Inheritance and fitness costs of resistance to Bacillus thuringiensis toxin Cry2Ad in laboratory strains of the diamondback moth, Plutella xylostella (L.)

The diamondback moth, Plutella xylostella (Lepidoptera: Plutellidae), is one of the main pests of Brassica crops worldwide. Management of P. xylostella is particularly challenging, as different field populations have readily acquired resistance to a wide range of insecticides, including Bacillus thuringiensis (Bt) toxins. In this study, a novel strain of P. xyllostela (Fuzhou-R2Ad) with 120-fold resistance to Bt Cry2Ad was selected in the laboratory, after screening for 66 generations from the susceptible strain Fuzhou-S. In the absence of Bt Cry2Ad toxin, the Fuzhou-R2Ad had significantly lower fitness as compared to the susceptible strain, which might be related to induced genetic changes to Bt toxins. We used several models to measure the dominance levels of insecticide resistance among different strains and found an incompletely recessive inheritance pattern of the Fuzhou-R2Ad resistance, which might be controlled by multiple genes. This study constitutes the first report of laboratory-acquired resistance to Cry2Ad toxin in P. xylostella. Our work presents further insights into the mechanism of Bt resistance and has immediate implications for the integrated pest management of P. xylostella globally.

calculated based upon the fitness of particular genotypes in insecticide-treated areas 23,24 . Last, a general formula has been proposed for dominance levels in relation to insecticide resistance 25 . Overall, dominance levels can be calculated for different traits, including insect fitness in insecticide-treated or untreated area D WNT . Although the dominance level could be estimated by D LC , D ML , D WT and D WNT , it may still be varied by environmental influences, genetic information and the selection of an insecticide resistance allele. Using D LC and D ML models, it has been shown that the resistance in Cry1Ac-selected strains was incompletely recessive in a field-derived population of DBM 6 . Pereira et al. has demonstrated a recessive inheritance of Cry1F resistance in European corn borer Ostrinia nubilalis, which was indicated by a dominance level D LC less than 0.11 16 . However, to our knowledge, there are no published studies that utilize the various models, especially D WNT , to fully evaluate the degree of the dominance.
In the present study, we evaluate, for the first time, the inheritance properties of a laboratory DBM strain with high resistance to Bt Cry2Ad, by comparing dominance of insecticide resistance between the susceptible strain, the positive and negative cross of the resistant strain and the backcross. Furthermore, we investigate levels of dominance and inheritance of resistance to Cry2Ad toxin in the hybrid, resistant and susceptible strains without selection pressure. Additionally, we estimate whether inheritance of Bt Cry2Ad resistance in P. xylostella is controlled by a single-gene or multiple genes. The results of this research have direct implication for resistance management of DBM to Cry2Ad, and can provide further information to advance the effective control of DBM globally.

