J. Dairy Sci. 87:2864-2867
© American Dairy Science Association, 2004.
Technical Note: Protein Conjugate-Based Immunoassay of Whole Milk Progesterone
J. S. Mitchell1,2,
Y. Wu1,
C. J. Cook1 and
L. Main2
1 Bioengineering Sector, HortResearch, Hamilton, New Zealand
2 Department of Chemistry, University of Waikato, Hamilton, New Zealand
Corresponding author: Y. Wu; e-mail:ywu{at}hortresearch.co.nz.
 |
ABSTRACT
|
|---|
The objectives were to develop a novel protein conjugate-based ELISA test for whole milk progesterone with a dynamic range capable of fully profiling estrous cycles in the dairy cow and to study effects of whole milk medium on antibody binding to progesterone-protein conjugates. A series of progesterone-4-ovalbumin conjugates with different length intermediate linkers were applied as coating antigens in an ELISA format to determine antibody-binding performance in whole milk. Use of an 18-atom linker gave higher binding than use of a 4- or 11-atom linker at certain conjugate concentrations, but no further increase was observed with increasing linker length. An ELISA constructed with the 18-atom linker conjugate gave a detection limit of 0.089 ng/mL progesterone and correlated well to an established radioimmunoassay procedure (r = 0.94). The assay has the distinct advantages of a wide linear range (0.1 to 100 ng/mL), allowing full profiling of bovine estrous cycles, use of whole milk directly without extraction or prior dilution, and employing more easily purified protein conjugates as coating antigens compared with commercial progesterone-enzyme conjugate for milk ELISA assays.
Key Words: progesterone • enzyme immunoassay • protein conjugate
Abbreviation key: PBS/T = phosphate-buffered saline with Tween-20, RIA = radioimmunoassay
|
INTRODUCTION
|
|---|
Progesterone in bovine whole milk is a well-established indicator of the bovine estrous cycle (Foulkes et al., 1982). Enzyme-linked immunosorbent assay techniques have been extensively used to measure progesterone for timing of AI and early pregnancy diagnosis. Many fully quantitative ELISA techniques currently available require milk samples to be pretreated or pre-diluted. They also often have narrow dynamic ranges (e.g., 1 to 30 ng/mL; Waldmann, 1993) that do not always allow for complete profiling of estrous cycles of individual cows where amplitudes of progesterone fluctuations may vary greatly from cow to cow. Furthermore, there is a dearth of studies on the effects of fundamental changes to coating antigens on antibody binding and assay performance in the whole milk medium; indeed, there have been no studies using progesterone-protein conjugates as coating antigens in milk progesterone ELISA.
Existing commercial ELISA, based on a competitive assay format, rely on the use of progesterone-enzyme conjugates (Waldmann, 1993). The use of such conjugates has certain significant disadvantages. These enzyme conjugates lack an efficient and convenient purification method. Time-consuming affinity chromatography is the only option to remove unbound enzyme that would otherwise contribute significant background signal. Enzymes are usually conjugated to analytes via the 
-amino groups of lysine residues. Most enzymes, however, have multiple residues available for conjugation. As the electrostatic and steric environment of the conjugated antigen greatly influences antigen-antibody interactions, changes to this environment arising from different attachment positions can greatly impair antibody-binding affinity to the ligand (Paek et al., 1993). In the production of the conjugate, the number of analyte molecules attached per enzyme molecule cannot be easily controlled; so, irregular binding, such as binding to multiple recognition sites, is possible, which changes the conditions for competition with the analyte in the sample (Paek et al., 1993). An alternative method that avoids these problems is to immobilize a protein conjugate of the target analyte and to create competition between this conjugate and free-sample analyte for binding to limited antibody (inhibition assay). Antibody binding detection is then possible via binding of an enzyme-labeled secondary antibody. Use of such a secondary antibody also increases specificity of label binding (Self and Cook, 1996). To address these challenges, we developed a simple ELISA test for whole milk progesterone, employing novel progesterone-4-protein conjugates with different length intermediate linkers as coating antigens.
