INTRODUCTION

Gram-positive pathogens cause a variety of different human diseases, which often present therapeutic difficulty because of bacterial antibiotic resistance (1). The treatment of hospital-acquired infections of Staphylococcus aureus and Enterococcus faecium is particularly challenging since many of these strains are multiply resistant to all classes of antibiotics (2). In search for a novel target of antibiotic therapy, we have begun to characterize the anchoring of surface proteins to the cell wall of S. aureus. Staphylococcal surface proteins are thought to be instrumental during the establishment and maintenance of human infections by either promoting the attachment to specific host tissues or preventing the phagocytic clearance of the invading bacteria (3, 4).

Cell wall anchoring of surface proteins in S. aureus requires both an NH₂-terminal signal (leader) peptide and a COOH-terminal cell wall sorting signal (5). The 35-residue sorting signal harbors an LPXTG sequence motif that has been found conserved within the sorting signals of more than 100 surface proteins of Gram-positive bacteria (6, 7) and serves as the recognition sequence for the proteolytic cleavage between its threonine (T) and glycine (G) residues (8). Previous work reported the purification of a surface protein that had been released from the staphylococcal peptidoglycan by lysostaphin digestion (9). The COOH-terminal peptide of this molecule was cut at engineered trypsin cleavage sites and characterized with ESI-MS¹ and Edman degradation, revealing the addition of two and three glycine residues to the threonine of the LPXTG motif (9). Because lysostaphin, a glycyl-glycine endopeptidase (10), cleaves the pentaglycine cross-bridges of the staphylococcal peptidoglycan predominantly at their central glycine residue (11), it was suggested that surface proteins could be anchored via an amide linkage between the carboxyl of threonine and the amino of the pentaglycine cross-bridge (9). Nevertheless, previous work left unresolved the chemical structure of the cell wall anchor of surface proteins. This question was addressed here, and we demonstrate that surface proteins are linked to a branched peptide (NH₂-Ala--Gln-Lys-(NH₂-Gly₅)-Ala-COOH). This branched anchor peptide is amide-linked to N-acetylmuramic acid, thereby tethering the peptide backbone to the glycan chains of the cell wall. Unlike the highly cross-linked wall peptides within the peptidoglycan, these branched anchor peptides are not substituted at their terminal D-alanine.

EXPERIMENTAL PROCEDURES

To express the hybrid protein Seb-MH₆-Cws, the coding sequence for the protein A cell wall sorting signal harboring the engineered methionine-histidine affinity tag was amplified from the chromosomal DNA of S. aureus strain 8325-4 (12) with the primers Seb-6His (AAGGTACCATGCATCACCATCACCATCACGCTCAAGCATTACCAGAAACT) and spa-2 (AAGGATCCTTATCATTTCAAATAAGAATGTGTT). The PCR product was digested with KpnI and BamHI and cloned into the corresponding sites of pSEB-FnBP (7) to generate pHTT4, which was transformed into S. aureus OS2 (5). The coding sequence of the 11 amidase was amplified from purified phage DNA (12) with the primers (AACATATGCAAGCAAAATTAACTAAAAAT and AAGGATCCCTAGTGATGGTGATGGTGATGACTGATTTCTCCCCATAAGTC), digested with NdeI and BamHI and cloned into the corresponding sites of the T7 expression vector pET-9a (13) to yield pHTT2. After transformation of Escherichia coli BL1 (DE3) pLysS (13), the resulting strain was employed for the induced expression of the 11 amidase.

S. aureus OS2 harboring pHTT4 was grown overnight in tryptic soy broth supplemented with chloramphenicol (10 µg/ml) and diluted 1:40 into the same medium. Generally 4 liters of culture were grown with shaking at 250 rpm and 37 °C for 5 h. Cells were harvested by centrifugation at 8000 × g for 15 min. Pellets were suspended in 100 ml of water, and the cell wall carbohydrates were extracted by the addition of 100 ml of an ethanol-acetone (1:1) mixture (14) and incubation for 30 min on ice. The cells were collected by centrifugation and washed with 300 ml of ice-cold water. Cell pellets were suspended in 30 ml of 0.1 M Tris-HCl buffer (pH 7.5 for lysostaphin and amidase digests, pH 6.8 for mutanolysin), and the peptidoglycan was digested with either lysostaphin (33 µg/ml, Ambi), amidase (67 µg/ml), or mutanolysin (333 units/ml, Sigma) and incubated for either 2 (lysostaphin) or 16 h (amidase and mutanolysin) at 37 °C. Peptidoglycan-released surface proteins were separated from the acetone-protoplasts (14) by centrifugation at 17,000 × g for 15 min, and the collected supernatant was filtered through a 0.2-µm pore size membrane. Solubilized Seb-MH₆-Cws surface protein was purified by affinity chromatography on Ni-NTA-Sepharose (Qiagen). Briefly, a column with 2 ml of bed volume was washed with 15 ml of equilibration buffer (50 mM Tris, 150 mM NaCl, pH 7.5), loaded with 30 ml of cell wall extract, washed first with 30 ml of wash buffer (50 mM Tris, 150 mM NaCl, 10% glycerol, pH 7.5) and then with 30 ml of equilibration buffer. Bound protein was eluted with 5 ml of elution buffer (0.5 M imidazole, 50 mM Tris, 150 mM NaCl, 10% glycerol, pH 7.5).

