Chapter 2 Using Microchip Gel Electrophoresis to Probe DNA–Drug Binding Interactions Nan Shi and Victor M. Ugaz Abstract Binding of small molecules with DNA plays an important role in many biological functions such as DNA replication, repair, and transcription. These interactions also offer enormous potential as targets for diagnostics and therapeutics, leading to intense interest in development of methods to probe the underlying binding events. In this chapter, we present a new approach to investigate the structural changes that accompany binding of DNA and small molecules. Instead of relying on conventional yet delicate single- molecule imaging methods, we show how a single microchip gel electrophoresis experiment incorporating both constant electric field and on-off actuation over a specific frequency range enables fundamental structural parameters (e.g., contour and persistence lengths) to be simultaneously determined. The microchip format offers an attractive combination of simplicity and scale-up potential that makes it amenable for high-throughput screening. Key words DNA binding, Contour length, Persistence length, Gel electrophoresis
1 Introduction Considerable efforts have been directed toward characterizing interactions between DNA and small molecule binding agents, owing to their potential usefulness as drug targets. Many of these compounds function by intercalation with DNA, binding to the major or minor groove of the double-stranded backbone [1–3], or via activity of substituent groups that enable external binding [4]. It is therefore of interest to understand and characterize the corresponding structural changes that are induced in the DNA complex upon these bindings. Some of the most commonly employed methods to probe binding interactions are based on experiments involving single DNA molecules, either via force measurement upon stretching with optical or magnetic tweezers or by direct imaging with atomic force microscopy (AFM) [5–7]. In the force measurement approach a DNA molecule large enough to be localized under optical Juan C. Stockert et al. (eds.), Functional Analysis of DNA and Chromatin, Methods in Molecular Biology, vol. 1094, DOI 10.1007/978-1-62703-706-8_2, © Springer Science+Business Media New York 2014
13
14
Nan Shi and Victor M. Ugaz
microscopy (typically λ-DNA) is stretched, and the corresponding force is measured as a function of displacement during both extension and relaxation. Care must be taken to ensure that the applied forces are small enough to maintain chemical equilibrium between the DNA and the bound drug compound, and to avoid introducing enthalpic effects that may distort the measurement. The resulting force-displacement data are then fit using the worm-like chain model Fp 1 x = 1 − kBT 4 L
−2
−
1 x + . 4 L
where F is the applied force, p is the persistence length of DNA– drug complex, x is the end-to-end displacement, L is the contour length of the complex, and kBT is the thermal energy (Boltzmann constant multiplied by the absolute temperature in Kelvin degrees) [8]. Although these experiments have played an instrumental role in providing fundamental insights about DNA–drug interaction mechanisms, they are delicate to perform and require sophisticated equipment (e.g., optical or magnetic tweezers) that is not readily available in all laboratories. The optically based format also inherently limits analysis to very-large-sized DNA molecules that can be directly visualized (i.e., contour lengths on the order of μm). We have recently explored DNA transport during gel electrophoresis in the entropic trapping (ET) regime, occurring when the average gel pore size is comparable to the size of the equilibrium random coil conformation of the DNA molecule. Our analysis revealed that a distinct mobility peak appeared at a specific actuation period when a periodically oscillating electric field was applied under conditions where ET-dominated transport occurs. We have developed a transport model that allows us to correlate the size of the DNA molecule with the position of the mobility peak. Since the molecular conformation of DNA is directly influenced by binding interactions, we are therefore able to apply an approach that uses microchip-based electrophoresis to gather structural information associated with DNA–drug complexation in a convenient format that circumvents the limitations of previous single-molecule methods.
2 Materials and Equipment Unless otherwise noted, all solutions are prepared using deionized (DI) water and stored at 4 °C. 2.1 Chemicals
1. DNA samples: 100 bp increment double-stranded DNA ladder (Cat # 170-8202; Bio-Rad; 100 μg/mL); single size 600 bp DNA fragment (NoLimits; Fermerntas; 500 μg/mL).
Microchip Gel Electrophoresis to Probe DNA Binding
15
2. Running buffer: 10× TBE (Cat # 161-0741; Bio-Rad). Dilute to 0.5× concentration. 3. DNA-binding compound: daunomycin hydrochloride (Cat # D8809; Sigma-Aldrich). Prepare an aqueous 1 mg/mL stock solution, and dilute to the desired concentration immediately before each experiment. 4. Intercalating dye: YOYO-1 iodide (Cat # 620722; Invitrogen). 5. Anti-photobleaching agent: 2-mercaptoethanol (BME) (Cat # M3148; Sigma-Aldrich). 6. Cross-linked polyacrylamide gel: Duracryl (30 % T, 2.6 % C; NextGen Sciences). 7. Photoinitiator: ReproGel Solution B (GE Health Care). 2.2 Microchip Gel Electrophoresis
Electrophoresis microchips (see Fig. 1) are fabricated following previously described methods [9]. The microchips consist of three primary components: (1) an etched glass microchannel (300 × 50 μm cross section) incorporating a separation channel with two side arms at one end for sample loading, (2) a silicon substrate containing an embedded electrode array, and (3) a PC board providing external connection to the on-chip electrodes (through wire bonding) via a 50-pin card edge connector and I/O block. The glass microchannels are bonded to the silicon substrate using UV-curable SK-9 Lens Bond (Summers Optical).
