Environ. Sci. Technol. 2005, 39, 3134-3142
Sorption and Related Properties of the Swine Antibiotic Carbadox and Associated N-Oxide Reduced Metabolites TROY J. STROCK, STEPHEN A. SASSMAN, AND LINDA S. LEE* Department of Agronomy, Purdue University, West Lafayette, Indiana 47907-2054
Carbadox (CBX) (methyl 3-[2-quinoxalinylmethylene]carbazate N1, N4 dioxide) is a chemotherapeutic growth promoter and antibacterial drug added to feed for starter pigs. Toxicity of CBX and at least one of its metabolites (bisdesoxycarbadox; DCBX) has resulted in a number of studies regarding its stability and residence time in edible swine tissue; however, little is known on its environmental fate pertinent to the application of antibiotic-laden manure to agricultural fields. We measured sorption of CBX and DCBX by soils, sediment, and homoionic clays from 10 mM KCl and 5 mM CaCl2 solutions, sorption of two N-oxide reduced metabolites (N4 and N1) by a subset of soils from 5 mM CaCl2, octanol-water partition coefficients (Kow) of CBX and all three metabolites, and CBX solubility in water and mixed solvents. Sorption appeared well-correlated to organic carbon (OC) for the soils (e.g., log (Koc, L/kg OC) ) 3.96 ( 0.18 for CBX). However, sorption was enhanced in the presence of K+, competitive sorption from the metabolites was observed, and sorption by clay minerals was large (≈ 105 L/kg for SWy-1). Sorption by clays was inversely correlated to surface charge density (e.g., sorption decreased from 105 to 10 L/kg as charge density increased from 1 to 2 µmolc/m2), similar to what has been observed for nitroaromatic compounds. In the absence of a clay surface, hydrophobic-type forces dominated with Kow values and reverse-phase chromatographic retention times increasing with the loss of oxygen from the aromatic nitrogens. Therefore, it is likely that both OC and clay contribute significantly to sorption of carbadox and related metabolites by soils with relative contributions most dependent on clay type.
swine tissue as well as its transport and transformation through the pig’s gastrointestinal tract (1-12). CBX has been shown to be rapidly metabolized in the pig’s liver and kidneys, producing both mono-N-oxides (methyl 3-[2-quinoxalinylmethylene]carbazate N4 and N1 oxides), DCBX, and quinoxaline-2-carboxylic acid (QCA) (Table 1). QCA has been reported to be the “principal metabolite” of CBX in pigs and has been shown to be non-mutagenic; however, both CBX and DCBX are hepatocarcinogens in rats (12). CBX was reported as not being detected in pig excreta, but the specific identities of the metabolites of CBX excreted by the pig other than QCA have not been published. The absolute concentrations of N4, N1, and DCBX in pig manure or in manure pits have not been reported, nor have their behavior or persistence in agricultural fields or in nearby waters or sediments been characterized. Although CBX has not been detected in manure (12), it can be introduced to manure pits by spilling of feed through the grating in the floors of the cages while pigs are rooting around in their food. Predicting the magnitude of sorption to soils and sediment and identifying the domains that control sorption can be important in assessing the risks associated with application of chemicals to agricultural fields. Little information is available in the literature about sorption of compounds containing aromatic N-oxides or about the contribution of the N-oxide oxygen to sorption relative to the lone electron pair of the reduced aromatic nitrogen. The N-oxide bond can be inferred to lower the octanol-water partition coefficient (Kow) of a compound by roughly an order of magnitude relative to the reduced aromatic nitrogen based on a comparison of quinoline and quinoline N-oxide Kow (13), but the effect of the N-oxide oxygen on water solubility (Sw), sorption, and ionizability is not clear. Therefore, determining the relationships between Sw, Kow, and Koc and evaluating competition by measuring sorption in the presence of related solutes can contribute valuable clues about the predominant sorption domains in soils, which can be useful in assessing bioavailability, recalcitrance, and mobility in the environment. In the current study, we measured sorption of CBX and DCBX by several soils, a sediment, and three homoionic clays from Ca2+ and K+ electrolyte solutions as well as sorption of CBX and reduced metabolites from a multiple-solute system in a subset of soils. We also determined Kow values and evaluated aqueous speciation of CBX and the reduced metabolites and measured CBX solubility in selected solvents. The reduced metabolites of CBX differ only in the absence of one or both N-oxides; therefore, by using the differences observed in their partitioning behavior between water and octanol, soils, and clays, we attempt to identify primary sorption domains and to elucidate the direct and/or indirect contributions of these N-oxides to sorption.
Introduction Carbadox (CBX) (methyl 3-[2-quinoxalinylmethylene]carbazate N1, N4 dioxide) is a chemotherapeutic growth promoter and antibacterial drug added to feed for starter pigs, an age class of pigs ranging in body weight from about 15 to 50 lb, at a concentration of 50 ppm for prevention of dysentery and improved feed efficiency. Because of the toxicity of CBX and at least one of its metabolites (bisdesoxycarbadox; DCBX), it has been subjected to a number of studies regarding its stability and residence time in edible * Corresponding author phone: (765)494-4772; fax: (765)496-2926; e-mail:
[email protected]. 3134
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 9, 2005
Materials and Methods Chemicals. CBX (CAS Registry No. 6804-07-5) was obtained from Sigma Chemical (St Louis, MO). All solutions with CBX and the N-oxide reduced metabolites were shielded from light due to their photosensitivity. Sodium dithionite (Na2S2O4) was obtained from Fisher Scientific (Hanover Park, IL). Acetonitrile (ACN) and methanol (MeOH) (Chrom AR HPLC) were from Mallinckrodt (Phillipsburg, NJ). Calcium chloride dihydrate (CaCl2), potassium chloride (KCl), and glacial acetic acid were purchased from Mallinckrodt and were of greater than 99% purity. All water was carbon-filtered and Millipore filtered to >18.0 µΩ‚cm. 10.1021/es048623q CCC: $30.25
2005 American Chemical Society Published on Web 04/02/2005
surface area. Kaolinite is a nonexpandable 1:1 clay with low CEC and low surface area. All clays were received in their Na-saturated form. Clays were then treated to be homoionic with K+ or Ca2+ at pH values approaching 4 and 8 to encompass the pH range of the soils used in this study. The clays were suspended in 50-mL polypropylene centrifuge tubes with 1 N calcium acetate or 1 N potassium acetate adjusted to the desired pH with acetic acid or NaOH. After 1 h equilibration, the samples were centrifuged (1750 RCF for 20 min) and decanted. This procedure was repeated four times, followed by three washes with 1 N CaCl2 or KCl. Two final washes were done with 0.01 N CaCl2 or 0.01 N potassium chloride. The sorbents were then air-dried for 48 h followed by oven drying at 65 °C for 18 h and gently crushed.
