Oligonucleotides (Supplementary Table S1) were synthesized by Hippo Technologies Co. Ltd., China. For the formation of DNA segments of EGFR and L858R, the sense and antisense sequences were added to one tube, and for the formation of the heteroduplex of EGFR and L858R, the EGFR sense sequences and L858R antisense sequences were added to the tube and vortexed briefly. The mixture was denatured by heating at 95 °C for 5 min and then cooling down (−0.1 °C/3 s) to 25 °C using the Bio-Rad CFX96 Real-Time System (Bio-Rad, United States).

A DNA capture probe (Supplementary Table S1) (Hippo Technologies Co. Ltd., China) was designed to hybridize with EGFR (fully complementary) and L858R (single-nucleotide mismatched), and it was modified with biotin at its 5′ends. Target DNA solutions were prepared in a 2× hybridization buffer (10× SSPE, 10× Denhardt’s, 10 mM EDTA, and 0.2% SDS). After heating to 95 °C for 5 min, the solution was mixed with the pre-warmed (2 min at 58 °C) capture probe and incubated at 58 °C for 4 h. The hybridization mix was then added to streptavidin magnetic beads (MedChemExpress). Streptavidin magnetic beads (Supplementary Figure S1) were prepared according to the manufacturer’s instructions. Briefly, 0.5 μL stock beads solution (10 mg/mL) was first washed three times with 200 μL wash buffer (1× TE, Yuanye Bio-Technology Co., Ltd. China; 1 M NaCl, GEN-VIEW Scientific, Inc. United States; 0.1% Tween-20, Sangon Biotech (Shanghai) Co., Ltd. China; pH 7.5), and the supernatant was removed after magnetic separation using a magnetic stand. Then, the magnetic beads were resuspended with a 20 μL wash buffer. The hybridization mix (20 μL) was added to the resuspended magnetic beads. After 1 h at 25 °C, the beads were removed and washed three times. The magnetic beads were then resuspended with 17 μL nuclease-free water.

We added 1 μL T7E1 (New England Biolabs, United States) and 2 μL 1× NE buffer (50 mM NaCl, 10 mM Tris-HCl, 10 mM MgCl2, 1 mM DTT) to the 17 μL heteroduplexes. Then, we performed T7EI digestion at 37 °C using the Bio-Rad CFX96 Real-Time System.

Native polyacrylamide gel (12%) was prepared by mixing thoroughly 30% acrylamide-bis-acrylamide (29:1) 2.4 mL, 5× TBE buffer (89 mM boric acid, 2 mM EDTA, 89 mM Tris, pH 8.3) 1.0 mL, 60 μL of 10% ammonium persulfate, 6 μL tetramethyl ethylenediamine, and deionized water 2.4 mL. We loaded 5 μL DNA samples into the prepared native polyacrylamide gel, and the gel electrophoresis was performed in 1× TBE buffer at 150 V for approximately 1 h. The gels were imaged on a Chem-I Doc XRS system (Bio-Rad, United States) after staining with Realsafe-Red nucleotide gel stain (Real-Times Biotechnology Co., Ltd., China) for 30 min.

Solid nanopores were prepared as described previously (Tian et al., 2023). Briefly, the chip (NORCADA Inc., Canada) was embedded in a customized flow cell filled with electrolytes of 10 mM Tris-HCl, 1 M KCl, and 1 mM ethylenediaminetetraacetic acid (EDTA, pH 8.0). Ag/AgCl was used as the reference electrode, and nanopores were prepared on the SiNx film using the electric pulse puncture method. Theoretical estimation of nanopore size can be made using the following formula: G = σ(4L/π d2+1/d)-1 (Yin et al., 2017; Kowalczyk et al., 2011). Where G is the conductance of the nanopore, which can be measured using I-V testing of the nanopore (Supplementary Figure S2), σ is the conductivity of the buffer solution (1 M KCl, 1 mM EDTA, 10 mM Tris, pH 7.6), σ at S·m^-1, L is the length of the SiNx film (in this work, it is 20 nm), and d is the nanopore diameter. The fabricated nanopore chips were integrated into polydimethylsiloxane (PDMS) microfluidic chambers. Ag/AgCl electrodes were placed in each reservoir and connected to an Axopatch 700B pA current amplifier (Molecular Devices) to apply the transmembrane bias and record ionic currents through the nanopore. Between the analysis of two samples, the nanopore was rinsed three times with alcohol and deionized water to ensure cleanliness and prevent cross-contamination. Each sample was tested three times to ensure reproducibility.

Nucleic acids were added to the cis side of the flow cell containing the solid-state nanopores. A bias voltage was applied to drive the nucleic acids through the pores for analysis. The current was measured using an Axopatch 200 B patch-clamp amplifier (Molecular Devices, CA, United States). The Clampfit and Origin software were used for signal extraction and data analysis, respectively.

All quantitative data are presented as means ± SD, statistical significance was evaluated using one-way ANOVA, followed by the Dunnett’s t-test.

