ANTIGEN-VERSUS ANTIBODY-IMMOBILIZED ELISA PROCEDURES BASED ON A BIOTINYL-ESTRADIOL CONJUGATE

J. sreroidBiochem.Vol.33,No. 6, pp. 1161-1166,1989 Printedin Great Britain

0022-4731/89 $3.00+ 0.00 PergamonPressplc

ANTIGEN- VERSUS ANTIBODY-IMMOBILIZED ELISA PROCEDURES BASED ON A BIOTINYL-ESTRADIOL CONJUGATE DAVID M. BODMER*,LOUISX. TIEFENAUER~and ROGERY. ANDREW Medical Bioanalytics Project, Paul Scherrer Institute, 5232 Villigen PSI, Switzerland (Received 2 February 1989)

biotinyl-6cr-estradiol derivative (Bio-E,) was synthesized and used as the key component in antigen- and antibody-immobilized ELISA techniques, and the relative merits of the two methods were compared. A precise and reproducible antigen-immobilization was achieved in avidin-coated microtiter plates with Bio-E,. This assay, when completed by the incubation with primary antibody and second antibody-peroxidase conjugate, has a very low detection limit (6 pg/ml estradiol) but required a long incubation time with primary antibody to reach equilibrium. At non-equilibrium conditions, using a high antibody concentration, the assay could be very fast and sensitive. In the antibody-immobilized assay, the Bio-E, was added to compete with the estradiol present in the calibrator or sample and visualized with a streptavidin-peroxidase conjugate. The detection limit is higher (34 pg/ml), but the specificity was superior and the incubation time to reach equilibrium shorter as compared to the antigen-immobilized assay. Therefore, the antibody-immobilized assay appeared to be ideal for the classical ELISA technique, whereas the antigen-immobilized method seemed to be best suited for automated assay systems using antibody in excess. Summary-A

INTRODUCTION The popularity of ELISA for the detection of antigens and antibodies in body fluids has been on the increase during the last few years [l-4]. This type of assays is based on the immobilization of antigen or antibody on a solid phase [5,6]. Various systems have been established to measure the concentrations of steroids [7-131. A method based on immobilized progesterone has been found to have advantages [14] over the other methods used to assay this hormone. Recently, a time resolved fluoroimmunoassay, based on the immobilization of estradiol-6-CMO-thyroglobulin, has been developed [ 151. So far, systematic and comprehensive studies are lacking in comparing the merits of the techniques using antigen- versus antibody-immobilized microtiter plates for estimating the concentration of steroids. Such information is requird for the development of methods for automated analysis in a clinical laboratory, where assay parameter such as incubation time, volume of the solution etc. are set by the manufacturers of the instrument. *Present address: Department of Drug Delivery Systems, Sandoz Ltd, 4002 Basel, Switzerland. tAuthor to whom all correspondence should be addressed. Abbreviufions: Bio-E,, biotinyl-estradiol; CMO, carboxymethyloxime; DMF, dimethylformamide; ELISA, enzyme linked immunosorbent assay; F(ab’),, antigen specific fragment of the second antibody molecule; NSB, non-specific binding; PBS, phosphate buffered saline; RIA, radioimmunoassay.

The present study, using estradiol as a model, is aimed at comparing the antigen- versus the antibodyimmobilized system with respect to kinetics, detection limits, precision and stability of the plate coatings. The relative advantage of each method is discussed with respect to practical use, and general suggestions are made for development of future ELISA systems. As a key reagent, a biotinyl-estradiol derivative was synthesized, and used to prepare an antigen., coated plate via previously immobilized avidin. Alternatively, antibody-coated plates can bind this same conjugate, which then can be detected with streptavidin-peroxidase. EXPERIMENTAL Materials

Avidin, streptavidin and protein A, horseradishperoxidase conjugates of avidin, streptavidin, extrAvidinrM, second antibody, F(ab’), (3.7 peroxidase units/ml), and protein A were obtained from Sigma, St. Louis, MO., U.S.A. Bovine non-fat dry milk (Rapilait) was purchased from a local food store (Migros), Switzerland. Anti-estradiol serum was collected from a rabbit immunized with estradiol6-CMO-bovine serum albumin. 6a-amino-estradiol was synthesized and analyzed as described before [16]. Unless noted otherwise, all other reagents and solvents were purchased from Merck, Darmstadt, F.R.G. or Fluka, Buchs, Switzerland. Microtiter plates

1161

DAVID M. BODMER et al.

