Real-Time Droplet DNA Amplification with a New Tablet Platform

Feb 6, 2012 - We present a novel droplet-based tablet platform for temporal polymerase chain reaction (PCR) in microliter droplets. The simple design ...
0 downloads 0 Views 419KB Size
Article pubs.acs.org/ac

Real-Time Droplet DNA Amplification with a New Tablet Platform Stephanie L. Angione,† Anuj Chauhan,‡ and Anubhav Tripathi*,† †

Center for Biomedical Engineering, School of Engineering and Division of Biology and Medicine, Brown University, Providence, Rhode Island, United States ‡ Department of Chemical Engineering, University of Florida, Gainesville, Florida, United States S Supporting Information *

ABSTRACT: We present a novel droplet-based tablet platform for temporal polymerase chain reaction (PCR) in microliter droplets. The simple design of the device does not require extensive processing or external equipment, which allows for greater ease of use and integration as a point-of-care diagnostic. We demonstrate its functionality to perform both PCR and reverse-transcription PCR for λ phage DNA and H3 influenza RNA with ramp rates and cycle times consistent with traditional PCR thermal cyclers. We additionally investigate the effect of performing PCR in small volumes on the reaction performance by specifically examining adsorption of reagents at the oil/water interface. We determined that adsorption of Taq polymerase at the biphasic interface reduces yield and impairs reaction performance at standard concentrations. Thus, microdroplet PCR reactions require additional polymerase to achieve sufficient amplification and we project for applications utilizing nanodroplets or picodroplets like digital applications, even greater concentrations of polymerase are required to achieve desired results. Following the adsorption investigation, we evaluated the sensitivity of λ phage PCR on our platform to be less than 2.0 copies/μL with an efficiency of 104.4% and similar sensitivity for reverse-transcription PCR for influenza H3 RNA.

T

passivation5,6 to lessen adsorption of reagents. Additionally, by carrying out the reaction in isolated droplets, PCR inhibition and carryover contamination can be mitigated.7 Droplet formats also allow for the possibility of increased throughput and sensitivity8,9 since droplets are discrete small volumes that can be easily adapted for parallel amplification at low concentrations. There are many examples of flow-through systems that utilize aqueous droplets surrounded in a nonaqueous oil or solvent. Several droplet PCR microdevices have been adapted from the traditional spatial thermal cycling format to include droplet-generating technologies for both three zones and two zones.9−12 There have been several pressure-driven flow systems that employ microdroplets, nanodroplets, or picodroplets8,13−15 to perform both PCR and reverse transcription (RT) PCR. These flow-through technologies have proved successful in generating highthroughput results as well as single-molecule sensitivity. Similar digital microfluidic platforms have used electrowetting technology to move microdroplets in place of pressure-driven systems,16 while other investigators have used actuating magnets to manipulate droplets containing magnetic beads to perform PCR temperature cycling or surface acoustic wave pump (SAW),17 to alleviate the need for pressure pumps.18,19 However, all of the above approaches are complex to create and

he ability to bring rapid diagnosis to developing countries has been an objective for many microfluidic technologies as well as a crucial challenge. There is a growing public health need for portable, low-cost diagnostic devices in resourcelimited settings, which presents many different design challenges.1 Rapid and sensitive diagnosis of diseases like human immunodeficiency virus (HIV), influenza, and bacterial infections by use of compact point-of-care devices has the potential to transform health care, in both hospital and field settings. Due to the typically low concentrations of pathogen nucleic acids in patient samples, many diagnostic assays depend upon DNA amplification though polymerase chain reaction (PCR), which exponentially amplifies the number of DNA molecules. Real-time monitoring of amplified DNA is critical for gene identification and mutation detection. Since PCR requires temperature cycling, it is typically carried out in benchtop units, which are slow and bulky and require relatively large volumes to perform the thermal cycling required. Recently, with the emergence of microfluidic PCR formats, smaller sample volumes are used and faster heating and cooling times have been achieved.2−4 Utilizing the concepts of both spatial and temporal microfluidic PCR, droplet-based microfluidic technologies have been emerging as alternatives to single-phase reactions. The advantage of droplet formats lies in low reagent usage and faster heat transfer and thermal equilibrium, as well as limiting interactions of channel walls with polymerase and DNA. Channel-based single-phase methods often require specific surface treatments or © 2012 American Chemical Society

Received: October 1, 2011 Accepted: February 6, 2012 Published: February 6, 2012 2654

