Ecotin, a microbial inhibitor of serine proteases, blocks multiple complement dependent and independent microbicidal activities of human serum

Ecotin is a serine protease inhibitor produced by hundreds of microbial species, including pathogens. Here we show, that ecotin orthologs from Escherichia coli, Yersinia pestis, Pseudomonas aeruginosa and Leishmania major are potent inhibitors of MASP-1 and MASP-2, the two key activator proteases of the complement lectin pathway. Factor D is the key activator protease of another complement activation route, the alternative pathway. We show that ecotin inhibits MASP-3, which is the sole factor D activator in resting human blood. In pathway-specific ELISA tests, we found that all ecotin orthologs are potent lectin pathway inhibitors, and at high concentration, they block the alternative pathway as well. In flow cytometry experiments, we compared the extent of complement-mediated opsonization and lysis of wild-type and ecotin-knockout variants of two E. coli strains carrying different surface lipopolysaccharides. We show, that endogenous ecotin provides significant protections against these microbicidal activities for both bacteria. By using pathway specific complement inhibitors, we detected classical-, lectin- and alternative pathway-driven complement attack from normal serum, with the relative contributions of the activation routes depending on the lipopolysaccharide type. Moreover, in cell proliferation experiments we observed an additional, complement-unrelated antimicrobial activity exerted by heat-inactivated serum. While ecotin-knockout cells are highly vulnerable to these activities, endogenous ecotin of wild-type bacteria provides complete protection against the lectin pathway-related and the complement-unrelated attack, and partial protection against the alternative pathway-related damage. In all, ecotin emerges as a potent, versatile self-defense tool that blocks multiple antimicrobial activities of the serum. These findings suggest that ecotin might be a relevant antimicrobial drug target.


Introduction
Ecotin is a canonical serine protease (SP) inhibitor first isolated from Escherichia coli [1]. Three unique features provide ecotin with unusually broad specificity, yet high affinity. Ecotin has: i) a "one-size-fits-all" methionine P1 residue [2] acceptable for the S1 pocket of many different SPs; ii) a peculiar binding mechanism whereby the ecotin homodimer "chelates" two SPs, each being tweezed between the primary binding site of one monomer and the secondary binding site of the other one [3,4] and iii) structural plasticity [5] enabling accommodation to a large set of SPs having different binding surfaces. Ecotin inhibits all three major pancreatic SPs, trypsin, chymotrypsin and elastase, and its function was first assumed to protect E. coli in its natural habitat, the colon [1]. Later, ecotin was shown to inhibit several plasma SPs, such as activated coagulation factor X (fXa) [6] and activated coagulation factor XII (fXIIa), as well as plasma kallikrein [7], but none of these enzymes were considered as physiologic targets.
Since then, ecotin orthologs were found in several microbes including human pathogens, such as Yersinia pestis¸Pseudomonas aeruginosa and Burkholderia pseudomallei [8,9] and even in eukaryotic organisms such as the pathogenic protozoa Trypanosomatida, including Leishmania major [10]. Ecotin orthologs from E. coli, Y. pestis, and P. aeruginosa were shown to inhibit neutrophil elastase (NE) secreted by neutrophils and macrophages during inflammation, and this activity was interpreted as a potential defense mechanism [11].
The complement system (CS) belongs to the humoral arm of the innate immune system and it is among the first defense lines against pathogenic microbes. It can be activated through three pathways, the classical (CP), the lectin (LP) and the alternative pathway (AP) [12]. The activity of all three pathways rely on specific SPs. Key enzymes of the LP are mannan-binding lectin (MBL)-associated serine protease (MASP)-1 and -2 [13], while MASP-3 is responsible for the activation of pro-factor D (pro-FD), the zymogen of the key AP-activating enzyme, factor D (FD) [14]. Activation of the LP and the AP is independent from the slowly developing adaptive immune response, therefore these two pathways can unleash an immediate attack the support provided from the National Research, Development and Innovation Fund of Hungary, financed under the 2018-1.2.1-NKP funding scheme. JD was supported by the grant KH130376 of the National Research, Development and Innovation Office (Hungarian Scientific Research Fund), https://nkfih.gov.hu GP and DS were supported by the European Union and the State of Hungary and co-financed by the European Regional Development Fund within the projects No. VEKOP-2.3.2-16-2017-00014, https://ec.europa. eu/regional_policy/en/funding/erdf MJ and NS were supported by the European Union and the State of Hungary and co-financed by the European Regional Development Fund within the projects No. VEKOP-2.3.3-15-2017-00021, https://ec.europa. eu/regional_policy/en/funding/erdf/ MJ was supported by the Institutional Excellence Program of the Ministry of Human Capacities of Hungary (20460-3/2018/FEKUTSTRAT), https://www. kormany.hu/en/ministry-of-human-resources The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. against invading microbes [15]. Accordingly, LP-and AP-inactivating capacity could provide the pathogens with substantial advantage during colonization of the host.
Interestingly, although the CS is a major, SP-dependent antimicrobial defense arm [15,16], ecotin has not been assessed as a CS-inhibitor. We studied the mechanism of action of ecotin in the past [17,18] and are investigating the molecular mechanisms of complement activation in the present [14,19,20]. When the crystal structures of our in vitro evolved, monospecific Schistocerca Gregaria Protease Inhibitor-related MASP inhibitors, (SGMI)-1 and SGMI-2 were solved in complex with MASP-1 and MASP-2, respectively [21], it revealed that the SGMIs and ecotin share analogous intramolecular interactions stabilizing the protease binding loop. Therefore, we tested E. coli ecotin as a LP inhibitor, and this has subsequently lead to the comprehensive study described here. This revealed that recombinant E. coli, Y. pestis, P. aeruginosa and L. major ecotin orthologs inhibit MASP-1-, MASP-2 and MASP-3 with varying relative potencies, while all four ecotin orthologs block LP-activation in normal human serum (NHS) with high and equal efficiency. We also demonstrated that E. coli ecotin inhibits LPactivation in mouse and rat sera as well.
Factor D (FD) is the initiator of the AP, and we have recently revealed that MASP-3 is the exclusive activator of pro-FD in resting human blood [14,22]. Here we show that E. coli ecotin readily blocks MASP-3-driven pro-FD activation. Using wild type E. coli cells and their derivatives lacking the eco gene (ecotin KO), we also show that endogenous ecotin protects bacteria against the attack of the LP and the AP, as well as against a complement-independent, heat resistant antimicrobial mechanism of heat-inactivated serum (HIS). All these findings unequivocally demonstrate that after 36 years of its discovery, ecotin emerges as a potent microbial defense factor modulating the immune response of the host by targeting several key proteases of the immune system. Accordingly, ecotin could be considered as a relevant antimicrobial drug target.

Results
Binding affinities of the four ecotin orthologs to the catalytic fragments of MASP-1, MASP-2 and MASP-3 have been determined in the form of dissociation constants, which, for reversible inhibitors are referred to as equilibrium inhibition constant (K I ) values. The values shown in Table 1 were measured as explained briefly below. In a series of samples increasing concentration of inhibitor was mixed with a preset concentration of enzyme, and after reaching binding equilibrium, the free enzyme concentration was determined through measuring the residual enzyme activity by adding an appropriate substrate. The obtained dose response curve was fitted to the appropriate kinetic model as explained in the Materials and methods section. The lower the K I value the higher the affinity is, and the differences can span orders of magnitudes. Therefore, for easier comparison of the affinities, we also illustrated them as -log 10 of the K I values in Fig 1. Some of the measurements delivered apparent K I values. For calculating the corresponding genuine K I values, we determined the K M values for each MASP enzyme-substrate pair (Fig 2.) (see Materials and methods). applied was below the K I determined in this study, therefore there is no conflict between the two observations.

Ecotin orthologs are potent MASP-2 inhibitors
Ecotin orthologs are potent inhibitors of MASP-2 with K I values in the 10 −8 -10 −9 M range (Fig 1, Table 1). E. coli ecotin inhibits MASP-2 with a K I of 11 nM, which is close to the SPRbased K D value of 24.7 nM reported by Gaboriaud et al. [23]. Importantly, most ecotin orthologs are as good, or even better MASP-2 inhibitors than SGMI-2, a monospecific, high-affinity inhibitor of the enzyme we developed previously via directed evolution [21].

