Burkholderia cenocepacia zinc metalloproteases influence resistance to antimicrobial peptides

The Burkholderia cepacia complex (Bcc) consists of at least 17 different species based on 16S RNA sequences, DNA–DNA homology, lipid and fatty acid analysis and phenotypic characteristics (Coenye et al., 2001, 2003; Vandamme et al., 1997; Vanlaere et al., 2008, 2009). The Bcc are important pathogens in people with cystic fibrosis (CF) as well as emerging opportunistic pathogens in non-CF patients (Govan et al., 2007; Mahenthiralingam et al., 2005, 2008). Some Bcc strains are transmitted between patients and have been referred to as epidemic strains (Chen et al., 2001; Coenye & LiPuma, 2003; Govan et al., 1993; LiPuma et al., 1990; Speert et al., 2002). Although strict infection control measures have reduced Bcc infections to only 4–7 % of CF patients (Govan et al., 2007; Mahenthiralingam et al., 2005), these infections can result in necrotizing pneumonia and sepsis that can lead to the rapid deterioration of the patient in what has been termed cepacia syndrome (Isles et al., 1984). In patients without cepacia syndrome, Bcc infections are also associated with increased morbidity and mortality (Corey & Farewell, 1996; Murray et al., 2008). Bcc strains are emerging as nosocomial pathogens in patients without CF, and outbreaks resulting in bacteraemia have been reported (Bressler et al., 2007). Bcc infections are difficult to treat due to these species' inherent resistance to antibiotics (Mahenthiralingam et al., 2005; Nair et al., 2004, 2005).

B. cenocepacia and Burkholderia multivorans are the most frequently isolated Bcc species from CF patients (Govan et al., 2007; LiPuma, 2005; Reik et al., 2005). B. cenocepacia accounts for most transmissible strains, although transmissible strains of B. multivorans and Burkholderia dolosa have also been reported in some CF centres (Biddick et al., 2003; LiPuma et al., 2001; McDowell et al., 2004; Speert et al., 2002). CF patients infected with B. cenocepacia have a much lower success rate after lung transplantation compared to CF patients infected with other Bcc species or not infected with Bcc (Alexander et al., 2008; Murray et al., 2008).

Several potential virulence factors have been characterized and implicated in the pathogenesis of Bcc lung infections (Mahenthiralingam et al., 2005; Vial et al., 2007). B. cenocepacia and at least four other Bcc species contain two distinct zinc metalloproteases, designated ZmpA and ZmpB. The zmpA and zmpB genes are always present in the same Bcc species but are not genetically linked. Species that lack both these genes have no detectable protease activity on skim milk agar, suggesting that these are the only extracellular proteases with casein activity (Corbett et al., 2003; Gingues et al., 2005; Kooi et al., 2006). Both ZmpA and ZmpB have pre-proenzyme structures and are autoproteolytically converted into a mature protease and a propeptide (Kooi et al., 2005, 2006). Recombinant forms of both proteases have been expressed and purified from Escherichia coli (Kooi et al., 2005, 2006). Although ZmpA and ZmpB are functionally similar, they share little sequence similarity with the exception of the conserved active site and zinc ligand motifs characteristic of zinc metalloproteases.

ZmpA and ZmpB have broad-spectrum activity and degrade substrates involved in tissue integrity or host defence, including type IV collagen, fibronectin, neutrophil α_1-proteinase inhibitor and α₂-macroglobulin. ZmpA cleaves gamma interferon and ZmpB degrades transferrin, lactoferrin and human immunoglobulins IgA, IgG and IgM. The specificity of each enzyme is clearly different in that ZmpA generally cleaves its substrates into two to three peptides whereas ZmpB cleaves substrates into a large number of small peptides (Kooi et al., 2005, 2006). We have shown that both proteases contribute to virulence in a chronic respiratory infection model. A mutation in either gene in B. cenocepacia strain K56-2 results in decreased lung pathology compared to the parent strain; however, a mutation in zmpA also results in reduced ability to survive in vivo since zmpA mutants are cleared from the lungs whereas the parent strain maintains a persistent chronic infection (Corbett et al., 2003; Kooi et al., 2006).

