Pyrosequencing in a Microfluidic Flow-Through Device - American

Nov 2, 2005 - SE-100 44 Stockholm, Sweden, and Biotage AB, SE-752 28 Uppsala, ... is currently run in a 96-plate format, as commercialized by Biotage...
0 downloads 0 Views 398KB Size
Anal. Chem. 2005, 77, 7505-7511

Pyrosequencing in a Microfluidic Flow-Through Device Aman Russom,*,† Nigel Tooke,‡ Helene Andersson,† and Go 1 ran Stemme†

Department of Signals, Sensors and Systems, Microsystem Technology, Royal Institute of Technology, SE-100 44 Stockholm, Sweden, and Biotage AB, SE-752 28 Uppsala, Sweden

To explore genome variation meaningfully, there is a critical need for a high-throughput and inexpensive platform for DNA analysis. Pyrosequencing is a nonelectrophoretic bioluminometric DNA sequencing method that uses a four-enzyme mixture reaction to monitor nucleotide incorporation in real time. Currently, the commercialized pyrosequencing technique is limited to a 96-microtiter plate format. However, high throughput and inexpensive pyrosequencing is required to meet the need of screening large numbers of samples. We present here DNA pyrosequencing on a nanoliter-volume microfluidic platform. The microfluidic approach involves the trapping of the DNA on microbeads in an on-chip filter chamber and flowthrough of the pyrosequencing reagents to monitor the reaction in real time. Two single-nucleotide polymorphisms were successfully scored to evaluate the microfluidic platform. In addition to significantly reducing reagent costs, microfluidic systems promise to improve the read length by eliminating intermediate product accumulation by constant removal of unincorporated nucleotides and elimination of dilution effects at each reaction cycle in the current plate format. Although only one filter chamber was used in this study, the platform should be readily adaptable to parallel analyses of nanoliter samples using filter chamber arrays to obtain high-throughput DNA analysis. The need for new automated high-throughput DNA sequencing technologies has been steadily increasing with the progress of the Human Genome Project. With the completion of the human sequence, attention is shifting to understanding how genetic variation, such as single-nucleotide polymorphism (SNP), leads to disease. This massive increase in sequence information must be coupled to increasingly faster, less expensive, and more powerful technologies for DNA analysis. Sanger sequencing1 was described in 1977 and is by far the most common sequencing approach used at present. The method has been continuously improved and was the workhorse in the Human Genome Project. However, its use in SNP genotyping is hampered by its low throughput and relatively high cost per sample. * Corresponding author. E-mail: [email protected]. Fax: +1 724 2999. † Royal Institute of Technology. ‡ Biotage AB. (1) Sanger, F.; Nicklen, S.; Coulson, A. R. Proc. Natl. Acad. Sci. U. S. A. 1977, 74 (12), 5463-7. 10.1021/ac0507542 CCC: $30.25 Published on Web 11/02/2005

