Identification of four neutralizing antigenic sites on the enterovirus D68 capsid

ABSTRACT

Enterovirus D68 (EV-D68) is an emerging human pathogen associated with respiratory diseases and/or acute flaccid myelitis. Neutralizing antigenic sites of EV-D68 have not yet been comprehensively studied. In this study, we generated multiple neutralizing monoclonal antibodies (MAbs) directed against EV-D68 prototype or clinical strains. All these antibodies can inhibit EV-D68 attachment. The antibody epitopes were identified by selection and sequence analysis of prototype or clinical strain-derived neutralization-resistant mutants. The epitopes were then grouped into four distinct neutralizing antigenic sites (I to IV) by cross-neutralization analysis of the mutants with the MAbs and by spatial considerations. Site I, including residues 81, 85, and 87 of VP1 protein, is located in the VP1 BC loop, near the fivefold axis, and at the north rim of the canyon (the receptor binding site). Site II, involving residues 137, 139, and 142 of VP2, is situated in the VP2 EF loop and at the south rim of the canyon. Site III is composed of VP1 C-terminal residues 285 and 293 and resides on the south side of the canyon of neighboring asymmetric unit. Site IV contains residue 70 (βB strand) of VP2 from an asymmetric unit and residues 74 and 79 (BC loop) of VP3 from an adjacent unit and is located around the threefold axis. The four antigenic sites show various degrees of sequence variation. The identification of the four neutralizing antigenic sites on EV-D68 capsid provides a better understanding of the recognition of EV-D68 by neutralizing antibodies and viral evolution and immune escape.

IMPORTANCE

Enterovirus D68 (EV-D68) is an emerging respiratory pathogen associated with acute flaccid myelitis. Currently, no approved vaccines or antiviral drugs are available. Here, we report four functionally independent neutralizing antigenic sites (I to IV) by analyses of neutralizing monoclonal antibody (MAb)-resistant mutants. Site I is located in the VP1 BC loop near the fivefold axis. Site II resides in the VP2 EF loop, and site III is situated in VP1 C-terminus; both sites are located at the south rim of the canyon. Site IV is composed of residue in VP2 βB strand and residues in the VP3 BC loop and resides around the threefold axis. The developed MAbs targeting the antigenic sites can inhibit viral binding to cells. These findings advance the understanding of the recognition of EV-D68 by neutralizing antibodies and viral evolution and immune escape and also have important implications for the development of novel EV-D68 vaccines.

INTRODUCTION

Enterovirus D68 (EV-D68) is a non-enveloped, single-stranded RNA virus belonging to species D of the genus Enterovirus in the family Picornaviridae (1). EV-D68 infection mostly affects children and leads to mild to severe respiratory illness. When severe, the illness can be life threatening, requiring intensive care and ventilator support (2, 3). Moreover, EV-D68 infection is also reported to be associated with neurological complications, mainly acute flaccid myelitis (AFM) that causes limb weakness or paralysis in children (3–5). Major outbreaks of EV-D68 left many of the children who survived with lasting paralysis and disabilities, highlighting the importance of EV-D68 surveillance in a post-polio world (6, 7).

EV-D68 (prototype strain Fermon) was first identified from pediatric patients in the United States in 1962 (8) and has been rarely reported until the worldwide outbreaks in the last decade (2, 3). Particularly, a nationwide outbreak of EV-D68-induced severe respiratory illness occurred in the United States from August 2014 to January 2015, leading to 1,153 confirmed cases including 14 deaths (3, 9). Coincident with this EV-D68 outbreak, 120 AFM cases were reported, suggesting the association between AFM and EV-D68 (5). EV-D68 outbreaks were also detected in the United States in 2016 and 2018 (10, 11), accompanied by the identification of 153 and 237 confirmed AFM cases, respectively (12). Moreover, over the past few years, EV-D68 has become widespread all over the world and resulted in clusters or outbreaks of cases (13, 14). Phylogenetic analysis reveals that the circulating EV-D68 strains are genetically diverse and can be divided into four distinct clades (termed A, B, C, and D) (15).

EV-D68 has an icosahedral capsid formed by 60 copies of viral capsid structural proteins VP1, VP2, VP3, and VP4 (16–18). The VP1, VP2, and VP3 proteins are located on the external surface of the viral capsid, whereas VP4 protein is situated internally (16–18). Thus, VP1, VP2, and VP3 are believed to be responsible for stimulating humoral immune responses in vivo. The 3D structures of EV-D68 virion reveal star-shaped plateaus at the fivefold axes, deep surface depressions (termed the canyon) encircling each plateau, and propeller-like protrusions at the threefold axes (16–18). The reported EV-D68 receptors include sialic acid, neuron-specific intercellular adhesion molecule-5 (ICAM-5/telencephalin), and sulfated glycosaminoglycans (sGAGs) (19–22). The sialic acid receptor binds to the canyon (20), while the ICAM-5 and sGAG binding sites have not been identified.

