Nucleotide deletions in dengue virus 3’ untranslated region compromises protein translation of live attenuated dengue vaccine strains

The live attenuated tetravalent dengue vaccine candidate, TV003 has now completed phase III clinical trial and is now licensed in Brazil. It is composed of DENV-1 and -4 that were attenuated through 30-nucleotide deletion (Δ30) in the 3’ untranslated region (3’UTR); the DENV-4 Δ30 backbone was also used to construct a chimeric virus bearing the pre-membrane and envelope genes of DENV-2. Attenuation of DENV-3 required an additional 31-nucleotide deletion (Δ30/31) in the 3’UTR. The reason for why two stretches of 3’UTR nucleotide deletions were needed to attenuate DENV-3, however, has remained undefined. Herein, we examined the DENV-3 (Δ30/31) against its wild-type Sleman/78 parent, as well as a partially attenuated DENV-3 (Δ30). We found using protein and RNA pulldown approaches, that wild-type 3’UTR bound host translation proteins, including ribosomal proteins and non-canonical eukaryotic initiation factor-3D (EIF3D) and EIF3K to support viral protein expression. Reduced binding of EIF3D and EIF3K to the Δ30/31 but not Δ30 mutant significantly reduced viral genome translation and replication. Correspondingly, silencing several of these proteins, including EIF3D, reduced viral protein expression and consequently decreased Sleman/78 and the Δ30 mutant replication. Notably, silencing EIF4E showed no statistically significant difference in viral replication, suggesting a functional role of EIF3 complex in DENV-3 genome translation. We also show that whilst Δ30 attenuated DENV-1 and -4, this deletion alone did not compromise DENV-2 translation, which could, at least partly, explain the reactogenicity profiles of these mutants in clinical trials. Our findings clarify the interacting partners of DENV 3’UTR and even suggest a path forward for updating the vaccine components, if necessary, with more contemporaneous DENVs.