AN INTEGRATED MICROFLUIDIC PROCESSOR FOR SINGLE NUCLEOTIDE POLYMORPHISM-BASED DNA COMPUTING William H. Grover and Richard A. Mathies Department of Chemistry, University of California, Berkeley, CA 94720, USA ABSTRACT An integrated microfluidic processor is developed for performing DNA computations using single nucleotide polymorphisms (SNPs) to represent binary bits. Fluorescent input DNA “answers” are captured by hybridization to complimentary oligonucleotides bound to magnetic beads in the microfluidic device. The beads are suspended within the processor by a magnetic field, and integrated microfluidic valves and pumps circulate the input DNA through selected bead suspensions. Input DNA remaining after a series of capture/rinse/ release steps provides the solution to a Boolean logic problem. The improved capture kinetics and single-base specificity enabled by microfluidics make our processor well-suited for performing larger-scale DNA computations as well as haplotyping assays of disease traits. Keywords: DNA computing, hybridization array, single nucleotide polymorphisms, haplotyping The monolithic membrane valves and diaphragm pumps developed previously by our group [1] have catalyzed the development of complex integrated sample preparation and analysis systems for DNA sequencing [2], pathogen and infectious disease detection [3], automated evolution of RNA catalysts [4], and amino acid analysis for space exploration [5]. In the work presented here, these membrane valves and pumps are used to route DNA “information” through a series of capture and release computing operations [6]. The microfluidic processor shown in Figure 1 includes sixteen 1.1 µL capture chambers connected to common fluid bus channels by 32 bus valves. Magnetic beads derivatized with 11-base capture oligonucleotides are pumped into the chambers (Figures 2B and 2C) and held in suspension within each chamber by an external magnetic field. Complementary 11-base fluorescein-labeled “input” oligonucleotides are synthesized with polymorphic bases at positions 3, 6, and 9. The identity of each biallelic polymorphic base (A or T) is used to encode the value of a binary bit (TRUE or FALSE), and the complete population of input oligonucleotides represents all possible solutions to a three-bit Boolean logic problem. The computation consists of a series of four capture/rinse/release steps. In each step, perfectly-complementary input oligonucleotides are captured during circulation through the selected bead suspension (30 min., ~22 cycles through the suspension). Mismatched input oligonucleotides remain in solution and are eliminated when rinse buffer is pumped through the bead suspension. Fluorescence images of beads following SNP-specific hybridization are shown in Figure 3. The capture chamber is then heated and the released oligonucleotides are circulated through the next bead suspension on the opposite side of the device (Figure 2D). The four back-and-forth capture/rinse/release steps correspond to the four clauses in the Boolean satisfiability problem in Figure 4. If two or more different input oligonucleotides are captured in the same step, they must satisfy the Boolean OR of the requirements of the different capture oligonucleotides in the chamber. Similarly, input oligonucleotides retained through a series of capture/rinse/release steps must satisfy the
9th International Conference on Miniaturized Systems for Chemistry and Life Sciences October 9-13, 2005, Boston, Massachusetts, USA
214
0-9743611-1-9/µTAS2005/$20©2005TRF
Boolean AND of the requirements in each chamber. Fluorescence imaging of the capture beads after each step verifies the elimination of incorrect answers (Figure 4). Input oligonucleotides remaining at the end of the computation represent the solution to the satisfiability problem. By circulating the remaining oligonucleotides simultaneously through pairs of bead suspensions with capture oligonucleotides that differ only at a single polymorphic base, the final values of each bit are read out (Figure 5). The improved capture kinetics, step-to-step transfer efficiency, and single-base specificity enabled by microfluidics make possible this first use of single nucleotides as binary bits in a hybridization-based DNA computation. Our microfluidic processor can detect SNP mismatches with only 1% error. Step-to-step transfer efficiencies of 92% are observed while still maintaining SNP specificity. Thermodynamic calculations suggest that an 8-bit, 10-clause problem could be solved using SNPs and our microfluidic processor. Additionally, by performing AND- and OR-like operations on genomic DNA fragments, we anticipate that our processor could be used to perform SNP haplotyping assays of complex disease traits.
