Structural basis and analysis of hamster ACE2 binding to different SARS-CoV-2 spike RBDs

ABSTRACT

Pet golden hamsters were first identified being infected with the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) delta variant of concern (VOC) and transmitted the virus back to humans in Hong Kong in January 2022. Here, we studied the binding of two hamster (golden hamster and Chinese hamster) angiotensin-converting enzyme 2 (ACE2) proteins to the spike protein receptor-binding domains (RBDs) of SARS-CoV-2 prototype and eight variants, including alpha, beta, gamma, delta, and four omicron sub-variants (BA.1, BA.2, BA.3, and BA.4/BA.5). We found that the two hamster ACE2s present slightly lower affinity for the RBDs of all nine SARS-CoV-2 viruses tested than human ACE2 (hACE2). Furthermore, the similar infectivity to host cells expressing hamster ACE2s and hACE2 was confirmed with the nine pseudotyped SARS-CoV-2 viruses. Additionally, we determined two cryo-electron microscopy (EM) complex structures of golden hamster ACE2 (ghACE2)/delta RBD and ghACE2/omicron BA.3 RBD. The residues Q34 and N82, which exist in many rodent ACE2s, are responsible for the lower binding affinity of ghACE2 compared to hACE2. These findings suggest that all SARS-CoV-2 VOCs may infect hamsters, highlighting the necessity of further surveillance of SARS-CoV-2 in these animals.

IMPORTANCE

SARS-CoV-2 can infect many domestic animals, including hamsters. There is an urgent need to understand the binding mechanism of the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) variants to hamster receptors. Herein, we showed that two hamster angiotensin-converting enzyme 2s (ACE2s) (golden hamster ACE2 and Chinese hamster ACE2) can bind to the spike protein receptor-binding domains (RBDs) of SARS-CoV-2 prototype and eight variants and that pseudotyped SARS-CoV-2 viruses can infect hamster ACE2-expressing cells. The binding pattern of golden hamster ACE2 to SARS-CoV-2 RBDs is similar to that of Chinese hamster ACE2. The two hamster ACE2s present slightly lower affinity for the RBDs of all nine SARS-CoV-2 viruses tested than human ACE2. We solved the cryo-electron microscopy (EM) structures of golden hamster ACE2 in complex with delta RBD and omicron BA.3 RBD and found that residues Q34 and N82 are responsible for the lower binding affinity of ghACE2 compared to hACE2. Our work provides valuable information for understanding the cross-species transmission mechanism of SARS-CoV-2.

INTRODUCTION

Since severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) first emerged, it has continually evolved into new variants due to sustained global transmission. Among these SARS-CoV-2 variants, five variants of concern (VOCs) (alpha, beta, gamma, epsilon, delta, and omicron) have been ever since monitored very closely. Currently, the prevalence of the SARS-CoV-2 omicron variant and its sub-variants has surpassed any other VOCs to become the dominant strains. Meanwhile, omicron and its sub-variants can break through species barriers and infect more animals, such as mice and rats, by acquiring additional mutations in its viral spike (S) protein (1).

The gain of ability of a virus to bind to the receptor in another species is a prerequisite for inter-species transmission. The entry of SARS-CoV-2 into cells relies on its interaction between the S protein and human angiotensin-converting enzyme 2 (ACE2) that facilitates the viral entry and replication (2). The extended receptor-binding motif (residues 438–506) positioned in the receptor-binding domain (RBD, residues 319–541) of SARS-CoV-2 S protein contacts the bottom side of the small lobe of ACE2, the receptor for SARS-CoV, and the human coronavirus NL-63 (3–6). In many studies from our group and others, the possible broad host range of SARS-CoV-2 has been well demonstrated by assessing the binding ability of the SARS-CoV-2 RBD to the ACE2 orthologs from different species (7–10). Natural human-to-animal infections have been continually reported in various animals, such as dogs, cats, ferrets, tigers, minks, and white-tailed deer (11–15). In experimental challenges, the high susceptibility of multiple animals (hamsters, cats, dogs, minks, Egyptian fruit bats, white-tailed deer, raccoon dogs, and non-human primates) to SARS-CoV-2 has also been identified (8, 15–21).

