Anal. Chem. 1994,66, 1860-1865
This Research Contribution is in Commemoration of the Life and Science of
I. M. Kolthoff (1894- 1993).
Simultaneous Immunoassay Using Electrochemical Detection of Metal Ion Labels Fred J. Hayes,+ H. Brian Haisall,' and William R. Heineman' Department of Chemistry, University of Cincinnati, Cincinnati, Ohio 4522 1-0 172
The concept of a simultaneousdual analyte immunoassay based on two different metal ion labels is demonstrated. The model system consists of two proteins, human serum albumin (HSA) and immunoglobulin G (IgG). Bismuth and indium ions have been coupled to these proteins through the bifunctional chelating agent diethylenetriamine-pentaaceticacid (DTPA). A maximum molar labeling ratio of 6 1 and 1 0 1 was obtained for HSA and IgG, respectively. Following a competitive equilibrium between unlabeled and labeled protein for a limited amount of specific antibody immobilizedon polystyrene, the bound metal ion labels were released by acidification and detected by differential pulse anodic stripping voltammetry (ASV). Limits of detection for HSA and IgG are 1.8 and 0.6 pg/mL, respectively. Application of the dual immunoassay to human serum samples gave results that were comparable to those obtained by nephelometry. The major trend in immunoassay development over the past decade has been away from liquid-phase assays involving radioisotopic labels and toward fast, homogeneous or solidphase assays using nonisotopic labels.' Recently, there has been considerable interest in the development of immunoassays that permit the simultaneous determination of two or more analytes in a single sample.2 Obvious advantages include reductions in sample size, analysis times, and overall cost. Multiple testing is most attractive for analytes that are grouped in panels such as thyroid-function tests and allergens and can be particularly advantageous in cases where a ratio of compounds is used to aid in diagnosis. Dual analyte immunoassays based on the use of 1251-and 57Co-labeled tracers have been reported for folate and ~ o b a l a m i n lutropin ,~ and f~llitropin,~ as well as thyrotropin and t h y r o ~ i n .Recently, ~ dual assays were described using combinations of europium, terbium, or samarium as labels and time-resolved fluorescence detection.'js7 Other assays have Present address: The Procter & Gamble Co., Miami Valley Laboratories, Cincinnati, OH 452394707, (1) Gosling, J. P. Clin.Chem. 1990, 36, 1408-1427. (2) Hage, D. S. Anal. Chem. 1991.63, 206R-209R. (3) Gutcho, S.;Mansbach, L. Clin.Chem. 1977, 23, 1609-1615. (4) Wians, F. J.; Dev, J.; Powell, M. M.; Heald, J. I. Clin.Chem. 1986, 32, 887890. ( 5 ) Gow, S.M.; Caldwell, G . ; Toft, A. D.; Beckett, G . J. Clin.Chem. 1986, 32, 2191-2194. (6) Hemmila, I.; Holttinen, S.;Pettersson, K.; Lovgren, T.Clin.Chem. 1987,33, 2281-2283. (7) Vuori, J.; Rasi, S.;Takala, T.; Vaananen, K. Clin.Chem. 1991, 37, 20872092.
