Highly pathogenic coronavirus N protein aggravates lung injury by MASP-2-mediated complement over-activation

N proteins of SARS-CoV, SARS-CoV-2 and MERS-CoV interact with MASP-2

To investigate the binding between the N protein of SARS-CoV and MASP-2, and to delineate the interacting domains of the two proteins, lysates of human 293T cells expressing Flag-tagged full-length MASP-2, the N-terminal CUB1-EGF-CUB2 region, or the C-terminal CCP1-CCP1-SP region (Fig. S1A) were subjected to anti-Flag immunoprecipitation using anti-Flag antibody-conjugated agarose beads. The immunoprecipitates were next incubated with 293T cell lysates expressing GFP-tagged SARS-CoV N protein (GFP-SARS N) or truncated mutants in the presence of 2 mM CaCl₂ or 1 mM EDTA. The adsorbates were probed with anti-Flag or anti-GFP antibodies by immunoblotting. Associations between Flag-MASP-2 and GFP-SARS N were observed only in the presence of CaCl₂ (Fig. 1A), in agreement with the requirement of Ca²⁺ for MASP-2-MBL binding and MASP-2 auto-activation. The CCP1-CCP2-SP region of MASP-2 (Fig. 1A) and the N-terminal domain (residues 1-175) of the N protein (Fig. S1A and Fig. S1B) were crucial for the association, whereas negative controls, or other truncated regions did not bind, and GFP-tagged full-length or truncated N protein did not co-immunoprecipitate with mouse IgG conjugated beads (Fig. 1A and Fig. S1B).

• Download figure
• Open in new tab

Fig. 1.

The N proteins of SARS-CoV, MERS-CoV and SARS-CoV-2 bind to MASP-2. (A) Lysates of 293T cells transfected with indicated plasmids were subjected to anti-Flag immunoprecipitation in the presence of 2 mM CaCl₂ or 1 mM EDTA, and analyzed by immunoblotting. IgG immnuoprecipitates were used as a negative control. (B) Lysates of 293T cells transfected with indicated plasmids were mixed with human serum (HS) and mouse serum (MS) and subjected to immunoprecipitation with anti-Flag agarose beads. The absorbates were probed with indicated antibodies. Purified recombinant MASP-2 was loaded as a marker. (C)-(E) Lysates of 293T cells transfected with indicated plasmids were subjected to anti-Flag immunoprecipitation in the presence of 2 mM CaCl₂ and analyzed by immunoblotting.

SARS-CoV N protein can be detected in patient serum as early as 1 day after the onset of symptoms (22). To simulate the SARS-CoV N protein associations in serum, Flag-tagged N protein (1 ng/ml) was added to human or mouse serum and precipitated with anti-Flag antibody conjugated agarose beads. The SARS-CoV N protein interacted with MASP-2 derived from both human and mouse serum (Fig. 1B). Further truncation and deletion analysis showed that amino acid residues 116-124, which are located in a coil motif of the SARS-CoV N protein (residues 115-130), were indispensable for the interaction with MASP-2 (Fig. 1C). However, the association of MASP-2 with SARS-CoV NΔ321-323, a mutant that fails to form the N protein dimer (Fig. S1C), was not much affected (Fig. 1C).

The crucial motif in SARS-CoV N protein for MASP2 interaction (residues 116-124) share a high identity with the corresponding motif in SARS-CoV-2 N (115-123) and MERS-CoV N (104-112) (Fig. S1D), suggesting that the N protein of SARS-CoV-2 and MERS-CoV will also interact with MASP-2. As expected, exogenously expressed MERS-CoV N and SARS-CoV-2 N both associated with MASP-2 (Fig. 1D and 1E), and the Δ104-112 deletion mutant of MERS-CoV N exhibited the predicted reduced association (Fig. 1D). Therefore, a common motif across coronavirus N proteins is important for MASP-2 binding.

N proteins of SARS-CoV, MERS-CoV and SARS-CoV-2 potentiate MASP-2-dependent complement activation

The CCP1-CCP2-SP domains of MASP-2 are responsible for self-activation and substrate binding activity, which in turn mediates complement lectin pathway activation (23). The MASP-2: N protein association demonstrated above suggests that SARS-CoV N protein may regulate MASP-2 dimerization, activation and cleavage. To demonstrate MASP-2 dimerization, MASP-2-Flag-conjugated beads were incubated with lysates of 293T cells expressing MASP-2-Myc in the presence/absence of N protein and MBL. Binding of MASP-2-Myc to MASP-2-Flag was potentiated by the SARS-CoV N protein, at an approximately equal molar stoichiometry (Fig. 2A). Next, purified MASP-2-Flag was incubated with mannan and MBL with or without SARS-CoV N protein. Higher levels of the cleaved MASP-2 fragments (residues 445-686) resulting from MASP-2 auto-activation were produced in the presence of SARS-CoV N protein (Fig. 2B).