TABLE 1. Selected Properties of Carbadox and Reduced Metabolitesa
a Key: a, Ref 18. b, Ref 31. c, Single solute (CBX) system. d, Multisolute system (N4 reaction mixture). e, Calculated using SMILES in KowWin (21). f, Calculated using chemaxon.marvin.calculations.logPPlugin (22).
TABLE 2. Selected Soil Properties
soil EPA1a EPA14a Bloomfield-3b Drummer-7c Toronto-4b Eustis-25
pH in water organic CEC (1:2 g/ carbon clay sand silt (cmolc mL) (%) (%) (%) (%) kg-1) 7.3 4.3 6.4 6.9 4.4 5.5
0.22 0.48 0.36 2.39 1.30 0.76
6 64 8 36 20 4.2
94 2 81 18 12 91.6
0 34 11 46 68 4.2
1.1 18.9 4.4 24.1 11 nad
clay typee K > > S, V K > > S, V, I I>K S > > I, K S > > I, K K
a Means et al. (16). b Li and Lee (14). c Sampled in 1998 from the Purdue Agronomy Farm. OC determined by dry combustion. Cation exchange capacity determined by NH4OAc extraction at pH 7. Particle size analysis determined using hydrometer method (32, 33). d na, not available. e K, kaolinite; S, smectite; V, vermiculite; I, illite.
Soils. Five soils and one sediment previously used in several other studies and representing a range in physicochemical properties were selected (Table 2). Bloomfield-3, Toronto-4, and Drummer-7 soils are surface soils from Indiana (14); Eustis-25 soil is a surface soil from Gainesville, FL (15); EPA14 is a soil from an eroded hillside in southeast Ohio (16); and EPA1 is a freshwater sediment from the Mississippi River north of Monticello, MN. The soils and sediment (referenced as “soils” henceforth) were air-dried, gently crushed to pass a 2-mm sieve as necessary, thoroughly mixed, and stored in closed containers at room temperature prior to use. Soil characterization methods are detailed elsewhere (14-17). Clays. Clays used included Wyoming montmorillonite (Swy-1) and Arizona montmorillonite (SAz-1) standard reference minerals obtained from the Clay Mineral Society and a generic kaolinite sample. Montmorillonite is a dioctahedral smectite, which has is a group of expandable 2:1 clays with considerable cation exchange capacity (CEC) and large
Chemical Analysis. Analysis of sample concentrations was performed using external standard solutions and an automated Shimadzu high-performance liquid chromatography (HPLC) system equipped with a UV detector (SPD10A; λ ) 280 nm), a fluorescence detector (RF-10Axl; λex ) 287 nm and λem ) 310 nm) and a Supelcosil ABZ+ reversedphase column (4.6 × 150 mm, 120 Å pore size, 5 µm particles; Supelco, Bellefonte, PA) with a matching 3-cm guard column. Injection volumes were 10 µL. For simultaneous analysis of carbadox and its reduced metabolites, an isocratic mobile phase of 7/28/65 v/v/v ACN/MeOH/0.3% formic acid at a flow rate of 1 mL/min was used. All peaks were separable at this mobile phase composition with retention times of 4.2, 8.3, 10.6, and 17.1 min for CBX, N4, N1, and DCBX, respectively (Figure 1). When only CBX was present, a mobile phase of 15/85 v/v ACN/water at 1 mL/min sufficed with a CBX retention time of 5.5 min. In the clay studies, analysis of CBX and DCBX were performed using an isocratic mobile phase of 7/28/65 v/v/v and 10/32/58 v/v/v ACN/MeOH/ 0.3% formic acid, respectively, at a flow rate of 1 mL/min. Standard solutions were prepared in 1/1 ACN/MeOH. No matrix effect was observed through dilution and analysis of a stock solution containing CBX, N4, N1, and DCBX with either 1/1 ACN/MeOH, water, or the electrolyte matrixes used at either mobile phase composition. The limits of quantification (LOQ, 10× signal-to-noise) using UV detection for CBX, N4, N1, and DCBX were 35, 50, 27, and 38 µg/L, respectively. LOQ for CBX using fluorescence detection was 5 µg/L. Preparation of Carbadox Aromatic N-Reduced Metabolites. The N-oxide reduced metabolites were synthesized from CBX according to procedures described in Massy and McKillop (18). Briefly, DCBX was made by adding sodium dithionite to CBX in glacial acetic acid and stirring for 1 h. The yellow-brown reaction mixture was evaporated to dryness using a rotary evaporator. The solid was then washed with water, vacuum-dried, and placed in 1:1 toluene-ethanol containing activated charcoal. The charcoal was filtered out, and the remaining liquid was evaporated under N2 gas until a light brown solid precipitated. The precipitate was removed by filtration, vacuum-dried, and analyzed by HPLC. N4 was prepared by adding sodium dithionite to a dispersion of CBX in dilute sodium hydroxide. The solids were removed by filtration, washed with water, and resuspended in dilute sodium hydroxide. Titration of this mixture to pH 10.8 did not precipitate the N4 as described in Massy and McKillop (18); therefore, the solution was titrated to pH 8 and dried with a rotary evaporator. The solid was then vacuum-dried and analyzed by HPLC. The cleanup step for the N4 solid described in Massy and McKillop (18) was omitted. Solubility. CBX solubility was determined at 22 ( 2 °C in water as well as in 25, 50, 75, and 100 vol % aqueous acetonitrile solutions and 50/50 ACN/MeOH in duplicate on two different occasions. Trends in solubility with addition of an organic solvent can offer insight with regards to a VOL. 39, NO. 9, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3135
