Transcriptome changes associated with Tomato spotted wilt virus infection in various life stages of its thrips vector , Frankliniella fusca ( Hinds )

Persistent propagative viruses maintain intricate interactions with their arthropod vectors. In this study, we investigated the transcriptome-level responses associated with a persistent propagative phytovirus infection in various life stages of its vector using an Illumina HiSeq sequencing platform. The pathosystem components included a Tospovirus, Tomato spotted wilt virus (TSWV), its insect vector, Frankliniella fusca (Hinds), and a plant host, Arachis hypogaea (L.). We assembled (de novo) reads from three developmental stage groups of virus-exposed and non-virus-exposed F. fusca into one transcriptome consisting of 72 366 contigs and identified 1161 differentially expressed (DE) contigs. The number of DE contigs was greatest in adults (female) (562) when compared with larvae (first and second instars) (395) and pupae (preand pupae) (204). Upregulated contigs in virus-exposed thrips had BLASTX annotations associated with intracellular transport and virus replication. Upregulated contigs were also assigned BLASTX annotations associated with immune responses, including apoptosis and phagocytosis. In virus-exposed larvae, Blast2GO analysis identified functional groups, such as multicellular development with downregulated contigs, while reproduction, embryo development and growth were identified with upregulated contigs in virus-exposed adults. This study provides insights into differences in transcriptome-level responses modulated by TSWV in various life stages of an important vector, F. fusca. INTRODUCTION Persistent and propagative viruses replicate in different tissues of arthropod vectors, resulting in intricate interactions between them. Several studies have investigated the effects of such interactions on arthropod vectors [1–3]. Thrips transmitted Tomato spotted wilt virus (TSWV) (Tospovirus, Bunyaviridae) is a propagative and single-stranded RNA virus. TSWV is acquired by thrips during the early larval stages [4]; upon acquisition, the virus replicates in the midgut tissues and salivary glands [5–8]. The effects of TSWV replication on the fitness of its vectors Frankliniella occidentalis (Pergande) and Frankliniella fusca (Hinds) (Thysanoptera: Thripidae) have been extensively studied [9–14]. Since TSWV replicates in thrips as well as in host plants, it is difficult to determine whether TSWV-induced effects on thrips fitness are host plant-mediated indirect effects or direct effects of virus replication. Nonetheless, both negative and positive influences of TSWV on thrips biology and behaviour have been observed [12, 13, 15]. TSWV infection negatively affected the survival rate of F. occidentalis [16]. Our previous study documented the delayed adult emergence, decreased survival rate and reduced feeding ability of F. fusca when exposed to TSWV [12]. Virus-exposed F. fusca also developed into smaller adults when compared with non-virus-exposed adults [17]. In contrast, TSWV infection positively influenced the feeding, fecundity, survival and development rate of F. occidentalis [10, 13, 15, 18]. Similarly, increased oviposition in virus-exposed F. fusca was associated with TSWV infection [12]. Despite the availability of extensive evidence for TSWVinduced effects on thrips preference and fitness, the molecular changes associated with TSWV infection in thrips are just beginning to be explored through transcriptomic and Received 14 February 2017; Accepted 27 June 2017 Author affiliations: Department of Entomology, University of Georgia, Tifton, GA 31793, USA; Department of Plant Pathology, University of Georgia, Tifton, GA 31793, USA; Department of Plant Pathology, Kansas State University, Manhattan, KS 66506, USA. *Correspondence: Rajagopalbabu Srinivasan, babusri@uga.edu

