Anal. Chem. 1994, 66,4027-4033
Fluorescein Isothiocyanate-Labeled Protein G as an Affinity Ligand in Affinity/Immunocapillary Electrophoresis with Fluorescence Detection Oscar-Werner Relf,t Ralf Lausch,$ Thomas Scheper,$ and Ruth Freltag'gt Institut fur Technische Chemie, Universitat Hannover, Callinstrasse 3, 30 167 Hannover, Germany, and Institut fur Biochemie. Westfilische Wilhelms-Universitat Munster, Wilhelm Klemm Strasse 2, 48 149 Minster, Germany
Antibodies from human sera (h-IgG) were tagged with a fluorescent dye (fluorescein isothiocyanate, FITC) through the affinity reaction of FITC-labeled protein G with the Fc fragment of the antibodies. The complexes were quantified by capillary zone electrophoresis (CZE) within 1 min, Le., fast enough to prevent their dissociation during the measurement. Conditionsfor the affinity reaction and the CZE analysis could thus be optimized independently. When an FITC-labeled protein G concentration of mol/L was used, h-IgG concentrations between and mol/L were reproducibly quantified (STD < 2%), using an LIF detector. A correlation coefficient, 9,of 0.9988 was established between the peak height and the IgG concentration. Alternatively, h-IgG containing serum samples and the FITC-labeled protein G were simply injected into the CE capillary in consecutive zones, followed by the application of the electrical field. Within 2 min, the affinity complexes were resolved and the IgG content of the serum quantified (13 = 0.9986). The injection sequence was of no consequence. The measurements agreed well with those found in a single radial immunodiffusion (SRID) assay. In addition FITC-labeled protein G-tagged anti-h-IgG1 antibodies were used to detect the specific antigen of the involved antibody, namely, h-IgG1, in human sera. The exceptional discriminatory power of antibodies together with the fact that antibodies can be raised against virtually all proteins and many other biological macromolecules accounts for the popularity that is enjoyed by the immunoassays in the biosciences and in A similar specificity, if not the simplicity in engineering the desired analyte, is the mark of many bioaffinity reaction^.^ Assays used to detect analytes in complex matrices typically depend on a label, such as an enzyme or a fluorescent dye, linked to one of the reactants for specific detection. Many of these assays, e.g., the widely used ELISA (enzyme linked immunosorbent assay), tend to be time and labor consuming, unsuited to automation, and yield less than satisfactory assayto-assay reproducibility. In areas such as biotechnology, where + Universitlt Hannover. t
Westfalische Wilhelms-Universitfft Mdnster.
(1) Butt, W. R., Ed. Practical Immunoassay; Marcel Dekker: New York, 1984.
(2) Tijssen, P. Practice and Theory of Enzyme Immunoassays. In Laboratory Techniques in Biochemistry and Molecular Biology; Burdon, R. H., van Knippenberg, P. H., Eds.; Elsevier Biomedical Press: Amsterdam, New York, and Oxford, 1985; Vol. 15. (3) Sassenfeld, H. M. Tibtech 1990, 8, 88. 0003-2700/94/0366-4027$04.50/0 0 1994 American Chemical Society
on-line or quasi-on-line data are required for process monitoring and control, such assays are increasingly replaced by (immuno)affinity sensors and (immuno)affinity flow injection analyzers.M Cost and lackof long-term stability are, however, among the disadvantages of these latter systems. The union of the principle of immuno- and/or bioaffinitybased analysis and the methodology of capillary electrophoreses (CE) should yield a highly selective means for reproducible analyte quantification even in complex sample matrices, especially if laser-induced fluorescence (LIF) detection is However, little has been done in this area so far. Nielsen et al.lOJ1showed that the separation of the antibody-antigen (human growth hormone) complex from the singular components is possible. However, they used UV detection, which would be insufficient to quantify the immunocomplexes in a complex protein-containing matrix. Chen et al. have reported on the successful separation of antibodyantigen complexes by capillary isoelectric focusing (CIEF).12 More recently, Arentoft et al. used micellar electrokinetic chromatography (MECC) to separate imm~nocomplexes.