Rapid self-assembly of DNA on a microfluidic chip
© Zheng et al; licensee BioMed Central Ltd. 2005
Received: 21 July 2004
Accepted: 18 February 2005
Published: 18 February 2005
DNA self-assembly methods have played a major role in enabling methods for acquiring genetic information without having to resort to sequencing, a relatively slow and costly procedure. However, even self-assembly processes tend to be very slow when they rely upon diffusion on a large scale. Miniaturisation and integration therefore hold the promise of greatly increasing this speed of operation.
We have developed a rapid method for implementing the self-assembly of DNA within a microfluidic system by electrically extracting the DNA from an environment containing an uncharged denaturant. By controlling the parameters of the electrophoretic extraction and subsequent analysis of the DNA we are able to control when the hybridisation occurs as well as the degree of hybridisation. By avoiding off-chip processing or long thermal treatments we are able to perform this hybridisation rapidly and can perform hybridisation, sizing, heteroduplex analysis and single-stranded conformation analysis within a matter of minutes. The rapidity of this analysis allows the sampling of transient effects that may improve the sensitivity of mutation detection.
We believe that this method will aid the integration of self-assembly methods upon microfluidic chips. The speed of this analysis also appears to provide information upon the dynamics of the self-assembly process.
There has been a rapid growth in the number of applications that are based upon DNA self-assembly, ranging from DNA microarrays (e.g. Affymetrix ) in the life sciences, through conformation-based mutation detection methods [2, 3], to the ongoing development of DNA scaffolding methods of nanoassembly . The control of the degree of DNA hybridisation requires elaborate and time consuming sample preparation (eg ) with methods that may require hours to achieve hybridisation , and on the order of an hour even within miniaturised systems [1, 7]. However, a rapid method of controlling denaturation and renaturation within a microfluidic device would enable an inexpensive mutation detection method that could be performed within minutes.
Microfluidic devices or 'microchips' are photolithographically-defined networks of microchannels in glass where the microchannels are similar in size to conventional capillaries. These microchips provide compelling advantages in terms of speed, reagent usage and integration over conventional capillary or gel-based methods. The potential of the microfluidic chip has led to the use of terms such as "micro-total analysis systems" and "lab-on-a-chip". These microchips have been demonstrated in conjunction with a range of applications that integrate the polymerase chain reaction (PCR) and capillary electrophoresis (CE) methods with some reaching nanolitre or smaller scale volumes. A powerful advantage of the microchip approach is that it can implement much the same molecular biology protocols and reagents as used with conventional equipment, thereby allowing a wealth of established expertise to be transferred to the microscale.
Although the most effective method of mutation detection is sequencing, it is also by far the most expensive . The microarray  technique, although powerful, is still handicapped by significant false positive rates and high cost . Alternative methods based on DNA self-assembly are much faster than sequencing and these include single-strand conformation polymorphism (SSCP), denaturing high performance liquid chromatography (DHPLC) and heteroduplex analysis (HA). Although their cost has been shown to be far lower than sequencing, the achievable sensitivities (the percentage of mutations that are successfully detected) are only about 90 % [10, 9].
Microfluidic chips may enable extremely high throughputs and high levels of integration. The achievement of this goal has been hindered by the lack of successful integrations of methods of mutation analysis based on single-stranded DNA (ssDNA) and double-stranded DNA (dsDNA) – likely due to the difficulties in controlling the degree of hybridisation on chip without time consuming thermal processing. A great advantage would be provided by a method of enabling microchip-based control of a rapid DNA self-assembly process.
