Aedes albopictus (Skuse) is a potential vector for many arboviruses, including La Crosse virus (LACV), the leading cause of pediatric encephalitis in North America. Bacteria isolated from the midgut and diverticula of field-caught female Ae. albopictus were cultured and identified using 16S ribosomal RNA gene amplification and sequencing. Members of seven and six bacterial families were identified from the midguts and diverticula, respectively, with nearly half of the isolates identified to the family Enterobacteriaceae. Many are related to bacteria identified in other invertebrates, and several may represent previously unknown species or genera. Of the 24 isolated bacteria, 12 (50%) showed a significant reduction in infectivity of LACV for Vero cells. Inhibition of infectivity ranged from 0 to 44% and was not dependent on bacterial classification. The antiviral activity of these bacteria warrants further investigation as an alternate means to interrupt the LACV transmission cycle.
Over the last several decades, the Asian tiger mosquito Aedes albopictus (Skuse) has spread from Southeast Asia to infest parts of Europe, Africa, the Caribbean, and the Americas. This mosquito is a competent laboratory vector of >20 arboviruses, many of which have been isolated from field-caught mosquitoes (Gratz 2004). Ae. albopictus is a natural vector of the dengue viruses and chikungunya virus (Jumali et al. 1979, Pesko et al. 2009). As such Ae. albopictus has garnered attention as an important vector of arboviruses of public health significance and can serve as a model for the transmission of arthropod-borne viruses from many virus families.
La Crosse virus (LACV) is a member of the California serogroup, genus Orthobunyavirus, family Bunyaviridae. It is the leading cause of pediatric encephalitis in North America, with >90% of cases reported in children under 15 yr old. La Crosse encephalitis has a case fatality ratio of ≈0.3% for the ≈70 cases reported yearly (Haddow and Odoi 2009). Specific antivirals and vaccines against LACV do not currently exist. Whereas the principal vector of LACV is Aedes triseriatus (Say) (Watts et al. 1972), Ae. albopictus is a highly competent vector in the laboratory (Grimstad et al. 1989), and LACV has been isolated from field-caught Ae. albopictus collected in areas where human infection with LACV has been reported (Gerhardt et al. 2001).
The mosquito midgut harbors a diverse bacterial flora, and any pathogen entering the midgut must interact with both the midgut and resident bacteria. Once the bacterial flora of the vector species is understood, the organisms present may be genetically manipulated to prevent establishment of a viral infection in the vector (Riehle et al. 2007). Several studies have characterized digestive tract bacteria from mosquitoes, with 8–14 bacterial genera identified from Culex quinquefasciatus Say (Pidiyar et al. 2004), Anopheles gambiae Giles and Anopheles funestus Giles (Lindh et al. 2005), Ae. triseriatus (Demaio et al. 1996), and Anopheles stephensi Liston (Rani et al. 2009). Bacteria from four genera were also shown to inhabit Ae. albopictus from La Réunion (Zouache et al. 2009). However, in a small preliminary study, four bacterial species isolated from Cx. quinquefasciatus midguts did not affect susceptibility to infection with Japanese encephalitis virus (Mourya et al. 2002).
In this study, we characterized the midgut and diverticula bacterial flora of field-caught Ae. albopictus from western Virginia. We then tested these bacteria for an in vitro interaction with LACV to identify bacteria that may be used to interrupt transmission of the virus. We have shown the utility of the Ae. albopictus-LACV model for studying bacteria-virus associations, which may be broadly relevant to other important arboviruses.
Materials and Methods
Mosquito Collection and Isolation of Bacteria.
Wild, host-seeking Ae. albopictus from Radford, VA, were collected with battery-powered aspirators as they landed on the investigators who served as bait; this collection method was approved by the Radford University Institutional Review Board. Every effort was made to collect mosquitoes before they commenced feeding, although after dissection some were noted to have ingested small amounts of blood. The surface of the mosquitoes was sterilized in 70% ethanol for 10 min before dissection (Demaio et al. 1996). In this study, we were only interested in determining which bacteria might be found in the Ae. albopictus digestive tracts and did not characterize differences among individual mosquitoes. To that end, diverticula and midguts from up to nine individuals were dissected under a laminar flow hood into a drop of sterile 0.85% saline, pooled in 50 μl of sterile 0.85% saline, and homogenized with plastic pestles. A total of 25 μl of the homogenate was spread on nutrient agar (BD Biosciences, Sparks, MD) plates and incubated at room temperature for 2–3 d. Distinct colonies were individually picked and placed in 5 ml of nutrient broth (BD Biosciences) and incubated at room temperature without shaking for 2–3 d, or until turbidity was noted.
