Bronchoalveolar lavage (BAL) of 33 adults positive for SARS-CoV2 infection, diagnosed by real-time PCR on nasopharyngeal swab, were collected. N = 28 were admitted to the Intensive Care Unit (ICU) at the Luigi Sacco Hospital (Milan, Italy); N = 5 were admitted to the Intermediate Medicine ward (IMW) of the IRCCS Policlinico San Matteo Foundation. Failure of a trial with continuous positive airway pressure via a helmet was the indication for admission in ICU. Failure was defined as respiratory rate >30 breaths per minute and PaO₂ to FiO₂ ratio <150, or respiratory acidosis with pH<7.36 and PaCO₂ >50 mmHg, or agitation, or confusion. BALs method collection and demographic characteristics of the studied population can be found in the already published paper (10).

Two of collected BAL was cytospinned to produce spots that were stained using Papanicolaou, May-Grunwald-Giemsa and Ziehl-Neelsen methods. The remaining material was fixed in 10% buffered formalin and routinely processed to form cell-blocks. From the cell-blocks, 3-μm paraffin sections were stained by H&E for cytological examination.

Lung autopsy tissues from two patients died of SARS-CoV2 pneumonia were fixed in 10% buffered formalin for 48 h. Three-μm paraffin sections were stained by H&E. Immunohistochemistry reactions were performed on the most representative area by using anti-Cytokeratin 7 (CK7) (clone SP52, Ventana) and anti-α-SMA (clone 1A4, Ventana).

A549 and 16HBE cell line (purchased from ATCC^®) was cultivated in DMEM high glucose supplemented with 10% of FBS, 1% of penicillin-streptomycin solution and 1% of L-glutamine (all purchased by EuroClone). Cell were cultivated at 37°C with 5% of CO₂ and harvested when they reached the 80% of confluence.

Neutrophils were isolated from peripheral blood of healthy donors. Blood was stratified by Lympholyte^® Cell Separation Media (EuroClone, Milan, Italy) and after centrifugation at 450 g for 30 min mononuclear cell phase was eliminated to allow neutrophils isolation. A solution of 3% Dextran (200 kDa) in 0.85% NaCl was added to the remaining phases of neutrophils and erythrocytes followed by a 30 min incubation at room temperature, allowing the precipitation of erythrocytes. Supernatant was collected and washed with PBS (EuroClone). After centrifugation at 450 g for 10 min, pellet was resuspended with VersaLyse Lysing Solution (Beckman Coulter S.r.l., Milano) for 20 min at room temperature in the dark. After the addition of PBS, sample was centrifuged again at 450 g for 10 min. This passage was repeated until the pellet was found to be free of erythrocytes.

Regarding macrophages, we decided to use alveolar macrophages (AM) isolated from BAL of patients subjected to bronchoscopy for diagnosis purpose, that did not show any sign of lung infection and with a cytologic count of AM > 76% and lymphocytes < 14%. BAL were filtered with a sterile gauze, centrifuged at 450 g for 10 min and then cell pellet were cultured in suspension for 24 h in DMEM containing 10% FCS, P/S and L-glutamine.

To induce the release of NETs and to isolate them we followed a published protocol with some modifications (11). 2.5 x 10⁶ neutrophils were cultivated in 60 cm² plate with the addition of 100 nM PMA (Sigma-Aldrich S.r.l., Milan, Italy) for 4 h at 37°C (PMA-Neu). After that, supernatant was discarded, and the layer of NETs present onto the bottom of the plate was harvested using PBS and collected into falcon. After centrifugation at 450 g for 10 min, cell-free NETs-rich supernatant was divided in 1.5 mL Eppendorf and centrifuged at 18,000 g for 10 min at 4°C.

