Final Report: Chemical Degradation Pathways for the Natural Attenuation of Marine Biotoxins

EPA Grant Number: R831042
Title: Chemical Degradation Pathways for the Natural Attenuation of Marine Biotoxins
Investigators: Ferry, John L. , Moeller, Peter M.
Institution: University of South Carolina at Columbia , Center for Coastal Environmental Health Biomolecular Research (CCEHBR)
EPA Project Officer: Packard, Benjamin H
Project Period: September 1, 2003 through August 31, 2006
Project Amount: $404,403
RFA: Ecology and Oceanography of Harmful Algal Blooms (2002) RFA Text |  Recipients Lists
Research Category: Aquatic Ecosystems , Ecosystems , Water

Objective:

The overarching goal of the proposal was explore the fundamental fate and transport processes that govern the abiotic processing of marine toxins. Our particular focus was the toxins domoic acid, kainic acid, brevetoxin B, saxitoxin, and okadaic acid. The specific objectives of the study were to a) build a library of multivariate models for describing the half-life of a given toxin as a function of light intensity, suspended solids, and water quality during a bloom, b) to identify degradation products, for further toxicity evaluation or use as chemical markers of abiotic degradation in the field, and c) build databases of Koc with respect to water quality. We believe this knowledge critical for predicting the impact of a harmful bloom event, and also that it will yield valuable insight into the possible ecological function of marine toxins based on new understanding of their persistence in the environment. It also may suggest what variables in the water column need to be adjusted to remediate the impact of a bloom.

Summary/Accomplishments (Outputs/Outcomes):

Design and construction of a high throughput photo-system for multifactorial photochemistry: To be able to use multifactor techniques, we first had to work out an experimental set-up (Figure 1). Once this groundwork was laid background photolysis experiments were performed to determine the timeframe in which dissolved organic matter was stable. Using a central composite design, we are focused on four particular variables that are important to DOM photobleaching: Suwannee river dissolved organic matter (DOM), total Fe, NO3-, and salinity (Table 1). We look at each of the four factors at five different levels, with a minimum n=3 for each photolysis. In these experiments, the absorbance of DOM is monitored with time, and an experimental rate, kobs, obtained for each condition (Figure 2). This rate is then correlated against all single factors and all possible intrafactor and interfactor combinations. Each possible variable is assigned a coefficient, βx, and the hypothesis that βx is nonzero is tested (two sided ttest, 95% level of confidence) for all (Table 2).

Figure 1: A well characterized photochamber can replace the typical continually stirred tank reactor format. Small vials are periodically removed from the light chamber and analyzed, rather than periodic sampling of a larger vessel. One complete filling of the photochamber = 184 vials or 61 different CSTR style exps, n=3, for a single time point.

Table 1. Experimental conditions. Factors, variable codes, and concentrations for the four-factor central composite design; n = 3 for axial and factorial point experiments, n = 6 for center point experiments.

Factor (units) Factor Concentration Levelsa
Coded Factor Levels -2 -1 0 1 2
Factor x1: Fe(III) (μM)b 0 1.00 2.00 3.00 4.00
Factor x2: NO3-(μM) 0 14.70 29.50 44.20 59.00
Factor x3: DOM (mg/L) 0 7.50 15.00 22.50 30.00
Factor x4: Salinity (ppt) 0 8.75 17.50 26.25 35.00

aDenotes initial concentrations, b nominal concentration

Table 2. β values and corresponding t-test comparisons for the full factor model, at a 95% level of confidence.

Parameter Factor effect (βxest.) x 104 (Std. error) x 104 |tcalc| Prob >|t|
β0 y-intercept 114.00 6.87
β1Fe(III) 6.08 15.60 0.39 0.70
β2 NO3-2.66 3.75 0.71 0.48
β3*DOM 28.00 3.75 7.46 < 0.0001
β4salinity (S) 2.06 3.75 0.55 0.58
β11Fe(III)-Fe(III) 3.24 6.92 0.47 0.64
β22 NO3--NO3-0.16 1.36 0.12 0.91
β33* DOM-DOM -9.30 1.36 6.84 < 0.0001
β44 S-S -1.40 1.36 1.01 0.32
β12 Fe(III)-NO3-2.45 3.20 0.77 0.45
β13 Fe(III)- DOM 5.28 3.20 1.65 0.10
β14 Fe(III)-S -2.00 3.20 0.63 0.53
β23 NO3--DOM -1.30 1.42 0.89 0.38
β24 NO3--S 1.16 1.42 0.82 0.42
β34 DOM-S 0.381.42 0.27 0.79