1162

(Nunc-Immuno Plates I, 96 flat-bottom wells) were obtained from Nunc, Roskilde, Denmark. Methods Biotinyl-estradiol

Synthesis and purification of 3,17P-dihydroxy1,3,5(10)-estratrien-6a-N-(~-biotinyl)-amino-caproamide (Bio-E,): 6cr-amino-estradiol (5 mg, 17 pmol) was dissolved in 200~1 in DMF containing 2mg (45 pmol) triethylamine. 16.6 mg (30 pmol) biotinylE-aminocaproicacid-N-hydroxy(3-sulfo)succinimide ester (Pierce Chemical Company, Rockford, U.S.A.) was dissolved in 50~1 of DMF and added to the 6a-amino-estradiol solution. The reaction mixture was stirred at room temperature for 20 h. The excess biotin reagent was precipitated by the addition of 750 ~1 solvent mixture I (chloroform/methanol/acetic acid, 93 : 7: 1, v/v) and separated by centrifugation (3OOOg, 10 min). The precipitate was washed three times with solvent mixture I, and the pooled supernatants (1.5 ml) were purified by high pressure liquid chromatography on a Bio-Sil HP-10 column (BioRad Lab., Richmond, U.S.A.), using a gradient from 7 to 30% of methanol in solvent mixture I. After a first run, the partially purified material was rerun on the same chromatographic system. The pure product was eluted after 10.5 min with 51% yield (87pmol). U.V. I,,, (ethanol): 281 nm (6 = 2100). m.s. m/e: 400 (23%), 270 (100%) 182.5 m (400+270), 339 (IS%), 287 (12%), 226 (44%), 114 (72%). Quantities of Bio-E, are given in weight units of the whole molecule and do not represent “estradiol equivalent” units. Preparation

of antigen-immobilized

plates

Microtiter plates were coated with 5 pg/ml avidin dissolved in borate coating buffer (10 mM, pH 9.6) containing 10 mM sodium chloride and 0.01% thimerosal (200 PI/well). After an overnight incubation at room temperature, the coating solution was discarded and the plates were washed three times with blocking buffer (Tris-HCl, lOmM, pH 5.5, containing 0.1% bovine non-fat dry milk, 10 mM ethylendiaminetetraacetate and 140 mM sodium chloride). Antigen was bound to the avidin-coated plates by incubation of 150 p 1 Bio-E, (15 ng/ml) in assay buffer (PBS, containing 0.05% bovine non-fat dry milk and 0.01% thimerosal) at room temperature for 1 h. Maximum binding was reached after 30 min. The antigen solution was discarded, the plates were washed five times and used directly or kept covered with assay buffer containing 0.1% dry milk at 4°C until used. Preparation

of antibody-immobilized

plates

Microtiter plates were first coated with protein A (0.5 pg/ml) in 0.1 M acetate buffer pH 5.1. After incubation (overnight, room temperature) the plates

were washed three times with blocking buffer as described above. Rabbit-anti-estradiol, diluted in assay buffer, was then bound with virtually 100% yield to the plate by incubation at room temperature for 30min. Immunoassay procedures

Unless noted otherwise, all incubations were performed at room temperature in a total volume of 200 PI/well. Each wash step consisted of 5 fill/aspirate cycles with 50mM Tris-HCl, pH 7.8, containing 0.1% (v/v) Tween 20. Incubation of Bio-E,, antiestradiol (first antibody), calibrators and all peroxidase conjugates were done in the assay buffer mentioned above. Bound perioxidase activity was determined by incubating the washed plates for 30 min with 0.2% tetramethylbenzidine dissolved in acetate buffer (160 mM, pH 5.0) containing 0.012% hydrogen peroxide. The enzyme reaction was stopped by the addition of 50 ~1 of 1 N sodium fluoride dissolved in acetate buffer (160 mM, pH 5.0). Absorbances were read at 650 nm using an ELISAplate reader (EAR 400 FW, SLT Laboratories, Austria). Means and standard errors (n > 8) were computed for every concentration tested. Specificity was calculated from the ratio of the concentrations of estradiol to the tested substance at 50% inhibition of tracer binding. All values were corrected for NSB. Protocol of the antigen -immobilized

assay

Calibrators and controls (100 ~1) were added to an antigen-coated plate. 100 ~1 anti-estradiol in a final dilution of 1: 7000 containing 0.002% avidin in assay buffer was added. The plate was covered with a film, briefly shaken and incubated at 37°C for 1 h. After washing, 200 ~1 of diluted (1: 10,000) F(ab’),peroxidase was added. After shaking the plate was incubated at room temperature for 2 h and washed. The peroxidase activity was then determined as above. Antibody-immobilized

