key: cord-0738164-ofiq6pjq
authors: Zhou, Jie; Peacock, Thomas P.; Brown, Jonathan C.; Goldhill, Daniel H.; Elrefaey, Ahmed M.E.; Penrice-Randal, Rebekah; Cowton, Vanessa M.; De Lorenzo, Giuditta; Furnon, Wilhelm; Harvey, William T.; Kugathasan, Ruthiran; Frise, Rebecca; Baillon, Laury; Lassaunière, Ria; Thakur, Nazia; Gallo, Giulia; Goldswain, Hannah; Donovan-Banfield, I’ah; Dong, Xiaofeng; Randle, Nadine P.; Sweeney, Fiachra; Glynn, Martha C.; Quantrill, Jessica L.; McKay, Paul F.; Patel, Arvind H.; Palmarini, Massimo; Hiscox, Julian A.; Bailey, Dalan; Barclay, Wendy S.
title: Mutations that adapt SARS-CoV-2 to mink or ferret do not increase fitness in the human airway.
date: 2022-01-19
journal: Cell Rep
DOI: 10.1016/j.celrep.2022.110344
sha: e0a378f17ecbeb4a9469fcb79dd61a793672c530
doc_id: 738164
cord_uid: ofiq6pjq

SARS-CoV-2 has a broad mammalian species tropism infecting humans, cats, dogs and farmed mink. Since the start of the 2019 pandemic several reverse zoonotic outbreaks of SARS-CoV-2 have occurred in mink, one of which reinfected humans and caused a cluster of infections in Denmark. Here we investigate the molecular basis of mink and ferret adaptation and demonstrate the spike mutations Y453F, F486L, and N501T all specifically adapt SARS-CoV-2 to use mustelid ACE2. Furthermore, we risk assess these mutations and conclude mink-adapted viruses are unlikely to pose an increased threat to humans, as Y453F attenuates the virus replication in human cells and all 3 mink-adaptations have minimal antigenic impact. Finally, we show that certain SARS-CoV-2 variants emerging from circulation in humans may naturally have a greater propensity to infect mustelid hosts and therefore these species should continue to be surveyed for reverse zoonotic infections.

Introduction 32 SARS-CoV-2 is a betacoronavirus that is thought to have emerged from an animal source in 33 2019 and rapidly spread by human-to-human transmission across the globe causing the COVID-19 34 pandemic. SARS-CoV-2 is transmitted efficiently by the airborne route due to its ability to efficiently 35 enter cells in the upper respiratory tract. The spike glycoprotein is responsible for host receptor 36 binding and membrane fusion of coronaviruses. SARS-CoV-2 spike binds to host angiotensin-37 converting enzyme 2 (ACE2) via the receptor binding domain (RBD) and is activated by TMPRSS2 38 protease expressed at the apical surface of the airway epithelium to mediate fusion (Hoffmann et al., 39 2020b). In addition, compared to closely related coronaviruses, SARS-CoV-2 spike contains a tract of 40 basic amino acids at the S1/S2 cleavage site that can be recognised by furin, enabling spike to be 41 surface of ACE2 result in SARS-CoV-2 binding to human ACE2 with high affinity 46 Yan et al., 2020). SARS-CoV-2 shows a broad host tropism and can experimentally infect many animal 47 intranasally inoculated 4 naive ferrets with ferret P2 virus and compared levels of virus shed from the 131 nose to 4 ferrets previously inoculated with the same infectious titre of parental England/2/2020 virus 132 (the same donors from Figure 1A ). At days 1-2, the mean titre of Y453F virus 133 shed in nasal washes was significantly higher than that of the parental virus, as determined by both E 134 gene copy number and TCID50 ( Figure 2B ,C). Both groups of ferrets showed comparable patterns of 135 fever during infection, peaking between days 2-4, and the Y453F-infected ferrets trended towards 136 more weight loss over the course of the experiment ( Figure 2D ,E). The titre of parental virus shed and 137 fever in parental virus-infected animals approached that in the ferret P2 infected animals by days 3-4, 138 likely because the parental virus had gained ferret-adapting mutations, such as Y453F or N501T, by 139 this point (see Figure 1A ). Deep sequencing of the virus from the ferrets inoculated with the Y453F-140 containing ferret P2 virus showed the Y453F substitution was maintained in all 4 animals throughout 141 the course of infection ( Figure 2F ). The E gene substitution S6L, however, was rapidly selected against, 142