* indicates βx ≠ 0 at the 95% level of confidence

Figure 2. Suwannee River DOM photobleached the most rapidly at 350 nm. Absorbance was measured at t = 0, 5, 10, 20, 30, 40, and 50 hrs, and is shown normalized against the absorbance at t = 0. Experimental conditions were: 30.00 mg/L [DOM]o; 2.00 μM Fe(III); 29.50 μM NO3-; 17.50 ppt salinity; 26°C +/- 0.5; and light intensity of 765 W/m2 summed over the range 290 nm -800 nm.

The nature of the experimental design and sample handling allowed the elimination of some potential photobleaching mechanisms. If HO• had been related to photobleaching, salinity would have tested as a significant variable, because the term “salinity” included ions such as HCO3-/CO3 2- , Br-, and I-. These ions are all known to react rapidly with HO•, suggesting that high salinity should have resulted in lower rates of photobleaching. Since salinity did not test as significant, HO• did not play a role in photobleaching in this system. It is possible that oxidation by O2-• was responsible for at least some photobleaching. However, Fe(III) did not test as a significant variable, and given its high rate of reaction with superoxide (1.5x108 M-1s-1) that would have been expected if superoxide was an important contributor to the rate of photobleaching. Photogenerated hydrogen peroxide is unlikely to have contributed to bleaching because the absorbance of the samples was stable with time post-irradiation; relatively long-lived H2O2 would have been expected to continue bleaching for some time after the sample was removed from the light source. Given that other photogenerated oxidants are excluded from a significant role in photodegradation, these results suggest that photoinduced charge transfer within SRDOM may be responsible for the bulk of SRDOM photobleaching that occurs under a wide variety of environmental conditions, and may be important for other types of DOM as well. Additionally, it was determined that DOM had a half life of ~24 hours with regards to photobleaching at 350 nm. This work has been published in Environmental Science and Technology 2006, 40, 3717-3722 (Appendix 1).

Design of aqueous phase derivatization process for LMW acids: At a tangent to this project, we developed a a technique for the rapid, room temperature derivatization of aqueous carboxylic acids to the corresponding 2,2,2-trifluoroethyl amide derivative. 3-Ethyl-1-[3-(dimethylamino)propyl] carbodiimide hydrochloride (EDC) and 2,2,2-trifluoroethyl amine hydrochloride (TFEA) were added to aqueous samples of several acids of interest in environmental analytical chemistry. Amidization was essentially complete within 10 min, and subsequent liquid-liquid extraction of the resultant amides with methyl-tertbutyl ether demonstrated recoveries of over 85% (Figure 3). At pH 4-5 the starting materials were ionized and were not co-extracted with the derivative, yielding much cleaner samples than historically obtained from carbodiimide based techniques in organic solvents. The fluorinated amides produced had excellent chromatographic characteristics for gas chromatography and were easily detected by electron capture detection or electron impact mass spectrometry (Figure 4). This method is suggested as a sensitive alternative to more traditional acidification, extraction, and ex-situ derivatization techniques.

Figure 3. a) The benzoic acid derivatization is ~90 % complete within 5 minutes. b) Using a 10 minute reaction time, it was found that the optimal pH for the benzoic acid derivatization was at pH 5.

Figure 4. N-2,2,2 trifluoroethylbenzamide (benzoic acid derivative) mass spectrum and its significant fragments found using GC-IT-MS.

The use of carbodiimides with quaternary ammonium substituents eliminates the non-polar contaminants typically associated with the use of more traditional carbodiimides, such as dicyclohexylcarbodiimide. The resulting fluorinated amides are readily chromatographed and easily detectable by electron capture techniques. It should be noted that this method has been tested with carboxylic acids along with amino acids with only the carboxylic acids derivatizing. While this method is not applicable to amino acids, it is still useful for EPA mission. This work has been published in the Journal of Chromatography A 2007, 1145 (1-2), 241-245 (Appendix 1).

