Complement Activation Contributes to Severe Acute Respiratory Syndrome Coronavirus Pathogenesis

The complement system is a critical part of host defense to many bacterial, viral, and fungal infections. It works alongside pattern recognition receptors to stimulate host defense systems in advance of activation of the adaptive immune response. In this study, we directly test the role of complement in SARS-CoV pathogenesis using a mouse model and show that respiratory disease is significantly reduced in the absence of complement even though viral load is unchanged. Complement-deficient mice have reduced neutrophilia in their lungs and reduced systemic inflammation, consistent with the observation that SARS-CoV pathogenesis is an immune-driven disease. These data suggest that inhibition of complement signaling might be an effective treatment option following coronavirus infection.


S evere acute respiratory syndrome coronavirus (SARS-CoV) emerged in 2002 and
2003 from coronaviruses circulating in animal markets in China (1). Emergence of this novel virus led to a global outbreak of respiratory disease, with over 8,000 human cases and 10% mortality (2,3). In 2012, a new, related zoonotic coronavirus was identified in the Middle East, designated Middle East respiratory syndrome coronavirus (MERS-CoV), causing severe respiratory disease with greater than 35% mortality (www .who.int/emergencies/mers-cov/en) (4). Both SARS-CoV and MERS-CoV cause a range of disease from asymptomatic cases to severe acute respiratory distress syndrome (ARDS) and respiratory failure (5). Notably, metagenomics and synthetic virus recovery strategies have since revealed the existence of large pools of preepidemic SARS-like bat coronaviruses which replicate in primary human airway epithelial cells. These viruses are poised for emergence because they both efficiently use human ACE2 entry receptors and resist existing vaccines and immunotherapeutics (6,7). Due to the ongoing threat and continued emergence of new, highly pathogenic coronaviruses from animal reservoirs, a thorough understanding of the host-virus interactions that drive SARS-CoV pathogenesis will aid the public health response to current and future coronavirus outbreaks (8).
The importance of complement in SARS-CoV pathogenesis is controversial. Previous studies have investigated the role of known polymorphisms in the mannose-binding lectin (MBL) and MBL-associated serine protese-2 (MASP2) genes in SARS-CoV infection outcome following the 2003 outbreak but with conflicting results. One retrospective analysis showed that people with low or deficient serum MBL levels were more likely to become infected with SARS-CoV (9) than those with high MBL levels, suggesting that MBL and complement activation play a role in protecting the host from infection. However, a second study found no association between MBL haplotype and SARS-CoV infection status (10). Additionally, it was shown MBL can bind to the SARS-CoV Spike protein in vitro by some groups (11) but not by others (12). Examination of the role of the downstream complement gene MASP2 found no association between genotype and SARS susceptibility (13). Together, the results leave a general uncertainty about the role of complement in response to SARS-CoV infection.
Despite the existing body of literature, the role of complement in SARS-CoV pathogenesis has never been directly assessed in vivo. The complement system is an ancient arm of the innate immune response comprised of multiple proteins whose reactive cascade of cleavage products can coordinate the inflammatory response at the sites of infection and can be directly antimicrobial. Consisting of more than 30 soluble and cell surface-associated proteins, complement is a major component of innate immunity that functions to recognize and eliminate invading pathogens (14). Activation of the complement system occurs through multiple mechanisms that include three welldescribed pathways, the classical, lectin, and alternative complement activation pathways (15), and results in proteolytic processing of various components of the complement system, including C3, C4, and C5. Proteolytic processing of C3 generates an array of cleavage products that are involved in amplification of complement activity through formation of C3 and C5 convertases, opsonization of pathogens, and attraction and activation of leukocytes of both the innate and adaptive arms of the immune response. Several studies, including a recent study showing that complement blockade results in reduced disease in a MERS-CoV human DPP4 transgenic (hDPP4-Tg) mouse model (16), have elucidated protective and pathogenic roles for the complement system following infection by a variety of viral pathogens (17). Furthermore, the complement system has well-described roles in other pulmonary diseases (18), especially after influenza virus and respiratory syncytial virus infection (19)(20)(21).