Burkholderia spp. are highly resistant to antimicrobial cationic peptides that typically kill bacteria by membrane disruption. LPS structure has been shown to play an important role in resistance to polymyxins and other peptides in both B. cenocepacia (Loutet et al., 2006) and Burkholderia pseudomallei (Burtnick & Woods, 1999). Other mechanisms of bacterial resistance to antimicrobial peptides have been described, including alteration of surface charges, changes in membrane proteins, efflux pumps/transporters and proteolytic enzymes (Brogden, 2005; Hiemstra et al., 2004; Potempa et al., 2009; Yount & Yeaman, 2005). It is likely that Burkholderia spp. employ multiple mechanisms to achieve their high degree of resistance to antimicrobial peptides. Since ZmpA and ZmpB are both broad-spectrum proteases, it is possible that one or both may be able to inactivate antimicrobial peptides by proteolytic cleavage. In this study, we determined the ability of purified ZmpA and ZmpB to degrade representative α- and β-defensins, LL-37, elafin and secretory leukocyte protease inhibitor (SLPI). We also compared the resistance of B. cenocepacia wild-type and a zmpA zmpB mutant to antimicrobial peptides to determine if the absence of ZmpA and ZmpB renders B. cenocepacia more sensitive to microbial killing.

Strains.
B. cenocepacia K56-2 is a CF respiratory isolate (Mahenthiralingam et al., 2000). K56-2 zmpA zmpB has been previously described (Kooi et al., 2006). This mutant has a trimethoprim resistance cassette inserted into the BsiWI site of zmpA and a tetracycline resistance cassette inserted into the remaining SalI site in zmpB after deletion of the internal 1.4 kb SalI fragment. K56-2 zmpA zmpB was grown in medium supplemented with 200 µg tetracycline (Tc) ml^–1 and 100 µg trimethoprim (Tp) ml^–1.

Substrates.
Human neutrophil peptide-1 (HNP-1), human beta defensin-1 (hβD-1) and LL-37 were obtained from AnaSpec. Recombinant hβD-1 was obtained from Cell Sciences. Recombinant human SLPI was obtained from R&D Systems. Protamine sulfate salt from salmon, cathepsin G from human sputum, elastase from human leukocytes (HLE), lysozyme from human neutrophils, recombinant human elafin and succinyl-Ala-Ala-Ala-p-nitroanilide were purchased from Sigma.

Protease digestions.
Recombinant ZmpA and ZmpB were purified as previously described (Kooi et al., 2005, 2006). Using hide powder azure as a substrate (Rinderknecht et al., 1968), one unit of activity was defined as the amount of protease that produced a change of 0.1 A₅₉₅ per hour at 37 °C. Protease digestions of antimicrobial peptides were performed essentially as previously described (Kooi et al., 2005, 2006). One microgram of peptide was incubated at 37 °C with 2 U ZmpB in 25 mM MES pH 5.6, 1 mM CaCl₂ or ZmpA in 10 mM Tris pH 7.2 in a final volume of 50 µl for 16 h unless otherwise indicated. Recombinant ZmpB, recombinant ZmpA and substrate-only controls were included. Products of digested peptides were separated by Tricine/SDS-16 % PAGE (Schagger & von Jagow, 1987) and were visualized by Coomassie blue R staining.

Antimicrobial peptide killing assays.
Liquid protamine killing assays were performed essentially as described previously (Banemann et al., 1998), with minor modifications. Briefly, overnight cultures were grown for 16 h at 37 °C in L broth, subcultured at 100-fold dilution in L broth without antibiotics and grown to mid-exponential phase. Dilutions of the cultures were made in L broth or PBS. Serial twofold dilutions of the antimicrobial peptide were performed in water (pH 5.0). Fifty microlitres of each peptide dilution was mixed with an equal volume of bacterial suspension (10⁵ c.f.u. ml^–1). A PBS/bacteria mix was used as a reference control to determine a 100 % survival rate. After 1 h incubation at 37 °C, serial dilutions of the bacteria/peptide mixture were plated on LB agar plates. Results were recorded as percentage survival. Elafin (Simpson et al., 1999) and hβD-1 (Morrison et al., 1998) killing assays were performed as previously described except the hβD-1 bacteria/peptide mixtures were incubated for 2.5 h.

Inactivation of antimicrobial peptide activity.
Protamine (125 µg ml^–1) was treated with ZmpA (2 U) in 10 mM Tris pH 7.2 or ZmpB (2 U) in 25 mM MES pH 5.6, 1 mM CaCl₂ in a final volume of 50 µl for 16 h at 37 °C. Protamine (125 µg ml^–1)-only (no protease) and buffer-only controls were included. After incubation, the samples were serially diluted twofold (125 to 0.11 µg ml^–1) and 10⁵ c.f.u. E. coli DH5α were added, followed by incubation for 1 h at 37 °C with shaking. Samples were serially diluted and plated on L-agar plates to obtain surviving bacterial numbers.

