Polychromatic Microarrays: Simultaneous ... - ACS Publications

7 Mar 2006 - Advancing beyond the standard red/green microarray experiment, a panel of eight report- ers were linked to eight B. anthracis samples and...
1 downloads 0 Views 757KB Size
Anal. Chem. 2006, 78, 2478-2486

Polychromatic Microarrays: Simultaneous Multicolor Array Hybridization of Eight Samples Jason R. E. Shepard*

The Wadsworth Center, The New York State Department of Health, Albany, New York 12208

High-throughput microscale platforms have transformed modern analytical investigations. Traditional microarray analyses involve a comparative approach, with two samples, a known control and an unknown sample, hybridized sideby-side and then contrasted for genetic differences. The samples are labeled with separate dyes and hybridized together, providing a differential expression pattern based on the reporter intensities. In contrast, the fiber-optic microarray platform described herein is analyzed with a microscope, thereby enabling the use of virtually any reporter, including quantum dots. The instrumentation takes advantage of the narrow emission bands characteristic of quantum dots to perform multiplexed detection of Bacillus anthracis. Advancing beyond the standard red/green microarray experiment, a panel of eight reporters were linked to eight B. anthracis samples and simultaneously analyzed in a microarray format. The ability to employ an assortment of reporters, along with the capacity to simultaneously hybridize eight samples confers an unprecedented flexibility to array-based analyses, providing a 4-fold increase in throughput over standard two-color assays. A growing trend in bioanalytical investigations is to perform multiple assays in parallel, increasing the throughput while decreasing the assay time, sample volume, and cost. Assays based on methodologies such as flow cytometry, real-time PCR, and microarrays have proven valuable, due in part to their ability to explore multiple parameters simultaneously. Microscale systems allow enormous levels of throughput; microarray platforms can simultaneously address tens of thousands of gene targets in a single assay. DNA microarrays have matured to the point that numerous assay formats are now available, as are array platforms that perform multiplexed protein assays and polysaccharide analyses.1,2 Essentially, the experiments originally designed as a Southern blot gel-based method have now developed the capacity to analyze entire genomes in a single assay.3,4 While the data garnered from such experiments provide enormous amounts of information, microarray studies are limited in that only two samples can be tested simultaneously. Limited sample throughput * To whom correspondence should be addressed. E-mail: jason_shepard@ hotmail.com. (1) Evanko, D. Nat. Methods 2005, 2, 88-89. (2) Epstein, J. R.; Biran, I.; Walt, D. R. Anal. Chim. Acta 2002, 469, 3-36. (3) Mockler, T. C.; Ecker, J. R. Genomics 2005, 85, 1-15. (4) Gunderson, K. L.; Steemers, F. J.; Lee, G.; Mendoza, L. G.; Chee, M. S. 2005, 37, 549-554.

