Structure-based mechanism of lipoteichoic acid synthesis by Staphylococcus aureus LtaS

Abstract

Staphylococcus aureus synthesizes polyglycerol-phosphate lipoteichoic acid (LTA) from phosphatidylglycerol. LtaS, a predicted membrane protein with 5 N-terminal transmembrane helices followed by a large extracellular part (eLtaS), is required for staphylococcal growth and LTA synthesis. Here, we report the first crystal structure of the eLtaS domain at 1.2-Å resolution and show that it assumes a sulfatase-like fold with an α/β core and a C-terminal part composed of 4 anti-parallel β-strands and a long α-helix. Overlaying eLtaS with sulfatase structures identified active site residues, which were confirmed by alanine substitution mutagenesis and in vivo enzyme function assays. The cocrystal structure with glycerol-phosphate and the coordination of a Mn²⁺ cation allowed us to propose a reaction mechanism, whereby the active site threonine of LtaS functions as nucleophile for phosphatidylglycerol hydrolysis and formation of a covalent threonine–glycerolphosphate intermediate. These results will aid in the development of LtaS-specific inhibitors for S. aureus and many other Gram-positive pathogens.

AGram-positive bacterial pathogen, Staphylococcus aureus, is the most frequent cause of skin and soft tissue infections and bacterial sepsis (1, 2). Two frequently highlighted aspects of S. aureus infections are the increase in community-acquired infections and the appearance of multidrug-resistant strains (3, 4). Clones of methicillin-resistant S. aureus (MRSA) strains that are also resistant to the vast majority of other clinically approved antibiotics have been isolated (5). The glycopeptide antibiotic vancomycin is often seen as a last resort to treat such infections; however, strains with either intermediate (VISA strain) or high (VRSA strains) vancomycin resistance have already been reported (6–8). Because of the increasing difficulty in treating infections with multidrug-resistant S. aureus strains, new ways of inhibiting the growth of S. aureus are heavily sought after.

The bacterial-specific peptidoglycan structure and its synthesis enzymes are a common target for antibiotics. Lipoteichoic acid (LTA) is another abundant surface polymer present in the envelope of Gram-positive bacteria (9, 10). Its chemical structure can vary greatly; however, the majority of Gram-positive pathogens including group A and B streptococci, Enterococcus faecalis, S. epidermidis, Bacillus anthracis, and Listeria monocytogenes produce LTA of the same polyglycerol-phosphate type as present in S. aureus (9, 10). Using a genetic screen, the S. aureus gene encoding the enzyme responsible for polyglycerol-phosphate LTA synthesis was recently discovered; this protein of previously unknown function (SAV0719 in the S. aureus MU50 genome) was renamed LtaS for lipoteichoic acid synthase (11). Moreover, the same study provided experimental evidence that LtaS is required for staphylococcal growth under normal physiological conditions, indicating that this enzyme is a new potential drug target (11).

Experimental evidence suggests that the glycerol-phosphate head group of the membrane lipid phosphatidylglycerol (PG) serves as donor for the glycerol-phosphate units, which form the LTA backbone (12). While direct biochemical evidence is lacking, cleavage of the PG head group and subsequent glycerol-phosphate polymerization are presumed to be the reactions catalyzed by LtaS as depletion of this enzyme in S. aureus leads to a complete absence of polyglycerol-phosphate LTA and expression of LtaS in a heterologous Gram-negative bacterial host, which naturally lacks LTA, leads to the production of polyglycerol-phosphate polymers (11).

LtaS and its homologues in other Gram-positive bacteria are predicted to be polytopic membrane proteins with a large enzymatic domain (currently annotated as a sulfatase domain) located on the extracellular side of the bacterial membrane (Fig. 1A). In agreement with this topology prediction, a cleaved fragment of the LtaS protein containing the complete enzymatic “sulfatase” domain was detected in culture supernatant and cell wall-associated fractions (13, 14). The predicted topology not only fits with the widely accepted model that the polyglycerol-phosphate backbone of LTA is synthesized on the outside of the bacterial membrane (15) but also makes LtaS a more attractive drug target due to the extracellular location of the enzyme domain.