© 2005 American Chemical Society

Several methods have been developed to characterize SNPs, including allele-specific hybridization,2 molecular beacon probes,3 flap endonuclease digestion,4 5’-nuclease TaqMan,5,6 oligonucleotide ligation assay,7 primer extension assay,8,9 and mass spectrometry.9 However, DNA sequencing is still the most accurate and informative technique because it defines both the location and nature of the changes. Recently, a user-friendly DNA sequencing method called pyrosequencing was described.10 Pyrosequencing is a nonelectrophoretic, real-time DNA sequencing method based on a four-enzyme mixture reaction that uses light release mediated by luciferase-luciferin as the detection signal. The technique relies on the incorporation of nucleotides by DNA polymerase and the release of pyrosphosphate (PPi), which is converted to adenosine triphosphate (ATP) by sulfurylase and then to detectable light by luciferase.10 The key advantage of pyrosequencing is that it produces high-quality data because it is derived from sequencing by synthesis with an easy-to-detect light signal that provides information not only on, for example, a single base but also on its sequence context. The method is unique in that it permits simple sequencing, even of longer sequences, without a separation step, such as electrophoresis. Pyrosequencing is currently run in a 96-plate format, as commercialized by Biotage AB (formerly Pyrosequencing AB). A number of the SNP technologies described have been subject to scale reduction and automation to meet the needs of increased throughput and reduced costs of reagents and samples. The trend in miniaturization has been moving toward high-density microtiter plates and recently toward the development of massive parallel assays in solid-phase microarray formats and homogeneous assays performed in individual channels in microfluidic devices. Microarray technology involving the synthesis or immobilization of probes into arrays of spots on a glass slide or other (2) Howell, W. M.; Jobs, M.; Gyllensten, U.; Brookes, A. J. Nat. Biotechnol. 1999, 17 (1), 87-8. (3) Tyagi, S.; Bratu, D. P.; Kramer, F. R. Nat. Biotechnol. 1998, 16 (1), 49-53. (4) Lyamichev, V.; Mast, A. L.; Hall, J. G.; Prudent, J. R.; Kaiser, M. W.; Takova, T.; Kwiatkowski, R. W.; Sander, T. J.; de Arruda, M.; Arco, D. A.; Neri, B. P.; Brow, M. A. Nat. Biotechnol. 1999, 17 (3), 292-6. (5) Livak, K. J.; Flood, S. J.; Marmaro, J.; Giusti, W.; Deetz, K. PCR Methods Appl. 1995, 4 (6), 357-62. (6) Whitcombe, D.; Brownie, J.; Gillard, H. L.; McKechnie, D.; Theaker, J.; Newton, C. R.; Little, S. Clin. Chem. 1998, 44 (5), 918-23. (7) Landegren, U.; Kaiser, R.; Sanders, J.; Hood, L. Science 1988, 241 (4869), 1077-80. (8) Pastinen, T.; Kurg, A.; Metspalu, A.; Peltonen, L.; Syvanen, A. C. Genome Res. 1997, 7 (6), 606-14. (9) Ross, P. L.; Lee, K.; Belgrader, P. Anal. Chem. 1997, 69 (20), 4197-202. (10) Ronaghi, M.; Uhlen, M.; Nyren, P. Science 1998, 281 (5375), 363, 365.

Analytical Chemistry, Vol. 77, No. 23, December 1, 2005 7505

solid support is a powerful platform for high-throughput genotyping by allowing many thousands of target DNA-probe interactions to occur in parallel. However, there are some disadvantages associated with this technology, including high cost, lack of standardization, lack of reproducibility, and lack of flexibility. An alternative approach to high-throughput DNA analysis is through the application of microfluidics, which involves manipulation of fluids in micrometer-sized channels. Interest in the development of microfluidic devices has been encouraged by the number of fluid-based processes, such as pyrosequencing, that could benefit from miniaturization. In addition to significantly reducing reagent costs, microfluidic systems promise to improve the performance of pyrosequencing by eliminating intermediate product accumulation caused by the enzyme apyrase as well as the dilution effect of nucleotide addition at each reaction cycle in the current plate format. A miniaturized pyrosequencing instrument was presented by Zhou et al.11 By using capillaries connected to the reaction chamber to deliver the dNTPs and an external photomultiplier detector, the total apparatus could be made as small as a notebook computer. However, the total reaction volume used was 100 µL, which is 2 times the volume used in the commercially available pyrosequencing instruments. The main reason for the large volume is to avoid dilution of the sample. It would, therefore be desirable to develop a microfluidic flowthrough system to avoid a dilution effect and intermediate product accumulation. 454 Life Sciences introduced recently a “PicoTiterPlate” platform to perform picoliter-volume PCR reactions of up to 300 000 discrete reactions simultaneously.12 The pyrosequencing principle was applied with the platform to resequence the adenovirus genome. The use of beads in microfluidic devices for surface-based biochemical assays offers several advantages, such as increased surface area and facilitated liquid handling. Furthermore, a wide variety of conventional chemistries are available for attaching different molecules such as DNA, RNA, proteins, and other molecules to beads. Here, we used beads as the solid phase for pyrosequencing in a microfluidic device. To perform bead-based pyrosequencing on microscale, the beads, reagents, and products must be controlled both in time and in space. We constructed a flow-through filter chamber device that fulfills the dual purpose of capturing the beads at specific places while the reagents are added and the reaction products are removed in a controlled manner.13 The microfluidic flow-through device used in this study was recently evaluated for SNP analysis by allele-specific extension using pyrosequencing chemistry.14,15 The extension reaction was based on a single-step runoff reaction whereby extendable 3′ ends of a primer/template complex were extended to the end of the template in the presence of all four nucleotides (dNTPs). In the current strategy, base-by-base sequencing-by-synthesis is performed in the device. Briefly, DNA captured in the flow-through (11) Zhou, G.; Kamahori, M.; Okano, K.; Harada, K.; Kambara, H. Electrophoresis 2001, 22 (16), 3497-504. (12) Leamon, J. H.; Lee, W. L.; Tartaro, K. R.; Lanza, J. R.; Sarkis, G. J.; deWinter, A. D.; Berka, J.; Weiner, M.; Rothberg, J. M.; Lohman, K. L. Electrophoresis 2003, 24 (21), 3769-77. (13) Andersson, H.; van der Wijngaart, W.; Enoksson, P.; Stemme, G. Sens. Actuators, B 2000, 67, 203-8. (14) Ahmadian, A.; Russom, A.; Andersson, H.; Uhlen, M.; Stemme, G.; Nilsson, P. Biotechniques 2002, 32 (4), 748, 750, 752, 754. (15) Russom, A.; Tooke, N.; Andersson, H.; Stemme, G. J. Chromatogr., A 2003. 1014 (1-2), 37-45.