In this study, we assessed the role of the complement system in the pathogenesis of SARS-CoV infection. Building from a systems biology analysis that suggested that complement was modulated during SARS-CoV infection, we confirmed that complement was activated upon SARS-CoV challenge. Mice deficient in C3 (C3 -/-), the central protein of the complement signaling pathway, were protected from SARS-CoV-induced weight loss and had reduced pathology, improved respiratory function, and lower levels of inflammatory cytokines/chemokines in the lung and periphery. Importantly, the kinetics and magnitude of virus replication in C3 -/and wild-type mice were the same, showing that complement does not play a role in controlling virus replication. We observed complement deposition in the lungs of SARS-CoV-infected mice, suggesting that complement activation results in immune-mediated damage to the lung. Additionally, serum activation indicates that complement-mediated systemic inflammation may drive the pathogenic response to SARS-CoV infection. Together, the results indicate that complement plays a critical role in SARS-CoV pathogenesis and that inhibition of the complement pathway might be an effective therapeutic to coronavirus-mediated disease.

RESULTS
Complement is activated in SARS-CoV MA15-infected mice. While work by other laboratories has shown that C3-deficient mice are extremely susceptible to both H5N1 and H1N1 influenza virus infection (22), the role of complement in SARS-CoV infection has not yet been evaluated in vivo. Using a systems biology-based approach, we identified the complement pathway as a high-priority target for control of SARS-CoV MA15 (the mouse-adapted SARS-CoV) pathogenesis based on weighted gene correlation network analysis (WGCNA) of RNA transcripts in the lungs of mice infected with a lethal versus a sublethal dose of SARS-CoV MA15 (23). Because the complement signaling cascade is activated through proteolytic cleavage events, we also assessed lung proteomics samples for complement protein abundance. C4b, CfB, and C3 all had significantly higher abundances in the lungs of mice infected with a lethal dose of SARS-CoV MA15 than in those of mice infected with a sublethal dose (Fig. 1A). Complement activation is measured by detection of pathway component cleavage products. C3, the master regulator of complement signaling, is cleaved into C3a and C3b following creation of C3 convertase. C3 activation products (C3 fragments C3a, C3b, iC3b, C3dg, and C3c) were detected by Western blotting in lung tissue of SARS-CoV MA15-infected mice, but not in control mice, as early as 1 day postinfection (dpi) (Fig. 1B), confirming that SARS-CoV MA15 infection activates the complement pathway.

Multiple complement pathways contribute to SARS-CoV MA15-induced pathogenesis.
To test the importance of the complement signaling pathway in SARS-CoV pathogenesis, we infected C3 -/mice and C57BL/6J controls with SARS-CoV MA15. Control mice exhibited approximately 15% transient weight loss, with peak weight loss at day 3 postinfection (Fig. 2A). In contrast, the C3 -/mice were significantly protected from infection, with no significant weight loss evident at any time point. Surprisingly, viral titers in the lung were similar in C3 -/and C57BL/6J controls (Fig. 2B), indicating that the lack of disease in C3 -/mice is uncoupled from viral replication efficiency and that complement signaling is not necessary for SARS-CoV MA15 clearance from the lung. We further measured SARS-CoV MA15-induced disease by assessing respiratory function using whole-body plethysmography following infection of C57BL/6J and C3 -/mice. Enhanced pause (Penh) is a calculated measure of airway resistance that we have associated with airway debris following SARS-CoV MA15 infection (24). The 50% exhalation force (EF50) measures the exhalation force midbreath, which increases as breathing becomes more difficult. Finally, the ratio of peak expiratory flow (RPEF) is the time to peak expiratory flow and has been associated with wheezing following infection. All three metrics have been shown to change significantly following SARS-CoV MA15 infection, with Penh and EF50 increasing following infection and RPEF decreasing. Combined, these measurements show that SARS-CoV MA15-infected animals have altered exhalation patterns in their breathing. C3 -/mice exhibit a decreased change in Penh and EF50 levels following SARS-CoV MA15 infection relative to those of infected C57BL/6J control mice; however, RPEF values were similar between infection conditions ( Fig. 2C to E). Together, these data indicate that despite the lack of weight loss in C3 -/mice, the absence of the complement pathway did not alter host control of viral The respiratory function of SARS-CoV MA15-infected C57BL/6J and C3 -/mice and mock-infected mice was measured using a Buxco whole-body plethysmography system for Penh, a measure of calculated airway resistance (C), EF50, midbreath expiratory flow (D), and RPEF, the rate of peak expiratory flow (E). *, P Ͻ 0.05 between mock-infected mice and a given condition. (C to E) Three mice were used for each infection group, and two mock-infected mice were used.
