Structural and biochemical analyses of a Clostridium perfringens sortase D transpeptidase

The structure of C. perfringens sortase D was determined at 1.99 Å resolution. Comparative biochemical and structural analyses revealed that this transpeptidase may represent a new subclass of the sortase D family.


Introduction
Clostridium perfringens is an anaerobic, spore-forming bacterium found in a wide range of environmental conditions including soil, marine sediments and the intestinal tract of humans and other vertebrates (McClane, 2007;Brynestad & Granum, 2002;Grass et al., 2013). This highly pathogenic Gram-positive bacterium is the second most common cause of foodborne diseases in the US, with an estimated one million reports each year (Scallan et al., 2011). Additionally, C. perfringens isolates can be responsible for the development of non-foodborne human gastrointestinal diseases, including sporadic diarrhoea and antibiotic-associated diarrhoea (Collie et al., 1998). The pathogenesis of C. perfringens-derived foodborne illnesses typically originates from germination of the spores in raw and cooked food under oxygen-limiting conditions (Jenuja et al., 2010). Significantly, the formation of spores correlates with the survival mode of the organism, which allows it to resist extreme temperatures, including heat treatment and refrigeration (Jenuja et al., 2010;Strong et al., 1966;Traci & Duncan, 1974). Once digested, C. perfringens isolates germinate in the intestinal tract, where they produce C. perfringens enterotoxin (CPE), resulting in gastrointestinal illnesses (Jenuja et al., 2010).
The exact mechanisms contributing to the pathogeneicity of C. perfringens isolates remain unclear, with a number of factors identified to promote the survival of the organism (Orsburn et al., 2008). In common with many pathogenic Gram-positive bacteria, C. perfringens displays and anchors a diverse array of surface proteins on its cell wall with functions such as adaptation to extreme environmental conditions, evasion of the host immune system, and virulence (Hendrickx et al., 2011;Marraffini & Schneewind, 2006). The covalent attachment of many of these proteins is mediated by the socalled sortase enzymes, a unique family of membrane-bound, cysteine transpeptidases which were first identified in Staphylococcus aureus . Mechanistically, the sortase catalytic activity is best illustrated for S. aureus sortase A, where the enzyme recognizes a unique pentapeptide sorting motif (Leu-Pro-X-Thr-Gly, LPXTG, where X denotes any amino acid) within the cell-wall sorting signal (CWSS) region located at the carboxyl-terminus (Kruger et al., 2004;Ton-That et al., 1999). Upon recognition, sortase A (SrtA) cleaves the Thr-Gly peptide bond, leading to loss of the C-terminal glycine and the formation of a thioacyl intermediate (Kruger et al., 2004;Ton-That et al., 1999). Subsequently, the presence of oligo-glycine acts as a nucleophile to dissociate the acyl-enzyme intermediate and promote sortase-assisted covalent coupling of the free amino group of the oligo-glycine to the Thr carboxyl group (Suree et al., 2009;Weiner et al., 2010).
Beyond the versatile housekeeping enzyme SrtA, the sortase enzymes can phylogenetically be characterized into five other distinct classes (Suree et al., 2007). The latter classes display more specialized roles. For example, members of the sortase B family are catalytically important for iron acquisition Mazmanian et al., 2002Mazmanian et al., , 2003 and class C sortases are predominantly responsible for the assembly of pili, which are involved in microbial adhesion and biofilm formation (Spirig et al., 2011;Cozzi et al., 2012Cozzi et al., , 2013Khare et al., 2011;Manzano et al., 2008;Neiers et al., 2009;Persson, 2011;Wu et al., 2012). Much less is known about the class D, E and F enzymes. Class D sortases are thought to induce spore formation in an oxygen-limiting environment (Marraffini & Schneewind, 2006, 2007, and the study of Bacillus anthracis SrtD revealed an exclusive preference towards the Leu-Pro-Asn-Thr-Ala (LPNTA) signal motif (Marraffini & Schneewind, 2006). Interestingly, B. anthracis also expresses the class A sortases which recognize the canonical LP(A/N/K)TG signal motif. Despite the signal motifs only differing slightly in their sequences, both B. anthracis SrtA and SrtD function nonredundantly, indicating evolved specificity towards the respective signal motifs (Marraffini & Schneewind, 2006). Functional analyses further revealed that B. anthracis SrtD functions at different stages of sporulation, including the attachment of the acidic surface protein BasH to the peptidoglycans of developing forespores (precursor spores; Marraffini & Schneewind, 2006) and the presentation of the BasI surface protein on the envelope of pre-divisional sporulating cells (Marraffini & Schneewind, 2007).
In the current study, we present the crystal structure of a C. perfringens transpeptidase which belongs to the class D family of sortases, suggesting a potential role of this enzyme in C. perfringens spore formation. Biochemically, the recombinant C. perfringens SrtD (CpSrtD) is catalytically active, with a high preference for the Leu-Pro-Gln-Thr-Gly-Ser (LPQTGS) signal motif. Additionally, CpSrtD catalytic activity is also dependent on a metal cation, with the presence of magnesium appearing to enhance CpSrtD catalysis towards the LPQTGS signal motif. The structure of CpSrtD is distinct from the previously reported NMR structure of B. anthracis sortase D (Robson et al., 2012), leading us to propose C. perfringens sortase D as a new subclass of the D-type sortase family.

