Focused upon Hybridization: Rapid and High Sensitivity Detection of

Article pubs.acs.org/ac

Focused upon Hybridization: Rapid and High Sensitivity Detection of DNA Using Isotachophoresis and Peptide Nucleic Acid Probes Nadya Ostromohov,† Ortal Schwartz,‡ and Moran Bercovici*,†,‡ †

Faculty of Mechanical Engineering, TechnionIsrael Institute of Technology, Haifa 3200003, Israel Russell Berrie Nanotechnology Institute, TechnionIsrael Institute of Technology, Haifa 3200003, Israel

‡

Downloaded by TEXAS A&M INTL UNIV on September 5, 2015 | http://pubs.acs.org Publication Date (Web): August 25, 2015 | doi: 10.1021/acs.analchem.5b02547

S Supporting Information *

ABSTRACT: We present a novel assay for rapid and high sensitivity detection of nucleic acids without amplification. Utilizing the neutral backbone of peptide nucleic acids (PNA), our method is based on the design of low electrophoretic mobility PNA probes, which do not focus under isotachophoresis (ITP) unless bound to their target sequence. Thus, background noise associated with free probes is entirely eliminated, significantly improving the signal-to-noise ratio while maintaining a simple single-step assay requiring no amplification steps. We provide a detailed analytical model and experimentally demonstrate the ability to detect targets as short as 17 nucleotides (nt) and a limit of detection of 100 fM with a dynamic range of 5 decades. We also demonstrate that the assay can be successfully implemented for detection of DNA in human serum without loss of signal. The assay requires 15 min to complete, and it could potentially be used in applications where rapid and highly sensitive amplificationfree detection of nucleic acids is desired.

D

focusing of nucleic acids with molecular beacon probes has been demonstrated as a potential method for rapid sequence specific detection of DNA and RNA.22,23 However, the method suffers from a high noise floor associated with unreacted fluorescent probes, severely restricting the limit of detection. Several solutions for reduction of background noise have been recently demonstrated; Bahga et al. used bidirectional isotachophoresis (ITP) and a subsequent capillary zone electrophoresis step to separate the reaction products and achieved a limit of detection of 5 pM.24 Garcia-Schwarz and Santiago used DNA-functionalized hydrogels to capture excess probe and remove background signal and achieved a limit of detection of 2.8 pM with a 4 decade dynamic range.25,26 Eid et al. used a mobility spacer and sieving matrix to reduce the background signal and demonstrated a limit of detection of 220 fM with a 3.5 decade dynamic range.27 These methods, while effective, are based on co-focusing of the probes and targets in the ITP interface and thus require additional steps for elimination of excess (unhybridized) probes, which otherwise significantly limit the LoD. These separations add significant complexity to the setup in the form of in situ gel cross-linking, surface chemistry, or multiple assay steps. Carlsson et al. demonstrated that hybridization of DNA molecules to peptide nucleic acid (PNA) probes results in

irect detection of nucleic acid targets is typically achieved by hybridization with sequence specific probes, activating a transduction mechanism. For the majority of diagnostic applications, such detection is fundamentally limited by the low concentrations of the targets and the associated slow hybridization kinetics, typically on the order of several hours.1 PCR and other amplification methods are the gold standard in detection of nucleic acids, and they overcome reaction rate limitations by producing additional copies of the target sequence, thus driving the hybridization reaction to completion, whether in the bulk (e.g., with molecular beacons2−4) or on surfaces (e.g., with DNA arrays5−7). Amplification methods are by far the most sensitive and can achieve a theoretical limit of detection (LoD) as low as a single copy per mL.8 However, PCR reactions are time-consuming, suffer from an inherent amplification bias, require significant sample preparation, and a well-controlled environment.9−11 Furthermore, their use for detection of short targets, such as miRNA, presents additional challenges, such as additional extension steps required before amplification.12 At the same time, many practical applications do not require the sensitivity of a single copy.13−20 This results in a growing need for simpler and faster methods for sequence specific detection of nucleic acids, which are not based on amplification. Isotachophoretic focusing has been suggested as a method for simultaneous signal enhancement and acceleration of reaction kinetics, providing a 4 orders of magnitude improvement in reaction rates at pM concentrations.21 In particular, co© XXXX American Chemical Society

Received: July 7, 2015 Accepted: August 16, 2015

A

DOI: 10.1021/acs.analchem.5b02547 Anal. Chem. XXXX, XXX, XXX−XXX

Article


Analytical Chemistry

Figure 1. Schematic illustration of the assay. (a) A microfluidic channel connecting two reservoirs is initially filled with LE. The left reservoir is filled with a mixture of TE, DNA sample, and a high concentration of PNA probes. (b) In a control case, no targets are available to carry the probes into the ITP interface, and when an electric field is applied across the channel, all the probes remain in the TE zone. (c) In the presence of target, PNA probes rapidly bind to any matching DNA sequences. The negatively charged DNA and PNA-DNA hybrids electromigrate into the channel and focus at the ITP interface, while unbound, weakly charged PNA probes remain behind.