dx.doi.org/10.1021/ac202532a | Anal. Chem. 2012, 84, 2654−2661

Analytical Chemistry

Article

nontrivial to operate. Many require continuous flow of liquids in microchannels, often driven by pressure pumps4,20 or utilizing valving,3,4,21 both of which add complexity to the device and operation. Additionally, fabrication of these devices is often time-consuming and expensive. Many require photolithography and wet etching methods for glass, quartz, or plastic materials.2,3,22 As an alternative to continuous flow technologies, several investigators have employed stationary droplet platforms23−28 or other forms of stationary droplets like the SlipChip to perform PCR.29,30 Although droplet technologies help mitigate the adverse affects due to the large surface to volume ratios associated with microfluidics, use of an immiscible fluid introduces a new interface for possible adsorption of reagents. Thus, despite the obvious use of droplet PCR methods, little has been reported regarding the transport and behavior in droplets surrounded by oil. The majority of publications have focused on hardware and methodology for droplet-based PCR, and less insight has been given regarding differences in performance or amplification efficiency. Some investigators have provided analyses of heat transfer properties and temperature distributions,10 but little has been reported regarding fundamental mass transport. In examining the results of representative droplet-based technologies, like the radial PCR device designed by Schaerli et al.8 that utilizes picoliter-sized droplets, there is an apparent difference between chip-based amplification and thermal cycler controls. The amplification factor of the device-based PCR compared to thermal cycler was up to 52-fold lower. This indicates that there is an inherent difference between droplet PCR and conventional single-phase PCR, but contributing factors such as adsorption of reagents have not been rigorously investigated. Some investigators have looked at altering parameters, such as Wang and Burns,27 who investigated nanoliter-sized droplet performance parameters including concentrations of Mg2+, Taq polymerase, and DNA template as well as temperature hold times and cycling conditions. Insight into the fact that there are factors at work in dropletbased microreactions was established but not fundamentally investigated. Since droplet in oil technologies are growing and have found a niche in point-of-care diagnostic development, it is useful to understand the principles that govern the system and how various parameters differ for droplet PCR compared to bulk reactions. Here we present a novel droplet-based tablet platform that utilizes low power and a straightforward setup, thus offering an easy-to-use PCR chip that offers greater simplicity over other methods. The simplicity of the device will make it easy to integrate into a point-of-care device for PCR diagnosis of numerous diseases. We demonstrate the platform’s usefulness in both standard DNA amplification and reverse transcriptase PCR in real time for identification of H3 influenza RNA. Additionally, we explore the inherent differences in the droplet PCR format compared to traditional PCR.

Figure 1. (a) Schematic of droplet platform: The ITO chip with the compound droplet is surrounded by an imaging spacer and covered with a coverslip. The droplet is placed at the center of the chip and covered in mineral oil. The spacer seals the compound droplet in a chamber with the coverslip to prevent evaporation. Each surface in contact with the droplet is covered with Teflon tape to prevent adsorption of DNA polymerase. (b) Overall control system: The entire droplet platform system, including the microscope and fluorescence acquisition, is shown. The droplet is mounted on a custom-made stage to accommodate connection to the voltage supply and cooling by a fan. The entire system is controlled by a LabView program that provides temperature feedback and fluorescence acquisition. (c) Temperature cycling profile for five representative cycles: Black line, PID-controlled surface temperature; red line, calibrated droplet temperature, measured by thermocouple.



EXPERIMENTAL PROCEDURES The New “Tablet” Platform. The tablet platform, as shown in Figure 1a, consists of a droplet of PCR mix surrounded by mineral oil, producing a disk-shaped compound droplet. A hydrophobic adhesive imaging spacer provides the cylindrical wall to create a tablet-like chamber for the PCR reaction, which has a diameter of 9 mm and is 1 mm thick. The bottom surface of the tablet is an indium tin oxide (ITO) coated soda-lime glass microscope slide. ITO is an optically

transparent angstrom-thick thin film that provides rapid and time-dependent resistive heating. The top surface of the tablet is an optically clear glass. The glass surface is covered with 2655

dx.doi.org/10.1021/ac202532a | Anal. Chem. 2012, 84, 2654−2661

Analytical Chemistry

Article

The primer set 5′-GATTGCCAGGCTTAAATGAGTC-3′ and 5′-GTTTCCGGATAAAAACGTCGAT-3′ was used. Primers were obtained from IDT. SYBR Green I dye (Applied Biosystems) was used for real-time fluorescence detection. Each 50 μL PCR mix consisted of 1× Taq buffer, 200 μM dNTPs, 0.4 μM both forward and reverse primers, 0.05 unit of Taq polymerase (unless otherwise noted for the polymerase investigation), a dilution of λ phage DNA template (500 bp), and 0.1× SYBR Green I. PCR was performed for 40 cycles, both on the PCR tablet and a control in a conventional thermocycler (Bio-Rad). Each cycle consisted of three stages: 95 °C for 10 s, 49 °C for 20 s, and 72 °C for 40 s. An initial denature was done at the beginning of cycling for 30 s, and a final extension was performed for 60 s. Sizing of the amplicon was done with the Agilent Bioanalyzer 2100. H3 Viral RNA Reverse Transcription Polymerase Chain Reaction. For influenza H3 RNA amplification, synthetic viral RNA (vRNA) of full-length H3 RNA was used as template. The sequence of the vRNA was identical to the strain with NCBI accession number AF348176. Primers specific for the individual strain were designed and include 5′-CTTTTAAGATCTGCTGCTTGTCCT-3′ and 5′-AGAAACAAACTAGAGGCCTATTCG-3′. Primers were obtained from IDT. Each RT PCR reaction consisted of 50 μL of RT-PCR mix from the Superscript III RT PCR kit, which includes Superscript III reverse transcriptase and Platinum Taq DNA polymerase (Invitrogen). Both reverse transcription and PCR were carried out in the same tube or on the droplet platform. This included the 1× reaction mix, containing buffer, MgCl2 and dNTPs, and 0.2 μM of both the forward and reverse primers, and 0.1× SYBR Green I dye was utilized for fluorescence quantification. Cycling consisted of a 30 min RT step at 50 °C, followed by 40 cycles of 94 °C for 10 s, 54 °C for 20 s, and 68 °C for 40 s. An initial denature was done following the RT step, 10 min for off-chip controls and 2 min for platform droplet amplification.

