1
Miniaturised nucleic acid analysis

2
The application of micro total analysis systems has grown exponentially over the past few years, particularly diversifying in disciplines related to bioassays.

3
The primary focus of this review is to detail recent new approaches to sample preparation, nucleic acid amplification and detection within microfluidic devices or at the microscale level.

4
We also introduce some applications that have as yet to be explored in a miniaturised environment, but should benefit from improvements in analytical efficiency and functionality when transferred to planar-chip formats.

5
The studies described in this review were published in commonly available journals as well as in the proceedings of three major conferences relevant to microfluidics (Micro Total Analysis Systems, Transducers and The Nanotechnology Conference and Trade Show).

6
Although an emphasis has been placed on papers published since 2002, pertinent articles preceding this publication year have also been included.

Introduction

7
The polymerase chain reaction (PCR) has succeeded in revolutionising the analysis of nucleic acids.

8
Disciplines both directly and indirectly related to the life sciences can readily exploit PCR, and with various genome study initiatives increasingly making available sequence data, this trend is set to continue into the future.

9
Most striking has been the synonymous advent of real-time PCR and delivery of a robust quantitative gene measuring tool.

10
However, the post-genome era brings with it new problems and exposes old fissures that have undermined the application of PCR.

11
The scaling-up of pre-existing procedures for higher throughput would be inappropriate for continuous environmental monitoring and population studies.

12
Indeed, when attempted this approach reveals the necessary requirement to integrate gene analysis into a process that serves best to eliminate both sampling and experimental error.

13
Miniaturisation of PCR integrated bioassays potentially possesses all of the desired qualities and comes with additional benefits relating to sample tracking, contamination, and savings in finances and time.

14
Long ignored issues relating to the meaningfulness of data are starting to be questioned.

15
Biological matter is comprised from the unit of life, the cell, and gene analysis is portrayed in terms of gene abundance, presence, or absence, within a particular biological sample.

16
Tissues are heterogeneous, in that many types of different cells co-exist within biological samples.

17
However, most PCR experimental testing to date does not clearly relate the gene content measured to the contribution made by all specific cell types.

18
Increasingly the need to analyse from the single cell is gaining in popularity, and once again the packaging and process offered by integrated miniaturised PCR offers great promise.

19
Indeed, when considered alongside the potential to dissect the PCR process itself to understand more how to increase and maintain PCR efficiency, the miniaturisation of PCR and future PCR applications seem to become permanently entwined.

20
The majority of developments in miniaturisation have stemmed from institutes with strong analytical, engineering, fabrication and optical specialisations.

21
The influx of interdisciplinary interest offers distinct and critical benefits to the application driven life sciences.

22
Therefore, to allow researchers from engineering and life science backgrounds to reflect on the recent progress and to help traverse interdisciplinary boundaries, we have compiled a concise review of papers based around nucleic acid assay miniaturisation and microfluidic PCR, and set these against a framework of biological applications.

23
The reader is also directed to other reviews covering either a wide range of applications,1,2 or specific fields such as DNA and protein analysis,3 genomics and proteomics analysis4 or molecular diagnostics5.

Theoretical aspects

Description of the polymerase chain reaction (PCR)

24
The PCR is a three-step amplification process first introduced by Saiki and co-workers in .19856

25
This in-vitro enzyme-mediated method facilitates the generation of nucleic acid sequences based on the choice of specific primers.

26
During the first step—called denaturation—the hydrogen bonding stabilising the double stranded DNA template is broken to form two complementary single strands (see Fig. 1).

27
In order to provide the energy necessary to break the bonding, this step is commonly performed at temperatures between 94 °C and 96 °C.

28
The temperature is then lowered for the annealing step: primers (synthetic sequences of single stranded DNA—normally 17–25 nucleotides long—which flank the target region to be amplified) specifically bind to the complementary sequences of the DNA template.

29
At this stage, the temperature is primer-dependant and usually varies between 50 °C and 65 °C.

30
Finally the temperature is raised to allow extension: the template is typically replicated by a thermostable DNA polymerase.

31
This temperature is often close to 72 °C, which is related to the temperature optimum of the DNA polymerase in conjunction with retaining hybridisation specificity of primers.

32
The denaturation-annealing-extension cycle is repeated, typically between 25 and 40 times.

33
Although these three steps are performed at three different temperatures, it is not uncommon to combine the annealing and the extension into one operation.

34
Interestingly, at the end of the 1st cycle, the product obtained is defined only at one end, as the replication is a function of time at the extension temperature.

35
It is only at the end of the 2nd cycle that the specific product is first produced.

36
Another interesting point is that two types of replication occur at the same time: semi-specific products such as those from cycle 1 lead to linear replication, whereas truly specific products or amplicons that are defined at both ends lead to exponential replication.

37
Although the linear replication process is negligible in terms of amounts of DNA produced, it is a crucial intermediary step in PCR as it is through linear replication that a specific amplicon is obtained, thus rendering exponential amplification possible.

38
Since 1985, different parameters for the PCR have been extensively studied to optimise the reaction and maximise amplification yield throughout a number of constraints imposed by different bioassays.

39
The extraction of a thermostable enzyme, Taq, from the bacterium Thermus aquaticus was the first breakthrough.7–9

40
Prior to this discovery, the enzyme was heat labile and lost its activity after each denaturation step.

41
Consequently, the polymerase had to be replenished after each cycle which was arduous, but importantly increased the risks of contamination during PCR as the sample vial had to be re-opened, and made optimisation almost impossible to achieve.