7506 Analytical Chemistry, Vol. 77, No. 23, December 1, 2005

Figure 1. Principle of pyrosequencing. The four different nucleotides are added iteratively to the DNA template, which is incubated with a four-enzyme mixture. Incorporation of a nucleotide by the DNA polymerase leads to release of PPi, which is detected by ATP sulfurylase and luciferase in a coupled reaction. The added nucleotides are continuously degraded by apyrase, thus allowing addition of subsequent nucleotides.

chip and annealed to a short sequencing primer is exposed to a mixture of pyrosequencing reagents containing one of the four dNTPs. The DNA polymerase will only incorporate the nucleotide if its base is complementary to the base on the DNA template next to the 3’ end of the primer, thus triggering the bioluminescence cascade reaction. Since the identity of the added nucleotide is known, the sequence of the template can be determined by iteration with each of the four nucleotides with a washing step in between. We demonstrate that pyrosequencing based on microfluidics produces accurate SNP scoring data in nanoliter volumes. In addition, a preliminary study of pyrosequencing on a monolayer of immobilized beads in channels without the use of a physical barrier is presented. Single-base incorporation could be demonstrated, which is the first step toward developing a novel pyrosequencing on microarray format. Principle of Pyrosequencing. The principle of pyrosequencing is shown in Figure 1. Each of the nucleotides is added in turn to an extendable DNA substrate, consisting of a single-stranded DNA template to which a sequencing primer is annealed. Successful incorporation of the nucleotide is detected through an enzymatic cascade where released PPi is converted by sulfurylase into ATP, which in turn is used as substrate for the firefly luciferase. The result is emission of a quantitative light signal that can be measured by a CCD camera. A nucleotide will only be incorporated into the growing DNA strand if it is complementary to the base in the template strand. Unincorporated nucleotides are continuously degraded by apyrase, which has enabled sequencing by synthesis to be performed in solution in a microtiter plate. The pyrosequencing chemistry is based on a balance among the enzymes and substrates involved. Two reactions in the assay, (i) DNA polymerization by DNA polymerase and (ii) degradation of nucleotides by apyrase, are critical since the enzymes involved in these two reactions compete for the same substrate, dNTP. Slight changes in the kinetics of these reactions influence the whole sequencing reaction.

