Stable expression and functional characterisation of the diamondback moth ryanodine receptor G4946E variant conferring resistance to diamide insecticides

Diamides, such as flubendiamide and chlorantraniliprole, belong to a new chemical class of insecticides that act as conformation-sensitive activators of insect ryanodine receptors (RyRs). Both compounds are registered for use against lepidopteran species such as the diamondback moth, Plutella xylostella, a notorious global pest of cruciferous crops. Recently acquired resistance to diamide insecticides in this species is thought to be due to a target-site mutation conferring an amino acid substitution (G4946E), located within the trans-membrane domain of the RyR, though the exact role of this mutation has not yet been fully determined. To address this we have cloned a full-length cDNA encoding the P. xylostella RyR and established clonal Sf9 cell lines stably expressing either the wildtype RyR or the G4946E variant, in order to test the sensitivity to flubendiamide and chlorantraniliprole on the recombinant receptor. We report that the efficacy of both diamides was dramatically reduced in clonal Sf9 cells stably expressing the G4946E modified RyR, providing clear functional evidence that the G4946E RyR mutation impairs diamide insecticide binding.

of diamides has increased substantially, currently accounting for 8% of the global insecticide market 10 , and over that period these insecticides have become one of the primary chemical control agents against lepidopteran pests due to their favourable biological, ecological and toxicological attributes 6,11 .
The diamide insecticides flubendiamide and chlorantraniliprole both have a similar mode of action, and act to prolong RyR channel opening, resulting in uncoordinated muscle contractions in intoxicated pest insects 11,12 . Due to their novelty, relatively little is known about how diamides interact with insect RyRs and how their high insecticidal specificity is achieved. Studies with chimeric proteins (insect-mammal and insect-nematode chimeras) pinpointed the probable location of diamide binding within the transmembrane region of the receptor 13,14 . However there is growing evidence for different diamide-binding profiles across various insect orders [15][16][17] , suggesting there may be structural variability within the binding region.
The first diamide control failures were reported for the diamondback moth, P. xylostella, one of the most destructive pests of cruciferous crops, accounting for $4-5 billion pounds worth of annual worldwide crop losses 18 , and being the most extensively distributed of all Lepidoptera globally 19 . Initial reports of resistance came from the Bang Bua Thong district of Thailand just 18 months after flubendiamide was launched. The following year (2010) P. xylostella resistance to diamides was reported in Cebu, Philippines and in 2011 in Yin Ling and Chang Hwa, Taiwan and in the Guangzhou and Guandong provinces of Southern China 20 . Reports of resistance were followed by the discovery of a G4946E amino acid substitution ( Fig. 1) within the transmembrane spanning region of RyR of highly resistant populations of P. xylostella from Thailand (Bang Bua Thong) and the Philippines (Sudlon, Cebu) 21 . The same mutation was later independently found in diamide resistant populations from China 22 . Binding studies to native membranes isolated from resistant populations collected from Southern China with fluorescently labelled diamide probes indicated a decreased affinity for diamides in resistant individuals 23 . However, the study largely failed to provide convincing functional evidence that the G4946E mutation confers a significant level of target-site resistance.
In this present study we report the cloning, characterization and successful expression of the full-length P. xylostella RyR. The cell lines expressing the receptor were used to conduct a more in-depth analysis of the effects of the G4946E substitution on receptor sensitivity to diamide insecticides and on general channel function.

