Translating animal models of SARS-CoV-2 infection to vascular, neurological and gastrointestinal manifestations of COVID-19

The emergence of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) in December of 2019 marked the onset of an unprecedented worldwide pandemic that has had lasting impact. SARS-CoV-2 rapidly spread from its origins in the city of Wuhan, China, to become a global public health threat. A member of the beta coronavirus family, SARS-CoV-2 was initially associated with severe respiratory symptoms, including pneumonia, fever and acute respiratory distress syndrome (ARDS) (Martinez-Salazar et al., 2022; Pal et al., 2020). SARS-CoV-2 infects cells via the angiotensin-converting enzyme 2 (ACE2) receptor by using the mechanism described in Box 1. The primary driver of case fatality with COVID-19 is thought to be the pulmonary damage resultant from infection followed by an associated systemic cytokine storm (Box 2) and sepsis (Bao et al., 2020; Ketcham et al., 2021). However, it quickly became evident that the clinical presentation can be highly variable and even long-term, affecting not only the respiratory system but also the vascular, olfactory, neurological and gastrointestinal systems (Ostergaard, 2021).

The World Health Organization (WHO) declared ‘coronavirus disease 2019’ (COVID-19) a global pandemic in March of 2020 and, over the next three years, there would be over 7 million deaths worldwide attributed to the disease [see WHO, WHO COVID-19 dashboard: Number of COVID-19 cases reported to WHO (https://data.who.int/dashboards/covid19/cases?n=c, accessed 4 December 2023)]. Although the Public Health Emergency for COVID-19 was declared to be over in May 2023, the virus continues to evolve and appears to be transitioning towards endemicity, posing ongoing cyclical health risks worldwide. Reported hospital admissions and COVID-19 deaths in the USA have markedly declined compared to the early stages of the pandemic; however, the toll of COVID-19 on lives and long-term health consequences remains a sobering reality [see Centers for Disease Control and Prevention (CDC), National Center for Health Statistics: COVID Data Tracker (https://covid.cdc.gov/covid-data-tracker/#datatracker-home, accessed 22 January 2024)]. In addition to acute presentations, by January 2023, approximately one in ten Americans reported to experience post-acute sequelae of COVID-19 (PASC), also referred to as long COVID, with 27% of those affected noticing significant limitations in their day-to-day activities [see CDC (https://www.cdc.gov/nchs/covid19/pulse/long-covid.htm)].

While significant progress has been made in identifying treatment strategies for those with acute COVID-19, much of the literature is descriptive and gaps persist in understanding the mechanisms causing the highly variable disease presentations. Refining or expanding clinical treatment strategies and public health measures to combat this disease as it becomes endemic will require a deeper mechanistic understanding of the molecular disease processes. Adequate disease modeling is an important approach that may help to identify potential therapeutic targets, improving patient outcomes. Development of translational animal models of SARS-CoV-2 infection that accurately reflect certain aspects of human disease remains an important challenge for the field moving forward.

The major goals of this Review are to define some essential requirements of animal models for the study of vascular, neurological and gastrointestinal (GI) manifestations of COVID-19, and to assess the strengths and weaknesses of currently established models in these areas.

Experimental animal models are crucial for understanding disease mechanisms, and selecting the most appropriate species to effectively replicate the specific clinical disease aspects is important. Effective, translatable animal models should replicate human disease phenotypes, have molecular and physiological similarities to humans, be practical for research and, ultimately, facilitate the development of diagnostics, treatments and preventative strategies for human disease (Choudhary et al., 2022). While no single model meets all the listed criteria, the insights gained from any individual model species are vital for furthering our understanding of human disease. For understanding SARS-CoV-2, several animal models have been employed, including mice, ferrets, hamsters and non-human primates (NHPs), each of which offers unique advantages and insights into various facets of COVID-19 (de Melo et al., 2021; Kumari et al., 2021; Rutkai et al., 2022; Arce and Costoya, 2021; Becker et al., 2021) (Table 1).

Rodents, such as mice and rats, are frequently employed in translational studies due to their short life cycles, cost-effectiveness, physiological and genetic similarities to humans, and relative ease of genetic manipulation, making them valuable models of various human diseases, including COVID-19 (Bryda, 2013; Mouse Genome Sequencing Consortium, 2002). Standard inbred strains of mice, such as BALB/c and C57BL/6, initially posed challenges due to the inefficiency of SARS-CoV-2 binding to murine ACE2 receptors (Zhao et al., 2020). Infection of these mice with SARS-CoV-2 revealed no overt clinical manifestations, weight loss or mortality (Mohandas et al., 2020). Viral presence appeared contained to the lungs and was rapidly cleared (Mohandas et al., 2020). However, subsequent variants of concern (alpha, omicron, etc.) have shown evidence of binding to mouse ACE2 (Chen et al., 2022; Fumagalli et al., 2022). Two strategies have been employed to address this major limitation of murine models: 1) the generation of mice that transgenically express human ACE2 protein (Fumagalli et al., 2022) and, 2) the use of selectively adapted viral strains that have higher affinity for mouse ACE2 (Dinnon et al., 2020). Apart from mice, Syrian hamsters are also a commonly used rodent model for viral infections. These animals are naturally susceptible to SARS-CoV-2, eliminating the need for genetic engineering (de Melo et al., 2021).

Beyond rodents, ferrets are often used in respiratory disease research as their respiratory tract anatomy and physiology is comparable to humans (Belser et al., 2011). They have natural susceptibility to human viral pathogens, most notably influenza virus (van Riel et al., 2007). NHPs, such as rhesus and cynomolgus macaques, are also valuable for studying complex diseases and when testing potential vaccines or treatments due to their close genetic and physiological similarity to humans (Vallender and Miller, 2013). Despite limitations, such as specialized husbandry needs, longer lifespans and more ethical considerations, non-rodent models remain an indispensable aspect of pre-clinical research where rodent models fall short.

The varying natural susceptibility to SARS-CoV-2 of these animal models can be explained by differences in their ACE2 protein sequences compared to that of human ACE2 (Alexander et al., 2020). Species with ACE2 sequences more similar to the human one − especially in key conserved regions − are more likely to be susceptible to SARS-CoV-2 infection than those with less similar ACE2 (Alexander et al., 2020). Key amino acids (aa) of the ACE2 protein, and differences among different animal models and ACE2 in humans, are illustrated in Fig. 1.

In most patients, SARS-CoV-2 infection begins with the inhalation of virus-laden respiratory droplets. COVID-19 presents with a wide range of respiratory symptoms that vary in severity. While some individuals may experience mild symptoms or remain asymptomatic, others develop more severe respiratory distress. Common respiratory manifestations include cough, shortness of breath and chest discomfort (Cascella et al., 2023; Stokes et al., 2020). Pneumonia is a frequent complication and a significant contributor to respiratory distress in patients (Stokes et al., 2020; Cascella et al., 2023). Observed tissue changes are listed in Table 1. In severe cases, COVID-19 can lead to ARDS, characterized by profound hypoxemia (Box 2), difficulty breathing and bilateral lung infiltrates (Box 2) on chest imaging (Selickman et al., 2022).

