Downregulation of APN1 and ABCC2 mutation in Bt Cry1Ac-resistant Trichoplusia ni are genetically independent

ABSTRACT

The resistance to the insecticidal protein Cry1Ac from the bacterium Bacillus thuringiensis (Bt) in the cabbage looper, Trichoplusia ni, has previously been identified to be associated with a frameshift mutation in the ABC transporter ABCC2 gene and with altered expression of the aminopeptidase N (APN) genes APN1 and APN6, shown as missing of the 110-kDa APN1 (phenotype APN1¯) in larval midgut brush border membrane vesicles (BBMV). In this study, genetic linkage analysis identified that the APN1¯ phenotype and the ABCC2 mutation in Cry1Ac-resistant T. ni segregated independently, although they were always associated under Cry1Ac selection. The ABCC2 mutation and APN1¯ phenotype were separated into two T. ni strains respectively. Bioassays of the T. ni strains with Cry1Ac determined that the T. ni with the APN1¯ phenotype showed a low level resistance to Cry1Ac (3.5-fold), and the associated resistance is incompletely dominant in the background of the ABCC2 mutation. Whereas the ABCC2 mutation-associated resistance to Cry1Ac is at a moderate level, and the resistance is incompletely recessive in the genetic background of downregulated APN1. Analysis of Cry1Ac binding to larval midgut BBMV indicated that the midgut in larvae with the APN1¯ phenotype had reduced binding affinity for Cry1Ac, but the number of binding sites remained unchanged, and the midgut in larvae with the ABCC2 mutation had both reduced binding affinity and reduced number of binding sites for Cry1Ac. The reduced Cry1Ac binding to BBMV from larvae with the ABCC2 mutation or APN1¯ phenotype correlated with the lower levels of resistance.

IMPORTANCE

The soil bacterium Bacillus thuringiensis (Bt) is an important insect pathogen used as a bioinsecticide for pest control. Bt genes coding for insecticidal proteins are the primary transgenes engineered into transgenic crops (Bt crops) to confer insect resistance. However, the evolution of resistance to Bt proteins in insect populations in response to exposure to Bt threatens the sustainable application of Bt biotechnology. Cry1Ac is a major insecticidal toxin utilized for insect control. Genetic mechanisms of insect resistance to Cry1Ac are complex and require to be better understood. The resistance to Cry1Ac in Trichoplusia ni is associated with a mutation in the ABCC2 gene and also associated with the APN expression phenotype APN1¯. This study identified the genetic independence of the APN1¯ phenotype from the ABCC2 mutation and isolated and analyzed the ABCC2 mutation-associated and APN1¯ phenotype-associated resistance traits in T. ni to provide new insights into the genetic mechanisms of Cry1Ac resistance in insects.

INTRODUCTION

Genetic engineering of crops to express insecticidal proteins from the soil bacterium Bacillus thuringiensis (Bt) has revolutionized pest management practices in crop protection and reduced reliance on chemical insecticide sprays (1–3). Since the commercialization of transgenic Bt crops in 1996, adoption of Bt crops has contributed significant economic and environmental benefits (3–6). The widespread adoption of Bt crops has meanwhile imposed selection pressure for insect resistance to Bt toxins, which challenges the durability of Bt crops for insect control.

Genetic and molecular studies on insect resistance to Bt have advanced our knowledge on the complex mode of action and the mechanisms of resistance to Bt toxins in insects, which involve multiple but not well-understood midgut proteins and mutations of their genes in the intoxication pathways (7–9). Bt resistance is commonly caused by reduced binding of Bt toxins to the midgut receptors in insects (9–12). A better understanding of the molecular mode of action of Bt toxins and mechanisms of resistance to Bt toxins is urgently needed for development of resistance management strategies (3).

Cry toxins are a large family of insecticidal toxins produced in various Bt strains and are the primary Bt toxins utilized for insect pest control (3, 13). Cry1Ac is a major Bt toxin for Lepidoptera pest control in Bt sprays and Bt crops, and studies on the mode of action of Cry toxins have been mostly on the group of Cry1A toxins (12, 14). The intoxication pathways of Cry toxins in insects involve a complex cascade of toxin–midgut protein interactions in the sequential binding model of Cry toxins, so alterations of any steps in the pathway can potentially lead to Bt resistance (7, 11, 12). Midgut proteins that have been identified or suggested to function as Cry receptors include GPI-anchored aminopeptidases N (APNs) and alkaline phosphatases (ALP), the midgut cadherin-like protein, and ABC transporters (11, 15–24). For Cry1Ac resistance, mutations or changes in the expression of Cry toxin receptor genes have been associated with high-level resistance in lepidopteran pests (12, 25–30).

The cabbage looper, Trichoplusia ni, developed resistance to Bt sprays in commercial greenhouses (31). From the greenhouse-derived resistant T. ni populations, T. ni strains resistant to Cry1Ac, Cry2Ab, and Cry1F have been isolated, respectively (32–34). The resistance to Cry1Ac in T. ni has been identified to result from loss of toxin-binding sites in the larval midgut (33). Further molecular and genetic analyses have revealed that resistance to Cry1Ac in T. ni is associated with altered APN expression (35) and a frameshift mutation in the ABCC2 gene (36, 37). The resistance conferred by the ABCC2 gene mutation, which leads to premature termination of the protein synthesis, has been identified, but knockout of the ABCC2 gene alone in susceptible T. ni only caused resistance to Cry1Ac at levels significantly lower than the level of resistance in the resistant strain from greenhouses (37). The high-level resistance to Cry1Ac in T. ni involves an additional mechanism showing as an incompletely dominant trait (37). Although it has been known that high-level resistance to Cry1Ac in T. ni is associated with the ABCC2 gene mutation and downregulation of APN1 expression, which is biochemically shown as a lack of the 110-kDa APN1 protein (APN1¯ phenotype) in the midgut brush border membrane, whether the ABCC2 mutation and the APN1¯ phenotype in the larval midgut are genetically associated and the role played by APN1 in Cry1Ac-resistance evolved in the greenhouse T. ni populations remain unknown. In this study, the genetic association of the ABCC2 gene mutation and the APN1¯ phenotype in Cry1Ac-resistant T. ni was analyzed, the resistance traits associated with the ABCC2 mutation and associated with the APN1¯ phenotype were isolated in two separated T. ni strains, and the resistance to Cry1Ac associated with the ABCC2 mutation and with the APN1¯ phenotype was examined.