FIGURE 1. Chromatographic separation of a mixture of carbadox (CBX), N4-oxide (N4), N1-oxide (N1), and desoxycarbadox (DCBX) using a Supelco Supelcosil ABZ+ reversed-phase column (4.6 mm × 150 mm and 3 cm guard, 120 Å pore size) and a 7/28/65 acetonitrile/ methanol/0.3% formic acid mobile phase at 1 mL/min. compound’s sorption behavior and selection of extraction solvents. CBX crystals were added in excess to centrifuge tubes containing solvent, equilibrated for 5 days using an end-over-end rotator, centrifuged at 7000 rpm (∼8000g), and analyzed directly using HPLC-UV. Octanol-Water Partitioning Coefficients (Kow). Kow values were determined at 22 ( 2 °C following the method reported by Karickhoff and Brown (19) for CBX and for the N4 solid at various pH values (1). Octanol was extracted with 0.1 M NaOH, extracted twice with deionized water, and then distilled at 195 °C to eliminate residual water. In the determination of the Kow for CBX, extracted and distilled octanol was saturated with water preadjusted to the target pH values using 0.1% HCl or 0.1% NaOH. The water-saturated octanol phases were saturated with CBX and then equilibrated with the corresponding pH-adjusted octanol-saturated water at octanol:water volume ratios of 1:9 and 1:4 in triplicate and 2:3 in duplicate for 2 h (a total of eight determinations at each pH). Samples were centrifuged (1750g), and the aqueous phases were assayed directly using HPLC-UV with a mobile phase of 15/85 v/v acetonitrile/water using standards prepared in 1/1 v/v ACN/MeOH. Octanol phases were diluted 1/20 v/v 1/1 ACN/MeOH followed by a 1/10 v/v dilution with the HPLC mobile phase. Standards for analysis of the octanol phase were made in a similar fashion. Both sets of standards produced similar slopes (99% pure by HPLC-UV. Trace contamination by CBX and N4 were found in a stock solution made from this solid along with trace amounts of several unidentified peaks. The solid produced by the N4 synthesis procedure of Massy and McKillop (18) contained peaks with retention times matching CBX and DCBX, with two other peaks, the larger of which was presumed to be N4 and the smaller of which was presumed to be N1. CBX and DCBX were quantified using external CBX and DCBX calibration standards. After accounting for the weight of CBX and DCBX in the solid, the remaining mass was assumed to be N4 and N1, whose relative proportions were estimated by assuming that their UV absorption coefficients were the same. The absorption coefficient at 280 nm for DCBX is about 2.3 times higher than that for CBX; therefore, the difference in the absorption coefficients between the N1 and N4 oxides is likely less than a factor of 2. Repeated analysis of stock solutions prepared from the N4 reaction solid gave a composition of 0.18 g of CBX/g, 0.65 g of N4/g, 0.05 g of N1/g, and 0.13 g of DCBX/g of the N4 solid. The structures of these constituents were verified by LCelectrospray tandem MS using a minibore column of the same composition as the one used for HPLC-UV, producing molecular ion peaks corresponding to the respective parent ions in the order of elution expected from HPLC of CBX, DCBX, and N4 standards. The assignment of N4 and N1 was based on the assumption that N4 would be the principal product of the N4 synthesis procedure and on previously published chromatograms showing N1 to elute later than N4 (7). The structure of DCBX was also independently verified by gas chromatography/mass spectrometry and produced a mass spectrum with identical mass/charge peaks as those reported in Massy and McKillop (18).
Solubility. The average aqueous solubility of CBX was measured at 58 ( 4 mg/L at 22 ( 2 °C. Solubilities in ACN and MeOH are 89 ( 3 and 102 ( 21 mg/L, respectively. Massy and McKillop (18) also reported that carbadox solubility is low in organic solvents. CBX solubility is greater in binary solvent solutions than in either pure solvent. CBX solubility in 25, 50, and 75 vol % aqueous acetonitrile solutions are 263 ( 14, 629 ( 52, and 600 ( 22 mg/L, respectively, resulting in a parabolic solubility profile characteristic of compounds with log Kow values close to zero (20). A CBX solubility of 252 ( 8 mg/L was determined in the 50/50 ACN/MeOH extracting solutions. Octanol-Water Partition Coefficients (Kow). The Kow values determined for CBX, DCBX, N4, and N1 from the N4reaction mixture and for CBX in a single solute system are summarized in Table 1. No significant differences at the 95% confidence level were observed in Kow measurements as a function of pH for CBX or for the other N-oxide reduced metabolites; therefore, Kow values were averaged for each system (Table 1). Also no significant difference was observed in the average Kow of CBX between the single- and multisolute systems in the pH range of 2-8. For the N4 reaction mixture, the aqueous pH values at equilibrium were 2.15, 4.76, 5.99, 6.21, and 7.91, which varied from the initial targeted pH values. In the single-solute CBX Kow determination, pH values were not measured at equilibrium. Based on the Kow determinations for the mixture and the lack of difference in Kow at the highest pH as compared to the other pH values, the pH values at the time of the Kow measurement were likely different than the targeted values, especially for the highest pH. With removal of the N-oxides, Kow values increased as expected: DCBX > N1 > N4 > CBX, which correlated well to retention times (tRPLC) observed on a reversed-phase HPLC column (Figure 1): log tRPLC ) 0.35 log Kow + 0.56 with a goodness of fit (R2) of 0.995. The Kow values for CBX, N4, N1, and DCBX were also calculated based on constants assigned to the various functional groups using computer programs KowWin (21) and Chemaxon (22) (Table 1), which resulted in similar trends. Chemaxon predictions better matched those experimentally determined, although a lower Kow for the N1oxide was estimated relative to N4. KowWin did not differentiate the position of the N-oxide and predicted Kow values for CBX, N4, and N1 of about an order of magnitude lower than measured values. Speciation of CBX and Metabolites. To better understand speciation of CBX, N4, N1, and DCBX as a function of pH, the pKa estimator by Chemaxon (23) was used. The N-oxide functional groups in CBX, N1, and N4, the heterocyclic N1 and N4 in DCBX, and the N in the side chain (closest to the ring) were all estimated to become protonated under acidic conditions at pKa values below zero except the heterocyclic N4 in DCBX, which was estimated to have a pKa of 1.04. The