Development of Human Monoclonal Antibodies With Broad Reactivity for Rabies Postexposure Prophylaxis

ABSTRACT

Rabies is an acute lethal disease causing by the neurotropic virus rabies virus (RABV). Rabies immune globulin (RIG) as an indispensable component of rabies postexposure prophylaxis (PEP) always faces with great challenges in terms of costs, stability and safety. Our objective is to develop a novel and potential fully human monoclonal antibodies (mAbs) cocktail for the improvement of rabies PEP. The neutralizing fully human mAbs were screened by using fully humanized antibody mice (CAMouse^HG). Then, two mAbs 26–12 G and 5–7 G were selected with potential neutralizing activity to RABV by using fluorescent antibody virus neutralization test (FAVN), which specifically bind to antigenic sites I and III of RABV‐glycoprotein (RABV‐G), the key amino acid residues were further identified in position 336, 337 of 5–7 G and 226, 227, 228 of 26–12 G by using cross‐linking and mass‐spectrometry. Both mAbs are highly conserved across 8 RABV strains (distributing in 3 lineages: Asian, Cosmopolitan and Arctic‐related) and 1 IRKV strain, and showed high neutralizing potential. Moreover, the in vivo experiment demonstrated that our cocktail can protect Kunming mice from a lethal RABV challenge. Collectively, we generate two noncompeting fully human mAbs (26–12 G, 5–7 G) and obtained cocktail CAM001 with their mixture. The high‐potency and broad‐spectrum neutralization of the cocktail supports its utility in human rabies PEP as an efficacious and affordable alternative to RIG products, particularly in endemic areas.

1. Introduction

Human rabies, caused by RABV, is an acute progressive viral encephalomyelitis with almost 100% fatality following the onset of symptoms and 59 000 people are estimated to die each year, which mainly occurs in Africa and Asia, and 40% are in children [1]. Although rabies can be controlled in the animal reservoirs through mass vaccination and prevented through the appropriate prophylactic treatment in humans exposed to the virus. Approximately 15 million people per year in China are treated after exposure to bites or scratches of suspected RABV‐infected animals [2]. Upon the virus exposure, timely and proper PEP is necessary to effectively prevent onset of rabies. According to the guide of World Health Organization (WHO), standard rabies PEP consists of wound care, concurrent administration of rabies vaccine and RIGs to prevent the exposed patients, particularly those with category III exposures, from contracting the disease [3].

Traditional RIGs are polyclonal human rabies immune globulins (HRIGs) or equine rabies immune globulins (ERIGs), which are generated respectively from human and horse hyperimmune sera of rabies vaccination. Despite being highly effective, RIG products are still criticized for their immunogenicity, safety, sustainability, batch‐to‐batch variation and high cost [4]. WHO has been encouraging the development of RABV‐G mAbs to replace traditional RIGs for PEP [5]. G protein is the only viral antigen presented on the surface of rabies virions and is the principal target for hosts to develop neutralizing humoral immune response, thereby it is of paramount importance to identify neutralizing monoclonal antibodies that are able to recognize G proteins of multiple RABV lineages as much as possible [6]. To this end, mouse and human mAbs have been developed in last decade, and two of them have been approved for marketing, including SII RMAb (a human monoclonal antibody, mAb) for neutralizing AS III, produced by Mass Biologics and Serum Institute of India) [7] and NM57(a single mAb for AS I, produced by North China Pharmaceutical Company of China) [8, 9]. Compared with RIGs, mAbs have several advantages such as greater neutralization breadth, lower dosage, higher safety, lower production costs, ease of mass production, and consistency between batches [10].

As a guide to design of future biologics, WHO recommends developing mAb cocktail targeting two or more non overlapping AS to improve the efficacy and breadth of RABV neutralization and minimize the risk of failure due to escape mutants from 2018 [11]. Up to date, four mAbs cocktail are approved for marketing or in advanced clinical trials, among which, Rabimabs is composed of two mice antibodies M777‐16‐3 and 62‐71‐3 (produced by Zydus Cadila in India, Available in the Indian market since 2020) [12], SYN023 is a cocktail of two modified humanized antibodies CTB011 and CTB012 (produced by Synermore Biologics of China) [13], GR1801 is of two fully human bispecific antibody C11m and C34m (produced by Genrix Biopharmaceutical of China) [14], and CL184 is of two fully human mAbs CR57 and CR4098, which produced by Crucell [15], while the development was recently halted due to the lack of broad RABV coverage [11]. Currently, there is no available fully human antibody cocktail for RABV, thereby, it is still urgently needed to develop mAbs and cocktails, which would provide more reserves or options for antirabies PEP and reduce the risk of PEP failure.

In this study, we screened and characterized a pair of antirabies fully human mAbs 26‐12 G and 5‐7 G from CAMouse^HG, these two mAbs exhibit picomolar or nanomolar affinities towards G protein with different antigenic epitopes, outstanding potency and breadth of neutralization against RABVs. Moreover, the two mAbs' cocktail is capable to neutralize 8 RABV strains and 1 IRKV strain (belong to two species of lyssaviruses) [16]. The structure and sequence analysis further revealed that the identified key residues in binding are highly conserved across multiple species of lyssaviruses. Importantly, our cocktail CAM001 affords protection equal or lower to a standard dose of the HRIG at 20 IU/kg in PEP of a lethal rabies mice models. Taken together, the high‐potency and broad‐spectrum neutralization of our cocktail CAM001 is a promising candidate for next‐generation PEP of exposed humans.

