key: cord-0834294-mi7gk451
authors: Ulrich, Lorenz; Halwe, Nico Joel; Taddeo, Adriano; Ebert, Nadine; Schön, Jacob; Devisme, Christelle; Trüeb, Bettina Salome; Hoffmann, Bernd; Wider, Manon; Bekliz, Meriem; Essaidi-Laziosi, Manel; Schmidt, Marie Luisa; Niemeyer, Daniela; Corman, Victor Max; Kraft, Anna; Godel, Aurélie; Laloli, Laura; Kelly, Jenna N.; Breithaupt, Angele; Wylezich, Claudia; Veiga, Inês Berenguer; Gultom, Mitra; Adea, Kenneth; Meyer, Benjamin; Eberhardt, Christiane; Thomann, Lisa; Gsell-Albert, Monika; Labroussaa, Fabien; Jores, Jörg; Summerfield, Artur; Drosten, Christian; Eckerle, Isabella Anne; Dijkman, Ronald; Hoffmann, Donata; Thiel, Volker; Beer, Martin; Benarafa, Charaf
title: Enhanced fitness of SARS-CoV-2 variant of concern B.1.1.7, but not B.1.351, in animal models
date: 2021-06-28
journal: bioRxiv
DOI: 10.1101/2021.06.28.450190
sha: 95090d785a14f3f36530dc06a78b67822bbb984e
doc_id: 834294
cord_uid: mi7gk451

Emerging variants of concern (VOCs) drive the SARS-CoV-2 pandemic. We assessed VOC B.1.1.7, now prevalent in several countries, and VOC B.1.351, representing the greatest threat to populations with immunity to the early SARS-CoV-2 progenitors. B.1.1.7 showed a clear fitness advantage over the progenitor variant (wt-S614G) in ferrets and two mouse models, where the substitutions in the spike glycoprotein were major drivers for fitness advantage. In the “superspreader” hamster model, B.1.1.7 and wt-S614G had comparable fitness, whereas B.1.351 was outcompeted. The VOCs had similar replication kinetics as compared to wt-S614G in human airway epithelial cultures. Our study highlights the importance of using multiple models for complete fitness characterization of VOCs and demonstrates adaptation of B.1.1.7 towards increased upper respiratory tract replication and enhanced transmission in vivo. Summary sentence B.1.1.7 VOC outcompetes progenitor SARS-CoV-2 in upper respiratory tract replication competition in vivo.

†These authors contributed equally to this work. ‡These authors jointly lead this work.

Uncontrolled transmission of severe acute respiratory syndrome-coronavirus 2 (SARS-CoV-2) in the human population has contributed to the persistence of the coronavirus disease 2019 (COVID- 19) pandemic. The emergence of new variants in largely immunologically naïve populations suggests that adaptive mutations within the viral genome continue to improve the fitness of this zoonotic virus towards its host. In March 2020, the single amino acid change in the spike (S) protein at position 614 (S 614D to S 614G ) was identified in only a small fraction of sequenced samples but, within a few weeks, became the most predominant variant worldwide 1 . The fitness advantage conferred by this single amino acid change was supported by a major increase in infectivity, viral loads, and transmissibility in vitro and in animal models 2-4 .

In the second half of 2020, new SARS-CoV-2 variants of concern (VOCs) with a combination of several mutations emerged including the B.1.1.7 (also known as Alpha) first described in southeast England 5 , and the B.1.351 (also known as Beta) first identified in South Africa 6 . In spring 2021, B.1.1.7 rapidly became the prevailing variant in many regions of the world and a higher reproduction number was inferred from early epidemiological data [7] [8] [9] . In addition to the S 614G change, B.1.1.7 has 18 different mutations in its genome, with two deletions and six substitutions within S alone 10 . Some of the S mutations, such as N501Y and H69-V70del, were hypothesized to confer enhanced replication and transmission capability, but clear experimental evidence is lacking 11,12 . B.1.351 VOC has nine mutations in S, including N501Y and two in the S receptor-binding domain (RBD), K417N and E484K, the latter putatively responsible for escaping neutralization from convalescent patient plasma [13] [14] [15] . Whether and which spike mutations are solely responsible for the putative fitness advantage is unknown.

