Three decades of nanopore sequencing
David Deamer1, Mark Akeson1 & Daniel Branton2

A long-held goal in sequencing has been to use a and together with Sergey Bezrukov at the US National Institutes of
voltage-biased nanoscale pore in a membrane to Health (NIH), had established the conditions that were necessary to
measure the passage of a linear, single-stranded (ss) avoid spontaneous gating (pore closure) of the a-hemolysin channel3.
This was important, because spontaneous gating would have hindered,
DNA or RNA molecule through that pore. With the
or confused, observations of nucleic acid translocation through the
development of enzyme-based methods that ratchet nanopore. Kasianowicz was also collaborating with Bezrukov to
polynucleotides through the nanopore, nucleobase- investigate the effect of polyethylene glycol on pore conductance and,
© 2016 Nature America, Inc. All rights reserved.

by-nucleobase, measurements of changes in the consistent with earlier reports2, found that a pore radius of ~1.1 nm
current through the pore can now be decoded into a accounted for their results4.
In initial experiments, D.D. and Kasianowicz worked with a single
DNA sequence using an algorithm. In this Historical
a-hemolysin channel inserted into a lipid bilayer that separated two
Perspective, we describe the key steps in nanopore buffered KCl-filled compartments. RNA homopolymers (polyuridylic
strand-sequencing, from its earliest conceptualization or polyadenylic acid) were then added to the cis side of the membrane.
more than 25 years ago to its recent commercialization Immediately after addition of the polymers, the first encouraging results
and application. were observed. When a positive voltage bias greater than 80 mV was
applied to the trans compartment, numerous blockades—transient,
Nanopore sequencing (Box 1 and Fig. 1) has its origins in several millisecond time-scale reductions of the ionic current through the
laboratories during the 1980s (Fig. 2). In 1989, one of us (D.D.) jotted a-hemolysin channel—appeared. No blockades were detected when the
a seemingly implausible idea in his notebook (Fig. 3), suggesting that trans compartment was negatively biased. This was expected because the
it might be possible to sequence a single strand of DNA being drawn polyanionic RNA in the cis chamber would be inhibited from entering
through a membrane’s nanoscopic pore by electrophoresis. Around the the nanopore when the trans side was negative. These preliminary results
same time, George Church’s interest in scaling up DNA sequencing for were consistent with the hypothesis that a voltage applied across the
the Human Genome Project led him to propose that an electronically hemolysin pore could electrophorese single molecules of RNA through
monitored phage motor embedded in a bilayer might provide sequence the a-hemolysin channel.
information upon processing of double-stranded (ds) DNA. Hagan In February 1994, D.D. and D.B. together visited Kasianowicz to

Bayley’s interest in membrane proteins and the structure and assembly repeat and extend the earlier results. Further experiments showed that
of oligomeric transmembrane protein pores, such as a-hemolysin, also the average blockade duration was larger for longer RNA polymers
led him and his colleagues to ask if pores could serve as biosensors of compared with shorter polymers. This led to the hypothesis that the
molecules and cations1. duration of the blockades reflected the length of time required for the
Whereas Church decided to pursue other high-speed sequencing polymer to translocate through the nanopore from the cis chamber to the
methods, two of us (D.D. and D.B.) had independently noted trans chamber. Shortly after these initial experiments were completed,
publications suggesting that a-hemolysin from Staphylococcus aureus D.B. and D.D. received a personal communication from Langzhou
could form transmembrane channels wide enough (>1.2 nm) to Song and colleagues5, who were investigating the crystal structure of
accommodate a single strand of DNA2. Furthermore, D.D. had listened a-hemolysin. The early evidence from X-ray diffraction indicated that
to talks by John Kasianowicz at two scientific conferences and knew that the channel diameter in the a-hemolysin pore was between ~1.5 nm
he was working with a-hemolysin. In late 1993, D.D. visited Kasianowicz and ~2.5 nm. This information facilitated well-controlled translocation
at the US National Institute of Standards and Technology (NIST) to experiments. If the narrowest diameter of the channel was only ~1.5 nm,
test whether it was possible to electrophorese homopolymers of RNA it was reasoned that ssDNA, but not dsDNA, polymers would be able to
through an a-hemolysin pore. Kasianowicz had already gained much translocate from cis to trans.