Progesterone was conjugated via its A-ring 4-position through heterobifunctional linkers of 4-, 11-, 118-, and 25-atoms in length to ovalbumin using caproate unit increments according to methods previously described (Wu et al., 2002) with an average number of progester-one units per protein molecule of 4 for 4-, 11-, and 18-atom linkers and 2.3 for the 25-atom linker. The lower number for the 25-atom conjugate occurs because the progesterone-linker derivative is less soluble in the polar solvents needed for conjugation. These conjugates were immobilized on 96-well ELISA plates at concentrations ranging between 0 and 5 µg/mL in 0.05 M bicarbonate buffer (pH = 9.6; 100 µL per well, in quadruplicate), and the plates were then blocked with 1% (vol/vol) ovalbumin (200 µL per well) in PBS with Tween 20 (PBS/T). The PBS/T (75 µL per well) was added to the blocked-and-washed plate followed by whole milk of zero progesterone as determined by radioimmunoassay (RIA) (25 µL per well) as blank milk. Monoclonal anti-progesterone antibody (Sigma P1922; Sigma Chemical Co., St. Louis, MO) was then added (5 µg/mL in PBS/T, corresponding to approximately 1/1600 dilution; 100 µL per well), and plates were incubated for 1 h at room temperature. Secondary antibody horseradish peroxidase conjugate (Sigma A5795; 1/12,500 dilution in PBS/T; 200 µL per well) was then added after washing the plate and was incubated as for the monoclonal antibody before washing and color development with 3, 3', 5, 5'-tetramethylbenzidine color solution (100 µL per well) followed by color stopping (2 M sulfuric acid; 50 µL per well) and reading at 450 nm. All milk samples were preserved with potassium dichromate tablets and frozen upon collection. All samples were thawed at time of use and not refrozen.
Binding curves showed that overall antibody binding increased with increasing linker length in the milk medium up to 18 atoms but no further for 25-atom length. At 1 µg/mL coating concentration, increasing the linker from 4 to 11 atoms enhanced binding by 32% and incrementing from 11 to 18 atoms produced a further increase of 18% at 1 µg/mL. However, extending the linker to 25 atoms decreased binding by 96% compared with the 18-atom linker at 1 µg/mL, which is partly due to the decreased number of progesterone molecules attached per protein molecule caused by low linker solubility. To prevent this decrease in progesterone attachment, more polar linkers would need to be used, but these will not be so lipophilic and, therefore, may not extend into the milk medium so well. Use of a linker up to 18 atoms is an advantage for increasing antibody-binding signal.
The improved signal was expressed as a decreased detection limit when the conjugates were applied as coating antigens in a progesterone assay from spiked whole milk samples as described previously but with the conjugates immobilized at 1 µg/mL and samples and spiked standards taking the place of blank milk. The standards were run in triplicate or quadruplicate and prepared by spiking the blank milk. The detection limit was taken as the concentration reading corresponding to the absorbance reading that is 2 SD of the absorbance reading of the blank below the mean absorbance of blank milk. The conjugate with an 18-atom linker gave the lowest detection limit at 0.089 ± 0.016 ng/mL compared with 0.87 ± 0.08 ng/mL for the conjugate having an 11-atom linker and 2.7 ± 0.9 and 30 ± 10 ng/mL for the conjugates with a 4-atom linker and 25-atom linker, respectively. This demonstrates that increasing linker length from 11 to 18 atoms improves the detection limit in the whole milk medium. These results are particularly interesting given that our previous studies in aqueous medium have shown no difference in binding or detection limits with increasing linker length in ELISA (Wu et al., 2002). Speculatively, it is possible that the lipophilic linker chains of the conjugates are more soluble in the milk medium than in the aqueous medium and so project the progesterone units more effectively, allowing for higher antibody binding and improved detection limit. The long 25-atom linker is possibly entwining around the progesterone and blocking antibody binding.
Two-tailed unpaired t-tests were applied to both the antibody binding data and the detection limit data, and all antibody binding data were statistically different from the other values compared (P < 0.001). For detection limits, the means for 11- and 18-atom linkers were highly significantly different (P < 0.001). The higher detection limit errors in the 4- and 25-atom linker conjugates were caused by low curve slope and low antibody binding, respectively, and made some detection-limit comparisons with them statistically difficult, though by one-tailed t-test, the 18-atom linker detection limit is lower than both the 4- and 25-atom linkers (one-tailed t-test P values of 0.05). In the case of detection limit errors, a 4-parameter logistic regression was fitted using Sigma Plot Version 8.0, and the standard errors were calculated by inputting the fraction-bound standard error at the nearest point to the detection limit in the working region of the assay curve into the rearranged 4-parameter logistic equation to translate standard error in the fraction bound to standard errors in the progesterone concentration. Assay curve data were fitted to a 4-parameter logistic regression as described previously (except where R2 < 0.98, in which case a smoothed line of best fit was applied). All curves compared used the same type of curve fitting.
