Environ. Sci. Technol. 2010, 44, 7238–7244
Intermediates and Reaction Pathways from the Degradation of Microcystin-LR with Sulfate Radicals M A R I A G . A N T O N I O U , * ,†,‡ ARMAH A. DE LA CRUZ,§ AND D I O N Y S I O S D . D I O N Y S I O U * ,† Department of Civil & Environmental Engineering, University of Cincinnati, 765 Baldwin Hall, Cincinnati, Ohio 45221-0071, Department of Environmental Engineering, Technical University of Denmark (DTU), Miljoevej, Building 113, 2800 Kgs. Lyngby, and Office of Research and Development, National Exposure Research Laboratory, U.S. Environmental Protection Agency, 26 West Martin Luther King Drive; Cincinnati, Ohio 45268
Received January 5, 2010. Revised manuscript received March 15, 2010. Accepted March 29, 2010.
Degradation of the cyanotoxin microcystin-LR (m/z 995.5) using sulfate radical-based advanced oxidation technologies (AOTs) and identification of reaction intermediates formed during treatment were investigated in this study. To the best of our knowledge this is the first study on the degradation and identification of reaction intermediates for any cyanotoxin with SO4•-. Tandem mass spectrometry designated the formation of nine (as m/z) reaction intermediates with four of them (m/z 1011.5, 1027.5, 1029.5, and 1045.5) having multiple peaks in the TIC chromatogram. New peaks that were not observed with hydroxyl radical formed during photocatalytic oxidation (PCO) have been detected such as m/z 1045.5. The initially formed intermediates involved the oxidation of the unsaturated bonds of MC-LR especially the diene bonds located on the chainoftheAddaaminoacid.Subsequentintermediatesimplicated the oxidative cleavage of small functional groups (i.e., sCOOH), up to the complete removal of the Adda chain. The electrophilic character of SO4•- is proven by the multihydroxylation of the aromatic ring. Toward the end of treatment, simultaneous oxidation of the Adda chain and the cyclic structure occurred without the formation of linear products.
Introduction Cyanotoxins are naturally occurring toxic metabolites produced and released by a third of the 150 cyanobacterial species currently identified (1-3). Their persistence in natural waters (4), high solubility and chemical stability makes them a significant health hazard for the ecosystem (2).Worldwide human and animal episodes of cyanobacterial contamination has placed them in the hazardous contaminant lists of environmental protection agencies worldwide (5-7). A myriad of technologies have been tested against cyanotoxins (8-16) with advanced oxidation technologies (AOTs) being among the most promising and efficient for water detoxification (14, 17, 18). Microcystin-LR (MC-LR), being the most toxic and frequently found derivative of the hepatotoxic group * Address correspondence to either author. E-mail:
[email protected] (D.D.D.);
[email protected] (M.G.A.). † University of Cincinnati. ‡ Technical University of Denmark (DTU). § U.S. Environmental Protection Agency. 7238
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 19, 2010
of microcystins, has been used as the model contaminant in many studies testing potential treatment processes. Most of the AOTs tested for cyanotoxin removal usually result in the formation of hydroxyl radicals (HO•) as the main active oxidizing species (12-14, 17-21, 21). In our recent studies, we have tested alternative oxidizing species, sulfate radicals (SO4•-) (22) that have oxidizing abilities comparable to HO• (23, 24). Different activation processes (catalyst, radiation, and thermal) of the oxidants persulfate (PS) and peroxymonosulfate (PMS), were tested with various efficiencies (22). To comprehend sulfate radical attack (SRA) in greater detail, the reaction intermediates and pathways that MC-LR follows with SO4•- were investigated herein. Photolytic decomposition of PS was selected for the production of SO4•because it is an excellent radical production technique (25). Studies have shown that activation of PS and PMS, results in the formation of SO4•- in a broad pH range (1e pH e 10.5) (26-28). Because SO4•- is a strong e- acceptor (23, 29), evident from their ability to oxidize OH-, Cl-, and H2PO4- to their corresponding radical species (25), it has been used to study the complex free-radical chemistry of the DNA through the formation of nucleobase radical cations (29, 30). Even so, studies on the reactivity of SO4•- with amino acids are limited (31-33), therefore the second order reaction rate constants (kx/SO4) of most of the amino acids that comprise the structure