key: cord-1052306-mwxkvwaz
authors: Li, Wei; Schäfer, Alexandra; Kulkarni, Swarali S.; Liu, Xianglei; Martinez, David R.; Chen, Chuan; Sun, Zehua; Leist, Sarah R.; Drelich, Aleksandra; Zhang, Liyong; Ura, Marcin L.; Berezuk, Alison; Chittori, Sagar; Leopold, Karoline; Mannar, Dhiraj; Srivastava, Shanti S.; Zhu, Xing; Peterson, Eric C.; Tseng, Chien-Te; Mellors, John W.; Falzarano, Darryl; Subramaniam, Sriram; Baric, Ralph S.; Dimitrov, Dimiter S.
title: High potency of a bivalent human VH domain in SARS-CoV-2 animal models
date: 2020-09-04
journal: Cell
DOI: 10.1016/j.cell.2020.09.007
sha: f47b365dd878bf22ee98ef4c1f73698a66d1145b
doc_id: 1052306
cord_uid: mwxkvwaz

Novel COVID-19 therapeutics are urgently needed. We generated a phage-displayed human antibody VH domain library from which we identified a high-affinity VH binder ab8. Bivalent VH, VH-Fc ab8 bound with high avidity to membrane-associated S glycoprotein and to mutants found in patients. It potently neutralized mouse adapted SARS-CoV-2 in wild type mice at a dose as low as 2 mg/kg and exhibited high prophylactic and therapeutic efficacy in a hamster model of SARS-CoV-2 infection, possibly enhanced by its relatively small size. Electron microscopy combined with scanning mutagenesis identified ab8 interactions with all three S protomers and showed how ab8 neutralized the virus by directly interfering with ACE2 binding. VH-Fc ab8 did not aggregate and did not bind to 5300 human membrane-associated proteins. The potent neutralization activity of VH-Fc ab8 combined with good developability properties and cross-reactivity to SARS-CoV-2 mutants provide a strong rationale for its evaluation as a COVID-19 therapeutic.

The global outbreak of a severe acute respiratory distress (SARS) coronavirus 2 (SARS-CoV-2) associated disease 2019 requires rapid identification of therapeutics and vaccines. While many vaccines are in clinical development, the time to market can be relatively long and immunogenicity can be limited for high-risk groups (Amanat and Krammer, 2020) . Alternatively and complementarily, antibodies can be used as safe and effective prophylactics and therapeutics (Pelegrin et al., 2015) . Convalescent plasma from COVID-19 patients inhibited SARS-CoV-2 infection and alleviated symptoms of newly infected patients (Casadevall and Pirofski, 2020; Rojas et al., 2020) suggesting that potent neutralizing monoclonal antibodies (mAbs) may be even more effective.

SARS-CoV-2 genome shares more than 80% homology to the SARS-CoV . Similar to SARS-CoV, SARS-CoV-2 uses the spike (S) envelope glycoprotein to enter into host cells. The viral entry is initiated by the receptor binding domain (RBD) of the S protein binding to its receptor, angiotensin-converting enzyme 2 (ACE2), leading to conformational change of the S2 subunit and formation of six helical-bundle resulting in membrane fusion between viral and host cells Yan et al., 2020) . The SARS-CoV RBD contains immune-dominant epitopes that can elicit neutralizing antibodies conferring protection to SARS-CoV infection (He et al., 2005) . A recent bioinformatics study showed that SARS-CoV-2 RBD has several B cell epitopes (Grifoni et al., 2020) . SARS-CoV-2 RBD based immunogens were able to elicit neutralizing sera in animals (Quinlan et al., 2020) . Thus, SARS-CoV-2 RBD is a good target for developing potent neutralizing mAbs. We and others have identified such potent neutralizing human mAbs targeting the RBD of SARS-CoV (Zhu et al., 2007) and the middle east respiratory syndrome coronavirus (MERS-CoV) (Ying et al., 2014a) . Recently, several groups have reported the isolation of potent neutralizing antibodies from convalescent human donors but all are in an Immunoglobulin G1 (IgG1) format with a molecular mass of about 150 kDa Ju et al., 2020; Rogers et al., 2020; Shi et al., 2020; Zost et al., 2020) .

Antibody domains and fragments such as Fab (fragment antigen binding, molecular weight of 50 kDa), scFv (singe-chain variable fragment, 30 kDa) and V H (heavy chain variable domain, 15 kDa) are attractive antibody formats as candidate therapeutics (Nelson, 2010) . For example, isotope labeled antibody fragments are more suitable for bio-imaging due to their better tissue penetration and faster clearance compared to full-size antibodies (Freise and Wu, 2015) . Single antibody domains (sAbd), e.g., camelid V H H (15 kDa) exhibit strong antigen binding and high stability (Harmsen and De Haard, 2007) . We and others have demonstrated that human IgG1 heavy chain variable domain (V H ) can be engineered to achieve high stability and affinity to antigens (Nilvebrant et al., 2016) , as exemplified by the V H , m36.4, targeting the human immunodeficiency virus type 1 (HIV-1) envelope glycoprotein co-receptor binding site (Chen et al., 2008a) . The V H domains small size could improve therapeutic efficacy for infectious diseases, such as COVID-19 because of greater penetration to sites of infection. The conformation of the SARS-CoV-2 S trimer is dynamic with only one RBD in the "up" conformation presenting neutralizing epitopes while epitopes in the other two RBDs may be masked . Small V H s may achieve binding to the cryptic RBD epitopes during the dynamic "breathing" of the S trimer . In addition, V H s may have J o u r n a l P r e -p r o o f an advantage for treatment of respiratory virus infections because V H s could efficiently penetrate tissue, especially when using direct delivery through inhalation (Detalle et al., 2016) .

To identify potent neutralizing V H s against SARS-CoV-2, we panned our large (10 11 clones) and diverse phage-displayed human V H antibody library against recombinant RBD. Several V H binders were isolated and screened for their affinities, ACE2 competition and stabilities. One of those V H s, ab8, in an Fc (human IgG1, crystallizable fragment) fusion format, showed potent neutralization activity and specificity against SARS-CoV-2 both in vitro and in two animal models. To our knowledge, this is the first report for high potency of a human antibody domain (V H ) in two animal models of infection.

We generated a large phage-displayed human V H library where heavy chain complementarity-determining regions (HCDR1, 2, 3s) were grafted into their cognate positions of a stable scaffold based on the germline V H 3-23 ( Figure S1A) . It was panned against recombinant RBD antigens with two different tags (avi-his and human IgG1 Fc tag) which were sequentially used to avoid phage enrichment to tags and related epitopes. The quality of the RBD used for panning was confirmed by ACE2 binding (Figure S1B and C). After three rounds of panning, a panel of V H binders was obtained. Among the highest affinity binders, we selected one, V H ab8, which did not aggregate during a six-days incubation at 37°C as tested by dynamic light scattering (DLS) (Figure S1D ). To increase the V H ab8 avidity and extend its in vivo half-life, it was converted to a bivalent antibody domain by fusion to the human IgG1 Fc (V H -Fc ab8) (Figure S1E ).

V H ab8 bound to SARS-CoV-2 RBD and S1 with half-maximal binding concentrations (EC 50 s) of 10 nM as measured by ELISA (Figure 1A and D) and an equilibrium dissociation constant (K D ) of 19 nM as measured by the biolayer interferometry (Blitz system) ( Figure 1B) . The relatively fast dissociation rate constant (k d = 4.1× 10 -3 S -1 ) was significantly (23-fold) decreased by the conversion to a bivalent Fc fusion format (k d = 1.8× 10 -4 S -1 ) ( Figure   1E ) resulting in high avidity. V H -Fc ab8 bound to SARS-CoV-2 RBD and S1 subunit of S protein with EC 50 s of 0.40 nM and 0.20 nM, respectively, and a K D of 0.54 nM ( Figure 1E ). It specifically bound to 293T cells expressing S, but not to control 293T cells ( Figure 1C and Figure S2A ). The binding of V H -Fc ab8 was higher than that of IgG1 CR3022, an anti-SARS-CoV antibody cross-reactive with SARS-CoV-2 (Tian et al., 2020) . The V H -Fc ab8's halfmaximal FACS measured binding concentration (FC 50 ) of 0.07 nM was higher than that of recombinant human ACE2-Fc (FC 50 = 0.52 nM) ( Figure 1F ). These data demonstrate that ab8 selected by an isolated RBD can bind to cell surface associated native S trimer. The binding of V H -Fc ab8 to the S protein was significantly improved compared to that of the V H ab8 through avidity effect.

