Anal. Chem. 1996, 68, 18-22
Microchip Electrophoretic Immunoassay for Serum Cortisol Lance B. Koutny, Dieter Schmalzing, Todd A. Taylor, and Martin Fuchs*
PerSeptive Biosystems, Framingham, Massachusetts 01701
An immunoassay performed using a microchip electrophoretic system is described. Separation and quantitation of free and bound labeled antigen in a competitive assay are carried out in channels micromachined into fused silica substrates. Such microchips are attractive because of their small size, ruggedness, and amenability to automated handling. The assay achieves the determination of cortisol in blood serum over the range of clinical interest (1-60 µg/dL) without the need for extraction or other sample preparation steps. The separation is performed in less than 30 s. Very high throughput is possible by operating the assay in multiple channels in parallel. These characteristics make microchip electrophoretic systems a promising technology for the rapid analysis of clinical samples. Electrophoretic immunoassays based on the separation of free and bound forms of antigen or antibody by capillary electrophoresis (CE) have recently been demonstrated. CE-based immunoassays in various stages of development for the measurement of insulin,1,2 human growth hormone,3 chloramphenicol,4 and opiates5 have been reported. In previous work,6,7 we described a CE-based immunoassay for the determination of cortisol in serum. These assays rely on the high specificity of the antibody/antigen recognition to achieve the necessary selectivity to operate on complex biological samples. High sensitivity is achieved by the use of fluorescent labeling, together with laser-induced fluorescence detection. CE provides rapid and efficient separation and quantitation. Such an approach has several major benefits. A single electrophoretic separation and quantitation achieves results that generally require multiple washing and other steps. Furthermore, fast separations can be achieved by the use of high electric field strengths. Also, CE lends itself readily to automation and requires few moving parts. A significant disadvantage of CE performed in capillaries is the serial nature of the analysis, operating sequentially on individual samples. To overcome this limitation, the use of arrays of capillaries has been explored.8-10 Such arrays allow many (1) Schultz, N. M.; Kennedy, R. T. Anal. Chem. 1993, 65, 3161-3165. (2) Schultz, N. M.; Huang, L.; Kennedy, R. T. Anal. Chem. 1995, 67, 924929. (3) Shimura, K.; Karger, B. L. Anal. Chem. 1994, 66, 9-15. (4) Blais, B. W.; Cunningham, A.; Yamazaki, H. Food Agric. Immunol. 1994, 6, 409-417. (5) Chen, F.-T. A.; Evangelista, R. A. Clin. Chem. 1994, 40, 1819-1822. (6) Schmalzing, D.; Nashabeh, W.; Yao, X.-W.; Mhatre, R.; Regnier, F. E.; Afeyan, N. B.; Fuchs, M. Anal. Chem. 1995, 67, 606-612. (7) Schmalzing, D.; Nashabeh, W.; Fuchs, M. Clin. Chem. 1995, 41, 14031406.
18 Analytical Chemistry, Vol. 68, No. 1, January 1, 1996
separations to be carried out simultaneously and hence enable throughput to be increased by the size of the array. However, the handling of bundles of capillaries poses severe operational challenges for automated systems. An alternate approach to solving this problem is based on adapting micromachining technology to the fabrication of microfluidic systems. Photolithography, combined with chemical etching and wafer bonding, can be used to construct enclosed channels in glass or fused silica substrates suitable for electrophoretic separations.11-13 A major advantage of this approach is that multiple channels of identical geometry are easily generated on a single substrate. This yields closely matched separation systems, integrated in a form that can be readily handled. Furthermore, the overall geometry can be replicated with high precision from substrate to substrate. Other advantages include system design flexibility with regard to channel size and layout and the ability to make channel junctions of various types.14-16 Precise volumes of fluids can be manipulated in these systems for injection of sample as well as for mixing of reagents. The separation of labeled15 and native16 amino acids, oligonucleotides,17 DNA restriction fragments,14 and inorganic ions18 in these kinds of devices has been reported. Microchip electrophoresis of proteins has lagged behind these other