Anal. Chem. 1999, 71, 4697-4700
A General Method To Perform a Noncompetitive Immunoassay for Small Molecules G. Giraudi,* L. Anfossi, I. Rosso, C. Baggiani, C. Giovannoli, and C. Tozzi
Dipartimento di Chimica Analitica, Universita` di Torino, Via Giuria 5, 10125 Torino, Italy
A new general method to perform a noncompetitive immunoassay for low-molecular-mass analytes (less than 6000 Da) is described and checked using cortisol as a model system. The method is based on the use of a “polydentate ligand” (cortisol-poly(L-lysine) conjugate) able to block the antibody sites unoccupied by the analyte, followed by the replacement of an antibody-bound analyte by an enzyme-labeled analyte (cortisol-horseradish peroxidase), and permits the direct measurement of the analyte bound sites. The observed signal shows a nearlinear correlation with the analyte concentration. The characteristics of interactions between the analyte and polydentate ligand with the specific antibody were studied to perform a preliminary evaluation of the noncompetitive immunoassay for cortisol. The noncompetitive assay was compared with a competitive immunoassay obtained under the same conditions and using the same reagents. The results of the experiments showed a lower detection limit for the noncompetitive model (0.15 ng mL-1 rather than 0.72 ng mL-1), emphasizing that the model is successful. Moreover, as the polydentate ligand is prepared from the same hapten used for the immunogen synthesis, this type of noncompetitive immunoassay appears generally applicable to all small molecules for which antibodies have been obtained. Since the 1960s, when the first written works described chemical analysis using antibodies or binding proteins,1 many different techniques have been reported to perform immunoassays. Although a wide variety in the choice of label, separation method, and procedure is seen, immunoassay methods are of two basic types: competitive and noncompetitive. The competitive method involves the use of a limited concentration of a specific antibody and a labeled (or immobilized) antigen. Various noncompetitive technologies have been described; among these, the most widely used is the sandwich method, which is based on two different antibodies (both used in excess) able to link the antigen. One, immobilized directly or indirectly onto a solid phase, is the “capture antibody”, while the other, with different epitope specificity, is labeled and used as a tracer. Mathematical modeling has shown that noncompetitive assays are potentially more sensitive (i.e., having a lower analyte detection * Corresponding author: (e-mail)
[email protected]. (1) Yalow, R. S.; Berson, S. A. J. Clin. Invest. 1960, 39, 1157-1175. 10.1021/ac981282c CCC: $18.00 Published on Web 09/18/1999
© 1999 American Chemical Society
limit) than competitive assays by orders of magnitude.2 Although lower detection limits for small molecules would be very interesting in different fields (small molecules include metabolic products such as hormones, drugs, toxins, pesticides, and environmental pollutants), until now, immunoassays for low-molecular-mass analytes continue to employ competitive formats because the analytes are too small to permit simultaneous binding of two antibodies (as in the sandwich noncompetitive method). The difference between competitive and noncompetitive assays lies in the detection of antibody occupancy:3 in the competitive one, the analyte-unbound sites instead of the analyte-bound sites are measured. As a consequence, the large number of unbound sites at low analyte concentration is difficult to distinguish from the zero value, just where a large number of unbound sites is present. Thus, the development of a noncompetitive format for haptens requires the skill of detecting the analyte-bound sites of the antibody. Some attempts have been made to develop noncompetitive assays for small molecules,4,5 but they are limited to certain chemical structures6 or they require the development of antimetatype antibodies.7,8 We propose a model that is based on the blocking of free sites of the capture antibody (unbound sites) by means of a “blocking reagent”. This is a large molecule able to bind to different sites at the same time, so that it is more strongly bound than the analyte to the immobilized antibody. In this way, when an enzyme-labeled analyte is added, competition for the antibody occupancy allows the removal of the analyte molecules, but not the blocking reagent. Consequently, the measured signal is associated with analyte bound sites and it is almost linearly correlated to the analyte concentration. EXPERIMENTAL SECTION Reagents. Cortisol (F) and poly(L-lysine) hydrobromide (polylys) were obtained from Sigma (St. Louis, MO). Horseradish peroxidase (EC 1.11.1.7, type VI-A) (HRP) was obtained from BioRad (Hercules, CA). The goat IgG able to recognize rabbit (2) Jackson, T.; Ekins, R. J. Immunol. Methods 1986, 87, 13-20. (3) Ekins, R. P. In Alternative Immunoassays; Collins, W. P., Ed.; John Wiley and Sons: Chichester, U.K., 1985; pp 219-237. (4) Self, C. H.; Dessi, J. L.; Winger, L. A. Clin. Chem. 1994, 40, 2035-2041. (5) Eramenko, A. V.; Bauer, C. G.; Makower, A.; Kanne, B.; Baumgarten, H.; Scheller, F. W. Anal. Chim. Acta 1998, 358, 5-13. (6) Pradelles, P.; Grassi, J.; Creminon, C.; Boutten, B.; Mamas, S. Anal. Chem. 1994, 66, 16-22. (7) Barnard, G.; Kohen, F. Clin. Chem. 1990, 36, 1945-1950. (8) Mares, A.; De Boever, J.; Osher, J.; Quiroga, S.; Barnard, G.; Kohen, F. J. Immunol. Methods 1995, 181, 83-90.
