RNAi is a major antiviral response in mosquitoes. The only other described mosquito immune response to arbovirus infection is mediated by the Toll antimicrobial pathway [26
]. RNAi is a highly conserved mechanism that is stimulated by the presence of an invading virus and controls viral replication through the sequence-specific degradation of the virus RNA. To study RNAi during SINV infection of Ae. aegypti
, we have engineered a double subgenomic SINV to express B2 protein, a potent VSR [13
]. In a recently published study, SINV-B2 and ONNV-B2 were shown to cause mortality in injected Ae. aegypti
and An. gambiae
mosquitoes, respectively [10
]. We show that mosquitoes infected in a more natural manner (per os
) with a B2 expressing SINV demonstrate increased viral titers, higher levels of viral dissemination from the midgut, and greatly enhanced virus-induced mortality in Ae. aegypti
, Ae. albopictus
, and Cx. tritaeniorhynchus
mosquitoes. In our system, the B2 protein is translated only in infected cells, avoiding potential off-target effects associated with transient dsRNA-mediated silencing of the RNAi pathway. Tschuch et al found that introduction of siRNA specific for green fluorescent protein (GFP) into human cells that did not express GFP non-specifically perturbed expression of more than 200 genes [27
]. A similar non-specific dsRNA-mediated regulation of gene expression has been described in sandfly (Lutzomyia longipalpis
) cell culture and the marine shrimp, Litopenaeus vannamei
]. Although similar experiments have not been performed in mosquito cells, introduction of dsRNA could have a similar effect. Detectable, yet not statistically-significant increases in viral titer have been observed when control experiments injecting β-gal dsRNA and virus into mosquitoes have been performed [7
]. While the potential for non-specific response of the mosquito immune system to dsRNA is intriguing and should be studied further, examining the RNAi response using a SINV expressing a viral RNAi suppressor is superior to dsRNA-mediated interference of the RNAi pathway.
Three lines of experimental evidence suggest that the B2 protein was functional in RNAi suppression when expressed during TE/3'2J/B2 virus infection. First, in vitro dicing experiments show inhibition siRNA accumulation in cell lysates derived from TE/3'2J/B2 virus-infected Aag2 cells. The presence of B2 protein inhibits the accumulation of biotinylated siRNAs, presumably by binding to the synthetic dsRNA and sequestering from Dicer-2. The presence of siRNAs in mock- and TE/3'2J/GFP-infected lysates provides evidence that Aag2 cells have a functional RNAi mechanism. Also, this shows that inhibition of siRNA accumulation is specific to TE/3'2J/B2 virus infection. The second line of evidence comes from Northern blot analysis of small RNAs in mosquito cells. Considerably less SINV-specific siRNAs accumulated in cell culture and mosquitoes infected with TE/3'2J/B2 virus compared to TE/3'2J and TE/3'2J/GFP virus infection. The dsRNA formed by viral replicative intermediates may be bound by B2 protein, protecting the dsRNA from detection by the RNAi machinery.
Finally, virus titers observed in Aag2 cells and adult Ae. aegypti
mosquitoes were much higher when B2 protein was expressed during infection. This agrees with previous data showing that inhibition of the RNAi pathway allows for arboviruses to replicate more efficiently in mosquitoes [6
]. By injecting mosquitoes with dsRNA targeting Dicer-2 or Argonaute-2 after an infectious bloodmeal, Campbell et al [6
] were able to show that SINV titers in individual mosquitoes increased significantly by day four as compared to β-gal dsRNA injected controls. The same effect was not seen at day seven and the authors suggest this may be due to a stimulation of the antiviral response by this time point or degradation of the dsRNA triggers via decay [6
]. A similar general phenomenon was seen with ONNV infection of An. gambiae
mosquitoes, with a detectable increase in virus titer up to six days post infection [7
]. This difference may be explained by the inoculation route as both dsRNA and ONNV were administered intrathoracically, bypassing any infection barriers associated with the midgut and ensuring introduction of virus and dsRNA into the hemocoel [7
]. A significant increase in SINV titers was observed at both four and seven days post infectious bloodmeal in mosquitoes ingesting TE/3'2J/B2 virus. The RNAi response is continuously inhibited by B2 protein as it is produced in infected mosquito cells. dsRNA intermediates or secondary structure of the virus genome will not be recognized by the RNAi machinery, allowing virus replication to continue unabated.
Our data indicate that SINV becomes pathogenic to mosquitoes when RNAi is suppressed during virus infection. Pathology and mortality have been associated with alphavirus infection of mosquitoes, but the pathology is specific to the midgut or salivary glands and the mortality can, in some cases, be attributed to intrathoracic inoculation of large amounts of virus [1
]. TE/3'2J/B2 virus-associated mortality was infection route- and mosquito species independent: significantly more Ae. aegypti
died when exposed to TE/3'2J/B2 virus either orally or via injection and Ae. albopictus
and Cx. tritaeniorhynchus
were susceptible to TE/3'2J/B2 virus following intrathoracic injection.