RESULTS

The ABCC2 gene mutation and the APN1¯ phenotype in TnCry1Ac-R strain segregated independently

To examine the genetic association of the ABCC2 gene mutation with the APN1¯ phenotype in the Cry1Ac-resistant T. ni, an F₂ population was generated from the cross between the resistant TnCry1Ac-R strain and the susceptible Cornell strain. The ABCC2 gene (referred to as resistance gene R₁) in the F₂ individuals was genotyped for its two alleles s1 and r1 from the susceptible and resistant parents. The result of genotyping from 81 F₂ T. ni individuals showed that the genotypes in the F₂ generation followed the Mendelian inheritance with the genotype ratio of s1s1:s1r1:r1r1 = 1:2:1 (Table 1). The presence or absence of the 110-kDa APN1 protein in larval midgut brush border membrane vesicles (BBMV) from the 81 F₂ larvae was determined as phenotype APN1⁺ or APN1¯, respectively. The gene controlling this APN1 phenotype was referred to as resistance gene R2, and the alleles from the susceptible and resistant parents were s2 and r2, respectively. As APN1 is expressed in both the homozygous susceptible allele genotype and heterozygous genotype (35), the presence of the 110-kDa APN1 (APN1⁺) represents the genotype s2s2 or s2r2, whereas absence of the 110-kDa APN1 (APN1¯) indicates the genotype r2r2. The numbers of APN1⁺ and APN1¯ individuals determined from the 81 F₂ larvae were 59 and 22, respectively, which is statistically consistent with the predicted ratio of 3:1, i.e. (s2s2 +s2 r2) : r2r2 = 3:1. Therefore, the result from phenotyping of the 81 F₂ individuals indicated that the s2 and r2 alleles also followed the Mendelian inheritance (Table 1).

TABLE 1.

Analysis of the genetic association of APN1 expression with the ABCC2 gene mutation in T. ni

Frequency of ABCC2 genotypes and frequency of APN1 phenotypes in F2
Gene/phenotype	ABCC2			APN1 phenotype
	(alleles: s1 and r1)			APN1+			APN1-
Genotypea	s1s1	s1r1	r1r1	s2s2 and s2r2			r2r2
Observed	19	44	18	59			22
Expected	20	41	20	61			20
P valueb	0.79			0.61

^a The alleles s1 and r1 are the wild-type and mutant alleles of ABCC2, respectively. The alleles s2 and r2 represent the alleles associated with the APN1 phenotype APN1⁺ (s2) and APN1¯ (r2), respectively.

^b P values were calculated by χ² tests.

Analysis of the inheritance of these two genes in this dihybrid F₂ population showed that the R1 and R2 followed Mendelian independent assortment. The observed numbers of APN1⁺ individuals (genotypes = s2 s2 and s2r2) with the three genotypes of the R1 gene, as well as the APN1¯ individuals (genotype = r2 r2) with the three genotypes of R1 gene, were all statistically consistent with the prediction from the Mendelian independent assortment for two independent genes (Table 1). Therefore, the genetic linkage analysis confirmed that the ABCC2 mutation and the expression of the 110-kDa APN1 are genetically independent in T. ni.

The ABCC2 mutation and APN1¯ phenotype in the TnCry1Ac-R strain were isolated in two separate T. ni strains

To separate the APN1¯ phenotype and the ABCC2 mutation (r1) in two strains, single-pair families were generated by crossing between the TnCry1Ac-R strain and the Cornell strain for selection of the mutant ABCC2 genotype (r1r1) and the APN1¯ phenotype, respectively, from the single-pair segregating populations from and after the F₂ generations (Fig. 1). By selections of single-pair families having the highest frequency of the APN1¯ phenotype with the Cornell strain ABCC2 alleles (s1s1) from the segregating populations and also of the single-pair families having the highest frequency of the APN1⁺ phenotype with the TnCry1Ac-R ABCC2 alleles (r1r1) for four consecutive generations (Fig. 1), two new T. ni strains were obtained. A strain having the homozygous ABCC2 gene mutation (genotype r1r1) from the TnCry1Ac-R but having normal APN1 expression as in the Cornell strain (APN1⁺ phenotype) was named TnCry1Ac-R1. The other strain having the ABCC2 allele from the Cornell strain (genotype s1s1) but lacking the 110-kDa APN1 in the larval midgut BBMV as the TnCry1Ac-R strain (APN1¯ phenotype) was named TnCry1Ac-R2. The ABCC2 alleles in these T. ni strains were confirmed to be the homozygous resistant allele in TnCry1Ac-R1 and to be the homozygous susceptible allele in TnCry1Ac-R2 by polymerase chain reaction (PCR) analysis and DNA sequencing of the PCR fragments from the mutation site in the ABCC2 gene (GenBank acc. nos. MZ571761 and MZ571762). Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) analysis of the larval midgut BBMV proteins confirmed that larvae of the TnCry1Ac-R2 strain lacked the 110-kDa APN1 protein, but larvae of the TnCry1Ac-R1 strain had a normal level of the 110-kDa APN1 in the midgut BBMV (Fig. 2).

Fig 1.