Competition with human ACE2 for binding to RBD is a surrogate indicator for antibody neutralization activity.

V H -Fc ab8 outcompeted human ACE2-Fc with a half-maximal inhibitory concentration (IC 50 ) of 1.0 nM ( Figure   2A ). Note that the V H -Fc ab8 was much more effective in outcompeting ACE2-Fc than V H ab8, consistent with its enhanced binding. ACE2 can also block V H ab8 for binding to RBD ( Figure S2B ) and cell surface associated S ( Figure S2C) . V H -Fc ab8 also significantly decreased the kinetics of ACE2 binding as measured by Blitz ( Figure   2B ). V H -Fc ab8 did not bind to the SARS-CoV RBD ( Figure 2C ) and did not compete with CR3022 for binding to RBD ( Figure 2D ). The CR3022 epitope is located in a conserved region on the RBD core domain distal from the ACE2 binding interface, as seen in the crystal structure of the Fab CR3022-RBD complex . These results indicate that the ab8 epitope may overlap with the ACE2 binding site on RBD.

Currently, nine prevalent RBD mutants were found in COVID-19 patients (Priyanka et al., 2020) . Six of these mutations (F342L, N354D. N354D/D364Y, V367F, R408I, W436R) are located in the RBD core domain and three, K458R, G476S and V483A are in the receptor binding motif (RBM) (Figure 3A) . V H -Fc ab8 bound to all mutants similarly to wild type RBD as measured by ELISA ( Figure 3B ). To map the ab8 epitope, we also generated several mutations in non-conserved positions compared to SARS-CoV spanning the footprint of ACE2 on RBM (N439A, G446L, L455A, F456A, A475I, F486A, Q493A, Q498A, N501A, Y505A) ( Figure 3C ). Most of these mutants retained V H -Fc ab8 binding except F486A, F456A and A475I ( Figure 3D and 3E) . The F486A significantly decreased binding without affecting the overall RBD conformation (Figure S2C and S2D) indicating that F486 directly interacts with ab8. The F456A and A475I mutations decreased the binding by 15% and 40%, respectively, but they also affected the RBD conformation ( Figure S2C and S2D) . These results suggest that a portion of the V H ab8 epitope could be in the RBM distal loop tip where the F486 is located at ( Figure 3F ).

To explore structural aspects of SARS-CoV-2 neutralization by V H ab8, we performed negative stain electron microscopic analysis of the complex formed between the S protein ectodomain and V H ab8 or soluble ACE2 (Figure 4) . The density maps showed that both V H ab8 and ACE2 were in a quaternary conformation in which two of the protomers in the trimer are in the "down" conformation with the third one in the "up" conformation ( Figures   4A and 4B) , similar to the quaternary conformation of the reported ACE2-bound S ectodomain (PDB ID: 6VYB) (Walls et al., 2020) . One molecule of the V H ab8 was observed bound to each RBD domain ( Figure 4A ). In the ACE2-S complex, one molecule of ACE2 was bound to the S protein trimer, straddling one "up" and one "down" RBD region ( Figure 4B ). There appears to be a noticeable shift of the "up" RBD domain when it is bound to V H ab8 ( Figure 4A ). This shift is not observed when ACE2 is bound to the trimer ( Figure 4B) . Superposition of the two density maps reveals that the binding site of V H ab8 directly overlaps with the ACE2 one, precluding simultaneous occupancy on the S protein ectodomain ( Figure 4C ). We also found that when ACE2 was added subsequent to the J o u r n a l P r e -p r o o f addition of V H ab8, only the V H ab8 bound state was observed, further confirming the ACE2 competition with V H ab8. To better understand the spatial relationship between the site of V H ab8 binding and that of ACE2 binding, we created a molecular model for ACE2 bound S trimer by aligning the RBD region of the crystal structure of SARS-CoV-2 RBD bound ACE2 (PDB ID: 6M0J) (Lan et al., 2020) to the "up" RBD region in the cryo-EM structure of the trimer (PDB ID: 6YVB) (Wrapp et al., 2020) . Superposition of this chimeric structure with the density map of V H ab8-bound S protein trimers reveals that the bound ACE2 has extensive overlap with the space occupied by bound V H ab8 ( Figure 4D ). The direct spatial overlap between bound V H ab8 and ACE2 provides a structural mechanism for the observed effect of ab8 on blocking ACE2 binding. The structural findings also showed that the RBM distal loop, which has F486 at its tip, is directly covered by the footprint of the bound V H ab8, consistent with the epitope mapping results showing that F486 is a direct contacting residue for ab8.

We used four different assays to evaluate V H -Fc ab8 mediated inhibition of SARS-CoV-2 infection in vitro: a βgalactosidase (β-Gal) reporter gene-based quantitative cell-cell fusion assay (Xiao et al., 2003) ; an HIV-1 backbonebased SARS-CoV-2 pseudovirus assay ; and two different replication-competent virus neutralization assays (a luciferase reporter gene assay and a microneutralization (MN)-based assay) (Scobey et al., 2013; Yount et al., 2003) . V H -Fc ab8 inhibited cell-cell fusion much more potently than V H ab8 ( Figure 5A ). The inhibitory activity of V H -Fc ab8 was also higher than that of ACE2-Fc. The control anti MERS-CoV antibody IgG1 m336 did not show any inhibitory activity. V H -Fc ab8 neutralized pseudotyped SARS-CoV-2 virus (IC 50 = 0.03 µg/ml) more potently than ACE2-Fc (IC 50 = 0.40 µg/ml) and V H ab8 (IC 50 = 0.65 µg/ml) ( Figure 5B ). The pseudovirus neutralization IC 50 for ACE2-Fc in our assay is comparable to the one reported by Changhai Lei et al.

(0.03-0.1 µg/ml) (Lei et al., 2020) . Interestingly, the maximum neutralization by V H ab8 was only 50% compared to the 100% by V H -Fc ab8 and ACE2-Fc, which was also observed for another antibody S309 (Pinto et al., 2020) . The complete neutralization by V H -Fc ab8/ACE2-Fc emphasizes the role of bivalency and related avidity in neutralization (Klasse and Sattentau, 2002) . Furthermore, in the reporter gene assay V H -Fc ab8 neutralized live SARS-CoV-2 with an IC 50 of 0.04 µg/ml ( Figure 5C ), which is much lower than that for ACE2-Fc (IC 50 of 6.1 µg/ml) and V H ab8 (IC 50 = 29 µg/ml). ACE2-Fc seemed to be much less potent against the live virus compared to the pseudovirus, which is also observed by others (IC 50 = 12.6 µg/ml) and may relate to the S expression levels and RBD/S conformation on the virus surface. We also confirmed the high V H -Fc ab8 live virus neutralization potency by a microneutralization (MN) assay-100% neutralization (NT 100 ) at 0.1 µg/ml ( Figure 5D ).

The NT 100 from the MN assay (0.1 µg/ml) was close to the IC 100 (0.2 µg/ml) from the reporter gene assay suggesting consistency in the live virus neutralizing activity of V H -Fc ab8 obtained with two independent assays at two different laboratories. These results suggest that V H -Fc ab8 is a potent neutralizer of SARS-CoV-2, which correlates with its strong competition with ACE2 for binding to RBD.

To evaluate the prophylactic efficacy of V H -Fc ab8 in vivo, we used a recently developed mouse ACE2 adapted SARS-CoV-2 infection model, in which wild type BALB/c mice are challenged with SARS-CoV-2 carrying two J o u r n a l P r e -p r o o f mutations Q498T/P499Y at the ACE2 binding interface in the RBD . It was shown that in this model, the aged BALB/c mice exhibited more clinically relevant phenotypes than those seen in hACE2 transgenic mice . Groups of 5 mice each were administered 36, 8, 2 mg/kg V H -Fc ab8 prior to high titer (10 5 pfu) SARS-CoV-2 challenge followed by measurement of virus titer in lung tissue 2 days post infection. V H -Fc ab8 effectively inhibited SARS-CoV-2 in the mouse lung tissue in a dose dependent manner ( Figure 6A ). There was complete neutralization of infectious virus at the highest dose of 36 mg/kg, and statistically significant reduction by 1000-fold at 8 mg/kg. Remarkably, even at the lowest dose of 2 mg/kg it significantly decreased virus titer by 10fold (two tailed, unpaired t test, p = 0.0075). To exclude possible effects of residual ab8 on viral titration, we performed another experiment in which mouse lungs were perfused with 10 ml of PBS before harvesting for titration.