Due to their high susceptibility to SARS-CoV-2, the hamster model of SARS-CoV-2 infection has also been developed and used to explore the pathogenesis of COVID-19 and to evaluate vaccine and therapeutic candidates (16, 22). In January 2022, a natural hamster-to-human infection of the SARS-CoV-2 VOC delta was first identified in Hong Kong (23). The source of the delta outbreak was identified as a pet hamster imported from the Netherlands, resulting in two independent zoonotic infections in humans in Hong Kong, which suggests that some companion animals can be a secondary reservoir of SARS-CoV-2 (23). A recent study first evaluated the pathogenicity of the omicron sub-variants BA.2, BA.4, and BA.5 in golden hamsters and observed less pathogenicity compared to the delta isolate (24). Further, some omicron mutations have been detected in viruses from rodents, offering the possibility that omicron may have evolved in the rodent population (25–28). Revealing the molecular mechanism of recognition of the SARS-CoV-2 variants to animal ACE2 receptors can provide valuable clues for the impact of the variants on infection and transmission, thus helpful to pre-warning of the emergence of any dangerous variants.

Here, we evaluated the binding capacity of the RBDs of the SARS-CoV-2 prototype (PT) and eight VOCs to two hamster ACE2s (golden hamster and Chinese hamster). The two hamster receptors bound to the SARS-CoV-2 RBDs with slightly lower binding affinity than hACE2 and mediated similar transduction efficiency for all SARS-CoV-2 pseudoviruses into BHK-21 cells expressing hamster ACE2s and hACE2, respectively. Next, we determined the structures of RBDs from omicron BA.3 and delta strains, which can infect golden hamsters in real life, in complex with golden hamster ACE2 (ghACE2) at 2.6 and 3.0 Å, respectively. Our work sheds light on the potential susceptibility of hamsters to all SARS-CoV-2 variants, calling for further surveillance of SARS-CoV-2 from hamsters.

RESULTS

Comparison of the binding between hamster ACE2s and the RBDs of different SARS-CoV-2 viruses

To explore the transmission potential of different SARS-CoV-2 strains to hamsters, we investigated the binding capacity of the nine RBDs from the SARS-CoV-2 PT and VOCs (alpha, beta, gamma, delta, BA.1, BA.2, BA.3, and BA.4/BA.5) (Fig. S1) to the ACE2 orthologs from golden hamster and Chinese hamster. Porcine deltacoronavirus (PDCoV) RBD that uses aminopeptidase N as an invasion receptor was used as negative control.

We purified the RBDs and tested their binding to BHK-21 cells expressing hamster ACE2s with flow cytometry. We found that all of the tested SARS-CoV-2 RBDs bound to the ACE2 orthologs from golden hamster and Chinese hamster. Notably, the RBDs of the omicron sub-variants BA.1, BA.2, and BA.3 could bind to hamster ACE2-expressing BHK-21 cells with a substantially lower positive rate than that of hACE2, with 41.10%–56.12% hamster ACE2-expressing cells staining positive vs 84.23%–95.05% hACE2-expressing cells (Fig. 1).

Fig 1.

Characterization of binding of the RBDs from SARS-CoV-2 PT and the eight variants with hACE2, ghACE2, and chACE2 by FACS. The His-tagged RBDs from SARS-CoV-2 PT and the eight variants (alpha, beta, gamma, delta, BA.1, BA.2, BA.3, and BA.4/BA.5) were incubated with BHK 21 cells expressing eGFP-fused ACE2s. Anti-His/APC antibody was used to detect cells expressing His-tagged protein. Binding efficiencies are expressed as the percentage of cells positive for the SARS-CoV-2 RBDs among the ACE2-expressing cells from one representative experiment with three replicates. The PDCoV RBD was used as the negative control.

Next, the binding affinities of the two hamster ACE2s to the SARS-CoV-2 RBDs were determined by surface plasmon resonance (SPR). hACE2 was used as a positive control. As shown in Fig. 2, all three ACE2s were capable of binding to all of the SARS-CoV-2 RBDs, which was consistent with the flow cytometry results. The binding affinity of ghACE2 to the SARS-CoV-2 RBDs was similar to that of Chinese hamster ACE2 (chACE2). However, the binding affinity of the hamster ACE2s to the RBDs of SARS-CoV-2 PT and the alpha, beta, gamma, delta, and BA.4/BA.5 variants was approximately twofold to fivefold weaker than that of hACE2. In terms of the RBDs of the omicron sub-variants (BA.1, BA.2, and BA.3) that used to be the dominant strains at the early stage of the omicron pandemic, the hamster ACE2s display approximately 6- to 15-fold weaker binding affinity than hACE2.

