Structural and functional insights into the ATP‐binding cassette transporter family in the corn planthopper, Peregrinus maidis

The corn planthopper, Peregrinus maidis, is an economically important pest of maize and sorghum. Its feeding behaviour and the viruses it transmits can significantly reduce crop yield. The control of P. maidis and its associated viruses relies heavily on insecticides. However, control has proven difficult due to limited direct exposure of P. maidis to insecticides and rapid development of resistance. As such, alternative control methods are needed. In the absence of a genome assembly for this species, we first developed transcriptomic resources. Then, with the goal of finding targets for RNAi‐based control, we identified members of the ATP‐binding cassette transporter family and targeted specific members via RNAi. PmABCB_160306_3, PmABCE_118332_5 and PmABCF_24241_1, whose orthologs in other insects have proven important in development, were selected for knockdown. We found that RNAi‐mediated silencing of PmABCB_160306_3 impeded ovary development; disruption of PmABCE_118332_5 resulted in localized melanization; and knockdown of PmABCE_118332_5 or PmABCF_24241_1 each led to high mortality within five days. Each phenotype is similar to that found when targeting the orthologous gene in other species and it demonstrates their potential for use in RNAi‐based P. maidis control. The transcriptomic data and RNAi results presented here will no doubt assist with the development of new control methods for this pest.


INTRODUCTION
The corn planthopper, Peregrinus maidis, is a destructive pest of maize and sorghum. Its feeding behaviour results in chlorosis, stunting, wilting and can even lead to the death of the plant. It also can transmit maize mosaic virus and maize stripe virus. The combination of physical damage and viral infection can significantly reduce crop yield (Autrey, 1983). Insecticide treatment and host plant resistance are the main methods for P. maidis control. Though applying insecticides is convenient, resistance develops rapidly. Moreover, insecticides negatively impact ecosystems and human health. While developing host plant resistance is challenging, it is a more sustainable method than applying chemicals. However, host information on resistance to P. maidis and its associated viruses remains scarce, significantly limiting choices among cultivars (Singh & Seetharama, 2008). As such, new control methods are sorely needed.
The ATP-binding cassette (ABC) transporter family is one of the largest gene families. It contains many transport proteins responsible T A B L E 1 Overview of 46 ATP-binding cassette (ABC) transporters identified in Peregrinus maidis.
These transporters can be divided into two structural types, full transporters and half transporters, based on the number of transmembrane domains (TMD) and nucleotide-binding domains (NBD) they encode.
A full transporter consists of two TMDs and two NBDs; a half transporter is comprised of one TMD and one NBD and must form a homo-or hetero-dimer to function. The TMDs are less conserved, reflecting their involvement in the translocation of specific substrates.  (Pan et al., 2020) as well as in insecticide-treated Sogatella furcifera (Horváth) (Zhou et al., 2018), D. citri  and Nilaparvata lugens . In two of these studies, ABC proteins were inhibited by verapamil Pan et al., 2020), and in the study of Pan et al. (2020), ABC genes were also knocked down using

Transcriptome sequencing, assembly and annotation
Over 32 million raw reads were generated across eight libraries: PmMeg (middle-stage embryos), PmLeg (late-stage embryos), N1-3 (1st-3rd-instar nymphs), N4-5 (4th-5th-instar nymphs), SWF (shortwinged females), LWF (long-winged females), SWM (short-winged males) and LWM (long-winged males). The reads were trimmed, and de novo assembled into contigs. The resulting combined transcriptome assembly consisted of 23,707 contigs with an average length of 2,406 bp (summary in Table S1). The BUSCO result indicated that the assembly has a high degree of completeness since 95% of the Insecta single-copy orthologs were found ( Figure S1). Blast2GO was used to align contigs to the arthropod-specific section in the NCBI non-redundant protein database and perform gene ontology (GO) mapping and annotation.
The numbers of total contigs, contigs with blast hits, contigs with GO mapping and contigs with annotations are shown in Figure S2. There were 18,237 BLASTx-annotated contigs, and most of the top BLASTx hits came from the brown planthopper, N. lugens ( Figure S3).

