Microcapsule Biosensors Using Competitive ... - ACS Publications

Swetha Chinnayelka, and Michael J. McShane* .... microcapsules were obtained by dissolving the MnCO3 cores using 0.1 M hydrochloric acid (HCl) solutio...
0 downloads 0 Views 356KB Size
Download PDF
Anal. Chem. 2005, 77, 5501-5511

Microcapsule Biosensors Using Competitive Binding Resonance Energy Transfer Assays Based on Apoenzymes Swetha Chinnayelka† and Michael J. McShane*,†,‡

Institute for Micromanufacturing, and Biomedical Engineering Program, Louisiana Tech University, Ruston, Louisiana 71272

This paper reports the first demonstration of a fluorescence resonance energy transfer based glucose sensor, wherein a competitive binding (CB) assay is encapsulated into polyelectrolyte microcapsules. The work supports the concept that microcapsules are superior to hydrogel systems or other matrixes for competitive-binding-based system, as they provide free movement of the sensing elements within the capsule interior while constant total sensing assay concentration is maintained. The transduction approach employed in these preliminary experiments is also a novel CB system based on a model apoenzyme, apo-glucose oxidase (AG), which is highly specific to β-Dglucose, as the model target-binding protein. The glucose sensitivity of the fluorescein isothiocyanate (FITC)-dextran and tetramethylrhodamine isothiocyanate-AG encapsulated in microcapsules showed 5 times greater specificity for β-D-glucose over other sugars, with sensitivity (change in intensity ratio) in the range of 2-6%/mM. It was observed that the sensitivity and range of the response can be tailored by controlling the assay concentration using different FITC-dextran molecular weight and total capsule concentration. The findings support the concepts of using microcapsules to encapsulate CB assays for reversible and stable sensors and the use of apoenzymes as specific molecular recognition elements in CB assays. Further, characterization results for microcapsule glucose sensors demonstrate their suitability for monitoring physiological glucose levels. Frequent biochemical monitoring of patients is necessary to prevent, diagnose, and treat a progressive disease early enough to avoid secondary complications. Therefore, in recent years, interest toward developing biosensors for continuous monitoring with high accuracy and specificity has increased. Glucose monitoring systems, in particular, have generated special interest because of the escalating worldwide impact of diabetes and the potential to minimize the undesirable effects of the disease through “tight” glucose control.1,2 While some minimally invasive devices for continuous glucose monitoring have reached the * Corresponding author. Phone: (318) 257-5100. Fax: (318) 257-5104. E-mail: [email protected]. † Institute for Micromanufacturing. ‡ Biomedical Engineering Program. (1) Mansouri, S.; Schultz, J. S. Biotechnology. 1984, 885-890. (2) Griffiths, D. G.; Hall, G. Trends Biotech. 1993, 11, 122-130. 10.1021/ac050755u CCC: $30.25 Published on Web 07/22/2005

© 2005 American Chemical Society

market, these have not reached widespread use for a number of reasons;3 therefore, optical sensors are still being pursued as alternatives, as they may transduce chemical concentration information into an optical signal that may be analyzed spectroscopically.4-9 While glucose monitors are driving sensor development, it is expected that such devices, if available, could also be modified to monitor other biochemicals, hormones, etc. for managing other medical conditions. Glucose sensors based on fluorescence spectroscopy are being investigated by many groups due to their high sensitivity and superior specificity in the case of enzymatic sensors.4-10 One of the approaches that has been deeply explored for glucose monitoring using fluorescence techniques is based on the changes in resonance energy transfer (RET) that occur due to the competitive binding (CB) of fluorescent ligand and analyte to occupy binding sites on a fluorescent receptor (lectin or other protein). The advantages of using competitive binding and RET techniques are the nonconsumption of analyte, the absence of reaction byproducts, and that the ratiometric nature of the RET analysis method allows variations in instrumental parameters, assay component concentrations, and measurement configuration to be internally compensated.11 Schultz has reported various sensors based on the combination of competitive binding and RET concepts, i.e., glucose competes with fluorescein isothiocyanate (FITC)-dextran (FD) to occupy binding sites on tetramethylrhodamine isothiocyanate (TRITC)-Con A due to higher binding affinity.1,4,14 Thus, glucose concentration has been estimated by measuring the changes in distance between FD and TRITC-Con A in terms of RET efficiency. Recently, this work on glucose sensors has been extended into the near-infrared (NIR) by labeling the dextran and Con A with an NIR donor-acceptor pair.15,16 RETbased glucose sensors have also been developed using other (3) Koschinsky, T.; Jungheim, K.; Heinemann, L. Diabetes Technol. Ther. 2003, 5, 829-842. (4) Chinnayelka, S.; McShane, M. J. J. Fluorescence 2004, 14, 585-595. (5) Chinnayelka, S.; McShane, M. J. Biomacromolecules 2004, 5, 1657-1661. (6) Meadows, D. L.; Schultz, J. S. Talanta 1988, 35, 145-150. (7) Cesare, N. D.; Lakowicz, J. R. Proc. SPIE 2002, 4625, 152-159. (8) Russell, R. J.; Pishko, M. V.; Gefrides, C. C.; McShane, M. J.; Cote´, G. L. Anal. Chem. 1999, 71, 3126-3132. (9) Brown, J. Q.; McShane, M. J. Biosens. Bioelectron. 2004, 22, 212-216. (10) McShane, M. J. Diabetes Technol. Ther. 2002, 4, 533-538. (11) Chantal, M.; Lebrun, C.; Prats, M. Biochem. Educ. 1998, 26, 320-323. (12) Schultz, J. S.; Mansouri, S.; Goldstein, I. J. Diabetes Care 1982, 5, 245253. (13) Meadows, D. L.; Schultz, J. S. Anal. Chim. Acta 1993, 280, 21-30. (14) Ballerstadt, R.; Schultz, J. S. Anal. Chim. Acta 1997, 345, 203-212.

