Revisiting Quorum Sensing: Discovery of Additional Chemical and Biological Functions for 3-Oxo-N-Acylhomoserine Lactones

Revisiting quorum sensing: Discovery of additional chemical and biological functions for 3-oxo-N-acylhomoserine lactones

Gunnar F. Kaufmann, Rafaella Sartorio, Sang-Hyeup Lee, Claude J. Rogers, Michael M. Meijler, Jason A. Moss, Bruce Clapham, Andrew P. Brogan, Tobin J. Dickerson, and Kim D. Janda*

Department of Chemistry and The Skaggs Institute for Chemical Biology, The Scripps Research Institute, 10550 North Torrey Pines Road, La Jolla, CA 92037

Communicated by Sydney Brenner, The Salk Institute, La Jolla, CA, November 23, 2004 (received for review October 21, 2004)

Bacteria use small diffusible molecules to exchange information in a process called quorum sensing. An important class of autoinducers used by Gram-negative bacteria is the family of N-acylhomoserine lactones. Here, we report the discovery of a previously undescribed nonenzymatically formed product from N-(3-oxododecanoyl)-L-homoserine lactone; both the N-acylhomoserine and its novel tetramic acid degradation product, 3-(1-hydroxydecylidene)-5-(2-hydroxyethyl)pyrrolidine-2,4-dione, are potent antibacterial agents. Bactericidal activity was observed against all tested Gram-positive bacterial strains, whereas no toxicity was seen against Gram-negative bacteria. We propose that Pseudomonas aeruginosa utilizes this tetramic acid as an interference strategy to preclude encroachment by competing bacteria. Additionally, we have discovered that this tetramic acid binds iron with comparable affinity to known bacterial siderophores, possibly providing an Fig. 1. AHLs used by P. aeruginosa in quorum sensing. unrecognized mechanism for iron solubilization. These findings merit new attention such that other previously identified autoinducers be reevaluated for additional biological functions. patients become infected with P. aeruginosa, and, of those, the majority die of septicemia. These infections are especially trou- tetramic acid ͉ bactericidal agents ͉ evolution blesome because P. aeruginosa continues to grow more resistant to many antibiotics. Over the last 10 years, significant progress has been made in he term ‘‘quorum sensing’’ has been coined to describe the elucidating the molecular mechanisms underlying P. aeruginosa Tability of a population of unicellular bacteria to act as a pathogenicity. Two different AHLs, N-(3-oxododecanoyl) ho- multicellular organism in a cell-density-dependent manner; that moserine lactone 1, synthesized by LasI, and N-butyrylhomo- is, a way to sense ‘‘how many are out there’’ (1–3). Bacteria use serine lactone 2, synthesized by RhlI, have been identified as the small diffusible molecules to exchange information among them- main quorum-sensing signaling molecules in P. aeruginosa (Fig. selves (2). An important class of ‘‘quormones,’’ or autoinducers, 1) (6). Genes regulated by this mechanism encode enzymes such is the family of N-acylhomoserine lactones (AHLs) used by as elastases A and B, catalase, and superoxide dismutase. Gram-negative bacteria. Variation in N-acyl chain length and the Furthermore, quorum sensing also has been demonstrated to BIOCHEMISTRY oxidation state of AHLs provide for bacterial strain specificity in control the expression of other virulence factors as well as the the signaling process and subsequent synchronization of gene formation of structures known as biofilms (7). Here, we dem- expression. Depending on their acylation pattern, most AHLs onstrate that N-(3-oxododecanoyl) homoserine lactone 1 per- diffuse freely across the bacterial cell membrane. Upon reaching forms a previously unrecognized role: The autoinducer itself and a critical threshold concentration, they bind to their cognate a corresponding degradation product derived from an unusual receptor proteins, triggering the expression of target genes; for Claisen-like condensation reaction function as innate bacteri- example, in the case of Vibrio fischeri, genes located on the lux cidal agents. Furthermore, the AHL degradation product tightly operon are transcribed that are responsible for the production of binds essential metals such as iron, possibly providing a previ- bioluminescence (4). Indeed, this account of V. fischeri was the ously unrecognized primordial siderophore. first description of a bacterial population acting in a concerted fashion to achieve a common goal: the simultaneous expression Materials and Methods of a particular set of genes. General Synthetic Methods. Unless otherwise stated, all reactions Pseudomonas aeruginosa is a common environmental micro- were performed under an inert atmosphere with dry reagents organism that has acquired the ability to take advantage of and solvents and flame-dried glassware. Analytical TLC was weaknesses in the host immune system to become an opportu- performed by using 0.25-mm precoated silica gel Kieselgel 60 nistic pathogen in humans (5). Most prominent is the role of P. F254 plates. Visualization of the chromatogram was by UV aeruginosa in patients suffering from cystic fibrosis because absorbance, iodine, dinitrophenylhydrazine, ceric ammonium lung-defense functions are severely impaired. As a result, infec- molybdate, ninhydrin, or potassium permanganate as appropri- tions with P. aeruginosa and the damage caused by the inflam- ate. All 1H NMR spectra were recorded on either a Bruker matory infection process increase the mortality rate in cystic fibrosis patients. Additionally, nosocomial infections by P. aeruginosa, particularly in burn victims, cause serious complica- Abbreviation: AHL, N-acylhomoserine lactone. tions because of the wide variety of virulence factors and *To whom correspondence should be addressed. E-mail: [email protected]. inherent antibiotic resistance. Significant percentages of burn © 2004 by The National Academy of Sciences of the USA

