Insights into the venom composition of the ectoparasitoid wasp Nasonia vitripennis from bioinformatic and proteomic studies

Abstract

With the Nasonia vitripennis genome sequences available, we attempted to determine the proteins present in venom by two different approaches. First, we searched for the transcripts of venom proteins by a bioinformatic approach using amino acid sequences of known hymenopteran venom proteins. Second, we performed proteomic analyses of crude N. vitripennis venom removed from the venom reservoir, implementing both an off-line two-dimensional liquid chromatography matrix-assisted laser desorption/ionization time-of-flight (2D-LC-MALDI-TOF) mass spectrometry (MS) and a two-dimensional liquid chromatography electrospray ionization Founer transform ion cyclotron resonance (2D-LC-ESI-FT-ICR) MS setup. This combination of bioinformatic and proteomic studies resulted in an extraordinary richness of identified venom constituents. Moreover, half of the 79 identified proteins were not yet associated with insect venoms: 16 proteins showed similarity only to known proteins from other tissues or secretions, and an additional 23 did not show similarity to any known protein. Serine proteases and their inhibitors were the most represented. Fifteen nonsecretory proteins were also identified by proteomic means and probably represent so-called ‘venom trace elements’. The present study contributes greatly to the understanding of the biological diversity of the venom of parasitoid wasps at the molecular level.

Introduction

The venom glands of parasitoid wasps manufacture various proteins and peptides that influence the arthropod host’s physiology. In some cases, parasitoids do not kill their target instantly but instead keep them alive to serve as a fresh food supply for their offspring. In order to achieve this, they have to target the host’s immunity, physiology, mobility, reproductive capacity and even behaviour.

Our knowledge of venom proteins and peptides involved in these processes is rather limited. The most studied species is the solitary pupal endoparasitoid Pimpla hypochondriaca (Hymenoptera: Ichneumonidae) that parasitizes a number of lepidopteran species including the tomato moth, Lacanobia oleracea (Lepidoptera: Noctuidae). Early attempts to identify biologically active components from P. hypochondriaca venom focused on their enzyme activity combined to specific inhibitor sensitivity, for instance L-3,4-dihydroxy-phenylalanine (L-DOPA) oxidizing activity/phenylthiocarbamide sensitivity (Parkinson & Weaver, 1999) or angiotensin-converting enzyme-like activity/captopril sensitivity (Dani et al., 2003).

With the availability of a cDNA library from the P. hypochondriaca venom-synthesizing gland, it became possible to characterize cDNAs encoding an arthropod-specific phenoloxidase (Parkinson et al., 2001), possibly involved in the previously discovered L-DOPA oxidizing activity. This was the first of an impressive series of venom constituents from which the encoding cDNAs became available, some of them by random sequencing: a 28 kDa serine protease (Parkinson et al., 2002a)), a reprolysintype metalloprotease (Parkinson et al., 2002b)), the large subunit of the paralytic heterodimeric polypeptide pimplin (Parkinson et al., 2002c), venom proteins (vpr) 1–3, enzymes similar to trehalase and laccase (Parkinson et al., 2003) and no fewer than seven cysteine-rich venom proteins (Parkinson et al., 2004). In fact, the attempts to characterize cDNAs from venom glands of endoparasitic wasps go back to the early 1990s, with the identification of two venom protein-encoding cDNAs from Chelonus sp. near curuimaculatus (Jones et al., 1992; Krishnan et al., 1994).

Another very successful search was for venom components from the endoparasitoid wasps Cotesia rubecula and Cotesia congregata (Hymenoptera: Braconidae), parasitizing Pieris rapae and Manduca sexta (Lepidoptera), respectively. These species differ from P. hypochondriaca in the fact that they also inject polydnaviruses into their lepidopteran hosts in order to overcome the host immune defences and to work in synergy with the wasp’s venom (Amaya et al., 2005). Very recently some venom constituents of two Microctonus spp. (Hymenoptera: Braconidae) were identified by combining mass spectrometric analysis of tryptic peptides from sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS-PAGE) separated proteins, cDNA library sequencing and direct cDNA pyrosequencing (Crawford et al., 2008). Microctonus hyperodae and Microctonus aethiopoides differ from most other endoparasitoids in that they lay their eggs in adult weevils rather than in their immature stages.

The ectoparasitoid wasp Nasonia vitripennis (Hymenoptera: Pteromalidae) has likewise been subjected to several studies aimed at understanding the roles of venom during parasitism (Rivers et al., 1993, 1999, 2002a, 2004, 2005). Ectoparasitoid wasps develop outside the host body, although they are frequently attached or embedded in the host’s tissues. There is a good understanding of how N. vitripennis venom as a whole functions: it inhibits host cellular immune responses (Rivers et al., 2002b), depresses respiratory metabolism (Rivers & Denlinger, 1994), stimulates increments in lipid levels within haemolymph and fat body (Rivers & Denlinger, 1995) and ultimately induces death. However, the characterization of the venom components involved in these processes has only just begun (Rivers et al., 2006).

With the N. vitripennis genome sequences available, we tackled the venom composition by two different approaches. First, we searched the transcripts of venom proteins by a bioinformatic approach using amino acid sequences of known hymenopteran venom proteins. Second, we performed proteomic analyses of crude Nasonia venom, implementing both an off-line two-dimensional liquid chromatography matrix-assisted laser desorption/ionization time-of-flight (2D-LC-MALDI-TOF) mass spectrometry (MS) and a two-dimensional liquid chromatography electrospray ionization Fourier transform ion cyclotron resonance (2D-LC-ESI-FT-ICR MS) setup.

Results

Bioinformatic study

To screen the N. vitripennis genome for homologues of known hymenopteran venom proteins, we constructed a query file comprising 383 amino acid sequences in fasta-format by a keyword-based selection. This query file was blastP-searched against the N. vitripennis NCBI RefSeq database using a stand-alone blast software package, with a significance cut-off of E⁻⁷. This resulted in a first raw data output of 192 hits. Our blastP-search was further manually extended using the National Center for Biotechnology Information (NCBI) online version of the software package against some interesting venom proteins from parasitoid wasps that were lacking in our initial query file. This resulted in four additional hits. One of these hits, C1q-like venom protein, was the subject of a specific and extensive study, described elsewhere in this supplement (de Graaf et al., 2009).

