The Trypsin Inhibitor Panulirin Regulates the Prophenoloxidase-activating System in the Spiny Lobster Panulirus argus

Background: The melanization reaction is an essential immune response in arthropods that should be tightly regulated. Results: A novel competitive and tight-binding trypsin inhibitor, panulirin, that inhibits the melanization response to lipopolysaccharides was found in lobster. Conclusion: Panulirin regulates serine peptidases in the pathway toward the activation of the prophenoloxidase enzyme. Significance: Serine peptidase inhibitors play a key role in controlling the immune response in arthropods. The melanization reaction promoted by the prophenoloxidase-activating system is an essential defense response in invertebrates subjected to regulatory mechanisms that are still not fully understood. We report here the finding and characterization of a novel trypsin inhibitor, named panulirin, isolated from the hemocytes of the spiny lobster Panulirus argus with regulatory functions on the melanization cascade. Panulirin is a cationic peptide (pI 9.5) composed of 48 amino acid residues (5.3 kDa), with six cysteine residues forming disulfide bridges. Its primary sequence was determined by combining Edman degradation/N-terminal sequencing and electrospray ionization-MS/MS spectrometry. The low amino acid sequence similarity with known proteins indicates that it represents a new family of peptidase inhibitors. Panulirin is a competitive and reversible tight-binding inhibitor of trypsin (Ki = 8.6 nm) with a notable specificity because it does not inhibit serine peptidases such as subtilisin, elastase, chymotrypsin, thrombin, and plasmin. The removal of panulirin from the lobster hemocyte lysate leads to an increase in phenoloxidase response to LPS. Likewise, the addition of increasing concentrations of panulirin to a lobster hemocyte lysate, previously depleted of trypsin-inhibitory activity, decreased the phenoloxidase response to LPS in a concentration-dependent fashion. These results indicate that panulirin is implicated in the regulation of the melanization cascade in P. argus by inhibiting peptidase(s) in the pathway toward the activation of the prophenoloxidase enzyme.

Several cytotoxic molecules are produced during melanogenesis, which have recently been demonstrated to be effective to combat infections (4,6). These intermediates can be also highly deleterious to the host if they are produced uncontrollably (5,7). Therefore, arthropods have developed several means to regulate the melanization reaction, both spatially and temporally, to avoid damage to the host.
Location of the PO response seems to be an important regulatory mechanism for insect defense against infections (8). In this sense, components of the proPO-activating system may associate to form a large noncovalent complex, which localizes the melanization to the surface of invading microorganisms or at the injury site (8 -10). This complex ensures a high local concentration of quinone products where necessary, whereas it avoids their dissemination. In addition, the stickiness of activated phenoloxidase promotes its deposition on pathogen or anomalous surfaces, where it assists localized melanization (5).
The melanin formation can be controlled even at late stages of the melanogenesis. For example, the presence of melanization inhibition proteins has been described in plasma from crustaceans (11) and insects (12). These proteins hamper the synthesis of melanin from quinones but have no effect on PO enzyme activity. Interestingly, melanization inhibition proteins from crustaceans and insects share a similar molecular mass (43 kDa) but are remarkably different in their primary structure (11).
The most straightforward solution for controlling the proPO-activating system could be that enzymatic components, such as PO and peptidases, exist inactive as zymogens (3,4). However, it is reasonable to expect the occurrence of mechanisms to control their activities once they become active. Several phenoloxidase inhibitors have been characterized in insects (13)(14)(15). Also in insects, genetic evidence suggests that serpin-type peptidase inhibitors are involved in regulating the melanization cascade (16). However, the peptidases inhibited by serpins have only been identified in the beetle Tenebrio molitor (17) and the tobacco hornworm Manduca sexta (16). Serpins regulate the proPO system in insects by inhibiting both the ppA (18 -20) and peptidases upstream of the ppA in the cascade (10,16,21).
Until the present report, pacifastin from the crayfish Pacifastacus leniusculus (22,23) was the only known peptidase inhibitor regulating the proPO system in crustaceans, despite the fact that this system has been investigated in a variety of crustaceans for decades. Pacifastin regulates the activity of the ppA. It is a heterodimeric protein (155 kDa) composed of two covalently linked subunits, each encoded by two different mRNAs. The light chain (44 kDa) contains the inhibitory domains, whereas the heavy chain (105 kDa) is instead related to transferrins (24).
In our first study on the proPO-activating system in spiny lobster, we indicated the presence of trypsin-inhibitory activity in the hemocyte lysate (25). Here, we describe the purification and some molecular and biological properties of a novel trypsin inhibitor that we have named panulirin. The low similarity to other protein inhibitors found through amino acid sequence comparison suggests the finding of a new class of peptidase inhibitor. Panulirin is a 5.3-kDa basic peptide (pI 9.5) composed of 48 amino acid residues, which contains six cysteine residues engaged in disulfide bridges. It is a competitive, reversible and tight-binding inhibitor of trypsin. Experimental evidence indicated that panulirin is implicated in the regulation of the proPO-activating system in the spiny lobster. 2), Triton X-100, protamine sulfate, glucose, sodium citrate, and calcium chloride were all obtained from Merck. Sephadex G-50 Superfine was from Amersham Biosciences. The HiTrap SP HP column, the low molecular weight calibration kit for SDS electrophoresis, and the low molecular weight gel filtration calibration kit were from GE Healthcare. The substrates S-2251 (Val-Leu-Lysp-nitroanilide), S-2238 (Phe-Pip-Arg-p-nitroanilide), and S-2586 (MeO-Suc-Arg-Pro-Tyr-p-nitroanilide) were from Chromogenix AB (Mondal, Sweden).

