Analyst, November 1996, Vol. I21 (1 565-1 568) 1565 Precipitation Flow Injection lmmunoassay for Human Immunoglobulin G* Andrea Hackera, Gerald Giibitza?* Juan-Manuel Fernandez-Romerob, Maria D. Luque de Castroh and Miguel Valcarcelh A-801 0 Graz, Austria h Depui-tment of Analytic-al Chemistry, Faculty cv Science, University of Cordoba, E- I4004 Coi-doha, Spain Institute of Pharmac.eutica1 Chemistry, Karl-Franzens-Uni-ciersity of Graz, The development of a precipitation flow injection immunoassay is described. This approach is based on the 'immunoprecipitin reaction, whereby the precipitate formed by binding of the sample antigen (HIgG) to fluorescein-labelled antibodies is retained on a filter built in on-line in a flow-injection system. After dissolving the precipitate with sodium hydroxide solution the liberation of fluorescein-labelled antibodies results in a fluorescence signal that is directly proportional to the concentration of HIgG.The instrumental set-up is very simple and fully automated. One assay cycle takes about 10 min. The RSD ranged between 1 and 3%, depending on concentration. The detection limit was about 140 fmol. The recovery from spiked serum samples was between 96 and 102%. Keywords: Human ininiunoglobulin G; immunoprecipitin reaction ; pi.ec*ipitation flow injection immunoassay Introduction Immunoassay methods are becoming increasingly important in biomedical and environmental analysis because of their high selectivity and sensitivity. Classical immunoassay techniques are usually time consuming, handling is difficult and automa- tion requires high instrumental complexity.Flow injection immunoassays (FIIA), which have attracted great interest in recent years,' have advantages over conventional immu- noassay techniques with regard to speed, precision, sensitivity and simplicity of automation. Regarding sample throughput, FIIA techniques cannot yet compete with automated im- munoassay analysers. However, the advantage is the tlexibility if many different types of assays have to be carried out and rapid results from a single assay are required. Precipitation flow injection analysis was introduced recently.4.5 A precipitate formed in situ by a reaction is retained on a filter which is arranged on-line in the FI system. After dissolution of the precipitate, the liberated component causes a signal at the detector.This approach has been successfully applied to the determination of both inorganic and organic corn pound^.^-^ We have adapted this principle to the development of a precipitation flow injection immunoassay making use of the well known fact that precipitation occurs during the immuno binding reaction, which can be accelerated by the addition of polyethylene glycol. The basic principle of the so-called immunoprecipitin reaction has been used in classical immunoassays7 and subsequently in continuous flow and F1 systems with nephelometricg-9 and turbidimetric l o detection. respectively. Since these detection methods are not very sensitive, we aimed to improve the * Presented at the VlIth Inkmational Symposiurn on Luminescence Spectrometry in Biomedical Analysis.Sophia Antipolis, Nice, France. April 17-1 9, 1996. To whom correspondence should be addresscd. sensitivity by developing an immunoassay in combination with sensitive fluorescence detection. A very rapid homogeneous enzymatic immunoassay for human immunoglobulin G (HIgG) with fluorescence detection using horseradish peroxidase- labelled antibodies together with leuco-diacetyldichlorofluor- escein was published by Kelly and Christian. I 1 This paper describes the development of a simple precipita- tion flow injection immunoassay with fluorescence detection for HIgG using fluorescein-labelled antibodies in the form of a non-compe ti tive assay. Experiment a1 Materials Human IgG and anti-human IgG (Fab specific) fluorescein isothiocyanate (FITC) conjugate were purchased from Sigma (Deisenhofen, Germany).Polyethylene glycol (PEG) 6000 was obtained from Fluka (Buchs, Switzerland) and disodium hydrogenphosphate dodecahydrate, sodium chloride and