backbone at equal spacing to the DNA bases. This results in a weakly charged, chemically and biologically stable molecule, capable of sequence specific binding to DNA and RNA, offering higher stability and hybridization rates compared to standard DNA probes.30 PNA provides improved distinction between closely related sequences and increased discrimination against single point mutations. The decrease in melting temperature generated by a single mismatch in a PNA-DNA (or PNA-RNA) complex is significantly larger than the difference in melting temperature for the same mismatch in a corresponding DNA-DNA complex. Therefore, sequence discrimination is more efficient for PNA probes recognizing DNA than for corresponding DNA probes.31 PNAs were also shown to be resistant to proteases, peptidases, and nucleases,32 and they have a low binding affinity to proteins.33 Owing to its hybridization properties and specificity, PNA can be utilized as a highly selective biosensor for nucleic acid detection. Figure 1 presents a schematic illustration of the assay. We inject the sample and a high concentration of fluorescently labeled PNA probes into the TE reservoir of an anionic ITP setup, allowing probes to rapidly bind to any matching target sequences present. The buffer system is chosen such that the electrophoretic mobility of the TE is higher than that of the free (unhybridized) probes but lower than that of the PNA-DNA hybrids. Therefore, once an electric field is applied, excess (unbound) PNA probes remain behind in the TE zone, while the negatively charged PNA-DNA hybrids focus at the ITP interface, resulting in a fluorescent signal. Hence, a fluorescent signal is obtained only in the presence of target sequences “carrying” the PNA probes to the interface. This allows a highly sensitive, direct detection of target nucleic acids while completely eliminating background noise associated with unbound probes.

additional drag, thus allowing the separation of free and hybridized DNA under capillary electrophoresis.28 However, to the best of our knowledge, the electrophoretic properties of PNA sequences have not been further utilized for diagnostic applications. We here leverage this concept and present a novel assay (Principle of the Assay section), supported by a simple analytical model (Theory section), which combines PNA probes with selective ITP focusing for complete elimination of background fluorescence associated with unbound probes. This results in a simple, single-step assay for amplification-free sequence-specific detection of nucleic acids based on a homogeneous reaction (i.e., in the bulk). In Sensitivity and dynamic range section we characterize the assay and demonstrate its ability to provide a limit of detection of 100 fM and 5 decades of dynamic range. In Signal dependence on target length section we study the effect of target length on the signal, and demonstrate a detection threshold of 17 nucleotides (nt). Furthermore, in Experimental Section we provide a description of a methodology for application of the assay directly to serum samples, and in Detection in spiked serum section we demonstrate its use for detection of DNA sequences spiked in human serum, illustrating the compatibility with relevant biological samples.

■

PRINCIPLE OF THE ASSAY ITP is an electrophoretic separation and preconcentration technique. In peak mode ITP, analytes of interest are focused at the interface between a high electrophoretic mobility leading electrolyte (LE) and a low mobility trailing electrolyte (TE). The sample is typically mixed with the TE, and an initial interface between the LE and TE is established. In order to focus, sample ions must have an intermediate electrophoretic mobility between those of the LE and TE. When an electric field is applied, such ions overspeed the slow trailing ions and accumulate at the migrating LE-TE interface, creating a highly concentrated zone.29 In this assay, we couple ITP based focusing of nucleic acids with PNA probes. PNA is an artificial DNA analogue in which the natural negatively charged deoxyribose phosphate backbone has been replaced by a synthetic neutral pseudo-peptide backbone. The four natural nucleobases are retained on the

■

THEORY Electrophoretic mobility of DNA-PNA complexes. Consider a single stranded DNA target of length LD (measured in number of nucleotides) to which a PNA probe of length LP is hybridized. We denote the unhybridized section of the DNA as L, where L = LD - LP. Using a similar approach as used by B


Article

Analytical Chemistry Table 1. Details of the Buffer Solutions Used as the Leading and Trailing Electrolytes Composition Leading Electrolyte Counter Ion Trailing Electrolyte Counter Ion

Savard et al.,34 we model the PNA-DNA conjugate as a singlestranded DNA with a nonzero mobility drag-tag. The mobility of the complex can be modeled as the sum of two contributions: the single stranded DNA section and the dragtag section,


μPD = μD

L α 0 + μPD L+α L+α

6.5

μLE = −65.8 × 10−9 [m2 V−1 s−1]

6.8

μTE = −17 × 10−9 [m2 V−1 s−1]

(4)

Eq 4 represents the dependence of the amount of sample focused at the interface on the length of the target, assuming the condition in eq 3 is satisfied. For short targets, the electrophoretic mobility of the complex may be lower than that of the TE, but as the length of the target increases, so does the electrophoretic mobility of the complex, according to eq 1. When the mobility of the unhybridized section of the DNA target overcomes the drag added by the PNA probe, the mobility of the complex exceeds the mobility of the TE, leading to focusing of the complexes. The threshold length, L*, is obtained by equating eq 4 to zero, yielding

(1)

L* = α

0 μPD − μT

μT − μD

+ LP

(5)

For finite sample injection (a finite volume of sample injected between the LE and the TE zones),37 the focusing rate (eq 4) is irrelevant, as the total amount of focused sample at the interface is fixed by the initial conditions. Thus, all complexes containing targets above the threshold length, L*, will focus at the interface regardless of their length.

