Model ecosystem evaluation of the environmental impacts of the veterinary drugs phenothiazine, sulfamethazine, clopidol, and diethylstilbestrol.

Four veterinary drugs of dissimilar chemical structures were evaluated for environmental stability and penchant for bioaccumulation. The techniques used were (1) a model aquatic ecosystem (3 days) and (2) a model feedlot ecosystem (33 days) in which the drugs were introduced via the excreta of chicks or mice. The model feedlot ecosystem was supported by metabolism cage studies to determine the amount and the form of the drug excreted by the chicks or mice. Considerable quantities of all the drugs were excreted intact or as environmentally short-lived conjugates. Diethylstilbestrol (DES) and Clopidol were the most persistent molecules, but only DES bioaccumulated to any appreciable degree. Phenothiazine was very biodegradable; sulfamethazine was relatively biodegradable and only accumulated in the organisms to very low levels. Data from the aquatic model ecosystem demonstrated a good correlation between the partition coefficients of the drugs and their accumulation in the fish.

produced are used for agricultural purposes, primarily as feed additives, and two-thirds of the 60 million tons of feed produced commercially contain medication, with 75% of these requiring legal withdrawal times before the treated animals can be marketed.
In addition to the deliberate additives listed above there are accidental additives which contaminate feed such as polychlorinated biphenyls (PCBs), persistent organochlorine insecticides, plasticizers, and flame retardants (PBBs).
The environmental fates and degradative pathways for nearly all of these substances are little known as are the possible levels of toxic effects, bioconcentration factors, and food chain relationships of the parent compound and its metabolites on the living elements of the ecosystem. The model ecosystem studies discussed here were developed to model the environmental impact of a feed lot on a sewer, an adjacent pond, or other aquatic drainage. Four representative radiolabeled veterinary drugs, the anthelmintic phenothiazine, the coccidiostat Clopidol, the bacteriostat sulfamethazine, and the growth promoter diethylstilbestrol were chosen for evaluation; the asterisks (*) in the structure denote the radiolabels.

Radioassay
Liquid scintillation was used for analysis of the concentrations of radiolabeled compounds in samples of water, feces, urine, and tissues of organisms. The cocktail was composed of 100 g naphthalene, 5 g diphenyl oxazole (PPO), made up to 1 liter with 1,4-dioxane. Quench corrections were made by using the channels ratio method (3). Radioautographs of thin layer chromatography plates (0.25 mm thick, fluorescent silica gel GF-254 from E. Merck) were made by using Eastman no-screen x-ray film. The plates were evaluated quantitatively by scraping 1 cm x 2 cm sections or fluorescent spots into vials of scintillation fluid and counted.

Model Metabolites
Preparations of the model compounds were as follows: DES-mono-p-D-glucuronide was isolated from rabbit urine (4); the structure was confirmed by mass spectromety. Acetyl-DES was prepared by using pyridine and acetic anhydride per equivalent of DES. DES was obtained from Sigma Chemical Company. Phenothiazine sulfoxide was prepared by adding one equivalent of H202 to phenothiazine in acetone solution and allowing the sulfoxide to crystallize slowly. The product decomposed at 240°C; the literature value is 250°C (5). Infrared spectroscopy revealed the sulfoxide bond stretching absorbance at 1078 cm-'.
Phenothiazine sulfone was prepared by using peracetic acid as the oxidizing agent (6). The compound melted at 260NC (literature mp 258°C), and the compound absorbed in the infrared at 1157 and 1288 cm-'.
Phenothiazone was obtained from Dr. G. D. Koritz, and was prepared by the oxidation of phenothiazine with FeCl3 (7). Phenothiazine was obtained from Eastman Chemical Company.
Clopidol was provided by Dow Chemical Company.
N4-acetyl sulfamethazine was prepared by using acetic anhydride in acetic acid (8). N4methyl sulfamethazine was synthesized by re-fluxing sulfamethazine in methanol with KOH and excess methyl iodide. Sulfamethazine was furnished by Dr. R. F. Bevill. The compounds and their derivatives were separated on TLC plates by using the solvent systems listed in Table 1. Chromogenic detection methods were also employed as aids in locating model metabolites on TLC plates (Table 1).