Photochemical Degradation of Domoic Acid: Our strategy has been to apply multifactor experimental techniques for modeling the environment to the degradation of the toxins listed. We are focused on four particular variables that are important to water quality and the probability of an algal bloom: dissolved organic matter (DOM), total Fe, total PO43-, and NO3-. We have been using the central composite experimental design to map the relevant concentrations required for each of these variables for each experiment (Table 3). We look at each of the four factors at five different levels, with a minimum n=3 for each photolysis. A typical experiment involves observing the photodegradation of a given toxin under the conditions outlined by the experimental design (Figure 5). In these experiments, the concentration of toxin is monitored with time, and an experimental rate, kobs, obtained for each condition. This rate is then correlated against all single factors and all possible intrafactor and interfactor combinations. Each possible variable is assigned a coefficient, βx, and the hypothesis that βx is nonzero is tested (two sided ttest, 95% level of confidence) for all (Table 4).

Figure 5. Domoic acid rapidly photodegrades with t1/2>< 20 hrs. Experimental conditions shown: λex = 300-800 nm; 0.96 μM domoic acid, 32 ppt salinity, 5 mg/L DOM, 2 μM Fe(III), 35 μM NO3- , and 2 μM PO43-

This screening strategy enabled the identification of DOM, total Fe, and the interaction term between total Fe and total PO43- as important for promoting or inhibiting the indirect photodegradation of domoic acid. The identification of an interaction term between the two inorganic nutrients led us to conduct a more detailed molecular investigation, and the development of an NMR based technique for measuring the association constants between Fe(III) and organic ligands in dilute solution. We found the association constant is approximately 1x1011. Although this number seems large, it is much smaller than the association constant for the interaction of Fe(III) with PO43- (~ 1x1040). We attributed the preserving effect of the Fe-PO43- interaction a function of competitive binding, with PO43- effectively removing Fe(III) from the system and reducing its impact as a photocatalytic oxidant. It was also possible to apply this analysis to generate a crude model for predicting domoate photodegradation. No direct photodegradation of domoic acid was observed. This work has been published in Environmental Science and Technology 2006, 40, 2200-2205 (Appendix 1).

Photochemical Degradation of Kainic Acid and LC-MS/MS versus the cELISA: Kainic acid direct photolysis (0.96μM kainic acid with 32 ppt salinity) was essentially non-existent and did not contribute significantly to the overall photodegradation (Figure 6). Under the conditions tested KA did photodegraded slowly over the time period of the experiment (Figure 7). The apparent loss was first order in kainic acid (as indicated by analysis of ln([KA]t/[KA]o vs time). The observed rate constant (kobs) was obtained by a linear least-squares analysis of the relationship between [KA]t/[KA]o and time for all experimental conditions tested (Table 5).

Samples containing both kainic and domoic acid were photodegraded and then quantified using LCMS/ MS and a commercially available cELISA from BioSenseTM Laboratories. LC-MS/MS was used to quantify both kainic and domoic acid, while the cELISA quantified only domoic acid. As in the previous photostability studies, the first order loss of the toxins was plotted (ln([x]t/[x]0) vs time) to determine rate constants for comparison between LC-MS/MS and cELISA (Figure 8). The result was an approximately 20-40% decrease in the rate of domoic acid disappearance for domoic acid quantified using cELISA (Table 6). This experiment showed that cELISA does quantify more domoic acid in the presence of kainic acid than does the LC-MS/MS. To determine if this increase in quantified domoic acid was due to degradation product buildup or the presence of kainic acid, samples containing either domoic acid or kainic acid that had not been photolyzed were assayed with cELISA. Concentration was then determined from the response. When the concentrations were compared, the results yielded a 1:1 ratio of domoic acid to kainic acid (Figure 9).

Figure 6. A plot of the direct photolysis of kainic acid degradation over time (0.96 μM kainic acid and 32 ppt salinity). Kainic acid alone shows no significant degradation in simulated seawater.

Figure 7. Plot of kainic acid degradation over time (0.96 μM kainic acid, 32 ppt salinity, 5 mg/L DOM, 17.5 μM NO3-, 4 μM Fe(III), and 2 μM PO43-). The t1/2 of kainic acid under the above conditions is approximately 71 hours.