www.pnas.org͞cgi͞doi͞10.1073͞pnas.0408639102 PNAS ͉ January 11, 2005 ͉ vol. 102 ͉ no. 2 ͉ 309–314 Downloaded by guest on September 30, 2021 AMX-500 or DRX-600 spectrometer (Billerica, MA) at 500 and of the extract was performed by using the described reverse- 600 MHz, respectively. All 13C NMR spectra were recorded on phase HPLC system (see above). The gradient used was as a Bruker AMX-500 spectrometer at 125 MHz. Optical rotations follows: t ϭ 0, 50% solvent B (0.1% trifluoroacetic acid in were determined at 598 nm in a conventional 10-cm cell by using acetonitrile) in solvent A (0.1% trifluoroacetic acid in water); t ϭ a PerkinElmer 241 MC polarimeter. MALDI-Fourier trans- 5, min 50% solvent B; t ϭ 25 min, 80% solvent B; t ϭ 35 min, form-MS experiments were performed on an IonSpec-Fourier 100% solvent B; t ϭ 50 min, 100% solvent B. Fractions were transform mass spectrometer (Lake Forest, CA). Electrospray collected every minute and analyzed by electrospray ionization- ionization-MS experiments were performed on an API 100 MS. Peaks with mass corresponding to tetramic acid 4 were PerkinElmer Sciex single-quadrupole mass spectrometer. Ana- found to elute at a retention time of 14.7 min and were confirmed lytical reverse-phase HPLC was performed on a Hitachi L-5000 by coinjection with authentic standard samples. series instrument equipped with a Vydac-C18 analytical column, a UV detector at 254 nm, and mobile phases composed of Kinetic Assays for the Formation of 3-(1-Hydroxydecylidene)-5-(2- mixtures of acetonitrile͞water (0.1% trifluoroacetic acid). Hydroxyethyl)Pyrrolidine-2,4-Dione 4. The assay was initiated by the addition of a solution of AHL (5 mM in DMSO) to phosphate General Procedure for Synthesis of Tetramic Acids. As a represen- buffer at 25°C (200 mM, pH 7.4͞10% DMSO cosolvent; 1 ml tative example, the synthesis of (S)-3-(1-hydroxydecylidene)-5- total volume). Continuous monitoring of the reaction was per- (2-hydroxyethyl)pyrrolidine-2,4-dione 4 is described. To a vial formed over 2,000 s by using a Hewlett Packard 8452A UV- containing (S)-3-oxo-C12-AHL 1 (111 mg, 0.374 mmol) in dry visible spectrophotometer. Tetramic acid product formation was MeOH (1 ml), NaOMe in MeOH (0.5 M, 0.747 ml, 0.374 mmol) determined spectrophotometrically by using an extinction coef- was added at room temperature under argon. After stirring at ficient of 13,900 MϪ1⅐cmϪ1 at ␭ ϭ 278 nm. The acquired data 55°C for 3 h, the reaction mixture was passed through acidic were fit to a two-state model in which competing reactions were ion-exchange resin (Dowex 50WX2–200, Ϸ2cm3) and further assumed to be nonequilibrating. Data analysis was performed by eluted with MeOH (20 ml). The combined filtrate was concen- nonlinear curve-fitting algorithms by using the program KALEI- trated under reduced pressure, and the resulting residue was DAGRAPH 3.6.2 (Synergy Software, Reading, PA). recrystallized (EtOAc͞hexanes) to give an off-white small crys- talline solid (80 mg). This product then was purified by prepar- Antibacterial Assays. The following bacterial strains were pur- ative HPLC to give tetramic acid 4 as white fluffy solid. Tetramic chased from the American Type Culture Collection (ATCC) acids were purified further by reverse-phase HPLC using a unless otherwise stated: Bacillus cereus ATCC 11778, B. cereus Vydac 214TP101522 column (Hesperia, CA) at a flow rate of 10 ATCC 13061, B. cereus ATCC 14579, B. cereus ATCC 27348, ml͞min with detection at 230 nm on a dual-pump Rainin Bacillus licheniformis 5A36 (from Bacillus Genetic Stock Center, Dynamax HPLC system (Rainin Instruments). In each prepar- Columbus, OH), Bacillus mycoides ATCC 6462, Bacillus subtilis ative separation, recrystallized material (40 mg) was dissolved in ATCC 6051, Enterococcus faecalis ATCC 29212, Escherichia coli 5 ml of 45:45:10 acetic acid͞water͞DMSO, filtered through a ATCC 35150, Listeria monocytogenes ATCC 43251, P. aeruginosa 0.45-␮m poly(vinylidene