Full crop protection from an insect pest by expression of long double-stranded RNAs in plastids

See allHide authors and affiliations

Science  27 Feb 2015:
Vol. 347, Issue 6225, pp. 991-994
DOI: 10.1126/science.1261680

Bypassing a plant's defense for pest defense

Colorado potato beetles can skeletonize the leaves on a potato plant, devastating crop yields. Insecticides are increasingly useless as the beetle evolves resistance. Zhang et al. used RNA interference to take down this beetle (see the Perspective by Whyard). Success required shifting production of the double-stranded RNA to the plastids to evade the plant's own RNA management mechanisms. The insect's own RNA interference mechanisms then inactivated two everyday genes that the beetle can't do without.

Science, this issue p. 991; see also p. 950


Double-stranded RNAs (dsRNAs) targeted against essential genes can trigger a lethal RNA interference (RNAi) response in insect pests. The application of this concept in plant protection is hampered by the presence of an endogenous plant RNAi pathway that processes dsRNAs into short interfering RNAs. We found that long dsRNAs can be stably produced in chloroplasts, a cellular compartment that appears to lack an RNAi machinery. When expressed from the chloroplast genome, dsRNAs accumulated to as much as 0.4% of the total cellular RNA. Transplastomic potato plants producing dsRNAs targeted against the β-actin gene of the Colorado potato beetle, a notorious agricultural pest, were protected from herbivory and were lethal to its larvae. Thus, chloroplast expression of long dsRNAs can provide crop protection without chemical pesticides.

Double-stranded RNA (dsRNA) fed to insects can be taken up by midgut cells and processed into small interfering RNAs (siRNAs) by the insect’s Dicer endoribonuclease (13). If the sequence of the fed dsRNA matches that of an insect gene, gene silencing by RNA interference (RNAi) disrupts expression of the insect’s gene (3, 4). By targeting essential insect genes, dsRNAs can be developed into highly species-specific insecticides (4). However, although expression of dsRNAs targeted against insect genes in transgenic plants (1, 2, 58) has impaired growth and development, complete protection of the plants and efficient killing of the insects have not been achieved. dsRNAs at least 60 base pairs (bp) in length are required for efficient uptake and biological activity in the target insect (3), but the plant’s own system for producing small RNAs (9) prevents the accumulation of high amounts of long dsRNA. The major processing products of dsRNA cleavage by Dicer are 21-bp siRNAs, but these had little (10) or no effect when fed to insects (3). Thus, rapid turnover of dsRNAs in the plant limits the efficacy of transgenic RNAi-based anti-insect strategies.

The plastids (chloroplasts) of plant cells are derived from formerly free-living cyanobacteria, a group of prokaryotes that lack an RNAi pathway. We reasoned that chloroplasts might be capable of stably accumulating long dsRNAs, in which case dsRNA expression from the plastid genome could provide better protection against insect pests than dsRNA expression from the nuclear genome. To test the feasibility of stable dsRNA expression in plastids, we transformed the tobacco (Nicotiana tabacum) plastid genome with three different types of dsRNA constructs (Fig. 1A and fig. S1). In ptDP constructs, the dsRNA is generated by transcription from two convergent (dual) promoters. In ptSL constructs, the dsRNA is also produced from two convergent promoters, but each strand is additionally flanked by sequences forming stem-loop–type secondary structures, which increase RNA stability in plastids (11). In ptHP constructs, hairpin-type dsRNA (hpRNA) is produced by transcription of two transgene copies arranged as an inverted repeat (Fig. 1A). We targeted the Colorado potato beetle (Leptinotarsa decemlineata; CPB), a notorious insect pest of potato and other solanaceous crops (e.g., tomato and eggplant). Both larvae and adults feed on foliage, skeletonize the leaves, and, if left uncontrolled, completely destroy the crop. In many areas of the world, the beetle has no natural enemies, and chemical pesticides are the main method of CPB control. However, since the middle of the 20th century, CPB has developed resistance to all major insecticide classes (and therefore has been branded an “international superpest”) (12).

