Enzyme Nanostructures with Maskless

Antonio Peramo, Ashley Albritton, and Garrett Matthews. Langmuir 2006 22 (7), 3228-3234 ..... electron transfer systems. Hiroko I. Karan. 2005,131-178...
0 downloads 0 Views 320KB Size
Download PDF
AC Research Anal. Chem. 1997, 69, 2619-2625

Accelerated Articles

Generation of Biotin/Avidin/Enzyme Nanostructures with Maskless Photolithography Narasaiah Dontha, Wilbur B. Nowall, and Werner G. Kuhr*

Department of Chemistry, University of California, Riverside, California 92521

Micrometer-sized domains of a carbon surface are modified to allow derivatization to attach redox enzymes with biotin/avidin technology. These sites are spatially segregated from and directly adjacent to electron transfer sites on the same electrode surface. The distance between these electron transfer sites and enzyme-loaded domains must be kept to a minimum (e.g., less than 5 µm) to maintain the fast response time and high sensitivity required for the measurement of neurotransmitter dynamics. This is accomplished through the use of photolithographic attachment of photobiotin using an interference pattern from a UV laser generated at the electrode surface. This will allow the construction of microscopic arrays of active enzyme sites on a carbon fiber substrate while leaving other sites underivatized to facilitate electron transfer reactions of redox mediators, thus maximizing enzyme activity and detection of the enzyme mediator. The ultimate sensitivity of these sensors will be realized only through careful characterization of the carbon electrode surface with respect to its chemical structure and electron transfer properties following each step of the enzyme immobilization process. The characterization of specific modifications of micrometer regions of the carbon surface requires analytical methodology that has both high spatial resolution and sensitivity. We have used fluorescence microscopy with a cooled CCD imaging system to visualize the spatial distribution of enzyme immobilization sites (indicated by fluorescence from Texas Red-labeled avidin) across the carbon surface. The viability of the enzyme attached to the surface in this manner was demonstrated by imaging the distribution of an insoluble, fluorescent product. An atomic force microscope was used to obtain high-resolution images that probe the heterogeneity of the enzyme sites.

(1) Morgan, H.; Pritchard, D. J.; Cooper, J. M. Biosens. Bioelectron. 1995, 10, 841-6. S0003-2700(97)00209-6 CCC: $14.00

© 1997 American Chemical Society

Control of the architecture of immobilized molecular layers is an important area of research with potential applications in the field of biosensing, cell control and guidance, and nanotechnology.1 Photopatterning is one of the techniques used to achieve spatially distributed biomolecules, i.e., enzymes, antibodies, and nucleic acids, which are used in the development of biochips.2 Typically, photolithographic techniques have been used to immobilize proteins using various surface-bound photoactivatable reagents. Biotin/avidin chemistry has been used extensively for protein immobilization.3 Once biotin is on the surface, it is very simple to immobilize any biomolecules with an avidin label. Streptavidin is a tetramer protein that has four identical binding sites for biotin. The binding of biotin to streptavidin is almost irreversible with the binding strength comparable with a covalent bond (Ka ) 1 × 1015 M-1).3,4 Because of this strong interaction, the complex is virtually unaffected by extreme pH, temperature, and organic solvents and other denaturing agents. The tetravalency of avidin for biotin allows the construction of a “molecular sandwich” that allows the surface-bound avidin to be coupled to a biotinylated enzyme that has the appropriate characteristics (i.e., substrate consumption and product formation) needed for the construction of a biosensor. Caged biotin, a derivative of biotin that does not bind avidin, can be covalently bound to a substrate5 and then photochemically activated to induce binding. Biotin/avidin chemistry was used to make protein structures on the order of 350-500 µm simply by covering the entire substrate with caged biotin, illuminating the substrate through a mask, thereby photodeprotecting biotin in only those regions exposed to light and allowing avidin binding to occur only at the photodeprotected biotin. The avidin binding properties can subsequently be revealed photochemically by uncaging the biotin by exposure to 350 nm UV light through a (2) Fodor, S. P. A.; Read, J. L.; Pirrung, M. C.; Stryer, L.; Lu, A. T.; Solas, D. Science 1991, 251, 767-73. (3) Wilchek, M.; Bayer, E. A. Anal. Biochem. 1988, 171, 1-32. (4) Fuccillo, D. Biotechniques 1985, 3, 494-501. (5) Pirrung, M. C.; Huang, C. Y. Bioconjugate Chem. 1996, 7, 317-21.

