Manduca sexta Serpin-6 Regulates Immune Serine Proteinases PAP-3 and HP8

Analogous to blood coagulation and complement activation in mammals, some insect defense responses (e.g. prophenoloxidase (proPO) activation and Toll pathway initiation) are mediated by serine proteinase cascades and regulated by serpins in hemolymph. We recently isolated Manduca sexta serpin-6 from hemolymph of the bacteria-challenged larvae, which selectively inhibited proPO-activating proteinase-3 (PAP-3) (Wang, Y., and Jiang, H. (2004) Insect Biochem. Mol. Biol. 34, 387–395). To further characterize its structure and function, we cloned serpin-6 from an induced fat body cDNA library using a PCR-derived probe. M. sexta serpin-6 is 55% similar in amino acid sequence to Drosophila melanogaster serpin-5, an immune-responsive protein. We produced serpin-6 in an Escherichia coli expression system and purified the soluble protein by nickel affinity and hydrophobic interaction chromatography. The recombinant protein specifically inhibited PAP-3 and blocked proPO activation in vitro in a concentration-dependent manner. Matrix-assisted laser desorption ionization-time of flight mass spectrometry indicated that the cleavage site of serpin-6 is between Arg373 and Ser374. Serpin-6 is constitutively present in hemolymph of naïve larvae, and its mRNA and protein levels significantly increase after a bacterial injection. The association rate constant of serpin-6 and PAP-3 is 2.6 × 104 m-1 s-1, indicating that serpin-6 may contribute to the inhibitory regulation of PAP-3 in the hemolymph. We also identified the covalent complex of serpin-6 and PAP-3 in induced hemolymph by immunoaffinity chromatography and mass spectrometry. Furthermore, immulectin-2, serine proteinase homologs, proPO, PO, attacin-2, and a complex of serpin-6 and hemolymph proteinase-8 were also detected in the proteins eluted from the immunoaffinity column using serpin-6 antibody. These results suggest that serpin-6 plays important roles in the regulation of immune proteinases in the hemolymph.

Phenoloxidase (PO) 1 participates in several insect physiological processes, including melanogenesis, cuticle sclerotization, wound healing, and other defense responses (2,3). It catalyzes the formation of quinones that are precursors of melanin. Melanin and proteins may cross-link to form a capsule around invading parasites. Additionally, quinones might have antimicrobial effects (4). Proteolytic activation of prophenoloxidase (proPO) in insects is mediated by a largely unknown serine proteinase pathway triggered by microbial surface molecules, such as lipopolysaccharide, peptidoglycan, and ␤-1,3-glucan (3,5). ProPO-activating proteinase (PAP), also known as proPOactivating enzyme, cleaves proPO and causes its activation.
So far, we have isolated three PAPs from the tobacco hornworm, Manduca sexta (6 -8). They all cleave proPO at Arg 51 but require an auxiliary factor to generate active PO. We have purified and characterized the "cofactor" as a high M r complex of serine proteinase homolog-1 and -2 (SPH-1 and -2) but do not yet understand its mechanism (9,10). A similar phenomenon was reported in the beetle Holotrichia diomphalia (11). In contrast, Bombyx mori proPO-activating enzyme does not appear to need any cofactor for proPO activation (12).
Here, we report the cDNA cloning of M. sexta serpin-6 and its recombinant expression in Escherichia coli. Biochemical analysis demonstrated that recombinant serpin-6 inhibited PAP-3 efficiently. The mRNA and protein levels of serpin-6 were upregulated in hemocytes and fat body after a bacterial challenge. Moreover, we characterized the hemolymph proteins bound to serpin-6.

EXPERIMENTAL PROCEDURES
Insects and Collection of Hemolymph, Hemocytes, and Fat Body-M. sexta eggs were purchased from Carolina Biological Supply, and the larvae were reared as described previously (34). Day 2 fifth instar larvae were injected with 1 ϫ 10 7 E. coli cells, or 0.1 mg Micrococcus luteus suspended in 50 l of phosphate-buffered saline. Hemolymph and fat body samples were collected at 12 and 24 h after the microbial challenge.
