key: cord-0861708-ue6ti33w
authors: Li, Dapeng; Edwards, Robert J; Manne, Kartik; Martinez, David R.; Schäfer, Alexandra; Alam, S. Munir; Wiehe, Kevin; Lu, Xiaozhi; Parks, Robert; Sutherland, Laura L.; Oguin, Thomas H.; McDanal, Charlene; Perez, Lautaro G.; Mansouri, Katayoun; Gobeil, Sophie M. C.; Janowska, Katarzyna; Stalls, Victoria; Kopp, Megan; Cai, Fangping; Lee, Esther; Foulger, Andrew; Hernandez, Giovanna E.; Sanzone, Aja; Tilahun, Kedamawit; Jiang, Chuancang; Tse, Longping V.; Bock, Kevin W.; Minai, Mahnaz; Nagata, Bianca M.; Cronin, Kenneth; Gee-Lai, Victoria; Deyton, Margaret; Barr, Maggie; Holle, Tarra Von; Macintyre, Andrew N.; Stover, Erica; Feldman, Jared; Hauser, Blake M.; Caradonna, Timothy M.; Scobey, Trevor D.; Rountree, Wes; Wang, Yunfei; Moody, M. Anthony; Cain, Derek W.; DeMarco, C. Todd; Denny, ThomasN.; Woods, Christopher W.; Petzold, Elizabeth W.; Schmidt, Aaron G.; Teng, I-Ting; Zhou, Tongqing; Kwong, Peter D.; Mascola, John R.; Graham, Barney S.; Moore, Ian N.; Seder, Robert; Andersen, Hanne; Lewis, Mark G.; Montefiori, David C.; Sempowski, Gregory D.; Baric, Ralph S.; Acharya, Priyamvada; Haynes, Barton F.; Saunders, Kevin O.
title: The functions of SARS-CoV-2 neutralizing and infection-enhancing antibodies in vitro and in mice and nonhuman primates
date: 2021-02-18
journal: bioRxiv
DOI: 10.1101/2020.12.31.424729
sha: db3dbfeec25904ea17b4ec1b572c2670e952e6cc
doc_id: 861708
cord_uid: ue6ti33w

SARS-CoV-2 neutralizing antibodies (NAbs) protect against COVID-19. A concern regarding SARS-CoV-2 antibodies is whether they mediate disease enhancement. Here, we isolated NAbs against the receptor-binding domain (RBD) and the N-terminal domain (NTD) of SARS-CoV-2 spike from individuals with acute or convalescent SARS-CoV-2 or a history of SARS-CoV-1 infection. Cryo-electron microscopy of RBD and NTD antibodies demonstrated function-specific modes of binding. Select RBD NAbs also demonstrated Fc receptor-γ (FcγR)-mediated enhancement of virus infection in vitro, while five non-neutralizing NTD antibodies mediated FcγR-independent in vitro infection enhancement. However, both types of infection-enhancing antibodies protected from SARS-CoV-2 replication in monkeys and mice. Nonetheless, three of 31 monkeys infused with enhancing antibodies had higher lung inflammation scores compared to controls. One monkey had alveolar edema and elevated bronchoalveolar lavage inflammatory cytokines. Thus, while in vitro antibody-enhanced infection does not necessarily herald enhanced infection in vivo, increased lung inflammation can occur in SARS-CoV-2 antibody-infused macaques.

The severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has caused a global pandemic with over 58 96 infected with SARS-CoV-1 ~17 years prior to sample collection (Figures 1A-B, S1 and S2) . We isolated and 102 characterized 1,737 antibodies that bound to SARS-CoV-2 S or nucleocapsid (NP) proteins ( Figure 1C ; Table S1 ).

We selected 187 antibodies for further characterization, and examined neutralization of SARS-CoV-2 pseudovirus 104 and replication-competent SARS-CoV-2. Forty-four of 81 recombinant RBD antibodies exhibited neutralization 105 when assayed in 293T/ACE2 cell pseudovirus, SARS-CoV-2 microneutralization, or SARS-CoV-2 plaque 106 reduction assays (Figures S3A-F; Tables S2-S3 ).