Analytical Chemistry, Vol. 69, No. 14, July 15, 1997 2619

mask. FITC-avidin or Texas Red avidin was then immobilized onto the surface, and the pattern was fluorescently detected. Alternatively, photobiotin [a nitro(aryl)azide derivative of biotin], has been used primarily to label proteins and nucleic acids.6,7 Exposure of photobiotin to UV light induces a photochemical reaction that generates a nitrene upon irradiation, which readily inserts into C-H bonds but also other chemical groups as well. Pritchard et al.8 immobilized avidin to the surface of thiolcovered gold substrates and then added photobiotin, which binds strongly to the immobilized avidin. Illumination of this surfacebound photobiotin attached an antibody (either rat or rabbit IgG) at specific sites on this substrate. Photolithographic masks used in this case created patterns with a spacing of 1.5 µm, near the diffraction limit of the light used. Hengsakul and Cass9 demonstrated that photobiotin will bind covalently to an organic surface when it is exposed to intense UV light (350-370 nm). They illuminated the substrate (either a polystyrene microtiter plate or a nitrocellulose membrane) through a mask (50-mesh grid) and formed a pattern corresponding to structures roughly 0.6 mm on a side. Biotin attached to the substrate in this manner was bound with avidin and then with a biotinylated alkaline phosphatase or horseradish peroxidase. This demonstrated that an enzyme could be photolithographically patterned onto a surface and remain active. Similar strategies can be used in the development of techniques to produce micrometer-sized spatial segregation of sites used for enzyme immobilization from sites of electron transfer on an electrode surface. In this work, micrometer-sized domains of the carbon surface are chemically modified to attach enzymes using biotin/avidin technology. Photolithographic techniques using a laser diffraction pattern, obtained in a matter very similar to the procedures used to create finely spaced optical gratings,10,11 are used to attach photobiotin to micrometer-sized regions of the carbon surface. Either fluorophore-tagged avidin or avidinconjugated enzymes could be attached to this spatially patterned biotin with essentially no loss in spatial resolution. The pattern of spatially segregated enzyme immobilization sites was visualized with 500 nm resolution using fluorescence microscopy and a cooled CCD imaging system. A higher resolution image of the enzyme and its product was obtained using atomic force microscopy (AFM), where the heterogeneity of the enzyme attachment within each biotin structure could be visualized. Ultimately, this type of photolithographic diffraction technique should allow the construction of microscopic arrays of active enzyme sites on a electrode surface while leaving other sites underivatized to facilitate electron transfer reactions of redox mediators to maximize enzyme activity and the detection of the enzyme mediator. EXPERIMENTAL SECTION Chemicals. Phosphate buffer (150 mM NaCl and 100 mM Na2 HPO4) was prepared with DI water (E-Pure, Barnstead, Debuque, IA) before adjusting pH. Photobiotin [N-(4-azido-2nitrophenyl)-N-(3-biotinylaminopropyl)-N-methyl-1-3-propanediamine)] acetate salt (Sigma, St. Louis, MO), Texas Red-avidin, (6) Lacey, E.; Grant, W. N. Anal. Biochem. 1987, 163, 151-8. (7) Forster, A. C.; McInnes, J. L.; Skingle, D. C.; Symons, R. H. Nucleic Acids Res. 1985, 13, 745-61. (8) Pritchard, D. J.; Morgan, H.; Cooper, J. M. Angew. Chem., Int. Ed. Engl. 1995, 34, 91-3. (9) Hengsakul, M.; Cass, A. E. G. Bioconjugate Chem. 1996, 7, 249-54. (10) Ilcisin, K. J.; Fedosejeves, R. Appl. Opt. 1987, 26, 396-400. (11) Fedosejevs, R.; Brett, M. J. Appl. Opt. 1989, 28, 1877-80.