Isolation of Full-length Serpin-6 cDNA by Library Screening-Once confirmed by BLAST search to be serpin cDNA fragments, the inserts in the PCR-derived clones were labeled with [␣-32 P]dCTP (Multiprime DNA labeling system, Amersham Biosciences) and utilized as probes to screen the cDNA library in ZAP2 (Stratagene). Positive plaques were purified to homogeneity and subcloned by in vivo excision of pBluescript phagemids. Complete nucleotide sequences were determined as described above. Sequence comparison was performed using ClustalW 1.7 program from MacVector 6.5 software (Genetics Computer Group, 1998).
Expression and Purification of Soluble Serpin-6 from E. coli Cells-To amplify a cDNA fragment encoding the mature serpin-6, a PCR reaction was performed using forward primer (j815: 5Ј-TAA CCA TGG AAT GTT TCT CC-3Ј) and reverse primer (j816: 5Ј-TAT GCA TGC TTA TTT CTT AGG GT-3Ј). The PCR product was digested with NcoI-SphI and cloned into the same sites in plasmid H6pQE60 (35). The resulting recombinant plasmid, serpin-6/H6pQE60, was sequenced to confirm correct insertion and sequence. The expression, extraction, and affinity purification of serpin-6 were performed under nondenaturing conditions as previously described (29,35). The recombinant protein eluted from nickel-nitrilotriacetic acid-agarose (Qiagen) was adjusted with ammonium sulfate to a final concentration of 1.0 M. After centrifugation at 15,000 ϫ g for 30 min, the cleared protein solution was separated by hydrophobic interaction chroma-tography on a 5-ml phenyl-Sepharose column (Amersham Biosciences) and eluted at 1.0 ml/min for 40 min with a linear, descending gradient of 1.0 -0 M ammonium sulfate in 10 mM potassium phosphate, pH 6.4. Concentration and buffer exchange of the serpin-6 fractions were performed in a Centriprep-30 (Millipore). The purified protein was stored at Ϫ70°C in 20 mM Tris-HCl, pH 7.5. A rabbit polyclonal antiserum was prepared against the recombinant serpin (Cocalico Biologicals Inc.).
Inhibition of ProPO Activation by PAP-3 and SPHs-ProPO was purified from M. sexta larval hemolymph as described before (36). Purified PAP-3 (100 ng/l, 1 l), SPHs (50 ng/l, 1 l), serpin-6 (150 ng/l, 10 l), and buffer (7.4 l, 20 mM Tris-HCl, pH 8.0) were preincubated at room temperature for 10 min. In a control, serpin-6 was replaced with the buffer (10 l). ProPO (10 l, 10 g/ml) was separately added to the wells, and the reaction mixtures were then placed on ice for 1 h. A mixture of proPO (10 l) and buffer (20 l) was used as another control. PO activity was determined using dopamine as a substrate (7).
Inhibition of ProPO Activation by Serpin-6 in a Fraction of the Larval Hemolymph-The induced hemolymph was fractionated with 0 -50% saturation of ammonium sulfate to obtain the proPO activation system (29). Recombinant serpin-6 (10 l, at final concentrations of 0 -150 g/ml) was incubated with the plasma fraction (10 l, 1:5 diluted in H 2 O) in the presence of M. luteus (1 l, 1 g/l). As controls, 10 l of buffer or the fraction without M. luteus were added individually and incubated at room temperature for 10 min prior to the PO activity assay.
Kinetic Inhibition Assay-A progressive curve method was employed to determine the association rate constant (37). Purified PAP-3 (1.8 M) was mixed with IEARpNA solutions containing serpin-6 at different molar ratios. Absorbance at 405 nm was monitored in the kinetic mode on a tunable VERSAmax microplate reader (Molecular Devices). The progressive curves were first analyzed according to Equation , in naïve and induced larval hemolymph, the plasma concentrations of serpin-6 were estimated by SDS-PAGE, immunoblot analysis, and densitometry using the purified recombinant serpin-6 as a standard (35).
