Crystal Structure of Streptococcus pyogenes Sortase A

Sortases are a family of Gram-positive bacterial transpeptidases that anchor secreted proteins to bacterial cell surfaces. These include many proteins that play critical roles in the virulence of Gram-positive bacterial pathogens such that sortases are attractive targets for development of novel antimicrobial agents. All Gram-positive pathogens express a “housekeeping” sortase that recognizes the majority of secreted proteins containing an LPXTG wall-sorting motif and covalently attaches these to bacterial cell wall peptidoglycan. Many Gram-positive pathogens also express additional sortases that link a small number of proteins, often with variant wall-sorting motifs, to either other surface proteins or peptidoglycan. To better understand the mechanisms of catalysis and substrate recognition by the housekeeping sortase produced by the important human pathogen Streptococcus pyogenes, the crystal structure of this protein has been solved and its transpeptidase activity established in vitro. The structure reveals a novel arrangement of key catalytic residues in the active site of a sortase, the first that is consistent with kinetic analysis. The structure also provides a complete description of residue positions surrounding the active site, overcoming the limitation of localized disorder in previous structures of sortase A-type proteins. Modification of the active site Cys through oxidation to its sulfenic acid form or by an alkylating reagent supports a role for a reactive thiol/thiolate in the catalytic mechanism. These new insights into sortase structure and function could have important consequences for inhibitor design.

adhesins or invasins and/or interfering with various arms of the host innate or specific immune defenses. The vast majority of these virulence proteins are retained at the bacterial surface after secretion by a mechanism that involves the covalent linkage of target proteins to the peptidoglycan layer of the cell wall. This linkage is catalyzed by membrane-associated transpeptidases called sortases (1,2). Proteins destined for cell-surface attachment contain a sorting signal recognized by these enzymes. As this mechanism is unique to Gram-positive pathogens, inhibiting the reaction is an attractive target for the development of novel antibacterials (3,4). The sortase-mediated transpeptidation reaction is also being increasingly used in a variety of biotechnology applications (5)(6)(7)(8).
The sorting signal that targets proteins for cell surface attachment is located at the C terminus of substrates and comprises a pentapeptide motif, typically LPXTG (where X is any amino acid), followed by a hydrophobic region and a tail of positively charged residues that locates the substrate to the cell surfacefollowingsecretion (2,9).Inonecurrentmodelofsortasedependent transpeptidation, the LPXTG motif is specifically recognized by the enzyme (10), and the thiolate group of an essential active site Cys attacks the scissile Thr-Gly bond (1,11). Two additional absolutely conserved residues are located in the active site, a His and an Arg (12,13). The His may act as a general acid to protonate the tetrahedral intermediate, thus facilitating collapse of the transition state and formation of an acyl-enzyme intermediate. Meanwhile, the Arg is thought to either stabilize the short-lived oxyanion-transition state (14) or to be involved in substrate recognition via a hydrogen bond (15). This reverse-protonation mechanism of catalysis, involving a Cys thiolate nucleophile and a His imidazolium general acid (but not a pre-organized ion pair), remains somewhat controversial as the pK a values of the Cys and His in Staphylococcus aureus sortase A (Sa-SrtA, 2 the most extensively studied sortase enzyme) have been measured as ϳ9.4 and 6.2-7.0 (16,17), respectively. Also, in existing structures of Sa-SrtA (initially determined by NMR (18) and then x-ray crystallography (12)) the two residues are not positioned in an orientation that would support this chemistry. It has been suggested that active Sa-SrtA (containing the Cys-184 thiolate and His120 imidazolium forms) may only represent a minor percentage of total enzyme present at physiological pH (17), and this would negate detection of the charged state by either biochemical or structural techniques. The acyl-enzyme intermediate formed during the reaction (19) is resolved through nucleophilic attack by an amino group of a branched chain peptidoglycan precursor (17,20), incorporating the substrate protein into the cell wall. Recombinant Sa-SrtA catalyzes the in vitro transpeptidation of LPXTG-containing peptides to pentaglycine (17,21), the branched side chain of lipid II, the cell wall precursor to which sortase A substrates are attached in this organism (22,23).
In addition to providing the first molecular structure of a sortase enzyme (18), NMR has also been used to probe the dynamics of Sa-SrtA and determine the overall binding orientation of the sorting signal peptide to the protein (24,25). These studies revealed that Ca 2ϩ acts as an allosteric activator by stabilizing the structure of the ␤ 6 /␤ 7 loop in Sa-SrtA in a closed, substrate binding state and that the binding site for the LPXTGsorting signal comprises residues from ␤ 4 and␤ 7 and the ␤ 3 /␤ 4 and ␤ 6 /␤ 7 loops. A crystal structure of Sa-SrtA bound to the same peptide is also consistent with the latter (12). Detailed mechanistic studies of SrtA-type enzymes have continued to focus on Sa-SrtA in attempts to rationalize the extensive kinetic and structural data (9 -11, 14 -17, 19, 21, 26 -28).
Although a single sortase (known as a housekeeping sortase and often, although not always, referred to as sortase A (SrtA)) is responsible for attachment of the majority of cell surface proteins containing a sorting signal, searches of Gram-positive genome sequences frequently reveal the presence of one or more additional sortases (29). Although genes encoding SrtA and its substrate proteins do not appear to be closely linked in bacterial genomes, genes encoding functionally related but distinct sortases tend to be clustered with their substrate proteins and can have divergent sorting motifs (30 -32).
