Introduction

Olfaction plays a pivotal role in intra- and inter-specific interactions by directing insects towards food or prey, mating partners, oviposition sites, and away predators as well as toxic compounds1. The specialized organ for olfaction in insects is the antenna, on which hair-like, multi-pore sensilla are situated and peripheral olfactory signaling events occur. Olfactory receptor neurons (ORNs) and their auxiliary structures are located at the roots of the antennae2, and the entire olfactory system is dependent to a great extent on receptors expressed at the peripheral ORNs. Starting with perception of semiochemicals and ultimately ending with the translation of olfactory signals into behavior, the entire process requires orchestration of the insect’s sophisticated olfactory system at various levels. Several types of olfactory proteins are believed to participate in the selective detection and, once they have conveyed information, the rapid inactivation of trace amount of odorants, i.e. odorant receptors (ORs), ionotropic receptors (IRs), gustatory receptors (GRs), odorant binding proteins (OBPs), chemosensory proteins (CSPs) and sensory neuron membrane proteins (SNMPs)3.

Insect ORs are seven-transmembrane domain proteins with a reversed topology compared to the G-protein coupled ORs in vertebrates4,5. ORs play a central role in converting semiochemicals into electrical signal, functioning as a heterodimer with a divergent, conventional ORx and a highly conserved noncanonical OR co-receptor Orco in fruit fly, OR2 in moths and OR7 in mosquitoes3. The OR genes are expressed in the olfactory neurons housed within the olfactory sensilla (found mainly on the antenna)6.

GRs are also seven-transmembrane domain proteins, but they are more ancient than ORs. GR genes are expressed in the gustatory neurons housed within the gustatory sensilla (found on the labia, maxillary palps, antennae, legs and genitalia)7. GRs can respond to tastants such as sugars, bitter substances, CO2 and some contact pheromones8,9,10,11.

IRs belong to the ionotropic glutamate receptor (iGluR)-like protein family and can be activated by small molecules like acetates and amine-like volatile compounds12,13. It has been proven that IRs are involved in chemosensation14,15 and other functions, i.e. regulation of the circadian clock in Drosophila melanogaster 16 and induction of physical defense in Daphnia pulex 17. IRs usually contain three transmembrane domains (TMDs), a bipartite ligand-binding domain with two lobes and one ion channel, and have been proposed to act as dimmers or trimers of subunits coexpressed in the same neuron12. However, they aren’t expressed in chemosensory neurons that express ORs or Orco14.

OBPs are the liaisons between external cues and ORs18, and they selectively bind hydrophobic odorant chemicals and transport them to the surface of the dendrites of ORNs19,20,21. OBPs also function in the recognition of specific odors through activation of the ORx/Orco complex20. Another class of odorant binding proteins, CSPs, are small soluble proteins expressed predominantly in the sensilum lymph as well as in non-olfactory tissues. It is clear that CSPs bind odorant or pheromone compounds22,23,24, but their olfactory mechanisms areas yet poorly studied.

SNMPs are insect membrane proteins that are known to associate with pheromone sensitive ORNs in Lepidoptera and Diptera25. There are two types of SNMPs, SNMP1 and SNMP225. In moth, the subtype SNMP1 is coexpressed with pheromone receptors (PRs) in pheromone-responsive neurons25, whereas the subtype SNMP2 is confined to sensilla support cells25,26,27,28.

The diamondback moth, Plutella xylostella (Lepidoptera: Plutellidae), is a destructive insect pest distributed worldwide that can cause considerable damage in cruciferous crops. It is estimated that the total loss caused by P. xylostella is about US$4-5 billion annually29. Although a bioinformatics analysis of the whole-genome sequence has explained the evolutionary success of P. xylostella with regard to its expansion in gene families associated with the perception and detoxification of plant defense compounds/insecticides at the genetic and molecular levels30, the peripheral olfactory mechanisms that contribute to the fitness of this insect pest remain poorly understood. Identification of genes expressed in the antennae will supply baseline information to understand their likely function in odorant perception in P. xylostella and insects adaptation to various host plants.