Detection of Serpin-Enzyme Complexes by Immunoblot Analysis-PAP-3 (1 l, 100 g/ml) and recombinant serpin-6 (2 l, 100 g/ml) were incubated at room temperature for 10 min and then treated with SDS reducing sample buffer at 95°C for 5 min. SDS-PAGE and immunoblot analysis were performed as described previously (1).
Determination of Cleavage Site in Serpin-6 -PAP-3 (12 l, 100 g/ml) was mixed with serpin-6 (12 l, 280 g/ml) for 10 min at room temperature. The reaction mixture was subjected to MALDI-TOF mass spectrometry (8). The molecular mass of a peak that was absent in the control spectra of serpin-6 and PAP-3 was compared with calculated values of carboxyl-terminal peptides to deduce the cleavage site in serpin-6.
RNA Extraction and Reverse Transcription-PCR Analysis-Total RNA samples were extracted from fat body and hemocytes of naïve and induced M. sexta larvae using Micro-to-Midi total RNA purification system (Invitrogen). First-strand cDNA synthesis was performed using 2-4 g total RNA, 10 pmol oligo(dT) 17 , and 200 units of Moloney murine leukemia virus reverse transcriptase (Invitrogen) at 37°C for 1 h. M. sexta ribosomal protein S3 cDNA was used as an internal standard to normalize the templates in a preliminary PCR experiment. After template adjustment, PCRs were performed to detect relative levels of serpin-6 cDNA. The primers used were j812 (5Ј-TGA TGA CTG CGT ATA AGG TA-3Ј) and j820 (5Ј-CTT CTC GAC TCA TGT AGC TCT CT-3Ј) for serpin-6 and k504 (5Ј-CGC GAG TTG ACT TCG GT-3Ј) and k501 (5Ј GCC GTT CTT GCC CTG TT-3Ј) for ribosomal protein S3. The thermal cycling conditions were: 94°C, 30 s; 50°C, 30 s; 72°C, 60 s. PCR cycle numbers were empirically chosen to show comparable band intensity and avoid saturation. After separation by 1.3% agarose gel electrophoresis, intensities of the PCR products were quantified using Kodak Digital Science one-dimensional gel analysis software (35).
Immunoaffinity Chromatography and MALDI-TOF Mass Fingerprint Analysis-To couple serpin-6 antibodies to protein A-Sepharose, 4.8 ml of the rabbit antiserum, 2.4 ml of resin (Sigma), and 19.2 ml of phosphate-buffered saline were incubated at room temperature for 1 h with gentle shaking. The resin was washed with 10 volumes of 0.2 M sodium borate, pH 9.0, and reacted with dimethylpimelimidate at a final concentration to 20 mM. After 2 h, 0.2 M ethanolamine, pH 8.0, was added to terminate the coupling reaction for another 2 h. For isolating serpin-6-proteinase complexes, 15 ml of cell-free hemolymph was collected from naïve or bacteria-injected larvae (fifth instar, day 3). M. luteus, diethylthiocarbonate, and 1-phenyl-2-thiourea were added to the plasma at final concentrations of 1 g/l, 10 mM, and 1 mM, respectively, to activate the serine proteinase system yet prevent hemolymph melanization. After incubation at room temperature for 30 min, 1 mM phenylmethanesulfonyl fluoride and 0.5 ml proteinase inhibitor mixture (Sigma catalog number P-8849) were added to block the proteinases in the activated hemolymph for 10 min. The plasma samples were centrifuged at 5000 ϫ g for 15 min at 4°C to separate debris from the supernatant. After binding with the affinity matrix (1.0 ml) for 8 h at 4°C with gentle agitation, the unbound proteins were removed by loading the suspension into an empty Poly-Prep column (Bio-Rad). The resin was sequentially washed with 20 ml, 1 M NaCl, and 20 ml, pH 6.8, 10 mM sodium phosphate to reduce nonspecific binding. Bound proteins were eluted with 50 mM glycine HCl, pH 2.5, and the fractions (0.5 ml) were neutralized with 50 l, 1.0 M sodium phosphate, pH 8.0. The affinity-purified proteins were subjected to SDS-PAGE, immunoblot, and peptide mass fingerprint analyses (10). The mass profiles were analyzed using Mascot (matrixscience.com/) to identify significant matches in GenBank. The monoisotopic peptide masses were also compared with the theoretical masses of M. sexta hemolymph proteinases, 2 predicted by MS-digest (prospector.ucsf.edu/).

