Anal. Chem. 2003, 75, 6813-6819
Electrophoresis-Assisted Active Immunoassay Victor N. Morozov* and Tamara Ya. Morozova
Beckman Institute for Biomedical Research, 28835 Single Oak Drive, Temecula, California 92590, and Institute of Theoretical and Experimental Biophysics, Russian Academy of Sciences, Pushchino, Moscow region, Russia 142290
An active assay can be defined as that in which diffusioncontrolled reactions are replaced by active delivery of analytes to probe molecules. The present paper describes an electrophoresis-assisted version of an active ELISA performed in tubes or wells with a dialysis membrane attached to their bottoms. The permeability of such a membrane to small ions allows us to apply electric field perpendicular to the membrane surface and to rapidly transport and concentrate charged macromolecular analytes in its vicinity. Probe molecules were either adsorbed or covalently linked to a modified surface of a membrane from regenerated cellulose. An active assay was performed both in separate cells and in 96-well microplates. It was demonstrated that the active assay format allows one (i) to reduce assay time to minutes instead of hours, (ii) to increase sensitivity by a factor of 10-300, and (iii) to capture within 10 min up to 70% of all the analyte molecules present in 0.36 mL of solution. Heterogeneous immunoassay procedures are widely used in clinical diagnostics, analysis of food and water contamination, and fundamental studies of proteins. Some of these applications (e.g., emergency analysis, rapid monitoring of pathogens in the environment) would sufficiently benefit from acceleration of such assays and from an increase in their sensitivity. It has been shown1-4 that in heterogeneous assays the rate of signal growth is often limited not by an association rate constant but by the rate of diffusion of analyte molecules through an unstirred layer. Several approaches to overcome such a diffusion limitation have been described. Thus, intensive agitation accelerates signal growth 5-7 times.2 Assay in a flow-through system comprising a porous substrate5 or a bed of packed functionalized beads6 is another way to reduce the diffusion limitations. While solving the diffusion problem, these techniques cannot, however, overcome the natural sensitivity limit determined by the dissociation constant of the probe-analyte complex. * Corresponding author. Fax: 909-695-0562. E-mail: vmorozov@ beckmaninstitute.org. (1) Butler, J. E. In Immunochemistry of Solid-Phase Immunoassay; Butler, J. E., Ed.; CRC Press: Boca Raton, FL, 1991; pp 3-26. (2) Franz, B.; Stegemann, M. In Immunochemistry of Solid-Phase Immunoassay; Butler, J. E., Ed.; CRC Press: Boca Raton, FL, 1991; pp 277-283. (3) Myszka, D. G.; He, X.; Dembo, M.; Morton, T. A.; Goldstein, B. Biophys. J. 1998, 75, 583-594. (4) Stenberg, M.; Nygren, H. J. Immunol. Methods 1988, 113, 3-15. (5) Poulsen, F.; Bjerrum, O. J. In Immunochemistry of Solid-Phase Immunoassay; Butler, J. E., Ed.; CRC Press: Boca Raton, FL, 1991; pp 251-259. (6) Hayes, M. A.; Polson, N. A.; Phayre, A. N.; Garcia, A. A. Anal. Chem. 2001, 73, 5896-5902. 10.1021/ac034733o CCC: $25.00 Published on Web 11/11/2003
© 2003 American Chemical Society
Since the limit of sensitivity is determined by the minimum detectable fraction of occupied probe molecules, it can be increased by employing more efficient labeling techniques and more sensitive detection methods. Another way to increase sensitivity was developed by Nanogen, Inc. (San Diego, CA). In their method, charged DNA or protein molecules are concentrated by a local electric field in the vicinity of microelectrodes, thus highly accelerating formation of specific complexes and boosting sensitivity.7-9 However, this remarkable technology has serious drawbacks. As the reaction between antigen and antibody goes in a thin layer of a gel deposited directly on microelectrodes, protons, hydroxyls, and other products of electrochemical reactions may affect the reaction, and local changes in pH may result in recharging the analyte ions and pulling