Amidase and lysostaphin-solubilized Seb-MH₆-Cws were precipitated with 7% trifluoroacetic acid (v/v), solutions were incubated on ice for 30 min, centrifuged for 15 min at 17,000 × g, and the collected pellets were washed in acetone and dried. After affinity chromatography, the mutanolysin-solubilized Seb-MH₆-Cws protein was dialyzed against 4 liters each of 50 mM of NH₄HCO₃ for 1, 3, and 24 h and then dried under vacuum. The dried pellets of the affinity-purified Seb-MH₆-Cws proteins were suspended into 600 µl of 70% formic acid. A crystal of CnBr was added, and the reaction was incubated for 16 h at room temperature. The cleaved peptides were dried under vacuum, washed twice with 100 µl of water, and suspended in 1 ml of buffer A (6 M guanidine hydrochloride, 0.1 M NaH₂PO₄, 0.01 M Tris, pH 8.0). A 1-ml column of Ni-NTA-Sepharose was pre-equilibrated with 10 ml of buffer A and loaded with the CnBr-cleaved peptides. The column was washed successively with 10 ml of buffer A, 10 ml of buffer B (8 M urea, 0.1 M NaH₂PO₄, 0.01 M Tris, pH 8.0), and 10 ml of buffer C (same as buffer B, but pH 6.3). The anchor peptides were finally eluted with 2 ml of 0.5 M acetic acid. The eluate was divided into equal volumes, one of which was desalted on C-18 matrix prior to MALDI-MS, whereas the other was subjected to RP-HPLC purification. Preparation of anchor peptides from 4 liter of staphylococcal culture (5 × 10¹³ colony forming units) typically yielded 1 nM (2 µg) of purified compound. Because of the relative small yield of anchor peptides, we chose to analyze our compounds by the more sensitive MALDI-MS and ESI-MS rather than fast atom bombardment mass spectrometry. The latter method has been used extensively in the past for the study of bacterial peptidoglycans (11, 15-17).

A C-18 matrix cartridge (Analtech) was prewashed with 10 ml of CH₃CN, followed by 10 ml of 0.1% trifluoroacetic acid in water. One ml of the affinity chromatography eluate was loaded on the cartridge, washed with 10 ml of 0.1% trifluoroacetic acid, and eluted with 3 ml of 60% CH₃CN. The desalted anchor peptides were dried under vacuum and suspended in 20 µl of CH₃CN:water:formic acid (50:50:0.1). MALDI-MS spectra were obtained on a reflectron time-of-flight instrument (PerSeptive Biosystems Voyager RP) in the linear mode. Samples (0.5-1.0 µl) were co-spotted with 1.0 µl of matrix (-cyano-4-hydroxycinnamic acid, 2 mg/200 µl CH₃CN:water:trifluoroacetic acid (70:30:0.1)) and mass measured using an external calibration.

For further purification of anchor peptides, the affinity chromatography eluate was subjected to RP-HPLC on a C-18 column (2 × 250-mm, C18 Hypersil, Keystone Scientific). The separation was carried out at 40 °C with a linear gradient from 99% H₂O (0.1% trifluoroacetic acid) to 99% CH₃CN (0.1% trifluoroacetic acid) in 90 min at a flow rate of 0.2 ml/min. The elution of anchor peptides was monitored at 215 nm, and 1-min fractions were collected. Vacuum dried fractions were generally re-dissolved in 50 µl of water:CH₃CN:formic acid (50:50:0.1). A Perkin-Elmer Sciex API III triple quadrupole mass spectrometer was tuned and calibrated by flow injection (10 µl/min) of a mixture of PPG 425, 1000, and 2000 (3.3 × 10⁵, 1 × 10⁴, and 2 × 10⁴ M, respectively) in water:methanol (1:1) containing 2 mM ammonium formate and 0.1% CH₃CN. Calibration across the m/z range 10-2400 was effected by multiple ion monitoring of eight PPG solution signals (typically the singly charged ions at m/z 58.99, 326.25, 906.67, 1254.92, 1545.13, 1863.34, and 2010.47 and the doubly charged ion at m/z 520.4). The ion spray voltage operated at 4.5 kV using hydrocarbon-depleted air for the spray nebulization ("zero" grade air, 40 p.s.i., 0.6 liter/min), and spectra were generated with a curtain gas produced from the vapors of liquid nitrogen. Samples (10 µl) were introduced into the ionization source by flow injection. ESI-MS spectra were obtained at instrument conditions sufficient to resolve the isotopes of the PPG/NH₄⁺ singly charged ion at m/z 906 with 40% valley, an orifice voltage of 60, and step size during data collection of 0.3 Da. Deconvolution of the series of multiply charged ions and calculation of peptide or protein molecular weight was achieved with the Hypermass^TM computer program. Daughter ion spectra were obtained using degraded mass resolution to improve the sensitivity of detection, and a step size of 1 Da was used for data collection. Under these conditions, the isotopes of the PPG/NH₄⁺ singly charged ion at m/z 906 were not resolved from one another.

