Parasitic insect-derived miRNAs modulate host development

Parasitic wasps produce several factors including venom, polydnaviruses (PDVs) and specialized wasp cells named teratocytes that benefit the survival of offspring by altering the physiology of hosts. However, the underlying molecular mechanisms for the alterations remain unclear. Here we find that the teratocytes of Cotesia vestalis, an endoparasitoid of the diamondback moth Plutella xylostella, and its associated bracovirus (CvBV) can produce miRNAs and deliver the products into the host via different ways. Certain miRNAs in the parasitized host are mainly produced by teratocytes, while the expression level of miRNAs encoded by CvBV can be 100-fold greater in parasitized hosts than non-parasitized ones. We further show that one teratocyte-produced miRNA (Cve-miR-281-3p) and one CvBV-produced miRNA (Cve-miR-novel22-5p-1) arrest host growth by modulating expression of the host ecdysone receptor (EcR). Altogether, our results show the first evidence of cross-species regulation by miRNAs in animal parasitism and their possible function in the alteration of host physiology during parasitism. The moth Plutella xylostella during its larval stage is the host of the endoparasitic wasp Cotesia vestalis. Here the authors show that the parasitoids deliver microRNAs to their hosts through their symbiotic virus and specialized cells leading to induced developmental delay.

P arasitic insects, particularly parasitic wasps, are a large group of animals that lay eggs in or on the bodies of other arthropods that serve as hosts for offspring development. More than 20% of all insect species are parasitic wasps, making them among the most species-rich animal groups on Earth 1 . Successful development of wasp offspring usually results in death of the host. As such, many parasitic species are used as biological control agents for management of pest species. The moth Plutella xylostella (order Lepidoptera: family Plutellidae) is a worldwide pest of cruciferous plants 2 and its larval stage is the natural host of the endoparasitic wasp Cotesia vestalis (family Braconidae). Many parasitic wasps kill their hosts by producing effector molecules that disable growth and immune defences. When laying eggs, C. vestalis introduces into P. xylostella effectors that derive from three sources: (1) venom produced in a venom gland, (2) a symbiotic polydnavirus named C. vestalis bracovirus (CvBV), which expresses virulence genes after infecting host cells, and (3) cells called teratocytes that are produced during embryogenesis and released into the body of the host when wasp eggs hatch [3][4][5][6] .
Most known effector molecules produced by C. vestalis and other parasitic wasps are proteins. However, eukaryotes including insects and some viruses produce microRNAs (miRNAs) that are small (20-22 nts) non-coding RNA (snRNA) 7 . miRNAs regulate a range of cellular processes by repressing translation or promoting the decay of target transcripts through imperfect pairing with complementary sites in 3′-untranslated regions (UTRs) 7 . Several intracellular pathogens promote their own survival by modulating the miRNAs produced by hosts 8 . Some non-viral microbial pathogens also produce snRNAs that target host genes, while some hosts produce miRNAs that affect pathogen gene function [9][10][11] . However, it is unknown whether parasitic animals such as insects produce miRNAs that affect hosts.
Recent sequencing of the C. vestalis and P. xylostella genomes provide the references needed for identifying C. vestalis-produced miRNAs and assessing their roles in parasitism of P. xylostella. Here, we report that C. vestalis teratocytes express several miR-NAs and CvBV encodes several miRNAs that are expressed in infected host. We further show that one teratocyte-produced miRNA and one CvBV-produced miRNA arrest host growth by modulating expression of the host ecdysone receptor (EcR).

