A Biofilm Matrix-Associated Protease Inhibitor Protects Pseudomonas aeruginosa from Proteolytic Attack

To identify proteins enriched in the biofilm matrix of P. aeruginosa, we modified a protocol used by Absalon and colleagues in Vibrio cholerae (11). Briefly, the extracellular proteins of mature P. aeruginosa PAO1 biofilms were biotinylated prior to biofilm disruption. The biotinylated proteins were then purified and identified using quantitative mass spectrometry. This “extracellular matrix proteome” was compared to the proteome of the total biofilm, in which the cells in the biofilm were lysed prior to the biotinylation of the proteins. In comparison to the extracellular matrix proteome, the “total biofilm proteome” contains both intracellular and extracellular proteins. Therefore, matrix proteins will be higher in abundance in the extracellular matrix proteome than in the total biofilm proteome. For a negative control, this analysis was also performed in the absence of the biotinylation agent for both the extracellular matrix and total biofilm proteomes.

This analysis was conducted twice, and a total of 857 unique proteins were identified in both runs (Fig. 1). The candidate matrix proteins were identified on the basis of the following criteria: (i) the candidate matrix proteins had greater than or equal to two-fold spectrum counts in the extracellular biofilm proteome relative to the total biofilm proteome analysis; (ii) the candidate proteins had at least 10 spectrum counts in the extracellular biofilm proteome; and (iii) the candidate proteins had greater than or equal to two-fold spectrum counts in the biotinylated extracellular biofilm proteome relative to the biofilm proteome without biotinylation. By comparing the proteomes using these high-stringency criteria, we identified 60 proteins that are enriched in the biofilm matrix in both analyses (Fig. 1; see Tables S1 and S2 in the supplemental material). On the basis of their predicted biochemical activities, we categorized 19 of the 60 proteins as potentially playing a role in the protection of the biofilm matrix (Fig. 1 and Table S1). Of these 19 proteins, 12 have predicted oxidoreductase activity based on the InterProScan software package (12), and another four proteins may play a role in redox processes. Of the remaining three proteins predicted to play a role in the protection of the biofilm matrix, two have functions in facilitating protein folding, and one is a serine protease inhibitor, which we chose to further investigate. It is important to note that our classification of these proteins as potentially playing a protective role in the biofilm matrix has not been verified, and it is possible that these proteins provide other functions in the matrix.

To validate our proteomic analysis, we focused on ecotin, a serine protease inhibitor (9) that is encoded by PA2755 (referred to herein as eco). Ecotin was enriched approximately four-fold in the matrix proteome (Table S1). Originally identified in Escherichia coli, ecotin inhibits enzymes, such as trypsin, chymotrypsin, and neutrophil elastase (9). While found in a wide variety of Gammaproteobacteria, most bacteria, including P. aeruginosa, that carry an eco gene do not produce proteases of the class that ecotin inhibits, which suggests that ecotin does not inhibit a bacterium’s own proteases (13, 14). Instead, it has been suggested that ecotin may protect bacteria from external proteolytic attack in metazoan hosts. Supporting this hypothesis, ecotin in E. coli can protect against neutrophil elastase-mediated cell death in vitro (15). Furthermore, ecotin and ecotin-like proteins have been shown to be important for the virulence of Burkholderia pseudomallei (13) and the infectivity of Leishmania major (16). We, therefore, hypothesized that ecotin in the biofilm matrix may protect P. aeruginosa biofilms against proteases such as neutrophil elastase, an enzyme the bacteria commonly come into contact with during chronic biofilm-based infections (10) via the following two mechanisms: (i) ecotin may protect proteins in the matrix from proteolytic degradation by neutrophil elastase, and (ii) ecotin may protect biofilm cells from neutrophil elastase-mediated death. Both possibilities would help preserve biofilm integrity and are not mutually exclusive.

Since P. aeruginosa has been shown to have strain-to-strain differences in its production of biofilm matrix components (17), we examined ecotin expression in a subset of the sequenced P. aeruginosa strains in the Pseudomonas Genome Database, all of which contain an eco gene (18) and are able to produce the P. aeruginosa exopolysaccharides Psl and Pel (17). Via immunoblotting, we found that ecotin is expressed in all nine of the tested clinical and environmental isolates. In stationary phase, the expression levels for these strains were at a level similar to that of our laboratory PAO1 reference strain (Fig. S1).

Next, we determined whether ecotin expression or its presence in the biofilm matrix changes during the course of biofilm growth. Using a culturing methodology similar to that used in the proteomic analysis, we used immunoblotting to test for the presence of ecotin in the total biofilm and the extracellular matrix of biofilms grown for 48, 96, and 144 h. While ecotin is present at approximately equal levels in the total harvested biofilm biomass at all three time points of development, it was found in the extracellular matrix at high levels only in the older 144-h biofilms (Fig. 2A). This increase in ecotin in the extracellular matrix was associated with a corresponding increase in the amounts of matrix-associated Psl exopolysaccharide (Fig. 2B).

