Early circulating strain of SARS-CoV-2 causes severe pneumonia distinct from that caused by variants of concern

To analyze the molecular pathogenesis of SARS-CoV-2, a small animal model such as mice is needed: human ACE2, the receptor of SARS-CoV-2, needs to be expressed in the respiratory tract of mice. We conferred SARS-CoV-2 susceptibility in mice by using an adenoviral vector expressing hACE2 driven by an EF1α promoter with a leftward orientation. In this model, severe pneumonia like human COVID-19 was observed in SARS-CoV-2-infected mice, which was conrmed by dramatic inltration of inammatory cells in the lung with ecient viral replication. An early circulating strain of SARS-CoV-2 caused the most severe weight loss when compared to SARS-CoV-2 variants of concern, although histopathological ndings, viral replication, and cytokine expression characteristics were comparable. We found that a distinct proteome of an early circulating strain infected lung characterized by elevated complement activation and blood coagulation, which were mild in other variants, can contribute to disease severity. Unraveling the specicity of early circulating SARS-CoV-2 strains is important in elucidating the origin of the pandemic.

Early circulating strain of SARS-CoV-2 causes severe pneumonia distinct from that caused by variants of concern Yusuke Matsumoto Tokyo Metropolitan Institute of Medical Science Michinori Kohara (  kohara-mc@igakuken.or.jp ) Tokyo Metropolitan Institute of Medical Science inserted in the rightward orientation, a viral pIX gene located downstream of the inserted unit was coexpressed with the transgene, and a fusion protein consisting of the N-terminal part of transgene product was expressed. These pIX products may be one of the main causes of adenovirus-induced immune responses. Interestingly, the EF1α promoter did not activate the pIX promoter in this adenoviral vector 17 . The EF1α promoter with a leftward orientation resulted in a reduced antiviral response and maintained prolonged transgene expression 17 . Thus, we generated an adenoviral vector expressing hACE2 under the EF1α promoter with a leftward orientation (rAd5 pEF1α-hACE2-L) (Fig. 1a). We rst examined whether intranasal administration of rAd5 pEF1α-hACE2-L affects the body weight of BALB/c mice, and found that administration at 5×10 7 or 2.5×10 8 focus forming units (FFU) per mouse did not cause any decrease in body weight (Fig. 1b). Next, BALB/c mice were intranasally administered with rAd5 pEF1α-hACE2-L at 1⋅10 7 , 5⋅10 7 or 2.5⋅10 8 FFU/animal. Five days after administration, mice were further intranasally inoculated with an early circulating strain of SARS-CoV-2 isolated in Japan (originated from Wuhan; Wu-2020 strain) at 1⋅10 5 plaque forming units (PFU) per mouse (Fig. 1c). As a result, the 5⋅10 5 FFU rAd5/mouse group showed a marked reduction in body weight, which peaked at 5-6 days post-infection (dpi) with SARS-CoV-2 Wu-2020. Although the 2.5⋅10 8 FFU rAd5/mouse group showed weight loss until 4 dpi of SARS-CoV-2 infection, this group showed recovery of body weight after 5 dpi. The 1⋅10 7 FFU rAd5/mouse group showed a slight reduction of weight followed by rapid recovery. Therefore, we determined that 5⋅10 7 FFU rAd5/animal was the most suitable dose for assessing disease severity.