1860
Analytical Chemistry, Vol. 66,No. 11, June 1, 1994
included use of enzymes,8 colored latex particle^,^ and flow cytometry. l o Multianalyte immunoassays have typically been performed using antibodies or antigens with different chemical labels. The main limitation to this approach is the simultaneous detection of different labels in the final step of the procedure. Although multichannel scintillation counters permit mixtures of radioisotopes to be quantitated, the instrumentation is expensive, and the radioisotopes are environmentally problematic. In assays using different enzymes or fluorescent labels, the optimum conditions for detection, such as pH or enhancement solution, can be significantly different for each. The different sensitivities of the labels may adversely affect the detection limits of the analytes. For multianalyte immunoassays using different chemical labels, it is desirable to obtain signals using the same detection scheme in which the labels exhibit essentially the same sensitivity under the conditions chosen for detection. These problems associated with the common labels used for immunoassay prompted the development of arrays of individual reaction zones on a single solid phase for multianalyte assays." This has the advantage of using only one label, but does require a detection system capable of serial or simultaneous quantitation of signals from each spatially resolved zone. This concept is the basis for microspot multianalyte immunoassays proposed by Ekins. l 2 Examples of assays using this format include those for theophylline and hemoglobin13 and for a series of a1lerger1s.l~A simultaneous immunoassay of four analytes in serum using a fluorescent europium chelate was described re~ent1y.l~ We are exploring the concept of multiple analyte immunoassay based on multiple metal ion labels that are detectable by voltammetric techniques. Metal ions can be readily coupled to proteins through the use of bifunctional chelating agents. Much of the work in this area has focused on the development of highly stable radiolabels for antibodies used in nuclear (8) Blake, C.; AI-Bassan, M. N.; Gould, B. J.; Marks, V.; Bridges, J. W.; Riley, C. Clin.Chem. 1982, 28, 1469-1473. (9) Hadfield, S.G.; Lane, A.; McIllmurray, M. B. J . Immunol. Methods 1987, 97, 153-158. (10) McHugh,T. M.; Wang, Y. J.;Chong, H. 0.J. Immunol. Methods 1989,116, 21 3-2 19. (11) Kricka, L. Clin.Chem. 1992, 38, 327-328. (12) Ekins, R.; Chu, F.; Biggart, E. J. Clin.Immunoassay 1990, 13, 169-181. (13) Donohue, J.; Bailey, M.; Gray, R. Clin.Chem. 1989, 35, 1874-1877. (14) Brown, C. R.; Higgins, K. W.; Frazer, K. Clin.Chem. 1985, 31, 15OC-1505. (15) Kakabakos, S.E.; Christopoulos, T. K.; Diamandis, E. P. Clin.Chem. 1992, 38, 338-342.
0003-2700/94/0366-1860$04.50/0
0 1994 American Chemical Society
medicine.I6 This technology can be adapted to the development of stable, but releasable, nonisotopic metal ion labels for immunoassay. The method of detection chosen for these labels is the electrochemical technique of stripping vo1tammetry.l7J8 The advantages of stripping voltammetry include the capability to determine simultaneously four to six metals at concentrations down to 10-lo M. Additionally, the required instrumentation is relatively inexpensive compared to that necessary for the spectroscopic techniques used for trace metal analysis. The metal ion labeling strategy combined with stripping voltammetry offers an attractive way for the development of immunoassays in which multiple analytes are determined simultaneously. In 1982 Doyle et al. reported a competitive heterogeneous immunoassay for HSA using In3+ as the releasable label.19 The purpose of this assay was to introduce a nonisotopic labeling strategy. We have now expanded that labeling concept to permit two analytes to be quantitated simultaneously. Here, we illustrate this assay concept using Bi3+and In3+ as metal ion labels and two proteins, human serum albumin (HSA) and immunoglobulin G (IgG), as model analytes. These two metals were chosen because they exhibit strong binding with DTPA: log Kf= 35.6 for Bi3+ and log Kf= 29 for In3+.20 They are also absent from most samples of biological origin and are detectable by anodic stripping voltammetry (ASV). Determinations of HSA and IgG in serum are often performed in clinical laboratories as an aid in the diagnosis of certain diseases, and the IgG to albumin ratio in cerebral spinal fluid is used in diagnosis of multiple sclerosis and damage to the central nervous system.21