EXPERIMENTAL PROCEDURES
Preparation of Hemocyte Lysate Supernatant-The spiny lobster hemolymph was obtained from the fourth walking leg coxa using sterile and precooled modified Alsever anticoagulant solution (AC) containing 27 mM sodium citrate, 115 mM glucose, 336 mM NaCl, 9 mM EDTA, pH 7 (26). The hemolymph was centrifuged immediately after collection at 700 ϫ g for 10 min at 4°C, and the supernatant was discarded. The hemocyte pellet was washed twice with cold AC, suspended in the corresponding lysis buffer (see particular experiments below), and disrupted by sonication three times at 40 watts for 10 s each. The clarified lobster hemocyte lysate (LHL) was obtained by centrifuging the homogenate at 4,000 ϫ g for 30 min at 4°C.
Determination of Trypsin-inhibitory Activity-Trypsin activity was determined using 0.9 mM BAPNA (1 ϫ K m ) as substrate (27). The nominal trypsin concentration was determined at 280 nm using an extinction coefficient (E 280 1% ) of 14.4 kDa (28), and 23.3 kDa molecular mass. A stock solution of bovine trypsin was prepared at 10 mg/ml in 1 mM HCl, 20 mM CaCl 2 , pH 3, whereas the stock solution of BAPNA (125 mM) was in DMSO. For the assay, 20 l of trypsin were mixed with 180 l of assay buffer (0.1 M Tris-HCl, pH 8, 150 mM NaCl, and 20 mM CaCl 2 ) in a well of a 96-well plate. The reaction was started by the addition of 50 l of 4.5 mM BAPNA in the assay buffer. The pNA released was measured kinetically at 405 nm for 10 min at 37°C in an ELx808 IU microplate reader (BioTek Instruments, Winooski, VT). Initial velocities were obtained using the kinetic application of the program KC4 version 3.4 (BioTek Instruments). One unit of trypsin activity was defined as the amount of trypsin causing the release of 1 mol of pNA/min. The extinction coefficient of p-nitroanilide at 405 nm for a volume of 250 l was 6.8 mM Ϫ1 , as determined empirically.
Trypsin-inhibitory activity was calculated at dilutions where the inhibitory percentage fell between 25 and 70%. One unit of inhibitory activity (IU) was defined as the amount of inhibitor producing 50% inhibition of 2 units of the trypsin (29). Specific activity was defined as the inhibitory activity/mg of protein.
Where indicated, the fractional activity (V i /V o ) was determined as the ratio between the initial velocity in the presence (V i ) and absence (V o ) of inhibitor.
Protein Concentration Determination-The protein concentration was determined by the Lowry method (30), using bovine serum albumin (BSA) as a standard. Samples containing HEPES above the interfering level with the Lowry assay were first precipitated with deoxycholate-trichloroacetic acid (31).
Influence of Ionic Strength on Inhibitory Activity-Hemocytes from four centrifuge tubes containing 50 ml of hemolymph/ anticoagulant (1:1, v/v) were collected and washed as above and then pooled to a final volume of 50 ml made up with AC. The homogeneous hemocyte suspension was divided into 6-ml aliquots and centrifuged. Finally, each pellet was suspended in 50 mM Tris-HCl, pH 7.5, containing various concentrations of NaCl (from 0 to 650 mM). The hemocytes were lysed and clarified as described earlier, and the inhibitory activity of trypsin for each fraction was determined. Where indicated, nucleic acids were precipitated by adding 0.1% protamine sulfate (final concentration) to lysed hemocytes before the clarification step.
Purification of Peptidase Inhibitor-The LHL was obtained in lysis buffer containing 450 mM NaCl and treated with protamine sulfate as above. The supernatant (10 ml at 9.5 mg/ml) was fractionated by gel filtration chromatography in a Sephadex G-50 Superfine column (2.6 ϫ 65 cm), equilibrated with 25 mM HEPES, pH 8.2, 100 mM NaCl, 0.01% Brij 35 (w/v) (buffer A). The flow rate was 0.8 ml/min, and fractions of 4 ml were collected for determining trypsin-inhibiting activity. The gel filtration column was calibrated with molecular mass standards (carbonic anhydrase (29 kDa), ribonuclease A (13.7 kDa), and aprotinin (6.5 kDa)). The pooled inhibitory fraction was applied to a 5-ml HiTrap SP HP column equilibrated with buffer A. The bound proteins were eluted with 135 ml of a linear NaCl gradient (100 -500 mM) in the same buffer at 0.5 ml/min. Protein elution was monitored at 280 nm. The inhibitory peak was further purified by reversed phase HPLC in a Knauer Smartline HPLC system (Germany), using a Discovery BIO Wide Pore C5 column (4.6 ϫ 250 mm, 5 m; Supelco) equilibrated with 0.1% (v/v) TFA in water (solvent A). The elution system comprised solvent A and 0.07% TFA in 70% acetonitrile (solvent B). Separation was performed with a linear gradient of solvent B from 5 to 80% over 55 min at a flow rate of 1 ml/min. The absorbance was monitored at 214 nm.
Determination of the Equilibrium Dissociation Constant (K i )-The concentration of active trypsin was determined by active site titration with 4-nitrophenyl 4Ј-guanidinobenzoate (33). The time to reach the equilibrium of trypsin-inhibitor complex was determined by incubating fixed concentrations of trypsin (48 g/ml) and inhibitor (0.9 g/ml) at room temperature for 0, 5, 10, 30, and 60 min before the addition of substrate. The active concentration of inhibitor was determined by titration against a fixed concentration of active site-titrated trypsin (1.5 M) assuming an equimolar binding between the enzyme and the inhibitor and E 0 /K i app Ն 100 (34). In addition, inhibitory activity was determined at different substrate concentrations (0.5, 1.0, 1.5, and 2.0 K m ) to demonstrate substrate-induced dissociation.
To determine the apparent dissociation constant (K i app ), trypsin (80 nM) was mixed with increasing inhibitor concentrations (2.4 -288 nM), and the residual trypsin activity was determined. The K i app value was calculated by fitting the experimental data to the quadratic Morrison equation for tight-binding inhibitors (35), using GraphPad Prism version 5 for Windows (GraphPad Software, San Diego, CA).
