WO2023077429A1

WO2023077429A1 - TRIMERIC ANTIBODIES AGAINST SARS-CoV-2 SPIKE PROTEIN

Info

Publication number: WO2023077429A1
Application number: PCT/CN2021/129009
Authority: WO
Inventors: Peixiang MA; Guang Yang; Richard A. Lerner; Min QIANG
Original assignee: Shanghaitech University
Priority date: 2021-11-05
Filing date: 2021-11-05
Publication date: 2023-05-11
Also published as: CN116406375A

Abstract

The present disclosure provides trimetric antibodies and fragments having binding specificity to the SARS-CoV-2 spike protein's receptor binding domain (RBD). The trimeric antibodies and fragments have strong affinity and potent neutralization ability against the SARS-CoV-2 virus and various mutant forms. In particular, these trimeric antibodies have enhanced neutralization capabilities than their counterpart dimers and can more broadly neutralize new viral variants. The trimeric antibodies and fragments thus may be used for preventing or treating SARS-CoV-2 viral infection.

Description

TRIMERIC ANTIBODIES AGAINST SARS-CoV-2 SPIKE PROTEIN

BACKGROUND

The continuous spreading of SARS-CoV-2, a novel coronavirus and cause of the coronavirus disease 2019 (COVID-19) , has led to over 205 million confirmed cases and more than 4.3 million deaths around the world as of 12 ^th August, 2021. Even though several approved vaccines and antibody drugs have curbed the infection speed at the end of 2020, the rapid spread and emergence of several dominant new SARS-CoV-2 lineages increased concerns for the effectiveness of current neutralizing antibodies (nAbs) and vaccines. Uncontrolled transmission promoted the virus evolution, which generated several prevalent strains, including these variants of concerns (VOC) B. 1.1.7 (alpha) , B. 1.351 (beta) , B. 1.617.2 (delta) , and P. 1 (gamma) .

Tracking the virus evolution, D614G is the first identified dominant mutation to increase SARS-CoV-2 infectivity without enhancing disease severity and immune-escape by antibodies, which has a vital role in the viral evolution and emergence of further variants. N501Y independently emerged in different variants, including B. 1.1.7 variant (501Y. V1) first identified in September 2020 in UK, B. 1.351 variant (501Y. V2) first reported in December, 2020 in South Africa, and P1 variant (501Y. V3) identified in Brazil. N501Y mutation confers a ～10 times fold increase of affinity between RBD and hACE2, however, unlike D614G mutation, neutralization effect of immunized sera and nAbs were affected by the N501Y variants. In addition, B. 1.351 variant caused more severe disease and in-hospital mortality. The variant B. 1.617.1 gained attention in India in February, 2021 has three key mutations in spike: L452R, E484Q, and P681R; another similar VOC, B. 1.617.2, with mutations L452R, T478K, and P681R shows apparently increased transmissibility, is highly prevalent in India, and now spreads globally. In addition, B. 1.351 and P. 1 sera showed significantly reduced neutralization against B. 1.617.2, indicating that individuals infected previously by these variants may be more susceptible to reinfection by B. 1.617.2.

Among the three N501Y versions of VOC, E484K is a key mutation mediating immune escape against nAbs or immunized nAbs. By evaluating the effect of mRNA vaccines on the 10 globally circulating strains of SARS-CoV-2, Wilfredo et al. reported that 5/10 pseudoviruses harboring K417N/T, E484K, and N501Y were highly resistant to neutralization of immunized nAbs. The multiple mutants (E484K+K417N+N501Y) completely abolishe the binding to Bamlanivimab (LY-CoV555) , which has been approved with an Emergency Use Authorization (EUA) . Based on the binding epitopes, et SARS-CoV-2 nAbs are divided into three classes: class 1 are ACE2 competing nAbs that bind in receptor-binding motif (RBM) of spike; class 2 cross-reacts with SARS-CoV and binds the base of RBD; class 3 are N-terminal domain (NTD) recognizing nAbs. Many highly potent neutralizing mAbs in class 1 and class 3 showed reduced or lost inhibitory activity against viruses containing an E484K spike mutation. For B. 1.617.2 variant, most approved nAbs maintained the neutralization activities with small, up to 5-fold reduction, while LY-CoV555 was severely reduced.

Monoclonal antibodies (mAbs) targeting the viral surface proteins have shown excellent neutralization efficacy in previous treatment of SARS, MERS and Ebola, and therefore are of particular interest to combat the current pandemic. Since the outbreak of COVID-19, the spike glycoprotein has been the main antigen targeted for development of therapeutic mAbs. Most neutralizing antibodies bind to the receptor binding domain (RBD) of the viral spike protein. Some non-RBD binding antibodies to the N-terminal domain (NTD) also appear to potently neutralize SARS-CoV-2.

Neutralizing antibodies have been derived from multiple sources, including memory B cells from SARS-CoV-2 convalescent patients, previous SARS neutralizing antibodies, immunized humanized H2L2 mice, nanobodies from alpaca, and single domain human antibodies from a pre-established library. New and better antibodies are needed for prevention and treatment of the infections by SARS-CoV-2. Especially, there is an urgent need for variant resistant neutralizing antibodies.

SUMMARY

The present disclosure provides antibodies, fragments thereof, and trimeric forms thereof, capable of binding to the SARS-CoV-2 spike protein. In some embodiments, provided is a trimeric antibody comprising three fusion polypeptides, each fusion polypeptide comprising an antigen binding domain fused to a trimerization domain, wherein the antigen binding domain has binding specificity to the receptor binding motif (RBM) of the receptor binding domain (RBD) of the SARS-Cov-2 spike protein.

In some embodiments, the trimerization domain is capable of mediating stable association of the trimeric antigen binding molecule. In some embodiments, the trimerization domain comprises an amino acid sequence selected from the group consisting of SEQ ID NO: 103, 107 and 111, preferably SEQ ID NO: 103.

In some embodiments, the trimeric antibody further comprises a peptide linker between the antigen binding domain and the trimerization domain. In some embodiments, the peptide linker is a flexible linker, preferably comprising the amino acid sequence of SEQ ID NO: 122 or 123. In some embodiments, the peptide linker is from 5 to 50 amino acid residues in length, preferably from 5 to 20 amino acid residues in length.

In some embodiments, the antigen binding domain binds to at least one of amino acid residues selected from the group consisting of 438, 447-456, 489-495, and 507-508 of SEQ ID NO: 126. In some embodiments, the antigen binding domain comprises a fragment of the present disclosure.

Also provided are methods for treating or preventing a SARS-CoV-2 viral infection in a subject, comprising administering to the subject an effective amount of the antibody or fragment thereof disclosed herein. In some embodiments, the subject suffers from a COVID-19 symptom.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 Selection of scFv antibodies targeting spike protein. a. Work flow of the panning process against S-RBD. b. Output vs. panning round for the antigen S-RBD-hFc during three rounds of screening. c. Phage ELISA results of 22 unique antibodies with positive readouts (OD ₄₀₅ ratio S-RBD-hFc/hFc > 2) .

FIG. 2 Analysis of antibody binding to cell surface-expressed trimeric spike protein. a HEK293T cells transfected with expression plasmid encoding the full-length spike of SARS-CoV-2 were incubated with purified IgG4 antibody and stained with PE labeled anti-human IgG4 Fc secondary antibody, then analyzed by FACS. Positive binding cells populations were labeled as positive pop. b-g FACS of antibodies binding to SARS-CoV spike, HCoV-229E spike, HCoV-HKU1 spike, HCoV-NL63 spike, HCoV-OC43 spike, MERS spike. h FACS of antibodies binding to non-transfected HEK293T cells. Cells stained with only secondary antibody were used as negative control (NC) . i Mean fluorescent intensity (MFI) of antibodies for SARS-CoV-2 spike binding.

FIG. 3 Competitive ELISA of antibodies with hACE2 and binding kinetics to the spike protein. a The three antibodies were titrated for competition with hACE2-ECD for binding to S-RBD and the fitting curves are shown. b-d Binding kinetics with wild-type (WT) S-RBD were measured by biolayer interferometry (BLI) . Biotinylated S-RBD was loaded onto the SA biosensor for detection of binding kinetics with S-B8 (b) and S-E6 (d) , while S-RBD amine coupled onto AR2G sensor was utilized for S-D4 (c) , with detection on an Octet. All curves were fitted by a 1: 1 binding model using the Data Analysis software (Forte Bio) . e-g Binding kinetics with N501Y S-RBD were measured by biolayer interferometry (BLI) . The method used was same as above, binding curves of S-B8 (e) , S-D4 (f) and S-E6 (g) were shown. h The association-rate (k _on) , dissociation-rate (k _off) and dissociation constant (K _D) of the three competitive antibodies to WT S-RBD and N501Y S-RBD are shown.

FIG. 4 Inhibition of syncytium formation by the antibodies. a Representative images of SARS-CoV-2 and SARS-CoV spike-mediated syncytium formation with hACE2 expressing cells 48 hours after co-culture. b, d Syncytium number calculation and inhibition rates when treated with 10 nM of hACE2 competitive antibodies are shown. S-RBD was used as the positive control. c, e Syncytium number and inhibition after treatment with antibodies and S-RBD at different concentrations are shown. The inhibition rates at 10 nM and 1 nM are summarized in f. Bars=400 μm. Error bars indicate SD, *P < 0.05, **P < 0.01, ***P < 0.001, determined by Student's T-test.

FIG. 5 Neutralization assay for the hACE2 competitive antibodies. a Neutralization ability of the three hACE2 competitive antibodies to WT SARS-CoV-2 pseudovirus was tested and fitted. c A microneutralization assay was adopted for testing of the three antibodies. NT ₅₀ and HillSlope for each antibody on authentic SARS-CoV-2 are summarized in b and d. e, f Neutralization ability of the three hACE2 competitive antibodies to SARS-CoV-2 N501Y+D614G mutant pseudovirus was tested and fitted, NT ₅₀ and HillSlope are shown. g Neutralization of three antibodies to real virus B. 1.1.7 strain, NT ₅₀ and HillSlope for each antibody are shown in h.

FIG. 6 Structural characterization of S-B8 and S-E6 integrating with SARS-CoV-2 S-RBD. Crystal structures are shown in ribbon representation with residues of interest in stick mode. The epitope surface on the RBD involved in interaction with the heavy and light chains of the antibodies are in orange and yellow, respectively. S-RBD is shown in white, S- B8 in blue and light blue for heavy and light chains, S-E6 heavy and light chains in orange and pink, and hACE2 in green. SHM residues are shown as semi-transparent magenta spheres and highlighted with magenta labels in c-h. a Structure comparison of S-B8 and S-E6 compared to hACE2 binding to the RBD in the same relative view. b Surface representation of S-RBD epitope residue interactions with S-B8 and S-E6. FRH1-CDRH1 and CDRH2 from both S-B8 (blue) and COVA2-39 (cyan, PDB 7JMP) are shown for comparison. c S-B8 CDRH1 and CDRH2 interaction with RBD. d Interaction between S-B8 and RBD ridge. e Interaction between S-E6 and RBD ridge. f Comparison to IGHV3-53 binding mode A. CC12.3 (pink for heavy chain and light pink for light chain) in complex with SARS-CoV-2 S-RBD (PDB 6XC7) illustrating the hydrogen bonding between the ₃₂NY ₃₃ motif and S-RBD. g Interaction between S-E6 and RBM mid-region. h Interaction between S-E6 and RBM on the opposite side of the S-RBD ridge.

FIG. 7 FACS analysis of antibody binding to cell surface-expressed mutated spike protein. HEK293T cells transfected with expression plasmid encoding the mutated full-length spike of SARS-CoV-2 were incubated with the three hACE2 competitive IgG4 antibodies. The cells were then stained with FITC labeled anti-human IgG Fc secondary antibody and analyzed by FACS. Cells stained with only secondary antibody were set as negative control (NC) . Positive binding cells populations were labeled as positive pop. a Mut1: D215H, b Mut2: S247R, c Mut3: D614G, d Mut4: N439K+D614G, e B. 1.1.7: spike of U. K. variant, f Mut5: N501Y+D614G, g Mut 6: N501Y+E484K+D614G, h Mut 7: N501Y+E484K+K417N+D614G, i Mut 8: N501Y+K417N+D614G. Grey line: NC, Red line: S-B8, Green line: S-D4, Blue line: S-E6. j Mean Fluorescent Intensity of Abs to each mutant are listed.

FIG. 8 Binding kinetics to the N501Y S-RBD. A-C. Binding kinetics with N501Y S-RBD were measured by biolayer interferometry (BLI) . Biotinylated N501Y S-RBD was loaded onto the SA biosensor for detection of binding kinetics with S-B8 (A) and S-E6 (C) , while N501Y S-RBD amine coupled onto AR2G sensor was utilized for S-D4 (B) , with detection on an Octet. All curves were fitted by a 1: 1 binding model using the Data Analysis software (Forte Bio) . D. The association-rate (k _on) , dissociation-rate (k _off) and dissociation constant (K _D) of the three competitive antibodies to N501Y S-RBD are shown.

FIG. 9 Neutralization to N501Y pseudovirus for the three hACE2 competitive antibodies. Neutralization ability of the three hACE2 competitive antibodies to SARS-CoV-2 N501Y+D614G mutant pseudovirus was tested and fitted, NT ₅₀ and HillSlope are shown in the lower panel.

FIG. 10 Neutralization to B. 1.1.7 authentic virus for the three hACE2 competitive antibodies. Neutralization ability of the three hACE2 competitive antibodies to authentic virus of B. 1.1.7 strain was tested and fitted, NT ₅₀ and HillSlope are shown in the lower panel.

FIG. 11 Neutralization to K417N+E484K+N501Y pseudovirus for S-D4 antibody. Neutralization ability of S-D4 antibody to SARS-CoV-2 K417N+E484K+N501Y+D614G mutant pseudovirus was tested and fitted, NT ₅₀ and HillSlope are shown in the lower panel.

FIG. 12 Neutralization to B. 1.351 authentic virus for the S-D4 antibody. Neutralization ability of S-D4 antibody to authentic virus of B. 1.351 strain was tested and fitted, NT ₅₀ and HillSlope are shown in the lower panel.