EV-D68 has posed a serious threat to global public health. However, there is no approved vaccine or antiviral drug against EV-D68 infection. Several experimental EV-D68 vaccines, including inactivated whole virus and virus-like particle (VLP) vaccines, have been developed and shown to induce neutralizing polyclonal antibodies that can protect mice against lethal EV-D68 infection (23–26). However, antigenic sites involved in the neutralization of EV-D68 have not yet been adequately studied. In addition, several EV-D68-specific neutralizing monoclonal antibodies (MAbs) have been reported, and their binding epitopes (especially contact residues) were identified by cryo-electron microscopy (cryo-EM) studies (18, 27, 28). However, the “functional” residues for the antibody epitopes, which are essential for antibody recognition and neutralization, have not been fully verified by experiments. Neutralizing antigenic sites in enteroviruses have been identified mainly through analysis of neutralizing MAb-selected resistant mutants or through assessment of the antigenicity and immunogenicity of viral peptides (29–33). The term “a neutralizing antigenic site” refers to a cluster of neutralizing antibody epitopes, in which each epitope overlaps with at least one other. In this study, to fully define neutralizing antigenic sites of EV-D68, five MAbs raised against EV-D68 prototype strain and four MAbs against EV-D68 clinical strains were used to select neutralization-resistant mutants. The mutants were sequenced and then classified into different groups by cross-neutralization studies. Finally, four distinct neutralizing antigenic sites (termed I to IV) were identified and mapped onto the 3D structure of EV-D68 mature virion. It is a little surprising that all the MAbs directed against the four antigenic sites could inhibit viral binding to cells. The results provide a better understanding of the recognition of EV-D68 by neutralizing antibodies and of the viral evolution and immune escape.

RESULTS

Generation and characterization of anti-EV-D68 neutralizing MAbs

In a previous study, two EV-D68-specific neutralizing MAbs, 2H12 and 8F12, were isolated from mice immunized with recombinant EV-D68 VLP [derived from clade B strain US/MO/14-18950 (hereafter MO50)] and characterized (27). In the current study, to prepare more neutralizing anti-EV-D68 MAbs, groups of BALB/c mice were immunized with purified inactivated Fermon (prototype strain) or inactivated US/MO/14-18947 (hereafter MO47; clade B strain from the 2014 EV-D68 outbreak), and the resulting hybridomas were screened by neutralization assay. A total of seven new MAbs were obtained and designated MAbs 1 to 7. The previously obtained MAbs 2H12 and 8F12 were now designated MAbs 8 and 9, respectively. The information of the MAbs is summarized in Fig. 1A. MAbs 1 to 5 were raised against inactivated Fermon, and MAbs 6 and 7 against inactivated MO47. The isotype of MAb 6 was IgM, while all the other MAbs were IgG. Neutralizing activities of the MAbs were measured by standard neutralization tests with EV-D68 strains Fermon and MO47. MAbs 1 to 5 potently neutralized the homologous strain Fermon with neutralization concentrations (the lowest MAb concentration that could completely inhibit cytopathic effect) ranging from 0.16 to 2.5 µg/mL but failed to neutralize the heterologous strain MO47 even at 500 µg/mL. MAb 6 effectively neutralized the homologous virus strain MO47 but did not neutralize the heterologous strain Fermon. MAb 7 strongly neutralized the homologous MO47 strain and weakly neutralized the heterologous strain Fermon. In line with our previous findings (27), MAb 8 neutralized the MO47 strain (nearly homologous to strain MO50) only. MAb 9 effectively neutralized both Fermon and MO47 strains. Together, six antibodies, MAbs 1 to 5 and 9, could potently neutralize prototype strain Fermon, while four antibodies, MAbs 6 to 9, could effectively neutralize clinical strain MO47.

Fig 1.

Characteristics and neutralization mechanisms of anti-EV-D68 MAbs. (A) Characteristics of anti-EV-D68 MAbs. Neutralization concentration is defined as the lowest antibody concentration that can fully inhibit cytopathic effect (CPE). Conc., concentration. (B and C) Attachment inhibition assay; 1 × 10⁵ TCID₅₀ of EV-D68 strain Fermon (B) or MO47 (C) was incubated with various amounts of the indicated anti-EV-D68 MAb or control (Ctr) antibody (anti-SARS-CoV-2 antibody 3A2) for 1 h. The virus/antibody mixtures were cooled and then allowed to bind to pre-cooled RD cells at 4°C for 2 h. After washing, RNA was isolated for real-time reverse transcription-PCR (RT-PCR) analysis for EV-D68 VP1 as well as β-actin as an internal control. For each antibody treatment, viral RNA levels relative to those for the only virus-infected samples are shown. Data are mean ± SEM of triplicate wells. Each symbol represents one well of 24-well cell culture plate. Results were analyzed by ordinary one-way analysis of variance to determine statistical significance between the virus-only and antibody-treated groups. ns, no significant difference (P ≥ 0.05); *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.

In our previous study, we have demonstrated that MAbs 8 and 9 can block EV-D68 attachment onto cells, the first step of viral infection (27). In this study, we investigated whether the newly prepared MAbs could inhibit EV-D68 attachment. Briefly, 1 × 10⁵ 50% tissue culture infectious dose (TCID₅₀) of EV-D68 Fermon or MO47 was incubated with various amounts of the antibodies prior to binding to cooled rhabdomyosarcoma (RD) cells at 4°C, and after extensive washing, the levels of viral RNA were determined by real-time reverse transcription-PCR (RT-PCR). As shown in Fig. 1B, pre-treatment with MAbs 1 to 5 could inhibit the binding of EV-D68 Fermon viral particles to RD cells in an antibody dose-dependent fashion. Similarly, MAbs 6 and 7 effectively inhibited the binding of EV-D68 MO47 (Fig. 1C). Particularly, MAb 6 showed the strongest effect in blocking virus adsorption (Fig. 1C), in line with the fact that it was the most potent neutralizing antibody (Fig. 1A). Together, MAbs 1 to 9 could inhibit viral binding to cell, albeit with different efficiencies.

