Oxidized elafin and trappin poorly inhibit the elastolytic activity of neutrophil elastase and proteinase 3

Authors


J. G. Bieth, INSERM U 392, Faculté de Pharmacie, 74 route du Rhin, 67400 Illkirch, France
Fax: +33 3 90 24 43 08
Tel: +33 3 90 24 41 82
E-mail: jgbieth@pharma.u-strasbg.fr

Abstract

Neutrophil proteinase-mediated lung tissue destruction is prevented by inhibitors, including elafin and its precursor, trappin. We wanted to establish whether neutrophil-derived oxidants might impair the inhibitory function of these molecules. Myeloperoxidase/H2O2 and N-chlorosuccinimide oxidation of the inhibitors was checked by mass spectrometry and enzymatic methods. Oxidation significantly lowers the affinities of the two inhibitors for neutrophil elastase (NE) and proteinase 3 (Pr3). This decrease in affinity is essentially caused by an increase in the rate of inhibitory complex dissociation. Oxidized elafin and trappin have, however, reasonable affinities for NE (Ki = 4.0–9.2 × 10−9 m) and for Pr3 (Ki = 2.5–5.0 × 10−8 m). These affinities are theoretically sufficient to allow the oxidized inhibitors to form tight binding complexes with NE and Pr3 in lung secretions where their physiological concentrations are in the micromolar range. Yet, they are unable to efficiently inhibit the elastolytic activity of the two enzymes. At their physiological concentration, fully oxidized elafin and trappin do not inhibit more than 30% of an equimolar concentration of NE or Pr3. We conclude that in vivo oxidation of elafin and trappin strongly impairs their activity. Inhibitor-based therapy of inflammatory lung diseases must be carried out using oxidation-resistant variants of these molecules.

Abbreviations
Lys-(pico)

lysyl-(2-picolinoyl)

MeOSuc

methoxysuccinyl

NE

human neutrophil elastase

pNA

p-nitroanilide

Pr3

human neutrophil proteinase 3

RBB–elastin

remazol-Brilliant Blue–elastin

Many amino acid residues of proteins are susceptible to oxidation by reactive oxygen species. Methionine, the most sensitive of amino acids to oxidation, is readily transformed into a mixture of the S- and R-epimers of methionine sulfoxide. The latter may be recycled by methionine sulfoxide reductases in the presence of thioredoxin, which itself may be regenerated by thioredoxin reductase in an NADPH-dependent reaction. Excessive methionine sulfoxide production and/or a defect in its recycling is believed to be involved in age-related diseases and in shortening of the maximum life span [1].

Oxidative processes also take place in lung infection and inflammation, where they are used, in conjunction with proteolytic enzymes, to kill bacteria and destroy foreign substances in the phagolysosome of polymorphonuclear neutrophils. The membrane of these phagocytes contains an NADPH oxidase, which transforms molecular oxygen into the short-lived superoxide anion. Superoxide dismutase transforms the latter into H2O2, an oxidant that further yields hypochloride in the presence of neutrophil myeloperoxidase. Aliphatic amines transform hypochloride into chloramines, which are potent and long-lived oxidants [2].

In inflammatory lung diseases, such as chronic bronchitis, emphysema or cystic fibrosis, excessive recruitment, activation or lysis of neutrophils results in the extracellular release of neutrophil elastase (NE; EC 3.4.21.37), proteinase 3 (Pr3; EC 3.4.21.76) and cathtifin G, three neutral serine proteinases that have been shown in vitro to cleave lung extracellular matrix proteins, including elastin, collagens, fibronectin and laminin. These enzymes are thought to be responsible for lung tissue destruction [2,3].

Nature has designed potent protein proteinase inhibitors to prevent local proteolysis caused by accidental neutrophil proteinase release during normal breathing, where inhalation of micro-organisms and air pollutants always takes place. These proteins include α1-proteinase inhibitor (also called α1-antitrypsin; a 53-kDa protein that inhibits the above three enzymes) [3]; α1-antichymotrypsin (a 68-kDa protein that specifically inhibits cathtifin G) [4]; mucus proteinase inhibitor (also called secretory leukoprotease inhibitor, or SLPI; an 11.7-kDa protein that inhibits NE [5] and cathtifin G, but not Pr3 [3]); and elafin and its precursor trappin-2 (also called pre-elafin and referred to as trappin throughout this article; that inhibit NE and Pr3 [6], but not cathtifin G [3]). The two former proteins are mainly synthesized in the liver and reach the lung via the blood circulation. They are irreversible inhibitors that belong to the serpin family. Their interaction with proteinases is characterized by a single constant – the association rate constant (inline image) [7]. The two latter molecules are synthesized in the lung and belong to the canonical type of inhibitors that interact reversibly with their target enzymes, the reaction being described by an association and a dissociation rate constant and inline image an equilibrium dissociation constant Ki = kdiss/kass[5,8].

