C5aR inhibition of nonimmune cells suppresses inflammation and maintains epithelial integrity in SARS-CoV-2–infected primary human airway epithelia

Background Excessive inflammation triggered by a hitherto undescribed mechanism is a hallmark of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infections and is associated with enhanced pathogenicity and mortality. Objective Complement hyperactivation promotes lung injury and was observed in patients suffering from Middle East respiratory syndrome-related coronavirus, SARS-CoV-1, and SARS-CoV-2 infections. Therefore, we investigated the very first interactions of primary human airway epithelial cells on exposure to SARS-CoV-2 in terms of complement component 3 (C3)-mediated effects. Methods For this, we used highly differentiated primary human 3-dimensional tissue models infected with SARS-CoV-2 patient isolates. On infection, viral load, viral infectivity, intracellular complement activation, inflammatory mechanisms, and tissue destruction were analyzed by real-time RT-PCR, high content screening, plaque assays, luminex analyses, and transepithelial electrical resistance measurements. Results Here, we show that primary normal human bronchial and small airway epithelial cells respond to SARS-CoV-2 infection by an inflated local C3 mobilization. SARS-CoV-2 infection resulted in exaggerated intracellular complement activation and destruction of the epithelial integrity in monolayer cultures of primary human airway cells and highly differentiated, pseudostratified, mucus-producing, ciliated respiratory tissue models. SARS-CoV-2–infected 3-dimensional cultures secreted significantly higher levels of C3a and the proinflammatory cytokines IL-6, monocyte chemoattractant protein 1, IL-1α, and RANTES. Conclusions Crucially, we illustrate here for the first time that targeting the anaphylotoxin receptors C3a receptor and C5a receptor in nonimmune respiratory cells can prevent intrinsic lung inflammation and tissue damage. This opens up the exciting possibility in the treatment of COVID-19.


C5aR inhibition of nonimmune cells suppresses inflammation and maintains epithelial integrity in SARS-CoV-2-infected primary human airway epithelia
Wilfried Posch, PhD, a * Jonathan Vosper, PhD, b * Asma Noureen, PhD, a Viktoria Zaderer, MSc, a Christina Witting, a Giulia Bertacchi, MSc, a Ronald Gstir, PhD, c Przemyslaw A. Filipek, PhD, c G€ unther K. Bonn, PhD, c Lukas A. Huber, MD, c,d Rosa Bellmann-Weiler, MD, e Cornelia Lass-Fl€ orl, MD, a and Doris Wilflingseder, PhD a Innsbruck, Austria Background: Excessive inflammation triggered by a hitherto undescribed mechanism is a hallmark of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infections and is associated with enhanced pathogenicity and mortality. Objective: Complement hyperactivation promotes lung injury and was observed in patients suffering from Middle East respiratory syndrome-related coronavirus, SARS-CoV-1, and SARS-CoV-2 infections. Therefore, we investigated the very first interactions of primary human airway epithelial cells on exposure to SARS-CoV-2 in terms of complement component 3 (C3)-mediated effects. Methods: For this, we used highly differentiated primary human 3-dimensional tissue models infected with SARS-CoV-2 patient isolates. On infection, viral load, viral infectivity, intracellular complement activation, inflammatory mechanisms, and tissue destruction were analyzed by real-time RT-PCR, high content screening, plaque assays, luminex analyses, and transepithelial electrical resistance measurements.
Results: Here, we show that primary normal human bronchial and small airway epithelial cells respond to SARS-CoV-2 infection by an inflated local C3 mobilization. SARS-CoV-2 infection resulted in exaggerated intracellular complement activation and destruction of the epithelial integrity in monolayer cultures of primary human airway cells and highly differentiated, pseudostratified, mucus-producing, ciliated respiratory tissue models. SARS-CoV-2-infected 3-dimensional cultures secreted significantly higher levels of C3a and the proinflammatory cytokines IL-6, monocyte chemoattractant protein 1, IL-1a, and RANTES. Conclusions: Crucially, we illustrate here for the first time that targeting the anaphylotoxin receptors C3a receptor and C5a receptor in nonimmune respiratory cells can prevent intrinsic lung inflammation and tissue damage. This opens up the exciting possibility in the treatment of COVID-19. (J Allergy Clin Immunol 2021;147:2083-97.) Key words: SARS-CoV-2, complement, anaphylatoxin receptors, primary human airway epithelial cells Early events occurring directly after severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) transmission to respiratory tissues can influence the outcome in the context of disease severity-in some patients, infection with coronavirus disease 19  results in excessive activation of the immune response at epithelial/immune barriers and the generation of a proinflammatory milieu. 1,2 The development of a cytokine storm and acute lung injury, causing acute respiratory distress syndrome (ARDS), are potential undesirable consequences of the disease. ARDS accompanied by systemic coagulopathy are critical aspects of morbidity and mortality in COVID-19. 3,4 These overshooting immune responses triggered by incoming viruses result in extensive tissue destruction during severe cases, resulting in tissue injury and multiorgan failure. 5,6 Complement may be among the factors responsible for the immune overactivation, because complement deposition and high anaphylatoxin serum levels have been reported in patients with severe/critical disease. 7 As recently shown in transcriptome analyses of bronchoalveolar lavages of patients, the complement system was among the most significantly upregulated intracellular pathways following SARS-CoV-2 infection. 8 In addition, the transcriptomes of primary normal human bronchial epithelial cells infected in vitro with SARS-CoV-2 revealed an enriched complement signature. 8 Very recently, Ramlall et al 9 identified in addition to type I IFN-and IL-6-dependent inflammatory responses, a robust engagement of complement and coagulation pathways following SARS-CoV-2 infection. Here, we further analyzed local complement activation at the protein level in a highly differentiated primary human 3-dimensional (3D) respiratory model on in vitro infection using a SARS-CoV-2 clinical specimen. 10,11 In such cultures, complete destruction of epithelial integrity was observed on viral infection and was associated with exacerbated complement component 3 (C3) production, membrane attack complex (MAC) formation as well as detachment of cells and release of virus into the supernatants. Virus infection was detected not only in outermost layers but virus particles were seen to penetrate deeply into the pseudostratified epithelium as well, with a concomitant C3 activation or probably also deposition on dead or damaged cells being observed in these layers. Besides C3 activation, significantly higher levels of C3a desArg, tissue damage, and proinflammatory cytokines, such as IL-6, monocyte chemotactic protein 1 (MCP-1), IL-1a, and RANTES, were detected in SARS-CoV-2-infected tissue models. Although C3aR blocking alleviated viral infection and reduced the inflammatory response in airways, C5aR antagonism completely restored tissue integrity, significantly downmodulated local complement production, and associated therewith returned C3a and proinflammatory levels back to normal. C5aR antagonism was also shown to correlate with low viral loads. Our results point toward a novel therapeutic intervention strategy whereby complement inhibitors are administered to high-risk patients during an early stage of disease progression to prevent an excessive inflammatory response and its associated pathological consequences.