Determination of Inhibitor Specificity-The inhibitory activity was evaluated against peptidases belonging to three mechanistic classes: metallo, cysteine, and serine. The pNA released was determined at 405 nm, as described earlier for trypsin. In the case of carboxypeptidase A, the activity was determined kinetically at 254 nm for 3 min at room temperature in a reaction mixture composed of 85 l of 25 mM Tris-HCl, pH 7.5, 0.5 M NaCl (reaction buffer), 1 ml of 1 mM hippuryl-L-Phe, and 15 l of the enzyme at 2.2 M. All enzymatic activities were determined under initial velocity conditions. After 30 min of incubation of each enzyme with 100-fold molar concentrations of panulirin, the inhibitory activity was evaluated by determining the fractional activity using the assay conditions described above.
Disulfide Reduction and Carbamidomethylation-Panulirin was mixed with 13.3 l of a solution containing 0.75 M Tris, pH 8.0, 33.3 l of 6 M guanidinium chloride, and 4.5 l of 1.3 M DTT. The reaction was incubated for 2 h at 37°C. After incubation, 25 l of 0.8 M iodoacetamide was added and allowed to react for 25 min at room temperature. The reaction was stopped with 10 l of pure formic acid. This solution was passed through a ZipTip-C 4 microcolumn (Millipore) for desalting prior to enzymatic digestion. The protein was eluted with 2.5 l of 60% acetonitrile (v/v).
Enzymatic Treatment-Samples of the reduced and carbamidomethylated panulirin were separately enzymatically digested with trypsin, chymotrypsin, and endoproteinase Glu-C. The eluates from the ZipTip were dissolved in 60 l of 0.19 M Tris, pH 8.0, and the enzyme was added at 50:1 (trypsin) or 25:1 (chymotrypsin) substrate/enzyme ratios in each case. For the enzymatic treatment with Glu-C, the eluate was dissolved in 60 l of 0.25 M ammonium hydrogen carbonate, pH 8.0, and the endoproteinase was added in a 50:1 ratio.
Chemical Treatment-Reduced and carbamidomethylated panulirin (10 g) was desalted through ZipTip-C 4 (Millipore), dried in a SpeedVac, and resuspended in 30 mM hydrochloric acid containing 6 M guanidine hydrochloride. The solution was transferred to a 1-ml vial (Pierce) and kept under vacuum and at 104°C for 12 h.
Mass Spectrometry-ESI-MS and ESI-MS/MS spectra were acquired using a QTof-1 TM mass spectrometer (Micromass) fitted with a Z-spray nanoflow electrospray ion source operated at 80°C with a drying gas flow at 50 liters/h. The analyzer was calibrated in a wide mass range (50 -2,000 Da) using a reference mixture of sodium and cesium iodides. Intact protein and peptide digest samples were loaded into the borosilicate nanoflow tips and submitted to 900 and 35 V of capillary and cone voltage, respectively. To acquire the ESI-MS spectra, the first quadrupole was used to select the precursor ion within a window of approximately 3 Thomson. Argon gas was used in the collision chamber at ϳ3 ϫ 10 Ϫ2 pascal pressure, and collision energies between 15 and 48 eV were set to fragment precursor ions. Data acquisition and processing were performed using MassLynx version 3.5 (Micromass).
N-terminal Sequence-Panulirin and a tryptic digestion-derived peptide (1550.2 Da) were directly sequenced on a Shimadzu PPSQ-31A (Shimadzu, Kyoto, Japan) automated gas phase sequencer. Samples were dissolved in 10 l of a 37% CH 3 CN (v/v) solution and applied to TFA-treated glass fiber membranes, precycled with Polybrene (Aldrich). Data were recorded using the Shimadzu PPSQ-31A software.
Sequence Analysis-The physical-chemical properties of panulirin were determined using the ProtParam tool (available on the ExPASy Web site).
Influence of Panulirin on PO Response to LPS-PO activity was determined spectrophotometrically by recording the formation of dopaminechrome and derivatives from dopamine as substrate (36). In brief, 20 l of 0.25 mg/ml LHL or pooled fraction eluted at the void volume from the gel filtration chromatography of the LHL (F1) at 0.025 mg/ml was mixed in flat bottom microplate wells with 100 l of 50 mM Tris-HCl buffer, pH 7.5, 50 mM CaCl 2 , and 50 l of 0.1 mg/ml LPS. The control experiment was LPS-free water instead of LPS. PO activity was assessed continuously at 490 nm and 37°C immediately after the addition of 50 l of 0.6 mg/ml dopamine.
In order to evaluate the influence of panulirin on the PO response to LPS, 20 l of LHL fraction depleted of trypsininhibitory activity (F1) at 0.025 mg/ml was incubated for 15 min with 50 l of 2-fold serial dilutions of purified panulirin (16.2 IU/ml starting inhibitory activity), and PO activity was assessed as described above.

RESULTS
Influence of Ionic Strength on Trypsin-inhibitory Activity of the LHL-Preliminary observations led us to suspect that the extent of trypsin-inhibitory activity in the lysate could be related to the ionic strength in the lysis buffer used to prepare the LHL. Therefore, we evaluated the inhibitory activity of LHL obtained in lysis buffer containing different concentrations of NaCl (0 -650 mM). A direct relationship between trypsin-inhibitory activity in the LHL and NaCl concentration was found. This effect occurs for concentrations as high as 450 mM. Further increase in NaCl concentration did not affect the inhibitory activity (Fig. 1).
