Type II Secretion-Dependent Degradative and Cytotoxic Activities Mediated by Stenotrophomonas maltophilia Serine Proteases StmPr1 and StmPr2

Abstract

Stenotrophomonas maltophilia is an emerging opportunistic pathogen that primarily causes pneumonia and bacteremia in immunocompromised individuals. We recently reported that S. maltophilia strain K279a encodes the Xps type II secretion system and that Xps promotes rounding, actin rearrangement, detachment, and death in the human lung epithelial cell line A549. Here, we show that Xps-dependent cell rounding and detachment occur with multiple human and murine cell lines and that serine protease inhibitors block Xps-mediated rounding and detachment of A549 cells. Using genetic analysis, we determined that the serine proteases StmPr1 and StmPr2, which were confirmed to be Xps substrates, are predominantly responsible for secreted proteolytic activities exhibited by strain K279a, as well as the morphological and cytotoxic effects on A549 cells. Supernatants from strain K279a also promoted the degradation of type I collagen, fibrinogen, and fibronectin in a predominantly Xps- and protease-dependent manner, although some Xps-independent degradation of fibrinogen was observed. Finally, Xps, and predominantly StmPr1, degraded interleukin 8 (IL-8) secreted by A549 cells during coculture with strain K279a. Our findings indicate that while StmPr1 and StmPr2 are predominantly responsible for A549 cell rounding, extracellular matrix protein degradation, and IL-8 degradation, additional Xps substrates also contribute to these activities. Altogether, our data provide new insight into the virulence potential of the S. maltophilia Xps type II secretion system and its StmPr1 and StmPr2 substrates.

INTRODUCTION

The Gram-negative bacterium Stenotrophomonas maltophilia, prevalent in the aqueous environment, is now emerging as a prominent opportunistic and nosocomial pathogen (1, 2). S. maltophilia infects an array of host tissues and organs, including the respiratory tract, blood, bone, soft tissue, eye, urinary tract, heart, and brain. However, pneumonia is the most common infection associated with S. maltophilia, followed by bloodstream infection as the second most common (1, 2). S. maltophilia, though relatively nonvirulent for healthy individuals, can cause serious infection in immunocompromised individuals. Therefore, S. maltophilia poses the biggest threat for patients with severe burns, cystic fibrosis, and HIV, as well as those undergoing chemotherapy or immunosuppressive therapy (3). Prolonged hospitalization in intensive care units and long-term antibiotic treatment are also considered risk factors for developing pneumonia and bacteremia, for which mortality rates can range from 23 to 77% and 14 to 69%, respectively (1, 4). Community-acquired S. maltophilia infections, especially in the high-risk populations mentioned, have been reported (5), and recently, a community-acquired skin infection was reported for the first time in an immunocompetent individual (6). The multidrug-resistant nature of S. maltophilia makes treatment of infections highly difficult (7), and in recent years, an increased resistance to the preferred antibiotic trimethoprim-sulfamethoxazole has been reported (8–10). These findings emphasize the need to better understand the virulence mechanisms employed by S. maltophilia.

The current understanding of S. maltophilia pathogenesis is limited, although studies have begun to investigate S. maltophilia infection in mammalian and nonmammalian animal models (11–18). In the murine lung, S. maltophilia induces inflammation and neutrophil recruitment, and it has been shown to grow and persist in certain mouse strains (11, 13–15). In all infection models, S. maltophilia strains have displayed various degrees of virulence (13, 15–18), but the molecular mechanisms mediating S. maltophilia pathogenesis have yet to be determined. The presence of virulence traits that commonly promote infection by other genera have been suggested by the sequenced genome of the clinical isolate K279a, as well as several additional sequenced genomes (19, 20). These possible virulence determinants include fimbriae, a siderophore, lipopolysaccharide, secretion systems, and secreted degradative enzymes (20).

Type II secretion (T2S) is one of the six secretion mechanisms found in Gram-negative bacteria. The T2S apparatus is comprised of 12 core proteins, including a cytosolic ATPase (T2S E), inner membrane platform proteins (T2S F, L, and M), pseudopilins that form a pilus-like structure (T2S G, H, I, J, and K), an inner membrane peptidase (T2S O), an outer membrane “secretin” or pore (T2S D), and a protein thought to act as a bridge for the inner and outer membrane factors (T2S C). T2S occurs through a multistep process, where T2S substrates are first translocated across the inner membrane into the periplasm, predominantly through the Sec pathway. The T2S apparatus then promotes translocation of the substrates through the outer membrane secretin into the extracellular milieu via the piston-like motion of the pseudopilus (21). Our laboratory has recently investigated the functionality of two T2S systems in S. maltophilia; i.e., Gsp T2S and Xps T2S (21). By constructing and characterizing gsp and xps mutants of the clinical isolate K279a, we concluded that Xps T2S promotes detrimental effects on the human lung epithelial cell line A549. Specifically, S. maltophilia supernatants caused cell rounding, detachment, and actin rearrangement and exhibited cytotoxicity of A549 cells in an Xps T2S-dependent manner (21). It remains to be determined whether Gsp T2S is functional, as gsp mutants do not lack any of the observed activities exhibited by strain K279a. Analysis of culture supernatants from strain K279a by SDS-PAGE revealed that at least seven proteins are secreted in an Xps T2S-dependent manner (21). T2S substrates, such as degradative enzymes, are commonly associated with the virulence of Gram-negative pathogens (22, 23), and the genome of strain K279a encodes such factors as proteases, lipases, esterases, DNases, RNases, and fibrolysins (19). Here, we report that two serine proteases (StmPr1 and StmPr2) are predominantly responsible for the T2S-mediated effects on A549 cells but also for newly described Xps-mediated degradation of both extracellular matrix (ECM) proteins and the cytokine interleukin 8 (IL-8).

MATERIALS AND METHODS

Bacterial strains and media.

The multidrug-resistant isolate K279a (American Type Culture Collection [ATCC] strain BAA-2423) served as the wild-type (WT) S. maltophilia strain for these studies (Table 1). Mutants used in this study are also listed in Table 1. S. maltophilia strains were routinely cultured at 37°C on Luria-Bertani (LB) agar (Becton, Dickinson, Franklin Lakes, NJ) or LB broth (21). In order to obtain supernatants for enzymatic assays and other analyses, S. maltophilia strains were cultured in a version of buffered yeast extract (BYE) broth consisting of 10 g per liter of yeast extract, 54.9 mM N-(2-acetamido)-2-aminoethanesulfonic acid buffer, and 39.2 mM potassium hydroxide at pH 6.85 to 6.95. Growth in this broth promoted higher levels of protease activity in supernatants (data not shown) than did the previously used version of BYE broth that was supplemented with α-ketoglutarate, l-cysteine, and ferric pyrophosphate (21). When appropriate, media were supplemented with chloramphenicol at 10 to 30 μg/ml, gentamicin (Corning, Tewksbury, MA) at 20 μg/ml, tetracycline at 20 μg/ml, norfloxacin at 5 μg/ml, 1 mM isopropyl β-1-thiogalactopyranoside (IPTG), and/or 10% sucrose. Bacterial growth was monitored by measuring the optical densities of cultures at 600 nm (OD₆₀₀) using a DU 720 spectrophotometer (Beckman Coulter, Indianapolis, IN). Escherichia coli strain DH5α (Life Technologies, Carlsbad, CA) served as a host strain for cloning and propagation of recombinant plasmids. E. coli cells were routinely grown on LB agar or in LB broth at 37°C unless otherwise noted. When appropriate, the media were supplemented with ampicillin (Research Products International, Mt. Prospect, IL) at 100 μg/ml, chloramphenicol at 30 μg/ml, gentamicin at 5 μg/ml, tetracycline at 10 μg/ml, and/or 10% sucrose. All chemicals were from Sigma-Aldrich (St. Louis, MO) unless otherwise noted.