Ten of forty-one NTD antibodies neutralized in the 293T/ACE2 pseudovirus and plaque reduction assays, 108 with the most potent antibody neutralizing pseudovirus with an IC50 of 39 ng/mL (Figures S3G-I; Tables S4-S5) .

In addition, 5 non-neutralizing NTD antibodies enhanced SARS-CoV-2 pseudovirus infection in 293T/ACE2 cells 110 by 56% to 148% (Figure 1D) . Infection of replication-competent SARS-CoV-2 nano-luciferase virus (Hou et al., 111 2020) also increased in the presence of each of the 5 non-neutralizing NTD antibodies ( Figure 1E) . Analysis 

We compared the phenotypes and binding modes to S protein for five infection-enhancing RBD antibodies 134 and three RBD antibodies that lacked infection enhancement to elucidate differences between these two types of 135 antibodies. The selected RBD antibodies neutralized SARS-CoV-2 pseudovirus and/or replication-competent virus 136 in ACE2-expressing cells (Figures 2A and S5) , despite five of these antibodies mediating infection enhancement 137 in ACE2-negative, FcR-positive TZM-bl cells (Figures 1F-L, 2A, and S5 (Figures 2A and S10) . The remaining two RBD 151 antibodies, DH1045 and DH1047, cross-reacted with both SARS-CoV-1 and SARS-CoV-2 S (Figures 2A, S4, 152 S31). DH1047 also reacted with bat and pangolin CoV spike proteins (Figures 2A, S4, S31 Figure 2E) . Thus, S protein antibody epitopes and binding modes were associated with FcR-167 independent, infection-enhancing activity of NTD antibodies. The five neutralizing antibodies bound the same 168 epitope as antibody 4A8 (Wrapp et al., 2020a) , with three of the five having the same angle of approach as 4A8 169 ( Figure S11) . Interestingly, the NTD antibodies with the same angle of approach as 4A8, were also genetically 170 similar to 4A8, being derived from the same VH1-24 gene segment (Table S8) 

NTD antibodies also segregated into two clusters based on their ability to block each other ( Figure 3A ).

Neutralizing NTD antibodies blocked each other and formed one cluster, while infection-enhancing/non-185 neutralizing NTD antibodies blocked each other and formed a second cluster (Figures 3A, 3C , S12 and S13).

NSEM reconstruction of SARS-CoV-2 S trimer bound with Fabs of neutralizing NTD antibody DH1050.1 and 187 infection-enhancing NTD antibody DH1052 confirmed that the two antibodies could simultaneously bind to 188 distinct epitopes on a single SARS-CoV-2 S trimer ( Figure 3D ). DH1054 was unique as it was able to block both 189 infection-enhancing and neutralizing NTD antibodies (Figures 3C and S13 ).

NTD antibodies did not compete with RBD antibodies for binding to S trimer ( Figure 3A ). This result gave 191 rise to the notion that in a polyclonal mixture of antibodies, the SARS-CoV-2 S trimer could bind both RBD and 192 NTD antibodies. To determine the potential for this complex to form, we liganded SARS-CoV-2 S trimer with Fabs 193 of each type of antibody and visualized the complex using NSEM. NSEM showed that neutralizing RBD antibodies 194 9 could also bind to the same S protomer as neutralizing NTD antibodies DH1050.1 or DH1051 ( Figure 3E ).

Moreover, we found that a single S protomer could be simultaneously occupied by two RBD antibodies (DH1043 196 and DH1047) and an NTD antibody (DH1050.1) ( Figure 3F) . Thus, the S trimer could simultaneously bind to 197 multiple RBD and NTD neutralizing antibody Fabs. 