Variants of concern (VOCs) have emerged showing evidence of altered virus characteristics 18 . VOCs have been associated with increased transmissibility, evasion of immunity from infection and vaccination, and reduced susceptibility to antibody therapies [19][20][21]  at a multiplicity of infection (MOI) of 0.001, and found that the growth kinetics were almost comparable among strains (Fig. 2a). To analyze their replication in mouse lungs, we inoculated the rAd5 pEF1α-hACE2-L administered BALB/c mice with SARS-CoV-2 via the intranasal route. Viral replication in the lungs was examined by quantitative real-time RT-PCR (qRT-PCR) for the detection of SARS-CoV-2 genome, and by plaque assay using Vero E6/TMPRSS2 cells. A clear increase in viral replication was observed with a peak at 2 dpi in all strains ( Fig. 2b and 2c), followed by a gradual decrease towards 7 dpi. The B.1.351 strain showed reduced genome copy numbers (signi cantly at 7 dpi), but other strains showed comparable genome copy numbers throughout the course of infection. Macroscopically, in Wu-2020-and B.1.1.7 strain-infected lungs, multiple dark red and brown lesions appeared on the surfaces of all lung lobes from 4 to 7 dpi (Fig. 2d, arrows). In contrast, the discolored lesions were restricted to the upper left lobe in P.1-and B.1.1351 strain-infected lungs at 7 dpi. The loss of body weight in mice did not correspond to the extent of viral replication and lung lesions (Fig. 2e). The B.1.351 strain, an exception, showed less viral replication and lesions in the lungs, so the degree of weight loss was lower throughout the course of infection. Although viral replication was similar in the B.1.1.7 and P.1 strains, the appearance of lung lesions was more pronounced in B.1.1.7, however, weight loss was relatively more pronounced in P.1. Mice infected with all four strains continued to lose weight until 5 dpi, but mice infected with three VOCs began to recover thereafter. Though viral replication and the extent of lung lesions were comparable in B.1.1.7-and Wu-2020-infected mice, mice infected with Wu-2020 did not regain weight, resulting in signi cant weight loss at 7 dpi compared with mice infected with VOCs (Fig.  2e).
The histopathological analysis demonstrated that severe pneumonia with thickened alveolar walls, in ammatory cell in ltration, hemorrhaging, and thrombus formation was remarkable in the dark red and brown lesions of the lung (Fig. 3a, hematoxylin and eosin staining; HE). Even in areas where discoloration was not obvious macroscopically, thickened alveolar walls and mild in ltration of in ammatory cells were observed (Supplementary Fig. 1, HE). Immunohistochemistry using an antibody against SARS-CoV-2 nucleocapsid (N) protein showed that the viral antigen was present in the lung epithelial cells ( Fig. 3a and supplementary Fig. 1). The viral antigen was stained most prominently at 2 dpi. Few viral antigens were found in lesions with remarkable cellular in ltration.
Next, we examined cytokine expression in the lung using a multiplex bead array. In ammatory cytokine, such as IL-6, was signi cantly elevated in all strains compared to mock-infected animals (Fig. 3b). Furthermore, IL-1b, IFN-g, IL-12, MIP-1b, MIP-2, LIF, KC, IL-10, MCP-1, M-CSF, G-CSF, and GM-CSF were signi cantly elevated and VEGF was decreased, at least in a strain compared to mock-infected mice. In B.1.351-infected lungs, the expression of some cytokines, such as MIP2, LIF, and MCP-1, was relatively low compared to other strains. This is correlated with the manifestation of low viral replication and mild weight loss. For infection by other strains, there was no clear evidence of a relationship between cytokine levels and disease severity.
The clinical severity of COVID-19 is not always associated with increased levels of pro-in ammatory cytokines and other in ammation markers 25 . To survey the molecules associated with disease severity, tandem mass tag (TMT) peptide labeling, combined with mass spectrometry (MS) quantitative proteomics in mouse lung at 7 dpi with SARS-CoV-2 was performed. The TMT-based quantitative proteomic method was approved previously for comparison of protein levels across multiple organs in human COVID-19 autopsy cases 26 . TMTpro 12-plex MS revealed distinct lung proteomes associated with infection by SARS-CoV-2 strains ( Fig. 4a and Supplementary Fig. 2a). Gene ontology (GO) enrichment analysis of signi cantly (p<0.05) up-and down-regulated (2-fold) proteins showed that the proteome of Wu-2020-infected lung was distinct from those of other variants (Fig. 4b). Immune response-related factors, such as regulation of complement activation, immune effector process, as well as platelet degranulation and regulation of blood coagulation, were enriched in the proteins that changed signi cantly in the proteome of Wu-2020-infected lungs (Fig. 4b). In contrast, the proteomes of VOCinfected lungs were associated with structural organization, such as the development of extracellular structures and changes in matrix organization, as well as nuclear DNA replication. Up-regulation of proteins associated with complement activation, e.g., C3a anaphylatoxin chemotactic receptor, complement components and complement factors was prominent in Wu-2020-infected lungs (Fig. 4c).