Here, we investigated the fitness of the VOCs B.1.1.7 and B.1.351 compared to the predominant parental strain containing the S D614G substitution -henceforth wt-S 614G -(i) in relevant primary airway culture systems in vitro, and (ii) in ferrets, Syrian hamsters, and two humanized mouse models to assess specific advantages in replication and transmission and to 

To assess whether VOCs B.1.1.7 and B.1.351 exhibit a replicative advantage, we first infected primary nasal airway epithelial cell (AEC) cultures at an MOI of 0.02 at different ambient temperatures. No increased fitness was observed for clinical isolates of B.1.1.7 nor B.1.351 in comparison to a wild-type SARS-CoV-2 wt-S 614G clinical isolate (Fig. 1A,B) . Because SARS-CoV-2 B.1.1.7 became the prevailing variant over progenitor variants (e.g., B.1.2) in many regions of the globe, we performed both individual replication and direct competition experiments to understand the potential replicative advantage conferred by mutations inside and outside the S region. Direct competition experiments in primary nasal and bronchial AEC cultures showed that clinical B.1.1.7 isolates and an isogenic variant with the B.1.1.7 spike (wt-S B.1.1.7 ) had replication kinetics similar to wt-S 614G when starting at equal titer ratios (Fig. 1C,D) . Accordingly, individual B.1.1.7 spike mutations showed comparable replication kinetics to wt-S 614G (Extended Data isolate (Extended Data Fig 2B) . Overall, the phenotype of both VOCs showed no major change in replication in human AEC cultures and were neutralized in part by serum of heterologous convalescent and vaccinated patients as reported in larger serological studies [13] [14] [15] .

Groups of six Syrian hamsters were each inoculated intranasally with a mixture of two SARS-CoV-2 strains comprising comparable genome equivalents in three one-to-one competition experiments: B.1.1.7 vs B.1.351, B.1.351 vs wt-S 614G , and B.1.1.7 vs wt-S 614G . All experimentally infected "donor" hamsters were strictly kept in isolation cages to prevent intergroup spill-over infections. Each donor hamster was co-housed with a naïve "contact I" hamster 1 day post infection (dpi), creating six donor-contact I pairs to evaluate shedding and transmission (Extended Data Fig.  3A ). On 4 dpi, donor hamsters were euthanized, and six subsequent transmission pairs were set up by co-housing each contact I hamster with a naïve contact II hamster (Extended Data Fig. 3A ).

In two competition experiments, wt-S 614G and B.1.1.7 outcompeted B.1.351 in nasal washings of the donor hamsters from 1 dpi until euthanasia 4 dpi. Genome copies reached up to 10 9 /mL for wt-S 614G and B.1.1.7, whereas B.1.351 viral loads were 10-fold lower at corresponding time points (Fig. 2, Extended Data Fig. 4) . Consequently, transmission was dominated almost every time by the competing variants wt-S 614G (Fig. 2) In the B.1.1.7 vs wt-S 614G competition, no clear fitness difference was observed in virus replication in nasal washes of donor hamsters, and both variants were detected at all time-points in each donor with B.1.1.7 specific copies ranging from 10 5 to 10 9 genome copies/mL, and wt-S 614G from 10 5 to 10 8 genome copies/mL (Fig. 3) . Of note, B.1.1.7 was dominant over wt-S 614G in the donor hamsters at 1 dpi, but ratios were balanced by the endpoint at 4 dpi. In organ samples of the donor hamsters, the highest viral loads were confirmed in the LRT with at least 10-fold higher genome loads for B.1.1.7 in comparison to wt-S 614G for 5 out of 6 animals (Extended Data Fig. 9 ).

Sequential transmission to contact I and II animals was highly efficient for both variants, which were simultaneously found in nasal washings of almost all contact hamsters, and equal dominance of B.1.1.7 and wt-S 614G in 3 out of 6 transmission groups (Fig 3, Extended Data Figs. 5A, 6A) . No clear advantage for one variant over the other was noted in the viral loads of the URT and LRT at experimental endpoints (Extended Data Fig. 9 ).

High SARS-CoV-2 replication in hamsters induced a rapid humoral immune response, as shown by serum reactivity in RBD-based ELISA and virus neutralization tests (VNT) (Extended Data Figs. 10, 11) . Sera of all B.1.1.7 vs wt-S 614G contact I animals (20 days post co-housing with donors) were positive in both assays, while in the contact II group (17 days post co-housing with contact I) four out of six reacted positive in both tests, one negative only in the VNT and one negative in both assays. Reduced neutralization of B.1.351 by the sera of the hamsters of the three competition experiments is justified in part by the impaired replication and transmission of this virus and by the reduced cross-reactivity of the sera (Extended Data Figs. 10, 11).

