Sensory gene identification in the transcriptome of the ectoparasitoid Quadrastichus mendeli

Sensory genes play a key role in the host location of parasitoids. To date, the sensory genes that regulate parasitoids to locate gall-inducing insects have not been uncovered. An obligate ectoparasitoid, Quadrastichus mendeli Kim & La Salle (Hymenoptera: Eulophidae: Tetrastichinae), is one of the most important parasitoids of Leptocybe invasa, which is a global gall-making pest in eucalyptus plantations. Interestingly, Q. mendeli can precisely locate the larva of L. invasa, which induces tumor-like growth on the eucalyptus leaves and stems. Therefore, Q. mendeli–L. invasa provides an ideal system to study the way that parasitoids use sensory genes in gall-making pests. In this study, we present the transcriptome of Q. mendeli using high-throughput sequencing. In total, 31,820 transcripts were obtained and assembled into 26,925 unigenes in Q. mendeli. Then, the major sensory genes were identified, and phylogenetic analyses were performed with these genes from Q. mendeli and other model insect species. Three chemosensory proteins (CSPs), 10 gustatory receptors (GRs), 21 ionotropic receptors (IRs), 58 odorant binding proteins (OBPs), 30 odorant receptors (ORs) and 2 sensory neuron membrane proteins (SNMPs) were identified in Q. mendeli by bioinformatics analysis. Our report is the first to obtain abundant biological information on the transcriptome of Q. mendeli that provided valuable information regarding the molecular basis of Q. mendeli perception, and it may help to understand the host location of parasitoids of gall-making pests.

Sensory genes play a key role in the life of parasitoids, such as foraging, oviposition site selection, and mating partners 1 . There are two major chemosensory mechanisms through olfaction and taste in which chemical signals are detected by one of the large multigene families that encode chemosensory proteins (CSPs), gustatory receptors (GRs), ionotropic receptors (IRs), odorant-binding proteins (OBPs), sensory receptors (ORs) and sensory neuron membrane proteins (SNMPs) 2 . The function of CSPs and OBPs is the first step in the recognition of chemical stimuli from the outside environment 3 . Chemoreceptors (such as GRs, IRs and ORs) are involved in the recognition and identification of various chemical signals and environmental odors to modulate chemical perception 4 . SNMPs are involved in cell signal transduction 4 .

Materials and methods
Insects. Branches of saplings damaged by L. invasa were collected from Guangxi University (108°29′ E, 22°85′ N), Nanning City, Guangxi Zhuang Autonomous Region, in October 2018. Specimens were placed in a glass container filled with water to retain freshness and transferred to a sealed net cage (length × width × height = 40 cm × 40 cm × 80 cm) with 70-80% relative humidity and a natural light photoperiod maintained at 27 ± 1 °C. The water in the glass container was replaced daily until the end of the emergence of Q. mendeli. The emerged Q. mendeli were collected daily using 50-mL plastic tubes. One day later, the tubes were immediately placed into liquid nitrogen and stored at − 80 °C. Six groups of female Q. mendeli adults (a group of twenty) were prepared for RNA extraction.

RNA sequencing.
A NanoPhotometer spectrophotometer (Thermo Fisher Scientific, Massachusetts, USA) and the Nano6000 Assay Kit for the Agilent Bioanalyzer 2000 system (Agilent Technologies, California, USA) were applied to check the purity and integrity of the total RNA, respectively. After total RNA extraction, magnetic beads with Oligo dT (Thermo Fisher Scientific, Hampton, USA) were used to enrich mRNA, and then, a fragmentation buffer was added to make it a short fragment. The fragments were sequenced on an Illumina HiSeq4000 (Illumina, California, USA).
Transcriptome data analysis. Reads obtained from the sequencing machines included dirty reads containing adapters or low-quality bases, which affected the subsequent assembly and analysis. De novo transcriptome assembly was carried out with the short read assembly program Trinity v3.0 30 . Basic annotation of unigenes includes protein functional annotation, pathway annotation, COG/KOG functional annotation and Gene Ontology (GO) annotation. To annotate the unigenes, we used the BLASTx program (http:// www. ncbi. nlm. nih. gov/ BLAST/) with an E-value threshold of 1e−5 for the NCBI nonredundant protein (Nr) database (http:// www. ncbi. nlm. nih. gov), the Swiss-Prot protein database (http:// www. expasy. ch/ sprot), the Kyoto Encyclopedia of Genes and Genomes (KEGG) database (http:// www. genome. jp/ kegg) 31,32 , and the COG/KOG database (http:// www. ncbi. nlm. nih. gov/ COG). Protein functional annotations could then be obtained according to the best alignment results 33 .
