Bio-MEMS Technologies and Applications - Wang and Soper (Eds) Part 15 pdf

Furthermore, advancements in microfabrication technolo-gies allow the fabrication of submicrometer-sized or even nanometer-sized channels as well as new, on-chip optical detection scheme

A method, which combined the optical trapping and microfluidic-based droplet generation, was developed by He et al (2005) This work described

a method for selective and controlled encapsulation of a single cell or sub-cellular structure into a picoliter or femtoliter volume of an aqueous droplet surrounded by an immiscible phase Once the selected moiety was encased within the droplet, rapid laser photolysis within the droplet was demon-strated (Figure 15.20) The process confined the cell lysate within the pL volume of the droplet With further development, the droplet manipulations could be combined with droplet fusion for initiating chemical reactions, such

as derivatization reactions The biggest advantage of such an approach is that the cell lysate and cell biocomponents are confined within a small volume, minimizing the dilution and diffusion of the analytes of interest

15.3 Single-Molecule Detection in Microfluidic Devices

Initial studies on SMD (single-molecule detection) in microfluidic devices have indicated how powerful this marriage can become (Dorre et al 1997;

FIGURE 15.20

Single-cell enzymatic assay within an aqueous droplet in soybean oil (a) A mast cell was encapsulated in an aqueous droplet that contained the fluorogenic substrate FDG (b) Prior to photolysis of the cell, there was little fluorescent product within the droplet because the intrac-ellular enzyme β-galactosidase was physically separated from FDG by the cell membrane (c, d) After laser induced cell lysis (c), β-galactosidase catalyzed the formation of the product fluores-cein, which caused the droplet to become highly fluorescent (d) (Reprinted with permission from He, M., Edgar, J.S., Jeffries, G.D.M., Lorenz, R.M., Shelby, J.P., and Chiu, D.T (2005) Selective encapsulation of single cells and subcellular organelles into picoliter- and femtoliter-volume

droplets Analytical Chemistry 77(6): 1539–1544 © 2005, American Chemical Society.)

0 s

Effenhauser et al 1997; Fister et al 1998; Mathis et al 1997) Higher sensi-tivities, minimized sample consumption, increased throughput and speed, and full automation of the assays were pointed out as the most appealing benefits To this point, genomic-based studies are probably the area that has gained the most from the powerful combination of SMD and microfluidics Many new tools for DNA sequencing, mapping and sizing have been devel-oped over the last 10 years with some already commercialized or being introduced to the market as we speak SMD and microfluidics have also proven invaluable in studies of protein folding kinetics (Lipman et al 2003) and mechanistic studies of DNA–enzyme interactions (Lee et al 2006; van Oijen et al 2003) Furthermore, advancements in microfabrication technolo-gies allow the fabrication of submicrometer-sized or even nanometer-sized channels as well as new, on-chip optical detection schemes (e.g., zero-mode waveguides or near field scanner) (Levene et al 2003; Tegenfeldt et al 2001), which significantly extend the range of capabilities of this powerful tech-nique In the following sections, a description and discussion of some of the applications of SMD in microfluidics will be presented Anyone interested

in this topic should also consult the following reviews: de Mello (2003), Dittrich and Manz (2005), and Tegenfeldt et al (2004)

15.3.1 DNA Fragment Sizing

Separation of DNA molecules distinguished by their size (in base pairs [bp])

is a fundamental requirement in most molecular biology assays For example, the isolation of certain fractions of DNA samples based on size may be required for further experiments, such as the preparation of bacterial artificial chromosome (BAC) or plant artificial chromosome (PAC) libraries DNA siz-ing is also an important tool in a wide range of diagnostic applications, such

as restriction fragment length polymorphism (RFLP) Conventional DNA sizing techniques typically rely on gel electrophoretic sorting, which are time-consuming processes, require the use of relatively large amounts of sample, and have limited sensitivity and resolution To alleviate some of these limi-tations, various microfluidic approaches with SMD have been proposed Two general approaches can be distinguished The first one, practical for sizing shorter DNA fragments, uses on-chip microcapillary gel electrophoresis with highly sensitive single-molecule fluorescence burst counting for detection In the second approach, the electrophoretic step is completely eliminated from the assay as single DNA molecules are directly detected and sized using specially designed microfluidic or nanofluidic chips This second approach

is especially useful for sizing very long (e.g., genome-sized) DNA fragments The size of the DNA is transduced by quantitatively measuring the fluores-cence signal generated from a single DNA molecule traveling through a focused laser beam The DNA is made fluorescent by adding to it an inter-calating dye, which acts like a molecular light switch, in which the fluores-cence is “turned on” when the dye is incorporated into the double-stranded