Fig. 1.

S. aureus LtaS is efficiently cleaved and localizes to cell wall and supernatant fraction. (A) Schematic representation of LtaS. LtaS is synthesized as a 74.4-kDa protein and cleaved after amino acid 217 following an ALA motif (13, 14). Part of the 49.3-kDa extracellular eLtaS domain (amino acids 245–604) is currently annotated as a sulfatase domain. (B) Cleavage efficiency and subcellular localization of LtaS. Midlog cultures of S. aureus strains RN4220 and COL were fractionated into supernatant (SN), cell wall (CW), and combined membrane and cytoplasmic (C/M) fractions, and LtaS and control proteins SdrD (cell wall anchored), Hla (secreted), SrtA (membrane), and L6 (cytoplasmic) were detected by Western blot using polyclonal rabbit antibodies as indicated on the Right of each section. Sizes of protein standards in kilodaltons are shown on the Left.

To provide insights into the enzyme mechanism of LtaS, we have crystallized and solved the structure of the complete extracellular LtaS domain (eLtaS) at 1.2-Å resolution and of eLtaS bound to glycerol-phosphate at 1.6-Å resolution. This structural information, combined with mutagenesis analysis, allowed us to propose a reaction mechanism for LtaS and to define structural features that distinguish lipoteichoic acid synthases from sulfatases.

Results and Discussion

LtaS Is Efficiently Processed in S. aureus.

In a very recent study we identified the S. aureus enzyme responsible for polyglycerol-phosphate LTA synthesis and renamed this protein of previously unknown function LtaS (11). Proteomics studies performed before a function was ascribed to this protein revealed that at least part of the enzyme is cleaved, releasing a 50-kDa C-terminal fragment into the supernatant and cell wall fraction (13, 14). N-terminal protein sequencing identified the cleavage site after residues ²¹⁵Ala-Leu-Ala²¹⁷ preceding the annotated sulfatase domain, which ranges from amino acids 245 to 604 (13) (Fig. 1A). To determine the extent to which the protein is processed, a cell wall fractionation experiment was performed and LtaS and fractionation-control proteins were detected by immunoblotting. In all S. aureus strains tested (RN4220, COL, Newman, SH1000), LtaS was processed very efficiently and the 50-kDa C-terminal fragment (eLtaS) could be detected in both supernatant and cell wall fractions (Fig. 1B and data not shown). This indicates that despite the fact that LtaS is synthesized as a membrane protein, the C-terminal eLtaS domain may also function as extracellular processed enzyme, to catalyze the polyglycerol-phosphate LTA backbone synthesis.

Overall Structure of the eLtaS Domain.

To provide experimental evidence for the proposed LtaS enzyme activity and gain insight into a possible reaction mechanism, we overexpressed and purified the complete eLtaS domain starting from amino acid 218 as an N-terminal His-tag fusion protein and determined the 3D structure by x-ray crystallography. The eLtaS structure was refined to 1.2 Å and covered all residues except the last 6 C-terminal amino acids, which were not visible in the density map. Data collection, phasing, and refinement statistics can be found in Table S1. The structure revealed that the overall fold of eLtaS is indeed sulfatase-like and shows striking similarity to human sulfatases and the prokaryotic Pseudomonas aeruginosa arylsulfatase (Fig. 2) (16, 17). eLtaS folds into 1 hemispheric unit made up of an α/β core and a C-terminal part composed of 4 anti-parallel β-strands and a long α-helix (Fig. 2 A and B). The α/β core is a shared structural feature of proteins in the alkaline phosphatase family clan, which also includes sulfatases, phosphodiesterases, and metalloenzymes such as phosphoglycerate mutase (PGM) (Fig. 2 C and D). However, with the exception of sulfatases, all other clan members, for which structural information is available, lack the C-terminal anti-parallel β-strands and α-helix and instead have a second globular domain joined to the core by 2 linker sequences. Therefore, sulfatases are the closest structural homologues to eLtaS and comparing our structure with the 1.3-Å P. aeruginosa arylsulfatase structure (16) showed a root mean square deviation (RMSD) of 2.95 Å over 309 Cα's (Fig. 2E).

