(in press). 11. Pratt JJ. Steroid immunoassay in clinical chemistry [Re-. viewJ. Chin Chem 1978 ... noassays. In: Masseyeff RF, Albert WH, Staines NA, eds. Meth-.
CLIN. CHEM. 41/9, 1403-1406 (1995)
#{149} Oak
Ridge Conference
Solution-Phase Immunoassay for Determination of Cortisol in Serum by Capillary Electrophoresis Dieter
Schmalzing,
Wassim
Nashabeh,
and Martin
Fuchs’
We describe a competitive solution-phase immunoassay for serum cortisol determination that involves capillary electrophoresis (CE) combined with laser-induced fluorescence for separation and quantification. A polyclonal antibody preparation and fluorescein-labeled cortisol are used as assay reagents. 8-Anilino-1 -naphthalenesulfonic acid was added to the serum sample during the assay to promote the release of cortisol from endogeneous binding proteins. Conditions for the rapid separation of free and bound labeled antigen by CE were developed. Aspects of assay performance are evaluated in this report. The resulting assay protocol allows for the analysis of serum samples without extraction or other sample preparation steps. Indexing Terms: laser-induced fluorescence/endogenous proteins/polyclonal antibody/methods comparison
binding
Capillary electrophoresis (CE) is a high-resolution separation technique in which the separation of analytes is based on differences in their electrophoretic mobility.2 The molecules to be separated migrate in an electric field with velocities that depend on molecular size and net charge. Besides being a powerful separation tool, CE has a number of favorable operating characteristics such as small scale, high speed, and simple instrumentation. CE is therefore being examined as a useful separation method for rapid and efficient immunoassays (1). Initial results on CE-based immunoassays for the measurement of insulin (2), human growth hormone (3), chioramphenicol (4), and opiates (5) have been reported. In previous work (6), we demonstrated feasibility of a CE-based immunoassay for the determination of cortisol in serum. As shown in these reports, CE can provide a number of advantages for performing immunoassays. CE requires only a small volume of material for separation and hence offers very low sample and reagent consumption. Also, a single CE separation replaces multiple wash steps and the signal readout typically required in conventional immunoassays. The CE approach therefore lends itself readily to automation. The high resolving power of CE can in principle allow the simultaneous determination of multiple analytes in a single capillary. Furthermore, an array of channels
PerSeptive Biosystems, Framinghatn, MA 01701. ‘Author for correspondence. Fax 508-383-7883. 2Nonstandard abbreviations: CE, capillary electrophoresis;
AMPD, 2-amino-2-methyl-1,3-propanediol; droxymethyl)methyl-3-aminopropanesulfonic 8-anilino-1-naphthalenesulfonic acid. Received
May 8, 1995; accepted
June
TAPS, acid; 14, 1995.
N-tris(hyand
ANS,
operated in parallel can be used to obtain high sample throughput. Recently, we have shown the ability to perform assay separations in micromachined channels with significantly higher speed than in capillary systems (Frederick Conference on Capillary Electrophoresis, Frederick, MD, Oct. 25-26, 1994). Such systems offer the potential for high throughput because of the ease with which multiple separation channels can be fabricated on a single substrate. On the other hand, the performance of immunoassays in the CE format poses new requirements. To achieve the sensitivity needed in many assays and to prevent interferences with matrix components, fluorescent labeling is desirable. Hence, labeled antibody or antigen must be synthesized and purified. Moreover, the proteins and other components present in biological matrices such as urine or serum present challenges to CE. For reproducible results, adsorption of sample components to the capillary wall must be prevented by the use of separation conditions tailored for these highly proteinaceous sample matrices while also achieving the overall objective of separation of free and bound fractions. The determination of the serum cortisol concentration (typically in the range of 10-600 gfL or 30-1700 nmol/L) is important for the assessment of adrenocortical activity and the diagnosis of adrenal malfunctions such as Addison disease and Cushing syndrome. In this paper, we present results on an improved CE-based immunoassay protocol for the determination of cortisol in serum. In contrast to the earlier study, we now use a polyclonal antibody rather than an Fab fragment. Interference from endogenous binding proteins in the serum with the CE separation has been eliminated.