2620 Analytical Chemistry, Vol. 69, No. 14, July 15, 1997

Figure 1. Experimental arrangement for the generation of a laser diffraction pattern for photopatterning. See text for details.

and alkaline phosphatase-avidin D (Vector Laboratories, Burlington, CA) were used as received. The electrode substrate was a glassy carbon plates, 1 mm thickness (Alfa Aesar). Silver epoxy (APS, Peabody, MA) was used to mount the electrodes to locally constructed aluminum holders (Matt McCormick, UC Riverside). Generation of Interference Pattern. A 10 mW 325 nm HeCd laser (Omnichrome, Chino, CA) was used to photoactivate the photobiotin. Figure 1 shows the experimental arrangement used to produce an interference pattern for photopatterning. The UV (325 nm) light beam was sent through a 50% beam splitter and reflected off two mirrors to make two parallel beams of equal intensity. The parallel beams were sent through a 50 cm focal length lens and the substrate (a photobiotin-coated glassy carbon electrode) was placed near the focal point, where the two laser beams recombine. The merging of two beams generates a firstorder interference pattern with 10 µm spacing when the angle between the two beams is ∼2°, according to the equation, nλ ) 2D sin(θ/2), where D is the spacing between adjacent maxima in the interference pattern, λ is the wavelength of light, θ is the angle between the laser beams, and n is order. The interference pattern spacing can easily be changed by modifying the angle between the beams. Exposure of the photobiotin to 325 nm light (10 mW laser) was done at different time intervals to optimize the contrast of the pattern (times were varied between 20 and 300 s). Preparation of Carbon Surface. Glassy carbon electrodes (GCEs) were polished using 0.3 µm and 0.05 µm alumina powder in deionized water and Metacloth on a home-built polishing apparatus. The carbon surfaces were cleaned thoroughly with deionized water to remove any residual polish after each polishing and then were air-dried. Photobiotin was deposited onto the electrode surface by applying 10 µL of photobiotin (1 mg/mL in DI water) on the polished GCE and the DI water was allowed to evaporate for 2-3 h. The GCE was placed in the diffraction apparatus (Figure 1), and the interference pattern was applied to the photobiotin-covered carbon surface for 20-300 s. A schematic diagram of the derivatization of the carbon surface is shown in Figure 2. This chemistry is identical to that described by Hengsakul and Cass.9 Afterward, the electrodes were washed with phosphate buffer to remove any unbound photobiotin. Spatially Localized Immobilization of Texas Red-Avidin (TR-Avidin). Photobiotin-coated, patterned electrodes prepared in this manner were treated with a 1% solution of Tween-20, a detergent used to diminish nonspecific adsorption. After 1 h, the electrode was placed in a TR-avidin solution (0.25 mg/mL

Figure 2. Attachment of enzymes through light-directed derivatization using photobiotin. Upper scheme: photobiotin deposited onto a glassy carbon surface is irradiated with a diffraction patterned using UV laser light. Photobiotin attaches to the surface in the pattern illuminated by the diffraction pattern of the laser. Unexposed photobiotin is unreacted and can be washed off the surface. Lower scheme: subsequent reaction of the biotinylated electrode with avidin and biotinylated enzyme. Enzyme attaches only where biotin was patterned onto the surface.