Ecotin orthologs are highly potent MASP-3 inhibitors
Except Y. pestis ecotin, which has a K I value of 8.0×10 −7 M, the other three orthologs are highly potent inhibitors of MASP-3 with K I values in the 10 −9 -10 −10 M range (Fig 1, Table 1). E. coli ecotin, with its 0.5 nM K I value, is a~20-fold more efficient MASP-3 inhibitor than TFMI-3, our monospecific MASP-3 inhibitor developed by phage display [14]. Previously, no natural MASP-3 inhibitors have been identified. Our findings suggest that by inhibiting MASP-3, ecotin expressing microbes could disrupt the activatory connection between the LP and the AP [14]. The 3-4 orders of magnitude higher inhibitory potency of the tested ecotin orthologs on MASP-2 and MASP-3 versus MASP-1 might reflect the higher evolutionary conservation of MASP-2 and MASP-3, which are present in all vertebrates, while MASP-1 is apparently missing from birds and most fish [24][25][26].
It is worth noting that, by using SPR, Gaboriaud et al. measured a much weaker, 600 nM affinity between E. coli ecotin and MASP-3 [23]. An important difference between the two experiments is that they used a single MASP-3 SP domain construct, which lacked even the SP-structure stabilizing activation peptide. In this study, we used a MASP-3 catalytic fragment construct containing two complement control protein (CCP) modules connected to the SP domain through the activation peptide. The observed thousand-fold difference in affinities  Table 1 were measured at thermodynamic equilibrium as described in the Materials and methods. For tight-binding inhibition assays, enzyme and inhibitor concentrations need to be low, close to the K I value, and reaching the equilibrium requires several hours-long incubation. Then, small synthetic substrates are added, and their conversion rates report on the concentration of the free enzyme. In these short readout time assays the small substrates compete only with the relatively high-off-rate inhibitors and in this case the measured apparent K I value needs to be corrected with K M (see K M values in Fig 2). On the other hand, small substrates do not compete with the low off-rate tight-binding inhibitors for enzyme binding.

Inhibitor
KI (M) Data are the average of at least two independent measurements ± SEM. N.D. No inhibition could be detected. a should correspond to the more stable, native structure of the MASP-3 SP domain in our experiments.

E. coli ecotin inhibits MASP-3 through a "slow-binding" mechanism, and blocks MASP-3-driven pro-FD conversion in vitro
The ecotin-MASP K I values in Table 1 were measured at thermodynamic equilibrium using small synthetic substrates. These values do not necessarily predict how efficiently ecotin would compete with pro-FD, the large natural substrate of MASP-3, which could bind the enzyme stronger than small substrates. To test this, first we applied a chromatography-based assay using purified recombinant proteins [22] and measured how E. coli ecotin hinders pro-FD conversion. Low sensitivity of the method required high enzyme and inhibitor concentrations, which facilitate the rapid attainment of equilibrium. We therefore applied short incubation time. While ecotin inhibited the pro-FD conversion in a concentration-dependent manner, it was only slightly more efficient than TFMI-3, which has 20-fold lower affinity (Fig 3). Extending the pre-incubation time between MASP-3 and ecotin significantly increased the binding affinity, which, in agreement with the 0.5 nM equilibrium K I value (Table 1), resulted in a stoichiometric linear titration (Fig 4). This incubation time dependence revealed a slow-binding complex formation mechanism. Slow-binding, which is quite common for high-affinity complexes, suggests that the structure of the enzyme and/or the inhibitor needs to go through a slow conformational change during the maturation of the complex.

E. coli ecotin inhibits activation of pro-FD in NHS
We also tested how E. coli ecotin inhibits pro-FD conversion in NHS using an assay format already described [14,22]. Briefly, the assay was developed for in situ detection of the pro-FD converting capacity of plasma or serum samples. As these complex biological samples contain many different proteins including endogenous pro-FD and FD, the assay relies on recombinant, fluorescently labeled pro-FD added to the plasma or serum sample. The kinetics of the conversion was followed by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) combined by measuring the fluorescence intensity of the pro-FD and/or FD band. For testing how ecotin can block endogenous pro-FD conversion, the inhibitor was pre-incubated with the serum for 10 minutes before labeled pro-FD was added. In accordance with the previously mentioned slow-binding mechanism, thermodynamic equilibrium between MASP-3 and ecotin could not be reached, and in the first hour, the rate of pro-FD activation was almost unaffected. However, pro-FD activation was almost completely inhibited after~6 hours indicating formation of the high-affinity equilibrium ecotin-MASP-3 complex. While 1 μM TFMI-3 prolonged the 3 hour serum half-life of pro-FD to~40 hours [14], 1 μM of the slow-binding E. coli ecotin prolonged it to 10 hours (Fig 3).

E. coli ecotin is a moderate inhibitor of FD and of the AP-type C3 convertase, C3bBb
We tested whether E. coli ecotin inhibits FD and/or C3bBb, the two activator proteases of the AP. FD-inhibition was measured in a simple photometric enzyme-inhibition assay using the small molecule synthetic substrate, Z-Lys-SBzl (see Materials and methods), and the corresponding K I value is 4.06 μM.
Combined inhibitory activity of ecotin on FD and C3bBb was tested in an SDS-PAGE based densitometry-analyzed C3-cleavage assay, and it was compared to that of the small molecule pan-specific complement inhibitor, FUT-175. In this assay, isolated FB and C3b serve as components for the formation of the pro-convertase, C3bB, isolated C3 serves as substrate, and C3bBb formation and C3-clevage is initiated by adding isolated FD to the mixture containing no inhibitor (positive control), 10 μM ecotin, or 100 μM FUT. In the positive control approximately 45% of the C3 was cleaved to C3b and C3a in the first 10 minutes (Fig 5A and  5E), after which no further cleavage occurred due to the short half-life of C3bBb. Therefore, we quantified and compared the C3b-contents of the 10-minute samples (Fig 5F). At 10 μM concentration E. coli ecotin reduced C3b-formation by 20% (Fig 5C and 5F), while 100 μM FUT-175 exerted 80% inhibition (Fig 5D and 5F).

E. coli ecotin is a weak inhibitor of C1s while it is inactive on C1r
By using catalytic fragments of C1s and C1r and their appropriate synthetic substrates, we tested how ecotin inhibits these key enzymes of the CP. We could not detect any inhibition of C1r, while ecotin inhibited C1s with a 25.15±4.45 μM K I value representing a very weak inhibition. These findings are in agreement with previous SPR-based results of Gaboriaud et al. [23] who immobilized ecotin on the chip and by injecting up to 2 μM C1s or C1r catalytic fragments, did not detect any interaction.

Effects of ecotin orthologs on the three complement pathways in NHS
We measured the inhibitory potency of ecotin variants on the three individual CS pathways using NHS.
In accordance with their MASP inhibition profile, ecotin orthologs readily blocked mannan-triggered LP-driven C3-deposition in 2% NHS (Fig 6A). Our second-generation in vitro evolved MASP-2 inhibitor, SGMI-2 [13,21] was used as an internal reference to assess relative  Table 1) in a small synthetic substrate-based in vitro enzyme-inhibition test, where enzyme and inhibitor reached thermodynamic equilibrium. We measured how efficiently ecotin inhibits MASP-3 under non-equilibrium conditions by competing with pro-FD, the natural substrate of the enzyme. We found that ecotin inhibits MASP-3 driven pro-FD conversion in a concentration-dependent manner (■), but only slightly more effectively than the 20-fold lower affinity MASP-3 inhibitor TFMI-3 (▲) (see Table 1). After pre-incubating ecotin with MASP-3 to approach thermodynamic equilibrium, the interaction of the two proteins became stronger and stoichiometric (•) in accordance with the subnanomolar K I value (see Table 1). This suggests that the interaction reaches equilibrium slowly. Data points are the average of two independent measurements. Error bars represent the SEM. Dotted lines are used to visually emphasize the trends. LP-inhibitory efficiencies of the inhibitors. The IC 50 value range of the ecotin orthologs is 84-190 nM, the highest value being essentially identical with that of SGMI-2 (194 nM) ( Fig 6D). Overall, this is in accordance with the K I values of the inhibitors towards MASP-2. While the IC 50 range is narrow, it is notable, that the most efficient LP inhibitor is P. aeruginosa ecotin, which has the best-balanced MASP inhibitory profile. While the other three orthologs are stronger MASP-2 inhibitors, they are significantly weaker MASP-1 inhibitors. Apparently, inhibitors having moderate affinities toward both MASP-1 and MASP-2, can be similarly or more efficient LP inhibitors than more specific, high-affinity MASP-2 inhibitors. We observed the same phenomenon for our first-generation sunflower trypsin inhibitor (SFTI) related MASP inhibitor (SFMI) compounds [27]. All ecotin orthologs provided complete LP-inhibition in the low micromolar range.
LPS-or zymosan-triggered, AP-driven C3-deposition was measured in 16.7% (6-fold diluted) NHS representing a serum concentration about 8-fold higher than the one used for the CP-and LP-activation assays. All ecotin orthologs exerted potent AP-inhibition in this test, but required a higher, 10 μM inhibitor concentration (Fig 6F). This result is in agreement Approximately 45% of C3 was cleaved when no inhibitor was present (positive control) (A). When no FB and FD were present, no C3-cleavage occurred (negative control) (B). In (C) the sample contained 10 μM E. coli ecotin, while in (D) 100 μM FUT. As the AP-type C3bBb C3-convertase has a short half-life, no C3-clevage occurs after 10 minutes (E), hence, we compared the amounts of C3b generated in 10 minutes after adding FD (F). At 10 μM, E. coli ecotin decreased the production of C3b by 20% (C, F), while at 100 μM, FUT, the small-molecule general complement-inhibitor, decreased C3b-production by 80% (D, F). https://doi.org/10.1371/journal.ppat.1008232.g005 with the above mentioned synthetic substrate based and SDS-PAGE based observations, where ecotin was shown to be a micromolar FD-inhibitor, and a moderate inhibitor of de novo C3bBb production as well as of C3bBb-driven C3b production ( Fig 5F).
As we showed that E. coli ecotin is a very weak C1s inhibitor, which does not inhibit C1r at all, we already anticipated that this ecotin ortholog will not inhibit the CP. This was verified and we also found that none of the ecotin orthologs inhibited IgG-triggered CP-driven C3-deposition in 2% NHS (Fig 7). This indicates that the other ecotin orthologs are also ineffective against C1s and C1r.