Isolation of the ABCC2 mutation and downregulated APN1 from the TnCry1Ac-R strain into two separate T. ni strains, TnCry1Ac-R1 and TnCry1Ac-R2. An F₂ population from the cross of the TnCry1Ac-R strain with Cornell strain was generated, and homozygous ABCC2⁺ (s1s1) and homozygous population ABCC2^- (r1r1) individuals were selected to generate multiple single-pair families for the s1s1 and r1r1 groups, respectively. The progeny larvae were sampled to phenotype the APN1 in midgut BBMV, and the family that had the highest frequency of the APN1⁺ phenotype (s2s2 or s2r2) in the r1r1 group (left) and the family that had the highest frequency of the APN1¯ phenotype (r2r2) in the s1s1 group (right) were selected. The selected families were used to generate multiple single-pair families again, followed by APN1 expression phenotyping. This selection scheme of single-pair cross families by APN1 phenotyping was repeated till a homozygous ABCC2⁺/APN1¯ (s1s1r2r2) strain and a homozygous ABCC2^-/APN1 +strain (r1r1s2s2) were obtained.

Fig 2.

SDS-PAGE analysis of proteins of midgut BBMV prepared from individual larvae of the T. ni strains TnCry1Ac-R (shown in lanes 1 and 2), TnCry1Ac-R2 (lanes 3 and 4), Cornell (lanes 5 and 6), and TnCry1Ac-R1 (lanes 7 and 8). The rectangles of dotted lines indicate the gel regions with BBMV proteins from 85 to 120 kDa that were exercised from the gel for proteomic analysis. The asterisk (*) indicates the 110-kDa APN1 protein band.

Confirmation of the altered expression of APN1 and APN6 in TnCry1Ac-R2 larvae by LC-MS/MS

Larval midgut BBMV proteins having molecular weights from 85 to 120 kDa were collected by excising the gel slices from the SDS-PAGE gel from two individual larvae from each of the Cornell, TnCry1Ac-R, TnCry1Ac-R1, and TnCry1Ac-R2 strains (Fig. 2). Proteomic analysis of the proteins in the gel slices identified 70–118 proteins from the individual larval samples, with at least two specific peptides identified and with quantitative abundance data. The APN1 protein was identified only from the Cornell and the TnCry1Ac-R1 strains, but not in the TnCry1Ac-R and TnCry1Ac-R2 strains (Table 2). In contrast, APN6 was identified only in the TnCry1Ac-R and TnCry1Ac-R2 strains, but not in the Cornell strain and the TnCry1Ac-R1 strain (Table 2). The APN1 protein accounted for 24% and 27% of the proteins in the gel slices from two larvae of the Cornell strain and accounted for 34% and 32% of the proteins in the gel slices from two larvae of the TnCry1Ac-R1 strain. The APN6 protein accounted for 3.5% and 3.1% of the proteins in the gel slices from two larvae of the TnCry1Ac-R strain and 3.6% and 3.8% in the gel slices from two larvae of the TnCry1Ac-R2 strain (Table 2).

TABLE 2.

Relative abundance of APN1 and APN6 detected in midgut BBMV proteins from 85 to 120 kDa in SDS-PAGE gel

T. ni strain	Replicate	APN1	APN6
Cornell	1	0.24	Undetected
	2	0.27	Undetected
TnCry1Ac-R	1	Undetected	0.035
	2	Undetected	0.031
TnCry1Ac-R1	1	0.34	Undetected
	2	0.32	Undetected
TnCry1Ac-R2	1	Undetected	0.036
	2	Undetected	0.038

Cry1Ac resistance in larvae of the TnCry1Ac-R1 and TnCry1Ac-R2 strains

The resistance to Cry1Ac associated with the ABCC2 mutation and the resistance with the APN1¯ phenotype were examined by bioassays of Cry1Ac toxicity in the TnCry1Ac-R1 and TnCry1Ac-R2 strains, respectively, in comparison with their original TnCry1Ac-R strain (Table 3). Resistance to Cry1Ac was observed in both larvae of the TnCry1Ac-R1 and TnCry1Ac-R2 strains. However, larvae from the TnCry1Ac-R1 strain showed a moderate level of resistance, with the resistance ratio being 91.5-fold, significantly lower than the resistance level of the parent resistant strain TnCry1Ac-R (826-fold). The resistance level in the TnCry1Ac-R2 strain was low, with the resistance ratio being only 3.5-fold.

TABLE 3.

Bioassays of Bt Cry1Ac toxicity in larvae of susceptible and resistant T. ni strains and F₁ families

Strain	Slope (SE)	LC50 (µg/mL) (95% C.I.)	RR	Da
Cornell (s1s1s2s2)	2.98 (0.28)	0.19 (0.15–0.24)	1
TnCry1Ac-R (r1r1r2r2)	2.65 (0.30)	157 (128–194)	826
TnCry1Ac-R1 (r1r1s2s2)	2.48 (0.26)	17 (12 to 26)	91.5
TnCry1Ac-R2 (s1s1r2r2)	3.57 (0.42)	0.67 (0.56–0.81)	3.5
F1 (♀︎TnCry1Ac-R x ♂︎TnCry1Ac-R2) (r1s1r2r2)	3.59 (0.42)	3.7 (2.3–6.5)	19.6	- 0.37
F1 (♂︎TnCry1Ac-R x ♀︎TnCry1Ac-R2) (r1s1r2r2)	4.15 (0.53)	3.7 (3.1–4.3)	19.3	- 0.38
F1 (♀︎TnCry1Ac-R x ♂︎TnCry1Ac-R1) (r1r1r2s2)	2.66 (0.28)	67 (44–104)	354	0.23
F1 (♀︎TnCry1Ac-R x ♂︎TnCry1Ac-R1) (r1r1r2s2)	2.65 (0.29)	80 (65–99)	421	0.39
F1 (♂︎TnCry1Ac-R x ♀︎TnCry1Ac-R1) (r1r1r2s2)	2.01 (0.19)	54 (45–67)	284	0.03
F1 (♂︎TnCry1Ac-R x ♀︎TnCry1Ac-R1) (r1r1r2s2)	2.31 (0.24)	117 (84–167)	617	0.73

^a D (degree of dominance) was calculated using the Stone formula (38).

