Organoid Models of SARS-CoV-2 Infection: What Have We Learned about COVID-19?

: Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) causes coronavirus disease 2019 (COVID-19), which was classiﬁed as a pandemic in March 2020. As of 22 January 2022, globally more than 347 million cases of COVID-19 have been diagnosed, with 5.6 million deaths, making it the deadliest pandemic since the inﬂuenza pandemic in 1918. The clinical presentation of COVID-19-related illness spans from asymptomatic to mild respiratory symptoms akin to inﬂuenza infection to acute symptoms, including pneumonia necessitating hospitalisation and admission to intensive care units. COVID-19 starts in the upper respiratory tract and lungs but in severe cases can also involve the heart, blood vessels, brain, liver, kidneys and intestine. The increasing global health and economic burden of COVID-19 necessitates an urgent and global response. Understanding the functional characteristics and cellular tropism of SARS-CoV-2, and the pathogenesis that leads to multi-organ failure and death, has prompted an unprecedented adoption of organoid models. Successful drug discovery and vaccine development rely on pre-clinical models that faithfully recapitulate the viral life cycle and the host cell response to infection. Human stem cell-derived organoids fulﬁll these criteria. Here we highlight the role of organoids in the study of SARS-CoV-2 infection and modelling of COVID-19 pathogenesis.


Modelling Human Infectious Disease with Organoids
Organoids are defined as three-dimensional structures grown from stem cells and consist of organ-specific cell types that self-organize to recapitulate key features and functional characteristics of tissues in a dish [1]. Organoids can be initiated from two main types of stem cells: (1) pluripotent embryonic stem (ES) cells or synthetically induced pluripotent stem (iPS) cells and (2) organ-restricted stem cells. Pluripotent stem cell generated organoids were made more accessible by the discovery by Takahashi and Yamanaka that somatic cells could be taken back ("induced") to pluripotency [2]. Organoids generated from organ-restricted stem cells were first described by the Clevers laboratory in 2009 [3] using mouse intestinal epithelium. Using variations of this intestinal epithelium protocol, organoids have since been established from many organs, especially epithelial tissues. A PubMed search on the term "organoid" shows an exponential increase in publications since 2009 (Figure 1a). Earlier use of the term in the period of 1965-1985 was initiated by developmental biologists to describe tissue explants. Organ-restricted stem cell-derived The infectious disease field was slower on the uptake of organoid technology. Early studies by the Clevers [8] and Estes [9] laboratories demonstrated the utility of tissue stem cell-derived organoids to model respiratory and intestinal infections, respectively, while the Zika virus (ZIKV) outbreak demonstrated the utility of brain organoids established from iPS cells in understanding this disease [10]. However, the adoption of organoids for infectious disease modelling was largely overlooked until the COVID-19 pandemic that has had a devastating impact socially and economically worldwide. This pandemic has triggered the urgent development of medical strategies to tackle COVID-19 and has exposed the limitations of conventional, so called "gold-standard" tissue culture methods used by virologists since the mid 1960′s. COVID-19 is a systemic disease and in this review, we highlight the advances made and the insights gained that are attributable to the adoption of organoid technology [11][12][13][14]. Half the articles with the PubMed search term "organoids AND infection" for 2021 were COVID-19 studies (Figure 1b,c). The era of organoids modelling infectious disease has arrived [15].

Human Coronaviruses
Coronaviruses (CoVs) are enveloped, non-segmented, positive-sense single stranded RNA (ssRNA) viruses that belong to the Coronaviridea family and can infect humans and animals [16]. Seven CoVs are known to infect humans, four (human (H)CoV-NL63, HCoV-OC43, HCoV-229E and HCoV-HKU) of which cause mild, seasonal respiratory tract diseases. Three of the seven have emerged over the last 20 years causing severe human disease. Severe acute respiratory syndrome CoV (SARS-CoV), a beta coronavirus, emerged in late 2002 in Guangdong Province, China and spread rapidly from a single index case to five countries and led to large outbreaks, many in tertiary hospitals, with a case fatality of 9.6%. The virus spread to a further seven countries and, in all, resulted in 8098 cases globally with 774 deaths, including many healthcare workers [17]. SARS-CoV is thought to have originated from bats and transmitted from bats to wild animals, such as civets, and then to humans. The epidemic ended in 2004, despite human-human transmission, but possibly resulting from the relatively low infectivity of the virus. In 2012, Middle Eastern Respiratory Syndrome CoV (MERS-CoV) emerged, causing outbreaks of severe pneumonia in Saudi Arabia and South Korea that were associated with a high fatality rate of 34% [18,19]. Imported cases have been reported in several countries, The infectious disease field was slower on the uptake of organoid technology. Early studies by the Clevers [8] and Estes [9] laboratories demonstrated the utility of tissue stem cell-derived organoids to model respiratory and intestinal infections, respectively, while the Zika virus (ZIKV) outbreak demonstrated the utility of brain organoids established from iPS cells in understanding this disease [10]. However, the adoption of organoids for infectious disease modelling was largely overlooked until the COVID-19 pandemic that has had a devastating impact socially and economically worldwide. This pandemic has triggered the urgent development of medical strategies to tackle COVID-19 and has exposed the limitations of conventional, so called "gold-standard" tissue culture methods used by virologists since the mid 1960's. COVID-19 is a systemic disease and in this review, we highlight the advances made and the insights gained that are attributable to the adoption of organoid technology [11][12][13][14]. Half the articles with the PubMed search term "organoids AND infection" for 2021 were COVID-19 studies (Figure 1b,c). The era of organoids modelling infectious disease has arrived [15].

Overview of SARS-CoV-2 2.1. Human Coronaviruses
Coronaviruses (CoVs) are enveloped, non-segmented, positive-sense single stranded RNA (ssRNA) viruses that belong to the Coronaviridea family and can infect humans and animals [16]. Seven CoVs are known to infect humans, four (human (H)CoV-NL63, HCoV-OC43, HCoV-229E and HCoV-HKU) of which cause mild, seasonal respiratory tract diseases. Three of the seven have emerged over the last 20 years causing severe human disease. Severe acute respiratory syndrome CoV (SARS-CoV), a beta coronavirus, emerged in late 2002 in Guangdong Province, China and spread rapidly from a single index case to five countries and led to large outbreaks, many in tertiary hospitals, with a case fatality of 9.6%. The virus spread to a further seven countries and, in all, resulted in 8098 cases globally with 774 deaths, including many healthcare workers [17]. SARS-CoV is thought to have originated from bats and transmitted from bats to wild animals, such as civets, and then to humans. The epidemic ended in 2004, despite human-human transmission, but possibly resulting from the relatively low infectivity of the virus. In 2012, Middle Eastern Respiratory Syndrome CoV (MERS-CoV) emerged, causing outbreaks of severe pneumonia in Saudi Arabia and South Korea that were associated with a high fatality rate of 34% [18,19]. Imported cases have been reported in several countries, including the United Kingdom, the US, Germany, Italy, Greece, The Netherlands, Malaysia and South Korea. The largest Organoids 2022, 1 4 outbreak occurred in 2014 in Saudi Arabia, with over 500 hospitals involved within a few months. In 2015, a large outbreak across 16 hospitals arising from a single index case occurred in South Korea. The outbreak was finally stopped by quarantining over 17,000 individuals [18]. MERS-CoV is also thought to have originated in bats.
These two highly pathogenic HCoV and the large pool of SARS-CoV-like viruses discovered in bats [20] meant that further zoonotic CoV epidemics and pandemics were highly likely.

