In the eye of the storm: SARS‐CoV‐2 infection and replication at the ocular surface?

Abstract Severe acute respiratory syndrome coronavirus 2 (SARS‐CoV‐2) first emerged in December 2019 and spread quickly causing the coronavirus disease 2019 (COVID‐19) pandemic. Recent single cell RNA‐Seq analyses have shown the presence of SARS‐CoV‐2 entry factors in the human corneal, limbal, and conjunctival superficial epithelium, leading to suggestions that the human ocular surface may serve as an additional entry gateway and infection hub for SARS‐CoV‐2. In this article, we review the ocular clinical presentations of COVID‐19 and the features of the ocular surface that may underline the overall low ocular SARS‐CoV‐2 infection. We critically evaluate the studies performed in nonhuman primates, ex vivo organ culture ocular models, stem cell derived eye organoids and the differences in infection efficiency observed in different parts of human ocular surface epithelium. Finally, we highlight the additional work that needs to be carried out to understand the immune response of the ocular surface to SARS‐CoV‐2 infection, which can be translated into prophylactic treatments that may be applied to other organ systems.


| INTRODUCTION
Since the discovery of yellow fever virus in 1901, more than 200 virus species that can infect humans have been discovered. Three to four new species are discovered every year, 1  were reported by officials in Wuhan, China. 4 This was soon followed by reports of clusters of cases within families 5,6 and infections of healthcare workers, 7 which indicated human-to-human transmission.
In January 2020, the pathogen named severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), belonging to the family of coronaviruses was identified. The virus spread quickly in both hemispheres and the World Health Organization (WHO) declared coronavirus disease 2019 (COVID-19) a public health emergency on the 30th January 2020 and subsequently a pandemic on 11th March 2020. As per 25th January 2021, 97 831 595 confirmed cases, including 2 120 877 deaths were reported globally.
Coronaviruses (CoVs) belong to the Coronaviridae family of order Nidovirales. They are classified into four genera: Alpha-, Beta-, Gamma-or Delta-coronavirus. The CoVs genome is 26-32 kb long, consisting of enveloped, single stranded positive-sense RNA that is 5 0 capped and 3 0 polyadenylated. 8 This is the largest RNA genome reported to date. 9 The CoV particles are spherical in shape with 50-200 nm diameter. In accordance with their parasitic nature, CoVs like many other viruses rely on host cells to complete their life cycle ( Figure 1). Following entry into the cells, the viral genome RNA is translated to generate all the proteins needed for RNA replication and transcription, a process, which requires involvement of several host proteins. 8 The CoVs RNA genomes are characterized by the presence of two large open reading frames (ORFs) 1a and 1b, which comprise more than 2/3 of the genome and encode 23 putative proteins including a 5 0 -replicase, spike (S), envelope (E), membrane (M), and nucleocapsid (N)-3 0 proteins ( Figure 1). The structural proteins S, E, M, and N are common to all known coronaviruses. 10 They function during host cell entry, viral morphogenesis, and virion release. The S protein is a membranous glycoprotein, which is essential for viral entry and defining host range, tissue tropism and virulence. M protein is localized at intracellular membrane structures and facilitates the change in the shape of the viral particle and its attachment to the nucleocapsid. The envelope protein (E) supports the manufacture and emission of particles. Lastly, the nucleocapsid protein (N) supports the replication of CoV genes by helping attach the genome to the replicationtranscription complex. The interaction between the M and E proteins and nucleocapsid results in budding through the membrane ( Figure 1). SARS-CoV-2 is a novel enveloped RNA Betacoronavirus2, which results in COVID-19 infectious disease and primarily affects respiratory tissues, although increasing evidence indicates that multiple other organs including heart, 11 kidneys, 12 gut, 13 liver, 14 and the neurological system 15 may be affected. SARS-CoV-2 is transmitted through zoonosis with a primary reproduction number of approximately 2.7, which indicates an exponential growth rate of infections. For symptomatic patients, clinical prognosis can be divided into three patterns: mild illness with upper respiratory tract symptoms, nonlife-threatening pneumonia, or severe pneumonia with acute respiratory distress syndrome (ARDS) that begins with mild symptoms for around a week, then progresses to necessitate advanced life support. 16 The newly sequenced SARS-CoV-2 genome is 29.9 kb and shares 82% identity with SARS-and MERS-CoV. 17 Phylogenetic analyses of 160 human SARS-CoV-2 genomes performed in March 2020 indicated three central variants, named A, B and C, which differ from each other by few amino acid changes. 18  Recent clinical evidence has suggested that COVID-19 may present with ocular symptoms as well as pathological changes in the ocular surface in a minority of patients. [21][22][23] Importantly, work published earlier this year by our group and others has shown the presence of SARS-CoV-2 entry factors ACE2 (angiotensin-converting enzyme 2) and TMPRSS2 (Transmembrane Serine Protease 2) in the superficial corneal, limbal and conjunctival epithelium, which led to suggestions of ocular surface serving as an additional entry gateway and potentially a replicative hub. 24,25 In this article, we review the COVID-19 ocular clinical presentations and the features of ocular surface, that may underline the overall low ocular SARS-CoV-2 infection. We critically evaluate the studies performed in nonhuman primates, ex vivo organ culture ocular models, stem cell derived eye organoids and the differences in infection efficiency observed in different parts of human ocular surface epithelium. Finally, we highlight the additional work that needs to be carried out to understand the immune response of the ocular surface to SARS-CoV-2 infection, which can be translated into novel prophylactic treatments that may be applied to other organ systems.