TABLE 1.

S. maltophilia strains used in this study

Strain	Description	Reference
K279	Clinical isolate from blood	21, 59
K279a(pBstmPr1-his)	K279a carrying pBstmPr1-his	This study
K279a(pBstmPr2-his)	K279a carrying pBstmPr2-his	This study
NUS1	gpsF mutant of K279a	21
NUS1(pBstmPr1-his)	gpsF mutant carrying pBstmPr1-his	This study
NUS1(pBstmPr2-his)	gpsF mutant carrying pBstmPr2-his	This study
NUS4	xpsF mutant of K279a	21
NUS4(pBstmPr1-his)	xpsF mutant carrying pBstmPr1-his	This study
NUS4(pBstmPr2-his)	xpsF mutant carrying pBstmPr2-his	This study
NUS4(pBxpsF)	Complemented xpsF mutant	21
NUS5	stmPr1 mutant of K279a	This study
NUS5(pBstmPr1)	Complemented stmPr1 mutant	This study
NUS6	stmPr2 mutant of K279a	This study
NUS6(pBstmPr2)	Complemented stmPr2 mutant	This study
NUS7	stmPr1 stmPr2 mutant of K279a	This study
NUS7(pBstmPr1)	stmPr1 stmPr2 mutant complemented with stmPr1	This study
NUS7(pBstmPr2)	stmPr1 stmPr2 mutant complemented with stmPr2	This study

DNA and protein sequence analysis.

S. maltophilia K279a genomic DNA was isolated as previously described (21). Primers used for sequencing and PCR were obtained from Integrated DNA Technologies (Coralville, IA). Primer names and sequences are listed in Table S1 in the supplemental material. DNA and protein sequences were analyzed using Lasergene (DNASTAR, Madison, WI). BLASTN and BLASTP homology searches were done using GenBank at the NCBI. Signal sequence prediction was performed using the SignalP 4.1 server.

Mutant construction and genetic complementation.

Mutants of strain K279a were constructed using the gene replacement vector pEX18Tc as described previously (21, 24, 25). A mutant (NUS5) with a deletion of the entire stmPr1 coding region (SMLT_RS03270) was obtained using PCR overlap extension protocol as previously described (26). The 5′ and 3′ regions that flank stmPr1 were PCR amplified from K279a DNA with Platinum Pfx polymerase (Life Technologies) using the primer pair SK241 and SK242 and the primer pair SK243 and SK244, respectively. A Flp recombination target (FRT)-flanked chloramphenicol cassette was PCR amplified from pKD3 (27) using primers SK239 and SK240. The overlap extension PCR mixture contained 50 ng each of the three previous PCR products. Three cycles of PCR were done using Platinum Pfx polymerase before the addition of the primer pair SK241 and SK244 for an additional 25 cycles. A band corresponding to the correct size (∼2.3 kb) was gel purified and digested with BamHI and PstI (New England BioLabs [NEB], Ipswich, MA) and ligated into pEX18Tc digested with the same enzymes, yielding pEXΔstmPr1::frt-cat-frt. The newly made plasmid was moved into E. coli S17-1 (28) and then mobilized from there into S. maltophilia K279a via conjugation (21). To select for S. maltophilia carrying the pEXΔstmPr1::frt-cat-frt mutagenizing construct, which provides resistance to tetracycline from the vector backbone and resistance to chloramphenicol from the mutated gene, transconjugants were plated on LB agar supplemented with tetracycline, chloramphenicol (10 μg/ml), and norfloxacin. Strain K279a is resistant to norfloxacin (21); therefore, this antibiotic was added to select against outgrowth of the S17-1 E. coli strain. Resistant colonies were streaked onto LB agar supplemented with 10% sucrose (wt/vol) and chloramphenicol (10 μg/ml) to select for cells in which recombination and loss of the pEX18Tc vector has occurred. Replacement of stmPr1 with chloramphenicol-flanked FRT sites was confirmed by PCR using mutant NUS5 DNA as a template with the primer pair SK246 and SK247. To perform Flp-mediated excision of the chloramphenicol cassette in the stmPr1 mutant, pBSFlp (27) was electroporated into NUS5 and transformants were selected on LB agar supplemented with gentamicin and IPTG (1 mM). Individual colonies were patched onto LB agar containing either chloramphenicol (10 μg/ml), gentamicin, or no selection. Colonies that were either chloramphenicol or gentamicin sensitive were streaked onto LB agar with 10% sucrose (wt/vol). Deletion of stmPr1 was confirmed by PCR using the primer pair SK246 and SK247.

A mutant (NUS6) specifically lacking stmPr2 (SMLT_RS04110) was obtained by separately PCR amplifying the 5′ and 3′ ends of the gene and flanking DNA from K279a DNA using the primer pairs SK132-SK134 and SK135-SK133, respectively. Each of the generated fragments was ligated into pGEM-T Easy (Promega, Madison, WI), and then the two resulting plasmids were digested with SmaI and SalI (NEB). A trimolecular ligation was performed by placing a gentamicin resistance (Gm^r)-containing cassette obtained from pX1918 (29) that had been digested with PvuII and HincII (NEB) between the beginning and end of stmPr2. The plasmid thus obtained (pGstmPr2::Gm^r) carried a 1,602-bp deletion in the central stmPr2 coding region. The stmPr2 fragment was ligated into pEX18Tc by digestion with EcoRI (NEB), yielding pEXstmPr2::Gm^r. Plasmid pEXstmPr2::Gm^r was moved into E. coli S17-1 and then mobilized into S. maltophilia K279a via conjugation. Transconjugants were selected on LB agar supplemented with tetracycline, gentamicin, and norfloxacin. Resistant colonies were streaked on LB agar supplemented with 10% sucrose (wt/vol) and gentamicin. Mutation of stmPr2 was confirmed by PCR using the primer pair SK132 and SK133.

A mutant (NUS7) lacking both stmPr1 and stmPr2 was constructed by mobilizing the pEXstmPr2::Gm^r plasmid carried by the E. coli S17-1 strain into mutant NUS5 containing the stmPr1 deletion. Transconjugants were selected as described above, and mutation of stmPr2 was confirmed by PCR using the primer pair SK132 and SK133.

For trans-complementation of mutants lacking stmPr1 (NUS5 and NUS7), a 2.2-kb PCR fragment containing the stmPr1-coding region plus its 258-bp promoter region was PCR amplified from K279a DNA using primers SK251 and SK252. To complement mutants lacking stmPr2 (NUS6 and NUS7), a 1.74-kb PCR fragment containing the stmPr2 coding region plus its 262-bp promoter region was amplified from K279a DNA with primers AD01 and AD02. The stmPr1 and stmPr2 fragments were digested with ApaI and SacI (NEB) and cloned into pBBR1MCS (30) digested with the same enzymes, yielding pBstmPr1 and pBstmPr2. pBstmPr1 was electroporated (21) into the stmPr1 mutant NUS5 and the stmPr1 stmPr2 mutant NUS7, and pBstmPr2 was electroporated into the stmPr2 mutant NUS6 and NUS7. Transformants were selected on LB agar supplemented with chloramphenicol (30 μg/ml). Cm^r clones were confirmed as carrying pBstmPr1 or pBstmPr2 by PCR using the primer pair SK251 and SK252 or the primer pair AD01 and AD02, respectively.

Detection of secreted proteins and activities.

