Proteinases of common pathogenic bacteria degrade and inactivate the antibacterial peptide LL-37

Authors


Summary

Effectors of the innate immune system, the anti-bacterial peptides, have pivotal roles in preventing infection at epithelial surfaces. Here we show that proteinases of the significant human pathogens Pseudomonas aeruginosa, Enterococcus faecalis, Proteus mirabilis and Streptococcus pyogenes, degrade the antibacterial peptide LL-37. Analysis by mass spectrometry of fragments generated by P. aeruginosa elastase in vitro revealed that the initial cleavages occurred at Asn-Leu and Asp-Phe, followed by two breaks at Arg-Ile, thus inactivating the peptide. Proteinases of the other pathogens also degraded LL-37 as determined by SDS-PAGE. Ex vivo, P. aeruginosa elastase induced LL-37 degradation in human wound fluid, leading to enhanced bacterial survival. The degradation was blocked by the metalloproteinase inhibitors GM6001 and 1, 10-phenantroline (both of which inhibited P. aeruginosa elastase, P. mirabilis proteinase, and E. faecalis gelatinase), or the inhibitor E64 (which inhibited S. pyogenes cysteine proteinase). Additional experiments demonstrated that dermatan sulphate and disaccharides of the structure [ΔUA(2S)-GalNAc(4,6S)], or sucroseoctasulphate, in-hibited the degradation of LL-37. The results indicate that proteolytic degradation of LL-37 is a common virulence mechanism and that molecules which block this degradation could have therapeutic potential.

Introduction

Epithelial surfaces of various organisms are continuously exposed to bacteria. During recent years the innate immune system, based on antibacterial peptides originally described in silk worms (Steiner et al., 1981), has been attributed important roles in the initial clearance of bacteria at biological boundaries susceptible to infection (for references see reviews by Selsted and Ouellette, 1995; Lehrer and Ganz, 1999; Schröder and Harder, 1999; Boman, 2000). Cathelicidins, consisting of a conserved N-terminal domain and a C-terminal antibacterial region, constitute one family of such peptides. In humans, the cathelicidin hCAP-18 is processed by proteinase 3 (a serine proteinase) to generate the active peptide LL-37, which exerts antibacterial activity against both Gram-negative and Gram-positive bacteria (Sörensen et al., 2001). LL-37 has been isolated from neutrophils (Zanetti et al., 1995) and subpopulations of lymphocytes and monocytes (Agerberth et al., 2000). This peptide is also found in seminal plasma (Malm et al., 2000), in the lung (Bals et al., 1998; Agerberth et al., 1999), and in keratinocytes during inflammation (Frohm et al., 1997). Furthermore, it has been reported that LL-37 has synergistic effects with α-defensin, another major antibacterial peptide found in the azurophilic granules of neutrophils (Nagaoka et al., 2000). The significance of cathelicidins for bacterial clearance is exemplified by recent findings indicating that the mouse antibacterial peptide CRAMP protects the skin from invasive bacterial infection (Nizet et al., 2001).

Pathogens, however, appear to overcome the innate immune defence systems giving rise to various infections. In some conditions, such as chronic ulcers of the skin, bacteria such as P. aeruginosa, E. faecalis, and P. mirabilis, manage to infect and persist for long periods of time. During infection, these pathogens, which are frequently found in chronic ulcers (P. aeruginosa; 20–30%, E. faecalis 80%, P. mirabilis 15%) (Hansson et al., 1995), use a combination of virulence strategies. For instance, release of various proteinases, such as elastase and alkaline proteinase of P. aeruginosa (Morihara et al., 1965; Fukushima et al., 1989; Okuda et al., 1990), may modulate host responses involving kallikreins, coagulation factors, complement, cytokines, and antiproteinases (for a review, see Travis et al., 1995). Interestingly, degradation of complement, antiproteinases, and matrix components in vitro by the proteinases, closely resembles the degradation pattern of these molecules seen in wound fluid in vivo (Schmidtchen, 2000, and unpublished data).

The well known bacterial persistence in chronic skin ulcers and the fact that antiproteinase deprivation and uncontrolled proteolysis characterize these ulcers raise the possibility that bacterial proteinases may affect antibacterial peptides. In this study, data are presented showing that bacterial proteinases, such as P. aeruginosa elastase, rapidly degrade and inactivate LL-37, leading to enhanced bacterial survival. Furthermore, we show that this proteolytic degradation may be modulated by proteinase inhibitors and by sulphated poly- and di-saccharides.