Trappin is a 9.9-kDa protein formed of two proteolytically cleavable domains. Four disulfide bridges stabilize its 6-kDa C-terminal inhibitory domain, named elafin in this article [9,10]. Its N-terminal domain, the so-called cementoin domain, contains four repeats, with a Gly–Gln–Asp–Pro–Val–Lys consensus sequence homologous to a putative transglutaminase substrate motif. The trappin molecule may therefore be covalently attached to other proteins [11]. These inhibitors are also antimicrobial [12,13] and thus participate in innate immunity [14].

We have recently used the Pichia pastoris expression system to prepare elafin and trappin in high yields. The two full-length recombinant inhibitors were found to be virtually indistinguishable in their kinetic constants for the inhibition of NE and Pr3: both were fast-acting inhibitors with kass = 2–4 × 106 m−1·s−1 and formed very stable inhibitory complexes with kdiss and Ki in the 10−4·s−1 and 10−10 m range, respectively [15].

In inflammatory lung diseases, activated or lysed neutrophils do not only release proteinases but also the aforementioned oxidants. The present article reports the kinetic consequences of inhibitor elafin and trappin oxidation on their interaction with NE and Pr3. It also examines the effect of insoluble elastin on the inhibitory properties of the native and oxidized inhibitors.

Results

Oxidation decreases the affinity of elafin and trappin for NE and Pr3

We oxidized elafin and trappin using either N-chlorosuccinimide, a classical reagent for surface-exposed methionine residues [16] or with the myeloperoxidase/H2O2/halide system, the neutrophil's oxidation device [17]. Figure 1 shows the effect of increasing concentrations of native and oxidized elafin and trappin on the activity of a constant concentration of NE and Pr3. Straight inhibition curves were obtained with the native inhibitors, in agreement with the low values of Ki[15], as compared with the enzyme concentrations used in the present assays [18]. In contrast, the curves describing the inhibition of NE by the oxidized inhibitors were concave, indicating a significant decrease in affinity [18]. The inhibition of Pr3 was even more dramatically affected: an equimolar solution of enzyme + oxidized inhibitor yielded only about 50% inhibition.

Figure 1.

Inhibition of neutrophil elastase (NE) and proteinase 3 (Pr3) by native and N-chlorosuccinimide-oxidized elafin (A) and trappin (B). Increasing concentrations of inhibitor were added to constant concentrations of enzyme, and the residual enzymatic activities were measured using appropriate substrates. (□), Native inhibitors + NE or Pr3; (○), (Δ), oxidized inhibitors + NE or Pr3, respectively.

To express the oxidation effect in a quantitative manner, we measured the equilibrium dissociation constant, Ki, for the complexes formed of oxidized elafin or trappin and NE or Pr3. Oxidation was carried out with either N-chlorosuccinimide or myeloperoxidase. The Ki values were determined from inhibition curves, such as those shown in Fig. 1. These curves were analyzed using Eqn (1):

image(1)

where a is the relative enzyme activity (rate in the presence of inhibitor/rate in its absence), [E]0 and [I]0 are the total enzyme and inhibitor concentrations, respectively, and K = Ki if the substrate (S) does not dissociate EI during the 20–60 s assay of enzymatic activity or K = Ki(1+[S]0/Km) if there is partial dissociation of EI by S so that E, I, S are in equilibrium with ES and EI. Substrate-induced dissociation experiments (see below) showed that dissociation of the oxidized inhibitor–NE complex was slow enough to be insignificant during the 20–60 s time period used to measure the activities of the inhibitory mixtures. Therefore, the K of Eqn (1) is substrate-independent and equals Ki. In contrast, dissociation of the oxidized inhibitor–Pr3 complex was very fast, so that E, I, S and their complexes were in equilibrium following the time required to mix the reagents. Hence, K is substrate-dependent and equals Ki(1 + [S]0/Km). As shown in Table 1, oxidation of elafin and trappin significantly increases the Ki (decreases the affinity) for its complexes with NE and Pr3. Oxidation by N-chlorosuccinimide or myeloperoxidase yields inhibitors whose Ki values are not significantly different from each other.