Fig 2.

Binding affinity assays between the RBDs from SARS-CoV-2 PT and the eight variants with hACE2, ghACE2, and chACE2 by SPR. (A) The mFc-tagged hACE2, ghACE2, and chACE2 were individually captured by anti-mouse IgG (mIgG) antibodies immobilized on the CM5 chip and then sequentially tested for binding with serially diluted RBDs. The raw and fitted curves are displayed in dotted and solid lines, respectively. (B) The binding affinity between the different RBDs and the three ACE2s is shown. Mean ± SD represents the mean and standard deviation of three independent experiments.

Cell entry of SARS-CoV-2 variant pseudoviruses mediated by hamster ACE2s

Next, to determine the efficiency of different SARS-CoV-2 variants infecting cells via binding to hamster ACE2s, nine vesicular stomatitis virus (VSV)-based pseudoviruses, including the SARS-CoV-2 PT and eight variants, were prepared. We transfected BHK-21 cells with plasmids containing enhanced green fluorescent protein (EGFP)-tagged ACE2s (hACE2, ghACE2, and chACE2, respectively). The EGFP-positive cells were sorted into 96-well plates and then incubated with similar amounts of all nine pseudotyped SARS-CoV-2 viruses (as determined by quantitative real-time PCR). Non-transfected BHK-21 cells were used as a negative control.

We found that ACE2s from both golden hamster and Chinese hamster supported SARS-CoV-2 pseudotyped virus entry into ACE2-expressing BHK-21 cells, whereas non-transfected BHK-21 cells did not (Fig. 3). Notably, the pseudoviruses of the SARS-CoV-2 PT and the eight variants displayed similar efficiencies for infecting BHK-21 cells expressing ghACE2, chACE2 and hACE2, with the delta strain displaying the highest efficiency.

Fig 3.

Infectivity of the pseudoviruses of SARS-CoV-2 to BHK-21 cells expressing hamster ACE2s. BHK-21 cells expressing different ACE2s and untransfected BHK-21 cells were infected with the pseudotyped SARS-CoV-2 PT and eight variants (alpha, beta, gamma, delta, BA.1, BA.2, BA.3, and BA.4), respectively. Untransfected BHK-21 cells were used as a negative control. The data represent the results of three replicates. All data are presented as the mean ± SD.

Complex structures of ghACE2 bound to the RBDs of the delta variant and omicron sub-variant BA.3

To further investigate the molecular basis for the interaction between hamster ACE2 and SARS-CoV-2 RBDs, we solved the complex structures of delta RBD-ghACE2 (Fig. S4) and BA.3 RBD-ghACE2 (Fig. S5). The density maps of delta S-ghACE2 and BA.3 S-ghACE2 were determined by cryo-EM reconstructions. We further focus refined the stably associated delta RBD-ghACE2 region to 3.0 Šresolution and the BA.3 RBD-ghACE2 region to 2.6 Šresolution (Table S1). The cryo-EM structures clearly show the binding details of the RBD-ghACE2 interfaces (Fig. 4). Compared to the delta RBD-hACE2 complex, the delta RBD-ghACE2 and BA.3 RBD-ghACE2 cryo-EM structures exhibit overall similar conformations, with root mean square deviations of 1.108 Š(765 Ca atoms) and 1.073 Šfor (759 Ca atoms), respectively. Overall, the delta RBD-ghACE2 and BA.3 RBD-ghACE2 complexes buried a smaller surface area than that of hACE2, 1603 Ų vs 1512 Ų vs 1728 Ų, respectively (Fig. 5A through C), which explains the weaker binding of ghACE2 to the delta and BA.3 RBDs.

Fig 4.

The complex structures of ghACE2 bound to SARS-CoV-2 delta RBD and BA.3 RBD. (A) The overall complex structures of hACE2 bound to the delta RBD, ghACE2 bound to the delta RBD, and ghACE2 bound to BA.3 RBD. The binding between the RBDs and ACE2s is mainly composed of two patches of interactions, with patch 1 and patch 2 indicated by dashed boxes. hACE2, ghACE2, delta RBD, and BA.3 RBD are colored in pale green, cyan, magenta, and yellow, respectively. (B and C) Detailed interaction of hACE2 with the delta RBD, ghACE2 with the delta RBD, and ghACE2 with the BA.3 RBD in patch 1 and patch 2. Residues involved in the interaction are labeled, and H-bonds and salt bridges are shown as yellow dotted lines with a cutoff of 3.5 Å.