Results

Bacterial proteinases degrade and inactivate LL-37

To examine the bactericidal effects of intact LL-37, the activity of the peptide against S. pyogenes, E. faecalis, P. mirabilis and P. aeruginosa was determined. The bacterial strains were separately incubated with 0–20 µg ml−1 of LL-37. Pseudomonas aeruginosa, E. faecalis, P. mirabilis, and S. pyogenes were effectively killed by the peptide at or above concentrations of 2–10 µg ml−1(Fig. 1). We then investigated the effects of P. aeruginosa elastase on LL-37. Initial experiments indicated that the peptide was rapidly degraded, with no detectable intermediate peptides on SDS-PAGE (not shown). Thus, a titration was performed and proteinase amounts yielding detectable amounts of LL-37 degradation products were used in further experiments. The addition of 30 mU of proteinase (∼25 ng protein) to 10 µg peptide (molar ratio enzyme to peptide ∼1:2500) yielded intermediate fragments of approximately 3–4 kDa after an incubation period of 1–30 min (Fig. 2A, upper panel). Longer incubations yielded no visible peptides in the SDS-PAGE system used (16.5%). The proteinase amounts used correspond to enzyme levels found in ∼5–10 µl of conditioned media from a P. aeruginosa culture grown overnight in TH-medium (as determined by an azocasein assay). Next, the effect of these LL-37 intermediates was investigated. In bactericidal assays, E. faecalis bacteria were incubated with the peptide fractions (corresponding to 10 µg ml−1), and those containing the 3–4 kDa forms (Fig. 2A, 5–30 min incubations) exerted residual bactericidal activity (40–60% bacteria killed), whereas subsequent incubations (1–20 h) contained no such activity (not shown). The finding that the 30 min incubation, which was devoid of intact peptide, exerted residual activity (∼40% of control), suggests that the larger ∼4 kDa degradation product is bactericidal. Finally, structural analysis of LL-37 fragments generated by P. aeruginosa elastase was performed. Liquid chromatography–mass spectrometry time of flight (LC-MS TOF) and MS-MS TOF analysis of the material corresponding to the 1 min to 1 h degradations (Fig. 2A) yielded major peptides of masses 3708.859, 3178.637, 2822.450, 2326.169 and 1401.748 Da. A correspondence between the peptide intensities of the major 3–4 kDa bands on electrophoresis (Fig. 2A) and the peak heights on mass spectrometry was noted (not shown). The P. aeruginosa cleavage points were determined and are indicated (Fig. 2B). Interestingly, the proteinase acted on a region of LL-37 that has been attributed with antibacterial activity (Oren et al., 1999).

Figure 1.

Bactericidal effect of LL-37. 2 × 10 6 colony-forming units (CFU) ml −1 of E. faecalis (□), S. pyogenes (s), P. mirabilis (–), or P. aeruginosa ( ● ) were incubated with LL-37 at indicated peptide concentrations for 2 h ( S. pyogenes, P. mirabilis and E. faecalis ) or 4 h ( P. aeruginosa ). CFUs were determined, and representative experiments are shown.

Figure 2.

Pseudomonas aeruginosa elastase degrades and inactivates LL-37.

A. LL-37 (10 µg) was incubated with 30 mU of P. aeruginosa elastase (PAELA) for various periods of time (shown on the x -axis). Equal aliquots of the incubations were then analysed by SDS-PAGE (16.5% Tris-tricine gel). A volume corresponding to 2 µg of LL-37 was loaded on the gels. Molecular mass markers are indicated to the left.

B. The LL-37 degradation products were analysed by mass spectrometry and the cleavage points are indicated. The proposed antibacterial region of LL-37 is underlined below.

In C and D, the bactericidal effect of LL-37 in the absence or presence of PAELA was determined.

C. 2 × 10 6 colony-forming units (CFU) ml −1 of P. aeruginosa were incubated with PAELA only (75, 150 or 300 mU), with LL-37 (0.2 µg) only, or with PAELA (70, 150 or 300 mU) and LL-37 (0.2 µg) at 37°C for 4 h (50 µl reactions). Plates were streaked and CFUs determined, and the inhibition of LL-37 mediated bacterial killing by added PAELA is indicated. A representative experiment (of three) is shown.

D. The inset illustrates the experiment in C using 150 mU of PAELA, 0.2 µg LL-37, or 150 mU PAELA and 0.2 µg LL-37 (PAELA + LL-37). 150 mU amounts to the PAELA secreted in 50 µl by overnight P. aeruginosa cultures.

Having proven that fragmentation of LL-37 leads to loss of antibacterial activity (using E. faecalis as the test organism), we examined whether proteinase production was protective for P. aeruginosa. In this context, various experimental approaches were tested. Analyses of P. aeruginosa growth in TH or LB medium before and after addition of proteinase inhibitors (GM6001 or 1, 10-phenantroline), showed that these, at levels necessary for inhibition of elastase (50–100 µM), inhibited bacterial growth, which prevented us from performing MIC analyses under conditions necessary for P. aeruginosa growth. The effect of supplementing P. aeruginosa with proteinase was therefore investigated. Addition of elastase, corresponding to the concentrations found in overnight cultures of P. aeruginosa, resulted in significant reduction of LL-37-mediated bacterial killing (Fig. 2C and D). Next, to show that degradation abolished bacterial binding, the uptake of fluorescent labelled LL-37 by bacteria was studied using fluorescence microscopy and flow cytometry. As shown inFig. 3B, Texas red-conjugated LL-37 rapidly bound to and stained P. aeruginosa bacteria. After an incubation of the labelled LL-37 with P. aeruginosa elastase, no binding of LL-37 to the bacteria could be detected (Fig. 3D). Both incubations were analysed directly and, hence, identical amounts of Texas red label were added in both cases to the bacteria (this was easily verified by UV-illumination). The images in Fig. 3B and D were recorded using identical settings for the microscope and CCD camera software. To quantify the fluorescence more stringently, we analysed the fluorescence distribution of bacteria incubated with intact labelled LL-37 (Fig. 3E, rightmost histogram) and bacteria incubated with LL-37 that had been subjected to prior treatment with P. aeruginosa elastase (Fig. 3E, leftmost histogram). Non-degraded LL-37 gave rise to a distinct fluorescence, whereas the fluorescence histogram of bacteria incubated with degraded LL-37 did not differ from that of unlabelled bacteria (not shown).

