Vigneronetal.BMCMicrobiology2012,12(Suppl1):S14 http://www.biomedcentral.com/1471-2180/12/S1/S14 RESEARCH Open Access Host gene response to endosymbiont and pathogen in the cereal weevil Sitophilus oryzae Aurélien Vigneron1, Delphine Charif2, Carole Vincent-Monégat1, Agnès Vallier1, Frédérick Gavory3, Patrick Wincker3, Abdelaziz Heddi1* Abstract Background: Insects thriving on nutritionally poor habitats have integrated mutualistic intracellular symbiotic bacteria (endosymbionts) in a bacteria-bearing tissue (the bacteriome) that isolates the endosymbionts and protects them against a host systemic immune response. Whilst the metabolic and physiological features of long-term insect associations have been investigated in detail over the past decades, cellular and immune regulations that determine the host response to endosymbionts and pathogens have attracted interest more recently. Results: To investigate bacteriome cellular specificities and weevil immune responses to bacteria, we have constructed and sequenced 7 cDNA libraries from Sitophilus oryzae whole larvae and bacteriomes. Bioinformatic analysis of 26,886 ESTs led to the generation of 8,941 weevil unigenes. Based on in silico analysis and on the examination of genes involved in the cellular pathways of potential interest to intracellular symbiosis (i.e. cell growth and apoptosis, autophagy, immunity), we have selected and analyzed 29 genes using qRT-PCR, taking into consideration bacteriome specificity and symbiosis impact on the host response to pathogens. We show that the bacteriome tissue accumulates transcripts from genes involved in cellular development and survival, such as the apoptotic inhibitors iap2 and iap3, and endosomal fusion and trafficking, such as Rab7, Hrs, and SNARE. As regards our investigation into immunity, we first strengthen the bacteriome immunomodulation previously reported in S. zeamais. We show that the sarcotoxin, the c-type lysozyme, and the wpgrp2 genes are downregulated in the S. oryzae bacteriome, when compared to aposymbiotic insects and insects challenged with E. coli. Secondly, transcript level comparison between symbiotic and aposymbiotic larvae provides evidence that the immune systemic response to pathogens is decreased in symbiotic insects, as shown by the relatively high expression of wpgrp2, wpgrp3, coleoptericin-B, diptericin, and sarcotoxin genes in aposymbiotic insects. Conclusions: Library sequencing significantly increased the number of unigenes, allowing for improved functional and genetic investigations in the cereal weevil S. oryzae. Transcriptomic analyses support selective and local immune gene expression in the bacteriome tissue and uncover cellular pathways that are of potential interest to bacteriocyte survival and homeostasis. Bacterial challenge experiments have revealed that the systemic immune response would be less induced in a symbiotic insect, thus highlighting new perspectives on host immunity in long-term invertebrate co-evolutionary associations. *Correspondence:[email protected] 1INSA-Lyon,UMR203BF2I,INRA,BiologieFonctionnelleInsecteset Interactions,Bat.Louis-Pasteur20ave.AlbertEinstein,F-69621Villeurbanne, France Fulllistofauthorinformationisavailableattheendofthearticle ©2012Vigneronetal;licenseeBioMedCentralLtd.ThisisanopenaccessarticledistributedunderthetermsoftheCreative CommonsAttributionLicense(http://creativecommons.org/licenses/by/2.0),whichpermitsunrestricteduse,distribution,and reproductioninanymedium,providedtheoriginalworkisproperlycited. Vigneronetal.BMCMicrobiology2012,12(Suppl1):S14 Page2of15 http://www.biomedcentral.com/1471-2180/12/S1/S14 Background In Drosophila melanogaster, microbe recognition leads Bacterial intracellular symbiosis (endosymbiosis) is wide- to signal production via four pathways (Toll, Immune spread in invertebrates and exhibits a large variety of Deficiency (IMD), JNK, and JAK/STAT) [19-21]. Each phenotypes, ranging from mutualism to pathogenesis. pathway responds to particular types of pathogens, i.e. Endosymbionts are transmitted vertically for hundreds Gram-positive bacteria and fungi for Toll and Gram- of host generations and affect the host biology in many negative bacteria for IMD. Signalling through the Toll ways, including reproduction, physiology and behavior receptor activates a set of phosphorylating reactions [1-4]. The outcome of the association depends on the involving complex adaptors. An inhibitor protein, called interactional networks between the host and bacterial Cactus, is degraded, thus releasing its associated nuclear partners, which sometimes interfere concomitantly with factor protein, called