Fine mapping and interaction analysis of a linear rabies virus neutralizing epitope

Microbes and Infection 12 (2010) 948e955 www.elsevier.com/locate/micinf

Original article

Fine mapping and interaction analysis of a linear rabies virus neutralizing epitope Kun Cai a, Jian-nan Feng b, Qin Wang a, Tao Li a, Jing Shi a, Xiao-jun Hou a, Xiang Gao a, Hao Liu a, Wei Tu a, Le Xiao a, Hui Wang a,* a

State Key Laboratory of Pathogens and Biosecurity, Beijing Institute of Microbiology and Epidemiology, No. 20 Dongdajie, Fengtai District, Beijing 100071, China b Beijing Institute of Basic Medical Science, No. 27 Taipinglu, Haidian District, Beijing 100850, China Received 18 March 2010; accepted 17 June 2010 Available online 25 June 2010

Abstract A novel human antibody AR16, targeting the G5 linear epitope of rabies virus glycoprotein (RVG) was shown to have promising antivirus potency. Using AR16, the minimal binding region within G5 was identified as HDFR (residues 261e264), with key residues HDF (residues 261e263) identified by alanine replacement scanning. The key HDF was highly conserved within phylogroup I Lyssaviruses but not those in phylogroup II. Using computer-aided docking and interaction models, not only the key residues (Asp30, Asp31, Tyr32, Trp53, Asn54, Glu99, Ile101, and Trp166) of AR16 that participated in the interaction with G5 were identified, the van der Waals forces that mediated the epitopeeantibody interaction were also revealed. Seven out of eight presumed key residues (Asp30, Asp31, Tyr32, Trp53, Asn54, Glu99, and Ile101) were located at the variable regions of AR16 heavy chains. A novel mAb cocktail containing AR16 and CR57, has the potential to recognize non-overlapping, non-competing epitopes, and neutralize a broad range of rabies virus. Ó 2010 Institut Pasteur. Published by Elsevier Masson SAS. All rights reserved. Keywords: Rabies virus; G5-HDF; Fine mapping; Modeling; Interaction

1. Introduction Rabies caused by the rabies virus (RV) affects the central nervous system, causing encephalopathy, and is considered to be virtually 100% fatal once symptoms are evident [1]. RV belongs to the family Rhabdoviridae and genus Lyssavirus. The viral genome is a non-segmented negative-strand RNA that produces five monocistronic mRNAs encoding the

Abbreviations: RV, rabies virus; RVG, rabies virus glycoprotein; BG, bacterial ghost; ELISA, enyzme-linked immunosorbent assay; PBS, phosphate-buffered saline; HRP, horseradish peroxidase; PBS, phosphate-buffered saline; mAb, monoclonal antibody; FAVN, fluorescent antibody virusneutralizing test; Ig, immunoglobulin; DMEM, Dulbecco’s modified Eagle medium; FBS, fetal bovine serum. * Corresponding author. Tel.: þ86 10 66948532. E-mail address: [email protected] (H. Wang).

nucleoprotein, phosphoprotein, matrix protein, glycoprotein, and viral RNA-dependent RNA polymerase [2]. The glycoprotein is capable of inducing and binding virus-neutralizing antibodies, which confers immunity against a lethal infection challenge with the virus [3,4]. According to World Health Organization (WHO) guidelines, category 3 exposures to rabies, which are defined as either single or multiple transdermal bites or contamination of mucous membranes with saliva of a rabid animal, require rabies postexposure prophylaxis (PEP) [5]. Rabies PEP includes administration of both vaccine and anti-rabies immunoglobulin (RIG) [6]. Although human rabies Ig (HRIG) is exclusively used in developed countries, it has inherent potential health risks. HRIG can display batch-to-batch variation and may be of limited availability in cases of sudden mass exposures [7]. Equine anti-rabies immunoglobulin (ERIG) is produced in developing countries but its quality falls

1286-4579/$ - see front matter Ó 2010 Institut Pasteur. Published by Elsevier Masson SAS. All rights reserved. doi:10.1016/j.micinf.2010.06.005