2.3 Imaging and Detection
Electrophoretic transport of the fluorescently labeled DNA is monitored using an Axioskop 2 microscope (Zeiss) with HBO 100 mercury arc lamp illumination, a fluorescein isothiocyanate (FITC) filter set, and a long working distance 10× objective. An ORCA-ER CCD camera (Hamamatsu) is used for image acquisition. A motorized x − y translation stage is employed to enable synchronized positioning of the microchip and actuation of the camera shutter (Openlab, PerkinElmer).
2.4 UV Curing
An Omnicure Series1000 spot curing system with collimating lens attachment (EXFO) is used for gel casting inside the microchip.
2.5 Function Generator
An Agilent 33220A function generator interfaced with a voltage amplifier (Trek Model 603) is used to generate waveforms for application of time-varying electric fields. Wave functions with different periods are designed using the Agilent IntuiLink Waveform Editor and then upload and stored in the device. During the experiments, the waveforms are monitored using a Hewlett-Packard 54603B oscilloscope.
16
Nan Shi and Victor M. Ugaz
a stationary microscope
sample motion direction t1 t2 t3
stage motion direction tn
t1
t2
t3
tn
b
Fig. 1 Experiment setup. (a) Overview of the image collection and analysis procedure using whole channel scanning detection. In each scan, a motorized stage carrying microchip enables a sequence of snapshots to be taken by a CCD camera interfaced with a fluorescence microscope. These individual snapshots are then assembled into a single composite image depicting all migrating DNA in the separation channel. A trace of fluorescence intensity versus position along the microchannel center line is then extracted for computation of mobility and diffusion coefficients. (b) Top view of an assembled electrophoresis microchip. The cross-sectional dimensions of all channels are 275 (width) × 45 (height) μm. An electrode array is patterned on the silicon substrate (the “floor” of the microchannel) to enable on-chip collection and focusing of DNA samples. The microchips are mounted on printed circuit boards that allow the on-chip electrodes to be individually addressed after wire bonding
Microchip Gel Electrophoresis to Probe DNA Binding
17
3 Methods 3.1 Gel Casting
For the sub 1 kb DNA lengths employed here, prepare 6.75 % T, 2.6 % C cross-linked polyacrylamide gels by dilution from the 30 % T stock. The final gel reagent mixtures are 7 μL 30 % T gel stock solution, 18.5 μL Solution B photoinitiator, 1.6 μL 10× TBE buffer, and 4 μL DI water. Before injecting this gel solution, the microchannel is first rinsed with Rain-X Anti-fog (SOPUS Products) followed by DI water. After injection, the polymerized gel interface is positioned by masking the glass surface above the injection ports with black tape. The microchip is then exposed to UV illumination for 1 min (intensity = 625 mW/cm2), after which the black tape is removed and un-polymerized gel solution in the channel is replaced with 0.5× TBE buffer. The microchip is then exposed to UV again for another 11 min to complete polymerization of the gel inside the separation channel (see Note 1).
3.2 Sample Focusing and Injection
Samples are prepared by combining 3 μL of the DNA sample, 3 μL of YOYO-1 intercalating dye (Invitrogen) diluted to one tenth of the stock concentration (except in experiments involving titration of YOYO-1) (see Note 2), 1.5 μL of β-mercaptoethanol, 1.5 μL of 10× TBE buffer, and 5 μL of an appropriate dilution from the daunomycin stock solution. The total volume is adjusted to 15 μL with DI water, and the mixtures are incubated for at least 30 min. Samples are then loaded into the microchip using a syringe and focused at the surface of internal electrode arrays by applying a low 1–1.2 V DC potential as described previously [10].
3.3 Microchip Gel Electrophoresis
After the sample is adequately focused, the potential is increased to 23.4 V DC across the separation channel to drive the DNA into the gel (1.56 cm distance between separation electrodes). A complete experiment begins by performing constant field electrophoresis, followed by electrophoresis under pulsed field conditions where the potential is switched on and off at periods ranging from 2 to 7 ms at 0.5 ms increments (see Note 3). At least 3 scans are performed to image the electrophoretic DNA motion under each field condition, and the displacements of the migrating zones between successive images are measured to determine the mobility. Two parameters can be tuned within each imaging scan: the scan interval (i.e., the time between successive scans) and the scanning length (i.e., the distance along the channel over which imaging is performed). Typical values in our experiments are a 90 s scan interval and 1.1 cm scan length.
