Comparison of Three Enzyme Immunoassays for Measuring 17B-Estradiol in Flushed Dairy Manure Wastewater

Natural steroidal estrogens are an environmental concern becauselow nanogram per liter concentrations in water can adverselyaffect aquatic vertebrate species by disrupting the normal functionof their endocrine systems. There is a critical need to accuratelymeasure estrogens in dairy wastes, a potential source of estrogenssuch as 17ß-estradiol, to assess the risk of estrogencontamination of agricultural drainage waters resulting fromland application. Commercially available enzyme immunoassay(EIA) kits have been used for measuring 17ß-estradiolin livestock manure, but it is not known if different EIAs providesimilar results. We compared three EIAs by measuring 17ß-estradiolin two samples of flushed dairy manure wastewater (FDMW). Themeasured concentrations of 17ß-estradiol in FDMW differedaccording to the immunoassay used. The differences were attributedto a matrix interference associated with coextracted humic substances.Future research should develop methods that enable routine measurementof 17ß-estradiol in livestock wastes by more conclusiveanalytical techniques such as gas chromatography-massspectrometry.


Introducción

Dairy farms in the United States generate approximately 21.5million Mg of recoverable manure solids each year that mustbe managed in a way that does not adversely impact the environment(USEPA, 2001). Typically, dairy wastes are applied to nearbypasture and croplands as soil amendments because they containvarious plant nutrients, including N, P, and K. However, agriculturaldrainage waters may become contaminated with natural steroidalestrogen hormones such as 17ß-estradiol when livestockwastes are land-applied (Shore et al., 1995; Nichols et al., 1997,1998; Bushee et al., 1998; Finlay-Moore et al., 2000;Dyer et al., 2001).

Estrogen contamination of waterways is a concern because lowconcentrations (10-100 ng L^-1) of these chemicalsin water can adversely affect the reproductive biology of vertebratespecies such as fish, turtles, and frogs by disrupting the normalfunction of their endocrine systems (Panter et al., 1998, 2000;Tyler et al., 1998; Irwin et al., 2001; Oberdorster and Cheek, 2001).For example, 17ß-estradiol concentrations of30 ng L^-1 induced vitellogenin (an egg yolk precursorprotein that is normally produced only by adult females) synthesisand abnormal testicular growth in male fathead minnows (Pimephalespromelas) after 21 d of laboratory exposure (Panter et al., 2000).However, research evaluating the in situ effects of manure-borneestrogens on wildlife is limited. Irwin et al. (2001) reportedthat vitellogenin production by female painted turtles (Chrysemyspicta) in ponds was significantly affected by estrogens in beefcattle runoff compared with turtles in ponds unexposed to beefcattle runoff.

Clearly, it is important to have accurate information aboutthe occurrence of estrogens in manure so that any estrogen contaminationof waterways resulting from dairy waste disposal can be preventedor minimized. Estrogen characterization of dairy wastes is nota trivial task, however, due to the low concentrations thatmust be measured, the difficulties associated with extractingestrogens from manure, the chemical complexity of the resultingextract matrix, and the potential for degradation losses tooccur during sample storage (Raman et al., 2001). A varietyof quantitative EIAs have been used for the determination of17ß-estradiol in manure-impacted surface and groundwater and in livestock wastes (Nichols et al., 1997; Bushee et al., 1998;Peterson et al., 2000; Finlay-Moore et al., 2000).The popularity of EIA for estradiol analysis is attributableto widespread commercial availability, ease of use, pg mL^-1 detection limits, and a lack of alternative quantitation methods.However, a variety of interferences, arising from poor standardization,cross-reactivity, and matrix effects associated with proteinbinding, humic substances, and endogenous enzymes, can adverselyaffect the quality (accuracy, precision, reproducibility) ofthe data produced (Wood, 1991; Maxey et al., 1992; Nunes et al., 1998;Huang and Sedlak, 2001). Thus, depending on samplecomplexity and EIA reagents, antibodies, and protocol, a potentialexists for different EIA systems to yield dissimilar and/orinaccurate results. The objective of this study was to determineif three different commercially available 17ß-estradiolEIAs yielded similar estimates of the endogenous concentrationof 17ß-estradiol in flushed dairy manure wastewater.

Sample Collection

Many dairies use hydraulic flushing for manure management, followedby primary treatment (mechanical screening or sedimentation,or both) to remove coarse solids. The liquid fraction of flusheddairy manure after settleable solids are removed is referredto as FDMW (Wilkie et al., 2004). A 1-L grab sample of FDMWwas collected from the University of Florida Dairy ResearchUnit located at Hague, FL, and immediately (<1 h) transportedto the laboratory for liquid-liquid ether extraction.Two weeks later, a second 1-L sample of FDMW was collected andprocessed in a similar manner. The total solids content of thesesamples was determined by a standard method (American Public Health Association, 1998).The first and second FDMW samplescontained an average of 0.57 and 0.62% total solids, respectively.

