Functional Immunomics of the Squash Bug, Anasa tristis (De Geer) (Heteroptera: Coreidae)

The Squash bug, Anasa tristis (De Geer), is a major piercing/sucking pest of cucurbits, causing extensive damage to plants and fruits, and transmitting phytopathogens. No genomic resources to facilitate field and laboratory studies of this pest were available; therefore the first de novo exome for this destructive pest was assembled. RNA was extracted from insects challenged with bacterial and fungal immunoelicitors, insects fed on different cucurbit species, and insects from all life stages from egg to adult. All treatments and replicates were separately barcoded for subsequent analyses, then pooled for sequencing in a single lane using the Illumina HiSeq2000 platform. Over 211 million 100-base tags generated in this manner were trimmed, filtered, and cleaned, then assembled into a de novo reference transcriptome using the Broad Institute Trinity assembly algorithm. The assembly was annotated using NCBIx NR, BLAST2GO, KEGG and other databases. Of the >130,000 total assemblies 37,327 were annotated identifying the sequences of candidate gene silencing targets from immune, endocrine, reproductive, cuticle, and other physiological systems. Expression profiling of the adult immune response was accomplished by aligning the 100-base tags from each biological replicate from each treatment and controls to the annotated reference assembly of the A. tristis transcriptome.


Introduction
The squash bug, Anasa tristis (De Geer), is a major pest of squash, pumpkin, watermelon, cucumber and cantaloupe production, causing substantial economic losses throughout its range [1]. Squash bug feeding causes extensive damage to stems resulting in wilting, fruit discoloration and pre/postharvest spoilage. Possibly more important, A. tristis vectors the causal bacterial agent of Cucurbit yellow vine disease (CYVD), Serratia marcescens [2]. This serious disease was first recorded in Texas and Oklahoma in 1988 and is now rapidly spreading through the West and Midwest. The epidemiology of CYVD ranges from little impact in some years to regional crop failure in others. Natural enemies of the squash bug include the scelionid wasp egg parasitoid Gryon pennsylvanicum (Ashmead), the tachinid parasitoid of adult squash bugs, Tricopoda pennipes) [3 5], and several predatory arthropods [6]. A. tristis adults enter diapause in the Fall, and in the Midwest A. tristis adults terminate diapause in May then proceed to feed on host plants [7,8]. A. tristis are univoltine exhibiting one generation each summer season beginning and ending with overwintering adults [8]. There are few effective biological agents or cultural practices for controlling this highly destructive pest aside from insecticides, thus development of more effective control measures are much needed. In their natural environments insects are subjected to a high incidence of opportunistic microbial infection [9] and parasitization [10]; thus immunoevasion and immunosuppression of the insect host is a strategy widely in use by parasitoids, nematodes, trypanosomes, bacteria and viruses [11]. Experimental immunosuppression of insects via malnutrition [10], or by dietary deficiencies of selenium [12 15] or of ascorbic acid [16 18] can enhance entomopathogen virulence. Immunosuppression of host larvae by injection of double stranded RNA directed against immune system components also increases susceptibility to microbial entomopathogens [19 23]. Field application of RNAi targeted specifically against insect pests is a promising new control approach [24 26]. Demonstration of field efficacy on a commercial scale has already been demonstrated by in situ oral vaccination of honey bees against the Israeli Acute Paralysis Virus by inclusion of RNAi against the virus in feeding solutions [27,28]. However, delivery of a sufficient quantity of RNAi to adversely impact pest life processes is far more challenging in the case of piercing/sucking insect pests, requiring plant-based expression of RNAi [29 32], although soaking or spraying with RNAi may become feasible in the future for some crops [26].
Deployment of RNAi against pest insects requires the discovery and generation of species specific gene silencing targets which disrupt tissues of targeted pests [31 33], or block central metabolic pathways such as arginine kinase [34] or mitochondrial Rieske iron-sulfur protein [35]. Immunosuppression as a biological control strategy [36] would first require discovery of viable, easily accessible gene silencing targets from among the squash bug immune system. The objective of this study therefore was to identify candidate immunosuppressive gene silencing targets suitable for control of A. tristis. The sequences of many inducible and constitutively expressed immune system components orthologous to those of other insect species were identified, and in addition a transcriptome was compiled which contains the vast bulk of A. tristis unigenes enabling a wide range of studies using this pest insect.

