Assessment of Structural Features of the Pseudomonas Siderophore Pyochelin Required for Its Ability to Promote Oxidant-Mediated Endothelial Cell Injury

Archives of Biochemistry and Biophysics Vol. 393, No. 2, September 15, pp. 236 –244, 2001 doi:10.1006/abbi.2001.2517, available online at http://www.idealibrary.com on

Assessment of Structural Features of the Pseudomonas Siderophore Pyochelin Required for Its Ability to Promote Oxidant-Mediated Endothelial Cell Injury Jon J. DeWitte,* Charles D. Cox,† George T. Rasmussen,* and Bradley E. Britigan* ,‡ ,§ ,1 ‡Department of Internal Medicine and Research Service, VA Medical Center—Iowa City and *Department of Internal Medicine, †Department of Microbiology, and §The Free Radical Research Program of the Department of Radiology, University of Iowa College of Medicine, Iowa City, Iowa 52242

Received January 22, 2001, and in revised form July 5, 2001; published online August 24, 2001

We previously showed that iron chelated to the Pseudomonas aeruginosa siderophore pyochelin enhances oxidant-mediated injury to pulmonary artery endothelial cells by catalyzing hydroxyl radical (HO •) formation. Therefore, we examined pyochelin structural/chemical features that may be important in this process. Five pyochelin analogues were examined for (i) capacity to accentuate oxidant-mediated endothelial cell injury, (ii) HO • catalytic ability, (iii) iron transfer to endothelial cells, and (iv) hydrophobicity. All compounds catalyzed similar HO • production, but only the hydrophobic ones containing a thiazolidine ring enhanced cell injury. Transfer of iron to endothelial cells did not correlate with cytotoxicity. Finally, binding of Fe 3ⴙ by pyochelin led to Fe 2ⴙ formation, perhaps explaining how Fe 3ⴙ–pyochelin augments H 2O 2-mediated cell injury via HO • formation. The ability to bind iron in a catalytic form and the molecule’s thiazolidine ring, which increases its hydrophobicity, are key to pyochelin’s cytotoxicity. Reduction of Fe 3ⴙ to Fe 2ⴙ may also be important. © 2001 Academic Press Key Words: Pseudomonas aeruginosa; iron; hydroxyl radical; spin trapping; endothelial cell; lung; hydrogen peroxide; hydrophobicity; free radical.

Pseudomonas aeruginosa is a leading cause of nosocomial pneumonia and also appears to be a contributor to the pathogenesis of chronic lung disease in cystic fibrosis (1). P. aeruginosa infection can lead to exten1 To whom correspondence and reprint requests should be addressed at Department of Internal Medicine, The University of Iowa Hospitals and Clinics, 200 Hawkins Drive, SW 54 GH, Iowa City, IA 52242. Fax: 319-356-4600. E-mail: [email protected].

236

sive tissue damage, characterized histologically by microvascular (endothelial) injury, thrombosis, and a neutrophil-predominant inflammatory response (2, 3). Endothelial cells are susceptible to damage by a variety of reactive oxygen species. Such oxidants, derived from neutrophils and other inflammatory cells, have been implicated in the pathogenesis of various forms of acute and chronic lung injury (2). In response to P. aeruginosa and other bacteria, neutrophils are stimulated to convert O 2 to superoxide anion ( •O 2⫺), which then dismutes to hydrogen peroxide (H 2O 2). This “respiratory burst” is an important defense mechanism against many bacteria (4). Superoxide and H 2O 2 react in the presence of some iron chelates to generate hydroxyl radical (HO •) via the Haber–Weiss reaction (5). In this reaction, iron serves as a catalyst, cycling between the ferric (Fe 3⫹) and ferrous (Fe 2⫹) state, leading to production of HO •, the most cytotoxic of the oxygen free radicals (5, 6). Hydroxyl radical formation has been incriminated as an important contributor to host injury in a wide array of inflammatory states (5, 7, 8). Not all iron chelates are capable of acting as catalysts of the Haber–Weiss reaction (7–11). In nonpathogenic states, practically all extracellular iron is bound to transferrin or lactoferrin. When bound to either of these proteins, iron is noncatalytic (7–11). To grow, pathogenic bacteria must be able to acquire iron from host iron chelates. Similar to the approach of other microbes, P. aeruginosa secretes two siderophores termed pyoverdin and pyochelin (12–15). These siderophores can scavenge iron from some host sources, making it available for use by the organism (12–15). Our laboratory has shown that iron bound to pyochelin, but not pyoverdin, can catalyze the forma0003-9861/01 $35.00 Copyright © 2001 by Academic Press All rights of reproduction in any form reserved.

