Interaction of β-1,3-Glucan with Its Recognition Protein Activates Hemolymph Proteinase 14, an Initiation Enzyme of the Prophenoloxidase Activation System in Manduca sexta*

A serine proteinase pathway in insect hemolymph leads to prophenoloxidase activation, an innate immune response against pathogen infection. In the tobacco hornworm Manduca sexta, recombinant hemolymph proteinase 14 precursor (pro-HP14) interacts with peptidoglycan, autoactivates, and initiates the proteinase cascade (Ji, C., Wang, Y., Guo, X., Hartson, S., and Jiang, H. (2004) J. Biol. Chem. 279, 34101–34106). Here, we report the purification and characterization of pro-HP14 from the hemolymph of bacteria-injected M. sexta larvae. The zymogen, consisting of a single polypeptide with a molecular mass of 68.5 kDa, is truncated at the amino terminus. It is converted to a two-chain active form in the presence of β-1,3-glucan (a fungal cell wall component) and β-1,3-glucan recognition protein-2. The 45-kDa heavy chain contains four low-density lipoprotein receptor A repeats, one Sushi domain, and one unique cysteine-rich region, whereas the 30-kDa light chain contains a serine proteinase domain, which was labeled by [3H]diisopropyl fluorophosphate. Pro-HP14 in the plasma strongly binds curdlan, zymosan, and yeast and interacts with peptidoglycan and Micrococcus luteus. Addition of autoactivated HP14 elevated phenoloxidase activity level in the larval plasma. Recombinant M. sexta serpin-1I reduced prophenoloxidase activation by inhibiting HP14. These data are consistent with the current model on initiation and regulation of the prophenoloxidase activation cascade upon recognition of pathogen-associated molecular patterns by specific pattern recognition proteins.

Innate immunity is essential for the wellbeing of vertebrates and invertebrates. Key features of this defense system include pattern recognition of infectious agents, stimulation of extra-and intracellular signaling pathways, immediate cellular/humoral responses, and induced synthesis of immune factors (e.g. antimicrobial peptides) (1)(2)(3). Serine proteinase cascades in plasma, such as the human complement system and horseshoe crab hemolymph coagulation pathway, mediate many of the defense responses (4,5). In insects, extracellular serine proteinases generate spätzle, phenoloxidase (PO), 2 and plasmatocyte spreading peptide via limited proteolysis (4,6). In Drosophila, a proteinase cascade activates spätzle to induce antimicrobial peptide synthesis by the Toll pathway. While Persephone (a clip-domain serine proteinase) is a component of the cascade (7), its activating proteinase and protein substrate remain unclear. Neither is it known how Gram-positive bacteria trigger the cascade activation (8) or how melanization often associates with malfunctioning of this proteinase pathway (9 -11).
Microbial infection or tissue damage often leads to proteolytic activation of prophenoloxidase (pro-PO) in insects. Active PO catalyzes the formation of quinones, reactive intermediates for melanization (12)(13)(14). Quinones may also participate in wound healing and sequestration of parasites or pathogens. To minimize the cytotoxicity of quinones to host tissues/cells, conversion of pro-PO to PO occurs as a local reaction involving a system of interacting proteins (e.g. pattern recognition receptors, serine proteinases, serine proteinase homologs, serpins, and pro-PO). Although some of these molecules have been characterized at the molecular level, the constituents, order, initiation, and regulation of the proteinase pathway are still poorly understood. Our current knowledge is mainly limited to pathogen recognition and pro-PO activation, the first and last steps of the cascade (6,15).