2478 Analytical Chemistry, Vol. 78, No. 8, April 15, 2006

is an important problem for clinical diagnostics because more tests are being required for patients presenting a suspicious or unknown illness. For example, a number of microorganisms exist that are capable of causing respiratory infections presenting as general flu-like symptoms. The battery of tests required to determine the cause of an individual illness, let alone during seasons or situations in which increased numbers of patients present, often takes days to perform and could quickly overrun the current system in a time of crisis. Standard microarray experiments, such as gene expression analysis or clinical diagnostics, employ dual sample hybridizations, with a two-dye reporter system.2,5,6 A common microarray experiment is to compare samples for differentially expressed genes in an effort to identify genetic markers of disease such as cancer.7,8 In this type of experiment, a genomic sample from cancerous tissue is labeled with a green dye and is then compared to a red-dyed, noncancerous sample. The labeled materials from the two samples are pooled, simultaneously hybridized to an array and compared for differences. Although this assay can analyze many gene targets in parallel, only two samples are directly compared at one time. The limitations of this dual-sample assay format are, in large part, experimental limitations of the instrumentation, analysis system, and reporters used for most conventional array analyses.9-11 Generally, the hybridization patterns of the labeled genomic samples are analyzed by laser excitation. Laser excitation provides higher excitation energy than white light sources and, as a result, yields greater fluorescent emissions. Expansion of such a system beyond a two-sample experiment would require further laserexcitation wavelengths for the additional fluorescent reporters. Four-color assays are now common for such methodologies as real-time PCR and have been demonstrated for array-based analyses with advanced four-laser scanning systems.12 More elaborate systems with additional lasers, providing more excitation wavelengths, would be expected to substantially increase the cost and size of scanning instruments. Further, reporters in current (5) Schoumans, J.; Ruivenkamp, C.; Holmberg, E.; Kyllerman, M.; Anderlid, B.-M.; Nordenskjold, M. J. Med. Genet. 2005, 42, 699-705. (6) Bae, J.-W.; Rhee, S.-K.; Nam, Y.-D.; Park, Y.-H. Nucleic Acids Res. 2005, 33, e113. (7) Pinkel, D.; Albertson, D. G. Nat. Genet. 2005, 37, S11-S17. (8) Weigelt, B.; Peterse, J. L.; van’t Veer, L. J. Nat. Rev. Cancer 2005, 5, 591602. (9) Martinez, M. J.; Aragon, A. D.; Rodriguez, A. L.; Weber, J. M.; Timlin, J. A.; Sinclair, M. B.; Haaland, D. M.; Werner-Washburne, M. Nucleic Acids Res. 2003, 31, e18. (10) Wentzell, P. D.; Karakach, T. K. Analyst 2005, 130 (10), 1331-1336. (11) Jaiswal, J. K.; Simon, S. M. Trends Cell Biol. 2004, 14, 497-504. (12) Lindroos, K.; Sigurdsson, S.; Johansson, K.; Ronnblom, L.; Syvanen, A.-C. Nucleic Acids Res. 2002, 30, e70. 10.1021/ac060011w CCC: $33.50

© 2006 American Chemical Society Published on Web 03/07/2006

Figure 1. A SEM of beads (∼3 µm) in the etched wells of an optical fiber array. The array employs beads coated with ss-DNA and loaded on the etched fiber bundle face. In this case, the beads were loaded in low density, leaving unoccupied wells for future addition of other bead types if necessary.

use have potentially serious limitations. Standard organic fluorophores can cause problems that include insufficient detection capabilities and reporter photobleaching.13,14 Standard reporters provide relatively weak signals for low-abundance targets, thus limiting the type of assay that can be performed. Organic fluorophores also have demonstrated substantial photobleaching upon continual exposure to excitation light, especially with more intense laser excitations. In attempts to address these issues, various novel reporter moieties, such as doped nanoparticles, quantum dots, and dendrimers, have been employed.14-18 However, these advanced reporter species have only been incorporated into array experiments thus far in the standard two-color format. The present report describes a high-density fiber-optic microarray platform, quantum dot reporters, and a standard epifluorescent microscope combined into a system for simultaneous multicolor analysis of eight different samples. The array platform used in this study is an etched optical fiber bundle embedded with microbead sensors described in detail previously.19-23 Briefly, a fiber-optic bundle with a 1-mm diameter (13) Gao, X.; Fu, X.; Li, T.; Zi, J.; Luo, Y.; Wei, Q.; Zeng, E.; Xie, Y.; Li, Y.; Mao, Y. J. Biochem. Mol. Biol. 2003, 36 (6), 558-564. (14) Zhou, X.; Zhou, J. Anal. Chem. 2004, 76, 5302-5312. (15) Li, Y.; Cu, Y. T. H.; Luo, D. 2005, 23, 885-889. (16) Stears, R. L.; Getts, R. C.; Gullans, S. R. Physiol. Genom. 2000, 3, 93-99. (17) Gerion, D.; Chen, F.; Kannan, B.; Fu, A.; Parak, W. J.; Chen, D. J.; Majumdar, A.; Alivisatos, A. P. Anal. Chem. 2003, 75, 4766-4772. (18) Liang, R.-Q.; Li, W.; Li, Y.; Tan, C.-y.; Li, J.-X.; Jin, Y.-X.; Ruan, K.-C. Nucleic Acids Res. 2005, 33, e17. (19) Epstein, J. R.; Ferguson, J. A.; Lee, K. H.; Walt, D. R. J. Am. Chem. Soc. 2003, 125, 13753-13759. (20) Epstein, J. R.; Lee, M.; Walt, D. R. Anal. Chem. 2002, 74, 1836-1840. (21) Ferguson, J. A.; Steemers, F. J.; Walt, D. R. Anal. Chem. 2000, 72, 56185624. (22) Walt, D. R. Science 2000, 287, 451.

