Evanescent Wave Long-Period Fiber Bragg Grating as an Immobilized

An immunosensor using a long-period grating (LPG) was used for sensitive detection of antibody-antigen reactions. Goat anti-human IgG (antibody) was ...
0 downloads 0 Views 163KB Size
Download PDF
Anal. Chem. 2000, 72, 2895-2900

Evanescent Wave Long-Period Fiber Bragg Grating as an Immobilized Antibody Biosensor Matthew P. DeLisa,† Zheng Zhang,‡ Mira Shiloach,† Saeed Pilevar,§ Christopher C. Davis,§ James S. Sirkis,‡ and William E. Bentley*,†

Department of Chemical Engineering and Center for Agricultural Biotechnology, Department of Electrical Engineering, and Smart Materials and Structures Research Center, University of Maryland, College Park, Maryland 20742

An immunosensor using a long-period grating (LPG) was used for sensitive detection of antibody-antigen reactions. Goat anti-human IgG (antibody) was immobilized on the surface of the LPG, and detection of specific antibodyantigen binding was investigated. This sensor operates using total internal reflection where an evanescent field interacts with bound antibody immobilized over the grating region. The reaction between antibody and antigen altered the LPG transmission spectrum and was monitored in real time as a change in refractive index, thereby eliminating the need for labeling antigen molecules. Human IgG binding was observed to be concentration dependent over a range of 2-100 µg mL-1, and equilibrium bound antigen levels could be attained in ∼5 min using an initial rate determination. Binding specificity was confirmed using human interleukin-2 and bovine serum albumin as controls, and nonspecific adsorption of proteins did not significantly interfere with detection of binding. Antigen detection in a heterogeneous protein mixture and in crude cell lysate from Escherichia coli was also confirmed. Moreover, regeneration of the LPG surface via diethlyamine treatment resulted in ∼80% removal of bound antigen. Subsequently, fibers reexposed to antigen retained greater than 85% of the initial signal after five consecutive regeneration cycles. Fiber-optic biosensors based on evanescent wave interactions have commonly been used in medical and environmental diagnostics. Detection and quantification of biomolecular interactions have benefited greatly from the incorporation of fiber-optic technology.1 A variety of optical sensing strategies have been developed, which are capable of detecting a broad range of biologicals including, but not limited to antibodies,2,3 nucleic acids,4,5 toxins,6 pesticides,7,8 explosives,9 and numerous other small molecules. * To whom all correspondence should be addressed: (telephone) (301) 4054321; (fax) (301) 314-9015; (e-mail) [email protected]. † Department of Chemical Engineering and Center for Agricultural Biotechnology. ‡ Smart Materials and Structures Research Center. § Department of Electrical Engineering. (1) Yeung, D.; Gill, A.; Maule, C. H.; Davies, R. J. Trends Anal. Chem. 1995, 14, 49-56. (2) Lu, B.; Lu, C.; Wei, Y. Anal. Lett. 1992, 25, 1-10. (3) Cush, R.; Cronin, J. M.; Stewart, W. J.; Maule, C. H.; Molloy, J.; Goddard, N. J. Biosens. Bioelectron. 1993, 8, 347-364. 10.1021/ac9912395 CCC: $19.00 Published on Web 06/01/2000