Recently, we reported the cDNA cloning, recombinant expression, and functional elucidation of Manduca sexta HP14, an immune-responsive hemolymph proteinase (16). This Sushi-domain enzyme contains seven Cys-rich regulatory domains and a serine proteinase domain. Incubation with peptidoglycan leads to autoprocessing of the recombinant pro-HP14 and generation of an amidase activity. Supplementation of hemolymph with the zymogen enhanced pro-PO activation in response to Micrococcus luteus. These data suggest that pro-HP14 binds to the Gram-positive bacteria, autoactivates, and triggers the immune proteinase pathway in M. sexta. To further explore the physiological function of HP14, we purified and characterized pro-HP14 from the hemolymph of M. sexta larvae injected with microbes. In the presence of a fungal cell wall component and its recognition protein, autoproteolysis of pro-HP14 yielded active HP14 that enhanced pro-PO activation in the plasma. Its binding to microbial surface components and inhibition by M. sexta serpin-1I were also investigated.

EXPERIMENTAL PROCEDURES
Insect Rearing, Bacterial Challenge, and Hemolymph Collection-M. sexta eggs were purchased from Carolina Biological Supply. The larvae were reared on an artificial diet (17). Day 2, fifth instar larvae were injected with a mixture of formaldehyde-killed Escherichia coli (3 ϫ 10 7 cells), M. luteus (30 g), and curdlan (30 g) in 50 l of H 2 O. Hemolymph was collected from cut prolegs of the larvae at 24 h after the immune challenge. Individual hemolymph samples were immediately mixed with equal volume of pH 7.0, 100% saturated (NH 4 ) 2 SO 4 to prevent melanization.
Purification of Pro-HP14 from Induced M. sexta Hemolymph-All procedures for pro-HP14 purification were carried out at 4°C using buffers filtered through a 0.22-m membrane. The induced hemolymph sample (55 ml, stored at Ϫ80°C) was thawed as 50% (NH 4 ) 2 SO 4 suspension. After centrifugation at 17,400 ϫ g for 20 min, the protein precipitate was collected and dissolved in 50 ml of buffer A (1 mM benzamidine, 0.01% 1-phenyl-2-thiourea, 0.5 M NaCl, 10 mM potassium phosphate, pH 6.8). The solution was centrifuged at 17,400 ϫ g for 30 min to remove flocculent materials. Saturated (NH 4 ) 2 SO 4 solution was slowly added to the supernatant to a final saturation of 35%. After centrifugation, the precipitate was collected and dissolved in 15 ml of buffer A.
For rapid removal of (NH 4 ) 2 SO 4 , the protein sample was passed through a Sephadex G25 column (2.5 cm i.d.ϫ 16 cm) containing buffer A. Protein fractions were pooled (ϳ40 ml) and applied to a hydroxylapatite column (2.5 cm inner diameter ϫ 7 cm, Bio-Rad) equilibrated with the same buffer. After washing with 100 ml of buffer A, bound proteins were eluted with a linear gradient of 10 -150 mM potassium phosphate in buffer A at 0.5 ml/min for 6 h. Fractions were analyzed by SDS-PAGE and immunoblotting using HP14 antibody. The flowthrough fractions were combined and precipitated with 50% saturated (NH 4 ) 2 SO 4 . Following centrifugation, the pellet was dissolved in 3 ml of buffer B (pH 7.5, 20 mM Tris-HCl, 0.5 M NaCl, 0.01% Tween 20, and 1 mM benzamidine) and separated on a Sephacryl S100-HR column (2.5 cm inner diameter ϫ 120 cm) equilibrated with buffer B.
The pro-HP14 fractions were combined, supplemented with 1 mM CaCl 2 and MgCl 2 , and loaded onto a concanavalin A-Sepharose column (5 ml). After washing, pro-HP14 was eluted from the lectin column with buffer B containing 1 mM CaCl 2 , 1 mM MgCl 2 , and 0.4 M methyl-␣-Dmannopyranoside. Eluted proteins were diluted with 10 volumes of buffer C (0.01% Tween 20, 2 mM benzamidine, 10 mM potassium phosphate, pH 6.4) and applied to a dextran sulfate-Sepharose CL-6B column (5 ml) (21). After washing with 30 ml of buffer C, a linear gradient of 0 -1.0 M NaCl in buffer C was employed to elute the bound proteins at 1 ml/min for 30 min. Twenty fractions of 1.5 ml were collected and analyzed by SDS-PAGE. The purified pro-HP14 was stored at Ϫ80°C before use. Prior to functional analyses, the purified protein was passed through a Sephadex G25 column (1.5 cm inner diameter ϫ 15 cm) equilibrated with pH 7.5, 20 mM Tris-HCl, 20 mM NaCl to remove benzamidine.