~~ Again, UV detection was employed. Karger et al.14 as well as Schulz and Kennedy15 finally adapted the fluorescence immunoassay to CE. None of the above assays has, however, been used to quantify analytes in real-life samples, and most were aimed for a specific and predestined problem. Moreover, the confinement of the CE separation parameters to conditions found suitable for (immuno)affinity complex formation exercised at present unnecessarily limits the application of fluorescence (immuno)affinity capillary electrophoreses (FACE) in general. In this paper, efforts toward a universally (4) Bradley, J.; StkMein, W.; Schmid, R. D. Process Control Qual. 1991,l. 157. (5) Karube, J. Novel Immunosensor Systems. In Biotechnology Vol.2, Biosemors andEnvironmenta1 Biotechnology; Hollenberg, C. P., Sahm, H., Eds.; Gustav Fischer Verlag: Stuttgard and New York, 1988. (6) Freitag, R.; Scheper, T. Chem. Ind. 1992, 9, 62. (7) HemmilP, I. Clin. Chem. 1985, 31, 359. (8) Hernandez, L.; Eskalona, J.; Joshi, N.; Guzman, N. J . Chromatogr. 1991, 559, 183. (9) Wu, S.;Dovichi, N. J. J. Chromatogr. 1989, 480, 141. (10) Nielsen, R. G.; Rickard, E. C.; Danta, P. F.; Sharknas, D. A.; Sittampalam, G. S . J. Chromatogr. 1991, 539, 171. (1 1) Grossman, P. D.; Colburn, J. C.; Lauer, H. H.; Nielsen, R. G.; Riggin, R.M.; Sittampalam, G. S.; Rickard, E. C. Anal. Chem., 1989, 61, 1186. (12) Chen,S. M.; Shively,J. E.; Lee,T. D. TheProteinSociety,SecondSymposium, San Diego, CA, Aug 1988; Forward Press: Seattle, WA, 1988; Abstract 908. (13) Arentoft, A. M.; Frokar, H.; Michaelsen, S.;Sorensen, H.; Sbrensen, S.J. Chromafogr.A 1993, 652, 189. (14) Karger, B. L.; Foret, F.; Schmalzing, D.; Shimura, K.; Szoko, E. Presented at the 5th International Symposium on HPCE, Orlando, FL, Jan 1993. (15) Schultz, N. M.; Kennedy, R. T. Anal. Chem. 1993, 65, 3161.
Analytical Chemistry, Vol. 66, No. 22, November 15, 1994 4027
applicable FACE are described. The potential of the assay for real-life applications is demonstrated by the quantification of the IgG content of human serum samples.
EXPERIMENTAL SECTION Capillary Zone Electrophoresis. All separations were performed on a Beckman P/ACE 2210 instrument (Beckman, Palo Alto, CA) supplied with an argon laser for fluorescence detection. Excitation and emission wavelength of the laser detector were 488 and 520 nm, respectively. For data collection, data analysis, and system control, the System Gold software from Beckman (Palo Alto, CA) on a PS2 computer (IBM) was used. The electrophoresis buffer was a 0.05 M borate buffer, adjusted to pH 10.5 with 1 N NaOH. The buffer was prepared daily from a sterile stock solution and passed through a 0.2-pm filter (Sartorius AG, Gottingen, Germany) prior to use. Fused silica capillaries were from ChromatographieService (Langerwehe, Germany). Capillary dimensions were 50 pm inner diameter and 27 cm length, Le., 20 cm from the capillary inlet to the detector. Prior to the first use, capillaries were treated with 0.1 N NaOH for 15 min, followed by a 5-min rinse with deionized water and a 5-min rinse with electrophoresis buffer. Between runs, the capillaries were cleaned and equilibrated with successive 3-min rinses of 0.1 N NaOH, deionized water, and buffer. Sample injection was done by applying a pressure of 50 mbar for 2 s to the sample valve placed at the grounded end of the capillary. The injected volume was approximately 5 nL. Voltages of 10 or 15 kV were applied as indicated. A temperature of 22 "C was maintained throughout. Chemicals. FITC-protein G (5 charges), human-IgG, and fine chemicals were from Sigma; bulk chemicals were from Fluka. The serum samples were kindly donated by the Medizinische Hochschule Hannover (Hannover, Germany) and by the Red Cross (Springe, Germany). Purification of FITC-Protein G. For purification and removal of any remaining unreacted FITC, the commercially obtained FITC-protein G was dissolved in concentrations of 1 mg/mL in a 50 mM phosphate buffer (pH 7.4). Volumes of 0.5 mL were loaded onto a 2-ml fast desalting column (Pharmacia, Uppsala, Sweden) followed by 2-6 column volumes of the phosphate running buffer. The column was integrated into an FPLC system consisting of two P 500 pumps, the 2141 UV detector, the Superrac fraction collector, and an MV7 injection valve (all Pharmacia, Uppsala, Sweden). The elution was monitored by an Erma 7215 UV/VIS detector at 280 nm (Erma, Japan), and 250-pL fractions were collected. The