■

EXPERIMENTAL SECTION PNA probes. We use two types of PNA probes in the experiments. For the sensitivity and specificity experiments we designed a 14-mer PNA probe, complementary to a unique section of 16S rRNA of E. coli,40,41 Cy5-Lys-O-CGTCAATGAGCAAA-Lys (probe I), synthesized by Panagene (Daejeon, Korea). In order to maintain an overall neutral or positive charge of the probe, we used a positively charged Cyanine 5 (Cy5) dye as a label, and added lysines on either end to improve solubility. The choice of a cationic fluorophore is also important in ensuring that any free dye remaining from the synthesis process will not focus under ITP. We chose a dye with a far red emission spectrum (peak at 670 nm), as there is typically less background fluorescence at these wavelengths, and higher signal-to-noise ratio can be obtained.42 To demonstrate focusing of shorter targets we used a 17-mer 5-Carboxytetramethylrhodamine (TAMRA)-labeled PNA probe, without an addition of lysines on its residues, TAMRA-OO-ATTCGTTGGAAACGGGA (probe II), which is available as a catalog item from BioSynthesis (Lewisville, Texas). Similar to Cy5, TAMRA is a cationic dye, which has a single positive charge and does not focus under anionic ITP. Design of ITP buffers. While the PNA probes, in particular those with additional lysines, are expected to be neutral or positively charged, in practice we observed significant focusing of free PNA probes when using an anionic ITP chemistry having a very low mobility trailing ion. We thus empirically designed a buffer system which focuses PNA-DNA complexes but not free (unbound) PNA probes. For all experiments the LE buffer was composed of 200 mM HCl and 400 mM Bis(2-

(2)

where μa, μL, μT are respectively the effective mobilities of the analyte, the leading ion (in the LE zone), and the trailing ion (in the adjusted TE zone). σLE and σw are respectively the conductivities in the LE and in the well, A is the cross section area of the channel, x is the axial distance along the channel, and cwa is the concentration of the analyte in the well. From eq 2, the mobility of the TE should clearly be minimized in order to maximize the sample accumulation at the ITP interface. In theory, the PNA probes are expected to be neutral, and thus should not focus under any ITP chemistry. However, in practice, we observe that the PNA probes have a net negative charge. Therefore, focusing of PNA-DNA complexes while rejecting free PNA probes requires the design of a specific ITP system in which the trailing ions have a sufficiently high mobility to overspeed the free PNA probes, but not the PNA-DNA complexes. This condition can be summarized as μP < μT < μPD

Effective Mobility

⎛ ⎞ σ LE w L α 0 Na(x) = ⎜μD + μPD − μT ⎟ c Ax ⎝ L+α ⎠ μ σw a L+α L

where μD and μ0PD are respectively the free solution mobilities (in absolute value) of the DNA and of the hybridized PNADNA section (equivalent to the mobility of the complex when their length matches exactly, L = 0), and α is the effective friction coefficient representing the friction of the drag-tag in terms of number of DNA monomers. The mobility of a DNA molecule in free solution is essentially independent of DNA size, and equals the mobility of a single DNA monomer.35 In this work, we use an estimate of μD = 38 × 10−9 m2 V−1 s−1, based on measurements by Stellwagen et al.36 We obtain the values of α and μ0PD from experimental measurements. ITP focusing of DNA-PNA hybrids. The accumulation rate of the sample at the ITP interface in peak mode ITP with semi-infinite sample injection (sample mixed in the TE reservoir37) is determined by the influx of sample ions into the moving ITP interface. Assuming the mobility of the analyte is uniform across the reservoir and the adjusted TE zone, the amount of accumulated sample, Na, is given by38,39 ⎛ μ − μ ⎞ σ LE T ⎟⎟ w cawAx Na(x) = ⎜⎜ a μ ⎝ ⎠σ L

pH Value

200 mM HCl 400 mM Bistris 10 mM MES 20 mM Bistris

(3)

where μP is the effective mobility of the free PNA probes. Substituting the complex’s mobility into eq 2 and rearranging, we obtain C


Article



Figure 2. Experimental results showing the sensitivity, dynamic range, and specificity of DNA detection using fluorescently labeled PNA probes. (a) Sensitivity and dynamic range for initial concentrations of target between 100 fM and 100 nM. Each bar represents an area average of the intensity profile of the fluorescent signal registered 18 mm from the TE reservoir. We injected into the TE a fixed concentration of 10 nM of PNA probes, and varied the concentration of a 200 nt DNA target between 100 fM and 100 nM. In control case I, no targets or probes were added to the reservoir. When probe was added but no targets were present (control II), the signal was not affected. The results demonstrate a linear increase in signal (log S = 1.23 log cPD + 19.0, R2 = 0.99) over a dynamic range of 5 decades (100 fM to 10 nM) with a limit of detection of 100 fM. At target concentration of 100 nM, no more probes are available to react, and the signal saturates. (b) The signal was calculated by integrating the area under the curve for intensity values greater than 20% of the peak. (c) Specificity demonstration for complementary and random targets 120 nt in length. The control bar represents no target and 10 nM probe added to the TE. Initial concentration of targets is 10 pM. All error bars correspond to 95% confidence on the mean using at least 5 repeats.

