Mechanistic basis for potent neutralization of Sin Nombre hantavirus … – Nature.com

A germline-revertant form of SNV-42 neutralizes SNV

Previous sequence analysis determined that SNV-42, which is encoded by human antibody variable region gene segments IGHV3-48*03/IGLV1-40*01, is remarkably close in sequence to the germline-encoded sequence, with a 97 or 99% identity to the inferred heavy and light chain variable gene sequences, respectively16. To understand whether somatic mutations are necessary for potent neutralization activity, we aligned the SNV-42 coding sequence with the inferred germline gene segments and reverted all mutations in the antibody variable regions to the residue encoded by the inferred germline gene (Supplementary Fig. 1). We then performed a neutralization assay to compare the potency of SNV-42 and the germline-revertant (GR) form of that antibody (denoted as SNV-42GR). We did not detect a change in the IC50 value between SNV-42 (IC50=21.4ngml1) and SNV-42GR (IC50=14.8ngml1) (Fig. 1a). Given that some of the residues in those regions are non-templated and thus cannot be reverted, we did not alter the junctional regions of SNV-42. These results indicate that many of the residues in the antibody paratope that are critical for SNV neutralization are encoded by IGHV3-48/IGLV1-40 germline genes. We also measured the KD values for the affinity matured and the germline reverted forms of SNV-42 to the recombinant SNV Gn head domain (Fig. 1b and Supplementary Table 1) using bio-layer interferometry (BLI). SNV-42 bound to GnH with sub-picomolar affinity, while SNV-42GR demonstrated sub-nanomolar affinity (9.21010M). However, this difference in affinity does not appear to impact the neutralization potency.

a, Representative neutralization curves of SNV-42, germline reverted (GR) SNV-42 and negative control DENV 2D22 to VSV/SNV determined through real-time cellular analysis using the Vero CCL-81 cell line. IC50 values were calculated on the basis of a nonlinear regression and error bars denote means.d. The assay was performed three independent times with similar results. b, Affinity measurements of SNV-42 and SNV-42GR for binding to SNV GnH ectodomain, measured by bio-layer interferometry. Representative curves and KD values are shown for SNV-42GR, while the KD value for SNV-42 could not be determined because the Koff could not be measured. Dashed line indicates dissociation step at 300s. c, Representative neutralization curves of SNV-42, SNV-42GR, positive control (oligoclonal mix of SNV-reactive antibodies) and DENV 2D22 to mutant VSV/SNV viruses. Error bars denote means.d. The assay was performed three independent times with similar results. d, SNV-42 binding in the presence of SNV M-segment mutant constructs determined by flow cytometry. The percent binding (% WT) of each mAb to the mutant constructs was compared to the WT SNV construct. An oligoclonal mix of SNV-reactive antibodies was included to control for expression of each mutant construct. The data are shown as means.d.; from left to right, n=9, 9, 9, 9, 9, 9, 6, 9, 9, 9, 9, 9, 9 and 6 technical replicates. The assay was performed three to four independent times with similar results. e, Top view of escape mutants mapped to the ANDV Gn/Gc spike (PDB: 6ZJM). The blue residues designate escape mutants. Gn is shown in white and Gc is shown in grey. All numbering for SNV sequences was based on GenBank KF537002.1.

Identifying potential escape mutants for antibodies is a crucial part of therapeutic development, and methods of immune evasion employed by hantaviruses are not well understood. To identify the critical binding residues involved in the recognition of SNV Gn by SNV-42, we used two different methods to identify escape mutants resistant to neutralization mediated by SNV-42. First, we implemented a high-throughput, single-passage neutralization escape mapping method using a real-time cellular analysis (RTCA) cell-impedance-based technology. We identified escape mutants in 32 of 88 replicates tested for escape, as manifested by cytopathic effect (CPE) in the presence of neutralizing concentrations of SNV-42 (Supplementary Fig. 2). We sequenced the gene encoding Gn in the virus in the supernatants in 6 wells. The neutralization-resistant viruses contained Gn gene mutations encoding K357Q or T312K alterations (Fig. 1c,e). To further identify escape mutants, we also selected a similar escape mutant (T312A) by serial passaging of a recombinant vesicular stomatitis virus (VSV/SNV) in increasing concentrations of SNV-42. We expressed recombinant forms of Gn with these mutations on the surface of cells and tested binding of SNV-42 to the mutant Gn constructs in flow cytometric binding assays. All three mutations completely ablated mAb binding, further supporting that these two residues are critical binding contacts (Fig. 1d). The binding of SNV-42 was not impacted by escape mutations selected for other SNV-neutralizing mAbs recognizing different antigenic sites (SNV-53 and SNV-24). Taken together, these two methods identified critical residues on SNV Gn that may be under pressure by some antibodies elicited in the human immune response to infection. However, SNV field strains with mutations at T312 or K357 have not been reported.