Dorsal-related Immunity Factor many cellular features such as metabolism, apoptosis (DIF), which translocates into the nucleus and induces and immunity [5-7]. antimicrobial peptide genes. The Imd protein is Insectslivingonunbalancednutritionaldietshouseso- upstream of two separate pathways. The first pathway called obligate endosymbionts, which interfere in the involves a protein from the mitogen-activated protein early stages of host embryogenesis with the differentia- (MAP) 3 kinase family, the dTAK1 (Drosophila trans- tion ofspecializedhostcells(thebacteriocytes)thatiso- forming growth factor b activated kinase 1) associated late the endosymbionts and protect them from the host with dTAB2 (Drosophila TAK1 Binding) [22] and immune systemic response [6,8]. In addition to the pri- requiring the potential E3 ubiquitin ligase dIAP2 (Droso- mary endosymbiont, which is fixed in all host popula- phila Inhibitor of APoptosis2) [23-25]. The latter tionsandisessentialforhostfitnessandsurvival,insects appears to be a good candidate for activating the IKK mayintegrate, duringtheirevolutionaryhistory,second- (inhibitor kB kinase) signalosome proteins, which in ary endosymbionts that are facultative and have an turn phosphorylate the Relish (Rel family) transcrip- impacton other biologicaland ecologicalfeaturesofthe tional factor. The second pathway controls the cleavage host[9,10].Evidenceofsymbionteliminationanddispla- of Relish. The “Drosophila Fas-associated death-domain- cementhasalsobeenreportedinweevils[11,12]andsus- containing protein” (dFADD), which is homologous to pected in other insect groups where multiple bacterial the mammalian adaptor protein that interacts with the speciesarecoexistingwithinasinglehostlineage[13,14]. complex “tumor necrosis factor receptor 1” (TNF-R1) to Once established within the host, endosymbionts can recruit pro-caspase-8, links IMD to the caspase “death- experience severe genome size reduction due to relaxed related ced-3/Nedd2-like” (DREDD) in order to build evolutionary pressures on the genes that are unnecessary the “adaptor” complex that allows the activation of cas- or redundant with respect to the host functions [15-17]. pases and apoptosis [26,27]. This pathway may end with As reported in Sodalis, the secondary endosymbiont of a proteasome-independent proteolytic cleavage of Relish, the tsetse fly, gene mutation and deletion processes can probably by the DREDD protein [28,29]. The Relish also affect cell membrane components and genes encod- cleavage dissociates the Rel and the Ankyrins and allows ing Microbe-Associated Molecular Patterns (MAMPs) for processing of the nuclear transcriptional factor. [18]. As these elements are essential for bacterial per- To investigate immune and cellular processes in the ception by the host immune system, the complexity of bacteriome tissue, we have used cereal weevils as a sym- molecular cross-talk between partners may evolve biotic system [6,30]. These crop pests include three spe- according to the level of bacterial genomic degeneration cies (i.e. Sitophilus oryzae, Sitophilus zeamais and and, hence, according to the age of the association. Sitophilus granarius) that all have in common an intra- However, while physiological and evolutionary aspects cellular symbiosis with a Gram-negative g-Proteobacter- of insect endosymbiosis have been thoroughly investi- ium, called Sitophilus primary endosymbiont (or SPE) gated over the past decades, very little is known about [31,32]. Sitophilus insects provide an interesting system the molecular mechanisms that permit the establish- for studying host immune responses to symbionts as ment of symbiosis and then the maintenance and the their association with SPE was established relatively regulation of symbiotic intracellular bacteria. Important recently (less than 25 MY ago), probably by endosym- questions concern, first, how endosymbionts are recog- biont replacement [11,12,17]. The endosymbiont gen- nized and tolerated by the host immune system, sec- ome has not experienced severe gene deletion [17,33]. It ondly how cellular pathways are regulated to prevent encodes functional secretion systems [34] and genes bacteriocyte cell disorders and death due to chronic encoding cell wall elements (unpublished data). Using infection with endosymbionts and, thirdly, how does suppressive subtractive hybridization (SSH), we have endosymbiosis influence host immunocompetence direc- already identified several immune-relevant genes of S. ted at pathogens? zeamais species and we have demonstrated that weevil Vigneronetal.BMCMicrobiology2012,12(Suppl1):S14 Page3of15 http://www.biomedcentral.com/1471-2180/12/S1/S14 