2. Methods

2.1. Cell Lines and Viruses

Mouse Neuro‐2a (N2a) and Baby hamster kidney (BHK‐21) cells were cultured in Dulbecco's modified eagle medium (DMEM) (Corning, 10‐092‐CVRC) and minimum essential medium (MEM) (Corning, 10‐010‐CV) respectively, supplemented with 10% fetal bovine serum (Gibco, 10099141), 100 U/mL penicillin G, and 100 μg/mL streptomycin. All cells were routinely cultured at 37°C in 5% CO₂. RABV strains SRV9 (attenuated strain, AAM18963.2), SHJDD14 (OR122986.1), ZJZSD1601 (MG383893.1), NeiMeng927A (EU284095.2), NMALSF01 (MW411451.1), GDMM D16 (MW411450.1), HuND02 (KT221115.1), HuNPB01 (GU186386.1), CVS‐11 (fixed strain, EU126641.1) and IRKV strain IRKV‐TH China 12(JX442979.1) [17] were propagated in BHK‐21 and stored at −80°C. The eGFP‐SRV9 is a recombinant SRV9 expressing eGFP (enhanced green fluorescent protein), previously developed by the reverse genetic system [18]. Virus titers were determined as TCID₅₀ per milliliter by direct immunofluorescence assay (DFA) in N2a cells.

2.2. Generation of mAbs From CAMouse^HG Mice

The CAMouse^HG was developed by using a new humanized antibody transgenic mouse strategy, particularly in IgH transgene, we deployed mouse IgH 5'‐enhancer and all IgM expression sequences and its regulatory elements among human VDJ sequences and human gamma constant regions and LCR to constitute a functional BCR structure and perform normal signal transductions in vivo, so the mature B‐cells can switch in full human IgG‐expressing and secreting plasma cells. In combination breeding with human Igκ and human Igλ transgenes, mouse endogenous IgH and Igκ knockouts, finally we established a new transgenic mouse line containing IgH, Igκ and Igλ transgenes with endogenous IgH and Igκ inactivated, called CAMouse (China Antibody Mouse) [19, 20, 21]. Five CAMouse^HG mice were subjected subcutaneous with RABV G protein [22] 50 μg/mouse at for points simultaneously, RABV‐G (GenBank: ADJ29911.1), residues 1‐439, replacing residues 73‐79 and 117‐125 with the Gly‐Gly‐Ser‐Gly‐Gly linker, respectively, was individually sub‐cloned into the pFastBac1 vector) in complete Freund’ adjuvant (Sigma, F5881) with 10 μg of CpG (Invivogen, tlrl‐1826) and 1% (v/v) Alhydrogel (Invivogen, vac‐alu‐250), then successively boosted with G protein in incomplete Freund's adjuvant plus CpG and Alhydrogel once every 2 weeks and repeated twice [23]. After 14 days of the third immunization, the mice were boosted with lively attenuated SRV9 (100 μL of 10^7.5 TCID₅₀/mL) in the right gastrocnemius muscle, once 2 weeks for total two times. Following the last booster immunity at day 66 and serum neutralization antibody check, the spleen cells were prepared and fused with SP2/0 cells at a ratio of 1:3 by electrofusion (BTX, ECM2001+). The fused cells were cultured in ClonaCell‐HY Medium D (STEMCELL Technologies, 03804) at concentration of 3 × 10⁵ cells/mL for 10 days. The growing cell clones were transferred to 96‐well plates, and the supernatant of each clone was collected after 3–4 days culture for the further analysis [24].

2.3. Indirect Immunofluorescence Assay (IFA)

To identify the reactivity of obtained mAbs, BHK‐21 cells were seeded into 96‐well plates and infected with CVS‐11 according to the routine procedures. The cells were fixed after incubation for 72 h, and the virus was detected with the generated hybridoma cell supernatant as primary antibodies and FITC‐labeled goat anti‐human IgG (abcam, ab97224) as the secondary antibody. In parallel, commercial FITC‐labeled anti‐RABV mAb conjugate (FUJIREBIO, 800‐092) was used as the positive control [25].

2.4. Viral Neutralization Test

To observe the neutralizing activity of anti‐RABV mAbs, 100 μL of hybridoma cell supernatant was added into the 96‐well plates, then, the eGFP‐SRV9 (100 TCID₅₀/50 μL) was added to each well and the plate was further incubated for 1 h. Next, 50 μL of BHK‐21 cell suspension were added to each well at a final density of 4 × 10⁵ cells/mL. After incubation at 37°C for 48 h, the fluorescence was examined under a microscope, and the hybridoma cells able to completely neutralize the virus were selected. Standard FAVN of WOAH was further carried out to test the neutralizing activity of mAbs. Briefly, 50 µL of sample (serum, hybridoma cell supernatant or mAbs), the reference positive (0.5 IU/mL) and negative controls (WOAH Reference Laboratory for Rabies, China) were distributed into four consecutive wells of a 96‐well plate and serially diluted. Then, CVS‐11 (200 TCID₅₀/50 μL) was incubated for 1 h, and BHK‐21 cell suspension were incubated for further 48 h. After fixing and incubating with FITC‐labeled anti‐RABV mAb conjugate (FUJIREBIO, 800‐092) for 30 min, the plates were examined under a fluorescence microscope and the titers were presented in IU/mL by comparing the results to that of positive standard [26]. The modified fluorescent antibody virus neutralization (mFAVN) assay was performed to test the neutralizing breadth of the mAbs, in which, the CVS‐11 was respectively substituted with the other 7 RABV strains and 1 IRKV strain to confirm that our mAbs can neutralize RABV strains with different genetic diversities.

