|Home | About | Journals | Submit | Contact Us | Français|
To analyze PASP in terms of its gene distribution and expression, its corneal pathologic effects, its enzymatic properties, and the protectiveness of the immune response to this protease.
Twenty-five strains of P. aeruginosa were analyzed for the PASP gene and secreted protein by PCR and Western blot analysis, respectively. Active recombinant (r)PASP (10 μg/20 μL) or heat-inactivated rPASP was intrastromally injected into rabbit corneas. Pathologic changes were monitored by slit lamp examination (SLE) and histopathology. Purified rPASP was assayed for cleavage of collagens and susceptibility to TLCK. Rabbit antibody to rPASP was produced and tested for enzyme inactivation, and actively immunized rabbits were challenged by intrastromal injection of active rPASP (5 μg).
All 25 strains of P. aeruginosa analyzed were positive for the PASP gene and protein. SLE scores of eyes injected with active rPASP were significantly higher than control eyes at all postinjection times (PI; P ≤ 0.004). Histopathologic studies documented the destruction of the corneal epithelial layer and portions of the corneal stroma at 9 hours PI, and polymorphonuclear (PMN) leukocyte infiltration into the cornea by 24 hours after active rPASP injection. PASP cleaved type I and IV collagens and was susceptible to TLCK inhibition. PASP was present in the cytoplasm and periplasm, but only secreted PASP was enzymatically active. A high antibody titer (ELISA titer ≥ 10,000) was produced, but this antibody did not protect against active rPASP challenge.
PASP is a commonly produced Pseudomonas protease that can cleave collagens and cause corneal erosions.
The pathogen Pseudomonas aeruginosa is a Gram-negative bacterium that causes opportunistic infections, especially in patients with cancer, cystic fibrosis, and burns.1–4 It is also documented to cause severe corneal infections, most commonly in association with the use of contact lenses and, at a lesser frequency, after eye injury or surgery.5,6 P. aeruginosa continues to be the leading cause of contact lens–associated bacterial keratitis in the United States.7,8
The virulence of P. aeruginosa is mediated by multiple mechanisms including the production of a wide array of extra-cellular proteases.9 There are several well characterized proteases—namely, elastase A and B, alkaline protease, and protease IV (PIV).10–13 Proteases produced by Pseudomonas can directly damage host tissues and can also indirectly damage the host by activating destructive host responses, such as host matrix metalloproteases.14 Pseudomonas elastase B (LasB) and alkaline protease (AP) are metalloproteases that can degrade a variety of host defense molecules, including complement and surfactant proteins.15–17 PIV, a serine protease, could be the most potent of the characterized Pseudomonas proteases and, unlike the other characterized proteases, is produced by essentially all clinical isolates.18 In contrast to the virulence retention of LasB- and/or AP-deficient mutant,19,20 the loss of the PIV gene has been shown to significantly reduce corneal virulence, and complementation of the mutated gene restored full virulence.21
Tissue damage caused by Pseudomonas proteases occurs independently of viable bacteria and can continue after bacteria are killed by antibiotic therapy. Thus, inhibition of these enzymes by chemical or specific immune therapy would be beneficial in protecting against corneal damage. Immunization against LasB and AP elicits neutralizing antibodies that have been shown to be protective, though to a limited degree, against the intrastromal challenge of whole bacteria.22 Efforts to develop an antibody capable of neutralizing PIV, however, have been unsuccessful, probably because of the low immunogenicity of this protease.23
In addition to the well characterized proteases, P. aeruginosa produces two other proteases: modified elastase and P. aeruginosa small protease (PASP).24 Modified elastase has been identified, but its biochemical properties or virulence potential have not been described. For PASP, one study of its molecular properties and possible importance to corneal virulence has been reported. PASP, as secreted into the culture medium, was found to have a molecular mass of 18.5 kDa. The PASP gene of P. aeruginosa strain PA103 is greater than 99% identical with a gene designated as PA0423 of strain PAO1, a finding that could suggest conservation of the gene among P. aeruginosa strains. DNA sequences of no known function, yet homologous (80%–86%) to PASP, have been detected in Escherichia coli, Pseudomonas putida, and Pseudomonas syringae. Preparations of PASP were found to produce corneal epithelial erosions after injection into rabbit corneas.24
The purpose of the present study was to determine the distribution of the PASP gene among P. aeruginosa strains, the production of this protease among clinical isolates, and the immunogenicity of PASP. Also included is a more detailed analysis of the effects of PASP on the rabbit cornea and its interaction with collagens. The results show that PASP is produced by all tested strains of P. aeruginosa and can cleave collagens and cause corneal erosions. The enzyme is shown to be in an inactive form in the cytoplasm and periplasm, but active after secretion.