• Download figure
• Open in new tab

Fig. 2.

The N proteins induce MASP-2 auto-activation and C4 cleavage. (A) Lysates from cells expressing indicated proteins were mixed with purified N or MBL and subjected to anti-Flag immunoprecipitation and analyzed by immunoblotting. (B) Purified MASP-2-Flag was incubated with/without N, MBL, and mannan at 37°C for 12 hr. Cleaved MASP-2 was probed with an anti-Flag antibody. (C) Purified MASP-2 and N proteins with/without anti-N monoclonal antibody were incubated with pre-conjugated MBL in mannan-coated plates at 4°C. Binding of MASP-2 was detected with an anti-MASP-2 antibody. (D) C4 was incubated with indicated components at 37°C for 0, 2, 6, 12 and 24 hr. C4 and a cleaved C4alpha’ fragment were detected with anti-C4alpha chain antibody. (E) C4 was incubated with indicated components at 37°C for 1 hr. C4 cleavage was detected as mentioned above. Ab, anti-MASP-2 antibody. (F)-(H) C4b deposition in relation to the concentration of N proteins of SARS-CoV, HCoV-229E, SARS-CoV-2, and MERS-CoV.

To investigate the effect of SARS-CoV N protein on the MBL-binding capability of MASP-2, purified MBL and MASP-2 (24) were incubated in mannan-coated microplate wells at 4°C to avoid MASP-2 activation, and the dynamics of MASP-2: MBL binding were assessed using an anti-MASP2 antibody. Compared with human serum albumin (HSA), a high-concentration component in binding buffer used as a negative control for N protein, the binding of MASP-2 to MBL was significantly enhanced in the presence of SARS-CoV N protein at a relatively low concentration (1%∼0.1% of MASP-2) and Ca²⁺, and the potentiation was effectively abrogated by anti-N protein antibody (Fig. 2C). Moreover, SARS-CoV N protein bearing the Δ116-124 or Δ321-323 deletion or N protein from the less pathogenic human coronavirus 229E-CoV showed little or no effect on MASP-2: MBL binding compared with the full-length SARS-CoV N protein (Fig. 2C). These results indicated that SARS-CoV N protein-potentiated MASP-2 activation is dependent not only on MASP-2 association but also on N protein dimerization.

MASP-2 cleaves complement components C4 and C2 to generate C3 convertase in the lectin pathway of the complement system upon activation. Next, the effect of the SARS-CoV N protein on LP complement activation was assessed by C4 cleavage. Purified C4 was incubated with MASP-2, Mannan, and MBL in the presence/absence of equal molar N protein, and C4 cleavage was observed to be significantly potentiated by the SARS-CoV N protein in the presence of mannan and MBL (Fig. 2D and Fig. S2A). Accordingly, mutants of SARS-CoV N protein (Δ116-124 and Δ321-323) as well as N protein from H229E-CoV failed to promote MASP-2-mediated C4 hydrolysis (Fig. 2E and Fig. S2B). Notably, an anti-MASP-2 monoclonal antibody or C1INH, an inhibitor of MASP-2 (25), blocked SARS-CoV-potentiated C4 hydrolysis, suggesting that N protein-potentiated C4 cleavage was dependent on MASP-2 activation (Fig. 2E and Fig. S2B). Moreover MERS-CoV N was also found to potentiate C4 cleavage (Fig. S2C). These results indicated that N protein prompts C4 cleavage and therefore complement activation by MASP-2 association and activation.

The impact of SARS-CoV N protein on complement activation via the lectin pathway was further investigated by complement deposition assays. Purified C4 was incubated with immobilized MBL-MASP-2 complex in the presence of indicated N protein. The N protein of SARS-CoV, MERS-CoV and SARS-CoV-2 but not the H229E-CoV N potentiated C4b deposition, which was dependent on the activity of MASP-2, in a dose-dependent manner (Fig. 2F-2H). Then, immobilized mannan was incubated with C1q-depleted serum (to eliminate the classical pathway) (26) in the presence/absence of SARS-CoV N protein, and the deposited C3 fragments (C3b, iC3b and C3dg) were detected by an anti-activated C3 antibody. In concert with C4b deposition, the deposition of activated C3 was evidently increased along with the increase of SARS-CoV N protein levels up to ∼40 nM (Fig. S2D), suggesting an enhanced activity of C3 convertase. In addition, SARS-CoV N protein had little or no effect on activated C3 deposition in calcium-free buffer containing EGTA, which suggests that SARS-CoV N protein-potentiated C3 activation occurs through the LP but not the AP pathway, in which C3 activation is Ca²⁺-independent (Fig. S3E). As shown in Figure S2D and S2E, C3b deposition was decreased in the presence of a high concentration of N protein, possibly due to further cleavage of C3b by soluble inhibitors in serum (such as factor H and factor I) when the surface is coated with high densities of C3b (27, 28). We further tested the deposition of the C5b-9 complex. As a result of amplified complement cascades, significantly increased deposition of the complex was induced by SARS-CoV N protein at a much lower concentration (Fig. S3F), similar to that observed in patient sera (22).