experience working with a-hemolysin in collaboration with Bayley, This hypothesis was tested by Kasianowicz at NIST using a mixture of
ssDNA and dsDNA controls designed and synthesized by D.B.’s group
1Department of Biomolecular Engineering, University of California, Santa Cruz, at Harvard. The mixture of single- and double-stranded synthetic DNA
California, USA. 2Department of Molecular & Cellular Biology, Harvard University, molecules was sent to Kasianowicz, who placed it on the cis side of a
Cambridge, Massachusetts, USA. Correspondence should be addressed to D.D. membrane that contained a known number of a-hemolysin pores. (In
some experiments, more than one pore was inserted in the membrane to
Received 6 January 2015; accepted 5 November 2015; published online 6 May more accurately quantify the amounts of DNA translocated to the trans
2016; doi:10.1038/nbt.3423 side.) The blockades produced in a given time interval were counted at


A nanopore achieves Processive control of DNA strand single-nucleobase translocation is demonstrated discrimination among using the phi29 DNA polymerase nucleobases in a DNA strand DNA sequencing is accomplished with the phi29 DNA polymerase Oxford Nanopore and the MspA pore Technologies (ONT) is Marina Corral Spence/Nature Publishing Group Church. but many translocate. but not dsDNA molecules passed through the pore. These results provided the first convincing evidence an ssRNA oligomer comprising 70 cytosine nucleotides followed by 30 that single-stranded nucleic acid polymers can be captured by the adenine nucleotides. Because the ionic conductivity through the narrowest region Figure 1 Nanopore sequencing. coli genome is and colleagues file patent segments are with Spike Willcocks. Deamer. All rights reserved. but this experiment was an MspA nanopore and a helicase enzyme that steps along the carried out using exceptionally low concentrations of ATP to slow the © 2016 Nature America. MinION device at the Advances in assembled de novo with application for nanopore distinguished in single David Norwood and Genome Biology and Technology 99. Even though individual nucleobases were not resolved. ssDNA could discriminate between purine and pyrimidine bases. polynucleotide. about oligomeric structure or composition7. The next important question was whether the a-hemolysin nanopore When the results were tallied it was clear that. It turned out that the variations in current levels were not Publication of the crystal structure of a-hemolysin 5 shortly due to different current-blocking properties of individual nucleobases. after the experiments showing ssDNA translocation through a but instead to the fact that the cytosine-containing portion of the nanopore6 made it possible to interpret the translocation results molecule forms a narrow single-stranded helix that can translocate in detailed molecular terms.4% accuracy using sequencing RNA molecules Gordon Sanghera meeting in February MinION alone 1995 1996 1999 2001 2005 2008 2012 2014 2015 First results of DNA Solid-state nanopores Single molecules of ONT releases MinION transport through the are produced and DNA are detected by to early users in the α-hemolysin nanopore used to detect DNA Jens Gundlach's group MAP program using the MspA nanopore Figure 2 Milestones in nanopore DNA sequencing. 1b) reveal the sequence nanopore (green) that provides the only path through which ions or of nucleobases in the translocating strand. Translocation aperture region of the nanopore serves as ­its sensing region. NIST. Santa Cruz. The approach was to synthesize ssDNA molecules. and quantitative PCR techniques were used at Harvard to quantify Distinguishing purine and pyrimidine segments the number of DNA molecules that passed to the trans side of the pore. two distinct current levels were clearly visible in each through the pore’s nanoscopic aperture. Branton Purine and pyrimidine founded by Hagan Bayley Clive Brown introduces the An E. the narrowest polynucleotides can move from the cis to the trans chamber. 1a). the other experiments were needed to determine whether nanopore results showed that an a-hemolysin nanopore can provide information sequencing would actually be possible. the M. In 1997. The experimental demonstration of through a-hemolysin as a helix. In nanopore strand-sequencing. Furthermore. thus ratcheting the DNA through the nanopore helicase activity. (a) Two ionic solution-filled chambers of the nanopore is particularly sensitive to the presence of the are separated by a voltage-biased membrane. When the 100 mer was added to the cis side of the electrical field around a voltage-biased pore and can be electrophoresed a-hemolysin pore. A constant voltage bias (trans side Current (pA) 80 positive) produces an ionic current through the nanopore and drives ssDNA or ssRNA in the cis chamber through the pore to A cis 70 the trans chamber.g. (b) Portion of a record showing the ionic current through a nanopore The diagram and current trace shown in Figure 1 are from an measured by a sensitive ammeter. of the polynucleotide through the nanopore is controlled by an enzyme (red). An enzyme (e. thereby increasing the duration of each level to illustrate one nucleobase per step. Thus. Inc.. the nanopore serves as a biosensor a b 100 and provides the sole passage through which the ionic solution 90 on the cis side of the membrane contacts the ionic solution on the trans side (Fig. whereas the adenine portion forms a­­ oligomer transport and single-molecule detection was the first step larger-diameter helix that must be unwound before the single strand can npg suggesting that nanopore sequencing might be feasible. A single-stranded nucleobase’s mass and its associated electrical field. NATURE BIOTECHNOLOGY VOLUME 34 NUMBER 5 MAY 2016 519 . a polymerase or helicase) 60 Current levels is tightly bound to the polynucleotide such that its step-wise movement controls and ratchets the nucleotides through 0 10 20 180 mV trans Time (s) the small-diameter nanopore. number of ionic current blockades matched the number of translocated and began to work on this question. the experiment carried out in a horizontal bilayer apparatus 7 using stepping rate is usually 30 bases per second. moved from NIH to the University of California. as predicted. nucleobase by nucleobase. the ionic polynucleotide (black) is electrophoretically driven through an MspA current levels through the nanopore (Fig. blockade.A. HISTORICAL PERSPECTIVE Box 1 Nanopore sequencing—the basics In nanopore sequencing. the resolution that can be achieved.

