16S rRNA sequencing
To assess the broader composition of gut microbiota associated with B. dorsalis populations in five mango growing regions in Kenya, high throughput sequencing of the bacterial 16S rRNA gene was carried out for adult samples collected in 2016 and third stage larval samples collected in 2018, retrieved from Kent variety mangoes, with inclusion of a laboratory reared (icipe) colony for comparison. Infested mangoes were collected from farms in Embu (S 0° 28' 56.6" E 37° 34' 55.5"), Muranga (S 0° 42' 50.0" E 37° 07' 03.4"), Nguruman (S 1° 48’ 32” E 36° 03’ 35”), Makueni (S 2° 21’ 18.9576” E 38° 11' 26.376") and Kitui (S 01° 21' E 38° 00').
At each sampling, infested mangoes were washed in distilled water, dissected and placed on sterile sand in ventilated cages at 27 ± 2 °C and 70% humidity to allow third stage larvae to burrow and pupate in sand. Puparia were retrieved from sand through sieving and maintained in sterile petri dishes in ventilated Perspex cages until eclosion. A proportion of third instar larvae were directly retrieved from the fruit for gut dissection. Emerging one day old adult flies from respective sites were collected for gut dissections.
Guts were dissected in sterile phosphate buffered saline (PBS) (140 mM NaCl, 2.7 mM KCl, 10 mM Na2HPO4·7H2O, and 1.8 mM KH2PO4 [pH 7.4]) after surface sterilization of the specimens. The selected larvae and adult flies were surface sterilized as described previously [63]. Dissected guts were homogenized using pestles in 1 ml microfuge tubes containing 300 µl PBS.
A total of five adult specimens per site sampled in 2016, and four larval specimens per site sampled in 2018 were randomly selected for DNA extraction. In addition, five adult flies and four larval specimens were included from the International Centre of Insect Physiology and Ecology (icipe) fruit fly laboratory colony. The sampled colony was derived from infested mango collected from different farms across Kenya and maintained for more than 40 generations with frequent wild infusions in the laboratory at 27 °C and 60% relative humidity. Adult flies were fed on a diet consisting of 3 parts sugar and 1-part enzymatic yeast hydrolysate ultrapure (USB Corporation, Cleveland, Ohio, USA), and water on pumice granules. For each generation, fresh mango domes were used as oviposition receptacles, from which embryos were washed in distilled water before inoculation on larval rearing diets [64]. Same age and generation of laboratory reared flies were used for this study.
DNA extraction and high throughput sequencing were carried out as previously described [63]. Larval DNA was sequenced at the Centre for Integrated Genomics, University of Lausanne, Switzerland and adult DNA at the Macrogen Europe Laboratory, the Netherlands. Adult and larval sequence sets were therefore analyzed separately.
16S rRNA gene sequence analysis
Sequence reads were quality checked and pre-processed in QIIME2 [65] as described previously [63]. A total of 638,815 sequence reads from adult specimens and 56,425 from larval specimens that were retained after removal of spurious reads and all reads shorter than 240 and 272 nucleotides in length respectively, were subjected to further analysis. These sequences clustered into 235 OTUs (adult) and 402 OTUs (larval). Of these, 50 OTUs (adult) and 94 OTUs (larval) survived low count and interquartile range-based variance filtering to eliminate OTUs that could arise from sequencing errors and contamination. Taxonomic assignment, OTU variance filtering and beta diversity measures were carried out as previously described [63]. Differential abundance of bacterial genera was evaluated using the differential gene expression analysis based on the negative binomial distribution (DESeq 2) tool [66].
Bacterial isolation
Cultivable bacteria were isolated from gut homogenates of both larvae and adult flies collected during 2018 from the aforementioned sites in Kenya. Larvae and adults were retrieved from infested mangoes and their guts dissected and homogenized as described above. An aliquot of 5 µl of the fourth serial dilution of each homogenate was inoculated under aerobic conditions on brain heart infusion (BHI) solid media using the spread plate technique [67] and incubated at 37 °C for 14 hours. Representative colony forming units (cfu) were selected based on morphology and clonally propagated up to four times to ensure purity on BHI agar plates.
Bacterial isolates identification
Pure isolates were sub-cultured in BHI broth and incubated at 37 °C for 16 hours on a shaking platform at 300 revolutions per minute (rpm). Bacterial cells were harvested from media then washed thrice in PBS by centrifugation at 10000 rpm for 10 min at 10 °C, each time discarding the supernatant.
DNA extraction from bacterial cells and PCR amplification were carried out as described previously [63] with slight variations in the primers used i.e. the 28F (5’-GAGTTTGATCNTGGCTCAG-3’) and 519R (5’-GTNTTACNGCGGCKGCTG-3’) primer pair, as well in the cycling conditions, where, following the initial denaturation, 35cycles of 30 s at 95 °C, 40 s at 54 °C and 1 min at 72 °C were run, followed by the final elongation step. Direct Sanger sequencing in both forward and reverse directions was done for all amplified samples. Sequence alignments were performed using Clustal W in Geneious 8.1.9 software [68]. Homology searches using BLAST against the 16S ribosomal RNA sequence database at the National Center for Biotechnology Information (NCBI) were done to infer identity and similarity of isolates to subject sequences in the database.