We observed that the primary epitopes of DH1041 and DH1043 were centered on the Receptor Binding Motif 237 (RBM; residues 483-506) of the RBD (Figures 4A-B , S17 and S18), providing structural basis for the ACE-2 238 blocking phenotype of these antibodies. While DH1041 utilized its heavy chain complementarity determining 239 regions (CDRs) to contact the RBM, the DH1043 paratope included both its heavy and light chains. (Figure 5A) . Throughout the four days of infection, DH1052-infused 283 mice exhibited similar levels of body weight loss and higher survival than mice given negative control IgG (2/9 284 control mice died while 0/10 DH1052-treated mice died) (Figures 5B-C) . In addition, DH1052-treated mice 285 exhibited lower lung hemorrhagic scores, lower lung viral plaque-forming unit (PFU) titers and lower lung tissue 286 subgenomic RNA levels compared to control IgG-infused mice (Figures 5D-F) 

Antibody infusion resulted in human antibody concentrations ranging from 11 to 238 μg/mL in serum at day 2 298 post-challenge (Figures 5H-I and S23A-D) . Sera with DH1050.1 neutralized SARS-CoV-2 pseudovirus at a mean 299 ID50 titer of 19 (Figure 5J) , and neutralized SARS-CoV-2 replication-competent virus at a mean ID50 titer of 192 300 ( Figure 5K) . In contrast, the presence of DH1052 or control antibody CH65 in serum did not neutralize SARS-301 CoV-2 (Figures 5J-K) . Four of 5 macaques that received DH1052 had comparable lung inflammation to control 302 CH65-infused macaques four days after infection (Figures 5L, S24 and S25A) . However, one macaque (BB536A) 303 administered DH1052 showed increased perivascular mononuclear inflammation, perivascular and alveolar edema 304 13 (Figure S25B) , and multiple upregulated bronchoalveolar fluid (BAL) cytokines (Figures S26-S27 ) compared to 305 either control antibody-infused animals or the four other monkeys in the DH1052-treated group. Immunohistologic 306 analysis with markers of macrophage subsets demonstrated alveolar and perivascular infiltration of M2-type 307 macrophages in both monkey BB536A with histologic appearance of alveolar edema and in control antibody-308 treated monkey BB785E (Figure S28) .

In contrast, macaques administered DH1050.1, a neutralizing NTD antibody, had lower lung inflammation 310 than CH65-infused macaques (Figures 5L, S24 and S25A ) and fewer infiltrating macrophages (Figure S28 ).

Infusion of either neutralizing NTD DH1050.1 or in vitro infection-enhancing antibody DH1052 reduced viral 312 nucleocapsid antigen in the lung (Figures 5M, S24 and S25A) . Envelope (E) gene subgenomic RNA (sgRNA) and 313 nucleocapsid (N) gene sgRNA in the BAL were also reduced in macaques that were administered DH1050.1 or 314 DH1052 compared to macaques treated with negative control antibody (Figures 5N-O and S23G) . In nasal swab 315 fluid, macaques showed reduced E and N gene sgRNA when neutralizing antibody DH1050.1 was infused (p < 316 0.05, nonparametric exact Wilcoxon test) (Figures 5P-Q, S23E-F and S23H-I) . With DH1052 infusion, there was 317 a trend to viral control but not significant.

Since DH1052-mediated infection enhancement in vitro increased as the antibody concentration increased 319 ( Figures 1D-E) , we performed an additional challenge study in 6 additional cynomolgus macaques with either 30 320 mg/kg of DH1052 or CH65 control antibody ( Figure S29A ). After challenge, the infection-enhancing, non-321 neutralizing NTD antibody DH1052 again showed a trend to suppress BAL viral load (Figures S29B-D) , and 322 significantly reduced viral replication in nasal swab samples (p < 0.05, nonparametric exact Wilcoxon test) 323 ( Figures S29E-G) compared to the negative control antibody CH65. Moreover, DH1052-treated macaques showed 324 equivalent lung inflammation (Figures S29H-I) , antigen expression (Figures S29J-K) , and cytokine expression 325 ( Figure S30 ) compared to CH65 control-treated macaques. Thus, with high dose (30mg/kg) of DH1052 antibody, 326 there was no enhanced lung pathology or elevated BAL cytokine levels post-SARS-CoV-2 challenge. These results 327 raised the hypothesis that the lung pathology seen in monkey BB536A was rare and may not have been caused by 328 antibody infusion.

