Correspondence Anal. Chem. 1996, 68, 1966-1970
Magnetic Permeability Measurements in Bioanalysis and Biosensors Christine Berggren Kriz,†,‡ Kajsa Rådevik,† and Dario Kriz*,†,§
Chemel Research Institute, Research Park IDEON, S-223 70 Lund, Sweden, and Analytical Chemistry and Pure and Applied Biochemistry, Chemical Center, University of Lund, P.O. Box 124, S-221 00 Lund, Sweden
A new transducer concept in bioanalysis and biosensors, utilizing measurements of magnetic permeability, is reported. A model system based on dextran ferrofluid, concanavalin A immobilized to a carrier (Sepharose), and glucose was used to demonstrate the feasibility of this approach. Direct ferromagnetic detection of the dextran ferrofluid was achieved by using a measuring coil (transducer) in a Maxwell bridge. A sensitivity of 21 µV/(µg Fe/mL) and a rsd value of 3.8% were obtained (n ) 5). It was also demonstrated that a small, non-ferromagnetic metabolite (glucose) could be detected using a competitive approach. With an increasing concentration of glucose (20-40 mM), we observed a decrease in the response (0.59-0.11 mV). Reference measurements performed on Sepharose without the biorecognition element, concanavalin A, showed no significant response (0.01 mV). Some potential advantages and drawbacks of this novel type of magnetic transducer are discussed. The advantages include very low interference from the sample matrix, as the transducer is only sensitive to ferromagnetic substances, which rarely are present in samples. In addition, it is suggested that these transducers should be free from fouling. The new transducers are proposed to provide the basis for a new group of affinity biosensors suitable for in vivo and in vitro use. Research on biosensor technology has, during the last two decades, been very intensive, as shown by the huge amount of publications and patent applications submitted worldwide.1 The reasons for these efforts are the potential advantages that such techniques might offer, complementing already established analytical procedures, such as high-performance liquid chromatography (HPLC), bioanalytical procedures such as radiolabeled or enzyme-labeled immunoassays (RIA or ELISA), and spectrophotometric bioassays. Some advantages often discussed in the literature include simplicity and rapidity, reduction of the need for sample pretreatment, low cost of the analytical equipment, avoidance of the use of hazardous chemicals, the wide range of †
Chemel Research Institute. Analytical Chemistry, University of Lund. Pure and Applied Biochemistry, University of Lund. (1) Janata, J.; Jasowicz, M.; DeVaney, D. M. Anal. Chem. 1994, 66, 207R228R. ‡ §
1966
Analytical Chemistry, Vol. 68, No. 11, June 1, 1996
analytes which may be measured, and excellent selectivity and sensitivity.2-4 However, affinity biosensors have shown poor results so far, since the operational principle involves only a binding reaction between the analyte and the recognition element in the biosensor. Thus, no analyte conversion occurs. This makes it necessary to use either various labeling techniques or masschange-sensitive transducers such as potentiometric, optical, and piezoelectric transducers.5 These transducer types are often very sensitive to nonspecific interactions with the matrix. Parallel to the development of biosensors, the potential of biomagnetic separation techniques has increasingly been recognized, mostly because of their efficiency, simplicity, mild operating conditions, and low cost.6 It would therefore appear very attractive to extend the use of magnetic phenomena into biosensor (and bioassay) technology. The novel type of magnetic transducers suggested, when used in combination with ferromagnetic labels, should exhibit very low interference due to the sample matrix, as they would only be sensitive to ferromagnetic substances, which rarely are present in samples. In addition, no fouling of the transducer would be expected to occur. The disadvantages include the laborious preparation of the ferromagnetic labels. Earlier attempts in this direction include measurements of induced currents, caused by ion migration, in nerves.7 The drawbacks of these earlier efforts include the nongeneral nature of the approach, the complex response signal due to the huge number of nervestimulating substances which affect the nerve, and interference from electronic noise in the surroundings. Analytical applications have also been reported using metal ion complexes as probes for nuclear magnetic resonance (NMR) techniques.8 The relative magnetic permeability (µr) is a constant, specific for a given material, which provides a measure of a material’s ability to contain and contribute to an externally applied magnetic field. For most materials, this constant is close to 1, but for some, (2) Turner, A. P. F. Curr. Opin. Biotechnol. 