DNA molecule and is “turned off” when the dye is in the surrounding fluid Because the loading of these dyes to double-stranded DNA is quantitative, the intensity of the fluorescence signal is proportional to the size of the DNA molecule On-chip capillary gel electrophoresis combined with a single-mol-ecule fluorescence burst counting for the separation and detection of fluo-rescently labeled DNA fragments was introduced by Haab and Mathies (1999) On-chip microcapillary gel electrophoresis for DNA separations offers several advantages over conventional gel and capillary gel electrophoreses These include shorter separation times, improved control of the sample injection plug, reduction in sample and buffer consumption, and increased throughput via high-density parallelism (e.g., 384 analyses running at the same time) (Aborn et al 2005; Tian et al 2005) On the other hand, using SMD significantly improves the sensitivity of the assay For example, 100 to

1000 bp DNA sizing ladder can be separated in approximately 300 s with a mass detection limit of approximately 1000 molecules using a microchip with

an effective separation length of 4 cm (Haab et al 1999) The unique feature

FIGURE 15.21

Scanning electron micrographs of the straight (a) and pinched (b) detectors Fluorescence images

of continuously injected fluorescein electrophoresing through the straight (c) and pinched (d) detectors The electrophoresis medium was 3% LPA and 1 × TAE The separation channel current was 3.2 µ A, and each cross-channel carried a focusing current of 1.6 µ A (current ratio 0.5).

(Reprinted with permission from Haab, B.B and Mathies, R.A (1999) Analytical Chemistry

71(22): 5137–5145 © 1999, American Chemical Society.)

Detection

Detection

44 µm 44 µm

of Haab’s design was the capability of focusing the DNA sample through a confocal detection volume in order to increase the percentage of DNA mol-ecules that could be detected without increasing the background Focusing

of the sample was achieved by physical narrowing (or tapering) of the channel dimensions and electrodynamic focusing (a sheath flow delivered shown to be 3 times higher than that of normal chip-based separations

A second SMD approach toward DNA sizing is based on direct measure-ment of fluorescent signals of DNA molecules labeled with intercalating dyes Because the amount of the intercalating dye bound to the DNA molecule is proportional to the size of the molecule, the length of the DNA molecule can

be found simply by quantifying the total fluorescent signal detected from the molecule The first application of this concept in a microfluidic format was demonstrated by Quake and coworkers (Chou et al 1999) The authors suc-cessfully used confocal detection of DNA flowing through a simple T-channel fabricated in PDMS Sizing of DNA molecules ranging in size from 2 to 200 kbp was achieved in less than 10 min using only 28 femtograms (approxi-mately 3000 molecules) of DNA (Figure 15.22) There are two important features of this approach that clearly differentiate it from gel electrophoresis methods First, the sizing time is independent of the length of the DNA molecules, whereas in gel electrophoresis it increases proportionally with the fragment length and can take days to size genomic DNA Second, the reso-lution of the assay increases with increasing DNA length, as more fluorescent signal is generated by longer molecules In the case of electrophoresis, longer DNAs typically result in reduced sizing resolution

FIGURE 15.22

(a) Optical micrograph of T-channel device The large channels have lateral dimensions of 100

µ m, which narrow down to 5 µ m at the T-junction The depth of the channels is 3 µ m (b) Analysis of ladder To test the upper length limit of the device, a ladder was analyzed Peaks corresponding to 50, 100, 150, and 200 kbp can clearly be resolved (Inset) The peak height measurement is linear even out to 200 kbp (Reprinted with permission from Chou, H.-P.,

Spence, C., Scherer, A., and Quake, S (1999) Proceedings of the National Academy of Sciences of

the United States of America 96(1): 11–13 © 1999, The National Academy of Sciences of the USA.)