Analytical Chemistry, Vol. 77, No. 17, September 1, 2005 5501

glucose-binding and genetically engineered proteins;17-19 these carry the disadvantage of low dissociation constant (∼0.1-1 µM), which generally makes them inappropriate for glucose monitoring at normal physiological levels. The RET-based approaches using Con A have been demonstrated in solution phase and with fiber optic probes. As a further development, a novel “smart tattoo” system was proposed, which could be implanted in the skin and interrogated remotely using harmless visible or near-infrared excitation light.8,20-22 The glucosesensitive Con A/dextran assay was entrapped in poly(ethylene glycol) microspheres, but there remain some problems in the production of uniform and strongly responsive glucose-sensitive microspheres, particularly in the areas of leaching stability and reversibility. We recently reported on polymeric glucose-sensitive microcapsules comprising TRITC-Con A and FITC-dextran (FD) multilayer films constructed using affinity binding and the layerby-layer (LbL) self-assembly technique.4 However, this approach has two primary disadvantages: (1) the signal levels and magnitude of response are limited, due to the entrapment of effectively low concentrations of the sensing molecules in the thin capsule walls and (2) the same concerns over Con A toxicity and nonspecific binding. To overcome the limitations of Con A, an alternative approach was recently developed using an inactive form of the enzyme glucose oxidase (apo-GOx) as the glucose-binding protein, which is highly specific to β-D-glucose.7,23 The glucose sensitivity of the TRITC-apo-GOx (TAG) and FD assay was estimated to be 3 times greater than previously published systems and, unlike the high binding affinity (low kd) in the case of GBPs, the dissociation constant was observed to be in the millimolar range (30-33 mM), making it more suitable for monitoring physiological glucose. To move solution-phase assays toward minimally invasive implantable sensors using the competitive binding transduction schemes, appropriate encapsulation methods are required. Here, we discuss a novel and facile approach, LbL self-assembly of photocross-linkable polyelectrolytes, for the fabrication of microcapsules that are used for carrying the apoenzyme/CB ligand RET assay elements. The major advantage in using microcapsules for competitive-binding assays is that they offer the potential for free movement of the ligand, receptor, and analyte during the process of competitive displacement. Furthermore, the semipermeable walls of the microcapsules facilitate maintenance of a constant concentration of the larger sensing assay elements while allowing analyte transport. (15) Ballerstadt R.; Polak A.; Beuhler A.; Frye J. Biosens. Bioelectron. 2004, 19, 905-914. (16) Ballerstadt, R.; Gowda, A.; McNichols, R. Diabetes Technol. Ther. 2004, 6, 191-200. (17) Tolosa, L.; Gryczynski, I.; Eichorn, L. R.; Dattelbaum, J. D.; Castellano, F. N.; Rao, G.; Lakowicz, J. R. Anal. Biochem. 1999, 267, 114-120. (18) Ge, X.; Tolosa, L.; Rao, G. Anal. Chem. 2004, 76, 1403-1410. (19) Ye, K.; Schultz, J. S. Anal. Chem. 2003, 75, 3451-3459. (20) McShane, M. J.; Rastegar, S.; Pishko, M. V.; Cote´, G. L. IEEE Trans. Biomed. Eng. 2000, 47, 624-632. (21) McShane, M. J.; Russel, R. J.; Pishko, M. V.; Cote´, G. L. IEEE Eng. Med. Biol. Mag 2000, 19, 36-45. (22) Ibey, B. L.; Meledeo, M. A.; Gant, V. A.; Yadavalli, V.; Pishko, M. V.; Cote´, G. L.; Proc. SPIE 2003, 4965, 1-6. (23) D′Auria, S.; Herman, P.; Rossi, M.; Lakowicz, J. R. Biochem. Biophys. Res. Commun. 1999, 263, 550.

5502

Analytical Chemistry, Vol. 77, No. 17, September 1, 2005

Figure 1. Schematic of the RET glucose assay based on competitive binding between dextran and glucose for binding sites on apo-GOx.5

It is believed that this is the first demonstration of a system wherein a fluorescent assay for glucose is produced by encapsulating a competitive binding assay within hollow nanoengineered polymeric microcapsules; this approach enables the reuse and potential implantation of the assay while many characteristics of solution-phase behavior are retained. It is noteworthy that the sensing approach retains the advantages of competitive binding assays (no analyte consumption) and RET-based sensors (ratiometric, high sensitivity) while adding the high specificity of enzymatic sensors. This report describes the investigation of the loading properties and glucose sensitivity of FD/TAG complexes encapsulated in polyelectrolyte microcapsules. The loading efficiency and glucose sensitivity of the microcapsules containing FD/ TAG complexes were studied using confocal microscopy, UVvis absorbance, and fluorescence spectroscopy. The effects of donor-acceptor pair concentration and dextran molecular weight on the loading, leaching, and sensor response were investigated, as these factors have previously been shown to influence assay properties.24,25 Such microscale systems with encapsulated dextran/apo-GOx molecules may potentially be used for in vivo glucose sensing.20,21,26 More importantly, these experiments lay the groundwork for an expansion of this concept to similar designs for sensing a wide variety of analytes and controlled drug delivery applications, where microcapsules can be used to encapsulate apoenzyme/competitive ligand binding systems. SYSTEM DESCRIPTION A schematic of the assay design is illustrated in Figure 1. Briefly, apo-GOx is prepared from GOx by removing the cofactor FAD, which results in the loss of catalytic activity of the enzyme GOx while the binding specificity of GOx toward β-D-glucose is retained.5 When apo-GOx is tagged with TRITC and exposed to FD, strong fluorescence peaks due to energy transfer between FITC and TRITC are observed (Figure 1c). Because of the high (24) Rolinski, O. J.; Birch, D. J. S.; McCartney, L. J.; Pickup, J. C. Spectrochim. Acta 2001, 57, 2245-2254. (25) Meadows, D. L.; Schultz, J. S. Biotech. Bioeng. 1991, 37, 1066-1075. (26) McShane, M. J.; Brown, J. Q.; Guice, K. B.; Lvov, Y. M. J. Nanosci. Nanotech. 2002, 2, 411-416.

