Introduction

Hepatitis A virus (HAV) is an epidemiologically important pathogen that causes acute hepatitis in humans [12]. It is a positive-strand RNA virus belonging to the family Picornaviridae, genus Hepatovirus [16].

Picornavirus particles contain 60 protomers arranged as 12 pentamers. Each protomer is composed of four capsid proteins, viral protein (VP) 1, VP2, VP3 and VP4. The HAV particle also contains 60 protomers, made up of the same capsid proteins, but the VP4 of HAV is truncated and smaller than the VP4 proteins found in other picornaviruses [5].

Unlike other picornaviruses [3], the antigenic structure of HAV has not been completely characterized. Its unique features, the difficulty of obtaining a high virus yield in tissue culture and the strict conformational dependence of its own neutralizing antigenic structure have hampered investigations of HAV morphology. Generating neutralizing antibodies in response to individual structural proteins, synthetic peptides or expressed polypeptides is reportedly difficult [7, 10, 15]. Fragmental structural proteins cannot maintain the function of the neutralizing site. Studies using escape mutants have generated information about the antigenic structure of HAV [17, 19] that supports the notion of an immunodominant neutralizing site involving residues of VP1 and VP3, and that a second, potentially independent site involves residue 221 of VP1.

We generated anti-idiotypic antibodies (anti-Id) by immunization with an HAV-specific human neutralizing monoclonal antibody to investigate the antigenic structure of HAV. Immunization with an antibody (Ab1) can induce anti-antibody antibodies (Ab2s), and those specific for idiotopes of Ab1 are considered anti-Ids. Idiotopes are confined to the variable region of immunoglobulin molecules, and they are crucial for antigen recognition [21]. Among the four known types of anti-Ids (Ab2α, Ab2β, Ab2γ and Ab2ε) [2], Ab2β binds directly to the idiotope within the paratope of Ab1. Ab2β has the potential to mimic the structure of the antigenic epitope complementary to the Ab1 paratope and it has an antigen-like function. Several reports have used anti-Ids to identify viral/bacterial antigenic structure or cellular receptors [1, 9].

Here, we describe the properties of our anti-Ids, designated mAb2 94-2 and 94-7, which mimic the HAV surface structure and allowed unique insight into the antigenic structure of HAV.

Materials and methods

Cell line and viruses

The GL37 line was derived from African green monkey kidney cells and established to support the optimal growth of HAV [20]. GL37 cells were cultured in Eagle’s minimal essential medium (Nissui, Tokyo, Japan) supplemented with 50 μg/ml of gentamicin (Biological Industries, Kibbutz Beit Haemek, Israel) and 10% fetal bovine serum (FBS, GIBCO, Invitrogen Corporation, Auckland, NZ).

Human tissue-culture-adapted hepatitis A virus strains KRM003 (genotype IIIB), KRM031 (genotype IA) and TKM005 (genotype IB) were isolated from patients with hepatitis A [18] and propagated in GL37 cells. Each strain was purified from cell extracts by differential centrifugation, chloroform extraction, RNase, DNase and protein K digestion, extraction with a mixture of 2-ethoxyethanol and 2-buthoxyethanol, and gel filtration. After sucrose gradient centrifugation, the fraction containing intact particles was selected.

The HAV strain KRM003 was propagated and purified as described above. Purified virus was inactivated with 0.025% formalin at 37°C for 12 days and then diluted with 10 mM Tris–HCl, 2 mM EDTA, 100 mM NaCl, 0.02% Tween 80, 0.1% NP40, 0.2% BSA, 0.03% NaN3 to 3 mg/ml and stored at 4°C.

Preparation of anti-HAV antibodies

Syngeneic anti-HAV human monoclonal antibodies designated KF6 and KF94 provided by Dr. Yasushi Kuwahara, Denka Seiken Co. Ltd. [4] corresponded to Ab1s derived from the blood of a convalescent patient with hepatitis A who was infected with strain IIIB. Here, the parental antibody was KF94, which cross-reacted with a mouse monoclonal antibody that recognized the HAV immunodominant site (data not shown).

