Multiple mutations in the nicotinic acetylcholine receptor Ccα6 gene associated with resistance to spinosad in medfly

Spinosad is an insecticide widely used for the control of insect pest species, including Mediterranean fruit fly, Ceratitis capitata. Its target site is the α6 subunit of the nicotinic acetylcholine receptors, and different mutations in this subunit confer resistance to spinosad in diverse insect species. The insect α6 gene contains 12 exons, with mutually exclusive versions of exons 3 (3a, 3b) and 8 (8a, 8b, 8c). We report here the selection of a medfly strain highly resistant to spinosad, JW-100 s, and we identify three recessive Ccα6 mutant alleles in the JW-100 s population: (i) Ccα63aQ68* containing a point mutation that generates a premature stop codon on exon 3a (3aQ68*); (ii) Ccα63aAG>AT containing a point mutation in the 5′ splicing site of exon 3a (3aAG > AT); and (iii) Ccα63aQ68*-K352* that contains the mutation 3aQ68* and another point mutation on exon 10 (K352*). Though our analysis of the susceptibility to spinosad in field populations indicates that resistance has not yet evolved, a better understanding of the mechanism of action of spinosad is essential to implement sustainable management practices to avoid the development of resistance in field populations.

Spinosad is an insecticide widely used for the control of insect pest species, including Mediterranean fruit fly, Ceratitis capitata. Its target site is the α6 subunit of the nicotinic acetylcholine receptors, and different mutations in this subunit confer resistance to spinosad in diverse insect species. The insect α6 gene contains 12 exons, with mutually exclusive versions of exons 3 (3a, 3b) and 8 (8a, 8b, 8c). We report here the selection of a medfly strain highly resistant to spinosad, JW-100 s, and we identify three recessive Ccα6 mutant alleles in the JW-100 s population: (i) Ccα6 3aQ68* containing a point mutation that generates a premature stop codon on exon 3a (3aQ68*); (ii) Ccα6 3aAG>AT containing a point mutation in the 5′ splicing site of exon 3a (3aAG > AT); and (iii) Ccα6 3aQ68*-K352* that contains the mutation 3aQ68* and another point mutation on exon 10 (K352*). Though our analysis of the susceptibility to spinosad in field populations indicates that resistance has not yet evolved, a better understanding of the mechanism of action of spinosad is essential to implement sustainable management practices to avoid the development of resistance in field populations.
Spinosad is a natural product mixture of two compounds, spinosyn A and spinosyn D, produced by the bacteria Saccharopolyspora spinosa and used as insecticide since 1997 1 . It affects the insect central nervous system causing involuntary neuronal excitation that produces muscle contraction, tremors, paralysis and death 2 . Spinosad acts in insects as an allosteric agonist of acetylcholine by binding to nicotinic acetylcholine receptors (nAChRs) that function as neurotransmitter ligand-gated ion channels 3 . nAChRs are composed by the homomeric or heteromeric assembly of five ligand-binding nAChR subunits circularly gathered to form the channel. A typical nAChR subunit contains an extracellular N-terminal domain including six loops (A-F) involved in the acetylcholine binding site, four transmembrane segments (TM1-4, of which TM2 is part of the receptor channel pore), and a large intracellular domain (between TM3 and TM4) 4 . Insect species have between 10-16 subunits in their genome, 10 in Drosophila melanogaster (α1-7 and β1-3), and 11 in C. capitata (α1-8 and β1-3) 5 . The specific binding of spinosad to α6 subunits is believed to mediate death in insects 6-8 . nAChR α6 subunits show a high degree of conservation both in amino acid identity and genomic structure among insect species and also with the α7 subunits in vertebrates 9 . The insect α6 gene contains 12 different exons (exon 1-12), with two variants of exon 3 (exon 3a and exon 3b) and generally three variants of exon 8 (exon 8a, 8b and 8c) (Fig. 1A). These mutually exclusive variants of exon 3 and exon 8, together with frequent A-to-I pre-mRNA editing, confers the capacity to generate a huge diversity of mRNA products. Interestingly, exon 3 contains the acetylcholine binding loop D, exon 8 includes part of the TM2 domain involved in the formation of the pore, and some of the editing sites described in α6 are located in the proximity of the acetylcholine binding pocket 10,11 . However, whether this α6 mRNA diversity leads to functionally distinct receptors and/or affect the interaction with spinosad remains elusive.
High levels of spinosad resistance in the field have been recorded in several pest species including the American serpentine leafminer Liriomyza trifolii 12 , the diamondback moth Plutella xylostella 13,14 , the western Laboratory strains. All strains were reared in the laboratory as described in Magaña et al. (2007) 31 .