FIG. 13 Trimeric antibody shows enhanced neutralizing activity to wild-type SARS-CoV-2 pseudovirus. A. Scheme of monomer, dimer and trimer antibodies bound to spike protein. B. SDS-PAGE analysis of reduced form mono-, di-, and tri-S-E6. C. SEC-HPLC analyses demonstrate the homogeneity of mono- (cyan) , di- (purple) , and tri-S-E6 (orange) . D. Neutralization ability of the different valents of S-E6 to WT SARS-CoV-2 pseudovirus was tested and fitted, trimer S-E6 was constructed by fusing CH1 to T4F tag with 3 (G4S) linker. E. Neutralization ability of three trimer S-E6 with different trimerization tags to WT SARS-CoV-2 pseudovirus was tested and fitted, the tags were fused to CH1 with 3 (G4S) linker. F. Neutralization ability of three trimer S-E6 with different lengths of rigid linkers to WT SARS-CoV-2 pseudovirus was tested and fitted, the linkers were used to fuse T4F tag. G. Construction of monomer, dimer and trimer of antibodies. For the trimer antibody, heavy chain plasmid was constructed by fusing CH1 domain (dark blue) with trimerization tag (light orange) with linker (gray) , the C-terminal His tag (dark orange) was added for antibody purification.

FIG. 14 Trimeric S-E6 shows restored binding avidity to the spike RBD of B. 1.351 (Beta Strain) . Binding kinetics of dimer S-E6 (A) or trimer S-E6 (B) with wild-type (WT) S-RBD were measured by biolayer interferometry (BLI) . Binding kinetics of dimer S-E6 (C) or trimer S-E6 (D) with spike RBD of B. 1.351 were measured by BLI. The fitting curves were shown as grey dotted line. E. The association-rate (k _on) , dissociation-rate (k _off) and dissociation constant (K _D) of the antibodies to WT S-RBD and B. 1.351 S-RBD are shown. All curves were fitted by a 1: 1 binding model using the Data Analysis software (Forte Bio) .

FIG. 15 Trimeric S-E6 shows restored neutralizing ability to B. 1.351 (Beta Strain) PSV. A. Neutralization abilities of the different valents of S-E6 to SARS-CoV-2 B. 1.351 pseudovirus were tested and fitted. B. Neutralization ability of three trimer S-E6 with different trimerization tags to SARS-CoV-2 B. 1.351 pseudovirus was tested and fitted. C. Neutralization ability of three trimer S-E6 with different lengths of rigid linkers to SARS-CoV-2 B. 1.351 pseudovirus was tested and fitted.

FIG. 16 Trimer format of nAbs from different classes show variable neutralizing enhancement to SARS-CoV-2 B. 1.351 PSV. Neutralization abilities of the dimer or trimer antibodies from

class

1, 2 and 3 to B. 1.351 SARS-CoV-2 pseudovirus were tested and fitted. S-B8 (A) , S-D4 (B) , S-B2 (C) and S-D9 (D) are from class 1. S309 (E) is from class 2. CoV2-2489 (F) is from class 3. G. NT ₅₀ values fold increase upon trimerization.

FIG. 17 Trimer format of nAbs from class 1 show variable neutralizing enhancement to SARS-CoV-2 B. 1.617.2 PSV. Neutralization abilities of the dimer or trimer antibodies from class 1to SARS-CoV-2 B. 1.617.2 pseudovirus were tested and fitted. S-B8 (A) , S-D4 (B) , S-E6 (C) , S-B2 (D) and S-D9 (E) are from class 1. F. NT ₅₀ values fold increase upon trimerization.

FIG. 18 shows the binding preference of trimeric S-E6 and S-B8. Binding kinetics of Trimer S-E6 (A) or trimer S-B8 (B) with wild-type (WT) S-RBD were measured by biolayer interferometry (BLI) with Linoleic acid (Red) or without Linoleic acid (blue) . The supplement of Linoleic acid significantly decreased the binding of antibody to the Spike protein, indicating the antibody prefer to bind the up conformation.

DETAILED DESCRIPTION

In order that the present disclosure may be more readily understood, certain terms are first defined. Additional definitions are set forth throughout the detailed description.

The term “antibody” as referred to herein includes whole antibodies and any antigen binding fragment (i.e., “antigen-binding portion” ) or single chains thereof. Whole antibodies are glycoproteins comprising at least two heavy (H) chains and two light (L) chains inter-connected by disulfide bonds. Each heavy chain is comprised of a heavy chain variable region (abbreviated herein as V _H) and a heavy chain constant region. The heavy chain constant region is comprised of three domains, C _H1, C _H2, and C _H3. Each light chain is comprised of a light chain variable region (abbreviated herein as V _L) and a light chain constant region. The light chain constant region is comprised of one domain, C _L. The V _H and V _L regions can be further subdivided into regions of hypervariability, termed complementarity determining regions (CDRs) , interspersed with regions that are more conserved, termed framework regions (FR) . Each V _H and V _L is composed of three CDRs and four FRs, arranged from amino-terminus to carboxy terminus in the following order: FRl, CDRl, FR2, CDR2, FR3, CDR3, and FR4. The variable regions of the heavy and light chains contain a binding domain that interacts with an antigen. The constant regions of the antibodies can mediate the binding of the immunoglobulin to host tissues or factors, including various cells of the immune system (e.g., effector cells) and the first component (Clq) of the classical complement system.

The term “antigen-binding portion” of an antibody (or simply “antibody portion” or “fragment” ) , as used herein, refers to one or more fragments of an antibody that retain the ability to specifically bind to an antigen (e.g., a LAG-3 protein) . It has been shown that the antigen-binding function of an antibody can be performed by fragments of a full-length antibody. Examples of binding fragments encompassed within the term “antigen binding portion” of an antibody include (i) a Fab fragment, a monovalent fragment consisting of the V _L, V _H, C _L and C _H1 domains; (ii) a F (ab') ₂ fragment, a bivalent fragment comprising two Fab fragments linked by a disulfide bridge at the hinge region; (iii) a Fab' fragment, which is essentially a Fab with part of the hinge region (see, FUNDAMENTAL IMMUNOLOGY (Paul ed., 3. sup. rd ed. I993) ; (iv) a Fd fragment consisting of the V _H and C _H1 domains; (v) a F _v fragment consisting of the V _L and V _H domains of a single arm of an antibody, (vi) a dAb fragment (Ward et al., (1989) Nature 341: 544-546) , which consists of a VH domain; (vii) an isolated complementarity determining region (CDR) ; and (viii) a nanobody, a heavy chain variable region containing a single variable domain and two constant domains. Furthermore, although the two domains of the F _v fragment, V _L and V _H, are coded for by separate genes, they can be joined, using recombinant methods, by a synthetic linker that enables them to be made as a single protein chain in which the V _L and V _H regions pair to form monovalent molecules (known as single chain F _v (scFv) ; see e.g., Bird et al. (1988) Science 242: 423-426; and Huston et al. (1988) Proc. Natl. Acad. Sci. USA 85: 5879-5883) . Such single chain antibodies are also intended to be encompassed within the term “antigen binding portion” of an antibody. These antibody fragments are obtained using conventional techniques known to those with skill in the art, and the fragments are screened for utility in the same manner as are intact antibodies.

The terms “monoclonal antibody” or “monoclonal antibody composition” as used herein refer to a preparation of antibody molecules of single molecular composition. A monoclonal antibody composition displays a single binding specificity and affinity for a particular epitope.

The term “human antibody” , as used herein, is intended to include antibodies having variable regions in which both the framework and CDR regions are derived from human germline immunoglobulin sequences. Furthermore, if the antibody contains a constant region, the constant region also is derived from human germline immunoglobulin sequences. The human antibodies of the disclosure can include amino acid residues not encoded by human germline immunoglobulin sequences (e.g., mutations introduced by random or site-specific mutagenesis in vitro or by somatic mutation in vivo) . However, the term “human antibody” , as used herein, is not intended to include antibodies in which CDR sequences derived from the germline of another mammalian species, such as a mouse, have been grafted onto human framework sequences.

As used herein, an antibody that “specifically binds the SARS-CoV-2 spike protein” or “has specificity to the SARS-CoV-2 spike protein” is intended to refer to an antibody that binds to the SARS-CoV-2 spike protein but does not substantially bind to non-SARS-CoV-2 spike proteins. Preferably, the antibody binds to a SARS-CoV-2 spike protein with “high affinity” , namely with a K _D of 1 × 10 ^-7 M or less, more preferably 5 × 10 ^-8 M or less, more preferably 3 × 10 ^-8 M or less, more preferably 1 × 10 ^-8 M or less, more preferably 3 × 10 ^-9 M or less or even more preferably 1 × 10 ^-9 M or less.

The phrases “an antibody recognizing an antigen” and “an antibody specific for an antigen” are used interchangeably herein with the term “an antibody which binds specifically to an antigen. ”

Various aspects of the disclosure are described in further detail in the following subsections.

Trimeric Antibodies

The instant discovered that trimeric formats of the anti-spike RBD antibodies exhibited higher neutralization activities than their dimer counterparts. Also importantly and yet unexpectedly, the trimeric format neutralized the immune-escape strain B. 1.351 (beta) potently with an NT ₅₀ of 36 pM and completely neutralized another immune-escape strain, B. 1.617.2 (delta) , with an NT ₅₀ of 7 pM.

Example 3 systematically compared the effects of different trimeric tags, linkers between antibody and trimeric tags for the antibody neutralization. It was discovered that fusing with the T4F tag with a flexible linker (e.g., G4S) resulted in significantly increased binding avidity of S-E6 to spike RBD. Also surprisingly, it neutralized the immune-escape variant (E484K, K417N, N501Y, and D614G) pseudovirus (PSV) ultrapotently with NT ₅₀ of 0.036 nM, and shows an 8.9-fold increase of potency against B. 1.617.2 PSV. By applying this trimeric format engineering method to other RMB-engaging nAbs, similar neutralizing effects enhancement resulted. The NT ₅₀ values can improve up to 30-fold.

Based on the interaction epitopes, nAbs have been divided into three classes (Chen, et al., Nature Medicine 2021, 27) . Class 1 antibodies block soluble hACE2 binding and bind multiple proximal sites in the receptor binding motif (RBM) of the RBD. Class 2 nAbs bind the base of the RBD of the spike protein. Class 3 nAbs recognize the NTD of the spike protein. The instant data show that class 1 antibodies achieved the best enhancement by reformatting to trimers. Class 2 and class 3 antibodies, however, showed similar NT ₅₀ with impaired inhibition rates.

The sequence of the spike protein is known (Table A) . Amino acids 11-304 constitutes the N-terminal domain (NTD) , and amino acid 319-541 constitute the receptor binding domain (RBD) . Amino acid residues within the receptor binding motif (RBM) include 438, 447-456, 489-495, and 507-508 (shown in boxed in Table A) .

Table A. Sequence and domains of SARS-CoV-2 Spike Protein (SEQ ID NO: 126)

In accordance with one embodiment of the present disclosure, therefore, provided is a trimeric antibody that includes an antibody or fragment fused to a trimerization domain.

As demonstrated, the antibody or antigen-binding fragment has binding specificity to the SARS-CoV-2 spike protein, and preferably binds to the receptor binding motif (RBM) of the receptor binding domain (RBD) (class 1 neutralization antibody) . The RBM is known to include amino acid residues 438, 447-456, 489-495, and 507-508 of SEQ ID NO: 126. Therefore, in some embodiments, the antibody or antigen-binding fragment has binding specificity to the SARS-CoV-2 spike protein and binds to at least one of residues 438, 447-456, 489-495, and 507-508 of SEQ ID NO: 126.

In some embodiments, the antibody or antigen-binding fragment includes VH and VL sequences as exemplified in the instant application. For instance, the antibody or fragment thereof comprises a heavy chain variable region (VH) comprising heavy chain complementarity determining regions CDRH1, CDRH2, and CDRH3, and a light chain variable region (VL) comprising light chain complementarity determining regions CDRL1, CDRL2, and CDRL3. In some embodiments, the CDRH1, CDRH2, CDRH3, CDRL1, CDRL2, and CDRL3, respectively, comprise the amino acid sequences of (a) SEQ ID NO: 5-10; (b) SEQ ID NO: 27-32; (c) SEQ ID NO: 49-54; (d) SEQ ID NO: 71-76; or (e) SEQ ID NO: 89-94. In some embodiments, the antibody or fragment thereof does not bind to the SARS-CoV spike protein.

The antibody or antigen-binding fragment in the trimeric antibody can take any format known in the art, such as a Fab fragment, a scFv fragment, or a nanobody. When a Fab fragment is used, as demonstrated herein, the VH and VL further include CH1 and CL, respectively. In some embodiments, the trimerization domain is fused to the CH1. In some embodiments, the trimerization domain is fused to the CL.

A trimerization domain is a peptide sequence that is capable of mediating stable association of a trimeric molecule. Trimerization domains are known in the art, such as the domains in trimeric proteins responsible for mediating association of the trimeric protein.

Example trimerization domains include the T4 bacteriophage fibritin trimerization motif (T4F) , the GCN4 trimeric leucine zipper motif (GCN4) , and the human collagen XVIII derived homotrimerization domain (TIE) . Example sequences are provided in SEQ ID NO: 103, 107 and 111. A preferred trimerization domain is T4F (SEQ ID NO: 103) . In some embodiments, the trimerization domain is not longer than 100 amino acids, or not longer than 90, 80, 70, 60, or 50 amino acids.

In some embodiments, the fusion protein further includes a peptide linker between the antibody or antigen binding fragment and the trimerization domain. In some embodiments, the peptide linker is flexible, such as SEQ ID NO: 122 or 123.

In some embodiments, the distance between antibody or antigen binding fragment and the trimerization domain is not longer than 100 amino acids, or not longer than 90, 80, 70, 60, 50, 40, 30, 25, 20, 15 or 10 amino acids. In some embodiments, the peptide linker is from 5 to 50 amino acid residues in length, preferably from 5 to 20 amino acid residues in length.

Antibodies Against the SARS-Cov-2 Spike RBD

The present disclosure also provides antibodies and fragments thereof capable of binding to the SARS-CoV-2 spike protein. As demonstrated in the accompanying experimental examples, these antibodies have excellent neutralization abilities.

The present disclosure provides antibodies and fragments having specificity to the SARS-CoV-2 spike protein, in particular the receptor binding domain (RBD) . The antibodies of the disclosure are characterized by particular functional features or properties of the antibodies.

SARS-CoV-2 is a single-stranded RNA-enveloped virus. Its entire genome is 29, 881 bp in length (GenBank no. MN908947) , encoding 9860 amino acids. Gene fragments express structural and nonstructural proteins. The S, E, M, and N genes encode structural proteins, whereas nonstructural proteins, such as 3-chymotrypsin-like protease, papain-like protease, and RNA-dependent RNA polymerase, are encoded by the ORF region.

A glycosylated S (spike) protein covers the surface of SARS-CoV-2 and binds to the host cell receptor angiotensin-converting enzyme 2 (ACE2) , mediating viral cell entry. When the S protein binds to the receptor, TM protease serine 2 (TMPRSS2) , a type 2 TM serine protease located on the host cell membrane, promotes virus entry into the cell by activating the S protein. Once the virus enters the cell, the viral RNA is released, polyproteins are translated from the RNA genome, and replication and transcription of the viral RNA genome occur via protein cleavage and assembly of the replicase-transcriptase complex. Viral RNA is replicated, and structural proteins are synthesized, assembled, and packaged in the host cell, after which viral particles are released.

