Gene silencing in Tribolium castaneum as a tool for the targeted identification of candidate RNAi targets in crop pests

RNAi shows potential as an agricultural technology for insect control, yet, a relatively low number of robust lethal RNAi targets have been demonstrated to control insects of agricultural interest. In the current study, a selection of lethal RNAi target genes from the iBeetle (Tribolium castaneum) screen were used to demonstrate efficacy of orthologous targets in the economically important coleopteran pests Diabrotica virgifera virgifera and Meligethes aeneus. Transcript orthologs of 50 selected genes were analyzed in D. v. virgifera diet-based RNAi bioassays; 21 of these RNAi targets showed mortality and 36 showed growth inhibition. Low dose injection- and diet-based dsRNA assays in T. castaneum and D. v. virgifera, respectively, enabled the identification of the four highly potent RNAi target genes: Rop, dre4, ncm, and RpII140. Maize was genetically engineered to express dsRNA directed against these prioritized candidate target genes. T0 plants expressing Rop, dre4, or RpII140 RNA hairpins showed protection from D. v. virgifera larval feeding damage. dsRNA targeting Rop, dre4, ncm, and RpII140 in M. aeneus also caused high levels of mortality both by injection and feeding. In summary, high throughput systems for model organisms can be successfully used to identify potent RNA targets for difficult-to-work with agricultural insect pests.


Results
T. castaneum screen reveals potential target genes in D. v. virgifera. We verified the lethality and dose response of the 50 genes selected from the iBeetle database by injecting a range of dsRNA concentrations (250 ng/μl, 1 ng/μl, 0.1 ng/μl and 0.01 ng/μl) into T. castaneum larvae (Table 1 and Supplementary Tables 1 and 3). Injection of 250 ng/μl dsRNA caused mortality rates of over 90% in over 75% of the tested target genes (Table 1). Dose responses over time for highly-lethal gene targets ncm, Rop, dre4, and RpII140 injected into T. castaneum larvae appear in Fig. 1. These targets showed significant mortality at doses down to 0.01 ng/μl within 14 days of dsRNA application, except RpII140 which showed a significant reduction in the survival at a doses of 1 ng/µl dsRNA.
Homologs for all 50 T. castaneum lethal genes were identified within the 1 st instar transcriptome of D. v. virgifera (Supplementary Table 1) via TBLASTN searches using T. castaneum NCBI RefSeq protein accession IDs in Table 1. The diet feeding bioassays of 50 D. v. virgifera dsRNAs identified 21 dsRNAs with significantly higher percent mortality, compared to YFP dsRNA negative control (p < 0.001, marked with an asterisk in Fig. 2A and Supplementary Table 1). Of the 50 dsRNA targets tested in D. v. virgifera, 36 also showed significant Growth Inhibition (Fig. 2B). Larval transcript expression levels were compared to the bioassay outcomes for all 50 D. v. virgifera genes; however, no correlation between gene expression and lethality or growth inhibition were found ( Supplementary Fig. 1). In this nine-day bioassay, dsRNAs targeting RpII140, dre4,ncm, CG34184, Rop and Rpb7 transcripts, which showed more than 60% mortality ( Fig. 2A, highlighted in green), were selected for further characterization.
To further confirm target sensitivity and probe the efficacy of the sub-regions of the selected targets, additional dsRNAs were designed. In most cases, the additional sequences, or versions, were located within the initially-tested active dsRNA region. For example, ncm-1 v1 and ncm-1 v2 dsRNAs represent non-overlapping sequences within ncm-1 dsRNA region (Supplementary Materials, Sequence 4). Rpb7-1 v1 is a sub-region or   Figure 3 shows that Rop, RpII140, and dre4 transgenes conferred high levels of root protection in multiple, independent T 0 generation maize lines. The qualitative differences between hpRNA expressing plants and negative controls were also evident from plant photos (Fig. 4)     Feeding bioassays using Rop, dre4, ncm, and RpII140 dsRNAs ( Fig. 6B) caused similar mortality rates in M. aeneus as observed in the injection assays. Significant mortality was induced six days after feeding on dsRNA for dre4 (63.33% ± 12.47) and after 10 days for Rop (46.67% ± 9.43), ncm (80.00% ± 16.33) and RpII140 (80.00% ± 16.33) (Fig. 6B, and Supplementary Tables 4 and 7). All treatments showed at least 90% mortality after two weeks of treatment (Fig. 6B).

