Characterization of anti-E30 NAbs 6C5 and 4B10
Two antibodies, 6C5 and 4B10, were generated by immunizing BALB/c mice with formaldehyde-inactivated E30 mature virions. To investigate the serotype specificity of 6C5 and 4B10, we propagated, purified Echovirus 3 (E3), Echovirus 6 (E6), Echovirus 11 (E11) and Coxsackievirus B3 (CVB3) virions, and separately examined their binding abilities to each antibody by enzyme-linked immunosorbent assay (ELISA). The ELISA experiments showed that both 6C5 and 4B10 bind E30, but neither react with E3, E6, E11 or CVB3, suggesting that both antibodies are serotype specific (Fig. 1a). Surface plasmon resonance (SPR) assays demonstrated that 6C5 and 4B10 both exhibit tight binding to E30 with affinities of 1.51 and 2.88 nM, respectively (Fig. 1b). To explore whether these two antibodies recognize different or the same patch of epitopes, we performed a competitive SPR assay (see “Methods”) and the result indicated that the binding of one antibody blocks the attachment of the other (Fig. 1c), which raises the possibility of 6C5 and 4B10 binding to the same epitope or at least partially overlapped epitopes. In line with the binding results, cell-based viral neutralization investigations revealed that both antibodies could efficiently neutralize E30 infection as intact antibodies or Fab fragments with 50% neutralizing activities in the nanomolar ranges, but neither 6C5 nor 4B10 could protect against other EV-Bs (Fig. 1d, e). To further verify the potency of the two antibodies, groups of murine E30 antisera with exceptionally high titers were subjected to a competitive binding efficiency test against 6C5 and 4B10. These sera showed high blocking rates against 6C5 and 4B10 binding to the E30 virions ranging from 60% to 80%, reflecting that epitopes of 6C5 and 4B10 are dominant in antisera (Fig. 1f, g). In general, functional receptor attachment is capable of dissociating viral capsid protein interactions to trigger genome release3,16,17,29,30. However, both intact antibodies and Fab fragments of 6C5 and 4B10 destabilized slightly E30 virions by 1–3 °C (Fig. 1h and Supplementary Fig. 1). Given that E30 virions continues to exist as mature virion at physiological temperatures even after slight destabilization by 6C5 and 4B10 (Supplementary Fig. 2), destabilization of the virus is unlikely to be their mechanism of neutralization.
Both 6C5 and 4B10 block viral binding to its receptors
As with many other EV-Bs, E30 is believed to utilize two types of receptors to gain entry into host cells: CD55 for viral attachment; and FcRn for viral uncoating16,31,32. To verify if these two molecules act as receptors and directly bind to E30, the extracellular domains of human CD55 and FcRn were prepared for the binding assay. SPR assays indicated that both CD55 and FcRn interact with the E30 virion with similar binding affinities of ~2 μM, about 1000-fold lower than those of 6C5 and 4B10 with E30 (Fig. 2a). To investigate whether 6C5 or 4B10 interferes with the binding of E30 to CD55/FcRn, we performed four sets of competitive binding assays by exposing the E30 virions to the receptor first and then to the antibodies or the other way around. E6 NAb 1A1 and EVD68 receptor ICAM533 molecules were used as negative controls and neither showed any binding to E30 (Fig. 2b, c). In contrast, binding of 6C5 or 4B10 completely blocked the attachment of the two types of receptors to E30. Moreover, both of the receptors could be displaced from E30 and replaced by either 6C5 or 4B10 (Fig. 2b, c). To further verify these results in a cell-based viral infection model, real time reverse transcription polymerase chain reaction (RT-PCR) was carried out to quantify the amount of virus remaining on the host cell surface after exposure to antibodies either before or after viral attachment to cells at 4 °C. Consistent with the competitive binding results, both 6C5 and 4B10 efficiently prevented E30 attachment to the cell surface and could displace the viral particles that had already attached to the cell surface in a dose-dependent manner (Fig. 2d). Intriguingly, when used together, the anti-viral effect of 6C5 and 4B10 as a result of the disruption of the virus-receptor interaction, was greater than those of the sum of their individual anti-viral activities (Fig. 2d). Thus, the NAbs augment each other’s anti-viral activity, functioning in a complementary manner.