Sixty proteins with no signal peptide as predicted by the SignalP 3.0 Server (Bendtsen et al., 2004) were excluded from our list of potential venom proteins. After establishment of the conserved domain architecture of the remaining 135 proteins, all except one could be placed in one of 11 protein clusters for further analysis: acid phosphatases (14 records), antigen 5-like proteins (six records), chitinases (five records), cysteine-rich venom proteins (eight records), γ-glutamyl transpeptidases (two records), laccases (four records), phenoloxidase-like proteins (eight records), lipases (16 records), serine proteases (56 records), metalloproteinases (10 records) and miscellaneous (trehalase, one record; icarapin, one record; angiotensin-converting enzyme, one record; aspartylglucosaminidase, one record; and calreticulin, one record). Based on an in-depth comparison with the query records (conserved domain architecture and sequence identity) we were able to narrow further the hits found and to retain 59 of them (not shown) for confirmative reverse transcriptase PCR (RT-PCR) analyses with Nasonia venom reservoir tissue and subsequent sequencing. This resulted in the sequencing of 21 transcripts of secretory proteins of the reservoir (see Supporting Information Table S1).

Proteomic study

An initial proteomic analysis on crude venom extracts showed relatively high concentrations of organic compounds interfering with tryptic digest, lyophilization and liquid chromatography (LC) separation. After sample clean-up, the strong cation exchange (SCX)-separation profile normalized (Fig. 1) and fractions 5–13 were further analysed on reversed phase (RP)-LC-MALDI-TOF MS as mentioned in the Experimental procedures. A combined database search of all tandem mass spectrometry (MS/MS) spectra from these nine fractions resulted in the identification of 23 peptides with ion scores above the extensive homology threshold, accounting for 14 proteins. Six additional peptides could be assigned to these proteins as individual mascot scores were 20 or higher; a subsequent search against the entire NCBI database resulted in no higher scores. No identifications were rejected after manual inspection of the MS/MS spectra.

Figure 1.

Two-dimensional chromatography performed on an Ultimate Plus Dual-Gradient Capillary/Nano LC system (Dionex-LC Packings, Hercules, CA, USA), preceding the MALDI-TOF mass spectrometry analyses. In the first dimension (A), peptides were separated on a strong cation exchange (SCX) column and the effluent was fractionated at 5 min intervals using a Probot spotting device (Dionex-LC Packings). (B) A 10 µl SCX-fraction was loaded onto a C18 microguard column and subsequently separated on a PepMap C18 analytical column. The reversed phase liquid chromatography effluent was spotted every 30 s on a MALDI-plate using the Probot spotter. The chromatogram shown is of the SCX fraction 10. mAU, milliabsorbance units.

LC-ESI-FT-MS analysis on these nine SCX fractions resulted in the identification of 91 unique peptides, accounting for 51 proteins, including the 14 proteins identified with LC-MALDI-TOF. In addition, we also included the SCX flow-through fraction in the LC-ESI-FT MS analysis, which finally resulted in 990 MS/MS spectra identified as N. vitripennis peptides, of which only six were decoy hits. After removing redundant peptide identifications, 258 unique peptide sequences were retained, accounting for 76 unambiguously identified proteins with a less than 1% false positive rate. Twenty-three of the 29 previously identified peptides using LC-MALDI-TOF MS were also identified with LC-ESI-FT MS representing all 14 proteins. In silico prediction of the presence of a signal peptide enabled us to subdivide the found proteins into 61 secretory (true venom) proteins (Supporting Information Table S2) and 15 nonsecretory proteins (Supporting Information Table S3).

Data overlapping and nomenclature

Overall this study yielded 79 unique venom proteins that are listed in Table 1. From our initial list of 196 blast-searched hits, nine proteins were subsequently confirmed by the proteomic study; but only three of them belong to the group of 21 blast hits with additional RT-PCR data (thus, six blast-searched hits that were not retained for RT-PCR analysis were eventually confirmed by 2D-LC-ESI-FT-ICR MS; see Fig. 2). Only one protein, γ-glutamyl transpeptidase-like venom protein 1, was found by all of the techniques (blast-search, RT-PCR, 2D- LC-MALDI-TOF MS and LC-ESI-FT MS). The venom proteins were named in accordance with their homologues, the protein to which they showed the most resemblance or to a characteristic conserved domain. Remaining proteins were named alphabetically, eg ‘venom protein D’.

Table 1.

