Long-Wavelength Fluorescence Polarization Immunoassay

Sep 5, 2007 - The versatility of the fluorescence polarization immunoas- say (FPIA) is increased by using two long-wavelength labels, Nile Blue and a ...
0 downloads 0 Views 177KB Size
Download PDF
Anal. Chem. 2007, 79, 7424-7430

Long-Wavelength Fluorescence Polarization Immunoassay: Determination of Amikacin on Solid Surface and Gliadins in Solution Marı´a Lourdes Sa´nchez-Martı´nez, Marı´a Paz Aguilar-Caballos, and Agustina Go´mez-Hens*

Department of Analytical Chemistry, University of Co´ rdoba, Campus of Rabanales, Marie-Curie Annex building, 14071-Co´ rdoba, Spain

Fluorescence polarization immunoassay (FPIA) is a homogeneous technique that has a wide acceptance for the determination of low molecular weight antigens such as therapeutic and abuse drugs, as shown by the instrumentation and the numerous

commercial kits available. However, two limitations of this technique are the relatively high detection limits obtained and the very scarce availability of methods for the determination of macromolecules. These drawbacks, which restrict the applicability and versatility of FPIA, can be ascribed to the fluorescent features of fluorescein, which is the main label used in conventional FPIA. Although this fluorophore has a high quantum yield, the detection limits obtained in FPIA depend on the overlapping of the fluorescent emission of fluorescein with the background signal from the sample matrix and on the narrow slits required to avoid the light scattering caused by its narrow Stokes shift. Also, the use of fluorescein as label in FPIA limits the versatility of this technique because its short lifetime (4 ns) is appropriate for the determination of haptens, but it is excessively short to obtain a suitable difference between the fluorescence polarization value of the free and bound tracer in the case of macromolecules.1 The detection limits in FPIA can be improved using fluorescent labels with excitation and emission wavelengths longer than those of the sample background, and with a wide Stokes shift to avoid light scattering, whereas the determination of macromolecules by FPIA requires the use of labels with longer lifetimes than that of fluorescein. Another aspect that has not been assayed in FPIA, although it can attain a notable reduction in the reagent consumption, is the use of dry reagent chemical technology, in which a drop of few microliters of immunoreagents is spotted onto a planar support, dried, and stored until use. At the moment of the analysis, only the addition of buffered standards or samples is required. However, dry reagent technology in FPIA requires the use of a long-wavelength fluorophore as label, which shows a Stokes shift wider than that of fluorescein. This fluorophore is unsuitable because of the high light scattering obtained. The aim of the research presented here has been the improvement of FPIA features, mainly the detection limit and cost, using a solid surface instead of a solution, and the application of this technique to the determination of macromolecules in real samples. Two long-wavelength fluorophores, Nile Blue (NB) and a ruthenium(II) chelate, have been chosen as alternative labels to fluorescein, to synthesize for the first time the corresponding tracers for amikacin and gliadins, respectively, which have been used as analyte models.

* To whom correspondence should be addressed. E-mail: [email protected]. Fax: +34-957218644.

(1) Terpetschnig, E.; Szmacinski, H.; Lakowicz, J. R. Anal. Biochem. 1995, 227, 140-147.

The versatility of the fluorescence polarization immunoassay (FPIA) is increased by using two long-wavelength labels, Nile Blue and a ruthenium(II) chelate. The first label has been used to study the potential of FPIA on a solid surface using dry reagent technology. The aminoglycoside antibiotic amikacin has been used as an analyte model, and the method has been applied to the analysis of serum samples. The second label has been used to show the practical application of FPIA to the determination of macromolecules, using gliadins as an analyte model, which have been determined in gluten-free food. Very low amounts of anti-amikacin antibodies and amikacin-Nile Blue tracer were immobilized onto nitrocellulose membranes, for the development of the amikacin method, and the consumption of reagents is lower than in conventional FPIA. Only the addition of the standard or sample extract at an adequate pH is required at the analysis time. The analyte displaces the tracer from the tracer-antibody immunocomplex, obtaining a decrease in the fluorescence polarization proportional to the analyte concentration. The gliadin-Ru(II) chelate tracer shows a relatively long lifetime, which allows the observation of differences in fluorescence polarization values between the tracer-antibody complex and the tracer alone. The dynamic range of the calibration graphs for both analytes is 0.5-10 µg mL-1 and the detection limits are 0.1 and 0.09 µg mL-1 for amikacin and gliadins, respectively. The study of the precision gave values of relative standard deviations lower than 5 and 1.5% for the amikacin and gliadin methods, respectively. Amikacin was determined in human serum samples using a previous deproteinization step with acetonitrile, obtaining recovery values in the range 83.4-122.8%. The gliadin method was applied to the analysis of gluten-free food samples by using a previous extraction step. The recovery study gave values between 94.3 and 105.0%.