Table 5. Rates of kainic acid photolysis under relevant environmental conditions. All samples contain 0.96μM kainic acid along with 32 ppt salinity.

Figure 8. A typical plot of toxin degradation over time (0.96 μM domoic acid, 0.96 μM kainic acid, 32 ppt salinity, 5 mg/L DOM, 17.5 μM NO3-, 2 μM Fe(III), and 1 μM PO43-). The domoic acid quantified using the ELISA technique is ~1.5 slower than the rate found using LC-MS.

Figure 9. A comparison of the relative concentrations of domoic acid and kainic acid, as calculated from the cELISA test, when the toxins were run separately shows that there is a 1:1 response on the cELISA for kainic acid. The conditions are 0.96 μM toxin and 32 ppt salinity, n = 6 replicates.

Table 6. A comparison of the half-lives found using LC-MS/MS and cELISA to quantify domoic acid. All samples contain 0.96 μM domoic and kainic acid along with 32 ppt salinity.

The photodegradation of kainic acid was also evaluated using this technique. It did not appear to undergo significant photodegradation under our conditions. The binding constant for kainic acid with iron (K1 = 611.10 and K2 = 793.98) was also determined. No significant binding takes place, supporting the fact that Fe(III) does little if anything to promote kainic acid photodegradation. A mixture of kainate and domoate has been exposed to the experimental matrix and their loss quantified. The resulting samples have been sent to Moeller’s lab for toxicity evaluation, using a commercially available ELISA kit for domoic acid, to probe for the correlation between the loss of the parent toxins and loss of toxicity. The ELISA kits indicated that domoic acid was much more persistent than what was found using the LC-MS technique to quantify. On average the estimated half-life of domoic acid more than doubled when the samples were quantified by ELISA as opposed to the half-lives found using LC-MS. This would seem to indicate that either a) kainic acid is giving a false positive and/or b) the photodegradation products from domoic acid are causing interference. This work has been published in the Journal of Agricultural and Food Chemistry 2007, 55, 9951-9955 (Appendix 1).

Adsorption of Domoic Acid: Here our strategy looks at the adsorption capacity of the different toxins to several varieties of clay/sediment. Surface loadings were chosen based on their relevance to marine snow particulate densities and also clay-based adsorption technologies for harmful algal bloom control (0-100 mg/L). Solid-water binding (Kd) constants from the adsorption isotherms (Figure 10) were determined to be in the non-surface area normalized range of 100-2800 mol/kg. The addition of Fe(III) (0-4 μM) and dissolved organic matter (0-10 mg/L) was also probed (Figures 11-12).

Figure 10. Domoic acid ([DA]0 = 0.96 μM) weakly absorbs to Na-montmorillonite (sieved to 38 μm particle size) in 32 ppt salinity.

Figure 11. The fraction of domoic acid adsorbed to the surface of Na-montmorillonite (constant loading of 60 mg/L) greatly increases with the addition of Fe(III).

Figure 12. The fraction of domoic acid adsorbed to the surface of Na-montmorillonite (constant loading of 60 mg/L) greatly increases with the addition of DOM.

Isotherms were been built for Na-montmorillonite, Ca-montmorillonite, well-crystallized kaolin, poorly crystallized kaolin, kaolinite, diatomeous earth, Gulf of Mexico sediment, Santa Barbara Basin sediment, and Bread and Butter Creek sediment with domoic acid. All adsorption experiments were performed under a constant salinity of 32 ppt. The effect of Fe(III) on adsorption was tested with diatomeous earth. It was found that domoic acid does not adsorb to diatomeous earth be it in simulated seawater (Instant Ocean), natural seawater, distilled water, or in the presence of Fe(III). All other clays were found to have weak binding affinities for domoic acid. Na-montmorillonite, Gulf of Mexico sediment and wellcrystallized kaolin (a constant 60 mg/L solid loading and 32 ppt salinity) were tested for the effect of Fe(III) and DOM. It was found that in all cases tested both the addition of Fe(III) and DOM enhanced the adsorption of domoic acid to the solids. This work was published in the Journal of Environmental Monitoring 2007, 9, 1373-1377 (Appendix 1).