In a similar approach as for hamsters, six donor ferrets were inoculated with a mixture of wt-S 614G and B.1.1.7 at comparable genome equivalents and sequential transmission was followed in naïve contact I and II ferrets (Extended Data Fig. 3B ). B.1.1.7 rapidly became the dominant variant in nasal washings from 2 dpi with up to 10 5 viral genome copies per mL (Fig. 4) . Correspondingly, the nasal concha of donor ferrets revealed high replication in the nasal epithelium and up to 100fold higher load of B.1.1.7 (up to 10 8.5 viral genome copies per mL) than wt-S 614G (up to 10 6.5 viral genome copies/ml) (Extended Data Fig. 12 ). While histopathological analysis clearly indicated viral replication within the nasal epithelium of the donor ferrets (Extended Data Fig. 13 ), no severe clinical signs were observed (Extended Data Figs. 5, 6) . Transmission to contact I animals was only detected in two ferret pairs, from which only one contact I ferret transmitted the virus to the contact II ferret. However, in each of these three transmission events, B.1.1.7 variant was vastly dominant and replicated to similarly high titers as in donor ferrets ( Fig. 4 and Extended Data Fig.   12 ). Virus shedding contact ferrets seroconverted by 21 dpi (Extended Data Figs. 10, 11).

To assess further adaptation of B.1.1.7 to human ACE2, four hACE2-K18Tg mice, which overexpress hACE2 in respiratory epithelium 16 , were inoculated with a mixture of SARS-CoV-2 wt-S 614G and B.1.1.7 at comparable genome equivalents (Fig. 5A ). Each inoculated mouse was cohoused with a contact hACE2-K18Tg mouse at 1 dpi. A complete predominance of B.1.1.7 in the oropharyngeal samples of all four inoculated mice was observed from 1 to 4 dpi with up to 10 6 viral genome copies/ml. This increased replicative fitness of B.1.1.7 over wt-S 614G was further reflected throughout the respiratory tract with higher genome copies in nose, lungs, olfactory bulb, and the brain at 4 dpi ( Fig. 5A ). All inoculated mice showed body weight loss at 4 dpi, but no weight loss was observed in contact mice (Extended Data Fig. 14A ). Only B.1.1.7 viral genomes were detected in lungs of contact mice (Extended Data Fig. 14B ).

To differentiate the influence of the spike mutations from those of the rest of the genome of B.1.1.7, a similar competition experiment was performed between wt-S 614G and wt-S B.1.1.7 .

Transgenic hACE2-K18Tg mice were inoculated with a mixture of wt-S B.1.1.7 and wt-S 614G at comparable genome equivalents and placed with a contact hACE2-K18Tg mouse at 1 dpi.

Interestingly, the replicative advantage of wt-S B.1.1.7 was less clear, especially in the lungs, where both wt-S B.1.1.7 and wt-S 614G variants showed comparable RNA levels (Fig. 4B ). While none of the contact mice lost weight, two of the four mice had detectable viral RNA in the lung 7 days after contact and only wt-S B.1.1.7 was detected (Extended Data Fig. 14) . The results indicate that the S B.1.1.7 spike mutations contribute partially to the replication advantage in the URT and to enhanced transmission of B.1.1.7 using mice that express high levels of human ACE2 in a K18-dependent manner.

We next used hACE2-KI homozygous mice, which express hACE2 in place of mouse ACE2 under the endogenous mouse Ace2 promoter 4 . In contrast to hACE2-K18Tg mice, hACE2-KI mice have a physiological expression of hACE2 with no ectopic expression of hACE2 in the brain, and no expression of mouse Ace2, which has been shown to be permissive to the spike mutation N501Y contained in S B.1.1.7 17 . Three groups of hACE2-KI mice were inoculated intranasally with a relatively low dose (10 4 PFU/mouse) of either SARS-CoV-2 wt-S 614G , B.1.1.7, or wt-S B.1.1.7 (n=8/group) as single virus infection (Fig. 6A ). Significantly higher viral genome copy numbers were observed in mice infected with B.1.1.7 or with wt-S B.1.1.7 compared to wt-S 614G in oropharyngeal swabs 1 dpi, in the nose 2 dpi, and olfactory bulb 4 dpi (Fig. 6B ,C). Of note, virus titers in the nose and lungs showed SARS-CoV-2 persistence 4 dpi in 3 out of 4 mice infected either with B.1.1.7 or with wt-S B.1.1.7 , but not for hACE2-KI mice inoculated with wt-S 614G (Fig.   6D ). No difference in lung histopathology score was observed between groups (SI Table 1 ).