Analytical Chemistry, Vol. 71, No. 20, October 15, 1999 4697
antibody and also the rabbit antiserum for cortisol were a gift of G. Bolelli (Servizio di Fisiopatologia Della Riproduzione, Policlinico S. Orsola, Bologna, Italy). All other chemicals were obtained from Merck (Darmstadt, Germany) and were of analytical grade. Cortisol-3-carboxymethyloxime (F-3-cmo) was previously prepared according to the method described in the literature.9 Conjugates between HRP and F-3-cmo were prepared by the N-hydroxysuccinimide-activated ester method,10 using an enzymesteroid molar reaction ratio of 1:4. The conjugates were purified by low-pressure gel filtration on Sephadex G25, and then the enzyme concentration and substitution ratio (i.e., the mean molar ratio of cortisol to HRP in the conjugate) were determined by spectrophotometric measurements as previously described.10 Synthesis of Poly(L-lysine)-Cortisol Conjugates. The polylys was previously labeled with trinitrobenzenesulfonic acid (TNBS) (molar reaction ratio 1:50 between TNBS and lysine residues), according to Sashidar et al.11 F-3-cmo was activated using the N-hydroxysuccinimide method and added to an aqueous solution (0.15 M NaHCO3) of polylys in a molar ratio of 1:1 between steroid and lysine residues. The precipitate obtained was washed with 0.15 M NaHCO3 to remove the unreacted steroid derivative and then redissolved with 10 M HCl. The solution was diluted and neutralized by means of 1 M NaOH to a final pH of 7-8. Stored at 4 °C, the latter is stable for 1 week. Conjugate concentration was determined by measuring the TNBS absorbance at 420 nm. Blocking Effect. The immobilization of the capture antibodies (rabbit anti-cortisol, diluted 1:40000) into the wells of microtiter plates previously coated with a goat IgG with anti-rabbit capabilities was performed according to the described procedure.12 A 0.2mL aliquot of a solution of polylys-F (concentrations: 0, 0.5, 1, 2, 3, 4, 6 nM) in phosphate buffer (sodium monohydrogen phosphate-sodium dihydrogen phosphate 0.1 mol L-1, sodium chloride 50 mmol L-1, gelatin 0.1% m/v, pH 7.4) was dispensed in duplicate into each well and incubated overnight at room temperature. The wells were washed three times with a 0.05% Tween 20 solution, and then 0.2 mL of the cortisol-HRP conjugate was added (0.15 µg mL-1 in phosphate buffer) and incubated for 10 min at room temperature. After further washing, the color development was carried out with tetramethylbenzidine and hydrogen peroxide,13 and then the absorbance of the wells was read at 450 nm after blocking the color reaction with sulfuric acid. Nonspecific binding of the F-HRP tracer was evaluated by replacing the capture antibody with the dilutant buffer whereas the nonspecific binding of nonconjugated polylys was evaluated replacing the polylys-F conjugate with a solution of polylys. The blocking effect of polylys-F was evaluated from the percentage ratio of the absorbance measured with and without polylys-F added. Exchange Reactions. The kinetics of the binding of enzymelabeled cortisol to the capture antibody and the rate for the exchange reactions between analyte and tracer or polydentate (9) Janovsky, S. C.; Shulman, F. C.; Wright, G. E. Steroids 1973, 23, 49-64. (10) Giraudi, G.; Baggiani, C.; Giovannoli, C. Anal. Chim. Acta 1997, 337, 9397. (11) Sashidar, R. B.; Capor, A. K.; Ramana, D. J. Immunol. Methods 1994, 167, 121-127. (12) Giraudi, G.; Rosso, I.; Baggiani, C.; Giovannoli, C. Anal. Chim. Acta 1999, 381, 133-146. (13) Giraudi, G.; Giovannoli, C.; Baggiani, C.; Rosso, I.; Coletto, P.; Dolci, M.; Grassi, G.; Vanni, A. Anal. Commun. 1998, 35, 183-18.