The total length of the spike protein is 1273 aa and consists of a signal peptide (amino acids 1-13) located at the N-terminus, the S1 subunit (14-685 residues) , and the S2 subunit (686-1273 residues) ; the last two regions are responsible for receptor binding and membrane fusion, respectively. In the S1 subunit, there is an N-terminal domain (14-305 residues) and a receptor-binding domain (RBD, 319-541 residues) ; the fusion peptide (FP) (788-806 residues) , heptapeptide repeat sequence 1 (HR1) (912-984 residues) , HR2 (1163-1213 residues) , TM domain (1213-1237 residues) , and cytoplasm domain (1237-1273 residues) comprise the S2 subunit.

Three potent neutralizing antibodies were discovered in this example. Interestingly, they do not cross-react with the SARS-CoV spike protein (FIG. 2b) , which is likely due to differences in the epitope (～70%difference) in the receptor binding site between SARS-CoV and SAR-CoV-2 targeted by these antibodies. In addition, neither of the three antibodies bind to other five HCoVs (FIG. 2c-g) . Moreover, autoreactivity assay in a HEp-2 cell ruled out that S-E6 and S-D4 originate from autoimmune responses, whereas S-B8 showed weak to moderate autoreactivity, which was increased in the S-B8 putative germline antibody.

The structural studies on S-E6 and S-B8 revealed several striking features of these combinatorial antibodies. The primary immune response to viral infection is followed by a secondary response that generates functionally better antibodies, where the binding energy can be refined by somatic hypermutation. The secondary immune response is for later encounter of the same antigen, and is the basis of vaccination. In cases of pandemics, such as SARS-CoV-2, avian influenza or Ebola virus, if the infection is not dealt with by the immune system in the first few days, the patient has a high probability of dying, and as a consequence, the immune system will not have enough time to refine the immune response. Consistently, neutralizing antibodies isolated from SARS-CoV-2 convalescent patients contain only a few amino-acid mutations that may be a result of weak B cell stimulation due to rapid viral clearance. Neutralizing antibodies isolated from convalescent patients shortly after infection may then possibly not be fully refined (matured) . In comparison, S-B8 and S-E6 exhibited higher levels of SHM, many of which are involved in specific interactions with SARS-CoV-2 RBD (S-RBD) . Nine of 13 SHM residues in S-B8 and eight of 22 in S-E6 are located in the antibody-antigen interface. While some of these SHM residues only use their peptide backbone, others rely on specific side chains for S-RBD binding. Interestingly, SHM in CDRH1 of S-E6 generates a ₃₃NY ₃₄ sequence that is similar to the ₃₂NY ₃₃ motif in IGHV3-53/3-66 antibodies, which are the most frequent germlines used in targeting the S-RBD, indicative that the combinatorial antibody library and the maturation process can yield alternate antibody solutions

The instant inventors screened a combinatorial human antibody library with the SARS-CoV-2 S RBD, and identified three highly potent antibodies that selectively bind the S protein and neutralize authentic SARS-CoV-2 virus. Compared to neutralizing antibodies from COVID-19 patients with generally low somatic hypermutation (SHM) , these antibodies contain over 13-22 SHMs, many of which are involved in specific interactions in their crystal structures with SARS-CoV-2 spike RBD.

The three antibodies, S-B8, S-D4 and S-E6, all competed strongly with the human ACE2 protein in binding to the S RDB, in a dose-dependent manner (FIG. 3) . Moreover, they showed potent neutralization against wild-type SARS-CoV-2 PSV infection in a dose-dependent manner (FIG. 5) . Also importantly, these antibodies were able to bind to certain natural mutants of the spike protein, such as D215H, S247R, D614G, N439K+D614G, N501Y+D614G, E484K+N501Y+D614G, K417N+E484K+N501Y+D614G, and K417N+N501Y+D614G (FIG. 7) . These data demonstrate that these antibodies can be suitable therapeutic agents against these natural SARS-CoV-2 variants.

In accordance with one embodiment of the present disclosure, therefore, provided is an antibody or fragment thereof that has specificity to the SARS-CoV-2 spike protein. The antibody or fragment includes a heavy chain variable region (VH) comprising heavy chain complementarity determining regions CDRH1, CDRH2, and CDRH3, and a light chain variable region (VL) comprising light chain complementarity determining regions CDRL1, CDRL2, and CDRL3.

In some embodiments, the CDRH1, CDRH2, CDRH3, CDRL1, CDRL2, and CDRL3 have the CDR sequences of those antibodies disclosed herein, such as those provided in Tables 1A-1E.

In one embodiment, the reference antibody is S-B8, which has a VH sequence of SEQ ID NO: 1 and a VL sequence of SEQ ID NO: 2. Its CDR sequences are SEQ ID NO: 5-10 according to Kabat numbering, and SEQ ID NO: 11-16 according to Chothia numbering. In some embodiments, the antibody or fragment thereof of the disclosure has CDRH1, CDRH2, CDRH3, CDRL1, CDRL2, and CDRL3 having the amino acid sequences of SEQ ID NO: 5-10, respectively. In some embodiments, the antibody or fragment thereof of the disclosure has CDRH1, CDRH2, CDRH3, CDRL1, CDRL2, and CDRL3 having the amino acid sequences of SEQ ID NO: 11-16, respectively.

In one embodiment, the reference antibody is S-D4, which has a VH sequence of SEQ ID NO: 23 and a VL sequence of SEQ ID NO: 24. Its CDR sequences are SEQ ID NO: 27-32 according to Kabat numbering, and SEQ ID NO: 33-38 according to Chothia numbering. In some embodiments, the antibody or fragment thereof of the disclosure has CDRH1, CDRH2, CDRH3, CDRL1, CDRL2, and CDRL3 having the amino acid sequences of SEQ ID NO: 27-32, respectively. In some embodiments, the antibody or fragment thereof of the disclosure has CDRH1, CDRH2, CDRH3, CDRL1, CDRL2, and CDRL3 having the amino acid sequences of SEQ ID NO: 33-38, respectively.

In one embodiment, the reference antibody is S-E6, which has a VH sequence of SEQ ID NO: 45 and a VL sequence of SEQ ID NO: 46. Its CDR sequences are SEQ ID NO: 49-54 according to Kabat numbering, and SEQ ID NO: 55-60 according to Chothia numbering. In some embodiments, the antibody or fragment thereof of the disclosure has CDRH1, CDRH2, CDRH3, CDRL1, CDRL2, and CDRL3 having the amino acid sequences of SEQ ID NO: 49-54, respectively. In some embodiments, the antibody or fragment thereof of the disclosure has CDRH1, CDRH2, CDRH3, CDRL1, CDRL2, and CDRL3 having the amino acid sequences of SEQ ID NO: 55-60, respectively.

In one embodiment, the reference antibody is S-B2, which has a VH sequence of SEQ ID NO: 67 and a VL sequence of SEQ ID NO: 68. Its CDR sequences are SEQ ID NO: 71-76 according to Kabat numbering, and SEQ ID NO: 77-82 according to Chothia numbering. In some embodiments, the antibody or fragment thereof of the disclosure has CDRH1, CDRH2, CDRH3, CDRL1, CDRL2, and CDRL3 having the amino acid sequences of SEQ ID NO: 71-76, respectively. In some embodiments, the antibody or fragment thereof of the disclosure has CDRH1, CDRH2, CDRH3, CDRL1, CDRL2, and CDRL3 having the amino acid sequences of SEQ ID NO: 77-82, respectively.

In one embodiment, the reference antibody is S-D9, which has a VH sequence of SEQ ID NO: 85 and a VL sequence of SEQ ID NO: 86. Its CDR sequences are SEQ ID NO: 89-94 according to Kabat numbering, and SEQ ID NO: 95-100 according to Chothia numbering. In some embodiments, the antibody or fragment thereof of the disclosure has CDRH1, CDRH2, CDRH3, CDRL1, CDRL2, and CDRL3 having the amino acid sequences of SEQ ID NO: 89-94, respectively. In some embodiments, the antibody or fragment thereof of the disclosure has CDRH1, CDRH2, CDRH3, CDRL1, CDRL2, and CDRL3 having the amino acid sequences of SEQ ID NO: 95-100, respectively.

Also provided, in one embodiment, is an antibody or fragment thereof, wherein the antibody or fragment thereof has specificity to the SARS-CoV-2 spike protein, and competes with an antibody or fragment thereof of the present disclosure in binding to the SARS-CoV-2 spike protein, or binds to the same epitope as the antibody or fragment thereof. In one embodiment, the antibody or fragment thereof is a blocking antibody or fragment thereof. In one embodiment, the antibody or fragment thereof is a non-blocking antibody or fragment thereof.

In various embodiments, the antibody can be, for example, a human antibody. In other embodiments, the V _H and/or V _L amino acid sequences can have at least 85%, 90%, 95%, 96%, 97%, 98%, or 99%sequence identify to the sequences set forth above. An antibody having V _H and V _L regions having high (i.e., 80%or greater) homology to the V _H and V _L regions of the sequences set forth above, can be obtained by mutagenesis (e.g., site-directed or PCR-mediated mutagenesis) of nucleic acids of V _H and/or V _L amino acid sequences, followed by testing of the encoded altered antibody for retained function (i.e., the functions set forth above) using the functional assays described herein.

The CDR regions recited in this disclosure can also be changed to each of its biological variants. A biological variant of CDR sequence is derived from the original sequence by one, two or three amino acid addition, deletion and/or substitutions. In some embodiments, the substitution is conservative amino acid substitution.

A “conservative amino acid substitution” is one in which the amino acid residue is replaced with an amino acid residue having a similar side chain. Families of amino acid residues having similar side chains have been defined in the art, including basic side chains (e.g., lysine, arginine, histidine) , acidic side chains (e.g., aspartic acid, glutamic acid) , uncharged polar side chains (e.g., glycine, asparagine, glutamine, serine, threonine, tyrosine, cysteine) , nonpolar side chains (e.g., alanine, valine, leucine, isoleucine, proline, phenylalanine, methionine, tryptophan) , beta-branched side chains (e.g., threonine, valine, isoleucine) and aromatic side chains (e.g., tyrosine, phenylalanine, tryptophan, histidine) . Thus, a nonessential amino acid residue in an immunoglobulin polypeptide is preferably replaced with another amino acid residue from the same side chain family. In another embodiment, a string of amino acids can be replaced with a structurally similar string that differs in order and/or composition of side chain family members.

Bi-functional Molecules

In some embodiments, the present disclosure provides bifunctional or bispecific molecules comprising an anti-spike protein antibody/fragment linked to at least one other functional molecule, e.g., another peptide or protein (e.g., another antibody or ligand for a receptor) to generate a bifunctional or bispecific molecule that binds to at least two different binding sites or target molecules. Thus, as used herein, “bispecific molecule” includes molecules that have three or more specificities. In a preferred embodiment, the bispecific molecule comprises a first binding specificity for the SARS-CoV-2 spike protein and a second binding specificity for a triggering molecule that recruits cytotoxic effector cells that can kill the SARS-CoV-2 virus. Examples of suitable triggering molecules are CD64, CD89, CD16, and CD3. See, e.g., Kufer et al., Trends in Biotech. 22 (5) : 238-44, 2004.

In some embodiments, the second function/specificity can be for an anti-enhancement factor (EF) , e.g., a molecule that binds to a surface protein involved in cytotoxic activity and thereby increases the immune response against the target virus or an infected cell. For example, the anti-enhancement factor can bind a cytotoxic T cell (e.g. via CD2, CD3, CDS, CD28, CD4, CD40, or ICAM-1) , other immune regulatory molecules (e.g. via PD-1, PD-L1, CTLA-4, CD122, 4-1BB, TIM3, OX-40, OX40L, CD40L, LIGHT, ICOS, ICOSL, GITR, GITRL, TIGIT, CD27, VISTA, B7H3, B7H4, HEVM, BTLA, KIR, CD47 or CD73) or other immune cell, resulting in an increased immune response against the virus or an infected cell.

Bifunctional/bispecific molecules also encompass bi-epitopic ones, which have a first specificity to one portion of a target antigen and a second specificity to another portion of the same antigen. The other portion may or may not overlap with the first portion. In some embodiments, the binding to other portion may not, on its own, has the intended blocking activity, but enhances the activity of the first specificity. The enhancement, without being bound by any particular theory, may be due to tighter binding or stabilized conformation. In some embodiments, both bindings can independently exhibit the desired activities.

Bifunctional molecules that include not just antibody or antigen binding fragment are also provided. As a tumor antigen targeting molecule, an antibody or antigen-binding fragment specific to the spike protein, such as those described here, can be combined with an immune cytokine or ligand optionally through a peptide linker. The linked immune cytokines or ligands include, but not limited to, IL-2, IL-3, IL-4, IL-5, IL-6, IL-7, IL-10, IL-12, IL-13, IL-15, GM-CSF, TNF-α, CD40L, OX40L, CD27L, CD30L, 4-1BBL, LIGHT and GITRL.

Bispecific molecules can come in many different formats and sizes. At one end of the size spectrum, a bispecific molecule retains the traditional antibody format, except that, instead of having two binding arms of identical specificity, it has two binding arms each having a different specificity. At the other extreme are bispecific molecules consisting of two single-chain antibody fragments (scFv's) linked by a peptide chain, a so-called Bs (scFv) ₂ construct. Intermediate-sized bispecific molecules include two different F (ab) fragments linked by a peptidyl linker. Bispecific molecules of these and other formats can be prepared by genetic engineering, somatic hybridization, or chemical methods. See, e.g., Kufer et al., supra; Cao and Suresh, Bioconjugate Chem. 9 (6) : 635-44, 1988; and van Spriel et al., Immunol. Today 21 (8) : 391-7, 2000; and the references cited therein.

Combinations and Pharmaceutical Compositions

In one embodiment, the treatment methods can further include administration of an effective amount of another agent. In some embodiments, the anti-spike protein antibody or fragment is co-administered with an effective amount of the another agent. In some embodiments, the second agent is also an anti-spike antibody of fragment thereof. In some embodiments, the second agent is co-administered with the antibody or fragment thereof simultaneously or sequentially.

In some embodiments, the second agent is effective in reducing or inhibiting cytokine release storm. In some embodiments, the second agent is a corticosteroid. Non-limiting examples include methylprednisolone (in particular in patients with a rheumatic disease) , dexamethasone (in particular in patients with FHLH) .