difluoride) filter, and purified by using PAO1 (generous gift from B. Iglewski, University of Rochester, a gradient of 40–60% solvent B (0.085% trifluoroacetic acid in Rochester, NY), Staphylococcus aureus ATCC 25923, Staphylo- acetonitrile) in solvent A (0.1% trifluoroacetic acid in water) coccus epidermidis ATCC 12228, Streptococcus pyogenes ATCC over 30 min. Pure fractions were identified by electrospray 49399, and Salmonella typhimurium ATCC 13311. ionization-MS on a Sciex API-150EX single quadrupole mass A bacterial colony was picked and grown overnight in the spectrometer (DP 10 V, FP 50 V) operated in multichannel growth medium and at the temperature recommended by analysis mode, pooled, and lyophilized. In all cases, the homo- ATCC. On the next day, the culture was diluted 1:1,000 in fresh geneity of pure pooled products was verified by analytical growth medium and grown until an OD650 of 0.1 was reached. reverse-phase HPLC using a C4-functionalized Vydac The culture then was diluted 1:200 in fresh growth medium. 218TP5415 column running a gradient of 20–80% B over 30 Aliquots (198 ␮l) were added into a 96-well microtiter plate 1 min; H NMR (600 MHz, CD3OD): ␦ 0.90 (t, J ϭ 7.0 Hz, 3 H), containing the test compound dissolved in 2 ␮l of DMSO 1.23–1.43 (m, 12 H), 1.66 (qt, J ϭ 7.4 Hz, 2 H), 1.71–1.82 (m, 1 (Sigma) at 100ϫ the desired concentration. Control experiments H), 1.96–2.08 (m, 1 H), 2.72–2.94 (m, 2 H), 3.62–3.78 (m, 2 H), contained only DMSO. The microtiter plate was sealed 13 3.97 (br s, 1H); C NMR (125 MHz, CDCl3): ␦ 14.1, 22.6, 25.9, (BreatheEasy, Research Products International) and incubated 29.2, 29.4, 31.8, 32.9, 34.0, 61.3, 62.0, 100.5, 175.0, 190.0, 195.3. on a shaker at the appropriate temperature overnight. The next ϩ MALDI-Fourier transform-MS for C16H28NO4 (M ϩ H ) cal- day, OD650 was measured by using a ThermoMax plate reader 25 culated 298.2017, found 298.2013. [␣] d ϭϩ14.0 (c ϭ 7.20, (Molecular Devices). EC50 values then were determined by MeOH). nonlinear curve-fitting using KALEIDAGRAPH 3.6.2 (Synergy Soft- ware). Reported values are the average of a minimum of three Detection of 3-(1-Hydroxydecylidene)-5-(2-Hydroxyethyl)Pyrrolidine- replicates. 2,4-Dione 4 from P. aeruginosa Culture. P. aeruginosa PAO1 cells were grown for 24 h in Luria–Bertani broth (2 liters) containing Electron Microscopy of Gram-Positive Bacterial Cells Exposed to 50 mM Mops (pH 7.4). The cells were harvested by centrifuga- Tetramic Acid 4. B. cereus (ATCC 11778) was grown overnight at tion and resuspended in BugBuster (50 ml; Novagen). After 30°C. The next day, the cells were diluted 1:500 and grown at incubation for1hatroom temperature, the cells were removed 30°C. After 3 h, tetramic acid 4 or AHL 1 was added to a final by centrifugation, and an aliquot of the supernatant (20 ml) was concentration of 100 mM (1% DMSO). Control cultures poured into a separatory funnel containing brine (1 liter) and contained only 1% DMSO. After further incubation (5 h) on CHCl3 (1 liter). The layers then were separated, and the aqueous a shaker at 30°C, the B. cereus samples were fixed in 2% layer was acidified to pH 2 with 6 M HCl. The aqueous mixture paraformaldehyde and 2.5% glutaraldehyde in 0.12 M caco- then was extracted with CHCl3 (2 ϫ 1 liter), and the organic dylate buffer (pH 7.3) and pelleted by using a tabletop layers were combined and concentrated. The resulting residue microfuge, and the resulting pellets were postfixed in cacody- was dissolved in hot acetonitrile (100 ml) and filtered through a late-buffered 1% osmium tetroxide. After dehydration in 0.45-␮m poly(vinylidene difluoride) filter. The filtrate then was graded ethanol series followed by propylene oxide, the samples concentrated, redissolved in hot acetonitrile (6 ml), and filtered were embedded in EMbed 812͞Araldite 502 (Electron again. Upon cooling, the solution was brown and turbid. Analysis Microscopy Sciences, Fort Washington, PA). Thin sections