affinity of the glucose toward apo-GOx, the addition of glucose will result in the displacement of dextran from apo-GOx, accompanied by a decrease in the energy transfer efficiency (Figure 1d), as indicated by a stronger FITC peak relative to TRITC. This work employs the elegant LbL self-assembly approach for the fabrication of microcapsules, which are used as microscale packages for the RET assay elements.27 The wall permeability can be adjusted by proper choice of materials and treatments; for example, microcapsules with walls comprising diazoresin (DAR) and poly(styrene sulfonate) (PSS) have been shown to possess decreased permeability to macromolecules following exposure to UV light due to the conversion of weak interactions between the neighboring layers in the capsule walls to covalent bonds.28-30 Recently, this concept has been utilized for the stable encapsulation of glucose oxidase into microcapsules.28 Thus, we intend to achieve the stable encapsulation of sensing elements (FD/TAG) within microcapsules by exploiting the photosensitive properties of DAR. The resulting microcapsule comprises an optical glucose sensor, with sensing materials encapsulated in the hollow capsule interior. EXPERIMENTAL SECTION Materials. FD (2 MDa, 500 kDa), GOx (G-2133), sodium poly(styrene sulfonate) (PSS, 1 MDa), poly(allylamine hydrochloride) (PAH, 70 kDa), β-D-glucose, mannose, R-D-glucose, sucrose, sodium bicarbonate, dimethyl formamide, ammonium sulfate, and sodium acetate buffer were obtained from Sigma. TRITC (Molecular Probes) was used to label apo-GOx using a standard amine labeling procedure.31 Diazoresin (Diazo-10, 4-diazodiphenylamine/ formaldehyde condensate hydrogen sulfate-zinc chloride salt, DAR) was purchased from PC Associates, NJ. All reagents were used as received. MnCO3 (5 µm) particles were prepared as previously described.32 Instrumentation. A UV-vis absorbance spectrometer (Perkin-Elmer Lambda 45) was used to collect absorbance spectra. The slit size (4 nm) and scanning speed (480 nm/min) were held constant throughout all the experiments. A scanning fluorescence spectrometer (QM1, Photon Technology International) was used to collect fluorescence emission spectra by exciting the sample at 480 nm. A 100-W longwave UV lamp (Blak-ray Model B 100AP, Entela) was used to irradiate the microcapsules for photo-crosslinking PSS and DAR layers in the capsule walls. Confocal images were taken with a Leica TCS SP2, equipped with a 63× oil immersion objective and Ar/ArKr, HeNe lasers. Counts and sizes of microcapsules were obtained with a Beckman Coulter counter (Z2) using a 100 µm aperture. Methods. Solution-Phase Experiments. All RET experiments were initially performed in deionized (DI) water. Changes in RET between FD and TAG were monitored using a fluorescence (27) McShane, M. J.; Lvov, Y. M. In Dekker Encyclopedia of Nanoscience & Nanotechnology; Schwartz, J. A., Contescu, C. I., Putyera, K., Ed.; Marcel Dekker: New York, 2005; p 1-24. (28) Zhu, H.; McShane, M. J. Langmuir 2005, 2, 424-430. (29) Srivastava, R.; Quincy, J. B.; Zhu, H.; McShane, M. J. Biotech. Bioeng. in press. (30) Zhang, Y.; Yang, S.; Guan, Y.; Xu, J. Cao, W. J. Macromolecules 2003, 36, 4238-4240. (31) Svensson, H. P.; Kadow, J. F.; Vrudhula, V. M.; Wallace, P. M.; Senter, P. D. Bioconjugate Chem. 1992, 3, 176-181. (32) Zhu, H.; Stein, E.; Lu, Z.; Lvov, Y.; McShane, M. J. Chem. Mater. 2005, 17, 2323-2328.

spectrometer, by exciting the sample at 480 nm and collecting emission across the range 500-650 nm. Initially, 270 pM of FD (donor) was added to 1 mL of DI water. Fluorescence spectra were then collected after each titration of TAG into the sample solution. To show that the glucose has higher affinity toward apoGOx over FD, RET changes were measured after each titration of 100 mg/mL β-D-glucose solution into the sample containing FD/TAG complexes. Sensitivity curves of fluorescence peak ratio versus glucose concentration were constructed. The effect of dextran molecular weight was studied by performing the same sensitivity experiments with equal molar concentrations (270 pM) of 500 kDa and 2 MDa FITC-dextrans. The effect of FD/TAG complex concentration on the displacement behavior was investigated by repeating the same procedure at different total concentrations of FD-500 kDa/TAG complexes and maintaining approximately constant 1:1 ratio of glucose moieties on FD-500kDa:TAG. To demonstrate the specificity of the sensor for glucose, the RET glucose sensitivity measurements were repeated with the titration of mannose, R-D-glucose, and sucrose into a solution of FD-500kDa/TAG complexes, with maintaining all other parameters constant. Fabrication of Diazoresin-Based Hollow Microcapsules. Solutions of PSS (anionic), PAH (cationic), and DAR (cationic) used for assembling {PSS/PAH} and {PSS/DAR} multilayers were prepared in DI water at 2 mg/mL. To obtain monodispersed samples, MnCO3 particles (5 µm) dispersed in DI water were sonicated for 10 min, prior to LbL assembly. As the core particles are positively charged, they were coated with one bilayer of {PSS/ PAH} and then three bilayers of {PSS/DAR}, followed by one layer of {PSS/PAH/PSS} as the outer layer, as illustrated in Figure 2a-d. During each adsorption step, the particles were suspended in the polymer solution for 15 min, followed by rinsing with DI water three times to remove excess polyelectrolyte. The final architecture of the shell structure was {(PSS/PAH)(PSS/DAR)3(PSS/PAH/PSS)}. It is noteworthy that the weak interaction between PSS/DAR layers could result in the disintegration of the capsule wall during the core dissolution process. To avoid this potential problem, initial and final bilayers of PSS/PAH were used to obtain a more stable shell structure.28 Hollow microcapsules were obtained by dissolving the MnCO3 cores using 0.1 M hydrochloric acid (HCl) solution for 20 min, followed by rinsing in DI water 4 times to remove excess HCl solution, as depicted in Figure 2e. Encapsulation of Sensing Chemistry in DAR-Based Microcapsules. Microcapsules prepared as described are permeable to macromolecules such as enzymes prior to UV irradiation but become impermeable after photo-cross-linking.28 To encapsulate FD-500kDa/TAG and FD-2MDa/TAG complexes in the hollow shells, the capsule suspension was split into two different batches with approximately equal number of capsules (∼107). In the first case, the microcapsule suspension was mixed with the 0.95 µM FD-500kDa solution for 5 min, followed by the addition of 12.5 µM TAG solution. The capsule suspension was incubated in the mixture of FD/TAG solution for 60 min, which allowed enough time for diffusion of loading molecules into the capsule interior (Figure 2f). Confocal microscopy was used to visualize the loading process of FD/TAG complexes into microcapsules. A similar procedure was followed for complexes containing larger dextran, Analytical Chemistry, Vol. 77, No. 17, September 1, 2005

5503

Figure 2. Fabrication of microcapsules and encapsulation of FD/TAG complexes.