pKa value for losing a proton from the NH group in the side chain and becoming negatively charged was estimated to be 9.61, 9.64, 9.75, and 9.77 for CBX, N1, N4, and DCBX, respectively. Therefore, even at a pH of 7.61 (CBX pKa - 2 pH units) only 1% of CBX (and less of the metabolites) exists as an anion in the aqueous phase. Chemaxon’s pKa estimates are in agreement with the absence of any pH dependence on the Kow values measured in the pH range between 2 and 8. Sorption by Soils. Multiple concentration sorption isotherms for CBX, N4, N1, and DCBX were generated from single- and multi-solute systems in various electrolyte solutions to determine the sorption magnitude and shed light on potential mechanisms. Soils reflected a range in characteristics that can be important to the sorption process including organic carbon (OC) content (0.22-2.39%), pH (4.3-7.3), clay content (1.8-64%), clay type (2:1 vs 1:1 layer clays), cation exchange capacity (CEC) (1.1-24.1 cmolc/kg), VOL. 39, NO. 9, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3137
FIGURE 2. Carbadox (CBX) sorption isotherms measured for CBX only (single), CBX and DCBX (binary), and the N4-reaction mixture containing CBX, N4, N1, and DCBX (N4 mixture) in 5 mM CaCl2 and CBX only from 10 mM KCl on Toronto-4 and Drummer-7 soils. and texture (sand to silty clay loam) (Table 2). Representative isotherms for CBX and DCBX are shown in Figures 2 and 3, respectively. Sorption model fits along with applied concentration ranges, m:V, and percent recoveries are summarized in Tables 3-5 for CBX, DCBX, and N1/N4, respectively. The percent recovered (mass recovered in aqueous and solvent extracts relative to applied) varied slightly over the Ci range and between solutes and soils, but recovery ranges were similar in the different electrolyte solutions for a given solute-soil combination. Chromatographic peaks were observed only for the compounds applied to the soils; no transformation products were observed. Freundlich model fits for the single-solute CBX data in 5 mM CaCl2 for 5 of the 6 soils produced N values close to or including unity in their 95% confidence intervals (Table 3). The one exception was the Eustis-25 soil (N ) 0.68), which had the lowest CBX sorption and highest sand content. Eustis soil is from an area that is frequently burned, thus contains small charcoal pieces that may have contributed to sorption nonlinearity. Using the linear Kd values (excluding Eustis25), an average log Koc of 3.96 ( 0.18 was estimated for CBX in 5 mM CaCl2. For DCBX, all sorption isotherms were nonlinear with Freundlich N values ranging between 0.58 and 0.84 (Table 4, Figure 3). Therefore, comparison of DCBX sorption to other solutes and between electrolytes was done at an equilibrium aqueous concentration (Cw) of 1 µg/mL, which is within the range for all DCBX-soil combinations and at an isotherm point where Kf and Kd values can be compared independent of linearity (Kd ) KfCwN-1; therefore, at Cw ) 1 µg/mL, Kd ) Kf). DCBX was generally sorbed 4050% less than CBX with a resulting log Koc value of 3.78 ( 0.17. Sorption magnitudes of the mono N-oxides, which were only measured from the multi-solute system in 5 mM CaCl2 (Table 5), were similar to those observed for CBX and DCBX. Sorption magnitude and nonlinearity of N4 was almost identical to what was observed for DCBX. For N1, which was only about 5% of the N4-reaction mixture, aqueous-phase concentrations of N1 at equilibrium were often near the 10× 3138
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 9, 2005
FIGURE 3. Desoxycarbadox (DCBX) sorption isotherms measured for DCBX only (single), CBX and DCBX (binary), and the N4-reaction mixture containing CBX, N4, N1, and DCBX (N4 mixture) in 5 mM CaCl2 on Bloomfield-3, Toronto-4, and Drummer-7 soils. signal-to-noise threshold, resulting in greater scatter between replicates and large standard errors, thus limiting further comparison with other solutes. For three soils, sorption of CBX and DCBX was also measured from 5 mM CaCl2 solutions containing multiple solutes and from single-solute 10 mM KCl solutions. CBX sorption from 5 mM CaCl2 decreased 10-35% in the presence of the N-oxide reduced metabolites (Table 3, Figure 2) with the effects being significant at the 95% confidence level for the Bloomfield-3 and Toronto-4 soils (not Drummer-7). Conversely, competitive sorption of DCBX was only significant in the Drummer-7 soil with sorption suppressed by ≈30% (Table 4, Figure 3). Sorption was also affected by the electrolyte matrix (10 mM KCl vs 5 mM CaCl2). CBX sorption from a KCl solution was greater by 117% (i.e., factor of 2.17), 119%, and 46% on Toronto-4, Drummer-7, and EPA14 soils, respectively. Similarly, but to a lesser extent, DCBX sorption from KCl was 77%, 11%, and 37% greater with Toronto-4, Drummer-7, and EPA14 soils, respectively (Table 4). Suppression of sorption in the presence of similar solutes and the inorganic cation effect on sorption are indicative of site-specific interactions involving soil cation exchange sites. Since CBX and its N-oxide reduced metabolites exist as
TABLE 3. Linear and Freundlich Model Fits to CBX Isotherms Measured from 5 mM CaCl2 or 10 mM KCl Solutions soil
solute system
Bloomfield-3 single
applied concn range (mg/L) 0.1-10.4
electrolyte solution
m:V % (g/mL) recovered
5 mM CaCl2 2:35
binary with DCBX 0.9-18.7
5 mM CaCl2 1:10
N4 mixture
1.0-7.6
5 mM CaCl2 1:10
single
0.1-11.6
5 mM CaCl2 2:35
binary with DCBX 0.9-18.7
5 mM CaCl2 1:20
N4 mixture
1.0-7.6
5 mM CaCl2 1:20
single
1.0-4.0
10 mM KCl
single
0.1-10.4
5 mM CaCl2 2:35
binary with DCBX 0.9-18.7
5 mM CaCl2 1:20
N4 mixture
1.0-7.6
5 mM CaCl2 1:30
single
1.0-4.0
10 mM KCl
single
0.1-11.6
5 mM CaCl2 2:35
single
1.0-4.0
10 mM KCl
EPA1
single
0.12-14.5
5 mM CaCl2 2:35
Eustis-25
single
1.5-7.6
5 mM CaCl2 2:35
Toronto-4
Drummer-7
EPA14
a
Standard error.
