A Venom Serpin Splicing Isoform of the Endoparasitoid Wasp Pteromalus puparum Suppresses Host Prophenoloxidase Cascade by Forming Complexes with Host Hemolymph Proteinases*

To ensure successful parasitism, parasitoid wasps inject venom along with their eggs into their hosts. The venom serves to suppress host immune responses, including melanization. Venom from Pteromalus puparum, a pupal endoparasitoid, inhibits melanization of host hemolymph in vitro in a dose-dependent manner. Using assay-guided fractionation, a serpin splicing isoform with phenoloxidase inhibitory activity was identified as P. puparum serpin-1, venom isoform (PpS1V). This serpin gene has 16 predicted splicing isoforms that differ only in the C-terminal region. RT-PCR results show that the specific serpin isoform is differentially expressed in the venom gland. Recombinant PpS1V (rPpS1V) suppresses host prophenoloxidase (PPO) activation rather than inhibiting the phenoloxidase directly. Pulldown assays show that PpS1V forms complexes with two host hemolymph proteins, here named Pieris rapae hemolymph proteinase 8 (PrHP8) and P. rapae prophenoloxidase-activating proteinase 1 (PrPAP1), based on gene sequence blasting and phylogenetic analysis. The role of rPrPAP1 in the PPO activation cascade and its interaction with rPpS1V were confirmed. The stoichiometry of inhibition of PrPAP1 by PpS1V is 2.3. PpS1V also inhibits PPO activation in a non-natural host, Ostrinia furnacalis, through forming a complex with O. furnacalis serine protease 13 (OfSP13), an ortholog to PrPAP1. Our results identify a venom-enriched serpin isoform in P. puparum that inhibits host PPO activation, probably by forming a complex with host hemolymph proteinase PrPAP1.

mechanism for encapsulating and killing parasitoid eggs (19 -21). Possibly because of the very small sizes of parasitoids, a limited number of components from their virulence factors have been identified and characterized (22)(23)(24)(25)(26).
Pteromalus puparum is a generalist endoparasitoid wasp that parasitizes the pupal stage of several butterfly species, including the small cabbage white butterfly, Pieris rapae, an agricultural pest (27,28). Other virulence factors found in some parasitoids, such as PDVs, virus-like particles, or teratocytes, have not been recorded in P. puparum (29). As the major maternal virulence factor in P. puparum, P. puparum venom regulates host development and metabolism (30) and suppresses cellular (27,28,31) and humoral immunity of the host (32)(33)(34). Here we report a serpin gene with 16 predicted splicing isoforms in P. puparum and show that one of these isoforms is a venom protein, which inhibits host PPO activation by forming complexes with host hemolymph proteinases.
To increase the throughput of our PO activity assay, the protocol was modified as follows. The substrate (L-dopa), elicitor (Micrococcus luteus), and inhibitors (phenylthiourea, venom, and other proteins) were mixed in wells of a 384-well plate. The diluted hemolymph samples were loaded on another 384-well plate. After placing one plate on top of the other and sealing them with tape, reaction samples were mixed simultaneously by centrifugation. For samples without inhibitor, spectrophotometric monitoring at 470 nm documented a ϳ5-min lag phase before absorbance rapidly rose to a plateau (Fig. 1B). Samples with venom had longer lag phases, ϳ20 min for 1 venom reservoir equivalent (VRE), ϳ35 min for 2 VREs, and ϳ40 min for 4 VREs. Absorbances rose more slowly and to lower final absorbances in venom-exposed hemolymph samples. Absorbance at 120 min was chosen for further statistical analysis ( Fig. 1C; analysis of variance (ANOVA): df ϭ 6, F ϭ 47.245, p Ͻ 0.001). Results show that P. puparum venom inhibits host hemolymph melanization in vitro in a dose-dependent manner.
Venom Serpin Isoform Fraction and Identification-In a previous study, 70 venom proteins were identified in P. puparum (37). Protease inhibitors, serine proteases, a serine proteinase homolog, a ␤-1,3-glucan-binding protein, and several venom proteins with no similarities to other known proteins were present and might have a role in the host's melanization inhibition. To determine which of these many proteins are responsible for the melanization inhibitory activity, an assay-guided venom fractionation program was conducted.
