Recovery of anosmia in hamsters infected with SARS-CoV-2 is correlated with repair of the olfactory epithelium

Severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) is responsible for a pandemic affecting billions of people worldwide. Apart from the extreme global economic impact, the pandemic will likely have a lasting impact through long-term sequelae not yet fully understood. Fully understanding the mechanisms driving the various symptoms and sequelae of SARS-CoV-2 infection will allow for the eventual development of therapeutics to prevent or treat such life-altering symptoms. In this study, we developed a behavioral test of anosmia in SARS-CoV-2-infected hamsters. We find a moderately strong correlation between the level of anosmia and the score of histological damage within the olfactory epithelium. We also find a moderately strong correlation between the level of anosmia and the thickness of the olfactory epithelium, previously demonstrated to be severely damaged upon infection. Thus, this food-searching behavioral test can act as a simple and effective screening method in a hamster model for various therapeutics for SARS-CoV-2-related anosmia.

www.nature.com/scientificreports/ to assess olfactory function directly. This simple behavioral test, with the results correlated to OE damage, has significant translational potential; therapeutic or preventative efficacy can be determined much more quickly.

SARS-CoV-2 challenge and olfactory behavioral test in hamsters.
All hamsters inoculated intranasally with 10 5 50% tissue culture infectious dose (TCID 50 ) of SARS-CoV-2 survived infection, featuring only a mild, transient weight loss and no development of fever ( Fig. 1) as previously described. SARS-CoV-2-inoculated hamsters were tested for anosmia at 2, 3,5,8,17,21,35, and 42 dpi. Mock-inoculated hamsters were tested for a baseline at 2 dpi and again at 42 dpi. The behavioral test consisted of burying a Teddy Graham cookie in one corner of a cage and measuring the time required for a hamster to find the cookie. Hamsters tested at 2, 3, and 5 dpi required significantly increased time to find the hidden cookie in comparison to the mock group (Fig. 2). Of note, some infected hamsters did not find the cookie within the maximum time provided in this study (300 s), and they did not show interest in the cookies, even when placed directly in front of them. Although not significantly different, the times to find the cookie of the remaining groups gradually trended shorter towards that of mock group, indicating a slow recovery of the sense of smell.
Histopathological analysis of olfactory epithelium. After behavioral testing, lung and nasal samples were collected for viral titration and histology, respectively. As expected, virus was detected from lungs with a high titer at 2 and 3 dpi, and decreased at 5dpi, and then virus was cleared from the lungs by 8 dpi (Fig. 3). Histology was performed on coronal nasal samples (Fig. 4) in order to quantify damage to the OE using two separate approaches: scoring of gross histopathological damage (Fig. 5) and direct measurement of OE thickness (Fig. 6). A marked increase in histological scores were seen in all four turbinate regions through 8dpi (Fig. 5). OE thickness was significantly reduced through 17 dpi for the nasal septum (S) and lateral turbinate (LT), and 8 dpi for the medial turbinate (MT) (Fig. 6). OE thickness was not significantly impacted in the dorsal turbinate (DT) (Fig. 6).
To determine whether correlations exist between the histopathological change in the OE and the time to find the hidden cookie, we checked Pearson's correlation between the time to find the cookie, the OE thickness, and the histology score in the OE (Table 1). We see a significant, moderately strong positive correlation between the time to find the cookie and the histology score. We also see a significant, mildly negative correlation between the time to find the cookie and the OE thickness.  www.nature.com/scientificreports/