The sources of P. aeruginosa strains used herein were described previously.18,24 Cultures were grown in M9 minimal medium containing 60 mM monosodium glutamate, 1 mM MgSO4 and 1% glycerol at 37°C for 20 hours.13 The bacteria were removed by centrifugation at 5000g for 15 minutes. The supernatants were filtered through a 0.22-μm filter and concentrated 50-fold with an ultrafiltration cell with a 10-kDa molecular mass cutoff filter (Amicon; Millipore, Billerica, MA). Total protein concentrations of supernatants were determined by bicinchoninic acid (BCA) assay (Sigma-Aldrich, St. Louis, MO).
Genomic DNA from all Pseudomonas strains tested was purified with a genomic DNA isolation kit (Qiagen, Valencia, CA). Two sets of primers were designed based on the PASP sequence from Pseudomonas strain PA103. One set amplified the full-length PASP gene (573 bp, forward primer: 5′-ATGCTGAAGAAGACCCTTGCCGCG-3′; reverse primer: 5′-TTACTGGCGAAGCCTTCGACGGA-3′) and the other amplified a portion of the gene (173 bp, forward primer: 5′-TCACCATCAAG-GCCAAGCTGATCGGCC-3′; same reverse primer). The PCR conditions were as follows: 100°C for 5 minutes; add Taq polymerase; 94°C for 1 minute; then 30 cycles of 94°C for 20 seconds, 55°C for 20 seconds, and 68°C for 1 minute. Products were electrophoresed on a 1% agarose gel and stained with ethidium bromide.
The construct expressing rPASP was described by Marquart et al.24 Briefly, the PASP gene from Pseudomonas strain PA103 was cloned into plasmid pHAT10 (Clontech, Mountain View, CA) with a histidine affinity tag (HAT) at the N terminus of the protein. The resulting pHAT10-pasp plasmid was expressed in Escherichia coli, and rPASP was purified from inclusion bodies on an affinity column (Talon; Clontech) under denaturing (8 M urea) conditions. The refolding of rPASP was achieved by gradual removal of urea from the suspension. The amount of refolded rPASP in the urea-free buffer (100 mM NaH2PO4, 10 mM Tris-HCl, pH 7.0) was quantified by BCA assay and the activity was verified by zymography using gelatin as substrate as described by Caballero et al.25
Heat-denatured rPASP from E. coli (100 μg) was injected SC into New Zealand White rabbits (Myrtle’s Rabbitry, Inc., Thompson Station, TN) in a 1:1 (vol/vol) mixture with Freund’s complete adjuvant. The rabbits (n = 3) were boosted twice by injecting 50 μg rPASP mixed with Freund’s incomplete adjuvant. Mock-immunized rabbits (n = 3) received adjuvant only. The anti-sera were collected 7 to 10 days after the injections. Antibody titers were determined by ELISA, with rPASP as the plate-bound antigen.
ELISA was performed as previously described.23 The antibody titer represented the midpoint of the titration curve and was expressed as the mean titer ± SEM.
Protein samples of culture supernatants (30 μL, 2 mg/mL) were electrophoresed on 12% SDS-PAGE gels, as described by Marquart et al.24
Osmotic shock and cell lysis were performed according to the method of Engel et al.,13 with some modifications. PA103-29 was grown in 500 mL dialyzed tryptic soy broth at 37°C for 24 hours. Cell lysis was achieved by sonication. The culture supernatant was concentrated 50-fold. The total protein concentration was determined, and samples with a similar amount of total protein were assayed by both Western blot analysis and zymography.
