Characterisation of SEQ0694 (PrsA/PrtM) of Streptococcus equi as a functional peptidyl-prolyl isomerase affecting multiple secreted protein substrates.

Peptidyl-prolyl isomerase (PPIase) lipoproteins have been shown to influence the virulence of a number of Gram-positive bacterial human and animal pathogens, most likely through facilitating the folding of cell envelope and secreted virulence factors. Here, we used a proteomic approach to demonstrate that the Streptococcus equi PPIase SEQ0694 alters the production of multiple secreted proteins, including at least two putative virulence factors (FNE and IdeE2). We demonstrate also that, despite some unusual sequence features, recombinant SEQ0694 and its central parvulin domain are functional PPIases. These data add to our knowledge of the mechanisms by which lipoprotein PPIases contribute to the virulence of streptococcal pathogens.


Introduction
In order to interact with their environments, bacteria translocate significant numbers of proteins across their plasma membranes, either for eventual release (secretion) or for localisation within the cell envelope. [1][2][3] In pathogens, this 'secretome' plays a vital role in host-pathogen interactions and consequently the mechanisms of protein translocation are of much interest as 'virulence-associated' functions. Proteins exported by the Sec translocase emerge on the extracytoplasmic side of the plasma membrane as unfolded proteins and the subsequent correct folding of these proteins is therefore critical to their functioning. In 'diderm' bacteria (those with outer membranes), a variety of periplasmic chaperones are required to allow protein folding in the periplasm and/or translocation across or into the outer membrane. 1,4 In monoderm Grampositive bacteria, secreted proteins fold at the membrane-wall interface with the assistance of a range of accessory components of the Sec translocase. 5 These include proteins belonging to the peptidyl-prolyl Isomerase (PPIase) family, which assist protein folding by catalysing cis-trans isomerisation of the peptide bond preceding proline residues. 6,7 In many Gram-positive bacteria, these PPIases are N-terminally lipid-anchored lipoproteins, presumably because the localisation of a PPIase peripheral to the plasma membrane surface places it in an optimal position to engage with substrate proteins emerging from the Sec translocon. 8 Several lipoprotein PPIases have been shown to have significant roles in bacterial physiology, notably PrsA in Bacillus subtilis. 9 Moreover, in some pathogens PPIases have been shown to affect virulence, 7 including PrsA of Bacillus anthracis, 10 Enterococcus faecalis EF0685 and EF1534, 11 Listeria monocytogenes PrsA2, 12,13 Streptococcus pneumoniae SlrA and PpmA 14 and Streptococcus pyogenes PrsA. 15 Some of these PPIase belong to the cyclophilin subfamily (e.g. S. pneumoniae SlrA; E. faecalis EF1534) but many belong to the parvulin subfamily, 16 including the members of PrsA family that appear to be ubiquitous in Firmicute genomes.
Streptococcus equi is the causative agent of the widespread equine disease Strangles. 17,18 We have previously shown that the PrsA homologue of S. equi (UniProt: C0M9L5, originally denoted PrtM) plays a significant role in S. equi virulence, both in an air interface tissue culture model, a mouse model and, most significantly, in the equine host. 19 PrtM is here referred to as SEQ0694, based on its annotation in the S. equi genome. 17 To further investigate the role of SEQ0694 we have here characterised the recombinant protein as a functional PPIase and used a proteomic approach to demonstrate that SEQ0694 likely influences the folding and activity of multiple secreted a Department of Applied Sciences, Faculty of Health & Life Sciences, proteins of S. equi, including at least two putative virulence factors.

Materials and methods
Bacterial strains and growth S. equi strain 4047 (wild type) and its isogenic mutant strain (DprtM 138-213 ) with a deletion of codons 138 to 213 in seq0694 (i.e. lacking the central domain of SEQ0694, ESI, † Fig. S1) are described in Hamilton et al. 19 S. equi strains were grown in Todd Hewitt media. Escherichia coli TOP10 and BL21 were grown in LB media.