Since TSWV replicates in thrips as well as in host plants, it is difficult to determine whether TSWV-induced effects on thrips fitness are host plant-mediated indirect effects or direct effects of virus replication.Nonetheless, both negative and positive influences of TSWV on thrips biology and behaviour have been observed [12,13,15].TSWV infection negatively affected the survival rate of F. occidentalis [16].Our previous study documented the delayed adult emergence, decreased survival rate and reduced feeding ability of F. fusca when exposed to TSWV [12].Virus-exposed F. fusca also developed into smaller adults when compared with non-virus-exposed adults [17].In contrast, TSWV infection positively influenced the feeding, fecundity, survival and development rate of F. occidentalis [10,13,15,18].Similarly, increased oviposition in virus-exposed F. fusca was associated with TSWV infection [12].
Despite the availability of extensive evidence for TSWVinduced effects on thrips preference and fitness, the molecular changes associated with TSWV infection in thrips are just beginning to be explored through transcriptomic and Downloaded from www.microbiologyresearch.orgby IP: 54.70.40.11On: Mon, 08 Apr 2019 21:56:58 proteomic approaches.Few studies have characterized transcripts [19][20][21] and proteins [22] induced by TSWV infection in thrips.All of the studies were conducted exclusively with one thrips species, F. occidentalis.F. fusca is the primary vector of TSWV in many crops, including peanut (Arachis hypogeae L.) in the southeastern United States [23,24].Both F. occidentalis and F. fusca are mainly flower feeders; F. fusca, however, seems to adapt much better to foliar feeding and colonizes seedlings more efficiently than F. occidentalis of hosts such as peanut [25][26][27].Thus far, nine thrips species have been identified as TSWV vectors [28,29].However, studies on TSWVinduced molecular changes in thrips have been limited to F. occidentalis.Jacobson et al. [30] examined interactions between TSWV isolates and different isofemale lines of Thrips tabaci (Linderman) from several locations, and demonstrated variation in vector competence among T. tabaci isofemale lines, as well as variation in transmission by T. tabaci among TSWV isolates [30].These results revealed that even within one thrips species, thrips-TSWV interactions could be different.Therefore, it is likely that TSWV-induced molecular changes in thrips could also vary with thrips species, and they need to be examined.
For successful TSWV transmission, early instar thrips larvae must acquire the virus.If adults acquire TSWV for the first time, they cannot transmit the virus.Such stage-specific acquisition and inoculation of TSWV by thrips [4] could potentially lead to varying molecular interactions at each life stage.Information on TSWV-induced stage-specific changes in thrips at transcript levels is limited and confined to one species.Very recently, Schneweis et al. [20] documented the stage-specific responses of F. occidentalis to TSWV [20].Given the ecological and host utilization differences among various Tospovirustransmitting thrips species, it is critical to examine the differences in TSWV-induced effects among different vector species.
The main objective of this study was to examine the transcriptome-level responses of virus-exposed F. fusca at various life stages.Total RNA samples from virus-exposed and non-exposed thrips at the larval, pupal and adult stages were sequenced using a HiSeq Illumina pair-end sequencing platform.Reads from all of the life stages were used to perform one de novo assembly and differential expression analysis was conducted between the virus-exposed and nonvirus exposed life stages of thrips.

Transcriptome assembly and differential expression analysis
Quality reads from virus-exposed and non-virus exposed thrips from all life stages were combined and assembled into 72 366 contigs.The de novo assembly consisted of an average contig length of 2161 bases and N50 of 3702 bases (Table 1).Following the assembly, clean reads from each experimental replicate were mapped back to the assembled contigs.The number of reads mapping to each contig was measured and normalized expression values were calculated across experimental replicates.In general, the expression values showed concordance among replicates (Fig. S1, available in the online Supplementary Material).Subsequently, 1161 differentially expressed (DE) contigs with a false discovery rate (FDR)adjusted P-value of <0.001 and a log 2 fold change (FC) !1 in expression levels were identified (Fig. 1).The number of DE contigs in larvae (first and second instars), pre-and pupae, and female adults was 395, 204 and 562, respectively (Table S1).Of all the DE contigs, only 8 (0.7 %) were common among three life stages, while 14 (1 %) to 85 (7 %) of the contigs were common between two life stages.The clear majority of contigs were unique to each life stage.Among the three life stages, larvae had the greatest number of unique contigs (90 %), followed by adults (81 %) and pupae (46 %) (Fig. S2).Validation of differential expression analysis by real-time quantitative reverse transcriptase-PCR (qRT-PCR) demonstrated that the FC direction of 15 randomly selected contigs in 2 experimental replicates was 86 and 73 %, respectively, similar to that of RNA-Seq (Table S2).
Provisional annotations of the complete list of differentially expressed contigs Among 1 161 DE contigs, 967 contigs contained BLASTX matches to the National Center for Biotechnology Information non-redundant (NCBI nr) sequence database.Gene ontology (GO) analysis of the complete list of DE contigs identified a total of 38 GO terms associated with biological process, molecular function and cellular component categories at level 5 (Fig. 2).Under the biological process, the cellular macromolecule metabolic process was the most dominant GO term, consisting of 15 % of contigs, followed by the protein metabolic process (14 %), embryo development (10 %) and gene expression (9 %).In the molecular function category, the annotated contigs were mostly associated with nucleotide binding (30 %), RNA binding (15 %) and kinase activity (11 %).Only two GO terms, namely the plasma membrane and intracellular component, were assigned to the cellular component category.Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis mapped the DE contigs to 61 biochemical pathways (Table S3).The top 10 KEGG pathways included purine metabolism, glycolysis, glyoxylate metabolism and others (Table 2).

