Age-dependent pathogenic characteristics of SARS-CoV-2 infection in ferrets

While the seroprevalence of SARS-CoV-2 in healthy people does not differ significantly among age groups, those aged 65 years or older exhibit strikingly higher COVID-19 mortality compared to younger individuals. To further understand differing COVID-19 manifestations in patients of different ages, three age groups of ferrets were infected with SARS-CoV-2. Although SARS-CoV-2 was isolated from all ferrets regardless of age, aged ferrets (≥ 3 years old) showed higher viral loads, longer nasal virus shedding, and more severe lung inflammatory cell infiltration and clinical symptoms compared to juvenile (≤ 6 months) and young adult (1–2 years) groups. Transcriptome analysis of aged ferret lungs revealed strong enrichment of gene sets related to type I interferon, activated T cells, and M1 macrophage responses, mimicking the gene expression profile of severe COVID-19 patients. Thus, SARS-CoV-2-infected aged ferrets highly recapitulate COVID-19 patients with severe symptoms and are useful for understanding age-associated infection, transmission, and pathogenesis of SARS-CoV-2.

. However, the magnitude of impact of SARS-CoV-2 infection is by far the greatest due to the signi cantly larger number of human cases, and consequently, higher mortality.
Despite strict precautionary measures, such as social distancing policies and restriction of social gatherings, the number of SARS-CoV-2 infections continues to grow exponentially with a proportional increase in mortality 5 . Fortunately, several licensed COVID-19 vaccines with high e cacy have been developed in record time. It is expected that by the end of next year all populations will have access to these safe and effective vaccines. However, given the strong correlation between severe COVID-19 and increasing age, it is imperative to understand the etiology of severe disease and determine the most effective treatment strategies for high-risk groups. Although patients with COVID-19 are distributed among all age groups, the majority of patients with severe COVID-19 fall within the 30 to 79 years of age (87%) grouping, while younger patients aged 10 to 19 years only comprised about 1% of total cases 6 .
Moreover, higher morbidity and mortality rates have consistently been observed in aged human populations throughout the COVID-19 pandemic 7 . Similarly, mouse-adapted SARS-CoV showed strong age-dependent disease phenotypes in a mouse model 8,9 . Furthermore, when compared to the young adult population, the aged population is generally more susceptible to respiratory infections and shows a poor prognosis 10 . Thus, this indicates that age is a critical factor for COVID-19 viral disease severity.
In addition to the various clinical symptoms reported for COVID-19, systematic experimental studies are needed to further dissect the diverse disease manifestations among different age groups. While human clinical studies are highly valuable, a number of limitations, including ethical issues, behavioral and environmental variables, and the medical history of the patients, may impede identi cation of the fundamental cause of the disease in a timely manner. Hence, this necessitates the development of an appropriate animal model to aid in understanding transmission and pathogenesis, as well as elucidating host immune responses to SARS-CoV-2. The current hACE2 transgenic mouse model, which expresses the human ACE2 entry receptor for SARS-CoV-2, showed weight loss and virus replication in lungs following SARS-CoV-2 infection. Although some mice showed neurological symptoms which lead to fatal infection in a virus dose-dependent manner 11,12 , other clinical symptoms of infection, such as body temperature, sneezing, coughing, and lethargy, are di cult to monitor in mouse model. Further, the neurologic-related mortality also brings the suitability of this animal model into debate, as central nervous system infection is rarely observed in COVID-19 patients, underscoring the need for an animal model that could better represent the human disease state. Following the fortuitous discovery that ferrets have natural susceptibility to human in uenza viruses, ferrets have been recognized as a useful animal model for study of respiratory viruses, such as respiratory syncytial virus, parain uenza viruses, and SARS coronavirus [13][14][15][16] . Ferrets have a respiratory tract histo-anatomically analogous to that of humans, with similar anatomic proportions of the upper and lower respiratory tracts, density of submucosal glands in the bronchial wall, and number of generations of terminal bronchioles 15 , further supporting the adequacy of this model for study of human respiratory viral infections.