E. coli BL1(DE3) pLysS, pHTT2 was grown overnight in LB medium containing chloramphenicol (30 µg/ml) and kanamycin (50 µg/ml). The culture was diluted 1:50 into 500 ml of the same medium and incubated with 250 rpm shaking at 37 °C. When the A₆₀₀ reached 0.2, T7 polymerase was induced by the addition of 1 mM isopropyl-1-thio--D-galactopyranoside and the culture was incubated for another 4 h. Cells were harvested by centrifugation at 8000 × g for 15 min and suspended in 20 ml of F buffer (50 mM Tris-HCl, 10 mM EDTA, 20% sucrose, 1 mM dithiothreitol, pH 8.0). Lysozyme (Sigma) was added to a final concentration of 0.1 mg/ml, and the suspension was sonicated on ice with 30 10-s pulses over 30 min. The lysate was centrifuged for 10 min at 14,000 × g to collect inclusion bodies harboring staphylococcal amidase. The pellets were extracted with 30 ml of a solution containing CH₃CN:isopropanol:water (3:3:14) and centrifuged for 10 min at 14,000 × g, and the supernatant was discarded. The pellets were suspended in 20 ml of buffer B, and insoluble material was removed by centrifugation for 10 min at 14,000 × g. Proteins soluble in the supernatant were filtered through a 0.2-µm pore size membrane prior to affinity chromatography. A 2.5-ml Ni-NTA-Sepharose column was pre-equilibrated with 15 ml of buffer B, loaded with the filtrate, washed first with 30 ml of buffer B, and then with 30 ml of buffer C. Amidase was eluted with 5 ml of buffer E (same as buffer B, but pH 4.5). The purity and molecular mass of staphylococcal amidase were determined by 12% SDS-PAGE analysis, which was in agreement with the predicted mass. As judged on Coomassie or silver-stained SDS-PAGE, the amidase preparation was 100% pure. The eluate was dialyzed against 1 liter of 50 mM Tris-HCl, 1 M urea, 0.005% Tween 80, pH 7.5, for 24 h at 4 °C without stirring, followed by a second dialysis against the same buffer without urea for 16 h at 4 °C and slow stirring.

RESULTS

An experimental scheme was developed that allowed for the selective purification of COOH-terminal peptides harboring the cell wall anchor structure of surface proteins (Fig. 1). Seb-MH₆-Cws is a recombinant protein consisting of the NH₂-terminal signal (leader) peptide and mature region of staphylococcal enterotoxin B (18) with a fused COOH-terminal cell wall sorting signal of protein A (5). At the fusion joint between these two sequences, a methionine (M) residue followed by six histidines (H₆) was inserted. This engineered insertion serves as an affinity tag for the rapid purification of solubilized surface protein by chromatography on nickel-Sepharose. Purified proteins were cleaved with CnBr at their methionine residues, thereby separating the COOH-terminal anchor peptides from the remainder of the polypeptide chains. The anchor peptides were further purified by a second round of affinity chromatography on nickel-Sepharose and analyzed by MALDI-MS.

