Microfluidics and microarray based approaches to biological analysis 2

Separation Channel1 3 2 4 Sample reservoir Buffer reservoir Sample waste Buffer waste Separation Channel 1 3 2 4 Sample reservoir Buffer reservoir Sample waste Buffer waste Sample re

CHAPTER 2 MICROFABRICATED DEVICES FOR NUCLEIC ACID AMPLIFICATION AND ELECTROPHORETIC SEPARATION

2.1 Introduction

The interest in electrophoretic separations on microfabricated devices has grown dramatically over the past few years due to the many advantages it offers The short plugs, good dissipation of Joule heating, and high field strengths result in extremely rapid separations that consume only picoliter sample volumes

2.1.1 Theory of Electrophoretic Based Separations

Electroosmotic Flow

As shown in Figure 2.1, the inner walls of fused silica capillaries possess an intrinsic negative charge due to the presence of weakly acidic silanol groups (-SiOH) (pKa ~ 5.3).1 Cations in solution build up near the capillary surface to balance this charge, thus forming an electrical double layer Upon the application of an electric field across the length of the capillary, the cations in the diffuse portion of the double layer migrate towards the cathode Since these cations are hydrated, they induce a bulk flow of solution within the capillary towards the cathode The magnitude of this electroosmotic flow (EOF) is generally described by the Scholuchowski equation:

shear, and is a function of the deprotonation of the silanols, ion adsorption onto the surface and the ionic strength of the buffer

+ _

+ + + + + +

N

+

+ +

compact layer

diffuse layer

+

+ + + + + +

N

+

+ +

compact layer

diffuse layer

Figure 2.1 Schematic representation of the electrical double layer at the capillary

inner wall Positive, negative and N signs respectively represent cations, anions, and non-charged species

Electrophoretic Flow

The basis for the electrophoretic process is the differential migration of sample ions relative to solvent molecules under the influence of an externally applied electric field

There are two main contributors to the ion mobility, the applied electric force (Fe) and

the frictional force (Ffr) that the molecules experiences as it moves through the buffer

solution The applied electric force (Fe) depends on the charge q of the particular ion

and electric field E:

The mobility is then the electrophoretic velocity normalized to the electric field strength:

The primary means of material transport on chips are the electrokinetic phenomena, i.e electrophoretic and/or electroosmotic effects Buffer and sample flows within the channel manifold are precisely controlled through potentials applied to the reservoirs The intersection between the injection and separation channels has been called the injection cross since its volume defines the sample volume injected into the separation channel The integrated injectors are usually either cross-channel injectors, formed by orthogonally intersecting the separation channel with a channel connecting the sample

to waste as shown in Figure 1.1, or twin-T injectors, where the two arms of the sample

to waste channel are offset to form a larger injection region as shown in Figure 2.2

Separation Channel

1

3

2 4

Sample reservoir

Buffer

reservoir

Sample waste

Buffer waste

Separation Channel

1

3

2 4

Sample reservoir

Buffer

reservoir

Sample waste

Buffer waste

Sample reservoir

Buffer

reservoir

Sample waste

Buffer waste

Separation Channel

1

3

2 4

Sample reservoir

Buffer

reservoir

Sample waste

Buffer waste

of the technology developed for the semiconductor industry can be transferred directly

to chip fabrication using insulating substrates.2 Although slight variations in glass fabrication techniques exist, the general fabrication aspects are similar (Figure 2.3) Generally a positive photoresist is spin-coated on top of the substrate, and the channel design is transferred to the substrate using a photomask Following exposure and development of the photoresist, the channels are etched into the substrate in a dilute HF/NH4F bath Because the photoresist used to mask wet etches does not adhere well

to glass, a sacrificial layer that adhere well to both glass and photoresist is often used

as intermediate layer (Figure 2.3) The access holes can be drilled on the etched substrate or another blank glass wafer To form the closed network of channels, the cover plate is bonded to the substrate over the etched channels Finally cylindrical reservoirs, to hold buffers and samples, are affixed onto the extremity of the channels

Figure 2.3 Schematic diagram of the photolithographic process used for making chips