310 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0408639102 Kaufmann et al. Downloaded by guest on September 30, 2021 were cut transversely, mounted on parlodion-coated copper slot grids, and stained with uranyl acetate and lead citrate; images were documented on a Philips CM100 electron micro- scope (FEI, Hillsborough, OR) using Kodak SO163 film. Negatives were scanned at 600 dots per inch by using a Fuji FineScan 2750 and converted to TIFF format for subsequent handling in PHOTOSHOP (Adobe Systems, San Jose, CA).

Iron Binding to Tetramic Acid 4. Stock solutions of ferric chloride (10 mM) and tetramic acid 4 (10 mM) were prepared in spectroscopic-grade methanol. Aliquots of both the ferric chloride and tetramic acid 4 stocks then were diluted to 150 and 500 ␮M, respectively, with 4:1 MeOH͞0.1 M NaOAc buffer solution (pH 7.4), with and without EDTA (100 ␮M). The absorbances of the formed complexes were measured at 440 nm in triplicate in the presence and absence of EDTA. Samples containing varying ligand and ferric chloride concentrations in methanol also were analyzed by MS as described above. Results and Discussion Our studies were initiated to examine the lifetime of AHL 1 in an aqueous environment. Incubation of compound 1 in water produced an undocumented compound in addition to the expected hydrolysis product 3. Structural characterization of this anomalous molecule revealed it to be 3-(1-hydroxydecylidene)- 5-(2-hydroxyethyl)pyrrolidine-2,4-dione 4, a compound belong- ing to a class of antibacterial compounds known as tetramic acids (Fig. 2) (8). The mechanism of formation is a Claisen-like intramolecular alkylation of the ␤-ketoamide moiety (Fig. 3). Specifically, the ␣-carbon between the ketone and amide moi- eties is deprotonated, leading to an anion that cyclizes intramo- lecularly on the lactone generating the tetramic acid motif. Tetramic acid 4 was found to be very stable, with no decomposition or detectable reversion to compound 1; most importantly, compound 4 also was detected in P. aeruginosa culture. To assess the generality of this phenomenon, a variety of AHLs with varying acyl chain lengths were prepared and incubated in buffer. This chemical reaction was not limited to compound 1 because all tested 3-oxo-AHLs with varying chain lengths also underwent this intramolecular rearrangement (Table 1). Sur- prisingly, although previous studies have attempted to measure the kinetics of AHL hydrolysis, the formation of 4 has never been reported, to our knowledge (9). Fig. 2. Representative natural products containing a tetramic acid motif. Interestingly, various tetramic acids have been demonstrated to possess my- BIOCHEMISTRY AHL 1 and Tetramic Acid 4 Are Innate Bactericidal Agents. Given that cotoxic, antibacterial, antiviral, and antioxidant activities. P. aeruginosa employs AHL 1 as the principal autoinducer and the known bactericidal activity of tetramic acids, we hypothe- ionophore, thereby dissipating the transmembrane change in pH sized that compound 4 could have a biological function in the and leading to cell lysis (13). The efficiency of this mechanism context of bacterial viability. We were guided by the observation is a function of the high hydrophobicity of reutericyclin, thereby that lipid-substituted antibiotics have been shown to localize near the cell membrane of Gram-positive bacteria (10). Indeed, favoring partitioning into the cytoplasmic membrane. Based on significant antibacterial activity was observed against all tested the similarities between this compound and 4, including the long Gram-positive bacterial strains in the presence of 4 after over- hydrophobic carbon chain and tetramic acid functionality, it is night incubation (Table 2), whereas no toxicity was observed plausible that 4 also would operate by means of a similar against P. aeruginosa or other tested Gram-negative bacteria mechanism. (Fig. 4). The effective concentration of 4 is biologically relevant, Notably, compound 1 also displayed cytotoxicity (Table 2) because high concentrations of compound 1 (Ͼ600 ␮M) have suggesting a dual role for 1 in P. aeruginosa communities as both been detected previously in P. aeruginosa biofilms (11). Further- quorum-sensing molecules and as an interference mechanism more, 4 has comparable activity (average EC50 ϭ 23 ␮MorϷ60 against bacterial competitors. Here, cytotoxicity cannot be at- ng͞ml) to other known antibacterial compounds. tributed simply to tetramic acid because the length at which the Similar to previously described tetramic acids and related experiment was conducted was such that tetramic acid formation analogs, sensitivity to compound 4 was apparent only in Gram- is expected to be minimal (see below). Indeed, although signif- positive cells with little variation present in EC50 values across icant cell death was observed after overnight incubation with different species and genera (12). Analogous observations have AHL 1 (Fig. 4C), the bacteria recovered after additional incu- been made in the case of the natural product reutericyclin, a bation, suggesting that although AHL 1 is bactericidal, upon tetramic acid isolated from the sourdough isolate Lactobacillus hydrolysis, the resulting compound 3 is tolerated by bacteria. reuteri LTH2584. This compound has been demonstrated to be Interestingly, resistant B. cereus strains known to express a bactericidal in Gram-positive bacteria by acting as a proton- penicillinase or cephalosporinase in conjunction with a penicil-