EXPERIMENTAL SECTION Materials. All chemicals were used without further purification. Chemicals and supplies used were as follows: human serum albumin, human IgG, protein standard (5.0 g/dL albumin and 3.0 g/dL globulin), and HEPES buffer (Sigma Chemical Co., St. Louis, MO); QCS unassayed normal and abnormal control sera (lyophilized) (Ciba Corning Diagnostics Corp., Irvine, CA); purified polyclonal rabbit antibody to human albumin and human IgG (BoehringerMannheim, Indianapolis, IN); DTPA (G. Frederick Smith Chemical Co., Columbus, OH); InC13 (99.999%), Bi(NO3)3.5H20 (99.999%), ultrapure nitric acid (99.999%), and ultrapure hydrochloric acid (99.999%) (Aldrich Chemical Co., Milwaukee, WI); Spectra/Por dialysis membrane tubing (MWCO of 12000-14000), 12 X 75 mm sterile polystyrene culture tubes, and Tween 20 (Fisher Scientific, Cincinnati, OH); protein dye reagent concentrate (Bio-Rad Laboratories, Richmond, CA); l / 4 in. diameter polystyrene balls (Polysciences, Inc., Warrington, PA). All other chemicals used Hnatowich, D. J. Nucl. Med. Biol. 1990, 17, 49-55. Heineman, W. R.;Mark, H. B., Jr.; Wise, J. A.;Roston, D.A. In Lnborafory Techniques in Electroanalytical Chemistry; Kissinger, P. T., Heineman, W. R., Ed%; Marcel Dekker: New York, 1984; Chapter 19. Wan& J. Stripping Analysis; VCH Publishers: Deerfield Beach, FL, 1985. Doyle, M. J.; Halsall, H. B.; Heineman, W . R. Anal. Chem. 1982, 54, 2318.,** LJLL. (20) Martell, A. E.; Smith, R. M. Critical Stability Constants; Plenum: New York, 1979; Vol. 1, p 281. (21) Moriarty, G. In Clinical Chemistry; Theory, Analysis, and Correlation, 2nd ed.; Kaplan, L. A., Pesce, A. J., Eds.; Mosby: St. Louis, MO,1989; Chapter 39.
were reagent grade and supplied by Fisher Scientific. Solutions were prepared with distilled, deionized water (Barnstead Nanopure System). Human serum samples were provided by the University of Cincinnati Medical Center Diagnostic Immunology Laboratory. CAUTION: Universal precautions for handling biohazards were exercised in the handling of human serum samples. Instrumentation. Voltammetric analyses were performed with a PAR 174Apolarographicanalyzer (EG & G Princeton Applied Research, Princeton, NJ). A PAR 303 static mercury drop electrode (SMDE), Ag/AgCl reference electrode, and Pt auxiliary electrode comprised the electrochemical cell. Voltammograms were recorded on a H P 7015B x-y recorder (Hewlett-Packard Co., Sunnyvale, CA). pH measurements were made with a glass-body combination pH electrode (Fisher Scientific, Cincinnati, OH). Spectrophotometry was done with a H P 8452 diode array spectrophotometer (HewlettPackard Co., Sunnyvale, CA). Labeling Procedure. DTPA was coupled to proteins covalently using the bicyclic anhydride method.22.23 The anhydride of DTPA was prepared as described.22.23 The chelate was coupled to HSA and IgG at various molar reaction ratios by dissolving 100 mg of protein in 5.0 mL of 0.10 M HEPES (pH 7.0). The solid anhydride was added and the reaction allowed to proceed for 1 h at room temperature. This solution was dialyzed against 0.25 M citrate buffer (pH 5.5) to remove unreacted DTPA. A 50 M excess of metal ion was added to the protein solution, which was then incubated for 1 h at room temperature. InC13is soluble in the citrate buffer and was added directly to the protein solution. A 10 mM bismuth stock solution was prepared by first dissolving Bi(N03)3.5H20 in 1.0 M HNO3, adding sodium citrate to give 0.25 M, and adjusting the pH to 5.5 with 1.0 M NaOH. Excess and nonspecificallycomplexed metal ions were removed bydialysisagainst0.15 Mcitratebuffer (pH 5.5). Thelabeled protein was dialyzed against 10 mM phosphate buffered saline, pH 7.2 (PBS), containing 0.15 g/L NaN3, and stored at 4 OC. These solutions were found to be stable for several months with no detectable loss of the labels. Determination of Labeling Ratio. The concentrations of the labeled protein solutions were determined using UV absorption spectrophotometry and the molar absorptivities of HSA and IgG or with the Bio-Rad protein assay reagent using pure protein standards. The molar labeling ratios were determined by diluting a known amount of In3+- or Bi3+labeled protein in 1.O M hydrochloric acid to release the label, which was then quantitated by ASV by generating a standard calibration curve from solutions of known amounts of metal ion and native protein in the same electrolyte or by the method of standard additions. Characterization of Label Release. In3+-and Bi3+-labeled HSA were adsorbed passively, and separately, onto the walls of the polystyrene tubes from a 10 pg/mL solution in PBS (pH 7.2). The tubes were incubated with various concentrations of hydrochloric and perchloric acid, and the dissociation (22) Hnatowich, D. J.; Layne, W. W.; Childs, R. L. Inf.J. Appl. Radiat.Isot. 1982, 33, 327-332. (23) Hnatowich, D. J.; Layne, W.W.;Childs, R. L. Science 1983,222,613615.