the channel by flowing 3.5% bleach (sodium hypochlorite) and 1 M NaOH through the channel for 5 min each. For experiments with human serum, we used a finite injection of sample introduced between the LE and TE. Figure 4a shows a schematic of the finite sample injection ITP experiments. We placed the sample into the West reservoir and applied vacuum to the South reservoir to obtain a finite sample injection of 4 nL. We vacuumed the content of the West reservoir, filled it with 10 μL TE, and applied a voltage of 600 V between the West and East reservoirs. For the positive serum we spiked 100 nM DNA into human serum (Sigma-Aldrich, St. Louis, MO) containing 1 μM PNA and incubated the mixture for 15 min. No DNA was spiked into the serum in the negative serum experiments. For the treated positive serum experiments, 4% proteinase K (Life Technologies, Grand Island, NY) and 2 mM EDTA (Sigma-Aldrich, St. Louis, MO) were added to the serum before incubation at 37 °C. In the positive buffer experiments the injection consisted of LE containing 1 μM PNA, spiked with 100 nM DNA. Experimental setup. We use a high sensitivity photomultiplier tube (PMT)-based point detection system similar to the setup described by Bercovici et al.45 Briefly, our experimental setup consists of a microfluidic chip, laser excitation, high voltage supply and a custom point confocal fluorescence system (see Figure S-1 in the SI). We performed all experiments using an Eclipse-Ti inverted microscope (Ti−U, Nikon Instruments, Melville, NY), on a commercially available microfluidic chip (NS95x, PerkinElmer, Waltham, MA). For

hydroxyethyl)-amino-tris(hydroxymethyl)-methane (bistris). The TE buffer was composed of 10 mM 4-Morpholineethnesulfonic acid (MES) and 20 mM bistris. We used 1% of 1.3 MDa polyvinylpyrrolidone (PVP) in the LE for suppression of electroosmotic flow (EOF),43 and 30% acetonitrile in the TE to improve PNA solubility. All chemicals were obtained from Sigma-Aldrich (St. Louis, MO). Table 1 summarizes the composition of the buffers, the pH values for the LE and TE zones, and the resulting effective mobility of the leading and trailing ions, as numerically calculated using SPRESSO.44 A list of target sequences, detailed considerations in setting probe concentrations, and a description of the experimental setup are all provided in details in the Supporting Information (SI) Sections S-2-S-4. Assay description. For each experiment we initially filled the microfluidic channel with LE, by filling the LE reservoir with 10 μL of LE and applying vacuum to the TE reservoir. We spiked DNA targets at concentrations between 100 fM and 10 nM into a TE buffer containing 10 nM of fluorescently labeled PNA probes. We incubated the mixture for 15 min at 37 °C, and injected 10 μL of it into the TE reservoir, after rinsing it with deionized water (DI). We then applied a voltage of 600 V across the channel using a high voltage sourcemeter (2410, Keithley Instruments, Cleveland, OH, USA). We used Matlab (R2012b, Mathworks, Natick, MA) to control the PMT and the sourcemeter and record the data during the experiments. The detector was located at a distance of 18 mm from the TE reservoir. Between experiments we cleaned D


Article


Analytical Chemistry experiments with probe I we used a Cy5 filter-cube (41008, 620/60 nm excitation, 700/75 nm emission, and 660 nm dichroic mirror, Chroma, Bellows Falls, Vermont, USA), and a water immersion objective (60x, CFI Fluor, N.A. = 1.0, Nikon Instruments, Melville, NY). We used a 640 nm 40 mW laser (Cube, Coherent Inc., Santa Clara, CA) to excite the Cy5 labeled PNA probes. We expanded the laser beam to 8 mm diameter, and delivered it to the objective through the microscope’s photoactivation port. We used an 800 μm pinhole located at the image plane to reject out of plane light, and collected the light on a PMT (H6780−20, Hamamatsu Photonics, Hamamatsu, Japan) connected to the computer through an A/D converter (C8908, Hamamatsu Photonics, Hamamatsu, Japan). Using a set of known dye concentrations (Dylight650, Thermo Scientific, Waltham, MA), we characterized the sensitivity of the PMT at each operating voltage. In experiments, we normalized any PMT signal values by the sensitivity level of its operation voltage. This allows comparison between data points taken using different operating voltages. For experiments with probe II, and finite injection ITP, we used a metal halide light source (Intensilight, Nikon, Japan), a TRITC filter-cube (49004, 545/30 nm excitation, 620/60 nm emission and 565 nm dichroic mirror, Chroma, Bellows Falls, Vermont, USA), and a 20x objective (CFI Plan Fluor, NA = 0.5, Nikon Instruments, Melville, NY). The images were captured using a CCD camera (iXon3, Andor Technology, Belfast, NIR, UK), and processed using MATLAB (R2012b, Mathworks, Natick, MA).

Figure 3. The electrophoretic mobility of the complex increases with the length of the target, resulting in a higher influx into the ITP interface and thus higher signals for longer targets. Probe I contains (positively charged) lysine groups, which lower the mobility of the target and result in a higher threshold of detectable target lengths (50 nt) compared to that of probe II (17 nt). We used fixed target concentrations of 10 pM for targets complementary to probe I and 10 nM for targets complementary to probe II. For each of the data sets, we use eq 4 to extract the values of α and μ0PD. Dashed lines correspond to the model when using these parameters. The quality of the fit indicated the model is well suited describing the effect of PNA drag and target length on the signal. The parameters obtained from the experimental results yield an α of 189 and μ P0D of 12.8 × 10−9 m2 V−1 s−1 for probe I, and an α of 118 and μ0PD of 16.7 × 10−9 m2 V−1 s−1 for probe II.