Structures of E30 in complex with its NAbs 6C5 and 4B10
To define the key epitopes and atomic determinants of the interactions between E30 and its two NAbs precisely, structural investigations of E30 in complex with Fab fragments from 6C5 and 4B10 were conducted. Cryo-EM micrographs of the formaldehyde-inactivated E30 virions (see the coordinated submission by Wang et al.15) in complex with 6C5/4B10 Fab fragments were recorded using an FEI Titan Krios electron microscope equipped with a Gatan K2 Summit detector (Supplementary Fig. 3). The cryo-EM structures of E30-6C5 and E30-4B10 complexes were determined at resolutions of 3.1 and 3.7 Å, respectively (Fig. 3a and Supplementary Fig. 3). The cryo-EM electron density maps were of high quality; allowing us to build the models of these two complexes (Fig. 3b). E30 capsid proteins exhibit no notable conformational changes upon binding to 6C5 or 4B10 with RMSDs of 0.3 and 0.4 Å, respectively, between the 6C5/4B10 Fab bound and unbound states of E30.
The 6C5 Fab fragment binds to the E30 viral surface within the pentameric building blocks at a site near the fivefold axis (Fig. 3c). The position of this binding site is similar to those observed previously for 11G1 antibody bound to Enterovirus D68 (EVD68) and NAb 1D5 bound to Coxsackeivirus A6 (CVA6)34. Five 6C5 Fab molecules attach to the north rim of the canyon that constructs the “star-shaped” protrusions surrounding the mesa [see coordinated submission by Wang et al.15]. When compared to the 6C5 Fabs, the Fabs of 4B10, contacting the south wall of the canyon, are comparatively more spread out (Fig. 3d). Interestingly, the 6C5 and 4B10 Fabs adopt distinctly different configurations upon binding to the E30 (Fig. 3c–e). When viewed down the fivefold axis, individual 4B10 Fabs stand vertically in an in-canyon attachment mode, resembling the strategy used by the uncoating receptor FcRn for its association with E30 [see coordinated submission by Wang et al.15]. In contrast, the 6C5 Fab inclines backward by ~50° crossing the canyon from close to a fivefold axis toward an adjacent threefold axis (Fig. 3c–f), covering a vast region of the surface, despite having a similar interaction area of ~800 Å2.
In the 6C5 Fab and 4B10 Fab bound E30 structures, variable domains of the light chain and heavy chain contribute ~39%, 61%, and ~92%, 8% of the protein-protein interactions, respectively, with 6C5 predominantly contacting VP1, whereas 4B10 largely binding to VP2 (Fig. 4a). The interaction patch in 6C5 comprises all six complementary determining regions (CDRs): L1 (residue 30 and 31), L2 (residue 51), L3 (residue 90), H1 (residue 32 and 33), H2 (residue 54 and 55), H3 (residues 99-101) and the light chain framework region (LFR, residue 33 and 48). The epitope recognized by 6C5 contains 11 residues, primarily locating in the VP1 BC loop (E82, K83, V84, D86, E87, D89, Y91), VP1 DE loop (T130), VP1 EF loop (K156, E159), and the VP1 HI loop (T229) (Fig. 4b). However, the epitope of 4B10 mainly includes residues 137, 138, 159, 161 and 163 of VP2 EF loop, residues 260 and 268 of VP1 C-terminal loop and residue 234 of VP3 C-terminus (Fig. 4b). Tight bindings between the antibodies and E30 are chiefly due to extensive hydrophilic interactions, including hydrogen bonds and salt bridges (Supplementary Table 2 and Table 3). Although the footprints of 6C5 and 4B10 on the E30 surface do not overlap completely, these two Fab fragments clash sterically due to proximity and the inclined posture of 6C5. These structural observations are consistent with the results of the competitive binding studies (Fig. 1c).