Materials and Methods
Cry Toxin. Cry2Ad 27 . The Fuzhou-S strain has been kept for over 150 generations under greenhouse conditions without exposure to insecticides, with individuals reared on potted radish seedlings (Raphanus sativus L. var. sativus) under the condition of 25° ± 1 °C, 65 ± 5% RH and 16 L:8D photoperiod.
A resistant strain was derived from the Fuzhou-S strain, by exposing the 3rd instar larvae of DBM to R. sativus leaves treated with Cry2Ad toxin. Fresh and untreated R. sativus leaves were dipped into the Cry2Ad toxin protein solution at LC 75 concentration for 10 s, and excess solution was wiped off with filter paper. After 48 h, the surviving larvae were then selected, allowed to pupate and chosen for production of further progeny 28 . Similar to the Fuzhou-S strain, the resistant Fuzhou-R2Ad strain has been maintained for about 70 generations in the laboratory without any exposure to insecticides except for Cry2Ad.
Bioassay. Following the procedures as outlined above, R. sativus leaves (ca. 10 mm diameter) were treated with five gradient concentrations of Cry2Ad solution. After drying, leaves were fed to the 3 rd -instar P. xylostella larvae that had previously been starved in clear plastic cups (78 mm (top) and 51 mm (bottom) in diameter, 82 mm height) for 2 h 29-31 . Each concentration was tested for a batch of 12 DBM larvae, and the experiments were independently repeated three times with 10 leaves in each replicate. In a control group, larvae were fed with leaf disks (ca. 10 mm diameter) that had been treated with distilled water containing 0.2% Triton X-100.
The treated larvae were then transferred to a climate chamber at 25° ± 1 °C, 65 ± 5% RH, and a 16 L:8D cycle. After 48 h, fresh untreated R. sativus leaves were added. Mortality of larvae was recorded after 72 h, and a toxicity regression curve was developed to estimate the value of LC 50 with 95% confidence intervals.
Hybridization. After pupation, each pupa was transferred individually into a collection tube for further eclosion. Emerged adults were sexed, and used for production of a F1 generation through reciprocal mass crosses. For one cross, 30 Fuzhou-R2Ad females were allowed to mate with 30 Fuzhou-S males in one laying cage (100 mm diameter and 80 mm height). For a second cross, 30 Fuzhou-S females were paired with 30 Fuzhou-R2Ad males 32 , larvae from the two parental colonies were defined as F1 (Fuzhou-R2Ad♀ × Fuzhou-S♂) and F1' (Fuzhou-R2Ad♂ × Fuzhou-S♀), and subject to the above bioassays. Subsequently, F2 progeny was obtained through single-pair crosses between F1 progeny, and a backcross (BC) was produced by pairing a F1 hybrid with the Fuzhou-S strain (F1 × Fuzhou-S). Lastly, 20 susceptible adults (i.e., 10 females and 10 males) were mixed with 20 resistant adults (10:10 sex ratio) for a pooled hybrid (R × S). Dominance of Cry2Ad toxin resistance in F1, F1' and BC hybrids were determined based on the probit analysis (visualised by slopes of log dose-probit line (LD-P line)), LC 50 value and corresponding 95% confidence limits.
Fitness tests. Newly-hatched larvae from Fuzhou-S, Fuzhou-R2Ad, F1 and F1' hybrid populations were randomly chosen, and individualized on potted turnip sprouts (ca. 40 mm diameter). On a daily basis, development of P. xylostella was monitored and the relevant biological parameters, including mortality, pupation rate, eclosion rate, and adult sex ratio, were recorded. Single-pair crosses of P. xylostella adults were conducted in 60 mm Petri dishes lined with moist filter paper, and mated females were allowed to lay eggs on the moist filter paper. Mated females were fed with 10% honey solution, and fecundity of each strain was recorded until all moths died.
Eggs were individually collected and incubated in Petri dishes, and egg eclosion rates were computed. Net population growth rate (R 0 ) was determined, defined as the ratio of new larvae (N n+1 ) to the initial number (N n ). The relative fitness of the resistant strain was calculated by: www.nature.com/scientificreports www.nature.com/scientificreports/ = R R Relativefitness (resistant or hybrid strain)/ (susceptiblestrain) 0 0 Data analysis. For each bioassay, LD-P line, LC 50 value, 95% confidence limits and the relative standard deviation were assessed. Two LC 50 values are considered to be significantly different (P < 0.05) if their 95% confidence intervals do not overlap 33 . Based on the LC 50 , the resistance ratio was defined as the ratio between the LC 50 value of Fuzhou-R2Ad, F1 or BC and that of the susceptible strain (i.e., Fuzhou-S). Degree of dominance (D) at LC 50 was calculated by: where LC R , LC RS and LC S represent lethal concentrations for resistant homozygotes, heterozygotes, and susceptible homozygotes, respectively. The value of D ranges from −1 to 1, representing a complete recessive towards an absolute dominance. Furthermore, D LC , was calculated by: which is equal to (D + 1)/2 34 . Hence, the D LC value varies between 0 (recessive resistance) and 1 (dominant resistance). We equally applied the D WT model to evaluate relative fitness of dominance under Bt insecticide selection. D WT was calculated by: where W TSS , W TRS and W TRR represent the relative fitness at a specific insecticide concentration for susceptible homozygotes, heterozygotes, and resistant homozygotes, respectively. If susceptible and resistant strains are considered as homozygous genotypes, D WT will be taken as h 23,35 . In a similar fashion as D LC , the h value ranges from 0 to 1 (i.e., from completely recessive to completely dominant resistance).
Another approach was used to assess dominance. For instance, D WNT value was calculated by: where W NTSS , W NTRS and W NTRR represent relative fitness in the absence of insecticide for susceptible homozygotes, heterozygotes, and resistant homozygotes, respectively 25 . When the D WNT value is 0.5, resistance is called co-dominant. D WNT values ranging from 0 to 0.5 demonstrate partial recessive, while D WNT values between 0.5 to 1 refer to partial dominance.
To test the genetic mode of inheritance, the expected mortality (E) of BC and F2 under a certain concentration of insecticide was estimated according to Georghiou's method 36 in which W 1 , W 2 , W 3 represent the actual mortality of Fuzhou-S, Fuzhou-R2Ad, and F1, respectively, for a given dose of insecticide. Chi-square test was employed to compare observed and expected mortality of BC and F2 37 . All of the above analyses, including one-way ANOVA with post-hoc Tukey's honestly significant difference, were performed by using data processing system (DPS) V9.01, while figures were developed using Prism Graphpad 6.