Table 1.  Kinetic constants describing the inhibition of neutrophil elastase (NE) and proteinase 3 (Pr3) by oxidized elafin and trappin. The data for the native inhibors are from Zani et al. [15]. The kdiss and Ki values are experimental, whereas the kass values are calculated. MPO, myeloperoxidase/H2O2/Cl; NCS, N-chlorosuccinimide; ND, not determined.
EnzymeInhibitorOxidantKi (m)kass (m−1·s−1)kdiss (s−1)
  1. a Calculated assuming that dissociation is terminated in 30 s or less, which corresponds to a t½ ≤ 6 s.

NEElafinNone8.0 ± 0.5 × 10−113.7 ± 0.1 × 1063.2 ± 0.1 × 10−4
NCS5.7 ± 0.6 × 10−91.1 ± 0.3 × 1066.3 ± 0.6 × 10−3
MPO4.0 ± 0.6 × 10−9NDND
NETrappinNone3.0 ± 1.0 × 10−113.6 ± 0.5 × 1061.1 ± 0.2 × 10−4
NCS9.2 ± 2.7 × 10−91.0 ± 0.3 × 1069.0 ± 3.1 × 10−3
MPO5.8 ± 0.9 × 10−9NDND
Pr3ElafinNone1.2 ± 0.1 × 10−103.3 ± 0.03 × 1064.0 ± 0.3 × 10−4
NCS2.9 ± 0.2 × 10−8ND≥ 0.1a
MPO2.5 ± 0.2 × 10−8NDND
Pr3TrappinNone1.8 ± 0.6 × 10−102.0 ± 0.1 × 1063.7 ± 1.1 × 10−4
NCS5.0 ± 2.0 × 10−8ND≥ 0.1a
MPO3.5 ± 0.5 × 10−8NDND

Oxidized elafin and trappin form unstable complexes with NE and Pr3

Is the above-observed increase in Ki caused by an increase in the dissociation rate constant, kdiss, a decrease in the association rate constant, kass, or an effect on both parameters (Ki = kdiss/kass)? To answer this question, we measured kdiss by extensively diluting equimolar enzyme–inhibitor solutions into highly concentrated substrate solutions and following the hydrolysis of substrate as a function of time. The complexes formed of NE and native or N-chlorosuccinimide-oxidized elafin and trappin gave progress curves that were initially concave, indicating continuous release of free enzyme, that is, complex dissociation. After a time, the curves became linear, indicating that the enzyme–inhibitor–substrate system had reached its steady state (Fig. 2A). Comparison of the time required to reach this steady state, and of the steady-state rates, clearly shows that the NE-oxidized inhibitor complexes dissociate much faster than the NE-native inhibitor ones. Quantitative calculation of kdiss confirms this (Table 1). The complexes formed of Pr3 and oxidized elafin and trappin were found to dissociate within the mixing time because no presteady state was visible (Fig. 2, curves 1 and 2). Hence, kdiss could not be calculated for these systems but is estimated to be greater than 0.1 s−1 (Table 1 legend). Thus, the oxidation of elafin and trappin leads to a > 250-fold increase of kdiss of their complexes with Pr3. We conclude that the oxidation of elafin and trappin renders the inhibitors unable to form stable complexes with NE and Pr3. Similar results were observed with trappin. Calculation of kass for the NE-oxidized elafin and trappin complexes using the measured values of Ki and kdiss shows that oxidation also decreases the rate constant of enzyme inhibition by factors of three to four. Thus, the deleterious effect of elafin and trappin oxidation on the affinity of the inhibitors for NE is caused by both an significant increase in kdiss and a moderate decrease in kass.

Figure 2.

Substrate- and dilution-induced dissociation of enzyme–inhibitor complexes. The complexes were diluted 100-fold into a concentrated substrate solution ([S]0 = 13.4 Km) and the release of product was recorded as a function of time. (A) Neutrophil elastase (NE)–inhibitor complexes. (B) Proteinase 3 (Pr3)–inhibitor complexes. The inhibitor was N-chlorosuccinimide-oxidized trappin (curves 1) or elafin (curves 2), and native trappin (curves 3) or elafin (curves 4).

Oxidation of Met at P1′ is responsible for the decreased affinities of oxidized elafin and trappin

Elafin and the inhibitory domain of trappin each have two methionine residues (M25 and M51 for elafin, and M63 and M89 for trappin). M25 and M63 are the P1′residues of the inhibitors' active centers. Mass spectrometry of the two proteins oxidized by N-chlorosuccinimide or myeloperoxidase showed that oxidation increased the m/z by 32 Da, indicating that their two methionine residues had been converted into methionine sulfoxide (Fig. 3).

Figure 3.

Mass spectra of native and oxidized elafin (A) and trappin (B). The peaks at m/z = 6000.975 and 9913.063 are assigned to the native inhibitors, whereas the peaks at m/z = 6032.038 and 9948.639 are assigned to the dioxidized inhibitors.

