Systematic mutational analysis of epitope-grafted ED3’s immunogenicity reveals a DENV3-DENV4 bi-serospecific ED3 mutant

Dengue viruses are classified into four serotypes (DENV1∼4), and the severe forms of dengue disease, the dengue hemorrhagic fever and shock syndrome, are caused by sero-cross-reacting antibodies. However, the residue determinants of the serospecificity and sero-cross-reactivity are yet to be identified. Here, we report an epitope grafting mutational analysis of the serospecificity and cross-serospecificity of the envelope protein domain 3 (ED3; 107 residues, ∼11.6kDa), which contains two major putative epitopes of DENVs. To this end, we constructed ED3 from DENV3 (3ED3) and DENV4 (4ED3), and six epitope-grafted variants, where we transferred epitope 1 (L304I, K305D, V309M, and S310A) and/or epitope 2 (D383N, K384S, K387T, and N389H) of 4ED3 onto 3ED3 and vice versa. Mice immunization using 3ED3 and 4ED3 generated serotype-specific antisera, as expected. Similarly, most epitope-grafted ED3s produced antisera serospecific to the template ED3 with little or no cross-recognition of ED3 of the serotype from which the epitopes were taken. This result indicated that a mere grafting of the epitope was not sufficient to transfer serospecificity, contrary to our expectations. However, one epitope grafted ED3 mutant, where epitope 1 of 3ED3 was grafted onto 4ED3 (4ED3epi1), generated antisera that was serospecific to both 4ED3 and 3ED3. The 4ED3epi1 is thus a chimeric ED3 that produces antisera possessing serospecificity to both 3ED3 and 4ED3. The 4ED3epi1 provides a unique tool for analyzing serospecificity and cross-reactivity in dengue, and we hope it will serve as a template for trivalent and eventually tetravalent antisera.


Introduction
Immunogenicity and serospecificity of 3ED3 and 4ED3 82 3ED3 and 4ED3 have a ~70% sequence similarity, and their structures are very similar, if not 83 identical ( Figure 1) [18,33]. However, despite being similar 3ED3 was slightly more 84 immunogenic over 4ED3 in mice model (Figure 2; Suppl. Fig. S1-2) and produced highly 85 specific antisera, in agreement with our previous report [26]. Namely, antisera raised against 86 3ED3 reacted only with 3ED3, and anti-4ED3 sera reacted only with 4ED3, with no or minimal   and anti-epitope-grafted 4ED3 sera against 4ED3 variants. Cross-recognition of 3ED3 variants 104 by anti-4ED3 sera (c) and cross-recognition of 4ED3 variants by anti-3ED3 sera (d) are shown.
These observations indicated that the serospecificity transfer through epitope-grafting onto ED3s 126 was scaffold-dependent. For example, we would conclude that epitopes 1 and 2 are essential for 127 the serospecificity of 3ED3 because anti-3ED3 sera interaction with epitope-grafted 3ED3s was 128 weak, and antisera raised against the epitope-grafted 3ED3 showed reduced recognition of 3ED3 129 ( Figure 2a). However, antisera raised against 4ED3 sera exhibited a titer almost identical against 130 all three epitope-grafted 4ED3s (Figure 2b).

131
Finally, let us consider sero-cross-recognition of 3ED3 and 4ED3 by antisera raised against the 132 epitope-grafted ED3s. In most cases, no sero-cross-recognition was observed using epitope 133 grafted mutants (Figure 2c-d; Table 2). The only exception was the anti-4ED3 epi1 sera, which 134 showed a significant recognition of 3ED3 (Figure 2c), in addition to recognizing 4ED3. This 135 observation indicated that the serospecificity of DENV3 (3ED3) was almost entirely transferred 136 onto DENV4 (4ED3) through grafting of epitope 1 (Figure 2d; Table 2). However, anti-epitope-137 grafted 3ED3 sera did show sero-cross-recognition of 3ED3 and 4ED3, again illustrating the fact 138 that the serospecificity transfer was scaffold-dependent.

139
The identification of 4ED3 epi1 as a chimeric ED3 possessing bi-serospecificity to both 3ED3 and 140 4ED3 is an important finding from an application viewpoint. This is because 4ED3 epi1 may serve 141 as a tool for further exploring serospecificity and sero-cross reaction among DENV serotypes.  were buried in 3ED3's interior and were thus not expected to affect serospecific recognition of 150 3ED3 by anti-3ED3 sera [26]. The remaining S 309 A substitution was the only residue. Thus, it is 151 tempting to consider that the S309A mutation was the main factor behind the reduced detection 152 of anti-3ED3 sera by 3ED3 epi1 and low antibody titers of anti-3ED3 epi1 sera against 3ED3 ( Figure   153 1; Table 1). However, this interpretation would need to be confirmed by single mutation analysis 154 in the future.  Table 1. Putative epitope residues on 3ED3 and 4ED3. 157 Residues substituted on ED3s (residues grafted onto 3ED3 from 4ED3 and vice-versa); ASA: 158 solvent Accessible Surface Area; ASA was calculated using 3vtt.pdb and 3we1.pdb; side-chain.

161
On the other hand, all the four residues in epitope 2 are surface exposed and cause a significant 162 modification in the size and properties of the side-chains (K 384 S, K 387 T, and N 389 H). Grafting of 163 epitope 2 from 4ED3 onto 3ED3 template (3ED3 epi2 and 3ED3 epi1,2 ) induced noticeable local   (Table 2).   Fig. S3). Furthermore, this interpretation is fully corroborated with the observation that 191 antisera raised against 4ED3 epi1 could equally recognize 3ED3, 4ED3, and 4ED3 epi1 , suggesting 192 that Ser 309 is responsible for the 4ED3 epi1 chimeric ED3 detecting both anti-ED3s.

197
In summary, this is most likely the first systematic examination of the effect of epitope grafting