indicating that this substitution could have been an adaptation to cell culture, selected in Vero cells 143 during isolation and amplification of the virus from nasal wash ( Figure 2G ). Several further 144 substitutions, all present at very low levels in the inoculum, rapidly grew to fixation in all 4 Y453F-145 infected ferrets. These encoded mutations in spike at D614N, in N protein at R68P and in the NSP2 146 protein at T632I ( Figure 2G ). It is unclear whether these substitutions are all bona fide ferret 147 adaptations or mutations hitchhiking as part of a selective sweep. Spike D614N may exert a similar 148 effect to the ubiquitous SARS-CoV-2 human adaptation D614G, to non-specifically enhance ACE2 149 binding by promoting the spike open conformation (Juraszek et al., 2021). Overall, these data suggest 150 that Y453F adapts the virus to ferret infection, but also further adaptations may arise during ongoing 151 adaptation in mustelid hosts. 152 Y453F enhances cell entry using the mustelid ACE2 receptor 153 J o u r n a l P r e -p r o o f Next, we tested whether Y453F and the other mustelid associated spike mutations improved 154 the use of the otherwise suboptimal ferret ACE2 . We created a library of spike 155 expression constructs, generated lentivirus-based pseudoviruses and assessed the entry of these into 156 cells transiently expressing ACE2 from human, ferret or rat, or empty vector, as previously described 157 . We note that ferret ACE2 differs from that of mink by only two amino acid 158 residues that are distal to the spike interaction interface, and therefore can be considered 159 representative for both mustelid species (See Supplementary Figure S1 ). 160

While WT (D614G) spike uses ferret ACE2 poorly for entry (>10-fold less well than human 161 ACE2), the adaptations Y453F, N501T or F486L, as well as full Cluster 5 spike (Δ69/70, Y453F, D614G, 162 I692V, M1229I), all allowed SARS-CoV-2 spike expressing pseudoviruses to enter into human-or ferret-163 , but not rat-, ACE2 expressing cells with much greater efficiency ( Figure 3A , B). A nearby substitution, 164 L452M, which has also appeared in at least one mink farm outbreaks (Lu et al., 2021) has no effect 165 suggesting this is not a specific adaptation to mink ( Figure 3A ). The effect of Y453F was not dependent 166 on the presence or absence of D614G, as Y453F in a 614D background showed a similar effect ( Figure  167 3C). Consistent results were also seen using a cell-cell fusion assay (Supplementary Figure S2A,B) . 168

Examining the structure of the spike RBD/ACE2 interface, each of these mink/ferret-adaptations is 169 close to residues that differ between human and mustelid ACE2, as others have previously modelled 170 (Welkers et al., 2020). For example, Spike-Y453F lies close to ACE2-34 (histidine in human ACE2, 171 tyrosine in mustelid), Spike-N501T lies close to ACE2-354, and Spike-F486L lies between ACE2 residues 172 79, 82 and 24 ( Figure 3D ). 173

Viruses containing Y453F mutation are attenuated for replication in primary human airway 174 epithelial cells 175

To assess the impact of the Y453F mutation on the replication of virus in human airway 176 epithelium, we infected primary human bronchial cells cultured at an air liquid interface with a mix of 177 the parental and ferret P2 viruses at a low multiplicity of infection (MOI) of around 0.1 ( Figure 4A) . 178 virus significantly outcompeted Y453F, with less than ~5% of reads by 48 hours post infection 180 containing Y453F. 181

Although the Y453F containing virus is highly similar to that which circulated in mink early in 182 the pandemic, the most prominent zoonotic spillover from mink was the Cluster 5 virus, which further 183 contained D614G and Δ69-70. D614G and Δ69-70 are thought to potentially enhance virus infectivity 184 in some backgrounds (Meng et al., 2021). Therefore, we performed a similar competition experiment 185 between a mixed inoculum of 40% Cluster 5 isolate and 60% early B.1 lineage, D614G containing virus 186 ('WT'; IC19). Again, we observed that the Y453F-containing Cluster 5 was outcompeted, constituting 187 only ~10% of reads by 24 hours post-infection ( Figure 4B ). 188

Finally, to further confirm that the attenuation in human cells of the Y453F containing viruses, 189 particularly the ferret-adapted strain, wasn't due to other changes in the genome (such as E S6L 190 described above) we generated by reverse genetics (RG) two viruses on a Wuhan-hu-1, both carrying 191 the D614G mutation in spike, WT (D614G), while the other additionally contained Y453F (D614G + 192 Y453F). As with the ferret adapted P2 virus and Cluster 5 isolate we saw that the Y453F + D614G RG 193 virus produced less infectious virus upon replication in the primary human airway cells as compared 194 to the otherwise isogenic WT (D614G) virus, significantly so at 24 hours post-infection ( Figure 4C ). 195