Insects, Infections and RNA Isolation
Colonies were founded by collection of adult squash bugs, Anasa tristis (De Geer) (NCBI Taxonomy ID:

Illumina Sequence Generation and Assembly Procedures
RNA was extracted from insects challenged with bacterial and fungal immunoelicitors, insects fed on different cucurbit species, and insects from all life stages from egg to adult. All treatments and replicates were separately barcoded for subsequent analyses, then pooled for sequencing in a single lane using the Illumina HiSeq2000 platform. RNA pools were submitted to University of Missouri Bond Life Sciences Center DNA Core for Illumina GAII sequencing Libraries were constructed TruSeq RNA sample preparation kit (#RS-930-2001, Illumina Inc., San Diego, CA, USA). Libraries were constructed according to the standard Illumina RNA-seq protocol (Part# 1004898 Rev. A, rev Sept 08; Illumina Inc., San Diego, CA, USA) from the pooled PCR products except for the fragmentation step as detailed [37,38].
Over 211 million ~100-base long Illumina single end reads were first cleaned of low quality bases Cold Spring Harbor Laboratory, Cold Spring Harbor, NY, USA). The remaining reads were screened for homology to contaminants (predominantly rRNA) leaving 115,101,048 reads. After trimming, filtering and cleaning the remaining reads were assembled into a de novo reference transcriptome using a local installation of the Broad Institute Trinity assembly algorithm [39]. The assembly was annotated using NCBIx NR, NCBI GVBRL, BLAST2GO [40,41], KEGG and other databases. Of the >130,000 total assemblies 37,327 were annotated . Contigs and singletons <80 bases were not analyzed. Annotation of KEGG orthologies (KOs) and metabolic pathway mapping was accomplished using the utilities provided by Kyoto Encyclopedia of Genes and Genomes. All sequence data discussed in this manuscript have been deposited in NCBI GenBank.

Identification of Differentially Regulated Genes
Expression profiling of A. tristis immune system responses were analyzed by normalizing the number of times sequence tags aligning with an mRNA was detected in controls [38]. To accomplish expression profiling the 100-base tags from each treatment, controls and replicates were independently aligned to the annotated assembly of the A. tristis reference transcriptome. The final assembly output was piped into a tab-delimited file that was imported into an Excel spreadsheet, which includes for each assembled contig the number of reads and the list of unique names for each read, to facilitate counting the contribution of different libraries for the final assembly. Genes those were upregulated or down regulated by a factor of three and that had an e-value less than 0.0001 are shown in Tables.

Anasa Tristis RNA-Seq
A total of 211,532,043 reads were acquired from the 11 barcoded input cDNA libraries. After quality trimming and filtering with FASTx 54.4% of the initial reads, 115,101,048 reads, remained for the assembly (Table 1). These were assembled using the program Trinity [39]. Over 211 million 100-base tags generated in this manner were trimmed, filtered, and cleaned, then assembled into a de novo reference transcriptome using the Broad Institute Trinity assembly algorithm. Annotation of this assembly [40] resulted in the identification of 37,327 unigenes (Table 1) of which the majority were genuine orthologs of Arthropoda ( Figure 1). The read length of those over 300 bp averaged 1588 bp (range 301 bp to 16,368 bp) ( Table 1). In comparison to other insect species this is likely an overestimate of the actual number of genes present in A. tristis. However a better estimate will not be available without alignment to a completed A. tristis reference genome. Genome size estimates of squash bug adults from this colony were estimated to be 1726 ± 29.2 Mb ( ) and 1782 ± 14.6 Mb ( ) [42]. a Trinity Assembly [40]. b Significant BLAST score (e > 10 6 ) using BLAST2GO [41].

