Environ. Sci. Technol. 2007, 41, 738-744
Analysis of Lagoon Samples from Different Concentrated Animal Feeding Operations for Estrogens and Estrogen Conjugates S T E P H E N R . H U T C H I N S , * ,† MARK V. WHITE,† FELISA M. HUDSON,‡ AND DENNIS D. FINE‡ Ground Water and Ecosystems Restoration Division, National Risk Management Research Laboratory, U.S. Environmental Protection Agency, P.O. Box 1198, Ada, Oklahoma 74821-1198, and Shaw Environmental and Infrastructure, P.O. Box 1198, Ada, Oklahoma 74821-1198
Although Concentrated Animal Feeding Operations (CAFOs) have been identified as potentially important sources for the release of estrogens into the environment, information is lacking on the concentrations of estrogens in whole lagoon effluents (including suspended solids) which are used for land application. Lagoons associated with swine, poultry, and cattle operations were sampled at three locations each for direct analysis for estrogens by GC/ MS/MS and estrogen conjugates by LC/MS/MS. Estrogen conjugates were also analyzed indirectly by first subjecting the same samples to enzyme hydrolysis. Solids from centrifuged samples were extracted for free estrogens to estimate total estrogen load. Total free estrogen levels (estrone, 17R-estradiol, 17β-estradiol, estriol) were generally higher in swine primary (1000-21000 ng/L), followed by poultry primary (1800-4000 ng/L), dairy secondary (370550 ng/L), and beef secondary (22-24 ng/L) whole lagoon samples. Swine and poultry lagoons contained levels of 17Restradiol comparable to those of 17β-estradiol. Confirmed estrogen conjugates included estrone-3-sulfate (2-91 ng/ L), 17β-estradiol-3-sulfate (8-44 ng/L), 17R-estradiol-3sulfate (141-182 ng/L), and 17β-estradiol-17-sulfate (7284 ng/L) in some lagoons. Enzymatic hydrolysis indicated the presence of additional unidentified estrogen conjugates not detected by the LC/MS/MS method. In most cases estrogen conjugates accounted for at least a third of the total estrogen equivalents. Collectively, these methods can be used to better determine estrogen loads from CAFO operations, and this research shows that estrogen conjugates contribute significantly to the overall estrogen load, even in different types of CAFO lagoons.
Introduction Endocrine-disrupting chemicals (EDCs) can exert environmental effects at low ng/L levels and are becoming of increasing concern worldwide (1-4). Some of the most potent EDCs include the natural estrogens 17β-estradiol (E2β) as * Corresponding author phone: (580)436-8563; fax: (580)436-8703; e-mail:
[email protected]. † U.S. Environmental Protection Agency. ‡ Shaw Environmental and Infrastructure. 738
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 3, 2007
well as the less active estrone (E1), 17R-estradiol (E2R), and estriol (E3) (5, 6). Although these compounds can be degraded biologically, they have been detected in sewage treatment effluents and receiving surface waters at ng/L levels (7-9). In addition to sewage treatment plants, concentrated animal feeding operations (CAFOs) have been identified as potentially important sources for the release of natural estrogens into the environment, but information is lacking on the concentrations of estrogens in lagoons which are used for land application for different CAFO operations. Recent reviews (10, 11) of the occurrence and environmental fate of manure-borne estrogens note that although estrogens are often degraded or transformed quite readily under aerobic conditions in the laboratory (see also refs 12-14), other laboratory studies indicate persistence, and field studies demonstrate estrogen transport to receiving waters following land application. Both reviews point out the need for additional research on the environmental fate of manureborne estrogens, and Hanselman et al. (10) also call for additional research on the types and amounts of estrogens in animal manure as well as for standardized methods to detect low levels of estrogenic compounds in these complex matrices. Given that information is lacking on distribution of estrogens in CAFO waste, the case is even more so with estrogen conjugates (10). Estrogens are excreted as either free estrogens or as sulfate or glucuronide conjugates, with the conjugated forms being biologically inactive. However, estrogen conjugates can be readily deconjugated to produce the active free estrogens, and the free estrogens can undergo reversible transformations as well (Figure 1). Numerous potential estrogen conjugates are possible because a sulfate or a glucuronide group can be attached to the phenoxy or hydroxyl group of each epimer. Estrogen conjugates are readily hydrolyzed in sewage treatment plants, although the sulfate forms are more recalcitrant than the glucuronide forms, and residual low levels of sulfate estrogen conjugates can be found in sewage treatment plant effluents and receiving waters (15-19). In contrast, there is virtually no information on the distribution of estrogen conjugates in CAFO waste lagoons and their environmental fate following land application. Unlike sewage treatment plants, CAFO lagoons typically function as holding reservoirs or anaerobic reactors, and waste effluent generally receives no additional treatment prior to land application. Estrogen conjugates are also expected to be more polar and therefore more mobile in the soil (10). The major objectives of our work therefore were to determine the concentrations of free estrogens in whole effluents (including suspended solids) used for land application from different representative CAFO waste lagoons, to determine estrogen conjugate levels by direct LC/ MS/MS methods as well as indirect methods using enzymatic hydrolysis, and to provide an overall assessment of the contribution of estrogen conjugates to the total estrogen loads.
