Anal. Chem. 2000, 72, 5618-5624
High-Density Fiber-Optic DNA Random Microsphere Array Jane A. Ferguson, Frank J. Steemers, and David R. Walt*
The Max Tishler Laboratory for Organic Chemistry, Department of Chemistry, Tufts University, Medford, Massachusetts 02155
A high-density fiber-optic DNA microarray sensor was developed to monitor multiple DNA sequences in parallel. Microarrays were prepared by randomly distributing DNA probe-functionalized 3.1-µm-diameter microspheres in an array of wells etched in a 500-µm-diameter optical imaging fiber. Registration of the microspheres was performed using an optical encoding scheme and a custom-built imaging system. Hybridization was visualized using fluorescent-labeled DNA targets with a detection limit of 10 fM. Hybridization times of seconds are required for nanomolar target concentrations, and analysis is performed in minutes. Miniaturized high-density arrays of oligonucleotide probes are proving to be powerful tools for advanced diagnostic and genetic analysis applications. Recent reviews illustrate the rapid proliferation of DNA biochips and describe the impact that the technology poses for the future.1-3 Large-scale DNA microarrays provide significant advantages relative to conventional electrophoretic techniques4 for discovering disease-associated mutations and sequences,5,6 in monitoring gene expression,7,8 and in screening for targets known to play a role in diseases.9 The advent of DNA biochips has also brought one of the goals of the Human Genome Project closer by increasing the speed and efficiency of producing genetic maps.10,11 Despite these successes, DNA biochip fabrication methods remain a challenge for creating cost-effective devices. Present fabrication methods for preparing biochip arrays include techniques such as light-directed combinatorial chemical synthesis6,12-15 (1) Lockhart, D. J.; Winzeler, E. A. Nature 2000, 405, 827-836. (2) Brown, P. O.; Botstein, D. Nat. Genet. 1999, 21 (1 Suppl), 33-37. (3) Walt, D. R. Science 2000, 287, 451-452,. (4) Baba, Y. J. Chromatogr., B: Biomed. Appl. 1996, 687, 271-302. (5) Shuber, A. P.; Michalowsky, L. A.; Nass, G. S.; Skoletsky, J.; Hire, L. M.; Kotsopoulos, S. K.; et al. Hum. Mol. Genet. 1997, 6, 337-347. (6) Hacia, J. G.; Brody, L. C.; Chee, M. S.; Fodor, S. P. A.; Collins, F. S. Nat. Genet. 1996, 14, 441-446. (7) deSaizieu, A.; Certa, U.; Warrington, J.; Gray, C.; Keck, W.; Mous, J. Nat. Biotechol. 1998, 16, 45-48. (8) DeRisi, J.; Penland, L.; Brown, P. O.; Bittner, M. L.; Meltzer, P. S.; Ray, M.; et al. Nat. Genet. 1996, 14, 457-460. (9) Needels, M. C.; Jones, D. G.; Tate, E. H.; Heinkel, G. L.; Kochersperger, L. M.; Dower, W. J.; et al. Proc. Natl. Acad. Sci. U.S.A. 1993, 90, 1070010704. (10) Lander, E. S. Science 1996, 274, 536-538. (11) Wang D. R.; Fan, J. B.; Siao, C. J.; Berno, A.; Young, P.; Sapolsky, R.; et al. Science 1998, 280, 1077-1082. (12) Wallraff, G.; Labadie, J.; Brock, P.; DiPietro, R.; Nguyen, T.; Huynh, T.; et al. ChemTech 1997, 27, 22-32.