© 2000 American Chemical Society

A common approach used in optical fiber immunosensor fabrication is chemical modification of a silica surface allowing for immobilization of a high density of antibody molecules over the surface of a fiber. A variety of immobilization methods have been reported including avidin bridging,10 covalent immobilization using heterobifunctional cross-linking agents,11-13 and noncovalent attachment via adsorption and gel entrapment.14 Subsequently, a target antigen specific to the immobilized antibody is incubated with an auxiliary label, such as a fluorophore or radioisotope, prior to detection. Finally, a competitive or sandwich immunoassay is carried out on the fiber surface generating a real-time fluorescence (or radioactive) signal as labeled antigen specifically interacts with immobilized antibody. One disadvantage of these approaches is the need for antigen labeling or additional reagents, which can be both time-consuming and expensive. A surrogate methodology is to monitor the change in refractive index that occurs within the evanescent field upon binding of the target antigen on a derivatized surface.15,16 Direct detection of molecular interactions can be carried out while avoiding the use of auxiliary chemical components. The use of long-period gratings (LPGs) is one alternate approach in which antibody-antigen interactions can be monitored while circumventing the attachment of a label. Therefore, label-free assays (4) Graham, C. R.; Leslie, D.; Squirrell, D. J. Biosens. Bioelectron. 1992, 7, 487493. (5) Pilevar, S.; Portugal, F.; Davis, C. C. Anal. Chem. 1998, 70, 2031-2037. (6) Ogert, R. A.; Brown, J. E.; Singh, B. R.; Shriver-Lake, L. C.; Ligler, F. S. Anal. Biochem. 1992, 205, 306-312. (7) Bier, F. F.; Schmid, R. D. Biosens. Bioelectron. 1994, 9, 125-130. (8) Eldefrawi, M. E., Eldefrawi, A. T., Anis, N. A., Rogers, K. R., Wong, R. B., Valdes, J. J. In Immunoanalysis of Agrochemicals; Nelson, J. O., Karu, A. E., Wong, R. B., Eds.; ACS Symposium Series 586; American Chemical Society: Washington, DC, 1995; pp 197-209. (9) Ligler, F. S.; Golden, J. P.; Shriver-Lake, L. C.; Ogert, R. A.; Wijesuria, D.; Anderson, G. P. Immunomethods 1993, 3, 122-127. (10) Chiolerio, F.; Fillipini, E.; Magnaghi, S.; Acerboni, R.; Malcovati, M. Macromol. Recognit. 1987, 2, 303-307. (11) Bhatia, S. K.; Shriver-Lake, L. C.; Prior, K. J.; Georger, J. H.; Calvert, J. M.; Bredehorst, R.; Ligler, F. S. Anal. Biochem. 1989, 178, 408-413. (12) Ahluwalia, A., DeRossi, D., Ristori, C., Schirone, A. and Serra, G. Biosens. Bioelectron. 1991, 7, 207-214. (13) Shriver-Lake, L. C.; Donner, B.; Edelstein, R.; Breslin, K.; Bhatia, S. K.; Ligler, F. S. Biosens. Bioelectron. 1997, 12, 1101-1106. (14) Subramanian, A.; Kennel, S. J.; Oden, P. I.; Jacobson, K. B.; Woodward, J.; Doktycz, M. Enzyme Microb. Technol. 1999, 24, 26-34. (15) Schlatter, D.; Barner, R.; Fattinger, Ch.; Huber, W.; Hubscher, J.; Hurst, J.; Koller, H.; Mangold, C.; Muller, F. Biosens. Bioelectron. 1993, 8, 109-116. (16) Spinke, J.; Oranth, N.; Fattinger, Ch.; Koller, H.; Mangold, C.; Voegelin, D. Sens. Actuators B 1997, 38-39, 256-260.

