Antimicrobial Peptide Conformation as a Structural Determinant of Omptin Protease Specificity

ABSTRACT

Bacterial proteases contribute to virulence by cleaving host or bacterial proteins to promote survival and dissemination. Omptins are a family of proteases embedded in the outer membrane of Gram-negative bacteria that cleave various substrates, including host antimicrobial peptides, with a preference for cleaving at dibasic motifs. OmpT, the enterohemorrhagic Escherichia coli (EHEC) omptin, cleaves and inactivates the human cathelicidin LL-37. Similarly, the omptin CroP, found in the murine pathogen Citrobacter rodentium, which is used as a surrogate model to study human-restricted EHEC, cleaves the murine cathelicidin-related antimicrobial peptide (CRAMP). Here, we compared the abilities of OmpT and CroP to cleave LL-37 and CRAMP. EHEC OmpT degraded LL-37 and CRAMP at similar rates. In contrast, C. rodentium CroP cleaved CRAMP more rapidly than LL-37. The different cleavage rates of LL-37 and CRAMP were independent of the bacterial background and substrate sequence specificity, as OmpT and CroP have the same preference for cleaving at dibasic sites. Importantly, LL-37 was α-helical and CRAMP was unstructured under our experimental conditions. By altering the α-helicity of LL-37 and CRAMP, we found that decreasing LL-37 α-helicity increased its rate of cleavage by CroP. Conversely, increasing CRAMP α-helicity decreased its cleavage rate. This structural basis for CroP substrate specificity highlights differences between the closely related omptins of C. rodentium and E. coli. In agreement with previous studies, this difference in CroP and OmpT substrate specificity suggests that omptins evolved in response to the substrates present in their host microenvironments.

IMPORTANCE Omptins are recognized as key virulence factors for various Gram-negative pathogens. Their localization to the outer membrane, their active site facing the extracellular environment, and their unique catalytic mechanism make them attractive targets for novel therapeutic strategies. Gaining insights into similarities and variations between the different omptin active sites and subsequent substrate specificities will be critical to develop inhibitors that can target multiple omptins. Here, we describe subtle differences between the substrate specificities of two closely related omptins, CroP and OmpT. This is the first reported example of substrate conformation acting as a structural determinant for omptin activity between OmpT-like proteases.

INTRODUCTION

Bacterial proteases are key virulence factors involved in host-pathogen interactions. Omptins are outer membrane (OM) proteases commonly found in Gram-negative pathogens. They play important roles at the host-pathogen interface by processing or degrading a variety of host and bacterial proteins (1–10). Many omptins are found among pathogens of the Enterobacteriaceae family (11). The prototypical omptins are Escherichia coli OmpT and Yersinia pestis Pla, which inactivate antimicrobial peptides (AMPs) and activate plasminogen into plasmin, respectively (4, 6). Omptins are β-barrel proteases embedded in the OM and activated by lateral interaction with lipopolysaccharide (LPS) (12–15). Omptins possess a variety of attributes reminiscent of both serine and aspartic proteases that distinguish them into their own protease class (14, 16, 17). The omptin catalytic residues consist of two dyads nested within the extracellular active site groove. The Asp-His dyad is responsible for activating a nucleophilic water, which attacks the scissile peptide bond, and the Asp-Asp dyad stabilizes the catalytic intermediate (18, 19). Previous work has shown that omptins have strong sequence specificity for cleaving substrates at dibasic motifs (20–22). This preference for dibasic motifs is due to the negatively charged S1 and S1′ specificity pockets (12, 14). Despite these shared attributes, omptin family members cleave a variety of different substrates. The omptin active site is bordered by five less-conserved surface loops that contribute to omptin substrate specificity (12, 14). The influence of surface loops on substrate recognition was well characterized for E. coli OmpT and Y. pestis Pla, which process plasminogen into active plasmin to different extents. The shorter L3 and L4 loops of Pla appear to better accommodate plasminogen into the active site groove than the longer loops of OmpT (14, 23) (see Fig. S1 in the supplemental material).

