Article pubs.acs.org/est
Oxidation of Microcystin-LR by Ferrate(VI): Kinetics, Degradation Pathways, and Toxicity Assessments Wenjun Jiang,† Long Chen,‡ Sudha Rani Batchu,† Piero R. Gardinali,† Libor Jasa,§ Blahoslav Marsalek,§ Radek Zboril,∥ Dionysios D. Dionysiou,⊥ Kevin E. O’Shea,† and Virender K. Sharma*,‡ †
Department of Chemistry and Biochemistry, Florida International University, 11200 SW Eighth Street, Miami, Florida 33199, United States ‡ Department of Environmental and Occupational Health, School of Public Health, Texas A&M University, 1266 TAMU, College Station, Texas 77843-1266, United States § Institute of Botany, Academy of Sciences of the Czech Republic, Lidická 25/27, 657 20 Brno, Czech Republic ∥ Regional Centre of Advanced Technologies and Materials, Department of Physical Chemistry, Faculty of Science, Palacky University, Slechtitelu 11, 783 71 Olomouc, Czech Republic ⊥ Environmental Engineering and Science Program, Department of Biomedical, Chemical and Environmental Engineering (DBCEE) 705 Engineering Research Center, University of Cincinnati Cincinnati, Ohio 45221-0012, United States S Supporting Information *
ABSTRACT: The presence of the potent cyanotoxin, microcystin-LR (MC-LR), in drinking water sources poses a serious risk to public health. The kinetics of the reactivity of ferrate(VI) (FeVIO42−, Fe(VI)) with MC-LR and model compounds (sorbic acid, sorbic alcohol, and glycine anhydride) are reported over a range of solution pH. The degradation of MC-LR followed second-order kinetics with the bimolecular rate constant (kMCLR+Fe(VI)) decreasing from 1.3 ± 0.1 × 102 M−1 s−1 at pH 7.5 to 8.1 ± 0.08 M−1 s−1 at pH 10.0. The specific rate constants for the individual ferrate species were determined and compared with a number of common chemical oxidants employed for water treatment. Detailed product studies using liquid chromatography−mass spectrometry/mass spectrometry (LC−MS/MS) indicated the oxidized products (OPs) were primarily the result of hydroxylation of the aromatic ring, double bond of the methyldehydroalanine (Mdha) amino acid residue, and diene functionality. Products studies also indicate fragmentation of the cyclic MC-LR structure occurs under the reaction conditions. The analysis of protein phosphatase (PP1) activity suggested that the degradation byproducts of MC-LR did not possess significant biological toxicity. Fe(VI) was effective for the degradation MC-LR in water containing carbonate ions and fulvic acid (FA) and in lake water samples, but higher Fe(VI) dosages would be needed to completely remove MC-LR in lake water compared to deionized water.
■
problematic cyanotoxins.4 More than 80 variants of MCs, differing only by the methylation pattern and two variable amino acids, have been identified.5 MC-LR is the most abundant of the microcystins accounting for 46.0−99.8% of total concentration of MCs in HABs.6 MC-LR consists of three D-amino acids, alanine (Ala), methylaspartic acid (MeAsp), and glutamic acid (Glu), two unusual amino acids (3-amino-9-methoxy-2,6,8-trimethyl-10phenyldeca-4,6-dienoic acid (Adda) and N-methyldehydroalanine (Mdha)), and two L-amino acids (leucine, Leu and arginine, Arg) (Figure S1, Supporting Information). The