The laboratory strain C was established from C. capitata collected at non-treated experimental fields of the Instituto Valenciano de Investigaciones Agrarias (València, Spain) in 2001. It is maintained in our laboratory without exposure to insecticides. The malathion-resistant strain W-4 Km and the lambda-cyhalothrin resistant strain W-1Kλ are spinosad-susceptible strains maintained under regular selection in the laboratory, as previously reported 33,35 . The spinosad resistant strain JW-100 s derives from individuals collected in Xàbia in 2007  (Supplementary Information, Table S1), and has been obtained by laboratory selection performed by treating the  adult rearing diet with spinosad: 1 to 10 ppm for 48 h from generations 1 to 29; 100 ppm for 48 h from generation  30 to 49; 100 ppm for 72 h from generation 50. From generation 24, adults were starved for 24 h before selection. At the beginning of the selection process field derived adults laid eggs on apples that were introduced in the rearing cages. At generation 10, reciprocal crosses were performed between the spinosad selected population and the malathion resistant strain W-4 Km with the objective of adapting the spinosad resistant strain to lay eggs on the net of the rearing cages, facilitating the rearing and selection processes. For topical application, adult flies (3-5 days old) were anesthetized with CO 2 and treated with a 0.5 μl drop of insecticide solution in acetone or only acetone (used as control) on the dorsal thorax by using an automatic microapplicator 900× (Burkard Manufacturing Co., Hertfordshire, United Kingdom). Three to four replicates of 10-15 adults per dose (calculated as μg of insecticide per g of fresh weight of insect, assuming an average weight of 10 mg) were performed. After topical treatment, insects were placed in the ventilated plastic dish containing water and rearing diet. The mortality was recorded after 48 h.

Chemicals.
The synergists PBO, DEF and DEM were diluted in acetone and applied topically to adult flies as described above. The doses applied (0.5 μg PBO, 1 μg DEF or 1 μg of DEM per insect) showed no mortality on adults. Acetone was used as a control. After 2 h, the flies were treated topically with spinosad as previously described. The mortality was recorded after 48 h. For all bioassays performed, alive and dead individuals were frozen with liquid nitrogen and stored at −80 °C.

Reciprocal crosses.
Adults of the C and JW-100 s strains were collected and their sex determined immediately after adult emergence. Males and females from each strain were placed separately into ventilated plastic dishes and maintained with water and rearing diet for two days. Reciprocal crosses between virgin individuals (100 C males x 100 JW-100 s females; and 100 JW-100 s males x 100 C females) were performed to obtain the F1 generation. The progeny of both reciprocal crosses was kept in the absence of selection pressure at standard conditions to produce the F2 generation. The susceptibility to spinosad in adults of the F1 and F2 generations was tested by feeding bioassays using concentration-mortality response for F1 and a discriminating concentration of 10 ppm for F2. Data analysis. Mortality data were used to estimate the concentrations/doses needed to cause 50% mortality (LC 50 or LD 50 ) by probit analysis using the computer program POLO-PC (LeOra Software14), which automatically corrects for control mortality using Abbott's transformation 36 . Resistance ratios (RR = LC 50 (field or lab strain)/LC 50 (C strain)) and synergistic ratios (SR = LD 50 (synergist −)/LD 50 (synergist+)) were considered significant if their 95% fiducial limits did not include 1 37 . Mortality data when using discriminating concentrations were subjected to arcsine square root transformation and compared by ANOVA followed by Dunnett´s test. Abbott's formula was used to correct mortality data for natural (non-treated) response 36  was carried out using the RevertAid H Minus First Strand cDNA Synthesis Kit (Thermo Fisher Scientific). The cDNA obtained was diluted with nuclease-free water and stored at −20 °C. Ccα6 codifying region was amplified through PCR in a volume of 25 μl using 0.4 μM of forward and reverse oligonucleotides (FnACh6-ex1, 5′-TCGCTGTTTGCCGTGTTGATCTTT-3′; and RnACh6-ex12, 5′-TTGGACTATTATGTGCGGAGCTGA-3′) (Sigma-Aldrich), 1.4 U of Expand High Fidelity DNA polymerase (Expand High Fidelity PCR System, Roche), 10x PCR Buffer II, 2.5 mM MgCl 2 , 0.8 mM dNTPs (Thermo Scientific), and 4 μl of the corresponding cDNA as template. PCR conditions were as follows: an initial denaturation step at 95 °C for 5 min; 40 cycles of 95 °C for 30 s, 60 °C for 30 s, and 72 °C for 2 min (with an increase of 1 s for every cycle); and a final step of 72 °C for 7 min for fully extension. PCR products were analysed by electrophoresis on 1% agarose gel (Agarose D2, Conda Pronadisa) and the bands of interest were purified from the gel with Ultrafree ® -DA Centrifugal Filter Units (Millipore, Ireland) following manufacturer's instructions. DNA sequencing was performed at