Analytical Chemistry, Vol. 66,No. 11, June I , 1994
1001
of the metal complex was allowed to reach equilibrium. The released metal ions were then quantitated by ASV. Titration of Immobilized Antibody. Capture antibodies for HSA and IgG were adsorbed passively, and separately, onto the walls of the polystyrene tubes from solutions of 5.0 pg/mL of the partially purified polyclonal antibodies in 10 mM HEPES (pH 7.5) with a 2-h incubation at room temperature. The tubes were rinsed with 10 mM phosphate buffered saline, pH 7.2, containing 0.05% Tween 20 (PBST). Immobilized antibody was incubated for 2 h at room temperature with increasing concentrations of labeled protein in PBST. The tubes were rinsed 3 times with distilled, deionized water and the bound labels released with 1.0 M HCl. The released metal ions were then quantitated by ASV. Immunoassay Protocol. Capture antibody for single analyte assays was immobilized on the surface of the polystyrene tubes by passive adsorption using concentrations of 5.0 pg/mL of partially purified polyclonal antibody in 10 mM HEPES (pH 7.5) and a 2-h incubation at room temperature. For dual assays, a mixture of 2.5 pg/mL of each antibody in solution was used and eight polystyrene balls were added per tube to immobilize sufficient antibody. The tubes were then rinsed once with PBST. Samples containing a fixed concentration of labeled protein and various concentrations of protein standards or dilutions of human serum in PBST were incubated in the tubes for 2 h at room temperature. The tubes were rinsed 3 times with PBS (pH 7.2) and once with distilled, deionized water and then filled with 1.0 M hydrochloric acid and allowed to equilibrate for 5 min. This solution was transferred to the electrochemical cell and the metal ions were detected by differential pulse ASV. Instrumental conditions were as follows: 4.0 min of argon purge, 10 min deposition at -750 mV versus Ag/AgCl with stirring and 30 s quiet time, positive potential scan at 2.0 mV/s, 25 mV pulse modulation amplitude for In3+, 10 mV pulse modulation amplitude for Bi3+,and 10 pA output full scale. Detection time from the point of release for In3+ and Bi3+in the dual assay was 20 min for each sample. The bound and free labeled antigens were separated easily using polystyrene as the solid phase. Correlation Studies. The single analyte immunoassays for HSA and IgG were compared to the Bio-Rad protein assay using pure protein solutions. Samples and standards for the Bio-Rad assay were prepared by diluting various volumes of lOmg/mLstocksolutions inPBS (pH 7.2). Theconcentrations ranged from 0 to 1.O mg/mL which are in the linear range of the standard Bio-Rad assay. The standards and samples for the immunoassay procedure were prepared from these solutions by dilution in PBST and ranged from 0 to 10 pg/mL. Dual immunoassays were performed with 10000-fold dilutions of human serum samples using dilutions of the Sigma protein standard to develop the standard curves. The values of the standards ranged from 0 to 5 pg/mL and 0 to 3 pg/mL for HSA and IgG, respectively. The results of these assays were compared to those obtained by nephelometry. RESULTS AND DISCUSSION Our assay format is a direct competitive heterogeneous immunoassay (Figure 1). Antibodies specific for HSA and IgG are adsorbed passively onto polystyrene. The unlabeled 1862
Analytical Chemistry, Vol. 66,No. 11, June 1, 7994
Add sample
Immobilize antibody
Incubate
1
Detect by ASV
Rinse
Figure 1. Schematicof thedualdtrect competitiveimmunoassayformat. Table 1. Coupling of DTPA to HSA and I g W
reaction ratio DTPA:protein 5:l 1O:l 20:l 25:l 5 01
labeling ratio DTPA:HSA
labelin ratio
DTP~I~G 3 5
10 10
Experimental values were rounded to whole numbers.