Results
Cloning and characterisation of a functional RyR from P. xylostella. The cloned P. xylostella RyR cDNA in this study encodes a 5118 amino acid protein (NCBI accession AFW97408), with a high degree of similarity to other successfully expressed insect RyRs including those from Drosophila melanogaster (78.5%) and Bombyx mori (91.5%) 13,24 . It shares all of the features described previously for the RyR isolated from P. xylostella in other studies [25][26][27] , including a GXRXGGGXGD selectivity filter motif in the transmembrane domain and MIR (212-393), SPRY (664-803, 1091-1214, 1551-1693) and RIH domains (2232-2454) in the cytosolic region. The lepidopteran specific amino acids at positions N(4953), N(4955), N(4966), L(4981), L(5012), N(5044), T(5095) are also present 28 . Overall a large number (over 200) of silent SNPs were found within our wild type version of the RyR (ROTH WT), with a clustered distribution (Fig. 2b). Four alternative splice sites were identified, three of which correspond to previously described splice sites IS2, IS3 and IS10 in Wang et al. 26 who also reported an additional seven splice variants. Multiple alignments of the P. xylostella RyR sequences with other insect RyRs suggest that the alternative splice forms identified in the moth are most likely not representative of the prevalent form of the receptor. This hypothesis is supported by the report of alternative splicing frequency published by Wang et al. 26 . Similar polymorphism in a lepidopteran RyR gene was also described in the tobacco budworm Heliothis virescens 29 . Functional analysis of transiently expressed RyR channels. Spodoptera frugiperda Sf9 cells transformed with the P. xylostella ryanodine receptors were loaded with FURA2, a fluorescent Ca 2+ indicator, and were exposed to caffeine (30 mM), followed by high concentrations of flubendiamide or its somewhat more water soluble analogue flubendiamide sulfoxide (Fig. 3). Sf9 cells expressing the ROTH WT or the G4946E RyR were able to respond to repeated caffeine applications, as demonstrated by a brief elevation of intracellular Ca 2+ concentration measured at 72 h post transfection. Flubendiamide sulfoxide also showed reversible activation of the calcium release in cells expressing ROTH WT RyR. Conversely, exposure of the ROTH WT RyR cells to flubendiamide resulted in permanently increased Ca 2+ levels and abolished any further caffeine evoked activation for the duration of the experiment, suggesting a fixing of the receptor in a permanently open state. Sf9 cells expressing the G4946E RyR variant did not show any calcium release when exposed to flubendiamide (Fig. 3). The more water soluble analogue, flubendiamide sulfoxide also failed to induce any release of calcium (data not shown). Additionally, unlike ROTH WT RyR cells, the G4946E variant remained sensitive to caffeine after exposure to flubendiamide. In the control un-transfected cells and cells exposed only to Cellfectin, caffeine and diamides failed to evoke any such responses.
Tritium labelled ryanodine binding experiments on isolated mixed-membrane preparations from Sf9 cells transiently expressing ROTH WT or G4946E RyRs gave B max and K d values of 2447 ± 425 dpm (244.9 ± 42.5 fmol/mg) and 12.89 ± 4.5 nM for WT and 2671 ± 401 dpm (267 ± 40 fmol/mg) and 14.69 ± 4.18 nM for the G4946E variant (Fig. 4). The K d values for both constructs were higher than seen for native preparations of rabbit skeletal muscle RyR (2.04 ± 0.12 nM) and H. virescens RyR (3.82 ± 0.39 nM), which also have much higher B max values; 4.57 ± 0.32 pmol/mg (rabbit) and 2.41 ± 0.17 pmol/mg (H. virescens) 30 . However in a different study of H. virescens native membrane preparations, the respective K d value was 13.9 ± 3.8 nM, making it comparable to the Sf9 expressed P. xylostella RyRs, while the B max values were between 1-1.5 pmol/mg 31 . Apparent differences in K d and B max with those reported in other studies could likely be explained by the different experimental conditions. It is clear that the presence of the G4946E mutation does not appear to have any significant effects on the P. xylostella RyRs ability to bind ryanodine, since both K d and B max values are similar. Functional analysis of stably expressed RyR channels. Single Sf9 cell clones stably expressing either the wild-type (Sf9-wtRyR) or the G4946E (Sf9-mRyR) RyR channel were isolated after a few cycles of clonal selection. Intracellular calcium signals induced by 30 mM caffeine in Sf9-wtRyR increased rapidly to the maximum in less than one minute after application and then decreased to baseline levels (Fig. 5). The transient nature of the calcium signals induced by millimolar caffeine concentrations was comparable to those measured previously in isolated H. virescens neurons 31 . This may be explained by compensatory calcium re-uptake mechanisms of the sarco-/endoplasmic reticulum, in particular the Ca 2+ -ATPase (SERCA) 32,33 . The longer time scale of the caffeine-induced calcium transients in the  Sf9-wtRyR cells, as compared to the isolated H. virescens neurons may indicate that excitable cells such as neurons and muscle cells can fine-tune intracellular calcium concentrations more efficiently than can the Sf9 cells lacking specific RyR accessory proteins.