Characterization of M. sexta Pro-HP14-MALDI-TOF mass spectrometry of pro-HP14 was carried out at Nevada Proteomics Center. Briefly, the purified protein (2 g) was separated by 10% SDS-PAGE under reducing condition and stained with Coomassie Blue. Individual protein bands were excised from the gel, treated with 100 mM iodoacetamide, and digested with sequencing-grade trypsin (25 l and 5 ng/l) (Promega) in 25 mM ammonium bicarbonate at 37°C for 16 h (22). In a parallel experiment, gel slices containing 2 g of pro-HP14 were reacted with iodoacetamide and then 100 ng of Glu-C (Princeton Separation) at 30°C for 16 h. After desalting with a ZipTip C18 (Millipore), the peptides were analyzed on an ABI 4700 MALDI-TOF/TOF mass spectrometer (Applied Biosystems). Mass data were acquired in reflector mode within the range of 800 -4,000 Da, and 2500 laser shots were averaged for each mass spectrum. Molecular masses of the peptides from pro-HP14 were calculated (prospector.ucsf.edu/) and compared with those of observed monoisotopic peaks to locate their positions. The isoelectric point of pro-HP14 was measured on a precast PhastGel IEF-9 (Amersham Biosciences) (23).
Optimization of Conditions for Pro-HP14 Autoactivation-The general conditions for curdlan-induced pro-HP14 activation were: purified pro-HP14 (200 ng, 10 l), curdlan (10 g, 1 l), ␤GRP2 (40 ng, 2 l), CaCl 2 (100 mM, 1 l), and buffer D (2 l) incubated at 37°C for 1 h. One of these conditions was changed as specified in Fig. 3 to test the effects of ␤GRP2/curdlan amount, incubation time/temperature/pH, and NaCl concentration. To assess cleavage extent, the reaction mixtures (8 l) were separated by 10% SDS-PAGE under reducing condition and visualized by silver staining. The effect of pH was investigated by replacing buffer D with 3 l, 1:5 diluted Polybuffer 96 (Amersham Biosciences) adjusted to various pH.
Determination of the Proteolytic Cleavage Site in HP14-HP14 generated under the optimal conditions was resolved by SDS-PAGE and transferred to a polyvinylidene difluoride membrane. After staining with Ponceau S (Sigma), the catalytic chain of HP14 was subjected to automated Edman degradation at Harvard Microchemistry Facility.
Binding of Plasma Pro-HP14 to Microorganisms-Following 0 -35% ammonium sulfate fractionation and centrifugation, hemolymph (0.5 ml) from the injected larvae was dissolved in 200 l of 20 mM Tris-HCl, 0.5 M NaCl, 1 mM CaCl 2 , 1 mM MgCl 2 , 0.01% 1-phenyl-2-thiourea, pH 7.5. The plasma fraction was incubated with M. luteus (2 ϫ 10 9 cells), E. coli (2 ϫ 10 9 cells), S. cerevisiae (5 ϫ 10 7 cells), or microbial cell wall components (1 mg and 100 l) at room temperature for 30 min. After centrifugation at 13,000 ϫ g for 1 min, the supernatants were removed. The pellets were washed five times with 50 mM Tris-HCl, 5 mM CaCl 2 , pH 7.8. Bound proteins were eluted with 40 l of 2 ϫ SDS sample buffer at 100°C for 5 min. As controls, the microbial cells/polysaccharides only were treated similarly. The supernatants, bound proteins, and control samples were subjected to 10% SDS-PAGE and immunoblot analysis using 1:2000 diluted HP14 antiserum as the first antibody. Distribution of pro-HP14 in the supernatant and pellet was estimated by comparing relative intensities of the immunoreactive bands.