Figure 3.

Binding of Texas red-labelled LL-37 to bacteria is abolished by Pseudomonas aeruginosa elastase. B shows the red fluorescence of bacteria (5 × 10 6  ml −1 ) stained with intact Texas red-LL-37 (10 µg ml −1 ) and D shows bacteria stained with Texas red-LL-37 subjected to prior treatment with P. aeruginosa elastase. Images in B and D were recorded using identical instrument settings. The corresponding Nomarski images are shown in A and C. E shows the staining of bacteria quantified by flow cytometry. The rightmost histogram shows bacteria labelled with intact Texas red-LL-37, whereas the leftmost histogram shows bacteria labelled with the degraded peptide. Results shown are representative of three separate experiments.

Finally, the effects of other bacterial proteinases on LL-37 were investigated. Incubation of LL-37 with E. faecalis growth medium (containing gelatinase), as well as purified S. pyogenes cysteine proteinase yielded similar fragments as those generated by P. aeruginosa elastase, as determined by SDS-PAGE. Proteus mirabilis proteinase did not generate these intermediates. However, for all bacteria, longer incubation periods yielded almost complete digestion of LL-37(Fig. 4A). Incubation of LL-37 with wound fluid from a patient with a P. aeruginosa infected chronic ulcer, also yielded almost complete degradation. Fluid from a surgical (sterile) wound did not exhibit this effect (Fig. 4B). Analysis of additional acute (three samples) and chronic, infected, wound fluids (two samples) from other patients yielded similar results (not shown).

Figure 4.

Bacterial proteinases and chronic wound fluid degrade LL-37. LL-37 alone is run in the far left lanes of panels A–C.

A. LL-37 (1 µg) was incubated with P. aeruginosa elastase (PAELA), E. faecalis growth medium containing gelatinase (EFGEL), P. mirabilis growth medium containing a 50 kDa metalloproteinase (PMMP), or S. pyogenes cysteine proteinase (SPCP) at the indicated periods of time. The proteinase activity added (as determined by an azocasein assay) corresponded to 6 mU (PAELA) and 3 mU (for the other proteinases) per 10 µl reaction.

B. LL-37 (1 µg) was incubated for 4 h at 37°C with chronic (CWF) or acute (AWF) wound fluid (7% in phosphate-buffered saline in a total volume of 15 µl) respectively.

C. LL-37 (1 µg) was incubated with PAELA (12 mU), or with 6 mU each of EFGEL, PMMP, SPCP for 1 h in the absence or presence of inhibitors. GM6001; 100 µM GM6001, PHE; 2 mM 1, 10-phenantroline, E64; 10 µM E64. The material was analysed by SDS-PAGE on 16.5% Tris-tricine gels. Molecular mass markers are indicated to the left.

