Asian Citrus Psyllid RNAi Pathway – RNAi evidence

Diaphorina citri, known as the Asian citrus psyllid, is an important pest of citrus because it transmits a phloem-limited bacteria strongly implicated in huanglongbing (citrus greening disease). Emerging biotechnologies, such as RNA interference, could provide a new sustainable and environmentally friendly strategy for the management of this pest. In this study, genome and functional analysis were performed to verify whether the RNAi core genes are present in the Asian psyllid genome and if the RNAi machinery could be exploited to develop a management strategy for this pest. Analyses of RNAi-related genes in the Asian citrus psyllid genome showed an absence of sequences encoding R2D2, a dsRNA-binding protein that functions as a cofactor of Dicer-2 in Drosophila. Nevertheless, bioassays using an in Planta System showed that the Asian citrus psyllid was very sensitive to ingested dsRNA, demonstrating a strong RNAi response. A small dose of dsRNA administered through a citrus flush was enough to trigger the RNAi mechanism, causing significant suppression of the targeted transcript, and increased psyllid mortality. This study provides evidence of a functional RNAi machinery, which could be further exploited to develop RNAi based management strategies for the control of the Asian citrus psyllid.

program. In silico genome analysis were performed on the ACP genome to identify core RNAi-related genes known to play key roles in small non coding RNA pathways. This was followed by functional studies based on oral bioassays performed on adult ACP, in order to confirm that the RNAi machinery was functional. In view of potential field application, the 'in Planta System' (iPS) was used for the feeding bioassays 17 . Five ACP-specific dsRNAs designed to target genes associated with different biological process were screened in these feeding bioassays. Using the iPS, ACP, was shown to be very sensitive to ingested dsRNA, and demonstrated a strong RNAi response. The results of this study provide valuable ground information for future studies on the efficacy of RNAi for the control of ACP, as well as for conducting further genome-wide screens to identify genes that could be used for the development of pest management methods against ACP.

Results
RNAi core genes in the Asian Citrus Psyllid. All sequences representing the core RNAi genes were identified in the genome of ACP (Table 1), except R2D2, the cofactor of Dicer-2 in the siRNA pathway. Regarding Dicers in the miRNA pathway, one complete sequence fragment of Dicer-1 (NW_007377676.1), representing the two RNase III domains, was recovered, as well as ~ 62% of the putative full-length sequence of Drosha (NW_007377741.1). In contrast to Dicer-1, only a partial sequence of Dicer-2 (siRNA pathway, NW_007379804.1) was retrieved, lacking the RNase III domain. With respect to Argonaute proteins, complete or almost complete sequences were identified for Argonaute-1 (miRNA pathway, NW_007377764.1), Argonaute-2 (siRNA pathway, NW_007379505.1) and Aubergine (piRNA pathway, NW_007378348.1), while a smaller fragment for Argonaute-3 (piRNA pathway, NW_007377777.1) was also detected. Finally, complete sequences representing the Microprocessor subunit DGCR8 (NW_007377449.1) and Loquacious (NW_007377608.1) were identified. A phylogenetic tree was constructed for Dicers, Argonaute proteins and dsRNA-binding cofactors that included the ACP sequences and made a comparison with corresponding sequences of organisms of different taxonomic groups (Figs 1, 2 and 3). The phylogenetic trees were meant to confirm the identity of the identified genes from the ACP genome, by comparing the identified RNAi-related genes from the ACP genome with their homologs in other species, hence, providing added confirmation of the identity of these genes.

