A Mouse-Adapted SARS-CoV-2 Induces Acute Lung Injury and Mortality in Standard Laboratory Mice

Adaptation of SARS-CoV-2 MA via Serial Passaging In Vivo

We previously developed a recombinant mouse adapted strain of SARS-CoV-2 (SARS-CoV-2 MA) capable of utilizing mACE2 for viral entry by remodeling the spike and receptor binding interface via reverse genetics (Dinnon et al., 2020). Although SARS-CoV-2 MA mediated infection of wild-type mice, young adult mice did not display the major clinical manifestations or hallmarks of ALI (Dinnon et al., 2020). To improve the model, we used experimental evolution in vivo via serial passage of SARS-CoV-2 MA in the lungs of young adult BALB/cAnNHsd mice (herein referred to as “BALB/c” mice) every 2 days to select for more virulent strains (Roberts et al., 2007). With passage, we observed a linear decrease in body weight over time achieving greater than 10% body weight loss on 2 days post infection (dpi) by passage ten (P10) (Figure 1 A). We confirmed the virulence of the virus population generated at P10 using a plaque purified clonal isolate (SARS-CoV-2 MA10) from this passage in young adult BALB/c mice (Figure 1B). Deep sequencing of mouse lung’s total RNA from the 10 passages, plaque purified SARS-CoV-2 MA10, and four additional plaque purified passage 10 viruses was performed to identify the changes responsible for the increased pathogenicity and rare variants. In addition to the spike Q498Y/P499T substitutions engineered into the parental SARS-CoV-2 MA, SARS-CoV-2 MA10 included 5 additional nucleotide changes, all resulting in nonsynonymous coding changes (Figures 1C and 1D; Table S1). These mutations emerged in an ordered fashion and include changes in nonstructural protein 4 (nsp4) (C9438T), nsp7 (A11847G), nsp8 (A12159G), spike (S; C23039A), and open reading frame 6 (ORF6; T27221C). Some sequence heterogeneity was observed across the plaque purified viruses, although SARS-CoV-2 MA10 had the fewest mutations and most represented the viral population found at passage 10 (Table S1). The SARS CoV-2 MA10 maintained the ability to utilize non-human primate ACE2 and replicated and formed plaques in Vero E6 cells (Figure 1E), consistent with utility for viral propagation and titration. Importantly, SARS-CoV-2 MA10 was also attenuated compared to wild-type SARS-CoV-2 (SARS-CoV-2 WT) in primary human bronchiolar epithelial cells (HBEs) (Figure 1F), suggesting decreased fitness in human cells. Collectively, these observations led to the choice of SARS-CoV-2 MA10 for subsequent studies.

SARS-CoV-2 MA10 Causes Acute Lung Injury in Young BALB/c Mice

To gain insight into the dose-dependent pathogenic potential of SARS-CoV-2 MA10, we performed dose ranging studies in 10-week-old BALB/c mice infected with either PBS (mock), 10⁵ plaque-forming unit (PFU) of the parental SARS-CoV-2 MA, or 10², 10³, 10⁴, and 10⁵ PFU SARS-CoV-2 MA10. We observed a dose-dependent increase in morbidity and mortality over the course of 14 days with SARS-CoV-2 MA10 (Figure 2 A). Mortality rates of 20% and 60% were recorded for infection with 10⁴ and 10⁵ PFU, respectively. Notably, infection with 10² PFU of SARS-CoV-2 MA10 produced an increased weight loss as compared to 10⁵ PFU infections with the parental SARS-CoV-2 MA strain, highlighting the increased pathogenicity gained through passaging. To best capture severe disease phenotypes without excessive mortality, we proceeded with 10⁴ PFU of SARS-CoV-2 MA10 as the standard infection dose for young adult BALB/c mice.

To characterize the pathogenesis of SARS-CoV-2 MA10 in young BALB/c mice, we examined the kinetics of disease in mice through 7 days post infection using an intranasal inoculating dose of 10⁴ PFU. SARS-CoV-2 MA10-infected mice rapidly lost weight and reached maximum weight loss at day 4 (losing 16% of starting weight) (Figure 2B). At 5 dpi, the weight loss trajectories of infected mice diverged, with many mice recovering body weight juxtaposed to mice that continued to lose weight, collectively resulting in a ∼15% mortality rate (Figure 2C). At the time of necropsy, acute stage lung damage was noted grossly as firm, red, heavy lobes that were scored based on the extent of congestion-related discoloration (Sheahan et al., 2020a) (indicative of edema and diffuse alveolar damage) that peaked at 4 dpi and remained high through 7 dpi (Figure 2D). Virus replication in the lung peaked 1–2 dpi and was absent in most surviving mice by 7 dpi (Figure 2E). Viral replication in the upper respiratory tract (measured by viral titer in the nasal cavity) remained high on 1–3 dpi but was non-detectable in most mice by 5 dpi (Figure 2F).