Kaufmann et al. PNAS ͉ January 11, 2005 ͉ vol. 102 ͉ no. 2 ͉ 311 Downloaded by guest on September 30, 2021 Fig. 3. Reaction of 3-oxo-AHL 1 to generate lactone hydrolysis product 3 (path a) and 3-(1-hydroxydecylidene)-5-(2-hydroxyethyl)pyrrolidine-2,4-dione 4 (path b).

linase (ATCC 13061 and 27348, respectively) were not affected formation 4. This hypothesis is concurrent with recent discus- by exposure to 1, yet displayed comparable sensitivity to tetramic sions about a more complex role of lactonases in vivo (17). acid 4 as other tested B. cereus strains (Table 2). We acknowledge Second, this finding raises the enticing evolutionary prospect that the observed antibacterial activity of 4 is relatively modest that 3-oxo-AHLs and their corresponding tetramic acids were in comparison with other known more potent bactericidal selected for as a result of their fortuitous cytotoxic abilities and agents; however, the fact that this compound is nonenzymatically resistance to lactonase degradation, respectively. produced from a molecule designed to play a role in cell–cell signaling is significant. Furthermore, even in the instance that 3-(1-Hydroxydecylidene)-5-(2-Hydroxyethyl)Pyrrolidine-2,4-Dione 4 only a small percentage of the competitor population is killed by Binds Iron with High Affinity. Iron plays an essential role in 4, this characteristic then should provide a competitive advan- physiological processes and the pathogenesis of bacteria (18, 19). tage for P. aeruginosa, thereby insuring the survival of the colony Many bacteria are known to produce siderophores to sequester or biofilm. iron, an element that, although essential for their growth, has The identification of AHL-hydrolyzing enzymes, or lacto- poor solubility under physiological conditions (20). P. aeruginosa nases, in B. cereus 240B1 led to the hypothesis that certain synthesizes two siderophores, pyoverdin and pyochelin, and has bacteria have evolved quorum-sensing signaling interference strategies, or so-called ‘‘quorum quenching’’ (14, 15). The hy- Table 2. Cytotoxicity of tetramic acid 4 and AHL 1 in various drolysis of AHLs not only prevents cell density-dependent bacterial cell lines signaling events from occurring, but also presumably could ␮ ␮ mitigate the cytotoxic effect of AHLs and furthermore prevent Bacteria* 4 EC50, M 1 EC50, M the formation of tetramic acid 4. Indeed, B. cereus ATCC 14579 Gram-positive cells, previously shown to express lactonase (16), do not show B. cereus (11778) 8.3 22.1 sensitivity to compound 1, yet are still affected by compound 4 B. cereus (14579) 14.8 Ͼ100 (EC50 ϭ 13.4 Ϯ 1.6 ␮M). This finding advocates two distinct B. cereus (13061) 12.5 Ͼ100 implications: the first is that, as previously hypothesized, lacto- B. cereus (27348) 18.3 Ͼ100 nases may have evolved in certain Gram-positive bacteria to B. licheniformis† 36.1 n.d. interrupt quorum-sensing signaling. However, our findings also B. mycoides (6462) 13.8 n.d. suggest that these enzymes have a previously unrecognized B. subtilis (6051) 13.9 31.3 function: to abrogate the toxic effects of 1 and prevent the E. faecalis (29212) 33.7 n.d. L. monocytogenes (43251) 34.1 n.d. S. aureus (25923) 26.7 80.2 Table 1. Summary of kinetic data for the decomposition of S. epidermidis (12228) 15.8 n.d. AHL 1 S. pyogenes (49399) 55.3 40.8 Ϫ1 Ϫ1 Gram-negative Compound khydrolysis,s ktetramic acid,s E. coli (35150) Ͼ100 Ͼ100 ϫ Ϫ6 ϫ Ϫ6 3-Oxo-C4-AHL 8.06 10 5.97 10 P. aeruginosa‡ Ͼ100 Ͼ100 ϫ Ϫ5 ϫ Ϫ6 3-Oxo-C6-AHL 3.07 10 2.29 10 S. typhimurium (13311) Ͼ100 Ͼ100 Ϫ5 Ϫ6 3-Oxo-C8-AHL 1.48 ϫ 10 1.28 ϫ 10 Ϫ6 Ϫ6 n.d., not determined. 3-Oxo-C12-AHL 1 4.77 ϫ 10 1.49 ϫ 10 *ATCC numbers are given in parentheses. † In each entry, Cn refers to the number of carbon atoms present in the N-acyl Strain 5A36 from Bacillus Genetic Stock Center. chain. ‡PAO1 from B. Iglewski.

312 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0408639102 Kaufmann et al. Downloaded by guest on September 30, 2021 Fig. 4. Electron micrographs of B. cereus ATCC 11778 after5hofincubation with DMSO (A), 100 ␮M tetramic acid 4 (B), or 100 ␮MAHL1(C). Cells in both B and C show signiﬁcant morphological changes.

