Unveiling the substrate specificity of meprin beta on the basis of the site in protein kinase A cleaved by the kinase splitting membranal proteinase.

The kinase splitting membranal proteinase (KSMP) is a metalloendopeptidase that inactivates the catalytic (C) subunit of protein kinase A (PKA) by clipping off its carboxyl terminal tail. Here we show that this cleavage occurs at Glu332-Glu333, within the cluster of acidic amino acids (Asp328-Glu334) of the kinase. The Km values of KSMP and of meprin β (which reproduces KSMP activity) for the C-subunit are below 1 μM. The Km for peptides containing a stretch of four Glu residues are in the micromolar range, illustrating the significant contribution of this cluster to the substrate recognition of meprin β. This conclusion is supported by a systematic study using a series of the C-subunit mutants with deletions and mutations in the cluster of acidics. Hydrophobic amino acids vicinal to the cleavage site increase the Kcat of the proteinase. These studies unveil a new specificity for meprin β, suggesting new substrates that are 1-2 orders of magnitude better in their Km and Kcat than those commonly used for meprin assay. A search for substrates having such a cluster of acidics and hydrophobics, which are accessible to meprin under physiological conditions, point at gastrin as a potential target. Indeed, meprin β is shown to cleave gastrin at its cluster of five glutamic acid residues and also at the M-D bond within its WMDF-NH2 sequence, which is indispensable for all the known biological activities of gastrins. The latter meprin cleavage will lead to the inactivation of gastrin and thus to the control of its activity.

The presence of a kinase splitting membranal proteinase (KSMP) 1 in the brush-border membranes of the rat small intestine was demonstrated as early as 1979 (1). This proteinase was shown to clip the catalytic (C) subunit of PKA, yielding a distinct cleavage product (CЈ) that was found to be devoid of the kinase activity. The biochemical characterization of KSMP as a proteinase revealed that it is an intriguing enzyme with a combination of the following unique features. (a) KSMP cleaves the C-subunit when it is free but not when inhibited by its regulatory (R) subunits, as in the R 2 C 2 complex (1,2). (b) This cleavage could not be simulated by other proteinases (trypsin, chymotrypsin, clostripain, and papain (2)), suggesting that its specificity is not due merely to an interdomain exposure in C. (c) The proteinase was found to single out and selectively cleave the C-subunit in the presence of the large number of other proteins found in crude extracts of different tissues (brain, liver, or muscle) (3). (d) KSMP was found to cleave the Csubunit in its native conformation but not if the kinase is pre-denatured (2). (e) It distinguishes between the "open" and "closed" conformations of the C-subunit (4) that were recently identified by x-ray crystallography of this kinase (5).
The cleavage of the C-subunit by KSMP leads to the removal of the carboxyl terminus tail of this kinase and seemed to occur at a distinct site (6 -8). Interestingly, two other kinases, the EGF-and the insulin-receptor kinases (which share certain sequence homology with the C-subunit (9)) were also shown to undergo a specific and conformation-dependent cleavage by KSMP (6, 10 -12). In both receptor kinases, it was shown that the KSMP cleavage occurs at the carboxyl-terminal part of the molecules. The specific and restricted character of the KSMP cleavage of C, as well as the EGF-and the insulin-receptor kinases, suggested the existence of a common structural motif that is recognized by the proteinase. Indeed, inspection of the primary sequences of the three kinases revealed that they share a stretch of acidic amino acid downstream from their common protein kinase core, raising the possibility that this stretch is an important biorecognition element for KSMP. This suggestion was supported by three additional findings. (i) The polyglutamic acid effectively inhibits the KSMP cleavage of the C-subunit (K i ϭ 6 M) (7). (ii) The monoclonal antibodies against a branched polyamino acid with exposed clusters of Glu cross react with the C-subunit but not with its KSMP cleavage product (CЈ) while a monoclonal anti-idiotype of these antibodies specifically binds to the active site of KSMP and inhibits it (7). (iii) The immunochemical mapping of the C-subunit with epitope-specific antibodies narrowed down the cleavage site location to the short region accommodating the cluster of acidic residues in C, i.e. Asp 328 -Glu 334 (8).
We have recently shown that C-degrading activity of KSMP can be reproduced by the ␤-subunit of rat meprin (13). Meprin is a membranal metalloendoproteinase found in the intestinal and renal brush-border membranes of the mouse (14,15), rat (16 -18), and man (19,20). Meprins belong to the astacin family of endopeptidases (21)(22)(23)(24) and are usually composed of two types of subunits, ␣ and ␤, that exist as homo-and heterotetramers bound to each other through disulfide bridges (25,26). In spite of a large number of studies on meprins, the substrate specificity and the physiological assignment are not established yet even though several suggestions have been set forth.