Ingestion of ACP specific dsRNA induces RNAi effects and mortality in ACP.
Five ACP-specific dsRNAs were designed to target genes associated with different biological processes, namely: (1) Energy mobilization: Arginine Kinase (AK, accession number: GU797832.1) 18 ; (2) Cell metabolism: pterin-4-alpha-carbinolamine dehydratase (PCDB1, accession number: DQ673423) 19  The dsRNAs were evaluated regarding efficacy to induce RNAi in ACP using the iPS feeding bioassay 17 . ACP were given feeding access for 15 days on citrus flush treated separately with each of the five designed dsRNAs, with a non-ACP dsRNA (dsGFP), and with no dsRNA (water control). ACP overall mortality scored at the end of the 15 days period varied depending on dsRNA ingested (Fig. 4A). Statistically significant mortality was induced only in insects which fed on flushes treated with dsAK (52%) and dsSOD (46%). Non-significant, lower mortality was observed in insects which fed on flushes treated with dsPCDB1 (8%), dsTOM (15%) and dsVIT (16%), which were not statistically different from the control dsGFP (4%) and water (5%) (Fig. 4A). In general, insect mortality started at 6 dpf, increasing overtime reaching a peak at 13 dpf (data not shown). Gene expression analyses were conducted on insects that fed on flush treated with dsAK and dsSOD inorder to confirm that the significant insect death observed was due to an RNAi effect. Insects collected at 5 dpf showed a 58% reduction in mRNA levels for AK mRNA and 69.6% for SOD compared to ACP at starting point ("Time 0") ( Fig. 4B). Insects that fed on flush treated with dsGFP or water did not show significant changes in AK expression levels. Together, these results show that ingestion of dsRNA can trigger the RNAi mechanism in ACP, leading to suppression of the target gene expression causing significant mortality, if the appropriate gene is targeted.