Anti-HAV rabbit hyperimmune serum was prepared by immunization with the inactivated HAV.

Generation of mouse monoclonal anti-KF94 antibodies (mAb2s)

Hybridoma clones were selected from the fusion products of SP2/0 myeloma (Riken Cell Bank, Tsukuba, Japan) and spleen cells of BALB/c mice immunized with KF94.

The affinity of culture fluid from each hybridoma clone to KF94 was tested using an enzyme-linked immunosorbent assay (ELISA) (Table 1). Polystyrene microtiter plates (F96 CERT. MAXISORP, Nunc Laboratories, Roskilde, Denmark) were coated with 50 μl of appropriately diluted KF94, or with anti-HAV-negative human serum obtained from a healthy individual. Sera or specimens derived from human sources were collected after obtaining informed consent from the donors.

Table 1 Summary of ELISA used in this study

Nonspecific binding in the wells was blocked with bovine serum albumin. Culture fluid (50 μl) was applied to plates coated with either KF94 or anti-HAV-negative human serum. Bound antibody was detected by incubation with a horseradish-peroxidase-conjugated (HRPO) anti-mouse IgG (MBL, Nagoya, Japan) followed by o-phenylenediamine (Tokyo Chemical Industry Co, Ltd, Tokyo, Japan) substrate. Thirty minutes later, the reaction was stopped with 2 N H2SO4. Absorbance was measured at 492 nm in an ELISA plate reader (Corona, Hitachinaka, Japan). The procedures for color development and absorbance measurement subsequent to adding conjugated antibody were identical in every subsequent ELISA.

Antibodies that reacted with KF94 and not with anti-HAV-negative human serum were identified as KF94-specific mAb2s. The isotype of the mAb2 was determined using a mouse monoclonal antibody isotyping test kit (Serotec Ltd., Oxford, UK). The ascitic fluid of each mAb2 was obtained by intraperitoneal injection of hybridomas into pristane-primed BALB/c mice. Animals were handled and cared for in accordance with the “Guidelines for Animal Experimentation at NIID.”

Cross-reactivity between the selected mAb2s was determined by competitive inhibition ELISA using biotin-labeled mAb2s and KF94-coated plates (Table 1). Briefly, non-labeled mAb2 was applied to KF94-coated plates, followed by biotin-labeled specific or non-specific mAb2. The mAb2 was labeled with biotin using EZ-Link NHS-Biotin (Pierce, Rockford, IL, USA) according to the manufacturer’s instructions and detected using HRPO avidin (MBL, Nagoya, Japan).

Biotin-labeled mAb2s were also used in mAb2-HAV binding ELISA (Table 1). Serially diluted biotin-labeled mAb2s were applied to plates that were prepared as follows: polystyrene microtiter plates were coated with 50 μl of anti-HAV rabbit hyperimmune serum, and nonspecific binding was blocked with bovine serum albumin. Fifty microliters of formalin-inactivated HAV strain KRM003 (20 ng/ml) was added, and the plates were incubated overnight at 4°C. Bound biotin-labeled mAb2s were detected as described above.

Inhibition assays

We determined the ability of the mAb2 to inhibit the binding of KF94 to HAV (Table 1). A mixture of diluted Ab1 and mAb2 was incubated at 4°C overnight and then applied to HAV-coated plates. Bound Ab1 was detected using a HRPO anti-human IgG (Dako, Glostrup, Denmark).

We examined the ability of inactivated HAV to inhibit the binding of KF94 to mAb2 (Table 1). Serially diluted inactivated HAV was incubated in KF94 plates overnight at 4°C, and then the wells were emptied. Biotin-labeled mAb2 (diluted 1:1,000) was applied to the wells, followed by HRPO avidin. Inhibition rates were measured using the formula:

Inhibition rate (%) = 100 × (absorbance without inhibitor − absorbance with inhibitor)/absorbance without inhibitor.