with 0.3 µM of FD-2MDa and 12.5 µM of TAG. In each case, the capsules were irradiated, while still in the FD/TAG loading solution, using a UV lamp for 10 min to cross-link the DAR and PSS layers of the capsule walls. Ultraviolet exposure was observed to result in a decrease of permeability of the multilayer wall, which stopped the outward diffusion of the encapsulated molecules from the microcapsule. The loaded capsules were then rinsed in DI water three times to remove residual loading solution (Figure 2g). Estimation of the Encapsulation Efficiency. The procedure outlined below is the standard method used to calculate the loading efficiency of sensing assay elements in microcapsules. It is well-understood that encapsulation of the sensing elements is based on diffusion from the loading solution, and the loading rate is therefore directly proportional to the initial concentration of the loading solution. However, estimation of the final amount of loading solution encapsulated in the microcapsules is important to study the effects of ligand molecular weight on sensor response. The FD and TAG concentrations in the loading solution (Cload) were determined using UV-vis absorbance spectroscopy, prior to the addition of hollow microcapsules. As described above, following suspension of the microcapsules in loading solution and exposing them to UV light, the capsules were rinsed three times and the supernatants were collected after each rinse. For each component, the concentrations in all the supernatants (e.g., Crinse ) Crinse1 + Crinse2 + Crinse3) were determined by UV-vis and subtracted from Cload to estimate the amount of the encapsulated (Cencap) material. Encapsulation efficiency was estimated as the ratio of the total number of FD or TAG moles loaded to number of capsules N in the corresponding sample.

Cencap ) Cload - Crinse E)

Cencap N

(1)

Stability of Encapsulation. For potential use as sensors, it is important for the encapsulated material to be retained in the microcapsule over time; it is expected that leaching will depend on the molecular weight of the potentially diffusing species. An experiment for the assessment of stability of the encapsulated material was performed on two sets of capsules, loaded with FD500kDa/TAG and FD-2MDa/TAG complexes, respectively. To 5504

Analytical Chemistry, Vol. 77, No. 17, September 1, 2005

estimate the percentage of loaded material lost from the capsules, leaching studies were performed by separating the supernatant solution from microcapsules via centrifugation and measuring the fluorescence from the supernatant. After each measurement, the supernatant solution was placed back with the microcapsule suspension in DI water. All samples were covered and stored at 4-8 °C under dark conditions, until the completion of all the experiments. Three replicate measurements were collected for each sample at 0, 2, 4, 10, 20, 30, and 50 h. Standard solutions of 26 nM FD-500kDa, 8 nM FD-2MDa, and 333 nM TAG were also measured at each point in time to correct for instrumental drifts over time and also to estimate the supernatant concentration as a percentage of a standard. The FD peak was obtained by exciting the supernatant at 480 nm and collecting emission from 500 to 650 nm, while the TAG peak was obtained by exciting at 543 nm and collecting from 555 to 650 nm. FD and TAG peaks of the standard solutions were obtained in a similar manner. From loading efficiency and capsule counts, the FD and TAG concentrations present were calculated, and the percentages of the FD and TAG leached were estimated. Sensor Response in Microcapsules. A fluorescence spectrum of the microcapsules loaded with the FD/TAG complexes dispersed in DI water was collected as the starting point. Fluorescence spectra were then collected after each addition of 100 mg/mL glucose to the microcapsules. The change in FITC to TRITC peak intensity ratio was calculated from each spectrum and plotted with respect to the increments in glucose concentrations. To observe the effect of molecular weight on sensor response and also to determine the feasibility of tailoring the sensor response, glucose sensitivity experiments were repeated for two different ligand sizes and three different concentrations of capsules. In the first case, capsules loaded with (FD-500kDa)/ TAG and (FD-2MDa)/TAG were assessed. Next, three different concentrations of (FD-500kDa)/TAG-loaded capsules (∼106, 2 × 106, 3 × 106 capsules/mL) were considered. Reversibility. To test the reversibility of these sensors, loaded microcapsules were suspended in glucose solutions in random order with respect to concentration. To achieve this, FD/TAGloaded microcapsules (4 × 105) were dispersed in DI water (0.5 mL), and the glucose concentration of the suspension was increased by addition of glucose stock solution. The glucose levels were changed by rinsing the capsules to remove glucose after

Figure 3. Normalized fluorescence spectra with the addition of (a) TRITC-apo-GOx (TAG) to FD-500kDa and (b) β-D-glucose into the sample containing FD/TAG complexes. (c) Percentage change in the FITC/TRITC intensity peak ratio with the addition of glucose. Error bars denote one standard deviation of three replicate measurements.

each measurement and then again adding glucose to a new concentration. In each case, after addition of glucose to the desired concentration and measurement of the fluorescence, the suspension was centrifuged, the supernatant (glucose) solution was removed, and, finally, an equal volume of fresh DI water was added. The change in FITC/TRITC peak ratio was obtained at each step by collecting a fluorescence scan. This procedure was repeated to cover the glucose concentration in the range of 0-90 mM, in the order of 0, 4.15, 0.35, 9, 2.8, 0.02, 5.5, 19, 13.56, 60, 26.45, and 95mM. RESULTS AND DISCUSSION Solution-Phase Response Tests. Fluorescence spectra collected during titration of TAG into a sample containing FD are shown in Figure 3a. From these spectra, it can be observed that the FITC/TRITC peak intensity ratio (515 nm/575 nm) decreases with each addition of TAG, indicating an increase in energy transfer due to close proximity of FD and TAG molecules (direct