Increasing evidence suggests that COVID-19 is more than just a respiratory disease, with some experts classifying it as a vascular disorder (Adu-Amankwaah et al., 2021; Liu et al., 2021; Siddiqi et al., 2021). Individuals with pre-existing vascular comorbidities, such as hypertension, coronary artery disease and diabetes, often experience more severe disease and worse outcomes when infected with SARS-CoV-2 (Geca et al., 2022; Huang et al., 2020; Szarpak et al., 2022). Pre-existing heart failure is associated with an almost 50% increased mortality, and patients with cardiovascular disease (CVD) and increased levels of troponin (Box 2) have been found to be associated with a 10% increase in case fatality ratio (CFR) compared with patients without such comorbidities (Bhatt et al., 2021; Dan et al., 2020). A growing body of literature recognizes vascular injury and resulting dysfunction as a crucial mechanism underlying COVID-19 clinical manifestations (Connors and Levy, 2020; Iba et al., 2020; Levi et al., 2020; Liu et al., 2021). This vascular involvement in COVID-19 pathogenesis underscores the importance of understanding and modeling these aspects of the disease.

Cardiovascular impacts of SARS-CoV-2 infection have been well-documented. A study conducted in Wuhan, China in early 2020 found that nearly 20% of hospitalized patients developed cardiac injury (Shi et al., 2020). These patients experienced more complications and had a higher mortality rate (51.2% vs 4.5%, P<0.001), even after accounting for age and pre-existing conditions (Shi et al., 2020). During the acute phase of infection, clinical reports have documented various common manifestations of cardiovascular injury (Table 1) (Adu-Amankwaah et al., 2021). The exact mechanisms remain unclear, but current reports suggest that endothelial infection play a central role. Endothelial infection may have multiple effects, including (i) disruption of the renin-angiotensin-aldosterone system (Cojocaru et al., 2023; Violi et al., 2020; Zipeto et al., 2020), which has been implicated as a potential mediator of the relationship between cardiovascular disease and COVID-19 severity (Coto et al., 2021); (ii) immune system activation (Giamarellos-Bourboulis et al., 2020); (iii) increased endothelial permeability, with resulting vascular leakage (Joffre et al., 2022) and, (iv) development of a pro-thrombotic environment (Fig. 2) (Adu-Amankwaah et al., 2021; Siddiqi et al., 2021; Nguyen et al., 2022). Endothelial injury can be further compounded by activation of toll-like receptors (TLRs) (Mantovani et al., 2023), resulting in increased production of reactive oxygen species (ROS) (Siddiqi et al., 2021; To et al., 2017). All of these contribute to symptoms such as ARDS, myocardial injury, thromboembolism and vasoplegia (Box 2) (Siddiqi et al., 2021; Adu-Amankwaah et al., 2021).

Beyond the acute phase of infection, COVID-19 survivors continue to face cardiovascular risks, including cerebrovascular disorders, heart disease and heart failure (Table 1) (Xie et al., 2022). Several studies have shown similarly increased rates of cardiovascular disease, post-discharge mortality and rehospitalization of initially hospitalized patients (Ayoubkhani et al., 2021; Daugherty et al., 2021). Patients hospitalized for COVID-19 infection who had not been admitted to the ICU had three times higher rates of cardiovascular disease, eight times higher rates of mortality and four times higher rates of rehospitalization compared with people not diagnosed with COVID-19 (Ayoubkhani et al., 2021). There appears to be a reciprocal relationship between CVD and COVID-19 infection by which CVD serves as a risk factor for and increases the severity of infection, and by which COVID-19 infection serves as a risk factor for and increases in severity of CVD.

The use of rodent models in studying the respiratory, pulmonary and vascular aspects of COVID-19 has provided valuable insights. The B6.Cg-Tg(K18-hACE2)2Prlmn/J (also known as K18-hACE2) mouse is one of the best-established transgenic mouse models for studying SARS-CoV-2 (Oladunni et al., 2020; Arce and Costoya, 2021; Yinda et al., 2021; Dong et al., 2022). This mouse line has originally been developed in 2006 as a model for studying the original SARS-CoV, a related coronavirus with >70% homology to SARS-CoV-2 (McCray et al., 2007). These mice express the human ACE2 gene (and, subsequently, the human ACE2 protein) under the control of the keratin 18 (KRT18; hereafter K18) promoter. This leads to expression in epithelial cells − including those of the airway − but may also enable expression in the central nervous system (CNS), GI system and kidneys (Laboratory, 2023; Carossino et al., 2022; Rabbani et al., 2022). Following intranasal challenge with SARS-CoV-2, these mice developed severe pneumonia characterized by manifestations detailed in Table 1 (Oladunni et al., 2020; Arce and Costoya, 2021; Yinda et al., 2021). This mouse model exhibited elevated levels of local and systemic chemokines (Box 2) and cytokines (Box 2), similar to the cytokine storm of human patients with COVID-19 (Ye et al., 2020; Wang and Ma, 2008; Parsons et al., 2005). K18-hACE2 mice are considered to be an ACE2-overexpression model and infection of these mice with SARS-CoV-2 can be highly lethal, making them an attractive model for acute severe disease (Yinda et al., 2021). There also appear to be minimal acute cardiac or vascular disease complications, probably due to the lack of K18 promoter-driven hACE2 expression in the heart and endothelium, but long-term effects merit further study (Yinda et al., 2021).

Another transgenic model, the Hfh4-ACE2 mouse, was developed to study W1V1-CoV, a SARS-like virus (Dinnon et al., 2020; Menachery et al., 2016; Jiang et al., 2020). Under the control of the forkhead box J1 (FOXJ1; hereafter HFH4/FOXJ1) promoter, these mice express human ACE2 primarily in lung ciliated epithelial cells, but expression has also been noted in the brain, liver, kidney and GI tract (Menachery et al., 2016). When infected with SARS-CoV-2, these mice display clinical signs in two distinct clusters, i.e. <50% showed significant weight loss, noticeable respiratory distress and neurological symptoms that progressed to lethality, whereas the remaining mice stayed asymptomatic and survived acute challenge (Dinnon et al., 2020; Jiang et al., 2020). Edema and necrosis were observed in some cardiomyocytes, while whole-body plethysmography (Box 2) indicated normal respiratory function (Dinnon et al., 2020; Jiang et al., 2020), suggesting − as it does not seem to replicate human disease − that this mouse model is less suitable for studying respiratory manifestations.