Purified P. aeruginosa LasB and AP were purchased from Elastin Products Company, Inc. (Owensville, MO) and United States Biochemical (Cleveland, OH), respectively. LasA was obtained from Dennis Ohman (Medical College of Virginia Campus of Virginia Commonwealth University, Richmond, VA). LasA, LasB, AP, and rPASP activities were verified by zymography. Recombinant (r)PIV was purified as previously described.18 PIV activity was verified by the hydrolysis of the chromogenic substrate N-(p-Tosyl)-Gly-Pro-Lys 4-nitroanilide acetate (Sigma-Aldrich).13 Fluorescein-conjugated substrates tested (Molecular Probes, Eugene, OR) included gelatin from pig skin, type I collagen from bovine skin, and type IV collagen from human placenta. Each substrate was suspended in a buffer provided by the manufacturer and incubated with rPASP (6 μg), rPIV, LasB, AP, or LasA (0.1 μg each), at 37°C in a 96-well plate after the procedures recommended by the manufacturer. Fluorescence was measured at various time points in a fluorescence microtiter plate reader with excitation at ~495 nm and emission at ~515 nm. Background values derived from the enzyme-free control specimens were subtracted from the fluorescence measurements.
Fluorescein-conjugated gelatin, as mentioned, was incubated with rPASP (10 μg) alone or plus an inhibitor. Fluorescence was measured after 24 hours at 37°C. Controls included gelatin alone and inhibitor plus gelatin alone.
IgG antibodies to rPASP were purified from rabbit anti-sera with protein A agarose beads (Sigma-Aldrich). The purity of IgG was verified by SDS-PAGE analysis. Anti-PASP IgG (1.3 mg/mL) or IgG purified from sera of mock-immunized rabbits (1.2 mg/mL) was incubated with active rPASP (10 μg/20 μL) in a 1:1 (vol/vol) mixture at room temperature overnight. BODIPY FL casein (Molecular Probes) was then added to the mixture, and fluorescence was measured at the various time points.
Specific pathogen-free New Zealand White rabbits used in these studies were maintained according to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. The rabbits were anesthetized by SC injection of a mixture of xylazine (100 mg/mL; Butler Co., Columbus, OH) and ketamine hydrochloride (100 mg/mL; Fort Dodge Animal Health, Fort Dodge, IA). Proparacaine hydrochloride (Bausch and Lomb, Tampa, FL) was topically applied to the eyes before intrastromal injections. Active rPASP (10 μg/20 μL) was injected to analyze pathologic changes, or an equal amount of heat-inactivated rPASP, prepared by boiling for 2 hours at 100°C, was injected as the negative control. Active rPASP (5 μg/10 μL) was also used to challenge actively immunized rabbits. Slit lamp examinations (SLEs) of rabbit corneas were performed by two masked observers at various postinjection (PI) times, with a scoring system ranging from 0 for normal eyes to 28 for a maximally inflamed eyes, as previously described.26,27 Corneal erosions were stained with fluorescein (Fluorets; Chauvin Pharmaceuticals, Aubenas, France) and examined under blue light. Photographs were taken with a slit lamp microscope (Topcon SL-7E; Koaku Kikai, Tokyo, Japan).
Rabbit corneas were intrastromally injected with 10 μg in 20 μL active rPASP or heat-inactivated rPASP (four eyes per group). Rabbits were killed via an intravenous overdose of pentobarbital (Sigma-Aldrich). Whole eyeballs were harvested at 9 (n = 2) and 24 (n = 2) hours PI and fixed in 10% neutral buffered formalin (EK Industries, Joliet, IL). After fixation, specimens were processed as previously described.28,29 Erosions in corneas injected with active rPASP were observed grossly and marked with dye. Microsections were made around the dyed area and stained with hematoxylin and eosin. The stained specimen slides were examined microscopically and photographed under both low and high magnifications.
Mean and SEM were calculated with commercial statistical analysis software (SAS, Cary, NC). For comparison of SLE scores, nonparametric one-way analysis of variance (Kruskal-Wallis test) was performed. P ≤ 0.05 was considered significant.