Ethics statement
Written informed consent was obtained from all donors of leftover nasopharyngeal/oropharyngeal specimens and EDTA blood by the participating clinics. The Ethics Committee of the Medical University of Innsbruck (a copy is attached to the proposal, ECS1166/2020) approved the use of anonymized leftover specimens of patients with COVID-19 for scientific purposes.
Cell culture of tissue models Human airway epithelia. Normal human bronchial epithelial (cat# CC-2540S) and small airway epithelial (cat# CC-2547S) cells are available in our laboratory and routinely cultured in air-liquid interface (ALI) as described. 10,11 Briefly, cells were cultured in a T75 flask for 2 to 4 days until they reached 80% confluence. The cells were trypsinized and seeded onto GrowDexT (UPM)-coated 0.33-cm 2 porous (0.4-mm) polyester membrane inserts with a seeding density of 1 3 10 5 cells per Transwell (Costar, Corning, New York, NY). The cells were grown to near confluence in submerged culture for 2 to 3 days in specific epithelial cell growth medium according to the manufacturer's instructions. Cultures were maintained in a humidified atmosphere with 5% CO 2 at 378C and then transferred to ALI culture. The epithelium was expanded and differentiated using airway media from Stemcell (Germany). The number of days in development was designated relative to initiation of ALI culture, corresponding to day 0. In addition, for real-time RT-PCR analyses of uninfected (UI) and SARS-CoV-2-infected 3D tissues and supernatants, apical-out lung organoids were generated. Briefly, cryopreserved lung organoids derived from human normal lung tissue (left superior lobe) were acquired from Hubrecht Organoid Technology (HUB), Utrecht, The Netherlands (HUB-07-A2-051). These organoids were cultured according to the previously described protocol. 12,13 Because the SARS-CoV-2 receptor angiotensin-converting enzyme 2 is predominantly expressed on the apical side of the airway epithelium, 14 apical-out lung organoids were generated as recently described to expose the apical side of the lung organoids to facilitate viral infection. Apical-out organoids were further differentiated into multiciliated cells by using differentiation medium as previously reported by Zhou et al. 15 Vero/TMPRSS2. VeroE6/TMPRSS2 is an engineered VeroE6 cell line expressing high levels of TMPRSS2 and is highly susceptible to SARS-CoV-2 infection and is used to expand SARS-CoV-2 viruses from repositories as well as patient isolates. The cell line was obtained via the Center for AIDS Reagents (National Institute for Biological Standards and Control) and is described in Matsuyama et al. 16 Transepithelial electrical resistance measurement Transepithelial electrical resistance (TEER) values were measured using EVOM volt-ohm-meter with STX-2 chopstick electrodes (World Precision Instruments, Stevenage, UK). Measurements on cells in ALI culture infected or not with SARS-CoV-2 plus/minus antagonists (1 mM) or treated with recombinant C5a (1 mg/mL) were taken immediately before the medium was exchanged. For measurements, 0.1 mL and 0.7 mL of medium was added to the apical and basolateral chambers, respectively. Cells were allowed to equilibrate before TEER was measured. TEER values reported were corrected for the resistance and surface area of the Transwell filters.