The complement system has been shown to be involved in the severity of human COVID-19 27,28 . Upregulation of proteins involved in platelet degranulation and blood coagulation, e.g., kininogen, bronectin, plasminogen activator inhibitor-1, coagulation factor XII and plasma kallikrein, was also remarkable in Wu-2020-infected lung tissue ( Fig. 4d and f). These factors are considered to work in concert and contribute to COVID-19 pneumonia via dysregulation of thrombus formation. Up-regulation of minichromosome maintenance complex component (MCM)2, 3, 4, 5, 6, and 7, which are related to nuclear DNA replication, was observed in SARS-CoV-2 infection, regardless of strain (Fig. 4g). MCM2-7 act as replicative DNA helicases that unwind the DNA duplex template as a hetero-hexameric complex 29 . The involvement of the MCM family in immune responses against viral infection is still poorly characterized. However, MCM up-regulation is correlated with proliferation and maintenance of leukocytes 30,31 , suggesting that the MCM family is involved in the activation of in ltrating cells in the COVID-19 pneumonia observed in lungs infected with all of the virus strains. Structural organizationrelated proteins, such as collagens, were down-regulated in VOC-infected lungs (Fig. 4e). Collagen deposition is a hallmark of lung brosis 32 and has been con rmed in the lungs of COVID-19 patients 33 . It is considered that collagen deposition may be correlated with mild disease onset based on the recovery of body weight in VOC-infected mice. Pathway analysis based on Wikipathways (https://www.wikipathways.org/index.php/WikiPathways) supports the enrichment of complement and coagulation cascades, as well as the blood-clotting cascade, in Wu-2020-infected lungs ( Supplementary  Fig. 4). We established a system to recapitulate COVID-19-like pneumonia in mice infected with SARS-CoV-2 after inducing hACE2 with rAd5 pEF1α-hACE2-L. When rAd5 pEF1α-hACE2-L was used, there were few abnormalities in protein expression ( Supplementary Fig. 2a), suggesting that this adenoviral vector has low cytotoxicity. Mice infected with the Wu-2020 strain developed diffuse pneumonia.
Histopathologically, thickened alveolar walls, hemorrhaging, and in ltration of in ammatory cells were prominent. The SARS-CoV-2 N antigen was found in alveolar epithelial cells, not in the lesions in ltrated by in ammatory cells; rather, the antigen was concentrated in areas that retained relatively normal alveolar structure. These ndings were consistent with human COVID-19 autopsy cases in early 2020 34,35 .