4698
Analytical Chemistry, Vol. 71, No. 20, October 15, 1999
ligand and tracer were performed on the previously immobilized antibody. The labeled cortisol (0.2 mL, 0.15 µg mL-1) was incubated in the wells for different periods of time (0-60 min) at room temperature. Then the wells were washed and the color development steps were performed as described above. In the case of the exchange reactions, the capture antibody binding sites were previously saturated by incubating the wells overnight with 0.2 mL of a phosphate buffer solution containing an excess of cortisol (10 µg mL-1) or polylys-F (4 nM), and then the mixture was washed as described above. Assays for Cortisol. The noncompetitive assay was performed by a two-step procedure. A 0.1-mL aliquot of a solution of polylys-F (4 nM in phosphate buffer) and 0.1 mL of a solution of cortisol (concentrations: 0, 0.1, 1, 10, 100, and 1000 ng mL-1 in phosphate buffer) was added to the wells containing the capture antibody (immobilized at a dilution of 1:10 000, 1:20 000, 1:40 000, 1:80 000). After 2 h of incubation at room temperature, the wells were washed three times, and then cortisol-peroxidase (0.2 mL, 0.15 µg mL-1 in phosphate buffer) was dispensed and incubated for 10 min at room temperature. After washing the wells and allowing the color development take place, the absorbance reading was taken. The Competitive assay was performed by incubating 0.1 mL of a solution of cortisol (0, 0.25, 1.0, 2.5, 5.0, 10, 20, 40, 80, and 160 ng mL-1 in phosphate buffer) and 0.1 mL of cortisol-HRP (0.4 µg mL-1) overnight at room temperature in the wells coated with the anti-cortisol antibody (immobilized at a dilution of 1:40 000). After washing, the color development took place and the absorbance reading was taken. The selected conditions were previously optimized to achieve maximum sensitivity with the reagents used. The calibration curves for both noncompetitive and competitive assays were fitted by a four-parameter logistic equation (Fig.P software from Biosoft, Cambridge, U.K). RESULTS AND DISCUSSION Measurement of the analyte-bound sites of the antibody can be done theoretically using a reagent able to occupy the free binding sites (i.e., not bound to the analyte) and to prevent the tracer from reaching them. In this way, adding the tracer, it can replace only the analyte, so that the corresponding signal is associated with the analyte-previously bound antibody sites. Blocking Effect. The blocking effect of polylys-F was evaluated as the ability to inhibit the binding of cortisol-HRP to the immobilized anti-cortisol antibody. The polylys-F conjugates with different molecular masses show a different effectiveness in blocking the capture antibody sites (see Figure 1). However, conjugates with higher molecular weight (g129 300) are able to block up to 60-65% of the free sites even after the addition of labeled cortisol. The incomplete blocking of the antibody sites, even at the highest polylys-F concentration, is probably due to the sticklike structure of the polylys-F conjugates, together with their great size, which prevents an efficient occupation of the immobilized antibody. Choosing the polylys-F (MW 418 400) conjugate, we have compared the exchange rates of both cortisol and polylys-F conjugate with the tracer, when this is added after the saturation of the capture antibody. As shown in Figure 2, the tracer quickly replaces the bound cortisol so that the curve is not very different from the one obtained by saturating the free antibody binding
Figure 1. Interaction between cortisol-polylysine conjugates of different molecular weights (polylys MW: ], 19 200; 3, 53 500; 0, 129 300; O, 418 400) and the immobilized capture antibody. The lower signal percentage means that polylys-F is occupying the antibody sites, which cannot bind the tracer, and indicates a higher blocking effect. The effect of nonspecific interaction between polylysine (MW 418 400) and the anti-cortisol antibody is also represented (b).
Figure 2. Binding curve of the labeled cortisol to the antibody as a function of the incubation time. The antibody sites were free (∇) or previously saturated by adding an excess of F (0) or polylys-F (O).