The serotype M1, SF370 strain of the important human pathogen Streptococcus pyogenes expresses two distinct sortases, SrtA (Spy1154) and SrtC2 (Spy0129). S. pyogenes SrtA (Sp-SrtA) is the general housekeeping sortase and is responsible for anchoring all proteins containing the LPXTG sorting signal to the cell wall, including various key virulence proteins, such as the M-protein, protein G-related ␣ z M-binding protein (GRAB), and ScpA (31,33). The second sortase (SrtC2) catalyzes the polymerization of cell surface pili that mediate specific S. pyogenes adhesion to clinically relevant human tonsil and skin cells (34,35). Recent studies in our laboratory 3 have shown that Sp-SrtA is responsible for anchoring these pili in the cell wall. This appears to be equivalent to reports that in Corynebacterium diphtheriae two sortases mediate pilus assembly (36) and suggests the role of sortases in the mechanism of pili assembly may be conserved in all Gram-positive bacteria.
To further understand structure/function relationships and the in vivo role of these enzymes, the three-dimensional structure of sortase A from the serotype M1, SF370 strain of S. pyogenes has been determined, and its transpeptidase activity with functionally relevant substrates has been demonstrated. The structure has been determined in two states, one in which the active site Cys is reduced and one where it is oxidized to its sulfenic acid derivative. A long cleft in the structure, with the catalytic Cys at its center, forms the binding site for the sorting signal and, likely, the branched side chain cell-wall precursor that acts to resolve the acyl-enzyme intermediate. Furthermore, the conformation of the loops surrounding the active site are all well defined in the structure (including the ␤ 6 /␤ 7 loop) providing for the first time a complete atomic description of the active site and substrate binding regions in a SrtA enzyme. These results form a basis for further studies to probe the catalytic mechanism and substrate specificity of this important family of enzymes.

EXPERIMENTAL PROCEDURES
Gene Cloning-The DNA sequence encoding SrtA residues Val-82-Thr-249 (SrtA ⌬81 ) was amplified from serotype M1 S. pyogenes strain SF370 genomic DNA using the primers 5Ј-CTTAGGATCCGTCTTGCAAGCACAAATGG-3Ј (forward) and 5Ј-ATGTTCTCGAGCTAGGTAGATACTTGGT-TATAAGA-3Ј (reverse). This construct removes the predicted secretion signal sequence (residues Met-1-Ala-43), other predicted disordered regions at the N terminus, and also introduces appropriate restriction sites for subsequent cloning (BamH1 and Xho1 sites, underlined). The PCR product was digested with the appropriate enzymes and cloned directly into a modified version of pET28a (cut with the same enzymes, vector originally from Novagen) in which the nucleotides between the Nde1 and BamH1 restriction sites are removed. The resulting construct was verified by DNA sequencing and transformed into Escherichia coli BL21 (DE3) cells for protein expression.
Protein Expression and Purification-Cultures of BL21 (DE3) harboring the Sp-SrtA ⌬81 ⅐pET28a vector were grown in 1 liter of Luria-Bertani (LB) media supplemented with kanamycin (50 g/ml) at 37°C with shaking to an A 600 between 0.4 and 0.6. Protein expression was induced by the addition of isopropyl 1-thio-␤-D-galactopyranoside to a final concentration of 1 mM. Cells were grown for a further 3-4 h before harvesting by centrifugation. Cell pellets were resuspended in 50 mM HEPES, 150 mM NaCl, 10 mM imidazole, pH 7.5, supplemented with 5 mM 4-(2-aminoethyl)-benzenesulfonyl fluoride hydrochloride (AEBSF) and lysed by sonication. The lysate was centrifuged, and the supernatant was applied to two pre-equilibrated 5-ml Hi-Trap chelating columns (loaded with nickel, GE Healthcare) joined sequentially. Sp-SrtA ⌬81 was eluted with an imidazole gradient (10 -500 mM) over 15 column volumes. Fractions containing Sp-SrtA ⌬81 (as identified by SDS-PAGE) were pooled and concentrated. The protein was then injected onto a Hi-Load 16/60 Superdex 75 column pre-equilibrated in 20 mM Tris-HCl, 150 mM NaCl, pH 7.5. Eluted fractions containing Sp-SrtA ⌬81 were pooled and subjected to a repeat metal chelation purification step (as above), which was necessary to remove traces of contaminating protein. Pure Sp-SrtA ⌬81 was then exchanged into 20 mM Tris, 150 mM NaCl, pH 7.5, and concentrated to 20 mg/ml by ultrafiltration before further study.