Dominance of resistance to Cry1Ac associated with the ABCC2 mutation and the APN1¯ phenotype in T. ni

The dominance of Cry1Ac resistance by the ABCC2 mutation in the APN1¯ phenotype background was analyzed by bioassays of F₁ families from reciprocal crosses between TnCry1Ac-R and TnCry1Ac-R2, together with bioassays in the TnCry1Ac-R and TnCry1Ac-R2 strains. The dominance of resistance associated with the APN1¯ phenotype in the ABCC2 mutation background was analyzed by bioassays of the F₁ family from TnCry1Ac-R x TnCry1Ac-R1, together with the bioassays in TnCry1Ac-R and TnCry1Ac-R1 strains (Table 3). The LC₅₀s of Cry1Ac in the two F₁ families from ♀︎TnCry1Ac-R x ♂︎TnCry1Ac-R2 and ♂︎TnCry1Ac-R x ♀︎TnCry1Ac-R2 were both 3.7 µg/mL, and the LC₅₀s in the TnCry1Ac-R and TnCry1Ac-R2 strains were 157 and 0.67 µg/mL, respectively. The degree of dominance (D) of the ABCC2 mutation-associated Cry1Ac resistance was calculated in the two F₁ families from reciprocal crosses to be −0.37 and −0.38, respectively, indicating incompletely recessive inheritance of the ABCC2 mutation-associated resistance in the normal APN1⁺ phenotype background without maternal effect. The LC₅₀s of Cry1Ac in two F₁ families from ♀︎TnCry1Ac-R x ♂︎TnCry1Ac-R1 were 67 and 80 µg/mL, respectively, and in two F₁ families from ♂︎TnCry1Ac-R x ♀︎TnCry1Ac-R1 were 54 and 117 µg/mL, respectively. The LC₅₀s in the TnCry1Ac-R and TnCry1Ac-R1 strains were 157 and 17 µg/mL, respectively. Therefore, the degrees of dominance (D) of the APN1¯-associated Cry1Ac resistance calculated from the two F₁ families of ♀︎TnCry1Ac-R x ♂︎TnCry1Ac-R1 were 0.23 and 0.39, respectively, and the degrees of dominance calculated from the other two F₁ families of ♂︎TnCry1Ac-R x ♀︎TnCry1Ac-R1 were 0.03 and 0.73, respectively, showing incompletely dominant inheritance of the APN1¯ phenotype-associated Cry1Ac resistance in the mutant ABCC2 background without maternal effect (Table 3).

Binding of Cry1Ac to midgut BBMV from the TnCry1Ac-R1 and TnCry1Ac-R2 strains

Assays for binding of ¹²⁵I-labeled Cry1Ac to the midgut BBMV preparations from the TnCry1Ac-R1 and TnCry1Ac-R2 strains showed the presence of specific binding of Cry1Ac to the BBMV from the two strains, in contrast to the binding assay with the BBMV from the original TnCry1Ac-R strain, which exhibited no specific binding for Cry1Ac (Fig. 3). With the BBMV preparations from the Cornell, TnCry1Ac-R1 and TnCry1Ac-R2 strains, the amount of ¹²⁵I-Cry1Ac bound to the BBMV increased with the increasing concentrations of BBMV in the binding suspension (Fig. 3). The amount of ¹²⁵I-Cry1Ac that bound to the BBMV from the TnCry1Ac-R2 strain was similar to that from the Cornell strain (Fig. 3A and B). However, binding of Cry1Ac to the BBMV from the TnCry1Ac-R1 strain was significantly lower than that from the Cornell strain (Fig. 3C). With the BBMV at 0.4 mg/mL, 43% of the Cry1Ac was bound to the BBMV from the Cornell strain larvae, but only 26% of the Cry1Ac was bound to the BBMV from the TnCry1Ac-R1 strain under the same binding condition.

Fig 3.

Binding of ¹²⁵I-Cry1Ac (0.75 nM) at increasing concentrations of midgut BBMV from the control Cornell strain (panel A), TnCry1Ac-R2 (panel B), TnCry1Ac-R1 (panel C), and TnCry1Ac-R (panel D). ● Total binding of ¹²⁵I-Cry1Ac to BBMV. ■ Nonspecific binding of ¹²⁵I-Cry1Ac to BBMV determined in the presence of 200-fold excess of unlabeled Cry1Ac. Error bars represent standard errors of the mean from two replicates. The difference between the total binding and the nonspecific binding indicates specific binding.

Further quantitative analysis of Cry1Ac binding to the BBMV by homologous competition binding assays (Fig. 4) allowed to determine the equilibrium dissociation constants (K_d) and concentration of specific binding sites (R_t) for the binding of Cry1Ac to the BBMV. The K_d values for the Cornell, TnCry1Ac-R2, and TnCry1Ac-R1 strains were 2.20 ± 0.48 nM (mean ± SD), 3.39 ± 0.51 nM, and 3.37 ± 0.94 nM, respectively. Therefore, the binding affinity of Cry1Ac to the BBMV from the TnCry1Ac-R2 and TnCry1Ac-R1 strains was slightly reduced, compared to the binding to the BBMV from the Cornell strain. The R_t values for the Cornell, TnCry1Ac-R2 and TnCry1Ac-R1 strains were determined to be 3.34 ± 0.60 pmol/mg (mean ± SD), 3.22 ± 0.42 pmol/mg and 1.25 ± 0.29 pmol/mg, respectively. The results showed that the concentration of binding sites for Cry1Ac on the BBMV from the TnCry1Ac-R2 strain was similar to that of the susceptible Cornell strain, but the number of binding sites on the BBMV from the TnCry1Ac-R1 strain was significantly reduced in comparison with that of the Cornell strain.

Fig 4.