In order to address numerous critical scienti c questions ranging from basic virology to the development and assessment of novel drugs and vaccines for COVID-19, we have recently established a ferret model for SARS-CoV-2 infection and transmission that highly recapitulates pathological aspects of the human infection 17 . SARS-CoV-2-infected ferrets exhibited elevated body temperatures and virus replication was readily detected; infected ferrets shed the virus through nasal washes and in saliva, urine, and fecal specimens; SARS-CoV-2 was readily transmitted to naïve direct-contact ferrets, but less e ciently to naïve indirect-contact ferrets; and acute bronchiolitis was observed in infected lungs 17 . Thus, SARS-CoV-2 replicates e ciently in the respiratory tracts of ferrets without prior adaption. Compared to small mammalian models, various clinical signs found in COVID-19 patients are also observable in ferrets, including elevated body temperatures, nasal discharge, sneezing, and shedding of the virus through nasal washes, saliva, urine, and feces 17 . Moreover, investigation of viral transmission of SARS-CoV-2 revealed the ability to spread to naïve ferrets through direct-contact and indirectly through respiratory droplets 17 . Also, we have recently developed an aged ferret infection model (≥4 years old, equivalent to 70 years old in humans) for the emerging human severe fever with thrombocytopenia syndrome virus (SFTSV) that fully recapitulates human clinical manifestation 18 . SFTSV infection exhibits a severe clinical manifestation and increased fatality rate in patients 50 years and older, which was recapitulated in SFTSV-infected aged ferrets, as evidenced by severe symptoms, such as high temperature, weight loss, severe thrombocytopenia, and death.
To further explore the age-related disease severity observed in COVID-19 patients, we infected ferrets divided into three different age groups with SARS-CoV-2: ferrets under 6 months to simulate juveniles/children (G1), 1 to 2-year-old ferrets to simulate young adults (G2), and ferrets more than 3 years old to simulate patients over 50 years old (G3). To compare disease severity among these three groups, clinical symptoms, viral load in the respiratory tract, and lung histopathology were examined. Furthermore, RNA sequencing (RNA-seq) analysis was performed with lung tissues from SARS-CoV-2infected young adult or aged ferrets to compare differences in global and dynamic gene expression. As a result, the aged (G3) ferret group showed higher viral load and more severe clinical symptoms than juvenile (G1) and young adult (G2) ferret groups, and expressed high levels of gene sets related to type I interferon (IFN), activated T cells, and M1 macrophage responses. Thus, SARS-CoV-2-infected aged ferrets considerably resemble COVID-19 aged patients with severe symptoms. This demonstrates the ability of this animal model to recapitulate the age-dependent etiology of severe COVID-19, and indicates the feasibility of its use in the development of novel therapeutics and vaccines for the exact target group.

Clinical features of SARS-CoV-2 infection among ferrets of different ages
In order to elucidate the clinical manifestations of SARS-CoV-2 infection across age groups, ferrets (n=9/group) were inoculated with 10 5.8 of 50% tissue culture infective dose (TCID 50 )/mL of NMC-nCoV02 strain through the intranasal (IN) route. While ferrets less than 6 months of age (G1) showed no increase in temperature, both the G2 (1-2 years old) and G3 (more than 3 years old) groups of SARS-CoV-2 infected ferrets showed elevated temperatures at 2-6 dpi, where the G3 group showed a prolonged elevated temperature even at 10 dpi (Fig. 1A). This trend was also associated with changes in body weight, where the G1 group showed less than 5% weight loss during the entire SARS-CoV-2 infection period, while the G2 and G3 groups showed a maximal 10% weight loss at 6 dpi, followed by a rapid recovery of the G2 group from 6 dpi but not the G3 group (Fig. 1B). To compare clinical manifestations of SARS-CoV-2 infection, we developed an arbitrary scoring method to describe clinical symptom (CS) values based on a 20-minute observation period of cough, rhinorrhea, and reduced activity. These CS values were compared among ferret groups as described in Table 1 and Table S1. The G2 and G3 groups showed the highest CS values of 4.17 and 4.67 at 4 dpi, respectively, while the G1 group showed a maximal CS value of 1 at 2 dpi before quickly recovering within 4 days, thus exhibiting a mild to asymptomatic infection (Table 1 and  Table S1). Particularly, the G3 group showed a prolonged period of high CS value that lasted until 10 dpi. In addition, aged contact ferrets showed the highest and most prolonged CS value compared to young adult contact ferrets. However, juvenile contact ferrets did not show any symptoms of clinical disease during the course of infection (Table S2). Of note, there was no loss in body weight among contact groups. While a light elevation of body temperatures was observed among contact ferrets co-housed with aged and young adult groups, no increase in body temperatures was noted in contact ferrets in the juvenile group (Fig. S1). Collectively, these results revealed that aged ferrets exhibit more severe and persistent clinical features of SARS-CoV-2 infection compared to younger ferrets.