| COVID-19 AND OCULAR MANIFESTATIONS
The first report of ocular manifestation among COVID-19 patients was provided by a member of the national expert panel on pneumonia, who was infected during his inspection in Wuhan, despite wearing an N95 mask. 26 This expert did not wear an eye protection or face

Significance statement
The newly discovered SARS-CoV-2 responsible for the COVID-19 pandemic has been shown to affect many organs, including the lung, kidney, gut, and brain, among others. Recent work has shown that a small percentage of COVID-19 patients present with ocular symptoms. In the present study, the authors critically evaluate the growing literature on COVID-19 and ocular surface as well as the cell, organ, animal, and patient-specific models used to elucidate the impact of SARS-CoV-2 on the ocular surface. The authors' comprehensive review suggests that ocular symptoms are most likely encountered in the initial phase of infection, with conjunctiva showing the highest propensity for infection.
shield and several days prior to developing pneumonia, he reported redness of his eyes, which led to the suggestion that unprotected exposure of the eyes might have allowed the virus to infect the body. 26 In a case-series, 38 COVID-19 positive patients treated at a hospital in Hubei Province, China were retrospectively reviewed for ocular symptoms. 23 Of these patients, 31.6% had manifestations consistent with conjunctivitis, such as epiphora, chemosis, conjunctival hyperemia, or increased secretions. The study also reported that patients with ocular symptoms were more likely to have higher white blood cell, neutrophil counts, procalcitonin, C-reactive protein as well as lactate dehydrogenase than patients without ocular symptoms, which suggest that ocular symptoms were more likely to appear in patients with severe pneumonia. 23 However, limitations of this study were the relatively small sample size, sampling of one eye only and absence of detailed ocular examinations to exclude other ocular diseases. A larger study of confirmed COVID-19 patients indicated a range of ocular symptoms including itching, redness, tearing, discharge and foreign body sensation; however, only 3 out of the 121 patients yielded positive results from the conjunctival swabs and only one of these was among the eight identified as having ocular symptoms. 21 A low frequency of conjunctival congestion (0.8%) was also reported in the largest retrospective study of 1099 confirmed COVID-19 patients that were hospitalized in 552 hospitals across 30 provinces in China. 27 Similarly, a low frequency of viral RNA detection in tears and conjunctival secretions has been reported 28  To ascertain the impacts of COVID-19 on the ocular surface, a prospective observational study assessed 38 confirmed COVID-19 patients and 31 healthy controls. 30 While no significant differences were observed regarding age and gender between the two groups, conjunctival impression cytology revealed decreased density and enlargement of goblet cells, squamous changes, and increased presence of neutrophils in the COVID-19 patients. Together these data demonstrate that pathological alterations may be present in the ocular surface at the beginning of COVID-19 without significant ocular manifestations. Importantly, the neurological impact on vision has not been fully addressed. A retrospective study of the clinical manifestations of 214 patients with COVID-19 in three hospitals in Wuhan reported that 1.4% (3/214) had visual impairment. 31 The study did not elaborate on the nature of those visual impairments. 31 There were also two reports of patients diagnosed with COVID-19 after presenting with diplopia and ophthalmoparesis from cranial nerve palsy. 32 Overall, the frequency of viral detection in tears and conjunctival sample of COVID-19 patients published to date is low and mostly in the initial phase of the infection. Furthermore, when the ocular symptoms were observed, conjunctivitis seems to be the most common ophthalmologic sign related to coronavirus infection.