replication or completely abolish respiratory disease following SARS-CoV MA15 infection.
In order to determine which arm of the complement pathway contributes to SARS-CoV MA15 pathogenesis, we infected knockout mice lacking components upstream of C3. C4-deficient mice lack the ability to signal through both the classical and lectin pathways, while fB-deficient mice lack the ability to signal through the alternative pathway. Both mouse strains showed reduced weight loss relative to that of infected control mice (see Fig. S1 in the supplemental material) at 3 days postinfection; however, neither C4 -/nor fB -/mice reproduced complete protection from weight loss observed in C3 -/controls. Together, the results suggest that multiple arms of the complement pathway may be activated and contribute to SARS-CoV-mediated disease through C3 activation.
Reduced lung pathology in C3 -/mice. Analysis of SARS-CoV MA15-infected lung sections showed that the absence of C3 resulted in reduced, but still significant, lung pathology. At day 2 postinfection, only minor effects on lung disease were observed with airway denudation and debris, the main histopathological phenotypes at this early time point; levels of pathology were similar between wild-type and knockout mice ( Fig. 3; Table S1). C3 -/mice exhibited more airspace inflammation, including eosino- phils, at 2 dpi than their wild-type controls, although this relationship was reversed later in infection. At 4 dpi, C57BL/6J mice displayed pronounced lung pathology, including inflammatory cells in the large airway and parenchyma, perivascular cuffing, thickening of the interstitial membrane, and low levels of intra-alveolar edema. In contrast, C3 -/mice showed reduced scores in these areas, consistent with the improved respiratory function observed in Fig. 2. Notably, the lung pathology results were not as pronounced as the complete absence of weight loss, suggesting a possible distinction between lung disease and overall pathogenesis. We also considered whether the decrease in SARS-CoV MA15 pathogenesis in C3 -/mice was due to reduced lung damage in the absence of complement pathway signaling. To investigate this possibility, we looked for signs of complement deposition on SARS-CoV MA15 lung tissue. At both 2 and 4 dpi, we observed scattered positive staining for complement in the lungs of SARS-CoV MA15-infected mice, suggesting that local tissue damage might contribute to SARS-CoV pathogenesis (Fig. 4). Interestingly, staining was consistently found in the parenchyma of the lung and not in the large airways, which are the other main site of SARS-CoV MA15 replication. No positive staining was observed in the lungs of C3 -/mice.