For confocal microscopy, 2 x 10⁶ PMA-Neu or non-activated neutrophils were washed with PBS and fixed with 4% of paraformaldehyde for 10 min. After three washes with PBS, cells were treated with blocking solution (1% BSA in PBS) for 1 h at room temperature. Cell were than stained with anti-H3 polyAb (dilution 1:100 – PA5-31954 - Invitrogen - Life Technologies) or anti-MPO polyAb (dilution 1:50 – DOM0001G - Invitrogen) or anti-HNE mAb (dilution 1:50 – MA1-40220 - Invitrogen) in 0.5% BSA in PBS for 1 h at room temperature. After three washes in PBS, secondary mAbs (anti-mouse and anti-rabbit IgG H&L, Alexa Fluor^® 488 - Abcam; anti-rabbit IgG H&L, Alexa Fluor^® 647 - Invitrogen) was added in 0.5% BSA in PBS for 1 h at room temperature. Neutrophils without treatment with PMA were used as control. Coverslips were than mounted using ProLong™ Gold Antifade Mountant with DAPI (Invitrogen) and analyzed with confocal microscopy (Fluoview FV10i, Olympus).

To quantify NETs, we measured free DNA using Cell Death Detection ELISAPLUS kit (Roche) and Citrullinated Histone H3 (CitH3) (Clone 11D3) ELISA Kit (Cayman).

To study the induction of EMT on A549 by PMA-Neu or pure NETs, 0.2 x 10⁶ A549 were cultured on 12-well plate for 24 h. Then, 2.5 x 10⁶ PMA-Neu or pure NETs derived from the same number of cells were added in each well and after 24 h supernatants were discarded and A549 were lysed to extract all proteins or total RNA to perform western blot and real-time PCR (RT-PCR) analysis. As positive control we treated A549 with 10 µg mL^-1 of TGF-β.

Cell death was evaluated labeling A549 with propidium iodide (PI) after 24 h of incubation with 100 nM PMA, 2.5 x 10⁶ PMA-Neu or NETs derived from the same number of cells and analyzed with flow cytometry.

A wild strain of SARS-CoV2 was isolated in VERO E6 cell line from a nasal swab of a COVID-19 infected patient. Virus was propagated and titrated to prepare a stock to be stored at –80°C to be used for all the experiments at a concentration of 100 TCID50. The amount of 100TCID50 was chosen on the basis of virus titration previously performed according to the procedure reported (12).

To mimic the airway microenvironment in vitro, 0.5 x 10⁶ A549 cells were cultured onto transwell inserts (0.4 µm-pore size - Corning Costar) in air liquid interface (ALI) to allow a good polarization of cells. After 14 days, different conditions were prepared:

Only A549 as control

A549 + SARS-CoV2 (added onto A549 in 20 µL to not modify the polarization of cells)

A549 + 2.5 x 10⁶ neutrophils + SARS-CoV2

A549 + 0.5 x 10⁶ AM + SARS-CoV2

A549 + 0.5 x 10⁶ AM + 2.5 x 10⁶ neutrophils + SARS-CoV2

After 48 h, medium in the lower compartment was collected to perform cytokines/NETs quantifications, transwell inserts were cut around the membrane edges to lyse cells to extract all proteins or total RNA.

Cells were lysed with lysis buffer (50 mM Tris-HCl [pH 7.4], 150 mM NaCl, 10% glycerol, 1% NP-40, protease inhibitor cocktail (Sigma Aldrich) and phosphatase inhibitor (Roche)), gently vortexed for 20 min at 4°C and centrifuged for 15 min at 13,200 rpm at 4°C. Supernatants were quantified by Pierce™ BCA Protein Assay Kit (Thermo Fisher Scientific).

Twenty micrograms of proteins from A549 cell extracts were loaded and separated in 8% SDS-PAGE. After electrophoresis, the gels were transferred to polyvinylidene difluoride membranes (Millipore), therefore blocked (5% no fat milk in 0.1% Tween 20 TBS) and incubated with the primary Ab (1:1000 in TBST + 2% BSA; overnight at 4°C or 2 h at room temperature): anti-E-Cadherin [M168] (ab76055, Abcam), anti-α-SMA [E184] (ab32575, Abcam), and anti-β-Actin (MAB1501R, Chemicon). After wash, the membranes were incubated with the appropriate HRP-conjugated secondary Ab (1:5000 in TBST + 2% BSA; 2 h at room temperature; anti-mouse A4416 and anti-rabbit A0545, Sigma). The immunoreactivity was detected by ECL reagents (Amersham), acquired with the ChemiDoc imaging system (Image Lab, Bio-Rad).