5618 Analytical Chemistry, Vol. 72, No. 22, November 15, 2000
and high-speed robotic printing.8,16,17 While these biochips have generated useful new genetic information and diagnostic capabilities, they suffer from both large feature and overall substrate sizes and require a new fabrication protocol to change any of the probe sequences in the array. An alternative to biochips is an approach based on DNA probes bound to microspheres. The process of functionalizing microspheres with DNA probes is well established.9,18-23 Different probefunctionalized microspheres can be analyzed simultaneously by labeling each microsphere type and mixing them. Identification of reactive microspheres in solution requires encoding each microsphere via chemical,9 spectrometric,18,21 electronic, or physical means.24 Once identified, the microspheres are sorted and typically analyzed using flow cytometry18 or confocal microscopy.21 Although the production of such microspheres is less costly and elaborate than biochip array production, the analysis of microspheres in solution is more complex. In this paper, we describe DNA array sensors that combine the ease of array production with reduced substrate size by coupling microspheres to high-density optical fiber arrays. The optical fiber substrate permits simultaneous and repetitive monitoring of the entire microsphere array. Single-stranded oligonucleotide probes are first immobilized on different optically encoded microspheres. The pure encoded, oligonucleotide functionalized microsphere populations are then mixed and randomly distributed and fixed in micrometer-sized wells on the fiber-optic substrate. Each microsphere is then decoded using image-processing techniques to allow positional registration of the entire array. The (13) Cronin, M. T.; Fucini, R. V.; Kim, S. M.; Masino, R. S.; Wespi, R. M.; et al. Hum. Mutat. 1996, 7, 244-255. (14) Chee, M.; Yang, R.; Hubbell, E.; Berno, A.; Huang, X. C.; Stern, D. Science 1996, 274, 610-614. (15) McGall, G.; Labadie, J.; Brock, P.; Wallraff, G.; Nguyen, T.; Hinsberg, W. Proc. Natl. Acad. Sci. U.S.A. 1996, 93, 13555-12560. (16) Schena, M.; Shalon, D.; Heller, R.; Chai, A.; Brown, P. O.; Davis, R. W. Proc. Natl. Acad. Sci. U.S.A. 1996, 93, 10614-10619. (17) Shalon, D.; Smith S. J.; Brown, P. O. Genome Res. 1996, 6, 639-645. (18) Fulton, R. J.; McDada, R. L.; Smith, P. L.; Kienker, L. J.; Kettman, J. R. Clin. Chem. 1997, 43, 1749-1756. (19) Hakala, H.; Lonneberg, H. Bioconjugate Chem. 1997, 8, 232-237. (20) Hakala, H.; Heinonen, P.; Iitia, A.; Lonnberg, H. Bioconjugate Chem. 1997, 8, 378-384. (21) Egner, B. J.; Rana, S.; Smith, H.; Bouloc, N.; Frey, J.; Brocklesby, W. S.; et al. Chem. Commun. 1997, 8, 735-736. (22) Van Ness, J.; Kalbfleisch, S.; Petrie, C. R.; Reed, M. W.; Tabone, J. C.; Vermeulen, N. M. J. Nucleic Acids Res. 1991, 19, 3345-3350. (23) Storhoff, J. J.; Elghanian, R.; Mucic, R. C.; Mirkin, C. A.; Lestinger, R. L. J. Am. Chem. Soc. 1998, 120, 1959-1964. (24) Czarnik, A. W. Curr. Opin. Chem. Biol. 1997, 1, 60-66. 10.1021/ac0008284 CCC: $19.00
© 2000 American Chemical Society Published on Web 10/14/2000
resulting fiber-optic array contains high-density, randomly distributed micrometer-sized features. This system combines the simplicity of microsphere sensor production with the convenience of biochip array analysis techniques and the small size of an optical fiber substrate. In this paper, we describe the preparation and performance characteristics of an array containing 25 oligonucleotide probes attached to encoded microspheres. EXPERIMENTAL PROTOCOL Internal Encoding. Aliquots (200 µL) of stock (1 mL of stock beads contains 5.8 × 109 beads in 0.01% merthiolate in water) 3.1µm-diameter amine-modified poly(methylstyrene)divinylbenzene microspheres (Bangs Laboratories, Inc. Carmel, IN) were filtered and washed with dry tetrahydrofuran (THF) and then placed in a microcentrifuge tube. A 200-µL aliquot of europium(III) thenoyltrifluoroacetonate‚3H2O dye [Eu-dye (Acros)] in THF was added to the beads. Eu-dye concentrations of 0, 0.001, 0.01, 0.025, 0.05, 0.1, 0.5, and 1 M were used. The microsphere/dye suspension was shaken (VWR Vortex Genie II) for 2 h. The suspensions were filtered separately (Millipore type HVLP) and washed