Secugen Genomic DNA extraction, PCR and sequencing of Ccα6 exon 3a and exon 3b. Genomic DNA extraction from thorax and abdomen of single individuals was carried out through tissue homogenization and DNA isolation using DNeasy Blood & Tissue kit (Qiagen), following manufacturer's instructions. DNA was quantified and 100-200 ng were used as template for PCR in a volume of 10 μl, along with: 0.4 μM of forward and reverse oligonucleotides, 0.5 U AmpliTaq Gold (Applied Biosystems), 10x PCR buffer II, 2 mM MgCl 2 , and 0.64 mM dNTPs. Oligonucleotides used for exon 3a amplification were FnACh6ex3a-intron (5′-ACGAAGGCGAAATAAGTTTCAAGT-3′) and RnACh6ex3a-intron (5′-GCTGCGTGAAAACCATTGAAATCG-3′). Oligonucleotides used for exon 3b amplification were FnACh6ex3b-intron (5′-CTTTTTATGTTCAATGTCTTCTGC-3′) and RnACh6ex3b-intron (5′-TTTGGTGCCACTTCGTATGCATGA-3′). For both amplifications cycling conditions consisted on an initial denaturation step at 95 °C for 5 min, followed by 40 cycles of 95 °C for 30 s, 57 °C for 30 s, and 72 °C for 25 s, and a final step of 72 °C for 7 min to fully extend all PCR products. Sequencing was performed as previously described, directly from PCR products, using FnACh6ex3a-intron or RnACh6ex3b-intron oligonucleotides. 10. The presence/absence of K352* mutation was detected through a PCR-RFLP method, taking advantage of the generation of an MseI restriction site in the presence of this mutation. Exon 10 was amplified in a volume of 10 μl using 100-200 ng of genomic DNA as template, 0.4 μM of FnACh6-996 (5′-TAAATCCGTTTTCCTGCAATGGCT-3′) and RnACh6-1111 (5′-CCTTTAATTCCAATTCCTTCATGC-3 ′) oligonucleotides, 0.5 U AmpliTaq Gold (Applied Biosystems), 10x PCR buffer II, 2.5 mM MgCl 2 , and 0.64 mM dNTPs. PCR conditions were as follows: an initial denaturation step at 95 °C for 5 min; followed by 40 cycles of 95 °C for 30 s, 60 °C for 30 s, and 72 °C for 25 s; and a final step of 72 °C for 7 min to complete the extension. PCR products were then digested at 37 °C for 2 hours with 2 U of MseI restriction enzyme (New England BioLabs) and

PCR-RFLP for K352* detection on exon
10x CutSmart ® buffer (New England BioLabs). Digestion products were finally analysed by electrophoresis on a 2.5% agarose gel composed of a 1:1 (vol/vol) mix of conventional agarose D2 and low melting temperature agarose (NuSieve TM GTG TM Agarose, Lonza). Two superposed bands of 58 bp in the agarose gel indicate a homozygous genotype for the K352* mutation. A single non-digested band of 116 bp indicates that the individual do not carry the mutation. Finally, a heterozygous individual produce the non-digested band (116 bp) and the two digested bands (58 bp) (see Supplementary  Multiplex PCR methods to detect AG > AT and Q68* mutations on exon 3a. A combination of multiplex PCRs was designed to detect the mutations observed in exon 3a (AG > AT and Q68*) from genomic DNA without the sequencing step. Multiplex PCR-1 to detect both mutations at the same time was performed in a volume of 10 μl. Oligonucleotides used in this multiplex were: 0.4 μM of a forward oligonucleotide specific for the alleles with the splicing site mutation AG > AT and the linked A200T mutation, including also a mismatch on 5′ region to confer more stability to the primer (FnACh6ex3a-splicing, 5′-ATTATTTATATGACGAAAAGATTC-3′) 39 ; 0.4 μM of a reverse oligonucleotide specific for the alleles with the Q68* mutation, including a destabilizing mismatch on the third nucleotide close to the 3′ end to avoid unspecific annealing on wild-type sequences (RnACh6ex3a-Q68Stop, 5′-CCACGCATTTGTGGTCAGAATTTA−3′); and 0.1 μM of FnACh6ex3a-intron and RnACh6ex3a-intron oligonucleotides, common to all the alleles ( Supplementary Information, Fig. S3). Other components required for the PCR were: 100-200 ng of genomic DNA, 0.5 U AmpliTaq Gold, 10x PCR buffer II, 2 mM MgCl 2 , and 0.64 mM dNTPs. Cycling conditions consisted on a denaturation step at 95 °C for 5 min, followed by 40 cycles of 95 °C for 30 s, 50 °C for 30 s, and 72 °C for 20 s, and a final step of 72 °C for 7 min to complete the extension. PCR products were finally analysed by electrophoresis on a 2.5% agarose gel composed of a 1:1 (vol/ vol) mix of conventional agarose D2 and low melting temperature agarose. Multiplex PCR using these four oligonucleotides generated four different patterns of bands ( Supplementary Information, Fig. S2B and S3) that allowed us to discriminate among: 1-homozygous genotypes without any of the two mutations (+/+) (320 bp band); 2homozygous for the splicing mutation (AG > AT/AG > AT) and heterozygous with no mutations in one allele and the splicing mutation in the other (+/AG > AT) (two bands of 320 bp and 209 bp); 3-homozygous for the Q68* mutation (Q68*/Q68*) or heterozygous with no mutations in one allele and Q68* mutation in the other (+/Q68*) (two bands of 320 bp and 158 bp); and 4-heterozygous with the splicing mutation in one allele and Q68* mutation in the other (Q68*/AG > AT) (three bands of 320 bp, 209 bp and 158 bp). Thus, although this multiplex PCR allows the detection of the two mutations observed