Fig. 2.

eLtaS and sulfatases share a similar topology. (A) eLtaS ribbon representation in rainbow colors, blue to red from N to C terminus. (B) eLtaS topology diagram in rainbow colors with β-strands represented as arrows and helices as cylinders. (C) Pseudomonas aeruginosa arylsulfatase ribbon model in rainbow colors. (D) Bacillus stearothermophilus phosphoglycerate mutase model in rainbow colors. (E) Superimposed models of eLtaS and P. aeruginosa arylsulfatase (eLtaS in blue and sulfatase in gray). (F) Superimposed models of eLtaS and phosphoglycerate mutase (eLtaS in blue and phosphoglycerate mutase in purple).

Location and Description of the Active Site Pocket.

Given the high degree of structural similarity between eLtaS and sulfatases (Fig. 2E), we rationalized that the location of the eLtaS active site must be similar to that in sulfatases. In the P. aeruginosa arylsulfatase the catalytic residue is located at the N terminus of α-helix B (16), which is equivalent to the N terminus of helix α3 in eLtaS (Fig. 2B). In the sulfatase, a posttranslationally modified cysteine residue, Cα-hydroxyformylglycine (HFG51), acts as the key catalytic residue for sulfate-ester hydrolysis. One hydroxyl group of HFG is activated by a Ca²⁺ ion and the other deprotonated by a histidine (H155) with the Ca²⁺ coordinated by the side chains of 3 aspartic acids (D13, D14, D317) and one asparagine residue (N318). Another histidine (H211) functions as proton acceptor to release the sulfate moiety from its substrate with 2 lysine residues (K113, K375) holding the sulfate group in position (Fig. S1). From the structural superposition, several of these key residues could be readily identified in eLtaS with H416 equivalent to H211 and 3 of 4 cation-binding residues (D317, N318, and D13) equivalent to D475, H476, and E255 in eLtaS. However, the HFG-modified cysteine is replaced by a threonine (T300) while H155, which deprotonates HFG, is absent in eLtaS (Fig. S1). Another striking difference is the absence of lysine residues in eLtaS, which in the sulfatase bind to the sulfate group (Fig. S1). In summary, on the basis of the structural superposition we propose that the location of the eLtaS active site is similar to that of sulfatases. There are clear differences between the active sites, including the replacement of the catalytic HFG residue with a threonine, the absence of key sulfate binding residues, and differences in the coordination of the active site metal ion (see below), which confer different substrate specificity to eLtaS. In addition, shorter loops in eLtaS between β4 and α9 and between β6 and α13 (Fig. 2B) place the proposed active site of eLtaS closer to the surface. This proposed active site pocket lies within a relatively flat surface, which may be related to the proposed association with and enzymatic function on the membrane bilayer.

The Active Center of eLtaS Can Bind a Mn²⁺ Cation.

Among proteins in the alkaline phosphatase clan, the B. stearothermophilus cofactor independent PGM catalyzes the isomerization of 2- and 3-phosphoglycerates and recognizes a substrate that shows similarity to the glycerol-phosphate head group of PG, the presumed substrate of LtaS. The structure of the B. stearothermophilus PGM has been determined (18) and similar to sulfatases and eLtaS comprises a conserved α/β core domain. However, in PGM the catalytic serine residue (S62) and the side chains of 3 additional amino acids (D444, H445, and D12) coordinate a Mn²⁺ cation in a distorted square pyramid, which is different from that of sulfatases that bind Ca²⁺ in an distorted octahedral bipyramidal coordination (Fig. 3). A superposition of the eLtaS structure with the α/β core domain of the B. stearothermophilus PGM (calculated RMSD of 2.25 Å over 193 Cα's) (Fig. 2F) revealed that the side chains of T300, D475, H476, and E255 in eLtaS superimpose with the side chains of S62, D444, H445, and D12 in PGM (Fig. 3 A, E, and F), suggesting that eLtaS could also bind a Mn²⁺ cation. Indeed, in an anomalous difference Fourier map collected at the absorption edge of Mn²⁺, a clear density corresponding to the position of the cation was observed (Table S1) (Fig. 3D), confirming that eLtaS can bind Mn²⁺. The distance between the Mn²⁺ and the oxygen atom of T300 is 2.19 Å, suggesting a deprotonated state of the active site threonine (19). It is notable that apart from the divalent cation binding residues, the rest of the catalytic center between PGM and eLtaS is very different.

