The research team, led by Biko K. Muita and John M. K. Roberts of the CSIRO Health and Biosecurity Research Unit, focused on a class of advanced dsRNA designs known as loop-ended dsRNA, or ledRNA. Unlike conventional linear dsRNA, ledRNA constructs carry single-stranded terminal loops on both ends of the double-stranded stem, protecting the vulnerable ends from exonuclease attack while preserving the double-stranded region that the cellular RNAi machinery requires. The team paired this architecture with two distinct sequence modifications: a G-U wobble design, in which all cytosines in the sense strand were substituted with thymines to create G-U wobble base pairs throughout the stem, and an asymmetric bulge design, in which a single nucleotide deletion was introduced every 22 nucleotides in the sense strand, generating a corresponding bulge on the antisense strand without altering its sequence.

These designs were tested against two Varroa genes selected from previous work. The first, Peritrophin-A-like, or Pero, encodes a chitin-binding protein involved in maintaining the peritrophic membrane in the mite gut. The second, crustacean hyperglycemic hormone, or CHH, is a neuropeptide that regulates energy metabolism, osmoregulation, and stress responses. Both genes had previously been shown to be effectively silenced with conventional dsRNA, producing moderate levels of mite mortality, making them suitable benchmarks for evaluating whether advanced designs could push RNAi efficacy further.

Stability assays revealed a dramatic improvement. In 50 percent sucrose feeding solution, conventional dsRNA degraded rapidly, with band intensity falling by roughly 35 percent within the first hour, 75 percent by three hours, and becoming undetectable by six hours. Hairpin RNA showed a moderate improvement, retaining about 25 percent intensity at six hours and remaining detectable until the nine-hour mark. In contrast, ledRNA degradation was substantially slower, retaining about 25 percent intensity at nine hours and 18 percent at ten hours, remaining approximately three-fold more abundant than conventional dsRNA between four and six hours.

A similar trend emerged following delivery into adult worker bees. Conventional dsRNA declined rapidly, falling to roughly 18 percent of the initial signal by day two and becoming undetectable by day four. The ledRNA construct decreased to about 60 percent of its initial intensity by day two, 30 percent by day three, and 15 percent by day four. The asymmetric bulge construct showed a comparable pattern, while the G-U modified dsRNA exhibited the strongest persistence, retaining approximately 58, 44, and 40 percent of its initial signal on days two, three, and four respectively. Overall, the advanced constructs exhibited significantly greater persistence than conventional dsRNA, with the G-U design showing the most robust statistical improvement.

Biological efficacy was assessed through two delivery routes. In direct immersion bioassays, mites were submerged in dsRNA solution and then transferred onto white-eye stage pupae. At 48 hours, mite mortality was higher for both gene targets with the advanced designs compared with conventional dsRNA. Relative to the untreated control, which showed 37.5 percent mortality, the asymmetric bulge designs targeting Pero and CHH significantly increased mortality by 52.5 and 37.5 percentage points respectively, followed by the G-U design targeting Pero, which increased mortality by 30 percentage points. Conventional dsRNA showed no elevation in 48-hour mortality above the untreated control for either gene.

Gene silencing measured by qPCR confirmed these biological outcomes. For Pero, significant knockdown was observed with conventional dsRNA at roughly 36 percent reduction, the G-U design at 46 percent, and most strongly with the asymmetric bulge design at 85 percent reduction. For CHH, significant silencing was only detected with the G-U design at roughly 47 percent reduction, despite the observed effects of the bulge design on mite mortality. In oral-delivery bioassays, where dsRNA was fed to adult worker bees parasitized by feeding Varroa mites, the Pero-targeting designs produced strong silencing at 72 hours: 78 percent for conventional dsRNA, 82 percent for the G-U design, and 92 percent for the bulge design. No significant silencing was detected for CHH-targeting designs at the same timepoint.

Small RNA sequencing of mites treated with the advanced designs revealed effective processing into functional siRNAs. Coverage spanned 100 percent of the target region for both genes, consistent with siRNA production being confined to the region targeted by the dsRNA design rather than distributed across the full locus. For the Pero gene, the G-U treatment produced higher-depth coverage, with mean read depth peaking at 565, while the bulge design reached 415. For CHH, the G-U design generated substantially higher-depth coverage, peaking at 5,412, compared with 587 for the bulge design. Conventional dsRNA coverage was very low for Pero and largely restricted to a narrow region for CHH. Across most size classes, antisense reads predominated for the advanced designs, with the bulge design showing 65 and 85 percent antisense reads for Pero and CHH respectively, whereas conventional dsRNA read distributions were skewed toward the sense strand.

The siRNA length distributions further highlighted design and target dependent differences. For Pero, siRNAs derived from the advanced designs showed dominant peaks at the 23 and 24 nucleotide size classes, whereas CHH-derived siRNAs exhibited a shift toward 22 and 23 nucleotide peaks. In most cases, the predominant antisense size classes were accompanied by corresponding sense peaks, consistent with Dicer processing of dsRNA into sense-antisense siRNA duplexes. The authors suggest that the presence of 24 nucleotide antisense-biased siRNA populations from the Pero-targeting designs could be evidence of RNA-dependent RNA polymerase-associated amplification, a secondary siRNA mechanism known from nematodes, although several RdRP homologs identified in Varroa require further functional investigation.

The findings provide the first demonstration of these advanced dsRNA molecules effectively targeting Varroa mites and suggest that structurally optimized designs could address key barriers to field application of RNAi-based biopesticides. The G-U design showed comparatively consistent performance, coinciding with robust target-specific siRNA production and an antisense bias, potentially because G-U wobble pairing introduces an imperfect duplex that alters substrate recognition and cleavage kinetics by arthropod Dicer-2. The asymmetric bulge design, which has been used to increase RNAi efficacy in transgenic plants by shifting siRNA size-class profiles, did not produce a discrete 22-nucleotide siRNA peak in this system, likely reflecting fundamental differences in dsRNA processing between plants and arthropods. The authors caution that confirmation in colony and field-level settings will be necessary to verify practical utility, but the laboratory results support further investigation of structurally optimized dsRNA designs for hive applications, potentially enabling more rapid knockdown of mite outbreaks with longer stability within the hive environment and honey bee host.

Subject of Research: RNAi-based pesticide engineering for Varroa destructor

Article Title: Novel dsRNA designs increase the stability and bioefficacy of RNAi-based pesticides for Varroa destructor

Article References: Muita, B. K., Damayo, J. E., Remnant, E. J., Smith, N. A., Greaves, I. K., Ghodke, A. B., Kumaran, N., & Roberts, J. M. K. (2026). Novel dsRNA designs increase the stability and bioefficacy of RNAi-based pesticides for Varroa destructor. Frontiers in Insect Science, 6, Article 1868457. https://doi.org/10.3389/finsc.2026.1868457

Image Credits: AI Generated

DOI: 10.3389/finsc.2026.1868457

Keywords: Varroa destructor, RNA interference, dsRNA, honey bee, pest control, biopesticide, ledRNA, gene silencing, Novel, designs, increase, stability