thoroughly with MeOH. The beads were stored in 0.01% Tween (essential for preparation and storage to prevent the beads from clumping together) in ultrapure water until use. External Encoding. Ten microliters of stock beads was rinsed25 with BT buffer (0.1 M boric acid, 0.1 M NaOH, 0.13 M NaHCO3, 0.01% Tween, pH 9). The beads were suspended in 100 µL of BT buffer, and then 5 µL of dye solution [Cy5 monofunctional dye (Amersham) or TAMRA (Molecular Probes)] in DMF was added. A Cy5 stock solution was prepared by adding 25 µL of DMF to a commercially available vial containing lyophilized Cy5. Dilutions of 0, 0.01, 0.05, 0.1, and 0.3 of the Cy5 stock and TAMRA concentrations of 0, 0.1, 0.4, and 3 mM were used. The beads were shaken for 2 h and then rinsed three times with BT buffer and three times with phosphate buffer containing Tween (0.02 M phosphate buffer, pH 7.4, 0.01% Tween 20). Cyanuric Chloride Activation. DNA probes were synthesized with a 5′-amino-C6 modifier (Glen Research) in the Tufts Physiology Department using an ABI synthesizer. 5′-Amino-terminal oligonucleotide probe (20 nmol) was dissolved in 180 µL of 0.1 M sodium borate buffer (SBB, pH 8.3). Oligonucleotide activation was initiated by adding 40 nmol of cyanuric chloride in 40 µL of acetonitrile. After 1 h, unreacted cyanuric chloride was removed by three cycles of centrifugal ultrafiltration (Microcon 3, Amicon) and recovered in 200 µL of 0.1 M SBB. DNA Functionalization. Five microliters of stock beads was rinsed with 0.02 M phosphate buffer (pH 7). A 150-µL sample of 5% glutaraldehyde26 in phosphate buffer was added to the beads. The beads were shaken for 1 h and then rinsed three times with phosphate buffer. A 150-µL sample of 5% polyethyleneimine (PEI) was then added to the beads.22 The beads were shaken for 1 h and then rinsed three times with phosphate buffer and three times with 0.1 M SBB. A 100-µL sample of 150 µM cyanuric chlorideactivated oligonucleotide probe22 in SBB buffer was added to the beads, and the resultant mixture was shaken overnight. The probe solution was removed25 and saved for reuse. The beads were then (25) All rinsing procedures entailed placing the centrifuge tube containing the beads and solution into a microcentrifuge (IEC Micromax RF) at 8000 rpm for 3 min. Liquid over the beads was removed using a pipet. (26) Walt, D. R.; Agayn, V. I. Trends Anal. Chem. 1994, 13, 425-430.
rinsed three times with SBB buffer. The remaining amine groups were capped with succinic anhydride to prevent nonspecific binding.22 A 100-µL sample of 0.1 M succinic anhydride in 90% DMSO, 10% SBB was added to the beads. The beads were shaken for 1 h and then rinsed three times with SBB buffer and three times with TE buffer (10 mM Tris-HCL, pH 8.3, 1 mM EDTA, 0.1 M NaCl, 0.1% SDS). Microwell Formation. Imaging fiber bundles (500-µm diameter) containing 6 × 103 individual fibers were chemically etched according to a previously detailed procedure.27 Briefly, the distal tip of a polished fiber is dipped in a buffered acid solution. The depth of the etched wells is controlled by the time exposed to the acid. After the desired time, the fiber tip is dipped in water and thoroughly rinsed. Array Formation. A 5-µL sample of probe-functionalized beads was stored in 40 µL of TE buffer. After selecting the desired probefunctionalized microspheres, 1 µL of each bead suspension was placed in a microcentrifuge tube and vortexed. A 0.05-µL aliquot of this mixture was placed onto the distal face of the imaging fiber containing the microwells. After solvent evaporation (∼3 min), the distal tip of the fiber was wiped with an antistatic swab to remove excess beads. When a new sensor is desired, sonicating the fiber tip in water for 3 min will remove the beads and regenerate the substrate (Sonogen Automatic Cleaner). Analysis Setup and Protocol. The imaging system, described previously, consists of a light source, an inverted microscope, and a modified Olympus epifluorescence microscope/charge-coupled device camera (Photometrics PXL).28 A fiber chuck held the imaging fiber in a fixed position while electronically controlled filter wheels switched between the analytical wavelength and the encoding wavelengths, enabling complete analysis and identification