Fig. 3.

Coordination of a Mn²⁺ cation in the active center of eLtaS. (A) Coordination bonds between Mn²⁺ in eLtaS and the side chains of residues T300, D475, H476, and E255. (B) Coordination bonds between Ca²⁺ in the P. aeruginosa arylsulfatase and side chains of residues D13, D14, HFG51, D317, and N318. (C) Overlay of A and B. (D) Anomalous difference density map of Mn²⁺ bound to eLtaS. (E) Coordination bonds between Mn²⁺ in B. stearothermophilus PGM and side chains of residues D12, S62, D444, and H445. (F) Overlay of A and E.

Site-Directed Mutagenesis of Active Site Residues Confirms Their Importance for in Vivo LtaS Function.

To confirm that the proposed active site residues are indeed essential for LtaS enzyme function, we performed alanine substitution mutagenesis on the following residues: H416, T300 (presumed catalytic and metal binding residue), E255, D475, and H476 (metal binding residues). We also mutated a number of residues (D458, H409, and G298) that are not predicted to be involved in catalysis as controls. LtaS variants were tested for in vivo function by expressing the mutant alleles from an anhydrotetracyline-inducible promoter in S. aureus strain ANG499 in which the chromosomal ltaS copy is placed under the tightly controlled isopropyl thiogalactoside (IPTG)-inducible Spac promoter (Fig. 4A). In the absence of IPTG, strain ANG499 does not produce LTA and ceases to grow ≈4 h after IPTG removal unless a functional LtaS protein is produced from a different locus (Fig. 4 B–D; ref. 11). LtaS variants E255A, T300A, T300V, H416A, D475A, and H476A with amino acid substitutions in key positions are nonfunctional as seen by their inability to complement the LTA synthesis and growth defect of strain ANG499 after IPTG removal (Fig. 4 B and C). Control variants LtaS-D459A, LtaS-H409A, and LtaS-G298A retained activity (Fig. 4 B and D). However, LtaS-G298A with an amino acid substitution in very close proximity to the active site threonine (T300) displayed reduced activity, as seen by the production of lower amounts of LTA. Immunoblot analysis of supernatant and cell wall fractions revealed that all LtaS variants were expressed in S. aureus, albeit at different levels. The observed reduction in protein level cannot be the reason for the lack of complementation, as we know from expression studies that small but detectable amounts of LtaS protein with wild-type activity would be sufficient for complementation (data not shown). Interestingly, concomitant with a loss of function, we also observed a decrease in LtaS processing and accumulation of full-length protein (Fig. 4B), the reasons for which are still unclear. Taken together, these results show that amino acids predicted from the structure to be essential for enzyme function are indeed required for LtaS activity in vivo.

Fig. 4.

Amino acid substitutions of key active site residues render LtaS inactive. (A) Schematic representation of chromosomal organization in S. aureus strains expressing different LtaS variants. (B) S. aureus strains expressing LtaS variants with amino acid substitutions in key active site residues are unable to produce LTA. (Top) S. aureus cell wall extracts were separated on 15% PAA gels and LTA was detected by Western blot using a monoclonal polyglycerol-phosphate-specific antibody. (Middle and Bottom) S. aureus cell wall extracts and supernatant fractions were separated on 10% PAA gels and LtaS protein was detected by Western blot. Locations of full-length LtaS protein and cleaved eLtaS fragment are indicated to the Right by a shaded and a solid star, respectively. Amino acid substitutions in LtaS are indicated at the Top and sizes of protein standards in kilodaltons are on the Left. (C and D) Bacterial growth curves. Expression of the indicated LtaS variant was induced by the addition of 300 ng/ml anhydrotetracycline and bacterial growth recorded. (C) Inactive LtaS variants. (D) Active LtaS variants together with a positive control (solid circle, wild-type LtaS) and a negative control (solid square, empty vector).