Staining and high content screening
To visualize SARS-CoV-2 infection in monolayers and 3D tissue models, cells were infected with clinical specimen of SARS-CoV-2 and analyzed for characteristic markers on days 2 and 5 postinfection (2 dpi and 5 dpi). For inhibition of C3aR (SB 290157, Calbiochem) and C5aR (mix of W-54011 and DF2593, 1:1, Calbiochem), commercially available antagonists were obtained from Sigma-Aldrich (St Louis, Mo). Tissue cultures were treated 2 hours before infection with 1 mM C3aR-and C5aR antagonists at the basolateral site. After infection, cells and 3D-cell cultures were fixed with 4% paraformaldehyde. Intracellular staining was performed using 13 Intracellular Staining Permeabilization Wash Buffer (103; BioLegend, San Diego, Calif). Antibodies to detect cilia (acetylated tubulin-Alexa488, Cell Signaling Technologies, Danvers, Mass, or wheat germ agglutinin-680, Thermofisher Scientific, Waltham, Mass), mucus (Muc5AC-Alexa647, Cell Signaling Technologies), nuclei (Hoechst 33342, Cell Signaling Technologies), actin (phalloidin-Alexa647, Cell Signaling Technologies), and complement C3 (C3-FITC, Agilent Technologies, Santa Clara, Calif) were used. To investigate terminal complement complex (TCC) formation, the anti-C5b-9 (aE11 clone, Abcam, Cambridge, UK) antibody recognizing a neoepitope exposed on polymerized C9 when incorporated in the human TCC was used. Neoantigens were shown to be expressed in both MAC and fluid phase (SC5-9) complex. 17 Intracellular SARS-CoV-2 was detected using Alexa594-labeled SARS-CoV-2 antibodies against S1 and N (both Sino Biological, Beijing, China). The Alexa594labeling kit was purchased from Abcam. Cells and 3D cultures were thoroughly washed following the staining procedure using the permeabilization buffer, sterile-filtered D-PBS, and finally mounted in Mowiol. To study these complex models using primary cells cultured in 3D and to generate detailed phenotypic fingerprints for deeper biological insights in a high-throughput manner, the Operetta CLS System (PerkinElmer, Waltham, Mass) was applied. Spot analyses, nanotube evaluation, and absolute quantification for C3-, SARS-CoV-2-, or C3-/SARS-CoV-2-containing cells were done using the Harmony software (Perkin Elmer, Rodgau, Germany). For quantification analyses, at least 1000 cells per condition were analyzed as indicated.
Real-time RT-PCR for absolute quantification of SARS-CoV-2 SARS-CoV-2 RNA (140 mL) was extracted using FavorPrep Viral RNA Mini Kit (FAVORGEN, Ping-Tung, Taiwan), according to manufacturer's instructions. SARS-CoV-2 levels were quantified using the primer-probe combinations designed by the Centers for Disease Control and Prevention (CDC) 2019-nCoV Real-Time RT-PCR Diagnostic Panel. Sequences specific to 2 distinct regions of the Nucleocapsid (N) gene, N1 and N2, and for the detection of a human housekeeping gene, Ribonuclease P published on the CDC website (https://www.cdc.gov/coronavirus/2019-ncov/ lab/rt-pcr-panel-primer-probes.html), were used. Single target assays of all 3 targets were performed in combination with the Luna Universal Probe One-Step RT-qPCR Kit (New England BioLabs, Ipswich, Mass). For absolute quantification using the standard curve method, SARS-CoV-2 RNA was obtained as a PCR standard control from the National Institute for Biological Standards and Control UK (Ridge, UK). The SARS-CoV-2 RNA was used in 10-fold dilutions according to the manufacturer's instructions. All runs were performed on a Bio-Rad CFX 96 instrument and analyzed by the Bio-Rad CFX Maestro 1.1 software (Bio-Rad Laboratories, Hercules, Calif).
Real-time RT-PCR for relative quantification of complement factors mRNA expressions of C3, C5, C3aR, C5aRI, and C5aRII were analyzed by real-time RT-PCR using commercially available gene-specific primer pairs (PrimePCR, Bio-Rad, D€ usseldorf, Germany). mRNA expression of Glycerinaldehyd-3-phosphat-dehydrogenase (PrimePCR, Bio-Rad) served as internal control to quantify the relative gene expression of target genes. The Luna Universal One-Step RT-qPCR kit (New England Biolabs) was used for target amplification, and runs were performed on the CFX96 real-time detection system (Bio-Rad). The cycling conditions were as follows: 10 minutes at 558C (room temperature), 1 minute at 958C, 40 cycles: 10 seconds at 958C, 30 seconds at 608C. The results were analyzed using the gene expression software of the cycler (CFX Manager Software, DDCt method; BioRad).

Multiparameter flow cytometry
Expression of complement receptors C3aR, C5aRI, and C5aRII was additionally investigated by flow cytometry using a BD FACSVerse analyzer (BD Biosciences, Franklin Lakes, NJ). For this, human airway epithelia (HAE) cultures were treated with commercially available C3aR (SB 290157) and C5aR (mix of W-54011 and DF2593, 1:1) antagonists, both obtained from Sigma-Aldrich. Tissue cultures were treated for 2 hours with 1 mM C3aR or C5aR antagonists before infection with SARS-CoV-2 at a multiplicity of infection (MOI) of 0.1. After infection, cells were washed and cultured for 48 hours. For flow cytometric analyses, cells were harvested using the animal component-free cell dissociation kit (Stemcell, Vancouver, Canada) and stained with a fixable viability dye (BD Horizon, BD Biosciences) as well as antihuman C3aR, C5aRI, and C5aRII mAbs (Biolegend). After fixation, samples were analyzed using BD FACSuite v1.0.6 and BD FACS Diva v9.0 software (BD Biosciences).

Viruses
Clinical specimens from COVID-19-positive swabs (ethics statement, ECS1166/2018) and SARS-CoV-2 viruses from repositories (BEI Resources, Manassas, Va; Center for AIDS Reagents [National Institute for Biological Standards and Control]; Nr-52281, Nr-52282, NR-52286) were propagated according to the manufacturer's instructions and used subsequently to infect HAE tissue cultures.

Statistical analysis
Statistical analysis of differences in infection levels, TEER values, or cytokine production was performed using the GraphPad prism software and using 1-way ANOVA with Dunnett posttest or unpaired Student t test. All experiments (infection experiments, RT-PCR, FACS, plaque assays, TEER measurements) were independently repeated at least 3 times. For quantification of images, at least 2000 cells were analyzed.

SARS-CoV-2 infection significantly increases intracellular C3 levels in primary human airway epithelial cells
To monitor SARS-CoV-2 infection in primary normal human bronchial epithelial cells, monolayers and fully differentiated pseudostratified epithelia were infected using different clinical isolates derived from patients with SARS-CoV-2. 10,11 The clinical specimens were anonymized before use, and the study has been approved by the Ethics Committee of the Medical University of Innsbruck (approval no. ECS1166/2020). In all cases, clear cytopathic effects (CPEs) were apparent, with the attendant detachment/floating of cells and syncytium formation clearly observable in Vero/TMPRSS2 cells (see Fig E1 in this article's Online Repository at www.jacionline.org) as well as in human airway epithelial cells (Fig 1, A). Viral RNA copies of clinical specimens and after expansion in Vero/TMPRSS2-expressing cells were absolutely quantified by real-time RT-PCR, and virus titers were found to be 3.  [18][19][20] HAE cells in monolayers were susceptible to SARS-CoV-2 infection (red) using clinical specimens as a source of the virus (Fig 1, A), and surprisingly SARS-CoV-2 infection in these primary airway epithelial cells was accompanied by extensive induction of intracellular (IC) C3 (Fig 1, A, and C; green). Absolute quantification using less than 1000 cells per condition (UI vs infected with SARS-CoV-2) revealed significantly higher C3 levels in infected cultures, whereas UI cultures exhibited only background IC C3 levels (Fig 1, A). Expression of complement components C3 and C5 was analyzed by real-time RT-PCR in UI and SARS-CoV-2-infected HAE cultures (Fig 1,  . Consistent with the microscopic data, expression of C3 was significantly higher in SARS-CoV-2-infected cultures, whereas no differences in expression of C5 were observed (Fig 1, B). These analyses demonstrate that intracellular C3 is massively induced at protein and gene expression levels in SARS-CoV-2-infected, primary HAE cultures.