Ecotin orthologs do not inhibit the C4b2a-type C3 convertase
Potent LP-inhibitory activity of all tested ecotin orthologs demonstrated that ecotin remains functional in NHS i.e. it is not degraded by any serum protease. As the CP-and the LP-inhibitory ELISA test conditions differ only in the immobilized target, all ecotin orthologs must be Ecotin is shown to inhibit LP-and AP-activation in human, and LP-activation in mouse and rat serum. C3-deposition on mannan-coated plates from dilute sera was measured as a function of inhibitor-concentration. All four tested ecotin orthologs are equally potent inhibitors of LPactivation in 50-fold diluted NHS. Their effects are comparable to that of our previously evolved selective MASP-2 inhibitor, SGMI-2. (A) E. coli ecotin inhibits the LP in analogous tests using 60-fold diluted pooled sera of C57BL/6 and BALB/c mice. (B), and 60-fold diluted serum of Wistar rat (C). IC 50 values for human and rodent serum are listed in (D) and (E), respectively. In the about 10-fold more concentrate, 6-fold diluted NHS used for the alternative pathway inhibition tests, 10 μM of the ecotin orthologs provide ortholog-specific, but significant inhibition, L. major ecotin being the most potent one (F). Symbols represent the average of at least two measurements. Error bars represent the SEM. present in the CP-assay in functional form. Lack of CP-blocking activity therefore clearly demonstrates that ecotin orthologs are inactive against the C4b2a-type C3 convertase.

E. coli ecotin is a potent LP inhibitor in rat and mouse serum as well
To test whether ecotin can block mannan-triggered LP-activation in other mammalian species as well, we used pooled sera of C57BL/6 and BALB/c mice and an individual serum sample of a Wistar rat, and tested the inhibitory effect of E. coli ecotin in C3-deposition ELISA tests. E. coli ecotin proved to be effective LP inhibitor in all three tests with IC 50 values of 16 nM, 30 nM and 16 nM in C57BL/6 or BALB/c mice and Wistar rat, respectively ( Fig 6B, 6C and 6E). The results reveal that the physiologic and pathologic roles of LP inhibitory activity of ecotin can be readily tested in these animal models. Moreover, cross-species activity of ecotin suggests an ancient role in microbial defense that could manifest in broad phylogenic groups of hosts. Ecotin protects bacteria against human serum-mediated lysis