Production and purification of recombinant proteins
Genomic DNA from S. equi 4047 was isolated using a DNeasy extraction kit (Qiagen). To produce recombinant N-terminally His-tagged full-length SEQ0694 (rSEQ0694), the seq0694 ORF, minus the sequence encoding the signal peptide, was amplified from S. equi 4047 genomic DNA using the primer pair 5 0 GATC GATC À C À A À T À A À T À GTGTCAGTCTACAAATGACAATACAAGTG 3 0 (forward primer, NdeI site underlined) and 5 0 GATCGATC À C À T À C À G À A À GATATTTTTCTGACTTAGATTTAGAAGATTGAC 3 0 (reverse primer, XhoI site underlined) and KOD Hot Start polymerase (Merck Chemicals) according to the manufacturer's instructions. The amplified ORF was cloned into pET28a (Merck Chemicals) using NdeI-XhoI and expressed in E. coli BL21(DE3) grown at 37 1C with shaking at 200 rpm, to an absorbance of 0.6 at 600 nm, in LB medium supplemented with 100 mg mL À1 kanamycin. Induction was performed by the addition of isopropyl-1-thio-b-Dgalactopyranoside to a concentration of 240 mg mL À1 , followed by further incubation for 18 h at 30 1C with shaking at 100 rpm. rSEQ0694 was purified according to the method of Malik et al., 20 except that the purified protein was concentrated and the buffer exchanged into 18.2 MO cm À1 water using 10 kDa cut-off centrifugal concentrator units (Viva Science) The identity of rSEQ0694 was confirmed by peptide mass fingerprinting of trypsinized bands excised from Coomassie blue-stained SDS-polyacrylamide gels (see below). In addition to rSEQ0694, the section of the seq0694 ORF encoding the predicted parvulin domain of SEQ0694 (amino acids 148-242, ESI, † Fig. S1; rSEQ0694parv) was amplified using primer pair TGCCATAG À C À A À T À A À T À GACTACTCAGGTCACTACTCTAG ACAATG (forward, NdeI site underlined) and TGCCATAG À C À T À C À G À A À GTTAGGCTTTTTTGGTTACCTTAACA (reverse, XhoI site underlined), cloned, expressed and the protein purified as described above, except that 5 kDa, 6 mL cut-off concentrator units (Viva Science) were used.
The concentration of both purified proteins was determined using the Bradford Assay.

Protease-coupled peptidylprolyl isomerase (PPIase) assay
The standard protease-coupled PPIase assay 12,21 was employed using three peptide substrates having a consensus sequence Suc-Ala-X-Pro-Phe-pNA (Suc, succinyl; X = alanine, lysine or phenylalanine; pNa, paranitroaniline). Assays were performed by mixing 10 mL of purified rSEQ0694 (60 mg mL À1 ) or rSEQ0694parv (40 mg mL À1 ) (diluted in 20 mM HEPES, pH 7.4; 140 mM NaCl; 10% v/v glycerol), cyclophilin (positive control) or diluent alone (negative control) with 480 mL of buffer (20 mM HEPES, pH 7.4; 140 mM NaCl; 1 mM DTT) and allowing the mixture to equilibrate on ice for 5 min. 10 mL of ice-cold chymotrypsin (20 mg mL À1 in 0.001 M HCl; 0.002 M CaCl 2 ) was pipetted into a cuvette in a spectrophotometer (Spectronic Unicam Helios-a, Thermos Electron Corporation), zeroed at 390 nm. The 490 mL ice-cold assay mixture was quickly added to and mixed with the chymotrypsin, followed by 500 mL tetrapeptide substrate in ice-cold 20 mM HEPES, pH 7.4, 140 mM NaCl, 1 mM DTT, so as to give a final concentration of 37.5-75 mM peptide, and mixed quickly by pipetting. The final chymotrypsin concentration in the reaction mixture was 0.2 mg mL À1 . The rate of the reaction (cis-trans isomerization) was measured by following colour formation (absorbance at 390 nm) resulting from pNA release from the trans form of the tetrapeptide substrate by chymotrypsin, for a maximum of 6 min. Spectrophotometric readings were recorded automatically via Vision 32 (Unicam Ltd) software.
Reported kinetic data are given as the mean value of triplicate measurements for every condition. To ascertain if these data reflected true Michaelis-Menten kinetics, a Lineweaver-Burk plot was constructed and used to determine value of K m (calculated by reciprocalising the X intercept in the Line-weaver-Burk plot). The specificity constant (M s) was determined by dividing K cat by K ma .

