Serine protease inhibition by insect peptides containing a cysteine knot and a triple-stranded beta-sheet.

Three insect peptides showing high sequence similarity and belonging to the same structural family incorporating a cysteine knot and a short three-stranded antiparallel β-sheet were studied. Their inhibitory effect on two serine proteases (bovine α-chymotrypsin and human leukocyte elastase) is reported. One of them, PMP-C, is a strong α-chymotrypsin inhibitor (Ki = 0.2 nM) and interacts with leukocyte elastase with a Ki of 0.12 μM. The other two peptides, PMP-D2 and HI, interact only weakly with α-chymotrypsin and do not inhibit leukocyte elastase. Synthetic variants of these peptides were prepared by solid-phase synthesis, and their action toward serine proteases was evaluated. This enabled us to locate the P1 residues within the reactive sites (Leu-30 for PMP-C and Arg-29 for PMP-D2 and HI), and, interestingly, variants of PMP-D2 and HI were converted into powerful inhibitors of both α-chymotrypsin and leukocyte elastase, the most potent elastase inhibitor obtained in this study having a Ki of 3 nM.

In the last decade, naturally occurring serine protease inhibitors (1) have been the focus of many studies, mainly for two reasons: first, the target proteins control functions in a variety of fundamental proteolytic processes in humans and mammals (blood clotting, digestion, inflammation, fibrinolysis), in invertebrates such as insects (immune system, digestion, protection against their predators) or worms (protection against their host), and plants (protection against insect attack); second, low molecular weight inhibitors of serine proteases have been attractive tools for studying the general aspects of protein conformation and protein-protein interactions (2). In the present study, we report the inhibitory properties of three homologous peptides (primary and tertiary structures) toward ␣-chymotrypsin, trypsin, and human leukocyte elastase.
We have previously isolated two peptides, PMP-C and PMP-D2, from the brain and the fat body of the insect Locusta migratoria (3). These peptides are composed of 36 and 35 residues, respectively, and are cross-linked by three disulfide bonds. The Thr-9 of PMP-C has an uncommon O-glycosidic linkage to a single fucose moiety. There is 40% strict identity between PMP-C and PMP-D2 with conservation of the Cys positions (3). Moreover, they are located on the same peptidic precursor, and, by Northern blot analysis, it has been shown that the gene encoding this precursor is mainly transcribed in the fat body (4). In this paper, we describe the isolation and characterization of HI, a novel locust peptide.
Because the isolation from insect extracts is time-consuming and yields only small amounts of peptides, we have prepared, at a reasonable scale (5-10 mg), by solid-phase synthesis, PMP-D2 (5) and PMP-C with and without the fucose moiety. 1 Although they are small peptides with a high disulfide content, no sequence similarities could be found when comparing them with small toxins or small protease inhibitors. However, the milligram quantities of PMP-D2 obtained by solid-phase synthesis enabled us to study its tertiary structure by two-dimensional nuclear magnetic resonance, which showed interesting similarities with the tertiary fold of both -conotoxin GVIA, a calcium channel blocker, and the Ascaris chymotrypsin/elastase inhibitor (6). This prompted us to evaluate their protease inhibitory activity.
In the present paper, we report on the inhibitory activity of PMP-C, PMP-D2, and HI toward serine proteases (bovine ␣-chymotrypsin, human leukocyte elastase, and porcine trypsin) using the natural and synthetic peptides. Since the P1 residue (7) within the reactive site of serine protease inhibitors determines the specificity for the cognate enzyme, mutational or synthetic changes of the P1 residue and/or replacement of active site residues should greatly influence both the specificity and the potency of the inhibition. For that reason, we have prepared by solid-phase synthesis variants 2 of PMP-C, PMP-D2, and HI, in which one or two residues within the reactive site have been changed, and we have evaluated their inhibitory properties toward HLE, 3 ␣-chymotrypsin and trypsin.

