Development of a PCR‐RFLP assay to identify Drosophila melanogaster among field‐collected larvae

Abstract The fruit fly Drosophila melanogaster is a model organism to study several aspects of metazoan biology. Most of the work has been conducted in adult fruit flies, including laboratory and field‐derived specimens, but Drosophila melanogaster larvae recently became a valuable model to better understand animal physiology, development, or host–microbe interactions. While adult flies can be easily assigned to a given Drosophila species based on morphological characteristics, such visual identification is more intricate at the larval stage. This could explain the limited number of studies focusing on larvae, especially field‐derived samples. Here, we developed a polymerase chain reaction‐restriction fragment length polymorphism (PCR‐RFLP) assay that discriminates D. melanogaster from other ecologically relevant Drosophila species at the larval stage. The method, which targets the cytochrome oxidase I (COI) gene, was validated using laboratory‐derived larvae from seven D. melanogaster populations originating from different geographic areas as well as six Drosophila species. We further validated this PCR‐RFLP assay in a natural context, by identifying wild larvae collected in two locations in France. Notably, among all PCR‐RFLP profiles that matched the D. melanogaster species, 100% were correctly identified, as confirmed by COI sequencing. In summary, our work provides a rapid, simple, and accurate molecular tool to identify D. melanogaster from field‐collected larvae.

studies focussed on wild Drosophila larvae. In studies that did have this focus, researchers investigated larvae originating from wildcaught adults or relied on adult emergence from field-collected larvae to provide taxonomic identification (Durisko, Kemp, Mubasher, & Dukas, 2014;Godoy-Herrera & Connolly, 2007;Pino et al., 2015).
Wild fruit fly larvae can be distinguished at the family level-such as Drosophilidae and Tephritidae larvae-based on size or spiracles arrangement, but such visual distinction is more intricate between drosophilid species (Figure 1; Van Timmeren, Diepenbrock, Bertone, Burrack, & Isaacs, 2017). Indeed, such morphological details are not available for all the species, remain difficult to see unless performing a time-consuming observation of all the larvae, and could vary according to environmental conditions. In this context, we developed a molecular tool that allows a rapid and accurate identification of D. melanogaster species at the larval stage.

to identify
Drosophila species at the larval stage. PCR-RFLP is a rapid, affordable, and accurate tool, which was successfully applied for insect species identification, notably using mitochondrial DNA markers as a target (Kim, Tripodi, Johnson, & Szalanski, 2014;Salazar et al., 2002;Taylor, Szalanski, & Peterson, 1996). We selected the mitochondrial gene cytochrome C oxidase I (COI), as a large number of nucleotide sequences were available in public and laboratory repositories. For PCR-RFLP development, COI DNA sequences from several Drosophila species were digested in silico using a panel of commercially available restriction enzymes. We paid specific attention to discriminate Drosophila species that are frequently collected along with D. melanogaster in French stations monitored during fieldwork (Fleury, Gibert, Ris, & Allemand, 2009). In the end, we selected the enzyme MboII, as it provided a profile specific of D. melanogaster among a panel of ecologically related Drosophila species including D. immigrans, D. suzukii, D. subobscura, D. simulans, D. busckii, and D. hydei. The method was then applied in vivo on DNA isolated from laboratory-derived and field-collected larvae. We validated our method by COI sequencing and confirmed the accuracy of species assignment using our PCR-RFLP assay.

| Drosophila rearing and maintenance
Laboratory populations of different species from the Drosophila genus (Diptera: Drosophilidae) were maintained at 21°C with a 12 hr light-dark cycle on a diet composed of 73.3 g of cornmeal F I G U R E 1 Larvae from various Drosophila species. Third-instar larvae were collected as they climbed up to the tube prior to pupation. Mouth hooks at the anterior part and posterior spiracles are visible. Individuals were observed under a binocular at 20 × magnification (Moulin-Giraud, France), 76.6 g of dry inactivated yeast (Lynside), 8.8 g of agar (VWR chemicals), 55.5 ml of 96% ethanol, and distilled water up to 1 L. One exception was Drosophila suzukii flies, which were allowed to lay eggs for 2 days on Nutri-Fly ™ medium (Genesee Scientific) prior transfer of the eggs on standard diet for maintenance.
The day prior to depositing the traps, fresh organic fruits (La Vie Claire, organic market in Lyon) were bought to limit insecticide levels. The fruit baits were lacerated using a sterile scalpel to promote rotting and then placed in a perforated plastic container. Two-thirds of banana and ~10 cherries were placed in each container on a thin layer of sawdust to keep humidity. Traps were closed and kept in a sealed box overnight to prevent any contamination by Drosophila from the environment. The next day, traps were humidified with tap water and suspended to the low branches of fruit trees using an iron wire, away from direct sunlight to avoid drying. Traps remained on the field for 3 days. On the third day, traps were collected and insects present inside were removed. The traps were closed with a lid and stored in the laboratory at ~25°C. Late third-instar larvae were collected when they escaped the fruit bait prior to pupariation. All the larvae were collected using clean forceps disinfected in 70% ethanol. Larvae were rinsed for 2 min in 2.6% bleach followed by 2 min in 70% ethanol and 2 min in sterile, 1 × PBS to limit external DNA contamination. Larvae were observed under a stereomicroscope, and only Drosophila-like larvae were selected for species identification. The selection criteria were as follows: (a) the absence of thoracic legs distinctive of Diptera larvae and (b) the absence of body pigmentation combined with the presence of visible branched anterior spiracles and posterior spiracles with dark orange ring at their tip, which are indicative of drosophilid, late third-instar larvae.
Larval guts were dissected and kept at −80°C for further analysis.
Total genomic DNA was isolated from corresponding individual carcasses (i.e., what remains after gut removal) prior to PCR-RFLP.