Diminished infiltration of the lungs of a select immune population of infected C3 -/mice. In order to identify and quantitate inflammatory cells in the lung, we performed flow cytometry at 4 dpi. In parallel with humans exhibiting lung pathology, C3 -/mice exhibited significant pulmonary infiltration following SARS-CoV MA15 infection, but this inflammation was reduced relative to that observed in wild-type mice. SARS-CoV MA15-infected C57BL/6J and C3 -/mice had similar total cell counts as well as similar percentages of CD45-positive cells in their lungs (data not shown). Consistently with what was observed in human SARS-CoV patients (25), lymphopenia was observed in the lungs of both SARS-CoV MA15-infected C57BL/6J and C3 -/mice with reduced percentages of B cells (Fig. 5A) and CD4 T cells relative to those in mockinfected mice following infection. Despite similar overall lymphocyte levels, small but significant differences were observed in levels of T cell activation between infected C57BL/6J and C3 -/mice; both CD4 and CD8 T cells in C3 -/mice expressed more Ki-67 ( Fig. 5C and D), an intracellular marker of proliferation, than those in C57BL/6J controls. Analysis of myeloid cells in the lung showed that infected C56BL/6J mice had significantly higher levels of neutrophils, particularly nonactivated neutrophils, in the lung than infected C3 -/mice ( Fig. 5B and E). Furthermore, inflammatory monocytes, which have previously been associated with increased SARS-CoV MA15 pathogenesis (26), were significantly increased in the lungs of wild-type mice but not C3 -/mice (Fig. 5B). Finally, we observed significantly more dendritic cells and alveolar macrophages in the lungs of SARS-CoV MA15-infected C3 -/mice than in the lungs of infected C57BL/6J mice. Together, although C3 -/mice produced a robust immune cell infiltration following SARS-CoV infection, they had significant reductions in both inflammatory monocytes and neutrophils relative to controls; both cell types that are associated with SARS-CoV pathogenesis (27). Conversely, the presence of activated T cells is associated with recovery following infection (28).
In addition to examining inflammatory cells, we evaluated the vascular integrity of the lung following SARS-CoV MA15 infection in the presence and absence of C3. We observed no differences in the numbers of platelets present in the bronchoalveolar lavage (BAL) fluid between C57BL/6J and C3 -/mice at either 2 or 4 days postinfection, indicating that the absence of C3 does not appear to significantly alter vascular permeability following infection with SARS-CoV (Fig. S2B).
Cytokine and chemokine levels are significantly decreased in the lungs of C3 -/mice. To further investigate the inflammatory response to SARS-CoV MA15 infection, we measured cytokine and chemokine protein levels in the lung in the presence and absence of complement signaling. Multiple protein expression patterns were observed in response to infection. MIP1a, MIP1b, and MCP1 are all highly expressed in the lung following SARS-CoV MA15 infection of both C57BL/6J and C3 -/mice (Fig. 6A), indicating that some inflammatory signaling remains intact in the absence of C3. Granulocyte colony-stimulating factor (G-CSF), interleukin 6 (IL-6), tumor necrosis factor alpha (TNF-␣), and IL-1a comprised a group of cytokines and chemokines that were more highly produced in the lungs of C57BL/6J mice than in C3 -/mice (Fig. 6B), all peaking at 2 days postinfection. Notably, these cytokines all have a role in the production, recruitment, or differentiation of neutrophils, consistent with the flow cytometry results in Fig. 5B. With the exception of RANTES, all inflammatory cytokines and chemokines were measured at the highest levels at 2 dpi, indicating that the host immune response is triggered quickly following infection with SARS-CoV MA15. Together, these results indicate that the absence of complement has an impact on the magnitude of some cytokines and chemokines in the lung; however, robust production can occur in either the presence or the absence of C3.
SARS-CoV MA15 induces systemic complement activation. The absence of complement signaling resulted in reduced SARS-CoV MA15 pathogenesis, as measured by weight loss and a partial reduction of respiratory dysfunction, pathology, immune infiltration, and cytokine responses in the lung. We hypothesized that systemic disease coupled with no change in viral titer might also drive important elements of complement-mediated disease. Therefore, we examined sera from wild-type and C3 -/mice for signs of systemic disease following infection. Western blot analysis showed increased levels of C3a-derived fragments in the serum, indicating systemic complement activation in SARS-CoV MA15-infected mice at 2 dpi (Fig. 7A). Given this result, we next examined cytokine and chemokine protein levels for markers of inflammation in the sera of SARS-CoV MA15-infected mice. Both MCP-1 and RANTES levels were elevated in the serum following infection, regardless of mouse genetic background (Fig. 7B). However, numerous cytokines and chemokines, such as IL-5, G-CSF, and KC (keratinocyte chemoattractant or CXCL1) were present in significantly higher abundance in the lungs of C57BL/6J mice than in those of C3 knockout mice (Fig. 7C). We further examined the possibility that SARS-CoV MA15 infection leads to complement deposition outside the lung and found no signs of increased complement staining in the kidney (Fig. 7D). Although no complement deposition was seen, the presence of both activated complement and inflammatory cytokines in the sera likely contributes to a systemic inflammatory response that drives SARS-CoV MA15-mediated weight loss following infection.