Because the accumulation of ecotin in the matrix was temporally associated with the accumulation of Psl (Fig. 2), we speculated that ecotin interacts with Psl. To test this hypothesis, we used a coimmunoprecipitation method previously described to identify CdrA as a Psl-binding protein (19). Using anti-Psl antibody-coated beads, we precipitated Psl and any interacting proteins from cell-free supernatants of eco-overexpressing strains with wild-type or psl mutant backgrounds. Ecotin was then detected by immunoblotting. Ecotin coprecipitated with anti-Psl antibodies in the presence, but not in the absence, of Psl (Fig. 3). For a negative control, we examined whether another protein when overexpressed would coimmunoprecipitate with Psl. While Tse1, an effector of the type VI secretion system (20), was expressed, it did not precipitate with anti-Psl antibodies even in the presence of Psl (Fig. 3). In addition, using the proteomic analysis that identified ecotin as a matrix-interacting protein, we failed to identify ecotin as a protein enriched in the matrix of a biofilm lacking Psl. In the two separate runs, only nine and two spectral counts were obtained for ecotin in the biotinylated extracellular biofilm proteome of a Psl-negative biofilm. Since these counts are below our threshold of 10 counts, ecotin was eliminated from further analysis of matrix-enriched proteins for the Psl-negative biofilm. Together, these results confirm that ecotin binds to the biofilm matrix via the exopolysaccharide Psl and validates our screening methodology for matrix proteins.

Since a protein may lose its functionality outside the cell when interacting with the matrix, we sought to determine whether matrix-bound ecotin is functionally active in a cell-free system. To obtain matrix-bound ecotin, we immunoprecipitated Psl from cell-free supernatants of cells that were overproducing both ecotin and Psl. As a control, Psl was immunoprecipitated from cell-free supernatants of Δeco cells that were overproducing only Psl. The activity of neutrophil elastase was then tested in the presence of ecotin-bound or ecotin-free matrix material. With the ecotin-bound matrix, neutrophil elastase activity was inhibited 52.1% ± 10.2% (mean ± standard deviation [SD]; n = 5; P < 0.01) compared to that with the ecotin-free matrix (Fig. 4A). This partial inhibition is likely due to experimental technicalities: the immunoprecipitated ecotin-bound matrix contained only 16.5 ± 3.2 nM ecotin (n = 3) compared to the 100 nM neutrophil elastase added to the experiment. Furthermore, this inhibition is likely not due to the release of ecotin from the matrix, since approximately equal amounts of ecotin were bound to the matrix before and after the elastase activity assay by immunoblotting (Fig. 4B). Nonetheless, these results show that matrix-associated ecotin can inhibit neutrophil elastase, suggesting that ecotin in the biofilm matrix can protect other matrix proteins from proteolytic degradation.

Since ecotin has previously been shown to protect planktonic E. coli from neutrophil elastase-mediated killing (15), we looked at whether ecotin could protect P. aeruginosa planktonic and biofilm cells from neutrophil elastase. Using planktonic cells, we compared the effect of neutrophil elastase treatment on the viability of wild-type versus Δeco cells (Fig. 5). While wild-type cells decreased in viability when treated with neutrophil elastase (a decrease of 0.93 ± 0.14 log₁₀-transformed CFU/ml [mean ± SD]; n = 3), Δeco cells had significantly more cell death (a loss of 2.47 ± 0.25 log₁₀-transformed CFU/ml; n = 3; P < 0.01). This difference in viability is due to neutrophil elastase, since wild-type and Δeco cells grew to similar levels in the absence of neutrophil elastase (Fig. 5). Furthermore, this defect in Δeco cells could be rescued by complementing the strain with an ectopic copy of eco driven under its native promoter. When treated with neutrophil elastase, the complemented strain had a decrease in viability that was indistinguishable from that of wild-type cells (a decrease of 1.48 ± 0.47 log₁₀-transformed CFU/ml; n = 3; P = 0.10).

While our data (Fig. 4 and 5) support our hypothesis that ecotin in the biofilm matrix can protect the biofilm against neutrophil elastase, to more directly test this hypothesis, we treated static glass-grown biofilms with neutrophil elastase and determined the percentage of cells killed by the treatment based on viability (propidium iodide) staining. As expected, we saw that indeed Δeco biofilms were more susceptible to killing by neutrophil elastase than wild-type biofilms were (Fig. 6A). In the absence of neutrophil elastase treatment, Δeco biofilms had an equivalent number of dead cells as that in wild-type biofilms (P = 0.57; n = 3). In comparison, after neutrophil elastase treatment, Δeco biofilms had significantly more cell death than wild-type biofilms did (P < 0.01; n = 3). Furthermore, complementation of eco rescued the biofilm susceptibility defect (P = 0.70; n = 3). These results show that ecotin can protect the biofilm from neutrophil elastase-mediated killing. This experiment, however, does not separate the contribution of the cellular ecotin in the periplasm from that of the extracellular matrix-bound ecotin.