Figure 2. Principle of microfluidic-based pyrosequencing. (a,b) Beads carrying the DNA template are captured in the filter chamber. (c,d) The pyrosequencing mixture containing one of the four nucleotides is delivered into the chip by dispensing a precise volume and the reaction is monitored in real time. (e) After 1 min, the reaction is terminated by adding a washing buffer and applying pressure. After the washing step, a new reaction cycle is started with a mix containing one of the four nucleotides. (f) The chip is regenerated by washing with water and applying pressure from the outlet. The chip can then be filled with new beads for another assay.

In this report, a miniaturized microfluidic system of the pyrosequencing technology is presented. Apyrase is excluded in the microfluidic system, thus eliminating many of the current limitations in pyrosequencing, such as undesired product accumulation. The principle of solid-phase pyrosequencing in the microfluidic device is shown in Figure 2. Briefly, DNA immobilized on beads is captured in the reaction chamber of the device and is exposed to the reagents containing one of the four nucleotides for one reaction cycle. Iteration with each type of nucleotide followed by a washing step is performed. The reaction is detected in real time. MATERIALS AND METHODS Microfabrication of Filter Chamber and Monolayer Device. The flow-through device (Figure 3A) was created using standard photolithographic procedures and bulk micromachining of silicon as described in ref 13. The two-level-mask fabrication process involves deep reactive ion etching (DRIE) and anodic bonding. The 100-mm-diameter, 500-µm-thick p-doped silicon (100) wafers were used as the starting material. First, the front side was patterned and etched using DRIE (Surface Technology Systems) to define the inlet channel, the reaction chamber, the filter, the waste chamber, and the outlet channel. To seal the

device, a 300-µm-thick Pyrex glass wafer was anodically bonded to the front side. The back side was then patterned and a second DRIE step used to create fluid connectors. The silicon-glass stack was sawn into 9 × 5 mm chips. External polyethylene (PE) tubes were used as fluid connectors to the outlet of the chip. A guide wire was used to align the PE tube with the fluid openings on the chip during the tube fixing process. The silicon-glass stack was briefly heated to melt the PE tube locally and thus fix it onto the chip. To give additional strength to the assembly, the interface between the chip and the PE tubes was covered with epoxy glue. The microfluidic monolayer device (Figure 3B) was manufactured in the same way as described above. The front side of the silicon substrate was overexposed during the lithographic patterning to remove the pillars that made up the reaction chamber. This was followed by an etching step. The back side was then patterned and etched to create the fluid connectors. The silicon monolayer device with the cavity etched was sawed into 9 × 5 mm chips. Sample Preparation. The oligonucleotides (Table 1) were immobilized on nonmagnetic streptavidin-coated 5.5-µm beads (2 × 106 beads/PCR) (Bangs Laboratories, Inc.) by incubating 10 µL of the oligonucleotide or 1:1 mixture of two oligonucleotides, giving a final concentration of 1 µM, with 20 µL of beads in 30 µL of binding buffer (10 mM Tris-HCl, 2 M NaCl, 1 mM EDTA, 0,1% Tween 10) for 30 min with mixing. After a washing step with 10 mM Tris-acetate, pH 7.6, the beads bearing the immobilized oligonucleotides were resuspended in 30 µL of annealing buffer (100 mM Tris-acetate, pH 7.6, 50 mM magnesium acetate), and 1 µL (3 µM) of the primer NUSPT (Table 1) was added. Hybridization was performed by incubating at 95 °C for 4 min and then cooling to room temperature. Microcontact Printing in the Monolayer Device. The prepolymer and curing agent (Sylgard 184 Dow Corning) for poly(dimethylsiloxane) (PDMS) stamps were mixed in a 10:1 ratio according to the manufacturer’s instructions. The mixture was degassed in a vacuum for 10 min and then poured onto the silicon master wafer. The master was structured as shown in Figure 3B and treated with C4F8 to make it hydrophobic. The mixture was then cured for 15 min at 140° C. The PDMS slab was peeled from the wafer and cut into pieces. The PDMS stamp was incubated in 50 µL of 1 mg/mL biotin-bovine serum albumin (BSA) (SigmaAldrich, Inc.) for 45 min. Excess solution was removed, and the stamp was dried in a stream of nitrogen gas. The stamp was aligned under a microscope facing down to the surface of the cavity and incubated for a period of 10-15 min. Fifty microliters of bead suspension containing the DNA was incubated for 1 h on

Figure 3. Scanning electron microscope photograph of (A) the flow-through filter chamber device and (B) the chip with a cavity used to immobilize beads without the use of physical barrier.

