key: cord-0688322-yb173b3l authors: Guo, Jiazheng; Zhang, Jun; Du, Peng; Lu, Jiansheng; Chen, Lei; Huang, Ying; Yu, Yunzhou; Xie, Qing; Wang, Rong; Yang, Zhixin title: Generation and characterization of humanized synergistic neutralizing antibodies against SARS‐CoV‐2 date: 2022-05-07 journal: J Med Virol DOI: 10.1002/jmv.27801 sha: 9f0cc271481d8d3dc26e2aca1fd0df66a3b74579 doc_id: 688322 cord_uid: yb173b3l The emerging coronavirus, severe acute respiratory syndrome coronavirus 2 (SARS‐CoV‐2), is the causative agent of coronavirus disease 2019 (COVID‐19), which has become a severe threat to global public health and local economies. In this study, several single‐chain antibody fragments that bind to the receptor‐binding domain (RBD) of the SARS‐CoV‐2 spike (S) protein were identified and used to construct human‐mouse chimeric antibodies and humanized antibodies. These antibodies exhibited strong binding to RBD and neutralization activity towards a SARS‐CoV‐2 pseudovirus. Moreover, these antibodies recognize different RBD epitopes and exhibit synergistic neutralizing activity. These provide candidate to combination use or bispecific antibody to potential clinical therapy for COVID‐19. transportation during home quarantine measures, made the collection of sufficient convalescent serum samples challenging. In this study, we constructed a murine phage antibody library against the RBD protein of SARS-CoV-2. Using antibody display technology, a panel of chimeric human-mouse neutralizing antibodies were generated and characterized. The results indicate that some mAbs exhibited favorable biological activity in vitro for the inhibition of viral entry into host cells and neutralizing SARS-CoV-2, providing the possibility for further therapeutic research. Phagemid pADSCFV-S was used to construct a single-chain antibody fragment (scFv) antibody phage library. The antibody eukaryotic expression vectors, pTSE-G1n and pTSE-K, which contain the constant region of the human heavy chain and light chain, respectively, were used to create the complete antibody molecules as previously described. 10 pNL4.3-luc-R − E − , a plasmid encoding an Env-defective, luciferaseexpressing HIV-1 genome, was kindly provided by Prof. Lihua Hou. pCAGGS-WSS was constructed to encode the wild-type SARS-CoV-2 S protein (deleting the 19AA of the C-terminal), which was constructed for preparation of the pseudovirus. Mice immunizations were performed in accordance with institutional regulations and guidelines. Five 6−8-week-old, specific pathogen-free BALB/c female mice (purchased from Beijing Laboratory Animal Center), were intraperitoneally administered 20 μg recombined SARS-CoV-2 RBD protein (Sino Biological Inc.) diluted in phosphate-buffered saline (PBS), followed by two similar immunizations 2 and 4 weeks later. An enzyme-linked immunosorbent assay (ELISA) was applied to determine the titer of the antisera in mice. The mice with the highest and specific activity against the RBD protein were selected for the construction of a mouse scFv-phage antibody immune library. A final booster dose of 20 μg RBD protein was administered 3 days before the spleen was removed from the hyper-immunized mice. Serum collected from the immunized mice was used as a positive control. The total RNA was extracted from the spleen tissue of the hyperimmunized mice using an RNA isolation kit (Omega). Variable light (VL) and variable heavy (VH) chain genes were amplified by RT-PCR and fused to the scFv gene using overlay-extended PCR. The scFv gene repertoire was digested with Not I/Sfi I and inserted into a phage display vector (pADSCFV-S). Competent Escherichia coli TG1 cells were transformed with the ligation mixture by electroporation. The final scFv antibody gene libraries were identified, aliquoted, and stored at −80°C. The resultant recombinant phage library was produced by the addition of a wild-type M13K07 helper phage. Selection of scFv phage clones from the library was accomplished via four successive rounds of affinity enrichment with the SARS-CoV-2 RBD protein. Both the input and output phages in each cycle were titrated on TYE