Provisional annotations of differentially expressed contigs at various life stages
Differentially expressed contigs at the larval stage In virus-exposed F. fusca larvae, 219 contigs were upregulated and 176 contigs were downregulated.Upregulated contigs were assigned to 13 GO terms, including ion transport, protein and lipid metabolism, and response to stress (Fig. 3a).Downregulated contigs were assigned nine GO terms under the biological process category, including multicellular organismal development, transport and DNA metabolic process (Fig. 3b).In the molecular function category, six and eight GO terms were assigned to upregulated and downregulated contigs, respectively (Fig. 3c).Contigs under the molecular function category included nucleotide binding, hydrolase and transporter activities.Under the cellular component category, three GO terms, such as intracellular membrane-bound organelles [Node score (NS) 6.86], cytoplasm (NS 6.52) and protein complex (NS 8), were assigned to both upregulated and downregulated contigs.The GOenrichment analysis demonstrated that DE contigs representing GO terms such as cellular amino acid metabolic process, catalytic activity, lipid binding and receptor activity were significantly (P<0.05)enriched (Table S4).
Differentially expressed contigs at the pupal stage Among the 204 DE contigs in virus-exposed F. fusca pupae, 150 were upregulated, while 54 were downregulated.In the biological process category, 14 GO terms, including growth, reproduction, translation and cell death, were assigned to upregulated contigs (Fig. 4a), while downregulated contigs were assigned to one GO term: metabolic processes.Under the molecular function category, downregulated contigs were assigned to ion binding.The upregulated contigs had nine GO terms, including GTPase, ATPase, oxidoreductase and isomerase activities (Fig. 4b).In the cellular component category, five GO terms, viz.mitochondrion (NS 6), endoplasmic reticulum (NS 7), ribosome (NS 13), protein complex (NS 9) and cytoskeleton (NS 5.72), were associated with upregulated contigs.Downregulated contigs were assigned a nucleus GO term (NS 5).Enrichment analysis revealed that GO terms such as vesicle-mediated transport, protein complex organization, hydrolase activity, RNAbinding and endoplasmic reticulum were overrepresented in virus-exposed pupae (Table S4).
Differentially expressed contigs at adult stage Differential expression analysis identified 478 and 84 contigs upregulated and downregulated, respectively, in virus-exposed F. fusca adults.Analysis of GO annotations identified 26, 14 and 3 GO terms under the biological process, molecular function and cellular component categories, respectively.Among the downregulated contigs, three, four and one GO terms were identified under the biological process, molecular function and cellular component categories, respectively.GO terms such as translation, viral process, cytoskeleton organization, protein transport, reproduction and growth were assigned to upregulated contigs under the biological process category (Fig. 5a), while contigs associated with oxidation reduction, organic substance metabolic process and primary metabolic process were assigned to downregulated contigs.In the molecular function category, GO terms like nucleotide binding, transporter activity, protein kinase activity, and peptidase activity were assigned to upregulated contigs.The downregulated contigs were associated with GO terms such as catalytic activity, ATP binding and zinc ion binding (Fig. 5b).Under the cellular component category, most of the upregulated contigs were associated with ribosome (NS 49), protein complex (NS 53), plasma membrane (NS 17), cytoskeleton (NS 22.16) and microtubule organizing centre (NS 6), while the downregulated contigs were associated with the nucleus (NS 8.08).The contigs associated with developmental process, reproduction, embryo development, structural molecule activity, RNA binding and cytoplasm were significantly overrepresented in virus-exposed adults (Table S4).
Color Key

Value
NAd InAd NLv InLv NPu InPu 0 2 4 6 Following GO analysis of DE contigs, the contigs were classified into categories that might be associated with virusvector interaction processes.DE contigs were classified into four categories, including (1) virus entry and movement, (2) virus replication, (3) vector response to virus infection,and (4) the effects of virus infection on vector fitness (Table 3).

