key: cord-348729-kejlm425
authors: Liu, Xiaoyu; Gao, Fang; Gou, Liming; Chen, Yin; Gu, Yayun; Ao, Lei; Shen, Hongbing; Hu, Zhibin; Guo, Xiling; Gao, Wei
title: Neutralizing Antibodies Isolated by a site-directed Screening have Potent Protection on SARS-CoV-2 Infection
date: 2020-05-04
journal: bioRxiv
DOI: 10.1101/2020.05.03.074914
sha: 
doc_id: 348729
cord_uid: kejlm425

Neutralizing antibody is one of the most effective interventions for acute pathogenic infection. Currently, over three million people have been identified for SARS-CoV-2 infection but SARS-CoV-2-specific vaccines and neutralizing antibodies are still lacking. SARS-CoV-2 infects host cells by interacting with angiotensin converting enzyme-2 (ACE2) via the S1 receptor-binding domain (RBD) of its surface spike glycoprotein. Therefore, blocking the interaction of SARS-CoV-2-RBD and ACE2 by antibody would cause a directly neutralizing effect against virus. In the current study, we selected the ACE2 interface of SARS-CoV-2-RBD as the targeting epitope for neutralizing antibody screening. We performed site-directed screening by phage display and finally obtained one IgG antibody (4A3) and several domain antibodies. Among them, 4A3 and three domain antibodies (4A12, 4D5, and 4A10) were identified to act as neutralizing antibodies due to their capabilities to block the interaction between SARS-CoV-2-RBD and ACE2-positive cells. The domain antibody 4A12 was predicted to have the best accessibility to all three ACE2-interfaces on the spike homotrimer. Pseudovirus and authentic SARS-CoV-2 neutralization assays showed that all four antibodies could potently protect host cells from virus infection. Overall, we isolated multiple formats of SARS-CoV-2-neutralizing antibodies via site-directed antibody screening, which could be promising candidate drugs for the prevention and treatment of COVID-19.

Coronavirus disease 2019 (COVID-19) is a worldwide epidemic of respiratory disease caused by the novel human coronavirus SARS-CoV-2 1,2 . Currently, over three million infected people have been identified in more than 200 countries and regions by laboratory testing, with an average mortality rate of approximately 6% (https://covid19.who.int/). The real number of infected cases is even higher, considering the detection limitation in many counties. Therefore, there is an urgent need to develop an effective vaccine and neutralizing antibody against SARS-CoV-2.

SARS-CoV-2, a single-stranded positive-sense RNA virus of the β-Coronaviridae family 3 . It shares 79% nucleotide sequence identity with SARS-CoV-1 4 . Both SARS-CoV-2 and SARS-CoV-1 infect host cells by directly interacting with the host angiotensin-converting enzyme-2 (ACE2) receptor through their spike glycoprotein expressed on the viral membrane and subsequently trigger the fusion of the cell and virus membrane for cell entry 5,6 . Spike glycoprotein exists as a homotrimeric complex on the viral membrane of coronaviruses 7 . Each spike monomer contains an S1 subunit and an S2 subunit 8 . The S1 subunit binds to ACE2 through its receptor-binding domain (RBD) to initiate cell recognition, whereas the S2 subunit anchors the spike protein to the viral envelope and responds to S1-induced cell recognition to mediate effective membrane fusion via a conformational transition 7 . These determined infection mechanisms indicated that blocking the interaction of SARS-CoV-2-RBD and ACE2 would cause a direct neutralizing effect against virus.

Neutralizing antibody is one of the most effective interventions for acute pathogenic infection 9 . Several approaches are reported to obtained SARS-CoV-2 neutralizing antibodies successfully. Among them, one approach is to screen the preexisting SARS-CoV-1 antibody repertoires by evaluating cross-reactivity 10 . An alternative approach is to clone the neutralizing antibody from the isolated SARS-CoV-2-RBD-specific single B cells from infected patients 11, 12 . However, the feasibility of these two strategies is quite limited due to the rare chance of accessing either SARS-CoV-1 antibodies or SARS-CoV-2-infected patients. Therefore, in vitro site-directed screening in a human antibody library would be more feasible and efficient.

The RBD is a relatively isolated domain of the S1 subunit with independent function 13 .

