Presence of D-alanine in an endopeptidase from Streptococcus pyogenes.

D-amino acids are commonly found in peptide antibiotics and the cell wall peptidoglycan of bacterial cell walls but have not been identified in proteins or enzymes. Here we report the presence of 6-7 A-alanine residues in an endopeptidase of Streptococcus pyogenes, a unique enzyme involved in surface protein attachment that we term LPXTGase. Using D-amino acid oxidase coupled with catalase for the deamination of D-alanine to pyruvic acid (a conversion unique to D-alanine), we were able to identify [14C]pyruvic acid in a [14C]alanine-labeled preparation of purified LPXTGase, which represents 27% of the amino acid composition. Because D-amino acids are not accommodated in ribosomal peptide synthesis, these results suggest that the same process used in assembling peptide antibiotics or a yet unidentified mechanism may synthesize the core protein of this endopeptidase.

In our previous report we described the unusual properties of an endopeptidase termed LPXTGase purified from Streptococcus pyogenes (1). This membrane-associated enzyme cleaves a common sequence motif, LPXTG, found near the C termini of precursor proteins programmed to be displayed on the cell surface and linked to the cell wall peptidoglycan. We discovered that the LPXTGase is highly glycosylated and contains hydrophobic amino acids with unusually high masses, which are not among the 20 common amino acids present in ribosomally translated proteins. The presence of unusual amino acids indicated the possibility that the core protein of the enzyme might be constructed entirely or in part through the well characterized nonribosomal peptide synthesis pathway, where amino acids are assembled into a peptide on amino acid-activating multienzyme templates. Examples are the syntheses of peptide antibiotics in Bacilli sp. (2)(3)(4), Streptomyces sp. (5)(6)(7), and Amycolatopsis sp. (8,9), synthesis of a phytoionophore in Pseudomonas syringae (10), and synthesis of the immunosuppressant cyclosporin A in the fungus Tolypocladium niveum (11). Unusual amino acids could also be incorporated into the ribosomally translated core protein through post-translational modifications, however. Another unusual feature of the core protein of LPXTGase is that only seven different amino acids represent 93% of the 61 common residues present in the core protein. It is particularly striking that alanine accounts for nearly 40% of the common amino acids found in the protein.
Over-representation of one or two amino acids within a peptide is a characteristic feature of nonribosomally constructed peptide antibiotics (4,5). As there exists an intriguing possibility that the core protein is indeed constructed at least in part by a nonribosomal peptide synthesis pathway, we considered whether some of the alanine residues might be in the D-form, a hallmark feature of nonribosomal synthesis of peptide antibiotics (11,12). This prompted us to search for the presence of D-alanine in the core protein of LPXTGase. In this report, we present strong evidence that alanine in the D-isoform is present in this bacterial enzyme.
Purification of [ 14 C]Alanine-labeled LPXTGase-About 3 g of the harvested cell pellets were suspended in 120 ml of 30 mM Mes buffer, pH 6.3. About 500 units of phage lysin from group C streptococcus (13) were added to the cell suspension, and the mixture was incubated with shaking at 37°C for 90 min. The resulting cell lysate was centrifuged at 10,000 rpm for 20 min, and the supernatant (cytosol) was saved. The pellets containing cell ghosts were suspended in 60 ml of 20 mM Tris buffer, pH 7.6, containing 0.2% Brij 35, the suspension was sonicated for 40 s, and the sonicated suspension was stirred at 4°C overnight. The suspension was centrifuged for 40 min at 10,000 rpm, and the supernatant (membrane extract) was collected. From the combined supernatants, [ 14 C]alanine-labeled LPXTGase was purified according to the procedure described previously (14). Briefly, the combined supernatants were applied to a DEAE-cellulose column equilibrated with 20 mM Tris buffer, pH 6.8, and the column was eluted with 20 mM Tris buffer, pH 7.6, containing 0.1% Brij 35. Fall-through eluant was concentrated by ultrafiltration using a YM-3 membrane, and the concentrated fallthrough fraction was applied to a Sephadex G-50 column (49 ϫ 2.5 cm) equilibrated with 20 mM Tris buffer, pH 7.6, containing 0.1% Brij 35. The column was then eluted with the same buffer. The gel filtration step was repeated. 14 C-labeled LPXTGase eluting from the second G-50 column was concentrated to 4 ml by ultrafiltration using a YM-3 membrane. The enzyme solution was lyophilized, and the dried enzyme was dissolved in 0.5 ml of water. Five volumes each of ethanol and ethyl acetate were added to the enzyme solution, and the mixture was kept overnight at Ϫ20°C. Flocculated enzyme aggregates were collected by centrifugation. The solvents were removed from the enzyme pellets by means of Speedvac. The dried enzyme was dissolved in 300 l of 80% phenol, after which the enzyme solution was mixed with 3 ml of 6 N HCl in capped glass vials. The enzyme was hydrolyzed by placing the vials in a 110°C oven for 22 h. The acid hydrolysate of the enzyme was lyophilized to remove HCl and phenol. The dried enzyme hydrolysate was dissolved in 1 ml of 50% ethanol, and the hydrolysate was banded on a 20 ϫ 20-cm silica gel plate. The plate was developed with a solvent mixture consisting of chloroform/methanol/water (1:2:1). An autoradiogram was made from the plate, and the 14 C-labeled alanine band was located. The alanine band was scraped off of the plate, silica gel was placed in a small column, and the column was eluted with 50% ethanol. The eluted material was dried by means of Speedvac, and the 14 Clabeled alanine was dissolved in 300 l of distilled water.