Finally, we performed competition experiments to compare the replication of the B.1.1.7 or wt-S B.1.1.7 with wt-S 614G in two groups of six hACE2-KI mice. We observed a complete predominance of B.1.1.7 variant and wt-S B.1.1.7 over wt-S 614G from 1 dpi in the URT. At 4 dpi, only B.1.1.7 or wt-S B.1.1.7 was detectable in lungs or nose (Fig. 5E ,F). No weight loss was observed in any of the hACE2-KI mice (Extended Data Fig. 15 ). Together, the two humanized mouse models support enhanced fitness of SARS-CoV-2 B.1.1.7 variant over its ancestor wt-S 614G , with increased replication and persistence in the URT and better systemic spread, mediated in part by changes located within the B.1.1.7 spike sequence. In binding assays in vitro, affinity between S B.1.1.7 and the ACE2 receptor was consistently higher than that of S 614G and ACE2 of humans, hamsters, and ferrets (Extended Data Fig. 16 ).

Epidemiological data indicate that new SARS-CoV-2 variant lineages with specific amino acid changes have a fitness advantage over contemporary strains. VOCs such as B.1.1.7 and B.1.351 are particularly concerning for their hypothesized ability to supersede progenitor strains and immune escape properties, respectively. In this comprehensive study, we provide experimental evidence that SARS-CoV-2 B.1.1.7 has a clear replication advantage over wt-S 614G in both ferret and two humanized mouse models. Moreover, B.1.1.7 exclusively was transmitted to contact animals in competition experiments, where ferrets and hACE2-K18Tg mice were inoculated with mixtures of B.1.1.7 and wt-S 614G . We have also shown that the molecular mechanism behind the fitness advantage of B.1.1.7 in vivo is largely dependent on a few changes in the S including three amino acid deletions (H69, V70, Y144), and six substitutions (N501Y, A570D, P681H, T716I, S982A, D1118H). In vitro, B.1.1.7 spike mutations increased the affinity to human ACE2 over 5fold, and 2-3-fold to ferret and hamster ACE2 indicative of an overall improvement in binding abilities rather than a specialization to human ACE2. In hACE2-KI mice, higher genome copies and/or titers of B.1.1.7 and wt-S B.1.1.7 compared to wt-S 614G were found in the URT (oropharynx, nose) and olfactory bulb. Increased replication and transmission of wt-S B.1.1.7 over wt-S 614G were also evident in hACE2-K18Tg mice. Transmission events are rare in mice, however, we observed transmission of B.1.1.7 and wt-S B.1.1.7 in 50% of the contact hACE2-K18Tg mice and no detection of wt-S 614G in any contact mouse. hosts support the hypothesis that the epidemiological advantage of B.1.351 may be due to immune escape as indicated by reduced efficiency in serum neutralization tests shown here and previously 14 . In immune -convalescent or vaccinated-populations, the immune escape advantage of B.1.351 may prove to be sufficient to compensate for the intrinsic reduced fitness and explains e.g., the low prevalence of this variant in regions with a mainly naïve population.

In biochemical assays, we found that the spike protein of B.1.1.7 had a higher affinity than wt-S 614G to the hamster ACE2 receptor. Yet, B.1.1.7 and wt-S 614G were comparable in their replication and transmission in Syrian hamsters in vivo suggesting that, in a model with a high basal rate of replication, the impact of a marginally fitter SARS-CoV-2 variant may not become apparent. Indeed, efficient simultaneous transmission of both variants to contact hamsters was observed in association with high viral loads in infected animals. In models supporting high replication, such as human AEC cultures and hamsters, only major improvement in replication and transmission can be detected when the variants compared already have a very high fitness. In contrast, in ferrets and hACE2-expressing mice, where the bottleneck for viral replication is more stringent, VOCs with a modest, but real increased ability to replicate and transmit can be identified.

The similar replication and transmission efficacy in the hamster are also in line with recent publications using the hamster model and VOCs 19 . The high replication levels observed from the infected hamster, however, might also indicate that the current variants of concern, e.g., B.1.1.7, are on their way to approaching the viral limit for fitness mutations in naïve hosts.