Endogenous ecotin protects bacteria in NHS against C3-deposition, membrane attack complex (MAC) formation and MAC-mediated cellkilling
We tested how endogenous ecotin protects bacteria against complement-mediated attack of NHS by comparing wild type and ecotin KO variants of two E. coli strains (ATCC 23505 and ATCC 12014) having different lipopolysaccharide (LPS) surfaces. Briefly, wild type or ecotin KO bacteria were incubated in 1-16% NHS for 30 min and deposition of activated C3 fragments and C5b-9 were measured by flow cytometry. Gating strategy is shown in a representative case in Fig 8. Cell death was also monitored by propidium iodide (PI) staining and proliferation-based viability tests.
The O-antigen of the ATCC 23505 strain is a mannan homopolymer [28], which activates the complement LP [29]. On the other hand, the more typical O-antigen (O55) of the smooth LPS of ATCC 12014 is composed of repeating pentasaccharide units lacking mannan groups but containing several acetylated carbohydrate components [30].
C3 and C5b-9-deposition was significantly higher on the ecotin KO versions of the two strains in terms of the mean fluorescence intensity (MFI) values of the C3 positive population and the ratio of C5b-9/PI positive cells (Fig 9A, 9C, 9D and 9F). The percentage of C3 positive cells was also higher in the ecotin KO versions of the two strains at lower serum concentrations. At the highest NHS concentration (16%) the percentage of C3 positive cells reached a plateau in both the wild type and ecotin KO samples. In further experiments and analyses the C3 MFI values of the C3 positive cells were determined and compared. In above 4% NHS, the lysis of the ecotin KO cells was so intense, that large amount of DNA was detected in the cell supernatant (Fig 10), and the ratio of the C5b-9/PI positive cells topped or even decreased Ecotin protects bacteria against human serum-mediated lysis ( Fig 9C and 9F) indicating that the highest-labeled population disintegrated. In contrast, on the wild type cells, only low level of C3b and C5b-9 was detected up to 8% NHS for the ATCC 12014 strain and up to the highest tested NHS concentration of 16% for the ATCC 23505 strain suggesting a highly efficient protective role of ecotin enabling only a sub-lethal level of Endogenous ecotin protects E. coli against C3 and C5b-9-deposition from NHS. The extent and the initiating pathway of CS activation were studied on two wild type and ecotin KO coli strains, the polymannan LPS carrying ATCC 23505 and the smooth LPS containing ATCC 12014. Cells were treated with increasing concentrations of NHS and labeled for deposited C3 fragments, and C5b-9 as described in Materials and methods. Mean fluorescence intensity (MFI) values of C3 positive cells (A, D) and percentage of C5b-9/PI positive cells inside the bacteria FSC-SSC gate were determined and is shown (C, F). Cells were treated with 2% NHS containing the indicated inhibitors, and the MFI of deposited C3 was determined as for panel A and D. MFI of deposited C3 without inhibitor was considered 100% complement activity (B, E). Data represent median +/-SD of three independent experiments. MFI of the C3-positive and C5b-9/PI positive ATCC 23505 ecotin KO cells was orders of magnitude higher than that of the wild type cells (A, C). This strain should trigger the LP, and indeed, SGMI-1 and SGMI-2, inhibiting MASP-1 and MASP-2, respectively, as well as exogenously added ecotin provided complete inhibition, while the anti-C1q mAb had no effect. The broad-specificity protease-inhibitor, FUT-175 was included as a control (B). Similar trends were found with the ATCC 12014 cells. This strain is less intensely attacked by the complement, and while it is also protected by its endogenous ecotin, the ratio of C3 MFI values and percentage of C5b-9/PI positive cells are smaller (D, F). All exogenously added inhibitors decreased C3-deposition on ATCC 12014 ecotin KO cells, but SGMI-1 and anti-C1q mAb were the most efficient. Therefore, this strain was attacked by all three, but dominantly by the classical and the alternative pathway (E).
https://doi.org/10.1371/journal.ppat.1008232.g009 MAC formation (Fig 9C and 9F). In fact, this low level MAC formation could be of utmost importance as it could open an exit route for the periplasmic ecotin through the outer membrane to reach the site of the surface-localized complement attack. The observed difference in the two wild type strains is apparently due to the different relative contributions of the three CS pathway to the attack, as explained below.
To assess which CS pathways are in action against the ecotin KO cells, we used exogenously added selective inhibitors: SGMI-2, which blocks only the LP [13], SGMI-1, which blocks both the LP and, when applied at higher concentrations, the LPS-triggered AP [13,19] and an anti-C1q monoclonal antibody (mAb) (MW1828, Sanquin), which blocks the CP. Ecotin was also applied exogenously as a control to test whether it provides comparable protection to that of the endogenous periplasmic inhibitor present in the wild type bacteria. We conducted these tests both in the context of the above mentioned C3-deposition assay as well as with the cell proliferation assays described below.
Briefly, in the proliferation-based cell-killing assay, constant amount of mid-log phase cells was treated for 1h with diluted serum either lacking or containing exogenously added inhibitors. The bacteria were then transferred into large excess of broth and the optical density (λ = 600 nm) was measured after 5h incubation.
In agreement with the C5b-9-deposition assay, the ecotin KO ATCC 23505 strain was efficiently attacked and killed by the serum, as almost no viable cell remained after an exposure to as low as 1% NHS. On the other hand, endogenous ecotin of the wild type ATCC 23505 strain provided perfect protection in a broad range of serum dilution up to 16% (Fig 11A).
We tested whether increased serum-susceptibility of the ecotin KO ATCC 23505 and ATCC 12014 cells is due only to the lack of endogenous ecotin. To do so, we transformed these cells with an isopropyl β-D-1-thiogalactopyranoside (IPTG)-inducible vector to restore periplasmic ecotin expression. Constructing the expression vector and measuring periplasmic ecotin levels of the cells is described in the Materials and methods. Briefly, ecotin levels were measured through chymotrypsin inhibitory activity of the isolated periplasmic fraction as follows: 20 nM final concentration of bovine chymotrypsin was mixed with serial dilutions of periplasmic fractions and the dilution corresponding to 50% enzyme inhibition was determined. We verified that the periplasm of ecotin KO cells does not contain ecotin. Isolated periplasmic fraction of wild type cells readily inhibited chymotrypsin. The periplasm of transformed but non-induced cells showed hardly detectable chymotrypsin inhibitory activity. Ecotin levels in the periplasmic fraction of transformed and IPTG-induced ATCC 12014 ecotin KO cells exceeded the wild type level 2.6-fold, while the corresponding ratio for the ATCC 23505 ecotin KO cells was 1.5-fold. Therefore, periplasmic ecotin level was restored or slightly increased over the wild type level in both strains.
Most importantly, in the ATCC 23505 ecotin KO strains, endogenously expressed ecotin perfectly restored wild type phenotype. Without induction, the vector-carrying ecotin KO cells were as susceptible to cell-killing as the untransformed ecotin KO cell. In contrast, IPTGinduced ecotin expression rescued these cells providing them with a level of protection practically identical with that of the wild type ( Fig 11A).
The flow cytometry experiments above demonstrated that the ATCC 23505 bacteria were attacked only by the LP, as both SGMI-1 and SGMI-2 could completely block C3-deposition, while the anti-C1q mAb provided no protection ( Fig 9B). Exogenous ecotin also provided full protection in agreement with the fact that wild-type ATCC 23505 cells and the rescued KO cells were completely protected by their endogenous ecotin (Figs 9B and 11A). The cell proliferation experiments yielded the same results: all LP-inhibitory compounds provided full protection for the cells, while the anti-C1q mAb had no activity ( Fig 12A).
Likewise, the flow cytometry measurements above showed that the ATCC 12014 bacteria were also attacked by the complement system, but, in terms of deposited C3, at a tenfold lower level (Fig 9D), and through different mechanisms (Fig 9E). Anti-C1q mAb and SGMI-1 provided almost complete protection against C3-deposition approaching the level achieved by the broad-specificity protease inhibitor FUT-175 [31] (344960, Sigma-Aldrich), while SGMI-2 had much lower protecting effects ( Fig 9E) demonstrating that LP-activation on this strain is negligible. On the other hand, high efficiency of the anti-C1q mAb treatment identified an essential role of CP activation, and the observed drop in C3-deposition can be interpreted as the cancelled contribution of the CP, plus the CP-triggered positive feedback loop provided by the AP, the latter one generating up to 90% of the deposited C3b [32]. High efficiency of SGMI-1 ( Fig 9E) cannot be explained by LP-inhibition, as we already concluded that LP was hardly triggered. However, it is in agreement with our previous finding that at the high concentration applied here, SGMI-1 is a strong inhibitor of LPS-triggered AP activity [19]. In all, high efficiency of SGMI-1 suggests that it inhibits both spontaneous AP-activation as well as the CPtriggered AP feedback loop.
The cell proliferation tests provided coherent results: endogenous ecotin of the wild type ATCC 12014 strain, that does not inhibit CP-activation, provided significant, but only partial protection for the bacteria (Fig 11B), most probably by partial inhibition of the AP amplification loop. When SGMI-1 or the anti-C1q antibody was added exogenously to the ecotin KO cells, these provided significantly stronger protection than SGMI-2, which had only a marginal effect (Fig 12B).
Again, importantly, in the ATCC 12014 ecotin KO strains, endogenously expressed ecotin perfectly restored wild type phenotype in the cell-killing assay. Without induction, the vectorcarrying ecotin KO cells were as susceptible to cell-killing as the untransformed ecotin KO cell, while ecotin expression induction rescued these cells providing them with a level of protection practically identical with that of the wild type cells (Fig 11B).

Endogenous ecotin protects bacteria against a complement-independent, heat-stable antimicrobial effector mechanism of serum
As the activity of the CS is heat sensitive, heat-inactivated serum (HIS) was used in the flow cytometry and cell proliferation assays as a typical negative control serving as a baseline to deconvolute complement-specific effects. In the flow cytometry assays, as expected, no C3-deposition was detected from HIS. Yet, based on the cell proliferation assays, and to our great surprise, the polymannan O9 LPS-carrying ecotin KO ATCC 23505 strain was attacked by HIS in a serum concentration dependent manner (Fig 11C). Lack of deposited C3 fragments verifies that this attack is complement-unrelated. Bactericidal activity of HIS has already been examined by others [33], but the nature of such effects remained enigmatic. The observed bactericidal effect could be ameliorated via supplementing the HIS with ecotin or with the  Fig 9, we tested the efficacy of exogenously added inhibitors also on the bactericidal effect. We used 1% NHS where the observed bactericidal effect of heat-stable factors (see Fig 11C) is negligible. Exogenously added SGMI-1 and -2, as well as ecotin rescued the ATCC 23505 ecotin KO strain, while anti-C1q mAb exerted minimal effect (A) supporting the dominant role of LP-activation. ATCC 12014 ecotin KO cells were rescued by the anti-C1q mAb, ecotin and SGMI-1, while SGMI-2 exerted a minor effect suggesting that all three complement pathways are in action, but the classical and alternative pathways dominate (B). We also attempted to identify the factors responsible for complement-independent bactericidal effects of 16% HIS. Only ecotin and the general serine protease inhibitor FUT-175 could diminish this attack (C). Data are from at least two parallel measurements, asterisks ( � ) representing p < 0.05 (twotailed Student's t-test) when comparing the OD values of samples treated with inhibitors to those of 1% NHS. As data in (C) were obtained from two parallels, no statistical significance was calculated. Error bars represent the SD. https://doi.org/10.1371/journal.ppat.1008232.g012 Ecotin protects bacteria against human serum-mediated lysis broad-specificity protease inhibitor FUT-175 [31]. Extraneous ecotin therefore ablated susceptibility of the cells to a bactericidal effect of HIS. However, our monospecific MASP-inhibitors, SGMI-1, SGMI-2 and TFMI-3 were all ineffective, as were the plasma kallikrein inhibitor PKSI-527 [34] and the FXIIa-blocking corn trypsin inhibitor [35] (Fig 12C). This inhibitor panel demonstrates that the complement-unrelated attack is protease-dependent, but is not driven by the contact system or by any moonlighting activity of the MASP enzymes.
We tested if the observed susceptibility of the ecotin KO ATCC 23505 cells to HIS concentration above 4% was due exclusively to the lack of endogenous ecotin, as explained for the other cell-killing tests. We found that in the ATCC 23505 ecotin KO strains, endogenously expressed ecotin perfectly restored wild type phenotype. Without induction, the vector-carrying ecotin KO cells were as susceptible to cell-killing as the untransformed ecotin KO cell. In contrast, IPTG-induced ecotin expression rescued these cells providing them with the same level of protection observed for the wild type (Fig 11C).