Activated complement plays a crucial role in the efficient phagocytosis of pathogens and cellular debris by C3b or C5b-mediated opsonization (29). To study complement-dependent phagocytosis, E. coli and mouse peritoneal macrophages were incubated together in diluted serum with or without SARS-CoV N protein. Bound C3 or its large fragment C3b, the product after C3 cleavage, was stained with FITC-labeled anti-C3c antibody, and the FITC-positive macrophages containing C3-conjugated E. coli were counted under a microscope. In concert with complement activation, complement-dependent phagocytosis by mice peritoneal macrophage in the presence of mouse serum containing complement component including MASP-2 was enhanced remarkably by SARS-CoV N protein compared with the HSA control (Fig. S3G). These findings indicated that SARS-CoV N protein effectively promoted the activation and opsonic effect of the complement system.

N proteins of SARS-CoV and MERS-CoV aggravates LPS-induced pneumonia by MASP-2-involved complement activation

Persistent activation of complement leads to uncontrolled inflammation. To investigate the effect of N protein-potentiated MASP-2 activation on inflammation, mice were pre-infected with adenovirus (1×10⁹ PFU) expressing SARS-CoV N (Ad-SARS N) or its mutant or adenovirus vehicle only (Ad-null). The mice were then challenged with LPS, which contains MBL-binding motifs, to activate LP and induce inflammation (30). 8 of 10 mice pre-exposed to adenovirus vehicle survived when challenged with 5 mg/kg LPS (Fig. 3A), while all 10 mice pre-exposed to Ad-SARS N died within 12 hr after LPS administration at the same dosage (Fig. 3A). Severe lung damage and massive inflammatory cell infiltration were also observed in dead mice (Fig. 3B). In concert with previous findings, the pre-infection of Ad-229E N and Ad-SARS N bearing the Δ116-124 or Δ321-323 deletion in the N protein, respectively, had significantly reduced effects on mouse mortality. Importantly, when anti-N, anti-MASP-2 antibody or C1INH was administered simultaneously with LPS administration in the Ad-SARS N pre-infected mice, both death rate and lung tissue inflammation induced by LPS were significantly reduced (Fig. 3A and 3B). These results collectively demonstrate that the SARS-CoV N protein greatly potentiated LPS-induced inflammation via MASP-2 activation, which thereby initiates LP-involved complement cascade activation.

• Download figure
• Open in new tab

Fig. 3.

N proteins potentiate LPS-induced pneumonia in mice. (A) BALB/c mice (10 mice/group) were infected with 1×10⁹ PFU Ad-SARS N/Ad-null or a saline control, and LPS (5 mg/kg) was given on the 6th days post infection (d.p.i.). Antibodies (200 μg/kg) or C1INH (4 mg/kg) was injected 30 min before LPS injection. The fatality of mice was noted. (B) Lung paraffin sections were analyzed by HE staining. (C) Mice were infected with 1×10⁸ PFU Ad-SARS N/Ad-null, and LPS (5 mg/kg) was given by on 6 d.p.i. Mice were sacrificed after LPS challenged for 6 hr. Frozen lung sections were stained with anti-C4b antibody. (D) SARS-CoV N and MASP-2 complex formation in frozen lung sections was measured by in situ PLA, as indicated by the red signals. Deposited C3 fragments were stained with FITC-labeled anti-C3c antibodies (green). Scale bar, 50 μm. (E) Mice were pre-infected with 1×10⁹ PFU Ad-MERS N/Ad-null and treated with LPS, anti-MASP-2 antibody or C1INH as mentioned above. The mice survival curve was plotted. (F) Masp2^-/- and Masp2^+/+ C57BL/6N mice were infected N-expressing adenovirus and injected with LPS as mentioned above. The mice survival curve was plotted.

To investigate the complement activation in lung of mice induced by LPS and N protein, mice pre-infected with Ad-SARS N (1×10⁸ PFU) were challenged with LPS (5 mg/kg). At 6 hr post LPS administration, the mice were sacrificed, and the paraformaldehyde-fixed lung tissue was subjected to immunohistochemical (IHC) staining with anti-C4b antibody or immunofluorescence analysis with FITC-labeled anti-C3c antibody. The C4b and activated C3 deposition in the lung were significantly increased in mice expressing SARS-CoV N protein (Fig. 3C and the middle panel of Fig. 3D), compared with the weak staining in the lungs of mice treated with LPS and Ad-null (Fig. 3C and the middle panel of Fig. S3A).