One of the many initial dangers that bacterial pathogens must confront during infection is host-produced AMPs, which are part of the innate immune system (24). These small (20 to 50 amino acids), cationic, and amphipathic peptides are primarily released from the host's epithelial cells and neutrophils (25–27). The cationic nature of AMPs contributes to their ability to interact with the anionic bacterial membrane and form pores, resulting in bacterial lysis (28–30). Besides their ability to directly lyse bacteria, AMPs are also known for their immunomodulatory properties that contribute to controlling infection (31). In mammals, there are two major families of AMPs: the cathelicidins and the defensins. Cathelicidins most often adopt an α-helical conformation upon interaction with the bacterial membrane and are far more susceptible to protease degradation than defensins, which are stabilized by three disulfide bonds (32, 33). Both humans and mice produce a single cathelicidin, LL-37 and cathelicidin-related antimicrobial peptide (CRAMP), respectively, which are 46% identical at the amino acid level and contain two dibasic motifs (see Fig. S2 in the supplemental material) (34–38). Circular dichroism (CD) showed that both AMPs are unstructured in water, although they adopt an α-helical conformation in the presence of the organic solvent 2,2,2-trifluoroethanol (TFE) or upon interaction with membrane phospholipids (35, 39, 40). The conformation of cathelicidins in host physiological fluids remains unclear, since these fluids are incompatible with CD experiments.

Despite the importance of AMPs in the host response, pathogens are capable of resisting the actions of AMPs. Bacteria have developed an extensive array of mechanisms to evade AMP killing. These include the use of LPS modifications, efflux pumps, capsules, and proteases (24). For example, the secreted proteases Staphylococcus aureus aureolysin and Pseudomonas aeruginosa elastase were shown to degrade and inactivate LL-37 (41, 42). As OM proteases with extracellular active sites, omptins are well positioned to take part in AMP cleavage and inactivation. An early study showed that OmpT readily degrades the AMP protamine (43). More recently, we showed that LL-37 is cleaved and inactivated to different extents by OmpT present in enterohemorrhagic (EHEC), enteropathogenic (EPEC), and uropathogenic E. coli (4, 5, 32). The enteric murine pathogen Citrobacter rodentium and the human pathogens EHEC and EPEC share numerous virulence factors and cause similar intestinal lesions (44). For these reasons, C. rodentium is used as a surrogate model to study EHEC and EPEC pathogenesis (45). C. rodentium produces the omptin CroP, which is 74% identical to E. coli OmpT at the amino acid level, suggesting similar substrate specificity (3, 17). In agreement, we showed that CroP degrades CRAMP and is a poor plasminogen activator, indicating that CroP is an OmpT-like protease (3, 17). In this study, we compared the substrate specificities of C. rodentium CroP and EHEC OmpT against the murine and human cathelicidins. We found that CroP degrades LL-37 and CRAMP more rapidly when these AMPs are in an unstructured conformation. Strikingly, α-helicity did not affect the rates of cleavage of these AMPs by OmpT. This study reveals an important difference in substrate recognition between C. rodentium CroP and E. coli OmpT, likely arising from differences within their active sites that may reflect the environmental pressures of specific host microenvironments in driving the molecular evolution of omptins.

MATERIALS AND METHODS

Bacterial strains, growth conditions, and reagents.

The bacterial strains and plasmids used in this study are listed in Table 1. DNA purification and genetic transformation were carried out according to standard procedures (46). Bacteria were grown overnight with aeration at 37°C in Luria-Bertani (LB) broth, subcultured 1/100, and grown to an optical density at 600 nm (OD₆₀₀) of 0.5 in N-minimal medium (pH 7.5) supplemented with 0.2% glucose and 1 mM MgCl₂, as previously described (4). When appropriate, chloramphenicol (Cm; 30 μg/ml) was added to the medium. LL-37 and CRAMP were synthesized with a purity greater than 90% (BioChemia). The fluorescence resonance energy transfer (FRET) substrate 2Abz-SLGRKIQIK(Dnp)-NH₂ was purchased from AnaSpec. CroP was purified and prepared in phosphate-buffered saline (PBS; pH 7.4), as previously described (17).

TABLE 1.

Bacterial strains and plasmids

Strain or plasmid	Description	Reference
Strains
C. rodentium
Wild type	DBS100 (ATCC 51459)	56
ΔcroP mutant	DBS100 ΔcroP	3
ΔcroP(pEHompT) mutant	DBS100 ΔcroP expressing ompT from pEHompT	This study
ΔcroP(pYCcroP) mutant	DBS100 ΔcroP expressing croP from pYCcroP	17
E. coli (EHEC)
Wild type	EDL933 O157:H7	57
ΔompT mutant	EDL933 ΔompT	4
ΔompT(pEHompT) mutant	EDL933 ΔompT expressing ompT from pEHompT	4
Plasmids
pYCcroP	pACYC184 containing croP under the control of its native promoter	17
pEHompT	pACYC184 containing ompT under the control of its native promoter	4

Proteolytic cleavage of AMPs.