The demonstration that meprin ␤ possesses a KSMP activity (13) raised the possibility that the specificity of this enzyme might be different from the one currently accepted. Furthermore, the protein and peptide substrates commonly used for meprins (27)(28)(29)(30) may mislead the search for its physiological  substrates and inhibitors, which have not yet been established  (for an excellent recent review, see Ref. 24). For example, systematic studies on the cleavage of peptide substrates by rat meprin did not point to any consensus motif but indicated that the hydrolysis may occur at peptide bonds adjacent to a hydrophobic (27), aromatic (19), or hydrophilic (31) amino acid residue. None of the substrates previously used for the systematic analysis of the meprin activity contained stretches of acidic amino acids.
This paper reports the identification of the KSMP cleavage site in the C-subunit of PKA and in peptides derived from the sequence of the C-subunit around that site. It illustrates in quantitative terms (K m and V max values) the importance of clustered acidic amino acid residues (Glu and Asp) for optimal KSMP and meprin ␤ cleavage. This conclusion is complemented using a series of C-subunit mutants with deletions and single-, double-, and triple-site mutations in the acidic residues within this cluster. On the basis of a search for proteins and peptide hormones that have a cluster of acidics and that may come in contact with meprin, we suggest gastrin as a potential physiological substrate for meprin ␤. Meprin ␤ is shown to cleave gastrin at its cluster of 5 glutamic acid residues and also its Met-Asp bond within the carboxyl-terminal sequence (WMDF-NH 2 ), which has been exceedingly well preserved during evolution (32) and is claimed to be indispensable for all the known biological activities of gastrins (33,34). The latter meprin cleavage will, therefore, lead to an inactivation of gastrin and thus to the control of its activity.

MATERIALS AND METHODS
Purification and Assay of KSMP and of the Catalytic Subunit of PKA-The purification and assay of KSMP (from rat kidney) (13) and of the catalytic subunit of PKA (35) (from bovine heart) were carried out as described earlier.
Purification of Meprin ␤ Precursor-The meprin ␤ precursor was expressed in the 293 human embryo kidney cell line as reported earlier (13). The KSMP purification procedure described before (13) was applied for the purification of the expressed meprin ␤ precursor, with some modifications. The meprin ␤ expressing clone was grown on the selective medium until a confluent monolayer was formed, collected with a rubber scraper, and washed three times by phosphate-buffered saline. The cells were resuspended in 10 mM Tris-HCl buffer, pH 7.1, containing 10 mM mannitol, and they were ruptured by ultrasound at 4°C. The membrane fraction (i.e. the pellet obtained by centrifugation of the lysis suspension at 50,000 ϫ g for 30 min) was resuspended in the same buffer to adjust the protein concentration to 1 mg/ml, and the membranes were then solubilized by adding octyl-␤-D-glucopyranoside to a final concentration of 1%. The subsequent purification steps were essentially the same as described for the purification of KSMP from rat kidney brush-border membranes (13) except that a Mono Q anionexchange chromatography step was applied instead of DEAE-Sephacell chromatography. The Mono Q column (HR 5 ϫ 5, Pharmacia, Sweden) was equilibrated with a 10 mM Tris-HCl buffer, pH 7.1, containing 1.5 mM MgCl 2 , 1 M ZnCl 2 , and 0.5% octyl-␤-D-glucopyranoside. The meprin ␤ preparation purified by Cu 2ϩ -chelating agarose was extensively dialyzed against the Mono Q equilibration buffer and then loaded on the column, which was developed with a 0 -500 mM NaCl gradient in the same equilibration buffer. Fractions containing the pure meprin ␤ precursor (identified by the Coomassie Blue staining of a band with a molecular mass of 105 kDa in SDS-PAGE) were found to be eluted from the column at ϳ300 mM NaCl and were pooled and used for the experiments described here.
Synthesis of Peptides and Their Purification-Peptides were synthesized by solid-phase peptide synthesis in the Chemical Services at the Weizmann Institute of Science, Rehovot. All the synthetic peptides were purified by reverse-phase HPLC before use and subjected to amino acid composition and sequence analysis to confirm their structure.
Determination of the KSMP Cleavage Site in C-A preparative scale cleavage of the C-subunit was carried out in a reaction mixture (final volume 5.5 ml) that contained 350 g of the C-subunit and 25 g of purified KSMP (13) in 20 mM Tris-HCl, pH 7.1, with the addition of 1.5 mM MgCl 2 and 0.15% octyl-␤-D-glucopyranoside. The reaction was allowed to proceed at 22°C for 80 min and then arrested by adding 50% trifluoroacetic acid to a final concentration of 0.5%. The resulting fragments were loaded onto Ultrasphere ODS (150 ϫ 4.6 mm) reversephase HPLC column (Beckman Instruments) and then eluted with a linear gradient of acetonitrile (0 -80%) in 0.1% trifluoroacetic acid. Two peptide peaks (which were not present at time zero) with retention times of ϳ43 and ϳ41 min were collected, rechromatographed on the same column, and sequenced in a gas-phase sequencing apparatus (Applied Biosystems). It should be noted that the amount of the major peptide (retention time ϳ43 min) was present in a 20-fold excess over the minor peptide (retention time ϳ41 min).