Results
Cry2Ad resistance ratio. The resistance to Cry2Ad developed slow, and increased 1.04 times at the 12 th generation as compared to the susceptible strain (Table 1). Resistance gradually increased over subsequent generations and by generation 37 a 8.70-fold increase was observed over the susceptible strain. In the 66 th generation, the relative resistance ratio was 120.59 (Table 1). Other fitness parameters, such as egg hatch rate, survival rate of the 2 nd -instar larvae, pupation rate, and female fecundity were significantly higher in F1 hybrid compared to F1' . And the relative fitness values of the positive cross F1 and negative cross F1' were 0.89 and 0.65, respectively.

Biological fitness parameters.
Inheritance properties. All experimental strains proved susceptible to Cry2Ad, and no significant difference was recorded in LC 50 values between F1 and F1' strains (Table 3). In the pooled hybrid (R × S), the LC 50 value was significantly lower than that of Fuzhou-R2Ad strain. Also, the overlap in 95% confidence limits of LC 50 between F1 and F1' strains confirmed that Cry2Ad resistance was autosomally inherited, without maternal effects and sex linkage.  (Table 4). Based on the relative DBM fitness (Table 2), the respective fitness values of F1 and F1' in insecticide-treated areas D WNT were 0.15 and 0.49. Hence, D, D LC , and D WNT parameters indicate that the genes conferring resistance to Cry2Ad in DBM selected strain was incompletely recessive. However, when subject to Cry2Ad at 25.32-202.60 μg/ml, DBM larvae had relatively high h values (0.56-0.71), suggesting an incomplete dominant inheritance of the Cry2Ad resistance.

Genetic mode of inheritance. LD-P lines and expected values were distinguishable for both BC and F2
crosses (Figs 1 and 2). A plateau was not reached neither after the 50% mortality of BC progeny nor at 25% or 75% mortality levels of F2 hybrids. Chi-square analysis showed that the resistance heredity in experimental DBM strains may be controlled by multiple genes (Tables 5 and 6).

