INTRODUCTION

Some peptidases act on peptides containing D-amino acids. Soluble Streptomyces carboxypeptidase DD catalyzes not only the transpeptidation reaction on the peptide intermediate in peptidoglycan biosynthesis, but also the hydrolysis of N,N-diacetyl-L-lysyl-(D-Ala)₂ in water (1). A D-peptidase has been purified and characterized from an actinomycete, although it is not strictly specific toward peptides containing D-amino acids (2). In Enterococcus, the vanX gene product, (D-Ala)₂ hydrolase, plays a role in vancomycin resistance (3). The chemically synthesized "D-enzyme" of human immunodeficiency virus type 1, in which all of the amino acids were replaced with the corresponding D-amino acids, displays D-stereospecificity (4). We discovered D-aminopeptidase (EC 3.4.11.19) from Ochrobactrum anthropi and found that its primary structure is similar to the -lactamases and penicillin-binding proteins (5, 6). The enzyme acts mostly on peptides with D-Ala at the NH₂ terminus to yield D-amino acids and does not act on D-amino acid derivatives with bulkier substituents. We proposed that D-aminopeptidase is a new "penicillin-recognizing enzyme" (1), based on its primary structure, inhibition by -lactam compounds, and the ability to catalyze peptide bond formation in organic solvents, although the enzyme does not show -lactamase activity (6, 7).

In this paper, we describe the screening of soil microorganisms for D-stereospecific endopeptidases using a synthetic peptide ((D-Phe)₄), characterization of the new enzyme alkaline D-peptidase (ADP),¹ as well as cloning and sequencing of the adp gene from Bacillus cereus strain DF4-B.

EXPERIMENTAL PROCEDURES

DEAE-Toyopearl 650 M, Butyl-Toyopearl 650 M, and HPLC G-3000 SW and ODS-80Ts columns were purchased from Tosoh Corp. (Tokyo, Japan); Superdex 200 was from Pharmacia (Uppsala); and Cosmosil 5C18-MS from Nacalai Tesque (Kyoto, Japan). Membrane filters (Diaflo Ultrafilter PM-30) and the Hollow Fiber cartridge system (Hollow Fiber Ultrafilter H10-P10) were obtained from Amicon, Inc. (Beverly, MA). Chemicals other than the peptides described below were from commercial sources and were used without further purification.

Peptide substrates used to screen and test the substrate specificity were synthesized from D- and L-Phe. NH₂ and COOH termini were protected by Boc (8) and methyl groups, respectively. Isobutyl chloroformate (9) and carbodiimide (10) condensed the monomer to a dimer and the dimer to a tetramer, respectively. The following peptide derivatives were synthesized: Boc-D-Phe, D-Phe tert-butyl ester, (D-Phe)₂·HCl, Boc-(D-Phe)₂, (D-Phe)₂ methyl ester·HCl, Boc-(D-Phe)₂ methyl ester, (D-Phe)₃·HCl, Boc-(D-Phe)₃, Boc-(D-Phe)₃ methyl ester, Boc-(D-Phe)₃ tert-butyl ester, (D-Phe)₄·HCl, Boc-(D-Phe)₄, Boc-(D-Phe)₄ methyl ester, L-Phe methyl ester·HCl, (L-Phe)₂ methyl ester·HCl, (L-Phe)₃·HCl, (L-Phe)₄·HCl, Boc-(L-Phe)₄, Boc-(L-Phe)₄ methyl ester, (D-Phe)₂-D-Tyr·HCl, D-Tyr-(D-Phe)₂·HCl, D-Phe-(L-Phe)₂·HCl, L-Phe-(D-Phe)₂·HCl, (D-Phe)₂-L-Phe·HCl, (L-Phe)₂-D-Phe·HCl, D-Phe-L-Phe-D-Phe· HCl, L-Phe-D-Phe-L-Phe·HCl, D-Phe-L-Phe·HCl, and L-Phe-D-Phe·HCl. The details will be reported elsewhere.²