The sketch shows three bases of a DNA strand being drawn through a nanopore by an applied voltage. Inc. It was soon shown that nanopore strand-sequencing. the immobilized strand greatly improved polynucleotide consistently produced two types of blockades6. All rights reserved. directed by in an otherwise homopolymeric DNA strand or in a heteropolymeric Jeffery Schloss. HISTORICAL PERSPECTIVE © 2016 Nature America. when Bayley’s and Ghadiri’s groups together sufficient time to identify the next nucleobase in the strand’s sequence. npg The ability to distinguish between purine and pyrimidine strands to discriminate among and identify individual nucleobases in a in a single RNA molecule generated growing interest in nanopore heteropolymeric strand of DNA14. Compared with a freely earlier puzzling observation that samples containing only one type of electrophoresing strand. making it possible to resolve an individual The fundamental question of whether a nanopore is capable of nucleobase. 1989). Figure 3 D. But all of these experiments depended the blockade signals can distinguish between several identical length on the DNA strands being immobilized in the a-hemolysin pore using ssDNA polynucleotides that differ only in their oligomeric sequence8. Because the between polynucleotide strands that traverse the nanopore 3ʹ to DNA strands were stably immobilized. This accounted for the fixed position in the pore’s sensing region. This program. Two pages of a notebook in which D.’s notebook. the signal-to-noise ratio. prevented the strand from completing its electrophoretically driven later experiments confirmed that the blockade signals can distinguish translocation through the small diameter of the pore. sketched the original nanopore sequencing concept are shown (dated June 25. and the expected base-specific electrical readout is shown below with each base affecting the current in such a way that the sequence of bases can be determined. found that a single adenine nucleotide substitution at a specific This proved to be a difficult challenge.D. location in a strand of poly-d(C) can be distinguished from all the cytosines by its characteristic effect on the ion conductance of Processive control of DNA translocation a-hemolysin12. regardless of whether they are located Genome Sequencing Technologies project. The voltage also produces an ionic current.8. The results pointed to the necessity of ratcheting the recognizing DNA with single-nucleobase resolution remained strand through the pore so that each step of the ratchet allowed unanswered until 2005. a terminal hairpin or a biotin-streptavidin complex whose width Further highlighting the sensitivity of the nanopore detection system.D. These initial experiments were soon followed by The pace of progress increased substantially in 2003 when the National further experiments showing that all four DNA nucleobases can be Human Genome Research Institute announced the Revolutionary distinguished from each other. eventually made over $50 million available in grants strand13. a nucleobase remained at a 5ʹ from those traversing 5ʹ to 3ʹ (refs.15 represented critical steps toward analysis of nucleic acids in many laboratories. This and subsequent demonstrations of a nanopore’s ability to explore methods for sequencing the three billion base pairs of the 520 VOLUME 34 NUMBER 5 MAY 2016 NATURE BIOTECHNOLOGY . 9–11).

Attempts to sequentially ratchet DNA through the nanopore sensor driven by A family polymerases were limited by the fact that these enzymes dissociate under the load imposed by the applied voltage required to drive DNA through the nanopore.000 times longer than were binary Klenow fragment–DNA complexes22. and would not translocate through. 4). depending on the volume Membrane of the amino acid side chains protruding into the channel cylinder5. 25 nucleotides Figure 5 Nanopore researchers at the Advances in Genome Biology and of the captured template strand were examined one by one in both the Technology Meeting. wild-type MspA. binding either the predicted. NATURE BIOTECHNOLOGY VOLUME 34 NUMBER 5 MAY 2016 521 . a porin from nanopore’s sensing region for >100 ms. Still. with single-nucleotide resolution by alternating the voltage applied Gundlach’s group then established that the mutant nanopores to the trans side of the membrane between –30 mV (an elongation npg mode wherein the 3ʹ-OH of the primer is accessible and binds to DNA polymerase in bulk phase) and +30 mV (a monitoring mode wherein the polymerase usually dissociates from the DNA. was to use processive diameter as in a-hemolysin but nearly an order of magnitude enzymes to control DNA movement through the nanopore16. Up to 500 synthetic DNA strands could be processed by a single nanopore. Inc. Florida. first proposed in 1998. HISTORICAL PERSPECTIVE M MspA spA -hemolysin -h α-hemolysin α Engineering protein nanopores to discriminate nucleobases It was apparent early in nanopore sequencing development that the wild-type a-hemolysin pore was not optimal as a DNA sequencing sensor. correctly. One solution. higher signal-to-noise ratio than can be achieved in <10 ms.6-nm-long aperture32. Marco Island. This sensing region was about the same The successful strategy. The sensing region in current levels obscures identification of individual bases. they demonstrated Early on. Subsequent experiments used ‘blocking oligomers’ that favored nanopore-controlled activation of phi29 DNAP–DNA complexes while preventing DNA polymerization in the bulk phase of the cis compartment23.31. proposed by Bayley and colleagues13–15. This level of translocation control was Jens Gundlach. They human genome for $1.A. All rights reserved. 