phosphate buffer) for 1 h, rinsed in DI and imaged for TR-avidin fluorescence. Spatially Localized Immobilization of Alkaline Phosphatase (APase; EC 3.1.3.1). Alternatively, the patterned electrode was incubated for 1 h in 1% Tween-20 and rinsed, and avidin-labeled APase solution (0.2 units/ml) was placed on the electrode and incubated for 1 h. After rinsing with phosphate buffer, a microscopic image of the APase-avidin-coated electrode was obtained. Next, a few drops of a solution of Vector Red was placed on the APase immobilized carbon surface. Vector Red is a substrate for APase which forms a fluorescent product upon hydrolysis and precipitates at the site of formation. The Vector Red solution was made according to the supplier’s protocol in 100 mM Tris-HCl, pH 8.2 buffer. After 30 min, the electrode surface was washed thoroughly and imaged for Vector Red fluorescence with the same filter set as that of Texas Red (em ) 595 nm and ex ) 615 nm). Optical and Fluorescence Microscopy. The surface of a TRA-modified glassy carbon electrode was positioned face-up toward either an Epiplan Neofluar 20X (air lens) or a 40× (water immersion) objective. Electrode surfaces were imaged with an epifluorescence microscope (Zeiss Axioskop, Thornwood, NY) equipped with a 100 W Hg arc lamp for epi-illumination and a 50

W halogen lamp for transmitted illumination.12 All images were collected in a darkened room with a cooled Thompson 7895B CCD (Class 2, 512 × 512 Metachrome II UV coated chip, MPP mode, Photometrics Ltd., Tucson, AZ) that was operated at -45 °C. Images were collected through a Photometrics NU-200 controller (16 bit, 40 khz A/D, Macintosh IIci configuration) and saved on a Macintosh IIci. Subsequent data processing was performed with IP-LAB 2.1.1c (Signal Analytics, Vienna, VA) and/or Spyglass Transform (Spyglass Software, Champaign, IL) imaging software. Microscopic surface features were brought into focus with light from the Hg arc lamp after it passed through two neutral-density filters (100-fold attenuation) with a 1 ms charge-coupled device (CCD) acquisition time and a camera gain setting of 1. Fluorescent images of modified carbon surfaces were obtained by passing the light from the Hg arc lamp through an excitation filter specific for the Texas red absorption band (595 nm) and collecting all fluorescence at wavelengths greater than 615 nm with a 1 s. camera collection time. All fluorescence images were acquired in the central zone of the CCD. Atomic Force Microscopy. A laser interference pattern was used to immobilize APase onto the carbon surface as described above. The product of APase hydrolysis of Vector Red was allowed to accumulate for 30 min; the surface was scanned with contact-mode AFM performed on a Burleigh Personal AFM, ARIS 3500. Silicon tips (pyramidal shape), also supplied from Burleigh (Burleigh Ins. Co.; Fishers, NY), were used as the cantilever. The reference force of the cantilever, all gains, and filters were optimized prior to each scan. Postprocessing of the images was done using True Image SPM software (Burleigh Instrument Co.; Fishers, NY). The carbon surface was derivatized as previously described with biotin, avidin, alkaline phosphatase, and then Vector Red stain. The modified carbon surface was imaged after each step: photopatterning with photobiotin (i.e., the initial step in the procedure), immobilization of avidin, biotinylated enzyme and after deposition of the insoluble product of hydrolysis of Vector Red. All AFM images were obtained with a 100 nm pyramidal-shaped silicon tip. RESULTS AND DISCUSSION Previously, we observed that derivatization of a carbon electrode can seriously degrade the electron transfer characteristics of the electrode surface.12 The objective of this work is to generate a protocol to allow spatial segregation of sites used for enzyme immobilization from sites of electron transfer on the same electrode surface. The distance between electron transfer sites and enzyme-loaded domains must be kept to a minimum (e.g., less than 5 µm) to maintain the fast response time and high sensitivity required for rapid measurements (i.e., neurotransmitter dynamics). This can be accomplished through the use of photolithographic techniques that activate surface-bound photolabile reagents resulting in the spatially directed immobilization of biopolymers. One system that has been used efficiently for this purpose uses photobiotin, a nitro(aryl) azide derivative of biotin. The reactivity of photobiotin is well documented, and optimal conditions for its use have been described.13 Briefly, a solution of photobiotin is completely dried on an organic surface, and the chemistry proceeds upon illumination with light with a (12) Hopper, P.; Kuhr, W. G. Anal. Chem. 1994, 66, 1996-2004. (13) Elsner, H. I.; Mouritsen, S. Bioconjugate Chem. 1994, 5, 463-7.

