Structural and Functional Characterization of Cleavage and Inactivation of Human Serine Protease Inhibitors by the Bacterial SPATE Protease EspPα from Enterohemorrhagic E. coli

EspPα and EspI are serine protease autotransporters found in enterohemorrhagic Escherichia coli. They both belong to the SPATE autotransporter family and are believed to contribute to pathogenicity via proteolytic cleavage and inactivation of different key host proteins during infection. Here, we describe the specific cleavage and functional inactivation of serine protease inhibitors (serpins) by EspPα and compare this activity with the related SPATE EspI. Serpins are structurally related proteins that regulate vital protease cascades, such as blood coagulation and inflammatory host response. For the rapid determination of serpin cleavage sites, we applied direct MALDI-TOF-MS or ESI-FTMS analysis of coincubations of serpins and SPATE proteases and confirmed observed cleavage positions using in-gel-digest of SDS-PAGE-separated degradation products. Activities of both serpin and SPATE protease were assessed in a newly developed photometrical assay using chromogenic peptide substrates. EspPα cleaved the serpins α1-protease inhibitor (α1-PI), α1-antichymotrypsin, angiotensinogen, and α2-antiplasmin. Serpin cleavage led to loss of inhibitory function as demonstrated for α1-PI while EspPα activity was not affected. Notably, EspPα showed pronounced specificity and cleaved procoagulatory serpins such as α2-antiplasmin while the anticoagulatory antithrombin III was not affected. Together with recently published research, this underlines the interference of EspPα with hemostasis or inflammatory responses during infection, while the observed interaction of EspI with serpins is likely to be not physiologically relevant. EspPα-mediated serpin cleavage occurred always in flexible loops, indicating that this structural motif might be required for substrate recognition.


Introduction
Enterohemorrhagic Escherichia coli (EHEC) cause severe diseases in humans worldwide. Shiga toxins are regarded as their main virulence factor. However, EHEC possess various further virulence factors that mediate adherence or interfere with host defense [1,2]. One of these additional virulence factors is the plasmid-encoded extracellular serine protease EspP which belongs to the serine protease autotransporter of Enterobacteriaceae (SPATE) family [3]. Five subtypes of EspP have been described (EspPa-EspPe [4,5], from which the translocation-competent and proteolytically active subtype EspPa (Uniprot Accession Number: Q7BSW5) is associated with highly virulent strains and isolates from patients with severe disease [4,6]. EspPa exhibits serine protease activity. In addition to porcine pepsin A and EHEC-Hemolysin [3,7], EspPa cleaves the human plasma proteins apolipoprotein A-I, the complement factors C3 and C5, and coagulation factor V [3,8,9]. EspPa-mediated cleavage of complement factors has been demonstrated to significantly reduce complement activation [9]. In addition, the degradation of factor V has been suggested to interfere with blood coagulation possibly leading to prolonged bleeding during EHEC infection [3].
The E. coli secreted protease, island-encoded (EspI) is a further member of the SPATE family and is secreted by Shiga toxinproducing E. coli (STEC) [8]. Notably, EspI has been found in less pathogenic E. coli serotypes [8,10,11]. The physiological function of EspI is yet unknown and to date only two substrates have been identified, namely porcine pepsin A and human apolipoprotein A-I [8].
Serine protease inhibitors (serpins) are structurally closely related proteins which modulate different important protease cascades by irreversible inactivation of serine proteases. They are involved in inflammatory host defense, complement activation, and blood coagulation [12,13]. Serpins share an exposed reactive center loop (RCL) that serves as a pseudosubstrate for the target protease. Cleavage of the reactive serpin bond initiates a conformational rearrangement of the serpin structure that leads to distortion and inactivation of the target protease by formation of an irreversible covalent serpin-protease complex [14]. a1-protease Inhibitor (a1-PI, Uniprot Accession Number: P01009) is the archetypal member of the serpin family and the most abundant serpin in human plasma. Its main physiological target is neutrophil elastase [15]. a1-antichymotrypsin, (a1-AC, Uniprot Accession Number: P01011) which is closely related to a1-PI, [16,17] mainly inhibits cathepsin G and mast cell chymases [15,18]. a2antiplasmin (a2-AP, Uniprot Accession Number: P08697) is the main physiological inhibitor of plasmin and thus influences fibrinolysis following blood coagulation [19,20]. Antithrombin III (ATIII, Uniprot Accession Number: P01008) inhibits thrombin, FIXa, and FXa -proteases of the blood coagulation pathwaywhich is considerably faster in the presence of its cofactor heparin [21][22][23][24]. Angiotensinogen (AGT, Uniprot Accession Number: P01019) is a non-inhibitory serpin that does not target proteases [25]. Via proteolytic processing by renin, AGT releases the vasopressor peptide angiotensin I which is further converted to angiotensin II [26,27]. An overview of serpin functions and nomenclature is given in Table 1.
Notably, we describe here the specific cleavage of various serpins from human plasma by the bacterial serine protease EspPa and compare this activity with the related SPATE EspI. Presented data further support the hypothesis that EspPa mediates virulence by interaction with key regulatory proteins of host defense and blood coagulation. In addition, we developed a photometrical assay for the analysis of serpin activity and applied matrix assisted laser desorption ionization-time of flight-mass spectrometry (MALDI-TOF-MS) and electrospray ionisation-fourier transform mass spectrometry (ESI-FTMS) for the direct elucidation of proteolytic cleavage sites.