Molecular Cloning and Structural
Features of M. sexta Serpin-6 -We designed degenerate primers encoding five internal peptides of M. sexta serpin-6 (1). Because the order of these peptides was unknown, we used different combinations of the primers to amplify cDNA fragments by PCR and cloned the products. Sequence analysis confirmed the cDNA inserts encoded different regions of a serpin. Using the longest one (642 bp, amplified by j802 and j803) as a probe, we screened ϳ6 ϫ 10 5 plaques in an induced M. sexta fat body cDNA library and identified 78 positives. We carried out plaque purification and in vivo excision for five of the positive clones. Complete sequence analysis indicated that their cDNA inserts were 1.5 or 2.1 kb. The long form contained an additional 0.6-kb sequence in the 3Ј-untranslated region (Fig. 1). A polyadenylation signal at the end of the coding region is probably responsible for the short transcript. We identified eight nucleotide substitutions in the open reading frames, one of which was nonsynonymous; Thr 109 is replaced by Asn 109 . Such nucleotide differences may represent allelic variations in the serpin-6 gene.
The longest cDNA contains a complete open reading frame spanning nucleotides 113-1531. Its corresponding protein is 412 residues long and includes the five peptide sequences of M. sexta serpin-6 ( Fig. 1). Following a predicted 17-residue secretion peptide, the mature protein starts with a Gln, consistent with the observation that serpin-6 is blocked at the amino terminus (1), presumably by pyroGlu. The predicted mature protein (395 residues) has a calculated molecular mass of 44,957 Da, smaller than the experimental value (46,710 Ϯ 10 Da). The difference (ϳ1153 Da) may result from N-linked gly- cosylation at Asn 100 and/or Asn 312 ; serpin-6 binds to concanavalin A, and Asn 100 is covalently modified (1). The calculated isoelectric point of serpin-6 is 6.1, higher than the experimental value of 5.4.
A BLAST search of GenBank TM revealed that M. sexta serpin-6 is most similar in amino acid sequence to D. melanogaster serpin-5 and A. gambiae serpin-9 (Fig. 2). It is 39% identical and 55% similar to Drosophila serpin-5, which may regulate the Toll and Imd pathways (38,39). M. sexta serpin-6 is 36% identical and 53% similar to A. gambiae serpin-9 but only 25 and 27% identical to M. sexta serpin-1J and sepin-3, respectively. While the reactive center loops of serpins are hypervariable, our multiple sequence alignment showed that the loop sequences of M. sexta serpin-6, D. melanogaster serpin-5, and A. gambiae serpin-9 are strikingly similar (Fig. 2). The predicted scissile bond is located between Arg and Ser residues, suggesting that these three serpins inhibit proteinases with a trypsin-like specificity (e.g. PAP-3). The loop sequences are less similar to the corresponding regions in M. sexta serpin-1J and serpin-3. M. sexta serpin-3 contains a 20-residue extension at the amino terminus, which is absent in the other serpins.