so- dium hydroxide from Merck (Vienna, Austria). All other buffer components were of analytical-reagent grade. Preparation of Human IgG Samples Different concentrations of samples were prepared by diluting a stock standard solution ( 5 mg ml- 1) with dilution buffer (SO mmol l-1 Na2HP04, pH 7.0, containing 0.9% NaCl). The stock standard solution was stored at -18 "C and the dilutions used were freshly prepared. The highest concentration was 68.3 pmol ml- l . Preparation of Labelled Anti-human IgG Stock standard solution (4.3 mg ml- l ) was diluted with dilution buffer to give a 10-fold molar excess of the highest concentra- tion of antigens.Determination of Human IgG in Serum Samples Serum samples were diluted 1 : 10000 with dilution buffer. The IgG content was calculated from the relative peak area using known amounts of authentic IgG as a standard. For recovery studies, serum samples with known IgG contents were spiked with two different amounts of IgG (680 and 340 nmol ml-1). Determination of Detection Limit A solution of 4.6 pmol ml-1 was prepared and treated with a 10-fold molar excess of anti-human IgG to this concentration. Apparatus A schematic diagram of the instrumental set-up is shown in Fig. 1. It consists of a solvent pump [Minipuls 3 (Gilson,1566 Analyst, November 1996, Vol.121 Villiers-le-Bel, France)], a Gilson liquid handler 221 XL, an automated Rheodyne (Cotati, CA, USA) injection valve, a switching valve controlled by the autosampler, a PTFE reaction coil (2 m X 0.76 mm id) immersed in a water-bath (37 "C), a metallic filter ( 5 pm pore size) and a fluorescence detector [Perkin-Elmer (Norwalk, CT, USA) LS-SB]. An Axxiom (Moorpark, CA, USA) 737 system was used for data acquisi- tion. Once a week the filter should be cleaned in an ultrasonic bath to avoid clogging. Assay Procedure Before starting, the system was equilibrated with carrier buffer ( S O mmol 1-1 Na2HP04, pH 7 , containing 6% PEG 6000 and 0.9% NaCl) at 0.5 ml min-'. The wavelengths were set to he, 485 nm (slit IS) and he, 521 nm (slit 20). The first sample was pre-mixed by the autosampler.The sequences of events are given in Table 1 . After measurement, the system should be washed with doubly distilled water. Results and Discussion Although heterogeneous FIIAs3 are very sensitive and precise, the need to prepare an immunoreactor and its limited stability are certain drawbacks. a .i C U b d Fig. 1 Instrumental set-up. a, Autosampler with an automated injection valve; b, pump; c, switching valve, controlled by the autosampler; d, water- bath (37 "C) with the reaction coil; e, metallic filter; f, fluorimeter; g, personal computer for data aquisition; I. carrier buffer; 11, dissolution reagent. Carrier buffer c 0.5 mi min-' + 0 0 Dissolution buffer d t = 7 Filter arranged on-line Premixing 0 0 Homogeneous FIIAs, although less sensitive, are easier to handle and more flexible.A serious problem with homogeneous FIIAs is detection, as a basic requirement is a change in the detection properties of the immuno-complex compared with the free antigens and antibodies. This can be solved, for example, by using an exitation energy-transfer principle in fluorescence detection, l 2 which, however, requires a complicated labelling technique. Simple detection techniques, which are based on the immunoprecipitin reaction, are nephelometric8.9 and turbidi- metric1() detection, but they are not very sensitive. To enhance the sensitivity, we applied the principle of the immunoprecipitin reaction for the development of an FIIA using a precipitation FI system in combination with a fluores- cence detector (Fig.1). HIgG and fluorescein-labelled antibodies are pre-mixed by an autosampler and injected automatically. The carrier buffer containing PEG, which is known to enhance the immunopre- cipitin reaction,"' 3 transports the mixture through a reaction coil to a metallic filter arranged on-line in the FI system. The precipitate formed is retained on the filter and excess labelled antibodies are washed through by the carrier buffer. The immunoprecipitate is dissolved by