Comparison of viral titer and shedding period among different age groups of ferrets To evaluate whether the clinical features seen in the ferret groups were associated with the degree of virus replication, we measured infectious viral titers from nasal washes (Fig. 1C). SARS-CoV-2 was isolated from all infected ferrets, regardless of their ages, from 2 to 6 dpi, whereas ferrets in the G2 and G3 groups continued to shed infectious virus until 8 dpi. Comparison of viral titers revealed that the G3 group showed signi cantly higher viral titers (3.5 to 4.9 log 10 TCID 50 /mL) from 2 to 6 dpi than the G1 and G2 groups. While the G2 group showed higher viral titers at 2 dpi than the G1 group, comparable viral titers were observed thereafter. These results clearly demonstrate that aged ferrets (G3) shed higher amounts of infectious virus in nasal discharges for a longer period of time than younger ferrets.
Because gastrointestinal involvement has also been documented during coronavirus infections of animals and humans 19,20 , we collected fecal specimens from the infected ferrets and performed qRT-PCR to assess SARS-CoV-2 viral loads and shedding periods from the intestines of ferrets of different ages (Fig. 1E). While viral RNAs were detected in fecal specimens of all three groups from 2 to 4 dpi, the G3 group showed the highest viral RNA copy number from 2 to 8 dpi, followed by the G2 group. On the other hand, the G1 group showed a small peak of viral RNA at 2 dpi, which rapidly declined thereafter to an undetectable range at 6 dpi. To further evaluate the viral titer in respiratory organs of infected animals, three ferrets from each group were euthanized at both 3 and 5 dpi for measurement of viral titers in the nasal turbinate and lungs. As expected, the G3 group showed the highest viral titers among the groups at a maximum of 5.2 log 10 TCID 50 /g in nasal turbinate at 3 dpi, and the G2 group also showed signi cantly high viral titers compared with the G1 group (Fig. 1F). Consistently, the G3 group showed a signi cantly higher viral titer (2.6 log 10 TCID 50 /g) in lung tissues compared with the G1 and G2 groups at 3 dpi (Fig.   1G). These results demonstrate that the aged ferret group shows signi cantly higher SARS-CoV-2 titers in the respiratory tract compared to juvenile and young adult ferrets, correlating with the clinical disease severity.
Comparison of SARS-CoV-2 transmission in ferrets by age As viral titers differed signi cantly among different age groups, naïve ferrets (n=3/group) were co-housed and placed in direct contact (DC) with infected ferrets two days after the primary infection to compare the transmission e ciency of SARS-CoV-2 ( Fig. 1D, Fig. S1 and Table S2). Virus was detectable in nasal washes of the DC ferrets as early as 3 days post-contact (dpc), and ferrets co-housed with the G3 group showed the highest viral titers at all time points with a peak of 4.10 log 10 TCID 50 /mL at 5 dpc. The DC ferrets co-housed with G2 and G3 groups shed infectious virus up to day 9 post-contact, whereas the DC ferrets exposed to G1 ferrets showed the lowest viral titers all throughout the co-housing period, reaching an undetectable range of infectious virus after 7 dpc (Fig. 1C, right panel). This showed that as the aged ferret group harbored high SARS-CoV-2 titers in their respiratory tracts, they were able to effectively transmit the virus to naïve ferrets by direct contact. As seen with children who appear to be less infectious than adults infected with SARS-CoV-2 due to their mild clinical manifestation of disease 21 , the infected juvenile ferret group carried low virus titer and was not readily infectious to naïve ferrets upon direct contact.