42
The introduction of a thermostable enzyme obviated this cumbersome procedure and also enabled the automation of the process.

43
Ensuing Taq polymerase engineering succeeded in enhancing processivity (which is defined for polymerases as the number of nucleotides added per unit to a strand during one sequence of binding and dissociation from the primer template) and fidelity (capability of a polymerase to faithfully replicate a DNA strand without expansion or deletion in the number of base pairs, nor substitution of the bases (AT ↔ CG)).10

44
In this study, a recombinant Taq polymerase-thioredoxin displayed an increase in processivity when compared to the wild-type Taq polymerase (from 50–80 nt to >300 nt).

45
Other enzymes such as Pfu,11 KlenTaq,12Vent13 or Tfi14 have been applied to PCR.

46
Significantly, some enzymes are nowadays modified to block their activity until they are heat activated.

47
This is for example the case of PfuUltra Hotstart DNA polymerase, which is commercialised by Stratagene.

48
In this instance, heat labile antibodies neutralise the enzyme activity until they are denaturated during the initial PCR denaturation step.

49
Such a procedure limits the formation of non-specific products, as it impairs the enzyme activity until thermocycling temperatures are reached.

50
Priming methods comparing poly(A) tails and random hexamers for cDNA synthesis were studied by Resuehr and co-workers,15 while Halford et al. studied the effect of primer-dimer formation over the PCR yield.16

51
Protocols to optimise PCR reactions are nowadays readily available.17,18

52
PCR has also been studied from a more theoretical approach and statistical models describing the reaction are often based on the branching theory.19

53
Based on this theory, the number of molecules generated by a PCR reaction starting with a single molecule after n cycles is: (1 + p)n, where p is the probability for a molecule to duplicate.

54
Such an equation leads to two consequences.

55
Firstly, the value p is positively linked to the PCR yield.

56
Secondly, for large values of p, the distribution of the DNA population becomes multinodal, presenting multiple probability peaks.

57
This is due to the amplification during the initial cycle becoming critical and a failure for the original molecule to replicate having a lasting impact.

58
It is consequently essential to optimise the PCR parameters in order to obtain the most favourable conditions and enhance the PCR yield.

Strand-displacement based amplification

59
Although PCR is the most popular DNA amplification technique, other methods also exist20 and are potentially suitable for microfluidic-based devices.

60
For example, Single Strand Displacement Amplification (SSDA) is an isothermal technique first introduced by Walker et al. in .199221

61
Comparably to PCR, this method allows exponential amplification of double-stranded DNA.

62
The Multiple Strand Displacement Amplification (MSDA) technique is similar to SSDA, with the only difference being that during extension of the complementary strand, new annealing sites are generated for the first primer.

63
Although MSDA can be applied with specific primers, it has also been used with random hexamers, leading to the amplification of entire genomes.22

64
Such an amplification process produces tree-like molecules leading to amplified molecules of different lengths.

65
Indeed the generated strands are not cleaved from the strand molecule from which the amplification originated.

66
A few products using this technique are commercially available, such as GenomiPhi DNA amplification offered by Amersham Biosciences, and REPLI-g from Molecular Staging.

67
Notomi et al. developed another method for DNA amplification: Loop-mediated isothermal AMPlification (LAMP).23

68
This technique is based on MSDA but instead of using a set of two specific primers, or random hexamers, it involves 3 sets of specific primers.

Other DNA amplification techniques

69
Lizardi and co-workers used circularisable oligonucleotide priming to investigate amplification of circular DNA templates.24

70
This method, called Rolling-Circle Amplification (RCA), produces a single-stranded product initialised by only one primer.

71
Although this technique might seem limited as it only applies to circular templates with an appropriate promoter, a wider variety of applications are technically possible as circular templates include viruses and plasmids.

72
RCA has been coupled to other techniques such as MSDA,24 or RCA itself (Circle-to-Circle Amplification).25

73
The Ligase Chain reaction was initially introduced by Landefren et al. in 1988 and is based on the ability of two oligonucleotides to hybridise next to each other given an appropriate complementary DNA target molecule.26

74
Although this method was introduced as a single-base-modification detection method, it has since been used for the sole purpose of DNA replication.27

75
The merging of PCR with gene micro-array applications has been demonstrated by Solid Phase DNA Amplification (SPA).28,29

76
In this method, the primers are surface-bound and thus do not freely diffuse in solution.

77
A Monte Carlo Lattice model was proposed by Mercier and co-workers to discribe solid phase amplification.19

78
As the amount of DNA generated is directly dependent on the number of primers immobilised at the surface, SPA would highly benefit from on-chip applications where area-to-volume ratios are greatly enhanced.

79
The application of PCR to define specific sequences for subsequent use as micro-array probes has also recently been described by Greiner et al30.

Substrates used and surface treatments to minimise bio-molecule adsorption

Substrates

80
Although microfluidic devices were originally structured in silicon or glass, polymers such as poly(dimethylsiloxane) (PDMS), Poly(methylmethacrylate) (PMMA) and Polycarbonate (PC) have increasingly been used as alternative substrate materials.

81
The search for new substrate materials is still a burgeoning area, and many recent microfluidic devices for PCR have been fabricated using SU-8, PEEK, PTFE, Cyclic-Olefin-Copolymer (COC) or Zeonor (polymer commonly used for compact discs and DVDs).

82
For a more detailed overview of some of the materials used, the reader should refer to references in Table 1.

Surface treatment

83
The efficiency of biological reactions is often limited by interactions between the microchannel surfaces and the bio-molecules present in solution.