AMP cleavage assays were carried out as previously described with a few modifications (4, 5). Either the specified bacterial strain (∼3 × 10⁸ CFU/ml) or purified CroP (14 μg/ml) was incubated with the indicated AMP (400 μg/ml). Cleavage assays were carried out in the indicated buffers (pH 7.4) at room temperature with samples taken at the designated time points. As negative controls, AMPs were incubated with buffer only for 60 min. Samples resolved by Tris-Tricine PAGE were mixed with Tris-Tricine sample buffer (Bio-Rad) prior to boiling for 5 min. Peptide degradation products were resolved on 10 to 20% Tris-Tricine SDS-PAGE gels (Bio-Rad), fixed with 5% glutaraldehyde, and stained with Coomassie blue G250.

Liquid chromatography (LC)-MS/MS.

For identification of the CroP cleavage sites, LL-37 and CRAMP were incubated with purified CroP, as described above. Peptide digests were desalted using C₁₈ ZipTip pipette tips (Millipore), dried in a SpeedVac, and resuspended in 50 μl of 1% formic acid, 1% acetonitrile, and 98% water. Peptide digests (5 μl) were subjected to reversed-phase column chromatography (self-packed C₁₈ column, 75-μm internal diameter by 150-mm length) coupled to an Easy-nLC II system (Thermo Fisher Scientific). Elution was performed using a linear gradient of solvent A (0.2% formic acid and 99.8% water) and solvent B (0.2% formic acid, 80% acetonitrile, and 19.8% water) at a flow rate of 600 nl/min. Tandem mass spectrometry (MS/MS) data acquisition was performed on an LTQ Orbitrap Velos mass spectrometer (Thermo Fisher Scientific) equipped with a nanoelectrospray ion source. Data acquisition was accomplished using a seven-scan event cycle comprised of a full-scan MS for scan event 1. The mass resolution for MS was set to 60,000 (at m/z 400) and used to trigger the 10 additional MS/MS events acquired in parallel in the linear ion trap for the six most intense ions. The mass-over-charge ratio range was from 300 to 2,000 for MS scanning with a target value of 1,000,000 charges and from ∼1/3 of parent m/z ratio to 2,000 for MS/MS scanning with a target value of 10,000 charges. The data-dependent scan events used a maximum ion fill time of 100 ms and 1 microscan. Target ions already selected for MS/MS were dynamically excluded for 25 s. The peak list files were generated using Extract-MSn software (minimum mass set to 600 Da, maximum mass set to 6,000 Da, and no grouping of MS/MS spectra). Protein database searching was performed with Mascot 2.3 (Matrix Science). The mass tolerances for precursor and fragment ions were set to 10 ppm and 0.6 Da, respectively.

FRET activity assay.

The activity of CroP under the various buffering conditions was assessed by incubating purified CroP (0.35 μg/ml) with the FRET substrate (4 μg/ml), as previously described (17). Triplicate reactions were monitored with an excitation at 325 nm and measured fluorescence at 430 nm in a black 96-well microtiter plate (Costar). Reactions were carried out over 60 min with shaking between measurements on a BioTek FLx800 multidetection microplate reader equipped with an injector. Initial background fluorescence was subtracted from values for normalization. Relative activity was determined by calculating the area under the curve relative to that of a FRET assay performed in PBS.

Circular dichroism experiments.

CD experiments were performed on a Jasco J-810 spectropolarimeter (Easton, MD) using a quartz cuvette with a path length of 0.1 cm. Spectra were recorded from 260 to 195 nm with 200 μg/ml of LL-37 and CRAMP in the indicated buffers. Each sample was scanned three times at 20°C using a bandwidth of 1 nm, a time response of 2 s, and a scan rate of 100 nm/min. Spectra, representing the averages of three scans, were corrected by subtracting the background spectrum of the appropriate buffer, and values were converted from ellipticity to mean residue ellipticity (MRE; degree × cm² × dmol⁻¹).