Measurements of the Kinetic Parameters-The kinetic constants K m and v max were measured as described by Cleland (36). Assuming a single substrate mechanism for the reaction, the relationship between the velocity of the reaction and the substrate concentration is given by the equation, where v is the rate of cleavage, K m and v max are Michaelis constants, and [S] is the initial concentration of the substrate. Briefly, the rate of cleavage was measured in each case by monitoring the formation of the cleavage products and then plotting the activity versus the initial substrate concentration in a double reciprocal plot. The equation describing the line, fitted to the experimental values by the least square method, was used for calculating K m , V max , and other dependent constants. For the determination of K cat values, V max was expressed in molar concentration of product formed per second. The molar concentration of the enzyme was calculated assuming molecular masses of 100 kDa for activated meprin ␤ and 85 kDa for KSMP.
Trypsin Activation of the Meprin ␤ Precursor-Purified meprin ␤ precursor was activated by trypsin (sequencing grade, Boehringer Mannheim). The preparative scale reaction mixture (500 l total volume) contained 0.05 mg/ml meprin ␤ precursor, 0.2 g/ml trypsin (the trypsin:protein ratio was 1:250) in 20 mM Tris-HCl, pH 7.1, supplemented with 1.5 mM MgCl 2 and 0.1% octyl-␤-D-glucopyranoside. The reaction was allowed to proceed for 5 min at 37°C, and the trypsin cleavage was then arrested by adding soybean trypsin inhibitor in a 10-fold excess (w/w) over the added trypsin. Complete trypsin inhibition was ascertained in each experiment by running an appropriate control; a substrate was incubated with the same mixture of trypsin and its inhibitor in the absence of the meprin ␤ precursor.
Cleavage of Synthetic Peptides-Cleavage of synthetic peptide substrates was performed in a 20 mM Tris-HCl buffer, pH 7.1, with an addition of 1.5 mM MgCl 2 and 0.1% octyl-␤-D-glucopyranoside. Analytical scale reaction mixtures were 30 l final volume, preparative scale reaction mixtures were 200 l, and the minimal amount of a peptide in the reaction mixture was not below 0.5 nmol. The peptide concentrations in the reaction mixtures ranged from 1 to 200 M, and the reaction was allowed to proceed for a time interval within the linear region of dependence on time. The cleavage was arrested by adding trifluoroacetic acid to the reaction mixture to a final concentration of 0.5%, and then its volume was adjusted to 500 l by 0.1% trifluoroacetic acid and injected to a reverse-phase HPLC column (LiChroCART 125-4 LiChro-Sphere RP-18, Merck, Germany). A flow rate of 1 ml/min was used, and elution of the bound material was achieved using a linear gradient of acetonitrile. The effluent was monitored simultaneously at 210 and 280 nm with a diode-array detector (Hewlett-Packard). A quantitative conversion of the peak area into the amount of peptide in it was calculated from a calibration curve. A quantitative amino acid analysis of the eluted peaks was run along with measuring the area of the peaks. The enzyme activity was defined as micromole of product released by 1 mg of enzyme/1 min.
Construction of Mutants and Their Translation in a Cell-free System-The wild-type murine C ␣ -subunit gene cloned into the pRSET-B vector (Invitrogen) under control of T7 polymerase promoter was a generous gift from Dr. S. S. Taylor (Univerity of California, San Diego). Site-directed mutations were introduced by oligonucleotide-directed mutagenesis of a uracil-containing single-stranded Kunkel template (37). The translation of the coding sequences was carried out in the TNT-coupled transcription/translation rabbit reticulocyte expression system (Promega), as recommended by the manufacturer, and per-formed in the presence of [ 35 S]methionine (Amersham, UK).