optically transparent Teflon fluorinated ethylene propylene (FEP) tape to prevent adsorption of PCR reagents to the surface of the glass. A Teflon surface was chosen on the basis of evidence that Taq polymerase does not adsorb to fluorinated surfaces like Teflon.31 On the ITO side of the chip, a thermocouple is placed at the center of the tablet. A droplet (1−4 μL) of PCR mix is placed at the center of the tablet on the FEP tape-covered side directly above and opposite the thermocouple. Four times as much mineral oil is then placed on top of the droplet. A glass coverslip also covered with FEP tape is then placed on the imaging spacer, effectively sealing the droplet of PCR mix inside the larger droplet of oil. Presently, the tablet platform is flipped and placed on a custom-made microscope stage with a small cooling fan. The PCR reaction can be monitored in real time from the top and bottom surfaces. The platform we present is unique and specifically advantageous in its inherent simplicity. It does not require any extensive fabrication methods, photolithography, or wet etching techniques, which are time-consuming and expensive and must be done in clean-room conditions with harsh chemicals. The fact that the PCR surface and heater are encompassed by the ITO-coated glass means that there are no external heating devices and no heater fabrication is necessary. Imaging is also extremely simple and straightforward since ITO is optically transparent. The platform is also very easy to use, both to assemble and to run PCR reactions. The computer interface is very user-friendly and easily adapted to many different cycling conditions or PCR reactions. Additionally, by implementing a fluorinated tape surface, which is removable between reactions, there are no specialized cleaning steps required between samples, and the removable surface eliminates sample carryover. By utilizing this platform with small volumes (1−4 μL), real-time PCR can be done utilizing a standard inverted microscope and photomultiplier tube (PMT) at a fraction of the cost of a specialized real-time PCR instrument without compromising sensitivity and improving the time required for temperature cycling. Instrumentation and Experimental Setup. Figure 1b shows the instrumentation required to operate the tablet platform. The ITO tablet platform is controlled by a customwritten proportional integral derivative (PID) temperature controller in LabVIEW (National Instruments). The thermocouple controls the temperature of the ITO surface by adjusting the voltage according to the feedback information. Temperature control was achieved via a computer with a data acquisition card (National Instruments) and USB voltage controllers from Matsusada Inc. Both temperature and fluorescence signals are collected during program operation. Since the ITO is deposited on only one side, the temperature of the reaction mixture is lower than the thermocouple reading on the ITO side. The temperatures were calibrated with a thermocouple placed in an oil droplet on the reaction surface. Materials and Preparation. Tablet Platform Preparation. Indium tin oxide (ITO) coated soda-lime glass microscope slides were obtained from SPI Supplies, and transparent Teflon fluorinated ethylene propylene (FEP) tape was obtained from McMaster. Hydrophobic adhesive imaging spacers were purchased from Grace Biolabs, and adhesive type E thermocouples from Omega. The cooling fan was obtained from Imbpapst and the coverslips from Fischer. λ Phage Polymerase Chain Reaction. A 106 bp amplicon of λ phage DNA (New England Biolabs) was amplified by PCR.



RESULTS AND DISCUSSION Temperature Cycling. Figure 1c shows the measured temperature inside the PCR droplet. The ITO tablet platform was used to perform temperature cycling for the PCR mix via the PID temperature controller. The temperature cycling program displayed is 95 °C for 10 s, 49 °C for 20 s, and 72 °C for 40 s. The maximum error of the system was found to be 1.3 ± 0.7 °C at the extension isothermal step. This is likely due to the fact that the droplet is heated through the glass surface, so error is generated in the feedback system. For the most temperature-dependent step, the annealing phase had an average error of 0.1 ± 0.4 °C, which is exceptionally low and better than most traditional thermal cyclers. The average cycle time is 117 ± 1.5 s, and the ramp rate for heating is 1.6 ± 0.2 °C/s and for cooling is 1.8 ± 0.1 °C/s. The ramp rate of traditional benchtop thermal cyclers is ∼2.0 °C/s. Although a representative five cycles is displayed, cycling was done out to 40 cycles for each PCR droplet. For RT-PCR for H3 RNA, the average error for the reverse transcription step was 0.8 ± 0.5 °C. The annealing step at 54 °C for the RT-PCR had a average error of 0.3 ± 0.5 °C. The measured temperature cycling is clear proof that the tablet platform can offer PCR cycling with ramp rates and cycle times consistent with or better than those observed for traditional bulky PCR thermal cyclers. DNA Amplification Controls. DNA amplification was carried out on the tablet platform with the cycling conditions 2656

dx.doi.org/10.1021/ac202532a | Anal. Chem. 2012, 84, 2654−2661

Analytical Chemistry

Article

Figure 2. continued volume, droplet control and tablet amplified samples: The gel plot inset displays the full-volume thermal cycler control with 0.01 ng/μL in lane 2; droplet and tablet samples in lanes 3 and 4, respectively; and full-volume, droplet, and tablet negative controls in lanes 5−7. The 106 bp amplicon is clearly visible in lanes 2−4, and the light band in lanes 5−7 represents unused primers. Concentrations of the amplicon in lanes 2, 3, and 4 are 292.9, 221.4, and 251.5 nM, respectively. The shading in lane 4 is background fluctuation.