bacteriomes exhibit a specific local immune expression weevils house both the integrated endosymbiont SPE that allows symbiont persistence within the bacteriocyte and the facultative endosymbiont Wolbachia [3]. To cells [6]. avoid any side effects from Wolbachia, the “Bouriz” S. Here,wehavestudiedthesiblingS.oryzaespecies.We oryzae strain was chosen because it harbors SPE only. have enlarged the panel of genes potentially involved in SPE-free aposymbiotic insects were obtained as host-symbiontinteractionthrough the constructionand described previously [37]. the sequencingof7differentlibrariesfromwholelarvae Bacteriomes were dissected from fourth instar larvae andfrombacteriomes(i.e.SSH,non-normalizedandnor- in Buffer A (25nM KCl, 10nM MgCl2, 250nM Sucrose, malized libraries).Bioinformatic analysisof26,886ESTs 35nM Tris/HCl, pH=7.5), and stored at -80°C prior to has generated 8,941 unigenes. The results of qRT-PCR RNA preparation. experimentsstronglysupportthegeneexpressionprofile To identify genes involved in the immune response, previously reported for the S. zeamais bacteriome [6], we challenged fourth instar larvae with the intracellular uncover new genes involved in the immune system, bacteria Salmonella typhimurium (Salmonella, Strain apoptosis, vesicular trafficking and cell-growth in the 12023G). About 105 bacteria were injected into the wee- bacteriome tissue, and broaden the proposal that endo- vil hemolymph, using a Nanoject II apparatus (Drum- symbiosis may influence the host immune response in mond, Broomall, PA). The larvae were incubated for 3, long-termhost-symbiontcoevolution. 6 or 12 hours at 27.5°C and 70% rh and then stored at -80°C until required for RNA preparation. Methods This work has been conducted in parallel with two Library constructions other invertebrate models (i.e. Armadillidium vulgare/ Details of material and conditions used for library con- Wolbachia and Asobara tabida/Wolbachia) with the structions are summarized in Table 1. object of identifying conserved and divergent immune Total RNA was extracted with TRIzol Reagent (Invi- pathways and to determine whether invertebrates have trogen, Cergy-pontoise, France), following the manu- selected common strategies to control their symbionts facturer’s instructions. RNA was purified using the and to discriminate between symbionts and pathogens RNeasy mini kit (QIAGEN, Alameda, CA) following [35,36]. the “RNA Clean Up” protocol. After purification, the RNA concentration of each sample was measured with ® Insect manipulation and sample preparation a Nanodrop spectrophotometer (Thermo Scientific, Insects used in this study were reared on wheat grains Wilmington, DE) and total RNA quality was checked at 27.5°C and at 70% relative humidity (rh). Sitophilus by electrophoresis. Table 1Libraries description andconstruction method. Library Type Origin Status Presence Description Numberofindividuals/bacteriomessampled of of andpooled(quantityofRNAusedfromsamples) infection symbiont Host SSH1 Subtraction Whole infected no Salmonella+vs.Salmonella- Salmonella-:10uninfectedaposymbioticlarvae response larvae (10µg) to pathogen SSH2 Subtraction Whole Not no Salmonella-vs.Salmonella+ Salmonella+:15infectedaposymbioticlarvae: larvae infected 5collected3hafterinfection(3.33µg),5after 6h(3.33µg)and5after12h(3.33µg) Host SSHA Subtraction Bacteriome Not yes Withsymbiontvs.without Withsymbiont:200symbioticbacteriomes(10 response infected symbiont µg) to symbiont SSHB Subtraction Bacteriome Not no Withoutsymbiontvs.with Withoutsymbiont:640aposymbiotic infected symbiont bacteriomes(10µg) SO Non- Bacteriome Not yes Poolofbacteriomeswith 170symbioticbacteriomes(10µg) normalized infected symbiont AO Non- Bacteriome Not no Poolofbacteriomeswithout 578aposymbioticbacteriomes(10µg) normalized infected symbiont NOR Normalized Whole infected yes PoolofSymbioticLarvae/ 10uninfectedaposymbioticlarvae(2µg)/10 larvae Aposymbioticlarvae/ uninfectedsymbioticlarvae(2µg)/15infected Aposymbioticlarvaeinfected aposymbioticlarvae:5collectedafter3hof during3h,6hand12h infection(2µg),5after6h(2µg)and5after12h (2µg) Vigneronetal.BMCMicrobiology2012,12(Suppl1):S14 Page4of15 http://www.biomedcentral.com/1471-2180/12/S1/S14 Libraries prepared from bacteriome tissue using treatment with duplex specific nuclease (DSN), as SO (symbiont-full bacteriome) and AO (symbiont-free described by [41]. Normalized cDNA was purified using bacteriome) Libraries (see Table 1) were prepared using a QIAquick PCR Purification Kit (QIAGEN, Alameda, the Creator SMART cDNA Library Construction kit CA), digested with restriction enzyme Sfi1, purified (BD (Clontech/BD Biosciences, PaloAlto, CA), following the Chroma Spin - 1000 column), and ligated into a pAL manufacturer’s instructions. cDNA was digested with 17.3 vector (Evrogen, Moscow, Russia) for E. coli Sfi1, purified (BD Chroma Spin – 400 column) and then transformation. ligated into a pDNRlib vector for E. coli transformation. SSH EST sequencing and data processing SSHA (symbiont-full/symbiont-free bacteriome), SSHB All clones from the libraries were sequenced using the (symbiont-free/symbiont-full bacteriome), SSH1 (Chal- Sanger method (Genoscope, Evry, France) and were lenged/Non-Challenged with S. typhimurium) and SSH2 deposited in the GenBank database. A general overview (Non-Challenged/Challenged with S. typhimurium) were of the EST sequence data processing is given in Figure performed by Evrogen (Moscow, Russia). In order to 1. Raw sequences and trace files were processed with reduce the number of false-positive clones in the SSH- Phred software [42,43] in order to remove any low qual- generated libraries, the SSH technology was combined ity sequences (score < 20). Sequence trimming, which with a mirror orientation selection procedure [38]. Puri- includes polyA tails/vector/adapter removal, was per- fied cDNA were cloned into the pAL16 vector (Evrogen, formed by cross_match. Chimerical sequences were Moscow, Russia) and used for E. coli transformation. computationally digested into independent ESTs. Normalized library Clustering and assembly of the ESTs were performed NOR was prepared by Evrogen (Moscow, Russia). Total with TGICL [44] to obtain unique transcripts (unigenes) RNA was used for ds cDNA synthesis using the composed of contiguous ESTs (contigs) and unique SMART approach [39]. SMART prepared amplified ESTs (singletons). For this purpose, a pairwise compari- cDNA was then normalized according to [40]. Normali- son was first performed using a modified version of zation included cDNA denaturation and reassociation, megablast (minimum similarity 94%). Clustering was Figure1Sequencetreatment(A)andfunctionalannotationprocedure(B). Vigneronetal.BMCMicrobiology2012,12(Suppl1):S14 Page5of15 http://www.biomedcentral.com/1471-2180/12/S1/S14 performed with tclust, that works via a transitive maintained at 27.5°C and 70% rh for 6 hours. 5 samples approach (minimum overlap: 60bp to 20bp maximum of 25 pooled bacteriomes were dissected and then fro- from the end of the sequence). Assembling was carried zen at -80°C until RNA extraction. out with CAP3 (minimum similarity 94%). Total RNA extraction and cDNA synthesis To detect unigene similarities with other species, sev- Total RNA from whole larvae was extracted with the eral blasts (with high cut-off e-values) were performed TRIzol Reagent (Invitrogen, Cergy-pontoise, France), fol- against the following databases: NCBI nr (blastx (release: lowing the manufacturer’s instructions. RNA was incu- 1 March 2011); e-value < 5, HSP length > 33aa), Refseq bated with 1 U/g of RQ1 RNase-Free DNase (Promega, genomic database (blastn, e-value < 10), Unigene divi- Charbonnières-les-Bains, France) for 30 min, at 37°C. sion Arthropods (tblastx, #8 Aedes aegypti, #37 Ano- Total RNA from bacteriomes was extracted with RNA- ® pheles gambiae, #3 Apis mellifera, #3 Bombyx mori, #53 queous -Micro (Ambion, Applied Biosystems, Austin, Drosophila melanogaster, #9 Tribolium castaneum; e- TX), which allows for a better RNA yield from small tis- value < 5). Gene Ontology annotation was carried out sue samples. After purification, the RNA concentration ® using blast2go software [45]. In the first step (mapping), was measured with a Nanodrop spectrophotometer a pool of candidate GO terms was obtained for each (Thermo Scientific, Wilmington, DE) and the RNA qual- unigene by retrieving GO terms associated with the hits ity was checked on an agarose gel electrophoresis. obtained after a blastx search against NCBI nr. In the Reverse-transcription into the first cDNA strand was second step (annotation), reliable GO terms were carried out using the First strand Synthesis System for selected from the pool of candidate GO terms by apply- the RT-PCR kit (Invitrogen, Cergy-pontoise, France). ing the Score Function (FS) of Blast2go with ‘permissive Real-time RTPCR transcript quantification annotation’ parameters (EC-weight=1, e-value-filter=0.1, Quantitative measurements were performed on RNA GO-weight=5, HSP/hit coverage cut-off = 0%). In the samples originating from 5 independent replicates. ® third step of the annotation procedure, the pool of GO Quantification was performed with a LightCycler 480 terms selected during the annotation step was merged system