3.4 Image Acquisition and Analysis
Individual snapshots acquired over the length of the separation channel during a single scan are joined together using Panavue software to produce a composite picture of the instantaneous i n-gel
18
Nan Shi and Victor M. Ugaz
band positions (see Fig. 1). A MATLAB code is then used to extract the intensity versus position profile along the center line of the separation channel in the corresponding electropherograms (the center line profile is used to avoid any sidewall band distortion). Each intensity peak is fitted to a Gaussian profile in order to obtain its center and variance for subsequent calculation of mobility and diffusion coefficients. 3.5 Simplified Model to Correlate Peak Period and Molecule Size
We have developed a transport model that describes electrophoretic transport of DNA in the ET regime [11]. Briefly, a trap time (dependent on gel pore morphology, electric field strength, and DNA size) is introduced to express the mechanism by which the DNA must overcome a local energy barrier to travel between neighboring pores in the gel: mi , j mN
=
t mig t mig + t trap
(1)
When the electric field driving force is switched on and off, the trap time becomes discontinuous and can be expressed in terms of a probability distribution composed of discrete intervals that collectively contribute to the overall electrophoretic mobility: mi , j mN
=
t mig
G 1 − exp − g off 2 t mig + G / 2 t mig G G 3G + g off − g on (2) exp − g off − exp − t mig + 3G / 2 2 2 2
Here, τmig = Rg/(μNE) = C1M(1 + υ)/E, γon, and γoff are denoted as Kramers escape rate (local energy trap) with and without electric field, which are closely related to molecule size (see Notes 4 and 5 for a detailed description of the governing equations and values of the constants). This local quantity is then integrated over the gel pore size distribution to obtain an overall mobility corresponding to the experimentally measured value. Both experimental data and numerical integration of Eq. 2 yield a maximum mobility at a specific value of the electric field switching period that satisfies the following resonance condition:
G g off = 1 2
(3)
A key point to notice is that the field-off escape rate γoff assumes a characteristic value for different-sized molecules in the same gel network and electric field conditions. This size-dependent resonance condition is the basis for our experimental approach.
Microchip Gel Electrophoresis to Probe DNA Binding
19
a
log µ
-4.5
300 500 600
-5
800 2
4
6
8
10
(ms) b 2
-4.7
0.25 0.50 0.75
-4.8
log µ
(nm2)
-4.9 -5 -5.1
1
3
5
7
9
11
(ms) Fig. 2 Stochastic resonance in gel electrophoresis. DNA fragments respond to the local energy barrier (related to macromolecular size and gel pore morphology) by displaying mobility peak at a size-dependent electric field actuation period Γmax. This behavior can be exploited by using the period at peak mobility to obtain information about the DNA molecule size. This is accomplished by constructing a calibration curve of mobility peaks obtained using a DNA ladder standard. (a) Measured mobility of a DNA ladder sample (300–800 bp) during electrophoresis in a 6.75 % T polyacrylamide gel under a pulsed electric field at different pulse times (1–10 ms) with a field time-averaged value of 15 V/cm. A sizedependent shift in the mobility peak is evident. (b) Our transport model reveals that the field-off Kramers escape rate plays a governing role in determining the resonance condition (Γmax/2)γoff = 1. The calculated γoff–1 are 2.7, 3.7, and 4.6 kBT for σ2 = 0.25, 0.5, and 0.75 nm2, respectively. The resonance conditions are roughly satisfied in three gel morphologies (obtained by applying different UV intensities during curing). Our model captures the effect of the gel morphology (i.e., the variance of the Gaussian pore size distribution, σ2) on the mobility peak. As the pore size becomes broader, the mobility peak becomes larger. Mobility of a 600 bp dsDNA fragment is shown in (b)
We first perform microchip gel electrophoresis across a series of electric field switching periods in order to identify the condition associated with a maximum electrophoretic mobility (see Fig. 2a). The resonance condition in Eq. 3 yields a value of γoff that then enables the molecule size (i.e., radius of gyration Rg) to be determined,
20
Nan Shi and Victor M. Ugaz
which is in turn a function of its persistence length (p) and contour length (L) [11]:
Rg =
1 p 2 pL 1 − 1 − exp ( −L / p ) 2 L
(4)
Figure 2b shows electrophoretic mobility data we obtained for a 600 bp dsDNA fragment. The native sample (Rg = 62.2 nm) displays a mobility peak at an electric field actuation period of about 5.5 ms. As the DNA size decreases, the period at maximum mobility becomes smaller. Our transport model is also able to quantitatively capture these experimental observations (see Fig. 2b). The mobility measured under constant field conditions enables us to extract the Kuhn segment number:
N k = L / 2p
(5)
This is accomplished by first establishing a μ versus Nk relationship using the native 100 bp dsDNA ladder sample (Nk = L/lk; L and lk are the contour and Kuhn segment lengths, respectively; L = 0.34 nm per base pair, lk = 2p = 100 nm for the native DNA) [12]. This calibration is then used to determine Nk from the constant field mobility of the bound complex (see Fig. 3). To demonstrate our method, we evaluated Eqs. 4 and 5 simultaneously to determine L and p for a 600 bp dsDNA fragment interacting with the binding agents daunomycin (an anticancer drug compound) and YOYO-1 (a widely used intercalating dye) (see Fig. 4). Compared with the native double-stranded DNA (p = 50 nm and L = 204 nm), our results suggest these compounds display different binding mechanisms. For example, our observations indicate that daunomycin tends to shorten the contour length and increase the DNA rigidity (higher persistence length), in apparent disagreement with other literature reporting that L increases [13]. We