Analytical Chemistry, Vol. 72, No. 13, July 1, 2000 2895

using LPGs not only can reduce the time and cost of the immunoassay but also make possible on-line monitoring and detection of important biological molecules. Long-period gratings in a single-mode fiber, having a period on the order of hundreds of micrometers, couple light from the forward-propagating mode into several forward-propagating cladding modes. The resulting transmission spectrum shows characteristic loss peaks. Since any variation in the core/cladding guiding properties modifies the transmission spectrum characteristics, LPGs have been used as strain, temperature, and refractive index sensors.17-21 Since LPGs are extremely sensitive to changes in the refractive index at the sensor surface, real-time monitoring of antibody-antigen interactions is a logical extension. The work described here demonstrates the feasibility of using LPGs to develop a fiber-optic biosensor. Transmission peaks shift with changes of the refractive index of the ambient medium in the vicinity of the fiber cladding. The ability of LPGs to couple light from the fiber core to the fiber cladding allows light to directly probe the aqueous environment surrounding the fiber cladding. Derivatizing LPGs with amino, carboxylate, or biotin groups allows for immobilization of antibodies followed by measurement of the bioreaction between bound antibodies (receptor) and free antigens as a change in refractive index that further results in a spectral shift of the LPG. By demodulating the spectral shift, these gratings can be used to probe specifically for a desired antigen with sensitivity (down to ∼12 nmol L-1) comparable to other label-free methods (surface plasmon resonance, interferometry) based on refractive index measurements.3,22,23 The specificity and sensitivity of LPG immunosensors renders them ideal tools for use in medical diagnosis, environmental monitoring, bioprocessing, and other applications. EXPERIMENTAL SECTION Reagents. Analytical grade reagents and sterile deionized water were used for preparing all solutions. Human IgG, goat antihuman IgG (anti-hIgG), protein A, bovine serum albumin (BSA), human albumin, and purified chloramphenicol acetyltransferase (CAT) were obtained from Sigma (St. Louis, MO). Bovine insulin was obtained from JRH Biosciences (Lenexa, KS), and human interleukin-2 (hIL-2) was obtained form CYTImmune Sciences Inc. (College Park, MD). Purified recombinant green fluorescent protein (rGFPuv) was obtained from Clontech (Palo Alto, CA), and purified GroES, GroEL, and DnaK were supplied by StressGen Biotechnologies Corp. (Victoria, BC, Canada). (3-Aminopropyl)triethoxysilane (APTS) was obtained from Aldrich (Milwaukee, WI), and glutaraldehyde was obtained from Fluka (Ronkonkoma, NY). Escherichia coli crude cell lysate was obtained from high cell density cultivation of E. coli strain W3110 [pTH-GFPuv/CAT] grown in a 5 L BioFlo III fermentor (New Brunswick Scientific, (17) Bhatia, V.; Vengsarkar, A. M. Opt. Lett. 1996, 21, 692-694. (18) Bhatia, V. Opt. Express 1999, 4, 457-466. (19) Patrick, H. J.; Williams, H. J.; Kersey, A. D.; Pedrazzani, J. R.; Vengsarkar, A. M. IEEE Photon Tech. Lett. 1996, 8, 1223-1225. (20) Patrick, H. J.; Kersey, A. D.; Bucholtz, F. J. Lightwave Technol. 1998, 16, 1606-1612. (21) Patrick, H. J., Chang, C. C. and Vohra, S. T. Electron. Lett. 1998, 34, 17731775. (22) Lundstro ¨m, I. Biosens. Bioelectron. 1994, 9, 725-736. (23) Fattinger, C.; Koller, H.; Schlatter, D.; Wehrly, P. Biosens. Bioelectron. 1993, 8, 99-107.