Materials and Methods Field Sites and Sampling. Eight separate CAFO lagoons were sampled from selected commercial swine, poultry, and cattle operations in south central United States from late Feb to early Apr 2006. All of these lagoons are used directly for land application and include swine sow (SS), swine finisher (SF), and swine nursery (SN) primary lagoons and dairy (DA) and beef feedlot (BF) secondary lagoons. Two separate poultry (laying chicken) operations with primary lagoons were also sampled (PO1 and PO2). The second poultry operation had 10.1021/es062234+ CCC: $37.00
2007 American Chemical Society Published on Web 01/04/2007
FIGURE 1. Structures of estrogens showing pathways and potential locations for formation of estrogen conjugates. another set of barns that was serviced by a series of lagoons, and the final tertiary lagoon (PO2T) was also sampled. For each lagoon, the first sample was taken adjacent to the lagoon discharge point using a stainless steel submersible pump (Grundfos Pumps Corporation, Clovis, CA) equipped with a float and suspended to a depth representative of effluent intake for land application. Two additional samples were taken from locations 20-100 m from the discharge point and equidistant to the shoreline. Lagoon water was pumped through Teflon tubing at 600 mL/min through a sample filter bypass system into a flow-through cell for 15 min prior to sampling. Samples for estrogen and estrogen conjugate analyses were collected unfiltered into 2-L glass media bottles which were immediately sealed without headspace and placed on ice. These samples were processed within 12 h of sample collection. Additional samples were obtained and analyzed for several other parameters (see the Supporting Information). Chemicals. Samples were analyzed for E1, E2R, E2β, E3, and ethinyl estradiol (EE2) as well as 13-16 separate estrogen conjugates. EE2 is not expected to be present in CAFO lagoons, but it is part of the analytical suite used in our method and helps to identify sample and analytical problems. A description of the sources of analytes and associated chemicals used in the methods is provided in the Supporting Information. Direct Estrogen Analysis in Samples and Sediments. Each lagoon sample was mixed and transferred to a 500-mL centrifuge bottle and centrifuged at 13700g for 1 h at 20 °C. Liquids were decanted, and the centrifugate solids were frozen at -20 °C. Liquid samples were filtered through a 90-mm glass fiber filter (Millipore APFC, particle retention > 1.2 µm, Cat. #APFC09050, Millipore Corporation, Bedford, MA) and split into two aliquots. The first aliquot was preserved with formaldehyde (1% final w/v) and stored at 4 °C overnight. The second aliquot was treated by enzyme hydrolysis using β-glucuronidase/arylsulfatase from Helix pomatia as de-
scribed by Lee et al. (20) prior to preservation with formaldehyde. Free estrogens (E1, E2β, EE2, and E3) were determined the next day by SPE, derivatization, and GC/ MS/MS analysis as previously reported (21), with the exception that E2R was also included in the suite of analytes and that carbon dioxide was used instead of methane (22) as the chemical ionization reagent gas. Method detection limits (MDLs) for 25-mL aliquots of the lagoon sample were 12, 4, 20, and 8 ng/L for E1, E2R, E2β, and E3, respectively. EE2 was not detected in any of the lagoons (MDL ) 20 ng/L). The centrifugate solids in each centrifuge bottle were freeze-dried for 4 d and extracted by adding 10 mL of 50/50 v/v methanol/ acetone and sonicating for 10 min. The extract was transferred to a 15-mL centrifuge tube, and the extraction was repeated two additional times. The solvent was evaporated under N2 at 40 °C, and 5.0 mL of 50/50 v/v methanol/acetone was added to