD. Furthermore, unraveling the thrombotic aspects of CLD could hold significant promise for developing novel strategies to combat liver injury. The rapid advancements in NETosis research offer novel strategies for preventing and treating liver injury in cholestasis.

Supplementary Material

hc9-8-e0513-s001.docx ^{(21.7KB, docx)}

hc9-8-e0513-s002.docx ^{(753.1KB, docx)}