Materials Chemicals
The acetonitrile, trifluoroacetic acid of HPLC quality, methyl tertbutyl ether, and dimethylformamide (DMF) were from SDS (Seltz, France). The N-ethyldiisopropylamine (DIEA) was obtained from Merck (Darmstadt, Germany). The Wang (p-benzyloxybenzyl alcohol) resins came from Novabiochem (Meudon, France), and N ␣ -Fmoc amino acids derivatives were purchased from Millipore (St. Quentin Yvelines, * This work was supported in part by EEC Contract BIO2CT930073. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 2 We prefer using "variant" instead of "mutant" (which refers to a mutational change) since they are obtained by solid-phase synthesis. 3 The abbreviations used are: HLE, human leukocyte elastase; RP-HPLC, reverse phase-high performance liquid chromatography; Fmoc, N-(9-fluorenyl)methoxycarbonyl; DMF, dimethylformamide; DIEA, ethyldiisopropylamine; pNA, para-nitroanilide. It should be mentioned that PMP-C, PMP-D2, and HI are given names of the peptides, rather than abbreviations. France) and Neosystem Laboratoire (Strasbourg, France). The BOP reagent (benzotriazolyl-N-oxy-tris-(dimethylamino)phosphonium hexafluorophosphate) and trifluoroacetic acid were from Neosystem Laboratoire.

Methods
Isolation of the Natural Peptides (PMP-D2, PMP-C, and HI) from the Hemolymph of the Insect L. migratoria and Primary Structure Determination of HI The hemolymph of fifth instar larvae (1800) of the African locust L. migratoria (kindly supplied by the Laboratoire de Biologie Générale de l'Université Louis Pasteur, Strasbourg, France) was collected directly into Eppendorf tubes and was centrifuged at 900 ϫ g for approximately 30 s at 4°C to separate the hemocytes from the plasma and was kept on ice. Plasma samples were pooled (70 ml) and applied (3 ml per cartridge) onto C18 Sep-Pak cartridges (Waters). The elution was performed with increasing concentrations of acetonitrile in water (18,36, and 60%). The 36% acetonitrile fractions were purified by C18 RP-HPLC using an analytical column (0.46 ϫ 25 cm) packed with 5-mwide pore (30 nm) Vydac particles. A linear gradient from 0 to 60% acetonitrile in 0.1% trifluoroacetic acid over 60 min, at a flow rate of 1 ml/min was used. The peaks were collected manually and, if necessary, repurified on the same column with a shallower gradient (a stepwise gradient from 5 to 11% (or 15%) acetonitrile over 10 min and from 11% (or 15%) to 35% (or 39%) acetonitrile over 40 min). They were estimated to be 95% pure on analytical RP-HPLC, and the amount of natural PMP-D2, PMP-C, and HI obtained from 70 ml of hemolymph was 1 mg, 1 mg, and 60 g, respectively.
The authenticity of the PMP-C and PMP-D2 was assessed by electrospray mass spectrometry on a VG BioTech BIO-Q mass spectrometer and by coelution with PMP-C and PMP-D2 obtained after isolation from the brain and the fat body of the same insect (3).
The peptide referred to as HI was reduced and alkylated by 4-vinylpyridine (as described in Ref. 3) and was then subjected to automated Edman degradation on an Applied Biosystems Sequencer, model 471A in the liquid pulse mode, which yielded a 35-amino acid sequence (Fig. 1). The molecular mass of HI, as determined by electrospray mass spectrometry, was 3716.43 Ϯ 0.18 Da, which is in excellent agreement with the sequencing data (deduced mass M ϭ 3722.28 Da minus 6 for 3 disulfide bridges).

Solid-phase Peptide Synthesis
The synthesis of PMP-D2 (5) and PMP-C 1 was performed as described elsewhere. HI and all the synthetic variants of PMP-C and PMP-D2 were synthesized manually, via Fmoc chemistry, starting with approximately 0.25 mmol of Wang (p-benzyloxybenzyl alcohol) resins. Amino acids were coupled as N ␣ -Fmoc derivatives with the following side chain protections: trityl for Asn, Cys, and Gln; t-butyl for Asp, Glu, Ser, and Thr; t-butoxycarbonyl for Lys, and Trp; 2,2,5,7,8-pentamethyl chroman-6-sulfonyl for Arg; and Met was coupled unprotected. The coupling reactions were achieved in DMF using 3 eq of BOP to activate the carboxyl function, 3 eq of protected amino acid, and 9 eq of DIEA. The completion of the couplings was assessed by a qualitative ninhydrin test. If necessary, double couplings were achieved with 5 eq of amino acids. The removal of the Fmoc protecting group was achieved with 25% piperidine in DMF. The side chain deprotection and the cleavage of the peptide from the resin (100 mg) was done with the standard trifluoroacetic acid procedure: trifluoroacetic acid/ethanedithiol/thioanisol/H 2 O/phenol (10 ml/0.25 ml/0.5 ml/0.5 ml/0.75 g) at room temperature for 2 h. The trifluoroacetic acid solution was vacuumfiltered into cold methyl tert-butyl ether (30 ml). The methyl tert-butyl ether suspension was centrifuged for 5 min at 2000 ϫ g at room temperature. The supernatant was discarded, and the precipitate was resuspended in methyl tert-butyl ether (30 ml). This procedure was repeated twice, and the precipitate was finally dissolved in 10% acetic acid and lyophilized.