2.5. Expression of Human mAbs

Double‐antibody sandwich ELISA was used to identify isotypes of the mAb light chains. Briefly, 1000‐fold diluted anti‐human lambda light chain antibody and mouse anti‐human kappa light chain antibody were used for ELISA coating (abcam, ab99832, ab79115). The anti‐human lambda‐ or kappa‐light chains peroxidase antibodies (abcam, ab99808, ab99811) were used for detection of primary antibodies. Total RNA from hybridoma cells was extracted using RNAprep Pure Cell/Bacteria Kit (TIANGEN, DP430) and reverse‐transcribed into cDNA (TaKaRa, 2641Q) according to the manufacturer's instructions. The forward mixed and reverse primers were listed in Table S1, and the PCR products were analyzed and sequenced. The recombinant human IgG1 mAbs were synthesized using DNA fragments encoding heavy‐ and light‐chain genes (Tsingke, TSE102) and cloned into modified pcDNA3.4 vector respectively, which contains human constant regions of IgG1 or light chains. The pairs of heavy‐ and light‐chain expression vectors were transiently cotransfected into HEK293F cells by using EZ cell transfection reagent (Life iLab, AC04L092). After 7 days incubation, the cell supernatants were harvested and purified using HiTrap MabSelect PrismA (Cytiva, 17549854) [20].

2.6. Western Blot Analysis

Purified RABV‐G [22] was electrophoresed on SDS‐PAGE gel under both reduced and non‐reduced conditions, then transferred to nitrocellulose (NC) membrane followed by blocking for 1 h with 5% fat‐free milk and incubation with our mAbs at 0.5 μg/mL overnight at 4°C. After washing the NC membrane was incubated with DyLight 680‐labeled anti‐human IgG antibody (Invitrogen, SA5‐10114) for 1 h at room temperature (RT). The membrane was washed and the bands were scanned by Odyssey Infrared Imaging System (LI‐COR Biosciences, CLX).

2.7. Competitive ELISA

Two mAbs NM57 and CR4098 (both under clinical development), and all our selected 19 mAbs were labeled with Horseradish Peroxidase (HRP) according to the HRP conjugation kit (Sangon Biotech, D601047), and tested by ELISA in a matrix competition assay. ELISA plates were coated with RABV‐G at 0.5 µg/mL for overnight at 4°C, then blocked with 5% BSA (Bovine Serum Albumin) for 1 h at RT, incubated with 19 unlabeled mAbs at a concentration of 25 μg/mL, followed by the addition of a suitable concentration of HRP antibodies for 1 h. The plates were washed and developed with 3,3’,5,5’‐tetramethylbenzidine (TMB) (Solarbio, PR1200) as a substrate. The reaction was stopped with 2.5 M H₂SO₄, and the absorbance of each well was monitored at 450 nm.

2.8. Antibody Affinity Assay

The Octet QKE system was used to analyze the affinity of our mAbs to G protein based on BLI technology (SARTORIUS, ForteBio). All samples were diluted in kinetic buffer (1% BSA in phosphate buffered solution, PBS), and the buffer was served as negative control. The anti‐human IgG Fc biosensor (SARTORIUS, 18‐5060) was used for detection after 20 min of hydration in the kinetic buffer. The mAbs at 10 mg/L was immobilized on the sensor surface as the loaded antigen. The threefold diluted G protein were used for association and dissociation with the mAbs, and the parameters were measured by the changes in concentration of molecules at the interface of the sensor and media.

2.9. Antigenic Site Analysis

Chemical crosslinking and Mass Spectrometry were performed to analyze the AS of mAbs 26‐12 G and 5‐7 G. In brief, samples were prepared in conjugation buffer with a 10‐fold kDa concentration of RABV‐G and two mAbs (26‐12 G and 5‐7 G) in excess. Prepare DSS (disuccinimidyl suberate) in DMSO. 10‐fold molar excess of cross‐linker was added to samples and incubated for 30 min, then quench buffer was added and incubates at RT for 15 min. The Micellar enzymatic hydrolysis includes: cut the gel after SDS‐PAGE electrophoresis, wash with ultrapure water, decolorize with NH₄HCO₃ and ACN (acetonitrile) until the gel particles are completely white, and dry them in vacuum. Adding TCEP (trichloroethyl phosphate) to reduce the disulfide bond, and CAA (cyanoacetamide) to make it alkylated. Dehydrate again until colloids are completely dry. Digest the colloids with trypsin solution at 37°C, extract the colloids and evaporate the samples to dryness with a vacuum concentrator [27]. Sample desalting: The above product was acidified with TFA (Trifluoroacetic Acid), centrifuged to remove impurities, and desalted with C18 StageTips. After lyophilization, the sample was dissolved with 0.1% FA (Fatty Acid), and the protein concentration was determined by Bicinchoninic acid. Liquid chromotography with mass spectrometry identification and data processing: The on‐board time of each sample was 90 min. Cross‐linked peptides were identified using pLink2 software with a mass accuracy of 10 ppm for precursor ions and 20 ppm for fragment ions, and the results were screened by applying a 5% FDR (False Discovery Rate) threshold at the spectral level.