Extraction

For each wastewater sample, four aliquots (20 mL) of FDMW werepoured into separate 50-mL glass centrifuge tubes. Twenty millilitersof pesticide-grade ethyl ether (Fisher Scientific, Hampton,NH) was added to each tube for extraction of 17ß-estradiol.Liquid-liquid extraction with ether was used for samplepreparation because it is a traditional solvent of choice forsteroid extraction from biological samples; ether extractionis recommended for sample purification by the EIA manufacturersused in this study, and it has been used previously for extractionand purification of dairy waste samples for EIA analysis (Raman et al., 2001).

The tubes were shaken horizontally for 2 h followed by centrifugationat 500 x g for 5 min to facilitate layer separation. Three 4-mLaliquots (one for each assay) of the ether extract were subsampledfrom each tube and placed into separate 5-mL evaporation flasks.The ether was evaporated to dryness at 40°C under N₂. Thedried sample was immediately reconstituted in 1 mL of bulk assaybuffer that was purchased from each immunoassay manufacturer.The reconstituted samples were individually sonicated for approximately1 min to enhance solubilization in the assay buffer. The sampleswere poured into 1.5-mL micro-centrifuge tubes, capped tightly,and stored overnight (-20°C) before immunoassay analysis.

Immunoassay Description

Enzyme immunoassay kits for the quantitative determination of17ß-estradiol were purchased from Assay Designs (Catalogno. 900-008; Ann Arbor, MI), Diagnostics Systems Laboratories(Catalog no. DSL-10-4300; Webster, TX), and Immuno-BiologicalLaboratories (Catalog no. RE 52041; Minneapolis, MN). The immunoassaykits were designated A1, A2, and A3, respectively. The A1 immunoassay(Catalog no. 900-008) was selected because it has been usedpreviously for the quantification of 17ß-estradiolin dairy wastes (Raman et al., 2001). The A2 and A3 immunoassayswere selected based on their use of rabbit polyclonal antibodies(RPA) and the competitive assay principle, and a low cross-reactivitywith other steroids (Table 1).

Each of the EIAs used in this study were based on the competitivebinding principle, whereby 17ß-estradiol and a fixedamount of enzyme-labeled estradiol compete for RPA binding sites.However, the A2 and A3 assays use RPAs that are directly coatedonto the microplate wells, whereas the A1 microplate wells arecoated with goat anti-rabbit IgG to capture the 17ß-estradiol-RPAcomplex. The alkaline phosphatase, streptavidin-horseradishperoxidase, and horseradish peroxidase enzyme tracers used byA1, A2, and A3, respectively, represent commonly used enzymereagents for estrogen immunoassay (Table 1) (Meyer et al., 1990;DeBoever et al., 1995; Mares et al., 1995; Vos, 1996). As shownin Table 1, each immunoassay has a low (<5%) cross-reactivitywith other estrogen steroids.

Immunoassay Analysis

Each assay was performed according to the manufacturer's instructions.All standards and samples were assayed in duplicate and an averagevalue was used to generate standard curves and interpolate unknownsample concentrations. Microplate washing was performed withan EL_x50/8 strip washer (Bio-Tek Instruments, Winooski, VT)using the wash buffer reagents provided by each company. Theabsorbance values of each well were measured using an FL 600microplate reader (Bio-Tek Instruments). A four-parameter logisticequation was used for all calibration curves (Rodbard and Lewald, 1974).

Immunoassay performance characteristics including sensitivity,standardization, precision, and recovery of diluted and spikedsamples were evaluated on both days of wastewater analysis.Sensitivity is defined as the lowest measurable concentrationof 17ß-estradiol that can be distinguished from therespective 0 pg mL^-1 calibrator (95% confidence interval)associated with each EIA (Vadlamudi et al., 1991). Sensitivitywas calculated for each EIA by interpolation of the mean ofeight replicate samples of the respective 0 pg mL^-1 calibratorminus two standard deviations.

Standardization accuracy refers to the ability of each EIA toyield a correct measurement of 17ß-estradiol for aknown standard concentration. Standardization accuracy was evaluatedat three concentrations (1500, 750, and 375 pg mL^-1) bydiluting a 300000 pg 17ß-estradiol mL^-1 buffersolution (Assay Designs) with the respective 0 pg mL^-1 calibrator of each EIA. Three concentrations were measured toensure accurate recovery at different interpolation points alongthe calibration curve. A recovery percentage for each standardconcentration was calculated by dividing the measured sampleconcentration by the known sample concentration and multiplyingthe result by 100. The three resulting values were averagedto express EIA standardization accuracy.

