Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Pseudomonas aeruginosa is an opportunistic pathogen that can cause life-threatening infections in the immunocompromised patient. P. aeruginosa is particularly dangerous in burn wounds, in respiratory diseases including cystic fibrosis, and during cancer chemotherapy (1). It is also one of the most common pathogens in nosocommial pneumonia, and is frequently isolated in intensive care units. The incidence of infection caused by drug-resistant strains is increasing (2, 3).

Bacterial adherence to host tissue is the earliest and most critical stage of the infectious process (4). The major virulence-associated P. aeruginosa adhesin is the type-IV pilus, with other adhesins (e.g., flagella) also involved (5). Evidence also implicates lipopolysaccharide (LPS) as an adhesin (6). Several nonpiliated P. aeruginosa strains express multiple adhesins with an affinity for human respiratory mucins and/or lactoferrin (7).

Epithelial integrity plays a central role in mediating P. aeruginosa adherence. The basolateral surface contains more P. aeruginosa receptors (asialoGM1 glycolipids) than the apical surface. Disruption of tight junctions in these polarized epithelial cells allows access to asialoGM1, resulting in increased typeIV pili-mediated adherence (8, 9). A recently described type III secretion locus in P. aeruginosa produces factors that are cytotoxic to epithelial cells and macrophages (reviewed by Frank [10]). Thus, an additional mechanism of P. aeruginosa adherence may be the killing of select cells by type III exotoxins in polarized epithelium, allowing access to their basolateral surfaces (9, 11).

P. aeruginosa also produces a viscous exopolysaccharide (EPS; alginate) in the upper airways of patients with cystic fibrosis (CF). Although other virulence factors (e.g., toxins, hemolysins, and proteases) are involved (12), alginate is correlated with the poor prognosis among patients with CF (1, 13). Conversion of P. aeruginosa to the mucoid phenotype (alginate hypersecretion) is common among patients with CF who have chronic respiratory infection (14). Alginate has been implicated as an important adhesin for mucoid P. aeruginosa strains (15), and assists in binding to a variety of human epithelial cells (16). Alginate is also a major component of the intercellular matrix in bacterial biofilms, which confer resistance to phagocytosis and to antibiotics (17, 18). Indeed, the majority of human infections involve biofilms (19), and the presence of P. aeruginosa biofilms in the lungs of CF patients has recently been demonstrated (20). Biofilms are also intractable. Antibiotic concentrations needed to eradicate biofilm bacteria were 50 to 5,000 times higher than that needed to kill planktonic bacteria (21, 22). Together with LPS, these exopolymers may confer resistance to complement-mediated serum killing.

Bismuth-containing remedies have been in regular use for centuries. Recently, bismuth antibacterial activity has been enhanced up to 1,000-fold by certain lipophilic thiol agents, thereby significantly enhancing its potency and versatility as an antibacterial agent (23). Inorganic bismuth (e.g., bismuth nitrate or bismuth subsalicylate) is marginally active, requiring high concentrations (250 µM or 50 µg/ml) to affect bacterial outer membrane protein profiles, or to inhibit EPS expression in Klebsiella pneumoniae (24). In contrast, subinhibitory levels (3-5 µM or  1 µg/ml Bi³⁺) of bismuth dimercaprol (BisBAL) inhibited K. pneumoniae EPS by > 90% (25). Bismuth-ethanedithiol (BisEDT) showed similar subinhibitory activity at 1 µM Bi³⁺.

Although growth-inhibitory bismuth-thiol (BT) concentrations are also potentially toxic to mammalian cells, subinhibitory concentrations are better tolerated and retain potent antibacterial activities. In this study, we examine subinhibitory BT effects on several P. aeruginosa virulence determinants. Such information is critical for evaluating the potential efficacy of these compounds as prophylactic and therapeutic agents. Therapeutic interventions that inhibit EPS may impact several host:parasite interactions, including adherence and resistance to host factors. This report examines the myriad in vitro effects of BTs on P. aeruginosa factors associated with virulence.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Bacteria, Cells, and Media

Mucoid Pseudomonas strains were employed for studies involving alginate expression. These include P. aeruginosa FRD1, a mucoid cystic fibrosis isolate; P. syringae pv. syringae FF5 pPSR12, a constitutive alginate producer; and FF5.31, made alg by Tn5 insertion in algL (26). P. aeruginosa FRD1 was grown in M9 liquid medium. P. syringae strains were cultured in mannitol-glutamate defined agar medium (27). Escherichia coli strain O18:K1, and a serum-sensitive isovariant O18-:K1, were employed in serum sensitivity studies.