2.10. Statistics Analysis

Statistical analysis was performed using GraphPad Prism software (version 9.02). Data analysis was performed using Data Analysis v11.0 (ForteBio) and Data Analysis HT v11.0 software (ForteBio). The MEGA 7 program package was used to construct the phylogenetic trees using the neighbor‐joining method with 1000 bootstrap replicates. Statistical significance was defined as *p < 0.05, **p < 0.01. Cross‐linked mass spectrometry data was analyzed using the SIM‐XL program (1.5.7.1). The three‐dimensional structure model of protein complex was constructed by Amber modeling software (v.14). The protein interaction relationship was analyzed using LigPlot software (v.2.2).

2.11. Postexposure Prophylaxis in Mice

One hundred and ten Kunming mice were challenged at Day 0 by the intramuscular route (gastrocnemius muscle in right hind limbs) with 50 μL CVS‐11 (10^7.5 TCID₅₀/mL) [28]. Animals were randomly assigned to 11 study groups and given mAbs treatments or PBS by the end of Day 0. For five groups (10 mice per group), 1/10 human doses of rabies vaccine (a commercial vaccine for human use, freeze‐dried, strain PV2061(CDBIO), potency ≥ 2.5 International Units (IU) was administered intramuscularly (i.m.) in left hind limbs at Days 0, 3, 7, 14, and 28 postinfection (p.i.) alone or along with the HRIG (20 IU/kg, equivalent to 20 mg/kg) or mAbs CAM001 of an equimolar mixture of 26‐12 G and 5‐7 G (0.03, 0.06, and 0.12 mg/kg) administering i.m. in right hind limbs in a final volume of 50 μL at day 0. For the other six groups, which were not vaccinated, but only injected PBS, HRIG (20 IU/kg), negative mAb (without virus neutralizing activity) or our mAb CAM001 (0.03, 0.06, and 0.1 mg/kg). The challenged animals were observed twice a day and promptly euthanized at the onset of one of clinical signs of rabies (i.e., motoric deficit, lack of coordination, paresis, paralysis and, sensory dullness). The brain tissues were collected to confirm RABV infection by fluorescent antibody test (FAT), a standard method of WOAH guidance [29].

3. Results

3.1. Generation and Identification of Full Human Anti RABV‐G Hybridoma

To obtain the fully human monoclonal antibodies (mAbs) capable of neutralizing RABV, five CAMouse^HG mice were immunized and the titers of neutralizing antibodies were evaluated as described in Figure 1A. The spleen cells were isolated from immuned mice with higher neutralizing activity and fused with myeloma cells to obtain hybridoma cells by electrofusion. The fused cells were screened and cultured by semi‐solid medium to generate monoclonal hybridoma clones (Days 10–12). We selected 196 hybridoma cells that can bind to RABV by using IFA assay, the representative hybridoma cells with different binding capability was showed in Figure 1B. Then, the recombinant eGFP‐SRV9 was used to examine the virus‐neutralizing activity of these 196 hybridoma cells. As indicated in Figure 1C, 56 hybridoma cells were selected, which supernatant could completely neutralize RABV. After the further FAVN testing, 26 clones were finally obtained which secret the better antibodies with neutralizing activity higher than 0.5 IU/mL (Figure 1D).

Figure 1.

Discovery of fully humanized monoclonal antibodies against RABV. (A) Sequential immunization strategy of the humanized‐antibody transgenic CAMouse^HG. (B) Indirect immunofluorescence assay (IFA) screening of the culture supernatant of 5760 hybridoma cell lines. The different fluorescence intensity in representative samples indicated the different binding capability of the supernatant to RABV. (C) eGFP‐SRV9 virus neutralization test screening of the culture supernatant of 196 hybridoma cell lines. The heatmap is normalized, and the red indicated the hybridoma supernatant is 100% neutralizing of the virus. D: Detection of the neutralizing activity of the supernatant of 56 hybridoma cells by fluorescent antibody virus neutralization assay (FAVN).

3.2. Expression and Identification of the Recombinant Monoclonal Antibodies (mAbs)

Identification of light chain types by double antibody sandwich ELISA in Figure S1 showed that all of above 26 selected antibody clones were kappa(κ)‐light chain. The sequencing data indicated that 27‐12 F and 25‐6 C, 1‐4B and 22‐5H shared the same sequences. The heavy‐ and light‐chain of total 24 hybridomas were cloned and expressed in HEK‐293F cells by using human IgG1 isotype backbone as vector [30]. The FAVN results demonstrated that 19 of 24 recombinant mAbs had neutralization activity (Figure 2A).