K. Cai et al. / Microbes and Infection 12 (2010) 948e955

far short of the requirement, and it can produce adverse effects such as anaphylactic shock [5]. The need to replace these hyperimmune serum preparations is widely recognized [8]. Human monoclonal antibodies (mAbs) have been shown to protect rodents from a lethal RV challenge [9e11], and the mAb that recognizes the G5 epitope represents one of strategies against RV [12]. G5 is a linear epitope within the glycoprotein of rabies virus [3]. The G5 epitope is a highly conserved region of RVG, indicating that synthetic peptides encompassing this region may induce production of a broadly reactive virus-neutralizing antibody in immunized animals and human beings [3]. In our previous study, a novel human single chain variable fragment (ScFv) of mAb (clone AR16) was obtained from a phage display library with a repertoire of approximately 108 specificities [13], and this antibody recognized the G5 epitope of RV specifically. To overcome the instability of ScFv, a novel isomer-3 domain disulfide-stabilized antibody fragment (3d-dsFv) which had a 6xHis tag was developed, and it possessed higher affinity and stability. The refolded 3d-dsFv was able to specifically neutralize the RV CVS-11 strain [14]. In the present study, we identified the minimal binding region of linear epitope G5, using the 3d-dsFv product of mAb clone AR16 (referred to as AR16 hereafter) and Pepscan technology. Subsequently, different residues were introduced into synthetic peptides that mimicked the epitope and these were tested for loss of mAb binding. The binding ability between AR16 and the mutant epitope was analyzed in vitro, and bioinformatic analysis was used to reveal the nature of the interaction between AR16 and the variants of the G5 epitope. Finally, we adopted a novel monoclonal antibody cocktail which included AR16 and the human mAb CR57 [9,15] to analyze its additive neutralizing effect against rabies in vitro. 2. Materials and methods 2.1. Cells, viruses and antibodies BSR cells were provided by the State Key Laboratory of Pathogens and Biosecurity, Beijing, China. BSR cells were grown with 5% CO2 at 37  C in Dulbecco’s modified Eagle medium (DMEM, Gibco) supplemented with 10% fetal bovine serum (FBS). The CVS-11 strain of RV was gifted by Xianzhu Xia from Changchun, China. A monolayer of BSR cells was infected with CVS-11 for 1 h with 5% CO2 at 37  C. Fresh DMEM supplemented with 2% FBS was added and cells were incubated for another 72 h. Culture supernatants were harvested and stored at 80  C. Anti-RVG mAb CR57 was provided by North China Pharmaceutical Group Corporation [9,15,16]. The 3d-dsFv product of clone AR16 was named as AR16 in this article [14]. 2.2. Preparation and detection of anti-RVG AR16 antibodies The AR16 mAb was prepared and purified conveniently by its 6xHis tag on the C terminus as described previously [14].

949

Renaturation was performed in vitro, purified by nickel-nitrilotriacetic acid (Ni-NTA) column (HisTrap, Amersham), and stored at 80  C. Subsequently, the neutralizing titers of AR16 were detected by the rapid fluorescent focus inhibition test (RFFIT) as described previously [14]. 2.3. Analysis of key residues and amino acids within G5 linear epitope Six 8-mer overlapping peptides were synthesized (DQLVNLHD, LVNLHDFR, NLHDFRSD, HDFRSDEI, FRSDEIEH, and SDEIEHLV), and solved in 0.01 mM phosphate-buffered saline (PBS, pH 7.0) and coated (1 mg/ml) in 96-well plate overnight. The plate was then blocked with 3% bovine serum albumin (BSA) for 1 h at 37  C and washed with 0.1% PBS-Tween20 (PBS-T). AR16 protein was added at a concentration of 0.1 mg/ml into each well, and incubated overnight at 4  C. After washing with PBS-T, HRP conjugated anti-His antibody (1:5000, Pierce) was added (secondary antibody) and incubated for 1 h at 4  C. For color development, 3,30 ,5,50 -tetramethyl benzidine dihydrochloride (TMB2HCl, Sigma) was used after washing with PBS-T. The reaction was stopped with 2 M H2SO4 and read under 450 nM. The 18-mer peptide DQLVNLHDFRSDEIEHLV within RVG was used as positive control, and a non-specific 8-mer peptide was used as negative control (non-specific peptide for short in figure). Alanine replacement scanning of the key peptide was performed by ELISA subsequently. Four 8-mer peptides (with alanine replacement in different sites of the key peptide) were synthesized and coated (1 mg/ml) overnight in a 96-well plate at 4  C. AR16 was used as the primary antibody, and the HRP conjugated anti-His antibody (1:5000, Pierce) was used as the secondary antibody. The 8-mer peptide LVNLHDFR was used as positive control, and the non-specific 8-mer peptide was used as negative control (non-specific peptide for short in figure). 2.4. Key epitope alignment The minimal binding region of the AR16-specific epitope on RVG was aligned using glycoprotein amino acid sequences of the 220 rabies virus isolates and 7 Lyssavirus genotypes (1e7) (data from Genbank). 2.5. Analysis of AR16 binding to variants of the G5 epitope of RVG ELISA was used to determine binding of AR16 (containing a 6xHis tag) to two 8-mer synthesized mutant peptides, LVNLHDFH and LVNLHDFN. The peptides were coated in the wells of a 96-well plate. AR16 was used as the primary antibody, and HRP conjugated anti-His antibody (1:5000, Pierce) was used as the secondary antibody. The 8-mer peptide LVNLHDFR was used as a positive control, and a non-related 12-mer peptide was used as a negative control.

950

K. Cai et al. / Microbes and Infection 12 (2010) 948e955

2.6. Modeling the structure of AR16 bound to the key epitope All computer simulations were performed using Insight II (2005) software. The AR16 homology model was created by finding the closest homolog (using NCBI BLAST) for the light and heavy Fv regions (PDB database) and then creating the complementary determinant region (CDR) loops from canonical classes using a sequence-based algorithm with empirical rules. The peptide model was optimized by mechanics to achieve a stable conformation at ordinary temperatures. The spatial structure of AR16 and the G5 peptides was established and optimized with molecular mechanism and molecular dynamics under a consistent valence force field (CVFF). Steep breakdown (<0.05 kcal/ mol) and conjugate gradient (<0.01 kcal/mol) were used as standard to optimize the structure of the peptide. Molecular docking and interaction were used to predict the antigeneantibody interaction model from its unbound monomer components. Considering the solution accessibility and surface electrostatic potential distribution of the two models, the best interaction model was selected from 50 optimized models. 2.7. Competition ELISA The competition between the two human mAbs, CR57 and AR16, for binding to the G5 epitope was determined by ELISA. RV (CVS-11 strain) antigen was coated on 96-well plates. In preliminary experiments, 2 IU/ml CR57 showed saturated binding to rabies virus. In this experiment, CR57 (final concentration: 2 IU/ml) was selected to incubate with serially diluted AR16 (0, 0.5, 1, 1.5, 2, 2.5, 3 IU/ml) for 2 h at 4  C, as primary antibody. HRP conjugated anti-His antibody (1:5000, Pierce) was used as secondary antibody for the detection of AR16. Anti-human IgG F(c) mAb conjugated with HRP (1:3000, Sigma) was used as secondary antibody for the detection of CR57. Serially diluted AR16 proteins without CR57 were used as control.

