NANO LETTERS
An Asymmetric Polymer Nanopore for Single Molecule Detection
2004 Vol. 4, No. 3 497-501
Abraham Mara,† Zuzanna Siwy,*,‡,§ Christina Trautmann,‡ Jackson Wan,† and Fredrik Kamme† Johnson & Johnson Pharmaceutical Research and DeVelopment, LLC, 3210 Merryfield Row, San Diego, California 92121, Gesellschaft fu¨r Schwerionenforschung, Planckstr. 1, 64291 Darmstadt, Germany, and Silesian UniVersity of Technology, Strzody 9, 44-100 Gliwice, Poland Received December 8, 2003; Revised Manuscript Received February 2, 2004
ABSTRACT We describe a sensor capable of detecting single DNA molecules. The sensor is based on a single nanopore prepared in a polymer film by a latent ion track-etching technique. For this purpose, a polymer foil was penetrated by a single heavy ion of total kinetic energy of 2.2 GeV, followed by preferential etching of the ion track. DNA molecules were detected as they blocked current flow during translocation through the nanopore, driven by an electric field. The nanopores are highly stable and their dimensions are adjustable by controlling etching conditions. For detecting DNA, conical nanopores with opening diameters of 2 µm and 4 nm were used. The nanopore sensor was able to discriminate between DNA fragments of different lengths.
There is a major effort to use nanosized channels or pores for biomolecular sensing.1,2 A molecule driven through an appropriately sized pore by means of an electric field will temporarily block the pore, thereby reducing or blocking the current flow through the pore.3,4 It is expected that the chemistry and structure, including the length of polymeric biomolecules, will be reflected in the duration and magnitude of current changes. The first trials to use this technique to detect single DNA polynucleotides were made with R-hemolysin, a transmembrane protein, inserted in a lipid bilayer.5,6 The 3D structure of R-hemolysin was determined in detail by X-ray diffraction, revealing a 1.5 nm channel7 through which single-stranded DNA may translocate. Early work using R-hemolysin showed successful discrimination between homo- and block polynucleotides with different base composition.6,8,9 Later studies showed that the pore could differentiate between hairpin DNA molecules with different loop and stem lengths as well as detecting base mismatches in the stem.10 The lipid bilayer that supports R-hemolysin is a major limitation due to electrical leakage currents, high capacitance per unit area and the innate physical instability of the membrane. These limit the range of applications of the protein pore, which has stimulated efforts to produce synthetic nanopores. Hitherto two techniques have been successfully applied to produce sensors in the form of single * Corresponding author. Present address: University of Florida, Department of Chemistry, Gainesville, FL 32611-7200. E-mail: zuzanna@ chem.ufl.edu. † Johnson & Johnson Pharmaceutical Research and Development. ‡ Gesellschaft fu ¨ r Schwerionenforschung. § Silesian University of Technology. 10.1021/nl035141o CCC: $27.50 Published on Web 02/17/2004
© 2004 American Chemical Society
pores. The first technique employs a feedback-controlled sputtering system, based on irradiating the materials with (argon) ion beams of several keV energy.11,12 The preparation of pores in Si3N4 with diameters down to 1.8 nm was demonstrated. Another technique uses classical lithographic methods combined with micromolding of poly(dimethylsiloxane) (PDMS).13 This led to the creation of much larger channels, approximately 200 nm in diameter. Both types of sensors have been shown to record the passage of single DNA molecules, observed as single, well-separated blockages of the pore lasting several milliseconds each. A new promising technique for nanopore production in silicon oxide has recently been reported.14 Pores of ∼20 nm diameter were created by electron-beam lithography and anisotropic etching, and subsequently reduced in diameter, with nanometer precision, using a high-energy electron beam. A short report of a potential use of these nanopores as molecular sensors was presented during the 2003 Biophysical Society Meeting.15 Here we describe another nanopore system: A single conical pore created in a polymer substrate by chemically etching the latent track of a single energetic heavy ion. We demonstrate its use in detection and discrimination of a distribution of differently sized DNA fragments. Our macroscopic, easily handled pore is 12 µm long, much longer than the solid-state nanopores reported thus far.11-13 Its high aspect ratio is believed to contribute to the observed lengthy translocations, effectively providing a higher temporal resolution. This allowed detection of fragments an order of