RESULTS

EHEC OmpT cleaves LL-37 and CRAMP at similar rates.

First, we compared the abilities of OmpT to cleave LL-37 and CRAMP by incubating both AMPs in PBS in the presence of the EHEC wild-type, ΔompT, and ΔompT(pEHompT) strains. Cleavage of AMPs was monitored over a 60-min time course, and peptides were resolved by Tris-Tricine SDS-PAGE. Both LL-37 and CRAMP were cleaved by EHEC wild-type and ΔompT(pEHompT) strains as indicated by the formation of cleavage products over time (Fig. 1). In the presence of EHEC wild type, LL-37 and CRAMP cleavage products were observed at 15 min, and cleavages of LL-37 and CRAMP occurred at approximately similar rates over 60 min (Fig. 1). The EHEC ΔompT(pEHompT) strain overexpressing ompT completely cleaved LL-37 and CRAMP within 30 min (Fig. 1). As expected, LL-37 and CRAMP cleavage products were not observed for the EHEC ΔompT strain (Fig. 1), indicating that cleavage of LL-37 and CRAMP is ompT dependent. These data show that EHEC OmpT cleaves LL-37 and CRAMP at similar rates.

FIG 1.

Cleavage of LL-37 and CRAMP by EHEC OmpT. LL-37 or CRAMP was incubated with the indicated EHEC strain or PBS alone (Ctl.) for the indicated time. Resulting peptide cleavage products were resolved by 10 to 20% Tris-Tricine SDS-PAGE and visualized with Coomassie blue staining.

C. rodentium CroP cleaves LL-37 and CRAMP at different rates.

Next, we compared the cleavages of LL-37 and CRAMP by the C. rodentium wild-type, ΔcroP, and ΔcroP(pYCcroP) strains. Strikingly, we found that the C. rodentium wild-type and ΔcroP(pYCcroP) strains cleaved CRAMP much faster than LL-37 (Fig. 2). Complete degradation of CRAMP by C. rodentium wild-type and ΔcroP(pYCcroP) strains was observed at 60 and 30 min, respectively. In contrast, LL-37 was only partially degraded by both strains at 60 min. Cleavage products were not detected when peptides were incubated with the C. rodentium ΔcroP strain, indicating that cleavage of both AMPs by C. rodentium is mediated by CroP. These data suggest that either CroP or interference from the C. rodentium background results in the differential cleavage observed. To determine whether the C. rodentium background affects AMP cleavage, pEHompT was transformed into the C. rodentium ΔcroP strain. As shown in Fig. 2, the C. rodentium ΔcroP(pEHompT) strain cleaved LL-37 and CRAMP at similar rates and cleavage was complete by 30 min. These data show that the C. rodentium background does not affect the cleavage rates of LL-37 and CRAMP by OmpT. Taken together, these data suggest that the intrinsic properties of CroP are responsible for the more rapid degradation of CRAMP than of LL-37.

FIG 2.

Cleavage of LL-37 and CRAMP by C. rodentium CroP. LL-37 or CRAMP was incubated with the indicated C. rodentium strain or PBS alone (Ctl.) for the indicated time. Resulting peptide cleavage products were resolved by 10 to 20% Tris-Tricine SDS-PAGE and visualized with Coomassie blue staining.

Purified CroP cleaves murine CRAMP more rapidly than human LL-37.

Having observed a difference in the rates of cleavage of LL-37 and CRAMP by C. rodentium CroP in whole cells, we sought to determine whether this difference also occurs when peptides are incubated with purified CroP. LL-37 and CRAMP were incubated in the presence of purified CroP, and cleavage of both AMPs at various time points was analyzed by Tris-Tricine SDS-PAGE. When incubated with the purified protease, cleavage products of both AMPs were observed within 5 min (Fig. 3). However, purified CroP cleaved CRAMP at a much higher rate than LL-37. Consistent with previous results, CroP almost completely degraded CRAMP within 5 min (17). In contrast, LL-37 cleavage by purified CroP was incomplete at 60 min (Fig. 3). Cleavage was CroP dependent, as LL-37 and CRAMP incubated with buffer alone were stable over 60 min (Fig. 3). Altogether, these data clearly indicate that CroP is able to degrade both AMPs, although it degrades murine CRAMP more rapidly than human LL-37.

FIG 3.