The crystal structure of the SARS-CoV-2-RBD 14 and the SARS-CoV-2-RBD/ACE2 complex 15, 16 has already been reported. It presents quite similar interaction details compared to those of the previously determined SARS-CoV-1-RBD/ACE2 structure 17 .

Notably, a recent study performed a systematic bioinformatics analysis to predict the potential B cell epitope and T cell epitope of SARS-CoV-2 18 . The only predicted conformational B cell epitope in the RBD is located within the ACE2 interface (P491-Y505). This information suggests that the ACE2 interface of SARS-CoV-2-RBD might have high immunogenicity, which would be a suitable targeting epitope to develop SARS-CoV-2-specific antibodies with potent neutralizing function by in vitro screening.

In the current study, we selected the ACE2-interface of SARS-CoV-2-RBD as the targeting epitope to screen neutralizing antibody. We performed site-directed antibody screening by phage display and finally obtained one IgG antibody and three single domain antibodies with potent neutralizing activities for SARS-CoV-2. These neutralizing antibodies are promising candidate drugs for the prevention and treatment of COVID-19.

Plasmids and reagents. The coding sequences of ACE2 and the SARS-CoV-2 spike were cloned into the pLVX vector to construct stable cell lines. The coding sequences of truncated ACE2 (Q18-S740), SARS-CoV-1-RBD (P317-V510), SARS-CoV-2-RBD (P330-V524) and SARS-CoV-2-RBD mut were cloned into the pFUSE vector to obtain hFc-fusion proteins. SARS-CoV-2-RBD mut was designed by substituting key residues on SARS-CoV-2-RBD with Ala or Phe to disrupt its interaction with ACE2 (Table 1) .

The coding sequences of domain antibodies were also cloned into the pFUSE vector to obtain hFc-fusion proteins. The coding sequences of the 4A3 heavy chain variable region and light chain variable region were amplified by adding the IL-2 signal peptide and cloned into the expression vectors pFUSE-CHIg-HG1 and pFUSE2-CLIg-hk (Invivogen, San Diego, CA), respectively. All plasmids were identified by sequencing.

SARS-CoV-2-RBD-his protein was purchased from GenScript (GenScript, Nanjing), and GPC5-his protein was purchased from R&D (Minneapolis, MN). Phage display. The TG1 clone was picked and cultured overnight in 2YT medium at 37 °C. SARS-CoV-2-RBD-his or SARS-CoV-2-RBD-hFc proteins in PBS buffer were coated on ELISA plates at 4 °C overnight. The phage library (Tomlison I library and Domain antibody library) and the coated wells were blocked with PBST with 5% milk at room temperature for 1 h. The blocked phages were precleaned by negative antigens and then added into the coated wells to incubate for 1 h at room temperature. After washing 20 times with PBST, the SARS-CoV-2-RBD-binding phages were eluted with 100 mM triethylamine (Sigma-Aldrich, Xuhui, Shanghai). All eluted phages were collected and used to infect TG1 cells. After incubating with helper phages, the eluted phages were rescued with a titer of approximately 10 11~1 0 12 pfu/ml for the next round of screening.

ELISA. For direct ELISA, the indicated antigen (5 μg/ml) was coated on an ELISA plate at 4 °C overnight. After blocking, biotin-labeled ACE2, blocked phage or the indicated antibodies were added to the wells and incubated at 37 °C for 0.5 h.

Streptavidin-HRP (Thermo, Pudong New Area, Shanghai), a rabbit anti-M13 HRP antibody (for phage) (GE Healthcare, Milwaukee, WI), or a goat anti-human Fcγ HRP antibody (Jackson ImmunoResearch, West Grove, PA) was added. TMB and H2SO4