INTRODUCTION Cyanobacteria were among the earliest known organisms and have been critical to the evolution of Earth with the production of oxygen in the Earth’s early atmosphere and nitrogen fixation.1 Despite the vital role these organisms play in the environment, reports of adverse health effects related to cyanobacteria date back to the 19th century.2 These ubiquitous microorganisms are commonly associated with water eutrophication. Population increases, global warming, and runoff from agricultural practices and human waste have contributed to an increase in the frequencies and intensities of cyanobacterial (blue-green algae) blooms, often referred to as harmful algal blooms or HABs. Recently, HABs have become concerns to environmental and human health globally.3 Microcystins (MCs), a class of hepatotoxic monocyclic heptapetides, are commonly produced by cyanobacteria and are among the most © 2014 American Chemical Society
Received: Revised: Accepted: Published: 12164
December 26, 2013 September 9, 2014 September 12, 2014 September 12, 2014 dx.doi.org/10.1021/es5030355 | Environ. Sci. Technol. 2014, 48, 12164−12172
Environmental Science & Technology
Article
Table 1. Second-Order Rate Constants and Half-Lives for Oxidation of MC-LR by Different Oxidants at 22−25 °C pH 7.0 oxidant
species
ferrate(VI)a (pK3 = 7.23)f permanganateb chlorinec (pKa = 7.54)g chlorine dioxided ozonee
HFeO4− FeO42− MnO4− HOCl OCl− ClO2 O3
ferrate(VI)
H2FeO4 HFeO4−
ferrate(VI)
HFeO4− FeO42−
ferrate(VI)
HFeO4− FeO42−
k, M−1 s−1 MC-LR (3.9 ± 0.2) × 102 8.0 ± 2.0 3.6 × 102 1.2 × 102 6.8 1.0 4.1 × 105 sorbic alcohol H3CCHCH−CHCH−CH2OH (2.8 ± 0.1) × 103 (4.0 ± 0.2) × 102 sorbic acid H3CCHCH−CHCH−COOH (2 ± 0.2) × 102 (M1, M2) (4 ± 0.1) × 105 (M2) glycine anhydride H-(Gly)2+ (2 ± 0.1) × 10° (M1, M2) (4 ± 1) × 10−1 (M2)
kapp, M−1 s−1
t1/2
2.5 × 102
155 s
3.6 × 102 7.2 × 101
107 s 504 s
1.0 4.1 × 105
13.1 h 0.08 s
H3CCHCH−CHCH−COO− (1.4 ± 0.1) × 102 (M1)
(Gly)2 (1.7 ± 0.6) × 105 (M1)
This study and half-life at dose [Fe] = 1 mg L−1 or [FeO42−] = 2.2 mg L−1. bFrom Rodriguez et al.18 and half-life at dose [Mn] = 1 mg L−1 or [MnO4−] = 2.2 mg L−1. cValues at pH 7.2 taken from Acero et al.47 and half-life at [HOCl] = 1 mg L−1. dFrom Kull et al.22 and half-life at [ClO2] = 1 mg L−1; eFrom Onstad et al.17 and half-life at [O3] = 1.0 mg L−1. aModel (M1) represents reactions of HFeO4− with protonated and unprotonated species of substrate. bModel (M2) represents reactions of HFeO4− and FeO42− with protonated species of substrate; fAt 25 °C from Sharma et al.46 g At 25 °C from Morris.82 a
dioxide requires high doses, which limits the practical application because of the formation of chlorite and chlorate as byproducts.22 At doses commonly used in drinking water treatment, ozone and permanganate can oxidize MC-LR.17−20 Recently, the high-valent iron based oxidant, ferrate(VI) (FeVIO42−, Fe(VI)) has emerged as a novel oxidant/coagulant to remove contaminants in water.23−27 An excellent paper recently appeared28 that clearly demonstrated the complex redox interplay of Fe(VI), Fe(V), and Fe(IV) with generation of molecular oxygen and hydrogen peroxide. The detailed study elegantly elucidated the conversion and cascade of Fe(VI) to other oxidation states in aqueous media. The application of Fe(VI) to oxidize MC-LR in water is the focus of the present study. A limited number of reports have appeared on the oxidation of MC-LR by Fe(VI);29−32 previous studies simply demonstrated the degradation of MC-LR in deionized water without detailed kinetics or product studies. The objectives of our current study were to (i) determine the kinetics of Fe(VI) reactions of MC-LR and model compounds (sorbic alcohol, sorbic acid, and glycine anhydride (Figure 1) as a function of solution pH, (ii) identify oxidation products (OPs) and reaction pathways using liquid chromatography-high resolution mass spectrometry (LC-HRMS) technique, (iii) assess the toxicity of OPs against PP1 assay, and (iv) evaluate potential of Fe(VI) to treat MC-LR in waters containing carbonate and natural organic matter (i.e., fulvic acid, FA) and in lake water.