Adsorption of Saxitoxin: Saxitoxin was adsorbed to several varieties of clay/sediment. Surface loadings were chosen based on their relevance to marine snow particulate densities and also claybased adsorption technologies for harmful algal bloom control (0-250 mg/L). Adsorption constants (Kads) from the adsorption isotherms (Figure 13) were determined to be 3x104 – 36x104 L/mol. ΔG0ads, Γmax, and Kd were also determined.

Figure 13. A) Equilibrium was reached within 3.5 hours for both clays and sediments in DI water. B) Saxitoxin adsorption is best described by a Langmuir isotherm. Conditions: [STx]0 = 5-24 μM, adsorbent loading = 0.25 g/L, pH = 6.5.

Saxitoxin was adsorbed to Na-montmorillonite, Ca-montmorillonite, well-crystallized kaolin, poorly crystallized kaolin, kaolinite, powdered activated carbon, Gulf of Mexico sediment, and Bread and Butter Creek sediment. Adsorption experiments were performed in both DI water and 32 ppt salinity. For all adsorbents tested, ~90% of the saxitoxin was adsorbed under our conditions. Resuspension experiments were also performed to determine the possible release of saxitoxin from adsorbed sediments. Within the first 3 hours the system reached equilibrium and ~ 10% of the adsorbed saxitoxin was released. This work has been submitted to Environmental Science and Technology (Appendix 1).

Brevetoxin research: Brevetoxin B-2 was purchased. However, due to analytical difficulties it was not able to be consistently analyzed.

Okadaic acid research: Okadaic acid was commercially unavailable the first 2 years of the project and then when it did become commercially available was priced beyond our budget.

Conclusions:

Analytical methods:

Domoic acid and kainic acid samples were analyzed without any further sample treatment with an Agilent 1100 HPLC coupled to a Micromass-Quattro mass spectrometer equipped with an electrospray ion-spray (ESI) interface. Chromatographic separation was achieved using an Aqua Sep 5μm particle size, 10 cm by 2.1 mm (i.d.) column (ES Industries, West Berin, NJ) in conjunction with a corresponding Aqua Sep 5μm particle size, 1 cm by 3.2 mm (i.d.) guard column. The LC-MS/MS procedure was as follows: A mixture of 0.1% aqueous formic acid in DI water (A) and 0.1% formic acid in acetonitrile (B) was used as the mobile phase. The initial condition was 95:5 A:B for 3 minutes, followed by a linear gradient over 13 min ending at 5:95 A:B. The ratio of A and B was reset to the initial condition over the following 8 minutes to reestablish initial conditions. The flow rate was 200 μL/minute with a sample injection volume of 50 μL. A 6 minute solvent diversion was used to avoid salt contamination of the ion source. The MS operating conditions were set to a cone voltage of 30V, a collision voltage of 16 eV, a source block temperature of 100 °C, and a desolvation temperature of 350 °C. The mass spectrometer was run in multiple reaction monitoring mode (MRM) with a dwell time of 0.20 seconds.

Brevetoxin (PBTx-2) samples were analyzed without any further sample treatment with an Agilent 1100 HPLC coupled to a Micromass-Quattro mass spectrometer equipped with an electrospray ion-spray (ESI) interface. Chromatographic separation was achieved using an Chromegabond WR C18 5μ15 cm by 2.1 mm (i.d.) m particle size, column (ES Industries, West Berin, NJ) in conjunction with a corresponding Chromegabond WR C18 5μ1 cm by 3.2 m particle size, mm (i.d.) guard column. The LC-MS/MS procedure was as follows: A mixture of 0.1% aqueous formic acid in DI water (A) and 0.1% formic acid in acetonitrile (B) was used as the mobile phase. The initial condition was 90:10 A:B for 5 minutes, followed by a linear gradient over 1 min ending at 5:95 A:B. This ratio was held constant for 7 min. The ratio of A and B was reset to the initial condition over 1 min and then held for 7 min to reestablish initial conditions. The flow rate was 200 μL/minute with a sample injection volume of 50 μL. A 5 minute solvent diversion was used to avoid salt contamination of the ion source. The MS operating conditions were set to a cone voltage of 40V, a source block temperature of 100 °C, and a desolvation temperature of 350 °C. The mass spectrometer was run in single ion reaction mode (SIR) with a dwell time of 0.10 seconds. Brevetoxin was identified and quantified by analysis of the signal from the M + H ion (896.48 Da) and the sodium adduct (918.47 Da).