procedure

To antibody-coated plates, 100~1 of Bio-E, (150 pg/ml if not indicated otherwise) and 100 ~1 of estradiol calibrator in assay buffer were added. After various incubation times and washing, 200 ~1 of streptavidin-peroxidase conjugate (50 ng/ml) was added. For the reduction of NSB, Tween 20 (final concentration 0.5% v/v) was present in the assay buffer during this incubation step. The incubation was stopped after 30 min by washing and the enzyme activity was measured as described above. All incubations were performed at room temperature. RESULTS

Bio-E, has been successfully used for both the antigen- and the antibody-coated plates (Fig. 1, A and B). In procedure A, Bio-E, was fixed on an

ELISA based on biotinyl-estradiol

c)

Fig. 3. Binding kinetics of the components in the antibodyimmobilized procedure 8. Bio-E, (A) in a concentration of 150 pg/ml and streptavidin-peroxidase (0) (50 ng/ml) were incubated at room temperature.

blotin-estradiol

peroxidase

Fig. 1. ELISA procedures based on the biotinyl-estradiol conjugate. Bio-E, is used as the key component in both the antigen-immobilized (A) and the antibody-immobilized (B) procedure.

avidin-coated microtiter plate. Titration experiments proved a final concentration of 15 ng/ml (2.25 ng/well) to be optimal for Bio-E, with respect to detection limit, sensitivity and precision of the assay. Plates with immobilized antigen could be stored dry or covered with assay buffer at 4°C for four month without any loss of antibody binding activity. In procedure B, plates were coated with protein A and sub~uently with anti-estradiol. The plates were stable for several months, when stored dry or covered with assay buffer at 4°C. In procedure 8, Bio-E, was applied in competition to estradiol in the calibrator. The optimal concentration of Bio-E, is 150 pg/ml. The binding kinetics in the two procedures were completely different. In Fig. 2 the binding of the I

OJ2. 2.0-

1 5 A”------~---

B

60

1163

lx)

180

240

Fig. 2. Binding kinetics of the components in the antigenimmobilized procedure A. Anti-estradiol (0) was diluted 1:7000 in assay buffer and incubated at 37°C. F(ab’),peroxidase (0) was diluted 1: 10,000 and incubated at room temperature.

primary antibody to the antigen-coated plates (procedure A) is displayed as function of time. Although the incubation was at 37”C, equilibrium was reached only after 2 h. The binding of F(a~)~-~roxidase was even slower. This conjugate was chosen after comparative experiments with second antibodyand protein A-peroxidase conjugates. NSB was slightly lower with F(ab’),-peroxidase (12 f 1%) compared to 13 + 2% and 17 f 4% (n = 12) with the other peroxidase conjugates. In addition, the coefficient of variation (CV) was much lower using the F(ab’),peroxidase conjugate. For simplicity the last two incubation steps in the antigen-immobilized procedure (see Fig. 1A) were stopped after I and 2 h, respectively. Full equilibrium was not reached under these conditions. For prevention of non-specific binding, the primary antiserum was pretreated with 0.002% avidin. In the antibody-immobilized system (procedure B), the binding of the components was much faster (Fig. 3): Binding of Bio-E, to the antibody reached equilibrium after 10 min and binding of streptavidinperoxidase to Bio-E, was completed after 20min. Both incubations are done at room temperature. After comparing avidin-, extrAvidin- and streptavidin-peroxidase conjugates, streptavidin-peroxidase turned out to be most suitable because of low NSB. Variations in the concentration of the key components have differential effects on the assay sensitivity. Increasing the antibody concentration using antigen-coated plates, a reduced sensitivity is observed (Fig. 4). In this antigen-immobilized system an excellent 50%-intercept (180 pg/ml, corresponding to 18 pg/well of estradiol), but a low optical density is observed at low antibody concentration (Fig. 5). However, using a ten-fold higher antibody concentration, a high absorbance is obtained, albeit at a ninefold loss of sensitivity. With an appropriate incubation time, a sensitive calibration curve concomitant with a sufficiently high optical signal can be reached: At high antibody concentration (dilution 1: 2000) the 50%-intercept drops from 750 to 450pg/ml when changing the incubation time from 2 h to 15 min (see Fig. 5). Variation of the Bio-E, concentration has no effect on the 50%.intercept (data not shown).