contains ∼50 000 independent 3-µm optical fiber array “spots”. Each spot is situated at the end of an independent optical cable that is selectively etched and loaded with a microbead sensor. An image of the beads localized into the etched wells of the fiberoptic array is shown in Figure 1. The beads are coated with singlestranded oligonucleotides and are used in multiplexed DNA assays. As in conventional microarray formats, labeled target nucleic acids are hybridized to the array. A high-resolution image showing a partial view of the array hybridized with a single reporter is shown in Figure 2. The 3-µm array spots were colored and overlaid using the imaging software. The blue intensities signal the fluorescence response from the target hybridization complex, and the green intensities indicate the individual fiberoptic array spots. An important attribute of the fiber-optic array platform is the uniform spot morphology, since the fluorescence is imaged through the fiber bundle. Many array formats have problems with highly variable spot boundaries, which can complicate array analysis. Whereas traditional two-color array-based analyses involve a red/green assay, the present fiber-optic platform combines organic fluorophores with quantum dot reporters for multicolor analysis. Quantum dots are luminescent semiconductor nanoparticles that are suitable for integration with the fiber-optic platform because their properties include tunable emission profiles and a single excitation band.24-27 The fiber-optic array is viewed through a (23) Shepard, J. R. E.; Danin-Poleg, Y.; Kashi, Y.; Walt, D. R. Anal. Chem. 2005, 77, 319-326. (24) Chan, W. C. W.; Nie, S. Science 1998, 281, 2016-2018. (25) Bruchez, M., Jr.; Moronne, M.; Gin, P.; Weiss, S.; Alivisatos, A. P. Science 1998, 281, 2013-2016. (26) Gao, X.; Nie, S. Anal. Chem. 2004, 76, 2406-2410. (27) Han, M.; Gao, X.; Su, J. Z.; Nie, S. Nat. Biotechnol. 2001, 19, 631-635.

Analytical Chemistry, Vol. 78, No. 8, April 15, 2006

2479

Figure 2. High-resolution image of the fiber-optic microarray and the 3-µm array spots after hybridization to the 16S rRNA gene target. The fluorescent intensities were measured, colored, and overlaid using the imaging software. The hybridization response is shown in blue; the green shade illustrates the individual fiber-optic array spots.

conventional microscope, with a white light source and multiposition filter wheels, providing excitation across the visible region, along with the capability to shutter between different filter combinations for detection. This experimental design has been used to analyze eight different reporter dyes in parallel, allowing the platform to detect eight different Bacillus anthracis samples simultaneously. EXPERIMENTAL PROTOCOL Materials and Methods Three Cy-dye streptavidin conjugates were purchased directly from Amersham Biosciences (Piscataway, NJ), and three quantum dot streptavidin conjugates were from Quantum Dot Corporation (Hayward, CA). Two quantum dots, QD560-Adirondack Green and QD660-Maple Red, were obtained from Evident Technologies (Troy, NY) and were coupled to streptavidin (Promega, Madison, WI) following the manufacturer’s protocol. The following materials were obtained from the listed sources: microspheres, Bangs Laboratories, Inc. (Carmel, IN); synthetic oligonucleotides, Integrated DNA Technologies, Inc. (Coralville, IA); optical filters, Chroma Technologies (Brattleboro, VT); microscope instrumentation (Optical Analysis, Nashua, NH); fiber bundles (Schott Fiber Optics Inc., Southbridge, MA) Microarray Development and Assembly. Microarray fabrication was performed as previously described.20,21,23 Pools of ∼3µm oligonucleotide-functionalized microspheres were loaded onto the etched face of a fiber-optic bundle, total 1-mm diameter. These arrays were developed for B. anthracis detection. Each of 10 different sequences, directed toward the Bacillus genus, was attached to beads and added to the array. The arrays were fabricated with the beads present in low density; each bead type 2480