Discussion
A thorough understanding of pesticide resistance development in P. xylostella is crucial for an effective and sustainable management of this globally-important pest. Past research has shown that the development of Bt resistance depends on the particular Bt strain and the type of Bt toxin 38 . Induced by Bt subspecies kurstaki, the resistance ratio of P. xylostella strain NO was 30 times 3 . Another P. xylostella strain NO-95 selected with high resistance to Bt subspecies kurstaki has very low resistance to Bt subspecies aizawai 5 . In 2014, a Cry1Ie susceptible Ostrinia furnacalis strain of ACB-BtS was found to have cross resistance to Cry1Ab, Cry1Ac and Cry1F toxins 39 . Other work has shown that a given Bt toxin produced by the same Bt species may exhibit different impacts on a DBM strains/populations, due to the differential modes of action of the Bt toxins 40,41 . In this study, we determine that DBM resistance development to the Bt Cry2Ad toxin is possible, after laboratory-based screening for 5 years  Table 2. Population growth parameters of different P. xylostella strains. According to one-way with post-hoc Tukey's honestly significant difference, the same superscript letter following the numbers between rows of a given column indicates no significant difference between the strains at P > 0.05. The different upper and lower case letters stand for the significance with P < 0.01, and P < 0.05, respectively. *Relative fitness of the susceptible Fuzhou-S strain is defined as 1.
www.nature.com/scientificreports www.nature.com/scientificreports/ and 66 generations. The resulting Fuzhou-R2Ad resistant strain had 120.59 times higher levels of resistance than the susceptible Fuzhou-S strain.
When unexposed to Bt Cry2Ad toxin, the Fuzhou-R2Ad has significantly lower fitness as compared to the susceptible strain. Similar findings has been made with DBM populations in Hawaii, where Dipel 2X ® (a wettable powder formulation of B. thuringiensis subsp. kurstaki strain HD-l) resistant strain NO-QA exhibited reduced survival, egg hatching and mating rates 42 . Such reduction in fitness is possibly related to induced genetic changes   www.nature.com/scientificreports www.nature.com/scientificreports/ to Bt toxins, which may remain even in the absence of selection pressure 24,43 . Hence, it is possible that effective DBM pest control can still be attained for resistant populations by discontinuing Bt Cry2Ad applications.
Inheritance of Bt resistance in the diamondback moth is considered to occur autosomally 14,28,44 , and similar inheritance models have been recorded for the Asian corn borer Ostrinia furnacalis 38 , the southern house mosquito Culex quinquefasciatus 45 , and the cotton bollworm Helicoverpa armigera 46,47 . As one notable exception, Malaysian populations of P. xylostella exhibited maternal effects on Cry1Ac resistance development 3 . In the current research, we detect susceptibility to Cry2Ad in all experimental strains or crosses, and confirm this to be autosomal resistance to Cry2Ad, without maternal effects or sex linkage (Table 3).
Our work also show that the resistance inheritance to Cry2Ad toxin in DBM strains is incompletely recessive. This is clearly shown by the following parameters: D F1 values of 0.73 and 0.13, D LC values of 0.44 and 0.28, D WNT values of 0.15 and 0.49 for F1 and F1' respectively. D, D LC and D WNT values indicate that resistance to Cry2Ad in the Fuzhou strains of P. xylostella is partially recessive. Secondly, the effective dominance is negatively regulated by concentrations of the Bt toxin 48,49 , namely an incomplete recessivity of resistance at a high Cry2Ad level and an incomplete dominance at low concentrations of Cry2Ad protein. However, when DBM populations are treated with a low dose of toxin, the reduced selection pressure may cause bias because of the increased survival rate in the susceptible strain.
Our work constitutes the first report of Cry2Ad resistance in P. xylostella, sheds light upon Bt resistance development, and could guide further pest management interventions against a globally-relevant lepidopteran pest. Caution needs to be taken when extrapolating our findings, as our research is conducted under highly-artificial conditions with laboratory-reared individuals. Hence, one could still encounter an incompletely coincident resistance to Cry2Ad due to variations in DBM field populations 50 . Further, we postulate that resistant heredity in local    www.nature.com/scientificreports www.nature.com/scientificreports/ diamondback moth populations is conferred by multiple genes (Figs 1 and 2; Table 5). All of the above provide fundamental insights into the mechanism and evolution of Bt resistance, according to the neo-Darwinian theory 51 . Further investigation of Bt resistance genes through molecular biology approaches, including molecular marker selection, would be a great help for the genetic manipulation of the diamondback moth. Moreover, the knowledge obtained from this research could boost the effectiveness of pest management interventions and enable sustainable DBM control globally.