For collection of culture supernatants, single colonies from S. maltophilia strains grown on LB agar were inoculated into 4 ml of BYE broth and then incubated overnight at 37°C with 220 rpm shaking. Following a 1:10 dilution of the culture into 25 ml of BYE broth in 125-ml flasks, bacteria were grown with shaking at 37°C to early stationary phase (OD = 3.3 to 3.5). Supernatants were collected by centrifugation of the cultures at 5,000 × g for 30 min at 4°C, followed by filtration through 0.22-μm syringe filters (EMD Millipore, Billerica, MA). In order to visualize secreted proteins, the supernatants were precipitated with 10% (vol/vol) trichloroacetic acid at 4°C overnight, followed by centrifugation at 5,000 × g for 15 min at 4°C. The pellets were incubated with 100% ethanol at 4°C for 30 min and then centrifuged at 5,000 × g for 15 min at 4°C. The pellets were then resuspended in 1× Laemmli sample buffer (31) containing 5% β-mercaptoethanol. Samples were heated for 5 min at 95°C before separation on a 10% SDS-PAGE gel. After electrophoresis, protein bands were visualized by overnight staining with the SYPRO ruby total protein stain (Bio-Rad) and imaged with an AlphaImager system (Alpha Innotech, San Jose, CA).

To specifically track the secretion of StmPr1 and StmPr2 by S. maltophilia, a 6-histidine (6×His) tag was added to the C terminus of each protein. The plasmid constructs containing the tagged proteins were made using the procedure described above for trans-complementation, except that different 3′ primers were used in order to provide the 6×His tag. For pBstmPr1-his, the primer pair SK251 and AD03 was used, and for pBstmPr2-his, the primer pair AD01 and AD04 was used. pBstmPr1-his and pBstmPr2-his were electroporated into wild-type K279a, its xpsF mutant NUS4, and its gspF mutant NUS1. These constructs were also electroporated into the stmPr1 stmPr2 mutant NUS7, and their ability to restore serine protease activity in the stmPr1 stmPr2 mutant supernatant was confirmed (data not shown), indicating that the His tag does not interfere with protease activity. As described above, early-stationary-phase supernatants were then collected from these newly generated strains, and the proteins were precipitated and subjected to SDS-PAGE. The supernatant material was then transferred to a polyvinylidene fluoride membrane and proteins were visualized by Ponceau S (Bio-Rad) staining and immunoblotting with a mouse anti-6×His horseradish peroxidase (HRP)-conjugated antibody (Invitrogen).

Assays for secreted gelatinase, caseinolytic, and serine protease activities were performed as previously described (21). To evaluate degradation of ECM proteins by bacterial supernatants, 10 μg of human type I collagen, human fibrinogen, or human fibronectin was incubated with 25 μl of S. maltophilia supernatants for 16 h at 37°C. As a positive control for degradation, the ECM proteins were also incubated with 0.2 μg of trypsin. Degradation of the ECM proteins was evaluated by SDS-PAGE analysis and total protein staining with colloidal Coomassie.

Assays for host cell rounding, detachment, actin rearrangement, and viability.

The human A549 cell line (ATCC CCL-185) was routinely passaged in Roswell Park Memorial Institute (RPMI) medium (Corning) containing 10% fetal bovine serum (FBS) (Atlanta Biologicals, Lawrenceville, GA) (21). The HeLa, 3T3, and MLE cell lines were cultured using conditions suggested by the ATCC. The human HeLa cell line (ATCC CCL-2) and the murine 3T3 cell line (ATCC CRL-2752) were routinely cultured in Dulbecco's modified Eagle's medium (DMEM) (Corning) containing 10% FBS. The murine MLE 12 cell line (ATCC CRL-2110) was cultured in DMEM/F-12 (Corning) supplemented with 5% FBS, 2 mM glutamine (Life Technologies), 10 nM hydrocortisone (Corning), 10 nM beta-estradiol, and insulin-transferrin-selenium (ITS) universal culture supplement (Corning). All cells were maintained at 37°C in 5% CO₂.

In order to monitor the effects of bacterial supernatants, the mammalian cells were seeded in a 96-well tissue culture plate (BD Falcon, Franklin Lakes, NJ) at a density of 5 × 10⁴ cells/well and were allowed to adhere for 2 h. Once adherent, monolayers were washed twice with serum-free RPMI without FBS. Bacterial supernatants, obtained as described above, were then incubated with mammalian cells in serum-free RPMI for 3 or 24 h at 37°C with 5% CO₂ in triplicate. For cell rounding experiments, images were captured using a 40× objective on the EVOS imaging system (AMG; Life Technologies) (21). As before (21), to evaluate detachment of the mammalian cells following 3- or 24-h incubation with S. maltophilia supernatants, cells were washed three times with PBS to remove nonadherent cells, and then 100 μl of PBS was added to the wells. The remaining cells were quantified by measuring absorbance at 600 nm using the Synergy H1 plate reader (BioTek, Winooski VT). Effects on the actin cytoskeleton in A549 cells were monitored as previously described (21).

To assess the effect of protease inhibitors on cell rounding and detachment, A549 cells were seeded and washed as described above, but protease inhibitors were added to cells immediately prior to addition of the bacterial supernatants. To make the protease inhibitor cocktail, one EDTA-free complete protease inhibitor tablet (Roche, San Francisco, CA) was resuspended in 1.5 ml of double-distilled water (ddH₂O) (7× cocktail) and 10 μl of the cocktail was added to cells. The serine protease inhibitors phenylmethane sulfonyl fluoride (PMSF) (MP Biomedicals, Santa Ana, CA) and chymostatin were used at 0.1 mM resuspended in isopropanol and at 5 μM resuspended in dimethyl sulfoxide, respectively.

To examine the effect of S. maltophilia secreted proteins on host cell viability, A549 cells were seeded into a 96-well tissue culture plate and washed prior to addition of early-stationary-phase supernatants as described above. Bacterial supernatants were incubated with A549 cells for 24 h at 37°C with 5% CO₂. Cellular metabolism was evaluated at this time point using PrestoBlue (Life Technologies), as previously described (21).

Detection of IL-8 secretion and degradation.

To detect IL-8 secretion by epithelial cells during coculture with S. maltophilia, A549 cells were seeded in a 24-well tissue culture plate at a density of 5 × 10⁵ cells/well and allowed to adhere for 2 h in 1 ml of RPMI with 10% FBS. Prior to addition of bacteria in duplicate wells, monolayers were washed twice with 0.5 ml of serum-free RPMI without FBS. The bacterial inocula were prepared by resuspending bacteria grown on LB agar to an OD₆₀₀ of 0.215 in PBS. Following dilution in PBS, approximately 5 × 10⁵ CFU were added per well in order to obtain a multiplicity of infection (MOI) of 1. The bacteria and A549 cells were cocultured for 24 h at 37°C with 5% CO₂. A549 cells treated with PBS alone were also included as an uninfected control. CFU were monitored during coculture, and no differences in growth were observed between the WT and mutant strains (data not shown). The coculture media were then collected by aspiration at 2, 4, 6, 8, and 24 h in order to analyze the levels of secreted IL-8. The coculture media were centrifuged at 1,000 × g for 10 min to pellet any host cells, and then the supernatant was aliquoted and frozen at −80°C. Secreted IL-8 levels in the supernatant were determined using the human IL-8 ELISA Ready-Set-Go! kit (eBioscience, San Diego, CA). Plates (Thermo Scientific) were coated, and an enzyme-linked immunoabsorbent assay (ELISA) was performed according to the company's specified protocol. IL-8 levels were quantified by measuring absorbance at 450 nm and 570 nm using the Synergy H1 plate reader, where the values obtained at 570 nm were subtracted from those obtained at 450 nm. To evaluate degradation of IL-8 by bacterial supernatants, 2.5 ng of carrier-free recombinant human IL-8 (BioLegend, San Diego, CA) was incubated with 100, 50, or 10 μl of S. maltophilia supernatants for 16 h at 37°C. IL-8 was also incubated with 0.2 μg of trypsin as a positive control. IL-8 levels following incubation with bacterial supernatant were determined by ELISA as described above.

Statistical analysis.

Data were analyzed using analysis of variance (ANOVA) and Tukey's multiple comparison posttest (GraphPad Prism version 6.0; GraphPad Software).