Although most microchip devices fabricated to date use glass, there have been several reports of devices fabricated from a variety of polymeric substrates including poly(dimethylsiloxane) (PDMS),3 poly(methyl methacrylate) (PMMA),4 acrylic,5 and polycarbonate.6 The interest in polymeric microfluidic devices stems primarily from the fact that plastic chips are less expensive to produce, and can be disposed after single use

2.1.2.2 Injection Methods

Sample injection, by default, was the first functionality integrated into single substrates

by fabricating intersecting channels at right angles The formed cross was intended to create a geometrically defined sample plug with a volume close to that of the

intersection cross Later on, various sample injection methods were developed making use of both EOF and electrophoretic effects

Floating Injection

For the floating injection mode, only two electrodes are used Voltage is applied between the sample reservoir and the sample waste reservoir (Figure 2.4, left) Buffer and buffer waste reservoirs are left floating Consequently, the sample solution is pumped past the injection cross to the sample waste reservoir If the injection potential

is applied long enough to ensure that even the slowest moving component passes through the injection cross, the injection plug will have a composition representative of the sample to analyze To introduce the sample plug into the separation channel, the potentials are switched from the sample loading to the separation mode of operation, i.e potential is applied between buffer reservoir and buffer waste reservoir (Figure 2.4, right) However, since the potential in the separation channel is left floating during injection, the analyte is free to diffuse into the separation channel This problem is particularly important with small ions and molecules having high diffusion coefficients

Sample

reservoir

Buffer reservoir

Buffer waste

Sample waste reservoirSample Sample waste

Buffer reservoir

Buffer waste

Sample

reservoir

Buffer reservoir

Buffer waste

Sample waste reservoirSample Sample waste

Buffer reservoir

Buffer waste

Figure 2.4 Schematic of floating injection

Pinched Injection

The pinched injection is achieved by spatially confining the sample in the cross intersection before dispensing it into the separation channel.7,8 The sample flow between the sample reservoir and sample waste is electrokinetically confined by the incoming buffer streams from the buffer reservoir and buffer waste (Figure 2.5, left) The extent of sample focusing is regulated by the electric field strength in separation channel versus injection channel Once the sample flow has reached a steady state, the electric field is switched to a dispensing step serving also as the separation step (Figure 2.5, right) In this step, to prevent sample leakage into the separation channel and achieve a short axial extent sample plug, an electric field is also applied to sample buffer and sample waste to draw sample back from the intersection

Sample reservoir

Buffer reservoir

Sample waste reservoirSample

Buffer waste

Sample waste

Buffer reservoir

Buffer waste

Sample reservoir

Buffer reservoir

Sample waste reservoirSample

Buffer waste

Sample waste

Buffer reservoir

Buffer waste

Figure 2.5 Schematic of pinched injection

When the pinched and floating injections are compared, the pinched sample loading is superior in two areas: temporal stability and plug length The pinched sample injection

is independent of time, electrophoretic mobility, and electric field strength On one hand a smaller plug length leads to higher efficiency but on the other hand can be detrimental to the sensitivity as less analyte is injected

Gated Injection

In order to increase the amount of analyte injected into the separation channel, the gated injection was developed.9,10 The gated valve has a loading/separation mode where the sample flows from sample reservoir to the sample waste while the buffer flows from the buffer reservoir to the buffer waste to prevent sample leakage and provide continuous buffer supply into the separation (Figure 2.6, left) To make an injection, the field in the buffer reservoir is set to zero allowing a plug of sample to move into the separation channel (Figure 2.6, middle) In the subsequent separation step (Figure 2.6, right), the field is switched back to the loading/separation step The

buffer flow cuts the sample plug, and the injection valve returns to its original state The length of the injection plug is therefore a function of both the time of the injection and the electric field strength