Cocrystal Structure of eLtaS with Glycerol-Phosphate.

For PGM and sulfatases, high-resolution structures with bound substrate or substrate analogues made it possible to propose catalytic mechanisms for these enzymes (16, 17). To gain similar insight into the eLtaS reaction mechanism, we cocrystallized eLtaS and an inactive eLtaS-T300A variant with glycerol-phosphate (mimicking the head group of PG) and refined the structures to 1.6- and 1.7-Å resolution, respectively [Fig. 5(eLtaS) and Fig. S2 (eLtaS-T300A)]. Glycerol-phosphate bound and unbound as well as wild-type and eLtaS-T300A structures are basically identical; however, the cocrystal structures revealed several specific contacts between the side chains of active site amino acids and the glycerol-phosphate moiety (Fig. 5). The Mn²⁺ cation remained coordinated by the side chains of T300, E255, D475, and H476 in the cocrystal structure and was in addition coordinated by 1 oxygen atom of the phosphate group. The phosphate group was further stabilized by hydrogen bonding to H416 and W354 (Fig. 5A). Two hydrophobic residues F353 and L384 pack against W354 and H416, respectively, stabilizing their positions. The carbon chain of glycerol-phosphate is positioned at the end of a larger pocket (Fig. 5B) and the hydroxyl groups form hydrogen bonds to H347, D349, and R356 (Fig. 5A). We interpret this structure as a presubstrate hydrolysis state, where the catalytic threonine residue (T300) is ready to attack the phosphor atom and release glycerol-phosphate from a substrate forming a potential covalent glycerol-phospho-threonine intermediate. It is interesting to note that most of the glycerol-phosphate substrate-binding residues are located on the same loop. The importance of several of these residues (H347, D349, W354, and R356) for LtaS enzyme function was confirmed by alanine substitution mutagenesis, using the in vivo complementation assay (Fig. S3).

Fig. 5.

Cocrystal structure of eLtaS with glycerol-phosphate. (A) Ribbon representation of eLtaS with bound glycerol-phosphate (Right) with enlarged region showing side chains of active site amino acids, which make contact with glycerol-phosphate and the Mn²⁺ cation (Left). (B) Space-filled model of eLtaS with bound glycerol-phosphate.

Proposed Reaction Mechanism and Model for LtaS-Mediated PG Cleavage and LTA Synthesis.

On the bases of the high-resolution structures of eLtaS, the identification and location of the Mn²⁺ ion, and the positioning of the bound glycerol-phosphate, we propose the following reaction mechanism for cleavage of the glycerol-phosphate head group from PG (Fig. S4A): in the apo form of LtaS, T300 is bound to a Mn²⁺ cation with the hydroxyl group deprotonated. This notion is supported by a short coordination bond distance of 2.19 Å between Mn²⁺ and the oxygen side chain atom of T300 (Fig. 3A). The PG lipid is docked into the binding pocket, with the glycerol-phosphate head group positioned by specific interactions within the substrate-binding loop and the phosphate moiety bound to the Mn²⁺ in a geometry that would favor nucleophilic attack by the activated T300. A trigonal bipyramidal transition state is stabilized by the Mn²⁺ cation followed by the formation of a potential covalent glycerol-phospho-threonine intermediate as depicted in Fig. S4B and similarly to that proposed for PGM, although not observable in our crystal structures. The generated lipid diacylglycerol (DAG) product is protonated by H416 and released from the enzyme. After generation of the glycerol-phospho-enzyme intermediate, additional reaction steps are necessary for the production of polyglycerol-phosphate chains. Experimental evidence supports a model in which the polyglycerol-phosphate chain is extended by sequential addition of glycerol-phosphate moieties at the tip of the growing chain (20, 21). If this is the case, our proposed covalent glycerol-phosphate-threonine intermediate (Fig. S4B) would be resolved by nucleophilic attack of the terminal OH group of the growing chain, thereby generating an empty enzyme and an LTA chain extended by 1 glycerol-phosphate unit. However, how the recognition of the growing chain might occur is currently not clear and we cannot rule out that other proteins or the membrane portion of LtaS itself play a role in the extension process. A future challenge will be to address some of these issues, for instance, by using different purified protein components and lipid substrate in an in vitro assay system.