been shown to be able to use a variety of heterologous sid- stoichiometry. Analogous to these studies, we have envisioned a erophores of microbial origin. The P. aeruginosa genome con- similar mode of complexation for the compound 4–Fe3ϩ com- tains many homologues of iron-siderophore receptor genes, plex (Fig. 5). To verify our hypothesis, we performed studies by reflecting the enormous flexibility of this organism to use using varying concentrations of ferric chloride and tetramic acid various iron carriers. The ability to use such a wide variety of and observed the formation of a 4-Fe3ϩ complex by spectro- siderophores underscores the importance of iron for bacterial photometric analysis at 440 nm; the corresponding 3-oxo-C12- growth and survival as well as the need to compete with other AHL did not show significant absorption at this wavelength upon microorganisms within the environments they inhabit for this FeCl3 addition. The complex was further characterized by using essential metal (21, 22). electrospray ionization-MS, confirming the formation of a 3:1 The 3-acetyl-pyrrolidine-2,4-dione heterocycle found in com- 4-Fe3ϩ complex ([M ϩ H]ϩ m͞z ϭ 945.5). pound 4 and in many other naturally occurring tetramic acids has Additional support for our hypothesis was given through the been shown to efficiently chelate a variety of metal cations measurement of the apparent binding constant (Kd,app) for including iron (23). The role of metal binding in the bactericidal iron(III) with compound 4. These experiments were performed activity of tetramic acids is unclear, with some tetramic acids by using a previously described protocol based on the compe- displaying increased toxicity as a metal complex, whereas in tition between tetramic acid 4 and EDTA for iron and the facile others, the toxicity is attenuated by chelation of a metal ion (24). detection of 4-Fe3ϩ complexes by their characteristic absorption Given this disparity, we investigated the ability of AHL-derived (26). The loss of 4-Fe3ϩ absorbance at 440 nm upon addition of tetramic acid 4 to complex metals to evaluate the potential EDTA was used to calculate the equilibrium constant, presum- bioactivity of these complexes. Bearing in mind the crucial role ing the formation of a 3:1 tetramic acid–Fe complex, according that iron plays in bacterial physiology (see above), and the to the following equation: demonstrated potential of tetramic acids to bind this metal, we ϭ ͓ ͞ 3 focused our investigations on the chelation of iron by 4. Keq [Fe(4)3] EDTA] [Fe(EDTA)][4] Lebrun et al. (25) studied the complexation of the fungal ϭ ͞ tetramic acid metabolite tenuazonic acid (Fig. 2) with iron(III) Kd,EDTA Kd,tetramic acid. [1] and observed the formation an octahedral complex with a 3:1 ϩ Ϫ