SARS-CoV-2 Wu-2020, an early circulating strain, was shown to be highly pathogenic in mouse lung. The Wu-2020, B.1.1.7 and P.1 strains have comparable replication potentials in both Vero E6/TMPRSS2 cells and mouse lung (Fig. 2a-c). In addition, these strains induced a marked cytokine response (Fig. 3b), and infection led to similar histopathological ndings ( Fig. 3a and Supplementary Fig. 1). However, there was a clear difference in the lung proteome (Fig. 4b) between the Wu-2020 strain, which induced prolonged weight loss, and the other strains, which induced weight loss followed by recovery (Fig. 2e). The ndings showed that proteins involved in the complement system were elevated most markedly in cases of Wu-2020 infection (Fig. 4b and c). The release of proin ammatory complement peptides helps to recruit leukocytes to the lung and aids in the assembly of the terminal complex that damages vascular endothelium 27,28,36 . Increased levels of complement fragments is related to disease severity in COVID-19 patients 27,37 , which suggests that they are well suited for use as a marker for serious injury in Wu-2020infected lung. The altered blood coagulation system, which is manifested by the up-regulation of thrombosis-associated proteins, such as tissue factor (TF), coagulation factor XII and plasma kallikrein (Fig. 4f), can also be involved in disease severity in Wu-2020-infected lungs. TF initiates the extrinsic coagulation pathway to form thrombin in response to tissue injury and in ammation 38-41 . Coagulation factor XII is activated by polyphosphates released from platelets, and initiates an intrinsic coagulation cascade. [42][43][44] , which occurs with disease onset in acute respiratory distress syndrome 45 . Factor XII also activates plasma kallikrein, thereby increasing the formation of the proin ammatory peptide bradykinin 46 . Simultaneously, several inhibitory factors for complement activation, such as complement factor H and vitronectin (Fig. 4c), and for coagulation, such as plasminogen which activates brinolysis ( Fig. 4d), were up-regulated in Wu-2020-infected lungs; these factors may play a role for prevention of excessive tissue injury. These results indicate that molecular events in pneumonia lesions are altered in ways that cannot be observed by morphological observation. Even in these cases, the lung damage associated with elevated proteins that are related with complement activation, platelet degranulation and blood coagulation may result in the manifestation of severe symptoms, such as unrecovered weight loss after Wu-2020 infection.
In this study, levels of pro-in ammatory cytokines could not be used as markers of disease severity. Rather, the ndings showed that complement-related and blood coagulation factors may be key factors associated with COVID-19 severity. In addition, we observed 35-fold and 9-fold up-regulation of metallothionein-2 (Mt2) and Mt1, respectively, in Wu-2020-infected lung (Supplementary Fig. 3). Mt1/2, which are potently induced by heavy metals, other sources of oxidative stress and cytokines, facilitate metal binding and detoxi cation 47 . In response to GM-CSF, macrophages express Mts (Mt2 rather than Mt1), which are involved in antimicrobial responses and contribute to the production of reactive oxidative species 48 . We observed a correlation between disease severity and Mt1/2 amount, suggesting that Mt1/2 may act as a marker for COVID-19 severity. Additionally, we identi ed other potential biomarkers that may be correlated with disease severity, including tenascin, membrane-spanning 4-domains subfamily A member 6C and ste n-1/3 ( Supplementary Fig. 3), whose associations with COVID-19 have not been studied to date. Furthermore, abundance of the SARS-CoV-2 N protein in lungs infected with Wu-2020 was markedly higher than that in other strains (Supplementary Fig. 3). SARS-CoV-2 N protein has been shown to promote NLRP3 in ammasome activation 49 , and it is possible that SARS-CoV-2 N protein remaining in the lung may stimulate excessive in ammation. The amount of residual SARS-CoV-2 N protein in lesions may also be indicative of lung injury. It is possible that a comparison of SARS-CoV-2 strains that exhibit different pathogenicity may reveal the existence of novel biomarkers for disease severity.  55,56 . Our ndings revealed that there is a difference in the manifestation of symptoms associated with SARS-CoV-2 strains, and that an early isolated strain was highly pathogenic in the lung. The major difference between our ndings and previous studies is that the respiratory-speci c pathogenesis of SARS-CoV-2 was recapitulated in the mouse model using rAd5 pEF1α-hACE2-L in this study. SARS-CoV-2 infection involves extra-respiratory manifestations, including cardiac, gastrointestinal, hepatic, renal, and neurological symptoms. Disease severity in K-18 mice infected with B.1.1.7 and B.1.351 strains may be due to these extra-respiratory symptoms, as shown by the presence of neurological pathogenesis 57,58 . Comparative studies of human autopsy cases of each variant have not yet been performed. Some autopsy cases of patients infected with the B.1.1.7 and P.1 strains revealed no signi cant morphological or histopathological differences compared to early circulating strains 59,60 . However, our ndings showed that the elevation of several potent biomarkers may affect lung pathogenesis. Understanding the overall differences in an organ's proteome can help to unravel the pathology of emerging variants. With the accumulation of autopsy cases infected with VOCs, these changes in pathologies will be revealed.