plates. The general binding specificity of the selective phage clones was detected by a phage-ELISA. The clones which had an optical density (OD) value greater than or equal to 2.0 was selected to operate a DNA sequence analysis. The VH and VL sequences of the scFv was determined using the IMGT database (http://www.imgt.org/IMGTlect/). Human-mouse chimeric immunoglobulin G (IgG), humanized IgG, and purification VH and VL chain regions from the 12 selected scFv clones were amplified by PCR and recloned into the pTSE-G1n and pTSE-K plasmids for IgG1 heavy chain and light chain expression to express human-mouse chimeric IgG. To construct humanized IgG, the VH and VL amino acid sequences were analyzed on the website http:// www.abysis.org/abysis/; the amino acid residues with frequency less than 0.1 in the framework were replaced by high frequency in Homo sapiens to increase the degree of humanization according to the Z-score. The space structure of the humanized IgG was constructed with Swissmodel, and the accessible surface area of amino acid residue solution was analyzed to determine which amino acid residues could be humanized. Competitive binding experiments were performed using a ForteBIO ® Octet QK e System (Pall ForteBio Corporation) to determine whether the neutralizing antibodies blocked the binding between RBD and ACE2. Purified ACE2 in HBS-EP buffer (Cytiva) at 300 nM was loaded on an individual biosensor followed by a 1 min wash in HBS-EP buffer. Afterward, a 10 min association was performed in the sample plate with either 150 nM RBD or RBD combined with mAbs. The results were analyzed using Data Analysis Software 7.0 (Pall ForteBio Corporation). A standard virus neutralization assay approved by the national control authority was performed as follows. A panel of tested antibodies was serially diluted and mixed with 100 TCID 50 /well SARS-CoV-2 (Wuhan-hu-1 strain). The antibody-virus mixtures were incubated for 60 min and then added to 96-well plates containing confluent monolayers of Vero-E6 cells in triplicate. Following a 1 h adsorption at 37℃, the supernatants were removed and replaced with 200 µl/well cell culture medium. The plates were then incubated at 37℃ with 5% CO 2 for 3 days. To reduce human anti-mouse antibody (HAMA) reactivity, humanmouse chimeric antibodies were humanized using the website, http:// www.abysis.org/abysis/, and Swissmodel (The data of Z-score, the simulation diagram of the spatial trend of carbon atoms were shown in F I G U R E 1 The binding and neutralization of chimeric antibodies. (A) An enzyme-linked immunosorbent assay (ELISA) analysis of the binding between receptor-binding domain (RBD) and chimeric antibodies mhC3, mhC11, and mhA-1F. (B) Pseudovirus-based neutralization assay of the chimeric antibodies, mhC3, mhC11, and mhA-1F. Data were obtained from three separate experiments and shown as the mean ± SD, and the IC 50 value was obtained via nonlinear regression. Supporting Information : Figures 2 and 3) . Through sequence analysis and substitution of noncritical amino acids, humanization of the three antibody strains was successfully completed, and the degree of humanization was above 96% (Supporting Information: Table 1 ). Humanized VH and VL were also amplified and cloned into pTSE- FreeStyle™293-F cells to express humanized IgG1 SFC3, SFC11, and HSA-1F. Compared with human-mouse chimeric mAbs, the humanized mAbs were able to bind to the RBD protein with similar binding activity by ELISA (Figure 2A) . To further analyze the binding ability of antibodies, the determination of KD was performed by biolayer interferometry. The results are shown in Supporting Information: Figure 4 , the binding of antibodies to SARS-CoV-2 RBD showed remarkable binding kinetics. Compared with the other two antibodies, SFC11 decreased the most gently in the dissociation phase (Supporting Information: Figure 4B ). Apparently, considering the KD of the three antibodies, SFC11 bound to RBD much more strongly than SFC3 and HSA-1F (Table 1) . Further, a PBNA assay was performed to assess the neutralization ability of the antibodies. While the humanized