RESULTS

Xps T2S causes rounding and detachment of epithelial and fibroblast lines of human and murine origin.

In order to investigate whether Xps T2S can mediate morphological effects on cell types in addition to the human A549 cell line, we performed cell-rounding and detachment experiments with MLE (murine lung epithelium), 3T3 (murine embryonic fibroblasts), and HeLa (human cervical epithelium) cells. Upon incubation with WT strain K279a supernatant, all cell lines exhibited similar levels of rounding and detachment, with the exception that 3T3 cells displayed significantly more detachment than did the human cell lines when a 6.25% (vol/vol) dose of supernatant was used (Fig. 1A and B). The observed morphological effects on all of the cell lines was attributed to Xps T2S, as supernatant from the xpsF mutant NUS4 did not induce any cell rounding or detachment, and the complemented xpsF mutant supernatant had an effect similar to that of WT supernatant (Fig. 1A and C). Some of the cell lines treated with xpsF mutant supernatant displayed an increased number of adherent cells compared to medium alone (Fig. 1C; also, see Fig. 2A and 6). A possible explanation is that the bacterial supernatant, when noncytotoxic, provides a growth-stimulating factor and/or nutrients that promote cell growth. Alternatively, the noncytotoxic bacterial supernatant may promote increased cell adherence to the tissue culture plate compared to the medium control. Together, these data demonstrated that an Xps T2S substrate(s) causes rounding and detachment of both human and murine cells.

FIG 1.

Effect of WT and Xps mutant supernatant on various human and murine cell lines. (A) The A549, HeLa, MLE, and 3T3 cell lines were incubated for 3 h with 25% (vol/vol) supernatants from strain K279a (WT), the xpsF mutant NUS4 (xpsF), the complemented xpsF mutant (xpsF/pxpsF), or a medium control. Mammalian cell morphology was evaluated by phase-contrast light microscopy. Data are representative of three independent experiments. (B) Cell lines were incubated with WT supernatant at the indicated doses for 3 h and washed, and then the remaining adherent cells were quantified by measuring absorbance at OD₆₀₀. Data were normalized to the value for cells treated with medium alone, which was set at 100% adherence. An asterisk indicates a statistically significant difference from A549 and HeLa cells (P < 0.05). (C) Cell lines were incubated for 3 h with 25% (vol/vol) supernatants from the WT, the mutant, and the complemented mutant. Cell detachment was determined as described for panel B. Single and double asterisks indicate statistically significant differences from the WT and the xpsF strain, respectively (P < 0.05). For panels B and C, data are as means and standard errors of the means (SEM) from three independent experiments.

FIG 2.

Serine protease activity and the stmPr1 and stmPr2 genes of strain K279a. (A) A549 cells were incubated for 3 h with supernatants from WT K279a and the xpsF mutant NUS4 in the presence or absence of a protease inhibitor (PI) cocktail or the serine protease inhibitor PMSF or chymostatin. Cell detachment was analyzed as described for Fig. 1. Data are the means and SEM from three independent experiments. An asterisk indicates a statistically significant difference from WT with no inhibitor (P < 0.05). (B) Map of the stmPr1 and stmPr2 loci in strain K279a. Horizontal arrows denote relative size and orientation of the depicted genes. Annotated flanking genes are indicated, and accession numbers are given for nonannotated flanking genes. (C) Domain structure of StmPr1 and StmPr2. The signal peptide, the S8 peptidase or catalytically active domain (CATD), and the bacterial prepeptidase domain (PPC) domain are highlighted along with the predicted catalytic triad (Asp-His-Ser). The size ranges for the predicted proenzymes and mature proteins are indicated.

FIG 6.

Effect of StmPr1 and StmPr2 on the detachment and death of A549 cells. (A and B) A549 cells were incubated for 3 h (A) or 24 h (B) with supernatants from the WT, the xpsF mutant, and the indicated protease mutants and their complements. Cell detachment was analyzed as described for Fig. 1. Single and double asterisks indicate statistically significant differences from the WT and the respective mutant, and the phi (Φ) indicates a statistically significant difference from the stmPr2 strain (P < 0.05). (C) A549 cells were incubated for 24 h with supernatants and then cell viability was measured. Data were normalized to the result for cells treated with medium alone, which was set at 100% viability. Single and double asterisks indicate statistically significant differences from the WT and the respective mutant, and a phi (Φ) indicates a statistically significant difference from the xpsF strain (P < 0.05). For panels A to C, data are the means and SEM from three independent experiments.

Xps-mediated rounding of A549 cells is dependent on protease activity.

We next set out to identify what Xps T2S substrate(s) is responsible for the observed rounding and detachment of A549 cells, as these cells are the most physiologically relevant with regard to respiratory infection by S. maltophilia. We have previously shown that the supernatant from strain K279a exhibits multiple protease activities, including Xps-dependent serine protease activity (21). Thus, we incubated A549 cells with WT supernatant in the presence or absence of various protease inhibitors and monitored changes in cell morphology and attachment. A protease inhibitor cocktail, as well as the serine protease inhibitors PMSF and chymostatin, abolished the ability of the WT supernatant to cause cell rounding and detachment, giving a result that was comparable to that obtained with the xpsF mutant supernatant (Fig. 2A and data not shown). These results indicated that Xps-mediated rounding of A549 cells is due to a serine protease(s).

Serine proteases StmPr1 and StmPr2 are secreted by Xps T2S.

Early on, it was determined that a bronchial isolate of S. maltophilia secretes a 47-kDa serine protease (StmPr1) that, when purified from the bacterial culture supernatant, can cause rounding of human fibroblasts (32). Subsequent genomic analysis revealed that strain K279a carries the stmPr1 gene (SMLT_RS03270) (18, 19, 33), and it is now apparent that stmPr1 maps directly upstream of the first gene in the xps operon (Fig. 2B). Interestingly, strain K279a also carries the gene for a second serine protease, StmPr2, (SMLT_RS04110), which, like StmPr1, is a member of the S8 peptidase family (18). In the strain K279a genome, the stmPr2 gene is found approximately 200 kb from stmPr1 (Fig. 2B). Alignment of the 630-amino-acid (aa) protein StmPr1 and the 580-aa StmPr2 from strain K279a revealed that the two proteins share 45.8% sequence identity and 58% sequence similarity (see Fig. S1 in the supplemental material). StmPr1 and StmPr2 also have similar domain structures, as both proteins are predicted to exist as proenzymes and to contain a conserved catalytic triad (Asp-His-Ser) within the mature form of the protein (Fig. 2C) (32, 34). Both StmPr1 and StmPr2 also have a predicted signal sequence (Fig. 2C) (32, 34), suggesting that they are substrates for a T2S system and possibly contribute to the rounding and detachment of A549 cells that is induced by strain K279a supernatant.

To confirm whether one or both of these proteases are secreted via T2S, StmPr1 and StmPr2 were engineered with a C-terminal His tag and expressed on a plasmid within WT, xpsF mutant, and gspF mutant (NUS1) backgrounds. Supernatants from early-stationary-phase cultures were then analyzed by SDS-PAGE, Ponceau S staining, and immunoblotting. When a total protein stain was performed on the supernatant material, both the proenzyme (55 to 57 kDa) and mature forms (44 to 47 kDa) of StmPr1 and StmPr2 were detected in the samples collected from the WT strain and the gspF mutant but not in the supernatant from the xpsF mutant (Fig. 3A). Similarly, by immunoblotting, His-tagged StmPr1 and StmPr2 were observed only in WT and gspF mutant supernatants (Fig. 3B). Of note, the His-tagged proteins detected by immunoblotting were approximately 55 to 57 kDa, consistent with the proenzyme form of StmPr1 and StmPr2 (Fig. 3B). The absence of bands corresponding to the mature forms of StmPr1 and StmPr2 in the immunoblot suggests that the C terminus of the mature proteins may be altered or cleaved, thus preventing detection of the His tag. A similar result was recently observed for a C-terminally His-tagged protease secreted by the Vibrio cholerae T2S system (35). The His-tagged proteases were also not detected in bacterial lysate from the xpsF mutant by immunoblotting, suggesting that, in the absence of T2S machinery, StmPr1 and StmPr2 do not accumulate within an xpsF mutant (data not shown). Taken together, these data demonstrated that StmPr1 and StmPr2 are secreted through the Xps T2S system but not the Gsp T2S system.