7424 Analytical Chemistry, Vol. 79, No. 19, October 1, 2007

10.1021/ac070761l CCC: $37.00

© 2007 American Chemical Society Published on Web 09/05/2007

Nitrocellulose membranes spotted with the tracer-antibody mixture were used to prepare the immunoreagent strips by dry reagent technology for amikacin determination. The displacement format,2 in which the tracer is displaced by the analyte from the immunocomplex, gives rise to a decrease in the fluorescence polarization signal proportional to the analyte concentration. Longterm therapies at high doses with amikacin (its therapeutic range is 20-30 µg mL-1) can cause nephrotoxicity and ototoxicity due to its narrow safety range since it is toxic above 35 µg mL-1,3 so it is mandatory to have appropriate analytical methods for its determination. The availability of commercial kits for amikacin determination, which involve the use of fluorescein as label and are based on conventional FPIA in solution, allows the comparison of their analytical features with those obtained in the developed method. Some ruthenium,1,4,5 rhenium,6 and osmium7 chelates have been described as potential labels in FPIA. The best results have been obtained with ruthenium chelates, which have lifetimes of ∼500 ns. These chelates have been proposed as labels in immunoassays for the determination of human serum albumin1,4,5 and myoglobin,4 but their practical usefulness to the analysis of real samples has not been demonstrated yet. The study presented here involves the assessment of a ruthenium(II) chelate as label to develop a FPIA method in solution for the determination of gliadins in gluten-free samples. The relatively long lifetime of the complex makes possible the observance of polarization changes of the tracer when bound to an anti-gliadin antibody. The control of gliadin concentration in food samples is important since they can cause an allergic response in celiac patients. The limits for gluten-free foods have been settled by the Codex Alimentarius Norm,7 at 20 mg of gluten per food kilogram for original glutenfree samples or 200 mg of gluten per food kilogram for samples from which gluten has been removed. This organization recommends the use of an immunochemical method for gliadin determination. A number of ELISA methods have been described for this purpose,8-16 which involves several steps of incubation and (2) Ngo, T. Anal. Lett. 2005, 38, 1057-1069. (3) Peloquin, C. A.; Berning, S. E.; Nitta, A. T.; Simone, P. M.; Goble, M.; Huitt, G. A.; Iseman, M. D.; Cook, J. L.; Curran-Everett, D. Clin. Infect. Dis. 2004, 38, 1538-1544. (4) Durkop, A.; Lehmann, F.; Wolfbeis, O. S. Anal. Bioanal. Chem. 2002, 372, 688-694. (5) Szmacinski, H.; Terpetschnig, E.; Lakowicz, J. R. Biophys. Chem. 1996, 62, 109-120. (6) Guo, X. Q.; Castellano, F. N.; Li, L.; Lakowicz, J. R. Anal. Chem. 1998, 70, 632-637. (7) Terpetschnig, E.; Szmacinski, H.; Lakowicz, J. R. Anal. Biochem. 1996, 240, 54-59. (8) Draft Revised Standards for gluten-free foods. Report of the 25th session of the Codex Committee on Nutrition and Foods for special dietary uses, November 2003. http://www.codexalimentarius.net/. (9) Gabrovska, D.; Rysova, J.; Filova, V.; Plicka, J.; Cuhra, P.; Kubik, M.; Barsova, S. J. AOAC Int. 2006, 89, 154-160. (10) Nicolas, Y.; Denery-Papini, S.; Martinant, J. P.; Popineau, Y. Food Agric. Immunol. 2000, 12, 53-65. (11) Sorell, L.; Lo´pez, J. A.; Valde´s, I.; Alfonso, P.; Camafeita, E.; Acevedo, B.; Chirdo, F.; Gavilondo, J.; Me´ndez, E. FEBS Lett. 1998, 439, 46-50. (12) Seilmeier, W.; Wieser, H. Eur. Food Res. Technol. 2003, 217, 360-364. (13) Official Method 991.19. In Official Methods of Analysis of AOAC International, 17th ed.; Horwitz, W., Ed.; AOAC International: Gaithersburg, MD, 2000; pp 15-16. (14) Rumbo, M.; Chirdo, F. G.; Fossati, C. A. J. Agric. Food Chem. 2001, 49, 5719-5726. (15) Prabhasankar, P.; Sai-Manohar, R. J. Agric. Food Chem. 2002, 50, 74557460.