In the present study, we sequenced and analyzed the antennal transcriptome of P. xylostella adults using second-generation high-throughput Illumina RNA sequencing (RNA-seq). The purpose of our study was to identify olfaction-related genes which might be targets as a part of pest control strategies of this insect pest species that devastates cruciferous vegetables. We identified 118 candidate chemosensory genes encoding 54 ORs, 16 IRs, 7 GRs, 15 CSPs, 24 OBPs and 2 SNMPs. The sex-biased and tissue-specific expression patterns of 54 ORs and 16 IRs was also determined by semi-quantitative reverse transcription PCR. We reported the protein sequences of these chemosensory genes in Supplementary Dataset File.

Results

Sequencing and unigene assembly

By using Hiseq. 2000 sequencing approach, a total of 60,041,232 and 59,753,272 raw reads were obtained from the P. xylostella female and male antennae samples, respectively. After removing low quality and adaptor reads, female and male antennae yielded 54,430,716 and 54,059,300 clean reads and 4,898,764,440 nt and 4,865,337,000 nt clean nucleotides, respectively. After initial assembly, 124,488(mean length 278 nt) and 132,190 contigs (mean length 268 nt) were obtained from the female and male antennae libraries, respectively. Next, 62,278 female (mean length 555 nt) and 63,928 male unigenes (mean length 531 nt) were generated after contig connecting. These two unigene sets were then pooled together for further clustering, which yielded a final set of 59,844unigenesconsisting of 18,570 distinct clusters and 41,274 distinct singletons. The mean length of these unigenes was 660 nt, and N50 was 979 nt (Table 1).

Table 1 Summary of the Pluttela xylostella transcriptome assembly.
Full size table

Identification of candidate chemosensory receptors: ORs and GRs

All the unigenes were searched by blastx against nr database and further by tblastn using 63 ORs from B. mori as queries, 54 candidate OR genes were identified (Table 2). Of these, 23 were predicted to have full-length open reading frames (ORFs). The length of these 23 OR genes ranges from 376 to 473 amino acid residues, and the encoded proteins are estimated to have 5–7 TMDs, which is characteristic of typical insect ORs. The remaining 31 OR genes code for at least 163 amino acids and are predicted to have more than one TMD. A phylogenetic analysis was then performed using our candidate ORs and the ORs from other Lepidopteran insects including H. armigera, H. virescens and B. mori (Fig. 1).

Table 2 Candidate olfactory receptor and gustatory receptor unigenes.
Full size table
Figure 1
figure 1

Phylogenetic tree of candidate Lepidopteran ORs, including the PR and Orco clades. Pxyl: P. xylostella (red), Harm: Helicoverpa armigera (black), Hvir: Heliothis virescens (green), Bmor: Bombyx mori (blue). The clade shaded in blue indicates the Orco clade. The clade shaded in red indicates the PBP clade. The bootstrap value for phylogenetic tree construction is 1000.

Full size image

The OR co-receptor gene was easily identified because of extremely high conservation among species compared to other chemosensory receptors. Similar to other insect ORs, most P. xylostella (Pxyl) ORs are highly divergent and share low similarity with other Lepidopteran insect ORs, including ORs from H. armigera, H. virescens and B. mori. However, nine PxylORs had 33%~100% identity to previously characterized PRs from P. xylostella and B. mori. They formed a single subgroup in a phylogenetic tree of Lepidopteran ORs (Fig. 1). Seven of these nine PxylORs (PxylOR1 andPxylOR3-8) were predicted to have full-length ORFs. Two short sequences (PxylOR41and PxylOR45) were also clustered in the PR branch. PxylOR41 has high similarity to PxylOR4, and PxylOR45 has relatively high similarity to BmorOR6. 12 of the remaining PxylORs were clustered with their Lepidopteran orthologous genes in the phylogenetic tree. But most PxylORs appeared to be distantly related to the known insect ORs (Fig. 1). We named the Orco unigene PxylOR2 and the 7full-length candidate PR unigenes PxylOR1 and PxylOR3-PxylOR8. The other 46 OR unigenes were ranked in order of decreasing ORF length and named PxylOR9-PxylOR54. We also identified 7 candidate GRs and named them as PxylGR1-PxylGR7.