In contrast, the mouse-adapted strain of SARS-CoV-2 (SARS-CoV-2 MA10) involved passaging SARS-CoV-2 ten times in BALB/cAnNHsd (BALB/c) mice to enable infection of mice that express the native murine ACE2 receptor (Dinnon et al., 2020; Leist et al., 2020). This approach has shown promise in modeling human-like disease presentations in certain mouse strains, such as BALB/c (Dinnon et al., 2020; Leist et al., 2020). Young mice showed little to no overt clinical signs of infection, but viral replication was supported in lungs, and plethysmography showed changes indicative of airway obstruction and altered exhalation flow rate (Dinnon et al., 2020; Leist et al., 2020). Older (1-year-old) mice showed increased mortality, localized changes in cytokine and chemokine levels, and a more severe disease presentation (Dinnon et al., 2020; Leist et al., 2020). Infection of young (10-week-old) mice at higher challenge doses also led to lethality (Leist et al., 2020). Histological changes were consistent with those observed for COVID-19 in humans and included diffuse alveolar damage, exfoliated cells in small airways, fibrin deposition, loss of surfactant, inflammatory infiltrates, peribronchiolar lymphocytic hyperplasia (Box 2), vascular congestion (Box 2) and bronchial epithelial damage (Dinnon et al., 2020; Leist et al., 2020). In contrast, infection of C57BL/6 mice demonstrated lower clinical and histological scores, and reduced viral replication, as well as quicker restoration of lung function (Leist et al., 2020). A limitation in this model is that human variants of concern (VOC) often involve mutations of the spike protein, rendering a mouse-adapted virus less suitable for studying viral evolution, transmission and associated varying disease manifestations.

In wild-type Syrian hamsters, infection led to viral replication in both the upper and lower respiratory tracts, with peak viral loads at ∼3 days post challenge (Choudhary et al., 2022; Becker et al., 2021; Osterrieder et al., 2020; Imai et al., 2020). They exhibited lung abnormalities including inflammatory cells in the airways, alveoli and interstitial spaces, along with epithelial hypertrophy, with vascular lesions and cardiac injury additionally noted (Table 1) (Francis et al., 2021; Rizvi et al., 2022; Xue et al., 2022). These hamsters showed elevated levels of various molecular markers of cardiac disease, including cholesterol and triglycerides (Table 1) (Rizvi et al., 2022). This animal model also showed age-dependent differences, with younger hamsters experiencing a more robust immune response, reduced clinical severity and faster recovery (Osterrieder et al., 2020).

Although wild-type Syrian hamsters are more commonly used in SARS-CoV-2 research, COVID-19 modeling has also included Roborovski dwarf hamsters and transgenic hACE2-expressing hamsters. Following infection, dwarf hamsters developed severe clinical signs and evidence of diffuse alveolar damage and hyaline microthrombi (Box 2) in the lungs (Trimpert et al., 2020). K18-hACE2 hamsters, expressing human ACE2 under the control of the same KRT18 promoter as the K18-hACE2 mice, were susceptible to severe lethal disease (Golden et al., 2022). Weight loss began as early as day 2 post challenge and a large proportion succumbed to disease by day 5 (Golden et al., 2022). The infection resulted in mild pulmonary lesions but severe nasal cavity and CNS lesions, which will be discussed below.

The natural susceptibility of some hamster models to SARS-CoV-2 make them appealing models. One critical limitation to the use of hamsters compared with murine models, is the limited availability of research tools like sequencing databases, antibodies, primers, and other hamster-specific cellular and molecular probes (Miao et al., 2019). This may limit the breadth of studies that can be conducted with hamster models (Table 1).

Ferrets, analogous to hamsters, have ACE2 receptors that are naturally susceptible to SARS-CoV-2 binding, although the binding affinity is considered low, especially in the lower respiratory tract (Everett et al., 2021). Infection of ferrets with SARS-CoV-2 primarily impacts the respiratory tract, and both intranasal and intratracheal exposure results in productive infections (Everett et al., 2021; van de Ven et al., 2021). The highest viral loads are typically found in the upper respiratory tract, especially in the nasal cavity, supporting the idea of this as the primary site of viral replication in ferrets (Au et al., 2022; Ciurkiewicz et al., 2022). Meanwhile, in humans, SARS-CoV-2 replicates in the upper respiratory tract (nose, throat, and sinuses) as well as the lower respiratory tract (lungs, and airways) (V'Kovski et al., 2021). Infectious virus can be isolated from nasal samples 3-7 days post infection (dpi) onwards, and upper respiratory tract viral loads peak between 4 and 7 dpi (Everett et al., 2021; Au et al., 2022). Importantly, neither infectious virus nor viral antigen has been observed in the lungs (Au et al., 2022; Everett et al., 2021; Ciurkiewicz et al., 2022). This highlights the utility of ferrets for upper respiratory tract infection studies but limits their usefulness as a model of human lower respiratory tract infection. Notably, some strains of SARS-CoV-2 (such as the VOC omicron) have been shown to produce greater clinical manifestations within the upper respiratory tract, for which ferrets may be a useful model (Nori and Ghani Zghair, 2022).

Compared with small-animal models, NHPs present unique differences that can make them particularly useful as they offer the closest genetic, physiological, anatomical and developmental similarities to humans (Vallender and Miller, 2013). Several NHP species, including rhesus macaques, cynomolgus macaques and African green monkeys, have been utilized in previous research with related coronaviruses SARS-CoV and Middle East respiratory syndrome (MERS) (Rowe et al., 2004; Lawler et al., 2006; Yao et al., 2014; Kuiken et al., 2003; Qin et al., 2005; McAuliffe et al., 2004; de Wit et al., 2013).

Following SARS-CoV-2 challenge, rhesus macaques showed minimal clinical presentation, with some studies reporting mildly elevated scores while others noted no clinical signs of disease (Choudhary et al., 2022; Zheng et al., 2020; Shan et al., 2020; Speranza et al., 2022). Of those animals that did show clinical manifestations of disease, the latter returned to baseline within two weeks of challenge. However, viral RNA was detectable in the lungs and by nasopharyngeal swabs (Choudhary et al., 2022; Zheng et al., 2020; Shan et al., 2020; Speranza et al., 2022). Pneumonia presentation was variable, generally noted in the inferior lobes, with edema, hemorrhage, necrosis, alveolar wall thickening, endothelial damage, fibrosis and immune cell infiltration (Choudhary et al., 2022; Shan et al., 2020; Speranza et al., 2022). There was also evidence of myocarditis (Box 2), microthrombi, increased coagulation, cardiomyocyte disarray and necrosis following viral challenge (Feng et al., 2021; Rabbani et al., 2022). As with humans, these macaques showed age-related differences in disease presentation, with older animals exhibiting higher clinical scores, slower recovery, increased pulmonary infiltrates, and elevated levels of various cytokines (Speranza et al., 2022).

Studies using cynomolgus macaques showed similar results, with minimal clinical presentations in young, previously healthy animals (Urano et al., 2021; Goncalves et al., 2021; Salguero et al., 2021; Boszormenyi et al., 2021). In one study, pneumonia was confirmed in all animals by computed tomography despite the lack of clinical signs (Urano et al., 2021). In another, it was noted that, while there was no gross pathology, histologically there were changes in all animals (Salguero et al., 2021). These changes included alveolar necrosis, thickened alveolar walls, degeneration and sloughing of bronchiolar epithelium, and hyperemia in the caudal pulmonary lobes (Urano et al., 2021; Salguero et al., 2021; Boszormenyi et al., 2021). As with rhesus macaques, there were age-related changes, with elderly animals (23-30 years) having elevated clinical scores, longer periods of viral RNA detection, longer periods of pneumonia (including recurrence) and more-diffuse viral detection (including bronchi, lymph nodes, heart, liver and kidney) (Urano et al., 2021). Overall, rhesus and cynomolgus macaques exhibited similar infection kinetics, effects on organ systems and age-related differences of disease presentation that model that of mild COVID-19 disease in humans.

