Animal Models for COVID-19: Hamsters, Mouse, Ferret, Mink, Tree Shrew, and Non-human Primates

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is a novel coronavirus causing acute respiratory tract infection in humans. The virus has the characteristics of rapid transmission, long incubation period and strong pathogenicity, and has spread all over the world. Therefore, it is of great significance to select appropriate animal models for antiviral drug development and therapeutic effect evaluation. Here, we review and compare the current animal models of SARS-CoV-2.


INTRODUCTION
In early December of 2019, a new coronavirus disease (coronavirus disease 2019, COVID-19) caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) was reported in Wuhan, China . SARS coronavirus (SARS-CoV), Middle East respiratory syndrome coronavirus (MERS-CoV), and SARS-CoV-2 all belong to the Betacoronavirus genus of Coronaviridae family and have typical coronavirus genome structure . Bats are suspected to be the natural host of SARS-CoV-2 (Banerjee et al., 2019), while other animals such as turtles, pangolins, snakes, cats, rabbits, ferrets, and monkeys have been suggested as potential intermediate hosts (Liu et al., 2020a;Wu et al., 2020). The spike (S) protein on the surface of SARS-CoV-2 is responsible for binding angiotensin converting enzyme 2 (ACE2) receptor on target cell surface . The interaction between viruses and host-specific receptors is an important factor that limits species tropism of the pathogen (Douam et al., 2015).
Severe acute respiratory syndrome coronavirus 2 infection may cause mild to severe symptoms. Severe outcomes include acute respiratory distress syndrome (ARDS), sepsis, septic shock, multiorgan failure, and death (Harapan et al., 2020). According to one survey, 80.9% infections were mild, 4.7% eventually developed critical symptoms, and the overall case fatality rate was 2.3% (Zhang, 2020) which is lower than MERS (34.4%) and SARS (9.5%) (Munster et al., 2020b). However, some patients (especially elderly patients with diabetes or coronary heart disease) can deteriorate in a short period of time and rapidly progress to multi-organ failure or death (Chen N. et al., 2020;Zhou F. et al., 2020). Other conditions associated with high mortality in adults include hypertension and other preexisting respiratory diseases (Williamson et al., 2020). Similar to SARS and MERS, fast deterioration in COVID-19 patients has been linked to the development of cytokine storm (Huang et al., 2005;Zhou et al., 2014). Serum levels of pro-inflammatory cytokines, such as interleukin-6 (IL-6), IL-1β, IL-2, IL-8, IL-17, granulocytecolony stimulating factor (G-CSF), monocyte chemotactic protein-1 (MCP-1), interferon-γ-inducible protein-10 (IP-10), and interferon (IFN), are upregulated in most patients with severe symptoms Tan et al., 2020) as a result of excessive immune response to SARS-Cov-2 infection . Besides, it has been observed that most severe cases have lymphopenia, displaying a sharp decrease in the number of CD4 + T cells, CD8 + T cells, B cells, and natural killer cells. Total leukocyte number was not remarkably changed in such patients, but neutrophils were increased Tan et al., 2020), and the increase in neutrophil-to-lymphocyte ratio (NLR) has been associated with disease severity and poor clinical outcome (Cao, 2020). Such anomalies in cell counts and NLR will eventually disappear in convalescent patients .
Animal models are valuable tools for elucidating the pathogenesis of SARS-CoV-2 and evaluating antiviral therapeutics and vaccines. We searched the literature for currently used COVID-19 animal models. This review attempts to summarize progresses made since the discovery of SARS-CoV-2 in the development and application of relevant animal models, and discusses each model's strength and weakness ( Table 2). Details about each animal model are listed in Table 1, including gender, age, inoculation, replication, pathology, immune response, and clinical signs.

CHINESE TREE SHREW
Chinese tree shrew (Tupaia belangeri chinensis), a squirrel-like and rat-sized mammal, is genetically close to primates (Fan et al., 2013). As emerging experimental animal, tree shrews have the advantages of small size, low feeding cost, and short reproductive cycle, and have been used as effective experimental animals to replace primates in biomedical research and drug safety testing, especially for research targeting hepatitis C and B viruses (Yan et al., 1996;Feng et al., 2017).
Two teams have conducted SARS-CoV-2 infection of tree shrews. In one study, after nasal inoculation, tree shrews displayed no clinical symptom except for an increase in body temperature, and SARS-CoV-2 RNA became detectable in nasal, throat, and anal swabs, and one serum sample. Highest virus shedding (copy number 10 5.92 /ml) was recorded with a young tree shrew's nasal swab at 6 days post inoculation (dpi) and the highest viral load is 10 9.08 /ml obtained using pancreas of one adult tree shrew dissected at 14 dpi . In contrast,  failed to detect SARS-CoV-2 RNA in throat and anal swabs. In addition, lung was the main site for viral replication, followed by the digestive tissues and heart ( Table 1; Zhao et al., 2020). Sections of lung tissues showed local mild lesions, including thickened alveolar septa and interstitial hemorrhage, in both studies. Zhao et al. (2020) also found some scant histopathological changes in brain, heart, liver, trachea, pancreas, etc. (Table 1). Moreover, laboratory examination found that serum albumin decreased after infection, while blood urea nitrogen and alanine aminotransferase increased, similar to the impaired liver and renal function observed during COVID-19 progression (Ronco et al., 2020;Zhang C. et al., 2020). It appears that SARS-CoV-2 only achieves limited virus replication and shedding, as well as insignificant pathogenesis, in tree shrews.