Venom proteins discovered by the bioinformatic and/or proteomic approaches

Protein name	Accession number	Method(s)
Proteases and peptidases
Metalloprotease	XP_001604362	B1, B2
Angiotensin-converting enzyme	XP_001607198	B1, B2
Dipeptidylpeptidase IV	XP_001599462	P2
Serine protease	lcl|hmm408134	P2
Serine protease	lcl|hmm536144	P2
Serine protease	XP_001604852	P2
Serine protease	lcl|hmm315294	P2
Serine protease	XP_001600774	P2
Serine protease	XP_001600807	B1, P2
Serine protease	XP_001600838	B1, P2
Serine protease (SP50)	XP_001602769	B1, B2
Serine protease (SP97)	XP_001604873	P1, P2
Serine protease (SP101)	XP_001605133	B1, B2
Serine protease homologue (SPH21)	XP_001600151	B1, B2, P2
Serine protease/CLIP	XP_001600149	P2
Serine protease/CLIP	XP_001600178	B1, P2
Serine protease/CLIP	XP_001599920	B1, P2
Serine protease/CUB (SPH42)	XP_001602368	B1, B2
Protease inhibitors
Cysteine-rich/KU venom protein	XP_001604564	B1, B2
Cysteine-rich/Pacifastin venom protein 1	XP_001606763	B1, B2
Cysteine-rich/Pacifastin venom protein 2	XP_001606768	B1, B2
Cysteine-rich/TIL venom protein 1	XP_001607367	P1, P2
Cysteine-rich/TIL venom protein 2	XP_001607359	B1, B2
Kazal type serine protease inhibitor-like venom protein 1	XP_001604686	P2
Kazal type serine protease inhibitor-like venom protein 2	lcl|hmm867024	P2
Small serine proteinase inhibitor-like venom protein	XP_001607610	P2
Carbohydrate metabolism
Chitinase 5	XP_001602780	B1, B2
Glucose dehydrogenase-like venom protein	XP_001600948	P2
Trehalase	XP_001602179	B1, B2
DNA metabolism
Apyrase	XP_001603046	P2
Endonuclease-like venom protein	XP_001599850	P2
Inosine uridine-preferring nucleoside hydrolase	XP_001599345	P2
Glutathione metabolism
γ-Glutamyl transpeptidase-like venom protein 1	XP_001607488	B1, B2, P1, P2
γ-Glutamyl transpeptidase-like venom protein 2	XP_001604839	B1, B2, P2
γ-Glutamyl cyclotransferase-like venom protein	XP_001605740	P2
Esterases
Acid phosphatase	XP_001600562	B1, B2
Acid phosphatase	XP_001605498	B1, P2
Multiple inositol polyphosphate phosphatase-like venom protein	XP_001605344	P2
Arylsulphatase b	XP_001603886	P2
α-Esterase	XP_001602279	P2
Lipase	lcl|hmm642184	P1, P2
Lipase-like venom protein	lcl|hmm589104	P2
Recognition/binding proteins
β-1,3-Glucan recognition protein	XP_001607754	P2
Chitin binding protein-like venom protein	lcl|hmm464144	P2
General odorant-binding protein-like venom protein	XP_001604989	P1, P2
Low-density lipoprotein receptor-like venom protein	lcl|hmm94654	P2
Immune related proteins
Calreticulin	XP_001600192	B1, B2
C1q-like venom protein	XP_001608267	B1, B2
Immunoglobulin-like venom protein	XP_001608198	P1, P2
Others
Aminotransferase-like venom protein 1	XP_001607226	P1, P2
Aminotransferase-like venom protein 2	XP_001602041	P1, P2
Antigen 5-like protein	XP_001603611	B1, B2
Antigen 5-like protein	XP_001603715	B1, B2
Aspartylglucosaminidase	XP_001603206	B1, B2
Laccase	XP_001605351	B1, B2
Laccase	XP_001600917	B1, P2
Unknown
Venom protein D	lcl|hmm895104	P2
Venom protein E	XP_001603438	P2
Venom protein F	XP_001602939	P2
Venom protein G	lcl|hmm716344	P2
Venom protein H	lcl|hmm214614	P2
Venom protein I	lcl|hmm282804	P2
Venom protein J	lcl|hmm34174	P2
Venom protein K	lcl|hmm320864	P2
Venom protein L	lcl|hmm726814	P2
Venom protein M	lcl|hmm72024	P2
Venom protein N	lcl|hmm203054	P2
Venom protein O	lcl|hmm169294	P2
Venom protein P	XP_001606165	P2
Venom protein Q	XP_001607549	P2
Venom protein R	XP_001603774	P2
Venom protein S	XP_001606832	P2
Venom protein T	XP_001601803	P2
Venom protein U	lcl|hmm215044	P1, P2
Venom protein V	lcl|hmm734344	P1, P2
Venom protein W	XP_001602724	P1, P2
Venom protein X	XP_001605793	P1, P2
Venom protein Y	XP_001603600	P1, P2
Venom protein Z	XP_001607601	P1, P2

B, bioinformatic approach; B1, blast-search; B2, subsequent reverse transcriptase PCR; P, proteomic approach; P1, 2D-LC-MALDI TOF MS; P2, 2D-LC-ESI-FT-ICR MS.

Figure 2.

Overview of data from the bioinformatic and the proteomic studies. Both approaches yielded proteins with no signal peptide (nonsecretory proteins), which were eliminated from our list of potential venom proteins. The blast-searching hits (blue circle) needed further confirmation by reverse transcriptase PCR (green circle) or one of the proteomic studies. All 14 proteins identified by the 2D-LC-MALDI-TOF mass spectrometry (MS) set-up (red circle), were also discovered by 2D-LC-ESI-FT-ICR MS (yellow circle). The numbers in red represent the Nasonia venom proteins.

Types of venom proteins

We identified venom proteins from Nasonia that appear to fall into different broad functional categories, including (1) proteases and peptidases; (2) protease inhibitors; (3) carbohydrate metabolism; (4) DNA metabolism; (5) glutathione metabolism; (6) esterases, (7) recognition/binding proteins; (8) immune related proteins; and (9) others. These are described in more detail below.

Proteases and peptidases. Proteases and peptidases are hydrolases that specifically cleave the peptide bonds found in proteins and peptides. They are essential for the survival of all kinds of organisms, and are encoded for by approximately 2% of all genes (Barrett et al., 2001). They are very common components of venoms, and the ones that we found fall into the categories metalloproteinase, peptidase and serine protease.

Metalloproteinase. Snake venoms contain a variety of metalloproteinases that are highly toxic, resulting in severe bleeding by interfering with blood coagulation and haemostatic plug formation or by degrading the basement membrane or extracellular matrix components of the victims (Matsui et al., 2000). They are all members of the metzincin (‘met’-‘zinc’ in) family, characterized by a Zn²⁺-binding motif of HExxHxxGxxH and a distal located methionine (Rawlings & Barrett, 1995). Detailed examination of the amino acid sequence of the Nasonia venom gland metalloprotease that we found reveals a zinc binding domain of the reprolysin type (HELGHNLGxxHD) (Jia et al., 1996; Hati et al., 1999), except for two substitutions (N→L and G→N). The same HELGHLLNxxHD catalytic domain was also found in metalloproteases from two other parasitoid (Eulophidae) hymenopteran species, Melittobia digitata (AAU89117) (Cônsoli et al., 2004) and Eulophus pennicornis (ACF60599) (Price et al., 2009), and bears great resemblance to that of tick (Ixodes scapularis, AAP22067) salivary gland metalloprotease (HELAHNLGCxHD) (Francischetti et al., 2003) and P. hypochondriaca venom metalloprotease (HELGHVFSAPRD) (Parkinson et al., 2002b).

Peptidase. Two peptidases were found in Nasonia venom: angiotensin-converting enzyme and dipeptidylpeptidase IV. Peptidase family M2 angiotensin-converting enzyme (ACE, EC 3.4.15.1) is a membrane-bound, zinc dependent dipeptidase that catalyses the conversion of the decapeptide angiotensin I to the potent vasopressor octapeptide angiotensin II by removing two C-terminal amino acids (Ondetti & Cushman, 1982). ACE-like enzyme activity was detected in P. hypochondriaca venom by reverse-phase high performance liquid chromatography (HPLC) using the synthetic tripeptide Hip–His–Leu as a substrate (Dani et al., 2003). This activity was sensitive to captopril, an ACE inhibitor, and antiserum raised against recombinant Drosophila melanogaster ACE-like enzyme recognized a 74 kDa band in a Western blot of wasp venom. The authors suggested that ACE could be involved in the processing of peptide precursors in the venom sac.

Dipeptidylpeptidase IV is a highly glycosylated serine protease that cleaves N-terminal dipeptides from substrates with proline, alanine, or hydroxyproline in the penultimate position. It occurs in both soluble and membrane-bound forms, the human form being known as CD26 (Ulmer et al., 1990). In honey bees the membrane-bound enzyme is involved in the processing of melittin, the main constituent of the venom (Kreil et al., 1980).