previously described in a 2 μL droplet. These results were compared to a control with the same PCR mix in a conventional benchtop thermal cycler. Both a full-volume (50 μL) control and a droplet control were performed in the thermal cycler as described in Figure 2a. The droplet control consists of a droplet of the same experimental size covered in 4 times as much mineral oil amplified in a standard PCR tube. The electropherograms for the tablet, droplet control, and fullvolume samples are also shown in Figure 2a, which shows that amplification occurred for both the ITO tablet platform and the off-chip control for a DNA concentration of 0.01 ng/μL, all generating a similar yield with concentrations of 292.9, 221.4, and 251.5 nM for the full-volume, droplet control, and tablet samples, respectively. The peak displays the 106 bp amplicon that is also visible on the corresponding gel plot in Figure 2b from the Agilent 2100 bioanalyzer. Lanes 2, 3, and 4 display amplification for the full-volume, droplet control, and tablet PCR reactions, respectively. This also confirms that the in-tube droplet controls appropriately model amplification via the tablet platform. Negative controls were also performed for both the tablet platform and in-tube, which show no amplification of the 106 bp segment. In Figure 2b, lanes 5−7 display the corresponding negative controls for each platform. The negative control results are primers which have not been used in amplification. Overall, the tablet platform performs as well as off-chip thermal cycling with a fraction of the volume, at a faster speed. Real-Time Tablet PCR. Real-time PCR was conducted by fluorescence acquisition during the extension step of the PCR reaction with SYBR Green I. The characteristic qPCR amplification curve is shown in Figure 2b along with a negative control that displays no change in fluorescence, with minimal fluctuation in fluorescence of 0.006 (AU). The linear phase of amplification occurs very early for the DNA starting concentration of 0.01 ng/μL, with a calculated cycle threshold (Ct) value of 6.3 ± 1.0. The Ct value was found by determining the intersection where the fluorescence intensity crosses the critical background signal. The critical background signal was determined as 10 times the standard deviation of the background signal. Additionally, the slight decrease in fluorescence displayed at the plateau phase is attributed to photobleaching of the SYBR Green dye. However, the general trend of the PCR is as expected, demonstrating that the ITO platform is ideal for small-volume real-time PCR. Sensitivity and Efficiency. Although polymerase adsorption resulting in anemic cycles is an issue that must be addressed for droplet PCR, sensitivity and efficiency of amplification is not compromised on the droplet platform at appropriate polymerase concentrations. To determine the sensitivity of the tablet platform, serial dilutions of template concentrations were amplified via the tablet system. Figure 3a displays the real-time data for serial dilutions of 2.0 × 105, 2.0 ×

Figure 2. (a) Electropherograms of tablet, droplet control, and fullvolume PCR: time vs fluorescence plots. The peak near 38s represents the 106bp amplicon of the lambda phage PCR. The smaller earlier peak is the marker for the Agilent bioanalyzer DNA 1000 assay. It is evident that each sample was amplified sufficiently and within the same amplification factor based on DNA production. Baseline flucutation in the droplet electropherogram is an artifact of the instrument and represents a 2.6% baseline error. (Inset) Schematic of the droplet in oil control setup compared to a regular in-tube assay. The droplet control consists of a 2 μL droplet of PCR mix covered in 4 times as much mineral oil. Both the full-volume regular assay and the droplet control are then amplified by use of a traditional thermal cycler. (b) Real-time PCR of positive and negative controls: PCR amplification plot for 0.01 ng/μL with a Ct value of 6.7 cycles is displayed for on-chip amplification and the corresponding negative control, with no significant change in fluorescence. (c) Gel plot of full 2657

dx.doi.org/10.1021/ac202532a | Anal. Chem. 2012, 84, 2654−2661

Analytical Chemistry

Article

110%, demonstrating that the droplet system operates as well as conventional systems with very high sensitivity. Our system performs consistently if not better than other platforms, with the obvious advantage of remarkably high efficiency at the lowest functional polymerase concentration. For a comparison with other droplet systems, please refer to Table S-1 in the Supporting Information. Influenza H3 Reverse Transcription−Polymerase Chain Reaction. Clinically relevant concentrations of influenza RNA are 104−106 copies/ml for patient samples and therefore require highly sensitive assays for subtyping detection. We designed a reverse-transcriptase PCR to amplify a segment of H3 RNA for detection of RNA via the droplet PCR platform. From a starting concentration of 107 copies/μL, RT-PCR was carried out in a 2 μL droplet, and the real-time amplification plot is displayed in Figure 4. This template

Figure 3. (a) Real-time PCR plots for 1/10 dilution series of template: Threshold is 10× standard deviation of background. The lowest concentration amplified was 2.0 copies/μL, whose Ct was determined to be 28.8 ± 0.62 for triplicate experiments. (b) Efficiency plot of realtime data: log concentration value vs Ct values calculated from Figure 5. The data have a correlation of 99% with an intercept of 30.3 and a slope of −3.22, with a calculated efficiency of 104.4% from E = 10(−1/slope)−1.