Anti-HAV-cellular-receptor mouse monoclonal antibodies

Anti-HAV-cellular-receptor mouse monoclonal antibodies (anti-HAV-receptor antibodies), designated 190/4, 235/4 and 263/6, were induced by immunizing GL37 cells and selected by their ability to block HAV propagation in GL37 cells [11]. We confirmed that the anti-HAV receptor antibodies recognized a common receptor. Anti-HAV receptor antibody190/4 was conventionally conjugated with horseradish peroxidase and named 190/4C.

Binding of mAb2 to HAV cellular receptor

The binding of each mAb2 to GL37 cells was examined using an immunofluorescence assay. The positive controls were the anti-HAV-receptor antibodies, 190/4, 235/4 and 263/6, and the negative control was a normal mouse serum (NMS) that was confirmed to be negative for anti-HAV antibody.

Confluent, unfixed GL37 cell monolayers on glass cover slips were gently washed with phosphate-buffered saline (PBS) and then incubated with mAb2 or control antibodies for 1 h at 37°C in a moist chamber. The cover slips were washed with PBS and then incubated with fluorescently labeled anti-mouse IgG (MBL, Nagoya, Japan) for 1 h at 37°C in a moist chamber. The labeled anti-mouse IgG was removed, and the cover slips were washed three times in PBS. The cell surface was immediately observed using a fluorescence microscope.

We confirmed using competitive inhibition ELISA that mAb2 94-7 and anti-HAV receptor antibody 190/4 shared the same HAV cellular receptors.

The anti-HAV receptor antibody 235/4 was the positive competitor, and an anti-human Fc mouse monoclonal antibody (anti-human Fc mAb) was the negative competitor in this ELISA. An anti-human Fc mAb bound non-specifically to KF94 despite being simultaneously generated with the mAb2s 94-2 and 94-7.

Confluent GL37 cell monolayers in 96-well cell culture plates (Corning, NY, USA) were incubated with serially diluted mAb2s and positive or negative competitors for 2 h at 37°C. After washing with PBS, 50 μl of HRPO anti-HAV-receptor antibody 190/4C was added to the corresponding wells and incubated for 2 h at 37°C. The wells were washed, and then bound HRPO anti-HAV receptor antibody was detected. Competition rates were measured using the formula:

Competition rate (%) = 100 (absorbance without competitor − absorbance with competitor)/absorbance without competitor.

Infectivity assays

We performed TCID50-infectivity and immunofocus assays to investigate the mAb2-mediated protection of GL37 cells against HAV infection.

For TCID50-infectivity assays, GL37 cell monolayers in 96-well cell culture plates were incubated with 50 μl of serially diluted mAb2 for 1 h at 37°C in a CO2 incubator. Without washing, 25 μl of HAV, genotype IA, IB or IIIB, was added to the appropriate wells and incubated for 1 h at 37°C. Cells were infected with all HAVs at a multiplicity of infection of 100. Infected cells were maintained in Eagle’s minimal essential medium (EMEM) supplemented with 2 mM l-glutamine, 50 μg/ml of gentamicin, 2% fetal bovine serum and 0.15% of sodium bicarbonate (2% FBS-EMEM) for 3 days. Infected cells were fixed with 80% methanol containing 0.03% H2O2, and air-dried, and then HAV propagated in infected cells was detected using anti-HAV hyperimmune rabbit serum. The plates were washed and incubated with HRPO anti-rabbit IgG (MBL, Nagoya, Japan), and then the rate of blocking of infection was measured using the formula:

Blocking rate (%) = 100 × (absorbance without inhibitor − absorbance with inhibitor)/absorbance without inhibitor.