Materials and Methods
Pseudonymized residual sample material from voluntary blood donations from the Transfusion medicine of the University Clinics Münster was used. Blood donors approved prior to donation that residual sample material can be used for scientific studies. The Ethics Committee of the Medical Faculty of the University of Münster was informed and approved the study design.

Plasma fractionation
Plasma samples (fresh frozen plasma, FFP) were stabilized with 17-23% (v/v) citrate-phosphate-dextrose (CPD) and were derived from whole blood donations using standard separation procedures for blood banks.
Plasma was diluted with 20 mM Na 3 PO 4 buffer (pH 7.0) and depleted using HiTrap Protein A FF and HiTrap Blue HP (GE

Cleavage of Substrates
To determine cleavage of substrates by EspPa or EspI, fractionated plasma (25 mg) or serpins (5 mg or 10 mg) were incubated (15 h, 37uC) with 1.5 mg of purified protease in 30 mL PBS buffer. ATIII was incubated in the same way after addition of 25 mg/mL (4.8 units/mL) unfractionated heparin (Merck). Proteins were either separated via sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), digested in-gel and analyzed using matrix assisted laser desorption ionization-time of flight-mass spectrometry (MALDI-TOF-MS) or subjected directly to MS analysis.

SDS-PAGE
After denaturation, proteins were separated on a 7.5% SDS-PAGE gel using a glycine (19.2 mM) containing buffer [37] or on a 13.3% SDS-PAGE gel using a tricine (100 mM) containing buffer [38] and stained with Coomassie Blue.

Mass spectrometric analysis
In-gel-digests or incubation mixtures (0.5 mL) were mixed with 0.5 mL a-cyano-4-hydroxycinnamic acid (Sigma-Aldrich, 10 mg/ mL in 50% MeCN/1% trifluoroacetic acid) and 0.5 mL of the mixture were spotted on a MALDI target (MTP 384 target plate ground steel, Bruker). Samples were analyzed using a Bruker autoflex speed in positive mode.
To determine the accurate masses of the largest a2-AP fragment, the incubation mixture was desalted using ZipTip C 18 Pipette Tips as described before and measured using a Thermo

