OUP user menu

Bacterial diversity of field-caught mosquitoes, Aedes albopictus and Aedes aegypti, from different geographic regions of Madagascar

Karima Zouache, Fara Nantenaina Raharimalala, Vincent Raquin, Van Tran-Van, Lala Harivelo Ravaomanarivo Raveloson, Pierre Ravelonandro, Patrick Mavingui
DOI: http://dx.doi.org/10.1111/j.1574-6941.2010.01012.x 377-389 First published online: 1 March 2011

Abstract

Symbiotic bacteria are known to play important roles in the biology of insects, but the current knowledge of bacterial communities associated with mosquitoes is very limited and consequently their contribution to host behaviors is mostly unknown. In this study, we explored the composition and diversity of mosquito-associated bacteria in relation with mosquitoes' habitats. Wild Aedes albopictus and Aedes aegypti were collected in three different geographic regions of Madagascar. Culturing methods and denaturing gradient gel electrophoresis (DGGE) and sequencing of the rrs amplicons revealed that Proteobacteria and Firmicutes were the major phyla. Isolated bacterial genera were dominated by Bacillus, followed by Acinetobacter, Agrobacterium and Enterobacter. Common DGGE bands belonged to Acinetobacter, Asaia, Delftia, Pseudomonas, Enterobacteriaceae and an uncultured Gammaproteobacterium. Double infection by maternally inherited Wolbachia pipientis prevailed in 98% of males (n=272) and 99% of females (n=413); few individuals were found to be monoinfected with Wolbachia wAlbB strain. Bacterial diversity (Shannon–Weaver and Simpson indices) differed significantly per habitat whereas evenness (Pielou index) was similar. Overall, the bacterial composition and diversity were influenced both by the sex of individuals and by the environment inhabited by the mosquitoes; the latter might be related to both the vegetation and the animal host populations that Aedes used as food sources.

Key words
  • bacterial community
  • DGGE
  • quantitative PCR
  • Wolbachia

Introduction

All arthropod pests and vectors harbor a number of commensal and mutualistic microorganisms that have an impact on the ecology and behavior of their hosts (Buchner, 1965; Buchner, Moranet al.,2008; Buchner, Moyaet al.,2008). Indeed, it is well-known that microbial communities associated with insects can contribute to host reproduction and survival, community interactions, protection against natural enemies and vectorial competence (Buchner, 1965;Moranet al.,2008;Moyaet al.,2008;Gottliebet al.,2010;Oliveret al.,2010). However, such extended phenotypes were mostly shown in phytophagous arthropods, whereas research on hematophagous insects has been limited. Historically, this unawareness was partly due to the lack of data on the composition of native bacterial communities associated with the later group of insects. A few studies have, however, reported a number of bacterial species in some medically important hematophagous insects. A relevant example is the tsetse fly Glossina, which harbors the secondary symbiont Sodalis glossidinius, suspected to enhance vectorial competence (Cheng & Aksoy, 1999; Aksoy & Rio, 2005; Farikouet al.,2010). More recently, bacteria belonging to genera Enterobacter, Enterococcus and Acinetobacter were isolated in Glossina palpalis palpalis, but their role in the tsetse fly biology remains to be determined (Geigeret al.,2009).