Binding of Purified Pro-HP14 to Microorganisms-Purified pro-HP14 (1 g of 50 l) and CaCl 2 (100 mM of 5 l) were incubated with micro-bial cells or cell-surface components in the presence or absence of their binding proteins (200 ng and 10 l) at 37°C for 60 min. The total volume was adjusted to 65 l with buffer D. As described above, the supernatants and bound proteins were obtained and subjected to SDS-PAGE followed by silver staining or immunoblot analysis. Microbial cells or cell-wall components, pro-HP14, and recognition proteins were included as controls.
Stimulation of Pro-PO Activation System by HP14-Pro-HP14 (1 g and 50 l), curdlan (1-mg pellet), ␤GRP2 (200 ng and 10 l), and CaCl 2 (100 mM and 5 l) were incubated at 37°C for 90 min. After centrifugation, 2 l of the supernatant was incubated with plasma (1:10 diluted in buffer D, 5 l) and buffer D (10 l) on ice for 20 min. PO activity in the reaction mixture was assayed using dopamine as a substrate (25). The preincubation mixture without pro-HP14 or curdlan-␤GRP2 was used as negative controls.
Inhibition of HP14-induced PO Activation by Serpin-1I-To detect the formation of an SDS-stable complex of HP14 and serpin-1I, purified pro-HP14 (200 ng and 10 l), serpin-1I (3 g and 1 l), curdlan (10 g and 1 l), ␤GRP2 (40 ng and 2 l), and CaCl 2 (100 mM and 1 l) were incubated at 37°C for 90 min. In the controls, pro-HP14 or serpin-1I was replaced by the same volumes of buffer D. The reaction mixture and negative controls (10 l) were separated by 7.5% SDS-PAGE and subjected to immunoblot analysis using 1:1000 diluted serpin-1 antiserum.
To examine the effect of serpin-1I on pro-PO activation induced by HP14, the active proteinase was generated under the optimal conditions. After centrifugation, supernatant (2 l), serpin-1I (3 g and 1 l) and buffer D (9 l) were incubated on ice for 10 min. Induced plasma (5 l and 1:10 diluted) was added to the HP14-serpin-1I mixture and further incubated on ice for 20 min prior to PO activity assay. A mixture of the supernatant (2 l), buffer D (10 l), and induced plasma (5 l and 1:10 diluted) was included as a positive control, whereas plasma only was a negative control.

RESULTS
Purification and Characterization of M. sexta Pro-HP14 from the Induced Larval Hemolymph-Using polyclonal antiserum against HP14, we detected a protein doublet in the hemolymph, whose level increased after an immune challenge (16). The immunoreactive proteins with apparent molecular masses of 75 and 67 kDa were mostly found in 0 -35% ammonium sulfate fraction of the induced plasma. After desalting and buffer exchange, we resolved the fractionated hemolymph by hydroxylapatite chromatography. The flow-through contained the large protein, whereas the bound part included the small one (Fig. 1, A and B). Because the secreted recombinant pro-HP14 had an apparent M r of ϳ75,000 (16), we decided to isolate pro-HP14 from the unbound fractions, representing ϳ20% of the total proteins loaded onto the hydroxylapatite column. Following a concentration step, the pooled flow-through fractions were separated by gel-filtration chromatography on a Sephacryl S100-HR column. About 85% of the proteins were removed in this step. The combined pro-HP14 fractions were loaded onto a concanavalin A-Sepharose column. After nearly half of the loaded proteins were removed, pro-HP14 and other glycoproteins were eluted from the lectin column. The pooled pro-HP14 fractions were resolved by ion-exchange chromatography onto a dextran sulfate column. Pro-HP14 eluted from the column at ϳ0.25 M NaCl. With ϳ7 ϫ 10 3 -fold purification, 179 g of pro-HP14 was isolated from 55 ml of induced hemolymph. As judged by SDS-PAGE analysis, the proenzyme was essentially pure, and it migrated to the 75-kDa position under reducing condition (Fig. 1B).
