Ternary Complexes of Gentamicin with Iron and Lipid Catalyze

the formation of reactive oxygen species by gentamicin and iron cell-free systems (8, 9), and lipid peroxidation occurs in the early stages of aminogl...
0 downloads 0 Views 284KB Size
Download PDF
Chem. Res. Toxicol. 2005, 18, 357-364

357

Ternary Complexes of Gentamicin with Iron and Lipid Catalyze Formation of Reactive Oxygen Species Wojciech Lesniak,†,‡ Vincent L. Pecoraro,*,‡ and Jochen Schacht*,†,§ Kresge Hearing Research Institute, Department of Otolaryngology, Department of Chemistry, and Department of Biological Chemistry, University of Michigan, Ann Arbor, Michigan 48109 Received November 5, 2004

This study was designed to elucidate the mechanisms underlying the formation of reactive oxygen species (ROS) by aminoglycoside antibiotics which may be causally related to the toxic side effects of these drugs to the kidney and the inner ear. ROS formation by aminoglycosides in vitro requires iron and the presence of polyunsaturated lipids as electron donors. Electron spray ionization mass spectrometry (ESI-MS) confirmed earlier observations that gentamicin strongly binds to L-R-phosphatidylinositol 4,5-bisphosphate (PtdIns(4,5)P2), a membrane lipid rich in arachidonic acid. Studies using lipid-coated membranes (PIP strips) further indicated that iron ions and gentamicin can simultaneously bind to phosphoinositides with at least one phosphate group on the inositol ring, suggesting the existence of ternary complexes among gentamicin, iron, and phospholipids. Peroxidation of PtdIns(4,5)P2 by ferrous ions significantly increased in the presence of gentamicin, and EI-MS measurements indicated that oxidative damage to PtdIns(4,5)P2 was accompanied by the release of arachidonic acid. Arachidonic acid also forms a ternary complex with Fe2+/3+-gentamicin, confirmed by ESI-MS, that reacts with lipid peroxides and molecular oxygen, leading to the propagation of arachidonic acid peroxidation.

Introduction Gentamicin belongs to the family of aminoglycoside antibiotics produced by multiple strains of soil actinomycetes, including Streptomyces and Micromonospora. Natural and semisynthetic aminoglycosides share similar structures and are usually composed of aminated cyclitol, streptamine, or 2-deoxystreptamine rings linked to different aminosugars. The amino groups, which characterize these drugs along with hydroxyl groups, determine many of their biological and chemical properties (1). The success of aminoglycosides as chemotherapeutic agents results from their broad antibacterial spectrum against most Gram-negative bacteria, rapid bactericidal action, and very low cost. Gentamicin is currently the antibiotic of first choice in many developing countries and is still widely used in industrialized countries for the treatment of serious bacterial infections. The antibacterial actions of aminoglycoside antibiotics include damage to the plasma membrane and binding to 30S and 16S ribosomal RNA, inhibiting polypeptide chain elongation and inducing misreading of the genetic code (2). However, therapy with aminoglycoside antibiotics has two major adverse side effects: damage to the proximal tubules of the kidneys (nephrotoxicity) and to the sensory cells of the inner ear, causing loss of hearing and balance (ototoxicity) (1). Insight into the molecular mechanism of aminoglycoside toxicity comes from in vivo studies demonstrating * To whom correspondence should be addressed. (V.L.P.) Phone: (734) 763-1519. Fax: (734) 936-7628. E-mail: [email protected]. (J.S.) Phone: (734) 763-3572. Fax (734) 764-0014. E-mail: [email protected]. † Kresge Hearing Research Institute, Department of Otolaryngology. ‡ Department of Chemistry. § Department of Biological Chemistry.

that loss of sensory cells increased in the presence of iron supplementation (3) and decreased with the co-administration of antioxidants or iron chelators (4-6) or the overexpression of the scavenging enzyme copper/zinc superoxide dismutase (7). This evidence is complemented by in vitro observations that aminoglycosides can catalyze the formation of reactive oxygen species (8, 9) in the presence of transition metals such as iron (10) and copper (11). Arachidonic acid can serve as an electron donor in the formation of reactive oxygen species by gentamicin and iron cell-free systems (8, 9), and lipid peroxidation occurs in the early stages of aminoglycoside-induced hearing loss in inner ear tissues (12). Thus, both in vivo and in vitro evidence suggests that aminoglycosides can interact with iron and lipids in the formation of reactive oxygen species and the expression of toxicity. However, the binding constants of gentamicin-metal complexes suggest that gentamicin itself cannot compete for transition metals with naturally occurring bioligands. Thus, although copper complexes are redox-active, they may only play a marginal role in the cellular toxicity of aminoglycoside antibiotics (11). Likewise, gentamicin forms iron complexes with a 1:2 metal to ligand molar ratio and octahedral geometry as well as potentially redox-active iron-gentamicin 1:1 complexes, but the stability constants of these complexes (10) are also inferior to those of complexes with naturally occurring bioligands such as citrate, amino acids, peptides, and proteins. There is an excellent correlation between the degree of ototoxicity of aminoglycoside antibiotics and their binding affinity toward polyphosphoinositides (13), and interactions with phospholipids are also considered to be a factor in the nephrotoxic potential of these drugs (14).

10.1021/tx0496946 CCC: $30.25 © 2005 American Chemical Society Published on Web 01/27/2005

358

Chem. Res. Toxicol., Vol. 18, No. 2, 2005

Lesniak et al.