using the LightCycler Fast Start DNA Master with GO terms associated with the Interpro domain SYBR green I kit (Roche Diagnostics, Meylan, France). (InterproScan predictions based on the longest ORF). Data were normalized using the ratio of the target Finally, the Annex augmentation step was run to modu- cDNA concentration to that of the glyceraldehyde 3- late the annotation by adding GO terms derived from phosphate dehydrogenase (gapdh) gene and the riboso- implicit relationships between GO terms [46]. mal protein L29 (RPL29) gene. Primers were designed to amplify fragments with less than 250 bp and are Statistical analyses on libraries listed in the additional file 1. We have used the randomization procedure (with 500 The PCR reactions were carried out in LightCycler 96- random datasets) and the R statistic, described in [47], well plates, in a final volume of 10 μl, containing 2.5 μl to detect unigenes whose transcript abundance (number of cDNA samples (diluted five-fold) and 7.5 μl of Light ® of ESTs) in symbiont-free and symbiont-full bacteriome Cycler 480 SYBR Green Master 1 mix, together with libraries was statistically different (at a FDR of 5.5%). In 0.5 μl of 10 mM of each primer, 1.5 μl H2O and 5 μl of order to perform a functional enrichment analysis of the Mastermix. Quantification was realized as described by unigenes extracted from the SSH, we used the Fatigo [49]. Normalization and statistical pair-wise comparisons web tool [48] against the SO library. were determined using REST [50]. When comparing more than two modalities at the same time, the non- Transcriptomic study parametric Kruskal-Wallis test was used. RPL29 was Sample preparation shown to be the best housekeeping gene, with Best- Transcriptomic analysis was performed on larval bacter- keeper tool [51], and this has been used in graphical iomes, whole symbiotic and aposymbiotic larvae, non- representations. treated, mock-infected (injected with PBS), and injected with 105 E. coli (TOP10, Invitrogen, Cergy-pontoise, Results France). The E. coli bacterium was used here because it General characteristics of libraries: 8,941 weevil unigenes has been shown to efficiently induce the weevil immune were generated system [6], and this bacterium does not necessitate an To explore bacteriome cellular specificities and weevil L2 safety lab structure for manipulation. Larvae were immune responses to bacteria, we have constructed 7 then maintained at 27.5°C and 70% rh for 6 hours. For cDNA libraries from S. oryzae larvae. These libraries each modality, 5 samples of 5 pooled larvae were pre- comprise the 4 SSH libraries, SSHA, SSHB, SSH1 and pared and then frozen at -80°C. Bacteriomes were dis- SSH2, the 2 non-normalized libraries from symbiont-full sected from non-treated larvae that have been (SO) and symbiont–free (AO) bacteriomes and one Vigneronetal.BMCMicrobiology2012,12(Suppl1):S14 Page6of15 http://www.biomedcentral.com/1471-2180/12/S1/S14 normalized library (NOR) from whole aposymbiotic lar- The distribution of unigenes in the different libraries vae challenged, and not, with S. typhimurium (Fig. 2A). is presented in Figure 2A. More than 60% of the uni- The sequencing of all the libraries has generated genes were provided by the NOR library, showing the 26,886 readable ESTs with sequence mean lengths of importance of normalization for unigene number 520 ± 177 bp. Contigation analysis has generated 8,941 enrichment. Blast analysis has shown that most of the unigenes. The average length of unigenes was 620 ± 260 first hits were from Tribolium castaneum sequences. bp, which suggests that most of the unigenes were This result was as expected and is linked with the rela- obtained from low contigation of ESTs. Indeed, the ana- tively high phylogenetic proximity between Tribolium lysis of unigene compositions in ESTs showed that and Sitophilus. about 88% of unigenes were obtained from between one Only about 25% of the unigenes had no Blast annota- (singleton) to four ESTs and less than 3.5% of unigenes tion that corresponded to the UTR part of the cDNA. were assembled from more than 10 ESTs (Fig. 2B). This Following the Blast2go annotation procedure for High finding highlights a low quantitative sequencing depth Scoring Pair (HSP) coverage of 0%, 3845 unigenes pre- with the Sanger methodology and advocates next-gen- sented at least one GO term (Fig. 2C). After Interpros- eration sequencing (NGS) methods, such as Illumina, to can prediction and the Annex procedure, 3995 unigenes fulfill in silico quantitative analysis of this work. The GC presented at least one GO term association. content of total sequences is about 35%, which is very close to the genomic GC content of Tribolium casta- Analysis of libraries neum (34%), phylogenetically the closest Coleopteran One of the objects of this study was to unravel the species sequenced so far [52]. Sequences covered around genes involved in host-symbiont interactions within the 5.5 Mb against 14 Mb of predicted transcripts in bacteriome. For this purpose, an in silico subtraction Drosophila. was conducted between SO and AO libraries, which Figure 2 General description of libraries. (A) Table of ESTs and Unigene numbers presented for each library. The percentages of mitochondrialandrRNAsequencesarealsoprovided.