3 23 BIOCHEMISTRY By using the known affinity of EDTA for Fe (Kd ϭ 5 ϫ 10 M), we were able to determine the relative affinity (Kd,app)of compound 4 for Fe3ϩ to be 1.6 ϫ 10Ϫ29 M3. A direct comparison of iron affinity between this presumed bidentate chelator and known hexadentate chelators such as desferrioxamine, EDTA, and pyoverdin is not possible because the affinity of hexadenate chelators are measured in units of molarity, whereas bidentate chelators are measured in units of (molarity)3. However, a parameter termed ‘‘pM’’ has been used previously as a method of standardization and is given by

ϭ Ϫ 3ϩ pM log10[Fe ], [2]

Table 3. Afﬁnity constants for iron (III) chelators

Compound Kd pM

EDTA* 5.00 ϫ 10Ϫ23 M 25.3 DFO* 2.51 ϫ 10Ϫ26 M 28.6 Pyoverdin† 10Ϫ32 M35 Pyochelin† 10Ϫ5 M8 Tetramic acid 4 1.6 Ϯ 0.3 ϫ 10Ϫ29 M3 25.8

Measured in 20% 0.1 M NaOAc͞MeOH buffer, pH 7.4. Fig. 5. Proposed structure of the compound 4-Fe3ϩ complex demonstrating *See ref. 27. a 3:1 ratio of tetramic acid͞metal. †See ref. 28.

Kaufmann et al. PNAS ͉ January 11, 2005 ͉ vol. 102 ͉ no. 2 ͉ 313 Downloaded by guest on September 30, 2021 where [Fe3ϩ] is equal to the concentration of iron in the presence a population of cells synthesizes 3-oxo-AHLs. However, the of 1 mM total iron chelator and 1 ␮M total Fe3ϩ at pH 7.4 (25). autoinducer not only functions as a transcriptional regulator but By using these values (Table 3), compound 4 was deduced to also serves as a bactericidal agent. Although the complexation of have a roughly three times stronger affinity for iron(III) than critical metals such as iron may play a role in this process, further EDTA, but weaker than pyoverdin. Interestingly, the reported study is required into the mechanism and scope of the observed Ϫ5 Kd for the secondary siderophore, pyochelin (10 M), is bactericidal activity and the potential of 4 to act as a primordial quite poor, suggesting that 4 could compete for available iron in siderophore. The compilation of our data recommends that solution and provide an additional method for iron other autoinducers be reevaluated for the potential to perform solubilization. additional biological functions. In summary, we propose that P. aeruginosa utilizes compound 4 as an interference strategy to preclude encroachment by We thank Professor Barbara Iglewski for generously providing P. competing bacteria. In essence, this behavior parallels other aeruginosa PAO1 and Malcolm Wood and Theresa Fassel for expert known quorum-controlled processes in that an individual cell assistance with electron microscopy experiments. This work was sup- cannot produce sufficient quantities of tetramic acid to affect ported by National Institutes of Health Grant AI055781 and by The other bacteria; the cytotoxic properties of 4 only emerge when Skaggs Institute for Chemical Biology.