Phosphoinositides are enriched in the 1-stearoyl-2arachidonoyl-sn-glycerol species and predominantly localized to the inner leaflet of plasma membranes. Although quantitatively minor membrane lipids, they can reach millimolar local concentrations (15). Phosphoinositides are functionally important as second messengers and as a source of cellular arachidonic acid whose release is stimulated by aminoglycosides (16). Since phospholipids and free fatty acids also bind iron (17, 18), the possibility arises that ternary complexes among aminoglycosides, iron, and lipids might exist which would facilitate the formation of reactive oxygen species. The present study explores this hypothesis using binding assays, ESI-MS1 and analyses of lipid peroxidation.

Experimental Procedures Chemicals. Gentamicin sulfate was purchased from Spectrum Chemical (New Brunswick, NJ). Piperazine-N,N′-bis(2ethanesulfonic acid) (PIPES), N,N′-diethylpiperazine (DEPP), 2(morpholino)ethanesulfonic acid (MES), 4-(N-morpholino)butanesulfonic acid (MOBS), N,N′-diethyl-N,N′-bis(sulfopropyl)ethylenediamine (DESPEN), and N,N,N′,N′-tetraethylethylenediamine (TEEN) buffers were obtained from GFS Chemicals Inc. (Powell, OH). Cyclohexane, choloroform, and methanol were purchased from Fisher Scientific (Chicago, IL). L-R-Phosphatidylinositol 4,5-bisphosphate sodium salt (PtdIns(4,5)P2; from bovine brain), arachidonic acid (from porcine liver), arachidonic acid methyl ester, naphthalene-2,3-dicarboxaldahyde (NDA), sodium cyanide, ferrous sulfate (FeSO4; heptahydrate), and ferric chloride (FeCl3; hexahydrate) were acquired from Sigma Chemical Co. (St. Louis, MO). 55FeCl3 radionuclide (2 mCi, 74 MBq) in 0.5 M HCl was purchased from Perkin-Elmer (Wellesley, MA). Interactions of Gentamicin and Iron Ions with Phospholipids. Binding of gentamicin and iron ions to phospholipids was assayed with PIP strips (P-6001, Echelon Biosciences Inc., Salt Lake City, UT) containing individual wells with 15 different phospholipids (Figure 1). The membranes were blocked with 3% (w/v) fatty acid-free bovine serum albumin in buffer [100 mM Tris chloride (pH 8), 150 mM NaCl, and 0.1% (v/v) Tween-20] for 1 h at room temperature. The strips were then washed five times for 5 min each in Tris/NaCl (100 mM Tris chloride (pH 7.5) + 150 mM NaCl) and incubated for 1 h at room temperature in the same solution additionally containing 100 µM gentamicin sulfate. The strips were then washed as before and incubated for 10 min in the same buffer containing 500 µM sodium cyanide; this was followed by the addition of 1 mL of an aqueous solution of naphthalene-2,3-dicarboxaldahyde (final concentration 500 µM) and an incubation for 30 min. Finally, the strip was washed as before and dried. Derivatized gentamicin was detected using a Typhoon 9400 high-performance laser scanning system (Amersham Biosciences Corp., Piscataway, NJ) in the fluorescence mode with an argon laser operating at 457 nm as the source of excitation. To evaluate ferric ion binding, the 55Fe3+ isotope (as 55FeCl3, 13 µM) was applied as described for gentamicin but under omission of the derivatization step. The dried strip was exposed for 12 h to a GP storage phosphor screen (Amersham Biosciences Corp.) which was scanned with the Typhoon 9400 system. Electron Spray Ionization Mass Spectrometry. Mass spectra were obtained with a Micromass LCT, a benchtop timeof-flight mass spectrometer with electrospray and APCI ionization modes connected to an HPLC pump with a Rheodyne loop injector (Waters Corp., Milford, MA). The spectrometer was operated in either positive or negative ion mode and controlled 1 Abbreviations: AA-ME, arachidonic acid methyl ester; CD, conjugated diene; ESI-MS, electron spray ionization mass spectrometry; PtdIns(4,5)P2, L-R-phosphatidylinositol 4,5-bisphosphate; ROS, reactive oxygen species.