hypothesize that our observations may reflect alterations to the DNA structure associated with the presence of BME (see Subheading 2.1, item 5), an additive we introduce to inhibit photobleaching. Since BME functions as a strong reducing agent, the aldehyde group generated from oxidation of BME could induce covalent bonding between daunomycin and one of the dsDNA strands while maintaining hydrogen bonding with the other strand [14–16]. This formation of an adduct complex via a cross-linking mechanism other than hydrogen bonding (intercalation) may explain our observations of a more compact DNA structure. On the other hand, YOYO-1 appears to increase the DNA chain flexibility (smaller persistence length) while increasing the contour length, possibly reflecting the bis-intercalation binding mode. An important conclusion from these measurements is the fact that YOYO-1 does not appear to dramatically alter the DNA structure even at the highest loadings we tested, in agreement
21
max
log10 (µ)
– 4.598 – 4.596
(ms)
– 4.65
6
– 4.60
7
– 4.594
Microchip Gel Electrophoresis to Probe DNA Binding
5
2
4
6
8 10
2
500 bp 600 bp
4 4.5 5 5.5
pulse time (ms)
4 3
800 bp 700 bp
4.2 ms, 47.25 nm
400 bp
prediction
200 bp
experiment
300 bp
linear fit 25
35
45
55
65
75
85
Rg (nm) Fig. 3 Overview of electrophoretic DNA structure determination based on stochastic resonance phenomena. Measuring Γmax over a range of native DNA fragments of known length enables molecular size (Rg) of the bound complex to be extracted. Experimental measurements (crosses) are in agreement with our transport model predictions (open squares). Once this relationship is established (linear fit, dashed line), contour and persistence lengths of the DNA complex can be extracted from the Γmax data as described in the text. The inset shows evaluation of the transport model to locate the mobility peak associated with at 400 bp dsDNA fragment. The left half of the plot (dots) shows results from calculations over a coarse pulse time increment to locate the approximate position of the mobility peak. The right half of the plot (line) zooms in on the peak calculated using finer increments in pulse period for a more accurate determination
with previous experiments involving large DNA molecules. Our electrophoretically determined structural parameter measurements are also in good agreement with AFM data (see Fig. 5). 3.6 Conclusion
Although single-molecule imaging methods are inherently able to provide direct visual information about molecular structure, the sample being probed must necessarily be a large-sized macromolecule (e.g., λ-DNA). These techniques also require specialized equipment and expertise, making them inaccessible to many researchers. Gel electrophoresis, on the other hand, is a widely available technique employed in virtually any biological laboratory. A notable feature of the electrophoretically based approach is that it imposes no lower limit on the size of macromolecules that can be interrogated, enabling study of interactions between small DNA and binding agents that otherwise could not be probed at all. The microfluidic-based platform is also readily amenable for automation and parallel operation, offering a new and attractive avenue to perform high-throughput screening studies.
Nan Shi and Victor M. Ugaz Increased contour length
a
DNA backbone
Increasing binding agent concentration
Daunomycin Covalent bond Hydrogen bond
Decreased contour length
YOYO-1
b -4.4
DNA-daunomycin DNA-YOYO-1 Native DNA
log µ
22
-4.6
-4.8
-5 2
4
6
(ms)
8
10
Fig. 4 DNA binding tests with daunomycin and YOYO-1. (a) Schematic representation of the binding modes associated with each compound. (b) The resonant mobility peak is shifted to lower periods and higher values upon DNA-daunomycin binding, reflecting the more compact size of the bound complex. DNA-YOYO-1 tests reveal a shift in the opposite direction, indicating larger complex size. The DNA base pair concentration in all three samples is 80 μM. Daunomycin and YOYO-1 concentrations in the complexes are 60 μM and 100 μM, respectively
Fig. 5 AFM imaging directly confirms the electrophoretic measurements of conformational changes upon binding. Theoretical calculations yield the following contour and persistence lengths: L = 217 nm, p = 50 nm for native DNA; L = 144 nm, p = 53 nm for DNA-daunomycin; and L = 265 nm, p = 43.4 nm for DNA-YOYO-1. Average contour lengths of 220, 140, and 250 nm obtained from AFM are in good agreement. Representative AFM images are shown to illustrate the differences in contour length (500 × 500 nm field of view). Concentrations of all species are the same as in Fig. 4b
Microchip Gel Electrophoresis to Probe DNA Binding
23
4 Notes 1. The same total UV curing times are applied during gel casting even when different gel concentrations are employed. However, the duration of the initial illumination step to set the gel interface may vary depending on the UV intensity applied. 2. In order to record DNA transport without altering its internal structure, the sample is labeled with YOYO-1 at very low dye:bp ratio (1:5). Under these conditions, we find that the exposure time required to obtain clear images is at least 600 ms. 3. The operating parameters employed in the gel electrophoresis experiments (gel polymerization conditions, electric field strength and frequency, etc.) are selected based on application of our transport model to ensure that entropic trapping is the dominant transport mode within the macromolecule size range of interest [11]. First, the gel concentration is chosen such that the average pore size is comparable to sample molecule’s radius of gyration. Next, our transport model is applied to locate the corresponding electric field strength in the ET regime. Finally, the approximate activation time scale is calculated to determine the electric field actuation frequency range. 4. Numerical integration of the governing transport equations is performed using MATLAB, with upper and lower integration limits on the pore size distribution chosen based on previous hydrogel characterization studies [17]. −T DS − Felec L∗ 1 −1 5. (γoff)− 1 = C2 exp(−TΔS/kBT), (g on ) = C3 2 exp . E kBT Fitting constants are C1 = 1.5 × 10–6, C2 = 10–3, and C3 = 0.7, with ΔWS = –TΔS to characterize the energy barrier in hydrogel network: ΔWS = 0.85MkBT ln[(1/ri)1/υ − (1/rj)1/υ]. Here, υ Rg ~ M is fitted from Eq. 4, where M is the number of base pairs. The electrophoretic force Felec under an applied electric field of strength E is calculated as previously described [12], where ri and rj represent the pore sizes associated with one pore unit in our transport model with ri