In some embodiments, the second agent is a cytoablative therapy. Non-limiting examples include cyclophosphamide (in particular in patients with JIA and MAS) , etoposide (in particular in patients with FHLH) , rituximab (in particular in Epstein-Barr virus (EBV) -associated HLH) , antithymocyte globulin (in particular for patients at bone marrow transplant phase of FHLH therapy) , alemtuzumab (in particular in patients with FHLH or SLE-associated MAS) .

In some embodiments, the second agent is a T-cell modulator. Non-limiting examples include calcineurin (e.g., cyclosporine) which prevents production of IL-2, and abatacept, which inhibits CD28 signaling of T cells. In some embodiments, the second agent is an anti-GM-CSF inhibitor or antibody.

In some embodiments, the second agent is a cytokine inhibitor, such inhibitors targeting INFγ, IL-1β, IL-18, IL-33, IL-6, and/or TNF.

In some embodiments, the second agent targets the underlying disease or condition, such as SARS-CoV-2 infection. Non-limiting examples include lopinavir, ritonavir, oseltamivir (Tamiflu) , favipiravir, fingolimod, methylprednisolone, bevacizumab, chloroquine phosphate, chloroquine, hydroxychloroquine sulfate and remdesivir.

In another aspect, the present disclosure provides a pharmaceutical composition comprising an antibody of the present disclosure formulated together with a pharmaceutically acceptable earlier. It may optionally contain one or more additional pharmaceutically active ingredients, such as another antibody or a drug. The pharmaceutical compositions of the disclosure also can be administered in a combination therapy with, for example, an anti-viral agent, or a vaccine.

The pharmaceutical composition can comprise any number of excipients. Excipients that can be used include carriers, surface active agents, thickening or emulsifying agents, solid binders, dispersion or suspension aids, solubilizers, colorants, flavoring agents, coatings, disintegrating agents, lubricants, sweeteners, preservatives, isotonic agents, and combinations thereof. The selection and use of suitable excipients is taught in Gennaro, ed., Remington: The Science and Practice of Pharmacy, 20th Ed. (Lippincott Williams &Wilkins 2003) , the disclosure of which is incorporated herein by reference. Preferably, a pharmaceutical composition is suitable for intravenous, intramuscular, subcutaneous, parenteral, spinal or epidermal administration (e.g., by injection or infusion) . Depending on the route of administration, the active compound can be coated in a material to protect it from the action of acids and other natural conditions that may inactivate it. The phrase “parenteral administration” as used herein means modes of administration other than enteral and topical administration, usually by injection, and includes, without limitation, intravenous, intramuscular, intraarterial, intrathecal, intracapsular, intraorbital, intracardiac, intradermal, intraperitoneal, transtracheal, subcutaneous, subcuticular, intraarticular, subcapsular, subarachnoid, intraspinal, epidural and intrastemal injection and infusion. Alternatively, an antibody of the disclosure can be administered via a non-parenteral route, such as a topical, epidermal or mucosal route of administration, e.g., intranasally, orally, vaginally, rectally, sublingually or topically.

Pharmaceutical compositions can be in the form of sterile aqueous solutions or dispersions. They can also be formulated in a microemulsion, liposome, or other ordered structure suitable to high drug concentration.

The amount of active ingredient which can be combined with a carrier material to produce a single dosage form will vary depending upon the subject being treated and the particular mode of administration and will generally be that amount of the composition which produces a therapeutic effect. Generally, out of one hundred percent, this amount will range from about 0.01%to about ninety-nine percent of active ingredient, preferably from about 0.1%to about 70%, most preferably from about 1%to about 30%of active ingredient in combination with a pharmaceutically acceptable carrier.

Dosage regimens are adjusted to provide the optimum desired response (e.g., a therapeutic response) . For example, a single bolus can be administered, several divided doses can be administered over time or the dose can be proportionally reduced or increased as indicated by the exigencies of the therapeutic situation. It is especially advantageous to formulate parenteral compositions in dosage unit form for ease of administration and uniformity of dosage. Dosage unit form as used herein refers to physically discrete units suited as unitary dosages for the subjects to be treated; each unit contains a predetermined quantity of active compound calculated to produce the desired therapeutic effect in association with the required pharmaceutical carrier. Alternatively, antibody can be administered as a sustained release formulation, in which case less frequent administration is required.

For administration of the antibody, the dosage ranges from about 0.0001 to 100mg/kg, and more usually 0.01 to 5 mg/kg, of the host body weight. For example, dosages can be 0.3 mg/kg body weight, 1 mg/kg body weight, 3 mg/kg body weight, 5mg/kg body weight or 10 mg/kg body weight or within the range of 1-10 mg/kg. An exemplary treatment regime entails administration once per week, once every two weeks, once every three weeks, once every four weeks, once a month, once every 3 months or once every 3 to 6 months. Preferred dosage regimens for an antibody of the disclosure include 1 mg/kg body weight or 3 mg/kg body weight via intravenous administration, with the antibody being given using one of the following dosing schedules: (i) every four weeks for six dosages, then every three months; (ii) every three weeks; (iii) 3 mg/kg body weight once followed by 1 mg/kg body weight every three weeks. In some methods, dosage is adjusted to achieve a plasma antibody concentration of about 1-1000 μg/mL and in some methods about 25-300 μg/mL.

A “therapeutically effective dosage” of an antibody of the disclosure preferably results in a decrease in severity of disease symptoms, an increase infrequency and duration of disease symptom-free periods, or a prevention of impairment or disability due to the disease affliction. For example, for the treatment of tumor bearing subjects, a “therapeutically effective dosage” preferably inhibits tumor growth by at least about 20%, more preferably by at least about 40%, even more preferably by at least about 60%, and still more preferably by at least about 80%relative to untreated subjects. A therapeutically effective amount of a therapeutic compound can decrease tumor size, or otherwise ameliorate symptoms in a subject, which is typically a human or can be another mammal.

Uses and Methods

The antibodies, antibody compositions and methods of the present disclosure have numerous in vitro and in vivo utilities involving, for example, detection of a SARS-CoV-2 spike protein or preventing or treating SARS-CoV-2 viral infection. In a preferred embodiment, the antibodies of the present disclosure are human antibodies. For example, these molecules can be administered to cells in culture, in vitro or ex vivo, or to human subjects, e.g., in vivo, to enhance immunity in a variety of situations. Accordingly, in one aspect, the disclosure provides a method of modifying an immune response in a subject comprising administering to the subject the antibody, or antigen-binding portion thereof, of the disclosure such that the immune response in the subject is modified. Preferably, the response is enhanced, stimulated or up-regulated.

Preferred subjects include human patients infected with the SARS-CoV-2 virus or is at risk of developing SARS-CoV-2 infection.

The disclosure further provides methods for detecting the presence of a SARS-CoV-2 virus in a sample, or measuring the amount of the SARS-CoV-2 virus, comprising contacting the sample, and a control sample, with an antibody or an antigen binding thereof of the present disclosure, under conditions that allow for formation of a complex between the antibody or portion thereof and the SARS-CoV-2 spike protein. The formation of a complex is then detected, wherein a difference complex formation between the sample compared to the control sample is indicative the presence of a SARS-CoV-2 virus in the sample. Moreover, the antibodies of the disclosure can be used to purify SARS-CoV-2 spike proteins.

EXAMPLES

The following examples are included to demonstrate preferred embodiments of the disclosure. It should be appreciated by those of skill in the art that the techniques disclosed in the examples which follow represent techniques discovered by the inventors to function well in the practice of the disclosure, and thus can be considered to constitute preferred modes for its practice. However, those of skill in the art should, in light of the present disclosure, appreciate that many changes can be made in the specific embodiments which are disclosed and still obtain a like or similar result without departing from the concept, spirit and scope of the disclosure. More specifically, it will be apparent that certain agents which are both chemically and physiologically related may be substituted for the agents described herein while the same or similar results would be achieved. All such similar substitutes and modifications apparent to those skilled in the art are deemed to be within the spirit, scope and concept of the disclosure as defined by the appended claims.

Example 1. Identification of Human Antibodies

This example reports the selection of three potent SARS-CoV-2 antibodies, S-E6, S-B8 and S-D4, from a pre-pandemic human combinatorial antibody library.

Two of these antibodies showed highly potent neutralization effects with apparent neutralizing NT50 as low as 0.025 nM to pseudovirus and 0.25 nM to authentic SARS-CoV-2 virus. Antibody germline classification suggests that the most potent antibody, S-E6, is derived from a less common germline, IGHV4-31, compared to the most frequent IGHV3-53/3-66, IGHV1-2 or IGHV3-30 germlines reported so far. Furthermore, a high level of somatic hypermutation (up to 15 amino acids per molecule) was found for all three antibodies, despite being derived from different germline genes. X-ray structural studies revealed that these S-E6 and S-B8 antibodies bind to the receptor binding site, but with different angles of approach to the RBD and slightly different epitopes.

Materials and Methods:

Cell culture

The Vero cell line (

CCL-81 ^TM) was maintained in a DMEM/F-12k media (Gibco, C11330500CP) containing 10% (v/v) FBS (Gibco, 1600074) . The FreeStyle ^TM 293-F (HEK 293F, ThermoFisher Scientific, R79007) cell line was cultured in a Freestyle 293 expression media (ThermoFisher Scientific, 12338026) . For establishing the HEK293T/hACE2 stable cell line, HEK293T cells (

ACS-4500 ^TM) were transiently transfected with hACE2 fusion BFP encoding PB510 plasmid using PiggyBac Transposon System (System Biosciences, PB210PA-1) , followed by addition of 2 μg/mL puromycin 6 h post-transfection. The resulting cells were kept in puromycin-containing media for an extra 2 days. Positive cells with BFP expression were sorted by a flow cytometry instrument (BD FACS Aria III) . The sorted cells with overexpressed hACE2 were expanded and cultured in a DMEM media (Gibco, 10566016) supplemented with 10%FBS (v/v) and 10 μg/mL puromycin.

Expression and purification of recombinant SARS-CoV-2 spike RBD, human ACE2 and antibodies

The DNA sequences of codon-optimized SARS-CoV-2 Spike Receptor Binding Domain (S-RBD) and human ACE2 Extracellular Domain (hACE2-ECD) were cloned into a pFuse-Fc expression vector (Invivogen) . A thrombin cleavage sequence was inserted between the RBD and Fc to generate a cleavable human Fc tag for future studies. The SARS-CoV-2 S-RBD-hFc and hACE2-ECD-mFc proteins were heterologously expressed in HEK293F cells by transient transfection and cultured for 4 days, then purified by Mabselect columns (Cytiva, 17-5199-01) . Briefly, cell media with secreted Fc tagged recombinant proteins, S-RBD-hFc and hACE2-ECD-mFc, were loaded onto a Mabselect column that was pre-washed and equilibrated with a PBS buffer (150 mM NaCl, 20 mM sodium phosphate, pH 7.2) , and eluted using a pH 3.4 citrate acid buffer.

DNA sequences for the variable regions of the combinatorial antibodies were cloned into a full-length human IgG4 mutant construct (S228P) and expressed in HEK293F cells for 4 days and further purified by Mabselect chromatography. Purified recombinant proteins and antibodies were buffer-exchanged into a PBS buffer using centrifugal concentrators.

Function-guided phage panning

SARS-CoV-2 S-RBD specific scFv antibodies were selected from a combinatorial human monoclonal scFv antibody phage library (10 ¹¹ members) after two rounds of affinity enrichment against the biotinylated S-RBD protein immobilized on the streptavidin-coated magnetic beads (Pierce, 21925) , followed by a third round of competitive panning vs. hACE2-ECD protein. Briefly, phagemid (displaying the antibody library) binding to the antigen (S-RBD) was enriched at each cycle and eluted with Glycine-HCl (pH 2.2) in the first two rounds of screening. XL1-Blue cells were used to express and amplify the output phagemids for the next round of panning. To determine hACE2 competitive antibodies, a kinetic competitive panning method was adopted in the third round panning. Instead of the conventional pH 2.2 buffer, an elution buffer containing a saturated concentration of hACE2-ECD protein (200 nM; for S-RBD and hACE2-ECD binding, EC ₈₀ = 80 nM) was used to elute the phagemids twice. After three iterations, 96 positive colonies were selected and analyzed by phage ELISA. All of the positive clones were sequenced using Sanger sequencing. Both the DNA and protein sequences of CDR3 domains were analyzed using the international ImMunoGeneTics (IMGT) information platform (www. imgt. org) .

Phage ELISA

Avidin (Pierce, 21121) was diluted to a final concentration of 2 ng/μL in a PBS buffer (Sigma, C3041) . The resulting avidin solution was used to coat the 96-half well plates (25 μL/well) at 4 ℃ overnight. The coated plates were washed once with the PBS buffer (150 μL/well) followed by the addition and incubation of 25 μL biotinylated SARS-CoV-2 S-RBD-hFc solution (2 ng/μL) in each well at room temperature for 1 h. The PBST (PBS containing 0.05%Tween-20) buffer alone and the hFc solution (2 ng/μL) were used as the background and negative controls, respectively. After removal of the incubation solution, the resulting plates were rinsed once using the PBST buffer and incubated with a blocking solution containing 5%milk (v/v) in PBST (150 μL/well) at 37 ℃ for 1 h. After blocking and PBST washing (once) , 50 μL of phagemid-containing XL1-Blue culture medium supernatants (by centrifuging the third round panning output XL1-Blue cells at 3000 g, 15 min) mixed with 10 μL 5%milk (v/v) in PBST was added to each well and incubated at 37 ℃ for 1 h. The resulting plates were rinsed eight times using PBST before subjecting to horseradish peroxidase (HRP) detection. A solution containing the secondary antibody, anti-M13 bacteriophage antibody conjugated with HRP (dilution factor 1: 5000; Sino Biological, 11973-MM05T-H) , was added into the above plates (150 μL/well) and incubated at 37 ℃ for 1 h. Plates were then washed eight times with PBST followed by the addition of 50 μL ABTS solution (Roche, 11684302001) into each well. After ～10 min incubation at room temperature, the absorbance change at 405 nm in each well was measured on a microplate reader (Enspire, PerkinElmer) .

Competitive ELISA

Competition between the selected antibodies and hACE2 for binding to the SARS-CoV-2 spike protein RBD was measured. The recombinant hACE2-ECD was coated in PBS buffer at 2 ng/μL, 100 μL per well at 4 ℃ overnight, washed with PBS once, then blocked with 3%BSA in PBS. Biotinylated S-RBD (hFc tag removed by thrombin digestion) at a final concentration of 50 nM was incubated with 2-fold serial diluted S-B8, S-D4, and S-E6 antibodies (from 1 -133 nM) at 4 ℃ for 30 min, in which an IgG4e1 isotype antibody was used as the negative control. The S-RBD and antibody mixture was then added to the hACE2-ECD coated plates and incubated at room temperature for 1 h, followed by 4 washes with PBST. The hACE2-ECD bound S-RBD in the plate was detected using a Streptavidin-HRP conjugated protein.