Effect of chymotrypsin on rSEQ0694 and rSEQ0694parv recombinant proteins
To determine if chymotrypsin had any significant effect on the recombinant proteins, 10 mL purified rSEQ0694 (60 mg mL À1 ) or rSEQ0694parv (40 mg mL À1 ) was incubated with chymotrypsin (10 mL, 20 mg mL À1 ) in 880 mL assay buffer (20 mM HEPES, pH 7.4; 140 mM NaCl; 1 mM DTT) for 20 s, 2 min and 5 min at 0 1C. The reaction was stopped by the addition of 100 mL 10 mM PMSF and subsequent incubation for 5 min at 0 1C. Incubations containing rSEQ0694 or rSEQ0694parv incubated with PMSF-inactivated chymotrypsin, chymotrypsin with PMSF, chymotrypsin alone, recombinant proteins with PMSF and recombinant proteins alone served as controls. The reactions were analyzed by SDS-PAGE.

Proteomic & bioinformatic methods
To compare protein expression in S. equi 4047 and DprtM 138-213 , the strains were grown to mid-log phase in Todd Hewitt broth, harvested by centrifugation and total cell proteins prepared as described previously. 22,23 After removal of cells, supernatant proteins were precipitated with 100% (w/v) trichloroacetic acid, washed three times with ice-cold acetone and processed as for total-cell proteins. Two dimensional electrophoresis (2DE) and protein spot identification following trypsinolysis and mass spectrometry were performed as described previously. 22,23 Only proteins identified with Z2 peptide matches and Mascot total scores Z50 were included.

Physiological tests
Survival of S. equi strains in saline solutions was tested by resuspending early stationary phase cells in 0, 0.9%, 14.7% or 29.4% NaCl w/v essentially as described by Reffuveille et al. 25 Cell suspensions were sampled after 24 and 48 h by serial dilution to 10 À3 in the same medium and then plated on Todd Hewitt agar plates for enumeration of surviving colony forming units. Antibiotic sensitivity testing was performed by the standard disc diffusion method using discs containing ampicillin (10 mg per disc), penicillin G (6 mg per disc), streptomycin (500 mg per disc), norflaxacin (5 mg per disc) and vancomycin (30 mg per disc). Zones of inhibition were measured after 48 h incubation.

Results and discussion
rSEQ0694 encodes a functional PPIase Our earlier study of S. equi SEQ0694 19 confirmed that this lipoprotein is needed for full virulence but did not directly address its function. Bioinformatic analyses indicated that SEQ0694 exhibits significant pairwise homologies to members of the PrsA/parvulin family of PPIases. In Firmicutes, these proteins typically contain a central parvulin domain, flanked by N-and C-terminal domains with likely additional chaperone functions or roles in substrate recruitment, 13,26,27 although these flanking domains show limited sequence homology (ESI, † Fig. S1). Notably, the parvulin domains of streptococcal and lactococcal PrsA/PrtM family members have been noted to lack key conserved residues 13,28 (see below) and both L. lactis PpmA and S. pneumoniae PpmA apparently lack PPIase activity, 14,29 although it is notable that these proteins can complement some, but not all phenotypes, of a L. monocytogenes prsA2 mutant. 30 To confirm in vitro PPIase activity of SEQ0694, we produced full-length SEQ0694 as a recombinant protein, rSEQ0694 (ESI, † Fig. S2), for assay using a standard protease-coupled PPIase assay in which the rate of cis to trans isomerisation of a tetrapeptide substrate is measured through selective and colourigenic chymotrypsin hydrolysis of the trans isomer. 12,21 In addition we produced the central parvulin domain of SEQ0694 as a recombinant protein, rSEQ0694parv. Both recombinant proteins were assayed against three tetrapeptide substrates varying in the amino acid preceding the critical proline residue. Whereas no activity could be detected using tetrapeptide substrates containing lysine-proline or alanineproline bonds (data not shown), both rSEQ0694 and rSEQ0694parv were found to exhibit PPIase activity using Suc-Ala-Phe-Pro-Phe-pNA as substrate (Fig. 1). However, both recombinant proteins exhibited notably lower activities than the calf thymus cyclophilin used as a positive control.
Recombinant protein stability to chymotrypsin under the assay conditions was assessed. Significant cleavage of rSEQ0694parv by chymotrypsin was observed (ESI, † Fig. S3), whereas rSEQ0694 remained relatively stable for up to 5 min. This meant that although rSEQ0694parv showed an apparently faster rate of reaction compared with rSEQ0694 ( Fig. 1), enzyme kinetics could only be determined for the latter (Fig. 2)  are not conserved in rSEQ0694 (Fig. 3). However, a candidate Asp (D187) which might fulfil the role of the critical conserved Asp/Cys could be identified in rSEQ0694 (Fig. 3). Although a role of this Asp/Cys as a catalytic nucleophile is not yet fully resolved, 36 its conservation in rSEQ0694 is likely to be significant. Moreover, the conserved residues in bacterial PrsA proteins identified by Jakob et al. 26 are well conserved in SEQ0694 (ESI, † Fig. S1).

