INTRODUCTION

Endopeptidase EC 3.4.24.15 (EP24.15)¹ is a 75-kDa neutral metalloendopeptidase that cleaves on the C terminus side of specific hydrophobic residues in substrates of less than 17 amino acids (1, 2). EP24.15 is thought to be involved in the regulated metabolism of a number of neuropeptides including gonadotrophin-releasing hormone, bradykinin, and neurotensin and has more recently been implicated in the aberrant processing of the amyloid precursor protein in Alzheimer's disease (3, 4). EP24.15 is widely distributed in cells and tissues throughout the body, predominantly as a soluble form constituting about 80% of the total activity and as a minor membrane-associated form accounting for the remainder of the activity (5). High levels of EP24.15 activity have been localized, both catalytically and immunohistochemically, primarily to the brain, pituitary, and testis with lower levels in other tissues such as the liver, kidney, spleen, and lung (6).

Typical of metalloendopeptidases EP24.15 is inhibited by metal ion chelators and can be reactivated by divalent cations such as Zn²⁺ or Mn²⁺. Like many of the thermolysin-like neutral metalloendopeptidases, the protein's active site contains a zinc ion, which participates in the catalytic process (7) and has the classical HEXXH motif (8) typical of zinc metalloproteases (9). However, unlike the other thermolysin-like metallopeptidases such as angiotensin-converting enzyme, endothelin-converting enzyme, neutral endopeptidase, and the closely related endopeptidase EC 3.4.24.16 (EP24.16), EP24.15 displays a unique thiol activation. EP24.15 is activated by low levels of thiol-containing reducing agents (<0.5 mM DTT) and inhibited by high concentrations (>5 mM DTT) (1, 10-12). Inhibition has been attributed to the disruption of intramolecular disulfide bridges (13) and to the thiophilicity of the catalytic zinc ion (14).

Despite the thiol-induced increase in EP24.15 activity, studies examining inhibition by thiol-blocking agents are somewhat ambiguous. The specific and effective cysteine peptidase inhibitor E-64 was shown not to affect EP24.15 activity (12, 15), and partial inhibition was only observed at relatively high concentrations in a yeast homologue of EP24.15 (16). Inhibition by thiol-modifying agents such as iodoacetate, iodoacetamide, and N-ethylmaleimide vary from virtually zero to 100%, and inhibition by these agents has been shown to be a time-dependent process (17). Thus the precise mechanism(s) by which thiol-reactive agents modify the activity of EP24.15 remains unclear.

The sequence of EP24.15 contains 14 cysteine residues, 7 of which are conserved in EP24.16, a closely related enzyme (18). It has been suggested that cysteine residue 483, which lies 5 residues from the catalytic center, is responsible for the thiol dependence of EP24.15 (8). However, this cysteine is also present in EP24.16, which is not activated by reducing agents, thus raising doubts as to its involvement. The present studies were undertaken to determine the mechanism(s) by which EP24.15 is activated by thiols and to define the precise cysteine residue(s) involved. We have employed size exclusion chromatography and dynamic light scattering to examine EP24.15 in both a reduced and non-reduced state. We have also characterized the enzyme's ability to bind the specific inhibitor N-[1-(R,S)-carboxy-3-phenylpropyl]-Ala-Ala-Tyr-p-aminobenzoate (cFP) in the presence and absence of thiol reductants. Finally, we have generated a series of EP24.15 mutant enzymes where each cysteine residue was systematically mutated to a serine and analyzed the properties of these mutants in the presence and absence of a reducing agent. Through these studies we have determined that EP24.15 forms intermolecular disulfide bridges involving Cys-246, Cys-248, and Cys-253 and that thiol activation of the enzyme occurs by the reduction of these low catalytic activity multimers to the more active monomeric form.

EXPERIMENTAL PROCEDURES

The endopeptidase 24.15 inhibitor cFP was synthesized (Dr. J. Boublik and L. Lakat) and iodinated (M. Fullerton) at the Baker Institute. The specific quenched fluorescent substrate (QFS), 7-methoxycoumarin-4-acetyl-Pro-Leu-Gly-Pro-D-Lys-(2,4-dinitrophenyl), was synthesized by Auspep (Parkville, Victoria, Australia). DTT was purchased from Calbiochem (Alexandria, New South Wales, Australia). All other reagents were purchased from Sigma unless otherwise stated.