Virus entry and movement
Virus entry and movement are the initial processes of virus infection.In virus-exposed F. fusca, several contigs associated with virus entry and movement were identified.Contigs with the BLASTX annotations clathrin and adaptor proteins, which are involved in intracellular transport, were upregulated by 9-and 10-fold, respectively, in virus-exposed adults (Table 3a).Also, one upregulated contig in virusexposed adults was annotated as spike proteina viral fusion protein responsible for mediating the entry of coronaviruses.In addition, homologues of aminopeptidase N, a receptor for a plant virus in aphids, were upregulated in virus-exposed larvae and adults.

Virus replication
In virus-exposed adults, several contigs known to facilitate the replication of animal-and plant-infecting viruses were upregulated.Six contigs with an annotation associated with the replication of animal-infecting viruses, including 40 s ribosome protein 15, ankyrin repeat domain, histone deacetylase 1 isoform 1 and host cell factor 1 protein, were upregulated by at least sixfold (Table 3b).Also, an upregulated contig was annotated as 60 s acidic ribosomal proteins p0, which is associated with replication of a plant virus, Potato virus A.

Vector response to virus infection
Following TSWV-exposure, several contigs with functional annotations of transport, lipid and carbohydrate metabolic processes, and response to stress were upregulated in F. fusca at all life stages.GO categories under biological process, such as response to biotic and abiotic stimulus, were  specific to virus-exposed adults.Further, contigs associated with three immune pathwaysapoptosis, proteolysis and phagocytosiswere upregulated in virus-exposed thrips (Table 3c).The contigs associated with each pathway are described in detail below.

Apoptosis
A total of 17 homologous contigs that have been known to play a role in apoptosis were identified in virus-exposed F. fusca.Out of 17, 12 homologous contigs were present in virus-exposed adults, 4 were present in virus-exposed pupae and 1 was present in virus-exposed larvae.Some of the contigs were annotated as 40 s ribosomal protein s3, calmodulin, deoxyribonuclease I and cathepsin b (Table 3c).Homologues of cathepsin were also present in larvae and pupae.

Proteolysis
Eight and three contigs associated with proteolysis were upregulated in virus-exposed adults and larvae, respectively (Table 3c).Four homologous contigs of 26 s proteasome non-ATPase regulatory subunit 1 were upregulated in virus-exposed adults by 11-fold.Further, upregulated contigs were annotated as proteins and enzymes that mediate protein ubiquitination and degradation, including e3 ubiquitin-protein ligase ubr5, serine protease, f-box-only protein 11 and proteasome subunit beta type-7 in virusexposed adults.In larvae, homologues of serine proteases were upregulated by twofold.

Phagocytosis
BLASTX annotation identified the upregulation of homologous contigs of tubulin alpha chain and calreticulin family   proteins in virus-exposed adults and pupae.In virusexposed larvae, a contig annotated as beta-mannosidasea lysosomal enzyme associated with degradation of glycoproteinwas upregulated by twofold (Table 3c).
Further, contigs associated with signal transduction that trigger immune responses were present in the upregulated contigs of virus-exposed adults and larvae.In virus-exposed adults, contigs annotated as ribosomal protein s6 kinase, serine threonine protein kinases and ras-like GTP-binding protein were upregulated, while in larvae, contigs annotated as rho GTPase-activating protein 190-like and zinc finger protein zpr1-like were upregulated (Table 3c).

Effects of virus infection on vector fitness
Exposure to TSWV resulted in the downregulation of contigs associated with multicellular organismal development in F. fusca larvae.The contigs involved in neuron development and cell division were also downregulated by several folds in virus-exposed larvae (Table 3d).In virus-exposed adults and pupae, reproduction-, embryo development-, cell differentiation-and growth-related contigs were upregulated.Further, egg production-associated contigs, namely vitellogenin, were upregulated by ninefold in virus-exposed pupae.Also, a homologue of lipophorin precursor was upregulated in virus-exposed adults.