The Emergence of SARS-CoV-2
In December 2019, a novel CoV (initially named "2019-nCoV") emerged in Wuhan, Hubei Province, China, causing severe pneumonia [21]. The virus was found to share 80% homology with SARS-CoV and was subsequently renamed as SARS-CoV-2. Genomic sequencing suggested that the reservoir for SARS-CoV-2 is likely to be bats and pangolins [22][23][24]. The disease caused by SARS-CoV-2 was soon named coronavirus disease 2019 (COVID-19) and declared a fast-evolving pandemic by the World Health Organization (WHO) on 11 March 2020. This triggered an unprecedented global effort to investigate the characteristics of the virus to develop strategies to diagnose, prevent and treat COVID-19. As of 22 January 2022, more than 347 million cases of COVID-19 have been diagnosed worldwide, with 5.6 million deaths, making it the deadliest pandemic since the influenza pandemic of 1918.

SARS-CoV-2 Cell Entry
SARS-CoV-2 particles comprise four main structural proteins: spike (S), envelope (E), membrane (M) and nucleocapsid (N) (Figure 2). CoV S proteins are composed of two domains, the receptor binding (S1) and fusion (S2) domains. Viral entry is initiated by the docking of the S protein [25,26] via its receptor binding domain (RBD) in S1 to its cellular receptor. Like SARS-CoV, SARS-CoV-2 uses angiotensin-converting enzyme 2 (ACE2) as its cellular receptor [27,28]. In the respiratory tract, the ACE2 receptor is preferentially expressed on the apical surface of epithelial cells. Both SARS-CoV and SARS-CoV-2 infect by binding apically distributed ACE2 and during their egress from infected cells, are also released from the apical surface of respiratory cells [29][30][31][32]. This is significant as our understanding of viral entry, release and neutralization may be guided very differently by studying these events in polarized, compared to non-polarized, respiratory epithelial cells.
Notably, the S protein of SARS-CoV-2 shares only~73% homology with the S protein of SARS-CoV. The SARS-CoV-2 S protein has an insertion of four amino acids at the S1/S2 junction, known as the multibasic cleavage site (MBCS), rendering it more efficiently cleaved by cellular proteases [27,33]. Cleavage of the spike protein by the cellular serine protease, transmembrane protease serine 2 (TMPRSS2), is required to transform the SARS-CoV-2 spike protein into a fusogenic state, which initiates direct fusion between viral and cellular membranes [34] (Figure 2). The SARS-CoV-2 infection of primary cells is dependent on TMPRSS2 cleavage [33,34]. In continuous cell lines, such as Vero cells established from African Green Monkey kidney epithelium [35], the SARS-CoV-2 uses the endocytic entry pathway wherein the gradual endosomal acidification activates cathepsins that effectively prime the S protein to initiate entry [33,36]. Notably, SARS-CoV-2 binds to human ACE2 more strongly than SARS-CoV [27,28] due to several residue substitutions in its RBD that stabilize the binding domains [37]. Mutations to SARS-CoV-2 spike protein, especially in the RBD, have arisen during the COVID-19 pandemic with emerging variants outcompeting pre-existing circulating isolates. This poses a challenge to vaccine development and antiviral discovery.

Emergence of SARS-CoV-2 Mutants
Since the emergence and persistence of SARS-CoV-2 in the global population, the virus has undergone several mutational shifts towards variants carrying combinations of changes within the S1 domain of the S protein, in addition to those already occurring in the RBD, enhancing their affinity and binding to the human ACE2 receptor. These mutations resulted in greater virulence and transmissibility and are potentially able to evade neutralizing antibodies (NAb) targeting the RBD and receptor binding motif (RBM) of the spike protein [38][39][40][41] [45][46][47]. The most recent variant to emerge is Omicron (B.1.529) [48,49]. Being substantially more infectious than Delta, Omicron is not effectively inhibited by vaccine-induced neutralizing antibodies and, in individuals who have received only two doses of COVID-19 vaccines, is associated with reduced vaccine efficacy [50][51][52][53].

COVID-19 Is a Systemic Disease
SARS-CoV-2 is mainly transmitted through aerosolised respiratory droplets [54]. However, transmission by direct contact and through mucosal membranes is also possible, although they are less likely to transmit infection than close contact with an infected person [55,56]. The basic reproductive number (R0) of the ancestral Wuhan virus and early variants ranged from 2 to 4 [57,58]. For the Delta variant, the R0 is 97% higher than previous variants, making this virus substantially more transmissible [59]. However, with an R0 that is 4-fold higher than Delta, Omicron became the most transmissible SARS-CoV-2 variant to date [60].
The median incubation period for people who become symptomatic after contact with an infected person is 5 to 6 days for the Wuhan strain [61,62]. In contrast, the The virion schematic shows in more detail the S protein subunit 1 (S1) containing the receptor binding domain (RBD). The SARS-CoV-2 replication cycle shows two modes of viral entry (endocytic and TMPRSS2-mediated pathways). Both require docking to ACE2 at the cell surface and release of the viral genome into the cytoplasm via membrane fusion. The endocytic pathway is inhibited by (hydroxy)chloroquine and E64d, while TMPRSS2-mediated entry is inhibited by camostat. After synthesis of viral genome and proteins, viral particles are assembled and exit the infected cell. Figure 2 was created with BioRender.com.

Emergence of SARS-CoV-2 Mutants
Since the emergence and persistence of SARS-CoV-2 in the global population, the virus has undergone several mutational shifts towards variants carrying combinations of changes within the S1 domain of the S protein, in addition to those already occurring in the RBD, enhancing their affinity and binding to the human ACE2 receptor. These mutations resulted in greater virulence and transmissibility and are potentially able to evade neutralizing antibodies (NAb) targeting the RBD and receptor binding motif (RBM) of the spike protein [38][39][40][41] [45][46][47]. The most recent variant to emerge is Omicron (B.1.529) [48,49]. Being substantially more infectious than Delta, Omicron is not effectively inhibited by vaccine-induced neutralizing antibodies and, in individuals who have received only two doses of COVID-19 vaccines, is associated with reduced vaccine efficacy [50][51][52][53].

