Glycerin Suppression of Fluorescence Self-Quenching and

Radiodiagnostic Products, NCSR “Demokritos”, 15310 Athens, Greece ... For a more comprehensive list of citations to this article, users are en...
0 downloads 0 Views 180KB Size
Download PDF
Anal. Chem. 2007, 79, 647-653

Glycerin Suppression of Fluorescence Self-Quenching and Improvement of Heterogeneous Fluoroimmunoassay Sensitivity Panagiota S. Petrou, Christos Mastichiadis, Ion Christofidis, and Sotirios E. Kakabakos*

Immunoassay/Immunosensors Laboratory, Institute of Radioisotopes & Radiodiagnostic Products, NCSR “Demokritos”, 15310 Athens, Greece

Fluorescent labels find wide application in immunoassays and immunosensors as well as in protein and DNA chips. However, the use of fluorescent labels in applications requiring high detection sensitivity is limited by fluorescence self-quenching observed when a relatively high number of fluorescent compounds is introduced in the recognition molecule. Here we describe a simple method that suppresses effectively fluorescence self-quenching observed when highly labeled antibodies are used as labels in immunoassays. This was achieved by treating the microtitration wells after the completion of the immunoassay with a glycerin solution followed by 15-min incubation of the emptied wells at 37 °C. The remedial action of this method on self-quenching was studied through a noncompetitive immunofluorometric assay for rabbit γ-globulins employing a sheep anti-rabbit γ-globulin antibody labeled with fluorescein at molar ratios ranging from 1.0 to 17.4. The glycerin/thermal treatment increased the fluorescence signal measured directly onto the solid surface by 9.2-117% for the antibodies with molar ratios of 1.0-17.4, compared with the values obtained prior to treatment. Furthermore, fluorescence self-quenching was completely removed for labeling ratios up to 14.0. The assay sensitivity was improved 2-4 times by the glycerin/thermal treatment when heavily fluoresceinated antibodies are used as labels (molar ratio g5.6). The proposed method resulted also in increased fluorescence signals when labels other than fluorescein were used and improved considerably the detection of protein spots on silicon dies. Fluorescent compounds have been extensively used in the past as labels in immunohistochemical1 and immunoassay2 applications. More recently, the development of optical immunosensors3,4 as well as of DNA5-8 and protein chips,9-13 opened new application * Correspondingauthor.E-mail: [email protected]: ++302106515573. (1) Haugland, R. P. Handbook of Fluorescent Probes and Research Chemicals, 6th ed.; Molecular Probes, Inc.: Eugene, OR, 1996. (2) Hemmila, I. Clin. Chem. 1985, 31, 359-370. (3) Mastichiadis, C.; Kakabakos, S.E.; Christofidis, I.; Koupparis, M.A.; Willets, C.; Misiakos, K. Anal. Chem. 2002, 74, 6064-6072. (4) Rowe, C. A.; Scruggs, S. B.; Feldstein, M. J.; Golden, J. P.; Ligler, F. S. Anal. Chem. 1999, 71, 433-439. (5) Fodor, S. P. A. Science 1997, 16, 40-44. 10.1021/ac061492m CCC: $37.00 Published on Web 11/30/2006