Results
Cotesia vestalis encodes multiple predicted miRNAs. snRNA libraries were constructed from C. vestalis larvae and teratocytes. Sequencing yielded 12 (larval) and 15 million (teratocyte) raw reads, of which 74.6% mapped to the C. vestalis genome after quality filtering. In both libraries, read distributions showed two peaks at 22-23 nt and 28-29 nt, a typical pattern in most insect small RNA libraries ( Supplementary Fig. 1) 12 . In total, 38% of the mappable reads in the C. vestalis larval libraries and 10.2% of the mappable reads in the teratocyte libraries also mapped to miR-NAs/pre-miRNAs in miRBase (Supplementary Table 1). Prediction software identified a total of 35 pre-miRNAs specifically expressed in larvae, 70 specifically expressed in teratocytes, and 71 expressed in both (Fig. 1a, Supplementary Data 1). Copy number distributions indicated the most abundant predicted miRNAs in larvae and teratocytes were 21-23 nt in length (Supplementary Tables 2 and 3). More than 80% of the annotated pre-miRNAs were predicted to produce both 5p and 3p miRNAs, while the remainder were predicted to produce mature miRNAs from just one arm (Fig. 1b). More than 90% of the mature miRNAs identified in C. vestalis were classified as conserved or highly conserved homologs of miRNAs in miRBase, while 7% were classified as novel (Fig. 1b). Read counts normalized to the numbers of reads in each library and expressed as reads per million (RPM) revealed highly variable abundance of miRNAs (Supplementary Data 1). In larvae, the six most abundant miR-NAs were miR-1-3p, miR-184-3p, miR-275-3p-1, miR-2944, let-7, and miR-8, while in teratocytes they were miR-375-3p, miR-8,  miR-184-3p, miR-275, miR-7929-3p, and miR-3405-3p (Supplementary Data 1). Read mapping to the C. vestalis genome showed that 71% of miRNAs were intergenic while 55% were clustered (Fig. 1b).
CvBV-derived miRNAs are expressed in the parasitized hosts. Like all polydnaviruses (PDVs), CvBV persists as a provirus in the genome of all C. vestalis cells, while replication only occurs in ovary calyx cells of females 3  miR-bantam-3p-4 miR-bantam-5p-4 miR-bantam-5p-4 miR-n20-3p miR-n20-3p miR-n20-5p miR-n20-5p miR-n21-3p miR-n21-3p miR-n21-5p miR-n21-5p miR-n22-3p miR-n22-3p miR-n22-5p miR-n22-5p miR-n23-3p miR-n23-3p miR-n23-5p miR-n23-5p Results shown are mean relative abundance ± s.e.m. of each miRNA from three independent biological replicates P. xylostella larvae from three treatments: (1) 16 h post oviposition by C. vestalis, a time point when host tissues are infected by CvBV but parasitoid eggs have not hatched, (2) 120 h post oviposition when host tissues are infected by CvBV and teratocytes are also present, and (3) non-parasitized larvae (negative control (NC)). Results showed that some miRNAs were more abundant at 16 h while others were more abundant at 120 h ( Fig. 2b-h). In the non-parasitized host larvae, none of these miRNAs were detected ( Fig. 2b-h). These results indicated that several CvBVencoded miRNAs emerge in parasitized host larvae through viral infection of host tissues, expression in teratocytes, or both. To assess whether CvBV-encoded miRNAs are expressed in CvBVinfected host tissues, CvBV virions were collected from female wasps as previously described 16 . The CvBV collected from the reproductive tract of a single adult female is defined as one female equivalent (FE). We then either injected 0.1 FE of CvBV particles into individual third instar host larvae or added 20 FE of CvBV particles to cultures of Pxem_ZJU cells, which is an established P. xylostella cell line derived from embryos. Abundance of the 20 CvBV-encoded miRNAs was then assessed in each sample by real-time quantitative PCR (RT-qPCR) ( Fig. 2i and j). Primers designed to amplify miR-novel24-3p, miR-novel24-5p, and miR-novel25-5p yielded inconsistent results and were excluded from further analysis. For the remaining 17 miRNAs, all but miR-novel22-5p were detected at 24 h post oviposition at greater relative abundance than the P. xylostella U6 snRNA, which served as the internal normalizing control (Fig. 2i). Similar results were observed in CvBV-infected Pxem_ZJU cells, albeit relative abundance of several miRNAs was substantially higher in comparison to the U6 snRNA normalizer. Altogether, these data support the expression of CvBV-encoded miRNAs in CvBVinfected hosts.
