Use of External Magnetic Fields To Reduce Reaction Times in an

Anal. Chem. 2004, 76, 1715-1719

Use of External Magnetic Fields To Reduce Reaction Times in an Immunoassay Using Micrometer-Sized Paramagnetic Particles as Labels (Magnetoimmunoassay) Richard Luxton,† Jasvant Badesha,† Janice Kiely,‡ and Peter Hawkins*,†

Faculty of Applied Sciences, and Faculty of Computing, Engineering and Mathematical Sciences, University of the West of England, Bristol BS16 1QY, United Kingdom

The paper presents a rapid immunoassay system capable of quantifying analyte in complex biological and environmental media. Antibody-coated micrometer-sized paramagnetic particles are used as labels in an assay in which they bind quantitatively with an analyte and capture antibody molecules immobilized on a polyester disk to form a sandwich assay. The assay is performed in a simple reaction vessel, and reactions between labels, analyte, and antibodies are accelerated by positioning magnets alternately above and below the vessel. The bound paramagnetic particles are quantified using a simple flat, spiral, coil located just below the polyester disk. The electronic circuitry associated with the coil uses components that are inexpensive and readily available. The coil has been designed to respond only to particles bound on the disk and not to unbound particles still in the test solution. Unbound particles are pulled away from the disk by the magnet before readings are taken. The use of the reaction vessel with the cardiac markers CRP and CKMB is described. No sample preparation or washing step is used in the assays, and results can be obtained in less than 3 min after introducing the sample into the vessel with sensitivities in the normal clinical range. Immunoassays make use of the highly specific and sensitive interactions between antigens and their antibodies. Over the last 30 years, they have found applications in many diverse areas including clinical chemistry1 and environmental monitoring.2 The high specificity and sensitivity arises from the nature of the interactions between the antigens, which are usually the analyte molecules, and their antibodies. With a few exceptions (such as techniques based on surface plasmon resonance,3 vibrating devices,4 etc.), a label, reporter, or marker must be added if the interactions between the antibodies and the antigens are to be quantified. Several different types of labels have been used in * Corresponding author. E-mail: [email protected]. Fax: ++44 117 32 82904. † Faculty of Applied Sciences. ‡ Faculty of Computing, Engineering and Mathematical Sciences. (1) Anderson, D. J.; Guo, B.; Xu, Y.; Ng L. M.; Kricka, L. J.; Skogerboe, K. J.; Hage, D. S.; Schoeff, L.; Wang, J.; Sokoll, L. J.; Chan, D. W.; Ward, K. M.; Davis, K. A. Anal. Chem. 1997, 69, 165R-229R. (2) Richardson, S. D. Anal. Chem. 2003, 75, 2831-2857. (3) Mullett, W. M.; Lai, E. P. C.; Yeung, J. M. Methods 2000, 22, 77-91. 10.1021/ac034906+ CCC: $27.50 Published on Web 02/07/2004