Figure 1.
Anasa tristis transcriptome assembly Top BLAST hits across insects and other invertebrate species. Highest number of hits was against the insect species most closely related to A. tristis, and which has a sequenced genome, Acyrthosiphon pisum (green).
Genome Ontology annotation of these contigs demonstrated that the RNA pools constituting the assembly resulted in successful sampling of the major classes of cellular component, molecular function and biological processes. A complete set of cytoplasmic and mitochondrial tRNAs was assembled, and the majority of tRNA-acyltransferases. GO annotation revealed substantial acquisition of transcripts involved in reproduction and reproductive processes (2,043), central metabolism (7,818), cell and organellar proliferation, division and movement (9,580), physiological processes (3,571), and arthropod specific functions such as molting and pupation (911). KEGG maps generated using BLAST2GO demonstrated that major metabolic pathways were reasonably complete, constituting a majority of enzymes required. The experiments were specifically designed to sample A. tristis immune system processes and the annotation resulted in the identification of 394 GOs from this category, and in addition 169 immune system regulatory GOs, 76 coagulation, and two respiratory burst GOs were noted ( Figure 2). Contamination with squash and microbial sequences was expected because whole adults and nymphs actively feeding upon live cucurbits were used to generate RNA pools. Also an effort was made to filter and remove contaminating sequences before assembly of the transcriptome. Despite this only a single contaminating sequence annotating as a putative annexin ortholog of Cucumis melo was identified. This may be partially explained as consistent with the squash bug piercing/sucking mode of feeding which differs from foliar herbivores that retain large masses of partially digested plant material within their midguts posing a greater risk of contamination with plant sequences. Trace contamination with Wolbachia spp., Burkholderia spp., Pseudomonas spp., Proteobacteria, phytoplasmodia, mycoplasmodia and other bacterial or fungal species was observed. The sample preparation method would have allowed contamination with midgut floral sequences. Also identified were sequences annotating as Serratia marcescens, the squash bug vectored bacterial agent of Cucurbit yellow vine disease [2,43]. Phloem feeding insects are famously dependent upon obligate symbiotic bacteria to compensate for nutritional deficiencies inherent to this diet [44 46], thus it is highly likely that A. tristis also may possess symbiotes as these and other data suggest [47]. A substantial number of sequences had significant matches to kinetoplastids (Figure 1), indicating that the very recently established squash bug colony may harbor these parasites. A metagenomic study of A. tristis symbionts and other associated flora would confirm and expand on these tantalizing first hits. Finally, the vast bulk of remaining contigs with significant BLAST scores annotated, as expected, within the Arthropoda (Figure 1), Hemiptera, and in particular were orthologous with the only hemipteran insect having a completed/annotated genome, Acyrthosiphon pisum.
An explication of all the putative A. tristis unigenes identified in this survey would exceed the ambit of a single manuscript, therefore only key categories of identified immunity unigenes will be highlighted below.

Immunity Related Transcripts Identified in the A. Tristis Transcriptome
Many developmental, rare or environmental transcripts are not adequately sampled by insect transcriptome projects as the treatments needed to induce transcription or particular life stages were not included. Thus the inducible transcripts of the insect immune system are routinely missing from published transcriptomes. In this report a deliberate effort was made to sample transcripts from a variety of environmental insults, developmental stages and immune activation. As no entomopathogens are currently known for A. tristis, and because the RNA-seq method is extremely sensitive to contaminating entomopathogen nucleic acid sequences, only purified cell wall components of bacteria and fungi were used to elicit an immune response in lieu of live or intact entomopathogens. The immune system of adult squash bugs was activated by septic puncture using either bacterial or fungal elicitors [37]. Control insects were subjected a sterile puncture. Septic puncture of adult squash bugs with bacterial cell wall elicitors resulted in the increased transcription of 3,950 contigs, and decreased levels of 5,242 contigs. Septic puncture with fungal cell wall elicitors resulted in the increased expression of 2,434 transcripts, and decreased transcription of 6,021. Control adults receiving a sterile puncture expressed 1,491 unique transcripts, while 2,088 transcripts were unique to the bacterial elicitation, and 1,941 were unique to the fungal elicitation. An upper bound estimate of the number of genes directly involved in the insect immune response was reported by Brucker et al. [48] who identified 489 putative immune system genes by hidden Markov model search of the completed genome of the jewel wasp, Nasonia vitripennis.