Figure 4. Real-time reverse transcription PCR for H3 influenza: amplification plot for 107 and 102 copies/μL full-length influenza H3 RNA with corresponding Ct values of 15.1 and 29.

concentration was chosen to ensure amplification, but an additional experiment displays a sensitivity of 100 copies/μL RNA template in Figure 4. The results display an appropriate amplification curve with a Ct value of 15.1 for 107 copies/μL and 29 for 102 copies/μL. Our Ct values are similar to the amplification performed for H5 RNA with an amplicon of 114 bp in a 500 nL droplet utilizing a similar droplet system.26 The reported Ct value at a concentration of 2.4 × 107 copies/μL was approximately 15, and Ct was approximately 33 for 2.4 × 102 copies/μL. Role of Oil−Water Interface on PCR Reaction. The effect of droplet size was investigated to determine the effect of surface to volume ratio of the PCR reaction. There is a trade-off between the number of molecules of polymerase in the droplet and the amount of molecules that become adsorbed at the surface of the oil/water interface. Protein adsorption at the oil/ water interface is a well-known phenomenon as investigated by Beverung et al.32 It is therefore of interest to investigate the effect of interfacial adsorption of Taq polymerase in droplets for PCR, which has only been briefly examined.27 We investigated the effect of polymerase adsorption at the interface by examining various polymerase concentrations for three different droplet sizes, 1, 2, and 4 μL. These experiments were

104, 2.0 × 103, 2.0 × 102, 2.0 × 101, and finally 2.0 copies/μL. The critical background signal was calculated as described previously, and corresponding Ct values were found to be 12.4, 17.1, 20.0, 23.1, 26.2, and 28.8 for the series. Thus, it is apparent that single-molecule amplification is entirely possible with the droplet PCR system. Although our results are for short model DNA, high-sensitivity results are also possible for longer fragments of clinically relevant samples as displayed with the H3 RT-PCR. By plotting the Ct values for each of the serial dilutions, the efficiency of the PCR can be found. Figure 3b displays the efficiency of the reaction, with a linear regression fit of 99%. The slope of the line is −3.22, indicating the tablet PCR is 104.4% efficient. The higher than 100% efficiency indicates some error associated with creating dilutions, especially at low starting concentrations. We determined the lowest concentration to be subject to statistical fluctuations displayed by a Ct of 28.8 ± 0.62. Utilizing a slope of −3.3 for 100% efficiency, the efficiency was calculated as E = 10(−1/slope) − 1. Typical and acceptable efficiencies are between 90% and 2658

dx.doi.org/10.1021/ac202532a | Anal. Chem. 2012, 84, 2654−2661

Analytical Chemistry

Article

where D is the diffusion coefficient. When a typical protein diffusion coefficient of 5 × 10−7 cm2/s is utilized, adsorption of a monolayer occurs between 2.2 and 13.6 min for bulk enzyme between 0.05 and 0.02 unit/μL. Protein adsorption at an oil interface is attributed to a short induction time, when diffusion to the interface is important and the molecules irreversibly adsorb to the interface due to exposure of hydrophobic residues causing conformational changes.32 Equation 3 does not account for interfacial composition or hydrophobic interactions associated with an oil/water interface, so it is likely that the time required for a monolayer to form is more than the predicted values. To evaluate monolayer formation at the interface, the thermodynamic variable Γ for polymerase was calculated to be 0.0046 unit/mm2 or 3.65 × 108 molecules/mm2 from eq 2. Here, we have used the experimental data as shown in Figure 5 to evaluate CB ≈ 0.0074 unit/μL with C0 ≈ 0.027, 0.026, and 0.0215 unit/μL for the 1, 2, and 4 μL droplets, respectively. This was done by utilizing the bulk polymerase concentration associated with the same DNA production in the droplets as was achieved for the full-volume controls. A fixed DNA production concentration of 150 nM was utilized for evaluation. To verify this calculation, we assumed that the enzyme creates a monolayer at the biphasic interface and determined that the area occupied per molecule for a 2 μL droplet would be 2.74 × 10−9 mm2. In comparing this area per molecule to the area per molecule determined from the reported radius of gyration of native polymerase (38.3 Å), the area occupied is significantly smaller, only 1.88 × 10−10 mm2.33 However, it is known from the literature that the enzyme would not be in its native state but denatured at the interface with exposure of hydrophobic residues of the 832 amino acid chain, as it is accepted that proteins unfold at liquid/liquid interfaces.32,34 Studies of proteins at the oil/water interface account for the fact that the protein likely adopts different conformations. Specifically, at low concentrations, the protein/ oil interaction can lead to further unfolding of the protein structure.34 We can calculate the denatured Rg for Taq polymerase using the scaling law provided by Flory:35

done with the in-tube droplet controls, whose end-point DNA concentrations match the tablet-amplified droplets well as demonstrated in Figure 2a. The results displayed in Figure 5

Figure 5. Polymerase investigation: final DNA concentration after 40 cycles of PCR for droplet volumes of 1, 2, and 4 μL with enzyme concentrations ranging from 0.01 to 0.05 unit/μL. Each experiment was carried out in triplicate. The trends display a marked difference between full-volume (FV) reactions and droplet experiments.