in exon 3a, it has the handicap that it cannot discriminate between the +/AG > AT and AG > AT/AG > AT genotypes and between the +/Q68* and Q68*/Q68* genotypes. Differentiation between homozygous Q68*/Q68* and heterozygous +/Q68* individuals was carried out through multiplex PCR-2 ( Supplementary Information, Fig. S2C). A new forward oligonucleotide was designed on the Q68 region, specific for the wild-type sequences including an extra destabilizing mismatch on the third nucleotide close to the 3′ end to avoid unspecific annealing (FnACh6ex3a-wtQ68, 5′-ATTTTTTATAGGACGAAAAGAGTC−3′) ( Supplementary Information, Fig. S3). It was used together with the reverse oligonucleotide specific for the alleles with the Q68* (RnACh6ex3a-Q68Stop) and the two intronic oligonucleotides common to all alleles (FnACh6ex3a-intron and RnACh6ex3a-intron), to allow the amplification. PCR components for 10 μl of multiplex PCR-2 were as follows: 100-200 ng of genomic DNA, 0.1 μM of FnACh6ex3a-intron and RnACh6ex3a-intron oligonucleotides, 0.2 μM of FnACh6ex3a-wtQ68, 0.4 μM of RnACh6ex3a-Q68Stop, 0.5 U AmpliTaq Gold, 10x PCR buffer II, 2 mM MgCl 2 , and 0.64 mM dNTPs. Amplification conditions consisted on a denaturation step at 95 °C for 5 min, followed by 40 cycles of 95 °C for 30 s, 50 °C for 30 s, and 72 °C for 20 s, and a final step of 72 °C for 7 min to complete the extension. PCR products were analysed by electrophoresis on a 2.5% agarose gel composed of a 1:1 (vol/vol) mix of conventional agarose D2 and low melting temperature agarose. The different pattern of bands depending on the genotype allowed to differentiate among: 1-homozygous genotypes without the Q68* mutation (+/+) (two bands of 320 bp and 209 bp); 2-heterozygous with no mutations in one allele and Q68* mutation in the other (+/Q68*) (three bands of 320 bp, 209 bp and 158 bp); and 3-homozygous with Q68* mutation in both alleles (two bands of 320 bp and 158 bp) ( Supplementary Information, Fig. S2C).
Finally, multiplex PCR-3 was designed to distinguish between +/AG > AT and AG > AT/AG > AT individuals ( Supplementary Information, Fig. S2D). A reverse oligonucleotide specific for the wild-type alleles was designed on the splicing domain region (RnACh6ex3a-wtAG, 5′-GTCAGAATCTGATTCTTTTCGTCC−3′) ( Supplementary Information, Fig. S3). It was used together with the forward oligonucleotide specific for the alleles with the AG > AT mutation (FnACh6ex3a-splicing) and the two intronic oligonucleotides common to all alleles (FnACh6ex3a-intron and RnACh6ex3a-intron), to allow the amplification. PCR was performed in a volume of 10 μl containing: 100-200 ng of genomic DNA, 0.1 μM of FnACh6ex3a-intron and RnACh6ex3a-intron oligonucleotides, 0.4 μM of FnACh6ex3a-splicing, 0.1 μM of RnACh6ex3a-wtAG, 0.5 U AmpliTaq Gold, 10x PCR buffer II, 2 mM MgCl 2 , and 0.64 mM dNTPs. Cycling conditions consisted on a denaturation step at 95 °C for 5 min, followed by 40 cycles of 95 °C for 30 s, 55 °C for 30 s, and 72 °C for 20 s, and a final step of 72 °C for 7 min to complete the extension. PCR products were analysed by electrophoresis on a 2.5% agarose gel composed of a 1:1 (vol/vol) mix of conventional agarose and low melting temperature agarose. The pattern of bands obtained allowed to differentiate among: 1-homozygous genotypes without the AG > AT mutation (+/+) (two bands of 320 bp and 145 bp); 2-heterozygous with no mutations in one allele and AG > AT mutation in the other (+/ AG > AT) (three bands of 320 bp, 209 bp and 145 bp); and 3-homozygous with AG > AT mutation in both alleles (two bands of 320 bp and 209 bp, but with slight unspecific annealing of RnACh6ex3a-wtAG oligonucleotide that shows a weak band of 145 bp) ( Supplementary Information, Fig. S2D). An unspecific band of approximately 400 bp that does not interfere on the interpretation of the results was also visible in +/AG > AT and AG > AT/ AG > AT genotypes ( Supplementary Information, Fig. S2D).

Results
Spanish field populations show high susceptibility to spinosad. We first assessed the spinosad susceptibility of twelve C. capitata field populations collected in Spain ( Supplementary Information, Fig. S1 and Table S1) by concentration-response bioassays (Table 1) and by discriminating concentration assays (Supplementary Information, Table S2). We found that LC 50 values, which estimate the concentration killing 50% of the individuals, were always far below the concentration of the spinosad-based insecticide recommended for C. capitata treatments in Spain (240 ppm, Spintor Cebo 0.024% p/v, Dow Agrosciences) ( Table 1). In addition, we found that resistance ratios (RR) calculated for field populations in comparison to the susceptible laboratory control strain (C strain) were always below three-fold. Similarly, treatments with the discriminating concentration of 1 ppm of spinosad, which causes 88% mortality in the C strain, also caused mortality ≥77% in all field populations tested (Supplementary Information, Table S2). Thus, our initial screening demonstrated high susceptibility to spinosad in all medfly field populations analysed.