PYOCHELIN STRUCTURE AND CELL INJURY

tion of HO •/and accentuate •O 2⫺/H 2O 2-mediated endothelial cell injury (16 –18). Similar results were obtained using lung epithelial cells (19). In contrast, other iron chelates, such as Fe–nitrilotriacetic acid (NTA) 2 or Fe–EDTA, which are efficient HO • catalysts, failed to augment endothelial cell injury (17, 18). The reason for the ability of iron bound to pyochelin to enhance oxidant-mediated endothelial cell injury is not known. Due to its highly reactive nature, HO • diffuses only a negligible distance before encountering an oxidizable substrate (20). Therefore, in order to cause cytotoxicity, HO • must be generated in close proximity to key cellular targets. This suggests that the cytotoxic potential of pyochelin could relate to its ability to target redox active iron to critical target sites for injury on and/or within the cell. A mutant strain of P. aeruginosa has been isolated, which is unable to synthesize the pyochelin precursor salicylate (21). By providing this strain with specific salicylate derivatives, several pyochelin analogues have been produced and purified to homogeneity (22). The present communication reports experiments in which these analogues were used to assess structural features of the pyochelin molecule that may be associated with the ability of this compound to promote HO •mediated injury of pulmonary artery endothelial cells. MATERIALS AND METHODS Endothelial cells. Confluent monolayers of porcine pulmonary artery endothelial cells were obtained as previously described (17, 18) by plating 5 ⫻ 10 4 cells per well in 24- or 48-well tissue culture plates (CoStar, Cambridge, MA) and incubating them (37°C, 5% CO 2) for 4 –5 days in 0.5 mL Medium 199 (University of Iowa Cancer Center, Iowa City, IA) with 5–10% heat-inactivated fetal bovine serum, 2⫻ basal medium amino acids, BME vitamins, 2 mM Lglutamine, and 10 units/mL penicillin/streptomycin (Gibco, Grand Island, NY). Cells were used between passages 6 and 9 at 2–3 days postconfluence. Preparation of pyochelin. Pyochelin and its benzene ring-substituted analogues were purified to homogeneity from broth culture of P. aeruginosa strain IA602 [American Type Culture Collection (ATCC), Rockville, MD] which had been provided defined structural variants of salicylate as previously described (22). These molecules along with aeruginoic acid and HTC were extracted using dichloromethane extraction and thin-layer chromatography as previously described (22, 23). The assigned structural names for pyochelin and its analogues are as follows: pyochelin, 2-[2-(2-hydroxyphenyl)-2-thiazolin-4-yl]-3-methyl-4thiazolidinecarboxylic acid; 5-fluoro pyochelin, methyl 2-[2-(5-fluoro2-hydroxyphenyl)-2-thiazolin-4-yl]-3-methyl-4-thiazolidinecarboxylate; 4-methyl pyochelin; methyl 2-[2-(3-hydroxy-4-methylphenyl)-2-thiazolin-4-yl]-3-methyl-4-thiazolidinecarboxylate; 6-AZA pyochelin; methyl 2-[2-(3-hydroxy-2-pyridyl)-2-thiazolin-4-yl]-3-methyl-4-thiazolidinecarboxylate; HTC, 2-(2-hydroxyphenyl)-2-thiazoline-4-carboxylic acid; and aeruginoic acid, 2-(2-hydroxyphenyl)-2-thiazole-4-carboxylic acid.

2 Abbreviations used: NTA, nitrilotriacetic acid; EPR, electron paramagnetic resonance; HBSS, Hanks’ balanced salt solution; HTC, 2-(2-hydroxyphenyl)-2-thiazoline-4-carboxylic acid.