2012.. Having ~12 nucleobases contributing to the measured Figure 4 Sensing regions in MspA and a-hemolysin.4 nm to ~2. was to find a nanopore with a relatively short © 2016 Nature America. MspA is much shorter than in a-hemolysin. The an essential component in five publications that demonstrated practical photograph was taken shortly after Clive Brown’s plenary seminar introducing nanopore DNA strand sequencing24–28. Using these mutant MspA constructs. Sensing Although there are three particularly good sensing spots within the Sensing region 5-nm-long stem14. thereby placing the extended primer/template junction at the detecting aperture in a-hemolysin). these and other methods to slow Although initial tests indicated that ssDNA could not be detected translocation through the pore without true nucleobase-by-nucleobase by. (seated) D. Using an A family DNA polymerase. initially suggested in the first nanopore picoampere current modulations measured in a nanopore requires a sequencing patent16. Initial shorter (Fig. binary phi29 DNAP–DNA complexes were retained on the a-hemolysin pore >10. Jens Gundlach at the University of Washington experiments demonstrated that. Using synthetic candidate.15. whose crystal structure was later shown ssDNA transport through the pore was one of the key feats that made to have a funnel-like geometry that narrowed to a ~1.4 nm. Nanopore sequencing was an obvious tested a series of modified channels for this purpose. nanopore strand-sequencing a reality. an early prototype of the MinION sequencer developed by ONT. acids were fully capable of electronically detecting and characterizing Ghadiri’s group showed that primer strand elongation could be detected single molecules of ssDNA33. mutants of MspA in which negatively charged amino acids lining the Subsequent experiments by Ghadiri and colleagues19 progressed a wild-type pore were replaced by neutral or positively charged amino further step toward ratcheting.B. and D.D. pore by orders of magnitude. DNA strands immobilized in the modified pores. Experiments using phi29 DNA polymerase (phi29 DNAP) were more successful. it was clear that even the smallest voltage bias that drives that the four canonical nucleotides could be individually resolved in a DNA strand through the pore moves each base through the sensing DNA strands13–15. Under conditions where catalysis was prevented. Nucleobase identification using the small The second solution. only two consecutive nucleotide additions were observed20.000 or less. experiments with differing length oligomers region showed that all of the ~12 nucleobases that occupy the 5-nm-long stem during strand translocation are sensed and reflected in the current level of a blockade29. and multiple research groups were soon supported. Kristen Stoops (ONT) and M. region in less than ~1–10 ms. Achieving sensing aperture. In most of those cases. that MspA would therefore be a better detector of Klenow fragment17 or the Escherichia coli exonuclease I18 to the individual nucleobases in a DNA strand than a-hemolysin (personal DNA strands slowed their translocation through the a-hemolysin communication).2 nm diameter. At best. Early work in Roland Benz’s laboratory had the higher signal-to-noise ratio required that each base reside in the characterized the biophysical properties of MspA30. The crystal structure of a-hemolysin revealed it to be a mushroom-shaped heptamer with a 5-nm stem whose inside channel diameter varies from ~1. ≤0. was to engineer a-hemolysin to create a nanopore with a sensing region shorter than 5 nm.21. Attaining such slow ratchet-like Mycobacterium smegmatis. Left to right: (standing) 3ʹ-to-5ʹ and 5ʹ-to-3ʹ direction. engineered ratcheting proved to be insufficient. absent catalysis.

24. Since 2012. In the MAP user group. coli Delivering nanopore sequencing to the genomics community K-12 (ref. This to a reference E. There have now been several MinION chemistry versions and numerous base-calling algorithm updates that have improved device Reading DNA using phi29 DNAP coupled to the MspA nanopore performance. widely accessible nanopore human chromosome X cloned into a bacterial artificial chromosome. Sanghera. high-coverage read depth yields DNAP–MspA combination have been documented for both synthetic more accurate MinION sequencing data. and research details23.000 bases35. More recently. in the Nanopore Research Group at the The two essential components of a functioning nanopore sequencing University of California. followed shortly thereafter by publication of the conceptual insertion sequences35. FL. Marco Island.37–39 are able to analyze recorded revealed progressive ionic current levels that reflected a ‘C-A. likelihood algorithm to align M13mp18 genome segments35. Bayley joined Gordon making it possible to use only MinION data for full de novo assembly. The MinION is compact and portable. coli K-12. an international consortium DNA displacement through the a-hemolysin pore. back to camera) and Reza Ghadiri (Scripps Research Institute) at the April 2011 fragment lengths are in the range of 6. In October 