Although less common, African green monkeys (AGMs) have also been assessed as a potential SARS-CoV-2 model. Similar to both macaque species, AGMs showed mild clinical disease in young, previously healthy animals (Woolsey et al., 2021; Blair et al., 2021; Cross et al., 2020; Hartman et al., 2020). During the acute stage of infection, there were signs of systemic inflammation, with a transient decrease in lymphocytes and platelets, and an increase in neutrophils (Table 1) (Hartman et al., 2020; Woolsey et al., 2021). Histological investigations showed lung damage (detailed in Table 1) (Woolsey et al., 2021; Cross et al., 2020; Hartman et al., 2020). Age-related differences included development of ARDS leading to fatality (Blair et al., 2021). Lethal cases showed respiratory distress and evidence of systemic inflammatory response syndrome (SIRS) (Table 1) (Blair et al., 2021).

Overall, these three NHP models offer valuable insights into COVID-19 pathogenesis, immune responses and age-related differences. They closely resemble human disease presentation and are particularly useful for studying severe cases, cytokine profiles, cardiac involvement and vascular aspects of the disease. However, use of NHPs in research can be challenging due to the high cost of care, long breeding cycles, long life span and ethical concerns.

The prevalence of neurological symptoms associated with SARS-CoV-2 infection varies widely among reports, ranging from 36-82% (Ellul et al., 2020; Mao et al., 2020a; Liotta et al., 2020). Common neurological symptoms associated with SARS-CoV-2 infection include anosmia (loss of smell), dysgeusia (loss of taste), fatigue, myalgia (muscle pain) and headache (García-Azorín et al., 2020; Liotta et al., 2020; Merkler et al., 2020; Varatharaj et al., 2020; Yaghi et al., 2020; Lou et al., 2021; Oosthuizen et al., 2021). Various factors, including age, sex, pre-existing neurological disorders, comorbidities and the severity of COVID-19, influence the likelihood of experiencing neurological symptoms (García-Azorín et al., 2020; Liotta et al., 2020; Mao et al., 2020a).

The pathophysiology of neurological symptoms in COVID-19 remains complex and only partly understood. Autopsy studies have provided some insights into potential mechanisms (von Weyhern et al., 2020; Solomon et al., 2020). Lesions associated with acute hypoxia were common in the cerebrum and cerebellum, but it was noted that these lesions might be normal findings within an autopsied brain or a result of systemic acute hypoxia due to severe pulmonary disease (von Weyhern et al., 2020; Solomon et al., 2020). One study found only evidence of hypoxic changes (Solomon et al., 2020), while another observed encephalitis (Box 2), meningitis and neuronal cell death (von Weyhern et al., 2020). While there are conflicting reports as to whether SARS-CoV-2 is neuroinvasive, immunohistochemistry of the viral spike glycoprotein has shown positive results in cortical neurons (Song et al., 2021). The variability in neuropathological findings suggests that multiple factors, including host response and disease severity, contribute to the diverse neurological manifestations seen in COVID-19 patients.

Expression of ACE2 protein in neuronal cells is thought to be necessary for viral neuroinvasion, although secondary mechanisms of systemic spread, such as extracellular vesicle formation could also contribute (Tahyra et al., 2022; Wang et al., 2020). Blocking the ACE2 receptor in human brain organoids resulted in a significant inhibition of SARS-CoV-2 infection (Song et al., 2021; Fumagalli et al., 2022). This might explain why CNS infection in humans is relatively rare, but relatively common in transgenic mouse models that overexpress hACE2 (McCray et al., 2007; Dinnon et al., 2020; Kumari et al., 2021). Due to the artificial K18 or Hfh4-driven overexpression, the expression pattern of hACE2 in these mice does not mimic that of natural expression in humans; therefore, results obtained from studies involving these animal models should be interpreted with caution (Dinnon et al., 2020; Renn et al., 2021; Amruta et al., 2023). Although it is likely that the mechanism of fatal encephalitis is similar between these animal models, here we will discuss only the K18-hACE2 mouse, as there is a relative paucity of information about the Hfh4-hACE2 model (Dinnon et al., 2020; Kumari et al., 2021).

Following intranasal SARS-CoV-2 inoculation, a high proportion of K18-hACE2 mice develop dose-dependent, fatal, non-suppurative (Box 2) meningoencephalitis (Vidal et al., 2022; Fumagalli et al., 2022; Zheng et al., 2021). However, studies have shown that these mice also develop anosmia and other neurological symptoms, including seizures, tremors, proprioceptive defects and abnormal gait, which underscores their continued potential as a model for neurological COVID-19 (Ye et al., 2021; Zheng et al., 2021; Fumagalli et al., 2022; Kumari et al., 2021; Carossino et al., 2022).

Infection of K18-hACE2 mice was associated with high titers of viral RNA and infectious viral particles in the CNS (Fumagalli et al., 2022; Vidal et al., 2022). Histological examinations showed diffuse positive staining for SARS-CoV-2 proteins throughout the cerebrum, with limited immunopositivity in the cerebellum (Song et al., 2021; Fumagalli et al., 2022). Neurons themselves appeared to be the primary target, while astrocytes (Box 2) and microglia (Box 2) were associated with viral positivity in some but not all studies (Fumagalli et al., 2022; Song et al., 2021; Golden et al., 2020; Carossino et al., 2022). Evidence suggested that intracerebral spread occurred via neuronal synapses (Box 2) (Vidal et al., 2022). While epithelial cells were susceptible to viral infection, no viral protein was observed in the choroid plexus (Box 2) epithelium, suggesting that neuroinvasion was not occurring via the cerebrospinal fluid (CSF) (Ng et al., 2023; Vidal et al., 2022; Song et al., 2021).

SARS-CoV-2 encephalitis in K18-hACE2 mice was characterized by perivascular cuffing (Box 2), neuronal necrosis (Letsinger et al., 2023) and immune cell infiltration of the brain (Fumagalli et al., 2022; Vidal et al., 2022). Furthermore, many studies found evidence of activated microglial cells and astrocytes (Fumagalli et al., 2022; Vidal et al., 2022; Letsinger et al., 2023). Cerebral vasculitis (Box 2) and the resulting fibrin thrombi were associated with significant disruption and remodeling of the brain vasculature (Vidal et al., 2022; Letsinger et al., 2023; Song et al., 2021), which might model the hypoxic lesions observed in human COVID-19 patients; however, this might be a weak association as no ischemic necrosis was noted (Vidal et al., 2022). Analogous to human patients, viral infection of mouse neurons was associated with vacuolar degeneration (Box 2), neuronal necrosis, satellitosis (Box 2) and neurophagia (Box 2) (Vidal et al., 2022; Letsinger et al., 2023). Altogether, the current evidence suggests that K18-hACE2 mice acquire diffuse brain lesions with high viral loads, limiting their translatability.