Discussion
Microbial pathogens evolved a variety of complement evasion strategies. These mainly rely on the acquisition or expression of complement regulators and inhibitors, or on the degradation of complement components [36][37][38], demonstrating that complement inhibition is a crucial defense mechanism of pathogenic microbes. In this paper, we unequivocally prove that ecotin inhibits both antibody-independent CS pathways, the LP and the AP, and it also blocks an apparently complement unrelated antibacterial serum activity. These discoveries place previous findings of other research groups about ecotin as a potential virulence factor into completely new context.
Ecotin was discovered 36 years ago in E. coli. By now, ecotin orthologs have been identified in more than 200 species, and over 40 of these are well-documented opportunistic or obligatory pathogens including Trypanosoma cruzi, Leishmania major, Burkholderia pseudomallei, Yersinia pestis and Pseudomonas aeruginosa (summarized in Fig 13 and S1 Table). These updated numbers support the decade-old finding that ecotin genes are more frequent in pathogenic species than in non-pathogenic ones [39]. Ecotin producing organisms are widespread among Bacteria, especially within the Gammaproteobacteria subdivision, which contains more than a hundred species harboring ecotin orthologs. In this subdivision, the Enterobacteriaceae family is particularly rich in known pathogenic species expressing ecotin. Moreover, several members of the eukaryotic Euglenozoa genus including the human parasites Trypanosoma brucei and Leishmania major also express ecotin orthologs (S1 Table).
Ecotin has been identified as virulence factor for several pathogenic Gram negative bacteria such as Yersinia and Burkholderia species and two species of the eukaryotic unicellular pathogen, Leishmania as well [8][9][10]40].
The gastrointestinal pathogen, Yersinia pseudotuberculosis is a close relative of the etiologic agent of the bubonic and pneumonic plague, Yersinia pestis. When wild type and ecotin KO Yersinia pseudotuberculosis cells were injected into the bloodstream of mice in 1:1 ratio, after 5 days, the wild type cells were present in about 30-fold excess compared to the ecotin KO cells. It was hypothesized that the in vivo function of ecotin might be protection from neutrophildependent killing probably by inhibiting neutrophil elastase [8]. Pseudomonas aeruginosa is an opportunistic pathogen causing chronic lung infections. Its ecotin was found to be sequestered by the exopolysaccharide, PsI on the bacterial surface and again, it was shown to protect the cells against neutrophil elastase-mediated elimination of the bacteria [41]. This phenomenon is in agreement with the findings of others, who found, that ecotin expression is upregulated in E. coli and P. aeruginosa biofilms [42,43]. Since both pathogens form biofilms in certain diseases [44], this might be a mechanism by which ecotin expressing bacteria increase their protection against antimicrobial effectors of the host. For Burkholderia pseudomallei, the causative agent of melioidosis, ecotin was shown to be important for intracellular survival of the bacteria in murine macrophages, most probably by inhibiting several proteases of the early endosome and allowing for an escape of the intracellular pathogen into the cytosol. Importantly, compared to the wild type strain, intraperitoneal infection with the ecotin KO strain leads to increased time to death and greater survival rate of the infected mice [9].
Ecotin orthologs are present in several species of the eukaryotic parasite Trypanosomatidae family including Trypanosoma cruzi causing Chagas disease, Leishmania major causing zoonotic cutaneous leishmaniasis and Leishmania donovani the etiologic agent of visceral leishmaniasis or kala-azar. While it was shown that Trypanosoma cruzi is attacked by all three complement pathways, and several parasite proteins were found to attenuate this attack, ecotin was not identified as a complement inhibitor, but was suggested to protect the cells against neutrophil elastase and cathepsin G [45,46].
Similarly, all but one publications on Leishmania ecotin orthologs indicated that the main function of these proteins is protection against intestinal proteases in the gut of the insect vector, or against contact pathway-mediated bactericidal effects of the host [47,48], or downregulation of inflammatory monocytes and monocyte-derived cells [49]. We found only a single, very recent paper reporting that Leishmania donovani ecotin ortholog LnISP2 is not only a neutrophil elastase inhibitor, but also a MASP-2 inhibitor [40]. Notably, however, in that study, recombinant LnISP2 was 50-fold less efficient in providing complete MASP-2 / LP-inhibition, than our recombinant Leishmania major ISP2 ortholog in analogous tests.
In all, we demonstrated that ecotin inhibits i) LP-activation, ii) LPS / zymosan-triggered AP-activation, iii) MASP-3-based pro-FD activation and iv) an apparently complement-unrelated antimicrobial effect of HIS. All these functions of ecotin, as we showed, provide a proliferative advantage for the ecotin harboring cells.
While the molecular mechanism of LP-inhibition is obvious: ecotin potently inhibits MASP-2, the source of the observed heat-stable bactericidal activity, and thereby the protecting mechanism of ecotin needs to be identified, and the mechanism of ecotin-based AP-inhibition also requires further investigation.
We showed that E. coli ecotin is a 4 μM FD-inhibitor and that at 10 μM concentration it could provide up to 75% inhibition of AP-activation triggered by LPS-and zymosan surfaces. Moreover, in a system containing purified components, we also showed that, at a lower degree, ecotin blocked FD-driven C3bBb generation and C3bBb-driven C3b generation, demonstrating that E. coli ecotin is a weak inhibitor of FD, C3bBb, or both. Note, that in the serum tests, ecotin was pre-incubated with the serum, so it could reach binding equilibrium with the active FD contents of the serum. In the solution assay no such pre-equilibrium with FD was possible as the reaction was initiated by FD. Moreover, it is important to note that the above AP-activation systems are not readily comparable. LPS and zymosan-triggered AP-activation are complex, serum-based assays engaging many complement regulatory proteins. Moreover, APactivation in these tests occur on surfaces, and require a de novo C3b-generation phase offering a level of inhibitory intervention missing from the solution phase AP-convertase generating system containing an initial C3b pool provided as purified protein. Previously we showed that MASP-1 is essential for LPS-triggered, but not for zymosan-triggered AP-activation [19]. As E. coli ecotin is a micromolar MASP-1 inhibitor, this activity might contribute to the inhibition of LPS-triggered but not to zymosan-triggered AP-activation. As long as the exact molecular mechanisms of AP-activation on these surfaces remain unknown, it cannot be excluded that zymosan-triggered AP-activation also relies on a yet to be identified protease, inhibited by ecotin.
MASP-3 was for long considered as a protease lacking any natural serpin-type or canonical inhibitor. Accordingly, the most outstanding property of ecotin is its very high inhibitory potency on MASP-3. As we have recently showed, in resting blood, not perturbed by coagulation or inflammatory processes, MASP-3 is the exclusive activator of pro-FD [14]. As FD is the AP-initiator enzyme, this MASP-3 function is fundamental, and the outstanding MASP-3-inhibitory capacity of ecotin suggests an important biologic function. Note, however, that this ecotin function could not manifest in our AP tests, as in those assays FD was already present in the serum in activated form, or was provided as activated, isolated protein.
As mentioned, in the blood, due to the presence of activated MASP-3, FD circulates mostly in activated form. While systemic control of FD-activation in the blood through complete MASP-3 inhibition should be possible, it would require the continuous presence of large excess of ecotin, which is an unlikely situation in microbial infections. On the other hand, inhibition of MASP-3-driven pro-FD activation can be relevant at the much more common locations of primary infections, which also represent the only niche of normal microbiota, the periphery, including the mucosae. Several opportunistic pathogens, such as E. coli and P. aeruginosa colonize at first the mucosae, therefore the earliest encounter between the complement and the bacteria occurs in this environment. The composition of the complement, the concentration of its components, and therefore the mechanism of complement activation might be different in the extravascular environments from those in the blood [50][51][52]. For example, the level of activated FD at the periphery might be limited by the level of its activators, and if so, ecotin could provide a major defense mechanism against the AP. A similar difference between blood and extravascular environments might prevail for the levels of the LP components. Based on this, we suggest that the potent protective effects of ecotin we demonstrated in dilute serum against the LP, the AP and a complement-unrelated attack imply similar efficient protections in mucosal, interstitial and other extravascular environments.
In all, the activities of ecotin revealed here, together with its previously established neutrophil and macrophage controlling actions demonstrate, that ecotin blocks or attenuates a plethora of powerful antimicrobial attacks of the host innate immune system. These versatile innate immunity-attenuating activities of ecotin might have been evolved in the normal host microbiota to provide general protection for the microbes in the mucosa and other extravascular niches as long as they do not provoke specific IgG-response unleashing (among others) the ecotin-resistant classical complement pathway. The very same ecotin functions, when combined with other virulence factors, should provide important advantages for pathogens as well.
We believe that all previous studies conducted on the interactions between ecotin-harboring microbes and their hosts should be revisited and reevaluated in the light of these new findings. We suggest that ecotin should be put in the focal point of dedicated fundamental research, and as it can be a potential drug target, the new findings might open novel therapeutic options as well.

