Two novel proteins expressed by the venom glands of Apis mellifera and Nasonia vitripennis share an ancient C1q-like domain


Dirk C. de Graaf, Laboratory of Zoophysiology, Ghent University, K.L. Ledeganckstraat 35, B-9000 Ghent, Belgium. Tel.: + 32 9264 8732; fax: + 32 9264 5252, e-mail:


An in-depth proteomic study of previously unidentified two-dimensional polyacrylamide gel electrophoresis spots of honey bee (Apis mellifera, Hymenoptera) venom revealed a new protein with a C1q conserved domain (C1q-VP). BlastP searching revealed a strong identity with only two proteins from other insect species: the jewel wasp, Nasonia vitripennis (Hymenoptera), and the green pea aphid, Acyrthosiphon pisum (Hemiptera). In higher organisms, C1q is the first subcomponent of the classical complement pathway and constitutes a major link between innate and acquired immunity. Expression of C1q-VP in a variety of tissues of honey bee workers and drones was demonstrated. In addition, a wide spatial and temporal pattern of expression was observed in N. vitripennis. We suggest that C1q-VP represents a new member of the emerging group of venom trace elements. Using degenerate primers the corresponding gene was found to be highly conserved in eight hymenopteran species, including species of the Aculeata and the Parasitica groups (suborder Apocrita) and even the suborder Symphyta. A preliminary test using recombinant proteins failed to demonstrate Am_C1q-VP-specific immunoglobulin E recognition by serum from patients with a documented severe bee venom allergy.


The hymenopteran venom glands evolved from their ancestral female accessory reproductive glands and consequently are restricted to female bees and wasps. Depending on the insect's lifestyle, the venom has different biological functions. In the highly eusocial honey bee, Apis mellifera, the venom represents an important defence weapon of the queen and the worker bees. Queen bees use this weapon soon after emergence, during fights with other rival queens, whereas worker bees use it to protect the colony or themselves when they are exposed to dangers and predators (Schmidt, 1995). Honey bee venom is lethal for other insect species, whereas in higher animals a single sting will mostly cause only local reactions with pain, swelling, redness and itching. Massive attacks, for instance by the more aggressive Africanized honey bees, can be deadly for humans (Tunget & Clark, 1993). However, fatalities from bee stings are more frequently the result of the victim's hypersensitivity than envenomation (Vetter et al., 1999). In parasitoid wasps, which deposit their eggs inside or outside their arthropod prey, venom has a completely different function and mainly serves to promote development and survival of their progeny (Moreau & Guillot, 2005). Although the parasitoid lifestyle by definition will ultimately cause the death of the prey, this is not always the immediate result of envenomation. Instead, the venom of parasitoid wasps may affect the prey's physiology by causing paralysis (Moreau & Guillot, 2005), suppression of its immunity (Asgari et al., 2003), apoptosis in the ovaries (Falabella et al., 2007) and release (Rivers & Denlinger, 1994, 1995) or conversion of nutritional biomolecules (Dani et al., 2005).

As a consequence of the diverse biological functions of the venoms of hymenopteran species, different protein compositions can be expected. Indeed, the venom of several stinging social bees and wasps contains hyaluronidase, phospholipase A1 (PLA1) or A2 (PLA2) and a number of cytolytic or neurotoxic peptides, all aimed at causing as much damage as possible. Hyaluronidase (EC catalyses the hydrolysis of hyaluronan and chondroitin sulphates, the most abundant glycosaminoglycans of the vertebrate extracellular matrix, and therefore causes the spreading of venom constituents throughout the body (Kreil, 1995). PLA1 (EC and PLA2 (EC are acyl hydrolases that target, respectively, the sn-1 and sn-2 fatty acid on the glycerophospholipid backbone to produce glycerophosphocholines. Thus, they have the ability to disrupt the phospholipid packings from several types of biological membranes, leading to pore formation and/or cell lysis (Dotimas & Hider, 1987). Melittin is the principal toxic bee venom constituent, representing 50% of its dry weight. It is a cationic, haemolytic peptide of 26 residues with amphiphilic properties that make it water-soluble and yet it spontaneously associates with natural and artificial membranes (Raghuraman & Chattopadhyay, 2007). Apamin, an 18-amino-acid peptide neurotoxin, potently and selectively inhibits small conductance Ca2+-activated K+ channels in both the central nervous system and peripheral tissues (Garcia et al., 1991).