Figure 1. Gentamicin interacts with both iron and phosphoinositides. The ability of gentamicin and iron ions, separately as well as in combination, to bind to a variety of phospholipids was analyzed using lipid-coated PIP strips as described in the Experimental Procedures. (A) Reference for the location of lipids: 1, lysophosphatidic acid; 2, lysophosphatidylcholine; 3, PtdIns; 4, PtdIns(3)P; 5, PtdIns(4)P; 6, PtdIns(5)P; 7, phosphatidylethanolamine; 8, phosphatidylcholine; 9, sphingosine1-phosphate; 10, PtdIns(3,4,)P2; 11, PtdIns(3,5)P2; 12, PtdIns(4,5)P2; 13, PtdIns(3,4,5)P3; 14, phosphatidic acid; 15, phosphatidylserine; 16, blank. (B) Incubation with 100 µM gentamicin followed by staining for gentamicin. (C) Incubation with 100 µM gentamicin and 55Fe3+ followed by radioautography for 55Fe. (D) Incubation with 100 µM gentamicin and Fe3+ followed by staining for gentamicin. by MassLynex 4.0 SP1 software. Nitrogen was applied as a nebulizing gas at a flow of 7 L/h. The desolvation temperature was 150 °C, and the gas flow was 543 L/h. Samples of 10 µL were injected into a solvent stream of 50% methanol. The scan range was m/z 50-3000. Gentamicin C2 chloride was obtained from commercial gentamicin C sulfate using an HPLC separation described previously (19). Seven different preparations were analyzed: (1) PtdIns(4,5)P2; (2) PtdIns(4,5)P2 + gentamicin C2; (3) PtdIns(4,5)P2 + FeCl3; (4) FeCl3 + gentamicin C2; (5) PtdIns(4,5)P2 + FeCl3 + gentamicin C2; (6) PtdIns(4,5)P2 + FeSO4; (7) PtdIns(4,5)P2 + FeSO4 + gentamicin C2. The samples contained 20% (v/v) methanol (to decrease surface tension) and had a total volume of 200 µL with the following final concentrations of reagents: PtdIns(4,5)P2, 10 µM; FeCl3, FeSO4, 20 µM; gentamicin C2, 20 µM. The pH was adjusted with sodium hydroxide. All samples were analyzed either immediately after mixing or after 1 h of incubation at 37 °C. In the study of the interaction of arachidonic acid with ferric ions and gentamicin, the samples were prepared by mixing stock solutions of gentamicin C2, arachidonic acid, and ferric chloride to obtain final concentrations of arachidonic acid of 100 µM, gentamicin C2 of 100 µM, and FeCl3 of 50 and 100 µM (for each sample containing FeCl3). The pH was adjusted with NaOH. Four different assay preparations were analyzed: (1) arachidonic acid; (2) arachidonic acid + gentamicin C2; (3) arachidonic acid + FeCl3; (4) arachidonic acid + FeCl3 + gentamicin C2. For all investigated systems, spectra were collected in both positive and negative ion modes. Conjugated Diene Assay. Arachidonic acid peroxidation was measured spectrophotometrically via the detection of conjugated dienes (20). Reaction mixtures containing arachidonic acid and MOBS buffer (pH 7.4) were vortexed and then emulsified using a Bransonic 12 ultrasonic emulsifier (Smith Kline Co., Shelton, CT) for 5 min prior to use. MOBS buffer was used in place of traditional pH buffers, such as sodium phosphate, due to its lower affinity to ferrous and ferric ions (21). Gentamicin (1 mM) was added to the emulsified preparation, and reactions were initiated by the addition of 100 µM

Free Radical Formation by Gentamicin and Iron

Chem. Res. Toxicol., Vol. 18, No. 2, 2005 359

FeSO4 or 100 µM FeCl3, final volume 200 µL. Incubations were carried out at 37 °C for 1 h and then stopped by the addition of 1 mL of chloroform/methanol (2:1, v/v). After the mixtures were vortexed and centrifuged for 5 min, the organic (lower) phase was removed and evaporated under a stream of N2 at 45 °C. The residues were dissolved by the addition of 200 µL of cyclohexane. Spectra between 200 and 500 nm were recorded on a Milton Roy Spectronic 3000 diode array spectrophotometer (Rochester, NY). Absorbance was read at 238 nm, and a molar extinction coefficient of 2.52 × 104 M-1 cm-1 was used to compute the concentration of conjugated dienes in the samples. Molecular Modeling. The energy-minimized model structures of Fe3+-gentamicin C2 complexes were obtained using the commercial software Insight II 3.0.0 molecular modeling system. Energy minimizations were carried out using the esff force field in the Discover 3 module and entailed 300 steps of minimization to remove initial strain, followed by a 1 ps dynamics simulation with a femtosecond time step, and finally a maximum of 1000 steps of energy minimization to obtain the final structure. All calculations were carried out with a distance-dependent dielectric field.

Results Binding of Gentamicin and Iron to Phospholipids. The binding of gentamicin and iron to phospholipids was examined on PIP strips (Figure 1). Gentamicin bound most strongly to PtdIns(3,4,5)P3, PtdIns(3,4)P2, PtdIns(4,5)P2, and PtdIns(3)P (panel B). 55Fe bound to all anionic phospholipids but not to phosphatidylethanolamine, phosphatidylcholine, and lysophosphatidylcholine regardless of the presence or absence of gentamicin (panel C). In the presence of iron, gentamicin remained mostly bound to PtdIns(3,4)P2 and PtdIns(4,5)P2 (panel D), suggesting the possibility of simultaneous interactions among phosphoinositides, iron, and gentamicin. The results were verified in three repetitions. Degradation of PtdIns(4,5)P2 by Iron-Gentamicin. ESI-MS of PtdIns(4,5)P2 yielded a major [M - H]2ion at m/z 522 and a minor [M - H]- at m/z 1046, indicating doubly and singly charged species of PtdIns(4,5)P2, respectively (Figure 2A). A triply charged anion expected at m/z 347.9 was not detected. This agrees with evidence that the charge of PtdIns(4,5)P2 does not equal its number of phosphate groups (22). Incubation of PtdIns(4,5)P2 for 1 h at 37 °C in medium without iron and gentamicin did not affect the spectrum. Gentamicin C2 yielded signals at m/z 464, 322, 160, and 142 regardless of the presence or absence of PtdIns(4,5)P2 (Figure 2B). The signal at m/z 464 corresponds to monoprotonated gentamicin C2, and a doubly charged species was detected at m/z 233. Fragments of the molecule are indicated by signals at m/z 322 (gentamicin C2 minus the purpurosamine ring), m/z 160 (garosamine or 2-deoxystreptamine), and m/z 142 (purpurosamine). An additional signal at m/z 756 was only detected in the mixture of gentamicin C2 and PtdIns(4,5)P2 (1:2 molar ratio) and may represent a doubly charged PtdIns(4,5)P2-gentamicin C2 adduct. The presence of ferrous (FeSO4) ions in mixtures of PtdIns(4,5)P2 and gentamicin C2 produced additional signals derived from PtdIns(4,5)P2 (Figure. 3C). Among these, [M - H]- at m/z 303 and 283 indicated free arachidonic and stearic acid, respectively, in agreement with the fact that these are the two major fatty acids in the PtdIns(4,5)P2 preparation used in these experiments.