of the microspheres within minutes. Excitation light was sent into the proximal tip of the imaging fiber, and emission from the fluorescing molecules was captured and directed onto the CCD camera detector.28 Fluorescence measurements were acquired and analyzed using commercially available IPLab software (Scanalytics, Inc.). The fiber was not removed from the imaging system during testing, rinsing, or regeneration steps. The proximal tip of the fiber was secured in the fiber chuck of the imaging system and all solutions were brought to the fiber’s distal tip, which housed the microbead sensors. Images acquired immediately prior to each test while the fiber tip was in buffer were subtracted from the response images. Background signals from empty wells were then subtracted from signals generated during each test. Multiplex Analysis. Images were acquired for 1 and 0.5 s at wavelengths specific to each encoding dye. A 365-nm excitation filter and a 600-nm long-pass emission filter were used for the Eu-dye. A 620-nm excitation filter and a 670-nm emission filter were used for the Cy5 dye. A 530-nm excitation filter and a 580nm emission filter were used for TAMRA. The images acquired at the three wavelength pairs were used to positionally register each microsphere sensor. Once registration was complete, sample analysis was performed. All targets were labeled with fluorecein. A 495-nm excitation filter and a 530-nm emission filter were used during (27) Pantano, P.; Walt, D. R. Chem. Mater. 1996, 8, 2832-2835. (28) Bronk, K. S.; Michael, K. L.; Pantano, P.; Walt D. R. Anal. Chem. 1995, 67, 2750-2757.
Analytical Chemistry, Vol. 72, No. 22, November 15, 2000
5619
target analysis. The fiber’s distal tip was placed in a target solution for 5 min, then rinsed with TE buffer, and fluorescence images were acquired for 5 s while the fiber was in buffer. Overlay segments were drawn to select the beads bearing a hybridization signal using IPLab software. These overlay segments were copied and pasted onto each of the encoding images, and the selected beads’ identities were determined. The sensor was regenerated as described above by dipping the fiber tip in 90% formamide in TE then rinsing with TE buffer. This procedure was repeated for each of the target solutions. Hybridization in Real Time. A background fluorescence image was acquired at wavelengths specific to fluorescein (excitation 490 nm, emission 530 nm) with the fiber’s distal tip in buffer. The fiber’s distal tip was then placed in 4 µL of fluorescein-labeled target solution, and one image was acquired every minute for 6 min. Subsequently, the fiber was dipped in 90% formamide in TE buffer at room temperature to regenerate the sensor and a background image was taken with the fiber in buffer. The fiber was again placed in the target solution where images were acquired for another 6-min interval. Reproducibility Study. The fiber’s distal tip was placed in 4 µL of labeled-target solution for 5 min and rinsed with TE buffer, and a fluorescence image was acquired for 5 s with the fiber tip in buffer. The fiber tip was then dipped into 90% formamide in TE (room temperature) then rinsed with TE buffer to remove any hybridized target and regenerate the sensor. This procedure was repeated 100 times using the IL2 target and 5 times (intermittently during the IL2 tests) using the IL6 target. Kinetic Study. The fiber tip was placed in the target solution for a given time and then rinsed with TE, and a fluorescence image was acquired with the sensor in buffer. After data acquisition, the fiber was placed back in the target solution for a given time, rinsed, and analyzed in buffer. The sensor was monitored at elapsed times of 10, 20, and 30 s and 1, 2, 3, 4, 5, and 10 min. After a plateau was reached, the sensor was regenerated by dipping in a 90% formamide solution in TE (room temperature) and the test was repeated using a different concentration of target solution. Microsphere Sensitivity. The fiber’s distal tip was placed in 4 µL of target solution until the hybridization signal-to-noise ratio was 3. The signal was monitored after rinsing the fiber tip with TE buffer and acquiring a fluorescence image for 5 s while the fiber tip was in buffer. For the hour-long assays, a 