2011. sequencing device would require sustained effort from the private sector. As a result. Bayley returned to his native England to become professor In many of the early studies testing the MinION on viral and microbial of chemical biology at the University of Oxford.40. DNA molecules in the cis compartment of each nanopore are continuously captured as a helicase enzyme unwinds the double-stranded DNA into a continuous single strand that is drawn through the pore to produce an electrical signal containing sequence information (Fig. Another early test of the MinION sequencing accuracy used shotgun analytical methods to generate a whole-genome data set from E.0) logical to put the two together. They were soon rewarded with their first convincing from the five laboratories. Reading accuracy will depend in part on the chemistry and significantly larger ionic current differences between nucleobases than enzyme used to ratchet the DNA through the nanopore in a controlled did a-hemolysin nanopores34. coli genome into a single (Fig. HISTORICAL PERSPECTIVE Technology Officer Clive Brown during a plenary talk at the AGBT Meeting in 2012.4% identity at 30× coverage relative embarked on a focused internal strand sequencing effort in 2010. fashion. The electrical signals yield long continuous DNA sequence reads with lengths being a func- Figure 6 Hagan Bayley (center) conversing with Jene Golovchenko (Harvard. single-nucleotide strands23–26. blocking oligomers results from some MAP researchers. of which 512 are selected for monitoring by the software. with 2D read accuracies of 92–93%.3)35 and to >90% in unpublished from UC Santa Cruz to Seattle with phi29 DNAP. 2). common left). During analysis of the sample. They reported consistent results MspA nanopore. nucleobase calls using the phi29 As with routine shotgun sequencing. and it was in a reference sequence) ranged from 66% in June 2014 (release R6. All rights reserved. He and Liz Manrao of five laboratories associated with MAP compared results from the in the Gundlach laboratory at the University of Washington used same strain of E. USA) meeting in February 2012 unfinished regions of a genome. The ionic currents they developed algorithms for nanopore sequencing35. a reference sequence was required for genome assembly and experienced in the development and commercialization of intellectual read accuracy. Max Cherf traveled to 85% in November 2014 (release R7. it must compete © 2016 Nature America. CT47 (cancer/testis) gene family in a previously unresolved segment of Development of a high-throughput.6-Mb contig that achieved 99. with other high-speed sequencing methods both in terms of accuracy discriminated between all four different nucleobases and achieved and cost. 2014. A flow cell in the device has 2. using identical protocols for extraction. and combine multiple reads of the same DNA sample and achieve npg T’ trimer repeat within the template strand. improved algorithms have been developed. the average identity observed (the device—processive control at single-nucleotide precision and good proportion of bases in a single-molecule read aligned to matching bases discrimination among nucleobases—were now in place. Two years later. Santa Cruz. and a synthetic DNA substrate that he had successfully used to monitor In a recent study published online36. Again using only MinION data and culminated in the ONT MinION device first described by ONT’s Chief a different improved sequencing algorithm. These results were reported higher accuracies. Spike Willcocks and David Norwood to found ONT in 2005 One laboratory has now assembled the E. weighing about 100 g and con- trolled by a laptop computer with software developed by the company. biotech and point-of-care diagnostics. both strands of the dsDNA molecule can be analyzed—a so-called 2D read— to substantially improve base-calling accuracy. nanopore sequencing has by the two groups in a poster at the Advances in Genome Biology and already been used to generate scaffolds that resolve complex and often Technology (AGBT. For example. such as the position of repetitive (Fig. ONT licensed DNA strand sequencing patents in 2008. Because the DNA molecules are hairpins. Recently sequential reads of DNA template strands. in June. Together with a team genomes.000 to 48. and 4. In 2003. Samples are prepared by adding a hairpin loop to one end of double-stranded genomic DNA or cDNA fragments. meeting of NHGRI Principal Investigators.28 and native bacteriophage DNA27. tion of the fragmentation method. Amherst. as expected. If commercial nanopore sequencing is to succeed. a second laboratory has 522 VOLUME 34 NUMBER 5 MAY 2016 NATURE BIOTECHNOLOGY .048 individual protein nanopores. coli genome37. ONT released individual MinION devices in a large-scale collaborative MinION Access Program (MAP). each embedded in a separate stable membrane. This The academic experiments described in the preceding text were advance allowed Jain and colleagues35 to establish copy number of the accomplished using single nanopores in lipid bilayer membranes. phi29 DNAP translocation control in combination with an engineered shearing and library preparations. 1). 41). Muthukumar Murugappan (University of Massachusetts. Inc. thus establishing the variant calls approaching 99% accuracy were achieved using a maximum feasibility of DNA nanopore sequencing. property. For example. 5).

