Apart from causing reduced virulence levels in the vertebrate host, American genotype DENV2 strains have been reported to be less efficiently transmitted by Ae. aegypti
]. Specifically, Armstrong and Rico-Hesse (2003) reported that six American genotype DENV2 strains isolated from Trinidad, Venezuela, Mexico, and Peru poorly disseminated from midguts of two different Ae. aegypti
strains when compared to dissemination rates of nine DENV2 strains from Southeast-Asia
]. None of the six American genotype DENV2 strains used in their study contained the unique aa substitutions described in DENV2-QR94 and DENV2-PR159
]. Others have speculated that American genotypes have lower efficiency in formation of replicative intermediates than viruses from Southeast-Asia, which leads to lower fitness levels of American genotype viruses in mosquitoes and vertebrate hosts
]. This could be the case for DENV2-QR94, which showed reduced replication efficiencies in midgut tissue following oral acquisition and in secondary tissue following intrathoracic injection. Intrathoracically injected DENV2-PR159, however, generated a mean virus titer in HWE mosquitoes that was comparable to that of American-Asian genotype DENV2-Jam1409. Immuno-fluorescence studies suggested earlier that DENV2-QR94 poorly disseminated from midguts of Chetumal mosquitoes at 4–7 days pbm by generating significantly fewer head tissue infections than American-Asian genotype DENV2 strains
]. Here, we show diminished dissemination of DENV2-QR94 from midguts of Chetumal mosquitoes as well as from midguts of two other DENV2-competent mosquito strains, D2S3 and HWE. Similar observations were made when assessing the vector competence of Ae. aegypti
(strain HWE) for American genotype DENV2-PR159. Our study demonstrates that the mosquito midgut is a selective sieve for DENV2 from which we could readily select dissemination-competent virus variants. We isolated from mosquitoes two viral midgut escape mutants, EM33 derived from DENV2-QR94 and EM41 of DENV2-PR159 that efficiently disseminated from the midgut.
The low-passage DENV2-QR94 virus had seven aa changes that have not been found in other wild-type DENV2 previously sequenced. Two of those aa changes occurred in structural proteins, E89G in prM and T55S in E. prM acts as a chaperone for E during virion assembly and maturation by preventing E from undergoing premature conformational changes during excretion through the trans-Golgi network at low pH
]. The E89G aa mutation in prM leads to a change of charge at position X of the four residue furin cleavage site R-X-K/R-R
]. Furin cleavage occurs late during virion maturation and is involved in E rearrangement and overall virus infectivity. Thus, the E89G mutation may impact efficient furin cleavage of prM. Interestingly, Kelly and colleagues observed an E89G mutation in prM after 30 serial passages of the attenuated DENV2 S16803 Thailand vaccine strain in primary dog kidney cells
]. The midgut escape mutant EM33 had one additional aa substitution, E202K in the viral E protein, resulting in a change of charge from negative to positive polarity. The 202 residue is located within the fg loop, which is part of the hinge region connecting E domain I with domain II
]. An aa substitution at this position is assumed to affect virus pathogenicity
]. The E202K mutation in E has also been reported in DENV2-S16803 after 10 serial passages in primary dog kidney cells
]. Importantly, the high passage DENV2-PR159 strain had the identical E202K mutation seen in EM33, although the lysine in this position in DENV2-PR159 E protein did not lead to virus dissemination from the midgut as observed for EM33.
Two aa changes in EM41 were associated with dissemination from the midgut, Q77E and E93D. Q77E leads to a charge change in a residue of domain II of the E protein, 21 residues upstream of the fusion peptide. The negative charge change of Q77E may have compensated for the positive charge change of E202K in the PR159 strain. The other aa substitution, E93D, occurred in the proteinase encoding domain of NS3. Both Q77E and E93D have not been described before in DENV2. Based on the observations made by Kelly and colleagues, it may be possible that aa substitutions E89G in DENV2-QR94 and E202K in the PR159 strain are a result of virus passage in cell culture and might not have been present in the genome sequences of the original virus samples that had been collected from infected patients
Recently, we have shown that transgene-mediated suppression of the antiviral RNA interference pathway (RNAi) in Ae. aegypti
significantly increased mean DENV2-QR94 titers in midguts but did not lead to increased dissemination efficiency
]. Thus, the inability of DENV2-QR94 to disseminate from the mosquito midgut did not depend on the RNAi response of the mosquito. However, at this point it cannot be excluded that other innate immune pathways of the mosquito may contribute to the defect in dissemination of QR94
]. Furthermore, we speculate that the single aa substitutions in prM (E89G) and E (E202K) negatively affect the ability of DENV2 virions to attach to recognition sites at the midgut basal lamina that would enable the virions to traverse this barrier. Alternatively, the mutations could hamper efficient virion maturation and/or conformational changes of virion surface structures that prevent the virus from escaping from the midgut. DENV2-QR94 not only failed to efficiently disseminate from the mosquito midgut but also produced significantly lower mean virus titers in midguts and secondary tissue in comparison to DENV2-PR159 or the DENV2-Jam1409 control, implying that the overall replication potential of the virus was diminished. Since DENV2-QR94 also had two unique aa substitutions in the polymerase encoding NS5, it is possible that these residues accounted for the reduced replication efficiency of the virus. However, it will be important to confirm our observations concerning aa substitutions affecting midgut escape of DENV2 by mutational analysis using an American genotype DENV2-derived infectious cDNA clone.
Previously, it was suggested that point mutations in the 3′UTR of American genotype DENV2 strain QR94 might contribute to its reduced midgut dissemination and replication efficiency
]. The 3′UTRs (453 nt) of DENV2-QR94 and DENV2-PR159 vary in five nucleotide positions (Additional file
). When compared to the 3′UTR of American-Asian genotype DENV2-Jam1409, the variation amounts to 26 positions including a nine nucleotide insertion. However, DENV2-PR159 and EM41 have identical 3′UTR nucleotide sequences and the 3′UTR of DENV2-QR94 and EM33 differ in a single nucleotide. Thus, mutations in the 3′UTR may not be seen as a likely cause for the poor midgut dissemination efficiency of the parental viruses.