Fig 5.

Interface comparison among ACE2 orthologs binding to the RBDs and identification of determinants of lower affinity for ghACE2. (A–C) Binding interface of hACE2 with delta RBD (A), ghACE2 with delta RBD (B), and ghACE2 with BA.3 RBD (C) with contacting residues labeled accordingly. (D) Structural comparison of delta RBD-ghACE2 with delta RBD-hACE2 and BA.3 RBD-hACE2 with BA.3 RBD-ghACE2. Residue substitutions of delta RBD-hACE2, delta RBD-ghACE2, BA.3 RBD-hACE2, and BA.3 RBD-ghACE2 are colored in pink, green, magenta, and cyan, respectively. Involved residues are shown as sticks in corresponding colors. H-bonds and salt bridges are represented by red dashes. (E) The hydrophobic surface of ghACE2 was altered by the N82M substitution. The structures of delta RBD-hACE2, delta RBD-ghACE2, BA.3 RBD-hACE2, and BA.3 RBD-ghACE2 are shown in surface representation and colored from yellow for the most hydrophobic region, to white, to teal for the most hydrophilic region.

Key residues contributing to hydrogen bond (H-bond, 3.5 Å cutoff) and van der Waals (vdw) interactions (4.5 Å cutoff) between ghACE2 and the two RBDs were also identified and labeled, respectively (Table S2). The ghACE2 displays a smaller binding interface that directly interacts with the delta and BA.3 RBDs than that of hACE2 bound to the delta RBD, forming a total of 149 and 165 atomic contacts, including five or six H-bonds, respectively (Table S2).

The binding between the RBDs and ghACE2 is distributed on two patches (Fig. 4A). The interacting network of the RBDs in patch 1 is substantially different between ghACE2 and hACE2 (Fig. 4B). For the delta RBD-ghACE2 complex, there is no potential H-bond or salt bridge contact in the patch 1 interaction network. In patch 2, the residues Y449, G496, Q498, N501, and G502 from the extending loop of the external domain of the delta RBD form multiple hydrogen bonds with residues from α1 (D38, Y41, and Q42) and K353 from the β-hairpin (Fig. 4B). For the BA.3 RBD-ghACE2 complex, there are only two potential H-bonds (R453 and Q493 from the BA.3 RBD interact with Q34 and E35 from ghACE2) in the patch 1 interaction network. In patch 2, the residues Y449, R498, and G502 from the BA.3 RBD form four H-bonds with residues D38, Q42, and K353 from ghACE2 (Fig. 4B).

Although there are multiple different residues between the hACE2 and ghACE2 coding sequences (Fig. S2), there are only two different residues (Q34 and N82) in ghACE2 compared to hACE2 at the binding surface (Fig. S3). In particular, the Q34 found in most rodent ACE2 orthologs likely weakens the interaction with the delta RBD by losing an H-bond and a panel of vdw interactions (Fig. 4B; Table S2). In contrast, Q34 of ghACE2 forms 11 vdw interactions with the residues (Y453, R493, and S494) of the BA.3 RBD, whereas H34 of hACE2 forms twofold vdw interactions with the residues (L455, Y453, N417, and R493) of BA.3 RBD compared to the Q34 of ghACE2 (Fig. 5D; Table S2). The N82-linked glycan of ghACE2 was identified in the delta RBD-ghACE2 and BA.3 RBD-ghACE2 cryo-EM structures and could weaken its binding to the RBDs by interfering with the hydrophobic environment (Fig. 5E). Together, these clues suggest that the variant residues in ghACE2 contribute to the decreased binding of ghACE2 to SARS-CoV-2 RBDs.