2896

Analytical Chemistry, Vol. 72, No. 13, July 1, 2000

Edison, NJ). A 25-mL sample was harvested at an optical density (OD600) reading of ∼10. Following centrifugation for 10 min at 5000g and 4 °C, pelleted cell mass was resuspended in 20 mL of 50 mM Tris-HCl buffer (pH 7.5). After a second 10-min centrifugation at 5000g and 4 °C, the pellet was resuspended in 5 mL of Tris-HCl buffer and sonicated (model 550 Sonic Dismembrator, Fisher Scientific, Pittsburgh, PA) for 3 min and 0.5-s pulse intervals. Lysed cells were centrifuged for 10 min at 5000g and 4 °C. Last, supernatant was decanted and used to make 10 and 50% (v/v) crude cell lysate solution in 0.01 M phosphate-buffered saline (PBS, 0.137 M NaCl, pH 7.4). Preparation of Long-Period Fiber Gratings. Long-period gratings were fabricated by creating a periodic modulation of refractive index in the core of the fiber. This was accomplished by exposing a standard single-mode fiber with germanium-doped core and cladding (SMF-28 fiber, Corning) to UV light from a frequency-doubled argon laser (Innova 300 Fred, Coherent Inc., Santa Clara, CA) at a wavelength of 244 nm using a point-by-point writing method. The UV laser beam was first focused onto the stripped fiber. The fiber was then moved incrementally forward using a computer-controlled step motor. The UV radiation (100400 nm) reacted with the germanium-doped fiber core causing a refractive index change24 on the order of 10-4. To increase the photosensitivity, the optical fibers were first hydrogen loaded under a pressure of 1000 psi at 70 °C. Following UV light exposure, the LPGs were then annealed at 150 °C for 24 h to stabilize the LPG spectrum. Chemical Treatment of LPGs. Long-period gratings were cleaned by immersion in 200 mL of 5% nitric acid for 2 h at 90 °C followed by thorough rinsing in sterile deionized water (3×). Silanization of the LPG surface was performed by immersion in 200 mL of fresh 10% (v/v) APTS in water (pH 3-4) for 1 h at 75 °C. The treated LPGs were then dried in a convection oven (VWR/ Sheldon Manufacturing Inc., Cornelius, OR) for 4 h at 115 °C. Following cooling, dried fibers were treated with 1% (v/v) glutaraldehyde in water (pH 6-7) for 30 min at room temperature. After being rinsed in 0.01 M PBS (0.137 M NaCl, pH 7.4) for 15 min, fibers were incubated in 2 mL of 0.5 mg mL-1 goat antihuman IgG (antibody) in 10 mM sodium acetate (pH 5.4) overnight at 4 °C. Once immobilization of antibody was completed, fibers were inserted into the flow cell (4 cm long × 0.5 mm wide × 0.5 mm deep) and binding assays were carried out. Signal Processing System. To detect the spectral shift of LPGs in real time, a scanning fiber fabry perot (FFP, Micron Optics, Inc., Atlanta, GA) filter demodulation system was built.25 As shown in Figure 1, the light from the broad-band source, an erbium-doped fiber amplifier (EDFA, M1702, AT&T Bell Laboratories, Breiningsville, PA), was injected into a standard shortperiod fiber Bragg grating (FBG), an LPG sensor, a tunable FFP filter, and finally a photodetector (New Focus, Inc., Mountain View, CA). The FBG was used as a wavelength reference to compensate for the drift and hysteresis of the FFP filter. By scanning the FFP filter over the spectral range containing the two gratings and differentiating the transmission spectrum using a lock-in amplifier at the same time, the zero-crossings of the derivative function were detected by computer using a 12-bit data acquisition card (DT7102, (24) Shackleford, J. F. Introduction to Materials Science for Engineers, 3rd ed.; Macmillian Publishing Co.: New York, 1992. (25) Kersey, A. D. Proc. SPIE-Int. Soc. Opt. Eng. 1992, 2071, 30-48.

Figure 1. (a) Schematic of signal processing system used to detect spectral shifts of LPGs. (b) Schematic of LPG immunosensor principle.

Data Translation, Inc., Marlboro, MA). The voltage difference, ∆V, between the zero-crossings of the reference FBG and LPG was proportional to the spectral shift of the LPG and was recorded using a self-coded program within LabVIEW (National Instruments, Austin, TX). Unlike other laser probing techniques, this approach was not dependent on wavelength-independent losses in the LPG or intensity variations in the optical source. Immunoassays. As the refractive index was sensitive to changes in solution flow rate, human IgG or human interleukin-2 (control) was delivered to the flow cell housing the treated LPG via a peristaltic pump (Watson Marlow, Wilmington, MA) at a flow rate of 250 µL min-1, unless otherwise noted. Immunoassays were initiated by rinsing the LPG with 5 mL of 0.01 M PBS directly through the flow cell at a flow rate of 250 µL min-1. This provided a stable baseline over which signal was obtained. Immediately after the PBS wash, a 10-mL antigen-containing solution was then introduced to the flow cell. A subsequent 10 mL PBS washing (flow rate 250 µL min-1) was carried out to remove unbound antigen. Initial rate calculations were performed using a polynomial least-squares fit through the data obtained over the first 5 min of each immunoassay. RESULTS AND DISCUSSION Dose-Response Experiments. The feasibility of using LPGs for detection of antibody-antigen reactivity via direct binding assays was investigated using chemically modified fibers with goat anti-human IgG cross-linked to the surface. One advantage of using LPGs was that signal response was obtained continuously in a concentration-dependent fashion as opposed to fluorophore experiments in which signal is typically obtained periodically to prevent photobleaching. Continuously obtained response data generated during dose-response experiments is shown in Figure 2. In addition to monitoring maximum signal, kinetic analysis of binding was performed by calculating the slope of the response

Figure 2. Time profile of antigen (human IgG) binding to goat antihuman IgG immobilized on LPG surface. Experiments performed in three stages (indicated as vertical dotted lines) as follows: (a) prewashing with 5 mL of 0.01 M PBS providing baseline prior to antigen detection; (b) incubation with human IgG (flow rate 250 µL min-1), and (c) rinsing with 0.01 M PBS.