© 2004 American Chemical Society

immunoassays. Popular labels currently used include radioisotopes, fluorescent and chemiluminescent molecules, enzymes, gold particles, and colored latex beads. Lateral flow or immunochromatographic assays are relatively simple immunoassay techniques using colored latex beads, an example of which is the very successful Clear Blue one-step pregnancy test originally developed by Unilever. It is also an example of a device that with no sample preparation can produce the result of an analysis in a few minutes without the intervention of a skilled operator. There is a requirement for more such devices for use in the field in environmental monitoring or for rapid diagnostics such as point-of-care testing. However, simple lateral flow devices are restricted to applications where it is only necessary to determine whether the concentration of the analyte is above or below a threshold value. If more quantitative measurements are required, then elaborate measuring equipment has to be used, usually with different markers, and this adds considerably to the complexity of the technique. An example of this is a lateral flow system for C-reactive protein (CRP),5 which uses a fluorescent label and a laser system to scan the strips at the end of the assay. Other techniques, such as the enzyme-linked immunosorbent assay, cannot be readily adapted into rapid, portable, systems as they involve many time-consuming steps (sample separation, washing, and incubation) and methods using luminescent labels need optically pure test solutions to reduce light absorption, scattering, and fluorescence quenching effects. Washing steps are necessary to remove excess, unreacted (free) label and other unwanted components in the sample (free proteins, excess antibody, etc.) that are likely to interfere in the determination of the amount of bound label in the immunoassay. Coated paramagnetic particles (PMPs) are currently used in the purification and isolation of antibodies, antigens, and other proteins and more recently in immunoassays where they serve to isolate the target antigen within the interacting molecules. PMPs are available in a range of different diameters (typically, 0.1-20 µm) from several different suppliers. They have a paramagnetic core (usually iron oxide) with a suitable coating to which capture antibody molecules are attached. In a typical automated system,6 an excess of the coated PMPs is introduced into a test solution (4) Benes, E.; Gro¨schl, M.; Burger, W.; Schmid, M. Sens. Actuators, A 1995, 48, 1-21. (5) Ahn, J. S.; Choi S.; Jang, S. H.; Chang, H. J.; Kim, J. H.; Nahm, K. B.; Oh, S. W.; Choi, E. Y. Clin. Chim. Acta 2003, 332, 51-59.

Analytical Chemistry, Vol. 76, No. 6, March 15, 2004 1715

containing a mixture of the target antigens (analyte) and other species. Eventually, the antigens become attached to the antibodies on the PMPs. At this, or a later stage, an excess of the appropriately labeled secondary antibody (detecting antibody) is introduced into the test solution, which then reacts with the target antigen (analyte) on the PMPs. The PMPs with their attachments can be easily drawn to one side of the reaction vessel and held there by an external magnet while the remaining test solution is flushed away and replaced with clean buffer solution. The PMPs are resuspended in the buffer solution and the quantity of bound label determined using a technique appropriate for the label. Although the PMPs help to simplify the extraction and washing processes, several steps might still be required, which add to the total time for a complete assay. Another problem is that expensive equipment is usually required to determine the amount of bound label (such as photon-counting equipment for luminescent labels, radiation-counting equipment for radioactive labels, etc.). In previous papers, we have described an immunoassay system that uses the PMPs as the labels, so no additional label is required,7 and a simple magnetometer to determine the quantity of bound PMP labels.8 In this paper, this work is developed further and we describe a simple reaction vessel that uses two external magnets to speed up the reactions between the antigens and the antibodies giving rise to a rapid measuring system requiring no sample preparation or washing stages. The use of the system with the clinically important molecules, CRP and creatine kinase isoenzyme MB fraction (CKMB) is described. METHOD In our earlier papers,7,8 we described a sandwich immunoassay system for human transferrin in which the PMP labels were immobilized quantitatively on a strip of plastic at the end of the assay and a magnetometer of our design determined the quantity of immobilized PMPs. In this paper, we describe a one-step, rapid, sandwich, magnetoimmunoassay system based on a simple reaction vessel containing all of the necessary reagents for the assay including the coated PMPs (Figure 1). A magnetometer coil with a flat spiral design is used to detect the captured PMPs, and there is also an arrangement for applying external magnetic fields using two permanent magnets. The sample and PMPs labeled with the secondary antibody (detecting antibody) are introduced into the vessel, and at the end of the assay, the PMPs are captured quantitatively via the primary antibody immobilized on a polyester (poly(ethene terephthalate)) disk at the bottom of the vessel. The disk is held in place by a screw cap and sealed by a Neoprene O-ring so that a used disk could be easily removed and replaced by a fresh one. The clinically important biochemical markers CKMB and CRP are used to illustrate applications of the system. Figure 2 shows schematically the composition of the sandwich assay immobilized on the polyester disk at the end of the assay for CRP. Although Figure 2 shows only one antibody/antigen/ antibody bond linking the PMP to the polyester disk, electron microscope investigations show that many cross-linking bonds are (6) Clements, J. A.; Forrest, G. C.; Jay, R. F.; Jeffery, M.; Kemp, P. M.; Kjeldsen, N. J.; Rattle, S. J.; Smith, A. Clin. Chem. 1992, 38, 1671-1677. (7) Richardson, J.; Hawkins, P.; Luxton, R. Biosens. Bioelectron. 2001, 16, 989993. (8) Richardson, J.; Hill. A.; Luxton, R.; Hawkins, P. Biosens. Bioelectron. 2001, 16, 1127-1132.