84
This issue is even more exacerbated by the wide variety of possible substrate material (see Table 1) in conjunction with the vast range of biological matter used as analyte; consequently various surface treatments have been investigated.

85
One of the most widely used processes to prevent on-chip adsorption of molecules consists of silanisation, a method first introduced by Hjertén in .198558

86
For example, El-Ali et al. performed a gas phase silanisation of a SU-8 chamber surface in order to enhance its PCR compatibility.44

87
Similarly, Shin and co-workers studied different silane compositions to decrease protein adsorption at glass surfaces.59

88
Noteworthy is that even though silanisation has been successfully applied to microfluidic devices, criticisms regarding the reproducibility of such coatings have been argued.60

89
To circumvent this issue, several methods have been investigated, and for instance, different strategies were implemented to prevent protein adsorption in PDMS microfluidic devices.

90
Choi et al. favoured a mixture of poly(ethylene glycol) and poly(acrylic),61 while Xiao and co-workers used a polyacrylamide gel.49

91
Alternatively, Shin and co-workers coated PDMS micro-channels with parylene in order to prevent adsorption on the chip surface.62

92
An additional approach was introduced when Kirby et al. studied the effect of polarity and surface charge on cell adsorption by controllably modifying the zeta potential at the surface of their device39.

Sample preparation

93
The following section addresses recent studies in the areas of cell culturing, cell identification and cell trapping, cell lysis and DNA extraction within microfluidic formats.

Cell culture

94
Much attention has focussed on the transfer of cell culturing techniques to planar chip formats.

95
For example, Grodrian et al. applied the principle of segmented flow on chip for parallel cell culture by separating culture media plugs with immiscible fluid.63

96
Alternatively, Tan and co-workers recreated a tissue-like structure in a micro-fluidic environment using cell-matrix assemblies and biopolymers,64 while Schaak et al. have performed cell culture in droplets.65

97
To assess the viability of the cultured cells Goto and co-workers implemented a protocol to detect nitric oxide released by macrophages66.

Cell identification and cell trapping

98
Biological samples are most often present as a complex matrix.

99
It is therefore of paramount importance to be able to differentiate cells or targets of interest from other biological material.

100
Different approaches have been implemented on chip, where for example, Lin et al. introduced an in-vitro impedance measurement of cells to distinguish between abnormal cells and normal cells under different frequency domains.67

101
Alternatively, Emmelkamp and co-workers chose an optical detection technique for sorting cells based on their intrinsic fluorescence signal within a microfluidic structure.36

102
After having detected the cell of interest, the next step is usually some form of isolation and concentration.

103
To this extent, different techniques have been implemented on chip, such as dielectrophoresis (DEP).

104
Fuchs and co-workers have used DEP to trap individual particles among a few thousand cells suspended in a liquid on a microelectronic chip.68

105
Similarly, during a study of blood components and their inhibitory effects on PCR, Wolff and co-workers captured cells on electrodes by DEP while inhibitors were eluted by washing.69

106
An alternative to conventional DEP has been proposed by Lapizco-Encinas et al. and involved electrodeless DEP for cell trapping.70

107
Capillary electrophoresis was also used by Park et al. to demonstrate the capture and subsequent release of DNA fragments within PEEK tubing.50

108
Although DEP and electrophoresis are widely used techniques for cell trapping, other methods have also been investigated.

109
For example, Denoual et al. presented a microfluidic device in which cells could be directed to cell traps by the fluid flow.71

110
Lettieri and co-workers have also introduced a microfluidic device for performing controlled micro re-circulating flow capable of DNA trapping.72

111
Two flows (an electroosmotic flow and a pressure-driven flow) were opposed and molecules were trapped by adjusting the different velocities.

112
From another perspective Goubault et al. captured rare cells using magnetic filaments and adjusted the specific/non-specific discrimination ratio by optimising flow rates,73 while Lu and co-workers presented a rapid organelle separation by free flow isoelectric focusing.74

113
Electrophoresis has also been used to increase sample concentration, where Halle et al. introduced a micro-fluidic free-flow zone electrophoresis device to concentrate biological particles in buffer75.

Cell lysis and DNA extraction/manipulation

114
In order to extract the different nucleic acid analytes from within a cell, controlled lysis is usually performed.

115
Although regularly achieved by heating the cell,54,76,77 other lysis methods have been developed.

116
For example, strong ultrasonic fields were applied by Tachibana et al.,78 whereas Marmottant and co-workers presented an experiment during which gentle bubble oscillations were successfully applied to rupture lipid membranes.79

117
Lu and co-workers also presented a miniaturised device for controlled cell lysis in order to release sub-cellular materials.80

118
Chemical lysis was performed by both Schilling et al.,81 and Hong and co-workers.82

119
Once cell lysis has been performed, the issue of DNA extraction can be addressed.

120
For example, Reddy and co-workers presented an organic-aqueous liquid extraction on a micro-device in order to purify DNA obtained directly from cells.83

121
On the other hand, Bienvenue et al. implemented a sol-gel technique to extract DNA.34

122
Other groups have favoured more targeted approaches to isolate DNA from its medium.

123
For example, Washizu and co-workers used electrostatic fields for direct DNA manipulations84 and Mohanty et al. applied dielectrophoresis and electrical impedance spectroscopy for single cell manipulation within microfluidic systems.85

124
Finally, Sundararjan and co-workers demonstrated immobilisation and programmable release of single DNA molecules by electrical heating86.

