RES EARCH

CORONAVIRUS protein composed of the RBD of SARS-CoV-2

spike. Second, antibodies were isolated from
Studies in humanized mice and convalescent humans peripheral blood mononuclear cells (PBMCs)
of human donors previously infected with
yield a SARS-CoV-2 antibody cocktail SARS-CoV-2. VI mice elicited a robust immune
response against the SARS-CoV-2 spike protein
Johanna Hansen1*, Alina Baum1*, Kristen E. Pascal1, Vincenzo Russo1, Stephanie Giordano1, after immunization. Titers of mice blood collected
Elzbieta Wloga1, Benjamin O. Fulton1, Ying Yan1, Katrina Koon1, Krunal Patel1, Kyung Min Chung1, 7 days after the last boost were determined by
Aynur Hermann1, Erica Ullman1, Jonathan Cruz1, Ashique Rafique1, Tammy Huang1, enzyme-linked immunosorbent assay (ELISA)
Jeanette Fairhurst1, Christen Libertiny1, Marine Malbec1, Wen-yi Lee1, Richard Welsh1, Glen Farr1, (fig. S1). Mice with the highest titers were
Seth Pennington1, Dipali Deshpande1, Jemmie Cheng1, Anke Watty1, Pascal Bouffard1, Robert Babb1, used for antibody isolation. Spleens from these
Natasha Levenkova1, Calvin Chen1, Bojie Zhang1, Annabel Romero Hernandez1, Kei Saotome1, mice were subjected to biotin-labeled mono-
Yi Zhou1, Matthew Franklin1, Sumathi Sivapalasingam1, David Chien Lye2, Stuart Weston3, meric RBD antigen staining and fluorescence-
James Logue3, Robert Haupt3, Matthew Frieman3, Gang Chen1, William Olson1, Andrew J. Murphy1, activated single-cell sorting. In parallel, whole
Neil Stahl1, George D. Yancopoulos1, Christos A. Kyratsous1† blood was collected from three patients 3 to
4 weeks after a laboratory-confirmed poly-
Neutralizing antibodies have become an important tool in treating infectious diseases. Recently, two merase chain reaction (PCR) positive test for
separate approaches yielded successful antibody treatments for Ebola—one from genetically humanized SARS-CoV-2 and after showing symptoms of
mice and the other from a human survivor. Here, we describe parallel efforts using both humanized COVID-19. PBMCs were isolated by ficoll gradi-
ent and RBD-specific B cells were fluorescence-

Downloaded from http://science.sciencemag.org/ on September 28, 2020

mice and convalescent patients to generate antibodies against the severe acute respiratory syndrome
coronavirus 2 (SARS-CoV-2) spike protein, which yielded a large collection of fully human antibodies activated single-cell sorted. The first sets of
that were characterized for binding, neutralization, and three-dimensional structure. On the basis of antibodies derived from these platforms are
these criteria, we selected pairs of highly potent individual antibodies that simultaneously bind the described here.
receptor binding domain of the spike protein, thereby providing ideal partners for a therapeutic antibody To assess antigen-specific responses, natu-
cocktail that aims to decrease the potential for virus escape mutants that might arise in response rally paired heavy and light chain cDNAs were
to selective pressure from a single-antibody treatment. cloned from the mice and human-derived B
cells (8) and transfected into Chinese hamster

