SARS-CoV-2 disease severity and transmission efficiency is increased for airborne but not fomite exposure in Syrian hamsters.

Transmission of SARS-CoV-2 is driven by contact, fomite, and airborne transmission. The relative contribution of different transmission routes remains subject to debate. Here, we show Syrian hamsters are susceptible to SARS-CoV-2 infection through intranasal, aerosol and fomite exposure. Different routes of exposure presented with distinct disease manifestations. Intranasal and aerosol inoculation caused more severe respiratory pathology, higher virus loads and increased weight loss. Fomite exposure led to milder disease manifestation characterized by an anti-inflammatory immune state and delayed shedding pattern. Whereas the overall magnitude of respiratory virus shedding was not linked to disease severity, the onset of shedding was. Early shedding was linked to an increase in disease severity. Airborne transmission was more efficient than fomite transmission and dependent on the direction of the airflow. Carefully characterized of SARS-CoV-2 transmission models will be crucial to assess potential changes in transmission and pathogenic potential in the light of the ongoing SARS-CoV-2 evolution.

Introduction affects disease manifestation. In order to evaluate existing SARS-CoV-2 control measures it is 48 crucial to understand the relative contribution of different transmission routes. Because the 49 majority of cases have been observed in households or after social gatherings, transmission of 50 SARS-CoV-2 is believed to be driven mostly by direct contact, fomites, and short-distance 51 airborne transmission [6]. Airborne transmission can be defined as human-to-human 52 transmission through exposure to large droplets and small droplet nuclei that can be transmitted 53 through the air; whereas airborne transmission includes transmission through both large and 54 small droplets, true aerosol transmission occurs via droplet nuclei particles smaller than 5 µm. 55 Fomites are a result of infectious respiratory secretions or droplets being expelled and 56 contaminating surfaces. 57 In multiple hospital settings SARS-CoV-2 viral RNA has been consistently detected on surfaces 58 [7-12] and air-samples [8,9,[13][14][15][16][17][18][19][20]. Detection of infectious virus in air and surface samples has 59 been relatively limited, however infectious SARS-CoV-2 has been recovered from air samples accordance with the virological findings, no SARS-CoV-2 antigen was detected in the trachea or 113 lung of any fomite inoculated hamsters (N = 0/4). Viral antigen was detected in ciliated epithelial 114 cells of the nasal turbinates in one (N = 1/4) fomite inoculated hamster (Fig 2 c,g,k,m). No 115 SARS-CoV-2 antigen was detected in the esophagus or brain in any of the evaluated animals 116 (data not shown) nor in unexposed control tissues (Fig 2 d, h, l, m). 117 118 Fomite SARS-CoV-2 exposure displays delayed replication kinetics in the respiratory 119 tract and leads to less severe lung pathology 120 To determine the correlation between exposure route and subsequent respiratory tract 121 pathology, sections of lung, trachea and nasal turbinates were obtained for histopathological 122 evaluation at 1 and 4 DPI. Interestingly, nasal turbinate pathology was observed in a subset of 123 hamsters regardless of inoculation route at 1 DPI (Fig 3 a,  infiltration of rare or low numbers of neutrophils into the bronchiolar mucosa and focal interstitial pneumonia with minimal septal expansion by edema fluid and spillover of rare leukocytes into 139 the adjacent alveolar spaces (Fig 3 i, j, k). 140 By 4 DPI, infectious virus could be detected in the lung of all animals regardless of inoculation 141 route. No significant difference was observed between I.N. and aerosol or fomite exposed 142 animals (Fig 1 d; N = 4, ordinary two-way ANOVA, followed by Tukey's multiple comparisons 143 test, p = 0.4114 and p = 0.9201, respectively). An increase in the severity of both turbinate and 144 pulmonary pathology was observed in all evaluated hamsters regardless of the route of 145 inoculation. Interestingly, in both aerosol and I.N. inoculation routes, regions of olfactory 146 epithelium within the nasal turbinates were more severely affected, suggesting initial viral 147 attachment and replication in ciliated epithelium followed by targeting of the more caudal 148 olfactory epithelium during disease progression (Fig 3 m, n, o). At this timepoint, nasal