Figure 2. Degradation of PtdIns(4,5)P2 in the presence of iron and gentamicin. The TOF ESI-MS spectra were obtained from single compounds or mixtures in the positive mode to detect gentamicin or negative mode to detect PtdIns(4,5)P2. All samples were incubated for 1 h at 37 °C and pH 7.4 prior to analysis. Key: (A) 10 µM PtdIns(4,5)P2; (B) 10 µM PtdIns(4,5)P2 + 20 µM gentamicin, recorded in positive mode; (C) 10 µM PtdIns(4,5)P2 + 20 µM FeSO4 + 20 µM gentamicin, recorded in negative mode.

In contrast, the presence of ferric ions (FeCl3) did not influence the ESI-MS spectra of either PtdIns(4,5)P2 or gentamicin C2 or their combination (data not shown). Peroxidation of PtdIn(4,5)P2 and Arachidonic Acid. To probe the interactions with PtdIns(4,5)P2 further, we examined whether the lipid was oxidized in the presence of gentamicin and iron (Figure 3). Conjugated dienes were indeed formed in the presence of ferrous, but not ferric, ions, and the Fe2+-mediated reaction was further stimulated by the addition of gentamicin. When free arachidonic acid was used as the substrate in the peroxidation assays, both ferrous and ferric ions caused a small amount of diene formation (Figure 4). The amount of conjugated diene was higher than in the incubations with PtdIns(4,5)P2, but differences in substrate concentrations easily account for this apparent discrepancy. When diene formation is considered as a percent of potentially available substrate, the levels of reaction product are of a similar magnitude. The rate of the reaction significantly increased upon addition of 1 mM gentamicin, and the kinetics showed distinct differences between the effects of Fe2+ and Fe3+. The presence of Fe2+ resulted in a linear rate of diene formation for about 15 min, whereupon the reaction asymptotically approached an end point. In contrast, diene formation in the presence of Fe3+ had a lag phase of about 10 min, after which the reaction proceeded to the same end point as with Fe2+. The pH profile of arachidonic acid peroxidation in the presence and absence of the combinations of Fe2+, Fe3+, and gentamicin (Figure 5) showed that arachidonic acid alone underwent a small amount of peroxidation at low pH (Figure 5A). Increasing the pH resulted in a decrease of peroxidation, and at pH 7.4 and higher the formation

360

Chem. Res. Toxicol., Vol. 18, No. 2, 2005

Figure 3. Peroxidation of PtdIns(4,5)P2. Reaction mixtures were incubated for 1 h at 37 °C in 5 mM MOBS buffer at pH 7.4, followed by an assay for conjugated dienes as described in the Experimental Procedures. The incubations (200 µL) contained (A) 20 µM PtdIns(4,5)P2, (B) 20 µM PtdIns(4,5)P2 + 40 µM gentamicin, (C) 20 µM PtdIns(4,5)P2 + 20 µM FeSO4, (D) 20 µM PtdIns(4,5)P2 + 20 µM FeSO4 + 40 µM gentamicin, (E) 20 µM PtdIns(4,5)P2 + 20 µM FeCl3, and (F) 20 µM PtdIns(4,5)P2 + 20 µM FeCl3 + 40 µM gentamicin. The values are means ( SD of six independent experiments. Differences are significant (p < 0.05; one-way ANOVA with Student-Newman-Keuls post hoc test) except (A) vs (B) and (E) vs (F).

Figure 4. Gentamicin increases arachidonic acid peroxidation by ferric and ferrous ions. The kinetics of arachidonic acid peroxidation in the presence of Fe2+ or Fe3+ and gentamicin were followed by a conjugated diene assay as described in the Experimental Procedures. The reaction mixtures (200 µL) were incubated at 37 °C in 50 mM MOBS buffer at pH 7.4 and contained (b) 9 mM arachidonic acid + 100 µM FeCl3, (O) 9 mM arachidonic acid + 1 mM gentamicin + 100 µM FeCl3, (9) 9 mM arachidonic acid + 100 µM FeSO4, and (0) 9 mM arachidonic acid + 1 mM gentamicin + 100 µM FeSO4. The mean of the conjugated diene concentrations in the control samples (9 mM arachidonic acid in MOBS) was subtracted from each value. The combination of arachidonic acid with gentamicin alone did not show significant conjugated diene formation; the values fell into the experimental error. The values are means ( SD of three independent experiments.

of conjugated diene in incubations with arachidonic acid alone fell into the experimental error. The rate and pH profile of this oxidation was not affected by the addition of gentamicin. However, peroxidation was significantly enhanced by iron ions in a pH-dependent manner. With either ferrous or ferric ions, around 300 µM CD (conjugated diene) was detected after a 1 h incubation at pH 3.4. The level of conjugated diene gradually decreased with increasing pH values.

Lesniak et al.