Competitive binding of ¹²⁵I-Cry1Ac with increasing concentrations of unlabeled Cry1Ac as the competitor to T. ni larval BBMV to determine the equilibrium dissociation constants (K_d) and concentration of specific binding sites (R_t) for the binding of Cry1Ac to the midgut BBMV from the susceptible control strain (A), TnCry1Ac-R2 (B), and TnCry1Ac-R1 (C). The binding affinity of Cry1Ac to the BBMV from the TnCry1Ac-R2 and TnCry1Ac-R1 strains was slightly reduced, while the number of binding sites was reduced in BBMV from the TnCry1Ac-R1strain, in comparison with the Cornell strain.

Formation of oligomers of Cry1Ac after binding to midgut BBMV from the TnCry1Ac-R1 and TnCry1Ac-R2 strains

Analysis of ¹²⁵I-Cry1Ac bound to the T. ni BBMV by SDS-PAGE showed that the activated monomeric Cry1Ac became oligomeric after binding to the BBMV from the Cornell strain, as well as the BBMV from the TnCry1Ac-R2 and TnCry1Ac-R1 strains (Fig. 5). The activated Cry1Ac in solution was detected as a monomeric 65-kDa protein, but once specifically bound to the BBMV, regardless of the larval source of the 3 T. ni strains examined, the Cry1Ac protein was detected to be oligomeric with a molecular weight of about 250 kDa (Fig. 5).

Fig 5.

Autoradiography after binding assays of ¹²⁵I-Cry1Ac to T. ni larval midgut BBMV. Assays were performed using midgut BBMV from the Cornell strain (A), TnCry1Ac-R2 (B), and TnCry1Ac-R1 (C). Samples shown in lanes 1 were ¹²⁵I-Cry1Ac in solution used in binding assays. BBMV were incubated with ¹²⁵I-Cry1Ac in the absence (lanes 2, 4 and 6) or the presence (lanes 3, 5 and 7) of a 200-fold excess of non-labeled Cry1Ac for 1 hour, and pellets recovered by centrifugation were analyzed by 12% SDS-PAGE. Concentrations of BBMV in the binding mixtures were 0.1 (lanes 2 and 3), 0.3 (lanes 4 and 5) and 0.4 mg/mL (lanes 6 and 7). * indicates the ¹²⁵I-Cry1Ac in the monomer form. ▶ indicates the ¹²⁵I-Cry1Ac in the oligomer form.

DISCUSSION

The resistance to Cry1Ac in T. ni evolved in greenhouse populations is associated with the ABCC2 gene and downregulation of APN1 shown as the absence of the 110-kDa APN1 protein in the larval midgut (35, 36). The resistance-conferring mutation in the ABCC2 gene has been identified to be a 4-bp insertion in an exon, which leads to disruption of ABCC2 synthesis (37). However, knockout of ABCC2 in the susceptible T. ni strain only caused a low to moderate level of resistance to Cry1Ac, indicating the presence of additional mutations for the resistance (37). How the downregulation of APN1 is involved in the resistance remains to be understood. Genetic linkage analysis of the ABCC2 mutation and the APN1¯ phenotype in this study confirmed that the downregulation of APN1 and the ABCC2 gene mutation are not genetically associated (Table 1). Therefore, the tight association of the ABCC2 mutation and downregulated APN1 in Cry1Ac resistance in T. ni, which resulted from selection by Bt, is due to their functional roles required for the high-level resistance to Cry1Ac, rather than a genetic association between the two.

In this study, the ABCC2 mutation and APN1¯ phenotype from the TnCry1Ac-R strain were isolated into the TnCry1Ac-R1 and TnCry1Ac-R2 strains. This made it possible to examine the roles of the ABCC2 mutation-associated and APN1¯ phenotype-associated Cry1Ac resistance separately. Proteomic analysis of the larval midgut BBMV proteins confirmed the previous observation on the differed expression of APN1 and APN6 in the Cry1Ac-resistant TnCry1Ac-R strain from the Cornell strain (35). In the midgut BBMV proteins ranging from 85 to 120 kDa, the 110-kDa APN1 protein was present in the Cornell strain-originated allele but was lacking in the TnCry1Ac-R strain-originated allele, whereas the APN6 protein was detected in the TnCry1Ac-R strain-originated allele but was not detected in the Cornell strain-originated allele (Table 2).

Association of the ABCC2 mutation with resistance to Cry1Ac has been identified in several insect species (19, 36, 39, 40), and ABCC2 as a receptor for Cry1Ac has been functionally confirmed by overexpression of an ABCC2 in both cultured cells and in insects to introduce susceptibility to the toxin (41–43). However, knockout of ABCC2 by gene editing in lepidopteran larvae, including in T. ni larvae, normally causes only a low to moderate level of resistance to Cry1Ac (28, 30, 37). Therefore, it is important to understand the level of resistance conferred by ABCC2 mutations that are selected in Cry1Ac-resistant insect populations. However, ABCC2 alleles from Bt-resistant insect populations had not been isolated as a single trait to study their roles in resistance to Bt toxins prior to this study. In this study, the ABCC2 allele from the greenhouse-originated Cry1Ac-resistant T. ni strain was isolated into a new strain, and the Cry1Ac resistance conferred by the resistant ABCC2 allele was analyzed in the susceptible Cornell strain background and also in the APN1¯ background. Bioassay of Cry1Ac toxicity in the TnCry1Ac-R1 strain showed that the ABCC2 mutant allele-associated resistance to Cry1Ac is at a moderate level (RR = 91.5), significantly lower than the resistance level in the original TnCry1Ac-R strain (RR = 826) (Table 3). This observation is similar to the previous observation of low to moderate levels of resistance to Cry1Ac in T. ni larvae, with various mutations introduced in ABCC2 by CRISPR/Cas9 mutagenesis (37), and confirms that an additional mutation or mutations are involved in the high-level resistance.