237

Iron-loaded pyochelin and its analogues were generated by adding FeCl 3 (in 100% ethanol) to a solution of apo forms of the compounds at pH 5.0 to achieve 40% iron saturation based on the known stoichiometry of the iron/pyochelin ratio of 1:2 (23). Lowering the ratio of iron:chelator below the theoretic maximum was done to decrease possible presence of “free” iron in the system. Hydrophobicity. This was quantified for pyochelin and each derivative by measuring the octanol/water coefficient. The wavelength of each derivative’s maximum absorbance was determined via scanning ultraviolet–visible wavelength spectroscopy. Then, standard curves of absorbance vs concentration were prepared in water, using 5, 10, 15, and 20 ␮M concentrations of the iron-loaded derivative. Three 3-mL aliquots of 33 ␮M concentrations (initial) of each of the derivatives were prepared in scintillation vials, each with a stirring bar. Three milliliters of 1-octanol was added to each vial and mixed vigorously for 1 min. Every 15 min for 1 h, the vials were again shaken vigorously. The octanol phases were removed and 1-mL samples of the aqueous phases were removed. After measuring the absorbance of the samples, the concentrations of the pyochelin derivatives were determined against the standard curve. The octanol/water coefficient was calculated as the coefficient ⫽ ([pyochelin]initial ⫺ [pyochelin]final)/[pyochelin]final. Hydrophobic compounds are defined as having an octanol/water coefficient of ⬎1.0 and hydrophilic compounds are defined as having an octanol/water coefficient of ⬍1.0. Spin trapping. The ability to act as a HO • catalyst was measured by spin trapping in conjunction with electron paramagnetic resonance (EPR). Reaction mixtures in water of xanthine (100 ␮M, Sigma Chemical Co., St. Louis, MO), xanthine oxidase (0.06 U/mL, Boehringer-Mannheim, Indianapolis, IN), and the desired pyochelin analogue (2 ␮M Fe), along with 170 mM ethanol and 10 mM ␣-(4pyridyl-1-oxide)-N-t-butyl nitrone (4-POBN, Sigma Chemical Co.) were employed. These mixtures were transferred to quartz EPR flat cells (0.5 mL), which were placed in the cavity of the EPR spectrometer [Model E104 A EPR, Varian Associates (Palo Alto, CA) or Bru¨ker ESP300 (Karlsru¨he, Germany)] and resulting EPR spectra obtained. In this spin-trapping system HO • formation is manifested by the EPR detection of the 4-POBN/ •CHOHCH 3 spin adduct whose splitting constants are A N ⫽ 15.5G and A H ⫽ 2.6G (24, 25). Endothelial cell injury. Microtiter plate-adherent pulmonary artery endothelial cells were loaded with 51Cr, by incubation for 18 h (37°C, 5% CO 2) in Medium 199 (The University of Iowa Cancer Center, Iowa City, IA) containing 20 ␮Ci/mL 51Cr (1 mCi/mL sodium chromate, Amersham, Arlington Heights, IL), as previously described (17). Each well was washed three times, and 0.5 mL of Hanks’ balanced salt solution (HBSS) was added to each well except for maximum release controls which contained HBSS with 10% Triton X-100. Iron chelated to pyochelin derivatives (2.5–10 ␮M) with or without xanthine (100 ␮M) was added and preincubated for 30 min after which time xanthine oxidase (0.1 U/mL) or hydrogen peroxide (50 or 100 ␮M) was added. The wells were incubated for 3 h following which the supernatants were removed and radioactivity was determined by ␥-counter. Results of each experiment are expressed as percentage mean specific 51Cr released in triplicate samples where specific 51Cr release was defined as (test well 51Cr cpm ⫺ spontaneous release 51Cr cpm)/(maximum release 51Cr cpm ⫺ spontaneous release 51Cr cpm) ⫻ 100. Iron association with endothelial cells. Association of iron with endothelial cell monolayers after exposure to iron chelated to pyochelin, one of its analogues, or NTA was performed using a modification of a method we previously employed for the study of iron acquisition from low-molecular-weight chelating agents by adherent macrophages (26). Pyochelin, one of its analogues, was loaded with 59FeCl 3 (Amersham) in a manner analogous to nonradioactive iron described above. Porcine pulmonary artery endothelial cell monolayers maintained in 48-well plates were washed and placed in the presence of HBSS to which 7.4 ␮M [ 59Fe] chelated to pyochelin or one of its

238

DEWITTE ET AL.

FIG. 1. Structures of pyochelin and the five analogues employed in this study: 5-fluoro pyochelin, 4-methyl pyochelin, 6-aza pyochelin, HTC, and aeruginoic acid.

analogues was added. The monolayers were incubated for defined time periods (1– 4 h). The monolayers were then repetitively washed three times with phosphate-buffered saline to remove extracellular 59 Fe following which the endothelial cells were detached from the plate using 0.2% trypsin. 59Fe (cpm) associated with the cells was then determined by ␥-counter. In each experiment, the [ 59Fe]chelates were also added to wells which lacked endothelial cells to control for possible nonspecific adherence of the iron chelator complex to the plastic surface. These cpm values, if they exceeded background, were subtracted from the results of cell-containing wells when calculating cell-specific iron acquisition. To provide insight as to whether iron acquired from pyochelin and/or its analogues was predominantly associated with the cell plasma membrane or cytosol, in some cases cells were subjected to subcellular fractionation after the above incubation with the [ 59Fe]chelates. After 120 min of exposure to the [ 59Fe]chelates, the monolayers were washed three times with phosphate-buffered saline (PBS) to remove extracellular 59Fe. The monolayers were then removed from the plates using 0.25% trypsin and transferred to 12 ⫻ 75-mm propylene tubes. The cells were then pelleted (500g, 10 min) and lysed by three repetitive freeze (⫺70°C)/thaw cycles by immersion in liquid N 2. Suspensions of the lysed cells were then centrifuged at 125g for 5 min to pellet unbroken cells. The resulting supernatant was then removed to 2-mL Eppendorf tubes and centrifuged at 16,000g for 30 min to separate the plasma membrane (pellet) from the cytosol (supernatant). The supernatant was removed by gentle aspiration and transferred to a vial for determination of radioactivity by ␥-counter. The pellet was transferred to a separate vial by cutting the pellet-containing bottom of the microfuge tube with a clipper into a vial for similar ␥-counting. Because some cells were lost due to failure to lyse, results were expressed as the percentage of 59Fe counts in the cytosol and supernatant fractions of the lysed cells (i.e., not pelleted at 125g).