washing. However, the availability of a homogeneous assay simplifies the determination of these proteins in food samples. EXPERIMENTAL SECTION Apparatus. An SLM Aminco (Urbana, IL) model 8100 photon counting spectrofluorometer, equipped with a 450-W xenon arc lamp and two polarizers (Glan-Thompson calcite prism type), was used. The excitation and emission slits were set to provide 2- and 8-nm band passes, respectively. The conventional cell compartment of the instrument was used to obtain fluorescence polarization measurements in solution but it was replaced by a Spectronic FP-111 variable-angle front surface accessory fixed in the -24° angle position to perform solid surface fluorescence polarization measurements. Reagents. All reagents used were of analytical grade. NB, glutaraldehyde, and amikacin (Sigma-Aldrich, St. Louis MO) were used to synthesize one of the two long-wavelength tracers. The other tracer was synthesized using bis(2,2′-bipyridine)-4′-methyl4-carboxybipyridineruthenium N-succinimidyl ester bis(hexafluorophosphate) and gliadins (Sigma). Anti-amikacin antibodies and the fluorescein-amikacin tracer were supplied by Abbott (Amikacin reagent pack for TDX/TDXFLX, Abbott Laboratories). Antigliadin polyclonal antibodies (7.2 mg mL-1) (Sigma, G-9144) were used and diluted with a phosphate buffer solution (0.02 M, pH 7.0). A stock solution of amikacin (1.7 × 10-2 M) was prepared in phosphate buffer (0.02 M, pH 7.4), and working dilutions of amikacin, amikacin tracer, and anti-amikacin antibodies were prepared using the same buffer solution. Gliadin (Catalog No, G-3375, MW 50 000) standards were prepared by dissolving these proteins (5 × 10-5 M) first in the minimum amount of 75% ethanol, and then, phosphate buffer (0.02 M, pH 7.5) was added to volume. Working standard solutions were prepared daily by diluting adequately this solution with the same phosphate buffer solution. Nitrocellulose membranes (0.45-µm pore size, 20 × 20 cm) were supplied by Sigma. Organic solvents such as chloroform, methanol, and dimethylformamide (DMF) were supplied by Panreac (Panreac Quı´mica S.A., Barcelona, Spain). Hi-Trap size exclusion columns (GE Healthcare) were used to purify the synthesized gliadin-tracer. A carbonate buffer (0.1 M, pH 9.0) was prepared in 25% ethanol to dissolve gliadins for the performance of the synthesis step. Procedures. Synthesis of the Tracers. The amikacin tracer was synthesized by dissolving 4 mg of NB (10 µmol) in 960 µL of DMF and adding 40 µL of glutaraldehyde (10 µmol). The mixture was stirred for 90 min, incubated for 24 h at room temperature, and purified by thin-layer chromatography (TLC), using silica gel 60 F254 plates (Merck), which had been previously washed with methanol. A chloroform/methanol (2:1) mixture was used as mobile phase. A blank without glutaraldehyde was used to find the appropriate band for the intermediate. A blue band with Rf ) 0.64 was chosen, scrapped from the plate, and eluted using 1 mL of methanol. A volume (100 µL) of the purified extract of the NBglutaraldehyde intermediate was added to 6.5 µmol of amikacin, dissolved in 400 µL of phosphate buffer solution (0.02 M, pH 7.4). The mixture was incubated for 72 h at room temperature and purified by TLC using a chloroform/methanol (2:1) mixture as (16) Denery-Papini, S.; Samson, M. F.; Autran, J. C. Food Agric. Immunol. 2000, 12, 67-75.