them off the active layer. To diminish such effects, special low-conducting buffers need to be used and the time of reaction has to be limited to 1-2 min.7-9 Here we describe an improved version of an active electrophoresis-assisted immunoassay in which a semipermeable membrane is used to concentrate analyte molecules and to immobilize probe molecules. Electrodes are spatially separated from the reaction chamber so that electrochemical products cannot reach the membrane and the electrophoretic cell within the assay time scale. EXPERIMENTAL SECTION Materials. Goat anti-human IgG and goat anti-rabbit IgG conjugated with alkaline phosphatase, rabbit antibodies to hen egg white lysozyme and to ovalbumin, and goat anti-human IgG conjugated with aequorin were products of Chemicon International Inc. (Temecula, CA). Dichlorodimethylsilane, Sephadex G-25, and NaN3 were obtained from Fluka Chemie (Buchs, Switzerland). All other reagents and a dialysis membrane (MWCO 12.4 kDa) were from Sigma-Aldrich Co. (St. Louis, MO). NUNC MicroWell plates were purchased from Nalge Nunc Int. (Rochester, NY). The 96-well MultiScreen filtration plates were from Millipore Corp. (Denver, MA). Design of Cells for Electrophoretically Assisted ELISA. Two types of cells were used in this study. In the first one, a dialysis membrane was glued to the bottom of a commercial (7) Sosnowski, R. G.; Tu, E.; Butler, W. F.; O’Connell, J. P.; Heller, M. J. Proc. Natl. Acad. Sci. U.S.A. 1997, 94, 1119-1123. (8) Edman, C. F.; Raymond, D. E.; Wu, D. J.; Tu, E.; Sosnowski, R. G.; Butler, W. F.; Nerenberg, M.; Heller, M. J. Nucleic Acids Res. 1997, 25, 49074914. (9) Ewalt, K. L.; Haigis, R. W.; Rooney, R.; Ackley, D.; Krihak, M. Anal. Biochem. 2001, 289, 162-172.
Analytical Chemistry, Vol. 75, No. 24, December 15, 2003 6813
Figure 1. Schematic of apparatus for electrophoresis-assisted ELISA in 96-well microplate.
96-well MultiScreen filtration plate in which the microfilter membrane was removed with a sand paper. The plate was treated for 20-40 s in a low-pressure discharge in air using a home-built rf plasma apparatus10 with a power of 20-30 W. Dialysis membrane was also treated in the plasma discharge for 30-60 s. Cyanoacrylate glue was evenly distributed over the edges of the wells. The membrane was firmly pressed to the edges and allowed to settle for 5-7 min. The glued membrane was cut between the wells to reduce tension. In some experiments, separate electrophoretic cells were used. These were prepared from the standard 0.6-mL polypropylene microcentrifuge tubes in which conic bottom parts and caps were cut off with a blade. Dialysis membrane was either glued to the upper part of the tube as described above for the microplate or squeezed between the edges of the tube and the cap. In the latter case, a hole was punched through the cap to enable electric contact with electrode buffer. Apparatus To Run Electrophoretically Assisted ELISA. The apparatus to run the ELISA in a 96-well microplate is schematically presented in Figure 1. It consists of two electrode chambers, 0.75 L each. Electrodes are made as grids of parallel Pt wires placed at the bottom of the lower chamber and on plastic strips protruding from the covering plate into the upper chamber. An electrophoretic 96-well plate prepared as described above was placed into the lower chamber in such a way as to have wells plunged into electrode solution almost entirely. Each well was filled to the top with analyte solution, avoiding bubbles. Another similar microplate was used as an array of sockets to electrically connect the upper chamber with the wells of the electrophoretic plate. A dialysis membrane glued to a plastic frame was placed under an angle in the lower electrode chamber, as illustrated in Figure 1, to deflect bubbles formed on the electrode during EP and to prevent their accumulation on the electrophoretic plate. In some experiments, solution in the lower chamber was subjected (10) Morozov, V. N.; Morozova, T. Y.; Hiort, C.; Schwartz, D. C. J. Microsc. 1996, 183 (Pt. 3), 205-214.