Determination of EspPa and a1-PI activity
Potential functional consequences of the interaction between a1-PI and EspPa were analyzed by measuring the activities of both proteins after coincubation. To investigate effects of a1-PI on EspPa protease activity, both proteins were incubated together (15 h, 37uC) at equimolar concentrations. Preincubated (15 h, 37uC) EspPa or a1-PI were used as controls. The remaining EspPa protease activity was then determined by the incubation (15 h, 37uC) of an aliquot containing 1 mg EspPa (either preincubated alone or with a1-PI) with 2 mM of the chromogenic peptide substrate Suc-Ala-Ala-Pro-Leu-pNA (Bachem) in 100 mL PBS (pH 7.4) and 5% dimethyl sulfoxide (DMSO). Active EspPa releases para-nitroaniline (pNA) from the peptide which is detected at 405 nm using a FLUOstar Optima plate reader (BMG Labtech). PBS was used as a buffer control.
The effect of EspPa-mediated cleavage on a1-PI serpin activity was determined by coincubation (15 h, 37uC) of a1-PI and EspPa in a molar ratio of 4:1. Again, incubations (15 h, 37uC) of EspPa or a1-PI alone were used as controls. To assess remaining serpin activity of a1-PI, the coincubation mixture and controls were incubated (5 h, 37uC) with trypsin (Promega) at a molar ratio of a1-PI and trypsin of 4:1. Active a1-PI inhibits trypsin protease activity. The remaining serpin activity was therefore assessed indirectly by determination of reduced trypsin activity using aliquots of coincubation mixtures and controls containing 0.25 mg trypsin and incubation (2 h, 37uC) with 2 mM of the chromogenic peptide Bz-Arg-pNA (Bachem) in 100 mL PBS (pH 7.4) containing 5% DMSO. Active trypsin releases pNA and absorbance was measured at 405 nm using a FLUOstar Optima plate reader. PBS was used as a buffer control.

Purification of EspPa and S263A
EspPa and the inactive EspPa mutant S263A were purified from culture supernatants using ammonium sulfate precipitation and liquid chromatography. Purity was verified via SDS-PAGE (Fig. 1, lane 5 and 6). EspPa shows a band at ,104 kDa representing the intact EspPa and a band at ,80 kDa which was identified by MALDI-TOF-MS as autoproteolysis product. S263A samples showed a pronounced protein band at ,104 kDa and a weaker band at ,85 kDa which was identified as a truncated form of S263A. The autoproteolyis product of EspPa remains active even after long term incubation ( Figure S1). Proteolytic activity of purified EspPa and the inactive S263A were assessed using a chromogenic oligopeptide substrate. As expected, all EspPa samples were proteolytically active while S263A showed no proteolytic activity (Fig. S1).

Identification of EspPa substrates in plasma
To identify physiological relevant substrates of EspPa, fractionated plasma was incubated either with EspPa or the EspPa negative control S263A (Fig. 1, lane 3 and 4). Incubation with EspPa resulted in loss of a pronounced 50 kDa band in plasma and the occurrence of a degradation product with a molecular weight of ,45 kDa in SDS-PAGE. The according protein band was digested in-gel and subjected to MALDI-TOF-MS analysis and unambiguously identified as a1-PI (Aldente score 235.7, sequence coverage 69% to a1-PI (UniProtKB: P01009)).

