key: cord-0748779-n3tfu6mu
authors: Hoagland, Daisy A.; Møller, Rasmus; Uhl, Skyler A.; Oishi, Kohei; Frere, Justin; Golynker, Ilona; Horiuchi, Shu; Panis, Maryline; Blanco-Melo, Daniel; Sachs, David; Arkun, Knarik; Lim, Jean K.; tenOever, Benjamin R.
title: Leveraging the antiviral type-I interferon system as a first line defense against SARS-CoV-2 pathogenicity
date: 2021-01-29
journal: Immunity
DOI: 10.1016/j.immuni.2021.01.017
sha: ac8c2bd4b28bc873b65d66b4832b380c6891c6a6
doc_id: 748779
cord_uid: n3tfu6mu

The emergence and spread of SARS-CoV-2 has resulted in significant global morbidity, mortality, and societal disruption. A better understanding of virus-host interactions may potentiate therapeutic insights toward limiting this infection. Here, we investigated the dynamics of the systemic response to SARS-CoV-2 in hamsters by histological analysis and transcriptional profiling. Infection resulted in consistently high levels of virus in both the upper and lower respiratory tracts and sporadic occurrences in other distal tissues. A longitudinal cohort revealed a wave of inflammation including a Type-I interferon (IFN-I) response that was evident in all tissues regardless of viral presence, but was insufficient to prevent disease progression. Bolstering the antiviral response with intranasal administration of recombinant IFN-I reduced viral disease, prevented transmission, and lowered inflammation in vivo. This study defines the systemic host response to SARS-CoV-2 infection and supports use of intranasal IFN-I as an effective means of early treatment.

economies (Eisinger and Fauci, 2018) . Four major influenza A virus pandemics in the last century and seasonal epidemics have accounted for significant morbidity and mortality (Krammer et al., 2018) . Countless numbers of emerging and re-emerging pathogens, including Dengue, Ebola, Nipah, and Zika viruses, have caused global outbreaks as the climate changes and zoonotic incidences of transmission increase. As evidenced by the societal disruption caused by Severe Acute Respiratory Syndrome

Coronavirus 2 (SARS-CoV-2), we remain inadequately prepared for these disease outbreaks, despite the many advances of biomedical research.

All organisms have the capacity to respond to virus infection. CRISPR-based antiviral defenses of archaea and bacteria work through the acquisition of viral genetic material by the host that is then utilized as a template to guide a nuclease to other copies of the virus genome, causing its degradation (Barrangou et al., 2007) . Similarly in plants, arthropods and worms, Dicer proteins process double-stranded RNA from viruses into small genetic fragments that enable the RNA-induced silencing complex to bind and cleave virus-specific transcripts (Guo et al., 2019) . In contrast to these sequencespecific genetic defenses, vertebrates utilize a relatively non-specific pathogen detection surveillance system as a first-line defense to inhibit virus replication (tenOever, 2016) . In vertebrates, cellular recognition of virus infection culminates in the transcriptional induction of a family of cytokines, the Type-I and -III interferons (IFN-I and IFN-III), which initiate a cascade of events that ultimately renders cells less permissive to virus infection (Lazear et al., 2019; Mesev et al., 2019; Park and Iwasaki, 2020) . Factors (IRFs) (Maniatis et al., 1998) . Co-activation of NFκB and members of the IRF family (IRF1, IRF3, and IRF7) causes the transcriptional activation of Interferon beta (IFNB), a IFN-I, which is then transcribed, translated and secreted, signaling in both an autocrine and paracrine manner to invoke an antiviral state. IFN-I binding to its cognate cell surface receptor induces activation of a transcription factor complex termed Interferon Stimulated Gene Factor 3 (ISGF3), composed of STAT1, STAT2, and IRF9 (Stark et al., 2018) . ISGF3 is responsible for the enhanced expression of at least 150 genes which exert direct antiviral effects (Schneider et al., 2014; Schoggins and Rice, 2011) . This early IFN-I response limits virus replication and cell-to-cell spread and aids in coordinating the pathogen-specific adaptive immune response and the development of immune memory. Not unexpectedly, viruses have co-evolved to specifically block the IFN-I response, precisely because of their critical roles in viral clearance (Garcia-Sastre, 2017) . This is evident for all viruses that cause significant morbidity, including smallpox, measles, influenza, mumps, rubella, and ebola. Indeed, SARS-CoV-2, the etiological agent for the COVID-19 pandemic, encodes in its genome several IFN-I antagonists (Konno et al., 2020; Lei et al., 2020; Miorin et al., 2020; Thoms et al., 2020; Xia et al., 2020 ). An interesting feature of SARS-CoV-2 infection is that the virus modulates the immune response in a way that suppresses the IFN-I system but induces and sustains high levels of chemokine mRNAs, resulting in an imbalanced host response (Blanco-Melo et al., 2020; Qin et al., 2020; Zheng et al., 2020b) . This juxtaposition of low IFN-I and high chemokines may be responsible for the enhanced infiltration of neutrophils and monocytes to the respiratory tract, resulting in COVID-19 (Coperchini et al., 2020; Zheng et al., 2020a) .

Here we sought to comprehensively characterize the host response to SARS-CoV-2 by tracking disease systemically from the initial exposure of virus in order to better understand the biology of COVID-19. SARS-CoV-2 infection resulted in a wave of inflammation that culminated in a systemic response that included distal organs such as the brain and GI tract. Lung pathology upon SARS-CoV-2 infection in hamsters was J o u r n a l P r e -p r o o f similar to what has been described amongst COVID-19 patients and demonstrated lasting transcriptional changes that extended even after virus has been cleared (Carfi et al., 2020; Weerahandi et al., 2020) . Lastly, the administration of intranasal IFN-I reduced viral load and tissue damage, suggesting this approach may prove to be an effective early intervention towards respiratory disease caused by SARS-CoV-2.

