Ledprona is a 490bp dsRNA that is designed to target the L. decemlineata PSMB5 mRNA (GenBank accession XM_023158308.1). dsRNA was produced using GreenLight Bioscience’s proprietary cell-free bioprocessing platform. The 490bp long sequence comprises a 460bp dsRNA sub-sequence targeting PSMB5 mRNA, which is the bioactive sequence, and 15bp transcribed spacer (ITS) flanking sequences on both ends necessary for RNA transcription.

To characterize Ledprona and its processed siRNA products, small RNA (sRNA) was sequenced from two treatments including dsGFP (dsRNA negative control targeting green fluorescent protein mRNA) and Ledprona. Each sample consisted of three replicates, and each replicate included two second instar CPB larvae fed for 72h on leaves treated with the respective dsRNA. Prior to sequencing, total RNA was extracted from six individual larvae per treatment as described below. Equivalent amounts of RNA from two individuals were pooled, resulting in three pools per treatment. Those RNA pools were then used for sequencing library preparation. sRNA sequencing libraries were created by Amaryllis Nucleics (Oakland, CA, United States) using NEXTFLEX kit and sequenced using the Novaseq single end 100bp module. The raw sequence read data for the samples were initially checked for quality using FastQC (Andrews, 2010) and MultiQC (Ewels et al., 2016). Adapters were then trimmed using cutadapt (Martin, 2011), and the trimmed reads were once again checked for quality using FastQC. The trimmed reads were then mapped to PSMB5 mRNA using ShortStack (Johnson et al., 2016). The alignment file created by ShortStack was then parsed with pysam (Li et al., 2009) to obtain only those reads that aligned in the sense (SAM flag 0) or antisense (SAM flag 16) orientations.

Colorado potato beetles were from a colony founded by adults collected from untreated potato plots at Aroostook Research Farm, Presque Isle, ME, United States, and maintained under laboratory conditions (Galimberti and Alyokhin, 2018). Approximately half of the colony was replaced with field-collected beetles every summer to minimize selection toward adaptation to captivity and genetic drift. Beetles were kept on potted potato plants in wood and fine mesh cages (50×50×90cm) in a research greenhouse under 16L:8D photoperiod. Eggs were collected and kept in a growth chamber (Percival Scientific Inc., Perry, IA) at 25°C with a 16:8h L:D photoperiod until larvae hatched, at which point they were either used in experiments or returned to the colony (Galimberti and Alyokhin, 2018).

A single potato leaflet was dipped into 100ml of Ledprona or dsGFP (negative control) for complete coverage with each respective treatment. After air drying, each leaflet was placed into a Petri dish (90mm×15mm) with a damp paper towel on the bottom. Ten second instar larvae were placed on top of the leaflet using a brush. Fresh treated leaf material was added as needed to individual Petri dish if leaflet was defoliated or desiccated. Larval mortality was recorded daily for 8days. Larvae that did not move after being probed by a soft brush were considered dead. Petri dishes were arranged in randomized complete block design and kept in an environmental chamber at 24±1°C. The experiment was repeated five times. Serial dilutions of Ledprona in distilled water were prepared in the beginning of each replication.

Normality of collected data was tested using Shapiro–Wilk test (PROC UNIVARIATE, SAS Institute, 2018) and determined to be non-normal (W=0.74, p<0.0001). Subsequently, the data were transformed using rank transformations (Conover and Iman, 1989; PROC RANK, SAS Institute, 2018). Transformed data were subjected to repeated-measures ANOVA (PROC MIXED, SAS Institute, 2018). Treatment and day of experiment were used as the main factors. Tukey tests were used to separate means among the treatments. Since the interaction between the treatment and the day of observation was significant (see Section “Small RNA Length Distribution”), differences among the treatments were tested using SLICE option (PROC MIXED, SAS Institute, 2018).

Larval survival was adjusted using Abbott’s formula (Abbott, 1925). Lethal concentrations killing 50 and 90% of the exposed populations and lethal times required for 50 and 90% of the exposed population were calculated using probit analyses (PROC PROBIT, SAS Institute, 2018).