EspPa cleaves various serpins
To determine if further serpins are cleaved by EspPa, different serpins were incubated with EspPa or S263A and cleavage was monitored by SDS-PAGE. EspPa degrades a1-PI, a1-AC, and the were incubated (15 h, 37uC) with EspPa or S263A (1.5 mg). Degradation products were separated via SDS-PAGE using a glycine buffer (a, c, e, g, i) or tricine buffer system (b, d, f, h, j). Proteolytic serpin fragments formed by EspPa are indicated by an arrow. a, b a1-PI is degraded to a large and small fragment (,45 kDa and ,4 kDa, respectively), c, d cleavage of a1-AC in two fragments, e, f the AGT band with the highest molecular weight is cleaved in two fragments, g, h large and small fragments (,55-57 kDa and ,4-7 kDa) formed by a2-AP cleavage. i, j ATIII is not cleaved by EspPa. Incubation of a1-PI with EspPa leads to a weak formation of an inhibitory enzyme-serpin complex as marked by **. M, molecular weight marker, *, autodegradation product of EspPa. doi:10.1371/journal.pone.0111363.g002 non-inhibitory serpin AGT into a large (.40 kDa) and a small (, 10 kDa) fragment (Fig. 2 a-f), while incubation of a2-AP leads to several degradation products (Fig. 2 g, h). None of the incubations led to pronounced formation of an inhibitory serpin-enzyme complex. Interestingly, the anticoagulatory serpin ATIII was not degraded by EspPa (Fig. 2i, j). . Activity of EspPa and a1-PI after coincubation. a, Determination of EspPa activity. EspPa and a1-PI were preincubated (15 h, 37uC) at equimolar concentrations and remaining activity of EspPa was analyzed by incubation of an aliquot of the mixture with the chromogenic substrate Suc-Ala-Ala-Pro-Leu-pNA. Activity was measured via released para-nitroaniline and normalized to EspPa. n = 9 for EspPa and EspPa+a1-PI or n = 6 for a1-PI, respectively. b, a1-PI activity (measured as inhibitory potential on trypsin) after incubation with EspPa. a1-PI and EspPa or S263A were preincubated at a molar ratio of serpin:enzyme = 4:1. Remaining inhibitory activity of a1-PI on trypsin was analyzed by incubation at a molar ratio of a1-PI:trypsin = 4:1. Trypsin activity was measured via release of para-nitroaniline from the chromogenic substrate Bz-Arg-pNA. c, SDS-PAGE analysis of conincubations. a1-PI, EspPa, S263A, and trypsin were incubated as in b) and mixtures were separated via SDS-PAGE (12% SDS-PAGE gel, glycine buffer). M, molecular weight marker, *, EspPa autodegradation product, **, inhibitory complex of a1-PI and trypsin, +, trypsin was directly subjected to SDS-PAGE without incubation. doi:10.1371/journal.pone.0111363.g003

Activity of a1-PI and EspPa after incubation
We next determined the functional consequences of the coincubation of serpin and SPATE protease by use of the bona fide serpin a1-PI and EspPa. The remaining EspPa-activity following incubation with a1-PI was assessed in a photometrical assay using the chromogenic EspPa substrate Suc-Ala-Ala-Pro-Leu-pNA. Incubation with a1-PI had no influence on the proteolytic activity of EspPa (Fig. 3a), demonstrating that a1-PI does not target EspPa.  The remaining inhibitory potential of a1-PI following incubation with EspPa was analyzed using trypsin as a serpin target. Although neutrophil elastase is the physiological target for a1-PI, trypsin also forms an irreversible inhibitory complex with the serpin and can therefore be used as an indicator for a1-PI activity [40]. Active a1-PI inhibits the proteolytic activity of trypsin and consequently loss of a1-PI serpin activity results in high proteolytic activity in the assay. Trypsin activity was determined by photometrical detection of the cleavage of the trypsin substrate Bz-Arg-pNA.
Incubation of trypsin with a1-PI or a1-PI preincubated with S263A resulted in nearly complete loss of trypsin activity (Fig. 3b), demonstrating that the employed a1-PI shows high serpin activity and that the inactive EspPa mutant S263A does not affect a1-PI. In contrast, a1-PI preincubated with EspPa did not reduce trypsin activity in the following assay (Fig. 3b). This demonstrates that EspPa-mediated a1-PI cleavage leads to loss of the inhibitory serpin activity. Corresponding results were obtained using SDS-PAGE (Fig. 3c). Incubation of a1-PI with trypsin leads to the formation of a serpin-enzyme-complex (Fig. 3c, lane 10). After incubation with EspPa, a1-PI is not able to form this complex with trypsin. Instead, the large a1-PI fragment is further degraded by trypsin (Fig. 3c, lane 6). EspPa as well as S263A were completely degraded when incubated with trypsin, demonstrating that neither EspPa nor S263A directly interfere with trypsin activity (Fig. 3c,  lanes 2 and 4). In addition, a1-PI does not interact with S263A (no serpin enzyme complex) (Fig. 3c, lane 7) but is cleaved by EspPa (Fig. 3c, lane 5). The addition of trypsin to the mixture of a1-PI and S263A led to incomplete degradation and occurrence of several degradation bands in SDS-PAGE. This is due to the fact that degradation of S263A by trypsin and the inhibition of trypsin by a1-PI occur in parallel resulting in only incomplete S263A degradation (Fig. 3c, lane 8).

