The study complied with the Declaration of Helsinki and was approved by the Ethics Committee of Fudan University Shanghai Cancer Center (license number: 20180109-04). Written informed consent was obtained from all subjects. Patients admitted to the intensive care unit (ICU) from January 2018 to December 2018 were enrolled in the study. The inclusion criteria included: ages between 18 and 70 years, diagnosed with sepsis or sepsis-induced ARDS (ARDS) after surgery, and an expected ICU hospitalization longer than 24 h. The exclusion criteria were as follows: history of cardiopulmonary arrest before ICU admission; history of connective tissue disorders, such as vasculitis; pregnancy; and any history of vascular embolism. Patients were diagnosed with sepsis according to the third international consensus definition for sepsis (Singer et al., 2016). ARDS was diagnosed in patients with severe sepsis or septic shock receiving mechanical ventilation (MV) for hypoxemic respiratory failure, and they met the diagnostic criteria for acute lung injury-ARDS according to the 2012 Berlin criteria for the diagnosis of ARDS (Fan et al., 2018).

Demographic data were recorded from patients upon admission to the ICU. All patients received computed tomography (CT) to assess the severity of lung injury upon admission to the ICU. The PaO₂/FiO₂ ratio was also calculated at this time. The Acute Physiology and Chronic Health Status Scoring System II (APACHE II) score (Knaus et al., 1985) was calculated to determine the severity of sepsis.

Ten milliliters of peripheral venous blood were taken within 1 h of ICU admission. Peripheral blood neutrophil counts, lymphocyte counts, monocyte counts, platelet counts, hemoglobin concentration (g/dl), erythrocyte sedimentation rate (ESR), C-reactive protein (CRP) levels, albumin concentration activated partial thromboplastin time (APTT, in seconds), prothrombin time (PT, in seconds), fibrinogen, and D-dimer concentrations were obtained and recorded. Interleukin (IL)-6 (pg/ml) and IL-8 levels (pg/ml) were measured by ELISA. Morning fasting peripheral venous blood was taken from healthy controls (HCs).

In our study, human PMNs were isolated as previously reported (Brinkmann et al., 2010). Briefly, 10 ml of human blood was prepared with heparin (10 U/ml). Five milliliters of Histopaque1119 (Sigma-Aldrich) were added to a 15-ml Falcon tube, and 5 ml of whole blood was placed on the top. Cells were centrifuged at 800g for 30 min, and the yellowish phase was discarded. Meanwhile, the lower reddish phase containing granulocytes was transferred into Falcon tubes. The cells were washed with PBS and centrifuged at 300g for 15 min. At the same time, a 100% Percoll solution was prepared by mixing 18 ml Percoll with 2 ml 10× PBS. After centrifugation, the pellets were combined, and the sedimented cells were resuspended in 5 ml of PBS. Two ml of resuspension was layered onto the gradients and centrifuged at 800g for 20 min. After centrifugation, the top layer was removed, and the remaining white interface was collected into Falcon tubes. The Falcon tubes were washed with PBS and centrifuged at 300g for 10 min. Finally, sedimented cells were resuspended in 2 ml of PBS (with 95% purity).

C57BL/6J mice aged 6 to 8 weeks were obtained from the Animal Research Center of Fudan University (Shanghai, China). All animal experiments were conducted in accordance with the relevant guidelines and regulations approved by the institutional animal care and use committee of Fudan University. An experimental cecal ligation and puncture (CLP) sepsis mouse model was used (Rittirsch et al., 2009). Briefly, SPF C57BL/6J mice weighing 25 to 30 g were anesthetized by intraperitoneal injection of 1% pentobarbital sodium (1 mg/kg). The abdomen was disinfected with 70% ethanol, a 1- to 2-cm-long incision was made along the middle of the abdomen, and the abdominal cavity was opened. Then, the cecum was exposed, ligated with a 5-0 suture, and punctured using a 20-gauge needle. Once needle-sized feces were detected, the cecum was placed back to its original position, and the abdomen was closed. All animals received 0.5 ml/10 g of normal saline for rehydration. After surgery, the animals were placed in their cage and monitored every two hours until sampling. Control mice referred to healthy controls. Sham mice underwent the same surgical procedures without cecal puncture or ligation.