To isolate the contribution of the extracellular matrix-bound ecotin, we examined whether exogenous addition of ecotin can protect Δeco biofilms from neutrophil elastase-mediated cell death (Fig. 6B). We first purified recombinant ecotin and confirmed that this protein was capable of inhibiting neutrophil elastase in vitro (Fig. S2), similar to that previously seen by Eggers and colleagues (15). We then tested the ability of this protein to rescue neutrophil elastase-mediated killing of the Δeco biofilm by exposing the biofilms to this recombinant ecotin and then rinsing the biofilms to remove any ecotin that did not bind to the matrix before neutrophil elastase treatment. After neutrophil elastase treatment, Δeco biofilms exposed to recombinant ecotin had significantly fewer dead cells compared to Δeco biofilms not exposed to recombinant ecotin (P < 0.05; n = 3) and had an equivalent number of dead cells to that of Δeco biofilms that were not treated with neutrophil elastase (P = 0.90; n = 3). This rescue of neutrophil elastase-mediated cell death in Δeco biofilms via the addition of recombinant ecotin is dependent on Psl. Unlike Δeco biofilms, biofilms of a mutant strain that lack both ecotin and Psl production (Δeco ΔpslD mutant) that were exposed to recombinant ecotin had significantly more cell death after neutrophil elastase treatment compared to biofilms not exposed to neutrophil elastase (P < 0.05; n = 3). Therefore, Psl is required for the supplemented ecotin to protect the biofilm against neutrophil elastase-mediated cell death. This requirement for Psl is likely only for the extracellular portion, and not the periplasmic cellular portion, of ecotin in wild-type P. aeruginosa biofilms, since endogenous cellular ecotin in other bacteria protects against neutrophil elastase independent of Psl. These results further support our conclusion that ecotin is a Psl-interacting protein in the biofilm matrix and that matrix-associated ecotin is functional. Together, our results show that ecotin in the matrix can protect not only other matrix proteins but also biofilm cells from proteolytic attack.

The bacterial strains, plasmids, and primers used in this study are listed in Tables S3 to S5, respectively. Construction of the strains used in this study are described in Text S1. Unless otherwise noted, bacteria were grown at 37°C in lysogeny broth (LB) or in Vogel-Bonner minimal medium (43).

Primary antibodies were diluted in 5% milk in phosphate-buffered saline (PBS) with 0.05% Tween 20: anti-ecotin sera (raised against AKLDEKVPYPKADC; Covance) diluted 1/400, anti-Psl antibodies (MedImmune) (44) diluted 1/3,000, and anti-Tse1 antibodies (20) diluted 1/2,000. Psl immunoblotting was performed as previously described (45). For SDS-PAGE, proteins were separated on an 18% Tris HCl gel and transferred to nitrocellulose for immunoblotting.

Tube biofilms were grown as previously described (46) in amine-free medium [60 mM trisodium citrate, 15 mM K₂HPO₄, 50 mM (NH₄)₂SO₄, 1.33 mM MgSO₄, 140 µM CaCl₂, 8.5 µM ZnSO₄, 3.9 µM FeSO₄, pH 7.5] for 6 days at 25°C. For the total biofilm proteome analysis, the cells in the biofilm biomass were first lysed before the proteins in the sample were biotinylated. For the extracellular biofilm proteome, the extracellular proteins in the biofilm were biotinylated before the biomass was collected and the cells were removed from the matrix material. The biotinylated proteins were purified and digested with trypsin. The resulting peptides were then identified by liquid chromatography coupled to tandem mass spectrometry (LC-MS/MS).

Psl coimmunoprecipitations were performed as previously described with the following modifications (19). Briefly, anti-Psl antibodies (MedImmune)-conjugated magnetic protein A Dynabeads (Life Technologies) were incubated with cell-free supernatants from stationary-phase cultures. Proteins coprecipitating with Psl were eluted with Laemmli buffer and analyzed by immunoblotting.

For planktonic expression experiments, cells were resuspended in Laemmli buffer at 10¹⁰ CFU/ml and analyzed by immunoblotting. To determine when ecotin is associated with the biofilm matrix, tube biofilms were grown as described above to the desired time point. Half of the sample was lyophilized to obtain the dry weight. In the other half of the sample, the cells were removed from the matrix. The proteins in the supernatant were normalized based on their dry weight measurements and analyzed by SDS-PAGE.

Psl was precipitated on the magnetic beads as described above, and neutrophil elastase (Millipore) at 100 nM and the fluorogenic substrate (MeOSucAAPV-AMC; Millipore) at 0.25 mM were then added. Cleavage of the substrate was measured at an excitation wavelength of 380 nm and an emission wavelength of 460 nm (Ex380/Em460). After the experiment, the beads were recovered, and the proteins were analyzed by SDS-PAGE as described above.

For planktonic experiments, cells at an OD₆₀₀ of ~2.0 were treated with neutrophil elastase (Oxford Biomedical Research) at 250 µg/ml at 37°C for 2 h before viable-cell counts were determined. For biofilm experiments, biofilms grown for 20 h on glass slides were treated with 100 µM ecotin and 2 μM neutrophil elastase (Millipore) for 30 min. The biomass was stained with 5 μM Syto9 (Molecular Probes) and 30 μM propidium iodide (Sigma-Aldrich) for 15 min before imaging with a Zeiss LSM 510 confocal laser scanning microscope (Carl Zeiss).