Disulfide Bond Formation and Purification of the Synthetic Peptides
The crude peptides (1 mg/ml) were air-oxidized in water (pH 8 -9 adjusted with DIEA) without preliminary purification. The completion of oxidation (usually 18 h) was checked by HPLC, and the reaction was acidified with trifluoroacetic acid before application on a semi-preparative C18 RP-HPLC column (2.2 ϫ 25 cm) packed with 10-m-wide pore (300 nm) Vydac particles. Elution was performed at a flow rate of 10 ml/min with a linear gradient of 0 -60% acetonitrile in 0.1% trifluoroacetic acid over 60 min. The collected fractions were pooled, concentrated by lyophilization, and repurified using a shallower gradient (a stepwise gradient from 5-11% (or 15%) acetonitrile over 10 min and from 11% (or 15%) to 35% (or 39%) acetonitrile over 40 min).
Peptide purity was estimated to be 95% by analytical RP-HPLC using a stepwise gradient as for the purification of HI. Overall yields, based on resin, were usually between 2 and 3%. The authenticity of the synthetic peptides was checked by electrospray mass spectrometry. Table I shows the deduced mass from Edman degradation and the measured mass of the different variants (the 6-Da difference between the mass deduced from amino acid sequence and that measured by electrospray mass spectrometry is due to the involvement of the 6 Cys in 3 disulfide bonds).

Determination of the Net Peptide Content in Peptide Inhibitors
Peptide concentrations were deduced from the molecular mass and the extinction coefficient, a parameter measured as described by Van Iersel et al. (12).

Serine Protease Inhibition
Protease concentrations mentioned in this article refer to the concentration of active enzymes. The Michaelis constants K m were determined using standard procedures. Unless otherwise stated, all kinetic experiments were performed at 25°C in 50 mM Tris containing 20 mM CaCl 2 , pH 8.0, a solution that will be referred to as "the buffer" throughout the text.
Measurement of Equilibrium Dissociation Constant K i -K i was calculated from equilibrium titration experiments. 990-l reaction mixtures containing constant amounts of enzyme and increasing amounts of inhibitor in the buffer were incubated at 25°C for 15 min, an incubation time that was sufficient to ensure maximum enzyme inhibition under our experimental conditions. Reactions were started by the addition of 10 l of substrate stock solution, and the release of p-nitroanilide was followed at 410 nm using a Uvicon 941 spectrophotometer (Kontron) and recorded until the rate of substrate breakdown remained constant.
Trypsin inhibition was assessed by reacting 4 M enzyme with 16 M peptide for 30 min at 25°C and measuring the enzyme activity with 1 mM benzoyl-Arg-pNA.
Measurement of Association (k ass ) and Dissociation (k diss ) Rate Constants-Kinetics of HLE and chymotrypsin inhibition by the native and synthetic peptides were measured using the progress curves method (13). Enzyme was dropped into a spectrophotometer cuvette containing substrate and inhibitor in the buffer. Pseudo-first order conditions, [I] 0 Ն 10 ϫ [E] 0 ([I] 0 is the initial inhibitor concentration and [E] 0 the initial enzyme concentration), were used. The release of p-nitroanilide was followed at 410 nm using a Cary 2200 spectrophotometer (Varian) on line with an IBM PS2 Model 30 microcomputer. Nonlinear regression analysis of progress curves was done using the Enzfitter software (Biosoft, Cambridge, UK).

Isolation and Primary Structure Determination of a Novel
Peptide, HI, from the Hemolymph of L. migratoria-While purifying from the hemolymph PMP-C and PMP-D2 as reference compounds, a novel peptide, named HI, was isolated. The sequencing of HI, carried out by automated Edman degradation, showed a 72% strict identity with PMP-D2 (Fig. 1). The molecular mass, measured by electrospray mass spectrometry, was 3716.43 Da. This value differs by 6 Da from the data obtained by Edman degradation (3722.28 Da), thus suggesting that the 6 Cys are all involved in 3 disulfide bonds. Moreover, because of its high homology with PMP-D2, it is likely that the disulfide pairing is the same as in PMP-D2 (6).