Ex vivo experiments using human wound fluid and neutrophil extracts

The finding that wound fluid from infected ulcers degrades LL-37, prompted us to further investigate LL-37 degradation in wound fluids, and the effects on bacterial survival of P. aeruginosa proteinase production in these wound fluids. Various ulcer-derived P. aeruginosa, producing either undetectable levels of elastase or expressing high levels of elastase (three isolates each, of which one from each group is presented inFig. 5, see also Schmidtchen et al., 2001a) were grown in TH medium or in wound fluid obtained from sterile surgical wounds. Pseudomonas aeruginosa elastase was only detected (by zymographic analysis and immunoblotting) in supernatants from cultures of previously characterized elastase-expressing bacteria, and the levels of elastase were similar in TH and in wound fluid (Fig. 5A). The minor 50 kDa enzyme is alkaline proteinase (Schmidtchen et al., 2001a). Sterile wound fluid alone contained the human 92 and 72 kDa gelatinases (MMP-9 and MMP−2 respectively) (Fig. 5A). Having confirmed elastase production, the various wound fluids were then incubated with Texas red-labelled LL-37. As shown in Fig. 5B, significant degradation of LL-37 was only noted in the wound fluid containing elastase-producing P. aeruginosa (ELA +). Thus, the data show that wound fluid infected ex vivo with elastase-producing P. aeruginosa, has the capacity to rapidly degrade LL-37. On the basis of these observations, we examined whether the different P. aeruginosa isolates, when grown in wound fluid, exhibited variable sensitivity to exogenously added LL-37. In growth assays, overnight cultures of P. aeruginosa were inoculated (5%) in 200 µl of 10% human wound fluid (in phosphate-buffered saline). After an incubation period of 5 h, 0–30 µg of LL-37 was added and bacterial growth assessed (by measuring absorbance at 490 nm). Five and 10 µg LL-37 had no effect relative the controls (not shown). Whereas the elastase-producing P. aeruginosa was unaffected by LL-37 at all concentrations (in contrast to the results with the same isolate in the bactericidal assays shown in Fig. 1), the non-producing isolate was inhibited by LL-37 doses at or above 20 µg (100 µg ml−1) (Fig. 5C). This concentration of LL-37 exceeded the bactericidal concentration of the peptide (obtained in 10 mM Tris-buffer; 10–20 µg ml−1, see Fig. 1), indicating the presence of inhibitory substances in wound fluid, as previously described for plasma (Wang et al., 1998). These results were similar for all other isolates studied, showing that soluble elastase promotes bacterial survival in wound fluid ex vivo. The fact that neutrophils secrete hCAP18 which is processed into active LL-37, raised the question whether the intact proprotein is affected by P. aeruginosa proteinase. To address this, human neutrophil extracts were treated with conditioned medium from elastase-producing or elastase-deficient P. aeruginosa(Fig. 6). As shown by immunoblotting using polyclonal antibodies against LL-37, degradation of hCAP18 was only noted after treatment with elastase-producing P. aeruginosa. Likewise, addition of purified P. aeruginosa elastase to neutrophil extracts resulted in a complete degradation of hCAP18 and LL-37 (not shown). As assessed by zymography, the neutrophil extracts contained several gelatinases, similar to those detected in acute wound fluid, corresponding to neutrophil-derived MMP-2 and MMP-9 (not shown). Analogously to the results with LL-37 in wound fluid alone (Fig. 5A), very little degradation of hCAP18 was seen in the control incubations.

Figure 5.

P. aeruginosa elastase is secreted in human wound fluid, leading to degradation of LL-37 and enhanced bacterial survival.

A. Elastase-producing P. aeruginosa (ELA+) and an isolate devoid of elastase expression (ELA-) were grown to stationary phase in TH, or in wound fluid (WF) from surgical wounds. Then 0.5 µl of the material was analysed by zymography (10% polyacrylamide gels with 1 mg ml −1 gelatin). P. aeruginosa elastase (PAELA), which migrates as a ∼150 kDa multimer on zymograms ( Schmidtchen et al., 2001a ), is indicated to the right.

B. Texas red-labelled LL-37 (6 µg) was incubated with 5 µl of 50% sterile or infected wound fluids for the indicated periods of time. The material was analysed by SDS-PAGE on 16.5% Tris-tricine gels. WF, sterile wound fluid; WF + ELA+, wound fluid infected with elastase producing P. aeruginosa ; WF + ELA–, wound fluid infected with P. aeruginosa with no detectable elastase expression. In A and B, molecular mass markers are indicated to the left.

C. Elastase-producing and elastase-deficient P. aeruginosa isolates (ELA+ and ELA–, respectively) were inoculated in 200 µl of 10% wound fluid, 0, 5, 10, 20 or 30 µg of LL-37 was added after an incubation period of 5 h, and bacterial growth was measured (as absorbance at 490 nm) from this time point. The data represent the mean of double samples, and a representative experiment (of three) is shown. No difference was noted between 0, 5 and 10 µg of LL-37 and the results using 10 (empty bars), 20 (dashed bars), or 30 µg LL-37 (black bars) are shown.

Figure 6.

hCAP18/LL-37 from neutrophil extracts is degraded by elastase-producing P. aeruginosa . Neutrophil extracts (5 µl corresponding to ∼5 × 10 4 cells) were incubated with conditioned TH medium from elastase-producing or elastase-deficient P. aeruginosa isolates (+ and –, respectively, on top of panel). TH (C) was used as control. The material was run on 16.5% Tris-tricine gels after transfer of proteins to PVDF, immunoblotting was performed using polyclonal antibodies against LL-37. The positions of hCAP18 and LL-37 are indicated to the right and molecular mass markers (kDa) are indicated to the left.