In conclusion, we demonstrated that an early circulating SARS-CoV-2 speci cally induces the manifestation of severe symptoms and is associated with dramatically altered host responses. How pathogenicity was transformed from the initial strains to that observed in the VOCs needs to be elucidated. Detailed analyses of the pathogenicity of the early circulating strains will lead to a better understanding of the origin of the pandemic. Declarations Y.M., F.Y. and M.K. conceived, designed, coordinated and performed the study, contributed to data interpretation, data presentation and manuscript writing. T.S., T.M., K.Y. and N.Y. assisted with the animal experiments. N.Y. assisted with the qRT-PCR experiments for quantitation of the viral genome. A.T. assisted with the generation of the adenoviral vector. Y.M., A.E., K.Y. and Y.S. performed proteome analyses and contributed to data presentation.

Ethics statement
All experiments using mice were approved by the Tokyo Metropolitan Institute of Medical Science Animal Experiment Committee and were performed in accordance with the animal experimentation guidelines of the Tokyo Metropolitan Institute of Medical Science.

Cells and viruses
Vero E6/TMPRSS2 cells, which constitutively express human TMPRSS2 24 , and human embryonic kidney 293 (HEK293) cells were maintained in Dulbecco's modi ed Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS), penicillin and streptomycin, and G-418 (1 mg/mL, only in VeroE6/TMPRSS2 cells). All cells were cultured at 37°C in 5% CO 2 . Generation of rAd5 pEF1α-hACE2-L E1-and E3-deleted adenovirus derived from human adenovirus type 5 encoding expression units with a leftward orientation were used in this study, as described previously 17,61 . The hACE2 gene was cloned in the antisense orientation into pAxCAwtit2, the adenoviral cosmid vector that contains the left end of adenovirus type 5 with the E1 region substituted with an expression cassette containing the EF1α promoter and a multicloning site using an Adenovirus Dual Expression Kit (Takara Bio, Tokyo, Japan). The rAd5 pEF1α-hACE2-L was generated by transfecting pAxCAwtit2 encoding hACE2 into HEK293 cells by using CalPhos Mammalian Transfection Kit (Takara Bio). The rAd5 pEF1α-hACE2-L were puri ed using two rounds of cesium chloride gradient centrifugation, and the titers of the concentrated and puri ed virus stocks were determined using HEK293 cells and an Adeno-X Rapid Titer Kit (Takara Bio) according to the manufacturer's instructions.

Plaque formation assay
Vero E6/TMPRSS2 cells in six-well plates were washed with DMED-GlutaMAX, and inoculated with serially diluted SARS-CoV-2, and incubated at 37°C for 60 min with rocking every 15 min. After removing the viruses, cells were washed with DMEM-GlutaMAX and overlaid with agarose medium. After incubation of cells at 37°C for 2 days, the plaques were visualized with crystal violet staining and counted.

Viral RNA quanti cation
The left lung lobe from each mouse was homogenized in nine volumes of Leibovitz's L-15 medium (Thermo Fisher Scienti c, Waltham, MA, USA) using a Multi-Bead Shocker (Yasui Kikai, Osaka, Japan). Total RNA samples were extracted from 50 μL of the supernatant of lung homogenates using Isogen LS (Nippon Gene, Tokyo, Japan) according to the manufacturer's instructions. Fifty nanograms of total RNA was used for quantitating the SARS-CoV-2 genome. Viral RNA was quanti ed using a 1-step reverse transcription qRT-PCR, as described previously 62 . Viral loads were calculated as copies per 1 μg of total RNA.