Previous studies on the function of ABCC2 as a Cry1Ac receptor in cell culture have indicated that co-expression of the midgut cadherin together with ABCC2 drastically increases the cell sensitivity to Cry1Ac (42, 44), suggesting that complete knockout of the susceptibility of insects to Cry1Ac may require mutations in both ABCC2 and the cadherin gene. Consistent with the studies in cell culture, mutations in both the midgut cadherin and the ABCC2 genes have been identified in an H. virescens strain that is highly resistant to Cry1Ac (18, 19). Similarly, studies on ABCC2-associated Cry1Ac resistance by introducing knockout mutations in H. armigera and P. xylostella showed that knockout of ABCC2 or ABCC3, respectively, only resulted in a minimally low level of resistance, but knockout of ABCC2 and ABCC3 together was sufficient to confer high-level resistance to Cry1Ac (28, 30). These studies indicated that high-level resistance could be conferred by a mutation in ABCC3, in addition to ABCC2. Mutations in both ABCC2 and ABCC3 to confer high-level resistance to Cry1Ac have been identified in P. xylostella (28), whereas the Cry1Ac resistance in T. ni originated from a Bt-resistant greenhouse population is neither associated with the midgut cadherin (36, 45) nor with the ABCC1 and ABCC3 genes (37), but with downregulation of APN1 in addition to the mutation in ABCC2. The bioassay of Cry1Ac toxicity in TnCry1Ac-R2 larvae showed that Cry1Ac resistance with downregulation of APN1 isolated from TnCry1Ac-R strain is at a low level (RR = 3.5, Table 3). The isolation of the ABCC2 mutation and downregulated APN1 in separate T. ni strains enabled us to analyze the inheritance of the ABCC2 mutation-associated resistance in the background of the APN1¯ phenotype and the APN1¯ phenotype-associated resistance in the background of the ABCC2 mutation. The inheritance of the resistance associated with the ABCC2 mutation in the genetic background of the APN1¯ phenotype was incompletely recessive (D = −0.37 and D = −0.38 for the F₁s from the reciprocal crosses, Table 3). This is consistent with the previous finding that Cry1Ac resistance in TnCry1Ac-R is incompletely recessive (46). However, the APN1¯ phenotype-associated resistance trait showed incomplete dominance in the genetic background of the ABCC2 mutation (D = 0.23 and 0.39 and D = 0.03 and 0.73 for the F₁s from the reciprocal crosses, Table 3). This result confirmed the recent report that there exists an additional Cry1Ac resistance trait or traits besides the ABCC2 mutation in the TnCry1Ac-R strain and the additional resistance trait is incompletely dominant in the ABCC2-knockout background (37). Therefore, in TnCry1Ac-R strain larvae, the ABCC2 mutation-associated resistance mechanism alone confers a moderate level of resistance to Cry1Ac and the APN1 downregulation-associated resistance mechanism alone confers a low level of resistance. The ABCC2 mutation-associated resistance trait is incompletely recessive, but the APN1 downregulation-associated resistance trait is incompletely dominant. How the ABCC2 mutation and the change of APN expression and possibly other factors result in a high-level resistance to Cry1Ac requires to be mechanistically understood.

Binding of a Bt toxin to its specific receptors in the insect midgut is an essential event in the pathway of toxicity (47). Previous studies have identified that the Cry1Ac-resistant T. ni larvae lack specific binding sites for Cry1Ac in the midgut (33). The isolation of the ABCC2 mutation-associated resistance and APN downregulation-associated resistance in separate strains allowed analysis of the binding of Cry1Ac to the midgut BBMV from larvae with the two resistance traits, respectively (Fig. 3 and 4). The results from binding analysis are consistent with bioassay data (Table 3). The highly Cry1Ac-resistant TnCry1Ac-R strain lacks specific binding sites for Cry1Ac, but the TnCry1Ac-R1 strain had a reduced concentration of specific binding sites and also reduced binding affinity for Cry1Ac in the larval midgut, which leads to a moderate level of resistance to Cry1Ac. The Cry1Ac-R2 strain showed reduced binding affinity in the midgut for Cry1Ac, but the binding site numbers in the midgut remained similar to the susceptible Cornell strain, which explains the low level of resistance to Cry1Ac in this strain.

Oligomer formation of Cry toxins by interaction with the midgut receptors is critical for the toxicity of Cry toxins in insects (7, 48, 49). The midgut BBMV from both TnCry1Ac-R1 larvae and TnCry1Ac-R2 larvae showed reduced binding to Cry1Ac, but the Cry1Ac bound to the BBMV from the two strains formed oligomers similarly to the oligomer formation observed with the BBMV from the susceptible Cornell strain (Fig. 5). Therefore, oligomerization of Cry1Ac by interaction with midgut receptors in T. ni can occur in the absence of ABCC2 or absence of the 110-kDa APN1 in the midgut cells. This observation is not surprising as oligomer formation of the toxin is commonly facilitated by interaction with the midgut cadherin protein in the sequential binding model (7, 11, 12). Our previous study has also shown that knockout of the cadherin gene in T. ni does not result in a decrease in larval susceptibility to Cry1Ac (25). However, oligomerization of Cry1Ac in cadherin-knockout T. ni larvae has not been examined. What midgut proteins in T. ni facilitate the oligomerization of Cry1Ac remain to be understood. There could be an alternate mechanism present in insects, in addition to the cadherin (11). Nevertheless, the lack of ABCC2 or APN1 in larvae affects the binding of Cry1Ac to the midgut brush border, leading to resistance, but oligomer formation of the toxin is not disrupted. The greenhouse-originated high-level Cry1Ac resistance in T. ni required the ABCC2 mutation-associated resistance trait and the APN1 downregulation-associated resistance trait, and the two resistance traits were isolated and studied in this study. The results in this study further indicated the APN1-associated resistance to Cry1Ac as an incompletely dominant trait in T. ni from the greenhouse-evolved resistant populations. What gene or additional genes together with the ABCC2 mutation are involved in the molecular mechanism of Cry1Ac resistance requires to be understood.