The neuroinvasive mechanism of SARS-CoV-2 in K18-hACE2 mice remains unclear and might depend on the route of infection (Bleau et al., 2015; Li et al., 2020; Bilinska et al., 2021; Meinhardt et al., 2021; Patrì et al., 2021; Rhea et al., 2021; Bauer et al., 2022; Fumagalli et al., 2022; Fontes-Dantas et al., 2023). Viral infection of olfactory epithelium has been observed to progress to neuronal necrosis in the olfactory bulb, leading to diffuse viral encephalitis. This suggests that retrograde viral migration in olfactory nerve axons is the route of viral neuroinvasion (Ye et al., 2021; Vidal et al., 2022; Letsinger et al., 2023). Notably, viral encephalitis was circumvented in K18-hACE2 mice inoculated via aerosolized virus (Fumagalli et al., 2022). Additionally, viral protein, viral RNA, infectious particles and infection-associated lesions were not detected in the brain of aerosol challenged mice (Fumagalli et al., 2022). The absence of viral RNA in the serum of both aerosol- and intranasal-challenged mice (Winkler et al., 2020; Kumari et al., 2021; Zheng et al., 2021; Fumagalli et al., 2022) suggests that viral translocation across the blood−brain barrier (BBB) does not play a major role in neuroinvasion in the K18-hACE2 mouse, although the possibility cannot be completely excluded (Li et al., 2020; Meinhardt et al., 2021; Vidal et al., 2022; Fumagalli et al., 2022). Moreover, the differences in neuroinvasion between intranasal inoculation and aerosolization might reflect a bolus or dose phenomenon, with aerosolization representing a more natural and translatable route of infection and subsequent disease progression.

In multiple different inbred strains of mice, intranasal challenge of SARS-CoV-2 MA10 led to a decrease in markers of BBB integrity (Amruta et al., 2023). Specifically, in BALB/c mice, there was an increase in microglial cell activation within cortical regions compared to those in unchallenged control mice (Amruta et al., 2023). BALB/c mice also had increased perivascular cuffing; however, fatal encephalitis did not develop and viral proteins were not detected in the brain of any observed mice (Amruta et al., 2023). MA10 was capable of inducing neuroinflammation and neuropathogenesis without causing diffuse CNS infection (Amruta et al., 2023), making it a more suitable model for acute neuro-COVID-19 compared to K18-hACE2 and Hfh4-hACE2 mice. Additionally, MA10 infection in mice deficient in caveolin-1, a protein essential for BBB permeability, was found to result in increased BBB permeability, disruption of BBB tight junctions, and deficits in learning and memory, highlighting the versatility of MA10 for use in mice of different genetic backgrounds (Trevino et al., 2024).

Other mouse models have been developed by using exposure to different viral variants and even components of the virus itself. For instance, infusion of SARS-CoV-2 spike protein into the brains of outbred mice induced late cognitive impairment, dysfunction and neuroinflammation (Fontes-Dantas et al., 2023). Additionally, infection with the SARS-CoV-2 Beta variant in C57BL/6 mice led to CNS infiltration of monocytes and microglial cell activation (Vanderheiden et al., 2024).

In hamsters, SARS-CoV-2 primarily manifested as a respiratory disease, with limited reports of neurological symptoms (Luan et al., 2020; Monchatre-Leroy et al., 2021; Käufer et al., 2022). However, some studies noted transient anosmia and ageusia − akin to clinical human symptoms − in infected hamsters (Xydakis et al., 2020; Golden et al., 2022; Carpenter et al., 2023; Bryche et al., 2020; de Melo et al., 2021; Käufer et al., 2022). This might be due to neurological deficits, or to damage to the olfactory epithelium in the nasal turbinates and associated loss of cilia necessary for odor detection (de Melo et al., 2021). While no viral protein was found in Syrian hamster brains by immunohistochemistry, isolated brain homogenates at day 3 post challenge showed infectious viral particles that had cleared by day 6, suggesting a transient and non-productive CNS infection (Bryche et al., 2020; Monchatre-Leroy et al., 2021; Käufer et al., 2022). Viral RNA was found in various brain regions, including the olfactory bulb, brainstem, cerebral cortex and cerebellum, as well as in the serum (Monchatre-Leroy et al., 2021; de Melo et al., 2021). Notably, one study observed early accumulation of proteins associated with neurodegenerative disorders in cortical neurons, but no evident neurodegeneration (Käufer et al., 2022). This was associated with microgliosis in the olfactory bulb following viral clearance, indicating ongoing neuroinflammatory processes. However, viral antigens have not been found in the brain, suggesting that diffuse infection CNS and COVID-19 disease does not occur in this hamster model (Bryche et al., 2020; de Melo et al., 2021).

Roborovski dwarf hamsters seemed similarly resistant to CNS infection (Trimpert et al., 2020). No viral RNA signals were detected in the brain of infected hamsters but blood leukocytes observed in cerebral vessels suggested that, in this model, viral dissemination to the brain may occur through the vasculature (Trimpert et al., 2020). Hamster cerebellar organotypic slice cultures were used to examine SARS-CoV-2 infection in the absence of the BBB, and viral replication was limited to 1 dpi, with a subsequent decline in viral RNA (Lamoureux et al., 2022). Likewise, there was no evidence of cell-to-cell transmission of the virus or neuronal infection, unlike what was observed during human autopsies (Stein et al., 2022).

In contrast, transgenic K18-hACE2 hamsters developed severe lesions in both nasal cavity and CNS in response to infection with SARS-CoV-2 (Golden et al., 2022). The nasal cavity showed extensive damage and inflammation of the olfactory epithelium; widespread brain disease included multifocal gliosis, meningitis, perivascular cuffing and neuronal necrosis (Golden et al., 2022). Viral RNA was detected in the olfactory epithelium, cerebral cortex and cerebellum at day 5 post challenge and persisted at a reduced level until 15 dpi (Golden et al., 2022). The presence of viral antigen in neurons and microglial cells confirmed true CNS infection in this model (Golden et al., 2022). These findings suggest that K18-hACE2 hamsters, like K18-hACE2 mice, are highly susceptible to CNS-lesion-associated disease (Golden et al., 2022). Overall, since widespread expression of human ACE2 is likely to make it susceptible to viral infection of the CNS, the K18-hACE2 hamster might be a useful model to further assess certain aspects of neurological COVID-19, such as anosmia and ageusia as well as mechanisms of viral neuroinvasion.