Microfluidic DNA amplification

125
Since its introduction PCR sustains a tremendous impact in several fields, including clinical diagnostics, medical sciences and forensics.

126
However this technique presents several drawbacks.

127
For example, the costs associated with each reaction are still quite high (mostly due to the price of the enzyme), even if current parallel processing has attenuated costs.

128
It is also well documented that PCR is highly prone to contamination due to its efficiency and high sensitivity.87,88

129
Additionally, a common advantage of miniaturised systems over macro apparatus is their reduced thermal mass: as they present less inertia to temperature changes ramping rates for heating or cooling are drastically increased, and this gives rise to more rapid gene amplification, and may present benefit for some of the conformation fluorescence resonance energy transfer (FRET) probes used in real-time PCR.

130
Whilst several of the more successful and reproducible applications of PCR are manifested in qualitative analyses, increasingly the requirements of PCR testing are placing a burden on elucidating a robust means to accurate and precise gene quantitation.89

131
For example, the definition of therapeutically significant disease thresholds could be used to more effectively treat infections or other diseases.

132
Therefore the cumbersome processes that endure sample losses and sample degradation, ultimately compromise gene quantitation can be circumvented by employing closed μTAS.

Well-based amplification

133
A microfluidic approach to DNA amplification would present several advantages.

134
Not only could system features easily be integrated to ease complex sample handling and tracking, but also possible risks of contamination would be reduced.

135
Indeed several operations (such as sample preparation, PCR and DNA detection) could ultimately be integrated into one device.

136
By rendering the system quasi-autonomous, interactions with the sample would be minimised and the number of possible sources of contamination would be greatly lessened.

137
These attractive features explain why the miniaturisation of the PCR apparatus has been of a growing interest in the scientific community.

138
The first on-chip PCR devices were presented by Northrup and co-workers in 199390 and Wilding et al. in .199491

139
Both devices consisted of a silicon chip with a micro-well in which the sample was loaded (see Fig. 2).

140
The entire chip was heated and cooled to provide the adequate thermocycling conditions.

141
Since its introduction, this model has been replicated by numerous groups, sometimes with major amendments.

142
These include integration of electrophoretic product sizing on a monolithic device,92 as well as optimisation of conditions to provide faster PCR (under 7 minutes),93 single-molecule detection,94 real-time applications95 and multiplexing.96,97

143
The use of micro-arrays96,98,99 and drop-metering100 provided these micro-devices with high throughput (more than 10,000 parallel reactions have been reported101–104) and better fluid-handling capabilities, whereas thermocycling conditions, such as temperature uniformity105 and heat transfer optimisation,106 were closely studied.

144
A high versatility in terms of thermocycling conditions was shown by Kajiyama and co-workers when they presented independently-addressable wells.107

145
In addition, a variety of applications were successfully demonstrated, such as ligase chain reaction108 and isothermal nucleic acids amplification (NASBA).109

146
There have only been few devices based on a well approach that differed significantly from the original design presented by Northrup.

147
For example, Oda et al. addressed the invasive characteristics of heating elements conventionally used by replacing the heaters with a transparent contact-free heating technique based on infra-red wavelengths.110

148
This device has since been developed further.35,111

149
Pal and co-workers performed PCR using induction heating, another non-contact heating method.112

150
In this paper the chips were mounted onto a ferrous ring, while the inductor was made of a copper wired wound around a ferrite core.

151
By alternating the current going through the inductor, magnetic fields were produced.

152
They then induced Eddy currents in the ferrous ring, which generated localised heat without any physical contact.

153
In 2003, Krishnan et al. presented an original device relying on convection to provide thermocycling conditions.113

154
Two heaters sandwiched the Rayleigh-Bénard cell and were maintained at constant temperatures, namely 97 °C at the bottom and 61 °C at the top.

155
Due to the difference in temperature, a convection flow developed in the cell, with an upward flux in the centre and downward fluxes on the sides.

156
As the sample moved up and down in the cell, PCR conditions were provided and DNA was amplified.

157
This device was later modified by Wheeler and co-workers.114

158
It is to be noted that these devices are fairly flexible in terms of design: the well geometry can be tailored to a specific application, or arrays of wells are easily manufactured.

159
However, they still do not take full advantage of lithographic techniques.

160
For example, it would be difficult to implement efficient mixing structures.

161
In addition, because the temperature of the entire chip needs to be altered for thermocycling, they display unnecessary inertia.

Continuous flow-through based amplification

162
A dynamic, moving-sample system particularly addresses previously described issues relating to full process integration.

163
In the case of continuous-flow chips for example, transition times to change temperature depends only on the sample pumping rate and the time the sample needs to reach temperature equilibrium.

164
Also heat inertia is reduced to a minimum because only the sample thermal mass needs to be taken into consideration rather than the chip plus sample as in the case of microwell PCR, with a direct consequence of enhanced rates of heating and cooling.

165
One approach to continuous flow-through based amplification was introduced in 1994 by Nakano and co-workers when they used capillary-based systems.

166
Their concept consisted of a capillary going through three different oil baths in a loop.115

167
The length of capillary in each bath determined the duration of the annealing, extension and denaturation steps.

168
The sample was continuously pumped through the capillary unidirectionally and DNA amplification occurred as the sample was going through each temperature cycle.