120 100 80 60 40 20 0

Pulse height (V)

1.5 1.0 0.5 0.0

0 50 100 DNA size (kbp)

150 200

(a)

(b)

from cross channels) (Figure 15.21) The detection efficiency of this assay was

Further improvements to direct single-molecule sizing of DNA samples was provided by the use of submicrometer-sized fluidic channels (Foquet et al 2002) (Figure 15.23) By using ultrasmall channels, high signal-to-noise ratios (SNR)were achieved even for very fast flow speeds, up to 5 mm/s, thereby shortening the analysis time per molecule to less than a few milliseconds

FIGURE 15.23

(a) SEM showing the cross-section of channel (10 µ m width, 270 nm height) The width appears smaller in this image because the channel is viewed from a side angle (b) Photon burst histogram obtained for a mixture of several DNA fragments The red curves correspond to the fitting of a set of six Gaussian peaks by a least-squares method The positions of the peaks depend on the size of DNA fragments; the peak area is proportional to the relative concentration

of each fragment (c) Plot of the burst size as a function of the (known) fragment size Error bars correspond to the standard deviation of the peak sizes The dashed line is a linear least-squares fit The intercept was 256 photons, and the slope was 2.498 photons/bp The correlation coefficient is 0.9998 (Reprinted with permission from Foquet, M., Korlach, J., Zipfel, W., Webb,

W.W., and Craighead, H.G (2002) Analytical Chemistry 74(6): 1415–1422 © 2002, American

Chemical Society.)

Burst size (photons)

2 µm

300 275 250 225 200 175 150 125 100 75 50 25 0

0 2000 4000 6000 8000 10000 12000 14000

2.0

DNA fragment size (bp)

80000 70000 60000 50000 40000 30000 20000 10000 0

(c)

15.3.2 Sequencing of Single DNA Molecules

The Sanger reaction combined with electrophoretic separations has become

a workhorse for most DNA sequencing projects However, this technology has inherent limitations, especially in terms of cost, read length, and sensi-tivity For example, it took over 9 months at a total cost of over $3 billion to

make conventional Sanger sequencing rather impractical for massive com-parative genomic studies Deciphering the genomes of thousands of people will make it easier to track down genetic risk factors for many genetic disorders (e.g., diabetes or heart diseases) and eventually lead to a discovery

of new medical treatments Inexpensive DNA sequencing technologies could

be used to monitor the environment for specific microorganisms, including biowarfare agents Applications of single-molecule detection to DNA sequencing has monumental potential to alleviate the cost and speed issues associated with classical Sanger sequencing Two examples of SMD in micro-fluidic formats for DNA sequencing are briefly described below

Sequencing by controlled enzymatic digestion performed in microfluidic channels was proposed by Dorre and colleagues (1997) In this assay, DNA

is first synthesized with each nucleotide labeled with a fluorescent dye The DNA is fixed within a flow stream and an exonuclease enzyme is used to sequentially cleave the terminal nucleotide, which is then released into the flow and detected (Figure 15.24) As long as the nucleotides can be detected

in the order that they were released, the sequence of the DNA of interest can

FIGURE 15.24

The principle of single-molecule sequencing (1) A bead loaded with a labeled DNA molecule

is held by a trap laser (D 1064 nm) inside a transparent microstructure (2) The DNA is degraded sequentially by an exonuclease The liberated monomers are transported to the detection focus via electroosmotic flow (EOF) (3) Passing the focus, the labeled monomers are excited by a laser and emit photon bursts that are recorded with respect to their fluorescence characteristics (wavelength, lifetime) (Reprinted with permission from Dorre, K., Brakmann, S., Brinkmeier, M., Han, K.-T., Riebeseel, K., Schwille, P., Stephan, J., Wetzel, T., Lapczyna, M., Stuke, M., Bader,

R., Hinz, M., Seliger, H., Holm, J., Eigen, M., and Rigler, R (1997) Bioimaging 5(3): 139–152 ©

1997, John Wiley and Sons.)

Trap

Excitation laser

Labeled DNA, immobilized on a carrier particle EOF

sequence genomes from only five persons (http://www.ornl.gov/sci/ techresources/Human_Genome/project/whydoe.shtml) These limitations

be established The speed of this assay is, in principle, limited only by the rate of an enzymatic digestion (100 to 1000 bases/min) The complete detec-tion of all dye-labeled monomers, which are cleaved off of the isolated DNA template during the sequencing reaction, is an essential requirement This can be addressed by use of a microfluidic system with a confocal multiele-ment setup (Dorre et al 1997; Dorre et al 2001) A linear array of adjacent multimode optical fibers, each connected to its own avalanche photodiode detector, is set at each microfluidic channel The entire cross-section of the microchannel is illuminated at the detection area This arrangement gener-ates a number of overlapping femtoliter detection volumes across the chan-nel and ensures high detection efficiency