Inactivation of elafin anti-protease activity.
Recombinant ZmpA or ZmpB (4 U) in a final volume of 100 µl or buffer controls were incubated with elafin (5 µg) for 16 h at 37 °C. Half of the samples were removed to confirm digestion by Tricine/SDS-PAGE and the other half were retained for elastase inhibition assays based on Cooley et al. (2001) with some modifications. Briefly, serial dilutions of treated elafin were incubated with porcine pancreatic elastase (1.25 µM) at final elafin : elastase molar ratios of 1 : 4 to 1 : 128 in PBS, 0.05 % Tween-20. A porcine pancreatic elastase only (no elafin, no protease) reference control was set up in parallel. After 30 min incubation at 37 °C, 77 µl of 20 mM Tris/HCl, pH 7.4, 0.5 M NaCl, 0.1 % PEG and 3 µl 50 mM substrate succinyl-Ala-Ala-Ala-p-nitroanilide was added. Changes in A₄₁₅ were monitored. The A₄₁₅ of the elastase-only control was set at 100 % and the % elastase activity of the samples were determined as follows: (A₄₁₅ of the sample/A₄₁₅ of the elastase-only control)x100.

Statistical analysis.
Statistical analysis was performed using one-way analysis of variance (ANOVA) with Dunnett's multiple comparisons test or Student's t-test. A P-value less than 0.05 was considered significant.

Ability of ZmpA and ZmpB to use antimicrobial peptides as substrates
Defensins are important components of lung innate immunity. HNP-1 is expressed predominantly by neutrophils and hβD-1 is constitutively expressed by mucosal epithelial cells (Brogden, 2005; Brown & Hancock, 2006; Mookherjee & Hancock, 2007). LL-37 is a cathelicidin expressed by both neutrophils and epithelial cells. To determine if ZmpA and ZmpB were able to use these antimicrobial peptides as substrates, recombinant forms of ZmpA and ZmpB were incubated with the peptides, after which the incubation mixtures were separated by Tricine/SDS-PAGE. Neither ZmpB nor ZmpA cleaved HNP-1 (Fig. 1a). Interestingly, ZmpB but not ZmpA degraded hβD-1 (Fig. 1b), and ZmpA but not ZmpB degraded LL-37 (Fig. 1c). Due to the small size of hβD-1 and LL-37 it was not possible to determine by Tricine/SDS-PAGE if these peptides were cleaved more than once by the proteases. Elafin and SLPI are peptides with antimicrobial activity, with the ability to eliminate pulmonary pathogens, as well as inhibitors of neutrophil elastase (Doumas et al., 2005; Hiemstra et al., 2004; Sallenave, 2000). Both ZmpB and ZmpA digested elafin and SLPI (Fig. 1d, e). As with hβD-1 and LL-37, the elafin cleavage products were not visible on the gel. The SLPI cleavage products were, however, clearly resolved by Tricine/SDS-PAGE. The masses of the peptides following incubation with ZmpA or ZmpB were different (Fig. 1e). Similar results have been observed with other host protein substrates, confirming that these proteases have different specificities (Kooi et al., 2005, 2006). Protamine is a fish antimicrobial peptide that is similar in size and charge to the defensins. Salmon protamine was digested by both ZmpB and ZmpA into different-size fragments (Fig. 1f). Intact protamine was not visible on the gel, possibly due to aggregation, which may have prevented migration into the gel. Lysozyme is a naturally occurring antibacterial protein that is found in saliva, mucus, nasal secretions, serum and the lysosomes of neutrophils and macrophages. Neither ZmpB nor ZmpA was active against human neutrophil lysozyme under the conditions employed (Fig. 1g). The protease digestion assays were performed for 16 h for maximum sensitivity; however, digestion of LL-37 by ZmpB and ZmpA, respectively, was determined to occur in less than 1 h (data not shown).