To establish which methionine residue leads to a decrease in inhibitory activity upon oxidation, M25L elafin and M63L trappin (two variants with a nonoxidizable leucine residue at P1′) were prepared. These variants inhibited NE and porcine pancreatic elastase, but did not react with Pr3. In addition, their affinity for NE was lower than that observed with the wild-type inhibitors (Table 2). Oxidation of the two variants with N-chlorosuccinimide and myeloperoxidase increased their m/z value by 15 Da, indicating oxidation of M51 and M89 of M25L elafin and M63L trappin, respectively. On the other hand, oxidation of M25L elafin and M63L trappin did not significantly affect their Ki for NE (Table 2). We therefore conclude that the oxidant-promoted alteration of the Ki of elafin and trappin is caused by the oxidation of their P1′ methionine residue.

Table 2.  Effect of M25L elafin and M63L trappin oxidation by N-chlorosuccinimide on their affinity for neutrophil elastase (NE).
InhibitorKi (m)
  1. a From Zani et al. [15].

Elafin
 Wild-typea0.8 ± 0.05 × 10−10
 M25L mutant0.8 ± 0.2 × 10−9
 Oxidized M25L mutant1.0 ± 0.1 × 10−9
Trappin
 Wild-typea0.3 ± 0.1 × 10−10
 M63L mutant0.9 ± 0.2 × 10−9
 Oxidized M63L mutant1.3 ± 0.2 × 10−9

Elastin impairs the inhibition of NE and Pr3 by native and oxidized elafin and trappin

NE and Pr3 are both able to solubilize fibrous elastin [19]. We used remazol-Brilliant Blue (RBB)–elastin to investigate their elastolytic activity in the absence and presence of native and N-chlorosuccinimide-oxidized elafin. Preliminary experiments were designed to compare the interaction of NE and Pr3 with this fibrous substrate.

About 50% of the enzymes were immediately adsorbed onto fibrous elastin following mixing of the reagents and stirring. After 1 min, 70% of the enzymes were adsorbed. Adsorption was complete after 10 min.

The affinity of elastin for NE or Pr3 was assessed by adding a constant concentration of enzyme to increasing concentrations of elastin, stirring for 10 min, centrifugating the suspensions and measuring the concentration of unbound enzyme using a synthetic substrate. Both NE and Pr3 gave hyperbolic saturation curves, as shown in Fig. 4A. Double reciprocal plots of the data (not shown) were linear, indicating that saturation conformed to classical reversible receptor–ligand binding, that is R+L ⇌ RL (where R represents elastin and L represents NE or Pr3). The binding curves may therefore be described by the following equation:

image(2)

where [R]0 is the total concentration of elastin and [R]0.5 is the concentration for which 50% of enzyme is bound. Non-linear regression analysis based on Eqn (2) gave [R]0.5 values of 0.77 ± 0.12 and 1.12 ± 0.25 mg·mL−1 for NE and Pr3, respectively. The two enzymes therefore have similar affinities for elastin.

Figure 4.

(A) Binding of constant concentrations of neutrophil elastase (NE) (○) and proteinase 3 (Pr3) (□) to different concentrations of insoluble elastin. The curves are theoretical and were generated using Eqn (2) with [R]0 = 0.77 and 1.12 mg·mL−1 elastin for NE and Pr3, respectively. (B) Kinetics of solubilization of elastin by NE (Δ) and Pr3 (bsl00046).

To measure the elastolytic activity of enzyme ± inhibitor mixtures, we used an elastin concentration of 3 mg·mL−1, which is well above the [R]0.5 value. Under these conditions, elastin solubilization by NE or Pr3 was linear, with time, up to an absorbance of at least 0.45 that is, up to at least 30% elastolysis (Fig. 4B). Thus, activity measurements were very reliable. The elastolytic activity of NE was found to be 1.9-fold higher than that of Pr3 (Fig. 4B).

Enzyme–inhibitor mixtures were also tested in the kinetic mode. Enzyme was added to an elastin plus inhibitor suspension to allow it to compete between substrate and native or oxidized inhibitors. The inhibition was assessed using either equimolecular concentrations of enzyme and inhibitor, or a 10-fold molar excess of inhibitor over enzyme. Figure 5 shows the results of competition experiments carried out with 1.5 µm NE or Pr3 and 1.5 µm native or oxidized elafin. The elastolytic activity of NE was found to be inhibited much less by oxidized elafin than by the native inhibitor (Fig. 5A), and the elastolytic activity of Pr3 was found to be almost insensitive to oxidized elafin (Fig. 5B). Native and oxidized trappin behaved in a similar way. With a 10-fold molar excess of inhibitor over enzyme, we observed full inhibition of both proteinases by the native inhibitors, but only ≈ 80% inhibition of NE and 50% inhibition of Pr3 by the oxidized inhibitors.