EspPa cleaves inside the reactive center loop
The loss of activity of a1-PI but not EspPa is based on cleavage of a1-PI without formation of an inhibitory serpin-enzymecomplex. To further understand how EspPa-mediated cleavage affects the inhibitory function, we determined the cleavage sites in a1-PI and the other serpins included in this study. To this end, large and small fragments of cleaved serpins were separated using SDS-PAGE, in-gel-digested and subjected to MALDI-TOF-MS analysis. Figure 4 shows the peptide mapping of EspPa cleavage products of a1-PI. The large a1-PI fragment consists of the Nterminal part of the serpin (Fig. 4a and b), while the C-terminal part from residue 383 to 418 forms the small fragment (Fig. 4 a,  and c). EspPa cleavage occurs at the active site of the serpin between 382 Met and 383 Ser as demonstrated by the occurrence of the non-tryptic peptide 19(SIPPEVK) and the complete sequence coverage for the small fragment (Fig. 4c). Sequence coverage of degradation products of the other serpins are given in Figure S2.

Direct MALDI-TOF-MS analysis of small fragments
Not all cleavage sites can be identified via in-gel-digest. Tryptic peptides might be too small when cleavage occurs close to lysine or arginine residues or when several cleavage sites are in close proximity to each other. As all small fragments formed by EspPacleavage show a molecular weight below 10 kDa, we applied direct MALDI-TOF-MS analysis to determine the exact mass of the small serpin fragments to elucidate and confirm cleavage sites (Fig. 5). For the small a1-PI fragment we observed a signal for the proton adduct of the a1-PI sequence 383 Ser-418 Lys (m/z 4133.333) confirming the cleavage site determined via in-gel-digest. In addition, signals representing the Na + adduct and the oxidized Na + adduct of the according a1-PI fragment sequence were observed (Fig. 5a). a1-AC shows a similar spectrum with a pronounced signal at m/z 4623.419 demonstrating cleavage Cterminal of 383 Leu at the reactive bond (Fig. 6b), which is in good accordance with data from in-gel-digest ( Figure S2). For AGT, we already observed three bands in SDS-PAGE (intact AGT and two non-proteolytic fragments) when incubated without protease (Fig. 2e). Accordingly, signals of two small AGT fragments were observed in MALDI-TOF-MS (Fig. 5c, right lane). Incubation with EspPa led to degradation of intact AGT and occurrence of the corresponding small fragment in MALDI-TOF-MS ( Fig. 2e and Fig. 5c, left lane). For a2-AP, proteolytic cleavage into several fragments is observed in SDS-PAGE (see Fig. 2g and 2h and Fig. 5d) after incubation with EspPa. Four distinct signals are seen in the MS spectrum indicating 4 cleavage sites. As the resolution for the signal at m/z 5308.3 is too low to determine the monoisotopic mass, we measured this sample in addition via nanospray-ESI-FTMS. Table 2 summarizes EspPa cleavage sites and their positions within the respective serpin. Measurement of a2-AP after incubation with EspPa via nanospray ESI-FTMS is described in Table 3.
a1-PI and a1-AC are cleaved at their reactive bonds (position of reactive sites are described in [41,42]), leading to loss of serpin function. In both molecules the reactive bonds are exposed in the RCL and serve as pseudosubstrates for the targeted proteases. In case of EspPa, the serpins are not able to form a stable inhibitorenzyme-complex and therefore release the intact EspPa after cleavage. Although AGT as non-inhibitory serpin does not contain a reactive bond, it is structurally closely related to the other serpins and is also cleaved in the RCL, indicating that a reactive bond is not necessary for EspPa-mediated serpin degradation. This is further underlined for a2-AP, which is cleaved at four positions outside the RCL (for RCL position see [43]). Cleavage sites are located at the Nand C-terminal extensions 25 aa downstream the N-terminus and 46, 31, and 30 aa upstream the C-terminus (see Table 2). Intriguingly, both the Nand C-terminal extensions are vital for the functional relevant binding of a2-AP to other proteins [19,44,45].