Total RNA was isolated from cells using miRNeasy Mini Kit (Qiagen). RNA concentration and purity were evaluated using a spectrophotometer (Nanodrop 2000, Thermo Scientific, Madison, WI, USA). cDNA was retrotranscribed from 1 µg of total RNA using LunaScript RT SuperMix Kit (NEB), according to the manufacturer’s instructions. Relative levels of Alpha Smooth Muscle Actin (ACTA2) and E-Cadherin (CDH1) mRNA were assessed using SYBR^® Green Luna^® Universal qPCR Master Mix (NEB) and normalized to the levels of glyceraldehyde-3- phosphate dehydrogenase (GAPDH) mRNA. All reactions were performed on an LC480 Real-Time PCR system (Roche Diagnostics, Vienna, Austria) according to the manufacturer’s recommendations. Each experiment was performed in triplicate. The threshold cycle (Ct) was defined as the fraction cycle number at which fluorescence exceeded the given threshold. Relative gene expression level quantification was compared with internal standards and analyzed using the 2^−∆∆Ct method.

To quantify cytokines released in the alveolar in vitro model, ELISA assays were performed. We quantified IL8 with SimpleStep ELISA^® Kit (Abcam), TGF-β (Abcam) and IL1β with human IL-1β/IL-1F2 Immunoassay (R&D Systems) was titered using a commercial enzyme-linked immunosorbent assay kit (Human IL-1β/IL-1F2 Immunoassay, R&D Systems) following the manufacturer’s instructions and the results were expressed as pg ml^-1. All determinations were measure in same session. For IL8 quantification in BAL of COVID-19 patients we referred to already published quantifications (10).

In vitro results presented in this study were analyzed by one-way ANOVA followed by Dunnet test, or by unpaired t-test. Results were obtained from three/four independent experimental replicates represented as mean ± SD. BAL analyses were done by Mann-Whitney test, while correlation analyses were done calculating Spearman coefficient. Data are represented as median (interquartile range – IQR). All analyses were carried out with a GraphPad Prism 6.0 statistical program (GraphPad software, San Diego, CA, USA). A value p < 0.05 was considered statistically significant.

Research and data collection protocols were approved by the Institutional Review Boards (Comitato Etico di Area 1) (prot. 20100005334) and by IRCCS Policlinico San Matteo Foundation Hospital (prot.20200046007). Written informed consent was obtained by all conscious patients in accordance with the Declaration of Helsinki. For unconscious patients next of kin were informed about all procedures and the inform consent was obtained from patients after recovery; for non-survivors was waived in accordance with the Italian law (Decreto legislativo 211/2003 art 5) (13).

In agreement with recent papers (8, 14–16), measuring NETs in BAL of mild (IMW) and severe (ICU) patients, we found that ICU have a significant higher amount of NETs compared to IMW patients ( Figure 1A ). Interestingly, when patients were divided in survivors and non-survivors, non-survivors group presents more NETs compared to patients who survived ( Figure 1B ). Finally, we assessed a significant direct correlation between NETs and neutrophil counts ( Figure 1C ), as well as the levels of IL8 ( Figure 1D ), most known cytokine related to neutrophil recruitment. Figure S1 are representative images of immunohistochemical analysis of BAL of severe COVID-19 patients showing that the major cellular components are neutrophils, macrophages, epithelial cells and cellular/nuclear debris.

The histopathological examination of lung tissues obtained from two patients who died for COVID-19 pneumonia demonstrated exudative and proliferative features of diffuse alveolar damage, corresponding to an early/intermediate phase of the disease (17). Moreover, a subset of pneumocytes is double positive for epithelial and mesenchymal markers, CK7 and α-SMA, respectively ( Figure 2 ), suggesting establishment of lung fibrosis through EMT mechanism.