mAbs also inhibited pseudovirus infection, the neutralization ability was slightly lower compared with the human-mouse chimeric mAbs. In addition, SFC3, SFC11, and HSA-1F could neutralize SARS-CoV-2 pseudovirus infection with 50% neutralization at 8.193, 20.81, and 9.638 nM ( Figure 2B ). Since the S1 protein-bonded cells express the viral receptor ACE2 through RBD, we investigated whether the neutralizing antibodies can inhibit RBD and ACE2 binding. Figure 3 shows that while SFC3 and HSA-1F were able to block the binding between RBD and ACE2, the SFC11 neutralizing antibody could not. This finding indicates that SFC11 may has a different epitope. Furthermore, a competitive binding assay was performed to confirm if any of the three mAbs competitively bound RBD. As shown in Figure 4A , even when the binding between SFC3 and RBD reached saturation, SFC11 and HSA-1F still combine with RBD to increase the signal. Moreover, the reverse was also observed ( Figure 4B,C) . These results reveal that three antibodies did not bind to RBD competitively. Interestingly, we found that the binding of SFC11 to RBD was reduced when SFC3 was former immobilized in biosensors compared to SFC11 being immobilized as former. We speculated that the reason may be that the binding of SFC3 to RBD would cause a certain steric hindrance to the binding of SFC11. Previously, Tian et al. divided the neutralizing antibody sites on the RBD into four main regions (Supporting Information: Figure 5 ). 11 To further confirm the binding region to RBD, we performed competitive binding between CB6 12 (Class 1), REGN10933 13 (Class 1), P2B-2F6 7 (Class 2), REGN10987 13 (Class 3), S2A4 14 (Class 4), and MW05 15 with our antibodies (Supporting Information: Figure 6 ). The summary of the results was shown in Figure 4D . SFC3 only had no competition with S2A4 (Class 4). We speculated that SFC3 had partial epitope overlap with these reference antibodies, or that SFC3 had strong steric hindrance, as in Figure 4B . It was very clear that HSA-1F only competed with antibodies in Class 1, and its binding site to RBD would be like with antibodies in Class 1. Interestingly, there was no competitive binding between antibody SFC11 and antibody of Class 1-Class 4, but only with antibody MW05. 15 These suggested that the binding site of SFC11 to RBD may partially overlap with the MW05 but has no overlapping epitope with the reference antibodies in Classes 1-4. In conclusion, the three Since the three antibodies targeted different regions of the RBD, we next tested whether the combined antibodies (1:1 ratio) could enhance their neutralizing activity with a PBNA. The results shown in Figure 5A show that mhA-1F combined with mhC3 or mhC11 could enhance the neutralization activity compared with Figure 1B . However, mhC3 combined with mhC11 could not enhance the neutralization activity. For humanized mAbs, synergistic neutralizing activity was also observed. As shown in Figure 5B , HSA-1F combined with SFC3 or SFC11 could both enhance the neutralization activity, especially when HSA-1F was combined with SFC3, compared with Figure 2B . However, the combination neutralization activity of three antibodies was much lower than HSA-1F combined with SFC3, So, the combination of three antibodies was not considered in the following test. We speculate that mhA-1F and HSA-1F may play an important role in the synergistic neutralization activity. To elucidate the inhibitory capacity of our anti-RBD antibodies against SARS-CoV-2, virus micro-neutralization assay was performed to test the ability of binding and neutralizing the SARS-CoV-2 epidemic strain (Wuhan-hu-1 strain). As shown in Figure 6A However, there were also several special neutralizing mAbs, such F I G U R E 3 SFC3 and HSA-1F blocked the binding between RBD and angiotensin-converting enzyme 2 (ACE2). Competitive binding assay between RBD and mAbs. In the legend box, the samples separated by slash "/" respectively represent "Loading" and "Association" stages of sample molecules. IgG, immunoglobulin G; mAbs, monoclonal antibodies; RBD, receptor-binding domain. F I G U R E 4 (See caption on next page) GUO ET AL. | 7 as the 4A8 targeted N-terminal domain of the S protein (NTD). 