Analytical Chemistry, Vol. 77, No. 23, December 1, 2005

7507

Table 1. SNPs Analyzed and Sequencing Primer Used in the Study template 1 2 3 4 5

5′ modification

sequences

biotin biotin biotin biotin biotin

5’-GCTGGAATTCGTCGTAACTGGCCGTCGTTTTACAAC-3’ 5’-GCTGGAATTCGTCGCAACTGGCCGTCGTTTTACAAC-3’ 5’- GCTGGAATTCGTCGTACTGGCCGTCGTTTTACAAC-3’ 5’-GCTGGAATTCGTCATACTGGCCGTCGTTTTACAAC-3’ 5’-GATGACTGTAACCCCAGTCATGGTGCACCTTTAGACTGG CCGTCGTTTTACAACG-3’

primer seq primer

the substrate. The unbound beads were then removed by careful washing in annealing buffer (100 mM Tris-acetate, pH 7.6, 50 mM magnesium acetate) solution. Two different streptavidin-coated beads were used: 5.5-µm nonmagnetic beads (Bangs Laboratories, Inc.) and 2.8-µm paramagnetic beads (Dynabeads M-280 streptavidin, Dynal A.S., Oslo, Norway). The cavity containing the DNA was sealed with PDMS to facilitate fluid flow and optical detection of the light signals. Pyrosequencing. Filter Chamber. Bead solution (5 µL) carrying single-stranded DNA with annealed primer was captured in the reaction chamber (50 nL). A dispensing unit with five compartments was filled with four separate pyrosequencing mixtures (excluding apyrase), one for each nucleotide, and a washing buffer containing annealing buffer. The pyrosequencing mixture contained exonuclease-deficient (exo-) Klenow DNA polymerase (Amersham Biosciences, Uppsala, Sweden), 0.7 unit/ µL; purified luciferase (BioTherma, Dalaro¨, Sweden), 27 ng/µL; recombinant ATP sulfurylase, 1.7 munits/µL; 75 mM Tris-acetate (pH 7.6); 0.5 mM EDTA; 35 mM magnesium acetate; 0.1% (w/v) BSA (BioTherma); 0.7 mM dithiothreitol; 10 µM adenosine 5′phosphosulfate; poly(vinylpyrrolidone), 0.4 µg/µL (360 000); Dluciferin (BioTherma), 100 ng/µL. The nucleotides at each dispensing compartment were dATP alpha S, 100 µM; dCTP, dGTP, and dTTP, 30 µM. Two hundred nanoliters of pyrosequencing mixture containing the desired nucleotide was delivered to the chip by the ink-jet dispenser. The reaction procedure was carried out in a dark box, and an external CCD camera monitored the reaction in real time. The reaction was terminated after 1 min by a washing step. The washing step included dispensing annealing buffer and simultaneously applying back pressure with a pump (parameter AB, Stockholm, Sweden) to remove all the reagents. The procedure was repeated with the next nucleotide. The data were obtained in Excel and graphic format. After each reaction, the microfluidic device was regenerated by applying pressure at the outlet to remove the beads from the reaction chamber. Monolayer Device. The same setup procedure was employed for the immobilized beads. After the immobilization of the beads, the chip was sealed with PDMS and placed on the setup facing down for optical detection. A total of 200 nL of the pyrosequencing reaction mixture, together with the desired nucleotide, was introduced into the inlet of the device by automatic dispensing. The light emitted during the reaction was detected with the CCD camera RESULTS AND DISCUSSION Chip Design and Liquid Handling. One of the main challenges in microfluidic systems is to obtain rapid and homo7508 Analytical Chemistry, Vol. 77, No. 23, December 1, 2005