Protease Inhibition by Natural and Synthetic PMP-C, PMP-D2, and HI-Preliminary experiments have shown that only PMP-C inhibits HLE. By contrast, chymotrypsin was found to interact with the 3 peptides. Interestingly, none of the peptides was found to inhibit porcine trypsin in our experimental conditions. Fig. 2 shows the effect of increasing quantities of synthetic PMP-D2 on constant quantities of chymotrypsin. Substrate was added to equilibrium mixtures of enzyme and inhibitor. The release of p-nitroanilide versus time was stable after 20 -30 s indicating that E, I, S (enzyme, inhibitor and substrate, respectively), and their complexes have reached their equilibrium. Calculation of the best estimate of the substrate-dependent equilibrium constant K i(app) was performed by nonlinear regression analysis of the experimental data based on Equation 1 (13): where a, the enzymic fractional activity, is the ratio of the velocity in the presence of inhibitor to that in its absence. The true K i (Table II) was deduced from K i(app) using the following relationship: where[S] 0 is the initial substrate concentration. The K i value for natural PMP-D2 was shown to be identical with that of the synthetic peptide (Table II). Equilibrium dissociation constants governing the interactions between chymotrypsin and synthetic HI and between HLE and PMP-C (natural and synthetic nonfucosylated) were determined using similar equilibrium titration experiments (Table II).
Since PMP-C (natural and synthetic nonfucosylated) binds chymotrypsin very tightly, K i was obtained through k ass and k diss . These parameters were determined using the progress curves method (13,14). A typical curve illustrating chymotrypsin inhibition by nonfucosylated PMP-C is shown in the inset of Fig. 3A. The curve is biphasic, i.e. the pre-steady state release of product is followed by a steady state, confirming that PMP-C reversibly interacts with chymotrypsin. Since no significant decrease of the initial substrate concentration occurred during the progress of the reaction and, since I 0 Ն 10 ϫ E 0 , product accumulation versus time can be described by the following equation: where P is the product concentration, v z is the rate of substrate hydrolysis at t ϭ 0, v s is the steady state velocity. k, the apparent first order rate constant governing the pre-steady state was calculated from the experimental data by nonlinear regression analysis using Equation 2. Chymotrypsin inhibition was studied using synthetic nonfucosylated PMP-C concentrations varying from 0.14 to 0.72 M. Fig. 3A shows the effect of [I] 0 , the initial inhibitor concentration on the apparent rate constant k; the linear increase of k strongly suggests that no reaction intermediate accumulates (within the range of inhibitor concentrations used) and that E and I interact according to a simple bimolecular and reversible mechanism. Fig. 3B shows the effect of the initial substrate concentration on k; the linear increase of k with 1/F (F ϭ 1 ϩ [S] 0 /K m ) indicates that inhibitor and substrate compete for the binding to the enzyme. Hence, k and I 0 are related as follows (15): where K m is the Michaelis constant. k ass , the second order rate constant was calculated from the slope of the linear curve shown in Fig. 3A using K m ϭ 23 M. k diss , the first order dissociation rate constant, is given by the intercept of the curve with the ordinate. The values of k ass , k diss , and K i (K i ϭ k diss / k ass ) of fucosylated and nonfucosylated PMP-C are shown in Table II; these kinetic constants are similar, indicating that the fucose moiety does not affect chymotrypsin inhibition. The inhibition properties (toward chymotrypsin and HLE) of the variants of PMP-C, PMP-D2, and HI were examined using similar methods, and the results are given in Table II.