1 Introduction Considerable efforts have been directed toward characterizing interactions between DNA and small molecule binding agents, owing to their potential usefulness as drug targets. Many of these compounds function by intercalation with DNA, binding to the major or minor groove of the double-stranded backbone [1–3], or via activity of substituent groups that enable external binding [4]. It is therefore of interest to understand and characterize the corresponding structural changes that are induced in the DNA complex upon these bindings. Some of the most commonly employed methods to probe binding interactions are based on experiments involving single DNA molecules, either via force measurement upon stretching with optical or magnetic tweezers or by direct imaging with atomic force microscopy (AFM) [5–7]. In the force measurement approach a DNA molecule large enough to be localized under optical Juan C. Stockert et al. (eds.), Functional Analysis of DNA and Chromatin, Methods in Molecular Biology, vol. 1094, DOI 10.1007/978-1-62703-706-8_2, © Springer Science+Business Media New York 2014
13
14
Nan Shi and Victor M. Ugaz
microscopy (typically λ-DNA) is stretched, and the corresponding force is measured as a function of displacement during both extension and relaxation. Care must be taken to ensure that the applied forces are small enough to maintain chemical equilibrium between the DNA and the bound drug compound, and to avoid introducing enthalpic effects that may distort the measurement. The resulting force-displacement data are then fit using the worm-like chain model Fp 1 x = 1 − kBT 4 L
−2
−
1 x + . 4 L
where F is the applied force, p is the persistence length of DNA– drug complex, x is the end-to-end displacement, L is the contour length of the complex, and kBT is the thermal energy (Boltzmann constant multiplied by the absolute temperature in Kelvin degrees) [8]. Although these experiments have played an instrumental role in providing fundamental insights about DNA–drug interaction mechanisms, they are delicate to perform and require sophisticated equipment (e.g., optical or magnetic tweezers) that is not readily available in all laboratories. The optically based format also inherently limits analysis to very-large-sized DNA molecules that can be directly visualized (i.e., contour lengths on the order of μm). We have recently explored DNA transport during gel electrophoresis in the entropic trapping (ET) regime, occurring when the average gel pore size is comparable to the size of the equilibrium random coil conformation of the DNA molecule. Our analysis revealed that a distinct mobility peak appeared at a specific actuation period when a periodically oscillating electric field was applied under conditions where ET-dominated transport occurs. We have developed a transport model that allows us to correlate the size of the DNA molecule with the position of the mobility peak. Since the molecular conformation of DNA is directly influenced by binding interactions, we are therefore able to apply an approach that uses microchip-based electrophoresis to gather structural information associated with DNA–drug complexation in a convenient format that circumvents the limitations of previous single-molecule methods.
2 Materials and Equipment Unless otherwise noted, all solutions are prepared using deionized (DI) water and stored at 4 °C. 2.1 Chemicals
1. DNA samples: 100 bp increment double-stranded DNA ladder (Cat # 170-8202; Bio-Rad; 100 μg/mL); single size 600 bp DNA fragment (NoLimits; Fermerntas; 500 μg/mL).
Microchip Gel Electrophoresis to Probe DNA Binding
15
2. Running buffer: 10× TBE (Cat # 161-0741; Bio-Rad). Dilute to 0.5× concentration. 3. DNA-binding compound: daunomycin hydrochloride (Cat # D8809; Sigma-Aldrich). Prepare an aqueous 1 mg/mL stock solution, and dilute to the desired concentration immediately before each experiment. 4. Intercalating dye: YOYO-1 iodide (Cat # 620722; Invitrogen). 5. Anti-photobleaching agent: 2-mercaptoethanol (BME) (Cat # M3148; Sigma-Aldrich). 6. Cross-linked polyacrylamide gel: Duracryl (30 % T, 2.6 % C; NextGen Sciences). 7. Photoinitiator: ReproGel Solution B (GE Health Care). 2.2 Microchip Gel Electrophoresis
Electrophoresis microchips (see Fig. 1) are fabricated following previously described methods [9]. The microchips consist of three primary components: (1) an etched glass microchannel (300 × 50 μm cross section) incorporating a separation channel with two side arms at one end for sample loading, (2) a silicon substrate containing an embedded electrode array, and (3) a PC board providing external connection to the on-chip electrodes (through wire bonding) via a 50-pin card edge connector and I/O block. The glass microchannels are bonded to the silicon substrate using UV-curable SK-9 Lens Bond (Summers Optical).
2.3 Imaging and Detection
Electrophoretic transport of the fluorescently labeled DNA is monitored using an Axioskop 2 microscope (Zeiss) with HBO 100 mercury arc lamp illumination, a fluorescein isothiocyanate (FITC) filter set, and a long working distance 10× objective. An ORCA-ER CCD camera (Hamamatsu) is used for image acquisition. A motorized x − y translation stage is employed to enable synchronized positioning of the microchip and actuation of the camera shutter (Openlab, PerkinElmer).