To understand the structural basis for SNV Gn recognition of SNV-42, a construct encoding the SNV GnH head domain (residues 21377) was produced recombinantly, purified and complexed with the Fab component of SNV-42. The SNV GnHSNV-42 complex was subjected to size exclusion purification and crystallized, and the structure was determined to approximately 1.8 resolution.

One complex of SNV GnSNV-42 was observed in the asymmetric unit of the unit cell (Fig. 2). The structure of SNV GnH has not been reported previously and consists of a compact fold formed of three domains: domains A and B and a -ribbon domain (Fig. 2). Despite a relatively low level of sequence identity (ranging from 43 to 63%), the SNV GnH is very similar to previously characterized hantavirus Gn glycoproteins5,6,7,14, where the equivalent regions of MAPV GnH, ANDV GnH, PUUV GnH or HNTV GnH exhibit root-mean-square deviations of 0.7, 0.9, 1.0 or 1.0 over equivalent C residues, respectively (Fig. 2b). Regions of SNV GnH that exhibit the greatest level of structural deviation from other GnH structures are in solvent exposed loop regions, consistent with these areas of the molecule being naturally flexible or requiring stabilizing contacts from the higher-order (GnGc)4 assembly.

a, Structure of the GnFab complex. The Fab is displayed with the backbone of the light and heavy chains coloured light grey or dark grey, respectively. The CDR loops are thicker and coloured according to the key in c. The Gn is displayed as a ribbon diagram with each of the three domains coloured according to the key in c. The two N-linked glycosylation sites are displayed in green and the location of the two previously described escape mutants (T312K and K357Q) are displayed in orange. Inset is a zoomed panel of the binding site with the side chains of the two escape mutant residues displayed. b, The backbone of the SNV GnH in pink overlaid on several previously reported GnH crystal structures from different hantavirus species in grey. These include Andes orthohantavirus (PDB ID 6Y5F), Maporal orthohantavirus (PDB ID 6Y62), Puumala orthohantavirus (PDB ID 5FXU) and Hantaan orthohantavirus (PDB ID 5OPG). Of note is the capping loop, indicated, which was replaced in SNV GnH with a much shorter GGSG linker to aid crystallogenesis. c, A domain schematic of the Sin Nombre glycoprotein precursor protein that is cleaved at the WAASA cleavage site to form Gn and Gc. The crystallized GnH region is outlined in bold and coloured according to domain. Transmembrane regions are displayed in dark grey and N-linked glycosylation sites displayed in green. The sequence of the capping loop between residues 8699 is displayed alongside the shorter GGSG linker that has been used in its place for this experiment.

Consistent with the epitope mapping analysis (Fig. 1d), SNV-42 binds to domain B and the E3-like domain of SNV Gn (Fig. 2a). The residues implicated in antibody escape identified above, T312 and K357, form key hydrogen bonding interactions with CDRH3 and CDRH1/3, respectively. These hydrogen bonding interactions appear to be perturbed when the mutations T312K and K357Q are modelled, and some rotomeric configurations of K312 may sterically interfere with the antibody, providing a structural rationale for antibody escape (Supplementary Fig. 4). The epitope comprises a large glycan-independent interface, which occludes ~8002 of buried surface area. While all complementarity-determining regions (CDRs) contribute to the epitope, residues comprising the CDRH3 of SNV-42 form the bulk of the interaction through insertion of a 9-residue-long loop into a cleft formed on the SNV Gn surface (Supplementary Fig. 3). CDRH3 possesses a low number of sequence somatic mutations from the putative germline, with only a single amino acid change from the germline D-gene (IGHD5-12*01). This change is one of only five amino acid changes from the germline-encoded sequence present in the paratope region including CDRH1 (T36), CDRH1 (E38), CDRH2 (R57) and CDRH3 (T112) that were all originally encoded as serine residues, plus CDRL1 (Y38) that was originally encoded as aspartate. However, these mutations do not impact the neutralization potency of SNV-42 (Fig. 1a and Supplementary Fig. 1). Interestingly, none of these five paratope residues were observed to sterically hinder antigen recognition when modelled back to the germline-encoded sequence (Supplementary Fig. 5).