DISCUSSION
In this study, we examined the transcriptome-level changes associated with TSWV infection in F. fuscaan important vector of TSWV in the southeastern United States.This study also increases our understanding of the stage-specific responses of F. fusca to TSWV infection.The DE contigs identified in this study provide hypotheses about putative genes associated with TSWV movement and replication, the response of thrips to TSWV infection and the effect of TSWV infection on thrips fitness.

Differentially expressed contigs associated with thrips biology and metabolism
The upregulated contigs in virus-exposed F. fusca had GO annotations such as organization of cytoskeleton, metabolic process, transport and binding activities.Viruses interact and reorganize host cytoskeleton components, including actin filaments and microtubules, for intercellular trafficking and various infection processes, including assembly and exit [31][32][33][34].Viruses also exploit host resources, including nucleic acids and proteins, to reproduce viral particles.They increase the availability of the nucleotide pool by altering cellular metabolisms and increasing the rate of RNA breakdown [35,36].In this study, the KEGG pathway analysis mostly identified metabolism-associated pathways such as carbohydrate and lipid metabolisms in the upregulated contigs of virus-exposed thrips at all life stages.TSWV RNA segments are enclosed by host-derived lipid envelopes, suggesting that increased lipid metabolism in thrips could fa cilitate envelope formation in replicating virions.Manipulation of host metabolism to facilitate virus replication has also been observed in other viruses, with an example being human cytomegalovirus [37].

Differentially expressed contigs associated with virus movement and replication
To successfully initiate virus replication, viruses must enter host cells.Several animal-infecting viruses, including members of Bunyaviridae, such as nairoviruses and hantaviruses, enter host cells through receptor-mediated endocytosis [38][39][40].Clathrin and adaptor proteins are important components of receptor-mediated endocytosis [41].
Contigs with annotations pertaining to clathrin and adaptor proteins were upregulated several-fold in virus-exposed adults.Therefore, it is possible that clathrin-mediated endocytosis could be influencing TSWV infection in thrips.