Nature of the state of iron (ferric vs ferrous) bound to pyochelin. Formation of ferrous iron (Fe 2⫹) following the addition of ferric iron to a solution of pyochelin was monitored using spectrophotometric detection of the presence of a Fe 2⫹ complex of the ferrous iron chelator bathophenanthrolein sulfate (Sigma). Each pyochelin derivative (40% saturated with FeCl 3) was added to water (pH 7.0) containing bathophenanthrolein (1 mM) to achieve final iron concentrations of 40, 60, and 80 ␮M. After 3 h at room temperature, absorbance of the triplicate samples was measured at 535 nM, which reflects Fe 2⫹– bathophenanthrolein complex formation. Controls using similar iron concentrations of FeCl 3 (ferric iron) and Fe 2SO 4 (ferrous iron) were employed as negative and positive controls for the presence of Fe 2⫹, respectively. Statistical analysis. A paired Student’s t test was used for all statistical analyses. Results were considered significant at P ⬍ 0.05.

RESULTS

Iron-loaded pyochelin catalyzes the formation of HO • and accentuates endothelial cell injury (16, 17). The present study was designed to assess pyochelin features critical to this process. Five analogues of pyochelin have been previously purified from broth culture: 5-fluoro pyochelin, 4-methyl pyochelin, 6-aza pyochelin, HTC, and aeruginoic acid (22). These differ from pyochelin either by the lack of the thiazolidine ring (HTC and aeruginoic acid) or by the presence of additions to the benzene ring of pyochelin (Fig. 1). These pyochelin analogues were used to explore the relationships between pyochelin structure, HO • catalysis, and endothelial cell injury.

PYOCHELIN STRUCTURE AND CELL INJURY

239

of the above pyochelin derivatives to act as an HO • catalyst. Therefore, their relative ability to augment the production of HO • formed during the reaction of xanthine and xanthine oxidase, which generates •O 2⫺ and H 2O 2, was assessed. Hydroxyl radical formation was measured using spin trapping in conjunction with EPR spectrometry. As shown in Figs. 4A and 4B, for the most part each pyochelin derivative was similar to pyochelin in its ability to catalyze the production of HO • (manifested as the magnitude of the 4-POBN/ • CHOHCH 3 spin adduct). Aeruginoic acid appeared slightly inferior in this regard relative to the other agents but still resulted in the generation of considerable HO • (Fig. 4B). These data do not suggest that variability in the extent of HO • formation is responsible for the inability of iron bound to HTC or aeruginoic acid to promote endothelial cell injury. Interaction of Ferripyochelin and Hydrogen Peroxide

FIG. 2. Mean ⫾ SEM (n ⫽ 4 or 8) endothelial cell injury (specific 51 Cr release) resulting from the 3-h exposure of porcine pulmonary artery endothelial cell monolayers to 50 or 100 ␮M H 2O 2 alone and in the presence of iron-loaded pyochelin and its analogues: 5-fluoro pyochelin, 4-methyl pyochelin, 6-aza pyochelin, HTC, and aeruginoic acid (2.5 ␮M). Asterisks denote a statistically significant (paired t test) increase in 51Cr release relative to the H 2O 2 alone-treated control. *P ⬍ 0.0005.

Our current (Fig. 4) and earlier results (17, 18) demonstrating that pyochelin and H 2O 2 react to form HO • and damage endothelial cells pose an additional question and suggested an alternative hypothesis for the inability of HTC and aeruginoic acid to promote cell

Ability of Pyochelin and Derivatives to Augment Oxidant-Mediated Endothelial Cell Injury We first examined the extent to which iron bound to each of the above pyochelin derivatives varied in its ability to promote oxidant-mediated endothelial cell injury. Porcine pulmonary artery endothelial cells that had been preloaded with 51Cr were exposed to H 2O 2 or xanthine/ xanthine oxidase in the absence and presence of pyochelin or one of its analogues. The presence of 4-methyl pyochelin, 6-aza pyochelin, and 5-fluoro pyochelin, as well as pyochelin itself, significantly increased endothelial cell damage, resulting from exposure to either H2O 2 or xanthine/xanthine oxidase (Figs. 2 and 3). There was no statistically significant difference between the magnitude of injury mediated by the four compounds. In contrast to these results, neither iron chelates of HTC nor aeruginoic acid significantly augmented endothelial damage above that resulting from exposure to H2O 2 or xanthine/xanthine oxidase alone (Figs. 2 and 3). Ability of Pyochelin Analogues to Catalyze Hydroxyl Radical Formation The most straightforward explanation for the above results was that there were differences in the capacity

FIG. 3. Mean ⫾ SEM (n ⫽ 4 or 8) of endothelial cell injury (specific 51 Cr release) resulting from the 3-h exposure of porcine pulmonary artery endothelial cell monolayers to 100 ␮M xanthine/0.1 U/mL xanthine oxidase alone and in the presence of iron-loaded pyochelin or one of its analogues: 5-fluoro pyochelin, 4-methyl pyochelin, 6-aza pyochelin, HTC, and aeruginoic acid (10 ␮M). Asterisks denote a statistically significant (paired t test) increase in 51Cr release relative to the xanthine/xanthine oxidase alone-treated control. *P ⬍ 0.001.