We next tested RBD neutralizing antibodies that also mediated infection enhancement in TZM-bl cells 332 expressing FcRI or FcRII in a SARS-CoV-2 acquisition mouse model (Figures 6A-B) . Aged (Figures 2A, S4A-B, S5E-H and S31) . Both of these RBD antibodies mediated FcR-dependent, in 345 vitro SARS-CoV-2 infection enhancement of TZM-bl cells that lacked ACE2 expression (Figures 1F-L) . To 362   1 and 6E) . After antibody infusion at 10 mg of antibody per kg of macaque body weight, serum human IgG 363 concentrations reached 11-228 μg/mL at day 2 post-challenge (Figures 6F-G and S23A-D) . The same macaque 364 serum containing the RBD antibodies exhibited a wide range of neutralization potency (ID50 titers) against SARS-365 CoV-2 pseudovirus or replication-competent virus, commensurate with the neutralization potency of each antibody 366 (Figures 6H and 6I) . Infusion of RBD antibody DH1041, DH1043, or DH1047 resulted in vivo protection from 367 SARS-CoV-2 infection. In macaques administered DH1041, DH1043, or DH1047, lung inflammation was reduced 368 and lung viral antigen was undetectable compared to control (Figures 6J-K, S24 and S25A ). E gene sgRNA and N 369 gene sgRNA were significantly reduced in the upper and lower respiratory tract based on analyses of 370 bronchoalveolar lavage fluid, nasal swabs, and nasal wash samples (Figures 6L-O and S23E-I) .

RBD antibody DH1046, a weaker neutralizing Ab compared to DH1041, DH1043 or DH1047 (Figure 2A) , 372 did not enhance sgRNA E or N in BAL or nasal swab samples (Figures 6L-O and S23E-I) , but protected only a 373 subset of infused monkeys. Three monkeys treated with RBD antibody DH1046 exhibited the same or lower levels 374 of lung inflammation compared to monkeys that received control IgG (Figure 6J) . Two DH1046-infused monkeys 375 had increased lung inflammation scores of 8 and 10 due to increased total areas of inflammation compared to 376 control antibody monkeys (Figures 6J, S24 and S25) , but had no evidence of perivascular or alveolar edema nor 377 evidence of abnormal BAL cytokines (Figures S26 and S27) . Thus, these two animals had more lung involved 378 with inflammatory macrophage infiltration but did not have pathological evidence of vascular leakage. 379

Comparing the DH1046 group to the control IgG group, viral nucleocapsid antigen in the lung was reduced 380 (Figures 6K, S24 and S25A ) . Cross-blocking activity of RBD and NTD neutralizing antibodies tested by surface plasmon resonance (SPR). Soluble, stabilized SARS-CoV-2 S trimer (S-2P) was captured by the antibody on the Y-axis followed by binding by the antibody on the X-axis. Antibody binding was considered competitive (red squares) if the binding antibody did not react with the captured S protein.

(B) 3D reconstruction of simultaneous recognition of SARS-CoV-2 S-2P trimer by two RBD antibodies DH1041+DH1047, or DH1043+DH1047. All three antibodies are SARS-CoV-2 infection-enhancing in ACE2-negative/FcγR-positive cells, but neutralizing in ACE2positive/FcγR-negative cells. (C) Cross-blocking activity of neutralizing antibodies and infection-enhancing NTD antibodies tested by SPR. SARS-CoV-2 S-2P trimer was captured by the antibody on the Y-axis followed by binding by the antibody on the X-axis. Antibody binding was considered competitive (red squares) if the binding antibody did not react with the captured S protein. 

SARS-CoV challenge despite their capacity to mediate FcgammaRII-dependent entry into B cells in vitro

Antibody-dependent enhancement and 598 SARS-CoV-2 vaccines and therapies

A Mouse-Adapted SARS-CoV-2 Induces Acute Lung Injury

Cell and animal models of SARS-CoV-2 603 pathogenesis and immunity

High-throughput isolation of immunoglobulin genes from 606 single human B cells and expression as monoclonal antibodies

Co-evolution of a broadly neutralizing HIV-1 antibody and founder virus

Potent neutralizing antibodies against multiple epitopes on SARS-CoV-2 spike

An infectivity-enhancing site on the SARS-CoV-2 spike protein is targeted by 615 COVID-19 patient antibodies. bioRxiv