1716 Analytical Chemistry, Vol. 76, No. 6, March 15, 2004

Figure 1. Diagram of the reaction vessel with the detecting coil, the moveable gantry, and the magnets.

Figure 2. Schematic diagram showing the components of the sandwich immunoassay for CRP.

actually formed between an individual PMP and the disk. The sandwich assay for CKMB has a similar arrangement. MATERIALS AND PREPARATIONS Activation of the Polyester Disks. Polyester disks (15 mm diameter and 1 mm thick) were first washed for 1 min in methanol at room temperature, irradiated with ultraviolet radiation (wavelength, 312 nm) for 10 min, and then washed again in methanol. Meanwhile, a solution of polymerized glutaraldehyde was prepared by mixing 5 mL of 5% glutaraldehyde (Sigma-Aldrich) with 500 µL of 0.1 M NaOH, leaving at room temperature for 30 min, and then neutralizing the solution by adding 0.1 M HCl. The polyester disks were placed in the polymerized glutaraldehyde solution for 30 min and then washed with methanol for 1 min. Immobilization of CRP or CKMB Antibody on the Activated Polyester Disks (Capture Antibody). Rabbit anti-human CRP or goat anti-human CKMB (Randox Laboratories Ltd.) was diluted 1:500 times in 0.1 M bicarbonate buffer (pH 9.7) containing 2% methanol and 0.5% glutaraldehyde. A 35-µL sample of the solution was applied to each activated polyester disk and allowed to incubate for 4 h at room temperature. The disks were washed in 50 mM phosphate-buffered saline, pH 7.4 (PBS). Unreacted active sites were blocked by placing the disks in a PBS solution containing 1% bovine serum albumin (BSA; Sigma-Aldrich) and 1 M glycine (BDH) for 1 h to prevent nonspecific binding. Immobilization of CRP or CKMB Antibody on the PMPs (Detector Antibody). Rabbit anti-human CRP antibody (DAKO

Figure 3. (A) Block diagram of the phase-locked loop circuitry used to determine the resonant frequency fn of the coil. (B) Diagram showing the construction of the flat spiral coil and illustrating how the electromagnetic field produced by the coil is affected mainly by the paramagnetic particles immobilized on the polyester disk and not by free particles in suspension in the buffer.

A/S) or monoclonal anti-human CKMB antibody (Randox Laboratories Ltd.) was diluted 1:1000 times in 0.1 M phosphate buffer (pH 8.1). A 10-µL sample of the PMPs (protein G-coated 2.8-µmdiameter Dynabead M-280 from Dynal UK Ltd.) was washed 3 times in phosphate buffer. It was then mixed with 30 µL of the appropriate antibody solution and incubated for 10 min at room temperature in a slowly rotating sample mixer (Dynal UK Ltd.). The PMPs were washed 3 times in phosphate buffer before being resuspended in 20 mM dimethyl pimelimidate (Sigma-Aldich) and incubated at room temperature with rotational mixing for 30 min to promote cross-linking. The reaction was stopped by suspending the PMPs in 50 mM tris(hydroxymethyl)aminoethane buffer (pH 7.3). Unoccupied sites on the PMPs were blocked using PBS/ BSA as described earlier. Standards of CRP and CKMB in Serum. Standards were prepared by diluting stock solutions of human CRP in serum (160 µg/mL, DAKO A/S) and human CKMB in serum (10 µg/mL, Randox Laboratories Ltd.) in PBS/BSA to give a range of solutions with known concentrations. Design and Construction of the Magnetometer Sensing Coil. The magnetometer coil had a flat spiral design. It was mounted in a hole cut in the center of a small platform made from fiberglass board. A rigid and self-supporting coil was made by winding 40 turns of 0.04-mm-diameter enameled copper wire (Comax) on a plastic former, coating the wire with cyanoacrylate glue as it was wound, and then removing the former when the glue had set hard. The finished coil had an internal diameter of 0.5 mm. In operation, the reaction vessel was placed on the platform with the polyester disk next to the coil. The presence of PMPs on the disk causes the inductance, L, of the coil to increase. (Figure 3) The coil forms a resonant circuit with a capacitor, C, connected in parallel. The increase in L produces a decrease in resonant frequency of the LC circuit from f0, the resonant