Mosquitoes are vectors of a large number of animal and human pathogens, including parasites and viruses. During the last few years, Madagascar and other neighboring islands have experienced severe epidemics of arboviruses, notably chikungunya and dengue. The species Aedes albopictus and Aedes aegypti have expanded over the Indian Ocean Islands (Fontenille & Rodhain, 1989;Salvan & Mouchet, 1994;Delatteet al.,2008;Sanget al.,2008;Bagnyet al.,2009,ab;) and have been identified as the primary vectors responsible for these outbreaks (Schuffeneckeret al.,2006; Vazeilleet al.,2007; Delatteet al.,2008; Ratsitorahinaet al.,2008; Sanget al.,2008). As for all insects, the successful spreading of mosquitoes worldwide might be partly linked to their symbiosis with microorganisms, notably with bacteria. However, little is known about the current composition of mosquito-associated microbial communities, and consequently, their potential contribution to the host behaviors is mostly ignored. Investigations have been performed to screen bacterial communities in mosquitoes reared under laboratory conditions or collected in the fields, using culture and nonculture methods. These studies have focused mainly on the gut microbial communities of two mosquitoes, Anopheles and Culex, and these revealed the presence of diverse bacterial groups including known genera such as Acinetobacter, Aeromonas, Asaia, Bacillus, Enterobacter, Flavobacterium, Lactoccocus, Pantoea, Pseudomonas, Microbacterium, Staphylococcus and Stenotrophomas (Pumpuniet al.,1996;Straifet al.,1998;Pidiyaret al.,2004;Faviaet al.,2007;Tereniuset al.,2008;Raniet al.,2009). These surveys highlighted that the relative abundance and the composition of mosquito-associated bacteria varied depending on the developmental stages and laboratory-reared or wild targeted populations. For Aedes mosquitoes, Demaio (1996) reported the occurrence of cultivable bacteria belonging to Enterobacter, Klebsiella, Pseudomonas and Serratia in the midgut of wild Aedes triseriatus. Most recently, this inventory was extended to Acinetobacter, Asaia, Bacillus, Comamonas, Delftia, Pantoea and Wolbachia detected in A. aegypti or A. albopictus, reared in insectaries (Gusmaoet al.,2007, 2010; Crottiet al.,2009; Zouacheet al.,2009b).

The aim of this study was to survey the composition of bacterial communities associated with wild Aedes mosquitoes and to explore whether the bacterial diversity is related to host ecology. To that end, we used culture and nonculture methods to describe the bacterial composition and diversity of A. albopictus and A. aegypti, males and females, caught from ecologically contrasted regions of Madagascar.

Materials and methods

Location and characteristics of survey areas

The sampling regions were selected for their different ecoclimatic characteristics (Table 1) and because they were sites of chikungunya or/and dengue epidemics (Ratsitorahinaet al.,2008; Randrianasoloet al.,2010), although no such virus infection was detected in the sampled population (data not shown). Vegetation and animals of the sampling sites are reported in Table 1.

View this table:
1

Ecological characteristics of mosquito Aedes sp. capture sites

The Analamanga region (Tsimbazaza Park, Ambohidratrimo and Ankazobe) is located in the centre of Madagascar at an altitude of 1200–1500 m. This region has a highland climate with two seasons: a hot and rainy period from October to March (21 °C average and about 200 mm of precipitation per month), followed by a cold and dry period (with temperatures down to 10 °C and rainfall not exceeding 20 mm month−1). The mean relative humidity in this region is high (77.5% in 2008). The Zoological and Botanic Park of Tsimbazaza is located in the centre of Antananarivo town at 1250 m altitude. Ambohidratrimo Hill is located 25 km to the northwest of Antananarivo with an altitude of 1300 m. Ankazobe is 80 km from the northern limits of Antananarivo at 1500 m altitude. This site is transitional, connecting the central and the western regions. It is surrounded by the nature reserve of Ambohitantely. The climate is wetter and colder than the other towns in the centre.

The Atsinanana region (Toamasina) is on the east cost of Madagascar at sea level. The climate is particularly hot and humid: the mean annual rainfall is about 3200 mm with rain all year, the mean annual temperature is 25 °C with a minimum of 18 °C from June to August, and relative humidity is around 87% all year.

The Boeny region (Mahajanga, Andranofasika and Ankarafantsika natural reserve) has an arid tropical climate characterized by a warm summer (mean temperature of 27 °C) with moderate rainfall (mean precipitation is about 400 mm year−1) and high relative humidity (81%) from November to March. Mahajanga is in the northwest of Madagascar, 600 km from Antananarivo in Edge Sea at a 22 m altitude. There are mango trees, bushes and flowers near dwellings in the town. The Andranofasika village is about 110 km from Mahajanga town and 5 km from the National Park of Ankarafantsika.