display the final DNA concentration for each droplet at varying polymerase concentrations as measured on the Agilent 2100 bioanalyzer after 40 cycles of PCR. The data display the wellknown phenomenon of protein adsorption at the water/oil interface. The only volume that displayed amplification for all the polymerase concentrations examined, 0.01−0.05 unit/μL, was the full-volume 50 μL PCR controls. The effect of interfacial adsorption was pronounced for all the droplets investigated, although significant differences between the droplet volumes were not observed. The reason not much difference is observed between 1, 2, and 4 μL droplets is that the relevant controlling parameter is the surface to volume ratio, which is A/V ∼ 3/R for a spherical interface. Thus it is the radius, not the volume, that contributes and this does not vary appreciably in the small volume changes considered. The respective radii for 1, 2, and 4 μL droplets are 0.62, 0.78, and 0.98 mm. We examined the data points by developing the following mass balance equation to calculate Γ, the interfacial enzyme concentration: CBV = C0V − ΓS

R g = R 0N ν

where N is the number of residues, R0 is a constant related to persistence length, and ν is the scaling factor, which is dependent on solvent quality. If we utilize accepted values for R0 = 2.08 ± 0.19 Å and ν = 0.598 ± 0.029,36 we determine that the random coil radius of gyration for Taq polymerase is 116.0 ± 10.7 Å. The error for the radius of gyration and subsequent calculations was done by standard propagation of error formulas. For verification we utilized these values of R0 and ν for a smaller protein, phosphoglycerate kinase (PGK), whose native Rg is reported as 23 Å and unfolded Rg as 78 Å.37 Using the length of PGK as 417 residues, we determined the unfolded Rg for PGK to be 76.7 ± 7.1 Å, which correlates with the reported value. Thus, using the calculated unfolded Rg = 116.0 ± 10.7 Å, the area/molecule for Taq polymerase in our system is 1.69 × 10−9 ± 3.12 × 10−10 mm2, which provides a more reasonable approximation of the interfacial coverage of Taq polymerase. It would then seem that the enzyme is at least partially unfolded at the interface in a denatured conformation, as would be expected. It should be noted, however, that the values for R0 and ν are from correlations of small-angle scattering studies of proteins, which provide approximate values

(1)

where CB is the effective or bulk enzyme concentration, C0 is the initial enzyme concentration, V is the volume of the droplet, and S is the surface area of the droplet. If a spherical droplet shape is assumed, the mass balance reduces to Γ = d(C0 − CB)/3

(2)

where d is the diameter of the droplet. The loss of enzyme is assumed to occur early in the PCR reaction, most likely during the initial denaturation phase. This is evidenced by utilizing the following expression32 to calculate the time t for Taq polymerase to adsorb at the interface: Γ(t ) = 2CB Dt /π

(4)

(3) 2659

dx.doi.org/10.1021/ac202532a | Anal. Chem. 2012, 84, 2654−2661

Analytical Chemistry

Article

results require. It is also of interest that for a 150 nL droplet, a concentration of 0.175 unit/μL Taq polymerase was employed as optimal, which by utilizing our calculated Γ value and mass balance equation provides a bulk enzyme concentration greater (0.13 unit/μL) than required to accommodate loss at the interface.27 This indicates that for several of these technologies there may be additional adsorption to microchip surfaces. In order to reduce the relatively large concentration of polymerase required to achieve an appropriate yield for microliter to picoliter droplets, a blocking protein such as bovine serum albumin (BSA) can be added to the PCR master mix to occupy adsorption sites at the interface and reduce the loss of polymerase from the bulk.

for evaluation. Additional studies provide insight into the fact that different denatured states of proteins will demonstrate different Rg values, up to 10−25% different for certain proteins.38 Additionally, it is of interest that, at the oil/water interface, a decrease in interfacial area occupied by the protein was observed and thought to occur due to extension of the protein molecules and better solvency of the hydrophobic regions, resulting in greater extension into the oil phase.34 This observation for β-casein may account for our result of approximately 61.7% ± 11.4% monolayer formation at the polymerase concentrations utilized. These results indicate that, for droplet applications involving enzymatic reactions like PCR, higher concentrations of enzyme are required. The typical full-volume PCR reaction utilizes 0.02−0.025 unit/μL polymerase, whereas our results indicate that, for a 1 μL droplet to get maximum amplification, 0.04− 0.045 unit/μL is required, as displayed in Figure 5 and in greater detail in Figure S-1 (Supporting Information). It is apparent that DNA production plateaus near 0.02 unit/μL for the full-volume reactions, but this is not the case for the significantly smaller 1 μL volume. It is well-known that, at low enzyme concentrations, PCR results in anemic cycles.39 An anemic cycle refers to a cycle where Taq polymerase molecules are exceeded by primed templates, resulting in templates that do not get amplified. It is thought that anemia begins around cycle 15 for most protocols, which we project can be significantly lower for low enzyme droplets, resulting in either little or no amplification. For other droplet applications that utilize nanoliter or even picoliter droplets, even greater concentrations of polymerase are required. Our results indicate that the majority of adsorption of Taq polymerase likely occurs at the oil/water interface and not to the Teflon surface. The experiments displayed in Figure 5 were done without an adsorptive surface present, and our in-tube droplet controls match our on-chip amplification results very consistently (Figure 2). While there is evidence that Taq polymerase adsorbs to Teflon, the indicated results are for concentrations far above the operating parameters for PCR (0.3 mg/mL, or 24.0 units/μL) and the authors subsequently report Teflon as an optimal surface for droplet-based PCR.31 Additionally, with the hydrophobic nature of Teflon, it is likely that a thin layer of oil surrounds the disk-shaped droplet on both surfaces, due to the higher affinity the surface holds for oil over water and the working volume of oil as 4 times that of the PCR droplet. This affinity is affirmed by the contact angles for water and oil on Teflon surfaces, with water having a contact angle near 120° and oil near 60°.40 It has also been reported that, at the Teflon/oil/water/air interface, typically oil will displace water on the Teflon surface, resulting in water displacement from the solid so that the resulting interfacial interactions are limited to the oil/water interface.41 This would then indicate that adsorption for our tablet-based PCR occurs almost exclusively at the oil/water interface. However, it should be noted that adsorption to surfaces is a larger concern for channel-based microfluidic PCR, due to the fact that either droplets or a single phase is in greater contact with the adsorptive surface due to the high surface to volume ratio of these devices. Table S-1 (Supporting Information) displays a summary of several droplet PCR platforms with their representative sensitivities, efficiencies, and polymerase concentrations as provided. It is of note that most investigators are utilizing significantly higher enzyme concentrations than our optimized