5′-GTAAAACGACGGCCAGT-3’

geneous mixing. Mixing in such systems is typically dominated by diffusion, and a pure diffusion-based mixing process can be very inefficient, particularly in solutions containing macromolecules or particles that have very low diffusion coefficients. The PPi that initiates the luminometric pyrosequencing reaction is released into the liquid phase, and this presents particular challenges when performing pyrosequencing in a microfluidic device, particularly regarding mixing and detection. Therefore, it is important to address the issues of sample delivery and smallvolume liquid handling to prevent loss of detectable light in the microfluidic approach. Precise liquid handling was achieved by creating a stop-flow condition during the extension reaction, while flow-through was applied in the washing step to remove all the remaining nucleotides. The stop-flow feature was achieved using a dispenser based on ink-jet technology and capillary filling. A schematic illustration of the dispenser designed for precise pneumatic delivery of nanoliter volumes (>50 nL) is shown in Figure 4A. The dispenser was devised with five cartridge compartments and had a conelike shape with an angle that ensured the dispensed liquid droplet hit the same spot from all five compartments, one for each nucleotide mixture and a washing buffer. Hence, the setup facilitated automation without the use of a robot arm or other moving parts, thus enabling miniaturization of the system. In addition, no external mixing or other active fluid control mechanism was required. Once the liquid drop was delivered to the inlet channel, the entire chip was filled by capillary force. During capillary filling of the system with reagents, there was a net liquid flow that was forced through the bead pack and toward the outlet. This filling procedure happened typically in a few seconds, which is 1 order of magnitude faster than the reaction kinetics that produce the detectable PPi in the microfluidic device. Therefore, only a low amount of PPi was generated and lost to the outlet during the filling procedure. Figure 4B shows a picture of a pyrosequencing reaction taken after 12 s. The light produced is concentrated around the filter chamber where the beads are captured. The captured DNA was exposed to the liquid passing through the reaction chamber, and consequently, all the PPi molecules were released inside the reaction chamber. Once the device was filled and PPi production started, transport of PPi only occurred through diffusion (which is rather slow). Therefore, PPi was gradually accumulated in the reaction chamber. Thus, the loss of signal to the outlet was negligible, and reliable real-time analysis could be performed in the filter chamber device using the stop-flow feature. Sequencing. Two SNP variants were used to evaluate pyrosequencing in the microfluidic filter chamber device (Table 1). The sequencing primer was designed to hybridize with the template

Figure 4. (A) Schematic representation of the microdispenser used for precise pneumatic delivery of nanoliter-volume reagents. The dispenser has five cartridge compartments, and a liquid droplet from each compartment is delivered to the inlet channel of the chip. (B) Picture of the pyrosequencing reaction. The signal intensity is highest at the reaction chamber.

to give incorporation of an invariable nucleotide before the variable position, thus providing an internal control. Pyrosequencing was performed by five iterative dispensations to score each SNP position. Figure 5 illustrates the pyrosequencing data obtained for the (C/T) SNP variant (templates 1 and 2). The dispensing order was (CTAGC), as shown in Figure 5A. First, a noncomplementary nucleotide (dCTP) was added as a negative control and followed by a complementary positive control (dTTP). The incorporation was successful, as expected, and a light signal corresponding to one base was detected in all three variants. The height of the first peak was used to normalize the remaining signals. After each cycle, washing buffer was dispensed and back pressure was applied simultaneously to remove the reagents. After the first positive incorporation, the SNP position was interrogated. In this case, the first incorporation was followed by the addition of the nucleotide dATP. For the homozygous samples, a signal indicating the incorporation of one base was obtained in the case of homozygous T (template 2) as shown in Figure 5B, III. In the case of homozygous C (template 1, Figure 5B, I), the nucleotide was not complementary and no signal was generated in this cycle. At the fourth reaction cycle, the nucleotide dGTP was added and was incorporated in the homozygous C sample (Figure 5B, I), producing the corresponding light signal. In the heterozygous sample (Figure 5B, II, template 1/2), both alleles were present, and a signal corresponding to 0.5 nucleotide incorporation was expected and obtained, for each allele. As a postposition control, the nucleotide dCTP was added in the fifth cycle, and a light signal corresponding to one base was produced in all samples. Figure 6 shows similar results for the second SNP template (templates 3 and 4), which included a (G/A) SNP variation. The dispensing order of the nucleotides was (GACTG), and the expected sequence data were A, C, or T followed by G incorporation. The