(22K): Fig. 1. Ability of B. cenocepacia ZmpA and ZmpB to digest antimicrobial peptides. One microgram of peptide was incubated with 2 U ZmpA or ZmpB in a final volume of 50 µl and products were separated by Tricine/SDS-16 % PAGE. (a) HNP-1 (3.4 kDa), (b) hβD-1 (3.9 kDa), (c) LL-37 (4.5 kDa), (d) elafin (6.0 kDa), (e) SLPI (11.7 kDa), (f) protamine (major component is 5.1 kDa) and (g) lysozyme (14.7 kDa). Lanes: 1, ZmpB only; 2, ZmpB and antimicrobial peptide; 3, antimicrobial peptide only; 4, ZmpA and antimicrobial peptide; 5, ZmpA only. Gels were scanned with a Umax scanner.

Sensitivity of K56-2 and K56-2 zmpA zmpB to antimicrobial peptides
Since at least one of the B. cenocepacia zinc metalloproteases could digest hβD-1, LL-37, SLPI, elafin and protamine, we next determined if there was a difference in susceptibility to killing by these antimicrobial peptides between a protease-negative mutant (K56-2 zmpA zmpB) and the wild-type strain (K56-2). The zmpA zmpB double mutant was used in most of these assays since one protease with activity against the antimicrobial peptide could render it inactive. K56-2 and K56-2 zmpA zmpB were equally resistant to hβD-1 (50 µg ml^–1) (data not shown). Elafin and SLPI at concentrations up to 2 µM had no effect on the viability of either strain (data not shown). Although K56-2 was consistently more resistant to LL-37 than K56-2 zmpA zmpB, no significant difference was found (data not shown). A difference was observed in protamine susceptibility, however, in that K56-2 was significantly more resistant to protamine than was K56-2 zmpA zmpB (Fig. 2) (ANOVA, P<0.05). Similar results were observed with a K56-2 zmpB mutant that still produces active ZmpA.

(38K): Fig. 2. Protamine liquid killing assay. K56-2, K56-2 zmpA, K56-2 zmpB, K56-2 zmpA zmpB and E. coli DH5α were incubated with various concentrations of protamine and the percentage of surviving cells was determined; 100 % survival corresponds to 0 mg protamine ml–1. K56-2 killing by protamine was significantly different from K56-2 zmpB and K56-2 zmpA zmpB (P<0.01, ANOVA, Dunnett's multiple comparisons test). The results represent the mean±SEM of four independent assays, each performed in duplicate.

Since previous studies have demonstrated that Burkholderia species are very resistant to β-defensins and other antimicrobial peptides (Baird et al., 1999; Burtnick & Woods, 1999; Loutet et al., 2006; Sahly et al., 2003), it is possible that mutations in zmpA and/or zmpB are insufficient to result in increased susceptibility to the antimicrobial activity of these defensins, due to myriad resistance mechanisms. It is also possible that the cleavage products generated by ZmpA or ZmpB retain antimicrobial activity. To test this hypothesis, we attempted to perform killing assays with peptides following incubation with either ZmpA or ZmpB using E. coli DH5α to test for killing. Treatment of protamine with either ZmpA or ZmpB decreased its ability to kill E. coli (Fig. 3). There was a significant difference between the antimicrobial activity of protamine and that of protamine treated with ZmpB, at concentrations of protamine ranging from 0.9 to 3.6 µg ml^–1, and protamine treated with ZmpA at 0.9 µg protamine ml^–1 (Fig. 3). No differences in killing were observed at concentrations above 3.6 µg ml^–1 or below 0.9 µg ml^–1 (data not shown).

(23K): Fig. 3. Ability of ZmpA and ZmpB treatment to decrease protamine antimicrobial activity. Protamine was incubated with 2 U ZmpA or ZmpB, or buffer controls. The residual antimicrobial activity against E. coli DH5α was determined using a liquid killing assay. Survival of the control without protamine was set at 100 %. ZmpB-treated protamine and buffer-treated protamine were significantly different at 0.9, 1.8 and 3.6 µg ml–1, and significantly different between ZmpA-treated protamine and buffer-treated protamine at 0.9 µg ml–1 (ANOVA, P<0.05). Values represent the mean±SD of triplicate assays.