Venom proteins from ϳ1000 venom reservoirs were separated on an anion exchange column ( Fig. 2A). In total, 182 frac-tions were collected. Of these, fractions 76 -81 showed strong inhibitory activity. The fractions with PO inhibitory activity were pooled and separated on the same column using a slower gradient (e.g. 20 -25% buffer B in 40 min; Fig. 2B), leading to a collection of 95 fractions. The peak of inhibitory activity did not correspond to the peak of absorbance, indicating that a low abundance venom component is responsible for inhibiting melanization.
Fractions 36 -45 were pooled, and proteins were separated by SDS-PAGE followed by silver staining (Fig. 2C). Two major bands (at 55 and 77 kDa) and several minor bands were present. Simultaneously, the pooled fraction was digested by trypsin and analyzed by LC-MS/MS. After searching against the P. puparum transcriptomic database, a splicing isoform of serpin (comp44322_c1_seq8; from UniGene GECT01032828.1) (37) was identified with the highest score (Fig. 2, D and E). As this isoform is identified in partially isolated venom, it was named P. puparum serpin-1, venom isoform (PpS1V). In total, 14 trypsin-digested peptides matched PpS1V, and three of these 14 were unique to this isoform. No isoform-specific tryp-FIGURE 1. Venom from P. puparum inhibits P. rapae hemolymph PO activity. A, effect of P. puparum venom on PO activity of P. rapae hemolymph from larvae and pupae. B, dose-dependent suppression of host hemolymph PO activity by P. puparum venom. Dopa chrome or dopamine chrome (melanization intermediates) was monitored at A 470 every 5 min for 2 h. C, summary of absorbance A 470 at 120 min from B. All values are presented as the mean (n ϭ 3). Error bars represent S.E. Asterisks indicate significant differences from PBS as the control. **, p Ͻ 0.01; ***, p Ͻ 0.001.
sin-digested peptides matched other splicing isoforms. The best blastp match for PpS1V in the NCBI Protein Database is Nasonia vitripennis ovalbumin-related protein X isoform X12 (XP_001606111), which is a splicing isoform of gene LOC100122505. Ovalbumin is the main protein of egg white and belongs to the serpin superfamily (38). N. vitripennis gene LOC100122505 has 18 splicing isoforms in the NCBI Protein Database. All of these 18 isoforms have a serpin domain (cd00172). . Partial separation and identification of PpS1V. A, chromatograph of total P. puparum venom by Bio SAX column. Venom protein (ϳ1000 RVEs) was eluted with a gradient of buffer B from 0 to 100% at a flow rate of 0.5 ml/min. B, chromatograph of pooled fractions 76 -81 from A on the same column with a shallow gradient of buffer B from 20 to 25% in 40 min. Fractions 36 -45 were denoted as phenoloxidase inhibitor (POI) and are indicated by the red bar. These fractions were pooled together for further SDS-PAGE and LC-MS/MS analysis. C, SDS-PAGE analysis of pooled fraction POI followed by silver staining. D, amino acid sequence of PpS1V with peptides identified by MS (highlighted residues). Trypsin cleavage sites are demarcated in blue. The first isoform-specific tryptic peptide of PpS1V is colored in red. E, MS/MS map marked with b ions (blue) and y ions (red) for the first isoform-specific tryptic peptide of PpS1V. F, SDS-PAGE analysis of pooled fraction POI followed by immunoblotting using antibody (Ab) against PpS1V.
Sequence Analysis-The PpS1V transcript (comp44322_c1_ seq8) is 1806 bp long with a 1197-bp open reading frame (ORF), a 220-bp 5Ј-non-coding region, and a 389-bp 3Ј-non-coding region containing a poly(A) tail. The ORF encodes a protein of 399 amino acids with a predicted signal peptide consisting of the first 19 residues. The calculated molecular mass of the mature protein without the signal peptide is 42.1 kDa, and the calculated isoelectric point is 4.8. To determine whether this isoform was caused by arbitrary assembly, the sequence was confirmed by PCR using cDNA from female adults and specific primers spanning the whole ORF (forward primer, 5Ј-GCG-TTAGCGTCTGGAACTCA-3Ј; reverse primer, 5Ј-AAACA-GATTGAGTTTGCGGA-3Ј). The results confirmed that this isoform is a true transcript in wasp female adults.