Differentially expressed contigs associated with immune response
Insects rely on innate immunity to fight against invading pathogens [42,43].Upon viral infection, several immune pathways are launched in insects, including the RNAi, Toll, JAK/STAT [21], phagocytosis, apoptosis [22], proteolysis [20] and JNK pathways [19].When compared with F. fusca adults, fewer innate immunity-related contigs were upregulated in virus-exposed larvae and pupae.Viruses are known to suppress immune responses regulated by their hosts [44,45].In virus-exposed adults, along with antiviral immune responses, contigs known to inhibit immune genes were also upregulated (Table 3c).For instance, cactin is a negative regulator of cactus gene that is associated with the activation of the Toll pathway [46].We documented the upregulation of a homologue of cactin in virus-exposed F. fusca adults.Further, dicer is a member of RNase III family, which cleaves viral particles and activates the RISC of the RNAi pathway, was also upregulated [47].In virusexposed adults, homologous contigs of RNase III inhibitor were upregulated by 11-fold.In addition, heat-shock protein 70, which inhibits apoptosis, was also upregulated in virusexposed adults [48,49].No. of contigs     Differentially expressed contigs associated with thrips fitness Exposure to TSWV affected the expression of several fitness-related contigs in thrips.We observed upregulation of contigs associated with egg production, embryo development and growth in virus-exposed F. fusca adults and pupae.In our previous study, we demonstrated that virusexposed F. fusca produced significantly more eggs than non-virus-exposed adults [12].This study seems to provide evidence at the transcriptome level for upregulation of oviposition in F. fusca.In addition, in this study, developmentrelated contigs were downregulated in virus-exposed larvae.
In contrast to our results, Badillo-Vargas et al. [22] compared the relative abundances of proteins between virusexposed and non-virus exposed F. occidentalis, and documented the upregulation of proteins associated with development in virus-exposed F. occidentalis larvae.This discrepancy concerning the effects of TSWV infection on thrips development could be due to differences in approach (proteomics versus transcriptomics), variations in the TSWV accumulation in the larvae used for each study and differences in the TSWV acquisition access time provided (AAP) to the larvae.In this study, F. fusca larvae were reared on TSWV-infected leaflets until they were used for RNA extraction.Badillo-Vargas et al. [22] only provided thrips with a three-hour AAP and moved thrips to their original rearing hosts, healthy green bean pods.This presumably also limited the indirect effects of virus-infected plant material on thrips.Feeding on virus-infected plants alone could alter gene expression in insect vectors.Wang et al. [43] demonstrated that white black planthopper reared on virus-infected plants had 700 genes that were differentially expressed compared to healthy control groups [43].
TSWV infection increased nutrients such as free amino acids in virus-infected plants [12,50].TSWV infection also affects host-plant quality by decreasing water storage capacity and photosynthesis [51].Thus, extended periods of exposure to TSWV-infected leaflets could have negatively affected the development of F. fusca larvae in this study.Our earlier study also demonstrated direct negative effects of TSWV infection on F. fusca survival rate and developmental time [12].The current study provides evidence at the transcriptome level for some of the observed macrolevel fitness effects of TSWV infection on F. fusca.
Comparison of the transcriptome-level responses of TSWVexposed F. fusca to F. occidentalis identified some similarities and differences between two thrips species.In this study and in Zhang et al. [21], exposure to TSWV induced upregulation of the contigs associated with vesicle-mediated endocytosis, insect development and lipid metabolism.
When compared with F. fusca, more immune systemrelated pathways, such as the Toll pathway, JAK-STAT and RNA interference, were documented in F. occidentalis by Zhang et al. [21].Unlike in our study, Zhang et al. [21] pooled thrips from all life stages and performed DE analysis between TSWV-exposed and non-exposed F. occidentalis without biological replicates.stage-specific DE analysis of Based on functional annotations, TSWV exposure induced the expression of similar contigs in both F. fusca and F. occidentalis, but the contigs were differentially regulated in each species.For example, the GO term 'receptor activity' associated with contigs specific to F. occidentalis was enriched at the pupal and adult stages, but it was only enriched in F. fusca larvae.Proteolysis-associated contigs were downregulated in F. occidentalis larvae in response to TSWV infection, but there was an upregulation of serine proteases in F. fusca larvae.Insect cuticle-related contigs were one of the enriched contigs in F. occidentalis adults, but they were mostly downregulated in TSWV-exposed larvae and pupae.In F. fusca, the contigs associated with the BLASTX annotation for cuticular proteins were upregulated in virusexposed larvae and adults.The observed differences in the profiles of DE transcripts between F. fusca and F. occidentalis could be due to innate differences in the thrips species, virus isolates, experimental designs and DE analyses used in the studies.Schneweis et al. [20] exposed F. occidentalis larvae to TSWV for only 3 h.In contrast, F. fusca were reared on TSWV-infected leaflets until they were used for RNA extraction.Further, we pooled the first and second instars for the larval stage, the pupae and prepupae for the pupal stage, and female adults up to 2 d old for the DE analysis.However, in F. occidentalis, DE analysis was performed with first instar larvae, prepupae and 24 h-old adults (male and female).Schneweis et al. [20] carried out differential expression analysis using genome-guided assemblies, while we first performed de novo assembly on F. fusca reads by pooling reads from all of the life stages and then conducted differential expression analysis.
The DE analysis in this study identified several contigs with putative roles in viral processes, thrips development, growth and reproduction.The contig functions need to be validated through gene-silencing experiments.Once proven, these candidates could be evaluated as likely targets for pest management.For instance, the contigs associated with egg production, virus movement and virus replication identified in this study could potentially be exploited for RNAi-mediated thrips and TSWV management.Unlike other economically important insect vectors, the genomic resources for thrips are very limited.The development of F. occidentalisexpressed sequence tags by Rotenberg et al. [52] initiated an effort to build sequence resources and annotate putative genes in thrips [53].The transcriptomic data generated in this study have already been submitted to the NCBI (SUB2623511).The transcriptomic data generated in this study will further enrich the genomic resources got thrips and facilitate functional genomic studies in other thrips species and other insects.
TSWV infection in subsamples of a virus-exposed thrips colony was tested by qRT-PCR.A total of 30 samples (two thrips pooled for each sample) were used for the study.QRT-PCR was conducted using N-gene-specific primers as previously described [55][56][57].All the tested thrips were positive for TSWV, and the TSWV N-gene copy numbers in thrips ranged from 4316 to 6 200 000.
Total RNA extraction and library preparation for sequencing Total RNA was extracted from virus-exposed or non-virusexposed larvae (first and second instars), pupae (including pre-pupae) and female adults (up to 2 d old) (six treatments in total).For each treatment, about 35 individual thrips were pooled and treated as one experimental replicate.A total of three experimental replicates were used per treatment (N=105 individual thrips per treatment).Virusexposed and non-virus-exposed thrips were reared under the same laboratory conditions as previously indicated, and all thrips were collected on the same day for RNA extraction.Total RNA was extracted using the RNeasy plant mini kit (Qiagen) and sent to Georgia Genomic Facility of the University of Georgia for RNA-Seq library preparation.Illumina sequencing libraries were constructed using TruSeq RNA sample preparation kits using at least 1 µg of the total RNA.Briefly, mRNAs (polyadenylated RNAs) were selected using oligo-dT.Subsequently, mRNAs were fragmented and first-and second-strand cDNA were synthesized