6814
Analytical Chemistry, Vol. 75, No. 24, December 15, 2003
to intensive stirring using a stir bar placed on a perforated plastic pedestal (not shown in Figure 1), protecting the lower Pt electrode from damage. The bubble deflector was removed in such cases. Electrophoretically Assisted ELISA. Coating was performed for 1 h at room temperature or overnight at 4 °C from 50 µL of 5-20 µg/mL coating proteins dissolved in different buffers. After washing 4 times with TBS-T buffer (20 mM Tris-HCl, pH 7.5, 0.15 M NaCl, 0.05% Tween-20), the wells were blocked for 20 min in 3% nonfat milk dissolved in TBS, rinsed 2 times with water, and filled with analyte solution. In most cases, 5 mM Gly-Gly buffer, pH 8.5, with 0.1% Tween-20 (GG-T) was used to prepare the analyte solutions. In some experiments other 5-20 mM buffers (imidazole, borate, Tris, MES, Gly, acetate) and other blocking substances (0.5% casein, 0.1% Hb, 0.1% BSA, 1% dialyzed defatted milk) were used in EP. Such low concentrated buffers allow application of high voltage needed for rapid EP concentrating without generation of excessive heat. The plate was then installed into a holder at the top of the lower electrode chamber. The upper electrode chamber with the connecting 96-socket plate was placed over the electrophoretic plate and filled with the electrode buffer to the level providing electric contact with the Pt wires on the cover plate. EP was performed at a constant current (typically 1.5-2 mA/ cell) with voltage of 100-300 V applied for 10 min in most cases. After electroconcentrating, the wells were washed 4 times with TBS-T, rinsed 2 times with water, and filled with another solution to perform the second EP cycle if indirect ELISA was run. After 4 final washes with TBS-T, the wells were filled with 0.1 mL of the standard pNPP solution (prepared from Sigma Fast tablet set) and incubated at room temperature for 5-30 min with intensive (∼1500 rpm) agitation. A standard microplate reader (MR 700, Dynatech Laboratories, Inc.) was used to measure adsorption at 405 nm. Quantification of light emission from aequorin was made with a microtiter plate luminometer (Labsystems Luminoscan). Immobilization of Proteins on Regenerated Cellulose. Different techniques were tested to increase protein binding to the dialysis membrane. Activation of the Membranes with Plasma. Both edges of commercial dialysis tubing were cut off with a scissor, and two pieces of the membrane were separated with pincers. Each piece was placed into the plasma chamber with internal side up and treated for 1 min in plasma discharge as described above. The same plasma treatment procedure was also used to activate membranes glued to microplates or attached to separate EP cells. Hydrophobization in DDS Vapor. Plasma-treated dialysis membrane (free or glued to a microplate) was placed in a plastic bag filled with dry nitrogen. A weak stream of nitrogen was passed over a droplet of DDS placed in a 1.5-mL Eppendorf tube and then introduced into the bag. The membrane was kept for 5-7 min in a nitrogen atmosphere containing DDS vapor. NHS/EDC Activation. Carboxylate groups formed in plasma discharge on the membrane surface were activated for 7-10 min in a freshly prepared solution of 0.2 M EDC and 50 mM NHS in water.11 After quick removal of the solution, the surface was covered with a coating solution. PEI/GA Activation. Plasma-treated membrane was rinsed with water and placed for 20 min into a freshly prepared 0.2% PEI (11) Johnsson, B.; Lo¨fås, S.; Lindquist, G. Anal. Biochem. 1991, 198, 268-277.
Figure 2. Time course of antigen-antibody reaction in active and passive assay. DDS-treated dialysis membrane was coated in hIgG solution (20 µg/mL in 10 mM acetic buffer, pH 5.0) for 1 h. Goat antihIgG-AP was diluted 40 000-fold with 5 mM Gly-Gly, pH 8.5, 0.1% Tween-20. A 0.2- and 0.46-mL aliquot of this solution was placed into each EP cell for passive (empty triangles and squares) and active assay (filled circles and squares), respectively. The values presented on the right ordinate are calculated from OD measured after 1-h reaction with pNPP in the dark.