Three nonmucoid strains of P. aeruginosa, PAO1, PA103, and PA103 exoU::Tn5, were also studied. PAO1 was cultured at 37°C in trypticase soy broth and on trypticase soy agar (TSA) plates (BBL, Cockeysville, MD) in the presence of varying concentrations of BisEDT. A human bronchial epithelial cell line, 16HBE14o, was maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum (FBS) under 95% room air and 5% CO₂. The tissue culture plates were coated with collagen 2 h before seeding. These cells retain differentiated epithelial morphology and functions, including the presence of tight junctions and cilia, and monolayers generate transepithelial resistance (28). Human lung adenocarcinoma A549 cells (ATCC CCL185) were grown and maintained as described (29). These cells retain alveolar type II epithelial cell morphology, including the presence of lamellar bodies and expression of surfactant protein C (29).

BTs

BisBAL was prepared as described (23). BisEDT was prepared in a similar manner by combining Bi(NO₃)₃ (Sigma Chemical Co., St. Louis, MO) with commercially prepared ethanedithiol (Sigma) in propylene glycol (ACS; Sigma) at a 1:1 molar ratio. Hydrochloric acid was added to solubilize and stabilize stock solutions. BisEDT was added directly to growth media at concentrations from 0.5-10 µM (0.1-2 µg/ml Bi³⁺).

Exopolysaccharide Expression

Mucoid P. aeruginosa FRD1 was cultivated in 10 ml of M9 medium with increments of BisBAL for 17 h at 37°C with shaking at 300 rpm in 150-ml flasks. Cultures were diluted with 30 ml of 0.9% saline and centrifuged at 10,000 × g for 30 min at 4°C. Supernatants were decanted and pellets were washed twice with distilled water and dried at 60°C for 17 h in preweighed microfuge tubes. The supernatants were dialyzed exhaustively against distilled water (12 kD cutoff). Alginate was assayed using the carbazole assay for uronic acids (32) and expressed as µg/ml/mg dry weight.

For P. syringae alginate studies, agar medium was impregnated with BisEDT at incremental concentrations, and plates were seeded on the same day at 10⁷ colony-forming units (cfu)/ml and grown for 72 h at 28°C. Bacteria were harvested using a cotton swab and resuspended in sterile 0.9% saline. A portion was removed for viable bacterial counts and total protein (33). Alginate was extracted as described for K. pneumoniae EPS extraction (34). Briefly, EPS was extracted from whole bacterial cultures with citrated Zwittergent (Calbiochem, LaJolla, CA), and assayed for uronic acid content (35). P. syringae alginate was expressed as uronic acid/mg protein. Bacteria were enumerated by standard agar plating.

LPS Extraction

LPS was isolated using a microextraction method as described previously (36). Briefly, strain PAO1 was grown on TSA plates containing varying BisEDT concentrations at 37°C for 24 h. Bacteria were harvested with sterile cotton swabs, resuspended and washed with phosphate-buffered saline (PBS), and adjusted to a turbidity of 0.4 at 650 nm. A 1.5-ml aliquot of the suspension was centrifuged for 10 min in a microcentrifuge at 14,000 rpm. The pellet was solubilized in 50 µl of lysing buffer (2% sodium dodecyl sulfate, 4% 2-mercaptoethanol, 10% glycerol, 1 M Tris, pH 6.8, and bromphenol blue). Heat denaturation (100°C for 10 min) was followed by proteinase K digestion (60°C for 60 min). Samples were loaded on a 12% polyacrylamide denaturing gel. For visualization of the LPS, gels were fixed with 0.9% periodic acid, 40% ethanol, and 5% acetic acid overnight, then stained with a silver stain kit according to the manufacturer's (Geno Technology, Inc., St. Louis, MO) instructions.

Extracellular Proteins

Secreted proteins from PA103 were isolated as described (37). Briefly, PA103 and PA103 exoU::Tn5 were grown with vigorous shaking in 5 ml MINS medium for 17 h. Bacterial number was normalized by OD₆₀₀ and confirmed by standard agar plating. The sample was centrifuged at 6,000 × g at 4°C for 20 min. Supernatant protein was precipitated with ammonium sulfate (55% final concentration) on ice for 2 h. Samples were centrifuged at 13,000 × g at 4°C for 20 min and resuspended in 500 µl Laemmli buffer. Twenty microliters of the lysate were loaded on a 10% Tris-HCl Ready-Gel (Bio-Rad, Hercules, CA). Similar results were obtained when loading was normalized to total protein, as determined by the BioRad protein assay.