HAMSTER
Hamsters are widely used in the research of respiratory viruses (Schaecher et al., 2008;Iwatsuki-Horimoto et al., 2018). Hamster ACE2 binds tightly to the SARS-CoV-2 S protein and mediates its entry . Therefore, hamsters are a potential infection model for studying the pathogenesis of SARS-CoV-2 infections.

Syrian Hamsters (Mesocricetus auratus)
Syrian hamsters started developing rapid breathing, lethargy, ruffled fur, and weight loss at 2 dpi of SARS-CoV-2. The mean viral loads in the nasal turbinate, trachea, and lung were consistently the highest among all tested tissues ( Table 1; Chan et al., 2020). Sia et al. (2020) detected SARS-CoV-2 replication in the duodenum but observed no obvious structural damage or inflammatory infiltration, which is similar to virus replication in the epithelial cells of the terminal ileum and colon of COVID-19 patients.
After infection, Syrian hamsters showed similar pathological manifestations as human COVID-19 pneumonia. Lung tissue displayed focal diffuse alveolar destruction, hyaline membrane formation, and mononuclear cell infiltration (Table 1; Chan et al., 2020). In addition, neutrophil extracellular traps, an important pathogenetic trigger in severe COVID-19 patients (Middleton et al., 2020), was observed in lungs of infected Syrian hamsters (Becker et al., 2021). Computed tomography (CT) detected patchy ground glass opacity with regions of lung consolidation at 2 dpi, and the most severe lung changes occurred at 7-8 dpi (Imai et al., 2020).
Severe acute respiratory syndrome coronavirus 2 infection could cause different degrees of multiple organs lesions in Syrian hamster, including spleen, lymph nodes, kidney, adrenal gland, ovary, etc. Among them, the adrenal gland showed focal to multifocal inflammation . Splenic atrophy and myocardial degeneration could also be observed (Chan et al., 2020), which may be related to the occasional reports of heart failure and lymphopenia in human patients (Chen N. et al., 2020;Huang et al., 2020).
A study found that high-dose intranasal inoculation of SARS-CoV-2 (5 × 10 5 TCID 50 ) in hamsters resulted in severe infection, whose primary manifestation was high levels of virus replication, severe extensive pneumonia, and partial mortality (Tostanoski et al., 2020). The high-dose hamster model recapitulates severe COVID-19 and might help to study the pathogenesis of serious clinical disease and to test antiviral agents. Disrupting of adaptive immunity is also a strategy for developing severe disease models. Cyclophosphamide-treated and RAG2 knockout (KO) hamsters infected with low doses of virus developed more severe and prolonged disease (Brocato et al., 2020). Low-dose virus-infected RAG2 KO hamster, which was marked by hemorrhage and severe edema in lung and rapid onset of disease (by 3 dpi), is a uniformly lethal model (Table 1). It demonstrates that the absence of functional B cells and/or T cells exacerbates pathogenesis at an early stage of infection. The gene expression of the cytokine/chemokine profiles in the lungs of infected hamsters showed a time-dependent trend. IFN-γ and pro-inflammatory cytokines, IL-4, IL-6, TNF-α, and IL-12p40, were induced at 2 dpi and reached peak levels at 4 dpi (Chan et al., 2020). The lymphoid necrosis observed in hamster's spleen might be associated with its innate immune pattern with high levels of cytokines, similar to the cytokine storm in human severe infection (Chan et al., 2020).
Neutralizing antibodies induced by primary SARS-CoV-2 infection protected hamsters from subsequent infection. Moreover, passive serum transfer also protected naive hamsters against viral replication in the lung (Imai et al., 2020). SARS-CoV-2 was transmitted efficiently between individual hamsters by direct contact and via aerosols (Sia et al., 2020). Therefore, Syrian hamsters can be used for research on viral transmissibility. At present, the replicative abilities, pathogenicity, transmissibility, and susceptibility to neutralization of some virus variants (P.1, D614G, and a variant with a deletion at the S1/S2 junction of the spike protein) have been evaluated in hamsters Imai et al., 2021;Plante et al., 2021;Wang et al., 2021).
Moreover, the hamster model appears to mimic the gender bias and age-dependent differences in COVID-19 patients Yuan et al., 2021). Younger animals launched earlier and more severe lesions and a faster recovery after infection . Male hamsters showed relatively severe clinical symptoms and more severe lung damage, while female hamsters behaved like human asymptomatic carriers with lower levels of virus replication. Higher receptor binding domain (RBD)-specific IgG levels in female hamsters indicated that the stronger humoral immune response in female contributes to reducing disease severity (Yuan et al., 2021). Therefore, Syrian hamster is an appropriate animal model for developing gender-based and age-different treatments of COVID-19.
It would be interesting to incorporate into these models known risk factors for severe COVID-19, such as metabolic syndrome, respiratory disease, hypertension, and old age, which might produce valid models of COVID-19 with comorbidities (Williamson et al., 2020). For example, the detrimental impact of a continuous Western diet on COVID-19 outcome has been proved in Syrian hamster (Port et al., 2021).