The term wildtype is used to describe any given genetic sequence that does not contain mutations. Since individuals usually carry two copies of each gene, the genetic sequence of the two copies may be identical (homozygous) or may differ (heterozygous). DNA is normally double stranded, but under some conditions (e.g. high temperature), melts into single strands. Under other conditions, such as a lower temperature, these single strands will self-assemble into the double-stranded form again. The resulting double-stranded DNA is referred to as a homoduplex if the sequences are perfectly complementary, or a heteroduplex if the sequences are nearly complementary (i.e. a mutant sequence paired with a wildtype sequence). The misfit in a heteroduplex creates a "bulge" or "bubble" where the bases do not match and this affects the shape of the assembled molecule, typically lowering its velocity during electrophoretic movement, i.e. the heteroduplexes typically migrate more slowly than the homoduplexes. Any heterozygous sample will generate four different duplexes, two homoduplexes and two heteroduplexes, although the molecules often co-migrate so that fewer than 4 separate electropherogram peaks are resolved.
In the HA method, electrophoretic conditions are chosen in order to enhance the velocity differences between the duplexes so that the process of duplex self-assembly can be used to determine the presence of a heterozygous state (hence indicating the presence of a mutation). In SSCP, isolated strands of ssDNA find near-complementary sequences on the same strand, with the result that the strand folds upon itself in a sequence dependent manner forming new conformations. This is a simplistic description since ssDNA without self-similar sequences, and homoduplex dsDNA, may also take complex forms. Techniques such as HA that aim to separate homoduplex fragments from heteroduplex fragments often use some combination of thermally and chemically denaturing conditions to cause the partial melting of the duplex, resulting in a shift in mobility or chromatography column retention time that increases with the degree of mismatch.
Many medical diagnostics could be implemented on microchips if an effective implementation of a highly sensitive mutation analysis method could be integrated with PCR/CE. Considerable work has been done in developing SSCP  and HA [2, 3, 12]. An excellent review of such methods has been produced by Jin et al. . The main drawback is the lower sensitivity of these methods. In macroscopic work Kozlowski and Krzyzosiak  and Kourkine et al.  greatly improved their sensitivities by combining SSCP and HA methods to develop capillary-based electrophoretic techniques with sensitivities of 90–94 % for SSCP and 75–81 % for HA. In a landmark analysis, Kourkine et al. achieved 100 % sensitivity by analysing denatured and non-denatured fragments in tandem. Despite being highly effective, the additional sample preparation required by these methods (i.e. separately preparing both single and double stranded DNA and maintaining this strandedness) complicates their implementation on microchips.
In this work, we present an electrophoretic method in which DNA is denatured in a microchip (with formamide) and, depending upon the sequence of applied voltages, can be prepared with a widely varying degree of hybridisation (i.e. from almost entirely ssDNA to almost entirely dsDNA). Given the small volumes involved within the microchip, diffusion time plays a small role and the reassembly process can be fast, with dsDNA obtained within minutes. The rapidity of the manipulation possible on this system allows some investigation of the dynamics of the reassembly, indicating that there are well-defined intermediate states where both ssDNA and dsDNA exist in the reassembly process.
We have applied our methods to the H63D and S65C mutations from the HFE gene associated with hereditary hemochromatosis (HH). The denaturation technique used enables a combined microchip-based method of HA and SSCP analysis.
Simultaneous Analysis of ssDNA and dsDNA
Reassembly of dsDNA
As will be described in the following section, by varying the electrophoretic parameters we can control the relative amount of dsDNA formed – a significantly larger amount could be obtained.
Dynamics of DNA Reassembly
After demonstrating that reassembly occurred within the microchannels after extraction from the formamide-rich sample well, it was of interest to investigate how the sequence and timing of the sample extraction and analysis might affect the degree of rehybridisation. In the work presented thus far we used a 60 s injection (although 20 s would probably have sufficed) as a means of drawing sample directly from the sample well to the intersection, from whence it could be analysed.