1994, 5, 49-53. (3) Scheller, F.; Schubert, F. Biosensors; Elsevier: Amsterdam, 1992. (4) Karube, I.; Sode, K. In Biotechnology Focus, 2nd ed.; Finn, R. E., Prave, P., Eds.; Hanser Publishers: Munich, Germany, 1990; pp 175-197. (5) Sethi, R. S. Biosens. Bioelectron. 1994, 9, 243-264. (6) Ugelstadt, J.; Berge, A.; Ellingsen, T.; Schmid, R.; Nilsen, T. N.; Moerk, P. C.; Stenstadt, P.; Hornes, E.; Olsvik, O. Prog. Polym. Sci. 1992, 17, 87161. (7) Babb, C. W.; Coon, D. R.; Rechnitz, G. A. Anal. Chem. 1995, 67, 763769. (8) Aime, S.; Botta, M.; Ermondi, G.; Fasano, M.; Terreno, E. Magn. Reson. Imaging 1992, 10, 849-854. S0003-2700(95)01227-3 CCC: $12.00
© 1996 American Chemical Society
As a model system, concanavalin A (Con A) immobilized to a carrier (Sepharose) was chosen as the biorecognition element for a magnetic affinity biosensing device. Con A has previously been used as the recognition element in potentiometric affinity biosensors.10 It belongs to the family of lectins and is frequently used as a model system with a high affinity for various sugars, such as glucose and dextran. The interactions with glucose and dextran have been used in an optical transducer-based competitive assay.11 We chose to demonstrate the new magnetic transducer principle in both the direct detection mode and the competitive mode. As an analyte in the direct mode we used a ferromagnetic dextran ferrofluid, and in the competitive mode the same dextran ferrofluid was used as a ferromagnetic competitor with a non-ferromagnetic analyte (glucose).
Figure 1. Three different approaches to detect an analyte by specific interaction with a carrier-bound recognition element, followed by a separation step and measurement of changes in magnetic permeability using a coil: (a) Direct detection. The analyte itself is ferromagnetic (M). (b) Sandwich detection. Non-ferromagnetic analytes require the use of ferromagnetic markers (M). (c) Competitive detection. The non-ferromagnetic analyte and a ferromagnetic competitor (M) compete for the binding sites on the carrier-bound recognition element.
such as elemental iron, nickel, cobalt, gadolinium, and manganese, it is very high. These materials are called ferromagnetic. In addition to the aforementioned elements, phases, chemical compounds, alloys, and semiconductors containing of these elements are also ferromagnetic. Three main approaches for bioanalysis based on the principle of detecting changes in the magnetic permeability are proposed (see Figure 1). These approaches were suggested earlier by our group9 but are now practically demonstrated for the first time. The easiest approach utilizes a direct detection. This is possible if the analyte per se is ferromagnetic. In the second approach, a non-ferromagnetic analyte is detected in a sandwich manner. One epitope of the analyte interacts with the carrier-immobilized biorecognition element, and a second epitope interacts with a ferromagnetic marker. For small, non-ferromagnetic metabolites lacking a second epitope, a competitive procedure is the method of choice. In this case, a suitable ferromagnetic competitor is used. This approach could also be applied with macromolecules. The magnetic permeability of a material inside a coil influences the inductance of the coil. It is thereby possible to detect changes in magnetic permeability using inductance measurements. As an example, the inductance for a long coil is described by eq 1:
L ) (µrµ0A/l)N2
(generally inside the coil)
(1)
where µr represents the relative magnetic permeability of the material in the coil, µ0 ) 4π × 10-7 V s A-1 m-1 is the permeability of a vacuum, A is the cross section area, l is the length of the coil, and N is the number of turn on the coil. To measure the inductance, and thus indirectly the relative magnetic permeability, the coil can be placed in a Maxwell bridge. We report here a new magnetic transducer for bioanalysis and biosensors, based on the measurement of magnetic permeability. (9) Kriz, D.; Kriz, C. B. Swedish patent application SE 9502902-1, 1995; pp 1-16.