Toxicity Methodology
The compounds and their model metabolites were each tested for lethal effects to the species of organisms to be used in the model ecosystem. Three-liter glass containers were each filled with two liters of standard reference water. Various predetermined concentrations of compound were added in small volumes of appropriate solvents (acetone, methanol, or water), and air was bubbled in for 8 hr. The organisms were added and observed for lethal or toxic effects after 24 and 48 hr.

Aquatic Model Ecosystem
A 3-day, 2-liter aquatic model ecosystem was used to determine relative uptake and degradation of each compound (9). The ecosystem con-tained 2 liters of standard reference water and the following organisms: Oedogonium cardiacum, Daphnia magna, Culex pipiens quinquefasciatus, Physa sp., and Gambusia affinis. Following equilibration, the compounds were added in minimum amounts of appropriate solvents. Two days later the system was dismantled, organisms were analyzed by grinding and extracting with appropriate solvents, theh combusting the residues to determine unextractable radioactivity. The water was extracted, then HCl was added until pH 2 was reached; refluxing and extraction produced a water-hydrolyzed fraction.

Dosing
Labeled compounds were administered orally in olive oil to Swiss white mice: '4C-DES at 0.5 mg/kg, "C-phenothiazine at 2 mg/kg, 35Ssulfamethazine at 100 mg/kg. One day old chicks were fed 0.0125% '4C-Clopidol in their feed, and injected subcutaneously at 0.05 mg/kg "'C-DES in propylene glycol.

Metabolism Cages
Mouse feces and urine were collected and separated within the "econo metabolism unit"  (10). The daily urine samples were made up to 25 ml with methanol and two 1 ml aliquots were taken for counting. Extraction of feces and urine followed the usual procedures (11), utilizing appropriate solvents.

Model Feedlot Ecosystem
A 33-day terestrial-aquatic model ecosystem was used to follow the qualitative and quantitative fate of the drugs being evaluated. The system was comprised of 15 kg of white quartz sand and 7 liters of standard reference water in a 10-gal aquarium. The biotic components of the system were: Sorghum vulgare, the alga Oedogonium cardiacum, the water flea Daphnia magna, the mosquito larva Culex pipiens quinquefasciatus, the snail Physa sp., the mosquito fish Gambusia affinis, and a complement of microbes and zooplankton. The detailed methodology for the model ecosystem has been described previously (12). The following modifications were made to facilitate the tracing of a veterinary drug through the model ecosystem: a 10 cm x 18 cm x 27 cm cage constructed from 0.25 in. wire mesh was supported from the top of the aquarium using glass rods, three mice or baby chicks, dosed as described for the metabolism cage'study, were placed in the cage and supplied with food and water. The cage was positioned over the sandwater interface to allow optimal input of excretory products into the aqueous phase without inciting an extensive algal bloom. The excretion rate data from the metabolism cage experiment determined the duration of the animals' confinement over the model ecosystem. The most suitable location for the cage was where one-fourth of the excrement fell directly into the water, and three-fourths onto the terrestrial phase (Fig. 1). The maximum allowable excretory input was that of three 1-day chicks or three 20-g mice in the suspended cage for 3 days following treatment. The daphnia were quite sensitive to excess chick excrement in the water. The system was maintained at constant temperature (25 + 1°C) and photoperiod (12 hr diurnal cycle of 5000 ftcandles) in a Percival environmental plantgrowth chamber. The level of radioactivity was monitored by withdrawal of 1 ml aliquots of water for counting. On day 26, 300 Culex larvae were added; on day 30, 50 of them were removed, as well as 50 Daphnia, for quantitative and qualitative analysis of radiolabel in their bodies. Three Gambusia were added to feed on the remaining Culex larvae and Daphnia for 3 days. On day 33 Gambusia, algae, and snails were removed for analysis. Extraction procedures were as described above for the aquatic ecosystem. One liter of twice-filtered water was extracted with a solvent, taken to pH 2 with concentrated HCl, and refluxed, and extracted again, to yield fractions called "water-unhydrolyzed" and "water-hydrolyzed". The identification of metabolites using TLC was then completed for extracts of the water and all organisms in the model ecosystem.