Materials and Methods Materials Polyimide-coated fused silica capillary columns (50 m i.d., 360 m o.d.) were from Polymicro Technologies (Phoenix, AZ). Cortisol 3-(O-carboxymethyl)oxime, cortisol, cortisol-deficient human serum, 1-ethyl-3[3-(dimethylamino)propyl]carbodiimide hydrochloride, N-hydroxysuccinimide, 2-amino-2-methyl-1,3-propane-
diol (AMPD), and N-tris(hydroxymethyl)methyl-3aminopropanesulfonic acid (TAPS) were supplied by Sigma (St. Louis, MO). Cortisol serum calibrators were obtained from Incstar (Stillwater, MN). 5-[(5-Aminopentyl)thioureidyl] fluorescein (fluorescein cadaverine) was from Molecular Probes (Eugene, OR). Rabbit polyclonal anti-cortisol antibody was purchased from Fitzgerald (Concord, MA). 8-Anilino-1-naphthalenesulCLINICAL CHEMISTRY, Vol. 41, No. 9, 1995
1403
fonic
acid, ammonium
(Milwaukee,
salt
(ANS),
was
from
Aldrich
WI).
2.
Apparatus
ri
antibody
CE separations were performed on fused silica columns coated in the laboratory with a coating similar to the siloxanediol/polyacrylamide coating of Schmalzing et al. (7). The CE system consisted of a P/ACE instrument Model 5500 (Beckman, Fullerton, CA) fitted with an argon laser source. Excitation was at 488 nm and detection at 520 nm.
3.
papaun
-
0+
A
protein
antigen
labelled antigen
Fab
optiona
4.
Qf
+
analyte
+
Procedures sample
of cortisol. Cortisol was labeled with fluorescein according to Pourfarzaneh et al. (8). The purity and integrity of the labeled compound was assessed by thin-layer chromatography, CE, and mass spectromeLabeling
incubation
o
+
o-
+
19.F
+
+
try.
Assay
vials were pipetted 20-FL solutions and 80 L of a solution labeled antigen and 2.75 mmol/L ANS (9) in water. To each was added 80 L of polyclonal anticortisol antibody diluted 1:40 in water. After 30 mm of incubation at room temperature, each sample was directly measured by CE. CE separation. The coated 50-sm columns had a total length of 27 cm and an effective length (to detector) of 20 cm. The samples were pressure-injected for 8 s at the negative electrode. The applied voltage was 30 kV with 20 mmo]/L TAPS/AMPD (pH 8.8) as the separation buffer. The current was - 17 A. Between runs, the column was rinsed for 0.5 miii with 20 mmol/L NaOH and 0.5 mm with buffer. aliquots
Results Assay
protocol.
Into small
of undiluted of 2.5 nmol/L
serum
and DiscussIon Format
Figure 1 schematically shows the format of our CE-based competitive immunoassay. Fluorescein-labeled tracer and antibody are added in specific amounts to the sample to be analyzed. Fab fragments can optionally be used, as will be discussed. After equilibrium is established in the free solution, a small volume is injected into the capillary, whereupon free and bound labeled tracer are separated by CE and quantified by measurement of fluorescence intensity. In principle, both signals can be used for quantification. The competitive format is necessary, because cortisol is a neutral molecule of low molecular mass (362 Da). Its binding to the much heavier antibody (150 kDa) or antibody fragment (50 kDa) does not produce a readily observable mobility difference between free and complexed antibody. In contrast, the free labeled antigen separates readily from the labeled antigen/antibody complex under a variety of conditions. Choice of Antibody for Immunoassay In our previous work on a CE-based immunoassay for cortisol (6) we used an Fab fragment derived from a mouse monoclonal antibody. Fab fragments were pro1404
CLINICAL CHEMISTRY, Vol.
41, No. 9, 1995
5.