Insects used in this experiment were taken from the same colony described in Section “Dose Response Laboratory Bioassay” (Galimberti and Alyokhin, 2018). Details of the experimental design are shown in Supplementary Table 1. For each treatment, twenty potato leaflets (roughly 60mm×85mm) were dipped into 50ml of Ledprona or dsGFP (negative control) solution at 255×10⁻⁵g/L. After air drying, each leaflet was placed into a Petri dish (100mm×15mm) on moisture filter paper (7cm size; treated with 500μl of deionized water). One second instar CPB larva was placed on top of the leaflet using soft tip forceps and held at room temperature. Leaflets were replaced with treated or untreated leaflets, according to Supplementary Table 1. Fresh leaf material was added to the Petri dish if the leaflet was consumed or desiccated. Each treatment had twenty replicates (N=20), and the assay was repeated twice (total N=40). The experiment was carried out for 9days, and the number of dead larvae was observed on 2, 3, 4, 6, 7, and 9days after larvae were exposed to dsRNA. One-way ANOVA was performed and followed by Least Significant Difference test to calculate the smallest significant between two means using Agricultural Research Management (ARM) software (GDM Solutions, Inc., Brookings, SD).

A greenhouse trial was conducted at the University of Maine Aroostook Research Farm in Presque Isle, Maine, United States. Insects used in this experiment were taken from the same colony described in Section “Dose Response Laboratory Bioassay” (Galimberti and Alyokhin, 2018). One potato plant (v. Katahdin) was grown per pot with six replications per treatment. Treatments were mixed in water to achieve 187Lha⁻¹ spray volume and applied to the entire plant with a spray bottle. Each plant was infested with 10 late 1st instar CPB larvae after treatments were allowed to dry on the leaves. Treatments included Ledprona doses of 0.3, 0.6, 1.2, 2.5, and 4.9g ai ha⁻¹, an untreated control, and spinosad (Blackhawk, Corteva Agriscience) at 0.8g ai/ha⁻¹ as a positive control. Evaluation of larval survival was taken at 1, 3, 7, and 14days after treatment. Whole plant percent defoliation ratings were visually estimated 14days after treatment. One-way ANOVA was performed and followed by Tukey’s Honest Significant Difference (HSD) test to separate treatment means using ARM software (GDM Solutions, Inc., Brookings, SD).

The alignment between the 460bp dsRNA homologous to the PSMB5 mRNA is shown in Figure 1. The mRNA encoded by PSMB5 is 1,010 nucleotide (nt) long. The coding sequence of the mRNA ranges from nt positions 106–945. A local sequence alignment performed using Emboss Water (Rice et al., 2000) between Ledprona, and the proteasome subunit beta type-5 mRNA reveals that Ledprona bears 100% identity to the coding sequence of the mRNA. Ledprona has an 18nt gap in the alignment with PSMB5 mRNA which represents a deletion (position 631–648) introduced during dsRNA design.

To investigate the length distribution of sRNA sequences originating from CPB, second instar larvae were exposed to Ledprona and dsGFP. As expected, dsGFP-treated larvae had a low number of sRNA reads that mapped to the PSMB5 mRNA sequence, ranging from 18 to 40nt (an average of less than 5 across all lengths and 11 that are 21bp long). However, larvae fed on Ledprona showed an abundance of sRNA identical to the PSMB5 mRNA sequence (Figure 2). The total sRNA distribution across all samples is shown in Supplementary Figure 1. Figure 2 shows a peak of 21nt sRNA at least three times higher than other sRNA sizes. The three replicate average shows ~14,000 21nt RNAs matching the PSMB5 from the sense strand; however, we observed almost half of this number (~8,500 21nt RNA) from the antisense strand.