Ethics statement
All studies using human serum were conducted in strict conformity with the WMA Declaration of Helsinki. Experimental protocols were approved by the local ethics committee (permission number: TUKEB 9190-1/2017/EKU). Informed written consent was obtained for the isolation of peripheral venous blood from the donors.
All procedures using mouse serum were carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. The protocol was approved by the Institutional Animal Care and Use Committee at the Icahn School of Medicine at Mount Sinai Center for Comparative Medicine (registration number: D16-00069).
All procedures using rat serum were performed in strict accordance with guidelines set by the National Institutes of Health (USA) and the Hungarian law on animal care and protection. The protocol was first approved by the Institutional Ethical Committee for Animal Care and Use of Semmelweis University and then approved by the Government Office of Pest County Directorate of Food Chain Safety and Animal Health (registration number: PEI-001/3948-6/ 2014).

Cell lines and recombinant proteins
Ecotin KO variants of E. coli BL21 (DE3) Star (Invitrogen), ATCC 12014 and ATCC 23505 were made according to [53]. Recombinant ecotin variants were expressed in E. coli BL21 (DE3) Star (ecotin KO) and purified to homogeneity. Recombinant enzymes were expressed and purified according to [54,55]. SGMI-1, SGMI-2 and TFMI-3 were expressed and purified as described in [14,21,56]. The K I values of ecotin variants against MASP-1 and MASP-2 were determined on synthetic substrates. The K M values of MASP-1 and MASP-2 for Z-Lys-SBzl were determined (Fig 2). Complement activity was measured in C3-deposition ELISA assays in 2% NHS as described in [13,27]. Similar tests were also conducted with mouse and rat serum as well.

Reagents
All common laboratory reagents were from Sigma-Aldrich and Merck. High-binding microtiter plates were from Greiner Bio-One.

Construction of ecotin knockout E. coli strains
The ecotin gene of E. coli BL21 (DE3) Star (Invitrogen), ATCC 12014 and ATCC 23505 (American Type Culture Collection, ATCC) was replaced with the chloramphenicol acetyltransferase gene (EC 2.3.1.28) using the recombineering plasmid pKD46-RecA as described [53,57]. The plasmid was electroporated into the bacteria, recombinant clones were isolated and the loss of the heat-sensitive plasmid was verified. Successful replacement at the targeted genomic locus was confirmed by PCR, restriction analysis and DNA sequencing using specific primers. The chloramphenicol resistance gene was kept to serve as a stable selection marker for the strains BL21 (DE3) Star ecotin KO, ATCC 12014 ecotin KO and ATCC 23505 ecotin KO. No ecotin could be detected in the periplasm of the recombinant (ecotin KO) clones.

Construction of ecotin expression vectors
The gene coding for the mature E. coli ecotin sequence (UniProt: P23827) had originally been cloned into the pT7-7 expression vector [17]. Genes coding for the mature ecotin protein sequences of Y. pestis (UniProt: A4TNJ2), P. aeruginosa (UniProt: B7UZW2), and L. major (corresponding to the ISP2 ecotin ortholog) (UniProt: Q4QFD4) were codon-optimized for expression in E. coli. Cleavage sites of frequently used restriction endonucleases were removed with silent mutations.
The four ecotin genes were obtained as a single synthetic construct (Life Technologies) in which each variant was flanked by cleavage sites of two different restriction endonucleases. The plasmid was digested with the appropriate enzymes in separate reactions. The genes were isolated and amplified by PCR using specific primers that also incorporated NcoI and HindIII sites. The genes of Y. pestis and P. aeruginosa ecotins were sub-cloned into the pET-25b(+) periplasmic expression vector (Novagen) with NcoI and HindIII restriction endonucleases. The L. major ecotin gene was sub-cloned into a modified pETM-11 cytoplasmic expression vector with NcoI and HindIII restriction endonucleases. The L. major ecotin was produced with an N-terminal His 6 -tag which could be cleaved off with TEV protease.

Expression of ecotin orthologs
E. coli BL21 (DE3) Star ecotin KO cells were transformed with the expression vectors. Transformed cells were grown in LB medium to OD 600 nm = 4 at 37˚C. Recombinant protein expression was induced with 0.5 mM isopropyl-β-D-thiogalacto-pyranoside. After 5 hours, the cell cultures were harvested via centrifugation (4500 g, 4˚C, 10 min).

Purification of E. coli, P. aeruginosa and Y. pestis ecotin orthologs
Cells were resuspended in 100 mL / liters of cell culture 2.5 mM MgCl 2 and frozen at -20˚C overnight. After thawing, the samples were centrifuged (48000 g, 4˚C, 20 min) and the supernatants corresponding to the periplasmic fractions were sonicated to disrupt any residual cell debris. Then, 65% (m/v) ammonium-sulphate was added gradually to the sample and the precipitated proteins were collected via centrifugation (48000 g, 4˚C, 20 min). The pellet was resuspended in 2.5 mM HCl, and the sample was dialyzed against 2.5 mM HCl overnight at 4˚C. The dialyzed sample was centrifuged (48000 g, 4˚C, 20 min) and the pH of the ecotincontaining supernatant was set to 7.9 with Tris base. The sample was loaded onto a GE HiTrap Q HP anion-exchange column equilibrated with 10 mM Tris-HCl pH 7.9 buffer. In the case of the Y. pestis ortholog, the ecotin was the major component of the flow-through. In contrast, the E. coli and P. aeruginosa ecotin orthologs were retained by the column and were eluted using a linear NaCl gradient. Fractions containing the ecotin orthologs were further purified with RP-HPLC on a Phenomenex Jupiter 10u C4 300A column equilibrated with water, 0.1% TFA, and eluted with a linear gradient of acetonitrile, 0.09% TFA. The ecotin containing fractions were lyophilized and the proteins were dissolved in water.

Purification of His 6 -tagged L. major ecotin (ISP2)
The cells were resuspended in 100 mL per liter of cell culture 50 mM Na-phosphate, 300 mM NaCl pH 7.5 buffer and the suspension was frozen at -20˚C. After thawing, the sample was sonicated and centrifuged (48000 g, 4˚C, 20 min). The protein was purified on Bio-Rad Profinity IMAC resin equilibrated with 50 mM Na-phosphate, 300 mM NaCl, pH 7.5. The protein was eluted with the same buffer supplemented with 250 mM imidazole and dialyzed against 10 mM Na-phosphate pH 7.5 at room temperature. The His 6 -tag was cleaved off with His 6 -tagged TEV protease. L. major ecotin was separated from the His 6 -tagged protease and the His 6 -tag by a second Ni 2+ affinity chromatography. The inhibitor was further purified on a GE HiTrap SP HP cation-exchange column equilibrated with 10 mM Na-phosphate pH 7.5 buffer. The protein was eluted with a linear NaCl gradient.

Quality control of the inhibitors
Purity of the proteins was confirmed via SDS PAGE. The correct molecular weight of the proteins was verified via mass spectroscopy (ESI-MS).

Determination of the inhibitor concentration
Concentration of ecotin samples was determined similarly as described elsewhere [11]. Briefly: bovine trypsin (Sigma Aldrich) was dissolved in 10 mM HCl, 10 mM CaCl 2 and active-site titrated using 4-methylumbelliferyl p-guanidinobenzoate [60]. The active concentration of each ecotin stock was determined by titration against the active-site titrated trypsin. Incremental amounts of the inhibitors were incubated with 1 μM trypsin for 10 minutes in 50 mM Tris-HCl pH 8.0, 10 mM CaCl 2 , 0.005% Triton X-100. Residual enzyme activities were measured using 1 mM Nα-Benzoyl-DL-arginine p-nitroanilide. Inhibitor concentration was determined by linear regression analysis.