2.4 UV Curing
An Omnicure Series1000 spot curing system with collimating lens attachment (EXFO) is used for gel casting inside the microchip.
2.5 Function Generator
An Agilent 33220A function generator interfaced with a voltage amplifier (Trek Model 603) is used to generate waveforms for application of time-varying electric fields. Wave functions with different periods are designed using the Agilent IntuiLink Waveform Editor and then upload and stored in the device. During the experiments, the waveforms are monitored using a Hewlett-Packard 54603B oscilloscope.
16
Nan Shi and Victor M. Ugaz
a stationary microscope
sample motion direction t1 t2 t3
stage motion direction tn
t1
t2
t3
tn
b
Fig. 1 Experiment setup. (a) Overview of the image collection and analysis procedure using whole channel scanning detection. In each scan, a motorized stage carrying microchip enables a sequence of snapshots to be taken by a CCD camera interfaced with a fluorescence microscope. These individual snapshots are then assembled into a single composite image depicting all migrating DNA in the separation channel. A trace of fluorescence intensity versus position along the microchannel center line is then extracted for computation of mobility and diffusion coefficients. (b) Top view of an assembled electrophoresis microchip. The cross-sectional dimensions of all channels are 275 (width) × 45 (height) μm. An electrode array is patterned on the silicon substrate (the “floor” of the microchannel) to enable on-chip collection and focusing of DNA samples. The microchips are mounted on printed circuit boards that allow the on-chip electrodes to be individually addressed after wire bonding
Microchip Gel Electrophoresis to Probe DNA Binding
17
3 Methods 3.1 Gel Casting
For the sub 1 kb DNA lengths employed here, prepare 6.75 % T, 2.6 % C cross-linked polyacrylamide gels by dilution from the 30 % T stock. The final gel reagent mixtures are 7 μL 30 % T gel stock solution, 18.5 μL Solution B photoinitiator, 1.6 μL 10× TBE buffer, and 4 μL DI water. Before injecting this gel solution, the microchannel is first rinsed with Rain-X Anti-fog (SOPUS Products) followed by DI water. After injection, the polymerized gel interface is positioned by masking the glass surface above the injection ports with black tape. The microchip is then exposed to UV illumination for 1 min (intensity = 625 mW/cm2), after which the black tape is removed and un-polymerized gel solution in the channel is replaced with 0.5× TBE buffer. The microchip is then exposed to UV again for another 11 min to complete polymerization of the gel inside the separation channel (see Note 1).
3.2 Sample Focusing and Injection
Samples are prepared by combining 3 μL of the DNA sample, 3 μL of YOYO-1 intercalating dye (Invitrogen) diluted to one tenth of the stock concentration (except in experiments involving titration of YOYO-1) (see Note 2), 1.5 μL of β-mercaptoethanol, 1.5 μL of 10× TBE buffer, and 5 μL of an appropriate dilution from the daunomycin stock solution. The total volume is adjusted to 15 μL with DI water, and the mixtures are incubated for at least 30 min. Samples are then loaded into the microchip using a syringe and focused at the surface of internal electrode arrays by applying a low 1–1.2 V DC potential as described previously [10].
3.3 Microchip Gel Electrophoresis
After the sample is adequately focused, the potential is increased to 23.4 V DC across the separation channel to drive the DNA into the gel (1.56 cm distance between separation electrodes). A complete experiment begins by performing constant field electrophoresis, followed by electrophoresis under pulsed field conditions where the potential is switched on and off at periods ranging from 2 to 7 ms at 0.5 ms increments (see Note 3). At least 3 scans are performed to image the electrophoretic DNA motion under each field condition, and the displacements of the migrating zones between successive images are measured to determine the mobility. Two parameters can be tuned within each imaging scan: the scan interval (i.e., the time between successive scans) and the scanning length (i.e., the distance along the channel over which imaging is performed). Typical values in our experiments are a 90 s scan interval and 1.1 cm scan length.
3.4 Image Acquisition and Analysis
Individual snapshots acquired over the length of the separation channel during a single scan are joined together using Panavue software to produce a composite picture of the instantaneous i n-gel
18
Nan Shi and Victor M. Ugaz
band positions (see Fig. 1). A MATLAB code is then used to extract the intensity versus position profile along the center line of the separation channel in the corresponding electropherograms (the center line profile is used to avoid any sidewall band distortion). Each intensity peak is fitted to a Gaussian profile in order to obtain its center and variance for subsequent calculation of mobility and diffusion coefficients. 3.5 Simplified Model to Correlate Peak Period and Molecule Size
We have developed a transport model that describes electrophoretic transport of DNA in the ET regime [11]. Briefly, a trap time (dependent on gel pore morphology, electric field strength, and DNA size) is introduced to express the mechanism by which the DNA must overcome a local energy barrier to travel between neighboring pores in the gel: mi , j mN
=
t mig t mig + t trap
(1)
When the electric field driving force is switched on and off, the trap time becomes discontinuous and can be expressed in terms of a probability distribution composed of discrete intervals that collectively contribute to the overall electrophoretic mobility: mi , j mN
=
t mig
G 1 − exp − g off 2 t mig + G / 2 t mig G G 3G + g off − g on (2) exp − g off − exp − t mig + 3G / 2 2 2 2
Here, τmig = Rg/(μNE) = C1M(1 + υ)/E, γon, and γoff are denoted as Kramers escape rate (local energy trap) with and without electric field, which are closely related to molecule size (see Notes 4 and 5 for a detailed description of the governing equations and values of the constants). This local quantity is then integrated over the gel pore size distribution to obtain an overall mobility corresponding to the experimentally measured value. Both experimental data and numerical integration of Eq. 2 yield a maximum mobility at a specific value of the electric field switching period that satisfies the following resonance condition:
G g off = 1 2
(3)
A key point to notice is that the field-off escape rate γoff assumes a characteristic value for different-sized molecules in the same gel network and electric field conditions. This size-dependent resonance condition is the basis for our experimental approach.