0.6-mL centrifuge tube was filled and capped. A hole was drilled in the cap to enable the fiber tip to be placed in the target solution while preventing evaporation. Hybridization at Elevated Temperature. An array was created using a 450-mm-long fiber. The proximal end of the fiber was connected to the imaging system, and the distal end was held in a vertical micropositioner. The buffer and target temperatures were controlled by a water bath. Testing was performed as described above. RESULTS Microsphere Encoding. Before attaching oligonucleotides to the microspheres, a family of dye-encoded microspheres was created. Fluorescent dyes were used to encode the microspheres (Figure 1a). Europium(III) thenoyltrifluoroacetonate‚3H2O (λex/ λem ) 365/615) (Eu-dye), Cy5 (λex/λem ) 620/700), and 5-(and6)-carboxytetramethylrhodamine, succinimidyl ester (λex/λem ) 5620 Analytical Chemistry, Vol. 72, No. 22, November 15, 2000
Figure 1. (a) Scheme used to attach DNA probes and encoding dyes to the microsphere. PEI is employed to increase the surface functional groups. (b) Atomic force micrograph (AFM) image of microspheres in wells. (c) Overview of array system. A 0.05-µL drop of water containing a mixture of thousands of beads is placed on the distal tip of an imaging fiber. Micrometer-sized wells etched into the fiber tip serve as host to the beads. Dipping the fiber tip into a labeledtarget solution produces a signal only on those fibers bearing the DNA probe complementary to the target in solution.
535/580) (TAMRA, SE) were chosen for this demonstration. The dyes were incorporated by exploiting the chemical properties of the amino-modified polystyrene microspheres (Figure 1a). The polystyrene microspheres swell in THF, enabling hydrophobic dyes to penetrate the microsphere and become entrapped when the microsphere contracts upon solvent removal.29 Eight distinguishable microsphere families were prepared by entrapping different Eu-dye concentrations inside the microspheres. The absorption and emission spectra of the dyes are not compromised within the microsphere’s environment, and their concentration remains constant over time. In addition to internal entrapment, the microspheres’ amine-modified surface permitted coupling to amine-reactive dyes.30 Different concentrations of Cy5 and TAMRA (29) Bangs, L. B. Uniform Latex Particles; Seragen Diagnostics, Inc.: Indianapolis, IN, 1984.
Table 1. Sequences Used in the Arraya.
a The sequences are listed 5′-3′. Each probe has a 5′-(NH -(CH ) -) functionality for cyanuric chloride activation and attachment to the 2 2 6 microspheres. Each complementary target has a 5′-fluorescein label.
were attached to the surface amine groups of the eight Eu-dyed beads. In this manner, a library of 100 spectroscopically distinguishable microsphere types was prepared using various combinations of the three dyes. Microsphere encoding was carried out prior to oligonucleotide attachment because reaction with the amine-reactive dyes after probe attachment affected the hybridization reaction. On the other hand, the oligonucleotide probes on the surface of the microspheres were not affected by subsequent internal encoding with Eu-dye. DNA Attachment. After the encoded microsphere library was in hand, we functionalized each encoded microsphere with a different single-stranded DNA probe. The probe sequences are shown in Table 1. A protocol used previously to create a single core fiber-optic DNA array was modified to prepare the DNAmicrosphere sensors.31 Cyanuric chloride-activated single-stranded DNA probes22 were covalently bound to encoded amine-functionalized microspheres (Figure 1a). When the cyanuric chlorideactivated probes were attached directly to the amine-modified polystyrene microspheres, detectable fluorescent signals were generated upon hybridization to labeled targets. However, by first modifying the microspheres with PEI22 before DNA functionalization, the signal increased 10-fold because the number of attachment sites available was amplified. Nonspecific binding of (30) Haugland, R. P. Handbook of Fluorescent Probes and Research Chemicals; Molecular Probes, Inc.: Eugene, OR, 1996. (31) Ferguson, J. A.; Boles, T. C.; Adams, C. P.; Walt, D. R. Nat. Biotechnol. 1996, 14, 1681-1684.