When an aliquot of LHL prepared in the lysis buffer without NaCl was treated with 0.1% protamine sulfate (w/v), a well known precipitant of nucleic acids, the trypsin-inhibitory activity increased from 7.4 to 38.60 IU/ml. In a different experiment, no significant differences (p Ͼ 0.05) were found between the inhibitory activity in protamine-treated (55.4 Ϯ 1.11 IU/ml; mean Ϯ S.E., n ϭ 5) and untreated (54.0 Ϯ 0.31 IU/ml; mean Ϯ S.E., n ϭ 5) LHL prepared in 450 mM NaCl, indicating that protamine sulfate did not impair the inhibitory activity under the experimental conditions used. Taken together, these results might suggest that trypsin inhibitors in the LHL are bound to nucleic acids, probably electrostatically, and that this binding is abrogated by high ionic strength. To confirm whether panulirin binds to nucleic acids, we isolated DNA from the hemocytes using TRI reagent according to the manufacturer's instructions. Panulirin (25 l at 8 M) was incubated with 25 l of 0.3 g/l DNA for 1 h at room temperature using buffers lacking NaCl. The control experiment was also incubated but with reaction buffer instead of DNA. Thereafter, the trypsin activity was determined as described above in a mixture containing 80 nM trypsin and 80 nM panulirin diluted from the above incubations. The inhibitory activity in the control experiment was 78%, and it dropped to 9.5% in the sample previously incubated with DNA. The corresponding DNA concentration in the absence of panulirin had no influence on trypsin activity (data not shown). These results suggest that panulirin is able to bind to nucleic acids at low ionic strength.
Isolation of Panulirin-Panulirin was purified by gel filtration on Sephadex G-50 followed by cation exchange chromatography on a HiTrap SP-Sepharose HP column (Table 1). Reversed zymography revealed a band with trypsin-inhibitory activity of around 16 kDa ( Fig. 2A). Hence, it was decided to fractionate the LHL first by gel filtration chromatography in Sephadex G-50 (30 kDa exclusion limit), presuming that the possible targets of inhibitors in the LHL are endogenous peptidases over 30 kDa that would elute at the void volume along with other components of the proPO-activating system. A homogenous suspension of washed hemocytes was evenly aliquoted, and each aliquot was lysed in 50 mM Tris-HCl, pH 7.5, containing various concentrations of NaCl (0 -650 mM). Trypsin-inhibitory activities were determined at dilutions producing 25-70% inhibition of 48 g/ml trypsin in the assay using 0.9 mM BAPNA as substrate. One unit of inhibitory activity (IU) was defined as the amount of inhibitor-containing sample producing 50% inhibition of 2 units of trypsin. Values represent the mean of at least three replicates plus S.D. (error bars).
A single peak of around 5 kDa showing trypsin-inhibitory activity eluted from the gel filtration between two major fractions, leading to a high degree of fractionation (Fig. 2B). We also found that precipitation of nucleic acids with protamine sulfate improved the resolution between the inhibitory peak and the fraction eluting at the void volume (not shown), which allowed us to fractionate higher volumes of LHL per chromatographic step. The inhibitory fraction (205-242 ml) was pooled and applied to cation exchange chromatography. Two trypsin-inhibiting peaks were identified (Fig. 2C). The second major peak with the strongest trypsin-inhibitory activity (66 -74 ml) was pooled and used throughout this study. Reversed phase HPLC analysis on a Supelco C 5 column (Fig. 2D) showed that this fraction, hereafter referred as panulirin, was around 95% pure. Isolated panulirin from reversed phase HPLC was used for N-terminal and MS analysis.
Characterization of Panulirin-Trypsin Interaction-The study was partially based on the strategies described by Bieth (34), using the steady state approach. Preliminary experiments showed concave inhibition curves in the plot between the fractional activity (V i /V o ) versus increasing concentration of panulirin at constant trypsin and substrate concentrations, which could mean that the incubation time for association of trypsin and panulirin was incomplete or that association was completed but the inhibition was reversible with [E o ]/K i Յ 10 (34). Therefore, the time dependence to reach the association between trypsin and panulirin was determined. The inhibitory activity remained constant because no significant differences (p Ͼ 0.05) were observed in the fractional activities among each incubation time evaluated (Fig. 3A), suggesting that completed association between trypsin and panulirin was reached upon mixing, and therefore, the concave inhibition curves observed probably describe a reversible interaction. It is worth mentioning that all progress curves in this experiment were linear (data not shown). On the other hand, we found that the fractional activity increased with substrate concentration at constant a Inhibitory activity: 1 unit of inhibitory activity (IU) was defined as the amount of protein needed to inhibit 2 units of trypsin activity. One unit of trypsin activity was defined as the enzyme activity that produces 1 mol of pNA/min under specified conditions. 6 ϫ 65 cm); 9.5 ml of LHL (10 mg/ml) treated with 0.1% protamine sulfate (w/v) was loaded onto the column previously equilibrated with 25 mM HEPES, 100 mM NaCl, pH 8.2, 0.01% Brij 35 (w/v) (buffer A) and eluted with the same buffer at 0.8 ml/min. Fractions of 4 ml were collected, and 10 l of each was assayed for inhibitory activity against 48 g/ml trypsin using 0.9 mM BAPNA. C, fractions from the gel filtration containing trypsin-inhibitory activity were combined and applied to a 5-ml HiTrap SP-Sepharose HP column equilibrated in buffer A. Bound proteins were eluted with 135 ml of a linear NaCl gradient (100 -500 mM) in buffer A. Fractions of 1 ml were collected during gradient elution and assayed for trypsin inhibition as above. D, the inhibitory peak from cation exchange was finally purified by reversed phase C 5 column (4.6 ϫ 250 mm) equilibrated with 0.1% (v/v) TFA in water. Bound samples were eluted with a linear gradient of acetonitrile from 3.5 to 56% over 55 min at 1 ml/min. The absorbance was monitored at 214 nm. trypsin and inhibitor concentrations (Fig. 3B), indicating substrate-dependent inhibition and thus confirming that the interaction is reversible and competitive (34). The active concentration of panulirin was determined by titration with a constant concentration of active site-titrated trypsin (1.5 M). At this trypsin concentration, the plot between the residual trypsin activity and inhibitor concentration tracks linear behavior (Fig. 3C), allowing the determination of an active inhibitor concentration of 22.5 M, which corresponded to 97% of the nominal concentration of purified panulirin.