The perfusion did not affect to any significant degree the infectious virus in the lungs ( Figure 6B ). The V H -Fc ab8 completely neutralized the virus in the lungs at 36 mg/kg and significantly reduced infectious virus at 8 mg/kg. V H -Fc ab8 also reduced viral RNA in the lungs ( Figure 6C ). These results demonstrate the neutralization potency of V H -Fc ab8 in vivo. They also suggest that the double mutations Q498T/P499Y on RBD did not influence V H -Fc ab8 binding and contribute to the validation of the mouse adapted SARS-CoV-2 model for evaluation of neutralizing antibody efficacy.

Recently hamsters were demonstrated to recapitulate clinical features of SARS-CoV-2 infection (Chan et al., 2020) (Imai et al., 2020) . To evaluate the V H -Fc ab8 efficacy in hamsters, it was intraperitoneally administered either 24 hours before (prophylaxis) or 6 hours after (therapy) intranasal 10 5 TCID 50 virus challenge. In the therapeutic group, the rationale for administration of the antibody six hours post viral infection is based on the replication cycle length of 5-6 hours after initial infection for SARS-CoV in VeroE6 cells (Keyaerts et al., 2005) .

Six hours after challenge with a high dose of 10 5 TCID 50 , approximately the same number of susceptible cells could become infected and likely produce much more infectious virus, which would need to be neutralized by the antibody to prevent subsequent cycles of infection. Nasal washes and oral swab at 1, 3, 5 days post infection (dpi) and different lung lobes at 5 dpi were collected. V H -Fc ab8 decreased viral RNA by 1.7 log in the lung when administered prophylactically. The lung viral RNA decrease in the therapeutic groups was slightly lower (by 1.2 log) ( Figure 6D) . Interestingly, the viral RNA load in the therapeutic groups was to some extent tissue location dependent ( Figure 6F ). The variation of the viral load in different lung lobes may relate to nonuniform antibody transport and viral spread inside the lung. Remarkably, V H -Fc ab8 alleviated hamster pneumonia and reduced the viral antigen in the lung (H&E staining, Figure 7A and C and immunohistochemistry Figure 7B and D). The control hamsters exhibited severe interstitial pneumonia characterized by extensive inflammatory cell infiltration, presence of type II pneumocytes, alveolar septal thickening and alveolar hemorrhage. Both prophylactic and therapeutic treatment of V H -Fc ab8 reduced the lesions of alveolar epithelial cells, focal hemorrhage and inflammatory cells infiltration. V H -Fc ab8 also reduced the shedding from mucosal membranes including in nasal washes and oral swabs ( Figure S4 ). The decrease in viral RNA in nasal washes and oral swabs were not as large as the decrease observed in the lung tissue, similar to a recent finding in hamsters (Imai et al., 2020) . Overall, the J o u r n a l P r e -p r o o f prophylactic treatment was more effective than the therapeutic treatment in decreasing viral load in nasal washes and oral swabs. Notably, prophylactic administration of V H -Fc ab8 effectively reduced the infectious virus in the oral swab at 1 dpi, while the post-exposure treatment did not (Figure S4C and G) . Interestingly, viral reduction (except the viral titer in the oral swab at 1 dpi) was more effective at 3 and 5 dpi compared to that at 1 dpi, likely due to the infection peak occurring before day 3 as reported in hamsters (Sia et al., 2020) . A striking finding is that V H -Fc ab8 given therapeutically at as low dose as 3 mg/kg can still decrease viral loads in the lung, nasal washes and oral swabs ( Figure S5 ).

We measured the V H -Fc ab8 concentrations at both doses (10 and 3 mg/kg) in the sera at 1 dpi and 5 dpi in the post-exposure treatment groups ( Figure S5C ). The higher dose (10 mg/kg) resulted in higher antibody concentration and better inhibitory activity than the lower dose (3 mg/kg). The relatively high concentration of V H -Fc ab8 five days after administration also indicates good pharmacokinetics. Furthermore, we also compared the V H -Fc ab8 concentration in both the sera and lung with that of IgG1 ab1, which has a similar affinity to SARS-CoV-2 and similar degree of competition with the receptor ACE2 as V H -Fc ab8 . We found that the concentration of V H -Fc ab8 in hamster sera is significantly higher than that of IgG1 ab1 at 1 and 5 dpi after postexposure administration of the same dose of 10 mg/kg ( Figure 7E ), possibly indicating more effective delivery of V H -Fc ab8 from the peritoneal cavity to the blood than that of IgG1 ab1. We also found that the V H -Fc ab8 concentration in all hamster lung lobes was higher than that of the IgG1 ab1 ( Figure 7F ), suggesting that V H -Fc ab8 appears to penetrate the lung tissue more effectively than IgG1 ab1. These results indicate that the in vivo delivery of V H -Fc ab8 may be more effective than that of full-size antibodies in an IgG1 format.

The V H -Fc ab8 propensity for aggregation was measured at 37°C by dynamic light scattering (DLS), which detects particle size distributions in the nanometer range (Stetefeld et al., 2016) . It displayed a single peak at 11.5 nm which is the size of a monomeric V H -Fc protein ( Figure S6A ). The absence of large-size peaks corresponding to large molecular weight species (aggregates) in solution, indicates that V H -Fc ab8 is highly resistant to aggregation at high concentration (4 mg/ml) and relatively long times of incubation (6 days) at 37°C. The V H -Fc ab8 propensity for aggregation was also evaluated by size exclusion chromatography (SEC), which showed that >96% of V H -Fc ab8 was eluted in a peak at a position corresponding to a monomeric state with a molecular weight of 80 kDa ( Figure   S6B ).

Antibody nonspecificity and polyreactivity can be an obstacle for developing an antibody into a clinically useful therapeutic. Polyreactivity may not only cause off-target toxicities and interfere with normal cellular functions, but may also reduce antibody half-life (Chuang et al., 2015) . To test for potential polyreactivity of V H -Fc ab8, a Membrane Proteome Array (MPA) platform was used, in which 5,300 different human membrane protein clones were separately overexpressed in 293T cells in a matrix array achieving a high-throughput detection of binding by FACS. V H -Fc ab8 did not bind to any of those proteins ( Figure S6C ), demonstrating its lack of polyreactivity and nonspecificity. Interestingly, we did not detect V H -Fc ab8 binding to the human FcγRIA, which is probably due to the relatively low expression level of FcγRIA on HEK-293T cell surface without concomitant expression of the common γ chain (Van Vugt et al., 1996) . In addition, we found that V H -Fc ab8 bound to the FcγRs much weaker than IgG1 (Figure S7 ), likely due to the different conformation in the lower hinge region for Fc fusion proteins compared to that of IgG1s (Ying et al., 2014b) . For the Fc fusion proteins (even with the same hinge sequence as IgG1), binding to FcγRs may be different from that of IgG1, and can be affected by the fusion partners (Lagassé et al., 2019) . The importance of antibody binding to FcγRs for therapeutic or prophylactic efficacy or toxicity in SARS-CoV-2 infection is unknown.

Neutralizing mAbs are promising for prophylaxis and therapy of SARS-CoV-2 infections. Recently, many potent neutralizing antibodies from COVID19 patients were identified that neutralize pseudovirus with IC 50 s ranging from 1 to 300 ng/ml, and replication-competent SARS-CoV-2 with IC 50 s from 15 to 500 ng/ml Ju et al., 2020; Rogers et al., 2020; Shi et al., 2020; Zost et al., 2020) . By comparison, the V H -Fcab8 reported here exhibited comparable or better neutralizing potency against SARS-CoV-2 pseudovirus and live virus (IC 50 s of 30 ng/ml and 40 ng/ml respectively). Of note, IC 50 s can vary widely between different assays and laboratories because there is no generally accepted standardized assay. In addition, there are many factors that contribute to potency and efficacy in vivo. Animal models are a more comprehensive and likely more reliable predictor of potential efficacy in humans than in vitro neutralization assays.