Acid Hydrolysis of [ 14 C]Alanine-labeled LPXTGase and Purification of [ 14 C]Alanine-The
Detection of D-Alanine-In an initial experiment, 60 l of an aliquot of 14 C-labeled alanine sample (18,200 cpm) was introduced into a microcentrifuge tube, then 4 l of 1 M Tris buffer, pH 8.5, 0.2 unit of D-amino acid oxidase (EC 1.4.3.3, Sigma catalog no. A 5222) in 2 l, and 0.2 unit of catalase (EC 1.11.1.6, Sigma catalog no. C 30) in 2 l were added, and the reaction mixture was incubated for 2 h at room temperature. Authentic L-alanine with similar radioactivity was subjected to D-amino acid oxidase and catalase in an identical manner. At the end of the reaction, 300 l of ethanol was added to the microcentrifuge tubes, and these were kept at Ϫ20°C. The resulting precipitates consisting of enzymes and excess buffer salt were pelleted by centrifugation, and the supernatants containing radioactivity were concentrated to about 20 l. The reaction products, along with untreated L-[ 14 C]alanine and [ 14 C]pyruvic acid (Amersham, CFA 85), were loaded on a silica gel plate, and the plate was developed with chloroform/methanol/water (1:2:1). Radioactive spots were detected by exposure of the plate to an x-ray film. To the remainder of the [ 14 C]alanine sample (90,800 cpm) in a microcentrifuge tube, 1 M Tris buffer, pH 8.5, was added to a final concentration of 50 mM after which 2 units each of D-amino acid oxidase and catalase were added to the tube. The reaction mixture was incubated for 6 h at room temperature, and radioactive reaction products were prepared and separated by silica gel TLC as above. Reaction products a and b shown in Fig. 3B were eluted from the plate with 70% ethanol. Products a, b, and authentic [ 14 C]pyruvic acid were loaded on a TLC plate, and the plate was developed with ethyl acetate/propionic acid/water (2:1:1).
Phenylhydrazone Derivatives of D-Amino Acid Oxidase-Catalasetreated D-and L-Alanine and Pyruvic Acid-10 mol each of D-and L-alanine in 100 l of distilled water were placed in microcentrifuge tubes after which 1 unit each of D-amino acid oxidase and catalase were added, and 1 M Tris buffer, pH 8.5, was added to a final concentration of 50 mM. The reaction was carried out at room temperature for 3 h. Enzymes and excess buffer were then precipitated from the reaction mixtures by adding 4 volumes of ethanol. After removing the precipitates by centrifugation, the supernatant containing the reaction products was placed in microcentrifuge tubes and was concentrated to 50 l. 20 mol of phenylhydrazine in 100 l of 90% ethanol was added to the reaction products from D-and L-alanine and to 10 mol of pyruvic acid in 50 l of 50% ethanol, pH adjusted to 7 with 1 N NaOH, and the mixtures were placed in a 110°C oven for 15 min. Aliquots (10 l) of the resulting phenylhydrazone derivatives were loaded on a silica gel plate, and the chromatogram was developed with a solvent mixture consisting of ethyl acetate/isopropyl alcohol/water/concentrated ammonium hydroxide (60:30:10:1). Phenylhydrazone derivatives were detected by viewing the plate under UV light. ]Alanine-labeled cytosol and membrane extract were combined and applied to a DEAE-cellulose column, and fall-through fraction was obtained as described under "Experimental Procedures." The fall-through solution was concentrated by means of ultrafiltration, and the concentrate was chromatographed on a Sephadex G-50 column (49 ϫ 2.5 cm) as described under "Experimental Procedures." LPXTGase activity was determined by measuring cleavage of bead-bound 125 I-labeled LPXTG-peptide as described previously (13). The peak enzyme fractions were concentrated and chromatographed again on the same G-50 column. From each fraction, a 50 l aliquot was used for enzyme assay, and 200 l was used for 14 C counts. extract of these cells, which also contained free alanine, was 1.26 ϫ 10 7 cpm. Fig. 1 shows the elution pattern of enzyme activity from the second Sephadex G-50 column. The enzyme activity profile, as detected by cleavage of a bead-bound 125 I-labeled LPXTG-containing peptide (13), coincided precisely with the 14 C activity profile. The total radioactive counts of the purified LPXTGase were 1.42 ϫ 10 5 cpm. Consistent with our previous findings that the LPXTGase does not contain any aromatic amino acid and thus does not absorb UV light at 280 nm, the enzyme peak showed no UV absorption at 280 nm, verifying the purity of the enzyme.