Novel Trypsin Inhibitor Regulates Immune Response in Lobsters
Having the active concentration of trypsin and inhibitor, we proceeded to determine the inhibitor dissociation constant. For this purpose, constant concentrations of trypsin at 80 nM were mixed with increasing concentrations of inhibitor, and the residual trypsin activity was determined. The concave inhibition curve obtained (Fig. 3D) indicated experimental conditions of [E o ]/K i app between 1 and 10 (34), which further demonstrated the reversible interaction between panulirin and trypsin. The K i app value obtained by fitting the data to the Morrison equation (Fig. 3C), was converted to true K i value by the equation, K i app ϭ K i (1 ϩ [S]/K m ), taking into account the competitive nature of the inhibitor (35). The real K i was 8.6 Ϯ 0.81 nM, indicating that panulirin binds trypsin with relatively high affinity.
Inhibitory Specificity-It is widely accepted that most inhibitors are specific for one of the four mechanistic classes of peptidases (37,38), although a few inhibitors have shown a broader specificity, for instance, against serine and cysteine peptidases or against serine and metallopeptidases (37)(38)(39). Therefore, we assayed the inhibitory activity of panulirin against papain and carboxypeptidase A as representative of cysteine and metallopeptidases, respectively. Thereafter, the selectivity of panulirin was evaluated against a wider panel of serine peptidases, which included chymotrypsin, elastase, subtilisin, thrombin, and plasmin. In all cases, the inhibitory activities after a 30-min incubation of an at least 100-fold molar excess (active concentration) of panulirin over each peptidase was below 10% (not shown), thus indicating that panulirin did not inhibit the enzymes assayed.
Determination of the Primary Structure of Panulirin-Molecular weight determination by MS of untreated inhibitor revealed the presence of a main protein with a molecular mass of 5,367.1 Da. The first 19 cycles with no contaminating residues of the N-terminal Edman degradation were clearly SYKARSXTAYGYFXMIPPR, where X represents the cysteine residues. Furthermore, panulirin, when treated with DTT at pH 8.0, showed a shift of mass signal of 6 mass units, which suggested the existence of three disulfide bridges and six cysteines. This was later confirmed by peptide alkylation of the cysteines with carbamidomethyl groups, which added 6 times (342 Da) to the molecular mass of panulirin. Enzymatic digestions, after carbamidomethyl alkylation of the cysteines with trypsin or chymotrypsin, were performed. The enzymatic cleavages were mainly found in the residues Lys and Arg for trypsin and in the residues Trp, Phe, Ile/Leu, and with good frequency in Arg for chymotrypsin. The manual interpretation of such tryptic and chymotryptic MS/MS spectra revealed the amino acid sequences of the ion species [680. 30 (Fig. 4A). Additionally, the enzymatic digestion with endoproteinase Glu-C was incomplete, suggesting the absence of glutamic acid in the primary structure. Furthermore, the interpretation of the partial acid hydrolysis MS/MS spectra disclosed the amino acid sequences of ion species [769. 35, 4ϩ] and [625.26, 3ϩ] (Fig. 4A). Finally, the N-terminal sequence of a tryptic fragment with a molecular mass of 1,550.2 Da gave the amino acid sequence of the first 10 residues (ARGHIXXSSP) that help to reveal the presence of an Ile residue and to confirm the primary structure of panulirin (Fig. 4A). The complete amino acid sequence of panulirin has been deposited in the Swiss Protein database Uniprot with accession number B3EWX6.
Sequence Analysis-Sequence similarity search at NCBI databases using BLASTP with default parameters retrieved only three hits, which corresponded to hypothetical proteins from Aspergillus niger, with bit score of 32 and Expect (E) values greater than 6. Thereafter, we searched at the MEROPS database, which is devoted to peptidases and peptidase inhibitors (40). The hits retrieved showing closer sequence relationships with panulirin were unassigned peptidase inhibitor homologues belonging to the I63 family according to the MEROPS classification (39). However, no significant relationship was found for the hits retrieved because the lower E value obtained was 0.67. According to the MEROPS developers (41), to include a sequence in a family it must be related directly or indirectly (transitive relationship) to the type-example of the family in a statistically significant way (i.e. E value below 10 Ϫ10 ) in the alignment using BLAST on the MEROPS database. Therefore, these results suggest that panulirin represents a new family of peptidase inhibitors.
Sequence analysis using the ProtParam tool revealed that panulirin is a basic peptide with a theoretical pI of 9.5, aliphatic index of 42.7, and extinction coefficient of 11.8 mM Ϫ1 cm Ϫ1 , assuming that the three pairs of Cys residues form cystines.
Later, several sequences showing high identities (Ͼ60%) with panulirin were found at the expressed sequence tag database from the hemocytes of the spiny lobster Panulirus japonicus. An ORF identified in the PJ_EST01_03A01 mRNA coding for a putative P. japonicus trypsin inhibitor (PjTI1) was deposited as a third party annotation in GenBank TM with accession number KC154047. The N terminus of the translated sequence showed properties attributable to a signal peptide, as assessed by the SignalP program, with a predicted cleavage site located between positions 22 and 23 (VHG-DP). The TBLASTN 2.2.25ϩ homology comparison between panulirin and PjTI1 showed 65.9 as a maximal score value covering 93% of panulirin sequence (Fig. 4B).
Influence of Panulirin on PO Response-We studied first the PO response to LPS in the LHL. Surprisingly, it was found that, conversely to that described in other arthropods, the phenoloxidase activity increased slightly in the presence of LPS (Fig. 5A). 10-Fold or lower concentrations of LPS produced negligible activity compared with the control (not shown). Also in Fig. 5A, it can be observed that the progress curve of the reaction showed a lag phase, which is probably due to the cascade mechanism of the proPO system. The lag phase might represent the time required since the activation of the system  NOVEMBER 1, 2013 • VOLUME 288 • NUMBER 44 until the conversion of proPO into active PO, the final component of the cascade. This behavior rules out the possibility of LHL contamination as a possible cause of the small difference found between the response to LPS and LPS-free water.