1. Schauder, S. & Bassler, B. L. (2001) Genes Dev. 15, 1468–1480. 15. Dong, Y. H., Wang, L. H., Xu, J. L., Zhang, H. B., Zhang, X. F. & Zhang, L. H. 2. Miller, M. B. & Bassler, B. L. (2001) Annu. Rev. Microbiol. 55, 165–199. (2001) Nature 411, 813–817. 3. Fuqua, C., Parsek, M. R. & Greenberg, E. P. (2001) Annu. Rev. Genet. 35, 16. Dong, Y. H., Gusti, A. R., Zhang, Q., Xu, J. L. & Zhang, L. H. (2002) Appl. 439–468. Environ. Microbiol. 68, 1754–1759. 4. Dunlap, P. V. (1999) J. Mol. Microbiol. Biotechnol. 1, 5–12. 17. Roche, D. M., Byers, J. T., Smith, D. S., Glansdorp, F. G., Spring, D. R. & 5. Lyczak, J. B., Cannon, C. L. & Pier, G. B. (2000) Microbes Infection 2, Welch, M. (2004) Microbiology 150, 2023–2028. 1051–1060. 18. Cox, C. D. & Adams, P. (1985) Infect. Immun. 48, 130–138. 6. Pearson, J. P., Passador, L., Iglewski, B. H. & Greenberg, E. P. (1995) Proc. 19. Gensberg, K., Hughes, K. & Smith, A. W. (1992) J. Gen. Microbiol. 138, Natl. Acad. Sci. USA 92, 1490–1494. 2381–2387. J. Clin. Invest. 112, 7. Smith, R. S. & Iglewski, B. H. (2003) 1460–1465. 20. Payne, S. M. (1994) Methods Enzymol. 235, 329–344. 8. Ghisalberti, E. L. (2003) Studies Nat. Products Chem. 28, 109–163. 21. Poole, K. & McKay, G. A. (2003) Front. Biosci. 8, d661–d686. 9. Yates, E. A., Philipp, B., Buckley, C., Atkinson, S., Chhabra, S. R., Sockett, 22. Takase, H., Nitanai, H., Hoshino, K. & Otani, T. (2000) Infect. Immun. 68, R. E., Goldner, M., Dessaux, Y., Camara, M., Smith, H., et al. (2002) Infect. 4498–4504. Immun. 70, 5635–5646. Adv. Heterocycl. Chem. 57, 10. Dong, S. D., Oberthur, M., Losey, H. C., Anderson, J. W., Eggert, U. S., Peczuh, 23. Henning, H.-G. & Gelbin, A. (1993) 139–185. M. W., Walsh, C. T. & Kahne, D. (2002) J. Am. Chem. Soc. 124, 9064–9065. 24. Gandhi, N. M., Nazareth, J., Divekar, P. V., Kohl, H. & de Souza, N. J. (1973) 11. Charlton, T. S., de Nys, R., Netting, A., Kumar, N., Hentzer, M., Givskov, M. J. Antibiot. 797–798. & Kjelleberg, S. (2000) Environ. Microbiol. 2, 530–541. 25. Lebrun, M. H., Nicolas, L., Boutar, M., Gaudemer, F., Ranomenjanahary, S. 12. Ga¨nzle, M. G., Ho¨ltzel, A., Walter, J., Jung, G. & Hammes, W. P. (2000) Appl. & Gaudemer, A. (1988) Phytochemistry 27, 77–84. Environ. Microbiol. 66, 4325–4333. 26. Wang, J., Buss, J. L., Chen, G., Ponka, P. & Pantopoulos, K. (2002) FEBS Lett. 13. Ga¨nzle,M. G. & Vogel, R. F. (2003) Appl. Environ. Microbiol. 69, 1305–1307. 529, 309–312. 14. Dong, Y. H., Xu, J. L., Li, X. Z. & Zhang, L. H. (2000) Proc. Natl. Acad. Sci. 27. Martell, A. E. (1977) Critical Stability Constants (Plenum, New York). USA 97, 3526–3531. 28. Ratledge C. & Dover, L. G. (2000) Annu. Rev. Microbiol. 54, 881–941.

314 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0408639102 Kaufmann et al. Downloaded by guest on September 30, 2021