SNV-42 is highly specific to SNV and did not demonstrate reactivity to or neutralize any other hantaviruses tested previously16. Assessment of sequence conservation at the epitope provides a structural rationale for this observation, since only 12 of 20 residues in the SNV-42 epitope were conserved with ANDV and 8 of 20 with HTNV. Furthermore, among these non-conserved residues, there exist non-complementary side chains which would probably sterically preclude mAb recognition (Supplementary Fig. 6).

Previous integrative cryo-electron tomography (cryoET) and X-ray crystallography analyses of ANDV, PUUV, HTNV and Tula virus (TULV) have revealed that the ultrastructure arrangement of the hantaviral (GnGc)4 is well conserved and consists of a tetramer of GnGc heterodimers. The GnH forms the most membrane-distal region of the spike and shields fusion loops located in domain II of the Gc5,7,13. To assess the location of the SNV-42 epitope in the context of the higher-order hantaviral GnGc lattice, we overlayed the SNV Gn subcomponent of our complex onto a previously reported (GnGc)4 assembly of ANDV (PDB: 6ZJM) (Fig. 3a,b). This analysis demonstrates that SNV-42 binds to the membrane-distal region of the lattice. While spatially distinct, the SNV-42 epitope is proximal to and slightly overlaps with the epitope of the weakly neutralizing mAb, HTN-Gn114, the only other structurally characterized anti-Gn mAb (Fig. 3c). In contrast to HTN-Gn1, SNV-42 binds in an orientation that is relatively perpendicular to the membrane (Fig. 3b,c). We note that each of the SNV-42 epitopes on the (GnGc)4 tetramer is mutually accessible for binding. Furthermore, unlike for HTN-Gn1, these sites are equally accessible in a cryo-electron microscopy (cryoEM)-derived model of the entire virus with the location of the (GnGc)4 spikes mapped onto the virion surface (Fig. 3a).

a, An EM-derived model of a Sin Nombre virion decorated in Fab fragments of mAb SNV-42. The virion model is derived from previously reported cryoET data of Tula virus5. The Gc is coloured blue and the Gn coloured pink or purple for the head or stalk regions, respectively. The light or heavy chains of the Fab are coloured light or dark grey, respectively. The zoomed inset displays nine individual glycoprotein spikes with the central spike surface rendered at higher resolution. The Fab fragments bound to the central spike are displayed as a backbone trace. b, Top view (left) and side view (right) of the Sin Nombre glycoprotein spike bound to Fab fragments of SNV-42. This assembly model is based on the previously reported ANDV glycoprotein spike tetramer (PDB: 6ZJM). The location of two SNV-42 escape mutants (T312K and K357Q) are displayed in orange and the equivalent locations of other previously reported antibody escape mutants are displayed in red. The complete list of antibody escape mutants and the species they apply to are detailed in Supplementary Table 2. To enable visualization of all epitopes, two loops that are not resolved in this SNV Gn structure (residues 8699 and 221229) were replaced by their equivalents from a previously reported ANDV Gn structure (PDB: 6ZJM). c, The equivalent view of a hantavirus glycoprotein spike bound to Fab HTN-Gn1, a previously reported neutralizing antibody that binds to HNTV14.

Previous epitope mapping of a panel of human SNV Gn- and Gc-specific antibodies revealed a series of epitopes spanning across solvent-accessible surfaces of the higher-order (GnGc)4 spike15,16. Integration of these data with putative epitopes predicted on the surface of other New and Old World hantaviruses indicates a broad distribution of epitopes across both the Gn and Gc glycoproteins (Fig. 3b). While immunodominant regions that are targeted during infection and immunization have yet to be identified, one such epicentre exists at the membrane-distal region of the GnH glycoprotein and co-localizes with our structurally elucidated SNV-42 epitope.

The role of bivalent interactions in the neutralization potency of hantavirus antibodies has yet to be described. To determine how the avidity effects impact the potency of SNV-42, we performed a neutralization assay comparing SNV-42 as a full-length IgG, Fab and F(ab)2. The F(ab)2 form was included to rule out any contributing steric effects of the fragment crystallizable (Fc) domain in neutralizing the virus. We saw no difference in the neutralizing activity between the full-length IgG form and the F(ab)2 form; however, the Fab form of SNV-42 did not demonstrate detectable neutralizing activity for VSV/SNV (Fig. 4a).