Cloning, expression and purification of CpSrtD
Codon-optimized C. perfringens sortase D (CpSrtD; CPE_RS01475) cDNA encoding residues 23-187 was synthesized at GeneArt AG and cloned into the pET-28a expression vector (Novagen) at the NdeI and XhoI restriction sites to generate an N-terminally His 6 -tagged recombinant protein.

Differential scanning fluorimetry (DSF)
Protein stability was determined across a series of conditions encompassing a range of different buffers/pH values and salts as previously described (Seabrook & Newman, 2013). Recombinant CpSrtD was found to be most stable in a buffer consisting of 50 mM MES pH 6.5, 200 mM NaCl (Supplementary Fig. S2).

Crystallization
Crystallization experiments of recombinant CpSrtD were set up at both 281 and 293 K using the Netherlands Cancer Institute (NKI) dual screen set (Newman et al., 2005). Recombinant CpSrtD was prepared at a concentration of 20 mg ml À1 in a 50 mM MES pH 6.5, 200 mM NaCl buffer formulation, and sitting-drop vapour-diffusion experiments were then set up using 200 nl protein solution and 200 nl reservoir solution. Protein crystals successfully grew under several conditions at both 281 and 293 K, with the best buffer formulation for growing native CpSrtD crystals consisting of 200 mM ammonium acetate, 100 mM bis-tris chloride pH 5.5, 25%(w/v) PEG 3350, yielding crystals after 1 d of incubation at 281 K. Under this condition, the protein crystal adopts a thin plate morphology with dimensions of approximately 350 Â 600 mm ( Supplementary Fig. S3a).

Data collection and structural determination
360 1 images were obtained on the MX-2 microfocus beamline at the Australian Synchrotron from a crystal that had been cryocooled to 100 K. The reflections were indexed using XDS (Kabsch, 2010) and scaled using AIMLESS (Evans, 2011). ClustalW was used to align the C. perfringens sortase sequence with the sequence from PDB entry 3g66 (Neiers et al., 2009), and CHAINSAW (Stein, 2008) was then used with PDB entry 3g66 (with an estimated sequence identity of 21%) to obtain a model for Phaser (McCoy et al., 2007), which was used to obtain the initial phases. The initial Phaser output LLG was 21.5, with a Z-score of 5.8. Two molecules were placed in the asymmetric unit and the final output values for the solution were LLG = 90.7, TFZ = 7.2, with an R value of 56.8. The model was initially rebuilt using Buccaneer (Cowtan, 2006) and subsequently rebuilt manually using Coot (Emsley et al., 2010) and refined using REFMAC (Murshudov et al., 2011). The data were 99.5% complete to a resolution of 1.99 Å and the final model had an R work of 17.6% and an R free of 21.3% (see Table 1 for crystallographic statistics). According to the PDB report, 97% of the residues are in the most favoured region of the Ramachandran plot and 3% are in the allowed region, with no outliers.

