Next-generation transgenic cotton: pyramiding RNAi and Bt counters insect resistance

Abstract

Transgenic crops producing insecticidal proteins from the bacterium Bacillus thuringiensis (Bt) are extensively cultivated worldwide. To counter rapidly increasing pest resistance to crops that produce single Bt toxins, transgenic plant 'pyramids' producing two or more Bt toxins that kill the same pest have been widely adopted. However, cross-resistance and antagonism between Bt toxins limit the sustainability of this approach. Here we describe development and testing of the first pyramids of cotton combining protection from a Bt toxin and RNA interference (RNAi). We developed two types of transgenic cotton plants producing double-stranded RNA (dsRNA) from the global lepidopteran pest Helicoverpa armigera designed to interfere with its metabolism of juvenile hormone (JH). We focused on suppression of JH acid methyltransferase (JHAMT), which is crucial for JH synthesis, and JH-binding protein (JHBP), which transports JH to organs. In 2015 and 2016, we tested larvae from a Bt-resistant strain and a related susceptible strain of H. armigera on seven types of cotton: two controls, Bt cotton, two types of RNAi cotton (targeting JHAMT or JHBP) and two pyramids (Bt cotton plus each type of RNAi). Both types of RNAi cotton were effective against Bt-resistant insects. Bt cotton and RNAi acted independently against the susceptible strain. In computer simulations of conditions in northern China, where millions of farmers grow Bt cotton as well as abundant non-transgenic host plants of H. armigera, pyramided cotton combining a Bt toxin and RNAi substantially delayed resistance relative to using Bt cotton alone.

Keywords: Bacillus thuringiensis; Helicoverpa armigera; RNA interference; genetic engineering; juvenile hormone; sustainability.

Figure 1

Fragments of H. armigera genes and vectors used to transform cotton for RNAi affecting juvenile hormone (JH). (a) Gene fragments JHA from HaJHAMT and JHB from HaJHBP (see Figures S1–S4 for sequences). (b) Vectors used for Agrobacterium‐mediated transformation of cotton. We modified plasmid pCAMBIA2300 so dsRNA expression was controlled by promoter PRP from cotton leaf curl virus upstream from fragment JHA, JHB or GFP (negative control) in the sense orientation, an intron from the potato GA20‐oxidase gene, the same fragment in the antisense orientation and a nopaline synthase (NOS) terminator (see Experimental Procedures).

Figure 2

Effects on transcription and JH concentration of feeding by susceptible (SCD) larvae of H. armigera on leaves from four types of cotton plants: non‐transgenic parent (W0), transgenic control (GFP), transgenic cotton producing dsRNA from HaJHAMT (JHA) or HaJHBP (JHB). (a) Transcription of HaJHAMT relative to actin. (b) Transcription of HaJHBP relative to actin. Larvae were not fed JHB cotton in (a) or JHA cotton in (b). Relative transcription was significantly lower for either JHA or JHB cotton than for W0 or GFP (P = 0.001 in each case, Tukey's HSD). (c) JH concentration was significantly lower for JHA and JHB than either W0 or GFP (P < 0.05 in each case, Tukey's HSD). No significant difference occurred between W0 and GFP in relative transcription or JH concentration (P > 0.5 in each case, Tukey's HSD). Bars show means and SE with an average of 4.6 replicates per bar.

Figure 3

Mortality of susceptible (SCD) and resistant (SCD‐r1) larvae of H. armigera fed leaves from seven types of cotton plants: non‐transgenic parent (W0), transgenic control (GFP), B. thuringiensis (Bt), RNAi (JHA and JHB) and pyramids (Bt + JHA and Bt + JHB). (a) 2015. (b) 2016. Each bar shows the mean and SE for mortality (%) based on three replicates of 50 larvae per replicate (n = 150 per bar, total n = 3900). For a given type of cotton and year, asterisks indicate a significant difference between insect strains (t‐tests, ***: P < 0.001, **: P = 0.01, no asterisks: NS, P > 0.20). For a given insect strain and year, different letters indicate significant differences between types of cotton; upper case for SCD and lower case for SCD‐r1 (Tukey's HSD, P < 0.05). SCD was not tested on pyramids in 2015.

Figure 4

Development time for susceptible (SCD) and resistant (SCD‐r1) larvae of H. armigera fed leaves from seven types of cotton plants: non‐transgenic parent (W0), transgenic control (GFP), B. thuringiensis (Bt), RNAi (JHA and JHB) and pyramids (Bt + JHA and Bt + JHB). (a) 2015. (b) 2016. Each bar shows the mean and SE for days from neonate to pupa based on three replicates (mean sample size per bar = 72 pupae, total n = 1884 pupae). For a given type of cotton and year, asterisks indicate a significant difference between insect strains (t‐tests, ***: P < 0.001, **: P = 0.0012). For a given insect strain and year, different letters indicate significant differences between types of cotton; upper case for SCD and lower case for SCD‐r1 (Tukey's HSD, P < 0.05). SCD was not tested on pyramids in 2015.

Figure 5

Computer simulations of the evolution of resistance by H. armigera to Bt cotton alone, a sequence of Bt cotton followed by RNAi cotton and a pyramid of Bt + RNAi cotton. We used a deterministic model with two alleles (r conferring resistance and s susceptibility) at each of two independently segregating loci. Locus one controlled survival on Bt cotton and locus two controlled survival on RNAi cotton. The initial resistance allele frequency was 0.05 for Bt cotton and 0.001 for RNAi cotton. The time to resistance is the number of years until the population fitness on transgenic cotton exceeded 0.50. The time to resistance for the sequence is the sum of the time for resistance to Bt cotton and RNAi cotton. (a) Realistic scenario: dominance of resistance (h) = 0.5 for Bt cotton and RNAi cotton and no fitness cost. (b) Optimistic scenario: dominance of resistance (h) = 0.5 for Bt cotton and 0.2 for RNAi cotton and minor, additive fitness cost. Each r allele reduced fitness on non‐transgenic host plants by 0.05; fitness of doubly resistant homozygotes (r ₁ r ₁ r ₂ r ₂) on non‐transgenic host plants = 0.80. (c) Pessimistic scenario: identical to (a) except initial resistance allele frequency for resistance to RNAi = 0.01.