To investigate whether a mustelid-adapted SARS-CoV-2 crossing back into the human 197 population would have a large impact on re-infections or vaccine-breakthrough we next tested 198 whether the mutation at Y453F facilitated escape from antibody neutralization. Surprisingly, Y453F-199 containing 'Ferret P2' virus was significantly more easily neutralised by convalescent first wave 200 antisera than wild type requiring only 0.6 as much antisera for a 50% neutralisation titre ( Figure 5A ). 201

We further investigated the relative antigenicity of Y453F, this time using the above-described RG 202 viruses and antisera from health care workers who had received two doses the of Pfizer-BioNtech-203 J o u r n a l P r e -p r o o f BNT162b2 vaccine. Again, we saw the Y453F-containing virus was more readily neutralised by 7 of the 204 10 vaccinee sera, although the difference was not significant ( Figure 5B ). 205

We next performed pseudovirus neutralisation assays with the previously described first wave 206 convalescent antisera against pseudoviruses expressing the common mustelid adaptations or with full 207

Cluster 5 spike. The B.1.351 (Beta) spike showed a significant, ~5-fold drop in mean NT50 ( Figure 5C Many circulating variants of concern show a greater ability to enter via mustelid ACE2. 211

Following worldwide circulation of SARS-CoV-2, a number of 'variant of concern' and 'variant 212 To investigate whether these variants may be more able to infect mink or ferrets than the 219 progenitor lineage B or B.1 viruses through better use of mustelid ACE2, we again used pseudoviruses 220 expressing these variant spike proteins and normalised entry to human ACE2 ( Figure 6 ). We found 221 that nearly all variants of concern tested could better utilise mink ACE2 than WT (D614G only) Brazil) showed no improved usage of ferret ACE2. It appears L452R, E484K and N501Y may promote 229 use of ferret ACE2, while K417N/T may result in a greater reduction in ferret ACE2 usage relative to 230 human ACE2. Overall, these data suggest multiple circulating variants of concern may be able to infect 231 mustelid hosts with only minimal, or indeed without, further adaptation. 232

In this study, we have performed a full risk assessment of mustelid hosts, such as mink and 234 ferrets, as reservoirs for the emergence of antigenic variants or new variant of concern. We have 235

shown earlier strains of SARS-CoV-2 are poorly adapted to mustelid ACE2 and therefore quickly gains 236 adaptations, such as Y453F, N501T or F486L to utilise mustelid ACE2. However, Y453F in particular, 237 negatively impacts replication kinetics of SARS-CoV-2 in human cells, potentially explaining why the 238 Danish mink-origin outbreaks did not propagate further following the culling of the mink. to remain effective against mustelid-adapted strains. Finally, we have shown that several VOC strains, 242 or VOC-associated mutations, partially adapt SARS-CoV-2 spike to mustelid ACE2. Therefore, it is likely 243 VOC lineages will continue to infect mink farms and risk spilling back over into humans. 244

Except for the Danish mink-adapted SARS-CoV-2 spillback, Y453F is found rarely in humans 245 with very few isolates reported in GISAID and only a single report of the mutation arising in 246 immunocompromised patients -this is despite Y453F having been shown in several studies to 247 enhance human ACE2 binding, in a similar manner to the VOC-associated mutations N501Y or L452R 248 suggesting a significant reporting lag. None of the four WHO-designated variants of concern have yet 260 been associated with mink farm outbreaks. It remains to be seen whether these VOCs would replicate 261 in mink/ferrets without any further adaptation, but we have shown that the most common mustelid 262 adaptations would be unlikely to have a large effect on VOC antigenicity. It will be key in the coming 263 years to continue to closely survey farmed mink and to sequence and share any SARS-CoV-2 genomes 264 from these animals in a timely manner as SARS-CoV-2 could still adapt in unexpected ways in mink 265 (Goldhill and Barclay, 2021). 266

This work also suggests that, particularly when investigating spike RBD mutants, ferrets (or 267 indeed mink) are poor models for humans, as mustelid ACE2 is poorly utilised by non-adapted SARS-268