Lungs were fixed in 10% neutral formalin solution for 24 h. Then, sections of the lungs were embedded in conventional paraffin and stained with hematoxylin and eosin. A semiquantitative scoring system was used to evaluate the severity of lung injury. The system evaluates four items, including alveolar congestion and hemorrhage, alveolar wall/hyaline membrane formation, neutrophil infiltration or aggregation in alveoli or blood vessels, and the degree of inflammatory cell infiltration (Randrianarison et al., 2008). Two pathologists blinded to the results scored the specimens. Specimens were scored as follows: 0, no injury; 1, mild injury (25%); 2, moderate injury (50%); 3, severe injury (75%); and 4, very severe injury (almost 100%) for each item. Then, each item was added for a possible maximum of 16 points (highest severity) (Uzunhan et al., 2016).

The wet-to-dry weight ratio was calculated to evaluate the degree of edema of the lung tissue (Planes et al., 2010). For wet/dry ratio calculations, the surface water of the lung tissue was absorbed with filter paper and placed in a clean and dry glass test tube to obtain the wet weight of the specimen. Then, each specimen was desiccated at 70°C for 48 h. Finally, the PaO₂/Fio₂ was calculated before euthanasia.

PMNs from healthy controls were initially cultured in RPMI (Gibco BRL) mixed with 5% FBS (Gibco BRL). For in vitro studies, PMNs isolated from healthy controls were treated for a duration of 6 h with plasma or platelets from healthy controls, patients with sepsis, or patients diagnosed with sepsis-ARDS. Platelets isolated from HC were treated with plasma from HC, sepsis, and sepsis-induced ARDS for 30 min at 37°C for platelet activation. Platelets were also cocultured with plasma from the three groups to induce NET formation.

For the two-procedure stimulation of PMNs, platelets were first pretreated with plasma from ARDS for 30 min to induce their activation. Then, activated platelets were cocultured with PMNs from healthy controls and plasma from ARDS patients. Platelets were pretreated with FLLRN peptide (500 mM, Anaspec) for 30 min to block protease-activated receptor 1 (PAR-1) signaling (Holinstat et al., 2011). ARDS plasma was treated with antithrombin III (Sigma-Aldrich, SML2370) or dabigatran (Sigma-Aldrich, A2221) for 30 min for thrombin inhibition. Recombinant thrombin (0.01 U, Calbiochem) was used as a positive control for healthy control platelet activation. For the combined stimulation, dabigatran was used in both ARDS plasma for platelet activation and subsequent PMN stimulation.

To explore the impact of neutrophil depletion and NET degradation on the progression of sepsis-induced lung injury, DNase (Sigma-Aldrich, D5025), NE inhibitor (ab142369), anti-Ly6G (Abcam, ab238132), and Cl-amidine (Sigma-Aldrich, 506282) were used as previously described (Williams et al., 2001; Cummins et al., 2008). Briefly, mice were injected intravenously with 65 U DNase (Sigma-Aldrich) in 100 μl 0.9% sodium chloride solution 6 h after cecal puncture. For human samples, DNase was added after PMNs were cultured. NET degradation is being induced through DNase treatment in the experiment. NETs were identified as structures positive for citH3. NET degradation in peripheral neutrophils was quantified by examining 100 cells in a double-blind experimental procedure.

Pulmonary blood vessels from ARDS mice were carefully separated, and the in situ thrombus attached to the inner wall of the blood vessel was detected under a microscope. The in situ thrombus was then carefully detached and fixed in 4% formaldehyde solution. Then, immunofluorescence staining was performed to observe NET formation. NET analysis was performed by immunofluorescence using confocal microscopy as previously reported (Barnes et al., 2020; Hisada et al., 2020). NETs were stained with anti-myeloperoxidase (diluted 1:100; Abcam 134132), anti-histone H3 (diluted 1:100, citrulline R2+R8+R17; Abcam 5103), anti-neutrophil elastase (diluted 1:100; Abcam 254178), anti-Ly6G (diluted 1:100, ab238132), and anti-tissue factor (diluted 1:100, Abcam 228968) antibodies. Alexa Fluor 488 (1:200, Life Technologies A-21206) and Alexa Fluor 555 (1:200, Life Technologies A-21432) were used as secondary antibodies. DNA was stained with DAPI (Sigma-Aldrich). The percentage of NET-releasing cells was calculated by examining 100 cells in a double-blind experimental procedure (Nie et al., 2019).