The in vivo association of MASP-2 with the N protein was further assessed by in situ proximity ligation assay (PLA) with anti-N and anti-MASP-2 antibodies (the left panels of Fig. 3D and S3A), and red signals, which were only detectable when SARS-CoV N and MASP-2 bind to each other, were abundantly observed in pulmonary cells from mice expressing SARS-CoV N but not negative control animals (Fig. S3A). These results confirmed an in situ direct binding and activation of MASP-2 by SARS-CoV N in mice lung tissue.

Similarly, mice challenged with LPS suffered serious pneumonia and 100% die within 24 hr when pre-infected with adenovirus expressing MERS-CoV N protein (Ad-MERS N) but not its Δ104-112 mutant (Ad-MERS N^Δ104-112) for 7 days, which was partially rescued by C1INH and anti-MASP-2 antibody (Fig. 3E). These results indicate that the N protein is employed by both SARS-CoV and MERS-CoV to promote complement activation through the MASP-2-mediated lectin pathway.

The pathogenicity of N protein was further investigated using Masp2 knockout (KO) mice (KO region was indicated by Fig. S3B) with MASP-2 protease activity deficiency by the same way. Mice were pre-infected with Ad-SARS N or Ad-MERS N (1×10⁹ PFU) for 5 days, and then challenged with LPS. Compared with wild-type mice, Masp2 KO mice survived longer and had a higher survival rate (Fig. 3F), which may be attributed to a compromised complement activation because of MASP-2 deficiency.

Complements cascade is over-activated in lungs of COVID-19 patients

Because the SARS-CoV and SARS-CoV-2 induce only mild lung damage in mice, and investigation with live SARS-CoV, including mouse adapted SARS-CoV, is not allowed by related regulations in China, we try to get the clinical evidence that the over-activation of complement LP pathway occurred in COVID-19 patients. The paraformaldehyde-fixed lung tissue of patients who died of COVID-19 were collected and subjected to IHC staining with MBL, MASP-2, C4alpha, C3 or C5b-9 antibodies. All these complement cascade components demonstrated strong IHC staining signals in the patient lung tissue (Fig. 4A-4E), which suggested that the complement components were deposited in type I and type II alveolar epithelia cells, as well as inflammatory cells, some hyperplastic pneumocytes, and exudates in alveolar spaces with necrotic cell debris. Notably, SARS-CoV-2 N protein was also observed in the lung tissue of died patients by autopsy (31). Further, significantly increased serum C5a level was also observed in COVID-19 patient, particularly in severe cases (Fig. 4F). These results collectively indicated that complement pathways were aggressively activated in the lungs of COVID-19 patients, which may be attributed to SARS-CoV-2 N protein.

• Download figure
• Open in new tab

Fig. 4.

Complements activation in COVID-19 patients. (A-E) Paraformaldehyde-fixed lung tissues from postmortem autopsy was used for paraffin tissue sections and IHC staining with indicated antibodies. Microphotography was carried out by Olympus BX52 microscope under a 10× objective. (F) Serum C5a from healthy people, mild or severe COVID-19 patients were analyzed by ELISA.

Complement-targeted therapy shows a promising curative effect against COVID-19

It has been widely accepted that the excessive inflammatory and cytokine storm greatly contribute to the severity and lethality of COVID-19, which may be attributed to the unrestrained activation of complement pathway. Therefore, downregulation of MASP-2 as well as its downstream signal molecules, such as the potent anaphylatoxin C5a, may provide a new approach to control the pneumonia induced by the SARS-CoV-2.

Based on our finding and its potential and tempting application in COVID-19 therapy, a recombinant C5a antibody (BDB-001 injection by Staidson Biopharmaceutics), which was in phase II clinical trial for hidradenitis suppurativa, was rapidly approved by National Medical Product Administration (NMDA) for phase II clinical trial for the treatment of COVID-19 (2020L0003). “A multicenter, randomized double blind placebo-controlled trial in mild COVID-19 patients” and an open label “two-cohort clinical trial in patients with severe and critical COVID-19” were carried out simultaneously under the approval from ethic committee of Huoshensan Hospital with the informed consent of the patients. While the larger dataset from more patients in the COVID-19 cohort will be reported elsewhere when the trial is finished, here we reported the first 2 cases administrated with anti-C5a antibody in the open label trial.