I
ovary (CHO) cells to produce recombinant fully
n the setting of the current coronavirus these two fundamentally different approaches human antibodies. Cultured supernatants con-
disease 2019 (COVID-19) pandemic, there were independently exploited to develop fully taining secreted antibodies were subjected
has been urgency to develop potent anti- human antibody treatments for the lethal to high-throughput screening for RBD protein
viral treatments, and early efforts have infectious disease caused by the Ebola virus: binding. Thousands of antibodies were isolated
hearkened back to the days of Emil von Genetically humanized VelocImmune (VI) mice and subsequently screened for binding affinity
Behring, who won the Nobel prize for show- (3, 4) generated an Ebola antibody cocktail to RBD monomer and dimer, epitope diversity,
ing that antibodies can be transferred in treatment (5), whereas sorting B cells from a ability to block angiotensin-converting enzyme
serum. However, technological advances over recovered patient yielded a single–therapeutic 2 (ACE2) receptor binding to RBD, and ability
the last century have allowed for the progres- antibody treatment (6). to neutralize vesicular stomatitis virus (VSV)–
sion from using convalescent serum to the In this work, we describe parallel high- based SARS-CoV-2 spike pseudoparticles [pVSV-
utilization of recombinant fully human anti- throughput efforts using both mice and hu- SARS-CoV-2-S(mNeon)]. Screening yielded
bodies. The proposal to genetically humanize mans to generate antibodies against the spike >200 neutralizing monoclonal antibodies (mAbs)
the immune system of mice (1) has provided protein of severe acute respiratory syndrome with broad potency ranges, dozens of which
an efficient source of naturally selected, fully coronavirus 2 (SARS-CoV-2). The ability to displayed neutralization potency in the pico-
human antibodies. For example, such mice derive antibodies using genetically human- molar range.
have been used to develop checkpoint inhibi- ized VI mice as well as B cells derived from More than 200 of the VI mouse and human-
tors for immune oncology (2) as well as Food convalescent patients enabled us to isolate a derived antibodies isolated in the primary
and Drug Administration (FDA)–approved very large collection of fully human antibodies screen neutralized VSV-based SARS-CoV-2
antibodies for the treatment of rheumatoid with diverse sequences, binding properties, spike pseudoparticles at >70% with ~2 mg/ml
arthritis, cardiovascular disease, cutaneous and antiviral activities. The prospective goal of expressed antibodies. The antibody varia-
squamous cell carcinoma, and allergic dis- of these parallel efforts was to generate a large ble regions were sequenced by next-generation
eases such as asthma and atopic dermatitis. and diverse collection so as to allow for the sequencing, and the repertoire for heavy and
More recently, the ability to sort individual B selection of pairs of highly potent individual light chain pairs was identified (Fig. 1). The
cells from previously infected human patients antibodies that could simultaneously bind predominant lineage of VI mouse antibodies
and molecularly clone the antibody genes from the critical receptor binding domain (RBD) utilized VH3-53 paired with VK1-9, VK1-33,
these B cells has led to an independent source of the spike protein, thereby providing ideal or VK1-39, whereas our human-derived anti-
of human antibodies, albeit limited to anti- partners for a therapeutic antibody cocktail bodies utilized VH3-66 paired with VK1-33 or
bodies that target infectious agents. Recently, that would have the potential to decrease the VH2-70 paired with VK1-39. Notably, VH3-53
likelihood of virus escape mutants that might usage has recently been reported for another
1
2
Regeneron Pharmaceuticals, Inc., Tarrytown, NY 10591, USA. arise in response to selective pressure from human-derived potent neutralizing anti-
National Centre for Infectious Diseases, Tan Tock Seng
single-antibody treatments (7). body against SARS-CoV-2 spike protein (9–11),
Hospital, Yong Loo Lin School of Medicine, Lee Kong Chian
School of Medicine, 16 Jalan Tan Tock Seng, Singapore Anti–SARS-CoV-2 spike antibodies were which indicates that combining the VI mouse
308442, Singapore. 3Department of Microbiology and generated with the following two methods. approach with the human platforms allows
Immunology, University of Maryland School of Medicine, First, VI mice were immunized with a DNA plas- the expanded capture of common rearrange-
Baltimore, MD 21201, USA.
*These authors contributed equally to this work. mid that expresses SARS-CoV-2 spike protein ments found in potent neutralizing SARS-
†Corresponding author. Email: [email protected] and then were boosted with a recombinant CoV-2 mAbs seen in humans. Further analysis

Hansen et al., Science 369, 1010–1014 (2020) 21 August 2020 1 of 5

RES EARCH | R E P O R T

Downloaded from http://science.sciencemag.org/ on September 28, 2020

Fig. 1. Paired antibody repertoire for human- and mouse-derived SARS-CoV-2 neutralizing antibodies. (A and B) Variable (V) gene frequencies for paired heavy
(x axes) and light (y axes) chains of isolated neutralizing antibodies to SARS-CoV-2 for VI mice (A) (N = 185) and convalescent human donors (B) (N = 68). The color
and size of the circles correspond to the number of heavy and light chain pairs present in the repertoires of isolated neutralizing antibodies. Neutralization is defined
as >70% with 1:4 dilution of antibody (~2 mg/ml) in VSV-based pseudoparticle neutralization assay.