Serine protease. Serine proteases are endopeptidases in which one of the amino acids of the active site is a serine. The proteins that we found all belong to the S1 peptidase family, the trypsins, which underwent the most predominant genetic expansion yielding the enzymes responsible for vital processes such as digestion, blood coagulation, fibrinolysis, development, fertilization, apoptosis and immunity (Page & Di Cera, 2008). Insect serine proteases are involved in the activating cascade of Toll signalling, in the context of dorsoventral axis formation as well as innate immune response, and of pPO signalling (Jang et al., 2008). Serine protease homologues, which lack one of three catalytic domains yet have similar amino acid sequences to active serine proteases, were shown to participate in several physiological processes of arthropods, including insects. In bee and wasp venoms, both active serine proteases (Parkinson et al., 2002a; Yamamoto et al., 2007) and serine protease homologues (Asgari et al., 2003a) have been discovered and may potentially be implicated in the hosts’ embryonic development (Cônsoli et al., 2004) and immunity (Asgari et al., 2003a; Zhang et al., 2004). With 15 members found, the serine protease protein family is the best represented of all Nasonia venom constituents.

Protease inhibitors. Despite their life-giving functions, enzymes that break down proteins are potentially very damaging in living systems, so their activities need to be kept strictly under control. Several distinct mechanisms exist for the control of excessive peptidase activity, important amongst which are the interactions of the enzymes with proteins that inhibit them (Rawlings et al., 2004). Several protease inhibitors were found in Nasonia venom.

Cysteine-rich venom protein. The cysteine-rich venom proteins (CVPs) that we found have cysteine arrangements resembling those of toxins or serine protease inhibitors that fall in established protein families, each with their typical motif. The Kunitz type motif usually has a peptide chain of around 60 amino acid residues and is stabilized by three disulphide bridges with the bonding pattern of 1–6, 2–4, 3–5. This motif was first seen in the bovine pancreatic trypsin inhibitor-like proteinase inhibitors, but its occurrence in toxins from various venomous animals, including snakes, lizards, spiders, cone snails and sea anemones is also well documented (Morrissette et al., 1995; Yuan et al., 2008; Matsunaga et al., 2009). The main ancestral function of Kunitz type proteins was the inhibition of a diverse array of serine proteases, but some of them have the ability to block ion channels, especially the voltage-gated potassium channels, which are essential for regulation of various physiological processes such as blood coagulation, fibrinolysis, host defence and action potential transduction. In P. hypochondriaca, CVP2 was suggested to be involved in the inhibition of previously found serine proteases (Parkinson et al., 2004). As several trypsin-like proteases were likewise found in the present study (see above), the biological function of Nasonia cysteine-rich/Kunitz (KU) venom protein might be similar.

The Pacifastin type motif has so far only been found in serine protease inhibitors of insects and crustaceans. The smallest inhibitors of this family are 35–38 residues long, and are made of one single cysteine-rich motif, whereas the largest inhibitor (Pacifastin from Pacifastacus leniusculus) contains multiple sets of this motif. In P. hypochondriaca CVP4 and Locusta migratoria migratorioides PI5 (CAD11969) the Pacifastin type motif occurs in triplicate, whereas the Nasonia cysteine-rich/Pacifastin venom proteins that we discovered in the present study are characterized by four repetitions of the same motif.

The trypsin inhibitor-like (TIL) type motif typically contains 10 cysteine residues that form five disulphide bonds. The cysteine residues that form the disulphide bonds are 1–7, 2–6, 3–5, 4–10 and 8–9. In the venom of P. hypochondriaca, CVP1 seems to be involved in stabilization or inhibition of venom phenoloxidase whilst it is stored in the venom sac. However, inhibitors acting on the hosts’ enzymes have also been reported. Indeed, the endoparasitoid C. rubecula produces a small venom protein, Vn4.6, that interferes with the activation of host haemolymph prophenoloxidase (Asgari et al., 2003a). It has only 65 residues and lacks a TIL domain. Microplitis demolitor bracovirus carried by the wasp M. demolitor produces an inhibitor Egf1.0 that also targets the host’s phenoloxidase cascade (Beck & Strand, 2007). Egf1.0 has a cysteine-rich motif with similarities to the TIL domain.

Serine protease inhibitor. In addition to the cysteine-rich venom proteins (see above) we found also two Kazal-type serine protease inhibitors and a small serine protease inhibitor. Several common structural features are observed in the family of single or multidomain Kazal-type inhibitors. These include a characteristic cysteine distribution pattern, a typical VCGxD sequence motif and highly homologous three-dimensional structures (Schlott et al., 2002). They have been found in different insect species, including Bombyx mori (Zheng et al., 2007), Triatoma infestans (Lovato et al., 2006), Galleria melonella (Nirmala et al., 2001) and Rhodnius prolixus (Friedrich et al., 1993). The small serine protease inhibitor that we found shows resemblance to that of Schistocerca gregaria (Gaspari et al., 2002).

Carbohydrate metabolism. In the functional category ‘carbohydrate metabolism’ the enzymes chitinase and trehalase were discovered by genome mining. The third member of this category, glucose dehydrogenase, has not been associated with insect venoms before and was found in the proteomic study.

Chitinase. Krishnan et al. (1994) were the first to describe an active chitinase stored in the chitin-lined venom reservoir of the adult female Chelonus sp. This chitinase is part of the venom injected from the ovipositor by the adult female. The enzyme is also released by teratocytes from Toxoneuron nigriceps (Hymenoptera: Braconidae) (Cônsoli et al., 2007). Teratocytes are cells that only occur in certain parasitic wasps and are derived from the serosal membrane of the wasp egg. They are released at the time the neonate larvae ruptures the egg covering. Subsequently, these cells circulate in the host haemolymph and synthesize proteins that may alter host physiology in support of endoparasitoid development (Rana et al., 2002). N. vitripennis do not release this type of embryonic cell, and the observed chitinase transcript shows the most resemblance to the Chelonus sp. chitinase gene.

Glucose dehydrogenase. Glucose dehydrogenase (EC 1.1.1.47) belongs to a family of oxidoreductases, specifically those acting on the CH-OH group of donor with NAD+ or NADP+ as acceptor. This enzyme participates in the pentose phosphate pathway.

Trehalase. Random sequencing of cDNAs from a venom gland library of P. hypochondriaca revealed a transcript that encodes a 61 kDa with extensive sequence similarities to trehalases, possibly having a nutritional function (Parkinson et al., 2003). This Pimpla protein served as a query record to find a Nasonia venom trehalase by blast-searching.