DISCUSSION
The complement system is a critical part of the host immune response to bacterial and viral infection. Originally identified in the 1900s as a heat-sensitive, nonspecific complement to the more specific adaptive immune pathways (29), the complement system is one way that the innate immune system detects and responds to foreign antigens. Because of its potential to damage host tissues, the complement system is also tightly regulated through a number of inhibiting proteins that are constitutively present in the serum (30). It has previously been shown that complement pathway signaling is critical for the protective host immune response to various bacterial infections (31) as well as some influenza virus and flavivirus infections (22,32,33). Furthermore, viruses, including herpesviruses, poxviruses, astroviruses, flaviviruses, and retroviruses, encode genes to help them evade detection by the complement system (17), strong evidence that complement is important in the host antiviral response. The host factors that drive protective (22) or pathogenic (34) complement-associated responses in viral infection are not well understood. Of particular concern, the anaphylatoxins C3a, C4a, and C5a are produced during activation of the complement signaling cascade; they have potent proinflammatory properties and can trigger inflammatory cell recruitment and neutrophil activation (35). C3a and C5a blockade has been proposed as a treatment for acute lung injury (36), and anti-C5a antibody has been shown to protect mice from infection with influenza virus (34) and, more recently, MERS-CoV (16). Complement recognition is important for the control of paramyxoviruses (37), dengue virus (38), and human T lymphotrophic virus type 1 (HTLV-1) (39), and many more viruses have developed means of evading detection by the complement system (17). In contrast, the data presented here, in conjunction with recent findings for Ross River virus (40,41), influenza virus (34), and well-established autoimmune disease (42), demonstrate that complement system activation can also lead to exacerbated disease.
Previous reports clearly established the ability of mannose-binding lectin (MBL) to bind to the SARS-CoV spike protein (11), dependent on an N-linked glycosylation site; however, the role of complement signaling in SARS-CoV pathogenesis was unclear (9,10,13,43). In this study, we demonstrate that the complement system is activated following SARS-CoV MA15 infection (Fig. 1B). However, we did not observe any change in viral titer in C3 -/mice (Fig. 2B), indicating an important difference between in vivo and in vitro studies and the use of viral pseudoparticles. The absence of complement signaling resulted in protection from SARS-CoV MA15-induced weight loss, as shown through the use of both C3-deficient mice ( Fig. 2A) and activation pathway-specific knockout mice (Fig. S1). Respiratory function in C3 knockout mice was improved relative to that of control mice, although significant changes in Penh and RPEF were still observed, indicating that the elimination of complement signaling did not completely remove the effects of SARS-CoV MA15 infection. While analysis of the cellular inflammatory response to SARS-CoV MA15 infection revealed modest changes in histopathology and overall inflammatory cell recruitment to the lungs, significant differences were observed in pathogenic inflammatory monocyte and neutrophil populations, indicating that complement signaling contributes to the broader immune response to infection. Immunohistochemical staining revealed that SARS-CoV MA15 infection induced complement deposition in the lung (Fig. 4), similar to that associated with pathogenesis in Ross River virus-infected mice (41) and some influenza virus infections (34), and it is likely that complement deposition contributes to pulmonary disease and inflammatory cell recruitment.