To our knowledge V H -Fc ab8 is the first human antibody domain whose activity was validated in two animal models. In the mouse ACE2 adapted SARS-CoV-2 infection model, V H -Fc ab8 significantly decreased infectious virus by 10-fold at 2 days post infection even at a very low dose of 2 mg/kg ( Figure 6A ). It also exhibited both prophylactic and therapeutic efficacy in a hamster model. It not only reduced the viral load in the lung and alleviated pneumonia; but it also reduced shedding in the upper airway (nasal washes and oral swab), which could potentially reduce transmission of SARS-CoV-2. Impressively, V H -Fc ab8 was active therapeutically even at 3 mg/kg. The finding that V H -Fc ab8 persisted for 4 days post administration at significant levels indicates that the pharmacokinetics of V H -Fc ab8 is comparable to that of a full size antibody; the half-lives of Fc fusion proteins were reported to vary from those of IgG1s and can range from hours to days (Unverdorben et al., 2016) . The molecular weight of V H -Fc ab8 (80 kDa) is half of that of full-size IgG1 which suggests an advantage in terms of smaller quantities needed to be produced compared to those for IgG1s to reach similar number of molecules and efficacy. In addition, it was shown that decreasing binder's size exponentially increases its diffusion through normal and tumor tissues (Jain, 1990) . Thus, decreasing the size two-fold can increase diffusion through tissues by four-fold. We found that after administration at the same dose, the concentration of V H -Fc ab8 was higher than that of IgG1 ab1 in both hamster sera and lung tissue. This result might suggest that the V H -Fc ab8 diffusion from the peritoneal cavity to the blood and penetration of lung may be faster than that of IgG1 ab1. This may further explain its efficacy at low doses in animals. Although the low dose showed efficacy in the small animal models, it should be noted that in humans higher doses could be required to achieve comparable degree of efficacy. Another caveat is that in the J o u r n a l P r e -p r o o f hamster post-exposure experiment, the V H -Fc ab8 was administered at a time (six hours) when the first round of virus replication was likely completed (Keyaerts et al., 2005) , but before the infection peak at 1-2 days (Sia et al., 2020) . Because it inhibits infection of new cells, its administration at around the infection peak or after may not be as effective unless it also kills infected cells in vivo which is under investigation.

Recently antibody domains including human V H and camelid V H H were reported having varying neutralization potency (Chi et al., 2020; Sun et al., 2020; Wrapp et al., 2020; Wu et al., 2020a) . Compared to those domains, V H -Fc ab8 is unique in terms of potency, aggregation resistance and specificity. V H -Fc ab8 exhibited good developability properties including stability at high concentrations and long incubation at 37°C, as well as absence or very low aggregation. In addition, V H -Fc ab8 did not bind to the human cell line 293T even at high concentration (1 µM) which is about 1754-fold higher than its K d indicating absence of non-specific binding to many membraneassociated human proteins. A similar result was obtained by the membrane protein array assay showing that V H -Fc ab8 did not bind to any of 5,300 human membrane-associated proteins, indicating its lack of non-specificity and thus low potential for off-target toxicity when used in vivo. Besides, unlike camel V H Hs, the V H ab8 sequence is fully human and therefore likely less immunogenic than that of camelid V H Hs.

Multiple structures are now available for the SARS-CoV-2 S protein trimer in complex with various neutralizing antibodies, offering insight into antigenic epitopes and inhibitory mechanisms critical for S protein neutralization. Epitopes on the SARS-CoV-2 S protein RBD have emerged as effective targets, as evidenced by the action of several RBD binding antibodies including CR3022, B38, C105, CB6, H014, and S309 (Barnes et al., 2020; Lv et al., 2020; Pinto et al., 2020; Shi et al., 2020; Wu et al., 2020b) . While B38, C105, and CB6 directly compete with ACE2 for binding sites on the RBD surface, H014 occupies a position distinct from these binding sites, precluding ACE2 binding via steric inhibition . S309 targets the RBD of the S protein both in closed and open S protein conformations, exhibiting a different mechanism of neutralization (Pinto et al., 2020) . A recent study of the structure of the S protein trimer in complex with the nanobody H11-D4 (PDB ID: 6Z43) revealed full occupancy of the nanobody on all three RBDs in a "one up and two down" conformation (Huo et al., 2020) , similar to what we report here. Our structural analysis demonstrates that the location of the V H ab8 bound to the trimeric S ectodomain directly overlaps the region that would be occupied by ACE2 when bound to the S protein. The ACE2 blocking is likely the major mechanism of the V H -Fc ab8 neutralizing activity, which is significantly augmented by avidity effects due to its bivalency. The narrow neutralization concentration range in the live virus neutralization (10-200 ng/ml for 0%-100% neutralization) ( Figure 5D ) indicates a plausible cooperative neutralization mechanism, probably due to the synergistic binding of V H molecules in V H -Fc ab8 to RBDs. Due to its small size, V H may facilitate targeting occluded epitopes on RBD that are otherwise inaccessible to full-length IgGs, which is important because the SARS-CoV-2 S protein is conformationally heterogenous, exposing neutralizing epitopes to varying degrees . The structural analysis shows that V H ab8 is able to simultaneously target all three RBD epitopes in both "up" and "down" conformations, which may provide a structural basis for a unique cooperative neutralization mechanism for V H -Fc ab8. V H -Fc ab8 with a long flexible linker between V H and Fc may allow two J o u r n a l P r e -p r o o f V H molecules to bind simultaneously two protomers in the same S trimer or cross-link two different protomers from different S trimers.

The ab8 epitope is distal to the CR3022 epitope, explaining its lack of competition with CR3022. The ab8 contact residue F486 (L472 in SARS-CoV) is not conserved which likely explains its lack of cross-reactivity to SARS-CoV. From the GISAID and NCBI databases, we found nine mutations in RBD with relatively high frequencies in current circulating SARS-COV-2. Six of them are in the core domain (F342L, N354D, N354D/D364Y, V367F, R408I and W436R) and three in the RBM (K458R, G476S, V483A). The core domain mutations are far away from the ab8 epitope, thus these mutations do not affect V H -Fc ab8 binding to RBD. Those three RBM mutations also did not affect ab8 binding although they are close to the ab8 epitope, suggesting that these mutations may not affect ab8 neutralizing activity although neutralization of whole virus carrying these mutations is needed to definitely demonstrate this possibility. Interestingly, V H -Fc ab8 effectively inhibited the mouse ACE2 adapted SARS-CoV-2 with a Q498T/P499Y mutation in RBD, indicating that this double mutation also does not affect V H -Fc ab8 binding to RBD. These results suggest that V H -Fc ab8 may be a broadly crossreactive SARS-CoV-2 neutralizing antibody.

In conclusion, we identified a fully human antibody V H domain that shows strong competition with ACE2 for binding to RBD and potent neutralization of SARS-CoV-2 in vitro and in two animal models. This potent neutralizing activity combined with its specificity and good developability properties warrants its further evaluation for prophylaxis and therapy of SARS-CoV-2 infection. Our elucidation of its unique epitope and mechanism of neutralization could also help in the discovery of more potent inhibitors and vaccines. Hamsters were bled at one and five dpi for measuring antibody concentrations in sera by SARS-CoV-2 S1 ELISA. Sera was diluted 1:100 and binding was detected by using the goat anti human IgG-HRP. (F). Viral RNA levels in different lung lobes. RNA quantity was presented as the TCID 50 equivalence. Experiments were performed in duplicate and the error bars denote ± SD, n =2.

Detailed methods are provided in the online version of this paper and include the following:

• KEY RESOURCES TABLE 

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Dimiter Dimitrov (mit666666@pitt.edu).

All requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact author. This includes antibodies, viruses, plasmids and proteins. All reagents will be made available on request after completion of a Material Transfer Agreement.

Antibody nucleotide sequence has been deposited to GenBank with an accession number of MT943599. The antibody is only allowed for non-commercial use. All data supporting the findings of this study are available within the paper and are available from the corresponding author upon request.

Vero E6 (CRL-1586, American Type Culture Collection (ATCC) and 293T (ATCC) were cultured at 37°C in Dulbecco's Modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum (FBS), 10 mM HEPES pH 7.3, 1 mM sodium pyruvate, and 100 U/mL of penicillin-streptomycin. 293T stably expressing SARS-CoV-2 and human ACE2 was cultured in DMEM medium containing 200 µg/ml Zeocin. HEK293F and expi293F were cultured in FreeStyle 293 serum free medium (ThermoFisher, Cat#12338018) and Expi293™ Expression Medium J o u r n a l P r e -p r o o f (ThermoFisher, Cat# A1435103), respectively. The SARS-CoV-2 spike pseudotyped HIV-1 backboned virus is packaged in 293T cells after transfecting pNL4-3.luc.RE and pcDNA3.1 S plasmids. The SARS-CoV-2 (US_WA-1/2020) and SARS-CoV2/Canada/ON/VIDO-01/2020 obtained from Centers for Disease Control and Prevention were propagated in Vero E6 cells. The recombinant SARS-CoV-2-SeattlenLuc virus and the mouse ACE2 adapted SAR-CoV-2 virus (carrying a Q498T/P499Y mutation in RBD) recovered by the reverse genetics was produced in VeroE6 cells. All work with infectious SARS-CoV-2 was performed in Institutional Biosafety Committee approved BSL3 facilities using appropriate positive pressure air respirators and protective equipment.