To gain insight into the impact of infection on lung physiology, pulmonary function was measured over time via whole body plethysmography (WBP). As compared to control mice, infected mice exhibited a loss in pulmonary function as indicated by significant changes in PenH and Rpef, measures of airway obstruction, and EF50, a measurement of exhalation flow rate (Figures 2G–2I).

Histopathologic analyses at 2, 4, and 7 dpi revealed early multifocal damage to conducting airway epithelia (including bronchioles) that corresponded to viral antigen staining, which was intense on 2 dpi, waned by 4 dpi, and was absent by 7 dpi (Figure 2J). Often, bronchial damage progressed to segmental epithelial denudation with an accumulation of inflammatory cells, sloughed epithelial cells, cellular debris, fibrin deposition, and plasma proteins in the airway lumens. Later post-SARS-CoV-2 MA10 infection, airway epithelia became hyperplastic with regeneration. The distal alveolar ducts and sacs were markedly altered by infection, displaying hallmarks of diffuse alveolar damage (DAD) and multifocal positive labeling of pneumocytes for viral antigen at early time points after infection. Histologic changes included hypercellular thickening of the alveolar septae caused by infiltrating immune cells, pneumocyte degeneration and necrosis, congestion of small vessels and capillaries, endothelial activation, increased neutrophils with extravasation, exudation of proteinaceous fluid and fibrin with occasional organization into hyaline membranes, and increased numbers of alveolar macrophages. Although later time points featured increased numbers of lymphocytes organizing around bronchioles, lymphocytic cuffing was not a prominent pathologic feature in comparison to findings induced by other respiratory viral pathogens. Importantly, the most severe, lingering damage over the time course was in the alveolar region.

The pathology of SARS-CoV-2 MA10-infected lungs was blindly quantified utilizing two metrics of ALI (Schmidt et al., 2018; Matute-Bello et al., 2011). First, diffuse alveolar damage (DAD) was assessed based on the degree of cellular sloughing and necrosis. SARS-CoV-2 MA10 induced DAD as early as 2 dpi and was maintained through 7 dpi (Figure 2J). Second, the American Thoracic Society (ATS) has generated a small animal model ALI scoring scheme that assesses neutrophil presence in the interstitium and alveolar space, hyaline membrane formation, protein accumulation, and alveolar septal thickening. Consistent with DAD scores, ATS ALI scores were increased in SARS-CoV-2 MA10-infected mice at 2 dpi and increased through 7 dpi (Figure 2K). Immunohistochemistry (IHC) staining for viral nucleocapsid revealed intense staining at 2 dpi and lack of staining by 7 dpi (Figure 2L), consistent with the lung viral titer data (Figure 2D). At 2 dpi, viral antigen was detected in conducting airway epithelia and in the alveoli, consistent with alveolar type II pneumocyte distribution patterns.

Increased Morbidity and Mortality in Old Mice after SARS-CoV-2 MA10 Infection

Because SARS-CoV-2 and other emerging human coronaviruses exhibit an age-dependent increase in disease severity, we investigated whether SARS-CoV-2 MA10 infection of aged mice resulted in an increased disease severity. In comparison to young mice, 1-year-old mice were highly susceptible to SARS-CoV-2 MA10, with high morbidity and nearly 100% mortality when infected with 10⁴ and 10⁵ PFU (Figure 3 A). Although mice infected with 10³ PFU rapidly lost weight with very few animals surviving, those infected with 10² PFU did not exhibit disease signs and all survived, suggesting a threshold of virus >10² was necessary to cause significant disease in 1-year-old mice. Accordingly, we selected the lowest dose that caused severe disease (10³ PFU) as the standard infection dose for 1-year-old mice. With this dose, the kinetics of weight loss were similar to young BALB/c mice. However, unlike infected young adult mice, all aged mice continued to lose weight over time and ultimately lost 30% of their starting weight, succumbing to infection and/or reaching the criteria for humane euthanasia (Figure 3B). Overall, we observed increased mortality starting on day 4 after infection with only ∼15% survival by day 7 (Figure 3C). Thus, data presented at late time points such as 6 or 7 dpi are biased toward rare survivors.