Roborovski Dwarf Hamster (Phodopus roborovskii)
Roborovski dwarf hamster, the smallest hamster in the world, has also proven to be a highly susceptible model of SARS-CoV-2. Hamsters infected with 10 5 plaque forming units (PFU) SARS-CoV-2 developed a rapid onset of fulminant clinical disease, with a sudden fall of body temperatures at 1-2 dpi and severe acute diffuse alveolar damage and hyaline microthrombi in the lung (Table 1; Trimpert et al., 2020). The lungs, jejunal, and kidneys of Roborovski dwarf hamsters contained high virus loads. Interestingly, Trimpert et al. (2020) found that Roborovski dwarf hamsters infected with low doses (5,000 PFU) still developed infection, but hyaline thrombi were not observed in such hamsters. It appears that Roborovski dwarf hamsters are highly susceptible to SARS-CoV-2 even at low doses, and mimic the clinical and pathological outcome observed in severe cases of COVID-19 ( Table 2; Trimpert et al., 2020). Because the amino acid sequence at the ACE2-RBD interface of Roborovski dwarf hamster differs from the Syrian hamster by only one amino acid (Figure 1), this minor difference is inadequate to explain its high susceptibility to SARS-CoV-2 and its fatal outcomes. Perhaps it is related to the difference in the host's immune response. Unfortunately, there is not enough research to clarify this issue due to the lack of dedicated tools and reagents for Roborovski dwarf hamster.

Chinese Hamster (Cricetulus griseus)
Chinese hamsters are smaller than Syrian hamsters, and have been tested for SARS-CoV-2 infection . Decreases in body weights post infection were significant and infected hamsters failed to regain lost weight by the end of the experiment at 14 dpi. Histopathology revealed pneumonia and alveolar damage similar to what has been observed in Syrian hamsters ( Table 1; Bertzbach et al., 2020). However, the course of bronchitis and pneumonia described in Chinese hamsters was milder compared to Syrian hamsters, and pneumonia was more prolonged. Evident acute alveolar damage was present until 14 dpi with persistence of viral RNA . Chinese hamsters are highly susceptible to diabetes, which may be used to establish COVID-19 model with preexisting diabetes (Gerritsen, 1982).
Owing to the similarities to COVID-19 patients regarding clinical symptoms and histopathology, hamster-based models have become important tools for the study of SARS-CoV-2. Various drugs, antibodies, and vaccines are already being tested in Syrian hamsters and progresses have been reported (Kaptein et al., 2020;Kreye et al., 2020;Rogers et al., 2020;Kim et al., 2021;Ku et al., 2021).

MOUSE
Compared with other lab animals, mice offer many practical advantages including small sizes, multiple well-established strains, clear genetic background, highly available research tools, and ease of genetic manipulation. Due to the low affinity of mouse ACE2 (mACE2) for S protein of SARS-CoV-2, mice cannot be efficiently infected with wild type viruses . Two main approaches were adopted for circumventing the incompatibility of mACE2 with SARS-CoV-2 S protein: (1) the use of mouse-adapted SARS-CoV-2 strains; (2) expression of human ACE2 (hACE2) in mice.

Mouse-Adapted SARS-CoV-2 Strains
SARS coronavirus clinical isolates successively passaged in the respiratory tract of mice generated mouse-adapted strains MA15 Long production time and high cost; suboptimal SARS-CoV-2 replication.
AdV-hACE2 mice Fast production at low-cost. hACE2 expressed non-systematically; mild bronchial inflammation associated with AdV; no extrapulmonary manifestations.

Ferret
Display symptoms of cough and fever, and transmission between animals.
Virus replication only in the upper respiratory tract; mild lung pathological damage.

Mink
Severe to fatal infections; lung lesions highly similar to those in human patients.
Not widely used; aggressive animal.
African green monkey Robust viral replication; lethal outcome in aged; manifestations of ARDS.
High cost; limited animal sources.
Rhesus macaque Phylogenetically close to human; symptoms similar to human infections; age-dependent increase in disease severity.
Non-lethal outcome; high cost.
Cynomolgus macaques Phylogenetically close to human; asymptomatic infection in aged animals.
Non-lethal outcome; high cost.
Common marmoset Phylogenetically close to human; smallest monkey allowing easy handling; breed well in captivity.
Relatively resistant to SARS-CoV-2 infection. The 20 amino acid residues at the interface between human ACE2 and SARS-CoV-2 RBD are shown on the right . Each amino acid is identified by a different color. and v2163, which caused lung damage and death in mice (Roberts et al., 2007;Day et al., 2009). Gu et al. (2020) used the same strategy and produced mouse-adapted SARS-CoV-2 strain MASCp6, which was infectious to mice and caused inflammatory responses and moderate pneumonia in both young and aged mice. The adaptive mutation of the strain, Asn501 to Tyr, at the key site in the RBD of viral S protein increased the binding affinity of the protein to mACE2. Another mouse-adapted strain, HRB26M, only caused mild pathological changes in lung, same as MASCp6 . This method enables fast establishment of mouse model of infection to provide a large number of experimental animals that can be directly applied to the evaluation of therapeutic and vaccine candidates. However, wild type mice infected with mouse-adaptive virus strains may not necessarily reflect the situation of wild type SARS-CoV-2 infection ( Table 2; Daniloski et al., 2020). Dinnon et al. (2020) used reverse genetics to remodel the S and mACE2 binding interface and constructed a recombinant virus (SARS-CoV-2 MA) that utilizes mACE2 for entry. This recombinant strain replicated in the respiratory tract and was cleared within 4 dpi, causing mild to moderate pneumonia ( Table 1). Leist et al. (2020) obtained a new mouse-adapted SARS-CoV-2 virus with increased pathogenicity, SARS-CoV-2 MA10, via serial passage of SARS-CoV-2 MA in the lungs of mice. MA10 infection caused acute lung injury and partial mortality in young BALB/c mice. The mortality rate of standard infection dose (10 4 PFU) was 20% . Compared to mice with MA infection, more severely impaired lung function, significant changes in PenH (enhanced pause) and Rpef (the fraction of expiration time at which the peak occurs) at 2 dpi, was observed in MA10 infected adult mice. Besides, the levels of several pro-inflammatory cytokines (Table 1) in the lungs of aged adult mice with SARS-CoV-2 MA or MA10 infection, rather than in the serum, increased at 2 dpi. This more localized cytokine/chemokine responses may explain the longer time of virus clearance in aged mice than in young mice Leist et al., 2020). The above two mouse-adapted strains exhibit an age-dependent disease severity in mice similar to COVID-19 patients ( Table 2; Dinnon et al., 2020;Leist et al., 2020). In comparison to BALB/c mice, the infection of MA 10 led to less severe and non-lethal disease in C57BL/6J mice, revealing host genetic background could affect SARS-CoV-2 disease susceptibility . Recently, a mouseadapted strain WBP-1 was found to produce lethal lung infections in mice. Q493K and Q498H substitutions in S protein of WBP-1 increased the binding affinity toward mACE2 (Huang et al., 2021).