2.9. Statistical analysis Statistical significance was determined by the Student’s t test or the chi-square test. A P value of <0.05 was considered significant. 3. Results 3.1. Prepared AR16 antibody could neutralize RV The preparation and detection of AR16 were described previously [14]. AR16 was successfully expressed as inclusion bodies in a prokaryotic expression system. Its expression level was approximately 53% of the total cellular proteins (Total Lab 2.01 software). After denaturization/renaturation and purification, the purity of the AR16 obtained was >90%. The neutralization titer of AR16 against RV was determined as 85 IU/mg. 3.2. Minimal neutralizing linear epitope within G5 of rabies virus was identified as “G5-HDF” using AR16 To identify the core amino acids within the binding region of G5 [3], an 8-mer overlapping peptide ELISA was performed. AR16 showed significant reactivity (P < 0.01) with three peptides: LVNLHDFR, NLHDFRSD, and HDFRSDEI. This implied a minimal binding region: HDFR (aa 261e264), within the RVG (Fig. 1A). Subsequently, to more accurately define the critical residues within the confirmed minimal binding region, we performed alanine replacement scanning through the LVNLHDFR peptide to identify the key residues within the core peptide that were crucial for binding of the AR16. The mutations of aa 261 (H to A)/262 (D to A)/263 (F to A) within the peptide inhibited binding of the peptide to AR16 (P > 0.05), and the mutation of aa 264 (R to A, LVNLHDFA) exhibited the same amount of binding (P > 0.05) as the maternal peptide LVNLHDFR (Fig. 1B). Thus, the central HDF (aa 261e263) triplet was shown to be the minimal key residues within G5, and the novel epitope containing the core HDF residues was named G5-HDF.

2.8. Detection of virus-neutralizing ability The neutralizing ability of the AR16/CR57 cocktail was detected by a fluorescent antibody virus-neutralizing test (FAVN) [17]. CR57 (final concentration: 1 IU/ml) was incubated with serially diluted AR16 (0, 0.5, 1, 1.5, 2, 2.5, 3 IU/ml); 100 TCID50 CVS-11 was added and incubated at 4  C for 2 h. These mixtures were added to BSR cell monolayers in 96-well plates and cultured at 37  C for 48 h. The cells were fixed with acetone at 4  C and dried. Mouse anti-rabies virus nucleoprotein (RVN) mAb conjugated with FITC (1:5000, Chemicon) was added and incubated for another 1 h at 4  C. Surplus FITC was removed thoroughly, and the fluorescent focus units (FFUs) were counted by fluorescent microscopy. CR57 (1 IU/ml) or serial concentrations of AR16 (0, 0.5, 1, 1.5, 2, 2.5, 3 IU/ml) incubated with 100 TCID50 CVS-11, were used as controls.

3.3. The key residues HDF (aa 261e263) within glycoprotein of Lyssaviruses of phylogroup I were conserved To explore the potential range of RV that could be neutralized by AR16, we analyzed the conservation of the target protein. The minimal binding region of RVG sequences of 220 genotype 1 rabies virus isolates (samples included human isolates, bat isolates, and isolates from canines or from domestic animals most likely bitten by rabid canines) were aligned to assess the conservation of the epitope (Table 1). Frequency analysis of the amino acids at each position within the minimal binding region revealed that the critical residues constituting the epitope were highly conserved (data not shown). Position 2 (D262) was completely conserved; the amino acids at positions 1 (H261) and 3 (F263) were

K. Cai et al. / Microbes and Infection 12 (2010) 948e955

A

analyzed. The key residues HDF (aa 261e263) were conserved within genotypes 1, 4, 5, 6, and 7 which belonged to phylogroup I, but not conserved within genotypes 2 and 3 which belonged to phylogroup II [18]. Residue at 264 was confirmed not to be conserved. This finding indicated that the residues at position 264 of G5-HDF might be used to explore the range of protection by AR16 antibodies.