Microchip Gel Electrophoresis to Probe DNA Binding
19
a
log µ
-4.5
300 500 600
-5
800 2
4
6
8
10
(ms) b 2
-4.7
0.25 0.50 0.75
-4.8
log µ
(nm2)
-4.9 -5 -5.1
1
3
5
7
9
11
(ms) Fig. 2 Stochastic resonance in gel electrophoresis. DNA fragments respond to the local energy barrier (related to macromolecular size and gel pore morphology) by displaying mobility peak at a size-dependent electric field actuation period Γmax. This behavior can be exploited by using the period at peak mobility to obtain information about the DNA molecule size. This is accomplished by constructing a calibration curve of mobility peaks obtained using a DNA ladder standard. (a) Measured mobility of a DNA ladder sample (300–800 bp) during electrophoresis in a 6.75 % T polyacrylamide gel under a pulsed electric field at different pulse times (1–10 ms) with a field time-averaged value of 15 V/cm. A sizedependent shift in the mobility peak is evident. (b) Our transport model reveals that the field-off Kramers escape rate plays a governing role in determining the resonance condition (Γmax/2)γoff = 1. The calculated γoff–1 are 2.7, 3.7, and 4.6 kBT for σ2 = 0.25, 0.5, and 0.75 nm2, respectively. The resonance conditions are roughly satisfied in three gel morphologies (obtained by applying different UV intensities during curing). Our model captures the effect of the gel morphology (i.e., the variance of the Gaussian pore size distribution, σ2) on the mobility peak. As the pore size becomes broader, the mobility peak becomes larger. Mobility of a 600 bp dsDNA fragment is shown in (b)
We first perform microchip gel electrophoresis across a series of electric field switching periods in order to identify the condition associated with a maximum electrophoretic mobility (see Fig. 2a). The resonance condition in Eq. 3 yields a value of γoff that then enables the molecule size (i.e., radius of gyration Rg) to be determined,
20
Nan Shi and Victor M. Ugaz
which is in turn a function of its persistence length (p) and contour length (L) [11]:
Rg =
1 p 2 pL 1 − 1 − exp ( −L / p ) 2 L
(4)
Figure 2b shows electrophoretic mobility data we obtained for a 600 bp dsDNA fragment. The native sample (Rg = 62.2 nm) displays a mobility peak at an electric field actuation period of about 5.5 ms. As the DNA size decreases, the period at maximum mobility becomes smaller. Our transport model is also able to quantitatively capture these experimental observations (see Fig. 2b). The mobility measured under constant field conditions enables us to extract the Kuhn segment number:
N k = L / 2p
(5)
This is accomplished by first establishing a μ versus Nk relationship using the native 100 bp dsDNA ladder sample (Nk = L/lk; L and lk are the contour and Kuhn segment lengths, respectively; L = 0.34 nm per base pair, lk = 2p = 100 nm for the native DNA) [12]. This calibration is then used to determine Nk from the constant field mobility of the bound complex (see Fig. 3). To demonstrate our method, we evaluated Eqs. 4 and 5 simultaneously to determine L and p for a 600 bp dsDNA fragment interacting with the binding agents daunomycin (an anticancer drug compound) and YOYO-1 (a widely used intercalating dye) (see Fig. 4). Compared with the native double-stranded DNA (p = 50 nm and L = 204 nm), our results suggest these compounds display different binding mechanisms. For example, our observations indicate that daunomycin tends to shorten the contour length and increase the DNA rigidity (higher persistence length), in apparent disagreement with other literature reporting that L increases [13]. We hypothesize that our observations may reflect alterations to the DNA structure associated with the presence of BME (see Subheading 2.1, item 5), an additive we introduce to inhibit photobleaching. Since BME functions as a strong reducing agent, the aldehyde group generated from oxidation of BME could induce covalent bonding between daunomycin and one of the dsDNA strands while maintaining hydrogen bonding with the other strand [14–16]. This formation of an adduct complex via a cross-linking mechanism other than hydrogen bonding (intercalation) may explain our observations of a more compact DNA structure. On the other hand, YOYO-1 appears to increase the DNA chain flexibility (smaller persistence length) while increasing the contour length, possibly reflecting the bis-intercalation binding mode. An important conclusion from these measurements is the fact that YOYO-1 does not appear to dramatically alter the DNA structure even at the highest loadings we tested, in agreement
21
max
log10 (µ)
– 4.598 – 4.596
(ms)
– 4.65
6
– 4.60
7
– 4.594
Microchip Gel Electrophoresis to Probe DNA Binding
5
2
4
6
8 10
2
500 bp 600 bp
4 4.5 5 5.5
pulse time (ms)
4 3
800 bp 700 bp
4.2 ms, 47.25 nm
400 bp
prediction
200 bp
experiment
300 bp
linear fit 25
35
45
55
65
75
85