We screened the ability of microorganisms to hydrolyze (D-Phe)₄ in LB medium (11) in enriched cultures at 30 °C. (D-Phe)₄ was dissolved in Me₂SO (10%, w/v) and then added to 2 ml of LB medium containing soil samples. The mixture was then aerobically shaken for 2 days. A loopful of the culture broth was transferred to the same medium and aerobically incubated under the same conditions. A small portion of the culture broth was streaked onto a plate of the same medium containing 1.5% agar and incubated at 30 °C overnight. Strains forming clear zones around the colonies were isolated, and (D-Phe)₄ degradation in the liquid culture was monitored by TLC (chloroform/methanol/acetic acid = 8:2:1 or 10:2:1) and visualized with ninhydrin. (D-Phe)_n (n = 1-4) were well separated by TLC with the solvent. A bacterial strain isolated from a soil of Kanagawa Prefecture, Japan completely degraded the substrate by forming (D-Phe)₂.

Taxonomical studies of strain DF4-B showed that it was B. cereus because the Gram-positive rods were aerobic, spore-forming, motile, catalase-positive, egg yolk-positive, and lysozyme-resistant, and they formed acid from glucose (12). Details will be reported elsewhere.²

ADP activity was routinely assayed at 30 °C by measuring the production of (D-Phe)₂ from (D-Phe)₄. The reaction mixture was composed of 1 µmol of (D-Phe)₄, 10 µl of Me₂SO, 50 µmol of Tris-HCl, pH 9.0, and 1 µmol of MgSO₄, and the assay was started by addition of the enzyme in a total volume of 500 µl. The reaction was terminated with 50 µl of 2 N HClO₄, and the amount of (D-Phe)₂ formed was estimated with a Waters 600E HPLC apparatus equipped with a Cosmosil 5C18-MS reverse-phase column (4.6 × 150 mm) at a flow rate of 1.0 ml/min, using 35% methanol in 5 mM KH₂PO₄/H₃PO₄ buffer, pH 2.9. Absorbance of the eluate was monitored at 254 nm. One unit of enzyme activity is defined as the amount of enzyme that catalyzes the formation of 2 µmol of (D-Phe)₂ from 1 µmol of (D-Phe)₄.

The substrate specificity was examined qualitatively by thin-layer chromatography first and then was quantitatively assayed by the following methods. (a) The enzyme activity toward peptide substrates was measured as described above with 2.5 units of enzyme. The amounts of (D-Phe)₂, D-Phe, L-Phe-D-Phe, Boc-D-Phe, and Boc-(D-Phe)₂ were quantitatively assayed by HPLC on a Cosmosil 5C18-MS reverse-phase column (4.6 × 150 mm) at a flow rate of 1.0 ml/min, using the following solvent system of 5 mM KH₂PO₄/H₃PO₄ buffer, pH 2.9: methanol = 13:7 or 9:11 (v/v). To determine the kinetic constants of the enzyme for the peptide substrate, a reaction mixture containing 0.1-12.5 µmol of the substrate, 50 µmol of Tris-HCl, pH 9.0, 1 µmol of MgSO₄, 10 µl of Me₂SO, and 100 µl of the enzyme solution in a total volume of 500 µl was used. (b) The enzyme activity toward -lactam compounds was determined by measuring the consumption of the substrate in a reaction mixture containing 100 µmol of potassium phosphate buffer, pH 7.0, 2.5 µmol of -lactam compound, and 0.46 units of the enzyme in a total volume of 1 ml. After incubation at the same temperature, 100 µl of the reaction mixture was added to 900 µl of 100 mM KH₂PO₄/H₃PO₄ buffer, pH 2.9, to stop the reaction. After centrifugation (15,000 × g, 10 min), the amount of -lactam compounds in the supernatant was determined by HPLC on an ODS-80Ts column (46 × 150 mm) with a solvent of 25% methanol in 5 mM KH₂PO₄/H₃PO₄ buffer, pH 2.9, and of 50 mM potassium phosphate buffer, pH 7.0, for ampicillin and penicillin G, respectively. To determine the kinetic constants for -lactam compounds, potassium phosphate buffer, pH 7.0, was used in place of Tris-HCl, pH 9.0, and MgSO₄ and Me₂SO were omitted. The enzyme solution used was 0.038 units (0.31 µg) for (D-Phe)₄, (D-Phe)₃, and L-Phe-(D-Phe)₂; 0.06 units (0.49 µg) for ampicillin and penicillin G; 0.19 units (1.48 µg) for (D-Phe)₂-L-Phe and L-Phe-D-Phe-L-Phe; and 1.0 unit (8.1 µg) for (D-Phe)₂. A Hanes-Woolf plot was used for the estimation.