D. A second. Nevertheless.J.. DNA at a range of coverages.2 nm for transmembrane pore. Inc. Proc. Kasianowicz. T.000 of the through which it has already translocated has been demonstrated51. Protein Eng. 13770–13773 (1996). J.to threefold fewer insertion errors35. & Bayley. possible isoforms of the Drosophila melanogaster Dscam1 gene by When single molecules of native DNA can be read multiple times nanopore sequencing42 shows that transcriptome characterization to achieve high accuracy reads. 6. and assembled de novo the sequence of E. 1859–1866 (1996). Meller. Rev. J.html. so it is likely that other applications New publications demonstrating the utility of nanopore sequencing will emerge. J. Natl. HISTORICAL PERSPECTIVE sequenced M13 bacteriophage DNA to 99% accuracy at moderate through a graphene layer may be short enough to yield ionic current coverage.. ratio for base determination. Using short read length second. stable ionic insulator. J. two papers were published that detailed use of MinION nanopore sequencers for Ebola outbreak that “.E.48 used by ONT’s base-caller hosted by Metrichor 97. J. All rights reserved. users will be able to directly analyze can now be accessible to a much broader range of investigators epigenetic modifications. The recent identification of ~8. M. experiments and modeling suggest that a nanopore pierced translocation through a biological nanopore.. In fact.M. Proc.. S. J..J....J. Aksimentiev. Schulten. it is sufficiently flexible to provide information about a variety of with existing methods39. USA 101.46. Kasianowicz. D. two challenges persist. calling in the clinic44. I. Wang. indicated that ONT’s future products would utilize mutants of the © 2016 Nature America.W. Lett.W. J. Dynamics and to be only ≤0. HG007827 and portable device is a major technological achievement. Natl. Even though MspA’s limiting aperture has been estimated 4. UK). Brandin. Acad. 12377–12382 (2005). Menestrina. & Branton. Orientation layer of graphene49 or other atomically thin material that can serve as a discrimination of single-stranded DNA inside the alpha-hemolysin membrane channel. For instance. For instance. the most common errors are deletions (~0. Branton. Although this has not yet been achieved. Branton. Some of the intervals may be so short that they are Reprints and permissions information is available online at http://www. Science 274. ACKNOWLEDGMENTS Conclusions and outlook The authors are funded by the National Human Genome Research Institute of The commercial development of nanopore strand-sequencing as a the National Institutes of Health award numbers HG006321. & Deamer. combating infectious disease outbreaks at the point of care. Natl.H. Brandin.nanopore sequencing could become the ideal tool for the low.. A. Determination of RNA orientation during nm thick.P. J.. Given an effective sensing length of ~1. Because a single layer of graphene is only ~0. J. & Deamer. nucleobases in a strand of DNA contribute to each of the measured 2352–2355 (1993). Some of these may be groundbreaking. H. Microsecond time-scale discrimination among polycytidylic acid.A. so that only the most likely sequence can Biophys. Dunning. H. molecular dynamic simulations suggest that free energy of polymers partitioning into a nanoscale pore. Gundlach. and affecting each current level substantially diminishes the signal-to-noise polyuridylic acid as homopolymers or as segments within single RNA molecules. The base multiplicity 7. polyadenylic acid. surveillance on site in Liberia53 and Guinea54 West Africa. 1079–1084 (2000). alternatively can be realized by repeated resequencing of the same native genomic spliced transcription occurs at multiple places that are often located fragment multiple times. The first problem is that enzyme turnover is stochastic. Brandin. it is not surprising that up to five neighboring bases individual polynucleotide molecules using a membrane channel. A. coli and lambda blockade levels to which only 1–2 nucleobases contribute50. S. R. be indistinguishable from a repetitive sequence of identical bases. D. & Meller. D. Kasianowicz. farther apart than the read lengths of most current high-throughput recapturing a segment of DNA multiple times in the same nanopore sequencing platforms. D. The success of nanopore sequencing to detect youtube. E.M. Golovchenko. E. are appearing monthly. or therapeutic monitoring”43. Huang. translocations and inversion in mutant pancreatic cancer Technology Officer. Walker. ONT has not disseminated information about tumor-associated alterations and is further complicated by low tumor which nanopores are incorporated in the several different flow cell versions shipped cellularity (the relative proportion of tumor to normal cells in a sample) to MAP participants.. require testing different biological nanopores or pores through a single 10. & Troll. Bezrukov. Biol.. J. suggesting that the MinION and its successor because nanopore sequencing does not alter the DNA strand it reads. in most biopsy samples. 90. early detection. Natl. Characterization of the sensing region. M.075/aligned read base) npg 1. USA 93. be determined. Nyamwanda. 