The immunofocus assay was a modified radioimmunofocus assay [13] as follows: We mixed 50 μl of mAb2 (diluted 1:20 in PBS supplemented with 2% FBS) with an equal volume of HAV, genotype IA, IB or IIIB, containing 40–60 focus-forming units (FFU) in 2% FBS-EMEM. GL37 cell monolayers in six-well cell culture plates (Falcon, Franklin Lakes, NJ, USA) were inoculated with 100 μl of the mAb2-HAV mixture and incubated for 1 h at 37°C in a 5% CO2 environment to allow mAb2 and virus adsorption to the cells. Without removing the inoculum, the GL37 monolayer was overlaid with 5 ml of Dulbecco’s modified Eagle medium (Nissui, Tokyo, Japan) containing 0.6% agarose ME (Iwai Chemicals Company, Tokyo, Japan), 2% FBS and 0.22% sodium bicarbonate. After the agar had solidified, the cultures were placed upside down and incubated at 37°C for 10–12 days at 37°C in 5% CO2. The agarose overlay was discarded, and the cells were fixed with 1.5 ml of 80% methanol containing 0.03% H2O2 for 1 h at 4°C. Anti-HAV rabbit hyperimmune serum (1 ml of 1:2,000 dilution) was added to each well and incubated overnight at 4°C. The wells were washed with PBS and then incubated with 1 ml of HRPO anti-rabbit IgG for 2 h at 37°C. The plates were washed once again, and then HAV foci were detected using 1.5 ml of DAB substrate [0.5 mg/ml diaminobenzidine, 0.03% (NH4)2Ni(SO4)2, 0.03% CoCl2, 0.03% H2O2 in PBS]. Infection was considered blocked if the input FFU was reduced by ≥50%.

Results

Binding properties of mAb2s

Among the hybridoma clones secreting mAb2, two stable clones were selected by KF94-binding ELISA. The ascitic fluids of the selected clones, designated mAb2 94-2 and 94-7 were specific for KF94, and they did not react with either KF6 or HAV-negative human serum. The titers of mAb2 94-2 and 94-7, given as the reciprocal of the endpoint dilution required to generate maximal absorbance in KF94-binding ELISA, were 102,400 and 25,600, respectively. Both were categorized as IgG1 subclass, κ chain. They did not cross-react or bind to HAV.

Interactions between mAb2s, KF94 and HAV

Figure 1a shows that both mAb2s inhibited KF94 binding to HAV. At a 1:1,000 dilution, mAb2 94-2 and 94-7 inhibited 61.1 and 57.7% of the binding, respectively, and the inhibition rate gradually decreased with higher dilutions. The inhibition ability was specific for KF94. The mAb2s did not affect anti-HAV human mAb KF6 binding to HAV.

Fig. 1
figure 1

Interactions between KF94, mAb2s and HAV. Anti-HAV antibodies KF6 and KF94 were incubated with serially diluted mAb2 94-2 or 94-7 (filled circle, KF94 and 94-2; open circle, KF94 and 94-7; filled square, KF6 and 94-2; open square, KF6 and 94-7; (a), and then binding to inactivated HAV was determined by ELISA. The mAb2s (diluted 1:1,000) inhibited KF94 binding to HAV by 60%. The mAb2s did not interfere with KF6 binding to HAV, thus confirming their specificity. Anti-HAV antibody KF94 was incubated with serially diluted inactivated HAV, and binding of preincubated KF94 to mAb2 94-2 (filled circle) or 94-7 (open circle) was determined by ELISA. Inactivated HAV interfered with mAb2 binding to KF94 (b)

Inactivated HAV inhibited KF94 binding to mAb2s (Fig. 1b). The inhibition rates were proportional to the concentration of inactivated HAV. KF94 binding to the mAb2s 94-2 and 94-7 was inhibited by 69.1 and 78.5%, respectively, at the maximal concentration of inactivated HAV. Binding of mAb2s or HAV to KF94 reduced the affinity of KF94.

Binding of mAb2 to HAV cellular receptor

Binding of mAb2 94-7 and anti-HAV receptor antibodies 190/4, 235/4, and 263/6 to GL37 cells was confirmed by immunofluorescence staining (Fig. 2). Neither mAb2 94-2 nor NMS bound to GL37 cells.