unique hydrophobic Adda side chain is critical to the binding and biological activity associated with MCs.7 The uptake of microcystin-contaminated water can pose serious chronic and acute toxic effects on humans and animals since these toxins are inhibitors of type 1 and type 2A protein phosphatase (PP1 and PP2A)8,9 and tumor promoters.10,11 The World Health Organization has established a provisional guideline of 1 μg L−1 MC-LR for drinking water.12 Removal of MC-LR at the source or in drinking water treatment plants is of utmost importance. Treatment of extracellular MC-LR may be divided into two categories: physical removal (e.g., activated carbon, coagulation−flocculation−sedimentation, and sand and membrane filtration) and degradation removal (e.g., UV-based advanced oxidation technologies and photocatalytic and chemical oxidation).13,14 Applications of physical removal methods usually require replacement of materials (e.g., membranes and activated carbons) and/or cleaning because of fouling or saturation. MC-LR is stable under natural sunlight and resistant to degradation by UV radiation.15 Photocatalytic degradation of MC-LR using titanium dioxide (TiO2) is promising; however, it is most effective under UV irradiation which can be costly and suffers from the low quantum yields of TiO2.16 Furthermore, this technology requires separation of the catalyst after treatment. A variety of chemical oxidants have been employed for drinking water treatment including chlorine, chlorine dioxide, chloramine, ozone, and permanganate (Table 1).13,17,18 Chlorine reacts with MC-LR but leads to chlorine substitution, which creates undesirable, potentially toxic chlorinated byproducts.19,20 Moreover, chlorine can react with bromide ion to form HOBr, which can generate toxic brominated byproducts.21 Chloramine is not effective for the degradation of MC-LR. Degradation of MC-LR by chlorine
■
EXPERIMENTAL SECTION
A description of reagents used and the procedure for treatment studies is given in Text S1 (Supporting Information). Experimental details of kinetics, products, and toxicity studies are given below. 12165
dx.doi.org/10.1021/es5030355 | Environ. Sci. Technol. 2014, 48, 12164−12172
Environmental Science & Technology
Article
Waters Alliance 2695 Analytical HPLC instrument connected to 996 PDA was applied for chromatographic measurements. The column used was a Spherisorb ODS column (25 cm × 4.6 mm i.d., 5 μm particle size) and the mobile phase was a mixture of 50% methanol−50% 20 mM ammonium acetate aqueous buffer solution. Chromatograms were analyzed at 238 nm. The lower detection limit of MC-LR was 0.2 μM. Product Studies. In this set of experiments, a solution containing 3.33 μM MC-LR and 66.6 μM Fe(VI) in buffered deionized water at pH 8.0 was magnetically stirred. After completion of the reaction indicated by the disappearance of the characteristic Fe(VI) color, a 1.0 mL sample was subjected to analysis. The concentration of MC-LR was determined using a Varian ProStar HPLC system equipped with a ProStar 410 autosampler and a ProStar 335 photodiode array detector at 238 nm. The stationary phase consisted of a Luna RP C18 column (250 × 4.6 mm i.d., 5 μm) and a mobile phase mixture of 0.05% (v/v) trifluoroacetic acid (TFA) in acetonitrile (40%) and 0.1% (v/v) TFA aqueous solutions (60%). The measurement of formation of •OH was conducted using the specific hydroxyl radical trapping agent, terephthalic acid (TA), which was added to the Fe(VI) reaction mixture. TA is commonly used to selectively trap hydroxyl radical with the formation of 2hydroxy terephthalate acid, 2-HTA with characteristic fluorescence.36 The 2-HTA was excited at 315 nm, and fluorescence measurements were made at 425 nm. A QExactive mass spectrometer (Thermo Scientific, San Jose, CA) equipped with a Hypersil Gold column (100 mm × 2.1 mm, 3 μm) (Thermo Scientific, San Jose, CA) and a heated electrospray ionization source operating in the positive ionization mode with the following parameters was utilized for