© 2007 American Chemical Society

areas that increased considerably the utilization of fluorescent labels. So far, fluorescein is the most widely used label due to its high absorptivity and quantum yield when conjugated to biomolecules.1,14 Besides, the great variety of the commercially available fluorescein derivatives and its low MW, which enables multiple labeling of biomolecules, facilitated further its applications. However, multiple labeling of biomolecules with fluorescein is not always beneficial in terms of signal increase. As has been reported, decreased fluorescence signals have been observed when proteins are labeled with multiple fluorescein molecules.1,14-16 This phenomenon has been attributed to the small Stokes shift of fluorescein that permits energy transfer from one fluorescent molecule to another resulting in self-quenching.1 Fluorometric analysis indicated that self-quenching occurs in solutions of free fluorescein at concentrations higher than 1.8 × 10-6 M.17 However, in the case of fluorescein conjugated to proteins, quenching occurs at considerably lower concentrations, due to the localization of the fluorescent molecules in a confined space. Thus, concerning antibody molecules in solution, quenching has been observed for labeling ratios (F/P) higher than 5 for intact molecules1,14 and 2-3 for Fab fragments.18 Similar quenching effects have been also reported for other fluorescent dyes such as Cy5 and Cy7.19 In order to surpass these drawbacks, novel dyes (6) Schena, M.; Heller, R. A.; Therault, T. P.; Konrad, K.; Lachenmeier, E.; Davis, R. W. Trends Biotechnol. 1998, 16, 301-306. (7) Potyrailo, R.; Conrad, R. C.; Ellington, A. D.; Hieftje, G. M. Anal. Chem. 1998, 70, 3419-3425. (8) Epstein, J. R.; Leung, A. P.K.; Lee, K.-H.; Walt, D. R. Biosens. Bioelectron. 2003, 18, 541-546. (9) Schuderer, J.; Akkoyun, A.; Brandenburg, A.; Bilitewski, U.; Wagner, E. Anal. Chem. 2000, 72, 3942-3948. (10) Bernard, A.; Michel, B.; Delamarche, E. Anal. Chem. 2001, 73, 8-12. (11) Christodoulides, N.; Tran, M.; Floriano, P. N.; Rodriguez, M.; Goodey, A.; Ali, M.; Neikirk, D.; McDevitt, J. T. Anal. Chem. 2002, 74, 3030-3036. (12) Delehanty, J. B.; Ligler, F. S. Anal. Chem. 2002, 74, 5681-5687. (13) Fall, B. I.; Eberlein-Koenig, B.; Behrendt, H.; Niessner, R.; Ring, J.; Weller, M. G. Anal. Chem. 2003, 75, 556-562. (14) Diamandis, E. P.; Christopoulos, T. K. Fluorescence immunoassays. In Immunoassay; Diamandis, E. P., Christopoulos, T. K., Eds.; Academic Press: San Diego, CA, 1996. (15) The, T. H.; Feltkamp, T. E. W. Immunology 1970, 18, 865-873. (16) The, T. H.; Feltkamp, T. E. W. Immunology 1970, 18, 875-881. (17) Imasaka, T.; Kadone, H.; Ogawa, T.; Ishibashi, N. Anal. Chem. 1977, 49, 667-668. (18) Der-Balian, G. P.; Kameda, N.; Rowley, L. Anal. Biochem. 1988, 173, 5963. (19) Gruber, H. J.; Hahn, C. D.; Kada, G.; Riener, C. K.; Harms, G. S.; Ahrer, W.; Dax, T. G.; Knaus, H.-G. Bioconjugate Chem. 2000, 11, 686-704.

Analytical Chemistry, Vol. 79, No. 2, January 15, 2007 647

that have similar absorption and emission maximums with the former ones but are less prone to self-quenching have been reported1,20,21 and introduced in the market. Nevertheless, selfquenching has been observed even with these dyes at high labeling ratios.20 An interesting approach reported in the literature in order to overcome self-quenching effects of conventional fluorescent compounds is the use of silver island films deposited on glass substrates.22,23 Following this approach, the fluorescence intensity of surface-bound fluorescent labels was enhanced and selfquenching was effectively reduced in the case of oligonucleotides highly labeled with fluorescein.22 However, concerning fluoresceinlabeled protein molecules directly adsorbed onto the silver surface, self-quenching was only partially removed for a labeling ratio as low as 3.23 In a previous report we presented,24 a simple procedure that increased considerably the fluorescence intensity measured directly onto the solid surface and improved the sensitivity of heterogeneous immunometric assays performed in microtitration wells using a fluorescein-labeled antibody. According to this report, short treatment of the wells after the completion of the immunoreaction with glycerin solution resulted in significant enhancement and stabilization of the fluorescence signal determined directly onto the solid surface. Maximum plateau values were achieved when the emptied wells were left for 45 min at room temperature or incubated for 15 min at 37 °C. The increase in the fluorescence values was ascribed to the viscous and hydrophilic character of glycerin, whereas the additional effect of thermal treatment was due to the increase of glycerin concentration onto the solid surface due to water evaporation. However, since this result was obtained using a commercially available fluorescein-labeled antibody with a fixed labeling ratio, solid conclusions about the possible remedial effect of the glycerin treatment on fluorescence self-quenching could not be extracted. Here, we provide experimental evidence that glycerin/thermal treatment of the solid surface after completion of the immunoreaction could release self-quenching of fluorescein-labeled antibodies. In fact, self-quenching was completely released for antibodies labeled with fluorescein at molar ratios up to 14 whereas partial release was observed at higher ratios. The effectiveness of the proposed procedure to release self-quenching, to enhance fluorescence signal, and to improve assay sensitivity was evaluated through a noncompetitive sandwich immunoassay for rabbit γ-globulins employing an anti-rabbit IgG antibody labeled with fluorescein at molar ratios ranging from 1 to 17.4. In addition, the possible remedial action of the proposed procedure on the fluorescence self-quenching of other fluorescent compounds when used as labels in immunoassays is demonstrated through appropriate model assays employing antibodies labeled with three different dyes of the Alexa Fluor (AF) series. Finally, the (20) Panchuk-Voloshina, N.; Haugland, R. P.; Bishop-Steward, J.; Bhalgat, M. K.; Millard, P. J.; Mao, F.; Leung, W.-Y.; Haugland, R. P. J. Histochem. Cytochem. 1999, 47, 1179-1188. (21) Anderson, G. P.; Nerurkar, N. L. J. Immunol. Method 2002, 271, 17-24. (22) Malicka, J.; Gryczynski, I.; Lakowicz, J. R. Anal. Chem. 2003, 75, 44084414. (23) Lakowicz, J. R.; Maliska, J.; D’Auria, S.; Gryczynski, I. Anal. Biochem. 2003, 320, 13-20. (24) Petrou, P. S.; Georgiou, S.; Christofidis, I.; Kakabakos, S. E. J. Immunol. Methods 2002, 266, 175-179.