Mosquito collection

Mosquitoes were collected between February and May 2008. Two methods were used to capture adult mosquitoes: during peaks of biting activity, a tube was used to capture insects landing on the human body or nets were used to capture insects near the grass. Aedes specimens, males and females, were identified using morphological characteristic keys (Ravaonjanahary, 1978). Captured adults were separated according to species and sex and stored in tubes containing silica gel. For each tube, the species, date, location, geographical position, and type of site was recorded. Only nonblooded mosquitoes were used for the analysis.

Bacterial isolation

Only live mosquito specimens from the field were used. Individuals were anaesthetized at 4 °C, rinsed three times in sterile water, surface disinfected in 70% ethanol for 10 min and rinsed five times in sterile water and once in sterile 0.8% NaCl. Two adult mosquitoes per sample were crushed in 150 μL sterile 0.8% NaCl. Homogenates (10 μL) were streaked on plates of modified Luria–Bertani and PYC agar media (Zouacheet al.,2009b). After incubation at 26 °C, single distinct colonies were reinoculated onto fresh agar plates of the corresponding medium. Colonies were streaked to check for purity and stored in 25% glycerol at −80 °C until use.

Genomic and plasmid DNA extractions

Mosquitoes were surface disinfected as described above, and then individually crushed in 200 μL of extraction buffer (2% hexadecyltrimethyl ammonium bromide, 1.4 M NaCl, 0.02 M EDTA, 0.1 M Tris pH 8, 0.2% 2-β-mercaptoethanol) heated to 60 °C. Homogenates were incubated for 15 min at 60 °C and proteins were extracted with chloroform : isoamyl alcohol (24 : 1, v/v). DNA was precipitated with isopropyl alcohol, pelleted by centrifugation for 15 min at 12 000 g, washed with 75% ethanol, dried and then dissolved in 30 μL of sterile water.

For bacterial isolates, genomic and plasmid DNA were extracted using the DNeasy Tissue Kit and QIAprep Spin Miniprep Kit, respectively (Qiagen, France).

Diagnostic PCR, amplified ribosomal DNA restriction analysis (ARDRA) and quantitative PCR amplification

Diagnostic PCR amplification was performed with primers listed in Table 2 using a T Gradient Thermocycler (Biometra, France). Reactions (25 or 50 μL volumes) contained genomic DNA template (1 μL), 200 μM of each dNTP, 500 nM of each primer, 0.025 mg mL−1 of T4 gene 32 protein (Roche, France) and 0.5 U of Expand polymerase in 1 × reaction buffer (Roche). PCR products were purified using QIAquick PCR Purification Kit (Qiagen). ARDRA was performed to screen the rrs genes of bacterial isolates in 20 μL reactions containing 200 ng of DNA, 1 × Buffer Tango and 10 U of each endonuclease RsaI and HhaI (Fermentas, France). DNA fragments were separated on 1% or 2% agarose gels stained with ethidium bromide.

View this table:
2

Primers used in this study

Real-time quantitative PCR was performed using the LightCycler apparatus (Roche). The 20-μL reaction mixture contained 1 × LightCycler DNA Master SYBR Green I (Roche), primers at 300 nM (for wsp) or 200 nM (for actin) (see Table 2) and 10 ng of template DNA. The amplification program was 10 min at 95 °C followed by 40 cycles of 15 s at 95 °C, 1 min at 60 °C and 30 s at 72 °C. Standard curves were constructed using a dilution series (101–108 molecules) of the pQuantAlb plasmid (Tortosaet al.,2008) containing wsp and actin fragments.

Denaturing gradient gel electrophoresis (DGGE)

Ingeny PhorU (Apollo Instruments, Compiègne, France) systems were used for DGGE analysis of the V3 PCR products as published (Zouacheet al.,2009a). The 6% acrylamide gels contained a linear chemical gradient of urea and formamide from 35% to 65% urea and 40% deionized formamide (v/v). PCR products (2 μg) were run in 1 × TAE at 60 °C for 17 h at 100 V, and then gels were immersed in SYBR Green for 30 min, rinsed in distilled water and photographed under UV. Bands were excised, washed three times with sterilized water and then 30 μL of water was added to the tubes, which were heated to 60 °C for 30 min and kept overnight at 4 °C. The eluate (2 μL) was used for PCR amplification, and then amplicons were cloned and sequenced as described below.