Gross pathological evaluations at necropsy revealed macroscopically detectable discoloration of lung tissue that achieved maximal severity on days 4 and 5 after infection (Figure 3D). Virus replication in aged mice peaked 1 to 2 dpi (5.3 × 10⁶ PFU/tissue and 1.2 × 10⁷ PFU/tissue, respectively), values similar to young adult mice. In contrast with young adult mice, in which virus was cleared by 7 dpi, significant levels of infectious virus remained in the lungs of aged mice at later time points (Figure 3E). Low levels of infectious virus were present in the serum at 2 dpi (Figure S1 A). Minimal virus was found in the heart, which may reflect residual virus from the serum, and virus was not detected in brain at the time of peak lung titer (2 dpi) (Figure S1A). Viral protein was not detected in the heart, liver, small intestine, kidney, or spleen (Figures S1B–S1F). Old mice also exhibited viral titers in the nasal cavity over the first 3 days of infection (peak on 3 dpi at 2 × 10⁴ PFU/tissue), consistent with young adult mice (Figure 3F). Infection with SARS-CoV-2 MA10 also disturbed lung function in aged mice in a similar, but more prolonged, manner compared to young mice with significant changes in PenH, Rpef, and EF50 at 2–5 dpi (Figures 3G–3I).

Histological analyses revealed severe DAD and higher ATS ALI scores at later time points throughout the lung in 1-year-old mice (Figures 3J and 3K), consistent with the more pronounced interstitial congestion, epithelial damage, immune cell infiltration, and edema in the older animals (Figure 3L). Viral antigen was detected in small airways and alveolar regions at 2 and 4 dpi (Figure 3L). Viral RNA was also detected in the olfactory epithelium at 2 dpi (Figure 3M), concordant with nasal cavity viral titers (Figure 3F). At 4 dpi, the olfactory epithelium was severely damaged, likely contributing to reduced nasal cavity viral titers, and infection had spread to the Bowman’s gland in the submucosa (Figure 3F).

Many viral diseases are associated with a systemic cytokine storm. We analyzed the chemokine and cytokine responses in the serum and lungs of 1-year-old BALB/c mice at 2 and 4 dpi (Figures S2 A and S2B). At 2 dpi, several proinflammatory cytokines were elevated in the lungs of SARS-CoV-2-infected mice, whereas few were elevated systemically in the serum. For instance, IL-6, IL-1α, IL-1β, TNF-α, MCP-1, and IFN-γ were highly elevated in the serum and/or lungs of infected mice, similar to reports in humans (Sinha et al., 2020). It remains uncertain as to whether these elevated cytokines contribute to severe disease outcomes after infection.

Ameliorated Disease and No Mortality in C57BL/6J Mice after SARS-CoV-2 MA10 Infection

C57BL/6 is the most commonly used mouse strain and is the genetic background for the majority of genetically engineered mice (Bryant, 2011). Because host genetic background dependent differences in disease susceptibility have been described for many infectious diseases including SARS-CoV (Noll et al., 2020; Rasmussen et al., 2014; Gralinski et al., 2015; Graham et al., 2015; Manet et al., 2020), we evaluated SARS-CoV-2 MA10 infection in young adult C57BL/6J mice. In comparison to BALB/c mice, 10-week-old C57BL/6J mice exhibited less severe disease and only the two highest doses (10⁴ and 10⁵ PFU) were associated with significant weight loss, but no mortality after infection (Figure 4 A). Therefore, we performed a detailed analysis of the 10⁴ PFU infectious dose over 7 days for a direct comparison between young BALB/c and C57BL/6J mice. After infection with 10⁴ PFU, C57BL/6J mice exhibited a transient 10%–15% weight loss (peaked on 3 dpi and 4 dpi) (Figure 4B) without mortality (Figure 4C). Gross congestion scores in lungs at the time of harvest never rose above a score of 1 (roughly 25% lung involvement) and declined from 3 dpi until 7 dpi (Figure 4D). A clear peak in viral replication in the lungs was observed on 2 dpi (1.6 × 10⁶ PFU/tissue) that was ∼1 order of magnitude lower than peak titers observed in BALB/c mice. After 2 dpi, titers decreased steadily and were not detectable by 7 dpi (Figure 4E). In addition, viral loads in the nasal cavity were relatively low through 3 dpi (2 × 10³ to 1.2 × 10⁴ PFU/tissue) and were undetectable by 4 dpi (Figure 4F). Changes in lung function as measured by WBP were similar in pattern for the two mouse strains, but C57BL/6J mice exhibited restored lung function to near baseline levels by 5 dpi, whereas abnormalities in BALB/c mice persisted until 7 dpi (Figures 4G–4I). Similarly, the patterns of histologic changes were equivalent in C57BL/6J mice to young BALB/c mice (Figure 4L), but the magnitudes of the acute lung injury scores were dramatically attenuated in C57BL/6J mice (Figures 4J and 4K).