Conservation of Rop, dre4, and RpII140 proteins across three coleopteran insects. The domain
analysis of Rop, RpII140, and dre4 showed that the domain architecture of these three proteins is conserved between the three species examined here ( Supplementary Fig. 2

Discussion
Plant-delivered RNAi has recently been developed as an insect control method with the advantage of species-specific gene silencing. The success of RNAi approaches to crop protection will largely depend on a robust oral RNAi response in the target insect and the efficacy of the individual RNAi targets. This is because the transcript knockdown via RNAi seems to be fairly rapid and may occur as early as 24 hours or less after dsRNA feeding 17 . Thus, an RNAi-induced lethal phenotype largely depends on gene target attributes including the time-course of gene silencing, the function of the targeted gene product and the turnover of the target-encoded protein. Additional determinants of RNAi-induced insect mortality can also depend on the number of gene paralogs with substituting functions, gene expression patterns, and other factors. Based on these technical considerations, the most reliable method for identification of RNAi targets that are highly lethal on short timescale may maize genotype Construct # events tested proportion of plants "passed"** SEM     Table 8).
ScIEntIFIc REPORTS | (2018) 8:2061 | DOI:10.1038/s41598-018-20416-y be empirical screening in the pest of interest or, alternatively, in a model insect when pest genomic information is lacking. In this study, we selected 50 T. castaneum target genes, the majority of which were previously identified in iBeetle to be lethal 46 , in order to test if corresponding orthologous RNAi targets are lethal in the maize pest insect D. v. virgifera. The selected genes were first tested in serial dilutions in T. castaneum to identify highly efficacious targets at a low dose of dsRNA. Of the 50 RNAi gene targets, high proportion showed mortality or growth inhibition. Further, the RNAi targets that showed high lethality in D. v. virgifera were also highly effective at low dose in T. castaneum. Thus, pre-screening for high levels of RNAi efficacy in a model insect can greatly increase the probability of finding efficacious RNAi targets in another species. The utility of these gene targets in plant-directed RNAi was demonstrated by the root protection to WCR feeding in maize expressing Rop, RpII140, and dre4 hairpin dsRNA. Not only did several individual T 0 integration events for each RNAi construct show high levels of root protection, each group of these constructs showed efficacy that was significantly different from the negative controls. RNAi targets Rop, RpII140, and dre4 encode polypeptides with essential functions in important biological processes. The current list of plant-expressed dsRNAs that are known to confer root protection in maize includes: 1) vacuolar ATPase subunit A, V-ATPase A 15 , 2) vacuolar protein sorting gene of the (ESCRT-III) Endosomal Sorting Complex Required for Transport-III, Snf7 29 (also called Vps32 or shrub in Drosophila), 3) vacuolar ATPase subunit C, V-ATPase C 34 , and 4) smooth septate junction proteins dvssj1 and dvssj2, which correspond to the orthologs snakeskin (ssk) and mesh, respectively 36 . Interestingly, RNAi target relationships become apparent even among the small number of root-protective RNAi targets for D. v. virgifera. Like Snf7 and V-ATPase A, Rop is involved in vesicular traffic within the cell, while RpII140, and dre4 are involved in transcription 47,48 . More specifically, dre4 (homolog of SPT16) is part of the FACT (facilitates chromatin transactions) complex and acts as a  49,50 . RpII140 is part of the DNA-directed RNA polymerase II subunit that catalyzes the transcription of DNA into RNA 51,52 . The Drosophila Ras opposite (Rop) is a homolog of the yeast Sec. 1 protein (also known as Unc-18 from C. elegans or the rat Munc-18/n-Sec. 1/rbSec. 1 gene) and essential for vesicle trafficking and membrane fusion 53 . These housekeeping processes are essential to cell viability and represent opportunities to source additional genes for plant-delivered or environmentally-supplied RNAi for insect control. To determine if the observed lethal effect of the identified RNAi target genes identified in T. castaneum and validated in D. v. virgifera can also induce RNAi in another coleopteran pest, we selected an important oilseed rape pest M. aeneus. As RNAi has not yet been demonstrated in M. aeneus, we first analyzed transcriptomic data for RNAi pathway genes 54 . All major RNAi pathway genes (siRNA, miRNA, and piRNA) that we searched for are present in M. aeneus, supporting a functional RNAi mechanism. As M. aeneus bioassays relied on field-collected beetles, RNAi experiments were limited to the four most active target genes from the above D. v. virgifera assays: Rop, dre4, ncm, and RpII140. Both injection and feeding of dsRNA led to high mortality rates in M. aeneus. The injection of dre4 dsRNA showed the most rapid effect. Subsequent quantitative qPCR further confirmed the suppression of the targeted mRNA levels. A significant level of transcript knockdown,, within mRNA extracted from whole beetles, suggests a systemic RNAi response in M. aeneus. Further, the feeding responses of M. aeneus demonstrate an environmental RNAi response in this pest. In total, these observations revealed a functional RNAi pathway, oral/environmental and systemic RNAi responses in M. aeneus, and confirmed the sequence-specific sensitivity for targets identified in T. castaneum.