Figure 2.

Expression and Identification of Recombinant Monoclonal Antibodies (mAbs). (A) Recombinant mAbs FAVN results. All antibodies were diluted to 0.1 mg/mL, the polyclonal human immunoglobulins (HRIG, CTBB, 100 IU/mL) was diluted to 20 IU/mL. (B) Western blot analyses of the binding of the indicated antibodies to reducing and nonreducing proteins. The recombinant mAbs were boiled within reducing or nonreducing buffer, respectively. Two others human mAbs NM57 and CR4098, which bind to nonreducing protein, were used as positive control. M represents protein marker.

To identify the cognate AS is conformational or liner, the SDS‐PAGE gels were performed with RABV‐G under reducing or nonreducing conditions, and probed with above recombinant mAbs by Western blot. As demonstrated in Figure 2B, 18 mAbs only bind with G protein under nonreducing condition, while 1‐12 G could bind under both nonreducing and reducing conditions. Collectively, 19 recombinant mAbs were obtained with acceptable neutralization activity, and all of which have conformational AS, besides 1‐12 G has both.

3.3. Determination of Antigenic Recognition Sites of mAbs on RABV‐G

Antibody competition studies were performed to determine the conformational AS recognized by the selected neutralizing mAbs. Two reference antibodies NM57 and CR4098, which were previously identified to recognize G protein AS I and III respectively [31, 32], were used as probes to map the specificity of our mAbs. As demonstrated in Figure 3, all 19 tested mAbs were clustered into 2 groups with 1‐12 G, 23‐5 G and 26‐12 G classified as AS I group, according to the reciprocal competitions with NM57, while 9 out of 19 mAbs including 1‐4B, 4‐1 A, 4‐7 C, 8‐2B, 9‐9 C, 10‐8 A, 13‐3D, 15‐2B and 19‐5 C as AS III group, according to the reciprocal competitions with CR4098. In addition, the binding of the other 7 mAbs could be completely blocked by CR4098, including 5‐7 G, 15‐3D, 15‐11B, 17‐1 C, 22‐2 F, 25‐6 C and 30‐9 A, however, these mAbs could not block the binding of CR4098 completely, especially for 17‐1 C, it could not influence the binding of CR4098 at all. Taken into account that when two AS overlap, or even when the areas covered by the arms of the two antibodies' overlap, the competition is almost complete and mutually cross‐competitive, while the weak or one‐way inhibition may reflect a decreased affinity due to steric or allosteric effects [6]. Thus, based on latter explanation, these 7 mAbs could form a third cluster that recognizes a distinct, hereafter dubbed AS III.A, and the more detail of the AS of mAb 5‐7 G would be further analyzed by the chemical crosslinking and mass spectrometry.

Figure 3.

RABV‐G antigenic site mapping using cross‐competition ELISA‐based binding studies. mAbs cross‐competition matrix performed by ELISA on the 19 isolated mAbs and two reference mAbs with known antigenic epitope specificity (NM57 and CR4098). The percentage of binding inhibition of the HRP antibodies (upper row) by the unlabeled antibodies listed in the left column is shown. Results are classified using color shading codes with values ≥ 80% in red, < 80% and ≥ 50% in yellow, and no shading for values < 50%. CTBB: the polyclonal human immunoglobulins, HRIG.

3.4. Identification of Broad‐Spectrum of RABV Neutralizing mAbs

12 out of 19 our mAbs were selected to identify the broad‐spectrum neutralizing efficacy against 8 RABV strains (SHJDD14, ZJZSD1601, NeiMeng927A, NMALSF01, GDMMD16, HuND02, HuNPB01, CVS‐11) and 1 IRKV strain (IRKV‐TH China 12), based on their neutralizing potency and distinct antigenic recognition on RABV‐G. Two mAbs NM57, CR4098, and HRIG were employed as positive control. The FAVN results demonstrated that 3 mAbs which clustered in AS I neutralize all the above 9 strains, among which, 26‐12 G had the best neutralizing effect with the activity ranged from 1215 to 14391 IU/mg on 8 strains, except CVS‐11 with 405 IU/mg (Figure 4A). As a comparison, NM57 (AS I) also neutralized all 9 strains, although the activity was significantly lower than that of 26‐12 G (Figure S2). Meanwhile, the other 9 mAbs clustered in AS III could neutralize the 8 strains except IRKV‐TH China 12, among which, 5‐7 G showed the best neutralizing efficacy with the activity ranged from 1215 to 6313 IU/mg, except the slightly lower on NMALSF (702 IU/mg) (Figure 4A). As a comparison, CR4098 (AS III) can neutralize 8 strains but not IRKV‐TH China 12, and almost all the activity was significantly lower than that of 5‐7 G (Figure S2). Moreover, the efficacy of HRIG was also lower than that of our candidates, although it can neutralize all 9 strains.

Figure 4.