Identification of neutralizing antigenic sites in EV-D68 prototype strain Fermon, defined by neutralization-resistant variants

The epitopes of the MAbs were identified by selection of escape mutants. MAbs 1 to 5 and 9 were used to select neutralization-resistant mutants from wild-type Fermon virus stock. For each plaque-purified escape mutant, its P1 (capsid protein) gene was sequenced and then compared with that of wild-type Fermon strain to identify the mutations. Most variants were found to have only single amino acid changes, making it easy to identify the specific mutation which was responsible for the neutralization resistance phenotype. The information of representative Fermon-derived escape isolates is summarized in Fig. 2A. The isolate 1.1, selected with MAb 1, had an amino acid change at residue 87 (H1087 to Y, H1087Y) of VP1, respectively. Note that viral capsid residues are numbered from 1001, 2001, and 3001 for VP1, VP2, and VP3, respectively. Isolates 2.1 and 2.2, selected with MAb 2, also carried mutations at amino acids 1085 (G1085E) and 1087 (H1087Y), respectively. MAb 3-selected isolate 3.1 possessed a K2070E alteration. MAb 4-selected isolate 4.1 also possessed the K2070E alteration. Isolate 5.1, selected with MAb 5, had a Q3079H change. MAb 9-selected isolates 9.1 and 9.2 had T2137N and G2142E substitutions, respectively. It is worth noting that MAb 1-selected mutants and MAb 2-selected mutants contain a common mutation at position H1087, suggesting that the epitope of MAb 1 overlaps with that of MAb 2. The mutants selected with MAbs 3 and 4 carry a common mutation at position K2070, indicating that the epitopes recognized by MAbs 3 and 4 overlap.

Fig 2.

Identification of neutralizing antigenic sites in EV-D68 prototype strain Fermon, defined by neutralization-resistant variants. (A) A summary of amino acid changes for the Fermon-derived escape variants. Escape mutants were selected by passaging strain Fermon in the presence of neutralizing MAbs, e.g., mutant 1.1 was selected by antibody 1. Viral capsid residues are numbered from 1001, 2001, and 3001 in VP1, VP2, and VP3, respectively. AA, amino acid. (B) Fold increase or decrease (prefixed with a minus sign) in neutralization concentration of the MAbs against the Fermon-derived escape mutants (mut), relative to wild-type (WT) Fermon strain. Pink highlighting, more than 10-fold decrease. (C–E) Model of EV-D68 Fermon biological protomer (PDB: 4WM8) showing the location of the neutralization escape mutations for group A mutants (1.1, 2.1, and 2.2) (C), group B mutants (3.1, 4.1, and 5.1) (D), and group C mutants (9.1 and 9.2) (E). For panel D, two adjacent (threefold related) Fermon protomers are shown, and distances between the indicated residues are shown in Å. The residues within the sites A, B, and C were shown in blue, magenta, and green, respectively. Note that the possible position of residue G1085 (not resolved in the Fermon structure) is indicated by dotted arrow. Black oval, triangle, and pentagon represent the twofold, threefold, and fivefold axes, respectively.

Each Fermon-derived escape isolate was tested for resistance to neutralization by each of MAbs 1 to 5 and 9 by standard neutralization assay. If neutralization of a given MAb against an isolate was reduced more than 10-fold compared with wild-type strain, the isolate is considered to be able to escape neutralization by the antibody; otherwise, the isolate is sensitive to neutralization by the antibody. The results are summarized in Fig. 2B. One can see that the escape mutants and MAbs fall into three non-overlapping groups, designated A, B, and C. Group A contained mutants selected with MAbs 1 and 2; group B contained variants selected with MAbs 3 to 5; group C contained variants selected with MAb 9. Specifically, the isolate 1.1 (H1087Y mutation) was resistant to neutralization by MAbs 1 and 2 but sensitive to neutralization by the other MAbs. Similarly, the isolates 2.1 and 2.2, which had mutations at G1085 or H1087, were resistant to MAbs 1 and 2. These results indicated that mutations at residues G1085 and H1087 conferred resistance to neutralization by MAbs 1 and 2, and thus, the two residues formed a neutralizing antigenic site (termed site A). The isolate 3.1 (K2070E mutation) was able to escape neutralization by MAbs 3 to 5 but sensitive to the other MAbs. Moreover, the isolates 4.1 (K2070E mutation) and 5.1 (Q3079H mutation) were only resistant to MAbs 3 to 5. Thus, the second antigenic site (termed site B) is composed of residues K2070 and Q3079 from different proteins. The isolates 9.1 (T2137N mutation) and 9.2 (G2142E substitution) were only resistant to their selecting antibody (MAb 9), indicating that residues T2137 and G2142 belonged to a neutralizing antigenic site (termed site C). Taken together, these results suggest the presence of three distinct neutralizing antigenic sites in the Fermon virus.