Fig. 1 Expression of dsRNAs in plastids.

(A) Map of transformation vectors for dsRNA expression from the plastid genome. The cassettes designed to produce the three different types of dsRNAs (ptDP, ptSL, and ptHP) are schematically depicted below the map, along with the expected structures and sizes of the dsRNAs. The selectable marker gene aadA is driven by the psbA promoter (PpsbA) and the 3′ untranslated region (3′ UTR) of the rbcL gene (TrbcL) from Chlamydomonas reinhardtii. DNA sequences from CPB target genes are shown in orange. SL1 and SL2, stem-loop–encoding sequences; Prrn, tobacco rRNA operon promoter; TrrnB, rrnB terminator from E. coli; intron, first intron from the potato GA20 oxidase gene. (B) Northern blot analysis of dsRNA accumulation in transplastomic tobacco and potato lines; 5 μg of total RNA was loaded in each lane, and the band sizes of the RNA marker are given at the left. The ethidium bromide–stained gel prior to blotting is shown below each blot. The asterisk indicates a shorter-than-expected transcript present in Nt-ptHP-ACT+SHR lines. Accumulation of some larger RNA species is likely due to read-through transcription, which is common in plastids (22, 23). Note that transplastomic lines generated with the same construct show identical transgene expression levels due to targeting by homologous recombination. (C) Comparison of dsRNA accumulation levels in leaves and tubers of transplastomic potato lines expressing ACT dsRNA; 5 μg of total cellular RNA was loaded per lane. The ethidium bromide–stained gel prior to blotting is shown below the blot. (D) Analysis of ACT dsRNA accumulation by Northern blotting. The amount of total RNA loaded in each lane is given (in μg). The ethidium bromide–stained gel prior to blotting is shown below the blot as a loading control. Note that 10 times as much RNA was loaded for the nuclear transgenic lines. (E) Analysis of siRNA accumulation by Northern blotting. Note that siRNAs accumulate only in the nuclear transgenic plants but not in the transplastomic plants, confirming that the dsRNAs produced in the plastid stay put. Thus, although the CPB ACT sequence used has some similarity to the potato ACT gene (66% over a stretch of 226 nucleotides), it cannot even theoretically silence the plant’s endogenous ACT gene because the chloroplast-produced dsRNAs do not leak out into the cytosol.

As essential target genes, the CPB ACT and SHR genes were chosen. ACT encodes β-actin, an essential cytoskeletal protein, and SHR encodes Shrub (also known as Vps32 or Snf7), an essential subunit of a protein complex involved in membrane remodeling for vesicle transport. Disruption of these genes when the insects are fed in vitro synthesized dsRNAs induces mortality with high efficacy (3, 13). To test longer dsRNAs and test for a possible synergistic action, we also produced an ACT+SHR fusion gene. To confirm the activity of these dsRNAs in the beetles, we synthesized the dsRNAs (ACT, SHR, ACT+SHR fusion, and GFP as a control) by in vitro transcription, painted them onto young potato leaves, and fed the leaves to second-instar CPB larvae. All three insect gene–derived dsRNAs reduced larval growth (fig. S2). The ACT dsRNA was more effective than the SHR dsRNA, and the ACT+SHR dsRNA was the least effective dsRNA (fig. S2), indicating that targeting two insect genes with the same dsRNA does not necessarily enhance insecticidal activity.

We first evaluated the three strategies for in vivo dsRNA production (ptDP, ptSL, and ptHP constructs; Fig. 1A and fig. S1C) with the ACT+SHR fusion gene in tobacco plants, because chloroplast transformation is relatively routine in this species. Transplastomic tobacco lines were produced by particle gun–mediated chloroplast transformation and purified to homoplasmy by additional rounds of regeneration and selection. Transgene integration into the plastid genome by homologous recombination and elimination of all wild-type copies of the highly polyploid plastid genome were confirmed by restriction fragment length polymorphism analyses and inheritance assays (figs. S1E and S3A). All transplastomic lines (referred to as Nt-ptDP-ACT+SHR, Nt-ptSL-ACT+SHR, and Nt-ptHP-ACT+SHR) displayed no visible phenotype and were indistinguishable from wild-type plants, both under in vitro culture conditions and upon growth in the greenhouse (fig. S3, B and D), indicating that dsRNA expression in the chloroplast is phenotypically neutral.