Transpeptidation Activity-Sp-SrtA ⌬81 activity was determined using an HPLC-based assay previously developed for Sa-SrtA (14,21). For initial activity determination, Sp-SrtA ⌬81 was incubated with 1.0 mM Abz-LPETGG-Dap(Dnp)-NH 2 and 2.0 mM NH 2 -Ala 2 -OH in assay buffer (300 mM Tris, 150 mM NaCl, 5 mM CaCl 2 , pH 7.5). Dialanine is the simplest form of cross-bridge between subunits in the peptidoglycan layer of S. pyogenes, equivalent to pentaglycine in S. aureus (37). Reactions were performed at 37°C (total volume 100 l) and quenched by removal into a half-volume of 1.2 M HCl (final HCl concentration ϭ 0.4 M). Samples were purified by reverse-phase HPLC using an analytical C18 column (4.6 ϫ 250 mm, 3 m, Vydac, Inc.) mounted on an Agilent 1200 HPLC system equipped with a fraction collector. 64 l of the quenched reaction was injected onto the column, and the reaction products were separated using a linear gradient from 0 to 45% acetonitrile, 0.1% trifluoroacetic acid over 25 min at a flow rate of 1 ml/min. Dnp-and Abz-containing peaks were detected using UV absorption (355 nm) and fluorescence (excitation ϭ 318 nm, emission ϭ 420 nm), respectively. Putative reaction products eluted at retention times of ϳ14.3 min (detected by UV at 355 nm), ϳ17.3 min (detected by fluorescence), ϳ18.3 min (detected by fluorescence), and ϳ24.3 min (detected by UV and fluorescence) and were collected using a fraction collector. The measured area under the both the substrate (Abz-LPETGG-Dap(Dnp)-NH 2 ) and product (GG-Dap(Dnp)-NH 2 ) peaks was used to determine the concentration of product. Linearity of activity was established out to 2.5 h. To determine steady-state kinetic parameters, assay length and enzyme concentration were chosen to yield product conversions of 1-10%. 5 M Sp-SrtA ⌬81 was incubated with 2 mM NH 2 -Ala 2 -OH, whereas the concentration of Abz-LPETGG-Dap(Dnp)-NH 2 was varied from 62.5 M to 8 mM. Reactions were initiated by the addition of enzyme. After incubation for 150 min at 37°C, the reactions were quenched and analyzed by HPLC as described above. All data points were collected in triplicate, and the overall assay was run in duplicate. To determine the effects of Ca 2ϩ on the reaction, the assay was repeated in assay buffer as above but with Ca 2ϩ omitted. Raw data were fitted to a modified version of the Michaelis-Menten equation incorporating substrate inhibition (Equation 1) using GraFit Version 4.03 (Erithacus Software), where V max is the apparent maximal enzymatic velocity, K m is the apparent Michaelis constant, and K i is the apparent inhibitor dissociation constant for unproductive substrate binding. The data fit this equation with a 2 value of 2.89 ϫ 10 Ϫ7 (5 mM Ca 2ϩ ) and 1.27 ϫ 10 Ϫ6 (no Ca 2ϩ ).
The identity of the transpeptidation product was determined by electrospray mass ionization-mass spectrometry using an Agilent 1100 MSD Trap SL mass spectrometer in positive ion mode. The nebulizer pressure was set at 20.0 p.s.i., dry gas at 7 liters/min, and dry temperature set to 325°C. The mass range was set between 120 -1200 m/z, with a target mass of 750. Samples were infused at 20 ml/h. Data were collected for 2 min and then averaged to provide the total signal.
Alkylation of Cys-208-The haloalkylating reagent 6-bromoacetyl-2-dimethylaminonaphthalene (badan) reacts with free thiols (and preferentially with ionized thiolate anions) forming a stable covalent thioether bond (38). The reaction generates a significant increase in fluorescence intensity on binding. To determine whether Sp-SrtA ⌬81 contains a reactive thiol group, the protein was incubated with badan for 1 h at pH 7.0. The change in fluorescence intensity at 525 nm (excitation at 380 nm) over this time was monitored. To map the location of any modifications, the intact mass of the resulting sample was determined by matrix-assisted laser desorption/ionization time of flight (MALDI-TOF) mass spectrometry followed by peptide mass-fingerprinting, cleaving the protein with trypsin (Proteomics facility, John Innes Centre).
Crystallization-Conditions supporting the growth of Sp-SrtA ⌬81 crystals were initially identified using the hanging drop method of vapor diffusion at 20°C and commercially available crystallization screens. 1 l of protein solution was mixed with 1 l of precipitant solution. Diffraction quality Sp-SrtA ⌬81 crystals were grown using 100 mM Tris, pH 6 -7, 20 -30% (w/v) polyethylene glycol 8000, and 200 mM sodium acetate. Crystals grew to their maximum dimensions in 4 -5 days.