240

DEWITTE ET AL.

FIG. 4. (A) EPR spectra resulting from the reactions of xanthine and xanthine oxidase in the presence of 4-POBN and ethanol (X/XO alone) as well as when supplemented with iron-loaded pyochelin (⫹PCH), 5-fluoro- pyochelin (⫹5-F PCH), 4-methyl- pyochelin (⫹4-M PCH), and 6-aza-pyochelin (6-N PCH) providing 2 ␮M Fe. The EPR spectra observed in the presence of iron-loaded pyochelin and its analogues are those of the 4-POBN/ •CHOHCH 3 spin adduct, reflecting HO • formation. Results are representative of three or more separate experiments. (B) EPR spectra resulting from the exposure of iron-loaded (2 ␮M Fe) HTC (top), pyochelin (middle), and aeruginoic acid (bottom), in the presence of 4-POBN and ethanol to the xanthine/xanthine oxidase system. The EPR spectra observed in the presence of iron-loaded pyochelin and its analogues are those of the 4-POBN/ •CHOHCH 3 spin adduct, reflecting HO • formation. Results shown are representative of three or more separate experiments.

injury. Pyochelin has been reported to bind iron in the ferric (3⫹) but not ferrous (2⫹) oxidation state (23). Conversely, ferrous iron is needed in order for HO • to be generated upon reaction with H 2O 2. Thus, it was unclear how the interaction of H 2O 2 with Fe 3⫹-loaded pyochelin resulted in HO •-mediated cell injury. One possible explanation was that endothelial cells reduce ferric to ferrous iron. However, this proved to be an unsatisfying explanation as the mixture of ferri-pyochelin and H 2O 2 in the absence of cells was found to generate HO • (Fig. 5). An alternative possibility was that the process of iron binding to pyochelin resulted in the generation of Fe 2⫹ from Fe 3⫹. Bathophenanthrolein absorbs at 535 nm upon binding Fe 2⫹, allowing detection of the formation of the Fe 2⫹– bathophenanthrolein complex to be utilized as a means of quantifying Fe 2⫹ (16). Pyochelin

was 40% loaded with Fe 3⫹ (FeCl 3) following which it was mixed with bathophenanthrolein and formation of Fe 2⫹ was monitored. As shown in Fig. 6 after 3 h of incubation, a significant portion of the iron initially added as the Fe 3⫹–pyochelin complex became detectable as Fe 2⫹. No increase in absorbance over time was observed with the incubation of FeCl 3 in the absence of pyochelin, indicating that the process was dependent on the presence of pyochelin and not due to a bathophenanthrolein-mediated disruption of the equilibrium between Fe 3⫹ and Fe 2⫹ in solution. Substitution of any of the five pyochelin analogues for pyochelin yielded similar results (data not shown). These data suggest that pyochelin is able to reduce ferric iron to a form that behaves in a fashion akin to ferrous iron. However, the difference between the cytotoxicity of aeruginoic acid and HTC relative to pyochelin and the

PYOCHELIN STRUCTURE AND CELL INJURY

241

FIG. 5. EPR spectra resulting from the reaction of ferripyochelin (10 ␮M) and H 2O 2 (25 ␮M), in the presence of 10 mM 4-POBN and 170 mM ethanol (top). Also shown are results obtained with H 2O 2 alone (bottom). The EPR spectrum observed in the presence of ferripyochelin and H 2O 2 is that of the 4-POBN/ •CHOHCH 3 spin adduct, reflecting HO • formation. Results shown are representative of four separate experiments.