648

Analytical Chemistry, Vol. 79, No. 2, January 15, 2007

advantages this procedure could offer to protein chips and to fluorescence microscopy are demonstrated using two differently labeled anti-rabbit IgG antibody preparations (fluorescein/protein ratio of about 5 and 14, respectively) spotted onto silicon dies coated with rabbit IgG. EXPERIMENTAL SECTION Materials. Affinity purified sheep-anti rabbit IgG antibody was a product of N.C.S.R. “Demokritos”. Rabbit IgG, fluorescein isothiocyanate (FITC, isomer I), bovine serum albumin (BSA, Cohn fraction V, RIA grade), horseradish peroxidase (HRP)labeled goat anti-rabbit IgG antibody, 2,2′-azinobis(3-ethylbenzthiazolinesulfonic acid) diammonium salt (ABTS), and 3-aminopropyltriethoxysilane were purchased from Sigma Chemical Co. (St. Louis, MO). Goat anti-rabbit IgG AF488 (F/P ) 6.5), rabbit anti-fluorescein AF594 (F/P ) 5.6), and goat anti-mouse IgG AF350 (F/P ) 7.9) were from Molecular Probes, Inc. (Eugene, OR). Mouse IgG was from OEM Concepts (Toms River, NJ), and goat anti-mouse IgG was from Chemicon International Inc. (Temecula, CA). All other chemicals were from Merck (Darmstadt, Germany), except as otherwise indicated. White opaque and transparent microtitration plates were from Nunc A/S (Roskilde, Danmark). FITC Labeling of Sheep Anti-Rabbit IgG Antibody at Different Molar Ratios. The affinity-purified anti-rabbit IgG antibody was labeled with fluorescein according to a previously published protocol.25 The antibody solution (1 mg/mL) was first extensively dialyzed against 0.05 M carbonate/bicarbonate buffer, pH 9.2. Then, the dialysis tube was transferred to a beaker containing a 0.15 g/L FITC solution in 0.05 M carbonate/ bicarbonate buffer, pH 9.2 (100 mL of FITC solution/mg of antibody). The reaction was carried out at 4 °C under continuous stirring, and the FITC solution was replaced every 24 h by a freshly prepared one. In order to obtain antibody labeled with fluorescein at different molar ratios, volumes of 1 mL were withdrawn from the dialysis tube at several time internals and the nonreacted FITC molecules were removed from the antibody solution through filtration by a PD 10 column (Pharmacia LKB, Uppsala, Sweden) eluted with 0.1 M PBS buffer, pH 8.0. The protein content of each labeled antibody preparation was determined by the Bradford method, whereas the amount of bound fluorescein was calculated by measuring the absorbance of each labeled antibody at 495 nm in 0.1 M PBS buffer pH 8.0 and using an extinction coefficient value of 68 000 cm-1 M-1.24 After determination of the protein content as well as of the molar ratio of fluorescein per antibody molecule (F/P) for the different preparations, BSA and NaN3 were added to each labeled antibody solution at a final concentration of 10 and 0.5 g/L, respectively. The labeled antibody solutions were stored at 4 °C. Determination of Fluorescein-Labeled Antibodies’ Binding Capacity. The binding capacity of fluorescein-labeled antibody preparations was determined with respect to that of unlabeled antibody through ELISA in wells coated with rabbit IgG. According (25) Bluestein, B. I.; Craig, M.; Stundtner, L.; Urcicuoli, C.; Walczak, I.; Luderer, A. Evanescent wave immunosensors for clinical diagnostics. In Biosensors with fiber optics; Wise, D. L., Wingard, L. B., Eds.; Humana Press: Clifton, NJ, 1991. (26) Singh, A. K.; Kilpatrick, P. K.; Carbonell, R. G. Biotechnol. Prog. 1996, 12, 272-280.