In vitro thioacyl intermediate formation
A solution containing 70 mM CpSrtD was incubated with a 15 mM solution of a peptide comprising of the first 16 aminoacid residues of amyloid-(A 1-16 ) fused at the C-terminus to a variety of sortase signal motifs in the presence of MES reaction buffer (50 mM MES pH 6.5, 200 mM NaCl, 1 mM TCEP) for 3 h at 316 K. The reaction was quenched by adding nonreducing NuPAGE loading buffer (Life Technologies). Following SDS-PAGE, resolved protein samples were transferred onto nitrocellulose membranes for Western blot analyses of thioacyl intermediate formation using an antibody against A (WO2). Equal loading was determined by Western blot using anti-His 5 antibody (Qiagen).
To analyze the impact of different metal ions on the catalytic activity of CpSrtD, recombinant protein was first incubated with 100 mM EDTA for 2 h at room temperature (RT). EDTA-treated CpSrtD was then diluted tenfold before being tested for catalytic activity in the presence of 10 mM metal ions for 3 h at 316 K. The reaction was quenched by the addition of nonreducing NuPAGE loading buffer. SDS-PAGE and subsequent Western blot analyses were then performed as above.

Dynamic light scattering (DLS)
A 20 ml aliquot containing 20 mg ml À1 CpSrtD was dispensed into each well of a black 384-well microplate with an optically clear base (Corning). Measurements were collected at 293 K using 5 s acquisitions and allowing the attenuation and laser power to be automatically set by the DLS system (DynaPro Plate Reader, Wyatt). The resulting distributions were derived from regularization fits to the average of 50 correlation curves using the DYNAMICS software (Wyatt) and are displayed as the intensity of light scattered as a function of the hydrodynamic radius ( Supplementary Fig. S4).

Overall structure of C. perfringens sortase
In this study, we report the crystal structure of a C. perfringens sortase that was solved at 1.99 Å resolution by molecular replacement (Table 1) jF obs j À jF calc j = P hkl jF obs j and is calculated using all data; R free is the R factor based on 5% of the data that were excluded from refinement. } R.m.s.d. is the root-mean-square deviation from ideal values (Engh & Huber, 1991). the crystallographic asymmetric unit (Supplementary Fig.  S3b). The final maps derived from the X-ray data showed clear density for 160 residues (28-187), only lacking the first five residues along with the N-terminal hexahistidine tag and thrombin cleavage site. Alignment of the two monomers revealed a r.m.s.d. value of less than 0.53 Å , with a slight conformational difference of the turn within the 1-2 helixturn-helix structure (Supplementary Fig. S3c). Each monomer displayed the typical eight -strands that form a -barrel structure (Fig. 1). This distinct barrel structure is also present in other sortases (Fig. 2a) and serves as a hallmark of this family of enzymes. A second distinctive feature observed in all sortase enzymes is the surface presentation of the conserved active site, comprising of a catalytic cysteine that is surrounded by a histidine and an arginine residue (Fig. 2a). Previous studies have demonstrated that the presence of both histidine and arginine are necessary for efficient catalysis by the conserved cysteine residue (Frankel et al., 2007;Clancy et al., 2010). These key residues are also present on the surface of C. perfringens sortase, with the catalytic cysteine located within the 7 strand at position 171 (Fig. 1). The adjacent arginine is found in the 8 strand at position 178, while the histidine is positioned within the 3-4 loop at residue 109 (Fig. 1).
Initial sequence analyses revealed that the C. perfringens sortase belongs to the class D subfamily 5 of transpeptidases (Dramsi et al., 2005), which is also referred to as class E of sortases (SrtE; Spirig et al., 2011). Comparative sequence analyses of the C. perfringens sortase D with the previously identified S. aureus sortase A (SaSrtA) and sortase B (SaSrtB), Streptococcus pneumoniae sortase C-2 (SpSrtC2) and B. anthracis sortase D (BaSrtC) revealed approximately 18, 25, 26 and 28% identity in their amino-acid sequences (Fig. 2b). The sparse similarity between the amino-acid sequences of the C. perfringens sortase and members of classes A-D of the sortase family further highlights that the C. perfringens transpeptidase may belong to a new class.