Column chromatographies were done at temperatures lower than 5 °C. Potassium phosphate buffer, pH 7.0, containing 0.1 mM EDTA was used throughout the purification. The strain was aerobically cultivated at 30 °C for 48 h and shaken at 150 rpm in 352 batches in 2-liter flasks, each containing 400 ml of nutrient broth, which consisted of 1% meat extract, 1% polypepton, and 0.5% NaCl, in tap water, pH 7.0 (total of 140 liters). Cells were removed by centrifugation at 9500 × g for 10 min. The supernatant was concentrated with an Amicon Hollow Fiber cartridge system (0.4 kg/cm²), and a crude enzyme was obtained. The pellet formed between 50 and 80% ammonium sulfate saturation was dissolved in 0.01 M buffer and dialyzed with the same buffer. The enzyme solution was added to the first DEAE-Toyopearl column (6.0 × 35 cm), which had been equilibrated with 0.01 M buffer. The enzyme was eluted with 0.1 M buffer containing 0.1 M NaCl, and the active fractions were concentrated by ammonium sulfate. The dialyzed enzyme solution was added to the second DEAE-Toyopearl column (5 × 20 cm), which had been equilibrated with 0.01 M buffer. The column was eluted with a linear gradient of 0.01 M buffer (1.2 liters) to 0.1 M buffer containing 0.1 M NaCl (1.2 liters). The active fractions were then brought to 30% ammonium sulfate saturation. The enzyme solution was added to the first Butyl-Toyopearl 650 M column (5 × 20 cm). The active fractions were eluted with a linear gradient of ammonium sulfate (30 to 0% saturation) in 0.01 M buffer. The active fractions were brought to 30% ammonium sulfate saturation. The enzyme solution was added to the second Butyl-Toyopearl 650 M column (5 × 12 cm). The active fractions that eluted with a linear gradient of ammonium sulfate (30 to 0% saturation) in 0.01 M buffer were combined and brought to 80% ammonium sulfate saturation. The mixture was centrifuged at 28,000 × g for 20 min, and the resulting pellet was dialyzed against the same buffer. The enzyme solution was added to the third DEAE-Toyopearl column (5 × 12 cm). The enzyme was eluted with a linear gradient of 0.01-0.1 M buffer containing 0.2 M NaCl. The active fractions were combined, dialyzed, concentrated by ultrafiltration, and applied to a column of Superdex 200 equilibrated with 0.05 M buffer containing 0.1 M NaCl. The column was eluted by fast liquid protein chromatography (Pharmacia) at 1.0 ml/min, and the active fractions were collected and concentrated by ultrafiltration.

¹H NMR spectra were measured with a JEOL EX 400 apparatus. Protein was assayed, SDS-polyacrylamide gel electrophoresis was done, and the M_r of the enzyme was estimated as described previously (6). Purified ADP (720 µg) was digested with lysyl-end peptidase (5.0 µg; Wako, Tokyo, Japan) at 30 °C for 16 h. The digest was separated by HPLC on a reverse-phase column (ODS-80Ts) in a 10-80% linear gradient of acetonitrile containing 0.1% trifluoroacetic acid at a flow rate of 0.5 ml/min with continuous monitoring of the absorbance at 215 nm. To determine the NH₂-terminal amino acid sequence, the enzyme samples were covalently bound to Sequelon-arylamine and Sequelon-diisothiocyanate membranes and then analyzed with a Prosequencer 6625 automatic protein sequencer (Milligen/Biosearch). The molecular mass of the enzyme was estimated with a PE-Sciex API III triple quadrupole mass spectrometer equipped with an ionspray ion source in the positive ion mode (Sciex, Thornhill, Ontario, Canada).