Immunoblotting for TF determination was performed as previously described (Eriksson et al., 2016). Briefly, total protein was extracted from samples, and the protein concentration was measured using BCA protein assay reagent (Beyotime, Shanghai, China). Then, the extracted protein was separated by sodium dodecyl sulfate polyacrylamide gel electrophoresis and detected by immunoblotting with specific antibodies against tissue factor (Abcam, ab228968).

Total RNA was extracted by TRIzol reagent (Invitrogen, MA, USA), and cDNA was synthesized with a Transcriptor First Strand cDNA Synthesis Kit (Roche, Basel, Switzerland) according to the manufacturer’s instructions. Primers specific for mouse TF mRNA were detected using forward primers spanning exons 4 and 5 (5′-TCAAGCACGGGAAAGAAAAC-3′) and reverse primers located within exon 5 (5′-CTGCTTCCTGGGCTATTTTG-3′).

Platelets were stained with anti-annexin/ANXA5 antibody (Abcam, ab63556) and anti-CD62P antibody (Abcam, ab234221). Anti-CD61-PerCP antibody (BD Biosciences, 347407) was chosen as the platelet-specific marker, and CD61-CD11b positivity was used as a marker of PMN/platelet aggregation. For neutrophils, CD11b (Abcam, 133357) was used as a specific marker. PMNs were identified as CD11b-positive events (Yoon et al., 2007; Adan et al., 2017; Kang et al., 2018).

MPO-DNA complexes were used to quantify NET release in vivo and in vitro. MPO-DNA complexes were quantified by capture ELISA as previously described (Aldabbous et al., 2016; Petersen et al., 2018).

A myeloperoxidase activity assay (ab105136) was used to test MPO activity. MPO activity was quantified as previously described (Huang et al., 2016).

Cell-free DNA (cf-DNA) in serum and culture supernatant was detected by a Quant-iTPicoGreen^® dsDNA kit (Invitrogen) according to the manufacturer’s instructions. The amount of DNA was measured by the fluorescence intensity (480 and 530 nm).

Thrombin concentration was measured in ex vivo plasma obtained from the blood of HC, sepsis, and ARDS patients by a TAT complex ELISA kit (Abcam, ab108907) according to the manufacturer’s instructions (Zhao et al., 2017).

Patient demographics and treatments were compared using a chi-square test or Student’s t test. Survival data were calculated by the Kaplan-Meier method and compared with the log-rank test. Correlations were analyzed by means of Spearman’s or Pearson’s tests. NET cutoff points were generated and analyzed using X-tile 3.6.1 software (Yale University, New Haven, CT, USA), which identified the cutoff with the minimum P values from log-rank ×2 statistics for survival (Camp et al., 2004). All laboratory experiments were performed in triplicate. A P value <0.05 was considered statistically significant. Statistical analyses were performed with SPSS 20.0 (SPSS Inc. Chicago, IL, USA).

Twenty-four and 16 patients diagnosed with sepsis and ARDS, respectively, were included in the study. The healthy control group consisted of 40 age-matched volunteers. More severe lung damage was observed in patients with ARDS than in the healthy control and sepsis groups ( Figure 1A ), which was accompanied by higher neutrophil numbers and higher serum levels of the inflammatory cytokines IL-6 and IL-8 in ARDS patients ( Table 1 ). To explore whether NETs were associated with lung damage in ARDS patients, we detected NET levels in peripheral blood by immunostaining neutrophils with MPO and CitH3, which are biomarkers of NET formation. Compared with patients from the healthy control and sepsis groups, higher levels of NETs per field in the blood of ARDS patients were observed [7 (3–13) vs. 3 (1-5), P<0.001, 7 (3-13) vs. 4 (3-7), P=0.023, respectively, Figures 1B, C ]. As expected, the serum level of MPO-DNA complexes was significantly increased in ARDS patients compared with healthy controls and sepsis patients [1.3 (0.6-1.7) vs. 0.6 (0.2-0.9), P<0.001, 1.3 (0.6-1.7) vs. 1.0 (0.7-1.2), P=0.011, respectively, Figure 1D ]. High NET levels, with four NETs per field as the cutoff value, and high MPO-DNA levels were further found to be associated with worse survival in ARDS patients in our study ( Figures 1E, F ). Our data also showed a strong correlation of MPO-DNA complex levels with PaO₂/Fio₂ and APACHE II scores (r=-0.395, P<0.001, r=0.344, P<0.001, respectively) ( Figures 1G, H ). Overall, the high NET levels and related poor survival in ARDS patients in our study indicate a potential role of NETs in the pathology of ARDS.