Analytical Chemistry, Vol. 79, No. 19, October 1, 2007

7425

mobile phase. The tracer band (Rf ) 0.05) was scrapped from the plate and eluted using 1 mL of methanol. The global yield was 2%, which was calculated using 624) 8.12 × 104 M-1 cm-1 for NB in methanol. The synthesis of the gliadin-tracer was carried out by mixing 125 µL of a DMF solution containing 5 mg of the commercial succinimidyl ester of the ruthenium complex with 1 mL of a 5 × 10-5 M gliadin solution prepared by dissolving the proteins in a solution constituted by carbonate buffer (0.1 M, pH 9.0) and ethanol 25%. The reaction mixture was stirred at room temperature in the dark for 2 h, and then, it was let to stand at 4 °C for 4 days. The synthesized tracer was purified by size exclusion using a HiTrap desalting column, which was previously equilibrated with 30 mL of phosphate buffer (0.02 M, pH 7.0). After sample application, 10 mL of the same phosphate buffer solution were added and 400-µL fractions were collected. The tracer was found to be between fractions 4 and 5, which were the first ones colored to appear. In both fractions, the estimated label/protein ratio was 6:1, which was calculated by using 451 ) 1.41 × 104 M-1 cm-1 for the ruthenium complex in phosphate buffer (0.02 M, pH 7.0) and the Coomassie Blue method for the protein concentration. Preparation of the Immunoreagent Strips. Nitrocellulose strips were cut into rectangular pieces of 1.5 × 3.0 cm, and the spot position was adjusted using a ring mark. A mixture (5 µL) of antiamikacin antibodies (7.6 × 10-8 M) and the long-wavelength tracer (9.6 × 10-8 M) in phosphate buffer solution (0.02 M, pH 7.4) was spotted onto the membrane and dried in an oven at 37 °C for 5 min. These membranes were stored at 4 °C, and they were stable for 2 months. Determination of Amikacin in Serum Samples. Each serum sample (500 µL) was treated with 500 µL of acetonitrile. The suspension formed was centrifuged for 10 min at 2500 rpm, and 400 µL of the supernatant was diluted with phosphate buffer (0.02 M, pH 7.4) up to 2 mL. Spiked samples were prepared by adding few microliters of an amikacin solution and waiting for 30 min, being the treatment used the same than that described above. A volume of this solution (5 µL) was spotted onto the immunorreagent strip. The strip was dried at 37 °C for 5 min, let cool to room temperature in a desiccator for 9 min, and fixed to the sample holder. The fluorescence polarization was measured using 626 and 674 nm as excitation and emission wavelengths. The normalized signal (B/B0), in which B is the fluorescence polarization obtained for the analyte and B0 that for the blank signal, was used as the analytical parameter. The calibration graph was obtained in a similar way using aqueous standards with amikacin concentrations in the range 0.5-10 µg mL-1. Each sample or standard was assayed in triplicate. Determination of Gliadins in Gluten-Free Food Samples. Each food sample (5.0 g) was extracted twice with 30 mL of 0.4 M NaCl for 10 min. Then, the suspension was centrifuged for 15 min at 2500 rpm, and the residue was extracted twice with 20 mL of 70% aqueous ethanol for 10 min. Each suspension was centrifuged for 15 min at 2500 rpm, and both extracts were combined and raised up to 50 mL with 70% aqueous ethanol. A volume (1 mL) of these extracts was mixed with gliadin-tracer and antibodies and the solution added to a final volume of 2 mL. Spiked samples to carry out the recovery study were obtained by adding a minimum volume of gliadin solution. These samples were let to stand 7426