High mortality rates observed in the M. aeneus feeding bioassay indicate a clear potential of utilizing RNAi as alternative control method for M. aeneus. A possible commercial approach for pollen beetle protection is via topical application by spraying dsRNA onto the host plants. Studies in the Colorado potato beetle (Leptinotarsa decemlineata) have shown that foliar application of dsRNA mediated plant protection that last for at least 28 days 55 , making this approach viable for further development. However, unlike L. decemlineata, the larval stages of M. aeneus reside inside the flower buds and might be not be exposed to typical foliar application. Thus, the development of plants expressing dsRNA constructs provides a promising alternative.
As RNAi targeting of genes described here was effective in three different coleopteran species, additional coleopteran pests might also be sensitive to their knockdown (e.g., other Diabrotica species, Colorado potato beetle (Leptinotarsa decemlineata), or wireworms (Coleoptera: Elateridae). The high level of identity at amino acid level across most of these proteins, with few or no 21-mer nucleotide identities, suggests that the RNAi response in Coleoptera may be generalized to the target type but is very specific to a target sequence 44,45 . We were able to identify gene orthologs in the preliminary transcriptome of M. aeneus that showed a robust lethal RNAi response. This enabled the transcript to be scanned for specific dsRNA trigger sequences devoid of potential off-target transcript matches, which can be as short as < 100 bp 16,34 . Taken together, the model organism screen approach in Tribolium followed by further refinement of potent RNAi targets in Diabrotica has enabled the selection of robust and specific RNAi targets for the difficult-to-work-with pollen beetle pest insect.

Conclusions
This research identified in the model insect T. castaneum four novel RNAi targets that induce high mortality in D. v. virgifera feeding bioassays. We used all four target genes for the production of transgenic dsRNA crops, three of which show root protection against D. v. virgifera larvae by RNAi. Moreover, we showed, for the first time, functional RNAi in field collections of the agricultural pest M. aeneus. Our data demonstrate the feasibility of RNAi applications for combating pollen beetle and potentially other pests. Further, the identified gene targets can be used to better understand RNAi responses in non-pest insects. Taken together, lethal RNAi target genes identified in T. castaneum showed high activity in two additional coleopteran pest species that can pave the way to a new generation of species-specific plant protection. The principle of target identification that we explored in this study may be applied to other pest insects with the goal of developing insect resistance strategies.

Methods
Insect rearing. Wild-type T. castaneum San Bernardino beetles were maintained as described previously 56 .