The basal rate of replication is an important factor in the assertion of a variant over a contemporary variant in a naïve population. Some individuals with higher bioaerosol exhalation levels can initiate disproportionate numbers of transmission events, possibly because of higher viral load in the URT, and are therefore called "superspreaders" 20 . The hamster model might thus resemble the human "superspreader scenario" since there are no clear indications of a predominant variant in this particular competitive transmission experiment. In the ferret and hACE2-KI models, more restrictions are found, e.g., to the URT. Therefore, these models more closely mimic the situation in humans with a dominance of mild infections. While the transmission events were not high overall (3 out of 8 pairs in ferrets, and 4 out of 8 in hACE2-K18Tg mice), the almost exclusive transmission of VOC B.1.1.7 relative to wt-S 614G mirrored the higher transmission rate in the human population to some extent, where B.1.1.7 has been responsible for more than 90% of recent infections in most countries in Europe 21 .

Overall, our study demonstrates that multiple complementary models (hAEC cultures, hamsters, ferrets, homozygous hACE2-KI and hemizygous hACE2-K18Tg mice) are necessary to comprehensively evaluate and fully chisel different aspects of human SARS-CoV-2 infection and the impact of emerging VOCs on the course of the ongoing pandemic. The combined analysis of our results allows a clear conclusion supporting a fitness advantage of B.1.1.7 and a concomitant disadvantage of B.1.351, which is in line with the observed epidemiological dominance and immune escape phenotype of these VOCs. Importantly, and reassuringly, despite the apparent fitness differences of these VOCs, there is no indication of increased pathology nor complete immune escape from humoral immunity. 

Authors declare that they have no competing interests.

All data are available in the main text or the supplementary materials. 

Extended Data Figures 1 to 17 Extended Data Tables 1 to 3 Legends for Supplementary Information: SI Table 1 and SI 20 . Isolation and maintenance of primary nasal and bronchial AEC cultures were performed according to manufacturer's guidelines or as previously described 24, 25 . Individual SARS-CoV-2 infections with contemporary virus isolates were conducted at either 33°C or 37°C as described elsewhere using an MOI of 0.02 26 , while all competition experiments and replication kinetics of individual isogenic variants were performed with an MOI of 0.005 as described previously 24 . The viral load quantification of individual SARS-CoV-2 infections with contemporary virus isolates was performed using the NucliSens easyMAG (BioMérieux) and quantitative real-time PCR (RT-qPCR) targeting the E gene of SARS-CoV-2 as described 27, 28 . The competition experiments nucleic acids were extracted using the Quick-RNA Viral 96 kit (Zymo research) and the RT-qPCR primers described in Extended Data Table 2 . The viral replication of individual isogenic variants was monitored via plaque assay.

Viruses released into the apical compartments were titrated by plaque assay on Vero E6 cells, as previously described 24, 29 . Briefly, 2x10 5 cells/ml were seeded in 24-well plates 1 day prior to titration and inoculated with 10-fold serial dilutions of virus solutions. and mixed with 100 plaque-forming units of the specific SARS-CoV-2 strain. Each 24-well was incubated with serum/plasma-virus. After 1 hour at 37°C, the supernatants were discarded, and cells were washed once with PBS and supplemented with the overlay solution also used for the plaque assay. After 3 days at 37°C, the supernatants were discarded, and cells were fixed using a 6% formalin solution and stained with crystal violet. All dilutions were tested in duplicates. Serum dilutions with an average plaque reduction of 50% or 90% (PRNT50/90) are referred to as titers.

Plaque reduction neutralization test (PRNT) (Geneva)

Vero-E6 cells were seeded at a density of 4x10 5 cells/mL in 24-well cell culture plates. All serum/plasma were heat-inactivated at 56°C for 30min and serially diluted in Opti-Pro serum free medium starting from 1:10 until 1:2560 if necessary. Serum/plasma were mixed with 50PFU of variant isolates and incubated at 37°C for 1h. All samples were run in duplicate and for each neutralization experiment an infection control (no serum/plasma) and a reference serum was used to ensure reproducibility between different experiments. Vero-E6 cells were washed 1x with PBS and inoculated with the virus serum/plasma mixture for 1h. Afterwards, the inoculum was removed and 500uL of the overlay medium also used for the plaque assays was added. After incubation for 3 days at 37°C, 5% CO2, the overlay medium was removed, cells were fixed in 6% formaldehyde solution for at least 1h, plates were washed 1x with PBS and stained with crystal violet. Plaques were counted in wells inoculated with virus-serum/plasma mixtures and compared to plaque counts in infection control wells. The 90% reduction endpoint titers (PRNT90) were calculated by fitting a 4-parameter logistics curve with variable slope to the plaque counts of each serum/plasma.