In contrast to caffeine, the diamide insecticides chlorantraniliprole and flubendiamide-sulfoxide evoked increased intracellular calcium concentrations that persisted during the time period of the measurement. This may be explained by the mechanistic hypothesis that diamides preferentially bind to the calcium-conducting conformational state of the RyR and stabilize the open channel due to the low nanomolar equilibrium dissociation constant of the diamide-RyR complex 34 .
Measurements of the integrated calcium signals as a function of the diamide concentration, gave characteristic dose-response curves, from which apparent EC 50 concentrations for flubendiamide-sulfoxide and chlorantraniliprole were calculated (Fig. 6). These values represent the effector concentration causing half-maximal integrated calcium signals in the cells. The EC 50 of flubendiamide sulfoxide was 245 nM± 46 nM (standard error) and the EC 50 of chlorantraniliprole was 17 nM± 2 nM (standard error). It should be noted that the EC 50 values are not only determined by the compound's affinities to the RyR, but also by their availability at the intracellular target site as influenced by their physico-chemical properties.
The efficacy of both diamides was reduced dramatically in the Sf9-mRyR cells expressing the G4946E RyR. It was possible to measure a full dose-response curve of chlorantraniliprole on the G4946E RyR, from which an EC 50 value of 3715 nM± 776 nM (standard error) was calculated. The limited solubility of the flubendiamide sulfoxide at concentrations exceeding 20 μ M prevented the measurement of complete dose-response curves and, consequently, an accurate EC 50 value could not be determined beyond an approximation of > 20,000 nM.  The results show that chlorantraniliprole had roughly two orders of magnitude less effect on the P. xylostella RyR channel with the G4946E substitution, indicating a drastic reduction of the receptor´s affinity to the diamide and making it very likely that the G4946E substitution is responsible for the observed resistance to diamides in P. xylostella.

Discussion
Diamide insecticides such as chlorantraniliprole and flubendiamide are a new class of insecticide recently introduced to the market to control a broad range of herbivorous pest insects, particularly of the order Lepidoptera. An over reliance on the use of diamide insecticides against diamondback moth has led to a rapid resistance development, particularly in Asia where the field efficacy of these compounds has now been seriously compromised. Larvae collected from the Philippines and Thailand in 2012 were found to be over 200-fold resistant to both chlorantraniliprole and flubendiamide compared to susceptible strains 21 . Non-synonymous mutations in each of the resistant strains that in both cases lead to a glycine to glutamic acid substitution (G4946E) in the protein were identified. The independent evolution of the same amino acid substitution, within a highly conserved region of the RyR channel, the C-terminal membrane-spanning domain, in two geographically separated resistant strains of P. xylostella strongly suggested a causal association with diamide resistance. The same mutation was subsequently independently identified in field populations of P. xylostella from China, 303-658 fold resistant to chlorantraniliprole 23 . A recent genotyping study 35 has confirmed the global presence of the G4946E mutation in ten different countries where diamide insecticides have largely failed to control diamondback moth populations, a significant correlation further highlighting the likely role of this target-site mutation in conferring resistance to diamides.
Radioligand binding studies with a tritiated flubendiamide derivative [ 3 H] PAD1 using P. xylostella thoracic microsomal membrane preparations from susceptible and resistant (Philippine) strains 35 has provided further compelling evidence for the involvement of the RyR G4946E mutation on both diamide specific binding and its concentration dependent allosteric modulation of [ 3 H] ryanodine binding 17,34 . In contrast to the susceptible strain, with reported K d and B max values of 2.7 ± 0.23 nM and 8.3 ± 0.19 pmol mg −1 , for the Philippine strain saturation binding with the tritiated flubendiamide analogue was not reached and meaningful equilibrium kinetics could not be calculated, suggesting that the G4946E mutation confers target-site resistance to diamide insecticides. EC 50 values for ryanodine binding stimulation in the Philippine strain were at least 100-fold higher, with resistance ratios of > 450 and 159 fold for flubendiamide and chlorantraniliprole, respectively. Reciprocal crosses of the Philippine strain, homozygous for G4946E, with a susceptible laboratory strain yielded F1 progeny with a diamide susceptible phenotype, suggesting an autosomal, recessive mode of inheritance. Subsequent back-crosses with the parental lines indicated a near monogenic inheritance for diamide resistance in the Philippine strain. The data reported in this current study, whereby we directly investigated binding of diamide insecticides to recombinant RyR proteins, now unambiguously provides clear functional evidence that the G4946E RyR mutation greatly impairs diamide insecticide binding and is responsible for the resistant phenotype exhibited by the resistant P. xylostella strains.