frequency for a blank disk, to fn, the resonant frequency for a disk with n PMPs immobilized on it. We have shown that the decrease in frequency is directly related to n; i.e., f0 - fn ) Kn, where K is a constant that depends on the construction of the coil.9 In these experiments, f0 is ∼5 MHz and the decrease in resonant frequency (f0 - fn) is relatively small (typically, 10-1000 Hz). The magnetometer uses a phase-locked loop (PLL) circuit to measure reliably this frequency decrease. Figure 3A shows a block diagram of the PLL circuit, which is described in more detail in ref 9. At resonance, the current flowing to a LC circuit is in phase with the applied voltage. A voltage-controlled oscillator drives the LC resonant circuit, and a phase detector determines the difference in phase between the applied voltage and the current. The phase detector feeds a dc error signal back to the oscillator through a low-pass filter, thus ensuring that the output frequency of the oscillator is always locked onto fn. The PLL circuit design uses inexpensive, readily available, components. The frequency fn was read by a meter (Tecstar FC2500) which was connected to a personal computer via a RS232 link so that the variation of fn with time could be displayed on the computer screen. A flat spiral design was adopted for the coil after numerous experiments with several different designs of coil and reaction vessel. This design was used because it has a simple arrangement for the reaction vessel with the polyester disk forming the base. Another advantage for this design of coil is that it produces an electromagnetic field that decreases rapidly with distance away from its surface (Figure 3B). Thus, PMPs immobilized on the disk are close to the coil and have a strong effect on the inductance, whereas free PMPs further away in the buffer solution do not. An approximate value for the variation in the magnetic field Hz along the central axis, z, of the coil can be determined by applying Biot-Savart’s law10 to the coil. For a circular coil with a single turn of radius a, the magnetic field produced by the coil carrying a constant current decreases with z by

Hz ) ka2/(z2 + a2)1.5

(1)

where k is a constant that depends on the current passing through the coil and the magnetic permeability of the medium surrounding the coil. A tightly wound, spiral coil consisting of m turns may be approximated to m concentric coils with each successive coil increasing in radius by w, the diameter of the wire. As the same current passes through each turn of the coil, the total field produced by the spiral coil is

Hz )

∑ k(a + {m - 1}w) /[z 2

2

+ (a +{m - 1}w)2]1.5 (2)

where the summation is from m ) 1 to m. This equation shows that the magnetic field strength falls rapidly with distance along the central axis of the coil. An experiment was conducted in air to investigate the change in sensitivity of the magnetometer to PMPs immobilized on a polyester disk as the disk was moved away from the detecting coil. The decrease in resonant frequency of the coil was noted when the disk was placed on the coil. A (9) Hawkins, P.; Luxton, R.; Macfarlane, J. Rev. Sci. Instrum. 2001, 72, 237242. (10) Bleaney, B. I..; Bleaney, B. Electricity and Magnetism; Oxford University Press: London, 1963; Chapter 5, p 132.

Analytical Chemistry, Vol. 76, No. 6, March 15, 2004

1717

Figure 4. Decrease in resonant frequency of the detecting coil as a polyester disk coated with paramagnetic particles is moved away from the coil in air. The points are the experimentally determined values, and the line is from a theoretical study as described in the text.