Fig. 2
figure 2

Binding of mAb2s to GL37 cells. MAb2 94-7 and anti-receptor antibodies (190/4, 235/4 and 263/6) bound to GL37 cells were detected by immunofluorescence assay. MAb2 94-2 and normal mouse serum (NMS) were undetectable

We performed competitive inhibition ELISA to confirm that mAb2 94-7 and anti-HAV receptor antibody shared the same HAV cellular receptor. Anti-HAV receptor antibody 235/4 and mAb2 94-7 competed with the HRPO anti-HAV receptor 190/4C for binding to HAV cellular receptors. Figure 3 shows that the inhibition rates were proportional to the concentrations of the competitors. At a 1:100 dilution, the inhibition rates of the positive competitor, mAb2 94-7 and 94-2, and the negative competitor were 93.2, 59.2, 20.8 and 18.5%, respectively. The inhibition rate of mAb2 94-7 was lower than that of the anti-HAV-receptor antibody 235/4, but higher than that of the mAb2 94-2 or the negative competitor.

Fig. 3
figure 3

Binding of mAb2s to GL37 cells in competition with anti-HAV receptor antibody. Positive competitor (filled square) and mAb2 94-7 (open circle) interfered with HRPO anti-receptor antibody 190/C binding to GL37 cells by recognizing the common HAV receptors of GL37 cells. The competitive inhibition rates of mAb2 94-2 (filled circle) and of negative competitor (open square) were equally low

MAb2-mediated protection of GL37 cells from HAV infection

The rates of mAb2 94-7-mediated blocking of GL37 cell infection with HAV in TCID50-infectivity assays were 65.8, 54.1, and 86.0% for genotypes IA, IB, and IIIB, respectively. Blocking rates are expressed as dose-response curves (Fig. 4). On the other hand, mAb2 94-2 enhanced HAV propagation rather than protecting the cells.

Fig. 4
figure 4

MAb2-mediated protection of GL37 cells from HAV infection determined by TCID50-infectivity assay. GL37 cells were protected from HAV infection by mAb2 94-7, but not by mAb2 94-2. Combinations of mAb2 and HAV genotypes: filled circle 94-2 and IA; filled square 94-2 and IB; filled triangle, 94-2 and IIIB; open circle, 94-7 and IA; open square, 94-7 and IB; open triangle, 94-7 and IIIB

Immunofocus assays showed that the number of immunofoci of genotype IIIB strain KRM003 was reduced by 83.3% in the presence of mAb2 94-7 diluted 1:40. The results were similar for genotypes IA and IB, which reduced the number of immunofoci by 71.1 and 77.8%, respectively, at the same dilution. In contrast, mAb2 94-2 did not reduce the numbers of immunofoci.

Discussion

We generated the anti-idiotypic antibodies mAb2 94-2 and 94-7 by immunizing a mouse with anti-HAV neutralizing antibody KF94. The mAb2s were specific for the parental anti-HAV antibody KF94 and did not cross-react. They inhibited the binding of KF94 to HAV (Fig. 1a), and inactivated HAV competitively inhibited the binding of KF94 to the mAb2s (Fig. 1b), suggesting that the mAb2s and HAV bound to the paratope of KF94. Each mAb2 recognized different idiotopes within the paratope and could bind to KF94 as a surrogate of HAV.

These data indicated that the mAb2s mimic an HAV neutralization site that is complementary to the paratope of KF94. However, mimicry of the neutralization site by mAb2s might be incomplete, because the mAb2s inhibited KF94-HAV binding by only about 60%.

The characteristics of the mAb2s differed with respect to their affinity for the HAV-susceptible cell line GL37. The mAb2 94-2 neither bound to GL37 cells (Fig. 2) nor inhibited HAV infection (Fig. 4). The mAb2 94-2 seemed to have mimicked a portion of the antibody-binding site in the HAV neutralization site. The mimicked antibody-binding site interfered with KF94-HAV binding but did not influence virus-cellular receptor interaction.