of overlaid sequences (fig. S2) showed strong tion. The antibodies bound specifically and tion of SARS-CoV-2 in VeroE6 cells (Fig. 2, B to
overlap in the repertoire of isolated kappa with high affinity to monomeric SARS-COV-2 D). All neutralization assays generated similar
chains between VI mouse and human-derived RBD [dissociation constant (Kd) = 0.56 to potency across the four mAbs, and no combi-
antibodies. Although the repertoire of lambda 45.2 nM] and dimeric SARS-COV-2 RBD (Kd = nations demonstrated synergistic neutrali-
chains did not overlap well, that may be because 5.7 to 42.8 pM). Because recombinant ACE2 zation activity (Fig. 2, C and D). As previous
only two lambda mice were included in this receptor is being considered as a COVID-19 studies indicate pseudoparticles contain-
trial. The average complementarity-determining therapeutic (13), we tested the potency of re- ing the SARS-CoV-2 spike are precleaved by
region (CDR) lengths (fig. S2D) for heavy chains combinant dimeric human ACE2-Fc (hACE2-hFc) furin-like proteases at the polybasic S1-S2
was similar between VI mouse and human- in our neutralization assay. Although recom- cleavage site during biogenesis in HEK293T
derived antibodies, with average lengths of 13 binant ACE2 was able to mediate neutraliza- cells, we assessed the impact of this cleavage
and 14.5 amino acids, respectively. The average tion of the VSV-based spike pseudoparticles on mAb neutralization potency. Spike-stabilized
kappa CDR length (fig. S2E) was the same for as previously reported, its potency was reduced pseudoparticles (fig. S3A) with a monobasic
VI mouse and human-derived antibodies at by more than a factor of 1000 compared with cleavage site (FurMut) in the S1-S2 interface or
9 amino acids, and the lengths were similar that of the best neutralizing mAbs (Fig. 2, A deleted region (FurKO) were produced as pre-
for lambda chains (fig. S2F), with an average and B). viously described (14, 15). No differences were
length of 11.1 and 10.6 amino acids, respectively. A smaller collection of four antibodies was observed in neutralization of either FurMut-
Approximately 40 antibodies with distinct chosen for further analyses to determine whether or FurKO-containing pseudoparticles relative
sequences and potent neutralization activ- the above binding data to RBD reflected bind- to wild-type (WT) in Vero cells (fig. S3B). No-
ities were chosen for additional characteri- ing to trimeric spike protein, whether neutral- tably, stabilized pseudoparticles had compa-
zation, as described below. The neutralization ization potencies noted in the above assays rable or greater infectivity to those with WT
potency of these mAbs spanned the single-digit were consistent with those seen in other assays cleavage sites in Vero cells, whereas substan-
to triple-digit picomolar range in the VSV-based including with SARS-CoV-2, and whether these tial loss of infectivity was observed in Calu-3
pseudoparticle assay. Antibodies shown to antibodies retained neutralization activity against cells (fig. S3C). Authentic SARS-CoV-2 with a
cross-neutralize SARS-CoV-1 and SARS-CoV-2 pseudoparticles with mutations in the S1-S2 natural deletion of the S1-S2 junction also had
spike proteins were weakly neutralizing (12). cleavage site. The binding affinity of these defects in infectivity in Calu-3 but not in Vero
So instead of focusing on cross-neutralizers, four antibodies against trimeric SARS-CoV-2 cells (16), which implicates differential prote-
we focused on nine of the most potent neu- spike (Kd = 37.1 to 42.8 pM) largely paral- ase usage between these two cell types. To in-
tralizing mAbs, with neutralization potencies leled the affinity against the RBD (table S3). vestigate the mechanism of neutralization, we
ranging from 7 to 99 pM (Fig. 2A and table Additionally, the potent neutralizing activity generated antigen-binding fragments (Fabs)
S1). All of these neutralizing mAbs bound to of these four antibodies was confirmed in for the four antibodies. We compared immuno-
the RBD of SARS-CoV-2 spike and blocked its the additional neutralization assays, includ- globulin G (IgG) with corresponding Fabs side
ability to interact with ACE2 with double-digit ing neutralization of pVSV-SARS-CoV-2-S by side for their ability to neutralize pseudo-
picomolar median inhibitory concentrations (mNeon) in the human lung epithelial Calu-3 typed VSV (fig. S4). The IC50s of all the Fabs
(IC50s) (table S1), which supports ACE2 block- cell line, neutralization of replicating VSV- were shifted compared with those of their
ade as the primary mechanism for neutraliza- SARS-CoV-2-S in Vero cells, and neutraliza- parental IgGs, which indicates that bivalent