Cloning and sequencing

PCR products were purified using the MinElute PCR Purification Kit (Qiagen), and cloned in the PCR®2.1-TOPO® vector according to the TOPO TA 2.1 Kit (Invitrogen, France). Clones containing DNA inserts were sequenced at Genoscreen (Lille, France). Sequences were analyzed with the blastn program at NCBI (http://www.ncbi.nlm.nih.gov/).

DGGE fingerprints and statistical analyses

Each band was considered as an operational taxonomic unit (OTU). Images acquired with Fisher Bioblock Scientific System (Fisher, Ilkirch, France) were analyzed using gelcompar II version 5.1 packages (Applied Maths, Kortrijk, Belgium). The software carries out a density profile analysis for each lane and calculates the relative contribution of each band to the total band intensity in the lane, with a reference pattern included in all gels. Relative intensity in the profile of each band or OTU (Pi) was calculated by the relative area under the peak in the profile (Pi=ni/N, where ni is the area under the peak i, and N is the sum of the areas for all peaks within the profile). The relative intensity of each band was used to calculate (primer v6 software) Shannon–Weaver (H′=−ΣPi log Pi where Pi=ni/N) and Simpson (1−λ′=1−{ΣiNi(Ni−1)}/{N(N−1)}) diversity indices. We estimated the evenness of the numbers of bacterial species in each sample using Pielou's index (J′=H′/logS, where logS=Hmax). Statistical analyses were performed using splus software and/or r packages.

Results

Collection of mosquitoes and bioecology

To collect Aedes adult mosquitoes, larval development sites were used as indicators (Table 1). Larvae refuges of A. aegypti consisted of natural sites (holes in trees or rocks, wet leaves of bamboo or palm trees and coconuts) outside cities and villages, whereas larvae refuges of A. albopictus were natural and artificial sites (containers or flowerpots) near habitations. Adults were collected around these larvae breeding sites.

Both species were found to be exophilic (which do not enter inside habitations). Except for natural reserves of the Ankarafantsika and the Andranofasika villages, A. albopictus was predominant in all sites sampled (Table 1). Indeed, a total of 137 females and 35 males were caught in the neighborhood of the tourist attraction site of Ambohidratrimo, named ‘Le Palais des Rois’. In Tsimbazaza Park, the presence of bamboo, bushes and many animals creates a favorable environment for A. albopictus development; a total of 823 females and 62 males were captured in this site. In contrast, only 93 females and 95 males were trapped in Ankazobe that has a colder climate (Table 1). The rainy and hot climate throughout the year in Toamasina allows uninterrupted development of A. albopictus, but it was difficult to capture adults during the active rainy season: 320 females and 30 males were trapped in the town itself. The two Aedes species were found in the Boeny region, but in different sites (Table 1). Aedes albopictus was predominant in urban areas, with 290 females and 20 males. Aedes aegypti was the major species found in village (Andranofasika) and forest zones (Ankarafantsika), although few individuals were caught: 13 females and 12 males.

Cultivable bacteria

To search for cultivable bacteria in mosquitoes, insects originating from the Boeny region were chosen as the two species under study were both present in the area: A. albopictus at the Mahajanga site and A. aegypti at the Ankarafantsika site. For the two media used, 22 colony types were obtained from males and 10 from females of A. albopictus. Only four colony types were recovered from A. aegypti females. Two to four representatives of each colony type were used for genomic DNA extraction and PCR amplification of the rrs gene. ARDRA of entire rrs gene amplicons revealed a total of 13 distinct patterns (not shown). Sequencing of the rrs gene of each isolate and blastn analysis allowed identifying two phyla: Proteobacteria and Firmicutes (Table 3). Bacteria belonging to the genus Bacillus were present in all the specimens of both sexes and species. In addition, one isolate from an A. albopictus female was an Agrobacterium sp. whereas isolates of the genera Acinetobacter and Enterobacter were found in A. albopictus males. For all isolates, the sequence similarities were between 98% and 100% with respect to the rrs sequences of type strains reported in databases.

