Penicillinase-releasing Protease of BaciZZus Zicheniformis 749 SPECIFICITY FOR HYDROXYAMINO ACIDS*

The membrane penicillinase of Bacillus licheniformis 749/C differs from the exopenicillinase in that it has an additional 24 amino acid residues and a phosphatidylserine at the NH2 terminus (Yamamoto, S., and Lampen, J.O. (1976) J. Biol. Chem. 251, 4095-4101). The conversion of the membrane penicillinase to the exo form is probably carried out by a specific penicillinase-releasing protease (PR-protease) whose properties are generally consistent with the properties of penicillinase secretion. The substrate specificity of the PR-protease was determined by identifying the NH2 and COOH termini of the peptides produced by hydrolysis of ribonuclease B and beef insulin. The enzyme hydrolyzed only peptide bonds involving the carboxyl groups of serine or thrombine. Similar bonds in synthetic di- or tripeptides of L-serine were not cleaved. The existence of seryl-lysine and threonyl-glucamic acid bonds in the protease-susceptible (phospholipopeptide) region of the membrane penicillinase and the presence of only lysine or glutamic acid at the NH2 terminus of the exoenzyme released in vivo are consistent with the specificity of PR-protease; hence, we propose that this enzyme has an essential role in the formation of exopenicillinase. The PR-protease is a potential tool for protein sequence determination because of its narrow and novel substrate specificity.

The membrane penicillinase ofBacillus licheniformis 7491 C differs from the exopenicillinase in that it has an additional 24 amino acid residues and a phosphatidylserine at the NH* terminus (Yamamoto, S., and Lampen, J. 0. (1976) J. Biol. Chem. 251, 4095-4101). The conversion of the membrane penicillinase to the exo form is probably carried out by a specific penicillinase-releasing protease (PR-protease) whose properties are generally consistent with the properties of.penicillinase secretion. The substrate specificity of the PR-protease was determined by identifying the NH2 and COOH termini of the peptides produced by hydrolysis of ribonuclease B and beef insulin. The enzyme hydrolyzed only peptide bonds involving the carboxyl groups of serine or threonine. Similar bonds in synthetic di-or tripeptides of L-serine were not cleaved.
The existence of seryl-lysine and threonyl-glutamic acid bonds in the protease-susceptible (phospholipopeptide) region of the membrane penicillinase and the presence of only lysine or glutamic acid at the NH, terminus of the exoenzyme released in uiuo are consistent with the specificity of PR-protease; hence, we propose that this enzyme has an essential role in the formation of exopenicillinase.
The PR-protease is a potential tool for protein sequence determination because of its narrow and novel substrate specificity. Bacillus licheniformis 749/C produces both a membranebound and an exopenicillinase (1,2). The hydrophobic membrane enzyme differs from the hydrophilic exopenicillinase in that it contains an additional 24 amino acid residues (only aspartic acid, asparagine, glutamic acid, glutamine, glycine, and serine) and a phosphatidylserine at the NH, terminus (3, 4). The sequence of the phospholipopeptide segment has been determined (5). The membrane-bound enzyme is an intermediate in the formation of exopenicillinase (6) and is readily converted to it in uiuo at neutral or alkaline pH (7,8). The released exopenicillinase always has lysine or glutamic acid at its NH, terminus (8,9), suggesting that a common proteolytic * This work was supported by United States Public Health Service Grant AI-04572 from the National Institute of Allergy and Infectious Diseases. processing of the membrane penicillinase takes place under various conditions of secretion (7).
We have detected a protease in Bacillus licheniformis 749 and 749/C which is capable of converting the membrane penicillinase to the exoenzyme and have purified it to apparent homogeneity (7,8). Several other neutral and alkaline proteases produced by these organisms have very little activity iti this regard (7). The penicillinase-releasing enzyme (PR-protease) is a serine protease with an approximate molecular weight of 21,000 and a pH optimum of 7.0 to 9.0. It requires CaZ+ or MgL+ for stability and its general properties and sensitivity to inhibitors are consistent with the characteristics of penicillinase secretion (7,8). In this communication we describe' the substrate specificity of the PR-protease" as determined by end group analysis of the peptides produced from ribonuclease B and beef insulin by digestion with the protease. DISCUSSION Conversion of the membrane penicillinase to the exoenzyme in a culture of Bacillus licheniformis 749 or 749/C is probably accomplished by a serine protease which we have recently described (7). The other neutral and alkaline proteases formed by these organisms have, at most, weak activity in converting the membrane enzyme to the exo form (7). Since the exoenzyme released in uiuo under a variety of conditions always has lysine or glutamic acid at its NH, terminus (8,9), it was presumed that the PR-protease hydrolyzes only a specific type of peptide bond. In addition, the PR-protease shows only limited proteolytic activity on methylated casein (7). These two observations led us to examine the substrate specificity of the PR-protease in order to clarify its role in the overall process of penicillinase secretion.