Differential lung histopathology of infected ferrets of different ages COVID-19 has most commonly been shown to be associated with a spectrum of lung damage. The aged ferrets showed considerable in ammation affecting most parts of lungs compared to the juvenile and young adult ferrets that showed only mild to moderate in ammation. Notably, all aged ferrets showed more than 50% lung damage at 5 dpi, ( Fig. 2A and Fig S3). To further compare the extent of pulmonary damage among ferrets of different ages, RNAscope in situ hybridization and histopathological examination were conducted (Fig. 2, Fig. S2, and S3). Based on RNAscope analysis of SARS-CoV-2, we could nd more virus infected cells in young adult and aged ferrets compared with the juvenile group ( . To rule out differential ACE2 expression among age groups, which may affect disease manifestation and virus replication kinetics in ferrets, ACE2 RNAscope analysis was performed on ferret lung sections. This showed that there was no detectable difference of the ACE2 expression among different age groups (Fig. S4A-H). These results demonstrate that the severity of lung damage is closely associated with the number of SARS-CoV-2 RNApositive cells in the lungs, but not with ACE2 receptor expression levels.
Differential SARS-CoV-2 antibody neutralization titers and serum IgG antibody among ferrets of different ages To compare the serum neutralization antibody (NAb) titers among ferrets of different ages, blood was collected from each group of ferrets at 12 and 21 dpi. At 12 dpi, all tested ferrets showed NAb titers of 32-64 without signi cant differences among the groups. However, at 21 days, the NAb titers were at least four-fold higher in the G3 group (GMT 256) than in the G1 group (GMT 64) (Fig. 3A). Furthermore, the G3 group showed signi cantly increased NAb titers at 21 dpi compared to 12 dpi, suggesting continuous activation of the immune response as long as 21 dpi (Fig. 3B). To further evaluate the serum IgG antibody titer, we conducted an ELISA on serum from each ferret. ELISA results revealed that the young adult (G2) and aged adult (G3) groups showed increased anti-SARS-CoV-2 antibody titers at 21 dpi (Fig.   3C). However, there were no statistical differences of anti-SARS-CoV-2 antibody titers in the juvenile group at 12 and 21 dpi. Two ferrets of the juvenile group (G1, n=3) exhibited a decrease of anti-SARS-CoV-2 antibody titers, and only one ferret showed a similar anti-SARS-CoV-2 antibody titer at 12 dpi. In addition, IgG antibody titers in serum of direct contact ferret groups of different ages were measured by ELISA. While the juvenile contact group showed only minimally detectable IgG at 12 dpi (Fig. 3D), young adult or aged ferrets demonstrated moderate or marked increase in IgG titers, respectively. This indicates that similar to severe COVID-19 patients 22 , aged ferrets display severe clinical symptoms and high viral titers in the respiratory tract as well as high serum NAb and IgG responses.

Transcriptional pro le of immune-related genes in lung tissues of SARS-CoV-2-infected ferrets
To gain a comprehensive understanding of the transcriptional pro le among ferrets of different ages following SARS-CoV-2 infection, we performed RNA sequencing (RNA-seq) analysis of lung tissues from G1, G2, and G3 ferret groups at 2 and 5 dpi, and compared the results with those of non-infected middleaged ferrets (control, n=3). We rst analyzed the overall variation of the samples using principal component analysis (PCA). Distinct clusters were observed among G1, G2, and G3 groups at 2 dpi ( lled circle, lll), which were also clearly separated from that of control ferrets ( lled triangle, p) (Fig. 4A). Intriguingly, age-dependent clustering at 2 dpi was largely attributed to principal component (PC) 2, which, in the majority, was composed of interferon-stimulated genes (ISGs), such as IRF7, ISG15, and OAS1 ( Fig  4B, Table S3). These ndings indicate that genes related to IFN responses are highly enriched in aged ferrets compared to juvenile ferrets during the early stage of SARS-CoV-2 infection (2 dpi). In contrast, samples at 5 dpi (open circle) did not display any particular clustering by age in PCA. This phenomenon could be explained by the recovery of several animals (one ferret in each of the G2 and G3 groups, and two ferrets in the G1 group) from SARS-CoV-2 infection, which were clustered with the control group.
Gene Set Enrichment Analysis (GSEA) revealed that DEGs of the G3 aged group were highly enriched with gene sets related to the anti-viral innate immune response as well as the adaptive immune response (T cell activation) at 2 dpi (Fig. 4C). These immune activation features were maintained even at 5 dpi (Fig.  S5B). In contrast, tissue remodeling-related gene sets, such as trabecula formation, chondrocyte proliferation and corni cation, were highly enriched in the G1 juvenile group both at 2 and 5 dpi (Fig. 4C and Fig. S5B). These gene sets may be closely associated with matrix remodeling during the recovery of lung epithelial injury in the juvenile group.