Identification of the broad‐spectrum neutralizing activity and affinity of recombinant mAbs. (A) The neutralizing activity of the selected 12 recombinant mAbs and HRIG (100 IU/mL, CTBB) against 9 RABVs were tested using FAVN assay. All antibodies were diluted to 0.1 mg/mL, HRIG (100 IU/mL, CTBB) was diluted to 20 IU/mL. Results are classified using color shading codes with values ≥ 100 IU/mL in red. < 100 IU/mL and ≥ 50 IU/mL in yellow, and no shading for values < 50 IU/mL, “−” means no neutralization. (B) The neutralizing activity of the diverse combination of we selected mAbs against 9 rabies viruses were tested using FAVN assay. All antibodies were diluted as above. Results are classified using the plus. “+++” represents a neutralizing potency of ≥ 100 IU/mL, “++” represents a neutralizing potency between 50 IU/mL~100 IU/mL, and “+” represents a neutralizing potency of< 50 IU/mL, “−” means no neutralization. (C) Analysis of the affinity of selected recombinant monoclonal antibodies to RABV‐G. RABV‐G was loaded into the HIS 1 K biosensor via a His‐tag. The dynamic dissociation of antigen‐antibody complexes was monitored using a ForteBio macromolecular interaction instrument based on BLI technology. The process of antibody binding to antigen is limited to 600 s.

We then selected our mAbs for cocktail combination according to their broad‐spectrum neutralization titers. Three AS III candidate mAbs 4‐7 C, 5‐7 G, 30‐9 A, and one AS I candidate 26‐12 G were selected. The three AS III mAbs were respectively mixed with 26‐12 G at a 1:1 ratio for FAVN detection. As demonstrated in Figure 4B, the mixture of 26‐12 G and 5‐7 G showed the best neutralization effect as compared with the others, with potential neutralizing potency above 100 IU/mL (1000 IU/mg) for all 9 strains, only with a little lower for NeiMeng927A (92.32 IU/mL, 923.2 IU/mg). As expected, we identified that the antibody with higher neutralizing potency usually dominates when two antibodies with different AS are combined, thereby the cocktail exhibits a higher neutralizing potency, but not the synergistic effects, meanwhile, the cocktail also exhibits the broad‐spectrum of both two included antibodies.

To identify whether the neutralizing activity of mAbs would be benefit from the high affinities towards G protein by physically blocking the initial steps of cellular entry, we performed affinity analysis on 12 candidate mAbs and 2 positive controls (NM57, CR4098), the data indicated that the affinity of 13‐3D and 26‐12 G was best with 1 pM and 1.96 pM respectively (Figure 4C and Figure S3). Intriguingly, there seems no exact correlation between the neutralization and affinity, such as the 1‐12 G, with the high affinity (11.5 pM) while the low neutralizing activity (78 IU/mg). This is in agreement with the previous finding for another RABV antibody screening [15]. Collectively, we identified two novel mAbs 26‐12 G and 5‐7 G, and confirmed their cocktail is with the satisfied neutralizing activity and broad‐spectrum reactivity in vitro.

3.5. Analysis of the Amino Acid Sites Conservation of 26‐12 G and 5‐7 G on Rabv‐G

To further clarify the specific key residues of the two mAbs 26‐12 G and 5‐7 G on G protein and their interaction models (Figure 5). The specific AS residues of the mAbs were defined by using cross‐linking and mass‐spectrometry analysis. As demonstrated in Figure 5A, we firstly analyzed the cross‐linked sites between G protein and our mAbs by using Amber Modeling Software, the results demonstrated that 26‐12 G mainly cross‐linked to position K226 of G protein. After analyzing the hydrogen bond in molecular docking by LigPlot software, we found K226, L227, C228, G229 and L231 would be the key amino acid sites to influence the binding of mAbs. We then mutated these residues to alanine, western‐blot data showed that 26‐12 G failed to bind with G protein when mutated at positions 226, 227 and 228 respectively (Figure 5B), while the mutation at the other two positions has no effects. Thereby, the K226, L227 and C228 on G protein AS I are the key residues in the binding with 26‐12 G. The similar analysis was performed in 5‐7 G, which mainly cross‐linked to position K342 (Figure 5A), and the further analysis revealed that R333, T334, N336, and E337 might play potential roles. After mutated the above residues, 5‐7 G failed to bind with G protein when mutated at positions 336 and 337 residues (Figure 5B), thereby N336 and E337 of G protein are considered as the key residues in the binding with 5‐7 G.

Figure 5.

Identification and visualization of critical amino acid residues for mAbs 26‐12 G and 5‐7 G to RABV‐G. (A) The specific amino acid residues of the mAbs were defined by using cross‐linking and mass‐spectrometry analysis. Amber software modeling was performed to analyze the spatial position relationship of the interaction between G protein and the mAbs, combined with LigPlot software analysis and amino acid point mutation technology, to determine the key amino acid recognition of G protein by antibodies. Green represents the structure of G protein (PDB: 6LGX), Orange represents the mAb 5‐7 G heavy chain, yellow represents the mAb 5‐7 G light chain, dark blue represents the mAb 26‐12 G heavy chain, and light blue represents the mAb 26‐12 G light chain, pink represents AS III, magenta represents AS I. (B) Western blot analysis of the reactivity of CVS‐11‐G protein mutants with mAbs 26‐12 G and 5‐7 G. the mutation sites in mAbs 26‐12 G and 5‐7 G were as indicated. M represents protein marker. (C, D) Level of amino acid residue conservation in AS I and III as calculated by the analysis of the G protein sequences from 3126 RABVs. Pie charts show the detailed distribution of amino acid usage at each critical position of 26‐12 G and 5‐7 G. Conservation analysis of amino acid residues in AS of mAb 26‐12 G (C). Conservation analysis of amino acid residues in AS III of mAb 5‐7 G (D).