Patient #1, a 54 years old male resident in the city of Wuhan, was admitted to Dongxihu Hospital on the 4th days after the onset of symptoms (5 days of illness) with fever. Infection of SARS-CoV-2 was confirmed by rRT-PCR for SARS-CoV-2, and chest CT scan showed bilateral opacities. The disease was getting worse from day 9 of illness, with high fever (38.7°C∼39.8°C), SpO2 <93% on room air, pneumonia progression on CT scan, and severe hepatic damage. Prednisone (40 mg/day for 4 days) was administrated on day 10 through 13 of illness but the condition deteriorated, so the patient was transferred to Wuhan Huoshenshan Hospital, a new hospital established urgently for severe COVID-19 patients, on the day 14 of illness (Fig. 5A), and was considered as severe case in critical condition with moderate ARDS (oxygenation index <150), respiratory rate 30/min,and the Percutaneous oxygen saturation (SpO2) drop to 77% when exposed to room air for 1 minute. Supportive care was provided, including high-flow nasal oxygen (HFNO) to target SpO2>95% (Fig. 5B, left). He was also administrated with the antibiotics (moxifloxacin) and human serum albumin, while prednisone was discontinued. Treatment with anti-C5a monoclonal antibody (BDB001) was initiated on the morning of day 15 of illness. The antibody was given intravenous in 250 ml saline, at a dosage of 300 mg/day, on day 1, 2, 3, 5, 7, 9, 11 and 13. No adverse event was observed and the clinical condition improved in the next days, with normal body temperature in the evening of the same day (Fig. 5C, left), increased oxygenation index (PaO₂/FiO₂) (Fig. 5B, left) and lymphocyte cell number (Fig. 5C, left), decreased C reactive protein (CRP) concentration (Fig. 5C, left), and significantly improved hepatic function (shown by decreased ALT, AST, and increased total serum protein and serum albumin concentration) (Fig. 5D, left). The fraction of inspiration O₂ and the gas flow rate of high-flow nasal oxygen (HFNO) was eventually decreased from the highest (80%, 40 L/min to 30%, 20 L/min) to target a SpO2>95% (Fig. 5B, left). Because the high risk of taking the patient in critical condition to CT scan in the temporarily established hospital (the long distance and the cold raining weather without HFNO), no CT scan just before anti-C5a administration was available for evaluation. Nevertheless, the pneumonia 20 days after the 1st dose is observed obviously improved than 11 days after the 1st dose (Fig. S4, upper panel).

• Download figure
• Open in new tab

Fig. 5.

Treatment of COVID-19 patients with anti-C5a antibody. (A) Timeline of illness onset, SARS-CoV-2 RNA detection and hospitalization of the two patients. (B) Flow rate, fraction of inspiration (FiO₂) of high-flow nasal oxygen, percutaneous oxygen saturation (SpO2) and oxygenation index (PaO2/FiO2) or SpO2/FiO2 in patient #1 (Left) and patient #2 (Right). (C) Body temperature (TEMP), C reactive protein (CRP) level and blood lymphocyte number (LYM) changes in patient #1 (Left) and patient #2 (Right). (D) Hepatic function changes in patient #1 (Left) and patient #2 (Right). ALT: alanine aminotransferase; AST: aspartate aminotransferase; TP: total protein; ALB: albumin.

Patient #2 was a 67 years old male who was admitted to the Sixth Hospital of Wuhan on the 5th day after the onset of symptoms (6 day of illness) with fever and cough. CT scan showed opacity on the superior lobe of left lung (Fig. S5A). Infection of SARS-CoV-2 was also confirmed by rRT-PCR at day 11 of illness. Anti-viral (Arbidol) and antibiotics (moxifloxacin) was administrated together with other supportive treatment. Condition was worsening by day 8 with severe cough and high fever (39.7°C) (Fig. 5C, right). Methylprednisolone (40 mg/day, for 7 days from the 8th day after illness) showed little if any improvement of symptoms. The patient was transferred to Wuhan Huoshenshan Hospital on day 10 (Fig. 5A) and the illness continued getting worse as shown by SpO2 (<90% on room air on day 14), high fever (>39°C on day 10-13) (Fig. 5C, right) and pneumonia on chest CT on day 11 of illness (2 days before the 1st dose) (Fig. S4, lower panel). The patients reported severe cough, chest tightness and dyspnea. HFNO has to be given to maintain SpO2>95%. Anti-C5a was administrated on the morning of illness day 14, and continued as in the patient #1. Subsequently, a normal temperature was observed on the same day (Fig. 5C, right). Cough, dyspnea and oppression in chest were reported to be getting better the next day (Day 15 of illness). A significant decreased C reactive protein (CRP) level, significantly increased white blood cell and lymphocyte number were also observed in the next days (Fig. 5C, right). Hepatic function was gradually improved (Fig. 5D, right, data on day 15 are not available). Flow rate and Fraction of inspiration O2 necessary to maintain SpO2>95% gradually decreased, with the increasing SpO2/FiO2 (Fig. 5B, right). Chest CT on day 26 of illness (12 days after the 1st dose) also showed alleviated pneumonia (Fig. S4, lower panel).