View this table:
3

Phylogenetic affiliation of isolates and sequences obtained from Aedes sp.

DGGE fingerprints and phylogenetic affiliation of bacterial sequences

To investigate the whole bacterial community of the two Aedes species, PCR-DGGE fingerprints of hypervariable V3 regions were produced. For each sampling site, females and five males (four males for A. aegypti) were analyzed individually. DGGE profiles varied between individuals of the same sex whether from the same site or not (Fig. 1). Banding patterns also differed between females and males of both A. albopictus and A. aegypti. To compare the DGGE profiles better, we analyzed them with gelcompar software and then by principal component analysis (PCA) using r software. In terms of the bacterial communities they host, females and males of A. albopictus from all collection sites are distinct, the first two axes explaining >43.8% of the total variability in PCA (Fig. 2).

1

DGGE profiles of bacterial communities of Aedes albopictus (a–e) and Aedes aegypti (f) from different regions of Madagascar. W, Wolbachia strain wAlbB from the Aa23 cell line used as an internal gel migration control. L, ladder used as an external gel migration control. Numbers correspond to cloned and sequenced bands (Table 4).

2

Principal component analysis (PCA) of male and female Aedes albopictus from the same collection site (a–d). F, females; M, males. (a) PCA of individuals from Ambohidratrimo. (b) PCA of individuals from Ankazobe. (c) PCA of individuals from Toamasina. (d) PCA of individuals from Tsimbazaza Park. Individuals are represented by dots. Individuals of the same sex are encircled. The percentage indicated within parentheses corresponds to the variance explained by each principal component.

To explore whether the mosquitoes' environment influences the bacteria they host, PCA was performed on the DGGE band profiles from males and females separately. For males (Fig. 3a and c), the type of vegetation (Table 1) may explain the differences because (1) individuals from urban areas (Mahajanga, Antananarivo and Toamasina) characterized by bushes and fruit trees are different from those from suburban areas (Ambohidratrimo and Ankazobe) surrounded by bamboo (PCA1, 17% of total variability); and (2) individuals from Ankazobe that is mainly a natural habitat are distinct from those from the touristic site of Ambohidratrimo (PCA3, 9.9% of variability). Although weaker (PCA3, 9.8% of total variability) for females, in addition to vegetation, differences between sites (Fig. 3b and c) can be linked to the hosts available to bite (Table 1). For instance, poultry were currently found in Toamasina and Ankazobe whereas Mahajanga is the only site where there is extensive ovine and bovine rearing. In contrast, Tzimbazaza Park is well-frequented by tourists and hosts a diverse range of vertebrates. In addition to humans, Ambohidratrimo may host natural fauna.

3

Principal component analysis (PCA) of Aedes albopictus collected from different sites in Madagascar. Individuals are represented by dots. Individuals from the same collection site are encircled. Percentages correspond to the variance explained by each principal component (PC). (a) PCA of A. albopictus females. AmF, females from Ambohidratrimo (birds, reptiles); AnF, females from Ankazobe (poultry); MF, females from Mahajanga (ovine and bovine); PaF, females from Tsimbazaza Park (lemurs, birds and reptiles); TF, females from Toamasina (poultry). The two axes explain 17% (PC1) and 9.9% (PC3) of the total variability. (b) PCA of A. albopictus males. AmM, males from Ambohidratrimo (bamboo hedge); AnM, males from Ankazobe (bamboo forest); MM, males from Mahajanga; PaM, males from Tsimbazaza Park; TM, males from Toamasina (vegetation of the three cities corresponds to bushes and fruit trees). The two axes explained 9.8% (PC3) and 8.2% (PC4) of the total variability. (c) Map of Madagascar showing sites of Aedes collection. The abbreviations used in the map, after names of collection sites, correspond to those used in PCA panels.