Hansen et al., Science 369, 1010–1014 (2020) 21 August 2020 2 of 5

RES EARCH | R E P O R T

binding, cross-linking, and steric hindrance in the ADCP assay (fig. S5 and table S3). Al- spike protein RBD, we performed hydrogen-
might all augment neutralization. though REGN10934 was able to mediate both deuterium exchange mass spectrometry (HDX-
Although the role of antibody effector func- ADCC and ADCP, it was not as strong of an in- MS) with the same nine antibodies (Fig. 3),
tion in protection against SARS-CoV-2 is yet ducer in those assays as the other three mAbs which revealed where each of the antibodies
unknown, it has been well established that it (fig. S6 and table S4). Further identification contacts the surface of the RBD and allowed
plays an important role in mAb therapeutic of mAb epitopes through high-resolution struc- comparison with the ACE2 binding site on the
efficacy against other viruses such as Ebola tural analysis may help illuminate the relation- RBD (Fig. 3). As might be expected, most of
and influenza viruses (17–19). Effector cells ship between specific epitopes and effector our neutralizing antibodies contact the RBD
including macrophages and monocytes have function of anti-spike mAbs. in a manner that overlaps the RBD residues
also been shown to be important for antibody- A prospective goal of our effort was to iden- that comprise the ACE2 interface; further-
mediated protection from SARS-CoV-1 infec- tify highly potent individual antibodies that more, the antibodies can be grouped on the
tion (20). To understand whether our lead could be paired in a therapeutic antibody cock- basis of their pattern of contacting the RBD
antibodies are capable of mediating effector tail, aiming to decrease the potential for de- surface. Comparing the cross-competition bind-
function, we assessed both antibody-dependent creased efficacy caused by variants arising as ing assays with the HDX-MS results provides
cellular cytotoxicity (ADCC) and antibody- the pandemic spreads or by virus escape mu- structural insights into the mechanism by which
dependent cellular phagocytosis (ADCP) activ- tants that might be selected for in response to noncompeting pairs of antibodies can simul-
ity in primary human cell bioassays utilizing pressure from a single-antibody treatment (7). taneously bind the RBD and can thus be ideal
natural killer (NK) cells and monocyte-derived Thus, we examined our nine most-potent neu- partners for a therapeutic antibody cocktail.
phagocytes. All four lead antibodies demon- tralizing antibodies in cross-competition bind- REGN10987 and REGN10933 represent such
strated the ability to mediate ADCC and ADCP, ing assays (fig. S7) and identified several pairs a pair of antibodies: REGN10933 targets the

Downloaded from http://science.sciencemag.org/ on September 28, 2020

albeit to slightly different degrees. REGN10987 of noncompeting mAbs with picomolar neu- spike-like loop region on one edge of the ACE2
displayed superior ability to mediate ADCC rel- tralization potency that could potentially be interface. Within that region, the residues that
ative to the other three mAbs, whereas it per- combined to form antibody cocktails. To fur- show the most notable HDX protection by
formed similarly to REGN10989 and REGN10933 ther study the binding regions of our mAbs on REGN10933 face upward, which suggests that

Fig. 2. Neutralization potency of anti–SARS-CoV-2 spike mAbs. (A) Serial pVSV-SARS-CoV-2-S(mNeon) in Calu-3 cells. (C) Neutralization potency of
dilutions of anti-spike mAbs, IgG1 isotype control, and recombinant dimeric individual anti-spike mAbs and combinations of mAbs against replicating
ACE2 (hACE2.hFc) were added with pVSV-SARS-CoV-2-S(mNeon) to Vero VSV-SARS-CoV-2-S virus in Vero cells. Cells were infected with a multiplicity
cells, and mNeon expression was measured 24 hours after infection as a of infection (MOI) 1 of the virus and stained for viral protein 24 hours after
readout for virus infectivity. Data are graphed as percent neutralization relative infection to measure infectivity. (D) Neutralization potency of individual
to virus-only infection control. (B) Neutralization potency of anti-spike mAbs, anti-spike mAbs and combinations of mAbs against SARS-CoV-2-S virus
recombinant dimeric ACE2, and IgG1 isotype control against nonreplicating in VeroE6 cells.