FIG 3.

Secretion of StmPr1 and StmPr2 by WT K279a and T2S mutants. His-tagged StmPr1 (Pr1) and StmPr2 (Pr2) were exogenously produced from a plasmid in WT strain K279a, the xpsF mutant NUS4, and the gspF mutant NUS1. Supernatants from these strains were analyzed by SDS-PAGE, followed by Ponceau S stain (A) and anti-His immunoblotting (B). The proenzyme (Pro) and mature forms of StmPr1 and StmPr2 are indicated. The migration of molecular mass standards (in the rightmost and leftmost lanes) is indicated to the left of the images (in kilodaltons). Data are representative of three independent experiments.

StmPr1 and StmPr2 are responsible for the gelatinase, caseinolytic, and serine protease activities in K279a.

To begin characterizing the relative importance of StmPr1 and StmPr2 for S. maltophilia secreted activities, we constructed mutants of strain K279a lacking stmPr1 (NUS5), stmPr2 (NUS6), or both stmPr1 and stmPr2 (NUS7). All three mutant strains grew similarly to the WT strain in broth culture (data not shown), indicating that the mutations do not result in a general growth defect. Supernatants from the WT, the xpsF mutant, and the three protease mutants were then collected and analyzed by SDS-PAGE and total protein staining. As expected, the stmPr1 mutant lacked a 57- and a 47-kDa protein species corresponding to the proenzyme and mature form of StmPr1, the stmPr2 mutant lacked a 55- and a 44-kDa protein species corresponding to the proenzyme and mature form of StmPr2, and the stmPr1 stmPr2 mutant was missing all four protein species (Fig. 4A, left). In the WT culture supernatant, the 47-kDa mature form of StmPr1 was the most abundant of the four protein species (Fig. 4A, left). Secretion of StmPr1 and StmPr2 in the respective stmPr1 or stmPr2 mutant background was restored by providing stmPr1 or stmPr2 in trans, and the stmPr1 stmPr2 mutant was complemented with either stmPr1 or stmPr2 (Fig. 4A, right). Not surprisingly, expression of stmPr1 and stmPr2 from their native promoters on a plasmid resulted in overproduction of the proteases in the complemented mutants. This was especially true for the stmPr1 complementation; i.e., only one-third of the total sample was loaded for the stmPr1-complemented mutants' supernatants (Fig. 4A, right). Interestingly, secreted StmPr1 was predominantly in the mature form, while secreted StmPr2 was predominantly in the proenzyme form, and this was observed for both chromosomally encoded and plasmid-coded protein (Fig. 4A).

FIG 4.

Expression and enzyme activities of StmPr1 and StmPr2 in K279a culture supernatant. (A) Supernatants collected from the WT, the xpsF mutant NUS4, the stmPr1 mutant NUS5, the stmPr2 mutant NUS6, and the stmPr1 stmPr2 mutant NUS7, as well as the indicated complemented mutants, were subjected to SDS-PAGE analysis, and secreted proteins were visualized by SYPRO ruby staining. The arrowhead indicates the proenzyme of StmPr1 (∼57 kDa), and the double arrowhead indicates the proenzyme of StmPr2 (∼55 kDa). The single asterisk indicates the mature form of StmPr1 (∼47 kDa), and the double asterisk indicates the mature form of StmPr2 (∼44 kDa). The migration of molecular mass standards (in kilodaltons) is indicated to the left of the gel images. Data are representative of three independent experiments. (B) The WT, the indicated mutants, and the complemented mutants were inoculated onto plates containing 4% gelatin. Gelatinase activity was evaluated after 3 days at 37°C by measuring the zone of clearing surrounding the area of growth. (C and D) Supernatants from the WT, the indicated mutants, and the complemented mutants were incubated with azocasein for 30 min (C) or N-succinyl-Ala-Ala-Pro-Phe-pNA (D) for 60 min at 37°C to evaluate caseinolytic and serine protease activity, respectively. Single and double asterisks indicate statistically significant differences from the WT and the respective mutant strain, and the phi (Φ) indicates a statistically significant difference from the xpsF and stmPr1 stmPr2 strains (P < 0.05). For panels B to D, data are the means and SEM from three independent experiments.

We have previously shown that hydrolysis of gelatin, azocasein, and the serine protease substrate N-succinyl-Ala-Ala-Pro-Phe-p-nitroanilide (pNa) by strain K279a supernatant is completely dependent on a functional Xps T2S system (21). To evaluate the contribution of StmPr1 and StmPr2 to the ability of S. maltophilia to hydrolyze these substrates, we performed in vitro protease assays with the various protease mutant strains. The stmPr1 mutant and the stmPr2 mutant did not exhibit any loss in the ability to hydrolyze gelatin or azocasein (Fig. 4B and C). However, the stmPr1 stmPr2 double mutant exhibited significantly less gelatinase and caseinolytic activity than the WT strain, demonstrating that both StmPr1 and StmPr2 contribute to the hydrolysis of these substrates (Fig. 4B and C). In support of these results, complemented mutant strains had gelatinase and caseinolytic activity above WT levels (Fig. 4B and C). We did not observe a significant difference in gelatin or azocasein hydrolysis between the xpsF mutant and the stmPr1 stmPr2 mutant (Fig. 4B and C). This result indicated that StmPr1 and StmPr2 are predominantly responsible for the secreted gelatinase and caseinolytic activity exhibited by strain K279a.

When N-succinyl-Ala-Ala-Pro-Phe-pNa was used as a substrate, supernatant from the stmPr1 mutant displayed an approximately 60% reduction in hydrolytic activity (Fig. 4D). Serine protease activity was completely restored in the complemented stmPr1 mutant supernatant (Fig. 4D), confirming a role for StmPr1 in secreted serine protease activity. In contrast, when supernatant from the stmPr2 mutant was incubated with N-succinyl-Ala-Ala-Pro-Phe-pNa, WT levels of substrate hydrolysis were observed (Fig. 4D). However, significantly less serine protease activity was seen in supernatant from the stmPr1 stmPr2 mutant than in that from the stmPr1 mutant supernatant, indicating that StmPr2 is responsible for the residual serine protease activity observed in the stmPr1 mutant supernatant. Furthermore, supernatant collected from the stmPr1 stmPr2 mutant complemented with either stmPr1 or stmPr2 exhibited WT levels of serine protease activity, providing additional evidence that both StmPr1 and StmPr2 hydrolyze N-succinyl-Ala-Ala-Pro-Phe-pNa (Fig. 4D). Similar to the gelatinase and caseinolytic assays, we did observe that complementation of mutants with stmPr1 or stmPr2 resulted in serine protease activity higher than that of the WT strain when early time points (0 to 30 min) were evaluated (data not shown). Overall, these data demonstrated that StmPr1 is predominantly responsible for the serine protease activity in strain K279a supernatant, but StmPr2 does play a lesser role. The greater role for StmPr1 may be attributed to the higher abundance of StmPr1 than StmPr2 in the supernatants (Fig. 4A). Finally, we observed that supernatants from the xpsF mutant and stmPr1stmPr2 mutant were equally impaired in hydrolysis of N-succinyl-Ala-Ala-Pro-Phe-pNa (Fig. 4D), indicating that secreted serine protease activity, like the secreted gelatinase and caseinolytic activities, was completely dependent on StmPr1 and StmPr2.

