Spodoptera

We built OBP and OR neighbor-joining trees based on Lepidoptera data sets. The OBP data set contained the 43 complete amino acid sequences deduced from the genome of B. mori, together with the largest OBP repertoires characterized within noctuid moths (7 sequences from H. virescens and 5 from Spodoptera exigua) and outside noctuid moths (13 from M. sexta and 4 from Plutella xylostella) (Accession numbers available in additional file 6). Signal peptide sequences were removed following predictions of cleavage site location made by SignalP 3.0. The OR data set contained 58 amino acid sequences from B. mori (6 sequences were removed from the alignment because of their short length) and the 21 sequences characterized from H. virescens, completed with subsets of sequences characterized within noctuids (3 sequences from Mythimna separata) and outside noctuids (4 from P. xylostella, 3 from Diaphania indica and 3 from E. postvittana). Amino acid sequences were aligned using ClustalW2 [87 (link)]. Unrooted trees were constructed by the neighbour-joining method, with Poisson correction of distances, as implemented in MEGA4 software [88 (link)]. Node support was assessed using a bootstrap procedure base on 1000 replicates, and nodes supported by a bootstrap value under 70% were collapsed to an horizontal line when drawing cladograms.

Upon retrieval of demultiplexed sequencing data, fastq files were trimmed of adapters using trim_galore \—paired—small_rna—dont_gzip—quality 0—length 26/34 (trim_galore 0.4.4 https://github.com/FelixKrueger/TrimGalore). Note that the—length parameter was set to 26 for the Towne IE2F PRO-Seq library that was prepared with 4N UMI RNA adapaters and 34 for the remaining libraries that used 8N UMI RNA adapters. This change results in the minimum processed read length being 18 bp regardless of the length of the UMI. For alignment of reads, two different concatenated genomes were utilized. For Towne datasets, an HCMV Towne genome sequence (GenBank accession number FJ616285.1) [66 (link)] concatenated to the human genome sequence (UCSC assembly hg38) was utilized. For TB40/E datasets, an HCMV TB40/E genome sequence (GenBank accession number KF297339.1) [67 (link)] concatenated to Spodoptera genome (WGS number JQCY02) and human genome (UCSC assembly hg38) sequences were utilized. The concatenated genome fasta files were used to construct a bowtie index using the bowtie-build -f function. Datasets were aligned using bowtie \—trim5 4/8—trim3 4/8—minins 26/34—maxins 600—fr—best—allow-contain—sam—fullref—threads 80—chunkmbs 5000. The—trim5,—trim3 and–minins parameters were decided by the length of the UMIs on the RNA adapters. Output .sam files were then deduplicated using dedup -n 4/8 -f -p where -n specifies the length of the UMI on each RNA adapter (https://github.com/P-TEFb/dedup). Mapped, deduplicated reads in output .bed files were then parsed out based on the genome to which they aligned and finally converted to bigwigs using the bedtools [68 (link)]. The Towne PRO-Seq datasets were normalized to total deduplicated reads mapping to the viral genome (control and Flavo groups were normalized separately). For datasets that incorporated spike-in controls, data were normalized as described in Ball et. al. [69 ]. Refer to S1 Table for details about the calculations performed to normalize the data. Normalization factors were applied via multiplication by genome coverage values in bedGraph files. PRO-Seq reads aligning to annotated HCMV Towne (FJ616285.1) [66 (link)] and TB40/E (KF297339.1) [67 (link)] genomes were viewed in the UCSC Genome Browser [70 (link)] and assembled in Towne and TB40/E track hubs [71 (link)], respectively. Both hubs have the open reading frame annotations identified by ribosomal profiling data generated human fibroblasts infected with the HCMV Merlin strain [47 (link)]. The additional ORFs identified by this method were located in Towne and TB40/E genomes using Snapgene and represented in the hub as a bigbed file. The TB40/E hub also contains HCMV Merlin mRNA-Seq data at 72 h pi (GSE41605). The two replicates of the Merlin mRNA-Seq datasets (GSM1020246 and GSM1020249) were aligned to the TB40/E genome. The SRA files for these datasets were downloaded using fastq-dump Afterwards, and trim galore was used in succession to remove–-adapter CTGTAGCCACCATC and—adapter AAAAAAAAAA. The trimmed fastq files from the two replicates were concatenated. Afterwards, a hisat2 index for the TB40/E genome was constructed using hisat2-build -f. The trimmed, concatenated fastq file was then mapped to TB40/E using hisat 2 \ -q–threads 80. The output .sam file was converted to a bigwig for display on the TB40/E hub.

Spodoptera littoralis colonies are maintained at ICA-CSIC, reared on artificial diet, and kept at 22 ± 1 °C and >70% RH, with a photoperiod of 16:8 h (L:D) in a growth chamber. The bioassays were conducted as described [52 (link)]. The upper surface of Capsicum anuum (1.0 cm²) were treated with 10 μL of the test substance. The products were tested at an initial dose of 5 µg/µL (50 µg/cm²). Five Petri dishes with two sixth-instar S. littoralis larvae (>24 h after moulting) were allowed to feed in a growth chamber (until 75% larval consumption of control disks or tretament disks, environmental conditions as above). Each experiment was repeated 2 times. Feeding inhibition was calculated by measuring the disk surface consumption (digitalized with https://imagej.nih.gov/ij/, accessed on 23 March 2020) as % FI = [1 − (T/C) × 100], where T and C represent feeding on treated and control leaf disks, respectively. The antifeedant effects (% FI) were analyzed for significance by the nonparametric Wilcoxon signed-rank test. Extracts and compounds with an FI > 70% were further tested in a dose-response experiment (4–5 serial dilutions) to calculate their relative potency (EC₅₀, the effective dose to give a 50% settling reduction) from linear regression analysis (% FI on Log-dose).

The eggs of common cutworm (Spodoptera litura) were purchased from Sumika Technoservice (Hyogo, Japan) and were hatched in the laboratory. The hatched neonates were reared on an artificial diet (Insecta LFS, Nosan Co., Kanagawa, Japan) until they were used for experiments in a biotron under a controlled temperature of 25 ± 2 °C and a 16 h photoperiod of fluorescent light. To assay insect performance on plants, S. lycopersicum or the HexVic-deficient IL11-1 plants were challenged with the third instars for 7 days. To assay insect performance on the artificial diet, newly hatched neonates were fed the diet mixed with HexGlc at 0.25 μg·g^–1, which is close to the average amount of HexVic detected in (Z)-3-hexenol-exposed tomato plants. The insects were weighed on a precision balance 7 days after the experiments.