Immunohistochemistry
The mice lungs were xed in 10% neutral buffered formalin, embedded in para n, sectioned at a thickness of 4-μm, stained with HE, and subjected to routine histological examination. Para n block sections were also used for staining of the SARS-CoV-2 N protein. Antigen retrieval was performed by autoclaving sections in 10 mM citrate buffer (pH 6.0) for 10 min, and then the sections were immersed in 0.3% hydrogen peroxide in methanol at room temperature for 30 min to inactivate endogenous peroxidase. The sections were blocked with BlockAce (DS Pharma Biomedical, Osaka, Japan) for 15 min, and incubated overnight at 4°C with 2 mg/mL of rabbit anti-SARS-CoV-2 N protein monoclonal antibody [HL344] (GenTex, Inc., CA, USA). Secondary labeling was performed by incubation at RT for 30 min with EnVision+ System-HRP labeled Polymer Anti-Rabbit (Dako Denmark A/S, Glostrup, Denmark), followed by color development with ImmPACT DAB Peroxidase Substrate (Vector Laboratories, Burlingame, CA, USA) at RT for 10 min. Nuclear staining was performed with hematoxylin solution.
TMTpro 12-plex MS analysis Lysates extracted from left lung lobes were processed and digested by using an EasyPep Mini MS Sample Prep kit (Thermo Fisher Scienti c) according to the manufacturer's protocol. Three mouse lungs were pooled, and 25 µg of peptides from each sample were labeled with 0.25 mg of TMTpro mass tag labeling reagent (Thermo Fisher Scienti c) according to the manufacturer's protocol. After TMT labeling, the 8 sample channels were combined in equal proportions, dried using a speed-vac, and resuspended in 0.1% TFA. Samples were fractionated into 8 fractions using High pH Reversed-Phase Peptide Fractionation Kit (Thermo Fisher Scienti c) according to the manufacturer's protocol. One microgram of peptide from each fraction was analyzed by LC-MS/MS on an EASY-nLC 1200-connected Orbitrap Fusion Lumos Tribrid mass spectrometer (Thermo Fisher Scienti c) equipped with an FAIMS-Pro ion mobility interface (Thermo Fisher Scienti c). Peptides were separated on an analytical column (C18, 1.6 µm particle size ´ 75 µm diameter ´ 250 mm, Ion Opticks) using 4-hr gradients (0% to 28% acetonitrile over 240 min) with a constant ow of 300 nL/min. Peptide ionization was performed using a Nanospray Flex Ion Source (Thermo Fisher Scienti c). FAIMS-Pro was set to three phases (-40. -60, and -80 CV) and a '1 sec cycle for a phase' data-dependent acquisition method was used where the most intense ions in every 1 sec were selected for MS/MS fragmentation by HCD. MS raw les were analyzed using a Sequest HT search program in Proteome Discoverer 2.4 (Thermo Fisher Scienti c). MS/MS spectra were searched against the SwissProt reviewed mouse reference proteome (UniProt). TMTpro-based protein quanti cation was performed using the Reporter Ions Quanti er node in Proteome Discoverer 2.4.

Statistical analysis
Statistical analyses were performed with Prism software (version 9.1.2; GraphPad, San Diego, CA, USA).
Statistical signi cance was assigned when p values were <0.05. Inferential statistical analysis was performed by one-way analysis of variance (ANOVA), followed by Tukey s test.   Histopathological analyses and cytokine levels of mouse lung infected with SARS-CoV-2 variants (a) Histopathologic ndings with HE staining and detection of SARS-CoV-2 N protein in mouse lungs (left lobe) infected with SARS-CoV-2. (b) Left lung homogenates were used for measurement of multiplex cytokines and chemokines using the Bio-plex suspension array system. Data represent means and SD, n=5. *p<0.05, **p<0.01. The colors of the asterisks indicate the following: black (vs mock) and gray (vs Wu-2020).

Figure 4
Proteomic landscape of SARS-CoV-2 infected mouse lungs (a) Volcano plots for mouse lung proteome of the indicated group compared adenovirus-infected/SARS-CoV-2 non-infected (mock) mice. Up-regulated (Log 2 ³ 1) and down-regulated (Log 2 £ -1) proteins and p value < 0.05 indicate threshold lines. The numbers of up-and down-regulated proteins are 403 and 411