to the protocol, unlabeled or fluorescein-labeled sheep anti-rabbit IgG antibody competed with an HRP-labeled goat anti-rabbit IgG antibody for binding onto immobilized rabbit IgG. More specifically, transparent microtitration wells were coated with 100 µL of a rabbit IgG solution (0.5 mg/L in 0.05 M carbonate/bicarbonate buffer, pH 9.2) overnight at room temperature. The wells were washed twice with 300 µL of a 0.01 M Tris-HCl buffer, pH 8.25, 9 g/L NaCl (washing solution A) and blocked with 300 µL of a 10 g/L BSA solution in 0.1 M NaHCO3, pH 8.5 (blocking solution), for 2 h at room temperature. After washing as previously, 50 µL of unlabeled or fluorescein-labeled sheep anti-rabbit IgG antibody solution at concentrations ranging from 0 to 20 mg/L and 50 µL of HRP-labeled goat anti-rabbit IgG solution diluted in assay buffer (0.1 M Tris-HCl buffer, pH 8.25, containing 10 g/L BSA) were added subsequently in each well. The wells were incubated for 2 h at room temperature under shaking and then washed four times with 300 µL of washing solution A containing 0.5 mL/L Tween 20 (washing solution B). After that, 100 µL of HRP substrate solution (0.03% H2O2, 1.9 mM ABTS in 0.1 M citrate-phosphate buffer, pH 4.5) was added per well. Following incubation for 30 min at room temperature while shaking in the dark, the absorbance of the wells at 405 nm was measured using the Multiscan RC microtitration plate reader (Labsystems). Calibration curves were obtained by plotting the absorbance at 405 nm against the concentration of unlabeled or fluorescein-labeled sheep anti-rabbit IgG antibody. Based on these curves (Figure S-1, Supporting Information), the concentration of each labeled antibody preparation that should be used to provide the same immunoreactivity with the unlabeled antibody was determined. The calculations were based in all cases on the mean value obtained using at least three different concentrations of the labeled antibodies that provided 60-40% inhibition. Noncompetitive Immunofluorimetric Assay for Rabbit IgG. The effect of the glycerin treatment onto the solid-phase immobilized antibodies was studied through a noncompetitive immunoassay for rabbit IgG. White opaque microtitration wells were coated with 100 µL of a 5 mg/L sheep anti-rabbit IgG solution in 0.05 M carbonate/bicarbonate buffer, pH 9.2 (coating buffer), overnight at room temperature. After washing the wells twice with 300 µL of washing solution A, 300 µL of blocking solution was added per well. The wells were then incubated for 2 h at room temperature and washed as previously noted. In each well, was then added 100 µL of a rabbit IgG solution (0-250 µg/L) in assay buffer, and the wells were incubated for 2 h at room temperature under shaking. After washing the wells four times with 300 µL of washing solution B, 100 µL of a fluorescein-labeled sheep antirabbit IgG solution in assay buffer was added per well. The wells were incubated for 2 h at room temperature under shaking and then washed as previously noted. In each well, 100 µL of a 30% (v/v) glycerin solution in 0.05 M carbonate/bicarbonate buffer, pH 9.2 was added, and the wells were incubated for 3 min at room temperature. The glycerin solution was then discarded, and the emptied wells were incubated for 15 min at 37 °C. The fluorescence bound to the bottom of the wells was measured using the Ascent Fluoroscan microtitration plate reader (Labsystems) with an excitation filter at 485 nm and an emission filter at 538 nm. Assays with Alexa Fluor-Labeled Antibodies. Goat antirabbit IgG AF488-labeled antibody was tested through a noncom-

petitive immunoassay for rabbit IgG using the same experimental conditions employed for the homemade fluorescein-labeled sheep anti-rabbit IgG antibodies. Goat anti-mouse IgG AF350-labeled antibody was tested though a noncompetitive immunoassay for mouse IgG. The assay was performed using the same conditions as the noncompetitive assay for rabbit IgG, whereas the fluorescence measurements were performed using an excitation filter at 340 nm and an emission filter at 405 nm. Rabbit anti-fluorescein AF594 antibody was tested in wells coated with 100 µL of fluorescein-labeled rabbit IgG (10 mg/L solution in coating buffer) and then washed and blocked as described previously for the noncompetitive assay for rabbit IgG. Then, in each well was added 100 µL of a 5 mg/L AF 594-labeled rabbit anti-fluorescein antibody solution in assay buffer, and they were incubated for 2 h at room temperature under shaking. Subsequent steps, e.g., washing, treatment with glycerin, etc., were as for the noncompetitive immunoassay for rabbit IgG. The fluorescence bound to the bottom of the wells was measured using an excitation filter at 584 nm and an emission filter at 612 nm. The fluorescein-labeled rabbit IgG was prepared as described previously for the sheep anti-rabbit IgG antibody through a 24-h incubation with the FITC solution. The F/P molar ratio of this preparation was 4.7. Detection of Spots on Silicon Dies through Fluorescence Microscopy. A silicon nitride wafer was activated with 3-aminopropyltriethoxysilane according to a published protocol27 and was incubated with a 10 µg/mL rabbit γ-globulin solution in coating buffer, for 1 h at room temperature. The surface was then washed with distilled water, dried under a N2 stream, and immersed in blocking solution for 1 h at room temperature. After that, the wafer was washed with distilled water and dried as previously. Drops of the differently labeled antibodies (2 µL) were then deposited on the surface, in proximity to each other in order to permit microscopic observation in the same optical field and allowed to react with the immobilized antigen for 1 h at room temperature in a humidity chamber, in order to avoid drying of the labeled antibody solutions. The wafer was washed extensively first with washing solution B and then with distilled water and dried. Fluorescence images of the spots were taken using the Axioskop 2 plus epifluorescence microscope (Carl Zeiss) facilitated with a Sony Cyber-shot digital camera after coverage of the surface first with a 0.05 M carbonate/bicarbonate buffer, pH 9.2 and then with a 90% glycerin solution in the same buffer. Coverslips were used in both cases. RESULTS AND DISCUSSION FITC Labeling of Antibody. The labeling of antibody with FITC was performed under mild conditions following a published protocol.25 According to this, the antibody solution is placed in a dialysis bag and dialyzed against a FITC solution for 48 h at 4 °C. Following this protocol, we found that ∼8 fluorescein molecules could be introduced per antibody molecule. For the purposes of our study, it was necessary to increase the labeling ratio. This was achieved by daily replacement of the FITC solution since in preliminary experiments a reduction of the reaction rate was observed when the incubation with the same FITC solution (27) Misiakos, K.; Kakabakos, S. E.; Petrou, P. S.; Ruf, H. H. Anal. Chem. 2004, 76, 1366-1373.