CE
(+)()
Of 0-F
I
free
D
=
=
I
complex
0
I
matrix components
Fig. 1. Competitive CE immunoassay scheme.
duced by enzymatic digestion of the intact monoclonal antibody and were further fractionated by HPLC (10) to yield preparations that migrated as single isoforms by CE. When used for the assay, the single Fab isoform results in a signal for the complex of the Fab fragment with the labeled antigen that is sharp and well resolved. This allows the complex signal to be easily quantified and has the further advantage that separation space is conserved for other possible signals, thereby making multianalyte assays feasible. However, significant quenching of the fluorescence on complex formation was observed for the fluorescein label used. For this reason, only the free labeled antigen signal was used for quantification. With the complex no longer being significant for quantification, we decided to investigate the use of intact monoclonal and polyclonal antibodies for this assay. Fig. 2 shows a comparison of the complex signal that results with (a) a single Fab isoform, (b) an intact monoclonal antibody, and (c) a rabbit polyclonal antibody. The greater degree of heterogeneity in the intact monoclonal and particularly in the polyclonal antiserum leads to a broadening of the complex signal. Although quantification of the complex signal becomes progressively more difficult, the free labeled antigen signal is unaffected by the type of antibody so that the choice of antibody is left open subject to the other requirements of the assay. The use of intact antibody eliminates the digestion and purifi-
IS
Ag
*
(0
C 0(
(I)
a a C a, a ‘5 a, 0 U-
1.5
0.0
2.0
Time (mm)
Time (mini Fig. 2. Comparison of the complex signal (C*) produced with three different antibody preparations. Conditions: 8-s pressure injection at the negative electrode; coated fusedsilica column, 27/30cm x 50 pin; pH 8.8(20 mmol/L TAPS/AMPD),30 kV, 17 pA, laser-induced antigen.
cation hence
fluorescence(excitatIon488
nm, emIssion 520 nm). Ag,
steps involved in generating Fab fragments simplifies reagent preparation.
and
CE of Cortisol in Serum Figure 3 shows electropherograms obtained in determinations of serum cortisol within the clinically significant range of 10-600 gfL (0.03 1.7 g.&moLfL)with a polyclonal antibody. At an applied field strength of 1100 V/cm the free labeled antigen was well separated from the complex. In a run time of -2 miii, wellresolved signals are obtained for the labeled cortisol and for an internal standard (fluorescein) used to correct for slight variations in injection volume. The complex does not migrate past the detector in this time frame and is purged from the capillary between runs. There was no noticable dissociation of the complex during the separation, as evidenced by the sharp, symmetrical, labeled antigen peak, which returns cleanly to the baseline. No cleanup or extraction of the serum samples before the CE runs was performed. Despite the high salt and protein content of serum, which can cause problems for CE, the conditions used permit reproducible separations with reliable signals. Migration time reproducibility between both individual runs and different capillaries was 0.5% over months of running assays. The cortisol-releasing agent ANS (9) was added to the serum samples to displace bound cortisol (11). In our earlier work, despite the use of ANS, 1 h or longer -
incubation
was
required
to prevent
severe
broadening
Fig. 3. CE immunoassay profiles with a polyclonal antibody. Conditions as in Fig. 2. IS, internal standard. of the unbound labeled cortisol peak, an effect attributed to cortisol-binding proteins. With the protocol used in this study, i.e., ANS combined with a highaffinity polyclonal antibody (reported affinity constant 1.0 X 1010 IJmol) and the effective ninefold dilution of the serum (vs threefold previously), sharp peaks were obtained on performing CE immediately after mixing the reagents without interference from the cortisolbinding proteins in the serum. CE measurements could then be used to obtain the time course of the binding reactions. Our data on the time course of equilibration showed that 30 min was sufficient to obtain stable signals. Quantification The peak areas of free labeled antigen relative to the internal standard were used to establish the calibration curve for cortisol in serum (a typical curve is shown in Fig. 4). Because of the very high run-to-run reproducibility, normalization for migration time was not necessary. With the antibody and labeled antigen titered as shown, the operating range of the assay is suitably centered on the clinically relevant range of cortisol concentration. A broader operating range was observed with the monoclonal Fab fragments (6), which we attribute to a higher affinity for the labeled cortisol than for the native compound with this antibody (12). Table 1 presents accuracy data for the assay, based on analytical recovery of cortisol added to sera with three different starting cortisol concentrations. The average measured recovery ranges from 56% to 117%, with the best performance at lower concentrations (average recoveries ranged from 96% to 102% for 15 gfL initial cortisol). CLINICAL CHEMISTRY, Vol. 41, No. 9, 1995
1405
With regard to accuracy and precision, better results have been reported for several other cortisol immunoassays (8, 9, 13). With our present experimental setup, the incubation temperature was not well controlled, and the serial analysis of the samples may have introduced variability in incubation time. Nevertheless, our results demonstrate many attractive features of CEbased assays. We believe the speed, microvolume scale, and high separating power of CE show promise for rapid and quantitative determination of analytes of diagnostic relevance in body fluids such a serum.