In contrast, the venoms of parasitoid wasps contain a number of proteins with completely different physiological functions. The protein γ-glutamyl transpeptidase (EC of the endophagous braconidAphidius ervi has a negative impact on the reproductive activity of its host by inducing apoptosis of the cells in the germaria and ovariole sheath (Falabella et al., 2007). The endoparasitoid wasp Cotesia rubecula injects different maternal factors that interfere with host immune reactions: a serine proteinase homologue inhibits melanization of the haemolymph (Asgari et al., 2003) and a calreticulin-like protein inhibits haemocyte spreading behaviour, thus preventing encapsulation of the developing parasitoid (Zhang et al., 2006). Lastly, aspartylglucosaminidase-like protein represents one of the main venom proteins of Asobara tabida. It has been suggested that hydrolysis of its glucosylasparagine substrate leads to the production of aspartate, which is a known excitatory neurotransmitter of the host (Drosophila) nervous system, and thus might be involved in the temporary paralysis of the fly (Moreau et al., 2004).

The present study was initially undertaken to identify a number of unidentified spots from a previous gel-based proteomic study of the venom of the honey bee (Peiren et al., 2005). By using a three-fold protein concentration we succeeded in identifying one additional venom protein characterized by an ancient C1q-like domain. Remarkably, homologous genes of this C1q-like venom protein (C1q-VP) were also found in a broad range of hymenopterans including species of the Aculeata and the Parasitica groups (suborder Apocrita) and even the suborder Symphyta.

Results and discussion

Honey bee venom proteomics

The abundance of PLA2 in honey bee venom (12% of its dry weight) interferes with gel-based proteomic applications. PLA2 protein heterogeneity, with different isoforms of 16 kDa (unglycosylated), 18 kDa (containing mannose, fucose and N-acetylamine) and 20 kDa (with additional N-acetylgalactosamine) (Altmann et al., 1991), can easily be distinguished on two-dimensional (2D) gels when up to 1 mg of protein is loaded (Peiren et al., 2005). Once above this threshold, the 2D-gel image becomes largely obscured by distortions in the alkaline isoelectric point (pI) range because of PLA2 and other abundant proteins and peptides, as shown in Fig. 1A. Nevertheless, these overloaded 2D-gels enabled us to improve the spot intensity and resolution of some minor proteins in the neutral pI range, which remained hardly detectable in the preceding study (Peiren et al., 2005). Seven protein spots were retained for further analyses: four corresponding to previously unidentified spots (spots 2, 3, 5 and 7) and three new ones (spots 1, 4 and 6).

Figure 1.

Gel-based proteomics of honey bee venom. (A) Overloaded 2D-gel stained with Coomassie Brilliant Blue G-250. Numbered spots were excised and subjected to in-gel tryptic digestion and mass spectrometric analysis. Spot 6 corresponds to Apis mellifera C1q-like venom protein (Am_C1q-VP). (B) Mass spectrometric identification of Am_C1q-VP. The main figure contains the peptide mass fingerprint, showing some phospholipase A (PLA2) peptides, as well as two peptides that were unambiguously identified as C1q-VP peptides. T indicates a trypsin autolysis fragment. The insert shows the tandem mass spectrometry spectrum of the ion at m/z 1201.60.