Teratocytes deliver miRNAs into the host. CvBV-encoded miRNAs expressed in infected host tissues have the potential to interact directly with host gene products, whereas miRNAs expressed in C. vestalis teratocytes would have to enter host cells after secretion. To investigate whether miRNAs expressed in teratocytes are secreted, we collected teratocytes from parasitized P. xylostella larvae and placed them into primary culture followed by RT-qPCR analysis of the 30 most abundant miRNAs detected in snRNA teratocyte libraries. Primers specific to the C. vestalis β-tubulin were used as a control for detection of teratocyte contamination in sample media ( Supplementary Fig. 2a). Ct values indicated that each miRNA was detected at greater abundance in teratocytes than in the medium used to culture teratocytes (Fig. 3a). However, the ratio of abundance for selected miRNAs including miR-34-5p and miR-275-5p-2 were less dissimilar between teratocytes and conditioned medium, suggesting that these factors are potentially secreted into the medium at higher abundance than the other most abundant miRNAs we detected in teratocytes (Fig. 3a). Since extracellular miRNAs are often associated with exosomes, we examined uptake of fluorescently labeled exosomes by the Pxem_ZJU cell line, using methods described by Bruns et al. 17 Results indicated uptake of teratocyte-produced exosomes by Pxem_ZJU cells (Fig. 3b). We then assessed whether any of the 30 most highly expressed miRNAs in teratocytes were present in Pxem_ZJU cells after coculture with teratocytes for 6 h followed by removal of teratocytes and processing of Pxem_ZJU cells for RT-qPCR analysis. PCR analysis of C. vestalis 18S RNA was first used to confirm that RNA isolated from Pxem_ZJU cells after co-culture and separation from teratocytes contained no C. vestalis RNA (Supplementary Fig. 2b). RT-qPCR analysis then showed that 20 teratocyte-miRNAs, including miR-375-3p, miR-281-3p, and miR-9-5p-1, were more abundant in Pxem_ZJU cells that were co-cultured with teratocytes than in control Pxem_ZJU cells that were cultured without teratocytes (Fig. 3c). We further noted that miR-375-3p, miR-281-3p, and miR-9-5p-1 are C. vestalis miR-NAs that are not encoded in the CvBV genome. Since many miRNAs are conserved between species, we next compared miRNAs identified from C. vestalis to the inventory of miRNAs previously identified from P. xylostella (Fig. 3d, Supplementary Data 1) 12,18 . The comparison showed that the sequence of miR-9-5p-1 in C. vestalis and that in P. xylostella are the same. In contrast, miR-281-3p and miR-375-3p from C. vestalis and P. xylostella exhibited four and six mismatches, respectively (Fig. 3d). These mismatches allowed us to design primers for distinguishing between the C. vestalis and P. xylostella variants for miR-281-3p and miR-375-3p by cloning and sequencing amplified products. In Pxem_ZJU cells co-cultured with teratocytes, sequencing of a 100 randomly selected miR-281-3p and miR-375-3p clones indicated that 77% and 87%, respectively, were the C. vestalis isoform (Supplementary Fig. 2c and Supplementary Data 1). The same assay using hemocytes collected from P. xylostella at 5 days post oviposition indicated 91% of miR-281-3p and 82% of miR-375-3p were the C. vestalis variants (Fig. 3d). Altogether, these results strongly suggested that miRNAs released from teratocytes were taken up by Pxem-ZJU cells and hemocytes from parasitized host larvae.
Identification of wasp miRNA targets in the host. We predicted candidate target genes for the 30 most abundant miRNAs expressed in teratocytes and/or encoded by CvBV using the 3′-UTRs for all P. xylostella messenger RNA (mRNA) genes with complete coding sequences and three prediction programs: RNAhybrid, Targetscan, and miRanda. We classified P. xylostella genes as high-confidence candidate targets for a given miRNA if predicted by at least two of the prediction programs (Supplementary Data 2). A total of 91 candidate P. xylostella target genes were identified with six teratocytes abundant miRNAs and six CvBVencoded miRNAs having no predicted targets. Among these host genes, two candidates were predicted to be targeted by two or more C. vestalis miRNAs. The targets were the P. xylostella ecdysone receptor (EcR, JN796249.1), which was the predicted target for Cve-miR-281-3p, Cve-miR-31-5p-1, and Cve-miR-novel22-5p, and prophenoloxidase 1 (proPO 1, GU199189.1), which was the predicted target for Cve-miR-750-3p and Cve-miR-31-5p-1. These observations suggested that teratocytes and CvBV may cooperate with each other in producing miRNAs that target the host EcR and proPO 1 genes. The overall list of predicted host gene targets also suggested that C. vestalis miRNAs expressed in teratocytes or encoded by CvBV preferentially interact with host factors associated with immunity (prophenoloxidase and prophenoloxidaseactivating proteinase 3, IMD-like protein), the nervous system (neuropeptide-like peptide, short neuropeptide F, nicotinic acetylcholine receptor alpha 6, glutamate-gated chloride channel), protein phosphorylation (protein phosphatase 2 and 4), apoptosis (caspase-1, -5, -9, apoptosis regulator buffy), and cell adhesion (cadherin-like protein and stromal cell-derived factor 2-like).