Identification of candidate IRs

IR sequences in the P. xylostella antennal transcriptome were identified based on similarity to known IRs of Lepidopteran insects, B. mori, C. pomonella, H. armigera, H. virescens and S. littoralis. Sixteen candidate IRs were identified by bioinformatic analysis, and five unigenes were predicted to have a full-length ORFs. The insect IRs typically have three TMDs. Of the 16 candidates IRs, 15 are predicted to have at least one TMD (Table 3). Twelve of the 16 putative IRs are at least 48% identical to the corresponding Lepidopteran orthologous IRs in S. littoralis and C. pomonella. The remaining four unigenes have relatively low similarity to other insect IRs: CL2177.Contig2 has 35% identity with IR1 of S. littoralis, unigene13888 has 31% identity with IR75 of C. pomonella, CL4692.Contig1 has only 25% identity with IR60a of D. melanogaster, and CL5979.Contig2 has only 24% identity with IR7c of D. melanogaster (Table 3). Phylogenetic analyses suggested that the prediction of IRs was credible. In a neighbor-joining tree of insect IRs, all candidate PxylIRs were clustered in a separate clade with their Lepidopteran orthologs (Fig. 2). All of these 16 candidate IR unigenes were named based on their homology to known IRs. For example, the IR Unigene 19385 has 55% similarity withIR75q2 and CL1791. Contig1 had 64% similarity to IR75q2. So, we named Unigene 19835 PxylIR75q2.2.

Table 3 Candidate ionotropic receptor unigenes.
Full size table
Figure 2
figure 2

Phylogenetic tree of candidate IRs from Pluttela xylostella and other insects. Pxyl: P. xylostella (red), Harm: Helicoverpa armigera (black), Hvir: Heliothis virescens (green), Bmor: Bombyx mori (blue), Slit: Spodoptera littoralis (purple), Cpom: Cydia pomonella (cyan).The clade shaded in red indicates the IR8a/IR25a clade. The bootstrap value for phylogenetic tree construction is 1000.

Full size image

Identification of putative OBPs

We identified 24 unigenes encoding OBPs from the antennal transcriptome of P. xylostella, including 3pheromone binding proteins (PBPs) and 3 general odorant binding proteins (GOBPs) (Table 4). Twenty-two of these 24 unigenes were predicted to have signal peptides, and 19 have full length ORFs. Signal peptide sequences were not detected in the remaining two putative OBPs due to incomplete N-terminal sequences. All 24 putative OBPs had high similarity to known Lepidopteran OBPs. The PBP and GOBP sequences were clustered in a separate clade in the OBP neighbor-joining tree (Fig. 3). Three candidate OBPs were classified into a PBP subgroup in the phylogenetic tree. They share 66%~100% similarity with previously characterized Lepidopteran PBPs and thus were named PBPs. We also found two GOBPs in the antennal transcriptome of P. xylostella and named them PxylGOBP1 and GOBP2. A new GOBP (PxylGOBP1.2) was identified that has 77% identity with PxylGOBP1. It was clustered in the GOBP clade and distinguished from other OBPs in the phylogenetic tree. The other 18 candidate OBPs are obviously distinct from the PBP and GOBP clades and have relatively lower similarity to OBPs from other Lepidopteran insects. Most candidate OBP sequences, such as PxylOBP2, PxylOBP3, and PxylOBP7, are closely clustered with at least one Lepidopteran ortholog, in congruence with the blastx results. Some candidate OBP sequences such as PxylOBP6, PxylOBP9, PxylOBP11 and PxylOBP17 are not clustered with OBPs from other Lepidopteran insects (Fig. 3). A possible reason may be that the orthologs of these PxylOBPs have not been identified in other Lepidopteran insects.

Table 4 Candidate odorant binding protein unigenes.
Full size table
Figure 3
figure 3

Phylogenetic tree of candidate Lepidopteran OBPs, including the GOBP and PBP clades. Pxyl: Pluttela xylostella (red), Harm: Helicoverpa armigera (black), Hvir: Heliothis virescens (green), Bmor: Bombyx mori (blue). The clade shaded in blue indicates the PBP clade. The clade shaded in red indicates the GOBP clade. The bootstrap value for phylogenetic tree construction is 1000.

Full size image

Identification of candidate CSPs

Bioinformatic analysis led to the identification of 15 different sequences encoding candidate CSPs (Table 5). All 15 unigenes were predicted to have signal peptides and 14 have a full length ORFs. Four candidate PxylCSPs (PxylCSP1-4) match the previously identified P. xylostella CSP sequences31. The other 11 candidate CSP sequences have at least 35% identity with known CSPs from other insects, and we named them according to the length of the coding region in descending order. In a neighbor-joining tree, all 15 sequences form a cluster with Lepidopteran orthologous genes (Fig. 4).