169
Based on a comparable concept, Curcio and co-workers performed a high-throughput DNA analysis system in a Teflon tube by separating different sample plugs with immiscible organic plugs.116

170
A similar experiment was performed in a capillary by Park et al.117

171
Friedman and co-workers proposed a static sample approach to nucleic acid amplification in .1998118

172
They enclosed the PCR sample inside a capillary, the ends of which were blocked by rubber pads.

173
Each capillary was coated with a transparent Indium Tin Oxide (ITO) layer, ensuring that thermal cycling conditions could be independently controlled.

174
Finally Chiou et al. presented a bi-directional sample pumping system.119,120

175
In this device, the capillary was rested on three heating blocks and the sample was pumped back and forth to provide thermal cycling.

176
An optical system stopped the pumping mechanism when the sample reached the desired temperature zone.

177
The first continuous-flow PCR chip presented in 1998 by Kopp and co-workers provided the basis for subsequent developments.121

178
It consisted of a serpentine channel that passed through three heating zones maintained at constant temperature by copper blocks (see Fig. 3).

179
Remarkably, specifically amplified products were detected after 20 cycles in 1.5 minutes.

180
Device improvements have included optimisation of the PCR conditions to prevent any cross-contamination during injection series,122 modification of the design to combine RNA Reverse-Transcription and PCR123 as well as a sensitivity study.124

181
Applications such as high throughput and single molecule amplification125 and infectious biological agent monitoring126 have also been reported.

182
Different approaches have been used to simulate the temperature distribution in continuous-flow PCR chips, such as finite element analysis,127 CFD-ACE+ (computational fluid dynamics and multi-physics software)128,129 and Lattice Boltzmann simulations.130

183
Although the continuous-flow approach allows for fast heating and cooling rates (thus improving product specificity) parallelisation is not easily applicable as it would significantly complicate the design and most likely increase the footprint of the PCR chip.

184
In addition, this type of device is restrictive in terms of number of cycles: the number of meanders is decided during the design procedure and cannot be modified to increase the number of cycles once the chip is manufactured.

185
However, Liu et al. circumvented this drawback by manufacturing a rotary device,131 where a12nL-sample was introduced in the chip and subsequently pumped at a rate to enable 2–3 revolutions per minute.

186
Integrated heaters provided the temperature zones necessary for thermocycling.

187
Although the number of cycles could easily be adjusted (by changing the number of revolutions) the manufacturing process required additional and complex steps.

Novel approaches to DNA amplification

188
Although the majority of microfluidic devices for PCR utilise batch or continuous-flow strategies, new approaches for DNA amplification have started to emerge.

189
For example, an on-chip sample-shunting technique has been introduced by Auroux et al.43,132

190
This approach, based on shunting the sample back and forth over heating zones in a straight channel by applying alternate pressure, combines the cycling flexibility of the well-chips with the quick temperature transition of the continuous-flow PCR micro-structure (see Fig. 4).

191
A theoretical evaluation of a system based on this concept was provided by Bu and co-workers.133

192
A sample-shunting device using ferrofluidics for actuators has also been introduced by Hardt et al.134

193
From another perspective, Pollack and co-workers used an electrowetting process to control droplets containing PCR reagents and demonstrated that such fluid handling did not inhibit DNA amplification.135

194
A few approaches have also been developed on macro-scale apparatus and could also be of interest for microfluidic PCR.

195
For example, high-speed PCR could be achieved on chip by using a microwave-assisted technique first demonstrated by Fermér et al.136

196
DNA amplification in a water-in-oil emulsion, presented by Nakano et al.,137 could be pushed to further limits thanks to the better fluid handling provided by micro-devices.

197
In addition, since cells can be manipulated in a microfluidic environment (see section ‘Sample Preparation’), another application could be in situ (meaning inside a cell) DNA amplification.

198
This technique, first introduced in 1990 by Haase et al.,138 has since been optimised139 and has been combined to reverse transcription140 and to LAMP amplification,141 offering enhanced flexibility in terms of possible applications.

Inhibitors and enhancers

199
One of the main concerns when performing on-chip PCR is the loss in efficiency.

200
Although this problem can be sometimes traced to inhibitory materials used during micro-device fabrication (e.g., residual chromium from chrome masks in glass chip fabrication142 or thermocouples110), the most predominant problem is the elevated adsorption of the PCR components onto the chip surface due to the increase of the surface-to-volume ratio in a micrometer-scale environment.

201
For example, Erill and co-workers pinpointed that the component the most prone to adsorption was the Taq polymerase enzyme.143

202
Although surface coating strategies such as silanisation are possible solutions (see section ‘Surface treatment’), their effectiveness is sometimes questionable due to a lack of reproducibility.60

203
A second approach, called dynamic passivation, includes passivating reagents directly in the PCR mixture.

204
The most common additives are polymers such as polyethylene glycol (PEG)144 and polyvinylpyrrolidone (PVP)144 or proteins such as bovine serum albumin (BSA).143

205
Interestingly, Robinson et al. studied the influence of the molecular weight of PVP over the coating efficiency, as well as the effect of analytes in solution over BSA adsorption.145

206
Additional dynamic coatings have also been studied by Giordano et al. and include hydroxyethylcellulose (HEC), epoxy (poly)dimethylacrylamide (EPDMA), and diaminobutane (DAB).146

207
Cui and co-workers investigated the effect of single-walled carbon nanotubes on the PCR yield.147

208
Results showed that the nanotubes increased the amount of DNA amplified when they were introduced in concentrations below 3 µg/µL.

209
Finally, Yoon and co-workers examined the influence of pH, potential and electrode polarity in an electrowetting–based microdevice in order to prevent bio-molecule adsorption148.