We determined by MALDI-TOF mass spectrometry that the molecular mass of "75-kDa" pro-HP14 from the induced plasma was actually 68,545 Ϯ 69 Da, smaller than the theoretical value of the mature proenzyme (residues 1-649, 71,774 Da) (16). Peptide mass fingerprint analysis indicated that 28 predicted trypsinolytic fragments (800 -4,000 Da) were apparent on the mass spectrum, covering 59% of the sequence from residue 68 to the carboxyl terminus ( Table 1). The identification of 644 FWTDEY 649 demonstrated that the proenzyme was intact at the carboxyl terminus and, therefore, truncated at the amino terminus to account for the mass reduction. Due to amino-terminal blocking, automated Edman degradation failed to yield any sequence from pro-HP14. However, we detected two peptides (760.34 and 949.46 Da) that are identical in mass to pyroGlu 62 LSNCR 67 (760.337 Da, carboxy-amidomethyl Cys (CAM-C)) and 68 ISQWQCK 74 (949.457 Da, CAM-C). Right before Gln 62 , there is a recognition site of intracellular processing enzymes: Arg 58 -Ser-Arg-Arg 61 (26).
Was pro-HP14 truncated somewhere before Gln 62 , and, after trypsin cleavage at Arg 61 , the exposed Gln 62 spontaneously became pyroGlu 62 ? Although it is possible, evidence suggests this is not the case: 1) there is no other known processing site in the first LDL r A repeat or in the following linker (residues 44 -65); 2) the first Gln in 106 QCQYNWFR 113 did not become pyroGlu 106 under the same condition (Table 1); 3) Glu-C digestion generated a peptide (1805.835 Da) with a mass identical to that of pyroGlu 62 LSNCRISQWQCKD (1805.802 Da, 2 CAM-C). Therefore, the 75-kDa pro-HP14 starts at pyroGlu 62 and Gln 62 -pyro-Glu 62 conversion occurred before trypsin digestion.
The calculated mass of pro-HP14 (residues 62-649) is 65,183 Da, smaller than the experimental value (68,545 Da). Because pro-HP14 binds concanavalin A, glycosylation appears to be responsible for the mass difference of 3,362 Da. We predict that the post-translational modification occurs at N 145 ET, N 383 GT, and/or N 582 GT, because peptide masses containing these were not found ( Table 1). The isoelectric point of pro-HP14 was determined to be 5.7, slightly higher than the calculated value (5.5) from its amino acid sequence (residues 62-649).
Cleavage Activation of Pro-HP14 Induced by Microorganisms and Their Binding Proteins-To test whether or not microorganisms could activate pro-HP14, we incubated purified pro-HP14 with various microbial cells or cell-surface molecules in the presence/absence of their binding proteins. SDS-PAGE analysis indicated that pro-HP14 was cleaved after incubating with yeast, curdlan, or zymosan A in the presence of ␤GRP2 (Table 2). However, laminarin (a soluble form of ␤-1,3- glucan) did not lead to pro-HP14 cleavage activation. E. coli did not cause such a change with or without immulectin-2, and neither did M. luteus. Whereas incubation of pro-HP14 with peptidoglycan did lead to pro-HP14 cleavage, the efficiency was lower than with ␤-1,3-glucan and ␤GRP2.
Proteolytic processing of pro-HP14 occurred when insoluble ␤-1,3glucan and ␤GRP2 were present at the same time ( Fig. 2A). Because pro-HP14 cleavage did not happen after incubation with curdlan or ␤GRP-2 alone, binding-induced autoactivation (rather than a contaminating proteinase) is responsible for the limited proteolysis. Cleaved HP14 was separated into 45-and 30-kDa bands by SDS-PAGE under reducing condition, whereas, under nonreducing condition, HP14 migrated as a single band at ϳ70 kDa (Fig. 2B). This agrees with our prediction based on the deduced sequence and domain structure of HP14: an interchain disulfide bond links the 30-and 45-kDa polypeptides (16). Although the 45-kDa heavy chain was blocked at the amino terminus, we determined the first ten amino acid residues of the light chain as: Val-Leu-Gly-Gly-Glu-Arg-Ala-Gln-Phe-Gly. This perfectly matched the sequence after the predicted proteolytic activation site (Gly-Thr-Glu-Leu*Val-Leu-Gly-Gly). Affinity labeling by [ 3 H]DFP demonstrated that the light chain was catalytically active (Fig. 2C). In other words, the 30-kDa band represents the proteinase domain at the carboxyl terminus.