switching to 1 mol 1-1 Table 1 Sequence of events of the assay Step Event* Flow ratel Time/ ml min--I min 1 2 The first sample or standard is pre- mixed by the autosampler Inject 75.4 pl of a mixture of 130 1.11 of standard or sample and 195 1.11 of labelled anti-human HIgG 0.5 3 Excess of labelled anti-human IgG reaches the detector; change flow rate 0.7 next solution to inject (1 mol 1-I NaOH) 0.7 7 Change flow rate 0.5 4 5 Change to dissolution reagent 6 Change to carrier buffer 8 Next injection Autosampler starts to pre-mix the * All steps are automated.0 0 2 5 7 8 9.5 10 0.5 ml min-' / t = O &O &O Carrier buffer 0.5 ml min-' New injection 0 --- 0 t = 8 t = 9.5 t=10 0.7 ml min-' t = 2 + Human immunoglobulin G (sample) FlTC labelled anti-human immunoglobulin G 0' Fig. 2. Scheme of the assay procedure. 0Analyst, November 1996, Vol. 121 I567 Incubation time/s Fig. 3. Influence of the incubation time: stopped-flow experiments with nephelornetric detection at a flow rate of 5.22 U 1 - I (0.5 ml min-l). Flow is stopped for various intervals after the immuno-complex reaches the detection cell.Fig. 4 Influence of flow rate; continous flow experiments with nephelo- metric detection at different flow rates were made using a PTFE reaction coil of 2 m X 0.76 mm id. 2.06 pmol c 11111 0 s 10 1.44 pmol k 1.03 pmol NaOH, resulting in a fluorescence signal caused by the liberated fluorescein-labelled antibodies (Fig. 2). Kinetic studies were performed to determine the optimum reaction conditions. Stopped-flow and continuous flow experiments with nephe- lometric detection showed that an incubation time of 90 s is sufficient for quantitative reaction (Fig. 3). A flow rate of 0.5 nil min-1 and a reaction coil of 2 m X 0.76 mm id were found to meet these requirements. The dependence of the signal intensity on the flow rate is shown in Fig.4. During the dissolution process with NaOH the flow rate was increased to 0.7 ml min-1. Comparison of the peak area of the signal for the excess antibodies and the signal for the dissolved antibodies showed an approximately 95% degree of reaction. The influence of PEG concentration on the reaction rate was also studied. Increasing amounts of PEG increased the reaction rate; however, clogging was a major problem. Hence a concentration of 6% PEG in the carrier buffer was used. 3 6 12 15 10 0 5 Molar excess of labelled antibodies Fig. 5 antibodies; continuous flow experiments with fluorimetric detection. Dependence of the reaction yield on the excess of labelled ----- 0 5 10 0 5 I0 0 5 100 5 100 5 10 Timdmin Signals obtained for a decreasing concentration of HIgG.Fig. 61568 Analyst, November 1996, Vol. 121 WE 100- $ 80- 3 60- 40 - 20 - Table 2 Intra- and inter-assay reproducibility for HIgG during 3 d using FITC-labelled anti-human IgG in a 10-fold molar excess over the highest concentration of HIgG Results (day 1) Results (day 2) Results (day 3) Results over 3 d x* A* x * x* ( n = 7) (T* Orcl ( n = 7) (T* or,1 ( n = 7) (T* B,,l ( n = 21) (T* 0x1 Blank 0.14 0.001 0.80 0.14 0.001 0.80 0.14 0.001 0.80 0.14 0.001 0.80 206.5 fmol per injection 0.29 0.006 1.99 0.27 0.008 3.02 0.30 0.008 2.54 0.29 0.014 4.96 413.0 fmol per injection 0.50 0.011 2.24 0.47 0.013 2.68 0.48 0.010 2.01 0.49 0.014 2.96 I .03 pmol per injection 0.89 0.014 1.56 0.87 0.011 1.23 0.88 0.013 1.47 0.88 0.014 1.55 I .44 pmol per injection 0.98 0.012 1.18 0.96 0.010 1.01 0.98 0.011 1.10 0.97 0.012 1.28 2.06 pmol per injection 1.39 0.017 1.21 1.39 0.018 1.27 1.4 0.014 1.01 1.39 0.016 1.16 * Peak area in cm2.1607 120 01 1 I 0 500 1 000 1 so0 2000 2500 Hurnan IgG/fniol per injection (75.4 pl) Fig. 7 Calibration curve for HIgG. Phosphate buffer of pH 7 was found to be optimum as the carrier buffer. For the dissolution of the precipitate, HCl and NaOH were tested. NaOH was chosen because of the higher fluorescence yield of the fluorescein label at high pH. Another parameter influencing the immunoreaction is tem- perature. The best results were obtained under physiological conditions. Therefore, the reaction coil was maintained at 37 "C. In addition, the