| The ocular surface
The ocular surface is commonly described as a continuous epithelium of the cornea, conjunctiva, and associated glands, which exudes and is bathed in the tear film ( Figure 2). All these components are required to act in concert to maintain a hydrated and protective barrier for the exposed surface of the eye. The tear film acts as the first line of defense, is complex and contains three layers secreted by different glands and tissues, that is, a superficial lipid layer, an aqueous component, and mucins. The lipid-rich secretion, which prevents evaporation of the aqueous middle layer, is secreted by the meibomian glands of the eyelids. 33 The lacrimal and accessory glands provide an aqueous lacrimal fluid of water, proteins, and electrolytes. The cornea and conjunctiva epithelial cells, conjunctival goblet cells and lacrimal gland produce the mucins and electrolytes, which constitute the innermost mucous layer of the tear film. 34,35 Transmembrane and soluble mucins interact to form the glycocalyx layer on top of the epithelia. The cornea and conjunctiva epithelia, through tight junctions, form a physical barrier and substrate for the tear film. And microvilli help anchor the tear film to the epithelia. 36,37 An integral element of the ocular surface functional unit is the lacrimal system. This has been described to include the main and accessory lacrimal glands, meibomian gland, cornea, limbus, conjunctiva, tear film, eyelids, nasolacrimal drainage duct, and interconnecting innervation. 38 The lacrimal gland itself not only secrets the aqueous tears (including water, electrolytes, protein, and mucus) maintaining a moist environment, but also secretes the microbicidal proteins lysozyme, lactoferrin, phospholipase, lacritin, immunoglobulins and cytokines, and via plasma cells, IgA to help protect against pathogens. 39 The tear film is drained via the efferent tear ducts through the nasal lacrimal passages formed of the lacrimal canaliculi, and sac, and nasolacrimal duct, linking the tear duct to the nasal passages, and into the surrounding cavernous vascular system. These passages are lined by epithelium with most epithelial cells presenting microvilli and some reports of motile cilia.
Goblet cells reside within this epithelium, sometimes grouping to form mucous glands. 40