To assess the relative host response to SARS-CoV-2, we infected the golden hamster between mock infection (PBS) and these two respiratory viruses ( Figure 1A ). Aligning captured reads from each animal to both the host and viral genomes demonstrated comparable viral loads between IAV and SARS-CoV-2 at the time of analysis suggesting any differences in the host response were due to virus biology ( Figure 1B ).

To identify the unique aspects of the host response to SARS-CoV-2, we identified differentially expressed genes (DEGs) comparing IAV and SARS-CoV-2-infected cohorts and assessed all host genes with a log2 fold change greater than two to identify enriched biological processes ( Figure 1C and Table S1 ). SARS-CoV-2 induced a greater inflammatory signature than that of IAV, exemplified by high mRNA levels of Ccl5, Gzmb, and Il1rn, consistent with what has been observed in both patients and the ferret model (Blanco-Melo et al., 2020) . Moreover, expression of the neutrophil markers Fcgr3, Ccrl2 and Cf2 were significantly increased in response to SARS-CoV-2 infected hamsters as compared to IAV infection, consistent with what has been reported amongst COVID-19 autopsies (Table S1 ) (Skendros et al., 2020) .

In evaluating these data, it also became apparent that, like many immune-related genes, Ifnb1 was not annotated in the golden hamster genome. Given this, we next performed a de novo build on these mRNA-Seq data sets and furthermore a BLASTn J o u r n a l P r e -p r o o f search for conserved Ifnb1 genomic sequences shared by other rodents. We identified a transcript that was 85% similar to Ifnb1 at the nucleotide level with the Chinese hamster (Cricetulus griseus) ( Figure S1A-B) . At a protein level, this same transcript was more closely related to Ifnb1 from the white-footed mouse (Peromyscus leucopus) showing a 65% identity match ( Figure S1C ). To ensure this gene was functional, we cloned the cDNA and expressed it in golden hamster fibroblasts (BHK-21 cells) and demonstrated that both transfection and transfer of supernatant resulted in robust induction of host ISGs in hamster, but not human-derived, cell lines ( Figure S1D -E). In aligning the Mesocricetus auratus interferon beta 1 transcript (MaIfnb1), we found no significant induction in response to either SARS-CoV-2 or IAV-infected animals ( Figure   S1F ). Taken together, these data suggest that SARS-CoV-2 induces high chemokine expression in the context of a muted IFN-I response.

To examine how different routes of infection influence disease progression, we exposed naïve hamsters to SARS-CoV-2-contaminated fomites, infected animals, or direct inoculation via the eye or nose ( Figure 2A ). First, we verified that our viral stocks of Figure 2D ). Despite inducing a similar host response, as measured by Isg15 and Irf7, increasing inoculum doses to as high as 10^5

pfu not only failed to increase overall replication but tended to result in lower levels of sgN and nsp14 ( Figure 2D -E and S2B-C). To determine the reliability of intranasal infection in this outbred animal model, we infected 12 hamsters and found that lung viral titers, as measured by plaque assay, were consistently 10^6 pfu at four days post infection, with just a single outlier ( Figure 2F ). Reproducibility was further corroborated by qRT-PCR for sgN and a concomitant host response as measured by Isg15 mRNA ( Figure 2G -H).

In an effort to minimize any confounding factors that may arise with a non-physiological viral challenge, we assessed lung pathology in response to inoculation with 100 pfu.

Lungs were harvested at days two, four, six, eight and fourteen post infection and tissue sections were stained with antibody for viral N, and hematoxylin and eosin (H&E) ( Figure 3A -B). The most robust N protein expression appeared at four days post infection and occurred predominantly in the epithelial cells comprising the air interface of the lung ( Figure 3A ). Despite N protein levels diminishing dramatically beyond four days post infection, viral protein was still evident at fourteen days, consistent with pathological assessment of the H&E-stained tissue sections. In agreement with earlier reports, epithelial degeneration in bronchioles and alveolar necrosis was apparent, as well as type-II pneumocyte hyperplasia and cytopathy, peaking at four and eight days post infection, respectively ( Figure 3C ) (Imai et al., 2020) . There was also evidence of alveolar, perivascular, and bronchial inflammation, peaking between four and eight days post infection ( Figure 3D and S3A). Inflammation, as defined by cumulative histopathological score, was maximal on day eight post infection ( Figure 3E and Table   J o u r n a l P r e -p r o o f S2). Moreover, atypical adenomatous hyperplasia of the alveoli was observed at fourteen days post infection, even after the majority of inflammation had subsided ( Figure 3B , Table S2 , Figure S3B ). Further analysis of H&E staining of lung tissue at day three post infection revealed several distinct phenomena. We observed apoptosis in the bronchial epithelium, as well as nested accumulations of neutrophils in the bronchioles of infected hamsters ( Figure 3F -G). Furthermore, severe vascular edema was observed, even in sites of the lung with relatively little inflammation ( Figure 3H ).

Taken together, these data highlight the value in this small animal model, as an unmodified clinical strain of SARS-CoV-2 causes progressive infection in the lower respiratory tract ultimately resulting in severe pneumonia as is observed in COVID-19

patients.

to the lower respiratory tract.

We next examined the transcriptional response in the respiratory tract of hamsters infected with 100 or 10,000 pfu, respectively, over the same time course in which we had observed progressive lung pathology. Total RNA from lungs was assayed by qRT-PCR for sgN and host Isg15 levels, at the indicated days post infection ( Figure S4A -B).