Spot 6 contained PLA2 as well as peptides that were unambiguously identified as peptides derived from a hypothetical protein (gi|66524376|) that we named Apis mellifera C1q-like venom protein (Am_C1q-VP). The MASCOT score was 52 and the P-value was 8.5e−005. Figure 1B shows the peptide mass fingerprint as well as a tandem mass spectrometry (MS/MS) spectrum from the major peak in the fingerprint spectrum at mass-to-charge ratio (m/z) 1201.6, showing the identification of the peptide with sequence ISDGVTFSGYR obtained from this protein. The other six analysed spots yielded no additional identification.

Sequence analysis and comparisons

The full 492 bp coding sequence (stop codon included) of Am_C1q-VP was successfully amplified from bee venom gland cDNA and cloned in the pET prokaryotic expression vector and then sequenced. The obtained nucleotide sequence matched perfectly with the record XM_623476, representing the mRNA sequence of the hypothetical protein (gi|66524376|) found in the proteomic study (see above), and was deposited in GenBank (accession number FJ502840). The deduced amino acid sequence represents a polypeptide of 163 amino acids with a signal peptide cleavage site between positions 19 and 20 (Fig. 2A). The mature protein has three putative N- or O-linked glycosylation sites as determined by NetNGlyc 1.0 and NetOGlyc 3.1 (Julenius et al., 2005). Study of its conserved domain architecture revealed a C1q domain (cl02466) referring to the subunit of the C1 enzyme complex that initiates the activation of the serum complement system in higher organisms. BlastP searching revealed strong identity (score up to 196, E-value: 5e−49) with only two hypothetical proteins: XP_001608267 from the jewel wasp, Nasonia vitripennis (Hymenoptera), and XP_001951534 from the green pea aphid, Acyrthosiphon pisum (Hemiptera). These proteins have a sequence identity of 59.4 and 36.7%, respectively, when compared to Am_C1q-VP. Their mature protein starts at the same relative position (Fig. 2A) and both are characterized by a C1q domain. We further name these two proteins Nv_C1q-VP and Ap_sC1q-VP, respectively (we will only demonstrate that Nv_C1q-VP is likewise expressed in the venom glands – see below – and therefore refer to the Acyrthosiphon pisum homologue as ‘similar to’ C1q-VP; sC1q-VP).

Figure 2.

(A) Multiple sequence alignment by ClustalW of C1q-like venom protein (C1q-VP) homologues of Apis mellifera, Nasonia vitripennis and Acyrthosiphon pisum. The signal peptide is underlined and N- or O-glycosylation sites are marked in grey. In two highly conserved regions (boxes 1 and 2) degenerate primers were developed. (B) Deduced amino acid sequences of A. mellifera C1q-VP (Am_C1q-VP) with the corresponding cDNA sequences of Ap. mellifera and N. vitripennis beneath for comparison. =, position with identical nucleotides in the two transcripts; 0, position with different nucleotides in the two transcripts. Selected primer annealing sites are underlined.

We also amplified and sequenced the coding sequence of Nv_C1q-VP (GenBank accession number FJ502841) from a N. vitripennis venom gland cDNA preparation. There was a 100% match with the coding sequence of the predicted mRNA sequence XM_001608217. Nv_C1q-VP was annotated as a new venom constituent of N. vitripennis.

In higher organisms, C1q is the first subcomponent of the classical complement pathway and a major link between innate and acquired immunity. The human C1q molecule is composed of six copies of three polypeptide chains (A, B and C). Each of these chains has a short N-terminal region (containing a half-cysteine residue involved in interchain disulphide bond formation), followed by a collagen like sequence and a C-terminal globular C1q region of approximately 135 residues (Kishore & Reid, 1999). In recent years, many additional human proteins with C1q domains have been identified, including adiponectin, C1qTNF (C1q and tumor necrosis factor related protein), precerebellin, alpha-1 type X collagen, elastin microfibril interfacer 1, gliacolin, multimerin and otolin. There are 15 highly conserved residues within the C1q domain, of which eight are invariant within the human gene set and predicted to cluster within the hydrophobic core of the protein (Tang et al., 2005). In the Apis, Nasonia and Acyrthosiphon (s)C1q-VPs, these core residues are likewise largely conserved (11/15, 11/15 and 10/15, respectively; not shown) and clearly define these three insect proteins as members of the C1q domain-containing (C1qDC) protein family. With the finding of C1qDC-encoding sequences in the sea urchin Strongylocentrotus purpuratus, their occurrence may be extended to an invertebrate species (Tang et al., 2005). However, insect C1qDC proteins have so far not been reported.