The recombinant proteins SARS-CoV-2 RBD-his, RBD mutants, RBD-Fc, ACE2-hFc were subcloned into pcDNA3.1 expression plasmids, and expressed in expi293F cells. Proteins with his tag were purified by Ni-NTA affinity chromatography and protein with Fc tag purified by protein A chromatography. Protein purity was estimated as >95% by SDS-PAGE and protein concentration was measured spectrophotometrically (NanoVue, GE Healthcare).

V H ab8 antibody was identified by panning of the phage library. V H -Fc ab8 were constructed by fusing V H to human IgG1 Fc with the native IgG1 hinge. IgG1 ab1 was obtained by our lab through panning of a Fab phage library.

MERS-CoV-specific IgG1 m336 and SARS-CoV antibody IgG1 CR3022 sequences from other groups were subcloned into the pDR12 plasmid for expression. V H ab8 (in a phagemid pComb3x with a Flag tag) was expressed in HB2151 E. coli and purified by Ni-NTA affinity chromatography. All other IgG1 were expressed in expi293 cells and purified with protein A chromatography.

For the mouse model, BALB/c mice purchased from Envigo (BALB/cAnNHsd, stock# 047, immunocompetent, 11-12 months of age, female) were used for all experiments. They are drug/test naïve and negative for pathogens. BioFresh bedding with Crinkle bedding added. Hamsters have access to food and water ab libitum. Food is Lab Diet 5P00 ProLab RMH300. Cages are changed weekly or as needed and spot cleaned. For experiment, hamsters were intraperitoneally treated with V H -Fc ab8 either 24 hrs before (prophylaxis) or 6 hrs (therapy) after intranasal challenge of 1×10 5 TCID 50 of SARS-CoV-2. Nasal washes and oral swabs were collected at day 1, 3 and 5 post infection (dpi). Hamsters were bled at 1 and 5 dpi. All hamsters were euthanized on 5 dpi. At euthanasia, lungs were collected for RNA isolation. For viral titer determination, VeroE6 cells TCID 50 assay was used. For testing viral RNA, viral RNA RT-qPCR was used. For testing antibody concentration at sera and lung, SARS-CoV-2 S1 ELISA was used. For histopathology, 10% formalin fixed and paraffin embedded tissues were processed with either hematoxylin and eosin stain (H&E) or immunohistochemistry (IHC). Lung lobes were scored based on pathology using microscopy.

CR3022. The SARS-CoV-2 S and the anti-SARS-CoV antibody IgG1 CR3022 and genes were synthesized by IDT (Coralville, Iowa). MERS-CoV-specific IgG1 m336 antibody was expressed in human mammalian cell as described previously (Ying et al., 2014a) . Briefly, IgG1 m336 light chain and heavy chain Fd were subcloned into the pDR12 vector containing dual promoters and a IgG1 Fc cassette. The recombinant plasmid was sequenced and transfected into expi293 cells for expression. The human angiotensin converting enzyme 2 (ACE2) gene was ordered from OriGene (Rockville, MD). The RBD domain (residues 330-532) and S1 domain (residues 14-675) and ACE2

(residues 18-740) genes were cloned in frame to human IgG1 Fc in the mammalian cell expression plasmid pcDNA3.1. The RBD protein with an AviTag followed by a 6×His tag at C-terminal was subcloned similarly. These proteins were expressed with Expi293 expression system (Thermo Fisher Scientific) and purified with protein A resin (GenScript) and by nickel-nitrilotriacetic acid (Ni-NTA) resin (Thermo Fisher Scientific). The Fab CR3022 antibody gene with a His tag was cloned into pCAT2 plasmid (developed in house) for expression in HB2151 bacteria and purified with Ni-NTA resin. Protein purity was estimated as >95% by sodium dodecyl sulfatepolyacrylamide gel electrophoresis (SDS-PAGE) and protein concentration was measured spectrophotometrically (NanoVue, GE Healthcare).

Unlike camel V H Hs, which naturally evolved to be autonomously stable, human V H is usually unstable and easy to aggregate in the absence of V L (Li et al., 2016; Nguyen et al., 2000) . However, human V H can be selected or engineered with high stability and solubility. To facilitate identification of stable V H binders, we chose engineered germline V H 3-23 as our library scaffold (Chen et al., 2008b) . Our human V H phage display library was made by grafting heavy chain CDR1, 2, 3 genes derived from 12 healthy donors' peripheral blood monocytes (PBMCs) and J o u r n a l P r e -p r o o f splenocytes (Takara, Cat. No. 636525) into their cognate positions of a stable scaffold (based on the germline V H 3-23) in a manner similar to the method we previously described but without mutagenesis of CDR1 (Chen et al., 2008a) . Briefly, CDRs were PCR-amplified by using primers with degenerated adaptors covering CDRs edge regions from diverse V H families in one end, and with sequences annealing to the V H 3-23 framework (FR) regions in the other end. The PCR products were then assembled by overlapping extension PCR by using primers with homologous ending. The whole V H was assembled by overlapping FR1-CDR1-FR2-CDR2 and FR3-CDR3-FR4 fragments. After assembly, the V H fragment was Sfi I digested followed by ligated into Sfi I linearized pComb3x

phagemid. The recombinant phagemid was then purified, desalted and concentrated for electroporation of bacteria TG1, from which the V H phage particles were rescued and produced. The library size was determined by tittering transformants. The library quality (diversity) was checked by randomly Sanger sequencing hundreds of V H clones and also evaluated by panning of diverse antigens. This library contains very large number of clones (10 11 ). For panning, the V H library was alternatively panned against biotinylated RBD-his and RBD-Fc proteins. RBD biotinylation occurred through biotin ligase (BirA) mediated enzymatic conjugation of a single biotin on AviTag (GLNDIFEAQKIEWHE) (Fairhead and Howarth, 2015) . The panning was for 3 rounds with input antigens of 10 µg RBD-his, 2 µg RBD-Fc and 0.5 µg RBD-his for the 1 st , 2 nd and 3 rd round, respectively. The panning process begun with incubation of antigens with 10 12 V H phage particles followed by washing with phosphate-buffered saline (PBS) containing 0.1% Tween-20. Bound phage pulled down by streptavidin-M280-Dynabeads were rescued by log-phase TG1 cells with the M13KO7 helper phage. After the 3 rd round panning, positive clones were selected by soluble expression monoclonal (SEM) ELISA followed by sequencing (Chen et al., 2008b) . V H binders were further screened for their binding affinity, stability and ACE2 competition. For conversion to Fc-fusion, the V H gene was subcloned into pSecTag B vector containing human IgG1 Fc fragment. V H -Fc ab8 was expressed as described above.

Enzyme-Linked Immunosorbent Assays (ELISAs). For detection of RBD biotinylation efficacy, horseradish peroxidase (HRP) conjugated streptavidin was used. For conformation of function of RBD-his after biotinylation, 100 ng ACE2-Fc was coated into the plates followed by addition of serially diluted biotinylated RBD-his. HRP conjugated streptavidin was used for detection. For other ELISAs, the SARS-CoV-2 RBD (residues 330-532) protein was coated on 96-well plates (Costar) at 100 ng/well in PBS overnight at 4 o C. For screening SEM ELISA, clones randomly picked from the infected TG1 cells were incubated with immobilized antigen. Bound phages were detected with HRP-conjugated mouse anti-FLAG tag Ab (Sigma-Aldrich). For the V H -Fc binding assay, HRPconjugated goat anti-human IgG Fc (Sigma-Aldrich) was used for detection. For the competition ELISA with hACE2, 2 nM of human ACE2-mouse Fc was incubated with serially diluted V H , or V H -Fc, and the mixtures were added to RBD coated wells. After washing, bound ACE2-mouse Fc was detected by HRP-conjugated anti mouse IgG (Fc specific) (Sigma-Aldrich). For evaluation of ACE2 blocking of V H ab8 binding to RBD, 10 nM V H ab8 was incubated with coated RBD in the presence of various concentration of ACE2-His (Sino Biological), and the bound V H ab8 was detected by HRP conjugated anti FLAG antibody. For evaluation of conformational changes of the epitope mapping RBD mutants, we used a mouse polyclonal anti SARS-CoV-2 RBD antibody (Sino biological, Cat.