Xps T2S, StmPr1, and StmPr2 promote the degradation of ECM proteins.

Purified StmPr1 and a growing number of S8 serine proteases that are secreted by environmental and pathogenic bacteria have collagenolytic activity (32, 36–40). Thus, to evaluate the relative importance of Xps, StmPr1, and StmPr2 in the degradation of collagen as well as other ECM proteins, degradation assays were performed by incubating human type I collagen, fibrinogen, and fibronectin with supernatants from WT strain K279a, the xpsF mutant, and the protease mutant strains. Compared to the medium control, supernatant from the WT strain degraded all three ECM proteins, with type I collagen being the least susceptible to degradation (Fig. 5). Degradation of the ECM proteins was predominantly dependent on Xps T2S, as the xpsF mutant supernatant was greatly impaired in substrate breakdown, and the complemented xpsF mutant exhibited WT levels of degradation (Fig. 5). Interestingly, the α chain of fibrinogen (∼70 kDa) still was degraded when fibrinogen was incubated with the xpsF mutant supernatant (Fig. 5), suggesting the existence of a protease activity that is not Xps dependent. Degradation of the α chain was also not dependent on Gsp T2S, as the gspF mutant supernatant behaved like the WT supernatant (see Fig. S2 in the supplemental material). Taken together, these data indicated that the Xps T2S system of strain K279a is primarily responsible for ECM degradation, but some degree of ECM protein degradation also occurs in an Xps- and Gsp-independent manner.

FIG 5.

Degradation of ECM proteins by WT, xpsF mutant, and protease mutant supernatants. Human type I collagen, fibrinogen, and fibronectin were incubated at 37°C for 16 h with 25 μl of supernatant from the WT, the xpsF mutant NUS4, and its complement and from the indicated protease mutants and their complemented derivatives. ECM protein degradation was analyzed by SDS-PAGE and total protein staining with Coomassie. ECM proteins incubated with medium alone or 0.2 μg of trypsin were included for comparison. The type I collagen γ, β, α1, and α2 chains are indicated, as are the fibrinogen α, β, and γ chains. The migration of molecular mass standards (in kilodaltons) is indicated to the left of gel images. Data are representative of three independent experiments.

Turning to the contributions of StmPr1 and StmPr2 to the degradation of fibrinogen and fibronectin, the stmPr1 mutant gave less extensive breakdown of the ECM proteins, as evidenced by the presence of more high-molecular-weight bands in the samples treated with mutant supernatant than in those treated with WT supernatant (Fig. 5). In contrast, stmPr2 mutant supernatant and WT supernatant yielded the same breakdown products (Fig. 5). These results indicated that StmPr1 is predominantly responsible for the ECM protein degradation observed with WT supernatants. However, supernatant from the stmPr1 stmPr2 mutant caused less degradation of fibrinogen and fibronectin than did the stmPr1 mutant supernatant, uncovering a partial role for StmPr2 (Fig. 5). With the type I collagen substrate, it was more difficult to discern differences in degradation between the mutant and WT supernatants, because the WT supernatant did not induce the formation of breakdown products but rather resulted in a slight decrease in the abundance of the various collagen chains (Fig. 5). However, incubating type I collagen with supernatant from the complemented mutants did result in visible breakdown products, indicating that both StmPr1 and StmPr2 can degrade collagen (Fig. 5). With all ECM proteins tested, we saw more extensive degradation by the complemented strains than the WT, which can be attributed to the high levels of StmPr1 and StmPr2 secreted by the complemented strains (Fig. 4A). Complementing the stmPr1 stmPr2 mutant with stmPr1 versus stmPr2 resulted in different degradation patterns, with StmPr1 inducing more extensive breakdown overall (Fig. 5). Finally, we observed that supernatant from the stmPr1 stmPr2 mutant caused more degradation of fibrinogen and fibronectin than did xpsF mutant supernatant (Fig. 5), indicating that Xps T2S substrates in addition to StmPr1 and StmPr2 contribute to degradation of these ECM proteins. Altogether, these data indicated that both StmPr1 and StmPr2 can degrade human type I collagen, fibrinogen, and fibronectin, but StmPr1 is predominantly responsible for the degradation of the ECM proteins.

StmPr1 and StmPr2 are primarily responsible for the detachment and death of A549 cells.

To evaluate the role of StmPr1 and StmPr2 in Xps-mediated rounding and detachment of host cells, we tested the effect of supernatants from K279a WT, the xpsF mutant, and the protease mutant strains on A549 cells. When lower doses of supernatant (6.25% or 12.5%) were incubated with the host cells for 3 h, the stmPr1 mutant and stmPr2 mutant, but not their complemented derivatives, caused significantly less detachment than the WT (Fig. 6A). At the 12.5% dose, the stmPr2 mutant supernatant caused more detachment than did the stmPr1 mutant supernatant (Fig. 6A), indicating that StmPr1 contributes more than StmPr2 does to A549 cell detachment. At the 25% dose of supernatant, the stmPr1 mutant and the stmPr2 mutant caused WT levels of detachment (Fig. 6A), indicating that deletion of stmPr1 or stmPr2 alone is not always sufficient to eliminate supernatant-induced detachment. However, at the 25% dose, as well as at the two lower doses, the stmPr1 stmPr2 mutant was as impaired as the xpsF mutant, and this lack of detachment was complemented with either stmPr1 or stmPr2 (Fig. 6A). Together, these data indicated that after a 3-h incubation, the Xps-mediated detachment of A549 cells is completely dependent on StmPr1 and StmPr2. We previously reported that actin rearrangement occurs in rounded A549 cells that had been treated with strain K279a supernatant for 1 to 3 h and that the observed rearrangement is dependent on Xps T2S (21). Consistent with the impaired ability to trigger protease-mediated rounding and detachment of A549 cells, the protease mutant supernatants were also less able to induce actin rearrangement (data not shown). When A549 cells are evaluated after a 24-h incubation with supernatant, the xpsF mutant is completely impaired in the ability to cause cell detachment (21), and similar results were observed with a stmPr1 stmPr2 mutant when a 12.5 or a 6.25% dose of supernatant was incubated with cells (Fig. 6B). However, when a 25% dose of supernatant was applied, the stmPr1 stmPr2 mutant did cause detachment of A549 cells (Fig. 6B). Together, these data indicated that there are other T2S substrates besides StmPr1 and StmPr2 that contribute to A549 rounding and detachment, but StmPr1 and StmPr2 are still predominantly responsible for the observed detachment at 24 h.

We previously reported that S. maltophilia strain K279a supernatants cause the death of A549 cells, resulting in a 30 to 40% drop in viability of the monolayer (21). To investigate whether this Xps-mediated cytotoxicity is also dependent on StmPr1 and StmPr2, we evaluated A549 cell viability after a 24-h incubation with supernatants from K279a WT, xpsF mutant, and protease mutant strains. Supernatant from the stmPr1 mutant or the stmPr1 stmPr2 mutant, but not the stmPr2 mutant, caused significantly less cell death than did WT supernatant (Fig. 6C). Indeed, the stmPr1 and stmPr1 stmPr2 mutants were as defective for cytotoxicity as was the xpsF mutant (Fig. 6C). The ability of the stmPr1 mutant to cause cell death was rescued by providing stmPr1 in trans, and a WT phenotype was restored to the stmPr1 stmPr2 mutant with either stmPr1 or stmPr2 (Fig. 6C). Altogether, these data demonstrated that the previously reported Xps-mediated cytotoxic effect on A549 cells can be attributed to StmPr1 and StmPr2, with StmPr1 playing a more predominant role, perhaps due to its greater abundance in the supernatant.