In contrast to the rise, plateau, and clearance of subgenomic viral replication, which was maximal between days two and four post infection following challenge with 100 pfu, a 10,000 pfu challenge resulted immediately in high RNA levels ( Figure S4A ). This was followed by a general decline of viral transcripts through day six post infection ( Figure   S4A ). Furthermore, there was an increase and decline of the host response to infection with the lower dose inoculum of virus, as measured by Isg15 levels, that mirrors the kinetics of viral replication. For the higher dose virus inoculum, Isg15 levels remained elevated through day eight post-infection, after viral RNA was largely cleared ( Figure   S4B ). Viewed together, these data suggest that the amount of viral inoculum can influence the host response to infection and should be considered when analyzing data that are extrapolated to human disease.

To further characterize the full antiviral response to infection, we performed mRNA-Seq on a longitudinal study and assessed the levels of a curated list of ISGs in both trachea and lungs on the indicated days, at the same low and high viral challenge doses ( Figure   4A -B and S4C). The host response to SARS-CoV-2 correlated to viral load as defined by virus reads per million in the trachea ( Figure 4A ). However, despite the elevated transcription of many ISGs, the host response was insufficient to contain virus replication within the upper respiratory tract, culminating in high viral load in the lower respiratory tract on day four post infection ( Figure 4B ). This response to virus was different when the viral challenge was increased by two orders of magnitude, where we observed no delay in ISG production in neither the trachea nor the lung ( Figure 4A -B).

Regardless, the host response was unable to restrict viral propagation. These same data also demonstrate that irrespective of inoculum dose, SARS-CoV-2 infection appears to limit the induction of many of the classical ISGs, in agreement with reports from different studies (Konno et al., 2020; Lei et al., 2020; Lucas et al., 2020; Thoms et al., 2020; Miorin et al., 2020) .

In addition to inducing an IFN-I response, infected cells activate other transcriptional programs aimed at reducing viral load. DEGs from both trachea and lung were enriched for transcripts involved in an inflammatory response, which included the recruitment of the adaptive immune system as determined by the enrichment of relevant gene ontology (GO) annotations ( Figure 4C -D, Table S3 ). In the trachea, the induction of genes associated with a cellular response to virus, to stress, and cytokines all mirrored virus levels, generally peaking two to six days post infection and dissipating thereafter ( Figure 4C ). Gene enrichment analyses suggested significant changes to cellular composition as denoted by the induction of transcripts involved in cilium assembly, microtubule bundle formation, and cytoskeleton organization, which is consistent with the H&E-staining that suggested that SARS-CoV-2 may cause loss of cilia or ciliated cells ( Figure 4C and S4D). Similar gene enriched pathways were observed in infected lungs, although these tended to be sustained for the first eight days of disease progression ( Figure 4B and 4D). By day fourteen, SARS-CoV-2 infection in both the trachea and the lungs was resolved, as measured by a lack of DEGs beyond the J o u r n a l P r e -p r o o f regeneration of cilia ( Figure 4C -D, and Table S3 ). Taken together, these findings outline the transcriptional response to disease progression in SARS-CoV-2 infected hamsters and may be used as a resource to better understand the molecular biology underlying the antiviral response in COVID-19 patients.

SARS-CoV-2 infection is capable of disrupting aspects of IFN-I biology, but can also lead to hyper-inflammation (Manson et al., 2020) Given this, we generated a curated list of inflammatory cytokines and compared the same transcriptional data sets from our longitudinal study ( Figure 5A -B). Consistent with our earlier data, infection initiated with high inoculum dose resulted in rapid induction of host inflammatory cytokines, in contrast to the response accompanying a low pfu challenge. Furthermore, the peak of the host response in trachea preceded that observed in lungs, by approximately four days ( Figure 5A -B). For example, the most robust induction of host genes in response to SARS-CoV-2 was chemokines, which are responsible for orchestrating lung immune cell infiltration and the respiratory distress that is the hallmark of COVID-19 (Tay et al., 2020; Vabret et al., 2020) . In response to a viral challenge of 100 pfu, levels of Ccl4

and Ccl5 peaked on day two in the trachea, in contrast to the lungs, where peak levels were not achieved until day six ( Figure 5A -B, S5A). The potent induction of these chemokines together with Cxcl10 and -11 suggested recruitment of activated T cells and macrophages to the respiratory tract. This was supported by an observed increase in Xcl2 transcripts that likely originated from T cells or NK cells present in the tissue in addition to the increased expression of Cd8a, Cd4, and Cd3g ( Figure 5A -B, S5B-C).

Ccl2, Ccl7, Ccl8 and Itgam ( Figure S5C ). In addition, cytokine and chemokine expression levels were not always equivalent between the upper and lower respiratory tracts. For example, Tnf and Il1b were induced to high levels in the upper respiratory tract but were generally absent in the lower airways. In contrast, Il36a, Cxcl6, and Ccl22

reached higher levels of expression in the lungs compared to the trachea ( Figure 5A -B).

These differences likely reflect the heterogenous composition of these tissues and J o u r n a l P r e -p r o o f differences in immune cell subsets that were recruited to these sites, given that overall viral levels were comparable in these airway compartments at different times during infection.

Aligning this longitudinal RNA-Seq data against Ifnl and Ifnb1 revealed that, at a high virus inoculating dose, detectable levels of both IFNs were only present in the trachea one day post infection, with no reads for either transcript in the lung or any other tissue at any time post infection ( Figure 5C -D). Comparable to the ISG profile, cytokine expression also decreased by day fourteen, which correlated with the peak of virus spike protein-specific IgG antibodies in serum, previously shown to elicit protection against re-infection ( Figure 5A -B and 5E) (Imai et al., 2020; Tostanoski et al., 2020) .