C. perfringens sortase recognizes the LPQTGS signal motif for transpeptidation
One notable feature that is observed in most sortases is their ability to preferentially recognize a specific signal motif for catalysis. To identify the signal motif preferred by C. perfringens sortase to achieve efficient catalysis, we performed a series of in vitro transpeptidation reactions using a substrate that consists of the first 16 amino-acid residues of the amyloid-(A 1-16 ) peptide fused at the C-terminus with the LPETG, LPNTGS, LPQTGS or LAETG sorting motifs. This panel of substrates represents signal motifs that are recognized by the different classes of sortase family, including class A (LPETG), class D (LPNTGS and LPQTGS) and class E (LAETG). Western blot analyses using anti-A antibody to detect the CpSrtD-substrate thioacyl intermediate revealed no CpSrtD catalytic activity towards the class E signal motif (Fig. 3a, lane 5, top panel) and minimal activity towards either the LPETG (Fig. 3a, lane 2, top panel) or LPNTGS (Fig. 3a, lane 3, top panel) signal motifs. In contrast, recombinant CpSrtD showed a strong preference towards the LPQTGS motif (Fig. 3a, lane 4, top panel).

CpSrtD catalysis is temperature-dependent
Heat-resistant C. perfringens spores can be produced at a faster, more efficient rate by incubating the isolates at 316 K (Garcia-Alvarado et al., 1992). To determine the optimal temperature at which C. perfringens SrtD achieves maximal transpeptidation activity in vitro, we analyzed the efficiency of the enzyme to catalyze the formation of the thioacyl inter- The overall structure of C. perfringens sortase D. The secondary structure of C. perfringens sortase D monomer A is represented by red 3 10 -helices and -helices and yellow -strands (PDB entry 4d70). The N-and C-termini of the enzyme are indicated. The conserved catalytic triad consisting of His109 (blue), Cys171 (purple) and Arg178 (orange) is shown. The yellow -strands form the -barrel structure which is typically observed in the sortase family of enzymes (right-hand side). Figures were generated using PyMOL (v.1.5.0.4; Schrö dinger).  mediate at different temperatures (Fig. 3b). Recombinant CpSrtD is highly inefficient in forming a thioacyl intermediate with the A 1-16 -LPQTGS substrate when incubated at RT (lane 2) or at 303 K (lane 3). However, the transpeptidase activity of the enzyme can be improved by incubating the CpSrtD at higher temperatures (Fig. 3b, lanes 4-6), with the maximal catalytic efficiency being observed at 316 K (lane 5).

CpSrtD activity is dependent on the presence of metal cation
Previous studies have demonstrated that S. aureus SrtA requires Ca 2+ ion for its catalytic activity (Naik et al., 2006). To assess whether CpSrtD activity is also dependent on a metal cation, we first investigated the effect of EDTA on the basal activity of the enzyme. Addition of EDTA reduced the ability of CpSrtD to form thioacyl intermediates with the A 1-16 -LPQTGS substrate in a concentration-dependent manner (Fig. 3c), indicating metal-iondependent catalysis. Interestingly, only EDTA at concentrations of 20 mM and higher demonstrated sufficient chelating properties which lead to a reduced CpSrtD transpeptidase activity (Fig. 3c,  lanes 6-9), suggesting the possibility of a tightly bound metal cation to the enzyme.
To further identify the specific metal cation(s) responsible for catalysis, we selected and examined the effect of a panel of divalent and trivalent cations on CpSrtD-mediated transpeptidation (Fig. 3d). The addition of 10 mM CaCl 2 (lane 2) or MnCl 2 (lane 7) had no impact on the basal activity of CpSrtD, while the presence of CoCl 2 (lane 3), CuCl 2 (lane 4), FeCl 3 (lane 5) or NiCl 2 (lane 8) reduced the ability of CpSrtD to form thioacyl intermediates. Interestingly, 10 mM ZnCl 2 (lane 9) completely inhibits CpSrtD transpeptidase activity. In contrast, the addition of MgCl 2 increases CpSrtD catalytic activity as demonstrated by the increased amount of thioacyl intermediate (lane 6), suggesting the potential importance of the Mg 2+ cation for CpSrtD activation and catalysis.