In the human C1q molecules, the N-terminal collagen-like domains from the three polypeptide chains (A, B and C) form a base stalk that then forms a bouquet-like arrangement with six branches, each represented by a heterotrimer of the C-terminal globular C1q regions of the A, B and C chains (Tang et al., 2005). The C1q-VPs that we found contain no additional conserved domains, suggesting other structural features. Nevertheless, recent studies have reported the direct interaction of the globular C1q domain of human C1q with the lipopolysaccharides from Gram-negative bacteria (Salmonella typhimurium) in a calcium-dependent manner (Roumenina et al., 2008). It seems interesting to further explore possible involvement of the C1q-VPs in the cellular immune response of insects, particularly as opsonizing agents during phagocytosis. In humans, the interaction between complement component C1q and calreticulin is well documented (Pagh et al., 2008). Monocyte endocytic receptor CD91 is implicated in the endocytosis of apoptotic neutrophils via interactions with C1q and calreticulin. The addition of C1q and calreticulin significantly enhanced the phagocytosis of apoptotic cell debris by monocyte-derived cells in systemic lupus erythematosus patients (Donnelly et al., 2006). The recent discovery of calreticulin in the N. vitripennis venom gland transcriptome (de Graaf et al., 2009) made it worthwhile to investigate whether the interaction between Nv_C1q-Vp and calreticulin also exists. Venom-derived calreticulin from the endoparasitoid wasp Cotesia rubecula was found to inhibit host haemocyte spreading behaviour (Zhang et al., 2006), possibly by competing for binding sites with the host haemocyte calreticulin, which mediates early-encapsulation reactions.

Expression studies in Apis and Nasonia

We examined the spatial distribution of Am_C1q-VP in honey bee workers and drones (mucus gland) by reverse transcription PCR using separated organs and tissues (Fig. 3A). The expected 492 bp band was not only detected in venom glands, but also in hypopharyngeal glands, salivary glands, brains, thoracic muscles, midguts, haemocytes and deviscerated abdomens. Am_C1q-VP was likewise expressed in the male accessory reproductive glands or mucus glands. The spatial distribution of Nv_C1q-VP in N. vitripennis was performed using whole tagmata (head and abdomen). The expected 483 bp band was found in PCR reactions with abdominal cDNA of both sexes, not with head cDNA (Fig. 3B). The female abdomen yielded only a faint band on agarose gel when compared to the dissected venom glands.

Figure 3.

Spatial distribution of the C1q-like venom protein (C1q-VP) homologues as determined by reverse transcription PCR. (A) In honey bees C1q-VP is expressed in venom glands (VeG), hypopharyngeal glands (HyG), salivary glands (SaG), brains (Br), midguts (MG), haemocytes (He), deviscerated abdomens (dAb) and mucus glands (MuG; drones), not in thoracic muscles (Mus). A cDNA control (cDNA_ctl) based on the profilin mRNA sequence (gi|148231751|), and a genomic DNA control (gDNA_ctl) developed in intron 2 of icarapin (DQ485318), were run in parallel. (B) In male and female Nasonia vitripennis wasps C1q-VP is expressed in the abdomens (Abd; faint band in female abdomen was only visible with the naked eye), not in the heads (Hea). The PCR reaction with dissected venom glands is given for comparison. Nasonia vitripennis cDNA_ctl was developed in ribosomal protein L13 (gi∣156544547∣) and the gDNA_ctl in venom γ glutamyl transpeptidase-like 1 (one primer at the intron/exon transition, the second primer in the exon).