Ferrets have been shown to comprise viral RNA present within various brain regions, including the olfactory bulb, cerebrum and cerebellum following both intranasal and intratracheal challenge, although neuroinvasive processes appear to be more efficient following intranasal challenge (van de Ven et al., 2021). This suggests viral dissemination via the circulatory system and subsequent brain seeding, which is in alignment to some human studies (Jacobs et al., 2022; Stein et al., 2022). In some ferret studies, viral RNA in the brain was detected as early as day 3 and until 7 dpi (Schlottau et al., 2020; Everett et al., 2021; Au et al., 2022; Ciurkiewicz et al., 2022). Despite this, there have been no reports of ferrets displaying any evidence of SARS-CoV-2-associated neurological signs or neuropathy (Schlottau et al., 2020; Everett et al., 2021; Monchatre-Leroy et al., 2021; van de Ven et al., 2021; Au et al., 2022; Ciurkiewicz et al., 2022). While ferrets may provide insight into the neuroinvasive mechanisms of SARS-CoV-2 and viral transmission, they do not appear to be a suitable model for the neurological disease manifestations of COVID-19 (Kutter et al., 2021; Kim et al., 2020; Richard et al., 2020).

In the brain of NHPs, significant lesions have been observed post viral infection (Choudhary et al., 2022; Philippens et al., 2022; Rutkai et al., 2022). Multiple inoculation methods, including standard intranasal and aerosol methods as well as intracranial exposure, have been used (Jiao et al., 2021; Beckman et al., 2022; Choudhary et al., 2022; Philippens et al., 2022; Rutkai et al., 2022); however, to date, no significant differences in brain pathology have been noted based on route of infection (Rutkai et al., 2022).

Neural pathology has primarily been observed in rhesus macaques (Rutkai et al., 2022); however lesions have also been observed in cynomolgus macaques (Philippens et al., 2022) and African green monkeys (Rutkai et al., 2022). One study found that basal ganglia, brainstem and cerebellum exhibit most significant injury and pathology among infected animals (Rutkai et al., 2022). Observations of multifocal acute microhemorrhages, primarily within these regions, were common but not always associated with ischemic injury, and often lacked evidence of thrombi or vascular injury (Rutkai et al., 2022). Neuronal changes were most often found in the cerebellum and brainstem, with marked neuronal and surrounding cell injuries, possibly due to hypoxia in these brain regions, as indicated by markers in the vasculature of the brain stem and basal ganglia (Rutkai et al., 2022). Neuronal cell death was also noted in the entorhinal cortex (Box 2), hippocampus, thalamus and midbrain (Jiao et al., 2021). Evidence of microglial activation was observed in various brain regions, including the pituitary gland, olfactory bulb and cerebellum (Philippens et al., 2022). In aged rhesus macaques, activated microglial cells were associated with neuronal synaptic engulfment and myelin degradation, indicating a potential link between SARS-CoV-2-associated neuroinflammation and neurodegenerative processes (Beckman et al., 2022). There was also evidence of astrocyte activation (Rutkai et al., 2022; Beckman et al., 2022), glial cell hyperplasia (Jiao et al., 2021) and elevated brain cytokines, indicating further neuroinflammatory processes (Jiao et al., 2021).

There is limited evidence for vascular dissemination being the pathway of neuroinvasion in NHPs; however, some studies observed perivascular cuffing without significant parenchymal lymphocyte infiltration (Choudhary et al., 2022; Rutkai et al., 2022), while another study found evidence of T-cell infiltration of the pituitary gland and multiple parts of the brain parenchyma ∼35 dpi (Philippens et al., 2022). This suggests that translocation of immune cells from the peripheral blood supply to the brain is possible in response to infection (Philippens et al., 2022), but whether the virus can move similarly has not been shown. In contrast, two studies support retrograde axonal transport via the olfactory nerve as the primary mechanism of neuroinvasion (Jiao et al., 2021; Beckman et al., 2022). In a study of rhesus macaques, intranasal infection led to detection of viral RNA and antigen in the nasal mucosa, olfactory tract and CSF as early as 1 dpi (Jiao et al., 2021). Viral RNA and antigen were observed migrating into the olfactory trigone (Box 2) by 4 dpi and into the entorhinal cortex by 7 dpi (Jiao et al., 2021).

While there is evidence for neuroinvasion in NHPs, viral replication may not be well supported (Jiao et al., 2021). Detection of viral protein in NHP brains largely appears to be limited to vascular cells, but some studies have found viral proteins in areas of inflammation associated with neurons, glial cells and immune cells (Rutkai et al., 2022; Philippens et al., 2022; Choudhary et al., 2022; Beckman et al., 2022). These differences may have been due to the age of the animals in the study, as older rhesus macaques displayed a higher level of viral burden in neurons, astrocytes and microglial cells when compared to their younger counterparts (Beckman et al., 2022). Analyses of viral RNA in the CNS were variable, with some studies finding scant viral RNA in a small number of animals in multiple regions in the brain (Rutkai et al., 2022; Philippens et al., 2022), while others found none (Choudhary et al., 2022; Jiao et al., 2021). None of these studies observed cerebral sub-genomic RNA, which is indicative of viral replication (Jiao et al., 2021; Choudhary et al., 2022; Philippens et al., 2022; Rutkai et al., 2022). Altogether, NHPs have exhibited neuropathology associated with SARS-CoV-2 infection that make them suitable models for acute neurological COVID-19 studies.

Gastrointestinal (GI) symptoms, including symptoms of diarrhea, nausea or vomiting, loss of appetite and abdominal pain have been observed in a subset of patients (de Oliveira et al., 2020; Jin et al., 2020; Mao et al., 2020b; Pan et al., 2020; Sultan et al., 2020; Carnevale et al., 2021; Din et al., 2021). Several mechanisms have been proposed to explain GI involvement in COVID-19, including direct viral infection of GI epithelium, secondary damage due to a cytokine storm and gut microbial dysbiosis (Lamers et al., 2020; Lehmann et al., 2021; Serban et al., 2021; Yeoh et al., 2021; Zuo et al., 2020).

Viral RNA detected in patient fecal samples suggests that the intestine may be an extra-pulmonary site of SARS-CoV-2 replication and that fecal-oral transmission may be possible (Chen et al., 2020; Zuo et al., 2020). Small-intestinal epithelial cells express the ACE2 receptor and are susceptible to viral replication (Shang et al., 2020; Xiao et al., 2020; Zhang et al., 2020; Patankar et al., 2021). The disruption of normal cellular function or death of enterocytes due to viral infection is thought to be one of the causes of GI symptoms observed in some patients (Lehmann et al., 2021). Additionally, a cytokine-storm-induced prothrombotic state can also lead to microthrombi and ischemic infarcts in multiple body systems, including the GI tract (Savla et al., 2021; Serban et al., 2021; Wu et al., 2022). Mesenteric thrombosis and ischemia are rare COVID-19 complications that can significantly increase morbidity and mortality (Serban et al., 2021; Wu et al., 2022).