sites by means of cortisol-HRP. Contrarily, in the case of the conjugate, the majority of the antibody sites are not available for binding with the tracer even after 30-40 min, when practically all the cortisol has been replaced. As a consequence, the polylyscortisol conjugate can be used as a blocking reagent in the noncompetitive assay just described by choosing a time condition for the tracer incubation that permits the widest substitution of the analyte and, at the same time, ensures the effectiveness of blocking. The preliminary selected condition was 10 min for tracer incubation. Moreover, the binding reaction rate for the polydentate ligand to the capture antibody (that is, the blocking reaction) was evaluated and the finding showed that it can be considered complete after 2 h (data not shown). Noncompetitive Assay. A calibration curve for cortisol was carried out to evaluate the effectiveness of the model described in performing a noncompetitive assay for small molecules. Theoretically, our model is based on three steps: first, the capture of the analyte by means of the specific antibody, then the block of the unbound antibody sites by the polylys-F conjugate and, finally, the removal of the analyte by exchange with the enzymelabeled analyte. Although a reduced sensitivity would have been expected from the simultaneous incubation of the analyte and the blocking reagent together, when compared to their being added sequentially (because of their competition for the antibody
Figure 3. Calibration curve for the described noncompetitive assay of cortisol (antibody dilution 1:40 000). The inset shows the nearlinear trend of the curve, and the bars represent standard deviation of the data.
occupancy), no advantages arose from the sequential addition of the reagents. This may be due to the difference in the diffusion rate of cortisol (a small molecule, MW 362.5) and polylys-F (MW higher than 418 400) which reduces the competition when the simultaneous incubation is performed. Otherwise, it may be due to the fact that the antibody-cortisol complexes have a high dissociation rate which allows the fast establishment of the equilibrium between cortisol and polylys-F, even if the latter is added afterward. Indeed, a simultaneous 24-h incubation of cortisol and polylys-F gives a very flat curve, as expected (data not shown). Figure 3 shows the calibration curve (antibody dilution: 1:40 000), where the signal represents the difference between the measured absorbance and the one measured for the zero standard. Because the effectiveness of the polylys-F block was not complete, we obtained a nonspecific signal, due to the 35-40% of unblocked antibody sites. The direct correlation between measured signal and analyte concentration is confirmed by the calibration curve (see Figure 3), meaning that the suggested model enables us to perform a noncompetitive assay (i.e., to measure the analyte-bound sites) for low-molecular-mass analytes. Besides, contrary to that observed for the competitive assay, as far as the immobilized antibody dilution is concerned, lowering the antibody concentration worsens the assay sensitivity (as shown in Figure 4). Furthermore, the least detectable concentration (i.e., the cortisol concentration corresponding to the absorbance equal to that of the zero standard plus three times its standard deviation) was measured without any optimization in the method, using the same reagents and anti-cortisol antibody dilution employed in the optimized competitive assay. The comparison between the value measured in the noncompetitive assay with that of the competitive assay for cortisol reveals a worse sensitivity for the competitive format (0.72 ng mL-1 rather than 0.15 ng mL-1). This limited increase of sensitivity can be attributed mainly to the high level of the nonspecific signal due to the incomplete blockage of the antibody sites by the polylys-F. In fact, as previously discussed, the sensitivity increase is related to the possibility of measuring near-zero signals at zero analyte concentration. It is worthwhile noting that the polydentate ligand proposed can be easily adapted to all small molecules, requiring only a functional group (commonly a carboxylic group) that can react Analytical Chemistry, Vol. 71, No. 20, October 15, 1999
4699
Figure 4. Detection limits (cortisol concentration corresponding to the absorbance equal to that of the zero standard plus three times its standard deviation) measured with the proposed method as a function of the immobilized antibody dilution.
with amino groups. The required molecule can be the same hapten derivative used for the synthesis of immunogens and tracers (or solid-phase antigens) in competitive immunoassays. As a consequence, as long as a competitive assay for the hapten exists, there is always a modified hapten available with the suitable functional group to synthesize the polydentate ligand. The main interest of the proposed method lies in the possibility of obtaining similar results with analytes observed at very low concentrations in real
4700
Analytical Chemistry, Vol. 71, No. 20, October 15, 1999
matrixes. Usually, these analytes require the use of high-affinity antibodies. Consequently, the substitution between the analyte and the tracer (obtained with cortisol simply by washing with a Tween solution and adding the tracer) might require, in these cases, stronger washing (for example, with higher ionic strength or different pH). However, the described method is based on the different exchange rates between the polydentate ligand and the analyte from the antibody complexes. This difference is expected independently from the affinity of the antibody used, because the polydentate ligand behaves as a polydentate molecule (able to bind to different antibody sites at the same time), which always shows a better affinity than the corresponding analyte for the specific antibody. Finally, it is interesting to mention an improvement in the assay performance, particularly in sensitivity, even in a preliminary step of the study. Further investigations will be directed toward the optimization of the method used (reagent’s concentration, time incubation, etc.) to obtain maximum sensitivity and, contemporarily, also toward the choice of different polydentate reagents (not based on the use of poly(L-lysine) characterized by a higher stability.
Received for review November 18, 1998. Accepted July 19, 1999. AC981282C