The integration onto a microchip of an effective means of mutation detection is perhaps one of the most important technological barriers to the implementation of microchip-based medical diagnostics. The best means of attaining sufficiently high sensitivity is by integrating several existing methods of microchip-based mutation detection. The capillary-based analysis procedure developed by Kourkine et al.  is likely to be highly effective in conjunction with the microchip analysis of prepared samples, but since the procedure is based upon the thermal processing (95°C and snap cooling) of diluted PCR products, the integration of this processing onto the microchip may be problematic. The present method allows for such integrations, thereby enabling the mutation analysis throughputs predicted by Medintz et al.  – throughputs as much as 100 times higher that those presently attainable. Another issue is that of signal to noise ratios – rather than dilute our sample (possibly weakening its signal strength) we can analyse the sample essentially undiluted. Moreover, we can enhance the signal strength, as we choose, for either the ssDNA or the dsDNA.
As demonstrated here, this method also allows on-chip comparisons of one type of DNA with another. A common problem encountered with HA methods is that they cannot distinguish homozygous mutant from homozygous wildtype – the present technique would allow an on-chip comparison of these samples to produce heteroduplexes that will then indicate the mutational status.
The on chip denaturation is produced through the addition of formamide. The melting temperature for this sequence of DNA following the addition of formamide was found to be approximately room temperature, as determined by
Tm = 81.5 + 16.6(log M) + 0.41 (% G + C) - 0.72 (% formamide) (1)
where Tm is the melting temperature in degrees Celsius, M is the monovalent salt molarity, (% G + C) is the percent of the guanine and cytosine in the DNA strand of interest, and (% formamide) is the percentage of formamide added . The melting of DNA was confirmed by forming heteroduplexes on-chip.
The ability to quickly re-hybridise on chip allows for rapid investigation of self-assembly mechanisms. In addition, this re-hybridisation enables the formation of duplexes made from a sample and a set of DNA references – i.e. DNA self-assembly within a microchip could be used to form duplexes that, under electrophoretic analysis, would show the results of comparing the sample DNA with each type of DNA in the reference set. This could avoid the need for DNA sequencing.
The rapidity of our method appears to provide additional information upon short-lived conformations. Although we have added a thermal re-annealing step as part of our PCR protocol, that step does not affect the results of analysis after adding formamide – i.e. by re-annealing on-chip the thermal reannealing is not needed. The thermal re-annealing stage was added to allow the direct comparison of heterozygous samples from the PCR with heterozygous samples after on-chip reassembly. After adding formamide, the electropherogram of the first separation analysis following any long injection shows a clearly defined transient peak. For H63D samples the transient peak is a single peak, whereas for S65C the transient peak is clefted. We have found that the transient peaks vary in size significantly depending upon the electrophoretic and PCR protocols used. Initially we had assumed that this transient peak indicated that the reassembly of the DNA was not 'random' but instead hybridised first in a high-melting point region, and only slowly thereafter. In this model, the presence of the split-peak would provide information upon the location of the mutation. This suggests that mutation S65C is within the higher melting point domain, while the H63D is not. However, as determined by the Meltmap program (generously provided by L. Lerman (MIT)), neither the H63D nor S65C mutations were within the high melting point region of the exon (data not shown).
Several research groups have reported artefacts that arise from ssDNA-primer interactions [14, 17–19]. Kourkine et al. [14, 18] reported that primer-ssDNA complexes can give rise to extra peaks during SSCP. They performed tests with samples of PCR-amplified DNA with and without the removal of the PCR-primers after the amplification step and found that the presence of primers led to the appearance of extra peaks . A reduction in primer concentration during PCR also proved to be effective in minimizing the appearance of these peaks. Kozlowski and Krzyzosiak  have reported similar effects and suggested that the primer-ssDNA complex may have a different mobility simply because of its changed mass, or perhaps due to a change in conformation induced by the binding. In the context of SSCP, they discussed two approaches for dealing with this effect 1) remove it through purification so as to obtain simpler profiles or 2) use the effect to advantage by achieving higher sensitivity in the detection of mutations. Hennessy et al.  performed similar tests and reported that variations in primer concentration are the likely source of irreproducible SSCP profiles. They too suggested that this effect could be used to increase the sensitivity of SSCP.