Given the extensive viral replication and consequent host response to SARS-CoV-2 infection observed in hamsters, we next examined tissues distal to the respiratory tract.

Specifically, we examined SARS-CoV-2 infection in the olfactory bulb, the four lobes of the brain (frontal, parietal, temporal, and occipital lobes, herein denoted simply as brain), and small intestine, but found inconsistent evidence for virus replication, as measured by viral reads per million which were orders of magnitude lower than the Figure 4 , Figure S6A ). However, despite the low levels of viral reads respiratory tract ( in tissues outside of the lungs, a strong antiviral transcriptional response was evident, as measured by enrichment score for IFN-I signaling ( Figure 6A -C). In brain tissue, the proportion of DEGs was highest at two days post-infection, which correlated with the peak viral load observed in the upper respiratory tract ( Figure 6D , S6B). Genes associated with response to virus, response to cytokines, IFN signaling, and innate immunity appeared prior to the peak differential gene expression in the lower respiratory tract, suggesting that inflammation, or possibly infection (led by chemokine spread or virus), spread bi-directionally from the trachea. Additional gene signatures in the brain included those associated with translation initiation, nonsense-mediated decay, and mRNA metabolism, suggesting a distinct transcriptional response that in some cases was sustained for the duration of the experiment. Consistent with evidence for inflammation in the absence of consistent infection, we observed the most sustained IFN response in the olfactory bulb ( Figure 6B , 6E, S6B). In addition to classical ISGs, gene signatures for both a stress response and an adaptive immune response were evident, which may account for acute olfactory impairment (anosmia) reported in both mild and severe cases of COVID-19 in both humans and hamsters (Cantuti-Castelvetri et al., 2020; Carignan et al., 2020; Imai et al., 2020) . The most modest inflammatory response in the tissues assessed was in the small intestine, in which a mild antiviral gene signature was observed, which correlated with the time of peak viral load in the respiratory tract; similar etiologies have been reported in a subset of COVID-19 patients (Yang and Tu, 2020) .

As many of the transcriptional signatures observed would require detection of pathogenassociated molecular patterns (PAMPs), we next performed more sensitive assays to determine if low levels of virus replication could be detected as reported by others (Imai et al., 2020; Tostanoski et al., 2020) . To this end, we subjected the same RNA from the mRNA-Seq data set to qRT-PCR for sgN. Using this assay, we were able to amplify a PCR product after fewer than 40 cycles in olfactory bulb, brain, and small intestine at days one, two and four post infection ( Figure 6G ). Moreover, sgN transcript could be detected as far out as eight days post infection in the olfactory bulb, suggesting the presence of replicating virus, albeit at 1/200 th of the level of lung at four days post infection. To corroborate these results, we assessed each aforementioned tissue for infectious particles by plaque assay of tissue homogenate at three days post infection.

We found that sgN transcript levels did not consistently correlate with infectious particles at the level of detection allowed by plaque assay ( Figure 6H ). For example, in the olfactory bulb, ~400pfu per 0.1mg of tissue were detected only in a single animal, yet sgN remained consistent at a Ct score of ~28 ( Figure 6H ). Together with the apoptosis observed in the bronchial epithelium and vascular edema in the lungs of infected hamsters, these data suggest the dissemination of virus-derived PAMPs may be responsible for the systemic inflammation observed across tissues ( Figure 3F , 3H, and Figure 6 ). Further studies will be needed to substantiate these results, but if these J o u r n a l P r e -p r o o f data accurately reflect the inconsistencies of tissue tropism and inflammation in COVID-19 patients, they could point to the underpinnings of diverse clinical manifestations.

The elevated ISGs observed across all investigated tissues may reflect the capacity of IFN-I to limit tissue tropism. To determine whether IFN-I could provide a prophylactic benefit against COVID-19, we decided to test it intranasally in our hamster model. To this end, we first evaluated murine and human IFNβ alongside commercially available universal IFN (IFNα A/D) in the BHK-21 hamster cell line and found IFNα A/D to be the most potent ( Figure S7A ). To examine antiviral efficacy in vivo, we administered IFNα A/D to hamsters intranasally and assessed the transcriptional response in the respiratory tract by mRNA-Seq and further corroborated this data by qRT-PCR ( Figure   7A , S7B). A robust IFN-I response was induced, comprised of canonical ISGs including:

Isg15, Mx1-2, Ifit2-3, Cxcl10-11 and Oas1-3, amongst others (Table S4) fewer neutrophils compared to their controls; increased mitosis was also observed in the bronchial epithelium, suggesting these hamsters were closer to disease resolution ( Figure 7M ). In conjunction with increased Il10 expression at day three and decreased interleukin-6 expression at day six, these data suggested that the introduction of exogenous intranasal IFN-I resulted in a modified immune cell infiltrate, resulting in lower viral load.

To further ascertain the prophylactic value of IFNα A/D, we applied it to a more clinically relevant transmission model where treated animals were exposed by direct contact rather than being infected intranasally. To this end, animals were treated with either PBS or IFNα A/D and co-caged with infected animals for ten hours ( Figure S7E ). In agreement with our direct infection model, IFNα A/D given prophylactically in this transmission model resulted in a greater than two log reduction of viral titers in infected animals and prevented transmission altogether in three of the five animals ( Figure S7E ). Figure 7F-G) . These data could be further corroborated by measuring the levels of sgN mRNA from total lung ( Figure S7H ).

As treatment with IFN-I may be prohibitive for various economic and geographical reasons, we next examined whether a comparable antiviral response could be achieved by intranasal administration of a PAMP that would induce endogenous IFN-I. The advantages of using a PAMP, such as a mimetic for dsRNA (poly I:C), include ease of manufacturing and storage, as well as affordability. Poly I:C administration intranasally showed comparable antiviral activity to IFNα A/D ( Figure S7I ).