Discussion
In arthropods it is well established that the core RNAi machinery consists of Argonaute endonucleases, Dicer enzymes and dsRNA binding proteins (Table 1) [23][24][25][26] . Three different RNAi pathways can be distinguished in insects, based on the types of Dicers or Argonautes involved and based on the small RNAs involved. Thus, the siRNA pathway is activated by exogenous dsRNA and involves Dcr2/R2D2 and Ago2 (RNAi core machinery: siRNA pathway, Table 1). The miRNA pathway consists of nuclear Dicer (Drosha/Pasha), cytoplasmic Dicer (Dcr1/Loqs) and Ago1 as core proteins (RNAi core machinery: miRNA pathway; Table 1). The piRNA pathway is also involved in defence against transposable elements and is characterized by Argonaute proteins of the Piwi class (Aub, Ago3) and its independence of Dicer ( 27 RNAi core machinery: piRNA pathway; Table 1).
Of interest was the finding that the ACP genome lacked sequences encoding R2D2, a dsRNA-binding protein that functions as a cofactor of Dicer-2 in Drosophila. In Drosophila, r2d2 is an essential gene for both RNAi and the innate immune response against RNA viruses 28,29 . Since RNAi is an efficient process in ACP after dsRNA feeding 17,[30][31][32] , it follows that R2D2 may not be necessary for this process in ACP or that it escaped detection. A similar finding was reported by Swevers et al. 33 where r2d2 was not found in the gut transcriptome of the colorado potato beetle known to have an efficient RNAi process following dsRNA treatment 10,34,35 . Presumably a different dsRNA-binding cofactor may compensate for the absence of R2D2 in ACP. Interestingly, BLASTX searches using Tribolium R2D2 as a query against the ACP genome resulted in the identification of another Loquacious, "Loquacious-like" (NW_007377608.1, Table 1 and Fig. 3). In Drosophila, it was demonstrated that Loquacious can act as a cofactor for both Ago1-and Ago2-RISC 36 . Further studies to explore the interaction of Loquacious in the RNAi pathway of ACP may be warranted. With the exception of R2D2, all RNAi-related genes that were searched for in the genome of ACP were identified, often encoding full-length open reading frames (ORFs). This then suggests that ACP has the three functional small RNA pathways, i.e. miRNA, siRNA and piRNA.
Besides these core genes, many different factors were identified that play a role in the efficiency of RNAi (Tables 1 and 2). The idea behind this approach was that RNAi cofactors have co-evolved with the core RNAi machinery and therefore will show similar phylogenetic profiles specific to each arthropod clade [37][38][39][40][41] . The picture that emerges is that the different RNAi pathways, while they have distinct components, are intimately integrated with other essential cellular processes such as translation, RNA processing, cytoskeleton function, transcriptional regulation, protein turnover, protein trafficking, splicing, nuclear import and export, DNA repair, and other mRNA degradation pathways 39,[41][42][43][44] . It was beyond the scope of the present article to investigate the presence of all factors implicated in RNAi in the ACP genome database. However, the identification of the selected RNAi related/associated genes in the ACP genome suggests that these genes do not constitute limiting factors for RNAi efficiency in ACP.
The genomic identification and analysis of core genes for the RNAi pathway, combined with RNAi bioassay results clearly demonstrate that an RNAi response is active in ACP. Similar results from RNAi studies in ACP have also been reported 17,[30][31][32]45 . However, the results presented here also indicate that, apparently, ACP is very sensitive to RNAi, as a small dose of dsRNA (100 ng) administered through a citrus flush (new growth tissue between 0.25 g to 0.35 g) could induce significant RNAi responses (target gene suppression and death).
Sensitivity in ACP could be due to different factors: (1) an absence or reduced dsRNase activity in the digestive tract and hemolymph; (2) efficient cell uptake of dsRNA, or (3) both. There is no biochemical analysis reported so far showing the presence of nucleases (or degradation of dsRNA) from psyllid salivary enzymes, or in the hemolymph, as shown in aphids 12 . Furthermore, previous studies reported that ingestion of siRNA induced gene silencing and mortality in psyllids 46 . These reports indicate that siRNA can remain intact and functional after being ingested by ACP. Also, these studies show that ACP cells are able to take up siRNA from the environment, and the siRNA could move systemically through the insect body. In Ecdysozoa, two different dsRNA uptake systems have been described so far. On one hand, there are the systemic RNA interference deficient (SID) transmembrane channel-mediated proteins, which were discovered in Caenorhabditis elegans 47 . SID-1 homologous genes have been reported from many but not all insects. For the other transmembrane proteins, SID-2 and SID-5 48 , and the tyrosine kinase SID-3 49 , no insect homologs have been reported as of yet. These SID proteins are necessary in the systemic RNAi response in C. elegans; for SID-1 and SID-2 it is thought to be through their involvement in dsRNA uptake from the C. elegans intestine 47,50 . A second mode of uptake of dsRNA in insects is endocytosis. In Drosophila melanogaster, no sid-1 homolog is present, and dsRNA uptake by receptor-mediated endocytosis has been demonstrated for this species 51,52 . Furthermore, in the Colorado potato beetle, endocytosis has been shown to be involved in dsRNA uptake, in addition to the SID-1 uptake route 53 . SID-1 was previously shown to occur in Anopheles gambiae (Diptera); Apis mellifera (Hymenoptera); Acyrthosiphon pisum (Hemiptera); Diaphorina citri (Hemiptera); Bombyx mori (Lepidoptera), Bombus terrestris (Hymenoptera); Drosophila melanogaster (Diptera); Leptinotarsa decemlineata (Coleoptera); Pediculus humanus (Phthiraptera); Tribolium castaneum (Coleoptera). The phylogenetic tree was constructed using the Neighbor-Joining method with MEGA, version 6, software. The numbers associated with the branches refer to bootstrap values (confidence limits) resulting from 1000 replicate resamplings. two psyllid species, ACP, Diaphorina citri and potato psyllid Bactericera cockerelli L. 54 , using degenerate primers to the conserved motif of sid-1-like genes. These sequences included conserved 'FYDXHD, MFFSFM, LDDD, and PVF' amino acid regions. A more thorough examination of SID-1 and other transporters, described from this study provides a better understanding of the genes involved with the RNAi mechanism in psyllids. Analyses of dsRNA movement post oral ingestion from plant tissue in ACP and the glassy-winged sharpshooter, Homalodisca vitripennis (Gemar)(Hemiptera: Cicadellidae), reported by Hunter et al. 54 showed dsRNA detection in the fat body post feeding access. However, while SID-1 is a key component in the uptake of dsRNA in some arthropods 55 , the systemic movement in arthropods is not solely dependent upon the presence of SID-1 47,56 . It is interesting to note that in two closely related coleopteran insects, Tribolium castaneum (Tc) and Diabrotica virgifera (Dv), the former has three different sid-1-like genes, which do not seem to be necessary for dsRNA uptake 57 whereas the latter only has two, which are both confirmed as being involved 58 . Further studies will therefore be needed to elucidate the mechanism of dsRNA uptake and systemic spread in ACP. However, the fact that the RNAi machinery is present and can be induced by low amounts of dsRNA delivered through plants and iPS, offers some intriguing possibilities for future applications in pest management.
In conclusion, this study provides evidence of a functional RNAi machinery in the Asian citrus psyllid. Given that ACP has all the important RNAi core genes needed to present a significantly strong response to ingested dsRNA, management strategies based upon RNAi appear to be a suitable approach to successfully reduce this pest in a highly specific manner. Nevertheless, further genome-wide screens to identify better genes, which could induce stronger RNAi effects in ACP will be required for the development of a proper and effective management method for this pest.