Endopeptidase 24.15 activity was assayed using a specific QFS, 7-methoxycoumarin-4-acetyl-Pro-Leu-Gly-Pro-D-Lys-(2,4-dinitrophenyl), similar to that described by Tisljar et al. (19). Recombinant endopeptidase 24.15 (0.5 µg) was assayed in duplicate in 2.5 ml of 25 mM Tris-buffered saline, pH 7.4, with or without 0.1 mM DTT. Substrate (0.5 mg/ml in dimethyl sulfoxide) was added to a final concentration of 4.5 µM, and the assay tubes were incubated at 37 °C for 30 min. The reactions were stopped by the addition of 25 µl of 100 mM ZnCl₂, and the tubes were cooled to room temperature. Fluorescence was read on a Perkin-Elmer LS-5 luminescence spectrometer (excitation, 314; emission, 418 nm).

Testes were removed from male Sprague-Dawley rats (280 g), and the tissues (7.3 g) were homogenized using a Polytron kinematic homogenizer in an equal volume to weight of ice-cold 25 mM Tris-buffered saline, pH 7.4. The homogenate was centrifuged for 5 min at 1,000 × g in a microcentrifuge at 4 °C. The supernatant (equivalent of 0.1 µl) was assayed for EP24.15 activity using the QFS assay, and 250 µl (diluted up to 400 µl with 25 mM Sørensen buffer, pH 7.0 (Na₂PO₄·NaH₂PO₄) (20)) was fractionated by size exclusion chromatography as described for the inhibitor binding studies. 50 µl of each fraction was assayed for EP24.15 activity using the QFS activity.

Recombinant EP24.15 (100 µg in 100 µl) was reduced by incubation with 2-mercaptoethanol (2 µl) for 1 h at 37 °C. Alkylation of EP24.15 (100 µg in 100 µl) was achieved by a 1-h incubation at room temperature with 4-vinylpyridine (2 µl). Size exclusion chromatography was performed using a Waters HPLC system, comprising two 510 pumps, a model 680 gradient controller, a U6K injector, and a model 441 UV absorbance detector. The size exclusion column (TSK-GEL SW3000, Toya Soda Co., Japan) was eluted with 100 mM ammonium acetate, pH 7.0, isocratically at a flow rate of 0.5 ml/min. Absorbance of the emerging peaks was measured at 214 nm. The column was calibrated with thyroglobulin, myosin, human -globulin, bovine serum albumin, and L-tyrosine.

Recombinant EP24.15 (20 µg) was incubated with ¹²⁵I-labeled cFP (1 ng, 2.5 × 10⁶ cpm) and cold iodinated cFP (2 µg) prepared by the IODO-GEN method (21) in 200 µl of 25 mM Sørensen buffer, pH 7.0, in the presence or absence of 1 mM dithiothreitol, for 15 min at room temperature. The mixture was then fractionated by size exclusion chromatography as described above, except the TSK-GEL SW3000 column was eluted with 25 mM Sørensen buffer, pH 7.0, at a flow rate of 0.5 ml/min. Fractions were collected at 0.5-min intervals and counted on a  counter (RIASTAR, Packard Instrument Co., Canberra, Australia). The column was calibrated prior to use as described previously.

Double-stranded site-directed mutagenesis of rat EP24.15 was performed on a pGEX-2 (Pharmacia Biotech Inc.) derived EP24.15 expression vector pG-24.15 (22) modified for rapid screening of mutations by the addition of a unique restriction endonuclease site (EcoRI) that replaces the ApaI restriction site on the plasmodia. Oligonucleotide primers were synthesized with mismatches coding for an amino acid change of Cys to Ser for each of the 14 Cys residues present in the protein. Oligonucleotides were 5-phosphorylated with T4 polynucleotide kinase (New England Biolabs Inc., Beverley, MA) and annealed to the double-stranded expression vector plasmid. Primers were extended and ligated in a single reaction, and the resulting plasmid DNA was selected for the ApaI to EcoRI mutation by digestion with ApaI (to cleave the wild-type DNA). Plasmid DNA containing the desired mutation was transformed into competent DH5 bacterial cells and plated overnight on ampicillin plates to yield single colonies. The plasmid DNA was purified (Mini-Prep, QIAGEN, Inc., Santa Clarita, CA) and cleaved with EcoRI to screen for mutations. Mutations were confirmed by double-stranded template dideoxy sequencing with Sequenase (U. S. Biochemical Corp.). The double and triple Cys mutations were prepared in an identical manner. Expression and purification of the mutant proteins for biochemical characterization were as described (22). Purification to homogeneity was assessed by SDS-polyacrylamide gel electrophoresis, and protein was quantitated by Bradford assay using bovine serum albumin as a standard (Pierce) (23). Yields of expressed protein were similar for all of the mutations; aliquots were stored at 80 °C for subsequent study.