Transgenic Mice
Various antiviral agents and vaccines have so far been evaluated in transgenic mouse models for COVID-19 (Liu et al., 2020b;Ma et al., 2020;Colunga Biancatelli et al., 2021). After intranasal inoculation with SARS-CoV-2, transgenic mice expressing hACE2 by the mACE2 promoter showed symptoms of girdling and weight loss, as well as virus replication in the lungs . Infected mice progressed to moderate interstitial pneumonia with diffuse lesions at 5 dpi, and the pneumonia became mild with focal lesion areas at 7 dpi, while other organs displayed no obvious lesions ( Table 1). In addition, a large number of macrophages and CD3 + T lymphocytes were found in the alveolar interstitium of hACE2 mice, and their numbers gradually increased with the course of infection . Another stable mouse model was generated, replacing the endogenous mACE2 with hACE2, by using CRISPR/Cas9 knockin (KI) technology (Sun S.H. et al., 2020). These hACE2 mice are highly susceptible to SARS-CoV-2 and demonstrate more robust virus replication in lung than hACE2 transgenic mice . Moreover, hACE2-KI mice (C57BL/6 background) with intratracheal inoculation of SARS-CoV-2 developed severe pneumonia with hyaline membranes-like changes, and can be used as an animal model for SARS-CoV-2-induced ARDS (Hong et al., 2021).
Two transgenic mouse models with hACE2 expression driven by the heterologous promoters showed lethal disease Winkler et al., 2020). After intranasal inoculation of the virus, K18-hACE2 mice and HFH4-hACE2 mice (hACE2 expression under the control of human cytokeratin 18 promoter and HFH4/FOXJ1 promoter, respectively) lost weight and generated typical interstitial pneumonia. K18-hACE2 and HFH4-hACE2 mice with SARS-CoV-2 neuroinvasion would succumb to the infection. Severe pneumonia in K18-hACE2 mice was characterized by high levels of pro-inflammatory cytokines and chemokines (IFN-β, IL-6, IL-11, CXCL10, etc.) and substantial inflammatory cell infiltration in the lung. Lung consolidation caused by infection led to reduced exercise tolerance and impaired lung function in those mice . HFH4-hACE2 mice with SARS-CoV-2 infection showed increased levels of creatine kinase, which may be related to edema and necrosis in some cardiomyocytes observed in heart tissue of mice .

Viral Vector-Mediated Delivery Systems
Human ACE2 expression in lung tissues of BALB/c mice was also achieved by intranasal administration of replication-defective adenoviruses (AdV) encoding hACE2 (AdV-hACE2). Five days later, mice were inoculated with SARS-CoV-2 and displayed weight loss, high virus replication in the lungs and pulmonary damage (Hassan et al., 2020). Lower levels of SARS-CoV-2 RNA could be detected in brain, liver, and spleen, but not in kidney, gastrointestinal tract, or serum ( Table 1). The tissue distribution of virus infection may be related to the expression of hACE2 in different tissues in addition to the natural tropism of the virus (Hassan et al., 2020). Interestingly, Hassan et al. (2020) also found that there was no substantial difference between nasal inoculation alone and intranasal-intravenous combined routes in lung infection. It indicates that virus inoculation by a systemic route is not required in this animal model. Infected mice also showed a substantial increase in neutrophil infiltration in perivascular and alveolar locations after 8 dpi, and mice given anti-IFNARI1 monoclonal antibody showed more severe infection with an increase in macrophage infiltration at 8 dpi, similar to severe pneumonia in COVID-19 patients (Hassan et al., 2020). Those AdV-hACE2 mice treated with anti-Ifnar1 monoclonal antibody had relatively high levels of proinflammatory cytokines and chemokines such as IL-6, CCL2, CCL5, CXCL10, CXCL11, IFN-λ, and IFN-β in lung, indicating that inhibiting type I IFN signaling can aggravate inflammation and promote infection.  also used AdV-hACE2 for enabling SARS-CoV-2 infection of BALB/c and C57BL/6 mice, and found that infection in the two strains followed almost the same disease course. Infected BALB/c mice showed ruffled fur, hunching, and difficulty breathing after 2 dpi, and lost up to 20% of body weight at 4-6 dpi, while C57BL/6 mice lost about 10-15%. Histological examination showed that the most severe changes in the lungs were observed at 5 dpi . Compared with AdV-Empty transduced mice, expression of multiple genes were upregulated in lungs of AdV-ACE2 transduced mice infected with SARS-CoV-2, such as several cytokines and chemokines, including TNF, IFN-γ, IL-10, IL-15, IL-6, CCL2, CXCL10, and platelet-derived growth factor subunit B (PDGFb), which was consistent with observations in COVID-19 patients ( Table 1; Huang et al., 2020). The effects of several candidate vaccines and antiviral therapeutics (polyinosinicpolycytidylic acid, remdesivir, 1B07 anti-S protein monoclonal antibody, etc.) have been evaluated in this animal model (Alsoussi et al., 2020;Hassan et al., 2020;. Venezuelan equine encephalitis replicon particles (VEEV-VRP) also can be used as an exogenous system for delivering hACE2 (Zhang Y.N. et al., 2020). After SARS-CoV-2 infection, this VEEV-VRP-hACE2 model developed interstitial pneumonia without obvious clinical symptoms, and the virus replication was limited to the lung ( Table 2). A human neutralizing antibody CB6 showed strong protective effects against the virus in this model (Zhang Y.N. et al., 2020).
However, delivery of hACE2 with viral vector-mediated delivery systems can cause extra lesions in lung of mice, as mild bronchial inflammation induced by AdV and small areas of alveolar septal thickening induced by VEEV-VRP (Hassan et al., 2020;Zhang Y.N. et al., 2020). It remains a problem that the lung disease caused by viral vector may be confused with the pneumonia caused by SARS-CoV-2. The expression and tissue distribution of hACE2 is not systemic in this kind of models, and may vary among different mice. Consequently, hACE2 mouse model using viral vector-mediated delivery systems is not ideal for simulating systemic SARS-CoV-2 infection ( Table 2).

