Article pubs.acs.org/est
Proton and Iron Binding by the Cyanobacterial Toxin Microcystin-LR Annaleise R. Klein,*,† Darren S. Baldwin,‡ and Ewen Silvester† †
Department of Environmental Management and Ecology (DEME), La Trobe University, Albury-Wodonga Campus, Victoria, Australia, 3690 ‡ Murray Darling Freshwater Research Centre (MDFRC), CSIRO Land and Water, La Trobe University, Albury-Wodonga Campus, Victoria, Australia, 3690 S Supporting Information *
ABSTRACT: Microcystins (MCs) are a group of hepatotoxins produced by cyanobacteria that have not had their functional role or the environmental factors that trigger production clearly determined. One suggestion is that microcystins are siderophores (i.e., ligands with an extremely high affinity with iron, typically with stability constants substantially greater than 1025). In this work, we explore proton and iron binding with microcystin-LR (MC-LR). Using UV−visible spectroscopy and a HPLC peak retention time-based method, the two acid dissociation constants associated with the carboxylic groups of MC-LR were determined to be: pKa1 = 2.17 and pKa2 = 3.96. Cyclic voltammetry provides evidence for the formation of at least two FeIII-MC-LR complexes, with the FeIII reduction peak significantly shifted to more reducing potentials in the presence of MC-LR. These complexes have been interpreted as a rapidly formed initial complex (Complex 1) and a more stable, and slower forming, Complex 2. The stability constant for FeIII-MC-LR (Complex 2) was estimated to be approximately 1013 in 60% v/v MeOH/water at 0.1 M ionic strength. The electrochemical experiments provide no evidence for the formation of a complex between Fe2+ and MC-LR. Given that most MC-LR is released only upon cell lysis, and coupled with the moderate strength of the stability constant with FeIII determined in this study, it appears unlikely that that MC-LR is an extracellular siderophore. If MC-LR is involved in iron regulation in cyanobacteria, it is more likely as a shuttle for iron across the cell membrane or in intracellular processes.
■
INTRODUCTION Microcystins (MC) are a class of hepatotoxins produced by a number of cyanobacteria (blue-green algae) including Microcystis, Planktothrix, Anabaena, and Nostoc;1 with MC-LR the most common variant out of more than 80 currently identified.2 Microcystin is toxic to both humans and domestic stock, and represents a threat to water supplies worldwide.3 Despite being widely studied, the functional role of MC and the environmental factors that trigger MC production remain elusive (for a review on the biological role of hepatotoxin biosynthesis and regulation in cyanobacteria refer to Ginn et al.4 and references cited therein). The prediction that mcyH in the microcystin gene cluster functions as a transporter gene may indicate an extracellular role for MC,5 possibly as an intercellular intraspecies signaling chemical.6 However, given the observation that significant levels of MC only occur following cell lysis (e.g., Jones and Orr7) an intracellular function is more commonly assumed, with MC production shown to respond to factors such as oxidative stress, light, and iron limitation.8−13 Correspondingly, the ferric uptake regulator (Fur) has been found to bind to regions of the MC gene cluster. With Fur responding to many factors such as iron availability, light, oxidative stresses, acidic stresses, the inter regulation of iron and nitrogen metabolisms, and the homeostasis of other metals (e.g., Zn, Mg, and Ni);10 a simple correlation between either iron, light, or oxidative stress, and MC production/functionality would appear unlikely. © 2013 American Chemical Society
While a more general role for MC is becoming more commonly suggested,14 the speculation of Utkilen and Gjomle15 that one of the functions of MC is as an intracellular siderophore for the protection against products from the Fenton reaction remains plausible. Despite MC lacking typical siderophoric binding sites (hydroxamate, catecholate, and αhydroxy carboxylic acid16), the structure still contains functional groups which are able to bind to cations (e.g., amino, carboxyl, nitrogen, and oxygen), with a ring structure that is ordinarily favorable for chelation.13,17,18 Indeed, a number of MC variants have been shown to form moderately strong complexes with a number of cations, including Cu2+, Zn2+, Pb2+, Cd2+, and Hg2+;17−20 however, there is only limited evidence that MC can actually bind iron. An Fe2+−MC-LR complex was identified in a study using cryospray ionization Fourier transform ion cyclotron resonance mass spectroscopy (CSIFTIC-MS),19 but to date the stability constant of any Fe−MC complex has yet to be reported. In the current study we explore proton and Fe binding by MC-LR. The first two carboxylic acid dissociation constants of MC-LR (required for the subsequent calculation of iron−MCLR stability constants) were determined using UV−visible Received: Revised: Accepted: Published: 5178