We therefore suggest that the transient peak is due to the pairing of one product strand with one primer as a result of the renaturation process. The primer-ssDNA complex is primarily ssDNA with a small region of dsDNA at the end(s) of the strand. It is therefore expected to migrate with similar mobility as the ssDNA peaks. The disappearance of the transient peaks with the subsequent short injections may be a result of the complementary single strand binding and displacing the primer. However, the presence of the transient peaks may still provide useful information. The differences in the transient peaks (cleft versus no cleft) between S65C and H63D suggest that their shape may be dependent on the position of the mutation and that the position greatly affects the transport of the transient form of DNA. Thus, the phenomenon of the transient may be a general behaviour that could provide additional mutational information. In corroboration of past work by others [17, 19], it therefore appears that the primer effects do provide mutational information. Moreover, this effect can be produced or avoided depending on whether the desire is to avoid the more complex profiles or to use them to achieve higher sensitivity.
We have developed a method of rapidly disassembling and re-assembling DNA within a microfluidic chip, allowing us control over the relative amount of ss and dsDNA and enabling the performance of rapid hybridisations under electrophoretic control. It has been reported that, when combined, HA and SSCP can provide sensitivities of 100% (e.g. ). In our work to date we have tested a large number of samples, predominantly of HFE, BRCA1 and BRCA2 sequences, and representing approximately several dozen different sequences. All samples containing a mutation have had their mutational status detected by at least one method. We expect then that the sensitivity of the combined methods will be close to 100%. We are now applying this method as part of a study of the application of DNA self-assembly based mutation detection methods (HA and SSCP) to the implementation of highly integrated microchips for performing medical diagnostics.
The present work is also an early step towards directing and studying DNA self-assembly within microfluidic systems. The method applied here could be improved significantly by shortening the injection and separation channels and ultimately may even assist in providing the control needed to direct the assembly of DNA-based nanosystems within microfluidic channels.
Volunteers who had given informed consent donated lymphocytes from which DNA was extracted and purified by using phenol-chloroform-isoamyl alcohol extractions  or the QIAmp DNA Blood kit (QIAGEN, Mississauga, ON). The purified DNA was solubilized in a Tris-EDTA buffer (TE, pH 8.0) and stored at 4°C. All genotypes were confirmed on an ABI Prism 377 Slab Gel Sequencer (Applied Biosystems, Streetsville, ON), using an ABI Prism BigDye Terminator v3.0 Ready Reaction Cycle Sequencing Kit with AmpliTaq DNA Polymerase (Applied Biosystems).
Primers Used for PCR
Amplicon – Primer
HFE Exon 2 – forward
5'-TCA GAG CAG GAC CTT GGT CTT TCC-3'
HFE Exon 2 – reverse
5'-CAT ACC CTT GCT GTG GTT GTG ATT-3'
PCR reagents (polymerases, buffers and primers) were obtained from Invitrogen (Burlington, ON). GeneScan™ polymer was used for microchip electrophoresis and obtained from PE Applied Biosystems (Foster City, CA). A polymer consisting of 5% GeneScan polymer and 10% glycerol (5GS10G), commonly used for SSCP, was made. Tris borate (Fisher Scientific, Fairland, NJ) with EDTA (Merck KGaA, Darmstadt, Germany) was used as the running buffer in concentrations of 1× and 0.1×. Glycerol (Sigma, Saint Louis, MO) is also added to each in 10% and 1% concentrations respectively (1 × TBE10G and 0.1 × TBE1G). De-ionised formamide (minimum 99.5%) was obtained from Sigma (F9037, Saint Louis, MO). The formamide was aliquotted and kept frozen until required.