Adult M. aeneus were collected from fields with flowering Brassica napus plants in the surroundings of Giessen and kept on greenhouse-grown rape plants in a climate chamber with a 16-hour photoperiod and a day/night temperature of 24/18 °C. Non-diapausing D. v. virgifera eggs were purchased from Crop Characteristics, Inc. (Farmington, MN). D. v. virgifera eggs were washed from soil with water and surface-sterilized with 10% formaldehyde for three minutes 57 . The eggs were then rinsed with water and hatched on artificial diet at 28 °C, as described previously 58,59 . D. v. virgifera total RNA was isolated and purified from approximately 0.9 g of whole first-instar larvae by following TRI REAGENT-Protocol (Sigma, St. Louis, MO, USA). The TruSeq stranded mRNA library preparation kit (Illumina, San Diego, CA, USA) was used to prepare mRNA libraries for 1 st instar D. v. virgifera by following instructions from the manufacturer's recommended protocol. In brief, poly-A containing mRNA was purified from 4 µg of total RNA using poly-T oligo-attached magnetic beads. Purified mRNA fragments were subsequently fragmented into smaller pieces (about 300 bp average length) using divalent cations under elevated temperature. Next, SuperScript II reverse transcriptase (Thermo Fisher Scientific, Waltham, MA, USA) and random primers were used to copy the mRNA into first strand cDNA. The cDNA was further converted into double-stranded cDNA (ds cDNA). These ds cDNA fragments then underwent A-tailing and then ligation to indexed Illumina adapters. Lastly, adapter-ligated library products were cleaned up and enriched with 15 cycles of PCR and purified. The purified, enriched libraries were normalized to 2 nM concentration, denatured with sodium hydroxide, and diluted in hybridization buffer. Paired-end sequencing (2 × 151) was carried out on Illumina HiSeq. 2000 according to Illumina's recommended protocol, yielding 87 million reads.

RNA extraction, library preparation and next-generation sequencing. Total
Transcriptome assembly and gene expression analysis. T. castaneum raw reads from Illumina HiSeq were processed by CASAVA software (Illumina) for demultiplexing and removal of the primers attached to the reads. Fastq-mcf was used to trim the adaptors attached to the reads. The trimmed reads ranged from 18 to 30 bp, and the 20-24 bp reads were considered potential siRNAs for further analysis. These filtered reads were mapped to transgene sequence using Bowtie software with no mismatch allowed (Langmead et al. 2009). The mapped 21-and 24-nt reads were visualized using Integrative Genomics Viewer software (Broad Institute, Cambridge, MA, USA).
The transcriptome sequence data generated by Vogel, et al. 45 was used for the identification of M. aeneus genes. Briefly, paired-end reads (2 × 100 nt) were acquired from Illumina HiSeq. 2500 with the error rate < 0.001 for 88% of bases. Quality of obtained reads was checked by fastQC (0.11.4). Trimming was performed by Trimmomatic v.0.36 (parameter: slidingwindow:4:5; leading:5; trailing:5; minlen:25). Sequences shorter 25 bp were discarded. The transcriptome de novo assembly was performed using Trinity (v.2.3.2.). Various assembly combinations were performed and analyzed by transrate (v1.0.3.). Resulting transcripts were aligned with the NCBI NR database by BLASTX search with an E-value cutoff of 1 × 10 −4 . The resulting BLAST hits were processed using Blast2GO software to classify transcripts into GO term categories, including molecular function, biological process, and cellular component. Additionally, the transcripts were translated in all six frames by transeq (EMBOSS package) and aligned by BLAST to the COG database with minimum protein identity of 50% and a protein coverage of at least 75% and lower than 125%.
To calculate gene expression levels and significance of expression differences, pairwise comparisons were performed in Cuffdiff, which is part of the Cufflinks package (2.2.1). Cuffdiff was used with geometric normalization and a threshold criteria for a false discovery rate (FDR) of 0.01. The expression levels were expressed as log2-fold-change of Fragments Per Kilobase per Million mapped reads (FPKM)-normalized count data.
D. v. virgifera raw sequencing data was processed using fastq-mcf to remove adaptors and low quality sequencing data with Q30 cut off. The transcriptome de novo assembly was performed on trimmed reads using Trinity (v.2.0.2) that generated 69,840 transcripts. This de novo transcriptome is used for further analyses. D. v. virgifera and M. aeneus. Based on the genome sequence of T. castaneum ver.Tcas5.2, ortholog proteins of M. aeneus were identified by using NCBI BLASTP with an E-value of 0.01. The output were further filtered by a minimum protein identity of 50% as well as a protein coverage of at least 75% and lower than 125%. Resulting hits were ranked by score; redundant and overlapping sequences were removed.