The residues conferring resistance to neutralization were mapped onto the 3D structure of mature Fermon virion (PDB: 4WM8) (16). As shown in Fig. 2C through E, these residues are all exposed on the virion surface. Specifically, antigenic site A, including residues G1085 and H1087, is located in the VP1 BC loop and near the fivefold symmetry axis of the virus (Fig. 2C). For antigenic site B, residue 2070 resides in VP2 B β-strand (βB), while residue Q3079 is situated in the VP3 BC loop (Fig. 2D). The distance between K2070 and Q3079 within one protomer is about 37 Å (Fig. 2D), which is larger than the size (about 30 × 20 Å) of the antigen-binding site of the antigen-binding fragment (Fab) (34). By contrast, the distance between K2070 and Q3079 from two adjacent (threefold related) protomers is much shorter (about 12 Å). Therefore, antigenic site B is likely to span the boundary between the threefold-related protomers in the Fermon virion. Antigenic site C, including residues T2137 and G2142, is situated in the VP2 EF loop and at the edge of the canyon (the putative receptor binding site) (Fig. 2E).

Identification of neutralizing antigenic sites in EV-D68 clinical strain MO47, defined by neutralization-resistant variants

MAbs 6 to 9 were used to select escape mutants from wild-type MO47 virus stock. The information of representative MO47-derived escape mutants is summarized in Fig. 3A. Isolates 6.1 and 6.2, selected with MAb 6, had single amino acid changes at positions 1081 and 1085, respectively. MAb 7-selected isolate 7.1 had a double mutation at residues 1285 and 1293. MAb 8-selected isolate 8.1 had a single amino acid substitution at position 2139. MAb 9-selected isolate 9.3 carried a G2142R change, while isolate 9.4 had double mutations at residues 2139 and 2142.

Fig 3.

Identification of neutralizing antigenic sites in EV-D68 clinical strain MO47. (A) Position of neutralization escape mutations in the capsid of the MO47-derived escape mutants. Escape variants were selected by passaging strain MO47 in the presence of neutralizing MAbs. AA, amino acid. (B) Fold increase or decrease (prefixed with a minus sign) in neutralization activity of the MAbs against the MO47-derived escape mutants, compared with wild-type MO47 strain. Pink highlighting, more than 10-fold reduction. (C-E) Model of EV-D68 MO47 biological protomer (PDB: 6CSG) showing the position of the neutralization escape mutations for group D mutants (6.1 and 6.2) (C), group E mutant (7.1) (D), and group F mutants (8.1, 9.3, and 9.4) (E). The residues within the sites D, E, and F were shown in blue, cyan, and green, respectively. Note that the possible position of residue R1085 (not resolved in the MO47 structure) is indicated by dotted arrow.

Each MO47-derived variant was analyzed for resistance to neutralization by each of MAbs 6 to 9 by standard neutralization tests. The results are summarized in Fig. 3B. The variants and MAbs could be divided into three separate groups (termed D, E, and F). Group D contained mutants selected with MAb 6; group E only included MAb 7-selected variant; group F contained variants selected with MAbs 8 and 9. Specifically, the variants 6.1 (S1081L mutation) and 6.2 (R1085Q change) were only resistant to their selecting antibody (MAb 6). Therefore, antigenic site D included residues 1081 and 1085. The variant 7.1 only escaped neutralization by its selecting antibody (MAb 7), and thus, residues D1285 and N1293 formed antigenic site E. The variants 8.1, 9.3, and 9.4, which had mutation(s) at positions T2139 and/or G2142, were resistant to MAbs 8 and 9. Moreover, compared to mutant 9.3 with the G2142R substitution, the second-site mutation at residue 2139 made isolate 9.4 more resistant to MAb 9. Thus, both residues 2139 and 2142 were included in MAb 9 epitope and belonged to antigenic site F. Together, these results demonstrate that there are three separate neutralizing antigenic sites in the MO47 virus.

Antigenic sites D to F of strain MO47 were mapped onto the structure of mature MO47 virion (PDB: 6CSG) (17). Antigenic site D, comprising residues S1081 and R1085, is situated in the VP1 BC loop and near the fivefold axis of the virus (Fig. 3C). Antigenic site E, composed of residues D1285 and N1293, is located at the C-terminal part of VP1 (Fig. 3D). Antigenic site F, comprising residues T2139 and G2142, is located in the VP2 EF loop and at the south rim of the canyon (Fig. 3E).