Sample reservoir

Buffer waste

Sample waste

Buffer

reservoir reservoirBuffer

Buffer reservoir Sample waste

Sample waste

Sample reservoir reservoirSample

Buffer waste

Buffer waste

Float

Sample reservoir

Buffer waste

Sample waste

Buffer

reservoir reservoirBuffer

Buffer reservoir Sample waste

Sample waste

Sample reservoir reservoirSample

Buffer waste

Buffer waste Float

Figure 2.6 Schematic of gated injection

The gated injection differs from the pinched injection on several counts With the gated injector, the sample migrates electrophoretically down the separation column and

is cleaved by restoring the flow of buffer from the buffer reservoir The pinched sample loading pumps the sample through the injection cross, and the plug, which is injected onto the separation channel, is the sample which resides in the injection cross With the pinched sample loading, the amount loaded onto the separation column is time independent and has no electrophoretic bias The gated injector is both time dependant and electrophoretically biased, but allows for injecting more analyte and is therefore appropriate for samples of low concentrations

2.1.2.3 Advantages

Both electrophoretic migration of ions and electroosmotic flow velocity are linearly dependent on the axial electric field strength applied While in the case of pressure-driven flow the external force is applied across the whole cross section of the tube leading to a parabolic flow profile, in electroosmosis the external force can only be exerted to a thin sheet of fluid close to the wall, thus leading to a plug-flow profile The short plugs, good dissipation of Joule heating, and high field strengths result in extremely rapid separations that consume only picoliter sample volumes

2.1.3 Combined Microchip Based Electrophoretic Separation and Micro Polymerase Chain Reaction

Micro CE dramatically increases the speed of nucleic acid separation However, prior

to separation, if present in low amounts, the nucleic acid of interest should be amplified by Polymerase Chain Reaction (PCR) to obtain enough material for detection PCR has revolutionized bioscience due to its ability to exponentially and specifically amplify DNA templates from very small starting concentrations Miniaturization offers improved thermal energy transfer compared to conventional macrovolumes, resulting in a greatly increased speed of thermal cycling and reduced amount of expensive reagents used A PCR-CE combination reduces the contamination problem, decreases the risk of infection, and allows for faster execution

of the analysis through reduced manual manipulations There have been several reports

on integration of DNA amplification and electrophoretic separation on a single microfabricated chip These devices contain small reaction wells, which were thermocycled to generate amplicons, followed by the injection/separation/detection steps in the interconnected microchannel network

Woolley et al reported the PCR amplification of DNA in a microfabricated chamber

containing silicon heaters that could be directly interfaced with an electrophoretic chip for PCR product analysis.11 Oda et al reported a noncontact heating thermocycling

using inexpensive IR for rapid PCR amplification, the detection was later made by conventional slab gel electrophoresis or CE.12 These approaches, without a large heating/cooling block, significantly increase the rate of thermal cycling, but the temperature control becomes more difficult than for conventional systems Multiplex reactions can also be performed on-chip By filling the sample reservoir with the PCR mixture and cycling the whole CE chip in a conventional thermocycler, rapid amplification was achieved and the PCR products were later analysed on the same chip.13 An integrated system combining fast on-chip DNA amplification by local thermocycling followed by microchip electrophoretic sizing was reported.14 An interesting alternative to a microchamber reaction format has been proposed by Kopp

et al who demonstrated a continuous-flow PCR system.15 A PCR cocktail was pumped continuously through a serpentine glass channel, periodically passing the three temperature zones to perform denaturation, annealing and amplification steps Although total reaction time was relatively long (50 min for 20 cycles), multiple simultaneous reactions can be carried out by sequential introduction of separate reactions in each loop A microfabricated structure for integrated PCR amplification

and CE separation was reported by Burns et al.16 They developed a device that used microfabricated channels, heaters, temperature sensors and fluorescence detectors to analyze nanoliter-size DNA samples The device was capable of measuring aqueous reagent and DNA containing solutions, mixing the solutions together, amplifying or digesting the DNA to form discrete products, and separating and detecting these products An important aspect of the device was the ability to meter a small, accurate

volume of fluid by means of a hydrophobic patch and injected air More recently,

Lagally et al developed an integrated monolithic system incorporating several 280 nL