On-chip DNA sequencing can be also performed using sequencing-by-synthesis approaches (Braslavsky et al 2003; Kartalov and Quake 2004) A primed DNA template is anchored within a microfluidic reactor and its position is recorded using high-sensitivity digital cameras connected to a microscope The DNA template is then exposed to a mixture of a known type of standard nucleotide, its fluorescently tagged analog, and DNA polymerase If the tagged nucleotide is complementary to the template base next to the primer’s end, the polymerase extends the primer with it and a fluorescence signal is detected after a washing step Iteration with each type

of nucleotide reveals the DNA sequence By imaging many reactors, the sequence of multiple DNA molecules can be obtained in parallel The proof

of principle of sequencing-by-synthesis with an average read length of 3

bp has been demonstrated in a fully integrated PDMS microfluidic system omy of material and integration under the lab-on-a-chip paradigm, and has the ability to obtain sequence information from millions of independent molecules in parallel

15.3.3 Other SMD Bioassays On-Chip

Many types of genomic studies, such as comparative studies of differences among species or among individuals within a given species, do not require single-base resolution provided by DNA sequencing DNA mapping, which provides low resolution information about DNA sequence, is especially use-ful in such cases as it is much faster and less expensive for long, genome-sized DNA An interesting approach to DNA mapping using SMD in micro-fabricated devices is called direct linear analysis (DLA) (Chan et al 2004) with intercalating dyes and with sequence-specific fluorescent tags (e.g., fluorescent peptide nucleic acids [PNAs]), which target 7 to 8 bp of DNA

In the next step, the sample is introduced into a microfluidic chip containing micropost arrays and tapered channels By interaction with microposts, DNA molecules uncoil and finally stretch into a linear form in the tapered region

of the channel Individual DNA molecules are then interrogated with laser (Figure 15.25) This approach to DNA sequencing has advantages in

econ-(Figure 15.26) In the first step, double-stranded DNA molecules are labeled

In order to match a typical confocal detection volume of less than 1 fL, channels with dimensions of less than 1 µm × 1 µm have to be fabricated Some

of the techniques that have been successfully used for fabrication of submi-crometer-sized and nanometer-sized channels for SMD are listed in Table 15.2 With submicrometer or nanometer channel dimensions, the detection vol-umes on the order of tens of attoliters can be generated, which are roughly

100 times smaller than the effective observation volume of typical confocal setups Thus, nearly all of the individual molecules can be detected rapidly with high signal-to-noise ratios This is extremely important for the SMD-based assays, which depend on 100% sampling efficiencies, such as the DNA sequencing-by-digestion assays described in the previous section Higher counting efficiencies also improve the statistical accuracy of any single-mol-ecule characterization because all molsingle-mol-ecules are counted and contribute to the analysis Another advantage of extremely small detection volumes is that higher sample concentrations can be used without a risk of detection errors resulting from multiple occupancy of the detection volume This is especially useful for studying biochemical processes, which occur efficiently only at much higher than the pico- or nanomolar concentration regime Thus, nano-fabricated devices offer the potential to study these processes at physiolog-ically more appropriate concentrations Finally, extremely higher linear velocities can be used with submicron channel shortening analysis time per molecule, significantly reducing analysis time (Foquet et al 2004)

TABLE 15.2

Techniques Used for Fabrication of Submicrometer-Sized and Nanometer-Sized Channels

Fabrication

technique

Channel characteristics

Example

Sacrificial layer 350 nm wide × 250 nm

tall channels in fused silica

DNA fragment sizing

(Foquet et al 2002; Foquet et al 2004) Focused ion beam

(FIB)

individual 150–900 nm trapezoidal trenches

in silicon

Electrophoretic behavior of single, fluorescently labeled DNA molecules

(Campbell et al 2004)

Reactive ion etch

(RIE)

500 nm square cross-section channels in fused silica

Detection of individual quantum dots conjugated with organic fluorophores

(Stavis et al 2005)

Nanoimprint

lithography (NIL)

Array of 10 nm–200 nm trenches

Sizing of genome-length (>1 million bp) DNA

(Cao et al 2002a,b; Tegenfeldt et al 2004a,b) Polymer fibers as

sacrificial layer

Suspended glass nanotubes, 100–500

nm diameter

Detection of single fluorescently labeled proteins

(Verbridge et al 2005)