Microchip Loading and Electrophoresis
The microchip was loaded with 5GS10G polymer without any pre-treatment. The sample well was loaded with 2.6 μL of 0.1 × TBE1G followed by 0.4 μL of DNA sample and mixed. The remaining wells were loaded with 3 μL of 1 × TBE10G. In the case of the Y-chip, 0.4 μL of wildtype DNA was added to 2.6 μL of 0.1 × TBE1G in the first sample well and mixed. The second sample well was filled with 2.6 μL of 0.1 × TBE1G and 0.4 μL of homozygous mutant. The operation of the μTK (injection and separation) was automated through the use of the LabVIEW interface. LIF detection took place 76 mm downstream from the intersection. We have found that the reproducibility of the peak arrival times is within 2 per cent from one run to the next. As such we have not needed to introduce size standards.
The sample DNA was brought from the sample well to the intersection and onto the sample waste well by applying 500 V/cm for 60 s. No initial injection was done with the Y-chip prior to denaturation. During this process the buffer and buffer waste well are left electrically disconnected. In doing so the intersection of the two (three) channels is filled with the sample DNA. This stage is referred to as an injection due to the injection of DNA into the separation channel in the sharply defined volume of the intersection of the channels.
Immediately following injection, the DNA caught within the intersection is separated by applying 714 V/cm for 180 s between the buffer and buffer waste wells. During this step, the sample and sample waste wells are left electrically disconnected. The effective separation distance was 76 mm from the intersection.
After the initial run on the 4-port chip, 1.5 μL of the sample mixture was removed and 1.5 μL of formamide was added and mixed. Following Howley et al. , this is sufficient to denature the DNA with a melting temperature of approximately 25.7°C. Since Fig. 4 clearly shows formation of heteroduplexes, we take this to indicate that the temperature was high enough to allow strands to interchange. Another run was then done with the same parameters as above. In the case of the Y-chip, denaturation of each sample was done immediately following the addition of the samples to the wells. A voltage of 400 V was applied between the sample and sample waste wells during a 60 s injection followed by a separation of 180 s. Subsequent electrophoretic runs followed with 10 s injections at 125 V/cm and 180 s of separation at 714 V/cm for both the 4-port and Y-chip. No additional mixing of the two samples for the Y-chip were required
We gratefully acknowledge the support of the Natural Sciences and Engineering Research Council of Canada.
- Anderson RC, Su X, Bogdan GJ, Fenton J: A miniature integrated device for automated multistep genetic assays. Nucleic Acids Res. 2000, 28: E60-10.1093/nar/28.12.e60.View ArticleGoogle Scholar
- Footz T, Somerville MJ, Tomaszewski R, Elyas B, Backhouse CJ: Integration of Combined Heteroduplex/Restriction Fragment Length Polymorphism Analysis on an Electrophoresis Microchip for the Detection of Hereditary Haemochromatosis. Analyst. 2004, 129: 25-31. 10.1039/b309931h.View ArticleGoogle Scholar
- Footz T, Somerville MJ, Tomaszewski R, Sprysak KA, Backhouse CJ: Heteroduplex-based genotyping with microchip electrophoresis and dHPLC. Genet Test. 2003, 7: 283-293. 10.1089/109065703322783635.View ArticleGoogle Scholar
- Keren K, Berman RS, Buchstab E, Sivan U, Braun E: DNA-templated carbon nanotube field-effect transistor. Science. 2003, 302: 1380-1382. 10.1126/science.1091022.View ArticleGoogle Scholar