EXPERIMENTAL SECTION Chemicals. Con A Sepharose CL-4B and Sepharose CL-4B were purchased from Pharmacia Biotech (Sweden). Dextran ferrofluid was manufactured according to previously described procedures.12 The iron content was determined by atomic absorbance measurements to be 10 mg/mL, and the dry weight content was 78 mg/mL. The size distribution of the particles in the dextran ferrofluid was examined by gel chromatography and was found to be heterogeneous and in the range of 20-200 nm. All other chemicals used were of analytical grade. Binding Procedure. Before use, the Con A Sepharose gel was washed 10 times with 5 mL of binding buffer (0.02 M tris(hydroxymethyl)aminomethane hydrochloric acid, pH 7.4, and 0.5 M sodium chloride) to remove the storage buffer (0.1 M sodium acetate, 1 M sodium chloride, 1 mM calcium chloride, and 0.01% sodium azide) from the gel. Incubations were performed in test tubes containing 500 µL of the gel suspension (Con A Sepharose or Sepharose), binding buffer, and various amounts of dextran ferrofluid and glucose. The total volume of the suspension was 5 mL. After overnight incubation, the tubes were centrifuged at 3500g for 5 min, and then the supernatant was removed and measured for turbidity. The gel was washed with binding buffer and quantitatively removed into a measuring vessel, which was made of glass (5 mm diameter × 10 cm) and specially designed to be inserted into the measuring coil. Turbidity Measurements. After incubation with Con A Sepharose or Sepharose (in the direct or the competitive detection mode), the dextran ferrofluid content in the supernatant was measured. This was performed using an Ultraspec III spectrophotometer from Pharmacia LKB (Sweden) at a wavelength of 400 nm. The amount of bound dextran ferrofluid was calculated by recording the difference in the turbidity of the supernatant before and after incubation and assuming that this concentration difference had accumulated in the gel phase. A linear relationship was observed between the turbidity (at 400 nm) and the dextran ferrofluid concentration (0-150 µg Fe/mL). Magnetic Permeability Measurements. The transducer used in this work measures changes in the magnetic permeability of materials and comprises a coil which is a part of a balanced Maxwell bridge, with two variable resistances. These are needed because both phase and amplitude have to be balanced. A sinusoidal wave of frequency 200 kHz and amplitude of 2 Vp-p is (10) Janata, J. J. Am. Chem. Soc. 1975, 97, 2914-2916. (11) Schultz, J. S.; Sims, G. Biotechnol. Bioeng. Symp. 1979, 9, 65-71. (12) Wikstro ¨m, P.; Flygare, S.; Gro¨ndalen, A.; Larsson, P.-O. Anal. Biochem. 1987, 167, 331-339.
Analytical Chemistry, Vol. 68, No. 11, June 1, 1996
1967
Figure 2. Diagram of the transducer setup. Measuring coil L4 (transducer) has a length of 11 mm, a radius of 10 mm, and 30 turns of closely packed wire and is part of a balanced Maxwell bridge. The other components in the Maxwell bridge are R1 ) 2860 Ω (10 kΩ trimpotentiometer), C1 ) 1 nF, R2 ) 180 Ω, R3 ) 27 Ω (100 Ω trimpotentiometer), R4 ) 0.44 Ω, and L4 ) 4.9 µH. A sinusoidal wave of 200 kHz and 2 Vp-p is fed into the bridge. The voltage difference measured over the bridge is further processed by a differential operational amplifier circuit, rectified, and finally recorded.