The laboratory-selected strain JW-100 s is resistant to spinosad. In order to obtain a spinosad-resistant strain, we adapted one of the populations with higher LC 50 , Xàbia 2007, to laboratory conditions and selected the strain over successive generations with increasing concentrations of spinosad in the adult rearing diet (Table 2). It was not until generation F25 that we observed a moderate increase in the RR (5.79-fold). After generation F25, we observed that resistance levels increased rapidly, with RR reaching 321-fold at generation F29. We set the selection pressure at 100 ppm of spinosad in the diet from generation F30, but we could not perform feeding bioassays to calculate LC 50 ratios after generation F36 because of the limited solubility of spinosad in water solutions (235 ppm at pH 7 and 290 ppm at pH 5 for spinosyn A, SPINOSAD Technical Bulletin, Dow AgroSciences LLC). Accordingly, we performed topical bioassays using acetone as a solvent from generation F35, after which we were able to achieve an RR of over 1000-fold at generation F45 compared to the susceptibility of the C strain. We observed similar levels of resistance until the last generation evaluated (F81). We named the laboratory-selected spinosad-resistant strain JW-100 s.
The resistance in JW-100 s strain is not reverted by inhibitors of detoxification enzymes and does not confer cross-resistance to other insecticides. To decipher the molecular mechanisms associated with spinosad resistance in JW-100 s, we tested the role of detoxification enzymes using the synergists PBO (inhibitor of cytochrome P450 s), DEF (esterase inhibitor) and DEM (inhibitor of glutathione S-transferases) ( Table 3). Topical treatment with PBO or DEF on JW-100 s produced a slight reduction on LD 50 , with a synergistic ratio (SR) of 1.91 and 1.73, respectively, whilst treatment with DEM did not influence the susceptibility of To further explore the mechanisms associated with resistance, we next tested the cross-resistance of the JW-100 s strain to the insecticides imidacloprid, fipronil, malathion and lambda-cyhalotrin (Table 4). Feeding treatments with discriminating concentrations of malathion and lambda-cyhalotrin revealed a high susceptibility of the JW-100 s strain to these two insecticides. On the contrary, we observed high tolerance to imidacloprid and fipronil after feeding treatments (RR of 6.5-and 17.5-fold when compared to the C strain, respectively). However, we did not observe this tolerance after topical treatments, suggesting that it could be related to a detoxification mechanism in the digestive tract, bypassed under topical application. These results demonstrated that the main mechanism responsible for the resistance to spinosad in JW-100 s did not confer cross-resistance to the insecticides imidacloprid, fipronil, malathion or lambda-cyhalotrin.   Table 2. Selection of resistance to spinosad to obtain the JW-100 s strain. a Selection concentrations (SC) used in the selection process, in ppm of spinosad in the diet. The absence of treatment is indicated as "−". b Assays were performed with Spintor Cebo until F13. Spinosad 88% technical grade insecticide was used since F14.  expression of alternative exons 3a/3b and 8a/8b, but not 8c. We also observed conserved A-to-I RNA editing sites 4, 5, 6 and 7 9 . Interestingly, we identified two homozygous point mutations generating premature stop codons in all of the five individuals of the JW-100 s strain analysed. One point mutation was located among the double peaks corresponding to the simultaneous sequencing of exons 3a/3b ( Supplementary Information, Fig. S4A). We determined the precise location of this mutation at exon 3a (nt: C202T; aa: Q68*) by PCR amplification and direct sequencing of exon 3a and exon 3b from genomic DNA ( Supplementary Information, Fig. S4A). The second point mutation corresponded to the coding sequence of exon 10 (nt: A1054T; aa: K352*) ( Supplementary Information, Fig. S4B). We next analysed the presence of these point mutations in a larger cohort of individuals (n = 48) from generation F85 of the JW-100 s strain, using direct sequencing, PCR-RFLP and multiplex PCR on genomic DNA. Our analysis demonstrated the existence of two alleles: Ccα6 3aQ68-K352* carrying the two premature stop codon mutations at exon 3a and exon 10; and Ccα6 3aQ68* carrying only the stop codon mutation at exon 3a (Supplementary Information,  Table S3). Remarkably, 92% of the non-treated individuals of the JW-100 s strain analysed from F85 were homozygous for Ccα6 3aQ68*-K352* , and the rest were heterozygous for both Ccα6 3aQ68*-K352* and Ccα6 3aQ68* alleles (Table 5). We observed similar results among individuals from generation F85 that survived a treatment of 100 ppm of spinosad in the diet. On the other hand, we did not observe any of these Ccα6 mutant alleles in three different laboratory strains highly susceptible to spinosad (C, W-4 km and W-1 kλ) 31,33 (Table 5). These results suggested that Ccα6 3aQ68*-K352* and Ccα6 3aQ68* alleles were associated with spinosad resistance in JW-100 s laboratory strain.
To further correlate the presence of the described alleles and the resistance to spinosad in JW-100 s strain, we recorded the frequency of the two alleles in non-treated individuals and in individuals surviving spinosad exposure at generations F0, F25, F29 and F48. Interestingly, we identified an additional new allele in generation F25 from sequencing the genomic region of exon 3a. This new allele, Ccα6 3aAG>AT , carries a mutation at the 5′ canonical splicing dinucleotide of exon 3a, changing the AG splice site to AT (AG > AT), together with another point mutation at the coding region of exon 3a (nt: A200T) ( Supplementary Information, Fig. S4C). This allele does not carry the mutations 3aQ68* or K352* ( Supplementary Information, Table S3). Chromatograms obtained from individuals homozygous for the Ccα6 3aQ68* allele demonstrated that this allele produced isoforms containing the exon 3b. On the other hand, chromatograms from individuals homozygous for the Ccα6 3aAG>AT allele demonstrated that this allele produced full-length 3b transcripts, but also produced incomplete 3a transcripts by skipping exon 3a and joining exons 2 and 4 without altering the coding frame of the gene.