Figure 5.

Kinetics of solubilization of elastin by 1.5 µm neutrophil elastase (NE) (A) and proteinase 3 (Pr3) (B) in the absence (Δ) or presence of 1.5 µm native (□) or N-chlorosuccinimide-oxidized (○) elafin. The order of addition of the reactants was elastin + inhibitor + enzyme (competition experiment).

While the above data are in overall agreement with the results obtained using synthetic substrates, they also indicate that elastin hinders the inhibition of both enzymes by the native and the oxidized inhibitors. To demonstrate this, we used Eqn (1) with K = Ki(1 + [R]0/[R]0.5) to calculate the percentage of inhibition that would have been observed if the system behaved like classical competitive inhibition. Table 3 compares this theoretical inhibition with the observed inhibition derived from the progress curves shown, for example, in Fig. 5. It was found that (a) the observed inhibition is lower than that with the theoretical inhibitor, regardless of the enzyme, the inhibitor and the state of oxidation of the latter, indicating that elastin does not simply act as a competing substrate but also hinders the inhibition process, (b) Pr3 is much more resistant to inhibition by native elafin than NE, although the two enzyme–inhibitor systems have similar kinetic constants (Table 1) and (c) oxidized elafin and trappin are very poor inhibitors of NE and Pr3.

Table 3.  Theoretical and observed inhibition of the elastolytic activity of neutrophil elastase (NE) and proteinase 3 (Pr3) by native and N-chlorosuccinimide oxidized elafin and trappin. [NE] = [Pr3] = [elafin] = [trappin] = 1.5 µm; [remazol-Brilliant Blue–elastin] ([RBB–elastin]) = 3 mg·mL−1. The theoretical percentage of inhibition was calculated using Eqn (1) (competitive inhibition) with K = Ki(1 + [R]0/[R]0.5). Ki values are from Table 1, [R]0 is the total concentration of elastin (3 mg·mL−1) and [R]0.5 is the elastin concentration at which 50% of enzyme is bound ([R]0.5 = 0.77 and 1.12 mg·mL−1 for NE and Pr3, respectively). The observed percentage of inhibition is that resulting from competition experiments, such as those shown in Fig. 5.
EnzymeInhibitor Percentage inhibition
TheoreticalObserved
NEElafinNative98.493.0
Oxidized87.325.0
TrappinNative99.088.5
Oxidized84.030.5
Pr3ElafinNative98.355.0
Oxidized76.310.0
TrappinNative98.080.0
Oxidized70.019.0

Discussion

The active site of serine proteinase inhibitors is composed of about eight surface-exposed amino acid residues, labeled P5 to P3′, which interact with subsites S5 to S3′ of the proteinase's active center. S1–P1 and S1′–P1′ interactions play an important role in inhibitor specificity and potency [20]. In elafin/trappin, P1 represents Ala and P1′ represents Met. Oxidation of the latter residue to methionine sulfoxide leads to a decrease in the affinity (1/Ki) of the two inhibitors for NE and Pr3. This decrease is significantly more pronounced for Pr3 than for NE and is mainly the result of an important increase in kdiss, the rate constant for the dissociation of the inhibitory complexes (Ki = kdiss/kass). The complexes formed of native elafin or trappin and NE or Pr3 have similar kdiss values, which correspond to a half-life of dissociation of 36–105 min [15]. The oxidation of elafin and trappin down-shifts the half-life of there complexes with NE to ≈ 1.3–1.8 min. On the other hand, the oxidized inhibitor–Pr3 complexes are so unstable that they relax ‘instantaneously’ when diluted into a substrate solution. This means that their half-lifes are not longer than a few seconds. The reason why oxidation renders the inhibitory complexes so unstable is not clear. Methionine sulfoxide is bulkier than methionine. Perhaps steric hindrance prevents easy binding of the methionine sulfoxide residue at the S1′ subsite of the active centers of NE and Pr3. The fact that the S1′ subsite of Pr3 is significantly smaller than that of NE [21] might then explain why (a) Pr3 is more sensitive to inhibitor oxidation than NE and (b) Pr3 does not react with the Met→Leu mutants.