We optimized the reaction conditions for pro-HP14 activation. At 15 min after 200 ng of pro-HP14 was incubated with 10 g of curdlan and 10 ng of ␤GRP2, nearly half of the proenzyme was converted to the two-chain active HP14 (Fig. 3). The cleavage reaction was Ͼ90% complete at 45 min. Further increase in ␤GRP2 or curdlan level had little effect on the proteolytic activation. Ionic strength of the reaction mixture negatively impacted the molecular interactions: 80% of the cleavage was blocked in the presence of 200 mM NaCl. KCl, Na 2 SO 4 , and NaAc had a similar effect (data not shown). The optimum pH for pro-HP14 autoactivation was ϳ7.5-8.5, and the cleavage was more complete at 37°C than 25°C (data not shown). At the physiological pH (6.7) and ionic strength (equivalent of 200 mM NaCl) (27), the proteolysis in the

Microbial cells or surface molecules
Binding protein larval plasma is estimated to occur at 15% of the optimal rate determined in vitro. Binding of Pro-HP14 to Microorganisms in the Presence or Absence of Plasma Proteins-To better understand the role of pro-HP14 in immune responses, we tested its binding with various microbial cells or cell-surface molecules using 0 -35% ammonium sulfate fraction of the plasma. In the presence of high salt and 1-phenyl-2-thiourea, pro-HP14 activation and protein cross-linking were prevented. Pro-HP14 in the fractionated hemolymph bound to the Gram-positive bacteria and yeast but not E. coli (Fig. 4B). Peptidoglycan and ␤-1,3-glucan appeared to be responsible for the binding. Although the reduction of pro-HP14 in M. luteus-and peptidoglycan-treated supernatants were relatively small, strong interactions between pro-HP14 and curdlan/zymosan completely depleted the proenzyme in the solution phase (Fig. 4A). The binding of pro-HP14 to S. cerevisae greatly reduced the proenzyme in the supernatant.

E. coli
We further examined whether the purified pro-HP14 can directly associate with microbes in the presence or absence of their recognition proteins. Immunoblot analysis showed that most pro-HP14 remained in supernatants treated with E. coli, M. luteus, or S. cerevisae (Fig. 5A and Table 3). Because supplementing the mixtures with immulectin-2 or ␤GRP2 did not increase the amount of binding, other plasma proteins must have contributed to the binding of pro-HP14 to M. luteus or yeast (Fig. 4). In the presence of ␤GRP2, zymosan or curdlan caused larger reduction of pro-HP14 in the supernatants than yeast did (Fig. 5A). This reduction was partly due to the proteolytic processing of pro-HP14.
Although ␤GRP2 did not seem to enhance the binding of pro-HP14 to S. cerevisae, its coexistence with curdlan induced the proteolytic processing of pro-HP14 (Fig. 5B).
Initiation of Pro-PO Activation System in the Larval Hemolymph by Autocleaved HP14-To test whether or not HP14 triggers the pro-PO activation pathway in the plasma, we first used curdlan and ␤GRP2 to completely activate pro-HP14 under the optimal conditions (Fig. 3). After removing the elicitor and associated ␤GRP2 by centrifugation, the supernatant containing the two-chain HP14 was added to plasma from the naïve or bacteria-injected larvae. After incubation, high levels of PO activity developed in the control and induced hemolymph samples (Fig.  6). The enzyme activities (0.06 and 1.21 units) were low in the plasma controls and became slightly higher when pro-HP14 was added (0.32 and 1.59 units), thus, the proteinase zymogen had only a small effect on pro-PO activation. In another control, after the supernatant of curdlan and ␤GRP2 mixture was incubated with the two plasma samples, higher PO activities (0.52 and 3.18 units) were detected, suggesting that the supernatant may contain trace amounts of curdlan and ␤GRP2 to activate the proteinase cascade. Nevertheless, such activity increases were much lower than those in HP14-treated plasma samples (20.5 and 43.4 units). These results are consistent with the conclusion that HP14 initiated the pro-PO activation system in the hemolymph.