Intra-assay precision refers to the within-run reproducibilityof the 17ß-estradiol signal that is produced for aparticular sample in an EIA. We evaluated precision by calculatingthe percent coefficient of variation observed between duplicatemeasurements corresponding to the four neat wastewater samples.The four resulting % CV values were averaged to express precision.

Recovery of diluted and spiked samples is a gauge of the linearrelationship between 17ß-estradiol measured in dilutedor spiked samples relative to the neat samples. Dilution recoverywas measured by diluting each of the four neat wastewater sampleswith an equal volume of the respective 0 pg mL^-1 calibratorof each assay. Spiked recovery was measured by spiking the neatwastewater samples with an equal volume of the second greatestrespective 17ß-estradiol calibrator from each EIA(i.e., A1, 7500 pg mL^-1; A2, 2000 pg mL^-1; A3, 1000pg mL^-1). The second greatest calibrators were used forspiking to ensure that the resulting spiked sample concentrationswould be interpolated from the mid-portion of the calibrationcurve of each assay. Dilution and spiked recovery was expressedas a percentage by dividing the measured concentration of thediluted or spiked sample by the theoretically expected concentrationof the diluted or spiked sample, and the result was multipliedby 100.

Data Analysis

The experimental design was a two-way factorial (three immunoassaymethods x two FDMW samples) with four replications. Experimentaldata were analyzed using the General Linear Model program ofSAS with a separation of sample means by Duncan's new multiplerange test (SAS Institute, 2000).

A summary of the immunoassay performance characteristics fromeach FDMW analysis is shown in Table 2. The measured sensitivitydata corresponding to the first wastewater sample were similarto or better than the manufacturer's data for each EIA. However,the sensitivity data corresponding to the second analysis weresomewhat larger for each assay. The average EIA sensitivityfor both analyses was 62, 14, and 26 pg mL^-1 for the A1,A2, and A3 assays, respectively. The sensitivity data demonstratethe exceptionally low 17ß-estradiol concentrationsthat can be measured using EIA.

 

* Flushed dairy manure wastewater.
 

Recovery data shown in Table 2 demonstrate that the A1 and A2assays were relatively well standardized for both analyses.The calibration of the A3 assay appeared to be somewhat lessaccurate for each individual analysis since it overestimatedby 36% and underestimated by 25%, respectively, the standardconcentrations for the first and second analysis. Overall, however,the average recovery percentage for both analyses was 105, 98,and 106% for the A1, A2, and A3 immunoassays, respectively.Therefore, it seems that each of the EIAs was reasonably wellstandardized.

Each assay also showed a high degree of intra-assay precisionbetween duplicate samples. The % CV for both analyses averaged8, 7, and 9%, respectively, for the A1, A2, and A3 assays. Thelow % CV values indicate that the chemical reactions involvedin generating the 17ß-estradiol signals for each EIAwas highly reproducible within the analytical run.

The recovery of diluted samples ranged from 66 to 128%, dependingon the EIA and day of analysis (Table 2). The recovery of dilutedsamples for both analyses averaged 79, 119, and 124%, respectively,for the A1, A2, and A3 assays. In contrast to diluted samples,recovery improved markedly when the neat samples were spikedwith 17ß-estradiol. The recovery of the spiked samplesaveraged 92, 95, and 91%, respectively, for the A1, A2, andA3 immunoassays. Overall, the recovery of diluted and spikedsamples demonstrates a reasonably linear recovery of 17ß-estradiolat the different interpolation points evaluated from the standardcurve.

Although some minor differences were encountered between assaysregarding standardization accuracy, intra-assay precision, andrecovery of diluted and spiked samples, the measured concentrationof 17ß-estradiol in both sets of FDMW samples differedaccording to the EIA used (Fig. 1). The A1 assay consistentlymeasured the greatest 17ß-estradiol concentrationsand the A2 assay measured the lowest. The average concentrationof 17ß-estradiol in the first wastewater sample measuredwith the A1, A2, and A3 immunoassays was 526, 161, and 332 ngL^-1, respectively, and 1310, 181, and 356 ng L^-1,respectively, in the second wastewater sample.