SARS-CoV-2 infection mobilizes intracellular C3 in both infected cells and neighboring cells
Next, we characterized the distribution of intracellular C3 in infected HAE monolayer cultures. We found that IC C3 was produced by infected cells and in particular also their neighboring cells (Fig 1, C; green). High-content image analyses revealed that viral particles (Fig 1, C, red, arrows) were transmitted in C3containing, activated cells via membrane nanotube formation to neighboring cells (Fig 1, C), a process also used by HIV-1 to exacerbate infection. 21 To quantify nanotubes and C3/SARS-CoV-2containing nanotubes in UI and SARS-CoV-2-infected cells, digital phase contrast microscopy on a total number of 5000 cells and concomitant analyses using the Harmony software (Perkin Elmer) were performed (Fig 1, C, bar graphs). This revealed a significantly higher number of nanotubes and C3/SARS-CoV-2 nanotubes in SARS-CoV-2-infected versus UI cells (Fig 1, C, bar graphs). This indicates that SARS-CoV-2 is shuttled via C3-containing nanotubes to neighboring cells.
Immunofluorescence staining revealed a significant infection of the tissue models with SARS-CoV-2 (Fig 2, C, and Fig E3, red), with both ciliated and mucus-producing cells being affected by the virus (Fig E3; red). From infected foci, the C3 signal spread over the epithelia to neighboring cells (Fig 2, C; green) as was also observed in HAE monolayers. Quantitation of cell numbers in the respiratory tissue models revealed a significant loss of cells in SARS-CoV-2 versus UI on 2 dpi (Fig 2, D, first chart, table): 42.3% versus 2.6% of the analyzed cells were positive for C3 in SARS-CoV-2-infected versus UI cultures (Fig 2, D, second chart, table); 27% were positive for SARS-CoV-2 in infected cultures; and 3.3% showed background fluorescence staining in UI (Fig  2, D, third chart). About half of the SARS-CoV-2-positive cells (15%) were also stained positive for C3 in infected cultures (Fig 2, D, fourth chart, table). A total number of more than 5000 cells was analyzed for each condition. 3D and xyz analyses of the 2 signals revealed not only a superficial localization of SARS-CoV-2 and C3 but also that the infection and intracellular complement mobilization penetrated deep into the tissue layers as illustrated in Figure 2, C (upper right image). In contrast, only background signals for SARS-CoV-2 and C3 were apparent in the UI control (Fig 2, C and D) and these epithelia displayed a higher degree of structural integrity (Fig 2, C, lower right image, UI).
Disruption of epithelial integrity is contingent on extensive C3 mobilization, TCC formation, virus release, and C3a secretion Infection of respiratory tissue models with clinical specimens of SARS-CoV-2 and the concomitant considerable IC C3 mobilization resulted in substantial disruption of the epithelial integrity as assessed by measuring TEER (Fig 3, A; left), release of SARS-CoV-2 into the medium (Fig 3, A, middle), C3a desArg secretion (Fig 3, A, right), and imaging of culture models (Fig 3,  B). In UI tissue culture, TEER values ranged from 1101 to 1358 V/cm 2 , whereas this range dropped to 270 to 710 V/cm 2 in SARS-CoV-2-infected cells (Fig 3, A; left). SARS-CoV-2-infected 3D cultures released significantly higher virus particles into the medium (Fig 3, A, left and middle) probably due to extensive syncytium formation 22 and damage of cells as also described in postmortem samples from individuals who died from COVID-19. 23 Moreover, infected cultures secreted significantly higher levels of the anaphylatoxin C3a compared with mock-treated uninfected cultures (Fig 3, A, right, UI). Image analyses of SARS-CoV-2-infected cells revealed the disruption of ciliated structures, which were clearly visible in UI models (Fig 3, B, wheat germ agglutinin, orange). UI cultures were stained positive for cilia (Fig 3, B; upper panel, orange) and nuclei (Fig 3, B; upper panel, blue). Furthermore, SARS-CoV-2 was not detected in UI  cells when using Alexa594-labeled antibodies recognizing the spike 1 (S1) or nucleocapsid (N) proteins (Fig 3, B; upper panel, red). Only background levels of C3 were detectable in UI cultures (Fig 3, B, upper panel, green). In contrast, ciliated structures were destroyed in SARS-CoV-2-infected cultures and became apparent already at 2 dpi (Fig 3, B; middle panel, orange) and were aggravated by ongoing infection and illustrated at 5 dpi (Fig 3, B; lower panel, orange). Infected cultures stained positive for the S1 and N proteins of SARS-CoV-2 (Fig 3, B; red), and C3 (Fig 3, B; green) was heavily induced in such tissue models. To further characterize whether MAC membrane channels might be involved in decreased epithelial resistance, we stained UI and SARS-CoV-2-infected cultures using an anti-C5b-9 antibody recognizing a neoepitope exposed on polymerized C9 when incorporated in the human TCC. 17 The C5b-9 antibody colocalized in many SARS-CoV-2-infected cells (Fig 3, C, white arrows, and Fig 3, D, SARS-CoV-2), whereas in UI cultures only background signals were obtained using these antibodies (Fig 3, C and D, UI). Brightfield images (Fig 3, C, left), staining of acetylated tubulin-recognizing ciliary structures (Fig 3, C, right, pink), and C5b-9-single positive cells (Fig 3, C, right, green arrows) additionally illustrated rounding up of cells and loss of structured cilia in infected cultures. All experiments were independently repeated at least 3 times, and quantitation was performed in 4 different fields on at least 2000 cells (Fig 3, D). These data revealed that local complement activation in pseudostratified HAE cells correlated with the disruption of the epithelial integrity following SARS-CoV-2 infection.