Adherence Assay

Adherence to epithelial cells was assessed using a modified radiolabeling assay as described (38). Briefly, A549 (2.5 × 10⁵) or 16HBE14o (4.5 × 10⁵) cells were seeded onto 12-well microtiter plates and incubated 17 to 48 h until confluent monolayers were formed. Radiolabeled bacteria were layered onto cells at 100 bacteria/epithelial cell and incubated for 1 h at 37°C. Cells were washed with PBS to remove nonadherent bacteria and lysed with 1 N NaOH. Lysates were counted in a liquid scintillation counter. Adherence was calculated as the ratio of lysate:input radioactivity, normalized to the mean of controls and expressed as a percentage. All samples were performed in triplicate and experiments repeated at least twice.

Electron Microscopy

Pelleted bacterial cells were fixed in 2.5% (vol/vol) glutaraldehyde in 0.1 M sodium cacodylate buffer (pH 7.2) for 24 h at room temperature, post-fixed in buffered 1% (wt/vol) osmium tetroxide, dehydrated in graded ethanol, and embedded in LX112 resin (Ladd Corp., Burlington, VT). Thin sections were stained with uranyl acetate and lead citrate and examined with a Zeiss EM 10 (Zeiss, Thornwood, NY) or a Philips EM300 transmission electron microscope (Philips, Acht, The Netherlands), both operating at 60 kV under standard conditions with anticontaminators in place.

Bacteria were pelleted at 5,000 × g for 3 min and prepared for freeze fracture by fixation in 2.5% glutaraldehyde in 0.1 M sodium cacodylate buffer (pH 7.4) for 1 h at room temperature. Samples were infiltrated in 25-27% (vol/vol) glycerol in 0.1 M sodium cacodylate buffer, frozen in liquid-cooled Freon 22, fractured in a double replica device at -115°C in a Balzer's freeze-etch unit (a modified BAF 301, fitted with a 550 turbo pump) (Balzer's, Hudson, NJ), and platinum-shadowed and carbon-coated via electron beam guns. Replicas were cleaned with 0.5% sodium hypopchlorite, mounted on copper grids, and examined with a Zeiss EM 10 transmission electron microscope.

Minimal Inhibitory Concentration, Minimal Bactericidal Concentration, and Cytotoxicity Determination

Minimal inhibitory concentration (MIC) and minimal bactericidal concentration (MBC) of bismuth-thiols were established in broth cultures in accordance with NCCLS standards, with some variation. MIC was expressed as the dose inhibiting growth for 24 ± 2 h. The MBC was the dose conferring a 99.9% reduction in viability, as determined by standard agar plating. The subinhibitory concentration was defined as the dose at 30-50% of the MIC that affected various structural or enzymatic processes.

Confluent cultures of A549 and 16HBE14o cells were exposed to various concentrations of BisEDT and vector-only control (propylene glycol) for 24 h. Cytotoxicity was determined by assessing mitochondrial dehyrogenase activity with the MTT (3-[4,5-Dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide Thiazolyl blue) based assay per the manufacturer's instructions (Sigma). MTT activity was assayed on serial dilutions of each cell line to ensure linearity of the assay. All samples were performed in triplicate and values were expressed as percent of activity as related to untreated controls.

Serum Sensitivity

To ascertain the BT effect on serum sensitivity, P. aeruginosa PAO1, E. coli O18K1, and isogenic mutant O18^-:K1 were incubated on agar plates impregnated with subMIC BisEDT (0-10 µM; MIC, 12-15 µM) for 24 h at 37°C, harvested and washed twice in minimal essential medium (MEM), adjusted to 10⁷ cfu/ml, mixed in a total of 90% normal human serum for 1 h at 37°C, and plated on agar media to assess viability. In some trials, serum was heated at 56°C for 30 min to inactivate complement components.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

MIC and MBC of BT

The MIC and MBC of two BTs, BisEDT and BisBAL, were determined for strain PAO1 in broth culture. The MIC and MBC for BisEDT, at a 1:1 molar ratio of Bi³⁺:thiol, were 2.8 and 5.3 µM, respectively. The molar ratio of Bi³⁺:thiol did not significantly affect the efficacy of BisEDT (23). The MIC and MBC for BisBAL at a 2:1 molar ratio (23) were 15.5 and 28.8 µM, respectively. Subinhibitory concentrations were adjusted to 30-50% of the MIC, as reported previously (25).