Data Collection and Structure Determination-For data collection, crystals were cryo-preserved by transferring into 100 mM Tris (pH 6 -7), 35% (w/v) polyethylene glycol 8000, and 200 mM sodium acetate before plunging into liquid nitrogen. Diffraction data were collected using either a Home-Lab system (R-AXIS IV ϩϩ detector with X-rays from a Rigaku Micro-max007 generator focused with Osmic optics) or at the Diamond Light Source, UK (station I03 equipped with an ADSC Q315 3 ϫ 3 CCD detector) as shown in Table 1. Crystals were maintained at cryogenic temperatures during data collection. Data were processed with MOSFLM (39) and scaled with SCALA (Ref. 40, as implemented within the CCP4 suite (41)). 5% of the data were set aside for the calculation of R free . The initial structure of S. pyogenes sortase A was determined in space group P2 1 2 1 2 by molecular replacement using PHASER (Ref. 42, as implemented in the CCP4 suite), with the Sp-SrtA ⌬81 ⅐Cys-ox dataset (the first dataset collected, see Table  1); this structure contains one molecule per asymmetric unit. The search model comprised a monomer of the crystal structure of Sa-SrtA (PDB entry 1T2P (12)), pruned to C␥ atoms in non-conserved regions using CHAINSAW (43). The structure was manually rebuilt and refined with iterative cycles of COOT (44) and REFMAC5 (45) to produce a final model (subsequently called Sp-SrtA ⌬81 ⅐Cys-ox). In this structure Cys-208 was found to be oxidized to a sulfenic acid form. To produce unmodified (reduced) Sp-SrtA, crystals were soaked in cryoprotectant solution supplemented with 10 mM dithiothreitol overnight before freezing in liquid nitrogen. The resulting dataset from this crystal (Sp-SrtA ⌬81 ⅐Cys-red, see Table 1) was also in P2 1 2 1 2. Finally, a single crystal was further soaked in fresh reducing cryoprotectant solution containing 10 mM dithiothreitol and 100 mM Leu-Pro-Ser-Thr-Gly peptide (Peptide Specialty Laboratories GmbH) for 24 h. The resulting dataset revealed a different crystal form (called Sp-SrtA ⌬81, in spacegroup P2 1 ; see Table 1). The Sp-SrtA ⌬81 ⅐Cys-ox structure was used to solve the Sp-SrtA ⌬81 ⅐Cys-red and Sp-SrtA ⌬81 datasets (again using PHASER), and repeated cycles of COOT and REFMAC5 generated final models (subsequently referred to as Sp-SrtA ⌬81 ⅐Cys-red and Sp-SrtA ⌬81 ). During the final stages of refinement, anisotropic B-factors were refined for Sp-SrtA ⌬81 ⅐Cys-red and Sp-SrtA ⌬81 ; an isotropic B-factor model was used for Sp-SrtA ⌬81 ⅐Cys-ox. The final model of Sp-SrtA ⌬81 ⅐Cys-ox comprises residues 82-211 and 214 -249 of the native sequence and 159 water molecules (residues 73-81 in the structure are derived from the vector used for expression); Sp-SrtA ⌬81 ⅐Cys-red comprises residues 82-210 and 214 -249 of the native sequence and 108 water molecules (residues 73 -81 in the structure are derived from the vector used for expression); Sp-SrtA ⌬81 comprises residues 89 -249 in one molecule, 87-249 in the second molecule of the asymmetric unit, 1 molecule of HEPES, and 351 water molecules. Final refinement statistics are given in Table 1. For analysis, MOL-PROBITY (46) and LSQMAN (47), respectively, were used to generate Ramachandran plots and superimposed structures from which root-mean-square deviations (r.m.s.d.) based on C␣ atoms were determined. Protein structure figures have been prepared with PYMOL (48). The coordinates and structure factors for Sp-SrtA ⌬81 , Sp-SrtA ⌬81 ⅐Cys-ox, and Sp-SrtA ⌬81 ⅐Cysred have been deposited with the Protein Data Bank with accession codes 3FN5, 3FN6, and 3FN7, respectively.
Modeling of the Substrate Complex-A model of the Sp-SrtA ⌬81 ⅐LPSTG complex was generated by docking the peptide into the refined crystal structure using (a) recent mutagenesis data from Sa-SrtA (10, 15), (b) known features of sequence conservation within the LPXTG motif, and (c) the requirement for the scissile peptide bond to be positioned appropriately in the active site as guides. Hydrogen atoms were added consistent with pH 7, and the complex was soaked in a 10-Å layer of water. The system energy was minimized using DISCOVER Version 2.98 (Accelrys) for a total of 3000 steps with a harmonic tethering potential applied between the protein backbone atoms and their initial coordinates (the tethering potential was scaled between 100 and 0.5 kcal per Å 2 during the minimization). Non-covalent interactions were truncated using a switching function at 15 Å. Interactive model manipulation used InsightII (2005). The coordinates of the Sp-SrtA ⌬81 ⅐LPSTG complex are available as supplemental material.

RESULTS
Protein Design and Production-The S. aureus sortase A deletion mutant Sa-SrtA ⌬59 , a construct lacking the N-terminal 59 residues, catalyzes the cleavage of LPXTG-containing peptides and subsequent transpeptidation to triglycine in vitro (18). Sequence alignments (not shown) revealed the equivalent catalytic domain in Sp-SrtA comprises residues 82-249. Expressed and purified Sp-SrtA ⌬81 was examined by liquid chromatography/mass spectrometry using a Thermo LTQ-FT mass spectrometer (PINNACLE Lab, Newcastle University) revealing a molecular mass of 20,325.3 Da compared with the theoretical mass of 20,326.0 Da, which assumes the N-terminal methionine is lost (in a small proportion of the sample the N terminus displayed an ␣-N-6-phosphogluconyl modification (an increase in mass of 178 Da), as has been observed for other proteins with equivalent His tags (49)).