An unexpected second peak of 33bp RNAs was observed in both sense and antisense strands. We found that for two out of three replicates (Figure 2; Rep 1 and 3) the largest peak was for 21nt, but one of the replicates (Rep 2) showed 33nt RNAs count higher than the 21nt peak. To assess if this double-peak (22nt and 33nt) was inherent to the sRNA population of the insect, and to check if these could be due to other non-coding RNA such as piRNA, we performed two analyses. First, we mapped the siRNA reads from dsGFP-treated larvae to the control dsGFP sequence to see if we observe a double-peak independent of the sequence. We observed only a single 21nt peak as shown in Figure 3, indicating that the 33nt peak is not an inherent sRNA population. Second, to assess if these 33nt sequences were piRNAs, we performed two analyses. We looked for the signature nucleotide bias observed in ping-pong derived piRNAs (Vodovar et al., 2012) in the 33nt sequences produced in response to the Ledprona treatment. As indicated by the sequence logo image in Figure 4, we found an inconclusive 1U/10A signature as highlighted. The information score (bits) in the sequence logo was low (<0.1), and the difference in enrichment for U at position 1 and T at position 10 was not considerably different from other bases as shown in Figure 4. To further confirm this, we ran PingPongPro (Uhrig and Klein, 2019) on 33nt reads mapped to PSMB5 for the Ledprona-treated larvae. This software tool predicts ping-pong-derived piRNA signatures in sequences along with their FDR values. We did not observe any signature with FDR<0.05 (recommended cutoff) in all three Ledprona-treated replicates (Supplementary Table 5). These findings, in combination with the fact that these 33nt peaks are identical to the PSMB5 gene, lead us to conclude that these 33nt peaks observed are not piRNAs and might be variation observed with a single replicate of sequencing.

The distribution of sRNA that aligned to the PSMB5 mRNA showed that 21nt RNAs were most abundant. Those were likely the Dicer products responsible for silencing PSMB5 in CPB. To confirm that those sRNAs were a product of Ledprona and to assess whether those molecules originated preferentially from one or more locations, the 21nt sRNAs identified were aligned PSMB5 mRNA (Figure 5). Figure 5 shows the 21nt RNAs that mapped along the length of PSMB5 mRNA sense and antisense strands (in the 5′ to 3′ orientation). We observed that 100% of the 21nt RNAs sequenced mapped to PSMB5 mRNA, confirming Ledprona as the sRNA origin. There were no hotspot siRNA locations identified, with some regions showing higher peaks of 21nt alignment than others but they were variable along the mRNA and across biological replicates. In addition, we also identified the expected gap in the alignment of Ledprona sRNAs and PSMB5 mRNA that corresponded to the 18nt absent from the dsRNA sequence (nt 631–648).

Larval mortalities caused by Ledprona consumption were statistically different among different concentrations (F11,48=46.71, p<0.0001). Mortality also increased as the time progressed (F7,336=179.93, p<0.0001), and the interaction between the treatment and the day of observation was also significant (F77,336=6.46, p<0.0001; Figure 6). Comparably few larvae died within 1day following treatment, after which mortality progressed proportionally to the concentration of Ledprona. Most control larvae, as well as the larvae treated with the two lowest concentrations, survived until the end of the experiment. Performance of the three highest concentrations was similar, especially starting on the fifth day after the treatment. Other concentration responses were between those extremes.

Ledprona acted slowly. Mortality did not exceed 40% until the fourth day after treatment even on leaflets treated with the highest concentration of Ledprona (Figure 6). Therefore, LC values were not calculated for the first 3days of observation. For other days, all log-probit models were significant (Table 1). As expected, LC values declined as time progressed. No differences were detected on the last 2days of the experiment, as evidenced by overlaps in 95% fiducial limits (Table 1). Model provided a good fit on the fourth day of the experiment, but there was a large difference between LC50 and LC90. Furthermore, fiducial limits around LC90 were very wide. Therefore, its estimate was likely not very reliable. On days 6 and 7, the data did not fit log-probit models, probably due to its high variation. On days 8 and 9, the fit was good. Because Ledprona appears to act slowly, a minimum of 8days post treatment is needed for a reliable toxicological estimate.