Purification of [ 14 C]Alanine-labeled LPXTGase-
Purification of [ 14 C]Alanine from Acid Hydrolysate of LPXT-Gase-Purified LPXTGase was hydrolyzed with 6 N HCl as described under "Experimental Procedures," and the acid hydrolysate was subjected to silica gel thin layer chromatography using chloroform/methanol/water (1:2:1) as the running solvent. Fig. 2 shows that the 14 C radioactivity was nearly exclusively present in alanine (1.14 ϫ 10 5 cpm). A very faint radioactive band was seen at rf 0.75, which corresponded with glutamic acid. No other radioactive bands corresponding to the rest of the LPXTGase-constituent amino acids were detected. Mobilities of non-radioactive, LPXTGase-constituent amino acids on silica gel thin layer chromatography with the same solvent mixture were previously determined (results not shown).
Presence of D-Alanine in the LPXTGase-Treatment of Dalanine with D-amino acid oxidase coupled with catalase re-sults in the deamination of D-alanine to produce pyruvic acid, a reaction that does not occur when L-alanine is similarly treated. In Fig. 3A, D-amino acid oxidase-catalase-treated [ 14 C]alanine from the enzyme was loaded onto spot 1, similarly treated authentic L-[ 14 C]alanine onto spot 2, untreated L-[ 14 C]alanine onto spot 3, and authentic [ 14 C]pyruvic acid was loaded onto spot 4 at the origins of a silica gel thin layer plate, and the plate was developed with chloroform/methanol/water (1:2:1). As shown, D-amino acid oxidase-catalase treatment of the [ 14 C]alanine sample produced two radioactive products, a and b, of which the mobility of product a is identical with that of pyruvic acid. Product b is consistent with the mobility of the dimeric form of pyruvic acid. The yield of these products was low as reflected in the quantity loaded on the plate.
To verify the presence of the dimeric form of pyruvic acid, the remaining (80% of the total) [ 14 C]alanine from the enzyme was incubated with an excess amount of D-amino acid oxidase and catalase for a prolonged time to ensure conversion of all Dalanine to pyruvic acid, and the reaction products were chromatographed on a silica gel plate in the same manner (Fig. 3B). The same products a and b were again detected but in higher yield. These two products were eluted from the plate and along with authentic pyruvic acid were applied to a silica gel plate, which was developed with ethyl acetate/propionic acid/water (2:1:1) (Fig. 3C). The main radioactive spots from both products a and b showed the same mobility as authentic pyruvic acid, but minor secondary radioactive spots were also observed. The fact that products a and b from Fig. 3B gave rise to nearly identical radioactive spots on subsequent chromatography us- ing a different solvent (Fig. 3C) indicates that these two products are inter-convertible. The total radioactive counts in products a and b in Fig. 3B were 15,200 and 9,400 cpm, respectively, which combined represent 27% of the total radioactivity in the [ 14 C]alanine sample. As there are 24 alanine residues in the LPXTGase, 6 -7 alanine residues are in the D-isoform. These numbers are based on the assumption that all D-alanines present in the alanine sample were converted to pyruvic acid after D-amino acid oxidase and catalase treatment.
Phenylhydrazone Derivatives of D-Amino Acid Oxidase-Catalase Treatment of D-and L-Alanine and Pyruvic Acid-We wished to verify that pyruvic acid is present in both monomeric and dimeric forms resulting in derivatives a and b. Track 1 of Fig. 4 shows phenylhydrazone derivatives of the reaction products arising from D-alanine after treatment with D-amino acid oxidase and catalase. Derivatives a and b are clearly present in this track. No phenylhydrazone derivative is seen when Lalanine treated with D-amino acid oxidase-catalase was reacted with phenylhydrazine (track 2). Track 3 shows phenylhydrazone derivatives arising from authentic pyruvic acid, which are identical to the migration pattern of derivatives a and b in track 1. Taken together, these results indicate that pyruvic acid exists both in monomer and dimer forms. DISCUSSION D-Amino acid oxidase (EC 1.4.3.3) deaminates D-amino acid to the corresponding ␣-keto acids, but the enzyme is not reactive with L-amino acids (15). We show here that at least 27% of the alanine present in LPXTGase became deaminated by Damino acid oxidase, verifying that LPXTGase does in fact contain D-amino acids. Based on radioactive count data, the estimated number of D-alanine residues in LPXTGase is 6 -7. The possibility exists, however, that the actual number is higher because pyruvic acid decomposes at 165°C, and the hot air stream used during loading of the D-amino acid oxidase reaction products on the TLC plates may have decomposed some pyruvic acid.