■

RESULTS AND DISCUSSION Sensitivity and dynamic range. Figure 2a presents experimental results demonstrating the limit of detection, sensitivity, and dynamic range of the assay. Each data point in the figure corresponds to the area under the curve of the registered signal for intensity values greater than 20% of the peak, as illustrated in Figure 2b. While the peak value of the fluorescence signal may depend on the dispersion of the ITP interface (for example due to nonuniform EOF46), the area under the curve indicates the total amount of sample accumulated at the interface and is not sensitive to such disturbances, making it a more robust figure of merit. The choice of a fixed threshold of 20% of peak value is rather conservative (some of the area is lost), but it allows robust evaluation of the area even for noisy signals and, as shown in the figure, results in a consistent linear signal across the range of concentrations (until saturation). We performed two control experiments: In control case I no targets or probes were present in the reservoir. The signal obtained in this case is the baseline signal corresponding to detectable inherent contamination in the buffers. In control case II, we added a high, 10 nM, concentration of fluorescently labeled probes to the TE. Despite the high initial probe concentration, the signal detected in this case is similar to control case I, confirming that the chosen ITP chemistry prevents free PNA probes from focusing at the ITP interface. As shown in Figure 2a, the addition of 100 fM targets results in a distinguishable signal above the baseline of the control cases and defines the LoD of the assay for a 200 nt long target (yielding a two sample t test p-value of less than 0.01). As we increase the concentration of the target, the detected signal increases proportionally up to a concentration of 10 nM, indicating that the number of PNA probes delivered to the ITP interface is proportional to the number of DNA targets present.

This provides a quantitative measure, with a dynamic range of 5 orders of magnitude in target concentration. At target concentrations higher than the initial 10 nM concentration of the probe, the signal saturates, indicating that the reaction is complete and there are no more probes available for the hybridization reaction. We emphasize that saturation is due to the limited number of PNA probes in the reservoir (10 nM), and not due to sensor saturation; as described in the Experimental Section, at each concentration the PMT sensor was operated at an appropriate voltage to avoid its saturation. Figure 2c presents experimental results verifying the specificity of the assay at room temperature, using a 10 nM probe concentration. We performed the experiments using complementary and random targets of equal lengths (120 nt) at a fixed concentration of 10 pM, and we compared those to control II (10 nM of PNA probes and no DNA). While the complementary sequence results in a signal which is 2 orders of magnitude higher than the control, the signal obtained for the random targets cannot be distinguished from the control. Signal dependence on target length. The sensitivity experiments in the previous section were performed using a 200 nt DNA target. However, as discussed in the Theory section, the electrophoretic mobility of the DNA-PNA complex is expected to reduce with shorter target lengths. This results in a threshold length below which the drag added by the PNA probe hybridized to the target is too high, and the complex mobility falls below that of the trailing ion. Characterization of E


Article



Figure 4. Demonstration of the assay’s compatibility with human serum. (a) Schematic illustration of the channel geometry and of the setup for finite sample injection. The microfluidic channel is isotropically etched, and it has a uniform depth of 12 μm. The length (L) and width (W) of each of its sections are presented in the figure. We initially filled the channel with LE, placed the sample into the West reservoir, and applied vacuum to the South reservoir to fill the West channel with sample. We then vacuumed the content of the West reservoir, filled it with TE, and applied a constant voltage of 600 V between the East and West reservoirs. (b) Experimental results showing the obtained signals. The untreated positive serum contained human serum spiked with 1 μM PNA and 100 nM DNA. The treated positive serum sample also contained 4% proteinase K and 2 mM EDTA. No DNA was added in the negative serum case, and the positive buffer experiments served as a positive control in which the serum was replaced by LE buffer. (c) Electric current traces for each of the experiments showing the repeatability of the experiments using serum.

In some applications, the DNA may be fragmented causing the target sequence of interest to be part of DNA strands of different lengths. For long enough fragments (above 100 nt), the signal is independent of the length, and it allows quantitation of the initial concentration of the target. However, the presence of shorter fragments would hinder the ability to accurately quantify this concentration, due to their different accumulation rates. This limitation can be overcome by injecting a finite amount of sample between the LE and the TE zones. In this configuration the focusing rate is no longer important, and all fragments satisfying the threshold condition given by eq 5 would focus and contribute equally to the signal. Detection in spiked serum. Our characterization experiments (Sensitivity and dynamic range sections and Signal dependence on target length sections) were performed in ideal buffer conditions. To show the feasibility of applying the assay to a real sample, we here demonstrate detection of DNA directly from human serum, and compare the obtained signals to those of a clean sample. We chose to use a finite injection scheme, which allows us to process undiluted serum despite its relatively high concentration of strong acids which would disrupt ITP if injected in the TE.47 Figure 4a presents a schematic of the finite injection scheme used in this set of experiments and lists the relevant dimensions of the channel. Figure 4b presents the signal obtained for four sets of experiments: a negative control composed of a serum sample with PNA probes but no DNA targets (“Negative Serum”), an untreated serum sample containing both PNA and DNA (“Untreated Positive Serum”), a serum sample containing both