identification of reaction intermediates based on accurate mass measurements using a mass tolerance of 5 ppm: sheath gas flow, 30 arbitrary units; auxiliary gas flow, 20 arbitrary units; spray voltage, 4.4 kV; capillary temperature, 300 °C; heater temperature, 250 °C. For the identified products, targeted MS/ MS experiments were performed at 17500 resolutions using the following parameters: normalized collision energy (% NCE), 20−50; automatic gain control (AGC) target, 2 × 104, maximum injection time (IT), 40 ms; isolation width, 2.0 m/ z. The mobile phase was (A) acetonitrile and (B) 0.1% formic acid in water. Gradient solution was 15−25% A for 6 min followed by a linear increase to 40% for 15 min and 80% for 20 min and a decrease to 15% for 22 min, and held constant for additional 3 min. Toxicity Tests. PP1 assays were conducted using a previously published procedure.37 PP1 enzyme solution was diluted to ∼0.05 unit μL−1 with a buffer (pH 7.4) of 1.10 mM Na2EDTA, 50 mM Tris-HCl, 2.0 MnCl2, 1.0 mM dithiothreitol, and 0.5 mg mL−1 BSA. A substrate solution was prepared by dissolving 0.1052 g of p-nitrophenyl phosphate into 20 mL buffer containing 50 mM Tris-HCl, 20 mM MgCl2, and 0.2 mM MnCl2 at pH 8.1. Ten microliters of test sample was mixed with 10 μL of PP1 enzyme solutions in a 96-well polystyrene plate. The plate was incubated at 30 °C for 3 min, followed by an addition of 180 μL solution of substrate. The concentration of p-nitrophenol was monitored at 15 min intervals for 60 min at 405 nm on a BioTek Synergy 2 Microplate Reader. The standard curve was plotted as % inhibition of PP1 activity versus MC-LR concentration.37 The detection limit in the PP1 assay was 5 nM.
Figure 1. Apparent second-order rate constants as a function of pH for the oxidation of MC-LR, sorbic alcohol, sorbic acid, and cyclic peptide. Solid lines were drawn using model. (Experimental conditions: [glycine anhydride] = 5.0 × 10−2 M; [sorbic acid] = 5.0 × 10−3 M; [sorbic alcohol] = 5.0 × 10−2 M; n = 6 for model compounds or n = 3 for MC-LR; error bars represent standard deviation of runs.)
Kinetic Studies. A stopped-flow spectrophotometer (SX.18MV, Applied Photophysics, UK) equipped with a photomultiplier was used to perform kinetic studies of Fe(VI) with model compounds, sorbic alcohol, sorbic acid, and glycine anhydride. Kinetics measurements were made under pseudo first-order conditions in which [substrate] ≫ [Fe(VI)]. The obtained curves were analyzed using a nonlinear least-squares algorithm in the SX.18MV software. The pseudo-first-order rate constants were an average of at least six runs at various concentrations at each pH. Corrections to pseudo-first-order rate constants were carried out for the spontaneous decay of Fe(VI) at the different pH values. Corrections to the rate constants were ≤5%. The kinetic modeling of the rate constants was conducted using SigmaPlot 12 (2011−2012). The rate constants for the reaction between Fe(VI) and MCLR were determined using pseudo-order conditions in which Fe(VI) was in excess. The approach used was similar to the one used for determining oxidation of micropollutants by Fe(VI) that accounts the self-decomposition of Fe(VI) (eq 1)33 d[MC‐LR] d[MC‐LR] = −k[Fe(VI)][MC‐RL] ⇒ dt [MC‐LR] ⎛ [MC‐LR]t ⎞ = −k[Fe(VI)]dt ⇒ ln⎜ ⎟ ⎝ [MC‐LR]0 ⎠ = −k
∫0
t
[Fe(VI)]dt
(1)
where [MC-LR]0 and [MC-LR]t are the concentrations of MCLR at time zero and t, respectively, and k is the second-order rate constant for the reaction. The reaction at time t was stopped by adding the excess hydroxylamine as a quenching reagent. The second-order rate constant of Fe(VI) with hydroxylamine is of the order of 104 M−1 s−1.34 The concentration of hydroxylamine was 200 times higher than MC-LR concentration, and therefore, the reaction could be stopped instantaneously with the addition of the quenching reagent. The approach used here restricted the determination of the rate constants within the pH range of 7.5−10.0. At pH < 7.5, the reaction was too fast to apply this method. Moreover, the decay of Fe(VI) without MC-LR at pH < 7.5 was too fast to accurately measure [Fe(VI)] using absorbance measurements. The concentration of MC-LR was monitored using HPLC.35 A 12166