applications, since detection has relied on laser-induced fluorescence, and labeling of proteins without introducing heterogeneity is difficult. We recently described direct UV absorbance detection of underivatized proteins separated in channels fabricated in fused silica substrates.19 These studies demonstrate that significantly faster separations can be obtained compared to those done in capillaries because the devices permit the use of short separation lengths with high electric field strength and because the injection methods used result in geometrically small injection plugs. (8) Mathies, R. A.; Huang X. C. Nature (London) 1992, 359, 167-169. (9) Huang, X. C.; Quesada, M. A.; Mathies, R. A. Anal. Chem. 1992, 64, 21492154. (10) Ueno, K.; Yeung, E. S. Anal. Chem. 1994, 66, 1424-1431. (11) Manz, A.; Harrison, D. J.; Verpoorte, E. M. J.; Fettinger, J. C.; Paulus, A.; Lu ¨ di, H.; Widmer, H. M. J. Chromatogr. 1992, 593, 253-258. (12) Harrison, D. J.; Manz, A.; Fan, Z.; Lu ¨ di, H.; Widmer, H. M. Anal. Chem. 1992, 64, 1926-1932. (13) Jacobson, S. C.; Hergenro¨der, R.; Koutny, L. B.; Warmack, R. J.; Ramsey, J. M. Anal. Chem. 1994, 66, 1107-1113. (14) Woolley, A. T.; Mathies, R. A. Proc. Natl. Acad. Sci. U.S.A. 1994, 91, 1134811352. (15) Fan, Z.; Harrison, D. J. Anal. Chem. 1994, 66, 177-184. (16) Jacobson, S. C.; Koutny, L. B.; Hergenro ¨der, R.; Moore, A. W.; Ramsey, J. M. Anal. Chem. 1994, 66, 3472-3476. (17) Effenhauser, C. S.; Paulus, A.; Manz, A.; Widmer, H. M. Anal. Chem. 1994, 66, 2949-2953. (18) Jacobson, S. C.; Moore, A. W.; Ramsey, J. M. Anal. Chem. 1995, 67, 20592063. (19) Koutny, L. B.; Taylor, T. A.; Schmalzing, D.; Nashabeh, W.; Yao, X.-W.; Fuchs, M. HPCE ‘95, Wu ¨ rzburg, Germany, Jan 29-Feb 2, 1995. 0003-2700/96/0368-0018$12.00/0
© 1995 American Chemical Society
We wished to investigate whether a similar speed advantage might be possible for electrophoretic immunoassays performed in a microchip format. Other attractive aspects of microfabricated devices for immunoassays include their small size, ruggedness, and ease of handling. In this paper, we present results on performing a competitive immunoassay for the determination of cortisol in serum using a micromachined electrophoresis system. We demonstrate that the immunochemical measurement of a clinically relevant analyte can be achieved in such a device. EXPERIMENTAL SECTION Micromachining. Microchip separation devices were made using photolithography and chemical etchants to produce channel structures in a fused silica wafer. Access holes were drilled at channel terminals through the etched wafer. A second fused silica wafer was bonded to the etched wafer to produce enclosed channels. After bonding, the wafer was cut into individual separation chips. To perform the microfabrication, a film of chromium (400 Å) was sputtered onto the fused silica substrate (75 mm diameter × 0.4 mm; Hoya, Tokyo, Japan). Photoresist (Shipley 1811, Newton, MA) was spin-coated onto the wafer and baked at 90 °C for 25 min. The resist was patterned by exposing it to UV (365 nm) radiation through a contact-aligned photomask (Advanced Reproductions, Wilmington, MA) and developing it in Microposit developer (Shipley). The chrome was removed using K3Fe(CN)6/ NaOH (Chrome Etch, Shipley). The resulting mask pattern was etched into the fused silica by immersing the wafer in NH4F/HF (1:1) etchant at 50 °C. The depth of etching was controlled by monitoring etching time and measured with a profilometer (Mitutoyo). The microchip utilized in this study was etched to a depth of 28 µm, yielding a channel width of 66 µm at the top of the channel because of the isotropic etching conditions. The cross-sectional area of the channel is equivalent to that of a cylindrical capillary with an i.d. of 44 µm. Photoresist was removed with acetone, and the remaining chrome was dissolved using K3Fe(CN)6/NaOH. Access to the channel terminals was provided by laser-drilled holes through the etched wafer. A second fused silica wafer was bonded to the etched wafer to enclose the channels. Both wafers were immersed in 50 °C NH4OH/H2O2 and rinsed in H2O. They were then placed in contact and thermally bonded. Initial bonding took place at 200 °C (2 h), followed by final bond formation at 1000 °C (overnight). Invidual reservoirs cut from glass tubing were attached with silicone adhesive (Dow