Proteomic analyses to identify putative SEQ0694 substrates
Having established that rSEQ0694 is a bona fide PPIase in vitro, we were interested to further explore the nature of its substrates. As SEQ0694 is a lipoprotein, we hypothesized that its substrates would be secreted proteins emerging from the Sec translocase, which need to fold rapidly en route to secretion. Misfolded proteins are typically turned over rapidly by extracytoplasmic proteases such as HtrA family members. 37 Proteomic approaches have therefore been used to identify extracytoplasmic proteins for which folding is dependent on a lipoprotein PPIase. 9,12,38,39 Thus we used proteomics to analyse differential protein expression in the proteomes of S. equi 4047 and an isogenic mutant, DprtM 138-213 , expressing a SEQ0694 N + C domains fusion protein lacking much of the central parvulin domain of SEQ0694 19 (Fig. 3). Note that as the seq0694 mutant strain was originally designated DprtM 138-213 19 for consistency we have retained this designation. Master 2D PAGE gels from 6 matched gel pairs (ESI, † Fig. S4) were analysed for differential protein expression and significant spots identified by mass spectrometry (Tables 2 and 3). Of the detectable total cell proteins, 12 differentially expressed proteins in 10 spots were identified ( Table 2). The changes were primarily in cytoplasmic enzymes (e.g. enolase) which, because the proteins fold in the cytoplasm, may reflect general responses to stress due to lack of fully functional SEQ0694 (see below). Four of these proteins were also detected in the cell-free supernatant proteins (Table 3). In the cell-free supernatant proteomes, 13 proteins in 17 spots were found to be differentially expressed. As expected, the majority of these are proteins predicted to be either secreted or cell envelope localised and because of this could be plausible substrates for SEQ0694 ( Table 3). As multiple proteins were found to be absent from the cell-free supernatant proteome of the mutant strain DprtM 138-213 , we hypothesise that SEQ0694 is likely to influence folding and secretion of multiple substrates rather than a specific substrate. Interestingly, two previously reported virulence factors of S. equi were notably absent from the cell-free supernatant proteome of the DprtM 138-213 mutant: the truncated fibronectin-binding protein FNE [40][41][42] and IgG endopeptidase IdeE2. 43 FNE is noted to be misannotated as a pseudogene in the strain 4047 genome 17 due to a misplaced start methionine. Our data therefore confirm the expression of FNE by strain 4047. SEQ0882, a putative DNase virulence factor homologous to S. pyogenes DNAse 44 was also absent from the cell-free supernatant proteome of the DprtM 138-213 mutant.
Cumulatively, these proteomic changes likely explain, at least in part, the attenuation of the DprtM 138-213 mutant. 19 Fig. 2 Kinetic analysis of rSEQ0694. The K cat for rSEQ0694 was determined to be 583.75 s À1 and the Km 100 mM. Calculated K cat /K m is 5.84 Â 10 6 M À1 s À1 . a Data from protease-coupled assays where substrate is a colourigenic tetrapeptide Succ-Ala-X-Pro-Phe-pNA in which X is the amino acid indicated in the Table. b Estimation from Fig. 1 in Heikkinen et al. 33 . c Subsequently Weininger et al. 49 have reported that PpiD is inactive as a PPIase using modified substrates in a protease-free assay. d Data from a protease-free assay using the tetrapeptide Succ-Ala-Ala-Pro-Phe-2,4-difluroanilide as substrate.