January 30, 2013 April 11, 2013 April 15, 2013 April 15, 2013 dx.doi.org/10.1021/es400464e | Environ. Sci. Technol. 2013, 47, 5178−5184
Environmental Science & Technology
Article
range after mixing of 1.93 ≤ swpH ≤ 7.02. The fitted pKa values were optimized by manually adjusting the retention times of the three protonation states of MC-LR, as well as the carboxylic acid dissociation constants (pKa1, and pKa2), to minimize the sum of squares. The extracted carboxylic acid pKa values were corrected for solvent and ionic strength according to the methods of Ruiz et al.22 and Rived et al.,23 using citrate pKa values in 60% v/v MeOH/water from Subirats et al.24 to calculate citric acid dissociation for ionic strength correction (Text S1; SI). UV−visible Spectroscopic Determination of MC-LR Carboxylic Acid pKa Values. An initial MC-LR solution was prepared in a jacketed reactor (25 °C; stirred) at approximately 10 μM by ∼200-fold dilution of MC-LR stock (1.875 mM in 100% MeOH) in 0.1 M NaCl. The pH was adjusted over the range 1.66 < swpH < 6.03 by addition of either NaOH solutions (0.5−0.005 M) or 0.1 M HCl, prepared in 0.1 M NaCl (chloride common ion fixed at 0.1 M), using a 721 NET Titrino controlled with TiNET software. The reactor solution was pumped to a quartz flow cell using a peristaltic pump (Ismatec) with spectra recorded after each pH adjustment (5 min equilibration) over the wavelength range 200−340 nm using a Varian Cary 1E spectrophotometer controlled by Varian Cary WinUV (v4.10) software. After correction for dilution and background, spectra were analyzed by multivariate curve resolution (MCR) techniques using Unscrambler X v10.1 (CAMO software AS, Oslo, Norway) to determine component spectra and relative concentrations. pKa values for acid dissociation constants were fitted by least-squares minimization using GRFit.25 Cyclic Voltammetry of MC-LR. Electrochemical experiments were conducted at a relatively high MC-LR concentration of 0.3 mM. All experiments were conducted in ≥60% v/ v MeOH/water to prevent adsorption of MC-LR to the reactor and electrodes,26 and inhibit the formation of an unidentified precipitate that, in preliminary experiments, occurred when iron was added to MC-LR in near pure aqueous solutions. All experiments were performed using 0.1 M NaNO3 as the background electrolyte, and at a constant temperature of 25 °C. Cyclic voltametric (CV) studies were carried out using an Autolab PGSTAT12 potentiostat (Eco Chemie) controlled with NOVA 1.7 software. The electrochemical cell consisted of a standard 3-electrode configuration with a platinum rod counter electrode (Metrohm), a Ag/AgCl (3 M KCl; Metrohm) reference electrode, and a boron-doped diamond (BDD) working electrode (Windsor Scientific). A BDD working electrode was chosen in favor of more tradition materials (e.g., Pt, glassy carbon) as BDD has negligible adsorption of organic compounds/proteins;27 a previous study found that MC-LR adsorbed onto a glassy carbon working electrode.28 Air was excluded from the solution both by purging with (prior to recording CVs) and providing a blanket of high purity argon that had passed through three traps containing ∼10% H2SO4, ∼ 10% NaOH with ∼1% pyrogallol, and ∼60% v/v MeOH/water with 0.1 M NaNO3. Electrochemical measurements were conducted in solutions of: background electrolyte, Fe2+ in background electrolyte, MCLR in background electrolyte as well as mixtures of Fe2+ and MC-LR in background electrolyte. Cyclic voltammograms were recorded in these solutions at pH intervals across the range 3 < s −1 wpH < 8 (from low to high) at scan rates of 10 and 100 mV s . s At the specific wpH values of 3, 4, and 6, CVs were recorded at scan rates of 100, 50, 20, 10, and 5 mV s−1. Electrochemical
spectroscopy and HPLC. The Fe binding properties of MC-LR were determined by cyclic voltammetry using a boron-doped diamond (BDD) electrode.