Cleavage of the Mutants and Quantitation of Their Cleavage Products-
The wild-type C-subunit and its mutant forms were translated in the rabbit reticulocyte lysate TNT system and subjected to proteolysis by KSMP. The cleavage reaction (15 l) contained 2 l of the standard translation reaction mixture, 0.01 g of purified KSMP preparation in a 20 mM Tris-HCl buffer, pH 7.1, with an addition of 1.5 mM MgCl 2 and 0.1% octyl-␤-D-glucopyranoside. The cleavage was allowed to proceed at 22°C. The components in the reaction mixture were separated by SDS-PAGE, the gel was dried, and the 35 S-labeled protein bands were visualized by autoradiography. Quantitation of the cleavage was done by densitometric analysis of the bands corresponding to the intact and KSMP-cleaved products. The KSMP activity was determined by quantitation of the C-subunit to CЈ conversion using the equation, activity (%) ϭ CЈ/(C ϩ CЈ). Calculated from several time points, the rate of cleavage was extrapolated to time zero. The initial cleavage rate of wild-type C-subunit was taken as 100%, to which the cleavage rates of the mutants were compared.

KSMP Cleaves the C-subunit at the E 332 -E 333 Bond within Its
Cluster of Acidic Amino Acids-To get an insight into the molecular basis of the specific cleavage of the C-subunit by KSMP, it was essential to determine exactly the bonds cleaved in the kinase by this proteinase. This recently became possible upon achieving a purification of KSMP to homogeneity (13).
To identify the exact sites of the KSMP cleavage in the C-subunit, we carried out this cleavage on a preparative scale, and then separated the cleavage products by reverse phase HPLC and sequenced them. Two new peaks were shown to be formed, a major and a minor peak (HPLC data not shown). In quantitative terms, the major peak was about 20-fold more abundant than the minor one. As seen in Table I, the major peak had the sequence EEIRVSINEK*GKEFS resulting from a cleavage of the E 332 -E 333 bond within the cluster of acidic amino acids of this kinase, i.e. residues D 328 -E 334 . This is in agreement with earlier experiments in our laboratory, which implicated the involvement of this cluster of acidic amino acids in the recognition of the C-subunit and of other kinases by KSMP (7, 10, 11) and with our more recent immunological mapping of the cleavage site with specific anti-peptide antibodies (8). The minor peak had the sequence DFLKK (Table I), which fits the stretch 25 DFLKK 29 in the amino-terminal region of the C-subunit (Fig. 1).
KSMP and Meprin ␤ Have Identical K m and V max values for the C-subunit-In an attempt to gain further support to our finding that the KSMP cleavage of the C-subunit can be reproduced by meprin ␤, we compared the K m and V max values of KSMP with meprin ␤ in the cleavage of this kinase. A preparation of KSMP obtained from rat kidney and recombinant meprin ␤ purified from transfected 293 cells (13) were used in this comparison. The cleavage was performed at different concentrations of the C-subunit as described under "Materials and Methods," securing a linear dependence of the proteolysis with time and with the concentration of the proteinase. Each of the reaction mixtures was subjected to SDS-PAGE followed by Coomassie Blue staining. The extent of proteolysis was determined by computing densitometry of these gels, monitoring the amount of the clipped C-subunit (i.e. CЈ) formed. Plotting the KSMP (C-subunit-degrading) activity versus the concentration of the C-subunit in double reciprocal coordinates gave straight lines (Fig. 2) from which the K m and V max values were calculated. Average values for these K m and V max values were calculated from three independent experiments and were found to be K m ϭ 0.44 Ϯ 0.03 M and V max ϭ 19 Ϯ 9 nmol/min ϫ mg for KSMP and were K m ϭ 0.58 Ϯ 0.07 M and V max ϭ 22.8 Ϯ 2.7 nmol/min ϫ mg for the expressed meprin ␤.
The Substrate Recognition of KSMP Analyzed by the Cleavability of the C-subunit Mutants-In order to evaluate the structural requirements of KSMP for its substrate recognition and cleavage, we constructed a series of the C-subunit mutants, with deletions or substitutions in the cluster of acidic amino acids of the C-subunit encompassing its KSMP cleavage site. The first step toward this analysis was to set up an adequate expression system for monitoring the cleavage of the C-subunit and its mutants since we found out that the wildtype C-subunit itself, though fully active as a protein kinase (38), is not cleaved if expressed in bacteria (data not shown) 2 . The rabbit reticulocyte lysate translation system using [ 35 S]methionine was found to be appropriate for that purpose. The wild-type C-subunit produced in this system was cleaved a Amino acid residue that was identified, but the yield could not be determined quantitatively.
b Amino acid residue that was not positively identified.