DNA detection methods

Sequence independent detection methods

210
Due to cost-effectiveness, the use of intercalators (molecules that upon binding to double-stranded DNA exhibit significant enhancements in their fluorescence quantum efficiencies) in combination with gel electrophoresis is probably the most widely used method in biological laboratories for the detection of post-PCR products.

211
With this technique, the DNA molecules are effectively labelled with an intercalator dye and subsequently separated according to their sizes.

212
Popular intercalating dyes include ethidium bromide and SYBR® Green, but other intercalators have also been studied such as BEBO,149 YOYO150 and TO-PRO-.3151

213
The use of intercalators to detect DNA molecules presents two main advantages: it enables real-time detection152,153 and it is highly versatile, as the dyes have been selected to bind selectively to dsDNA.

214
However, indiscriminate binding is also a major drawback: both specific and non-specific products will generate the same type of signal, making it difficult to differentiate between them.

215
Integrating a size separation method or denaturing gradient profile often assists the identification of authentic amplicons.

216
One of the most widely used on-chip detection techniques for DNA molecules is capillary electrophoresis (CE).

217
CE was first demonstrated on chip by Manz et al154. and has since then been applied in many fields,155 including the analysis of PCR products.156

218
Recent applications include the detection of short tandem repeats by Mitnik and co-workers,157 while Tseng et al. studied the effect of bubble cells on the resolution and the sensitivity of DNA separation during CE.158

219
Efforts to develop matrices for on-chip CE with lower viscosity comprised a novel sieving buffer based on hydroxypropylmethylcellulose presented by Han et al.,159 as well as a study of a series of poly(ethylene oxide) matrices by Xu and co-workers.160

220
A high-throughput device capable of genotyping simultaneously 384 samples in less than 325s was presented by Emrich et al.161

221
Other sequence-independent detection methods have also been investigated.

222
For example, Messina et al. introduced a high-frequency sensor that enabled the differentiation between pre-amplification and post-PCR products.162

223
Zangmeister and co-workers developed a novel fluorescence detection method based on the displacement of a fluorescent-tagged oligomer by the untagged target molecule163.

Sequence-specific detection methods

Real-time probes

224
It is the authors' belief that, although real-time probes are not yet widely used for on-chip detection due to complex detection system requirements, they will soon be routinely implemented for on-chip quantitative and qualitative real-time PCR (RQ-PCR).

225
In this section the reader will find a brief overview of some of the commercially available real-time probes.

Hydrolysis probes

TaqMan® probes

226
The selectivity of real-time PCR was greatly improved with the introduction of fluorogenic probes (oligonucleotides whose sequences are complementary to part of the target molecules),164 such as the TaqMan® or hydrolysis probes.

227
The detection principle is based on the quenching phenomenon known as Fluorescence Resonance Energy Transfer (FRET).

228
FRET corresponds to the transfer of energy from a fluorescent reporter molecule (R) to a quencher molecule (Q), leading to light emission by Q at a longer wavelength than the donor.

229
After hybridisation (and during annealing) the probe is cleaved due to the 5′ nuclease activity of the enzyme.

230
As FRET is impaired, the reporter's emission can be detected.

231
RQ-PCR optimisation using TaqMan® probes has been performed and include the study of the best possible combinations between commonly used fluorophores and quenchers by Marras et al.,165 as well as quantitative co-amplification of multiple targets (multiplexing) by Wang and co-workers.166

232
Based on a similar approach, Gilliland et al. introduced PCR quantitation by competitive PCR167.

Cycling probe technology

233
Duck et al. developed in 1990 the Cycling Probe Technology (CPT).168

234
This detection method is based on a chimeric DNA-RNA-DNA probe labelled at one end with a reporter and at the other end with a quencher.169

235
Upon hybridisation with the target sequence, the enzyme RNaseH specifically cleaves the RNA region of the probe, resulting in fluorescence emission.

236
This method has been implemented on chip with success in Professor Harrisson's group170–172.

DNAzyme

237
A detection method similar to the CPB and also based on a hybridisation probe was recently released by BD Biosciences: the BD Qzyme Assay.173

238
The innovative component is the DNAzyme, a catalytically active oligonucleotide capable of cleaving nucleic acid substrates at specific locations.

239
The DNAzyme operates in coordination with a DNAzyme-specific fluorogenic target that is cleaved during the PCR process, and consequently the fluorescent signal enables the real-time monitoring of DNA amplification.

240
Since the introduction of the DNAzyme probes, several applications have been reported in the literature such as quantitation of nucleic acid sequences in real-time,174 study of the probe fluorescence signal175 and elaboration of a circular probe176.

MGB technology

241
A new probe approach based on Minor Groove Binders (MGB) has recently been introduced by Epoch Biosciences.

242
MGB are flat molecules that can fit in the minor groove and interact with the phosphate-sugar backbone either by hydrogen bonds or by hydrophobic interactions.

243
Such characteristics were proved to stabilise DNA hybridisation, resulting in shorter primers retaining higher melting temperatures177 and increased probe specificity.178

244
Similar MGB technology is now used both by Applied Biosystems for their TaqMan® chemistry and by Epoch Biosciences for their MGB Eclipse Probe Systems.

245
MGBs are used in conjunction with fluorogenic probe technologies and are gaining in popularity with the increasing requirement to place probes over exact regions of target sequences, such as exon-exon boundaries used in splicosomic studies.