Sp-SrtA ⌬81 Is an Active Transpeptidase-The catalytic activity of Sp-SrtA ⌬81 was tested using an assay established previously for Sa-SrtA but using dialanine as the transpeptidation substrate rather than tri-or pentaglycine, reflecting the different cell wall cross-bridges in the different parent organisms. A plot of initial rates for both the assays with and without Ca 2ϩ against substrate concentration are shown in Fig. 1 (with the appropriate fit), and the derived steady-state kinetic parameters are summarized in Table 2 (values for Sa-SrtA ⌬24 are also given for comparison (15)). As Sp-SrtA ⌬81 is not activated by Ca 2ϩ , the results of the assays in the absence of this ion are further described here. Relative to Sa-SrtA ⌬24 , Sp-SrtA ⌬81 has an 80-fold reduction in apparent k cat (k cat app ) and a 10-fold reduction in apparent K m (K m app ), leading to an apparent k cat /K m (k cat / K m app ) that is 8-fold lower. Overall apparent poor catalytic effi-  ϭ 65.95, b ϭ 69.80, c ϭ 39.87 a ϭ 66.03, b ϭ 69.89, c ϭ 40.15 a ϭ 39.28, b ϭ 59 ciency is a general property of sortases in vitro and is attributable to the substrates used being poor mimics of true in vivo substrates (the sorting motifs are part of a larger protein, and the transpeptidation substrate is part of a cell wall precursor) and is also a reflection of the reverse protonation catalytic mechanism, where a small fraction of the enzyme is in the catalytically competent protonation state. Interestingly, Sp-SrtA ⌬81 is subject to substrate inhibition at peptide concentrations Ͼ4 mM, a feature not observed for Sa-SrtA; this is not expected to be physiologically relevant. Formation of the correct transpeptidation product was determined by analysis of the reaction products by electrospray mass ionization-mass spectrometry. The predominant peak in the HPLC fractions corresponding to the product was 720.3 Da, the ϩ1 charge state of Abz-LPET-AA-OH (theoretical molecular mass of 719.3 Da).
Sp-SrtA ⌬81 Can Be Alkylated on Cys-208-To investigate whether Sp-SrtA ⌬81 contains a reactive thiol group, the protein was incubated with badan. An increase in fluorescence during incubation (at 525 nm) was suggestive of an alkylation event. MALDI-TOF mass spectrometry was used to compare the intact mass of Sp-SrtA ⌬81 pre-and postincubation with badan and revealed an increase in weight of 214.1 Da, supporting modification of the protein at a single site (the expected increase in molecular mass for each covalently attached badan is 211 Da). To characterize the site of protein modification, preand post-badan-incubated Sp-SrtA ⌬81 were digested with tryp-sin, and the resulting peptides were analyzed by mass spectrometry. A peptide comprising residues EVTLVTCT-DIEATER changes by a mass of 211 Da post-incubation. Under these experimental conditions the only badan-reactive group in this peptide is the side chain of Cys-208.
Overall Structure of Sp-SrtA ⌬81 -Except where noted, the structure of Sp-SrtA ⌬81 (P2 1 form) is used as the reference model for further discussion as this likely represents the active structure in solution and has been determined to the highest resolution. The two molecules in the asymmetric unit of Sp-SrtA ⌬81 overlay with an r.m.s.d. of 0.2 Å (all 161 C ␣ atoms from residues 89 -249) and can, therefore, be considered essentially identical. An example of the final electron density map, including the active site residues, is shown in Fig. 2a.
The structure of Sp-SrtA ⌬81 adopts the 8-stranded ␤-barrelfold unique to sortases. The overall structure is very similar to that of Sa-SrtA ⌬59 (12, 18) despite only sharing 24% sequence identity in the core catalytic domain (not counting the 19 residue C-terminal extension in Sp-SrtA); see Fig. 2b. The average r.m.s.d. for the overlay of 107 equivalent C ␣ atoms is 1.54 Å (Sp-SrtA ⌬81 chain A on the 3 molecules of the asymmetric unit of Sa-SrtA ⌬59 (PDB code 1T2P)). Although the core of the ␤-barrel fold is very well conserved, there are significant differences in the connecting loop regions and the N/C termini. The largest of these are located to the active site region and include the loop connecting ␤ 6 /␤ 7 and ␤ 7 /␤ 8 (see Fig. 2b); there are also shifts in the position of the ␤ 2 /␤ 3 loop, ␤ 3 , and the ␤ 3 /␤ 4 loop, which are close to the active center. Based on structural alignment, Sp-SrtA ⌬81 has an additional 23 residues at the C terminus compared with Sa-SrtA ⌬59 . These residues form an extension to ␤ 8 , a turn, and a short ␣-helix which lies across the external face of the ␤-sheet formed by ␤ 1 /␤ 2 ␤ 5 /␤ 6 before proceeding toward the active site end of the molecule where they contribute to the structure of the extended active site cleft (see below and Fig. 2, b and c) along with the loop immediately after ␤ 4 (which contains a three-residue insertion compared with Sa-SrtA ⌬59 ).
The core structure of Sp-SrtA ⌬81 is also similar to SrtB from S. aureus and Bacillus anthracis (50). S. aureus SrtB overlays on Sp-SrtA ⌬81 with an r.m.s.d. of 1.57 Å (103 equivalent C ␣ atoms). As with the comparison to Sa-SrtA ⌬59 , the structure of many connecting loops between the core ␤-strands of Sp-SrtA ⌬81 and S. aureus SrtB differ, and in this case many of these will be due to the different substrates that these proteins recognize.