CoV-2 spike. Thus, it is not a given that adaptation to human ACE2 will also result in increased 269 infectiousness, transmissibility or pathogenicity in the ferret model. However, ferrets remain a useful 270 model for investigating non-RBD phenotypes though care should be taken to use previously ferret 271 adapted viruses to prevent rapid adaptation. 272

Limitations of the study 273 Although this work has taken a multidisciplinary approach with experiments often performed 274 in parallel over different sites a number of limitations remain. One of which is the overreliance on 275 ferrets and ferret ACE2 as a substitution for mink/mink ACE2. Another limitation is that only Y453F 276 containing viruses were followed up for in depth virological study, therefore this work cannot make 277 J o u r n a l P r e -p r o o f strong conclusions about the zoonotic potential of the other mink-adaptations that often appear, 278 particularly N501T which has subsequently appeared on several human SARS-CoV-2 variant lineages. 279

Finally, this work has used entry of pseudovirus in ferret ACE2 overexpressing cells as a proxy for ACE2 280 usage rather than quantitively measuring the direct interaction between Spike RBD and ACE2 binding. 

Further information and requests for resources and reagents should be directed to and will 551 be fulfilled by the lead contact, Wendy Barclay (w.barclay@imperial.ac.uk). 552

J o u r n a l P r e -p r o o f Novel plasmids generated in this study are available upon request to the lead contact without 554 restriction (for academic research) or with a completed Materials Transfer Agreement (for commercial 555 work). 556

Raw sequencing reads have been deposited to the NCBI SRA and are publicly available as of 558 the date of publication. Accession numbers for the associated Bioproject are listed in the key resources 559 Any additional information required to reanalyze the data reported in this paper is available 562 from the lead contact upon request 563

Ferrets (Mustela putorius furo) -All ferret studies were performed in a containment level 3 565 laboratory, using a bespoke isolator system (Bell Isolation Systems). Outbred female ferrets (16-566 20 weeks old) weighing 750-1,000 g were used. 

The early SARS-CoV-2 strain, England/2/2020 (VE6-T) was previously isolated by Public Health 594

At ICL, 100 mm dishes of 293Ts were transfected using 678 lipofectamine 3000 (Thermo) with a mixture of pCSFLW, pCAGGS-GAGPOL and spike proteins 679 expressed in pcDNA3.1. After 24 h supernatant was discarded and replaced

and 72 h post-transfection, passed through a 0.45 µm 681 filter, aliquoted and frozen at -80 o C. At the Pirbright Institute pseudovirus was

Cells were transfected using polyethyleneimine (PEI) with a mixture of pCSFLW, p8

SARS-CoV-2 spikes expressed in pcDNA3.1. As before supernatant was discarded and replaced at 24 h 684 post-transfection then harvested and pooled at 48 and 72h. Supernatant was clarified by low-speed 685 centrifugation

Pseudovirus assays at ICL were performed as previously described

Briefly 10mm diameter dishes of 293T cells were transfected with 1 µg of ACE2 of empty vector using 688

24 h later cells media was replaced, and cells were resuspended by scraping and 689 plated into 96 well plates and overlayed with pseudovirus. 48 h later cells were lysed with reporter 690 lysis buffer (Promega) and assays were read on a FLUOstar Omega plate reader (BMF Labtech) using 691 the Luciferase Assay System

At Pirbright assays were performed largely as previously described

BHK-21 cells were transfected with 500 ng of ACE2 or empty vector (pDISPLAY) using TransIT-694

Mirus Bio) according to the manufacturer's recommendation. 24 h later, media was removed, and 695 cells were harvested following the addition of 2mM EDTA in PBS, resuspended in DMEM and plated 696 into white-bottomed 96 wells plates (Corning)

Glo substrate (Promega) diluted 1:2 with PBS and read on a GloMax Multi+ Detection System 699 (Promega). CSV files were exported onto a USB flash drive for analysis

Antisera/pseudovirus mix 702 was then incubated at 37 o C for 1 h then overlayed into 96 well plates of 293T-ACE2 cells. Assays were 703 then lysed and read as described above

Cell-cell fusion assay

Cell-cell fusion assays were performed as described elsewhere

WT or mutant SARS-CoV-2 spike proteins. BHK-21 target cells stably 708 expressing rLuC-GFP-8-11 (target cells) were co-transfected with ACE2 expression constructs. Target 709 cells were co-cultured with effector cells 24 h post-transfection and quantification of cell-cell fusion 710 was performed 24 h later with the Renilla luciferase substrate