DNA targets to the amine-functionalized microsphere surfaces was prevented by capping unreacted amines with succinic anhydride.22 The resulting encoded probe-functionalized microspheres can be stored for months and mixed in any desired combination to create or alter the DNA sensor array. Each milliliter of stock solution contains ∼6 × 109 microspheres, enabling functionalization of billions of beads at once. Even after a 20× dilution, a 1-µL volume of microsphere solution contains enough beads to produce hundreds of different arrays. Microsphere-Based Fiber-Optic Sensors. We previously reported an array consisting of randomly distributed independently addressable micrometer-bead sensors using an imaging optical fiber substrate.32 This system employed imaging fibers consisting of 6000 individually clad fibers that were melted and drawn together to form a coherent, 500-µm-diameter bundle.33 The imaging fiber substrate is a critical component of the array’s facile production and assay. The compositional difference between the core and cladding of each fiber enables the cores to be etched selectively, providing for the simultaneous formation of 6000 3.5µm-diameter wells in the surface of the fiber tip within seconds.27 A droplet of water containing a mixture of the encoded 3.1-µmdiameter DNA-functionalized microsphere sensors is placed onto the fiber bundle tip. Individual beads settle spontaneously into the wells as the water droplet evaporates to produce a randomly (32) Michael, K. L.; Taylor, L. T.; Schultz, S. L.; Walt, D. R. Anal. Chem. 1998, 70, 1242-1248. (33) Pantano, P.; Walt, D. R. Anal. Chem. 1995, 67, 481A-487A.
Analytical Chemistry, Vol. 72, No. 22, November 15, 2000
5621
Figure 2. CCD images of a small section of a fiber-optic array containing 25 different probe-functionalized microspheres. A random selection of 12 images obtained after hybridization to one labeled target. The target used for each experiment is listed to the left of each frame. (The numbers correspond to the number assignment in Table 1.) The beads bearing a signal were identified by their encoding pattern. Between each test, the sensor was regenerated with 90% formamide in TE.
distributed array of thousands of microsphere sensors. Excess microspheres are removed from the fiber tip while surface-binding interactions between the beads and the wells hold each microsphere in place (Figure 1b). The fixed location of each microsphere in the wells of the imaging fiber enables rapid identification of the sensing chemistry attached to each bead by measuring its spectroscopic encoding signature with an optical-imaging system.28 The ease of fabrication and the intrinsically small feature size of individual probe elements are two attractive components of the present array. Controlling Array Formation. One of the primary advantages of this system is the ability to alter the types of microspheres contained in an array. The density of microspheres in solution controls the number of occupied wells. With dilute solutions, empty wells remain after the initial array production. Additional microspheres bearing different probes can be added to the unoccupied sites or to the original microsphere solution at any time to create a more diverse array. If a different selection of beads is desired, sonicating the fiber tip removes the beads from the wells, enabling a new sensor array to be made in the same substrate. Optical Imaging and Analysis System. By coupling the imaging fiber bundle to a detection system fitted with a CCD camera, we can resolve individual fibers in the array and the microsphere residing in the well at each fiber tip, while simultaneously viewing the entire array. Hybridization was visualized 5622
Analytical Chemistry, Vol. 72, No. 22, November 15, 2000
using fluorescent-labeled complementary targets (Figure 1c). When DNA targets hybridize to their complementary probe sequences on the microspheres, a fluorescent signal is generated. The microspheres containing these signals are selected, and the identity of the probe on each bead is determined by its spectroscopic signature. Hybridization Specificity in a Multiplex Assay. To demonstrate this microsphere array system, we selected 13 sequences from disease-related genes (oncogenes and cystic fibrosis) and disease states (lymphocyte and cytokine expression) (Table 1).13,16,19,31,34 To create 25 completely specific probes, the reverse complements of the sequences were synthesized. An array sensor was created with the 25 different probes, each attached to a different encoded microsphere. The encoding signals were used to positionally register the microspheres in the array. After registration, the array was interrogated with each of the 25 target solutions with sensor regeneration between each test (Figure 2). The fiber tip was dipped into a fluorescent-labeled target solution. After a specified time, the fiber tip was removed from the target solution, rinsed, and then placed in buffer. Microspheres bearing a complementary probe display a fluorescent signal due to the hybridized labeled target. Replicates of each bead type located randomly within the array yield redundant information which (34) Healey, B. G.; Matson, R. S.; Walt, D. R. Anal. Biochem. 1997, 251, 270279.