Furthermore, Gene Set Variation Analysis (GSVA) with public gene sets revealed that genes related to B cell response and T cell response were predominantly enriched in the aged group (G3) at 2 dpi, compared to the juvenile (G1) or the young adult (G2) ferrets (Fig. 4D), which is in agreement with a recent clinical study that showed stronger antibody and T cell responses in severe COVID-19 patients than in patients with mild disease 23,24 . Moreover, other immune-related gene sets, including IFN response, macrophage activation and NK cell activation, were also highly enriched in the aged group at 2 dpi (Fig. 4D). Notably, gene sets related to type I IFN responses and activated M1 macrophages were signi cantly enriched in aged ferrets at the earlier stage of SARS-CoV-2 infection (Fig. 4E), suggesting a marked activation of immune response-associated in ammation promptly followed by the initial infection in aged ferrets. Moreover, chemokines, such as CCL4, CCL5 and CXCL10, were markedly upregulated in most aged ferret at 2 dpi compared to other groups. Similarly, high expressions of IFNB1, IFNG, IL1B, IL2, IL7, and TNF were also observed at 2 dpi (Fig. S6). At 5 dpi, chemokine and cytokine expressions then returned to normal levels. These data suggest that aged ferrets have increased expression of in ammatory cytokines and chemokines during the early phase of infection. Finally, we investigated whether the SARS-CoV-2infected aged ferrets also reproduced the natural course of severe COVID-19 as seen in humans. In fact, the gene sets upregulated in postmortem lung tissue of COVID-19 patients 25 or PBMCs from severe COVID-19 patients 26 were also highly enriched in aged ferrets (G3), especially at 2 dpi, compared to the juvenile (G1) and young adult (G2) ferrets ( Fig. 4F and G). These transcriptional pro les of immunerelated genes indicate that the SARS-CoV-2-infected ferret model re ects the immunological properties of age-dependent severe COVID-19 in humans.

Discussion
SARS-CoV-2 has already had devastating effects on the global community affecting myriad aspects of our lives. While effective vaccines will be readily available soon, the exact timeline is unclear and although there is nally hope on the horizon, things are expected to get worse in the meantime, as thousands of people die every day and are separated from their families, pushing the national health care system to the brink. Moreover, as part of current social distancing policies, student attendance in academic institutions is restricted in many countries and is being replaced with online classes. However, recent studies have suggested that although children and young adult populations may predominantly exhibit asymptomatic SARS-CoV-2 infections with low pathogenicity, they may still be carriers of infectious viruses 27,28 . In contrast, the majority of patients with severe morbidity and mortality are reportedly elderly people who have limited social activities compared to other age groups [29][30][31] . It became evident early on that the wide range of SARS-CoV-2 disease severity and the sequelae of infection correlated with the age of the a icted person and presence of pre-existing medical conditions. Thus, to investigate the differential and diverse clinical manifestations in COVID-19 patients of different ages, COVID-19 age-related disease severity was replicated in ferrets of three different age groups. Although currently there is no apodictic formula to calculate age equivalency between ferrets and humans, various reports indicate that average life span of domesticated ferrets is between 5 and 7 years. However, given their short life span and the observed onset of serious health problems as early as 3-4 years of age in most ferrets, a majority of veterinarians considered ferrets to be geriatric at 3 years of age 32 . Thus, although we cannot accurately correlate considering lifespan and developmental stages of both ferret and human, we believed that ferrets more than 3 years old could represent the aged human population of human.