Identification of key residues responsible for the decreased binding of ghACE2 to SARS-CoV-2 RBDs

Next, we explored the mechanism that leads to the weaker binding of ghACE2 to SARS-CoV-2 RBDs compared to hACE2, especially the amino acids vital to the binding of the RBDs of dominant SARS-CoV-2 variants (delta variant and the omicron sub-variants BA.1, BA.2, BA.3, and BA.4/BA.5). Therefore, we introduced the Q34H, N82M, or Q34H-N82M mutations to ghACE2 to mimic the key variant residues of hACE2. Sequence alignment of the key residues in ghACE2 involved in the interaction with the delta RBD from 28 rodent species was performed. The residues at positions 34 and 82 differ most frequently between the rodent ACE2s and hACE2 (Fig. S3). Mutational analysis was further conducted by transiently expressing ghACE2 carrying the Q34H and/or N82M mutations in human embryonic kidney (HEK) 293T cells. We found that the binding affinities of the RBDs from SARS-CoV-2 PT, delta, BA.1, BA.2, and BA.3 to ghACE2 were enhanced by approximately twofold through the introduction of the Q34H mutation. Similarly, ghACE2-N82M displayed increased binding to the six RBDs tested by approximately twofold as well (Fig. 6). Together, these results suggest that residues H34 and M82 play important roles in the receptor binding of SARS-CoV-2.

Fig 6.

Binding affinity assays between different SARS-CoV-2 RBD variants and ghACE2 mutants by SPR. (A) The mFc-tagged ACE2s from ghACE2 and its mutants (Q34H, N82M, and Q34H-N82M) were captured by anti-mIgG Fc antibodies immobilized on the CM5 chip and sequentially tested for binding to serially diluted SARS-CoV-2 PT and the eight variants (alpha, beta, gamma, delta, BA.1, BA.2, BA.3, and BA.4/BA.5). (B) The binding affinities of the eight RBDs to ghACE2 and its mutants are shown. Mean ± SD represents the mean and standard deviation of three independent experiments.

DISCUSSION

Natural human-to-animal transmissions of SARS-CoV-2 have been continually reported in different mammalian species, including dogs, cats, ferrets, tigers, minks, and white-tailed deer (11–15). However, there are fewer examples of SARS-CoV-2 being transmitted back to humans from animals, such as farmed minks in the southeastern Netherlands, cats in Thailand, and white-tailed deer in Canada (21, 29). It was first reported that pet golden hamsters could acquire SARS-CoV-2 infection in real-life settings and infect humans in Hong Kong (23). The SARS-CoV-2 host range is mostly genetically determined by the interaction of S protein with ACE2 (3, 9, 30–32), and we, therefore, determined the molecular basis of cross-species recognition of the SARS-CoV-2 delta variant and omicron BA.3 S protein to ghACE2.

In addition to supporting replication of SARS-CoV (33, 34) and some other respiratory viruses, including human metapneumovirus (35), human parainfluenza virus 3 (36), and influenza A virus (37), the hamster model has also been developed to explore the pathogenesis of COVID-19 and evaluate potential vaccine and treatment candidates (16, 22). It has been reported that golden hamsters are permissive for infection by the SARS-CoV-2 PT, delta, and omicron BA.1 strains, and both delta and BA.1 display similarly high transmissibility via contact and non-contact transmission (38). The omicron BA.4 and BA.5.2 strains (GISAID: EPI_ISL_14035969 and EPI_ISL_14035968) have recently been newly detected in golden hamsters. Consistently, we identified that the ACE2s of golden hamster and Chinese hamster could efficiently bind to the RBDs of SARS-CoV-2 PT and eight VOCs (alpha, beta, gamma, delta, BA.1, BA.2, BA.3, and BA.4/BA.5). Moreover, VSV-based pseudoviruses incorporating the S protein of different SARS-CoV-2 variants could invade BHK-21 cells expressing hamster ACE2s and hACE2 with similar transduction efficiency, suggesting golden hamsters and Chinese hamsters are susceptible to all SARS-CoV-2 strains, not just the delta and omicron variants that have been documented (38).