Figure 3. Real-time hybridization. An IFNG probe array was placed in 4 µL of IFNG target solution and fluorescence images were acquired every minute for 6 min. After regenerating the sensor, the experiment was repeated.
contributes to the array’s reliability. Approximately 20% (1295 out of 6000) of the wells were occupied with an average of 50 replicates of each bead type in the array. The resulting data enabled the correct identification of each target solution. We have also demonstrated single base pair mismatch differentiation by conducting the hybridization under conditions of higher stringency such as elevated temperature where only the fully complementary sequences should hybridize. The K-ras wildtype sequence (probe 6) was hybridized to two targets: one target contained the wild-type complement and a second target contained a single base mismatch at position 7. When hybridization was conducted at 53 °C, similar signals were generated from the hybridized complementary fluorescent-labeled target at both room temperature and at elevated temperature. On the other hand, the single base mismatch target produced a 50% lower signal than the fully complementary target at room temperature and provided fluorescence only at background levels when hybridization was carried out at 53 °C. Monitoring Hybridization in Real Time. Each microsphere’s fixed position made possible a hybridization study in real time (Figure 3). A DNA array containing identical beads was placed on the imaging system. The distal tip of the fiber bearing the microsphere sensors was placed in a labeled-target solution. Emission due to the hybridized labeled target was captured every minute for several minutes. In the small region of the imaging fiber selected for this study, ∼70 microspheres contained the probe complementary to the target in solution. Each microsphere was monitored independently and simultaneously. Signals from 40 beads were averaged to provide the kinetic data shown in Figure 3. Using a target concentration of 100 nM, hybridization was detected immediately, as seen by the steep slope of the response curve. The downward slope during the last 2 min of the experiment resulted from photobleaching of the fluorescent label Reproducibility and Regenerability. While the sensor remained on the imaging system, it could be regenerated by dipping the fiber tip into a room-temperature formamide solution. The
Figure 4. Kinetics. Four target concentrations were used to generate kinetic curves. The rate of hybridization was determined by dividing the signal generated at the final time in the assay by the final time (in minutes). The intensity per unit time was then plotted yielding a linear kinetic curve (inset).
same microspheres were assayed multiple times by placing the regenerated fiber into the target solution and repeating the experiment. As seen in Figure 3, the signal from the microspheres returns to background and the sensor can be used for multiple analyses with comparable results. Multiple assays of the same DNA array sensor (containing both IL2 and IL6 beads) were performed over several days. IL2 targets were consecutively hybridized and dehybridized 100 times. The stable performance with less than 8% standard deviation of the signal intensity exemplifies the robust nature of the DNA microspheres. At periodic intervals during the 100-assay IL2 test, microspheres carrying the IL6 probe in the same array were tested to see whether regeneration affected their response. Both probe types showed no compromise in response during the tests. Each array can be used for multiple tests since it is regenerated quickly and easily. The ability to reuse a single array hundreds of times significantly increases throughput and decreases the cost of each array. In addition, this reusability provides some interesting possibilities in regard to using the arrays for expression profiling. Hybridization Kinetics. Hybridization of the complementary target to the probe on the microsphere is rapid, predictable, and consistent. Figure 4 shows the hybridization kinetics at three target concentrations. Plotting the initial rates of the curves, Figure 4 inset, yields a straight line that can be used to determine unknown concentrations of target solutions. Sensitivity of the Microspheres. An important consideration with DNA analysis systems is the amount of sample needed for testing. For a specified absolute number of DNA molecules, the smaller the sensing volume, the higher local concentration of DNA and the less a sample needs to be amplified for detection. Small sample volumes, 4 µL of target solution and 10 µL of rinsing solution, are required with this system since only the tip of the 500-µm-diameter fiber is dipped into the solution. Sample concentrations of 100 pM are detectable in 10 min, eliminating extensive sample amplification. Hybridizations of nanomolar target solutions require several seconds and femtomolar solutions require hours for a detectable signals (Table 2). Analytical Chemistry, Vol. 72, No. 22, November 15, 2000