Analytical Chemistry, Vol. 79, No. 2, January 15, 2007

649

Figure 1. Effect of the reaction time on the degree of labeling of sheep anti-rabbit IgG antibody with fluorescein.

exceeded 24 h, possibly due to hydrolysis of the reactive isothiocyanate groups. As is shown in Figure 1, for a 24-h reaction time, a molar ratio of 6.9 was achieved. Each additional 24-h incubation with fresh FITC solution resulted in insertion of 2-3 fluorescein molecules/antibody. Thus, a molar ratio of ∼17 fluoresceins/antibody molecule was achieved after 6.5 days of reaction. By repeating this labeling procedure for different reaction times, 24 different fluoresceinated antibody preparations with F/P ratios ranging from 1 to 17.4 were obtained. Binding Capacity of the FITC-Labeled Antibody. Extensive labeling of antibodies with FITC could result in considerable loss of their binding capacity,24 possibly due to coupling of fluorescein molecules on amine groups28,29 in the vicinity of the antibody binding sites. The fluorescently labeled antibodies in our study were immobilized through immunoreaction. Thus, comparison of the fluorescence signal provided by the antibodies labeled at different molar ratios just on a protein concentration basis could lead to false conclusions. For this reason, the relative binding capacity of the differently labeled antibody preparations was determined with respect to that of the unlabeled molecule through a competitive enzyme immunoassay for anti-rabbit IgG antibody. It was found that introduction of up to 3 fluorescein molecules/ antibody did not practically affect its binding capacity compared with that of the unlabeled antibody. Incorporation of more than 3 fluoresceins/antibody molecule resulted in gradual decrease of antibody binding capacity, reaching a 75% loss of its original value for the antibody labeled at 17.4 F/P ratio (Table S-1, Supporting Information). Based on these results, the concentrations of the differently labeled antibodies that should be used in a noncompetitive immunoassay for rabbit IgG in order to ensure the same immunoreactivity between the different preparations were calculated. Relative binding capacity values of the labeled antibody preparations (expressed as percent binding capacity of the unlabeled antibody) as well as the respective antibody concentrations used in the study are provided in the Supporting Information, Table S1. Effect of Glycerin on the Fluorescence Self-Quenching of Immobilized Labeled Antibodies. The effectiveness of the glycerin/thermal treatment to suppress self-quenching between (28) Banks, P. R.; Paquette, D. M. Bioconjugate Chem. 1995, 6, 447-458. (29) Schnaible, V.; Przybylski, M. Bioconjugate Chem. 1999, 10, 861-866.

650 Analytical Chemistry, Vol. 79, No. 2, January 15, 2007

Figure 2. Fluorescence signal values obtained for a 50 µg/L rabbit IgG solution with the differently labeled sheep anti-rabbit IgG preparations in a noncompetitive immunofluorometric assay for rabbit IgG after the completion of the assay (squares), after treatment of the wells with a 30% (v/v) glycerin solution (up triangles), and after glycerin/thermal treatment (down triangles). Each point corresponds to the mean value of three replicates; the error bars represent (SD.

fluorescein labels attached to the same molecule was evaluated using the differently labeled goat-anti rabbit IgG preparations as detection antibodies in a noncompetitive sandwich immunoassay for rabbit IgG. After completion of the immunoreaction, the wells were incubated with a 30% (v/v) glycerin solution, which has been reported to provide optimum fluorescence enhancement and stabilization, as well as assay repeatability.24 The glycerin solution was discarded, and the wells were incubated for 15 min at 37 °C. In Figure 2, for simplicity reasons, the fluorescence values obtained without glycerin treatment, after treatment with glycerin solution, and after subsequent incubation of the glycerin-treated emptied wells at 37 °C were presented for a single rabbit IgG concentration (50 µg/L). As shown, in the untreated wells, the fluorescence values were almost linearly increased up to a molar ratio of ∼5, whereas a 45% increase in fluorescence signal was achieved using antibodies labeled at molar ratios up to 14.0, where plateau signal values were obtained. A similar fluorescence signal pattern was observed after glycerin treatment, however; in this case, the fluorescence values obtained using antibodies labeled at molar ratios higher than 5 were significantly increased. Thermal treatment of the glycerin-treated wells resulted in additional increase of the fluorescence values. In fact, as shown in Figure 2, the fluorescence values obtained increased linearly with the molar ratio up to 14.0 fluoresceins/antibody molecule. Taken overall, the increase in the fluorescence signal ranged from 9.2% for the antibody with the molar ratio of 1.0 to 117% for the antibody with molar ratio 17.4. In order to provide more solid experimental evidence and elucidate whether the increase in the fluorescence values was due to dequenching, 24 different labeled antibody preparations, obtained by three independent labeling procedures, were tested through the noncompetitive immunoassay for rabbit IgG. The fluorescence values obtained with these preparations before and after glycerin/thermal treatment were expressed as signal per fluorescein molecule and normalized toward the fluorescence value obtained using the antibody preparation with molar ratio