Equilibrium binding assays
The equilibrium inhibitory constant (K I ) values of ecotin against the catalytic fragments of MASP-1, MASP-2, MASP-3, C1s and C1r as well as against recombinant FD were determined according to the method of Green and Work with the modifications of Empie and Laskowski et al. [61,62]. The measurements were done in 20 mM HEPES, 145 mM NaCl, 5 mM CaCl 2 , 0.05% Triton X-100 pH 7.5 buffer in 200 μl final volume using a BioTek Synergy H4 multimode microplate reader. A fixed concentration of protease was incubated with increasing amounts of the inhibitor in separate samples. Enzyme concentrations were set based on preliminary experiments. After reaching equilibrium, the amount of uninhibited enzyme was quantitated using a thioester substrate combined with an appropriate co-substrate. In most cases, 250 μM Z-L-Lys-SBzl hydrochloride thioester substrate and 500 μM 5,5'-dithiobis-(2-nitrobenzoic acid) (DTNB) as photometric co-substrate were used. For C1s and C1r, 250 μM Z-Gly-Arg-SBzl was combined with 500 μM DTNB. For the tight-binding E. coli ecotin / MASP-3 interaction the 250 μM Z-L-Lys-SBzl hydrochloride thioester substrate was combined with 40 μM of the high-sensitivity Thiol Fluorescent Probe IV co-substrate. The fluorescence of this co-substrate increases 470-fold upon reacting with thiols [63]. The K I values were determined from at least two parallel measurements by nonlinear regression analysis in OriginPro software using the following equation: ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi where [E], [E] 0 and [I] 0 represent the molar concentration of the free enzyme, the total enzyme, and total inhibitor, respectively. Based on the literature [7], a k on of 5.3 x 10 5 M -1 s -1 was used to estimate k off values and subsequently the half-lives of the complexes. For enzyme-inhibitor pairs where the half-life of the complex was comparable with, or shorter than the length of the measurements, only apparent K I values (K I � ) could be determined because there was a genuine competition between the inhibitor and the substrate for enzyme-binding. The K I values were then calculated by the following equation: where [S] is the molar concentration of the substrate and K M is the Michaelis constant [64]. For this calculation, the K M values of the three MASPs for Z-L-Lys-SBzl were determined by us. For FD, a K M of 3.79 mM, while for C1s, a K M of 190 μM was used, based on the literature [65,66].

Determination of the K M values
The

Analysis of the interaction between E. coli ecotin and MASP-3 using pro-FD as a substrate
These experiments were done as in [22] with

Cy3-labeled pro-FD cleavage in NHS
These experiments were done based on the already described method [22]. A total of 4 μl 1 mg/ml Cy3-labeled pro-FD was added to 40 μl NHS, and the mixtures were incubated at 37˚C for 24 hours. Samples were withdrawn at every hour in the first 8 hours, at 10 h, 12 h and then at 24 h on pro-FD conversion. The inhibitors were added to the serum samples at 1 μM final concentration and incubated for 10 minutes prior to addition of the labeled pro-FD. The reaction in the withdrawn samples was immediately stopped by 10-fold dilution with SDS-PAGE sample buffer and heating for 2 min at 95˚C. A total of 2 μl of the cleavage assay samples containing 18 ng Cy3-labeled protein were run on SDS-PAGE under reducing conditions. The gels were scanned with a Typhoon laser scanner (GE Healthcare). Band intensities were quantified by densitometry. Because the two bands were partially overlapping, only the time point at which the intensity of the Cy3-pro-FD band equaled that of the Cy3-FD band could be reliably determined on the gel, and this time point was considered as the half-life of Cy3-pro-FD.

Testing FD-and C3bBb-inhibiting capacity of ecotin in an in vitro AP convertase generating system
An in vitro AP convertase generating system was set up by mixing 100 nM C3b, 30 nM FB and 1 μM C3 that served as substrate for the freshly generated C3bBb-type C3-convertase. This mixture was pre-incubated in 50 mM Tris-HCl, 200 mM NaCl, pH 7.45 buffer at room temperature, for 30 minutes without (positive control) or with inhibitors, before starting AP-convertase generation by adding 10 nM FD. One of the samples contained 10 μM ecotin, while the other contained 100 μM of the pan-specific serine proteinase inhibitor, FUT-175, that inhibits all CS proteases. A sample containing neither FD nor FB served as negative control. From the mixtures that received FD samples were withdrawn at various time points, and were immediately treated with mercaptoethanol and SDS containing gel loading sample buffer at 95˚C. C3b-generation from C3 was assessed by SDS-PAGE. C3 and C3b were applied onto the gel as controls. Extent of C3-cleavage and C3b-generation for the inhibitor containing samples were compared to those in the positive control sample containing no inhibitor and to the samples taken directly after starting the AP-convertase generation by adding FD. This way, we could deconvolute the quantity of C3b added to enable the preliminary C3bB-formation from the C3b generated by the convertase. The results were analyzed via ImageJ software.

Preparation of pooled NHS
Blood was drawn from 10 healthy donors after informed written consent into Vacutainer serum tubes (BD Diagnostics) to produce fresh serum. Blood samples were incubated for 1 h at room temperature to allow complete coagulation. The serum was collected, pooled, and stored at −80˚C until use.

Preparation of pooled mouse serum
Blood was collected from five C57BL/6 and five BALB/c mice at the age of 6-10 weeks by venipuncture of the facial vein. Blood samples were incubated for 0.5 hours at room temperature to allow coagulation to be completed. The samples were centrifuged at 14,000 rpm for 20 minutes at 4˚C. Serum supernatants were pooled, snap frozen in liquid nitrogen and stored at -80 C until use.

Preparation of rat serum
Blood was collected from an individual euthanized Wistar rat. The serum preparation was done as described for human serum preparation.

Complement ELISA assays
Classical pathway (CP) inhibition measurements from NHS. CP activity was measured through detection of C3-deposition as described previously [27], with the following modifications: 1. Greiner high-binding microtiter plates were coated with 10 μg/mL IgG in 15 mM Na 2 CO 3 pH 9.6, and incubated overnight at 4˚C. 4. These mixtures were pre-incubated in single loose tubes (National Scientific Supply Company) for 30 min at room temperature before transferring onto the activator surface.
Polyclonal rabbit anti-human C3c antibody (A0062; DakoCytomation) was used as the primary antibody in 2000-fold dilution and HRP-conjugated anti-rabbit antibody (A1949, Sigma) was used as the secondary antibody in 40,000-fold dilution.
The signal produced by non-inhibited NHS was considered to be 100% activity, while 0% complement activity was defined as the signal produced by 2% NHS treated with 50 μM FUT-175 inhibitor. Results were obtained from four parallel measurements. In all serum tests, the signal developing reaction on the plate was stopped when optical density of the highest signal intensity non-inhibited reaction was between OD 0.5-1.0. The typical background signal was around OD 0.05.
Lectin pathway (LP) inhibition measurements from NHS. LP activity was measured through detection of C3-deposition as described previously [27], with the following modifications: 1. Greiner high-binding microtiter plates were coated with 10 μg/mL mannan in 35 mM NaHCO 3 , pH 9.6, and incubated overnight at 4˚C. 4. These mixtures were pre-incubated in single loose tubes (National Scientific Supply Company) for 30 min at room temperature before transferring onto the activator surface.
Polyclonal rabbit anti-human C3c antibody (A0062; DakoCytomation) was used as the primary antibody in 2000-fold dilution and HRP-conjugated anti-rabbit antibody (A1949, Sigma) was used as the secondary antibody in 40,000-fold dilution.
The signal produced by non-inhibited NHS was considered to be 100% activity. 0% complement activity was defined as the signal produced by 2% NHS treated with 50 μM FUT-175 inhibitor. Results were obtained from four parallel measurements.
LP inhibition measurements from rat serum. In these tests, 60-fold diluted serum was used, otherwise the experiments were performed as described above, because the anti-human C3c polyclonal primary antibody cross-reacts with rat C3c. The signal produced by non-inhibited serum was considered to be 100% activity. 0% complement activity was defined as the signal produced by 60-fold diluted rat serum treated with 50 μM FUT-175 inhibitor. Results were obtained from four parallel measurements.
LP inhibition measurements from pooled mouse serum. Nunc MaxiSorp plates were coated with 10 μg/mL mannan in PBS overnight at 4˚C. Wells were blocked with 5 mg/ml BSA in TBS pH 7.4 and washed with TBS, 0.05% Tween-20 (wash buffer) subsequently. Pooled mouse serum was diluted 60-fold in TBS, 5 mM MgCl 2 , 5 mM CaCl 2 , 0.05% Tween-20 containing serial dilutions of E. coli ecotin. These mixtures were pre-incubated for 15 min at room temperature before transferring 100-100 μl onto the plate. The plates were incubated at 37˚C for 30 minutes and then washed with wash buffer. 100-100 μl HRP-conjugated goat antimouse C3 IgG (Cappel #55557) diluted 5000-fold in TBS, 5 mg/ml BSA, 0.05% Tween-20 was applied to the plates and incubated for 1 hour at room temperature. The plates were washed and 100 μl/well TMB substrate (Sigma) was added to the wells. The reaction was stopped by adding 50 μl/well 1 M HCl and the signals were read at 450 nm.
The signal produced by non-inhibited serum was considered to be 100% activity. 0% complement activity was defined as the signal produced by serum treated with 50 μM FUT-175 inhibitor. Results were obtained from three parallel measurements.
Alternative pathway inhibition measurements from NHS. Alternative pathway assays were carried out as described for the LP and CP inhibition measurements from NHS, but with the following important modifications: 1. the plates were coated with 100 μg/mL zymosan or with 10 μg/mL LPS (Sigma Aldrich).
3. The final dilution of NHS was 6-fold.
Calculation of the half maximal inhibitory concentration (IC 50 ) values. IC 50 values were calculated with using OriginPro software. Data was fitted using the following dose response function: where A min and A max represent the minimum and maximum of the fitted curve, respectively, [I] is the inhibitor concentration, p is the Hill slope and y is the degree of C3 deposition.