13472–13477 (2004). 177–190 nanopore sequencing.. 7. tumor suppressor genes diluted 1:100 with wild-type sequences shows CsgG52 nanopore. 70. (1986). 90.. base adducts. Acad. perhaps more important. D. portable DNA sequencers for monitoring and molecular relapse. current level to improve nanopore sequencing accuracy will probably Proc.-H.A.be). J. Current noise reveals protonation kinetics and number of ionizable sites in an open protein ion channel. Ionic channels formed by Staphylococcus aureus alpha-toxin: voltage- ratcheting mechanism will be an important step toward more accurate dependent inhibition by divalent and trivalent cations. Bezrukov. 655–662 (1994). Clive Brown. HG003703. within genomic DNA obtained from only a single cell.. Membr. Reducing the number of nucleobases affecting each 9. Mathé. A. Brutyan. & Branton. DNA heterogeneity and phosphorylation unveiled by single-molecule electrophoresis. Macromolecules 29. whose resolution would increase nanopore AUTHOR CONTRIBUTIONS sequencing accuracy to the level needed for single-nucleotide variant The three authors contributed equally to writing the manuscript. K. Acad.2 nm contribute to 5. the nanopore sequencing instruments will soon be routine next-generation accuracy usually achieved by high coverage of PCR-amplified material sequencing instruments. Phys. During final production of this Historical Perspective. generation sequencing poses similar problems for the detection of Notes added in proof: Until recently.Z. A. 190–199 (2006). a heptameric the current levels45. A pore-forming protein with with about two. Krishnasastry.. Song.. 8517–8522 (1996). polymers such as RNA and proteins.R.. NATURE BIOTECHNOLOGY VOLUME 34 NUMBER 5 MAY 2016 523 . They also demonstrated the algorithm’s Although nanopore sequencing technology was developed for ability to accurately classify sequence variants at far lower coverage than DNA. Sci. several nucleobases extending over a length of ~1. Nivon. L. usually by applying dynamic programming. Structure of staphylococcal alpha-hemolysin..nature. Biophys.com/watch?v=nizGyutn6v4&feature=youtu.A. 77. Akeson. Vodyanoy. 3227–3233 (1999). USA 102. so that an enzyme provides an imperfect ratchet in which COMPETING FINANCIAL INTERESTS the intervals between each advance of the DNA are variable. In an 8 March 2016 presentation for MAP participants (www.. problem is that several 3. Rapid nanopore discrimination between single polynucleotide molecules.J. USA Viterbi algorithm47.6 nm long32. J. ONT’s Chief deletions.. current levels. An improved a metal-actuated switch. These papers underscore level detection of cancer-associated… [sequence variants] needed for the future importance of inexpensive. Proc. base in the sequence. G. B. & Kasianowicz. simultaneously influence the current levels as a single DNA strand Sci. strand breaks and depurination than had previously been the case. E. D. Sci. It is during The authors declare competing financial interests: details are available in the online these time intervals that the nanopore current identifies each successive version of the paper. such as the 8.35 11. L. M. Sci. 2.com/ lost or overlooked in system noise and some of the long intervals may reprints/index. in many genes. (Oxford. traverses through MspA’s limiting aperture24. Acad. Nelson. & Kasianowicz. Butler. J. et al.

© 2016 Nature America. PLoS One 25.. Norris. 349–353 (2012). A complete bacterial genome assembled de Chem. W. 35. Int. F1000Res.. Pavlenok. T. Int. Proc.L. et al. Niederweis. Natl. Fan. Methods 4. Laszlo.L. Trias. et al. Garaj. Chem. J. Olasagasti. J. M. Emerg. Reading DNA at single-nucleotide resolution with a mutant MspA 46. Nat. generated on the MinION™ portable single-molecule nanopore sequencer. Niederweis. D. & Bayley. Y. Nature 516.L. et al. Theory 13. Ed. S. 33. Nucleobase recognition in ssDNA at the central constriction of Nat. J. 16. A. All rights reserved. H.M. K. R. 1087–1091 (2015). 31. B. et al. 54. Ashkenasy. Derrington. & Branton. Bayley. IEEE Trans. Nano Lett. Proc. 48.H.. M. Infect. S. J. et al. 32. S. et al.R. et al. Lett. Nanopore DNA sequencing with MspA. (2015). A. Peters. et al.J. base in an individual DNA strand: a step toward DNA sequencing in nanopores.795.. et al. 30.. Chu. 20647–20652 (2008). Nat. Mol. 6960–6968 (2012). Cloning of the mspA gene encoding a porin from Mycobacterium 52. Stoddart. Nat. 136. & Bayley. 102. 933–945 (1999). 18910–18915 (2013). USA 105. L37–L39 (2012). Methods 12. M. Acad. Real-time.A.. B. G. DNA base-calling from a nanopore using a methylcytosine. Baldarelli. & Ghadiri. Real-time 43. Nature doi:10. González-López. B. N. Viterbi. Xie. Schreiber. Gershow. Inc.. US patent DNA reads.R. Improved data analysis for the MinION nanopore sequencer. Nanopores discriminate among five 50. & Benz. Chem. USA 106. 