CONCLUSIONS We present a simple high-efficiency PCR tablet platform capable of amplifying DNA with a starting concentration of 2.0 copies/μL and full-length influenza RNA with 100 copies/μL. Additionally, we determined that loss of polymerase at the biphasic interface in droplet-based techniques reduces amplification yield and sensitivity. These results highlight the potential of this device to be useful for point-of-care applications with the obvious requirement that the need for a microscope be eliminated. We have additionally highlighted a fundamental difference between droplet-based amplification and batch reactions that can help future designs for PCR microdroplet devices account for polymerase loss and optimize the amount required when utilizing microliter- to picoliter-sized droplets for PCR.



ASSOCIATED CONTENT

S Supporting Information *

One table, with a summary of droplet PCR methods and relevant parameters for comparison with our results, and one figure, with graphs of full-volume and 1 μL polymerase experiments that correspond with Figure 5 for further clarification. This material is available free of charge via the Internet at http://pubs.acs.org.



AUTHOR INFORMATION

Corresponding Author

*Email: [email protected].



ACKNOWLEDGMENTS A.T. acknowledges the support of the National Science Foundation (CBET 0653835), and S.L.A. acknowledges the financial support of the Brown University Graduate Fellowship and RISG NASA Graduate Fellowship for this research.



REFERENCES

(1) Chin, C. D.; Linder, V.; Sia, S. K. Lab Chip 2007, 7 (1), 41−57. (2) Dettloff, R.; Yang, E.; Rulison, A.; Chow, A.; Farinas, J. Anal. Chem. 2008, 80 (11), 4208−4213. (3) Lagally, E. T.; Medintz, I.; Mathies, R. A. Anal. Chem. 2001, 73 (3), 565−570. (4) Park, N.; Kim, S.; Hahn, J. H. Anal. Chem. 2003, 75 (21), 6029− 6033. (5) Slyadnev, M. N.; Lavrova, M. V.; Erkin, M. A.; Kazakov, V. A.; Ganeev, A. A. J. Anal. Chem. 2008, 63 (2), 192−198. (6) Shin, Y. S.; Cho, K.; Lim, S. H.; Chung, S.; Park, S. J.; Chung, C.; Han, D. C.; Chang, J. K. J. Micromech. Microeng. 2003, 13 (5), 768− 774. 2660