Cleavage of serpins by EspI
Purified EspI samples showed a protein band at ,110 kDa (intact EspI) as well as two EspI autoproteolysis products at ,50 and 45 kDa, respectively. Similar to EspPa, autoproteolysis products remain active. Serpins were incubated with purified EspI in the same way as described for EspPa. Incubation of a1-PI and a1-AC with EspI led to degradation of these serpins. Notably, EspI also forms a pronounced inhibitory complex with both protease inhibitors resulting in only incomplete serpin degradation (Fig. 6 a-d). In contrast to EspPa, EspI does not cleave a2-AP and AGT (Fig. 6e-h). Cleavage of ATIII occurred only with very low efficiency (Fig. 6i) and might not be relevant under physiological conditions.
To determine the cleavage sites of a1-PI and a1-AC, we subjected incubation mixtures of serpins and EspI to direct MALDI-TOF-MS analysis. Serpin cleavage occurred at the reactive bond leading to signals at m/z 4155.400 (a1-PI, 20 ppm deviation according to calculated m/z) and 4623.509 (a1-AC, 19 ppm deviation according to calculated m/z), respectively (data not shown).

Conclusions
EspPa is an EHEC virulence factor that belongs to the SPATE family. As suggested for SPATEs in general, EspPa most likely mediates its virulence via cleavage and inactivation of host proteins. Here, we present a method for the rapid determination of EspPa-mediated cleavage sites in various human plasma serpins via MALDI-TOF-MS as well as a photometrical assay to analyze serpin functionality after proteolytic cleavage. Concerning the functional consequences, degradation of a2-AP might lead to bleeding disorders. This serpin is the primary physiological inhibitor of plasmin and deficiency has been shown to result in uncontrolled fibrinolysis and severe hemorrhagic complication [44,45]. a2-AP harbors a 42 aa N-terminal and a 55 aa C-terminal extension [19,46]. While the N-terminal extension is cross-linked to fibrin, the very C-terminal 491 Lys residue mediates binding to plasmin [47]. EspPa cleaves between 47 Leu and 48 Thr releasing part of the N-terminal extension and at three different sites inside the C-terminal extension leading to release of a polypeptide containing 491 Lys. Together, this most likely leads to loss of function of a2-AP. The role of a1-PI in thrombosis is not well understood. However, a1-PI is able to inhibit activated protein C. In pediatric ischemic stroke patients elevated levels of a1-PI have been found and were discussed to contribute to this thrombotic disease in children [48,49]. ATIII is the main anticoagulatory serpin. Although it is able to interfere with virtually all proteolytic coagulation factors, its main targets are thrombin, FIXa, and FXa. Intriguingly, it is the only serpin in this study that is not cleaved by EspPa. Despite the structural similarity of serpins, EspPa specifically cleaves only selected serpins. More specific, procoagulatory serpins such a2-AP and a1-PI are efficiently degraded while the anticoagulatory ATIII is not affected at all. Together with data demonstrating that EspPa cleaves coagulation factor V [3], this underlines the hypothesis that interference with blood coagulation (and possibly also inflammatory host responses) [50] might be one of the major functions of EspPa which might contribute to formation of hemorrhages observed during EHEC infection.
Having a closer look at EspPa cleavage sites, it is notable that more than 70% (5 of 7) of cleavage sites identified in this study occur after Leu. This is in good accordance to already reported EspPa cleavage sites [3,9,7,51], indicating that substrate cleavage is most favorable C-terminal to Leu. In a2-AP, cleavage also occurs after 459 Ser. This residue, however, is positioned next to 460 Leu after which EspPa cleaves, too. The second non-Leu cleavage site is C-terminal to 382 Met in a1-PI. The 382 Met-383 Ser bond, however, is the reactive bond exposed in the RCL and required to react with target proteases. Similarly, a1-AC is cleaved at the reactive bond that consists of a Leu-Ser motif which is also located in the exposed RCL. Cleavage of the non-inhibitory AGT shows that a reactive bond is not strictly required for substrate recognition by EspPa but cleavage also occurs inside the corresponding reactive center loop. In contrast, a2-AP is not cleaved in the RCL but inside the Nand C-terminal extensions which are vital for a2-AP functionality. Though the crystal structure of a2-AP has only been solved for a N-terminally truncated murine form, it seems that the C-terminal extension consists of a flexible loop because it could not be modeled into electron density maps [52]. Perhaps, this structural flexibility seen in the reactive center loops and in the C-terminal extension of a2-AP is required for substrate recognition by EspPa. Figure 7 shows crystal structures of the serpins that are cleaved by EspPa [52][53][54][55].  