To identify the bacterial community in these mosquito samples, representative DGGE bands were excised from the gel, cloned and sequenced as numbered in Fig. 1. The V3 fragment size obtained varies from 165 to 196 bp, giving only an indication of bacterial phylogenetic affiliation. blast analyses indicated that sequences belonged to Bacteroidetes (2.6% of the sequenced bands), Firmicutes (10.5%) and Proteobacteria (86.9%). At the genus level, sequences were affiliated mostly with Acinetobacter, Asaia, Pseudomonas and an uncultured Gammaproteobacterium (Table 4). Some other bacteria detected included the genera Bradyrhizobium sp., Delftia sp., Herbaspirillum sp., Rhizhobium sp. and Stenotrophomonas sp. as well as members of the Enterobacteriaceae (uncultured Citrobacter sp., Enterobacter sp., Pantoea sp., Shigella sp. and Yokenella sp.). An uncultured Streptococcaceae bacterium and members of the genus Staphylococcus were also identified (Table 4). As expected, sequences of the control bands corresponding to Wolbachia V3 amplicons were seen exclusively in A. albopictus (Fig. 1a–f).

View this table:
4

Phylogenetic affiliation of sequences obtained from Aedes sp. in DGGE analysis

Bacterial diversity analysis

We evaluated the bacterial diversity and evenness in A. Albopictus from the different sampling sites. Considering all the sampling sites, the Shannon–Weaver (H′) index varied from 1.16 to 2.45 and the Simpson diversity (1−λ′) index varied from 0.63 to 0.89. The Pielou's index (J′) was between 0.80 and 0.86 (Table 5). Statistical analyses for all indices showed that there was a significant difference (P<0.01, Tukey) linked to the sex for individuals from Tsimbazaza Park only. In addition, Shannon–Weaver and Simpson diversity indices varied between sampling sites. In particular, significant differences (P<0.01, Tukey) were found between samples from Ankazobe, Mahajanga and Tsimbazaza Park. The regions Ambohidratrimo and Toamasina had intermediary values (Table 5). No differences in evenness between sampling sites were observed with Pielou's index.

View this table:
5

Diversity indices and evenness values of Aedes albopictus

Wolbachia prevalence and density in A. albopictus

Usually, A. albopictus harbors two Wolbachia strains named wAlbA and wAlbB (Sinkinset al.,1995). Diagnostic PCR using wsp primers against the subset (685 of a total of 1905) of wild A. albopictus revealed double infection in 99% females (n=413) and 98% males (n=272); four females and six males found were singly infected with wAlbB strain (not shown).

Wolbachia's density was estimated by quantitative PCR targeting the wsp gene with primers designed to be strain specific toward wAlbA and wAlbB strains and the host gene encoding the cytoskeleton protein actin (Table 2). The relative numbers of bacterial genes per host gene are given as the copy number ratio of Wolbachia wsp to host actin. Overall, the relative numbers of the wAlbA strain varied from 0 to 5.19 per female (Fig. 4) and from 0 to 1.67 × 10−2 per male (Supporting Information, Fig. S1). The wAlbB density was also extremely variable, between 4.56 × 10−4 and 5.16 per female (Fig. 4) and from 9.42 × 10−3 to 1.16 per male (Fig. S2). In general, Wolbachia strains wAlbA and wAlbB were significantly (P<0.05, Tukey) more abundant in females than in males. Interestingly, Wolbachia's density in females varied depending on either the bacterial strains present or the mosquitoes' geographical origin (Fig. 4). The relative density of strain wAlbA was significantly higher (P<0.05, Tukey) than that of wAlbB in females from Tsimbazaza Park only. The densities of each Wolbachia strain in females were compared between sampling sites. Results indicated that wAlbA strain was more abundant (P<0.05, Tukey) in Tsimbazaza Park than in Mahajanga, whereas wAlbB strain predominated (P<0.05) in Ambohidratrimo compared with Mahajanga and Tsimbazaza Park. Differences in Wolbachia densities in males were not statistically significant between sites, probably due to a high interindividual variability.

4

Relative density of Wolbachia in Aedes albopictus females from different sites in Madagascar. The relative numbers of Wolbachia are given as the copy number ratio of wsp to host actin. wAlbA (black) and wAlbB (grey) strains were measured in five female individuals per sampling site. Bars indicate SEs.