Others
Severe acute respiratory syndrome coronavirus 2 must be handled in biosafety level three (BSL-3) facilities, which limits the development of urgently needed antiviral agents. Recombinant lentiviral viruses pseudotyped with SARS-CoV-2 S protein as the sole viral envelope protein has been demonstrated to be able to simulate SARS-CoV-2 virus infecting cells through S protein binding to hACE2 (Nie et al., 2020). Fan et al. (2018) reported successful infection of mice transgenic for MERS-CoV receptor using MERS-CoV S pseudovirus. hACE2 transgenic mice with pseudotyped SARS-CoV-2 infection have been applied in COVID-19 research Bai et al., 2021). Such pseudoviruses can be handled in BSL-2 facilities and would greatly facilitate studies targeting both SARS-CoV-2 S protein in the context of infection, as well as other SARS-CoV-2 proteins in pathogenesis. Interestingly, recombinant subunit 1 of SARS-CoV-2 S protein (S1SP), rather than the live virus, can induce alveolar and systemic inflammation in K18-hACE2 mice, while the intact S protein provokes minimal or no responses (Colunga Biancatelli et al., 2021). This protein-induced model provides another way to avoid the limitation of BSL-3 facilities for COVID-19. Hassert et al. (2020) utilized an mRNA-based transfection system for transiently delivering hACE2 into type I interferon receptor 1 deficient (Ifnar1) mice to study the adaptive immune response to SARS CoV-2. hACE2 mRNAs mainly target the lung and liver of mice, and a neutralizing antibody response was detected in those mice after SARS-CoV-2 infection (Hassert et al., 2020). The transfection efficiency of mRNA-ACE2 remains an issue in vivo according to the data presented (Hassert et al., 2020).

FERRET
Ferrets are naturally susceptible to SARS-CoV-2 infection. Infected ferrets showed mild clinical symptoms similar to human, including elevated body temperature, decreased activity, cough, and rhinorrhea Park et al., 2020;Shi et al., 2020).
The virus replicated efficiently in the upper respiratory tract of ferrets for up to 8 days . Viral RNA shedding was detectable up to 18 dpi in high-dose group (5 × 10 6 PFU) (Ryan et al., 2021). RNA could be detected in ferret's nasal wash, blood, nasal washes, urine, and fecal specimens, indicating diverse routes of shedding . Immunohistochemistry detected viral antigens in nasal turbinate, trachea, lung, and intestine , while histology analysis revealed severe lymphoplasmacytic perivasculitis, vasculitis, and mild peribronchitis in the lung ( Table 1; Shi et al., 2020). Compared with pathological changes in COVID-19 patients with pulmonary edema and diffuse alveolar damage, virus-induced damage to ferret lungs is milder (Xu Z. et al., 2020). The mild lung pathology in the highdose group ferrets was similar to the medium-dose group, but more extensive. No obvious lesions were observed in other tissue, except for multifocal inflammatory cell infiltration was observed in the portal areas of liver (Ryan et al., 2021). Ferrets are fully protected from acute lung pathology following re-challenge of the virus, which indicates naturally acquired immunity helps ferrets against reinfection (Patel et al., 2021;Ryan et al., 2021).
After direct contact with infected ferrets, all healthy naive ferrets showed moderate increases in body temperature and decreases in activity. Naive ferrets in indirect contact with infected ferrets through permeable partition did not show obvious symptoms, but some became positive for viral RNA in nasal washes and fecal specimens. It suggests that SARS-CoV-2 can be transmitted between ferrets through direct contact and airborne route, similar to transmission between humans Richard et al., 2020). These results indicate that the ferret model is excellent for studying SARS-CoV-2 transmission and mild infection ( Table 2).