FIG. 1. Sequence analysis of the peptides released from the C-subunit upon KSMP cleavage. Two peptides (not present at time zero) were formed: a major peptide and a minor peptide. They were rechromatographed on the same column and sequenced. Their sequences (underlined) displayed a very good match with the indicated stretches in the C-subunit (compare with the sequence within the framed area). It should be noted that the major peptide was formed in a 20-fold excess over the minor peptide. The question mark in the sequence of the major peptide indicates an amino acid residue that was not positively identified (cysteine), whereas the two amino acids (Glu and Arg) that are not in bold were identified, but their yield was too low to be determined unequivocally (see also by KSMP with the formation of the clipped product (CЈ, ϳ34 kDa). However, as seen in Fig. 3A, a mutagenic deletion of the whole cluster of acidic amino acids (Scheme 1, ⌬DDYEEEE) in the stretch D 328 -E 334 led to a complete lack of cleavage. The same lack of cleavability occurred also upon removal of the stretch of the four glutamic acid residues (⌬EEEE), i.e. the E 331 -E 334 segment. In contrast, the deletion of D 328 -Y 330 (as in the mutant denoted ⌬DDY) did not significantly affect the KSMP cleavage rate though it slightly reduced this rate (Fig.  3A). These results point to the importance of the cluster of the four glutamic acid residues for the interaction of the proteinase with the C-subunit although a reduction in the local negative charge, or a distortion in the conformation of the kinase as a result of these deletions, may also be responsible for the lack of cleavage. In addition, it should be kept in mind that in this case, where the cluster of acidic amino acids is part of a tail with which the large lobe of the C-subunit embraces its smaller lobe (cf. Fig. 5 in Ref. 4), the loss of cleavability may result from the removal of a segment of the polypeptide backbone, which may well prevent amino acid residues following the deletion from reaching their counterparts in the core segment with which they interact in the wild-type enzyme structure. This kind of "shortened rope" effect may, of course, have a detrimental structural outcome in many deletion studies except when the deletion involves a loop that does not interact with the core of the protein. 2 To avoid some of the difficulties mentioned above and to obtain a more accurate assessment of the individual contribution of each amino acid residue to the KSMP recognition and C-cleavage, we carried out an alanine scanning of the glutamic acid residues within the cluster of acidics in the C-subunit (Scheme 1). Quantitative analysis of the initial rates of cleavage of these mutants showed that there is only a ϳ20% reduction in the cleavage rate when a single glutamic acid residue is mutated into an alanine, even if the Glu to Ala mutation results in a discontinuity in the sequence of the negatively charged residues. A double Glu to Ala substitution as in the E332-333A mutant retained ϳ20% of the initial cleavage rate (Fig. 3B). However, replacement by alanine of three or of all four glutamic acid residues of the C-subunit (E331-333A and E331-334A) resulted in a complete lack of cleavage (Fig. 3B).
Use of Synthetic Peptides Derived from the C-subunit to Establish the Substrate Specificity of KSMP-To complement our results on the specificity of KSMP that were based on the sequence of the cleavage products and on mutations of the C-subunit, we attempted to elucidate the recognition of KSMP by the use of synthetic peptides derived from the segment in the C-subunit that encompasses the KSMP cleavage site. We prepared four synthetic peptides whose sequences were derived from the following segments in the C-subunit: 1) K 319 -I 335 , 2) F 327 -I 335 , 3) S 325 -Y 330 , and 4) Y 330 -I 335 (Fig. 4). Each of these peptides was cleaved by a pure KSMP preparation, and the cleavage products were resolved by reverse-phase HPLC and analyzed by a determination of their amino acid composition or sequence. As seen in Fig. 4, KSMP cleaves the peptides between acidic amino acids in positions that correspond either to the E 332 -E 333 bond, or to the D 328 -D 329 bond in C. The cleavage 2 The C-subunit expressed in E. coli is not myristylated and does not undergo the same post-translational modifications that occur in the mammalian enzyme. The lack of cleavage in spite of full catalytic activity is intriguing and is currently being studied in our laboratory. cleavage of the C-subunit mutants carrying deletions at the cluster of acidic amino acids. The C-subunit and the indicated mutants were translated in a rabbit reticulocyte lysate translation system in the presence of [ 35 S]methionine. The cleavage was allowed to proceed for the indicated time periods as described under "Materials and Methods," and the products (if any) were separated by SDS-PAGE. The protein bands were then visualized by autoradiography, and the resulting autoradiographs were subjected to densitometric scanning for quantitation. The initial rates of cleavage (measured in arbitrary units) were compared as a percentage of the initial rate of cleavage of the wild-type C-subunit (taken as 100%). B, the effect of substitutions by alanine of the various acidic amino acids in the cluster of acidics on the cleavage of the the C-subunit mutant by KSMP. The cleavage and its quantitation were carried out as described in panel A.
site of the peptides in the middle of the cluster of glutamic acid residues is identical to the KSMP cleavage site detected in the C-subunit and described above (E 332 -E 333 ). The lack of cleavage of the D 328 -D 329 bond may be due to steric hindrance, or an unfavorable juxtaposition of this D-D bond of the C-subunit and the KSMP active site. Interestingly, the three-dimensional structure of the C-subunit established by x-ray crystallography showed that, while the backbone of the cluster of acidic amino acids in the C-subunit exhibits relatively high B factors and thus a greater flexibility and availability (5), the D 328 -D 329 bond may be less available for interaction with KSMP since D 329 forms a salt bridge with K 47 (39) that would neutralize the negative charge of D 329 and also restrict its flexibility.