On the other hand, mAb2 94-7 competed with the anti-HAV-cellular-receptor antibodies for binding to GL37 cells (Figs. 2, 3). The binding of mAb2 94-7 to GL37 cells partially blocked HAV infection (Fig. 4). We postulate that the mAb2 94-7 mimicked the part of the neutralization site that contains functional antibody-binding and cellular-receptor-binding sites.

The speculation that mimicry of the HAV neutralization site by mAb2 would be incomplete also explains why mAb2 94-7 could not totally block HAV infection. MAb2 94-7 was capable of blocking the infectivity of different genotypic strains (Fig. 4) and thus seems to mimic a common receptor-binding site among genotypes IA, IB and IIIB. Thus, these genotype strains might infect GL37 cells via a common receptor. Among the three genotypes, the rate at which mAb2 94-7 blocked infection was highest against strain IIIB (Fig. 4), which might be because KF94 was prepared from a patient infected with HAV strain IIIB.

Anti-idiotypic antibodies induced by immunization with an anti-Sindbis-virus neutralizing antibody competed with the virus for cellular receptors [22]. This suggested that a crucial receptor-binding site exposed on the viral surface is recognized by Ab1. We also speculate that the HAV receptor-binding site is exposed on the viral surface, because mAb2 94-7 competed with HAV for cellular receptors. Furthermore, the syngeneic mAb2 94-2 mimics part of the antibody-binding site. These data indicate that the antibody- and receptor-binding sites mimicked by mAb2s are exposed on the viral surface and are in close vicinity or overlap, thus comprising an epitope that could induce KF94.

Unlike other members of the family Picornaviridae [3], very little is understood about HAV neutralization sites and relationships between antibody- and receptor-binding sites. However, all published data support the notion that major and minor immunodominant neutralization sites exist on HAV virions and empty capsids [17, 19]. The immunodominant neutralization sites of native particles appear to be conformational and generally differ from those of denatured particles or isolated HAV structural proteins. Antibodies elicited by immunization with native or formalin-inactivated virus have broad neutralizing activity against different strains [20]. In contrast, the development of neutralizing antibodies in response to individual structural proteins, synthetic peptides, or expressed uncleaved precursors and polypeptides is problematic [7, 10, 15]. Thus, analysis of HAV neutralization sites using such probes is not simple. Analysis of HAV neutralization sites has mainly depended on the use of neutralization-escape mutants generated by serial passage of the virus in cultured cells in the presence of neutralizing mouse monoclonal antibodies [17, 19]. Although HAV strains isolated from various parts of the world belong to a single serotype [18], neutralization-escape mutants can be produced experimentally. These mutants are expected to possess a replicative advantage and survive more efficiently than wild-type virus as a result of the arrangement of surface neutralizing antibody-binding sites. However, their in vivo replication is restricted compared with wild-type virus, although they appear to be equally stable in vitro [14]. Furthermore, although the rate of substitution throughout the HAV genome is high, most of the mutations are silent [6]. Presumably, alterations in neutralizing-antibody-binding sites, such as in neutralization-escape mutants, might arise during natural infection. However, constraints probably prevent this from occurring in nature. Our results indicated that one such constraint depends on the relationship between antibody- and receptor-binding sites on the viral surface. These sites might be closely located or overlap within an immunodominant neutralization site. Therefore, mutations on the antibody-binding site might result in deterioration of the receptor-binding site. Such deterioration would consequently reduce the affinity of the HAV receptor-binding site, which would result in neutralization escape mutants being unable to replicate to significant levels in vivo. Alterations in antibody-binding sites of poliovirus are thought to be responsible for the affinity of the receptor-binding site [8]. The relationship between the antibody- and receptor-binding sites shown by mAb2s might partly explain why only one serotype of HAV exists.

To our knowledge, we are the first to analyze the HAV surface using anti-Ids to mimic receptor- and antibody-binding sites within HAV neutralizing sites. Our anti-Ids could not completely mimic the viral surface structure, but they remained functionally intact. Anti-Id antibodies may be considered obsolete, but they nevertheless represent an effective tool with which to structurally analyze the viral surface.