Digestion of the staphylococcal cell wall with lysostaphin resulted in the solubilization of Seb-MH₆-Cws, and most of these polypeptides migrated uniformly on SDS-PAGE with a mass of approximately 30,000 Da (Fig. 1). Some of the lysostaphin-released material appeared as distinct species with higher molecular mass, which is likely caused by an incomplete digestion of the peptidoglycan. After CnBr cleavage of Seb-MH₆-Cws, three major anchor peptides were identified by MALDI-MS with m/z 1726, 3836, and 3867 in addition to ions with lesser intensity (m/z 1668, 1759, and 3777; Fig. 2A). To purify individual anchor peptides, we subjected the affinity-purified sample to RP-HPLC on C-18 resin (Fig. 2B). Two major absorption peaks eluted at 24-29% and at 34-37% CH₃CN, respectively. The material that eluted at 24-29% CH₃CN consisted of two different peptides with an observed average compound mass of 1665.42 and 1722.58, as measured by ESI-MS (Fig. 2C). Amino acid analysis and Edman degradation confirmed the structure of these two peptides to be NH₂-HHHHHHAQALPET-Gly-Gly-(Gly)-COOH (calculated mass 1665.76 and 1722.81), indicating that they differed by the addition of a COOH-terminal glycine residue (mass 57) (Table I). (To facilitate identification of residues within anchor peptides, all amino acids specified by the mRNA coding sequence are printed in the single-letter code, whereas linked cell wall residues are indicated in the three-letter code.) This result confirmed our previous observation that COOH-terminal anchor peptides can be liberated from the staphylococcal cell wall by the cleavage with lysostaphin between the second and the third glycyl residue of the pentaglycine cross-bridge (9).

Table I. Edman degradation of anchor peptides Aliquots of HPLC-purified anchor peptides were analyzed by electrospray ionization mass spectrometry and Edman degradation. The picomolar concentration of observed phenylthiohydantoin-coupled amino acid peaks per Edman sequencing cycle is reported. Residue 1722 Da 2235 Da 2713 Da 1-His (Ala)a 99 791 (709) 2231 (195) 2-His 138 796 (173) 1893 (64) 3-His 160 773 1889 4-His 164 749 1779 5-His 183 672 1350 6-His 170 351 986 7-Ala 84 212 577 8-Glu 72 196 401 9-Ala 102 207 626 10-Leu 56 163 331 11-Pro 28 66 118 12-Glu 29 67 121 13-Thr 19 54 14-Gly 31 73 126 15-Gly 45 116 218 16-Gly 37 146 347 17-Gly 193 462 18-Gly 165 574 a During cycle 1, the 2235-Da amidase-released anchor peptide yielded two phenylthiohydantoin-coupled amino acids (histidine and alanine) at equimolar concentrations, whereas the compounds with mass 1722 Da and 2713 Da did not.

The anchor peptides that eluted from the RP-HPLC at 34-37% CH₃CN were ESI-MS measured to be peptides with masses of 3777, 3830, and 3858 (data not shown). Edman degradation and amino acid analysis revealed that these compounds consisted of COOH-terminal anchor peptides with additional 16 upstream residues of Seb sequence (NH₂-VDSKDVKIEVYLTTKKGTMHHHHHHAQALPET-Gly-Gly-(Gly)-COOH). The observed compound masses can thus be explained as anchor peptides with two and three glycyl residues linked to the threonyl of the LPXTG motif, similar to the peptides described above. This result indicated that the CnBr cleavage at the engineered methionine-histidine site occurred with reduced efficiency (50-60%) as compared with the cleavage at other methionyl residues (data not shown).

Muralytic amidases cleave the bacterial wall peptides at their amide linkage between the amino of L-alanyl and the carboxyl of N-acetylmuramic acid (19, 20). Since no such enzyme was available to us, we cloned, overexpressed, and purified the amidase gene of staphylococcal phage 11 (21) in E. coli.² Amidase cleavage of the staphylococcal peptidoglycan solubilized Seb-MH₆-Cws as two distinct species, both of which migrated more slowly on SDS-PAGE than the lysostaphin-released counterpart (Fig. 1). Edman degradation of amidase-solubilized Seb-MH₆-Cws (without prior CnBr cleavage) revealed the expected NH₂-terminal sequence (ESQPDP). In contrast to the sequencing data obtained for the lysostaphin-released species, we observed a half-molar amount of phenylthiohydantoin-coupled alanine in addition to the NH₂-terminal glutamic acid after the first cleavage cycle (data not shown). This result indicated that some of the amidase-released species harbored two amino-terminal residues, glutamic acid and alanine, consistent with a branched peptide structure and with the cleavage of the amide bond between L-alanyl and N-acetylmuramic acid by the added amidase (22).

After CnBr cleavage and affinity purification, the amidase-released anchor peptides were subjected to MALDI-MS (Fig. 3A). Five major ions with m/z 2235, 2717, 2758, 4335, and 4857 were identified and these compounds were further purified by RP-HPLC (Fig. 3B). An average ESI-MS mass of 2235.05 was observed for the anchor peptide eluting at 26% CH₃CN (Fig. 3C). When subjected to Edman degradation, this compound yielded the sequence NH₂-A/HHHHHHAQALPET-Gly-Gly-Gly-Gly-Gly-COOH and two phenylthiohydantoin-coupled residues, histidine and alanine, were identified at equimolar concentrations during cycle 1 (Table I). This result is consistent with the structure of a branched peptide consisting of surface protein linked via the pentaglycine cross-bridge to the -amino of lysine in the wall peptide and a calculated average mass of 2235.35 as shown in Fig. 4. An NH₂-terminal alanyl of the wall peptide, liberated via amidase cleavage, is likely linked to D-isoglutamine and presumably no further amino acid was released from the wall peptide due to the nature of its iso-peptide bond (23) (see Fig. 4 for a structural model). The amino acid analysis was consistent with the presented structure (Table II).