To further explore the role of NETs in ARDS, a mouse model of sepsis-induced lung injury was established, and healthy control and sham mice were compared. Compared with control and sham mice, specimens from ARDS mice showed significantly increased lung injury, as defined by diffuse disruption of the alveolar wall and massive inflammatory cell infiltration, which could be significantly alleviated by DNase treatment ( Figure 2A and Supplementary 1G ). As expected, an increase in the lung wet/dry weight ratio was observed in ARDS mice compared with control and sham mice at 6, 9, and 12 h after model establishment ( Figure 2B ). A decrease in PaO₂ was observed in ARDS mice compared with control and sham mice at 3, 6, 9, and 12 h after model establishment ( Figure 2C ). When ARDS mice were treated with DNase, their increased lung wet/dry weight ratio and decreased PaO₂ were partially reversed ( Figures 2B, C ).

To further check the level of NETs in ARDS mice, biomarkers of NETs and NETosis were detected. We observed significantly elevated levels of NETs in the lung tissues ( Figure 2D , Supplementary Figures 1E, F, H ), MPO/CitH3 expression in peripheral blood neutrophils ( Figure 2E ), cell-free plasma DNA ( Figure 2F ), and serum MPO-DNA complexes ( Figure 2G ) and a significantly increased number of cells releasing NETs ( Figure 2H ). Increased NET levels, indicated by elevated biomarkers mentioned above, were then significantly decreased by DNase treatment ( Figures 2F–I ). Higher NET levels are usually associated with impaired NET degradation. We, therefore, compared NET degradation among control, sham, ARDS, and ARDS mice treated with DNase. Our results showed significantly decreased NET degradation in ARDS mice, which could be reversed by DNase treatment ( Figure 2I ). Anti-Ly6G antibody and PAD4 inhibitor CI-amidine, known to deplete neutrophils, were also found to significantly inhibit NETs levels in ARDS mice as indicated by reduced serum levels of cell-free DNA ( Supplementary Figure 1A ), MPO ( Supplementary Figure 1B ), and MPO-DNA complexes ( Supplementary Figure 1C ) as well as reduced expression of CitH3 in lung tissues of ARDS mice after treatment ( Supplementary Figure 1D ). We also stained NETs with neutrophil elastase (NE) and NE inhibitor to detect NET formation in different groups. Additionally, MPO activity was assessed in each group. In summary, the findings shown above suggest a potential role for NET inhibition in protecting against sepsis-induced lung injury ( Supplementary Figures 2A–C ).

TF, as one of the primary initiators of blood coagulation, has been shown to be closely associated with NET formation. To check the expression and distribution of TF in ARDS mice, lung thrombi of ARDS mice were isolated, and the expression of TF and CitH3 was detected. Fluorescence imaging showed clear exposure of NETs to TF, as indicated by the coexpression of TF and CitH3 in lung thrombi ( Figure 3A ). To explore factors causing the increased TF level and its exposure to NETs, neutrophils from healthy controls were cocultured with plasma from healthy controls, sepsis patients, and ARDS patients. TF and CitH3 expressions were detected on neutrophils to determine TF/CitH3 colocalization. We observed massive NET formation in neutrophils cultured with ARDS plasma ( Figures 3B, C ), in which TF expression by neutrophils was significantly increased, and NETs were exposed ( Figure 3B ). Enhanced TF expression at both the protein and mRNA levels in the ARDS plasma culture group was also shown through Western blot and qPCR analyses ( Figures 3D, E ). Significantly decreased TF expression upon DNase treatment ( Figures 3D, E ) further demonstrated increased TF in the ARDS group exposed to NETs. The thrombin–antithrombin (TAT) complex, considered one of the biomarkers of immunothrombosis, was then detected by ELISA. Significantly enhanced TAT complex levels were found in the ARDS group ( Figure 3F ). Moreover, when treated with DNase or the anti-TF, the enhanced TAT was fully reversed ( Figure 3F ). Overall, the findings described above demonstrated that the plasma of ARDS patients contributes to TF-enriched NET formation, which plays a key role in the immunotherapy of ARDS patients.