Novel Trypsin Inhibitor Regulates Immune Response in Lobsters
We also tested the PO response to LPS in the F1 fraction, which is devoid of trypsin-inhibitory activity (see Fig. 2B). The PO response to LPS increased significantly in the F1 fraction compared with LHL under similar experimental conditions (Fig. 5B), suggesting that panulirin might be involved in regulating the phenoloxidase response to LPS. In addition, the phenoloxidase activity found in F1 confirmed our assumption that components of the proPO system responsible for recognizing microbial elicitors leading to melanization response in the LHL all have a molecular weight mass above 30 kDa.
However, it is conceivable that factors regulating the PO response other than panulirin (such as peptidase inhibitors, melanization inhibition proteins, or phenoloxidase inhibitors), although currently unknown in Panulirus argus, were also absent from F1, helping to explain the differences in PO activity. To ascertain whether the increment on PO response to LPS was due to the lack of panulirin, F1 was incubated with constant LPS concentrations and 2-fold dilutions of decreasing concentrations of panulirin for 15 min at room temperature before determining PO activity. It was found that PO response to LPS decreased in a dose-response fashion with increasing panulirin concentration, whereas controls without LPS remained similar for each inhibitor concentration (Fig. 6). This result suggests that panulirin is implicated in the regulation of peptidase(s) that are in the pathway toward the activation of proPO into PO and therefore involved in the regulation of the proPO-activating system in the spiny lobster.

DISCUSSION
Peptidases intervene in several immune response mechanisms of invertebrates, such as coagulation, melanization, activation of the Toll receptor, and complement-like reactions (42). Because the presence of peptidases in biological systems usually implies the occurrence of peptidase inhibitors to maintain homeostasis (43), peptidase inhibitors to control such processes are likely to occur (e.g. avoiding unnecessary activation of PO zymogen).
Peptidase inhibitors from the Kazal, serpin, Kunitz, ␣-macroglobulin, and pacifastin families have been identified in arthropods so far, and some are thought to be involved in immunity (43)(44)(45)(46). However, the exact physiological function of most of them remains unknown.
We have previously reported the presence of trypsin-inhibitory activity in the hemocytes of the spiny lobster P. argus (25). Recent evidence indicates the existence of genes encoding Kazal, Kunitz, and serpin type inhibitors in the hemocytes of a closely related species, the spiny lobster Panulirus japonicus (47). Therefore, it is conceivable to expect the occurrence of these inhibitors also in P. argus. In the current study, we present the finding and purification of a trypsin inhibitor from the hemocytes of P. argus, but surprisingly, it showed no sequence similarity with any known protein. Because it has been suggested that the finding of a new peptidase inhibitor with no sequence homology to any existing inhibitor family will lead to the building of a new family (41), we propose that this protein, named here panulirin, represents a new family of peptidase inhibitors.
In this sense, it is worth mentioning that the first systematic organization of peptidase inhibitors in families was accom-  plished for standard mechanism inhibitors of serine peptidases (48). This was mainly based on primary sequence homology, although the topology of disulfide bridges and their relationship to the reactive site were also considered (37,48,49). Currently, protein inhibitors of peptidases are organized in a hierarchical classification system that attempts to overcome some disadvantages of previous classification approaches (39 -41). The system is composed of three main levels: inhibitor unit, family, and clan. A family consists of protein sequences that are homologous, whereas membership of a clan is determined by similarities in protein tertiary structures, and hence all of the members of a clan will share a similar protein fold despite limited sequence similarity (39 -41). This classification system is implemented at the MEROPS database, which is now widely used.
A putative trypsin inhibitor from P. japonicus (PjTI1; GenBank TM accession number KC154047) showing high homology with panulirin was identified by reverse searching in an expressed sequence tag database. The high value in identity found (64%) between panulirin and PjTI1 supports our primary sequence elucidation results and strongly indicates the occurrence of genes encoding panulirin-like inhibitors in the Panulirus genus. Interestingly, both the signal peptide and the cleaving site predicted in PjTI1 are highly similar to those described in defensin-like peptides from P. japonicus (50) and P. argus (26).
Panulirin is constitutively expressed in the hemocytes of P. argus at ϳ3% of total protein. It is a non-glycosylated basic peptide composed of 48 amino acid residues containing six cysteine residues forming disulfide bridges. The small protein size resembles that of Kunitz (51) and pacifastin-inhibitory domains (52), whereas the presence of six cysteine residues forming disulfide bridges is a typical feature found across most inhibitory units of serine peptidase inhibitors, regardless of the family or the inhibitory mechanism (39,48,52,53).
Three different types of natural protein inhibitors of serine peptidases can be distinguished based on their mechanism of action: standard mechanism canonical inhibitors, non-canonical inhibitors, and serpins (38,54). We found that panulirin is a reversible, competitive, and tight-binding inhibitor of trypsin. Hence, it should be either a canonical or non-canonical inhibitor. The serpin possibility was ruled out because they are much larger proteins (350 -500 amino acid residues), which interact with the cognate peptidase as irreversible suicide substrates through a trapping mechanism (38,55,56). It has been stated that standard mechanism inhibitors are by far the most prevalent of the three classes of inhibitors of serine peptidases (54). In 2004, the MEROPS database grouped 48 families of peptidase inhibitors, of which 19 were serine peptidase inhibitors that obey the standard mechanism (39). Since then, around nine new families of inhibitors that probably act through such a mechanism have been added (41). On the other hand, the noncanonical inhibitors are much less abundant, and the few existing are solely known for thrombin and factor Xa (38,57). Taking the above together, it is reasonable to suggest that panulirin is a standard mechanism canonical inhibitor. However, confirming this assumption will require further studies. Standard mechanism canonical inhibitors interact with the cognate enzyme through an exposed binding loop of convex shape having similar or canonical conformation, which is complementary to the concave active site of the enzyme (37,38,49,53,54). The loop is made up of 6 -11 contiguous amino acid residues (the reactive site region) (54). Its central part contains the most exposed P1-P1Ј peptide bond (Schechter and Berger nomenclature (58)), called the reactive site, which is recognized by the peptidase in a substrate-like manner (37,38,55).