Table II. Amino acid analysis of anchor peptides Aliquots of HPLC-purified anchor peptides were analyzed by amino acid analysis. Samples were hydrolyzed in hydrochloric acid and liberated amino acids were quantitated by HPLC and are indicated as the picomolar amount per sample. On the basis of these measurements the residue per mole amount (res./mol) of each amino acid was calculated. Residue 1722 Da 2235 Da 2713 Da Observed Calculated Observed Calculated Observed Calculated pM res./mol pM res./mol pM res./mol Ala 204 2.0 256 4.8 1231 4.1 Glu/Gln 202 2.0 104 1.9 808 2.7 Gly 325 3.3 315 5.9 1563 5.1 His 542 5.4 258 4.8 1611 5.3 Lys 0 19 0.4 264 0.9 Leu 100 1.0 55 1.0 252 0.8 Pro 114 1.1 87 1.6 330 1.1 Thr 130 1.3 85 1.6 329 1.1

To characterize the entire structure of the compound with mass 2235, we performed CID of the triply charged parent ion (m/z 746) in an MS/MS experiment. The resulting daughter ion spectrum was examined for the presence of breakdown products (15, 24, 25) and compared with our structural model (Fig. 4). The structure of the branched peptide was revealed by the presence of the daughter ions at m/z 201, 258, 342, 399, and 982. The ion at m/z 201 was interpreted to be Ala--Gln. The observation that no glutamine was released by Edman degradation during cycle two suggests that the two amino acids are linked by the characteristic isopeptide bond. The ion at m/z 399 comprises the entire stem peptide (NH₂-Ala--Gln-Lys-Ala-COOH), whereas the ion at m/z 342 represented the loss of Ala from this structure (NH₂-Ala--Gln-Lys-COOH). Evidence for the bond between the fifth glycyl and the -amino of lysyl was provided by the fact that Edman degradation during cycles 19 and higher (i.e. after the fifth glycyl) did not release phenylthiohydantoin-coupled amino acids (Table I) and by the presence of the ions at m/z 258 (NH₂-Lys-(NH₂-Gly)-Ala-COOH) and 982 (NH₂-Lys(NH₂-H₆AQALPET-Gly₅)-COOH), which comprise this bond. A summary of the observed daughter ions and their interpretation is presented in Table III.

Table III. Summary of daughter ions produced during MS/MS of the 2235-Da amidase-released anchor peptide Observed m/za Charge state Calculated m/zb  obs-calc c Proposed structure Ion typed 83.4e +3 83.4 0 HH a2 109.5 +1 110.1  0.6 H a1 182.7 +1 184.2  1.5 AL  e 201.0 +1 200.2 +0.8 Ala-Gln b2 257.7 +1 256.3 +1.4 Lys-(NH2-Gly)-Ala  e 275.4 +1 275.3 +0.1 HH b2 342.0 +1 342.4  0.4 Ala-Gln-Lys (+15) c3/b3 399.0 +1 399.5  0.5 Ala-Gln-Lys-Ala b4 412.5 +1 412.4 +0.1 HHH b3 447.3 +2 448.0  0.7 H6A b7 512.1 +2 512.1 0 H6AQ b8 547.8 +2 547.6 +0.2 H6AQA b9 588.9 +1 588.7 +0.2 Ala-Gln-Lys-(NH2-Gly3)-Ala y7 604.5 +2 604.1 +0.4 H6AQAL b10 678.9 +3 679.1  0.2 Lys-(NH2-H6AQALPET-Gly5)-Ala c20 693.3 +2 694.3  1.0 H6AQALPE (H2O) a12 717.3 +2 717.3 0 H6AQALPE b12 745.8 +3 Parent ion 824.4 +1 823.8 +0.6 H6 b6 882.0 +1 881.9 +0.1 AQALPET-Gly3  e 909.9 +2 910.4  0.5 H6AQALPET-Gly5 b18 981.9 +2 981.6 +0.3 Lys-(NH2-H6AQALPET-Gly5) y20 1028.7 +1 1030.1  1.4 Ala-Gln-Lys-(NH2-PET-Gly5)-Ala y12 1089.3 +2 1089.7  0.4 Parent ion 58 Da  f 1206.0 +1 1207.3  1.3 H6AQAL b10 a See Fig. 4A for relative intensity of daughter ions. b Calculations are based on average masses according to the MacBioSpecTM program. c The difference between the observed and calculated masses of daughter ions. d Nomenclature refers to NH2- and COOH-terminal cleavage fragments according to Roepfstorff and Fohlman (51) and Biemann (52, 53). Using the same nomenclature, the prime () labeled cleavage fragments are predicted to arise from the second NH2 terminus of the branched anchor structure. e Fragments thought to arise by two cleavages are calculated as the sum of the residue masses. f Interpreted to result from a rearrangement of the parent ion.