ABC transporters and phylogenetic analysis
Forty-six putative ABC transporters were identified within the combined P. maidis transcriptome assembly (Table 1), and the numbers of each subfamily member were compared with those found in other insects ( Table 2)

RNAi of PmABCB_160306_3, PmABCE_118332_5 and PmABCF_24241_1
The identity of each RNAi target (PmABCB_160306_3, PmABCE_118332_5  used the deduced amino acid sequences of L. striatellus ABC transporters as queries (Sun et al., 2017) because L. striatellus is one of the most closely related species in which ABC transporters have been identified (Huang et al., 2017). Only using the two ABCA sequences identified in this species as queries could possibly limit the number of identifiable ABCA transporters. Therefore, we further queried the transcriptome with 11 ABCA sequences from Lygus hesperus (Hull et al., 2014). Interestingly, the additional query sequences had no impact-the top hits were still the two ABCA members we previously identified. Given the level of assembly completeness indicated by BUSCO ( Figure S1), there may only be two ABCA members in P.

maidis.
Seven ABCB transporters were identified in P. maidis, which is about the average in insects (Table 2). ABCB subfamily is associated with insecticide transport and/or resistance (Dermauw & Van Leeuwen, 2014). For example, in L. striatellus, ABCB2 was overexpressed consistently in chlorpyrifos-, deltamethrin-and imidaclopridresistant strains, indicating its involvement in resistance to these insecticides (Sun et al., 2017). ABCC transporters play a role in insecticide transport and/or resistance as well (Dermauw & Van Leeuwen, 2014). Generally, hemipteran insects have fewer ABCC transporters (Table 2), which is also the case for P. maidis-six ABCC members were identified within the P. maidis transcriptome. Although it is outside the scope of this work, given that insecticide treatment is F I G U R E 4 Effects of RNAi on Peregrinus maidis. Fifth-instar nymphs were injected with buffer (20% phenol red solution), dsRNA of target genes and EGFP. (a) The knockdown was confirmed by semiquantitative PCR on the third day post injection. RPL10 was used as a reference gene. The band intensities were quantified using ImageJ, and shown above the bands (times compared to that in the buffer treatment). (b) Injection of PmABCB_160306_3 dsRNA disrupted ovary development. (c) Localized melanization distant from the injection site (marked by the white stars) was found in the PmABCE_118332_5-dsRNA treatment. (d) Knockdown of PmABCE_118332_5 and PmABCF_24241_1 led to death during moulting.
the main method for P. maidis control, it is worth studying how ABCB and ABCC transporters participate in insecticide transport and/or resistance in P. maidis.
Two ABCD, one ABCE and three ABCF transcripts were identified in P. maidis, similar to those found in many other insects ( We identified 17 ABCG and eight ABCH transporters within the P. maidis transcriptome (Table 2). In arthropods, ABCG subfamily is comparatively large (Dermauw & Van Leeuwen, 2014), so the number of G-subfamily members we identified was in line with expectations.
ABCH subfamily has been identified only in arthropods and zebrafish thus far (Dermauw & Van Leeuwen, 2014), and more ABCH members were found in hemipteran insects ( Table 2). The larger number of Hsubfamily members appeared to be due to a gene expansion within a single clade, which was reported in L. striatellus (Sun et al., 2017), Bemisia tabaci (Tian et al., 2017), D. citri (Wang et al., 2019) and also found in P. maidis (Figure 2).