small ring-shaped cardboard spacer of known thickness was now placed between the coil and the polyester disk and the decrease in resonant frequency again noted. This experiment was repeated using different thicknesses of cardboard spacers, and the results are plotted in Figure 4. These results show that, in air, the sensitivity of the magnetometer to the PMPs falls rapidly with distance from the coil. The line drawn through the experimental values in Figure 4 is an attempt to show how well eq 2 predicts the fall in sensitivity of the magnetometer with distance between the PMPs and the detecting coil. Equation 2 was applied by taking the experimentally determined value for the decrease in resonant frequency with the disk place on the coil (i.e., the PMPs are 1 mm away from the coil) and using eq 2 and the dimensions of the coil to determine how the decrease in resonant frequency changes from this value as the disk is moved away from the coil.. An assumption here is that the decease in resonant frequency of the coil behaves in a way similar to the decrease in magnetic field produced by the coil. Figure 4 indicates that, for measurements made in air, eq 2 describes reasonably well how the magnetometer is less sensitive to PMPs that are further away from the coil than ones that are closer. It is recognized, however, that eq 2 must be an oversimplification of the true situation because it applies strictly to the magnetic field along the axis of the coil and provides no information about the field off the axis of the coil: it is likely that, away from the central z-axis of the coil, the magnetic field decreases much more rapidly with distance from the coil so that the lines of force run close to the surface of the coil as illustrated in Figure 3B, which means that PMPs immobilized on the disk interact much more strongly with the magnetic field than those further away. In addition, eq 2 applies only to the magnetic field produced by the coil and ignores the electric field perpendicular to the magnetic field produced by the high-frequency current passing through the coil. When the reaction vessel (Figure 1) is placed on the coil, the electric field will be affected by the permittivity of the polyester disk and its capture layer before it reaches the PMPs, which accounts for the small, and reproducible, fall in resonant frequency observed when a blank disk with no PMPs is placed on the coil. The electric field penetrating into the test buffer solution will be strongly affected by the relatively high permittivity of the solution and its low electrical impedance. When 1718

Analytical Chemistry, Vol. 76, No. 6, March 15, 2004

these effects on the electrical field are taken into account, the sensitivity of the magnetometer coil to PMPs suspended in the buffer is likely to fall much more rapidly than in air as shown in Figure 4. The distance between the top of the liquid in the reaction vessel and the coil is ∼4.5 mm (Figure 1), and according to Figure 4 with air in the vessel, the magnetometer coil is ∼8 times more sensitive to PMPs immobilized 1 mm from the coil on the polyester disk than to unbound PMPs at the top of the vessel. For reasons given above, this relative sensitivity is likely to be much more than 8 with a buffer solution in the vessel. Unfortunately we were unable to determine a value for the relative sensitivity in a buffer solution because we could not perform a simple experiment like the one described above to determine the decrease in sensitivity of the magnetometer to PMPs in suspension in a buffer solution at different distances from the coil with the equipment available. The conclusion of these experiments is that, with this coil design, the unbound PMPs only have to be pulled to the top of the test solution using a magnet and they will not interfere with the readings of the bound PMPs on the polyester disk so their complete removal from the solution is not essential. Use of Two External Magnets To Accelerate the Immunoassay. Two button magnets (10 mm in diameter and 5 mm high from Farnell in One) were mounted above and below the reaction vessel on a small plastic gantry (Figure 1) that could be moved vertically in both directions through a rack-and-pinion arrangement driven by a small electric motor. This mechanism alternately moved the lower magnet (A) up toward the polyester disk and then the upper magnet (B) down toward the top of the reaction vessel. Optoelectronic switches positioned near the ends of the rack (not shown in Figure 1) stopped the motor when the magnets were at certain predetermined distances from the reaction vessel. A programmable timer circuit controlled the motor and the time each magnet spent near the reaction vessel. Alternately moving the magnets toward the reaction vessel in this way considerably reduced the total time for the immunoassay as illustrated in Figure 5A, which shows an assay for a sample containing 4 ng/mL CKMB. A polyester disk coated with the CKMB capture antibody was fitted to the reaction vessel and 0.8 mL of PBS/BSA introduced into the vessel. The motor was set in motion and the magnets took 36 s to execute a complete cycle. Modulations were observed in the resonant frequency of the coil that reflected the positions of the magnets relative to the coil. A higher frequency was observed when magnet A was close to the coil and a lower frequency when it was further away. On a part of the magnet cycle when magnet A was closest to the coil, 50 µL of the PMPs labeled with CKMB antibody and 50 µL of the CKMB standard were introduced into the reaction vessel. Magnet A attracted the PMPs through the test solution to the surface of the antibody-coated polyester disk. Antibody-coated PMPs that have reacted with CKMB antigen form an immune complex on the surface of the polyester disk with the antigen acting as a bridge between the detector and capture antibodies (Figure 2). The presence of PMPs on the disk caused a decrease in the resonant frequency of the coil. On the next part of the cycle, magnet A moved away from the disk, magnet B moved to the top of the reaction vessel, and unbound PMPs were attracted to the top of the test solution. As explained earlier, the resonant frequency of the coil was now an indication of the number of PMPs in the assay