Identification of substances that inhibit the degradation of LL-37

To find out whether the degradation of LL-37 could be inhibited, the effects of proteinase inhibitors were studied. Results showed that degradation was abolished by the metalloproteinase inhibitors GM6001 and 1, 10-phenantroline (which both inhibited P. aeruginosa elastase, E. faecalis gelatinase and P. mirabilis proteinase) (Fig. 4C, PAELA, EFGEL and PMMP), or by the inhibitor E64 which inhibits the S. pyogenes cysteine proteinase (Björck et al., 1989) (Fig. 4C, SPCP). Previous studies showed that the antibacterial peptide α-defensin specifically bound to glycosaminoglycans, preferably chondroitin sulphate B (CS-B), and that this binding destroyed the antibacterial effect of the peptide (Schmidtchen et al., 2001). Therefore, we investigated whether LL-37 could bind to various glycosaminoglycans. As assessed by slot-binding experiments, LL-37 preferably bound to CS-B and CS-E(Fig. 7A), and subsequently, in a functional assay, we found that the various glycosaminoglycans blocked the bactericidal effects of the peptide (Fig. 7B). An octasulphated disaccharide, sucrosoctasulphate (SOS), also blocked the bactericidal activity of LL-37. Having shown that LL-37 binds to sulphated polysaccharides, we wanted to investigate whether these polysaccharides, their constituting disaccharides or related carbohydrates, exerted protective effects on LL-37 vis-à-vis added bacterial proteinase. Indeed, intact polysaccharides, preferably CS-B and CS-E, were able to protect LL-37 from degradation by P. aeruginosa elastase (Fig. 7C). Interestingly, a protective effect was noted after addition of trisulphated CS-disaccharides of the structure ΔUA(2S)-GalNAc(4,6S) (Fig. 7D), which are found (as co-polymers) in the CS-B fraction used here (Rodén et al., 1973). Notably, a related disaccharide (ΔUA(3S)-GalNAc(4,6S)) was recently identified in CS-E (Kinoshita et al., 2001). Sucroseoctasulphate also exhibited similar protective effects at almost equimolar concentrations of disaccharide relative the peptide (Fig. 7E). Sucrose did not exhibit these effects (not shown). These results show that anionic molecules of various sulphation patterns inhibit the degradation of LL-37. No effect on elastase activity (as determined by an azocasein assay) by either the CS variants or sucroseoctasulphate was detected (not shown).

Figure 7.

Sulphated carbohydrates bind to and inhibit LL-37 and protect the peptide from P. aeruginosa elastase-mediated degradation.

A. LL-37 (0.2–5 mg) was applied onto nitrocellulose filters which were incubated with iodinated CS-B (∼10 mg ml −1 ). A buffer control (C) and different polysaccharides (2 mg ml −1 ) were added for competition of binding; CS-B, A, C, E (chondroitin sulphate B, A, C, and E) followed by autoradiography.

B. For analysis of possible inhibition of LL-37 activity, 2 × 10 6  CFU ml −1 E. faecalis were incubated with LL-37 (10 µg ml −1 ) for 2 h in the presence of the glycosaminoglycans CS-B (r), CS-A (▪), CS-C (d), CS-E (▪), or the disaccharide sucroseoctasulphate (▵) at the indicated concentrations. Colony-forming units were determined and representative experiments (of three) are shown.

C–E. LL-37 (1 µg) was incubated with P. aeruginosa elastase (12 mU) in a total of volume of 15 µl for 1 h without or together with inhibitor. In C, the inhibitors were: buffer; the glycosaminoglycans CS-B (chondroitin sulphate B); CS-A (chondroitin sulphate A); CS-C (chondroitin sulphate C); and CS-E (chondroitin sulphate E). In D, the inhibitors were: buffer; the disaccharides 0S; DUA-GalNAc; 2S-4S [DUA(2S)-GalNAc(4S)]; 2S-6S [DUA(2S)-GalNAc(6S)]; 4,6S [DUA-GalNAc(4,6S)]; and 2S-4,6S [DUA(2S)-GalNAc(4,6S)] (50 nmoles of disaccharide per reaction). In E, the inhibitors were buffer and sucroseoctasulphate (SOS) of the indicated amounts. The material was analysed by SDS-PAGE (16.5% Tris-tricine gels). Molecular mass markers (kDa) are indicated to the left, and LL-37 alone is run in the far left lanes.

Discussion

Bacterial evasion of peptide-based innate defences may prove to be important for the initial colonization of epithelial surfaces. Recently, pathways aimed at inactivating the innate immune system have been uncovered. For example, proteinases of common pathogenic bacteria, such as P. aeruginosa, E. faecalis and S. pyogenes were shown to release dermatan sulphate chains, leading to inactivation of the antibacterial peptide α-defensin (Schmidtchen et al., 2001b). Independent findings showed that the metalloproteinase lasA from P. aeruginosa induced release of another sulphated polysaccharide, heparan sulphate, leading to inactivation of innate defence and increased mortality during P. aeruginosa infection (Park et al., 2001). Staphylococcal resistance to defensins has recently been related to modification of bacterial membrane lipids with l-lysine (Peschel et al., 2001). During Shigella infections, it was shown that this pathogen downregulates the expression of LL-37 and human β-defensins, with bacterial DNA as a potential regulator (Islam et al., 2001). Additional acterial defence mechanisms directed against antibacterial peptides may involve modifications of outer cell wall components, such as lipopolysaccharide (Ernst et al., 1999) or phosphorylcholine (Lysenko et al., 2000) and the modulation of efflux pumps (Shafer et al., 1998).