It should be mentioned that the peptides 319 KGPGDTSNFD-DYEEEEI 335 and 325 SNFDDY 330 , but not the peptide 327 FD-DYEEEEI 335 , were found to have an additional (minor) cleavage site between F 327 and D 328 (Fig. 4). The rate of this minor cleavage is at least ten-fold slower than the rate of cleavage between the two aspartic acid residues in these peptides.
The Cleavage of the C-subunit-derived Synthetic Peptides by KSMP Is Reproduced by Meprin ␤-Our recent demonstration that the cleavage of the C-subunit by KSMP can be reproduced by meprin ␤ (13) prompted us to study the specificity of the ␤-subunit of meprin and to compare it with the substrate specificity of KSMP from rat kidney. The clone of human kidney fibroblasts constitutively expressing rat meprin ␤ (13) was used as a source of the enzyme. The substrate specificity of the ␤-subunit purified from these cells was assayed after activation of the precursor enzyme by a limited trypsin digestion (13), resulting in the formation of a catalytically active enzyme with an apparent molecular weight of 100 kDa (Fig. 5, lane 2). The peptides used as substrates for the rat kidney KSMP (see above) were also used in this study of the meprin ␤ specificity. Again, the resulting cleavage products were subjected to analysis by amino acid composition or sequencing and shown to be identical to the cleavage products formed by KSMP. This finding indicates that the ␣-subunit of meprin, which is present in the rat kidney KSMP preparations (13), is not essential for either the cleavage of C or for the cleavage of these peptide substrates. While substrates for the ␣-subunit of meprin are known (26,40), a systematic study aimed at establishing its substrate specificity will have to await the separate stable expression of this subunit.
For a more quantitative comparison between KSMP and meprin ␤, the kinetic parameters (K m and K cat values) of these two enzyme preparations were measured with a synthetic peptide (YEEEEI) containing a single cleavage site to simplify the measurements and their processing. As seen in Table II, the K m and K cat values of KSMP and meprin ␤ with this substrate were found to be quite similar, further supporting our conclusion (13) that meprin ␤ and KSMP are closely related, if not identical.
The Kinetic Parameters of Meprin ␤ Obtained with Peptide Substrates Point to a New Specificity Profile for this Metalloendopeptidase-As a member of the astacin family of metalloendopeptidases, meprin ␤ should have an arylamidase activity (24) and have a preference for bonds flanked by neutral or hydrophobic amino acid residues. Therefore, we assayed the meprin ␤ expressed in the 293 cell line (13) with some synthetic peptides derived from the C-subunit sequence that carried alanine substitutions of the hydrophobic amino acid residues at the edges of the cluster of the four glutamic acid residues in the C-subunit, which are recognized by both KSMP and meprin ␤ (Fig. 6). For the purpose of comparison, Table II also quotes quantitative data obtained in other laboratories regarding the substrate specificity of rat meprin (26,27). This comparison clearly shows that the K cat values obtained for meprin substrates that were proposed to be physiological targets of this proteinase (LHRH, substance P, and bradykinin, which do not contain clusters of acidics) are of the same order of magnitude of the K cat values of meprin ␤ when cleaving the peptide YEEEI derived from the C-subunit (Table II). However, the K m values of meprin for three synthetic peptides containing the cluster of the four Glu residues were found to be 20 -80-fold lower than the values obtained for the hormonal substrates studied earlier in other laboratories (27). Furthermore, the catalytic efficacy of meprin ␤ for these acidic peptides is significantly higher (12-100-fold, as reflected in its K cat /K m values) than the comparative values obtained for the hormones LHRH, substance P, and bradykinin (Table II).
It should be noted that substitution of either the Tyr or the Ile residues by alanine in the peptide YEEEEI does not significantly affect the K m values of meprin ␤, suggesting a relatively low contribution of these hydrophobic residues in creating the affinity between the enzyme and its peptide substrates. However, the Tyr to Ala substitution somewhat decreases (ϳ2.5fold) the catalytic efficacy of the cleavage. The possible contribution of aromatic amino acid residues (such as Phe and Tyr) in enhancing the K cat of meprin ␤ is probably reflected better in the cleavage of SNFDDY, a peptide substrate of this proteinase derived from the cluster of acidic amino acids in the C-subunit that is cleaved between the two aspartic acids. As seen in Table  II, the kinetic parameters of meprin ␤ for the YEEEEI and SNFDDY peptides are considerably distinct. The SNFDDY peptide was found to have a substantially reduced affinity (K m ϭ 80 M), which we believe is associated with a reduction of the overall charge of the cluster of acidics. However, due to an increasingly high rate of cleavage, the K cat /K m ratio for SN-FDDY is within the range determined for the high affinity substrates (Table II).