PCR chambers etched into a glass structure, connected to microfluidic valves and hydrophobic vents for sample introduction and immobilization during thermal cycling.17 They used DNA sample of very low concentration and observed the stochastic PCR amplification and analysis of single-molecule DNA templates and were able to detect single-molecule amplification Recently, Northup and co-workers designed a portable system containing a miniature analytical thermal cycling instrument in which a silicon-micromachined reaction chamber with integrated heaters and optical windows was used to conduct PCR amplification However, this particular device was not interfaced with any electrophoretic separation channel, but instead characterization of the amplified product was accomplished by real-time fluorescence monitoring.18 This device is especially useful for field use Real-time PCR is in that case especially convenient as it does not require time-consuming post PCR manipulation and processing of the reaction with slab gel or CE, hybridization to DNA arrays or mass spectrometry This strategy was applied for the very rapid detection of bacteria.19 Efforts have been made to minimize the time needed for on-chip PCR amplification to less than 240 seconds.20 Polymeric chips were used, thus reflecting the wider arrays of applications of polymeric materials to microfabricated devices A hybrid poly(dimethylsiloxane) (PDMS)-glass microchip for functional integration of DNA amplification and gel electrophoresis has also been reported Thermoelectric heating/cooling was used in this case Such devices can be disposable due to their inexpensive and relatively simple fabrication In addition to a decrease in amplification time, there is a wide interest in trying to increase the density of reactions carried out at

the same time and Nagai et al reported a high throughput PCR amplification in a

silicon based array.21

2.2 Results and Discussion

2.2.1 Method Development Based on Conventional Capillary Electrophoresis

2.2.1.1 Optimization of Separation Condition with Conventional Capillary Electrophoresis

Slab gel electrophoresis was previously the method of choice to separate PCR amplified DNA fragments However, slab gel electrophoresis, followed by stain or probe detection, is often time-consuming and labor intensive Capillary electrophoresis (CE) is an alternative approach, which offers many advantages Because of the high surface-to-volume ratio, the efficiency of heat dissipation is much higher than in slab gel electrophoresis Thus, much higher electric fields can be applied during electrophoresis, resulting in faster separations CE also offers the possibility of full automation, avoiding time-consuming pouring and loading, as well as increased precision and the possibility to quantitate results

The separation matrix chosen was based on entangled cellulose polymers since these are known to provide good sieving ability allowing for easy filling of capillaries and channels The monointercalating fluorophore Thiazole Orange (TO) was chosen as intercalating dye because it provided the most sensitive detection among monomeric dyes22,23,24 and because of its compatibility with the ion argon laser excitation line (488 nm) Using 1×TAE (89mM Tris, 89mM Acetic acid, 2mM Disodium EDTA), 0.75 % Hydroxyethylcellulose (HEC) and 0.25 µg/mL TO, the two Z and W female bird genes

could be baseline resolved at 12kV as shown in Figure 2.7 Similarly to the mammalian X and Y genes, the Z and W genes are bird sex genes, but whereas mammalian females have two X genes and males have one X and one Y gene, male birds have two Y genes and female birds have one X and one Y gene The Z male bird genes appear as only one single peak because the two genes are of the same size (Figure 2.8) The Z gene, common to the male and female, migrates faster and thus has

a smaller number of bp than the W gene present only in the female species High separation efficiencies of 1 × 106 and 8.8 × 105 theoretical plates per meter were obtained for the female bird genes The separation of the two genes depends on the polymer concentration but an increase in the HEC concentration did not result in any improvement in separation efficiency and resulted only in an increase in polymer viscosity and was therefore more difficult to flush into the capillary The use of other cellulosic polymers was also attempted Methylcellulose (MC) was very difficult to solubilize and needed overnight stirring Neither hydroxypropylmethylcellulose (HPMC) nor MC could achieve the high separation efficiencies obtained for HEC

Figure 2.7 Capillary electrophoretic separation of female bird genes; concentration:

10 ng/µL; buffer: 1xTAE, 0.75 % HEC, 0.25 µg/mL TO; separation voltage: 12 kV; separation capillary: 50 µm i.d × 70 cm total length (effective length 57 cm)