Here, we show for the first time that proteinases of the clinically significant bacterial species P. aeruginosa, E. faecalis, P. mirabilis and S. pyogenes degrade the major human antibacterial peptide LL-37 and that this degradation leads to loss of LL-37 binding to bacteria and, ultimately, abolished bacterial killing. Pseudomonas aeruginosa elastase was shown to specifically hydrolyse LL-37. The P1′ position was preferentially occupied by hydrophobic amino acids (L, I, F), whereas the P1 position was less specific (N, D, R, E, K). The P1′ specificities corresponded well with those reported for the M4 peptidase family (thermolysin family) to which P. aeruginosa elastase belongs. Enterococcus faecalis gelatinase, as well as the 50 kDa metalloproteinase of P. mirabilis are also members of this group, having similar specificities (Rozalski et al., 1997; de Kreij et al., 2000). In this context, it should be noted that, in addition to cationic amino acids, most antibacterial peptides contain a significant (∼50%) portion of hydrophobic amino acids (Hancock and Scott, 2000). Therefore, considering the data discussed above, and taking the specificities of the M4 peptidase family into account, it is conceivable that bacterial proteinases may have profound effects on the actions of antibacterial peptides. This notion is further supported by the fact that human α-defensin was also degraded by P. aeruginosa elastase (unpublished results).

From the perspective of chronic ulcers and considering recent observations identifying defensins in wound fluids (Frohm et al., 1996), as well as findings showing that cutaneous injury induces the release of LL-37 in response to infection (Dorschner et al., 2001), it is still an enigma how certain pathogens, such as P. aeruginosa, E. faecalis or P. mirabilis may persist and colonize for long periods of time in these ulcers. In this study, we present several lines of evidence indicating a pivotal role for P. aeruginosa elastase in the evasion of LL-37 mediated bacterial killing. First, the finding that elastase production made P. aeruginosa completely resistant to the action of LL-37 during growth in human wound fluid ex vivo suggests that during P. aeruginosa infection of human wounds, release of elastase inactivates LL-37 and protects bacteria from the action of the antibacterial peptide. Second, P. aeruginosa elastase also induced degradation of the proprotein hCAP18 in neutrophil extracts. Third, we found that chronic wound fluid collected from P. aeruginosa infected leg ulcers, but not acute sterile wound fluid, had the capacity to degrade LL-37. Taken together, these findings correspond well with recent data showing that chronic venous ulcers are devoid of LL-37 and contain low levels of hCAP18, in contrast to normally healing wounds (Frohm-Nilsson, 2001). Because all chronic ulcers contain bacteria able to secrete proteinases, it is plausible that the observed lack of LL-37/hCAP18 is due to degradation by bacterial proteinases.

From a therapeutic point of view, inhibition of bacterially induced proteolysis could lead to increased levels of LL-37 in vivo. As hCAP-18 is processed to LL-37 by extracellular cleavage with proteinase 3, a serine proteinase (Sörensen et al., 2001), the use of specific metalloproteinase inhibitors, such as GM6001, could represent a possible treatment option in future in vivo studies. Considering the action of glycosaminoglycans and sulphated disaccharides, their effects also merits some consideration. The fact that these molecules block the bactericidal effects of LL-37, as well as protect the peptide from proteolysis, highlights the complexity and sometimes dual roles of bacterial virulence mechanisms. For example, it has been shown that bacterial proteinases are able to release CS-B, which may both inhibit (Schmidtchen et al., 2001b) and, as shown here, block antibacterial peptide degradation induced by the same proteinases. Other peptides, such as the antibacterial and highly cationic peptide PR-39, induce synthesis of the proteoglycan syndecan (Gallo et al., 1994), which, in turn, has the capacity to inactivate the peptide (Park et al., 2001). It is also worth considering that the functions of CS-B and the other glycosaminoglycans may go beyond mere inhibition of antibacterial activity. Recent data show that components of innate immunity, such as α−defensin, PR-39 and LL-37, act as potent chemoattractants (Huang et al., 1997; Yang et al., 2001). Hence, glycosaminoglycan-induced stabilization of these cationic peptides could be of biological significance in a highly proteolytic environment, perhaps underlying the herein noted remarkable stability of LL-37 in non-infected human wound fluids (which are rich in neutrophil elastase and various MMPs). In this context, it can be noted that growth factors, such as basic fibroblast growth factor, are bound to and protected from proteolysis by sulphated carbohydrates, including glycosaminoglycans and sucroseoctasulphate (Folkman et al., 1991; Klagsbrun, 1992), and that non-infected human wound fluid contains significant levels of CS-B (Penc et al., 1998).

In summary, we have presented data showing that the major human antibacterial peptide LL-37 is degraded and inactivated by proteinases of several significant human pathogens. This observation and the finding that proteinase inhibitors, glycosaminoglycans and sulphated disaccharides protect LL-37 from proteolytic degradation could be utilized in the search for novel antibacterial strategies.