fed into the bridge. The bridge is balanced if the following equations are fulfilled: L4 ) R2R3C1 and R4 ) R2R3/R1. For the values of the electrical components in the bridge, see the legend to Figure 2. The voltage difference measured over the Maxwell bridge is further processed by a differential operational amplifier circuit and rectified. The final output signal (in millivolts) is recorded by a chart writer. The introduction of ferromagnetic materials inside the coil causes an increase in the voltage difference over the Maxwell bridge. RESULTS AND DISCUSSION Characterization of the Magnetic Transducer. Different approaches may be utilized for the determination of the inductance of a coil. These include, for instance, resonance frequency measurements of a parallel circuit containing the coil and a capacitor, various bridge measurements, transient measurements, inductive coupling between two coils, and impedance measurements. We initially attempted resonance frequency measurements, but this method gave poor results. The sensitivity observed was about 100-1000 times less than that obtained using the Maxwell bridge approach. The output signal from the Maxwell bridge was proportional to the concentration of the ferromagnetic dextran ferrofluid solution (expressed in µg Fe/ mL) that was placed within the measuring coil. Measurements on 1.5 mL buffer solutions (0.02 M tris(hydroxymethyl)aminomethane hydrochloric acid, pH 7.4, and 0.5 M sodium chloride) containing increasing concentrations of ferromagnetic dextran ferrofluid (0-50 µg Fe/mL) yielded a sensitivity of 21 ( 4 µV/ 1968 Analytical Chemistry, Vol. 68, No. 11, June 1, 1996
(µg Fe/mL). A relative standard deviation (rsd) of 3.8% was obtained from five measurements on samples containing 30 µg Fe/mL. We therefore decided to use this approach for the subsequent experiments. A commercially available dextran ferrofluid (D-8517, Sigma) was also tested. However, this was not ferromagnetic because it contained the iron in the form of iron hydroxide, Fe(OH)3, and not as magnetite, Fe3O4. Consequently, no response was observed when this dextran ferrofluid was placed within the measuring coil. Kinetics of the Dextran Ferrofluid Binding. To study the kinetics of the dextran ferrofluid binding to Con A Sepharose, the turbidity of the supernatant was followed. The decrease of the dextran ferrofluid concentration in the solution was obtained, and the accumulation in the Con A Sepharose gel could thus be calculated. As a reference, Sepharose without immobilized Con A was used. No accumulation of dextran ferrofluid in the gel could be detected in this control experiment. In Figure 3, the concentration of bound dextran ferrofluid in the Con A Sepharose gel is plotted against time. As can be seen from the graph, the binding increases in a logarithmic manner. The kinetics was found to be slow, and in the subsequent experiments, incubation times of 20 h were used. It has been reported13 that Con A binds low molecular weight carbohydrate ligands with a second-order rate constant of 104-105 M-1 s-1. This is lower than would be expected for a diffusion-controlled process but does not explain the long (13) Goldstein, I. J.; Poretz, R. D. In The Lectins [Properties, Functions, and Application in Biology and Medicine]; Liener, I. E., Sharon, N., Goldstein, I. J., Eds.; Academic Press: London, 1986; pp 60-61.
Figure 3. Kinetic study of the binding of dextran ferrofluid. Con A Sepharose (500 µL), dextran ferrofluid (56 µg Fe/mL), and binding buffer (0.02 M tris(hydroxymethyl)aminomethane hydrochloric acid, pH 7.4, with 0.5 M sodium chloride) were used. The total volume was 5 mL. The supernatant content of dextran ferrofluid was followed during a 20 h period by measuring the turbidity at 400 nm. The accumulation in the Con A Sepharose was calculated and plotted.
Figure 4. Response (in mV) obtained with a Maxwell bridge as a function of various initial concentrations of dextran ferrofluid (expressed in µg Fe/mL) incubated with Con A Sepharose (upper curve) and with Sepharose (lower curve).