These findings were extended by TBlastN search (search of a translated nucleotide database using a protein query) of different expressed sequence tag (EST) databases of selected tissues. Am_C1q-VP was found in the RIKEN full-length enriched honey bee cDNA library of the head (HGSC, 2006). The identified sequence (DB736445.1) gave an identity of 99.35% with the query protein. Nv_C1q-VP matched two sequences in the N. vitripennis ESTs (GenBank: ES634296.1 and ES644094.1) and one in the Nasonia giraulti ESTs (GenBank: ES618760.1). One EST each from N. vitripennis and N. giraulti came from pupal/adult libraries, whereas the remaining N. vitripennis EST originated from a larval library. Expression of Nv_C1q-VP was also found in tiling array data (cDNA hybridized to an array of genomic sequences) of adult female reproductive tracts and male testes (J. Colbourne, unpubl. data). These data suggest that Nv_C1q-VP has a wide spatial and temporal pattern of expression, rather than being uniquely expressed in female venom glands.

With this information, the possible function of C1q-VP should be seen in a totally different light. It has recently been suggested that in-depth proteomic studies of honey bee venom will discover more proteins whose functions do not correspond with the principal function of venom as a defence weapon (de Graaf et al., 2010). This group of proteins is called ‘venom trace elements’. It concerns proteins present in extremely low quantities and with only a local function in the venom duct or the reservoir (maturation and stabilization of the secretions, protection and recovery of the tissue) or which are normal cell components, released by leakage of the gland tissue. The fact that C1q-VP was initially discovered as a tiny protein spot in an overloaded 2D-gel, and that it was produced by the venom glands of Apis mellifera and N. vitripennis– two species in which the venoms have completely different functions – suggests that it represents such a venom trace element. Its occurrence in other tissues of these insects further supports this view. However, despite this, we have kept the name C1q-like venom protein, as it refers to the secretion in which it was first discovered. Dipeptidyl peptidase IV represent another venom trace element. It is a membrane-anchored enzyme responsible for the processing of a major bee venom constituent, melittin, and thus not supposed to be found in this venomous secretion. Nevertheless, it was recently added to the International Union of Immunological Societies list of allergens ( as Api m 5, indicating that even at this low abundance a protein can be of immunological significance.

Occurrence of C1q-VP in other species

Using a degenerate oligonucleotide primer set designed in two conserved domain boxes, PCR was performed with genomic DNA from six additional hymenopteran species, including five from Aculeata and one from Symphyta (sawflies). All reactions resulted in a 176 bp amplicon. The PCR products were sequenced and the deduced amino acid sequences showed a high level of identity and were easily alignable (Fig. 4A), with expected high levels of nucleotide divergence reflecting saturation at silent codon positions amongst divergent taxa. We were unable to identify C1q-VP-encoding genes (by Blast searching) in other insect genomes, except in the Acyrthosiphon pisum genome, suggesting that this particular domain-containing gene is highly divergent in other taxa.

Figure 4.

(A) Multiple sequence alignment by ClustalW of a highly conserved amino acid sequence of different hymenopteran species derived by PCR on genomic DNA using degenerate primers. Taxonomic position (suborders) of each species (family names are provided when determination was not conclusive until the species level) is provided, together with the group (Gr) to which it belongs (Acu, Aculeata; Par, Parasitica). (B) Unrooted neighbour-joining tree, showing the associations between different species of the Aculeata group.