degree of conversion was found to be dependent on the excess of labelled antibodies (Fig.5). As the blank signal increased with increasing antibody excess, proba- bly caused by adsorption on the metallic filter, a 10-fold molar excess of labelled antibodies was used as a compromise. Several types of filters were checked. Since filters with pore siLes > 0.5 pm did not quantitatively retain the precipitate, the choice was limited. A serious problem was clogging of the filters by the fine amorphous precipitate. A basic requirement was stability at high pH, as 1 mol 1-I NaOH is used for the dissolution of the precipitate. Among the materials tested, only the metallic filter met these requirements. A disadvantage of the metallic filters, however, is the adsorption of a certain amount of labelled antibodies, resulting in a blank signal, which limits the sensitivity.Fig. 6 shows typical signals for various concentrations of HIgG. The first peak corresponds to the excess labelled antibodies and the second to the liberated antibodies; the latter is directly proportional to the concentration of HIgG. The resulting calibration curve is shown in Fig. 7. There is a linear range of more than one order of magnitude with a typical correlation coefficient of 0.997. The detection limit, defined as a signal that is three standard deviations of the blank signal, was found to be 140 fmol per injection (4.6 pmol ml-1 sample). The standard deviation of the blank signal is given in Table 2. The precision of the assay was checked both within one assay series on one day and in the course of several days (Table 2).The intra-assay reproducibility was 1-3% and the inter-assay reproducibility was 1-5% (RSD). Preliminary experiments with serum samples showed that there are no interferences and no sample pretreatment is necessary. The recoveries from serum samples spiked with 340 and 680 nmol ml-' were 102 and 96.2%, respectively, with a reproducibility of about 2% (RSD). Conclusion Precipitation flow injection immunoassay has been shown to be a simple and sensitive approach for the analysis of biological samples. The instrumental set-up is very simple and easily automatable. No stopped-flow is necessary. One assay cycle takes about 10 min. This time could be shortened considerably by using a multiple reaction system. An array of three reaction coils and filters connected in parallel may shorten the time for one assay to about 3.5 min.Further, this approach would allow successive assays for different analytes within one series. The method shows excellent precision, ranging between 1 and 3% for series measured on a single day and between 1 and 5% for continuous measurements over 3 d. The detection limit for HIgG was 140 fmol per injection or 4.6 pmol ml-I of sample. Detection limits in this range are certainly not relevant for IgG analysis; however, this model assay was developed with a view towards future applications to several other analytes in trace amounts. The recovery from serum samples was between 96 and 102%. References 1 2 3 4 5 6 7 8 9 10 11 12 13 Puchades, R., Maguieira, A., Atienza, J., and Montoya, A., Crit. Rev. Anal. Chem., 1992, 23, 301. Pollema, C. H., Ruzicka, J., Lernmark, A., and Christian, G. D., Microchern. J., 1992, 45, 121. Gubitz, G. and Shellurn, C., Anal. Chini. Acta, 1993, 283, 421. Valcarcel, M., and Gallego, M., Trends Anal. Chem., 1989, 8, 34. Martinez, P., Gallego, M., and Valcarcel, M., Anal. Chem., 1987, 59, 69. Kuban, V., FreJenius' J . Anal. Chem., 1993, 346, 873. Marrack, J. R., and Richards, C. R., Immunology, 1971, 20, 1019. Killingsworth, L. M., Britain, C. E., and Woodard, L. L., Clin. Chem., 1975,21, 1465. Sternberg, J. C., Int. Clin. Prod. Rev., 1984, 3, 16. Worsfold, P. J., Hughes, A., and Mowthorpe, D. J., Analy~t, 1985, 110, 1303. Kelly, T. A., and Christian, G. D., Talanta, 1982, 29, 1109. Lim, C. S., Miller, J. M., and Bridges, J. W., Anal. Chzm. Acta, 1980, 114, 183. Creighton, W. D., Lambert, P. H., and Miescher, P. A., J. fmmunol., 1973, 111, 1219. Paper 61031 48J Received May 7, 1996 Accepted July 4, 1996