each centrifuge tube to dissolve the residue. A 2-mL aliquot was diluted with water to 25 mL, and this sample was then extracted and derivatized for free estrogen analysis as described above. Direct Estrogen Conjugate Analysis. Each lagoon sample was mixed and transferred to a 500-mL centrifuge bottle and centrifuged at 10 800g for 20 min at 20 °C. Liquids were decanted and filtered as described previously, but centrifugate solids were not analyzed. Estrogen conjugates were analyzed using modifications of the SPE and LC/MS/MS methods reported by Gentili et al. (17). For SPE, 2-mL reversible cartridges containing 500 mg of Carbopack X (Supelco, Bellefonte, PA) were used. Carbopak X is a graphitized carbon black with a particle size of 60/80 mesh and a surface area of 240 m2/g. The cartridges were washed with 10 mL of 80/20 v/v methylene chloride/methanol containing 5 mM tetramethylammonium chloride, followed by 5 mL of methanol, 20 mL of water acidified with HCl to pH 2, and 5 mL of water. Sample volumes of 100 mL were transferred to serum bottles and spiked with 10 µL of 1 ng/µL 17β-estradiol-2,4,16,16d4-3-sulfate (E2β-d4-3S) as an internal standard. After 100 VOL. 41, NO. 3, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
739
FIGURE 2. Aqueous estrogen concentrations in lagoon samples associated with various types of CAFOs as determined by GC/MS/ MS. Abbreviations are as described in the text. Data shown are mean with standard deviation from three locations for each lagoon. mL of sample passed through the cartridge, the cartridge was washed with 50 mL of water, 10 mL of methanol acidified with 50 mM formic acid, 5 mL of methanol, and 10 mL of 80/20 v/v methylene chloride/methanol. Next the cartridges were reversed, and conjugated estrogens were eluted using 10 mL of 80/20 v/v methylene chloride/methanol containing 5 mM tetramethylammonium chloride. The flow of solvent through the cartridge was ∼1 drop/s. The extract was collected in 15-mL centrifuge tubes, and the solvent was evaporated to dryness under N2. The extract residue was dissolved in 1.00 mL of 10/90 v/v acetonitrile/water. Each extract was filtered into a 2-mL autosampler vial using a 0.2-µm disposable syringe filter and acidified to pH 3 by adding 4 µL of concentrated acetic acid. For the LC/MS/MS analysis, 10 µL of extract was injected by an Agilent 1100 autosampler (Agilent, Wilmington, DE). A 2.1 × 100 mm LC column and 2.1 × 10 mm guard column, both containing 3.5 µm Waters Xbridge C18 packing material, were used for the LC separation (Waters, Millford, MA). The mobile phase consisted of (A) 25 mM methylamine in water and (B) acetonitrile. The initial solvent composition was 97.5% A followed by a gradient starting upon sample injection to 70% A over 15 min and then to 20% A over 5 min. A composition of 20% A was maintained for 8 min after which the composition was changed back to 97.5% A. The composition was held at 97.5% A for 30 min before the next sample was injected. The column flow was 200 µL/min during the entire LC run. A stream of 150 µL/min of acetonitrile was added to the column flow before LC eluent entered the electrospray source. A divert valve was used as a solvent dump during the first 5.5 min of the run and the last 32 min of the run. During the divert valve dump, acetonitrile continued to flow into the ion source. A Finnigan TSQ Quantum Ultra mass spectrometer with an Ultramax source was used for the tandem mass spectrometry (Finnegan, San Jose, CA). Negative ion MS/MS was used with a spray voltage of 2000 V and an ion transfer tube temperature of 350 °C. Multiple reaction monitoring (MRM) was done using parent and product ions and collision energies optimized for each estrogen conjugate (see the Supporting Information).