In this study, we demonstrate the age-associated pathogenesis of SARS-CoV-2 infection using a ferret model. Comparison of clinical symptom values and virus load analysis showed that the aged ferret model fully recapitulates clinical manifestations of COVID-19 in humans (Table 1 and Fig. 1). Recent reports of human patients with SARS-CoV-2 infection in China indicate that aged and comorbid patients carry high virus loads and experience di culty recovering from severe pneumonia, which leads to high mortality 33 . This nding is well in line with the results of infected ferrets of different ages, although none of the aged ferrets died from SARS-CoV-2 infection in this study. Speci cally, aged ferrets showed higher lung damage score, increased virus loads in their respiratory tracts and higher virus shedding compared to the juvenile and young adult ferrets. Moreover, RNAscope in situ hybridization clearly demonstrated higher numbers of SARS-CoV-2 RNA-positive lung cells and in ltrating in ammatory cells in the aged ferret group than in the juvenile and young adult groups, which was ultimately correlated with severe viral pathogenesis in the aged ferret group. While differential ACE2 expression among age groups might affect disease manifestation and virus replication kinetics in ferrets, we found no detectable difference of the ACE2 expression among different age groups. On the other hand, aged ferrets showed high expressions of chemokines (CCL4, CCL5, and CXCL10) and pro-in ammatory cytokines (IFNB1, IFNG, IL1B, IL2, IL6, and IL7) in the early phase of infection.
High expressions of chemokines and pro-in ammatory cytokines may contribute to the aberrant in ammatory response, which in turn causes severe pulmonary pathologies in aged adult ferrets. These ndings partially correlate with recent reports of severe COVID-19 cases with increased expression of proin ammatory cytokines and chemokines, which are associated with pulmonary in ammation and extensive lung damage 34,35 . Cytokine storm is potentially life-threatening event related to COVID-19. Patients with severe COVID-19 often exhibit acute respiratory distress syndrome, a consequence of cytokine storm resulting from the marked expression of a combination of immune-active molecules. Thus, the pathology of SARS-CoV-2 in infected aged ferrets considerably re ects that of aged COVID-19 patients, making it a valuable animal model to understand the age-dependent viral pathogenesis of SARS-CoV-2. Of note, directly infected ferrets showed the highest viral titer at 2 dpi, which then gradually decreased until 8 (the juvenile group) or 10 dpi (adult and aged ferret groups). However, animals infected through direct contact transmission showed moderate virus titers at 3 dpc which then peaked at 5 dpc in all groups with exception of the juvenile group. This differential phenomenon in infection and transmission groups may be explained by the initial infection dose. The direct infection groups were infected with a high dose of virus (10 6.0 TCID 50 /mL, which is almost the maximum titer in ferret respiratory tracts) and showed a gradual decrease of virus titer after the maximum titer was reached at 2 dpi. In contrast, ferrets infected by direct contact were presumably infected with lower titers, resulting in virus replication that the maximum titer was reached in naïve animals on 5 dpc. Therefore, the natural infection pattern of SARS-CoV-2 may be more similar to that of the direct contact groups. For example, the juvenile group was presumably exposed to lower amounts of infectious virus which were rapidly cleared in naïve animals.
The role of juveniles (<18 years of age) in the spread of SARS-CoV-2 has not been fully de ned. A recent study shows that children are not likely to be the source of SARS-CoV-2 transmission and outbreak, and thus, are minor drivers of the COVID-19 pandemic 20 . In contrast, juveniles typically exhibit the highest rates of in uenza virus infection and are considered to play a critical role in in uenza virus spread. Thus, it is believed that because juveniles exhibit mild clinical manifestations of COVID-19 they are less likely to transmit infectious SARS-CoV-2 than adults, who exhibit more severe symptoms. Supporting these reports, the present study shows that the infected juvenile ferrets carry low virus titers and are not readily infectious to naïve contact animals. This indicates that besides aged ferrets, juvenile ferrets can also potentially be a useful animal model in understanding the important scienti c aspects of the infrequent transmission of SARS-CoV-2 from children to other children and from children to adults.
Although the kinetics of antibody development in SARS-CoV-2 infected-ferrets were comparable among all three groups at 12 dpi, NAb titers were two-and four-fold higher in the young adult and aged ferret groups, respectively, than in the juvenile group (Fig. 3). It is noteworthy that the aged ferret group, which demonstrated more severe clinical symptoms along with the highest viral loads in the respiratory tract, showed a four-fold increase in NAb titer at 21 dpi over that at 12 dpi, suggesting a close association between clinical outcomes and antibody production. This result is also well within agreement with a recent study 35 reporting that SARS-CoV-2 serum neutralizing antibody levels were higher in severe SARS-CoV-2 patients than in asymptomatic or mild patients. Thus, SARS-CoV-2-infected aged ferrets can also be used to understand the immunological aspects underlying the high neutralizing antibody titer in patients with more severe COVID-19.