S. maltophilia induces IL-8 secretion by A549 cells followed by protease-mediated degradation of IL-8.

In order to begin to investigate the effects of Xps, StmPr1, and StmPr2 on the innate immune response, we incubated A549 cells with strain K279a at an initial MOI of 1 for 24 h and then monitored IL-8 levels in the culture supernatant by ELISA. IL-8 is a proinflammatory cytokine and neutrophil chemoattractant, and elevated levels of the murine homologues of IL-8, MIP-2 and GROα/KC occur in the lungs of mice infected with S. maltophilia (14). IL-8 was detected in the supernatant of WT-infected cells at 6 and 8 h postinfection; however, by 24 h postinfection, IL-8 was no longer detected (Fig. 7A). Supernatant from monolayers infected with the xpsF mutant had IL-8 levels comparable to WT-infected monolayers at 6 and 8 h postinfection. However, at 24 h, the IL-8 levels were still quite elevated and significantly higher than supernatant from WT-infected cells (Fig. 7A). The ability to decrease IL-8 levels at 24 h postinfection could be restored to the xpsF mutant through trans-complementation with xpsF (Fig. 7A). These results indicated that Xps does not contribute to the induction of IL-8 secretion by A549 cells but is responsible for the observed decrease in IL-8 levels at 24 h. We next incubated A549 cells with the protease mutants and evaluated IL-8 levels at 24 h to see if StmPr1 and/or StmPr2 is responsible for the Xps-mediated decrease in IL-8 levels. The stmPr1 mutant, but not its complement, was attenuated in its ability to cause a decrease in IL-8 levels, whereas the stmPr2 mutant behaved like the WT did (Fig. 7B). The stmPr1 stmPr2 mutant was also impaired, and its defect was complemented with stmPr1 but not stmPr2 (Fig. 7B). Thus, StmPr1, but not StmPr2, contributes to the decrease in IL-8 levels in the coculture medium. However, it is important to note that the relative levels of StmPr1 and StmPr2 produced during coculture are not known. StmPr1 is completely responsible for the Xps-mediated decrease in IL-8, as no significant differences in IL-8 levels were detected between the supernatant of cells infected the xpsF mutant, the stmPr1 mutant, and the stmPr1 stmPr2 mutant (Fig. 7B).

FIG 7.

Effect of Xps, StmPr1, and StmPr2 on strain K279a-induced secretion of IL-8 by epithelial cells. (A) A549 cells were cocultured with the WT, the xpsF mutant, and the complemented xpsF mutant, and then IL-8 levels in the infected-cell supernatants were quantified at the indicated time points by ELISA. (B) A549 cells were cocultured with the WT, the xpsF mutant strain, and the indicated protease mutants and their complements for 24 h, and secreted IL-8 was quantified by ELISA. (C) Recombinant IL-8 was incubated at 37°C for 16 h with various doses of supernatants from WT, and the indicated mutants and their complements. IL-8 levels were quantified by ELISA. IL-8 incubated with medium alone or 0.2 μg of trypsin was included for comparison. Single and double asterisks indicate statistically significant differences from the WT and the respective mutant, and the phi (Φ) indicates a statistically significant difference from the xpsF strain (P < 0.05). For panels A to C, data are the means and SEM from three independent experiments.

One possible explanation for the loss of IL-8 in the infected cell supernatant is protease-mediated degradation of IL-8. To directly test this theory, we incubated S. maltophilia strain K279a supernatants with purified recombinant IL-8 and evaluated the stability of IL-8 via ELISA. There was a dose-dependent degradation of IL-8 by WT supernatants (Fig. 7C). The degradation of IL-8 was dependent on Xps T2S, as the xpsF mutant supernatant did not cause any degradation of IL-8, and degradation of IL-8 by the xpsF mutant supernatant was rescued with xpsF (Fig. 7C). The stmPr1 mutant, but not its complement, was impaired in the ability to degrade IL-8 when 50 μl of supernatant was incubated with the chemokine (Fig. 7C). No impairment of IL-8 degradation was observed for the stmPr2 mutant. However, supernatant from the stmPr1 stmPr2 double mutant was more impaired than stmPr1 mutant supernatant at the 100-μl dose of supernatant (Fig. 7C). Additionally, IL-8 degradation by supernatant from the stmPr1 stmPr2 mutant was restored with either stmPr1 or stmPr2 (Fig. 7C). Together, these data demonstrate that StmPr2 does contribute to IL-8 degradation, but StmPr2-mediated IL-8 degradation was observed only when the protease was in high abundance. Therefore, it is possible that StmPr2 did not contribute to IL-8 degradation during coculture with A549 cells because the protease was not produced at high levels under coculture conditions (Fig. 7B). No difference was observed between the xpsF mutant supernatant and the stmPr1 stmPr2 supernatant at the 10 μl and 50 μl doses, but the supernatant from the stmPr1 stmPr2 mutant did cause significantly more degradation than did the xpsF mutant supernatant when 100 μl of supernatant was incubated with IL-8 (Fig. 7C). Thus, indicating that, in addition to StmPr1 and StmPr2, other Xps T2S substrates contribute to strain K279a-mediated degradation of IL-8. From these experiments, we can conclude that S. maltophilia strain K279a induces Xps-independent IL-8 secretion by A549 cells and that StmPr1 and StmPr2 both can degrade IL-8, but StmPr1 is primarily responsible for the observed Xps-mediated degradation.

DISCUSSION

Here, we demonstrate that Xps T2S-dependent rounding and detachment occurs in four cell types, including epithelial cells of murine and human origin and murine fibroblasts. We found that in A549 cells, cell rounding and detachment are dependent on serine protease activity, and we subsequently identified the serine proteases StmPr1 and StmPr2 as the Xps T2S substrates predominantly responsible for the effects on A549 cells. The repertoire of T2S substrates in Gram-negative bacteria frequently includes proteases, and T2S proteases are produced by a number of important plant (i.e., Xanthomonas spp.) and human (i.e., Legionella pneumophila, Pseudomonas aeruginosa, Vibrio spp., and enterohemorrhagic Escherichia coli) pathogens (22). Among these pathogens, serine proteases have been identified as T2S substrates in Xanthomonas campestris and Vibrio cholerae (41–43). BLASTP analysis revealed that StmPr1 produced by strain K279a is most homologous with an annotated serine protease of Pseudomonas geniculata (69% identity), followed by annotated serine proteases of Xanthomonas sacchari (62% identity) and X. campestris (57% identity). StmPr2 is also most related to an annotated serine protease of P. geniculata (96% identity), followed by annotated serine proteases of Xanthomonas hortorum and X. campestris (77% identity). StmPr1 and StmPr2 are the first documented Xps T2S substrates of S. maltophilia. However, there are likely yet-unidentified Xps substrates, as we previously reported that at least seven proteins are secreted in an Xps-dependent manner (21). Here, we also demonstrate that Xps T2S-dependent proteolytic activity contributes to the degradation of ECM proteins and IL-8, but it is still unclear whether the Gsp T2S is functional in strain K279a, as Gsp mutants did not lack any of these newly described activities.

In this study, we also investigated the relative contribution of StmPr1 and StmPr2 to the proteolytic activities exhibited by S. maltophilia strain K279a supernatant using a genetic analysis. This approach allowed us to determine that StmPr1 and StmPr2 are equally responsible for the Xps-mediated hydrolysis of gelatin and azocasein but that Xps-mediated serine protease activity, which was assayed using N-succinyl-Ala-Ala-Pro-Phe-pNa as a substrate, is almost entirely dependent on StmPr1. The higher abundance of mature StmPr1 compared to StmPr2 in the supernatant may account for this observation, but another possibility is that StmPr1 may have a higher affinity than StmPr2 for this substrate. StmPr1 also played a greater role than did StmPr2 in promoting morphological effects on A459 cells, including rounding, detachment, and actin rearrangement, as well as cell death. In support of our findings, StmPr1 purified from a S. maltophilia bronchial isolate hydrolyzed azocasein and N-succinyl-Ala-Ala-Pro-Phe-pNa and was sufficient to cause rounding of human fibroblasts (32). Also in line with our data, StmPr2 from the clinical isolate 19580, when purified from recombinant E. coli, exhibited caseinolytic activity (34).