curve during initial binding prior to signal saturation (Table 1). It was determined that the initial rate could be used to calculate the concentration of antigen; thus, for each concentration there was a corresponding characteristic slope. This calculation was made within the first 5 min of each experiment regardless of the saturation concentration of the LPG, allowing for very rapid determination of antigen levels. Moreover, since these results were independent of signal saturation, reliable fiber-to-fiber measurements were obtained. Treated LPGs bound human IgG in a concentration-dependent manner over a range of 2-50 µg mL-1. The dose-response results are presented in Figure 3. Maximum signal was reported as the absolute change in refractive index after subtracting the baseline Analytical Chemistry, Vol. 72, No. 13, July 1, 2000

2897

Table 1. LPG Fiber-Optic Immunoassay Results Using Human IgG (Antigen) and Goat Anti-Human IgG (Receptor) antigen concn (µg mL-1)

max signal (AU)

init rate (min-1)

human IgGa 2 5 10 20 20 (flow rate 100 µL min-1) 20 (flow rate 500 µL min-1) 50

0.8 ( 0.2 3.5 ( 0.2 7.3 ( 0.1 10.2 ( 0.7 10.6 ( 0.8 11.0 ( 0.9 12.0 ( 0.5

0.05 ( 0.01 0.15 ( 0.01 0.35 ( 0.02 0.88 ( 0.08 0.68 ( 0.04 0.92 ( 0.07 1.18 ( 0.15

a

Flow rate was 250 µL min-1 unless otherwise noted.

Figure 4. Time profile of human IgG binding (20 µg mL-1) at flow rates of 100, 250, and 500 µL min-1. Solid arrow indicates initiation of human IgG exposure to treated LPGs. Hollow arrowheads indicate initiation of 0.01 M PBS rinsing.

signal obtained during the prewashing step. Linearity between maximum signal and human IgG concentration was obtained for concentrations up to 20 µg mL-1, which was similar to previous reports.2 Beyond this range, the propensity for the response to increase proportionally with the dose decreased and reached its maximum point. Further, regression analysis of the maximum signal with initial rate data resulted in a linear correlation (r2 ) 0.92) indicating that the initial rate was a statistically reliable (and more rapid) indicator of antigen level. Experiments to determine the effect of nonspecific adsorption on binding signal were run by prewashing the LPGs with blocking agent (BSA in PBS, 2 mg mL-1). No difference was observed in the response curves obtained when BSA was used (data not shown); therefore, there was no interference from nonspecific adsorption. Interestingly and fortuitously, BSA was thus used as a control antigen in future work. Control proteins human IL-2 and BSA were used to demonstrate the specificity of the biosensor. In both cases, these proteins gave no appreciable signal at the concentrations investigated. Sensitivity was defined as the concentration on the response curve corresponding to a signal equal to that at the zero antigen concentration at two standard deviations26 and was calculated as 0.7 µg mL-1. Last, the use of protein A, an immunoglobulin binding protein containing four Fc binding

domains, was previously shown to improve antigen binding27 and was explored here. However, this was of no benefit and in fact was detrimental. Immunoassays for human IgG (20 µg mL-1) using LPGs to which the capture antibody was bound via protein A led to a 24% decrease in maximum signal compared to that obtained for covalently attached antibody (data not shown). Effect of Flow Rate on Antigen Binding. As noted earlier, all dose-response experiments were performed by delivering antigen at a flow rate of 250 µL min-1. To determine the effect of flow rate on binding response curves, flow rates of 100 and 500 µL min-1 were investigated (depicted in Figure 4) for detection of 20 µg mL-1 antigen. Importantly, flow rates of 250 and 500 µL min-1 resulted in similar peak signals (10.2 and 11.0) and initial binding rates (0.88 and 0.92 min-1), respectively, whereas a flow rate of 100 µL min-1 produced a peak signal of 10.6 but an initial binding rate of 0.68 min-1. Additionally, although the maximum signals were similar for all three of these response curves, this signal was obtained in ∼20 min for a flow rate of 500 µl min-1 compared to approximately 40 and 60 min for antigen delivery rates of 200 and 100 µL min-1, respectively. Thus, future experiments could be expedited by increasing the flow rate of antigencontaining solutions. Probing Heterogeneous Protein Solutions. For an efficacious biosensor, many practical applications require detection of antigens in complex, nonhomogeneous solutions. Therefore, the ability to detect human IgG in the presence of competitor proteins and in complex heterogeneous solutions was investigated and the results are presented in Table 2. First, human IgG (10 µg mL-1) was dissolved in a 0.01 M PBS solution containing 10 µg mL-1 of the following proteins: BSA, human albumin, insulin, hIL2, rGFPuv, and tobacco mosaic virus coat protein; as well as 1 µg mL-1 of the following proteins from E. coli: CAT, GroES, GroEL, and DnaK. Treated LPGs exposed to this mixture along with 10 µg mL-1 human IgG gave a peak signal of ∼8.3 units which is ∼1.0 unit higher than the signal obtained for a homogeneous solution having the same concentration of human IgG. This was likely a consequence of nonspecific adsorption and/or cross-