Hansen et al., Science 369, 1010–1014 (2020) 21 August 2020 3 of 5

RES EARCH | R E P O R T

the Fab region of REGN10933 binds the RBD

from the top direction, where REGN10933 will
have collisions with ACE2. To avoid competi-
tion with REGN10933, REGN10987 can only
bind to the HDX-defined protected regions
from the front or the lower left side (in the
front view of REGN10987 in Fig. 3). This would
be consistent with the neutralization data, as
REGN10987 would orient itself in a position
that has high probability to interfere with ACE2.
Confirming the above data, single-particle cryo–
electron microscopy (cryo-EM) of the complex
Fig. 3. HDX-MS determines mAb of SARS-CoV-2 spike RBD bound to Fab frag-
interaction on spike protein ments of REGN10933 and REGN10987 shows
RBD. 3D surface models for the that the two antibodies in this cocktail can
structure of the spike protein RBD simultaneously bind to distinct regions of the
domain showing the ACE2 RBD (Fig. 4 and table S5). A three-dimensional
interface and HDX-MS epitope (3D) reconstructed map of the complex with
mapping results. RBD residues nominal resolution of 3.9 Å shows that the two
Fab fragments bind at different epitopes on the

Downloaded from http://science.sciencemag.org/ on September 28, 2020

that make contacts with ACE2
(21, 22) are indicated in yellow RBD, which confirms that they are noncom-
(top). RBD residues protected by peting antibodies. REGN10933 binds at the top
anti–SARS-CoV2 spike antibodies of the RBD, extensively overlapping the binding
are indicated with colors that site for ACE2. On the other hand, the epitope for
represent the extent of protection, REGN10987 is located on the side of the RBD,
as determined by HDX-MS away from the REGN10933 epitope, and has
experiments. RBD residues in little to no overlap with the ACE2 binding site.
purple and blue indicate sites of We report notable similarities and consis-
lesser solvent exchange upon tencies in the antibodies generated from
antibody binding that have genetically humanized mice and from con-
greater likelihood to be antibody- valescent humans. The scale of the genetic-
binding residues. The RBD engineering approach used to create the VI
structure is reproduced from mouse (involving genetic-humanization of
PDB 6M17 (21). more than 6 Mb of mouse immune genes)
has resulted in the ability to effectively and
indistinguishably mimic the antibody responses
of normal humans. The genetically humanized–
mouse approach has the advantages that it
can potentially allow for further immuniza-
tion optimization strategies and that it can be
applied to noninfectious disease targets. By
combining the efforts from two parallel and
high-throughput approaches for generating
antibodies to the RBD of the SARS-CoV-2 spike
protein, we generated a sufficiently large col-
lection of potent and diverse antibodies that
we could meet our prospective goal of identi-
fying highly potent individual antibodies that
could be combined into a therapeutic anti-
body cocktail. Inclusion of such antibodies
into an antibody cocktail may deliver optimal
antiviral potency while minimizing the odds
of virus escape (7)—two critical, desired fea-
tures of an antibody-based therapeutic for
treatment and prevention of COVID-19. Such
an antibody cocktail is now being tested in
human trials (clinicaltrials.gov NCT04426695
and NCT04425629).

REFERENCES AND NOTES

Fig. 4. Complex of REGN10933 and REGN10987 with the SARS-CoV-2 RBD. (A) 3.9-Å cryo-EM map of 1. F. W. Alt, T. K. Blackwell, G. D. Yancopoulos, Trends Genet.
the REGN10933-RBD-REGN10987 complex, colored according to the chains in the refined model (B). RBD is 1, 231–236 (1985).
2. J. Larkin et al., N. Engl. J. Med. 373, 23–34 (2015).
colored dark blue; REGN10933 heavy and light chains are green and cyan, respectively; and REGN10987 3. A. J. Murphy et al., Proc. Natl. Acad. Sci. U.S.A. 111, 5153–5158
heavy and light chains are yellow and red, respectively. (2014).