Lung secretions also contain mucus proteinase inhibitor (SLPI), an 11.7 kDa NE inhibitor that shows some homology with elafin [5] and whose P1 and P1′ residues are Leu and Met, respectively [22]. Oxidation of SLPI also reduces its NE inhibitory capacity [23] as a result of methionine sulfoxide formation [8]. Table 4 compares the kinetic properties of native and oxidized elafin and SLPI. It can be seen that the two native inhibitors have very close Ki, kass and kdiss values and that the two oxidized inhibitors also have close affinities for NE. The only difference is that the oxidation of SLPI mainly depresses kass, whereas the oxidation of elafin mainly increases kdiss.

Table 4.  Comparison of the effects of N-chlorosuccinimide oxidation of elafin and the mucus proteinase inhibitor (SLPI) on their interaction with neutrophil elastase (NE).
InhibitorKi (m)kass (m−1·s−1)kdiss (s−1)
  1. a From Boudier & Bieth [8]. b From Table 1.

Native SLPIa9.2 ± 2.5 × 10−114.9 ± 0.5 × 1064.5 ± 0.8 × 10−4
Oxidized SLPIa1.1 ± 0.3 × 10−82.6 ± 0.3 × 1052.9 ± 0.5 × 10−3
Native elafinb8.0 ± 0.5 × 10−113.7 ± 0.1 × 1063.2 ± 0.1 × 10−4
Oxidized elafinb5.7 ± 0.6 × 10−91.1 ± 0.3 × 10−66.3 ± 0.6 × 10−3

Triggered neutrophils release reactive oxygen species as well as the lysosomal enzyme, myeloperoxidase. Therefore, the myeloperoxidase/H2O2/Cl system we have used to oxidize elafin/trappin is a good model for in vivo inhibitor oxidation in neutrophil-rich lung inflammatory fluids. This system yields oxidized inhibitors whose inhibition kinetic constants are indistinguishable from those observed with elafin/trappin oxidized with N-chlorosuccinimide, the classical reagent specific for surface-exposed methionine residues [16].

Oxidation does not fully abolish the inhibitory properties of elafin and trappin. This raises the following question: are the oxidized inhibitors still sufficiently potent to inhibit NE and Pr3 in lung inflammation? The in vivo potency of a proteinase inhibitor depends upon its in vivo concentration ([I]vivo) and the kinetic constants describing its inhibition of the target proteinase [24]. The absolute concentration of a protein in lung secretions is difficult to measure because this protein is collected by bronchoalveolar lavage, which dilutes it to an undefined extent. According to the reasoning of Ying & Simon [25], the elafin concentration in bronchial secretions would be 1.5–4.5 µm. If we assume that an inflammatory lung secretion contains 3 µm oxidized elafin and ≤ 3 µm NE + Pr3 and that there are no competing biological substrates present, we may calculate the percentage of free enzyme using Eqn (1) with, say, [E]0 = 0.3 µm, [I]0 = 3 µm and K =Ki from Table 1. This calculation shows that there is only 0.2% free NE and 1% free Pr3 in this lung secretion, indicating that, in the absence of competing substrates, oxidized elafin still binds NE and Pr3 tightly.

In the lung, the situation appears to be more complex: proteinases are released in a milieu that contains both substrates and inhibitors, which may compete for their binding. This raises the following question: are oxidized elafin and trappin able to prevent or at least to minimize NE- or Pr3-mediated proteolysis of insoluble extracellular matrix proteins, such as elastin, collagen, fibronectin and laminin? We have shown that the main effect of inhibitor oxidation is an increase in the rate of enzyme–inhibitor complex dissociation. As a consequence, if such a complex comes close to an insoluble protein substrate, a fraction of enzyme may be rapidly transferred to this substrate and proteolysis may take place. It should be emphasized that substrate insolubility provides high local substrate concentration and, hence, a strong ability to dissociate an inhibitory complex. Substrate-induced complex dissociation might be particularly important for the Pr3-oxidized inhibitor complexes, whose half-life of dissociation are a few seconds only.

We have used elastin as a substrate to verify the above hypothesis. This insoluble polymer was able to dissociate the native elafin–NE and elafin–Pr3 complexes, despite their low Ki and kdiss values, confirming the above assumption. We have used the measured ‘affinity’ of elastin for the two enzymes to calculate an apparent Ki, which was then used to calculate the inhibition based on simple competition between substrate and inhibitor for the binding of enzyme. This theoretical inhibition was significantly higher than the experimental one, again confirming the above hypothesis. The most important differences were found for the inhibition of Pr3 by native and oxidized elafin. The experiments were carried out with 1.5 µm elafin, which is within the physiological concentration range [25]. In an equimolar mixture of enzyme and oxidized elafin, NE and Pr3 are inhibited to the extent of 25% and 10%, respectively. This clearly shows that oxidized elafin is a poor inhibitor of the elastolytic activity of these two enzymes. Oxidized trappin is somewhat more potent because it inhibits the two proteinases to the extent of 30 and 19%, respectively. It may be anticipated that the oxidized inhibitors will also poorly protect other insoluble extracellular matrix proteins from proteolysis.