Water Solubility and Partition Coefficients
The value for water solubility of phenothiazine and DES was determined by radioassay at 25°C; determinations for Clopidol and sulfamethazine utilized unlabeled drugs. Distilled water. was used, and the pH of the water was not controlled. Partition coefficients were determined by using a system of 1-octanol and water (13). Table 2 gives these physicochemical parameters for the four drugs studied.

Results and Discussion
Toxicity Tests The evaluation of the drugs and their primary metabolites for acute lethal effects on the biotic components of the model ecosystem produced Table 3. Phenothiazine and N4-methyl sulfamethazine exhibited some degree of toxicity.

Metabolism Cage Studies
The comparison of excretion of DES by orally dosed mice and subcutaneously injected chicks is presented in Table 4. The dose injected into the chicks was eliminated considerably more slowly. The degree of metabolism by both animals is shown in Tables 5 and 6. A large proportion of the DES was excreted by the mouse in the feces, mostly as free DES and its metabolites with some as conjugates. All DES found in the urine was either conjugated or metabolized to more polar products. The chick excreted about 8% of the administered dose as free DES and 40% as hydrolyzable conjugates of the parent molecule. Other metabolism studies involving beef cattle, sheep, rabbits, cats, rats, and chickens have shown glucuronide and sulfate conjugates to be major metabolites. Our experience with these metabolites indicates they are quite short-lived in an aquatic environment and are easily hydrolyzed to release free DES. Phenothiazine was rapidly eliminated by the mouse (Table 4), only 14% of the dose being excreted as intact phenothiazine. Table 7 gives the results of the metabolism study.
The predominant metabolite was the primary sulfur-oxidation product, phenothiazine sulfoxide. The sulfone also occurs, as well as two major unknown polar metabolites thought to be leucophenothiazone (Rf = 0.05) and thionol (at the origin). Earlier work by use of colorimetric assay techniques found ring-hydroxylation products (leucophenothiazone, phenothiazone, thionol) and their conjugates to be the major metabolites in dairy cows (14) and rabbits, dogs, pigs, sheep, and horses (15).
The total dose of Clopidol ingested by chick during 24 hr is excreted somewhat more slowly than was determined with rats (16). About 37% of radiolabel excreted during 3 days was in the form of free Clopidol, with significant amounts of the a-hydroxylated product (23%) and the car- boxylic acid derivative (16%) (see Table 8). Acid hydrolysis revealed small amounts of conjugates of Clopidol and the a-hydroxyl product present. The metabolism agrees well with previous work (17), which identified the same major degradation products in rabbits.
Sulfamethazine was eliminated by the mouse according to the data shown in Table 4, somewhat slower than observed in sheep (18). The metabolism was primarily to the N4-acetylated product as shown by Table 9. This parallels the results of other studies with cows and sheep (8,18).

Model Aquatic Ecosystem
The fate of the four compounds is expressed as concentrations (ppm) of the parent molecule and detectable metabolites in each component of the small aquatic ecosystem (Table 10). The crucial    These indices were originally devised to reflect degree of bioconcentration and ease of biodegradation for a series of DDT analogs (19). They have since been utilized to assess the comparative environmental fate of many classes of insecticides, herbicides, fungicides, industrial chemicals, and heavy metals. The EM and BI values can be determined for the 33-day model feedlot ecosystem, as well as the 3-day aquatic model. Diethylstilbestrol concentrated to a considerable degree in the alga and snail and to a lesser extent in the fish. BI values ranged from 0.42 in the snail to 1.2 in the fish and 1.4 in the daphnia.
Phenothiazine concentrated less than DES in the snail but more in all the other organisms. However, the BI values were higher for phenothiazine, ranging from 0.6 to 9.4, indicating that it was more easily metabolized by the organisms. By comparison phenothiazine was concentrated in the body (due to high lipophilicity) more than DES, but was also a better substrate for enzymatic oxidation reactions, especially sulfoxidation.
Clopidol concentrated to fairly low levels and was metabolized very slowly as demonstrated by the absence of metabolites in the body extracts. The appearance of trace amounts of the primary degradation products in the water was the only evidence of metabolism. The polar derivatives in the water indicated that Clopidol was probably excreted by most organisms via a conjugation process.
Sulfamethazine failed to concentrate to levels high enough to analyze the organisms for metabolites. However, if all 35S label in the organisms were considered parent molecule, the EM values would all be less than 1.6. Sulfamethazine essentially did not bioconcentrate. Equal concentrations of polar and nonpolar products were found in the water after the 2-day exposure to the organism complex ( Table  10).