Analytical Chemistry, Vol. 79, No. 19, October 1, 2007

overnight to evaporate the solvent and to allow the equilibration of the added analyte with the sample matrix. Then, they were treated as mentioned above. The calibration graph was obtained using mixtures (2 mL) containing gliadin standards (0.5-10 µg mL-1), tracer (2.6 × 10-8 M), and anti-gliadin antibodies (4.5 × 10-8 M), which were prepared in phosphate buffer (0.02 M, pH 7.5). These mixtures were incubated for 7 min. Excitation and emission wavelengths used to obtain the fluorescence polarization values were 451 and 626 nm, respectively. The analytical parameter used was the normalized signal described above. All experiments were carried out at room temperature. Each standard or sample was assayed in triplicate. RESULTS AND DISCUSSION Choice of Membrane and Fluorophor for Amikacin FPIA on a Solid Surface. The compatibility of the solid support with the nature of the reagents is essential to attain their appropriate stability and performance. Both nitrocellulose membranes and chromatographic paper were assayed as supports for performing the proposed immunoassay method for amikacin determination. The use of strips simplifies the assay and allows a low consumption of reagents. Nitrocellulose membranes are often used in lateralflow assays17,18 to immobilize antibodies or other proteins, since these macromolecules are very stable on this support. The optical properties of both materials were evaluated by obtaining their excitation and emission spectra alone and in the presence of two tracers, a commercial FPIA tracer derived from fluorescein and the synthesized long-wavelength tracer involving NB. Figure 1 shows the emission spectra obtained for the nitrocellulose membranes using 490 (Figure 1A) and 626 nm (Figure 1B) as excitation wavelengths, which correspond to the maximum excitation of fluorescein and NB, respectively. The sharp peak that appears in Figure 1A at 520 nm could be attributed to light scattering phenomena, including a Raman effect, caused by the measurements of front-surface fluorescence. This peak overlaps the emission maximum of fluorescein, hindering the performance of fluorescence polarization measurements. However, the use of the NB-based tracer minimizes this shortcoming, as can be seen in Figure 1B. Although a dispersion peak also appears, the emission of the tracer can be measured at a longer wavelength. The use of a chromatographic paper as solid support gave rise to similar results, but the signal of the tracer was slightly lower, ∼80% of that measured using nitrocellulose membranes, so these membranes were chosen to perform the method. The long-wavelength tracer was synthesized by using glutaraldehyde as a linker of amino groups of amikacin and NB, in a way similar to the procedure used for the immobilization of aminecontaining compounds onto amine-functionalized supports.19 Primary amine groups react to aldehyde moieties to give Schiff bases, which could be further reduced. The separation of the glutaraldehyde excess was done by TLC using chloroform/methanol (2: 1) as mobile phase, and the product formed was eluted from the scrapped silica gel plates using methanol. The subsequent linkage (17) Mansfield, M. A. In Drugs of Abuse; Wong, R. C., Tse. H. Y., Eds.; Humana Press: Totowa, NJ. 2005; p 71. (18) Ferna´ndez-Sa´nchez, C.; Gallardo-Soto, A.; Rawson, K.; Nilsson, O.; McNeil, C. J. Electrochem. Commun. 2004, 6, 138-143. (19) Hermanson, G. T.; Krishna Mallia, A.; Smith, P. K. Immobilized Affinity Ligand Techniques; Academic Press: San Diego, CA, 1992; p 413.

Figure 1. Emission spectra obtained, using 490 (A) and 626 nm (B) as excitation wavelengths, for nitrocellulose membranes alone (1) and spotted with amikacin-fluorescein tracer (A.2) and amikacin-NB tracer (B.2).

to amikacin followed a similar mechanism. In both steps, blanks of reaction were carried out in order to identify the bands associated with the NB-glutaraldehyde intermediate and the final tracer. The synthesis gave a global yield of 2%, which was calculated taking into account the molar absorptivity for NB in methanol, as indicated above. Although the yield was relatively low, only one synthesis was needed to perform the experimental work described. Choice and Synthesis of Tracer for Gliadin FPIA in Solution. The potential usefulness of FPIA to gliadin determination in food samples was studied by using the commercial ruthenium(II) chelate bis(2,2′-bipyridine)-4-methyl-4carboxybipyridineruthenium N-succinimidyl ester bis(hexafluorophosphate), which has a long-wavelength emission and a wide Stokes shift, to label gliadins. These features minimize potential interferences from the sample matrix, which normally occur at shorter wavelengths, and the dispersion phenomena from scattered light. This chelate also fulfills the requirement of a relatively long lifetime, which allows the fluorescence polarization measurement of macromolecules.5 The pH conditions used for the tracer synthesis were similar to those described for the labeling of other proteins using succinimidyl esters of similar ruthenium complexes.1,4,5,20 The insolubility of gliadins in carbonate buffer required the use of ethanol to obtain a clear solution. The combination of carbonate buffer with 25% ethanol allowed the preparation of 5 × 10-5 M gliadins to carry out the synthesis step in an adequate medium. A 100-fold molar excess of reagent was used to ensure good labeling of the gliadins. The mixture was stirred for 2 h and let stand for 4 days to complete the labeling reaction. This long time together with the high molar reagent excess was used to ensure a good labeling efficiency because gliadins have few lysine residues available for the labeling reaction.21 Then, the mixture (20) Szmacinski, H.; Castellano, F. N.; Terpetschnig, E.; Dattelbaum, J. D.; Lakowicz, J. R.; Meyer, G. J. Biochim. Biophys. Acta 1998, 1383, 151-159. (21) Harris, J. D.; Taylor, G. A.; Blake, T. K.; Sands, D. C. Euphytica 1994, 76, 97-100.