excitation of TRITC with 480 nm light is minimal). To confirm that apo-GOx binds glucose with higher affinity than dextran, RET changes were monitored with the titration of 100 mg/mL β-Dglucose solution into the sample containing FD/TAG complexes. The FITC peak increased relative to the TRITC peak, as shown in Figure 3b, due to the displacement of dextran from apo-GOx by glucose molecules. This displacement results in a decreased average energy transfer from FITC to TRITC and is reflected as a decrease in the FITC/TRITC intensity ratio. For all RET experiments, sensitivity curves were obtained by plotting the percentage change in FITC/TRITC peak ratio versus glucose concentration (Figure 3c), and the slope of the linear region was calculated as a measure of response sensitivity. In the case where FD-500kDa was used as the competitive ligand, a total change of 17% in energy transfer was observed due to 40 mM glucose, with a sensitivity of 0.5%/mM in the region of interest (0-40 mM). The same procedure was repeated with FD-2MDa, from which it was observed that there was a 7.5% change in RET with the addition of 40 mM glucose. The increase in dextran molecular weight was associated with ∼2.5 times decrease in sensitivity and ∼3 times increase in the dissociation constant, as shown in the inset of Figure 3c. A key point to note for these data is that the calculated dissociation constants closely match reported values for glucose oxidase.33 Thus, the processes required for cofactor removal and labeling of apo-GOx do not destroy the glucose binding behavior of the enzyme. This suggests that the binding of glucose and dextran is occurring in the active site, as opposed to other regions of the enzyme that may be adhesive for sugars. Furthermore, this finding is critical for sensors intended for physiological monitoring, as the higher kd values (millimolar range) more closely match the relevant range for glucose than the micromolar dissociation constants possessed by glucose binding proteins. It is also noteworthy that the observed variation in the sensor response with different molecular weights of dextran could be due to an overall increase in FD/TAG complex concentration in the sample. To achieve equal RET efficiency with different molecular masses of FD, the ratio of the number of glucose residues (proportional to the molecular mass of FD) on FD to the number of binding sites on TAG must be identical. As the number of TAG molecules that can bind to FD is directly proportional to glucose residues on FD, the TAG concentration at which the energy transfer saturated during TAG titration increases proportionally with dextran molecular weight. This results in a total increase in FD/TAG complexes present in the assay, which translates directly to a change in sensor response. This explains why the glucose sensitivity experiments (Figure 3c) show that 80 mM glucose is required to reach the saturation point for the larger dextran, whereas the saturation was reached with 40 mM glucose for the smaller molecule. Additional experiments conducted with lower FD molecular masses (77 and 250 kDa) indicated that the sensitivity was increasing and the saturation concentration was lower with decreasing FD molecular mass (results not shown). These findings indicate that the sensitivity and range of the system can be “tuned” to some extent by controlling FD molecular mass. (33) Swoboda, B. E. P.; Massay, V. J. Biol. Chem. 1965, 240, 2209-2215.

Analytical Chemistry, Vol. 77, No. 17, September 1, 2005

5505

Figure 4. Percentage change in FITC/TRITC peak ratio with the addition of glucose into different FD/TAG complex concentrations. Lines are regression curves used only to clearly indicate the trend for each set of measurements. Error bars show one standard deviation of three replicate measurements.

Figure 5. Relationship of sensitivity (%/mM) and dissociation constant (kd) to concentration of assay components (constant mole ratio).

Effect of Concentration on Sensor Response. To further explore the tunability of the system, sensitivity experiments were repeated at four different concentrations of FD-500kDa/TAG complexes. For these tests, the total FD/TAG complex concentration was varied over a wide range, but the ratio of FD/TAG mass was held constant in all experiments. An initial fluorescence spectrum was collected in each case for FD/TAG complex concentrations of 38 pM/106 nM, 125 pM/298 nM, 275 pM/694 nM, and 1100 pM/2620 nM, respectively. Fluorescence spectra were collected after each addition of glucose to the FD/TAG complexes and, as expected, there was a change in the FITC/ TRITC peak ratio with the addition of glucose, which is plotted as percentage change in normalized peak ratio in Figure 4. In all cases, glucose solution was added until there was no further change in the FITC/TRITC peak ratio value, indicating a saturated response, except for the highest concentration, where the response did not reach saturation, even with the addition of 300 mM glucose. The sensitivity and dissociation constant (kd) for these four systems are plotted versus complex concentration over the range of 0-30 mM glucose concentration in Figure 5. It is evident that the concentration of complexes present dramatically affects both the sensitivity and the range of the response. With an ap5506 Analytical Chemistry, Vol. 77, No. 17, September 1, 2005

Figure 6. Percentage change in the normalized FITC/TRITC peak ratio values with the addition of different sugars. Error bars denote one standard deviation of three replicate measurements.

proximately 30 times increase in the FD/TAG complex concentration, the sensitivity was reduced by 4 times and dissociation constant was increased by approximately 5 times the value for the lowest concentration. The required glucose concentration to reach the saturation point was increased approximately 10 times, from 30 to 300 mM, when the FD/TAG complex concentration increased approximately 30 times (FD concentration from 38 to 1100 pM). These results agree with the previous studies on the effect of receptor/ligand ratio on sensor response using a single dimensionless equation.25 By comparing the results obtained in the current and previous sections, it can be concluded that the sensor response follows the same trend with the increase in dextran molecular weight and FD/TAG complex concentration, because the final condition in both cases is identical; i.e., there is an increase in FD/TAG complex concentration. Thus, the sensor properties can be tailored by varying the dextran molecular weight or FD/TAG complex concentration. Specificity. It is known that glucose oxidase exhibits a high degree of specificity for β-D-glucose; it has been reported that R-Dglucose, D-mannose, and D-fructose are also oxidized by the enzyme, but at least at a 20-fold reduced rate.34 This suggests that the binding specificity for the molecules is likely different; however, in the case of the apoenzyme, the specificity had to be experimentally determined, as no previous reports were found. To observe the effect of different sugars on the response of the FD/TAG glucose assay, the sensitivity experiments were repeated with β-D-glucose, mannose, R-D-glucose, and sucrose. For these experiments, the starting concentrations for the assay complexes were 270 pM:170 nM FD:TAG. To observe the specific nature of the sensor, the different sugar solutions were titrated into the sample in a stepwise manner until the RET reached saturation. The corresponding percentage changes in FITC/TRITC peak intensity ratios versus sugar concentration are plotted in Figure 6. It can be observed from the response curves that the maximum sensitivity was achieved with β-D-glucose, and there was ∼5-10 times lower sensitivity for mannose, sucrose, and R-D-glucose. Total percentage changes in peak ratio with addition of different (34) Wohlfahrt, G.; Witt, S.; Hendle, J.; Schomburg, D.; Kalisz, H. M.; Hecht, H. J. Acta Crystallogr. 1999, D55, 969-77.