Inhibition of HP14 by Serpin-1I-Although [ 3 H]DFP labeling, pro-HP14 auto-processing, and pro-PO activation demonstrated HP14 is an active serine proteinase, we failed to detect its amidase activity using Ala-Ala-Pro-Leu-p-nitroanilide, as we did with peptidoglycan-activated recombinant HP14 (16). To further demonstrate that HP14 is an active proteinase, we decided to test if HP14 could form an SDS-stable complex with serpin-1I. Among the 12 variants of M. sexta serpin-1, serpin-1I has a reactive site sequence (AL*SL) most similar to the proteolytic activation site for HP14 (EL*VL). After curdlan-activated HP14 was incubated with recombinant serpin-1I, we detected a complex using serpin-1 antibodies (Fig. 7A). The immunoreactive band, absent in the controls of serpin-1I and HP14 only, had an apparent M r (ϳ75 kDa) close to that predicted for the complex formed between the HP14 catalytic domain and serpin-1I. No such complex was formed with serpin-1J, a negative control with an Arg at the P1 position (data not shown). We further examined the effect of serpin-1I on pro-PO activation induced by active HP14. As shown in Fig. 6, supplementation of active HP14 greatly enhanced pro-PO activation in the induced hemolymph. Preincubation of activated HP14 with serpin-1I significantly suppressed pro-PO activation (Fig. 7B).

DISCUSSION
Innate immunity in vertebrates and invertebrates is activated by host proteins recognizing conserved surface determinants of pathogens (28).  Upon binding, these proteins stimulate plasma factors and blood cells to immobilize and kill the invading microorganisms. Although many defense reactions are mediated by extracellular serine proteinase pathways (4,5), limited information is available on how pathogen recognition leads to proteinase activation. A current model suggests that the binding of recognition proteins to a microbial surface induces a conformational change required for interacting with the first proteinase zymogen, leading to its autoactivation and pathway initiation (4,12,29). The lectin-mediated pathway for complement activation in mammals is an example for this model: mannose-binding protein and mannose-binding protein-associated serine proteinase (MASP) trigger the serine proteinase cascade (30). In this report, we provide biochemical evidence that the pathway for pro-PO activation in a lepidopteran insect is initiated in similar manner: ␤-1,3-glucan and ␤GRP2 together induced pro-HP14 autoactivation (Fig. 2), and active HP14 by itself started the pro-PO activation cascade (Fig. 6). These similarities in the pro-PO activation pathway in arthropods and the complement system in vertebrates suggest that proteinase autoactivation to trigger innate immune responses is an ancient evolutionary adaptation for defense against infection (4,14).
A few known arthropod initiation serine proteinases greatly differ in their domain and subunit organizations. The ␤-1,3-glucan-responsive proteinases, M. sexta HP14 and Tachypleus tridentatus factor G, trigger pro-PO activation and hemolymph coagulation cascades, respectively (4,5). In the insect, recognition of ␤-1,3-glucan is performed by a separate protein (␤GRP2), whereas in the horseshoe crab, a subunit of factor G executes the same function. This subunit, which associates with the proteinase subunit, is encoded by a separate gene. In contrast, the horseshoe crab factor C consists of a single polypeptide. This lipopolysaccharide-responsive enzyme, by strictest definition, is a patternrecognition proteinase. Although T. tridentatus factor C and M. sexta HP14 both contain a Sushi domain and a catalytic domain, the types of other structural modules (e.g. LDLr A repeat, C-type lectin) are strikingly different, and, unlike Factor C, pro-HP14 does not autoactivate in the presence of lipopolysaccharide.