Immune recognition of recombinant Am_C1q-VP

We were able to produce recombinant Am_C1q-VP in Escherichia coli and to purify it by affinity chromatography (Fig. 5). A preliminary test using a simple spot blot technique with recombinant protein failed to show any Am_C1q-VP-specific Immunoglobulin E (IgE) recognition by sera from patients with a compelling history of bee venom allergy (Fig. 6). However, a similar test with recombinant icarapin gave a positive reaction in four out of five bee venom allergic patients, all beekeepers (Peiren et al., 2006). This could indicate that Am_C1q-VP is not involved in the adverse immune response against bee venom or that the prevalence is too low to be detected with such a small number of human sera.

Figure 5.

Recombinant expression of Apis mellifera C1q-like venom protein (Am_C1q-VP). The lysates of unstimulated (U) and stimulated (S) cell cultures were separated by sodium dodecyl sulphate polyacrylamide gel electrophoresis [and subsequently stained with Coomassie Brilliant Blue R-250 (Coo)]. Anti-His (α-His) immunostaining served to localize the Am_C1q-VP band (indicated by an arrow head; cross-reactive bands are indicated by an asterisk). Purification on a Profinity IMAC Ni-Charged Resin yielded an almost pure recombinant protein sample.

Figure 6.

Venom-specific IgE immunorecognition of (recombinant) Apis mellifera C1q-like venom protein (Am_C1q-VP) as determined by spot blot analyses using sera from patients with bee allergy, wasp allergy or neither. Recombinant equine uterocalin served as irrelevant control protein. From each serum sample the total IgE, wasp venom-specific IgE and bee venom-specific IgE titre are given.

Experimental procedures

Animals, venom and tissue collection

Worker honey bees and drones (Apis mellifera carnica) were from the experimental apiary of Ghent University, Belgium. Bumble bees (Bombus terrestris) were kindly provided by Biobest NV, Westerlo, Belgium. All other hymenopteran species were from the Rochester University (USA) collection: Calliopsis puellae, Diadasia diminuta, Chrysididae, Pheidole tetra, N. vitripennis and Argidae.

Honey bee venom was collected by ‘manually milking’ as described elsewhere (Peiren et al., 2005). The N. vitripennis venom gland tissue preparation was made by dissecting the venom reservoirs and associated glands in phosphate buffered saline (PBS), pH 7.2, and separating secretions from the tissues by rapidly spinning. After removal of the supernatant, the pellet was resuspended in RNALater® (Ambion, Austin, TX, USA). The honey bee venom glands were removed from the abdomen and then cut free from the reservoirs and immersed in RNALater®. All other bee tissues were dissected under a binoculary microscope in liquid (RNALater®) and further stored in fresh solution. The different N. vitripennistagmata were also separated in RNALater® by precisely cutting along their transitions with a pair of scissors.

Gel-based proteomics

Two-dimensional gel electrophoresis was performed using immobilized pH gradient strips (pH 3–10; Bio-Rad, Hercules, CA, USA). Proteins were resolved by isoelectric focusing in the first dimension and sodium dodecyl sulphate polyacrylamide gel electrophoresis (15% T and 2.6% C) in the second dimension. Protein spots were visualized by Coomassie Brilliant Blue R 250 staining. Seven spots were retained for further processing: four previously unplaced spots (Peiren et al., 2005), together with three new ones. They were manually excised twice from different gels and pooled, and each sample was then subjected to in-gel digestion as described elsewhere (Scharlaken et al., 2007) using modified trypsin (Promega, Madison, WI, USA) reconstituted in 50 mM ammonium bicarbonate at pH 7.8. After digestion, peptides were extracted from the gel pieces with 60% acetonitrile containing 0.1% formic acid. For mass spectrometry, the samples were dried in a SpeedVac (Thermo Savant, Holbrook, NY, USA) and redissolved in 12 µl 0.1% formic acid.