FITC-protein G containing fractions were further purified, using a IgG-Sepharose column. A 50 mM phosphate buffer (pH 7.4) containing 1 M NaCl was used for elution. The purified protein was desalted, using a microdialyzer (Pierce Chemical, Rockford, IL) and lyophilized. Formation of the FITC-Protein G-h-IgG Complex. For the investigation of the complex formation between FITClabeled protein G and IgG, solutions containing 2000, 1000, 500,200, 20, and 2 pg/mL of h-IgG were prepared in a 100 mM phosphate buffer (pH 5.8). FITC-protein G concentrations were 300 pg/mL for the crude and 250 pg/mL for the purified substance in the same buffer. A 250-pL sample of each h-IgG concentrationwas mixed with 250 pL of the FITC4028
Analytical Chemistry, Vol. 66, No. 22, November 15, 1994
protein G preparation in question. The mixtures were incubated in the microvial for 4 min at 22 "C and immediately transferred to the capillary vial and injected, which took 1 additional min. For investigation of the dependence of the complex formation in the incubation time, 9- and 29-min incubations were also used, resulting in a total of 10 and 30 min, respectively, of incubation at the moment of detection. Incubation temperatures of 20,35, and 45 "C were evaluated for an incubation time of 5 min. The effect of the incubation buffer's salt concentration was studied by dissolving unpurified FITC-protein G (each sample: 120 pg/mL final concentration) and h-IgG (each sample: 500 pg/mL final concentration) in a 50 mM phosphate buffer (pH 6.0), containing 50,100,250,500, and 1000 mM NaC1. The FITC-protein G-h-IgG containing samples were incubated at 22 "C for 5 min and analyzed by CE. The influence of the buffer pH on the interaction between crude FITC-protein G and IgG was investigated, applying 250 pL of each substance in a 50 mM phosphate buffer (pH 6.0). The mixture was incubated for 5 min at 22 "C, and 20 pL was removed for analysis by CE. The pH value of the sample mixture was subsequently lowered to pH 2.5 by the addition of microliter volumes of 2 M phosphoric acid. Following the removal of 20 pL for CE analysis, the pH was raised back to pH 6.0 by adding microvolumes of 6 N NaOH. Again, 20 pL was removed for analysis. The pH value was monitored by pipetting 2-pL volumes of the sample mixture onto pH-indicator paper. Formationof FITC-protein G-anti-h-IgGt-h-IgC1 Complex. Samples of 750 pg/mL anti-h-IgG1and 130 pg/mL unpurified FITC-protein G were incubated for 5 min at 22 OC. The pH value was raised to pH 7.4 by the addition of microvolumes of 6 N NaOH. A total of 2 mL of human serum was dialyzed and lyophilized, followed by resuspension in 250 p L of the FITC-protein G-anti-h-IgG1 solution. This mixture was incubated at 28 "C for 12 h. After 0.1,0.5, 1.0, 6.0, and 12 h, 20 pL was removed and analyzed by CE. h-IgGI depleted serum was treated identically and used as a blank. Analysis of h-IgG in Human Serum by FACE. Prior to the analysis of IgG in human serum, standard solutions containing 1.0, 2.5, 5.0, 7,5 10.0, 15.0, 20.0, and 25.0 mg/mL h-IgG were prepared in incubation buffer (50 mM phosphate, pH 5.8). Human serum samples and standard solutions alike werediluted 1:lO with incubation buffer. Crude FITC-protein G (1600 pg/mL) was alsodissolved in incubation buffer. Using the capillary as theincubation site, the FITC-protein G solution was injected first by applying pressure for 2 s, followed by a 2-s pressure injection of an h-IgG standard or a serum sample. Immediately after the injection, the electric field was activated, and the separation took place. Experiments with a reverse injection order were also performed. The concentrations of h-IgG in the serums were calculated based on the calibration curve obtained by the analysis of the h-IgG standard solutions. Single Radial Immunodiffusion (SRID). The SRID used as the reference assay for quantification of IgG in human serum was performed according to Hobbs.16 (16) Hobbs, J. R. Association of Clinical Pathologists Broadsheet 68, 1970
r
f
-
*- G 5
-
8- F f
0
2-
:-
-
0
-
d
8
8 -
2
BHI
8;-
22
2 -
$ ;
z-
-
Ij:
f6
-
n
0
- 0
-