Discussion
Our knowledge about host control has been mostly based on functional proteins released by wasp-associated factors, such as venom, PDVs, and teratocytes. In this study, we report that C. vestalis-derived miRNAs are released into P. xylostella from teratocytes or CvBV, and that injection of two wasp-associated miRNAs delayed host molting and metamorphosis.
It has been more than a decade since the first discovery of miRNAs in Epstein-Barr virus 22 . More than 250 miRNAs have since been identified from both RNA and DNA viruses from diverse families 23 . However, no miRNAs have previously been described from PDVs. More than 70% of the encapsidated CvBV genome consists of non-coding sequence 13 . Unlike virulence genes dispersed in different PDV segments, the identified miR-NAs in CvBV reside on only one segment with miRNA homologs similarly residing on a single orthologous segment of CcBV and CsBV 14,15 . Alignment with other BVs further suggests this segment is unique to BVs associated with braconid wasps in the genus Cotesia. The higher number of miRNAs and two duplications of miRNA precursors in CvBV further suggest that alterations have occurred in the miRNA inventory due to divergence of C. vestalis from C. congregata and C. sesimiae. When it comes to the function of CvBV-coding miRNAs, CvBV-derived miRNAs can be grouped into two classes: those that are analogs of host miRNAs (P. xylostella) and those that are bracovirus-specific. For the analogs, the seed sequence is shared with host miRNAs, whereas CvBV miRNAs have potentially evolved to target new sites in hosts or function as regulators of wasp or viral gene expression. Most known functions of virus-encoded miRNAs are involved in controlling host immune responses, prolonging longevity of infected cells and regulating the lytic cycle of infected host cells 23 . Target candidates of CvBV-derived miRNAs suggest roles in modulating host immune responses, protein neurophysiology, and development, which is consistent with the known functions of BV in parasitism of hosts by wasps 5 . In addition, it is not surprising to find that CvBV-derived genes or miRNAs expressed in teratocytes given that all wasp cells contain the CvBV proviral genome. CvBV-derived miRNA in teratocytes may function as a form of dosage compensation for CvBV or act as a signal to initiate CvBV miRNA expression. miRNAs released by cells via extracellular vesicles (EVs, such as exosomes or microviscles) into circulation have been previously reported in studies of vertebrates in the context of serving as biomarkers for disease detection [24][25][26][27][28] . While the release of exocytotic vesicles by teratocytes is known 29 , the production of EVs by teratocytes has not been previously reported. Previous studies have reported the secretion of several proteins by teratocytes 30,31 . Our results add to this literature by supporting that teratocytes also secrete miRNAs. Detection of several teratocytederived miRNAs in the medium of teratocyte primary cultures led to the hypothesis that C. vestalis teratocytes also likely secrete miRNAs into the hemolymph of parasitized P. xylostella larvae. The large number of miRNAs produced by teratocytes are also likely to regulate host in various ways due to the potential of the target candidates.
Insect growth is a complex process involving multiple signal pathway networks. Juvenile hormone and ecdysone are key regulators of insect growth, molting, and metamorphosis. In parasitoid-host interactions, larval endoparasitoids often prolong the host larval stage, reduce body size, and disrupt metamorphosis 3,6 . Recent studies also show that miRNAs are effectors of several genes with roles in hormone signaling pathways 32,33 . In Drosophila, bantam, miR-14, and miR-34 modulate ecdysone signaling [34][35][36] . In lepidopteran insects, let-7 and miR-281 also play regulatory roles in molting and metamorphosis 19,37 . miR-14 from Drosophila and miR-281 from Bombyx mori have been implicated in regulating ecdysone signaling by limiting the expression of EcR 19,36 . In our study, miRNAs from both teratocytes and CvBV share EcR as their common target, implying a role of miRNAs in host developmental regulation. We demonstrated that both teratocytes-derived Cve-miR-281-3p and PDVderived Cve-miR-novel22-5p downregulated EcR gene expression and decreased the production of protein, which correlated with delayed host development. In our experiments, metamorphosis was delayed rather than arrested probably due to the injection of each host larva with only a single dose of each miRNA, whereas these factors are continuously produced in naturally parasitized hosts. Previous reports and our data jointly suggest an evolutionary conserved relationship between miR-281 family and the EcR in Lepidoptera. Thus, other parasitoid species are likely to take advantage of the same strategy to manipulate the development of their lepidopteran hosts. In addition, the newly functional confirmation of Cve-miR-novel22-5p from PDV indicated the convergent evolution of similar function in teratocyte and PDV, but the factors may cooperate in a successive manner to achieve the prolongation of host larval stage and in turn the completion of wasp offspring development.