No. 40592-MP01) and the human IgG1 CR3022 antibody. For measuring the binding of V H -Fc ab8 to RBD mutants, 100 ng RBD mutant was coated on 96-wells plates and incubated with V H -Fc ab8 with binding detected by using J o u r n a l P r e -p r o o f HRP conjugated anti human Fc antibody. To evaluate the binding of V H -Fc ab8 and IgG1 ab1 to human FcγRs, recombinant human FcγRIA, IIA, IIIA were coated on 96-wells plates followed by addition of biotinylated V H -Fc ab8 and IgG1 ab1. Binding was detected by the streptavidin-HRP. All colors were developed by 3,3′,5,5′tetramethylbenzidine (TMB, Sigma) and stopped by 1 M H 2 SO 4 followed by recording absorbance at 450 nm.

Experiments were performed in duplicate and the error bars denote ± 1 SD.

BLItz. Antibody affinities and avidities were analyzed by the biolayer interferometry BLItz (ForteBio, Menlo Park, CA). For measuring V H ab8 affinity, the RBD-Fc was mounted on the protein A sensor (ForteBio: 18-5010). 125 nM, 250 nM and 500 nM V H ab8 were used for association. For measuring avidity of V H -Fc ab8, biotinylated RBD-Fc was immobilized on streptavidin biosensors (ForteBio: 18-5019) for 2 min and equilibrated with Dulbecco's phosphate-buffered saline (DPBS) (pH = 7.4) to establish baselines. 50 nM, 100 nM and 200 nM V H -Fc ab8 were chosen for association. The association was monitored for 2 min and then the antibody was allowed to dissociate in DPBS for 4 min. The k a and k d were derived from sensorgrams fittings and used for K d calculation. For the competitive Blitz, 500 nM V H -Fc ab8 was loaded onto the RBD-Fc coated sensor for 300 s to reach saturation followed by dipping the sensor into a 100 nM ACE2-Fc or Fab CR3022 solution in the presence of 500 nM V H -Fc ab8. The association was monitored for 300 s. The signals from100 nM hACE2 or CR3022 binding to the RBD-Fc coated sensor in the absence of V H -Fc ab8 was independently recorded in parallel. Competition was determined by the percentage of signal in the presence of V H -Fc ab8 to signal in the absence of V H -Fc ab8 (< 0.7 is considered to be competitive) (Wu et al., 2020a) . (Agilent, cat. no. 200521) . Mutants were expressed and purified according to the abovementioned RBD purification procedures. ELISA was used to evaluate the binding of these mutants compared to the wild type RBD.

A. Expression and Purification. The codon optimized SARS-CoV-2 2P S protein ectodomain construct (GenBank: YP_009724390.1) was C-terminally tagged with 8xHis and a twin Strep tag and cloned into the mammalian expression vector pcDNA 3.1 (Synbio). HEK293F cells were grown in suspension culture using FreeStyle media (ThermoFisher) at 37 °C in a humidified CO 2 incubator (8% CO 2 ). Cells were transiently transfected at a density of 1 x 10 6 cells/ml using branched polyethylenimine (PEI) (Sigma) (Portolano et al., 2014) . Media was exchanged after 24 h and supplemented with 2.2 mM valproic acid. Supernatant was harvested by centrifugation after 4 days, filtered and loaded onto a 5 ml HisTrap HP column (Cytiva). The column was washed with buffer (20 mM Tris pH 8.0, 500 mM NaCl, 20 mM imidazole) and the protein was eluted with buffer (20 mM Tris pH 8.0, 500 mM NaCl, 500 mM imidazole). Purified protein was concentrated (Amicon Ultra 100 kDa cut off, Millipore Sigma) and loaded onto a J o u r n a l P r e -p r o o f Superose 6 column (Cytiva) equilibrated with GF buffer (20 mM Tris pH 8.0 and 150 mM NaCl). Peak fractions were pooled and concentrated to 1.3 mg/ml (Amicon Ultra 100 kDa cut off, Millipore Sigma).

Purified S protein ectodomain (0.04 mg/ml) was mixed with V H ab8 (0.02 mg/ml) or soluble ACE2 (0.02 mg/mL) and incubated on ice for 10 mins. For the competition experiment, the S protein (0.04 mg/ml) was first incubated on ice with V H ab8 (0.02 mg/ml) for 10 mins then followed by addition of ACE2 (0.02 mg/mL) for another 10 mins. The mixtures (4.8 µl) were applied to 300mesh copper grids coated with continuous ultrathin carbon. Grids were plasma cleaned using an H 2 /O 2 gas mixture for 15 s in a Solarus plasma cleaner (Gatan Inc.) prior to adding the sample. Samples were allowed to adsorb for 30 s before blotting away excess liquid, followed by a brief wash with MilliQ H 2 O. Grids were stained by three successive applications of 2% (w/v) uranyl formate (20 s, 20 s, 60 s). Grids containing S protein ectodomain with V H ab8, and S protein ectodomain mixed with both V H ab8 and soluble ACE2 were imaged using a 200 kV Glacios transmission electron microscope (ThermoFisher Scientific) equipped with a Falcon3 camera operated in linear mode. Using EPU automated acquisition software (ThermoFisher Scientific), 15-frame movies were collected at 92,000x magnification (corresponding to a physical pixel size of 1.6 -) over a defocus range of -0.5 to -3.0 µm with an accumulated total dose of 40 e -/Å 2 /movie. Grids containing purified S protein ectodomain (0.04 mg/mL) with soluble ACE2 (0.02 mg/mL) were imaged using a 200kV Glacios transmission electron microscope equipped with a Ceta 16M CMOS camera (ThermoFisher Scientific). Micrographs were collected at 92,000x magnification (physical pixel 1.6 -) over a defocus range of -0.5 to -3.0 µm with a total dose of 50 e -/Å 2 using EPU automated acquisition software.

C. Image Processing. Motion correction and CTF estimation were performed in RELION (3.1) (Scheres, 2012) .

Particles were picked by crYOLO (1.7.4) (Wagner et al., 2019) with pre-trained model for negative stain data. After extraction, particles were imported to cryoSPARC live (v2.15.1) (Punjani et al., 2017) and subjected to 2D classification and 3D heterogeneous classification. Final density maps were obtained by 3D homogeneous refinement. Figures were prepared using UCSF Chimera (Pettersen et al., 2004) . After washing, V H ab8 binding was detected by PE conjugated anti FLAG tag antibody.

To test antibody mediated inhibition of cell fusion, the β-galactosidase (β-gal)

reporter gene based quantitative cell fusion assay was used (Xiao et al., 2003) . In this assay, 293T-S cell expression of T7 RNA polymerase was achieved by infection with vaccinia virus VTF7.3, while 293T-ACE2 cell expression of T7 promoter controlled β-Gal was obtained by infection with vaccinia virus VCB21R. β-Gal will be expressed only after fusion of the two types of cells, which can be monitored by chromogenic reactions using β-Gal substrate. To assay cell-cell fusion, 293T cells stably expressing SARS-CoV-2 S (293T-S) cells were infected with T7 polymerase-expressing vaccinia virus (vTF7-3), and 293T cells stably expressing ACE2 (293T-ACE2) were infected with vaccinia virus (vCB21R Lac-Z) encoding T7 promotor controlled β-gal. Two hours after infection, cells were incubated with fresh medium and transferred to 37 °C for overnight incubation. The next day, 293T-S cells were pre-mixed with serially diluted antibodies or ACE2-Fc at 37 °C for 1 h followed by incubation with 293T-ACE2 cells at a 1:1 ratio for 3 h at 37°C. Then cells were then lysed, and the β-gal activity was measured using βgalactosidase assay kit (substrate CPRG, G-Biosciences, St. Louis, MO) following the manufacturer's protocol.

Fusion inhibition percentage (sample reading, F) was normalized by maximal fusion (reading, F max ) of 293T-S and 293T-ACE2 cells in the absence of antibodies using this formula: Fusion inhibition % = [(F max -F)/(F max -F blank )] × 100%, in which F blank refers to the OD reading of 293T-S and 293T incubation wells. Fusion inhibition percentage was plotted against antibody concentrations. Experiments were performed in duplicate and the error bars denote ± 1 SD.

Pseudovirus Neutralization Assay. Pseudovirus neutralization assay was performed based on previous protocols . Briefly, HIV-1 backbone based pseudovirus was produced in 293T cells by co-transfection with plasmid encoding SARS-CoV-2 S protein and plasmid encoding luciferase expressing HIV-1 genome (pNL4-3.luc.RE) using PEI. Pseudovirus-containing supernatants were collected 48 h later and concentrated using Lenti-X™ concentrator kit (Takara, CA). Pseudovirus neutralization assay was then performed by incubation of SARS-CoV-2 pseudovirus with serially diluted antibodies or ACE2-Fc for 1 h at 37 °C, followed by addition of the mixture into pre-seeded 293T-ACE2 cells. The mixture was then centrifuged at 1000 × g for 1 hour at room temperature.