To investigate the gene structure of PpS1V, its genomic sequence was retrieved from the assembly of P. puparum genome, 3 and the gene was named P. puparum serpin-1 (PpSerpin-1, GenBank TM accession number KX268468). Using available RNA sequencing data of P. puparum and 18 isoforms from N. vitripennis gene LOC100122505, 16 putative splicing isoforms were predicted for PpSerpin-1 and alphabetically named isoforms A-P; isoform O was the identified PpS1V in the isolated venom fractions (Fig. 3, A and B). These 16 isoforms share the first seven exons and differ in exon 8. This variable exon 8 includes the RCL, which determines the inhibitor selectivity of the serpin (Fig. 3, A and B). Although these 16 exon variants are different from each other in exon 8, the amino acid alignment showed that four positions were fully identical, and six more were identified as conservations in the variable region (Fig. 3C). These conserved positions may be important to the inhibitory function of serpin.
To reveal the possible evolutionary history of gene PpSerpin-1, phylogenetic analysis was performed using the amino acid sequence of PpS1V excluding the variable region coded by exon 8 (Fig. 4). If there were multiple splicing isoforms in serpins from other insects, the first isoforms were chosen for analysis. Our analysis shows that PpS1V is closely related to N. vitripennis antichymotrypsin-2 isoform X1 (XP_008201829.1), which is the first isoform of N. vitripennis gene LOC100122505. Alternative splicing was found in serpins from other parasitoid wasps, such as Copidosoma floridanum, Trichogramma pretiosum, and Orussus abietinus, and non-parasitoid hymenopterans, such as Apis mellifera, Athalia rosae, and Ceratosolen solmsi marchali. Splicing isoforms were also identified in serpins from several non-venom dipterans, coleopterans, lepidopterans, and hemipterans (Fig. 4).
Isoform-specific RT-PCR-To investigate the distribution of PpS1V expression, specific RT-PCR was performed using isolated heads, thoraces, ovaries, venom glands, and carcasses (abdomen without ovary and venom apparatus) from female adults (Fig. 5). Result showed that PpS1V is highly differentially expressed in venom gland, although weak bands were also present in thorax and ovary. This indicates that PpS1V is more likely to be a real venom isoform rather than leakage during sample processing. However, because of retention of the intron between exons 8N and 8O in transcript of isoform N, it is not possible to design specific primers for real time quantitative PCR of isoform O (e.g. PpS1V) (Fig. 5).
Effect of rPpS1V on Host Melanization-As serpins are primary regulators in PPO activation and PpS1V was identified with the highest score in the partially isolated venom fractions, we hypothesized that PpS1V could be responsible for the host's PPO inhibition in P. puparum venom. To test the function of PpS1V, we produced recombinant proteins and examined their effects on the host's PPO activation pathway.
The coding sequence of mature PpS1V was cloned into pFast-HTB and pET-28a vectors. rPpS1V was successfully expressed in both Escherichia coli and the baculovirus system (Fig.  6A). The protein expressed in E. coli showed an observed mass of 48 kDa, which was close to the predicted size of 45.7 kDa. The protein expressed in Sf9 cells showed an observed mass of 45 kDa, which was close to the predicted size of 44.3 kDa. The predicted size of rPpS1V expressed in E. coli was bigger than that in the baculovirus system. This was caused by a longer introduced linker to the His tag at the N terminus in pET-28a vector.