The cytokines and chemokines IL-5, IL-6, KC (CXCL1), and G-CSF have higher abundances in the lungs of SARS-CoV MA15-infected wild-type mice than in C3 -/mice and are all known to promote neutrophil recruitment. Indeed, significantly more neutrophils were observed in the lungs of SARS-CoV MA15-infected C56BL/6J mice than in the C3 -/mice (Fig. 5B). Interestingly, while there were fewer neutrophils present, the neutrophils found in the lungs of C3 -/mice infected with SARS-CoV MA15 had significantly more staining of major histocompatibility complex class II (MHC II) and the costimulatory molecules CD80 and CD86 (Fig. 5E), indicating a state of activation (44). Unpublished data from our laboratory have consistently demonstrated higher neutrophil counts in the lungs of mice with severe disease than in those of mice with only mild pathogenesis. Additionally, neutrophilia in human SARS-CoV patients was associated with a poor outcome of infection (45), and studies of native rat coronavirus (46) found both a protective and a pathogenic role for neutrophils following infection. Combined, these data demonstrate that the absence of complement provides significant improvements in pulmonary disease following SARS-CoV MA15 infection and suggest that a nonpulmonary cause might also contribute to the lack of weight loss in C3 -/mice.
Importantly, our data demonstrate that SARS-CoV MA15 infection activates the complement system systematically as well as in the lung (Fig. 7A and 1B). Wild-type C57BL/6J mice exhibited an increased abundance of serum cytokines and chemokines in response to SARS-CoV MA15 infection (Fig. 7C) in comparison to C3 -/mice. In particular, the pyrogenic cytokine IL-6 (47) is present at higher abundance in the lungs and sera of C57BL/6J mice than in those of C3 -/mice. IL-1␣ and TNF-␣ are also more abundant in the lungs of C57BL/6J mice, suggesting that wild-type but not C3 -/mice develop a fever response to infection that contributes to weight loss and respiratory dysfunction phenotypes. While the precise mechanism of complement activation following SARS-CoV MA15 infection is still unclear, it is likely through recognition of the viral spike glycoprotein and partially mediated by MBL (see Fig. S1 in the supplemental material).
The anaphylatoxins produced by the activated complement pathway, C4a, C3a, and C5a, have important immunostimulatory roles in vascular permeability and inflammatory cell recruitment (35,48). C3a and C5a in particular are noted for their roles in causing mast cell degranulation, initiating a cytokine storm, promoting vascular permeability, and contributing to acute lung injury (36,49,50). Furthermore, a C5a antibody blockade was recently shown to protect in a model of highly susceptible MERS-CoV mice (16). Although there are no reports of mast cell activation following SARS-CoV or MERS-CoV infection, activation has been observed both in vitro and in vivo following infection with influenza virus (51) and may occur following other severe respiratory infections, including coronaviruses. Mast cells release cytokines, including IL-6, IL-1, and TNF-␣, consistent with the inflammatory profile observed following SARS-CoV MA15 infection. While this work cannot definitively conclude that the complement anaphylatoxins and mast cell activation contribute to SARS-CoV MA15 pathogenesis, the data are consistent with this possibility, and the concept warrants further investigation. Complement pathway activation is a hallmark of bacterial infection, and genetic deficiencies in the complement pathway result in enhanced susceptibility to Streptococcus pneumoniae, Neisseria meningitidis, and Haemophilus influenzae infections (52), as well as to sepsis. Interestingly, it has also been shown that SARS-CoV MA15 infection stimulates TLR4 (53,54), which is classically known as the lipopolysaccharide (LPS) receptor (55,56) and important for recognition of many bacterial infections. Combined, these data suggest that host recognition of SARS-CoV MA15 infection may activate similar pathways recognizing a bacterial infection, leading to immune signaling cascades that cause systemic disease and enhance viral pathogenesis. While systemic activation of the complement pathway may be useful during a bacterial infection, it is less so during a localized acute viral infection, such as SARS-CoV. Furthermore, complement activation, in conjunction with the presence of neutrophils, is known to cause increased vascular permeability, a condition that is also observed following SARS-CoV infection (57)(58)(59) and was associated with poor outcome. Baseline complement activation also increases with age (60, 61), consistent with increased SARS-CoV morbidity and mortality in aged populations. Finally, it has previously been reported that serum C5a levels are predictive of ARDS development (62) and that, in the absence of complement, animals are protected from bacterially induced "shock lung" (63), data consistent with the pathogenic role that we have found for complement following SARS-CoV MA15 infection.