It has been reported that multiple ACE2 mutants have higher or lower affinity with the RBD of SARS-CoV-2 S protein, thereby promoting or resisting SARS-CoV-2 infection (6, 32). Recently, a study identified a series of ACE2 mutations, including those validated experimentally, occurred in the patient, and reported in cell lines, and their effects on SARS-CoV-2 infection (6). In this study, structural analysis revealed that the Q34 and N82 residues of ghACE2 are involved in RBD binding as part of patch 1, which displays a distinct interaction pattern with the delta RBD and BA.3 RBD compared to that of hACE2. Sequence analysis of ACE2s from various rodents revealed that the residues at positions 34 and 82 are highly polymorphic, with Q34 and N82 existing in many rodents (Fig. S3). In addition to the monoclonal antibodies, recombinant ACE2s with high affinity for SARS-CoV-2 RBD can also be used as the RBD blockers against COVID-19 (39, 40). Our recent study generated a binding-enhanced hACE2 as a therapeutic candidate against SARS-CoV-2 variants (40). However, many designed mutants of hACE2, including H34F, H34W, M82F, and M82W, weaken its binding affinity for the SARS-CoV-2 RBD in spite of their enhanced hydrophobicity or electrification (40), indicating the critical roles of H34 and M82 in the receptor binding of SARS-CoV-2. Here, our mutational analyses suggest that Q34 and N82 in ghACE2 are responsible for the decreased binding of ghACE2 to the RBDs from different SARS-CoV-2 strains compared to hACE2. The substitution of Q34 with histidine enhanced the capacity of ghACE2 to interact with the delta RBD and BA.3 RBD. The potential glycosylated N82 exists in the ACE2s of golden hamster (NYS), Chinese hamster (NYS), and rat (NFS). Indeed, the N82-linked glycosylation sites were identified in our cryo-EM structures of the delta RBD-ghACE2 and BA.3 RBD-ghACE2 complexes. The ghACE2-N82M mutant displayed stronger binding to the tested SARS-CoV-2 RBDs than that of the wild-type counterparts. The stronger binding capacity may be due to that the N82-linked glycan in ghACE2 would disrupt (by steric interference) a hydrophobic contact between M82 and F486.

Due to the high mutation rate of SARS-CoV-2, numerous variants have emerged and worsened the situation of the pandemic (https://www.cdc.gov/coronavirus/2019-ncov/variants/). As reported, omicron variants evolved independently of all other SARS-CoV-2 variants, and a possible animal origin was proposed to be related to rodents such as mice and rats (41). Three mutations (Q493R, Q498R, and N501Y) in omicron variants are responsible for the high affinity to mouse ACE2 but cannot bind to the PT RBD protein (1, 42). As rodents, golden hamster and Chinese hamster displayed approximately 6- to 15-fold lower binding affinity for their ACE2 receptors to the RBDs of omicron BA.1, BA.2, and BA.3 than that of hACE2. Moreover, several studies report that the golden hamster and Chinese hamster are susceptible to SARS-CoV-2 infection, including SARS-CoV-2 PT, delta variant, and omicron variants, and display clinical symptoms and histological changes (24). Although not always the case, natural reservoirs frequently co-evolve with their viruses and rarely show clinical symptoms (43–45). Thus, the hamsters are more likely to be the victims of omicron variants rather than to be the animal reservoir in which the omicron mutations accumulated. Hamsters are distributed worldwide as popular pet animals, so the high risk of SARS-CoV-2 infection in pet hamsters deserves our attention. Furthermore, the trade in pets may provide a route for the spread of SARS-CoV-2 across international boundaries. Future surveillance of SARS-CoV-2 and its variants in hamsters is crucial to prevent new variants that develop from long-term SARS-CoV-2 infection in hamsters from spilling back to humans.

MATERIALS AND METHODS

Cells

The HEK Expi293F cells (Gibco) were cultured at 37°C in SMM 293-TII Expression Medium with 5% CO₂ in a shaking incubator (140 rpm). BHK-21 cells (ATCC, ATCC CCL-10) and HEK 293T (ATCC CRL-3216) were cultured at 37°C in Dulbecco’s Modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum (FBS) at 5% CO₂.

Gene cloning

The S proteins and RBDs of the SARS-CoV-2 PT and the variants (alpha, beta, gamma, delta, BA.1, BA.2, BA.3, and BA.4/BA.5) used in SPR assays, flow cytometry (FACS), and cryo-EM analyses were expressed using Expi 293F cells. The cDNAs encoding the S protein RBDs from the SARS-CoV-2 PT and the variants (residues R319-F541) were synthesized with optimized codons and were cloned into pCAGGS vectors with a C-terminal six histidine tag. Similarly, ghACE2 (residues 19–615) used for cryo-EM analysis was also cloned into pCAGGS with a C-terminal six histidine tag using the EcoRI and XhoI restriction sites. Gene sequence encoding delta S protein (residues 1–1,205) and omicron BA.3 S protein (residues 1–1,205) were synthesized with optimized codons and were cloned into pCAGGS vectors with a Strep-II tag and a His-tag. The “6P” mutations (F814P, A889P, A896P, A939P, K983P, and V984P) were introduced to stabilize the profusion state (46).