Analytical Chemistry, Vol. 78, No. 8, April 15, 2006

was represented by between 15 and 50 replicates in the array, but for simplicity, all data were analyzed on the basis of the average response of 10 identical beads. The decision to analyze 10 replicate beads is based on previous work with the fiber-optic array.20 All PCR was performed with DNA isolated from B. anthracis, Sterne strain 34F2. Probe and primer sequences were derived from the GenBank database and were analyzed using assorted public domain, Web-based software tools. The hybridization results are in response to a single gene target, coding for the 16S rRNA gene of the bacterium. The probe sequence, primer sequences and PCR conditions used for this study: forward primer, 5′-CTg-ggA-TAA-CTC-Cgg-gAA-AC-3′; reverse primer, biotin-5′-Cgg-gTC-CAT-CCA-TAA-gTg-AC-3′; probe sequence, 5′gCT-AAT-ACC-ggA-TAA-CAT-TTT-gAA-3′. PCR. PCR was performed using Accuprime Supermix (Invitrogen Corporation, Carlsbad, CA). Asymmetric PCR was performed following manufacturer protocols for 50-µL reactions, except for the primer concentrations, for which 200 nM biotinylated reverse primer and 20 nM forward primer was used. One microliter of DNA template from the bacterial culture extractions was used. The amplification profile included an initial step at 94 °C for 5 min, followed by 30 cycles of 95 °C for 1 min, 48 °C for 1 min, and 68 °C for 1 min, and a final extension step at 72 °C for 7 min. Optical Filters. Filter optics for each reporter are listed as (excitation filter, dichroic beam splitter, emission filter), and individual filters are described in terms of (center wavelength ( band-pass). SP stands for short-pass filter: Cy2(480 ( 15, 505, 535 ( 20), Cy3(540 ( 12, 565, 585 ( 10), Cy5(640 ( 10, 660, 640 ( 10), QD525(460sp, 474, 535 ( 20), QD560(460sp, 474, 565 ( 10), QD585(460sp, 474, 585 ( 10), QD605(460sp, 474, 605 ( 10),

Figure 3. Assay design employing array hybridization and quantum dot reporters. The figure shows an image of the fiber-optic array (a) prior to hybridization and (c) post hybridization. (b and d) The assay starts with beads coated with ss-DNA, followed by hybridization of QD-labeled target DNA. The target DNA incorporates a biotin via the PCR process, which is linked to a streptavidin-QD conjugate to provide the signal upon formation of the hybridization duplex. The hybridization is highly specific, such that nonspecific hybridization to other beads in the array does not result in appreciable signal.

QD660(460sp, 474, 640 ( 10), QD705(460sp, 474, 715 ( 15), QD800(460sp, 474, 820 ( 20). Hybridizations. Experiments were performed in groups, on reusable fiber-optic arrays. The array positions of the 16S beads were decoded with a 10-s hybridization (10 µL of a 1 µM complementary sequence). For single-reporter hybridizations, asymmetric PCR products were combined 1:1 with hybridization buffer (0.2 M phosphate-buffered saline, pH 7.4 [PBS]). Twenty microliters of the combined solution was denatured at 95 °C for 5 min and then flash-frozen. The frozen, denatured PCR products were warmed to room temperature and then allowed to hybridize to the multiplexed array for 30 min. The streptavidin-reporter conjugates (10 µL; ∼100 nM concentrations) were allowed to incubate with the hybridized array for 20 min to ensure complete labeling. For multiplex hybridizations, the asymmetric PCR products were purified via Qiagen PCR purification kits and were mixed with the streptavidin-reporter conjugates (10 µL, ∼100 nM, 1 h). The labeled samples were then pooled in equal proportions (10 µL each; 80 µL total volume), warmed to 40 °C, and hybridized for 1 h. The array was then scanned at each reporter excitation/emission wavelengths to determine the hybridization pattern of each sample. All hybridizations, incubations, and washing steps were carried out at room temperature unless otherwise stated. Between assays, a formamide rinse (90% in PBS) was used to dehybridize the array. The PBS hybridization buffer was used to wash the array after each step. Image integration varied for the reporter panel as followed: Cy2, QD525, QD560, QD585, QD605 (50 ms each); QD660 (100 ms); Cy3 (250 ms); QD705, QD800 (500 ms); Cy5 (1 s). Image acquisition, analysis,