Effective reduction of local complement after SARS-CoV-2 infection by antagonizing of C3aR or C5aR on HAE cells
Next, we used our highly differentiated HAE model to investigate possible treatment options that might circumvent the induction of the synthesis of complement proteins by SARS-CoV-2, productive virus infection, and the concomitant secretion of high levels of anaphylatoxin and associated tissue disruption. To this end, we performed microscopic analyses and real-time RT-PCR of culture models as well as plaque assays of basolateral supernatants from the same cultures pretreated with C3aR or C5aR inhibitors on the basolateral side before addition of SARS-CoV-2 clinical isolates at an MOI 0.1 on the apical side. Before antagonizing C3aR and C5aR on HAE cells, we characterized C3aR, C5aR1, and C5aR2 expression on HAE monolayers by flow cytometry (see Fig E4 in this article's Online Repository at www.jacionline.org) and 3D tissue models by high content screening (see Fig E5 in this article's Online Repository at www.jacionline.org). High C5aR2 expression was determined on both monolayers and tissues (Figs E4 and E5), whereas C3aR and C5aR1 were expressed only on 3D models (Fig E5). After verifying the presence of the various anaphylatoxin receptors in our system, cells were infected for 2 days and then stained for SARS-CoV-2 S1/N (red), intracellular C3 (green), and nuclei (blue) for microscopic analyses (Fig 4, A). As demostrated before, SARS-CoV-2-infected HAE tissue models showed a high level of infection (Fig 4, A; SARS-CoV-2, red) and intracellular C3 induction (Fig 4, A; SARS-CoV-2, green) as well as disruption of the epithelium compared with the uninfected cultures (Fig 4, A; UI). Basolateral preincubation of epithelia with C3aR antagonist resulted in reduction of viral loads as well as of C3-positive cells and preserved a more intact morphology of the human airway epithelium (Fig 4, A; C3aR-Inh/SARS-CoV-2). In addition to imaging analyses, basolateral supernatants of pretreated and infected HAE cultures were harvested on 2 dpi to perform a plaque assay using TMPRSS2-expressing VeroE6 (VeroE6/ TMPRSS2) cells for determining viral infectivity from these cultures (Fig 4, B). Furthermore, the virus concentration was analyzed by quantitative real-time RT-PCR (Fig 4, C). Both plaque assay and real-time RT-PCR revealed a slight decrease in PFUs or SARS-CoV-2 copy numbers after treatment with C3aR antagonist (C3aR Inh/SARS-CoV-2), but compared with nontreated, infected cells (SARS-CoV-2) this reduction was statistically not significant (Fig 4, B and C).The most convincing results in terms of antagonizing SARS-CoV-2-mediated intracellular C3 production were achieved using a C5aR antagonist (Fig  4, A-D; C5aR Inh/SARS-CoV-2). Only background levels of virus-infected cells comparable to the background signal in UI cultures were detected (red) and tissue integrity was maintained in C5aR antagonist-treated, SARS-CoV-2-infected HAE cultures (Fig 4, A; C5aR Inh/SARS-CoV-2). Similar results were also obtained from plaque assays and RT-PCR (Fig 4, B and C). Pretreatment of HAE cultures with C5aR antagonist (C5aR Inh/ SARS-CoV-2) significantly reduced viral titers compared with infected cultures (SARS-CoV-2) in basolateral supernatants (Fig 4,  B) and also led to a reduced number of infective viral particles (Fig  4, C). Compared with SARS-CoV-2-infected cultures, we found a highly significant reduction in C3a desArg in both C3aR-and C5aR antagonist-treated and SARS-CoV-2 infected cells (Fig  4, D; C3aR Inh/SARS-CoV-2 and C5aR Inh/SARS-CoV-2). Blocking C5aR on HAE cells that were subsequently infected with SARS-CoV-2 resulted in a significant decrease in C3a de-sArg levels compared with HAE cultures pretreated with C3aR inhibitor and then infected, confirming the results seen in imaging analyses (Fig 4, D; C3aR Inh/SARS-CoV-2 and C5aR Inh/ SARS-CoV-2). Here, we illustrated that SARS-CoV-2-induced excessive local complement activation can be decreased by antagonizing C3aR and is completely reversed by blocking C5aR. N1 on 2 dpi. No viral RNA was analyzed in UI tissues. Four independent experiments were performed, and statistical significance was analyzed using unpaired Student t test. B, Plaque assays of SARS-CoV-2 supernatants from HAE cells were performed on Vero/TMPRSS2 cells. Assays were independently repeated 6 times, and unpaired Student t test was performed to evaluate statistical significance. C, Multilayered epithelia were infected using SARS-CoV-2. On 2 dpi, tissues were stained for Hoechst (blue), SARS-CoV-2 S1/N (red), complement C3 (green), and WGA (orange). Representative images of infected regions and

Protection against SARS-CoV-2 infection by inhibiting C5aR
To determine whether C3aR/C5aR antagonism is potentially an effective postexposure preventative mechanism, we infected differentiated HAE tissue models with SARS-CoV-2 clinical isolates and added the C3aR and C5aR antagonists following a 6-hour delay. Cells were again analyzed after 2 days. As depicted in Fig 5, A, SARS-CoV-2 efficiently infected the culture models and caused tissue disruption (Fig 5, A; SARS-CoV-2), whereas uninfected control cells displayed an intact tissue morphology as illustrated by a uniform phalloidin staining without disruptions (Fig 5, A; UI). Addition of C3aR inhibitor (Fig 5, A; C3aR Inh/ SARS-CoV-2) led to decreased viral load, but the reduction in local C3 production was not as obvious as that observed on preadministration of antagonist. C5aR antagonism resulted in decreased SARS-CoV-2 infection and lower C3 levels (Fig 5, A; C5aR Inh/ SARS-CoV-2). To investigate whether the delayed C3aR/C5aR antagonism is also associated with protection from SARS-CoV-2 infection, TMPRSS2-expressing VeroE6 (VeroE6/TMPRSS2) cells were infected using the supernatants from such differentially treated HAE cells. Cells were monitored for CPEs. Clear CPEs with detachment/floating and syncytium formation developed 1 dpi, and all cells were detached in SARS-CoV-2 and surprisingly also in C3aR antagonist-/SARS-CoV-2-exposed cells on 2 dpi. In contrast, antagonizing C5aR protected the cells from CPEs after SARS-CoV-2 infection and they looked similar to uninfected controls (Fig 5, B) and C5aR antagonist-treated control cells (not shown). Hence, we showed that inhibiting C5aR on fully differentiated HAE cells prevented subsequent productive SARS-CoV-2 infection in highly susceptible cells.