Pattern Recognition and Signal Transduction
The immune response is triggered by the presence of pathogen associated molecular patterns (PAMPs) such as bacterial cell wall or flagellar components in the presence of host damage associated molecular patterns and chemokines [49]. Pattern recognition receptors located on hemocytes and other tissues bind to the PAMPs activating one or more signal transduction pathways which lead to the synthesis of many antimicrobial enzymes and peptides. Several orthologs of PAMPS and pattern recognition receptors were identified within the A. tristis transcriptome. Among these were -1,3-glucan recognition protein 4a [JQ398676], peptidoglycan recognition protein S2 (JQ398681), C-type lectins, galectins, Dscams, hemolectin, scavenger receptor class B [KF578379], and leucinerich repeat proteins (Table 2). Elicitation with bacterial or fungal cell wall components via septic puncture led to the upregulation of several PAMPS. Expression of Dscam was upregulated 15-fold by bacterial elicitation, and 10-fold by fungal elicitation. The scavenger receptor ortholog was upregulated 149-fold by bacterial and 124-fold by fungal elicitation of A. tristis adults ( Table 2).
Three main conserved signal transduction pathways occur in insects: The toll pathway responsive to gram-positive bacteria and fungi; the imd pathway responsive to gram-negative bacteria; and the JAK/STAT pathway which responds to bacterial or viral infection [50]. Several components of these signal transduction pathways were identified within the assembly. The toll pathway components toll, toll-interacting protein, toll-like receptor 13, dorsal, NF--B inhibitor alpha, NF--B inhibitor-interacting Ras-like protein, spaetzle, snake, easter, and relish (Table 2). Among the imd pathway components identified were dFADD, JNK-interacting protein 1, JNK-interacting protein 3, (Table 2).  Several orthologs of yet another major insect signal transduction pathway, the eicosanoid pathway [51], were recognized within the A. tristis transcriptome assembly. Release of arachidonic acid by phospholipase A 2 allows the C20 fatty acid to enter a series of oxidative reactions yielding second messenger prostaglandins. These prostaglandins bind to receptors that in turn activate hemocytes, among other actions [52]. Within the Heteroptera, the classic model insect Rhodnius prolixus has been documented to possess several components of the eicosanoid pathway interacting with trypanosomal infection [53]. Contigs within the A. tristis assembly that annotated as orthologs of several enzymes within this pathway were 15-hydroxyprostaglandin dehydrogenase [NAD + ], peroxidasin-like, phospholipase A2 [JQ398686], Phospholipase A-2-activating protein, prostaglandin E synthase, prostaglandin reductase, prostaglandin E2 receptor ( Table 2). The transcript levels of these enzymes, however, remain unchanged by bacterial or fungal elicitation of adult squash bugs.

Melanization, Coagulation and Antimicrobial Activities
Upon infection and activation via the toll and imd signal transduction pathways insects produce a suite of antimicrobial peptides, enzymes and enzyme inhibitors which act to limit growth of the pathogen [50]. Analysis of immune-induced transcripts demonstrated that this is also the case with A. tristis (Table 3). Several contigs orthologous to antibacterial peptides of other insects induced by infection were identified. These included transcripts orthologous to a Triatoma brasiliensis defensin-like peptide.
[KF578378] for this defensin-like peptide induced in Anasa tristis by bacterial elicitation. A second transcript encoding an antimicrobial peptide orthologous to hemiptericin of Pyrrhocoris apterus was identified which was elevated 16.6-fold by bacterial elicitation. I propose that this peptide be named Additional antimicrobial enzymes and protease inhibitors were noted, including lysozymes and short pacifastin-like inhibitors (Table 3).
Melanization reactions catalysed by phenoloxidase are primary immune responses to microbial incursion and parasitization which shroud invaders with melanin, crosslink proteins and directly generate microbicidal free radicals [54 56]. Contigs annotating as the major insect melanization enzyme prophenoloxidase, as well as putative serpins and serine proteases within the prophenoloxidase regulatory cascade [50] were recognized (Table 3). Hemolymph coagulation is a well-known defence pathway [57] requiring the action of phenoloxidase, as well as other components. Several orthologs of putative coagulation pathways were noted in the assembly ( Table 3).
Generation of antimicrobial reactive oxygen and nitrogen species by hemocytes and other tissues acts to limit microbial growth [58]. A. tristis orthologs of the plasma membrane reactive oxygen generator NADPH oxidase driving the respiratory burst phenomenon [59] were identified ( Table 2). Also orthologs of the antimicrobial free nitrogen radical generator nitric oxide synthase [19] significantly upregulated 3.4-fold by bacterial elicitation were identified within the assembly. A dual oxidase ortholog of the Drosophila midgut which generates reactive oxygen species also was noted. Enzymes responsible for inactivation of reactive oxygen species also were noted including superoxide dismutase, catalase, peroxidase, glutathione peroxidase, and others. Finally, the orthologs of iron storage and transporting proteins transferrin and ferritin were observed. While transcript levels of the ferritin ortholog were significantly upregulated 5.3-fold by bacterial and 3.7-fold by fungal elicitation the transcript levels of transferrin appeared to be unaffected by these treatments (Table 3).

RNAi Pathways
Although gene silencing via injected or per os RNAi has been demonstrated in several Hemipteran species few examples of successful Heteropteran gene silencing are published. Here the A. tristis orthologs of proteins required for RNAi are presented. Within the A. tristis assembly several orthologs of the miRNA processing pathways were identified (Table 4). An ortholog of the nuclear microprocessor complex subunit, DGCR8 was observed, while cytoplasmic subunits argonaute-1, and -2, aubergine, piwi, dicer-1, dicer-2, and a RISC-loading complex subunit were identified. None of the identified orthologs within these pathways appeared to be differentially regulated by microbial elicitation ( Table 4). As there are no known viral entomopathogens of A. tristis, no viral elicitation was attempted in this report. It should be noted however that viruses were present within the assembly and thus the insects used for the RNA pools may have had active viral infections (see below).