this threshold, as well as the dependence of the signal on target length, is important in determining the range of targets for which this assay could be applicable. Figure 3 presents the signal obtained in ITP as a function of the length of the target, for the two probes. In all experiments, the target concentration is 10 pM for targets complementary to probe I, and 10 nM for targets complementary to probe II, regardless of their length. We present the values after subtracting the signal value of the control case, and normalizing by the maximum value (obtained for the longest target). As expected, the electrophoretic mobility of the complex increases with the length of the target, resulting in a higher influx into the ITP interface and thus higher signals for longer targets. Importantly, we show the assay is able to detect sequences as short as 50 nt using probe I, and 17 nt for probe II. We attribute this difference in the detectable target length thresholds to the additional lysine groups present in probe I; while this chemical modification improves the solubility of the probe, it also adds significant drag and lowers the complex mobility, resulting in a higher friction coefficient, α. Notably, the experimental results follow the trend of eq 4, validating the relation between the mobility and the length of the target provided in eq 1. The fitted parameters obtained from the experimental results yield an α of 189 and μ0PD of 12.8 × 10−9 m2 V−1 s−1 for probe I, and an α of 118 and μ0PD of 16.7 × 10−9 m2 V−1 s−1 for probe II (full details on the fitting procedure are provided in SI section S-5). F


Article



longer sequences. Furthermore, the use of PNA probes reduces the dependence of the hybridization on the ionic strength of the sample, a problem commonly encountered with DNA probes.51 The ability to avoid amplification processes, which could be very sensitive to contaminations and amplification biases, together with the ability to detect very short sequences, which are difficult to amplify, could render this technique useful even if sample purification is required. However, direct detection from biological samples remains the ultimate goal. We demonstrated in this work the applicability of the assay to human serum, and we showed that signal levels can be equal to those obtained in a clean sample, provided that the sample is treated with proteinase K for degradation of nucleases and that a finite injection scheme is utilized. The limit of detection in biological samples is expected to be related to the focusing of autofluorescent proteins which would set a higher background value. Further work should be performed to quantify these limitations, but they may be overcome by the use of timeresolved fluorescence or reporting molecules such as quantum dots, whose excitation and emission spectra are significantly different than those of natural organic species. In this work we performed the assay at room temperature, and we demonstrated specificity only for a random sequence. However, since mismatches in PNA-DNA hybrids are known to generate a significant difference in the melting temperature of the complex,52 single nucleotide specificity likely can be achieved by maintaining an elevated temperature on-chip. Furthermore, as the PNA probe is capable of specifically binding to RNA as well as to duplex DNA,53,54 this assay may also be expanded for detection of RNA and dsDNA targets.

PNA and DNA treated with proteinase K and EDTA (“Treated Positive Serum”), and a positive control with LE buffer, PNA, and DNA (“Positive Buffer”). Importantly, no signal was observed in the negative serum sample, and the signal value we present is the standard deviation of the background. The untreated positive serum sample provided a clear signal above the background. However, it is lower by an order of magnitude compared to the positive buffer reference. This loss of signal can be attributed to the presence of nucleases in serum, resulting in partial degradation of the DNA,48,49 or nonspecific binding of DNA to proteins,50 competing for the hybridization reaction between the DNA and PNA. To prevent this loss of signal, we added to the serum sample 4% proteinase K (to digest proteins) and 2 mM EDTA (to decrease DNase activity), and incubated at 37 °C for 15 min. Indeed, the signal obtained for the positive treated serum is similar to the signal obtained for the clean buffer, indicating the assay can be successfully implemented for detection in serum without loss of signal. Electric current traces are a convenient independent method for quantifying the repeatability and robustness of an ITP assay. Figure 4c presents 9 overlaid electric current traces of the experiments performed with serum, showing good repeatability of the assay regardless of the proteinase K and EDTA treatment.

■

CONCLUSIONS We demonstrated a new assay leveraging the neutral backbone of PNA to achieve rapid and high sensitivity detection of nucleic acid sequences by ITP focusing. We showed that the use of PNA-based probes enables complete elimination of the signal of unhybridized probes in the ITP interface, significantly improving the limit of detection and extending the dynamic range. Furthermore, we demonstrated the applicability of the assay in human serum using a finite sample injection scheme. The most sensitive ITP-based assays demonstrated to date are those of Garcia-Schwarz et al. and Eid et al.26,27 GarciaSchwarz used a functionalized polyacrylamide region to retain excess probes, thus cleaning the ITP interface from residual signal. As the separation process is based on hybridization of the probes to matching sequences in the gel, the method could potentially be applied to targets of any length (including very short, 22 nt targets, as demonstrated by the authors). However, its main disadvantage is its reliance on in situ cross-linking of polyacrylamide in the channel, which adds significant complexity that may not be suitable for point-of-care analysis. In contrast, the method by Eid et al. uses a sieving matrix to separate between free and hybridized probes. Its advantage is in performing both the reaction and separation in the bulk, without requiring surface functionalization. However, since removal of free probes is based on separation by size, it is inherently limited to target lengths which are sufficiently longer than the probe. The main advantage of the assay presented here is its simplicity, as it requires only a straight channel, makes use of homogeneous reactions occurring in the bulk, and does not require a sieving matrix or surface functionalization of any kind. Yet, despite its simplicity, it provides a more than 2-fold improvement in LoD and a more than 10-fold improvement in dynamic range over these methods, while being directly applicable (without any chemistry modifications) to target lengths between 17 and 200 nt. Based on the fact that the free solution mobility of DNA remains essentially constant for longer sequences, we expect the assay to be equally effective for