MINK
Severe acute respiratory syndrome coronavirus 2 outbreaks occurred in mink (Neovison vison) farms in the Netherlands, with infected minks showing signs of respiratory disease, including watery nasal discharge and even severe respiratory distress ending in death (Oreshkova et al., 2020). The virus was suspected to be transmitted from infected farm workers and then spread among the minks (Oreshkova et al., 2020). Highly diverse mink SARS-CoV-2 variants, including Y453F, F486L, and N501T, are found and spill over to human (van Dorp et al., 2020;Oude Munnink et al., 2021). Escaped and wild minks infected with SARS-CoV-2 might become a virus reservoir, potentially constituting a pandemic threat (Koopmans, 2021).
After displaying clinical symptoms for 1 or 2 days, infected minks usually stopped eating and died on the next day . Histological examination revealed severe diffuse interstitial pneumonia with diffuse alveolar damage and formation of hyaline membranes in the lung Oreshkova et al., 2020). Viral RNA could be detected in the conchae, lungs, liver, distal large intestine, throat swabs, and rectal swabs (Oreshkova et al., 2020) and SARS-CoV-2 isolated from mink (MT396266) was highly similar to circulating human isolates ( Table 1; Sharun et al., 2020). Minks have transmitted SARS-CoV-2 to humans, and their high susceptibility to the virus suggest that they can easily become a natural reservoir (Enserink, 2020).
In laboratory settings, a SARS-CoV-2 strain isolated from human was used to infect minks intranasally . Infection caused severe pathological injury similar to COVID-19 in the respiratory tracts of minks, as autopsy revealed severe interstitial pneumonia with extensive and diffuse consolidation, pulmonary thrombosis, nasal mucosa damage in the olfactory region, and nasal cavities filled with mucinouspurulent secretion. Immunohistochemical analysis of lung lobes showed the widespread distribution of SARS-CoV-2 antigenpositive cells; and high levels of virus replication were detected in turbinates, soft palate, lung lobes, and other tissues ( Table 1; Shuai et al., 2020). The virus can be transmitted between minks via the air over more than 1 m distance (Kutter et al., 2021). Although minks infected in the laboratory did not die like infected farm minks, they lost 10-20% of body weight at around 8 dpi. In addition, a spike protein-based subunit vaccine candidate was evaluated using this mink model .

Rhesus Macaque (Macaca mulatta)
After inoculation with SARS-CoV-2 via a combination of intratracheal, intranasal, ocular, and oral routes, rhesus macaques showed changes in respiratory pattern (irregular, accelerated respiration), piloerection, and weight loss on 1 dpi (Table 1; Munster et al., 2020a). Pulmonary infiltrates were visible in lung radiographs of all animals, similar to what have been observed in patients with COVID-19. The symptoms could last 8-16 days and autopsy revealed pulmonary edema, pulmonary thrombosis, and interstitial pneumonia centered on terminal bronchioles (Munster et al., 2020a;Shan et al., 2020). Some rhesus monkeys generated neutralizing antibodies after infection, providing protection from secondary infection, which proved that humoral immunity produce a marked effect . Cytokine and chemokine levels in serum, such as IL-1ra, IL-6, IL-10, IL-15, MCP-1, macrophage inflammatory protein (MIP)-1β increased on 1 dpi, but minor decline was observed in transforming growth factor (TGF)-α on 3 dpi (Munster et al., 2020a). Virus loads were high in nasal swabs, throat swabs, and bronchoalveolar lavage fluids. Urogenital swabs remained negative in all animals, and only one rhesus monkey showed virus shedding from the rectum on 21 dpi. These observations indicate that the rhesus macaque infection model replicates moderate symptoms of COVID-19 patients. On the other hand, old macaques showed more severe diffuse interstitial pneumonia compared to young macaques and may thus be used for the study of more severe SARS-CoV-2 infection ( Table 2; Yu et al., 2020).

Cynomolgus Macaque (Macaca fascicularis)
In one study, cynomolgus macaques inoculated with SARS-CoV-2 displayed increased body temperature and weight loss (Lu et al., 2020). In another similar study, however, the inoculated animals showed no overt clinical symptoms, except for a serous nasal discharge in one aged animal on 14 dpi (Rockx et al., 2020). This difference may be related to different methods and doses of virus inoculation. Both studies reported diffuse interstitial pneumonia. Besides, Lu et al. (2020) also found inflammation in liver and heart, and mild lesions in mesenteric lymph nodes and kidney ( Table 1). By 14 dpi, there were virus-specific antibodies against the S and nucleocapsid proteins in the serum of all animals (Rockx et al., 2020). Viral shedding in upper respiratory tract of aged animals was prolonged to 21 dpi (Rockx et al., 2020). Compared with young adult animals, higher levels of viral RNA were detected in nasal swabs of aged animals. However, aged animals did not show more prominent lung infections or obvious clinical symptoms, which was different from rhesus macaques (Lu et al., 2020;Rockx et al., 2020).

African Green Monkey (Chlorocebus aethiops)
African green monkeys (AGMs) supported a higher level of viral replication than rhesus and cynomolgus macaques, for both SARS-CoV and SARS-CoV-2 (McAuliffe et al., 2004;Woolsey et al., 2021). After virus challenge, AGMs showed mild and various clinical symptoms, including decreased appetite, elevated body temperature, and even severe respiratory distress and death in two aged animals. Lung tissues developed extensive interstitial pneumonia with hyperemia and hemorrhage, and more severe lesions with ARDS was observed in two of four aged animals (Blair et al., 2021;Woolsey et al., 2021). Aged AGMs may become a lethal model to mimic serious COVID-19 manifestations, including ARDS. Mild liver inflammation and chronic bilateral glomerulonephritis was noted in extrapulmonary organs of infected AGMs. The changes of systemic cytokines in AGMs following infection is consistent with COVID-19 patients: IFNstimulated genes and IL-6 and IL-8 signaling upregulated (Woolsey et al., 2021). A study have evaluated the transmissibility and disease severity of SARS-CoV-2 B.1.1.7 variant in AGMs (Rosenke et al., 2021).