B. cereus genomic DNA was isolated as described (6). Escherichia coli cells were cultured in LB medium with 100 µg/ml ampicillin. Plasmids were purified using the QIAGEN plasmid purification kit. B. cereus genomic DNAs were partially digested with MboI and fractionated by sucrose density gradient ultracentrifugation (5-25%; 100,000 × g, 16 h). DNA fragments of ~3-6 kb were purified and ligated into BamHI-digested and dephosphorylated pUC118 by T4 ligase. Ampicillin-resistant transformants expressing ADP activity were directly selected by monitoring halo formation from (D-Phe)₄ because the host E. coli JM109 cells did not show ADP activity. One of ~17,000 transformants exhibited ADP activity, and it harbored a plasmid designated pBDP2 with a 5-kb DNA insert. For subcloning, pBDP2 was digested with EcoRI and SalI, and the resulting 1.3-kb fragment was introduced into the EcoRI and SalI sites of pUC118 to yield pBDP22.

B. cereus genomic cDNAs were completely digested with EcoRI and SalI and then size-fractionated by sucrose density gradient ultracentrifugation to give DNA fragments of ~1.8 kb. These fragments were ligated by T4 ligase into pUC118 that had been digested with EcoRI and SalI and dephosphorylated and then introduced into E. coli JM109. Ampicillin-resistant transformants were screened with the DNA insert of pBDP22 as a probe by colony hybridization. One clone (designated pADP1) with an ~1.8-kb DNA insert was selected for further analysis.

Colony hybridization and Southern blot hybridization using ³²P-labeled probes prepared by nick translation were performed as described by Maniatis et al. (11). The DNA sequence was determined by the dideoxy chain termination procedure (28) using -³⁵S-dCTP or with an automatic gene sequencer (ALF red sequencers, Pharmacia).

RESULTS

B. cereus was isolated from a soil sample and considered as a likely source of the enzyme. When the strain was cultured in 400 ml of nutrient broth, the enzyme activity was detected in the culture broth, and its formation was associated with the growth of the microorganism. We harvested the supernatant at 48 h to avoid lowering the specific activity of the enzyme by cell lysis.

Since the enzyme was constitutive, the substrate (D-Phe)₄ was omitted from the medium in large-scale culture. The enzyme isolated from the supernatant of 140 liters of culture broth was electrophoretically pure. A summary of the purification procedure for the enzyme is shown in Table I. The enzyme was purified ~300-fold with an 8% yield.

Table I. Purification of ADP from the culture broth of B. cereus strain DF4-B Step Total protein Total activity Specific activity Yield mg units units/mg % Culture broth 22,400 9280 0.414 100 Crude enzyme 15,600 6350 0.407 68 Ammonium sulfate 9960 4860 0.488 52 DEAE-Toyopearl (1st) 4240 4810 1.13 52 DEAE-Toyopearl (2nd) 2260 3730 1.42 40 Butyl-Toyopearl (1st) 440 2230 5.02 24 Butyl-Toyopearl (2nd) 375 2200 5.87 24 DEAE-Toyopearl (3rd) 16.8 1130 67.3 12 Superdex 200 6.13 768 125 8

The enzyme was judged to be homogeneous by SDS-polyacrylamide gel electrophoresis and HPLC on a TSK G-3000 SW column, as each of these procedures yielded a single band or a single peak. Fig. 1 (A and B) shows the results of SDS-polyacrylamide gel electrophoresis and the estimation of the M_r by HPLC, respectively. The M_r of the subunit calculated was ~36,000 as determined by SDS-polyacrylamide gel electrophoresis. That of the native enzyme was ~37,000 according to gel filtration chromatography, indicating that the native enzyme was a monomer. Mass spectrophotometry revealed that the M_r of the enzyme was 37,952. The absorption of the purified enzyme in 0.01 M potassium phosphate buffer, pH 7.0, was maximal at 281 nm.