Figure 2.8 Capillary electrophoretic separation of male bird genes (same separation

conditions as in Figure 2.7)

Some of the main advantages of capillary electrophoresis over slab gel electrophoresis may be attributed to its high heat dissipation and shorter separation times due to the use of higher electric fields To reduce the migration times of the fragments, without loss of resolution, the voltage was increased from 12 kV (170V/cm) to 20 kV (285V/cm) This resulted in a loss of efficiencies as shown in Table 2.1 In the range from 12 kV to 20kV, the two peaks were baseline resolved (Figure 2.9) However a decrease in peak height was observed at higher voltages Given that the TO intercalating dye was present in the running buffer and that the dye would react with the DNA as the latter migrated through the polymer, a higher voltage and consequently shorter migration time would lead to a lower interaction time between DNA and the dye Consequently, lower fluorescence intensities of the DNA fragments at higher voltages were observed The second peak exhibited a lower fluorescence intensity compared to the first peak, irrespective of the applied voltage This can be explained

by a depletion phenomenon.25 As the fragments migrated through the capillary, a depletion of dye occurred, resulting in a lower interaction between the dye and the DNA fragments from the first fragments to the last ones Separation at 15 kV offered a

relative short run time Relative standard deviations (RSDs) of the migration times for

six successive injections are listed in Table 2.2 The separated fragments can thus be

accurately determined with high efficiencies, and limits of detection of 0.1 ng/µL were

obtained (S/N>3)

Figure 2.9 Influence of separation voltage, (a) 12 kV, (b) 15 kV, (c) 18 kV and (d) 20

kV Other separation conditions are same as in Figure 2.1

Table 2.1 Separation efficiencies as function of the separation voltage, with

N=5.54(t/∆t1/2)2, where t is the migration time and ∆t1/2 is the measured full width of a

peak at half maximum

Female Male

Z W Z

Table 2.2 R.S.D.s for 6 consecutive runs

2.2.1.2 Influence of DNA: Dye Ratio on the Electrophoretic Separation

Simple (monomeric) intercalators such as ethidium bromide (EtBr) and TO, which

have reduced steric constraints, could potentially intercalate with a ratio of 1/1

(dye/bp) However, results from previous studies indicated a saturation maximum of

one intercalator molecule per two base pairs.26,27 The intercalating dye is expected to

serve a dual role in the separation process It improves the detectability of the DNA but

also acts to enhance resolution by uncoiling the DNA and improving its ability to

interact with the polymer matrix.28 The intercalation of positively charged TO causes

longer migration times of DNA fragments as the charge of the DNA becomes less

negative.27 TO also greatly enhances the native fluorescence of the DNA as it

intercalates in the double stranded (ds) DNA However, TO should be mixed with the

buffer, and not with the DNA sample, because preformed complexation of ds-DNA

with dyes often results in band broadening and tailing.27,29 In addition, DNA

precomplexed with dye will not be electrokinetically injected as effectively as

unstained DNA since the negatively charged fragments are partially neutralized by the

cationic dyes.23 Using

{1-(4-[3-methyl-2,3-dihydro-(benzo-1,3-oxazole)-2-methyidene]-quinolinium)-3-trimethyl ammonium propane diiode} (YO-PRO-1) as

dye, McCord et al observed that the dye: DNA ratio does not have much influence on

separation efficiency On the contrary, when using TO, the dye: DNA ratio is of

critical importance.27 Increasing the TO concentration from 0.25 to 1 µg/mL resulted

in broad peaks (Figure 2.10) and relatively low peak heights as previously observed

Figure 2.10 Influence of increasing TO concentration on separation efficiency;

sample concentration: 10 ng/µL, buffer: 1xTAE, 0.75 % HEC, 12 kV, 0.25 µg/mL TO; separation voltage: 12 kV; separation capillary: 50 µm i.d × 70 cm total length (effective length 57 cm)

2.2.1.3 Determination of Fragment Size

In order to determine the size of the analyzed fragments, the same buffer and separation conditions were applied when separating the 25 bp ladder The separation at