One has to remember, however, that when using submicrometer- and nanometer-sized channels, new issues can arise These include channel clog-ging, increased molecule-to-surface interactions, high levels of background caused by increased refraction from channel surfaces, interfacing nanometer-sized channels to macroscale instrumentation, and extremely high back pres-sures induced by nanometer-sized channels when hydrodynamic pumping

is used Some of these issues can be addressed by proper design of the nanofluidic devices For example, nanochannels can be accessed by microf-luidic channel networks, which are easier to interface to macroscale instru-mentation (Cao et al 2002a, b; Foquet et al 2004) Dense arrays of parallel nanochannels can be used to reduce clogging issues as well (Cao et al 2002a; Tegenfeldt et al 2004b) (Figure 15.27), whereas suspended glass nanotubes with extremely thin walls reduce the problem of substrate-related back-problems associated with molecule adhesion to the channel walls and elec-trokinetic forces can be used to drive the fluids eliminating the problems of high back pressures

FIGURE 15.27

(a) SEM image shows the “roof” and profile of the nanochannels fabricated with NIL and sealed

by the SiO2 sputtering process (b) Buried nanochannels under the sealing roof were shown to

be perfectly intact after sealing (c) CCD image of λ-phage DNA concatemers stretched in nanofluidic channels (Reprinted with permission from Cao et al 2002b, © 2002, American Institute of Physics.

30 µm

(c)

ground (Verbridge et al 2005) (Figure 15.28) Surface coatings can reduce the

less expensive and offer greater flexibility in fabrication strategies for mass producing microstructures

Many polymeric materials possess good light transmission properties for use in optical detection (see Figure 15.29a) Unfortunately, most polymers also show high fluorescent backgrounds in the visible range of the electro-magnetic spectrum as compared to glass (see Figure 15.29b) This high auto-fluorescence can be attributed to either the polymer itself or to different additives, such as plasticizers, commonly used in commercial-grade poly-mers (compare autofluorescence values for different types of PMMA in Fig-ure 15.29b) The autofluorescence of polymeric material is strongly dependent on excitation wavelength with higher autofluorescence values observed for shorter wavelengths (Piruska et al 2005; Wabuyele et al 2001) PDMS and PMMA (especially high purity grade) are the only polymeric materials used for microfluidic chip applications that come close to glass in terms of its low autofluorescence properties using a relatively wide range of excitation wavelengths (λexc > 480 nm) These properties of both PDMS and PMMA have already been widely used in SMD (Braslavsky et al 2003; Chou

et al 1999; Effenhauser et al 1997; Wabuyele et al 2003; Wabuyele et al

FIGURE 15.29

(a) The transmission spectra of the plastic materials taken at normal incidence (Reprinted with permission from Piruska, A., Nikcevic, I., Lee, S.H., Ahn, C., Heineman, W.R., Limbach, P.A., and

Seliskar, C.J (2005) Lab on a Chip 5(12): 1348–1354 © 2005, The Royal Society of Chemistry.) (b)

LIF background levels of different polymers measured at three different excitation wavelengths,

488 nm, 632.8 nm, or 780 nm The average fluorescence intensity (cps) of the polymers was normalized with respect to the value obtained for glass at the same excitation wavelength COC: cyclic olefin copolymer; PDMS: poly(dimethylsiloxane); PMMA: poly(methylmethacrylate); MG-PMMA: medical-grade PMMA; G-MG-PMMA: grey acrylic; C-MG-PMMA: clear acrylic; ABS: poly(acry-lonitirle-butadiene-styrene); PETG: polyethylene terephtalate glycolate; PS: polystyrene; PC: poly-carbonate; LX-PC: Lexan-grade PC; PET: polyethylene terephtalate; PU: polyurethane; PP: polypropylene; NY: nylon; PSU: polysulfone; HDPE: high-density polyethylene; LDPE: low-density polyethylene (Reprinted with permission from Shadpour et al 2005 , © 2005, Elsevier B.V.)

PMMA (0.25 mm) PMMA (3.0 mm) PMMA (5.0 mm)

Boro float glass (1 mm) COC (1 mm)

PC (0.25 mm)

PC (2.0 mm) PDMS

100

80

60

40

20

0

100

80

60

40

20

0

200 400 600 800 1000

200 400 600 800 1000

Wavelength/nm

10 9 8 7 6 5 4 3 2 1 0

T PS PC PET

-P PU PP NY PSU

488.0 nm 632.8 nm 780.0 nm