b
1:30
1:30
1:30
Kd ( SEa (R2)
27.0 ( (0.999) 92-104 20.7 ( 0.2 (0.995) 98-110 19.8 ( 0.5 (0.985) 103-106 154 ( 2 (0.998) 89-96 111 ( 2 (0.990) 95-106 99.2 ( 2.8 (0.982) 80-91 335 ( 19 (0.907) 68-83 144 ( 1 (0.999) 79-94 117 ( 1 (0.996) 80-90 118 ( 2 (0.995) 78-87 316 ( 7 (0.985) 97-104 80.8 ( 0.6 (0.999) 93-96 118 ( 2 (0.993) 88-100 16.1 ( 0.1 (0.999) 80-105 1.4( 0.1 (0.984)
Kf b (R2)
0.3c
93-103
25.1
(24.5-25.7)c
(1.000) 21.1 (20.4-21.9) (0.998) 18.8 (17.8-20.0) (0.989) 141 (129-159) (0.998) 105 (100-112) (0.992) 100 (91.2-107) (0.982) 243 (195-302) (0.980) 135 (126-148) (0.997) 120 (115-126) (0.995) 115 (95.5-138) (0.979) 251 (219-295) 0.992) 79 (74-83) (0.999) 115 (109-124) (0.996) 15.8 (12.0-21.4) (0.994) 2.5 (2.3-2.8) (0.981)
N 1.05 (1.04-1.07)c 1.00 (0.97-1.03) 1.06 (0.996-1.13) 0.98 (0.94-1.02) 1.1 (1.05-1.16) 1.06 (0.97-1.15) 0.76 (0.66-0.87) 0.89 (0.84-0.94) 0.89 (0.86-0.93) 0.92 (0.76-1.07) 0.84 (0.76-0.91) 1.02 (0.996-1.05) 0.96 (0.895-1.02) 1.07 (0.93-1.22) 0.65 (0.58-0.72)
µg1-N mLN g-1. c 95%low, 95%high.
TABLE 4. Freundlich Model Fits to DCBX Isotherms Measured from 5 mM CaCl2 or 10 mM KCl Solutions solute system
soil Bloomfield-3
Toronto-4
Drummer-7
EPA14
a
µg1-N mLN g-1.
applied concn range (mg/L)
electrolyte solution
m:V (g/mL)
% recovered
single
1.0-4.0
5 mM CaCl2
1:30
93-97
binary with CBX
0.9-19.1
5 mM CaCl2
1:10
91-98
N4 mixture
1.3-10.4
5 mM CaCl2
1:10
89-105
single
1.0-4.0
5 mM CaCl2
1:30
76-85
binary with CBX
0.9-19.1
5 mM CaCl2
1:20
88-93
N4 mixture
1.3-10.4
5 mM CaCl2
1:20
68-88
single
1.0-4.0
10 mM KCl
1:30
82-89
single
1.0-4.0
5 mM CaCl2
1:30
86-96
binary with CBX
0.9-19.1
5 mM CaCl2
1:20
91-100
N4 mixture
1.3-10.4
5 mM CaCl2
1:30
89-91
single
1.0-4.0
10 mM KCl
1:30
86-94
single
1.0-4.0
5 mM CaCl2
1:30
68-73
single
1.0-4.0
10 mM KCl
1:30
66-74
b
Kf a (R2) 12.9 (12.6-13.5) b (0.996) 12.6 (12.2-12.9) (0.998) 11.2 (10.3-12.0) (0.981) 85.1 (82.8-87.5) (0.999) 82.4 (77.6-87.1) (0.989) 75.9 (72.4-79.4) (0.930) 151 (115-199) (0.972) 143 (132-155) (0.995) 123 (120-127) (0.999) 100 (91.2-110) (0.992) 159 (151-162) (0.999) 43.7 (42.7-44.7) (0.999) 60.0 (57.5-63.1) (0.998)
N 0.8 (0.74-0.86)b 0.71 (0.69-0.73) 0.80 (0.74-0.87) 0.77 (0.75-0.80) 0.76 (0.72-0.81) 0.79 (0.74-0.84) 0.83 (0.63-1.02) 0.64 (0.59-0.68) 0.63 (0.62-0.64) 0.61 (0.55-0.67) 0.64 (0.62-0.67) 0.84 (0.82-0.87) 0.78 (0.74-0.82)
95%low, 95%high.