Recently, a transcriptome analysis study of COVID-19 patients demonstrated robust induction of type I IFN responses in severe COVID-19 patients compared to mild/moderate COVID-19 patients 25 . Transcriptome analysis during the early stage of SARS-CoV-2 infection in ferrets also revealed enrichment of genes involved in both innate and adaptive immune responses in aged ferrets compared to juvenile and young adult ferrets. Notably, the gene sets related to the type I IFN response and activated M1 macrophages were signi cantly enriched in SARS-CoV-2-infected aged ferrets at 2 dpi. Furthermore, infected aged ferrets also showed enrichment of genes also expressed in tissues of patients with severe COVID-19, such as lung tissues of post-mortem COVID-19 patients and PBMCs from patients with severe COVID-19 symptoms 25,26 . These results indicate that the SARS-CoV-2-infected aged ferrets not only recapitulate the clinical course of severe symptoms seen in COVID-19 patients, but also the corresponding alteration of transcriptional pro les. In addition, the aged ferrets demonstrated upregulation of genes related to early activation of an adaptive immune response, including T cell and B cell responses, which was maintained up to 5 dpi. These ndings may provide a clue to the mechanisms underlying the relatively weak SARS-CoV-2-speci c T cell responses and attenuated neutralizing antibody activity observed in asymptomatic or mild COVID-19 patients 37 . Thus, this study provides fundamental information of the in vivo gene expression dynamics of the host immune response as seen in hyperin ammatory responses provoked by severe SARS-CoV-2 infection in COVID-19 patients 26,38 .
Taken together, aged ferrets showed signi cantly higher virus loads and more severe lung pathology compared to juvenile and young adult ferrets. Moreover, these differences were closely associated with enhanced type I IFN responses and activated M1 macrophages as well as hyper-in ammatory responses in aged ferrets. This aged immune-competent ferret model demonstrates for the rst time age-dependent pathogenesis of SARS-CoV-2 infection, making it an invaluable animal model to understand the agedependency of COVID-19 pathogenesis and the detailed underlying mechanism of asymptomatic infection in juveniles and young adults.

Methods
Study design for age-dependent pathogenesis in ferrets SARS-CoV and SARS-CoV-2 antibody-free female ferrets over 36 months (3 years ≤ Age, n=9), 12-24 months (1 ≤ Age ≤ 2 years, n=9), and under 6 months (Age ≤ 6 months, n=9) were infected through the intranasal (IN) route with the NMC-nCoV02 strain (GISAID accession number: EPI_ISL_1069194) at a dose of 10 5.8 TCID 50 per ferret. Nasal washes and fecal specimens were collected every day for 10 days from each group of ferrets to measure viral titers. To measure the infectious live virus in the collected specimens, each sample was inoculated with Vero cells, which were then incubated for 4 days prior to virus isolation. To assess viral replication in various organs of ferrets following SARS-CoV-2 infection, ferrets (n=3/group) were sacri ced at 3 and 5 dpi to harvest their nasal turbinates and lung tissues with individual scissors to avoid cross-contamination. The left lung lobes from the harvested whole lungs were homogenized for virus titration in Vero cells and the right lung lobes were immediately xed in 10% neutral-buffered formalin solution for further histopathological examinations.

RNA-Sequencing
Total RNA was isolated using TRIzol reagent (Invitrogen) according to the manufacturer's instructions.