The gene sequence of PASP from P. aeruginosa strain PA103 has 99.5% identity to the sequence from strain PAO1 (P. aeruginosa genome project reference strain).24 To extend the study of the PASP gene distribution, we analyzed various P. aeruginosa strains (n = 25) for the presence of the PASP gene by PCR. Two sets of primers were used: one set to amplify the full-length 573-bp gene and the other to amplify a 173-bp PASP gene fragment. The genomes of all the P. aeruginosa strains tested were found to be positive for the entire PASP gene (Fig. 1A, lanes 3–14, 16–28) and the PASP gene fragment (Fig. 1B, lanes 3–14, 16–28).
The presence of PASP in culture supernatants was tested by Western blot analysis. The polyclonal antibody to PASP, produced by immunizing the rabbits three times with heat-denatured rPASP, had an ELISA titer of ≥ 10,000 (Fig. 2A). In contrast, the titer of the sera from mock-immunized rabbits was ≤ 120. Sera from mock-immunized rabbits, unlike sera from immunized rabbits, failed to produce positive reactions in an immunoblot assay using rPASP as the antigen (data not shown).
Anti-PASP antibody recognized rPASP as a 27-kDa band and also detected some smaller bands (Fig. 2B, lane 2). The purified rPASP has a molecular mass larger than native PASP (18.5 kDa), because of the presence of the HAT tag and retention of the signal peptide. And the smaller band (23-kDa) in purified preparations of rPASP was previously shown to have the N-terminal sequence expected for PASP.24 In all the P. aeruginosa strains tested, anti-PASP antibody recognized a protein band with the size of ~19 kDa, a size consistent with the molecular mass of native PASP (18.5 kDa; Fig. 2B, lanes 3–10, 12–20, 22–29). In some strains, the antibody reacted with proteins of lower molecular mass that were probably the breakdown products of PASP.
To determine whether active PASP is present inside the bacterial cell or periplasmic space, we tested P. aeruginosa strain PA103-29 for the PASP protein and proteolytic activity by Western blot analysis and zymography, respectively. The PASP protein was found in the periplasmic fraction as well as the cell lysate and culture supernatant (Fig. 3A). On zymogram, there was no protease activity present in the periplasmic fraction or cell lysate. Only the supernatant showed two species of protease activity, one at ~200 kDa and the other at ~80 kDa (Fig. 3B). The 200-kDa band corresponds to that reported for PIV and the 80-kDa band corresponds to PASP.24
Rabbit eyes (six per group) that were injected with 10 μg of active rPASP demonstrated profound pathologic changes, whereas eyes injected with an equal amount of heat-inactivated rPASP lacked such extensive changes (Fig. 4A). At 5 hours PI, eyes injected with active rPASP showed corneal erosions, mild chemosis, and moderate iritis. Control eyes with heat-inactivated rPASP appeared normal.
At 9 hours PI, eyes injected with active rPASP showed even more severe pathologic changes, including the enlargement of erosions and development of mild corneal infiltrate. Control eyes at 9 hours PI demonstrated mild chemosis and injection.
At 24 hours after injection of active rPASP, the corneal infiltrate was more extensive, but corneal epithelial erosions were healing or had healed. In the control eyes injected with inactive rPASP, pathologic changes were limited and consisted of mild chemosis, iritis, and injection.
SLE scores at any time point measured after active rPASP injection were significantly higher than that of control eyes injected with heat-inactivated rPASP (Fig. 4B, P ≤ 0.004). SLE scores increased consistently in eyes injected with active rPASP during the 24-hour experimental course. The rabbits were killed at 24 hours PI and the presence of contaminating bacteria was assayed. All corneas were free of bacterial contamination.
Histopathological analysis was performed on eyes injected with active or heat-inactivated rPASP (four eyes per group). At 9 hours PI, under low power, erosions were identified, and the continuum of the epithelial layer was found to be disrupted (Figs. 5A, 5B). Damage extending down into the corneal stroma was also observed (Fig. 5B). However, limited PMN infiltration was observed at this stage as examined under high power.