Figure 7. Confocal images of the microcapsules (a and b) in FD-500kDa/TAG complex loading solution with FITC and TRITC fluorescence; (c and d) with FITC and TRITC fluorescence, respectively, after UV exposure and rinsing in DI water; (e) phase transmission image of the loaded microcapsules; (f) overlay of FITC, TRITC fluorescence and transmission mode image. Fluorescence line scan of the capsules (g) during loading process (from a and b) and (h) after encapsulation process (from c and d).

sugars also follows the same trend as sensitivity, i.e., with the addition of 50 mM β-D-glucose, mannose, R-D-glucose, and sucrose solutions there is 16.9%, 6.0%, 5.9%, and 2.5% change in RET, respectively. The dissociation constants for β-D-glucose, mannose, R-D-glucose, and sucrose were observed to be 14, 24, 46, and 3 mM, respectively. It can be observed that, in the case of sucrose, the range of the linear region is reduced by more than 10 times compared to glucose. This large difference in sensor response is attributed to the significant variation of sucrose structure (disaccharide) from that of β-D-glucose (monosaccharide), which results in very weak hydrogen-bond interactions between sucrose and apo-GOx.34 The above results prove that the apo-GOx retains its binding specificity for β-D-glucose, and the presence of low levels of other sugars will not significantly interfere with accurate measurement of glucose concentration. Encapsulation of FD/TAG Complexes in Microcapsules. Microcapsules with {(PSS/PAH)(PSS/DAR)3(PSS/PAH/PSS)} shell architecture were prepared as described and used for encapsulating FD/TAG complexes. Both FD-500kDa/TAG and FD-2MDa/TAG complexes were encapsulated in parallel experiments. Confocal images of representative polyelectrolyte microcapsules loaded with the glucose assay (there was no obvious difference between capsules loaded with FD-500kDa and FD2MDa) are shown in Figure 7. It can be observed from the images of capsules in loading solution (Figure 7a,b) and the corresponding line scan in Figure 7g that there is an equal concentration of FD/TAG in the interior and exterior of the capsules. This was expected, because the semipermeable capsule walls allow diffusion of macromolecules. However, the exposure of the microcapsules to UV light results in the cross-linking of the adjacent DAR and PSS layers in the capsule wall, decreasing the permeability of the capsule walls to FD/TAG complexes. As the final step, the capsules were rinsed to remove excess FD/TAG complexes, yielding microcapsules loaded with the sensing assay (Figure 7c-f). It can be observed

from Figure 7c,d that the FITC and TRITC fluorescence is concentrated in the capsule interior and walls, indicating the immobilization of FD/TAG complexes inside and on the capsule walls. When UV light was not used to irradiate capsules before rinsing in DI water, the resulting capsules showed weak fluorescence from the capsule walls with no fluorescence from the solution in the interior of the capsules. High encapsulation in the walls could be due to a combination of electrostatics (residual charged residues of polyelectrolytes in walls), van der Waals forces, different solubility/partitioning, and physical entrapment of the sensing elements (upon cross-linking). It is likely that electrostatic forces of attraction between negatively charged FD/ TAG complexes and positive polymer layers in the capsule walls dominate all the other types of interactions, due to the relative magnitude of the forces. It is noteworthy that association with capsule walls is seen without the wall cross-linking, so the contribution of the latter effect is expected to be minimal. From Figure 7e,f, which clearly shows the capsule walls before and after encapsulation, and from the line scan (Figure 7h) of the loaded and rinsed capsules, it can be observed that the encapsulated FD/TAG complexes are distributed uniformly in the hollow interior and on capsule walls. The mismatch of the shapes of the line scans in Figure 7g,h is due to the loading protocol used. It can be observed from the intensity scales that the average intensity value inside the capsule (between 2.2 and 5.8 µm) is almost the same. There are no peaks at capsule walls for FD line scans in Figure 7g, because the dextran can readily diffuse into capsules without any entrapment in the walls. However, during the encapsulation of TAG after loading of FD, it was observed that a much stronger green fluorescence signal was associated with the capsule walls. It is believed that the dextran was attracted to the TAG, which was entrapped in the walls due to its structure and charge. Therefore, after rinsing, a fraction of the FD and TAG was entrapped in the capsule walls, which resulted in the strong peaks in FD-line scans in Figure 7g. It is noteworthy that the broad Analytical Chemistry, Vol. 77, No. 17, September 1, 2005

5507

Table 1. Encapsulation Efficiency Parameters for FITC-Dextran of Two Different Molecular Weights calculated parameters no. of FD moles and mass (pg)/capsule no. of TAG moles and mass (pg)/capsule FD/TAG mole ratio in a capsule no. of glucose residues/FD molecule no. of glucose moles/capsule no. of glucose residues/TAG molecule

peaks of TAG in Figure 7g are due to the presence of TAG loading solution, still present when the image was collected, which creates a concentration gradient near capsule walls. However, after the excess TAG is was removed during the rinsing process, the high intensity from the walls results in apparently sharper edges in the image (Figure 7h). The encapsulation efficiency of FD/TAG complexes was estimated by measuring the absorbance of the rinse solutions and obtaining the capsule count using the Coulter counter. The results, presented in Table 1, prove that each capsule contains ∼10-17 and 10-15 mol of FD and TAG, respectively, in the capsules loaded with FD-500kDa/TAG and FD-2MDa/TAG complexes (the mass of the loaded FD and TAG was on the same order of magnitude for the two sets of capsules considered). It can be observed from Table 1 that the FD/TAG mole ratio is slightly higher in the case of smaller dextran, due to encapsulation of more moles of FD500kDa relative to FD-2MDa. Using the molecular weights of dextran and glucose molecules, it was estimated that there are 1.27 × 10-13 and 2.16 × 10-13 total glucose residues in the capsules loaded with FD-500kDa/TAG and FD-2MDa/TAG complexes, respectively. The greater number of glucose residues present more binding sites for apo-GOx molecules, which is supported by the calculations in Table 1. It can be observed that there are more TAG molecules in the capsules loaded with FD-2MDa/TAG (1.07 × 10-15) compared to FD-500kDa/TAG (7.85 × 10-16), because of more glucose residues in the former case. Even though there is a 1.7 times increase in the number of glucose residues in FD-2MDa/TAG-loaded microcapsules over FD-500kDa/TAGloaded capsules, there is only 1.3 times increase in the corresponding TAG concentration. While this suggests that the apoGOx/dextran association does not scale linearly with the dextran length, further investigation on what factors affect encapsulation efficiency will be required to establish a clear mathematical relationship. However, these results do prove that the amount of sensing assay in the microcapsules does vary with molecular weight, and the sensor response characteristics depend on the encapsulated assay concentration, ligand size, and FD/TAG ratio. Leaching of Encapsulated Molecules from Microcapsules. Loss of FD and TAG molecules from microcapsules was quantified using fluorescence spectroscopy studies, where release of the encapsulated molecules from the microcapsules was observed as an increase in the fluorescence intensity of the supernatant. The results obtained from this experiment are given in Figure 8, where the percentage leached indicates the percentage of initially encapsulated FD and TAG molecules lost from the capsules to the supernatant solutions. These experiments were performed on capsules loaded with FD-500kDa/TAG and FD-2MDa/TAG complexes. It is clear from the data that the leaching is higher for the capsules loaded with FD-500kDa/TAG, but still less than 4%. This 5508