Conformation probes

Molecular Beacons

246
Operating on a comparable FRET principle as TaqMan® probes, Molecular Beacons were introduced by Tyagi and co-workers.179,165,180

247
They consist of single-stranded oligonucleotides arranged in a stem-loop structure, with a reporter probe attached to one of the stem arms and a quencher to the other.

248
The loop portion is complementary to part of the target sequence, whereas the stem portion flanks the loop and are self complementary and unable to hybridise to any of the target sequence.

249
To simplify, Molecular Meacons exist in two different states.

250
When they are in a loop-stem structure, the fluorescence signal is quenched by FRET or proximity effects.

251
On the contrary, during the annealing step, the loop part hybridises to the target molecule to separate the reporter and quencher dyes and a signal can be detected.

252
Molecular Beacons have been used in a microfluidic environment to detect the breast cancer gene BRCA1 by Culha and co-workers,181 and also as a mean to study DNA hybridisation kinetics and to perform DNA melting temperature analysis by Dodge et al.182

253
Based on a similar concept, Nazarenko and co-workers introduced self-quenched primers.183

254
These single-fluorophore labelled primers emit fluorescence with different intensities depending on their conformation: a weak emission is noticed when the primers are in a stem-loop configuration, whereas the fluorescence signal is at its maximum when the primers are incorporated in a double-stranded DNA helix.

Scorpions

255
Scorpion primers are an alternative to Molecular Beacons.184

256
A Scorpion primer consists of a three-part molecule based on the following model: (PCR primer) – (blocker molecule) – (probe).

257
During the PCR process, the primer hybridises to the target molecule and is then extended.

258
Upon denaturation, the extended-DNA-strand/target hybrid is dissociated and so are the two arms of the Scorpion.

259
Once the temperature reduced for the annealing/extension phase, the probe-part of the Scorpion anneals to the extended strand, the quenching process is impaired and fluorescence is detected.

260
Upon extension the probe is displaced and recovers its quenched conformation with a resultant drop in the fluorescence signal.

Hybridisation probes

LC probes

261
Roche Molecular Diagnostics have also developed a mutation analysis kit based on hybridisation probes.

262
As with ligase chain detection, the probes are specifically designed to target the amplicon at adjacent sites.

263
However, one probe is labeled at the 3′ end with fluorescein (the “donor” probe), whereas the two other probes are labeled with an acceptor dye on the 5′ end (either LC-Red 640 or LC-Red 705, each dye corresponding to a mutation possibility).

264
The detection principle is based on FRET: the single-base-mutation is detected depending on the wavelength at which the signal is detected, as LC-Red 640 and LC-Red 705 emit at two different wavelengths.

AEGIS platform

265
The latest developments in DNA diagnostic technologies include the expansion of the DNA alphabet.

266
EraGen Biosciences, Inc. developed the AEGIS (An Extended Genetic Information System) platform that extends the DNA alphabet by up to 8 new bases.

267
Two of the new bases, isoC and isoG, have already been combined to the natural DNA bases to produce the company's diagnostic technology.

268
The GENE-CODE system is a real-time nucleic acid detection system based on the reduction in fluorescent signal during PCR.

269
During the amplification process, a site-specific quencher is inserted opposite a complementary AEGIS base in one of the primers, resulting in signal diminution by FRET phenomenon.

270
This technology is still fairly new but clinical diagnostics assays based on the AEGIS bases isoC and isoG are already commercially available (such as the VERSANT HIV Assay developed by Bayer Diagnostics).

271
Although real-time PCR methods using hydrolysis probes, conformation probes or hybridisation probes are powerful techniques, they are inherently limited in some cases, as pointed out by Gardner and co-workers regarding TaqMan® probes.185

272
These probes are indeed highly specific, rendering the detection of viruses with substantial genetic variation among strains, such as the HIV virus, impractical.

RQ-PCR data analysis strategies

273
At the same time as the above techniques enabled quantitative PCR, the issue relating to data analysis became a paramount concern.

274
Not only were systematic errors in fluorescence measurements investigated,186 but different methods of analysis were also introduced.

275
The most widely used procedure of analysis is the threshold method.187

276
Based on the background fluorescence signal, a certain level of fluorescence signal (usually around 10 times the value of the average background fluorescence value during early cycles) is imposed as the threshold limit.

277
After plotting the fluorescence signal obtained during PCR against the number of cycles performed, the intersection of this curve with the threshold limit can be determined.

278
It is often called the threshold value, or CT value, and represents the number of cycles necessary to generate an exact amount of DNA.

279
With this method, CT values can be compared and DNA template concentrations can be calculated.

280
Although this method is well accepted among the scientific community, new approaches have been proposed.

281
For example, Rutledge and co-workers studied the mathematics of quantitative PCR and offered an alternative analysis tool based on DNA mass at threshold limit,188 whereas Liu et al. favoured the kinetic curves to extrapolate the initial amount of gene transcript.189

282
New algorithms have also been developed to promote automatic analysis of real-time PCR data190.

Single Nucleotide Polymorphism (SNP)

283
Whilst quantitative PCR assays present a significant challenge for microfluidic applications, the development of qualitative assays such as the analysis of genetic mutations has attracted much attention.

284
Single Nucleotide Polymorphisms (SNPs) are frequent non-random base substitutions, present at 1 per 1000 base pairs within the genome, and are avidly studied for association with diseases.