Non-specific peptide DQLVNLHD

#

Peptide

LVNLHDFR

*

NLHDFRSD

*

HDFRSDEI

*

FRSDEIEH

#

SDEIEHLV

#

18-mer peptide

* 0.0

.2

.3 OD450

.4

.5

.6

Non-specific peptide

Peptide

B

.1

LVNLADFR

#

LVNLHAFR

#

LVNLHDAR

#

LVNLHDFA

*

LVNLHDFR

* 0.0

.1

.2 OD450

951

.3

Fig. 1. Analysis of binding ability between AR16 and the linear epitope within RVG. 18-mer peptide DQLVNLHDFRSDEIEHLV within RVG was used as positive control, and non-specific 8-mer peptide was used as negative control (non-specific peptide for short). (A) AR16 was tested using an 8-mer peptide PEPSCAN-ELISA in the region of RVG’s ectodomain identified using overlapping 18-mer peptides. (B) Alanine replacement scan through the core 4-mer peptide containing the minimal binding region as identified in panel A. * P < 0.01 vs non-specific 8-mer peptide; # P > 0.05 vs non-specific 8-mer peptide.

conserved in 99.55% of the isolates, while only 1 out of 220 isolates observed the mutation: H261 to P261, or F263 to R263. A mutation (R to H) in 44.5% of the 220 isolates was observed at Position 4, revealing that this position was not conserved. Subsequently, the minimal binding regions of RVG sequences of Lyssavirus isolates (genotypes 2e7) were

3.4. R264 within G5-HDF had a minimal effect on the recognition of AR16 by rabies virus Approximately 44.5% tested rabies virus strains harbored the mutation R264 to H264 within the synthetic G5-HDF of RVG (Table 1). To determine the role of residue 264 in the recognition of AR16 by RV, and explore the protection range of AR16, a peptide-antigen ELISA was performed. The results showed that both of the peptides LVNLHDFH and LVNLHDFN could bind AR16 (P < 0.01) at the same level as the peptide LVNLHDFR (P > 0.05) (Fig. 2). These results suggested that AR16 has the potential to neutralize a broader range of RV effectively. 3.5. Eight residues within CDRs of AR16 were important for interaction with rabies virus Computer simulation was adopted to model the structure of the AR16-G5 epitope peptide complex, and to deduce a structural explanation for the neutralization of RV by AR16. The models of AR16 (Fig. 3A) and epitope peptide (Fig. 3B) were established successfully. AR16 contained framework regions (FRs, green) and 3 CDRs in heavy chain (CDR1, 31e35 aa; CDR2, 50e60 aa; CDR3, 99e104) and light chain (CDR1, 23e34 aa; CDR2, 50e56 aa; CDR3, 89e97) respectively. Key residues within the peptide were predicted, including His261, Asp262, Phe263 and Ser265 (purple) (Fig. 3C). The protocol captured the van der Waals contribution of interface residues to binding. In addition, charged residues within the antibody (gray) contacted charged residues within the peptide (purple) across the interface, and these

Table 1 Conservation of the AR16 minimal binding region within Lyssaviruses.a Genotype

Rabies

1

Mokola Lagos bat Duvenhage European bat 1 European bat 2 Australian bat

2 3 4 5 6 7

a

Key epitope (Percentageb) HDFR (55.5%) HDFH (43.6%) HDLH (0.45%) PDFH (0.45%) HNDR (100%) HNNR (100%) HDFH (100%) HDFH (100%) HDFH (100%) HDFH (76.5%) HDFN (23.5%)

The numbers of isolates were listed as follows: Rabies, 220; Mokola, 6; Lagos bat, 19; Duvenhage, 6; European bat 1, 14; European bat 2, 12; Australian bat, 17. b Percentage of occurrence of each key epitope is shown in parentheses.

Non-specific peptide

Peptide

Lyssavirus

LVNLHDFN

*

LVNLHDFH

#

*

LVNLHDFR

*

0.0

.2

.4

.6

.8

1.0

OD450

Fig. 2. Binding analysis between mutant peptide and AR16 in vitro. Synthesized peptides were coated respectively, AR16 was used as first antibody, and HRP conjugated anti-His antibody was used as second antibody for ELISA test. Non-specific 8-mer peptide (non-specific peptide for short) was used as negative control. * P < 0.01 vs non-specific 8-mer peptide; # P > 0.05.

952

K. Cai et al. / Microbes and Infection 12 (2010) 948e955

Fig. 3. The structure modeling of AR16 bound to the G5-HDF peptide. (A) The structure modeling of AR16. AR16 contained framework regions (FRs, green) and 3 CDRs in heavy chain and light chain respectively, including: VL-CDR1 (red), VL-CDR2 (orange), VL-CDR3 (gray), VH-CDR1 (blue), VH-CDR2 (purple), VH-CDR3 (yellow); (B) Modeling the structure of G5-HDF peptide; (C) Modeling the docking between AR16 and G5-HDF peptide. Coloring scheme: FRs of AR16 (yellow), CDRs of AR16 (red), key residues within AR16 (blue), G5-HDF peptide (green), key residues within G5-HDF (purple); (D) Modeling the interaction between AR16 and G5-HDF peptide. The hydrogen bonds were indicated with dash lines. Coloring scheme: FRs of AR16 (yellow), CDRs of AR16 (red), key residues within AR16 (white), G5-HDF peptide (green), key residues within G5-HDF (purple). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

residueeresidue contacts lead to several interaction possibilities, including: Asp31 (VH-CDR1):Phe263 (3.76  A), Tyr32 (VH-CDR1):Phe263 (6.03  A), Trp53 (VH-CDR2):His261 (6.69  A) and Trp166 (VL-CDR1):Ser265 (4.78  A). Asp30 (VH-CDR1)/Asn54 (VH-CDR2)/Glu99 (VH-CDR3)/Ile101 (VH-CDR3) within AR16, and Asp262 within the peptide also participated in the interaction through van der Waals force (Fig. 3D).

concentrations of 2.5 IU/ml; no increase in OD450 was observed at AR16 concentrations higher than 2.5 IU/ml. A similar tendency was monitored in the AR16 group, and there was no statistical difference in OD450 between the AR16 and AR16 þ CR57 (P > 0.05) when added to RV (Fig. 4B).