The optimal pH for the activity of the enzyme was measured in several buffers at various pH values (final concentration of 0.1 M): potassium phosphate, pH 6.4-7.9; Tris-HCl, pH 7.5-9.0; ethanolamine HCl, pH 8.5-10.9; and glycine/NaCl/NaOH, pH 9.9-12.8. When the velocity of the observed maximal activity at pH 10.3 was taken as 100%, the relative activities at pH 7.1, 7.5, 7.9, 8.5, 9.0, 9.5, 9.9, 10.3, 11.0, 11.3, 11.8, and 12.3 were 7.9, 15, 30, 46, 79, 91, 93, 100, 94, 80, 76, and 10%, respectively. The enzyme activity was maximal at 45 °C. About 60% activity remained after incubation at 43 °C in 0.1 M potassium phosphate buffer, pH 8.0, for 10 min. No activity was lost between pH 5.0 and 10.0 after incubation at 30 °C for 1 h in 0.05 M buffer at various pH values.

The substrate specificity of the enzyme was examined as shown in Table II. The enzyme was active toward (D-Phe)₃ and (D-Phe)₄, forming (D-Phe)₂ and D-Phe. The enzyme was also active toward tripeptides with D-Tyr at the COOH or NH₂ terminus and toward Boc-(D-Phe)_n (n = 2-4), forming Boc-D-Phe, (D-Phe)₂, and D-Phe. The enzyme had esterase activity toward D-Phe methyl ester and (D-Phe)₂ methyl ester. The products from Boc-(D-Phe)₃ tert-butyl ester were Boc-D-Phe, D-Phe, and D-Phe tert-butyl ester. The enzyme was not active toward L-Phe methyl ester, (L-Phe)₂ methyl ester, (L-Phe)₄, Boc-(L-Phe)₄, Boc-(L-Phe)₄ methyl ester, (D-Val)₃, (D-Leu)₂, and (D-Ala)_n (n = 2-5). These properties indicate that the enzyme is an endopeptidase that acts D-stereospecifically upon peptides composed of aromatic D-amino acids. On the other hand, a dimer was formed when D-Phe methyl ester and D-Phe amide were the substrates. Eight stereoisomers of the phenylalanine trimer were synthesized, and their effectiveness as substrates for the enzyme was tested. The enzyme recognized the configuration of the second D-Phe of tripeptides and catalyzed the hydrolysis of the second peptide bond from the NH₂ terminus. The calculated V_max/K_m values for the peptides containing L-Phe were lower than those for (D-Phe)₃. The enzyme also showed -lactamase activity toward ampicillin and penicillin G. The calculated V_max values of the enzyme for -lactam compounds were about the same as those for (D-Phe)₃ and (D-Phe)₄, while the K_m values were several hundred times larger. On the other hand, carboxypeptidase DD (13) and D-aminopeptidase (6) activities were undetectable.