uncharged molecules in the aqueous phase within the soil pH range represented with the pKa values of the acidic nitrogen atoms being e1, direct cation exchange is unlikely. Uncharged nitroaromatic compounds have also been observed to be sorbed much more strongly (several orders of
magnitude) to K-dominated clays and subsurface soils as compared to those same sorbents saturated with Ca (2428). The mechanisms proposed include (i) electron donoracceptor (EDA) interactions between the electron-accepting solutes and the electron-donating siloxane clay surface (26); VOL. 39, NO. 9, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3139
TABLE 5. Freundlich Model Fits to N4- and N1-Oxide Isotherms Measured from 5 mM CaCl2 and Containing CBX, DCBX, N4, and N1 at the Ratios Present in the N4-Reaction Mixture
N4
N1
a
m:V (g/mL)
applied concn range (mg/L)
soil Bloomfield-3
1.0-8.0
1:10
94-107
Toronto-4
1.0-8.0
1:20
92-102
Drummer-7
1.0-4.0
1:30
80-97
Bloomfield-3
0.07-0.57
1:10
91-118
Toronto-4
0.07-0.57
1:20
88-107
Drummer-7
0.08-0.31
1:30
53-59
µg1-N mLN g-1.
b
Kf a (R2)
% recovered 13.2
N
(12.6-13.8)b
(0.987) 79.4 (75.9-81.3) (0.995) 97.7 (87.1-110) (0.987) 16.2 (9.5-26.9) (0.896) 151 (64.6-339) (0.93) 85.1 (56.2-129) (0.988)
0.82 (0.77-0.88)b 0.81 (0.77-0.84) 0.67 (0.58-0.76) 0.88 (0.69-1.07) 1.01 (0.76-1.26) 0.80 (0.70-0.90)
95%low, 95%high.
TABLE 6. Sorption Coefficients (Kd, L/kg) for Carbadox and Desoxycarbadox Determined in Triplicate from a Single Solute System at One Concentration Prepared at pH Values Approaching 4 and 7 in 5 mM CaCl2 for the Ca-Clays and 10 mM KCl for the K-Clays carbadox
desoxycarbadox
claya
cation
pH
Kd (L/kg)
pH
Kd (L/kg)
Swy1 82 cmolc/kgb 1.09 µmolc/m2 b 0.01:35 g/mL
Ca Ca K K
4.22 6.75 4.34 6.93
45 900 ( 1700 30 600 ( 1600 98 600 ( 30 000 21 800 ( 1400
4.62 6.84 4.85 6.70
10 000 ( 500 6700 ( 400 160 000 ( 82 000 27 000 ( 1800
SAz1 125 cmolc/kgb 1.67 µmolc/m2 b 0.1:35 g/mL
Ca Ca K K
5.22 6.95 5.24 7.23
83.6 ( 1.8 75.7 ( 2.9 3260 ( 100 21 200 ( 2220
5.80 6.84 5.89 6.94
6.8 ( 0.6 24.2 ( 8.6 230 ( 40 750 ( 30
kaolinite ≈3 cmolc/kgc ≈3 µmolcm2 c 0.25:6 g/mL
Ca Ca K K
4.63 6.90 4.68 6.61
19.8 ( 9.7 4.8 ( 0.5 28.1 ( 0.6 31.6 ( 1.1
4.67 6.74 4.69 6.36
7.6 ( 3.6 1.4 ( 0.4 11.7 ( 0.5 9.4 ( 0.5
a Values reported under clay type is for cation exchange capacity (cmol /kg), charge density (µmol /m2), and the clay mass (g):solution volume c c (mL) at which sorption was measured. b Reported by Weissmahr et al. (26) for a reference kaolinite. c Reported by Sheng et al. (25) for Na-saturated Swy-2 and SAz-1
and (ii) inner-sphere complexation of the nitro group directly with the cation on the surface with the ring being stabilized by the neutral siloxane surface (28). In the former case, access to the siloxane surface is hindered in the presence of highly hydrated cations such as Ca2+, thus sorption is greater when K+ is on the exchange sites. In the case of an inner-sphere complexation mechanism, the large hydration shell of Ca2+ hinders inner-sphere complexation as compared to K+. Characteristic of both hypotheses is that sorption would decrease with increasing charge density (ratio of CEC to surface area), which has been observed (28, 29). Sorption by K- and Ca-Saturated Clays. To better ascertain mechanisms and the potential contribution of the clay fraction to site-specific sorption in soils and sediment, single-concentration Kd values were estimated for CBX and DCBX on three clays that vary in their charge densities and were prepared to be homoionic with Ca2+ or K+ at pH values approaching 4 and 7 (Table 6). For all cation-clay-pH combinations, no transformation of CBX or DCBX was observed and mass balance of the parent compounds was 100% in all cases. The low charge density Swy-1 clay generally exhibited a greater affinity for CBX and DCBX than did the higher charge density clays; very high sorption by Swy-1 resulted in large standard errors about the average Kd values in some cases. For the high charge density kaolinite, sorption was general quite low as compared to the montmorillonite clays. Across all clay-cation-pH combinations, CBX sorption was much higher than for DCBX with the exception of the 3140
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 9, 2005
K-Swy-1. For both CBX and DCBX on all three clay types, sorption was greatest on the K-clays with the exception of CBX at the higher pH on Swy-1. The greatest impact of the exchange cation on sorption was consistently on Saz-1. The effect of pH varied across clays. For Swy-1 and kaolinite, sorption was usually greater at the lower pH or there was little pH effect (e.g., CBX sorption by K-kaolinite). For the SAz-1 clay, sorption was generally greater at the higher pH or there was little pH effect (e.g., CBX sorption by Ca-SAz1). Interpretation of pH is complicated by the fact that some protons will exchange with Ca2+ and K+; the extent of exchange is both cation and clay dependent.