(B)Distributionofunigenes(UGs)asafunctionofthenumberofESTsinvolvedintheUG sequences.UGswithonlyoneESTaresingletons,UGswithmorethanoneESTarecontigs.(C)Blast2goannotationresults.Numberof sequencespresentingGOtermsassociationisgivenforeachstepofthefunctionalannotation.ThedifferentstepsaredescribedintheMethods section. Vigneronetal.BMCMicrobiology2012,12(Suppl1):S14 Page7of15 http://www.biomedcentral.com/1471-2180/12/S1/S14 evaluates statistical differences in unigenes prevalence in other hand, 4 sequences related to the cathepsin 1-like the presence or absence of the symbiont in the bacter- protein, the chemosensory protein, the ribosomal pro- iome tissue. This analysis identified 11 differentially tein L37 and the myoinositol oxygenase, all showed sig- expressed genes (Table 2). The most differentially nificantly higher expression in the symbiont-free expressed gene showed the first blastx hit with a cellular bacteriome. Finally, it is noteworthy that 4 sequences, Fatty-acid binding protein (FABP), and presented a caly- including 2 more expressed in the symbiont-full bacter- cin domain with the Interproscan tool. It is predicted iome and 2 more expressed in the symbiont-free bacter- that it would be upregulated in the presence of SPE. iome, have neither Blast annotation nor an Interproscan However, this first blastx hit presented a relative low e- definition domain. Such sequences cannot be used in value (i.e. 6e-05) and the predicted protein of the this state and require further characterization. sequence showed a weak similarity with the fatty-acid In addition to in silico subtraction, SSHA and SSHB protein (32% on 132 predicted amino acids). This find- libraries were also constructed with the aim of identify- ing highlights the need for additional work to clarify the ing genes involved in host-symbiont interactions. As annotation of this gene. As this gene was also reported described in the Methods section, we carried out a func- as being the most highly expressed in the bacteriome of tional enrichment analysis of SSHA and SSHB in order S. zeamais [30], it is referred to as the “Most Expressed to highlight major GO terms associated with these Gene in the weevil Bacteriome” (MEGwB). library sequences (see Additional file 2). Concerning the The subtraction has also identified two other SSHA library, three GO terms from biological processes sequences, which are highly expressed in the symbiont- (i.e. “transposition”, “cell division”, “DNA recombina- full bacteriome, when compared to the symbiont-free tion”) and one GO term from molecular functions (i.e. bacteriome. The first was related to methylmalonyl-CoA “transposase activity”) were significantly over-repre- decarboxylase (58% similarity based on predicted pro- sented. Concerning SSHB, five GO terms from biologi- tein) and the second was a transmembrane protein close cal processes (i.e. “digestion”, “nitrogen compound to the Tribolium transmembrane 41B protein. On the metabolic process”, “carbohydrate metabolic process”, Table 2Listofunigenes presenting statistically differentrepresentations in AO andSO libraries. Accession 1R Unigene 2Redundancy 2Redundancy 31stblastxresulton 3Accession 3e-value 3Coverage 3Max 4InterProScan numbers length inAO inSO TriboliumCastaneum identity predicteddomain (pb) FQ866673 16.45 664 24 103 allergenacaS13 XP_969762 6e-05 50% 32% IPR011038; (cellularFABP-like) IPR012674 FQ866935, 15.91 1307 304 440 NA NA NA NA NA noIPR FQ867818 FQ877624 5.21 723 21 0 NA NA NA NA NA NoIPR FQ884311 3.22 351 13 0 RPL37 XP_969650 3e-36 76% 94% IPR001569; IPR011331; IPR011332; IPR018267 FQ868370 2.9 525 17 1 Chemosensory NP_001039278 6e-33 75% 49% IPR005055 protein10 FQ862292 2.9 974 17 1 CathepsinL-like NP_001163996 2e-68 88% 48% noIPR proteinase FQ869260 2.73 138 11 0 NA NA NA NA NA NoIPR FQ865010 2.49 865 12 28 Gamma-subunit. XP_973308 2e-24 54% 58% IPR010625 methylmalonyl-CoA decarboxylase FQ884611, 2.48 1463 10 0 Myoinositol XP_966469 3e-133 6% 74% IPR007828 FQ867701 oxygenase FQ864415 2.17 704 0 6 Transmembrane XP_975236 1.8e-02 25% 42% NoIPR protein41B FQ863216 2.17 812 0 6 NA NA NA NA NA noIPR 1TheRstatistictest,with500randomdatasets,wasperformedtoevaluategeneswhoserepresentationinAOandSOlibrarieswasstatisticallydifferent. SequencesshowinganRstatistic>2weresignificant. 