| COVID-19, the ocular surface, and ocularsystemic transmission
As summarized above, SARS-CoV-2 infection of the ocular surface is observed at low frequency. 44 Currently the potential of the ocular surface as a mucous membrane to act as a site for infection, replication and transport of SARS-CoV-2 remains to be elucidated. As discussed, corneal and conjunctival epithelial cells with the appropriate receptor and protease may be able to act as a direct inoculation site from infected respiratory droplets ( Figure 3). However, the tear film and the fast drainage (approximately every 5 minutes) may provide a barrier for infection of the underlying epithelia. This may explain the scarce reports of conjunctivitis and absence of keratitis among COVID-19 infected patients. It has been postulated that the tear film, particularly the superficial lipid layer, may act as a barrier to prevent SARS-CoV-2 binding to the corneal/conjunctival epithelia entry receptors. Furthermore, the tear flow may provide an "ocular surface wash out" effect, preventing prolonged persistence of virus on the ocular surface. Nonetheless, if the virus makes its way to the ocular surface epithelium, through the tear film, then the latter through its lipid nature (and thus adherence of virus), tear flow and drainage may facilitate a second route of infection binding to receptors in and beyond the nasolacrimal system. 45,46 In addition, the conjunctival epithelium is known to secrete mucins that contain sialic acids, 47  route of infection has been tested in primates (see Section 3.4) resulting in a mild respiratory infection. 46 The rarity of conjunctivitis 22,28,29 and the infection of conjunctival explants 52 provide some evidence that the tear film is protective.
The role of the nasolacrimal ducts in linking the nasal cavityassociated lymphoid tissue with the ocular mucosal immune system contributes to the immunological interdependence between the ocular and respiratory systems. 47,53 This may be relevant given recent findings showing expression of SARS-CoV-2 entry genes in nasal epithelial cells together with innate immune genes. 54 The authors suggest the nasal epithelial cells may therefore have a role in viral infection, spread and clearance. 54 There have also been suggestions that cellular receptors in the system linking the ocular and respiratory systems may affect the tissue tropism of respiratory viruses, and proposals that mechanisms may exist which confer an ocular tropism to some viruses. 47 Thus, the nasolacrimal system as a route for migration of the  For the CoV genome to enter the host cell, the S1 unit of the spike protein needs to bind to a cellular receptor at the host cell surface ( Figure 1). Work published this year, indicated that the angiotensin converting enzymes 2 (ACE2) acts as a receptor for SARS-CoV-2. 25,57,58 In addition, viral entry also requires S priming by host proteases, which enable fusion of the viral and cellular membranes. 59 S priming is carried out by transmembrane protease serine type 2 (TMPRSS2). In the absence of proteases in the host cell surface, the virus can enter the cell through an endosomal pathway, which is at least 100 times less efficient than the primary route. 60 In this pathway, the S protein is instead activated by the endosomal cathepsin B/L. 60 A recent non-peer reviewed study, also suggests binding of the S protein to the CD147 cell surface receptor as a novel entry route for SARS-CoV-2 in human cells. 61 Accordingly, a recent clinical trial employing a humanized anti-CD147 antibody has shown a significant improvement in the case severity of critically ill patients. 62 Interestingly, CD147 is expressed on the cell surface of human corneal epithelium, stroma and endothelium and conjunctival epithelium. 63 Many epithelial tissues as well as various eye compartments (Table S2) have been shown to express ACE2 and TMPRSS2. In a recent study, 38 conjunctival samples from 8 healthy subjects and 30 patients were analyzed with bulk RNA-Seq sequencing for the expression of ACE2 and TMPSS2. 64 The authors reported that both ACE2 and TMPRSS2 mRNA and protein were not substantially expressed in the conjunctival samples and thus suggested that COVID-19 transmission via this pathway would be unlikely. 64 One limitation of this study is that 68% of the samples were collected from patients, with either melanomas, squamous cell carcinomas or papilloma, and so may not reflect the conjunctiva of most patients with COVID-19. Furthermore, the bulk RNA-Seq can only reflect the average gene expression; hence, if the SARS-CoV-2 entry genes were expressed in a relatively small fraction of cells in the conjunctival epithelium, they would be missed with this approach. These technical issues could also be an underlying factor for the findings presented by Ma and colleagues, who reported consistent expression of ACE2 in two out of three primary human conjunctival samples, but no detectable TMPRSS2 using RT-PCR. 65 In an effort to investigate at the single cell level the expression of ACE2 and TMPRSS2, single cell RNA-Seq studies have been performed indicating high ACE2 and TMPRSS2 expression in the upper respiratory system (nasal epithelial cells) and at lower level of expression in the superficial conjunctival epithelium. 54 In a subsequent study focused only on the human ocular surface, two groups including ours, reported co-expression of ACE2 and TMPRSS2 in the conjunctival, corneal and limbal epithelium 24,25 alongside a ubiquitous viral receptor heparin sulfate, which can facilitate the initial viral attachment. 66 The expression of ACE2 and TMPRSS2 in the superficial conjunctival and corneal epithelial surface was also corroborated by immunohistochemistry and Western blotting. 24,25 Importantly, we identified TNF, NFKβ and IFNG as upstream transcriptional regulators of ACE2 and TMPRSS2 in the superficial conjunctival epithelium, suggesting that pro-inflammatory signals may fine-tune the response of ocular surface epithelium to SARS-CoV-2 infection. 24 The presence of SARS-CoV-2 entry factors on the ocular surface is not the ultimate arbiter of infectability since the percentage of cells that express both entry factors may enhance or suppress viral entry into the eye. Our single cell RNA-Seq study published earlier this year, indicated that only 6.6% of cells of the superficial conjunctival epithelium express both entry proteins 24 ; hence, questions remain whether this is sufficient to enable robust SARS-CoV-2 entry into the ocular surface and moreover estab-  69 Published evidence suggests that lactoferrin prevents SARS-CoV binding to heparan sulfate proteoglycans. 70 If this was to prove true also for SARS-CoV-2, lactoferrin would provide a potential prophylactic solution for exposed surfaces such as the ocular, buccal, and nasal epithelium.

| Can SARS-CoV-2 infect the ocular surface?
The study of animal models that recapitulate human disease is an Rhesus macaques were also infected with MERS-CoV using a combination of intratracheal, intranasal, oral, and ocular surface routes. 72 The virus could be detected in conjunctival swabs by 3 days post infection, but it was no longer present by day 6, indicating that conjunctival testing should be done relatively early during the disease. using both types of explants as well as in vitro models of conjunctival and corneal epithelial surface is needed. In addition, work with pathological specimens is also necessary to understand the SARS-CoV-2 tropism in the human ocular surface. A recent study of 10 confirmed COVID-19 deceased individuals (median age 66, range 46-90 years old) indicated the positivity rate for SARS-CoV-2 to be 15% for conjunctiva, 5% for the anterior corneal surface and 15% for the vitreous. 74 In summary, the limited number of studies published to date indicate that SARS-Cov-2 may be able to infect the conjunctival epithelium, although the infection via the ocular surface route may result in a mild lung infection compared with inoculation via the trachea.
The concept of ocular surface infection is important not only for better prevention, detection, and treatment of COVID-19 patients, but also for testing the ocular tissues intended for transplantation. The above recent study performed by Sawant et al reported that of the 132 ocular tissues from 33-surgical donors, positivity rate for SARS-CoV2 was 13% (17/172). 74 The viral spike and envelope proteins were detected in the epithelial layers of the corneas that were procured without povidone-Iodine disinfection, strongly suggesting that the latter should become a norm for eye banks supplying corneas for transplantation purposes.