Cleavage of LL-37 and CRAMP by purified CroP. LL-37 or CRAMP was incubated with CroP and LPS or LPS alone (Ctl.) for the indicated time. Resulting peptide cleavage products were resolved by 10 to 20% Tris-Tricine SDS-PAGE and visualized with Coomassie blue staining.

LC-MS/MS analysis of LL-37 and CRAMP degradation products.

To further analyze the AMP degradation products generated by CroP, AMP samples incubated with purified CroP for 5 and 30 min were analyzed by LC-MS/MS. AMP samples incubated with buffer alone for 60 min contained the expected full-length LL-37 and CRAMP (data not shown). Purified CroP cleaved LL-37 at both dibasic motifs (R₇-K₈ and K₁₈-R₁₉) (Fig. 4), as previously reported for EHEC OmpT (4). The relative abundance of the N-terminal cleavage product (LLGDFFR) was stable between 5 and 30 min. In contrast, the abundance of the intermediate cleavage product (KSKEKIGKEFKRIVQRIKDFLRNLVPRTES) decreased between 5 and 30 min, while the abundance of the C-terminal cleavage product (RIVQRIKDFLRNLVPRTES) increased (see Fig. S3 in the supplemental material). These data indicate that CroP first cleaves the dibasic motif R₇-K₈ and subsequently cleaves the K₁₈-R₁₉ motif. Together, these data show that CroP and OmpT cleave LL-37 at the same dibasic sites.

FIG 4.

Analysis of LL-37 and CRAMP cleavage products by purified CroP. CroP-dependent LL-37 and CRAMP cleavage products were detected by liquid chromatography and analyzed by MS/MS. Shown is a schematic of the LL-37 and CRAMP amino acid sequences. Dibasic motifs are highlighted in red, major cleavage products are indicated by solid arrows, and a minor cleavage product is indicated by a dashed arrow.

Murine CRAMP contains the two dibasic motifs R₄-K₅ and K₁₅-K₁₆ (see Fig. S2 in the supplemental material). CroP primarily cleaved CRAMP at the second dibasic motif (K₁₅-K₁₆) (Fig. 4). Unlike what was observed with LL-37, CroP did not cleave CRAMP at the N-terminal dibasic motif (R₄-K₅) (Fig. 4). The minor cleavage product (KLVPQPEQ) was detected and revealed the monobasic motif Q₂₇-K₂₈ as a secondary cleavage site (Fig. 4). Analysis of the LC-MS/MS spectra showed no difference in the relative abundances of cleavage products at 5 and 30 min (see Fig. S4 in the supplemental material), confirming the rapid cleavage of CRAMP by CroP observed by Tris-Tricine SDS-PAGE (Fig. 2 and 3). Together, these LC-MS/MS data suggest subtle differences between the cleavages of LL-37 and CRAMP by CroP. In addition, they confirm that CroP cleaves CRAMP more rapidly than LL-37.

Influence of AMP conformation on CroP cleavage.

To determine the reasons for the differential cleavage of LL-37 and CRAMP by CroP, CD experiments in the far-UV range were performed with both AMPs. A previous study has shown that LL-37 is α-helical in PBS (39). Our data confirmed this result; the CD spectrum of LL-37 in PBS displays a maximum at 200 nm and minima at 208 and 222 nm, which are typical of α-helices (Fig. 5A). In sharp contrast, the CD spectrum of CRAMP in PBS exhibited a minimum at 200 nm that is indicative of an unstructured conformation (Fig. 5B). These data show that LL-37 and CRAMP have different conformations in PBS. They also suggest that recognition of these AMPs by CroP may be influenced by the conformation of the substrate.

FIG 5.

Effect of buffer on peptide conformation and CroP activity. (A and B) Circular dichroism of LL-37 (A) and CRAMP (B) was measured in PBS (black), 10 mM phosphate buffer (red), 1 mM phosphate buffer (blue), and distilled water (dH₂O) (green) in the far-UV spectrum from 200 to 260 nm. (C and D) LL-37 (C) and CRAMP (D) were incubated for 15 min with or without CroP in the indicated buffers at pH 7.4. Resulting peptide cleavage products were resolved by 10 to 20% Tris-Tricine SDS-PAGE and visualized with Coomassie blue staining.

CroP preferentially cleaves unstructured AMPs.