Inhibitor-based therapy of inflammatory lung diseases has been proposed in the last decade. For instance, aerosol-delivered α1-antitrypsin [26] and SLPI [27] have been shown to augment the anti-NE capacity of lung secretions. As elafin and trappin inhibit both NE and Pr3, they might be potential drugs in cystic fibrosis where enormous amounts of free NE and Pr3 are found in lung secretions [28]. However, the sensitivity to biological oxidation of the wild-type inhibitors prohibits their therapeutic use: oxidation-resistant variants must be designed. The Met/Leu variants described here can obviously not be used because they do not inhibit Pr3. The preparation of variants with less bulky amino acid residues at P1′ is now in progress.

Elafin is synthesized as trappin, a soluble 9.9-kDa protein whose N-terminal cementoin domain contains transglutaminase substrate motifs that allow it to be covalently attached to insoluble extracellular matrix proteins [11]. It is not unlikely that trappin forms insoluble complexes with such proteins. Under its insoluble form, this inhibitor might therefore be endowed with appealing properties. First, its bioavailability might be dramatically better than that of elafin and soluble trappin. Second, it might be more potent than the soluble inhibitor because insolubility ‘creates’ affinity, a concept classically used in affinity chromatography. Third, it might be less susceptible to oxidation than the soluble molecule because insoluble targets are more difficult to reach than soluble ones as they do not undergo brownian motion. Hence, soluble oxidant scavengers present in lung secretions [2] may more efficiently protect it from oxidation. Covalently bound trappin has not yet been identified in human lung structures. The foregoing view is nevertheless not pure conjecture because animal studies show that intratracheally administered trappin (but not elafin) is able to prevent NE-induced acute lung injury [29].

Experimental procedures

The source and active site titration of NE and Pr3 was the same as described previously [15].

Production of recombinant M25L–elafin and M63L–trappin

Using the elafin cDNA cloned into pGE-SKA-B/K (20 ng) as a template [15], PCR amplification was perforrmed according to the standard procedure of Higuchi et al. [30] to obtain cDNAs encoding M25L–elafin and M63L–trappin. For this purpose, forward primers 5′-CGACTCGAGAAAAGAGCGCAAGAGCCAGTCAA-3′ and 5′-CGACTCGAGAAAAGAGCTGTCACGGGAGTTCCT-3′ were used for amplification of the elafin and the trappin cDNA 5′ end, respectively, and reverse primer 5′CGAGCGGCCGCCCCTCTCACTGGGGAAC-3′ was used for the common 3′ end of elafin and trappin. Oligonucleotides 5′-GGTGCGCCTTGTTGAATCC-3′ (forward) and 5′-GGATTCAACAAGGCGCACC-3′ (reverse) were used to introduce the Met/Leu substitution (Leu codon: TTG). Amplified fragments were cloned into the pPIC9 vector and electroporated into P. pastoris yeast strain GS115 (his4) competent cells (Invitrogen, Carlsbad, CA, USA).

Both recombinant inhibitors were produced and purified by ion exchange chromatography, as described previously for wild-type elafin and trappin [15]. Each of the molecules migrated as a single band at 7 kDa (M25L–elafin) and 12 kDa (M63L–trappin) in a reducing SDS/PAGE gel, indicating homogeneity of each preparation.

Oxidation of inhibitors

We used either N-chlorosuccinimide [16] or the myeloperoxidase/H2O2/halide system [17]. In the former method, 5 µm inhibitor was reacted with 2 mmN-chlorosuccinimide (Sigma Aldrich, Saint Quentin Fallavier, France) at pH 8.5 (200 mm Tris/HCl). After 20 min at room temperature, 0.55 vol. of the reaction medium was diluted with 0.45 vol. of 100 mmN-acetylmethionine (Bachem, Bubendorf, Switzerland) to stop oxidation. The reaction products were removed by gel filtration on a Sephadex G-25 column (Pharmacia, Uppsala, Sweden), equilibrated and developed with a 5 mm ammonium bicarbonate solution containing 3 mm NaCl. The oxidized inhibitor solution was then lyophilized. In the latter method, 4 µm inhibitor was incubated with 3 nm myeloperoxidase (Athens Research and Technology, Athens, GA, USA) and 0.3 mm H2O2 (VWR International, Fontenay Sous Bois, France) dissolved in 200 mm sodium phosphate, 160 mm NaCl, pH 6.2. After 20 min at room temperature, the oxidation reaction was stopped with 0.36 µm human erythrocyte catalase (Sigma, St Louis, MO, USA). Preliminary experiments showed that the incubation times were sufficient to obtain maximal oxidation of the inhibitors.