high-quality sequencing data with a relatively high signal-to-noise ratio enabled unambiguous SNP calling. The results presented here were obtained using a chip with a reaction chamber volume of 50 nL and a total volume of dispensed reagents of 200 nL. As we have shown, five dispensations are more than enough for unambiguous SNP calling. Hence, it should be possible to analyze up to 50 samples in the microfluidic device using the same amount of reagents used in one standard pyrosequencing assay. The relatively high signal-to-noise ratio obtained indicates it should be possible to further miniaturize the system. We have obtained promising results using a reaction chamber volume of 12.5 nL and a dispensing volume of 50 nL (data not shown). However, the signal-to-noise ratio was too low to distinguish heterozygous samples with confidence. The background signal, which determines the detection sensitivity, is mainly due to PPi contamination in the pyrosequencing mixture. Thus, it is important to reduce the level of impurities in the reaction mixture if high-quality data are to be obtained from low reagent volumes. Each reaction step was allowed to run for 1 min followed by a washing step. The total reaction time was less than 15 s to achieve a maximum peak height. Hence, faster sequencing can be achieved by automating the washing step, which would reduce the cycling time. It should be possible to miniaturize the device further and fabricate it in arrays16 to perform many pyrosequencing reactions in parallel on the same chip and thus to increase throughput. The same type of CCD camera as in the original 96-plate format pyrosequencing was used in this study. Just a fraction of the area is used for a single chip. Thus, with appropriate lens system it should be possible to detect thousands of parallel microfluidic-based experiments without updating the (16) Andersson, H.; Wijnaart, W.; Stemme, G. Electrophoresis 2001, 22 (1), 24957.

Analytical Chemistry, Vol. 77, No. 23, December 1, 2005

7509

Figure 5. Pyrosequencing of templates 1 and 2 in the filter chamber device. (A) Short sequence of the two primed templates at the allelic position and the dispensing order of the nucleotide. The nucleotides in boldface type (C/T) indicating the SNP site are located in the second position from the end of the primer. The nucleotide dispensing order was “C-T-A-G-C”. (B) Pyrosequencing results for the three possible variants of the SNP site, shown as (I) homozygous C (template 1), (II) heterozygous C/T (representing 50/50 mixture of templates 1 and 2), and (III) homozygous T (template 2). The first dispensation was a mismatch, as negative control, while the second and last dispensations matched for all three variants, functioning as positive controls. Sequence data in proportion to the amount of bases incorporated were obtained for all variants.

detection system. However, in contrast to fluorescence detection where a powerful laser can be used to increase the signal, the signal from the luminescence assay is proportional to the amount of template in the reaction. Hence, further miniaturization will require more sensitive detection systems. DNA Pyrosequencing on an Immobilized Monolayer of Beads. We showed earlier that etched channels can be patterned with beads based on surface chemistry.17,18 The chemistry of internal channel surfaces of a silicon chip (Figure 3B) was 7510 Analytical Chemistry, Vol. 77, No. 23, December 1, 2005

Figure 6. Pyrosequencing of templates 3 and 4 in the filter chamber device. (A) Short sequence of the two primed templates at the allelic position and the dispensing order of the nucleotide. The nucleotides in boldface type (G/A) indicates the SNP site. The nucleotides were introduced in the following order: G-A-C-T-G. (B) Pyrosequencing results of the three possible SNP variants of templates 3 and 4: (I) homozygous A, (II) heterozygous G/A, and (III) homozygous G. The sequence data patterns are easily distinguished in all variants.

modified to enable the self-assembly of streptavidin-coated beads in areas coated with biotin-linked BSA. The cavity acted as a reaction chamber without any physical barriers and was connected to inlet and outlet channels allowing flow-through of the pyrosequencing reagents. The monolayer device containing the immobilized monolayer of activated beads was then sealed with PDMS to enable precise liquid delivery and optical detection. The same setup was used for liquid delivery and detection as for the (17) Andersson, H.; Jonsson, C.; Moberg, C.; Stemme, G. Talanta 2002, 56, 302-8. (18) Andersson, H.; Jonsson, C.; Moberg, C.; Stemme, G. Electrophoresis 2001, 22 (18), 3876-82.