mucosal 149 pathology was observed in all fomite inoculated animals. However, the pathology was less 150 severe as compared to I.N. and aerosol groups and focused primarily on regions of ciliated 151 mucosa, suggesting a delay in disease progression relative to aerosol and I.N. routes. Tracheal 152 inflammation was observed in all inoculation routes and varied from minimal to mild (Fig 3 q, r,  153 s). Moderate pulmonary pathology consistent with previously described SARS-CoV-2 infection 154 in Syrian hamsters [24] was observed in aerosol and I.N. inoculated animals at 4 DPI (Fig 3 u,  155 v) with less severe and less consistent pathology observed in the fomite inoculation group (Fig 3  156 w). Lesions were characterized as moderate, broncho-interstitial pneumonia centered on 157 terminal bronchioles and extending into adjacent alveoli. The interstitial pneumonia was 158 characterized by thickening of alveolar septa by edema fluid, fibrin and moderate numbers of 159 macrophages and fewer neutrophils. Inconsistent pulmonary pathology was observed for this 160 group with lesions ranging from minimal to moderate, which is in accordance with the 161 observation that some fomite exposed animals did demonstrate high viral loads in the lung at 4 162 DPI \ (Fig 3 w). No significant histopathologic lesions were observed in sections of mediastinal 163 and mesenteric lymph node, esophagus, duodenum, or colon, (data not shown) or any control 164 animal on 1 and 4 DPI (Fig 3 d , h, I, p, t, x). 165 Using a hierarchical clustering of lung pathology parameters (bronchiolitis, interstitial 166 pneumonia, tracheitis, pathology of the ciliated and olfactory epithelium) on both 1 and 4 DPI in 167 relation to the observed viral titers, a clear relationship existed between the respiratory 168 pathology at 1 DPI in the trachea, and viral load of trachea and lung, while pathology in the 169 nasal epithelial was more distantly related (Fig 3 y). Of note, viral load in the lungs at 4 DPI was 170 most closely associated with presentation of interstitial pneumonia. Fomite exposed animals 171 most closely resembled unexposed controls at 1 DPI and clustered together as a separate 172 group at 4 DPI due to the appearance of tracheitis, pathology in the ciliated epithelium without 173 distinct lower respiratory tract involvement (Fig 3 z). This implies that fomite SARS-CoV-2 174 exposure displays delayed replication kinetics in the respiratory tract and leads to less severe 175 lung pathology at 4 DPI compared to direct deep deposition of virus into the respiratory tract 176 To investigate the systemic immune response, cytokine specific ELISAs were performed on 181 serum at 4 DPI (Fig 4 a). Expression patterns were strikingly different depending on exposure 182 route for pro-inflammatory tumour necrosis factor (TNF)-α and anti-inflammatory IL-4 and IL-10. 183 Both I.N. and aerosol groups presented with increased levels of TNF-α at 4 DPI as compared to 184 unexposed animals, whilst the fomite exposed group demonstrated decreased levels; a 185 significant difference in serum levels was detected between I.N. and fomite exposed groups (N 186 = 4, Kruskal-Wallis test, followed by Dunn's multiple comparisons test, p = 0.0360). Adversely, 187 the IL-4 levels were markedly increased in all groups as compared to unexposed animals, yet 188 highest levels were seen in fomite exposed animals, the difference between unexposed and 189 fomite group reaching statistical significance (N = 4, Kruskal-Wallis test, followed by Dunn's 190 multiple comparisons test, p = 0.0109). Increased serum IL-10 was also observed in fomite 191 exposed animals and I.N. exposed animals, while a decrease was observed in animals after 192 aerosol exposure, resulting in a significant difference between aerosol and fomite exposed 193 hamsters (N = 4, Kruskal-Wallis test, followed by Dunn's multiple comparisons test, p = 0.0286). 194 While not significant, a trend of decreased serum levels of interferon (INF)-γ as compared to 195 uninfected animals, was observed. No significant differences were seen for serum levels of 196 interleukin (IL)-6. 