Corning). Instrumentation. A fused silica microchip was coupled with a power supply and fluorescence optics for the chip-based assay. High voltage was provided by a Spellman CZE 1000R power supply (Plainview, NY) through a switching circuit and resistor network (though the power supply is current limited, operating precautions should be taken when working with high voltage). Laser-induced fluorescence detection used an Omnichrome (Chino, CA) argon ion laser operating with ∼3 mW of output at 488 nm, focused into the channel at a 53° angle of incidence with a 10 cm focal length lens. A 20× microscope objective (Edmund Scientific, Barrington, NJ) collected fluorescence emission. The collected light was spatially filtered by a 2 mm i.d. aperture in the image plane and optically filtered by two 520 nm bandpass filters (520DF30 Omega Optical, Brattleboro, VT). A photomultiplier tube (Hamamatsu R928, Bridgewater, NJ) connected to a Keithley 614 electrometer (Cleveland, OH) detected the fluorescence signal. The signal was
digitized with a PC-controlled 20 bit data acquisition system (Data Translation 2804, Marlborough, MA) and analyzed using Caesar software (ADI, Alameda, CA). Capillary electrophoresis was carried out on a P/ACE Model 5500 (Beckman, Fullerton, CA), fitted with an argon ion laser source. Excitation was at 488 nm and detection at 520 nm. Capillary separations were performed with capillaries having a modified form of the siloxanediol/polyacrylamide coating of Schmalzing et al.20 Chemicals and Reagents. Ammonium hydroxide, hydrogen peroxide, cortisol, cortisol-3-(O-carboxymethyl)oxime, 1-ethyl-3[3-(dimethylamino)propyl]carbodiimide hydrochloride, N-hydroxysuccinimide, 2-amino-2-methyl-1,3-propanediol (AMPD), and 3-{[tris(hydroxymethyl)methyl]amino}propanesulfonic acid (TAPS) were supplied by Sigma. Hydrofluoric acid, ammonium fluoride and 8-anilino-1-naphthalenesulfonic acid ammonium salt (ANS) were from Aldrich (Milwaukee, WI). 5-[(5-Aminopentyl)thioureidyl]fluorescein (fluorescein cadaverin) was from Molecular Probes (Eugene, OR). Rabbit polyclonal anticortisol antibody was purchased from Fitzgerald (Concord, MA). Cortisol serum standards were obtained from Incstar (Stillwater, MN). Labeling of Cortisol. The labeling of cortisol with fluorescein was performed as in our earlier work.6 The purity and integrity of the labeled compound were assessed by TLC, CE, and mass spectrometry. Assay Protocols. In Water. Twenty microliter aliquots of cortisol standard solutions prepared in water in the range from 10-4 to 10-8 M and 20 µL of fluorescein-labeled antigen (10-7 M) were pipetted into small vials. A 20 µL aliquot of antibody (40× dilution) was added. After 25 min of incubation at room temperature, the samples were analyzed by microchip electrophoresis. In Serum. Twenty microliter aliquots of undiluted serum standards with 1, 4, 10, 30, and 70 µg/dL cortisol and 80 µL of a solution containing 2.5 × 10-8 M labeled antigen and 1.6 mM ANS21 were pipetted into small vials. A 80 µL aliquot antibody (40× dilution) was added. After 30 min of incubation, the samples were measured by microchip electrophoresis. Microchip Electrophoresis. Separations were carried out with a field strength of 800 V/cm in a 20 mM TAPS/AMPD (pH 8.8) electrolyte. The offset pinched injection scheme produced an injection volume of ∼0.75 nL. Capillary Separation. The coated 50 µm 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 field strength was 30 kV, and 20 mM TAPS/AMPD (pH 8.8) was used as the background electrolyte. RESULTS AND DISCUSSION Microchip Design. The design of the chip used in this study is shown in Figure 1. The substrate material is fused silica, which we have found to etch very cleanly, yielding sharp features. The optical properties of fused silica are also favorable. Compared to the designs used by other groups, the channels we fabricated are deeper, a typical channel depth being 28 µm. The chemical etchant that was used attacks the fused silica isotropically, yielding a channel width of twice the depth plus the photomask line width (10 µm), so that the width of the 28 µm deep channel is 66 µm. (20) Schmalzing, D.; Foret, F.; Piggee, C.; Carrilho, E.; Karger, B. L. J. Chromatogr. 1993, 652, 149-159. (21) Brock, P.; Eldred, E. W.; Woiszwillo, J. E.; Doran, M.; Schoemaker, H. J.Clin. Chem. 1978, 24, 1595-1598.