The full-length coding sequences of hACE2 (GenBank accession number: BAJ21180), ghACE2 (GenBank accession number: XP_005074266), chACE2 (GenBank accession number: XP_003503283), and three mutants of ghACE2 (Q34H, N82M, and Q34H-N82M) were cloned into pEGFP-N1 vectors used for FACS assays. The extracellular domain of hACE2, ghACE2, chACE2, and the three ghACE2 mutants fused to the Fc domain of mouse IgG (mFc) were separately cloned into pCAGGS vectors and used for SPR assays.

Protein expression and purification

The RBDs of the SARS-CoV-2 PT and the variants, ghACE2, delta, and omicron BA.3 S proteins were expressed in Expi293F cells after plasmid transfection using Sinofection Transfection Reagent (Sino Biological). At 5 days post transfection, the culture supernatants were collected and filtered through 0.22-µm filters. The soluble proteins were purified using 5-mL His-Trap HP columns (GE Healthcare). Then, the RBDs and ghACE2 were further purified by gel filtration using a HiLoad 16/600 Superdex 200 pg column (GE Healthcare) with a buffer consisting of 20 mM Tris-HCl (pH 8.0) and 150 mM NaCl. The delta and BA.3 S proteins were purified by gel filtration using a Superose 6 Increase 10/300 Gl column (GE Healthcare) with the same buffer as above.

To prepare the mFc-tagged ACE2 proteins, the pCAGGS plasmids containing the coding sequences of hACE2, ghACE2, chACE2, and three ghACE2 mutants (Q34H, N82M, and Q34H-N82M) were transiently transfected into HEK 293T cells. After 72 h, the supernatants containing the ACE2-mFc proteins were harvested, concentrated, and then used for SPR assays.

Flow cytometry

For the FACS assays, plasmids containing hACE2, ghACE2, and chACE2 fused to eGFP were transfected into BHK-21 cells. The cells were collected after 24 h and washed twice with phosphate buffer saline (PBS) (with 0.5% FBS) and then separately incubated with 1 µg/mL of the test proteins (the RBDs of SARS-CoV-2 PT and the eight variants) with histidine tags at 37°C for 30 min. After washing, cells were stained for 30 min at 37°C with anti-His/APC antibodies (1:500, Miltenyi Biotec, AB_2751870). Cells were then washed three times, and FACS data were acquired on a BD FACSCanto. Binding efficiencies are expressed as the percentage of cells positive for RBD-His among the GFP-positive cells (ACE2-expressing cells). The figures were generated using FlowJo V10 software.

SPR analysis

SPR-based measurements were performed using a BIAcore8000 system (GE Healthcare) with CM5 chips (GE Healthcare) at 25°C in single-cycle mode. For all measurements, PBST [1.8 mM KH₂PO₄, 10 mM Na₂HPO₄ (pH 7.4), 137 mM NaCl, 2.7 mM KCl, and 0.005% (vol/vol) Tween 20] was used as the running buffer. The CM5 biosensor chip was first immobilized with an anti-mIgG antibody (ZSGB-BIO, ZF-0513). Concentrated supernatants containing different mFc-tagged ACE2 proteins were then individually captured by the CM5 chip pre-immobilized with anti-mouse IgG antibodies at more than 600 response units. Gradient concentrations of the RBD proteins were then run across flow cell 2, with flow cell 1 set as a control. After each cycle, the sensor chip was regenerated using glycine (pH 1.7). Binding kinetics were analyzed with Biacore Insight software (GE Healthcare) using a 1:1 Langmuir binding model. The results were then visualized using OriginPro 9.1.

Production and quantification of pseudoviruses

The pseudoviruses of SARS-CoV-2 PT and the eight variants were constructed with a mCherry-encoding replication-deficient VSV vector backbone (VSV-ΔG-mCherry) and the coding sequence of the corresponding S proteins, as previously described (47, 48). Briefly, HEK 293T cells were transfected with 30 µg of S protein expression plasmid, and the VSV‐ΔG‐mCherry pseudovirus was added 24 h later. The inoculum was removed after incubating for 1 h at 37°C. After washing the cells with PBS, the culture medium was changed into DMEM supplemented with 10% FBS and 10 µg/mL of anti-VSV-G antibody (I1Hybridoma ATCC CRL2700). The pseudoviruses were harvested 20 h post inoculation, passed through a 0.45-µm filter (Millipore, Cat#SLHP033RB), aliquoted, and stored at −80°C.