SARS-CoV-2 has been shown to significantly alter the normal gut microbiome, resulting in a loss of bacterial richness and diversity (Gu et al., 2020; Yeoh et al., 2021). This dysbiosis increases the likelihood of illness from opportunistic pathogens, and has downstream effects on the immune system and the CNS (Fig. 3) (Li et al., 2019; Zuo et al., 2020; Yeoh et al., 2021; Wiertsema et al., 2021). The bidirectional interaction that occurs between the gut and the CNS, often referred to as the gut−brain axis, is primarily mediated via gut microbes, and the hormonal, endocrine and circadian signals relayed to the gut from the brain (Carabotti et al., 2015; Silva et al., 2020; Parker et al., 2020). Therefore, any disruption to normal gut microbiota can have downstream effects on the CNS (Fig. 3B). In addition, dysbiosis increases patient susceptibility to sepsis, hospital-acquired infections and organ failure, and has been associated with severe illness in hospitalized patients (McDonald et al., 2016).

One possible proposed mechanism of SARS-CoV-2-induced gut dysbiosis is virus-associated intestinal inflammation; however, it is still unclear whether dysbiosis leads to inflammation, inflammation leads to dysbiosis, or if there exists a cyclical relationship between the two (Clerbaux et al., 2022). Another proposed mechanism involves the dysregulation of ACE2 due to SARS-CoV-2 binding in the gut (Clerbaux et al., 2022). Regardless of mechanism, a loss of commensal symbionts known to have immunomodulatory effects, such as Faecalibacterium prausnitzii, Eubacterium rectale and Bifidobacteria, as well as enrichment of opportunistic pathogens was observed in human patients. This dysbiosis persisted even after patients had been medically cleared and recovered from COVID (Zuo et al., 2020). This type of dysbiosis is associated in other diseases with increased severity, suggesting that the gut microbiome plays a role by modulating the host immune response (Yeoh et al., 2021).

Few mouse studies have assessed the effects of SARS-CoV-2 infection on the gastrointestinal tract. One study that looked at gut microbiome diversity and richness during infection in the K18-hACE2 mouse model observed that high-dose inoculation led to greater disruption of microbial diversity and richness in the cecum, reduced relative abundance of Firmicutes and Actinobacter, and a reduced Firmicutes:Bacteroidetes ratio compared to control mice (Seibert et al., 2021). The increased abundance of families within the Firmicutes phylum was positively correlated with severity of viral infection in the lung (Seibert et al., 2021).

Syrian hamsters − although no overt clinical symptoms, such as diarrhea − had been observed, were also used to study SARS-COV-2-associated dysbiosis (Seibert et al., 2022; Sencio et al., 2022). Non-infectious viral RNA was found in the colon at 2 dpi, and little to no viral RNA was found in the small intestine and cecum (Sencio et al., 2022). Cytokines in the colon were upregulated, but no evidence of intestinal damage was observed (Sencio et al., 2022). Some evidence of intestinal barrier dysfunction was observed, with elevated intestinal fatty-acid-binding protein in the blood (Sencio et al., 2022). As with the mouse study and in human patients, changes in the gut microbiome composition were noted by 16S ribosomal RNA sequencing (Sencio et al., 2022; Seibert et al., 2022). On principal coordinate analysis, there was a clustering of gut microbiome composition across time, suggestive of a significant shift following SARS-CoV-2 infection (Seibert et al., 2022; Sencio et al., 2022), which correlated with disease severity and inflammation (Sencio et al., 2022). There was also an enrichment of opportunistic pathogens that positively correlated with viral RNA and pathology in the cecum, suggesting that microbiome changes may secondarily modulate disease severity in the gut (Seibert et al., 2022).

While studies into gastrointestinal manifestations of COVID-19 have not been performed in ferrets, rhesus macaques and cynomolgus macaques have been used to study SARS-CoV-2-associated gut dysbiosis (Sokol et al., 2021). While neither weight loss nor overt signs of pneumonia were noted, occasional and transient diarrhea was observed (Sokol et al., 2021). Viral RNA was detected in the rectal fluid of one cynomolgus and one rhesus macaque at 7 dpi (Sokol et al., 2021). Notably, one of these monkeys still had RNA-positive rectal samples at 26 dpi, suggesting that SARS-CoV-2 can persist in or shed through the gastrointestinal tract (Sokol et al., 2021). There were alterations in the fecal microbiota composition following infection with this disruption in the microbial ecosystem peaking at 10−13 dpi (Sokol et al., 2021). The relative abundance of Streptococcus species was strongly and positively correlated with select plasma chemokine levels, strengthening the link between gut dysbiosis and immune response (Sokol et al., 2021). Altogether, NHPs infected with SARS-CoV-2 showed evidence of gut microbiota disturbances and occasional diarrhea, with changes in gut microbiota composition associated with immune responses.

Considering the impact of altered gut microbiota composition on COVID-19 disease severity, it is important to conduct additional investigations that aim to detail the molecular mechanisms through which SARS-CoV-2 infections alter the gut microbiome and how these changes influence inflammation and COVID-19 severity. Furthermore, modulating the gut microbiota composition for potential mitigation of disease severity might prove to be an additional tool in disease management in a clinical setting. Animal models will be critical in testing hypotheses pertaining to the interaction between gut microbiota and COVID-19 severity.

Post-acute sequelae of COVID-19 (PASC), also referred to as long COVID, represents a significant concern in the context of SARS-CoV-2 infection. It encompasses the persistence or emergence of new clinical manifestations for 30 or more days following infection (Thaweethai et al., 2023). High-risk factors for PASC include being female, a member of an ethnic minority group, of low socioeconomic status, obese and having other comorbidities (Subramanian et al., 2022). Severe or mild acute COVID-19 illness can lead to PASC symptoms, suggesting that factors beyond viral infection, such as stress or other comorbidities, can contribute to the presentation of PASC symptoms (Hernandez-Romieu et al., 2022; Augustin et al., 2021). PASC prevalence is lower in those infected with the Omicron variant, previously vaccinated or treated with antiviral therapy, indicating that these factors influence its occurrence (Sherif et al., 2023).

Common symptoms associated with PASC include fatigue, post-exertional malaise, cognitive difficulties (often described as brain fog) and joint pain (Thaweethai et al., 2023). Fatigue and post-exertional malaise stand out as the most frequently reported and debilitating symptoms (Thaweethai et al., 2023). PASC prevalence estimates vary across different regions and populations, but it has been estimated that 6% of the adult population in the USA [see National Center for Health Statistics (CDC): Long COVID Household Pulse Survey (https://www.cdc.gov/nchs/covid19/pulse/long-covid.htm, accessed 3 September 2024], 2.9% of the entire population in the UK [see Office for National Statistics (ONS): Prevalence of ongoing symptoms following coronavirus (COVID-19) infection in the UK: 30 March 2023 (https://www.ons.gov.uk/peoplepopulationandcommunity/healthandsocialcare/conditionsanddiseases/bulletins/prevalenceofongoingsymptomsfollowingcoronaviruscovid19infectionintheuk/30march2023)] and 8.9% of 20,000 patients surveyed in Shanghai, China, reported experiencing PASC symptoms (Cai et al., 2023) might be experiencing PASC. To the global economy PASC is estimated to contribute to a loss of ∼1 tn US dollars every year (Al-Aly et al., 2024). Further investigation into the mechanisms of disease is crucial due to the detrimental impact of PASC on both clinical health and the economy, highlighting the need for additional research to identify effective treatments and preventatives.