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.analchem.5b02547. Theory of PNA-DNA hybridization, details of sequences of PNA probes and DNA targets used in the experiments, considerations in setting probe concentration, experimental setup, and fitting procedure (PDF)

■

AUTHOR INFORMATION

Corresponding Author

*Building 360, Room 406, Faculty of Mechanical Engineering, TechnionIsrael Institute of Technology, Technion City, Haifa 3200003, Israel. Tel: +972-4-829-3463. E-mail: mberco@ technion.ac.il. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS We thank Rebecca Khalandovsky for performing preliminary experiments in serum and for fruitful discussions. We gratefully acknowledge funding from the Israel Science Foundation Grant No. 512/12 and 1698/12. We also acknowledge support from FP7 Marie Curie Career Integration Grant No. PCIG09-GA2011-293576 and Technion Center for Security Science and Technology (CSST) Grant No. 2019459. G


Article



■

(36) Stellwagen, N. C.; Gelfi, C.; Righetti, P. G. Biopolymers 1997, 42, 687−703. (37) Rogacs, A.; Marshall, L. A.; Santiago, J. G. J. Chromatogr. A 2014, 1335, 105−120. (38) Khurana, T. K.; Santiago, J. G. Anal. Chem. 2008, 80, 6300− 6307. (39) Karsenty, M.; Rubin, S.; Bercovici, M. Anal. Chem. 2014, 86, 3028−3036. (40) Sun, C. P.; Liao, J. C.; Zhang, Y. H.; Gau, V.; Mastali, M.; Babbitt, J. T.; Grundfest, W. S.; Churchill, B. M.; McCabe, E. R. B.; Haake, D. A. Mol. Genet. Metab. 2005, 84, 90−99. (41) Mohan, R.; Mach, K. E.; Bercovici, M.; Pan, Y.; Dhulipala, L.; Wong, P. K.; Liao, J. C. PLoS One 2011, 6, e26846. (42) Berlier, J. E.; Rothe, A.; Buller, G.; Bradford, J.; Gray, D. R.; Filanoski, B. J.; Telford, W. G.; Yue, S.; Liu, J.; Cheung, C. Y.; Chang, W.; Hirsch, J. D.; Beechem Rosaria, P.; Haugland, J. M.; Haugland, R. P. J. Histochem. Cytochem. 2003, 51, 1699−1712. (43) Milanova, D.; Chambers, R. D.; Bahga, S. S.; Santiago, J. G. Electrophoresis 2012, 33, 3259−3262. (44) Bercovici, M.; Lele, S. K.; Santiago, J. G. J. Chromatogr. A 2010, 1217, 588−599. (45) Bercovici, M.; Han, C. M.; Liao, J. C.; Santiago, J. G. Proc. Natl. Acad. Sci. U. S. A. 2012, 109, 11127−11132. (46) Garcia-Schwarz, G.; Bercovici, M.; Marshall, L. A.; Santiago, J. G. J. Fluid Mech. 2011, 679, 455−475. (47) Psychogios, N.; Hau, D. D.; Peng, J.; Guo, A. C.; Mandal, R.; Bouatra, S.; Sinelnikov, I.; Krishnamurthy, R.; Eisner, R.; Gautam, B. PloS One 2011, 6, e16957. (48) Eder, P. S.; DeVine, R. J.; Dagle, J. M.; Walder, J. A. Antisense Res. Dev. 1991, 1, 141−151. (49) Zöllner, E.; Seibert, G.; Slor, H.; Zahn, R. Experientia 1981, 37, 548−550. (50) Hoch, S. O.; McVey, E. J. Biol. Chem. 1977, 252, 1881−1887. (51) Kuhn, H.; Demidov, V. V.; Coull, J. M.; Fiandaca, M. J.; Gildea, B. D.; Frank-Kamenetskii, M. D. J. Am. Chem. Soc. 2002, 124, 1097− 1103. (52) Igloi, G. L. Proc. Natl. Acad. Sci. U. S. A. 1998, 95, 8562−8567. (53) Jensen, K. K.; Ørum, H.; Nielsen, P. E.; Nordén, B. Biochemistry 1997, 36, 5072−5077. (54) Kuhn, H.; Demidov, V. V.; Nielsen, P. E.; Frank-Kamenetskii, M. D. J. Mol. Biol. 1999, 286, 1337−1345.