Generation of axenic lines
Bactrocera dorsalis embryos were collected from gravid females from the icipe B. dorsalis laboratory reared colony using perforated mango domes. Embryos were surface sterilized in 70% ethanol for 5 minutes, then dechorionated in a 7 % v/v sodium hypochlorite solution for 3 minutes in a fine mesh (Nitex Nylon100 µm) basket. Dechorionated embryos were rinsed three times in distilled water for 5 minutes each then flooded with absolute ethanol. Sterile materials were used in subsequent procedures in a sterile laminar flow hood. Using a fine camel hair brush, embryos from the bottom of the basket were transferred and spread out on 2 cm X 2 cm X 4 mm sponge cloth immersed in larval rearing diet [64] in flat base 30 mm X 100 mm cylindrical test tubes. Approximately 100 embryos were placed in each tube. Axenic control lines were derived at this step by plugging cotton wool up to 3cm from the top of the tube.
Generation of mono-association lines
An inoculum 50 µl of 1 X 104 cfu/ml of each isolate was introduced in triplicate per experiment directly onto the embryos before plugging the tubes with cotton wool.
Rearing and quality check of fly lines
All tubes were maintained at 27 °C and 70% humidity. A control group with an intact microbiome (whose embryos were not dechorionated) was included in triplicate in each experiment. The immature stages of all B. dorsalis lines were reared in axenic conditions.
To quality check axenic lines, random third stage larvae were retrieved from axenic control tubes per experiment and homogenized in 50 µl PBS. Five µl of this homogenate was plated on nutrient agar plates and incubated at 37 °C and checked for bacterial growth after 15 hours. A volume of 200 µl of sterile larval rearing diet was added to each tube every 24 hours after hatching.
No bacterial growth from third instar larvae was recorded on enriched media plates during quality check of axenic lines, inferring a strong elimination effect using this approach.
Mature third-stage larvae crawled upward and burrowed into the cotton wool plug to pupate. The time taken to the observation of at least 10 first instar larvae as well as to the observation of puparia on the cotton wool plug were recorded for each tube. Measurements of 1-day old puparia weight, dorsal to ventral length as well as width of the sixth segment were also recorded. Weights were recorded in triplicates of 20 puparia each from every fly line, whereas length and width measurements were recorded from 20 puparia per fly line. Length measurements were carried out under a Leica LAS EZ4D stereomicroscope (Leica Ltd., Switzerland). Cotton wool plugs with puparia were submerged in autoclaved distilled water at room temperature and carefully pulled apart to free puparia. The retrieved puparia were dried on sterile paper towel and maintained on sterile Petri dishes placed in sterilized ventilated Perspex cages until eclosion at 27 °C and 70% humidity. Eclosed adult flies were maintained under normal conditions. All data for development and puparia measurements were tested for normality using the Shapiro-Wilk’s test. Non-normal distributions were recorded in the embryo duration (W = 0.84, p < 0.001), larval duration (W = 0.92, p < 0.001), puparia length (W = 0.93, p < 0.001), puparia width (W = 0.96, p < 0.001) and puparia weight (W = 0.87, p < 0.001) datasets. All datasets conformed to homogeneity of variance as determined using the Levene’s test: embryo duration (F(6, 56) = 0.50, p = 0.81), larval duration (F(6, 56) = 1.24, p = 0.30), puparia length (F(6, 133) = 0.85, p = 0.54), puparia width (F(6, 133) = 1.31, p = 0.26) and puparia weight (F(6, 56) = 1.34, p = 0.25). Statistical significance in the datasets was therefore evaluated using the Kruskal-Wallis test followed by Dunn’s multiple comparisons post hoc test. All analyses were conducted in the R statistical software [69]. In addition, adult flies emerging from all mono associated lines were monitored for fitness for a period of 12 days post eclosion. No mortality was recorded in any of the B. dorsalis lines during this period.
Exposure to Metarhizium anisopliae
Dry conidia of M. anisopliae ICIPE69 were obtained from icipe’s Arthropod Pathology Unit Germplasm. Triplicates in groups of 20 newly emerged adult flies from each fly line were each exposed to 0.3 g of dry spores of the M. anisopliae ICIPE69 for 1 min in a contamination device made from a 50 ml falcon tube lined with velvet. Exposed flies were released in 10 cm x 10 cm x 10 cm ventilated cages and maintained on adult B. dorsalis rearing diet [70] and sterile water saturated on cotton wool at 27 °C and 70% humidity. A control set derived from unexposed flies was included in each treatment. All flies in this set (unexposed to fungus) in all B. dorsalis lines survived the duration of the experiment. The survival rates of B. dorsalis from the different fly lines were monitored daily after exposure to the M. anisopliae isolate. Pairwise comparisons of survival between all the fly lines and controls was evaluated using the Gehan-Breslow-Wilcoxon test. Survival curves for adult flies exposed to ICIPE69 were generated using the Kaplan-Meier method in the Graph Pad Prism software, version 7.00 for Windows [71].
Fly line microbiota
Gut tissues from a pool of five 2-day post eclosion adult flies per line from the L. lactis and the P. alcalifaciens inoculated B. dorsalis lines were processed as described above for high throughput sequencing, targeting the v3-v4 region of the bacterial 16S rRNA gene, and similarly analyzed in QIIME2-2018.11.