Increases in Serpin-6 mRNA and Protein Levels after a Bacterial Challenge-To understand the transcriptional regulation of serpin-6 gene in response to a microbial infection, we analyzed the relative mRNA levels in the normalized total RNA samples by reverse transcription-PCR (Fig. 3A). While the faint bands in naïve hemocytes and fat body represented a low, constitutive level of serpin-6 transcripts, a significant up-regulation was observed in both tissues after a bacterial challenge. Additionally, we examined the serpin-6 concentration in hemolymph by immunoblot analysis (Fig. 3B). A low level of serpin-6 was present in naïve M. sexta larval hemolymph. After injecting the larvae with killed E. coli or M. luteus, we detected a small decrease of the 46-kDa protein after 6 h, and the band intensity was higher than the control at 24 h post-injection. The observed protein level changes probably reflected con-sumption and replenishing of serpin-6 after the bacterial challenge. The concentrations of serpin-6 in naïve and induced hemolymph are estimated to be 15 and 30 g/ml, respectively.
Purification and Characterization of Recombinant Serpin-6 -To explore its biochemical functions, we produced serpin-6 in an E. coli expression system. The soluble serpin-6 (15 g/ml, ϳ50% of the total serpin-6 produced) was purified to near homogeneity by nickel affinity and hydrophobic interaction chromatography. Serpin-6 migrated as a 44-kDa single band on a 10% SDS-PAGE under the reducing condition (Fig. 4). Trypsinolytic peptide mass fingerprint analysis of the recombinant protein revealed 39 matching peptides, representing 77% of the overall sequence including the carboxyl-terminal fragment (data not shown).
A characteristic feature of a serpin-proteinase reaction is formation of a high M r , SDS-stable complex of the serpin and its cognate enzyme. We conducted an experiment to test whether serpin-6 can form such a complex with PAP-3. In the control of serpin-6 or PAP-3 only, PAP-3 antibodies recognized the two polypeptide chains of the proteinase but not serpin-6 ( Fig. 5A). After the two proteins were incubated together, the light chain of PAP-3 remained at the 21-kDa position, but the 37-kDa catalytic domain completely disappeared. Instead, a new immunoreactive band migrated to the 75-kDa position anticipated for a complex of serpin-6 and PAP-3 catalytic domain. This 75-kDa band was also recognized by serpin-6 antibodies (Fig. 5B). Therefore, the recombinant serpin-6 was an active inhibitor that formed a covalent complex with PAP-3.
After PAP-3 was incubated with the recombinant serpin at different molar ratios, the residual amidase activity decreased linearly as serpin-6 concentration increased (Fig. 6, A and B).
Complete inhibition occurred at a molar ratio of 2:1 (serpin-6: PAP-3). Moreover, recombinant serpin-6 also blocked proPO activation by PAP-3 and SPHs. To test whether serpin-6 could inhibit proPO activation in hemolymph, we obtained the 0ϳ50% ammonium sulfate fraction of M. sexta plasma, which contained all the components for proPO activation. Added serpin-6 at 45 and 140 ng/l blocked proPO activation by 50 and 90%, respectively (Fig. 6C). This result suggested that serpin-6 inhibits one or more of the serine proteinases in the proPO activation system.
To further characterize the serpin-proteinase reaction, we

FIG. 6. Concentration-dependent inhibition of M. sexta PAP-3 and proPO activation by serpin-6.
Purified recombinant serpin-6 was incubated with purified PAP-3 at different molar ratios for 10 min at room temperature. The residual activities (percent) were plotted against the corresponding molar ratios of M. sexta serpin-6 and PAP-3. A, amidase activity: the activity was then measured using IEARpNA substrate at 405 nm. B, proPO activation reaction: the above reaction mixtures were further incubated with proPO and SPHs on ice for 30 min prior to PO activity assay. C, proPO activation system: the plasma fraction containing different amounts of serpin-6 was incubated with M. luteus for 10 min at room temperature before PO activity assay. determined the second-order association rate constant (k a ) for the inhibition (Fig. 7). The k a (2.6 ϫ10 4 M Ϫ1 s Ϫ1 ) indicated that serpin-6 may contribute to the inhibitory regulation of PAP-3 in hemolymph.