Analytical Chemistry, Vol. 77, No. 17, September 1, 2005

500 kDa 10-17/25.4

5.1 × 7.9 × 10-16/125.2 6.5 × 10-2 2500 1.3 × 10-13 1.6 × 102

2 MDa 2.2 × 10-17/43.3 1.1 × 10-15/171.3 2.0 × 10-2 10000 2.1 × 10-13 2.0 × 102

Figure 8. Percentage of the initially encapsulated FD and TAG in the microcapsules lost to the supernatant; 500K-FD and 500K-TAG, and 2M-FD and 2M-TAG indicate the moles of FD and TAG in the supernatant corresponding to the capsules loaded with FD-500kDa/ TAG and FD-2MDa/TAG complexes, respectively.

result is intuitive, because the diffusivity scales as the inverse of the square root of molecular mass, and both free and complexed forms of FD-500kDa will have larger diffusivity than the corresponding free and complexed forms of FD-2MDa. It is also noteworthy that dextran molecules, which possess a linear structure with branches, are expected to diffuse more readily through the capsule walls compared to apo-GOx molecules (globular structure) because of the lower hydrodynamic volume. This trend can be observed in Figure 8, which shows that the amount of FD leached is always greater than the corresponding TAG concentration. And, as the TAG is bound to FD, relatively few TAG (2M-TAG) molecules can diffuse out of the capsule in the case where TAG is bound to FD-2MDa. It is very likely that the 4% loss in the loaded materials is due in part to molecules weakly trapped in or on the capsule walls, because there is no significant barrier to resist diffusion of the assay molecules. It is noteworthy that in all cases there is only a small amount of leaching, i.e., about 1-4% of the initially encapsulated molecules that occurs during the first 10 h of leaching studies, after which there was no change in the fluorescence intensities from the supernatant solutions. Thus, the photo-cross-linking technique appears to be an appropriate approach for stable encapsulation of the elements for the competitive binding assay. Sensor Response to Glucose Addition in Microcapsules. Glucose sensitivity experiments were performed on the microcapsules loaded with FD-500kDa/TAG and FD-2MDa/TAG complexes. The fluorescence spectrum was collected after each addition of β-D-glucose to microcapsules loaded with FD/TAG complexes, and the corresponding percentage change in the

Figure 9. Percentage change in the normalized FITC/TRITC peak ratio with the addition of glucose to the microcapsules loaded with FD-500kDa/TAG and FD-2MDa/TAG complexes. Error bars show one standard deviation of three replicate measurements. Table 2. Details of the Sample Used for Testing Sensor Response in Microcapsules

no. of FD moles (pmol) no. of TAG moles (pmol) no. of glucose residues/FD no. glucose moles (pmol) no. of glucose residues/TAG

500 kDa

2 MDa

20.5 315.5 2500.0 51219.5 162.3

6.7 329.5 10000.0 66659.5 202.0

FITC/TRITC peak ratio was calculated for each case, as shown in Figure 9. It can be seen that there is a linear increase in the peak ratio with the addition of glucose up to 30 mM for the capsules loaded with FD-500kDa/TAG complexes, with a sensitivity of 2.75%/mM. The signal saturation after 30 mM indicates that all the available binding sites on apo-GOx were occupied by glucose molecules. By closely examining the curves in Figure 9, it becomes obvious that there is a small difference between the glucose responses of the two sets of capsules. The sensitivity over the linear region (0-30 mM) for capsules loaded with FD-2MDa/ TAG complexes was 2.5%/mM. A possible reason for this is the difference in encapsulation efficiency, which will be discussed below in more detail. By using the encapsulation efficiency and back-calculating the number of FD and TAG molecules in both samples, it was found that the ratio (number of glucose moieties/ TAG) is approximately equal in the two sets of capsules loaded with FD-500kDa/TAG and FD-2MDa/TAG complexes (Table 2). Thus, the minimal variation in the sensor response of the two sets of microcapsules is attributed to the insignificant difference in the concentration of sensing assay elements. It was previously found that there is a considerable variation in the sensitivity and dissociation constant with different molecular masses of dextran for solution-phase experiments, which is due to the high FD/TAG complex concentrations and not merely due to the increase in molecular mass. In the case of microcapsules, however, the encapsulation efficiency is limited by the molecular mass of the loading materials and capsule size. Therefore, the loading procedure resulted in the encapsulation of almost equal mass of FD and TAG per capsule in the capsules, loaded with FD-500kDa and FD-2MDa (Table 1). As opposed to solution phase

Figure 10. Effect of capsule concentration on the change in FITC/ TRITC peak ratio with the addition of glucose solution. Error bars show one standard deviation of three replicate measurements.

experiments, there is no considerable variation in the sensing assay concentration with the encapsulation of FD-500kDa/TAG and FD-2MDa/TAG into microcapsules (Table 2). Thus, there is no significant difference in the sensor response of the two sets of capsules. This explanation agrees with the solution-phase experimental results (Figure 4 and 5), which show that there is a significant change in the sensor response mainly with the variation in FD/TAG complex concentration. Thus, it can be concluded that the sensor response is mainly dependent on the number of ligand and receptor molecules that are available, and not the size of the ligand molecule. The small difference in sensor response of the two sets of microcapsules could be due to the difference in the encapsulation efficiency or the behavior of dextran molecules with different molecular weights due to the conformation of dextran chains in the capsule interior. Effect of Capsule Concentration on Sensor Response. It was demonstrated in solution-phase experiments that the sensor response can be tailored by varying assay concentrations. To test the feasibility of tailoring the sensor response with the sensing assay encapsulated in microcapsules, the glucose sensitivity experiments were performed with FD-500kDa/TAG loaded capsules at three different concentrations: 1 × 106, 2 × 106, and 3 × 106 capsules/mL. Fluorescence scans were collected at each addition of glucose solution to the capsules, and the corresponding changes in the FITC/TRITC peak ratios for all three samples are plotted in Figure 10. It is immediately evident that with 3 times increase in capsule concentration, there is a decrease in the total change in the percentage of RET, from 73% to 58%, and also a higher glucose concentration is required to reach the FITC/ TRITC peak ratio saturation point. It can be seen from Table 3 that a 3 times increase in capsule concentration results in a 3.1 times decrease in sensitivity and 4 times increase in dissociation constant; this trend corroborates the solution-phase experimental results. These results prove that the sensitivity can be tailored by changing the competitive binding assay concentration, regardless of the environment of the assay. It can be observed from the microcapsule and solution-phase glucose sensitivity experiments that there is an approximately 2-fold increase in the sensitivity in the case of microcapsules. This Analytical Chemistry, Vol. 77, No. 17, September 1, 2005