The HP14 zymogen produced by baculovirus-infected insect cells  Not determined a Binding of pro-HP14 in plasma is represented as % of the total proenzyme in the supernatant. The relative levels of pro-HP14 were determined by immunoblotting and densitometry of the protein band. b Binding of pro-HP14 to microbial cells or cell-wall components is represented as % of the total purified zymogen in the unbound fraction. had two forms: the 87-kDa pro-HP14 was primarily intracellular, whereas the 75-kDa one was mostly in the medium (16). With an apparent M r similar to the purified hemolymph pro-HP14, the latter may also start with pyroGlu 62 . Indeed, the 75-kDa recombinant protein was blocked at its amino terminus and intact at the carboxyl terminus. In light of these findings, we further speculate that the 87-kDa protein represents the full-length pro-HP14 with five LDLr A repeats. Because the 87-kDa protein was not detected in M. sexta larval plasma, the function of its first LDLr A repeat seems dispensable. Orthologs of HP14 in D. melanogaster, A. gambiae, and Apis mellifera contain four instead of five LDLr A repeats at the amino terminus. On the other hand, the 67-kDa protein in plasma may represent pro-HP14 with the first two LDLr A repeats removed, antibodies specific for LDLr, Sushi, 7C, or proteinase domain all recognized this band. 3 Structural and functional comparisons of this form of pro-HP14 with the 75-kDa pro-HP14 are under active investigation.
The structure-function relations of human MASPs have shed light on our analysis of M. sexta HP14. Corresponding to the amino-terminal LDLr A repeats, the CUB-EGF-CUB region in MASP-1 and MASP-2 is responsible for interacting with mannose-binding protein and for forming homodimer (31). The following two Sushi domains in MASP-2 sta-bilize the structure of the serine proteinase domain at the carboxyl terminus and are directly involved in the enzyme autoactivation (32). In HP14, the second Sushi domain does not exist: between the Sushi and proteinase domains, there is another cysteine-rich region (16). Perhaps, the last (or seventh) Cys in this region forms the interchain disulfide bond with Cys 510 in the catalytic domain. If proven true, the part that contains the first six cysteine residues may represent the prototype of a unique domain structure stabilized by three disulfide bonds. We tentatively call that "Wonton domain." Structural and functional analyses are required to test if the Sushi and Wonton domains take part in the autoactivation of HP14 and interaction with its substrate.
One important feature of the pro-HP14 autoproteolysis induced by curdlan and ␤GRP2 is the high specificity: a single cleavage at the predicted activation site yielded two polypeptides (Fig. 3). In contrast, after recombinant pro-HP14 had been incubated with peptidoglycan, a series of low M r bands appeared in a time-dependent manner (16). This apparent low specificity could be caused by the lack of particular plasma factors in the activation reaction. In fact, formation of a macromolecular complex containing HP14 and other immune proteins was observed in M. sexta larval plasma incubated with microbial surface components (33,34). Consequently, further studies are needed to test if other proteins (e.g. peptidoglycan recognition proteins) are required for specific cleavage activation of pro-HP14 by peptidoglycan. Such experiments would further establish the physiological role of HP14 as a common entry point for responses against Gram-positive bacteria and fungi.
We demonstrated that HP14 is an active proteinase that can be trapped in a covalent complex with a suicidal inhibitor (Fig. 7). Although this is not to say that HP14 is the physiological target of serpin-1I (because, even at a high concentration, serpin-1I failed to completely block pro-PO activation), it did suggest a possible regulatory mechanism for pathway initiation. Moreover, the Leu residue preceding the cleavage site in serpin-1I (and HP14) provided important clues for elucidating the protein substrate of HP14. Because HP14 is a chymotrypsin-like enzyme preferring to cleave after Leu, we could test the ability of HP14 to activate several M. sexta hemolymph proteinase precursors (pro-HP2, pro-HP16, pro-HP19, pro-HP21, and pro-HP22) that have a Leu at the putative activation site (35). This will be our next step toward elucidating the pro-PO activation system in M. sexta.