Table 5 Candidate chemosensory protein unigenes.
Full size table
Figure 4
figure 4

Phylogenetic tree of candidate Lepidopteran CSPs. Pxyl: Pluttela xylostella (red), Harm: Heliocoverpa armigera (black), Hvir: Hethiothis virescens (green), Bmor: Bombyx mori (blue). The bootstrap value for phylogenetic tree construction is 1000.

Full size image

Identification of candidate SNMPs

SNMPs were first identified in pheromone-sensitive neurons of Lepidoptera31 and are thought to function in pheromone detection32. Two kinds of SNMPs (SNMP1 and SNMP2) have been identified in insects and transcripts corresponding to both were found in the P. xylostella transcriptome. The sequence of CL2414Contig2 is identical to the PxylSNMP1sequence published in Genbank. CL242.Contig4 has 70% identity with SNMP2 of O. furnacalis, and we annotated this sequence as P. xylostella SNMP2 (Table 6).

Table 6 Candidate sensory neuron membrane protein unigenes.
Full size table

Tissue- and sex-specific expression of candidate OR and IR genes in P. xylostella

To investigate the expression profile of PxylORs and PxylIRs between sexes and tissues, we determined the expression patterns of the 54 candidate ORs and 16 candidate IRs genes in the antennae and legs of male and female adult P. xylostella by semi-quantitative reverse transcription PCR (RT-PCR) (Fig. 5). As shown in Fig. 5, all of these 54 ORs were expressed in the antennae. PxylOR2 was expressed in male and female antennae and legs. The expression of the nine candidate PRs was observed only in antennae but not in legs. And of which, seven candidate PRs (PxylOR1, 3, 4, 5, 6, 7 and 41) had male-biased or male-specific expression patterns. Different from the other lepidopteran ORs, PxylOR8, was only expressed in female antennae. PxylOR45 was expressed in both male and female at a similar level. In other 44 general ORs PxylOR54 expression was much higher in female than in male antenna and the remaining 43 ORs were expressed in both male and female antennae at a similar level. In contrast to ORs, the expression of all IRs did not differ significantly between males and females. All of these 16 PxylIRs were expressed in the male and female antennae, but PxylIR7d.3 and PxylIR25a were also expressed in legs.

Figure 5
figure 5

Tissue- and sex-specific expression patterns of candidate PxylORs and PxylIRs. M: male antennae, F: female antennae, L: legs. PxylRPS3 is the reference.

Full size image

Discussion

In the present study, we profiled the antennal transcriptome of P. xylostella adults by RNA-seq technology and annotated 118 putative olfactory genes, including 54 putative ORs, 24 OBPs, 16 IRs, 15 CSPs, 7 GRs, and 2 SNMPs. Chemosensory genes have been identified in other Lepidopteran insects; 134 putative chemosensory unigenes were identified in the antennae of H. armigera, including 60 ORs, 34 OBPs, 19 IRs,18CSPs, 1 GR and 2 SNMPs, and 131 putative chemosensory unigenes were identified in H. assulta antennae, including 64 ORs, 19 IRs, 29 OBPs, 17 CSPs, and 2 SNMPs33. Our results are comparable with those from H. armigera and H. assulta in the number of genes identified. The identification of chemosensory genes from antennal transcriptomes was also reported for the moth M. sexta (91 genes, including 48 ORs, 18 OBPs, 21 CSPs and 4 IRs)34 and B. mori (138 genes, including 71 ORs, 20 OBPs, 16 CSPs and 31 IRs)35,36,37,38 and many other insect pests.

Insects utilize three groups of chemosensory receptors, ORs, IRs and GRs, to perform a variety of essential behaviors such as foraging, mating and oviposition. ORs are the centerpiece of peripheral olfactory reception and determine the sensitivity and specificity of odorant reception3. Due to the availability of insect genome databases and progress in sequencing technology, increasing numbers of OR genes have been identified from many Lepidopteran species. To date, 68, 64, 70 ORs have been identified in the genome databases of B. mori 38, Danaus plexippus 39 and Helioconius Melpomene 40, respectively. Recently, by using next-generation sequencing technology the antennal transcriptome of M. sexta was profiled, and 48 OR genes were identified34,41. In this study, we identified 54 ORs in the antennal transcriptome of adult P. xylostella. The number of ORs identified in this paper is less than that identified by You et al.30 in the genome database of P. xylostella. We might have missed some development-related OR genes because we only identified chemosensory genes in the adult antennae. Typical insect ORs are characterized by seven TMDs. We found less than seven TMDs in PxylORs, which is also observed in other Lepidopteran insects33,42,43. This is probably caused by the limited power of the software used for TMDs finding.