These data provided the first evidence for a damaging role of C5a in severely affected COVID-19 patients. The two severe patients are significantly benefited from anti-C5a monoclonal antibody therapy, which suggested that an aberrant complement cascade should be considered as a promising therapeutic target of COVID-19.

Cell culture and transfections

The 293T cell line was obtained from the Cell Resource Center of Peking Union Medical College. Cells were grown in Dulbecco’s modified Eagle’s medium (DMEM) (Invitrogen) supplemented with 10% heat-inactivated fetal bovine serum (FBS) (HyClone), 2 mM L-glutamine, 100 units/ml penicillin, and 100 µg/ml streptomycin. Cells were transfected with plasmid DNA using Lipofectamine 2000 (Invitrogen) according to the manufacturer’s protocol.

Vectors and epitope tagging of proteins

The N gene of SARS-CoV (GenBank Accession #AY274119) was amplified by RT-PCR from the SARS-CoV RNA of patient serum samples (Upstream primer: 5’-CGGAATTCCATATGTCTGATAATGGACCCCAA-3’; downstream primer: 5’-CGGGATCCTTATGCCTGAGTTGAATCAGC-3’) and cloned into the pcDNA3-based Flag vector (Invitrogen), pCMV-Myc (Clontech), pGEX-4T-2 (GE Health care), and the BglII and EcoRI sites of pEGFPC1 (Clontech). The N gene of MERS-CoV was chemically synthesized (Huaxinrcomm Technology Co., Ltd) and cloned into the pcDNA3-based Flag vector at the BamHI and EcoRI sites. The N gene of SARS-CoV-2 was chemically synthesized (General Biosystems (Anhui) Co. Ltd) and cloned into the pcDNA3.1-based HA vector at the KpnI and XbaI sites. The MASP2 gene was amplified from human hepatocyte cDNA library and inserted into corresponding vectors.

Immunoprecipitation and immunoblotting

Cell lysates were prepared in lysis buffer (50 mM Tris-HCl, pH 7.5, 1 mM phenylmethylsulfonyl fluoride, 1 mM dithiothreitol, 10 mM sodium fluoride, 10 µg/ml aprotinin, 10 µg/ml leupeptin, and 10 µg/ml pepstatin A) containing 1% Nonidet P-40. Soluble proteins were subjected to immunoprecipitation with anti-Flag M2 agarose (Sigma). The adsorbates were then separated by SDS-PAGE and transferred onto an Immobilon-P transfer membrane (Millipore) by semi-dry transblot (Biorad). The membrane was blocked by 5% Western-Blocker (Biorad). Immunoblot analysis was performed with horseradish peroxidase (HRP)-conjugated anti-Flag (Sigma), anti-β-actin (Sigma), anti-green fluorescent protein (GFP) (Clontech), anti-MASP-2 (Santa Cruz), anti-C4α (Santa Cruz), HRP-conjugated anti-Myc (Santa Cruz), and goat anti-mouse immunoglobulin G (IgG) (Amersham/Pharmacia) antibodies. The antigen-antibody complexes were visualized by chemiluminescence (GE Health Care).

Purification of SARS-CoV and MERS-CoV N protein

As described previously (40), pET22b-SARS/MERS-CoV N was transformed into the expression strain BL21 (DE3). After induction with 1 mM IPTG for 8 h, the bacteria were harvested by centrifugation and re-suspended in buffer A (25 mM Na₂HPO₄/NaH₂PO₄ (pH 8.0), 1 mM EDTA, and 1 mM DTT) before sonication. Soluble N protein in the lysate was purified with ion-exchange chromatography with SP-Sepharose Fast Flow (25 mM Na₂HPO₄/NaH₂PO₄ (pH 8.0), 1 mM EDTA, 1 mM DTT, and 0.35-0.5 M NaCl), followed by Superdex 200 gel filtration (GE Healthcare) and elution with buffer A. E. coli transformed with the vector pET22b was lysed as described above, and the eluate was used as a negative control for the purified N protein.Purified SARS-CoV-2 N-His was obtained from General Biosystems (Anhui) Co. Ltd.

