Anal. Chem. 1999, 71, 3150-3156
Operation of Acoustic Plate Mode Immunosensors in Complex Biological Media R. Dahint,* F. Bender, and F. Morhard
Angewandte Physikalische Chemie, Universitaet Heidelberg, Im Neuenheimer Feld 253, 69120 Heidelberg, Germany
Acoustic wave based immunosensors have proven to facilitate the in situ detection of marker-free proteins in real time. However, the vast majority of these studies focuses on the interaction of a single type of antigen with immobilized receptors in pure buffer solutions. In an effort to evaluate the potential of acoustic plate mode immunosensors for operation in more complex biological environments, antigen/antibody reactions have been studied in pure buffer solution, in the presence of cells, and in human serum. It has been observed that the devices do not respond to cell adsorption and that antigen/ antibody reactions can successfully be detected even if a thick layer of cells is deposited on the sensing surface. By varying the frequency of operation, it was shown that the sensitivity of the devices toward nonspecific protein adsorption is reduced at high frequencies of operation. Thus, spurious immunosensor response caused by nonspecific adsorption processes can be suppressed by appropriately selecting device frequency. Using immunoglobulin G with minimum cross reactivity with human serum proteins, antigen/antibody reactions have also been monitored in human serum. While the observed frequency shifts are comparable to those measured in pure buffer solutions, the binding process is accompanied by additional acoustic loss, indicating changes in the viscoelastic properties of the interfacial layer. The separation and identification of proteins from a mixture of similar species is of outstanding importance in medical diagnostics and biotechnology. Based on antigen/antibody recognition, immunoassays have been developed that successfully detect these molecules with high sensitivity and accuracy.1 However, as multiple incubation and washing steps are involved in this technique, it is rather time-consuming and not suitable for on-line monitoring of binding processes. Moreover, it has to be guaranteed that the introduction of labeled molecules, which are a key component of immunoassays, does not affect the behavior of the biochemical system. To overcome these drawbacks, the field of immunosensors evolved. Here, receptors are immobilized on a sensing element that directly responds to the biochemical reaction. Several sensor technologies have been investigated for real-time monitoring of * Corresponding author: (tel) +49-6221-544934; (fax) +49-6221-546199; (e-mail)
[email protected]. (1) Price, C. P.; Newman, D. J. Principles and Practice of Immunoassays; Stockton Press: New York, 1991.
3150 Analytical Chemistry, Vol. 71, No. 15, August 1, 1999
biospecific interactions.2 They may be divided into two classes: label-based and label-free techniques. Label-based techniques, such as fluorescence-based detection methods, facilitate very low detection limits. However, they require additional incubation steps andsjust as for immunoassayssit has to be ensured that the label itself does not impose changes on the behavior of the biosystem. Furthermore, used as an on-line technique, one has to differentiate between labeled molecules that have undergone a specific reaction and those that did not react. Thus, additional washing steps may be favorable, defeating the intention of real-time analysis. Label-free techniques include optical fiber based sensors, biosensors based on surface plasmon resonance (SPR),3 and acoustic wave based sensors.4 While the optical techniques are sensitive to variations in the index of refraction at the sensing surface, acoustic wave sensors respond to changes in the mechanical and acoustoelectric properties at the interface. Thus, different physical parameters are accessible by the different techniques, and an appropriate choice of the sensing element will be dictated by the specific analytical problem. While the optical fiber and SPR-based sensors perform well for many applications, they sometimes require a relatively expensive optical analytical system. Although SPR technology has enjoyed some commercial success as relatively large, laboratorysize instruments, SPR-based microsensors are not yet a reality. On the other hand, acoustic wave devices offer the opportunity to realize low-cost robust sensors in miniature packages. Operated at high frequencies, they are extremely surface sensitive.5 Thus, bulk liquid effects on the sensor response can