Identification of T. castaneum gene orthologs in
To identify T. castaneum proteins in D. v. virgifera, TBLASTN searches using candidate protein coding sequences were run against BLASTable databases containing the unassembled D. v. virgifera sequence reads or the assembled contigs. Significant hits to a D. v. virgifera sequence (defined as lower than <1 × 10 −20 for contig homologies and better than E-value of <1 × 10 −10 for unassembled sequence reads homologies) were confirmed using BLASTX against the NCBI non-redundant database. The results of this BLASTX search confirmed that the D. v. virgifera homolog candidate gene sequences identified in the TBLASTN search indeed comprised D. v. virgifera genes, or were the best hits to the non D. v. virgifera candidate gene sequence present in the D. v. virgifera sequences. In most cases, T. castaneum candidate genes, which were annotated as encoding a protein, showed unambiguous sequence homology to a sequence or sequences within the D. v. virgifera transcriptome. In few cases, partially-overlapping contigs were assembled into longer contigs using Sequencher ™ v4.9 (Gene Codes Corporation, Ann Arbor, MI).

Identification of genes involved in RNAi pathway. Identification of core RNAi and potential systemic
RNAi genes in T. castaneum was based on findings of Tomoyasu et al. 54 . The reference set of 17 proteins was aligned to the assembled and translated transcripts of M. aeneus and D. v. virgifera by BLASTP. Proteins with an identity greater than 50% and a minimum coverage of 75% were considered as homologs. Additionally, domain architecture was analyzed by ScanProsite and hits with equal or higher profile score than 10.0 indicated a domain occurrence. Hits with alignment scores below 8.5 are usually common, but were regarded as questionable and therefore excluded from further analysis.  Table 1) were generated with gene-specific RNAi primers including the T7 promoter sequence (TAATACGACTCACTATAGGGAGA) at the 5′, purchased from Sigma-Aldrich (St Louis, MO, USA). Ambion MEGAscript T7 kit (Thermo Fisher Scientific, Waltham, MA) was used to prepare dsRNA according to the manufacturer's protocol.
T. castaneum RNAi design. Unassembled D. v. virgifera sequence reads or the assembled contigs identified to contain either T. castaneum RNAi target orthologs or RNAi pathway targets were annotated with the location of the open reading frame (ORF) for ortholog based on BLASTX results from NCBI non-redundant database. Using the ORF location dsRNA sequence was designed to be between 200 and 500 base pairs with a %GC between 40 and 60 and a distance from the ATG and stop codon of greater than seventy base pairs.
Total RNA was extracted from D. v. virgifera eggs, larvae, and adults using the TRIzol (Life Technologies, Grand Island, NY) isolation method according to the manufacturer's instructions. cDNA was synthesized from 1 µg of total RNA using a SuperScript III reverse transcription kit (Thermo Fisher Scientific, Waltham, MA). Oligonucleotide primers for PCR template production were designed using VectorNTI (Invitrogen, Carlsbad, CA) or Primer3 software and contained a T7 promoter sequence at their 5´ ends. dsRNAs were synthesized using Ambion MEGAscript T7 (Thermo Fisher Scientific, Waltham, MA) dsRNA synthesis kit, according to the manufacturer's protocol. Sequences within the open reading frame of up to approximately 500 bp were selected for the initial dsRNA bioassay. The sequences of D. v. virgifera dsRNA amplicons appear in Supplementary Table 1. Synthesized dsRNA was quantified on NanoDrop 8000 spectrophotometer (Thermo Fisher Scientific, Waltham, MA) and diluted in 0.1 × TE to a working concentration of 12.5 ng/µl. Injection bioassays. T. castaneum larvae and M. aeneus adults were anaesthetized on ice before affixation to double-stick tape. Dorsolateral injection of 150 nl dsRNA [250 ng/µl] into each of the insects was performed using a pulled glass capillary (Item No: 504949, World Precision Instruments, Sarasota, FL) and the micromanipulator M3301 (World Precision Instruments, Sarasota, FL) under a dissecting stereomicroscope (n = 12, three replicates). Negative controls received an equivalent amount of water or dsRNA corresponding to the Galleria mellonella metalloproteinase inhibitor (IMPI) gene (AY330624), which is absent in beetles. T. castaneum larvae were kept on whole-grain flour with 5% yeast powder after injection, whereas M. aeneus beetles were kept in petri dishes with dried pollen and wet tissues as food and water sources. Survival rates were monitored and counted every 48 h for 14 days.