Purification and renaturation of MASP-2

Recombinant protein expression and renaturation were performed as described (24, 41). In brief, pET22b-MASP-2 was transformed into the expression host strain BL21 (DE3). After induction with 1 mM IPTG, cells were harvested and sonicated. The inclusion bodies were solubilized in 6 M GuHCl, 0.1 M Tris-HCl (pH 8.3), and 100 mM DTT at room temperature; the solubilized proteins were then diluted into refolding buffers containing 50 mM Tris-HCl, 3 mM reduced glutathione (Sigma), 1 mM oxidized glutathione (Sigma), 5 mM EDTA, and 0.5 M arginine. The protein samples were then cooled to 4°C. The renatured protein was dialyzed against 20 mM Tris, 140 mM NaCl, pH 7.4 at 4°C, concentrated with PEG8000, aliquoted, and stored at −70°C.

To obtain high-activity MASP-2, Flag-tagged MASP-2 was expressed in 293T cells, precipitated with anti-FLAG magnetic beads, and eluted with Flag peptide (Sigma). The concentration of MASP-2 was assessed using a BCA kit and immunoblot analysis, with the purified prokaryotic-expressed MASP-2 as a standard control.

MASP-2 auto-activation and C4 cleavage assay

Purified MASP-2 (8 nM) was incubated at 37°C in 20 mM Tris-HCl (pH 7.4), 150 mM NaCl, and 2 mM CaCl₂ with purified C4 (Calbiochem), recombined MBL (Calbiochem), mannan (Sigma), and SARS-CoV N protein at concentrations of 50 nM, 30 nM, 15 ng/ml, and 10 nM, respectively. The cleavage was followed by SDS-PAGE under reducing conditions, and the C4alpha’ fragments, which are left by C4a release after C4 cleavage and can indicate C4 activation, and MASP-2 were detected by immunoblot analysis with anti-C4alpha chain antibody (Santa Cruz) or anti-Flag antibody (Sigma). C4b containing C4alpha’ but not C4a was used to indicate C4alpha’ band position.

Complement deposition assay

The C4b deposition assay was performed using a human MBL/MASP-2 assay kit (Hycult biotech) (42). In brief, diluted serum was incubated in mannan-coated plates with high salt binding buffer overnight at 4°C and removed by washing, and the MBL-MASP-2 complex was captured. Purified C4 and N protein were added and incubated for 1.5 hr, and the deposited C4b was detected following standard protocols. The functional activity of LP and AP was assessed by ELISA as previously described (43). Nunc Maxisorb plates were coated with 10 μg mannan per well in 100 mM Na₂CO₃/NaHCO₃ (pH 9.6) at room temperature overnight. After each step, plates were washed three times with PBST (300 μl/well). Residual binding sites were blocked by incubation with 10 mM Tris-HCl (pH 7.4), 150 mM NaCl, and 2% HSA for 2-3 hr at room temperature. Serum samples were diluted 1:80 in 10 mM Tris-HCl (pH 7.4) containing 150 mM NaCl, 0.5 mM MgCl₂, 0.05% Tween-20, and 0.1% gelatin with or without 2 mM CaCl₂ and N protein. All samples and buffers were prepared on ice. The plates were then sequentially incubated for 1 hr at 4°C and for 1.5 hr at 37°C followed by washing. All incubation volumes were 100 μl. Complement binding was detected using antibodies followed by washing. Detection of C4, activated C3, and C5b-9 was performed using anti-C4α chain antibody (Santa Cruz), anti-activated C3 antibody (Santa Cruz), and anti-C5b-9 antibody (Calbiochem), respectively.

Antibody binding was detected using HRP-conjugated sheep anti-mouse antibody or donkey anti-rabbit antibody (R&D). Enzyme activity of HRP was detected using TMB incubation for 30-60 min RT, and the reaction was stopped with 2 M H₂SO₄. The OD was measured at 450 nm using a microplate reader.

MASP-2:MBL binding assay

Binding of MASP-2 to MBL was assessed by ELISA. As mentioned above, Nunc Maxisorb plates were coated with 10 μg mannan per well in 100 mM Na₂CO₃/NaHCO₃ (pH 9.6) at room temperature overnight and blocked with 2% HSA. MBL protein (1 μg/ml) was incubated in 10 mM Tris-HCl (pH 7.4), 150 mM NaCl, 5 mM CaCl₂, 100 μg/ml HSA, and 0.5‰ TritonX-100 at 4°C for 2 hr. Purified MASP-2 and N (or control) proteins were added to the wells at different times to obtain final concentrations of 0.2 mg/ml and 200 ng/ml, respectively. The plates were washed after 32 hr of incubation at 4°C, and the binding of MASP-2 were detected with anti-MASP-2 antibody followed by HRP-conjugated rabbit anti-goat antibody. The enzyme activity of HRP was detected using TMB incubation for 30-60 min at RT, and the reaction was stopped with 2 M H₂SO₄. The OD was measured at 450 nm using a microplate reader.