Detection of C3-deposition on E. coli by flow cytometry
Flow cytometry was used to assess C3-deposition on E. coli cells. E. coli ATCC 23505, ATCC 23505 ecotin KO, ATCC 12014 and ATCC 12014 ecotin KO cells were cultured in peptone/ beef extract (P/BE) broth, as recommended on www.atcc.org. At OD 600 nm = 0.5, cells were chilled on ice, centrifuged and washed twice with 10 mM HEPES, 145 mM NaCl, 5 mM CaCl 2 , 5 mM MgCl 2 pH 7.4 buffer (NHS-buffer). Cell density was measured, and 3.4 x 10 7 cells were transferred into 1.5 mL microcentrifuge tubes and pelleted at 15,000 g. The cells were suspended in 200 μL of either NHS or NHS-buffer (negative control) at appropriate dilution, and incubated at 37˚C for 30 minutes. After incubation, cells were washed twice and suspended in 100 μl of 1/500 polyclonal rabbit anti-C3c antibody (A0062, DakoCytomation) or in 100 μL NHS-buffer, mixed thoroughly and incubated at 37˚C for an hour. After washing the cells twice, 100 μl of 1/2000 Alexa Fluor 488 conjugated polyclonal goat anti-rabbit IgG (A21206, Invitrogen) was added to each sample and incubated at room temperature for 30 minutes in dark. Samples were measured on a Beckman Coulter CytoFlex instrument using CytExpert software. SSC threshold values were set to exclude most of debris, and to enable gating on bacterial populations, data from 20,000 cells in the bacteria gate were collected in each sample. All buffers used in flow cytometry samples were filtered (0.22 μm filter), to exclude any contaminating particles in the SSC-FSC range of the bacteria. Data analysis was performed using FlowJo software. Due to the intensive lysis altering the detected signal of opsonized cells (Fig  10), no statistical significance was calculated.

Detection of C5b-9 (MAC) deposition on E. coli by flow cytometry
Flow cytometry was used to measure C5b-9-deposition on the surface of E. coli cells. All steps were the same as described above, except that mouse anti-C5b-9 antibody (A239, Quidel) was used as the primary antibody and Alexa Fluor 647 conjugated rabbit anti-mouse antibody (A21239, Invitrogen) as the secondary antibody. Pore formation was also evaluated via adding propidium iodide (PI) as a DNA stain. The percentage of C5b-9/PI positive cells was determined. The sample preparation steps yielded some PI+ (single positive) cells, but as the percentage of this PI positive population was constant in all samples, it was considered as artifact, and was thereby excluded from data analysis. Single C5b-9 positive cells were present only in very low amounts, suggesting a transient state that would also turn PI positive upon longer incubation.

Bactericidal tests in NHS and HIS
Bactericidal tests were conducted based on the method of Muschel et al. [67], with modifications. Cells were grown to mid-log phase (OD 600 nm~0 .5) in P/BE broth, centrifuged, washed in NHS-buffer, and their density was determined via measuring the optical density at 600 nm, in Beckmann Coulter DU-7400 spectrophotometer. Cells were diluted in P/BE broth, then 2 x 10 7 cells were transferred to each test tube. NHS-buffer (control tubes), NHS or HIS (both diluted in NHS-buffer) was added to reach the appropriate final serum concentration. In the experiments testing the effects of exogenously added inhibitors, the serum was pre-incubated with the appropriate inhibitor for 30 minutes at room temperature. The cells were incubated with serum at 37˚C for 1h, then cells from the sample tubes were inoculated into 20-fold excess of fresh, serum-free P/BE broth. This dilution ceased complement activation and lowered the concentration of cell debris that could have interfered with the subsequent optical density measurements. After 5h of incubation, optical density (λ = 600 nm) of the samples was measured and compared to those of the control samples.

Complementation of the ecotin KO cells with E. coli ecotin provided on an IPTG-inducible expression vector
Protecting role of ecotin was tested in complementation experiments. As the two E. coli strains are not compatible with T7 promoter vectors, we sub-cloned the gene of wild type ecotin from the original pT7-7 [17] vector into the pTFPI-D2-pro-lib [14] phagemid vector using NdeI and XbaI restriction enzyme cleavage sites. The resulting vector, which allows for IPTG-inducible periplasmic expression of ecotin, was named pTac.Eco. The correct sequence of the vector was confirmed by DNS sequencing. Ecotin KO E. coli ATCC 23505 and ATCC 12014 cells were transformed with the pTac.Eco vector by electroporation. We tested that in P/BE broth, the untransformed cells and the transformed, non-induced cells grew with the same rate, and this rate did not decrease up to a level of 50 μM IPTG added to the transformed cells. Therefore, an IPTG concentration of 50 μM was used for ecotin expression induction in the above described bactericidal tests. Ecotin level in the periplasmic fraction was determined as follows. Wild type E. coli ATCC 23505 and ATCC 12014 cells, their ecotin KO versions, their ecotin KO pTac.Eco-transformed but non-induced versions and their ecotin KO pTac.Eco-transformed and 50 μM IPTG-induced versions were grown in P/BE broth up to OD 600 = 1.0. The periplasmic fraction was isolated similarly to the procedure described at the "Purification of E. coli, P. aeruginosa and Y. pestis ecotin orthologs" section in the Materials and methods, but in this case the pelleted cells were resuspended in 10 mL / 1 L culture ice cold dH 2 O and were frozen. The isolated periplasmic fractions were then diluted to the same protein content based on OD 280 value with 20 mM Tris-HCl, 5 mM CaCl 2 , pH 7.9 buffer. Then, serial dilutions of the buffered periplasmic fractions were mixed with 20 nM final concentration of bovine chymotrypsin (Worthington, #LS001432) and incubated for 1 hour. Free residual enzyme concentration was determined by 250 μM Suc-Ala-Ala-Pro-Phe-PNA substrate (Bachem AG, #1067440).

Statistical analysis of results
Where statistical significance is indicated, two-tailed unpaired Student's t-test was performed. Difference between two tested datasets was considered significant if the calculated P value was lower than 0.05.

Analyzing the phylogenic distribution of ecotin harboring organisms
Species were selected from the PFAM database [68] with the keyword "ecotin". Only species with proper identification were analyzed, isolates such as "Pseudomonas sp." were excluded. Multiplicates representing multiple entries/ecotin domains from the same microorganism were removed. Ecotin-expressing species were considered pathogenic if they are listed in at least one of the following databases: Global RPH (globalrph.com/bacteria), Bode Science Center (www.bode-science-center.com/center/relevant-pathogens-from-a-z.html), or American Biological Safety Association (my.absa.org/Riskgroups). Number of species within individual phylogenic groups was looked up in the data of Catalogue of Life (www.catalogueoflife.org/col).