Experimental procedures

Materials

LL-37, LLGDFFRKSKEKIGKEFKRIVQRIKDFLRNLVPRTES (MW 4492.08) and Texas-red labelled LL-37 were synthesized by Innovagen AB, Lund. The Texas-red label including an additional cysteine was coupled to the aminoterminus of LL-37. The purity and molecular weight of these peptides was confirmed by mass spectral analysis (MALDI.TOF Voyager). Both peptides exhibited identical bactericidal activities using E. faecalis as the test bacterium. 1, 10-Phenantroline, GM6001, E64, the chondroitin sulphate (CS) preparations CS-A, -C, and -E, and the CS and dermatan sulphate (DS) disaccharides DUA-GalNAc, DUA(2S)-GalNAc(4S), DUA(2S)-GalNAc(6S), DUA-GalNAc(4,6S), DUA(2S)-GalNAc(4,6S) were purchased from Sigma. CS-B (DS36) was obtained from Dr L.-Å. Fransson. The preparation and characteristics of this fraction have been described previously (Rodén et al., 1973; Fransson et al., 1979). Sucroseoctasulphate (potassium salt, MW 1287.55) was purchased from Toronto Research Chemicals. Pseudomonas aeruginosa elastase was purchased from Calbiochem. In some experiments, we used an equivalent elastase enzyme, generously supplied by Dr H. Maeda, Kumamoto University, Japan.

Bacterial culture and proteinases

Pseudomonas aeruginosa , E. faecalis, and P. mirabilis isolates were initially obtained from chronic venous ulcers and were grown for 18 h at 37°C (stationary phase) in Todd–Hewitt (TH) medium. The S. pyogenes strain AP1 (40/58) was from the World Health Organization Collaborating Centre for References and Research on Streptococci (Prague, Czech Republic), and was grown to stationary phase in C-medium ( Gerlach et al., 1983 ). Bacteria were pelleted by centrifugation and supernatants sterile-filtered (0.3 µm) and stored at −20°C. The cysteine proteinase of S. pyogenes was purified according to previous protocols ( Herwald et al., 1996 ). Growth culture supernatants from the E. faecalis (expressing gelatinase) and P. mirabilis strains (expressing a 50 kDa metalloproteinase) (not shown) were sterile-filtered and used directly. The activity of the proteinase preparations was determined using an azocasein assay ( Okamoto et al., 1997 ). For preparation of infected wound fluids, wound fluid from surgical wounds (1 ml diluted 1:1 with TH, denoted 50% WF in the text) was inoculated with 20 µl of overnight cultures of P. aeruginosa. Bacteria were pelleted by centrifugation and supernatants were stored at −20°C.

Wound fluids

Wound fluids (100–600 µl) from patients with chronic, infected, venous ulcers were collected under a Tegaderm dressing for 2 h as previously described (Schmidtchen, 2000). Sterile wound fluid was obtained from surgical drainages after mastectomy. Collection was for 24 h after operation. Wound fluids were centrifuged, aliquoted and stored at −20°C. The use of this material was approved by the Ethics Committee at Lund University (LU 509–01, LU 708–01). Informed consent was obtained from the patients.

Sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS-PAGE)

Electrophoresis was performed using the Ready Gel system (Bio-Rad) on 16.5% precast polyacrylamide gels with a 4% stacking gel (Tris-tricine Ready Gel, Bio-Rad). For analysis of LL-37 degradation, peptides solubilized in 10 mM Tris, pH 7.4, were incubated with sterile or infected wound fluids, purified proteinases and, when indicated, additional substances were added (see figure legends for details). The incubations were stopped by boiling (5 min) and samples were supplemented with Tricine sample buffer (Bio-Rad). After fixing, the gels were stained with 0.25% Coomassie brilliant blue, destained and dried between two cellophane sheets in a Gel air dryer (Bio-Rad), and scanned (Duoscan T1222; Agfa). Texas red-labelled LL-37 was visualized directly after electrophoresis by the use of UV-illumination and digital photography. Pictures were processed using the Adobe photoshop software (Adobe).

Zymography

Wound fluids and bacterial supernatants were mixed with sample buffer (0.4 M Tris HCl, 20% glycerol, 5% sodium dodecyl sulphate (SDS), 0.03% bromophenol blue, pH 6.8) and electrophoresed on 10% polyacrylamide gels (1 mg of bovine gelatine per ml of gel). To remove SDS, gels were incubated with 2.5% Triton X-100. Incubation was then performed for 18 h at 37°C in buffer containing 50 mM TrisHCl, 200 mM NaCl, 5 mM CaCl2, 1 µM ZnCl2, pH 7.5. Gels were stained with Coomassie blue G-250 in 30% methanol, 10% acetic acid for 1 h and destained in the same solution without the dye. Gelatinase-containing bands were visualized as clear zones against a dark background using using an Agfa Duoscan 1200 in the transillumination mode. Pictures were processed using the Adobe photoshop software (Adobe).