dx.doi.org/10.1021/es5030355 | Environ. Sci. Technol. 2014, 48, 12164−12172
Environmental Science & Technology
■
Article
RESULTS AND DISCUSSION Oxidation of MC-LR. Plots of ln{[MC-LR]0/[MC-LR]t} as a function of Fe(VI) exposure data ∫ 0t [Fe (VI)] were constructed to determine the second-order rate constants ((kMCLR+Fe(VI), M−1 s−1) for the reaction between Fe(VI) and MC-LR at different pH (Figure S2, Supporting Information). An increase in the bimolecular rate constants is observed with a decrease in pH (Figure 1) as observed for other systems.38,39 Figure 1 also includes the k as a function of pH for the reactions of Fe(VI) with sorbic alcohol, sorbic acid, and glycine anhydride. The reactivity of Fe(VI) with sorbic alcohol increased with decreasing solution pH, similar to the trend predicted by the density function theory (DFT) calculations for Fe(VI) reactivity.40 In the case of sorbic acid, there was a sharper decrease in rates observed under basic media compared to acidic media (Figure 1). The degradation of glycine anhydride (cyclic diglycine) decreases with increasing pH under alkaline conditions, but the rate was constant under acidic conditions (Figure 1). Significantly, the rate constant for the reactivity of Fe(VI) with the cyclic peptide is at least oneorder of magnitude slower than those of acyclic glycine and glycine−glycine.23,41 This suggests the amide bond of the cyclic structure is less reactive than the amine and amide groups of glycine and a simple peptide, diglycine, respectively. The kinetic measurements at different pH in Figure 1 were evaluated using the species of Fe(VI) and substrate, S (MC-LR, sorbic alcohol (H3CCHCHCHCHCH2OH), sorbic acid (H 3 CCHCHCHCHCOOH ⇌ H + + H 3 CCH CHCHCHCOO−, pKa = 4.7642), and glycine anhydride (H-(Gly)2+ ⇌ H+ + (Gly)2; pKa = 12.8843). Fe(VI) has three pKa values (H3FeO4+ ⇌ H+ + H2FeO4, pKa1 = 1.9;44 H2FeO4 ⇌ H+ + HFeO4−, pKa2 = 3.5;45 HFeO4− ⇌ H+ + FeO42−, pKa3 = 7.2346). The following model was used to interpret the data of the reactivity of Fe(VI) with S in Figure 1 (eq 1) k[Fe(VI)tot ] =
rate constants are presented in Table 1. Protonation of Fe(VI) and substrate introduced ambiguity; hence, rate constants could be fitted using either reactions of HFeO4− with protonated and unprotonated species of substrate (model M1) or reactions of HFeO4− and FeO42− with protonated species of substrate (model M2) (Table 1). The rate constants of sorbic acid were similar to MC-LR, which indicates that Fe(VI) may attack the double bonds in MC-LR. The cyclic peptide (glycine anhydride) reacted much slower than sorbic acid (Figure 1 and Table 1). This suggests that initial attack by Fe(VI) on MC-LR may involve different reaction sites such as carboxylic acid and diene. Nevertheless, there is a possibility of the reaction of Fe(VI) and different amide bonds in the heptapeptide of MC-LR. The comparison of the reactivity of different oxidants with MC-LR is presented in Table 1.17,18,22,47 Ozone, Mn(VII), and chlorine dioxide mediated degradation exhibit no significant pH dependence on rate constants.18 The apparent second-order rate constants for the oxidation of MC-LR by chlorine decreased with increase in pH,18 similar to oxidation by Fe(VI) (Figure 1). The species-specific rate constants for oxidation of MC-LR by HOCl and OCl− were similar to species of Fe(VI) (Table 1). At neutral pH, ozone had the highest reactivity among common oxidants listed in Table 1. Ozone can efficiently attack the double bond of MC-LR orders of magnitude faster at neutral pH than other oxidants. The rate of the ozone reaction with sorbic acid was much faster than that with amines and amides groups of MC-LR.17 The order of the reactivity with MC-LR was O3 > Mn(VII) > Fe(VI) > chlorine > chlorine dioxide (Table 1). The half-life (t1/2) for degrading MC-LR by O3 is