were added to detect the OD450 nm value. For capture ELISA, an anti-his antibody (5 µg/ml) was coated on an ELISA plate at 4 °C overnight. After blocking, soluble antibodies extracted from the periplasm of TG1 were added into the wells and incubated at 37 °C for 0.5 h. After washing, SARS-CoV-1-RBD-hFc, SARS-CoV-2-RBD-hFc or SARS-CoV-2-RBD mut-hFc protein (5 μg/ml) was added and incubated at 37 °C for 0.5 h. After washing, a goat anti-human Fcγ HRP antibody (Jackson ImmunoResearch, West Grove, PA) was added and incubated at 37 °C for 0.5 h. TMB and H2SO4 were added to detect the OD450 nm value. were performed under the approved standard operating procedures of Biosafety Level 3 laboratory. Live SARS-CoV-2 was isolated from throat swabs of SARS-CoV-2infected patients in Jiangsu Province and identified by sequencing (strai Beta CoV-JS27). Viruses were amplified in Vero E6 cells and made as working stocks at 10 5 pfu/ml. For the neutralization assay, Vero E6 cells were seeded into 96-well plates at 10 4 /well and cultured overnight. SARS-CoV-2 (100 TCID50) was pre-incubated with a series of diluted antibodies at 37 °C for 1 h. Then, the virus-antibody mixtures were added to seeded Vero E6 cells. Cytopathic effects (CPEs) were photographed 4 days later.

Statistical analysis. All group data are expressed as the mean ± standard deviation (SD) of a representative experiment performed at least in triplicate, and similar results were obtained in at least three independent experiments. All statistical analyses were conducted using GraphPad Prism 8.0. Two-tailed Student's t-test of the means was used for statistical analysis, with P *<0.05 defined as significant.

To screen potent neutralizing antibodies against SARS-CoV-2, we first analyzed the ACE2 interface of SARS-CoV-2-RBD ( Figure 1A ). We selected sixteen residues essential for the hydrophobic or electrostatic effects within the ACE2 interface of SARS-CoV-2 and made mutations ( Table 1) . The predicted structure model of SARS-CoV-2-RBD mut showed that the overall conformation of the RBD did not change after mutation ( Figure 1B ), but the surface property of the ACE2 interface had been changed ( Figure 1C ). We then fused SARS-CoV-2-RBD, SARS-CoV-2-RBD mut and SARS-CoV-1-RBD with a human Fc tag and performed purification ( Figure 1D ). The ACE2binding activity of purified SARS-CoV-2-RBD mut significantly decreased compared to that of SARS-CoV-2-RBD ( Figure 1E ), and a similar trend was also observed when we detected the binding of SARS-CoV-2-RBD and SARS-CoV-2-RBD mut on ACE2-CHO cells ( Figure 1F and Figure 1G ). These results indicated that these mutations successfully abolished ACE2 recognition by destroyed ACE2 interface of SARS-CoV-2-RBD. The purified SARS-CoV-2-RBD mut would be suitable to function as the negative antigen in our screening.

To obtain SARS-CoV-2-specific neutralizing antibody, we performed site-directed antibody screening by phage display. We utilized SARS-CoV-2-RBD-his and SARS-CoV-2-RBD-hFc as the positive antigens and GPC5-his and SARS-CoV-2-RBD mut-hFc as the negative antigens to execute the selection within a naive human scFv antibody phage library and a domain antibody phage library, respectively ( Figure 2A ).

After four rounds of screening, the antigen-binding activity of the eluted phage dramatically increased ( Figure 2B) . Notably, the eluted phage exhibited a stronger binding signal on SARS-CoV-2-RBD compared to that on SARS-CoV-2-RBD mut, especially those from the domain antibody library ( Figure 2C ), indicating an expected precleaning effect during selection. We then randomly picked 200 single clones from the 4 th round eluted phage and performed monoclonal phage ELISA. The positive binders were enriched significantly in both libraries ( Figure 2D ). All the positive binders were sequenced, and finally, we obtained nine enriched clones from the domain antibody library and one enriched clone from the scFv antibody library. Among them, nine domain antibodies bound to SARS-CoV-2-RBD specifically, whereas one scFv antibody (4A3) showed weak binding activity on SARS-CoV-1-RBD and SARS-CoV-2-RBD mut in the phage ELISA ( Figure 2E ).

The nine binders isolated from the domain antibody library contained only the antibody heavy chain variable region. We then fused them with a human Fc tag and performed purification. The 4A3 scFv binder was converted into a human IgG1 and purified as well ( Figure 3A ). Among all the purified antibodies, 4A12, 4D5, 4A10, 4C5 and 4A3

were selected for further evaluation due to their strong binding activities to both SARS-CoV-2-RBD ( Figure 3B ) and SARS-CoV-2 spike-overexpressing cells ( Figure 3C ), in addition to their promising expression yield and purity.