DNA metabolism. All members of this category were discovered by a proteomic approach. None of them had been found associated with the venom of parasitoids before.

Apyrases. Apyrase is a calcium-activated plasma membrane-bound enzyme (EC 3.6.1.5) that catalyses the hydrolysis of adenosine triphosphate (ATP) to yield adenosine monophosphate (AMP) and inorganic phosphate. Apyrase activity is present in the saliva of haematophagous arthropods (Faudry et al., 2006). It is related to blood-feeding because of apyrase’s ability to hydrolyse adenosine diphosphate (ADP), a key component of platelet aggregation. Recently, ADPases were also found in Brazilian snake venoms (Sales & Santoro, 2008).

Endonucleases. Endonucleases are enzymes that cleave the phosphodiester bond within a polynucleotide chain, in contrast to exonucleases, which cleave phosphodiester bonds at the end of a polynucleotide chain. They occur in nematocyst venom of the marine invertebrates Portuguese Man O’ War (Physalia physalis) (Neeman et al., 1980), East Coast sea nettle (Chrysaora quinquecirrha) (Neeman et al., 1981) and in snake venoms (Georgatsos & Laskowski, 1962). Our observation of an endonuclease in N. vitripennis venom is the first of its kind.

Inosine uridine-preferring nucleoside hydrolases. Inosine uridine-preferring nucleoside hydrolase (EC:3.2.2.1) catalyses the hydrolysis of all of the commonly occurring purine and pyrimidine nucleosides into ribose and the associated base, but has a preference for inosine and uridine as substrates. This enzyme was first discovered in the parasitic protozoon Crithidia fasciculata – which is deficient in de novo synthesis of purines – and serves to salvage the host purine nucleosides (Gopaul et al., 1996).

Glutathione metabolism. The idea that the venom of parasitoids has an impact on the host’s glutathione homeostasis is not new (Falabella et al., 2007). However, we were able to extend the enzyme families involved.

γ-glutamyl transpeptidase. Human γ-glutamyl transferase (EC 2.3.2.2) is an extracellular enzyme that is anchored to the plasma membrane of cells, where it hydrolyses and transfers γ-glutamyl moieties from glutathione (GSH) and other γ-glutamyl compounds to acceptors (Heisterkamp et al., 2008). The venom of the endophagous braconid Aphidius ervi contains a protein with a significant level of sequence identity with γ-glutamyl transpeptidase and it has been demonstrated that this enzyme has a negative impact on the reproductive activity of the wasp’s host by inducing apoptosis of the cells in the germaria and ovariole sheath (Falabella et al., 2007). Endogenous γ-glutamyl transpeptidase plays an important role in GSH homeostasis by removing the γ-glutamyl group from GSH, after which the resulting cysteinylglycine is cleaved by a membrane dipeptidase so that the released cysteine is transported into the cell and used for restoring the intracellular GSH concentration. Falabella et al. (2007) assumed that Aphidius venom γ-glutamyl transpeptidase might impact the delicate balance between intracellular and extracellular pools of GSH, possibly exposing the cell to oxidative stress and thus triggering apoptosis.

γ-glutamyl cyclotransferase. γ-glutamyl cyclotransferase (EC 2.3.2.4) is another enzyme that participates in GSH metabolism. It catalyses the chemical reaction in which the substrate (5-L-glutamyl)-L-amino acid is converted into 5-oxoproline and L-amino acid.

Esterases. A wide range of different esterases exists that differ in their substrate specificity, protein structure and biological function. Acid phosphatases and lipases are known to occur in hymenopteran venoms. We were able to extend this category with two interesting members: arylsulphatase and carboxyl-esterase, both found by proteomics.

Acid phosphatase. Two single domain and one multiple domain secretory acid phosphatases were found in the bioinformatic (XP_001600562) and the proteomic (XP_001605498; XP_001605344) studies. As only one out of 14 Nasonia acid phosphatases found by blast-searching was selected for subsequent RT-PCR confirmation, we cannot exclude the possibility that some venom acid phosphatases remained undiscovered.

Acid phosphatase activity has been demonstrated in relatively high levels in the venom of P. hypochondriaca, but nevertheless it plays no role in the in vitro anti-haemocytic activity displayed by whole venom preparations (Dani et al., 2005). Moreover, the pH optimum of Pimpla acid phosphatase is much lower than of the pupal Lacanobia oleracea host haemolymph (pH 6.7), suggesting that under these circumstances the enzyme would not be very active. Recently, the acid phosphatase cDNA from the venom apparatus of the endoparasitoid wasp Pteromalus puparum has been cloned and characterized (Zhu et al., 2008). The expression of this gene seems to be regulated at different developmental stages, with the highest level between 2 and 4 days after adult emergence. In honey bees it represents a major allergen (Api m 3) of the venom, responsible for bee sting allergy (Grunwald et al., 2006).

Arylsulphatase. The sulphatases constitute a conserved family of enzymes that specifically hydrolyse sulphate esters in a wide variety of substrates such as glycosaminoglycans, steroid sulphates, or sulpholipids. By modifying the sulphation state of their substrates, sulphatases play a key role in the control of physiological processes, including cellular degradation, cell signalling and hormone regulation (Diez-Roux & Ballabio, 2005). The venom of the black-necked spitting cobra Naja nigricollis was found to contain a high level of the enzyme arylsulphatase (Nok et al., 2003).

Esterase/lipase. Esterases and lipases act on carboxylic esters (EC: 3.1.1.–). The catalytic apparatus involves three residues (catalytic triad): a serine, a glutamate or aspartate and a histidine. One carboxyl-esterase and two lipases were identified in the present study. The carboxyl-esterase belongs to the acetylcholinesterase-juvenile hormone esterase family (Zera et al., 2002), but its exact function could not be predicted. Lipase activity has recently been found in the venom of P. hypochondriaca using the API ZYM semiquantitative colorimetric kit (Dani et al., 2005). Lipases perform essential roles in the digestion, transport and processing of dietary lipids in most living organisms. In general, it has been demonstrated that Nasonia venom induces alterations in the host’s lipid metabolism (Rivers & Denlinger, 1995).

Recognition/binding proteins. This represents a very interesting category of proteins that interact with a wide range of target molecules. Their biological function is probably equally diverse. They were all discovered by the proteomic approach and none have been found in association with insect venoms previously.