Common Marmoset (Callithrix jacchus)
Common marmoset, a New World monkey, has been used in research on a wide range of human diseases (Zühlke and Weinbauer, 2003;Adams et al., 2008). Compared with other nonhuman primates, it has smaller size, breeds well in captivity, and can be handled with ease. SARS-CoV-2 inoculated marmoset displayed no obvious clinical symptoms, with only one-third of animals showing slightly elevated body temperature ( Table 2; Lu et al., 2020). Viral RNA could be detected in the peripheral blood from 2 to 8 dpi, and in nasal, throat, and anal swabs during 2 weeks post viral inoculation. Slight infiltration of inflammatory cells into pulmonary septum and mild lesions in liver and spleen were observed in histopathological examination ( Table 1; Lu et al., 2020). Although no viral RNA was detected in tissue samples from the two tested marmosets, ultrastructural lesions in the lung and heart were observable under transmission electronic microscope (Lu et al., 2020). Additionally, virusspecific antibody was undetectable in the serum of marmosets during the experiment. These results indicate that, in contrast to cynomolgus and rhesus macaques, common marmoset is relatively resistant to SARS-CoV-2 infection and may not be a suitable animal model.

Susceptibility and Disease Severity
ACE2 is a crucial factor that limits the sensitivity of the host to SARS-CoV-2. We select 11 species currently being used as SARS-CoV-2 animal models, and analyze the amino acid sequence of their ACE2 to construct a phylogenetic tree (Figure 1). The amino acid residues shown in Figure 1 can form hydrogen bonds or salt bridges with S protein to stabilize the binding Liu et al., 2021). In primates, the 20 amino acid residues at the interface of human and the Old World monkeys, are exactly the same. But marmoset, a New World monkey, have lower residue conservation, which may explain its relatively resistance to SARS-CoV-2 infection. In addition, tree shrew are less susceptible to the virus .
The difference in ACE2 can partially explain the susceptibility of some species to viruses. Liu et al. (2021) found an interesting phenomenon: ACE2 of ferret and mink, which exhibit limited binding affinity with viral S1 protein, have the function of mediating the entry of SARS-CoV-2 in cell-level analysis. However, farm minks developed fatal infections, which constitutes a model of severe and potentially lethal infection of SARS-CoV-2 (Oreshkova et al., 2020). Therefore, the actual host-pathogen interaction is more complicated.
In summary, among the several animal models, three types of hamsters, ferret, mink, AGMs, rhesus, and cynomolgus macaque are highly susceptible to SARS-CoV-2, while marmoset, tree shrew, and wild-type mouse are relatively less susceptible.
Susceptible small animal models for COVID-19 are of great significance in evaluating antiviral agents and vaccines. Syrian hamster infected with high-dose virus and RAG2 KO hamsters developed lethal disease (Brocato et al., 2020;Tostanoski et al., 2020). Although wild-type mouse cannot be infected with SARS-CoV-2 in its natural state. Some mouse models (BALB/c mice infected with MA10, K18-hACE2, and HFH4-hACE2 mice) so far have reproduced severe respiratory disease that resemble pneumonia in severe COVID-19 in humans. In view of the time required for the development and production of transgenic mice, which usually takes from several months to years, approaches based on mouse-adapted or recombinant virus strains capable of infecting through mACE2, or expression of hACE2 in mouse lung cells are generally much quicker, and susceptible animals can be obtained in as fast as 2-3 weeks .

Lung Pathology
Typical pathological changes in the lung of COVID-19 patients include pulmonary edema with hyaline membrane formation, diffuse alveolar damage, type II pneumocyte hyperplasia, dilatation of capillary and occurrence of thrombosis, monocytes and macrophages within alveolar spaces, and interstitial mononuclear inflammatory infiltrates. Pulmonary fibrosis and consolidation in COVID-19 patients are usually milder compared to SARS patients, but mucus secretion tends to be more prominent in the former . The most common patterns of COVID-19 pneumonia on chest CT were ground-glass opacity and bilateral patchy shadowing .
Mild pulmonary abnormality of tree shrew failed to recapitulates the typical characteristics of COVID-19. On the other hand, hamsters are promising models in this aspect. Syrian hamster infected with SARS-CoV-2 showed moderate to lethal broncho-interstitial pneumonia and bilateral ground-glass opacity on CT (Table 1), similar to the pneumonia characteristics of COVID-19 (Imai et al., 2020;Tostanoski et al., 2020). The lung pathology of Chinese hamster was roughly similar to that of Syrian hamster, but the pneumonia was milder and prolonged . Notably, typical manifestations of ARDS, including severe acute diffuse alveolar damage and hyaline microthrombi, have been observed in infected Roborovski dwarf hamsters, providing the opportunity for studying SARS-CoV-2-related ARDS (Trimpert et al., 2020). Mild to severe interstitial pneumonia has been observed in different mouse models (Table 1). In contrast to K18-hACE2 transgenic mice, AdV-hACE2 mice develop lower titer viral replication and less obvious symptoms. Destroying the intact type I IFN response exacerbated lung pathology and clinical symptoms of AdV-hACE2 mice (Hassan et al., 2020;Rathnasinghe et al., 2020). Minks and ferrets, the closely related mustelids, displayed different patterns of viral replication . SARS-CoV-2 replicated efficiently in both the upper and lower respiratory tract of minks, but only in the upper respiratory tract of ferrets Shuai et al., 2020). Consequently, acute bronchiolitis in infected ferrets, is less similar to that of COVID-19 patients. Among all animals that have been used for modeling SARS-CoV-2 infection, three kinds of Old World monkeys are closest to human in terms of lung pathology, and common marmoset shows relatively milder lung damage.