was purified using size exclusion chromatography with Sephadex G-25 columns. The tracer appeared in the first orange-colored band, ∼1.5-2 mL after the elution buffer was added, indicating the presence of gliadins conjugated to the fluorophore. Optimization of Assay Procedures. Both immunochemical systems were optimized by applying the univariate method and using fluorescence polarization measurements. Each result was the average of three measurements. The analytical parameter used, except for antibody optimization, was the normalized signal, as indicated above. Thus, the lowest B/B0 values obtained correspond to the optimal experimental conditions. Tracer and antibody concentrations are interrelated variables. Tracer concentration chosen has to provide an adequate fluorescent signal for the accurate measurement of fluorescence polarization, but taking into account that high concentrations can limit the sensitivity of the assay. The amikacin-tracer was studied in the range 7.2 × 10-8-1.44 × 10-7 M, finding that 9.6 × 10-8 M gave the best results. Antibody concentration was then optimized by obtaining a dilution curve (Figure 2A) for the antibody at the optimum tracer concentration, with an antibody concentration of 7.6 × 10-8 M finally chosen as the optimum value. The pH of the amikacin system was studied in the range 6.5-8.2, finding that the most sensitive assay was at pH 7.2-7.6 (Figure 3A). A 0.02 M phosphate buffer solution of pH 7.4 was used to adjust this variable. The temperature for performing drying of the membranes after application of the samples was evaluated in the range 30-45 °C. At low temperatures, the membranes can remain wet after the drying time and the reproducibility between membranes can be poor. Also, the probability of collisional deactivation processes with solvent molecules is higher than in the case where the membranes are dry. On the other hand, excessively high temperatures can affect the conformation of the antibody leading to partial denaturation. It was found that at temperatures below 35 °C the sensitivity decreased, while the potential effects to the thermal denaturizing were slight at the conditions assayed. The influence of drying time was studied in the range 5-10 min, Analytical Chemistry, Vol. 79, No. 19, October 1, 2007

7427

Figure 2. Dilution curves for anti-amikacin (A) and anti-gliadin antibodies (B). Experimental conditions for (A): [tracer] ) 9.6 × 10-8 M, [phosphate] ) 0.02 M, pH 7.4, drying temperature 37 °C, drying time 5 min, and cooling time 5 min. Experimental conditions for (B) [tracer] ) 2.6 × 10-8 M, pH 7.0, [phosphate] ) 0.02 M, and incubation time 7 min.

Figure 3. Influence of pH on the amikacin (A) and gliadin (B) systems. Experimental conditions for (A): [amikacin] ) 1 µg mL-1, [antiamikacin] ) 7.6 × 10-8 M, [tracer] ) 9.6 × 10-8 M, [phosphate] ) 0.02 M, drying time 5 min, cooling time 5 min, and drying temperature 37 °C. Experimental conditions for (B): [gliadins] ) 1 µg mL-1, [anti-gliadins] ) 4.5 × 10-8 M, [tracer] ) 2.6 × 10-8 M, [phosphate] ) 0.02 M, incubation time 7 min.

finding that 5 min was enough to obtain reproducible results. The evaluation of cooling time after the drying process was done in the range 5-9 min, and it was found that the results were independent of this variable at times higher than 7 min. Different gliadin-tracer concentrations were assayed from 8.5 × 10-9 to 4.3 × 10-8 M, obtaining an adequate fluorescence intensity using 2.6 × 10-8 M tracer. The antibody concentration chosen as optimum was 4.5 × 10-8 M, which was obtained by assaying an antibody dilution curve at the optimum tracer concentration (Figure 2B). The optimization of the pH was carried out in the range 6.0-9.0, finding that the optimum pH was 7.08.0 (Figure 3B). The pH of the system was adjusted to pH 7.5 using a phosphate buffer. Study of the influence of buffer concentration was carried out in the range 1.3 × 10-3-0.1, and 0.02 M was chosen as the optimum value. 7428 Analytical Chemistry, Vol. 79, No. 19, October 1, 2007