285
Several techniques to detect SNPs have been implemented on chip.

286
For example, Kimbal and co-workers applied Temperature Gradient Gel Electrophoresis (TGGE) in a polycarbonate micro-device.191

287
Buch et al. combined standard gel electrophoresis to TGGE in their 2-D separation device.192

288
In the first dimension, molecules were separated by gel electrophoresis according to their sizes, while TGGE allowed a SNP analysis in subsequent perpendicular channels.

289
Using hydroxy-ethylcellulose as a sieving matrix, Liu and co-workers used TGCE and successfully separated a mixture of wild type DNA, mutant type DNA and the corresponding heteroduplexes.193

290
Ito et al. favoured Affinity Capillary Electrophoresis (ACE) to develop a gene mutation assay.194

291
Zhong and co-workers performed SNP genotyping by applying a combination of allele-specific oligonucleotides attached to the chip surface, detector probes and a thermostable DNA ligase.195

292
Another approach was presented by Russom et al.: they analysed SNPs by extending allele-specific fluorescently labelled nucleotides196.

Hybridisation based detection

293
DNA hybridisation is widely exploited to achieve high selectivity and sensitivity.

294
Different approaches have been applied on chip.

295
For example, Lai and co-workers developed an enzyme-linked immunosorbent assay on a compact disk platform,56 while Cai et al. determined by electrochemical impedance spectroscopy that DNA hybridisation is strongly frequency-dependent.197

296
Trau and co-workers applied PCR reactors combined to microarrays to genotype Chinese medicine plants.198

297
Magnetic beads were also used in microfluidics devices to immobilise by hybridisation alternatively the target template,54 the capture probe42 or various sensing agents.48

298
A different approach was introduced by McKendry et al.:199 using cantilever assays they were able to determine the sequences of unlabeled DNA molecules present at nanomolar concentrations.

Integrated systems

299
One of the main challenges in miniaturisation is the integration of functional components to perform several operations without the need for macro apparatus or manual user input.

300
A few groups have started to tackle this issue and a brief overview of the integrated systems available at the moment is given in this section.

301
A number of instances of integrated on-capillary systems have been recently presented in the literature.

302
For example, Li and co-workers developed a system that could perform cell counting, cell-lysis, PCR, on-column hybridisation and product detection.200

303
In addition, an integrated on-capillary arrangement capable of cell-lysis, reverse transcription PCR and product analysis was reported by Matsunaga et al.,77 while a set-up capable of PCR, sequencing and purification in capillaries was introduced by Hashimoto et al.201

304
Several examples of on-chip integrated systems have also been published in the literature.

305
One of the devices combining the most operations was presented by Liu et al.54,76

306
Their device enabled the following steps: cell capture, cell concentration, cell lysis, purification, PCR, SNP assaying and hybridisation.

307
In addition, valves, mixers and pumps were incorporated onto the chip.

308
Another system offering various capabilities—namely cell capture, cell lysis, purification and DNA amplification—was introduced by El-Ali and co-workers.44

309
Other structures have also been investigated, and for a more detailed overview of some of the most recent, the reader should refer to Table 2.

310
As can be seen in this table, the pre-PCR steps that are integrated the most include cell capture,44,54,76,82,202–205 cell lysis44,54,76,82,202–204,206,207 and sample purification.35,44,54,76,82,203,205,208

311
The well-approach PCR41,44,54,76,208–217 seems to be favored for integrated systems when compared to the continuous-flow process.35,123,125,206,218–220

312
This might be due to an increased simplicity in sample handling for the former method.

313
The physical elements the most integrated are heaters41,44,125,205,206,209–214,216–219 and valves.41,54,76,82,202,206,209,210,212,216–218

314
DNA amplification is mostly performed to increase the detection sensitivity either of a specific molecule among a heterogeneous population or to characterise and/or measure a given section of a target.

315
However, even though the merits of PCR are widely recognised, the process of DNA amplification impairs the speed, the cost-effectiveness and even sometimes the accuracy and the sensitivity of such detection methods.

316
To resolve this issue, U.S.

317
Genomics has developed a unique method with which single molecules from an un-amplified sample can be detected using the GeneEngine.

318
This system relies on isolating the target from the sample, fluorescently tagging it and finally injecting it into the test platform where each molecule is subsequently detected by laser illumination.

319
Using DirectMolecular Analysis, quantitation, SNP analysis and molecule interaction quantification can be performed by colour coincidence counting.

320
Analysis in less than one minute and femtomolar range sensitivity are claimed to be achieved.

321
Alternatively, DirectLinear Analysis enables the sequencing of single linear genomic DNA molecules.

322
Sequence-specific probes are first hybridised to the DNA molecule.

323
The strand is then stretched and the probes are detected linearly by the GeneEngine.

324
A spatially resolved map of specific sequence locations on the DNA can then be generated, leading to a ‘genetic barcode’ for each individual molecules.

325
Applications of such barcodes include genetic comparison of samples or populations, SNP detection and genome mapping or microorganisms.

Conclusion

326
The future of high throughput gene analysis will benefit greatly from the implementation of micro Total Analysis Systems, particularly for the analyses requiring multi-step procedures such as those including sample preparation.

327
Microfluidic devices will serve therefore to provide a level of standardisation not presently achievable with current laboratory standard operating procedures, especially those based upon minute quantities of samples.

328
As the development of microfluidic devices continues, newly devised microstructures need to utilise the better production and micro-engineering procedures.

329
Importantly however, the devices should also innately incorporate the requirements of each bioassay to enable them to be customised to the exact requirements of the specific bioassay and thus permit the optimum co-selection of device and choice of assay chemistry.