The mechanisms underlying PASC are complex, probably multifactorial. They might include immune dysregulation; microbiota dysbiosis; autoimmunity and priming; blood clotting and endothelial abnormalities; dysfunctional neurological signaling; reactivation of other viruses, such as the Epstein-Barr virus; and viral persistence (de Melo et al., 2021; Natarajan et al., 2022; Stein et al., 2022; Craddock et al., 2023). However, the exact mechanisms underlying the syndrome are unclear (Calabrese et al., 2021; Pretorius et al., 2021; Liu et al., 2022b; Peluso and Deeks, 2022; Phetsouphanh et al., 2022; Seeßle et al., 2022; Su et al., 2022). Evidence indicates that SARS-CoV-2 RNA or antigen can persist in humans for extended periods of time (Stein et al., 2022; Appelman et al., 2024; Proal et al., 2023). PCR analysis of tissues obtained from autopsies of 44 unvaccinated individuals who had died from COVID-19 between 26 April 2020 and 2 March 2021 revealed viral RNA in various parts of the body − including the brain, respiratory system, cardiovascular system and gastrointestinal system − up to 230 days after the initial infection (Stein et al., 2022). In one patient, subgenomic RNA − which is generated by the virus during active replication and signifies ongoing viral replication − was also detected in the lungs as late as 76 dpi (Stein et al., 2022). Despite these findings, no infectious virus was isolated from any samples obtained from the more chronically infected patients (Stein et al., 2022). Additionally, the presence of viral RNA in patient fecal samples up to 210 dpi suggests that the GI tract may be a viral reservoir (Natarajan et al., 2022). Analysis of patient blood samples revealed the presence of viral proteins but not RNA, suggesting that viral persistence occurs primarily within tissues (Proal et al., 2023). Altogether, the data suggest that viral persistence can be observed in multiple tissue types; however, the connection between viral persistence and PASC symptoms remains to be determined.

Significantly more IFN-gamma- and TNF-alpha-producing SARS-CoV-2-specific T-cells were found in the blood of patients who got infected with SARS-CoV-2 and had been diagnosed with PASC compared to those who recovered from COVID-19 without PASC (Littlefield et al., 2022). The increased frequency of these T-cells was also associated with increased systemic inflammation and decreased lung function, suggesting that these T-cells contribute to symptoms of pulmonary PASC, and the increased frequency of SARS-CoV-2-specific T-cells was attributed to a viral reservoir (Littlefield et al., 2022). Additionally, analysis of blood obtained from patients diagnosed with PASC or not at 8 months post infection with SARS-CoV-2 revealed markers of T-cell exhaustion in those diagnosed with long COVID (Yin et al., 2024). T-cell exhaustion suggests that ongoing viral stimulation occurs in patients with PASC symptoms (Yin et al., 2024). The same study found more SARS-CoV-2 antibodies in patients diagnosed with long COVID than in recovered patients, further supporting the idea of a viral reservoir and immune disturbances caused by persistent virus presence (Yin et al., 2024). However, in patients who had previously experienced COVID-19, those experiencing the PASC symptom of post-exertional malaise and those who had recovered, both had viral N protein within their skeletal muscle tissue (Appelman et al., 2024). There were no differences in the level of N protein between PASC and recovered patients, suggesting that factors other than viral persistence were playing a role in the pathophysiology of post-exertional malaise in PASC patients, highlighting the need to further explore mechanisms of PASC pathophysiology (Appelman et al., 2024).

The current lack of established animal models for PASC underlines the challenges associated with performing PASC studies using animals (Frere et al., 2022; Jansen et al., 2022). Overcoming those challenges could expedite the development of PASC models.

Efforts are underway to develop and validate a PASC animal model (Jansen et al., 2022; Usai et al., 2023). At 60 dpi BALB/c mice infected with SARS-CoV-2 MA10 showed signs of neuroinflammation and neuropathogenesis, suggesting this model to be suitable to investigate the long-term neuropathological changes associated with SARS-CoV-2 infection (Gressett et al., 2023 preprint). Such neurological insults could have enduring impacts on behavior and cognitive function, which can be explored in future long-term studies using behavioral and physiological assays. MA10 infection in mice also resulted in chronic pulmonary damage and fibrosis, symptoms as observed in humans recovering from diffuse alveolar damage (Dinnon et al., 2022). Infected hamsters had chronic inflammation in the olfactory bulbs and olfactory epithelium that correlated with a decrease in compulsive and anxiety-like behavior (Frere et al., 2022). NHPs have been shown to get acute neuroinflammation, neuronal and glial cell injury, and hypoxic injury associated with SARS-CoV-2 infection (Rutkai et al., 2022). Although the use of NHPs for long-term COVID research poses significant challenges, they appear to be a suitable model for neurological PASC since they show evidence of early neurological lesions that may have lasting effects and contribute to PASC symptoms. Murine models appear to be promising candidates for GI-PASC research because their gut microbiome can be manipulated more easily compared to other species (Hart et al., 2018; Park and Im, 2020; Manca et al., 2020). Further efforts to establish preclinical animal models for PASC will be essential in streamlining the development of therapeutics and preventatives.

Over the past 4 years, an exceptional amount of data investigating SARS-CoV-2 infection has highlighted how investment in animal models of infectious disease can be mobilized to understand the pathogenic processes and mechanisms involved in acute disease (Table 1). As the world population now turns to dealing with an ongoing endemic disease and its longer-term manifestations, it is important to re-examine the utility of current animal models. Understanding the mechanisms of residual disease in various biological systems remains a key challenge. While the acute disease processes most dramatically involve the immune and pulmonary systems, the subacute and post-acute effects on the cardiovascular, neurological and gastrointestinal systems are more challenging to study in animal models. This Review has highlighted animals that demonstrate efficacy as models for acute disease processes involving these organ systems; however, the suitability of these animals as models for post-acute disease processes remains unclear.

To understand the complexities of COVID-19 infection, new and continued innovation in both animal models and in vitro systems, such as organoid models, is essential. A significant obstacle in developing animal models for COVID-19 is the high cost and limited access to biocontainment research facilities. Creating animal and organ system models of COVID-19 that can be studied outside of this restrictive environment would enable more extensive research, thereby accelerating the validation and creation of effective models. Alternatives to animal models, such as human organoids, show potential use in the study of pathological processes associated with SARS-CoV-2 infection (Ng et al., 2023). The current primary limitation with organoids is that any organoid lesions associated with viral infection cannot be correlated with clinical symptoms. Organoids also provide limited integration of multiple organ systems in comparison with animal models. Therefore, animal models remain critical for understanding the interactions between organ systems and the immune system, and how these contribute to the clinical manifestations of COVID-19.

Information presented here aims to guide the design of upcoming studies that explore post-acute disease processes within the organ systems discussed, which will play a pivotal role in developing effective treatments and preventative strategies for disease associated with SARS-CoV-2 infection.