At 24 hours PI, histopathologic examination showed that the eyes injected with active rPASP had extensive PMN infiltration into the cornea (Fig. 5C). Active rPASP also induced PMN infiltration into the anterior chamber and limbus area (data not shown). The control eyes injected with heat-inactivated rPASP had very minimal PMN infiltration into the cornea (Fig. 5D), anterior chamber or limbus (data not shown), and the corneal structure appeared normal (Fig. 5D).
Collagens are important structural proteins in the cornea. Type I collagen is the major type of collagen in human corneal stroma, whereas type IV collagen is found in the basement membrane of the corneal epithelium.30 To test the susceptibility of type I and IV collagens to PASP digestion, fluorescence-based assays were performed. Fluorescein-conjugated gelatin and collagens were incubated with major known proteases of P. aeruginosa, including PASP, LasB, AP, LasA, and PIV (Fig. 6). All the proteases were able to digest gelatin, a hydrolyzed form of collagen, indicating their levels of enzymatic activity. PASP was shown to digest both type I and IV collagens. Among the other proteases, LasA, LasB, and AP demonstrated collagenase activity with at least one of the two collagens. PIV was found least efficient in digesting collagens.
The antibody produced by immunization with rPASP was tested for inhibition of rPASP activity. For the neutralization assay (Fig. 7A), 10 μg of active rPASP was incubated with approximately 25 μg of IgG purified from immune or mock anti-sera at room temperature overnight. The fluorescent casein substrate was added to the mixture, and then fluorescence was measured at various time points for 5 hours. The enzymatic activity of rPASP on casein was not inhibited by the anti-PASP IgG antibody. The IgG fraction of sera from mock-immunized rabbits also lacked an inhibitory effect on the enzymatic activity of rPASP.
The effect of PASP antibody on the activity of rPASP was also tested in vivo. The corneas of both mock- and rPASP-immunized rabbits (three rabbits per group) were intrastromally injected with 5 μg of active rPASP and corneal pathology was examined. At 7 hours PI, both rPASP- and mock-immunized rabbits showed equivalent ocular pathology, including epithelial erosion, iritis, chemosis, and injection (data not shown). The corneal infiltrate of both groups of rabbits was profound at 24 hours PI (data not shown). There was no significant difference in SLE scores of the two groups of rabbits at either 7 (P = 0.054) or 24 (P = 0.600) hours PI (Fig. 7B). All injected corneas were healed with minimal scarring at 5 to 7 days PI (data not shown).
The metalloprotease inhibitor EDTA and the serine protease inhibitor TLCK were tested for their ability to block the hydrolysis of fluorescein-conjugated gelatin by rPASP. Only TLCK demonstrated dose-dependent inhibition of rPASP activity (Fig. 8).
The present study shows that PASP can cleave collagens and cause corneal erosions and is secreted by all P. aeruginosa strains tested, including ocular and nonocular clinical isolates and laboratory strains.
The PASP gene has been identified only in P. aeruginosa strain PA103; however, the present study demonstrates that the PASP gene is commonly found in P. aeruginosa isolates. The presence of the PASP gene was confirmed by two PCR reactions targeting different portions of the gene. In addition to the distribution of the PASP gene among P. aeruginosa isolates, the secretion of PASP protein into P. aeruginosa culture supernatants was demonstrated by Western blot analysis of all the strains tested. The findings indicate that PASP, like PIV,18 is a product of a very high percentage of P. aeruginosa isolates, possibly all of them. The finding of PASP and PIV as products of possibly all strains is in contrast to LasB and AP, which are produced by a subset of P. aeruginosa strains.18,31,32 The strains tested in the present study varied in their geographic origins, having been isolated from multiple points of the United States. They also varied in terms of the anatomic site of the infection from which they were isolated. Of the 25 P. aeruginosa strains studied, there were eight nonocular clinical isolates, suggesting that PASP could have a role in Pseudomonas infections other than keratitis. Strain PAO1 and PA103 included in this study are known to be the prototype of the invasive and cytotoxic P. aeruginosa groups, respectively,33 and both strains were found to produce PASP.