To better understand if ABC genes could be good targets for
RNAi-based control of P. maidis, we tested PmABCB_160306_3, PmABCE_118332_5 and PmABCF_24241_1 via RNAi, since their orthologs in other insects have proven important during development.
Though semiquantitative PCR is not as informative as qPCR, the results suggest that the expression of target ABC genes was reduced by RNAi (Figure 4a). Interestingly, while the expression level of reference gene, Ribosomal protein L10 (RPL10), appeared to be consistent in all three treatments (buffer, EGFP-dsRNA and ABC-dsRNA), the expression level of target ABC genes varied between the negative controls (buffer and EGFP-dsRNA treatments). Although we strived to load the same amount of PCR product in every lane, it is impossible to completely rule out human error. However, we think a more likely explanation for the difference observed between the negative controls is a global effect on gene expression triggered by the EGFPdsRNA. A similar phenomenon was reported by Nunes et al. (2013). Specifically, they found that despite the lack of GFP sequence in the honeybee genome, feeding GFP dsRNA caused a global change in gene expression, with over 200 genes being up-regulated and over 400 being downregulated (Nunes et al., 2013). Since our knockdown phenotypes were consistent with those found in other species (Adedipe et al., 2019;Broehan et al., 2013) and no such phenotypic changes were found in either of the control treatments (buffer and EGFP-dsRNA), we are confident that RNAi-mediated silencing of the target ABC genes was indeed achieved.
RNAi silencing of PmABCB_160306_3 impeded ovary development. It should be noted that the dissected females had just emerged, so eggs were not seen in either treatment. Evidently, however, the ovarioles were reduced in the PmABCB_160306_3-dsRNA treatment ( Figure 4b), consistent with previous findings (Adedipe et al., 2019;Broehan et al., 2013). Another change worthy of attention is that the haemolymph appeared to be much oilier when we dissected the PmABCB_160306_3-dsRNA-injected planthoppers (personal observation). PmABCB_160306_3 is an ortholog of ABCB7. ABCB7 mutant medaka fish exhibited abnormal iron metabolism in erythrocytes and lipid accumulation in the liver (Miyake et al., 2008). The analogue to the liver in insects is the fat body in haemocoel. Therefore, the oilier haemolymph probably came from lipid accumulation caused by the knockdown of PmABCB_160306_3, which has not been pointed out in other insects (Adedipe et al., 2019;Broehan et al., 2013).
RNAi silencing of PmABCE_118332_5 and PmABCF_24241_1 resulted in high mortality within five days ( Figure 5), revealing the essential functions of these ABCE and ABCF genes. While we do not know their functions in P. maidis, in humans ABCE1 plays diverse roles in viral infection (Dooher & Lingappa, 2004;Liu et al., 2020;Zimmerman et al., 2002), translation (Khoshnevis et al., 2010;Mancera-Martinez et al., 2017), and ribosomal recycling (Barthelme et al., 2011;Mancera-Martinez et al., 2017;Pisarev et al., 2010). ABCE1, also known as RNase L inhibitor, negatively regulates the 2 0 ,5 0 -oligoadenylate synthetase-RNase L system, an innate immunity pathway responsible for antiviral effects (Bisbal et al., 1995). The localized melanization found in P. maidis injected with dsRNA targeting PmABCE_118332_5 (Figure 4c) might be due to disruption of a similar function. In humans, ABCF1 is involved in translation initiation (Tyzack et al., 2000) and recently there has been increased focus on its role in immunity (Arora et al., 2019;Lee et al., 2013). Sequence conservation between human and arthropod ABCE, as well as ABCF proteins (Dermauw & Van Leeuwen, 2014), hints at functional conservation. Therefore, it is not surprising that knockdown of PmABCE_118332_5 and PmABCF_24241_1 led to high mortality. As Broehan et al. (2013) expected, insects other than coleopterans are indeed susceptible to RNAi silencing of the ABCE and ABCF genes, which makes them ideal targets for pest control. Corrected mortality rate on the fifth day after injection F I G U R E 5 Corrected mortality rate on the fifth day post injection. Twenty to thirty nymphs were injected per treatment, and the experiment was performed three times. On the fifth day post injection, living insects were counted and the corrected mortality rate was calculated using Abbott's method (Abbott, 1925). Each bar represents the mean ± standard error. The different letters above the bars indicate significant differences ( p < 0.05), as determined by oneway ANOVA and Tukey's test.
PmABCE_118332_5 and PmABCF_24241_1 appear to be better targets for RNAi-based control than PmABCB_160306_3.
Knockdown of PmABCB_160306_3 indeed impacted the female reproductive system. However, since P. maidis feeding causes damage, it is possible for nymphs to damage the crop before they are impacted by RNAi silencing of PmABCB_160306_3. Delivering dsRNA of PmABCE_118332_5 and PmABCF_24241_1 via feeding will be our future work. The result will be informative, because delivery via feeding is closer to how dsRNA will be delivered in the field.
Transgenic plants can be a way to deliver dsRNA such as dsRNAexpressing rice against the brown planthopper (Zha et al., 2011), transgenic lettuce against the silverleaf whitefly (Ibrahim et al., 2017) and transgenic wheat against the grain aphid (Sun et al., 2019). Another possible way to deliver dsRNA is via a dsRNA-expressing symbiont.
The feasibility of this method has recently been demonstrated: dsRNAexpressing Rhodococcus rhodnii in the kissing bug (Whitten et al., 2016), BFo2α (an Enterobacteriales species) in the western flower thrip (Whitten et al., 2016) and Serratia symbiotica in the pea aphid (Elston et al., 2021). Since feeding would reduce the efficacy of silencing effects, it would be good to target more genes, for example, V-ATPase D (Yao et al., 2013).
In conclusion, we generated a comprehensive de novo transcriptome assembly for P. maidis by combining RNA sequence data from a wide range of life stages. We identified 46 members of the ABC transporter family and provided insights into the function of some via RNAi. Knockdown phenotypes showed that three of these have potential as targets for RNAibased control. Our transcriptomic data and RNAi results have potential to aid in the development of new control methods for P. maidis.