To measure the polydispersity of wild-type and C246S/C248S/C253S mutant EP24.15, dynamic light scattering was performed at 23 °C on a DynaPro 801 molecular sizing instrument (Protein Solutions, Charlottesville, VA). A 1 mg/ml solution of each enzyme in 25 mM Tris, pH 7.4, 100 mM NaCl, with and without 0.1 mM DTT was equilibrated for 1 h before repetitive measurements were taken over a period of 30-45 min. Analysis to determine particle radius distribution and molecular weight was performed using software supplied by the manufacturer.

2 µg of EP24.15 was incubated in a total volume of 200 µl of 1 mM Sørensen buffer, pH 7.4, with 10 nM cFP, 20,000 cpm of ¹²⁵I-cFP in the absence or presence of 0.1 mM DTT, at 37 °C for 30 min. 1 g of hydroxyapatite (Bio-Rad) was suspended in 6 ml of 1 mM Sørensen buffer, pH 7.0, and allowed to settle. The top layer was aspirated to remove free floating particles, and the volume made up to 6 ml again; this process was repeated 3 times. 200 µl of the hydroxyapatite suspension was added to the protein mixture and incubated on ice for 30 min with regular vortexing. The inhibitor enzyme-hydroxyapatite complex was precipitated by centrifugation at 7,500 × g for 5 min, and the supernatant was removed. The pellet was washed 3 times in 300 µl of 1 mM Sørensen buffer, pH 7.0, before being resuspended and counted along with the supernatant and the pooled wash fractions. The number of counts bound in the pellet fraction was taken as a percentage of the total counts.

RESULTS

Both recombinant and tissue-derived EP24.15 exhibited a marked activation (approximately 8-fold) in the presence of 0.1 mM DTT in comparison with the absence of reducing agent (Fig. 1).

Under non-reducing conditions, fractionation of recombinant endopeptidase 24.15 by size exclusion chromatography revealed aggregates of approximately 150, 240, and >300 kDa in addition to the monomeric species of 75 kDa (Fig. 2A). However, once subjected to reduction or reduction and alkylation, these aggregates dissociated, and EP24.15 eluted as a single species of 75 kDa (Fig. 2, C and D). If simply reduced prior to fractionation by size exclusion chromatography under non-reducing conditions, EP24.15 eluted in both the monomeric and an approximately 150-kDa aggregated form (Fig. 2B) suggesting a rapid and dynamic formation of multimeric forms.

Analysis of rat testicular soluble extract indicates that the formation of EP24.15 multimeric species is not simply a function of the recombinant protein in a highly purified and concentrated form. Size exclusion chromatography of the rat testes extract under non-reducing conditions showed that the vast majority of EP24.15 (determined following thiol activation of collected fractions) corresponded to a molecular mass much greater than 75 kDa (Fig. 3). However, when chromatographed under reducing conditions, the activity eluted as a single peak corresponding to the monomeric molecular mass of 75 kDa.

To determine whether the thiol activation of EP24.15 is a function of the protein's ability to bind substrate when either in the monomeric or multimeric forms, recombinant EP24.15 was incubated with the radiolabeled EP24.15-specific inhibitor cFP both in the presence and absence of a thiol-reducing agent. Fractionation by size exclusion chromatography under either non-reducing (Fig. 4A) or reducing (Fig. 4B) conditions showed that the inhibitor binds almost exclusively to the monomeric form with approximately a 5-fold increase in the amount of the radiolabeled cFP bound to EP24.15 in the presence of DTT. The increase in the amount of cFP bound under reducing conditions is consistent with increased amounts of monomer; hence thiol activation of EP24.15 likely reflects the conversion of an inactive multimer to an active monomer. The change in the degree of EP24.15 aggregation (reduced versus non-reduced) is also reflected in the UV trace of protein absorbance (214 nm).