U)
We thank Lance Koutny, Margaret helpful
Johns,
and Chris Burns for
discussions.
References 1. Afeyan NB, Regnier FE. Analysis utilizing isoelectric focusing.
i#{248}
1cy7
10
10.0
1
Cortisol Conc. (mol/L) Fig.4. Calibration curve for cortisol in serum measured immunoassay.
by the CE
Table 1. Accuracy of serum cortisol determination. Cortisol, pg/L Initial 15 15 15 15
Added 18 89 179 223
Found 18 85 171 228
100
96 96 102 117
18 89
79
89
125
179
179
100
125
223 18 89 179 223
214 21 50 129
96 117 56 72
164
74
300
21
Recovery, %
125
300 300
Initial intraassay precision data on sets of 10 replicate incubations showed CVs of 8.5%, 9.0%, and 16.2% for serum cortisol concentrations of 114, 83, and 38 gIL, respectively.
1406
CLINICAL
CHEMISTRY,
40:1819-22. 6. Schmalzing
Vol. 41, No. 9, 1995
D, Nashabeh W, Yao X-W, Mhatre R, Regnier FE,
Fuchs M. Capillary electrophoresis-based immunoassay for cortisol in serum. Anal Chem 1995;67:606-12. 7. Schmalzing D, Foret F, Piggee C, Carrilho E, Karger BL.
Afeyan
125
300
US Patent 5,376,249, 1994. 2. Schultz NM, Kennedy RT. Rapid immunoassays using capillary electrophoresis with fluorescence detection. Anal Chem 1993;65:3161-5. 3. Shimura K, Karger BL. Affinity probe capillary electrophoresis: analysis of recombinant human growth hormone with a fluorescent labeled antibody fragment. Anal Chem 1994;66:9-15. 4. Blais BW, Cunningham A, Yamazaki H. A novel imniunofluorescence capillary electrophoresis assay system for the det.ermination of chioramphenicol in milk. Food Agric Immunol 1994;6: 409-17. 5. Chen F-TA, Evangelista RA. Feasibility studies for simultaneous immunochemical multianalyte drug assay by capillary electrophoresis with laser-induced fluorescence. Cliii Chem 1994;
NB,
Characterization and performance of a neutral hydrophilic coating for the capillary electrophoretic separation of biopolymers. J Chromatogr 1993;652:149-59. 8. Pourfarzaneh M, White GW, Landon J, Smith DS. Cortisol directly determined in serum by fluoroimmunoassay with magnetizable solid phase. Clin Chem 1980;26:730-3. 9. Brock P, Eldred EW, Woiszwillo JE, Doran M, Schoemaker HJ. Direct solid-phase 1251 radioimmunoassay of serum cortisol. Cliii Chem 1978;24:1595-8. 10. Mhatre R, Nashabeh W, Schmalzing D, Yao X-W, Fuchs M, Whitney D, Regnier FE. Purification of antibody Fab fragments by cation-exchange chromatography and pH gradient elution [Tech Brief]. J Chromatogr (in press). 11. Pratt JJ. Steroid immunoassay in clinical chemistry [ReviewJ. Chin Chem 1978;24:1869-90. 12. Delaage R, Barbet J. Theory of multiple equilibria in immunoassays. In: Masseyeff RF, Albert WH, Staines NA, eds. Methods of immunological analysis, Vol. 1. Weinheim: VCH, 1993: 493-7. 13. Ogihara T, Miyai K, Nishi K, Ishibashi K, Kumahara Y. Enzyme-labeled iminunoassay for plasma cortisol. J Cliii Endocrinol Met.ab 1977;44:91-5.
