Sortase from Staphylococcus aureus Does Not Contain a Thiolate-Imidazolium Ion Pair in Its Active Site*

Many surface proteins are anchored to the cell wall by the action of sortase enzymes, a recently discovered family of cysteine transpeptidases. As the surface proteins of human pathogens are frequently required for virulence, the sortase-mediated anchoring reaction represents a potential target for new anti-infective agents. It has been suggested that the sortase from Staphylococcus aureus (SrtA), may use a similar catalytic strategy as the papain cysteine proteases, holding its Cys184 side chain in an active configuration through a thiolate-imidazolium ion interaction with residue His120. To investigate the mechanism of transpeptidation, we have synthesized a peptidyl-vinyl sulfone substrate mimic that irreversibly inhibits SrtA. Through the study of the pH dependence of SrtA inhibition and NMR, we have estimated the pKas of the active site thiol (Cys184) and imidazole (His120) to be ∼9.4 and 7.0, respectively. These measurements are inconsistent with the existence of a thiolate-imidazolium ion pair and suggest a general base catalysis mechanism during transpeptidation.

Gram-positive bacteria infect humans through an array of surface-associated proteins that promote bacterial adhesion, resistance to phagocytic killing, and host cell invasion. Many surface proteins are covalently anchored to the cell wall by the action of sortase enzymes, a family of novel transpeptidases (reviewed in Refs. [1][2][3][4][5]. The SrtA protein from Staphylococcus aureus is the most extensively characterized sortase enzyme (6) and anchors surface proteins that contain a C-terminal sorting signal consisting of a conserved LPXTG motif, a hydrophobic domain, and a tail of mostly positively charged residues (7,8). SrtA and related proteins may be excellent targets for new broad-spectrum anti-infective agents, because sortase-like enzymes and the LPXTG signal are universally conserved in Gram-positive bacteria (5,9), and sortase (Ϫ) strains of S. aureus (10 -12), Listeria monocytogenes (13,14), and Streptococcus gordonii (15) display defects in their virulence.
The structure of SrtA revealed a novel protein fold and localized the active site to a hydrophobic surface depression that contains two highly conserved and enzymatically important residues, His 120 and Cys 184 (SrtA numbering) (16 -18).
This catalytic dyad is reminiscent of the active sites of the papain cysteine proteases (Cys 25 -His 159 -Asn 175 , papain numbering) (19,20), suggesting that they are mechanistically related (21). In this model of SrtA function, the imidazole ring of His 120 promotes the formation of the Cys 184 thiolate, which then nucleophilically attacks the carbonyl carbon at the scissile Thr-Gly peptide bond in the LPXTG motif (22). After covalent linkage via a thioacyl bond to the threonine carbonyl group, the incoming amine of the cell wall precursor lipid II (23,24) may then be deprotonated by His 120 for attack on the covalent intermediate (16,18).
Previous work left unresolved how His 120 activates Cys 184 . In the papain cysteine proteases, the cysteine side chain is held in an active configuration through a thiolate-imidazolium ion interaction with the histidine (25). However, in the NMR structure of SrtA solved in the absence of its substrates, the side chains of Cys 184 and His 120 do not interact, arguing against the presence of an ion pair (16). To resolve this issue, and as a first step toward the design of a therapeutically useful anti-infective agent, we have synthesized a peptidyl-vinyl sulfone substrate mimic that inhibits SrtA. The pH dependence of SrtA inhibition and NMR studies preclude the presence of an ion pair in the active site, because His 120 and Cys 184 have pK a values of 7.0, and ϳ9.4, respectively.