Hansen et al., Science 369, 1010–1014 (2020) 21 August 2020 4 of 5

RES EARCH | R E P O R T

4. L. E. Macdonald et al., Proc. Natl. Acad. Sci. U.S.A. 111, or in part with federal funds from the U.S. Department of Health through a materials transfer agreement upon request at
5147–5152 (2014). and Human Services, Office of the Assistant Secretary for https://regeneron.envisionpharma.com/vt_regeneron/. For
5. K. E. Pascal et al., J. Infect. Dis. 218, S612–S626 (2018). Preparedness and Response, Biomedical Advanced Research and questions about how Regeneron shares materials, use the email
6. D. Corti et al., Science 351, 1339–1342 (2016). Development Authority, under OT number HHSO100201700020C. address preclinical.collaborations@ regeneron.com. This work is
7. A. Baum et al., Science 369, 1014–1018 (2020). Author contributions: J.H., A.B., B.O.F., A.H., E.U., A.R., T.H., J.F., licensed under a Creative Commons Attribution 4.0 International
8. X. Wang, B. D. Stollar, J. Immunol. Methods 244, 217–225 P.B., R.B., B.Z., M.Fra., A.J.M., N.S., G.D.Y., and C.A.K. conceptualized (CC BY 4.0) license, which permits unrestricted use, distribution,
(2000). and designed experiments. J.H., K.E.P., V.R., S.G., E.W., B.O.F., Y.Y., and reproduction in any medium, provided the original work is
9. Y. Cao et al., Cell 182, 73–84.E16 (2020). K.K., K.P., K.M.C., E.U., J.Ch., A.R., C.L., M.M., W.-y.L., R.W., S.P., properly cited. To view a copy of this license, visit https://
10. Y. Wu et al., Science 368, 1274–1278 (2020). J.Cr., B.Z., A.R.H., K.S., Y.Z., M.Fra., S.W., J.L., R.H., and M.Fri. creativecommons.org/licenses/by/4.0/. This license does not
11. T. F. Rogers et al., Science 369, 956–963 (2020). performed research. J.H., K.E.P., V.R., S.G., E.W., B.O.F., Y.Y., apply to figures/photos/artwork or other content included in the
12. M. Yuan et al., Science 368, 630–633 (2020). K.K., K.P., K.M.C., E.U., J.Ch., A.R., J.F., C.L., M.M., W.-y.L., G.F., article that is credited to a third party; obtain authorization
13. C. Lei et al., Nat. Commun. 11, 2070 (2020). D.D., J.Cr., P.B., R.B., N.L., C.C., B.Z., A.R.H., K.S., Y.Z., M.Fra., S.W., from the rights holder before using such material.
14. A. C. Walls et al., Cell 181, 281–292.e6 (2020). J.L., R.H., M.Fri., G.C., W.O., N.S., G.D.Y., and C.A.K. analyzed data.
15. D. Wrapp et al., Science 367, 1260–1263 (2020). S.S. and D.C.L. acquired reagents. J.H., A.B., A.R., B.Z., M.Fra., G.D.Y., SUPPLEMENTARY MATERIALS
16. S. Y. Lau et al., Emerg. Microbes Infect. 9, 837–842 (2020). and C.A.K. wrote the paper. C.A.K. acquired funding. Competing
science.sciencemag.org/content/369/6506/1010/suppl/DC1
17. D. J. DiLillo, P. Palese, P. C. Wilson, J. V. Ravetch, interests: Regeneron authors own options and/or stock of the
Materials and Methods
J. Clin. Invest. 126, 605–610 (2016). company. This work has been described in one or more pending
Figs. S1 to S7
18. E. O. Saphire, S. L. Schendel, B. M. Gunn, J. C. Milligan, G. Alter, provisional patent applications. G.C., W.O., A.J.M., N.S., G.D.Y., and
Tables S1 to S5
Nat. Immunol. 19, 1169–1178 (2018). C.A.K. are officers of Regeneron. Data and materials availability:
References (23–33)
19. E. O. Saphire et al., Cell 174, 938–952.e13 (2018). The structure of SARS-CoV-2 RBD in complex with REGN10933 and
MDAR Reproducibility Checklist
20. F. Yasui et al., Virology 454-455, 157–168 (2014). REGN10987 Fabs has been deposited in the Protein Data Bank,
21. R. Yan et al., Science 367, 1444–1448 (2020). Data S1
with PDB ID 6XDG. The corresponding cryo-EM map is available in
22. J. Lan et al., Nature 581, 215–220 (2020). the Electron Microscopy Data Bank, with ID EMD-22137. Sequences View/request a protocol for this paper from Bio-protocol.
of the nine characterized antibodies have been deposited in
ACKN OW LEDG MEN TS GenBank and are available in the supplementary materials. 30 May 2020; accepted 11 June 2020