magnitude smaller than reported with solid state systems, classifying fragments only a few hundred bases long. We believe our system to be the first artificial pore to successfully discriminate nucleic acids based on length at a level on par with the R-hemolysin studies. Production of Single-Nanopore Membranes. To fabricate the nanopore, we employed a track-etching technique, based on irradiation of a polymer film with heavy ions of total kinetic energy from hundreds to thousands of MeV, and subsequent chemical etching of the latent ion tracks.16 The number of pores created in this way is equal to the number of ions allowed to penetrate the foil. The size and shape of the pore produced can be tailored with a precision of nanometers by controlling the type of etchant, temperature, and duration of etching. Single-pore membranes were produced at the Gesellschaft fu¨r Schwerionenforschung in Darmstadt.17,18 In preparing the sensor we chose a conical shape for the pore. The resistance of an asymmetric pore is concentrated in the tip of the cone, which reduces the effective length of the pore. A conical pore also has a smaller resistance than a cylindrical pore of the same limiting aperture and length, generating higher ion currents for a given voltage. This allows the use of durable macroscopic foils while keeping the currents measurable with off-the-shelf equipment. Information about the translocation events, as well as the chemistry and structure of the biomolecules sensed, is present in the ion current signal. The amplitude of this signal is reduced when large molecules translocate the pore. Therefore, it is important that the current signal through the pore in the presence of any buffer in which the molecule is dissolved is very simple; the ideal case being that the signal is stable. Unlike many other biochannels, which fluctuate significantly at the millisecond time scale, the R-hemolysin pore is very quiet in KCl solutions. Moreover, it has been found that nanopores prepared in some polymer materials exhibit similar behavior.19-21 With a constant voltage applied, conical pores in poly(ethylene terephthalate) (PET), however, produce current fluctuations with the amplitude reaching as high as 100% of the signal. The nanopores in polycarbonate foils exhibit similar transport characteristics.22 Only one type of nanopores yet produced by the track-etch technique, those prepared in polyimide (Kapton 50 HN, DuPont), is characterized by an extremely stable current signal over a wide range of electrolyte concentration and pH.20,23 The unique transport properties of the Kapton pores were explained on the basis of an exceptionally smooth surface of the membrane after irradiation and etching, and also correlated with the chemical structure of the polyimide base material.20,23 Kapton was the material chosen for the nanopore due to its mechanical robustness, stability, and its aforementioned transport properties. Results. Examination of DNA Translocation through Kapton Nanopores. We studied a series of conical Kapton pores prepared as described in the Experimental Section, with the small opening d varying over the range 2-7 nm in diameter and holding the big opening constant at ∼2 µm.23 Figure 1 shows a scanning electron microscopy image of a 498
Figure 1. Scanning electron micrographs of a single pore in a Kapton foil showing (A) its big opening as viewed from the surface (due to the 10 µm thickness of the foil, the small opening of the channel is at a depth beyond the range of the SEM), and (B) a part of cross section of a similar pore; the part with the small opening of the pore is below scanning electron microscopy resolution and is not shown.