Analytical Chemistry, Vol. 69, No. 14, July 15, 1997

2621

wavelength of approximately 320-370 nm. Absorption of light at this wavelength causes photolysis of the nitroaryl moiety to generate a nitrene, which inserts readily into any target molecule (i.e., a glassy carbon surface).14 Photobiotin has been used to spatially direct the immobilization of avidin and biotinylated enzymes on nitrocellulose and polystyrene surfaces.9 This was accomplished simply by exposing the photobiotin-coated substrate to intense light through a mask consisting of a 50-mesh copper grid. This protocol required that the substrate be exposed to light for up to 30 min.9 Additionally, the use of a small grid as the mask has its own inherent problems. First, they are very difficult to work with because of their small size and fragile because of their fine grid size. The use of grids also requires a tight seal between the grid and substrate.15 This is needed because if the seal is not light tight, light will permeate the seal and portions on the substrate will get exposed even though the mask is over them. Pritchard et al. also generated micrometer-scale patterns on gold and silica substrates using photobiotin and masking technology.8 They functionalized gold and silica surfaces with amines and then coupled avidin to the surface-immobilized amine with 1-ethyl-3[3-(dimethylamino)propyl]carbodiimide (EDC). These surfaces were then incubated with photobiotin. After photobiotin bound to the avidin, the substrate was incubated with rabbit IgG or rat IgG. A mask with 1.5 µm divisions was then placed over the substrate, and the photobiotin was exposed to intense visible light (9 mW/cm2 of light greater than 300 nm), at which time the photobiotin binds to the antibody molecules. Goat anti-rabbit IgG labeled with tetramethylrhodamine isothiocyanate (TRITC) and/ or rabbit anti-rat labeled with IgG fluorescein isothiocyanate (FITC) were then added on the substrate, generating a fluorescent pattern showing the locations of each individual antibody.8 Morgan et al. demonstrated that the same techniques using a 3 µm mark-space mask on a silica surface could generate a pattern of antibodies that could be visualized with AFM. They showed that the spacing of the immobilized antibodies observed by AFM and fluorescence microscopy were very similar. In this work, we have immobilized photobiotin onto an electrode surface with the objective to provide regions of enzymemodified surfaces directly adjacent to unmodified areas (which ultimately will have facile electron transfer kinetics). An array of micrometer-sized wide lines can be formed by using the interference pattern produced when two parallel coherent light sources are combined at a surface at an angle. Experimentally, this is very similar to the procedures used to create finely spaced optical gratings.10,11 Basically, the laser beam is passed through a 50% beam splitter, and each beam is directed to recombine at the sample surface at approximately the same angle of incidence (Figure 1). The placement of the beams is fine-tuned to allow complete overlap of the mode structure of the laser spot. Higher order interference patterns are minimized through the use of highquality optics (1/10 wave surface flatness). This ensures that the variation between intensity maxima and minima in the first order will be several orders of magnitude larger than those formed with second and higher orders.10,11 This produces a well-defined pattern of lines across the electrode surface, where the spacing between points of positive interference (D) can be approximated (14) Staros, J. Trends Biol. Sci. 1980, 320-2. (15) Thompson, L.; Bowden, M. Introduction to Microlithography, 2nd ed.; American Chemical Society: Washington, DC, 1994.