' All of the experimental procedures, results, and data in the form of figures are presented in a miniprint format immediately following this paper. For the convenience of those who prefer to obtain the supplementary material in the form of 10 pages of full size photocopies, it is available as JBC Document Number 76M-1387. Orders should specify the title, authors, and reference to this paper, the JBC Penicillinase-releasing Protease of Bacillus licheniformis The type of peptide bond cleaved by the PR-protease was determined by end group analysis of the peptides produced from two small proteins of known sequence, i.e. insulin and pancreatic ribonuclease B. NH,-terminal analysis by the dansylation procedure of the peptides from PR-protease digests of insulin and ribonuclease B suggested that the PR-protease is not specific for the amino acid donating the amino group to the peptide bond hydrolyzed. Proline, leucine, valine, and histidine were identified as newly formed NH*-terminal residues from insulin. The peptides from ribonuclease had leucine or proline, valine, serine, threonine, and aspartic acid as new NH,-terminal residues. Thus there is no obvious pattern of specificity in the NH,-terminal amino acids of the peptides produced by the PR-protease.
The COOH-terminal residues of the peptides from insulin were identified as aspartic acid, serine, threonine, and alanine (Fig. 2) by the selective tritium exchange procedure (12). Asparagine (detected as aspartic acid after acid hydrolysis) and alanine are the COOH-terminal residues of the A and B chains of beef insulin (16). The peptides from ribonuclease B had serine, threonine, and valine as the COOH-terminal amino acids (Fig. 3); valine is the COOH-terminal residue of this protein. These results were substantiated by treatment of the peptides from ribonuclease B with carboxypeptidase A (Fig. 4), which released serine or threonine, or both, and valine. Thus the new COOH-terminal residues produced from both proteins were serine and threonine.
The amount of COOH-terminal asparagine and alanine did not increase during PR-protease cleavage of insulin, and these amino acids were not detected at the COOH terminus of the peptides from ribonuclease B. Similarly, the amount of COOH-terminal valine did not increase during cleavage of ribonuclease B, and COOH-terminal valine was not produced from insulin. These results make it highly probable that PR-protease hydrolyzed only those peptide bonds that involved the carboxyl group of either serine or threonine. Thus the specificity of PR-protease, like that of other specific serine proteases (17, 181, is primarily a function of the amino acids donating the carboxyl group to the peptide bond cleaved. In the amino acid sequences of beef insulin and ribonuclease B, all the new NH,-terminal amino acids produced by PRprotease are present in seryl or threonyl peptide bonds (16). Beef insulin contains 1 threonyl and 3 seryl residues and these are linked to the amino groups of proline, valine, leucine, and histidine, the 4 new NH,-terminal residues detected, so that all of the Ser-X or Thr-X bonds of insulin must have been hydrolyzed. This is probably not true for ribonuclease B, since alanine, glutamine, glycine, methionine, and phenylalanine are present in seryl or threonyl peptide linkage but were not detected as new NH, termini. Another restriction on the action of PR-protease is evident from the observation that a series of dipeptides in which the carboxyl group was donated by serine were not hydrolyzed by the PR-protease; these dipeptides included Ser-Leu, a linkage that was cleaved in insulin. The enzyme also failed to hydrolyze Ser-Ser-Ser and polyserine. The basis for the inability of the PR-protease to hydrolyze smaller peptides has not been investigated.
Paper chromatography of the peptides from ribonuclease B revealed the presence of a small amount of valine that was not detectable in the original denatured ribonuclease B or in incubations lacking the ribonuclease. This finding indicates the slow hydrolysis of the COOH-terminal Ser-Val-COOH bond of the ribonuclease, which is the only Ser-Val bond in the entire molecule (161, and is consistent with the proposed substrate specificity of PR-protease. Fig. 5 shows a schematic representation of the membrane penicillinase of Bacillus licheniformis 749/C including the sequence of the protease-susceptible phospholipopeptide region (5) which is lacking in the released hydrophilic exopenicillinase.
Hydrolysis of the membrane penicillinase by PR-protease yields the exoenzyme without loss of activity (7). Hence the exoenzyme portion (which carries the catalytically active site) must initially have been in a protease-resistant conformation, and the cleavage site should be in the phospholipopeptide segment of the membrane penicillinase.
The specificity of PRprotease for peptide bonds involving the carboxyl group of serine or threonine is in agreement with the presence of lysine or glutamic acid at the NH, terminus of the exoenzyme (8,9). Ser-Lys-Thr-Glu (residues 25 to 28) is the tetrapeptide segment from which the NH,-terminal lysine or glutamic acid of exopenicillinase is derived, and cleavage of Ser113-Lys2,i or Thr,,-Glu,, must have occurred. Other bonds potentially sensitive to PR-protease (Ser-Asn, Ser-Gly, and Ser-Glu) are present in the phospholipopeptide region of the membrane enzyme and these might also be split during the physiological release of exopenicillinase.
However, the segment from residues 19 through 28 (Fig. 5) which contains the two linkages known to be cleaved, is polar and highly charged and probably projects out of the membrane into the aqueous phase, while the segment from residues 1 to 18 is less polar and may even span the lipid bilayer (5). The Serf,-Lys,,; and Thr,,-Glu,, bonds may well be the only susceptible ones that are readily accessible for cleavage by the PR-protease and production of the exoenzyme.
The amino acid sequence of the membrane penicillinase (5) and the substrate specificity of the PR-protease together provide a ready explanation of exopenicillinase release and thus of the role of this protease in penicillinase secretion, at least from a chemical point of view. A complete picture of its physiological role can be obtained only when the properties of mutants which lack the PR-protease have been studied.
PR-protease has a novel and narrow substrate specificity. It is a potentially valuable tool for the determination of amino acid sequences.