proteins being measured compete with a known amount of labeled protein for the limited number of antibody binding sites. Excess protein is washed away and the bound metal ions are released by acidification and detected by differential pulse ASV. The signal generated by the released labels is indirectly proportional to the amount of unlabeled protein in the standard or sample. Molar Labeling Ratio. Various ratios of bicyclic anhydride to protein were used in the reaction to couple the chelate to HSA and IgG. The goal was to couple as many functional groups as possible to the protein without significantly altering its antigenic properties. This would enhance sensitivity by attaching the maximum number of metal ion labels to the proteins. The extent of DTPA conjugation per molecule of protein was determined by complexation with In3+ or Bi3+ and subsequent quantitation of these labels by ASV. The numbers of DTPA groups attached to HSA and IgG by the bicyclic anhydride reaction at various reaction ratios are presented in Table 1. Under theconditions used, the maximum numbers of chelates conjugated per molecule were 6 and 10 for HSA and IgG, respectively. These values agree with the results of Hnatowich et aL2*and other methods of coupling that involve protein amino groups such as those used by Meares et al.Z4 The maximally-labeled proteins were used in the immunoassays. Comparison of the signals generated at zero dose of unlabeled protein with those containing an equimolar mixture of labeled and unlabeled protein indicates no apparent loss in reactivity of the labeled proteins with the specific antibodies. An important step in the labeling procedure is the dialysis. Both In3+ and Bi3+ bind to various ligating moieties in both IgG and HSA in additon to the DTPA labels. Metal ions bound to these “other” sites are also released by acidification. Metal bound to these sites is coordinated weakly compared ~
(24) Meares, C. F.; McCall, M. J.; Rearden, D. T.; Godwin, D. A,; Diamanti, C. I.; McTigue, M. Anal. Etochem. 1984, 142.68-78.
U""
,
I
L
E
u
E
Y
n
n
a
0.00
0.20
0.40
0.60
0.80
1.00
0.01
Concentrrtlon of acid (M)
0.20
0.40
0.60
0.80
1
10
100
Concentration (pg /m L)
Figure 2. Release of In3+ from Immobilized labeled HSA with hydrochloric acid (W) and perchloric acid (0).Percent of labelreleased was based on amount released with 2.0 M HCI as 100%.
0.00
0.1
1.00
Concentration of acid (M)
Figure 3. Release of B13+ from immobilized labeled HSA with hydrochloric acid (D)and perchloric acid (0).Percent of label released was based on amount released wlth 2.0 M HCI as 100 % .