Our structure-function studies have established that the catalytic center of LtaS is located in a well-defined pocket close to the surface of the eLtaS domain, which we propose to be positioned on top of the membrane bilayer. The eLtaS domain is initially tethered by a linker region to the N-terminal transmembrane helices but then efficiently cleaved during S. aureus growth (Fig. 1). LtaS cleavage may have a regulatory function and, for instance, play a role in chain length determination or even lead to a switch in substrate specificity, as has been described recently for the Escherichia coli protein MdoB, which has an enzymatic activity similar to that of LtaS (22). Interestingly, in LtaS and homologues in B. subtilis and B. anthracis (13, 23, 24), the proteolytic cleavage site is immediately preceded by an AXA sequence motif, which is reminiscent of a type I signal peptidase cleavage site, a peptidase that in S. aureus is encoded by spsB (25). Cleavage of polytopic membrane proteins by signal peptidase I has been reported previously (26). However, typically the cleavage sites are located 3–7 aa after the hydrophobic core of an N-terminal signal peptide (27). Although the AXA motif in S. aureus LtaS is 40 aa from the N-terminal transmembrane domain, it is possible that the linker region is proximal to the membrane either by interacting directly with the eLtaS domain or via the close spatial proximity of eLtaS to the lipid bilayer. Both scenarios would allow recognition of the cleavage site by integral membrane peptidase like SpsB in S. aureus.

Conservation of Active Site Residues in LtaS Homologues.

The presented crystal structures allowed us to identify key active site residues in LtaS [T300 (catalytic residue and metal binding); E255, D475, and H476 (metal binding); H416 (substrate binding and leaving group protonation); and H347, D349, F353, W354, R356, and L384 (substrate binding)], which differ considerably from those found in sulfatases. A ClustalW2 alignment (http://www.ebi.ac.uk/Tools/clustalw2/) of proteins with significant similarity to LtaS (E values < e⁻⁴⁰) revealed that the metal binding residues E255, D475, H476, and H416 (proposed in protonating the leaving group) are completely conserved in all these proteins. The active site threonine is highly conserved but in a few cases replaced by a serine, which could also act as a nucleophile to form a covalent phospho-intermediate. Some variation is seen in the substrate-binding loop (347–356 HGDYKTFWNR in LtaS), although on the basis of our eLtaS crystal structure, we predict that at least 150 proteins that contain the consensus sequence HxD/NxxFW/YNR will recognize and hydrolyze PG. These results suggest not only that LtaS homologues exist in other Gram-positive bacteria but also that key structural features of the LtaS catalytic pocket, as revealed in this study, are conserved. As shown in this and a previous study, LtaS function is required for S. aureus growth under normal physiological conditions and proper cell division (11). Therefore, LtaS and its homologues could serve as targets for unique antibacterial drugs that would also target currently widespread drug-resistant Gram-positive pathogens.

Materials and Methods

Bacterial Strains.

A complete list of bacterial strains used in this study can be found in Table S2. E. coli and S. aureus strains were grown at 37 °C in Luria–Bertani (LB) and tryptic soy broth (TSB), respectively, supplemented with antibiotics as indicated in Table S2.

Plasmid and Strain Construction.