RESULTS AND DISCUSSION The analysis of affinity or immunocomplexes by capillary zone electrophoresis (CZE) faces some serious problems. The separation of the unbound components from the complex is required for quantification of the analyte. Optimal electrophoresis conditions may, however, be adverse to the stability of the affinity/immunocomplex proper, e.g., the salt concentrations and the pH value of the electrophoresis buffer and the application of the electric field per se. At near neutral pH on the other hand, where most affinity complexes show their highest stability, the CE separation is prolonged by low electroendosmotic flow. During the course of the ensuing slow separation, a certain extend of complex dissociationwill take place, as complex formation is a reversible reaction. According to the interpretation presented by Schulz
0
i
I
f -
and Kennedy,ls the two components rapidly move apart in the electrical field as soon as a complex is dissociated due to the differences in their electrophoretic mobilities. Moreover, the reactants are moved into a region containing a lower concentration of the respective complex partner, thereby decreasing the likelihood of complex reformation. During preliminary experiments (data not shown), we were able to establish the fact that whenever the heightlarea of the complex peak was evaluated as a function of some parameter such as the capillary length that influences mainly the duration of the separation, an asymptotic approach to a maximum signal value is observed with decreasing analysis time. If a separation can be achieved within 3 min or less, the maximum signal from the complex is detected, regardless of the specific electrophoretic conditions. This value is assumed by us to correspond Analytical Chemistry, Vol. 66, No. 22, November 15, 1994
4029
'Oo0
1
,*'
,A + '
/
loo0l
I
,I'
0 0
I
I
200
400
,
600
I
800
0
I
1000
-.. m'
+
,
------&..-.....
.-.
200
0
to a situation where a minimum of complex dissociation is enforced by the CE Conditions. The increased tendency for the adsorption of the protein to the inner surface of the silica capillary observed at a pH value of 7 and below may present another problem. In our case, e.g., no complete separation of unbound FITC-protein G from the FITC-protein G-h-IgG complexcould be achieved at neutral pH. The application of higher salt concentration to prevent the wall adsorption and to speed up the separation leads to increased Joule heating, which also influences the stability of the complex adversely during separation. Our rational in overcoming the above-mentioned problems was to opt for the fastest possible CZE analysis, regardless of whether the CZE conditions necessary to achieve this were also suitable to complex formation. The application of high pH electrophoresis buffers and short capillaries is one way to reach this objective. Due to the high electroendosmotic flow that prevails at pH values above pH 10 even at low salt concentrations, a fast separation of the complex and the unbound FITC-protein G is achieved and wall adsorption is prevented. Consequently, the complex is influenced by the electrophoretic conditions only for a short period of time; too short for the slow reaction kinetics to allow more than an insignificant and undetectable amount of complex dissociation to take place. Using these conditions, the application of FACE to the analysis of human sera became feasible. Formation of FITC-Protein G-h-IgG Complex. The formation of the affinity complex between FITC-labeled protein G and h-IgG strongly depends on the concentrations of both proteins. Using a molar excess of FITC-protein G (approximately 10" M), h-IgG in concentrations between 10" and lo-* M could be detected as a result of the formation of the affinity complex (Figure 1A-G). IgG concentrations of approximately 10" M constituted the detection limit, since the signal caused by the affinity complex peak was already hard to distinguish from the baseline noise in that case (Figure 1A). When a constant voltage of 15 kV was applied, the complex was seen as a single sharp peak 1.04 min after injection (Figure 1A-C) and after 2.05 min when a voltage of 10 kV was applied (Figure 1D-G). For a total of 28 (10 kV) and 40 (1 5 kV) experiments, relative standard deviations of 1.93% 4030
Analytlcal Chemlstty, Vol. 66,No. 22, November 15, 1994
I
600
,
-
.-.. 1
800
1
1000
h-IgG [pgiml]
h-IgG [pgiml] Flguro 2. h-IgG calibration curves recorded as a function-of the incubation time of h-IgG with FITCprotein 0. Incubation temperature, 5 O C ; (M)incubation time, 5 min; (+) 10 min; (A)30 mln.