Artificial diet bioassays. M. aeneus adults were kept without water for 24 h before treatment to ensure that beetles drank from a droplet of 5 µl dsRNA solution (1 µg/µl). After visual examination of dsRNA solution uptake, beetles were transferred to petri dishes with artificial diet and a wet tissue. The water-based diet (1% gelatin and 50% homogenized pollen) was mixed with dsRNA to a final concentration of 500 ng/cm 2 , to ensure continuous uptake of dsRNA (n = 10, three replicates). Negative controls received an equivalent amount of water or dsRNA corresponding to the Galleria mellonella metalloproteinase inhibitor (IMPI) gene (AY330624). As the recipe did not contain any antibiotics or antimycota, the diet was exchanged every two days to avoid fungal contamination. Mortality rates were checked every two days.
The D. v. virgifera feeding bioassays were conducted with neonate larvae (two to three larvae per well) in 128-well plastic bioassay trays (BIO-BA-128, C-D International, Pitman, NJ) that contained 1.5 ml of an artificial diet 60 . DsRNA in 0.1 × TE was surface-applied at 500 ng/cm 2 . Trays were sealed with Pull N' Peel Tab vented adhesive plastic sheets (BIO-CV-16, C-D International, Pitman, NJ) and held at 28 °C, ~40% Relative Humidity. The total number of insects exposed to each sample, the number of dead insects, and the weight of surviving insects were recorded after nine days. DsRNA targeting yellow fluorescent protein gene (YFP), 0.1XTE buffer, and water were used as negative controls. Growth Inhibition (GI) was calculated based on the average weights of all controls, as follows: = − GI [1 (TWIT/TNIT)/(TWIBC/TNIBC)], where TWIT is the Total Weight of live Insects in the Treatment; TNIT is the Total Number of Insects in the Treatment; TWIBC is the Total Weight of live Insects in the Background Check (negative control); and TNIBC is the Total Number of Insects in the Background Check (negative control). To assess the potency of active RNAi targets, four-fold serial dilutions of dsRNAs were bioassayed and the LC 50 (concentration at which 50% of the insects are dead) and GI 50 (concentration that causes 50% growth inhibition or GI) values were calculated using log-logistic regression analysis in JMP Pro from SAS Institute Inc; raw data appears in Supplementary  -1 v1, dre4-1 v2, ncm-1 v2, Rop-2 v3, RpII140 v1, and RpII140 v2 contained no 21-mer or longer matches to A. mellifera, B. terrestris and M. musculus.
Standard cloning methods were used in the construction of Gateway-enabled entry vectors (Invitrogen, Carlsbad, CA) containing RNAi hairpin expression cassettes. Hairpins were designed to include target sense and antisense sequences separated by a flexible linker. Expression of the hairpin was driven by the promoter from the maize ubiquitin 1 (Ubi-1) gene 62 and terminated by the 3′ untranslated region of the maize peroxidase 5 gene 63 . Each hairpin containing entry vector was recombined using Gateway technology (Invitrogen, Carlsbad, CA) with a destination vector harboring a selectable marker cassette to create binary vectors for maize transformation.

Development of transgenic plants. Binary expression vectors, based on pTI15955 plasmid from
Agrobact-erium 64 , contained hpdre4-1 v1, hpdre4-1 v2, hpRop-2 v3, hpncm-1 v2, hpRpII140 v1, and hpRpII140 v2 hairpins. Each of these plasmids were transformed into Agrobacterium tumefaciens strain DAt13192 (RecA-deficient ternary strain) 65 . Colonies were selected and plasmid DNA was isolated and confirmed via restriction enzyme digestion. Each binary vector was separately transformed into maize via Agrobacterium-mediated transformation of immature embryos isolated from the inbred line, Zea mays c.v. B104 using conventional methods with modifications 66 . Briefly, the immature embryos were incubated with a suspension containing Agrobacterium cells and surfactant and then were moved to solid medium plates followed by co-cultivation for 3-5 days. The treated embryos were transferred onto a medium containing antibiotics and compounds for selective isolation of genetically transformed corn tissues. The corn tissues were grown on selection medium until plants were regenerated.