The active form of PASP was found only in the culture supernatant, not in the cytoplasm or the periplasmic space, similar to PIV.13 The PASP sequence of PA103 appears to have a signal peptide consistent with that needed for type II secretion.24 The size of the PASP protein, as measured by SDS-PAGE, remains the same in the periplasm or the supernatant. The model of a larger proenzyme being processed into a smaller active enzyme does not apply to PASP. However, there could be some accessory molecules necessary for the maturation of PASP, especially as it exits the periplasmic space.
Purified active rPASP causes corneal erosions similar to native PASP.24 The erosion is dependent on the enzymatic activity of rPASP because the heat-inactivated form did not cause an erosion. The destruction of corneal epithelium has been documented by fluorescein staining and histologic analysis. Histologic analysis has also demonstrated that rPASP can produce a corneal erosion that extends into a significant portion of the stroma. The epithelial erosion and ulcer in the stroma are consistent with the fact that rPASP cleaves type I and IV collagens. LasB was previously reported to cleave both type I and IV collagens, whereas AP cleaves type I collagen.34 Our data confirmed these results in a different assay system. Also determined was that LasA can efficiently cleave both types of collagens whereas PIV cannot. The injection of rPASP into the cornea induced an inflammatory response that included the migration of PMNs into the corneal stroma, limbus, and anterior chamber. The mechanism by which PASP activates an inflammatory response is unknown; however, this could involve an interaction of PASP with a protease-activated receptor (PAR), as described for another P. aeruginosa protease.35 In principle, the pathologic changes in the eye could be the direct effects of PASP or a combination of PASP with activated host matrix MMPs and neutrophil proteases.
The present immunization studies demonstrate that rPASP, unlike PIV, is very immunogenic and anti-rPASP antibody reacts with both native and recombinant PASP. However, this antibody does not interfere with the catalytic site on PASP to neutralize PASP activity. The antibody to PASP afforded no protection in the rabbit eye after challenge with active rPASP. PASP has not yet been well characterized as an enzyme; its cleavage sites on target proteins and its catalytic site have yet to be determined. Amino acid sequence analysis and a structural computer model of PASP offer little information about the active site of PASP (data not shown). The preliminary inhibitor study presented herein indicates that PASP may be a serine protease because TLCK completely inhibits PASP activity. However, sequence homology search relates PASP to YceI-like proteins in the Gram-negative bacteria that are involved in the metabolism of polyisoprenoids. PASP does not have sequence homology to any of the serine proteases. There is a possibility that, in addition to the proteolytic activity, PASP is also involved in the metabolism of polyisoprenoids.
In the present study, the correlation between PASP and corneal virulence has been further strengthened. Among the major Pseudomonas proteases, LasB has been associated with corneal virulence during Pseudomonas keratitis, although multiple clinical isolates, including ocular isolates, are deficient in LasB.18,36 Also, Kernacki et al.37 showed that LasB was not detected in corneas infected with P. aeruginosa. Based on deletion mutant studies, both LasA and AP are not required for corneal virulence.19,20,38–40 PIV, like PASP, is commonly produced by P. aeruginosa strains and is associated with corneal virulence.21,26 The production of PIV and PASP by numerous P. aeruginosa strains suggests that the two proteases could have important roles in keratitis. PIV can cleave a wide spectrum of host proteins (e.g., IgG, complement, surfactants)13,41 and thus help protect the bacteria from host defenses. However, PIV is not efficient in cleaving collagens and does not cause erosions when expressed in the nonpathogenic P. putida during experimental keratitis.42 Preliminary studies of PASP indicate this enzyme is not very efficient in cleaving a variety of host proteins (e.g., IgG) (Tang A et al. IOVS 2007; 48:ARVO E-Abstract 725), but as shown herein, can readily cleave collagens and cause corneal erosions. It is possible that, for a typical strain of P. aeruginosa, PIV helps protect the bacterium from host defenses and PASP could directly contribute to the destruction of ocular tissue during Pseudomonas keratitis.
Supported by National Eye Institute Grant EY12961.
The authors thank Stephen Stray, PhD, for discussions and critical review of the research.
Disclosure: A. Tang, None; M.E. Marquart, None; J.D. Fratkin, None; C.C. McCormick, None; A.R. Caballero, None; H.P. Gatlin, None; R.J. O’Callaghan, None