Is Gastrin a Physiological Substrate of Meprin ␤?-In view of the finding reported here that meprin ␤ has a distinct preference for substrates containing a cluster of negatively charged amino acids, we carried out a search in the data base for peptides and proteins containing stretches of at least four acidic amino acid residues. This search revealed quite a few peptides and proteins with such stretches and, consequently, candidate substrates for meprin ␤. Among these, gastrin seemed to be of particular interest since it is found in the gastrointestinal tract and in the kidney where it can be exposed to meprin. Discovered in 1905 as an acid-stimulating factor (41), gastrin is now regarded also as an important growthstimulating hormone (42). Gastrin occurs in multiple hormonal forms that are produced as a result of proteolytic processing and may contain from 71 to 6 amino acid residues (32,43). Its most abundant forms (G-34 and G-17) contain a stretch of five glutamic acid residues (Fig. 7) (34). All known biological effects of gastrin reside in the conserved carboxyl-terminal tetrapeptide amide WMDF-NH 2 , which is common to all gastrins and also to the cholecystokinins (Scheme 2B).
Analysis of the fragments resulting from gastrin cleavage by either KSMP or meprin ␤ revealed proteolysis at three distinct sites (Fig. 7). Two adjacent cleavage sites were found within the cluster of glutamic acid residues. For the sake of simplicity, the kinetic parameters of the cleavages were measured on the gastrin fragments rather than on the whole molecule (Fig. 7). The first fragment, MEEEEEAY, accommodated the two alternative cleavage sites identified in gastrin 17. The affinity of meprin ␤ for this peptide was found to be very similar to that observed with the C-subunit-derived peptides (Table II). The FIG. 6. Determination of the kinetic parameters (K m and V max ) of trypsin-activated meprin ␤ for the synthetic peptide substrates. The cleavage reactions were allowed to proceed at different concentrations of peptides (from 5 to 100 M), and the amount of the product formed after different time intervals was quantitated by HPLC separation and measuring of the appropriate peak area. The activity of the meprin ␤ was expressed in mol of the product formed per 1 min by 1 mg of the enzyme. The activity versus the initial concentration of the peptide was plotted in double reciprocal coordinates. The calculation of the kinetic parameters (K m , V max , K cat , and their ratio) is described under "Materials and Methods."  cleavage of the carboxyl-terminal fragment of gastrin, EEAY-GWMDF, resulted in clipping off its last two residues (cleavage at the Met-Asp bond). Since any modification in the carboxylterminal tetrapeptide amide WMDF-NH 2 grossly reduces or abolishes all its known biological effects (33,34,44), this cleavage will inactivate gastrins. Notably the K m for the cleavage at this site was found to be substantially higher than the K m for the cleavage at the sites in the acidic cluster. However, the K cat for this cleavage was much higher, and thus, it had a comparable K cat /K m value (Table II). The decreased affinity for the EEAYGWMDF peptide compared with the site in the cluster of glutamic residues was actually expected in view of the findings reported here regarding the important contribution of clustered acidics to the affinity of meprin ␤ for its substrates.
It should be emphasized, however, that in spite of the fact that clusters of acidics are recognized and cleaved by meprin ␤ with a significantly lower K m , hydrophobic amino acid residues most likely play an important role in the cleavage of physiological substrates by this proteinase. They do so with a high K m but also with a high K cat and thus with a high catalytic efficacy. The evidence supporting the importance of hydrophobic amino acid residues is summarized in the following. (i) The protein substrate (C-subunit) with which KSMP was originally discovered (1-3), as well as the EGF-and insulin-receptor kinases (7,10,11), contain hydrophobic amino acid residues within or adjacent to their cluster of acidic amino acids. (ii) The dye 1-anilino-8-naphthalenesulfonate, which is known to bind to hydrophobic sites in proteins, inhibits the cleavage of the Csubunit by KSMP (45). (iii) While a copolymer composed of Glu and Tyr is cleaved by KSMP, a polymer of Glu amino acid residues alone is not cleaved by it and acts as a competitive inhibitor of this proteinase (7,45). (iv) The hydrophobic proteinase inhibitor chymostatin inhibits the cleavage of the Csubunit by KSMP though at relatively high concentrations (Ն 10 Ϫ3 M) (3). (v) In general, members of the astacin family of peptidases have been shown to cleave peptides containing hydrophobic amino acid residues and to possess an arylamidase activity (26). It is, therefore, possible that the degradation of gastrin by meprin ␤ may draw its affinity from the interaction of meprin ␤ with the cluster of acidics (K m ϭ 1.5 M) while the fast cleavage at the Met-Asp bond (which will inactivate gastrin) may originate from the high K cat value (63.4 s Ϫ1 ) for this cleavage. It is also possible that the cleavage at the cluster of acidics facilitates the subsequent evacuation of the active site.