To examine the potential neutralizing capabilities of our candidate antibodies, we detected whether they would disturb the binding between SARS-CoV-2-RBD and ACE2-positive cells. Three domain antibodies (4A12, 4D5, 4A10) and 4A3 IgG exhibited obvious inhibition in a dose-dependent manner, whereas the SARS-CoV-1-neutralizing antibody M396 seemed to have no effect ( Figure 4A and 4C). Considering the high similarity of the RBDs, we also evaluated the blocking effects of our antibodies on the binding between SARS-CoV-1-RBD and ACE2-CHO cells. None of our candidate antibodies showed inhibitory effects ( Figure 4B and 4D) . Notably, we suspected that clone 4A3, which showed weak cross-reaction with SARS-CoV-1-RBD in the phage ELISA ( Figure 2E ), might exhibit a certain blocking effect for the cell binding of SARS-CoV-1-RBD, but actually, it showed specific blocking on only the cell binding of SARS-CoV-2-RBD. We then performed SPR to further evaluate the affinities of these candidate antibodies. The results showed that the affinities of our antibodies ranged from 1.03 nM to 5.82 nM (Table 2 and Figure 5 ). Altogether, these results indicated that we obtained four antibodies that might have potential neutralizing functions against SARS-CoV-2.

Domain antibody 4A12 was predicted to have advantage for accessing all three

According to the recently reported cryo-EM structure, the SARS-CoV-2 spike trimer appears in two distinct confirmation states: a closed state with the three RBDs embedded and an open state with only one RBD extended for ACE2 binding. Since the extended RBD presents a complete exposed ACE2 interface that would be more easily captured by an antibody than the closed RBD, we simply wondered whether the remaining embedded ACE2 interfaces of both the open and closed spike trimers could be accessed by antibody. Because the reported cryo-EM structure of the SARS-CoV-2 spike trimer lacks some RBD residue information, we replaced its RBD with the determined structure of SARS-CoV-2-RBD to establish the remodeled SARS-CoV-2 spike trimer. We then analyzed the interspace between each embedded ACE2 interface and its neighboring monomer. The three interspaces in the closed trimer were quite uniform (from 17.2 Å to 19.5 Å). For the two interspaces of the open trimer, one did not change (19.5 Å), whereas the other was somehow occupied by the extended RBD (11.2 Å) ( Figure 6A ). The domain antibody exhibited a relatively smaller size and antigen interface because it contains only a heavy chain variable region with three complementarity-determining regions (CDRs) instead of six in IgG ( Figure 6B ). We then performed molecular docking to compare the binding patterns of 4A3 scFv and the domain antibodies on our remodeled spike trimers. Both 4A3 and the domain antibodies were predicted to recognize all three ACE2 interfaces of the closed trimer ( Figure 6C ). However, when docked with the open trimer, 4A3 scFv was predicted to target only two ACE2 interfaces. It could not access the embedded ACE2 interface occupied by the extended RBD, probably due to its larger size. Surprisingly, one of the domain antibodies, 4A12, was predicted to block all three ACE2 interfaces ( Figure 6D ).

These predictions suggested that the domain antibody might have the advantage of blocking all the ACE2 interfaces of both closed and open trimers, which might offer more effective protection against SARS-CoV-2. Nevertheless, detailed structural examination of the binding patterns of these antibodies with SARS-CoV-2-RBD and SARS-CoV-2 spike trimers is extremely important in our future investigations.

To evaluate the potential antiviral activities of our antibodies, we prepared SARS-CoV-2 pseudovirus by replacing the coding sequence of VSV glycoprotein with SARS-CoV-2 spike glycoprotein in a lentivirus packaging system ( Figure 7A ). We preincubated our candidate antibodies with SARS-CoV-2 pseudovirus and then added them to cultured ACE2-CHO cells. The domain antibodies 4A12, 4A10, and 4D5 and 4A3 IgG exhibited obvious neutralizing potencies, with IC50 values from 0.19 µg/ml to 1.13 µg/ml ( Figure   7B ). These results were consistent with the blocking pattern observed in the SARS-CoV-2-RBD and ACE2-CHO cell-binding assays ( Figure 4A and 4C). Although 4C5 did not show a blocking effect in our cell-binding assay, it still showed a mild neutralizing capability ( Figure 7B ).