To test whether dsRNAs stably accumulate in chloroplasts, we performed Northern blot analyses. The results revealed that all three types of expression constructs triggered production of substantial amounts of long dsRNAs (Fig. 1B), suggesting the absence of efficient dsRNA-degrading mechanisms from plastids. dsRNA accumulation levels in Nt-ptDP and Nt-ptSL plants were very similar, indicating that the terminal stem-loop structures added to the ptSL constructs do not increase dsRNA stability (Fig. 1, A and B). dsRNA accumulation levels in Nt-ptHP lines were even higher but included shorter-than-expected transcripts (Fig. 1B), possibly because the plastid RNA polymerase encounters difficulties transcribing sequences containing large inverted repeats. Therefore, we used the convergent promoter approach (ptDP constructs) for dsRNA expression in all subsequent experiments.

We next introduced the three target gene constructs (ACT, SHR, and ACT+SHR, integrated into the ptDP cassette; Fig. 1A) into the plastid genome of potato (14) (see supplementary materials), the main host of CPB, and isolated homoplasmic transplastomic lines (St-ptDP-ACT, St-ptDP-SHR, and St-ptDP-ACT+SHR; fig. S1, B and F). To be able to compare the level of protection from herbivory in transplastomic and nuclear transgenic plants, we introduced the identical transgenes (as hairpin constructs containing a spliceosomal intron that is posttranscriptionally removed) (15) into the nuclear genome by Agrobacterium-mediated transformation (St-nuHP; fig. S1D). Phenotypic analyses showed that all transplastomic and nuclear transgenic potato plants were indistinguishable from wild-type plants with regard to growth and tuber production (fig. S3, C and E, and figs. S4 and S5).

Northern blot analyses of transplastomic potato lines revealed that the accumulation levels of ACT dsRNAs were higher than those of SHR and ACT+SHR dsRNAs (Fig. 1B). Comparison to a dilution series of in vitro synthesized RNA revealed dsRNA accumulation levels in leaves of ~0.4% of the total cellular RNA for ACT, ~0.05% for SHR, and ~0.1% for ACT+SHR (fig. S6). By contrast, hybridization signals in the nuclear transgenic plants could be detected only upon overloading of the gels and/or overexposure of the blots, consistent with efficient degradation of dsRNAs into siRNAs by the plant’s endogenous RNAi machinery (Fig. 1D and figs. S7 and S8A). The presence of siRNAs in nuclear transgenic plants and their absence from transplastomic plants were directly confirmed by Northern blot analyses (Fig. 1E).

Because CPB larvae and adults feed on leaves but not on belowground potato tubers, only the leaves need to be protected from herbivory. Nonphotosynthetic tissues, such as tubers, express plastid to much lower levels than photosynthetic tissues, such as leaves (16, 17). Thus, despite whole-plant transformation, dsRNA production in tubers, where accumulation of transgene-derived RNA is unnecessary and perhaps undesired by the consumer, was below or near the detection limit (Fig. 1C and fig. S8B).