Synthesis of Vinyl Sulfone Inhibitor-The vinyl sulfone inhibitor was synthesized using solution phase methodology (see Fig. 1). The Leu-Pro-Ala tripeptide was synthesized by standard amino acid coupling chemistry using 1-[3-(dimethylamino)propyl]-3-ethylcarbodiimide hydrochloride (EDCI) and 4-(dimethylamino)pyridine (DMAP), starting with the carbobenzyloxy-protected amino acid Cbz-Leu-OH and N-tertbutoxycarbonyl (N-t-Boc) methyl esters of alanine and proline. L-Threonine was fully protected as the N-t-Boc-threonine methyl ester with the alcohol protected as the t-butyldiphenylsilyl ether (2) in three steps. The ester was reduced with diisobutylaluminum hydride (DIBAL-H) to the aldehyde, which was immediately reacted without purification with diethyl phenylsulfonylmethylphosphonate (3) to give the desired vinyl sulfone functionality (4) in good yield. Removal of the Boc group with trifluoroacetic acid followed by coupling the amine with the Cbz-pro-* This work was supported by National Institutes of Health Grant AI/GM52217 (to R. T. C. and M. E. J.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
ʈ Both authors contributed equally to this work. ** To whom correspondence may be addressed. tected tripeptide gave the desired tetrapeptide. Removal of the t-butyldiphenylsilyl group with HF gave the vinyl sulfone inhibitor (1). The inhibitor was purified by silica gel chromatography and the structure confirmed by 1 H and 13 C NMR spectroscopy and FABϩ (fast atom bombardment) mass spectrometry.
Enzyme Assays-Fluorescent measurements of SrtA activity in the presence of the vinyl sulfone inhibitor were performed in 96-well microtiter plates as described previously (26). Inhibition reactions contained 5 M SrtA and vinyl sulfone inhibitor (100, 300, or 600 M) in buffer R (50 mM Tris, pH 3.5-10, 150 mM NaCl, 5 mM glycine, 5 mM CaCl 2 ). d-QALPETGEE-e was dissolved in dimethyl sulfoxide and added to the reaction at a final concentration of 25 M, for a total reaction volume of 200 l. The final concentration of Me 2 SO in the reaction was kept below 5%. The reactions were incubated for 12 h at 37°C with gentle mixing in a GENios multiwell fluorimeter (Tecan, Durham, NC) (360 nm excitation filter, 465 nm emission filter). Fluorescence emission was recorded at 10-min intervals and blanked against a reference solution containing SrtA in buffer R. All enzyme assays were performed in triplicate, resulting in ϳ450 measurements for each inhibitor concentration assayed. The inhibition parameters K i and k i were solved simultaneously for each inhibitor concentration using the equation, where [P] t is the concentration of product at time t, [E] is the total concentration of SrtA, and [S] and [I] are the total concentrations of the fluorogenic substrate and inhibitor, respectively (27). K m and k cat (the Michaelis and first-order rate constants for the uninhibited SrtA transpeptidation reaction, respectively) were determined independently for each pH value (data not shown). Curves were fit using the program Sigmaplot2000 (SPSS version 6.0).
HPLC Analysis of Inhibitor Modification of SrtA-Thirty microliters of SrtA (95-100 M) in buffer I (50 mM Tris-HCl, 150 mM NaCl, and 5 mM CaCl 2 ) were adjusted to the desired pH with 1 M HCl or 1 M NaOH as needed and were incubated with a 20-fold molar excess of the vinyl sulfone inhibitor for 20 h at 37°C with gentle agitation. The reaction was stopped by adding 500 l of buffer A (0.1% trifluoroacetic acid in water), and the products were separated by reverse phase HPLC using a C18 column (Waters, Milford, MA) with the application of a gradient of 25-50% buffer B (90% acetonitrile, 10% H 2 O, 0.1% trifluoroacetic acid). NMR Spectroscopy-NMR experiments were carried out at 308 K on a Bruker DRX500 spectrometer equipped with a triple resonance probe using 15 N-and 13 C-labeled SrtA ⌬N59 and C184A SrtA ⌬N59 proteins (0.5 mM) in buffer N (50 mM Tris-HCl (pH 6.2), 100 mM NaCl, 20 mM CaCl 2 , 3 mM dithiothreitol, and 7% D 2 O). The chemical shifts of wild-type SrtA ⌬N59 have been reported previously, and the resonances of the single histidine side chain (His 120 ) in the C184A SrtA ⌬N59 mutant were readily assigned by reference to these data. For the pK a measurements, a series of two-dimensional 1 H-13 C heteronuclear single quantum correlation spectra were recorded to monitor the pH dependence of the 1 H-␦2-13 C-␦2 and 1 H-⑀1- 13 C-⑀1 resonances of the His 120 side chain (spectra recorded at pH values of 4.5-10). Chemical shifts in the NMR spectra were plotted as a function of pH and fit to the following equation, where ␦ HA and ␦ A are the chemical shifts of the fully protonated and deprotonated forms of the ionizable group, and ␦ obs is the observed chemical shift (28,29) . The data were fit using SigmaPlot2000.