REFERENCES

(1) Wetmur, J. G.; Fresco, J. Crit. Rev. Biochem. Mol. Biol. 1991, 26, 227−259. (2) Tyagi, S.; Kramer, F. R. Nat. Biotechnol. 1996, 14, 303−308. (3) Mhlanga, M. M.; Malmberg, L. Methods 2001, 25, 463−471. (4) Lewin, S. R.; Vesanen, M.; Kostrikis, L.; Hurley, A.; Duran, M.; Zhang, L.; Ho, D. D.; Markowitz, M. J. Virol. 1999, 73, 6099−6103. (5) Sassolas, A.; Leca-Bouvier, B. D.; Blum, L. J. Chem. Rev. 2008, 108, 109−139. (6) Heller, M. J. Annu. Rev. Biomed. Eng. 2002, 4, 129−153. (7) Ramsay, G. Nat. Biotechnol. 1998, 16, 40−44. (8) Palmer, S.; Wiegand, A. P.; Maldarelli, F.; Bazmi, H.; Mican, J. M.; Polis, M.; Dewar, R. L.; Planta, A.; Liu, S.; Metcalf, J. A.; Mellors, J. W.; Coffin, J. M. J. Clin. Microbiol. 2003, 41, 4531−4536. (9) Rahman, M. T.; Uddin, M. S.; Sultana, R.; Moue, A.; Setu, M. Anwer Khan Mod. Med. Coll. J. 2013, 4, 30−36. (10) Halford, W. P. Nat. Biotechnol. 1999, 17, 835−835. (11) Wright, P. A.; Wynford-Thomas, D. J. Pathol. 1990, 162, 99− 117. (12) Schmittgen, T. D.; Lee, E. J.; Jiang, J.; Sarkar, A.; Yang, L.; Elton, T. S.; Chen, C. Methods 2008, 44, 31−38. (13) Kessler, H. H.; Preininger, S.; Stelzl, E.; Daghofer, E.; Santner, B. I.; Marth, E.; Lackner, H.; Stauber, R. E. Clin. Vaccine Immunol. 2000, 7, 298−300. (14) Kuypers, J.; Wright, N.; Ferrenberg, J.; Huang, M. L.; Cent, A.; Corey, L.; Morrow, R. J. Clin. Microbiol. 2006, 44, 2382−2388. (15) Limaye, A. P.; Huang, M. L.; Leisenring, W.; Stensland, L.; Corey, L.; Boeckh, M. J. Infect. Dis. 2001, 183, 377−382. (16) Mohan, R.; Mach, K. E.; Bercovici, M.; Pan, Y.; Dhulipala, L.; Wong, P. K.; Liao, J. C. PLoS One 2011, 6, e26846. (17) Foxman, B. DM, Dis.-Mon. 2003, 49, 53−70. (18) Ewig, S.; Torres, A. Clin. Chest Med. 1999, 20, 575−587. (19) Koenig, S. M.; Truwit, J. D. Clin. Microbiol. Rev. 2006, 19, 637− 657. (20) Kawaguchi, T.; Komatsu, S.; Ichikawa, D.; Morimura, R.; Tsujiura, M.; Konishi, H.; Takeshita, H.; Nagata, H.; Arita, T.; Hirajima, S.; Shiozaki, A.; Ikoma, H.; Okamoto, K.; Ochiai, T.; Taniguchi, H.; Otsuji, E. Br. J. Cancer 2013, 108, 361−369. (21) Bercovici, M.; Han, C. M.; Liao, J. C.; Santiago, J. G. Proc. Natl. Acad. Sci. U. S. A. 2012, 109, 11127−11132. (22) Persat, A.; Santiago, J. G. Anal. Chem. 2011, 83, 2310−2316. (23) Bercovici, M.; Kaigala, G. V.; Mach, K. E.; Han, C. M.; Liao, J. C.; Santiago, J. G. Anal. Chem. 2011, 83, 4110−4117. (24) Bahga, S. S.; Han, C. M.; Santiago, J. G. Analyst 2013, 138, 87. (25) Garcia-Schwarz, G.; Santiago, J. G. Anal. Chem. 2012, 84, 6366− 6369. (26) Garcia-Schwarz, G.; Santiago, J. G. Angew. Chem., Int. Ed. 2013, 52, 11534−11537. (27) Eid, C.; Garcia-Schwarz, G.; Santiago, J. G. Analyst 2013, 138, 3117−3120. (28) Carlsson, C.; Jonsson, M.; Norden, B.; Dulay, M. T.; Zare, R. N.; Noolandi, J.; Nielsen, P. E.; Tsui, L. C.; Zielenski, J. Nature 1996, 380, 207. (29) Everaerts, F. M.; Beckers, J. L.; Verheggen, T. P. E. M. Isotachophoresis: Theory, Instrumentation and Applications; Elsevier: 2011. (30) Hyrup, B.; Nielsen, P. E. Bioorg. Med. Chem. 1996, 4, 5−23. (31) Egholm, M.; Buchardt, O.; Christensen, L.; Behrens, C.; Freier, S. M.; Driver, D. A.; Berg, R. H.; Kim, S. K.; Norden, B.; Nielsen, P. E. Nature 1993, 365, 566−568. (32) Demidov, V. V.; Potaman, V. N.; Frank-Kamenetskil, M.; Egholm, M.; Buchard, O.; Sönnichsen, S. H.; Nielsen, P. E. Biochem. Pharmacol. 1994, 48, 1310−1313. (33) Hamilton, S. E.; Iyer, M.; Norton, J. C.; Corey, D. R. Bioorg. Med. Chem. Lett. 1996, 6, 2897−2900. (34) Savard, J. M.; Grosser, S. T.; Schneider, J. W. Electrophoresis 2008, 29, 2779−2789. (35) Viovy, J. L. Rev. Mod. Phys. 2000, 72, 813. H


Focused upon Hybridization: Rapid and High Sensitivity Detection of

Recommend Documents