All of the amino acids in proteins exhibit the same steric configuration; thus, they are all L-amino acids. D-Amino acids have not been identified in proteins, although they do exist in nature in peptides from multicellular organisms (16). In bacteria for instance, D-amino acids are commonly found in association with the amino acids used to construct the peptidoglycan. For example, in the disaccharide-pentapeptide precursor for peptidoglycan synthesis one D-glutamate and two D-alanines are present prior to cross-linking after which, the terminal D-alanine is removed to facilitate the linkage. However, beside the bacterial peptidoglycan, D-amino acids have generally been associated with non-ribosomal peptide synthesis, in the production of peptide antibiotics (2)(3)(4)(5)(6)(7)(8)(9). Our finding of Dalanine in the LPXTGase enzyme still does not resolve whether the D-amino acid was incorporated through a non-ribosomal peptide synthesis pathway or was post-transcriptionally added to a ribosomally translated protein backbone. To date, the largest non-ribosomally synthesized peptide is Ͻ50 amino acids, and none have been found to exhibit catalytic activity. Because D-amino acids cannot be accommodated during ribosomal synthesis of a polypeptide chain, this strongly suggests that the 60-amino acid LPXTGase may be the first reported instance of a D-amino acid-containing enzyme that is not ribosomally synthesized.
One could argue, however, that the LPXTGase used in this paper might be contaminated by L-alanine oxidase or racemase. This is unlikely because the LPXTGase used in the amino acid analysis was highly purified as described previously (1). However, even if the LPXTGase were contaminated by a small amount of impurities, it is unlikely that L-alanine oxidase or racemase is a part of the impurity. The L-alanine racemase from Escherichia coli (and other species) contains 359 amino acid residues and 40 acidic amino acid residues. Thus, L-alanine oxidase or racemase cannot appear in the fall-through fractions in DEAE-cellulose chromatography or the M r ϭ 14,000 region of the G 50 chromatography used in purifying the LPXTGase. Furthermore, if L-alanine oxidase or racemase copurified with LPXTGase for inexplicable reasons, only free amino acids, and not amino acids contained in proteins, are substrates for these enzymes.
Although the deaminated product of D-alanine is pyruvic acid, we did observe an additional product. The keto-enol tautomerism of ␣-keto acids is well known (17). For example, pyruvic acid as a keto acid is expected to undergo keto-enol tautomerism, of which the enol form is an ␣-hydroxy acid that, as a rule, undergoes dimerization and polymerization. [ 14 C]pyruvic acid is generally shipped as a lyophilized powder in a sealed vial. When freshly dissolved in 50% ethanol and chromatographed, a single radioactive spot is usually observed (Fig. 3, A and B). However, after standing it may become FIG. 4. Phenylhydrazone derivatives of the products from Damino acid oxidase-catalase-treated D-alanine, L-alanine, and phenylhydrazone derivatives of pyruvic acid. Phenylhydrazone derivatives were prepared as described under "Experimental Procedures." Spot 1, phenylhydrazone derivatives of the products from Damino acid oxidase-catalase treatment of D-alanine. Spot 2, phenylhydrazone derivatives of the products from D-amino acid oxidase-catalase treatment of L-alanine. Spot 3, phenylhydrazone derivatives of pyruvic acid. Spot 4, phenylhydrazine. The running solvent was ethyl acetate/ isopropyl alcohol/water/concentrated ammonium hydroxide (60:30:10: 1). Letters a and b, left side, represent the phenylhydrazone derivatives of the monomeric (a) and dimeric (b) forms of pyruvic acid. oxidized, and additional spots will appear after chromatography (Fig. 3C). Thus, we interpret that the secondary product that we observed in our chromatograms arose from the dimerization of pyruvic acid. Whatever caused the appearance of the second spot from the D-amino acid oxidase-treated [ 14 C]alanine from LPXTGase, an identical second spot also appeared from authentic [ 14 C]pyruvic acid. Therefore, it seems safe to conclude that the second spot originated from pyruvic acid.
Thus, in summary, we have identified the presence of at least six D-alanine residues in the LPXTGase enzyme, accounting for ϳ10% of the total amino acids in this molecule. However, our results are inconclusive as to whether the D-alanine found in LPXTGase is the result of a post-translational modification of L-alanine to D-alanine or is incorporated non-ribosomally.