Mass spectrometric analyses were carried out on a 4700 Proteomics Analyzer matrix assisted laser desorption/ionization (MALDI) time-of-flight/time-of-flight (TOF/TOF) mass spectrometer (Applied Biosystems, Foster City, CA, USA). One µl of peptide solution was mixed with an equal volume of 10 mg/ml α-cyano-4-hydroxycinnamic acid (HCCA) in 50% acetonitrile containing 0.1% trifluoroacetic acid (TFA), and 1 µl was deposited on the target plate. Identification of proteins from MS and/or MS/MS MALDI data was performed using the MASCOT algorithm ( The database used was downloaded from GenBank and contained the known Apis mellifera coding sequences.

DNA, RNA isolation and cDNA synthesis

Genomic DNA was isolated using the Invisorb® Spin Plant RNA Mini Kit (Invitek GmbH, Berlin, Germany). RNA was extracted using the RNeasy Mini Kit (Qiagen, Valencia, CA, USA) following the protocol for purification of total RNA from animal cells using spin technology. For genomic DNA removal, an on-column DNase digestion with the RNase-Free DNase Set (Qiagen) was carried out according to the manufacturer's instructions. First-strand cDNA was synthesized using the RevertAid™ H Minus First Strand cDNA synthesis kit (Fermentas, St Leon-Roti, Germany). The kit uses RevertAid™ H Minus M-MuLV Reverse Transcriptase which, because of the lack of RNase H activity, does not degrade RNA in DNA-RNA hybrids during first strand cDNA synthesis.

Primer development and PCR

For amplification of transcripts from various cDNA sources, we developed primers at the extreme ends of the coding sequences with a melting temperature (Tm) of approximately 60 °C, using the formula Tm = 2(A + T) + 4(G + C). The different primer sets are listed in Table 1. For directional cloning of Am_C1q-VP in the prokaryotic expression vector, the forward coding sequence primer was preceded by four bases (CACC).

Table 1.  Primer sets used
GenePrimer sequenceAmplicon length (bp)
  1. Am, A pis mellifera; C1q-VP, C1q-like venom protein; CDS, primer set to amplify the coding sequence; cDNA_ctl, primer set to control the presence of cDNA; gDNA_ctl, primer set to control the presence of genomic DNA; Nv, Nasonia vitripennis; Y, R, N and S denote multiple nucleotides (Y, C/T; R, A/G; N, A/G/C/T; S, C/G).


In order to amplify gene fragments of potential (s)C1q-VP homologues from different hymenopteran species, degenerate primers were developed in two regions encoding highly conserved amino acid sequences (Fig. 2; boxes 1 and 2). In order to retrieve the nucleotide sequences with the lowest number of substitutions for primer development, we aligned the corresponding DNA sequences of Apis mellifera and N. vitripennis. The selected primer set is also given in Table 1.

All PCR reactions were carried out in a Eppendorf Mastercycler® (Eppendorf, Westburg, NY, USA). Following an initial step at 95 °C for 15 min, the reaction mixtures were subjected to 35 cycles consisting of denaturation at 93 °C for 1 min, primer annealing at 55 °C for 30 s, and DNA extension of 2 min 45 s at 72 °C. The reaction was completed by a final extension step at 72 °C for 10 min. When the degenerate primer set was used the number of cycles was set at 40 and the annealing temperature was lowered to 37 °C (1 min).

Sequencing and bioinformatics

Prior to the sequencing, we purified the obtained amplicon and plasmid DNA using the Illustra_ GFX_ PCR DNA and Gel Band Purification Kit (GE Healthcare, Piscataway, NJ, USA) and the GeneJEP™ Plasmid Miniprep Kit (Fermentas), respectively. DNA sequencing was performed on an Applied Biosystems 3130XL automated DNA sequencer (Perkin Elmer, Boston, MA, USA) with 50 cm capillaries filled with POP-7 polymer and using the ABI Prism BigDye V 3.1 Terminator Cycle Sequencing Kit (Applied Biosystems).