Maintaining epithelial integrity following SARS-CoV-2 infection by inhibition of C5aR on differentiated HAE cells
Because C5aR antagonism appeared to be the most promising strategy for protection against SARS-CoV-2 infection due to the downregulation of local complement activation, we performed more detailed analyses in such cultures to account for the absence of infection observed in susceptible indicator cells. We found that TEER values could be restored in SARS-CoV-2-infected HAE cells treated with C5aR inhibitor (Fig 6, A). Although tissue models were strongly disrupted on SARS-CoV-2 infection as indicated by a significantly lower TEER value compared with UI (Fig 6, A; UI), a highly significant improvement in epithelial integrity was observed in C5aR-treated as well as C5aR-and SARS-CoV-2-exposed cultures (Fig 6, A; C5aR Inh and C5aR Inh/SARS-CoV-2). Also, C3-containing nanotube levels were reduced to zero in HAE cultures pretreated with C5aR antagonist before SARS-CoV-2 infection comparable to UI and C5aR Inh/UI and in contrast to SARS-CoV-2-infected cultures (Fig 6, B). To further characterize whether C5a alone can mediate damage of HAE cultures, cells were treated with 1 mg/mL C5a for 24 hours, but not infected with SARS-CoV-2 (Fig 6, C). These experiments revealed that TEER values of HAE tissues significantly dropped on treatment with C5a compared with untreated control cells (Fig 6, B, UI), and microscopic analyses also showed disruption of tissue integrity of these cultures (see Fig E6 in this article's Online Repository at www.jacionline.org). This pointed toward a role of local complement activation in amplifying viral load and worsening tissue injury in patients with COVID-19.
C5aR antagonism correlates with downmodulation of the proinflammatory signature following SARS-CoV-2 infection Finally, we analyzed the proinflammatory signature of HAE tissue cultures during SARS-CoV-2 infection in the presence and absence of C5aR antagonist. As controls, tissue models were left untreated (UI) or treated with C5aR inhibitor only (C5aR Inh/UI) (Fig 6, D). As assessed by luminex technology and median fluorescence intensities, SARS-CoV-2 induced significantly higher levels of IL-6, MCP-1, IL-1a, and RANTES compared with uninfected control cultures (Fig 6, D; SARS-CoV-2 vs UI). IP10 levels fluctuated in various experiments following SARS-CoV-2 infection, with an overall modest decrease observed in C5aR Inh-pretreated cultures (Fig 6, D). Induction of all proinflammatory cytokines by SARS-CoV-2 infection was reversed by antagonizing C5aR, and such pretreated cultures showed levels similar to uninfected cultures (Fig 6, D; C5aR Inh/SARS-Cov-2 and UI, C5aR Inh/UI). Blocking C5aR may therefore be a possible treatment option that could prevent exaggerated anaphylatoxin and cytokine responses and eventually bring the viral load under control due to the maintenance of an intact epithelium.

DISCUSSION
The entire airway epithelium from the nasal cavity down to the alveoli constantly encounters pathogens and if invaders are able to breach the mucus-producing, ciliated, pseudostratified barrier, an immediate innate immune response is needed. In this context, complement proteins have been demonstrated to be produced and secreted by airway epithelial cells. 24 Clinical evidence has been provided that complement may play a pathogenic role in COVID-19, and recently Yan et al 8 illustrated that SARS-CoV-2 infection augments C3 gene transcription locally in lung epithelial cells and that SARS-CoV-2-induced C3 results in local biologically active C3 products. 8 Normally, complement should be Disruption of epithelial integrity is a combination of extensive C3 mobilization, C3a secretion, TCC formation, and virus production. A, TEER in V/cm 2 (left), SARS-CoV-2-RNA copy numbers/mL (middle), and C3a (right) levels were determined for both UI and SARS-CoV-2-infected 3D airway epithelia and plotted on a bar graph. Four to 8 independent experiments were performed, and statistical significance was analyzed using unpaired Student t test. B, 3D airway epithelia were fixed and stained for Hoechst (blue), SARS-CoV-2-S1/N (red), C3 (green), and WGA (orange) and analyzed at 2 (middle panel) and 5 dpi (lower panel). The upper row represents UI cells. Experiments were independently repeated 3 times, various fields were analyzed, and representative images are shown. C, Multilayered epithelia were stained using C5b-9 Ab   C5aR (mix of W-54011 and DF2593, 1:1) inhibitors added basolaterally and infected apically using SARS-CoV-2 (MOI 0.1-1). Cells were fixed and stained for Hoechst (blue), SARS-CoV-2 S1/N (red), and C3 (green) on 2 dpi. Two independent experiments are shown (upper and lower panels) B, Plaque assays of supernatants from UI and infected cultures plus/minus C3aR and C5aR inhibitor were performed on Vero/TMPRSS2 cells. Assays were independently repeated 3 times, and 1-way ANOVA using Dunnett posttest was performed to evaluate statistical significance. C, Viral RNA was analyzed in UI and infected tissues plus/minus inhibitors. The experiment was repeated thrice, and 1-way ANOVA using Dunnett posttest was performed. D, C3a level determination in C3aR Inh-and C5aR Inh-treated SARS-CoV-2-infected pseudostratified epithelia. Basolateral supernatants (2 dpi) were harvested and C3a levels determined using a BD Biosciences OptEIA Human C3a ELISA kit. C3a levels (ng/mL) were determined for each experimental condition and plotted on a bar graph. inhibitors. Multilayered epithelia were infected by apical addition of SARS-CoV-2 (MOI 0.1-1) and after 6 hours C3aR Inh or C5aR Inh was administered basolaterally. Cells (2 dpi) were fixed and stained for Hoechst (blue), SARS-CoV-2 S1/N (red), complement C3 (green), and phalloidin (orange) and then analyzed by confocal microscopy. Different experimental conditions are indicated by the white text underneath the images. B, BF microscopy analysis of VeroE6/TMPRSS2 cells reinfected with supernatants SARS-CoV-2-infected pseudostratified epithelia treated 6 hours postinfection with C3aR or C5a receptor inhibitors. Supernatants from SARS-CoV-2-infected pseudostratified epithelia treated 6 hours postinfection with C3aR or C5a receptor inhibitors were used to infect VeroE6/TMPRSS2 cells, which were fixed and visualized by BF confocal microscopy 2 dpi. Text underneath images indicates the corresponding experimental conditions. Independent experiments were repeated at least 3 times. BF, Brightfield.