5 mg). Degradation products were separated via SDS-PAGE using a glycine buffer (a, c, e, g, i) or a tricine buffer (b, d, f, h, j). a, b a1-PI is cleaved into two fragments (,45 kDa and ,4 kDa), c, d a1-AC is cleaved into two fragments, e, f AGT is not cleaved by EspI, g, h a2-AP is not cleaved by EspI, i, j ATIII is cleaved only with very low efficiency. Note the formation of inhibitory serpinenzyme-complexes after incubation with a1-PI and a1-AC. M, molecular weight marker, *, autodegradation product of EspI, **, inhibitory serpin-EspI-complex. Serpin fragments are indicated by an arrow. doi:10.1371/journal.pone.0111363.g006 EspI shows significant differences in substrate specificity compared to EspPa. a1-PI and a1-AC are also cleaved at their reactive bonds which should lead to loss of function of these serpins. However, serpin cleavage and release of the protease is not complete for EspI, most probably due to the pronounced formation of an inhibitory serpin-enzyme-complex of EspI with a1-PI and a1-AC. In contrast, EspPa completely degrades both serpins and forms only small amounts of the inhibitory complex only with a1-PI which does not significantly reduce EspPa activity. In addition, AGT and a2-AP, which are degraded by EspPa at positions other than the reactive bond, are not degraded by EspI. Concerning the functional differences of both SPATE proteases, EspPa is able to cleave serpins specifically within accessible loop structures and is notably not inhibited by the analyzed serpins, while EspI is only able to interact with the reactive bond of a1-PI and a1-AC. The latter interactions show equilibria between EspI inhibition and serpin degradation. Taking into account the high amounts of serpins such as a1-PI in plasma, EspI activity might be strongly reduced in this milieu in vivo, while serpin degradation and inactivation might be a relevant function of EspPa also during infection.
In summary, we established a rapid method to determine cleavage sites of small proteolytic fragments via MALDI-TOF-MS. Functional implications have been investigated in a newly developed photometrical assay using chromogenic peptide substrates. EspPa degrades and thereby inactivates different plasma serpins which, in case of a2-AP, might lead to bleeding disorders or in case of a1-PI and a1-AC might interfere with the acute phase reaction during inflammatory host response. Cleavage occurs in flexible regions most favorable C-terminal to Leu. Comparison of EspPa and EspI indicate different functions of this SPATE also in vivo. Figure S1 Activity of EspPa and S263A. a, Determination of EspPa and S263A activity directly after purification. EspPa or S263A was incubated (15 h, 37uC) with the chromogenic substrate Suc-Ala-Ala-Pro-Leu-pNA. Activity was measured via released para-nitroaniline and normalized to EspPa. PBS was used as control. n = 2, b, Determination of EspPa activity after preincubation. Purified EspPa was preincubated for 15 h at 37uC resulting in the formation of autoproteolysis products (see Fig. 3c, lane1). To assess remaining proteolytic activity of autoproteolysis products the preincubated sample was incubated with the chromogenic substrate Suc-Ala-Ala-Pro-Leu-pNA (15 h, 37uC). Again, activity was measured via released para-nitroaniline and normalized to EspPa. PBS was used as control. n = 2. (TIFF) Figure S2 Peptide mapping of EspPa cleavage products of the serpins. Serpin fragments were subjected to in-gel-digest and analyzed via MALDI-TOF-MS. Peptides of the large fragment are given in bold. Peptides of the small fragments are given in italics, a, sequence coverage of a1-AC fragments, b, sequence coverage of AGT, c, sequence coverage of a2-AP. Note that in the small fragments of AGT and a2-AP no serpin peptides were found. (TIF)

Supporting Information
Author Contributions