Platelets, as one of the key initiators of thrombosis, have been reported to contribute to NET formation (Caudrillier et al., 2012). We compared the activation status of platelets from different patients and found that platelets from ARDS patients showed increased levels of surface activation markers compared with platelets from sepsis patients or healthy controls ( Figures 4A, B and Supplementary Figures 3 , 4 ). We then continued to explore the contribution of platelets to the pathology of sepsis-induced lung injury. First, we observed a significant increase in the number of platelet/neutrophil aggregates (CD61+CD11b+) in PMNs isolated from ARDS patients compared to sepsis patients or healthy controls ( Figure 4C and Supplementary Figure 5 ). To explore whether activated platelets contribute to NET formation and TF exposure on NETs in ARDS patients, neutrophils from healthy controls were cocultured with platelets from healthy controls, sepsis patients, and ARDS patients. Then, TF and CitH3 expression was detected to evaluate the TF-enriched NETs of different culture groups. As expected, neutrophils cultured with ARDS patient-derived platelets showed significantly increased NET release ( Figures 4D, E ).

The above findings suggest that both plasma and platelets of ARDS patients contribute to TF-enriched NET formation, which subsequently induces immunothrombosis in ARDS patients. We then continued to confirm the complementary effects of plasma and platelets on the formation of TF-enriched NETs. PMNs from healthy controls were first stimulated with plasma from ARDS patients to induce NET formation, with plasma from healthy controls and sepsis patients as controls. Six hours later, activated platelets from ARDS patients were added to the culture system. Fluorescence imaging of cells from different culture groups was compared. We observed significantly increased NET formation ( Figure 5A ) and numbers of PMNs releasing NETs ( Figure 5D ). We next asked whether plasma from ARDS patients was able to induce platelet activation. Platelets from healthy participants were isolated and stimulated with plasma from healthy controls, sepsis patients, and ARDS patients. As expected, a higher activation level was observed on platelets stimulated with plasma from ARDS patients compared to plasma from healthy or sepsis controls ( Figures 5B, C ). In summary, factors in plasma contribute to the activation of platelets, which play a key role in TF-enriched NET formation and the subsequent immunothrombosis of ARDS patients.

Thrombin was reported to be necessary for platelet activation through protease-activated receptor-1 (PAR-1) (Petzold et al., 2020). We therefore explored whether thrombin is the factor in plasma that induces platelet activation in ARDS patients. We first detected thrombin levels in the plasma of ARDS patients. Compared with healthy volunteers or sepsis patients, patients with ARDS had significantly elevated thrombin levels ( Figure 6A ). When extra thrombin was added to platelets stimulated with plasma from ARDS patients, a higher activation level of those platelets was observed ( Figures 6B, C ). When thrombin was inhibited using thrombin inhibitors, such as antithrombin III or dabigatran and specific PAR-1 antagonism, the activation of platelets was significantly inhibited ( Figures 6B, C ), suggesting a key role for thrombin in the activation of platelets by ARDS plasma. When PMNs from healthy controls were stimulated with ARDS plasma and ARDS plasma-treated platelets, robust TF-enriched NETs were induced ( Figure 6D ), which were then significantly inhibited when thrombin inhibitors were added to the cultures ( Figure 6E ). PMNs stimulated with both ARDS plasma and ARDS-activated platelets also induced significantly higher TAT levels, which were fully inhibited upon treatment with either TF blocking antibody or DNase ( Figure 6F ), indicating an alleviation of immunothrombosis. Overall, our findings demonstrate that thrombin in plasma in the ARDS environment plays a key role in platelet activation and the subsequent TF-enriched NET induction and immunothrombosis of ARDS patients.