Table II. Substrate specificity of ADP OMe, methyl ester; pNA, p-nitroanilide; OtBu, tert-butyl ester. Substrate Relative activity Km Vmax Vmax/Km % mM units/mg units/mg·mM (D-Phe)6 1.8a (D-Phe)4 100a 0.398 199 500 (D-Phe)3 90a 0.127 130 1020 (D-Phe)2 0.2b 50.1 13.7 0.270 D-Phe-L-Phe <0.1b (D-Phe)2-L-Phe 14.9a 0.522 30.6 59.0 L-Phe-(D-Phe)2 119c 0.455 154 346 L-Phe-D-Phe-L-Phe 28.1c 1.63 66.0 41.0 D-Tyr-(D-Phe)2 83.6b (D-Phe)2-D-Tyr 83.6a D-Phe-OMe 15,a 1.8b D-Phe-NH2 0.1,a 0.1b D-Phe-pNA 4.2b Boc-(D-Phe)4 1.8,d 0.8e Boc-(D-Phe)3 3.2,d 1.1e Boc-(D-Phe)2 7.0d Boc-(D-Phe)3-OtBu 1.2,d 0.3e Boc-(D-Phe)4-OMe 0.5,d 0.2e Boc-(D-Phe)3-OMe 0.7,d 0.3e Boc-(D-Phe)2-OMe 1.4d Ampicillin 8.9f 73.1 262 3.58 Penicillin G 9.7f 48.9 250 5.11 a  Formation of (D-Phe)2. b  Formation of D-Phe. c  Formation of L-Phe-D-Phe. d  Formation of Boc-D-Phe. e  Formation of Boc-(D-Phe)2. f  Consumption of a -lactam compound was measured.

The following compounds were inert as substrates: (L-Phe)₄, (L-Phe)₃, D-Phe-L-Phe-D-Phe, D-Phe-(L-Phe)₂, (L-Phe)₂-D-Phe, (L-Phe)₂, L-Phe-D-Phe, L-Phe methyl ester, L-Phe amide, Boc-(L-Phe)₂, Boc-(L-Phe)₄ methyl ester, D-Leu p-nitroanilide, D-Ala p-nitroanilide, D-phenylglycine amide, (D-Ala)₅, (D-Ala)₄, (D-Ala)₃, (D-Ala)₂, (D-Val)₃, (D-Leu)₂, and DL-Ala-DL-Phe.

We investigated the effect of metal ions on the enzyme activity. The enzyme (30 units/ml) was dialyzed with 0.01 M potassium phosphate buffer, pH 7.0, containing 10 mM EDTA for 48 h and then diluted with the same buffer 20 times. Thereafter, 5 mM concentrations of cations were added in place of MgSO₄, and the enzyme activity was assayed by the standard procedure. The enzyme activity was enhanced by Mg²⁺ (138%), Mo³⁺ (130%), and Ba²⁺ (123%). We also measured the enzyme activity after incubation at 30 °C for 30 min with various compounds (at 5 mM unless otherwise noted). The activity was inhibited 94% by phenylmethylsulfonyl fluoride, 76% by Ag⁺, 74% by Fe²⁺, and 32% by Hg²⁺. The activity was not lost upon incubation with the following agents: Na⁺, K⁺, Ba²⁺, Mg²⁺, Mn²⁺, Sn²⁺, Pb²⁺, Ca²⁺, Ni²⁺, Co²⁺, Cd²⁺, Cu²⁺, Zn²⁺, Al³⁺, Fe³⁺, Cr³⁺, 5,5-dithiobis(2-nitrobenzoic acid), hydroxylamine, N-ethylmaleimide, EDTA, EGTA, 8-oxyquinoline, 2,2-dipyridyl, o-phenanthroline, Tiron (1,2-dihydroxybenzene-3,5-disulfonic acid), NaF, sodium azide, KCN, monoiodoacetate, 2-mercaptoethanol, dithiothreitol, DL-penicillamine, D-cycloserine, and p-chloromercuribenzoate. These results indicate that the enzyme is a serine peptidase.

We measured the time course of the (D-Phe)₄ degradation. As shown in Fig. 2, (D-Phe)₄ was hydrolyzed to (D-Phe)₂ and D-Phe. No (D-Phe)₃ was detected. These results coincide with the kinetic properties of the enzyme described above. The mode of action of the enzyme was examined with the synthetic substrates D-Tyr-(D-Phe)₂ and (D-Phe)₂-D-Tyr as shown in Fig. 3. When D-Tyr-(D-Phe)₂ was the substrate, D-Phe was released first, and then D-Tyr was slowly formed. When (D-Phe)₂-D-Tyr was used as the substrate, D-Tyr was released first, and then D-Phe was slowly formed. In both reactions, the second peptide bond from the NH₂ terminus of the substrate was hydrolyzed first. These results show that the enzyme acts as a D-stereospecific dipeptidyl endopeptidase.