2Unigeneredundancyisgivenforeachlibrary(AOandSO). 3Foreachunigene,wegaveblastxmatcheswithTriboliumcastaneum,theclosestgenome-sequencedinsect,phylogenetically,toSitophilus.Accessionnumbersof Triboliumrelatedsequences,e-valueofblastxhits,sequencescoverageandmaxidentitybetweenSitophilusandTriboliumsequencesarealsogiven. 4Interproscanpredicteddomainsaregiventocompletethecharacterizationofsequences. Vigneronetal.BMCMicrobiology2012,12(Suppl1):S14 Page8of15 http://www.biomedcentral.com/1471-2180/12/S1/S14 “polysaccharide metabolic process”, and “energy deriva- Differential gene expression in the bacteriome tissue tion by oxidation of organic compounds”) and nine GO We have compared the steady state levels of 29 genes in terms from molecular functions (i.e. “hydrolase activity”, the bacteriome and in the whole aposymbiotic larvae “ion binding”, “tetrapyrole binding”, “hydrolase activity, (Fig. 3). We preferred to use whole aposymbiotic larvae, acting on glycosyl bonds”, “monooxygenase activity”, rather than symbiont-free bacteriome tissue, as the con- “peptidase activity”, “heme binding”, “cation binding” trol because SSHB is prone to a lot of potential contam- and “hydrolase activity, hydrolyzing O-glycosyl com- ination from the gut. The total transcriptome of larvae pounds”) were significantly over-expressed. represented an average level of gene transcripts and this The SSHA yielded 55 unigenes with the eukaryotic was then used as the control. blast result. A detailed listing of these unigenes is pre- As described previously in S. zeamais [6], only Toll sented in Additional file 3. The remaining unigenes Interacting Protein (TollIP), as a potential negative regu- were related to prokaryotic assignation, which means lator of the vertebrate Toll pathway [53] and coleopteri- that the subtraction has been contaminated with sym- cin-A, as AMP, are upregulated in the bacteriome of S. biont DNA. Surprisingly, none of the 55 unigenes were oryzae. The sarcotoxin and genes described as having related to the immune response and only one, an aspar- lytic activity, such as wpgrp2 (weevil PeptidoGlycan tic proteinase, presented a high similarity (96%) with a Recognition Protein2), gnbp1 (Gram Negative Binding sequence found in S. zeamais [6]. Most of the SSHA Protein1) and c-type lysozyme, are significantly down- unigenes are referred to as metabolic or cellular regula- regulated in the bacteriome when compared to aposym- tion genes, suggesting high cellular activity in the sym- biotic larvae challenged, or not, with E. coli (Fig. 3 and biont-full bacteriome [30]. The functional enrichment 4). analysis has allocated, to the SSHA, the level 3 GO To gain a better understanding of immune regulation terms “transposition” (GO:0032196) and “transposase in the bacteriome, we have analyzed additional genes activity” (GO:0004803). This is probably due to the mas- identified in this work, which are branched at different sive presence of insertion sequences (IS) recently docu- levels of the signaling pathways, including imd and iap2 mented in the SPE genome [17]. (IMD pathway), and cactus and ecsit (Toll pathway) The 844 EST sequences from SSHB have provided 299 [23,54-56]. Intriguingly, the imd and iap2 genes, which unigenes potentially expressed specifically in the sym- activate AMP synthesis via the IMD pathway in Droso- biont-free bacteriome. Blastx annotations have identified phila, are highly expressed in the Sitophilus bacteriome. around 60% of these sequences as digestive enzymes. Moreover, the ecsit gene, which participates in Toll-sig- Functional analysis of SSHB has allocated the level 3 naling pathway activation in vertebrates [56,57], is also GO terms, such as “digestion” (GO:0007586), “nitrogen highly expressed in the bacteriome whereas the Toll compound metabolic process” (GO:0006807) or “hydro- inhibitor cactus is downregulated (Fig. 3). Taken lase activity” (GO:0016787). As these functions are together, these data suggest that both IMD and Toll dominant in the gut tissue, and as symbiont-free bacter- pathways are potentially initiated in the bacteriome, iomes are very thin, flat and intimately attached to the which appears to be in contrast with the low amounts intestine, contamination from the gut is highly probable of effector gene transcripts (e.g. AMP) in this tissue. while dissecting out the