Sequence similarity was investigated using the Blast algorithm (Johnson et al., 2008) available at or as a stand-alone software package. The conserved domain architecture was predicted with CDART (Geer et al., 2002), putative N- or O-linked glycosylation sites were determined with NetNGlyc 1.0 and NetOGlyc 3.1 (Julenius et al., 2005) and multiple sequence alignment was performed with the ClustalW program (Thompson et al., 1994).

Recombinant expression of Am_C1q-VP

Directional cloning of the coding sequence of Am_C1q-VP into the pET100/D-TOPO vector (Invitrogen, Carlsbad, CA, USA) was performed in accordance with the manufacturer's instructions. This vector allows the expression of recombinant proteins with an N-terminal tag containing the Xpress epitope and a six-His tag that enables further purification. The nucleotide sequences of the cloned PCR products were verified by sequencing of both strands.

Recombinant Am_C1q-VP was expressed in E. coli BL21 Star (DE3) containing plasmid pET100/D-TOPO-Am_C1q-VP. Bacteria were grown in 100 ml Luria Bertani (LB) broth medium at 37 °C to an optical density at 600 nm of 0.5. Subsequently, isopropyl-1-thio-β-D-galactopyranoside was added to a final concentration of 1 mM, and incubation was continued for 4 h. Cells were harvested and resuspended in 8 ml lysis buffer (6 M guanidine HCl, 20 mM sodium phosphate, 500 mM NaCl, pH 7.8). After a 10 min incubation at room temperature, the cell lysate was sonicated on ice with three five-second pulses at high intensity and centrifuged. The supernatant containing recombinant Am_C1q-VP was loaded on Profinity IMAC Ni-Charged Resin (Bio-Rad) that was equilibrated with binding buffer (8 M urea, 20 mM sodium phosphate, 500 mM NaCl, pH 7.8). The column was washed with the same buffer at declining pH-values (pH 7.8, 6.0 and 5.3) and eluted in 8 M urea, 50 mM sodium phosphate, 500 mM NaCl, pH 4.0 in five steps of 1 ml each. Sample purity was determined on a 10% SDS-PAgel that was stained with Coomassie Brilliant Blue R-250. Whole bacterial lysates of stimulated and unstimulated cells were run in parallel.

Immunoscreening on spot blots

Three groups of sera were selected from the University Hospital of Antwerp serum bank for a preliminary screening of the immune recognition of recombinant Am_C1q-VP: (1) sera from three honey bee allergic patients with a positive specific IgE (sIgE) test (ImmunoCAP FEIA; Phadia AB, Uppsala, Sweden) and skin test (ALK-Abello, Horsholm, Denmark) to honey bee venom (Fig. 6); (2) sera from two patients with a documented wasp venom allergy and with a negative sIgE and skin test to honey bee venom; and (3) sera from five control individuals without venom allergy (negative sIgE test to honey bee and vespid venom).

One µg recombinant protein (Am_C1q-VP or equine uterocalin, which served as irrelevant control; Suire et al., 2001) was spotted on a nitrocellulose sheet, air-dried and washed twice with deionized water before incubating in milk diluent (Sigma, St Louis, MO, USA) for 1 h at room temperature. Subsequently, the membrane was immersed with diluted human serum (1 : 4 in milk diluent) overnight at 4 °C. Indirect staining of specifically bound human IgE was performed in three steps: first with a monoclonal antihuman IgE antibody (1 : 1000, Sigma) for 3 h at room temperature, thereafter with a peroxidase conjugated rabbit anti-mouse IgG (1 : 8000, Sigma) for 1 h and finally by incubating in the chemiluminescent substrate (SuperSignal West Dura Extended Duration, Pierce, Rockford, IL, USA) for 10 min. Blots were washed with Tris-buffered saline after each treatment. The chemiluminescent reaction was visualized and analysed with a Chemi Doc system (Bio-Rad).


D. G. E. and W. J. S. would like to thank Chris Bridts and Christel Mertens for technical assistance.

Conflicts of interest

None declared.