β-1,3-glucan recognition protein. Pattern recognition molecules serve as biosensors in the activation of innate immune responses in both vertebrates and invertebrates. The family of β-1,3-glucan recognition proteins (βGRPs) is involved in the recognition of β-1,3-glucans of fungi and Gram-negative bacteria. βGRPs from several arthropods have been implicated in the activation of a protease cascade that leads to prophenoloxidase activation (Bilej et al., 2001) and the activation of induction of antimicrobial peptide genes (Kim et al., 2000).

Chitin-binding protein. The protein that we found has a chitin-binding peritrophin-A domain (pfam01607: CBM_14) typical for chitin-binding proteins, particularly peritrophic matrix proteins of insects and animal chitinases (Elvin et al., 1996). In honey bees, the venom duct and reservoir are characterized by an epicuticular lining (Bridges & Owen, 2005) and proteomic analyses of gland tissue revealed two endocuticular proteins (Peiren et al., 2008). A cuticular lining also exists in the N. vitripennis reservoirs (Ratcliffe & King, 1969). The chitin-binding protein that we found is probably also associated with the soft-type cuticula.

General odorant-binding protein. Odorant-binding proteins (OBPs) are the main proteins involved in the interaction between odorants and the elements of the sensillar lymph (Pelosi & Maida, 1995; Pelosi, 1996). OBPs include pheromone-binding proteins (PBPs) and general odorant-binding proteins (GOBPs), both of which have molecular masses around 16 kDa and contain six conserved cysteine residues paired in three disulphide bridges (Li et al., 2008). Hitherto, GOBPs had not yet been described in the context of insect venoms.

Low-density lipoprotein receptor. The protein that we found has two low density lipoprotein receptor class A domains (cd00112), cysteine-rich repeats that play a central role in mammalian cholesterol metabolism; the receptor protein binds LDL and transports it into cells by endocytosis. The same conserved domain was also found in the Locusta migratoria lipophorin receptor (Dantuma et al., 1999).

Immune related proteins. Rivers et al. (2002b) reported that Nasonia venom inhibits host cellular immune responses. Two members of this category may play a role in this process: calreticulin and C1q-like venom protein, the latter being handled elsewhere in this supplement (de Graaf et al., 2009). The third member, the immunoglobulin-like protein, has two conserved domains in common with immune related proteins of higher organisms, but may have a strikingly different biological function in insects.

Calreticulin. Calreticulin is a multifunctional Ca²⁺-binding chaperone in the endoplasmic reticulum and expression of the protein is tightly regulated at the transcriptional level. A protein with similarities to calreticulin was found in C. rubecula venom fluid (Zhang et al., 2006) and its associated polydna particles (Asgari et al., 2003b). It was suggested that it competes for binding sites with host haemocyte calreticulin, known to mediate early-encapsulation reactions (Zhang et al., 2006).

Immunoglobulin-like protein. It was interesting to find a protein with an immunoglobulin (IG) (smart00409) and an IG cell adhesion molecule (IGcam) domain (cd00931). This IG-like venom protein showed most resemblance to the ecdysone inducible protein L2 of Aedes aegypti (EAT46727; E-value: 3e-33) and Apis mellifera (XP_393019; E-value: 9e-33) and the neural/ectodermal development factor IMP-L2 precursor of Pediculus humanus corporis (EEB17941; E-value: 1e-32) and Culex quinquefasciatus (EDS42108; 2e-32). These four proteins belong to the same IG superfamily, first discovered in a differential screen designated to isolate genes encoding 20-hydroxyecdysone-induced secreted or membrane-bound proteins of Drosophila (Natzle et al., 1986). IMP-L2 refers to inducible membrane-bound polysomal transcripts (Natzle et al., 1986), but also to imaginal morphogenesis protein-late 2 (Honegger et al., 2008). The protein has been implicated in neural and ectodermal development in Drosophila (Garbe et al., 1993). It has recently been assigned as a putative homologue of vertebrate insulin and insulin-like growth factor (IGF)-binding protein 7 that binds to Drosophila insulin-like peptide 2 (Dilp2) (Honegger et al., 2008). IMP-L2 seems to be the first functionally characterized insulin-binding protein in invertebrates, and serves as a nutritionally controlled suppressor of insulin-mediated growth in the fruit fly. Thus, the IG-like venom protein that we identified in this proteomic study could be an excellent candidate responsible for the developmental arrest seen in Nasonia-envenomated flesh flies (Rivers & Denlinger, 1994).

Others. This category includes proteins that are difficult to categorize but that show strong resemblances to known proteins. Antigen 5-like protein, aspartylglucosaminidase and laccase were all discovered by a query-based blast search and thus, have been described in hymenopteran venoms before. In contrast, the finding of aminotransferase in Nasonia venom was the first of its kind. Finally, our study has revealed no fewer than 23 proteins that did not show similarity to any known protein. In Table 1, these unknown proteins are categorized separately. At present, prediction of their function is impossible and demands continued investigation.

Aminotransferase. Aminotransferases (EC 2.6.1.X) are pyridoxal 5′-phosphate-dependent enzymes, which are ubiquitous in nature and play an important amino group-transferring role in nitrogen metabolism in cells (Hwang et al., 2005). The two members of this protein family that we found belong to the kynurenine aminotransferase subgroup, involved in the transamination of kynurenine to an α-keto acid. The molecular and biochemical characterization of kynurenine aminotransferase from Aedes aegypti mosquitoes has been reported (Fang et al., 2002). So far, this enzyme has not been associated with venomous secretions.

Antigen 5-like protein. Antigen 5 represents one out of three major allergenic proteins of the venom from yellow jackets of the genus Vespula and from hornets of the genera Vespa and Dolichovespula. The amino acid sequences from 19 different wasp species are available (Hoffman, 2006) and the crystal structure of Ves v 5 has been solved (Henriksen et al., 2001), but despite this, the biological function of antigen 5-like proteins has not yet been identified. The two antigen 5-like venom proteins that we discovered are the first to be derived from parasitoid wasp species.

Aspartylglucosaminidase. Aspartylglucosaminidase or N4-(β-N-acetylglucosaminyl)-L-asparaginase (EC 3.5.1.26) is an amidohydrolase enzyme involved in the catabolism of N-linked oligosaccharides of glycoproteins. It usually cleaves asparagine from N-acetylglucosamines as one of the final steps in the lysosomal breakdown of glycoproteins. Aspartylglucosaminidase is one of the main components of venomous secretions of the endoparasitic wasp Asobara tabida (Moreau et al., 2004), although its activity could not be demonstrated.