1
400
---------
as a function
Figure 3' h-lgG
~
~
(A)45
OC.
~
~ ~
~,
: ~ ~
~~
".cT~$TT!
Of
the
o~
~
and 1.64%,respectively,werecalculated for the retention time of the complex. The broad peak with several spikes following the complex signal is caused by surplus FITC-protein G. The increase of the applied voltage for separation did not influence either the complex stability under CE conditions,the resolution, or the calibration curve, but allowed the separation to be completed within 2 min. The application of even higher voltages led to increased Joule heating, which influenced the complex formation and stability adversely. A linear relationship with a correlation factor r2of 0.9988 could be established between the h-IgG concentration and the peak height, as evaluated from the base line. For an h-IgG concentration of 500 pg/mL, a relative standard deviation of the peak height of 1.72% was determined. The peak area correlated less well with the h-IgG concentration due to the pronounced peak tailing, which caused overlaps with the signal of unbound FITC-protein G. This was especially eminent in the lower concentration range. For analytical purposes, the determination of the concentrations by the peak height of the affinity complex peak is recommended. FITC binds in different quantities and at different locations to protein G during the labeling process. Therefore, the charge and-to a lesser degree-the molecular mass of the FITCprotein G varies. Since the separation in CZE stronglydepends on the mass to charge ratio, a single protein tagged with various ratios of FITC causes several signals. The broad peak and the spikes seen in the signal of the surplus FITC-protein G are caused by such variations of the number of bound FITC molecules. Due to the predominance of the mass to charge ratio of the more homogeneous h-IgG in the affinity complex, only a single signal is observed in that case. No influence of the ratio of FITC bound to protein G on the complex formation could be observed. It would, nevertheless, be preferable to use a uniform FITC-protein G. We attempted to purify the commercial FITC-protein G and repeated the incubation and CZE experiments. The peak shape of protein G was still broad, but lacked the spikes (data not shown). Apparently, the commercial FITC-protein G was contaminated with unbound FITC and someunknown labeled proteins or peptides, which are removed by the IgG affinity column during purification. However, the complex formation, the resolution
~~
,~
-I r,:
C
Y 0
J
n
Figure 4. Influence of the pH value of the sample buffer on the interaction between FITC-protein G and h-IgG. (A) CZE of an 20-pL aliquot taken from a sample containing FITC-protein G and IgG in a 50 mM phosphate buffer, pH 6.0. The complex signal is observed at 1-01 min. (8)CZE of the same sample after the buffer’s pH had been lowered to 2.5 by the addltion of a mlcrovolume of 2 M phosphoric acid. No signal from the complex is observed; all signals are caused by the FITC-protein 0. (C) CZE of the same sample after elevation of the sample pH back to 6.0 by the addition of a microvolume of 6 N NaOH. The complex signal is once more present at 1.01 mln.