a, Representative neutralization curves of IgG1 and Fab forms of SNV-42 to VSV/SNV determined by RTCA using the Vero CCL-81 cell line. IC50 values were calculated on the basis of a nonlinear regression and error bars denote means.d. The assay was performed three independent times with similar results. b, Representative affinity curves of the F(ab)2 and F(ab) form of SNV-42 for binding to SNV GnH, measured by bio-layer interferometry. Representative curves and KD values are shown for SNV-42 F(ab), while the KD value for SNV-42 F(ab)2 could not be determined because the Koff could not be measured. Dashed line indicates the dissociation step at 300s. c, sEC1-EC2 blocking activity of neutralizing antibodies determined through a flow cytometric assay in which mAbs were added at saturating concentration before the addition of the soluble PCDH-1 domain labelled with Alexa Fluor 647 dye. High (50gml1), medium (10gml1) or low (0.5gml1) mAb concentrations were tested. Two-way analysis of variance (ANOVA) with Dunnetts multiple comparisons, ****P<0.0001; NS, not significant. The data are shown as means.d., n=9 technical replicates. The assay was performed two independent times with similar results. d, FFWO assay testing VSV/SNV post-attachment antibody neutralization in permissive (pH 5.5) conditions at 10gml1. Vero CCL-81 cells were used and GFP expression was measured to determine relative infectivity. The data are shown as means.d. of technical replicates, n=9. The assay was performed two independent times with similar results. One-way ANOVA with Dunnetts multiple comparisons, ****P<0.0001.

To further determine whether the lack of neutralizing activity was due to loss in avidity, we measured the KD values of the Fab and F(ab)2 forms of SNV-42 to SNV GnH using BLI (Fig. 4b). In concordance with the IgG form, the F(ab)2 bound SNV GnH with a sub-picomolar avidity, while the Fab form demonstrated a fast off rate and low KD value in comparison (4.08108M). Thus, the neutralization activity of SNV-42 requires bivalent binding to the (GnGc)4 assembly.

To investigate the possibility that bivalent binding of SNV-42 disrupts fusogenic conformational changes to the GnGc complex, we assessed the likelihood of SNV-42 to cross-link neighbouring epitopes on the (GnGc)4 assembly (Supplementary Fig. 9). This analysis suggests that inter-spike, but not intra-spike, bivalent binding may be plausible.

The hantavirus Gn protein probably plays several roles in facilitating the entry of hantavirus virions into host cells17. Gn forms a heterodimer with Gc and prevents the premature membrane insertion of the virus by covering the hydrophobic residues in the fusion loop5. Although the receptor-binding site is unknown, Gn is assumed to interact with attachment factors, including PCDH-117. To understand how SNV-42 engages with Gn to neutralize SNV, we investigated two mechanisms of interfering with viral entry: receptor blocking and fusion inhibition. Previous work has shown that the first extracellular cadherin-repeat domain (EC1) of PCDH-1 interacts with SNV Gn/Gc11, so we employed a flow cytometry-based competition-binding assay to test whether SNV-42 could block the interaction of a soluble recombinant form of the EC domains (sEC1-EC2). We showed that SNV-42 and SNV-42GR both block sEC1-EC2 binding to SNV Gn/Gc in a dose-dependent manner. However, we did not see complete blocking, even at the highest concentrations tested (50gml1; Fig. 4c). Notably, the Fab form of SNV-42 also did not block receptor binding, further suggesting that bivalent binding is required for receptor blocking and viral neutralization.

Although Gc is the canonical fusogenic protein, it is possible that targeting Gn may inhibit dynamic changes necessary to expose the fusion loop and promote viral entry18,19,20. We used a fusion from without (FFWO) assay to test fusion inhibition that can measure antibody-mediated neutralization post-attachment of the virion to the cell surface. SNV-42 and SNV-42GR significantly reduced VSV/SNV infection but did not completely inhibit viral fusion even at saturating concentrations (Fig. 4d). Further, although it is uncertain whether the hantaviral Gn remains bound to the Gc throughout the host-cell entry process, superposition of the SNV GnSNV-42 complex onto the structure of ANDV Gn bound to the near post-fusion state of ANDV Gc15 suggests that SNV-42 is also capable of recognizing a post-fusion state of the GnGc complex (Supplementary Fig. 10). While the precise transitions undertaken by the GnGc assembly are not well understood, it is plausible that bivalent binding of SNV-42 to the (GnGc)4 lattice interferes with the structural transitions required for entry and fusion. As SNV-42 does not mediate complete receptor blocking or neutralization post-attachment at high concentrations, the findings suggest that both mechanisms probably contribute together to cause the exceptional potency of the antibody.

Go here to see the original:
Mechanistic basis for potent neutralization of Sin Nombre hantavirus ... - Nature.com