Figure 5. Resonant frequency of the detecting coil before and after adding (A) 20 µL of a serum sample containing 40 ng/mL CRP and (B) 50 µL of a serum sample containing 4 ng/mL CKMB to the reagents in the reaction vessel. Each recording shows how the alternating magnets modulate the resonant frequency of the coil and how the reactions are speeded up as unbound coated PMPs are pulled to and from the polyester disk. The decrease in frequency f0 - fn is directly related to the number of magnetic particles immobilized on the polymer disk.

bound on the polyester disk, because the unbound PMPs near the top of the solution did not affect the resonant frequency of the coil. In subsequent up and down movements of the magnets, more PMPs interacted with the immobilized antibody through cross-linking reactions with the antigen. The interactions were probably aided by the optimum orientation of the reacting sites as the PMP moved through the buffer. These additional PMPs that became cross-linked to the polyester disk produced a further decrease in the resonant frequency. The process continued until all of the CKMB molecules in the test solution had been bound to the antibody-coated PMP or all the available binding sites on the polyester disk had been saturated and no further decrease in the resonant frequency was observed. In most cases, this point was reached in just four cycles of the magnets, so the assay reached completion in 2-3 min after adding the sample. This contrasts very favorably with our previous study using PMPs in the absence of magnets with human transferrin,7 where an overnight incubation at 4°C was required to produce a measurable (11) Agewall, S. Clin. Biochem. 2003, 36, 27-30. (12) Go ¨lbasi, Z.; Uc¸ ar, O ¨ .; Keles, T.; Sahin, A.; C¸ agli, K.; C¸ amsari, A.; Diker, E.; Aydodu, S. Eur. J. Heart Failure 2003, 4, 593-595.

Figure 6. Decrease in resonant frequency f0 - fn plotted against concentration for (A) CKMB and (B) CRP in the serum samples.

response. Figure 6A shows a graph of the decrease in resonance frequency (f0 - fn) plotted against the concentration of CKMB in the sample. The response of the assay starts to level off for CKMB concentrations above ∼10 ng/mL, which is probably caused by saturation of the available capture sites on the coated polyester disks. The detection limit for CKMB was ∼2 ng/mL. The normal level for CKMB as a cardiac marker is less than 3.5 ng/mL,11 so this response fits very well with a clinical application for measuring CKMB directly in blood. Experiments with CRP (Figure 5B) in the reaction vessel produced a similar modulated output from the magnetometer coil with the response leveling off for CRP concentrations greater than 20 ng/mL (Figure 6B) and a detection limit of ∼3 ng/mL. The normal level for CRP is 2.4 ( 0.18 µg/ mL,12 so in this application, the test samples would have to be diluted by ∼1:100 to bring them into the working range of the instrument. The technique we have described is a generic one that could be adapted for a number of similar assays. We are currently working on a simpler version of the apparatus that uses two fixed electromagnets to vary the applied magnetic field instead of the permanent magnets on the moving gantry. ACKNOWLEDGMENT The authors gratefully acknowledge the financial support for this project from Randox Laboratories Ltd., County Antrim, U.K. Received for review August 4, 2003. Accepted January 13, 2004. AC034906+ Analytical Chemistry, Vol. 76, No. 6, March 15, 2004

1719