Exopolysaccharide Inhibition

The effect of BTs on alginate expression was examined in two separate systems. BisBAL at a fraction of the growth-inhibitory concentration was shown to inhibit alginate production from a mucoid P. aeruginosa isolate (Figure 1). At 2 µM BisBAL, alginate per mg dry weight was reduced to 76% of control values. At 4 µM, alginate was at 33% of control levels. Lag times increased progressively with incremental BisBAL concentrations, with an increase of 3 h beyond untreated control cultures at 4 µM BisBAL. However, no effect on bacterial viability was detectable at 17 h, as assessed by standard agar plate counts (data not shown).

View larger version (14K): [in this window] [in a new window]   Figure 1.   BisEDT reduces EPS expression. Mucoid P. aeruginosa FRD1 (black boxes) were cultivated M9 medium with incremental amounts of BisBAL for 17 h at 37°C. Alginate was assayed using the carbazole assay and expressed as µg/ml/mg dry weight. P. syringae was cultivated in mannitol-glutamate defined agar medium impregnated with BisEDT for 72 h at 28°C. Alginate was extracted by the Zwittergent/citrate method, and determined by uronic acid content, expressed as uronic acid per mg protein for mucoid strain FF5 (black circles) and alginate minus mutant FF5.31 (black triangles).

Alginate expression was also inhibited in P. syringae FF5 pPSR12, a constitutive alginate producer, using BisEDT as agent (Figure 1). The response to BisEDT was bimodal. Relative to untreated controls, there was a sharp increase (269%) in alginate expression per mg protein at 0.2 µM BisEDT, followed by a 62% decrease at 0.6 µM. At 1 µM BisEDT, alginate per mg protein was reduced to less than 8% of control, though growth inhibition on the plates was also substantial. P. syringae FF5.31 (algL::Tn5), produced no measurable alginate (Figure 1). The difference between P. syringae on agar plates with and without BisEDT was obvious. Untreated bacteria were highly viscous and runny, whereas treated bacteria produced a dry sheen as a lawn on agar plates.

LPS Inhibition

To determine if BisEDT affects LPS expression, strain PAO1 was cultured on plates impregnated with varying concentrations of BisEDT. LPS was extracted and analyzed on sodium dodceylsulfate-polyacrylamide gel electrophoresis (SDS-PAGE). Because BT activity is antagonized by the presence of components in enriched agar medium (data not shown), somewhat higher BisEDT concentrations were used. As shown in Figure 2, BisEDT reduced the quantity of LPS per mg protein in a concentration-dependent manner on SDS-PAGE. Both high and low molecular weight LPS, as well as lipid A, were affected.

View larger version (79K): [in this window] [in a new window]   Figure 2.   BisEDT reduces LPS. LPS was extracted and purified from P. aeruginosa PAO1 cultured on TSA plates containing BisEDT at 0, 2, 4, and 6 µM, loaded on a 10% SDS-PAGE gel, and silver stained.

Ultrastructural Alterations

To determine the effects of BisEDT on P. aeruginosa ultrastructure, PAO1 was cultured in the presence of subMIC (1.5 µM) BisEDT and prepared for transmission electron microscopy. Freeze-fracture of P. aeruginosa revealed an array of "blemishes" (see arrows) on the outer surface of BisEDT-treated bacteria (Figure 3B), when compared with untreated bacteria (Figure 3A). The blemishes indicate loose attachment and blebbing of outer membrane. Thin sections confirmed membrane blebbing (Figure 3F) and revealed an aggregation of electron-dense material at the periphery of the protoplast (Figures 3D and 3F) when compared with controls (Figures 3C and 3E). The cytoplasmic effect is likely due to bismuth deposition and ribosomal aggregation to form deposits close to the plasma membrane. Blebbing occurred in a remarkable fashion, with long ribbons of material extending from bacterial surfaces, which is quite unusual for P. aeruginosa (39). Surface electron-dense deposits were also seen associated with the outer membrane and the blebs, presumably due to BisEDT.