Discussion
SARS-CoV-2-associated anosmia is perhaps one of the most well-known symptoms of COVID-19. While our previous work indicates that the anosmia is likely temporary, due to the gradual recovery of the severely damaged OE 16 , the long-term consequences of this anosmia remains unknown. Additionally, whether therapeutics can be developed to treat those suffering anosmia remains to be seen. Syrian golden hamsters are an excellent animal model through which to screen vaccines and therapeutics for SARS-CoV-2 infection 13 . Not only does infection consistently produce a transient acute phase of disease featured by a mild weight loss, but the hamsters themselves are outbred and immunocompetent, making the model much more relevant to COVID-19 in humans. Previous histology studies focusing on the OE damage within these hamsters requires intensive and time-consuming techniques. Moreover, as human histological samples are difficult to obtain, a behavioral test that reproduces the large amounts of human anosmia data is desirable. Thus, the objective of this study was to further develop a simple behavioral test to assess olfactory function directly to aid in the screening process of anosmia therapeutics. As expected, our SARS-CoV-2-inoculated hamsters developed a transient weight loss with viral titers cleared from the lungs by 8 dpi (Figs. 1 and 3). This is consistent with previous studies, but one additional explanation for why the hamsters lose weight may be that they are simply unable to smell their food due to the significant OE damage (Fig. 6). This weight loss paired with the increase in time to find the hidden cookie (Fig. 2), the marked pathology in the nose, and the significant thinning of the OE emphasize the effectiveness of the Syrian golden hamster as an anosmia model for COVID-19.  www.nature.com/scientificreports/ One of the main objectives of this project was to assess the olfactory function directly by using a behavioral test for the confirmation of anosmia in a hamster COVID-19 model. Ideally, the time taken for each hamster to find the hidden cookie would correlate with OE damage; longer searching times would correlate with OE damage. We performed a more intensive correlation test, using Pearson's correlation test (Table 1). We found a significant and moderately strong positive correlation between the time taken to find the cookie and the histology score for each turbinate region. This is to be expected, as the visible histological damage requires at least 8 dpi to be cleared, by which the behavioral results are no longer significantly different. Additionally, we see a significant and moderate negative correlation between the time to find the cookie and the OE thickness of each turbinate region. Again, this result was expected as regeneration of the OE thickness should be related to the recovery of the sense of smell. Overall, based on our histology scores and thickness in each OE region, the nasal septal region (S) is the most consistently correlated with the olfactory behavior tests. This indicates S may be the most suitable region for future histopathological analyses to asses OE damages induced by SARS-CoV-2 infection. It www.nature.com/scientificreports/ is important to note that the receptor for SARS-CoV-2, angiotensin-converting enzyme 2 (ACE2), may not be equally expressed between the nasal turbinate regions, accounting for the differences in severity of damage seen upon histology. Previous work indicates that the septal and dorsal regions may express higher levels of ACE2, rendering these zones more susceptible to SARS-CoV-2 infection 17 . Additionally, nasal turbinate scrolls are possibly more physically difficult for the virus to reach. However, as the nasal septal region is physically supported well by other supporting tissues and thicker bone structures, it is more resistant to acquired damage during sample preparation, as compared to the other regions. Thus, we propose that the nasal septal region will be a good target for future histopathological analysis of OE damage in COVID-19 patients. While we find that histological damage moderately correlates with olfactory function, previous work using a chemically induced anosmia has indicated that over 90% OE destruction is required before anosmia occurs 18 . This strengthens our claim that a behavioral test is more desirable to detect functional changes during screening of vaccines and therapeutics than relying solely on histology and provides further insight into a potential mechanism.
Previous research indicates that SARS-CoV-2-related anosmia may be due to infection of non-neuronal cells within the OE; most olfactory sensory neurons (OSNs) do not express ACE2 17,19 . These studies indicate that anosmia may be linked to infection of the sustentacular cells supporting the olfactory system, resulting in a rapid and severe deciliation of the OE 19,20 . Without cilia to detect odorants, the resulting anosmia occurs 20 . According to Butowt and von Bartheld, this is the most likely mechanism for this anosmia 21 . Based on our results, this deciliation mechanism is also likely. Comparing our results to that of others, there is a similar timeline of damage and regeneration. Significant anosmia occurs between 2 dpi through 5 dpi, followed by a gradual trend of recovery, with hamsters regaining seemingly normal olfactory function by 35 dpi (Fig. 2). Similarly, previous work in a Syrian golden hamster model demonstrates peak deciliation occurring as early as 2 dpi with OE recovery and functional cilia appearing back around 14 dpi 20 . Further work should be conducted directly examining any correlations between the ciliation of the OE and olfactory function, as measured by using this food behavioral test.
In the case of hamsters that did not find the cookie within the maximum time frame (300 s), we confirmed that hamsters were unable to detect the cookie by either directly handing it to them or by placing it immediately in front of them and watching for their reaction. All hamsters that had significantly higher search times were www.nature.com/scientificreports/ unable to find, or were completely uninterested in, the cookies in both of these situations. This fact helps to emphasize that these hamsters indeed developed severe anosmia. Thus, this study indicates that this anosmia behavioral test is suitable for direct assessment of the olfactory function. Moreover, this behavioral test can be used as a fast and simple indicator of the efficiency of vaccines or therapeutics against SARS-CoV-2 infection in the hamster model. This test will contribute to the speedy development of therapeutics to help aid those suffering from anosmia related to COVID-19.   (2,3,5,8,17,21,35, and 42 dpi), four hamsters were tested for any signs of anosmia. Testing occurred in the afternoon at approximately the same time each testing day. An empty housing cage was prepared with at least 3 cm of bedding. In one corner, a honey-flavored Teddy Graham (Nabisco) was buried about 1 cm below the surface of the bedding. In the corner diagonal to the buried cookie, a hamster was placed, the lid closed, and a timer started. The timer was stopped once the hamster had revealed and grasped the cookie, and the time recorded. All hamsters were given a maximum of 5 min (300 s) to find the cookie.

Viral titration.
Collected lung samples were homogenized with DMEM supplemented with 2% FBS to make a 10% homogenate. These samples were then diluted tenfold serially and inoculated into Vero E6 cells on 96-well plates. Infected cells were allowed to incubate for 72 h at 37 °C with 5% CO 2 , then fixed with 10% formalin and stained with 0.25% crystal violet to visualize the presence of cytopathic effect (CPE). TCID 50 values were then calculated and recorded using the Reed and Muench method 23 .
Histological analysis. At the time of sampling, collected muzzles were fixed in 10% buffered formalin for 7 days before removal from the BSL-3 facilities. Samples were prepared for histological analysis as previously described 16 . Briefly, olfactory bulbs and nasal tissue including the olfactory epithelium (OE) were extracted from the skull and decalcified with EDTA (10% w/v) and embedded in paraffin. Thin sections with 5 μm thickness were mounted on glass slides and stained with Hematoxylin and Eosin. To analyze the OE, coronal sections of the OE were divided into four areas along zonal organization as previously described 16 and two independent analyses were performed examining the histopathology of the OE for each group. Each of the four turbinate zones were examined and assigned a pathology score: 0-no damage (normal), 1-mild damage (damage reaches only through epithelial layer; basal layer remains untouched), 2-moderate damage (up to 75% of damage reaches basal cells), or 3-severe damage (over 75% of damage reaches basal cells). Scores were averaged between the two reviewers. To investigate the changes in OE thickness, two independent measurements were performed for the four different nasal turbinate regions as previously described 16 .