COVID-19 Is a Systemic Disease
SARS-CoV-2 is mainly transmitted through aerosolised respiratory droplets [54]. However, transmission by direct contact and through mucosal membranes is also possible, although they are less likely to transmit infection than close contact with an infected person [55,56]. The basic reproductive number (R0) of the ancestral Wuhan virus and early variants ranged from 2 to 4 [57,58]. For the Delta variant, the R0 is 97% higher than previous variants, making this virus substantially more transmissible [59]. However, with an R0 that is 4-fold higher than Delta, Omicron became the most transmissible SARS-CoV-2 variant to date [60].
The median incubation period for people who become symptomatic after contact with an infected person is 5 to 6 days for the Wuhan strain [61,62]. In contrast, the incubation period for Delta is 4 days and is shortened to 3 days for Omicron [63]. Both pre-symptomatic and asymptomatic transmissions have been reported [64,65]. Pre-symptomatic transmission occurs 1 to 3 days before symptom onset [66]. The viral load in the upper respiratory tract is highest in the first 3 to 5 days of infection [67], and although the loads are higher in symptomatic individuals, both pre-symptomatic and symptomatic persons can shed high viral loads, readily transmit infection and be associated with high secondary attack rates [68]. The viral load and duration of infectious virus shedding are higher for both the Delta and Omicron variants [69,70].
COVID-19 is a multisystem disease ( Figure 3) that starts in the upper respiratory tract and lungs but can also involve the heart, blood vessels, brain, liver, kidneys and intestine [71]. In their majority, patients with COVID-19 present with mild symptoms consisting of fever, cough, loss of taste and/or smell and dyspnea. Other less common symptoms include headache, sore throat, chills, diarrhea, chest pain and myalgias [72]. Whilst most cases resolve without any serious consequences, up to 20% of cases develop moderate to severe disease needing hospitalization. This is more likely to occur in unvaccinated individuals infected with either the Alpha or Delta variants [73,74]. The case fatality rate for infected individuals is approximately 1.9% [75], although this may rise to 8% for people aged 70 to 79 years and to 14.8% for people 80 years and older [76].
Organoids 2022, 1, FOR PEER REVIEW 5 incubation period for Delta is 4 days and is shortened to 3 days for Omicron [63]. Both pre-symptomatic and asymptomatic transmissions have been reported [64,65]. Pre-symptomatic transmission occurs 1 to 3 days before symptom onset [66]. The viral load in the upper respiratory tract is highest in the first 3 to 5 days of infection [67], and although the loads are higher in symptomatic individuals, both pre-symptomatic and symptomatic persons can shed high viral loads, readily transmit infection and be associated with high secondary attack rates [68]. The viral load and duration of infectious virus shedding are higher for both the Delta and Omicron variants [69,70]. COVID-19 is a multisystem disease ( Figure 3) that starts in the upper respiratory tract and lungs but can also involve the heart, blood vessels, brain, liver, kidneys and intestine [71]. In their majority, patients with COVID-19 present with mild symptoms consisting of fever, cough, loss of taste and/or smell and dyspnea. Other less common symptoms include headache, sore throat, chills, diarrhea, chest pain and myalgias [72]. Whilst most cases resolve without any serious consequences, up to 20% of cases develop moderate to severe disease needing hospitalization. This is more likely to occur in unvaccinated individuals infected with either the Alpha or Delta variants [73,74]. The case fatality rate for infected individuals is approximately 1.9% [75], although this may rise to 8% for people aged 70 to 79 years and to 14.8% for people 80 years and older [76].

Organoid Models of SARS-CoV-2
In vitro models for SARS-CoV-2 infection are essential for the understanding of host and viral factors supporting the cell tropism and the viral replication cycle. They are also needed for preparing viral inoculum for further research and pre-clinical evaluations of medical countermeasures against COVID-19. Early during the pandemic, virologists used classical cell lines such as the immortalized Vero cells and human continuous cell lines. The serious limitations of these simple tissue culture models are that they are not human or are derived from human cancers, and thus do not faithfully replicate healthy human physiology. The Vero cell line has been extensively used by virologists, as these cells have a homozygous gene deletion that causes the loss of the type I interferon (IFN-1) gene cluster and cyclin-dependent kinase inhibitor genes [77,78]. These cells are permissive to diverse viruses, yielding high titers of viral particles in the culture supernatant. However, the SARS-CoV-2 infection of Vero cells leads to the adaptation of the viral genome optimizing their replication in these cells. The MBCS is mutated or deleted in Vero cells [79]. The MBCS facilitates TMPRSS2-mediated entry, and given that this is the mode of entry into primary human epithelial cells [33,80,81], Vero cell-generated viral stocks that do not reflect authentic virus amplification/replication cycles may not recapitulate authentic infection.
The caveat to human cancer cell lines is that they are transformed with multiple changes to the cells' DNA and are also adapted to tissue culture [82]. Yet again, cancer cell lines are far removed from the original cancer and even more so from healthy human tissue. Interestingly, Calu-3 cells, established from a human lung adenocarcinoma, preserve the MBCS and better mimic the SARS-CoV-2 infection of primary epithelial cells than Vero cells [83]. Thus, Calu-3 might provide an intermediary culture system for SARS-CoV-2 propagation. Nonetheless, organoids can reproduce the pathology of COVID-19 of their specific tissue of origin and have been rapidly adopted during COVID-19 ( Figure 3). Here, we highlight the diverse human organoids employed in SARS-CoV-2 research.