A summary of neutralizing antigenic sites of EV-D68

The information of antigenic sites of the prototype strain Fermon and the clinical strain MO47 was put together for analysis and summarized in Fig. 4A and B and 5. Antigenic site A (including residues 1085 and 1087) of strain Fermon and site D (residues 1081 and 1085) of strain MO47 are both located in the BC loop of VP1 (around the fivefold axis of the virus) and at the north rim of the canyon (the putative receptor binding site). It is, therefore, reasonable to infer that site A and site D are actually the same site of EV-D68, renamed site I, including residues 1081, 1085, and 1087 (Fig. 4A and B and 5A). Antigenic site C (residues T2137 and G2142) of strain Fermon and site F (residues T2139 and G2142) of strain MO47 are both situated in the VP2 EF loop and at the south rim of the canyon. Therefore, site C and site F are actually the same site of EV-D68, renamed site II, involving residues 2137, 2139, and 2142 (Fig. 4A and B and 5B). Antigenic site E of strain MO47 is composed of residues 1285 and 1293 (VP1 C-terminal) and is renamed site III (Fig. 4A and B and 5C). The C-terminal of each VP1 lies across the neighboring asymmetric unit and resides on the south side of the canyon of neighboring unit (Fig. 4B). Antigenic site B (renamed site IV) of strain Fermon contains residue 70 (βB strand) of VP2 protein from an asymmetric unit and residue 79 (BC loop) of VP3 protein from an adjacent unit (Fig. 4A and B and 5D). A previous study showed that a protective antibody 15C5 raised against EV-D68 clinical strain STL-2014-12 (hereafter STL) also targeted site IV (a more detailed explanation in the Discussion) and selected escape mutants which had an L3074S change (18). Therefore, site IV includes residues 2070, 3074, and 3079 and is located around the threefold axes and probably spans the boundary between the pentamers in EV-D68 virion (Fig. 4A and B and 5D). Note that the four antigenic sites are renamed according to their distance from the fivefold axis.

Fig 4.

Antigenic structure of EV-D68. (A) A summary of neutralizing antigenic sites of EV-D68 strains Fermon and MO47. A previous study showed that by using MAb 15C5, mutants with an L3074S mutation were selected from wild-type EV-D68 clinical strain STL. (B) Surface diagram of two adjacent (twofold related) asymmetric units of EV-D68 MO47 (PDB: 6CSG) showing the location of the four neutralizing antigenic sites. The residues involved in EV-D68 antigenic sites I, II, III, and IV were shown in blue, green, cyan, and magenta, respectively. Note that several amino acid residues differ between EV-D68 Fermon and MO47 strains. The possible position of residue 1085 (not resolved in the structure) is indicated by dotted arrow. The C-terminus of an adjacent VP1 is indicated by a black arrow. The canyon region (surface depression) is outlined with white dashed lines. (C) Location of the previously reported neutralizing immunogenic (NIm) sites on HRV14 asymmetric unit (PDB: 7BG6) surface. NIm-IA, IB, II, and III sites are highlighted in blue, midnight blue, green, and magenta, respectively.

Fig 5.

Amino acid sequence alignment of VP1, VP2, and VP3 of EV-D68 strains Fermon, MO47, MO50, and STL and HRV14 strain 1059. The residues involved in EV-D68 antigenic sites I (A), II (B), III (C), and IV (D) are boxed in blue, green, cyan, and magenta, respectively. The residues within the previously reported HRV NIm-IA, IB, II, and III sites are boxed in blue, midnight blue, green, and magenta, respectively. Dots represent residues identical to those of Fermon strain. Red dashes are gaps. The secondary structure elements are shown at the top.

EV-D68 is biologically similar to human rhinovirus (HRV), since both viruses show acid sensitivity and a lower optimum growth temperature (33°C) (35). Thus, it is valuable to compare the neutralizing immunogenic (NIm) sites of EV-D68 and HRV. NIm sites of serotype HRV14, a representative of the major HRV group, have been extensively studied, and four sites were identified, including NIm-IA, NIm-IB, NIm-II, and NIm-III (33, 36). The HRV14 NIm sites were mapped on the 3D structure of one asymmetric unit of HRV14 native particle (PDB: 7BG6) (37) (Fig. 4C). HRV14 NIm-IA site is located within the VP1 BC loop and includes residues D1091 and E1095 (Fig. 4C and 5A), which is equivalent to site I of EV-D68. HRV14 NIm-IB site contains residues from two separate peptide segments of VP1: residues Q1083 and K1085 (VP1 βB strand) and residues D1138 and S1139 (VP1 DE loop) (Fig. 4C and 5A), which do not correspond to any of the antigenic sites of EV-D68. HRV14 NIm-II site is composed of residues S2158, A2159, E2161, and V2162 in the VP2 EF loop (Fig. 4C and 5B), corresponding to site II of EV-D68. HRV14 NIm-III site involves residues N3072, R3075, and E3078 in the VP3 BC loop (Fig. 4C and 5D), which is equivalent to site IV of EV-D68. It is worth noting that EV-D68 site III does not correspond to any of the four NIm sites of HRV14 (Fig. 4C and 5C). Together, EV-D68 sites I, II, and IV correspond to HRV NIm sites IA, II, and III, respectively.

In addition, we also compared the antigenic sites of EV-D68 with those of other enteroviruses, such as poliovirus, enterovirus 71 (EV71), coxsackieviruses CVA16, and CVA10. EV-D68 antigenic site I is equivalent to the principal antigenic site (amino acids 89 to 100 located within the BC loop) in VP1 of poliovirus type 3 (38) and the binding epitope of CVA16-specific antibody 9B5 (39). EV-D68 antigenic site II corresponds to the VP2 P28 region (residues 136 to 150) of enterovirus 71 (EV71) and coxsackievirus A10 (CVA10), which was previously identified as the linear neutralizing epitopes of the two enteroviruses (30, 40, 41). EV-D68 antigenic site IV corresponds to the binding epitope of CVA16-specific antibody 8C4 (39).

