Nanopore-Based Target Sequence Detection

The promise of portable diagnostic devices relies on three basic requirements: comparable sensitivity to established platforms, inexpensive manufacturing and cost of operations, and the ability to survive rugged field conditions. Solid state nanopores can meet all these requirements, but to achieve high manufacturing yields at low costs, assays must be tolerant to fabrication imperfections and to nanopore enlargement during operation. This paper presents a model for molecular engineering techniques that meets these goals with the aim of detecting target sequences within DNA. In contrast to methods that require precise geometries, we demonstrate detection using a range of pore geometries. As a result, our assay model tolerates any pore-forming method and in-situ pore enlargement. Using peptide nucleic acid (PNA) probes modified for conjugation with synthetic bulk-adding molecules, pores ranging 15-50 nm in diameter are shown to detect individual PNA-bound DNA. Detection of the CFTRΔF508 gene mutation, a codon deletion responsible for ∼66% of all cystic fibrosis chromosomes, is demonstrated with a 26-36 nm pore size range by using a size-enhanced PNA probe. A mathematical framework for assessing the statistical significance of detection is also presented.

The 12 bp binding γPNA (PNA Bio) has the sequence GAA*AGT*GAA*AGT where * indicates the location of a biotin backbone modification that was incorporated via coupling to free amines on a lysine residue.
For the 12 bp binding γPNA study, the 5.6 kbp DNA scaffold was first heated to 95 • C for 2 minutes, cooled to 60 • C, and then allowed to incubate with a 10-fold excess of γPNA to possible binding sites for 1 hr. Excess γPNA was then dialyzed out of the sample using a 20kDa MWCO dialysis cup (Thermo Scientific). This DNA/γPNA complex was subsequently labeled with a 10-fold excess of neutravidin (Thermo Scientific) to biotin sites. All molecules were stored at -20 • C prior to use.
Detection of multiple sites of a sequence in 5.6 kb DNA The linear 5.6 kbp dsDNA molecule was engineered to contain the 12 bp sequence interspersed at 25 sites within the DNA (Fig. S15). The purpose of this repetition is to boost the sensing signal for each scaffold, since the more occupied PNA sites there are, the longer the nanopore current is impeded, yielding a more easily detected signature. Instead of using bisPNA as the sequence-specific binding molecule, we used the smaller and more versatile γPNA. The work in [1] shows that positive detection and localization of these smaller PNAs is possible, but required a precision sub-4nm pore, and was shown to work with a salt gradient. In the absence of a salt gradient and with a larger pore, we sought to demonstrate positive detection of the presence of the probe by adding bulk to the size of each γPNA. To provide the option of increasing probe size, the γPNA had three biotin molecules incorporated via coupling to free amines on the backbone Lysine amino acid, and using the protein neutravidin as the bulk-adding payload that can bind to each biotin.
After testing complex stability by gel electrophoresis (Fig. S16a), the DNA/γPNAneutravidin (D/P-N) reagent was tested in a nanopore assay. Figures S16b-d show data comparing δG versus duration distributions for events from three separate experiments conducted on comparable pores. Specifically, DNA alone was tested in an 11 nm pore, neutravidin alone was tested with a 12 nm pore, and the D/P-N reagent was tested with an 11 nm pore, all pores being formed in 10 nm SiN membranes by controlled dielectric breakdown. DNA alone at 2.7 nM yielded 1301 events over 16 minutes, neutravidin alone at 210 nM yielded 2589 events over 11 minutes, and D/P-N complexes at 160 pM yielded 4198 events over 7.3 minutes, with a clear population of fast events present due to a significant amount of free neutravidin as background. Nonetheless, the largest δG events in the D/P-N experiment are attributed to D/P-N complexes (Fig. S16b), providing a simple criteria for tagging events in detecting DNAs that have one or more target 12 bp sequences bound by the γPNA-neutravidin probe.
The mathematical criteria derived can be used to assess confidence in detection. Using the criteria δG > 16 nS, 390 of the events in the D/P-N experiment are tagged resulting in Q(p) =9.29%. In the prior control experiments, only 0.46% of D and 0.16% of N events are flagged. Though the DNA/γPNA and γPNA-neutravidin controls were not run on this pore, they are unlikely to produce events with δG > 16 nS, even with folding of the DNA (γPNA create smaller bulges than bisPNA). Regarding false positives, one question is if there are complexes in which neutravidin non-specifically binds to the DNA scaffold; if so, such complexes could cause events with δG > 16 nS and thus produce a high false positive percentage. The 10X excess neutravidin-to-biotin conditions used do not show an aggregation effect in the presence of γPNA binding to the 25 targets, although aggregation is observed for 36X excess (Fig. S16a). Also, a separate gel showed no visual difference between DNA alone and DNA with 20X excess neutravidin in the absence of γPNA (Fig. S17). These findings suggest that such complexes are unlikely to exist, and therefore would not contribute appreciably to the false positive population. Collectively, these results suggest 0.46% can be used as a (max) false positive probability using the criteria δG > 16 nS for this nanopore. The 99% confidence test is satisfied, with Q * = 1.15% and Q(p) − Q * = 8.14% greater than Q(0) = 0.46%. The estimated time to results (first time to 99% confidence) is 18 seconds, exceedingly fast for this data set due to the high capture rate. We note that by exploiting the presence of multiple levels within events [2], which were observed in many D/P-N events including those with δG < 16 nS, a stronger flagging criteria could be identified to achieve a positive detection result in less time and/or at a lower concentration of molecules.
To test if there are complexes in which neutravidin non-specifically binds to the DNA scaffold, a gel was run and showed no visual difference between DNA alone and DNA with 20X excess neutravidin in the absence of γPNA (Fig. S17). These findings suggest that such complexes are unlikely to exist, and therefore would not contribute appreciably to the false positive population.  Figure S17: An EMSA assay shows that excess neutravidin (20X) does not bind non-specifically to the DNA scaffold. A 4-20% PAGE EMSA comparing: 1) high molecular weight sizing ladder, 2) the 5.6 kb DNA scaffold, 3) DNA/γPNA-neutravidin complex at 10x excess neutravidin to biotin (as in nanopore recording conditions), and 4) DNA with 20x excess neutravidin in the absence of γPNA. Compared to lanes 1 and 2, only 1/10 of the amount was loaded into lane 3, resulting in a band with lower intensity.