We determined the cleavage site of serpin-6 upon reacting with PAP-3 by MALDI-TOF mass spectrometry (Fig. 8). A major peak of 4639 Da was detected in the proteinase-inhibitor mixture but not in the control spectra of serpin-6 and PAP-3 only. This peak had exactly the same mass as the carboxylterminal peptide released from a cleavage of serpin-6 between Arg 373 and Ser 374 .
Isolation and Characterization of Hemolymph Proteins Associated with Serpin-6 -We activated the induced larval hemolymph with M. luteus and examined whether or not serpin-6 and PAP-3 form a complex in plasma. Immunoaffinity chromatography using serpin-6 antibodies allowed us to isolate serpin-6 and its associating proteins from the plasma sample. We analyzed the proteins eluted from the affinity column by immunoblot (Fig. 9) and trypsinolytic peptide mass fingerprint analyses. While most serpin-6 remained intact, part of it was either cleaved or covalently linked with PAP-3 or hemolymph proteinase 8 (HP8) (Fig. 9, B-D). Formation of the serpinproteinase complexes markedly increased in the induced hemolymph after a bacterial elicitation. The peptide mass fingerprint of the upper band had 12 mass fits with serpin-6 and 8 with the PAP-3 catalytic domain (Fig. 9A). The lower band represents the complex of serpin-6 and HP8: 11 and 5 mass peaks match those calculated from sepin-6 and HP8 proteinase domain sequences, respectively. These results indicate that serpin-6 is a physiological regulator of PAP-3 and HP8 during immune responses.
Immunoblot analysis indicated that several other immune proteins also tightly associated with the antibody column. These include: M. sexta immulectin-2, proPO, PO, SPH-1, SPH-2, and HP14 (Fig. 9, E-I). Peptide mass fingerprint analysis confirmed the presence of these proteins. For instance, peptide map of the 37-kDa immunoreactive band (Fig. 9E) covers 40% of immulectin-2 sequence. Additionally, 13 trypsinolytic peptides of the 15-kDa band (Fig. 9A) match 58% of attacin-2 sequence. While similar amounts of serpin-6 and immunlectin-2 were present in the naïve and induced hemolymph, we isolated more attacin-2, proPO, PO, SPHs, HP14, and serpin-6-proteinase complexes from the induced hemolymph (Fig. 9A). This result suggests that the secondary defense reactions involve more molecules in hemolymph from the bacteria-challenged larvae. DISCUSSION It is common in vertebrates and invertebrates that extracellular serine proteinase cascades mediate acute-phase responses upon microbial infection. These pathways are often regulated by irreversible inhibitors of the serpin superfamily (15,40,41). In this study, we have cloned and characterized the newly discovered serpin-6 from M. sexta hemolymph. It is constitutively synthesized in the larval fat body and hemocytes. After a microbial challenge, induced transcription and translation of serpin-6 gene lead to a higher protein level in the plasma. Recombinant serpin-6 formed covalent complex with PAP-3 and inhibited proPO activation by the purified enzyme and in the hemolymph. The in vivo half-life of PAP-3, calculated as 1/(k a [I]), is 2.4 min in naïve larval hemolymph and 1.2 min in induced hemolymph. Endogenous serpin-6 was identified in an SDS-stable complex with PAP-3 in the hemolymph after M. luteus stimulation. These results provide good evidence that serpin-6 contributes to the regulation of PAP-3 under physiological conditions. Along with our previous work on M. sexta serpin-1J and serpin-3 (8,29,30), we demonstrate in this report that multiple serpins inhibit a single proteinase in the insect hemolymph. Comparative kinetic analysis should allow us to evaluate relative importance of these three serpins in the regulation of PAP-1, PAP-2, and PAP-3.