When proteins were incubated with host hemolymph before adding elicitor, rPpS1V (0.5 g) from both E. coli and Sf9 cells showed activity to significantly suppress the melanization of host hemolymph ( Fig. 6B; ANOVA: df ϭ 7, F ϭ 125.63, p Ͻ 0.001). rPpS1V purified from E. coli suppresses the melanization of host hemolymph in a dose-dependent manner (Fig. 6C). When proteins were incubated with preactivated host hemolymph, only the PO inhibitor phenylthiourea (PTU) showed significant suppression of PO activity ( Fig. 6D; ANOVA: df ϭ 7, F ϭ 125.63, p Ͻ 0.001). There were no significant differences among other treatments without PTU ( Fig. 6D; ANOVA: df ϭ 6, F ϭ 2.01, p ϭ 0.14; PTU was excluded in this ANOVA). These results show that PpS1V and P. puparum venom inhibit the host hemolymph melanization by suppressing the PPO activation pathway rather than by directly inhibiting activated PO. Identification of Complexes with Host Hemolymph Proteinases-Serpin inhibits serine proteases by forming covalent complexes with the target protease (13). To identify potential target proteases of PpS1V in host hemolymph, pulldown assays were conducted followed by LC-MS/MS identification.
When PpS1V was incubated with preactivated hemolymph, there were two nonspecific bands of 37 and 78 kDa detected by fluorescence staining and three specific bands recognized by the His tag antibody (Fig. 7A). Two of these three specific bands were present with minor size differences at ϳ78 kDa, and the third band was below intact PpS1V with an apparent mass of 41 kDa. After PpS1V was incubated with naïve hemolymph 30 min before adding elicitor, one more specific band (64 kDa) was detected by fluorescence staining and recognized by the His tag antibody (Fig. 7A).
We hypothesized that PpS1V inhibits PPO activation of P. rapae hemolymph by forming a complex with PrPAP1. The complex was detected in the PpS1V pulldown sample (Fig. 7, E and F) but was not present in the control sample.
Interaction between rPpS1V and rPrPAP1-To confirm the function of PrPAP1 in the PPO activation cascade and its interaction with PpS1V, rPrPAP1 Xa was expressed in the baculovirus system. Activated PAPs, but not pro-PAPs, hydrolyze a colorimetric peptide substrate, acetyl-Ile-Glu-Ala-Arg-p-nitroanilide (IEARpNA) (39 -41). The culture medium containing rPrPAP1 Xa showed a high IEARase activity (data not shown), suggesting that rPrPAP1 Xa had already been activated for an unknown reason. Compared with culture medium harvested from eGFP-expressing recombinant baculoviruses, culture medium containing activated rPrPAP1 Xa strongly enhanced the melanization of P. rapae hemolymph with or without M. luteus ( Fig. 8 1 g) (Fig. 8). A characteristic feature of serpin-proteinase interactions is the formation of a covalent complex of a serpin with its target proteinase (44,47). Immunoblotting band intensities of serially diluted rPpS1V and rPrPAP1 Xa indicate that the concentration of rPrPAP1 Xa in culture medium was ϳ5 ng/l. After incubating rPpS1V with culture medium containing activated rPrPAP1 Xa , complexes above 70 kDa were detected by Western blotting using antibodies against His tag, PrPAP1, and PpS1V (Fig. 9, A-C), indicating that PpS1V forms a covalent complex with PrPAP1.
To further investigate the inhibition of PrPAP1 by PpS1V, we tested the hydrolysis inhibition of a colorimetric peptide substrate, IEARpNA, by PrPAP1. PrPAP1 activity decreased linearly as the PpS1V concentration increased (Fig. 10). The stoichiometry of inhibition is 2.3, indicating that PpS1V is an efficient inhibitor of PrPAP1.
PpS1V inhibits O. furnacalis PPO Activation via a Complex with OfSP13, an Ortholog to PrPAP1-We also tested the effect of PpS1V on the hemolymph melanization of a non-natural host, the Asian corn borer, O. furnacalis. rPpS1V showed strong inhibition of the melanization of O. furnacalis hemolymph ( Fig. 11A; ANOVA: df ϭ 2, F ϭ 133.818, p Ͻ 0.001) in a dose-dependent manner (Fig. 11B).