Figure 3. Normalized fluorescence signal values per fluorescein molecule for the differently labeled sheep anti-rabbit IgG preparations. (A) When they have been used as detection antibody in a noncompetitive immunofluorometric assay for rabbit IgG without glycerin treatment (black squares), after glycerin/thermal treatment (white squares). (B) When used in solution, (black squares) carbonate/ bicarbonate buffer, pH 9.2; (white squares) 90% (v/v) glycerin solution in the same buffer. Each point corresponds to the mean value of three replicates.

1.0, according to the following equation:

normalized signal ) (signal of Ab labeled at molar ratio x/molar ratio x) signal of Ab labeled at molar ratio 1 Assuming that, at molar ratio 1.0, there is no quenching effect, the signal drop per fluorescein that would be observed at higher labeling ratios can be directly correlated to the quenching between neighboring fluorescein molecules. As shown in Figure 3A for the solid-phase immobilized antibodies, before glycerin treatment decreased fluorescence, values were obtained for molar ratios higher than 3.7. After glycerin/thermal treatment, however, normalized values of 1 were obtained for molar ratios equal to or higher than 14.0, indicating that fluorescence self-quenching was effectively suppressed. The increase in fluorescence values as well as self-quenching suppression obtained by the glycerin/thermal treatment could be ascribed to the increase in the refractive index of the surrounding medium.1 The Foerster radius, e.g., the distance at which energy transfer is 50% efficient, is proportional to the fourth power of the

refractive index of the medium (n). Glycerin has higher refractive index (1.4748 at 20 °C) than water (1.3330 at 20 °C). Thus, the distance for efficient energy transfer between the acceptor and the donor molecule could be significantly reduced in the presence of glycerin, and as a consequence, the quenching effect will be less intense for a given number of labels per protein molecule. To elucidate this issue, the effect of glycerin on the fluorescence signals provided by the differently labeled antibodies in solution was also determined. For treatment of the immobilized antibodies, a 30% (v/v) glycerin solution was employed. Since, however, the final glycerin concentration on the solid after decanting the solution and thermal treatment is higher than 30% due to water evaporation, several glycerin concentrations ranging from 10 to 90% (v/v) were tested in solution. The concentrations of antibody employed were up to 1.7 × 10-9 M or lower so in all cases the total fluorescein concentration would be lower than the concentration (1.8 × 10-6 M) at which self-quenching has been observed in solutions of free fluorescein.17 It was found that, by increasing the glycerin content in the antibody solution from 10 to 90% (v/ v), the fluorescence values were constantly increased (Figure S-2, Supporting Information). Therefore, a 90% (v/v) glycerin solution was used for the final evaluation. As presented in Figure 3B, both in the absence and in the presence of 90% (v/v) glycerin, a drop in signal per fluorescein molecule for F/P molar ratios higher than 5 was observed. Though the drop was less pronounced in the presence of glycerin, the results obtained with the solid-phase immobilized antibodies cannot be attributed only to the effect of the increased refractive index of glycerin. It seems that additional mechanisms related to the nature of the solid and the distance of the labeled molecules from the surface should be considered. It is known that the fluorescence signal is strongly affected by the dye microenviroment and, especially, the local charge distribution as it is defined by other protein molecules present in the dye neighborhood.26 This microenvironment is quite different for the labeled molecules in solution as compared to the immobilized ones. Glycerin could act as a shield for the fluorescent molecules against a quenching effect that could arise due to interactions of the immobilized fluorescent molecules with neighboring protein molecules or with the solid itself. Analytical Characteristics of the Assay Using Highly Labeled Antibodies. The effect of glycerin/thermal treatment on the assay performance, when antibodies with high F/P ratios were used as labels, was determined using as labels two different antibody preparations, one of moderate molar ratio (5.6) and one with high molar ratio (14.0), to obtain rabbit IgG calibration curves. As is shown in Figure 4, before glycerin/thermal treatment, the calibration curves obtained with the two labeled antibody preparations were almost identical. After glycerin/thermal treatment, the fluorescence signal was increased considerably over the whole range of the calibrators concentration, especially for the antibody with molar ratio of 14.0. The intra-assay CVs did not alter significantly before and after treatment and ranged for the antibody with a molar ratio of 5.6 from 0.5 to 7.0% and for the antibody with a molar ratio of 14.0 from 0.8 to 7.6%. The detection limit was determined in all cases as the concentration of rabbit IgG corresponding to +3 SD of the mean blank signal derived from 16 determinations of the zero calibrator solution (not containing rabbit IgG). The detection limits determined before Analytical Chemistry, Vol. 79, No. 2, January 15, 2007