dx.doi.org/10.1021/ac202532a | Anal. Chem. 2012, 84, 2654−2661

Analytical Chemistry

Article

(7) Zhang, Y. H.; Ozdemir, P. Anal. Chim. Acta 2009, 638 (2), 115− 125. (8) Schaerli, Y.; Wootton, R. C.; Robinson, T.; Stein, V.; Dunsby, C.; Neil, M. A. A.; French, P. M. W.; deMello, A. J.; Abell, C.; Hollfelder, F. Anal. Chem. 2009, 81 (1), 302−306. (9) Markey, A. L.; Mohr, S.; Day, P. J. R. Methods 2010, 50 (4), 277− 281. (10) Mohr, S.; Zhang, Y. H.; Macaskill, A.; Day, P. J. R.; Barber, R. W.; Goddard, N. J.; Emerson, D. R.; Fielden, P. R. Microfluid. Nanofluid. 2007, 3 (5), 611−621. (11) Reichert, A.; Felbel, J.; Kielpinski, M.; Urban, M.; Steinbrecht, B.; Henkel, T. J. Bionic Eng. 2008, 5 (4), 291−298. (12) Link, D. R.; Kiss, M. M.; Ortoleva-Donnelly, L.; Beer, N. R.; Warner, J.; Bailey, C. G.; Colston, B. W.; Rothberg, J. M.; Leamon, J. H. Anal. Chem. 2008, 80 (23), 8975−8981. (13) Beer, N. R.; Hindson, B. J.; Wheeler, E. K.; Hall, S. B.; Rose, K. A.; Kennedy, I. M.; Colston, B. W. Anal. Chem. 2007, 79 (22), 8471− 8475. (14) Beer, N. R.; Wheeler, E. K.; Lee-Houghton, L.; Watkins, N.; Nasarabadi, S.; Hebert, N.; Leung, P.; Arnold, D. W.; Bailey, C. G.; Colston, B. W. Anal. Chem. 2008, 80 (6), 1854−1858. (15) Kumaresan, P.; Yang, C. J.; Cronier, S. A.; Blazei, R. G.; Mathies, R. A. Anal. Chem. 2008, 80 (10), 3522−3529. (16) Hua, Z. S.; Rouse, J. L.; Eckhardt, A. E.; Srinivasan, V.; Pamula, V. K.; Schell, W. A.; Benton, J. L.; Mitchell, T. G.; Pollack, M. G. Anal. Chem. 2010, 82 (6), 2310−2316. (17) Guttenberg, Z.; Muller, H.; Habermuller, H.; Geisbauer, A.; Pipper, J.; Felbel, J.; Kielpinski, M.; Scriba, J.; Wixforth, A. Lab Chip 2005, 5 (3), 308−317. (18) Tsuchiya, H.; Okochi, M.; Nagao, N.; Shikida, M.; Honda, H. Sens. Actuators, B 2008, 130 (2), 583−588. (19) Ohashi, T.; Kuyama, H.; Hanafusa, N.; Togawa, Y. Biomed. Microdevices 2007, 9 (5), 695−702. (20) Yu, C. Y.; Liang, W. S.; Kuan, I.; Wei, C. H.; Gu, W. G. J. Chin. Inst. Chem. Eng. 2007, 38 (3−4), 333−339. (21) Liu, R. H.; Yang, J. N.; Lenigk, R.; Bonanno, J.; Grodzinski, P. Anal. Chem. 2004, 76 (7), 1824−1831. (22) Koh, C. G.; Tan, W.; Zhao, M. Q.; Ricco, A. J.; Fan, Z. H. Anal. Chem. 2003, 75 (22), 6379−6379. (23) Kim, H.; Vishniakou, S.; Faris, G. W. Lab Chip 2009, 9 (9), 1230−1235. (24) Neuzil, P.; Pipper, J.; Hsieh, T. M. Mol. Biosyst. 2006, 2 (6−7), 292−298. (25) Neuzil, P.; Zhang, C. Y.; Pipper, J.; Oh, S.; Zhuo, L. Nucleic Acids Res. 2006, 34 (11), No. e77. (26) Pipper, J.; Inoue, M.; Ng, L. F. P.; Neuzil, P.; Zhang, Y.; Novak, L. Nat. Med. 2007, 13 (10), 1259−1263. (27) Wang, F.; Burns, M. A. Biomed. Microdevices 2009, 11 (5), 1071−1080. (28) Kim, H.; Dixit, S.; Green, C. J.; Faris, G. W. Opt. Express 2009, 17 (1), 218−227. (29) Ismagilov, R. F.; Shen, F.; Du, W. B.; Davydova, E. K.; Karymov, M. A.; Pandey, J. Anal. Chem. 2010, 82 (11), 4606−4612. (30) Ismagilov, R. F.; Shen, F.; Du, W. B.; Kreutz, J. E.; Fok, A. Lab Chip 2010, 10 (20), 2666−2672. (31) Prakash, A. R.; Amrein, M.; Kaler, K. V. I. S. Microfluid. Nanofluid. 2008, 4 (4), 295−305. (32) Beverung, C. J.; Radke, C. J.; Blanch, H. W. Biophys. Chem. 1999, 81 (1), 59−80. (33) Joubert, A. M.; Byrd, A. S.; LiCata, V. J. J. Biol. Chem. 2003, 278 (28), 25341−25347. (34) Maldonado-Valderrama, J.; Fainerman, V. B.; Aksenenko, E.; Galvez-Ruiz, M. J.; Cabrerizo-Vilchez, M. A.; Miller, R. Colloids Surf., A 2005, 261 (1−3), 85−92. (35) Flory, P. J. Principles of Polymer Chemistry; Cornell University Press: Ithaca, NY, 1953. (36) Fitzkee, N. C.; Rose, G. D. Proc. Natl. Acad. Sci. U.S.A. 2004, 101 (34), 12497−12502.

(37) Smith, L. J.; Fiebig, K. M.; Schwalbe, H.; Dobson, C. M. Folding Des. 1996, 1 (5), R95−R106. (38) Millet, I. S.; Doniach, S.; Plaxco, K. W. Unfolded Proteins 2002, 62, 241−262. (39) Mullis, K. B. Genome Res. 1991, 1, 4. (40) McCarthy, T. J.; Gao, L. C. Langmuir 2008, 24 (17), 9183− 9188. (41) Zettlemo, Ac; Aronson, M. P.; Lavelle, J. A. J. Colloid Interface Sci. 1970, 34 (4), 545.

2661

dx.doi.org/10.1021/ac202532a | Anal. Chem. 2012, 84, 2654−2661