Considering the low sequence conservation of the serpin family, we believe the high similarity among M. sexta serpin-6, D. melanogaster serpin-5, and A. gambiae serpin-9 is significant. In particular, their P4-P4Ј residues are nearly identical in this typically hypervariable region (Fig. 2). Since this is the site where serpins specifically interact with their target enzymes, we anticipate that M. sexta serpin-6 and the dipteran serpins may perform similar physiological functions. The expression profile of serpin-6 appears to be consistent with that of Drosophila serpin-5, a potential regulator of the Toll and Imd pathways (39,42).
Since serine proteinase genes greatly outnumber serpin genes in the D. melanogaster and A. gambiae genomes, we postulated that a single insect serpin may regulate multiple serine proteinases in one or more pathways (1). Genetic analyses provide support for such hypothesis: loss-of-function mutations in a serpin gene can lead to pleiotropic phenotypes (31)(32)(33). In this paper, we present direct biochemical evidence that serpin-6 regulates two proteinases (i.e. PAP-3 and HP8) in the hemolymph. While the involvement of PAP-3 in proPO activation is established, we have not yet elucidated the biological function of HP8. Of the 18 hemolymph proteinases we recently cloned, HP8 is most similar in sequence to H. di- omphalia proPO-activating factor-III, a Ca 2ϩ -dependent activating enzyme for proPO-activating factor-II precursor (44). Further analysis is required to test the effect of serpin-6 on blocking the proteolytic processing of M. sexta proSPH-1 and proSPH-2, as a step toward understanding the physiological roles of serpin-6 in regulating proPO activation. A high M r complex of SPH-1 and SPH-2 is structurally similar and functionally equivalent to H. diomphalia proPO-activating factor-II.
Association of hemolymph proteins to the serpin antibody column provides us a unique opportunity to examine a macromolecular complex formed in the insect defense response. This complex associated with serpin-6 specifically, since we did not detect any of these proteins bound to the control column coupled with the preimmune antibodies. 3 Due to the stringent washing conditions and absence of abundant hemolymph proteins in the eluate, direct or indirect associations of these proteins with serpin-6 appear to be strong and specific. Additionally, the composition of such complex appears to be quite stable. Many of these proteins were also identified in the bound fractions from the immunoaffinity columns for M. sexta serpin-4 and serpin-5, negative regulators of the proPO activation system. 4 Among the co-purified proteins, immulectin-2 bound to lipopolysaccharide of Gram-negative bacteria and stimulated the proPO activation (45). SPH-1 and SPH-2, which associated with immulectin-2 (9), were also identified as major proteins in this experiment (Fig. 9, E, G, and H). These two proteins are required by PAP to generate active PO. ProPO and PO were detected in the complex and so was HP14, an initiating proteinase of the proPO activation system (Fig. 9, F and I) (46). While these proteins all participate in melanization, we have also co-eluted several minor components that may be also involved in proPO activation and/or other defense reactions. These include HP17 and HP21. 3 Much to our surprise, M. sexta attacin-2 in the induced hemolymph also associated with the antibody column (Fig. 9A); we expected antimicrobial peptides to directly and independently interact with bacterial surfaces. The detection of attacin-2 suggests that recognition specificity and binding strength of the immune protein complex may be enhanced by associating with a bacteria-killing protein.
Taken together, our results strongly suggest that a bacterial entry may elicit a subset of hemolymph proteins to associate and form a non-covalent protein complex. The molecular interactions among pattern recognition proteins, HPs, SPHs, proPO, and other proteins in the complex ensure a localized defense reaction against the invading organisms. After exerting their functions, active serine proteinases are inhibited by multiple serpins to limit potential damage to the host tissues and cells. Serpin affinity chromatography has allowed us to observe the immune complex for the first time.  The reaction mixture of serpin-6 and PAP-3 was directly analyzed by MALDI-TOF mass spectrometry. A representative strong single-accumulation spectrum is presented with the mass values on top of the MH ϩ peaks. The spectrum was subjected to noise removal and calibrated with an external standard of horse apomyoglobulin. The control spectra of serpin-6 and PAP-3 alone do not contain corresponding mass peak.