OfSP13 is the best reciprocal hit to PrPAP1 when reciprocal blastn was performed between O. furnacalis (50) and P. rapae transcriptome (33). OfSP13 has already been demonstrated to cleave PPO into PO and play an important role in PPO activation (44,51). To test whether PpS1V forms a complex with OfSP13, we mixed rPpS1V with factor Xa-activated OfSP13 Xa and recorded a higher molecular weight complex by immunoblotting (Fig. 11C). Activity of activated OfSP13 Xa decreased linearly as PpS1V concentration increased (Fig. 11D). The stoichiometry of inhibition is 2.2 (Fig. 11D), similar to that of PrPAP1 by PpS1V. These results demonstrate that PpS1V inhibits O. furnacalis hemolymph melanization through forming a complex with OfSP13, an ortholog to PrPAP1.

Discussion
Inhibiting host melanization is an immunosuppression strategy in parasitoid wasps. However, quite diverse proteins could be recruited in different parasitoid systems for this convergent FIGURE 8. Effect of PrPAP1 on melanization of P. rapae hemolymph. Ten microliters of diluted P. rapae hemolymph was mixed with M. luteus (0.5 g), purified PpS1V, or 10 l of culture medium containing activated PrPAP1 (cultured Sf-900 II medium harvested from eGFP-expressing recombinant baculoviruses was used as control). After incubation at room temperature for 5-15 min, PO activity was assayed as described under "Experimental Procedures." All values are presented as the mean (n ϭ 3). Error bars represent S.E. Asterisks indicate significant differences from the control. **, p Ͻ 0.01; ***, p Ͻ 0.001.  function. Virulence proteins include serine proteinase homolog (24,49) and unrelated peptide Vn 4.6 (22) from Cotesia rubecula venom, extracellular superoxide dismutase (25) and serpin (23) from Leptopilina boulardi venom, and epidermal growth factor-like proteins (26,48,52) from Microplitis demolitor PDV. With an estimated 600,000 species, parasitoid wasps are one of the most abundant and diverse insect groups on earth (53). Such high diversity may present an untapped source of bioactive compounds with potential in pest control and drug discovery.
By assay-guided fractionation, a splicing isoform of serpin, PpS1V, was identified from P. puparum venom that inhibits host melanization. This serpin isoform was also detected in previous venom proteomic research of P. puparum (37). However, the previous study only used gene loci to calculate the expression level. Splicing isoforms were not considered, mainly because of the limitation of isoform expression estimation without a reference genome. Thus, PpS1V was not included in the reported 70-venom protein set.
In insects, serpin genes have evolved alternative splicing from a single gene to produce multiple functional serpins, which differ in RCL. Alternate splice forms were first reported in M. sexta serpin-1 with 12 isoforms (54,55) and identified in other insects such as B. mori (56), Choristoneura fumiferana (57), and Anopheles gambiae (58). Based on our phylogenetic analysis, alternative splicing seems to be common in insect serpins. It is also reasonable to propose that the venom isoform of PpSerpin-1 evolved after the invention of the venom apparatus in Hymenoptera.
For the PpSerpin-1 gene, 16 splicing isoforms were predicted by bioinformatics approaches. Except for the venom isoform PpS1V, the tissue distributions and functions of the other 15 splicing isoforms remain unknown.
It is thought that venom proteins are mainly recruited through gene duplication from non-venom proteins followed by natural selections for adaptive traits (59). Several cases of gene duplications have been described in parasitoid venoms, including Ci-48a-like proteins and reprolysin-like proteins from M. demolitor (60), RhoGAP domain-containing proteins from Leptopilina (61), and ␥-glutamyl transpeptidases from Aphidius ervi (62). However, gene duplication is not the only way venom has become diverse. The origins of venom proteins also include co-option of single copy genes (63), de novo synthesis of novel venom proteins from non-coding DNA, alternative splicing (64,65), and lateral gene transfer from microorganisms (66). P. puparum serpin-1 provides an example of venom recruitment through alternative splicing. Another example is the sarco/endoplasmic reticulum calcium ATPase from Ganaspis sp. 1 that has a venom isoform and a non-venom isoform (67). Pulldown results indicate that possible targets of PpS1V are hemolymph proteinase PrHP8 and PrPAP1. This is similar to serpin-1J in M. sexta (40,47,54,55,68). Manduca serpin-1J regulates PPO activation through inhibition of M. sexta PAP3 (MsPAP3) (40,54) and the Toll pathway through inhibition of MsHP8 (47). PrPAP1 is clustered with proteinases involved in the PPO activation cascade. The role of PrPAP1 in P. rapae melanization and its interaction with PpS1V were also confirmed. In contrast, PrHP8 is clustered with proteinases involved in Toll pathway activation. Thus, PpS1V may also have a role in regulation of the Toll pathway.