Transcriptome sequencing, assembly and annotation
Total RNA was isolated from a range of P. maidis life stages ( Figure S4 After quality checks using FastQC (Andrews, 2010), raw reads from each library were imported into Lasergene software (DNASTAR, Madison, WI) for quality trimming and de novo assembly into contigs using default settings. A summary of the combined transcriptome is shown in Table S1. The contigs were BLASTx-annotated (E value cutoff of 10 À5 ) using OmicsBox software (BioBam, Valencia, Spain).
The data are available at NCBI under BioProject PRJNA896602 with SRA SRR22197767 to SRR22197774 for the raw reads.

ABC transporters and phylogenetic analysis
The annotated P. maidis transcriptome was exported as a fasta file from OmicsBox, and imported into BlastStation (TM Software, Arcadia, CA) as a database. To search for ABC transcripts in this database, deduced amino acid sequences of L. striatellus ABC transporters (Sun et al., 2017) were used as queries. The P. maidis homologues were selected based on the E value and percentage of identity. Then, the homologous sequences and their deduced amino acid sequences (obtained at the ExPASy website https://web.expasy.org/translate/) were used as BLASTx and BLASTp queries, respectively at the NCBI website (https://blast.ncbi.nlm.nih.gov/Blast.cgi) to confirm the identity and integrity. The subfamily was double-checked using BLASTp at the FlyBase website (https://flybase.org/blast/index.html). Two distinct domains of ABC proteins, TMD and NBD, were predicted by Scan-Prosite (https://prosite.expasy.org/scanprosite/) and TMHMM Server (http://www.cbs.dtu.dk/services/TMHMM/). Sequences of the ABC transcripts are provided (Supporting Information).
The phylogenetic trees were built using MEGAX software (Kumar et al., 2018). The deduced amino acid sequences were aligned using the MUSCLE algorithm. Subsequently, 56 amino acid substitution models were tested. The model with the lowest Bayesian information criterion (BIC) score, considered to describe the substitution pattern the best, was chosen for the construction of phylogenetic trees. Then, trees were built using the maximum likelihood method with the best fitting substitution model and partial deletion treatment of gaps or missing data, and re-sampled using 1,000 bootstrap replicates.