| Genomic DNA isolation
Total genomic DNA was isolated from whole individual larvae (for laboratory larvae) or individual carcasses (for wild larvae) using the 96-well plate Animal DNA Mini-Preps Kit (Biobasic, NBS Biologicals) according to the manufacturer's instructions with some modifications described below. Briefly, frozen samples were ground dry using a TissueLyzer (Qiagen) with one sterile 5-mm stainless steel bead per tube for 20 s at 20 Hz. A second grinding step was performed after adding 300 μl of ACL lysis buffer and 20 μl of 20 mg/ml proteinase K per sample. Samples were then incubated overnight at 56°C. After purification according to the manufacturer's recommendations, DNA was eluted in 100 μl of DNase-free water (Gibco). An empty tube and a tube with a grinding bead alone were included as controls to monitor DNA cross-contamination between samples. DNA was stored at −20°C until use.

| In silico design of the PCR-RFLP assay
In silico design of PCR-RFLP was performed using CLC Bio main workbench software (version 7.9.1). Cytochrome C oxidase subunit I (COI) nucleotide sequences were downloaded from the National Center for Biotechnology Information (NCBI) and aligned using multiple sequence alignment (MUSCLE) tool with default parameters (Edgar, 2004). We included COI sequences from ecologically relevant, Drosophila-like species. We hypothesized that "contaminant" larvae (i.e., hardly distinguishable from D. melanogaster in the field) would likely belong to the Drosophilidae family. Therefore, we included COI sequences from Drosophilidae detected in the Rhône-Alpes region according to previous field surveys, although most of these species were sporadically observed (Withers & Allemand, 2012). We also in-  information Table S1). Sequences under 650 base pairs (bp) were filtered out, and only unique sequences (i.e., no duplicate) were retained for each species. These unique sequences were verified manually using the Basic Local Alignment Tool (BLAST) to ensure that they encode for COI. In silico digestion of the clean COI sequences was performed using the 1,562 commercially available restriction enzymes present in CLC Bio, which correspond to 340 distinct cutting sites. This method allowed us to attribute a PCR-RFLP profile number to each species.

| PCR-RFLP on individual larvae
DNA samples were diluted 1:50 in DNase-free water prior to PCR amplification, to limit potential PCR inhibition by molecules from The 20 μl-digestion mixture was composed of 1 μl of MboII (5 U/ μl), 2 μl of 10× reaction buffer, 7 μl of PCR product, and 10 μl of DNase-free water. A control sample without the restriction enzyme (replaced by water) was used as a nondigested control. The samples were incubated at 37°C for 1 hr prior to electrophoresis on a 2% agarose gel followed by UV exposure (Gel documentation Bio-Print, Vilber). The size of the bands was estimated using a 50bp ladder (Fermentas).

| In silico identification of D. melanogaster species using COI-MboII PCR-RFLP assay
The selected COI sequences were digested in silico using the set of restriction enzymes available in CLC software (version 7.9.1). We retained commercially available restriction enzymes that display, after a single digestion step, a D. melanogaster profile that would be distinct from Drosophilidae of the region (Withers & Allemand, 2012) and especially from the co-occurring Drosophila species present in the field stations. Among the potential candidate enzymes, AciI, AceIII, and MboII discriminated D. melanogaster from the Drosophila species detected in the two field stations. We selected MboII as it also provided a specific PCR-RFLP profile for the nontarget  Table S1). To draw a workable graphical representation of this analysis, we chose to focus on the ecologically relevant species cited above (Withers & Allemand, 2012). PCR-RFLP profiles from the 15 other members of the Drosophilidae family present in Rhône-Alpes region were different from D. melanogaster (Figure 2). In particular, D. melanogaster PCR-RFLP profile was clearly distinct from the six Drosophila species of ecological interest, demonstrating the accuracy of our method in this context. As justified above, the use of MboII digestion enzyme enabled us to specifically identify these six species, which presented a unique profile (Figure 2).