Analysis of Aquatic Model Ecosystem Results
A comparison of EM values for the fish Gambusia from the 3-day model aquatic system with octanol/water partition coefficients revealed an excellent correlation (r = 0.987). The log EM values were plotted vs. log octanol/water par-   (9). The correlation coefficient r = 0.9 F value = 11.13 indicated a high significance. Clearly, the bioconcentr chemicals by the fish is closely rel: lipid-partitioning properties of the che The unextractable radioactivi organisms of the model ecosystem is a the extent to which xenobiotic com totally degraded in vivo and the r atoms are reconstituted into tissue c It has been shown that there is a hig negative correlation between magnification of pesticides in model biota and per cent unextractable ra DDE had the lowest value determir and is well known to be virtually nor in living organisms (20).
The values for the unextractable r, for the drugs studied here are record 2. Sulfamethazine, phenothiazine, a: had very high values and DES was int

Model Feedlot Ecosystem
We compared two modes of intro( into'the ecosystem; oral dosing of i olive oil), and 'subcutaneous injecti chicks (in propylene glycol).
At the conclusion of the 33-day exr the mouse ecosystem DES constitutec tractable radioactivity while in t ecosystem DEA accounted for 25% of radioactivity.  Tables 11 and 12. The other ated to the organisms in the ecosystem also contained some Nmicals.
DES. ty of the Phenothiazine was considerably more t measure of biodegradable as only 4% of the extractable 14C ipounds are was in the form of the parent molecule. Table 13  radiolabeled shows the amounts of phenothiazine and its omponents. metabolites (mostly sulfoxide and polar comrh degree of pounds) in the water. None of the organisms conecological tained detectable levels of radioactivity on day ecosystem 33 of the experiment further proving the ease of tdioactivity.
degradation of phenothiazine to polar nonacned, 0.25%, cumulating compounds. idegradable Analysis of the water showed 16% of the extractable radioactivity was sulfamethazine; adioactivity however it did not accumulate to a very large led in Table  degree in any of the organisms. This may be atnd Clopidol tributed to sulfamethazine's moderately high ermediate.
water solubility and very low partition coefficient (see Table 2) which allow rapid elimination and minimal storage in lipoid tissues. The primary ducing DES metabolite in thq -water is Athe N4-acetyl mice (using sulfamethazine, while the organisms each conon of baby tained some sulfamethazine, its acetylated, and methylated derivatives as well as polar products  Table 15 'shows the distribution of metabolites in the water of the Clopidol ecosystems; the a-hydroxylation product o P is the primary metabolite. The organisms concentrated the "4C label in their tissues to the levels shown in Table 16. Bioconcentration to this degree is rather insignificant, inasmuch as autoradiography confirmed all the activity to be in the form of very polar metabolites; the one exception is that. the snail contained an appreciable quantity of the carboxylic acid derivative of Clopidol, which is also quite polar. Apparently the Clopidol molecule is easily excreted by the organisms exposed to it in the model ecosystems; the moderately high water solubility and low partition coefficient are consistent with the observations that Clopidol is not ac'cumulated in the body because it can be readily eliminated.