E. coli JM109 transformants expressing ampicillin resistance and ADP activity were screened by halo formation on LB agar containing (D-Phe)₄. Plasmid pBDP22 with a 1.3-kb insert conferred ADP activity in E. coli and was found to contain a lacZ-adp gene fusion, which encodes a -galactosidase (1-12 amino acids)-ADP (²⁵⁰GAT to the COOH terminus) hybrid protein, by sequence analysis. Genomic Southern hybridization using the DNA insert of pBDP22 as a probe indicated that the 1.8-kb EcoRI-SalI fragment contained the full-length adp gene. One clone (pADP1) that transformed E. coli into expressing ADP and that carried a 1.8-kb insert was further analyzed. Transcription in the plasmids appeared to be governed by an extant promoter of the adp gene because ADP activity was expressed when the direction of the transcription of the insert was opposite that of the original plasmid with pUC119.

The nucleotide sequence of the 1.8-kb EcoRI-SalI fragment revealed a single open reading frame (ORF) that probably initiates at the ¹⁰⁶ATG codon preceded by a potential ribosome-binding site (⁹¹GGAG). Translation of the ORF encoded a predicted protein of 388 amino acids with an M_r of 42,033, with an amino acid sequence identical to those obtained by NH₂-terminal amino acid sequencing of the six peptides prepared from purified ADP as shown in Fig. 4. Considering that ADP was secreted in the culture broth and that the M_r of the predicted ORF (42,033) was larger than those estimated by SDS-polyacrylamide gel electrophoresis and HPLC, ADP is synthesized with a signal peptide. In fact, the predicted ORF exhibits a positively charged NH₂ terminus, followed by a hydrophobic stretch with a high leucine content. This domain closely resembles those of the signal peptides of exported proteins in Bacillus species (14). The NH₂-terminal amino acid was suggested to be serine (²²³AGT) based on the mass spectrometry results (M_r 37,952) with purified ADP. This ORF (²²³AGT to ¹²⁶⁷AAG) encodes a protein with a calculated M_r of 37,926, which is in agreement with those estimated by other methods. The observed difference of 26 mass units between the M_r deduced from the primary structure and that calculated by mass spectrometry was probably caused by the formation of an oxazolidinone ring at the NH₂-terminal Ser. However, the exact molecular structure of the NH₂ terminus is not clear.

Alignment by the SWISS-PROT and NBRF-PIR data bases using the BLAST, FASTA, and DNASIS programs showed that the deduced primary structure of ADP is similar to those of carboxypeptidase DD from Streptomyces R61 (35.0% identical over 346 amino acids) (13), penicillin-binding proteins from Streptomyces (Nocardia) lactamdurans (28.1% identical over 263 amino acids) (15) and from Bacillus subtilis (28.5% identical over 309 amino acids) (16), class C -lactamases from Serratia marcescens (24.9% identical over 217 amino acids) (17), class C -lactamases from Enterobacter cloacae (25.1% identical over 191 amino acids) (18), fimbrial protein D from Dichelobacter nodosus (24.1% identical over 261 amino acids) (19), D-aminopeptidase from O. anthropi (27.5% identical over 182 amino acids) (20), and esterase from Pseudomonas sp. (30.5% identical over 154 amino acids) (21). Fig. 5 shows the results of the alignment of the primary structure of ADP with that of Streptomyces R61 carboxypeptidase DD. The sequence Ser-Xaa-Xaa-Lys is perfectly conserved in this class of enzymes, as shown in Fig. 6. A triad sequence similar to the His-Tyr-Gly commonly found in -lactamases could not be found (22).

DISCUSSION

In this study, we synthesized (D-Phe)₄ in eight steps to screen for a new microbial endopeptidase. A strain degrading (D-Phe)₄, indicated by a clear zone on LB agar plates, was isolated and identified as B. cereus. We purified the enzyme 300-fold to homogeneity from 140 liters of culture broth in 8% yield. We designated the enzyme as alkaline D-peptidase. It is the first D-stereospecific endopeptidase isolated with -lactamase activity.