We next used PCR-RFLP and multiplex PCR methods to detect the presence of the three mutations and thus identify the specific genotype of each individual ( Supplementary Information, Table S3). We called alleles that did not bear any of the described mutations '+' alleles. The frequencies of genotypes and alleles in each generation are shown in Table 5 and Supplementary Information, Table S4, respectively. We did not find any of the described mutations among the F0 individuals that generated the JW-100 s strain, including both non-treated individuals and the survivors at different spinosad concentrations. However, in generation F25, when we observed a small increase on the level of resistance (Table 2), we detected Ccα6 3aAG>AT and Ccα6 3aQ68* alleles with a frequency of 0.3 and 0.38, respectively, in non-treated individuals ( Supplementary Information, Table S4). Surprisingly, we did not detect the allele Ccα6 3aQ68*-K352* in this generation. Among the genotyped individuals that survived to a concentration of 6 ppm of spinosad in the same generation (F25), a low frequency (0.1) had a wild type allele (+/Ccα6 3aAG>AT ), while all the rest had both mutated alleles (Ccα6 3aAG>AT /Ccα6 3aAG>AT , Ccα6 3aQ68* /Ccα6 3aQ68* or Ccα6 3aAG>AT /Ccα6 3aQ68* ) ( Table 5). Four generations later (F29), when the resistance of JW-100 s increased to 321-fold, the frequency of Ccα6 3aAG>AT and Ccα6 3aQ68* alleles among the non-treated individuals increased (0.45 and 0.55, respectively), while + alleles were not detected ( Supplementary Information, Table S4). Among the survivors at a high spinosad dose (240 ppm), the frequency of Ccα6 3aQ68* allele (0.8) was higher than the Ccα6 3aAG>AT allele (0.2), and the Ccα6 3aAG>AT /Ccα6 3aAG>AT genotype was observed only in one individual among 30 analysed (Table 5 and Supplementary Information, Table S4). Finally, in generation F48 we found the Ccα6 3aQ68*-K352* allele at a high frequency (0.98), whilst the frequency of Ccα6 3aQ68* allele diminished (0.02) and the Ccα6 3aAG>AT allele was not detected (Table 5 and Supplementary Information, Table S4). Altogether, these results further demonstrated that alleles Ccα6 3aQ68*-K352* , Ccα6 3aQ68* and Ccα6 3aAG>AT were associated with spinosad resistance. It is worth  Table 3. Effect of synergists on the resistance of JW-100 s to spinosad by topical application. † Synergists: 0.5 μg PBO, 1 μg DEF and 1 μg DEM diluted in 0.5 μl acetone and topically applied (acetone was used as control without synergist). After 2 h the flies were treated with spinosad. ‡ Number of flies considered in the Probit analysis (including non-treated). § Lethal dose (LD 50 ) in µg/g of insect (fresh weight assuming an average weight of 10 mg) by topical application at 48 h (a 0.5 μL drop of insecticide solution in acetone or acetone alone was applied to the dorsal thorax of each fly by using an automatic microapplicator). Spinosad 88%, technical grade insecticide. ¶ Synergistic ratio (SR) = LD 50 (synergist −)/LD 50 (synergist+). The fiducial limits for SR were calculated according to Robertson and Preisler (1992). ♯ SR is significant (P < 0.05) if the 95% FL does not include 1. *Good fit of the data to the probit model (P > 0.05).
Spinosad resistance in JW-100 s is completely recessive and linked to Ccα6 3aQ68*-K352* and Ccα6 3aQ68* alleles. Finally, we studied the inheritance of spinosad resistance by performing reciprocal crosses of JW-100 s individuals from generation F83 with individuals of the susceptible C strain (Table 6). We confirmed that all individuals at F1 were heterozygous (+/Ccα6 3aQ68*-K352* or +/Ccα6 3aQ68* ) (  Table 4. Cross-resistance of JW-100 s strain to imidacloprid, fipronil, malathion and lambda-cyhalothrin. † Number of flies considered in the Probit analysis (including non-treated). ‡ Lethal concentration (LC 50 ) in ppm of insecticide in the diet for the feeding bioassays at 48 h. For malathion and lambda-cyhalothrin, % mortality (M) at two discriminant concentrations were tested. Lethal dose (LD 50 ) in µg/g of insect (fresh weight assuming an average weight of 10 mg) by topical application at 48 h (a 0.5 μL drop of insecticide solution in acetone or acetone alone was applied to the dorsal thorax of each fly by using an automatic microapplicator). § Resistance ratio (RR) = LC 50 (JW-100 s strain)/LC 50 (C strain) or LD 50 (JW-100 s strain)/LD 50 (C strain). The fiducial limits for RR were calculated according to Robertson and Preisler (1992). ♯ RR is significant (P < 0.05) if the 95% FL does not include 1. *Good fit of the data to the probit model (P > 0.05).