Pseudovirus infection assay

The pseudovirus particles of the SARS-CoV-2 PT and eight variants were normalized to the same amount for quantitation by qRT-PCR. BHK-21 cells were transfected with each of the pEGFP-N1-ACE2s plasmids, and 24 h later, EGFP-positive cells were sorted, reseeded in 96-well plates at 2 × 10⁴ cells/well, and cultivated for another 24 h. The BHK-21 cells were washed with PBS before adding the supernatant containing pseudovirus particles. At 15 h post transfection, imaging and analysis of fluorescent cells were performed using a CQ1 confocal image cytometer (Yokogawa, Japan). Each group contained three replicates. Untransfected BHK-21 cells were used as negative controls.

Cryo-EM sample preparation and data acquisition

For the ghACE2-bound S protein complex of delta or BA.3, the S protein was incubated with the purified ghACE2 at a 1:4 molar ratio (S trimer to ghACE2) overnight on ice, followed by purification by concentration and dialysis using an Ultracon concentrator (Millipore) with 20 mM Tris (pH 8.0) and 150 mM NaCl. A droplet (3.0 µL) of the purified complex at a concentration of 2 mg/mL was applied to glow-discharged C-flat R1.2/1.3 (300 mesh) holey carbon grids and subsequently vitrified using Vitrobot Mark IV (Thermo Fisher Scientific). For delta S/hACE2 and BA.3 S/hACE2 complex data sets, a total of 2,995 and 8,266 movies were respectively collected on a 300-kV Titan Krios transmission electron microscope equipped with a Gatan K3 detector and GIF Quantum energy filter. EPU software (Thermo Fisher Scientific) was used for automatic data collection. Movies were collected at a calibrated pixel size of 0.88 Å. The defocus range was between −1.0 µm and −2.0 µm. Each movie was dose fractionated into 32 frames with a total dose of 50 e⁻/Ų.

Image processing and 3D reconstruction

MotionCor2 (49) was used to correct the movies for drift, and contrast transfer function (CTF) parameters were determined using CTF estimation in the patch mode. Particle picking, extraction, and 2D classification were then performed using cryoSPARC (50). For the delta S/hACE2 complex, a total of 422,405 particles were extracted and used for initial reconstruction and heterogeneous refinement. Then, 141,067 particles were separated and showed ACE2 binding. This complex subset was further subjected to global and local CTF refinement as well as homogeneous refinement. Local refinement for the RBD-ACE2 region was performed and resulted in a density map at 2.96 Å resolution, which was finally sharpened by DeepEMhancer (51). For the BA.3 S/hACE2 complex, a total of 953,666 particles were extracted and used for initial reconstruction and heterogeneous refinement. Then, one subset comprising 416,754 particles was further subjected to global and local CTF refinement as well as homogeneous refinement. Local refinement for the RBD-ACE2 region was performed and resulted in a density map at 2.60 Å resolution, which was finally sharpened by DeepEMhancer (51).

Model building

The SARS-CoV-2 RBD-hACE2 structure (PDB: 6LZG) was docked into the cryo-EM density maps of the RBD/ACE2 region using CHIMERA (52), respectively. The models were manually corrected and refined iteratively using COOT and Phenix (53, 54). The stereochemical quality of each model was evaluated using MolProbity (55).

Quantification and statistical analysis

Flow cytometry analysis

All experiments were performed three times; one representative of each experiment is shown in Fig. 1.

Binding affinity analysis

K_D values of SPR experiments were obtained with BIAcore 8K Evaluation Software (GE Healthcare), using a 1:1 binding model. The values indicate the mean ± SD of three independent experiments.

DATA AVAILABILITY

The atomic structure coordinates have been deposited in the RCSB Protein Data Bank (PDB) with the accession codes 8KA8 and 8KC2. The corresponding electron microscopy maps have been deposited in the Electron Microscopy Data Bank with the accession codes EMD-37006 and EMD-37090.

SUPPLEMENTAL MATERIAL

The following material is available online at https://doi.org/10.1128/jvi.01157-23.

Supplemental material. jvi.01157-23-s0001.pdf.

Figures S1 to S5 and Tables S1 and S2.

ASM does not own the copyrights to Supplemental Material that may be linked to, or accessed through, an article. The authors have granted ASM a non-exclusive, world-wide license to publish the Supplemental Material files. Please contact the corresponding author directly for reuse.