All of the PxylORs identified in the antennal transcriptome are highly divergent and share low similarity with other Lepidopteran insect ORs. A study showed that the common ancestor of Lepidopterans had fewer OR genes but that there were multiple gene gains and few gene losses during the evolution of Lepidoptera. This phenomenon of gene family expansion is suggested to be associated with the adaption of Lepidopteran species to host plants44. We also identified 9 (PxylOR1, PxylOR3-8, PxylOR41 and PxylOR45) candidate PRs based on their similarity to previously characterized PRs. The antennal expression pattern of PoxylPRs is consistent with that of PRs in H. armigera 42 and S. littoralis 45. Among these 9 candidate PRs, 7 showed male-biased expression, and PxylOR5 was only expressed in male antennae. In contrast, PoxylOR8 was only expressed in female antennae. Sex and tissue-specific expression of chemosensory genes is very common among Lepidoperan pests. It was found in H. assulta 33 and H. armigera 42 that some of their antennal OR genes showed sex-biased expression pattern. The male-specific expression of PxylOR5 probably plays a role in locating females, while the female-specific expression PxylOR8 likely also has ecological significance, i.e. optimization of pheromone production and spatial dispersion of females among host plants46,47 and selection of oviposition sites.

We identified one Orco unigene, named PxylOR2, which has high similarity to HarmOR2, BmorOR2 and HvirOR2. Orco is highly conserved among all insect species3 and carries out similar functions in different insects48 by forming a ligand-gated ion channel49. Orco probably functions as a chaperone and forms a dimer with the other ORs in P. xylostella.

GRs can respond to tastants such as sugars, bitter substances, CO2 and some contact pheromones50. Thus, GRs play very important roles in food selection and feeding behaviors in insects. The first insect GRs were identified in the fruit fly, D. melanogaster 51. The number of Lepidopteran GRs varies greatly; there is one GR in Cydia pomonella 52 and H. armigera 42, 2 in M. sexta 34, 3 in Heliothis virescens 53 and 5 in Spodoptera littoralis 45,54. In the antennal transcriptome of adult P. xylostella we identified 7 GRs, which is more than those in the Lepidopteran insects mentioned above, but far less than the number found in the silkworm B. mori (65 GRs)55 and the oriental tobacco budworm H. assulta (18 GRs)56. GRs are mainly expressed in gustatory organs such as the proboscis and maxillary palps, rather than in antennae8. This is a possible reason why we identified only 7GRs in P. xylostella. Two GR genes, GR21a and GR63a have been proved to be putative CO2 receptors in the antennae of the fruit fly57,58. And in mosquitos, 3 putative CO2 receptor genes (GR22, 23 and 24) have been identified in the maxillary palps of different species59,60,61. The PxylGR1 was closely related to the GR22 in mosquito and GR21a in the fruit fly and predicted to be a candidate CO2 receptor.

IRs belong to an ancient chemosensory receptor family, and two subfamilies of IRs have been identified recently, i.e. the conserved ‘antennal IRs’ and the species-specific ‘divergent IRs’62. The first IR was identified in the coeloconic sensilla of Drosophila 14 and most Drosophila IRs have clear orthologs within the genus of Lepidoptera34,42,63. IRs are ligand-gated ion channels that mediate chemical communication between neurons14. In this study, we identified 16 IRs in the antennal transcriptome of P. xylostella and named them based on homologous sequences from other insects. Similar numbers of IRs have been identified from other Lepidopteran insects: 19 IRs were identified in the antennal transcriptomes of H. armigera and H. assult 33, 15 IRs in C. pomonella 52, 20 IRs in Chio suppresalis 43, and 12 IRs in S. litoralis 36. All of these IRs are expressed in antennae, but PxylIR7d.3 and PxylIR25a are also expressed in legs, which is different from the expression patterns of these genes in H. assulta 33. Coincidently, HarmIR25a, HarmIR75d, HarmIR75p and HarmIR76p are also expressed in the cotton bollworm legs42. The function of leg-expressed IRs remains unknown and deserves in-depth investigation.