Figure 5. Gentamicin-mediated arachidonic acid peroxidation is pH-dependent. The reaction mixtures (200 µL) were incubated at 37 °C for 1 h at different pH values, and arachidonic acid peroxidation was followed by the conjugated diene assay as described in the Experimental Procedures. The pH was varied by using the following buffers: PIPES (pH 3.4), DEPP (pH 4.4), MES (pH 5.4 and 6.4), MOBS (pH 7.4), DESPEN (pH 8.4 and 9.4), or TEEN (pH 10.4). Values are means + SD of six experiments. (A) Arachidonic acid peroxidation in the presence of gentamicin alone or iron alone: (*) 9 mM arachidonic acid only; (1) 9 mM arachidonic acid + 1 mM gentamicin; (b) 9 mM arachidonic acid + 100 µM FeCl3; (9) 9 mM arachidonic acid + 100 µM FeSO4. (B) Stimulation by gentamicin of iron-dependent arachidonic acid peroxidation. The values represent the differences between corresponding incubations with and without gentamicin in the presence of arachidonic acid and FeCl3 (b) or FeSO4 (9).

The stimulation of iron-dependent arachidonic acid peroxidation by gentamicin (Figure 5B) was also pHdependent. Its maximum, however, was different from the maximum of peroxidation with iron only. Gentamicin had its largest stimulatory effect between pH 6.4 and pH 8.4. To assess contribution of the carboxyl group to arachidonic acid peroxidation, the oxidation of arachidonic acid methyl ester was followed by the conjugated diene assay (Figure 6). Diene formation in the presence of Fe3+ alone or in combination with gentamicin remained low after 1 h at 37 °C. The presence of Fe2+ alone or in combination with gentamicin increased diene formation, but the level (50 to 60 µM) remained markedly below that of the diene formation from arachidonic acid (Figure 4). Complexes among Gentamicin, Iron, and Arachidonic Acid. Since arachidonic acid is an iron chelator (17), we next probed for interactions among arachidonic acid, gentamicin, and iron using TOF ESI-MS (Figure 7). A solution of arachidonic acid (Figure 7A, inset) was characterized by signals at m/z 327 and 342 which correspond to adducts with sodium and potassium, respectively. Addition of 0.5 molar equivalent of ferric ions resulted in the appearance of a novel series of peaks at m/z near 698 consistent with an adduct composed of arachidonic acid, Fe3+, and K+ in a 2:1:1 ratio (Figure 7A). In the system containing arachidonic acid, Fe3+, and gentamicin C2 (in a 1:1:1 molar ratio), the peaks around m/z 698 disappeared and a new signal at m/z 411 emerged (Figure 7B). This signal is consistent with a ternary complex of arachidonic acid-Fe3+-gentamicin C2 detected as a doubly charged molecule.

Free Radical Formation by Gentamicin and Iron

Figure 6. Arachidonic acid methyl ester is not a good substrate for peroxidation. Peroxidation of arachidonic acid methyl ester (AA-ME) was followed by the formation of conjugated dienes as in Figures 4 and 5 and as described in the Experimental Procedures. The reaction mixtures (200 µL) were incubated for 1 h at 37 °C in 50 mM MOBS buffer at pH 7.4 and contained (A) 9 mM AA-ME + 1 mM gentamicin, (B) 9 mM AA-ME + 100 µM FeCl3, (C) 9 mM AA-ME + 100 µM FeCl3 + 1 mM gentamicin, (D) 9 mM AA-ME + 100 µM FeSO4, (E) 9 mM AAME + 100 µM FeSO4 + 1 mM gentamicin. The values are means ( SD of three independent experiments. The mean of the conjugated diene concentrations in the control samples (AA-ME only in MOBS) was subtracted from each value. Differences are statistically significant (p < 0.05; one-way ANOVA with StudentNewman-Keuls post hoc test) except (A) vs (C).

Figure 7. Gentamicin forms ternary complexes with Fe3+ and arachidonic acid. TOF ESI-MS spectra were obtained as in Figure 2 and described in the Experimental Procedures. Gentamicin in this assay was gentamicin C2 obtained as described earlier (19). All spectra were recorded in positive mode. Incubations consisted of (A) 100 µM arachidonic acid + 50 µM FeCl3 (inset, 100 µM arachidonic acid alone) and (B) 100 µM arachidonic acid + 50 µM FeCl3 + 100 µM gentamicin C2 (inset, 100 µM Gentamicin C2 alone).

Discussion These studies indicate that gentamicin can form ternary complexes with iron ions and phosphoinositides or free arachidonic acid. The resulting complexes promote lipid peroxidation, a reaction that may underlie the molecular mechanism of the toxic side effects of aminoglycoside antibiotics. Interactions between the individual components of the proposed ternary complex have long been demonstrated. Aminoglycosides bind to polyphosphoinositides with high affinity (23, 24), visualized here in PIP strips and by ESIMS. The PIP strips also show that the binding is rather