Having established that long dsRNAs accumulate to high levels in leaves of transplastomic potato lines, we next tested whether dsRNA production in the chloroplast offers protection against CPB. To this end, the mortality of first-instar CPB larvae was determined upon feeding on detached leaves from wild-type, transplastomic, and nuclear transgenic potato plants for 9 consecutive days (Fig. 2A and fig. S9A). In addition, the weight of all surviving larvae was measured to follow their growth (Fig. 2, B and C; fig. S9, B to D; and fig. S10). The bioassays revealed that all transplastomic potato plants induced high mortality (Fig. 2, A and C, and fig. S9A). Consistent with the high expression level of ACT dsRNA (Fig. 1B and fig. S6) and the high efficacy of in vitro synthesized ACT dsRNA (fig. S2), the most potent insecticidal activity was conferred by the ACT dsRNA–expressing transplastomic plants that caused 100% mortality within 5 days. By contrast, none of the nuclear transgenic potato plants conferred any larval mortality (Fig. 2A and fig. S9A), in line with the earlier finding that short siRNAs fed to insects have only small effects or do not induce an RNAi response at all (3). However, all nuclear transgenic lines caused reduced growth of CPB larvae (Fig. 2B), presumably due to the small amounts of dsRNAs the plants accumulate (Fig. 1D and figs. S7 and S8). While none of the CPB larvae survived feeding on transplastomic St-ptDP-ACT plants, some of the larvae survived for 9 days on St-ptDP-SHR and St-ptDP-ACT+SHR leaves. However, these survivors suffered from very strong growth retardation (Fig. 2B and fig. S9, C and D).

Fig. 2 Feeding assays of CPB larvae on transplastomic and nuclear transgenic potato plants.

(A) Survivorship of first-instar larvae upon feeding on detached leaves of wild-type, transplastomic, and nuclear transgenic potato plants. (B) Growth of surviving larvae. The weight of survivors was determined after 3, 5, 7, and 9 days of feeding. Data are means ± SD (n = 30). Significant differences to the wild-type control were identified by analysis of variance. *P < 0.05, **P < 0.01, ***P < 0.001. The best-performing nuclear transgenic lines were included in the assay (see figs. S7 to S10). For assays with additional transplastomic lines, see fig. S9. Note that the weight of survivors in the assays with the transplastomic plants expressing ACT dsRNA (St-ptDP-ACT21) could only be measured until day 3, because all larvae were already dead at day 5 [see (A)]. (C) Example of a bioassay with detached leaves of wild-type potato plants and nuclear transgenic and transplastomic leaves expressing ACT dsRNA. Leaves were exposed to first-instar CPB larvae, replaced with fresh young leaves every day, and the photograph was taken at day 3. Note that almost no visible damage is seen in St-ptDP-ACT21 leaves.

To confirm that the killing of the CPB larvae by the transplastomic plants was due to induction of RNAi, we determined expression of the target genes in the larval gut after 3 days of feeding (i.e., when the larvae fed on the transplastomic plants were still alive). Already at this early stage, expression of β-actin and Shrub was suppressed in the insects (fig. S11, A and B). As expected on the basis of the mortality data (Fig. 2A), target gene suppression was strongest in larvae fed on St-ptDP-ACT plants. Moreover, accumulation of ACT-derived siRNAs was detected in gut tissue of larvae fed with transplastomic leaves, whereas accumulation in larvae fed with nuclear transgenic leaves was below the limit of reliable detection (fig. S11C).

CPB resistance of transplastomic potato plants was further assessed by determining the leaf area consumed by CPB larvae and adult beetles. Almost no visible consumption of leaf biomass occurred in St-ptDP-ACT leaves (Fig. 2C and fig. S12A), due to complete cessation of larval feeding after 24 hours, even prior to mortality (Fig. 2A). Similarly, adult beetles caused very little additional damage to St-ptDP-ACT leaves after 2 days (Fig. 3A). Finally, we exposed whole plants to second-instar larvae (which are generally less sensitive to insecticidal agents than first-instar larvae) and scored survival. This test resulted in 17% survival of the larvae after 5 days of feeding on St-ptDP-ACT plants (and 63% survival upon feeding on St-ptDP-SHR plants after 6 days; Fig. 3B, fig. S12, B to D, and fig. S13), presumably due to the initial larval growth and development on wild-type leaves. However, the larvae grew very poorly after transfer to the transplastomic plants, and the damage they caused to the leaves was very small (fig. S12, B to D). In nature, CPB larvae typically hatch and feed on the same plant, and therefore they would not enjoy a wild-type diet before feeding on the transplastomic plant.