of the complex peak, and the FITC-protein G signal were not improved by the application of the purified FITC-protein G. Therefore, the commercial FITC-tagged protein G was used without further purification in the following experiments. Figure 2 shows the influence of the incubation time on the complex formation. Even after an incubation time of 30 min the affinity reaction has not reached equilibrium, as the peak height continues to increase somewhat with incubation time. However, as observed before in the case of immunosensors and immunoflow injection analyzers, reaching the equilibrium is not necessary in such cases if standardized incubation times are used.” Here too a linear relationship is reproducibly observed between the h-IgG concentration and the peak height at all incubation times. Moreover, the increase in signal intensity observed after 30 min compared to that seen after 5 min is only small, since approximately 90%of the maximum peak height is already reached after 5 min. Reliable results can even be achieved with incubation times of less than 1 min. Therefore, with the general instability of the FITC-labeled protein G in mind, prolonged incubation times should and can be avoided. A strong impact on the complex formation was found for the incubation temperature (Figure 3). At 20 OC the complex is stable, while at higher temperatures lower complex concentrations were observed. At 35 OC the peak height decreased by about 30% compared with the results achieved at 20 O C ; at 45OC an even lower complex concentration was found. This indicates a negative impact on the temperature on complex formation kinetics and complex stability. For incubation temperatures of 35 and 45 OC,the relationship (17) Freitag, R.; Scheper, T.; Schiigerl, K. Enzyme Microb. Techno/.1991, 13, 1.
between the peak height and the h-IgG concentration ceases to be linear. No explanation was found for this repeatedly observed result. The influence of the pH value of incubation buffer on the affinity complex formation is shown in Figure 4A-C. At pH 6.0, complex formation took place. After lowering the pH of the sample to a value of 2.5, the complex dissociated, and nearly no complex was detected. This is comparable to the results achieved in affinity HPLC, where low pH values are used for elution of antibodies bound to the protein G immobilized on the column matrix.18 After raising the pH again, the signal of the affinity complex was once more present in the electropherogram. Furthermore, we observed an acceleration in complex formation when 40-100 mM of NaCl was added to the incubation buffer. The addition of more than 250 mM, on the other hand, had an adverse effect on the complex stability. Determination of IgG in Human Serum by FITC-Protein G. Protein G binds to all subgroups of human IgG, but does not bind to other human immunoglobulins such as IgA or IgM. Thereby, FITC-labeled protein G should be a suitable agent to allow a direct quantification of h-IgG in human serum. To reduce sample handling as far as possible, the injection mode suggested by Regnier19Jowas used. The serum sample or the standard solution and the FITC-protein G were injected consecutively into the capillary. The affinity complex between h-IgG and protein G was formed and at the same time separated from the unbound FITC-protein G. We found that, ~~
-~
~
~
(1 8) Pharmacia. Product Information.
(19) Bao, J.; Regnier, F. E.J . Chromofogr. 1992, 608, 217. (20) Regnier, F. E.Presented at the 6th International Symposium on HPCE,San Diego, CA, 1994.
Analytical Chemistry, Vol. 66, No. 22, November 15, 1994
4031
f
z 0
4
9
13
Figure 5. Application of FITC-protein G for the analysis of IgG in human serum of various patients. (A) IgG deficiency serum; (B) normal serum: (C) serum with abnormal high IgG concentration.
IgG Depleted Serum
t
IgG Deficiency Serum
Abnormal High IgG Conc. Serum
Normal Serum I
0
.
.
.
.
I
5
.
.
.
.
I
.
.
10 h-IgG [mglml]
.
.
,
15
.
.
.
.
,
20
Figure 8. Comparison of IgG concentrations measured in human serum by SRID and CZE using FITC-protein G.