Conservation analysis of antigenic sites of EV-D68

To evaluate whether the four antigenic sites are conserved, the capsid protein sequences from the prototype strain Fermon and the clinical strains MO47, MO50 (VLP vaccine strain), and STL were aligned and analyzed (Fig. 5). Note that numbering of amino acids is according to Liu et al. (16). For antigenic site I, residue 1081 is fully conserved among the four EV-D68 strains, while residues 1085 and 1087 are different between strains (Fig. 5A). This explains why MAbs 1 and 2, raised against strain Fermon, cannot neutralize the clinical strain MO47, while MAb 6, raised against strain MO47, cannot neutralize the prototype strain (Fig. 1A). For antigenic site II, residues 2137, 2139, and 2142 are completely conserved among the strains (Fig. 5B). For antigenic site III, residue 1285 is different between prototype strain and clinical strains (Fig. 5C), thus explaining why MAb 7 only weakly neutralizes the prototype strain (Fig. 1A), while residue 1293 is conserved among the four strains (Fig. 5C). For antigenic site IV, residues 2070 and 3079 are conserved among the four EV-D68 strains, while residue 3074 is different between prototype strain and clinical strains (Fig. 5D).

To further analyze the conservation of residues within the EV-D68 antigenic sites, a total of 1151 complete or near-complete EV-D68 VP1 sequences, 1109 EV-D68 VP2 sequences, and 1151 EV-D68 VP3 sequences were downloaded from NCBI database using BLAST as of August 2023. Note that the capsid protein sequences of almost all EV-D68 strains in the NCBI database were used for sequence alignment and conservation analysis. The results were shown in Fig. 6. The VP1 BC loop (antigenic site I), VP2 EF loop (antigenic site II), and VP1 C-terminus (antigenic site III) regions in different EV-D68 strains display extensive sequence variations (Fig. 6A and B). For antigenic site IV, VP2 βB region is highly conserved, while VP3 BC loop shows some sequence variation (Fig. 6B and C). The analysis also revealed that residues 2137 and 2139 within site II are completely conserved, while the other residues within the EV-D68 antigenic sites show various degrees of variation (Fig. 6D), suggesting viral evolution.

Fig 6.

Conservation analysis. (A–C) Sequence conservation analysis for EV-D68 VP1 (A), VP2 (B), and VP3 (C) proteins using the CLC Sequence Viewer and Vector NTI software. EV-D68 capsid protein sequences (1151 complete or near-complete VP1 sequences, 1109 VP2 sequences, and 1151 VP3 sequences) were obtained from NCBI database using BLAST as of August 2023. Locations of secondary structure elements (β-strands) are shown. (D) Analysis of the conservation of the residues within antigenic sites I to IV. Seq, number of sequences.

DISCUSSION

In this study, the neutralizing epitopes of EV-D68 prototype and clinical strains were successfully characterized by sequencing the neutralization-resistant variants. The resulting amino acid substitutions were then grouped by cross-neutralization assay and by spatial considerations into four distinct neutralizing antigenic sites (I to IV).

The four distinct sites are located on the outer capsid surface and thus highly accessible to antibody molecules. Our previous study reported that VP2 EF loop (site II) plays a key role in the recognition of MAbs 8 (previously designated 2H12) and 9 (8F12), revealed by the cryo-EM structures of EV-D68 MO47 mature virion complexed with 2H12 or 8F12 Fab (27). Residue T2139 within MO47 site II was found to be involved in the interaction with light chains of 2H12 or 8F12 (27), consistent with the finding that the T2139A mutation found in isolate 8.1 conferred escape from MAbs 8 (2H12) and 9 (8F12) (Fig. 3). Surprisingly, structural study shows that residue G2142 within MO47 site II is not directly involved in the interaction with 2H12 and 8F12 (27), whereas G2142R mutation found in isolate 9.3 escaped neutralization by 2H12 and 8F12 (Fig. 3). This is probably because the polar G2142 was mutated to the positively charged arginine with enlarged side chain, creating steric hindrance for the binding of 2H12 and 8F12 to site II of EV-D68. Moreover, VP1 GH loop is also involved in the interaction between EV-D68 virion and antibody 2H12 or 8F12 (27) and is spatially adjacent to VP2 EF loop, suggesting that VP1 GH loop should also be included in antigenic site II. In addition, Zheng et al. reported that an anti-EV-D68 protective antibody termed 15C5 interacts with the AB, BC, and HI loops of VP3 and the BC and HI loops of VP2 from adjacent protomer (18). They also found that all 15C5-resistant mutants had a L3074S change, which is situated at the VP3 BC loop (18). Obviously, the epitope of MAb 15C5 overlaps with that of our MAb 5 (also targeting VP3 BC loop), and thus, 15C5 epitope should belong to site IV (Fig. 4 and 5). Together, antigenic site IV contains residue 2070 (VP2 βB strand) from one protomer and residues 3074 and 3079 (VP3 BC loop) from an adjacent protomer.