solution with the pH adjusted to 9.2. The membrane was thoroughly washed with water, treated for 10 min in 0.2% glutaraldehyde dissolved in 0.15 M phosphate buffer, pH 7.5, and then washed again with the phosphate buffer and water. Borohydride Treatment. Reduction of carbonyl groups on plasma-treated membrane was performed for 1-2 h in 10 mg/ mL solution of NaBH4 in water. The membrane was thoroughly washed with water afterward. Titration of Carboxylate Groups. Membrane sample, 2425 cm2, was cut into thin strips 1-2 mm wide and protonated in HCl solution (pH 2) for 20-30 min. The strips were thoroughly washed with water and placed into 3 mL of 1 M KCl solution bubbled with nitrogen until at constant pH. Changes in pH were titrated back with 10 or 100 mM NaOH solutions under nitrogen. Conductivity Measurements. Probes, 3-4 µL, were taken from different depths in an EP cell. Conductivity was measured with a conductometer (YSI model 32 conductance meter, Yellow Spring Instruments Co., Inc., Yellow Spring, OH) using a homebuilt cell consisting of two platinized Pt wires, 0.1 mm in diameter, inserted into a PTFE tube with an internal diameter of 2 mm. The cell was calibrated with KCl solutions. Measurements of Electrophoretic Mobility. Mobility was estimated from distribution of AP activity in 2-µL aliquots taken from different depths in an electrophoretic cell after EP of antihIgG-AP solution for 2-4 min. RESULTS AND DISCUSSION Protein Concentration and Assay Acceleration in an Electric Field. Figure 2 presents an example of a dramatic acceleration of antigen-antibody reaction in active ELISA. Whereas intense agitation increases the rate only ∼7 times (in accordance with literature data2), an electric field accelerates the same reaction 1000-fold. It is seen in Figure 2 that a stronger field results in higher acceleration. Comparison of active and passive ELISAs could be performed at a short time scale characteristic for EP-assisted assay or in a long-term assay typical for standard ELISA. When compared at identical 10-min incubation time, active ELISA turns out to be ∼250 times more sensitive, as seen in Figure 3A. Of course, no
Figure 3. Comparison of signals in active and passive ELISAs. (A) Direct ELISAs in separate EP cells at identical (10 min) time of antigen-antibody reaction without agitation. Dialysis membrane was treated and coated with hIgG as in Figure 2. Stock of anti-hIgG-AP was diluted with GG-T. EP was made with 1.5 mA/cell (empty circles). In the passive ELISA (filled circles), identical cells were filled with the same solutions and the assay was performed in a similar way but with no voltage applied. Mean values of controls without coating are subtracted. (B) Indirect standard ELISA on a NUNC microplate (filled circles) and EP-assisted ELISA on DDS-treated membrane (empty circles). Both microplates were coated from 10 µg/mL solution of Ova in 0.15 M carbonate buffer, pH 9.5 at 4 °C overnight. The EP-assisted binding of primary rabbit anti-Ova and secondary antirIgG-AP (diluted 20 000-fold) was performed from GG-T buffer at 2 mA/well for 10 min. In the standard ELISA, each binding stage was performed for 1 h from the antibody solutions in 3% defatted milk in TBS with intensive (∼1500 rpm) agitation. Mean values of controls without the primary antibody are subtracted.
equilibrium is established during such a short time in the passive assay, which normally takes hours.1,12 It is hard to compare active and passive assay protocols in a long-run experiment since the limited buffering capacity of electrode buffers does not allow us to run EP for hours. So, we could only compare long-term standard ELISA with short-term active one. Even under such conditions, unfavorable for active ELISA (shorter time and a low-salt buffer, (12) Ekins, R. P.; Chu, F. In Principles and Practice of Immunoassay; Price, C. P., Newman, D. J., Eds.; Macmillan Reference Ltd.: London, 1997; pp 627648.