In summary, we identified a splicing isoform of serpin in P. puparum venom. This serpin isoform shows activity to inhibit the host's PPO activation through forming a complex with PrPAP1. This study provides a case of parasitoid venom evolution through alternative splicing and sheds light on the mechanisms by which parasitoid venom suppresses host immunity.

Experimental Procedures
Insect Rearing-Laboratory cultures of P. puparum and its host P. rapae were maintained at 25°C with a photoperiod of 14:10 h (light:dark) as described previously (27,28). Once emerged, the wasps were collected and held in glass containers without hosts and fed ad libitum on 20% (v/v) honey solution to lengthen life span. P. rapae was reared in a greenhouse on cabbages grown throughout the year. Larvae of O. furnacalis were maintained on an artificial diet at 28°C under a relative humidity of 70 -90% and a photoperiod of 16:8 h (light:dark) as described previously (44,51).
Venom Protein Extraction-Mated female wasps aged 3-7 days after eclosion were anesthetized on ice and then dissected in PBS on an ice plate under a stereoscope (Olympus). The venom reservoirs were washed several times using PBS and then transferred to an Eppendorf tube on ice. After centrifugation at 16,000 ϫ g and 4°C for 10 min, the supernatant was filtered with a 0.22-m Millipore (Billerica, MA) filter and stored at Ϫ80°C until use.
Hemolymph Collection-The plasma was collected from naïve P. rapae pupae or larvae on ice and diluted four times in anticoagulant (4 mM sodium chloride, 40 mM potassium chloride, 8 mM EDTA, 9.5 mM citric acid, 27 mM sodium citrate, 5% sucrose, 0.1% polyvinylpyrrolidone, and 1.7 mM PIPES, pH 6.8). After removing hemocytes by centrifugation at 3000 ϫ g and 4°C for 10 min, hemolymph samples were collected and stored at Ϫ80°C for further assays.
Phenoloxidase Activity Assay-For the PO activity assay of P. rapae hemolymph, 15 l of diluted hemolymph was mixed with 5 l of inhibitor (venom protein or saturated PTU) and 5 l of elicitor (0.1 g/l M. luteus). For the PO activity assay of O. furnacalis hemolymph, 2 l of larval hemolymph was mixed with 8 l of recombinant proteins. Then samples were placed on a rotary mixer at 25°C for around 10 min. After adding 800 l of substrate solution (20 mM dopa in PBS, pH 6.5), samples (200 l) were measured at A 470 in 96-well plates for 20 min (Varioskan Flash multimode reader, Thermo Scientific). One unit of PO activity was defined as 0.001 ⌬A 470 /min. To increase the throughput of the PO assay, the assay was further modified and conducted in 384-well plates. Ten microliters of diluted P. rapae hemolymph was mixed with 5 l of proteins, 5 l of elicitor (0.1 g/l M. luteus), and 5 l of substrate solution (50 mM L-dopa in PBS, pH 7.5). The plates were measured at A 470 at 25°C every 5 min for 2 h using a Varioskan Flash multimode reader. Statistical analyses were performed using Data Processing System (DPS) v13.5 (69). Comparisons of two samples were conducted using Student's t test. Comparisons of multiple samples were performed using ANOVA followed by Tukey's multiple comparison test.