in the case of a 2-s pressure injection for both the sample and the FITC-labeled reagent, the sequenceof injection was of no consequence since identical results were achieved with a reversed order of injections. Due to the dispersion effect between the injected samples and the parabolic flow profile of the solution during the pressure injection, the zones were already sufficiently mixed during injection. Further mixing due to differences in the electrophoretic velocities of the two reaction partners may have taken place after the electric field had been applied. As shown in Figure 5, the peak representing the affinity complex between h-IgG and FITC-protein G is not as well resolved from the surplus FITC-protein G peak as observed above when a premixed sample was analyzed. However, the h-IgG could be reliably quantified. The manual sample and reagent handling were much reduced; the automation of the entire assay becomes possible. Using the h-IgG standards, a correlation of r2 = 0.9986 between the peak height and the h-IgG concentration was found. On the basis of this 4032 AnalyticalChemistry, Vol. 66, No. 22, November 15, 1994
calibration, various human sera were analyzed. The determined IgG concentrationsof were comparable to the results achieved by single radial immunodiffusion (SRID) (Figure 6). With regard to theshort analysis time, the easeof handling, and the potential for automation, this type of fluorescence affinity capillary electrophoresis (FACE) awaits a number of applications in biotechnology or medicine. Antigen Detection by FITC-Protein G Tagged Anti-b-IgG1 Antibodies. Protein G binds to the Fc region of the antibody; the Fab region of the immunoglobulin is thus still capable of binding to its specific antigen. Antibodies tagged with FITCprotein G may therefore be used to quantify antigens even in complex matrices by making these antigens accessible to quantification by LIF. In the first step, the FITC-protein G-antibody complex has to be created using an equimolar ratio of both reactants and the optimum conditions predetermined in the above experiments. Any antibody left untagged would obviously not be able to promote antigen detection, whereas residual FITC-protein G can be tolerated in many cases. In our case, however, the antigen to be detected was also an antibody, namely, h-IgG1. Therefore, all unreacted FITC-protein G had to be removed in order to prevent its reaction with the antigen. In Figure 7A, the complex formation between FITC-protein G and anti-IgG1 antibodies is shown. The other substances present in the unpurified protein G did not interfere in the following experiments, apparently they were not related to protein G. The FITC-protein G-anti-h-IgG1 complex was added to concentrated human serum to allow the formation of the threeprotein “sandwich” complex for analysis of serum IgGl The optimalincubation conditions for this sandwich complex were found to be at a pH of 7.0-7.4. After a 1-h incubation period,
.
I.
1
-7
i !IB
'j A ::
I
8004
600
/ ,
.
.
o 0
20
40
60
80
100
300
400 f
'
500 I
'
600 t
'
700 l .
800 1 '
Incubation Time [min.] Flgure 8. Influence of the incubation time on the formation of the sandwlch complex between FIX-protein G-antl-h-IgG1 and h-IgG,. (0)Peak height of the sandwich complex (IgG1-aIgG1-ffi); (M) peak height of the residual FITC-protein G-anti-IgG, complex (aIgGl-PO).
0
9
0
n
Flgure 7. Quantification of h-IgGl in preconcentrated human serum via the formation of a sandwlch complex with FITC-protein Gantih-IgG,. (A)CZE analysis of the complex formed between FITC-protein G and anti-IgGl (aIgG1-PG). (e) CZE analysis of h-IgG, in preconcentrated human serum by formation of a sandwich complex with FITC-protein Ganti-IgG1 (IgG 1-aIgG 1-PG).
a peak most likely caused by the sandwich complex can be observed in the electropherograms (1.26 min), which increases steadily with increasing incubation time. The electropherogram obtained after a 6-h incubation is shown in Figure IB. This new peakcannot be caused by the FITC-protein G-anti-h IgG1,which appears as before after 2.1 1 min, nor by an affinity complex formed directly between the FITC-labeled protein G with the analyte IgG since such a complex was established to have the same mobility as the FITC-protein G-anti-hIgGI, Le., to appear after roughly the same time. The time requirement of the complex formation is a drawback of the method, for the association of the FITCprotein G-antibody complex to the antigen is governed by
slow reaction kinetics (Figure 8). There is a direct correlation between thedecreaseof the FITC-protein G-anti-h-IgG1 peak and the increase of the sandwich-complex peak with incubation time. Even after 12 h, the formation of the sandwich complex is not completed. The immuno reaction between the two antibodies can be accelerated by raising the temperature to more than 30 OC, but this has an adverse effect on the stability of the FITC-protein G-anti-h-IgG1 complex. The assay based on the sandwich formation does, however, present the most universal variety of FACE. FITC-protein G can be used to tag any antibody, which reaction in turn yields a reagent suitable for antigen detection even in complex samples such as human serum.
ACKNOWLEDGMENT Beckman (Miinchen, Germany) supported this work by allowing us access to a fluorescencedetector for our CE system. Samples of human serum were kindly donated by the Medizinische Hochschule Hannover (Hannover, Germany) and the Red Cross (Springe, Germany). Received for review April 5, 1994. Accepted August 9, 1994." ~
~
@Abstractpublished in Aduance ACS Abstracts, October 1, 1994.
Analytical Chemistry, Vol. 66, No. 22, November 15, 1994
4033