Ledprona was not very larvicidal at low concentrations. The lowest three concentrations never reached 50% mortality (Figure 6). Therefore, they were not used to calculate LT values. For all other concentrations, models were significant (Table 2). As expected, lethal times decreased as concentrations increased (Table 2). Good fit was observed only for the highest three concentrations tested, suggesting that a sufficiently high dose is needed for a trustworthy evaluation of LT values.

Second instar larvae showed 2.5% mortality after 7days and 10% mortality after 9days of constant feeding on potato leaves treated with dsGFP negative control at 255×10⁻⁵g/L. No mortality correction was used for data analysis because of the low background mortality. From day 6, all durations of Ledprona treatments at 255×10⁻⁵g/L were different from the negative control (Figure 7).

Larvae fed on Ledprona for 6h resulted in 50% mortality at 6days after initial exposure. The mortality increased to 85 and 90%, respectively, on days 7 and 9. Larvae fed on Ledprona for longer periods of time, 24, 48, 72h, and 9 consecutive days, showed more rapid death and higher final percent mortality (Supplementary Figure 2). In these treatments, 75 to 80% of the larvae were dead on day 6 and 97.5 to 100% mortality was observed after 3 additional days.

Molecular analysis of samples was performed in order to detect knockdown of PSMB5 mRNA and protein levels in larvae exposed to Ledprona. Relative RT-qPCR was used to estimate the level of PSMB5 transcript, and isotopically labelled synthetic peptides enabled absolute quantification of PSMB5 protein by mass spectrometry.

The 48% lower mRNA expression for larvae fed on Ledprona for as little as 6h was not significant when included in a multiple comparison (Sidak’s multiple comparison test; Figure 8A). However, it was significant when considered alone relative to the dsGFP control (t-test, p=0.0130). PSMB5 expression significantly decreased (72–89%) in larvae exposed to Ledprona for more than 24h. Figures 8B, 9B show the nanomolar ratio of PSMB5 protein relative to total protein after normalization to total protein recovered from beetle biomass. Larvae feeding for 6h did not show decrease in protein level after 72h (Figure 8B). Larvae exposed to Ledprona for 24h showed significantly lower (53%) PSMB5 protein level when considered alone (t-test, p=0.0297), but were not significant when included in a multiple comparison (Sidak’s multiple comparison test). On the other hand, larvae feeding for 24h and 72h showed 76–88% lower protein level compared to control. Although mRNA expression did not significantly vary for larvae fed for longer than 24h (Figure 8A), the protein level required 48h or longer exposure to display significant levels of knockdown (Figure 8B).

The duration of the RNAi effect on mRNA expression and protein levels was assessed by treating larvae with dsRNA for 24h and then transferring the larvae to untreated leaves and collecting samples after 12, 24, 48, or 72h. Figure 9A shows that 78% decrease in PSMB5 mRNA expression was detected as early as 12h after treatment and mRNA knockdown persisted for up to 72h (84% decrease), without recovery of mRNA levels. As expected, PSMB5 protein reduction took longer than mRNA knockdown. A decrease in PSMB5 was not observed after 12h or 24h (Figure 9B); however 47 and 95% reduction in PSMB5 was observed at 48h and 72h, respectively, after initial exposure.

Colorado potato beetle survival was 100% 1day after treatment with Ledprona while spinosad reduced survival to 50% (Tukey’s HSD test; Table 3; Figure 10). At 3 and 7days after treatment, the untreated control had 70% CPB survival while spinosad reduced survival to 0%. Ledprona did not influence survival compared to the untreated control at 1, 3, or 7days after treatment. However, Ledprona at all rates tested resulted in 0% survival at 14days after treatment, which was similar to spinosad (Table 3; Figure 10). Whole plant defoliation with Ledprona at 14days after treatment was reduced to less than 5% compared to 52% in the untreated control. Ledprona at all rates except the middle rate (1.2g ai ha⁻¹) reduced defoliation similar to spinosad (Figure 11).