Epistasis confers resistance to Bt toxin Cry1Ac in the cotton bollworm

Abstract

Evolution of resistance by insect pests reduces the benefits of extensively cultivated transgenic crops that produce insecticidal proteins from Bacillus thuringiensis (Bt). Previous work showed that resistance to Bt toxin Cry1Ac, which is produced by transgenic cotton, can be conferred by mutations disrupting a cadherin protein that binds this Bt toxin in the larval midgut. However, the potential for epistatic interactions between the cadherin gene and other genes has received little attention. Here, we report evidence of epistasis conferring resistance to Cry1Ac in the cotton bollworm, Helicoverpa armigera, one of the world's most devastating crop pests. Resistance to Cry1Ac in strain LF256 originated from a field-captured male and was autosomal, recessive, and 220-fold relative to susceptible strain SCD. We conducted complementation tests for allelism by crossing LF256 with a strain in which resistance to Cry1Ac is conferred by a recessive allele at the cadherin locus HaCad. The resulting F₁ offspring were resistant, suggesting that resistance to Cry1Ac in LF256 is also conferred by resistance alleles at this locus. However, the HaCad amino acid sequence in LF256 lacked insertions and deletions, and did not differ consistently between LF256 and a susceptible strain. In addition, most of the cadherin alleles in LF256 were not derived from the field-captured male. Moreover, Cry1Ac resistance was not genetically linked with the HaCad locus in LF256. Furthermore, LF256 and the susceptible strain were similar in levels of HaCad transcript, cadherin protein, and binding of Cry1Ac to cadherin. Overall, the results imply that epistasis between HaCad and an unknown second locus in LF256 yielded the observed resistance in the F₁ progeny from the complementation test. The observed epistasis has important implications for interpreting results of the F₁ screen used widely to monitor and analyze resistance, as well as the potential to accelerate evolution of resistance.

Keywords: Bacillus thuringiensis; F1 screen; allelism; cotton bollworm; genetically engineered crop; resistance management; second‐site noncomplementation; transgenic cotton.

Figure 1

Establishment of the resistant LF256 strain, which was started by crossing male #256 captured from Langfang, China, with a female from the resistant SCD‐r1 strain. Cadherin genotypes are shown, with r and s indicating resistant and susceptible alleles, respectively. Alleles r₁ and r₁₈ each have a premature stop codon. Based on the initial complementation test results, the second cadherin allele in the field‐captured male was tentatively named r_x (as shown above), to indicate its hypothesized role in resistance. However, analysis of resistant strain LF256 revealed that resistance in this strain was not conferred by alleles at the cadherin locus (see Section 2.1). In particular, r_x was not fixed in the LF256 strain, which refutes the hypothesized role of this cadherin allele in resistance

Figure 2

Survival at the diagnostic concentration of Cry1Ac of two resistant strains (LF256 and SCD‐r1), a susceptible strain (SCD) and the progeny from crosses between strains. Asterisks indicate 0% survival for the SCD strain and the F₁ progeny from the cross between the SCD strain and the SCD‐r1 strain

Figure 3

Alignment of polymorphic amino acids predicted from sequencing the cadherin gene in male #256, resistant strain LF256, and susceptible strain SCD of Helicoverpa armigera. Dashes indicate the amino acids are the same as in the r_x allele from male #256. Only LF256‐2 was identical to r_x. The red line shows the putative toxin‐binding region of HaCad. No mutations were found in SCD or LF256 in amino acids 1,422–1,440, which are especially important in binding (Zhang et al., 2017)

Figure 4

Analysis of the SCD (susceptible), LF256 (resistant), and SCD‐r1 (resistant) strains of Helicoverpa armigera. (a) Western blot of cadherin protein. (b) Ligand blot of binding of Cry1Ac to cadherin protein. Both blots show similar bands for LF256 and SCD, and no band (a) or a weaker band (b) for SCD‐r1