DNA sequencing technologies that use single unamplified DNA templates—sometimes called single-molecule sequencing or third-generation sequencing—avoid the biases introduced by PCR, and have the potential for producing very long sequences at low cost. However, sequence accuracy can be an issue.
Pacific Biosciences systems
The PacBio RSII system, released in 2010, was heralded as the first DNA sequencing method to sequence single unamplified molecules in real time. The sequencing templates are double-stranded DNA molecules that have single-stranded hairpin oligonucleotides ligated to each end (Figure 1A). A single sequencing primer is annealed to one of the hairpins.
Fig1. Single-molecule real-time sequencing using the PacBio system. (A) The template for PacBio sequencing is a double stranded DNA with single-strand hairpins (green) ligated to either end. A sequencing primer (red) anneals to one of the hairpins. The strand-displacing polymerase (gray) moves the template continuously round, producing concatenated copies of the whole sequence, as detailed in Travers et al. (2010). (B) The polymerase is anchored at the bottom of a well; the four phospho-labeled dNTPs diffuse freely. (C) When a nucleotide is incorporated into the growing chain, its fluorescent label remains at the bottom of the well much longer than the freely diffusing dNTPs. (A, adapted from Travers KJ et al. [2010] Nucl Acid Res 38:e159; PMID 20571086; B and C, adapted from Eid J et al. [2009] Science 323:133–138; PMID 19023044. Reprinted with permission from the AAAS.)
Sequencing takes place in a SMRT Cell, a fabrication containing 150,000 tiny wells (capacity per well = 10−21 liters) called zero-mode waveguides. A single DNA polymerase molecule is anchored to the bottom of each well. Dye-labeled dNTPs diffuse in and out of the wells (Figure 1B). The dye labels are on the terminal phosphates of the dNTPs, so that on incorporation the dye is lost and totally natural DNA is synthesized. A high- processivity, strand-displacement DNA polymerase is used, such as Φ29 or Bst polymerase. The synthesis point moves round and round the template, making long concatemers of the sequence and using both the sense and antisense strands as template in its journey round.
When the polymerase binds and then incorporates a dNTP, it increases the time the attached label spends at the bottom of the well, compared to the time spent by randomly diffusing dNTP molecules, as recorded by the laser imaging system (Figure 1C). The duration represents the real-time dynamics of the polymerase, and this may be different when template bases carry epigenetic modifications such as methylation. Thus the system has the unique ability to identify patterns of modification directly from the raw data. For the principles of the system and discussion of the templates, see Eid et al. (2009) (PMID 19023044) and Travers et al. (2010) (PMID 20571086) in Further Reading.
The PacBio RSII allows extremely long reads (20 kb or more), and both sample preparation times and run times are short, allowing the whole procedure to be completed in a single day. The error rate per nucleotide, at around 11%, is much higher than in competing systems, but the errors are random and can be compensated by allowing the polymerase to run through the same circular template many times. The throughput is lower and the cost per base higher than for many competing systems. However, the long reads make it ideal for de-novo sequencing of small bacterial and viral genomes, and for sequencing low-complexity regions or structural variants in human and other genomes.
Oxford Nanopore Technologies system
Oxford Nanopore are developing a competing third-generation system. In the MinION device, released to early-access users in 2014, a flow cell contains maybe 500 wells, each of which is spanned by a synthetic, electrically resistant membrane in which a single nanopore is anchored. The nanopores are made of modified α-hemolysin protein. Single-stranded DNA feeds through the 10 μm long × 1 μm wide pore. As the different sized nucleotides pass through, they block the ionic current flowing through the pore to different extents, potentially allowing each nucleotide to be recognized. In practice, the blocking effect depends on at least five contiguous nucleotides and must be decon voluted to identify individual nucleotides. To get signals for analysis, the passage of the DNA through the pore must be slowed down by several orders of magnitude, and this is achieved by coupling it to a relatively slow-moving processive enzyme. The test DNA is double-stranded; the leader end has a single-strand extension coupled to the motor enzyme, while the far end can be ligated to a hairpin oligonucleotide carrying a second motor enzyme, thus allowing both strands to be sequenced (Figure 2).
Fig2. Feeding test DNA through a nanopore. Oxford Nanopore’s sequencing strategy requires DNA templates to be ligated with two adaptors. The first adaptor is bound with a motor enzyme as well as a tether, whereas the second adaptor is a hairpin oligonucleotide that is bound by the HP motor protein. Changes in current that are induced as the nucleotides pass through the pore are used to discriminate bases. The library design allows sequencing of both strands of DNA from a single molecule (two direction reads). (From Reuter JA et al. [2015] Mol Cell 58:586–597; PMID 26000844. With permission from Elsevier.)
Nanopore sequencing offers great promise. Potentially the read length is limited only by the length of the test DNA. Sample preparation is simple, the process is quick, and the device is small and simple enough to be portable. The big problem is the error rate, with insertion, deletion, and substitution rates of 4.9%, 7.8%, and 5.1%, respectively, reported by Jain et al. (2015) (PMID 25686389; see Further Reading). Reuter et al. (2015) (PMID 26000844; see Further Reading) also report a high run-failure rate. However, the device has already demonstrated its use in sequencing a previously unresolved, highly- repetitive region of human chromosome X (see Jain et al. 2015), and with increases in reliability and accuracy, it has many possible applications.
Other technologies
Many alternative massively-parallel sequencing technologies are available or under development by a variety of companies. The following list is intended to give a flavor of the current technical ferment, without in any way claiming to be comprehensive.
• Complete Genomics (http://www.completegenomics.com)
• Genapsys (http://genapsys.com)
• Genia (http://www.geniachip.com)
• Gnubio (http://gnubio.com)
• Lasergen (http://lasergen.com)
• Nabsys (http://nabsys.com)
• Stratos (http://stratosgenomics.com)
• ZSG (http://www.zsgenetics.com)