In silico identification and manual curation of RNAi-related genes in ACP. A list of RNAi-related
genes was composed based on literature ( Table 2) that included the following categories: RNAi core machinery, auxiliary factors, dsRNA uptake, antiviral RNAi and nucleases. Amino acid sequences of homologs for these genes were found in GenBank for Tribolium castaneum, which has a well-annotated genome. Subsequently, the genome of ACP available on the i5k Workspace@NAL platform (https://i5k.nal.usda.gov/webapp/blast/) 59 , was searched against the genes in this list, using sequences of homologs as query. Hits (contigs) obtained from this search were subsequently used in BLASTX searches to verify their identity and to detect functional domains. Phylogenetic analysis of RNAi core genes from ACP genome. To search for ACP's RNAi homologous proteins in other species, a protein-protein BLAST search (BLASTp) was performed using the NCBI BLAST Service (http://www.ncbi.nlm.nih.gov/). The selected species were: Tribolium castaneum (Coleoptera), Leptinotarsa decemlineata (Coleoptera), Bombyx mori (Lepidoptera), Danaus plexippus (Lepidoptera), Acyrthosiphon pisum (Hemiptera), Bombus terrestris (Hymenoptera), Apis mellifera (Hymenoptera), Pediculus humanus corporis (Phthiraptera), Ixodes scapularis (Acari/Ixodidae), Drosophila melanogaster (Diptera) and Anopheles gambiae (Diptera). For protein alignment and phylogenetic tree construction, only the hit with the

Figure 4. Ingestion of ACP-specific dsRNAs induced insect mortality and gene suppression. (A) Adult
ACP were fed on citrus flushes treated with dsRNAs targeting five ACP genes and controls treatments (dsGFP and water). ACP mortality was monitored over a 15 days period. The asterisk indicates the first day where cumulative mortality was observed on flushes treated with dsAK or dsSOD, which showed statistical differences (P < 0.05) compared to the controls (dsGFP and water). (B) Adult ACP showed reduced levels of AK (Arginine kinase) and SOD (Superoxide Dismutase) mRNAs 5 days post feeding (dpf) on a flush treated with dsAK or dsSOD. Feeding on flushes treated with dsGFP or water did not alter target gene expression in psyllids. Bars represent the standard deviation. Different letters indicate statistically significant differences (P < 0.05).
lowest E-value was chosen. If no significant hit was found or the protein sequence obtained was incomplete, the organism was excluded from the phylogenetic tree. Amino acid sequence alignments and phylogenetic analyses were performed with the ClustalW program, integrated in the Molecular Evolutionary Genetics Analysis 6.06 software (MEGA 6.06) 60 . Phylogenetic trees and P-Distances were constructed using the Neighbor-Joining method 61 with MEGA 6.06 software. Bootstrapping 62 was used to estimate the reliability of phylogenetic reconstructions (1000 replicates). Synthesis of double stranded RNA. The synthesis of dsRNAs was performed using the MEGAscript RNAi kit (Thermo-Fisher Scientific) according to the manufacturer's instructions. The template DNA for dsRNA production was generated by PCR amplification using gene-specific primers containing T7 promoter sequences tailed at the 5′ end of each primer. All primer sequences are listed in Table 2. A dsRNA specific for Green fluorescent protein (dsGFP, AJ306911.1) was used as a negative control in the experiments. After synthesis, dsRNA concentrations were measured using NanoDrop ™ ND1000 spectrophotometer (Thermo-Fisher Scientific, Waltham, USA).