2622

Analytical Chemistry, Vol. 69, No. 14, July 15, 1997

by the equation nλ ) 2D sin(θ/2), where λ is the wavelength, θ is the angle of incidence, and n is order. For 325 nm light, at an angle of incidence, θ ) 45°, the minimum practical spacing is 0.46 µm between maxima. This type of spacing should be more than ideal for our situation, since any electroactive products generated at the immobilized protein site must diffuse to an adjacent, underivatized site for detection. The diffusion time for the redox mediator to travel between the enzyme and the electron transfer site (based on the Einstein diffusion equation) is on the order of a few microseconds. With a HeCd laser at 325 nm, an angle of ∼2° is needed to get 10 µm spacing for the first-order interference pattern (Figure 3). This approaches the upper limit of size for our hardware because generating an incidence angle less than 2° is very difficult. On the other hand, it is relatively easy to increase the angle between the beams and decrease the spacing of the interference pattern. We chose 10 µm spacing (leading to 5 µm features) for this work such that the substrate pattern could be easily characterized using optical microscopy, which is easily capable of characterizing micrometer-sized structures.12 Epifluorescence microscopy allows the excitation radiation to use the same optics as the collection system; in addition, this radiation is concentrated onto a welldefined region of the sample within the microscope’s field of view. Very high sensitivity was obtained through the use of a cooled CCD camera to image the fluorescence from the electrode surface. An average pixel size of 20 µm, when combined with a magnification of 40 in the microscope objective, generated an imaging resolution of 0.5 µm/pixel, near the diffraction limit of light-based systems. In these experiments, a thin layer of photobiotin was deposited onto the surface of glassy carbon electrodes through solvent evaporation. The biotin-coated GCE was exposed to the diffraction pattern formed with 325 nm light (∼1 W cm-2) for various times (30-200 s) to induce photolysis of the nitroaryl moiety to generate a nitrene, which inserts readily into the glassy carbon lattice.14 An exposure time of 60 s was found to reproducibly form a linear array of biotin on the GCE with a spacing of ∼10 µm. A very regular pattern of lines, closely resembling the intensity distribution expected for the laser diffraction pattern, could be visualized by optical microscopy as shown in Figure 3 (upper left). The distance between adjacent lines of bound photobiotin was examined by plotting the average intensity as a function of lateral distance (Figure 4, upper left). The average width of each photopatterned line (taken at half-maximum intensity) was 4.95 ( 0.19 µm (n ) 5). This corresponds to a standard deviation less than (0.5 pixel. TR-avidin was allowed to bind to the immobilized biotin after photopatterning. Since avidin has a high affinity for biotin (1015), they form a very strong complex. Texas Red, a fluorescent dye, was used to tag the avidin molecules so they could be imaged with a fluorescence microscope. The microscopic spatial distribution of the bound TR-avidin is shown in Figure 3 (bottom left). Virtually the same linear array was observed, but each individual line was not continuously labeled by TR-avidin. This is probably due to the heterogeneity of the glassy carbon surface,16 where the density of surface oxides may influence the availability of biotin for binding. Previously, we observed similar heterogeneity when carbon fiber surfaces were derivatized using biotin/avidin chem(16) McCreery, R. In Electroanalytical Chemistry; Bard, A., Ed.; Marcel Decker: New York, 1991; Vol. 17, pp 221-374.