Previous work showed that both the conformation and antimicrobial activity of LL-37 are salt dependent (39, 47). To modulate the conformation of LL-37 and CRAMP, different buffers (PBS, 10 mM and 1 mM sodium phosphate, or H₂O [pH 7.4]) of decreasing ionic strengths were used. CD spectra were recorded, and cleavage of AMPs by CroP was visualized by Tris-Tricine SDS-PAGE (Fig. 5). LL-37 adopted an α-helical conformation in 10 mM sodium phosphate, although the reduced minimum at 208 nm suggests slightly decreased α-helicity compared to PBS (Fig. 5A). In contrast, CD spectra of LL-37 in 1 mM phosphate buffer or H₂O lacked the minima at 208 and 222 nm and exhibited a minimum at 200 nm, indicating a transition from an α-helical to an unstructured conformation. The difference in intensity of the minima at 200 nm may suggest that LL-37 is somewhat less structured in H₂O than in 1 mM sodium phosphate (Fig. 5A). To determine whether CroP activity is affected by the different buffers, the activity of CroP in these buffers was measured using the FRET substrate 2Abz-SLGRKIQIK(Dnp)-NH₂, which is cleaved by omptins (4). Similar FRET activity was observed in PBS and 10 mM sodium phosphate (see Fig. S5 in the supplemental material). In contrast, the FRET activity was reduced by approximately 25% and 60% in 1 mM phosphate buffer and H₂O, respectively (see Fig. S5). These data indicate that lowering ionic strength decreases CroP activity. Incubation of LL-37 with CroP for 15 min in PBS generated the larger LL-37 cleavage product and a smaller cleavage product in low abundance (Fig. 5C). The abundance of this smaller cleavage product clearly increased when the assay was carried out in 10 and 1 mM sodium phosphate buffers and H₂O (Fig. 5C). Importantly, the intensity of this smaller fragment was markedly increased and correlated with a decreased intensity of the larger fragment when the assay was performed in H₂O (Fig. 5C). Taking into account the lower activity of CroP in 1 mM sodium phosphate and H₂O (see Fig. S5), these data clearly indicate that CroP activity against LL-37 increases with decreasing ionic strength. In addition, they suggest that the increased CroP activity correlates with the decreased α-helicity of LL-37. Thus, these data suggest that the transition of LL-37 from an α-helical to an unstructured conformation results in increased cleavage by CroP.

Similar experiments were performed with CRAMP. Regardless of the buffering system used, CD spectra of CRAMP exhibited a minimum at 200 nm, indicating that CRAMP remains unstructured (Fig. 5B). As expected from these data, incubation of CRAMP with CroP for 15 min resulted in its complete cleavage in all buffers used (Fig. 5D). Taken together, these data suggest that CroP has a strong preference for cleaving unstructured AMPs.

Increased α-helicity leads to a decrease in AMP cleavage by CroP.

TFE is a helix-promoting compound that is extensively used to characterize α-helical peptides (35, 48). Using the FRET activity assay, we found that TFE did not interfere with CroP activity when present at concentrations of 8% and lower (see Fig. S5 in the supplemental material). The addition of 4 and 8% TFE to LL-37 in PBS resulted in increases in the negative peaks at 208 and 222 nm, indicating slightly increased α-helical content of LL-37 (Fig. 6A). Upon incubation of LL-37 with CroP in the presence of 4 and 8% TFE, we observed a decreased abundance of the larger LL-37 cleavage product, indicating that TFE slightly decreases the cleavage rate of LL-37 by purified CroP (Fig. 6C). For CRAMP, the addition of 4% TFE induced the appearance of a small shoulder near 222 nm in the CD spectrum (Fig. 6B). Importantly, the addition of 8% TFE induced a more drastic change in the CRAMP CD spectrum: the minimum at 200 nm was replaced by minima at 208 and 222 nm. This indicates that in the presence of small amounts of TFE, CRAMP transitions from an unstructured to an α-helical conformation (Fig. 6B). The effect of altered CRAMP conformation on CroP activity was next assessed by cleavage assay. The presence of 4 and 8% TFE resulted in decreased CRAMP cleavage by CroP compared to that with buffer alone. CroP was still able to cleave CRAMP in the presence of 4% TFE, although to a slightly lesser extent than in the absence of TFE (Fig. 6D). Importantly, CroP cleavage of CRAMP was drastically reduced by the addition of 8% TFE (Fig. 6D). Together, these experiments show that the presence of TFE increases the α-helicity of both AMPs and, in turn, decreases the ability of CroP to cleave these AMPs. This is particularly obvious for CRAMP, when 8% TFE induces an α-helical conformation and virtually abolishes CroP cleavage. Therefore, we conclude that α-helicity is a structural determinant that negatively impacts the ability of CroP to cleave these AMPs.