binding times observed here. However, the pores of the Sepharose are only about 300 nm in diameter, which can be compared to the 20-200 nm dextran ferrofluid particles. Thus, the slow kinetics observed is not surprising. By using other immobilization matrixes for Con A and smaller dextran ferrofluid particles, it should be possible to significantly reduce this problem. Direct Detection of a Ferromagnetic Analyte. Figure 4 shows the response obtained from the Maxwell bridge as a function of the initial concentration of the ferromagnetic model analyte (dextran ferrofluid) in a sample solution. As expected, saturation is achieved for higher concentrations. To investigate
whether the observed increase in response was due to the specific interaction between Con A Sepharose and dextran ferrofluid, the Con A Sepharose was exchanged for Sepharose. In this case, no response could be observed, as can be seen from the lower curve in Figure 4. Both the supernatants (using turbidity measurements) and the gel phases (Con A Sepharose and Sepharose) were analyzed. The results obtained were compared and found to be almost identical (data not shown). These experiments served as a demonstration of a direct detection of ferromagnetic substances using a magnetic transducer in combination with a biorecognition element, such as might be applied in a bioassay and biosensor. In this particular case, ferromagnetic dextran ferrofluid and Con A Sepharose were used. However, ferrofluids are commercially available for biomagnetic separations in immunodiagnostics, molecular and cell diagnostics, and therapeutics. Such materials could easily be qualitatively and quantitatively analyzed using the described direct detection approach. The direct detection approach has, due to the small amount of ferromagnetic substances of interest, less potential compared to the sequential and the competitive detection modes. Here, almost any analyte, from small metabolites to nucleic acids, enzymes, proteins, and cells, can be detected. Thus, we wished also to demonstrate the use of the competitive detection mode. Competitive Detection of Glucose. Glucose and dextran ferrofluid were used in a competitive detection mode with Con A Sepharose as the biorecognition element. These experiments show that it is possible to measure a non-ferromagnetic analyte (glucose) by using a ferromagnetic competitor (dextran ferrofluid). First it was necessary to choose a fixed concentration of the competitor. The results from the direct detection of dextran ferrofluid were used. The highest sensitivity for the competitive detection was expected to be achieved using a ferrofluid concentration in the first part of the curve, where the slope is highest (Figure 4). For too low concentrations, though, increased errors were expected in the measurements. Therefore, an initial concentration of dextran ferrofluid of 42 µg Fe/mL was chosen for the subsequent competition experiments. The results obtained from the competition experiments are shown in Figure 5. When no analyte (glucose) was present in the sample, a high value of the magnetic permeability of the Con A Sepharose was observed. Upon increasing the concentration of glucose in the sample, the response decreased, due to the competition between glucose and dextran ferrofluid for the binding sites on Con A Sepharose. No significant response was observed when Sepharose was used in the reference measurements (lower curve). The competition curve shown in Figure 5 has a very sharp response decrease in the concentration range 20-40 mM. This dynamic range could potentially be used for the determination of glucose. By changing the binding conditions (such as pH or ion strength) or the initial concentration of dextran ferrofluid, the concentration interval for the dynamic range could be altered. In competitive assays in general, the differences in affinity constants for the analyte and competitor determine the detection interval of the analyte. CONCLUSIONS In this study, we have shown the feasibility of using magnetic permeability measurements in bioanalysis and biosensors. This approach should also be useful for the investigation of other recognition element-analyte systems (antibodies/antigens, recepAnalytical Chemistry, Vol. 68, No. 11, June 1, 1996
1969
to calculate its concentration. Obviously, the results presented in this preliminary investigation need to be supplemented by assays carried out on more complex and realistic samples. Work is presently being carried out in our laboratory in this direction. The main advantages of detection using magnetic permeability measurements are its simplicity compared with classical immunoanalysis methods, the possibility of real-time monitoring of the recognition element-analyte complex formation, the low or nonexistent interference from the sample, and the fact that hazardous radiolabels or unstable enzyme labels are not required. Further improvements in instrumentation and the use of smaller dextran ferrofluid particles should increase the sensitivity and lower the response time.
Figure 5. Response obtained with the Maxwell bridge in the competitive detection mode. Dextran ferrofluid (competitor) and glucose (analyte) competed for the binding sites on Con A Sepharose (upper curve). Reference experiments were performed using Sepharose (lower curve).
ACKNOWLEDGMENT The dextran ferrofluid used in this work was a kind gift from Dr. P.-O. Larsson, Pure and Applied Biochemistry, University of Lund, Sweden. Furthermore, the authors thank Dr. Richard Ansell for linguistic advice.
Received for review December 19, 1995. Accepted March 21, 1996.X AC951227T
tors/peptides) or DNA/DNA hybridization studies. The data obtained can be used to indicate the presence of an analyte and
1970
Analytical Chemistry, Vol. 68, No. 11, June 1, 1996
X
Abstract published in Advance ACS Abstracts, May 1, 1996.