To the best of our knowledge, this is the first evidence demonstrating the molecular mechanisms by which parasitoid wasps control host development via miRNA. Our findings indicate that endoparasitoids produce miRNAs with roles in altering host growth. Other miRNAs we identified may also have roles in regulating host metabolism and immune responses.

Methods
Insects. C. vestalis and P. xylostella used in the study derived from cultures that had been continuously maintained for 5 years and >100 generations. P. xylostella was reared on cabbage grown at 25°C, 65% relative humidity, and 14 h light: 10 h dark photoperiod. C. vestalis was reared by allowing adult females to parasitize P. xylostella third instars. Adult wasps were fed with 20% (w/v) honey solution.
Cell lines. The strongly adherent P. xylostella Pxem_ZJU cell line, which was derived from the P. xylostella, was maintained in 60 mm culture dish in TNM-FH culture medium plus 10% fetal bovine serum (FBS) at 27°C under ambient atmosphere 38 . The human embryonic kidney cell 293 (HEK293) cell line was bought from Thermo Fisher and maintained at 37°C in 5% CO 2 in DMEM basic medium (Gibco) medium plus 10% FBS. All cell lines were verified to be mycoplasma-free using the PCR-based mycoplasma detection kit (Kaiji, Nanjing, China). Mycoplasma testing was performed on all lines using Mycoplasma Hoechst Stain Kit (MP Biomedicals, Solon, OH, USA).
Small RNA libraries sequencing. Twenty second instar C. vestalis larvae, teratocytes from 200 parasitized P. xylostella larvae (see below), 20 third instar P. xylostella larvae, and 20 fourth instar P. xylostella larvae were collected separately and total RNA isolated using TRIzol reagent (Ambion, Foster City, CA, USA). Small RNAs (~18-30 nt) were isolated by the denaturing PAGE method of Lagos-Quintana et al. 39 Small RNA sequencing libraries were then constructed using the TruSeq Small RNA Library Preparation Kit (Illumina, San Diego, CA, USA). Library sequencing was performed by Illumina HiSeq 2000. miRNA genes were inferred by miRdeep2 software against the Rfam database of release 11.0 40 . Sequences that matched miRNA targets were predicted using RNAhybrid 41 , Tar-getScan 42 , and miRanda 43 with default parameters. To ensure the reliability, the genes that were predicted to be the targets by at least two algorithms were treated as the candidates for further analysis.
CvBV particles extraction and injection. To assess whether CvBV-encoded miRNAs are expressed in CvBV-infected host tissues, CvBV virions were collected from female wasps as previously described 16 . The ovaries of C. vestalis female adult wasps were dissected into prechilled PBS and the calyx was punctured individually. The calyx fluid with PBS was filtered using a 0.22 μm filter to remove cellular debris, and centrifuged at 20,000 × g for 1 h, the viral particle pellet was resuspended in PBS. The viral particles collected from one single adult female is defined as one FE. To infect host larvae, 0.1 FE CvBV particles were injected into P. xylostella larvae on the second day of the third instar, the expression level of CvBVencoded miRNAs was detected 6 h post injection by RT-qPCR. To infect Pxem_ZJU cells, 20 FE CvBV particles were added to a 35 mm culture dish containing 1 × 10 5 P. xylostella Pxem_ZJU cells, the expression level of CvBV-encoded miRNAs was detected 24 h post infection by RT-qPCR. Teratocyte collection and culture. Teratocytes from C. vestalis and P. xylostella hemocytes were collected using previously established methods 44 . In brief, P. xylostella larvae were dissected 5 days post-parasitism in culture dishes containing TNM-FH medium (HyClone, Logan, UT, USA) plus antibiotics (ampicillin and kanamycin, each at 100 µg/l). Opening the hemocoel of larvae released the C. vestalis teratocytes and P. xylostella hemocytes that were present into the medium where they settled at the bottom of the dish. After 30 min, most hemocytes had bound to the culture dish, while the much larger, nonadherent teratocytes were collected and transferred to another dish containing fresh medium. Teratocytes were further washed 5× in medium, transferred to a microfuge tube and then gently centrifuged at 500 × g for 5 min. The resulting pellet consisted of only teratocytes whose abundance was estimated using a hemocytometer. To obtain the conditioned medium, 1 × 10 5 cells teratocytes were collected from P. xylostella 5 days post oviposition, and cultured in TNM-FH medium with 10% exosomedepleted FBS. After 48 h, the teratocytes were collected with the culture medium into a microcentrifuge tube and centrifuged at 500 × g for 5 min, the supernatant and the teratocytes were collected separately for miRNA isolation. PCR amplification of the β-tubulin gene of P. xylostella was used as a marker for hemocytes while the C. xylostella 18S RNA gene was used as marker for contamination of hemocytes by teratocytes after purification.