Affinity determination by Biolayer Interferometry (BLI)

Binding affinities of S-D4 with SARS-CoV-2 wild-type or mutant S-RBD were performed by BLI on an Octet RED96 (Molecular Devices LLC, San Jose, CA, USA) using AR2G biosensors. The SARS-CoV-2 S-RBD fused hFc was first digested by thrombin to remove the Fc tag. The resulting S-RBD diluted in a PBS solution containing 0.02%Tween-20 and 0.05%BSA (PBST-B) (10 μg/mL) was loaded to the AR2G biosensor by amine coupling. The AR2G-S-RBD sensors were dipped into a PBST-B for 60 sec to establish a baseline, and then incubated with 2-fold serial diluted antibody solutions to record the progressive curves of association. Finally, sensors were incubated in a PBST-B buffer to record the progressive curves of dissociation. For S-B8 and S-E6 detections, S-RBD was first biotinylated before loading to a streptavidin (SA) sensor, the remaining procedure was same to that of S-D4. Sensor regeneration was performed by dipping the used sensors into a pH 3.4 citrate acid buffer, and equilibrated in a PBST-B buffer. Results were analyzed by ForteBio Data Analysis software.

Interaction of antibodies with cell surface expressed spike by FACS

In a flow-cytometry binding experiment, the spike protein of either full-length SARS-CoV-2 or SARS, which was conjugated with P2A-EGFP, was transiently transfected into a HEK293T cell. After 24 h cultivation, cells were collected and re-suspended in an ice-cold FACS buffer (PBS, 0.05 %BSA and 2 mM EDTA) . The spike protein expressing cells (50,000 cells per tube) were then incubated with different anti-S-RBD antibodies for 20 min at 4 ℃, and washed with 1 mL ice-cold FACS buffer, spun, and re-suspended in a 100 μL ice-cold FACS buffer containing the Alexa555 conjugated secondary antibody that recognizes human Fc (1: 800 v/v dilution, Life technology, A21433) . After incubating at 4 ℃ for 15 min, the cells were washed twice and re-suspended in a FACS buffer, and then sorted and analyzed on a flow cytometer (CytoFLEX S, Beckman Culter) to determine relative binding level by the antibodies to the cell overexpressing wild-type spikes. Mean fluorescence intensities of Alexa555 in eGFP-positive cells were recorded and analyzed to evaluate antibody binding.

Size-exclusion-high-performance liquid chromatography (SEC-HPLC)

Twenty μL of 0.5 μg/μL purified S-RBD antibodies were applied to an Agilent Bio SEC-5, 500A HPLC system. The mobile phase used PBS buffer (pH 7.2) running at a flow rate of 0.35 mL/min. Absorbance was analyzed and integrated by retention time and area under the curve (AUC) to determine the percentage of aggregation, monomer and degradants compositions.

Cell-cell fusion assay

The cell-cell fusion assay was established as follows. Briefly, hACE2 positive Vero cells (cells with endogenous hACE2 were sorted by FACS) were used as target cells. HEK293F cells that are transiently transfected with either SARS-CoV-2 spike-P2A-EGFP or SARS spike-P2A-EGFP were set as effector cells. The target Vero cells were first seeded into 24-well plates at a density of 1×10 ⁵/well and cultivated at 37 ℃ for 4 h, followed by addition of effector cells, HEK293F/SARS spike-EGFP or HEK293F/SARS-CoV-2 spike-EGFP, at a ratio of 2: 1, respectively. The co-cultures of cells were cultivated in a DMEM medium with 10%FBS, and treated with or without anti-SARS-CoV-2 spike antibodies at indicated concentrations. The recombinant SARS-CoV-2 S-RBD was used as a positive control. After cultivating at 37 ℃ for 6 h, the rates of cell-cell fusion were evaluated using a fluorescence microscope (EVOS M5000, Life Technologies) . Five fields for microscopic analysis were randomly selected in each treated group, the numbers of fused and unfused EGFP positive cells were counted.

Preparation of pseudovirus

HEK293T cells were co-transfected with both NL4-3 mCherry Luciferase plasmid (addgene, 44965) and pcDNA3.1 Wild-type or mutant SARS-CoV-2 spikeΔ19 plasmid (encoding SARS-CoV-2 spike protein, with 19 AA truncated in C terminal) using Lipofectamine 3000 (Invitrogen, L3000-015) following the manufacturer’s instruction. Pseudotyped particles were readily released into the supernatant. The supernatants containing SARS-CoV-2 pseudovirus were harvested at 48 h post-transfection, filtered (0.45 μm pore size, Sartorius, 16533-K) , and mixed with the Lenti-X Concentrator (Takara, 631231) overnight at 4 ℃. The mixture was then centrifuged at 1500 g for 45 min at 4 ℃. The cell pellets were collected and re-suspended in a DMEM medium and stored at -80 ℃ until use.

Pseudovirus-based neutralization assay

To detect the neutralization ability of selected antibodies against infection of coronavirus pseudovirus (PSV) , HEK293T/hACE2 cells were first seeded into 96-well white bottom plates at a density of 1×10 ⁴/well, and cultivated overnight. The PSV was pre-incubated with an equal volume of different concentrations of selected antibodies (dilution factor: 3.16, from 200 nM to 200 fM for S-B8 and S-D4, 200 nM to 6.3 fM for S-E6) in DMEM at 37 ℃ for 30 min. DMEM with or without PSV in the absence of antibodies were set as controls. After incubation, the PSV mixture was transferred to the culture plates containing HEK293T/hACE2 cells. The DMEM media containing PSV and antibodies were replaced with fresh media after 16 h treatment, cells were incubated for an additional 48 h. PSV infection efficacy was evaluated by luciferase activity using Bright-Lumi ^TM Firefly Luciferase Reporter Gene Assay Kit (Beyotime, RG015M) . Fifty microliter of luciferase substrate was added to each well, and the relative luminescence unit (RLU) values were measured on an Envision plate reader (PerkinElmer, Ensight) .

Authentic SARS-CoV-2 virus neutralization assay

The study was performed in the CL3 Facility of the University of Oxford operating under license from the HSE, on the basis of an agreed Code of Practice, Risk Assessments (under ACDP) and Standard Operating Procedures. In brief, this rapid, high-throughput assay determines the concentration of antibody that produces a 50%reduction in infectious focus-forming units of different authentic SARS-CoV-2 strains in Vero cells, as follows. Quadruplicate, 0.5log ₁₀ serial dilutions of antibody (11 steps from 100 nM to 1 pM) were pre- incubated with a fixed dose of SARS-CoV-2 (Victoria 01/2020 isolate) before incubation with Vero cells. A 1.5%carboxymethyl cellulose-containing overlay was used to prevent satellite focus formation. Twenty hours post-infection, the monolayers were fixed with 4%paraformaldehyde, permeabilized with 2%Triton X-100 and stained for N antigen using mAb EY 2A. After development with a peroxidase-conjugated antibody and True Blue peroxidase substrate, infectious foci were enumerated by ELISPOT reader. Data were analyzed using four-parameter logistic regression (Hill equation) in GraphPad Prism 8.3.

Autoreactivity assay

The autoreactivity assay was performed using a HEp-2 anti-nuclear antibodies (ANA) kit (Medical &Biological Laboratories Co., Ltd, 4220-12CN) according to the manufacturer’s instructions. Briefly, 35 μL of 0.1 mg/mL antibodies were loaded to the wells in a slide pre-seeded with fixed and permeabilized HEp-2 cells and incubated for 20 min at room temperature. Positive serum from autoimmune patients and negative serum from healthy donors provided by the kit were used as controls. After washing twice (5 min each) , the FITC-conjugated secondary anti-human antibody was incubated with the cells for 20 min at room temperature. The slide was then washed and mounted with a coverslip before observation on a fluorescent microscope (ZEISS, Axio Observer A1) with a 20× objective.

Protein production and structure determination

The coding sequence for receptor binding domain (RBD; residues 319-541) of the SARS-CoV-2 spike (S) protein was synthesized and cloned into a customized pFastBac vector, which was designed to fuse an N-terminal gp67 signal peptide and C-terminal His ₆-tag to the target protein. To express the RBD protein, a recombinant bacmid DNA was generated from the sequencing-confirmed pFastBac construct using the Bac-to-Bac system (Life Technologies) . Baculovirus was generated by transfecting purified bacmid DNA into Sf9 cells using FuGENE HD (Promega) , and subsequently used to infect suspension cultures of High Five cells (Life Technologies) at a multiplicity of infection (MOI) of 5 to 10. Infected High Five cells were incubated at 28 ℃ with shaking at 110 rpm for 72 h for protein expression. RBD protein that was secreted into the supernatant was harvested and then concentrated with a 10 kDa MW cutoff Centramate cassette (Pall Corporation) . The RBD protein was purified by affinity chromatography using Ni-NTA resin (QIAGEN) , followed by size exclusion chromatography on a HiLoad Superdex 200 pg column (GE Healthcare) , and buffer exchanged into 20 mM Tris-HCl pH 7.4 and 150 mM NaCl. Fabs were expressed in ExpiCHO cells and purified using CaptureSelect CH1-XL resin (ThermoFisher) and followed by size exclusion chromatography. The Fab/RBD complexes were formed by mixing the two components in an equimolar ratio and incubating overnight at 4 ℃ before setting-up crystal trials.

The Fab/RBD complexes were screened for crystallization using 384 conditions of the JCSG Core Suite (QIAGEN) on our robotic CrystalMation system (Rigaku) at The Scripps Research Institute. Crystals appeared in the first week, were harvested during the second week, and then flash-cooled in liquid nitrogen for X-ray diffraction experiments. Diffraction data were collected at cryogenic temperature (100 K) at beamline 23-ID-B of the Advanced Photon Source (APS) at Argonne National Laboratory with a beam wavelength of

and processed with HKL2000. Diffraction data were collected from crystals grown in conditions: 20%PEG 3350, 0.2 M sodium sulfate, pH 6.6 for the S-B8/RBD complex; and 20%isopropanol, 20%PEG 4000, 0.1 M citrate pH 5.6 for the S-E6/RBD complex. The X-ray structures were solved by molecular replacement (MR) using PHASER with MR models for the RBD and Fab from PDB 7JMW. Iterative model building and refinement were carried out in COOT and PHENIX, respectively. Epitope and paratope residues, as well as their interactions, were identified by using PISA program with buried surface area

as the criterion.

Data analysis and statistics

The results were expressed as means ± standard deviation (SD) unless otherwise indicated. Data analysis was performed by one-way analysis of variance (ANOVA) using Origin Pro 2019 statistical software or GraphPad Prism software. Significance was assumed at a P value < 0.05.

Results

Selection of antibodies against SARS-CoV-2 spike RBD

This example constructed and overexpressed the SARS-CoV-2 spike RBD (S-RBD) linked to human Fc (hFc) with a thrombin digestion site. After affinity purification, recombinant SARS-CoV-2 S-RBD was biotinylated, immobilized on streptavidin-coated magnetic beads, and panned against a combinatorial scFv antibody phage library containing 10 ¹¹ members generated from peripheral blood mononuclear cells (PBMC) of 50 healthy donors.

In the first two rounds, a pH 2.2 glycine-HCl solution was used to elute antibody-displaying phagemids bound to S-RBD. To enrich for antibodies that compete with hACE2, a “function-guided enrichment” strategy was used in the third round, where recombinant hACE2-ECD protein was used to elute S-RBD binding phagemids (FIG. 1a) . After three rounds of panning, S-RBD-specific antibodies were enriched (FIG. 1b) and 22 unique antibodies were selected that specifically bound to S-RBD-hFc (FIG. 1c) . Next-generation sequencing of the library revealed that 92%of human heavy-chain IGHV and 89%of the light-chain (IGLV and IGKV) germlines were covered, when aligned to the IMGT (international ImMunoGeneTics) database, enabling screening of antibodies encoded by diverse germlines. Of note, by analyzing ca. 400,000 sequences in the library (209,000 IGHV sequences, 93,000 IGKV sequences, and 87,000 IGLV sequences) , the random selected antibodies displayed low SHM levels as expected for a

library, with over 70%, 61%and 80%sequences of IGHV, IGKV, and IGLV, respectively, having no more than 3 amino acid mutations.

Selected anti-S-RBD antibodies retain binding to full-length spike

The scFv antibodies were then converted to full-length monoclonal antibodies (mAbs) by cloning into a human IgG4e1 (S228P) vector. HEK293F cells were adapted for expression of combinatorial antibodies that were secreted into culture supernatants. Three of the best performing antibodies, S-B8, S-D4, and S-E6, were purified to homogeneity with yields of 8.1, 9.6 and 17 mg/L, respectively, whereas S-RBD-hFc (IgG1) was 58 mg/L. To characterize interactions between the anti-S-RBD antibodies and full-length spike, HEK293T cells were transiently transfected with either SARS-CoV-2 spike-P2A-EGFP or SARS spike-P2A-EGFP. Flow-cytometry (FACS) showed that all three antibodies in full-length-IgG4 format retained their ability to bind full-length SARS-CoV-2 spike with no cross-reactivity with other HCoV spikes, including SARS-CoV (FIG. 2b) , HCoV-229E (FIG. 2c) , HCoV-HKU1 (FIG. 2d) , HCoV-NL63 (FIG. 2e) , HCoV-OC43 (FIG. 2f) , MERS-CoV (FIG. 2g) , or with non-transfected cells (FIG. 2h) .

Relevant sequences of the antibodies are shown in the tables below.

Table 1A. Antibody S-B8

Table 1B. Antibody S-D4

Table 1C. Antibody S-E6

Table 1D. Antibody S-B2

Table 1E. Antibody S-D9

Table 1F. Trimerization Domain

Antibody binding and competition with hACE2-ECD to SARS-CoV-2 S-RBD

To assess neutralization potential of the mAbs, this example investigated their ability to compete with ACE2-ECD for S-RBD binding by ELISA. S-B8, S-D4 and S-E6 all competed strongly with hACE2-ECD in a dose-dependent manner, with IC ₅₀ values of 12.9±1.5 nM, 7.1±0.4 nM, and 12.2±0.7 nM, respectively (FIG. 3a) . Competition between S-E6 and S-B8 or S-D4 for binding to S-RBD was also observed, indicating epitope overlap between the three antibodies. Kinetic parameters of on-rate (k _on) , off-rate (k _off) , and dissociation constant (K _D) for these antibodies were then determined by biolayer interferometry (FIG. 3b-d, and h) .