- Kozlowski P, Krzyzosiak WJ: Combined SSCP/duplex analysis by capillary electrophoresis for more efficient mutation detection. Nucleic Acids Res. 2001, 29: p. Art. No. e71.-10.1093/nar/29.14.e71.View ArticleGoogle Scholar
- Pappaert K, Vanderhoeven J, Van Hummelen P, Dutta B, Clicq D, Baron GV, Desmet G: Enhancement of DNA micro-array analysis using a shear-driven micro-channel flow system. Journal of Chromatography A. 2003, 1014: 1-9. 10.1016/S0021-9673(03)00715-5.View ArticleGoogle Scholar
- Ali MF, Kirby R, Goodey AP, Rodriguez MD, Ellington AD, Neikirk DP, McDevitt JT: DNA hybridization and discrimination of single-nucleotide mismatches using chip-based microbead arrays. Analytical Chemistry. 2003, 75: 4732-4739. 10.1021/ac034106z.View ArticleGoogle Scholar
- Gross E, Arnold N, Goette J, Schwarz-Boeger U, Kiechle M: A comparison of BRCA1 mutation analysis by direct sequencing, SSCP and DHPLC. Hum Genet. 1999, 105: 72-78. 10.1007/s004390051066.View ArticleGoogle Scholar
- Xiao W, Oefner PJ: Denaturing high-performance liquid chromatography: A review. Hum Mutat. 2001, 17: 439-474.View ArticleGoogle Scholar
- Arnold N, Gross E, Schwarz-Boeger U, Pfisterer J, Jonat W, Kiechle M: A highly sensitive, fast, and economical technique for mutation analysis in hereditary breast and ovarian cancers. Hum Mutat. 1999, 14: 333-339.View ArticleGoogle Scholar
- Tian HJ, Jaquins-Gerstl A, Munro N, Trucco M, Brody LC, Landers JP: Single-strand conformation polymorphism analysis by capillary and microchip electrophoresis: A fast, simple method for detection of common mutations in BRCA1 and BRCA2. Genomics. 2000, 63: 25-34. 10.1006/geno.1999.6067.View ArticleGoogle Scholar
- Tian HJ, Brody LC, Landers JP: Rapid detection of deletion, insertion, and substitution mutations via heteroduplex analysis using capillary- and microchip-based electrophoresis. Genome Res. 2000, 10: 1403-1413. 10.1101/gr.132700.View ArticleGoogle Scholar
- Jin LJ, Ferrance J, Landers JP: Miniaturized electrophoresis: An evolving role in laboratory medicine. Biotechniques. 2001, 31: 1332-1342.Google Scholar
- Kourkine IV, Hestekin CN, Buchholz BA, Barron AE: High-throughput, high-sensitivity genetic mutation detection by tandem single-strand conformation polymorphism/heteroduplex analysis capillary array electrophoresis. Anal Chem. 2002, 74: 2565-2572. 10.1021/ac020025b.View ArticleGoogle Scholar
- Medintz IL, Paegel BM, Blazej RG, Emrich CA, Berti L, Scherer JR, Mathies RA: High-performance genetic analysis using microfabricated capillary array electrophoresis microplates. Electrophoresis. 2001, 22: 3845-3856. 10.1002/1522-2683(200110)22:18<3845::AID-ELPS3845>3.0.CO;2-0.View ArticleGoogle Scholar
- Howley PM, Israel MA, Law MF, Martin MA: Rapid Method for Detecting and Mapping Homology between Heterologous Dnas - Evaluation of Polyomavirus Genomes. J of Biological Chem. 1979, 254: 4876-4883.Google Scholar
- Hennessy LK, Teare J, Ko C: PCR conditions and DNA denaturants affect reproducibility of single-strand conformation polymorphism patterns for BRCA1 mutations. Clinical Chemistry. 1998, 44: 879-882.Google Scholar
- Kourkine IV, Hestekin CN, Magnusdottir SO, Barron AE: Optimized sample preparation for tandem capillary electrophoresis single-stranded conformational polymorphism/heteroduplex analysis. Biotechniques. 2002, 33: 318-325.Google Scholar
- Kozlowski P, Krzyzosiak WJ: Optimum sample medium for single-nucleotide polymorphism and mutation detection by capillary electrophoresis. Electrophoresis. 2004, 25: 990-998. 10.1002/elps.200305782.View ArticleGoogle Scholar
- Sambrook J, Russell DW: Molecular cloning : a laboratory manual.3rd edition. Cold Spring Harbor, N.Y., Cold Spring Harbor Laboratory Press; 2001.Google Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.