Analytical Chemistry, Vol. 75, No. 24, December 15, 2003
6815
which inhibits the antigen-antibody reaction by a factor of 5-7), the latter remains ∼10 times more sensitive, as was found in many experiments. Results of one such experiment are presented in Figure 3B. How does EP accelerate the reaction and increase assay sensitivity? One can think of two mechanisms: (i) concentrating the molecules in the vicinity of the membrane surface and (ii) a direct delivery to the binding sites on the surface without concentration. A simple criterion, γ ) V/kassnCs (where V is the velocity of analyte motion to the membrane, Cs is the surface concentration of active molecules bound to the membrane surface, and kassn is the association rate constant) can be introduced to distinguish between the situation when analyte is delivered more rapidly than consumed (γ . 1) and when the delivery presents a bottleneck for the binding process (γ , 1). Our experiments have shown that anti-hIgG-AP molecules move through the GG-T buffer with an average velocity of V ) 3 mm/min, when a 1.5-mA current passes through the cell. Assuming that our membrane binds as much active protein as a typical hydrophobic surface, ∼50 ng/ cm2 (see ref 13), we estimate that Cs ∼ 3.3 × 10-13 M/cm2 for a protein with a molecular weight of 150 000. Even for the highest value of kassn, ∼1 × 106 M-1 s-1, known in the heterogeneous immunoassay,1 γ is ∼15, indicating that electroconcentrating should be the primary event in the EP process under our conditions. The slightly concave form of the curves describing signal increase in the active ELISA in Figure 2 may indicate that some electroconcentrating really precedes the binding. Electroconcentrating presents the most important difference between EP assay and the flow-through assay, where the assay is accelerated by forcing analyte solution through pores in a membrane5 or a layer of beads coated with capturing antibodies.6 EP not only allows acceleration of the assay but also makes it more sensitive since at the same initial concentration a larger fraction of capturing molecules becomes occupied in EP due to the higher local analyte concentration. This concentrating occurs directly at the reaction place, quickly and selectively, since neutral molecules, low molecular ions, and even macromolecules with the opposite charge will not be concentrated unlike solvent evaporation, which concentrates all the substances in solution. What level of concentration could be achieved in the EP assay? In the absence of binding protein, concentration will increase in the vicinity of a semipermeable membrane until a stationary distribution is reached when the electrophoretically driven flux of analyte ions is counterbalanced by a diffusion flux in the gradient of concentration. Such stationary concentration is described by the equation14
C(x) ) βLCo exp(-βx)
(1)
Here β ) V/D ) EeZ/kT, where x is the distance to the membrane surface, k is the Boltzmann’s constant, T is the temperature, and (13) Balcells, M.; Klee, D.; Fabry, M.; Ho ¨cker, H. J. Colloid Interface Sci. 1990, 220, 198-204. (14) Chin, C.; Dense, J. B.; Warren, J. C. J. Biol. Chem. 1976, 251, 3700-3705. (15) Schwesinger, F.; Ros, R.; Strunz, T.; Anselmetti, D.; Gu ¨ ntherodt, H. J.; Honneger, A.; Jermutus, L.; Tiefenauer, L.; Plu ¨ ckthun, A. Proc. Natl. Acad. Sci. U.S.A. 2000, 97, 9972-9977. (16) Hinterdorfer, P.; Baumgartner, W.; Gruber, H. J.; Schilcher, K.; Schindler, H.; Proc. Natl. Acad. Sci. U.S.A. 1996, 93, 3477-3481.
6816 Analytical Chemistry, Vol. 75, No. 24, December 15, 2003
Figure 4. Profile of conductivity of the GG-T buffer in EP cell after passing a current of 1.5 mA for 10 min. To keep this current constant, voltage was changed from 180 to 230 V when positive potential was applied to the electrode at the lower chamber (empty circles). No voltage changes were required to keep constant current at the negative potential at the bottom electrode (filled circles). Dashed line presents conductivity in the initial buffer solution.
e is the elementary charge. For the height of the cell, L ) 12 mm, the diffusion coefficient, D ) 4 × 10-11 m2/s (according to ref 4), and the measured velocity, V ) EeZD/kT ) 3 mm/min in the electric field of E ∼ 1500 V/m (at I ) 1.5 mA/cell), one can estimate that β ) V/D ) 1.25 × 106 m-1 for molecules the size of IgG. These molecules will be concentrated by a factor of C(x)0)/ Co ) βL ) 15 000 at the bottom of a cell in approximately τ ∼ L/V ) 4 min and 100[1 - exp(-βx)] ) 86% of all the analyte molecules will be confined within a layer of ∼1.6 µm. Thus, EP has a tremendous potential for boosting assay sensitivity. Of course, many factors are capable of reducing the concentrating effect: (i) membrane polarization is capable of reducing local electric field; (ii) surface roughness results in less analyte bound to the surface areas protruding through a thin concentrated analyte layer; (iii) repulsion of protein ions tends to resist analyte concentrating. A detailed discussion of all these factors goes beyond the scope of the present paper. We will consider only the first factor due to its importance for inhibition of convection and for changing solution composition in the vicinity of the membrane surface. Concentration Polarization at the Dialysis Membrane. This may be considered as one explanation for the