Discussion
A previous report suggested that the C. perfringens sortase described in this study belongs to the class E family of sortases (Spirig et al., 2011). At the   structural level, recombinant C. perfringens sortase demonstrated the classic -barrel configuration found in all sortase enzymes and displays the conserved catalytic cysteine at position 171 flanked by histidine and arginine residues (Fig. 1). Despite the previous report, our biochemical data indicate that the C. perfringens sortase does not belong to the class E family of transpeptidases owing to its inability to recognize and catalyze the LAETG motif (Fig. 3a, lane 5), the sorting signal motif preferred by this class of enzymes (Duong et al., 2012). Instead, the recombinant C. perfringens sortase demonstrated efficient catalysis towards a class D signal motif, LPQTGS (Fig. 3a, lane 4), highlighting the possibility that this C. perfringens sortase belongs to the class D family of enzymes (CpSrtD) that are responsible for spore formation under anaerobic conditions. This notion was further supported by additional biochemical analyses demonstrating that CpSrtD is most efficient in catalysis at 316 K (Fig. 3b), which is the temperature reported for optimally inducing spore formation in C. perfringens isolates (Garcia-Alvarado et al., 1992). Our findings on CpSrtD substrate selectivity are also supported by analysis of the C. perfringens strain 13 genome, which revealed that the sortase gene is clustered in the same operon as a hypothetical cell-wall anchor protein (CPE_RS01465) which possesses a C-terminal LPQTGS signal motif, thus underlining the possibility of the LPQTGS motif as one of the natural substrates of this enzyme (Shimizu et al., 2002).
Comparative sequence analyses revealed that C. perfringens sortase D is relatively distinct from the previously reported class D sortase isolated from B. anthracis (unfortunately referred to as BaSrtC; Robson et al., 2012), with a limited 28% identity in their amino-acid sequence (Fig. 2b). Superposition of the secondary structures reveals further differences between the two enzymes, with a calculated r.m.s.d. of 1.7 Å (Fig. 4a). The most notable difference is the presence of N-terminal -helices in C. perfringens SrtD which are absent in the B. anthracis SrtD structure. Previously, only class B and C sortases have been observed to display long N-terminal -helices (Fig. 2a). The N-terminal region of class C sortases plays an important role in catalysis, in which the -helices flank a flexible loop region that form the so-called 'lid' structure ( Fig. 2a), which is thought to be responsible in controlling the access of substrates to the SrtC catalytic site (Khare et al., 2011;Manzano et al., 2009;Neiers et al., 2009). Within the 'lid' structure in all class C sortases lies the conserved 'lid' domain consisting of Asp-Pro-Try/Trp/Phe (DPY/W/F; Cozzi et al., 2013;Khare et al., 2011;, and a point mutation of the key residue within the 'lid' domain is necessary for activation of the class C sortases in vitro (Cozzi et al., 2013). In contrast, recombinant CpSrtD is catalytically active in the wild-type form in vitro (Fig. 3), and sequence analysis revealed that the conserved 'lid' domain is not present in this enzyme (Fig. 2b). Therefore, it is possible that the N-terminal -helices present in C. perfringens sortase D possess different, albeit unknown, function(s). The role of the N-terminal -helices in sortase B is also unknown; however, both S. aureus and B. anthracis sortase B can be distinguished from the other classes of sortases, including C. perfringens sortase D, by the presence of an additional conserved residue within the active site (Jacobitz et al., 2014;Zhang et al., 2004). These class B sortases possess a conserved aspartate (at position 223 in SaSrtB; Fig. 2a) which has a suggested role in controlling substrate specificity without affecting the overall transpeptidation activity (Jacobitz et al., 2014). The overall structure of sortase B represents a nearly equal distribution of and structures, with eight -strands forming the -barrel core structure surrounded by several long and short -helices (Fig. 2a). Conversely, the S. aureus sortase A structure is predominantly made up of loops that connect the -barrel structure (Fig. 2a) is largely composed of a helix-turn-helix, the N-terminus of SaSrtA is unstructured (Fig. 2a). It is interesting to note that in contrast to the other classes of sortases, the N-terminus of SaSrtA is positioned on the opposite side to the