FIG 6.

Effect of AMP α-helicity on CroP activity. (A and B) Circular dichroism of LL-37 (A) and CRAMP (B) was measured in PBS with 4% TFE (red), with 8% TFE (blue), and without TFE (black) within the far-UV spectrum from 200 to 260 nm. (C and D) LL-37 (C) and CRAMP (D) were incubated with or without CroP for 5 min in the presence of the indicated amount of TFE. Resulting peptide cleavage products were resolved by 10 to 20% Tris-Tricine SDS-PAGE and visualized with Coomassie blue staining.

DISCUSSION

Omptin proteases are key virulence factors involved in the pathogenicity of several Gram-negative bacterial pathogens (6, 7, 9, 49). All omptins characterized so far preferentially cleave substrates at dibasic motifs (8). Omptins are divided into the OmpT-like and Pla-like subfamilies, based on amino acid sequence identity and ability to proteolytically activate plasminogen. The OmpT-like omptins, including E. coli OmpP and C. rodentium CroP, are 70 to 80% identical at the amino acid level. Although subtle differences between E. coli OmpT and OmpP were reported (21, 22), it is unclear whether OmpT-like proteases cleave the same substrates. By comparing the degradation of cathelicidins by E. coli OmpT and C. rodentium CroP, we found that peptide conformation is a structural determinant for CroP activity. CroP showed a strong preference for unstructured peptides and poorly cleaved α-helical cathelicidins. In contrast, OmpT cleaved the human and murine cathelicidins regardless of their α-helicity. This study reveals that OmpT-like proteases may vary in their substrate specificity and suggests significant structural differences between the active site grooves of CroP and OmpT, despite their high amino acid sequence identity.

Omptins preferentially cleave substrates at dibasic motifs (P1 and P1′ positions). This specificity is determined by the conserved and negatively charged S1 (E₂₇ and D₂₀₈) and S1′ (D₉₇) specificity subsites (12, 14). We previously reported that EHEC OmpT cleaves LL-37 at the dibasic motifs R₇-K₈ and K₁₈-R₁₉ (4). Here, LC-MS/MS analyses showed that purified CroP cleaves LL-37 at the same dibasic motifs previously reported for OmpT (Fig. 4). In addition, analysis of the cleavage products at 5 and 30 min revealed that CroP cleaves the R₇-K₈ motif before the K₁₈-R₁₉ motif. This preferential cleavage of the R₇-K₈ motif was not observed for EHEC OmpT, likely because cleavage of LL-37 by the EHEC ΔompT(pEHompT) strain was much faster and fully cleaved LL-37 fragments were obtained within 5 min (4). Though CRAMP also contains two dibasic motifs (R₄-K₅ and K₁₅-K₁₆), we found that most of the cleavage occurred at the K₁₅-K₁₆ motif (Fig. 4). In addition, a minor cleavage site (Q₂₇-K₂₈) was identified (Fig. 4). This indicates that CroP, like other omptins, can also cleave substrates at monobasic sites (20, 21). The ability of CroP to cleave at this monobasic site is supported by the presence of a phenylalanine at the P2 position (F₂₆-Q₂₇-K₂₈), which is also a preference of OmpT (20). The reason why CroP does not cleave CRAMP at the N-terminal R₄-K₅ dibasic motif is unclear. Several factors can account for this lack of cleavage. First, the R₄-K₅ motif is close to the N terminus of CRAMP (Fig. 4). Thus, the amino group of the N-terminal residue, which is positively charged at physiological pH, may interfere with binding of the R₄-K₅ motif into the CroP active site. Second, the R₄-K₅ motif is followed by two glycines (G₆-G₇) at positions P2′ and P3′. The S2′ pocket of OmpT poorly tolerates a Gly at P2′ (19, 20). The conserved S2′ pocket of CroP may also contribute to the lack of cleavage at the R₄-K₅ motif of CRAMP. Together, these data indicate that CroP and OmpT share substrate sequence specificity as a result of the conserved specificity subsites.