Panulirin did not inhibit papain and carboxypeptidase A, which were studied as type examples of cysteine peptidases and metallopeptidases, respectively. It is widely accepted that inhibitors of serine peptidase rarely exceed this mechanistic class (34,(37)(38)(39). On the other hand, selectivity of serine peptidase inhibitors against different serine peptidases is usually broader (37). In this sense, it was found that panulirin did not inhibit typical non-trypsin-like serine peptidases such as chymotrypsin, elastase, and subtilisin. This result was not completely unexpected because the reactive site (P1 residue) of serine peptidases inhibitors is usually the primary specificity determinant, being Arg or Lys in most trypsin-like inhibitors (53). However, panulirin did not inhibit trypsin-like peptidases, such as thrombin and plasmin, indicating a notable selectivity of panulirin among these closely related enzymes. In the case of canonical inhibitors, this ability could be explained by the negative influence of the inhibitor scaffolding (i.e. the remainder of the inhibitor molecule other than the reactive site region, which is known may influence the inhibitory specificity toward related peptidases) (54).
Because the panulirin sequence presents two Lys and five Arg residues, the reactive site (P1) position cannot be predicted with certainty without structural or experimental evidence. However, at this moment, some speculation is tempting at least for suggesting the most likely candidates. Although it is known that the reactive site region is the most variable region in the primary sequence of serine peptidase inhibitors, even among those belonging to the same family (53,59), several amino acid residues at certain positions around the reactive site are conserved in one or more families (38,48,60,61). Thus, we were encouraged to conduct a visual inspection of those conserved residues along the panulirin sequence. Considering that panulirin is probably a canonical inhibitor, the Lys 3 and Arg 5 residues from the N terminus and Lys 47 from the C terminus were discarded from the analysis because their location near the center of a loop is improbable. This first approach reduced the putative P1 position to Arg 19 , Arg 21 , Arg 31 , and Arg 33 residues.
The BPTI, antistasin, elafin, arrowhead, hirustasin, and chelonianin families present Cys at P2 (38,60), which occurs in the putative P1 Arg 21 and Arg 31 of panulirin. It is also known that Ala and Gly residues are very well conserved in P1Ј of sequences homologous to BPTI (62), being more likely Ala (61). Three of the putative P1 position residues (Arg 21 , Arg 31 , and Arg 33 ) matched this sequential motif in the panulirin sequence. Considering both observations, it was noticeable that putative P1 sites, Arg 21 and Arg 31 , have P2 and P1Ј amino acid residues that are conserved in the BPTI family. Surprisingly, we also found a marked similarity between the regions Cys-Lys-Ala-Arg (P2P1P1ЈP2Ј) of BPTI and the panulirin sequence fragment Cys 30 -Arg 31 -Ala 32 -Arg 33 , with a Lys 3 Arg 31 isofunctional substitution.
Considering the apparent similarity described above and that residues P3-P3Ј better describe the length and convex shape of the loop in canonical inhibitors (60), we selected for further analysis the Arg 31 residue in panulirin sequence as the putative P1 and thus the Trp-Cys-Arg-Ala-Arg-Gly fragment as the putative P3-P3Ј reactive site region. It was found that this fragment had no significant homology to any inhibitory unit because no hits were retrieved using BLAST at the MEROPS database, even at E values as high as 100. Similar results were obtained with larger sequence fragments (P7-P7Ј). This result could be explained by the hypervariability in primary sequence of the reactive site regions (59,63,64). Furthermore, this fragment lacks Pro at P3, which, along with Cys at P2, is a distinctive combination found in the BPTI family (60). Therefore, the similarity found could be rather fortuitous but might deserve future consideration. On the other hand, the reactive site region of BPTI is located toward the N terminus, whereas the sequence fragment analyzed in panulirin tends more to the C terminus.
A few other specific residues conserved in other families were also found. For instance, the putative P1 position Arg 21 has Pro in P3, which is conserved among soybean trypsin inhibitors (Kunitz) (38,60), or the presence of Ser at P1Ј of the same Arg 21 that, along with Pro at P3Ј, absent in Arg 21 , are conserved residues in the first and second domains of Bowman-Birk inhibitors (60,61). However, these minor similarities are divergent and less relevant at this moment.
The presence of conserved residues in the reactive region has usually served as a family signature and has helped to classify standard mechanism inhibitors within known families (54). We found that the P1 site could not be deduced with acceptable confidence through comparison with other families, thus establishing similarities with them, or vice versa. The main reason could be that panulirin represents a new family that is still only partially characterized. Once the P1 site becomes determined experimentally, the analysis of the reactive region will allow proper identification of the conserved residues, if any, shared with other inhibitor families. In addition, the expected appearance of new protein sequences encoding panulirin-like inhibitors will help to identify conserved residues within this family.
Finally, a panulirin model was constructed to assess its putative structure (Fig. 7). The core of 34 residues (the inner length between the cysteines at the ends plus the two adjacent serine residues; see underlined sequence of panulirin in Fig. 4B) of panulirin was modeled using the ESyPred3D Web server based on the solved structure of the bovine neutrophil ␤-defensin 12 (Protein Data Bank code 1bnb). This template shares 30.8% identities with panulirin using the ALIGN program (65). The model was refined with the program SSBond (66) to predict the disulfide bonds, which were Cys 7 -Cys 37 , Cys 14 -Cys 30 , and Cys 20 -Cys 38 . This disulfide pattern has been observed in ␤-defensins and other defensin-like peptides (i.e. antimicrobial peptide tachystatin A from horseshoe crab, defensin-like peptides from platypus, and ␤-defensins 2 and 3 from humans), which are part of the innate defensive system of animals. Panulirin probably folds in a core of 31 residues, leaving two tails of 6 and 10 residues at the N and C termini. The model showed that side chains from Arg 21 and Arg 31 residues are solvent-exposed, whereas in Arg 33 , the side chain is toward the protein core and hence probably not accessible to the protease reactive site.