RNA quality was assessed with Agilent 2100 Bioanalyzer using an RNA 6000 Nano Chip (Agilent Technologies), and RNA was quanti ed using an ND-2000 Spectrophotometer (Thermo Fisher Scienti c). Extracted RNAs were processed using the TruSeq Stranded mRNA Sample Prep Kit (Illumina) according to the manufacturer's instructions. High-throughput sequencing was performed as paired-end 150 sequencing runs using NovaSeq 6000 (Illumina). Raw reads were assembled and low-quality reads were ltered using Cutadapt (version 2.8). Filtered reads were aligned on a reference genome downloaded from Ensembl (MusPutFur1.0, Accession number: GCF_000215625.1) using STAR (version 2.7.1a) and annotated with additive human ortholog genes from the human reference database (Biomart database, GRCh38). Gene counts were normalized to valid library size and the dimensional reduction was performed by principal components analysis (PCA) using the top 2 principal components (PCs) throughout the whole samples. Two-sided Wald test was performed to analyze the differentially expressed genes (DEGs) according to each condition with DESeq2 (version 1.26.0) 39 . DEGs were determined according to cutoffs of a p-value < 0.05 and a log2 fold change > 0.5 for by day and > 1 for by age. To analyze functional pro les of each condition, gene set enrichment analysis (GSEA) was performed in Gene ontology: Biological process database (GO. BP) and speci c public gene set using clusterPro ler (version 3.14.3) [40][41][42] . To compare the gene set enrichment score for the speci c gene sets, gene set variation analysis (GSVA) (version 1.34.0) was performed 43 . To calculate the severe COVID-19 signature score, signi cantly upregulated genes in severe COVID-19 patients were used 25,26 .
RNAscope in situ hybridization and pathology SARS-CoV-2 RNA (Spike gene) and ACE2 were detected using the Spike-speci c and ACE2 probe (Advanced Cell Diagnostics, Cat. # 848561, # 848151) and visualized using RNAscope 2.5 HD Reagent Kit RED (Advanced Cell Diagnostics, Cat. # 322360). Lung tissue sections were xed in 10% neutralbuffered formalin and embedded in para n, according to the manufacturer's instructions, followed by counterstaining with Gill's hematoxylin #1 (Polysciences, cat # 24242-1000). For pathological examination, the embedded tissues were sectioned and dried for three days at room temperature. Histopathological examination was conducted by hematoxylin and eosin (H&E) staining. Slides were viewed using Olympus IX 71 (Olympus, Tokyo, Japan) microscope with DP controller software to capture images.

Histopathology scoring analysis
Quanti cation of SARS-CoV-2 RNA positive-and ACE2 positive cells. From 400x magni cation area, all positively stained cells (SARS-CoV-2 RNA positive-and ACE2 positive) were counted in triplicate and the average calculated. Ferret lungs were graded depending on the degree of in ammation observed in each individual lung at 3 and 5 dpi.

Serologic assay
The neutralizing antibody (NAb) assay against SARS-CoV-2 was carried out using a micro-neutralization assay in Vero cells. Collected ferret serum specimens were inactivated at 56°C for 30 min. Initial 1:2 serum dilutions were made with the medium, and two-fold serial dilutions of all samples were made to a nal serum dilution of 1 Observational clinical symptoms: Cough, rhinorrhea, movement, and activity. Score: 0; normal, 1: occasional, mild reduced activity, 2: frequent, reduced activity.
Page 19/23 * Scores were measured by observation of clinical symptoms for at least 20 minutes in each group of ferrets based on the following criteria: Cough: 0; no evidence of cough, 1; occasional cough, 2; frequent cough (score 2). Rhinorrhea: 0; no nasal rattling or sneezing, 1; moderate nasal discharge on external nares, 2; severe nasal discharge on external nares. Movement, activity: 0; normal movement and activity, 1; mild reduced movement and activity, 2; evidence of reduced movement and activity. Transcriptional pro le of immune-related genes in lung tissues of SARS-CoV-2-infected ferrets. Principal component analysis (PCA) scatter plot of gene expression of the SARS-CoV-2 infected ferrets showing distinct cluster according to ages, associated with interferon stimulated signature (A). Heatmap of agespeci c differentially expressed genes (DEGs) compared to control ferrets shows that genes related to innate immune response r were upregulated in 2 dpi (B). The 'Common' gene sets were composed of genes differentially upregulated in more than two groups, while 'Juvenile-speci c', 'Young adult-speci c' and 'Aged-speci c' gene sets were composed of genes uniquely upregulated in each group.

Figures
Representative immune-related genes were listed next to the heatmap. Bar plots with normalized enrichment score (NES) from enrichment analysis of representative Gene Ontology (GO) biological pathway shows anti-viral immune responses were enrich in both 2 dpi (C). Heatmap of gene set variation analysis (GSVA) with immune related GO biological pathway shows that various immune responses were upregulate in aged compared to juvenile or young adult ferrets (D). J, Juvenile; Y, Young adult; A, Aged.