Determination of the P1 Amino Acid by Synthetic Replacement of the Amino Acid at the Presumed Site-It is well documented that ␣-chymotrypsin inhibitors usually have bulky and aromatic amino acid residues such as Tyr, Phe, Leu, or Met as their P1 residue, while elastases have moderately large hydro- FIG. 1. A, amino acid sequence of the peptide named HI and comparison with the sequence of PMP-D2 (72% identity); B, sequence comparison between PMP-C and PMP-D2 (a gap is introduced to maximize the homology between the two sequences). phobic residues such as Leu, Met, Ala, Val as P1, but never Phe (1). Therefore, we hypothesized that the most likely P1 residue in PMP-C is Leu-30, which could account for its inhibitory effect toward both ␣-chymotrypsin and HLE. Leu-30 was in-deed the most probable location for the P1 amino acid, since it is the only Leu residue in the sequence of PMP-C (Fig. 1) and since it is located in an exposed and accessible region of the peptide, immediately after the third strand of the ␤-sheet. To test this hypothesis, we prepared a synthetic variant of PMP-C in which Leu-30 was replaced by a Val residue. The choice of Val was based on the following considerations. Val has never been reported as the P1 amino acid in chymotrypsin inhibitors; the change of Leu to Val should has little if any effect on the tertiary fold of PMP-C; small peptidic inhibitors are supposed to react in a substrate-like manner, and Val is a P1 residue found in the substrates of HLE (16).
As shown in Table II, the L30V variant lost most of its ability to inhibit ␣-chymotrypsin. However, it retained its weak activity toward HLE (Table II). The spectacular change in the inhibition of ␣-chymotrypsin observed with the L30V variant indicated that it was possible to lose specifically the inhibitory activity toward chymotrypsin, while retaining the anti-elastase property. Therefore, we concluded that the changed amino acid Leu was the P1 residue, the decrease in affinity being due to an inappropriate P1 residue and not to the misfolding of the PMP-C variant.

Confirmation of the Determined P1 Residue by Conversion of PMP-D2 and HI into Powerful ␣-Chymotrypsin Inhibitors-
The comparison of PMP-C with PMP-D2 and HI sequences enabled us to locate the reactive site of PMP-D2 and HI by simply superimposing them: C 28 TLKAC 33 , PMP-C; C 27 TRKGC 32 , PMP-D2; C 27 TRKAC 32 , HI.
Thus, the most probable reactive sites P1-PЈ1 in PMP-D2 and HI are Arg-29-Lys-30. To confirm this hypothesis, the presumed P1 residues (Arg) were replaced by Leu in PMP-D2 and HI in order to have the same reactive site Leu-Lys as in PMP-C. The potency of both R29L variants toward ␣-chymotrypsin inhibition was increased significantly, and they were both converted into HLE inhibitors. The PMP-D2 variant was found to be the strongest inhibitor of HLE obtained in this study (Table II).
When comparing the amino acids within the reactive site P3-PЈ3 of PMP-C and the R29L variants of HI and PMP-D2, one can notice the strict identity of the recognition site between PMP-C and the HI variant (CTLKAC), whereas PMP-D2 variant has a Gly in the PЈ2 position (CTLKGC) instead of Ala; these results are indicative of an Ala in position PЈ2 which may be preferable to Gly for ␣-chymotrypsin inhibition, while the contrary is true for elastase inhibition.
Design of a Better Elastase Inhibitor-Several elastase inhibitors have been reported to have a Met residue at PЈ1; for instance, mucous proteinase inhibitor (17) and Ascaris chymo-  trypsin/elastase inhibitor (18) have Leu-Met as P1-PЈ1 reactive site. Therefore, considering the strongest elastase inhibitor obtained in this study (PMP-D2 variant R29L), a double variant R29L/K30M (with Met as PЈ1) was designed in order to improve the binding to HLE. Indeed, the variant R29L/K30 M is 4-fold stronger than the previous one with a K i of 3 nM, confirming the importance of the Met in position PЈ1.
To investigate the role of the residues beyond the P1-PЈ1 bond, we have prepared a double variant of PMP-C: K31M/ A32G (which has the same P3-PЈ3, CTLMGC, as the most powerful elastase inhibitor R29L/K30M variant of PMP-D2) and examined its inhibitory activity; this variant is still 7-to 8-fold weaker HLE inhibitor than the double variant of PMP-D2 (a Met as PЈ1, in this case, does not increase the potency toward HLE). In contrast, the double variant of PMP-C is more effective toward ␣-chymotrypsin than the PMP-D2 double variant. Thus, the residues beyond the P1-PЈ1 bond seem to have an effect on the specificity toward proteases, and variants of PMP-D2 seem to be more specific toward elastases than the variants of PMP-C. DISCUSSION The results of this study clearly show that PMP-C and PMP-D2 differ significantly with respect to their selectivity toward serine proteases, even though they exhibit a high sequence homology (45%) and are structurally related. 4 It seems reasonable to hypothesize that a common ancestor might have adapted for specific and diverse biological functions by punctual mutations that do not affect the overall three-dimensional structure. In that respect, it is remarkable that a unique substitution in the reactive site of PMP-D2 (from Arg to Leu) is sufficient to restore the serine protease inhibitory activity.