In response to the global pandemic of COVID-19 there have been unprecedented efforts to develop therapeutic interventions, both in the sphere of vaccines and antivirals. Different strategies have been employed to target the SARS-CoV-2 virus, that include direct inhibition of viral replication by targeting the RNA-dependent RNA polymerase, protease, S protein, and/or blocking interactions with the ACE2 entry receptor and TMPRSS2 in conjunction with studies further defining the intricacies of these interactions (Dieterle et al., 2020; Douangamath et al., 2020; Han et al., 2020; Sorokina et al., 2020; Tay et al., 2020) . In addition, efforts are underway to repurpose FDA-approved drugs that may show therapeutic value in vivo (De Clercq and Li, 2016; Si et al., 2020) . In an effort to advance general countermeasures against viral outbreaks, our strategy has been to develop broad-spectrum antivirals rather than pathogen-specific interventions for each emerging and re-emerging global virus outbreak. Our approach focuses on exploiting the host response to combat virus infections, namely enhancement of the IFN-I response.

We found that the host response to SARS-CoV-2 in hamsters, albeit insufficient to protect against severe bronchopneumonia induced by the virus, is characterized by active IFN-I signaling. Other studies have identified a similar transcriptional response in human ACE2 (hACE2)-transgenic mice, in mice transduced with hACE2-encoding J o u r n a l P r e -p r o o f adenovirus-based vectors or by using a mouse-adapted strain of SARS-CoV-2 (Dinnon et al., 2020; Israelow et al., 2020; Sun et al., 2020; Winkler et al., 2020) . Additionally, we are able to show that this response extends beyond the respiratory tract in several organs distal to the site of infection for up to fourteen days. While it has been suggested that imbalanced IFN-I is in part responsible for the severe inflammation and aberrant neutrophil infiltration, IFN-I is also necessary to control virus titer in both hamsters and mice (Boudewijns et al., 2020; Sun et al., 2020) . Given this effect of the IFN-I response to limit viral replication, we decided to leverage recombinant IFN-I as a prophylactic and a therapeutic against SARS-CoV-2. IFN-I and/or -IIIs represent ideal candidates for inhibiting viral infection. In addition to their pleiotropic effects in directly targeting multiple stages of the virus replication cycle, IFNs also activate an appropriate immune response to clear virus, regardless of the pathogen (Wang and Fish, 2019) . It was recently demonstrated that IFN-I signaling, mediated by STAT2, can limit viral replication and dissemination in the hamster model while IFN-III signaling appeared less important. Moreover, it has also been demonstrated clinically that both the presence of auto-antibodies against IFNs and inborn errors of innate immunity, such as deficiencies in TLR3 and IRF7, are enriched in patients with severe COVID-19 pneumonia, suggesting the potential value of exogenous IFN treatment in certain patients whose own IFN response is impaired or obstructed (Bastard et al., 2020; .

However, administering IFN-I systemically has proven to have very low concordance because it can cause unwanted side effects. This dynamic likely contributed to the inability of IFN-I to reduce mortality in the SOLIDARITY trials (Dinnon et al., 2020; Pan et al., 2020) . To avoid these potential adverse effects, we administered IFN-I locally to the respiratory tract and achieved two-three log reduction in viral burden and kept the virus below the limit of detection in certain experimental settings. Furthermore, a recent phase-two clinical trial of inhaled, nebulized interferon beta-1a to treat COVID-19 patients resulted in decreased time to recovery (Monk et al., 2020) . Altogether, this supports intranasal delivery of IFN-I as a potential prophylactic and therapeutic against COVID-19.

The outcomes of SARS-CoV-2 are heterogeneous, with the greatest dichotomy observed between young people and those of advanced age (Williamson et al., 2020) .

While this disparity certainly has multiple underlying factors, numerous reports have implicated a misfiring of the antiviral response as being a major contributor to the severity of disease (Blanco-Melo et al., 2020; Hadjadj et al., 2020; Lucas et al., 2020; Yao et al., 2020) . As observed in both cell culture and animal models, it is clear that the host response to SARS-CoV-2 results in an early aberrant IFN response juxtaposed to an overproduction of chemokines (Blanco-Melo et al., 2020; Hadjadj et al., 2020; Lucas et al., 2020) . In the hamster model, SARS-CoV-2 infection results in the induction of an insufficient IFN-I response as it spreads from the upper to the lower respiratory tract.

While it remains unclear whether the heightened immunity of these tissues prevents virus infection and therefore restricts tropism in this hamster model, this would be consistent with the fact that non-respiratory organs that express ACE2 often show an absence of virus even in severe COVID-19 cases (Ziegler et al., 2020) . In any event, the combination of high pro-inflammatory cytokines and chemokines to low IFN-I levels is likely the molecular basis for the neutrophil nests observed in the lower respiratory tract. This conclusion is supported by the fact that exogenous administration of IFNα A/D reduces neutrophil presence in the lung.

We also observe that the inflammatory response initiated at the earliest site of replication, the upper respiratory tract, may be sufficient to induce systemic inflammation in distal tissues, where mature virions are present at vanishingly low levels if present at all, despite the robust detection of viral RNA. However, there have been reports of central nervous system invasion in SARS-CoV-2 PCR-positive deceased individuals (Meinhardt et al., 2020) . While it remains unclear what the source of the inflammatory response is in these distal tissues, and whether the source of inflammation may differ between cases of mild, moderate and severe disease, it is tempting to speculate that the secretion or spillover of subgenomic material due to its abundance, perhaps bound by N protein to form a protected ribonucleoprotein complex, may be a prevalent PAMP that is disseminated from the primary site of replication. In support of this idea, we find that detection of sgN by qRT-PCR does not always correlate with the J o u r n a l P r e -p r o o f detection of infectious virions. This concept warrants further study but has been suggested in the past for unrelated RNA viruses (tenOever et al., 2002; tenOever et al., 2004 ).