DAVID M. F%ODMER

1164

”

I

12.5

8

25

III

sb

500

ICAY

31

310 EsbaI

The calibration curve in the antibody-immobilized case is directly influenced by the Bio-E, concentration. There is a nearly linear relationship between the SO%-intercept and the Bio-E, concentration (see Fig. 6). The antibody concentration does not determine the sensitivity in this case (data not shown). Data concerning precision and specificity are summarized in Table 1. Precision is excellent in procedure A and B. The coefficient of variation is strictly dependent on the last incubation step in the antigenimmobilized procedure. A significant difference in the crossreactivity data between the two procedures is observed (see Table 1). DISCUSSION

In this study two main goals have been pursued: Firstly, comparison of the characteristics of antigenversus an antibody-immobilized ELISA systems, and secondly, the optimization of theses procedures with respect to the volumetric and temporal requirements of automated assay systems. Automation requires

IOW

100

@iJw

Fig. 4. Calibration curves using various anti-estradiol concentrations in the antigen-immobilized procedure A. Calibrator in various concentrations was incubated with the antibody at 37°C on antigen-immobilized plates (see Methods). The antibody was diluted I:2000 (O), I:4000 (O), 1:8000 (a), 1: 12,000 (B) and 1: 16,000 (A). The insert displays the SO%-intercept values as a function of relative antibody concentrations.

EP w/ml

al.

I

100150250

Esbww

L

et

OD. 2.0

J Fig. 5. Influence of incubation time and antibody concentration on the signal in the antigen-immobilized procedure A. Anti-estradiol was incubated at 37°C in dilutions of 1: 2000 (0) and 1: 20,000 (0). The dashed lines show the absorbance, the full lines the 50%~intercept concentrations of the corresponding calibration curves.

lwo

3100

ww

Fig. 6. Calibration curves using various Bio-E, concentrations in the antibody-immobilized procedure B. Calibrator in various concentrations was incubated together with Bio-E, for 30min at room temperature on antibodyimmobilized plates (see Methods). The Bio-E, concentration were 0.15 (A), 0.3 (m), 0.6 (U), 1.5 (0) and 3 (0) ng/ml. The insert displays the 50%~intercept values in function of Bio-E, concentrations.

not only good sensitivity but also high precision and fast kinetics. To achieve this goal, we combined the well-known avidin-biotin complex with an antigen structure displaying a reduced bridge effect, as recently exemplified [16] for new ‘251-tracer structures. This tracer, having an 6c(-amido-linkage, exhibits very little, if any, bridge effect. We therefore synthesized a biotinyl-estradiol as shown in Fig. 7. The c-aminocaproic acid spacer connecting the carbonyl group of the biotin moiety with the antigen part (6cr-amino-estradiol) improves the avidin-biotin interaction [ 17,181, facilitating binding in the deep binding pocket of the avidin molecule [6]. The positive effect of this spacer group was confirmed in our laboratory using biotinylated antibodies bearing a long and short spacer, respectively [ 191. Bio-E, was the key substance for both the antigenand the antibody-immobilized procedure (see Fig. 1). The remainder of the reagents was identical for both Table I. Comparison of the assay performance in the two ELISA Drocedures investieated (see Fig. 1) Assay parameters Sensitivity Detection limit (pg/ml) SO%-Intercept (pg/ml) Specificity Crossreactivity (%) of estrone estriol estradiol-3-glucuronide estradiol-17-glucuronide ethinyl estradiol Precision CVb (%) (n = 8) intraplate interplate

Antigen Antibody immobilized (A)1 immobilized (B) 6 40

17 4 3.6 0.12 0.07 6 5

34 310

3.6 0.6 0.1 0.09 0.0003

4 3

“See Fig. I. bCoefficient of Variation. All steps were performed under assay conditions (see Methods). The detection limit is defined as B, 5 2 standard deviations, the 50%~intercept as the estradiol concentrations at B/B, = 50% and the specificity (crossreactivity) as the ratio of the concentration of estradiol to the tested compound at B/B, = 50%

ELISA based on biotinyl-estradiol

“0

fl

cm

1:

”