Transcriptome assembly and differential expression analysis
Bioinformatics software at Georgia Advanced Computing Resource Center, UGA (https://wiki.gacrc.uga.edu/wiki/Software) were used to process raw reads, perform de novo assembly and conduct differential expression analysis.
Trimmomatic was used to trim adapter primers and adapter sequences from the raw reads [58].Subsequently, quality reads were produced by removing three bases at the beginning and end of each read, setting the minimum read length threshold to 50 bases and discarding reads if the average quality of four bases fell below 20.Quality reads from virusexposed and non-virus-exposed thrips from all life stages (treatments) and experimental replicates were pooled to perform a single de novo assembly using the Trinity program with the parameters '-kmer 25-minimum-contiglength 500 bp-Passafy' [59].To evaluate the completeness of the assembled contigs (overlapping DNA segments), the Core Eukaryotic Genes Mapping Approach (CEGMA) program was used.CEGMA contains a core set of 248 proteins that are highly conserved diverse eukaryotes.CEGMA identifies homologues of those conserved proteins in the assembly [60].With CEGMA, all the 248 core proteins were identified in F. fusca transcriptome [61,62].
For differential expression analysis, cleaned reads from each experimental replicate were first mapped to the de novo assembled transcriptome individually using Bowtie software [63].The RSEM program then estimated the number of reads corresponding to each contig across the replicates, thereby providing experimental replicate-specific expression values [64].The expression values were normalized as fragment per kilobase of transcript per million mapped reads and trimmed mean m-values were calculated across the replicates.Also, the concordance in expression values between replicates was estimated by subjecting the data to generalized linear mixed models using Proc GLIMMIX in SAS (SAS Enterprise 4.2; SAS Institute).Subsequently, the DE contigs between treatments were identified using the Benjamin and Hochberg multiple-testing adjustment method with an FDR-adjusted P-value set to <0.001 [65] and a log 2 FC (ratio of virus-exposed/non-virus exposed expression levels) !1.

Validation of differential expression analysis through qRT-PCR
To validate differential expression analysis, 15 contigs were randomly selected and their expression levels were compared between virus-exposed and non-virus-exposed F. fusca.QRT-PCR was only performed on female adults up to 2 d old.About 60 virus-exposed or non-virus-exposed female adults were pooled.Each thrips pool was treated as one experimental replicate.Two experimental replicates were included for each contig.Total RNA extraction, cDNA synthesis and qRT-PCR were performed as described previously [55][56][57].Each targeted contig included two technical replicates and mean Ct values were calculated.Subsequently, the expression level of each contig was normalized to the expression level of a thrips reference gene (actin).Normalized expression levels of targeted contigs were estimated using the Pfaffl analysis method, which included the PCR efficiency (E) and Ct values of the target contigs and the reference gene, E target ct (target) /E ref ct (ref) [66].
Provisional annotations of differentially expressed contigs DE contigs were characterized by searching for homologous sequences against the NCBI nr database using BLASTX program at an E-value threshold of 10 À10 .Further, Java-based Blast2GO software (https://www.blast2go.com)was used to study the functional annotations of DE contigs [67].Using the GO database, GO terms were assigned to the contigs under biological process, molecular function and cellular components with an E-value hit filter of 10 À6 and an annotation cutoff of 55.To obtain detailed functional information, the GO terms were classified at GO level 5 and a node-score of 5 for the complete list of DE contigs, whereas the GO terms were assigned to DE contigs from each life stage at multilevel with a 5-node score.Further, pathway analysis was performed using the KEGG database with Blast2GO.To identify enriched GO terms in DE contigs, functional enrichment analysis was performed using Fisher's exact test.The background was set to the complete list of DE contigs and significantly enriched GO terms in each life stage were identified at P<0.05.