Rg (nm) Fig. 3 Overview of electrophoretic DNA structure determination based on stochastic resonance phenomena. Measuring Γmax over a range of native DNA fragments of known length enables molecular size (Rg) of the bound complex to be extracted. Experimental measurements (crosses) are in agreement with our transport model predictions (open squares). Once this relationship is established (linear fit, dashed line), contour and persistence lengths of the DNA complex can be extracted from the Γmax data as described in the text. The inset shows evaluation of the transport model to locate the mobility peak associated with at 400 bp dsDNA fragment. The left half of the plot (dots) shows results from calculations over a coarse pulse time increment to locate the approximate position of the mobility peak. The right half of the plot (line) zooms in on the peak calculated using finer increments in pulse period for a more accurate determination
with previous experiments involving large DNA molecules. Our electrophoretically determined structural parameter measurements are also in good agreement with AFM data (see Fig. 5). 3.6 Conclusion
Although single-molecule imaging methods are inherently able to provide direct visual information about molecular structure, the sample being probed must necessarily be a large-sized macromolecule (e.g., λ-DNA). These techniques also require specialized equipment and expertise, making them inaccessible to many researchers. Gel electrophoresis, on the other hand, is a widely available technique employed in virtually any biological laboratory. A notable feature of the electrophoretically based approach is that it imposes no lower limit on the size of macromolecules that can be interrogated, enabling study of interactions between small DNA and binding agents that otherwise could not be probed at all. The microfluidic-based platform is also readily amenable for automation and parallel operation, offering a new and attractive avenue to perform high-throughput screening studies.
Nan Shi and Victor M. Ugaz Increased contour length
a
DNA backbone
Increasing binding agent concentration
Daunomycin Covalent bond Hydrogen bond
Decreased contour length
YOYO-1
b -4.4
DNA-daunomycin DNA-YOYO-1 Native DNA
log µ
22
-4.6
-4.8
-5 2
4
6
(ms)
8
10
Fig. 4 DNA binding tests with daunomycin and YOYO-1. (a) Schematic representation of the binding modes associated with each compound. (b) The resonant mobility peak is shifted to lower periods and higher values upon DNA-daunomycin binding, reflecting the more compact size of the bound complex. DNA-YOYO-1 tests reveal a shift in the opposite direction, indicating larger complex size. The DNA base pair concentration in all three samples is 80 μM. Daunomycin and YOYO-1 concentrations in the complexes are 60 μM and 100 μM, respectively
Fig. 5 AFM imaging directly confirms the electrophoretic measurements of conformational changes upon binding. Theoretical calculations yield the following contour and persistence lengths: L = 217 nm, p = 50 nm for native DNA; L = 144 nm, p = 53 nm for DNA-daunomycin; and L = 265 nm, p = 43.4 nm for DNA-YOYO-1. Average contour lengths of 220, 140, and 250 nm obtained from AFM are in good agreement. Representative AFM images are shown to illustrate the differences in contour length (500 × 500 nm field of view). Concentrations of all species are the same as in Fig. 4b
Microchip Gel Electrophoresis to Probe DNA Binding
23
4 Notes 1. The same total UV curing times are applied during gel casting even when different gel concentrations are employed. However, the duration of the initial illumination step to set the gel interface may vary depending on the UV intensity applied. 2. In order to record DNA transport without altering its internal structure, the sample is labeled with YOYO-1 at very low dye:bp ratio (1:5). Under these conditions, we find that the exposure time required to obtain clear images is at least 600 ms. 3. The operating parameters employed in the gel electrophoresis experiments (gel polymerization conditions, electric field strength and frequency, etc.) are selected based on application of our transport model to ensure that entropic trapping is the dominant transport mode within the macromolecule size range of interest [11]. First, the gel concentration is chosen such that the average pore size is comparable to sample molecule’s radius of gyration. Next, our transport model is applied to locate the corresponding electric field strength in the ET regime. Finally, the approximate activation time scale is calculated to determine the electric field actuation frequency range. 4. Numerical integration of the governing transport equations is performed using MATLAB, with upper and lower integration limits on the pore size distribution chosen based on previous hydrogel characterization studies [17]. −T DS − Felec L∗ 1 −1 5. (γoff)− 1 = C2 exp(−TΔS/kBT), (g on ) = C3 2 exp . E kBT Fitting constants are C1 = 1.5 × 10–6, C2 = 10–3, and C3 = 0.7, with ΔWS = –TΔS to characterize the energy barrier in hydrogel network: ΔWS = 0.85MkBT ln[(1/ri)1/υ − (1/rj)1/υ]. Here, υ Rg ~ M is fitted from Eq. 4, where M is the number of base pairs. The electrophoretic force Felec under an applied electric field of strength E is calculated as previously described [12], where ri and rj represent the pore sizes associated with one pore unit in our transport model with ri