Mapping of the G4946E substitution onto the recently published high resolution crystal structure of the (closed state) rabbit RyR1 channel 36,37 suggests that this residue in the WT P. xylostella channel potentially acts as a glycine hinge at the interface between the transmembrane spanning S4 helix and the S4-S5 linker domain (Fig. 1), hence residue changes at this position are likely to have a major impact on movement of the S5 and S6 helices that control opening and closing of the channel pore, and have a direct knock on effect on binding of diamide insecticides to the receptor. Three additional point substitutions (E1338D, Q4594L and I4790M) in the P. xylostella RyR were recently reported in a field population collected from Yunnan province, China 38 (corresponding to residues E1333, Q4548, I4744 in the channel expressed in this current paper), exhibiting up to 2128-fold resistance to diamide insecticides. It has been shown that the I4790M change (located midway along the RyR S2 helix) may lie in close proximity to the G4946 residue in the 3D structure of the RyR 35 and consequently these two residues may define the diamide binding site on the receptor. The isoleucine residue at position 4790 is specific to Lepidoptera, being a methionine in all other insects, spiders and mites, a leucine in nematodes and a cysteine in mammalian RyRs, so this residue may be responsible for the differential sensitivities of the P. xylostella and other insect RyR channels to flubendiamide and chlorantraniliprole chemistries 16,17 .
Chlorantraniliprole and flubendiamide have also been extensively used to manage the tomato leaf miner Tuta absoluta (Lepidoptera: Gelechiidae), an invasive pest of tomato crops that is rapidly expanding around the globe. High resistance levels of up to 2,414-and 1,742-fold for chlorantraniliprole and flubendiamide, respectively have been detected in a T. absoluta population originating from Sicily (Italy), suggestive of a similar target-site insensitivity in this species 39 . Cross resistance between flubendiamide and chlorantraniliprole has also recently been reported for the smaller tea tortrix, Adoxophyes honmai 40 , with resistance ratios of 105-fold and 77.2-fold respectively. Even though the level of relative activity by resistant individuals may vary among the two actives, ultimately the result is inadequate field control. Currently, resistant populations in most countries are still fairly localised and many crop production areas will still maintain susceptible populations. Significant resistance management efforts are therefore needed to protect diamide chemistry as a useful pest management tool, especially with the introduction over the next few years of additional manufacturers competitively selling multiple brands containing different diamide actives. It has been recognised that there will need to be a concerted and coordinated programme of insect monitoring for early detection of diamide tolerance within populations, coupled with the use of appropriate insecticide rotations with different modes of action chemistries, to alleviate this emerging problem 41 . The cell lines stably expressing RyR variants generated in this study also offer opportunities for a high throughput counter-screening of chemical libraries to seek for compounds overcoming the resistance.

Methods
Chemicals. All stand-alone chemicals used for the preparation of bacterial media were purchased from Sigma. The DMSO used for dilution of all active compounds was of analytical grade (purity ≥ 99%) and supplied by Sigma. [ 3 H] ryanodine (spec. activity 2 TBq mmol −1 ) was supplied by PerkinElmer. Flubendiamide, flubendiamide sulfoxide and chlorantraniliprole were of technical grade (purity > 98%) provided in-house (Bayer CropScience) and purchased as analytical standard from Fluka Chemicals (Buchs, Switzerland), respectively. Analytical grade caffeine was from ReagentPlus ® (Sigma). Cloning of P. xylostella RyR. Initially 3 small (approx. 500-1000 bp) fragments of the RyR were PCR amplified, using degenerate primer pairs based on the silkworm B. mori RyR sequence (NCBI accession number XM_004924859.1). Gene specific primers were then designed (based on the sequence information obtained from these fragments), and two large fragments, spanning the approximately 7 Kb gaps between the 3 initial fragments, were amplified, cloned and sequenced. The extreme 5′ and 3′ ends of the receptor were obtained through RACE.