MATERIALS AND METHODS

Insect strains

The inbred laboratory T. ni Cornell strain (46) was used as a Bt-susceptible strain in this study. The Cry1Ac-resistant T. ni strain TnCry1Ac-R used in this study was originated from a greenhouse Bt-resistant population (31) and has been backcrossed with the Cornell strain to introgress the resistance into the genetic background of the Cornell strain (33). The T. ni strains described above and generated in this study were maintained on an artificial diet as previously described (46).

Analysis of the genetic association of the ABCC2 gene mutation and APN1¯ phenotype in Cry1Ac-resistant T. ni

To analyze the genetic association of the ABCC2 gene mutation with the APN1¯ phenotype in the midgut in Cry1Ac-resistant T. ni, the TnCry1Ac-R strain was crossed with the susceptible Cornell strain, and the derived F₁ family was maintained on an artificial diet to obtain the F₂ population. The two ABCC2 gene alleles s1 and r1 from the susceptible Cornell strain and the resistant TnCry1Ac-R strain (GenBank nos.: MW595613.1 and MW595614.1) in F₂ individuals were determined by PCR analysis (37). Genomic DNA samples were prepared from the hemolymph individually collected from the proleg of the larvae (25), and the ABCC2 alleles were determined by PCR analysis using the primer set 5′-GCCCAAGATAGTCTGAAAGTAA-3′ and 5′-TACCAGCTTTTTAGTAGTCGTTAG-3′ to amplify a diagnostic 120-bp fragment for the s1 allele and a 102-bp fragment for the r1 allele, as previously reported (34). In total, 81 F₂ individuals were genotyped.

The phenotypes of APN1 expression shown as presence or absence of the 110-kDa APN1 in the larval midgut, the APN1⁺ phenotype and the APN1¯ phenotype, were examined by SDS-PAGE analysis of larval midgut BBMV proteins prepared from individual F₂ T. ni larvae (35). The same 81 mid-5^th instar larvae from the F₂ population used above for ABCC2 genotyping were dissected on ice to isolate the midgut tissue. BBMV was prepared from the individual midgut, and the presence or absence of the distinct 110-kDa APN1 in the BBMV proteins from individual larvae was examined by 7% SDS-PAGE analysis, as previously reported (34). The genetic association of the APN1 expression phenotypes with the ABCC2 mutation was analyzed based on Mendelian’s independent assortment of two independent genes—the ABCC2 gene (R1) and the gene for the downregulated APN1 (referred to as R2). It is known that the 110-kDa APN1 is highly expressed in homozygous susceptible T. ni (Cornell strain) and is also expressed at a reduced level in heterozygous F₁ larvae, but is not present in homozygous resistant T. ni (TnCry1Ac-R strain) (35). Therefore, positive detection of the 110-kDa APN1 in midgut BBMV indicates the homozygous susceptible allele genotype of R2 (s2s2) or the heterozygous genotype (s2r2). A negative detection of the 110-kDa APN1 indicates the homozygous resistant allele genotype (r2r2).

Separation of the ABCC2 mutation and the APN1¯ phenotype from the TnCry1Ac-R strain into two T. ni strains

With the finding from this study that the APN1¯ phenotype in the TnCry1Ac-R strain is genetically independent of the ABCC2 mutation, experiments to separate the APN1¯ phenotype and the ABCC2 mutation (r1) in two new strains were conducted by crossing the TnCry1Ac-R strain with the Cornell strain and selecting the ABCC2 mutant (r1r1) and the APN1¯ phenotype, respectively, from segregating populations in single-pair crosses from and after the F₂ generations (Fig. 1). First, T. ni individuals with homozygous ABCC2 wild-type (s1s1) and homozygous ABCC2 mutant (r1r1) were identified from the F₂ population by PCR analysis of larval hemolymph samples, as described above, and these individuals were used to prepare ABCC2⁺ (s1s1) single-pair families and ABCC2^- (r1r1) single-pair families (Fig. 1). Progeny larvae (6–9 individuals) from each single-pair family were analyzed to determine the APN1 expression phenotype in the midgut BBMV, as described above. The single-pair ABCC2^- (r1r1) family found to have the highest frequency of the APN1 +phenotype (ABCC2^-/APN1⁺) in the 6–9 larvae analyzed and the single-pair ABCC2⁺ (s1s1) family found to have the highest frequency of the APN1¯ phenotype (ABCC2⁺/APN1¯) were selected. The selected families were used to generate multiple single-pair families again, and the APN1 expression was again examined in progeny larvae from the individual single-pair families. This selection scheme of single-pair cross families by APN1 phenotyping was repeated for four consecutive generations till a stable homozygous ABCC2⁺/APN1¯ strain and a homozygous ABCC2^-/APN1 +strain were obtained. The ABCC2^-/APN1⁺ strain was named TnCry1Ac-R1, and ABCC2⁺/APN1¯ strain was named TnCry1Ac-R2, thereafter.

LC-MS/MS analysis of the APN1 and APN6 proteins in larval midgut BBMV from the TnCry1Ac-R1 and TnCry1Ac-R2 strains

Mid-fifth instar T. ni larvae from the newly generated strains TnCry1Ac-R1 and TnCry1Ac-R2, along with the parental strains Cornell and TnCry1Ac-R, were dissected on ice to separate the midgut, and midgut BBMV preparations from individual larvae were prepared as described above. Proteins of midgut BBMV from individual larvae were separated by 7.5% SDS-PAGE and stained with Coomassie brilliant blue R250.

To analyze the APN1 and APN6 proteins in the midgut BBMV preparations from the T. ni strains, SDS-PAGE gel slices containing BBMV proteins from individual larvae in the molecular weight range from 85 to 120 kDa were excised. The gel slices were processed by in-gel trypsin digestion, and the peptides derived were analyzed by nano LC-MS/MS on an Orbitrap Fusion Lumos Tribrid Mass Spectrometer coupled with the UltiMate 3000 RSLCnano System (Thermo Fisher Scientific) by service at the Proteomics Facility of Cornell University (Ithaca, NY). The MS/MS data were analyzed using the Proteome Discoverer 2.5 software package (Thermo Fisher Scientific) against the protein sequence database generated from the T. ni genome (http://www.tnibase.org) (50). The relative abundances of APN1 and APN6 in the samples were calculated from the abundance of total proteins identified in the gel samples by the Proteome Discoverer 2.5 software package.