Fractionation of Venom Proteins-Venom protein was diluted with buffer A (25 mM Tris-HCl, pH 7.5) to 2 ml and loaded onto a Bio SAX column (5 m, 4.6 ϫ 50 mm; Agilent Technologies, Wilmington, DE) using a Biologic Duo-Flow high performance chromatography system (Bio-Rad). Proteins were eluted at a flow rate of 0.5 ml/min with a gradient of buffer B (25 mM Tris-HCl, 1 M NaCl, pH 7.5) against buffer A. Proteins were monitored by absorbance at 280 nm. Fractions of 200 l were collected in a deep well plate (Eppendorf) using a BioLogic BioFrac fraction collector (Bio-Rad) and then desalted using a Zeba spin desalting plate (Thermo Scientific) according to the manufacturers' instructions. The desalted fractions were stored at Ϫ80°C until use.
Protein Identification of Isolated Venom Protein-To identify potential candidates with PO inhibitory activity, a pooled fraction of 200 l was digested by trypsin using the filter-aided sample preparation method (70). Then mass spectrometric analysis was performed using an Easy nLC HPLC system (Thermo Scientific) followed by Q-Exactive (Thermo Finnigan). In this study, samples were first loaded on a Thermo Scientific EASY column (5 m, 2 cm ϫ 100 m, C 18 ) and then separated on another Thermo Scientific EASY column (3 m, 75 m ϫ 100 mm, C 18 ) with a flow rate of 250 nl/min. Buffer A was water with 0.1% formic acid, buffer B was 84% acetonitrile with 0.1% formic acid, and the gradient was from 0 -50% buffer B in 50 min and then 50 -100% buffer B in 4 min. The chargeto-mass ratios of peptides and fractions of peptides were collected 10 times after every full scan. The resulting MS/MS spectra were searched against the translated P. puparum transcriptomic database (37) using Mascot software (71). The maximum number of missed cleavages was set as 2. Carbamidomethyl of cysteine and oxidation of methionine were set as fixed and variable modifications, respectively. Peptide confidence Յ0.01 was used to filter the peptide identification. This part of the work was done by Shanghai Applied Protein Technology Co., Ltd. (Shanghai, China).
Specific RT-PCR-Female wasps were dissected in Ringer's saline (182 mM KCl, 46 mM NaCl, 3 mM CaCl 2 , 10 mM Tris-HCl) with 1 unit/l RNase inhibitor (TOYOBO, Osaka, Japan) on an ice plate under a stereoscope (Olympus). The total RNA was extracted from different tissues of female wasps using TRIzol reagent according to the manufacturer's protocol and then was reverse transcribed using TransScript One-step gDNA Removal and cDNA Synthesis SuperMix (TransGen, China) with random primers. The isoform-specific primers were designed to span exon 7 and exon 8 using PerlPrimer V1.1.21 (72) and are listed in Table 1. The isoform-specific sequence of PpS1V was amplified by PCR using TransTaq HiFi DNA polymerase and confirmed by sequencing.
Sequence Analysis-Signal peptides were predicted using SignalP 4.1 (73). Gene structure was predicted by Splign (74) using transcript variants of N. vitripennis gene LOC100122505 and transcripts from the P. puparum transcriptome. Multiple amino acid sequence alignments were performed using MUSCLE v3.8 (75). Regions with low similarities were manually removed. Phylogenetic analysis was conducted by PhyML version 20131022 using default settings (76). For phylogenetic tree construction of PpS1V, the variable region at the C terminus was excluded. The structure of PpS1V was modeled by SWISS-MODEL (77,78) using Protein Data Bank code 2H4R as the template and visualized using PyMOL v1.7.0.0.
