Each of the mosquito strains examined were susceptible to DENV-2. However, the Chetumal and D2S3 strains were clearly more susceptible to DENV-2 infection than the Rex-D mosquitoes (Figure 1). DENV-2 replication in infected Rex-D mosquitoes was also slower than in Chetumal mosquitoes (data not shown). Unfortunately, we did not have a long colonized strain of Chetumal mosquitoes to compare to the newly colonized strain. Colonization has been shown to select for certain genotypes in Aedes aegypti . Thus the use of long-colonized mosquitoes as well as high passaged viruses from an area could confound the interpretation of the actual epidemic potential of DENV in that area. It would be prudent to study vector-pathogen interactions using field-relevant mosquitoes as well as low passage viruses from a specific area in order to draw more precise conclusions about the risk of dengue transmission in that area.
It was surprising that the Chetumal mosquitoes were as permissive to DENV-2 productive infection as the D2S3 strain, which was selected for DENV susceptibility . The susceptible phenotype is likely conditioned by different genetic mechanisms, which remain to be determined. However, the results do confirm previous studies demonstrating that mosquitoes from the eastern portion of the Yucatan Peninsula (an endemic area for dengue) are remarkably competent vectors for DENV .
DENV-2 infection was detected in midgut epithelial cells as early as 2 dpi. Viral titers and virus antigen then increased until 10 dpi when both virus titer and antigen decreased (Figure 2 and 3). In contrast, Q-PCR revealed that viral RNA copy number remained fairly constant even late into the EIP (Figure 3). Conversely, DENV-2 antigen was not cleared from salivary glands or neural tissues at equivalent time points. The former observation is of interest, because once a mosquito is productively infected with an arbovirus, she is capable of transmitting the virus for life. Obviously, any virus modulation in the salivary glands would be epidemiologically significant since transmission efficiency could be reduced. Modulation of virus in the midgut is provocative; either the virus or the epithelium is active in modulating the virus infection. The continued presence of DENV-2 RNA suggests a probable mechanism of persistence in vectors. DENV-2 as other arboviruses can persistently infect insect and human cell lines and arthropod vectors [20-23]. It is unlikely that the modulation is conditioned by the metabolic activity of the midgut epithelium, because a second uninfectious bloodmeal provided 14 days after DENV-2 infection of Aedes aegypti failed to reestablish virus production in the midgut (data not shown). The decline in infectious virus and viral antigen may be associated with an antiviral response, post-transcriptional or post-translational repression, or other self-limiting factors such as epithelial cells physiological status (i.e. cell aging). Further studies will be necessary to elucidate the molecular determinants of this phenomenon.
The mechanism(s) by which DENV disseminates from the midgut are not well understood. This study suggests that the mosquito tracheal system may be implicated in DENV dissemination (Figure 4 and Table 1). The presence of virus antigen in tracheal system significantly correlated with DENV-2 dissemination, mostly between 3 and 5 dpi. Tracheoles are part of the tracheal system, are 0.2 to 1 μm in diameter, are fluid-filled, and reach every tissue in the insect. The tracheal system has been reported to act as dissemination conduit for different viruses in insects. Notable examples are Autographa californica M nuclear polyhedrosis virus (AcMNPV) infecting Trichoplusia ni, nucleopolyhedrovirus infecting Bombyx mori, and Sindbis virus in Aedes albopictus [24,25]. The tracheal system is also suspected to be the midgut escape conduit for Venezuelan equine encephalitis virus (VEEV) in Ochlerotatus taeniorhynchus . However, this is the first report indicating that DENV may use the mosquito tracheal system during dissemination.
Unlike a number of other arboviruses, DENV displayed no tropism for midgut-associated (Figure 6-C) or flight muscles. Rift Valley Fever virus infects both skeletal and visceral muscles . Sindbis virus replicates in visceral muscles of Aedes albopictus . West Nile virus infects the contiguous muscles of the posterior and anterior midgut in Culex mosquitoes . The reasons for mosquito muscle refractoriness to DENV are unknown and require further analysis. Rift Valley fever virus and Sindbis alphavirus exhibit tropisms for cardia tissue [11,30], while DENV-2 was not routinely found in this tissue.
Interestingly, DENV-2 infection was detected in 36% of Chetumal mosquito salivary glands by 4 dpi. In contrast, 36% weren't detected until 10 dpi. The Chetumal mosquitoes would seem to be more vector competent than the Rex-D mosquitoes. Reportedly the EIP for DENV is between 7 and 14 days, and it can be affected by conditions such as temperature, humidity and viral dose [2,5,6,31]. Blood meals provided to mosquitoes in our studies contained 1.7 ± 0.7 × 107 PFU/ml, which is likely biologically relevant, since viremia titers of DENV-2 in humans may vary from 102 to 107 PFU/ml or 103 to 108 TCID50/ml [32,33]. In our observations, the lateral distal lobes were frequently the first site to become infected in the salivary glands, the medial and proximal lobes developed infection later (Figure 5). This pattern probably results from the higher virus receptor concentration in this region of the salivary glands, but this remain to be determined. The lateral distal lobes participate prominently in the secretion of enzymes and proteins involved in hematophagy [34,35], therefore infection of this region might promote early virus transmission. Other studies have demonstrated that DENV-2 can be transmitted by mosquitoes with . Thus, our results suggest that the EIP for DENVs may be shorter than the commonly reported (at least with some strains of the vector). There is no sensitive and reliable laboratory animal model for dengue, and currently used in vitro assays for transmission may fail to detect small but transmissible amounts of virus at earlier times. However further studies will be necessary to demonstrate that infectious virus is being transmitted early in the EIP and that the results can be extrapolated to low passaged DENV strains.
A shorter EIP for dengue would have important epidemiological consequences in dengue transmission. In nature, Aedes aegypti female mosquitoes feed frequently and almost exclusively on humans, which confers greater fitness than feeding on other hosts . These highly anthropophilic mosquitoes may ingest 2 to 3 bloodmeals during a single gonadotropic cycle, feeding with a frequency of 0.63 to 0.76 times a day . Aedes aegypti mosquitoes, with a very short EIP, persistently infected salivary glands, and the ability to feed multiple times within a gonadotropic cycle, would seem to be the ultimate DENV vector, which could help explain the epidemic potential of dengue.