Luminescence was read on a Glomax Multi+ Detection System (Promega). CSV files were exported on 712 a USB flash drive for analysis

Ferret infection study

Outbred female ferrets (16-717 20 weeks old) weighing 750-1,000 g were used. Four donor ferrets were inoculated intranasally with 718 200 μl of 10 5 p.f.u. of each virus while lightly anaesthetized with ketamine (22 mg kg−1)

Prior to the start of the study ferrets were confirmed to be seronegative to SARS-CoV-2. All 721 animals were nasal-washed daily, while conscious, by instilling 2 ml of PBS into the nostrils; the 722 expectorate was collected into disposable 250-ml sample pots. Ferrets were weighed daily post 723 , and body temperature was measured daily via subcutaneous IPTT-300 transponder

RNA extraction and sequencing

RNA was reverse transcribed using Superscript IV (Invitrogen) and PCR of the spike was 728 performed using KOD polymerase (Merck). For next generation sequencing RNA from virus-containing 729 samples were extracted using the QIAsymphony DSP Virus/Pathogen mini kit (Qiagen)

DNase-treated using the TURBO-free Kit (Invitrogen; (AM1907). cDNA was synthesised using the 731

NEB) before amplification by 732 the ARTIC Network protocol using the multiplexed primer scheme version 3. Fast5 files were 733 basecalled with guppy (v.5.0.7) with high accuracy calling (hac). The fastq files produced by Nanopore 734 sequencing were filtered with lengths 400 and 700 using Artic-ncov2019 pipeline v1

-medaka --medaka-model 737 r941_min_high_g360 --normalise 0" parameters was then used to process the filtered fastq files to 738 generate ARTIC V3 primer trimmed bam files and consensus genome sequences

Phylogenetic analysis

A table 746 of accession IDs and acknowledgement is given in Supplementary Table S1. A sequence with only 397 747 nucleotides (hCoV-19/mink/Spain/NV-2105/2021, EPI_ISL_1490748) was excluded from analysis. to the Wuhan-Hu-1 reference genome sequence (MN908947

Seven further sequences were excluded from further analysis as they lacked nucleotide data enabling 751 the determination of amino acid identity at spike positions 453

19/mink/Netherlands/NB-EMC-40-4/2020, EPI_ISL_577788; hCoV-756 19/mink/Denmark/mDK-56/2020, EPI_ISL_641448). Epidemiological lineages were determined using 757

Phylogenetic analysis was performed using the remaining 936 mink 759 genomes rooted on the Wuhan-Hu-1 reference genome (MN908947) with a general time reversible 760 model of nucleotide substitution, a proportion of invariant sites estimated from the data and a gamma 761 distribution describing among-site rate variation (GTR + I + ) built using

2014) with the phylogeny rooted on the sequence of the virus Wuhan-Hu-1. The maximum likelihood 763 phylogeny was plotted, alongside data on sampling location extracted from the virus name and amino 764 acid identity at spike positions 453, 486, 501 and 614

Quantification and statistical analysis

Statistics throughout this study were performed using one-way analysis of variance (ANOVA)

No statistical method was used to 768 predetermine sample size. Several genome sequences were manually removed from the phylogenetic 769 analysis and were described in the associated sections. The experiments were not randomized, and 770 the investigators were not blinded to allocation during experiments and outcome assessment

Binding Domain Reveals Constraints on Folding and ACE2 Binding. Cell 182, 1295-1310.e1220. 449 Tegally, H., Wilkinson, E., Giovanetti, M., Iranzadeh, A., Fonseca, V., Giandhari, J., Doolabh, D., Pillay, 450 S., San, E.J., Msomi, N., et al. (2021) . Detection of a SARS-CoV-2 variant of concern in South Africa. 451Nature 592, 438-443. 452 Thakur, N., Conceicao Here Zhou et al. show that common mink/ferret adaptations of SARS-CoV-2 enhance the use of the otherwise poorly used ferret ACE2 receptor, while weakening virus replication in human airway cells. However, many SARS-CoV-2 variants can intrinsically enter cells via ferret ACE2 implying these hosts may be more susceptible to novel strains.Highlights (89 char max): Y453F, F486L, and N501T often arise in SARS-CoV-2 Spike during ferret/mink adaptation  These mutations specifically adapt SARS-CoV-2 to use ferret ACE2  Common ferret or mink adaptions attenuate the virus in human airway cells  SARS-CoV-2 variants can use ferret ACE2 with any adaptation