SNVs serve as critical predictors of tumor progression and therapeutic resistance while enabling the identification of actionable drug targets for personalized treatment strategies (Łaźniak et al., 2023; Watanabe et al., 2018; Kim et al., 2020; Alanee et al., 2017). However, detecting SNVs remains technically challenging due to their low mutation frequency (<5%) and single-base resolution requirements (Li et al., 2021). Despite these obstacles, advances in molecular technologies are gradually overcoming these difficulties, enabling the precise identification of oncogenic mutations that drive tumorigenesis. In this study, we synthesized DNA fragments containing mutations in the epidermal growth factor receptor (EGFR) gene to serve as detection targets (Supplementary Table S1). The EGFR L858R point mutation, located in exon 21 and classified as a type II kinase domain mutation, involves a thymine-to-guanine substitution at codon 858 of the EGFR gene. This nucleotide change leads to the replacement of leucine with arginine at the corresponding position in the EGFR protein (Liu et al., 2024). Positioned within the N-terminal region of the activation loop, the mutation replaces a hydrophobic leucine residue with a bulkier, positively charged arginine, which is structurally accommodated in the active conformation of the kinase (Yun et al., 2007). Functionally, L858R is an activating mutation that enhances EGFR’s catalytic efficiency by 20–50 times compared to the wild-type protein (Sutto and Gervasio, 2013). As a key oncogenic driver, L858R plays a critical role in the development of several malignancies, including lung, colorectal, and head and neck cancers (Andrews et al., 2023; Castellanos et al., 2017). The mutation detection strategy leverages two distinct biochemical properties: 1) the T7 endonuclease I (T7E1) enzyme’s ability to specifically recognize and cleave single-base mismatches, particularly those involving cytosine residues (Guan et al., 2004); and 2) nanopore-based single-molecule sensing capabilities. To investigate the enzymatic specificity, we systematically evaluated T7E1 activity against EGFR/L858R heteroduplex structures. T7E1 digestion was performed at its optimal temperature of 37 °C for 30 min (Supplementary Figure S3). Under these conditions, the enzyme specifically cleaved EGFR/L858R heteroduplexes at single-base mismatch sites, producing two distinct DNA fragments, whereas wild-type and mutant homoduplexes remained intact (Figure 1A). This finding confirms the enzyme’s discrimination capacity while revealing previously unreported cleavage dynamics specific to EGFR/L858R heteroduplexes. The experimental design addresses a critical knowledge gap, as the structural interplay between T7E1 enzymatic activity and EGFR mutation-bearing duplexes had not been systematically characterized prior to this investigation.

T7E1 is a key enzyme used in genetic engineering and molecular biology, and it is a type of endonuclease that is commonly used for gene editing and mapping. Moreover, it can be used for single nucleotide polymorphism (SNP) detection owing to its ability to recognize and cleave DNA strands containing mismatched base pairs. The T7E1 enzyme is often employed in conjunction with other molecular tools, such as CRISPR/Cas9, to induce double-stranded DNA breaks at mismatched sites in the genome. The optimal temperature for the T7E1 digestion is 37 °C. In previous experiments, five to ten units of the T7E1 enzyme could achieve high cleavage efficiency for 100 ng of DNA in a 20 μL reaction system (Zou et al., 2023; Seol et al., 2024). In this experiment, we first performed T7EI digestion at 37 °C for 30 min using five units of T7E1 in a reaction volume of 20 μL (Figure 1A). The optimal enzymatic efficiency was achieved when the enzyme time was 30 min, whereas increasing the time to 1 h did not significantly improve the enzymatic efficiency of the T7E1 enzyme (Figure 1B). Increasing the enzyme concentration or cutting time may lead to the nonspecific cutting of fragments. The DNA concentration was then changed. The results indicate that although the amount of cleavage products decreased with lower DNA concentrations, more than half of the mismatched DNA at a concentration of 0.1 μM was still digested by five units of T7E1 at 37 °C for 30 min in a 20 μL reaction volume (Figure 1C). This suggests that T7E1 is sufficiently efficient for detecting low concentrations of EGFR/L858R heteroduplexes under these conditions (Figure 1D). Thus, using five units of T7E1 at 37 °C for 30 min provides adequate enzymatic efficiency for subsequent analysis.

Magnetic separation is an efficient technology for separating and purifying nucleic acids from chemical and biological suspensions (Berensmeier, 2006). For target DNA capture, a DNA capture probe modified with 5′-biotin was designed and custom-synthesized. The probe was fully complementary to EGFR but mismatched with L858R. To enhance binding specificity and stabilize interaction with a target containing a single-base mismatch, we designed a 130-bp probe (Supplementary Table S1). (Supplementary Table S1). Excess probes were added to mixtures containing the wild-type and mutant genes. After denaturation and gradual annealing, the probes hybridized randomly with the wild-type and mutant DNA. Streptavidin magnetic beads were used to separate the probes. Nucleic acids that were not bound to the probes were discarded along with the supernatant. If the sample contains a mutation, the probes can capture it and form single-base-mismatch heteroduplexes. Thus, a single-base mismatch could be recognized and cleaved by the T7E1 enzymes. The oligonucleotides combined with the magnetic beads through the probes were removed using magnetic separation, and the free fragment nucleotides were left in the supernatant. In this study, our results showed that 46 bp fragment nucleotides were detected by native polyacrylamide gel electrophoresis in the L858R-containing mixture after T7E1 enzyme digestion and magnetic separation (Figure 2). No band was detected for EGFR without L858R (Figure 2). In conventional detection, interference may arise not only from wild-type EGFR but also from other normal or mutated genes in the human genome. To further verify the specificity of our method, we conducted validation experiments using the K-RAS gene and its G12D mutant, which are unrelated to EGFR but commonly found in cancer samples. Results showed that no bands were observed for the K-RAS gene or samples containing the G12D mutation, indicating that the capture probe is specific to EGFR and cannot capture K-RAS or G12D (Supplementary Figure S4). However, native polyacrylamide gel electrophoresis cannot detect low concentrations of nucleotides. Therefore, we developed a highly sensitive nanopore detection method.