View larger version (138K): [in this window] [in a new window]   Figure 3.   Electron micrographs of P. aeruginosa. PAO1 was cultured overnight in the absence (A, C, and E) or presence (B, D, and F ) of BisEDT (1.5 µM). A and B are freeze-fracture replicas with the arrowheads indicating blemishes seen on the outer surface of BisEDT-treated bacteria. C, D, E, and F are transmission electron micrographs. Note the presence of electron dense material in the periphery (D), and cytoplasmic blebbing indicated by the arrows (F ), of the BisEDT-treated bacteria.

Inhibition of Post-translational Modification

The effect on protein processing was examined with P. aeruginosa type III secretion product ExoU. Strain PA103 expresses ExoU, with its well-characterized cleavage products (37). The secreted products from BisEDT-treated overnight cultures were analyzed by SDS-PAGE (Figure 4). Processed, lower molecular weight forms of ExoU predominated in untreated bacteria. In contrast, ExoU cleavage in the supernatant of BisEDT-treated bacteria was immature, as illustrated by the shift in the protein triplet from the lower to higher molecular weight forms, and less cleavage product. The exoU gene products were defined using a mutant bacteria deficient in ExoU (PA103exoU:: Tn5, Figure 4). This profile was obtained when the samples were normalized to bacterial numbers, suggesting that BisEDT did not overtly affect secretion.

View larger version (56K): [in this window] [in a new window]   Figure 4.   BisEDT alters processing of type III secretion products. PA103 was grown in MINS media to induce secretion of type III products in the presence of varying concentrations of BisEDT. The supernatants were separated on a 10% denaturing polyacrylamide gel and loading was normalized to the number of viable bacteria. Arrows indicate exoU gene products (58). Similar results were obtained when equal protein concentrations were loaded (data not shown).

Cytotoxicity

The toxicity on tissue culture cells was assessed using an MTT assay for two transformed cell types, a bronchiolar cell line (16HBE14o) and an alveolar cell line (A549), after a 24-h exposure to BisEDT. The MTT assay measures mitochondrial dehydrogenase activity spectrophotometrically (40). The 16HBE14o gave an LD₅₀ of 4.8 µM, whereas transformed A549 cells were more resistant at 13.3 µM BisEDT (Figure 5). There was virtually no effect on viability in either cell culture at 2 µM BisEDT. Cultures were rendered nonviable between 10 and 20 µM BisEDT.

View larger version (21K): [in this window] [in a new window]   Figure 5.   BisEDT LD50 determination on transformed bronchiolar and alveolar epithelial cells. 16HBE14o (black boxes) and A549 (black circles) cells were exposed to varying concentrations of BisEDT for 24 h and viability assessed using the MTT assay.

Reduced Adherence in Tissue Culture

The effect on adherence of subMIC-treated bacteria to epithelial cells was assessed employing a radiolabel adherence assay. PAO1 was cultured overnight in the presence of varying BisEDT concentrations and percent of adherent bacteria was calculated. Adherence of untreated PAO1 to a confluent monolayer of 16HBE14o and collagen matrix was at 24% and 14%, respectively. BisEDT reduced adherence up to 28% to cells (Figure 6A) and as much as 53% to matrix in a concentration-dependent manner (Figure 6B), achieving statistical significance (P < 0.05, analysis of variance) at 0.5 µM (100 ppb Bi³⁺) and maximal inhibition to epithelial cells at 1.5 µM.

View larger version (17K): [in this window] [in a new window]   Figure 6.   Culturing P. aeruginosa in subMIC BisEDT inhibits adherence. (A and B) PAO1 was cultured overnight in the presence of varying concentrations of BisEDT. Adherence to 16HBE14o cells (A) and extracellular matrix (B) was assessed by a radiolabel assay. (C) PAO1 was cultured overnight in trypticase soy broth. Bacteria and BisEDT were then added simultaneously to 16HBE14o cells and adherence was assayed. The radioactive lysate was counted and the percent adherence calculated when normalized to adherence using untreated-bacteria controls. Values represent the mean of n = 3 ± SD. *Indicates statistically significant from control (P < 0.05).

To determine the immediate effects of BisEDT on adherence, untreated PAO1 and BisEDT were added simultaneously to epithelial cells. Compared with the control, adherence capability was reduced 52.5% (Figure 6C). There was also a dose-related response in the treatment groups. Pretreatment of epithelial cells with BisEDT had no effect on adherence, indicating that the effect was on bacteria (data not shown). In addition, treatment of bacteria and epithelial cells with BisEDT at 2-8 µM for 1 h did not reduce their viability (data not shown).