The structure of S. pyogenes sortase A has also been determined in two states in a P2 1 2 1 2 crystal form. In the first of these, Sp-SrtA ⌬81 ⅐Cys-ox, additional electron density extending from the sulfur atom of the active site Cys-208 is consistent with the sulfenic acid form of this residue. After overnight soaking of a crystal in cryo-protectant solution supplemented with 10 mM dithiothreitol (resulting in the structure Sp-SrtA ⌬81 ⅐Cys-red) this extended density is no longer observed. Overall, the three crystal structures of the protein are virtually identical (r.m.s.d. Ͻ 0.41 Å for all overlays using C ␣ atoms of residues 92-207 and 216 -247). However, some re-arrangements in the active site region are observed, as discussed below.
Active Site Architecture in Sp-SrtA ⌬81 -Previous studies with Sa-SrtA identify Cys-208, His-142, and Arg-216 as the key cat-  Table 2. alytic residues in Sp-SrtA. In the structure of Sp-SrtA ⌬81 these three residues are clustered at the center of a long cleft (ϳ32 Å, see Fig. 2c), which could readily accept a protein substrate harboring an appropriate pentapeptide sorting signal. The orientation of these residues is consistent with the model of reverse protonation suggested from biochemical studies of Sa-SrtA.
Cys-208 appears well positioned to attack an incoming scissile Thr-Gly bond, and Arg-216 may be involved in orienting the Structural Changes on Cys-208 Oxidation in Sp-SrtA ⌬81 -Co-incident with modification of Cys-208 to its sulfenic acid form (as observed in Sp-SrtA ⌬81 ⅐Cys-ox) the ␤ 4 /␤ 5 , ␤ 6 /␤ 7 , and ␤ 7 /␤ 8 loops undergo significant re-arrangements compared with Sp-SrtA ⌬81 . Most significant are the changes to the ␤ 7 /␤ 8 loop (between residues Thr-207 and Ile-217), where a reorientation in the backbone dihedral angles between Thr-207 and Cys-208 results in the position of the sulfur atom of Cys-208 shifting by ϳ5.4 Å. The side chain of His-142 appears to rotate 145°around 2 (although the plane of the histidine in the density is difficult to definitively assign), and the position of Arg-216 also changes (see FIGURE 2. a, stereoview showing the arrangement of active site residues (Cys-208, His-142, and Arg-216) in the structure of Sp-SrtA ⌬81 . In mesh representation is the final -weighted 2 F obs Ϫ F calc ⅐F calc map contoured at 1.2. b, stereoview showing the overlay of Sp-SrtA ⌬81 (in gray) with Sa-SrtA ⌬59 (in blue). Secondary structure motifs and the positions of important active site residues are labeled (carbon atoms of active site residues are colored dark gray and yellow for Sp-SrtA ⌬81 and Sa-SrtA ⌬59 , respectively). The ␤ 6 /␤ 7 and ␤ 7 /␤ 8 loops are colored red/green in Sp-SrtA ⌬81 and cyan/magenta in Sa-SrtA ⌬59 respectively. c, surface representation of Sp-SrtA ⌬81 showing the active site cleft (in red). Active site residues are labeled: C, Cys-208; H, His-142; R, Arg-216. The region to the left of the highlighted residues represents the LPXTG binding site. To the right is the region predicted to bind the extended protein substrate and/or the branched side chain cell wall precursor. Fig. 3a). This oxidized Cys can be reduced in the crystal as shown by the structure of Sp-SrtA ⌬81 ⅐Cys-red. Residual electron density in the region of the ␤ 7 /␤ 8 loop in the Sp-SrtA ⌬81 ⅐Cys-red structure is best accounted for by building an alternate conformation for residues Thr-207 to Glu-215. One of these conformations is that found in Sp-SrtA ⌬81 ; the other is that found in Sp-SrtA ⌬81 ⅐Cys-ox. These two structures show there is some structural plasticity in the active site of Sp-SrtA, although the relevance of the conformational change between Sp-SrtA ⌬81 and Sp-SrtA ⌬81 ⅐Cys-ox is questionable as the enzyme with Cys-208 oxidized will be inactive.
Molecular Modeling of Substrate Binding to Sp-SrtA ⌬81 -Despite being soaked in high concentrations of LPSTG peptide, no electron density for this peptide was observed in the structure of Sp-SrtA ⌬81 . A similar result was obtained in studies with Sa-SrtA ⌬59 (12). Although attempts to crystallize a C208A mutant of Sp-SrtA ⌬81 proceed, a molecular model of the Sp-SrtA ⌬81 ⅐LPSTG complex has been constructed and is discussed below.

DISCUSSION
Sortases are transpeptidases that reside on the extracellular side of the cell membrane and catalyze the covalent linkage of proteins containing a specific signal to the cell surface. Sortase A, the product of open-reading frame Spy1154 in serotype M1 S. pyogenes strain SF370 (Sp-SrtA), catalyzes the linkage of various proteins and protein complexes (such as pili) to the cell wall in this organism. Sortase substrate proteins are frequently associated with pathogenesis, and therefore, sortases are required for virulence of Gram-positive pathogens (31,32,(51)(52)(53)(54)(55)(56). As they are attractive targets for the design of novel antibacterials, a thorough understanding of their structure and mechanism of action is essential.