To determine a suitable nanopore detection voltage, we synthesized a nucleic acid fragment with the same sequence as the free fragment (Supplementary Table S1) after enzyme digestion. The translocation of the DNA fragments into the nanopore was investigated at a series of voltages ranging from +100 to +500 mV. The nanopores employed in our experiments had an approximate diameter of 3 nm. In previous studies, the 3 nm nanopore demonstrated superior analytical performance for nucleic acid analysis (Alam et al., 2024). DNA fragments at a final concentration of 10 nM were added to the cis chamber of the flow cell. The translocation of molecules through nanopores requires overcoming free energy barriers, such as geometric confinement at the pore entrance and electrostatic repulsion (Chinappi et al., 2020). At low voltages, the electrophoretic force is insufficient to overcome these barriers, preventing effective entry into the nanopore. When voltages of +100 and +300 mV were applied, no identifiable signals were observed under voltage-clamp conditions. Therefore, there was no blocking signal in the current trajectory and no discernible signals were detected in DNA-free blank samples (Figure 3A-E). Consequently, a threshold voltage (500 mV) is necessary to lower the barrier via enhanced electric field energy, leading to a nonlinear relationship between the capture rate and the applied voltage (Grosberg and Rabin, 2010). The observed open-pore current was around 1,425 pA. Thus, a series of current signals was detected at 500 mV (Figure 3). The raw translocation traces are shown in Figure 3F, and the scatter plots are presented in Figure 3G. According to the Gaussian fit of the histogram, the average blocking amplitude at 500 mV was 1,091.92 ± 8.63 pA (Figure 3H). Under an applied voltage of 500 mV, the average residence time of the blocking signal was 0.19 ± 0.01 m (Figure 3I).

To evaluate the feasibility of nanopore detection of enzyme digestion products, we measured translocation events at a voltage of 500 mV. As negative controls, samples containing only buffer (Figure 4A) or enzyme (Figure 4B) showed no detectable pulses. Similarly, no signals were observed for wild-type EGFR (Figure 4C), as the absence of mismatches prevented enzymatic digestion. In samples containing the L858R-mutated EGFR gene, hybridization with the probe introduced a single-base mismatch, which enabled T7E1 to recognize and cleave the heteroduplex, releasing short nucleic acid fragments. After magnetic bead separation, free nucleic acids in the supernatant generated discernible pulse signals in nanopore detection (Figure 4D). The raw translocation traces and scatter plots are shown in Figure 4E. The amplitude of the current blockade was 1,125.27 ± 5.50 pA (Figure 4F). Histograms of the dwell times are shown in Figure 4G. Specificity was further validated using the K-RAS gene and its G12D mutant; in this case, probe incompatibility resulted in no digestion or nanopore signals (Supplementary Figure S5). These results confirm the system’s ability to discriminate single-nucleotide variations with minimal background interference, underscoring its potential for precise mutation detection.

Detecting SNVs presents several challenges, primarily due to the need for high sensitivity and specificity. The low abundance of certain variants, particularly in heterogeneous populations, makes it difficult to distinguish them from background noise. In our study, 50 μL of synthesized EGFR and L858R gene solutions (0.1 nM) were aliquoted into a series of centrifuge tubes, with L858R-to-EGFR concentration ratios set at 10%, 5%, 1%, 0.3%, and 0.1%. After magnetic bead enrichment and purification, enzyme digestion, and repurification, mutated genes were detected at levels as low as 0.1% (Figure 5). The raw translocation traces and scatter plots are shown in Supplementary Figure S6A–E. The amplitudes of the current blockade are shown in Supplementary Figure S6F–J. Supplementary Figure S6H–O shows the histograms of dwell times. This method demonstrates effective SNV detection, with a detection limit of 0.1%, compared with 0.5% for NGS (Yang et al., 2017). Although a positive correlation between nanopore signal intensity and variant allele frequency (VAF) was observed, under the current experimental conditions, the reproducibility and linearity of this correlation were limited, preventing the establishment of a fully reliable calibration curve or regression model. Therefore, the assay is currently presented as a qualitative method for detecting the target mutation. Further optimization may enhance its potential for quantitative applications in future studies.