5509

Table 3. Change in Sensitivity and Dissociation Constant with Varying Capsule concentration

capsules/mL

FD (mol)

TAG (mol)

sensitivity (% ∆ ratio/mM)

1.0 × 106 2.0 × 106 3.2 × 106

2.1 × 10-11 4.1 × 10-11 6.7 × 10-11

3.2 × 10-10 6.3 × 10-11 1.0 × 10-9

6.11 2.67 1.95

Figure 11. Percentage change in the normalized FITC/TRITC peak ratio with the addition of random glucose concentrations to the microcapsules loaded with FD-500kDa/TAG complexes. Error bars show one standard deviation of three replicate measurements.

observation is the result of the low concentration of mobile sensing assay elements, due in part to the entrapment of ∼70% of the total encapsulated molecules in the capsule walls. The apo-Gox and dextran molecules that are entrapped in the walls may not dissociate in the presence of glucose molecules; thus, the total number of mobile molecules will be far less than in solution-phase samples. This low concentration (∼30%) of freely moving FD/ TAG complexes is ultimately reducing the active sensing assay concentration, which results in the increased sensitivity, as predicted by theory and experimentally found in studying the effect of competitive binding assay concentration on response (Figure 5). The response time of this system was not measured separately, because the sensor response was observed to reach steady-state in approximately 1-2 min, which is the approximate time required for the collection of one fluorescence scan. There was no observable change in the response time with the variation in assay elements concentration and ligand molecular weight. The response kinetics will be the subject of future studies, but clearly the response occurs in a reasonable time and will not be a limiting factor for implementation of this system for glucose sensing. Reversibility of the Glucose Sensors. The reversibility of the fabricated sensors was tested by exposing FD-500kDa/TAG loaded microcapsules to random glucose concentrations. Fluorescence spectra were collected after each addition and removal of glucose solution from the microcapsules, and the corresponding changes in FITC/TRITC peak ratios are plotted in Figure 11. It is clear that there is a linear response in the 0-10 mM range, and a decreased yet measurable response up to 20 mM, which covers much of the region of interest for glucose monitoring in diabetics. Sensitivity with random glucose concentrations was calculated to be 2.5%/mM, with a total sensitivity of more than 5510 Analytical Chemistry, Vol. 77, No. 17, September 1, 2005

total change in peak ratio/100 mM glucose (%)

linear range (mM)

kd (mM)

73 66 58

0-10 0-30 0-40

5 12 19

50%, which is comparable to the sensitivity obtained with capsules exposed to stepwise increase in glucose concentrations. These results show that the microcapsule-based sensors are completely reversible, without any considerable loss in sensitivity. Further assessment of the forward and reverse response kinetics is currently being studied with a flow-through system. In the above demonstrated system, the use of prefabricated microcapsules for carrying the sensing assay is advantageous,8 because it allows (a) for the free movement of sensing elements during competitive binding process, which is critical for proper functioning of the sensor, which relies upon the equilibrium association and dissociation of free molecules, and (b) the maintenance of the constant concentrations of the ligand and receptor with continuously varying analyte concentration. The generic sensor design described here opens the door for a wide variety of analytes to be sensed using the versatile sensing technique, which involves selection of analyte-specific enzyme and a competing ligand; it can be considered a platform technology for development of biosensors based on competitive-binding and fluorescence RET-based techniques. Potential targets include neurochemicals such as glutamate and choline, lipids, and estrogen.35 The general approach to construction of specific “smart” responsive systems may be useful in other applications, such as drug delivery, wherein the apo-GOx/dextran dissociation determines release of an encapsulated compound in response to glucose.36 With specific reference to the glucose-sensing microcapsules described here, the sensors employing apo-GOx elements can potentially be used for in vivo glucose sensing without the concern of toxicity associated with Con A. It is also noteworthy that the sensors can be easily extended into the near-infrared (NIR) region by replacing the FITC-TRITC RET pair with an appropriate NIR donor-acceptor combination. It is also appropriate to note that these sensors could be interrogated using fluorescence lifetime measurements,7,37,38 which will be better suited to in vivo monitoring because they are less affected by the fluorophore concentration and optical properties of the medium. CONCLUSIONS A novel glucose-sensing system based on competitive binding, resonance energy transfer, and polymer microcapsule technology has been demonstrated. Microcapsules containing FD/TAG complexes showed a decrease in RET due to the addition of (35) Labaree, D. C.; Reynolds, T. Y.; Hochberg, R. B. J. Med. Chem. 2001, 44, 1802-1814. (36) Tanna, S.; Taylor, M. J.; Adams, G. J. Pharm. Pharmacol. 1999, 51, 10931098. (37) Tolosa, L.; Szmacinski, H.; G. Rao, G.; Lakowicz, J. R. Anal. Biochem. 1997, 250, 102-108. (38) Tolosa, L.; Malak, H.; G. Rao, G.; Lakowicz, J. R. Sens. Actuators B, 1997, 45, 93-99.

glucose, with a sensitivity ranging from 2 to 6%/mM over the range from 0 to 40 mM. It was found that the key variable in tuning the response of the system (sensitivity and dissociation constant) to glucose is the concentration of FD/TAG complexes. The assay comprising an apo-GOx recognition element showed 5 to 10 times more sensitivity to β-D-glucose compared to other sugars. On the basis of the sensitivity, specificity, dissociation constant values, and reversible response, this sensor system appears to be suitable for glucose monitoring in diabetic patients. These glucoseresponsive FD/TAG complexes can also potentially be used in the future for controlled drug delivery studies; e.g., insulin can be released due to apo-GOx/dextran dissociation in response to glucose. In vitro and in vivo studies to test the dynamic response,

biocompatibility, cytotoxicity, and stability of these sensors are in progress. ACKNOWLEDGMENT The authors gratefully acknowledge the National Science Foundation (NIRT Grant #0210298) and National Institutes of Health (R01EB000739). Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the view of the funding agencies. Received for review May 2, 2005. Accepted June 7, 2005. AC050755U

Analytical Chemistry, Vol. 77, No. 17, September 1, 2005

5511