Opsonocytophagic assay

Mouse cells isolated from peritoneal cavity were washed and inoculated with RPMI 1640 Media (10% FBS) in 96-well plates for 2 hr at 37°C. Serum was diluted by 0.781%, 1.562%, 3.125%, 6.25%, 12.5%, 25%, 50% and 100% with 1×PBS, 1mM CaCl₂, and 2mM MgCl₂. Diluted serum, SARS-CoV N protein (100ng/ml) and E. coli (the ratio to cells was 10:1) were added to each well and incubated for 30 minutes at 37°C. The eluate from pET22b-transformed E. coli was used as a negative control for purified N protein to exclude effects due to bacterial components that may active complement. The reaction was stop and cells were fixed with 10% neutral formalin. Complement C3c depositon was detected with FITC-C3c antibody, and the stained cells were counted. The points represent the mean values from two repeated wells. Error bars, mean±S.D. *P<0.05 and **P<0.01 by unpaired two-tailed Student’s t-test.

Animal experiments

Groups of BALB/c mice were provided from the experimental animal center of the Academy of Military Medical Sciences. Masp2^-/- (KO) mice and littermate control wild type (WT) mice were provided from Cyagen Biosciences Inc. All mice were maintained in the experimental animal center of the Academy of Military Medical Sciences (China). Mice (8-10/group) were infected three times (day 1, 2, 3) with 1×10^8-9 PFU Ad-N/Ad-null (Beijing BAC Biological Technologies) or a saline control via the tail vein, and LPS (5 mg/kg) was given via the tail vein on the 5-6th day. Anti-MASP-2 monoclonal antibody (200 μg/kg, HBT), anti-N monoclonal antibody (200 μg/kg, Sino Biolgical) or C1INH (4 mg/kg, Calbiochem) was injected via the tail vein 30 min before LPS injection. This study was performed with the approval of the ethics committee at the Beijing Institute of Biotechnology and conformed to the relevant regulatory standards.

Immunohistochemistry

Postmortem autopsy from the 4 patients died in Huoshenshan hospital was carried out by Dr. Xiuwu Bian under the approval of the hospital ethic committee and the family member of the death. The detailed results will be published elsewhere. Paraformaldehyde-fixed lung tissues was used for paraffin tissue sections and immunohistochemical staining with MBL (Santa Cruz), MASP-2 (Santa Cruz), C4α chain (Santa Cruz), C3 (Santa Cruz) or C5b-9 (Calbiochem) antibodies as described in the instruction manual.

Detection of C5a in COVID-19 patients

Sera from mild or severe COVID-19 patients were collected under the approval of hospital ethic committee. Server patients were defined as fever or suspected respiratory infection, plus one of respiratory rate >30 breaths/min, severe respiratory distress, or SpO2 <90% on room air. Patient with pneumonia and no signs of severe pneumonia are defined as mild cases. Sera from 10 mild patients (5 males and 5 females), with an average of 56±12.1 years old and an average 26.3±9.1 days of illness, and sera from 18 severe patients (7 males and 11 females), with an average of 72.4±10.1 years old and an average 40.15±12.74 days of illness, were assayed.

Sera from healthy people for physical were collected from the clinical laboratory. Serum C5a level was detected by double antibody sandwich ELISA (R&D Systems).

Anti-C5a antibody therapy

Mice-Human (IgG4) antibody against human C5a (Bdb-001 injection) was co-developed by Staidson Biopharmaceutical Co., Ltd and inflaRx GmbH. In this study, anti-human C5a antibody (Bdb-001) was provide by Staidson, which was manufactured in accordance with the Good Manufacture Protocol. The injection was approved by National Medical Product Administration (NMDA) of China for phase I clinical trial in healthy subjects in 2019. After the SARS-CoV-2 coronavirus induced pneumonia (COVID-19) outbreak, the antibody was approved by NMDA for phase II clinical trial for the treatment of COVID-19 on February 7, 2020 (2020L00003). Administration of the antibody to human was also approved by the Ethics Committee in Wuhan Huoshenshan Hospital. 300 mg anti-C5a antibody in 250 ml saline was administrated i.v. on day 1, 2, 3, 5, 7, 9 and 11. Arterial blood gas (ABG) test, C reaction protein (CRP), blood routine (blood RT) and Hepatic function were regularly determined. SaO2, blood pressure, heart beat were monitored as required.

Statistics

The data were presented as the mean ±SD and analyzed using the two-tailed Student t test or ANOVA. Survival curves were analyzed by Log-rank test, Gehan-Breslow-Wilcoxon test or parametric regression model. Differences were considered significant at *P <0.05, **P <0.01, or ***P<0.001 as indicated.