Chem. Res. Toxicol., Vol. 18, No. 2, 2005 361

specific for phosphoinositides with at least one monoesterified phosphate group on the inositol ring. The signal at m/z 756 in ESI-MS confirms the formation of a stable adduct between PtdIns(4,5)P2 and gentamicin. In contrast to the strong affinity of aminoglycosides to phosphoinositides (23, 24), binding constants of complexes between the drugs and transition-metal ions such as copper(II) or iron(II/III) are relatively weak. Although the combination of iron and gentamicin promotes lipid peroxidation in vitro (8, 9), aminoglycosides themselves should not be able to compete effectively with intracellular ligands for the “free” iron pool under physiological conditions (10, 11). On the other hand, various lines of evidence suggest that gentamicin-iron interactions should also exist in vivo: supplemental iron exacerbates aminoglycoside toxicity (3), and iron chelators attenuate it (4-6). The present data resolve this paradox by showing that the redox-active species may be a ternary complex among gentamicin, iron, and lipids. To form complexes in vivo, the potential chelators have to compete for the free iron, i.e., the “low molecular mass iron pool” available for cellular processes and sequestered by bioligands such as phosphates and amino acids (25). Membrane lipids and fatty acids readily bind iron (18), and the phosphate groups of PtdIns(4,5)P2 in particular should form a bidentate complex of high affinity. While all negatively charged phospholipids potentially bind iron, the ability to interact simultaneously with both iron and gentamicin is largely limited to PtdIns(3,4)P2 and PtdIns(4,5)P2. While we only followed PtdIns(4,5)P2 in detail, the structural similarities between phosphoinositides suggest that complexes with PtdIns(3,4)P2 would have similar characteristics. At a physiological pH of 7.4, the 1-, 3′′-, and 6-amino groups of gentamicin are positively charged and aminoglycoside-phosphoinositide binding is likely based on strong electrostatic interaction with the phosphate groups of PtdInsP2 (26). This binding mode could change in the presence of iron, but our results clearly indicate that iron does not prevent aminoglycoside interactions with mono- and diesterified phosphates. This finding suggests that iron-phospholipid complexes remain negatively charged, allowing simultaneous electrostatic interactions with gentamicin. Conversely, gentamicin does not displace Fe3+ from the lipids with the possible exception of PtdIns(3,4,5)P3. At physiological pH the 3- and 2′-NH2 groups of gentamicinswhich are not involved in binding to phosphoinositidessare deprotonated and may participate in the coordination sphere of iron ions. Iron chelation by nitrogen donors influences its redox potential and may explain the higher oxidative activity of the system containing gentamicin. The sum of these results implies that aminoglycoside antibiotics have one binding domain for phosphoinositides and another for iron as we hypothesized earlier (1). Such a ternary complex would provide an activated iron in proximity of arachidonic acid at the 2′-position of PtdIns(4,5)P2 that then can be released upon oxidative damage or oxidized to trigger the formation of ROS. In addition to accelerating the iron-mediated peroxidation of PtdIns(4,5)P2, gentamicin can also stimulate iron-mediated arachidonic acid oxidation. The small difference in the magnitude of stimulation should not be surprising because the iron exists in complexes with different coordination modes that may result in dissimilar redox potentials and consequent ROS formation. In the complex with arachidonic acid the carboxyl group

362

Chem. Res. Toxicol., Vol. 18, No. 2, 2005

Lesniak et al.

Figure 8. Structure of the proposed arachidonic acid-iron-gentamicin complexes. Energy minimization models and schematic representation of the proposed ternary complexes among gentamicin, iron, and arachidonic acid. The side chain of arachidonic acid does not influence the coordination mode of iron and was omitted for clarity of the illustrations. Key: (A) arachidonic acid-Fe-H4G at low pH (H4G ) gentamicin molecule with deprotonated 3-NH2 group (pKa ) 5.75); (B) arachidonic acid-Fe-H3G at higher pH (H3G ) gentamicin molecule with deprotonated 3-NH2 and 2′-NH2 (pKa ) 7.35) groups.

appears to be crucial for the peroxidation reaction since the arachidonic acid methyl ester is much less sensitive to oxidation. This difference in reactivity is in agreement with the fact that free fatty acids bind iron via their carboxyl groups (17). The lack of peroxide formation from the methyl ester also supports the contention that ferrous and ferric ions alone do not form a strongly redox-active species with gentamicin alone, corroborating the weak interactions in Fe2+-gentamicin 1:1 and octahedral 1:2 complexes with stability constants 0.002 and 0.2 M2, respectively (10). A ternary complex among arachidonic acid, gentamicin, and iron ions was also recognized by ESI-MS. Upon addition of gentamicin C2, the set of signals related to the Fe3+-(arachidonic acid)2 complex disappeared and a novel signal appeared at m/z 411. This is best explained by the displacement of one arachidonic acid from the arachidonic acid-Fe3+-arachidonic acid complex by gentamicin and the formation of an arachidonic acid-Fe3+gentamicin C2 complex. Although such a peak may result from aggregation of analytes during the ionization process in ESI-MS, it is the sum of our studies that supports this signal as a genuine ternary complex. This interpretation is also well supported by the pH profile of arachidonic acid peroxidation promoted by iron ions and gentamicin. The drug begins to stimulate the peroxidation of arachidonic acid around pH 5, and the maximal stimulation occurs between pH 6.5 and pH 8.5. This region corresponds to the deprotonation of the 3- and 2′amino groups of gentamicin with pKa values of 5.75 and 7.35, respectively (19), which can serve as electron donors in the coordination of iron ions. The energy-minimized models of the arachidonic acidFe3+-gentamicin C2 complexes using the above considerations, the protonation pattern of gentamicin, and the proposed metal binding side in this drug (10, 19) suggest two pH-dependent complexes. The binding mode of the iron complex with arachidonic acid and gentamicin at low