of MC-LR were not available (kAla/SO4 ) 1 × 107 M-1s-1 (33)). So far, it has been reported that SO4•- can selectively react with saturated amino acids via H-abstraction to form carbon center radicals (31, 34), can be selectivity added in unsaturated amino acids (26), can remove free carboxylic groups (32) and, when aromatic substituents are present, SO4•- react at diffusion control rates (32). A general reactivity order of SRA with functional groups is aromatic > sCH2- > sCOs > sCO2H (35). Linear dipeptides (aromatic and aliphatic) were found to react faster with SO4•- than their parent amino acids alone, suggesting some degree of activation because of the peptide bond (32). However, in the case of MC-LR the cyclic structure imposes limitations that prohibit the reaction with SRs based on mechanism described in the linear peptides (32). Therefore, in this study, we have investigated the different mechanistic pathways that MC-LR undergoes during SRA, by identifying the formed reaction intermediates with mass spectrometry. Based on the previous discussion, it would be of great interest to determine which groups of MC-LR react with SO4•-, whether the free carboxylic (Glu, MeAsp) and basic groups (Arg) of MC-LR participate in any of the degradation mechanisms or the bulky size of the toxin acts as a barrier. Finally, the study aims to identify the most susceptible amino acids/sites of MC-LR with SRA. To the best of our knowledge, this is the first study of its kind on the reaction intermediates of the cyanotoxin MC-LR with SO4•-.
Experimental Section Safety. Due to the adverse health effects caused by cyanotoxin exposure, all the experiments were conducted in an Advance SterilchemGARD III Class II Biological Safety Cabinet (The Baker Company, Sanford, ME). Appropriate laboratory clothing, safety gloves, and protective goggles were mandatory when handling toxin solutions. Materials. Solid MC-LR (0.5 mg, 96.4%) was purchased from Calbiochem (Gibbstown, NJ) and stored at -20 °C. A 482 mg/L standard of toxin solution was prepared by the addition of 1 mL of autoclaved Milli-Q Synthesis A10 water (Millipore Corp., Billerica, MA). Potassium persulfate (PS, 10.1021/es1000243
2010 American Chemical Society
Published on Web 04/23/2010
K2S2O8, Fisher) was the sulfate radical-generating oxidant and was used as received. Reactor Design. A 5 mL solution containing 20 µM of MC-LR 2000 µM of PS (without pH adjustments and unbuffered solutions, pHo 6.4) was placed in round reactor vessels (6 cm diameter) and irradiated by two long wave 15 W UVA (300 < λ < 400 nm) lamps (Cole Palmer, Vernon Hills, IL). The intensity of the lamps was measured with potassium ferrioxalate actinometry, and the photon flux (P) of the reactor was determined equal to 18.28 and 21.85 W for the λ ) 365 nm and λ ) 313 nm, respectively (22). LC, MS, and MS/MS Analysis. Samples were analyzed for the concentration of MC-LR with liquid chromatography (LC)/UV and for the reaction intermediates with LC/mass spectrometry (MS). An Agilent 1100 series quaternary LC equipped with a photodiode array detector (PDA) set at 238 nm was utilized for the quantification of the toxin. The analysis was conducted under reverse phase conditions with the stationary phase being a C18 Discovery HS, Supelco column (4.6 mm ×150 mm, 5 µm particle size) and the mobile phase consisted of a mixture of (A) 0.05% trifluoroacetic acid in acetonitrile and (B) 0.1% trifluoroacetic acid in Milli-Q water (36) For these analysis conditions, MC-LR eluted at 20.4 min. The same gradient was applied for the LC/MS analysis but the mobile phase was acidified with 0.1% formic acid and the stationary phase had a diameter of 2.1 mm (12). The column was at room temperature and a delay in the elution of MC-LR was observed (RT 22.30 min). A Thermo Finnigan LTQ-FT (linear ion trap with Fourier transform) ion cyclotron resonance (ICR) mass spectrometer was utilized for the MS and MS/MS identification of the reaction intermediates. Positive ions were analyzed using electrospray ionization (ESI) in the FT-ICR where accurate masses were acquired in full scan mode (300-2000 Da/scan) and 100 000 resolution at m/z ) 400. The ESI sheath gas was operated at 16 (unitless), the auxiliary gas at 0 (unitless) and the heated capillary temperature at 275 °C. The instrument was tuned with the MC-LR to achieve maximum sensitivity. For full scan LTQ-MS/MS spectra, the selected precursor ions were isolated with an isolation width of 3 Da and product ions were formed with normalized collision energy of 35.