Protein expression, purification, and biolayer interferometry assay SARS-CoV-2 spike protein expression plasmids were constructed to encode the ectodomain of spike protein S 614G or S B.1.1.7 (residues 1-1208, with a mutated furin cleavage site and K986P/V987P substitutions) followed by a T4 foldon trimerization domain and a polyhistidine purification tag. ACE2 protein (human, hamster, or ferret) expression plasmids were constructed to encode the ectodomain of ACE2 followed by a human IgG1 Fc purification tag. The recombinant proteins were expressed using the Expi293 Expression system (ThermoFisher Scientific) and purified with HisTrap FF columns (for polyhistidine-tagged spike proteins) or with HiTrap Protein A column (for Fc-tagged ACE2 proteins) in FPLC (Cytiva) system.

Binding affinity between the trimeric spike and dimeric ACE2 were evaluated using Octet Inoculated donor hamsters were isolated in individually ventilated cages for 24 hours.

Thereafter, contact hamster I was co-housed to each donor, creating six donor-contact I pairs (Extended Data Fig. 3A ). The housing of each hamster pair was strictly separated in individual cage systems to prevent spillover between different pairs. Four days post inoculation (dpi), the individual donor hamsters (inoculated animal) were euthanized. To simulate a second transmission cycle, the original contact hamsters (referred to as contact I) were commingled with a further six naïve hamsters (referred to as contact II), which equates to another six contact I and contact II pairs (Extended Data Fig. 3A ). These pairs were co-housed until the end of the study at 21 dpi. Because the first contact hamster (cage 6) in the competition trial wt-S 614G vs. B.1.1.7, died already 2 dpc, the second contact hamster for this cage was also co-housed with the donor hamster; thus the first and second contact hamsters in this cage were labeled contact Ia and contact Ib, respectively. To enable sufficient contact between the donor hamster and contact Ib hamster, which was commingled routinely on 4 dpi, the donor hamster was euthanized not before 7 dpi (instead of 4 dpi), when it finally reached the humane end-point criterion for bodyweight (below 80% of 0 dpi body weight).

Viral shedding was monitored by nasal washings in addition to a daily physical examination and body weighing routine. Nasal washing samples were obtained under a short-term isoflurane anesthesia from individual hamsters by administering 200µl PBS to each nostril and collecting the reflux. Animals were sampled daily from 1 dpi to 9 dpi, and afterwards, every other day until 21

dpi. Under euthanasia, serum samples and an organ panel comprising representative upper (URT) and lower respiratory tract (LRT) tissues were collected from each hamster. All animals were observed daily for signs of clinical disease and weight loss. Animals reaching the humane endpoint, e.g., falling below 80% of the initial body weight relative to 0 dpi, were humanely euthanized.

Ferret studies Similar to the hamster study, 12 ferrets (six donor ferrets and six transmission I ferrets) from the FLI in-house breeding were housed pairwise in strictly separated cages to prevent spillover contamination. Of these, six ferrets were inoculated with an equal 250 µl mixture of SARS-CoV-2 wt-S 614G and B.1.1.7. The inoculum was back-titrated and the ratio of each variant was determined by RT-qPCR. The wt-S 614G vs B.1.1.7 mixture held a 1:1.2 ratio with 10 5.875 TCID50 distributed equally into each nostril of donor ferrets. Ferrets were separated for the first 24 hours following inoculation. Subsequently, the ferret pairs were co-housed again, allowing direct contact of donor to contact I ferrets. All ferrets were sampled via nasal washings with 750 µl PBS per nostril under a short-term inhalation anesthesia. Donor ferrets were sampled until euthanasia at 6 dpi, which was followed by the introduction of one additional naïve contact II ferret per cage (n=6), resulting in a 1:1 pairwise setup with contact I and II ferrets (Extended Data Fig. 3B ). All ferrets, which were in the study group on the respective days, were sampled on the indicated days.

Bodyweight, temperature, and physical condition of all animals were monitored daily throughout the experiment. URT and LRT organ samples, as well as blood samples of all ferrets were taken at respective euthanasia timepoints.