In this work, we demonstrate that SARS-CoV MA15 infection activates the complement pathway and that complement signaling contributes to disease following infection. This disease is likely mediated by complement protein deposition in the lung as well as systemic complement activation and inflammation. Notably, the absence of C3 has no impact on viral titer, unlike what has been observed following influenza virus infection. Despite these differences, it is notable that MERS-CoV and H5N1 influenza virus-induced acute lung injury and pulmonary inflammation are reduced in mice that are treated with either a C3a receptor (C3aR) antagonist or antibodies to C5a (16,34). A similar treatment might be effective in mitigating SARS-CoV MA15-induced disease, as SARS, MERS, and influenza have common disease manifestations, including development of acute lung injury. Given the large array of zoonotic strains poised for cross-species transmission, broad-based inhibitors of emerging coronavirus infections are a high priority (64). Pinpointing the precise arms of the complement pathway that contribute to SARS-CoV will help further identify therapeutic targets while minimizing unnecessary inhibition of the immune response. This work suggests that investigation of anticomplement drugs for treatment of coronavirus infections is warranted and would pair well with direct antiviral therapeutics.

Viruses and cells.
Stocks of recombinant mouse-adapted SARS-CoV (MA15) (65) were propagated and their titers were determined in Vero E6 cells and stored as single-use aliquots at Ϫ80°C as previously described (66). Tissue titers of MA15 were determined by plaque assay on Vero E6 cells as previously described (66,67), with a limit of detection of 100 PFU. All experiments using live virus were performed in a class II biological safety cabinet in a certified biosafety level 3 laboratory with negative air pressure and redundant exhaust fans; personnel wore personal protective equipment, including Tyvek suits, hoods, and powered air-purifying respirators.
Mouse experiments. C57BL/6J (stock number 000664) and C3 -/-(stock number 003641) mice were purchased from Jackson Laboratories. fB Ϫ/Ϫ mice were generously provided by Charles Jennette (UNC), and C4 -/mice were provided by Mark Heise (UNC). All animal husbandry and experiments were performed in accordance with all University of North Carolina at Chapel Hill Institutional Animal Care and Use Committee guidelines. Age-matched (10-to 11-week-old) female mice were anesthetized with a mixture of ketamine-xylazine and intranasally inoculated with 50 l of phosphate-buffered saline (PBS) or 10 5 PFU of SARS-CoV MA15 diluted in PBS. Mice were monitored for disease signs and weighed at 24-h intervals.
Microarray and proteomics analysis. Microarray and proteomics analyses were performed on a time course and according to a dose-response study published by Gralinski et al. by following the same methods (23). The proteomics data (experiment SM001) are publically available through the PNNL (http://omics.pnl.gov) Web portal. Briefly the mean intensity (abundance) for each protein was then graphed as an average (5 mice for each infection, 3 mice for mock infection) for each group at each time point. Missing or absent values were not scored; however, if no value was observed in any of the samples at a time point, the sample was registered with a single 0, representing "not detected." Histological analysis. At the times indicated in the figures, mice were euthanized using an overdose of isoflurane, and lung tissue was fixed in 10% formalin. Tissues were embedded in paraffin, and 5-m sections were prepared. To determine the extent of inflammation and tissue pathology, tissues were stained with hematoxylin and eosin and scored in a blind manner from 0 (no sign of phenotype) to 3 (widespread and severe phenotype).
Platelet counts. For bronchoalveolar lavage (BAL), immediately following euthanasia, 1 ml of PBS was injected into the lung through the trachea by using a 22-gauge Exel Safelet catheter tip (Fisher). This fluid was then drawn back out and used for subsequent analysis. Two hundred fifty microliters of BAL fluid was used for absolute counting of gross cell types using an Abaxis VetScan HM5 analyzer.