Recombinant Protein Expression and Purification-Recombinant plasmids were generated using the ClonExpress One Step Cloning kit (Vazyme, China). To perform recombination cloning, both insert DNA fragments and linear pFast-Bac-HTB vector were amplified by PCR. The primers used are listed in Table 1. Linear pET-28a vector was generated by digestion with BamHI and XhoI (TaKaRa, Dalian, China), and linear pFast-Bac1 vector was generated by digestion with BamHI and KpnI (TaKaRa). In PrPAP1, the predicted activation site RSDR 122 was mutated into IEGR 122 using the Mut Express II Fast Mutagenesis kit (Vazyme, China) and plasmid pFast-Bac1-PrPAP1 as the template. IEGR is a preferred cleavage site for commercially available factor Xa protease, which theoretically cleaves PrPAP1 Xa at IEAR 122 and produces active PrPAP1. For protein expression in E. coli, recombinant pET-28a plasmids were transferred into BL21(DE3) and confirmed by sequencing. After growth in autoinduction medium containing 100 g/l kanamycin at 20°C for 48 h, E. coli cells were harvested by centrifugation at 10,000 ϫ g and 4°C for 10 min. For recombinant protein expression using the baculovirus system, recombinant pFast-Bac plasmids were used to generate recombinant baculoviruses according to the manufacturer's instructions. Sf9 cells (2 ϫ 10 6 cells/ml) in 100 ml of Sf-900 II serum-free medium (Invitrogen) were infected with the recombinant baculovirus and incubated at 28°C with shaking at 90 rpm in 250-ml flasks. The Sf9 cells were harvested 72 h after infection by centrifugation at 1000 ϫ g for 20 min at 4°C. For secreted protein expression, culture medium containing PrPAP1 Xa was harvested 96 h after infection. The recombinant proteins in both E. coli and Sf9 cells were purified using the His-Bind Purification kit (Novagen) according to the manufacturer's instruction. The concentration of protein was determined using method of Bradford (79).
Pulldown Assay-One milliliter of diluted P. rapae hemolymph containing 50 l of saturated PTU was incubated with 100 l of M. luteus (1 g/l) and 5 l of recombinant protein (2 g/l) at 4°C overnight. After centrifugation at 12,000 ϫ g at 4°C for 20 min, the supernatant was incubated with 25 l of cOmplete His tag purification resins (Roche Applied Science) at 4°C for 2 h. The hemolymph with resins was then loaded onto spin columns and spun at 1000 ϫ g for 2 min followed by reloading and spinning several times until all the sample had been loaded. The resins were washed three times with 300 l of washing buffer (1 M NaCl, 120 mM imidazole, 40 mM Tris-HCl, pH 7.9) before eluting with 50 l of elution buffer (1 M imidazole, 0.5 M NaCl, 20 mM Tris-HCl, pH 7.9). Eluted proteins were then analyzed by SDS-PAGE followed by Lumitein (Biotium, Hayward, CA) protein gel staining and immunoblotting.
Protein Identification in Pulldown Complexes-To identify potential target proteins in pulldown complexes, the excised gel slices were in gel-digested by trypsin and lyophilized separately followed by mass spectrometry on a 1DLC-LTQ-Velos instrument (Thermo Finnigan, San Jose, CA) as described previously (37). Briefly, samples were desalted on Zorbax 300 SB-C 18 (Agilent Technologies) and then separated on an RP-C 18 column (150-m inner diameter, 150-mm length) (Column Technology Inc., Fremont, CA). Buffer A, buffer B, and the gradient were the same as mentioned above. The charge-to-mass ratios of peptides and fractions of peptides were collected 20 times after every full scan. The resulting MS/MS spectra were searched using BioworksBrowser 3.3 (Thermo Electron, Bremen, Germany) against the translated P. rapae transcriptome with the manually added bait sequence (e.g. PpS1V). Carbamidomethylation of cysteine and oxidation of methionine were set as fixed and variable modifications, respectively. The number of maximum missed cleavages was set to 2. ⌬CN (Ն0.1) and cross-correlation scores (Xcorr; one charge Ն1.9, two charges Ն2.2, and three charges Ն3.75) were used to filter the peptide identification. The criterion unique identified peptides Ն2 was used to filter the protein identification. This part was done by Shanghai Applied Protein Technology Co., Ltd.
Author Contributions-Z. Y. conducted most of the experiments, analyzed the results, and wrote most of the paper. Q. F. and Y. L. conducted experiments on gene cloning and specific RT-PCR. S. X. and L. Y. conducted experiments for recombinant protein expression and purification. F. W. performed sequence analysis. G. Y. and C. A. conceived and designed the research. J. H. W. and C. A. provided input on data interpretation. G. Y. and J. H. W. wrote the paper with Z. Y.