Since PMP-C is a tight-binding reversible inhibitor and has a small and compact shape and an exposed binding loop, we propose to include it in the large group of the "small canonical tight-binding serine protease inhibitors." This group of protease inhibitors consists of 16 different families and includes peptides ranging from 29 to 120 amino acids (2). Interestingly, the peptides of this group are structurally unrelated, but share some properties such as hydrophobic cores (often maintained by disulfide bonds), stability toward unfolding, and, more remarkably, an exposed binding loop (containing the scissile P1-PЈ1 bond) that fits into the active site cleft of the serine protease. The specificity of serine protease inhibitors is significantly, but not exclusively, determined by the nature of the P1 residue in the reactive site. Although in most families of proteins, the active regions are highly conserved, in the serine protease inhibitors, there is no consensus sequence for the reactive site emerging yet. Indeed, retention of activity in these proteins exists even though the P1 residue has been changed (1). In some cases, substitutions lead to a predictable change in the inhibitory specificity (17, 19 -21).
Taking into account the variability of the P1 region, we have designed "variants" of PMP-C, PMP-D2, and HI by targeting precisely the amino acid replacement. The aim of this study was to determine the P1 residue of the peptides and to increase the affinity toward HLE, since a variety of elastase inhibitors have been shown to be effective in animal models of emphysema, acute respiratory distress syndrome, rheumatoid arthritis, cystic fibrosis, bronchitis, or acute pancreatitis.
We have proved that Leu-30 is actually the P1 position by modulating successfully the inhibitory properties of PMP-C (maintaining its activity toward HLE and decreasing it toward ␣-chymotrypsin) with the L30V variant. This result is unambiguously confirmed in the case of PMP-D2 and HI, where Arg-29 (equivalent to Leu-30 in PMP-C) was replaced by a Leu; the R29L variants are potent ␣-chymotrypsin inhibitors and they present a reasonable affinity toward HLE.
Surprisingly, the replacement of the P1 ϭ Leu-30 (in PMP-C) by Val increased the affinity toward HLE by a factor of 3 only. Indeed, several novel elastase inhibitors were obtained by replacing the P1 for Val (19 -21). Our results highlight the significant role of the nature of the amino acids within the reactive site other than the P1 side chain for elastase inhibition. We therefore evaluated the effect of the PЈ1 residue (large and basic side chain) by replacing the Lys-30 in the R29L variant of PMP-D2 (which was presently the most effective HLE inhibitor) by a Met (often found in PЈ1 position of HLE inhibitors), and the increased affinity of this double variant for HLE confirmed our previous observation. Up to now, the most appropriate P3-PЈ3 sequence for elastase inhibition is CTLMGC; however, it should be noticed that the double variant of PMP-D2 is more powerful than the equivalent double variant of PMP-C K31M/A32G; this is indicative of a sequence and/or conformational effect which influences the reactive site binding loop.
It is noteworthy that although PMP-C, PMP-D2, and HI have several Lys residues in their sequences, and PMP-D2 and HI have an additional Arg, none of these peptides inhibits porcine trypsin. This result is surprising, considering that many natural trypsin inhibitors have a Lys as P1 residue and that most synthetic substrates of trypsin have an Arg as P1 residue.
While this work was in progress, Boigegrain et al. (22) have isolated from the hemolymph of L. migratoria two peptides with sequences identical with PMP-C and PMP-D2. They have shown that both PMP-C and PMP-D2 are powerful inhibitors of ␣-chymotrypsin (K i of 0.25 nM and 0.12 nM, respectively) and weak to medium inhibitors of HLE (K i Ͼ 0.1 M and of 18 nM, respectively). These values are significantly different from ours. We have shown in this work that PMP-D2 interacts only weakly with ␣-chymotrypsin (K i of 1.5 M) and does not inhibit HLE, and, similarly, PMP-C has a K i of 0.12 M instead of a K i Ͼ 0.1 M toward HLE. We have no explanation for this discrepancy.
It will be interesting to further characterize which amino acid(s) outside the reactive site and which part(s) of the framework are essential for the protease inhibitory activity. This should help to design more affine and smaller peptidic analogues or peptidomimetics, which are required for a therapeutic use of these peptides.