Analytical Chemistry, Vol. 68, No. 1, January 1, 1996
19
Figure 1. Scale drawing of the channel geometry on the microchip.
Figure 3. Schematic drawing of the injection process for the offset pinched injector.
Figure 2. SEM of cleaved edge of chip substrate showing channel cross section.
Figure 2 is a scanning electron micrograph of the cut edge of an etched substrate, providing a cross-sectional view of a channel. The deeper etching means that the optical path length for detection is similar to that in the capillary tubes typically used for electrophoresis and eases the requirements for focusing of the excitation laser. Deeper channels may be expected to lead to increased power dissipation, raising the possibility of a temperature rise in the channel from Joule heating. In our case, the channel cross-sectional area is equivalent to that of a 44 µm i.d. capillary. With the low-conductivity buffer system used for the experiments reported here, the power dissipation is very low (20 mW over the 2.5 cm separation channel), and Joule heating in the chip is therefore insignificant. The design has three arms of 2.5 cm length and a sample arm of 0.5 cm, meeting at an offset junction. Two of the 2.5 cm arms are folded for compactness. The matched lengths provide equivalent resistances in the three arms, while the sample arm is short for rapid loading. For injection, we used a variant of the pinched injector first described by Jacobson et al.13 The arms of the junction were offset by 500 µm to define a larger injection volume. The voltages on the four arms were controlled to pinch off the flowing sample stream, preventing diffusion of sample into the separation channel and providing an injection volume which is independent of sampling time. The effective separation length 20 Analytical Chemistry, Vol. 68, No. 1, January 1, 1996
was 2.2 cm (injector to detector). The details of offset injector operation are shown schematically in Figure 3. The micromachined separation system we fabricated is compact (17 mm × 60 mm) and able to withstand repeated handling. The chip used for this study was operated for hours at a time over several months. After each use, the chip was flushed with water followed by methanol and air-dried. Microchip Assays. The immunoassay we describe here is performed using the competitive format. Antibody and labeled antigen are added in specific quantities to the sample to be analyzed. The sample is incubated during an equilibration step, during which the labeled antigen competes with native antigen (analyte) in the sample for a limited number of antibody binding sites. Thereafter, the free labeled antigen is separated from the complex and quantitated. A previously determined calibration curve is used to relate the amount of free labeled antigen measured to the concentration of the analyte in the sample. For this assay, we used a rabbit polyclonal anti-cortisol antiserum and a fluoresceinated cortisol derivative. With these reagents, the assay separation yields a sharp, well-defined peak for the free labeled antigen and a slowly migrating diffuse band for the complex. Quantitation is performed on the free labeled antigen. The complex signal is only detectable at concentrations well above that used for the assay because of quenching and the broadness of the peak,7 and hence it is not necessary to extend the analysis to include the complex signal. This reduces the analysis time and bypasses the issue of quenching of fluorescence in the complex by the antibody. We added a rapidly migrating internal standard (fluorescein) to compensate for variations in injection volume. For serum samples, the analysis is complicated by the presence of endogenous cortisol binding proteins. Most of the total cortisol in serum is bound to corticosteroid-binding globulin, albumin, and other proteins.22 To allow a determination
a
a
b
b
Figure 4. Immunoassay separations (IS, internal standard; Ag*, labeled cortisol; ordinate in arbitrary units). (a) Two replicate injections in a microchip; sample cortisol concentration, 2.5 × 10-6 M. (b) Single separation in a capillary; sample cortisol concentration, 1 × 10-6 M.
of the total cortisol concentration, the releasing agent 8-anilino1-naphthalenesulfonic acid was added to the serum during incubation. With the use of these reagents, the electrophoretic profile for the assay is shown for two replicate injections in Figure 4a for cortisol samples dissolved in water. For comparison, Figure 4b shows the profile for a similar incubation obtained in a capillary tube. Note the difference in time scale. Three major peaks are observed, corresponding to the internal standard, an impurity from (22) Dunn, J. F.; Nisula, B. C.; Rodbard, D. J. Clin. Endocrin. Metab. 1981, 53, 58-68.