Fig. 3 Consumption of detached leaves of potato plants by adult beetles, and survivorship of larvae upon feeding on whole plants.

(A) Leaf area consumed by freshly emerged adult beetles fed on leaves of wild-type potato plants, nuclear transgenic plants, and transplastomic plants expressing ACT dsRNA. As an additional control, leaves painted with in vitro synthesized GFP-derived dsRNA were included. Data are mean ± SD (n = 24 for St-wt, n = 12 for all other lines). (B) Survivorship of second-instar CPB larvae after feeding on whole plants at day 5 (see fig. S12).

To ultimately confirm that the plastid-expressed ACT dsRNA silences the actin gene in CPB larvae, we examined actin filaments in the larval midgut, hindgut, and Malpighian tubules by staining with fluorescein isothiocyanate (FITC)–labeled phalloidin. Already after 1 to 2 days of feeding on transplastomic leaves, the larvae displayed disorganized actin filaments, which were particularly obvious in the columnar cells of the midgut (fig. S14). Also, the intensity of phalloidin-FITC labeling progressively decreased with the time of feeding (fig. S14), which suggests that actin deficiency is the cause of death. RNA-dependent RNA polymerase (RdRP) genes are absent from the genomes of insects (18). Therefore, silencing signals are not amplified at the RNA level, and RNAi effects remain restricted to those cells that have taken up (or produced) silencing-inducing dsRNAs. Consequently, a continuous input of dsRNAs is required for efficient gene silencing in insects. Because of the low stability of dsRNAs expressed from the nuclear genome and their efficient processing by the plant’s own Dicer endoribonucleases, complete protection of plants from insect pests has not been accomplished (1, 2). Our findings underscore the importance of producing large amounts of long dsRNAs to achieve efficient protection. Whereas transplastomic ACT dsRNA–expressing plants cause 100% lethality to CPB larvae, SHR dsRNA–expressing plants are somewhat less efficient (70% mortality after 9 days; Fig. 2A). This correlates with the factor of ~8 lower accumulation levels of the SHR dsRNA. Because both constructs are driven by the same expression signals, we conclude that the SHR sequence chosen is less stable in potato plastids than the ACT sequence. Consequently, testing other fragments of the SHR gene seems an appropriate future strategy to improve the insecticidal efficiency of SHR dsRNA–expressing transplastomic plants.

As insect pests are developing increasing resistance to chemical insecticides and Bt toxins (12, 1921), RNAi technology is becoming a promising future strategy for plant protection. RNAi-activating dsRNAs can be chosen from a vast number of potential target genes. Moreover, the dsRNA approach provides plant protection without chemicals and without synthesis of foreign proteins in the plant. We have shown that plastids can be engineered to produce the quantities of dsRNA needed to control a major agricultural pest, the Colorado potato beetle. Shifting the target of transgenesis from the nucleus to the plastid removes the major hurdle on the way to exploiting transgenically delivered RNAi for efficient plant protection in the field (18).

Supplementary Materials

Materials and Methods

Figs. S1 to S14

Table S1

References (2437)

References and Notes

  1. Acknowledgments: We thank P. Endries for help with plant transformation, E. Maximova for help with microscopy, Y. Zhang for providing phalloidin-FITC, the MPI-MP Greenteam for plant cultivation, M. Kaltenpoth and B. Weiss for help in insect tissue preparation, and T. Krügel and the MPI-CE greenhouse team. This research was financed by the Max Planck Society. Plastid transformation vectors and transplastomic plants are available under a material transfer agreement. J.Z., S.A.K., D.G.H., R.B., and the Max Planck Society (MPG) have filed a patent application (EP14199415) that relates to the transplastomic production of dsRNAs for plant protection. Supplement contains additional data.
View Abstract

Stay Connected to Science

Navigate This Article