Serotype specific binding modes for 6C5 and 4B10
Functional characterization of NAbs 6C5 and 4B10 revealed that these antibodies are serotype specific, suggesting antigenic differences in the different subgroups of EV-Bs. A number of studies have revealed a role for several exposed loops engaged in the construction of the canyon of Enteroviruses in the serotype-specific differences [see coordinated submission by Wang et al.15]7,35. These loops include VP1 BC, GH loops, VP2 EF loop and VP1 C-terminal loop, which are also often mapped as neutralizing epitopes11,27,28,36. As a major structural marker, the VP1 BC loop not only contributes significantly to distinguishing E30 from other EV-Bs, but it is also the most divergent region with regards to primary sequence within EV-Bs (Fig. 4c and Supplementary Fig. 4). Overall, excluding 10% of the binding area contributed by conserved residues, the average conservation is only 26% in the 6C5 epitope (Fig. 4c and Supplementary Fig. 4). The specificity of VP1 BC loop both in sequence and configuration and it’s recognition by 6C5 explain the serotype-specificity of 6C5 (Fig. 4c, d). In contrast to 6C5, 4B10 buries 495 Å2 of the VP2 surface by interaction with VP2 EF loop and 200 Å2 of VP1 as well as 110 Å2 of VP3 via association with their C-terminal loops. Unlike the VP1 BC loop, the backbone Ca atoms of VP2 EF loop, VP1 C-terminal loop and VP3 C-terminus of E30 adopt similar conformations as those observed in other EV-Bs. However, the primary sequence of these regions varies across EV-Bs, indicating that the side-chain dependent interactions play critical roles in the recognition of the E30 antigenic determinants by 4B10 (Fig. 4c, d). Unexpectedly, the VP1 GH loop, harboring the widely reported major antigenic sites in EV-As27,37,38, is unlikely to contribute to the key epitopes in E30 due to the failure in obtaining NAbs targeting this loop despite many trials. In general, protective antibody diversity, such as 6C5 and 4B10 elicited by E30, is an important feature of the adaptive immune system, wherein the system protects hosts against viral infection by producing diverse protective antibodies. Since E30 elicits production of strong neutralizing antibodies like 6C5 and 4B10, it qualifies as a reasonable vaccine candidate [see coordinated submission by Wang et al.15].
Structural superimposition studies reveal steric clashes between 6C5/4B10 and receptors
Competitive binding assays demonstrated the abilities of 6C5 and 4B10 to effectively abrogate the interactions between E30 and its receptors FcRn and CD55 (Fig. 2b–d). Atomic structures of E30 in complex with FcRn/CD55 reveal that FcRn inserts into the viral canyon depression through primarily binding to VP1 GH, VP2 EF and parts of VP1 BC loop, while CD55 lies outside the canyon, adjacent to the “south wall” of the viral canyon [see coordinated submission by Wang et al.15]. FcRn presents a classical “in-canyon” recognition mode for most uncoating receptors, while CD55 exhibits a representative attachment strategy for many attachment receptors in picornaviruses. Superpositions of the E30-FcRn/E30-CD55 and E30-6C5 Fab/E30-4B10 Fab complex structures showed clashes between the two receptors and 6C5/4B10. Notably, the superimposition analysis reveals that 4B10 targets the canyon in a manner similar to FcRn (Fig. 5a, b). A number of receptors have been shown to insert themselves inside the viral canyon, whose conserved residues can, therefore, slip under the radar of the immune system, like KREMEN1, FcRn, and CD155 (major receptors for EV-As, EV-Bs, and EV-Cs, respectively)16,31,39,40,41. Unexpectedly, these receptor binding residues are remarkably non-conserved across receptor-dependent viruses42,43. Of the FcRn-binding residues, only VP1 Gly151, Gly207, and VP3 Gln238 are conserved and involved in tight interactions with FcRn. Most essential conserved residues for receptor binding present weak side-chain recognition signals, but control the local protein configuration, indicating that receptor binding is largely driven by side-chain independent interactions [see coordinated submission by Wang et al.15]. Such a strategy is likely to mitigate the constraints imposed by antigenic variation in receptor binding. For a relatively blunt antibody, the binding sites are usually outside the canyon – the “uncoating receptor” footprint, widely scattered in the most exposed regions with serotype-specific configurations, as those observed for E30 6C5, EV71 D6 and CVA6 1D5 (Figs. 5a, b and 6)27,34. In these cases, the neutralizing epitopes, at the most, partially overlap the receptor binding sites, and thus these viruses may evolve by mutating residues in the non-overlapped regions, thereby abrogating neutralization by their antibodies without impinging on receptor recognition. However, 4B10 directly inserts into the canyon and reaches the broad back region (Fig. 5c). Footprints of FcRn and 4B10 on the E30 surface reveal an overlapped patch with an area of ~100 Å2. Thus, the ability of the NAbs in preventing E30 from binding receptors can be attributed to steric clashes arising out of proximity and partially overlapping binding sites.