197 Irrespective of exposure route, all exposed animals seroconverted at 14 DPI as seen by the 198 presence of antibodies targeting the SARS-CoV-2 spike measured by ELISA (Fig 4 b). The 199 magnitude of humoral response was linked to the exposure route. I.N. exposure resulted in the 200 strongest, and significantly higher antibody response when compared to fomite exposure (N = 4, 201 Kruskal-Wallis test, followed by Dunn's multiple comparisons test, p = 0.0209). No significant 202 difference was observed between I.N. and aerosol exposed animals. Taken together this 203 suggests a predominantly anti-inflammatory immune response is mounted after fomite 204 exposure, as compared to aerosol exposure, which may protect from more severe outcome, yet 205 is also linked to a weaker, but still substantial, antibody response. 206 207 Viral shedding is exposure route dependent 208 To gain an understanding of route-dependent virus shedding patterns of SARS-CoV-2 in the 209 Syrian hamster, daily oropharyngeal and rectal swabs were taken until 7 DPI, after which swabs 210 were taken thrice weekly (Fig 4 c, d). Oropharyngeal swabs are a measurement of respiratory 211 shedding while rectal swabs assess intestinal shedding. Viral sgRNA, a marker of virus 212 replication [27], was detected in both swabs from all exposed animals on at least one day. 213 When comparing the overall respiratory shedding profile between the exposure routes, different patterns were observed. I.N. inoculation resulted in high viral loads starting at 1 DPI and 215 continued up until 6 DPI, before sgRNA levels started to decrease. In the aerosol inoculated 216 group, the peak of virus shedding was reached on 2 DPI and viral sgRNA levels decreased 217 immediately thereafter. In contrast, animals exposed through the fomite route demonstrated 218 different shedding kinetics as compared to aerosol and I.N. groups with an increase in viral 219 sgRNA shedding over multiple days, until peak shedding was reached at 5 DPI. While a trend 220 seemed present for higher individual peak shedding in I.N. and fomite groups, no significant Aerosol exposure led to overall less viral RNA in oropharyngeal swabs as compared to I.N. and 230 fomite exposure (N = 4, Kruskal-Wallis test, followed by Dunn's multiple comparisons test, p = 231 0.0263). In contrast, most commutative viral sgRNA was detected in rectal swabs of aerosol 232 exposed animals (Fig 4 f). Taken together, these data suggest that severity of disease is not 233 indicative of the duration and cumulative amount of virus shed after infection. 234 235 Early shedding profile may predict disease severity and corresponding immune 236 response 237 As we observed different impacts on disease profiles between exposure routes, we next 238 investigated potential predictability of disease through early shedding patterns. Cytokine 239 responses as a measurement of the immune status (4 DPI) were included in the correlations between early shedding (2 DPI), peak shedding, peak weight loss, lung titers and pathology at 4 241 DPI (Fig 3 g). Lung viral titers were positively correlated significantly with the amount of viral 242 RNA detected in oropharyngeal swabs at 2 DPI (Spearman correlation test, N = 12, p = 0.047). 243 Lung titers showed a positive relationship with upper and lower respiratory tract pathology and 244 weight loss. This suggests that early time point respiratory shedding (before disease 245 manifestation) may predict the acute disease manifestation. 246 Serum levels of IL-4, IL-6 and IL-10 did not show any significant correlations with parameters of 247 disease severity; however, a clear negative relationship could be seen in the correlations. TNF-248 α, negatively correlated to IL-4 and IL-10 levels (Spearman correlation test, N = 12, p = 0.048 249 and p = 0.049, respectively). A positive correlation between early rectal shedding and TNF-α 250 serum levels and olfactory pathology was observed (Spearman correlation test, N = 12, p = 251 0.0002 and p = 0.001, respectively) (Fig 4g). 252 253

Airborne transmission is more efficient than fomite transmission in the Syrian hamster 254
To investigate viral fomite contamination of caging, daily swabs were taken from surfaces in 255 cages containing one I.N. inoculated hamsters, up to 7 DPI ( Sup Fig 1 b, c). Viral gRNA was 256 detectable at 1 DPI in all samples, sgRNA was detectable for 7/8 (87.5%) bedding samples and 257 3/8 (37.5%) cage samples, and at 2 DPI in 8/8 cages for both samples. Viral sgRNA was 258 detectable at high concentrations up until 7 DPI, with peak concentrations seen on 2 and 3 DPI, 259 suggesting a robustly contaminated caging environment. 