829–833 (2014). novo using only nanopore sequencing data. Quick. Stoddart. M. J. 775–779 (2007). Nanopore monitoring of DNA polymerase function and stepwise single-nucleotide DNA strand sequencing detects structural variants in cancer.. J. M. Madoui. Genome assembly using Nanopore-guided long and error-free of individual polymer molecules based on monomer-interface interactions. portable genome sequencing for Ebola surveillance.H. H. 10. Lett.. & Loman. Acad. 30.L. Mikhailova. the alpha-hemolysin pore. Meller. Decoding long nanopore sequencing reads of natural DNA. Church. E. decoding algorithm. Natl. et al. Szalay. USA 107. 45. M. A. et al. P.. A. Wescoe. Maglia. 1401–1404 (2005). Proc.L. M.C.A. 44. 3435–3438 (2001). Cherf. Nucleotide discrimination with DNA immobilized in the MspA nanopore.1139236 (19 January 2016). BMC Genomics 16. 3633–3637 (2010). Maglia. a solid-state nanopore.T. npg 524 VOLUME 34 NUMBER 5 MAY 2016 NATURE BIOTECHNOLOGY . Microbiol. 24. Appl. Soc.M. Heron. 86. Biophys... Phys. Am.. Faller. Sequence-specific detection of individual DNA polymerase nanopores using PoreSeq.. 130.T.R. Eshleman. Characterization 38. S. Nat.. USA 110.782 (1998).A. Microbiol. et al. J. 13. & Gundlach. R. & Ghadiri.. et al. Error bounds for convolutional codes and an asymptotically optimum 18904–18909 (2013). 12192–12196 (2013). P. Nat.. Lieberman. 14. 33.. Benner.R. Acad. Chu. doi:10. Proc. Genome Biol. et al. Molecular dynamics study of MspA arginine mutants predicts 23.1075 (2015). A. Angew. Cockroft. I. 556–559 (2010). and analysis. 36.. 28. 21. Manrao. Chem. 16060–16065 (2010). J. Biotechnol. USA 110. Permeability of the cell wall of Mycobacterium smegmatis.. Nanotechnol. The structure of a mycobacterial outer. Acad. Nanopore sequencing as a rapidly deployable Ebola outbreak tool. 16.H. & Golovchenko. Biotechnol. Error rates for nanopore discrimination among cytosine. & Timp. Quick. Rev. E. Nat. D. membrane channel.-A. 818–820 (2008). Bhattacharya. Voltage-driven DNA translocations through a 51. Viterbi algorithm. Stoddart. 331–334 (2016). 5.2016. De novo sequencing and variant calling with 17.. Lu. 327 (2015). 106. at 5-Å precision.T. A single-molecule nanopore 41. 283–290 (1994). Derrington. 10106–10109 (2010). A..1080/ translocation through a protein nanopore. Butler. Biotechnol.. Hoenen. Cockroft. 2. Mol. Ther. J.. Quick. & Schulz. J.A. Sci. Mikhailova. HISTORICAL PERSPECTIVE 12. e25723 (2011). R. Structural and mechanistic insights into the bacterial amyloid smegmatis. Natl. Recognizing a single 33. 32. T. 26. Gigascience 3. Sci. J. 733–735 (2015). J. 47. Nature 487.. 132. 19. Sci. Acad. J. Biotechnol. J. S.. Recapturing and trapping single molecules with nanopore.. & Golovchenko. 39. Amorin. 14. J. F. A. 49. T. D. 53. & Ghadiri. 2. 22. L. 204 (2015). Derrington. J. & Akeson. MinION nanopore sequencing identifies the position and 18. Nivon. 5-hydroxymethylcytosine with nanopore MspA.. 40. Automated forward and reverse ratcheting of DNA in a nanopore slow DNA translocations and ion current blockades indicative of DNA sequence. J. M. G. Bolisetty. in complex mRNAs by nanopore sequencing.R. Szalay.A.A. N.J. 33.. Heron. Electrical pulse fabrication 27. Sci. nanopores. M. Detection and mapping of 5-methylcytosine and 6. Jain. D. Soc. Chem. 37.. M. Dis. et al. M.. discrimination in immobilized DNA oligonucleotides with a biological nanopore. 344–348 (2012).. et al. M. Molecule-hugging graphene C5-cytosine variants in DNA. J. Z. 296–300 nanopores. Nat. I. J. nanopore and phi29 DNA polymerase. (2010).J. Single-nucleotide 34. & Golovchenko. & Aksimentiev. 7702–7707 (2009).E. Processive replication of single DNA molecules in a nanopore 44. Single-molecule analysis of DNA-protein complexes using structure of a bacterial antibiotic resistance island. Cancer Biol. Nanotechnol.E. G. 5. Replication of individual DNA molecules under electronic control 42. Angew.M. Quinlan. et al.. Am. Golovchenko. Chem. Niederweis.H. D. Laszlo.A. Proc..Z. Multiple base. Am.. Soc. M. Natl. Accurate whole-genome sequencing and haplotyping from 10 to catalyzed by phi29 DNA polymerase. M. J. secretion channel CsgG. 49. Sci.R. Branton. of graphene nanopores in electrolyte solution. Nat. M. 203109 Biotechnol.  Ip. (2015).R.J.M. M. A. Workman. 22. H. 29. Niederweis. Loman. Acad. et al. A reference bacterial genome dataset device detects DNA polymerase activity with single-nucleotide resolution. recognition sites in a biological nanopore: two heads are better than one. Sci. 49. MinION Analysis and Reference Consortium: phase 1 data release 15. E. T. Liu. 798–806 (2010). I. Methods 12. Deamer. Rajadinakaran. Timp.J. 4. Single-molecule DNA detection with an engineered MspA protein nanopore. & Simpson. G. complexes in real time using a nanopore. J. Nanotechnol. Sánchez-Quesada. 260–269 (1967). S. 315–317 (2007). G. et al. Ed. ACS Nano 6. B. C. Nat. J. J. D. D. Natl. Comer. 1189–1192 (2004). 250-253 (2014).. Amorin. 30. N. Kuan. M. 17961–17972 20 human cells. Inf. & Gundlach. et al. 15384047. G. Phys. Proc. M. 22 (2014). Nat. Goyal.. Manrao. 16582–16587 (2014). W. & Branton. 190–195 (2012). 20. 351–356 (2015).M. Science 303. E. and hydroxymethylcytosine along individual DNA strands. & Kasianowicz. P. Schreiber. Pavlenok. Hornblower. Determining exon connectivity using a protein nanopore. USA 110.1038/nature16996 (3 February 2016). A. et al. Ashton. A. 718–724 (2007). Natl. Angew. & Graveley.