SARS-CoV-2 MA10 Cellular Tropism

To characterize the tissue and cellular tropism of SARS-CoV-2 MA10, BALB/c mouse nasal and lung tissues were probed for the SARS-CoV-2 viral nucleocapsid and tissue/cell type markers by RNA in situ hybridization (ISH) and immunohistochemistry. In the proximal conducting airways (e.g., trachea, bronchi), little, if any, SARS-CoV-2 MA10 infection was identified. However, robust viral infection was identified in terminal bronchioles that connect to the alveolar spaces (Figures 5A and 5B). In the SARS-CoV-2 MA10-infected mouse terminal bronchiolar epithelial region, expression of Scgb1a1, which is a secretory club cell marker, largely disappeared whereas Foxj1, a ciliated cell marker, persisted. IHC identified SARS-CoV-2 MA10 nucleocapsid expression in non-ciliated cells with occasional colocalization with CCSP (Figures 3L and 5C), suggesting that secretory club cells were infected by SARS-CoV-2 MA10 and subsequently lost Scgb1a1 expression. This cellular tropism of SARS-CoV-2 MA10 is different in human airways, perhaps reflecting different cell levels for ACE2 expression between human versus mice (i.e., ciliated versus secretory club), respectively (Zhang et al., 2020b).

In alveoli, ISH studies of mock-infected mice identified the two major epithelial cell types, i.e., AT1 (Ager expressing) and AT2 (Sftpc, Sftpb expressing) cells (Figures 5D and 5E). In SARS-CoV-2 MA10-infected mice, Sftpc and Sftpb expression characteristic of AT2 cells virtually disappeared, whereas Ager expression associated with AT1 cells persisted at 2 dpi. IHC identified occasional cells expressing a third AT2 cell marker (LAMP3) that also co-expressed the SARS-CoV-2 MA10 nucleocapsid (Figure 3F). Collectively, the loss of surfactant protein transcripts, but not AGER, staining and colocalization of a third AT2 marker (LAMP3) with virus, argues for selective infection by SARS-CoV-2 MA10 of AT2 in the alveolus. The finding that SARS-CoV-2 MA10-infected AT2 cells suppressed expression of selective cell-type-specific genes (e.g., Sftpc and Sftpb) is consistent with findings in infected human AT2 cells in vitro.

The nasal cavity of the mouse is comprised of ∼50% respiratory epithelium and 50% olfactory epithelium (Chamanza and Wright, 2015). As noted above (Figure 3M), SARS-CoV-2 MA10 RNA was detected in olfactory epithelium, as defined anatomically and by the olfactory sensory neuron marker (OSN) marker, Uchl1 (Figure S3 ). Notably, viral RNA was not detected in cells expressing Uchl1, indicating that SARS-CoV-2 MA10 likely infected sustentacular cells rather than OSNs. The selective olfactory infection by SARS-CoV-2 MA10 is likely associated with altered olfactory function commonly observed in subjects with COVID-19 (Wölfel et al., 2020; Lechien et al., 2020; Spinato et al., 2020).

Interferon Signaling Is Protective in SARS-CoV-2 MA10 Infection

We next tested whether the pathogenic SARS-CoV-2 MA10 virus could be used with genetically deficient mice to elucidate aspects of underlying molecular pathways and networks that regulate SARS-CoV-2 disease. Interferon signaling plays an important role in controlling and regulating disease severity after infection with many viruses, including coronaviruses (Mesev et al., 2019; Israelow et al., 2020). SARS-CoV-2 has also been reported to be sensitive to type I and III interferons in human cells in vitro (Felgenhauer et al., 2020; Vanderheiden et al., 2020) and mice in vivo (Israelow et al., 2020; Dinnon et al., 2020). Consequently, we infected C57BL/6J mice lacking the type I and II interferon receptors (IFNR DKO) and wild-type controls with 10⁴ PFU of SARS-CoV-2 MA10. IFNR DKO mice were more susceptible to SARS-CoV-2 MA10 as indicated by the prolonged weight loss compared to wild-type mice (Figure 6 A). Animals were harvested on planned harvest days. At 4 dpi, SARS-CoV-2 MA10 IFNR DKO mice displayed much higher congestion scores (Figure 6B), which were associated with higher viral titers on 2 and 4 dpi in IFNR DKO mice (Figure 6C). These data suggest that IFNs are important in limiting viral replication and assisting in virus clearance in vivo. Consistent with these data, lung function abnormalities were more pronounced and prolonged in infected IFNR DKO mice (Figures 6C–6E).