RESULTS AND DISCUSSION
To characterize the active site His 120 -Cys 184 dyad, we synthesized an irreversible inhibitor that consisted of the substrate recognition motif of SrtA (-Leu-Pro-X-Thr-Gly-, where X FIG. 1. A, molecular structure of the peptidyl-vinyl sulfone inhibitor Cbz-Leu-Pro-Ala-Thr-SO 2 Ph. B, strategy for synthesis of vinyl sulfone inhibitor. L-Threonine was fully protected as the N-t-Bocthreonine methyl ester, with the alcohol protected as the t-butyldiphenylsilyl ether (2), in three steps. The ester was reduced with diisobutylaluminum hydride (DIBAL-H) to the aldehyde, which was immediately reacted without purification with diethyl phenylsulfonylmethylphosphonate (3) to give the desired vinyl sulfone functionality (4). Removal of the Boc group with trifluoroacetic acid (TFA) followed by coupling the amine with the Cbz-protected tripeptide (Cbz-Leu-Pro-Ala) gave the desired tetrapeptide. Removal of the t-butyldiphenylsilyl group with HF gave the vinyl sulfone inhibitor (1).
is any amino acid) but replaced the scissile Thr-Gly amide bond with a vinyl sulfone group (C ϭ C-SO 2 Ph) (Fig. 1A). Alanine was placed at the X position to facilitate the synthesis, and the vinyl sulfone group was used because it has previously been shown to covalently modify the active site thiol in cysteine proteases (30 -34). This is relevant because mutagenesis studies have demonstrated the catalytic importance of SrtA residue Cys 184 (18), and several sulfhydryl-directed reagents block the activity of SrtA in vitro (17,18).
The efficacy of the vinyl sulfone compound was tested in vitro by determining how it altered the SrtA-catalyzed hydrolysis of an internally quenched fluorescent substrate analogue (d-Gln-Ala-Leu-Pro-Glu-Thr-Gly-Glu-Glu-e). Hydrolysis progress curves were generated by monitoring the increase in fluorescence that accompanies the cleavage of the substrate (18, 26).
The curves were consistent with the vinyl sulfone compound acting as an irreversible inhibitor of SrtA ( Fig. 2A) (35). Typical progress curves of the reaction in the presence of 100, 300, and 600 M inhibitor are shown in Fig. 2B. Fits of these data to Equation 1 by non-linear regression analysis determined the first-order rate constant of inactivation (k i ) and the dissociation constant of inhibitor binding (K i ). At pH 7, the K i of the inhibitor is 9 ϫ 10 Ϫ6 M, which is comparable with the measured K m of SrtA of 3 ϫ 10 Ϫ6 M for the fluorogenic substrate analogue (data not shown). The rate constant of inactivation (k i ) of the vinyl sulfone inhibitor is 4 ϫ 10 Ϫ4 min Ϫ1 . To date, only two other irreversible SrtA inhibitors have been characterized. Both contain the Cbz-Leu-Pro-Ala-Thr sorting signal mimic of the vinyl sulfone compound but utilize different reactive groups (diazomethane and chloromethane) (26). The first-order rate constants of the peptidyl-diazomethane (5.8 ϫ 10 Ϫ3 min Ϫ1 ) and -chloromethane (1.1 ϫ 10 Ϫ2 min Ϫ1 ) SrtA inhibitors are larger than the vinyl sulfone k i by ϳ10and 20-fold, respectively (26). This difference in reactivity is consistent with the higher electrophilicity of chloro-and diazomethane reactive groups and has been observed in inhibition studies of other cysteine proteases (36). For example, diazomethane and chloromethane inhibitors of human cathepsin L have second-order rate constants up to 100-fold higher than vinyl sulfone inhibitors (30,36). As the transpeptidation reaction of SrtA is extremely poor in vitro (k cat /K m ϭ 20.6 M⅐s Ϫ1 ) it probably does not reflect the efficiency of the enzyme in vivo, where it must complete the anchoring of surface proteins within the doubling time of the bacterium. The development of a quantitative in vivo assay for inhibition will be necessary to evaluate the efficacy of peptidyl inhibitors of sortase as anti-infective agents.