To determine which of the 14 cysteine residues in EP24.15 amino acid sequence (Fig. 5) may be involved in the formation of the multimers and hence the regulation of endopeptidase activity, a series of mutant enzymes were generated. The cysteine residues were converted systematically by site-directed mutagenesis to serine residues, and the activity and degree of thiol activation were analyzed using the QFS assay. The mutant enzymes of cysteine residues 246, 248, and 253 showed high activity in the absence of DTT, up to 8-fold greater than observed for the wild-type and other mutated enzymes (Fig. 6A). When assayed in the presence of 0.1 mM DTT, all three mutants showed only a modest activation in comparison with the substantial thiol activation observed for the wild-type and remaining mutant enzymes. Interestingly, cysteine residues 246 and 253 are unique to EP24.15 in contrast to EP24.16, which does not exhibit thiol activation, whereas cysteine residue 248 is conserved. Therefore, a double mutant (C246S/C253S) and a triple mutant (C246S/C248S/C253S) were generated in the same fashion. Activity assays demonstrated that both of these mutants have high intrinsic activity in the absence of a reductant and do not demonstrate enhanced activity in the presence of DTT (Fig. 6B).

Wild-type and C246S/C248S/C253S mutant EP24.15 were subject to dynamic light-scattering analysis to ascertain their size distribution in solution in the presence and absence of DTT. The wild-type enzyme exhibited polydispersity or aggregation in the absence of reductant, reflected by a wider distribution of radii. In comparison the reduced form appears monomeric based on the radius size and the calculated molecular weight. When the C246S/C248S/C253S mutant was subjected to the same measurements both the reduced and non-reduced forms appeared monomeric (Table I).

Table I. Dynamic light scattering of reduced and non-reduced EP24.15 Distribution of particle radii (±S.D.) and estimated molecular mass were determined by dynamic light scattering for the wild-type (WT) and triple cysteine mutants (1.0 mg/ml) in the presence and absence of 0.1 mM DTT. The estimated molecular mass of the various EP24.15 forms was calculated from the hydrodynamic radius of the particle. Sample Molecular mass Radius kDa nm ± S.D. WT  DTT 1423 12.8  ± 7.6 WT + DTT 96 4.2  ± 1.5 C246S/C248S/C253S  DTT 90 4.1  ± 1.2 C246S/C248S/C253S + DTT 79 3.9  ± 0.9

To assess the catalytic site availabilty, both C246S/C253S and C246S/C248S/C253S EP24.15 mutant enzymes as well as the wild type were incubated with ¹²⁵I-cFP in the presence and absence of 0.1 mM DTT. The enzyme-inhibitor complex was precipitated using hydroxyapatite, and the bound cFP was compared with the free cFP. As shown in Fig. 7, there was a 10-fold increase in the amount of inhibitor bound to the wild-type enzyme in the presence of DTT. However, inhibitor binding to both mutant enzymes was the same in both reduced and non-reduced situations and similar to reduced wild-type EP24.15. This result indicates that the catalytic site of both the C246S/C253S and C246S/C248S/C253S mutants is available for binding regardless of the redox state of the environment.

DISCUSSION

In this study, we have confirmed that both recombinant and tissue-derived EP24.15 activity is markedly increased in the presence of a reducing agent. We have also shown that both purified recombinant and tissue extracts of native EP24.15 form multimers of approximately 150, 240, and >300 kDa through intermolecular disulfide bridges, which upon reduction return EP24.15 to a monomeric form. The formation of these multimers is a rapid dynamic equilibrium process as can be seen by the rapid reformation of the multimers upon the removal of reducing agent. In addition we have used dynamic light scattering, a spectroscopic technique, to determine the distribution of reduced and non-reduced EP24.15 particle size in solution. The molecular weight is resolved from the hydrodynamic radius of the particle, and the range of dispersity is reflected in the standard deviation of the radius (24). The calculated molecular mass of the reduced samples and the triple Cys mutant (DTT) are within 20% of the apparent molecular mass of EP24.15 (75 kDa), and their radius sizes are almost identical. However, under non-reducing conditions the wild-type protein has a radius 2-3 times greater than that of the monomer and an increased calculated molecular weight confirming the formation of aggregates.