concentration effect being lower than theoretically predicted. Direct measurements of the buffer conductivities in aliquots taken from different depths in the EP cell reveal a notable redistribution of buffer components within the cell during EP. As seen in Figure 4, when current goes upward (plus at the electrode in the lower chamber), the electrolyte concentration increases 3-4 times at the bottom and decreases by ∼30% at the top as compared to the initial solution. The overall effect of such concentration polarization consists of redistribution of the electric field along the electrophoretic cell with the field at the bottom being ∼4 times lower (E ∼370 V/m) than it would be without the polarization. Analyte concentration will be reduced to the same extent. Since the polarization was reduced by ∼50% upon decrease in concentration of titrable acidic groups inside the membrane from 10 to 0.5 mM as a result of treatment with a mixture of EDC and ethanolamine, and since the polarization completely disappeared in 10 mM HCl solutions, one can attribute it to the cationexchanging properties of the dialysis membrane. Due to lower
mobility of anions in the membrane, the latter are accumulated at the boundary. Increased electrolyte concentration results in a drop of electric field and in lower heat production. When this happens over the bottom membrane (plus on the electrode in the lower chamber), an increase in temperature with height (by 3-6 °C when EP is performed in 5 mM Gly-Gly, pH 8.5, current, 1.5-2 mA/cell) produces a stable density gradient. Accumulation of more concentrated buffer below the socket membrane in the upper part of the cell upon reversal of the field is rapidly washed out by convection, as one can see in Figure 4. Inhibition of Convection. Though an electric field accelerates the assay even in the presence of convection, it takes a longer time and the bound analyte molecules are not uniformly distributed over the substrate surface. It is, therefore, preferable to inhibit convection. The stable density gradient formed due to concentration polarization, discussed above, could be used as an easy and convenient means to suppress convection when EP is used to collect negatively charged analytes, such as most of the IgG antibodies at pH >8. We also tested two other simple techniques to prevent convection: addition of dry Sephadex G-25 to the wells and formation of a density gradient. In the latter case, 70-100 µL of 10-20% glycerol or sucrose solution (prepared in the buffer used to dissolve analyte) was added into each well, topped with analyte solution, and slightly mixed with the pipet tip. Such a density gradient suppressed convection so effectively that accumulation of the concentrated buffer solution near the top of the EP cell became possible. Effect of Buffers and Other Solutes. Use of the membrane polarization discussed above as a means to inhibit convection requires choice of buffers in such a way as to provide a pH at which the analyte molecules have negative charge and move downward in the EP cell. Of course, buffer should have low conductivity to allow rapid analyte concentration without solution overheating. Since most IgG have pIs in the range of 6.2-8.5, we have chosen pH 8.5. We found that different buffers could be used in EP ELISA with almost similar results. Thus, signals obtained in direct active ELISA (hIgG on membrane, anti-hIgG-AP in solution) with four different buffers (10 mM borate, Tris, GlyGly, or glycine) of the same pH 8.5 varied by less than 2-fold. The presence of other charged macromolecules in the analyte solution may affect the EP-assisted assay. Thus, addition of 0.1% of Hb decreased signal in the EP ELISA ∼2-fold (as compared to that in 0.1% Tween-20), presumably due to direct interference with the antigen-antibody reaction or due to increased viscosity. In contrast to Hb, a dialyzed rabbit serum added to analyte solution (specific antibody from rabbit) in a concentration of 1% (v/v) resulted in no changes of the signal in EP ELISA, whereas addition of 10% serum decreased the signal 2-fold. Electrophoretic Washing. Simple estimates show that under typical EP conditions force applied to a protein molecule with Z ) 5 net charges is F ) eZE ∼ 0.001 pN. This force is 5 orders of magnitude below the level of forces necessary to break specific bonds (50-200 pN according to refs 15 and 16). Though this force is too low to directly affect breakage of even weak unspecific bonds, we found that the reversal of electric field for 2-3 min after electroconcentrating reduced the level of the signal in ELISA by 5-30% depending on the type of coating and on analyte concentration. Such “electrophoretic washing” can be attributed
Figure 5. Dependence of ELISA signal on pH of coating solution. Plasma-treated membranes were coated for 1 h from 10 µg/mL solutions of HEWL or Ova in different 10 mM buffers (acetic, pH 5.0; MES, pH 6.0; PIPES, pH 7.5; borate, pH 9.4). Rabbit anti-Ova and anti-HEWL were diluted 2 × 104 times in 5 mM Gly-Gly buffer pH 8.5 with 0.1% Hb. After EP at 1.5 mA/well for 10 min, anti-rIgG-AP conjugate diluted 2 × 104-fold with the same Gly-Gly buffer was electroconcentrated under similar conditions. The inset presents the same dependence on the DDS-treated membrane. Mean values of controls without the primary antibody are subtracted.