Full autopsy was performed on all animals under BSL3 conditions. The lung, trachea, and nasal conchae were collected and fixed in 10% neutral-buffered formalin for 21 days. The nasal atrium, decalcified nasal turbinates (cross-sections every 3-5 mm), trachea and all lung lobes were trimmed for paraffin embedding. Based on PCR results, tissue sections (3 μm) of all donors (day 6) and one recipient (# 8, day 20) were cut and stained with hematoxylin and eosin (H&E) for light microscopical examination. Immunohistochemistry (IHC) was performed using an anti-SARS nucleocapsid antibody (Novus Biologicals #NB100-56576, dilution 1:200) according to standardized avidin-biotin-peroxidase complex-method producing a red labelling and hematoxylin counterstain. Lung tissue pathology was evaluated according to a detailed score sheet developed by Angele Breithaupt (DipECVP) (SI Table 2 ). Evaluation and interpretation was performed by board-certified veterinary pathologists (DiplECVP) (AB, IBV).

Homozygous hACE2 knock-in mice (B6.Cg-Ace2 tm1(ACE2)Dwnt ; henceforth hACE2-KI) and

hemizygous transgenic mice (Tg(K18-hACE2)2Prlmn; henceforth hACE2-K18Tg) were described previously 4,16 . All mice were produced at the specific-pathogen-free facility of the Institute of Virology and Immunology (Mittelhäusern), where they were maintained in individually ventilated cages (blue line, Tecniplast), with 12-h/12-h light/dark cycle, 22 ± 1 °C ambient temperature and 50 ± 5% humidity, autoclaved food and acidified water. At least 7 days before infection, mice were placed in individually HEPA-filtered cages (IsoCage N, Tecniplast).

Mice (10 to 12 weeks old) were anesthetized with isoflurane and infected intranasally with 20 μl per nostril with the virus inoculum described in the results section. One day after inoculation, Table 2 ). Here, virus specific primers were used to realize a high analytical sensitivity (less than 10 genome copies/µl template) of the according PCR assays, also in samples with a high genome load of the non-matching virus.

The RT-qPCR reaction was prepared using the qScript XLT One-Step RT-qPCR ToughMix 

The N501Y spike substitution enhances SARS-CoV-2 transmission. bioRxiv

Recurrent emergence and transmission of a SARS-CoV-2 Spike deletion H69/V70. bioRxiv

Escape of SARS-CoV-2 501Y.V2 from neutralization by convalescent plasma

SARS-CoV-2 variants B.1.351 and P.1 escape from neutralizing antibodies

SARS-CoV-2 501Y.V2 escapes neutralization by South African COVID-19 donor plasma

Lethal infection of K18-hACE2 mice infected with severe acute respiratory syndrome coronavirus

Adaptation of SARS-CoV-2 in BALB/c mice for testing vaccine efficacy

CVnCoV protects human ACE2 transgenic mice from ancestral B BavPat1 and emerging B.1.351 SARS-CoV-2. bioRxiv

Comparing infectivity and virulence of emerging SARS-CoV-2 variants in Syrian hamsters

Exhaled aerosol increases with COVID-19 infection, age, and obesity

SARS-CoV-2 Cell Entry Depends on ACE2 and TMPRSS2 and Is Blocked by a Clinically Proven Protease Inhibitor

Rapid reconstruction of SARS-CoV-2 using a synthetic genomics platform

Disparate temperature-dependent virus-host dynamics for SARS-CoV-2 and SARS-CoV in the human respiratory epithelium

Well-Differentiated Primary Mammalian Airway Epithelial Cell Cultures

Propagation of respiratory viruses in human airway epithelia reveals persistent virus-specific signatures

Detection of 2019 novel coronavirus (2019-nCoV) by real-time RT-PCR

SARS-CoV-2 viral load in the upper respiratory tract of children and adults with early acute COVID-19

At 4 dpi, the donor hamsters were euthanized and the initial contact hamsters I were co-housed with one additional hamster (Contact II). (B) Timeline of the ferret experiment

A-C) Syrian hamsters were inoculated with comparable genome equivalent mixture of either wt-S 614G and B.1.1.7 (A), wt-S 614G and B.1.351 (B), or B.1.1.7 and B.1.351 (C). In hamsters, body weight was monitored daily until 13 dpi, afterwards every two days until 21 dpi and

D) and temperature (E) were monitored daily in ferrets until 12 dpi, and afterwards every 2 days

Grey dotted lines in E indicate the physiologic range for body temperature in ferrets

Viral genome load in upper (URT) and lower (LRT) respiratory tract tissues of Syrian hamsters in the competitive transmission experiment between SARS-CoV-2 wt-S 614G and B.1.351. Absolute quantification was performed by RT-qPCR analysis of tissue homogenates of donor, contact I and contact II hamsters in relation to a set of defined standards

Tissue samples were collected at the respective time of euthanasia (Euth.)