Respiratory Tract
COVID-19 was first recognized as causing severe pneumonia; thus, it was vital to identify the regions and cell types of the respiratory tract that are infected with SARS-CoV-2. The respiratory tract architecture changes from a pseudostratified barrier epithelium lining the nose to the delicate monolayer of alveoli, which are small air sacs at the extremities of the lung enabling the exchange of oxygen and carbon dioxide (CO 2 ) molecules into and out of the bloodstream (Figure 4). The first publications testing the respiratory cell tropism of SARS-CoV-2 used human airway epithelium (HAE) established from progenitor cells (basal cells) isolated from conductive airway tissues resected during surgery. Pseudostratified 3D tissues were differentiated at an air-liquid-interface (ALI) [84,85] or embedded in a matrix [85] (Figure 4). For ALI differentiation, the basal cells are seeded onto transwell membranes, and once confluent across the surface of the membrane, the apical medium is removed, exposing the cells to air. ALI-differentiation medium is refreshed in the basal chamber over 3-4 weeks to generate the pseudostratified tissue. Matrix embedded organoids are similarly seeded in an expansion phase to form cysts, and then changed to an organoid differentiation medium for 3-4 weeks [8,32]. HAE organoids are comprised of several cell types, including basal cells, ciliated cells with apical beating cilia and goblet cells secreting mucous onto the apical surface. These studies confirmed the SARS-CoV-2 infection of ciliated cells, and that viral entry and release occur from the apical surface and was soon confirmed by others [21,84,85]. Using a reporter virus system, RNA in situ hybridization and cultures established from different regions of the respiratory tract, Hou and colleagues [86] demonstrated that the ALI-differentiated nose epithelium expressed the highest levels of ACE2 and TMPRSS2 and showed the greatest SARS-CoV-2 infectivity, which gradually decreased towards the proximal respiratory tract (Figure 4). Early expression studies either did not detect or detected low levels [87] or did not report [85] ACE2 protein expression in respiratory tissues, despite robust protein expression reported in other human tissues such as the intestine analyzed in parallel [85,87]. More recent studies have confirmed ACE2 protein expression and SARS-CoV-2 infections in respiratory epithelia, identifying the differentiated ciliated cells as the dominant cell type infected [21,29,32,88,89] (reviewed extensively elsewhere, e.g., [90][91][92]).
Organoids 2022, 1, FOR PEER REVIEW 7 [87] or did not report [85] ACE2 protein expression in respiratory tissues, despite robust protein expression reported in other human tissues such as the intestine analyzed in parallel [85,87]. More recent studies have confirmed ACE2 protein expression and SARS-CoV-2 infections in respiratory epithelia, identifying the differentiated ciliated cells as the dominant cell type infected [21,29,32,88,89] (reviewed extensively elsewhere, e.g., [90][91][92]). With the urgency to discover and evaluate therapies of severe respiratory damage in COVID-19, researchers have sought to establish human alveolar models of SARS-CoV-2 infection ( Figure 4). Unlike primary upper respiratory cells with well-established and characterized ALI culture systems [91], establishing similar cultures of primary alveolar cells proved more difficult. The main cell types of the alveoli sacs are the type I and type II pneumocytes (AT1 and AT2, respectively). Flat AT1 cells mediate gas exchange, while the AT2 cells secrete surfactants. Rare AT2 cells behave as the stem cells of the epithelium and maintain close contact with underlying Wnt-producing fibroblasts. Wnt-producing cells are invariably found in the niche of stem cells. Daughter cells leaving this Wnt-rich niche transdifferentiate into AT1 cells. The AT2 stem cells maintain homeostasis and respond to injury to repair the tissue; they self-renew, recruit AT2 stem cells to the site of injury and generate AT1 cells [93,94]. Recapitulating this level of complexity in tissue culture cannot be mimicked using immortal alveolar cell lines or primary alveolar cells [95]. Recent studies using lung bud tip matrix-embedded organoids to initiate the ALI cultures resulted in patches of bronchiolar and alveolar tissues [96]. SARS-CoV-2 infection demonstrated that in the alveolar regions of the culture, it is the AT2-like cells that support SARS-CoV-2 infection [96]. This has important ramifications for pathogenesis, given that the AT2 cells repair the tissue.  With the urgency to discover and evaluate therapies of severe respiratory damage in COVID-19, researchers have sought to establish human alveolar models of SARS-CoV-2 infection ( Figure 4). Unlike primary upper respiratory cells with well-established and characterized ALI culture systems [91], establishing similar cultures of primary alveolar cells proved more difficult. The main cell types of the alveoli sacs are the type I and type II pneumocytes (AT1 and AT2, respectively). Flat AT1 cells mediate gas exchange, while the AT2 cells secrete surfactants. Rare AT2 cells behave as the stem cells of the epithelium and maintain close contact with underlying Wnt-producing fibroblasts. Wnt-producing cells are invariably found in the niche of stem cells. Daughter cells leaving this Wnt-rich niche transdifferentiate into AT1 cells. The AT2 stem cells maintain homeostasis and respond to injury to repair the tissue; they self-renew, recruit AT2 stem cells to the site of injury and generate AT1 cells [93,94]. Recapitulating this level of complexity in tissue culture cannot be mimicked using immortal alveolar cell lines or primary alveolar cells [95]. Recent studies using lung bud tip matrix-embedded organoids to initiate the ALI cultures resulted in patches of bronchiolar and alveolar tissues [96]. SARS-CoV-2 infection demonstrated that in the alveolar regions of the culture, it is the AT2-like cells that support SARS-CoV-2 infection [96]. This has important ramifications for pathogenesis, given that the AT2 cells repair the tissue.
ALI-differentiated cultures established from bronchial lavage or lung resections in a 96-well format have recently been established as a high-throughput research platform (for example, [97]). Several respiratory tissue-derived ALI cultures are now commercially available (including nose, bronchial and alveolar). Although the cost of the commercial cultures is prohibitive for most research groups, it should accelerate drug discovery and vaccine development undertaken by larger research entities. Respiratory and more complex organoid models on-a-chip ( Figure 4) add an additional dimension to the utility of organoid models for studying SARS-CoV-2 infectivity (reviewed elsewhere, for example [98]).
Alternative models for high-throughput platforms have been established using human ES cell-derived lung organoids (hESC-LO) [99]. Here, the ES cells were differentiated into the endoderm, then towards the lung progenitor lineage and finally into differentiated lung cells. RNA-seq on the resultant differentiated lung cells confirmed the presence of AT2 cells expressing ACE2 and TMPRSS2. The alveolar-like organoids were permissive to SARS-CoV-2 infection and upregulated similar cytokine/chemokine genes as observed in lung autopsy tissues of COVID-19 patients [99]. A screen of FDA-approved drugs yielded some targets for follow-up. The main advantage of hES-derived organoids, despite the inability to completely recapitulate human tissue, is that different tissue types can be established from the same ES line, providing a consistent genetic background [1].

Intestine Epithelium
The first publication using human organoids for SARS-CoV-2 infection was the result of the collaboration between the Clevers and Haagmans laboratories in the Netherlands [85], combining tissue-stem cell derived organoid expertise with virology. The Clevers laboratory identified Leucine-rich, repeat-containing G-protein coupled receptor 5 (Lgr5) as an exclusive maker of adult intestinal epithelial stem cells [100]. They showed that Lgr5 + stem cells generate the different cell types of the epithelium and self-organize into 3D structures that mimic the gut epithelium when embedded in the matrix and provided with stem cell niche growth factors [3,101]. Human small intestinal organoids (hSIOs) can be expanded indefinitely and cryopreserved, providing a renewable resource. The hSIOs can be cultured and embedded into a matrix or grown at ALI [9]. hSIOs [85] and tissues [87] abundantly express ACE2 and support infection by SARS-CoV and SARS-CoV-2 [85]. Methodologies to clone and genetically engineer hSIOs were already established in the Clevers laboratory and employed to reveal essential host factors for coronavirus infection, demonstrating the power of coupling CRISPR/Cas9 gene editing with organoids [33]. Moreover, the hSIOs together with human alveolar and airway ALI organoids revealed differences between the ancestral SARS-CoV-2 and the Alpha (B.1.1.7) variant. The Alpha variant produced a higher replicative fitness, producing high titers of virus later in infection [102].
The very high levels of ACE2 and TMPRSS2 [103] in hSIOs make them a permissive culture system enabling a wide infection dynamic range and a robust platform for the screening of anti-viral inhibitors. Additionally, there is potential for an oral-fecal route of transmission that requires further investigation. SARS-CoV-2 is an enveloped virus ( Figure 2) and would not be expected to survive the harsh gastric environment. However, COVID-19 patients develop gastrointestinal symptoms; SARS-CoV-2 infection is reported to alter the intestinal microbiome, and SARS-CoV-2 RNA is detected in fecal samples [104,105].