Enhanced Serum Sensitivity

To determine if BisEDT rendered P. aeruginosa more sensitive to host defenses, a serum sensitivity assay was performed. An E. coli K1 strain and its LPS-isogenic mutant O18 were included to assess the role of O-antigen. For both P. aeruginosa PAO1 (Figure 7A) and E. coli O18K1 (Figure 7B), increasing BT enhanced serum sensitivity. At the highest BT concentrations in agar media (10 µM), viability was reduced by over two log₁₀ units. An E. coli O18^- mutant was sensitive to serum without BT treatment (Figure 7B), but was further reduced in viability after BT treatment by two log₁₀ units. Heat inactivation of serum abrogated the BisEDT effect, indicating that it was complement-mediated.

View larger version (13K): [in this window] [in a new window]   Figure 7.   BisEDT enhances serum sensitivity of P. aeruginosa and E. coli. P. aeruginosa PAO1 (A), and E. coli O18K1 and O18-K1 (B) were grown on agar medium impregnated with BisEDT. Bacteria were resuspended in MEM (108 cfu/ml) and incubated with normal or heat-inactivated serum for 1 h at 37°C and plated for viability. (A) Viability of PAO1 in normal (black boxes) and heat-inactivated (black circles) serum. (B) Viability of E. coli O18+ in normal serum (black squares) and heat-inactivated serum (black circles), and O18- isovariant in normal serum (black triangles) and heat-inactivated serum (black diamonds). Values represent the mean of n = 3 ± SD.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

BTs affect a number of bacterial factors involved in virulence, including EPS expression, exoproteins, serum resistance, and adherence mediators. Collectively, these are among the primary virulence factors in bacteria. The data indicate that the bacterial surface is altered in several ways at apparently nontoxic BisEDT concentrations. Offending bacteria are disarmed and immune defenses augmented by subinhibitory BisEDT, with impact on complement binding, phagocytosis (25), and serum sensitivity. Biofilm formation may be kept in check with small BT doses administered topically, orally, or aerosolized (41). Given that most infections involve biofilms (19), BTs could impact a variety of disease processes.

Previous studies have demonstrated BT inhibition of EPS in K. pneumoniae (25) and staphylococci (42). The current report documents the effect on P. aeruginosa alginate, and supports the universality of BTs as EPS inhibitors. Bacterial EPS is also a major virulence factor in plant diseases. The plant pathogen, P. syringae, expresses alginate constitutively, and is useful in alginate studies. In contrast, mucoid P. aeruginosa rapidly loses alginate expression in vitro. In both species, alginate expression decreased markedly after BT treatment. The bimodal effect on P. syringae alginate expression (Figure 1) is similar to that seen with copper sulfate on the algD promoter (26). Growth inhibition confounded the data at higher BT levels. Yet clearly, alginate is inhibited at subMIC BT levels. Because alginate is a major virulence factor for P. aeruginosa (19), the prospect of preventing infection with BTs is intriguing.

Obvious changes in the bacterial surface were apparent after BT treatment. Electron micrographs of freeze-fractured bacteria reveal spotty blemishes, suggesting vesicle formation, as observed with other agents that perturb the outer membrane (e.g., subMIC gentamicin, EDTA, and small cationic peptides) (39, 43). These agents induce the release of spherical membrane vesicles containing periplasmic material such as proteases, lipases, and other virulence factors (39, 46). In contrast, BisEDT induced ribbon-like structures, which, like membrane blebs, appear to contain LPS. Because LPS is an important toxin (49), release by BT treatment may cause detriment to the host. Preliminary results indicate that less LPS is found in supernatants of BT-treated cultures. However, it is not replenished on the bacterial surface, which suggests that LPS synthesis is also inhibited. The quantity and biologic activity of released endotoxin during BT treatment are currently under investigation.

LPS is also a major source of metal binding sites in gram-negative bacteria. Langley and Beveridge have shown the precipitation of gold as intracellular elemental crystals in P. aeruginosa (52). Similarly, there is an apparent bismuth deposition, as electron dense material in the cytoplasm and in and around the outer membrane, perhaps due to LPS binding.