5623
Table 2. Microsphere Array Sensitivity: Time Required To Visualize Hybridization Using Different Target Concentrations [target]
hybridization time
100 pM 10 pM 100 fM 10 fM
10 min 30 min 4h 17 h
DISCUSSION Since fluorescein was used to label the DNA targets, we selected encoding dyes with spectral properties that would not overlap with the fluorescein spectrum. Covalently binding these dyes to the surface of the amine-functionalized microspheres yielded stable and reproducible signals. Unfortunately, such surface encoding reduces the number of amines available for the cyanuric chloride-activated oligonucleotide probe molecules. Therefore, the concentrations of the dyes were optimized to enable sufficient signals from both the encoding dyes and the hybridized targets. The finite number of surface amine groups reduces the range and number of dye combinations that can be generated with an external-labeling scheme. To increase the number of encoded microspheres, dyes also can be entrapped inside the beads. Lanthanide complexes are suitable for such internal encoding because they are readily soluble in organic solvents that can be used to swell the beads for internal loading. The dyes’ spectra are not compromised and their intensity remains constant once inside the microsphere. The DNA sequences employed in this work play important roles in immune function and cystic fibrosis diagnosis. Not only is detection of these sequences important but quantification in a sample can provide relevant clues to different disease states. Competition between labeled and unlabeled targets26 and monitoring the kinetics of hybridization are methods used to quantify analytes of interest. The detection of single base mismatches is an important feature of any genetic assay. The microsphere fiberoptic array described here completely discriminates between single base pair mismatches at elevated temperatures. This microarray has the shortest total assay time relative to other high-density DNA analysis systems and can monitor hybridization directly in the target solution. The high density of
5624
Analytical Chemistry, Vol. 72, No. 22, November 15, 2000
probes on each bead and the small bead size contribute to the short analysis time and sensitivity of the system. The microsphere fiber-optic system’s sensitivity has been demonstrated here to be 10 fM. Current work aims to demonstrate even higher sensitivity of the microsphere fiber-optic system. This system uses only 4 µL of sample, which further reduces the absolute number of DNA molecules required for an assay. The fiber-optic microarray also provides bead replicates, which improves confidence levels when signals from low target concentrations are analyzed. The DNA microarray presented here has smaller feature sizes and higher packing densities compared to other DNA arrays. We have demonstrated the fiber-optic microarray using a 500-µmdiameter imaging fiber with well diameters of 3.5 µm. Fibers have also been tapered to produce nanometer-scale wells serving as host to nanometer-diameter beads.32 Using longer fibers, the microarray sensor tip can be brought to the sample and used to sequentially test multiple solutions. Utilizing the imaging fiber’s remote sensing capabilities, arrays with nanometer dimensions potentially can be used for direct intracellular analysis. The advantages of this high-density randomly distributed micrometer-sized microsphere-based DNA array include costeffective production of the microbead arrays in seconds, highthroughput analysis, easy replacement with additional or different microspheres when different testing is desired, and facile regeneration of the sensor and substrate. In addition, the array can be brought to the sample solution rather than the solution being brought to the array. We are presently working on improving the sensitivity of the system to further reduce amplification requirements. With appropriate modifications, this general approach can be applied to the fabrication of libraries containing combinatorial peptides, antibodies, and other molecules. ACKNOWLEDGMENT The authors thank the National Institutes of Health for funding (GM 48142). F.J.S. acknowledges the Technology Foundation (STW), Technical Science Branch of The Netherlands Organization for Scientific Research (NWO), for a postdoctoral fellowship.
Received for review July 19, 2000. Accepted September 1, 2000. AC0008284