HN--C-N/ ii

ia

NH

5

I. I_ i 0 0 0 Fig. 7. Chemical structure of biotinyl-estradiol. Q Estradiol moiety. @ Spacer moiety. @ Biotin moiety.

k

systems. It was therefore possible to compare procedural differences of the two systems, without interference by changes in reagent composition. In addition, this approach is very versatile: Only E,-Bio and the primary antibody must be exchanged for another hormone, e.g. for progesterone. The rest of the protocol is common to all ELISA, a fact that may signi~cantly simplify the routine use of such assays in the clinical laboratory. The high affinity of the biotin moiety in the Bio-Et conjugate to the avidin-coated on the plate provides a precise and reproducible antigen-immobilization method. This coating is a base for the very low coefficient of variation. From a practical point of view this indirect antigen coating offers some advantages: avidin-coated plates can be prepared in advance and they can be stored for a long period. Because avidin is coated in excess, fluctuations in coating conditions do not have any direct impact on the precision of the antigen coating. Additionally, any biotinylated molecule can be bound to this plates irreversibly and quantitatively with high precision 1191.Regarding optimal concentration of Bio-E,, much more is needed in case of antigen- (2.25 nglwell) than in case of antibody-immobili~tion (15 pg/well). Thus an excess of immobilized estradiot is required for optimal antibody binding. This observation is in agreement with the findings presented elsewhere [20,21]. The authors demonstrated that antibody binding to antigen-coated plates is at an optimum, if the plate surface is saturated with antigen. Only under these conditions, divalent binding of an individual antibody molecule to two immobilized antigen molecule can be achieved, Differences in the sterical environment and diffusion properties inherent in the two procedures is revealed in the binding kinetics. Because of the high affinity of the antigen to the antibody, binding is assumed to be controlled by diffusion. The diffusion of the large antibody molecule is about ten times slower than that of the small Bio-E, molecule assuming that the diffusion constant is proportional to the square root of the molecular mass. Therefore, the binding of the antibody to the antigen-coated on the plates is expected to occur much slower than the binding of the small Bio-E, to the antibody-coated. This speculation is confirmed by our observations: an excess of antigen on the plastic surface is needed to enhance the probability of antibody binding. We

1165

assume that the length of the spacer connecting biotin with estradiol in the Bio-E, molecule is of great importance with respect to the kinetics. This type of sterical influence is now being investigated. The binding of the last component, the second antibody-peroxidase or the streptavidin-peroxidase conjugate, respectively, seems to be determinated by the avidities. The molecular masses are only different by a factor of about two so that the diffusion rate is comparable. However, the extremely fast binding of the strepta~din-~roxidases conjugate is due to the high affinity constant of streptavidin to biotin. The affinity constants of the second antibody moiety in the peroxidase conjugates to the rabbit antibody is lower by several orders of magnitude. The equilibrium of this binding is reached only after a few hours. Because peroxidase is quite unstable under assay conditions, the incubation has been stopped after 90 min at non~quilib~um conditions. A linear dependence of the 50%-intercept value from the Bio-E, concentration is only observed in the antibody-coated case (see Fig. 1B). A direct competition occurs between the “tracer” and estradiol present in the calibrator or sample solution. By increasing the Bio-E, concentration, statistically more of the steroid derivative will bind to the antibody: thus, at a given calibrator concentration, a higher signal is observed. This linear relationship (see insert Fig. 6) indicates that the Bio-E, concentration is the major determinating factor for the sensitivity. The concentration of Bio-E, immobilized on plates has an influence only on the intensity of the signal but not on the position of the calibration curve. Obviously, Bio-E, immobilized on the plastic surface does not bind to the antibody at the same rate as freely movable estradiol. Therefore, full equilib~um is only achieved when the incubation times exceeds the diffusional constraints imposed by the need of the antibody to reach the Bio-E,-coated surface. However, the concentration of the antibody determines the sensitivity of the calibration curve in the antigen-immobilized procedure. The more antibody binding sites are present at a given estradiol concentration the more antibody molecules wili bind to the antigen-coated plate. In this procedure the sensitivity can be adjusted by varying the concentration of the antibody. Actually, a limitation is given by a lower signal intensity with decreasing antibody concentrations. The observed lower specificity in the antigenimmobilized procedure is not fully understood. Recognition of the antigen by the antibody might be impeded by sterical hindrance. In summary, we conclude that both the antigenand the antibody-immobilized procedures result in precise assays. The fast kinetics and the high specificity of the antibody immobilized procedure appears to be best suited in most cases. In an automated system, the antigen-immobilized procedure probably may be prefered, because of the very short incubation

DAVIDM. BODMERet al.