Laccase. The crude venoms isolated from N. vitripennis (Abt & Rivers, 2007) and P. hypochondriaca (Parkinson & Weaver, 1999) were found to possess phenoloxidase (PO) activity, but only in the latter could this activity be attributed unambiguously to the arthropod-specific PO enzyme type, encoded by three genes (POI-III) that derived by gene duplication (Parkinson et al., 2001). In Nasonia, venom PO activity appears to play an important role in cell death, but in contrast to Pimpla venom POs, this is not related to the formation of toxic phenolic compounds. Indeed, no melanization occurs in Nasonia venom-intoxicated cell cultures, and host haemolymph does not melanize following envenomation (Rivers et al., 2002b)). Abt & River (2007) found that this PO activity can be associated with two proteins with estimated molecular weights of 160 and 68 kDa, as determined by SDS-PAGE separation followed by in-gel staining for PO activity using L-DOPA as a substrate. Laccase belongs to a group of proteins collectively known as multicopper oxidases and is hypothesized to play an important role in insect cuticle sclerotization. During this extracellular process, cuticular proteins are cross-linked into a matrix as a result of oxidative and nucleophilic reactions of catechols to their corresponding quinones (Suderman et al., 2006). Thus, laccase has similar PO activity to the enzyme phenoloxidase, which could not be found in the present study. Parkinson et al. (2003) have already suggested that the P. hypochondriaca venom laccase is also involved in L-DOPA oxidizing activity. Interestingly, the theoretical molecular weight of the laccase that we found, without its signal peptide, is exactly 68 kDa, thus corresponding perfectly well with the molecular weight of the lower band in the L-DOPA substrate gel of Nasonia venom (Abt & River, 2007).

Discussion

Our in-depth search for venom constituents, that combined genome mining with confirmative RT-PCR and two complementary mass spectrometric tools, revealed an unexpectedly rich composition of N. vitripennis venom. With 79 venom proteins identified, N. vitripennis has more characterized venom components than any other hymenopteran. Significant progress has been made in unravelling the venom composition of the other hymenopteran with a sequenced genome, the honey bee (Peiren et al., 2005). However, it seems reasonable to believe that the latter’s venom complexity is inferior to that of the Nasonia wasp. This is probably because of the differences in biological function of the venoms of honey bees and parasitoid wasps. In the former, venom must inflict pain as an act of defence; in the latter, venom must target the host’s immunity, physiology, mobility, reproductive capacity and even their behaviour in order to guarantee the progenies’ development.

Amongst the proteins identified by the proteomic studies, 15 seem to be nonsecretory and probably do not represent venom constituents. In venomic research it is not uncommon to find so-called ‘venom trace elements’ with only a local function in the venom duct or reservoir or released by leakage of the gland tissue (de Graaf et al., 2010). The finding of actin strongly indicates that some cytoplasmatic proteins might have been released while the venom sacs were spun-down during venom collection.

The limited data overlap between the bioinformatic and the proteomic studies can be explained partially by the fact that the bulk of the venom (proteomic study) is synthesized in the columnar cells of the acid gland, whereas the RT-PCR (bioinformatic study) focused only on a group of unique reservoir cells (Ratcliffe & King, 1969). Possibly these two compartments produce distinct protein sets, an assumption already made by others (King & Ratcliffe, 1969). We cannot exclude the possibility that some of the newly discovered proteins derived from the small clusters of cells of uncertain origin that were attached to the reservoirs during sampling.

Remarkably, half of the 79 identified venom proteins were not yet associated with insect venoms: 16 proteins showed similarity only to known proteins from other tissues or secretions, and an additional 23 did not show similarity to any known proteins. Our discovery will certainly facilitate access to this impressive untapped pharmacopoeia. In addition, several proteins known to occur in the venoms of P. hypochondriaca and Cotesia spp. were also found in Nasonia, suggesting that endoparasitoids and ectoparasitoids have more venom constituents in common than previously thought.

Experimental procedures

Chemicals and reagents

Dithiothreitol (DTT), Iodoacetamide (IAA), KH₂PO₄, KCl and dibasic ammonium citrate were purchased from Fluka (Buchs, Switzerland). Alfa-cyano-4-hydroxycinnamic acid for MALDI analyses was obtained from Sigma (St. Louis, MO, USA) and HPLC-grade acetonitrile (ACN) from Biosolve (Valkenswaard, the Netherlands). Trifluoroacetic acid (TFA) was from Pierce (Rockford, IL, USA) and formic acid from Panreac (Barcelona, Spain). Sequencing grade modified trypsin was obtained from Promega (Madison, WI, USA). The mass standard calibration mixture for MALDI-TOF calibration was purchased from Applied Biosystems (Foster City, CA, USA) and caffeine, L-methionyl-arginyl-phenylalanyl alanine acetate·H₂O (MRFA) and UltraMark for calibrating the FT-ICR and LTQ mass analysers were from Thermo Fisher Scientific (Waltham, MA, USA).

Bioinformatic screening of the Nasonia genome

Sequence similarity was investigated using the blast algorithm (Tatusova & Madden, 1999) available at http://www.ncbi.nlm.nih.gov or as a stand-alone software package. The conserved domain architecture was predicted with cdart (Geer et al., 2002), the presence of signal peptides was verified using the SignalP 3.0 Server (Emanuelsson et al., 2007) at http://www.cbs.dtu.dk/services/SignalP/ and multiple sequence alignment was performed with the ClustalW program (Thompson et al., 1994). The expression of selected proteins by the reservoir secretory cells was confirmed by RT-PCR. The procedure of tissue collection will be described in the next paragraph; cDNA was prepared following a procedure outlined in an earlier paper (Peiren et al., 2006). Primer sets are given in Supporting Information Table S1. Amplicons purified using the Illustra_ GFX_ PCR DNA and Gel Band Purification Kit (GE Healthcare, Diegem, Belgium) were sequenced on an Applied Biosystems 3130XL automated DNA sequencer (Applied Biosystems) with 50 cm capillaries filled with POP-7 polymer, using the ABI Prism BigDye V 3.1 Terminator Cycle Sequencing kit.