to that bound to DTPA and can "leak off" during the immunoassay procedure. Consequently, assay precision is significantly improved by restricting coordination to DTPA sites. This is accomplished by the dialysis step in which weakly coordinated Bi3+ and In3+ are removed. Release of Label. In order to detect the metal ion labels by ASV in the final step of the immunoassay, they must be released from the chelate. Acidification dissociates metal ions from DTPA chelates by protonation of the carboxylate and amine coordination sites. Dissociation can be assisted by use of an acid whose anion forms soluble complexeswith the metal ion. The acidity of the release solution as well as type of acid is critical to maximize the detection limit of the assay. The dependence of In3+dissociation from labeled HSA on acid concentration using two different acids, hydrochloric and perchloric, is illustrated in Figure 2. The dissociation of Bi3+ from labeled HSA using the same acids is illustrated in Figure 3. Hydrochloric acid releases the labels efficiently and is especially critical for the release of Bi3+, since chloride ion assists in the release by coordinating and forming a strong
Flgure 4. Titration of separatety lmmoblllzedantibody wlth In3+-labeled HSA (H) and BP+-labeled IgG (0)with a 2-h incubatlon.
complex with Bi3+.25Based on these data, 1.OM hydrochloric acid was chosen as the release solution for the immunoassay. This solution ;eleases the metal ion labels rapidly with essentially complete dissociation of the metal ion complexes within 2 min (data not shown). Similar behavior was observed for the dissociation of these two metal ions from labeled IgG, which is to be expected since DTPA is the coordinating agent in both cases. This means that the labels could be used interchangeably. Titration of Immobilized Antibody. Titration curves of immobilized antibody with metal-labeled protein were used to determine the optimum concentrations of labeled protein in the immunoassays. These curves are illustrated in Figure 4 for HSA and IgG using In3+-labeled and Bi3+-labeled proteins, respectively. Saturation of antibody binding sites is approached above 1.Opg/mL. Labeled analyteconcentrations greater than this were chosen for the immunoassays to maximize the signals generated by the metal ion labels. This was necessary since the level of metal ions released in the immunoassays is close to the detection limits of the electrochemical system used in the method development. Quantitationby ASV. The method of detection chosen for the metal ion labels was differential pulse ASV at a hanging mercury drop electrode. Typical stripping voltammograms for the simultaneous detection of low levels of In3+ and Bi3+ in 1.0 M HC1 are shown in Figure 5. The appropriateness of In3+and Bi3+as choices of labels is illustrated by the bottom voltammogram. A typical blank solution contains easily detectable levels of Cd2+, Pb2+, and Cu2+. The absence of In3+and Bi3+from the blank enables a lower limit of detection for the two labeling metal ions. Interassay precision (CV) for the quantitation of indium and bismuth using a 10-min deposition time ranges from 5 to 9%and 3 to 5% for 10-9 and M solutions, respectively. Limits of detection using conditions chosen for the immunoassay are 7.5 X 10-lo and 8.5 X M for indium and bismuth, respectively. Comparable results were obtained when the labels were switched (Le., Bi3+-labeled HSA and In3+-labeled BSA). ( 2 5 ) Meites, L. Handbook of Analytical Chemistry: McGraw-Hill New York,
1963; pp 1-39.
Analytlcai Chemistry, Voi. 66,No. 1 I, June
I, 1994
1063
800 Cd In
s-
El
Cu
Pb
n
600
1
400 0
x U
n
200
0 -1000
-800
-400
-600
-200
0
200
0
Potentinl (mV) versus AglApCl
Figure 5. Typical differential pulse stripping voltammograms for indium and bismuth in 1.0 M HCI using the HMDE and a 10-min deposition. Voltammograms are for blank electrolyte (bottom), 1.0 X M, and 1.0 X lo-* M (top) solutions of both metals. HSA
0
20
IgG
40
60
80
100
Concentration (pg/mL)
Flgure 6. Typical calibration curves generated from a dual analyte immunoassay for the simultaneous determination of HSA and IgG. Concentratlonsof Bi3+-labeledHSA and In3+-labeledIgG In the assay were 28.0 and 7.0 pg/mL, respectively.