Indirect ELISA against the RBD of SARS-CoV-2. Sera of ferrets (A) and hamsters (B) inoculated with a mixture of wt-S 614G and B.1.1.7 were tested for specific reactivity against the SARS-CoV-2 RBD-SD1 domain (wt-S amino acids 319-519) at the terminal endpoints for each animal

Extended Data Fig. 11. Virus neutralization titers of animal sera. SARS-CoV-2 100%

neutralization titers (VNT100) in serum samples (log2 titer) of sera from wt-S 614G and B.1.1.7, coinoculated hamsters (A) and ferrets (B), wt-S 614G and B.1.351 inoculated hamsters (C) and B

Sera from all donor, contact I and contact II hamsters (A) and ferrets (B) used in the competitive transmission experiment between wt-S 614G and B.1.1.7 was collected at respective timepoints of euthanasia and were tested against SARS-CoV-2 wt-S 614G and B.1.1.7 on VeroE6 cells. The sera of the donor, contact I and contact II hamsters used in the competitive transmission experiment between B.1.351 and wt-S 614G (C), as well as the sera of the B.1.351 and B.1.1.7 co-inoculated hamsters (D) were tested against wt-S 614G , B.1.1.7 and B.1.351. Extended Data Fig. 12. Viral genome load in upper (URT) and lower (LRT) respiratory tract tissue of ferrets in the competitive transmission experiment between SARS-CoV-2 B.1.1.7 and wt-S 614G . Absolute quantification was performed by RT-qPCR analysis of tissue homogenates of donor

Tissue samples were collected at the respective time of euthanasia (Euth.)

Extended Data Fig. 13. Histopathology of nasal conchae of donor ferrets inoculated with wt

Representative micrographs of hematoxylin and eosin staining of 3 μm sections of nasal conchae of donor ferrets 6 dpi. Insets show immunohistochemistry staining of SARS-CoV-2 with anti-SARS nucleocapsid antibody with hematoxylin counterstain. (A-D) The respiratory (A, B) and olfactory (C, D) nasal mucosa exhibited rhinitis with varying severity

Lesion-associated antigen was found in ciliated cells of the respiratory epithelium (A, B) and in sustentacular cells of the olfactory epithelium (C, D) in all animals at 6 dpi

Extended Data Fig. 14. Bodyweight and transmission in hACE2-K18Tg mice

A) Relative body weight in donor mice (left panel), and in contact mice (right panel) (n=4/group). (B) Pie chart illustrating the ratio of wt-S 614G (orange) with B.1.1.7 (dark blue

Bodyweight of hACE2-KI mice. Mice mock-infected or inoculated with 10 4 PFU of wt-S 614G , B.1.1.7, or wt-S B.1.1.7 (n = 8 until 2 dpi

Genome sequences of used SARS-CoV-2 variants. Colors of the variants represent respective viruses in the different experiments

SARS-CoV-2 B.1.1.7 -ORF1 assay SARS-CoV-2 B.1.1.7

614G vs B.1.1.7 (Mouse studies and human AEC cultures) SARS-CoV-2 wt-S 614G -S assay SARS-CoV-2 wt-S 614G

SARS-CoV-2 B.1.1.7 -S assay SARS-CoV-2 B.1.1.7

SARS-CoV-2 B.1.1.7 -S assay SARS-CoV-2 wt-S B.1.1.7 wt-S 614G vs B.1.351 (Hamsters) SARS-CoV-2 wt-S 614G -ORF1 assay SARS-CoV-2 wt-S 614G

SARS-CoV-2 B.1.1.7 -ORF1 assay SARS-CoV-2 B

B.1.1.7 vs B.1.351 (Hamsters) SARS-CoV-2 B.1.1.7 -S assay SARS-CoV-2 B.1.1.7

 Table 1 . Histopathological score of lungs of hACE2-KI mice infected with single virus inoculum of wt-S 614G , B.1.1.7, or wt-S B.1.1.7 . Hematoxilin and eosin stained slides of left lungs of hACE2-KI mice were scored by a board-certified veterinary pathologist, who was blinded to the identity of the specimen. The scoring scheme is detailed in SI Table 2 . Table 2 . Lung pathology scoring scheme.