The secondary structure of α-helical AMPs in physiological fluids remains unclear. It is acknowledged that most α-helical AMPs are unstructured in aqueous solution and adopt an α-helical conformation upon interaction with the bacterial membrane (40). Although unstructured in pure water, LL-37 was reported to be α-helical in most physiological buffers and in solutions containing the ion compositions of biological fluids (39). Our data clearly show that LL-37 is α-helical and CRAMP is unstructured in PBS (Fig. 5). This difference in conformation appears to be responsible for the differences in rates of AMP cleavage by CroP. The putative α-helical conformation of LL-37 in human biological fluids is consistent with the possibility that the OmpT active site groove is wide enough to fit a structured α-helix as well as unstructured peptides. This structural feature of OmpT is consistent with the OmpT cleavage of both LL-37 and CRAMP (Fig. 1). In contrast, our cleavage data suggest that the CroP active site restricts access to α-helical AMPs but is accessible to unstructured AMPs. This difference may indicate that the 26% difference in amino acid sequence (74% identity) between CroP and OmpT is important for structural specificity. The sequence of the five surface loops of omptins has been shown to determine substrate specificity (23). Therefore, the amino acids accounting for the difference between CroP and OmpT in substrate recognition are likely within the surface loops. Although the amino acid sequences of these loops are relatively well conserved between CroP and OmpT, a noticeable exception is the loop L4 of OmpT that contains two additional residues compared to CroP (see Fig. S1 in the supplemental material). The absence of these residues in CroP may contribute to a narrower CroP active site groove, resulting in the exclusion of α-helical substrates.

In a previous study, we showed that the serine protease inhibitor aprotinin inhibits CroP in a competitive manner (17). Lower concentrations of aprotinin were required to inhibit CroP activity than OmpT activity in whole-cell experiments (17). In light of our current findings, the difference in observed potency of aprotinin may be attributed to differences in the ability of the extended inhibitory loop of aprotinin to fit within the active site grooves of CroP and OmpT. In the present study, we show that the α-helicity of the substrate greatly influences its rate of cleavage by CroP but not by OmpT. Together, these studies suggest significant structural differences between the CroP and OmpT active sites, despite the 74% amino acid identity of the CroP and OmpT sequences and conservation of the specificity subsites.

These differences between the CroP and OmpT active sites may reflect the molecular adaptation of the E. coli and C. rodentium omptins to their respective infectious niches. This study may suggest that LL-37 and CRAMP adopt different secondary structures in their respective organisms, in vivo. C. rodentium is a natural mouse pathogen with strict tissue and host tropisms (50). It is possible that C. rodentium CroP evolved to be more selective for cleaving unstructured CRAMP over other peptides. In contrast, E. coli pathogenic and commensal strains have much broader tissue and host tropisms (4, 5, 51, 52). The broad substrate specificity of OmpT is likely beneficial for bacterial fitness in the various microenvironments encountered. Future work will reveal whether this structural specificity is unique to CroP or conserved among other OmpT-like proteases.

The different CroP and OmpT structural specificities emphasize the importance of molecular adaptation that fine-tuned omptins through the acquisition of different substrate specificities. Adaptive molecular evolution is not unprecedented among omptins. Epo, the omptin of the plant pathogen Erwinia pyrifoliae, is highly similar to Pla in amino acid sequence (77%) and similar to Pla in the ability to degrade serpins (53, 54). However, Epo failed to activate plasminogen due to sequence variation in two surface loops (53). Another example of molecular adaptation is found in Pla. In a recent study, Zimbler et al. showed that a single-amino-acid modification within Pla (I259T), which is found in modern Y. pestis strains, enhances the invasive capacity of Y. pestis during bubonic plague (55). Thus, it appears that the plasticity of the omptin active site contributes to the ability of omptins to cleave the different substrates that their host bacteria encounter.

In summary, the activity of omptins is controlled by their localization to the OM, activation through lateral interactions with LPS, and sequence substrate specificity. This study reveals that the substrate conformation is a structural determinant that influences cleavage rates. The differences observed for the closely related C. rodentium CroP and EHEC OmpT suggest that the OmpT-like proteases may have diverged to cleave different physiological substrates. Identification of these physiological substrates is critical for a better understanding of the pathogenicity of omptin-bearing bacteria.