active site (Fig. 2a). This indicates the natural orientation of SaSrtA on the bacterial cell wall, where it is likely that the active site is exposed to the surface and away from the cell wall. As previously highlighted for SaSrtB and BaSrtB (Zhang et al., 2004), the active site of the remaining classes of sortases could be partially buried when anchored onto the bacterial cell wall owing to the active site being positioned on the same plane of the protein as their N-termini. Analyses of the two class D enzymes further revealed the presence of a unique -helix structure within the CpSrtD loop that connects the 2 and 3 strands (Fig. 4b). In contrast, the 2-3 loop in B. anthracis sortase D is uninterrupted, and NOESY spectra suggested that the residues within this loop, as well as the residues in the loop that connects 4 and 1, exhibited resonance line broadening associated with protein oligomerization (Robson et al., 2012). By comparison, our crystallization studies indicated that CpSrtD is likely to exist in a monomeric form. This notion is supported by gel-filtration analyses, revealing a single peak with a molecular weight that corresponds to a C. perfringens sortase D monomer (Supplementary Fig. S1a). Additionally, dynamic light-scattering (DLS) experiments to measure the hydrodynamic radius of the enzyme also demonstrated that at the concentration used in the crystallization studies (20 mg ml À1 ) recombinant CpSrtD is likely to be monomeric ( Supplementary Fig. S4). It is not known whether the presence of the -helical structure in CpSrtD plays a role in forcing the enzyme to adopt a monomeric form; further mutational and structural studies would be required to elucidate this. The overall -helical structures within the CpSrtD crystal structure are indeed distinctive compared with B. anthracis sortase D. It is possible that the crystallization buffer conditions used in our studies have provided a more physiological environment (sodium chloride and ammonium acetate salts at about 200 mM, compared with the NMR conditions, which were just 20 mM HEPES buffer) that allows proper folding of surface secondary structures; further solution studies of the B. anthracis sortase D structure under more physiological buffer conditions would be important to provide a better comparative analysis.
Both class D enzymes can also be further differentiated at the catalytic level, where the sortases demonstrated a differential preference towards specific sorting signal motifs. As previously highlighted, the C. perfringens sortase D is catalytically active towards the LPQTGS signal motif, but performed inefficiently towards the LPNTGS signal motif (Fig. 3a), which is highly similar to the LPNTAS motif preferred by the B. anthracis sortase D (Robson et al., 2012). Additionally, both enzymes are catalytically active at different temperatures, with the C. perfringens sortase D demonstrating efficient catalysis at 316 K in vitro (Fig. 3b). Interestingly, while the B. anthracis sortase D is catalytically active at room temperature (Robson et al., 2012), our recombinant C. perfringens sortase D was highly inefficient when incubated at this temperature in vitro (Fig. 3b). How the LQPTGS signal motif is positioned within the CpSrtD catalytic cleft is unclear; however, comparative analysis with previous structural studies of the SaSrtA-LPAT complex have provided some insights (Zong et al., 2004). While the environment of the catalytic cleft in CpSrtD is equally rich in hydrophobic residues for substrate binding, it is relatively narrow owing to the position of a small helix within the 6-7 loop when compared with SaSrtA. As such, it is possible that CpSrtD requires a substrate-induced conformational change to promote proper binding of the LPQTGS motif, but additional structural studies will need to be performed to further elucidate this.
Overall, the C. perfringens sortase D reported in this study is structurally and catalytically distinct from the previously reported class D enzyme isolated from B. anthracis, suggesting that CpSrtD may represent a new subclass of the sortase D family. Our structural and biochemical analyses suggest further characterization of the biological roles of CpSrtD in promoting spore formation, which will provide further insights into the pathogenesis of foodborne illnesses derived from C. perfringens infections. Ultimately, these studies may lead to the development of new antimicrobial agents for controlling foodborne outbreaks associated with this highly pathogenic bacterium.