benzene-ring-substituted forms of pyochelin does not appear to be due to differences in the ability of the compounds to generate this reduced form of iron. Relative Hydrophobicity of the Pyochelin Analogues Variations in hydrophobicity would be another mechanism whereby pyochelin and its analogies could differ in their ability to target iron to eukaryotic cells. Increased hydrophobicity would be expected to enhance the ability of the chelator to place reactive iron in close proximity to a cell membrane thereby enhanc-

ing the potential for HO •-mediated damage. Consistent with this possibility, pyochelin, as well as all three of the benzene-ring-substituted pyochelin derivatives which augmented oxidant-mediated endothelial cell injury, was found to be hydrophobic (octanol:water partition coefficient ⬎1.0). Pyochelin exhibited an octanol: water partition coefficient of 2.05. The addition of the F and N to the benzene ring of pyochelin resulted in a slight decrease in pyochelin’s hydrophobicity. The octanol:water partition coefficients for 5-fluoro pyochelin and 6-aza pyochelin were 1.36 and 1.32, respectively. As expected, the addition of the methyl group to pyochelin (4-methyl pyochelin) increased hydrophobicity (octanol/water coefficient ⫽ 3.54). In contrast, HTC and aeruginoic acid exhibited octanol:water coefficients of 0.177 and 0.523, respectively, with values ⬍1.0 indicative of hydrophilic compounds. Acquisition of Iron from Pyochelin and Its Analogues by Endothelial Cells

FIG. 6. Formation of Fe 2⫹– bathophenathrolein, recorded as the absorbance at 535 nm, obtained following the incubation of similar concentrations of ferric iron (FeCl 3), ferricpyochelin (Fe PCH), or ferrous iron (Fe 2SO 4) in the presence of bathophenanthrolein (1 mM) for 3 h at 25°C. The magnitude of absorbance at 535 nm directly reflects the amount of Fe 2⫹ present.

We previously theorized that the hydrophobicity of pyochelin is a key factor that allows it to accentuate endothelial cell damage by effectively placing catalytic iron in close proximity to the cell membrane (17). To further test this hypothesis, we assessed the extent to which 59Fe became tightly associated with endothelial cells following exposure to [ 59Fe]pyochelin compared to [ 59Fe]HTC, [ 59Fe]4-methyl pyochelin, [ 59Fe]6-aza pyochelin, [ 59Fe]5-fluoro pyochelin, or [ 59Fe]aeruginoic acid. As shown in Fig. 7A, increasing amounts of 59Fe became associated with endothelial cell monolayer following exposure to pyochelin. The magnitude of iron which became cell-associated was similar to that of pyochelin when the endothelial cell monolayer was incubated with 59Fe chelates of 4-methyl pyochelin, 6-aza pyochelin, 5-fluoro pyochelin, HTC, or aeruginoic acid (Fig. 7B). Less cell-associated 59Fe was detected

242

DEWITTE ET AL.

of aeruginoic acid to efficiently accomplish this, in spite of its inability to enhance oxidant-mediated injury (Figs. 2 and 3), suggests that simple iron transfer efficiency is not directly linked to cytotoxic potential of the compounds. We also observed no difference in the subcellular distribution (cytosolic vs membrane) of iron acquired from each of the chelates by the endothelial cells (Table I). DISCUSSION

P. aeruginosa infections of the lung and other tissues are notable for their extensive tissue damage in the presence of a neutrophil-dominated inflammatory response (2). We previously showed that damage of pulmonary artery endothelial cells (17, 18) and airway epithelial cells (19) by various sources of •O 2⫺ and H 2O 2 is augmented in the presence of iron bound to the P. aeruginosa siderophore pyochelin. The mechanism responsible appears to involve the ability of the iron– siderophore complex to catalyze the formation of HO •. This capacity of ferripyochelin to augment H 2O 2mediated cytotoxicity has always been somewhat puzzling because the reaction of ferric iron and H 2O 2 should not produce HO •. This dichotomy now appears to be explained by the fact that the process of binding of ferric iron by pyochelin appears to result in the formation of iron which behaves as if it were in the ferrous form. Consistent with this, the reaction of ferripyochelin with H 2O 2 was shown to generate spin adducts which would be formed during HO • production. The source of reducing equivalents within the pyochelin molecule which allows reduction of ferric to ferrous iron is not known at present. Pyochelin reduction of iron may proceed in a manner similar to that mediated by trioglycolate (CD Cox, unpublished obser-

FIG. 7. (A) Endothelial cell-associated iron (mean ⫾ SEM pmol 59 Fe/10 6 cells) over time as a function of exposure of endothelial cell monolayers to 7.4 ␮M 59Fe chelated to pyochelin. (B) Comparison of the magnitude of cell-associated 59Fe following endothelial cell incubation for 120 min with 7.4 ␮M 59Fe chelated to 4-methyl pyochelin (4me), 5-fluoro pyochelin, 6-aza pyochelin, HTC, or aeruginoic acid relative to paired experiments using a similar concentration of 59Fe pyochelin. Results, expressed as percentage of 59Fe acquired from pyochelin (control) in the same experiment, are means ⫹ SEM from three or more separate experiments, each performed in triplicate. Results obtained with HTC were significantly lower than with pyochelin (P ⬍ 0.0001). None of the other four pyochelin analogues yielded values of cell-associated iron that were statistically different from native pyochelin.