Discussion

Our data illustrate the current distribution and preferential habitats of A. albopictus and A. aegypti, two major mosquito vectors of arbovirus, in seven localities of Madagascar (Table 1 and Fig. 3c). Aedes albopictus was found to be predominant in urban and suburban areas, whereas A. aegypti specimens were exclusively recovered in sylvan habitats showing weakly anthropophilic behavior (Table 1). In contrast to previous reports showing a high prevalence of A. aegypti in Mahajanga (Ravaonjanahary, 1978; Fontenille & Rodhain, 1989), we noted the current dominance of A. albopictus in this region. These data are in line with what is known on the undercurrent expansion of A. albopictus in Indian Ocean Islands and worldwide, affecting the density of sister taxon A. aegypti concomitantly (Salvan & Mouchet, 1994;O'Mearaet al.,1995;Delatteet al.,2008;Bagnyet al.,2009a, b, c;Paupyet al.,2010.

To examine whether the environment inhabited by the mosquitoes influenced the diversity of bacterial communities associated with wild mosquitoes, DGGE analysis was performed. Profiles varied between individuals and capture sites. This variation could be linked to environmental features, suggesting that some bacterial species that colonize mosquitoes may originate from the environment. Thus, vegetation used as food sources or resting and potential hosts for biting appear to be factors influencing the bacterial community associated with A. albopictus and A. aegypti. Bacterial communities associated with mosquitoes were mainly studied from laboratory-reared populations, which may not reflect those of wild populations. Indeed, it was shown that field-caught Anopheles mosquitoes harbor a greater bacterial diversity than laboratory populations (Raniet al.,2009). Studies on other insects such as the ground beetle Poecilus chalcites have also shown a higher bacterial diversity in wild populations in comparison with those from laboratories (Lehmanet al.,2009). In addition, it was demonstrated that either nutrition regime or breeding technique could affect the composition of insects' commensal microbial community (Raniet al.,2009; Zouacheet al.,2009a). Conversely, the bacterial populations can influence the behavior and the biology of insect hosts as well (Tsuchidaet al.,2004; Moran & Degnan, 2006). Generally, such extended phenotypes issuing from these reciprocal interactions are evidenced in symbioses between insects and their vertically transmitted endosymbiotic bacteria (Buchner, 1965; Moranet al.,2008). Actually, only a few bacterial symbionts horizontally acquired from the environment have been shown to significantly impact the insects' fitness. This is the case of the heteropteran stinkbug Riptortus clavatus which acquires the beneficial gut bacterial symbiont Burkolderia from the environment in each generation (Kikuchiet al.,2007). Other examples consist of gut microbiota that may contribute to nutrition and detoxification of some insects such as termites and the beetle Tenebrio molitor (Gentaet al.,2006; Warneckeet al.,2007), or provide protection against pathogens in Lepidoptera or desert locust (Dillon & Charnley, 2002; Raymondet al.,2008, 2009), albeit the environmental origin of these microbiota was not clearly established. Altogether, these studies highlighted the importance of taking into account environmental factors such as ecological niches when analyzing symbiotic microbiota associated with wild animal populations. Whether the bacterial communities found here may contribute to adaptive behavior and successful invasion of A. albopictus is under investigation.

At the genus level, several bacteria detected in this study are commonly described in soil and some have been found in hematophagous species of Culicidae, including A. triseriatus (Demaioet al.,1996), Culicoides sonorensis (Campbellet al.,2004), Culex quinquefasciatus (Pidiyaret al.,2004), Anopheles darlingi (Tereniuset al.,2008), Anopheles gambiae (Donget al.,2009), A. albopictus (Zouacheet al.,2009b) and A. aegypti (Gusmaoet al.,2007, 2010; Crottiet al.,2009). Intriguingly, three genera, Acinetobacter, Asaia and Pseudomonas, that are known to contain cultivable species were constantly found in the two species studied here. This suggests either a continuous acquisition through the environment or a vertical inheritance through generations. Interestingly, the genus Asaia was previously found in laboratory-reared Anopheles stephensi and A. aegypti, as well as in wild A. gambiae where it was demonstrated to be transmitted vertically (Faviaet al.,2007; Crottiet al.,2009; Damianiet al.,2010). Our results are the first description of Asaia sp. in natural populations of both A. albopictus and A. aegypti. The ability of Asaia to be inherited both paternally and maternally is attracting attention as a potential candidate for blocking transmission of mosquito-borne pathogens through paratransgenesis (Faviaet al.,2008). Functions have been suggested for some of the other bacterial genera isolated here. The genus Bacillus may probably be involved in cellulose and hemicellulose degradation in termites (reviewed in Konig, 2006). Members of the Enterobacteriaceae family are thought to provide an additional nitrogen source to the fruit fly Ceratitis capitata (Beharet al.,2005). A recent study has shown that an Acinetobacter sp. strain is able to inhibit a tobacco mosaic virus by producing an antiviral compound (Leeet al.,2009). Many other groups of bacteria detected for the first time in mosquitoes perform unknown functions. A better knowledge of the mosquito-associated bacteria will allow investigating their role in the host biology.