Comparison of the observed kinetic parameters for Sa-SrtA (with Ca 2ϩ ) and Sp-SrtA ⌬81 (without Ca 2ϩ ) reveals a 80-fold decrease in the k cat app for Sp-SrtA ⌬81 . Although this was unexpected, the difference in rates may be attributable to the assay conditions, which have been optimized for Sa-SrtA (even taking into account the differences in Ca 2ϩ concentration). Future studies will identify whether changes in assay conditions result in an improved rate of product formation by Sp-SrtA ⌬81 and a correspondingly higher k cat app . In contrast, K m app of Sp-SrtA ⌬81 for the synthetic peptide Abz-LPETGG-Dap-(Dnp)-NH 2 was much improved at 830 M compared with 8.8 mM for Sa-SrtA. For Sa-SrtA (which utilizes a ping-pong kinetic mechanism) it has previously been demonstrated that in the presence of an NH 2 -Gly 5 -OH nucleophile, acylation of the LPXTG substrate is the rate-limiting step in the overall reaction (17,19). Although it has not yet been shown that the same holds true for Sp-SrtA, it is reasonable to assume that it operates by similar kinetic and catalytic mechanisms.
Ca 2ϩ stimulates the activity of Sa-SrtA ϳ8-fold by binding to a pocket on the protein surface and modulating the structure and dynamics of the ␤ 6 /␤ 7 loop (18,25). The Ca 2ϩ is bound predominantly by two Glu residues and an Asp (on the ␤ 3 /␤ 4 loop) and a Glu on the ␤ 6 /␤ 7 loop. In contrast to Sa-SrtA, we have shown in this work that the activity of Sp-SrtA ⌬81 is not promoted by Ca 2ϩ . If anything, the presence of this ion actually mildly inhibits turnover (ϳ2-fold reduction in k cat app ). Comparison of the structures of Sp-SrtA ⌬81 and Sa-SrtA in the region of the Sa-SrtA⅐Ca 2ϩ binding pocket reveals they adopt very differ- (light gray, red, green) and Sp-SrtA ⌬81 ⅐Cys-ox (blue, cyan, magenta, and with carbon atoms of the residues in yellow)as described in the text. b, stereoview depicting the molecular model of the LPSTG peptide (carbon atoms in cyan, sequence from left to right) bound to Sp-SrtA ⌬81 . C ␣ trace of Sp-SrtA ⌬81 is shown in light gray with carbon atoms of highlighted residues, as described under "Results" and under "Discussion," in the text, in dark gray. ent conformations. Although this complicates direct comparison on a residue-per-residue basis, it is worth noting that at least two of the amino acids involved in Ca 2ϩ binding to Sa-SrtA do not have a structural equivalent in Sp-SrtA ⌬81 (E171T, E105K, Sa-SrtA numbering); in fact, the side chain of Lys-126 in Sp-SrtA ⌬81 (the structural equivalent of Glu-105) occupies the Ca 2ϩ position observed in Sa-SrtA.
The most complete model describing the catalytic mechanism of sortases comes from studies of Sa-SrtA. Early investigations identified a key role for Cys-184 (1,11,57,58). Structural studies on sortases which have to date been limited to Sa-SrtA, S. aureus sortase B, and B. anthracis sortase B suggest that formation of the acyl-enzyme intermediate may include the Cys as part of a Cys-His-Asn (18) or Cys-His-Asp (12,50) catalytic triad. However, in these structures the His is too far removed from the active site to enable chemistry without significant reorientation of side chains and movement of the main chain. The latter has been considered unlikely as the putative catalytic residues are anchored onto ␤-strands within a rigid ␤-sheet (12). This led to the proposal of a Cys-Arg dyad (with the Arg acting as a general base, modulating the ionization state of the Cys (12,13)) and relegated the role of the His to maintaining the structure of the active site but not participating in catalysis. This role is not consistent with kinetic and mutagenesis studies, which predict a critical role in the catalytic mechanism for a His residue (to protonate the leaving group) and suggest that the Arg acts not as a general base but either stabilizes the transient tetrahedral oxyanion intermediate (14) or is important for substrate binding and orientation (15).
In contrast to these apparent discrepancies, the structure of Sp-SrtA ⌬81 reveals a novel arrangement of these critical sortase active site residues, which is consistent with the Sa-SrtA kinetic and mutagenesis studies. In Sp-SrtA ⌬81 the Cys-His-Arg side chains adopt a conformation with Arg-216 on one side of the Cys-208 side chain and His-142 on the other; the N ␦1 atom of His-142 is oriented toward Cys-208. These residues appear to occupy ideal positions to enable their roles in the chemistry of the reaction (Arg-216 to orient the substrate or stabilize the oxyanion intermediate and His-142 to protonate the leaving group on acyl-enzyme formation). The solvent structure in the active site also suggests that the protein can support formation of a thiolate nucleophile. In this conformation the protein appears receptive to substrate LPXTG-containing proteins, and the catalytic residues are appropriately positioned to enable chemistry. If structure-based drug design is to have a contribution to the development of sortase inhibitors, it is essential that detailed, accurate templates (i.e. relevant protein structures) are available.