Here we present compelling evidence that intranasal administration, or induction of IFN-I, provides protection against SARS-CoV-2 infection, whether for prophylaxis or for early therapeutic use. These data reflect the inhibitory potential of IFN-I and identify the opportunity for intranasal administration of IFNs as broad-spectrum antivirals for SARS-CoV-2 and, potentially, other respiratory virus infections.

The transcriptional host response during early infection with SARS-CoV-2 and after viral clearance has remained largely undefined, partially due to limitations in clinical settings.

Here we define this response in young golden hamsters, which are all able to effectively clear virus after disease course. This comes with its own limitations, including the lack of representation of older animals and its extrapolation to human disease. Age of the animals and severity of disease may alter the antiviral host response and/or tissue 

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Benjamin R. tenOever (benjamin.tenoever@mssm.edu).

Plasmids generated in this study are available upon request.

The RNA-sequencing datasets generated during this study are available at NCBI GEO (GSE161200). The sequence of the Mesocricetus auratus Interferon Beta 1 gene annotated in this study is available on GenBank (MW017682). Additional Supplemental Items are available from Mendeley Data at http://dx.doi.org/10.17632/hstgdwvz9f.1

Golden hamsters, ranging in ages from 3-5 weeks, were purchased from Charles River Figures 1-6, 7A -H, and S2-S7. Seven male hamsters and three female hamsters per group were used to generate the data in Figure 7I -J. One male hamster and three female hamsters per group were used to generate the data in Figure 7K -M. For Figure   5E assessing in vivo Spike antibodies, male hamsters 8-9 weeks of age were used.

Vero E6 

Plaque assays were performed using Vero E6 cells on 12-well plates. Virus was diluted logarithmically in SARS-CoV-2 infection medium for a total volume of 200µl inoculum per well. Plates were incubated for one hour at room temperature, rocking plates every 10 minutes. An overlay of Modified Eagle Medium (Gibco) 4 mM L-glutamine (Gibco), 0.2% BSA (MP Biomedicals), 10 mM HEPES (Fisher Scientific), 0.12% NaHCO 3 , and 0.7% Oxoid agar (Thermo Scientific) was mixed and pipetted into each well. Plates were then incubated at 37°C for 48 hours and then fixed in 5% formaldehyde in PBS for 24 hours before removal from the BSL3 facility. Plaques were stained with crystal violet solution (1% crystal violet (w/v) in 20% ethanol (v/v)) for 15 minutes. Plaque assays of lung homogenates were frozen once before plaque assay was performed; plaque assays of other tissue homogenates were plaqued prior to any cryopreservation.

Before intranasal infection, hamsters were anesthetized by intraperitoneal injection with were also administered in a 100µl volume resuspended in PBS. In direct contact transmission experiments, index hamsters were intranasally infected with 1000pfu of SARS-CoV-2 and caged for 24 hours. Sentinel hamsters were then co-housed with one infected hamster in a clean cage until the experimental endpoint. In fomite transmission experiments, two hamsters were infected with 1000pfu SARS-CoV-2 and housed in cages for two days. They were then removed, and a single naïve hamster was moved into the cage for four days before the experimental endpoint. For ocular infections, 6000pfu of SARS-CoV-2 was administered in the eye (5µl in each eye) of anesthetized hamsters. Hamsters subjected to mRNA-Seq or histology were perfused with 60mL icecold PBS before harvest. A segment of the upper small intestine closest to the stomach was used for small intestine analysis. When whole brain was harvested from hamsters, J o u r n a l P r e -p r o o f the olfactory bulb was sectioned off for separate RNA extraction and analysis. The brain was cut in half longitudinally after removing hindbrain and half was used for RNA extraction and analysis.

Hamster tissues were homogenized in Lysing Matrix A homogenization tubes (MP Biomedicals) for 40 seconds at 6.00m/s for 2 cycles in a FastPrep-24 5G bead beating grinder and lysis system (MP Biomedicals). All tissues for mRNA-Seq analysis were homogenized directly in 1ml TRIzol (Invitrogen). If hamster lung samples were being plaqued for SARS-CoV-2 titer, they were homogenized in 1ml PBS. Homogenates were then centrifuged for five minutes at 10,000g. Supernatant was immediately frozen for later analysis by plaque assay. The cell pellet was resuspended in TRIzol and rehomogenized as above before RNA extraction. RNA was isolated by phenol/chloroform extraction according to manufacturer's instructions.

RNA was DNase treated using DNA-free DNA removal kit (Invitrogen). 1µg of total RNA was used for each SuperScript II Reverse Transcriptase reaction with oligo d(T) primers (Invitrogen). Resultant cDNA was diluted 1:20 before real-time quantitative polymerase chain reaction (RT-qPCR) analysis. RT-qPCR reaction was performed using KAPA SYBR® FAST qPCR Master Mix (2X) Kit (Roche) on a LightCycler 480 Instrument II (Roche, Cat# 05015243001). Relative mRNA and viral RNA levels were quantified by normalizing sgRNA to actin expression and normalizing lung RNA levels to mockinfected controls.