Primers and sequences are listed in Table S3. Plasmid pProEX-eLtaS was constructed for overexpression of the eLtaS domain fused to an N-terminal His-tag and used for purification and structure determination. Primer pair 5′-BamHI-SAV719-Cterm and 3-XbaI SAV0719 and RN4220 chromosomal DNA were used to amplify the eLtaS domain starting from base 652. The resulting PCR product was cut with BamHI and XbaI and ligated with pProEX HTb restricted with the same enzymes, yielding plasmid pPoEX-eLtaS. Plasmid pProEX-eLtaS was initially recovered in E. coli XL1 Blue (strain ANG558) and transformed for protein and seleno-methionine-labeled protein overproduction into E. coli Rosetta and B834, yielding strains ANG571 and ANG563, respectively. Plasmid pProEX-eLtaS-T300A (LtaS variant with Thr to Ala substitution at amino acid position 300) was constructed by QuikChange mutagenesis (Stratagene), using primer pair 5-SAV0719-T300A and 3-SAV0719-T300A. Plasmid pProEX-eLtaS-T300A was initially recovered in strain XL1 Blue and transformed for protein expression into the Rosetta strain, yielding strains ANG572 and ANG575, respectively. A panel of LtaS variants with single-amino-acid substitutions was initially created in the E. coli vector pOK-ltaS (11), which contains the full-length ltaS gene under its native promoter control. The QuikChange method was used to change amino acids E255, G298, T300, H347, D349, W354, R356, H409, H416, D458, D475, and H476 to alanines (A) and T300 to valine (V) (see Table S3 for primer sequences). To analyze the functionality of the different LtaS variants in S. aureus, mutant ltaS alleles were PCR amplified using primer pair 5-AvrII-SAV719–22bp and 3-BglII-SAV719 and cloned under tetracycline-inducible promoter control into AvrII and BglII sites of the S. aureus integration vector pitet (28). Plasmids were initially obtained in E. coli strain XL1 Blue (strains ANG1110–ANG1118 and ANG1175–ANG1180) and subsequently electroporated and integrated into the lipase gene geh of S. aureus strain ANG499 (11) with IPTG-inducible ltaS expression, yielding strains ANG1119–ANG1127 and ANG1181–ANG1186. DNA sequences of all plasmid inserts were verified by fluorescence automated sequencing at the Medical Research Council CSC Genomic Core Laboratory Facility, Imperial College London.

LtaS Antibody Production and S. aureus Cell Fractionation.

Standard procedures were used for LtaS antibody production and cell fractionation experiments; details can be found in SI Text.

LtaS Growth Complementation and LTA and LtaS detection in S. aureus.

LtaS growth complementation assays and LTA detection were essentially performed as previously described (11); for experimental details and LtaS detection see SI Text.

Protein Expression, Purification, and Crystallization Conditions.

Standard Ni-affinity and size exclusion chromatography was used for protein purification; for experimental details see SI Text. Protein crystals were grown by the vapor diffusion method from a solution containing 200 mM ammonium acetate, 30% PEG in 100 mM sodium citrate buffer pH 5.6 and 10 mg/ml eLtaS protein or 6 mg/ml Se-Met substituted eLtaS protein. Glycerol-phosphate cocrystals were grown from the same solution containing 30 mg/ml glycerol-phosphate, 10 mM MnCl₂ and 6–10 mg/ml eLtaS or eLtaS-T300A protein.

Data Collection, Processing, Structure Determination, and Refinement.

Native eLtaS crystals were transferred for cryoprotection into crystallization buffer containing 10% ethylene glycol and subsequently submerged into liquid nitrogen. Crystals of ligand-bound protein were frozen without prior buffer exchange. Data for native eLtaS and Se-Met substituted eLtaS were collected at the Daresbury synchrotron radiation facility, data sets for the glycerol-phosphate-bound eLtaS and eLtaS-T300A were collected at the Diamond synchrotron radiation facility, and the manganese anomalous data set for eLtaS was collected at the European synchrotron radiation facility. All data were integrated in Mosflm and scaled in Scala from the CCP4 package (29). The multiwavelength anomalous dispersion (MAD) method was used to obtain experimental phases from Se-Met eLtaS crystals. Se atom sites were calculated in Shelx (30) and refined in Sharp (31). The initial model was built in ARP/warp (32) and further model building and rebuilding was performed in Coot (33). Structures were refined in Refmac5 (29) and statistical numbers are listed in Table S1. To determine the identity of the active site metal ion, an anomalous difference Fourier map was calculated from a data set collected at an x-ray wavelength near the Mn²⁺ absorption edge. All numbers were generated in Pymol (http://pymol.sourceforge.net/) and coordinates of structures are deposited in the Protein Data Bank.