HCAP18/LL-37 detection in neutrophil extracts by immunoblotting

Neutrophils were prepared by routine procedures from blood of healthy human donors. The cells (1 × 107 cells ml−1) were disrupted by freeze-thawing and 5 µl of the super-natants were incubated with 5 µl of conditioned medium from elastase-producing P. aeruginosa, elastase-deficient P. aeruginosa or purified elastase (0.5 µg), TH was used as control. Incubation was performed at 37°C for various periods of time (see figure legend). The material was loaded onto 16.5% precast polyacrylamide gels and proteins/peptides were transferred to PVDF membranes (Immobilon, Millipore). Membranes were incubated for 1 h with rabbit polyclonal LL-37 antibodies (1:4500) (Innovagen AB, Lund), washed, and subsequently incubated (1 h) with HRP-conjugated secondary antibodies (1:1000). HCAP18/LL-37 was visualized using the ECL system (Boehringer).

Blotting assay for LL-37 binding

This was performed essentially as previously described (Schmidtchen et al. 2001b). LL-37 was applied onto nitrocellulose membranes (Hybond, Amersham). Membranes were blocked (PBS, pH 7.4, 0.25% Tween 20, 3% bovine serum albumin) for 1 h and incubated with radiolabelled DS (∼10 µg ml−1) for 1 h in the same buffer. Unlabelled polysaccharides (2 mg ml−1) were added for competition of binding. The membranes were washed (3 × 10 min) (PBS, pH 7.4, 0.25% Tween 20). Bas 2000 radioimaging system (Fuji) was used for visualization of radioactivity.

Mass spectrometry

Analysis of LL-37 fragments was performed by electrospray ionisation mass spectrometry on a Q-tof (Micromass, UK) using Masslynx software (performed by Innovagen AB, Lund).

Bactericidal assays

For antimicrobial assays S. pyogenes, E. faecalis, or P. aeruginosa were grown to mid-log phase in TH-medium. Bacteria were washed and diluted in 10 mM Tris-HCl, pH 7.5, containing 5 mM glucose. 50 µl of bacteria (2 × 106 colony-forming units ml−1) were incubated with intact or degraded LL-37, at the indicated concentrations (see figure legends). Incubations were carried out at 37°C for 2 h (S. pyogenes, E. faecalis) or 4 h (P. aeruginosa). To quantify the bactericidal activity, serial dilutions of the incubation mixture were plated on TH agar, incubated at 37°C overnight, and the number of colony-forming units were determined. In one experiment, the agar plates were placed on a light-board, photographed, and pictures further processed using Photoshop software (Adobe).

Bacterial growth assay

Pseudomonas aeruginosa were grown to stationary phase (OD ∼ 1) in TH-medium. Then 200 µl of 10% sterile wound fluid (in PBS) was inoculated with 10 µl of the bacterial suspension in 96-well plates (Falcon). After an incubation period of 5 h at 37°C, LL-37 was added and absorbance measured at 490 nm (using a Bio-Rad 550 microplate reader) after various periods of time. The absorbance value at T=5 h (addition of LL-37) was subtracted from the thereafter attained values (see figure legends for further details).

Fluorescence microscopy

Bacteria (5 × 106 ml−1) were incubated for 5 min on ice together with either intact Texas red-labelled LL-37 (10 µg ml−1) or with Texas red-labelled LL-37 that had been subjected to prior treatment with P. aeruginosa elastase (5 ng of elastase µg−1 LL37 for 4 h at 37°C). Complete degradation of the peptide was verified by electrophoresis of the material on 16.5% Tris-tricine gels followed by UV-transillumination. After washing twice with ice-cold PBS, the bacteria were mounted on slides for immediate visual inspection and recording of images. This was performed using a Nicon Eclipse TE300 inverted fluorescence microscope equipped with a Hamamatsu C4742-95 cooled CCD camera, a Plan Apochromat 60X objective, and a high N.A. oil-condenser.

FACS analysis

Flow cytometry measurements were performed using a FACSCalibur flow cytometer (Becton-Dickinson) equipped with a 15 mW argon laser tuned at 488 nm. The bacterial population was selected by gating with appropriate settings of forward scatter (FSC) and sideward scatter (SSC). The FL3 fluorescence channel (λem = 585 nm) was used to record the emitted fluorescence of Texas red. All experiments were performed at least three times using different preparations of bacteria and the results were analysed using the software CellQuest (Becton Dickinson).

Acknowledgements

This work was supported by grants from the Swedish Research Council (projects 13471, 7480, 14379 and 12613), the Royal Physiographic Society in Lund, the Welander-Finsen, Magnus Bergvall, Thelma-Zoegas, Crafoord, Alfred Österlund, Crafoord, and Kock Foundations, and HANSA MEDICAL AB. We thank Professor Lars-Åke Fransson and Professor Hiroshi Maeda for the generous gift of DS36 and P. aeruginosa elastase, respectively, and Victoria Rydengård and Mina Davoudi for expert technical assistance.

Ancillary