Extrapulmonary Pathology
Some COVID-19 patients, especially those with severe infection, have extrapulmonary symptoms, such as kidney injury, cardiac complications, and neurologic manifestations (dizziness, impaired consciousness, taste impairment, smell impairment, etc.) (Chen N. et al., 2020;Mao et al., 2020;Zhou F. et al., 2020). Although the main target organ of SARS-CoV-2 is lung, the virus has a broad organotropism and can be detected in some extrapulmonary organs, including the kidneys, liver, heart, and brain, according to the result of autopsy (Puelles et al., 2020). Pathology results showed a few interstitial mononuclear inflammatory infiltrates in heart tissue and severe acute tubular necrosis in kidney tissue . In addition, lymphocytic panencephalitis, meningitis, ischemic damage, and microinfarcts were reported in postmortem brain of COVID-19 patients (von Weyhern et al., 2020;Song et al., 2021).
Although some animal models develop extrapulmonary manifestations of disease after infection as previously mentioned (Table 1), there is still a lack of an ideal model that can mimic the systemic response to SARS-CoV-2 infection in human.
Acute tubular necrosis, mild focal myocardial degeneration, and adrenal medulla atrophy could be observed in extrapulmonary tissues of Syrian hamsters . Syrian hamster has been used in the study of myocardial pathology in COVID-19 . SARS-CoV-2 infection of cardiomyocytes, increased CCL2 expression and macrophage infiltration were reported in their heart tissues, which was confirmed in autopsy samples of COVID-19 patients. This study provides direct evidence that SARS-CoV-2 infects cardiomyocytes in vivo. Therefore, Syrian hamsters may be applied in pathological research of systemic SARS-CoV-2 infection.
In addition, SARS-CoV-2 is proved neurotropic in some mouse models. After intranasal inoculation of the virus, 6-weekold K18-hACE2 mice showed some neurological symptoms, such as tremors and ataxic gait, and all animals died by 6 dpi (Kumari et al., 2021). Those mice also exhibited encephalitis features including production of cytokines and chemokines, leukocyte infiltration, and neuronal cell death (Kumari et al., 2021). SARS-CoV-2 could infect cells within the nasal turbinate, eye, and olfactory bulb, which related to smell impairment induced by infection of these cells in COVID-19 patients (Kumari et al., 2021). Besides, Song et al. (2021) observed the remodeling of brain vasculature in infected regions in this model. In conclusion, the severe neurological disease observed in K18-hACE2 mice is closely linked with the infection of nervous system in human patients, and this mouse model provide a valuable tool for studying the pathogenesis of infection in nervous system. The primary manifestation of clinical disease in these hACE2 transgenic mouse models was encephalitis, rather than that severe pneumonia has been reported in some hamster model, which limits them to systemic SARS-CoV-2 models (Tostanoski et al., 2020). Neuroinvasion also could be found in mice which expressed adeno-associated virus (AAV)-hACE2 expression in brain and then infected with SARS-CoV-2 intraventricularly. Those mice developed weight loss and death after infection (Song et al., 2021).

Host Immune Cellular Profile
The counts of white blood cells in the peripheral blood of COVID-19 patients are normal or decreased in the early stage of onset with the decrease in lymphocytes, and the progressive decrease in lymphocytes and increased inflammatory factors can be observed in severe patients .
According to the whole genome sequencing, the immune and nervous systems of tree shrew have high homology with those of human (Fan et al., 2013). In addition, some cytokines of tree shrew are similar to their human counterparts in terms of structure and function, such as IFN-λ3 and CXCL12 (Li et al., 2013;Chen et al., 2014). However, instead of lymphopenia often observed in human patients, tree shrews in the adult group increased white blood cells, lymphocytes, monocytes, and granulocytes but old group had a significant decrease in monocytes . The overall pattern of immune cellular profile is different from that in humans. Shan et al. (2020) noticed leukocytes and neutrophils decreased, but lymphocytes increased in the blood of infected rhesus macaques. In another experiment, leukocytosis, neutrophilia, monocytosis, and lymphopenia were observed at 1 dpi, and then lymphocytes and monocytes returned to baseline, but the neutrophils continued to decline by 5 dpi (Munster et al., 2020a). In conclusion, rhesus macaques infected with SARS-CoV-2 showed neutropenia, lymphocytopenia, and increased NLR, similar to that observed in severe COVID-19 patients (Cao, 2020). And the number of T cells and monocytes in cynomolgus macaques peaked at 2 or 4 dpi, then gradually decreased and reached the lowest at 10 or 12 dpi, consistent with transient lymphocytopenia in rhesus macaques (Lu et al., 2020).

CONCLUSION
A variety of animal models for COVID-19 have been developed. Most animal models have mild to moderate infections. Minks, AGMs, certain mouse, and hamster models with critical infections can be applied to the pathogenesis study of severe COVID-19. Animals with low ACE2-binding affinity, such as mice, can be made susceptible to SARS-CoV-2 through engineering with expression of human hACE2 (transgenic technology or viral vector-mediated delivery systems) or by using adapted/recombinant virus strains.
The age-dependent increase in disease can be observed in some animal models. Researchers can consider age and other additional risk factors of COVID-19 to be incorporated into models to mimic human comorbidities. In addition, animal models administered with pseudovirus or viral proteins (S1SP) which are not restricted by BSL-3 facilities can be established and used for research on COVID-19 pathogenesis.

AUTHOR CONTRIBUTIONS
SS and YX conceptualized the review. SS drafted the manuscript. YX, SS, ML, YY, NK, YS, DT, NL, FW, and JL revised the manuscript critically for important intellectual content and approved the version of the manuscript to be published.