To further assess the broad‐spectrum reactivity of our cocktail, we analyzed the conservation of above amino acid sites in totally 3126 sequences from NCBI database, which represent the diversity of RABV isolates worldwide. For mAb 26‐12 G that recognizing at K226, L227, and C228, all three position is highly conserved, among which, K226 is 99.39%, L227 is 99.9%, and C228 is 99.91% (Figure 5C). For 5‐7 G recognizing at N336 and E337, although N336 involves relatively diverse amino acid changes, asparagine still has 94.63% conservation, E337 is very conserved, in which E has 99.68% conservation, D and K account for 0.29% and 0.03%, respectively (Figure 5D). Collectively, the recognition epitope for both 5‐7 G and 26‐12 G is highly conserved in RABV‐G.

3.6. The mAbs Cocktail Protects Mice From a Lethal RABV Infection

To investigate the neutralizing capacity of mAbs in vivo, the postexposure neutralizing capability of the cocktail CAM001 (26‐12 G:5‐7 G = 1:1) was evaluated against a lethal RABV infection in Kunming Mice models. Briefly, the mice (n = 10 per group) were challenged intramuscularly with a lethal dose of CVS‐11(5 LD₅₀), subsequently treated with our cocktail (0.03, 0.06, 0.12 mg/kg) and HRIG (20 IU/kg) alone or in combination with vaccine. As demonstrated in Figure 6, all the 10 mice (100%) that treated with PBS succumbed between Day 11 and 14, as well as the vaccine only group and the non‐neutralizing mAb group. The standard HRIG‐based PEP was effective in reducing the overall mortality with 4 out of 10 animals (40%) survived. Strikingly, the cocktail treated group (0.06 mg/kg and 0.12 mg/kg, equivalent to 1/166 and 1/333 of the administered HRIG) had 60% (6/10) and 70% (7/10) survival rate until 30 days post challenge, respectively, even the lower doses of 0.03 mg/kg protected 30% of the mice (3/10), and this efficacy is equal to a 20 IU/kg HRIG dose. Meanwhile, the similar survival rate was observed in cocktail alone and combined with vaccine groups, the cocktail treated group (0.03 mg/kg, 0.06 mg/kg and 0.12 mg/kg had 30% (3/10), 60% (6/10) and 70% (7/10) survival rate until 30 days post challenge.

Figure 6.

mAbs cocktail protect mice from a lethal RABV infection. Percent survival in Kunming mice infected with RABV CVS‐11 isolate and then left untreated or treated with the standard PEP (HRIG and vaccination) or with three experimental PEP protocols replacing HRIG with different doses of a cocktail of 26‐12 G and 5‐7 G mAbs. Kaplan–Meier survival curves are shown by plotting percent survival against time (in days).

In addition, the serum of treated survival mice was collected to exam whether the mAbs influence the endogenous response of challenged mice to vaccine, and the data showed that the antibody levels were diminished to 0.17–4.5 IU/mL (average 1.5) in cocktail treated alone group at 30 days exposure, while the levels increased to 4.5–364.5 IU/mL (average 90) in mice treated with mAbs and vaccine (Supplementary Table 2). This indicated that our cocktail mAbs had no interference on the endogenous response to vaccine. All these data provide strength evidence about the potency of CAM001 in the PEP models.

4. Discussion

Despite the development of rabies vaccines for over a century, rabies continues to impose a major burden upon public health, especially in the developing countries [33]. PEP provides effectively prevention, while the death is inevitable once PEP fails and the onset of clinical symptoms [34]. Clearly, RIG is indispensable in PEP, the mechanism of RIG action is based on passive antibody administration, which confers immediate protection through the neutralization of RABV at the site of infection, unlike vaccination where the stimulation of protective immunity is delayed. However, among the 20 million people who receive different form of PEP, less than 10% of patients with category III exposures receive RIG due to the high cost and low availability [2]. As strongly advocated by the WHO, mAbs have been proposed as a practical alternative of RIG to fulfill these unmet medical needs [3], which not only abating possible safety concern from applying RIG, like contamination in blood‐derived products, but also being free from animal welfare concern, as does in case or ERIG. Ideally, a cocktail would be the best chance at tackling a broad spectrum of domestic and wildlife and street RABVs, which consisting of at least two non‐competing mAbs that specifically targeting highly conserved G protein residues [35]. Humanized antibody animals provide the principle technological support for the development of fully humanized antibodies. Currently, there are totally 46 FDA licensed fully human antibody drugs having major impact on several diseases, among which, 28 of them are derived from humanized antibody mice. For rabies, there are two fully humanized antibodies up to date, Rabishield (SII RMAb) [7] and Ormutivimab (NM57) [9], while both are mono‐specific mAbs and only recognize one AS. In this study, we used our humanized antibody mice to generate the fully human mAbs, and selected a pair of antirabies mAbs 26‐12 G and 5‐7 G, with high affinity towards G protein, outstanding potency and breadth of neutralization against RABVs. We further identified the both mAbs bind to G protein AS I and AS III, respectively. Moreover, we obtained the cocktail CAM001 which can neutralize all 8 RABV strains and 1 IRKV strain, and actually 5‐7 G fails to neutralize IRKV‐TH China 12, while 26‐12 G can neutralize all 9 strains, especially shows very high neutralizing activity for IRKV‐TH China 12, as expected, the cocktail optimized the broad spectrum of single mAb and the higher neutralizing potency, the both mAbs can complement by each other, but not antagonistic.