Downloaded from http://science.sciencemag.org/ on September 28, 2020

The authors thank K. Tramaglini for program management. Regeneron materials described in this manuscript may be made Published online 15 June 2020
Funding: A portion of this project has been funded in whole available to qualified, academic, noncommercial researchers 10.1126/science.abd0827

Hansen et al., Science 369, 1010–1014 (2020) 21 August 2020 5 of 5

Studies in humanized mice and convalescent humans yield a SARS-CoV-2 antibody cocktail
Johanna Hansen, Alina Baum, Kristen E. Pascal, Vincenzo Russo, Stephanie Giordano, Elzbieta Wloga, Benjamin O. Fulton,
Ying Yan, Katrina Koon, Krunal Patel, Kyung Min Chung, Aynur Hermann, Erica Ullman, Jonathan Cruz, Ashique Rafique,
Tammy Huang, Jeanette Fairhurst, Christen Libertiny, Marine Malbec, Wen-yi Lee, Richard Welsh, Glen Farr, Seth
Pennington, Dipali Deshpande, Jemmie Cheng, Anke Watty, Pascal Bouffard, Robert Babb, Natasha Levenkova, Calvin Chen,
Bojie Zhang, Annabel Romero Hernandez, Kei Saotome, Yi Zhou, Matthew Franklin, Sumathi Sivapalasingam, David Chien
Lye, Stuart Weston, James Logue, Robert Haupt, Matthew Frieman, Gang Chen, William Olson, Andrew J. Murphy, Neil Stahl,
George D. Yancopoulos and Christos A. Kyratsous

Science 369 (6506), 1010-1014.

DOI: 10.1126/science.abd0827originally published online June 15, 2020

Downloaded from http://science.sciencemag.org/ on September 28, 2020

An antibody cocktail against SARS-CoV-2
There is an urgent focus on antibodies that target the severe acute respiratory syndrome coronavirus 2
(SARS-CoV-2) viral spike and prevent the virus from entering host cells. Hansen et al. generated a large panel of
antibodies against the spike protein from humanized mice and recovered patients. From this panel, they identified
several neutralizing antibodies, including pairs that do not compete for binding to the receptor binding domain. Baum et
al. focused in on four of these antibodies. All four are effective against known spike variants. However, by growing a
pseudovirus that expresses the spike in the presence of individual antibodies, the authors were able to select for spike
mutants resistant to that antibody. In contrast, escape mutants are not selected when pseudovirus is grown in the
presence of pairs of antibodies that either do not compete or only partially compete for binding to the RBD. Such a pair
might be used in a therapeutic antibody cocktail.
Science, this issue p. 1010, p. 1014

ARTICLE TOOLS http://science.sciencemag.org/content/369/6506/1010

SUPPLEMENTARY http://science.sciencemag.org/content/suppl/2020/06/15/science.abd0827.DC1
MATERIALS

RELATED http://stm.sciencemag.org/content/scitransmed/12/541/eabb5883.full
CONTENT
http://stm.sciencemag.org/content/scitransmed/12/549/eabb9401.full
http://stm.sciencemag.org/content/scitransmed/12/534/eabb1469.full
http://stm.sciencemag.org/content/scitransmed/12/546/eabc1931.full
REFERENCES This article cites 33 articles, 12 of which you can access for free
http://science.sciencemag.org/content/369/6506/1010#BIBL

Use of this article is subject to the Terms of Service

Science (print ISSN 0036-8075; online ISSN 1095-9203) is published by the American Association for the Advancement of
Science, 1200 New York Avenue NW, Washington, DC 20005. The title Science is a registered trademark of AAAS.
Copyright © 2020 The Authors, some rights reserved; exclusive licensee American Association for the Advancement of
Science. No claim to original U.S. Government Works
 PERMISSIONS http://www.sciencemag.org/help/reprints-and-permissions

Downloaded from http://science.sciencemag.org/ on September 28, 2020

Use of this article is subject to the Terms of Service

Science (print ISSN 0036-8075; online ISSN 1095-9203) is published by the American Association for the Advancement of
Science, 1200 New York Avenue NW, Washington, DC 20005. The title Science is a registered trademark of AAAS.
Copyright © 2020 The Authors, some rights reserved; exclusive licensee American Association for the Advancement of
Science. No claim to original U.S. Government Works