The 18-atom linker conjugate was selected to be the coating antigen in a new whole milk progesterone ELISA using the format described previously. The ELISA was fully validated to determine inter- and in-traassay variability, recoveries, and correlation vs. an established RIA method for 45 samples. The data obtained showed an interassay variability of 12.5% (3 separate assays on different plates of 6 different samples) and an intraassay variability of 5.5% (for 22 different samples across the progesterone range of 1.6 to 21 ng/mL, each assayed in triplicate), which are comparable with previous ELISA reports. The recoveries were computed as percentages of the assayed concentration relative to the spiked concentration of the assayed sample using spiked blank milk. The recoveries ranged from 90 to 111% with a mean of 102% at 5 different concentrations ranging from 1 to 50 ng/mL. Figure 1
shows the regression analysis between the ELISA and the RIA (Coat a Count Progesterone; Diagnostic Products Corp., Los Angeles, CA; validated for bovine milk by Srikandakumar et al., 1986). The correlation with RIA is significant (r = 0.94; P < 0.001; n = 45). The assay was then applied to the profiling of estrous cycles for individual cows, and 2 of those cycles are shown as examples in Figure 2. These samples were stored as previously described and were assayed in 3 batches with most samples assayed together in one batch. The wide log linear range (0.1 to 100 ng/mL; r = 0.99) of the assay allows plotting of the full estrous cycle pattern from trough to peak and allows the analyst to monitor milk progesterone fluctuations clearly and accurately. A further advantage to using such protein conjugates is that they are potentially more stable than the commercial antibody-coated surfaces after immobilization on ELISA plates because the proteins in the present format are used only as steroid attachment intermediates while the antibody functions as the main immunoassay component and so greatly affects the assay sensitivity. The protein conjugates can also be easily purified by simple dialysis, and the format of the assay avoids the need for an antigen-enzyme conjugate, which requires expensive affinity chromatography.
View larger version (12K):
[in this window]
[in a new window]
|
Figure 1. Regression analysis of the correlation between the ELISA test and the progesterone radioimmunoassay (Diagnostic Products Corp.) for whole milk samples taken at various times in the estrous cycle (n = 45) from 5 different cows. The correlation is highly significant (y = 0.81x + 1.1; r = 0.94; P < 0.001; percentage error in the slope of the regression line = 5.6%).
|
|
This research demonstrates that the use of an 18-atom intermediate linker can significantly increase antibody binding signal to progesterone-protein conjugate coating antigen over a 4-atom linker in the milk medium and that this is reflected in a lower assay detection limit compared with shorter linkers. The protein conjugate, when applied in a working ELISA, provides a valuable new tool for whole milk progesterone detection by obviating the need for any pretreatment or predilution of the milk, allowing a wide dynamic range for clear estrous cycle profiling and avoiding several inherent problems with using antigen-enzyme conjugates.
|
ACKNOWLEDGEMENTS
|
|---|
The authors thank Paul Johnstone of Sensortec for supplying milk samples and for helpful discussions and Dexcel Ltd. for provision of RIA services. J. Mitchell thanks the New Zealand Foundation for Research Science and Technology for a scholarship for this work.
Received for publication March 26, 2004.
Accepted for publication June 2, 2004.
|
REFERENCES
|
|---|
Foulkes, J. A., A. D. Cookson, and M. J. Sauer. 1982. AI in cattle based on daily microtitre plate enzyme immunoassay of progesterone in whole milk. Br. Vet. J. 138:515–521.[Medline]
Paek, S., L. G. Bachas, and W. Schramm. 1993. Defined analyte-enzyme conjugates as signal generators in immunoassays. Anal. Biochem. 210:145–154.[Medline]
Self, C. H., and D. B. Cook. 1996. Advances in immunoassay technology. Curr. Opin. Biotechnol. 7:60–65.[Medline]
Srikandakumar, A., R. H. Ingraham, M. Ellsworth, L. F. Archbald, A. Liao, and R. A. Godke. 1986. Comparison of a solidphase, no-extraction radioimmunoassay for progesterone with an extraction assay for monitoring luteal function in the mare, bitch, and cow. Theriogenology 26:779–793.
Waldmann, A. 1993. Enzyme immunoassay (EIA) for milk progesterone using a monoclonal antibody. Anim. Reprod. Sci. 34:19–30.
Wu, Y., J. S. Mitchell, C. J. Cook, and L. Main. 2002. Evaluation of progesterone-ovalbumin conjugates with different length linkers in enzyme-linked immunosorbent assay and surface plasmon resonance-based immunoassay. Steroids 67:565–572.[Medline]
TOP
ABSTRACT
INTRODUCTION
) and the progesterone radioimmunoassay (Diagnostic Products Corp.) (
) for 2 complete reproductive cycles. Error bars represent 1 SD of the mean of replicate determinations of individual milk samples. The ELISA determinations were completed in triplicate. Larger error bars on some points reflect that these points are averages of determinations on different plates with their errors combined by least squares addition.