3.6. AR16 and CR57 recognized non-overlapping, non-competing epitopes of RVG

To attain the necessary levels of safety, as well as to replicate the protective activities of HRIG, a cocktail consisting of several mAbs is necessary. To further investigate the combined neutralizing effect of the novel CR57/AR16 cocktail, we performed the FAVN assay. The neutralization rate of 1 IU/ml CR57 was approximately 40%. Mixed CR57/AR16 provided a higher neutralization rate than either CR57 or AR16 used alone (P < 0.01). AR16 was able to neutralize RV completely at a concentration of 2.5 IU/ml, and the mixture of CR57/AR16 was able to neutralize RV completely when the concentration of AR16 was 2.0 IU/ml (Fig. 5), suggesting that addition of CR57 to AR16 could elevate its neutralizing capacity.

CR57, which could recognize distinct antigenic site I [9,15], was different from AR16 which could recognize antigenic site VI. To investigate whether CR57 and AR16 compete or interfere with each other for binding to rabies virus, a competition ELISA was performed. The OD450 was stable in each well of mixed AR16 and CR57 proteins (P > 0.05) when anti-human IgG F(c) mAb was used to detect CR57 (Fig. 4A). When the HRP conjugated anti-His antibody was used to detect AR16, the OD450 increased with the concentration increase of AR16 within the cocktail (P > 0.05) up to

3.7. Novel CR57/AR16 cocktail has enhance neutralization activity for RV

K. Cai et al. / Microbes and Infection 12 (2010) 948e955

953

A 2.0 AR16 AR16+CR57(2IU/ml)

Inhibition % of infection

1.6

OD450

1.2 **

.8 .4

0.0

.5

1.0

1.5

2.0

2.5

3.0

3.5

1.0 #

OD450

*

60 * 40 20

AR16 AR16+CR57(1IU/ml) 0.0

.5

1.0

1.5

2.0

2.5

3.0

3.5

Fig. 5. Neutralisation rate detection of mixture CR57/AR16. The detection was performed by FVAN on BSR cells. CR57 (final concentration: 1 IU/ml) was incubated with serial concentration AR16 (0, 0.5, 1, 1.5, 2, 2.5, 3 IU/ml), 100 TCID50 CVS-11 was added and incubated at 4  C for 2 h, respectively. These mixtures were added to the BSR cell monolayer in 96-wells plate. Mouse antiRV N mAb conjugated with FITC (1:5000) was used as detection antibodies. CR57 (1 IU/ml) or serial concentration AR16 (0, 0.5, 1, 1.5, 2, 2.5, 3 IU/ml) incubated with 100 TCID50 CVS-11 was used as control. The fluorescent focus units were counted by fluorescent microscopy. * P < 0.01 vs AR16 at the same concentration; # P > 0.05 vs AR16 at the same concentration.

*

*

*

.8

#

Concentration of AR16 (IU/ml)

Concentration of AR16 (IU/ml) 1.2

#

*

80

0

0.0

B

*

100

* .6 * .4 *

.2

AR16 AR16+CR57(2IU/ml)

*

0.0

0.0

.5

1.0

1.5

2.0

2.5

3.0

3.5

Concentration of AR16 (IU/ml) Fig. 4. Competition analysis between AR16 and CR57. The activity was detected by competition ELISA. RV (CVS-11 strain) was coated in 96-wells plate as antigen. CR57 proteins (final concentration: 2 IU/ml) were incubated with serial diluted AR16 (0, 0.5, 1, 1.5, 2, 2.5, 3 IU/ml) at 4  C for 2 h, and these mixtures were used as first antibodies. Serial diluted AR16 proteins without CR57 were used as control. (A) The detection of CR57. Anti-human IgG F(c) mAb conjugated with HRP (1:3000) was used as second antibody; (B) The detection of AR16. HRP conjugated anti-His antibody (1:5000) was used as second antibody. * P > 0.05 vs AR16 at the same concentration; ** P > 0.05 between one and each other; # P > 0.05.

4. Discussion The potency of a rabies vaccine was dependent upon the dose of glycoprotein within the vaccine, whether in human or animals [8,19]. Thus, glycoprotein was a target molecule for designing anti-rabies therapeutic drugs. The antigenic structure of RVG was initially defined by Lafon et al. [20]. Since then, the antigenic sites had been mapped by identification of amino acid mutations in the glycoprotein of mAb-resistant variants [21e25]. The conformational antigenic site I was initially defined by only one mAb, 509-6, located at residue 231 [20,21]. Marissen et al. [15] redefined antigenic site I as a neutralizing epitope complex harboring both conformational and linear epitopes located at positions 226e231. Antigenic site II was a discontinuous conformational epitope comprising residues 34e42 and residues 198e200 [21,23]. Antigenic site III was a continuous conformational epitope at residues 330e338 and harbored two charged residues, K330 and R333,