Similar experiments were performed with hβD-1 (up to 50 µg ml^–1), elafin (up to 15 µg ml^–1) and SLPI (up to 30 µg ml^–1), but the commercial preparations used of these peptides had no antimicrobial activity against E. coli (data not shown). Elafin and hβD-1 have previously been shown to be active against Pseudomonas aeruginosa (Morrison et al., 1998; Simpson et al., 1999). Inactivation of the antimicrobial activity of either elafin or hβD-1 could be relevant in B. cenocepacia and Ps. aeruginosa co-infections. Therefore, we tested the antimicrobial activity of elafin digested with ZmpA or ZmpB, and hβD-1 digested with ZmpB, against Ps. aeruginosa strain PAO1 in comparison to the intact peptides. Although recombinant elafin was employed in this assay, we did not observe killing of PAO1 (data not shown). It has previously been demonstrated that the C-terminal domain of elafin contains the antimicrobial activity (Simpson et al., 1999). Since elafin is degraded into very small peptides by ZmpA and ZmpB, it is unlikely that these peptides would be of sufficient size to retain activity even if a preparation of elafin with antimicrobial activity was used in the assay. We were able to demonstrate killing of PAO1 by hβD-1; however, no difference in killing was observed between intact peptide and that previously incubated with ZmpB at the 50 µg ml^–1 concentration tested (data not shown). These data suggest that the proteolytic fragments generated by ZmpB do retain some antimicrobial activity.

ZmpA and ZmpB inactivate the anti-elastase activity of elafin
In addition to their antimicrobial properties, elafin and SLPI are inhibitors of host proteases, particularly human neutrophil elastase. To determine if cleavage by ZmpA or ZmpB reduced the ability of elafin to inhibit elastase, twofold serial dilutions of elafin, pre-incubated with ZmpA or ZmpB, were incubated with porcine pancreatic elastase (1.25 µM) at final elafin : elastase molar ratios of 1 : 4 to 1 : 128. Both ZmpA and ZmpB cleavage of elafin significantly decreased its anti-elastase activity (Table 1). Neither ZmpA nor ZmpB had any activity against the elastase substrate, succinyl-Ala-Ala-Ala-p-nitroanilide. Complete elafin cleavage in these assays was confirmed by Tricine/SDS-PAGE (data not shown). At higher elafin concentrations (1 : 4 elafin : elastase molar ratio) elafin inactivation of elastase was observed likely due to the presence of residual uncleaved elafin. The commercial SLPI preparation was found to be inactive since a SLPI : elastase molar ratio of 1 : 1 resulted in no significant decrease in elastase activity (data not shown).

Table 1. ZmpA and ZmpB inactivation of elafin elastase inhibition activity

Since inactivation of elafin could alter the protease/antiprotease balance in the lung, the ability of ZmpA and ZmpB to cleave neutrophil proteases was also examined. Cathepsin G is a chymotrypsin-like protease that is a major constituent of human neutrophil granulocytes. Both neutrophil elastase and cathepsin G have been found to be essential for neutrophil killing of bacteria (Reeves et al., 2002). Both cathepsin G and neutrophil elastase are present in the CF lung (Rees & Brain, 1995) and can cause tissue destruction. Under the digestion conditions employed, neither ZmpB nor ZmpA degraded cathepsin G or neutrophil elastase (data not shown). Bacterial proteases are an important mechanism of resistance to host defences and are increasingly being recognized as factors that contribute to antimicrobial peptide resistance. Many proteases have been reported to degrade LL-37, including Proteus mirabilis ZapA (Belas et al., 2004), Staphlylococcus aureus aureolysin (Sieprawska-Lupa et al., 2004), Ps. aeruginosa LasB, Enterococcus faecalis gelatinase, Streptococcus pyogenes SpeB (Schmidtchen et al., 2002), and a Bacillus anthracis metalloprotease (Thwaite et al., 2006). E. coli OmpT and OmpP proteases degrade protamine (Hwang et al., 2007; Stumpe et al., 1998). Fewer studies have examined the activity of bacterial proteases against other antimicrobial peptides but Pr. mirabilis ZapA has been shown to degrade hβD-1 (Belas et al., 2004) and LasB has been shown to cleave SLPI (Sponer et al., 1991). Our studies have shown that at least one of the B. cenocepacia ZmpA and ZmpB proteases had activity against hβD-1, LL-37, elafin and SLPI. Inactivation of these peptides by ZmpA or ZmpB could contribute to the inherent resistance of B. cenocepacia.