Plant material and insect colony.
Feeding Bioassay. We used the bioassay called 'in planta system' (iPS) 17 to deliver dsRNA to ACP. Briefly, citrus flushes were collected from potted plants, washed in 0.2% bleach for 10 min and rinsed for 3 min by submersion in autoclaved or filtered water. The base of stems were cut with a razor blade, ~45° angle, while submerged in water. The flush was then transferred to a 1.5 mL tube containing 0.5 mL water. A dsRNA solution (i.e. 100 ng of dsRNA in 300 μ L of water) was added to the tube, the tube top was wrapped with Parafilm ™ (American National Can ™ , Neenah, WI 54956) and placed under artificial lighting to stimulate absorption of the dsRNA solution. After the plant flush absorbed the entire dsRNA solution, the tube was filled with filtered water (0.45 μ m, Sterile Syringe Filters, Corning ® ). The treated flushes in the tubes were then transferred to a cage and 15 adult ACP, of mixed genders, were added to each cage (Fig. 5). Each treatment (ACP-dsRNA, dsGFP and water control)  consisted of four cages, and experiments were repeated three times. Mortality was scored daily for 15 days. Data for total mortality was analyzed with analysis of variance (ANOVA) and t-Test, with P < 0.05 of probability.

Sample collection, RNA extraction and quantitative real-time PCR (RT-qPCR).
To assess the extent of RNAi by ACP, the level of the mRNA was quantified by RT-qPCR. Each treatment was composed of three cages, with 15 adults in each cage. From each cage, three pools of 3 insects each were collected 5 days post feeding (dpf) on treated flush. At time point zero (the day the experiment was set up) three pools of 3 insects were collected from each treatment group. These samples represent "Time 0" and were used for comparative gene expression analysis. The experiment was repeated twice. ACP were collected using a mouth-operated aspirator (BioQuip) and transferred to 1.5 mL tubes containing 50 μ L of Trizol ® (Thermo-Fisher Scientific) and glass beads ≤ 106 μ m (Sigma-Aldrich). After maceration using a pestle, 950 μ L of Trizol ® was added to the tube and insects' total RNA was extracted following the manufacturer's instructions. An aliquot of 5 μ g of total RNA was treated with Turbo DNA-free ™ (Thermo-Fisher Scientific) to remove residual DNA. The cDNA synthesis was carried out using the SuperScript ™ III first-strand synthesis system (Thermo-Fisher Scientific), following the manufacturer's instructions. RT-qPCR reactions were performed in triplicates using Power SYBR ® Green Master Mix (Applied Biosystems, ThermoFisher Scientific). RT-qPCR reactions were carried out using a Rotor-Gene 6000 PCR machine (QIAGEN Corbett, Australia). The α -tubulin expression levels were used to normalize Ct values. The relative gene expression data were analyzed using the relative 2 −ΔΔCT method 63 . The expression levels of the target gene were relative to the expression of the same gene in the ACP collected at "Time 0" (which was assigned the value of 1.0). The data were statistically analyzed using analysis of variance (one way ANOVA) and t-Test at P < 0.05 of probability. All the primer sequences used for RT-qPCR are provided on the Table 3.
Data accessibility. All Table 3. Primers used for dsRNA production and RT-qPCR. $ The underlined sequences correspond to the T7 sequence. * The position of the primer is related to the gene.