The medium was replaced 4 hrs later. After 24 h, luciferase expression was determined by Bright-Glo kits (Promega, Madison, WI) using BioTek synergy multi-mode reader (Winooski, VT). Cells only and virus only wells were included and used for normalization. The 50% pseudovirus neutralizing antibody titer (IC 50 ) was calculated using Graphpad Prism 7. Experiments were performed in duplicate and the error bars denote ± 1 SD.

(MN) assay was used as previously described (Agrawal et al., 2016a; Agrawal et al., 2016b; Du et al., 2013; Du et al., 2014) . Briefly, serially three-fold and duplicate dilutions of individual monoclonal antibodies (mAbs) were incubated with 120 pfu of SARS-CoV or SARS-CoV-2 at room temperature for 2 h before transferring into designated wells of confluent Vero E6 cells grown in 96-well microtiter plates. Vero E6 cells cultured with medium with or without virus were included as positive and negative controls, respectively. MERS-CoV RBD-specific J o u r n a l P r e -p r o o f neutralizing m336 mAb (Ying et al., 2014a) were used as additional controls. After incubation at 37 o C for 4 days, individual wells were observed under the microcopy for the status of virus-induced formation of cytopathic effect.

The efficacy of individual mAbs was expressed as the lowest concentration capable of completely preventing virusinduced cytopathic effect in 100% of the wells.

Full-length viruses expressing luciferase were designed and recovered via reverse genetics as described previously (Scobey et al., 2013; Yount et al., 2003) .

Briefly, the SARS-CoV-2 RNA from infected cell culture was reverse-transcribled and constructed into the seven contiguous genomic cDNA subclones with interconnecting junctions, which were then BsaI/BsmBI digested and ligated into a full-length SARS-CoV-2 genome cDNA through the cohesive ends. A silent mutation of T15102A was introduced into a conserved region in nsp12 to differentiate our recombinant viruses from the circulating SARS-CoV-2 strains through Sanger sequencing. The reporter viruse was synthesized by replacing a 276-bp region in ORF7 with a GFP-fused nanoluciferase (nLuc) gene. After assembly into full-length cDNA, full-length RNA was in vitro transcribed and was electroporated into Vero E6 cells. Virus stocks were propagated on Vero E6 cells in minimal essential medium containing 10% fetal bovine serum (HyClone) and supplemented with penicillin/kanamycin (Gibico). Viruses were tittered in Vero E6 USAMRID cells to obtain a relative light units (RLU) signal of at least 20× the cell only control background. Ab or ACE2-Fc were serially diluted 4-fold up to eight dilution spots with at a starting dilution 100 µg/ml, and were incubated with SARS-CoV-UrbaninLuc and SARS-CoV-2-SeattlenLuc viruses at 37°C with 5% CO 2 for 1 hour. Then virus-antibody dilution complexes were added to the pre-seeding E6 USAMRID cells (20,000) in duplicate. Virus-only controls and cell-only controls were included in each neutralization assay plate. Following infection, plates were incubated at 37 °C with 5% CO 2 for 48 hours. Then cells were lysed and luciferase activity was measured via Nano-Glo Luciferase Assay System (Promega) according to the manufacturer specifications. SARS-CoV and SARS-CoV-2 neutralization IC 50 were defined as the sample concentration at which a 50% reduction in RLU was observed relative to the average of the virus control wells. Experiments were performed in duplicate and IC 50 was obtained by the non-linear fitting of neutralization curves in Graphpad Prism 7.

mouse ACE2 adapt SARS-CoV-2 variant was constructed by introduction of two amino acid changes (Q498T/P499Y) at the ACE2 binding pocket in RBD. Virus stocks were grown on Vero E6 cells and viral titer was determined by plaque assay . Groups of 5 each of 10 to 12-month old female BALB/c mice (Envigo, #047) were treated prophylactically (12 hours before infection) by intraperitoneal injection with 36, 8, or 2 mg/kg of V H -Fc ab8, respectively. Mice were challenged intranasally with 10 5 PFU of mouse-adapted SARS-CoV-2.

Two days post infection, mice were sacrificed and lung viral titer was determined by the plaque assay. To exclude the residual lung antibody impact on viral titration, mice were euthanized and perfused with 10 ml of PBS via cardiac puncture before lung harvest for viral titration. For virus titration, the caudal lobe of the right lung was homogenized in PBS. The resulting homogenate was serial-diluted and inoculated onto confluent monolayers of Vero E6 cells, followed by agarose overlay. Plaques were visualized via staining with Neutral Red on day 2 post J o u r n a l P r e -p r o o f infection. To measure the viral RNA in the lung, tissue homogenate lysed in Trizol LS (Thermofischer) was then processed with Thermofischer Trizol RNA isolation protocol followed by RT-qPCR using the QuantiFast Probe RT-PCR kit (Qiagen) to amplify a portion of upE gene. The 50% tissue culture infectious doses (TCID 50 ) equivalence were estimated by running serial dilutions of known TCID 50 standards.

Infection. SARS-CoV2/Canada/ON/VIDO-01/2020 was propagated on Vero'76 cells using DMEM with 2% FBS and 1µg/ml L-(tosylamido-2-phenyl) ethyl chloromethyl ketone (TCPK) trypsin. Infectious work with SARS-CoV-2 was approved by the Biosafety Protocol Approval Committee (BPAC) at the University of Saskatchewan and performed in the high containment laboratories at VIDO-InterVac. Male hamsters (9-week-old) were obtained from Charles River (Montreal, QC). For evaluations of prophylactic efficacy, all hamsters (n=7) were injected intraperitoneally with 10 mg/kg of V H -Fc ab8 24 hours prior to intranasal challenge of 50 µl/nare containing a total of 1×10 5 TCID 50 of SARS-CoV-2. For the therapeutic group, hamsters were infected as above and treated intraperitoneally with 10 mg/kg (n=3) or 3 mg/kg (n=4) of V H -Fc ab8 6 hours post-infection. Untreated hamsters were kept as a control. Nasal washes and oral swabs were collected at day 1, 3 and 5 post infection (dpi). Hamsters were bled at 1 and 5 dpi. All hamsters were euthanized on 5 dpi. At euthanasia, lung lobes were collected for virus titration and RNA isolation. For viral titer determination, nasal washes were diluted in a 10-fold dilution series and absorbed on Vero'76 cells in triplicates for 1 hour at 37°C. Inoculum was removed and replaced with fresh DMEM containing 2% FBS, penn/strep and 1µg/ml TPCK. Cytopathic effect was scored on day 3 and day 5 post infection.

The limit of detection is 13.6 TCID 50 . For testing viral RNA, viral RNA isolated from nasal and oral swabs using the QiaAmp Viral RNA mini kit (Qiagen) and the QuantiFast Probe RT-PCR kit (Qiagen) to amplify a portion of upE gene. For RNA levels in tissues, 30 mg of tissue homogenate in buffer RLT were processed with the RNeasy kit (Qiagen) followed by RT-qPCR as above. TCID 50 equivalence were estimated by running serial dilutions of known TCID 50 standards. For testing Ab8 concentrations post injection at hamster sera and lung tissue, SARS-CoV-2 spike-1 ELISA was used. S1 protein was coated at 1 µg/ml overnight at 4°C in PBS onto MaxiSorp plates (Nunc). The following day plates were blocked with 5% skim milk and 0.05%Tween20. Serum collected on day 1 and day 5 post-challenge was diluted 1:100 and absorbed for 1 hour at 37 °C. Plates were washed and goat anti human IgG-HRP was added. Plates were washed and subsequently developed with OPD (o-phenylenediamine dihydrochloride) substrate. Optical density was measured at 450 nm after 30 mins of incubation. For lung tissues, after blocking homogenates were diluted 1:10 and absorbed overnight at 4°C followed by detection with anti-human IgG-HRP and substrate as stated above. The control hamster lung homogenate was used for background correction. For histopathology on day 5 p.i., 10% formalin fixed and paraffin embedded tissues were processed with either hematoxylin and eosin stain (H&E) or immunohistochemistry (IHC) for detection of SARS-CoV2 antigen; in IHC after blocking tissue slides were treated with anti-nucleocapsid rabbit polyclonal antibodies followed with anti-rabbit HRP antibody. (Tucker et al., 2018) . The entire library of plasmids is arrayed in duplicate in a matrix format and transfected into HEK-293T cells, followed by incubation for 36 h to allow protein expression.