Heart
Some COVID-19 patients develop severe cardiovascular complications, which led to investigations to determine if this was the direct result of the SARS-CoV-2 infection of cardiac tissue or a consequence of systemic disease. Examination by digital PCR, Western blot, immunohistochemistry, immunofluorescence, RNAscope and transmission electron microscopy assays of postmortem cardiac heart tissues of SARS-CoV-2 positive patients who showed no signs of cardiac involvement revealed that cardiomyocytes (CM) were indeed infected. Variable patterns of cardiomyocyte injury were observed [106]. This and other studies [107] examining autopsy tissues confirmed the SARS-CoV-2 presence in the heart and CM injury. To examine the human cardiac tropism of SARS-CoV-2, researchers adopted CM cultures derived from human pluripotent ES cells (hESC-CM) or human iPS cells (hiPS-CM).
The differentiation of hESCs is directed towards mesoderm lineage and then differentiation into spontaneously contracting CM, achieving 85-95% purity [108,109]. hESC-CMs express ACE2, while ES cells, mesoderm and cardiac progenitor cells do not [109]. hESC-CMs also express TMPRESS2, cathepsin L and furin [108], mirroring the expression of healthy human heart tissue (left ventricle) [108]. hESC-CMs are susceptible to SARS-CoV-2 infection, which is inhibited by the ACE2 antibody, camostat and E64d, indicating that cell entry requires docking to ACE2 but can proceed via an endocytic pathway (sensitive to E64d) or TMPRSS2-dependent membrane fusion (sensitive to camostat) [108] (Figure 2).
The differentiation of hiPS cells yields three cardiac cell types: endothelial cells, CM and cardiac fibroblast [107]. Of these, only CM expressed ACE2; all cell types expressed cathepsin L and cathepsin B, and none expressed TMPRSS2. Only the CM were infected with SARS-CoV-2, which was inhibited by an ACE2 antibody and E64d, but not by the TMPRSS2 inhibitors camostat and aprotinin [107]. Consequently, hiPS-CM [107,110] do not recapitulate the TMPRSS2 expression pattern of human heart tissue [108]. The SARS-CoV-2 infection of purified hiPS-CM caused cessation of beating after 72 h [111]; however, it remains to be determined if the cardiac arrest seen in COVID-19 patients [71] is the direct effect of the SARS-CoV-2 infection of the CM.
Despite these caveats, pluripotent stem cell-derived cardiomyocytes have been widely adopted as a platform to identify inhibitors of SARS-CoV-2 infection [108,112] and to recapitulate aspects of the CM damage caused by SARS-CoV-2 infection in humans [107,[109][110][111]. The hES cells and reagents to differentiate them into CM are now commercially available. The indication that SARS-CoV-2 entry into CM may depend on TMPRSS2-mediated fusion or on the endocytic pathway ( Figure 2) certainly warrants the inclusion of CM models in cell-type platforms used for screening SARS-CoV-2 medical countermeasures [99].

Liver
Several continuous cell lines were screened and utilized for SARS-CoV-2 infection experiments, including Huh7 cells, a human hepatocellular carcinoma cell line. The susceptibility of this liver cancer cell line to SARS-CoV-2 infection in vitro and the liver comorbidities associated with COVID-19 in patients with severe disease [113] prompted an analysis of existing single-cell (sc) RNAseq databases for cell types within the liver that express ACE2 and TMPRSS2 and could potentially be infected by SARS-CoV-2 (for example, references [114,115]). An analysis across five different liver tissue types (human fetal, healthy, cirrhotic, tumor and adjacent normal) identified a single subpopulation of human liver that co-expressed ACE2 and TMPRSS2, namely the cholangiocyte-biased progenitor pool [115].
Indeed, using human liver organoids, the SARS-CoV-2 infection of cholangiocytes was subsequently confirmed [116]. Human liver organoids established from biliary duct fragments are bipotent progenitor cells that expand as cystic organoid structures and can be differentiated towards a biliary or hepatic cell fate [117,118]. In their expansion phase, the organoid cells express cholangiocyte markers [116,118], ACE2 and TMPRSS2 and support SARS-CoV-2 infection, leading to the formation of large syncytia [116]. Transcriptomic analysis by scRNAseq revealed the decreased expression of genes involved in barrier functions and bile acid transport following SARS-CoV-2 infection. This study in organoids supports the idea that liver damage in COVID-19 patients might be due to cholangiocyte injury induced by SARS-CoV-2 infection [116]. The advantage of human liver ductal organoids resides in their long-term expansion [118], providing a renewable source of primary cells for SARS-CoV-2 infection.
McCarron and colleagues [119] took organoid technology one step further. Nonalcoholic fatty liver disease (NAFLD) and nonalcoholic steatohepatitis (NASH) are becoming recognized as co-morbidities of COVID-19 [120]. The investigators have established and characterized bipotent ductal organoids from patients with NASH because animal models cannot recapitulate this human pathology. Transcriptomic and functional investigations confirmed that NASH patient-derived organoids recapitulate key features of human NASH. Infection with the SARS-CoV-2-VSV pseudovirus that expresses the SARS-CoV-2 S protein and requires ACE2 and TMPRSS2 for fusion with the cellular membrane was more efficient in NASH organoids than in organoids established from healthy tissues. Interestingly, increased permissiveness to SARS-CoV-2 was not associated with elevated ACE2 or TMPRSS2 expression but an upregulation of ubiquitin D, a known inhibitor of the interferon antiviral response, in the NASH organoids akin to human NASH liver tissue [119]. Establishing organoids from diseased tissues has the potential to reveal the mechanisms of comorbidities in COVID-19.

Brain
Initial symptoms, such as headache, dizziness and taste and smell impairment were commonly reported for SARS-CoV-2 patients, but in most cases resolved quickly. More serious indications of neurological disorder include seizures, encephalopathy and cerebrovascular disease [121]. MRI-based screens showed structural damage in the brains of recovering COVID-19 patients, indicating potential long-term neurological effects [122]. SARS-CoV-2 RNA was detected in the cerebral spinal fluid (CSF) of COVID-19 patients with neurological symptoms [123] and in autopsy brain samples [124]. To confirm or deny SARS-CoV-2 infection and identify target cell types in the brain, researchers turned to human brain organoids established from hES cells or hiPS cells. Initial results seemed conflicting, detecting SARS-CoV-2 virus in the neurons but not productive infection [125]. This was confirmed in a systematic test of SARS-CoV-2 neurotropism in region-specific brain organoids (cortical, hippocampal, hypothalamic and mid-brain), which also did not show productive SARS-CoV-2 infection [126]. Subsequent studies with brain organoids that contained choroid plexus epithelium were productively infected, but only in the choroid plexus region of the organoid [126,127]. Tropism for choroid plexus epithelium was confirmed using pure choroid plexus organoids [126]. It is proposed that productive SARS-CoV-2 infection leads to the damage of the epithelium and loss of barrier function, allowing viral entry into the central nervous system (CNS).
Another potential portal of entry into the CNS is from the nasal cavity through the olfactory bulb. An elegant study by Meinhardt and colleagues [128] showed that SARS-CoV-2 can enter the CNS by crossing the neural-mucosal interface in olfactory mucosa. The proximity of mucosal, endothelial and nervous tissue, including olfactory and sensory nerve endings, coupled with the robust SARS-CoV-2 infection of the nasal epithelium, might precipitate entry into the CNS via the olfactory tract. This may explain some of the symptoms of COVID-19, including loss of smell and taste.