cross section of a single asymmetric Kapton pore. We found the pores with d ∼ 4 nm to be most suitable for DNA translocation studies. For smaller pores the translocations were scarcely seen, while larger ones produced much shallower blockades. To perform ion current measurements in the presence of DNA, a conductivity cell was constructed in which two buffer compartments were separated by the film containing the nanopore. We examined the ability of our nanopore to detect individual double-stranded DNA molecules and to differentiate between DNA molecules of different lengths, using a 4.4 kb plasmid which was cut with restriction enzymes to produce two sticky-end fragments, 284 bp and 4.1 kb. A large ratio of DNA fragment lengths was utilized to enhance the separation in the expected distribution of blockage durations. It is interesting to note, possibly due to the increased stiffness of the double-stranded DNA and lack of self-annealing, that we saw virtually no permanent blockages of our pores. The recording cell contained 1 M KCl, buffered to pH 7.2 using phosphate buffered saline, and 0.01% Triton-X100 to prevent adsorption of DNA onto the Kapton film and surfaces of the cell. A voltage of 120 mV was applied across the nanopore, which is typical of the experiments with the R-hemolysin pore. The baseline current was 710 pA and no deflections in current flow indicative of pore blockades were seen (Figure 2A). DNA was then added to a final concentration of 0.5 µg/µL on the side of the Kapton film with the small pore opening. Addition of DNA to the side of the membrance containing the larger opening resulted in far fewer translocations, possibly due to entangling of the DNA molecules in the gradually narrowing taper of the pore. Figure 2B shows a current trace recorded in the presence of 284 bp and 4.1 kb DNA fragments. Numerous events were observed. These were clearly defined; some blockages were up to 70% of the total signal. We interpret these events as translocation of DNA molecules through the nanopore based on the correlation between addition of DNA to the recording cell and observation of Nano Lett., Vol. 4, No. 3, 2004
Figure 2. Current blockages caused by DNA. (A) Ion current versus time recording with 1 M KCl, (pH 7.2, 0.01% Triton-X100), as a control. (B) Ion current versus time recording with 1 M KCl solution (as in A) containing 284 bp and 4.1 kb dsDNA fragments, added to the side of the membrane with the small pore opening, sampled at 30 kHz. (C) A segment of the recording at higher resolution with examples of various shapes of blockages. Event I was followed within 1.1 ms by event II, which we classify as a double event, or “train”.
events. Due to the small opening of the nanopore, 4 nm, only a single double-stranded DNA molecule is expected to translocate at a time. The events differ in the depth (Figure 2B). This is possibly due to the DNA molecules assuming various conformations in the 4 nm diameter pore, resulting in various degrees of blockade. This phenomenon has also been observed with a much larger 200 nm pore13 and 3 nm Si3N4 pore.12 The wide distribution of ion current blockades and dwell times has been attributed to thermal bending of the DNA molecule including the possibility of creating a hairpin-like structure.12 Some double-events were observed, where the separation between two events was less than 2 ms (Figure 2B). As the restriction enzymes used produce sticky ends, and the salt concentration was high, we suggest that the ends of the DNA fragments may hybridize. The ends of this new hybrid would themselves have sticky ends, conceivably allowing for many such fragments to join end to end, creating a “train” of DNA fragments. If a DNA fragment that is hybridized translocates through the nanopore, one end of the trailing DNA fragment will come into close proximity of the pore opening, thus significantly increasing the probability that it will enter. The mechanical force exerted on the DNA concatamer as it translocates through the pore may be sufficiently strong to break the hybridization between the ends, resulting in two temporally closely associated, yet defined events. We also observed three and four events coupled. Analysis of Blockage Time Distribution. Due to the stability of the current signal through the Kapton pore we could employ a sampling rate of 30 kHz, filtered at 10 kHz. This high sampling rate enabled us to compare the time scale Nano Lett., Vol. 4, No. 3, 2004
Figure 3. Histograms of blockage durations. A histogram of blockage durations for the data from the recording in Figure 2 is shown in (A). The histogram has been fitted with the sum of two Gaussians. (B) A histogram of blockage durations recorded in the presence of blunt-ended 286, 974, and 4126 bp DNA fragments, fitted with three Gaussians.