Figure 3. Glassy carbon electrodes photolithographically derivatized with photobiotin using a laser diffraction pattern. All images are presented as raw data (the only data processing performed was a tilt correction). Each data point represents the CCD output and includes a system offset of 1100. Bar is 10 µm in all images. Upper left: white light reflected image of patterned photobiotin on a carbon substrate. Photobiotin was irradiated for 60 s with 325 nm light (∼1 W cm-2) in a diffraction pattern produced with an angle between the beams, θ, of 1.7°. The image was integrated for 0.001 s on the CCD camera. Lower left: fluorescence image of TR-avidin attached to a biotin-patterned carbon surface. Photobiotin was irradiated for 90 s as described above and then TR-avidin was bound to the surface-immobilized biotin. The fluorescence of TR-avidin was integrated for 1 s using appropriate excitation and emission filters (see text for details). Upper right: white light reflected image of the attachment of the avidin conjugate of alkaline phosphatase onto a biotin-patterned carbon surface. Photobiotin was irradiated for 60 s as described above and then APase-avidin was bound to the surface-immobilized biotin. The image was integrated for 0.001 s. Lower right: fluorescence image of the product of Vector Red hydrolysis by immobilized alkaline phosphatase patterned on a carbon surface. Photobiotin was irradiated for 60 s as described above, then APase-avidin was bound to the surface-immobilized biotin. A solution of Vector Red was applied to the derivatized surface for 30 min and the insoluble, fluorescent product was allowed to precipitate onto the surface. The fluorescence image was integrated for 1 s using appropriate excitation and emission filters (see text for details).

istry.12,17 In this case, the chemical architecture of the surface may play an even greater role in determining the binding properties of biotin, since only a very short tether (14 carbon units) was employed. This implies that avidin must closely approach the electrode surface for binding to occur. The TR-avidin line spacing was examined by plotting fluorescence intensity as a function of lateral distance (Figure 4, lower left). The average

width of each photopatterned line(taken at half-maximum intensity) was 5.05 ( 0.1 µm (n ) 5). This is not statistically different from that observed for photobiotin only, indicating that TR-avidin is binding specifically to the biotin photolithographically immobilized to the electrode surface. This also implies that there is no degradation of resolution in the formation of these nanostructures induced by biotin/avidin chemistry.

(17) Pantano, P.; Kuhr, W. G. Anal. Chem. 1993, 65, 2452-8.

Analytical Chemistry, Vol. 69, No. 14, July 15, 1997

2623

Figure 4. Lateral profiles of the diffraction pattern-induced derivatization of glassy carbon surfaces with photobiotin. The intensities of reflected light (upper figures) and fluorescence (lower figures) were plotted as a function of lateral distance across the electrode surface. The fluorescence intensities were background-corrected. All other conditions are identical to those presented in Figure 3. Upper left: white light reflected image of patterned photobiotin on a carbon substrate. Lower left: fluorescence image of TR-avidin attached to a biotin-patterned carbon surface. Upper right: white light reflected image of the attachment of the avidin conjugate of alkaline phosphatase onto a biotin-patterned carbon surface. Lower right: fluorescence image of the product of Vector Red hydrolysis by immobilized alkaline phosphatase patterned on a carbon surface.

To demonstrate the feasibility of functional enzyme attachment to a glassy carbon surface, biotin was photolithographically derivatized to the carbon as described above. An alkaline phosphatase conjugate of avidin (APase-avidin) was attached to the patterned biotin and then the electrodes were washed and the enzyme-derivatized carbon surface was imaged with light microscopy. As shown in Figure 3 (upper right), a pattern similar to that of biotin-only is observed. Additionally, the APase-avidin line spacing was examined by plotting intensity as a function of lateral distance (Figure 4, upper right). The average width of each photopatterned line (taken at half-maximum intensity) was 5.0 ( 0.16 µm (n ) 5), indicating that the enzyme-avidin complex is binding specifically to the biotin photolithographically immobilized to the electrode surface. The viability of the APase attached to the surface in this manner was demonstrated by imaging the distribution of the product of the alkaline hydrolysis of Vector Red (Figure 3, lower right). An insoluble, fluorescent product is formed when Vector Red is hydrolyzed by the enzyme. The distribution of the precipitated product shows the activity of the patterned enzyme because the fluorescent product is formed only where active enzyme is present. As shown in Figure 3 (lower right), the spatial distribution of the product of APase hydrolysis of Vector Red is similar to that of biotin-only (Figure 3, upper right) and to that of the bound TR-avidin (Figure 3, lower left). Additionally, the product distribution was examined by plotting fluorescence intensity as a function of lateral distance (Figure 4, lower right) and the average width of each photopatterned line (taken at halfmaximum intensity) was 5.15 ( 0.25 µm (n ) 5). This is not 2624