Since the NH₂-terminal amino acid of the natural enzyme could not be determined, probably because it was modified, the M_r of the enzyme was determined by mass spectrometry. The observed M_r of the enzyme was 37,952. Serine (²²³AGT) appears to be the first residue of the mature protein for the following reasons. Bacillus species seem to prefer small neutral amino acid residues at positions 1 and 3 of the signal peptide (14), especially at Ala³-Xaa-Ala¹, where cleavage occurs after the carboxyl-terminal alanine. However, both of the alanine residues are occasionally substituted by other amino acid residues with a short side chain. In fact, -mannosidase produced by Bacillus sp. has the sequence Ser³-Met-Ser¹-Ser⁺¹ (23), which is similar to the sequence Ser³-Val-Ser¹-Ser⁺¹ found in ADP (Fig. 4). However, assuming that the NH₂-terminal amino acid of the secreted mature protein is serine (²²³AGT), the signal peptide of ADP (39 amino acids) is longer than those of other exported proteins of bacilli (18-35 amino acids) (14). Therefore, ADP might be synthesized as a propeptide with a few extra residues between the signal peptide and mature regions, like the -amylase from B. subtilis (24) and the -lactamases from B. cereus (25) and Bacillus licheniformis (14).

We synthesized several derivatives of D-Phe oligopeptides. The enzyme showed D-stereospecific dipeptidyl aminopeptidase and endopeptidase activities. The calculated V_max/K_m values for (D-Phe)₃ and (D-Phe)₄ were ~1000 times higher than for (D-Phe)₂. The enzyme was active toward (D-Phe)_n, Boc-(D-Phe)_n, (D-Phe)_n methyl ester, D-Phe-NH₂, Boc-(D-Phe)_n methyl ester, and Boc-(D-Phe)_n tert-butyl ester, but not toward (D-Ala)_n (n = 2-4), (D-Val)₃, and (D-Leu)₂. The enzyme digested (D-Phe)₂-D-Tyr into (D-Phe)₂ and D-Tyr and digested D-Tyr-(D-Phe)₂ into D-Tyr-D-Phe and D-Phe. The enzyme attacked (D-Phe)₄, forming (D-Phe)₂, which was then slowly hydrolyzed to D-Phe. The product (D-Phe)₃ was not formed. The enzyme had an optimal pH of ~10.3, and it was stable in a relatively wide pH range. The enzyme was inhibited by phenylmethylsulfonyl fluoride, indicating that a serine residue is responsible for exerting the enzyme activity. Thus, the enzyme was named alkaline D-peptidase (D-stereospecific peptide hydrolase, EC 3.4.11.-). Although the enzyme showed -lactamase activity toward ampicillin and penicillin G, carboxypeptidase DD (13) and D-aminopeptidase activities were not detected. The kinetic measurements have clearly shown that the enzyme prefers (D-Phe)₃ and (D-Phe)₄ to the penicillins. ADP activity similar to that described in this study has not been reported, although some -lactamases and carboxypeptidase DD are known to be active toward acyclic depsipeptides (26, 27). Because the enzyme acts on aromatic ring-containing substrates and is excreted into the medium, the enzyme might have evolved to be active toward the synthetic substrates ampicillin and penicillin G.

We found that ADP (350 amino acids after modification) is homologous to the sequence of carboxypeptidase DD from Streptomyces R61 (13), penicillin-binding proteins, class C -lactamases, and D-aminopeptidase from O. anthropi (6). As shown in Fig. 6, the sequence Ser-Xaa-Xaa-Lys is perfectly conserved in this class of enzymes, and the consensus sequence is located ~60 residues from the NH₂ termini of most of the enzymes. Thus, we propose that alkaline D-peptidase from B. cereus could be categorized as a new member of the penicillin-recognizing enzymes, which include penicillin-binding proteins, -lactamases, and D-aminopeptidase.