Complement deposition staining. Lung sections from SARS-CoV MA15-, Ross River virus-, or mock-infected mice were stained for the presence of C3 by the Animal Histopathology and Laboratory Medicine Core at the University of North Carolina. SARS-CoV MA15 lung samples were tested from 20-week-old mice at 1, 2, 4, and 7 days after infection with 10 5 PFU of virus. Staining was performed using a goat anti-mouse C3 primary antibody (MP Biomedicals).
Immunoblot analysis. Mice were perfused with PBS, and then lung tissue was dissected and homogenized in lysis buffer (50 mM Tris [pH 8.0], 150 mM NaCl, 1% Nonidet P-40, 0.5% deoxycholate, and 0.1% sodium dodecyl sulfate [SDS] supplemented with Complete protease inhibitor cocktail [Roche]). Total protein concentrations were determined by using the Coomassie Plus assay kit (Pierce). Dilutions of serum or 20-to 30-g aliquots of protein were diluted in an equal volume of 2ϫ SDS sample buffer, and SDS-polyacrylamide gel electrophoresis was performed. Proteins were transferred onto polyvinylidene fluoride membranes (Bio-Rad). Membranes were blocked in 1ϫ PBS-5% milk-0.1% Tween 20 and incubated with goat anti-mouse C3 antibody (1:1,000; Cappel) overnight at 4°C. Membranes were washed in PBS-0.1% Tween 20 and incubated with rabbit anti-goat-horseradish peroxidase (1:10,000; Sigma) for 1 h at room temperature. After a washing step, proteins were visualized by enhanced chemiluminescence (Amersham) according to the manufacturer's instructions.
Flow cytometry. Following euthanasia at 4 days postinfection, mice were perfused with 10 ml of PBS via cardiac puncture. Lungs were dissected, minced, and incubated for 90 min with vigorous shaking at , and CD103-PerCP-eFluor710 (clone 2E7; eBioscience). After being stained, cells were fixed in 2% paraformaldehyde overnight and then stored in PBS until acquisition within 24 h. A minimum of 100,000 events were collected using an LSRII cytometer (Becton, Dickinson), and analysis was completed using FlowJo software version 10 (TreeStar). All samples were first evaluated through subsequent gates for (i) mononuclear cells, (ii) doublet exclusion, (iii) dead-cell exclusion based on uptake of a fixable live/dead cell discriminator (Invitrogen), and CD45-LCA expression before downstream analyses. Reported cell frequencies were normalized to the percentage of total CD45-LCA ϩ events, where appropriate.
Whole-body plethysmography. Respiratory function was measured using whole-body plethysmography as described by Menachery et al. (24). Briefly, mice were loaded into individual chambers and allowed to acclimate for 30 min before a 5-min measurement window. Measurements were recorded every 2 s for a total of 150 measurements per time point per mouse.
Cytokine and chemokine protein analysis. The small center lung lobe of each mouse was homogenized in 1 ml of PBS and briefly centrifuged to remove debris. Fifty microliters of homogenate was used to measure cytokine and chemokine protein abundance using a Bio-Plex Pro mouse cytokine 23-plex assay (Bio-Rad) according to the manufacturer's instructions.
Statistical analyses. Percent starting body weights, viral titers, and inflammatory cell numbers were evaluated for statistically significant differences by the Mann-Whitney test or Student's t test using GraphPad Prism software.
Accession number(s). The microarray data were previously deposited in the GEO database under accession number GSE33266 (23).

ACKNOWLEDGMENTS
Research was supported by grants from the NIAID of the NIH (AI100625 to R.S.B. and M.T.H., AI106772 to R.S.B., AI109761 to R.S.B., and K99AG049092 to V.D.M.). Animal histopathology was performed by the Animal Histopathology and Laboratory Medicine Core at the University of North Carolina, which is supported in part by an NCI Center Core Support Grant (5P30CA016086-41) to the UNC Lineberger Comprehensive Cancer Center. The UNC Flow Cytometry Core Facility is supported in part by Cancer Center Core Support Grant P30 CA016086 to the UNC Lineberger Comprehensive Cancer Center. The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.
The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.