Strain
Gen.   Table 6) were only slightly higher than the LC 50 value estimated for the control strain (0.58 ppm, Table 1), much smaller than that calculated at F36 for JW-100 s by feeding bioassays (157 ppm, Table 2). Moreover, the LC 50 values did not differ between the two reciprocal crosses, suggesting an autosomal rather than X-linked inheritance pattern. The resulting dominance values (D LC ) calculated in relation to this LC 50 were close to 0 (0.05 and 0.06, Table 6). These results indicated that spinosad resistance was inherited as an autosomal and almost completely recessive trait.

Spinosad
In order to analyse the linkage between spinosad resistance and Ccα6 3aQ68*-K352* and Ccα6 3aQ68* alleles, we next obtained F2 offspring from the reciprocal crosses, and we treated these individuals with a discriminating concentration of spinosad (10 ppm) causing approximately 90% mortality (Table 6). We genotyped all surviving individuals and a number of dead individuals from this treatment (Table 7). Notably, none of the 76 survivors carried the + allele, with a majority homozygous for the Ccα6 3aQ68*-K352* allele (67 individuals, frequency of 0.88), and a minority heterozygous for the two mutant alleles Ccα6 3aQ68* and Ccα6 3aQ68*-K352* (7 individuals, frequency of 0.09). Surprisingly, two individuals (frequency of 0.03) among the survivors were heterozygous for the mutation 3aQ68* and homozygous for the mutation K352* (as verified by sequencing genomic DNA), thus carrying the allele Ccα6 3aQ68*-K352 in combination with an allele not previously detected, Ccα6 K352* . On the other hand, all dead individuals genotyped from the F2 had at least one + allele, either in heterozygosis (42 individuals) or in homozygosis (18 individuals) (Table 7). Altogether, these results confirmed the linkage between resistance to spinosad and the Ccα6 3aQ68* and Ccα6 3aQ68*-K352* alleles in the JW-100 s strain, as well as the recessive character of the mutant alleles.

Discussion
In a number of insect species, high levels of spinosad resistance have been associated with alterations in the α6 gene affecting all the α6 isoforms. Studies on D. melanogaster have found that deletion of the nAChR α6 subunit gene, point mutations generating premature stop codons, or other mutations next to the Cys-loop motif result in resistance to spinosad 6,7,22,23 . Resistance to spinosad has also been associated in P. xylostella and B. dorsalis with truncated α6 transcripts due to mis-splicings, insertions or deletions 24,25,30 ; in P. xylostella with a 3-residue deletion in the TM4 domain 26 ; in T. absoluta with exon 3 skipping in α6 transcripts 27 ; and in F. occidentalis, T. absoluta and T. palmi with a point mutation G275E at a transmembrane domain 28,29,40 . Here, we demonstrate that the resistance to spinosad selected in the JW-100 s strain of medfly is associated with three different mutant alleles of the nAChR α6 subunit gene. The Ccα6 3aQ68*-K352* allele contains the mutations 3aQ68* and K352*, and would lead to isoforms truncated at exon 3a, and to 3b isoforms truncated at exon 10 (Fig. 1B). Thus, none of the Ccα6 transcripts from homozygous individuals for the Ccα6 3aQ68*-K352* allele could generate complete Ccα6 protein products. On the other hand, the Ccα6 3aQ68* allele contains the mutation 3aQ68* that would affect only the isoforms containing exon 3a while generating wild-type full-length Ccα6 isoforms with exon 3b (Fig. 1C). Similarly, the Ccα6 3aAG>AT allele, which contains the AG > AT mutation present in the 5′ splicing site of exon 3a, would produce full-length 3b transcripts and incomplete transcripts that skip exon 3a and join exons 2 and 4 (Fig. 1D). Hence, our results suggest that the absence of the isoforms containing exon 3a could be enough to confer resistance, despite the expression of full-length 3b isoforms.