Furthermore, we have demonstrated that inhibitor access to the catalytic site appears restricted when EP24.15 is multimerized in that inhibitor binding only occurs in the monomeric form. Thus, the observation that access to the particular cysteine residues involved may be restricted, along with the dynamic nature of these interactions, may help explain the ambiguous data obtained for the inhibition of EP24.15 by alkylating and other thiol-blocking agents.

Through systematic site-directed mutagenesis we have shown that the closely located cysteine residues 246, 248, and 253 are likely involved in forming these intermolecular disulfide bridges and hence involved in the thiol activation of EP24.15. Although Cys-248 is conserved in EP24.16, a closely related but non-thiol-activated peptidase, it is presumably acting in a cooperative manner with the EP24.15 unique Cys-246 and Cys-253 to confer the thiol dependence. These results are in agreement with the suggestion of Barrett et al. (25) that Cys-246 and Cys-253 were likely candidates for a role in the thiol activation of EP24.15, reflecting both their close location and sole appearance in EP24.15 upon sequence comparison with mitochondrial oligopeptidase. Our findings, however, are not consistent with the hypothesis that the cysteine, which is 5 residues from the catalytic center (Cys-483) and also present in EP24.16, is involved in the thiol activation (8).

We thus propose a mechanism for the thiol activation of EP24.15 where under oxidative conditions the enzyme aggregates through the formation of intermolecular disulfide bridges involving cysteine residues 246, 248, and 253. This multimeric form is inactive, suggesting that the disulfide bridging between one or more EP24.15 molecules blocks substrate access. The addition of a reductant disrupts the disulfide bridges by monomerizing the enzyme and thus removing the restriction on access of substrate to the catalytic site and therefore activating EP24.15. The formation of multimers through disulfide linkages involving other cysteines cannot be dismissed. However, only when the disulfide linkages involve Cys-246, -248, and -253 is the catalytic activity of EP24.15 affected. It is of interest to note that Cys-246, -248, and -253 lie within a region of charged residues (Fig. 5) consistent with the concept that this part of the molecule may be located at or near the enzyme surface.

To our knowledge, this is the first report of a peptidase being activated in such a manner, i.e. the reduction of multimers rather than their formation to promote activation. However, it is not the first report of thiol activation in which disulfides blocking the active site are reduced. Plant NADP-dependent malate dehydrogenase has been shown to be activated through thiol/disulfide interchange with reduced thioredoxin such that disulfide reduction at the N terminus leads to a conformational change to the active site, whereas the C-terminal end, which shields access to the catalytic residues, is opened by this reduction (26).

The physiological significance of the thiol activation of EP24.15 both intra- and extracellularly is still to be elucidated. We have examined the catalytic activity of rat EP24.15 in testes extract on both reduced and non-reduced size exclusion chromatography and have found a single peak of activity under reducing conditions corresponding to the monomeric form, whereas the activity is eluted in a higher molecular mass form when non-reduced. In addition, we have examined the activation of EP24.15 over a range of protein concentrations and exclude the possibility that multimer formation only occurs at high concentrations of recombinant protein. Indeed, we find similar or even greater thiol activation at lower concentrations (<0.1 µM) of EP24.15, similar to those that we calculate exist, for example, in the cell nucleus.² Thus nuclear EP 24.15 activity could be controlled by the well documented (27-29) regulated changes in nuclear redox potential. In the cell cytosol we would predict that EP24.15 is largely in the monomeric form, reflecting the general reductive environment of the cytosolic compartment. At the cell surface, however, which is generally oxidative, it is conceivable that enzymes such as the protein disulfide isomerase(s) or thioredoxin (30-32) may influence EP24.15 activity and thus ultimately regulate peptide signal delivery at the cell surface.

In conclusion, we have demonstrated that the mechanism by which EP24.15 is activated by thiols likely reflects the conversion of an inactive multimer (where substrate access to the catalytic site is blocked) into an active monomer. This conversion reflects the disruption of intermolecular disulfide bridges involving a cluster of cysteine residues at positions 246, 253, and possibly 248 of EC 3.4.24.15.