OBPs are believed to be directly involved in the activation of the ORx/Orco complex in the recognition of specific odors20. A total of 24 OBPs were identified in the antennal transcriptome of P. xylostella, including three GOBPs and three PBPs. The number of OBPs identified in the present study was comparable to those identified in transcriptomic analyses of H. armigera (34) and H. assulta (29)33, S. litura (21)64, S. littoralis (26)54, but fewer than those identified in B. mori (44)37. OBPs showed lineage-specific expansion and diversification; therefore, it is not surprising that there are some differences, or even big differences, in the number of OBPs. Previous studies have also shown that some insect OBPs and CSPs are expressed exclusively in non-antennae tissues or in larvae65. Therefore, different sampling and sequencing strategies may lead to different results. In a previous study, two GOBPs, GOBP1 and 2, were identified in P. xylostella antennae66. GOBPs were also found in the antennae of C. pomonella 67 and S. litura 68. The antennal P. xylostella GOBPs identified in this study have ecological significance, e.g. guiding P. xylstella to find better food69. The antennal S. litura GOBP1 can bind to plant odorants, while S. litura GOBP2 can bind to aldehyde-sex compounds and analogs68.

CSPs are a class of small soluble proteins expressed highly in the chemosensilla lymph70 and show high binding activity to odorants and pheromones71. We identified 15CSPs genes in the present study. The number of CSPs identified from P. xylostella was comparable to the number in B. mori (18)72, H. armigera (18) and H. assulta (17)33 and S. litura (18)64, but fewer than the number in M. sexta (21)34, Sesamia inferens (24)63 and S. littoralis (31) 45. Because CSPs are also expressed in tissues other than antennae73,74 and may participate in other physiological processes, it is possible that we have missed some CSPs in our antennal transcriptome analysis.

SNMPs are two-transmembrane domain proteins that share very high homology to members of the mammalian CD36 receptor family, which are thought to function in pheromone detection of Lepidopteran and Dipteran insects31. Two subtypes of SNMPs (SNMP1 and SNPM2) have been frequently identified in most insects, e.g. Helicoverpa armigera 33,42, Cnaphalocrocis medinalis 27, S. exigua75, S. litura 28, C. suppressalis 43, H. assulta 33, and in this study, P. xylostella. The expression of antennal SNMPs in P. xylostella suggests their role in pheromone detection, similar to what has been reported in D. melanogaster 32,76.

Conclusions

In summary, we identified 118 candidate olfactory genes that may function in odorant perception in the diamondback moth, P. xylostella by assembling and annotating transcriptomic sequence data. We carried out a comparative phylogenetic analysis to predict gene functions and examined the transcriptome patterns of the P. xylostella OR and IR genes. Genes with sex-biased and tissue-specific expression patterns, especially PxylOR5 and PxylOR8, are potential targets for environmentally-friendly management of this destructive insect pest. Our results lay the foundation for functional analysis of these receptors in both neurobiological and evolutionary studies.

Materials and Methods

Insect rearing

The laboratory-maintained P. xylostella was reared in the Institute of Plant Protection, Chinese Academy of Agricultural Sciences, Beijing, China. The larvae and adults were fed on Chinese cabbage and kept in cages at 27 ± 1 °C under a 16: 8 (L: D) photoperiod and 65 ± 5% relative humidity. Male and female larvae were distinguished at the last instar and placed in separate cages. Antennae of female or male adults were dissected at 1–3 days after adult emergence, immediately frozen in liquid nitrogen, and then stored at −70 °C until use.

Total RNA extraction

The frozen antennae were transferred to a liquid nitrogen-cooled mortar and ground with a pestle. One mL of TRIzol reagent was pipetted to the homogenate (Invitrogen, Carlsbad, CA, USA) and total RNA was extracted following the manufacturer’s instructions. Total RNA was resuspended in RNAse-free H2O, and RNA quantity was determined with a Nanodrop ND-2000 spectrophotometer (NanoDrop products, Wilmington, DE, USA). RNA integrity was assessed using an Agilent 2100 BioAnalyzer (Agilent Technologies, Englewood, CO, USA).