However, as the DprtM 138-213 mutant should still express a N + C domain fusion protein (lacking most of the parvulin domain), it may be that more dramatic proteome changes would be evident in an seq0694 null mutant, since a L. monocytogenes PrsA N + C construct partly complemented the proteome defect of a full prsA deletion 12 and an N + C fusion construct of B. subtilis PrsA partially restored secretion of an AmyQ reporter protein (although it did not restore viability to PrsA-depleted cells 27 ). In B. subtilis, the N and C domain is notable in driving dimerization of PrsA and, although lacking primary sequence homology, has structural similarity to other 'foldases' such as trigger factor. 26 Without structural characterisation of the N + C fusion encoded by the S. equi DprtM 138-213 mutant we cannot speculate whether this construct is likely to have a native-like conformation and functionality. However, it is notable that the sequence deletion removes not only the majority of the parvulin domain of SEQ0694 but also a conserved lysine of the Firmicutes PrsA protein N-domains. It is worth reemphasising that the partial deletion in the S. equi DprtM 138-213 mutant is sufficient to cause significant attenuation of virulence in the natural host. 19 It was interesting to note that SEQ1657, a cyclophilin PPIase lipoprotein (orthologous to S. pneumoniae SlrA 14 and L. lactis PpiA 29 ) was up-regulated in both the total cell and secreted proteins of the parental strain. Likewise, it was observed that the SEQ1171 sortase is up-regulated in the mutant strain, perhaps suggesting a need to remodel protein localisation within the mutant cell envelope.
As the proteomic data suggested a range of protein functions are likely to be perturbed in strain DprtM 138-213 , including stress responses, we performed several physiological tests. Although the mutant strain grows normally in nutrient rich broth, we observed pleiotropic changes including increased sensitivity to salt stress (ESI, † Table S1) and increased sensitivity to various antibiotics with diverse cellular targets (ESI, †  Fig. S5). Increased sensitivity to salt stress has previously been observed in a prsA mutant of E. faecalis 11 and a prsaA2 mutant  of L. monocytogenes. 30 A range of findings have been observed regarding antibiotic susceptibilities of other prsA mutants. Similar to our findings, a prsaA2 mutant of L. monocytogenes displayed increased sensitivity to bacitracin, penicillin and vancomycin but not gentamicin 30 and a mutant in Staphylococcus aureus prsA showed increased sensitivity to vancomycin. 45 However, a prsA mutant of E. faecalis was unaffected in its sensitivity to ampicillin and norflaxin, 11 in contrast to our findings. Cumulatively, our data suggest a general perturbation in cell envelope function in the DprtM 138-213 mutant, which likely reflects multiple changes in the extracytoplasmic proteome of the mutant (consistent with our proteomic data). This is conclusion is consistent with the pleiotropic effects of PrsA mutation in other Firmicutes. 11,28,30,45 Conclusions The data presented here confirm that rSEQ0694 is a moderately active PPIase, despite lacking conservation of several amino acids previously considered to be significant to the activity of other parvulin PPIases. This observation thus focusses attention on the conserved Asp/Cys identified as likely critical for catalysis. Furthermore, proteomic experiments confirm that loss of the lipoprotein PPIase activity in strain DprtM 138-213 affects multiple cell envelope proteins, including virulence factors, and is likely to generate diverse phenotypic effects. As strain DprtM 138-213 is attenuated, 19 these findings further suggest that streptococcal PPIases, and PPIases generally, 7 are interesting targets for novel therapeutic strategies. By analogy with other bacterial PPIases, it would also be of interest to determine whether the N-and C-terminal domains of SEQ0694 possess additional chaperone activities that contribute to post-translocational protein folding.