Kidney
ACE2 is strongly expressed in kidney tubules [87,129,130]; SARS-CoV-2 can be detected in urine [131] and in autopsy kidney tissue from SARS-CoV-2-infected patients [132,133], and COVID-19 patients show signs of kidney damage [71,134,135]. To determine if the kidney function decline reported for COVID-19 patients could be due to the infection of kidney cells by SARS-CoV-2, hES-derived kidney epithelium organoids were interrogated. The engineered kidney organoids expressed ACE2 and TMPRSS2 and were infected with SARS-CoV-2, producing infectious virus, and the infection was inhibited by soluble ACE2 [136,137]. Remdesivir, an RNA polymerase inhibitor that blocks viral replication, had an additive anti-viral effect when combined with soluble ACE2 in the kidney organoids [138].
An alternative approach, using normal human kidney proximal tubule epithelial cells, was to establish kidney epithelium organoids embedded in a matrix. The organoids expressed ACE2, and a pseudoviral assay was used to confirm that the SARS-CoV-2 S protein enabled pseudoviral entry into the cells [139]. Notably, conditional reprogramming (CR) conditions were used to establish long-term cultures of the primary kidney epithelial cells in 2D before initiating organoid formation and differentiation in 3D. The CR technique has been used to establish long-term cultures from a variety of tissue types without genetic manipulation [140]. An integrated 2D CR and 3D provides a physiologically relevant platform for drug discovery and safety evaluation to study kidney function and the innate immunity of kidney cells to viral infection.

The Eye-An Alternative Route of Entry
Although the respiratory system is considered the primary route for SARS-CoV-2 infection, and the nose epithelium the dominant initial site of infection, observations by ophthalmologists early in the pandemic raised the possibility that the eye was also a route of viral entry [141]. People contracted COVID-19 despite wearing a face mask to protect the respiratory route of entry, which led to investigations of the ocular surface as a site of infection by SARS-CoV-2. The ocular surface includes the cornea, sclera and limbus, and an examination of post-mortem samples demonstrated that these tissues express ACE2 and TMPRSS2 [142,143] and were positive for SARS-CoV-2 S protein by immunofluorescence in samples from SARS-CoV-2 positive cadavers [142]. To confirm that these tissue cell types were indeed susceptible to SARS-CoV-2 infection, cultures were established from cells isolated from eye regions of healthy donors and infected with SARS-CoV-2. The limbus cells were the most susceptible primary ocular tissue to SARS-CoV-2 infection [142]. Accessing human eye tissue relies on organ donation, and the cultures established have a limited lifespan and complexity. Human ES cell and iPS cell-derived retinal [144] and "whole eye" [142,145] organoids have been generated and confirm endogenous ACE2 and TMPRSS2 expression [142,146], and that SARS-CoV-2 infection is primarily in limbus-like cells [142,145]. Collectively, these data demonstrate the power of pluripotent stem cellderived organoid technology in establishing the eye as a potential mode of SARS-CoV-2 transmission and emphasizes the importance of face shields to prevent touching the face or eye, especially in the case of conjunctivitis.

Vasculature
The vasculature, like neuronal networks, is of great importance to our understanding of COVID-19 pathogenesis as it encompasses all organs. The endothelium lines the inner surface of all blood vessels and plays crucial roles in human physiology and pathophysiology, including COVID-19. The endothelium functions in both adaptive and innate immunity; it provides an anticoagulant surface and maintains blood vessel barrier integrity. The dysregulation of all these functions features in the sequelae of COVID-19 (reviewed extensively, e.g., [147]). However, it remains controversial whether endothelial damage is a direct effect of SARS-CoV-2 infection of the endothelial cells.
Given that severe COVID-19 is a multi-organ disease, the initial focus was on the affected organs. Nonetheless, endothelial damage is causally implicated in COVID-19 organ dysfunction (reviewed extensively, e.g., [147][148][149][150][151]). Early in the pandemic, it was recognized that circulating endothelial cells in COVID-19 patients, which is indicative of endothelial damage, correlated with increases in markers of disease severity and inflammatory cytokines [152,153]. Examinations of autopsy tissues from COVID-19 patients by immunohistochemistry revealed an accumulation of inflammatory cells associated with the endothelium in vascular beds of several organs, including the heart, small intestine and lung [154]. Further examination of kidney post-mortem tissues by electron microscopy showed viral inclusion bodies in the peritubular space and viral particles in endothelial cells of the glomerular capillary loops [154]. The productive SARS-CoV-2 infection of endothelial cells was confirmed using hiPS-derived capillary organoids. Infection of these blood vessel organoids was inhibited by soluble human ACE2, indicating ACE2-mediated viral entry [136]. In a similar vein, Yang and colleagues [155] used hES to generate eight different distinct cell types or organoids modelling the embryonic development of tissues derived from the three germ layers (definitive endoderm, mesoderm and ectoderm). Productive SARS-CoV-2 infection was confirmed in several tissue types, including endothelial cells [155]. Mouse models of SARS-CoV-2 also support the direct infection of lung and cardiac endothelial cells [156].
Nonetheless, the extent to which the direct infection of endothelial cells by SARS-CoV-2 contributes to vascular involvement in COVID-19 remains to be determined and may indeed be organ and tissue dependent. In vitro cultures of primary human umbilical vein endothelial cells (HUVECs) and human microvascular endothelial cells from the lung (HMVEC-L) did not support SARS-CoV-2 infection [157]. These HUVECs and HMVES-L cells expressed very low levels of ACE2 and TMPRSS2 and did not show any morphological changes upon virus exposure to their apical or basal surface. In agreement, Wagner and colleagues [158] could not demonstrate a productive SARS-CoV-2 infection of endothelial cell models established from the vasculature of diverse tissues, irrespective of the SARS-CoV-2 variant used for infection. Virus uptake occurred, but viral replication was not supported. Another study using HUVECs reported that cloned SARS-CoV-2 proteins cause vascular permeability [159]. Resolving these controversies requires further interrogation using physiological human model systems, including more complex organoid models, where the vasculature is embedded within the organoid. For example, hiPS-derived kidney organoids contain multiple renal cell types and structures, as well as the vasculature [160].
Further clear examples of vascular involvement came from COVID-19 infections of children. Although severe COVID-19 illness was rare in children, several cases of acute pediatric vasculitis of unknown origin were diagnosed, initially in Italy and England, and then globally [161][162][163]. It was described as a Kawasaki-like multisystem inflammatory syndrome [164]. Similar symptoms were later observed in adolescent and adult COVID-19 patients, who did not necessarily present with respiratory manifestations and often tested negative for SARS-CoV-2 by PCR on respiratory swab samples (e.g., [165,166]). The latter implies vasculitis occurs after active infection. Again, direct evidence of productive infection of the endothelium is yet to be determined in these patients.
Further vascular complications in COVID-19 include coagulopathy and resulting thromboembolic events, the exact underlying mechanisms of which are unknown; however, the prophylactic administration of heparin to patients requiring hospitalisation significantly reduced mortality [167,168]. Endothelial dysfunction is also implicated in long COVID-19 by a multicenter observational study of 798 patients [169]. Our understanding of the roles played by the vasculature are far from complete and will undoubtedly fuel the development of more complex organoids.

Other Organs
Many other organoids derived from tissue stem cells or human pluripotent stem cells have been adopted for SARS-CoV-2 research to define cell tropism and to understand pathogenesis and modes of transmission ( Figure 3). These include tonsils and the oral cavity, with saliva as a potential means of transmission, the pancreas, alveolar macrophages and gastric epithelium, to name a few (reviewed comprehensively elsewhere in addition to review articles cited here).