(26) Sloper, A. N.; Deacob, J. K.; Flanagan, M. T. Sens. Actuators B 1990, B1, 589.

(27) Anderson, G. P.; Jacoby, M. A.; Ligler, F. S.; King, K. D. Biosens. Bioelectron. 1997, 12, 329-336.

Figure 3. Fiber-optic immunoassays using LPGs coated with goat anti-human IgG and incubated with human IgG. Dose-response data and initial rate calculations were reported as the mean maximum normalized signal and mean initial rate, respectively, ( SE (n ) 4). Control antigens (human interleukin-2 or BSA) resulted in no appreciable signal for all concentrations investigated.

2898 Analytical Chemistry, Vol. 72, No. 13, July 1, 2000

Table 2. Fiber-Optic Immunoassays Probing Nonhomogeneous Solutions for Human IgG (10 µg mL-1) antigen-containing solutiona

max signal ( SEb

0.01 M PBS heterogeneous protein mixture 2.2 mg mL-1 total proteinc 2.2 mg mL-1 total proteinc + BSA 11.0 mg mL-1 total proteind

7.3 ( 0.1 8.3 ( 0.3 9.0 ( 0.4 7.3 ( 0.2 7.0 ( 0.3

a All antigen-containing solutions contained 10 µg mL-1 human IgG and were based in 0.01 M PBS solution. b The mean ( SE (n ) 3) of the maximum normalized signal. c 10% (v/v) crude cell lysate from E. coli culture. d 50% (v/v) crude cell lysate from E. coli.

reactivity of proteins with the LPG surface. Likewise, when probing for human IgG (10 µg mL-1) dissolved in ∼2.2 mg mL-1 total E. coli protein (10% (v/v) E. coli crude cell lysate in PBS), a signal of ∼9.0 units was obtained. To test the assumption that the elevated signal was due to nonspecific adsorption to the treated LPG surface, fibers were prewashed with BSA blocking solution (2 mg mL-1). As shown in Table 2, the additional blocking step led to a peak signal of 7.0 units, which coincided exactly with data obtained previously for this concentration of antigen (see Figure 2). Interestingly, when antigen (10 µg mL-1) dissolved in ∼11.0 mg mL-1 total E. coli protein (50% (v/v) E. coli crude cell lysate in PBS) was probed using the treated LPG, a signal of ∼6.7 was obtained without prewashing with BSA blocking solution. Additionally, this signal was obtained at a slower rate (0.44 min-1) as compared to 0.85 and 0.52 min-1 attained for 10 µg mL-1 antigen solutions in 2.2 mg mL-1 total E. coli protein and 2.2 mg mL-1 total E. coli protein with BSA blocking, respectively. It was believed that the increased viscosity due to the higher concentration of total protein led to a reduction in the mass-transfer rate and, consequently, a reduced initial binding rate. Phenomena of this nature would have to be accounted for when initial rate calculations of nonhomogeneous biological samples are performed. Regeneration of Surface Activity. Finally, regeneration of a biosensor is advantageous for facilitating semicontinuous on-line monitoring. Many techniques have been demonstrated to dissociate antibody-antigen complexes including pressure and electric field;28,29 however, the use of solvents is the most straightforward and least expensive. Previously, it was reported that dissociation of bound antigens was most effective using aqueous diethylamine (0.05 M, pH 11.5) displaying a reassociation ability of greater than 97% up to 8 times in succession.2 Therefore, regeneration of the treated LPG surface was performed using aqueous diethylamine treatment (Figure 5). A time profile of two consecutive regenerations is presented in Figure 5a. These results demonstrate the advantage of monitoring refractive index change, as it was possible to measure antigen binding as well as continuously track antigen stripping. The signal obtained upon reexposure to antigen is reported in Figure 5b as percent of the initial signal (including ( the standard error). Note, the initial rate for each successive stripping is also reported in Figure 5 relative to the initial rate calculated for the first exposure to antigen. As expected, there was very good agreement between signal measurements and the (28) Olsen, W. C.; Leung, S. K.; Yarmush, M. L. Biotechnology 1989, 7, 369. (29) Matousek, V.; Horejsi, V. Chromatographia 1982, 35, 285.