On the other hand, the strong positive charge of panulirin could explain several experimental findings of this study. First, it could be the cause of the discrepancy between the molecular mass found by gel filtration and ESI-MS (5 kDa) and that found by reversed zymography (16 kDa). In other experiments using SDS-PAGE, we found molecular masses of around 12 kDa (not shown). This difference could be due to the well known anomalous behavior showing basic proteins on SDS-PAGE-based electrophoresis.
In addition, such a high positive charge seems to be responsible for panulirin interaction with nucleic acids in the LHL, causing the trypsin-inhibitory activity to go unnoticed or be underestimated if the ionic strength in the lysis buffer is unsuitable. We have even found that the peak of inhibitor at 280 nm in the gel filtration chromatography ( Fig. 2A) disappears in similar chromatographic conditions if the LHL is prepared in a buffer lacking NaCl (not shown). The above findings suggest that the electrostatic interaction between panulirin and the nucleic acids in the LHL could be eventually stronger than its affinity for trypsin, even when our results indicated that panulirin bound to trypsin with relatively high affinity. Previous studies have shown that nucleic acids bind to cationic peptides and that the salt concentration influences such interaction (67,68). However, the in vitro panulirin-nucleic acid interaction observed in this work probably has no repercussions for the biological role of panulirin.
The major contributions to knowledge of the proPO system of crustaceans have been made by the group of Söderhäll in the FIGURE 7. Schematic three-dimensional representation of panulirin. The structure backbone was achieved by homology modeling. The core of 34 residues (the inner length between the cysteines at the ends) shows two ␤-strands stabilized by three disulfide bridges in the arrangement Cys 1 -Cys 5 , Cys 2 -Cys 4 , and Cys 3 -Cys 6 . The residues Arg 21 , Arg 31 , and Arg 33 are shown. The model structure was obtained using ESyPred3D and drawn with PyMOL. fresh water crayfish P. leniusculus, and therefore it has been widely used for building a general model of melanization cascade in crustaceans (2)(3)(4), where the regulatory role of pacifastin on the proPO system is depicted. Recently, cDNAs encoding pacifastin-related peptides have been described in the Chinese mitten crab Eriocheir sinensis (69) and in the swimming crab Portunus trituberculatus (70). They also seem to be implicated in the immune response because they are up-regulated upon microbial challenge, although their role in the proPO-activating system remains to be established. On the other hand, members of the pacifastin family are widely found among insects (45,71), but interestingly, they are not involved in the regulation of the melanization cascade (72).
Our knowledge about molecular components of the humoral immune response in the spiny lobster is still rather limited. In the current study, we have found that PO response to LPS is lower than that reported in other crustaceans like shrimps (73), black tiger prawns (74), and crayfish (75). The substantial increase in sensitivity of PO response to LPS in the LHL fraction depleted of trypsin-inhibitory activity and other molecules below 30 kDa obtained by gel filtration chromatography (F1) indicated that LPS considerably activate the proPO system of P. argus, but such a response is tightly regulated. Our results also demonstrated that panulirin is very involved in the regulation of PO response (see Fig. 6), although we cannot rule out the presence of other regulatory factors acting on phenoloxidase, peptidases, or the melanization reaction.
The in vivo significance of such a tight regulation of the melanization cascade in the LHL of spiny lobster is yet unknown and suggests the need for further studies, especially if we consider the well known importance of melanization response to host defense. One conceivable cause could be related to hemocyanin-derived phenoloxidase activity. It is known that hemocyanin from P. argus exerts phenoloxidase activity when it is partially hydrolyzed by trypsin (76). Furthermore, it has even been suggested that hemocyanin-derived phenoloxidase activity may be involved in host defense (77). Considering the aforementioned, it is conceivable that ppA or any other serine peptidase released from the hemocytes to the plasma in response to a microbial stimulus is capable of activating hemocyanin into phenoloxidase, thus contributing to the overall melanization reaction. In this probable scenario, tighter control of peptidase activity may be required to avoid excessive melanization and its potential deleterious effects to the host.
The ppA in P. argus seems to be a calcium ion-dependent trypsin-like serine peptidase (25). At this time, we suggest that panulirin may inhibit this enzyme. However, further studies are required to ascertain whether panulirin inhibits the calciumdependent ppA or other serine peptidase, if any, upstream of the cascade. It is worth mentioning that, unlike in insects, serine peptidase upstream of ppA in the ProPO cascade has not yet been identified in crustaceans, although its presence has been inferred in P. leniusculus (4). Once purified ppA or any other serine peptidase involved in the melanization response in P. argus becomes available, it will be possible to evaluate the relative affinity with panulirin and to identify where and how panulirin regulates the proPO activation pathway.
Because ppA in crustaceans are trypsin-like serine peptidases, commercial trypsin from mammals has been widely used to activate proPO in vitro in several investigations pursuing the isolation and characterization of proPO (78 -81). We have found previously that higher amounts of trypsin are required to activate prophenoloxidase in P. argus (25). Considering the significant presence of panulirin in the LHL, it is likely that this behavior was mostly due to the inhibitory activity of panulirin, which impose a higher concentration of trypsin to accomplish complete proPO activation.
Since the discovery of pacifastin more than 20 years ago (23,24) and up to the present report, the regulatory role of peptidase inhibitors in the proPO-activating system of crustaceans has not been proven for inhibitors other than pacifastin, although serpins have been recently suggested to also regulate the proPO cascade in crustaceans (82). However, until now, the role of serpins in regulating melanization cascade has been truly demonstrated only in insects (16). We have described here for first time how panulirin, a novel peptidase inhibitor, is involved in the regulation of the proPO cascade in lobsters. The results of this study, together with the recent discovery of novel genes encoding defensin-like antimicrobial peptides in lobsters (26,50), could indicate that Panulirus can be an attractive genus for deepening understanding of the immune system in crustaceans.