■
EXPERIMENTAL METHODS Nomenclature for pH Measurement in Mixed Solvent Systems. In accordance with the IUPAC convention the measured pH in a mixed solvent system is referred to as yxpH, where x refers to the solvent used for electrode calibration and y refers to the sample solvent; as commonly employed in this type of study w indicates water (aqueous) and s indicates a mixed solvent (e.g., methanol/water).21 In this work all pH calibrations were carried out in commercial aqueous buffers using a Metrohm microelectrode (6.0234.100); all pH measurement are therefore wwpH or swpH. Chemicals. Microcystin-LR (MW = 995.1) was purchased from Enzo Life Sciences. A 10 ppm (approximately 10 μM in 100% methanol) MC-LR standard solution was purchased from Sigma-Aldrich for HPLC calibration. (Caution: MC-LR is an irritant, very toxic, and may be carcinogenic.) Stock solution of Fe2+ was prepared from FeSO4.7H2O. Acid and base solutions were prepared from HCl, HNO3 and NaOH Titrisol (Merck) ampules. All other chemicals were analytical grade and were used without further purification. Eighteen (18) MΩ.cm MilliQ water was used for the preparation of all solutions; for electrochemical measurements, the Milli-Q water was first boiled and argon degassed. Quantitative Analysis of MC-LR by HPLC. A HPLCbased method was used for the quantitative analysis of MC-LR. The HPLC instrumentation was a Waters 600 Controller, 717plus autosampler and 2996 photodiode array detector. The separation system used a LiChrospher 100 RP-18 (5 μm) column (Merck) maintained at 30 °C with an (isocratic) eluant containing a 60:40 v/v mixture of 100% MeOH and phosphate buffer (76 mM KH2PO4 and 10 mM H3PO4; pH 3) at a flow rate of 1 mL min−1 and an injection volume of 50 μL. The diode array detector was set at a sampling frequency of 5 s−1 over the spectral range of 210 nm −400 nm (2 nm resolution); the absorbance at 238 nm was used for quantification of MCLR concentrations. pH-Dependent Stability of MC-LR. For pH-dependent stability measurements an initial solution of MC-LR was prepared at a concentration of approximately 15 μM in 0.8% v/ v MeOH/water (starting volume ∼5.1 mL; 125-fold aqueous dilution of ∼2 mM MC-LR in 100% methanol). This solution was stirred in a jacketed reactor (maintained at 25 °C with a Grant GR150 water bath-circulator) and the swpH adjusted to 3 by the addition of HCl. The stability of MC-LR was investigated across the swpH range of 3−10, with the pH increased by the addition of NaOH using a Metrohm 721 NET Titrino controlled by Metrohm TiNET software. Sacrificial samples (250 μL) were taken from the reactor 15 min after adjusting the pH and diluted 1:1 with 100% MeOH for HPLC analysis. At the completion of this experiment the MC-LR concentration was 9.2 μM. HPLC Determination of MC-LR Carboxylic Acid pKa Values. The pKa values of the two carboxylic acid groups of MC-LR were determined using HPLC from the dependence of the MC-LR retention time on eluant swpH . In this experiment the separation conditions were the same as that described for the analysis of MC-LR with the exception that the eluant was 60:40 v/v mixture of 100% MeOH and citrate buffer (1.44 ≤ w wpH ≤ 5.62; total citrate = 20 mM), corresponding to a pH 5179
dx.doi.org/10.1021/es400464e | Environ. Sci. Technol. 2013, 47, 5178−5184
Environmental Science & Technology
Article
measurements of mixtures of Fe2+ with MC-LR were made after completion of the MC-LR measurements. In order to minimize Fe2+ hydrolysis and/or oxidation the swpH of the MC-LR in background electrolyte was adjusted to 3, Fe2+ added to give a MC-LR:Fe ratio of 1:0.9 (Fe2+tot ≈ 0.27 mM), and the solution allowed to equilibrate for 1 h prior to recording CVs across the range 3 < swpH < 8. During the course of the MC-LR and Fe2+ mixture scans, MC-LR in 100% MeOH was periodically added to account for volume change through solution loss during electrode repolishing. All potentials in this work are relative to the Ag/AgCl electrode (KCl = 3 M; E1/2 = 0.207 V). Geochemical Calculations. Thermodynamic calculations were carried out using Geochemist’s Work Bench (GWB v8.0). For these calculations the acid dissociation and FeIII binding constants for MC-LR (from this work) were added to the default thermodynamic data set.29
■
RESULTS AND DISCUSSION
pH-Dependent Stability of MC-LR. The stability of MCLR was investigated over the swpH range 3 to 10 (0.8% v/v methanol/water) corresponding to a 4 h titration period, and then held at swpH > 8 for a further 20 h. Over this period there was minimal change in retention time or peak area of the samples analyzed by HPLC (Figure S1; SI). This is consistent with the findings of Yu et al.30 who report MC-LR stability from pH 1 to 12 (in water) with temperatures ranging from 10 to 150 °C for 6 h. We therefore conclude that the MC-LR should remain stable under the conditions employed in our experiments. Determination of MC-LR Acid Dissociation Constants. The acid dissociation for the two carboxylic groups of MC-LR can be described by eq 1; the intermediate species (H2MC±) is a zwitterion due to the protonated amine moiety associated with the arginine group of MC-LR. The acid dissociation of this amine group likely occurs at pH > 9,31 and has not been investigated in this work. + Ka1
± Ka 2
H3MC XooY H 2MC XoooY HMC
−
Figure 1. Data for the determination of MC-LR pKa values, where ○ indicates experimental values and indicates the fitted model. (a) HPLC method of determination, showing retention time versus swpH, using 60:40 v/v methanol:water (citrate buffer) eluant. (b) UV−vis spectroscopy method of determination. Fitted curve is based on component 1 as identified by mutivariate curve resolution (MCR) analysis (see text), showing distribution of protonated species versus s wpH, in