Because the vinyl sulfone warhead of the peptide inhibitor is expected to be most reactive toward cysteine thiolates, and Cys 184 is the only cysteine in SrtA, we measured the pH dependence of inactivation to determine the pK a of Cys 184 . The in vitro hydrolysis reaction was repeated at the three inhibitor concentrations over the pH range of 3.5-10, and the inhibition parameters K i and k i were determined by curve fitting. As shown in Fig. 2C, k i increases dramatically above pH 8, whereas the K i value is not significantly affected. To estimate the pK a of Cys 184 , a modified Henderson-Hasselbach equation (37), (where k min and k max are the minimum and maximum observed first-order inhibition constants) was modeled for the inhibitor data by fixing k min (4 ϫ 10 Ϫ4 min Ϫ1 at pH 6) and solving for k max over a range of pK a values. Plots of Equation 3 for pK a values of 9 (dotted line), 9.5 (dashed line), and 10 (solid line) are shown in Fig. 2C. The estimated the pK a of Cys 184 is ϳ9.4.
To confirm that the pH-dependent inhibition of SrtA results from the covalent modification of Cys 184 , we tracked the modification reaction by chromatography. The SrtA protein was incubated with a 20-fold molar excess of inhibitor for 20 h, and the reaction products were separated by HPLC. Fig. 3, A-C shows the results of incubating SrtA with the inhibitor at pH 9, 7, and 6, respectively. Two inhibitor-modified SrtA species elute on a C18 reverse phase HPLC column (peaks 1 and 2), and mass spectrometry indicates that both peaks contain SrtA covalently modified by a single inhibitor molecule. At pH values of 6 and 7, little modification occurs within 20 h with a slight excess of species 2 being produced. However, at pH 9.0, nearly all of the SrtA protein is converted to species 1 (Fig. 3A). An analysis of the pH dependence of modification indicates that the production of species 1 is most efficient at pH values nearing 9, whereas maximal production of the minor species 2 occurs at pH ϳ7 (data not shown). Because the enzyme is maximally inhibited by the vinyl sulfone compound at high pH values (Fig. 2C), where species 1 is almost exclusively present, the data strongly suggest that species 1 corresponds to the inhibitor-SrtA complex that forms during the inactivation. This hypothesis was substantiated by thiolating Cys 184 with a 2-fold excess of nitroxide spin label HO-225 ((1-oxyl-2,2,5,5-tetramethylpyrroline-3-methyl)-methanethiosulfonate) (38) and testing for its ability to react with the vinyl sulfone inhibitor at pH 9. As shown in Fig. 3D, preincubation with HO-225 prevented the formation of both species 1 and 2. These data are consistent with Cys 184 being the primary and most reactive site for modification by the vinyl sulfone at pH values above 8.
To investigate whether an imidazolium ion is present in the active site of SrtA, NMR was used to determine the pK a of the His 120 side chain. Because the chemical shifts of atoms within the imidazole are expected to be sensitive to the ionization state of the side chain, a series of 1 H-13 C heteronuclear single quantum correlation spectra were recorded using a 13 C-and 15 N-enriched sample of SrtA, and the chemical shifts of the 1 H-␦2 and 1 H-⑀1 atoms of His 120 were recorded at pH values between 4.5 and 10. Fig. 4 shows a plot of the chemical shift of the 1 H-⑀1 atom as a function of pH (a similar curve was obtained for the 1 H-␦2 atom). A fit of the titration data to Equation 2 (see "Experimental Procedures") indicates that the His 120 side chain has a pK a of 7.0 Ϯ 0.1 in the presence of 20 mM CaCl 2 . Because Ca 2ϩ binding near the active site has been shown to enhance the in vitro activity of SrtA 8-fold (16), we repeated the pH titration experiment in calcium-free buffer (buffer N with no calcium) to determine the effects of Ca 2ϩ on the ionization state of the active site His 120 residue. No significant changes in the pK a of His 120 were detected in the absence of calcium (pK a ϭ 6.6 Ϯ 0.1) (Fig. 4B). The data indicate that at physiological pH, the imidazole side chain is only partially protonated and that calcium does not stimulate the activity of SrtA by altering the ionization state of His 120 . Ca 2ϩ binding may stabilize substrate binding or the fold of the protein.