Analytical Features of the Methods. Calibration curves for amikacin and gliadins were obtained under optimum experimental conditions and using the normalized signal B/B0 as the analytical parameter. Experimental data from both systems were adjusted by nonlinear regression to a three-parameter sigmoidal curve with the equation y ) {a/[1 + e-(x-x0)/b]}, in which y is the normalized signal B/B0 and x is amikacin or gliadin concentration expressed in micrograms per milliliter. The values of a, b, and x0 are 0.779 ( 0.006, -5.2 ( 0.1, and 11.35 ( 0.06, respectively, for the amikacin system, and 1.42 ( 0.09, -14.4 ( 0.8, and 10.5 ( 0.9, respectively, for the gliadin system. The dynamic range of the calibration graphs (Figure 4) was 0.5-10 µg mL-1 for both amikacin and gliadin systems. The detection limits (LODs), calculated using standards and according to IUPAC recommenda-

Figure 4. Calibration curves for amikacin (A) and gliadin (B) determination. Experimental conditions for (A): [anti-amikacin] ) 7.6 × 10-8 M, [tracer] ) 9.6 × 10-8 M, [phosphate] ) 0.02 M, pH 7.4, drying temperature 37 °C, drying time 5 min, and cooling time 7 min. Experimental conditions for (B) [anti-gliadins] ) 4.5 × 10-8 M, [tracer] ) 2.6 × 10-8 M, [phosphate] ) 0.02 M, pH 7.5, incubation time 7 min.

tions,22 was 0.1 µg mL-1 for amikacin and 0.09 µg mL-1 for gliadin. The LOD value obtained for amikacin is lower than that reported (0.8 µg mL-1) using a commercial FPIA kit for amikacin determination based on a fluorescein tracer.23 The detection limit obtained for gliadins is higher than those reported by some ELISA methods,11,12 which are in the order of a few nanograms per milliliter. However, the proposed method is sensitive enough to perform the analysis of gluten-free food samples. The precision (percentage of relative standard deviation), calculated at two concentrations of each analyte, 0.5 and 1 µg mL-1 amikacin, and 0.5 and 5 µg mL-1 gliadins, expressed as the percentage of relative standard deviation, was 4.0 and 4.7%, respectively, for amikacin, and 0.7 and 1.4%, respectively, for gliadins. Although the precision values obtained for amikacin are higher than those reported for the commercial kit (1.4-2.1%),23 they can be considered acceptable taking into account the use of front-surface measurements. The selectivity of these FPIA methods basically depends on the antibodies. Anti-amikacin antibodies are quite selective, although the assay presents cross-reactivity toward its precursor kanamycin. Other aminoglycoside antibiotics, such as dibekacin, tobramycin, netilmycin, gentamycin, and vancomycin, do not interfere with amikacin determination when assayed at their upper therapeutic limit. Polyclonal antibodies developed in rabbit using gliadin as the immunogen can react with prolamin fractions of rye, barley, and oats, but they do not react with maize, potato, or rice prolamins. Thus, they can detect prolamins from different cereals that are toxic for patients suffering from celiac disease. Applications. The amikacin method was applied to the analysis of four different serum samples. Acetonitrile was used to remove proteins from the sample since they slightly modified the signal provided by the tracer. It was found that the assay was not affected by 20% of this solvent, but higher percentages gave rise to membrane shrinking. This resulted in low performance of (22) Long, G. L.; Winefordner, J. D. Anal. Chem. 1983, 55, 712A-714A. (23) TDx/TDxFLx Amikacin, Ref. 9508-60/-85, Abbott Laboratories, IL, U.S.A. http://www.abbottdiagnostics.com/.