bacteriomes. To extend this investigation to other cellular processes that are of interest to bacteriocyte homeostasis and sur- Transcriptomic study vival, we have analyzed three genes potentially involved The purpose of the transcriptomic study was to analyze in apoptosis activation and regulation, namely the Inhi- molecular and cellular specificities of the bacteriome bitor of APoptosis2 (iap2), the Inhibitor of APoptosis3 and to test the influence of symbiosis on the host (iap3), and the caspase-like gene. Whilst apoptosis inhi- immune response to bacterial pathogens. Analyzed bitor genes (i.e. iap2 and iap3) are highly expressed, the genes were retrieved from different libraries based on in caspase-like encoding gene is weakly expressed in the silico subtraction, experimental subtractions (SO, AO, bacteriome (Fig. 3 and 4). In line with this finding, the SSHA), and on the examination of genes involved in cel- RAt Sarcoma (Ras), calmodulin-1 and leonardo 14-3-3, lular pathways of potential interest to intracellular sym- which are all involved in cell growth and survival biosis, such as apoptosis, cell trafficking and immunity [58-60], are also upregulated in the bacteriome. Taken (NOR, SSH1). together, these data suggest that bacteriocyte cell path- In total, we have selected 29 genes (Additional file 4). ways are regulated to prevent cell death and to promote Except for MEGwB, all sequences presented more than cell survival. 60% similarity with their first hit on the blastx and/or Vesicular trafficking is also an important process in major Interproscan domains of the unigene predicted the bacteriocyte functions, both for metabolic protein. exchange between the host and the endosymbiont [30] Vigneronetal.BMCMicrobiology2012,12(Suppl1):S14 Page9of15 http://www.biomedcentral.com/1471-2180/12/S1/S14 Figure3Analysisofgeneexpressionprofilesinthebacteriome.TranscriptsofgeneswerequantifiedbyqRT-PCR.Bacteriomesdissected fromfourth-instarlarvaewerecomparedtowholeaposymbioticfourth-instarlarvae.Expressionofgeneswasnormalizedwiththeexpressionof theribosomalproteinL29.Eachboxrepresentsthemedian(boltline)andthequartiles(25%/75%)offiveindependentmeasurements. StatisticalanalysiswasperformedwiththeRESTpair-wisefixedreallocationrandomizationtest.Asterisksindicateasignificantdifferencebetween thebacteriomeandthecontrol(p-value<0.05). and for intracellular bacterial control by cellular autop- involved in endosomal maturation) and a gene encod- hagy [61]. Among the selected genes, we have tested ing for a Soluble NSF Attachment protein REceptor three genes involved in vesicular formation and traf- (SNARE, vesicle fusion) [62-64]. We have demon- ficking, these being the Ras related GTP-binding gene strated that all these genes are highly expressed in the (Rab7, late endosomes labelling), the hepatocyte bacteriome, when compared to the aposymbiotic larvae growth factor-regulated tyrosine kinase substrate (Hrs, (Fig. 3). Vigneronetal.BMCMicrobiology2012,12(Suppl1):S14 Page10of15 http://www.biomedcentral.com/1471-2180/12/S1/S14 Figure 4 Quantitative immune gene expression in symbiotic and aposymbiotic larvae of Sitophilus oryzae. (A) Transcript levels of immunegenesquantifiedbyqRT-PCRinwholeaposymbioticandsymbioticlarvae.Forbothsymbioticandaposymbioticlarvae,non-injected larvae,larvaeinjectedwithPBS,andlarvaeinjectedwithE.coliwereanalyzed.Resultsfromgeneexpressioninthebacteriomearereportedhere asanindicator.RepresentedexpressionofgeneswasnormalizedwiththeexpressionoftheribosomalproteinL29.Eachboxrepresentsthe median(boltline)andquartiles(25%/75%)offiveindependentmeasurements.Foreachsymbioticandaposymbioticstatus,thenon- parametricKruskal-Wallistestwasappliedinordertodetermineglobaldifferencebetweenthethreemodalitiestested(p-value<0.05), representedbyanasterisk.(B)Differentialexpressionratiosobtainedfromq-RT-PCRexperiments.Forgenespresentingsignificantdifferencesin expressionaftertheglobaltest(seeA),theprickingstresseffectwastestedbycomparinglarvaeinjected,ornot,withPBS.Theinfectioneffect wastestedbycomparinglarvaeinjectedwithPBSandlarvaeinjectedwithE.coli.TheRESTpair-wisefixedreallocationrandomizationtestwas applied.Foreachmodalitytested(notinjected,injectedwithPBSandinjectedwithE.coli),acomparisonbetweensymbioticlarvaeand aposymbioticlarvaewasappliedinordertoevaluatetheimpactofsymbiosisontheexpressionofimmunegenes.TheRESTpair-wisefixed reallocationrandomizationtestwasperformedbetweentheexpressionofgenesfromsymbioticandaposymbioticlarvae.Underlinedscores indicatesignificantdifferencesbetweenthetwomodalitiestested(p-value<0.05).Anup-arrowindicatesupregulatedgeneswhereasadown- arrowindicatesdownregulatedgenes.