StmPr1 and StmPr2 belong to the MEROPS peptidase family S8, also known as the subtilase family, which is the second largest family of serine proteases (44). S8 peptidases are synthesized as proenzymes that contain a propeptide domain, also known as an intramolecular chaperone (IMC) (45). From structural and biochemical studies performed with subtilisins from Bacillus subtilis, it has been determined that maturation of this family of proteases is dependent on three stages. First, the IMC mediates proper folding of the mature protease domains (46). Next, autoprocessing of the bond between the IMC and the mature protease domains occurs, yielding an IMC protease-inhibited complex. During this stage, the IMC remains associated with the protease and serves as a temporary inhibitor of protease activity (47). Finally, through a second autoproteolytic event, the IMC is released and degraded, which results in an active, mature protease (48–50). We detected both the proenzyme and mature forms of StmPr1 and StmPr2 in strain K279a supernatant. Interestingly, StmPr1 was predominantly present in the mature form, and StmPr2 was predominantly in the proenzyme form. An explanation for this observation includes the possibility that activation of StmPr2 through proteolytic cleavage occurs more slowly or is less efficient than that of StmPr1, perhaps due to the requirement of an additional secreted factor that promotes complete processing of StmPr2. Alternatively, the stmPr2 gene is annotated as having an I9 inhibitory element in the propeptide region, which is absent from the stmPr1 propeptide. In the region of the propeptide where the I9 element is predicted (residues 44 to 136 of StmPr2), there is also only 37.8% amino acid sequence identity between StmPr1 and StmPr2 (see Fig. S1 in the supplemental material). Thus, activation of StmPr1 and StmPr2 may be differentially regulated by their IMCs due to the presence or absence of the I9 inhibitory element. Future studies directly comparing the properties of purified StmPr1 and StmPr2 will help elucidate the relative potencies of StmPr1 and StmPr2 as well as determining whether StmPr1 and StmPr2 have distinct targets and functions or are truly redundant proteases.

We have found that both StmPr1 and StmPr2 contribute to the Xps-mediated degradation of the ECM components collagen, fibrinogen, and fibronectin, although StmPr1 is predominantly responsible for the degradation of ECM proteins by strain K279a. Collagen and fibronectin are major structural components of the ECM, which provides structure and elasticity to host tissues (51). Fibrinogen is primarily known for its role as a clotting factor during tissue injury (52); however, fibrinogen secreted by lung epithelial cells assembles into ECM fibrils when fibronectin is present (53). Collagenolytic activity has been reported for only a small subset of S8 peptidases primarily produced by environmental isolates (36–40). The properties of bacterial S8 peptidases have been extensively studied for their use in commercial applications, but the pathogenic potential of bacterial S8 peptidases is relatively unexplored (54). While numerous pathogens hijack host proteins to degrade ECM components, few bacterial proteases are known to be capable of directly degrading ECM proteins (55). StmPr1 was the first S8 peptidase produced by a respiratory pathogen shown to degrade collagen, fibrinogen, and fibronectin (32), and here we demonstrate that StmPr2 also contributes to these activities. Identification of ECM proteins as targets of StmPr1 and StmPr2 suggests that the rounding and detachment of A549 cells may be due at least in part to protease-mediated breakdown of cellular connections with ECM proteins. However, the possibility for intracellular targets, and additional extracellular targets remains. The ability of StmPr1 and StmPr2 to degrade ECM proteins and cause rounding of human lung cells may have significant clinical relevance, as S. maltophilia-induced lung tissue destruction has been reported in immunocompromised patients that succumbed to S. maltophilia infection (56).

Our data demonstrate that S. maltophilia strain K279a induces secretion of the proinflammatory chemokine IL-8 by the human lung cell line A549 in a time-dependent manner, with a decrease in IL-8 levels by 24 h postinfection. S. maltophilia-induced IL-8 secretion has also been observed in the human airway epithelial cell line 1HAEo- (15). These data have important clinical implications, as the immune response to S. maltophilia in cystic fibrosis patients is associated with decreased lung function (57). The decrease in IL-8 secretion by A549 cells observed at 24 h postinfection was due to StmPr1-mediated degradation. Similar results were reported for the MucD serine protease secreted by P. aeruginosa, which degraded IL-8 secreted by a human intestinal epithelial cell line during infection (58). In a murine model of S. maltophilia pneumonia, the IL-8 homologues GROα/KC and MIP-2 are also upregulated in the murine lung in a time-dependent manner (14). The murine immune response was characterized by an early and robust increase in these chemokines at 1 day postinfection followed by a decline at three days postinfection (14). It is possible that StmPr1, and potentially to a lesser extent StmPr2, may contribute to the observed decrease in GROα/KC and MIP-2 levels in the murine lung. Thus, investigating if and when StmPr1 and StmPr2 are produced in the lung will be important in understanding the S. maltophilia murine model of pneumonia.

Although StmPr1 and StmPr2 are completely responsible for the detrimental effects on A549 cells after a 3-h incubation with supernatants, we found that StmPr1 and StmPr2 did not entirely account for effects on A549 cells treated with supernatant for 24 h. These data indicate that an additional Xps T2S substrate(s) contributes to the morphological effects on A549 cells. Similarly, our findings suggest that additional proteases contribute to the degradation of ECM proteins and IL-8 by strain K279a. BLAST analysis revealed that strain K279a carries three additional S8 peptidase-encoding genes (SMLT_RS20900, SMLT_RS16790, and SMLT_RS19730) with homology to stmPr1 and stmPr2. While most of the proteolytic activity exhibited by strain K279a is dependent on Xps T2S, we did observe that the degradation of the fibrinogen α chain occurs in the absence of Xps T2S as well as Gsp T2S, indicating that additional secretion systems are important for fibrinogen α chain degradation.

A recent study examining the presence of predicted virulence genes in the genomes of both clinical and environmental S. maltophilia isolates found that stmPr1 is present in the newly sequenced genomes of strains RA8 and SKK35 in addition to strains K279a and R511-3 (20). Extending this analysis, we found that stmPr1 is present in 21 of the 28 current S. maltophilia genome assemblies. Because the 7 genomes that did not appear in the BLAST results are draft genomes, we cannot yet draw conclusions about the presence or absence of stmPr1 in these genomes. Our analysis revealed that two stmPr1 allelic variants are represented among sequenced strains, with 7 strains carrying the K279a variant (19) and 14 strains carrying a variant most similar to stmPr1 from the previously studied bronchial isolate (32) (see Fig. S3A in the supplemental material). Importantly, despite the approximately 30% difference in amino acid identity between the stmPr1 allelic variants, synteny is conserved in the stmPr1 locus in all complete S. maltophilia genomes. Additionally, we found that the stmPr2 gene is carried by 20 of the 28 S. maltophilia genome assemblies, with the same caveat that the 8 genomes missing from the BLAST results are draft genomes. In contrast to the allelic variation in StmPr1 sequences, StmPr2 identities among sequenced strains range from 94 to 100% when compared to strain K279a StmPr2 (see Fig. S3B in the supplemental material). Looking beyond the sequenced genomes, the presence of both stmPr1 and stmPr2 was confirmed by PCR analysis in the majority of cystic fibrosis clinical isolates, and strain K279a stmPr1 is the most common allelic variant of stmPr1 among these isolates (18). The conservation of stmPr1 and stmPr2 among clinical isolates, together with our experimental data, suggests roles for StmPr1 and StmPr2 in the virulence of S. maltophilia.