651

Figure 4. Typical calibration curves of noncompetitive sandwich immunoassay for rabbit IgG obtained using sheep anti-rabbit IgG antibody labeled with fluorescein at molar ratios 5.6 (squares) and 14.0 (triangles), respectively, before (open symbols) and after glycerin/ thermal treatment (closed symbols). Each point corresponds to the mean value of three replicates; the error bars represent (SD.

glycerin treatment were 0.5 and 0.6 µg/L for the antibodies with molar ratios of 5.6 and 14.0, respectively, whereas after glycerin/ thermal treatment, the detection limits determined were 0.3 and 0.15 µg/L, respectively. In other words, there was a 4-fold improvement in detection limit compared with those obtained without treatment when antibodies with high labeling ratios were used. It must be noted that, at least in our application, the use of more heavily labeled antibodies and the glycerin/thermal treatment did not cause an increase of the nonspecific binding signal (less than 1 RFU in all cases). Concerning the stability of the heavily labeled antibodies, it was found that they provided the same fluorescence values before and after glycerin/thermal treatment for at least 1 year after their preparation when stored at 4 °C in solution, indicating that there was no detachment of the labels from the antibody molecule or quenching during storage. Effect on Other Fluorescent Labels. A first attempt to investigate the possible effect of glycerin/thermal treatment on other fluorescent labels was performed by employing three different compounds from the AF series. Alexa Fluor dyes have been developed as alternatives to conventional fluorescent dyes and present decreased quenching, bleaching, and dependence of fluorescence by environmental factors.20 Three different commercially available antibodies, each one labeled with a different fluorescent compound, were tested after their binding onto respective immobilized antigens. The antibodies were supplied with a F/P ratio in the range that provided the maximum fluorescence values for each particular dye.20 However, as shown in Figure 5, in all cases, the glycerin/thermal treatment resulted in accumulative increase in the fluorescence intensity. More specifically, the percent increase gained after the glycerin/thermal treatment was approximately 10, 20, and 40 for the antibodies labeled with AF350, AF488, and AF594, respectively. On the other hand, treatment of the wells after completion of the immunoreaction only with the glycerin dilution buffer (control) did not affect significantly the fluorescence signals whereas the subsequent thermal treatment of these wells resulted in considerable loss of 652 Analytical Chemistry, Vol. 79, No. 2, January 15, 2007

Figure 5. Fluorescence signal values obtained with (1) goat antirabbit IgG antibody labeled with AF488, (2) goat anti-mouse IgG antibody labeled with AF350, and (3) rabbit anti-fluorescein antibody labeled with AF594, used as labels in the respective immunoassays. Black bars: signals received after the completion of the immunoreaction (washed and emptied wells). Diagonally crossed bars: signal after treatment with 30% (v/v) glycerin solution. Gray bars: signal after glycerin/thermal treatment. Hatched bars: signal of control wells treated with carbonate/bicarbonate buffer, pH 9.2. White bars: signal of control wells treated with carbonate/bicarbonate buffer, pH 9.2 and thermally treated at 37 °C. Each point corresponds to the mean value of three replicates; the error bars represent (SD.

the fluorescence signals of all three fluors. Thus, even in the case of these dyes, the glycerin/thermal treatment can result in increase and significant stabilization of the fluorescence signal. The effect could be even more impressive if the specific antibodies were labeled at molar ratios where quenching between neighbor labels occurs for this kind of dyes.20 Therefore, further investigation is required to this direction. Detection of Protein Spots. The potential application of glycerin treatment to protein chips in order to enhance the signal and to improve the detection sensitivity was also investigated. The study was performed by spotting manually solutions of two differently labeled antibodies, with molar ratios of 5.6 and 14.0, on silicon dies coated with rabbit IgG. After immunoreaction, microscope fluorescence images of the spots were first obtained by covering the surface with carbonate/bicarbonate buffer, pH 9.2 (Figure 6A). Then, the chips were washed, and images were received after covering the surface with a 90% (v/v) glycerin solution in the same buffer (Figure 6B). This solution was selected in order to increase the glycerin concentration in the surface without thermal treatment and because it is one of the most common mounting mediums in fluorescence microscopy for decades.30-32 As it concluded from Figures 6A and B, the presence of 90% (v/v) glycerin resulted in significant increase of the fluorescence signal obtained by both antibodies. Quantitative determination of fluorescence intensity of the two spots using the Image-Pro Plus software (Figure 6C) revealed that in the presence of glycerin the fluorescence signals provided by the antibodies with molar ratios of 5.6 and 14.0 were increased by approximately 85 and 170%, respectively. The ratio of the net signal provided in (30) Lennette, D. A. Am. J. Clin. Pathol. 1978, 69, 647-648. (31) Pratt, J. L.; Michael, A. F. J. Histochem. Cytochem. 1983, 31, 840-842. (32) Boeck, G.; Hilchenbach, M.; Schauenstein, K.; Wick, G. J. Histochem. Cytochem. 1985, 33, 699-705.