Usually, natural populations of A. albopictus have been found singly or doubly infected with Wolbachia (Kittayaponget al.,2000, 2002; Tortosaet al.,2010). When associated with A. albopictus, Wolbachia manipulates the reproduction of its host, inducing a density-dependent cytoplasmic incompatibility phenomenon, which increases the proportion of infected individuals in the population (Sinkinset al.,1995; Dobsonet al.,2001). Interestingly, Wolbachia was recently demonstrated to inhibit mosquito-borne pathogens in some circumstances (Moreiraet al.,2009; Bianet al.,2010; Glaser & Meola, 2010). Here, the survey of Wolbachia in A. albopictus wild populations revealed a high rate of double infection by Wolbachia wAlbA and wAlbB strains in both sexes. The densities of the two Wolbachia strains varied depending on the sex and the sampling region. These results are in accordance with previous data on high variability in Wolbachia densities in field populations (Ahantariget al.,2008; Uncklesset al.,2009). A few cases of single infection by wAlbB were also detected both in males and in females (Fig. 4). Loss of wAlbA strain in A. albopictus males' aging in the laboratory was recently reported in previously doubly infected populations from the Reunion island (Tortosaet al.,2010). Surprisingly, a different pattern was found in field populations of A. albopictus from Thailand, where single infection consists of either Wolbachia wAlbA or wAlbB strains (Kittayaponget al.,2000; Ahantariget al.,2008), suggesting that different factors may account for the prevalence of Wolbachia in this mosquito species, which in turn could potentially interfere with the extended population phenotype.

In conclusion, the results presented here highlight the link between the habitats and the bacterial diversity of wild mosquitoes. As pathogens transmitted by mosquitoes coexist with associated bacteria that can affect insect population dynamics and vectorial competence, characterizing the bacterial composition and diversity of A. albopictus and A. aegypti in their environment is a step forward in understanding the ecology and the multipartite interactions occurring in these two major vectors of arbovirus.

Supporting Information

Fig. S1. Relative density of WolbachiawAlbA in Aedes albopictus males from different collection sites in Madagascar.

Fig. S2. Relative density of WolbachiawAlbB in Aedes albopictus males from different sites in Madagascar.

Please note: Wiley-Blackwell is not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.

Acknowledgements

This paper is dedicated to the memory of Dr Jesus Caballero-Mellado (Centro de Ciencia Genómica, Cuernavaca, Morelos, Mexico) who left us in October 2010. We are grateful to Madagascar National Parks (formerly ANGAP) for authorizing collection of wild mosquitoes and to Biofidal-DTAMB Laboratory of IFR41 in University Lyon 1 for technical assistance. K.Z. was supported by PhD fellowships from the French Ministère de l'Education Nationale, de la Recherche et des Nouvelles Technologies. F.N.R. was supported by the Fondation pour la Recherche sur la Biodiversité (FRB, formerly IFB). This work was funded by grants ANR-06-SEST07 and FRB-CD-AOOI-07-012, and was carried out within the frameworks of GDRI ‘Biodiversité et Développement Durable à Madagascar’ and COST action F0701 ‘Arthropod Symbioses: from fundamental to pest disease management’.

References

View Abstract