Reproducibility
The   triplicate using three model ecosystems, each with three chicks fed 10 g of feed contaminated with "4C-Clopidol at 0.0125% for 3 days. The three systems were assayed independently to measure the degree of replicatability of results. As shown in Tables 15 and 16, the replicates were in very good agreement, in fact beyond our expectations, especially since the degree to which the '4C-contaminated chicken feed was spilled directly into the three systems was somewhat random and uncontrollable, despite every precaution to make feeding complete.
The maximum amounts of 14C entering the water phase after feeding Clopidol for 3 days ranged from 0.16 to 0.20 ppm on day 26 (average 0.19 ppm) and declined to 0.13 to 0.18 ppm after 33 days. There was good consistency in the amounts of intact Clopidol and its a-hydroxy and carboxylic acid derivatives found in the water phase (Table 15). The agreement between the three replicated systems seems extraordinary considering the extremely small quantities detected. We conclude that the environmental parameters measured are basically functions of Environmental Health Perspectives   intrinsic physical-chemical properties of the test compound (Fig. 2) and are relatively constant for a given compound.

Analysis of Model Feedlot Ecosystem Results
It can be concluded that chemicals of relatively high water solubilities and low partition coefficients do not accumulate to any great extent in the organisms of the 33-day terrestrial-aquatic model ecosystem. Such compounds tend not to be sequestered in fatty tissues and can be excreted rather easily by animals. Comparisions with data for industrial chemicals (21) and pesticides (19,22) indicate that compounds of lower water solubility and higher partition coefficients accumulate in organisms and biomagnify through food chains more than the chemicals evaluated here.
A second factor to consider is that of susceptibility to enzymatic degradation. The most lipophilic and least water-soluble compound we examined was phenothiazine. Despite its physical properties that could allow the compound to bioconcentrate and despite its having the highest EM in the short-term aquatic model ecosystem, phenothiazine failed to accumulate in the organisms in the 33-day model feedlot ecosystem. This was due to its oxidation to readilyexcretable polar compounds by the mixed function oxidases of the organisms. Clearly two parameters must be considered in any meaningful environmental evaluation of a chemical: (1) the water solubility/partitioning properties of the molecule and (2) functional groups present that will permit attack by degradative enzyme systems.

Conclusions
The model feedlot ecosystem is an adaptation of the terrestrialaquatic model ecosystem (12) and has been designed to screen veterinary drugs and feed additives for persistence in the environment and for food chain biomagnification. The method is quite reproducible with respect to rate of metabolic breakdown and degree of bioaccumulation, as demonstrated by the threereplicate experiment utilizing Clopidol. In combination with the aquatic model ecosystem and metabolism cage studies, the model feedlot ecosystem provides a precise quantitative and qualitative evaluation of the environmental fate of the compounds tested.
Examination of four synthetic veterinary drugs with considerable differences in biological, chemical, and physical properties provided valuable information about the environmental properties of the compounds.
Diethylstilbestrol is more resistant to degradation by the mouse or the chick than the other three drugs, despite the lower doses of DES administered. A significant portion of the DES excreted persisted in the water and organisms as the parent molecule. In view of its potency as a feminizing hormone and its known human carcinogenicity, DES may present a significant degree of environmental hazard. Clopidol is fairly stable environmentally but the parent molecule did not accumulate in any of the organisms in the model feedlot ecosystem. Therefore it is not likely to cause deleterious effects to nontarget organisms. Sulfamethazine has the highest water solubility and lowest octanol/water partition coefficient of the four durgs studied. It is readily excreted rather than stored in the body and is also susceptible to metabolism by the organisms.
Phenothiazine is quite lipophilic but is extremely susceptible to sulfoxidation and ring hydroxylation by both enzymatic and lightcatalyzed reactions. The resultant oxidation products are more water soluble and easily ex-creted. Because of its biodegradability, it poses little threat to the environment except for toxicity to some aquatic organisms ( Table 3).
None of the drugs evaluated was as recalcitrant as an organochlorine pesticide, but there was considerable variation in accumulation and biodegradation of the four compounds. The partition coefficient seems to be a good parameter to predict the fact of an organic molecule in a model ecosystem. The model feedlot ecosystem appears to be a useful tool for screening new veterinary drugs and feed additives for potential persistence and biomagnification.