Results and Discussion Direct Estrogen Analysis. The distribution of aqueous free estrogens, although varying in scale, showed similar trends in the swine and poultry primary lagoons, with E1 > E3 > E2R, E2β (Figure 2). Surprisingly, E2R levels were greater than E2β levels in these lagoons and substantially so for the swine sow and poultry lagoons. This is contrary to what would be expected, since swine and poultry generally excrete E2β rather than E2R (10), but similarities in levels of E2R and E2β 740
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 3, 2007
have been shown for other swine lagoons (23). The high levels of E2R relative to E2β in these lagoons probably occur due to transformations within the lagoons and therefore may not reflect the original ratios of estrogens excreted by the animals. E2β is readily oxidized to E1 by fecal bacteria, and this transformation has been observed in swine sow, nursery, and finisher lagoon samples before (21). A recent study has shown that this transformation is reversible under anaerobic conditions, and, in fact, racemization can occur and produce E2R from E1 which was originally produced from E2β (24). In some respects this could be considered as a deactivation step, in that E2R is less biologically active than E2β (5, 6). However, it is not known to what extent E2R remains stable in anaerobic CAFO lagoons, and the potential therefore exists for reverse transformation back to E1 and ultimately back to E2β. Reversible transformations among these three estrogens may also help to explain why the estrogen distribution trends are so similar between the different swine and poultry lagoons and indicates that it may not be possible to differentiate between different livestock sources of estradiol in environmental samples based on its stereochemistry alone, as has been suggested elsewhere (10). More importantly, these results suggest caution should be used when assessing the potential for environmental impact from specific estrogens in CAFO lagoons, since transformation to more or less active forms may occur. Aqueous estrogen concentrations in the swine lagoons were generally similar to what has been reported previously for other facilities, although there are differences. For example, mean concentrations of E1, E2β, and E3 in the swine sow lagoon were 9940, 194, and 6290 ng/L, respectively, whereas the corresponding mean concentrations were 17400, 2460, and 7830 ng/L in a separate large swine sow facility with two anaerobic digesters and two primary lagoons (25). Hence, E2β levels were about an order of magnitude less in this study. Some of the discrepancy may be due to a different composition of swine in the other facility (19920 gestating sows, 4980 farrowing sows, and 100 boars), since boars excrete much more estrogens in urine and feces than cycling sows (27, 28), but there are other possible explanations as well. Raman et al. (23) also found higher levels of E2β (3900 ng/L) in farrowing sow facilities in the southeastern United States, but E1 levels were somewhat less (5900 ng/L) than observed in this study. Similar results were observed for swine finisher lagoons in that study (E1 ) 1100-5900 ng/L, E2β ) 18003300 ng/L) compared to this study (E1 ) 1550 ng/L, E2β ) 120 ng/L). Estrogen concentrations in the swine nursery lagoon were comparable to what was observed previously for a separate operation with four lagoons, in which mean concentrations of E1, E2β, and E3 were 530, 47, and 198 ng/ L, respectively (25). Estrogen levels in the two poultry primary lagoons were similar to those in the swine nursery lagoon and decreased 100-fold in the tertiary poultry lagoon (Figure 2). Because PO2T serviced a different set of chicken houses than PO2, the data cannot be used to directly assess the efficacy of using sequencing lagoons for estrogen removal. However, each set of chicken houses was constructed and operated identically, so it is reasonable to assume that the estrogen load from land application of lagoon effluent would be much less if wastewater were obtained from tertiary rather than primary poultry lagoons. Even though the swine operations had the greatest amounts of suspended solids, total estrogen concentrations were generally not much higher (