S-B8, S-D4, and S-E6 exhibited K _D values of 170 pM, 120 pM and 110 pM, respectively, with S-D4 displaying a greatly reduced off-rate (FIG. 3c, h) . Compared to the wild-type spike RBD, binding affinities of the three antibodies to the N501Y mutant S-RBD that is observed in the UK variant B. 1.1.7 were similar with K _D values of 130 pM, 97 pM and 210 pM for S-B8, S-D4, and S-E6, respectively (FIG. 3e-h) . However, the binding abilities of all three antibodies were greatly affected by the E484K mutation in RBD that originally emerged in the South Africa variant B. 1.351 and Brazil variant P. 1, as shown by the dramatically decreased binding signals to E484K+N501Y RBD. For S-B8, binding to K417N+E484K+N501Y RBD is slightly greater than to E484K+N501Y RBD, whereas S-D4 and S-E6 exhibited similar weak binding to K417N+E484K+N501Y RBD.

This example then tested natural mutants of SARS-CoV-2 spike proteins that have been clinically associated with more severe illness and longer hospital stays, as well as the key amino-acid mutations of spike proteins in circulating variants such as B. 1.1.7 and B. 1.351 by FACS assay. Three mutant spike proteins (FIG. 7a-c) , i.e. D215H (mut 1) , S247R (mut 2) , and D614G (mut 3) , found in patients requiring treatment in an intensive care unit (ICU) , spike mutant N439K+D614G found in mink (mut 4, FIG. 7d) , B. 1.1.7 spike (FIG. 7e) , key mutation N501Y+D614G (mut 5, FIG. 7f) that enhances RBD affinity to hACE2 in rapidly spreading variants, E484K+N501Y+D614G (mut 6, FIG. 7g) found in B. 1.351 and P. 1 strains, K417N+E484K+N501Y+D614G (mut7, FIG. 7h) in the B. 1.351 strain, and K417N+N501Y+D614G (mut 8, FIG. 7i) , all displayed positive binding by FACS to S-B8, S-D4, and S-E6. These data indicate the antibody therapeutic potential against these natural SARS-CoV-2 variants isolated from severely affected patients and the circulating variants of concern.

By analyzing the mean fluorescent intensity (MFI) of each antibody to all of the mutants (FIG. 7j) , this example found that binding of S-B8 to both mut 6 (to 18%of mut 3, FIG. 7g) and mut7 (to 14.7%of mut3, FIG. 7h) decreased significantly. Similarly, S-D4 binding to mut 6 and mut 7 detected decreased to 33.2%and 35.6%of mut 3, respectively (FIG. 7j) . However, the MFI of S-E6 to mut 6 and mut 7 was similar to that of mut 3 (FIG. 7j) that differs from the BLI data and suggests some differences are detected on the spike expressed on cells versus that on the S-RBD for this antibody.

Inhibition of cell-cell fusion induced by SARS-CoV-2 spike and hACE2

The S2 subunit of the SARS-CoV-2 spike mediates membrane fusion in hACE2 expressing cells and is essential for virus infection. hACE2 binding to SARS-CoV-2 is stronger than to the SARS-CoV spike (K _D of 4.7 nM and 32 nM, respectively) . To test whether these antibodies could inhibit spike-mediated membrane fusion of cells, this example established a cell-cell fusion assay using Vero cells overexpressing hACE2 as target cells, SARS-CoV-2 spike-P2A-EGFP transient transfected HEK293F cells as effector cells, and SARS-CoV spike-P2A-EGFP cells as a negative control.

Spike-expressing HEK293F cells were mixed with S-B8, S-D4 or S-E6 at 10 nM or 1 nM just before adding to the Vero cells and syncytium formation observed 6 hours later. The SARS-CoV-2 spike induced significant cell-cell fusion as manifested by formation of larger EGFP positive cells, whereas the SARS-CoV spike barely induced syncytium formation (FIG. 4a) . All three antibodies inhibited cell-cell fusion by SARS-CoV-2 at both 10 and 1 nM with 10 nM being significantly more potent (FIG. 4b, c and f) . At 10 nM, S-D4 and S-E6 exhibited over 80%inhibition of cell-cell fusion, which was slightly greater than recombinant S-RBD; S-D4 and S-E6 were also more potent than S-B8 at 1 and 10 nM (FIG. 4d, e) .

Inhibition of SARS-CoV-2 pseudovirus and authentic virus

To test neutralization against SARS-CoV-2 virus, this example first assessed the antibodies in a pseudovirus (PSV) infection assay. Pseudotyped particles were pre-incubated with S-B8 and S-D4 (from 200 nM to 200 fM) and S-E6 (200 nM to 6.3 fM) , followed by infection of HEK293T/hACE2 cells. Luciferase activity resulting from infection was determined at 60 h post transfection. All three antibodies showed potent neutralization against wild-type SARS-CoV-2 PSV infection in a dose-dependent manner that went to completion. The NT ₅₀ values of S-B8, S-D4, and S-E6 in the pseudovirus neutralization were determined to be 2.2±0.2 nM, 0.48±0.03 nM, and 0.025±0.002 nM, respectively (FIG. 5a) in a 1: 1 interaction model with HillSlopes near 1.0 (FIG. 5b) .

This example next tested antibody neutralization of authentic SARS-CoV-2 virus [BetaCoV/Australia/VIC01/2020; GenBank MT007544.1 (Victoria/01/2020) , B VIC01] . Twenty hours after infection, intracellular virus was visualized and quantitated as percent infectivity. All three antibodies were capable of fully blocking infection by authentic virus B VIC01 (FIG. 5c) with NT ₅₀ values for S-B8, S-D4 and S-E6 of 0.88±0.14 nM, 2.04±0.31 nM and 0.15±0.06 nM, respectively (FIG. 5d) .

Due to the emergence of the N501Y mutation in the RBD of the B. 1.1.7 strain (U. K. variant) , this example also tested the neutralization abilities of the three antibodies to SARS-CoV-2 spike N501Y+D614G pseudovirus. All three antibodies appeared to display better neutralizing efficacy than to wild-type PSV. The NT ₅₀ values of S-B8, S-D4, and S-E6 in N501Y+D614G spike pseudovirus neutralization were determined to be 0.53±0.09 nM, 0.042±0.008 nM, and 0.021±0.003 nM, respectively (FIG. 5e) in a 1: 1 interaction model with HillSlopes near 1.0 (FIG. 5f) . Tests on authentic SARS-CoV-2 UK variant (B. 1.1.7) virus showed that all three antibodies maintain neutralization ability (FIG. 5g) , with NT ₅₀ values for S-B8, S-D4 and S-E6 of 2.16±0.28 nM, 1.88±0.20 nM and 1.15±0.14 nM, respectively (FIG. 5h) . Compared to their ability to neutralize wild-type SARS-CoV-2 authentic virus, the ability of S-B8 and S-E6 to neutralize the B. 1.1.7 variant decreased by approximately 2.4 and 7.6 fold, respectively (FIG. 5d, h) . However, S-D4 showed slightly better potency.

As one might expect from the decrease in their binding affinity to E484K and E484K+K417N variants, all three antibodies showed dramatically decreased neutralization ability when tested against the E484K+K417N+N501Y spike mutant pseudovirus. Antibody concentrations of 200 nM were still not sufficient to completely block PSV infection. A similar decrease in neutralizing ability was observed for all three antibodies when tested against the authentic B. 1.351 variant. However, S-D4 did show weak neutralization at high concentrations (>50 nM) against the B. 1.351 variant.

S-B8 and S-E6 bind the RBD and sterically block ACE2 binding

To elucidate the molecular recognition of S-B8 and S-E6 for SARS-CoV-2 S-RBD, x-ray structures of Fab+RBD complexes were determined to 2.25 and

respectively (Table 2) . Fab S-B8 and S-E6 bind the receptor binding site (RBS) with different approach angles (FIG. 6a) and sterically compete with ACE2 for RBD binding, consistent with the competition assay (FIG. 3a) . S-B8 interacts mainly using its heavy chain, which contributes 73%of the buried surface area (BSA, 737 of

) (FIG. 6b) and 12 of 16 polar contacts (Table 3) . S-E6 predominately uses its light chain, which contributes 63%of the BSA (530 of

) and 16 of 19 polar contacts (Table 3) . Light-chain dominant interactions are less common in antibodies.

Table 2. Crystallographic data collection and refinement statistics

^a Numbers in parentheses refer to the highest resolution shell.

^b R _sym = Σ _hkl Σ _i | I _hkl, i -<I _hkl> | /Σ _hkl Σ _i I _hkl, i and R _pim = Σ _hkl (1/ (n-1) ) ^1/2 Σ _i | I _hkl, i -<I _hkl> | /Σ _hkl Σ _i I _hkl, i, where I _hkl, i is the scaled intensity of the i ^th measurement of reflection h, k, l, <I _hkl> is the average intensity for that reflection, and n is the redundancy.

^c CC _1/2 = Pearson correlation coefficient between two random half datasets.

^d R _cryst = Σ _hkl | F _o -F _c | /Σ _hkl | F _o | x 100, where F _o and F _c are the observed and calculated structure factors, respectively.

^e R _free was calculated as for R _cryst, but on a test set comprising 5%of the data excluded from refinement.

Table 3. Hydrogen bonds and salt bridges identified at the antibody-RBD interface *

*Somatically hypermutated residues are highlighted in yellow.

IgBLAST analysis suggests S-B8 is derived from IGHV3-66, a germline that is highly similar to IGHV3-53. However, in S-B8, ₃₂NY ₃₃ in CDRH1 is mutated to ₃₂SH ₃₃ and ₅₃SCGS ₅₆ ( ₅₃TGGT ₅₆ in COVA2-39) in CDRH2 to ₅₃GDGN ₅₆. Intriguingly, CDRH1 and CDRH2, as well as FRH1, of S-B8 still bind to a similar region on SARS-CoV-2 S-RBD to that of binding mode B (FIG. 6b) . The ₃₂SH ₃₃ in S-B8 is part of a type I beta-turn (FIG. 6c) . V _H S32 side chain hydrogen bonds with RBD Q493 and the V _H T28 carbonyl oxygen. The V _H H33 imidazole forms a salt bridge with RBD E484 and a π-π interaction with Y489. The V _H ₅₃GD ₅₄ backbone in CDRH2 also forms two hydrogen bonds with E484, and V _H T28 and L31 make four hydrogen bonds with Q493 and S494 (FIG. 6c, Table 3) . F486 in the S-RBD ridge region is buried in a hydrophobic pocket (V _H W47, V _H I50, V _L Y91, V _L L94, V _L L96) between the heavy and light chains, while ₄₈₅GF ₄₈₆ and ₄₇₆GS ₄₇₇ on the RBD ridge interact with V _H Y52 and V _L Y27d via π-π interactions (FIG. 6d) .

S-E6 interaction with SARS-CoV-2 S-RBD

S-E6 is an IGHV4-31 antibody. Interestingly, SHM introduces a ₃₃NY ₃₄ sequence in a similar position to the ₃₂NY ₃₃ motif in CDRH1 in IGHV3-53/3-66 antibodies that interact with the same RBD site but in a different orientation compared to ₃₂NY ₃₃ of IGHV3-53 binding mode A. Nevertheless, V _H N33 still hydrogen bonds with RBD A475 carbonyl (FIG. 6e) , as does V _H N32 of IGHV3-53 in binding mode A (FIG. 6f) . V _H Y34 and V _H A97 form two hydrogen bonds with N487 of the S-RBD (FIG. 6e) , which differ from Y33 in IGHV3-53 antibodies (FIG. 6f) . F486, along with N487, interact with a hydrophobic pocket formed by V _H Y34, A97, H101 and I102 of S-E6 and also make π-π and cation-π interactions (FIG. 6e) . However, the S-E6 light chain contributes the majority of the buried surface with the RBD. CDRL2 ₅₀FDYR ₅₃ interact with the receptor binding motif (RBM) via multiple polar interactions (8 hydrogen bonds and 3 salt bridges) to E484, F490, L492, Q493 and S494 (FIG. 6g, Table 3) . Moreover, V _L F50 interacts with a nearby hydrophobic patch formed by L455, F456 and Y489 (FIG. 6g) , and V _L S27, G29, S67 and G68 form five hydrogen bonds with R403, N501 and Y505 on the other side of the RBS ridge (FIG. 6h, Table 3) . However, residue 501 is located at the edge of S-E6 epitope site with space that allows for accommodation of the tyrosine mutation at the site (FIG. 6h) . The neutralization of N501Y+D614G pseudovirus suggests that residue 501 is not critical in the binding site of S-E6 since no significant change in the neutralization potency is observed between wild-type and the mutant virus (FIG. 5e, f) .

SHM residues form specific interactions with the RBD

Most RBD-targeting neutralizing antibodies isolated from COVID-19 patients have minimal SHM, although some antibodies expressed from memory B cells several months after infection have increased SHM. The antibodies derived from the combinatorial antibody library in this study are highly mutated. S-B8 and S-E6 contain 13 and 22 SHM residues, respectively, several of which are in the antibody paratope (FIG. 6) , including V _H ₃₁LSH ₃₃, V _H ₅₀IT ₅₁, V _H ₅₃GD ₅₄, V _H N ₅₆, V _H D ₅₈ and V _L L94 in S-B8, and V _H N ₃₃, V _L V ₃₉, V _L ₅₀FDYR ₅₃ and ₆₅TR ₆₆ in S-E6 (FIG. 6c-g) . In summary, several SHM residues appear to be critical for interaction with SARS-CoV-2 RBD. The interaction with SHM residues appear to be mainly with the heavy chain in S-B8 or the light chain in S-E6.

Antibody autoreactivity

To investigate the origin of the three antibodies, a HEp-2 autoreactivity assay was performed. Neither S-D4 nor S-E6 showed a positive signal in the assay, suggesting that they are not derived from an auto-immune response, whereas S-B8 displayed weak to moderate autoreactivity. This example further generated an S-B8 putative germline antibody by mutating back all of the SHMs in the S-B8 heavy chain to the

IGHV3-66 sequence. The mutated antibody showed greater autoreactivity than S-B8 and no S-RBD binding up to 12.5 nM.

Antibody-dependent enhancement (ADE) activity assessment of the three antibodies

ADE occurs through two distinct mechanisms during viral infections, one via enhanced infection mediated by FcγRIIa expressed on monocytes and macrophages, and the other via enhanced immune activation caused by excessive Fc-mediated effector functions and immune complex formation. In these antibody constructs, this example adopted an engineered IgG4e1 (S228P) format to reduce the affinity to Fcγ receptors (FcγRs) . The ADE effects of the three antibodies were assessed in three cell lines expressing different levels of FcγR. The qPCR results revealed high level FcγIA and IIA, high level FcγIIA, and low level FcγIIB and IIIA for THP-1, K562, Raji, respectively. Treatment of Raji, K562 and THP-1 cells with a mixture of SARS-CoV-2 pseudovirus with different concentrations of S-B8, S-D4 and S-E6 showed no apparent ADE effects.