1µg of total RNA was enriched for polyadenylated RNA species and prepared for shortread next-generation sequencing using the TruSeq Stranded mRNA Library Prep Kit (Illumina) according to the manufacturer's instructions. Sequencing libraries were sequenced on an Illumina NextSeq 500 platform. Fastq files were generated using bcl2fastq (Illumina) and aligned to the Syrian golden hamster genome (MesAur 1.0, J o u r n a l P r e -p r o o f ensembl) using the RNA-Seq Alignment application (Basespace, Illumina). Differential gene expression was determined using the edgeR protocol in the Bioconductor package . Samples from the same tissue were analyzed together. Genes that did not have a sufficiently large count in a given tissue as described in were excluded from the analyses. Highly expressed genes in certain samples such as genes of viral origin were taken into consideration through normalization of the library sizes to an effective library size (Robinson and Oshlack, 2010) . Differential gene expression between control group and experimental groups were tested for using a negative binomial model with extended with quasi-likelihood methods (QLF) . All genes with a false discovery rate (FDR) < 0.05 were classified as differentially expressed genes. Statistical significance of enrichment of custom gene sets were determined using fry from the edgeR package on Bioconductor (Wu et al., 2010) .

Number of reads mapping to the viral genome (GenBank: MN985325.1) was performed using bowtie2 (Langmead and Salzberg, 2012) . For DotPlot analyses, significantly induced genes in each sample (log2 fold change > 1, FDR < 0.05) were used as query to search for enriched biological processes (BP) using STRING (Szklarczyk et al., 2019) . A reduced set of enriched gene ontology (GO) terms (FDR < 0.05) was obtained using REVIGO (allowed similarity = 0.5) (Supek et al., 2011) . Selected enriched GO terms were visualized by Dot plot using ggplot2 (H. Wickham. ggplot2: Elegant Graphics for Data Analysis. Springer. 2016).

The identification of the interferon beta gene from golden hamsters was achieved by de novo assembly using MEGAHIT (https://github.com/voutcn/megahit) (Li et al., 2015) . In brief, RNA derived from the lungs of hamsters treated with poly I:C were used to generate an Illumina mRNA library. Fastq files were then inputted into MEGAHIT for de novo assembly. Assembled contigs generated as a result were subsequently used as input for a BLASTx search against human IFNB1 protein (NP_002167.1). This search resulted in a single hamster contig that showed 46% identity to human IFNB1 (see Figure S1 ). This gene annotation has been deposited to GenBank (accession no. MW017682). The sequence was synthesized as a gBlock (IDT) and cloned into a J o u r n a l P r e -p r o o f doxycycline-inducible TetOne vector (Takara). Functionality was confirmed by transfecting the construct into BHK-21 cells cultured in normal growth medium supplemented with doxycycline 10ng/ml. Transfection was performed using Lipofectamine 2000 (Invitrogen) following manufacturer's instructions.

For H&E staining, lung tissues were aseptically harvested following whole-body perfusion and fixed in 10% neutral buffered formalin for 24 hours. Tissues were then embedded in paraffin and sectioned into 8 micron thick sections onto charged slides. 

The ELISA protocol was adapted form established protocol (Amanat et al., 2020) . 96well plates (Immulon 4 HBX, Thermo Fisher) were coated with 50µl of 2µg/ml J o u r n a l P r e -p r o o f recombinant RBD in PBS at 4°C overnight. After overnight incubation, the coating solutions were removed, and the plates were blocked with 100µl of 3% non-fat milk (americanBio) prepared in 0.1% Tween 20 containing PBS (PBS-T) for 1 hour at room temperature. To reduce the risk of containing live virus, serum samples were heated 1 hour for 56°C before use and was handled in a BSL-3 facility. Serial dilution of serum samples and dilution of secondary antibody was done in 1% non-fat milk prepared in PBS-T. After blocking, solutions were removed and 100µl of diluted serum was added following 2 hours incubation at room temperature. After the solutions were removed, the wells were washed with 250µl PBS-T three times. 100µl of 1:7500 diluted anti-Syrian hamster IgG HRP (ThermoFisher) secondary antibody was added and incubated for 1 hour at room temperature. After the solutions were removed, the wells were washed with 250µl PBS-T three times. Once the wells were completely dry, 100µl of SIGMAFAST o-phenylenediamine dihydrochloride (Millipore Sigma) solution was added and the reaction was stopped with 30µl of 3M HCl. The optical density at 490 nm was measured and the concentrations of the antibody were analyzed by area under curve (AUC) using Prism 8 (GraphPad).

Statistical significance was determined by Student's t test or Mann-Whitney test when data points did not follow normal distribution. In the case of three experimental groups, one-way ANOVA was applied, and the p-value was corrected for multiple comparisons using Dunnett's test. Statistical tests were performed with Prism 8.0 (GraphPad) software. For all RNA sequencing analyses, statistical significance of differential gene expression between groups was determined through the edgeR pipeline (see transcriptome analysis, methods section). For all data, a p-value < 0.05 or a false discovery rate < 0.05 was considered to be statistically significant. Statistical details for each experiment including statistical tests applied, p-values and number of replicates can be found in the figure legends.

J o u r n a l P r e -p r o o f Table S1 . Differentially expressed genes in SARS-CoV-2 infected hamster lungs compared to lungs infected IAV. Related to Figure 1 . All statistically significantly differentially expressed genes between SARS-CoV-2-and IAV-infected hamster lungs.

Genes are shown as ensembl gene ID next to known golden hamster gene name and human orthologs. Expression of each gene is shown as mean count, log2 fold change and adjusted p-value (padj). Table S2 ). Each point represents J o u r n a l P r e -p r o o f the score for one animal. Representative images of (F) apoptosis in the bronchial epithelium (400x magnification), (G) accumulation of neutrophils (200x magnification), and (H) severe vascular edema (100x magnification) from hamster lungs collected three days post SARS-CoV-2 infection (100pfu). See also Figure S3 and Table S2 . 