Sp-SrtA ⌬81 contains one reactive thiol group, the side chain of Cys-208, as determined by modification with badan. Comparison with an equivalent titration of Sa-SrtA suggests a similar pK a likely exists for the active site Cys in the resting state of both enzymes (data not shown). Determining whether Sp-SrtA does indeed operate by the same reverse protonation mechanism along with a full evaluation of the function and charge state of the active site residues requires monitoring during catalytic turnover and is the subject of ongoing work. Nonetheless, all indications are that Sp-SrtA operates by a mechanism similar to Sa-SrtA.
The ␤ 6 /␤ 7 loop of sortases is important in determining substrate selectivity, and therefore, this region is important for enzyme activity (10,15). In the structures of Sa-SrtA ⌬59 this loop is disordered (high B-factors or not modeled in crystal structures (12) or mobile in NMR structures (18,24,25)). As previously mentioned, the dynamics of this loop in Sa-SrtA are affected by Ca 2ϩ . Mutagenesis and NMR studies of Sa-SrtA have suggested Val-168 and Leu-169 (present on the ␤ 6 /␤ 7 loop) are important for binding the Leu-Pro region of the LPXTG peptide (15,25). These residues do not interact with the peptide in the published crystal structure (12). In the Sp-SrtA ⌬81 structure reported here, the ␤ 6 /␤ 7 loop is well defined in the electron density, and the B-factors are comparable with the core regions of the structure. The structure of Sp-SrtA ⌬81 provides the first housekeeping sortase structure with a defined position for the amino acids lining the active site and, therefore, a sound base for investigating substrate interaction.
In the absence of an Sp-SrtA ⌬81 -peptide complex crystal structure, a theoretical model has been constructed (as described under "Experimental Procedures") in an attempt to understand how the sorting signal might bind to the enzyme (see Fig. 3b). Although this model is only a prediction of a possible structure, a number of interesting features are apparent. In the model, hydrophobic residues located to the ␤ 6 /␤ 7 loop , along with residues Met-125, Val-186, Val-206, Ile-218, and the side chain of Arg-216 interact with the Leu-Pro moiety of the LPXTG motif, a region known to be critical for binding specificity to Sa-SrtA. The Thr-Gly moiety is oriented such that the scissile peptide bond faces the side chains of Cys-208 and His-142. In the structure of the Sa-SrtA ⌬59 (C184A)⅐LPETG complex (12) the carbon atom of the scissile bond is ϳ11 Å away from the N ␦1 atom of His-120 (the His-142 equivalent), and access to the substrate is blocked by a cluster of hydrophobic residues, which is not consistent with the mechanism for catalysis presented here. Analysis of the submitted coordinates and original experimental data (both are available from the PDB entry 1T2W) reveals very high B-factors for the peptide in the Sa-SrtA ⌬59 (C184A)⅐LPETG complex and very little electron density to account for the position described, although a different protocol for map generation has been followed (however, maps calculated by the Electron Density Server (59)) also confirm this observation). The molecular model of the Sp-SrtA ⌬81 ⅐LPXTG interaction will be a valuable tool for targeting residues for site-directed mutagenesis to further probe substrate specificity and catalysis.
Unexpectedly, in addition to the Sp-SrtA ⌬81 structure (where Cys-208 is reduced), the structure of the enzyme with Cys-208 oxidized to its sulfenic acid derivative has also been determined (Sp-SrtA ⌬81 ⅐Cys-ox). Although possibly no more than an artifact, this form may have physiological relevance. Sortases are cell surface virulence factors and as such are potentially targets for host immune systems. It is interesting to speculate that reactive oxygen species produced by cells of the innate immune system could inactivate sortase enzymes of invading bacteria by oxidation of the active site Cys, limiting their potential to establish an infection. Our laboratories are investigating this intriguing hypothesis. Furthermore, despite the structure of Sp-SrtA ⌬81 ⅐Cys-ox clearly being that of an inactive enzyme, it is notable that in this position Cys-208 adopts a similar orientation to Cys-184 in Sa-SrtA and suggests that modification of this residue can induce structural rearrangements in the active site, at least in Sp-SrtA. Realignment of active site residues has been observed upon acylation of Sa-SrtB with inhibitors (13,50). Such rearrangements could be important in forming a binding site for the second substrate, a region that may be located to the elongated cleft adjacent to the active site of Sp-SrtA ⌬81 (see Fig. 2c). The equivalent region in S. aureus SrtB and B. anthracis SrtB has been predicted to bind such molecules (50). To understand whether structural rearrangements are indeed relevant in sortase mechanism requires further structural studies, likely the presence of polypeptide and/or peptidoglycan substrates to trap functionally relevant states.

CONCLUSIONS
Sortase A from serotype M1 S. pyogenes strain SF370 catalyzes the transpeptidation of LPXTG-containing peptides to the cell-wall precursor mimic NH 2 -Ala 2 -OH in vitro, and activity is not stimulated by Ca 2ϩ . The structure of Sp-SrtA reveals a novel arrangement of active site residues in a sortase and represents the first structure for these enzymes where these residues are located in positions consistent with kinetic studies. The results support a role for a thiolate nucleophile in the sortase mechanism. Movement of a ␤-strand/loop region, which harbors the catalytic Cys residue (and may be a response to modification of this residue), suggests the sortase active site can accommodate rearrangements that may have physiological/ mechanistic relevance.