of our observed events with those reported using R-hemolysin pores.5,6 The low noise in the baseline measurement allowed reliable identification of blockages of even a few percent of the baseline signal’s depth. Figure 3A shows the blockage duration histogram of a 10 minute recording, a fragment of which was shown in Figure 2B. A total of 576 blockade events were recorded. In our analysis we considered only events consisting of at least two experimental ion current points, a total of 266 events. The multiple blockades were included in the analysis as separate events. The histogram presents the distribution of events shorter than 1.6 ms, fitted with a sum of two Gaussians, whose peaks correspond to the mean blockage times. The first peak, at 0.16 ms, was observed in all our recordings and its position did not change significantly for various experimental conditions. We attribute this to bumping of the polymer, i.e., a DNA molecule interacting with the pore but not translocating, as previously described with R-hemolysin.4 It is assumed that any time DNA is pulled to the pore but the interaction is not within a certain distance of the polymer’s end that it will not translocate. The second Gaussian (Figure 3A) has a peak at 0.42 ms, which we attribute to the translocation of the 284 bp DNA fragment. Again, this is in good agreement with previously reported experiments using R-hemolysin, which generated a blockade time of 0.29 ms for a 210 nt singlestranded DNA fragment. We have also recorded events of several milliseconds, which we attribute to the translocation of the 4.1 kb DNA 499
fragment. The blockade time scales linearly with the molecule length therefore the expected translocation time for a 4.1 kb fragment was between 5 and 6 ms. The distribution of blockage durations in this region was wide, making a determination of the average translocation time statistically unreliable. Detection of DNA Fragments of Various Lengths. To differentiate between DNA polynucleotides of different lengths, φ-X174 DNA was cut to yield 286, 974, and 4126 bp fragments. Figure 3B shows a histogram of current blockage durations recorded with a 4 nm Kapton pore in the presence of these DNA fragments. The histogram, from 0.07 to 3 ms, was fitted with the sum of three Gaussians. Events longer than 3 ms are included to show that the translocation times in this region were strongly scattered. The first peak appeared at 0.18 ms, corresponding to the DNA polymer bumping into the pore opening. The two subsequent distributions, at 0.37 and 1.07 ms, we ascribe to the translocations of the 286 and 974 bp fragments, respectively. The region where translocations of the 4.1 kb fragment were expected, at 5-6 ms, contained events which were too few and too scattered to be statistically tractable. The ratio of the mean blockage durations for the 286 and 974 bp fragments, as determined by the respective peaks of the Gaussians in the histogram, was 2.9. This is close to the ratio of the two fragments’ lengths (974/286 ) 3.41), suggesting that blockage duration scaled linearly with polymer length in this experiment. Discussion. We have evaluated the feasibility of using a latent ion track-etched nanopore in a stable polymer film for biomolecular sensing. The pore is inert and mechanically robust. It can be handled by hand and exchanged between laboratories. The shape and size of the pore is fully customizable. Our results indicate that a 4 nm nanopore in Kapton film is able to detect DNA molecules and to distinguish between DNA polynucleotides of different lengths based on their translocation times. We found that distributions for dsDNA pieces of up to approximately 1000 bp in length have well-defined peaks. Applying driving voltages similar to those typically used in R-hemolysin studies, we found that the blockage durations were similar to those observed with the R-hemolysin pore for similarly sized DNA fragments.5 This similarity suggests that the functional length of the Kapton nanopore, i.e., the length of the channel where presence of DNA will significantly alter the electrical resistance, was short compared to the foil thickness. The translocation speed observed with long polymer nanopores is slower than the one reported for Si3N4 nanopores,12 which offers a higher resolution of events for shorter DNA pieces. We will focus in the future on improving the operation of our sensor for longer DNA molecules. With sticky-end DNA fragments we observed a “train” effect, i.e., two or more temporally closely associated events. We propose that this effect corresponds to the translocation of double-stranded DNA fragments that were associated by hybridization of their sticky ends. Using blunt-ended DNA we did not observe this train effect. 500