Figure 5. Microchip immunoassay separations. Three replicate injections for each of three samples with cortisol levels as shown. (a) Cortisol in water and (b) serum samples: 800 V/cm; 20 mM TAPS/ AMPD (pH 8.8). The internal standard concentration was higher in the serum samples.
the dye, and the free labeled antigen. The complex is not detected. The overall profile is the same for the two systems, though greater resolution is obtained in the capillary. For the labeled antigen, 6000 plates (400 plates/s) are obtained in the microchip versus 73 000 (810 plates/s) in the capillary. This difference is due largely to the lower voltage (2 kV) used in the microchip compared to the capillary (30 kV). Using the microchip, the separation was accomplished in under 30 s, with resolution adequate to separate the species of interest. Figure 5 shows the variation of the labeled antigen signal with Analytical Chemistry, Vol. 68, No. 1, January 1, 1996
21
the sample cortisol concentration. Replicate injections are shown for three cortisol concentrations spanning the range of clinical relevance, in part a for water samples and in part b for serum samples. The amount of free labeled antigen detected becomes larger as the concentration of cortisol in the sample increases. With the high speed of separation, we examined the feasibility of making multiple determinations for each sample. Since the complex is not detected under the conditions used, we decided to begin subsequent separations without waiting for the complex to migrate out of the channel. This approach proved to be feasible. The migration of the labeled antigen through the slower migrating zone of complex from the previous injection produced no observable effects. We performed more than 100 injections from a single sample in this way, limited only by the eventual depletion of buffering capacity of the separation medium after ∼1 h of continuous operation. The reproducibility of labeled antigen determination for replicate injections was typically 1-2% for water samples and 3-6% for serum samples. Greater variability was observed for the serum samples, which we attribute to the presence of salts and proteins in the serum. Data obtained by performing frontal electrophoresis showed fluctuations in the steady state fluorescence signal (data not shown). Frontal electrophoresis can be expected to reflect the behavior of the sample stream during the injection process. Serum cortisol standards were used to construct calibration curves. A typical curve is shown in Figure 6. The error bars reflect (1 standard deviation based on nine replicate measurements for each incubation. Quantitation based on peak areas produced lower coefficients of variation than when peak heights were used. The position of the calibration curve is determined primarily by the reagent concentrations. With the protocol shown here, the working range of the microchip assay is well centered on the range of clinical interest (1-60 µg/dL or 30-1700 nM), matching the results of our previous study in capillaries using the same reagents.7 Assay precision is influenced by the detection sensitivity. The basic detection sensitivity of the laser-induced fluorescence systems used was comparable for the microchip and capillary formats. The incubation time for the assay is long compared to the separation time. The incubation time necessary for competitive assays can be handled by starting a series of assay incubations at short intervals. Following incubation, the assays are rapidly measured in sequence. A number of commercial clinical analyzers operate on samples in this way. In a microchip system, the
22 Analytical Chemistry, Vol. 68, No. 1, January 1, 1996
Figure 6. Calibration curve for microchip electrophoretic serum cortisol immunoassay. Ratio of free labeled antigen peak area to peak area of internal standard versus cortisol concentration in serum.
incubations could be performed off-chip or in multiple sample wells with subsequent measurement in rapid sequence. We have used the micromachining process to fabricate arrays of electrophoretic separation systems in a single substrate. With the high speed of the individual assay separations that we demonstrate in this report, such arrays can be operated either sequentially or simultaneously to obtain very high sample throughput. Furthermore, in an array format, the relatively long incubation time of the assay limits only the time to the first result and not the overall throughput. As such, these systems show great promise for the rapid processing of clinical samples. We conclude that microchip electrophoretic systems are rugged, compact analytical devices suitable for immunological assays. ACKNOWLEDGMENT We thank Fred Regnier and Noubar Afeyan for their advice and support. Received for review August 15, 1995. Accepted October 10, 1995.X AC9508311 X
Abstract published in Advance ACS Abstracts, November 15, 1995.