and processing was performed using IPLab software (Scanalytics, Rockville, MD) RESULTS Bead/Array Decoding. The fiber-optic array fabrication process is such that the positioning of oligonucleotide-functionalized beads is random. An aliquot of bead solution is added to the fiber bundle face, and the beads localize indiscriminately in the etched wells. Previous fiber-optic array designs used dye “bar codes” to register the bead positions in the array.20,21 Because this work employs every available optical channel for a reporter, bead registration is performed in a different manner, taking advantage of the fact that these arrays are reusable; over 100 sequential hybridizations have been performed, with negligible loss in performance.21 After bead addition, the bead locations were identified with a preexperimental hybridization to fluorescently labeled complementary oligonucleotides. These assays served to identify the position and sequence of each 16S bead. Hybridization targets were dehybridized by exposure to the organic denaturant formamide, thereby regenerating the array for the additional core experiments. Thus, all orthogonal optical channels were available for a reporter. Hybridization of a Single Reporter. Arrays were fabricated for detection of B. anthracis. The extracted DNA was subjected to asymmetric PCR with biotinylated primers, and the resulting amplicons were used for hybridization. The hybridization of biotinylated oligonucleotides were signaled by means of a series of streptavidin-reporter conjugates (Figure 3). An initial panel of 10 fluorescent reporters was employed, with emissions spanning Analytical Chemistry, Vol. 78, No. 8, April 15, 2006

2481

Figure 4. Single-reporter hybridization results. Each reporter was used individually but was scanned at all reporter wavelengths. The green peak is the response of the reporter at its excitation and emission; the remaining peaks are the intensities of the other excitation and emission combinations. In this case, three reporters had no signal bleed-through at the wavelengths of the other reporters, but one did (shown in red). For each assay, the measured intensities were normalized to that of the specific reporter emission.

the entire visible spectrum, including three fluorophores (Cy2, Cy3, and Cy5) and seven quantum dots (QD525, QD-Adirondack Green-560, QD585, QD-Maple Red-620, QD655, QD705, and QD800). For each dye used, the hybridization results are presented as the average response of 10 identical beads containing a target sequence coding for the 16S rRNA of the bacterium. The signal intensities were determined by subtraction of the average background intensity from the average hybridization intensity. Acceptable signal cutoff points were determined on the basis of net signal greater than 3× the standard deviation (3σ) of the background intensity. A series of consecutive single-sample, single-reporter experiments was performed, first hybridizing the biotinylated 16S target and then incubating with 1 of the 10 different streptavidin-reporter conjugates. The single reporter experiments were executed with hybridization first and then subsequent labeling to ensure the hybridization was identical and that the reporters themselves did not unduly influence the hybridization efficiency due to such factors as the larger size of the quantum dots. This procedure established that the amount of hybridizing target would be the same for each experiment, and then the streptavidin reporter step could be extended to ensure complete labeling. In each case, after the reporter had been linked to the array, the fluorescence intensities were measured at the excitation and emission wave2482