The anchor peptide with an observed ESI-MS average mass of 2713.48 eluted at 28% CH₃CN and upon Edman degradation also yielded a sequence of NH₂-HHHHHHAQALPET-Gly-Gly-Gly-Gly-Gly-COOH. However, in contrast to the compound with a mass of 2235, no release of phenylthiohydantoin-coupled alanine was observed during the first cycle. These observations are consistent with the linkage of surface protein to a cell wall tetrapeptide that is amide-linked to N-acetylmuramyl-(1-4)-N-acetylglucosamine (19, 26) (calculated average compound mass 2713.83). This was a surprising finding because no N-acetylglucosaminidase or N-acetylmuramidase had been added that would have released surface proteins with linked disaccharide. One explanation for this result may be that the well known endogenous N-acetylglucosaminidase of S. aureus (20, 27) had released surface proteins during the prolonged incubation with amidase. Nevertheless, the availability of the compound with mass 2713 provided us with an opportunity to study the linkage of anchor peptides to the glycan strands of the staphylococcal cell wall.

The structure of the 2713-Da anchor fragment was characterized by MS/MS using the parent ion at m/z 905. The resulting daughter ion spectrum was compared with our model (Fig. 5). The presence of N-acetylhexosamines in this structure was revealed by the characteristic ions at m/z 187, 168, and 138 (25). N-Acetylhexosamine was also indicated as the compound with mass 203, and the resulting branched peptide with linked N-acetylmuramic acid generated doubly and triply charged ions at m/z 1246 and 832, respectively (MurNAc-L-Ala-D--Gln-L-Lys(NH₂-H₆AQALPET-Gly₅)-D-Ala-COOH). The linkage of L-alanine in the stem peptide to the lactyl moiety of N-acetylmuramic acid could be demonstrated with the ions at m/z 1154 and 770 (lactyl-Ala--Gln-Lys(NH₂-H₆AQALPET-Gly₅)-Ala-COOH). Several other ions confirmed the sequence of the stem peptide to be identical to that of the amidase-released anchor peptide (Fig. 5). Although the data identify the presence of N-acetylhexosamine and N-acetylmuramic acid, they cannot determine the sequence of these two sugars in the disaccharide-linked anchor peptide. If this sample resulted from the solubilization by a glucosaminidase, its structure would likely be MurNAc-(Ala--Gln-Lys(NH₂-H₆AQALPET-Gly₅)-Ala-COOH)-(1-4)-GlcNAc. All daughter ions observed in this MS/MS experiment are listed in Table IV for a comparison of the observed and calculated masses with the proposed structures.

Table IV. Summary of daughter ions produced during MS/MS of the 2713-Da compound Observed m/za Charge state Calculated m/zb  obs-calc c Proposed structure 84.2 +3 83.4 +0.8 HH (a2) 126.5 +1 127.1  0.6 Propionyl-Ala 138.2 +1 138.1 +0.1 (3)d 168.2 +1 168.2 0 (2)d 186.8 +1 186.2 +0.6 (1)d 203.0 +1 204.2  1.2 GlcNAc 274.4 +1 276.3  1.9 MurNAc 302.0 +1 300.4 +1.6 Gln-Lys-Ala (28) 329.6 +1 328.4 +1.2 Gln-Lys-Ala 546.5 +1 547.6  1.1 H6AQA (b9)e 603.5 +2 604.1  0.6 H6AQAL (b10)e 701.9 +3 702.4  0.5 Gln-Lys-(NH2-H6AQALPET-Gly5)-Ala (43)f 721.7 +3 722.4 +0.7 Gln-Lys-(NH2-H6AQALPET-Gly5)-Alah 770.3 +3 770.8  0.5 Propionyl-Ala-Gln-Lys-(NH2-H6AQALPET-Gly5)-Alaf 832.1 +3 832.1 0 MurNAc-Ala-Gln-Lys-(NH2-H6AQALPET-Gly5)-Alag 905.0 +3 Parent ion 1051.7 +2 1053.2  1.5 Gln-Lys-(NH2-H6AQALPET-Gly5)-Ala (43)h 1082.9 +2 1083.2  0.3 Gln-Lys-(NH2-H6AQALPET-Gly5)-Alah 1154.3 +2 1155.7  1.4 Propionyl-Ala-Gln-Lys-(NH2-H6AQALPET-Gly5)-Alaf 1207.7 +1 1207.3 +0.4 H6AQAL (b10)e 1246.1 +2 1247.7 +1.6 MurNAc-Ala-Gln-Lys-(NH2-H6AQALPET-Gly5)-Alag 1286.9 +1 1287.3 +0.4 MurNAc-Ala-Gln-Lys-(NH2-PET-Gly5)-Alaf a See Fig. 5A for the relative intensity of daughter ions. b Calculations were performed as described in Table III. c The difference between the observed and calculated masses of daughter ions. d See Fig. 5A for an explanation of the predicted structure. e Nomenclature refers to NH2- and COOH-terminal cleavage fragments according to Roepfstorff and Fohlman (51) and Biemann (52, 53). f Calculated as residue mass. g Calculated as residue mass minus 17 (OH). h Calculated as residue mass plus 18 (H2O).