 

Because no differences were observed between EIAs when a puresolution of 17ß-estradiol was analyzed (standardizationaccuracy) (Table 2), the apparent difference between assayssuggests that an interference affected 17ß-estradiolquantitation in FDMW samples in one or more of the EIAs. A knownsource of interference with the EIAs is the presence of othersteroidal estrogens that are listed as cross-reactants in Table 1.It was noticed that the apparent concentrations of 17ß-estradiolin the wastewater followed in the same qualitative order (A1> A3 > A2) as the reported estrone cross-reactivity of the differentassays. Consequently, estrone was a suspected source of biasbetween assays. Hence, we measured estrone with an estrone EIA(Catalog no. DB 520 51; Immuno-Biological Laboratories). Similarestrone EIAs were not available from the other companies forcomparison.

Estrone concentrations were 562 and 781 ng L^-1 in thefirst and second wastewater samples, respectively. Based onthe cross-reactivity data shown in Table 1, estrone in the firstwastewater sample would have contributed approximately 26, 8,and 12 ng L^-1 of 17ß-estradiol signal to theA1, A2, and A3 assays, respectively. Likewise, estrone in thesecond set of wastewater samples would have contributed approximately36, 11, and 16 ng L^-1 to the 17ß-estradiol signal.If the estrone cross-reactivity data provided by the manufacturersare correct and the EIA measured estrone concentrations areaccurate, the large differences observed between assays do notappear to be caused by estrone cross-reactivity.

Other types of matrix interferences that are known to affectthe quality of EIA data are often associated with coextractedhumic substances. For example, Huang and Sedlak (2001) demonstratedthat certain types of humic substances extracted from surfacewater could give positive signals during 17ß-estradiolEIA. Presumably, the humic substances cross-react with the 17ß-estradiolantibody or adsorb to the estradiol enzyme conjugate in a mannerthat inhibits the competitive antibody binding and thus givea false-positive EIA signal. On the other hand, humic substancesmay cause false-negative EIA signals if they inhibit the competitivebinding of 17ß-estradiol to the antibody binding sites.

Ideally, the lack of agreement between immunoassays could bereconciled with a more conclusive measurement technique likegas chromatography-mass spectrometry (GC-MS) todetermine which assay provided the most accurate measurementof 17ß-estradiol in FDMW. Unfortunately, GC-MSquantification was not possible with these wastewater samplesdue to the extraordinary sample complexity associated with theether extracts and because the ng L^-1 sample concentrationsare several orders of magnitude lower than the detection limits(approximately 10 µg L^-1) associated with the onlypublished method for the GC-MS analysis of dairy wastes(Raman et al., 2001). A similar problem was reported by Raman et al. (2001),who tried to compare the endogenous concentrationof 17ß-estradiol in press-cake dairy solids measuredby the A1 EIA and GC-MS. Endogenous 17ß-estradiolcould not be measured by GC-MS due to the relatively poordetection limits. However, when 17ß-estradiol wasspiked into the press-cake samples, the A1 EIA and GC-MSmethods agreed well. Nevertheless, the spiked EIA and GC-MScomparison does not yield much information regarding bias ofthe A1 assay because an interference, if present, would havebeen greatly masked by dilution of the spiked samples.

Based on the large differences observed between EIAs in thisstudy, caution should be observed when interpreting the biologicalsignificance or ecological risk of 17ß-estradiol concentrationsin livestock wastes when measured by EIA. Immunoassays are potentiallyvaluable tools for the rapid screening of environmental samples.However, a better understanding of the artifacts and interferencesassociated with highly complex and variable livestock wastematrices is clearly needed. To better understand EIA limitations,it is critical that sensitive and reliable GC-MS or liquidchromatography-mass spectrometry (LC-MS)-based methodsbe developed as definitive reference methods.

Ether extraction and quantitation by EIA is a convenient methodfor measuring estrogens in FDMW. Although no differences wereobserved between EIAs when a pure solution of 17ß-estradiolwas analyzed, three EIAs gave different 17ß-estradiolresults for the same wastewater samples. The differences aremost likely caused by one or more matrix interferences associatedwith coextracted humic substances in the sample. The poor qualityof the ether extracts and low concentrations of 17ß-estradiolin the wastewater prevented GC-MS quantitation and thereforeit is not known which of the three EIAs yielded the most accuratemeasurement of 17ß-estradiol. Future research needsto develop better extraction and/or purification techniquesso that 17ß-estradiol and other estrogens can be measuredin FDMW by more conclusive techniques like GC-MS or LC-MSand to ensure that immunoassay results can be validated.

This research was supported by the Florida Agricultural ExperimentStation and a grant from the School of Natural Resources andEnvironment Mini-Grants Program, University of Florida, andapproved for publication as journal series number R-09877.