which played a role in viral pathogenicity [23e25]. Antigenic site IV and V had been described on the rabies virus glycoprotein of ERA strain [20,26,27]. Dietzschold et al. [3] described a unique linear epitope G5, a portion of the glycoprotein flanked by the amino acids from positions 254e275 located at antigenic site VI (264 aa). G5 was found on the ectodomain using the mAbs that bound to the denatured RVG [28], and dogs receiving rabies vaccination had developed antibody responses directed to this linear epitope [29]. A second site was identified by Ni et al. [30] between aa 249 and 268. Benmansour et al. [21] defined the linear epitope G1 (G, Gif), and also described the presence of a minor site (site a) located at positions 342e343, which was distinct from antigenic site III despite its close proximity [31]. A novel, linear B-cell epitope had been identified at the N terminus of the RVG between aa 14 and 19 [32]. In our previous studies, phage display technology was used to produce a ScFv from a human mAb against the G5 epitope, and the isomer from the same clone, AR16, was produced to optimize its stability [14]. Here, by using overlapping peptides that mimicked the G5 linear epitope, the minimal binding region HDFR (aa 261e264) was mapped precisely, and the amino acids essential for binding were determined for the first time. The central triplet of residues (H261, D262, and F263) was recognized by the novel AR16 antibody, and the core epitope containing the key residues HDF was named G5-HDF. These results were consistent with the computer docking model. Residue R264 had minimal involvement in the binding between AR16 and epitope. The reason for the difference between the importance of the residue R264 in the epitope G5 shown by us and that shown by van der Heijiden et al. [33] may due to the different affinity of antibodies, and it also

954

K. Cai et al. / Microbes and Infection 12 (2010) 948e955

may be related to the different donator of antibodies, because the virus-neutralizing antibody (VNA) induced in different animals (mice, dog, goat and human etc.) is usually different, and the different endosomatic environment (pH etc.) of different donator may effect the recognition site between antibody and epitope [34]. The confirmation of the residue R264 should be enforced by cocrystallization in the future. Recently, the overall structure of vesicular stomatitis virus glycoprotein (VSVG) was explored by Roche et al. [35,36]. Both of the VSV and RV belonged to rhabdovirus, and the conformation of RVG and the location of G5-HDF within RVG could be predicted and alignmented in future according to the similar conformational structure of VSVG. AR16 was possible to neutralize approximately 99% of rabies virus strains, including the strains harboring the mutation R264 to H264 within RVG (approximately 44.5% of RV stains), which could not be neutralized by 6-15C4 mAb [33]. The binding of AR16 to the 8-mer peptide LVNLHDFH supported our assumption. The central triplet residues HDF were also conserved in Lyssavirus isolates of genotypes 1, 4, 5, 6, and 7, which belonged to the same phylogenetic classification of Lyssaviruses [18]. We proposed that AR16 would be able to neutralize phylogroup I Lyssaviruses (genotypes 1, 4, 5, 6 and 7) but not those of phylogroup II (genotypes 2 and 3). The neutralizing ability of AR16 against Lyssaviruses isolates should be tested in vivo in future studies. Any anti-RV antibody cocktail should possess suitable neutralizing potency and non-overlapping epitope recognition [30,37]. The novel G5-HDF epitope for binding AR16 was a linear epitope located at antigenic site VI of RVG, and there was no epitope within antigenic sites I, II, and III that overlapped with the G5-HDF linear epitope, which made it one of the most potential cocktail components. The enhancement effect of neutralization was validated between the AR16 and CR57, which was one of the most potent human mAbs specific for antigenic site I [9,15,16]. The higher neutralizing titer provided by the novel combination mAb cocktail (including CR57 and AR16) revealed that the cocktail, in recognizing both non-overlapping and non-competing epitopes, could neutralize a broad range of RV. A cocktail of CR57 (against antigenic sites I) and CR4098 (against antigenic sites III) human mAbs has been developed [37]. Lafon et al. demonstrated an addition of a third anti-RVG mAb specific for antigenic site II would not improve the neutralizing ability because some mAbs that neutralize antigenic sites II or III competed with each other for binding of the glycoprotein [23e25]. However, Benmansour et al. [21] demonstrated that most of MAbs neutralize G protein recognized site II or III (nearly 97%), especially 2/3 of them were occupied by MAb which recognized site II. Therefore, the importance of MAb that neutralzed site II should not be underestimated at a clinical use as cocktails of polyclonal antibodies in therapeutics. Recently, the effectiveness of murine MAb cocktails was reported by Mu¨ller et al. [38], and the possibility of the cocktails of murine MAbs containing two major neutralizing epitopes (II, III) in PEP was recommended. So, we suggested that a combination mAb cocktail that included all four non-