Although the formation constants for these metal ions differ by several orders of magnitude, the values are so large that this difference has no effect on the results. Immunoassay. A significant problem encountered during this stage of the experimentation was the very low concentrations of metal ions in the release solution. This is due to the relatively poor surface to volume ratio of the immunoreactor vessels. In order to increase the surface area, and thereby increase the amount of In3+/Bi3+released from the surface, small polystyrene spheres with immobilized antibody on the surface were added to each reaction vessel in order to obtain the results described below. Standard calibration curves were generated for HSA and IgG in both single and dual assay formats using pdre standards. Typical curves that can be obtained are illustrated in Figure 6 for HSA and IgG using the dual analyte format. Detection limits vary depending upon the range of standards and the concentration of labeled protein used in the assay. The limits of detection for HSA and IgG in the dual analyteimmunoassay (HSA standards ranged from 0 to 5.0 pg/mL, IgG standards ranged from 0 to 3.0 pg/mL, In3+-labeled HSA was 2.5 pg/ mL, and Bi3+-labeled IgG was 1.3 pg/mL) were 1.8 and 0.6 pg/mL, respectively. These were determined from the signal corresponding to 3 times the standard deviation of the zerodose response subtracted from the zero-dose response. 1064
Analytical Chemistry, Vol. 66, No. 11, June 1, 1994
1 0
.
-
. 1000
. 2000
-
.
-
3000
. 4000
-
I 5000
Nephelometry (mg/dL)
Flgure 7. Correlation plots for HSA )(. and IgG (0)in human serum comparing the dual analyte immunoassay and results obtained by nephelometry. Each data point is the mean of two separate determinations: n = 8.
Analyses of laboratory prepared samples of HSA and IgG using the single analyte immunoassay were compared to those using the Bio-Rad protein assay reagent. There is good agreement between the two methods over the entire range of concentrations (0-1 .O mg/mL), illustrating that the metal ion labeling concept provides a reliable measure of analyte levels in solution: for HSA, n = 10, intercept = +0.03, slope = 0.92, and correlation coefficient = 0.996; for IgG, n = 10, intercept = -0.03, slope = 1.02, and correlation coefficient = 0.986. The voltammograms contained no evidence of protein leached into the release solution which could lead to electrode fouling. One expected effect would be distortion of the stripping waves by adsorbed protein. A second effect would be faradaic waves due to oxidation of the disulfide bonds present in these proteins. Since the protein is likely to adsorb on the electrode, this would be a surface wave and very sensitive to protein. Neither of these effects was observed in release solutions, including blank solutions. Analysis of Serum Samples. Although the purpose of this paper is to demonstrate the concept of this labeling strategy for dual analyte determinations in a general sense, an obvious application would be analyses of samples of biological origin such as serum. Consequently, analyses of serum samples were performed as a preliminary evaluation of the practical applicability of the dual assay format. Human serum samples were analyzed using the dual immunoassay and the results were compared with those obtained by nephelometry. The correlation plots for HSA and IgG are shown in Figure 7. There is generally good agreement between the two methods over the concentration ranges found in the samples: IgG data, correlation coefficient 0.87, standard error of x and y 197; HSA data, correlation coefficient 0.83, standard error of x and y 38 1. Anodic stripping voltammograms for immunoassays performed on serum samples were similar to those shown in Figure 5. They showed no evidence of interference from trace metals found in serum that complexed with HSA or IgG on the surface inside the immunoreactor vessel. Most metal ions in serum are bound to serum proteins, so exchange with the protein on
Table 2. Interassay Preclslon for Slgnals Generated In Single Analyle Immunoassays of Control Sera'
analyte
mean i (nA)
cv ( % )
abnormal serum HSA abnormal serum IgG normal serum HSA normal serum IgG
62.5 70.5 49.4 89.2
10.6 10.4 12.8 9.6
Assays were performed with Bi3+-labeledproteins (n = 10). Table 3. Interassay Preclslon for Slgnals Generated In Dual Analyte Immunoassays of Control Sera' analyte mean i (nA) (%)
cv
abnormal sera HSA abnormal seral IgG normal sera HSA normal sera IgG
45.2 62.5 38.4 75.8
23.1 18.1 26.5 15.7
Assays were performed with BP-labeled IgG and In3+-labeled HSA (n = 10).
the surface inside the immunoreactor vessel may be slow. Loosely bound metals would be washed out by the rinsing steps. Control sera were used to evaluate the interassay precision (CV) for the signals generated in single analyte and dual analyte immuoassays (Tables 2 and 3). The increased variability seen with the dual assay may arise from the use of the polystyrene balls which were necessary to immobilize a sufficient amount of antibody. This created difficulties in solution transfers and might have led to inadequate washing.