Sequence alignment and phylogenetic analysis. TransMembrane prediction using Hidden Markov Models 2.0 (http:// www. cbs. dtu. dk/ servi ces/ TMHMM) 34 . The signal peptides were predicted using SignalP 4.1 (http:// www. cbs. dtu. dk/ servi ces/ Signa lP/) 35 . Amino acid sequence alignment was performed using the ClustalW method implemented in Mega v7.0 36 . The Q. mendeli CSP, GR, IR, OR, OBP and SNMP nucleotide sequences were used as queries (BLASTx) in the GenBank database, and sequences from different insect species (i.e., Apis mellifera Ligustica, Bombyx mori Silk, C. floridanum, D. melanogaster, M. mediator, N. vitripennis, Solenopsis invicta Buren, Tribolium castaneum Herbst and Tric. Pretiosum Riley) and their amino acids were retrieved in the GenBank database and used to construct a phylogenetic tree. A. mellifera, B. mori, D. melanogaster and Trib. castaneum are model insects; S. invicta is a kind of social insect; C. floridanum, M. mediator and Tric. pretiosum were parasitoid from Hymenoptera. Amino acid sequences were aligned using the Muscle method implemented in Mega v7.0 37 . The resulting alignment was manually curated to remove gap-rich regions. Maximum-likelihood trees (for CSP, GR, IR, OR, OBP and SNMP) were constructed using IQ-TREE with the best-fitting substitution model 38 . Subsequently, trees were viewed and graphically edited in FigTree v1.4.3 39 and Adobe Illustrator CS6. Branch support was assessed using the bootstrap method based on 1000 replicates.
Identification of candidate chemosensory genes. In this study, 3 putative unigenes encoding CSPs were identified, named QM_comp07737, QM_comp08732 and QM_comp26540 (Additional file 6). The lengths of these unigenes were 509 bp, 608 bp and 280 bp, respectively (Additional file 7). Among these unigenes, QM_ comp07737 and QM_comp26540 were incomplete due to a lack of a 5′ or 3′ terminus (Additional file 7). QM_ comp08732 sequences were full-length putative CSP genes because they had complete ORFs and 4 cysteines,   2 and Additional file 9). The phylogenetic tree showed that QM_comp26540 shares a high homology and is closely clustered with MmedCSP1, which has been functionally characterized, and QM_comp07737 and QM_comp08732 did not branch clusters with any other insects; they may be specific CSPs of Q. mendeli (Fig. 2).
Identification of candidate gustatory receptors. Ten candidate GR proteins were identified from the data sets (Additional file 6). Among these unigenes, QM_comp03300, QM_comp23544, QM_comp24536 and QM_comp26507 were incomplete due to the lack of a 5′ or 3′ terminus (Additional file 7). QM_comp00164, QM_comp03333, QM_comp11847, QM_comp15910, QM_comp22611 and QM_comp22814 sequences were full-length putative GR genes because they had complete ORFs. These unigenes had molecular weights that ranged between 4 and 56 kDa and had a signal peptide sequence that ranged between 15 and 41 amino acids at the N-terminus (Additional file 8). Through a homology search with known proteins, the results showed that 79% of QM_comp15910 were orthologs of the proteins in Trichomalopsis sarcophagae Gahan. A phylogenetic tree based on the maximum likelihood method was constructed used the 10 GR sequences of Q. mendeli along with 191 GR sequences from 9 other species (i.e., A. mellifera, B. mori, C. floridanum, D. melanogaster, M. mediator, N. vitripennis, S. invict, Trib. castaneum and Tric. pretiosum) ( Fig. 3 and Additional file 9). The phylogenetic Identification of candidate ionotropic receptors. Twenty-one candidate IR proteins were identified from the data sets (Additional file 6). Among these unigenes, 13 unigenes were incomplete due to the lack of a 5′ or 3′ terminus (Additional file 7). Eight unigenes were full-length putative IR genes because they had complete ORFs. These unigenes had molecular weights that ranged between 5 and 12 kDa (Additional file 8). Through a homology search with known proteins, the results showed that 100% of the IRs in Q. mendeli were orthologs of the proteins in Asbolus verrucosus LeConte, and the orthologs of other IR sequences were also above 37% (Additional file 8 were chosen to construct a phylogenetic tree based on the amino acid sequences (Additional file 9). The phylogenetic tree showed that all candidate IR proteins were clustered with at least one Hymenoptera ortholog (Fig. 4).