overlapping, non-competing epitopes (IeIII, G5) could potentially be used to replace HRIG to provide complete protection, and the further studies will be needed. The effect of van der Waals forces between residueeresidue contacts was presumed to be the major molecular action between AR16 and the G5-HDF epitope in this study. The essential residues of the antibody were predicted for the first time: 7 of 8 presumed key residues (Asp30, Asp31, Tyr32, Trp53, Asn54, Glu99, and Ile101) within AR16 located at the variable regions of the heavy chain. The structure optimization of AR16 should reserve these residues to retain its binding activity. Only one residue (Trp166) located at the variable region of the light chain. Although the hypothetical interaction between W166 and the G5-HDF epitope was not essential, as demonstrated by our group and van der Heijiden et al. [33], but we supposed that some key residues located within the light chain might be beneficial to the conformation and stabilization of the antibody. The confirmation of interaction should be considered in further research based on the cocrystallization structure. Moreover, the construction of whole antibody or macromolecule antibody (bivalent or tetravalent miniantibody) based on AR16 could be produced in eukaryotic expression system in future. Acknowledgments We are grateful to Academician Xian-zhu Xia in Changchun Institute of Veterinary for the rabies virus CVS-11 strain. References [1] J.A. Wilkerson, Rabies update, Wilderness Environ. Med. 11 (2000) 31e39. [2] L.G. Schneider, H. Diringer, Structure and molecular biology of rabies virus, Curr. Top. Microbiol. Immunol. 75 (1976) 153e180. [3] B. Dietzschold, M. Gore, D. Marchadier, H.S. Niu, H.M. Bunschoten, L. Otvos Jr., W.H. Wunner, H.C. Ertl, A.D. Osterhaus, H. Koprowski, Structural and immunological characterization of a linear virus-neutralizing epitope of the rabies virus glycoprotein and its possible use in a synthetic vaccine, J. Virol. 64 (1990) 3804e3809. [4] T. Mebatsion, M.J. Schnell, K.K. Conzelmann, Mokola virus glycoprotein and chimeric proteins can replace rabies virus glycoprotein in the rescue of infectious defective rabies virus particles, J. Virol. 69 (1995) 1444e1451. [5] Human rabies prevention e United States, 1999. Recommendations of the Advisory Committee on Immunization Practices (ACIP), MMWR. Recomm. Rep. 48 (1999) 1e21. [6] K. Ray, M.J. Embleton, B.L. Jailkhani, M.K. Bhan, R. Kumar, Selection of single chain variable fragments (scFv) against the glycoprotein antigen of the rabies virus from a human synthetic scFv phage display library and their fusion with the Fc region of human IgG1, Clin. Exp. Immunol. 125 (2001) 94e101. [7] Centers for Disease Control and Prevention (CDC), Mass treatment of humans who drank unpasteurized milk from rabid cows e Massachusetts, 1996e1998, MMWR. Morb. Mortal. Wkly. Rep. 48 (1999) 228e229. [8] WHO Consultation on a Monoclonal Antibody Cocktail for Rabies Post Exposure Treatment. World Health Organization, Geneva, Switzerland, 2002. [9] B. Dietzschold, M. Gore, P. Casali, Y. Ueki, C.E. Rupprecht, A.L. Notkins, H. Koprowski, Biological characterization of human monoclonal antibodies to rabies virus, J. Virol. 64 (1990) 3087e3090. [10] K. Enssle, R. Kurrle, R. Ko¨hler, H. Mu¨ller, E.J. Kanzy, J. Hilfenhaus, F.R. Seiler, A rabies-specific human monoclonal antibody that protects mice against lethal rabies, Hybridoma 10 (1991) 547e556.

K. Cai et al. / Microbes and Infection 12 (2010) 948e955 [11] C.A. Hanlon, C.A. DeMattos, C.C. DeMattos, M. Niezgoda, D.C. Hooper, H. Koprowski, A. Notkins, C.E. Rupprecht, Experimental utility of rabies virus-neutralizing human monoclonal antibodies in post-exposure prophylaxis, Vaccine 19 (2001) 3834e3842. [12] H. Bunschoten, M. Gore, I.J. Claassen, F.G. Uytdehaag, B. Dietzschold, W.H. Wunner, A.D. Osterhaus, Characterization of a new virus-neutralizing epitope that denotes a sequential determinant on the rabies virus glycoprotein, J. Gen. Virol. 70 (1989) 291e298. [13] X.L. Zhao, J. Yin, W.Q. Chen, M. Jiang, G. Yang, Z.H. Yang, Generation and characterization of human monoclonal antibodies to G5, a linear neutralization epitope on glycoprotein of rabies virus, by phage display technology, Microbiol. Immunol. 52 (2008) 89e93. [14] K. Cai, H. Wang, S. Bao, J. Shi, X. Hou, X. Gao, H. Liu, J. Yin, Novel human 3-domain disulfide-stabilized antibody fragment against glycoprotein of rabies virus, Microbes. Infect. 10 (2008) 548e555. [15] W.E. Marissen, R.A. Kramer, A. Rice, W.C. Weldon, M. Niezgoda, M. Faber, J.W. Slootstra, R.H. Meloen, M. Clijsters-van der Horst, T. J. Visser, M. Jongeneelen, S. Thijsse, M. Throsby, J. de Kruif, C.E. Rupprecht, B. Dietzschold, J. Goudsmit, A.B. Bakker, Novel rabies virus-neutralizing epitope recognized by human monoclonal antibody: fine mapping and escape mutant analysis, J. Virol. 79 (2005) 4672e4678. [16] M. Prosniak, M. Faber, C.A. Hanlon, C.E. Rupprecht, D.C. Hooper, B. Dietzschold, Development of a cocktail of recombinant-expressed human rabies virus-neutralizing monoclonal antibodies for postexposure prophylaxis of rabies, J. Infect. Dis. 188 (2003) 53e56. [17] F. Cliquet, M. Aubert, L. Sagne´, Development of a fluorescent antibody virus neutralisation test (FAVN test) for the quantitation of rabies-neutralising antibody, J. Immunol. Methods 212 (1998) 79e87. [18] H. Badrane, C. Bahloul, P. Perrin, N. Tordo, Evidence of two Lyssavirus phylogroups with distinct pathogenicity and immunogenicity, J. Virol. 75 (2001) 3268e3276. [19] T.J. Wiktor, R.I. Macfarlan, K.J. Reagan, B. Dietzschold, P.J. Curtis, W.H. Wunner, M.P. Kieny, R. Lathe, J.P. Lecocq, M. Mackett, Protection from rabies by a vaccinia virus recombinant containing the rabies virus glycoprotein gene, Proc. Natl. Acad. Sci. U.S.A. 81 (1984) 7194e7198. [20] M. Lafon, T.J. Wiktor, R.I. Macfarlan, Antigenic sites on the CVS rabies virus glycoprotein: analysis with monoclonal antibodies, J. Gen. Virol. 64 (1983) 843e851. [21] A. Benmansour, H. Leblois, P. Coulon, C. Tuffereau, Y. Gaudin, A. Flamand, F. Lafay, Antigenicity of rabies virus glycoprotein, J. Virol. 65 (1991) 4198e4203. [22] C. Prehaud, P. Coulon, F. LaFay, C. Thiers, A. Flamand, Antigenic site II of the rabies virus glycoprotein: structure and role in viral virulence, J. Virol. 62 (1988) 1e7. [23] I. Seif, P. Coulon, P.E. Rollin, A. Flamand, Rabies virulence: effect on pathogenicity and sequence characterization of rabies virus mutations affecting antigenic site III of the glycoprotein, J. Virol. 53 (1985) 926e934.