Viral Sequences
Next generation sequencing approaches such as RNA-seq seem ideal methodologies to delineate and to sample the plant-pathogenic virome of plant feeding insect vectors such as squash bugs. Aphids and whiteflies are the major vectors of cucurbit viruses. Squash bugs are known vectors of Serratia marcescens, the bacterial pathogen of Cucurbit yellow vine disease [42,60,61]. The newly founded laboratory squash bug colony from which RNA pools used in this study were isolated was comprised of animals recently collected from local squash and zucchini fields. The animals also were occasionally fed with squash and zucchini cuttings brought in from the field or purchased at local organic outlets. Thus it was not entirely unexpected that RNA-seq revealed tags homologous to many RNA and DNA viruses known to infect insects, or to be vectored to plants by insect species.
To obtain a preliminary survey of the A. tristis virome all reads from the combined RNA pools that did not align to the Trinity assembly were BLASTx screened against the NCBI gbvrl database resulting in 915 significant hits logged. Major taxa of insect viruses detected within these hits included ascoviridae, iridoviridae, granuloviridae, entomopoxviridae, baculoviridae, nudiviridae, even ichnoviridae and bracoviridae. Among these were sequences of plant infecting viruses sequences were detected of viroids, bromoviridae, closteroviridae, genimiviridae, luteoviridae, potyviridae, secoviridae, tombusviridae and virgaviridae. The presence of several of the 17 cucurbit associated viruses such as Cucurbit aphid-borne yellows virus, Melon yellow spot virus, and Papaya ringspot virus within the squash bug colony was indicated [62]. Transcripts from DNA viruses also would have been sampled by this approach and thus even RNA/DNA viruses of vertebrates also were observed. These findings set the stage for a more comprehensive survey of plant pathogens harbored by squash bugs in the laboratory and the field. Confirmation that these viruses occur within wild populations of squash bugs would indicate that further prospecting for entomopathogenic viruses effective as biological control agents of A. tristis is warranted. Next generation deep sequencing of small RNAs extracted from infected cells or insects yields short interfering RNAs, which are products of the Dicer-dependent antiviral pathway, could enable the in silico reconstruction of viral transcripts or entire RNA/DNA virus genomes [63 67]. Phloem/xylem feeding insects can also be used to sample the viral diversity of plant populations on r- [68,69]. Viromes of insects [67,70,71], their food plants [72,73], their predators [74], and even their entomopathogens could be sampled simultaneously in toto. Transcriptomes and genomes have recently been completed for several cucurbits [75] which would allow concurrent pathogenomic monitoring of squash bug, virus and cucurbit host transcription during insect feeding, e.g., [76].

Conclusions
To construct a reasonably complete A. tristis immunotranscriptome, adults were subjected to microbial elicitation of the immune response. The resulting de novo assembly contained many A. tristis orthologs of immune system proteins known from other insect species, including those of phylogenetically related hemipterans. Key components of the entomopathogen recognition system, humoral and cellular immune responses, and second messenger regulatory networks were identified. Interestingly, a partial virome of A. tristis was noted within the de novo assembly, along with the presence of known insect transmitted bacterial species which will become the subject of future studies. Novel approaches are needed to target the unique mode of A. tristis feeding on the phloem of its cucurbit hosts. The de novo transcriptome generated in this report consists of the vast majority of transcript categories synthesized by this insect to support life processes, such as olfaction, neuroendocrinology, reproduction, digestion, and the immune response against bacterial and fungal elicitation. Targeted disruption of one or more of these transcripts, or the proteins that they encode, could reduce the severe economic impact of A. tristis on horticultural production.

Conflicts of Interest
The author declares no competing interests.

Disclaimer
The U.S. Department of Agriculture (USDA) prohibits discrimination in all of its programs and activities on the basis of race, color, national origin, age, disability, and where applicable, sex, marital status, familial status, parental status, religion, sexual orientation, genetic information, political beliefs, (Not all prohibited bases apply to all programs.) Persons with disabilities who require alternative means for communication of program information (Braille, large print, audiotape, etc.) should contact -2600 (voice and TDD). To file a complaint of discrimination, write to USDA, Director, Office of Civil Rights, 1400 Independence Avenue, S.W., Washington, D.C. 20250-9410, or call (800) 795-3272 (voice) or (202) 720-6382 (TDD). USDA is an equal opportunity provider and employer.