If a thiolate-imidazolium ion pair exists in the active site of SrtA, one would expect the thiolate of Cys 184 to perturb the ionization of His 120 (i.e. alter its pK a ). We ascertained the effect of Cys 184 on His 120 by repeating the NMR-pH titration experiment using the mutant C184A SrtA ⌬N59 . The results of this titration (Fig. 4A, open circles) were fit to Equation 2 to obtain a pK a of 7.3 Ϯ 0.1. The similarly measured pK a values for His 120 in the wild-type and mutant SrtA proteins suggest that its ionization state is independent of the Cys 184 side chain, a finding that is inconsistent with the existence of an ion pair between His 120 and Cys 184 .
A thiolate-imidazolium ion pair between the side chains of His 120 and Cys 184 would enhance the reactivity of the thiol toward electrophiles at weakly acidic and neutral pH (36) because the imidazole group would act to polarize the thiol. In this scenario, it is expected that the pK a value of the cysteine FIG. 4. A, pH titration of His 120 . Plot of the chemical shift changes in the 1 H-⑀1 resonance of His 120 as a function of pH. The titrations were performed in the presence of 20 mM calcium, and the data were fit to Equation 2 to obtain pK a values of 7.0 and 7.3 for wild-type SrtA ⌬N59 (solid circles) and C184A SrtA ⌬N59 (open circles), respectively. Analysis of the heteronuclear single quantum correlation NMR spectra over the same pH range (data not shown) indicates that the protein remains folded throughout the titration study. B, effect of calcium on the pK a of His 120 . The solid and open circles are titration data obtained for the wild-type SrtA ⌬N59 in the presence (solid circles) and absence (open circles) of 20 mM CaCl 2 . Fits to Equation 2 reveal the pK a of His 120 in the absence of calcium to be 6.6. thiol would be more acidic and the histidine imidazole would be more basic than their normal values of 7.85 and 6.0, respectively (39). For example, the anomalous pK a values of 3.3 (for the Cys 25 thiol) and 8.5 (for the His 159 imidazolium) support the existence of a thiolate-imidazolium ion pair in papain (40,41). In SrtA, however, our investigation argues against the use of a thiolate-imidazolium ion pair in the reaction mechanism.
The pH dependence of modification of Cys 184 by the vinyl sulfone inhibitor (Fig. 2C) shows a dramatic increase in the first-order rate of inactivation as the pH is raised from 7.5 to 10 (at pH 10 the inhibitor is 20 times more reactive than at pH 7.5). Because the inhibitor is expected to be more reactive toward a thiolate ion, the pH dependence of inhibition can be attributed to the deprotonation of Cys 184 . These data argue against the presence of an ion pair in the active site, because the pK a of Cys 184 is estimated to be ϳ9. 4.
NMR studies of the His 120 side chain have measured its pK a at 7.0 in the wild-type protein, which is inconsistent with the presence of an imidazolium cation at neutral pH. Moreover, our finding that the removal of the Cys 184 side chain ( C184A SrtA ⌬N59 ) has only a modest effect on the ionization state of His 120 side chain argues against the presence of an ion pair, because in the papain system the pK a of His 159 is lowered by 4.5 pH units upon the methylthiolation of the Cys 25 (25).
Although the thiolate-imidazolium ion pair is a common catalytic entity of cysteine proteases, it is not universal (42,43). The absence of an ion pair in SrtA suggests its catalytic mechanism may be similar to the viral 3C proteases (picornains), a structurally and mechanistically distinct group of cysteine proteases that perform general base catalysis (reviewed in Refs. 36 and 44). The crystal structures of the picornains of hepatitis A (45), rhinovirus (46), and poliovirus (47) show a similarity of three-dimensional structures and catalytic mechanisms to the serine proteases of the trypsin/chymotrypsin family and may be evolutionarily related (36,43,44). The pH-dependent alkylation of the active site cysteine of poliovirus protease 3C with iodoacetamide has measured its pK a at 8.86 (42), which is similar to the estimated pK a of the SrtA thiol. This is consistent with a reaction mechanism in which the cysteine nucleophile is uncharged at physiological pH, and the histidine functions as a general base.