pH involves two oxygen atoms donated by the carboxyl group of arachidonic acid and two electron donors from the gentamicin molecule, namely, the 3-NH2 group (pKa ) 5.75) and the glycosidic oxygen connecting the 2-deoxystreptamine and purpurosamine rings, located in equatorial positions (Figure 8A). Upon transition to a higher pH the 2′-amino group of the purpurosamine moiety undergoes deprotonation (pKa ) 7.35) and assumes an appropriate orientation to participate in the coordination sphere in the axial position as shown in Figure 8B. Under physiological conditions around pH 7.4, an equilibrium of these two forms will exist. Both ferric and ferrous ions will form complexes with the same coordination pattern. While the models depict gentamicin C2, it should be noted that all components of the commercial gentamicin C complex can engage in the same coordination mode. Gentamicin C1a, C2, C2a and C1 differ only in the methylation of the 6′ unit of the purpurosamine ring, and this moiety does not participate in metal binding. Arachidonic acid-Fe-H4G (H4G, gentamicin molecule with a deprotonated 3-NH2 group) and arachidonic acid-FeH3G (H3G, gentamicin molecule with deprotonated 3-NH2 and 2′-NH2 groups) complexes possess two or one H2O molecule, respectively, in the axial positions that can be readily displaced by molecular oxygen to initialize formation of reactive oxygen species. The initial steps of lipid peroxidation can be expected to follow established patterns of iron-dependent oxidations that can be promoted by chelators (27):

arachidonic acid-Fe2+-gentamicin + O2 T (arachidonic acid-Fe2+-O2-gentamicin T arachidonic acid-Fe3+-O2•--gentamicin) T arachidonic acid-Fe3+-gentamicin + O2•- (1) Superoxide anion would then dismutate to hydrogen

Free Radical Formation by Gentamicin and Iron

peroxide, which, in turn, interacts with complexed ferrous ion to produce hydroxyl radicals. Hydroxyl radicals are sufficiently reactive to abstract one of the bisallylic hydrogens of arachidonic acid and initialize peroxidation. The carbon radical thus formed can be stabilized through molecular rearrangements accompanied by conjugated diene formation. Alternatively, the carbon radical may interact with molecular oxygen to form a peroxyl radical (arachidonic acid-OO-) that can further propagate lipid peroxidation by abstracting a hydrogen atom from a methylene group of arachidonic acid. In an alternative but minor route in the in vitro system, arachidonic acid-Fe2+-gentamicin could react with lipid peroxides which are present as contaminants or being formed during sonication of the incubation mixtures, leading to alkoxyl radical production. The alkoxyl radical is also capable of abstracting bisallylic hydrogen from arachidonic acid to form a carbon radical so that the chain reaction of lipid peroxidation can continue. An interaction with preformed lipid peroxides is also likely for the system initially composed of arachidonic acid, Fe3+, and gentamicin which promotes arachidonic acid peroxidation after a lag phase. Ferric complexes are more inert in the reactions with lipid peroxides than ferrous complexes, but the arachidonic acid-Fe3+gentamicin complex can be reduced by lipid peroxides accompanied by peroxyl radical formation. ROS are now generally accepted as crucial participants in the mechanisms of aminoglycoside ototoxicity and nephrotoxicity (1), but the details of the formation of ROS remain to be elucidated. Aside from a direct stimulation of drug-initiated ROS formation and lipid peroxidation, potential pathways include the stimulation of ROSgenerating enzymes such as nitric oxide synthase or NADPH oxidase. Nitric oxide, for example, has been suggested to be involved in hair cell death induced by gentamicin application to vestibular tissues (28), and small Rho-GTPases that can stimulate NADPH oxidase are involved in the response of cell cultures to aminoglycosides (29). How such acute effects of drug treatment relate to the chronic (and clinically relevant) ototoxicity of aminoglycosides remains to be established, and it is possible that they differ. In support of such a notion, nitrotyrosine, a marker for peroxynitrite damage, was not found in animals chronically treated with kanamycin while lipid peroxidation was present (12). On the other hand, there is no reason to assume that drugs such as aminoglycosides should only exert their toxic actions through a single mechanism. Rather, multiple sites of drug attack seem reasonable, depending on the system under investigation, the concentration of the drug used, and the timing of the drug administration. Notwithstanding the potential contributions of additional pathways, our results allow us to propose a mechanism of the formation of reactive oxygen species and lipid peroxides by aminoglycoside antibiotics. Phosphoinositides serve as intracellular binding sites for aminoglycosides in vivo, providing phosphate groups (from the phospholipids) and amino groups and glycosidic oxygen (from the drug) as ligands for iron. The complexed iron forms a redox center, facilitating oxidative damage of phosphoinositides, triggering oxidation and possible release of arachidonic acid. Arachidonic acid, which also forms complexes with iron ions and aminoglycosides, can further promote oxidative reactions.

Chem. Res. Toxicol., Vol. 18, No. 2, 2005 363

Acknowledgment. This work was supported by Research Grant RO1 DC-03685 and Research Core Center Grant P30 DC-05188 from the National Institute on Deafness and Other Communicational Disorders, National Institutes of Health.