Extraction of Genomic DNA From Bacterial Samples.
A total of 1 ml of culture was pelleted at 16,000 × g for 2 min, resuspended in 400 μl of lysis solution (47 mM ethylenediaminetetraacetic acid, 25 mg/ml lysozyme), and incubated at 37°C for 1 h. A volume of 8 μl of proteinase K (20 mg/ml) was then added to the suspension and incubated an additional hour at 50°C. Genomic DNA was then extracted using the Wizard SV Genomic DNA kit (Promega, Madison, WI), according to the manufacturer's directions, and eluted in 200 μl of nuclease-free water.
The 16S Ribosomal Gene Amplification and Sequencing.
A volume of 2 μl of genomic DNA served as template for polymerase chain reaction (PCR) amplification in a 25-μl reaction containing 1× Taq polymerase buffer, 0.4 mM each dNTP, 50 pmol each primer, and 2.5 U Taq polymerase (Denville Scientific, Metuchen, NJ). The PCR program was as follows: 94°C for 5 min, followed by 35 cycles of 94°C for 30 s, 57°C for 30 s, 72°C for 90 s, and a final extension step of 72°C for 10 min. The universal primers used in this study (8f, 5′-AGAGTTTGATIITGGCTCAG-3′ and 1501r, 5′-CGGITACCTTGTTACGAC-3′) were previously used to identify bacteria from An. gambiae and An. funestus (Lindh et al. 2005). Amplicons were gel purified using the Wizard SV Gel and PCR Clean-Up System (Promega, Madison, WI) per the manufacturer's directions. A total of 1 μl of DNA from each sample was quantified using the Thermo Scientific Nanodrop 1000 spectrophotometer. The GenomeLab DTCS Quick Start Kit (Beckman Coulter, Fullerton, CA) was used to prepare sequencing reactions using 25 fmol of genomic DNA and the primers used for amplification, according to the manufacturer's directions. Sequencing was performed on a Beckman Coulter CEQ-8000 DNA sequencer, and electropherograms were visually inspected for base-calling accuracy. Sequences were submitted to GenBank, and a nucleotide BLAST search was performed for identification of samples.
Bacteria and Virus Interactions.
Bacteria were assayed in vitro for their ability to reduce LACV infectivity for African green monkey kidney (Vero) cells, as follows. A total of 1 ml of bacterial culture diluted to 0.5 McFarland standard (≈108 cells/ml) was pelleted at 14,000 × g for 1 min, resuspended in 1 ml of medium 199 (Mediatech Cellgro, Herndon, VA) containing 105 plaque-forming units (PFUs) of LACV, and incubated at room temperature for 30 min. Escherichia coli served as a control bacterial species. The bacteria, with any adherent virus, were removed by centrifugation at 14,000 × g for 1 min. A total of 300 μl of 10-fold serial dilutions (10−1, 10−2, and 10−3) of supernatant was added to confluent monolayers of Vero cells in six-well plates (Corning, Lowell, MA) and incubated for 1 h at 37°C and 5% CO2, with gentle rocking every 15 min. The supernatant was replaced with 1× medium 199, 0.8% gum tragacanth, 5% fetal bovine serum, and antibiotics (1.25 mg of amphotercin B and 0.75 mg of gentomycin per 100 ml medium), and incubated for 3 d at 37°C with 5% CO2. Plaques were visualized with 1 mg/ml crystal violet in 10% formalin. Plaques were counted in the well that gave a countable number, which varied depending on the inhibitory effect of the bacteria. Plaque numbers were compared with virus controls treated in the same manner without exposure to bacteria, and all experiments were performed in triplicate in groups of approximately six bacteria and a virus control run each time. Significance was determined by analysis of variance within test groups, with Dunnett's test performed post hoc using the no-bacteria virus as the control (JMP7 software, SAS Institute 2007).
Isolation and 16S PCR and Sequencing of Bacterial rRNA Gene.
Fifteen bacterial colonies from Ae. albopictus midguts and 11 colonies from the diverticula were selected based on individual colony morphology. Based on BLAST homology, the bacteria isolated from the midgut belong to the following bacterial families: Enterobacteriaceae (eight isolates), Enterococcaceae (one), Pseudomonadaceae (one), Bacillaceae (one), Flavobacteriaceae (one), Acetobacteraceae (one), and Sphingobacteriaceae (one). Bacteria from the diverticula fall into the following families: Enterobacteriaceae (three isolates), Leuconostocaceae (two), Microbacteriaceae (two), Paenibacillaceae (one), and Enterococcaceae (one). Specific identities based on the highest-scoring BLAST result with a full binomial designation are given in Table 1. Three isolates (AaMG6, AaD2, and AaD6) could not be sequenced because of failure of the PCR to amplify the 16S gene.