Figure 5. (a) Time profile of LPG generation (two cycles); (b) regeneration of LPGs coated with goat anti-human IgG and incubated with 20 µg mL-1 human IgG. LPGs washed with 0.05 M diethylamine (pH 11.5) prior to reexposure with antigen. The mean ( SE of the maximum signal or initial rate expressed as a percentage of the initial response (n ) 3) is shown. Key: black bar, signal; gray bar, initial rate.

initial rates calculated for the five consecutive regenerations. Only a slight decrease in signal level was observed, as greater than 85% surface activity was retained after five successive washes with diethylamine. This result was identically observed in the initial rate calculations. An average of 86% antigen removal over the course of the experiments was obtained (data not shown); however, the amount of unstripped antigen increased steadily with each successive regeneration. It was hypothesized that this slight increase in unstripped antigen directly led to the slight decrease observed in the signal obtained following reexposure to antigen. CONCLUSIONS In summary, a sensitive (approximately nanomolar) immunosensor for the specific detection of human IgG has been developed by immobilizing goat anti-human IgG antibody on the surface of a long-period fiber grating. The principle of operation is based on the change of refractive index in the evanescent field that occurs at the LPG surface upon antibody-antigen interaction. Since LPGs coupled light from the fiber core to the fiber cladding, direct probing of the aqueous environment surrounding the fiber cladding and quantitative measurement of antigen binding were Analytical Chemistry, Vol. 72, No. 13, July 1, 2000

2899

possible. Furthermore, as a consequence of monitoring refractive index shifts, label-free immunoassays were successfully performed. To our knowledge, this is the first published report on the use of long-period gratings for detection of antibody-antigen interactions. It is also one of the first reports to demonstrate that LPGs can detect extremely specific interactions occurring within the evanescent field and that these results are highly reproducible. Experimental results confirmed that antigen binding occurred in a concentration-dependent manner over a range of 2-100 µg mL-1. Although this range was adequate for demonstrating LPGs as a viable immunosensor, experiments are currently under way using alternate immobilization techniques in an effort to expand the linear range of antigen detection similar to those published for label-based (i.e., fluorophore) assays.12,13 Of note, peak signal was obtained in 40 min when antigen delivery flow rate was 250 µL min-1 and this improved to 20 min at a flow rate of 500 µL min-1. Moreover, calculation of the initial rate led to determination of antigen concentration after only 5 min. The feasibility of using the treated LPGs for detecting antigens in nonhomogeneous solutions, such as multiprotein mixtures and

2900

Analytical Chemistry, Vol. 72, No. 13, July 1, 2000

cell lysate from bacteria, was tested and results indicated that the LPGs were sufficiently sensitive to detect antigen binding in these highly complex aqueous mediums. An extension of this preliminary work is to develop LPGs capable of detecting recombinant proteins (i.e., human interleukin-2) produced in microorganisms such as E. coli. Additionally, the ability to regenerate surface activity will allow for on-line monitoring of bioprocesses in which these recombinant products are being synthesized at high levels. ACKNOWLEDGMENT Partial funding for this research was provided by the National Science Foundation ECSEL Coalition (to the University of Maryland), the United States Army (to W.E.B.), and the Army Research Laboratory (to C.C.D.).

Received for review October 29, 1999. Accepted March 1, 2000. AC9912395