Transmission of SARS-CoV-2 from Humans to Animals
Reverse zoonosis is defined as any disease that is spread from humans to animals and tends to occur when there is high prevalence of human infection. With the capacity of coronaviruses to jump species boundaries and adapt to new hosts [170], the reverse zoonosis of SARS-CoV-2 to a wide range of animal species [171] was not unexpected. There are, however, two main concerns with events of reverse zoonosis: firstly, the infected animals may develop severe disease and even die, and secondly, the population of animals in question can become a virus reservoir, from which reintroduction back into humans can take place [172].
During the COVID-19 pandemic, the transmission of SARS-CoV-2 from humans to companion, zoo and farmed animals has been documented [171,[173][174][175] (Figure 5). The first domestic animals to have naturally acquired SARS-CoV-2 infection were cats (Felis catus) [176,177], dogs (Canis lupus familiaris) [176,178] and pet ferrets (Mustela putorius furo) [179,180] living in the same household with COVID-19 patients. The impact of SARS-CoV-2 infections, through reverse zoonosis on domestic animal health, is unclear as the symptoms reported in these animals ranged from asymptomatic to severe respiratory illness [176]. The airborne transmission of SARS-CoV-2 between cats has also been reported [181].
main concerns with events of reverse zoonosis: firstly, the infected animals may develop se-vere disease and even die, and secondly, the population of animals in question can become a virus reservoir, from which reintroduction back into humans can take place [172].
During the COVID-19 pandemic, the transmission of SARS-CoV-2 from humans to companion, zoo and farmed animals has been documented [171,[173][174][175] (Figure 5). The first domestic animals to have naturally acquired SARS-CoV-2 infection were cats (Felis catus) [176,177], dogs (Canis lupus familiaris) [176,178] and pet ferrets (Mustela putorius furo) [179,180] living in the same household with COVID-19 patients. The impact of SARS-CoV-2 infections, through reverse zoonosis on domestic animal health, is unclear as the symptoms reported in these animals ranged from asymptomatic to severe respiratory illness [176]. The airborne transmission of SARS-CoV-2 between cats has also been reported [181].
Among captive wild animals in zoological settings, tigers (Panthera tigris) [170,182,183], lions (Panthera leo) [170,182], puma (Puma concolor) [184], snow leopards (Panthera uncia) [185], gorillas (Gorilla gorilla) [186,187] and otters (Aonyx cinereus) [188] were found to be naturally infected through contact with SARS-CoV-2-infected zoo staff. Animals displayed varying degrees of respiratory illness but recovered. Whether subsequent animal-to-animal transmission occurred is not known. The human transmission of SARS-CoV-2 to farmed minks has been documented in the United States, Canada and many European countries [171]. Minks exposed to the virus presented mild to severe respiratory distress because of interstitial pneumonia [189]. The origin of the SARS-CoV-2 outbreak could be traced back to the farmers and their close contacts who tested positive for the virus and displayed COVID-19 symptoms [171]. In the United States, the rapid emergence of the N501T (Asn 501 → Thr) mutation in the RBD of the S protein was observed in farmed minks and detected in 99% of SARS-CoV-2 sequences from minks [190]. Furthermore, Cai et al. [190] demonstrated that the N501T mutation occurred two months earlier in humans than in the minks, suggesting that the variant may have evolved in humans, and was then transmitted to farmed minks with subsequent sustained transmission among the highly susceptible mink population. Minks were suggested as an intermediate host for SARS-CoV-2 adaptation following the identification of a mink-associated variant in Denmark, thus supporting the rapid adaptation of the virus in this animal reservoir [190,191]. These findings were verified by subsequent virus transmissions from infected minks back to humans [192]. The mink mutant proved not to be too dangerous, as there was little evidence that it was more transmissible or more virulent in humans, nor that it could jeopardize the efficacy of therapeutics and vaccines [193]. However, as a precautionary measure, Denmark, the largest producer of mink fur, culled more than 17 million minks to control the spread of COVID-19 in minks and farm workers [193,194].
The human transmission of SARS-CoV-2 to farmed minks has been documented in the United States, Canada and many European countries [171]. Minks exposed to the virus presented mild to severe respiratory distress because of interstitial pneumonia [189]. The origin of the SARS-CoV-2 outbreak could be traced back to the farmers and their close contacts who tested positive for the virus and displayed COVID-19 symptoms [171]. In the United States, the rapid emergence of the N501T (Asn 501 → Thr) mutation in the RBD of the S protein was observed in farmed minks and detected in 99% of SARS-CoV-2 sequences from minks [190]. Furthermore, Cai et al. [190] demonstrated that the N501T mutation occurred two months earlier in humans than in the minks, suggesting that the variant may have evolved in humans, and was then transmitted to farmed minks with subsequent sustained transmission among the highly susceptible mink population. Minks were suggested as an intermediate host for SARS-CoV-2 adaptation following the identification of a mink-associated variant in Denmark, thus supporting the rapid adaptation of the virus in this animal reservoir [190,191]. These findings were verified by subsequent virus transmissions from infected minks back to humans [192]. The mink mutant proved not to be too dangerous, as there was little evidence that it was more transmissible or more virulent in humans, nor that it could jeopardize the efficacy of therapeutics and vaccines [193]. However, as a precautionary measure, Denmark, the largest producer of mink fur, culled more than 17 million minks to control the spread of COVID-19 in minks and farm workers [193,194].
The analysis of ACE2 protein conservation across species, the modelling of the S viral protein with ACE2 [195][196][197][198] and animal infection models [199][200][201][202][203][204][205][206][207][208][209][210][211] have highlighted the susceptibility spectrum of many animal species to COVID-19. Some livestock species displayed resistance to experimental infection with SARS-CoV-2 including cattle (Bos taurus) [212], and chickens, ducks and pigs [213,214]; however, as it takes only a few mutations or genetic recombination events for the virus to cross species and establish a new host, it is imperative that SARS-CoV-2 occurrence in domestic and farmed animals is monitored. A strategy to help mitigate or contain SARS-CoV-2 transmission could be the establishment of animal organoid cultures for screening the susceptibility of these animals to new SARS-CoV-2 variants.