(A) Pneumatic access holes
10 mm
(B) Bead suspension
Pneumatic channel
Diaphragm pump Bus valves for chamber selection Fluid bus channel
Input/output reservoirs
Figure 1: The microfluidic processor (A) and an oblique view of a single capture chamber containing a bead suspension (B). The polydimethylsiloxane (PDMS) membrane sandwiched between the fluidic and pneumatic wafers seals the capture chambers from the fluid bus channels. Applying a vacuum via the pneumatic access holes deflects the membrane away from the fluidic wafer and opens the bus valves. By selecting capture chambers on the left and right sides of the device, a 5.8 µL recirculating loop is formed.
!
"
#
$
Figure 2: Schematic of fluid flow within the microfluidic processor (A) when chambers on the right (B) or left (C) sides of the device are being loaded and when DNA “answers” are being transferred from one chamber to another (D).
9th International Conference on Miniaturized Systems for Chemistry and Life Sciences October 9-13, 2005, Boston, Massachusetts, USA
215
Perfect match:
SNP mismatch:
Capture: "IO AG7TC4CA7GT Input: &!- AC7TG!GA7CT
Capture: "IO AG7TC!CA7GT Input: &!- AC7TG!GA7CT
[NOT(B0) OR B2]
"IO AG!TC7CA7GT "IO AG7TC7CA4GT
AND
[B0 OR B1]
Figure 3: Fluorescence images of capture beads following 30 min. recirculation of perfectlycomplimentary or single-base mismatched fluorescent input DNA through the beads and 10 min. rinse with hybridization buffer. W represents T or A.
AND [NOT(B1) OR NOT(B2)] AND [NOT(B0) OR B1]
"IO AG4TC7CA7GT "IO AG7TC4CA7GT
"IO AG7TC!CA7GT "IO AG7TC7CA!GT
"IO AG!TC7CA7GT "IO AG7TC4CA7GT
step 1 6 answers remain
2 4 answers
3 2 answers
4 1 answer (correct)
4 (high contrast) 1 answer (correct)
Figure 4: Fluorescence images of capture beads after each of the four computation capture/rinse/ release steps (T = TRUE and A = FALSE for biotinylated capture DNA). The observed decrease in fluorescence corresponds to the elimination of input DNA representing incorrect answers to the problem. "IO AG!TC7CA7GT
"IO AG4TC7CA7GT
B0 TRUE
FALSE
D
"IO AG7TC!CA7GT
"IO AG7TC4CA7GT
B1 TRUE
FALSE "IO AG7TC7CA!GT
"IO AG7TC7CA4GT
Figure 5: Fluorescence images of the three “readout” steps. The pattern of fluorescence indicates that the remaining input oligonucleotides encode B0 = FALSE, B1 = TRUE, and B2 = FALSE, the correct answer to the problem in Figure 4.
B2 FALSE
TRUE
1. 2. 3. 4.
W.H. Grover, A.M. Skelley, C.N. Liu, E.T. Lagally, and R.A. Mathies, Sens. Act. B 89, 315 (2003). B.M. Paegel, R.G. Blazej, and R.A. Mathies, Curr. Opin. Biotech. 14, 42 (2003). E.T. Lagally et al., Anal. Chem. 76, 3162 (2004). B.M. Paegel and G.F. Joyce, “Microfluidic serial transfer circuit: Rapid and automated evolution of RNA catalysts,” Ninth International Conference on Miniaturized Systems for Chemistry and Life Sciences (µTAS), Boston, October 9-13, 2005. 5. A.M. Skelley et al., Proc. Natl. Acad. Sci. U.S.A. 102, 1041 (2005). 6. W.H. Grover and R.A. Mathies, submitted. Corresponding author: Richard A. Mathies,
[email protected]. This research was supported by generous donations from Affymetrix to the Berkeley Center for Analytical Biotechnology.
9th International Conference on Miniaturized Systems for Chemistry and Life Sciences October 9-13, 2005, Boston, Massachusetts, USA
216