Targeting “Immunogenic Hotspots” in Dengue and Zika Virus: A Novel Approach to a Common Vaccine

Diseases caused by Dengue (DENV) and Zika (ZIKV) viruses cause significant mortality and illness globally. Due to the high sequence similarity of the viral proteins and the purported cross-reactive immune responses against the viruses, we envisioned a common multi-epitope vaccine (MEV) against both viruses by adopting a novel approach of identifying “immunogenic hotspots”. These stretches of the structural and non-structural proteins are enriched with MHC class I and class II supertype-restricted T cell epitopes, and B cell epitopes, in addition to being highly conserved between different DENV serotypes and ZIKV. Such an approach ensures inclusion of multiple overlapping T and B cell epitopes common to both viruses, and also warrants high population coverage. Importantly, epitopes known to cause antibody-dependent-enhancement of infection have been excluded. These immunogenic hotspots have then been stitched together with linkers in-silico along with an adjuvant, CTxB to develop the MEV candidate. Four structural models of the MEV were selected on the basis of conformational preservation of CTxB, and their biophysical parameters, which also conserved the immunogenicity of the multiple epitopes. Importantly, each of the MEV candidates were found to interact with TLR4-MD2 complex by molecular docking studies, indicative of their ability to induce TLR-mediated immune responses.


Introduction 46
Vector-borne viral diseases, like Dengue fever and Zika viral fever cause severe illness, 47 significant mortality and resultant socio-economic burden worldwide, thus proving to be a major 48 public health concern of the modern era. The Dengue virus (DENV) has four serotypes (DENV1-4),

49
which are responsible for dengue fever, dengue hemorrhagic fever, and dengue shock syndrome in 50 humans [1]. Globally, it is the fastest-growing mosquito-borne disease, causing nearly 400 million 51 infections every year with a mortality rate of 20% when untreated (https://www.who.int/en/news-52 room/fact-sheets/detail/dengue-and-severe-dengue). On the other hand, the Zika virus (ZIKV) causes 53 the zika viral fever which, in addition to causing headache, muscle and joint pain, rash, edema of 54 extremities, retro-orbital pain and conjunctival hyperemia [2],often lead to congenital abnormalities in 55 unborn fetus [3]. Due to such long-lasting damage by ZIKV, WHO declared this virus a public health 56 emergency of international concern in 2016.

57
Being members of the genus Flavivirus, DENV and ZIKV share some common genetic as 58 well as phenotypic features. Both are borne by different species of the Aedes mosquito, and have 59 approximately 11 kb+ ssRNA genome that encodes three structural and seven non-structural proteins.

62
Importantly, the 4 serotypes of DENV share a high genetic similarity with ZIKV, especially in the E 63 protein and the NS3 and NS5 proteins [6]. following the first infection. Additionally, ADE has also been reported in ZIKV infections following 69 prior DENV infections and vice versa [12,13,14]. This could be due to binding of the non-neutralizing 70 or weakly neutralizing, but cross-reactive antibodies from the first infection, to the heterotypic virus predict epitopes for a predefined set of MHC class II alleles that gives maximum population 148 coverage [24]. The epitope length was set as 9mer and 15mer for MHC class I and class II molecules 149 respectively. The output result was in the form of percentile rank and the lower the rank of an epitope 150 the stronger the binding of that particular epitope. A cutoff of 2.0 percentile rank was used to get the 151 best predicted epitopes which was then used to compute the immunodominant regions of the proteins.

152
Prediction of linear and conformational B cell epitopes   246 Therefore, to select the best candidates from the non-structural proteins of the DENV serotypes and 247 ZIKV, we probed these proteins for sequence homology determination and phylogenetic relatedness.

248
Our studies as well as already published data [6]show that NS3 and NS5 have the highest sequence 249 homology(greater than 60% homology) between ZIKV and all the DENV serotypes. Therefore, we 250 selected these two proteins for further analysis of phylogenetic relatedness. The NS5 protein of ZIKV 251 and DENV4 share three common ancestors and the pair shares two common ancestors with DENV2,

252
indicative of high degree of relatedness between these three viral species. However, NS5 of DENV1 253 and DENV3 are phylogenetically separated from these three (Fig 1(A)). In comparison, NS3 of ZIKV 254 and DENV4 share two common ancestors, while that of DENV1 and DENV3 share a separate pair of 255 common ancestors, reflective of the greater phylogenetic difference between these two clades.

256
Additionally, further divergence is seen with the NS3 protein of DENV2 from the rest of the members 257 (Fig 1(B)). Thus, the data demonstrates that NS5 is more closely related across all the viral species 258 than NS3 and hence it was selected as a candidate for further vaccine design.  In contrast, the E protein yielded 2 such dominant immunogenic hotspot stretches (Table 2).

280
Interestingly, E protein, better known for eliciting humoral immune response, gave a 15mer conserved 281 sequence WLVHKQWFLDLPLPW within the 200-226 hotspot stretch (Fig 2(A)). This conserved 282 sequence has a few amino acid substitutions where L205M, K208R and L212F in DENV3, K210R in 283 DENV2and Q216E, L219H and L221I in ZIKV. Since, the majority of these substitutions are with 284 biochemically similar amino acids and are not in the anchor regions of the epitopes generated from 285 the stretch, this particular hotspot was selected for our MEV candidate. The second hotspot of the E 286 protein did not yield any conserved sequence and therefore was not considered further.