Before specificity testing, optimal antibody concentrations for screening were determined by using cells expressing positive (membrane-tethered Protein A) and negative (mock-transfected) binding controls, followed by flow cytometric detection with an Alexa Fluor-conjugated secondary antibody (Jackson ImmunoResearch Laboratories).

Based on the assay setup results, V H -Fc ab8 (20 µg/ml) was added to the MPA. Binding across the protein library was measured on an iQue3 (Ann Arbor, MI) using the same fluorescently labeled secondary antibody. To ensure data validity, each array plate contained positive (Fc-binding; SARS-CoV-2 S protein) and negative (empty vector) controls. Identified targets were confirmed in a second flow cytometric experiment by using serial dilutions of the test antibody. The identity of each target was also confirmed by sequencing.

For the mouse model, the statistical significance of difference between V H -Fc ab8 treated and control mice lung virus titers was determined by the two-tailed, unpaired, student t test calculated using GraphPad Prism 7.0. A p value < 0.05 was considered significant. ** p < 0.01. For the mice lung viral titer after perfusion, viral RNA and hamster lung viral RNA, statistical significance was determined by the Mann-Whitney U test. A p value < 0.05 was considered significant. ns: p > 0.05, *p < 0.05, **p < 0.01, ***p < 0.001. For comparing V H -Fc ab8 and IgG1 ab1 concentration, significance analysis was determined by the two-way ANOVA followed by Tukey test in GraphPad Prism 7.0. A p value < 0.05 was considered significant. ns: p > 0.05, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Immunization with inactivated Middle East Respiratory Syndrome coronavirus vaccine leads to lung immunopathology on challenge with live virus

Passive Transfer of A Germline-like Neutralizing Human Monoclonal Antibody Protects Transgenic Mice Against Lethal Middle East Respiratory Syndrome Coronavirus Infection

SARS-CoV-2 Vaccines: Status Report

Structures of Human Antibodies Bound to SARS-CoV

Spike Reveal Common Epitopes and Recurrent Features of Antibodies

Potent Neutralizing Antibodies against SARS-CoV-2 Identified by High-Throughput Single-Cell Sequencing of Convalescent Patients' B Cells

The convalescent sera option for containing COVID-19

Neutralizing Antibody and Soluble ACE2 Inhibition of a Replication-Competent VSV-SARS-CoV-2 and a Clinical Isolate of SARS-CoV-2

Simulation of the clinical and pathological manifestations of Coronavirus Disease 2019 (COVID-19) in golden Syrian hamster model: implications for disease pathogenesis and transmissibility

Human domain antibodies to conserved sterically restricted regions on gp120 as exceptionally potent cross-reactive HIV-1 neutralizers

Construction of a large phage-displayed human antibody domain library with a scaffold based on a newly identified highly soluble, stable heavy chain variable domain

Humanized Single Domain Antibodies Neutralize SARS-CoV-2 by Targeting Spike Receptor Binding Domain

Eliminating antibody polyreactivity through addition of N-linked glycosylation

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

Generation and Characterization of ALX-0171, a Potent Novel Therapeutic Nanobody for the Treatment of Respiratory Syncytial Virus Infection

A mouse-adapted SARS-CoV-2 model for the evaluation of COVID-19 medical countermeasures

A Truncated Receptor-Binding Domain of MERS-CoV Spike Protein Potently Inhibits MERS-CoV Infection and Induces Strong Neutralizing Antibody Responses: Implication for Developing Therapeutics and Vaccines

A conformation-dependent neutralizing monoclonal antibody specifically targeting receptor-binding domain in Middle East respiratory syndrome coronavirus spike protein

Site-specific biotinylation of purified proteins using BirA

In vivo imaging with antibodies and engineered fragments

A Sequence Homology and Bioinformatic Approach Can Predict Candidate Targets for Immune Responses to SARS-CoV-2

Properties, production, and applications of camelid single-domain antibody fragments

Identification of a critical neutralization determinant of severe acute respiratory syndrome (SARS)-associated coronavirus: importance for designing SARS vaccines

SARS-CoV-2 Reverse Genetics Reveals a Variable Infection Gradient in the Respiratory Tract

Neutralizing nanobodies bind SARS-CoV-2 spike RBD and block interaction with ACE2

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

Physiological Barriers to Delivery of Monoclonal Antibodies and Other Macromolecules in Tumors

An emerging coronavirus causing pneumonia outbreak in Wuhan, China: calling for developing therapeutic and prophylactic strategies

Occupancy and mechanism in antibody-mediated neutralization of animal viruses

Fc-Fusion Drugs Have FcγR/C1q Binding and Signaling Properties That May Affect Their Immunogenicity

Structure of the SARS-CoV-2 spike receptor-binding domain bound to the ACE2 receptor

Neutralization of SARS-CoV-2 spike pseudotyped virus by recombinant ACE2-Ig

Potent neutralization of SARS-CoV-2 in vitro and in an animal model by a human monoclonal antibody. bioRxiv : the preprint server for biology

Antibody Aggregation: Insights from Sequence and Structure. Antibodies (Basel) 5

Bat origin of a new human coronavirus: there and back again

Neutralizing Antibodies Isolated by a site-directed Screening have Potent Protection on SARS-CoV-2 Infection

Structural basis for neutralization of SARS-CoV-2 and SARS-CoV by a potent therapeutic antibody. science

Antibody fragments: hope and hype

Camel heavy-chain antibodies: diverse germline V(H)H and specific mechanisms enlarge the antigen-binding repertoire

Engineered Autonomous Human Variable Domains

Antiviral Monoclonal Antibodies: Can They Be More Than Simple Neutralizing Agents?

UCSF Chimera--a visualization system for exploratory research and analysis

Cross-neutralization of SARS-CoV-2 by a human monoclonal SARS-CoV antibody

Recombinant protein expression for structural biology in HEK 293F suspension cells: a novel and accessible approach

Mutations in Spike Protein of SARS-CoV-2 Modulate Receptor Binding

cryoSPARC: algorithms for rapid unsupervised cryo-EM structure determination

The SARS-CoV-2 receptor-binding domain elicits a potent neutralizing response without antibodydependent enhancement

Isolation of potent SARS-CoV-2 neutralizing antibodies and protection from disease in a small animal model

Convalescent plasma in Covid-19: Possible mechanisms of action

RELION: implementation of a Bayesian approach to cryo-EM structure determination

Reverse genetics with a full-length infectious cDNA of the Middle East respiratory syndrome coronavirus

A human neutralizing antibody targets the receptor-binding site of SARS-CoV-2

Pathogenesis and transmission of SARS-CoV-2 in golden hamsters

Dynamic light scattering: a practical guide and applications in biomedical sciences

Potent neutralization of SARS-CoV-2 by human antibody heavy-chain variable domains

Potent binding of 2019 novel coronavirus spike protein by a SARS coronavirus-specific human monoclonal antibody

Isolation of state-dependent monoclonal antibodies against the 12-transmembrane domain glucose transporter 4 using virus-like particles

Pharmacokinetic properties of IgG and various Fc fusion proteins in mice

FcR gamma-chain is essential for both surface expression and function of human Fc gamma RI (CD64) in vivo

SPHIRE-crYOLO is a fast and accurate fully automated particle picker for cryo-EM

Structural Basis for Potent Neutralization of Betacoronaviruses by Single-Domain Camelid Antibodies

Identification of Human Single-Domain Antibodies against SARS-CoV-2

A noncompeting pair of human neutralizing antibodies block COVID-19 virus binding to its receptor ACE2

The SARS-CoV S glycoprotein: expression and functional characterization

Structural basis for the recognition of SARS-CoV-2 by full-length human ACE2

Exceptionally potent neutralization of Middle East respiratory syndrome coronavirus by human monoclonal antibodies

Monomeric IgG1 Fc molecules displaying unique Fc receptor interactions that are exploitable to treat inflammation-mediated diseases

Reverse genetics with a full-length infectious cDNA of severe acute respiratory syndrome coronavirus

A highly conserved cryptic epitope in the receptor-binding domains of SARS-CoV-2 and SARS-CoV

A safe and convenient pseudovirus-based inhibition assay to detect neutralizing antibodies and screen for viral entry inhibitors against the novel human coronavirus MERS-CoV

Potent cross-reactive neutralization of SARS coronavirus isolates by human monoclonal antibodies

Potently neutralizing and protective human antibodies against SARS-CoV-2