The specific involvement of α6 isoforms in spinosad susceptibility has previously been analysed in D. melanogaster 7,8 . Taking advantage of the Gal4-UAS system, Perry et al. (2015) 8 demonstrated that the expression of any of the four different Dmα6 isoforms (3a8a, 3a8b, 3b8a or 3b8b) is sufficient to restore spinosad susceptibility in an α6-deficient strain of D. melanogaster. However, the rescue of resistance by expressing a single Dmα6 transcript using the Gal4-UAS system might be influenced by the particularities of the system, which usually confers a higher expression level than that in basal wild-type conditions. Moreover, the authors observed that leaky expression of the parental UAS constructs in the absence of Gal4 driver, particularly the UAS constructs containing 3a isoforms, is sufficient to partially rescue the susceptibility to spinosad in an α6-deficient strain. They propose that some isoforms could be "more responsive to spinosad or better able to form functional receptors than others" 8 . This hypothesis is in concordance with our results that suggest that removal of 3a isoforms could be sufficient to confer resistance to spinosad in the medfly. On the other hand, an alternative explanation is that the alteration of the expression of one isoform in medfly may interfere either with the expression of other isoforms or with the  correct assembly of nAChR receptors targeted by spinosad. Finally, we cannot completely discount that other mutations that escaped the limits of our analysis, for example mutations in non-encoding regions, could affect the transcript stability or the translation of full-length wild-type 3b isoforms in Ccα6 3aQ68* and Ccα6 3aAG>AT alleles. Thus, further experiments generating mutations affecting specific isoforms would be required to confirm that the alteration of only some α6 isoforms is sufficient to achieve resistance in the medfly. The frequency of the different alleles observed in the JW-100 s population changed during the selection process. Interestingly, Ccα6 3aAG>AT and Ccα6 3aQ68* alleles were the most frequent when the selection pressure was relatively low, but they were almost completely substituted by Ccα6 3aQ68*-K352 when the concentration of spinosad was raised to 100 ppm (see Tables 2 and 5). These changes may be explained by an unbalanced contribution of the different alleles to resistance and/or by the existence of an undetermined fitness cost associated with the different mutations. The unbalanced contribution to resistance between the Ccα6 3aAG>AT and Ccα6 3aQ68* alleles could explain the higher frequency of individuals carrying the allele Ccα6 3aQ68* in survivors to 240 ppm at generation F29 (Table 5). However, the decreased frequency of this allele detected in generation F48 may be indicative of an undetermined biological cost that could have influenced the selection in favour to Ccα6 3aQ68*-K352* . The fitness cost associated with mutations in the α6 gene has not been elucidated in depth in insect species yet. This gene is not essential for survival, as demonstrated in knockout strains of D. melanogaster that show no evidences of fitness disadvantage 6 . Moreover, quantification of biological parameters (development, fecundity, fertility and life span) in spinosad resistant strains of H. virescens 41 , P. xylostella 42 and H. armigera 43 only showed slight reductions in fitness, and no negative effect on fitness were reported for a F. occidentalis resistant strain 44 . However, reversion of resistance concomitantly with the relaxation of spinosad selection pressure, an indicator of biological cost, has been reported for M. domestica 45 , L. trifolii 12 , S. litura 46 , P. xylostella 42 , T. absoluta 47 and F. occidentalis 48 . In mammals, diverse nAChRs are present in multiple tissues and participate in different functions, including learning and other complex behavioural traits 49 . Thus, we cannot discount the possibility that the loss of function of truncated α6 transcripts may affect some behavioural pattern in resistant individuals, making them less competitive than wild-type individuals.
None of the mutations associated with spinosad resistance in this report were found in the analysis of the F0 individuals directly collected in the field. These results could be explained by a low frequency of the resistant alleles in the field. However, it is possible that mutations could have spontaneously occurred along the selection process in the laboratory. In addition, a distinct unknown mechanism may have contributed to spinosad tolerance at the first generations of selection. This mechanism could be responsible for the cross-resistance to imidacloprid and fipronil that was observed in feeding bioassays but not in topical treatments. Moreover, it could also account for the slight but significant effect of the synergists PBO and DEF on spinosad resistance. Remarkably, the malathion resistant medfly strain W-4 Km, crossed with the population collected in Xàbia at the beginning of the selection process (see material and methods section), shows a low but significant cross-resistance to spinosad and lufenuron that could be attributed to a detoxification mechanism 35 . Reports on low levels of spinosad resistance mediated by suspected detoxification are abundant in the scientific literature 21 .
There is limited sustainability of the use of insecticides in bait sprays for the control of medfly, according to the recent reports on the resistance to malathion and lambda-cyhalothrin detected in Spanish field populations 31,33 . Other insecticides are currently registered for bait treatments against medfly in citrus groves. Deltamethrin is used in lure and kill traps, but resistance to lambda-cyhalothrin may confer cross-resistance to this insecticide as demonstrated in a laboratory strain 33 . Our analysis of the susceptibility to spinosad in populations collected at different locations and in different years indicates that resistance has not evolved in Spanish field populations (Table 1 and Supplementary Information, Table S2), and as far as we know no previous reports exist on resistance to this insecticide in medfly field populations. We have demonstrated that the Ccα6 3aQ68*-K352 allele is almost completely recessive. In addition, other resistant alleles (Ccα6 3aAG>AT , Ccα6 3aQ68* and Ccα6 K352* ) are likely to be recessive, as all individuals surviving spinosad carried either one of these resistance alleles in homozygosis or two of them in heterozygosis (see Tables 5 and 7). Thus, the presence of at least one copy of the wildtype nAChR α6 gene is sufficient to avoid spinosad resistance. Accordingly, spinosad resistance in D. melanogaster α6 null strains 6 and in the Pearl-Sel strain of P. xylostella that produces mis-spliced α6 transcripts 13,24 are also recessive. The recessive nature  of the resistance alleles, together with a potential fitness cost for resistance, may be an impediment for their selection in the field. However, resistance to spinosad caused by residue substitutions at the conserved Cys-loop of the subunit may be incompletely dominant, as is the case for the P146S mutation recently reported in D. melanogaster 7 . Moreover, our study highlights the plasticity of the Ccα6 gene to acquire and support mutations. This plasticity could favour resistance evolution in the field, since different resistance alleles at low frequency could be combined in heterozygous individuals that would survive spinosad treatment. This, together with the successful laboratory selection of Xàbia field population to obtain the resistant strain JW-100 s, highlights that resistance could evolve if medfly management relies extensively on the use of this insecticide. This scenario underscores the need to implement resistance management strategies to counteract the selection of resistance in field populations of medfly.
Ethical approval. This article does not contain any studies with human participants or animals performed by any of the authors.

Data Availability
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.