Mass spectrometry

We used a Biflex MALDI-TOF spectrometer (Brucker, Wissembourg, France) equipped with a reflectron and a nitrogen laser (λ = 337 nm). The samples were mixed with 1 µL of a matrix formed of a saturated solution of α-cyano-4-hydroxycinnamic acid in H2O/acetonitrile (1 : 1, v/v). After vacuo dessication, measurements were performed in the positive linear mode. Calibration was carried out with insulin (m/z = 5734.4) and horse heart myoglobin (m/z = 16952.5).

Enzymatic methods

All kinetic measurements were carried out in 50 mm Hepes, 150 mm NaCl, pH 7.4, a solution called the buffer.

The rate of solubilization of fibrous elastin was measured using 3 mg·mL−1 RBB–elastin (particle size: 200–400 mesh) (Elastin Products Co., Owensville, MO, USA) suspended in the buffer at 37 °C. The suspension was stirred for 15 min before the addition of enzyme, inhibitor or complex. While stirring was continued, 500 µL samples of suspension were withdrawn at given time-points, mixed with 500 µL of 0.75 m acetate buffer, pH 4.0, centrifuged at 10 000 g for 10 min and read at 595 nm against a blank prepared from a reaction mixture where enzyme and inhibitor were absent. Full solubilization of 3 mg·mL−1 RBB–elastin corresponds to an absorbance at 595 nm of 1.55.

The kinetics of adsorption of NE or Pr3 to RBB–elastin was measured by adding enzyme (final concentration 1.5 µm) to 3 mg·mL−1 substrate, withdrawing samples from the stirred suspensions, centrifugating at 10 000 g and adding 10 µL of supernatant to 990 µL of 0.2 mm methoxysuccinyl-Ala-Ala-Pro-Val-p-nitroanilide (MeOSuc-Ala2-Pro-Val-pNA) or 0.29 mm methoxysuccinyl-lysyl-(2-picolinoyl)-Ala-Pro-Val-p-nitroanilide [MeOSuc-Lys-(pico)-Ala-Pro-Val-pNA] (Bachem) to measure the nonadsorbed NE or Pr3, respectively. The affinity of NE or Pr3 for RBB–elastin was measured by adding enzyme (final concentration 1.5 µm) to suspensions formed of variable concentrations of RBB–elastin, stirring for 10 min, centrifugating and measuring the enzymatic activities in duplicate, as described above.

The equilibrium dissociation constant, Ki, for the enzyme-oxidized inhibitor complexes, was measured by reacting increasing concentrations of oxidized inhibitors with constant concentrations of NE (70 nm) or Pr3 (190 nm). After 20 min at 25 °C, the residual NE and Pr3 activities were measured at 410 nm and 25 °C by following the breakdown of 0.2 mm MeOSuc-Ala2-Pro-Val-pNA or 0.29 mm MeOSuc-Lys-(pico)-Ala-Pro-Val-pNA, respectively. The assay times were 20–60 s. The data were fitted to Eqn (1)[18] by nonlinear regression analysis.

The dissociation rate constant, kdiss, of the enzyme-oxidized inhibitor complexes was measured by dissociating the complexes by both high dilution (100-fold) and high substrate concentration (13.4 Km). A 1 µm enzyme concentration was mixed with 1 µm inhibitor in the buffer. After 30 min at 25 °C, 10 µL of this mixture was added to 990 µL of a buffered substrate solution contained in a thermostated spectrophotometer cuvette. The substrate was 1.5 mm MeOSuc-Ala2-Pro-Val-pNA for the NE–inhibitor complexes and 0.1 mm MeOSuc-Lys-(pico)-Ala-Pro-Val-thiobenzylester [31] for the Pr3–inhibitor complexes. The latter reaction medium also contained 3 mm 4,4′-dithiodipyridine (Sigma Aldrich), which reacts with benzylthiol to form a complex that absorbs at 324 nm [32]. The hydrolysis of substrate was recorded until the absorbance varied linearly with time, indicating that the enzyme/inhibitor/substrate system had reached a steady state. These data were used to calculate the derivative curve representing the time-dependent release of free enzyme from the inhibitory complex. The dissociation rate constant, kdiss, could then be calculated from this curve, as described previously [15].

Acknowledgements

We thank ‘Vaincre la mucoviscidose’, the French cystic fibrosis foundation for financial support, Jean-Marie Strub for mass spectrometric analysis, and Philippe Mellet and Didier Rognan for valuable discussions.

Ancillary