CONCLUSIONS We have demonstrated a methodology using metal ion labels and ASV detection that is applicable for the simultaneous determination of proteins by immunoassay. This concept has been illustrated in a dual analyte immunoassay using HSA and IgG as model analytes. The bicyclic anhydride method of coupling DTPA to proteins, which has been demonstrated for radiolabeling of antibodies used in nuclear medicine, works well for labeling our tracer antigens with metal ions. It is a reliable and efficient means of conjugating several stable, but releasable labels per molecule of the model analytes without significantly altering their antigenic properties. Polystyrene as the solid phase permits easy separation of bound and free label and subsequent release of the bound label for detection with no leaching of protein into the final solution. The major drawback to this assay in its present format is the limitations imposed by the hanging mercury drop (HMDE) (26) Nustad, K.; Paus, E.; Bormer, 0. In Immunochemisfry of Solid Phase Immunoassay; Butler, J. E., Ed.; CRC Press: Boca Raton, FL, 1991;Chapter
used in the electrochemical detection system. The concentrations of the metal ion labels that are ultimately found in the release solution are very close to the detection limits of the instrumentation using a 10-min deposition time. The inherent imprecision at these levels leads to difficulty in accurately quantitating analyte levels in complex matrices such as human serum. Additionally, this limits the ability to addressquestions of matrix effects such as nonspecific binding and crossreactivity on accuracy and limits of detection. Various approaches could be taken to improve the detectability of the metal ion labels. One way would be the use of a flow system for stripping analysis based on a mercury film electrode (MFE). These electrodes have become more popular in ASV because of the higher sensitivities and higher resolution they provide compared to the HMDE." A second approach would be to increase dramatically the surface area of the solid phase, thereby immobilizing a larger amount of capture antibody and ultimately higher levels of bound label. An example of how this could be accomplished would be through the use of monosized polymer particles as the solid phase adsorbant.26 Another approach to improving the electrochemical signals would be to increase the number of labels attached per protein molecule. This concept has been demonstrated through modifications of proteins with polychelates for use as contrast reagents in magnetic resonance imaging.27 The methodology used with this dual immunoassay could potentially be applied to a large number of clinically important proteins in human serum and other body fluids. This immunoassay could be expanded to include more than two analytes by the use of other metal ions such as Ga3+ (log Kf = 2 5 ) and T13+(log Kf= 46). These metal ions are detectable by ASV and can be resolved from the In3+ and Bi3+ labels.
ACKNOWLEDGMENT The authors thank the University of Cincinnati for funds made available through the Biomedical Research Challenge and the University Research Council and the University of Cincinnati Medical Center Diagnostic Immunology Laboratory for the human serum samples. Scientific Parentage of the Authors. Fred J. Hayes Ph.D. under W. R. Heineman, Ph.D. under R. W. Murray, Ph.D. under R. C. Bowers, Ph.D. under I. M. Kolthoff; Willliam R. Heineman Ph.D. under R. W. Murray, Ph.D. under R. C. Bowers, Ph.D. under I. M. Kolthoff. Received for review November, 9, 1993. Accepted March 9, 1994."
14. (27) Sieving, P. F.; Watson, A. D.; Rocklage, S.M. Bioconjugafe Chem. 1990, I , 65-69.
e Abstract
published in Aduance ACS Abstracts, April 15,
1994.
Analvtical Chemistty, Vol. 66, No. 11, June 1, 1994
1065