Fig. 1 .
Fig. 1.Heat map displaying expression levels of differentially expressed contigs in Frankliniella fusca at log 2 fold change !1 and FDR<0.001.The expression levels of each contig (row) in each sample (column) are depicted with a colour scale in which green represents low expression, black represents medium expression and red represents higher expression.The columns in the map include: non-virusexposed adults (NAd), virus-exposed adults (InAd), non-virus-exposed larvae (NLv), virus-exposed larvae (InLv), non-virus-exposed pupae (NPu) and virus-exposed pupae (InPu).Three experimental replicates (~105 thrips) of female adult thrips up to 2 d old, larvae (first and second instars), and pre-pupae and pupae were used for the analysis.

Fig. 2 .
Fig. 2. Gene ontology (GO) terms assigned to the complete list of differentially expressed Frankliniella fusca contigs by Blast2GO under the biological process, molecular function and cellular component categories at GO level 5.

Fig. 3 .
Fig. 3. Pie charts displaying Gene Ontology (GO) terms assigned by Blast2GO to upregulated (a) and downregulated (b) contigs of virus-exposed Frankliniella fusca larvae (first and second instar) under the biological process category at multilevel with a 5-node score cutoff.The numbers after the GO terms represent node score values.The bar diagram demonstrates the number of contigs assigned to GO terms identified in upregulated and downregulated contigs of virus-exposed larvae under the molecular function category (c).at multilevel with a 5-node score cutoff.

Fig. 4 .
Fig. 4. Gene Ontology (GO) terms assigned to Frankliniella fusca contigs upregulated in virus-exposed pupae (pre-and pupae) under the biological process (a) and molecular function (b) categories by Blast2GO at multilevel with a 5-node score.The numbers following the GO terms represent node score values in the pie chart.

Fig. 5 .
Fig. 5. Gene Ontology (GO) terms mapped by Blast2GO to upregulated (a) contigs of virus-exposed Frankliniella fusca adults (females up to 2 d old) under the biological process category.Molecular function-associated GO terms (b) assigned to upregulated and downregulated contigs of virus-exposed F. fusca adults.GO terms in both categories were obtained with a node score value cutoff of 5 at multilevel.The numbers after the GO terms represent the node score values in the pie chart.

Table 3 .
[20]..occidentalisby Schneweis et al.[20]allowed for direct comparison of TSWV-induced responses between F. fusca and F. occidentalis at each life stage.Comparison of the findings revealed that in both species only a small percentage (1.6 % in F. fusca and <1 % in F. occidentalis) of contigs were differentially expressed following TSWV infection.In F. occidentalis, the highest distribution of DE contigs was in pupae (37 %), followed by larvae (36 %) and adults (26 %).In the case of F. fusca, more DE contigs were identified with adults (48 %) than larvae (34 %) and pupae (17 %).Also, while most of the DE contigs were upregulated by TSWV in F. fusca across life stages, in F. occidentalis, there were more downregulated than upregulated contigs in larvae and adults.Further, both studies demonstrated that the majority of TSWV-induced DE contigs were unique to each life stage.Only 0.6 % of DE contigs were shared by all life stages in F. fusca, while 0.7 % were shared in F. Unlike F. occidentalis, F. fusca larvae had the most unique contigs (90 %), followed by adults (81 %) and pupae (46 %).In F. occidentalis, the pupal stage had 83 % unique contigs, while the adult and larval stages had 78 and 68 % unique contigs, respectively. F performed to produce final cDNA libraries for sequencing.Sequencing was performed at the University of Texas Health Science Center, at San Antonio, Texas on llumina HiSeq 2000 platform using v3 paired-end 100 cycle sequencing settings.Six RNA-seq libraries were included in one lane and a total of three lanes were used.