FIG 6.
Maintenance of epithelial integrity and reduced proinflammatory response of SARS-CoV-2-infected cells treated with C5aR inhibitor. A, Multilayered epithelia were infected by apical addition of SARS-CoV-2 (MOI 0.1-1) and following 6 hours were then supplemented basolaterally with either 1 mM C3aR (SB 290157) inhibitor or 1 mM C5aR (mix of W-54011 and DF2593, 1:1) inhibitor. TEER measurement was performed on 2 dpi for each experimental condition, and the results were plotted on a bar graph. Experiments were performed thrice and statistical significance evaluated using 1-way ANOVA and Dunnett posttest. B, C3containing nanotubes were reduced to zero in C5aR inhibitor-treated cultures also on SARS-CoV-2 infection protective during viral infections, but with respect to COVID-19 local complement activation might worsen tissue injury. 25,26 Here, we found that exaggerated C3 production, C3a desArg secretion, and TCC formation in SARS-CoV-2-infected primary human airway epithelial tissue cultures initiated a highly inflammatory microenvironment, resulting in significant tissue damage despite the lack of other immune components. Also, neighboring cells not directly infected with SARS-CoV-2 showed elevated amounts of C3 as well as TCC staining, which might indicate dead or damaged cells. On interaction of SARS-CoV-2 with pseudostratified epithelia, a massive intracellular C3 production was observed with concomitant secretion of the anaphylatoxin C3a desArg, formation of TCC and release of virions, destruction of tissue culture integrity, and hyperactivation of proinflammatory cytokines (IL-6, MCP-1, IL-1a, RANTES). This could be reversed by blocking C5aR localized to primary human airway epithelial cells, which completely antagonized virus-induced tissue damage, virus replication as analyzed by viral RNA in supernatants and plaque assays, and local complement activation, even if added after infection of pseudostratisfied epithelia with SARS-CoV-2.
Treatment of 5 severely ill COVID-19 patients with the C3 antagonist AMY-1 or adjunct treatment with the C5 activation inhibitor eculizumab resulted in disease recovery and point toward a pathogenic role of locally produced complement during COVID-19. 27 A study using human immune cells and human C5aR1 knock-in mice demonstrated that anti-C5aR1 therapeutic mAbs prevented C5a-mediated activation and recruitment of cells and inhibited acute lung injury. 25 Worsening of injury by local complement activation in the airway epithelium was already shown for SARS-CoV infection. In addition, increased C3a and C5a levels in plasma and lung homogenates have been implicated in the pathogenesis of various lung conditions including cystic fibrosis and idiopathic pulmonary fibrosis. 28,29 Thus, increased anaphylatoxin levels resulted in downmodulation of regulators of complement activation such as CD55 and CD46 and changes in injury markers on HAE cells, such as decreased E-cadherin, and increased cleaved poly-ADP ribose polymerase. 28 Furthermore, in patients with idiopathic lung injury, increased soluble C5b-9 levels were analyzed in lung lavage fluid. 24 We, too, found a significantly enhanced induction of TCC formation in SARS-CoV-2-infected tissue models compared with UI controls, suggesting that cell damage also occurs by formation of MAC membrane channels in infected cells. In addition, inhibition of C3aR or C5aR in bleomycin-injured mice with fibrotic lungs led to downregulation of local complement production as well as disease progression. 24,30 Here, we demonstrated that local complement production and progression of SARS-CoV-2 infection could be arrested in primary HAE by C5aR antagonist. This was even the case if the C5aR blocker was administered several hours after SARS-CoV-2 infection. Besides decreasing locally produced complement, blocking C5aR resulted in the restoration of transepithelial electrical resistance, a marker of epithelial integrity. Furthermore, we observed a normalization of proinflammatory cytokine levels back to those of healthy untreated HAE tissue models or cultures incubated with the antagonists alone. In COVID-19-positive individuals, the levels of 10 circulating cytokines (IL-1a, IL-1RA, IL-6, IL-10, GM-CSF, IP-10, MCP-1, RANTES, TSLP, and TNF-a) were reported as being significantly elevated, and different expression profiles were associated with different disease severity. 31 Inflammation is among the first coordinated lines of defense following tissue damage by either infection or injury, but excessive immune responses convert the protective mechanism into a harmful one as observed in patients with COVID-19. 32 Besides the aforementioned biomarkers for disease severity, the anaphylatoxins C3a and C5a secreted by airway epithelial cells or C3 in mucus might also serve as valuable indicators for disease severity or progression. As illustrated in our highly differentiated, pseudostratified HAE, extremely high levels of C3a desArg are secreted from the airway epithelium on interaction with SARS-CoV-2. Both C3a desArg and C5a comprise important effector molecules attracting, activating, and regulating components of innate and adaptive immunity. 33 C5a was demonstrated to induce expression of IL-1b and IL-8 in mononuclear cells and augments the release of IL-6 and TNF-a. 34 Furthermore, several studies illustrated that C5a and C5aR may be responsible for inflammation in the lung, that is, during IgG-mediated inflammation or during ARDS, 35,36 and blocking C5aR attenuated IgG-complex-mediated inflammation. Although we were not able to detect C5a in basolateral supernatants of SARS-CoV-2-infected epithelia, this observation might be attributable to its previously described intracellular engagement and capacity to activate C5aR1 and induction of inflammasome activity. 37 Although this ''complement-metabolism-inflammasome'' axis (reviewed in Arbore et al 38 ) was first demonstrated in immune cells such as CD4 1 T cells, it is also conceivable that such pathway is also activated in the airway epithelium, showing expression of C5aR1 at the protein and the mRNA levels. 39,40 Herein, all anaphylatoxin receptors (C3aR, C5aR1, C5aR2) were shown to be expressed on 3D ALI tissue models. As recently reviewed in Kulkarni et al, 41 all components of lectin and alternative complement pathways are expressed in or produced by human airway epithelial cells, making it likely that on SARS-CoV-2 infection, exacerbating or chronic amplification of local synthesis of these complement proteins putatively propagates epithelial injury. Furthermore, we found a significantly higher expression of proinflammatory cytokines following stimulation of HAE tissue models with SARS-CoV-2, which is in line with C5a effector functions. Importantly, the upregulation of proinflammatory cytokines IL-6, MCP-1, IL-1a, or RANTES was efficiently blocked by C5aR antagonism. Besides C5a, C3a desArg may bind C5aR2 at a site distinct from the C5a binding site, [42][43][44] but this issue was controversially discussed over years 45 as well as the role of C5aR2 as either immune-activating or immune-dampening (reviewed in Li et al 46 ). Because we found (C5aR Inh/SARS-CoV-2). A total of 5000 cells were counted for their presence of nanotubes (DPC), C3, and SARS-CoV-2 under the indicated conditions. C, Addition of recombinant C5a significantly reduced the epithelial integrity as analyzed by TEER. Experiments were performed thrice and statistical significance evaluated using unpaired Student t test. D, Supernatants from SARS-CoV-2-infected pseudostratified epithelia subsequently treated with C5aR antagonists were harvested 2 dpi and cytokine profiles determined using Luminex technology. Levels of cytokines are expressed in median fluorescence intensities. Two independent experiments in duplicate were performed. Statistical significance was evaluated using 1-way ANOVA and Dunnett posttest. DPC, Digital phase contrast. = J ALLERGY CLIN IMMUNOL VOLUME 147, NUMBER 6 a highly elevated induction of C3a desArg on SARS-CoV-2 infection and abundant expression of C5aR2, some of the observed effects may be due to interaction of these complement components. More detailed analyses of these interactions during SARS-CoV-2 infection are necessary, but would have exceeded the scope of this study. Yet, these data provide a promising base for future translational studies of modifying C3a desArg or C5a/C5a desArg and C5aR1 or C5aR2 interactions.