To demonstrate a direct relation between NETs induced by SARS-CoV2 infection and EMT mechanism, we firstly assess if NETs are able to induce EMT in a monoculture of A549, an alveolar epithelial cell line. To induce NETosis we used (2, 5) (9)PMA ( Figure S2 ) and, as expected, NETs are visible as web-like structures and they are positive for the histone H3 ( Figure S2K ), HNE ( Figure S2N ) and MPO ( Figure S2R ), confirming literature data (18). Moreover, DAPI labeling further confirms that these proteins are associated to DNA ( Figures S2L, O, S ). Then, we treated A549 with 2.5 x 10⁶ neutrophils activated with PMA or with pure NETs isolated from the same amount of PMA-activated neutrophils. We evaluated the expression of two major proteins modified in EMT process: α-SMA (a mesenchymal marker) and E-cadherin (an epithelial marker). Western blot and RT-PCR analysis showed that α-SMA expression was significant up-regulated by PMA-Neu and NETs after 24 h compared to control cells ( Figure 3 ), as well as TGF-β treatment ( Figure S3B ). Regarding E-Cadherin, PMA-Neu and NETs significantly downregulated protein and mRNA expression after 24 h of treatment ( Figure 3 ), an effect that was similar to those obtained by TGF-β treatment ( Figure S3B ). Optical images confirmed that the modulation of proteins expression leads to a change in cell morphology ( Figure S3A ).

Knowing that NETosis can induce also cytotoxicity we evaluated A549 cell death after 24 h of treatment with PMA-Neu and NETs. The amount of cell death was assessed by PI staining acquired by flow cytometer. Analysis showed that PMA-Neu induced 33.70 ± 4.37% of cell death compared to control cells, in contrast to NETs ( Figure S3C ). To exclude that cytotoxicity observed after PMA-Neu treatment was due to the presence of PMA, we evaluated also PMA effect on A549 viability demonstrated that PMA induce only 8.25 ± 0.97% of cell death ( Figure S3C ).

Assessed that NETs are present at high level in BAL of severe COVID-19 patients and that they are able to induce EMT, we next set-up an airway in vitro model to connect SARS-CoV2 infection, production of NETs and EMT trigger. To make airway in vitro model we cultivated A549 cells at ALI, AM were seeded at the apical side of A549 cells, while neutrophils were added to the basolateral chamber of the transwell to mimic alveolar microenvironment (19, 20). Our in vitro model was then inoculated with SARS-CoV2 (Neu+AM+SARS-CoV2). As a control, A549 cells were cultured alone, or with SARS-CoV2, or with Neu+SARS-CoV2, or with AM+SARS-CoV2.

After 48 h of incubation, protein ( Figures 4A, B ) and mRNA expression analysis ( Figure 4C ) showed that the only condition that showed a complete expression pattern of EMT was Neu+AM+SARS-CoV2. In fact, even if α-SMA (mRNA and protein level) is significant up-regulated also in Neu+SARS-CoV2 condition, we did not see a down-regulation of E-cadherin. For SARS-CoV2 condition, even if at mRNA we observed a significant modulation of mRNA of α-SMA and E-Cadherin ( Figure 4C ), at protein level we did not see the same significance ( Figure 4B ).

To clarify if NETs are produced by neutrophils and if they are possibly involved in triggering EMT, we quantified NETs showing that there is high concentration of NETs in Neu+SARS-CoV2 and Neu+AM+SARS-CoV2 compared to control cells ( Figures 4D , S4C ). As expected, in SARS-CoV2 and AM+SARS-CoV2 we did not find any sign of NETs given the absence of neutrophils ( Figures 4D , S4C ).

Quantifying TGFβ ( Figure 4E ), most known pro-fibrotic factor, IL8 ( Figure S4A ) and IL1β ( Figure S4B ), the two major cytokines involved in inducing NETosis (21), we observed that they are present only in the treatment condition where are present AM (Neu+AM+SARS-CoV2 and AM+SARS-CoV2). However, assessed that in AM+SARS-CoV2 there is no EMT induction ( Figures 4A–C ), we can suggested that in this in vitro model are necessary the presence of all factors mentioned to have a complete EMT trigger.

All the results obtained with A549 cell line, was confirmed by using another cell type to perform in vitro model, 16HBE, human bronchial epithelial cell line that express ACE2 (22), in contrast to A549 that did not express the specific receptor for SARS-CoV2 ( Figure S5 ).