With our track-etched pores we have also recorded very rapid blockages, which appeared independent of the DNA length and experimental conditions. We interpret these as bumping of the polymer into the pore opening without subsequent translocation, as previously reported for the R-hemolysin pore. The inherent physical and chemical stability of the Kapton nanopore provides great freedom in evaluating the effect of changing parameters such as temperature and buffer conditions on the translocation of polynucleotides. Future efforts will also include the introduction of a “gate” element24 in order to provide active control over the translocation rate of a polynucleotide. We would like to emphasize that another advantage of using polymer matrix for creating a nanopore system is the possibility to modify chemically the surface of nanopores and ultimately control its surface chemistry.25 We think that nanopores in polymer films will be the starting point of sensors specific for given DNA sequences. Experimental Section. Preparation of Kapton Nanopores. Kapton foils of ∼12 µm thickness were irradiated with single uranium ions of 2.2 GeV energy (UNILAC, Darmstadt, Germany). Ion tracks were developed in sodium hypochlorite.20,23,26 The conical pore geometry was obtained by applying the etchant only on one side of the membrane, which was placed between two chambers of a conductivity cell. The other chamber was filled with a stopping medium, which neutralized the etchant as soon as the pores were etched through. 1 M KI was used as the stopping medium, reducing OCl- ions to Cl-. The conductivity cell completed a circuit driven by a custom current-to-voltage converter of pA sensitivity. Applying a voltage across the membrane enabled us to monitor and control the etching process: the current at the beginning of the etching process is zero and only when the etchant penetrates the foil does the current increase, gradually showing the unambiguous opening of the pore and corresponding increase of the pore diameter.26 For a pore with a subsequently verified limiting aperture of ∼5 nm, we observed a current of approximately 0.2 nA while etching with an applied potential across the membrane of 1.5 V. This actively monitored etching assured that nanometer size openings were obtained. The angle of the cone can be tailored by the pH of the sodium hypochlorite,27 and for our experiment we etched conical pores with an opening angle of ∼15 degrees. The large opening D, nominally ∼2 µm, is determined from known etching rates under relevant conditions. When the etching was completed, the effective diameter of the small opening was measured based on pore conductivity using a standard KCl solution.20,23 Independent verification of pore diameter has been established by measuring conductivity of the pore in the presence of poly(ethylene glycol).19 At present, we can successfully prepare pores with limiting apertures down to 2 nm repeatably. Cell. To perform ion current measurements in the presence of DNA, several generations of conductivity cells were constructed in house, with volumes ranging from 1 mL to 10 µL. We reduced the membrane surface exposed to buffer down to 0.2 mm2, which helped to lower capacitance and Nano Lett., Vol. 4, No. 3, 2004
minimize adsorption of DNA to the surface. Smaller volumes served to increase concentrations of DNA for a given mass of input. The fixed, fully submerged electrodes of later cell revisions reduced noise and increased repeatability. Easily flushable chambers facilitated complete buffer exchange and collection of buffers after a run; iterations included water passageways for temperature controlling purposes. The blunt ended data presented here were recorded with a cell of 10 µL volume and semi-permanently mounted silver/silverchloride electrodes. The data recorded with sticky ended fragments used a 40 µL chamber and off-the-shelf electrodes. Recording. The measurements were made using an Axopatch 200B with the signal digitized by a Digidata 1322A (Axon Instruments, CA). Data Analysis. The ion current time series were analyzed with QuB software available at www.qub.buffalo.edu/.28,29 DNA Preparation. The first type of DNA was generated from a 4.4 kb plasmid cut with the restriction endonucleases EcoRI and SapI (recognition sequences 5′-G∧AATTC-3′, 3′-CTTAA∧G-5′, and 5′-GCTCTTCN∧NNN-3′, 3′-CGAGAAGNNNN∧ -5′, respectively) to obtain two sticky-end fragments, 284 bp and 4.1 kb long. The fragments were purified by precipitation in phenol-chloroform-isoamyl alcohol (25:24:1 at pH 8.0). The second type of DNA used was bacteriophage phi-X174 (cs70 mutation), cut with XmnI (recognition sequences 5′-GAANN∧NNTTC-3′ and 3′CTTNN∧NNAAG-5′), resulting in three blunt ended fragments with lengths of 286, 974, and 4126 bp. All DNA was used at a stock concentration of 2 µg/µL in 1 M KCl, buffered to pH 7.2 with phosphate buffered saline (PBS), and with 0.01% Triton-X100. Acknowledgment. Z.S. was supported by the Foundation for Polish Science, the Alexander von Humboldt Foundation and a J&J PRD fellowship. The authors are very grateful to Dr. Ed Kaftan and Dr. Adrienne Dubin for their support in experiments. References (1) Wang, H.; Branton, D. Nature Biotechnology 2001, 19, 622. (2) Austin, R. Nature Materials 2003, 2, 567.