Structural insights into immune recognition of 618 the severe acute respiratory syndrome coronavirus S protein receptor binding domain

Cross-neutralization of SARS-CoV-2 by a human monoclonal 622 SARS-CoV antibody

Safety and Efficacy of the BNT162b2 mRNA Covid-19

Convergent antibody responses to SARS-CoV-2 in 628 convalescent individuals

COVID-19, MERS, and SARS in a nonhuman primate model

Isolation of potent SARS-CoV-2 neutralizing antibodies and protection from 634 disease in a small animal model

Antibody potency, effector function, and 637 combinations in protection and therapy for SARS-CoV-2 infection in vivo

Pandemic 639 Preparedness: Developing Vaccines and Therapeutic Antibodies For COVID-19

A 641 human neutralizing antibody targets the receptor-binding site of SARS-CoV-2

Epitopes required for 643 antibody-dependent enhancement of Ebola virus infection

How innate immune mechanisms contribute to antibody-enhanced 645 viral infections

Antigenicity of the SARS-CoV-2 Spike Glycoprotein

Molecular Mechanism for Antibody-Dependent Enhancement of Coronavirus Entry

Epitopes in Humans Elicited both Enhancing and Neutralizing 652 Effects on Infection in Non-human Primates

Structural and Functional Basis of SARS-CoV-2 Entry by Using Human ACE2

Antibody-dependent SARS coronavirus infection is mediated by 658 antibodies against spike proteins

Broad neutralization of SARS-related viruses by 661 human monoclonal antibodies

Structural Basis for Potent Neutralization of 664

Betacoronaviruses by Single-Domain Camelid Antibodies

Cryo-EM structure of the 2019-nCoV spike in the prefusion conformation

Antibody-dependent enhancement (ADE) of SARS-CoV-2 infection in recovered COVID-19 patients: 670 studies based on cellular and structural biology analysis. medRxiv

A 672 noncompeting pair of human neutralizing antibodies block COVID-19 virus binding to its receptor 673 ACE2

Antibodies and vaccines 675 against Middle East respiratory syndrome coronavirus

SARS-coronavirus replication in human peripheral monocytes/macrophages

Antibody-dependent enhancement of SARS coronavirus infection and its role in 681 the pathogenesis of SARS

Antibody-dependent infection of human macrophages by severe acute respiratory 684 syndrome coronavirus

Inborn errors of type I IFN immunity in patients with life-threatening 687 COVID-19

Structural basis for the neutralization of SARS-CoV-2 by an 690 antibody from a convalescent patient

Active replication of Middle East respiratory syndrome coronavirus and aberrant 693 28 induction of inflammatory cytokines and chemokines in human macrophages: implications for 694 pathogenesis

Potently neutralizing and protective human antibodies against 697 SARS-CoV-2

Rapid isolation and profiling of a diverse panel of human 700 monoclonal antibodies targeting the SARS-CoV-2 spike protein

(H-I) Day 2 serum neutralization titers shown as the reciprocal serum dilution that inhibits 50% (ID 50 ) of (H) pseudotyped SARS-CoV-2 replication in 293T/ACE2 cells or (J) SARS-CoV-2 replication in Vero cells. (J-K) Lung histopathology. Sections of the left caudal (Lc), right middle (Rm), and right caudal (Rc) lung were evaluated and scored for (J) the presence of inflammation by hematoxylin and eosin (H&E) staining, and (K) for the presence of SARS-CoV-2 nucleocapsid by immunohistochemistry (IHC) staining. Symbols indicate the sums of Lc, Rm, and Rc scores for each animal. While two monkeys in the DH1046 RBD antibody infusion group had higher overall lung pathology scores than controls, lung histology did not show alveolar or perivascular edema nor was BAL inflammatory cytokine levels elevated. Thus, these two animals had more lung involved with inflammatory macrophage infiltration but did not have pathological evidence of vascular leakage. Statistical significance in all the panels were determined using Wilcoxon rank sum exact test. Asterisks show the statistical significance between indicated group and CH65 control group: ns, not significant, *P<0.05, **P<0.01.