Vectashield, Vector Laboratories Inc.) that contain glycerin, very similar results in terms of signal increase were achieved. Our results indicate that, using highly labeled antibodies and a common mounting medium, it is possible to increase considerably both the fluorescence signal and the detection sensitivity in immunofluorescence microscopy.

Figure 6. Microscope fluorescence images of spots of fluoresceinlabeled sheep anti-rabbit IgG antibodies onto rabbit IgG-coated silicon wafer in carbonate/bicarbonate buffer, pH 9.2 (A) and in a 90% (v/v) glycerin solution in the same buffer (B). The left spots correspond to the antibody with molar ratio of 5.6 and the right to the antibody with a molar ratio of 14.0. Some satellite labeled antibody microspots could be also observed between the two main spots, especially in image B received after glycerin solution application due to signal enhancement. (C) Fluorescence intensity of the spots depicted in images A and B in (1) carbonate/bicarbonate buffer, pH 9.2, and (2) in a 90% glycerin solution in the same buffer. Black columns: background signal, White columns: net signal obtained with the antibody labeled at F/P 5.6, Hatched columns: net signal obtained with the antibody labeled at F/P 14.0. Background signal was determined from an area of the chip free of labeled antibody satellite spots.

the presence of glycerin by the antibody with a molar ratio 14.0 to that obtained by the antibody with A molar ratio 5.6, was very close to 2.5, which is the theoretically expected value if selfquenching is completely suppressed. It is noteworthy that the background signal value was the same in the presence of either carbonate/bicarbonate buffer or 90% glycerin solution. Therefore, the signal-to-background ratio was increased accordingly for the two antibodies, reaching a value of 4.5 from 2.6 and 11.0 from 4.3 in the presence of glycerin compared to carbonate/bicarbonate buffer for the antibodies with molar ratios of 5.6 and 14.0, respectively. These results indicate that the use of a 90% glycerin solution for the detection of protein spots through fluorescence microscopy could improve considerably the detection sensitivity. It is interesting to mention that although 90% glycerin is one of the most commonly used mounting mediums in immunofluorescence microscopy,30-32 so far there is no report concerning the enhancement effect of this medium in the fluorescence signal of fluorescein-labeled antibodies. Moreover, we found that using commercially available mounting medium formulations (e.g.,

CONCLUSIONS The glycerin/thermal treatment of fluorescein-labeled antibodies after the end of the immunoreaction increased considerably the fluorescence signal determined directly onto the solid phase and suppressed completely fluorescence quenching using antibodies labeled at molar ratios as high as 14.0. Concerning the analytical characteristics of the heterogeneous immunoassays, the proposed treatment resulted in ∼4-fold improvement in assay sensitivity using antibodies labeled at high molar ratios. Following the proposed method, interesting results were also obtained using fluorescent compounds other than fluorescein, indicating its potential application as a universal approach to increase fluorescence signal and suppress fluorescence self-quenching. Considerable signal and detection sensitivity improvement was also achieved concerning detection of labeled antibody spots immunoadsorbed onto silicon chips in the presence of 90% glycerin. In this case too, fluorescence self-quenching was effectively suppressed for antibodies labeled with up to 14.0 fluoresceins. It is true that by using a 90% glycerin solution in immunoassays performed in microtitration wells, the results obtained were very similar to those achieved using a 30% glycerin solution followed by thermal treatment at 37 °C; however, the high glycerin concentration affected negatively the repeatability of the measurements, possibly due to uneven well-to-well removal of the glycerin solution due to its high viscosity. Therefore, we do not suggest this procedure for assays in microtitration wells though it is very appropriate for fluorescence microscopy and detection of spots on flat surfaces. Given that the currently available fluorescence microarray scanners detect fluorescence directly onto the solid surface, the proposed method can find immediate application to the fast expanding field of protein and DNA chips, permitting signal increase and improving detection sensitivity. ACKNOWLEDGMENT This work was supported in part by the European Commission through the BRITE/EURAM III project BRPR CT 97 0393 (BOEMIS) and the GROW-2001 project G5RD CT 2002 00744 (MICROPROTEIN). SUPPORTING INFORMATION AVAILABLE Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org. Received for review August 10, 2006. Accepted October 23, 2006. AC061492M

Analytical Chemistry, Vol. 79, No. 2, January 15, 2007

653