Based on our pseudovirus experiment, we selected 4A3 IgG, 4A12 and 4A10 to evaluate their neutralizing activities on authentic SARS-CoV-2. The observations of CPE showed that 4A3, 4A12 and 4A10 exhibited complete protection at 7.5 µg/ml, 10 µg/ml and 37.5 µg/ml, respectively, when we performed a 4-day exposure to SARS-CoV-2 (100 TCID50) ( Figure 7C and 7D) . These protective effects were still quite stable when the exposure time was extended to 10 to 15 days (Table 3) .

Overall, we isolated several human monoclonal neutralizing antibodies against SARS-CoV-2 by site-directed screening strategy, which could be promising candidate drugs for the prevention and treatment of COVID-19.

Currently, several studies have obtained SARS-CoV-2 antibodies with neutralizing activity by phage display, including full length antibody 21 and domain antibody 22,23 .

Instead of using a mutant SARS-CoV-2-RBD with a disrupted ACE2 binding motif, one of these studies utilizing a captured ACE2/SARS-CoV-2-RBD complex to perform the pre-absorption which also obtained neutralizing antibodies as well 21 .

The RBDs mediate the ACE2 interaction for both SARS-CoV-1 and SARS-CoV-2 13 .

Many investigations utilize the RBD as the target region to produce neutralizing antibodies and vaccines 24-26 . It has been reported that immunizing rodents with SARS-CoV-2-RBD elicits a robust neutralizing antibody response without antibodydependent enhancement (ADE) 27 . According to the determined structure, the ACE2 interface of the RBD presents only a small portion of the whole domain 15 , and only those antibodies binding to the interface would directly interfere the interaction with ACE2. This would be one possible explanation for why not all the antibodies isolated with RBD binding activity showed neutralizing activity. Based on this information, we performed site-directed screening with the positive antigen SARS-CoV-2-RBD and negative antigen SARS-CoV-2-RBD mut to ensure that we obtained antibodies accurately targeting the ACE2 interface. As we predicted, most of our candidate antibodies exhibited a significant blocking effect for ACE2 recognition on cells ( Figure   4 ). Although the antibody 4C5 did not, it did show a relatively mild virus-neutralizing function compared to that of the other candidates (Figure 7) . These functional evaluations might prove the feasibility of our site-directed screening strategy, but further validations, such as epitope determination by cryo-EM, are still necessary.

The small size of the domain antibody enables several unique advantages, including high expression yield, enhanced tissue penetration, and hidden epitope targeting 28-30 . Phage display is a well-established strategy for in vitro antibody screening 41 . The antibody is amplified in the prokaryotic system quickly and efficiently 42 . We initially obtained ten enriched binders that all showed strong antigen-binding activity. However, when expressed in 293T cells, three of them lost antigen binding, and two of them exhibited poor yield. This effect is a common problem for phage display when shifting to a eukaryotic expression system. Other in vitro screening strategies, such as yeast display 43 or mammalian cell display 44 , may offer more stable selection due to their eukaryotic background. Antibodies selected from naive phage libraries usually require further optimization to improve their affinity. Even though the affinities of our antibodies are in a range (10 -9 M) similar to that of antibodies screened from SARS-CoV-2-infected patients 11 , we still need to perform affinity maturation to achieve more competitive affinity and neutralizing function against SARS-CoV-2. On the other hand, collecting the blood of infected donors to construct a SARS-CoV-2-specific phage library for antibody screening will be an alternative strategy as well.

In the current study, we established site-directed phage display screening, a feasible and efficient in vitro assay, to obtain neutralizing antibodies. In addition to screen neutralizing antibodies against viruses, this strategy could be widely used for isolating function-blocking antibodies targeting the essential domains of various antigens. 

Potent human neutralizing antibodies elicited by SARS-CoV-2 infection

Human monoclonal antibodies block the binding of SARS-CoV-2 spike protein to angiotensin converting enzyme 2 receptor

A 193-amino acid fragment of the SARS coronavirus S protein efficiently binds angiotensin-converting enzyme 2

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

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

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

Structure of SARS coronavirus spike receptorbinding domain complexed with receptor

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

Automated structure refinement of macromolecular assemblies from cryo-EM maps using Rosetta

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

Mammalian cell display for the discovery and optimization of antibody therapeutics

We thank our colleagues Dr. Yujie Sun and Dr. Ningning Wang for providing cell lines; 

The authors declare that they have no conflicts of interest with the contents of this