The structural studies on S-E6 and S-B8 revealed several striking features of these combinatorial antibodies. The primary immune response to viral infection is followed by a secondary response that generates functionally better antibodies, where the binding energy can be refined by somatic hypermutation. The secondary immune response is for later encounter of the same antigen, and is the basis of vaccination. In cases of pandemics, such as SARS-CoV-2, avian influenza or Ebola virus, if the infection is not dealt with by the immune system in the first few days, the patient has a high probability of dying, and as a consequence, the immune system will not have enough time to refine the immune response. Consistently, neutralizing antibodies isolated from SARS-CoV-2 convalescent patients contain only a few amino-acid mutations that may be a result of weak B cell stimulation due to rapid viral clearance. Neutralizing antibodies isolated from convalescent patients shortly after infection may then possibly not be fully refined (matured) . However, a recent study has shown higher levels of SHM several months after infection in some COVID-19 patients.

In comparison, S-B8 and S-E6 exhibited higher levels of SHM, many of which are involved in specific interactions with SARS-CoV-2 RBD (S-RBD) . Nine of 13 SHM residues in S-B8 and eight of 22 in S-E6 are located in the antibody-antigen interface. While some of these SHM residues only use their peptide backbone, others rely on specific side chains for S-RBD binding (FIG. 6c, d, e, and g) . Interestingly, SHM in CDRH1 of S-E6 generates a ₃₃NY ₃₄ sequence that is similar to the ₃₂NY ₃₃ motif in IGHV3-53/3-66 antibodies, which are the most frequent germlines used in targeting the S-RBD, indicative that the combinatorial antibody library and the maturation process can yield alternate antibody solutions. However, it is unclear how these SHM residues could have been raised specifically to the SARS-CoV-2 S-RBD, since the library was generated before the SARS-CoV-2 pandemic.

Of note, the heavy and light chains are randomly paired during the selection experiment. However, S-E6 is a light-chain dominant antibody and most of the SHM residues in the heavy chain are not involved in interaction with SARS-CoV-2 RBD. Thus, these findings raise fascinating questions about the original antigen (s) that elicited S-B8 and S-E6, at least to the heavy or light chains that dominate binding to SARS-CoV-2 RBD.

Example 2. Testing with new mutants

The following assays were conducted to test the antibodies for newer mutants.

Affinity to N501Y RBD determination by Biolayer Interferometry (BLI)

Binding affinities of S-D4 with SARS-CoV-2 N501Y S-RBD were performed by BLI on an Octet RED96 (Molecular Devices LLC, San Jose, CA, USA) using AR2G biosensors. N501Y S-RBD diluted in a PBS solution containing 0.02%Tween-20 and 0.05%BSA (PBST-B) (10 μg/mL) was loaded to the AR2G biosensor by amine coupling. The AR2G-N501Y S-RBD sensors were dipped into a PBST-B for 60 sec to establish a baseline, and then incubated with 2-fold serial diluted S-D4 antibody solutions to record the progressive curves of association. Finally, sensors were incubated in a PBST-B buffer to record the progressive curves of dissociation. For S-B8 and S-E6 detections, N501Y S-RBD was first biotinylated before loading onto a streptavidin (SA) sensor, the remaining procedure was same to that of S-D4. Results were analyzed by ForteBio Data Analysis software.

Binding kinetics with N501Y S-RBD were measured by biolayer interferometry (BLI) . Biotinylated N501Y S-RBD was loaded onto the SA biosensor for detection of binding kinetics with S-B8 (FIG. 8A) and S-E6 (FIG. 8C) , while N501Y S-RBD amine coupled onto AR2G sensor was utilized for S-D4 (FIG. 8B) , with detection on an Octet. All curves were fitted by a 1: 1 binding model using the Data Analysis software (Forte Bio) . The association-rate (k _on) , dissociation-rate (k _off) and dissociation constant (K _D) of the three competitive antibodies to N501Y S-RBD are shown in FIG. 8D. All three antibodies exhibited potent binding affinities to N501Y S-RBD.

Neutralization of SARS-CoV-2 N501Y Pseudovirus

SARS-CoV-2 N501Y+D614G coronavirus pseudovirus (PSV) was first prepared, HEK293T/hACE2 cells were then seeded into 96-well white bottom plates at a density of 1×10 ⁴/well, and cultivated overnight. The PSV was pre-incubated with an equal volume of different concentrations of selected antibodies (dilution factor: 3.16, from 200 nM to 20 fM for S-D4 and S-E6, 200 nM to 0.6 pM for S-B8) in DMEM at 37 ℃ for 30 min. DMEM with or without PSV in the absence of antibodies were set as controls. After incubation, the PSV mixture was transferred to the culture plates containing HEK293T/hACE2 cells. The DMEM media containing PSV and antibodies were replaced with fresh media after 16 h treatment, cells were incubated for an additional 48 h. PSV infection efficacy was evaluated by luciferase activity using Bright-Lumi ^TM Firefly Luciferase Reporter Gene Assay Kit (Beyotime, RG015M) . Fifty microliter of luciferase substrate was added to each well, and the relative luminescence unit (RLU) values were measured on an Envision plate reader (PerkinElmer, Ensight) .

Neutralization ability of the three hACE2 competitive antibodies to SARS-CoV-2 N501Y+D614G mutant pseudovirus was tested and fitted, NT ₅₀ and HillSlope are shown in the lower panel of FIG. 9. All three antibodies showed superior neutralization abilities.

Authentic B. 1.1.7 SARS-CoV-2 virus neutralization assay

The study was performed in the CL3 Facility of the University of Oxford operating under license from the HSE, on the basis of an agreed Code of Practice, Risk Assessments (under ACDP) and Standard Operating Procedures. In brief, this rapid, high-throughput assay determines the concentration of antibody that produces a 50%reduction in infectious focus-forming units of B. 1.1.7 authentic SARS-CoV-2 strains in Vero cells, as follows. Quadruplicate, 0.5log ₁₀ serial dilutions of antibody (9 steps from 316 nM to 31 pM) were pre-incubated with a fixed dose of SARS-CoV-2 (B. 1.1.7 variant) before incubation with Vero cells. A 1.5%carboxymethyl cellulose-containing overlay was used to prevent satellite focus formation. Twenty hours post-infection, the monolayers were fixed with 4%paraformaldehyde, permeabilized with 2%Triton X-100 and stained for N antigen using mAb EY 2A. After development with a peroxidase-conjugated antibody and True Blue peroxidase substrate, infectious foci were enumerated by ELISPOT reader. Data were analyzed using four-parameter logistic regression (Hill equation) in GraphPad Prism 8.3.

Neutralization ability of the three hACE2 competitive antibodies to authentic virus of B. 1.1.7 strain was tested and fitted, NT ₅₀ and HillSlope are shown in the lower panel of FIG. 10. All three antibodies showed superior neutralization abilities.

Neutralization of SARS-CoV-2 K417N+E484K+N501Y Pseudovirus by S-D4 antibody

SARS-CoV-2 K417N+E484K+N501Y+D614G coronavirus pseudovirus (PSV) was first prepared, HEK293T/hACE2 cells were then seeded into 96-well white bottom plates at a density of 1×10 ⁴/well, and cultivated overnight. The PSV was pre-incubated with an equal volume of different concentrations of S-D4 antibody (dilution factor: 3.16, from 200 nM to 0.6 pM) in DMEM at 37 ℃ for 30 min. DMEM with or without PSV in the absence of antibody were set as controls. After incubation, the PSV mixture was transferred to the culture plates containing HEK293T/hACE2 cells. The DMEM media containing PSV and antibody were replaced with fresh media after 16 h treatment, cells were incubated for an additional 48 h. PSV infection efficacy was evaluated by luciferase activity using Bright-Lumi ^TM Firefly Luciferase Reporter Gene Assay Kit (Beyotime, RG015M) . Fifty microliter of luciferase substrate was added to each well, and the relative luminescence unit (RLU) values were measured on an Envision plate reader (PerkinElmer, Ensight) .

Neutralization ability of S-D4 antibody to SARS-CoV-2 K417N+E484K+N501Y+D614G mutant pseudovirus was tested and fitted, NT ₅₀ and HillSlope are shown in the lower panel of FIG. 11. S-D4 showed superior neutralization abilities.

Authentic B. 1.351 SARS-CoV-2 virus neutralization assay

The study was performed in the CL3 Facility of the University of Oxford operating under license from the HSE, on the basis of an agreed Code of Practice, Risk Assessments (under ACDP) and Standard Operating Procedures. In brief, this rapid, high-throughput assay determines the concentration of antibody that produces a 50%reduction in infectious focus-forming units of B. 1.351 authentic SARS-CoV-2 strains in Vero cells, as follows. Quadruplicate, 0.5log ₁₀ serial dilutions of antibody (9 steps from 316 nM to 31 pM) were pre-incubated with a fixed dose of SARS-CoV-2 (B. 1.351 variant) before incubation with Vero cells. A 1.5%carboxymethyl cellulose-containing overlay was used to prevent satellite focus formation. Twenty hours post-infection, the monolayers were fixed with 4%paraformaldehyde, permeabilized with 2%Triton X-100 and stained for N antigen using mAb EY 2A. After development with a peroxidase-conjugated antibody and True Blue peroxidase substrate, infectious foci were enumerated by ELISPOT reader. Data were analyzed using four-parameter logistic regression (Hill equation) in GraphPad Prism 8.3.

Neutralization ability of S-D4 antibody to authentic virus of B. 1.351 strain was tested and fitted. The NT ₅₀ and HillSlope are shown in the lower panel FIG. 12 demonstrating the excellent neutralization ability of S-D4.

Example 3. Trimeric format of SARS-CoV-2 RBD-targeting antibodies retrieve ultrapotent neutralization ability against immune escape mutants

This example developed a trimeric form of anti-spike RBD antibodies. As compared to the monomeric and dimeric forms, the trimeric form significantly increased the binding avidity to spike RBD, and surprisingly, it neutralized the immune-escape strain B. 1.351 (beta) potently with an NT ₅₀ of 36 pM and completely neutralized the B. 1.617.2 (delta) strain with an NT ₅₀ of 7 pM. By applying this trimeric format engineering method to other RMB-targeting antibodies, they also achieved significant neutralizing effects. The NT ₅₀ values improved up to 30-fold.

Materials and Methods

Cell Culture

The FreeStyle ^TM 293F (HEK293F, ThermoFisher Scientific) cell line was cultured in Freestyle 293 expression media (ThermoFisher Scientific) . HEK293T/hACE2 stable cell line was maintained in high glucose Dulbecco Modified Eagle Medium (DMEM) (Gibco) containing 10% (v/v) fetal bovine serum (FBS) (Gemini) , with addition of 10 μg/mL puromycin.

Production and purification of recombinant multivalent antibodies

For multivalent antibodies engineering, taken S-E6 as example, we constructed trimeric antibodies of S-E6 by fusing T4 bacteriophage fibritin trimerization motif (T4F) , GCN4 trimeric helices, or TSE tag to the C-terminus of the S-E6 Fab heavy chain, respectively. For fusing trimerization tags with antibody, different linkers were tried, including variable length of rigid linkers (EAAAK, SEQ ID NO: 105) .

For monomer and trimer antibodies, the encoding sequence of Fab domain of S-E6 heavy chain was cloned into expression vector with 6×His tag on C-terminal. HEK293F overexpressing each recombinant protein was cultured for 4 days. Recombinant proteins were harvested from the supernatant. After centrifugation, HisTrap Excel HP column (GE Healthcare) and HiTrap Protein A HP column (GE Healthcare) were used to purify His-tag (monomer and trimer antibodies) and hFc-tag proteins (dimer antibodies) , respectively. The procedure followed manufacturer's instruction. Briefly, for His-tag proteins, after loading of the supernatant to a HisTrap column, the column was washed with a binding buffer (500 mM NaCl, 20 mM sodium phosphate, 20 mM imidazole, pH 7.4) , and eluted with 500 mM imidazole in PBS buffer. For hFc-tag antibodies, after elution, the recombinant antibodies were buffer exchanged and concentrated in PBS buffer (150 mM NaCl, 20 mM sodium phosphate, pH 7.2) using Ultracel membrane with molecular weight cutoff of 30 kDa (Merck Millipore) and stored at -80 ℃ until use.

Size-exclusion-high-performance liquid chromatography (SEC-HPLC)

Twenty μL of 0.5 μg/μL purified antibodies were applied to an Agilent Bio SEC-5, 500A HPLC system. The mobile phase used was PBS buffer (pH 7.2) running at a flow rate of 0.35 mL/min. Absorbance was analyzed and integrated by retention time and area under the curve (AUC) to determine the percentage of aggregation, monomer and degradants compositions. Signal was monitored by A280. Data processing was performed using the GraphPad Prism software.

Affinity determination by Biolayer Interferometry (BLI)

The binding affinities of multivalent antibodies with SARS-CoV-2 wild-type or mutant S-RBD were performed by BLI on an Octet RED96 system (Molecular Devices LLC, San Jose, CA, USA) at room temperature. In brief, biotin conjugated S-RBD or mutant S-RBD diluted in a PBS solution containing 0.02%Tween-20 and 0.05%BSA (PBST-B) (10 μg/mL) was captured on a SA-coated biosensor (ForteBio) . The baseline was recorded for 60 sec in a running buffer (PBST-B) , and then the sensors recorded the progressive curves of association stage for 300 sec in wells containing 2-fold serial diluted antibody solutions. In the dissociation step, sensors were incubated in a PBST-B buffer for 600 sec to record the progressive curves of dissociation. The average k _on, k _off and k _D values were calculated from all the binding curves based on the fitting with a 1: 1 Langmuir binding model.

Production of SARS-CoV-2 wild-type and mutant pseudovirus

Pseudovirions were produced by co-transfected HEK293T cells with the WT or mutant spikeΔ19 plasmid and NL4-3 mcherry Luciferase plasmid (addgene) by using Lipofectamine 3000 reagent (Invitrogen) . The virus-containing supernatants were harvested 48 h post-transfection, filtered through 0.45 μm filter, and mixed with Lenti-X Concentrator (Takara) overnight at 4 ℃, then centrifuged at 1500 g for 45 min to remove supernatants. The pellets were collected and stored at -80 ℃ until use.