| STEM CELL MODELS FOR UNDERSTANDING SARS-CoV-2 ENTRY AND/OR REPLICATION IN THE HUMAN OCULAR SURFACE
Given the impact of COVID-19 on human health, the development of therapies directed against the virus itself or against the physiological consequences of SARS-CoV-2 infection, is an urgent and unmet need.
Numerous academic and commercial organizations are engaged in the search for both novel and re-purposed small molecule antiviral therapies, vaccines, anti-inflammatory agents, convalescent plasma infusions, natural killer cells and speculative cell therapy approaches. [75][76][77] To enable and inform these worldwide efforts, a thorough understanding of SARS-CoV-2 tropism needs to be gathered systematically in different tissues and organ systems. Tissues obtained from COVID-19 infected patients have been widely used to understand the pathology of the disease; however, in most cases these represent the end stage of infection and do not permit multiple experimental questions to be investigated in detail.
Hence, lab made tissue models which can be generated with ease are needed to obtain new insights into SARS-CoV-2 infection and the response of each tissue to the infection. Both adult and pluripotent stem cells differentiated to tissue specific lineages are being used to generate organoids of different types (brain, lung, gut, kidney, etc) 78 and cell monolayers (eg, pancreatic beta cells). 79 In this section we review how stem cells are and can be used to understand how SARS-CoV-2 infects various parts of the ocular surface.
A whole eye organoid approach from human pluripotent stem cells containing retina, retinal pigment epithelium, ciliary margin, lens, cornea and iris as well as cell monolayers grown from limbal, corneal and conjunctival epithelium were recently used to understand effectiveness of SARS-CoV-2 infection in the ocular surface. 80 The authors of this study found that limbus was the most susceptible to infections, with central cornea exhibiting very low levels of viral replication, which corroborates previous findings of corneas being refractory to infection. 73 Furthermore, the authors concluded that even though type I and III interferons were not detected, human eye organoids mounted an immune response indicative of the NFKB mediated proinflammatory cytokine response. Notwithstanding these interesting findings, the cell differentiation approaches used for the generation of human eye organoids resulted in a mix of epithelial cells from various parts of the ocular surface. Furthermore, the cell monolayers grown from the limbal, corneal, and conjunctival epithelium were not subjected to a directed differentiation protocol or analyzed in detail to assess whether they represented a basal or superficial epithelium, which may complicate data interpretation. As an alternative, we sug-

| SUMMARY AND FUTURE DIRECTIONS
In this review, we have shown that the incidence of associated ocular signs and SARS-CoV-2 detection in tears and conjunctival samples to date is relatively low and mostly observed in the initial phases of COVID-19 infection. We have also evaluated the work performed in explant and cell models, which show that conjunctival and limbal epithelium are infected by SARS-CoV-2, but the corneal epithelium is refrac- showing that corneas are refractory to infection. 81 This illustrates that this field of research is still in the early stage, with multiple groups relying on different but limited tissue sources to fully understand how SARS-CoV-2 enters, replicates and/or productively infects the ocular surface. The response of ocular surface to SARS-CoV-2 infection and its mounting immune response needs to be studied in detail, as important insights, which can be translated into prophylactic treatments may be gained. Last but not least, the role of tears in viral binding and transmission needs to be investigated as lactoferrin, a multifunctional protein implicated in the innate immune response, which is highly present in tears, has been shown to prevent SARS-CoV binding to heparan sulfate proteoglycans, 70 providing a potential prophylactic solution for exposed surfaces such as the ocular, buccal and nasal epithelium.
It is also important not to forget the short amount of time elapsed

CONFLICT OF INTEREST
The authors declared no potential conflicts of interest.

DATA AVAILABILITY STATEMENT
Data sharing is not applicable to this article as no new data were created or analyzed in this study.