Venom sample preparation

Venom reservoirs from emerging N. vitripennis females were dissected in ice-cold physiological solution under binocular microscope. The reservoirs are firmly attached to the vagina, near the proximal end of the ovipositor shaft and can be unambiguously distinguished by their location, size, morphology (the reservoir has a small nick on the side opposite to the secretory region) and unique light scatter (Fig. 3). Often small clusters of cells of uncertain origin were attached to the reservoir and could not be removed without risking the eruption of the reservoir. The reservoirs were transferred several times to a fresh droplet of liquid in order to clean them from the outside. The content of 10 reservoirs (that were collected in 10 µl physiological solution) was finally released by centrifugation and separated from the reservoir secretory cells. This crude venom preparation was initially desalted using Vivaspin (Sartorius Stedim Biotech, Aubagne Cedex, France) centrifugal filters with a 5 kDa molecular weight cut-off by three consecutive washing steps in 200 µl pure Milli-Q (MQ) water. The sample was concentrated to a volume of 20 µl. The proteins were reduced by adding 2 µl of 50 mM DTT solution in MQ-water for 10 min at 100 °C, and alkylated by adding 2 µl 100 mM IAA solution in MQ-water for 2 h at 37 °C. Subsequently, the proteins were digested by adding 15 µl of 0.1 µg/µl trypsin solution. After overnight incubation at 37 °C, tryptic peptides were dried in a speedvac (Thermo Savant, Holbrook, NY, USA) and dissolved in 15 µl 5% ACN/0.1% formic acid (v/v).

Figure 3.

(A) Photograph of a dissected Nasonia venom reservoir. On the sketch in (B) we marked the characteristic nick (black arrow head) and the small cluster of cells of uncertain origin (c).

Two-dimensional liquid chromatography

Ten µl digested venom extract was loaded on an off-line 2D-LC-MALDI-TOF MS setup as previously described (Vanrobaeys et al., 2005). In this study, an Ultimate Plus Dual-Gradient Capillary/Nano LC system (Dionex-LC Packings, Hercules, CA, USA) was used. In the first dimension, peptides were separated on a SCX column, 150 mm × 0.3 mm, packed with POROS 10S (Dionex-LC Packings). The capillary pump generated a two-step linear gradient from 0 to 40% buffer SCX-B in 30 min and from 40 to 100% buffer SCX-B in 15 min at a 5 µl/min flow rate. Buffer SCX-A was 5 mM KH₂PO₄/5% ACN (v/v) in MQ-water; buffer SCX-B was 5 mM KH₂PO₄/5% ACN (v/v)/500 mM KCl in MQ-water. LC-effluent was fractionated at 5 min intervals using a Probot spotting device (Dionex-LC Packings). SCX-fractions were dried and dissolved in 15 µl 5% ACN/0.1% formic acid (v/v) and further unravelled via RP chromatography. Ten µl SCX-fraction was loaded onto a C18 microguard column, 2 mm × 800 mm (Dionex-LC Packings) at a 10 µl/min flow rate with buffer RP-A. After valve switching, peptides eluted from the trapping column and were separated on a PepMap C18 analytical column, 150 mm × 75 µm (Dionex-LC Packings), at a 250 nl/min flow rate. Again, a two-step gradient was generated: 0–50% buffer RP-B in 25 min and 50–100% buffer RP-B in 10 min. Buffers RP-A and RP-B were 5% ACN/0.05% TFA (v/v) and 80% ACN/0.05% TFA (v/v) in MQ-water, respectively. RP-LC effluent was spotted every 30 s on a MALDI-plate using the Probot spotter. Afterwards, 0.5 µl of a MALDI-matrix solution (10 mg/ml alfa-cyano-4-hydroxycinnamic acid dissolved in a 50% ACN/0.1% TFA/10 mM dibasic ammonium citrate) was spotted over the dried RP LC-fractions using the Freedom EVO robotic setup (Tecan, Männedorf, Switzerland). SCX and RP separations were monitored using, respectively, a 45 nl or 3 nl UV-flow cell at 214 nm.

LC-MALDI-TOF MS

MALDI-TOF/TOF measurements were performed on a 4700 Proteomic Analyzer (Applied Biosystems, Foster City, CA, USA). Prior to data acquisition, the instrument was calibrated with the mass standard calibration mixture containing angiotensin I, Glu-fibrinopeptide, adrenocorticotropic hormone (ACTH) (clip 1–17) and ACTH (clip 18–39). MS-spectra were acquired by collecting 2000 subspectra at a laser intensity of 4300. After all MS-spectra from a particular LC-run were collected, a job-wide interpretation method automatically selected five precursor peaks per MALDI-spot for MS/MS. MS/MS-spectra were acquired by accumulating 5000 subspectra at a laser intensity of 4900. Finally MS/MS-spectra were subjected to a MASCOT (Matrix Science, London, UK) database search using the GPS explorer v2.0 software (Applied Biosystems). Peptides were identified using a N. vitripennis Gnomon database. Carbamidomethylation on cysteine residues and oxidation of methionine were selected as fixed or variable modifications, respectively, and two tryptic miscleavages were allowed.

LC-ESI-FT MS

For LC-ESI-FT MS analysis, 5 µl from all SCX-fractions described above were separated on an Agilent 1200 chromatographic system equipped with a Zorbax 300SB-C18 trapping column, 5 mm × 0.3 mm, and a Zorbax 300SB-C18 analytical column, 150 mm × 75 µm (Agilent, Santa Clara, CA, USA). Samples were initially trapped via a capillary pump at a 4 µl/min flow rate. After valve switching, peptides were separated on the analytical column at a 300 nl/min flow rate by applying a 35 min linear gradient ranging from 2% ACN/0.1 formic acid to 80% ACN/0.1% formic acid in water. The LC-effluent was directly coupled to a Triversa NanoMate ESI source (Advion, Ithaca, NY, USA), working in nano-LC mode. Eluting peptides were sprayed into the mass spectrometer source using D-chips (Advion) on which a 1.55 kV voltage was supplied. The LTQ-FT Ultra mass spectrometer (Thermo Fisher Scientific) was tuned and calibrated with caffeine, MRFA and UltraMark before measurement. The FT-ICR mass analyser acquired MS-scans during the LC run at a 100 000 resolution. Using the preview mode, MS/MS fragmentation spectra were acquired in the LTQ XL Ion Trap for the three most intense ions. Raw LC-MS/MS data were subsequently analysed with the Sequest database searching algorithm implemented in the Bioworks v3.3.1 software (Thermo Fisher Scientific). MS/MS data were searched against the Gnomon protein databases from N. vitripennis, concatenated with a shuffled decoy database generated with the Decoy Database Builder software (Reidegeld et al., 2008). Carbamidomethylation on cysteine residues was allowed as a variable modification, two tryptic miscleavages were permitted and precursor mass and b and y ion tolerance were set to 10 ppm and 0.5 amu, respectively. According to the manufacturer’s guidelines, only peptide hits with XCorr values higher then 2.0, 2.5 and 3.5 for charge states +2, +3 and +4, respectively, were retained in the final peptide list as a good criterion for positive identification.