following exposure to [ 59Fe]HTC than that observed with the other pyochelin analogues (P ⬍ 0.05) (Fig. 7B). These experiments revealed that the most hydrophilic agent (HTC) exhibited the poorest ability to transfer iron to endothelial cells. However, the ability

TABLE I

Subcellular Distribution of Iron Acquired by Endothelial Cells from Pyochelin or Its Analogues % iron association Source

Membrane fraction

Cytosolic fraction

Pyochelin 5-Fluoro pyochelin 4-Methyl pyochelin 6-Azo pyochelin HTC Aeruginoic acid

92.1 ⫾ 1.8 93.0 ⫾ 1.6 91.6 ⫾ 1.0 89.7 ⫾ 1.0 88.2 ⫾ 1.8 91.5 ⫾ 1.6

7.9 ⫾ 1.8 7.0 ⫾ 1.6 8.4 ⫾ 1.0 10.3 ⫾ 1.0 11.8 ⫾ 1.8 8.5 ⫾ 1.6

Note. Shown is the mean ⫾ SEM percentage (n ⫽ 3) of 59Fe detected in the respective membrane- and cytosol-containing fractions of PPAEC following a 120-min incubation with 59Fe chelates of pyochelin or the pyochelin analogues noted. No significant differences were observed in the subcellular distributions of iron among experiments utilizing pyochelin or its analogues.

PYOCHELIN STRUCTURE AND CELL INJURY

vation) whose structure has some similarities to that of pyochelin. Although the sulfhydryl groups on pyochelin are good candidates for involvement in the process, more extensive study is required. Nevertheless, this capacity of pyochelin to reduce iron may be an integral part of its cytotoxic potential. It is worth noting, however, that our studies do not eliminate the possibility that cellular reductase activity may also contribute to the generation of ferrous iron or that an additional oxidant species resulting from the reaction of H 2O 2 with Fe 3⫹ bound to pyochelin such as a ferryl species (27) could contribute to ferripyochelin-mediated augmentation of H 2O 2-dependent endothelial cell injury. Pyochelin is relatively unique in structure relative to other bacterial siderophores (14). Among its unique properties is its hydrophobicity (14). In the present study, we employed five pyochelin analogues to assess the structural properties of pyochelin that allow it to augment oxidant-mediated endothelial cell injury. When assessed using a 51Cr release assay, the three benzene-ring-substituted analogues of pyochelin each exhibited a similar capacity to accentuate endothelial cell damage relative to the parent compound. In contrast, the two analogues which lacked the thiazolidine ring failed to increase oxidant-mediated cell injury. These data argue strongly that the presence of this thiazolidine ring is critical to the cytotoxic potential of pyochelin. How the loss of the thiazolidine ring leads to a loss of ability to enhance oxidant-mediated endothelial cell injury has not been definitively established. This ring is not required for bound iron to act as a catalyst for HO • formation (Fig. 4B). It is our current hypothesis that a major factor contributing to the loss of cytotoxicity in the absence of the thiazolidine ring is the decrease in the hydrophobicity of the molecule. All pyochelin analogues that were hydrophobic enhanced oxidant-mediated cytotoxicity and resulted in readily detectable acquisition of iron by the endothelial cells. Absence of the thiazolidine ring resulted in a hydrophilic compound and was associated with a lack of ability to enhance oxidant toxicity. However, only HTC differed in its ability to transfer iron to the endothelial cell monolayer, indicating that efficiency of iron transfer to the endothelium is not the sole requirement to enhance oxidant-mediated injury. Furthermore, no difference in cytosolic vs membrane distribution of iron in endothelial cells following incubation with each of the pyochelin analogues was observed, with most of the iron associated with the plasma membrane-containing fraction in each case. This is similar to the pattern we have previously observed with iron acquisition by a variety of cells from other low-molecular-weight chelating agents (Britigan et al., unpublished observations). These subcellular fractionation data must be interpreted with caution however. The cell lysis proce-

243

dure utilized for subcellular fractionation could have altered the physicochemical composition of the plasma membrane, leading to results that do not truly reflect the iron distribution in intact cells. It remains possible that iron transfer to an as yet unidentified key subcellular site is the determining factor in ferripyochelin cytotoxicity. Our data support a general role for hydrophobicity as a determinant of the cytotoxic potential of pyochelin. However, for those analogues that are hydrophobic (octanol:water partition coefficient ⬎1.0), no direct correlation was seen between the relative level of hydrophobicity and the amount of endothelial cell damage they engendered. This could reflect the relative insensitivity of the 51Cr release for detecting small differences in cell injury. To what extent pyochelin must be hydrophobic in order to exhibit its cytotoxic effect will likely require the generation of additional derivatives with less hydrophobicity. Previous work by other groups has demonstrated that iron bound to another hydrophobic iron chelate, 8-hydroxyquinoline, also exhibits a high propensity to cause cell injury, whereas more hydrophilic ones do not (28 –30). In summary, the present work has identified features of the P. aeruginosa-derived siderophore pyochelin which are likely important for its ability to augment oxidant-mediated endothelial cell injury. The present data suggest that the thiazolidine ring of pyochelin, perhaps via its ability to increase the hydrophobicity of the molecule and transfer iron to key sites on the endothelial cell surface, is critical for cytotoxic activity. In addition, reduction of ferric to ferrous iron upon chelation of ferric iron by pyochelin may be of importance, but this, as well as the chemical mechanism involved, will require further assessment. ACKNOWLEDGMENTS This material is based upon work supported by the Medical Research Service, Department of Veterans Affairs, and the Public Health Service (Grants HL 44275, AI 34954, and T32 AI70343) and was performed in part during the tenure of B. E. Britigan as an Established Investigator of the American Heart Association.