Adherence of P. aeruginosa to lung epithelial cells or collagen matrix in tissue culture was reduced significantly by BisEDT. The effect on adherence was on the bacterial rather than on the cellular side. Several factors may contribute to reduced adherence. The more moderate reduction of adherence seen suggests that pili expression is not affected. Furthermore, twitching motility on agar plates, which is mediated by type IV pili, was not affected by BTs (data not shown). Reduced adherence is likely due to the loss of LPS, which also mediates adherence (53, 54). A kink in the LPS mantle illustrated by the appearance of LPS-like ribbons extruding from bacterial surfaces (Figure 3F) subjects bacteria to complement-mediated, serum bactericidal activity after BT treatment. However, BisEDT also enhanced serum sensitivity of a rough LPS mutant of E. coli, suggesting that other factors were involved. Deposition of polycationic bismuth may promote aggregation of surface as well as intracellular materials. The gross release of aggregated surface material may alone explain the alterations in adherence and serum sensitivity. Alternatively, bismuth may affect electrophilic interactions with other surfaces. Grant and coworkers demonstrated that pH and cations affect bacterial adhesion in an in vitro system (38). The clinical relevance of the moderate decrease in adherence to epithelial cells is unclear. Nevertheless, the BT effect is pleiotrophic, and influences a number of surface structures and secretions, including proteins and alginate. Delineating the role of these diverse changes on adherence is currently under investigation in both cell culture and experimental models of infection.

Type III secretion products are involved in the cytotoxicity of various cell types, including epithelial cells (8, 37, 55), with most virulent P. aeruginosa expressing one of two type III secretion-exotoxins, ExoS or ExoU. Recent studies also demonstrate the association of type III secretion with increased mortality in the clinical setting (56, 57). ExoU is cleaved by bacterial proteases, because the cleavage products accumulate in the absence of epithelial cells (37). The results suggest that BisEDT inhibits the protease responsible for the cleavage of native ExoU. Identification of the protease and the effect of cleavage on cytotoxicity and virulence are still unknown. Nevertheless, BisEDT might reduce virulence in part by either mitigating cytotoxicity of ExoU, reducing bacterial protease activity, or decreasing adherence induced by damaged epithelium in response to these factors.

Several factors influenced BT action on bacteria, including culture media and genetic background. The culture medium effect is evident, requiring different BT concentrations to produce desired effects on agar or in liquid medium. The MIC for BisEDT against strain PAO1 in broth or agar is 2.8 versus 13 µM, respectively. Thus, agar medium reduces BT potency 4-fold or more. For strain-to-strain variance, compare BT levels that inhibit P. syringae versus P. aeruginosa alginate. The disparity reflects methodologic differences, because two different BTs were studied under two different conditions. BisEDT is 3-fold more potent than BisBAL against P. aeruginosa (MIC; 2.8 versus 9.7 µM, respectively). P. syringae was also more sensitive to BisEDT than was P. aeruginosa. Regardless of medium or strain, BTs inhibit alginate expression substantially at concentrations well below the MIC, and below the toxic threshold in tissue culture.

That BisEDT works at subinhibitory concentrations is noteworthy. Higher BT concentrations inhibit cell growth and exhibit eukaryotic cell toxicity (23). Given the somewhat narrow therapeutic ratio suggested by the cell culture data, potential applications for BTs rest largely on activity at low concentrations. Nebulization of BisEDT periodically into the lungs of patients with patients may curb or prevent chronic P. aeruginosa infection, provided there are no cumulative toxic effects. The toxicity of these compounds in animal models delivered by various modes of administration is currently being tested in animal models.

In summary, several major virulence factors in P. aeruginosa were mitigated by subinhibitory BT treatment. BTs promote serum sensitivity, phagocytosis, and enhanced complement binding to encapsulated bacteria (25), while reducing adherence to host tissue. Host defenses may be augmented by BTs in P. aeruginosa infections (e.g., cystic fibrosis, burn wounds). Taken together, the reduction in adherence, reduced alginate, LPS blebbing, and processing of secreted exotoxin expression could have a powerful impact on the outcome of infection, by reducing the pathogenicity of the offending organism and leaving them prone to a multitude of host defenses. The therapeutic ratio is small, and the cumulative effects of multiple treatments have not been studied thoroughly, nor has the biologic activity of LPS in supernatants. In addition, BTs work synergistically with other antimicrobial agents (e.g., aminoglycosides; P. Domenico, unpublished observations). Taken together, these data suggest that BTs may have a major impact on infectivity that should be explored further in animal models.