Discussion The magnitude of sorption exhibited by pure clays cannot be directly compared to the clay within a whole soil due to the presence of other sorption domains and the interactions between those domains (e.g., organic matter coatings on clay surfaces). However, several semi-quantitative comparisons can be made. On both soils and clays, CBX sorption was generally higher than DCBX, although greater differences were observed with the clays. For the kaolinitic soils (Eustis25, EPA1, Bloomfield-3, EPA14), sorption was less than for the smectite soils as observed between the kaolinite and the montmorillonite clays and was positively correlated to percent of clay. The kaolinitic soils were also coincidentally lower in OC than the smectite soils. Sorption was greatest by the two smectite-rich soils (Toronto-4 and Drummer-7),
which also had higher OC contents. K+-enhanced sorption on soils is also consistent with the trends observed on the clay minerals except in magnitude, which was much greater on the clays. Soil clays are typically not as crystalline as pure reference clays and contain a mix of clay types. Even for soils denoted as having primarily smectite clays, multiple types of smectite are likely present. Charge densities for smectites can vary considerably as exemplified by the two smectite clays used in this study. Also the soils used in this study were not made homoionic; therefore, not all native soil ions may have been exchanged with the applied electrolyte cation. Although sorption of CBX and DCBX is clearly affected by the resident cation, the actual sorption mechanism is still not clear. K-enhancement was observed whether or not the N-oxide groups were present; however, the magnitude of the effect was solute-sorbent dependent. For DCBX (no N-oxides), K-enhancement on the low charge density Swy-1 was several times greater than that observed for CBX. CBX sorption by Ca-SWy-1 was higher than for DCBX, but with K+ on the exchange sites, DCBX sorption exceeded that of CBX. A switch in sorption order did not occur for the other two clays. Both the N-oxides in the ring and the side chain nitrogens could serve in EDA interactions with the siloxane surface as well as have some direct complexation with the surface cations. Spectroscope studies are required to further delineate the relative contribution of the various nitrogen atoms. Organic matter and clay minerals are considered to be the two most reactive domains in soils. However, differentiating relative contributions between soil organic matter and clay is difficult at best, especially with only macroscopic study. In the absence of a clay surface, as is the case with partitioning between octanol and water (Kow) and between reverse-phase liquid chromatographic (RPLC) and water, CBX and the N-oxide reduced metabolites followed the trends expected based on chemical structure and assuming only hydrophobic processes: CBX < N4 < N1 < DCBX with a strong positive log-log correlation between Kow and tRPLC. Likewise, it is not unreasonable to assume that organic matter will also contribute significantly to sorption of CBX and its reduced metabolites. For many organic compounds containing weakly polar functional groups, sorption often appears to be well-correlated to soil OC and thus is often assumed to be the controlling sorption domain (30). The percent of OC of the six soils used in the study are coincidentally highly correlated to percent of clay (R2 ) 0.90 excluding the high clay EPA14 soil). Therefore, the apparent correlation between percent of OC and the sorption of both CBX and DCBX as reflected in relatively small standard deviations about the estimated log Koc value may be fortuitous. Without the additional studies with multiple solutes and different electrolytes, which is not typically done in characterization for regulatory purposes, OC-controlled sorption may have been concluded for these compounds. Environmental Significance. For many agricultural chemicals, sorption to soil OC through hydrophobic-type processes appears to describe much of the sorption observed across a multitude of soils (26) such that Koc-Kow correlations and soil OC are sufficient for reasonable predictions of sorption and transport. For carbadox (CBX) and its metabolites, sorption appeared well-correlated to organic carbon (OC) for the soils; however, sorption was enhanced in the presence of K+, competitive sorption from the metabolites was observed, and sorption by clay minerals was large. Both OC and clay appear to contribute significantly to sorption of carbadox and related metabolites by soils with relative contributions most dependent on clay type. Therefore, typical Koc-Kow correlations are likely to be insufficient for predicting sorption and transport of these antibiotics. Application of these antibiotic residues to soil in a manure matrix, which
will likely have a different cation composition than the native soils, along with the potential liming of fields after manure application further cloud predictions. Nonetheless, sorption data for CBX and its metabolites suggest that these compounds if present in the manure will be reasonably sorbed in fields where antibiotic-laden manure is applied, thus transport to groundwater and tile drains is likely to be minimal under most conditions.
Acknowledgments This work was funded in part by the U.S. Environmental Protection Agency National Risk Management Research Laboratory (Cincinnati, OH) under Cooperative Agreement 82811901-0 and the College of Agriculture, Purdue University.