Conclusions
We propose that the application of complement inhibitors represents a promising approach in SARS-CoV-2 disease management. Complement inhibitors (eg, AMY-101, eculizumab, or its variant ravulizumab) are currently in advanced stages of SARS-CoV-2 clinical trials and can be administered to the lungs through nebulizers. [47][48][49] Beside anaphylatoxin targeting, another yet unexplored option is the targeting of regulators of complement activation such as CD46 or CD55 in the lungs of patients with COVID-19 to reduce inflammation. 48,50 As illustrated herein and also reviewed in Jodele and Koehl, 7 complement may serve as a powerful new therapeutic target for COVID-19 treatment, because severely ill patients with COVID-19 with atypical ARDS showed an extensive complement activation both systemically and within the lungs. 7,8,[51][52][53] Crucially, we show here for the first time that targeting the anaphylotoxin receptors C3aR and C5aR in nonimmune respiratory cells can prevent intrinsic lung inflammation and tissue damage. This opens up the exciting possibility that complement inhibitors can be applied in the treatment of ARDS in patients with severe COVID-19. FIG E2. Determination of TCID50 of clinical SARS-CoV-2 isolates used in this study. The absolute levels of viral RNA derived from clinical specimens and then subsequently propagated in Vero/TMPRSS2 cells were quantified using real-time RT-PCR with SARS-CoV-2-specific primer/probe pairs for the nucleocapsid N1 and N2 regions. A representative PCR quantification from 3 different isolates using the N2 primer/probe set is illustrated on the left, with the standard curve shown on the right. Tables below depict log starting quantities and TCID50. Using these isolates, a TCID50 on Vero/TMPRSS2 cells was performed and copy numbers calculated to infect 50% of the tissue cultures. TCID50, Tissue culture infectious dose 50.
FIG E3. SARS-CoV-2 can infect both ciliated and mucus-producing cells in an HAE model. HAE cells were infected with SARS-CoV-2 at an MOI of 0.1. Cells (2 dpi) were fixed and stained for Hoechst (blue), SARS-CoV-2 S1/N (red), acetylated tubulin (green), and mucus-producing cells (orange) and then analyzed by confocal microscopy. On the left, a representative overview of HAE cells stained with Hoechst for nuclei (blue), acetylated tubulin (green), mucus-producing cells (orange), and virus (red) is depicted; on the right, the Hoechst channel was switched off to more clearly show viral distribution over the epithelium.
FIG E4. C5aR2 is the most abundantly expressed anaphylatoxin receptor on HAE monolayers. HAE monolayers were analyzed for their expression of C3aR, C5aR1, and C5aR2 by flow cytometry. Monolayers of HAE cells solely express higher levels of C5aR (dark gray) compared with isotype controls (light gray), whereas only low levels of C3aR and C5aR1 were analyzed. Experiments were repeated thrice independently.