Limitations of Drug Screens on Conventional Cell Lines
Hydroxychloroquine was one of the earliest drugs to be repurposed for COVID-19 treatment. In February 2020, Wang and colleagues were the first to report the effect of chloroquine as an antiviral treatment for SARS-CoV-2 infection [215]. They screened a variety of potential compounds in Vero cells in vitro for inhibition efficacy and cytotoxicity and showed that remdesivir and chloroquine were the best candidates. Shortly later, the same group demonstrated that hydroxychloroquine, a less toxic derivative of chloroquine, was as effective as chloroquine in inhibiting viral replication and infection [216]. While remdesivir performed better, there were many limitations for using remdesivir for COVID-19 treatment on a global scale. Remdesivir was still an experimental drug under patent protection of Gilead, which means the cost for treatment would be prohibitive. The manufacturing process is complicated, and the drug itself must be administered intravenously. Hydroxychloroquine, on the other hand, is widely and cheaply available. It can be taken orally and has been approved for long-term use in humans. These findings with hydroxychloroquine were confirmed by others using Vero and other cell lines, except in Calu-3 [217,218]. However, the urgency for a drug and political pressure led to human trials of hydroxychloroquine before pre-clinical testing was completed. Unsurprisingly, hydroxychloroquine failed clinical trial [219]. We now know that in mouse [220], hamster and nonhuman primate models [221,222] of COVID-19, chloroquine failed to protect against SARS-CoV-2 infection. Much time and money was wasted by this haste.
A detailed study of cellular factors necessary for SARS-CoV-2 entry ( Figure 2) using CRISPR/Cas9 gene editing in primary human epithelial organoids verified ACE2 and TMPRSS2 as essential factors for SARS-CoV-2 infection [33]. Infection was inhibited by the TMPRSS2 inhibitor camostat but not hydroxychloroquine, which inhibits receptormediated endocytosis. Furthermore, the Alpha variant of SARS-CoV-2 (B.1.1.7), SARS-CoV and MERS similarly depended on TMPRSS2 [33]. Had the early drug screens included primary epithelial cells, hydroxychloroquine would not have been considered a viable candidate for clinical trial at that time.
A further caveat to using Vero cells for antiviral drug screens and viral neutralization assays, which are considered by virologists as the "gold standard", is that, as already mentioned above, viral infection cannot induce an IFN-I response in these cells, and thus any effect of drugs or neutralizing antibodies may not translate to humans. Furthermore, Vero cells are not human-but monkey kidney-derived and may not predict toxicity to human cells.

Clinical Isolates and Viral Stocks
The authenticity of the cell model is paramount to ensuring the validity of pre-clinical tests, but so is the authenticity of the virus. As we have alluded to already, SARS-CoV, MERS and SARS-CoV-2 enter cells by docking to ACE2 followed by TMPRSS2-mediated cleavage of the S protein to initiate fusion. The fusion can be with the plasma membrane or from an endosome (Figure 2). Unlike SARS-CoV and CoV-MERS, SARS-CoV-2 has MBCS that facilitates TMPRSS2-mediated cleavage and, consequently, speed and efficiency of viral entry [80,223]. This is the preferred mode of entry into primary respiratory cells as these express high levels of TMPRSS2 but low levels of cathepsins [80]. SARS-CoV-2 stocks prepared in Vero cells can harbor mutations in or deletions of the MBCS [79,80,223,224] affecting the ability of the virus to infect human airway cells and viral transmission from infected to uninfected animals [225]. These mutations were rarely observed in clinical samples [226]. Propagation of SARS-CoV-2 in Calu-3 cells and primary airway epithelium preserves the authenticity of the S protein. Although Calu-3 is a cancer cell line and consequently far removed from primary airway epithelial cells, it expresses high levels of TMPRSS2 [80] and provides an intermediary renewable culture system for propagating viral stocks that, at least for the SARS-CoV-2 variants tested to date, preserves virus authenticity [83].

Accurate Pre-Clinical Models of SARS-CoV-2 Transmission and Respiratory Disease
ALI-HNE and matrix-embedded HNE organoids are proving to be excellent models of SARS-CoV-2 variant transmissibility [32,227,228]. Infection of ALI-HNE with an ancestral (Wuhan) clinical isolate [229] did not cause overt cytopathic effects, despite yielding a high titer virus from the apical surface. In stark contrast, infection with the Delta (B.1.617.2) variant induced large syncytia and extensive damage to the epithelium [32]. This difference between the viruses was corroborated by fusogenicity assays [228] and was attributed to the P681R mutation in the S protein of Delta [227]. Delta caused similar cytopathic effects in human alveolar organoids and in the lungs of hamsters, causing damage to the alveolar pneumocytes and inducing macrophage infiltration [227]. The fusogenicity and pathogenicity of Delta might underlie the increased transmissibility and rapid spread of Delta globally [230] and the severity of the illness caused.
Early studies with the most recent variant Omicron (B.1.529) in ALI-HNE show more rapid viral kinetics than Delta [231], which may underlie the increased transmissibility and rapid global spread of this variant [48]. However, compared to Delta, syncytia formation is diminished [52,231], as is pathogenicity in mouse and hamster models [232,233] and human lung alveolar organoids [52]. Reduced pathogenicity in human alveolar organoids might indicate decreased severity of disease in the real-world human experience with Omicron. Abdullah and colleagues [234] compared the 466 hospital COVID-19 admissions since 14 November 2021 to the 3962 admissions since 4 May 2020, prior to the Omicron outbreak, in the Omicron-driven fourth wave in the City of Tshwane, Gauteng Province, South Africa, its first global epicenter. COVID-19 disease severity was decreased in the Omicron wave. However, the extent that this decreased severity can be attributed to viral factors, as opposed to immunization status, for example, remains to be determined. Nonetheless, the organoid models were predictive of immune evasion by Omicron [52], the structural basis of which has recently been attributed to specific mutations in the S protein [235].
Collectively, the primary human respiratory organoids established from different regions of the respiratory tract offer insights into viral transmissibility and pathogenicity of SARS-CoV-2 variants as they emerge and the potential for immune evasion.

Conclusions and Perspectives
The SARS-CoV-2 pandemic has seen an unprecedented advancement in diverse fields: genomic epidemiology and modelling, vaccine and drug development, pre-clinical models and public health, to name a few. Additionally, it has revealed shortcomings in each of these as measures were introduced to control the pandemic. In the early phase of the pandemic, the focus was on the non-functional characteristics, such as genomics, phylogeny and evolution of SARS-CoV-2. It soon became evident that COVID-19 was more than a respiratory infection, and functional characteristics such as mechanisms of cell entry and cell tropism, host cell response to infection and the development of vaccines and drugs to treat COVID-19 became, and are still, paramount. Advances in our understanding of functional characteristics require authentic pre-clinical models.
The need to understand the underlying drivers of SARS-CoV-2 mutations, the clinical impact of emerging variants and the multi-organ clinical sequelae of severe COVID-19, has led to an exponential adoption of organoids from diverse tissues. Through organoid technology, we gained an understanding of potential viral entry points (nose, eye), a potential loss of epithelial barrier function to gain access to the brain/CNS (choroid plexus, olfactory bulb) and are beginning to understand the role of comorbidities (NASH). Early studies of autopsy tissue identified the organs harboring SARS-CoV-2. Tissue and region-specific organoids helped identify the target cells within the tissue of affected organs (e.g., lung, liver, kidney, brain and vasculature).
The limitations of classical virology tissue culture models were quickly revealed, which led to the use of more predictive organoid-based cultures in drug discovery and vaccine feasibility studies. The diverse human organoid pre-clinical models provide insight into transmissibility and pathogenesis. The next phase is organoid models with more complexity, incorporating components of the cellular and physical microenvironment of infected cells. Understanding the long-term effects of COVID-19, referred to as "long-COVID", for example, remains a challenge and will undoubtedly fuel further adoption and advancement of organoid models.
Author Contributions: All authors contributed to writing this review. All authors have read and agreed to the published version of the manuscript.