955

[24] P. Coulon, J.P. Ternaux, A. Flamand, C. Tuffereau, An avirulent mutant of rabies virus is unable to infect motoneurons in vivo and in vitro, J. Virol. 72 (1998) 273e278. [25] B. Dietzschold, W.H. Wunner, T.J. Wiktor, A.D. Lopes, M. Lafon, C.L. Smith, H. Koprowski, Characterization of an antigenic determinant of the glycoprotein that correlates with pathogenicity of rabies virus, Proc. Natl. Acad. Sci. U.S.A. 80 (1983) 70e74. [26] M. Lafon, J. Ideler, W.H. Wunner, Investigation of the antigenic structure of rabies virus glycoprotein by monoclonal antibodies, Dev. Biol. Stand. 57 (1984) 219e225. [27] A. Flamand, H. Raux, Y. Gaudin, R.W. Ruigrok, Mechanisms of rabies virus neutralization, Virology 194 (1993) 302e313. [28] M. Houimel, K. Dellagi, Peptide mimotopes of rabies virus glycoprotein with immunogenic activity, Vaccine 27 (2009) 4648e4655. [29] N. Johnson, K.L. Mansfield, A.R. Fooks, Canine vaccine recipients recognize an immunodominant region of the rabies virus glycoprotein, J. Gen. Virol. 83 (2002) 2663e2669. [30] Y. Ni, Y. Tominaga, Y. Honda, K. Morimoto, S. Sakamoto, A. Kawai, Mapping and characterization of a sequential epitope on the rabies virus glycoprotein which is recognized by a neutralizing monoclonal antibody, RG719, Microbiol. Immunol. 39 (1995) 693e702. [31] T.R. Luo, N. Minamoto, H. Ito, H. Goto, S. Hiraga, N. Ito, M. Sugiyama, T. Kinjo, A virus-neutralizing epitope on the glycoprotein of rabies virus that contains Trp251 is a linear epitope, Virus Res. 51 (1997) 35e41. [32] K.L. Mansfiled, N. Johnson, A.R. Fooks, Identification of a conserved linear epitope at the N terminus of the rabies virus glycoprotein, J. Gen. Virol. 85 (2004) 3279e3283. [33] R.W. van der Heijden, J.P. Langedijk, J. Groen, F.G. UytdeHaag, R.H. Meloen, A.D. Osterhaus, Structural and functional studies on a unique linear neutralizing antigenic site (G5) of the rabies virus glycoprotein, J. Gen. Virol. 74 (1993) 1539e1545. [34] P.J. Kankanamge, T. Irie, K. Mannen, T.S. Tochikura, A. Kawai, Mapping of the low pH-sensitive conformational epitope of rabies virus glycoprotein recognized by a monoclonal antibody #1-30-44, Microbiol. Immunol. 47 (2003) 507e519. [35] S. Roche, S. Bressanelli, F.A. Rey, Y. Gaudin, Crystal structure of the low-pH form of the vesicular stomatitis virus glycoprotein G, Science 313 (2006) 187e191. [36] S. Roche, F.A. Rey, Y. Gaudin, S. Bressanelli, Structure of the prefusion form of the vesicular stomatitis virus glycoprotein G, Science 315 (2007) 843e848. [37] A.B. Bakker, W.E. Marissen, R.A. Kramer, A.B. Rice, W.C. Weldon, M. Niezgoda, C.A. Hanlon, S. Thijsse, H.H. Backus, J. de Kruif, B. Dietzschold, C.E. Rupprecht, J. Goudsmit, Novel human monoclonal antibody combination effectively neutralizing natural rabies virus variants and individual in vitro escape mutants, J. Virol. 79 (2005) 9062e9068. [38] T. Mu¨ller, B. Dietzschold, H. Ertl, A.R. Fooks, C. Freuling, C. FehlnerGardiner, J. Kliemt, F.X. Meslin, R. Franka, C.E. Rupprecht, N. Tordo, A.I. Wanderler, M.P. Kieny, Development of a mouse monoclonal antibody cocktail for post-exposure rabies prophylaxis in humans, PloS. Negl. Trop. Dis. 3 (2009) e542.