Co-culture of teratocytes with host cells. For teratocytes co-cultured with host cells, 3 × 10 4 teratocytes were added to culture dishes containing 1 × 10 5 P. xylostella Pxem_ZJU cells. Two days later, teratocytes were collected and transferred to microfuge tubes as described above. Pxem_ZJU cells were collected by washing in PBS and then incubating in 0.25% trypsin (Sigma-Aldrich, St. Louis, MO, USA) for 30 min. The resulting cells were transferred to a microfuge tube and pelleted at 3000 × g. 18S RNA of C. vestalis was used to detect if there was contamination of teratocytes in P. xylostella Pxem_ZJU cells. Teratocytes and teratocytes cultured with P. xylostella Pxem_ZJU cells were used for the detection of miRNAs and functional validation, respectively.
Immunohistochemistry. For Alexa-NHS labeling, teratocytes were collected and cultured with water-soluble Alexa Fluor™ 568 NHS Ester (Thermo Fisher Scientific, Hudson, NH, USA) (1 μM) in TNM-FH medium with 10% FBS medium for 30 min. The medium was removed and the cells were washed with PBS three times followed by addition of fresh TNM-FH medium with exosome-depleted FBS and cultured for 6 h. The culture medium was then collected and subjected to exosome isolation using an exosome isolation reagent (Invitrogen). The Pxem_ZJU cells were adhered to an 8-chamber slide (Thermo Fisher Scientific) incubated with/ without the fluorescently labeled medium for 18 h. DAPI was added during the last 5 min. Slides were analyzed by confocal microscopy (LSM 800, Zeiss, Jena, Germany).
Detection of miRNA sources. To identify the sources of miRNAs, primers were designed based on the miRNAs of C. vestalis that did not cover the last variable nucleotides (Supplementary Table 4). Both Pxem_ZJU cells co-cultured with C. vestalis teratocytes and hemocytes of parasitized P. xylostella were collected for miRNA extraction. PCR-amplified fragments of interest were cloned into the pGEM-T Easy Vector I (Promega, Madison, WI, USA) and sequenced. For each group of miRNA homologs, 100 clones were randomly selected and sequenced. Based on sequence differences, the proportion of each miRNA originated from P. xylostella versus C. vestalis was then determined.
Quantitative real-time PCR of miRNAs and mRNA. miRNAs were extracted from P. xylostella hemocytes and Pxem_ZJU cells using the High Pure miRNA Isolation Kit (Roche, Mannheim, Germany). For RT-qPCR, total RNA was isolated using TRIzol reagent (Ambion) according to the manufacturer's instructions. The quality and concentrations of total RNAs were estimated by electrophoresis and NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific). Complementary DNAs (cDNAs) were synthesized using the NCode miRNA First-Strand cDNA Synthesis Kit (Invitrogen, Carlsbad, CA, USA). The universal RT-qPCR primer was provided by this kit. The miRNA-specific forward primers were designed based on mature miRNA sequences of C. vestalis by changing the U in the selected miRNA sequence to T and the Tm were adjusted to range from 50 to 55°C (Supplementary  Table 4) 45 . We also confirmed that teratocyte and hemocyte templates were not contaminated by teratocytes using PCR and specific primers for the C. vestalis 18S RNA gene. For mRNA detection, total RNA was isolated using TRIzol reagent (Ambion) according to the manufacturer's instructions. The quality and concentrations of total RNAs were estimated by electrophoresis and NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific). First-strand cDNAs were synthesized using the ReverTra Ace qPCR RT kit (Toyobo, Osaka, Japan) according to the manufacturer's instructions. Both β-actin (GenBank Accession Number AB282645) and β-tubulin (GenBank Accession Number EU127912) genes of P. xylostella were used as internal controls. Primer sequences used for RT-qPCR analysis are shown in Supplementary Table 4. The RT-qPCR analyses using the above cDNA templates and primers specific for particular miRNAs (Supplementary Table 4) were conducted using the CFX Connect real-time system (Bio-Rad, Hercules, CA, USA) and THUNDERBIRD qPCR Mix (Toyobo, Osaka, Japan). All RT-qPCR assays were performed using four internal replicates and at least three biological replicates under the following conditions: 95°C for 60 s and 40 cycles of 95°C for 15 s and 60°C for 30 s. Small nuclear RNA (snRNA) U6 gene of P. xylostella was used as the internal control, and RT-qPCR data were analyzed using the 2 −ΔΔCt method 46 or normalized to that of U6 snRNA using the 2 −ΔCt method 47 .