RNAi of PmABCB_160306_3, PmABCE_118332_5 and PmABCF_24241_1
To start, we compared the expression level using transcriptomic data ( Figure S5) and synthesized cDNA for the use in dsRNA synthesis.
Total RNA was extracted from pooled 20 short-winged females using the QIAGEN RNeasy mini kit (QIAGEN, Hilden, Germany), and reverse-transcribed into cDNA using the SuperScript III first-strand synthesis kit (Thermo Fisher Scientific Inc., Waltham, MA). The cDNA was used as a template in the following polymerase chain reaction (PCR).
To synthesize dsRNA, linear target fragments flanked by the T7 promoter sequences were needed. These were produced by nested PCR. Specifically, in the first-round PCR, the target fragment was amplified using cDNA as template (50 ng for each 10 μL reaction) and target-specific primers (as listed in Table S2). The first-round PCR product was confirmed by agarose gel electrophoresis. Then, the band was cut and soaked in sterile water (three times of the band weight) in a 1.5 mL microcentrifuge tube overnight. The overnight solution was the template in the second-round PCR. In the second round, the primers flanked by the T7 promoter sequences (as listed in Table S2) were used to be recognized by T7 RNA polymerase for dsRNA synthesis. The second-round PCR product was purified using the QIAquick PCR purification kit (QIAGEN, Hilden, Germany) and sequenced to confirm the identity. Then, 1 μg of it was used as template for dsRNA synthesis using the MEGAscript T7 transcription kit (Thermo Fisher Scientific Inc., Waltham, MA). The dsRNA was purified using the Direct-zol RNA miniprep kit (Zymo Research, Irvine, CA), and the concentration was determined using a NanoDrop 1000 Spectrophotometer (Thermo Fisher Scientific Inc., Waltham, MA).
Fifth-instar nymphs were chosen for injection based on previous studies (Adedipe et al., 2019;Broehan et al., 2013). Nymphs were anaesthetized on ice, placed in an injection arena (9-cm-diameter Petri dish containing a layer of 1% agarose), and injected with 40 nL of dsRNA ($120 ng) using a FemtoJet microinjector (Eppendorf, Hamburg, Germany). Buffer (20% phenol red solution; phenol red was mixed with the dsRNA solutions for the ease of injection) and EGFPdsRNA were injected as negative controls. Nymphs without treatment were kept to correct the mortality rates using Abbott's method (Abbott, 1925). Then, the nymphs were reared in a cup with corn leaves for three days ( Figure S6) and moved to a 2-week-old corn seedling for further observation. On the third day post injection, knockdown was confirmed by semiquantitative PCR. Three nymphs from each treatment were homogenized in 300 μL of QIAzol reagent (QIAGEN, Hilden, Germany) and kept at -80 C for semiquantitative PCR (RNA extraction and cDNA synthesis were conducted as described above; 50 ng of cDNA was used for each 10 μL PCR reaction). RPL10, a stably expressed gene, was the reference gene in semiquantitative PCR (Yao et al., 2013). The primers used are listed in Table S2. The PCR band intensities were quantified using ImageJ program (version: 1.53) according to the ImageJ User Guide (Ferreira & Rasband, 2012). On the fifth day post injection, living insects were counted, and the corrected mortality rate was calculated using Abbott's method (Abbott, 1925). The statistical significance of mortality rates was determined by one-way ANOVA and Tukey's test in Microsoft Excel 2016 (Microsoft, Redmond, WA). Virgin females from the PmABCB_160306_3-dsRNA treatment were also collected on the fifth day post injection, and stored at À80 C for ovary dissection.
Twenty to thirty nymphs were used in each treatment, and the experiment was performed three times.
If a kit is used without further explanation above, experiments are conducted according to the manufacturer's protocol. PCR was performed using MyTaq DNA Polymerase (Meridian Bioscience, Cincinnati, OH) with the cycling condition on a C1000 Touch Thermal Cycler (Bio-Rad Laboratories, Hercules, CA) as follows: 95 C for 1 min; 35 cycles of 95 C for 15 s, 55 C for 15 s and 72 C for 10 s; 72 C for 5 min. To confirm the PCR products and RNAi-mediated knockdown, agarose gel electrophoresis was conducted: 1% agarose gel was prepared in 40 mL of TBE buffer containing 1 μL of ethidium bromide; gel loading dye was added to the PCR products and loaded into the wells; 2 μL of HyperLadder™ 100 bp (Meridian Bioscience, Cincinnati, OH) was loaded as ladder; the gel was run at 90 V for 40 min.

CONFLICT OF INTEREST
The authors declare no competing interests.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are openly available in GenBank at https://www.ncbi.nlm.nih.gov/genbank/.