Surprisingly, very few amino acid residues within each antibody epitope were identified in this study and other study (18). For each neutralizing MAb, at least 10 mutants were isolated. However, sequencing showed that almost all of each MAb-selected isolates had the same sequences and harbored only one or two mutations compared with the sequences of wild-type strains (Fig. 2A and 3A). A possible reason for this phenomenon is that there are structural or functional constraints to variation of epitope residues (29). Several residues are important for the stability of viral capsid or involved in other biological function, such as receptor binding, and thus, the mutations at these positions may be very harmful or lethal to the virus (29). The few residues identified by neutralization-resistant variant analysis are “functional” residues for antibody epitopes, which are essential for antibody recognition and neutralization. By contrast, structural studies of the virus-antibody complex show that about 6–15 residues exposed on the surface of EV-D68 mature virion are in contact with each antigen-binding fragment (Fab) (18, 27), and thus, these residues constitute the “contact” or “structural” epitopes. Whether each of the contact residues is necessary for antibody neutralization must be verified by experiments, e.g., site-directed mutagenesis. Obviously, each of the functional residues and some of the contact residues are critical for antibody neutralization, and the number of functional residues is often far lower than the number of contact residues (29). In addition, the “functional” residues may not be exactly the same as contact residues. For instance, for the MAb 9 (8F12) epitope, residue T2139 is both a functional and a contact residue, while G2142 is a functional residue, but not a contact residue (27). A limitation of our study is that we did not experimentally test the effects of mutations at the functional epitope residues on EV-D68 viral replication and attachment due to time and resource constraints. This important scientific question needs further investigation.

Except the prototype strain Fermon, the circulating EV-D68 strains are divided into four distinct clades, including A, B, C, and D (15). In this study, MAbs were raised against inactivated EV-D68 vaccines based on the prototype strain Fermon and clade B strain MO47 or against the VLP vaccine based on clade B strain MO50. Neutralization-resistant mutants were selected from wild-type Fermon or MO47 virus stock. Thus, one limitation of this research was that only EV-D68 prototype strain and clade B strains were included in this study. We did not conduct research on the antigenic sites of other clade strains due to resource and time constraints. In addition, only limited number of MAbs were included in this study, leading to the concern that there may be other antigenic sites on the EV-D68 capsid that have not yet been identified. More in-depth research in this field is required.

In this study, we found that all the MAbs (nos. 1 to 5) raised against prototype strain Fermon failed to neutralize clinical strain MO47, while MAbs 6 to 8 raised against clinical strains were unable to effectively neutralize prototype strain Fermon (Fig. 1A). Moreover, our previous study demonstrated that pooled immune sera raised against inactivated MO47 viral particles potently neutralized the homologous strain MO47 (titer = 8,192) but weakly neutralized the Fermon strain (titer = 128) (25). In line with these findings, residues 1085, 1087, 1285, and 3074 within antigenic sites I, III, and IV are found to be different between prototype strain Fermon and clinical strains (Fig. 5). In addition, NCBI BLAST analysis revealed that the loop regions where the EV-D68 antigenic sites are located show varying degrees of sequence variation (Fig. 6). Together, these results indicate large antigenic differences between EV-D68 prototype strain and clinical strains, suggesting viral evolution.

Antigenic site I is located at the north rim of the canyon, while sites II and III at the south rim of the canyon (Fig. 4B). Since the canyon is shown to be the binding site of the sialic acid receptor (20), the neutralizing antibodies targeting sites I, II, or III could inhibit the binding of the virus to cellular receptor via steric hindrance. Indeed, our previous study has shown that MAbs 8 (2H12) and 9 (8F12) directed against site II can block the binding of EV-D68 to sialic acid receptor, which was revealed by hemagglutination inhibition test and cryo-EM studies of EV-D68/antibody immune complexes (27). In this study, as expected, we found that the MAbs 1, 2, and 6 targeting site I and the MAb 7 targeting site III could inhibit the binding of EV-D68 to cell surface in a dose-dependent manner (Fig. 1B and C). Surprisingly, we and others found that the MAbs 3, 4, 5, and 15C5 targeting antigenic site IV were able to block viral particles from attaching to host cells (18), despite the fact that site IV is located around the threefold axis and away from the canyon region (Fig. 4B). A recent study reported that rhinovirus C receptor, cadherin-related protein 3, binds around the threefold axis and not in the canyon (42). Whether there are unidentified EV-D68 receptors that bind around the threefold axis of viral capsid needs further investigation.

Altogether, the identification of the four neutralizing antigenic sites on EV-D68 capsid provides a better understanding of the recognition of EV-D68 by neutralizing antibodies and viral evolution and immune escape. These findings also have important implications for the development of novel EV-D68 vaccines.

MATERIALS AND METHODS

Cells and viruses

Human RD cells were grown in DMEM (Gibco, Thermo Fisher Scientific, USA) with 5% fetal bovine serum (FBS) at 37°C. Mouse myeloma SP2/0 cells were grown in RPMI 1640 medium (Gibco) with 10% FBS at 37°C. EV-D68 prototype strain Fermon and clade B strain US/MO/14-18947 (hereafter MO47) were purchased from ATCC and then expanded in RD cells at 33°C. Viral titers were determined by the TCID₅₀ assay using RD cells.

Antigens

Inactivated EV-D68 viral particles were prepared from RD cells infected with strains Fermon or MO47 as described previously (25). Total protein concentrations of the purified antigens were determined by Bradford assay.