cDNA Library construction and Illumina sequencing

Tenμg of total RNA, extracted from approximately 2000 antennae of 1–3 day old adult male or female moths. The cDNA library for each sample was prepared using the NEBNext® mRNA Library Prep Reagent Set for Illumina (NEB, Ipswich, MA, USA) following the manufacturer’s instructions. Poly-A RNA for each sample was fragmented in fragmentation buffer to a length of 200 nt–700 nt. Random hexamers were used to generate first-strand cDNA, and second-strand cDNA was synthesized using RNaseH and DNA polymerase I. The double-strand cDNA (ds cDNA) samples were purified with the QIAquick PCR Purification Kit (Qiagen, Hilden, Germany) and eluted with EB buffer. The short fragments were treated with T4 DNA Polymerase and T4 Polynucleotide Kinase for end-repair and dA-tailing, then sequencing adaptors with barcodes were ligated to the dA tail of ds cDNA using T4 DNA ligase. To select insert length, ds cDNA samples were separated by agarose gel electrophoresis and bands of approximately 200 bp were excised and purified with the QIAquick Gel Extraction Kit (Qiagen, Hilden, Germany). Paired-end sequencing of the library was performed on the Illumina HiSeq™ 2000 platform (Illumina, San Diego, CA, USA) at the Beijing Genome Institute (Shenzhen, China). The read length of each end was 90 bp. The male and female libraries were sequenced in one lane, and raw reads were then sorted by barcode sequence.

Unigene generation

Raw reads were pre-processed to remove low quality reads and reads containing adapter sequences and poly-A/T tails. The publicly available program Trinity was used to perform de novo assembly of clean reads to generate a set of transcripts77. The Trinity outputs were then clustered by TGICL (TGI Clustering tools)78. The final unigene dataset consists of uniformly clustered sequences and singletons.

Gene identification and functional annotation

Unigene sequences were first searched against protein databases like nr, Swiss-Prot, KEGG and COG, using blastx with an e-value cut-off of 1e−5  79. To identify more OR genes, 63ORs from B. mori were used as queries in tblastn searches of P. xylostella antennal unigenes. Unigene ESTs were predicted using ESTScan80. Signal peptides in the protein sequences were predicted using SignaIP 4.081. The TMDs of annotated genes were predicted using TMHMM Server Version2.0 (http://www.cbs.dtu.dk/services/TMHMM).

Phylogenetic analyses

Phylogenetic trees were constructed based on the amino sequences of the candidate olfaction genes and genes from the collected data sets. The OR datasets contained OR sequences identified from Lepidopteran insects (36from H. armigera, 18 from H. virescens and 63 from B. mori)38,42,82,83. The IR datasets contained IR sequences from H. armigera (11), S. littoralis (11), Cydia pomonella (10), B. mori (18) and D. melanogaster (64)36,42,52,62. The OBP datasets contained sequences from H. armigera (26), H. virescens (17) and B. mori (34)37,42. The CSP data set contained sequences from H. armigera (13)42, H. virescens (9)84 and B. mori (16)70. All amino acid sequences were aligned using ClustalW285. The unrooted neighbor-joining trees were constructed by the Jones-Taylor-Thornton(JTT) method with 1,000 bootstrap replications as implemented in MEGA5 software86.

Expression analysis of the candidate receptors by semi-quantitative reverse transcription PCR

To illustrate and compare the expression patterns of candidate receptors in male and female antennae, semi-quantitative RT-PCR was performed using cDNA prepared from male antennae, female antennae and legs (male and female mixture). Legs were used as a control to confirm the antennae-enriched expression of candidate receptors. Total RNA was extracted as described above. Prior to cDNA synthesis, RNA was treated with DNase I (Fermentas, Vilnius, Lithuania) to remove trace amounts of genomic DNA. The cDNA was synthesized using the First Strand cDNA Synthesis Kit (Fermentas, Vilnius, Lithuania) and was used as a template in PCR reactions with gene-specific primers. The housekeeping gene RPS3 was used as a control87. Primers were designed using the Primer Premier 5 software (PREMIER Biosoft International), and the sequences are available in Supplementary Table S1. PCR was performed with the Veriti Thermal Cycler (Applied Biosystems, Carlsbad, CA, USA) under the following conditions: 94 °C for 2 min, 33 cycles of 94 °C for 30 s, 55–60 °C for 30 s, and 72 °C for 30 s, and 72 °C for 10 min. The cycle number was reduced to 27 and 30 for Actin and OR2 amplification because of their high expression level. The experiment was repeated three times using three independently isolated RNA samples. PCR amplification products were run on a 2% agarose gel and verified by DNA sequencing.