Figure 7. Single-nucleotide incorporation using immobilized monolayered beads. Two types of beads, 5.5-µm nonmagnetic polystyrene beads (A) and 2.8-µm magnetic polystyrene beads (B) were tested. In both cases, a parallel experiment with a mismatching nucleotide (dTTP) addition was performed as a control for sequence specificity. Only the matched nucleotide addition (dCTP) gave a signal. Part of the DNA sequence (with the base to be sequenced seen in boldface type) is shown.

closed filter chamber device described above. The oligonucleotide template 5 was used in this study, and a mixture containing either the matching nucleotide dCTP or a mismatching dTTP was dispensed. As can be seen in Figure 7, a single-nucleotide incorporation using the immobilized beads could be detected. Two types of streptavidin-coated beads were used in this study: 5.5µm nonmagnetic polystyrene beads (Figure 7A) and 2.8-µm magnetic polystyrene Dynabeads (Figure 7B). As the reaction mixture containing the nucleotide dCTP flowed into the cavity of the monolayer device and came into contact with the DNA immobilized on beads, incorporation took place and the emitted light was detected. Another chip was prepared in parallel, and a mismatching nucleotide (dTTP) was added as a control for the sequence specificity. DNA polymerase should not incorporate the nucleotide, and consequently, no light was detected for the (19) Russom, A.; Haasl, S.; Ohlander, A.; Mayr, T.; Brookes, A. J.; Andersson, H.; Stemme, G. Electrophoresis 2004, 25 (21-22), 3712-9.

mismatch nucleotide. Multistep reactions could not be implemented because the open silicon chip used for the immobilization of the beads was too fragile to allow connection with tubing for the application of pressure. However, the results indicate the potential of performing base-by-base DNA sequencing on immobilized monolayer of beads in microfluidic cavities. The immobilization of the beads on the surface improves the accessibility of reagents to the sequencing template and should allow further miniaturization of the pyrosequencing technology. It took ∼30 s to reach a maximum peak height in the monolayer variant, which is a factor 2 slower as compared to the bead pack in the filter chamber. This is because the beads cover a larger area as compared with the filter chamber, where the beads are packed in a three-dimensional fashion. Therefore, the initiation of the reaction has to be controlled better to obtain synchronized reaction conditions in the monolayer device. The monolayer approach offers reduction in reagent use and flexibility in the size of the beads compared with the filter chamber device. We have earlier shown that monolayers of beads can withstand the forces generated by water flow in the channel18 and also withstand alkali treatment and heating.19 This should allow the integration of sample preparation steps and aid the development of pyrosequencing toward a bead-based microarray format. CONCLUSIONS We have successfully applied pyrosequencing on a microfluidic platform. Using the microfluidic platform, high-quality sequence information was obtained in nanoliter volumes. This dramatically reduced the enzyme and substrate consumption. The DNA template is immobilized on beads captured in the microfluidic device, thus enabling iterative washing. Hence, no intermediate inhibitory substances are accumulated after each reaction step, which promises increased sequence read length. Although only one flow-through device is used as a proof of concept, the microfluidic device can easily be mass-produced in arrays of filter chambers for parallel processing. The microfluidic approach can be integrated into microscale systems for increasingly automated sample processing from biological source to final data output. Integration of the detection would enable the development of a hand-held DNA analysis device for point-of-care application. ACKNOWLEDGMENT We thank Dr. Pål Nyre´n for many fruitful discussions about pyrosequencing. This work was supported by grants from VINNOVA. Received for review May 2, 2005. Accepted September 27, 2005. AC0507542

Analytical Chemistry, Vol. 77, No. 23, December 1, 2005

7511