For assembly of a full-length RyR ORF, four overlapping fragments (F1-F4) were amplified by PCR and cloned (Fig. 2a). Sequencing of individual clones for each of the four fragments revealed that the RyR obtained from the ROTH strain is polymorphic, with significant single nucleotide polymorphisms (SNP) between clones and alternative splice site variability (Fig. 2b). Due to the high number of SNPs, a PvuI restriction site had to be engineered back into the F2 fragment selected for assembly into the full-length ORF. For the preparation of the RyR G4946E variant, the glutamic acid (E) was introduced into fragment F4 prior to final assembly of the full-length ORF. Initial assembly of both the full-length ROTH WT and G4946E RyR variant was achieved through simultaneous ligation of the 4 pre-digested fragments F1-F4 (in an equal weight ratio) into the vector pcDNA 3.1(-), between the NotI and KpnI sites of the plasmids multiple cloning site (MCS). The final assembled ORF comprised 15357bp, coding for a 5118 amino acid protein The G4946E diamide resistance associated substitution 21 is located at amino acid position G4900 in the expressed version of the P. xylostella channel. The full-length ORF was then moved across into the expression vector pIZ/V5-His, pre-modified by engineering an NruI site at position 695 downstream of the MCS, within the V5 epitope. The full-length RyR ORF was excised from the pcDNA 3.1(-) plasmid using NotI and PmeI and ligated into the modified pIZ V5/His, between the NotI and NruI sites.
Degenerate PCR amplifications were carried out using gradient PCR and REDtaq mastermix (Sigma) in 25 μ l reactions consisting of 12.5 μ l mastermix, 2 μ l of each degenerate primer (20 mM), 1 μ l cDNA and water. Cycling conditions were 95 °C for 2 min followed by: 35 cycles of 95 °C for 20 s, 45-50 °C for 45 s and 72 °C for 3 min with a final extension time of 72 °C for 5 min.
The initial long fragment PCR amplifications (25 μ l) used Long Range PCR mix (Thermo) and consisted of 2.5 μ l reaction buffer containing MgCl 2 , 1 μ l cDNA, 1 μ l dNTPs (10 mM), 0.5 μ l of each primer (20 mM) and 2.5 U of enzyme mix. Cycling conditions were 94 °C for 2 min followed by: 35 cycles of 94 °C for 10 s, 50 °C for 20 s, 68 °C for 25 min, followed by a final extension of 68 °C for 15 min. The primer sequences for all of the cloning steps are shown in Tables 1, 2 and 3. All PCR amplified fragments were visualized on 1% (w/v) agarose gels and purified using a QiaQuick ® gel extraction kit (Qiagen), following the manufacturer's guide, prior to downstream applications.
All of the PCR-amplified fragments were ligated into a pJET 1.2 blunt end vector (part of CloneJET cloning kit (Thermo)) following the kit's protocol. Fastdigest ® restriction enzymes (Thermo) were used for all restriction digests, following the manufacturer's instructions. The full-length ligations into pcDNA 3.1(-) and pIZ/V5-His were done in 30 μ l reactions, using 3 μ l of 10 × ligation buffer and 10U of T4 DNA ligase (Thermo) for 2 h at room temperature. All plasmids were transformed into XL-10 gold E. coli (Agilent), recovered and grown at 30 °C to minimize potential rearrangements within the plasmid. Transfections grade plasmid preps were obtained using a Plasmid maxi prep plus kit (Qiagen) following the manufacturer instructions. All mutagenesis reactions were done using a Quikchange II mutagenesis kit (Agilent) following the kit's manual (mutagenesis primers are shown in Table 4).      Selection of stable cell lines. Cells were transfected as described above and subjected to selection with 300 μ g ml −1 zeocin by the dilution method, essentially as described in the manufacturer´s manual (Invitrogen). After a few successive rounds of clonal selection, zeocin-resistant cell clones were tested for functional RyR expression by calcium fluorescence measurements.

Sf9 Transfections and membrane preparation.
Calcium fluorescence measurements. Calcium