Identification of candidate odorant binding proteins. Fifty-six candidate OBP proteins were identified from the data sets (Table 4; Additional file 6). Among these unigenes, 10 unigenes were incomplete due to the lack of a 5′ or 3′ terminus (Additional file 7). Forty-six unigenes were full-length putative OBP genes because they had complete ORFs. These unigenes had molecular weights ranging between 10 and 17 kDa and had a signal peptide sequence ranging between 16 and 23 amino acids at the N-terminus (Additional file 8). Insect OBPs can be classified into classical OBPs (six-cysteine conserved signature) and Minus-C (missing C2 and C5) and Plus-C (carries an additional conserved cysteine located between C1 and C2 and after C6). QM_comp02388, QM_comp07285, QM_comp08846, QM_comp10855, QM_comp21133, QM_comp21238 and QM_comp24139 had four conserved cysteines, which were minus-COBPs. Other OBPs had six conserved cysteines, which are classic OBPs. Plus-COBPs were not found in hymenopteran species. Through a homology search with known   5 and Additional file 9). The phylogenetic tree showed that all candidate OBP proteins were clustered with at least one Hymenoptera ortholog (Fig. 5).
Identification of candidate odorant receptors. Thirty candidate OR proteins were identified from the data sets (Additional file 6). Among these unigenes, 21 unigenes were incomplete due to the lack of a 5′ or 3′ terminus (Additional file 7). Nine unigenes were full-length putative OR genes because they had complete ORFs. These unigenes had molecular weights ranging between 4 and 53 kDa and had a signal peptide sequence ranging between 16 and 23 amino acids at the N-terminus (Additional file 8). Through a homology search with known proteins, the results showed that 95% of QM_comp14333 were orthologs of the proteins in C. cunea, and the orthologs of other OR sequences were also above 32% (Additional file 8  6 and Additional file 9). The phylogenetic tree showed that all candidate OR proteins were clustered with at least one Hymenoptera ortholog (Fig. 6).
Identification of candidate sensory neuron membrane receptors. Two candidate SNMP proteins were identified from the data sets (Additional file 6). QM_comp21591 was incomplete due to the lack of a 3′ terminus (Additional file 7). QM_comp09081 was a full-length putative SNMP gene because it had complete ORFs.   (Fig. 7 and Additional file 9). The phylogenetic tree showed that QM_comp21591 fell into the same clade as the insect SNMP1 group, and QM_comp09081 fell into the same clade as the insect SNMP2 group (Fig. 7).

Discussion
In this study, the major sensory genes (i.e., CSPs, GRs, IRs, OBPs, ORs and SNMPs), which perhaps regulate Q. mendeli to locate its host L. invasa, are first reported, providing valuable information for exploring how parasitoids use sensory genes to locate gall-making pests. CSPs are widespread in the antenna and other chemical sensory organs of insects 40 and are involved in the chemical perception and related behavior of insects 41 . Compared to the total number of insects in the world, CSPs have been identified in only a few species of insects to different degrees, such as Coleoptera 42 , Hemiptera 43 , Hymenoptera 12 , Lepidoptera 44 and Orthoptera 45 , whose numbers show interspecific diversity. For example, the number of CSPs varies from 4 CSPs in Drosophila melanogaster to 22 in B. mori 46 . In this study, 3 candidate CSPs were identified, which are less than those in the parasitoids Ch. cunea (11), Sclerodermus sp. (10), Me. pulchricornis (8) and Tric. dendrolimi (7) 2,4,12,16 and are greater than those in the parasitoids Cot. chilonis (2) and Ap. ervi (2) 6,11 . Previous research confirmed that generalists seem to be specifically suited for the processing of odorant mixtures, and they respond in a similar manner to plant volatiles 47 . For example, Ch. cunea is a generalist and   www.nature.com/scientificreports/ QM_comp26540 may function as a chemosensor, which is involved in the process of recognizing plant volatiles from eucalyptus when Q. mendeli searches its host, L. invasa. GRs are widespread in gustatory organs of insects that respond to various taste-related soluble compounds, and cuticular hydrocarbons and odorants, such as sugars, amino acids, salts, bitter compounds, CO 2 and pheromones, can be recognized and combined by GRs 53,54 . To date, GRs in some model