Bioassays of susceptibility to Cry1Ac in larvae of the TnCry1Ac-R1 and TnCry1Ac-R2 strains

The Bt toxin surface overlay bioassay method on artificial diet (46) was used to examine the susceptibility to Bt Cry1Ac in neonates of the T. ni strains generated in this study. Briefly, a 200-µL aliquot of the Bt Cry1Ac solution from a twofold dilution series was spread on the surface of diet (5 mL diet with surface area ~7 cm²) in 30-mL cups, and 10 neonate larvae were placed in each cup with five replicate cups for each dose. Larval mortalities were recorded after 4 days, and the median lethal doses (LD₅₀s) of Cry1Ac in the T. ni strains were calculated by probit analysis using the software POLO (LeOra Software). The Bt toxin Cry1Ac used in this study was prepared from B. thuringiensis strain HD-73, using the same method as previously described (46).

Analysis of the dominance of Cry1Ac resistance associated with the ABCC2 mutation and the APN1 downregulation in T. ni

To examine the inheritance of Cry1Ac resistance associated with the gene for downregulation of APN1 (R2) in the background of the ABCC2 mutation (r1r1), the resistant strain TnCry1Ac-R (r1r1;r2r2) was reciprocally crossed with the TnCry1Ac-R1 (r1r1;s2s2) to generate F₁ families (r1r1;s2r2). Another set of F₁ families (s1r1;r2r2) were generated by reciprocally crossing the resistant strain TnCry1Ac-R (r1r1;r2r2) with the TnCry1Ac-R2 (s1s1;r2r2) to examine the inheritance of Cry1Ac resistance conferred by the ABCC2 mutation in the background of APN1 downregulation. The susceptibility of the F₁ families to Cry1Ac was determined, together with the susceptible Cornell strain (s1s1;s2s2), the resistant strain TnCry1Ac-R (r1r1;r2r2), and the new strains TnCry1Ac-R1 (r1r1;s2s2) and TnCry1Ac-R2 (s1s1;r2r2), using the bioassay procedures as described above. The dominance of Cry1Ac resistance associated with APN1 downregulation (R2) in the r1r1 background was analyzed using the bioassay results from the TnCry1Ac-R, TnCry1Ac-R1 and the F₁ families from TnCry1Ac-R x TnCry1Ac-R1. The dominance of R1 in Cry1Ac in the r2r2 background was analyzed using the bioassay results from the TnCry1Ac-R, TnCry1Ac-R2, and the F₁ families from TnCry1Ac-R x TnCry1Ac-R2. The degree of dominance (D) of the resistance was calculated using the Stone formula (38): D = [2 x (lg LC50 in rs) – (lg LC50 in rr) – (lg LC50 in ss)]/[(lg LC50 in rr) – (lg LC50 in ss)].

Analysis of Cry1Ac binding to midgut BBMV from the new T. ni strains with the ABCC2 mutation and APN1¯ phenotype

T. ni 5th instar larvae from the Cornell strain, TnCry1Ac-R, TnCry1Ac-R1, and TnCry1Ac-R2 were dissected on ice to isolate the midgut from other attached tissues, and the midgut tissue was cleaned in phosphate-buffered saline (PBS) and stored at −80°C after flash-freezing in liquid nitrogen. Approximately 12 grams of insect midgut tissue samples per strain were lyophilized and shipped from Cornell University to the University of Valencia. Midgut BBMV preparations were made from part of the lyophilized midgut samples using the differential magnesium precipitation method (51, 52) and quantified by measurement of protein contents using the Bradford method (53). To assess the BBMV preparation quality, the APN activities as an enzymatic marker in the initial homogenates and final BBMV preparations were analyzed to determine the enrichment of brush border membranes (54).

The Cry1Ac (25 µg) was labeled with 0.3 mCi of ¹²⁵I (PerkinElmer, Boston, MA) using the chloramine T method (55), and the specific activity obtained for the labeled proteins was 3.14 mCi/mg. PBS with 0.1% BSA was used as the binding buffer. To change the buffer in which BBMV were prepared and stored, prior to the binding assays, BBMV were centrifuged for 10 minutes at 16,000 × g at 4°C and resuspended in binding buffer. Cry1Ac binding assays were initially performed by incubation of increasing concentrations of BBMV with 0.75 nM of labeled Cry1Ac protein in a final volume of 100 µL for 1 hour at room temperature (RT), and nonspecific binding was determined by addition of 200-fold excess of unlabeled Cry1Ac protein (150 nM) in binding assays. Homologous competition binding assays were conducted by incubating 0.25 mg of BBMV with the labeled Cry1Ac protein and increasing amounts of the unlabeled protein in a final volume of 100 µL. After 1-hour incubation at RT, samples were centrifuged and washed with binding buffer, and the radioactivity in the final BBMV pellet was measured in a model 2480 WIZARD² gamma counter. Both binding and competition assays were performed twice for each strain. The equilibrium dissociation constant (K_d) and concentration of binding sites (R_t) were estimated from the homologous competition experiments using the LIGAND program (56). Values were defined as the mean ± standard deviation (SD).

To examine the oligomerization of Cry1Ac post binding to the midgut BBMV, the final BBMV pellets with ¹²⁵I-Cry1Ac bound from the binding assays above were resuspended in 10 µL of water, mixed with SDS-PAGE sample loading buffer, and then heated at 50°C for 5 minutes, immediately followed by 12% SDS-PAGE. To visualize the ¹²⁵I-Cry1Ac band images, the SDS-PAGE gels were dried, and radioactive Cry1Ac bands were imaged using a Fujifilm Imagine Plate screen with a Typhoon FLA-7000 system (GE Healthcare, USA).