The Prosequence of Meprin ␤ Contains Seven Acidic and Seven Hydrophobic Amino Acid Residues within a Stretch of Twenty Amino Acids and Differs from Meprin ␣ in This Respect-Meprins are synthesized as inactive proenzymes whose prosequences are removed when they are called upon to act. One of the plausible mechanisms for an auto-inhibition of enzymes is the blocking of the active site by a substrate-like prosequence, which can then be removed for the purpose of activation by the enzyme itself or by another enzyme, in response to an appropriate regulatory stimulus. In view of the specificity profile described above for meprin ␤, we looked into its prosequence, searching for features related to the specificity ensuing from this study and for a difference between meprin ␤ and meprin ␣ that might reflect the known difference in specificity between these two subunits (24). At the same time, we attempted to find out whether, in view of the specificity implicated in this study, there may be evidence supporting or disproving an autoinhibition mechanism of the type described above. Scheme 2 illustrates such a comparison for the rat, mouse, and human prosequences of meprins ␣ and ␤. From this comparison, it is evident that (i) the prosequence of meprin ␤ possesses a stretch of acidic and hydrophobic amino acids (in line with the specificity profile described above). This is especially prominent in the stretch D 55 -I 74 , say in the rat, in which out of 20 amino acids, seven are acidic (mostly Asp), seven are hydrophobic (mostly Ile and Leu), and not one is a basic amino acid. (ii) The parallel stretch in promeprin ␣ has six hydrophobics but only three acidics and one basic amino acid. (iii) Five of the acidic and hydrophobic residues in the ␤ prosequence come in pairs of DI or DL, raising the possibility that this repeated motif may have a distinct inhibitory significance. These suggestions regarding the difference in specificity between the ␤ and the ␣ subunits of meprin, and regarding the molecular basis of their autoinhibition, will hopefully shed light from a new angle on the biorecognition of these metalloendopeptidases. However, their ultimate proof will have to rely on molecular modeling studies (based on the astacin structure (46)) and possibly to await the determination of the three-dimensional structures of these precursors by x-ray crystallography. CONCLUSION This paper provides evidence to show that the cleavage of KSMP in the C-subunit occurs at E 332 -E 333 , within the cluster of acidic amino acids (D 328 -E 334 ) of this kinase. The K m values of KSMP and of meprin ␤ (which we recently showed to reproduce KSMP activity) for the C-subunit are below 1 M. The K m for peptides containing a stretch of four Glu residues are in the micromolar range, suggesting a significant contribution of this cluster of acidics to the biorecognition of meprin ␤. This conclusion is supported by experiments with a series of C-subunit mutants with deletions and mutations in the cluster of acidics. In addition, hydrophobic amino acids vicinal to the cleavage site seem to play an important role in the function of this proteinase, increasing its K cat .
These studies unveil a new specificity profile for meprin ␤, suggesting new candidate substrates that are 1-2 orders of magnitude better (lower in their K m and higher in their K cat ) than substrates commonly used for meprin ␤. Specifically, the search for substrates that have such a cluster of acidics and hydrophobics and are accessible to meprin ␤ under physiological conditions, pointed at the hormone gastrin as a potential target. Indeed, we show here that, at least in vitro, meprin ␤ cleaves gastrin at its cluster of five glutamic acid residues and also at the Met-Asp bond within its WMDF-NH 2 sequence SCHEME 2. A, alignment of prosequence segments of the ␣and ␤-subunits of meprin from different species. The acidic amino acid residues are shown in bold, and those conserved in the these species are put in a frame. Conserved hydrophobic amino acid residues are in an outlined font. The numbers indicate the positions of these residues in the sequence of the mouse ␣-subunit of meprin. B, comparison of the carboxylterminal fragments of gastrin and of cholecystokinin. The conserved carboxyl-terminal tetrapeptides that are identical in the two hormones are framed. whose unmodified structure has been claimed to be indispensable for all the known biological activities of gastrins (33,34). The latter meprin cleavage will lead to the inactivation of gastrin and thus to the control of its activity. In view of the fact that meprins have been implicated in key biological processes such as growth, development, and tissue remodeling, the search for such target substrates, and the demonstration that they occur in vivo, is quite an important task.