Average compound masses of 4342, 4821, and 4849 were measured by ESI-MS for amidase-released peptides eluting at 36-38% CH₃CN (Fig. 3). Similar to the observations made with the partial CnBr cleavage of lysostaphin-solubilized Seb-MH₆-Cws, these mass measurements are consistent with peptides harboring an additional 16 upstream residues of Seb fused to the amidase-released anchor structures as a result of an uncleaved methionine adjacent to the engineered histidine tag.

Muramidase cleaves the staphylococcal cell wall at the glycan chains (20), i.e. the 1-4 glycosidic bond between N-acetylmuramic acid and N-acetylglucosamine, and released Seb-MH₆-Cws as a spectrum of species with increasing mass due to linked peptidoglycan (Fig. 1). NH₂-terminal sequencing of the muramidase-solubilized Seb-MH₆-Cws also yielded two phenylthiohydantoin-coupled residues after the first cycle: glutamic acid and alanine. The presence of phenylthiohydantoin-alanine can be explained by the alkali release of ester-linked D-alanine from cell wall teichoic acid (28, 29) during the sequencing cycle. Because D-alanyl-decorated polyribitol teichoic acids are linked to N-acetylmuramic acid within the glycan chains, muramidase digestion of the staphylococcal peptidoglycan can solubilize surface proteins with attached teichoic acids (30, 31). Although several attempts were made to purify large quantities of the muramidase-released, CnBr-cleaved anchor species, we failed to obtain sufficient material for a more detailed chemical, ESI-MS, or MS/MS analysis.

DISCUSSION

The cell wall anchor structure of surface proteins in S. aureus has been revealed by a combination of mass spectrometry and molecular biology techniques. Surface proteins are tethered to the cell wall via an amide bond between the carboxyl of their COOH-terminal threonine and the amino of the pentaglycine cross-bridge, which is attached to the -amino of lysyl within a peptidoglycan tetrapeptide. This branched anchor peptide is linked to N-acetylmuramic acid, thereby attaching surface proteins to the glycan chains of the staphylococcal cell wall. We think that surface proteins may be linked to a precursor molecule of peptidoglycan synthesis rather than to the mature, assembled cell wall, because the staphylococcal peptidoglycan is highly cross-linked and has very few free pentaglycine amino groups (11, 23, 32-34). During the biosynthesis of bacterial peptidoglycans, a membrane-bound disaccharide-precursor molecule (undecaprenyl-PO₄-PO₄-MurNAc-(L-Ala-D--Gln-L-Lys-(NH₂-Gly₅)-D-Ala-D-Ala-COOH)-(1-4)-GlcNAc) is assembled in the cytoplasm (35, 36) and translocated across the membrane to serve as a substrate for transglycosylation and transpeptidation reactions (37, 38) (Fig. 6). Such a precursor molecule could also serve as the peptidoglycan substrate for the cell wall sorting reaction of surface proteins. To characterize a cell wall sorting intermediate consisting of surface protein with linked peptidoglycan precursor, we have begun to purify and characterize such a compound from staphylococcal protoplasts.³ The structural elucidation of such an intermediate may thus resolve the nature of the peptidoglycan substrate for the cell wall sorting reaction.

Digestion of the staphylococcal peptidoglycan with either lysostaphin or amidase solubilized surface proteins as species with uniform mass. This can be explained by our model of the cell wall anchor structure (Fig. 6). Both lysostaphin and amidase cleaved the branched anchor peptide at sites that are proximal to its linkage with the glycan strands. Hence, a single enzymatic cut completely solubilized surface proteins from the otherwise tightly cross-linked cell wall of S. aureus. In contrast, muramidase digestion released surface proteins as a large spectrum of fragments with increasing mass due to linked peptidoglycan fragments. If the anchor peptides were randomly linked to N-acetylmuramyl, i.e. anywhere along the glycan chains of S. aureus, the release of surface proteins would require no more than two muramidase cuts at neighboring sites. In such a scenario, one would expect surface proteins also to be solubilized as a uniform species rather than the observed spectrum of fragments. An alternative possibility, which would be consistent with the data presented here, is that surface proteins may be tethered to distinct sites of the glycan chains that cannot be directly cleaved by muramidase. Although an N-acetylmuramic acid-linked anchor peptide has been characterized in this report, its flanking sugar moieties and relative position within the glycan chains have not yet been identified.