Analytical Chemistry, Vol. 69, No. 14, July 15, 1997

statistically different from that observed for photobiotin only, indicating that the spatial distribution of the activity of the enzyme-avidin complex is virtually identical to that of the enzyme/avidin complex (as well as the biotinylated surface). The glassy carbon electrode surfaces were also examined with AFM after every step in the patterning procedure. No discernible pattern was observed after deposition of photobiotin, illumination of the surface with the UV interference pattern, attachment of TRavidin or even after attachment of the avidin conjugate of APase (data not shown). This indicates that only a very thin layer of photobiotin was initially deposited onto the glassy carbon surface and that illumination with UV light did not significantly alter the physical morphology of the electrode surface. A distinguishable spatial pattern of the immobilized APase was finally observed after extensive deposition of the Vector Red product (Figure 5). The AFM image of this surface shows considerable heterogeneity of the enzyme complex within each line (similar to that seen in the fluorescence images of TR-avidin and the Vector Red product, Figure 3). Since the photopatterned, biotinylated surface appears to be quite homogeneous within a given line (Figure 3, top), it would seem that avidin binding (as well as enzyme activity) is influenced by other factors, i.e., the local chemical architecture of the glassy carbon surface. We have observed similar heterogeneity in the distribution of avidin bound to biotin across a carbon, fiber surface.12 Since direct electrochemical oxidation of (18) Ratcliff, B. B.; Klancke, J. W.; Koppang, M. D.; Engstrom, R. C. Anal. Chem. 1996, 68, 2010-4. (19) Pillai, V. N. R. Synthesis 1980, 1980, 1-26. (20) Jacobs, J. W.; Fodor, S. P. A. Trends Biotechnol. 1994, 12, 19-26.

Figure 5. Atomic force microscope image of the product of Vector Red hydrolysis by the avidin conjugate of alkaline phosphatase immobilized to a glassy carbon surface derivatized with photobiotin via a diffraction pattern. Experimental details are identical to Figure 3. The AFM image is ∼45 µm × ∼58 µm. The higher resolution in this image clearly shows the heterogeneity of the patterned lines that could not be distinguished with optical microscopy. The reference force voltage was 5.0 V on a 100 nm pyramidal silicon probe tip. The image was processed with plane removal, line removal, and a median filter using True Image SPM software supplied by Burleigh Instrument Co.

a carbon surface is possible on a micrometer scale,18 it may ultimately be possible to control the avidin binding reaction by manipulating the chemical architecture of the surface within each line. It should also be possible to use this type of maskless photopatterning with a variety of photodeprotection chemistries. Spatial control of the derivatization of solid substrates has been achieved through the use of photolabile protecting groups, where such groups are removed after application of an appropriate wavelength of light to allow activation of a specific chemical synthetic step.2 Photolabile protecting groups, which are good chromophores that are very sensitive to light, must be relatively stable to the other chemical reagents that they will be exposed to in the ground state as part of the synthetic procedure.19 The selective removal of these photolabile groups has been directed through conventional masking technology to obtain structures with dimensions of 100 µm on a side.2,20 Similar chemistries may

be used with interference pattern-generated lithographic patterns to fabricate arrays of a wide variety of proteins, where the dimensions of the protein microstructures can be created on a submicrometer scale.

ACKNOWLEDGMENT This work was supported by the National Institutes of Health (Grant GM44112-01A1).

Received for review February 24, 1997. Accepted April 28, 1997.X AC9702094 X

Abstract published in Advance ACS Abstracts, May 15, 1997.

Analytical Chemistry, Vol. 69, No. 14, July 15, 1997

2625