| Species discrimination of Drosophila larvae in vivo using PCR-RFLP
Genetic divergence among populations from a given species can introduce variation in the restriction site and thus impact the accuracy of the PCR-RFLP method. In our effort to discriminate D. melanogaster from other species at the larval stage, we first investigated whether different populations of D. melanogaster (originating from various locations worldwide) exhibited the same PCR-RFLP profile.
PCR amplification of COI resulted in a single fragment of 709 bp for all the populations tested (Figure 3a). For all samples, MboII digestion produced two fragments of ~500 and ~300 bp. The PCR-RFLP profiles being identical across all the populations tested, the profile of D. melanogaster is thus particularly robust for identification ( Figure 3b).
Then, we tested the accuracy of the in silico predictions by performing the PCR-RFLP assay on larvae from six ecologically relevant Drosophila species. PCR amplification of COI showed a single band at the expected size of 709 bp in all the samples ( Figure 4a). All the MboII-digested profiles displayed between 2 and 3 bands according to the species, ranging from ~500 to ~50 bp, the band below 100 bp being hardly visible (Figure 4b).
In accordance with the in silico analysis, we observed a specific PCR-RFLP profile for D. melanogaster, distinguishable from the six other Drosophila species. Consequently, this result indicates that our method is relevant to identify D. melanogaster at the larval stage in this ecological context. In addition, six species-specific profiles were obtained, suggesting that this tool could also discriminate between the selected Drosophila species at the larval stage ( Figure 4b).

| PCR-RFLP implementation for species identification of wild larvae
We applied our PCR-RFLP method to identify species from wild larvae collected in two different sites (Igé and Reyrieux) along the Rhône Valley (France) (Figure 5a-b). Three to 15 larvae per trap were dissected and stored at −80°C prior to PCR-RFLP analysis.  We finally validated our PCR-RFLP technique using wild larvae collected on the field in two sites along the Rhône Valley, France.
Around 30% of the traps were positive for larvae, suggesting that our collection method could be optimized by dissecting the fruit, or by increasing bait attractiveness using, for instance, different fruit baits or adding yeast extract. Three Drosophila species (D. limbata, D. littoralis, and D. phalerata) harbored the same in silico PCR-RFLP profile than D. melanogaster. Among these species, only D. phalerata was known to develop on rotten fruits, and only this species was observed in the field station explored in this study (Withers & Allemand, 2012). Only 3.4% (7/207) of the collected larvae were misidentified by our PCR-RFLP method, and the misrecognition concerned the sole D. suzukii and D. busckii species. Although high, accuracy of our assay could thus be improved on the particular "contaminant" genus/species (here Phortica), for instance, by performing a second digestion on this subset of samples. Further in silico  Figure S3). Consequently, this enzyme could be used on COI amplicons in a second digestion step, if the presence of flies from by these four species is suspected in the area of study.
In summary, the PCR-RFLP is very robust and reliable for D. melanogaster identification, as all the 175 wild-caught larvae assigned by PCR-RFLP were confirmed following COI sequencing. This method allows a gain in time and money by eliminating the need of COI sequencing. This PCR-RFLP tool was initially designed to specifically identify D. melanogaster in a given ecological context, and it appeared to be also robust for the identification of D. immigrans and D. simulans larvae, even if the sample size should be increased to confirm this result. More generally, the pipeline of the method can be easily adapted to identify Drosophila in other ecological contexts or to target other species in view of the availability of target gene sequences.

ACK N OWLED G M ENTS
We are grateful to Claudine Weistroffer and Didier Penin for giving us access to their orchards for fly collection. We also thank Cristina Vieira-Heddi for providing D. melanogaster populations from various geographic origins, as well as two anonymous reviewers for their comments on the manuscript. This work was supported by the LABEX ECOFECT (ANR-11-LABX-0048) of Université de Lyon (ComEndovir project) within the program "Investissements d'Avenir" (ANR-11-IDEX-0007) operated by the French National Research Agency (ANR).

CO N FLI C T O F I NTE R E S T
None declared.

AUTH O R CO NTR I B UTI O N S
VR, FL, and NK designed the work. VR, PG, and NK collected field samples. HH designed the PCR-RFLP assay. VR, NK, HH, and MV performed the experiments. VR analyzed the results and drafted the manuscript. NK and FL provided critical reading and improved the article.