Bacteria and Virus Interactions.
In vitro interaction studies were performed to determine whether the bacteria isolated from Ae. albopictus digestive tissues could inhibit cell culture infection by LACV. Two isolates did not survive passage in the laboratory. Of the 24 isolates tested, 12 showed a significant reduction in the number of plaques that formed on Vero cells compared with virus not exposed to bacteria (Fig. 1). Three of the four test groups of bacteria displayed a significant reduction in LACV infectivity (group 1 [AaMG1, 3, 4, 5, 6, 7]: F = 14.7133; df = 6, 14; P < 0.0001; group 2 [AaMG8, 9, 10, 14, 18, 19, 20, E. coli]: F = 4.7046; df = 8, 17; P = 0.0035; group 3 [AaD1, 4, 5, 7, 9, 10]: F = 15.2131; df = 6, 14; P < 0.0001; group 4 [AaD2, 3, 6, 8]: F = 1.3565; df = 4, 10; P = 0.3158). The 12 bacteria that inhibited LACV infection of Vero cells were as follows (P values in parentheses): AaMG1 (0.267), AaMG3 (0.0048), AaMG4 (0.0007), AaMG5 (<0.0001), AaMG6 (<0.0001), AaMG7 (<0.0001), AaMG9 (0.0331), AaMG18 (0.0238), AaD5 (0.0401), AaD7 (0.0147), AaD9 (<0.0001), and AaD10 (0.0056).
The reduction in LACV infectivity ranged from no change to 44%. Four bacteria reduced the number of PFUs by ≈40%, as follows: AaMG5 (38.2%), AaMG6 (36.2%), AaMG7 (41.9%), and AaD9 (43.6%). The remaining bacteria reduced infectivity by <28%. Bacteria inhibiting LACV infectivity represented at least seven different bacterial families, including Gram-negative rods and Gram-positive cocci and rods. The enteric bacterium E. coli was used as a control and resulted in a nonsignificant (P = 0.9940) reduction in PFUs of 3.9% (Fig. 1).
To our knowledge, this is the first study to demonstrate an interaction between bacteria isolated from a mosquito and a virus transmitted by that mosquito. One half of the bacteria we isolated from Ae. albopictus significantly reduced infectivity of LACV; similarly high numbers (81%) of poliovirus-inhibiting bacteria were identified in fresh-water samples (Ward et al. 1986). Only a single study has previously attempted to block arbovirus infection with bacteria from a vector, and no significant interaction was found in vivo (Mourya et al. 2002). We did not identify a bacterium that completely blocks LACV infection, possibly because of our experimental design. We incubated the bacteria with the virus for 30 min because, in a newly blood-fed mosquito, this is the approximate time required for the peritrophic matrix to effectively block the virus from accessing the midgut cells (Jacobs-Lorena 1996) and for the mosquito to concentrate the blood meal and excrete most of its water (Wheelock et al. 1988). Bacteria with antipoliovirus activity inhibit the virus by ≥90% after a 2-day incubation (Ward et al. 1986), suggesting that a longer incubation period might increase viral inhibition. Hence, it is unlikely that a single bacterium could be used to prevent infection of the vector, given that a 44% reduction would likely still leave the vector infected. However, mosquitoes infected with a lower dose may take longer to become infective (Mahmood et al. 2006) and may therefore not survive the extrinsic incubation period (Anderson and Rico-Hesse 2006). We have only demonstrated inhibition of LACV in a cell culture assay, and we are currently investigating whether the bacterial inhibition of virus infectivity also occurs in living mosquitoes coinfected with these bacteria and LACV.