Proteases have been reported to enhance resistance of other bacterial pathogens to antimicrobial peptides. S. aureus strains that produce aureolysin are more resistant to LL-37 than strains that express no aureolysin activity (Sieprawska-Lupa et al., 2004). Clostridium botulinum protease-positive strains are resistant to protamine whereas protease-negative strains are susceptible (Hansen et al., 2001). We did demonstrate increased protamine susceptibility of a zmpA zmpB mutant compared to the B. cenocepacia wild-type strain, suggesting that these proteases could contribute to resistance. The level of susceptibility of the zmpB mutant and the zmpA zmpB mutant was similar to that of the E. coli control, whereas the zmpA mutant was intermediate between the double mutant and the wild-type strain. We also demonstrated that protamine treated with either ZmpB or ZmpA had decreased antimicrobial activity against E. coli.

Unfortunately, commercial preparations of hβD-1, elafin and SLPI did not have antimicrobial activity against B. cenocepacia; therefore, we were not able to determine if the zmpA zmpB mutant was more susceptible than the wild-type to these peptides or if cleavage of these peptides with ZmpA or ZmpB influenced their antimicrobial activity. Some increased susceptibility of the zmpA zmpB mutant to LL-37 was consistently observed; however, the differences were not statistically significant. It may be difficult to demonstrate large differences in susceptibility in the assay employed since dilution of mid-exponential-phase cultures for the killing assay would result in very low amounts of protease being present.

Although proteolytic enzymes may contribute to the resistance of B. cenocepacia to antimicrobial peptides, it is recognized that other factors, particularly LPS, likely play a greater role. The presence of 4-amino-4-deoxyarabinose (Ara-4N) moieties attached to phosphate residues in the lipid A backbone is believed to be involved in the resistance of B. cepacia to antimicrobial peptides (Cox & Wilkinson, 1991; Shimomura et al., 2003). A complete lipid A-core oligosaccharide was recently shown to be required for resistance to several unrelated antimicrobial peptides and for survival of B. cenocepacia K56-2 in a rat chronic infection model (Loutet et al., 2006).

Previously we have demonstrated that ZmpA and ZmpB cleave neutrophil α₁-proteinase inhibitor and α₂-macroglobulin (Kooi et al., 2005, 2006). In this study, we demonstrate that these proteases also have activity against elafin and SLPI. We determined that ZmpA or ZmpB cleavage of elafin reduced its anti-elastase inhibitory activity. A great deal of CF lung damage is thought to be due to increased neutropil elastase levels. Inactivation of elastase inhibitors such as α₁-proteinase inhibitor, SLPI and elafin by ZmpA and ZmpB could disrupt the balance of neutrophil elastase and its inhibitors, contributing to the elevated neutrophil elastase levels in the lung. Since ZmpA and ZmpB did not have any direct activity against neutrophil elastase or cathepsin G, ZmpA or ZmpB produced by B. cenocepacia in the lung would primarily influence the levels of the inhibitors of the neutrophil proteases. Neutrophil elastase and cathepsin G expressed by neutrophils and monocytes are implicated in antimicrobial defence by degrading engulfed micro-organisms inside the neutrophil phagolysosomes. Mice lacking neutrophil elastase (Belaaouaj et al., 1998, 2000) and cathepsin G (Reeves et al., 2002) are deficient in bacterial killing. Elevated levels of neutrophil elastase that occur in CF inflammatory lung disease, however, also have the ability to digest structural elastin, collagen and fibronectin, resulting in extensive tissue damage (Shapiro, 2002). This protease can also degrade the chemokine receptor CXCR1 [interleukin-8RA (IL-8RA)] on neutrophils, which in turn can disable the bacterial-killing capacity of these cells (Hartl et al., 2007).

Recently, B. cenocepacia zmpA expression was shown to be upregulated sixfold by growth in medium containing CF sputum, suggesting that ZmpA would likely be present in lungs of CF patients infected with B. cenocepacia (Drevinek et al., 2008). Although zmpB expression was not reported to be elevated in this study, expression analysis was performed on mid-exponential-phase cultures and zmpB is poorly expressed during this phase of growth. Previously, we have reported that both ZmpA and ZmpB contribute to the lung pathology observed in a chronic respiratory infection model. It is likely that ZmpA and ZmpB enhance the virulence of B. cenocepacia by a variety of mechanisms including degradation of antimicrobial peptides, resulting in decreased susceptibility to their killing activity, degradation of neutrophil protease inhibitors, resulting in increased inflammation, as well as direct destruction of tissue and other host defence proteins such as immunoglobulins, transferrin and lactoferrin.

This study was supported by operating grant MOP-42510 from the Canadian Institutes of Health Research to P. A. S. D. F. Viteri is acknowledged for excellent technical assistance.

Edited by: P. Cornelis