15 kV, with efficiencies ranging from 1 to 1.9 × 106 theoretical plate numbers, is shown in Figure 2.11 The Ogston model30,31 has been used to describe the mobilities

of DNA fragments of low numbers of bp in polymer solutions The mobility µ varies with the size of the DNA molecule: µ=µ0e-KCN, where µ0 is the mobility of DNA in free solution, K is a constant and N is the number of bp of the DNA fragment Using the data obtained with the 25 bp ladder, we plotted µ=f(N) as shown in Figure 2.12 This model can satisfactorily describe the mobilities of fragments of sizes ranging from 25 to the usual value of N=25032, with a correlation coefficient of 0.9978 For fragments with a larger number of base pairs, the reptation model has been developed.33,34,35 In that case, the DNA fragments move “snake-like” through the pores

of the polymer, and the mobilities are inversely proportional to the sizes of the DNA fragments: µ≈N-1 By plotting the mobilities as a function of 1/N for the fragments with sizes ranging from 275 to 500 bp, a straight line was obtained (Figure 2.13) with a correlation coefficient of 0.9938 Using this latter graph we could determine the sizes

of the two bird sexing genes Z and W, which respectively contained 350 and 368 base pairs

Figure 2.11 Capillary electrophoretic separation of 25 bp ladder (10 ng/µL); , buffer:

1xTAE, 0.75 % HEC, 0.25 µg/mL TO; separation voltage: 12 kV; separation capillary:

50 µm i.d × 70 cm total length (effective length 57 cm)

Ln(u)

-6.3 -6.25

-6.2 -6.15

-6.1 -6.05

1.6 1.65

1.7 1.75

1.8 1.85

1.9

1/N (*10E3 )

u 1s-1)

(*10E4cm2V-Figure 2.13 Mobility µ as function of the inverse of the number of base pairs (1/N)

2.2.1.4 Capillary Coatings

A chemical gel consists of a sieving matrix such as polyacrylamide, which is linked and/or chemically linked to the capillary wall, or channel wall The gel is prepared in the same manner as slab gels In CE and micro CE, the solution is pumped

cross-into the capillary/channel where it polymerizes in situ The media is permanently

“fixed” in the capillary When the gel is no longer functional, neither is the capillary, nor the microchip Hence, some physical gels have been developed These are, in the simplest terms, “viscous buffers”; solutions of hydrophilic polymers dissolved in an appropriate buffer The attractive characteristic of physical gels is that the polymers used are not cross-linked and therefore, they can be pumped out of the capillary at the end of each run, the most important of these polymers being hydroxyethyl cellulose, linear polyacrylamide (PA), and polyethylene oxide The disadvantage of using PA is that it must be used at a somewhat higher viscosity than some of the other polymers, and the monomers used in its manufacture are toxic Hence cellulose-based polymers were used throughout this work The most important prerequisite for the use of a gel in

a capillary is the complete elimination of the electroosmotic flow (EOF) If this condition is not adequately fulfilled, the gel is extruded out of the capillary by the EOF and separation becomes impossible The first widely used method, to date, for the functionalization of such capillaries was first described by Hjerten (Figure 2.14).36

Si O Si

SiOH SiO- K+

SiOH HCl

SiOH SiOH (MeO)3 Si (CH2)3 O CH3

CH 2

+

Si (CH2)3 O

CH3

CH2Si

Si

O O OMe

TEMED, Persulfate

Si (CH 2 ) 3 O

CONH2

CH 2

Si Si

O O OMe

Figure 2.14 Reaction scheme for coating of the capillary wall

The functionalisation of capillaries and channels, using this strategy, is straightforward and gives reproducible results However, the coating procedure can be lengthy and the coating is not stable at a high or low pH In addition, it involves the use of toxic chemicals such as acrylamide, which is detrimental when working with microchips and

a dynamic capillary coating was therefore investigated and compared with a permanent coating

Polyvinylpyrrolidone (PVP) has self-coating properties to the silica wall, reducing the EOF values significantly,37 as well as sieving properties for DNA sizing separations.38The main advantage of dynamic coatings such as PVP is the simplicity of preparation