260 To assess the potential risk of fomite transmission, we introduced sentinel hamsters to cages 261 after housing two I.N. infected animals for 4 days. (Fig 5 a). No signs of disease or weight loss 262 were observed in sentinel animals, but seroconversion was seen in 4 out of 8 animals (Fig 5 f) 263 at 21 days after exposure (DPE) to a contaminated cage, confirming that hamster-to-hamster 264 indirect transmission via fomites can occur (Fig 5 h). 265 Next, the efficiency and dynamics of airborne hamster-to-hamster transmission were assessed. 266 For this purpose, we designed a cage divider, which allowed airflow but no direct contact or 267 fomite transmission between animals. (Fig 5 b, Fig 5 c, d, and supplemental video). We used a 268 particle sizer to assess the effect of the cage divider on blocking particle flow. We observed that 269 cross-over of smaller particles (<10 µm) was blocked approx. 60%, whilst larger particles (>10 270 µm), were reduced over 85% on the sentinel side (Fig 5 d, e). 271 In the first experiment, one sentinel hamster was placed on the side of the divider downflow 272 from one infected animal (N = 8). In contrast to animals exposed directly to aerosolized virus, no 273 signs of disease or weight loss were observed in any of the sentinel animals (Fig 5 g). However, 274 all animals seroconverted. To assess the importance of directional airflow, airborne 275 transmission was also modeled for 4 transmission pairs housing the sentinel against the airflow 276 (Fig 5 b, c). Only one out of 4 of the sentinels placed against airflow seroconverted (Fig 5 g), 277 suggesting, as expected, that directional airflow is key to airborne transmission. When

283
Titers were comparable to those observed after direct inoculation. Together, this 284 suggests that hamster-to-hamster airborne transmission may present with 285 asymptomatic disease manifestation, yet the humoral immune memory is comparably 286 robust.

287
To investigate the transmission risk posed by animals after fomite or airborne transmission, the 288 respiratory shedding profile was determined. Viral shedding was demonstrated in 4 out of 8 sentinel hamsters after exposure to contaminated cages on multiple consecutive days. 290 Shedding was observed at 1 DPE, with peak viral sgRNA being seen at 4/5 DPE, like what was 291 observed in hamsters directly exposed to fomites (Fig 4 c). For airborne transmission, sentinels 292 downstream of airflow started shedding by 1 DPE, and all 8 animals had high amounts of viral 293 sgRNA in the oropharyngeal cavity by 2 DPE, which remained high until 6 DPE. This data 294 suggest that this indirect exposure route presents with a distinctly different disease 295 manifestation and shedding profile than direct aerosol exposure (Fig 5 i). Of note, commutative 296 viral shedding between infected airborne exposed animals showed no difference to those 297 infected through fomite transmission (Fig 5 j). These data imply that, whilst presenting with no or 298 very mild disease phenotypes, both routes of indirect exposure between animals create a 299 mimicry of asymptomatic carriers. 300 301 Discussion 302 SARS-CoV-2 transmission is driven by close proximity, confined environment, and the 303 frequency of contacts [28]. Infection with SARS-CoV-2 is believed to be driven by direct contact, 304 inhalation of virus within respiratory droplet nuclei, contact with droplet contaminated surfaces or 305 any combination between these exposures. Yet, the relative contribution of each of the potential 306 routes of exposure in relationship to human-to-human transmission has been elusive. Moreover, 307 the relationship between exposure route and the differential impact on disease severity has 308 been equally obscure. Animal models are essential to model experimental transmission under 309 Surprisingly, we demonstrate here that fomite transmission may still occur (4 out of 8) when 358 peak shedding of infectious virus has waned as previously shown [25], and environmental 359 contamination is expected to be reduced. Importantly, this implies that even with an increased 360 understanding of airborne transmission involvement at this stage of the pandemic, the risk of 361 fomite transmission in humans should not be underestimated. In particular, fomite transmission 362 may be more likely to occur in nosocomial settings that present a combination of fomite and 363 aerosol generating procedures and may potentially be further enhanced with more susceptible 364 hospital population [48,49]. 365 Within our transmission set-up we show a selective reduction of largest particles (>10 µm), but 366 that this exclusion was not absolute (Fig 5). Therefore, we cannot formally distinguish between 367 true aerosol transmission (droplet nuclei < 5 µm), droplet transmission (> 10 µm), or a 368 combination of these two. Previous studies have shown that SARS-CoV-2 can be transmitted 369 through the air in a ferret model over short and moderate distance [50,51] and in hamsters over 370 short distance [25,52]. In our study we were able to show a high efficiency of airborne 371 transmission with 100% of the sentinels becoming infected. When reversing the airflow from 372 uninfected animals toward infected animals, a sharp reduction in transmission was observed. 373