be drastically reduced. For example, a 300-MHz shear horizontally (SH) polarized acoustic wave penetrates only ∼30 nm into an aqueous solution, which is almost 1 order of magnitude less than the penetration depth of optical techniques utilizing He/Ne light sources.6 Among the various types of acoustic wave sensors, acoustic plate mode (APM) devices can be designed such that the sensing surface and the sensor electrodes are strictly separated, thus avoiding any corrosion problems, facilitating the (2) Go ¨pel, W.; Jones, T. A.; Kleitz, M.; Lundstro¨m, I.; Seiyama, T. Chemical and Biochemical Sensors. In Sensors: A Comprehensive Survey; Go ¨pel, W., Hesse, J., Zemel, J. N., Eds.; VCH: Weinheim, 1991; Vols. 2 and 3. (3) Malmqvist, M. Nature 1993, 361, 186-187. (4) Nieuwenhuizen, M. S.; Venema, A. Mass-Sensitive Devices. In Chemical and Biochemical Sensors: Part I; Go ¨pel, W., Jones, T. A., Kleitz, M., Lundstro ¨m, J., Seiyama, T., Eds.; VCH: Weinheim, 1991; Vol. 2, pp 647-680. (5) Martin, S. J.; Ricco, A. J.; Niemczyk, T. M.; Frye, G. C. Sens. Actuators 1989, 20, 253-268. (6) Stenberg, E.; Persson, B.; Roos, H.; Urbaniczky, C. J. Colloid Interface Sci. 1991, 143, 513-526. 10.1021/ac990119u CCC: $18.00
© 1999 American Chemical Society Published on Web 06/23/1999
derivatization of the sensing surface, and enhancing sensor stability.5 Many publications on immunosensors have appeared in the literature. The vast majority of them focuses on the detection of an isolated type of antigen in pure buffer solutions. In real applications, however, the antigen will be dissolved in a complex biological medium (such as in serum or whole blood) containing a variety of proteins and/or biological cells. All these species may affect the sensor response by specific or nonspecific interaction. It is the scope of this paper to take a first step toward the operation of APM sensors in complex biological media and identify experimental conditions under which these devices can successfully detect immunoreactions. For this purpose, APM sensors on ZXLiNbO3 and -65°Y-90°X quartz have been used to study antigen/ antibody reactions in pure buffer solutions, in the presence of cells, and in human serum. EXPERIMENTAL SECTION Chemicals. Ethanol and H2O2 (30%) were purchased from Baker (Gross-Gerau, Germany), H2SO4 (95-97%) was from Riedelde Hae¨n (Seelze, Germany), and mercaptoethanol and bovine serum albumin (BSA) were from Serva Feinbiochemica (Heidelberg, Germany). Phosphate-buffered saline (PBS), tris(hydroxymethyl)aminomethane (TRIS), and glutaraldehyde were obtained from Sigma Chemie (Deisenhofen, Germany), and (3-aminopropyl)trimethoxysilane was from ABCR (Karlsruhe, Germany). A PBS solution containing 0.1% Tween-20 (Merck, Darmstadt, Germany) is denoted as PBS/Tween-20. Both PBS and TRIS buffer were used at pH 7.5. Affinity chromatography purified goat anti-rabbit immunoglobulin (IgG) (2.3 mg/mL) and rabbit anti-horse IgG peroxidase (POD) conjugate (0.8 mg/mL) were bought as PBS solutions from Jackson Immuno Research, Inc. (West Grove, PA) as well as two IgG solutions with minimum cross reactivity with human serum proteins: goat anti-rabbit IgG (1.8 mg/mL) and rabbit anti-sheep IgG (1.8 mg/mL). Antibodies with the latter degree of purification were used in experiments with human serum. Human serum was obtained from human blood in a Sarstedt monovette (Sarstedt AG, Nu¨mbrecht, Germany) by centrifuging operation at 4000 cycles/min for 10 min. Cell-Tak cell and tissue adhesive was purchased from Becton Dickinson (Bedford, MA). Candida tropicalis (UCD 69-45) and Escherichia coli (strain K12 “wild type”, ATCC 23716) cell cultures were purchased from the Deutsche Sammlung von Mikroorganismen und Zellkulturen (Braunschweig, Germany). Devices. The APM sensors used in the experiments were fabricated on 0.5-mm-thick, optically polished crystals of ZXLiNbO3 and -65°Y-90°X quartz. For the ZX-LiNbO3 devices, a sensor design was selected in which the transmitting and receiving interdigital transducers (IDTs) utilize different geometries.7 By this technique, mode interference at higher frequencies is effectively reduced. The transmitter consists of an IDT with 4 fingers/period (4F-IDT) and a periodicity of λ ) 92.0 µm. For the receiver IDT, a 3F-IDT with λ ) 61.2 µm was chosen. The device is operated at ∼145 MHz, where the second harmonic response of the 3F-IDT and the third harmonic response of the 4F-IDT coincide. (7) Ros Seigel, R.; Harder, P.; Dahint, R.; Grunze, M.; Josse, F.; Mrksich, M.; Whitesides, G. M. Anal. Chem. 1997, 69, 3321-3328.