Luciferase assays. 3′-UTRs of target genes were obtained by 3′-RACE using specific primers (Supplementary Table 4) and the SMART RACE cDNA Amplification kit (Clontech, Mountain View, CA, USA). All PCR-amplified DNA fragments were cloned into the pGEM-T Easy Vector I (Promega) and sequenced. The 3′-UTR of Pxy-EcR was cloned into the pMIR-REPORT miRNA Expression Reporter Vector (Invitrogen). Mutations at the miR-281-3p seed sequence binding region were introduced at the miRNA target sites in the 3′-UTR of the P. xylostella ecdysone receptor gene (Pxy-EcR, GenBank accession number NM_001309151.1). The mutant construct was used as the negative control. Constructs were transfected into HEK293 cells using Lipofectamine 3000 (Invitrogen). Mimics of Cve-miR-281-3p, Cve-miR-31-5p-1, and Cve-novel22-5p were also transfected into the HEK293 cells. The miRCURY LNA miRNA mimic negative control (Exiqon 479903-001, Woburn, MA, USA) was used as negative control. The Dual-Luciferase Reporter Assay System (Promega) was used to measure the interaction between miRNAs and Pxy-EcR in each sample selected. At 48 h post transfection, the luciferase activity was measured according to the recommended protocol using a Thermo Scientific Varioskan Flash Microplate Reader (Thermo Fisher Scientific). miRNA agomir treatment. Cve-miR-281-3p and Cve-novel22-5p mimics synthesized by Sangon Biotech (Shanghai, China) were injected separately into the P. xylostella larvae on the first day of the third instar using a glass needle and an Eppendorf Femtojet (Eppendorf, Germany) with microcontroller (Narishige, Japan). The blank control and miRCURY LNA miRNA mimic negative control (Exiqon 479903-001) were designed following a universal rule 48,49 . The expression level of miR-281-3p in P. xylostella haemocytes was detected every 12 h post injection by RT-qPCR, as described above. The development of P. xylostella in the mimic-treated group (5 pmol), negative control group (5 pmol), and non-treated group was monitored using a Keyence VHX-2000C (Keyence, Japan) microscope at 12 h intervals. Total RNA at 0, 12, 24, 36 and 48 h post injection was extracted as previously described. For immunoblotting, larvae were processed using the One Step Animal Tissue Active Protein Extraction Kit (Sangon, Shanghai, China). Samples were then run on 10% SDS polyacrylamide gel electrophoresis (PAGE) gels and transferred to Immobilon-P (Millipore) membranes. After washing, the membrane was blocked with 2% BSA followed by incubation for 2 h with P. xylostella EcR monoclonal antibody (diluted 1:1200). Samples were then visualized on at room temperature for 1 h. After a further washing step, immunoreactive bands were visualized on X-ray film using a horseradish peroxidase-conjugated goat anti-rabbit secondary antibody (Abcam, Eugene, OR, USA) (1:2500) and ECL detection reagents (SuperSignal West Dura Extended Duration Substrate, Thermo Fisher Scientific).
Statistical analysis. All statistical analyses were performed using SPSS 16.0 software. Data are presented as mean ± s.e.m. Student's t test was used to determine the difference between two groups. p-value <0.05 was considered significant.
Data availability. The genome assembly data have also been deposited at DDBJ/ EMBL/GenBank under accession numbers LQNH00000000. The sequencing data files for the miRNA profiling and genome information of C. vestalis can also accessed from the Parasitica web portal (http://www.insect-genome.com/parasitica/).