We originally hypothesized that the observed mortality was caused by apoptotic death of a majority of infected cells in the mosquito. FHV has been shown to induce apoptosis in Drosophila
cell culture through the depletion of an intracellular inhibitor of apoptosis [31
]. Apoptosis in alphavirus-infected mosquito cell lines is dependent on the amount of viral RNA and infectious virus produced during infection [32
]. We show that considerably more SINV subgenomic RNA and 100-fold more infectious virus are produced in mosquitoes when B2 protein is expressed during infection. However, apoptosis could not be detected within infected cells in sections of virus-infected mosquitoes (data not shown). It is possible that cell death caused by TE/3'2J/B2 virus is via a non-apoptotic mechanism. Necrosis has been observed in midgut epithelial cells of Culiseta melanura
mosquitoes orally-infected with eastern equine encephalitis virus at times corresponding to peak midgut virus titers [1
]. Electron microscopy of infected cell morphology and detailed analysis of infected mosquito gene expression using microarray analysis may help to more clearly define the mechanism of TE/3'2J/B2 virus-associated mortality.
Behavioral changes have been suggested as a direct result of arbovirus infection [1
]. TE/3'2J/B2 virus infection of the brain and sensory organs may lead to changes in mosquito behavior that could eventually lead to death such as decreased nutrient and water uptake or inability to oviposit. Although not examined here, quantitative observation of behaviors such as blood feeding and oviposition may provide evidence for neurological effects associated with virus infection [36
The salivary glands are an important organ for successful transmission of arboviruses. If TE/3'2J/B2 virus infection leads to cytopathology in the salivary glands, transmission of the virus may be more efficient or could be hindered. It was suggested that SINV-associated pathology in Ae. albopictus
midgut-associated musculature and salivary glands could lead to a decrease in feeding success [4
]. If this is true, then transmission of TE/3'2J/B2 virus could be more efficient as mosquitoes take a longer time to probe the skin prior to imbibing blood. However, if salivation were compromised by virus-induced cytopathology, transmission of virus from the salivary glands would be less efficient due to decreased saliva inoculation volumes.
The B2 protein alone is likely not the mosquito mortality-associated factor. FHV was capable of replication to high titers when injected into Ae. aegypti
mosquitoes, but no mortality was associated with the infection [37
]. Also, transgenic Drosophila
flies that express B2 protein have been shown to be deficient in siRNA-mediated but not microRNA-mediated RNA silencing and are more susceptible to RNA virus infection and virus-associated mortality [16
]. This suggests that B2 protein by itself is not capable of causing mortality in dipterans, but that B2 protein in combination with an infecting RNA virus is capable of protecting virus replication from the influence of RNAi. Additionally, recent experiments show that a SINV expressing a B2 mutant incapable of binding siRNAs does not suppress RNAi in mosquitoes [10
], indicating that the siRNA binding activity of B2 is responsible for the effect observed in our experiments.
The implications of TE/3'2J/B2 virus-associated mortality are two-fold. First, unlike pathogenic viruses that do not require persistent infection of the host, arboviruses may not encode true suppressors of RNAi. B2 protein and many proteins produced by pathogenic plant viruses are dsRNA binding proteins and potent suppressors of the RNAi response. The dsRNA-binding protein NSs of La Crosse virus, an arbovirus transmitted by Ochlerotatus triseriatus
mosquitoes, was initially suggested to be a VSR in mammalian cells, but was later shown to be an interferon antagonist that did not interfere with RNAi in mosquito cells [39
]. Similar conclusions were made with the NS1 protein of influenza A virus, a non-vectored virus [41
]. To our knowledge, there has been no description of an arbovirus-produced protein that is a VSR in mosquito cells, and our data suggest that encoding a VSR may be detrimental to arbovirus transmission.
Second, mortality of TE/3'2J/B2 virus-infected mosquitoes suggests there may be a delicate balance between mosquito immune response and virus replication that allows for the persistent nature of arbovirus infection in the vector. In the model of Semliki Forest virus (genus Alphavirus
) regulation of RNA replication, production of negative-strand RNA, that serves as a template for full-length virus genome and subgenomic RNA, is restricted to the early phase of replication [43
]. Limiting the production of negative-strand RNA may allow for more efficient allocation of cellular resources to progeny virus production and may have evolved to exclude subsequent viruses from establishing infection. It was proposed that regulation of negative-strand RNA synthesis, in turn regulating full length and subgenomic positive-strand RNA, evolved to moderate virus-associated virulence in the mosquito vector [43
]. Our experiments with TE/3'2J/B2 virus suggest that the replicase proteins of SINV, which control the amounts of viral RNA through sequential cleavage of polyprotein complexes, may not be the sole regulators of virus RNA quantities. Rather, there appears to be a balanced effort from both the virus and the mosquito immune system to regulate replication so that the virus can persist in mosquitoes without causing significant adverse effects, allowing the virus to increase its transmission efficiency to a new host. Further studies could compare inhibition of siRNA accumulation at early times during TE/3'2J and TE/3'2J/B2 virus infection and may shed light on the potential cooperation of viral replicase complexes and RNAi response in regulation of virus RNA production in mosquito cells. Identifying key mosquito factors necessary for viral RNA regulation may lead to novel transgenic mosquitoes that over-express these factors and are, therefore, refractory to arbovirus infection.