Preparation of MAbs

MAbs 1 to 7 were prepared in this study using inactivated viral particles as immunogens. Specifically, adult female BALB/c mice (6 to 8 weeks old) were immunized by intraperitoneal injection of 5 µg of antigen (inactivated Fermon or MO47 viral particles) and 500 µg of aluminum hydroxide adjuvant (Invivogen, USA) three times at 2- to 3-week intervals. Three days after the last booster, spleen cells from immunized mice were fused with SP2/0 myeloma cells using polyethylene glycol 1450 (Sigma, USA). The resulting fused cells were selected in hypoxanthine, aminopterin, and thymidine (Sigma) culture medium. Eight to ten days later, hybridoma culture supernatants were screened by neutralization assay as described below, and positive hybridoma cells were subcloned by limiting dilution. MAb isotypes were determined by the SBA Clonotyping System-HRP kit (Southern Biotech, USA). Purified MAbs were prepared from ascites fluid through Protein G or Protein L columns (GE Healthcare, USA).

MAbs 8 (also designated 2H12) and 9 (also designated 8F12) were generated in our previous study using recombinant EV-D68 VLP [derived from clade B strain US/MO/14-18950 (hereafter MO50)] as the immunogen (27).

Generation and sequence analysis of neutralization-resistant mutants

A total of 3,000 TCID₅₀ of EV-D68 wild-type strains Fermon or MO47 was added to RD cells grown in 24-well plates. After 1 h of incubation at 33°C, supernatants were removed, and fresh culture media supplemented with 1 µg/mL of each MAb were added. After 3–4 days of culture, the infected cell cultures were harvested and used to infect fresh RD cells in the presence of 5 µg/mL of the MAbs. Supernatants were harvested 3–4 days later and subjected to another round of selection with 10 µg/mL of the MAbs. The resulting escape mutants were purified by one round of plaque purification on RD cells. Plaques were then picked and amplified in RD cells. Viral RNA was extracted using TRIzol reagent (Invitrogen, USA) and reverse transcribed using M-MLV reverse transcriptase (Promega, USA). The P1 coding region was amplified by PCR, sequenced, and analyzed for mutations.

Neutralization assay

Before neutralization assay, the viral titer (TCID₅₀/mL) of each escape mutant was determined. Hybridoma culture supernatants and purified MAbs were tested for neutralizing activities against wild-type or mutant strains of EV-D68 by micro-neutralization assay as described previously (24). Briefly, in 96-well plates, 50 µL/well of serially diluted antibody samples was mixed with 100 TCID₅₀/well of wild-type or mutant viruses and incubated at 37°C for 1 h. Next, 15,000 cells/well of RD cells were added and incubated for 3 days, followed by observation of cytopathic effect (CPE). Neutralization concentration is defined as the lowest concentration that could completely inhibit CPE.

Inhibition of viral binding by the MAbs

A 1.0 × 10⁵ TCID₅₀ of EV-D68 Fermon or MO47 was separately mixed with 10, 100, or 1,000 ng of anti-EV-D68 MAbs or control antibody (anti-SARS-CoV-2 MAb 3A2) (43). The mixtures were incubated at 37°C for 1 h and then cooled on ice. The cooled mixtures were added to pre-chilled RD cells grown in 24-well plates and incubated at 4°C for 2 h. The cells were washed twice with cold PBS and then subjected to RNA isolation using TRNzol Universal reagent (TIANGEN, China) and cDNA synthesis using PrimeScript RT reagent Kit (Takara, Japan). The cDNA samples were subjected to real-time PCR analysis using SYBR Premix Ex Taq kit (Takara) according to the manufacturer’s protocol. EV-D68 Fermon-specific primers were as follows: forward primer, 5′-CACCATACTCACAACTGTGGC-3′; reverse primer, 5′-AATGAAATGAATCCTGCTCCT-3′. EV-D68 MO47-specific primers were as follows: forward primer, 5′-CGAGAGCATCATCAAAACAGCGACC-3′; reverse primer, 5′-CACTGTGCGAGTTTGTATGGCTTCT-3′. β-Actin primers were as follows: forward primer, 5′-GGACTTCGAGCAAGAGATGG-3′; reverse primer, 5′-AGCACTGTGTTGGCGTACAG-3′. Data were analyzed using the 2^–ΔΔCt method with β-actin as the internal control.

Structural representation of MAb epitopes

The locations of the antibody epitopes were shown on the crystal structure of EV-D68 Fermon mature virion (PDB: 4WM8) (16) or the cryo-EM structure of EV-D68 MO47 mature virion (PDB: 6CSG) (17) using UCSF ChimeraX software (v1.5). The previously reported NIm sites (33) were mapped on the 3D structure of HRV14 native particle (PDB: 7BG6) (37).

Sequence alignment

The VP1, VP2, and VP3 protein sequences of EV-D68 strains Fermon (GenBank accession no. AY426531), MO47 (KM851225), US/MO/14-18950 (KM851228; hereafter MO50), and STL-2014–12 (KM881710; hereafter STL) and HRV strain 1059 (L05355) were aligned using Vector NTI software (v11.5.1) and CLC Sequence Viewer software (v8). Note that numbering of amino acids is according to Liu et al. (16).

For sequence conservation analysis for EV-D68 VP1, VP2, and VP3 proteins, complete or near-complete EV-D68 capsid protein sequences were obtained from NCBI database using BLAST as of August 2023 and aligned using CLC Sequence Viewer software (v8).

ETHICS APPROVAL

The animal studies were approved by the Institutional Animal Care and Use Committee at the Shanghai Institute of Immunity and Infection.