Nano Lett., Vol. 4, No. 3, 2004
(3) Colombini, M. J. Membr. Biol. 1980, 53, 79. (4) Bezrukov, S. M.; Vodyanoy, I.; Parsegian, V. A. Nature 1994, 370, 279. (5) Kasianowicz, J. J.; Brandin, E.; Branton, D.; Deamer, D. W. Proc. Natl. Acad. Sci. U.S.A. 1996, 93, 13770. (6) Akeson, M.; Branton, D.; Kasianowicz, J. J.; Brandin, E.; Deamer, D. W. Biophys. J. 1999, 77, 3227. (7) Song, L.; Hobaugh, M. R.; Shustak, C.; Cheley, S.; Bayley H.; Gouaux, J. E. Science 1996, 274, 1859. (8) Meller, A.; Nivon, L.; Brandin, E.; Golovchenko, J.; Branton, D. Proc. Natl. Acad. Sci. U.S.A. 2000, 97, 1079. (9) Howorka, S.; Cheley, S.; Bayley, H. Nature Biotechnology 2001, 19, 636. (10) Vercoutere, W.; Winters-Hilt, S.; Olsen, H.; Deamer, D.; Haussler, D.; Akeson, M. Nature Biotechnology 2001, 19, 248. (11) Li, J.; Stein, D.; McMullan, C.; Branton, D.; Aziz, M. J.; Golovchenko, J. A. Nature 2001, 412, 166. (12) Li, J.; Gershow, M.; Brandin, E.; Golovchenko, J. A. Nature Materials 2003, 2, 611. (13) Saleh, O. A.; Sohn, L. L. Nano Lett. 2003, 3, 37. (14) Storm, A. J.; Chen, J. H.; Ling, X. S.; Zandbergen, H. W.; Dekker C. Nature Materials 2003, 2, 537. (15) Storm, A. J.; van den Broek, D.; Lemay, S.; Dekker, C. Biophys. J. (Annual Meeting Abstracts) 2002: 51 a. (16) Fleischer, R. L.; Price, P. B.; Walker, R. M. Nuclear Tracks in Solids. Principles and Applications; University of California Press: Berkeley, 1975. (17) Spohr, R. Methods and device to generate a predetermined number of ion tracks, German Patent DE 2951376 C2 (filed 20.12.1979, issued 15.09.1983); United States Patent No. 4369370, 1983. (18) Ball, P. Nature Materials, nanozone news (31 October 2002). (19) Siwy, Z.; Gu, Y.; Spohr, H. A.; Baur, D.; Wolf-Reber, A.; Spohr, R.; Apel, P.; Korchev, Y. E. Europhys. Lett. 2002, 60, 349. (20) Siwy, Z.; Apel, P.; Baur, D.; Dobrev, D. D.; Korchev, Y. E.; Neumann, R.; Spohr, R.; Trautmann, C.; Voss K. O. Surf. Sci. 2003, 532, 1061. (21) Siwy, Z.; Fulin´ski, A. Phys. ReV. Lett. 2002, 89, 158101. (22) Baur, D.; Siwy, Z.; Spohr, R.; Trautmann, C., unpublished results. (23) Siwy, Z.; Dobrev, D.; Neumann, R.; Trautmann, C.; Voss, K. O. Appl. Phys. A 2003, 76, 781. (24) Siwy, Z.; Behrend, J.; Fertig, N.; Fulinski, A.; Martin, C. R.; Trautmann, C.; Neumann, R.; Tomil-Molares, E. Nanodevice for charged particle flow and method for producing same, German and US patent, registration on 25.09.2002, Nr 102 44 914.7. (25) Howorka, S. Upper Austrian Research Institute, Linz, Austria, private communication. (26) Siwy, Z.; Apel, A.; Dobrev, D. D.; Neumann, R.; Spohr, R.; Trautmann, C.; Voss, K. O. Nucl. Instr. Methods B 2003, 208, 143. (27) Trautmann, C.; Bruechle, W.; Spohr R.; Vetter, N.; Angert, N. Nucl. Instr. Methods B 1996, 111, 70. (28) Qin, F.; Auerbach, A.; Sachs, F. Biophys. J. 1996, 70, 264. (29) Qin, F.; Auerbach, A.; Sachs, F. Proc. Royal Soc. B 1997, 264, 375.
NL035141O
501