Our study has identified bacteria belonging to 10 bacterial families, with many of them coming from the family Enterobacteriaceae. At least 14 different genera were isolated based on 16S rRNA similarity to known bacteria. This diversity is similar to that seen in several other mosquito species. Bacteria from 20 genera were identified by biochemical and gas-chromatography analyses from An. gambiae and An. funestus (Straif et al. 1998). Fourteen total genera were also identified by 16S rRNA sequencing from An. gambiae and An. funestus, using both culture-dependent and -independent methods (Lindh et al. 2005). Similarly, in Cx. quinquefasciatus, nine genera were identified by culture, and several unidentified bacteria from two taxa (γ-proteobacteria and Firmicutes) were identified by cloning and sequencing of the 16S genes from total DNA (Pidiyar et al. 2004). In field-caught An. stephensi females, 97% of the bacteria isolated from the midgut belonged to the γ-proteobacteria (Rani et al. 2009). Finally, bacteria belonging to the Bacillus and Serratia genera and at least one yeast species were identified from the ventral diverticula of Aedes aegypti (L.), the yellow fever mosquito (Gusmao et al. 2007).
In a study of midgut bacteria in Ae. triseriatus, Demaio et al. (1996) identified a combined 13 genera from fourth-instar larvae, newly emerged adults, field-caught adults, and blood-fed adults. Common isolates between Ae. triseriatus and Ae. albopictus are members of the Bacillus, Erwinia, Pseudomonas, and Enterobacter genera. Several of these isolates in our study demonstrated a significant interaction with LACV, suggesting that a common control strategy centered on one or more of these bacteria could be devised. Of note, bacteria were isolated from midguts of freshly emerged adults (Demaio et al. 1996, Drancourt et al. 2000, Moll et al. 2001), indicating that the remodeling of the mosquito body that occurs during the pupal stage, while greatly reducing bacterial numbers, does not completely destroy all bacteria. Furthermore, mosquitoes emerging from pupae may acquire bacteria in the water in which they develop (Lindh et al. 2008). Bacteria resident in the midgut can then undergo a sizeable population expansion when the mosquito acquires a blood meal (Demaio et al. 1996, Pumpuni et al. 1996), which would increase the number of cells available to prevent virus infection. Thus, developing a bacterial mechanism, similar to the use of Bti to control mosquito populations, to interfere with LACV infection of the mosquito host may prove fruitful if applied during the larval stage.
The species listed in Table 1 represent the best-scoring BLAST match that has a full binomial designation. Many of these were isolated from environmental sources. However, other high-scoring bacteria have been identified from other invertebrate species, including many isolated from parts of the digestive tract. Indeed, many of the best matches were to bacteria that had not been identified to genus and/or species and were only identified by culture-independent methods (i.e., extraction of total DNA from a tissue and amplification, cloning, and sequencing multiple clones of the 16S rRNA gene). A list of similar sequences obtained from the best 25–30 BLAST matches is given in Table 2. These isolates may represent either groups of symbiotic bacteria common to many insects or simply organisms that are common in the environment and frequently establish replicating populations in insect digestive tracts. Furthermore, a number of our isolates may represent new species and/or new genera. It has been suggested that 16S rRNA homology of 99+% is indicative of the isolate belonging to a species and 97+% represents a member of a genus (Drancourt et al. 2000). Several of our isolates have identities <97%, and thus need to be further characterized using methods other than bioinformatic classification to definitively establish a full taxonomic identification.
Mosquito populations from disparate geographic areas are known to vary in their vector competence. Ae. triseriatus exhibits this phenomenon for LACV (Grimstad et al. 1977), and Ae. albopictus populations have been shown to vary in their ability to transmit dengue virus (Boromisa et al. 1987). Whereas this may be the result of genetic variation among these populations, it would be interesting to determine whether the midgut microbial flora covaries with vector competence. We are currently developing a PCR-based assay to identify bacterial genera or species to determine the frequency with which they are found in natural populations and whether geographic variation in the midgut flora is evident.
The identification of bacteria common to the digestive tracts of mosquitoes is the first step in developing a means to biologically control pathogens they transmit. An effective paratransgenic strategy is currently being used against Trypanosoma cruzi, the parasite that causes Chagas disease (Durvasula et al. 1997), and the concept has been demonstrated to kill Plasmodium parasites in Anopheles mosquitoes (Riehle et al. 2007). Both of these strategies have involved introducing an antiparasitic gene into the bacteria. Our study demonstrates that some bacteria may have an inherent virus-inhibiting ability that could be supplemented by genetic manipulation to prevent viral infection of the vector mosquito, although further characterization of the antiviral mechanism and improving its efficacy are required. Thus, blocking infection of the vector using these bacteria would prevent transmission to and disease in the human population.
We kindly thank Sally Paulson for providing the LACV used in this study, Rajeev Vaidyanathan for comments on the manuscript, and Jeremy Wojdak and Fred Singer for statistical assistance. This work was supported by a seed grant to J.R.A. from the Radford University Office of Sponsored Programs and Grants Management.