to a rapid active removal of unbound analyte molecules, preventing them from reassociation. Coating of Dialysis Membranes. Though many other ultrafiltration membranes could be used in an EP-assisted immunoassay, we have chosen a dialysis membrane from regenerated cellulose owing to its mechanical strength and a relatively high optical transparency, which allows us to directly measure OD on a standard microplate reader (wet membrane, 88 µm thick, has an OD of ∼0.1 at λ ) 405 nm). However, untreated membrane from regenerated cellulose binds proteins weakly and unevenly. Since the surface concentration of bound probe molecules is an important factor affecting sensitivity of any heterogeneous immunoassay, we tested several techniques to activate dialysis membrane. Hydrophobization is probably the simplest way to increase adsorption of proteins. Vapor deposition of a cyanoacrylate adhesive on the membrane surface is one easy way to do this. Exposure to a cyanoacrylate vapor makes the dialysis membrane slightly hydrophobic and notably increases protein binding. Thus, a short exposure (7 min) of a wet dialysis membrane to a vapor of a relatively nonvolatile octyl cyanoacrylate glue increased the membrane capacity to adsorb proteins ∼3 times. Exposure to a vapor of dimethyldichlorsilane is another simple technique for hydrophobization. Such “activation” increased hIgG adsorption ∼30-fold. Weak pH dependence of the hIgG coating in the range of pH between 5.0 and 9.5 in a low-salt buffer indicates that the hydrophobic forces are mostly responsible for the IgG binding on such a membrane. Treatment in Plasma Discharge. Adsorption on plasma-treated membrane, which was not exposed to cyanoacrylate vapor, shows strong dependence of the ELISA signal upon pH and salt concentration of coating solution. As illustrated in Figure 5, proteins with different isoelectric points show quite different dependence of coating efficiency upon pH. Whereas ovalbumin (pI ) 4.7) binds better at acidic pH, basic protein, hen egg Analytical Chemistry, Vol. 75, No. 24, December 15, 2003
6817
white lysozyme (pI ) 11.0), produces a better coating at basic pH. To the contrary, these two proteins show similar pH dependence on a hydrophobic surface as seen from the inset in Figure 5. It is well known17,18 that the surface of any plastic subjected to plasma discharge in air or oxygen acquires a variety of groups, including carboxylate, carbonyl, and peroxide ones. Our direct titration experiments revealed that the density of acidic groups was increased by ∼6 nM /cm2 after 30-s treatment in plasma. The presence of these negative charges results in attraction/repulsion between charged protein molecules and the negatively charged membrane surface affecting their adsorption or formation of chemical bonds. Similar pH dependence characterizes protein immobilization on a layer of carbohymethylated dextran in the BIAcore technique.11 Carbonyl groups introduced into the membrane surface in the plasma discharge17,18 can directly react with amine groups of proteins forming the Schiff bonds.19 Involvement of carbonyl groups in binding of proteins to a plasma-treated surface has been proven in our experiments with AP immobilization. It was found that 7 times less AP was bound to the plasma-treated membrane after reduction of carbonyl groups in a solution of sodium borohydride. The effect of plasma treatment on protein binding was not notably decreased even after 19-h storage at room temperature in water. This observation rules out the idea that radicals (formed directly in plasma or upon subsequent decomposition of peroxide groups) might be involved in covalent bonding of protein molecules. According to our measurements with a Linnik microinterferometer, plasma ablates the surface layer of the dialysis membrane at a rate of ∼0.1 µm/min in our conditions, “erasing” all memory of the previous treatments and immunoassays. We found that microplates could be reused at least 3-4 times. Coupling to carboxylic groups. Numerous carboxylates introduced into surface layer in plasma present another avenue for protein binding. Standard EDC/NHS procedure optimized upon development of the BIAcore technique,11 produced the highest signal in ELISA as seen from Figure 6. Coupling via PEI/GA. Binding of PEI to a plasma-treated surface was expected to decrease its negative charge or even recharge it. Indeed, maximum hIgG binding was shifted from pH