Results and Discussion This study focuses on the identification of the reaction intermediates formed during the chemical degradation of the cyanotoxin microcystin-LR with the PS/UV(300 < λ < 400 nm) system at neutral pH. The sulfate radicals were formed via homolytic cleavage of the peroxide bond of PS (22). PS is also activated by one electron transfer mechanisms from transition metals such as Fe2+, Cu2+ (37, 38). A study by Anipsitakis and Dionysiou has found Ag1+ to be the most efficient catalyst for the activation of PS but the system requires high concentrations of the catalyst (50 mg/L) and acidic pH (pH 3.0) to prevent Ag1+ precipitation (38). In addition, studies have shown that Ag1+ quenches sulfate radicals with high rates [kAg/SO4 ) 3.0 × 109 M-1s-1] (39, 40) and competes with MC-LR for reaction with the radicals (22). Therefore, homolytic decomposition of the oxidant was chosen over metal catalyst activation in this study. UV(300 < λ < 400 nm) radiation was also chosen over the commonly used in industrial applications UVC (λ ) 254 nm) because it is a less energy demanding source but can result in the production of SO4•- (28) and cannot photolyze MC-LR (10, 14). The PS/UV(300 < λ < 400 nm) system efficiently degrades MC-LR (Supporting Information (SI) Figure S1) (22). Since the objective of the study was to identify the oxidation products of SRA on MC-LR, preventing the formation of other competitive radicals was crucial. Formation of HO• was particularly of concern because they produce oxidation products similar to SO4•- (41, 42). The analytical methods
used herein can only detect stable and not transient intermediates and therefore it would have been difficult to distinguish which products resulted from the oxidation of which radical. HO• can form from the oxidation of hydroxyl groups and water by SO4•- or the photolytic dissociation of water. UV(300 < λ < 400 nm) cannot cause the photolysis of water, the reaction between SO4•- and water is negligible (pH 7, kH2O/SO4 < 6 × 101 M-1s-1 (23), whereas the reaction of SO4•- with OH- is significant only at basic pH conditions (43) (pH 11, kHO/SO4 ) 8.3 × 107 M-1s-1). Because of the abovementioned reasons, by using the PS/UV(300 < λ < 400 nm)/ pHMQ system to generate SO4•-, HO• quenching agents were not used. Based on our previous experience with HO• produced intermediates (12-15) and some preliminary studies on SO4•-, 22 masses (m/z) were selected for the mass accuracy analysis with the FT-ICR. SI Table S1 summarizes the “true” intermediates that were identified with the study. To distinguish between impurities of the MC-LR solution and true reaction intermediates, the “peak criteria” (12) were used in this study as well: the peaks had to have a signal-to-noise ratio of 3 and the difference between the peak areas of the initial and treated sample had to be at least two times, if the peak was found in the initial sample. Eventually, 9 m/z met our criteria and their areas with treatment time are listed in SI Table S1. For some of the intermediate masses, multiple peaks were detected and based on how well they were separated the peak areas are listed individually (i.e., m/z ) 1027.5) or as the overall peak area (i.e., m/z ) 1029.5). In the case of m/z ) 1043.5 is listed as peak no. 3, because two other peaks with the same m/z were detected but did not follow our criteria (SI Table S1). The MS/MS analysis was setup to obtain as many of the intermediate spectra as possible. However, in some cases, because of the concentration of the intermediates (area of the peaks was less than NL < 1.0 × 101), the compounds were not fragmented in a comprehensive and conclusive way. Multiple Peaks. Intermediates with m/z 1011.5, 1027.5, 1029.5, and 1045.5 had multiple peaks in the TIC chromatogram (SI Figure S2). Multiple peaks are formed from isobaric