18 In the present study, C3 and A-1F both bound to RBD and blocked the binding of RBD to ACE2. Although C11 could not block the binding of RBD to ACE2, C11 still displayed good neutralizing activity. It has been reported that there are four groups of epitopes associated with RBD. 6 (C) HSA-1F was immobilized on the biosensor. In the legend box, the samples were separated by slash "/" respectively to represent "Loading," "Association," and Reassociation stages of sample molecules. (D) was the identification of the binding regions to RBD of three mAbs by a BLIbased cross-competition assay. "√" means that the first antibody bound RBD competitively with the second antibody and "x" means that the first antibody was not competitive with the second antibody. mAbs, monoclonal antibodies; RBD, receptor-binding domain. Moreover, viral mutations may lead to a reduction in the neutralizing activity of mAbs; however, bispecific antibodies may inhibit this decline. Thus, the development of bispecific antibodies using C3, C11, and A-1F may have significance for future SARS-CoV-2 therapeutics. In this study, we identified several humanized neutralizing antibodies specific to SARS-CoV-2 via the ability to bind RBD of the SARS-CoV-2 S protein. Three of these humanized antibodies exhibited strong binding and neutralization activity through different RBD regions. Moreover, these antibodies exhibit synergistic neutralizing activity. Antibody therapies for the prevention and treatment of viral infections Effectiveness of convalescent plasma therapy in severe COVID-19 patients Adjunctive treatment with high-titre convalescent plasma in severely and critically ill COVID-19 patients-a safe but futile intervention. A comparative cohort study Treatment of 5 critically Ill patients with COVID-19 with convalescent plasma Convalescent plasma treatment of critically ill intensive care COVID-19 patients SARS-CoV-2 neutralizing antibody structures inform therapeutic strategies Human neutralizing antibodies elicited by SARS-CoV-2 infection A human monoclonal antibody blocking SARS-CoV-2 infection Identification of human single-domain antibodies against SARS-CoV-2 A murine monoclonal antibody with potent neutralization ability against human adenovirus 7 Distinct BCR repertoires elicited by SARS-CoV-2 RBD and S vaccinations in mice A human neutralizing antibody targets the receptor-binding site of SARS-CoV-2 Studies in humanized mice and convalescent humans yield a SARS-CoV-2 antibody cocktail Mapping neutralizing and immunodominant sites on the SARS-CoV-2 spike receptor-binding domain by structure-guided high-resolution serology Characterization of neutralizing antibody with prophylactic and therapeutic efficacy against SARS-CoV-2 in rhesus monkeys A noncompeting pair of human neutralizing antibodies block COVID-19 virus binding to its receptor ACE2 Structural basis for potent neutralization of betacoronaviruses by single-domain camelid antibodies SARS-CoV-2 cell entry depends on ACE2 and TMPRSS2 and is blocked by a clinically proven protease inhibitor The neutralizing antibody, LY-CoV555, protects against SARS-CoV-2 infection in nonhuman primates Genome-wide analysis of SARS-CoV-2 virus strains circulating worldwide implicates heterogeneity Evaluating the effects of SARS-CoV-2 spike mutation D614G on transmissibility and pathogenicity SARS-CoV-2 D614G variant exhibits efficient replication ex vivo and transmission in vivo SUPPORTING INFORMATION Additional supporting information can be found online in the Supporting Information section at the end of this article. How to cite this article The authors thank Prof. Lihua Hou for kindly providing the plasmid pNL4.3-luc-R − E − . The authors thank Prof. Jun Wu and Bo Liu for kindly providing RBD. This study was funded by the National Key R&D Program of China (2020YFC0841400). The authors declare no conflicts of interest. The data supporting the findings of this study can be obtained from the corresponding authors upon reasonable request. https://orcid.org/0000-0002-0092-5068