insects with genome reports have been identified, such as A. gambiae (76), B. mori (69), D. melanogaster (68) and N. vitripennis (58) 55,56 . In this study, 10 candidate GRs were identified, which was similar to other parasitoids, such as An. japonicus (8) 13 , Sclerodermus sp. (6) 2 and M. mediator (6) 8 . This could be attributed to the sequencing depth and species-functional specificity of GRs 53 . DmelGR5 and DmelGR64 in D. melanogaster are receptor proteins for sweet taste and are used to detect glucose, sucrose, maltose, maltitol and cottonseed sugar 57 . For Q. mendeli, females that were fed a honey solution or honey solution + young eucalyptus leaves lived for a longer time than those who underwent other treatments, including flowers, gall leaves, water, galled leaves + honey solution, no food and young leaves 23 . Therefore, GRs in Q. mendeli should play a key role in recognizing sugar and fresh eucalyptus leaves via various soluble compounds 58 . The phylogenetic tree showed that QM_comp00164 and QM_comp22611 were the same clade as the proteins in Trib. castaneum, QM_comp03300, QM_comp22814, and QM_comp26507 were the same clade as the proteins in D. melanogaster, and other proteins in Q. mendeli were the same clade as the proteins in N. vitripennis. Thus, GRs of Q. mendeli may share high homology and closely cluster with the proteins in D. melanogaster, N. vitripennis, and Trib. castaneum. Interestingly, QM_comp11847 was in the same clade as the sugar receptor NvitGR1, which was used to recognize the only source of nutrients from host Lucilia caesar L. for the offspring of N. vitripennis 59,60 . Thus, QM_comp11847 may be involved in recognizing host organisms and sugars, which helps Q. mendeli to quickly access energy from these molecules.
IRs, which evolve from the ionotropic glutamate receptor (iGluR), are a new class of sensory proteins mainly in taste organs/sensilla that respond to food components, such as sugars, salts, water and bitter compounds, and detect small temperature differences [61][62][63] . In this study, 21 candidate IRs were identified, which was more than in the parasitoids Me. pulchricornis (19) 12 65,66 . Taking the oviposition features into consideration, we deduced that female Q. mendeli may be capable of sensing surface heat on galls related to L. invasa damage, which requires further exploration. OBPs are crucial in insect olfactory perception and are the first step in the recognition of chemical stimuli from the outside environment 3 . In some model insects, OBPs have been identified to different degrees, such as B. mori (57) 67 , D. melanogaster (51) 68 , Trib. castaneum (46) 69 and A. gambiae (44) 70 . In this study, 56 candidate OBPs were identified, which was more than in the parasitoids Ae. bambawalei (54) 14 , Ch. cunea (25) 4 , Tric. dendrolimi (24) 16 , M. mediator (20) 8 , Me. pulchricornis (16) 12 , T. japonicum (15) 3 , Ap. ervi (15) 11 , Sclerodermus sp. (10) 2 , Cop. floridanum (8) 15 , Cot. chilonis (8) 6 and As. hispinarum (8) 17 . The number of Q. mendeli OBPs identified was less than that in the parasitoid Cot. vestalis (74) 5 . Previous studies revealed that OBPs in parasitoids play a key role in binding and transporting hydrophobic odorants from the environment to sensory receptors 71 . In Q. mendeli, a significant behavioral response to the gall volatiles d-limonene and decanal was observed (unpublished data). Relevant QmenOBPs function in chemical sensing of these volatiles characterizing L. invasa and its shelter host eucalyptus trees, which bioinformatics analysis could help to target. The phylogenetic tree showed that 15 QmenOBPs were the same clade as 5 DmelOBPs of D. melanogaster, and 15 QmenOBPs were the same clade as NvitOBP1 of N. vitripennis, and 3 CfloOBPs of C. floridanum, which suggested that most OBPs of Q. mendeli may share high homology and closely cluster with the proteins in C. floridanum, D. melanogaster, and N. vitripennis. The evolutionary relationship of Q. mendeli OBPs, as inferred in the phylogenetic tree, indicated that they are orthologous sequences due to the absence of monophyletic groups. Our results showed that QM_comp21238 is the same clade as MmedOBP10 of M. mediator, which is involved in the process of recognizing β-ionone and Nonanal when they find the location of their hidden host A. segetu 51,72 . Thus, QM_comp21238 may also