SARS-CoV-2 MA10 Allows Rapid Evaluation of Medical Counter Measurements

As previously shown for SARS-CoV-2 MA, mouse adapted viral strains allow for rapid testing of prevention and intervention strategies (Dinnon et al., 2020). The SARS-CoV-2 MA10 murine model adds to measurements of viral load the ability to evaluate changes in clinical parameters (weight loss, lung function, and pathologic changes) and mortality (Dinnon et al., 2020). Utilizing our previously described non-select BSL2 Venezuelan equine encephalitis viral replicon particle (VRP) system (Agnihothram et al., 2018; Dinnon et al., 2020), 10-week-old young adult and 1-year-old (“aged”) BALB/c mice were immunized with 10³ VRPs expressing SARS-CoV-2 WT spike (S), nucleocapsid (N), and GFP control, followed by a boost at 3 weeks, and challenged with SARS-CoV-2 MA10 4 weeks post boost (7 weeks post prime). Neutralization assays using nLuc expressing reporter virus revealed strongly neutralizing activity in the serum from mice at 3 weeks post boost from spike, but not serum from nucleocapsid or GFP vaccinated mice (Figures 7 A and S4 A). Of note, older (1-year-old) animals exhibited significantly reduced neutralization titers as compared to 10-week-old animals, capturing age related vaccine vulnerabilities often observed in human populations (Figure S4B). Notably, polyclonal sera had similar neutralization titers for both SARS-CoV-2 and SARS-CoV-2 MA containing two of three receptor binding domain changes present in MA10, suggesting that the SARS-CoV-2 MA10 model can accurately be used to test vaccine efficacy (Figure S4C). Only mice vaccinated with SARS-CoV-2 spike expressing VRPs exhibited disease protection as demonstrated by an absence of reductions in in body weight, protection from death (Figures 7B, 7C, S4D, and S4E), and total elimination of viral titers in the lower respiratory tract (lungs) (Figures 7D and S4F). Only one old mouse had detectable lung titer, corresponding to the mouse with the lowest serum neutralization titer. Interestingly, viral titers were still detectable on 2 and 4 dpi after infection in the nasal cavity of immunized young and old mice (Figures 7E and S4G), suggesting that mucosal immunity in the nasal cavity may be difficult to achieve by systemic immunization. Importantly, significant improvements in lung function were measured in both age groups in VRP-S vaccinated mice versus VRP-GFP or VRP-N controls (Figures 7F–7H and S4H–S4J).

Key Resources Table

Resource Availability

Experimental Models and Subject Details

Method Details

Quantification and Statistical Analysis

Data visualization and analyses were performed using build-in functions of GraphPad Prism. Specific statistical tests, numbers of animals, and definitions of center, dispersion and precision measures are mentioned in respective figure legends. For the characterization of SARS-CoV-2 MA10 in young and old BALB/c as well as young C57BL/6 mice the following statistical tests were used: mixed effect analysis followed by Sidak’s multiple comparisons was used to analyze weight loss and whole body plethysmography data; cell growth curves and cytokine / chemokine responses were analyzed by 2-factor ANOVA followed by Sidak’s multiple correction; gross lung congestions scores, lung and nasal titers, as well as DAD and ATS / ALI scores were analyzed by 2-factor ANOVA followed by Sidak’s multiple comparisons; survival rates were analyzed by log rank test. For the IFNR-DKO data the following statistical tests were used: weight loss data was analyzed using mixed effect analysis followed by Sidak’s multiple comparisons; 2-factor ANOVA followed by Tukey’s multiple comparisons was used for gross congestion scores and lung / nasal titer data; whole body plethysmography was analyzed via 2-factor ANOVA followed by Sidak’s multiple comparisons. VRP mouse data was analyzed as followed: weight loss and whole body plethysmography data was analyzed via mixed effect analysis followed by Sidak’s multiple comparisons; neutralization data was log transformed and analyzed via 1-factor ANOVA followed by Holm-Sidak’s multiple comparison; lung / nasal titer data as well as whole body plethysmography was analyzed by 2-factor ANOVA followed by Dunnetts multiple comparisons; unpaired, two-tailed Student’s t test was used for comparisons of serum IC₅₀ values from 10-week and 1-year-old vaccinated mice; Comparison of serum IC₅₀ values from 10-week-old spike vaccinated mice to neutralize SARS-CoV-2 WT versus SARS-CoV-2 MA was analyzed via Wilcoxon matched-pairs signed rank test.