Additionally, we found there is no exactly correlation between the neutralization activity and affinity, such as13‐3D has the high affinity (1 pM) but the low neutralizing activity (405 IU/mg), while, our candidate 5‐7 G has the high neutralizing activity (6313 IU/mg) but the lower affinity (1.62 nM). This is in agreement with the previous reports [35] that the neutralization ability of an antibody is determined by the structure and position of the AS, while the neutralization breadth is usually determined by the conserved of the recognition site and residues [36].

Noting the neutralizing antibodies are directed towards the G protein, the conservation of key amino acid residues in G protein is the major factor to influence the cross‐protective of mAbs against lyssavirus. RABV‐G mainly contains five neutralizing antigen sites (AS), AS I and AS III are the most popular ASs in human RABV antibodies [11, 37]. AS I includes both conformational and linear antigenic sites and is represented by a.a. 226‐231 [38]. AS III is a conformational AS containing a.a. 330‐338 [8]. Our analysis identified the key recognized residues 226‐228 for 26‐12 G and 336 ‐337 for 5‐7 G, and the cocktail CAM001 showed neutralizing activity to 9 strains, include 8 RABVs and 1 IRKV strain. To assess the broad‐spectrum capacity of our mAbs to more virus strains, we analyzed the conservation among 3126 strains in all the phylogroup I/II/III lyssaviruses for recognizing by 26‐12 G and 5‐7 G. For mAb 5‐7 G, N336 is polymorphic with 94.63% conservation, while E337 is highly conserved with 99.68% among 3126 stains, this effect is similar to that of Rabishield, which targets the key amino acids 336 in AS III and neutralizes 23 RABVs and 6 Bat‐RABVs, but not Duvenhage virus (belongs to lyssavirus duvenhage) [7]. Considering our 5‐7 G has one more residue 337 besides 336, the similar neutralizing efficacy would at least be speculated in our 5‐7 G, of course this would require more strains data to support. For the other cocktail component 26‐12 G, it targets the highly conserved residues 226‐228 in AS I, with 99.39%, 99.9% and 99.91% respectively among 3126 stains. As compared with the other AS I mAb RVC20, which recognizes the residues 226 and 231, with 99.73% and 67.65% conservation respectively among 2566 RABVs, and was reported neutralized all the phylogroup I lyssavirus, noting residue 231 is a polymorphic residue, thereby 226 plays principal function in broad neutralizing effect [6]. Fortunately, there are three highly conserved residues in our 26‐12 G, thereby, we proposed that 12‐26 G should has the better broad‐spectrum neutralizing capacity. Importantly, our cocktail showed the satisfied antiviral effect in vivo, which significantly optimized the protective effect of vaccine as compared with equivalent dose of HRIG, the survival rate increased to 60% from 40%. The high dose of cocktail even increased the survival to 70%. Similar results were reported at a 50‐fold reduction in the injected dose of CVS‐11, the survival rates for RVC20/RVC58 0.045 mg/kg (equivalent to HRIG 20 IU/kg) and HRIG group were 75% and 67%, respectively [6].

In conclusion, our humanized CAMouse^HG mice provide a valuable platform to obtain fully human reactive and neutralizing antibodies, after immune, hybridoma cells screening, sequencing and recombination expression, we successfully generates two noncompeting mAbs and mixed to cocktail CAM001, which show a potent prophylactic protective efficacy against RABV in vitro and in vivo, suggesting that the CAM001 mAbs cocktail offers a safe and cost‐effective alternative for PEP, and which would bolster the use of rabies PEP, particularly in endemic areas.

Author Contributions

M.W., X.P., W.X., L.F., H.Z., X.L., F.M., S.M., Q.L., and Y.X. conducted the experiments and analyzed the data. M.W., X.Y., C.T, L.G., and Y.L. designed the experiments. M.W., Q.L., C.T., L.G., and Y.L. wrote the paper. C.T., L.G., and Y.L. are corresponding authors. All the authors have read and approved the final manuscript for publication.

Ethics Statement

All mice experiments were approved by the Changchun Veterinary Institute of the Chinese Academy of Agricultural Sciences and the Animal Welfare and Ethics Committee of the Chongqing Academy of Animal Husbandry and laboratory animal‐Guideline for ethical review of animal welfare (GB/T 35892‐2018). All procedures were in accordance with AAALAC standards. Whenever possible, procedures are designed to avoid or minimize discomfort, distress, and pain to animals.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting information.

Supporting information.

Supporting information.

Supporting information.

Supporting information.

Supporting information.

Supporting information.

Data Availability Statement

Data will be made available on request.