Transgene copy number analysis. The described binary destination vector contained an herbicide tolerance gene (aryloxyalknoate dioxygenase; AAD-1 v3), under the expression regulation of a maize Ubi-1 promoter and a fragment containing a 3′ untranslated region from a maize lipase gene (ZmLip 3'UTR). DNA Real-time PCR analyses to detect a portion of the AAD1 coding region in gDNA were used to estimate transgene insertion copy number. The AAD levels were compared to the levels of single-copy native gene. Simple events (having one or two copies of transgene insertions) were selected for greenhouse bioassay. Additionally, PCR assays designed to detect a portion of the spectinomycin-resistance gene (SpecR; from the binary vector plasmids outside of the T-DNA) were used to determine if the transgenic plants contain extraneous integrated plasmid backbone. Samples for these analyses were collected from plants grown in environmental chambers at the V2-V3 growth stage. Maize leaf pieces approximately equivalent to two leaf punches were collected in 96-well collection plates (QIAGEN). Tissue disruption was performed with a KLECKO ™ tissue pulverizer (Garcia Manufacturing, Visalia, CA) in BioSprint 96 AP1 lysis buffer (supplied with a BioSprint 96 DNA Plant Kit; QIAGEN) with one stainless steel bead. Following tissue maceration, gDNA was isolated in high throughput format using a BioSprint 96 DNA Plant Kit and a BioSprint 96 extraction robot. gDNA is diluted 2:3 DNA:water prior to setting up the qPCR reaction. qPCR analysis. Transgene detection was performed by hydrolysis probe real-time quantitative PCR assay. Primers and probe to detect a portion of the SpecR gene (SPC1) and a segment of the AAD-1 herbicide tolerance gene (GAAD1) appear in Supplementary Table 10. Assays were multiplexed with reagents for an endogenous maize chromosomal Invertase gene (IVR1, GENBANK Accession No: U16123), which served as an internal reference sequence to ensure gDNA is present in each assay. PCR amplification was set up using LightCycler 480 Probes Master mix (Roche). PCR was as performed on LightCycler 480 Instrument (Roche) using fluorophore activation and emission for the FAM-and HEX-labeled probes. Cp scores (the point at which the fluorescence signal crosses the background threshold) are determined from the real time PCR data using the fit points algorithm (LightCycler 480 Software, Version 1.5) and the Relative Quant module (based on ΔΔC T method).
Root protection assays. The whole plant maize bioassays were conducted by following the protocol described in Dönitz, et al. 46 . In brief, the transgenic corn plants (T 0 , one plant per event) were planted into root trainer pots containing Metromix soil after reaching V2 or V3 stage. The plants were infested with 125-150 D. v. virgifera eggs and allowed to grow for two weeks. Two weeks after infestation, the plant roots were washed and rootworm feeding damage was scored using node-injury scale (NIS) ranging from 0 to 1 as compared to 0 to 3 described by Oleson, et al. 67 . Event DAS59122-7, expressing Cry34Ab1/Cry35Ab1, served as Bt positive control. The negative controls included non-transformed B104, B104 plants expressing either yfp hairpin dsRNA or YFP protein, and non-transgenic isoline of DAS59122-7, 7SH382. The constructs expressing RNAi targets were bioassayed at different times with both positive and negative controls included in each experiment.
Statistical analysis. Analyses of variance (ANOVA) for T. castaneum and M. aeneus bioassays were followed by a Holm-Sidak test with significance threshold of p < 0.05 using Daniel's XL toolbox for Excel, version 7.2.10 68 . Each experiment was compared to a control group and all experiments were conducted independently at least three times. For D. v. virgifera experiments, means comparisons were performed for all pairs using Tukey-Kramer HSD method in JMP Pro 11.1.1 (SAS, Cary, NC).
The T 0 root damage rating data are not normally distributed; hence they were converted into categorical data which follows binomial distribution. All T 0 events and control plants that showed a root rating of ≤ 0.5 were designated as "pass" and the events with root ratings > 0.5 to 1.0 were called "fail". To identify the constructs that provided better root protection, the proportion of plants which passed the bioassay was analyzed with the generalized linear mixed model procedure 69