This suggest that directional airflow plays an important role in the transmission of SARS-CoV-2. 374
This has also been observed in human-to-human transmission events, where transmission in 375 confined spaces (e.g. restaurant) was directed by airflow [16,53,54]. Control measures focused 376 on strategically designed room ventilation will directly aid the control of the pandemic [55,56]. 377 In this study, we showed the relative contribution of airborne and fomite transmission and the 378 impact of exposure route on disease. The hamster transmission model will be crucial to assess 379 the transmission and pathogenic potential of novel SARS-CoV-2 strains, in the light of the 380 continuing SARS-CoV-2 virus evolution [57]. In addition, this work will allow the development of 381 effective public health countermeasures aimed at blocking human-to-human transmission. The 382 findings of this study suggest that using more natural routes of transmission are highly suitable 383 for accurately assessing the transmission potential and pathogenicity of novel evolved strains 384 [57]. Additionally, these data strongly suggest that the Syrian hamster model would be very 385 . Briefly, non-anesthetized hamsters were exposed to a single exposure whilst 417 contained in a stainless-steel wire mesh cage. Aerosol droplet nuclei were generated by a 3-jet 418 collision nebulizer (Biaera technologies, USA) and ranged from 1-5 µm in size. Respiratory 419 minute volume rates of the animals were determined using the methods of Alexander et al. [59]. Airborne transmission was examined by co-housing hamsters (1:1) in specially designed cages 435 with a perforated plastic divider dividing the living space in half. This divider prevented direct 436 contact between the donor/primary infected and sentinel hamster and the movement of bedding 437 material. Regular bedding was replaced by alpha-dri bedding to avoid the generation of dust 438 particles. Donor hamsters were infected intranasally as described above and sentinel hamsters 439 placed on the other side of a divider afterwards. Hamsters were followed as described above 440 until 21 DPI. Experiments were performed with cages placed into a standard rodent cage rack, 441 under normal airflow conditions (Fig 5 c, d, e). Sentinels were either placed in the direction of 442 the airflow, or against it (Fig 5 b). 443 444

Fomite Transmission experiments 445
Fomite transmission was examined by infecting donor hamsters as described above by I.N. 446 inoculation. Two animals per cage were housed for 4 days. Regular bedding was replaced by 447 alpha-dri bedding to avoid the generation of dust particles. At 4 DPI, donors were euthanized, 448 and sentinel animals (2 animals per cage) were placed into the contaminated cage (Fig 5 a). 449 Hamsters were followed as described above until DPI 21; bedding and cages were left with a standard spray bottle through the cage inlet. The particle size range of the generated 457 particles was measured using a Model 3321 aerodynamic particle sizer spectrometer (TSI). The 458 cage was coated with two sprays at an interval of 30 seconds (s) and after a third spray the 459 sample port was opened, and a sample was analyzed. The cage was sprayed every 30 s and 460 five samples were analysed (5 runs, each 60 s) for both donor side (primary infected side) and 461 sentinel side. 462 463

Histopathology and immunohistochemistry 464
Necropsies and tissue sampling were performed according to IBC-approved protocols. Tissues 465 were fixed for a minimum of 7 days in 10% neutral buffered formalin with 2 changes. Tissues 466 were placed in cassettes and processed with a Sakura VIP-6 Tissue Tek, on a 12-hour 467 automated schedule, using a graded series of ethanol, xylene, and ParaPlast Extra. Prior to 468 staining, embedded tissues were sectioned at 5 µm and dried overnight at 42°C. Using