A serological assay to detect SARS-CoV-2 seroconversion in humans

CRISPR provides acquired resistance against viruses in prokaryotes

Autoantibodies against type I IFNs in patients with life-threatening COVID-19

Imbalanced Host Response to SARS-CoV-2 Drives Development of COVID-19

signaling restricts viral dissemination but drives severe pneumonia in SARS-CoV-2 infected hamsters

Neuropilin-1 facilitates SARS-CoV-2 cell entry and infectivity

Persistent Symptoms in Patients After Acute COVID-19

Anosmia and dysgeusia associated with SARS-CoV-2 infection: an age-matched case-control study

From reads to genes to pathways: differential expression analysis of RNA-Seq experiments using Rsubread and the edgeR quasi-likelihood pipeline

The cytokine storm in COVID-19: An overview of the involvement of the chemokine/chemokinereceptor system

Semi-quantitative Determination of Protein Expression using Immunohistochemistry Staining and Analysis: An Integrated Protocol

Approved Antiviral Drugs over the Past 50 Years

A Replication-Competent Vesicular Stomatitis Virus for Studies of SARS-CoV-2 Spike-Mediated Cell Entry and Its Inhibition

A mouse-adapted model of SARS-CoV-2 to test COVID-19 countermeasures

Crystallographic and electrophilic fragment screening of the SARS-CoV-2 main protease

Ending the HIV/AIDS Pandemic(1)

Ten Strategies of Interferon Evasion by Viruses

Small RNA-based antimicrobial immunity

Impaired type I interferon activity and inflammatory responses in severe COVID-19 patients

Identification of SARS-CoV-2 Inhibitors using Lung and Colonic Organoids

Syrian hamsters as a small animal model for SARS-CoV-2 infection and countermeasure development

Mouse model of SARS-CoV-2 reveals inflammatory role of type I interferon signaling

Innate immune recognition

SARS-CoV-2 ORF3b is a potent interferon antagonist whose activity is increased by a naturally occurring elongation variant

Colour Deconvolution -stain unmixing in histological imaging

Fast gapped-read alignment with Bowtie 2

Shared and Distinct Functions of Type I and Type III Interferons

Activation and evasion of type I interferon responses by SARS-CoV-2

MEGAHIT: an ultra-fast single-node solution for large and complex metagenomics assembly via succinct de Bruijn graph

Identification of Common Deletions in the Spike Protein of Severe Acute Respiratory Syndrome Coronavirus 2

Longitudinal analyses reveal immunological misfiring in severe COVID-19

It's DE-licious: A Recipe for Differential Expression Analyses of RNA-seq Experiments Using Quasi-Likelihood Methods in edgeR

Structure and function of the interferon-beta enhanceosome

Olfactory transmucosal SARS-CoV-2 invasion as a port of central nervous system entry in individuals with COVID-19

Decoding type I and III interferon signalling during viral infection

SARS-CoV-2 Orf6 hijacks Nup98 to block STAT nuclear import and antagonize interferon signaling

Safety and efficacy of inhaled nebulised interferon beta-1a (SNG001) for treatment of SARS-CoV-2 infection: a randomised, double-blind, placebo-controlled, phase 2 trial

Type I and Type III Interferons -Induction, Signaling, Evasion, and Application to Combat COVID-19

Dysregulation of Immune Response in Patients With Coronavirus

edgeR: a Bioconductor package for differential expression analysis of digital gene expression data

A scaling normalization method for differential expression analysis of RNA-seq data

Interferon-stimulated genes: a complex web of host defenses

Interferon-stimulated genes and their antiviral effector functions

Complement and tissue factor-enriched neutrophil extracellular traps are key drivers in COVID-19 immunothrombosis

Structural models of human ACE2 variants with SARS-CoV-2 Spike protein for structure-based drug design

Responses to Cytokines and Interferons that Depend upon JAKs and STATs

Generation of a Broadly Useful Model

REVIGO summarizes and visualizes long lists of gene ontology terms

STRING v11: J o u r n a l P r e -p r o o f protein-protein association networks with increased coverage, supporting functional discovery in genome-wide experimental datasets

The trinity of COVID-19: immunity, inflammation and intervention

The Evolution of Antiviral Defense Systems

Activation of TBK1 and IKKvarepsilon kinases by vesicular stomatitis virus infection and the role of viral ribonucleoprotein in the development of interferon antiviral immunity

Structural basis for translational shutdown and immune evasion by the Nsp1 protein of SARS-CoV-2

Ad26 vaccine protects against SARS-CoV-2 severe clinical disease in hamsters

Global virus outbreaks: Interferons as 1st responders

Post-discharge health status and symptoms in patients with severe COVID-19

Factors associated with COVID-19-related death using OpenSAFELY

SARS-CoV-2 infection of human ACE2-impaired function

ROAST: rotation gene set tests for complex microarray experiments

Evasion of Type I Interferon by SARS-CoV-2

Implications of gastrointestinal manifestations of COVID-19

Cell type-specific immune dysregulation in severely ill COVID-19 patients

Inborn errors of type I IFN immunity in patients with life-threatening COVID-19

Functional exhaustion of antiviral lymphocytes in COVID-19 patients

Impaired NLRP3 inflammasome activation/pyroptosis leads to robust inflammatory cell death via caspase-8/RIPK3 during coronavirus infection

SARS-CoV-2 Receptor ACE2 Is an Interferon-Stimulated Gene in Human Airway Epithelial Cells and Is Detected in Specific Cell Subsets across Tissues

This work was funded by the generous support of the Marc Haas Foundation and DARPA's PREPARE program (HR0011-20-2-0040). The views, opinions, and/or findings expressed are those of the author and should not interpreted as representing