1166

times and the possibility excess.

of the use of antibody

Acknowledgemenrs-The authors the technical assistance of Mr J.-P. Mr A. Schuler and Mrs M. Fuchs Vijayalaxmi in the preparation of

in

gratefully acknowledge

10. Turkes A., Dyas J., Read G. F. and Riad-Fahmy

11.

Dobler, Mr J.-J. Hefti, and the help of Dr C. this manuscript.

REFERENCES immunoI. Engvall E. and Perlmann P.: Enzyme-linked sorbent assay (ELISA). Quantitative assay of immunoglobulin G. Immunochemistry 8 (1971) 871-874. 2. Engvall E. and Perlmann P.: ELISA III. Quantitation of specific antibodies by enzyme-labelled anti-immunoglobulin in antigen-coated tubes. J. Immun. 109 (1972) 1299135. 3. Van Weemen B. K. and Schuurs A. H. W. M.: Immunoassay using antigen-enzyme conjugates, FEES Lett. 15 (1971) 232-236. 4. Van Weemen B. K. and Schuurs A. H. W. M.: Immunoassay using hapten- enzyme conjugates. FEBS Let?. 24 (1972) 77-81. CRC Press, Boca 5. Maggio E. T.: Enzyme-immunoassay. Raton, Florida (1980) pp. 53-70. 6. Tijssen P.: Practice and Theory of Enzyme Immunoassays. Elsevier, Amsterdam (1985) pp. 14-20, p. 22. 7. Van Weemen B. K. and Schuurs A. H. W. M.: The influence of heterologous combinations of antiserum and enzyme labeled oestrogen on the characteristics of estrogen enzyme immunoassays. Immunochemistry 12 (1975) 667670. M., Haryu A., Kurosaka K. and Nambara 8. Numazawa T.: Picogram order enzyme immunoassay of estradiol. FEBS Lett. (1977) 79 396398. D., Read G. F., Joyce B. G. and Walker 9. Riad-Fahmy R. F.: Steroid immunoassays in endocrinology. In Immunoassays for the 80s (Edited by A. Voller, A. Bartlett and D. Bidwell). University Park Press, Baltimore (1981) pp. 205-226.

13.

14.

15. 16.

17.

18.

19. 20.

21.

D.: Enzyme immunoassay for specific determination of the synthetic estrogen, ethinyl estradiol, in plasma. Clin. Chem. 27 (1981) 901-905. Shah H. P. and Joshi U. M.,: A simple, rapid and reliable enzyme-linked immunosorbent assay (ELISA) for measuring estrone-3-glucuronide in urine. J. Steroid Biochem. 16 (1982) 283-286. Roda A., Girotti S., Piacentini A. L., Preti S. and Lodi S.: Development of a sensitive, direct luminescent enzyme immunoassay for plasma estradiol-17/I. Anaiyf. Biochem. 156 (1986) 267-273. Marcus G. J. and Durnford R.: Estradiol assay by microtitre plate enzyme immunoassay. J. Steroid Biochem. 29 (1988) 207-212. Elder P. A., Yeo K. H. J., Lewis J. G. and Clifford J. K.: An enzyme-linked immunosorbent assay (ELISA) for plasma progesterone: immobilized antigen approach. Clin. Chim. Acta 162 (1987) 199-206. Lovgren T. N. E.: Time-resolved fluoroimmunoassay of steroid hormones. J. Steroid Biochem. 27 (1987) 47-51. Tiefenauer L. X., Bodmer D. M., Frei W: and’ Andres R. Y.: Prevention of bridge binding in immunoassay: A general estradiol tracer structure. J. Steroid Biochem. 32 (1989) 251-257. Costello S. M., Felix R. T. and Giese R. W.: Enhancement of immune cellular agglutination by use of an avidin-biotin system. Clin. Chem. 25 (1979) 1572-1580. Leary J. J., Brigati D. J. and Ward D. C.: Rapid and sensitive calorimetric method for visualizing biotinlabeled DNA probes hybridized to DNA or RNA immobilized on nitrocellulose: Bio-blots. Proc. N&z. Acud. Sci. U.S.A. 80 (1983) 40454049. Tiefenauer L. X. and Bodmer D. M.: Z. Analyt. Chem. 330 (1988) 342. Vos Q,, Klasen E. A. and Haaijman J. J.: The effect of divalent and univalent binding on antibody titration curves in solid-phase ELISA. J. Immun. Meth. 103 (1987)47-54. Lew A. M.: The effect of epitope density and antibody affinity on the ELISA as analysed by monoclonal antibodies. J. Immun. Mefh. 72 (1984) 171-176.