The disaccharide containing compound with mass 2713 was found in the amidase but not in the muramidase-digested sample; however, the source of this molecule could not be determined. One explanation for its occurrence is that the 2713-Da compound may have been solubilized from the staphylococcal peptidoglycan by a specific enzymatic activity, for example N-acetylglucosaminidase (20, 27). Alternatively, this compound may represent a peptidoglycan (disaccharide) precursor molecule that is linked to surface protein. Such precursors are generally attached to lipid; however, the trifluoroacetic acid treatment of the amidase-digested sample could have removed a phosphodiester-linked undecaprenyl decoration.

Because all cell wall anchor peptides of surface proteins identified here consist of unsubstituted tetrapeptides, we propose that surface proteins linked to a peptidoglycan precursor molecule could be incorporated into the cell wall by a transglycosylation reaction (39) (Fig. 6). The linked anchor structure may subsequently be cleaved at its terminal D-alanyl-D-alanine by a carboxypeptidase activity (38) to generate the mature anchor peptide unit. This model takes into account that all cell wall peptides are thought to first be synthesized as a pentapeptide precursor by the addition of D-alanyl-D-alanine to the muramyl-linked tripeptide in the bacterial cytoplasm (40). After assembly into the cell wall, the pentapeptides may be either cross-linked to generate substituted tetrapeptides or cleaved by carboxypeptidases, thereby yielding free (unsubstituted) tetrapeptides (38). Nevertheless, it also seems possible that pentapeptide-peptidoglycan precursor molecules may be cleaved by carboxypeptidases and that the resulting tetrapeptide precursors serve as substrates for the sorting reaction of surface proteins. Our current data cannot distinguish between these possibilities, but they may be addressed in the future by combining our mass spectrometric analysis of anchor structures with the antibiotic inhibition of peptidoglycan precursor synthesis.

A notable feature of the cell wall anchor structure of surface proteins in S. aureus is its lack of cross-linking with neighboring wall peptides. The regulated release of proteins from the bacterial surface has been proposed for several different Gram-positive organisms (41, 42). Such release of surface proteins is thought to be important during the pathogenesis of bacterial infections as a mechanism for either the escape from the host's immune surveillance or the selective alteration of bacterial surface properties. An extensive cross-linking between anchor peptides and the neighboring cell wall would require a multitude of muralytic enzymes to release surface proteins from the peptidoglycan. The demonstrated anchor structure, however, can easily be released by an enzymatic activity that selectively cleaves the glycan strands of the cell wall without an extensive solubilization of the bacterial peptidoglycan. We thus predict that surface proteins released into bacterial culture supernatants harbor COOH-terminally linked cell wall tetrapeptides with attached glycan fragments.

Murein (Braun's) lipoprotein is a structural element of the envelope of Gram-negative bacteria (43, 44). In contrast to Gram-positive surface proteins, which display their free NH₂-terminal domains on the bacterial surface, an NH₂-terminal lipid modification inserts Braun's lipoprotein into the inner leaflet of the outer membrane (the N-terminal cysteine contains thioether-linked, 2,3-palmitoylated glycerol as well as N-acyl palmitoyl) (45). The short polypeptides are thought to be assembled into a trimeric structure (46) and about one third of all murein lipoprotein is linked to the peptidoglycan (47, 48). This is accomplished via an amide linkage between the -amino of the lipoprotein's COOH-terminal lysine and the carboxyl of m-diaminopimelic acid in the wall peptide (17, 48). When engineered to harbor a COOH-terminal sorting signal, lipoproteins can also be linked to the cell wall of Gram-positive organisms; however, such polypeptides have not yet been identified to occur naturally in bacterial envelopes (49).

An important conclusion that can be drawn from the cell wall anchor structure is that the amide bond exchange reaction of surface protein sorting is not sensitive to the -lactam inhibition of the classical transpeptidation reaction catalyzed by penicillin-binding proteins (50). We tested this prediction and found that neither methicillin nor penicillin G could inhibit surface protein anchoring to the staphylococcal cell wall (data not shown). The sorting reaction may thus represent a novel target for antibiotic therapy if its amide bond exchange reaction proves to be essential for either viability or pathogenesis.