Analytical Chemistry, Vol. 78, No. 8, April 15, 2006

lengths specific to all 10 reporters. For example, the QD655 reporter was used in a single reporter assay, but the protocol was scanned for all 10 reporter excitations and emissions. Similar single-reporter assays were performed for the other 9 reporters. The results from 4 of the 10 single-reporter assays are shown in Figure 4, with the correct signal shown in green and signal bleedthrough shown in blue. Instances that the nonspecific signal bleedthrough was determined to be >3σ are marked in red. This figure shows three reporters, Cy2, Cy3, and QD655, for which no appreciable spectral overlap was measured, and QD800, for which substantial bleed-through signal (>3σ) with a neighboring reporter occurred, thus precluding the use of the entire 10-reporter panel in multiplexed hybridizations. The single-reporter hybridization results from the entire reporter panel are included in Table 1 (and graphically in the Supporting Information). The table summarizes the single reporter experiments and highlights the problems of spectral overlap between a few of the reporters. All signal values were normalized to the specific reporter signal, which is shown in green. Signal due to spectral bleed-through had values >3σ and are labeled in red. Because two of the reporters, QD585 and QD705, resulted in bleed-through with two neighboring reporters (Table 1a), they were logical choices to be eliminated, and the remaining multiplexed panel of eight reporters exhibited no spectral overlap (Table 1b).

Table 1. (a) Results from the Ten Single Reporter Hybridizations, Scanned Across the Multiplexed Panel; (b) By Eliminating Two Reporters, QD585 and QD705, the New Panel Exhibiting No Nonspecific Value >3σ

a Results normalized to the correct reporter response are shown in green, and those indicative of spectral bleed-through (>3σ) are shown in red.

Hybridization of Multiplexed Reporters. On the basis of the single reporter results, the eight reporters were combined in a multiplex assay. For multiplexed hybridizations, biotinylated samples labeled with their streptavidin reporters were hybridized to the array together. Similar hybridizations were performed with a range of reporter subsets, from standard duplex experiments up to eight samples simultaneously hybridized to the array. Figure 5 shows the results from multiplexed hybridization experiments in which four, six, seven, and eight different reporters were employed simultaneously. In these hybridization experiments, the residual signal from reporters not present in the assay was 50-fold greater than those published. In addition, if bead saturation is expected to be a problem, the flexible fabrication format of the fiber-optic array allows control over the number of beads, and arrays with more beads can be easily prepared. The merging of a microarray platform with quantum dots has allowed the level of multiplexed analysis to be increased 4-fold over standard methods. Quantum dots have other advantages, including increased brightness and photostability as compared to conventional dyes, and will play an important role in future array-based experiments. The present platform demonstrates the ability to perform analyses of genes coding for pathogenic microorganisms and will be an important tool for public health diagnostics. Although the present experiments function in a detection capacity and employ a PCR amplification step, the ability to employ multiple indicators should have an impact on future array-based studies. Non-PCR array analysis might not be capable of running eight samples efficiently, depending on the expression levels interrogated. Even so, hybridization of more samples in parallel would still increase the throughput and experimental significance relative to dual sample, comparative analyses. Experiments capable of multisample hybridization could probe multiple hypotheses in a single experiment, comparing multiple samples or parameters to a standard, a capability lacking in the current two-color format.34 Another advantage is that multiple standards can be included and can be employed as boundaries to the relevant data, thereby bracketing an unknown response. This multiplestandard capability is critical; the majority of microarray setups currently provide only ratiometric results, rather than any truly (33) Steinberg, G.; Stromsborg, K.; Thomas, L.; Barker, D.; Zhao, C. Biopolymers 2004, 73, 597-605. (34) Dudoit, S.; Shaffer, J. P.; Boldrick, J. C. , Working Paper 110, U. C. Berkeley Division of Biostatistics Working Paper Series 2002.

Analytical Chemistry, Vol. 78, No. 8, April 15, 2006

2485

quantitative value. The use of multiple standards will also help to ensure cross-platform comparability in terms of experimental variability but, most importantly, in terms of hybridization results. ACKNOWLEDGMENT The author acknowledges Drs. Israel Biran, Nick Cirino, and Adrianna Verschoor for reviewing the manuscript and helpful discussions. The author also thanks Evident Technologies for providing two of the quantum dots, QD560-Adirondack Green and QD660-Maple Red, and Dr. William Samsonoff of the Wadsworth

2486

Analytical Chemistry, Vol. 78, No. 8, April 15, 2006

Center Electron Microscopy Core facility for help with scanning electron microscopy used in Figure 1. SUPPORTING INFORMATION AVAILABLE Supporting Infomation as noted in text is available. This material is available free of charge via the Internet at http:// pubs.acs.org. Received for review January 3, 2006. Accepted February 9, 2006. AC060011W