292
Another 15mer hotspot from NS5 protein having single position amino acid differences highlighted.

293
(E) 12mer conserved hotspot from NS5 protein having single position amino acid differences 294 highlighted.

296
NS5, which is known to elicit strong T cell responses, gave three immunogenic hotspot 297 stretches ( Table 2) that were found in the similar regions across all viral species. Upon investigation 298 of the these stretches, the 450-500 amino acid stretch of NS5 yielded a 12mer CVYNMMGKREKK 299 and 15mer AKGSRAIWYMWLGAR sequence that was fully conserved containing overlapping 300 epitopes (Fig 2(B), 2(C)). Another 15mer sequence FLEFEALGFLNEDHW was obtained from the 301 same hotspot, but with two highly-related substitutions (F482Y in DENV3 and L491M in DENV1) 302 (Fig 2(D)). A fourth 12mer conserved sequence GWSLRETACLGK was mined out from the 303 hotspot stretch 750-769 (Fig 2(E)). This sequence had a substitution of two amino acids (L750I and 304 G757A) specifically in ZIKV which are biochemically similar as well. Moreover, these substitutions are not in the anchor region of the epitopes (important for the formation of stable pMHC complexes 306 and antigen presentation by APC) that are generated from these hotspots. Thus, these four 307 "Immunogenic Hotspots" were selected as potential candidates for generation of multi-epitope 308 vaccine.

310
'Immunogenic Hotspot' from the selected proteins   Fig 3). This 30mer sequence essentially is a combination of the two 332 15mer hotspots (Fig 2(C), 2(D)) that were obtained in the case of MHC class I epitopes having Like T cell epitopes, B cells epitopes were also identified from the candidate proteins. Online 343 epitope prediction tool ABCpred was used for linear B cell epitope prediction of the selected proteins.

344
The 16mer epitopes with threshold value more than 0.75 were selected and further analyzed to 345 identify the immunogenic hotspots.

346
DENV E and M proteins elicit an initial B cell response that generates neutralizing antibodies 347 against these two proteins. These antibodies against PreM and E proteins have been reported to

372
(AKGSRAIWYMWLGARFLEFEALGFLNEDHW) (Fig 3) previously described also contains 373 multiple linear B cell epitopes. This indicates that this region of NS5 is important for both cell 374 mediated and humoral immune response. were identified with the help of Discotope 2.0 server using the available 3D structures from PDB.

386
Only a few regions of the E protein were found to give consistent conformational epitopes.

399
Next, we analyzed the conformational epitopes derived from the NS5 protein of ZIKV,  (Table 4).

426
To increase the immunogenicity and the probability of uptake by cells, the adjuvant CTxB was   Table). Fig 4   449 shows the predicted tertiary structures of the four models (8, 9, 19 and 21) as visualized in Pymol.

450
Molecular docking of the individual MEV candidates with TLR4 showed thateach candidate MEV 451 interacts with different regions of the TLR4 and its associated complex MD2 (Fig 5). This is 452 indicative of the potential of the selected MEV candidates to activate TLR mediated immune 453 response.These four vaccine candidates were further probed for T and B cell epitope generation using 454 the respective epitope-prediction tools discussed above and it was found that the highly immunogenic 455 T and B cell epitopes have been preserved in these candidates (S7 Table). which have a high degree of sequence similarity, ADE is a matter of serious concern, with numerous 470 reports of increased virulence of subsequent heterotypic infections. Therefore, any vaccine made 471 against these viruses, must be specifically checked to exclude any potential ADE, while ensuring the 472 extension of the protection to all serotypes concerned.

473
Given the close phylogenetic relationship between DENV serotypes and ZIKV, we ventured 474 to design a common vaccine which will not merely incorporate multiple single T and B cell epitopes,

475
but rather include immunogenic stretches or hotspots, rich in multiple HLA supertype-restricted T cell 476 epitopes and B cell epitopes that are common to the pathogens, as well as completely conserved 477 among species. Only four such hotspots could fulfill the stringent criteria, which also specifically 478 excludedthose stretches that contain known ADE-associated epitopes.Notably, in the predicted MEV 479 candidates, only linear B cell epitopes could be included as the predicted conformational B cell 480 epitopes, were not found to be completely conserved sequences among the considered viruses. More 481 importantly, the contact residues of these predicted conformational epitopes varied in number among 482 the serotypes, which could purportedly lead to differences in antibody affinities against these epitopes.

483
Further investigation needs to be done to ascertain the contribution of this observation to ADE.

484
Furthermore, the designed MEV described here, also incorporates an adjuvant, the CTxB, 485 which can be takenup by cells via the GM1-gangliosides, and has been found to be recognized by 486 TLR4[29], thereby having the ability to initiate TLR mediated signaling pathways that can activate 487 the protective immune response. Lastly, the immunogenic hotspots described in this study include T 488 cell epitopes that are restricted to MHC class I supertypes and MHC class II alleles with maximum 489 population coverage, thus ensuring its efficacy among the global population. In sum, the MEV candidates described here can be considered as a stepping stone towards a common vaccine against 491 DENV and ZIKV.