Literature Cited (1) De Graaf, G. J.; Jager, L. P.; Baars, A. J.; Spierenburg, T. J. Some pharmacokinetic observations of carbadox medication in pigs. Vet. Q. 1988, 10, 34-41. (2) Aerts, M. M. L.; Beek, W. M. J.; Keukens, H. J.; Brinkman, U. A. Th. Determination of residues of carbadox and some of its metabolites in swine tissues by high-performance liquid chromatography using on-line pre-column enrichment and postcolumn derivatization with UV-Vis detection. J. Chromatogr. 1988, 456, 105-119. (3) Baars, A. J.; Jager, L. P.; Spierenberg, Th. J.; de Graaf, G. J. Residues of carbadox metabolites in edible pork products. Recent developments in toxicology: trends, methods and problems. Arch. Toxicol. Suppl. 1991, 14, 288-292. (4) Lauridsen, M. G.; Lund, C.; Jacobsen, M. Determination and depletion of residues of carbadox, tylosin, and virginiamycin in kidney, liver, and muscle of pigs in feeding experiments. J. Assoc. Off. Anal. Chem. 1988, 71, 921-925. (5) De Graaf, G. J.; Spierenburg, Th. J. Liquid chromatographic determination of carbadox and desoxycarbadox in medicated feeds and in porcine gastro-intestinal tract. J. Assoc. Off. Anal. Chem. 1985, 68, 658-660. (6) Lynch, M. J.; Mosher, F. R.; Schneider, R. P.; Fouda, H. G.; Risk, J. E. Determination of carbadox-related residues in swine liver by gas chromatography/mass spectrometry with ion trap detection. J. Assoc. Off. Anal. Chem. 1991, 74, 611-618. (7) Binnendijk, G. M.; Aerts, M. M. L.; Keukens, H. J.; Brinkman, U. A. Th. Optimization and ruggedness testing of the determination of residues of carbadox and metabolites in products of animal origin. J. Chromatogr. 1991, 541, 401-410. (8) Rose, M. D.; Bygrave, J.; Tarbin, J. A. Determination of quinoxaline carboxylic acid (metabolite of carbadox) in animal tissue by HPLC. Food Addit. Contam. 1995, 12, 177-183. (9) MacIntosh, A. I.; Lauriault, G.; Neville, G. A. Liquid chromatographic monitoring of the depletion of carbadox and its metabolite desoxycarbadox in swine tissues. J. Assoc. Off. Anal. Chem. 1985, 68, 665-671. (10) Hutchinson, M. J.; Young, P. Y.; Hewitt, S. A.; Faulkner, D.; Kennedy, D. G. Development and validation of an improved method for confirmation of the carbadox metabolite, quinoxaline-2-carboxylic acid, in porcine liver using LC-electrospray MS-MS according to revised EU criteria for veterinary drug residue analysis. Analyst 2002, 127, 342-246. (11) Miao, X. S.; March, M. E.; Metcalfe, C. D. A tandem mass spectrometric study of the N-oxides, quinoline N-oxide, carbadox and olaquindox, carried out at high mass accuracy using electrospray ionization. Int. J. Mass Spectrom. 2003, 230, 123133. (12) Freedom of Information Summary NADA 041-061 [online]. 1998. Available at http://www.fda.gov/cvm/efoi/section1/ 041061s013098.htm (accessed June 15, 2004). (13) Kollist-Siigur, K.; Nielsen, T.; Gron, C.; Hansen, P. E.; Helweg, C.; Jonassen, K. E. N.; Jorgensen, Ole; Kirso, U. Sorption of polycyclic aromatic compounds to humic and fulvic acid HPLC column materials. J. Environ. Qual. 2001, 30, 526-537. (14) Li, H.; Lee, L. S. Sorption and abiotic transformation of aniline and R-naphthylamine by surface soils. Environ. Sci. Technol. 1999, 33, 1864-1870. (15) Lee, L. S.; Rao, P. S. C.; Nkedi-Kizza P.; Delfino, J. J. Influence of solvent and sorbent characteristics on distribution of pentachlorophenol in octanol-water and soil-water systems. Environ. Sci. Technol. 1990, 24, 1654-661. VOL. 39, NO. 9, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3141
(16) Means, J. C.; Wood, S. G.; Hassett, J. J.; Banwart, W. L. Sorption of polynuclear aromatic hydrocarbons by sediments and soils. Environ. Sci. Technol. 1980, 14, 1524-1528. (17) Lee, L. S.; Nyman, A. K.; Li, H.; Nyman, M.; Jafvert, C. Initial sorption of aromatic amines by surface soils. Environ. Toxicol. Chem., 1997, 16, 1575-1582. (18) Massy, D. J. R.; McKillop, A. Selective monodeoxygenation of methyl-3-(2-quinoxalinylmethylene)carbazate N1, N4-dioxide (carbadox). Synthesis 1996, 1477-1480. (19) Karickhoff, S. W.; Brown, D. S. Determination of Octanol/Water Distribution Coefficients, Water Solubilities and Sediment/Water Partition Coefficients for Hydrophobic Organic Pollutants; EPA600/4-79-032; U.S. EPA: Athens GA, 1979. (20) Lee, L. S.; Bellin, C. A.; Pinal, R.; Rao, P. S. C. Cosolvent effects on sorption of organic acids by soils from mixed solvents. Environ. Sci. Technol. 1993, 27, 165-171. (21) Syracuse Research Corporation PhysProp Database [online]. Available at http://esc.syrres.com/interkow/webprop.exe?CAS) 91-19-0 (accessed July 21, 2004). (22) chemaxon.marvin.calculations.logPPlugin. ChemAxon Academic Package from chemaxon.com. (23) chemaxon.marvin.calculations.pKaPlugin. ChemAxon Academic Package from chemaxon.com. (24) Haderlein, S. B.; Weissmahr, K. W.; Schwarzenbach, R. P. Specific adsorption of nitroaromatic explosives and pesticides to clay minerals. Environ. Sci. Technol. 1996, 30, 612-621. (25) Sheng, G.; Johnston, C. T.; Teppen, B. J.; Boyd, S. A. Adsorption of dinitrophenol herbicides from water by montmorillonites. Clays Clay Miner. 2002, 50, 25-34. (26) Weissmahr, K. W.; Haderlein, S. B.; Schwarzenbach, R. P. Complex formation of soil minerals with nitroaromatic explosives and other π-acceptors. Soil Sc. Am. J. 1998, 62, 369-378.
3142
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 9, 2005
(27) Weissmahr, K. W.; Hildenbrand, M.; Schwarzenbach, R. P.; Haderlein, S. B. Laboratory and field scale evaluation of geochemical controls on groundwater transport of nitroaromatic ammunition residues. Environ. Sci. Technol. 1999, 33, 25932600. (28) Johnston, C. T.; de Oliveira, M. F.; Teppen, B. J.; Sheng, G.; Boyd, S. A. Spectroscopic study of nitroaromatic-smectite sorption mechanisms. Environ. Sci. Technol. 2001, 35, 37674772. (29) Johnston, C. T.; Boyd, S. A.; Teppen, B. J.; Sheng G. Sorption of nitroaromatic compounds on clay surfaces. In Handbook of Layered Materials for Catalytic Applications; Auerbach, S. M., Carrado, K. A., Dutta, P. K., Eds.; Marcel-Dekker: New York, 2004; pp 155-189. (30) Gerstl, Z. Estimation of organic chemical sorption by soils. J. Contam. Hydrol. 1990, 6, 357-375. (31) Kuhla, D. E. U.S. Patent 4006142, 1977; Chem. Abstr. 1977, 86, 190007. (32) Sparks, D. L., Ed. Methods of Soil Analysis: Part 1-Physical and Mineralogical Method, 2nd ed.; Agronomy Monograph 9; American Society of Agronomy: Madison, WI, 1986. (33) Sparks, D. L., Ed. Methods of Soil Analysis: Part 3-Chemical Methods, 2nd ed.; Agronomy Monograph 5; American Society of Agronomy: Madison, WI, 1996.
Received for review September 3, 2004. Revised manuscript received February 10, 2005. Accepted February 25, 2005. ES048623Q