ORs are thought to play critical roles in the perception of chemosensory stimuli by insects 54 . The number of ORs in parasitoids vary greatly 1,4 . In this study, 30 candidate ORs were identified, which was more than in the parasitoid Tric. dendrolimi (9) 16 , Mi. cingulum (9) 9 and Sclerodermus sp. (8) 2 . Previous studies revealed that the OR of M. mediator play an important role in recognizing plant volatiles, such as nonanal and farnesene, which provided a key start to manipulate and develop ORs in wasps to find hosts and use them as biological tools for pest control 73 . The phylogenetic tree showed that QmenORs were spread across the tree branches and clustered with homologous ORs from other species, which suggested that QmenORs may be functionally conserved. The RNAi investigation of the role of MmedOrco, the M. mediator ortholog of Drosophila Or83b, supported the assumption that this highly conserved gene plays a similar role in insects 73,74 . QmenORs may function as chemoreceptors to recognize plant volatiles from eucalyptus. Our results showed that QM_comp20892 is in the same clade as DmelOR10a of D. melanogaster, which plays a role in responding to odorants such as methylsalicylate and acetophenone 75,76 . For Q. mendeli, QM_comp20892 may be involved in the process of recognizing similar odorants or ligands.
SNMPs are involved in cellular signal transduction and play a role in odor detection 5 . Two SNMPs are normally broadly identified in different insects, e.g., parasitoids Ch. cunea and Sclerodermus sp. 2,4 . It has been reported that SNMP1 and SNMP2 are both expressed in antennae sensilla and have different expression patterns 4,77 . In Ch. cunea, CcunSNMP1 is a morphine receptor of neurons, and CcunSNMP2 is mainly expressed in supporting cells and the lymph of antennal sensilla 4 , while the location and expression patterns of SNMPs in Q. mendeli should be further studied since this information should be associated with their functions. SNMP1 of D. melanogaster is involved in pheromone detection and enhances the Ca 2+ responses served in signal transduction 78 . SNMP1 of M. mediator was determined to participate in both pheromone and general odor detection 79 . In contrast, the general functional mechanism of SNMP2 in parasitoids is still poorly understood. The phylogenetic tree showed that QM_comp21591 was the same clade as TcasSNMP1 of Trib. castaneum and QM_comp09081 was the same clade as CfloSNMP2 of C. floridanum, suggesting that QM_comp21591 and QM_comp09081 shall be SNMP1 and SNMP2 in Q. mendeli respectively. In addition, Q. mendeli is a uniparental www.nature.com/scientificreports/ parasitoid that is not required in the male search for mating 22 . Thus, the function of QM_comp21591 and QM_comp09081 may include an oviposition pheromone receptor rather than a sex receptor and a membrane protein with unknown functions, which needs to be further explored. Chemical detection involves a series of complicated processes that require participation and interactions by multiple cascades of sensory proteins. Insect sensory proteins are capable of functional cooperation and division. Firstly, OBPs and CSPs are both chemically binding proteins to various odorants and can also respond to the same chemicals, e.g., MmedOBP10 and MmedCSP1 are involved in the process of recognizing β-ionone and nonanal when they find the location of their hidden host A. segetu 51,72 . Secondly, ORs and IRs are both chemoreceptors, while there are differences in the process of recognizing odor substances 80 . For example, IRs are better at detecting long-lasting odor pulses, and they are less sensitive, suggesting that they are better at close-range odor detection. In contrast, ORs are more sensitive and better at resolving brief (low molecular flux) pulsed stimuli 80,81 . Moreover, features of functional organization have emerged between behavioral response profiles of OBPs and electrophysiological response profiles of ORs 75 . Therefore, the sensory genes in Q. mendeli should systematically act on the process of locating their gall-making host, and the biological functions of these genes and their products are still poorly known. Overall the sensory genes of the wasp reported here provide valuable insight into the molecular mechanisms of olfaction, which help pave the way for the host location of Q. mendeli in gall-making pests.