compounds that differ in the three-dimensional orientation of the bonds (stereoisomers) or have completely different structures (structural isomers). This is a common phenomenon during radical attack irrespective of the active species used (13, 44). As seen in SI Scheme S1, the transient steps of free radical attack on unsaturated carbon bonds can cause bond isomerization or the formation of different resonance configurations of the radicals (i.e., allylic radicals) (13, 44). Based on the mechanistic routes of SO4•- attack and the obtained MS/MS spectra, we have attempted to unveil the corresponding chemical structures. Starting with the 1011.5, there is 16 Da difference between MC-LR and this intermediate, which can result from the addition of atomic oxygen in the structure of MC-LR through hydroxyl substitution of unsaturated bonds. The possible sites of attack are the aromatic ring, the diene bonds of the Adda amino acid, and the ethylene group of the Mdha amino acid (see SI for nomenclature). To determine which intermediate is formed, the corresponding MS/MS data were analyzed. We have focused on the formation of specific ion fragments to determine the affected site. Among them m/z 599.3 (Arg-Adda-Gly-H; MeAsp-Arg-Adda-H) and 553.3 (Mdha-Ala-Leu-MeAsp-Arg-H). The 553.5 fragment was chosen because it excludes the Adda amino acid, which contains groups especially susceptible to SRA (23, 24). The MS/MS spectra of MC-LR have been extensively discussed in an earlier publication (12). However, since differences can be observed in the fragmentation patterns and intensities between different instruments, the MS/MS spectrum of MC-LR was determined for the LTQ-FT-ICRVOL. 44, NO. 19, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
7239
MS as well. Once again, it was observed that the initial fragments are caused from the removal of small neutral molecules (including H2O, CO, CO2, NH3) followed by the removal of amino acids and the formation of linear fragments (i.e., m/z 599.3) (SI Figure S3). The m/z ) 1011.5 with retention time (R.T.) ) 17.90 min intermediate, is believed to have resulted from the hydroxylation of the Adda amino acid (aromatic ring or diene) because of the formation of the 799.6; 698.5; 615.4 fragments that have 16 Da of difference between the MC-LR fragments of 783.5 (Ala-Leu-MeAsp-Arg-Adda-H), 682.4 (Arg-Adda-GluMdha-H) and 599.3, respectively. All these ion fragments contain unsaturated bonds from the Adda and Mdha amino acids, another site where the hydroxylation may have occurred. However, in both cases the 553.3 (Mdha-Ala-LeuMeAsp-H) was formed and supports that the hydroxylation occurred at the Adda (SI Figure S4). A similar spectrum was observed for the intermediate m/z 1011.5, at R.T. ) 20.1 min (SI Figure S5). Based on the MS/MS spectra we cannot definitely state which spectrum belongs to the aromatic ring and which to the diene bonds. Given that the second order rate constants with sulfate radicals for the aromatic ring is higher than that with the diene bonds, it is likely that the hydroxylation occurs mainly in the aromatic ring (kbenzene/SO4 ) 3.0 × 109 M-1s-1 (47), kallylic-OH/SO4 ) 7.4 × 108 M-1s-1 (45). Hydroxylated intermediates are observed in both SRA and HO• attack, however the mechanistic routes that lead to the stable intermediates differ (13, 25, 26, 35, 41). Electron spin resonance (ESR), a magnetic resonance technique based on the interaction of unpaired electron spins with an external magnetic field (46), has been used for the identification of transient radical species formed upon radical attack (25, 26). Consequently ESR has assisted in determining the oxidation mechanisms leading to stable intermediates (25, 26). Sulfate radicals react with the aromatic ring (and olefinic double bonds) and form short-lived sulfate radical adducts (lifetime