APM devices on -65°Y-90°X quartz exhibit a wide ranging spectrum of well-separated, SH polarized modes with almost identical insertion loss. Therefore, identical IDT geometries can be selected for transmitter and receiver without provoking substantial mode interference. For sensor design, 4F-IDTs with λ ) 36.0 µm were chosen, resulting in a lowest order mode at ∼93 MHz and a relatively clean APM spectrum up to ∼500 MHz. A detailed description of APM devices on rotated cuts of quartz crystals will be presented elsewhere.8 For the suppression of spurious signals caused by temperature variations, nonspecific adsorption processes, changes in bulk liquid properties, etc., a dual delay line configuration was used in which one acoustic propagation path serves as an internal reference. As LiNbO3 is a high-coupling piezoelectric material, the sensing surface of the ZX-LiNbO3 devices was coated with a 20-nm adhesive layer of chromium followed by deposition of 60 nm of gold in order to eliminate acoustoelectric interactions.9 Due to their much lower piezoelectricity, no such metallization was required for the quartz sensors.5 Both types of devices were mounted with the IDTs on the bottom surface so that the nonelectroded surface will be in contact with the liquid. A small thermometer (Pt 100) was fixed at the lower crystal surface in order to monitor the temperature of the device. Surface Preparation. A liquid-tight, single-chamber Teflon cell was mounted on the nonelectrode surface of the device covering both sensing and reference lines. After the gold surface was cleaned for 15 min in a freshly prepared 1:3 mixture of H2O2 (30%) and H2SO4 (96%) (so-called piranha solution) cooled to 50 °C, the cell was rinsed with ultraclean water (Millipore) and dried in a nitrogen stream. Caution: Piranha solution reacts violently, even explosively, with organic materials! If silane-coated sensors were reused for another measurement, 5 N NaOH was applied for at least 24 h prior to piranha solution treatment in order to remove the films. For covalent coupling of antibodies, the devices were first hydroxylated. This was achieved by exposing the sensing surface for 1 h to a 0.8% ethanolic solution of mercaptoethanol in case of the metallized LiNbO3 devices and to 5 N NaOH for the nonmetallized quartz sensors. After the surface was rinsed with ethanol, both lines were treated with a solution of 2% (3aminopropyl)trimethoxysilane in ethanol for 30 min and washed with ethanol. Now, the device was exposed to a 2.5% solution of glutaraldehyde in PBS. Upon removal of this solution after 30 min, the surface was carefully washed with PBS. Then, by tilting the device, only the sensing line was incubated with 20 µL of goat anti-rabbit IgG dissolved in 1 mL of PBS for 1 h. Finally, the sensing line was intensively washed with PBS/Tween-20 and PBS. To reduce nonspecific protein adsorption on both sensing and reference line, the sensor was tilted back to its original position and the two lines were exposed to 10 mg of BSA dissolved in 1 mL of TRIS buffer for 1 h. After careful rinsing with PBS/Tween20 and PBS, the sensor was ready for the immunosensor experiments. In some experiments, adhesion of cells to gold-coated substrates was promoted by the use of Cell-Tak, a formulation of (8) Bender, F.; Dahint, R.; Josse, F.; Ricco, A. J.; Martin, S. J., submitted. (9) Dahint, R.; Grunze, M.; Josse, F.; Andle, J. C. Sens. Actuators B 1992, 9, 155-162.
Analytical Chemistry, Vol. 71, No. 15, August 1, 1999
3151
Figure 1. Schematic view of the experimental setup. The two propagation paths of the dual delay line configuration can be addressed individually. As the IDTs are located on the lower crystal surface, analyte and electrodes are strictly separated.
polyphenolic proteins extracted from the marine mussel, Mytilus edulis. The adhesion proteins can be immobilized by adsorption from a neutral solution. After the surfaces were cleaned with piranha solution, 7 µg of the proteins was added to 1.5 mL of PBS in order to coat an area of ∼2 cm2. Incubation time was 1 h. APM Measurements. For the experiments, the device is connected to a network analyzer (HP 8752A) which records phase and signal amplitude at the output of the APM sensor relative to the input of the device (Figure 1). Two miniature relays allow one to individually address and monitor the two acoustic propagation paths (i.e., sensing and reference line). Device temperature is monitored by a small thermistor (Pt 100) mounted at the lower crystal surface. A liquid cell is mounted on top of the device and filled with ∼1.5 mL of PBS. To eliminate electromagnetic cross talk and adjacent mode interference, the amplitude frequency spectrum is Fourier transformed into the time domain, where spurious signals are numerically removed by gating. After transforming the data back into the frequency domain, a spectrum with clean and well-separated modes is obtained. As detailed in a previous publication,10 this technique significantly improves the reproducibility of the measurements. Next, one or more distinct modes of the transmission spectrum are selected and the appropriate frequencies of operation are (10) Schumacher, J.; Dahint, R.; Josse, F.; Grunze, M. In Proc. IEEE Ultras. Symp; Levy, M., Schneider, S. C., McAvoy, B. R., Eds.; IEEE: New York, 1994; pp 629-632.
3152 Analytical Chemistry, Vol. 71, No. 15, August 1, 1999
individually determined for both sensing and reference line. As the two propagation paths are almost identical, the deviations in the frequency of operation are usually small (