Table 1. Recoveries for Amikacin Added to Human Serum Samples serum sample 1 2 3 4

a

amikacin concentration (µg mL-1) added founda recovery (%) 20 25 30 20 25 30 20 25 30 20 25 30

20.3 ( 0.9 26 ( 2 33 ( 2 23 ( 2 29 ( 3 36.8 ( 0.6 20 ( 2 27 ( 1 33 ( 1 16.7 ( 0.9 28 ( 2 35 ( 2

101.5 104.0 110.0 115.0 116.0 122.8 100.0 108.0 110.0 83.4 113.3 116.6

Mean ( SD (n ) 3).

fluorescence polarization measurements, since membranes became more fragile and even deformed. The calibration graph was constructed using aqueous standards because there were not any matrix effects in the presence of 10% of serum matrix. Results were calculated by interpolating in the calibration graph. None of the analyzed samples contained amikacin. The recovery study was carried out adding three different amounts of analyte to the serum samples, so that the final concentration was within the therapeutic range, and subtracting the results from similarly prepared unspiked samples. Table 1 lists the recoveries obtained, which were in the range 83.4-122.8%. Several gluten-free food samples, namely, maize flours and maize bread loaf, were analyzed by the developed gliadin method. The low limit defined by the legislation required the use of a relatively high percentage of sample (5% w/v). The potential interferences from sample matrix, such as globulins and albumins, were removed before the extraction with 70% ethanol. Samples were extracted twice with 0.4 M sodium chloride to remove these proteins and other water- or salt-soluble compounds, which could Analytical Chemistry, Vol. 79, No. 19, October 1, 2007

7429

Table 2. Recoveries of Gliadins Added to Gluten-Free Food Samples samplea 1 2 3

added (µg g-1)

foundb (µg g-1)

recovery (%)

10 14 20 10 14 20 10 14 20

9.8 ( 0.2 13.2 ( 0.8 20.1 ( 0.4 10.4 ( 0.5 13.9 ( 0.9 21 ( 2 9.2 ( 0.4 14.2 ( 0.9 20 ( 2

98.0 94.3 100.5 104.0 99.3 105.0 92.0 101.4 100.0

a Name (composition; trade mark, manufacturer): 1, maize flour (maize starch; Maizena, Unilever Food Espan ˜a, S.A.) 2, maize flour (maize and potato starch, powdered skimmed milk, sugars, guar gum, pectin, glucono-∆-lactone, sodium hydrogen carbonate, soy flour, tartaric diacetyl ester glyceride, iron, vitamins (B1, B2, B3, and B6), and powdered yeast; Harisin, Sanavı´ S.A.); 3, maize bread (maize starch, water, sugar, vegetal gum, salt, sodium hydrogencarbonate, yeast, potassium sorbate; Diet Ra´disson, Pagesa). b Mean of three determinations ( SD.

interfere with the assay. The influence of the ethanol percentage was assayed in the system, and it was found that the B/B0 ratio was practically unaffected by a 35% level of this solvent. Fluorescence polarization increased in the presence and in the absence of gliadins at this ethanol concentration, leading to a slight decrease of B/B0. Thus, the calibration curve with standards was performed in the presence of this ethanol percentage. Table 2 summarizes the recoveries obtained, which were calculated by adding three different amounts of gliadin to each sample and ranged from 94.3 to 105.0%.

7430 Analytical Chemistry, Vol. 79, No. 19, October 1, 2007

CONCLUSIONS The results obtained show that the use of long-wavelength tracers increases the versatility of FPIA, allowing its use as a screening technique on solid surface and its application to the determination of macromolecules in real samples. Although the Stokes shift of the NB tracer is not excessively wide, it is enough to minimize light dispersion phenomena from the solid support. The low volume required for the preparation of immunoreagents make the reduction of the reagent consumption possible if compared to conventional FPIA assays carried out in solution. Also, the use of nitrocellulose membranes provides a stable immobilization of the antibody-tracer complex. The practical usefulness of the proposed system has been demonstrated by the analysis of human serum samples. The proposed FPIA method for gliadin determination is a fast and simple homogeneous assay that can be quite amenable to automation using the microplate format. This has been the first attempt to apply FPIA to the determination of proteins in food samples, which has been possible owing to the special luminescent features of the ruthenium(II) chelate used to synthesize the tracer. ACKNOWLEDGMENT Authors gratefully acknowledge the financial support to the Spanish Ministerio de Ciencia y Tecnologı´a (MCyT) (Grant CTQ2006-03263/BQU).

Received for review April 17, 2007. Accepted July 18, 2007. AC070761L