Implications:

This work has shown that domoic acid has a short half life in the photic zone which is dependent on several variables which interact together. Also, this work has shown that commercially available ELISA tests for domoic acid can be unreliable in the presence of kainic acid. This has a large impact economically as many of the fish which are deemed unusable by this test may actually be fine as domoic acid and kainic acid have been shown to be co-occurring in different macroalgae species. Additionally, domoic acid only weakly binds to sediments and clays indicating that this is not a strong removal mechanism of the toxin from the water column, especially in Fe(III) and/or DOM limited regions where these substituents are not available to enhance domoic acid’s adsorption. Additionally, saxitoxin readily adsorbs to sediments but upon resuspension is not easily released. This could be problematic for benthic organisms that ingest these sediments.


Journal Articles on this Report : 6 Displayed | Download in RIS Format

Other project views: All 15 publications 6 publications in selected types All 6 journal articles
Type Citation Project Document Sources
Journal Article Burns JM, Ferry JL. Adsorption of domoic acid to marine sediments and clays Journal of Environmental Monitoring 2007;9(12):1373-1377. R831042 (Final)
  • Abstract from PubMed
  • Abstract: Journal of Environmental Monitoring-Abstract
    Exit
  • Journal Article Burns JM, Schock TB, Hsia MH, Moeller PR, Ferry JL. Photostability of kainic acid in seawater. Journal of Agriculture and Food Chemistry 2007;55(24):9951-9955. R831042 (Final)
  • Abstract from PubMed
  • Abstract: ACS-Abstract
    Exit
  • Journal Article Burns JM, Hall S, Ferry JL. The adsorption of saxitoxin to clays and sediments in fresh and saline waters. Water Research 2009;43(7):1899-1904. R831042 (Final)
  • Abstract from PubMed
  • Full-text: ScienceDirect-PDF
    Exit
  • Abstract: ScienceDirect-Abstract
    Exit
  • Journal Article Fisher JM, Reese JG, Pellechia PJ, Moeller PR, Ferry JL. Role of Fe(III), phosphate, dissolved organic matter, and nitrate during the photodegradation of domoic acid in the marine environment. Environmental Science & Technology 2006;40(7):2200-2205. R831042 (Final)
  • Abstract from PubMed
  • Abstract: ES&T-Abstract
    Exit
  • Journal Article Ford QL, Burns JM, Ferry JL. Aqueous in situ derivatization of carboxylic acids by an ionic carbodiimide and 2,2,2-trifluoroethylamine for electron-capture detection. Journal of Chromatography A 2007;1145(1-2):241-245. R831042 (Final)
    R826399E02 (Final)
  • Abstract from PubMed
  • Full-text: ScienceDirect-PDF
    Exit
  • Abstract: ScienceDirect-Abstract
    Exit
  • Journal Article Hefner KH, Fisher JM, Ferry JL. A multifactor exploration of the photobleaching of Suwannee River dissolved organic matter across the freshwater/saltwater interface. Environmental Science & Technology 2006;40(12):3717-3722. R831042 (Final)
  • Abstract from PubMed
  • Abstract: ES&T-Abstract
    Exit
  • Supplemental Keywords:

    marine, estuary, adsorption,chemicals, particulates, organics, pathogens, bacteria, intermediates,ecosystem, indicators, restoration, regionalization, aquatic, habitat,conservation,environmental chemistry, biology, ecology, hydrology, limnology, modeling, monitoring, analytical, measurement methods, pacific coast, Atlantic coast, gulf coast, Chesapeake Bay, aquaculture, tourism, RFA, Scientific Discipline, Water, Ecosystem Protection/Environmental Exposure & Risk, Oceanography, algal blooms, Environmental Monitoring, marine ecosystem, dissolved organic matter, bloom dynamics, food web, marine biotoxins, chemical degradation, nutrient kinetics, natural attenuation, water quality, photocatalytic oxidation, algal bloom detection

    Progress and Final Reports:

    Original Abstract
  • 2004 Progress Report
  • 2005 Progress Report