References (1) Forge, A., and Schacht, J. (2000) Aminoglycoside Antibiotics, Audiol. Neuro-Otol. 5, 3-22. (2) Blanchard, S., Fourmy, D., Eason, R., and Puglisi, D. (1998) rRNA chemical groups required for aminoglycoside binding, Biochemistry 37, 7716-77241. (3) Conlon, B. J., and Smith, D. W. (1998) Supplemental iron exacerbates aminoglycoside ototoxicity in vivo, Hear. Res. 115, 1-5. (4) Song, B., Anderson, D., and Schacht, J. (1997) Protection from gentamicin ototoxcity by iron chelators in guinea pigs in vivo, J. Pharmacol. Exp. Ther. 282, 369-377. (5) Song, B., and Schacht, J. (1996) Variable efficacy of radical scavengers and iron chelators to attenuate gentamicin ototoxicity in guinea pig in vivo, Hear. Res. 94, 87-93. (6) Sha, S. H., and Schacht, J. (2000) Antioxidants attenuate gentamicin-induced free-radical formation in vitro and ototoxicity in vivo, D-methionine is a potential protectant, Hear. Res. 142, 3440. (7) Sha, S., Zajic, G., Epstein, C., and Schacht J. (2001) Overexpression of copper/zinc-superoxide dismutase protects from kanamycin-induced hearing loss, Audiol. Neuro-Otol. 6, 117-123. (8) Priuska, E., and Schacht, J. (1995) Formation of free radicals by gentamicin and iron and evidence of an iron/gentamicin complex, Biochem. Pharmacol. 50, 1749-1752. (9) Sha, S. H., and Schacht, J. (1999) Formation of reactive oxygen species following bioactivation of gentamicin, Free Radical Biol. Med. 26, 341-347. (10) Priuska, E., Clark-Baldwin, K., Pecoraro, V., and Schacht, J. (1998) NMR-studies of iron-gentamicin complexes and the implications for aminoglycoside toxicity, Inorg. Chim. Acta 273, 8591. (11) Lesniak, W., Harris, W. R., Yudenfreud-Kraviz. J., Schacht, J., and Pecoraro, V. L. (2003) Solution chemistry of copper(II)gentamicin complexes, Relevance to metal-related aminoglycoside toxicity, Inorg. Chem. 42, 1420-1429. (12) Jiang, H., Sha, S.-H., and Schacht, J. (2005) The NF-κB pathway protects cochlear hair cells from aminoglycoside-induced ototoxicity, J. Neurosci. Res. (in press). (13) Lodhi, S., Weiner, N., Mechigian, I., and Schacht, J. (1980) Ototoxicity of aminoglycosides correlated with their action on monomolecular films of polyphoshpoinositities, Biochem. Pharmacol. 29, 597-601. (14) Schanck, A., Brasseur, R., Mingeot-Leclercq, M. P., and Tulkens, P. M. (1992) Interactions of aminoglycosides with phosphatidylinositol, a 15N nuclear magnetic resonance study, Magn. Reson. Chem. 30, 11-15. (15) Stephens, L. R., Jackson, T. R., and Hawkins, P. T. (1993) Agoniststimulated synthesis of phosphatidylinositol(3,4,5)-trisphosphate: a new intracellular signaling system? Biochim. Biophys. Acta 1179, 27-75. (16) Nakashima, S., Tohmatsu, T., Shirato, L., Takenaka, A., and Nozawa, Y. (1987) Neomycin is a potent arachidonic acid release in human platelets, Biochem. Biophys. Res. Commun. 146, 820826. (17) Nalini, S., and Balasubramanian K. A. (1993) Studies on iron binding by free fatty acids, Indian J. Biochem. Biophys. 30, 224228. (18) Simpson, R. J., and Peters, T. J. (1987) Iron-binding lipids of rabbit duodenal brush-border membrane, Biochim. Biophys. Acta 898, 181-186. (19) Lesniak, W., McLaren, John.,Harris, W. R., Pecoraro, V. L., and Schacht, J. (2003) An isocratic separation of underivatized gentamicin components, 1H NMR assignment and protonation pattern, Carbohydr. Res. 338, 2853-2862. (20) Buege, J., and Aust, S. (1978) Microsomal lipid peroxidation, Methods Enzymol. 52, 302-310. (21) Kandegedara, A., and Rorabacher, D. (1999) Noncomplexing tertiary amines as “better” buffers covering the range of pH 3-11. Temperature dependence of their acid dissociation constants, Anal. Chem. 71, 3140-3144. (22) Hsu, F., and Turk, J. (2000) Characterization of phosphatidylinositol, phosphatidylinositol-4 phosphate, phosphatidylinositol-

364

(23) (24) (25) (26)

Chem. Res. Toxicol., Vol. 18, No. 2, 2005 4,5-bisphosphate by electrospray ionization tandem mass spectrometry, a mechanistic study, J. Am. Soc. Mass Spectrom. 11, 986-999. Schacht, J. (1979) Isolation of an aminoglycoside receptor from guinea pig inner ear tissues and kidney, Arch. Oto-RhinoLaryngol. 224, 129-134. Brasseur, R. (1990) Molecular description of biological membranes by computer aided conformational analysis, CRC Press, Boca Raton, FL. Symons, M. C. R., and Gutteridge, J. M. C. (1998) Free Radical and Iron: Chemistry, Biology and Medicine, pp 78-79, Oxford University Press, Oxford. Schacht, J., Lodhi, S., and Weiner, N. D. (1977) Effects of neomycin on polyphosphoinositides in inner ear tissues and

Lesniak et al. monomolecular films, in Membrane Toxicity (Miller, M. W., and Shamoo, A. E., Eds.) pp 191-208, Plenum Press, New York. (27) Stadtman, E. R. (1993) Fenton chemistry, Annu. Rev. Biochem. 62, 797-821. (28) Takumida, M., and Anniko, M. (2001) Nitric oxide in guinea pig vestibular sensory cells following gentamicin exposure in vitro, Acta Oto-Laryngol. 121, 346-350. (29) Bodmer, D., Brors, D., Pak, K., Gloddek, B., and Ryan, A. (2002) Rescue of auditory hair cells from aminoglycoside toxicity by Clostridium difficile toxin B, an inhibitor of the small GTPases Rho/Rac/Cdc42, Hear. Res. 172, 81-86.

TX0496946