REFERENCES 1. Pennington, J. P. (1990) in Principles and Practice of Infectious Diseases (Mandell, G. L., Douglas, R. G., Jr., and Bennett, J. E., Eds.), pp. 2199 –2205, Churchill Livingstone, New York. 2. Fick, R. B., Jr., and Hata, J. S. (1989) Chest 95, 206S–213S. 3. Bodey, G. P., Bolivar, R., Fainstein, V., and Jadeja, L. (1983) Rev. Infect. Dis. 5, 279 –313. 4. Root, R. K., and Cohen, M. S. (1981) Rev. Infect. Dis. 3, 565–598. 5. Lubec, G. (1996) J. Invest. Med. 44, 324 –346. 6. Halliwell, B., and Gutteridge, J. M. C. (1986) Arch. Biochem. Biophys. 246, 501–514. 7. Weiss, S. J. (1986) Acta Physiol. Scand. 548(Suppl.), 9 –37. 8. Miller, R. A., and Britigan, B. E. (1995) J. Invest. Med. 43, 39 – 49.

244

DEWITTE ET AL.

9. Winterbourn, C. C. (1983) Biochem. J. 210, 15–19. 10. Baldwin, D. A., Jenny, E. R., and Aisen, P. (1984) J. Biol. Chem. 259, 13391–13394. 11. Buettner, G. R. (1987) J. Biol. Chem. 262, 11995–11998. 12. Sriyosachati, S., and Cox, C. D. (1986) Infect. Immun. 52, 885– 891. 13. Cox, C. D., and Adams, P. (1985) Infect. Immun. 48, 130 –138. 14. Cox, C. D., Rinehart, K. L., Jr., Moore, M. L., and Cook, C. J., Jr. (1981) Proc. Natl. Acad. Sci. USA 78, 4256 – 4260. 15. Ankenbauer, R., Sriyosachati, S., and Cox, C. D. (1985) Infect. Immun. 49, 132–140. 16. Coffman, T. J., Cox, C. D., Edeker, B. L., and Britigan, B. E. (1990) J. Clin. Invest. 86, 1030 –1037. 17. Britigan, B. E., Rasmussen, G. T., and Cox, C. D. (1994) Am. J. Physiol. Lung Cell. Mol. Physiol. 266, L192–L198. 18. Britigan, B. E., Roeder, T. L., Rasmussen, G. T., Shasby, D. M., McCormick, M. L., and Cox, C. D. (1992) J. Clin. Invest. 90, 2187–2196. 19. Britigan, B. E., Rasmussen, G. T., and Cox, C. D. (1997) Infect. Immun. 65, 1071–1076.

20. Czapski, G. (1984) Isr. J. Chem. 24, 29 –32. 21. Ankenbauer, R. G., and Cox, C. D. (1988) J. Bacteriol. 170, 5364 –5367. 22. Ankenbauer, R. G., Staley, A. L., Rinehart, K. L., and Cox, C. D. (1991) Proc. Natl. Acad. Sci. USA 88, 1878 –1882. 23. Cox, C. D., and Graham, R. (1979) J. Bacteriol. 137, 357–364. 24. Ramos, C. L., Pou, S., Britigan, B. E., Cohen, M. S., and Rosen, G. M. (1992) J. Biol. Chem. 267, 8307– 8312. 25. Bynoe, L. A., Pou, S., Gottsch, J. D., and Rosen, G. M. (1991) Biochem. Biophys. Res. Commun. 179, 1305–1310. 26. Olakanmi, O., Stokes, J. B., and Britigan, B. E. (1994) J. Immunol. 153, 2691–2703. 27. Sutton, H. C., and Winterbourn, C. C. (1989) Free Radical Biol. Med. 6, 53– 60. 28. Balla, G., Vercellotti, G. M., Eaton, J. W., and Jacob, H. S. (1990) J. Lab. Clin. Med. 116, 546 –554. 29. Leanderson, P., and Tagesson, C. (1996) Carcinogenesis 17, 545– 550. 30. Jonas, S. K., and Riley, P. A. (1992) Free Radical Res. Commun. 17, 407– 418.