Pseudovirus-based neutralization assay

96-well white bottom plates were seeded with 1×10 ⁴ HEK293T/hACE2 cells per well. The following day, antibodies were subjected to a 3.16-fold dilution series. An equal volume of different concentrations of antibodies was mixed with PSVs of wild type or mutant. After incubation for 30 min at 37 ℃, the mixture was added to 96-well plates. After 16 hours cultivation at 37 ℃, the inoculum was replaced with fresh media, cells were incubated for additional 48 h. PSV infection efficacy was evaluated by luciferase activity using Bright-Lumi ^TM Firefly Luciferase Reporter Gene Assay Kit (Beyotime, RG015M) . NT ₅₀ values were calculated using the “log (inhibitor) vs. response” equation in GraphPad Prism software.

Results

Trimeric S-E6 shows enhanced neutralization ability than the full IgG format

It has been demonstrated that the binding and neutralization of RBS targeting antibodies are significantly reduced against the E484K containing spike mutations. Our computational modeling demonstrates that a trimeric antibody could simultaneously bind to the three monomers in the spike, a trimeric protein (FIG. 1A) . The trimeric antibody, therefore, was contemplated to have higher binding avidity and better neutralizing potency to the immune escape mutants.

Here, we used S-E6 as an example to optimize the system. In order to prove the concept, S-E6 was first engineered into a trimeric format by fusing the CH1 domain with a trimerization tag T4F with a (G ₄S) ₃ linker (FIG. 13G) . Both SDS-PAGE analysis and SEC-HPLC results indicate the trimeric S-E6 was well assembled and showed homogenous distribution (FIG. 13B, C) . The results showed that the trimeric S-E6 (NT ₅₀=0.0016±0.001 nM) was more potent than the dimer (NT ₅₀=0.036±0.004 nM) and monomer (NT ₅₀=1.3±0.4 nM) S-E6 in neutralizing SARS-CoV-2 wild-type pseudovirus (WT-PSV) (FIG. 13D) . These results demonstrate that the trimer-Fab format antibodies do have better neutralizing potency than the conventional IgG format.

To optimize the trimeric format, we first compared the trimeric efficiency of three trimeric domains, including T4 bacteriophage fibritin trimerization motif (T4F) , GCN4 trimeric leucine zipper motif (GCN4) , and human collagen XVIII derived homotrimerization domain (TIE) . All trimeric proteins yielded pure proteins on SDS-PAGE, but GCN4-tagged antibodies come up with two peaks in the HPLC-SEC separation, indicating non-homogeneous composition.

Table 4. Related Sequences

We further analyzed the neutralization capabilities of these trimmers. T4F-fused antibody showed the best neutralizing potency (T4F, NT ₅₀=0.0016±0.001 nM; TSE, NT ₅₀=0.0066±0.001 nM; GCN4, NT ₅₀=0.0073±0.001 nM) (FIG. 13E) . To compare the effects of different linkers, the flexible GS linker was replaced with different lengths of rigid linker (1-5 repeats of EAAAK; SEQ ID NO: 105) . The 3-and 5-repeated rigid linker mediated trimers (3rigid, NT ₅₀=0.038±0.002 nM; 5 rigid, NT ₅₀=0.20±0.035 nM) shows similar or decreased neutralizing ability, while the Trimer-1rigid S-E6 improves moderately (NT ₅₀=0.026±0.002 nM) (FIG. 13F) . Thus, we fused CH1 domain to T4F with 3 (G ₄S) linker in the following experiment (Table 5) .

Table 5. Neutralizing parameters of nAbs

Trimer S-E6 revived neutralizing potency to B. 1.351 (Beta Strain) PSV with enhanced binding avidity to S-RBD

Binding is the basic physical force for neutralization, so we determined the binding parameters of antibody-RBD. Kinetic parameters of on-rate (k _on) , off-rate (k _off) , and dissociation constant (K _D) for dimer/trimer S-E6 to wild-type S-RBD and B. 1.351 S-RBD were determined by biolayer interferometry (FIG. 14A, B) . Dimer and trimer S-E6 exhibited K _D values of 83.6 pM and 10.1 pM to WT S-RBD, respectively, with trimer S-E6 showing increased association rate (FIG 14E) . Similarly, the binding affinities of trimer S-E6 to the B. 1.351 S-RBD was also higher than dimer S-E6, with K _D values of 1.58 nM and 0.58 nM to WT S-RBD and B. 1.351 S-RBD, respectively (FIG. 14C, D and E) . The elevation of binding avidity suggests that the trimer S-E6 was more active to neutralize the B. 1.351 variant.

To test whether the trimer format of nAbs can restore the neutralizing potency to the immune escape variants, we first tested the neutralization of trimer S-E6 to the B. 1.351 PSV. The experimental design was similar to that of WT SARS-CoV-2 PSV in FIG. 13, different valents/trimerization tags/linkers encoded S-E6 were tested for neutralizing B. 1.351 PSV. As shown in FIG. 15, trimer S-E6 indeed shows markedly enhanced potency than the dimer and monomer ones, while the best modification is also the construct with 3 (G4S) linked T4F tag (Table 5) . Trimer S-E6 showed a NT ₅₀ of 0.036±0.022 nM, comparable to the potency of native form of S-E6 (IgG) on WT SARS-CoV-2 PSV (NT ₅₀=0.036±0.004 nM, FIG. 13D, Table 5) . It suggests that engineering IgG format nAb into trimer-Fab format, it restored the neutralizing potency to the immune escape variant B. 1.351.

Trimeric class 1 nAbs restored neutralizing potency to B. 1.351 (Beta Strain) PSV

Motivated by the success in restoring the S-E6 neutralization against B. 1.351, we further investigated whether the trimer reformatting is applicable to other S-RBD targeting nAbs. Class 1 nAbs are ACE2 competing ones that bind in RBM of spike, and some of them show decreased potency to B. 1.351. We used T4F tag fusing with triple (G4S) linker as the trimer format, which is the most potent trimer format confirmed by S-E6. We took several class 1 nAbs belonging to the top enriched IGHV germlines for trimer modification, including S-B2 (IGHV3-30) , S-D9 (IGHV5-51) , S-D4 (IGHV3-33) and S-B8 (IGHV3-66) (FIG. 16A-D) . In addition, a class 2 nAb, S309, was also engineered (FIG. 16E) . A NTD targeting Ab CoV2-2489 was selected as an example of class 3 nAb (FIG. 16F) .

Trimeric form of class 1 antibodies exhibited increased neutralizing potency than their corresponding IgG4 format, such as S-B8, S-D4 improved NT ₅₀ by 3.5-fold and 6.5-fold, respectively. As conserved site targeting antibodies, class 2 antibody S309-IgG4 and class 3 antibody CoV2-2489-IgG4 could neutralize the B. 1.351 PSV, with NT ₅₀ of 0.36±0.08 nM and 0.35±0.04 nM, respectively. All NT ₅₀ values fold increases upon trimerization, antibodies from Class 1 enhanced up to 30.5-fold (FIG. 16G) . For S-B2 antibody, the dimer format did not show any neutralization against B. 1.351 PSV even at 100 nM, but the trimer format inhibited the PSV infection with a NT ₅₀ of 0.1 nM for B. 1.351 PSV.

Interestingly, although similar NT ₅₀ were observed between dimer and trimer, the trimerizing engineering to antibodies from class 2 and class 3 impairs the neutralizing effects, showing inhibition rates at maximal concentration (200 nM) decreased. The inhibition rates of S309 decreased from 76%to 55%at 200 nM, the inhibition capability reduced by 28%, while CoV2-2489 inhibition decreased from 95%to 64%at 200 nM, the inhibition capability reduced by 32% (Table 6 and FIG. 16G) . But the inhibition rate of trimeric class 1 antibodies at 200 nM are improved in comparison to the dimer antibodies.

Table 6. Neutralizing parameters of nAbs

NT50 (B. 1.351) (nM)

NT50 (B. 1.617.2) (nM)

S‐E6 dimer	0.32±0.22	0.065±0.003
S‐E6 trimer	0.036±0.02	0.007±0.003
S‐B8 dimer	1.2±0.2	0.37±0.06
S‐B8 trimer	0.35±0.07	0.15±0.01
S‐D4 dimer	0.66±0.13	0.24±0.03
S‐D4 trimer	0.10±0.02	0.14±0.02
S‐B2 dimer	>100	0.23±0.02
S‐B2 trimer	0.10±0.02	0.067±0.004
S‐D9 dimer	～1.5±3.6	0.24±0.05
S‐D9 trimer	0.05±0.01	0.067±0.014
S309 dimer	0.36±0.08	N. A
S309 trimer	0.28±0.13	N. A
CoV2‐2489 dimer	0.35±0.04	N. A
CoV2‐2489 trimer	0.20±0.07	N. A

Trimeric class 1 nAbs restored neutralizing potency to B. 1.617.2 (Delta Strain) PSV

Due to the high transmission efficiency and more breakthrough cases of B. 1.617.2, we tested the potency of trimer antibodies on its PSV. Taken S-E6 as example, the neutralizing ability of trimer S-E6 is significantly higher than that of the IgG form, and the NT ₅₀ value is only 4.6-fold higher than that to wild-type SARS-CoV-2 PSV (FIG. 17A) . Neutralizing enhancement of trimer format was also observed on other class 1 antibodies (FIG. 17B-E) . For S-B2, S-B8, S-D4, S-E6 and S-D9, the increased neutralizing abilities are 2.5-fold, 1.5-fold, 8.5-fold, 3.5-fold and 3.5-fold, respectively (FIG. 17F and Table 6) . Meanwhile, the neutralization to B. 1.671.2 of four antibodies are near complete.

The variants spread faster and caused higher risk of severe disease. For example, the B. 1.1.7 (alpha) strain presented a 48%higher risk of severe disease than wild-type virus, the B. 1.351 (beta) strain triggered a 24%higher risk of severe disease, 49%higher risk of critical disease, and 57%higher risk of COVID-19 death. The B. 1.351 (beta) strain can escape the monoclonal antibodies targeting the viral spike glycoprotein (S) . The B. 1.617 was fully resistant to neutralization by bamlanivimab and partially resistant against neutralization by antibodies elicited by infection and vaccination with the Comirnaty/BNT162b2 vaccine. Taken together, the virus has developed a series of variants to escape the neutralization by nAb. Therefore, developing variant-resistant antibody is vital for the variants infection treatment and prevention. In this example, we proposed a strategy of engineering the nAbs into the trimeric format. We found an efficient trimeric format by fusing T4F with 3 (G4S) to the C-terminus of CH1 domain. The trimeric antibodies significantly increased the avidity to the spike protein. Finally, the trimeric formats successfully restored the neutralization capabilities to the immune escape strains, including B. 1.351 and B. 1.617.2.

Example 4. Conformation selection of SE6 protein

Linoleic acid can lock the SARS-CoV-2 spike protein in the down conformation. We investigated the conformation preference using the Octet.

Purified Spike trimer protein was immobilized on the SA sensor. Linoleic acid (Cat. L8134, Sigma) was dissolved in the running buffer. The baseline was recorded for 60 sec in a running buffer (PBST-B) , and then the sensors recorded the progressive curves of association stage for 300 sec in wells containing 2-fold serial diluted antibody solutions. In the dissociation step, sensors were incubated in a PBST-B buffer for 600 sec to record the progressive curves of dissociation. The sensor graph with or without the Linoleic acid was compared to get the binding preference. The average k _on, k _off and k _D values were calculated from all the binding curves based on the fitting with a 1: 1 Langmuir binding model.

As shown in FIG. 18, trimeric S-E6 and S-B8 preferentially bound to the up conformation. Binding kinetics of Trimer S-E6 (FIG. 18A) or trimer S-B8 (FIG. 18B) with wild-type (WT) S-RBD were measured by biolayer interferometry (BLI) with Linoleic acid (Red) or without Linoleic acid (blue) . The supplement of Linoleic acid significantly decreased the binding of antibody to the Spike protein, indicating the antibody prefer to bind the up conformation.

***

The present disclosure is not to be limited in scope by the specific embodiments described which are intended as single illustrations of individual aspects of the disclosure, and any compositions or methods which are functionally equivalent are within the scope of this disclosure. It will be apparent to those skilled in the art that various modifications and variations can be made in the methods and compositions of the present disclosure without departing from the spirit or scope of the disclosure. Thus, it is intended that the present disclosure cover the modifications and variations of this disclosure provided they come within the scope of the appended claims and their equivalents.

All publications and patent applications mentioned in this specification are herein incorporated by reference to the same extent as if each individual publication or patent application was specifically and individually indicated to be incorporated by reference.

Claims

A trimeric antibody comprising three fusion polypeptides, each fusion polypeptide comprising an antigen binding domain fused to a trimerization domain, wherein the antigen binding domain has binding specificity to the receptor binding motif (RBM) of the receptor binding domain (RBD) of the SARS-Cov-2 spike protein.
The trimeric antibody of claim 1, wherein the trimerization domain is capable of mediating stable association of the trimeric antigen binding molecule.
The trimeric antibody of claim 1, wherein the trimerization domain comprises an amino acid sequence selected from the group consisting of SEQ ID NO: 103, 107 and 111, preferably SEQ ID NO: 103.
The trimeric antibody of any one of claims 1-3, further comprising a peptide linker between the antigen binding domain and the trimerization domain.
The trimeric antibody of claim 4, wherein the peptide linker is a flexible linker, preferably comprising the amino acid sequence of SEQ ID NO: 122 or 123.
The trimeric antibody of claim 4 or 5, wherein the peptide linker is from 5 to 50 amino acid residues in length, preferably from 5 to 20 amino acid residues in length.
The trimeric antibody of any one of claims 1-6, wherein the antigen binding domain binds to at least one of amino acid residues selected from the group consisting of 438, 447-456, 489-495, and 507-508 of SEQ ID NO: 126.
The trimeric antibody of any one of claims 1-7, wherein the antigen binding domain comprises a heavy chain variable region (VH) comprising heavy chain complementarity determining regions CDRH1, CDRH2, and CDRH3, and a light chain variable region (VL) comprising light chain complementarity determining regions CDRL1, CDRL2, and CDRL3, wherein the CDRH1, CDRH2, CDRH3, CDRL1, CDRL2, and CDRL3, respectively, comprise the amino acid sequences of (a) SEQ ID NO: 5-10; (b) SEQ ID NO: 27-32; (c) SEQ
The trimeric antibody of any one of claims 1-7, wherein the antigen binding domain is a Fab fragment, a scFv fragment, or a nanobody.
One or more polynucleotides encoding the trimeric antibody of any one of claims 1-9.
A cell comprising the one or more polynucleotides of claim 10.
A composition comprising the trimeric antibody of any one of claims 1-9 and a pharmaceutically acceptable carrier.
A method for treating or preventing a SARS-CoV-2 viral infection in a subject, comprising administering to the subject an effective amount of the trimeric antibody of any one of claims 1-9.
The method of claim 13, wherein the subject suffers from a COVID-19 symptom.