A generalizable nanopore sensor for highly specific protein detection at single-molecule precision

Identifying and quantifying protein biomarkers is a pressing demand in precision and personalized medicine, as many hematological malignancies and solid tumors involve altered protein expression. A major challenge for protein detection in biofluids is nonspecific binding of assay reagents to surfaces and recognition elements, which raises background noise and limits dynamic range at low analyte levels. Single-molecule nanopore sensing using the resistive-pulse technique has advanced peptide and protein analytics, including detection of shape, size, post-translational modifications, enzymatic activity, unfolding, and protein fingerprinting. However, conventional nanopore approaches require protein entry into the pore lumen, imposing steric constraints that can weaken specific interactions, and preclude detection of proteins larger than the pore. Detecting single proteins outside the nanopore via a covalently attached external binder offers a practical alternative, but demands a generalizable transduction mechanism that converts binding into a specific electrical signature and an easily adaptable architecture across diverse binder–analyte pairs. Previous sensors faced limitations from binder size, steric hindrance at the pore opening, and accessibility of the interaction interface. To address these issues, the authors propose a modular class of sensing elements comprising an antibody-mimetic protein binder (monobody) tethered to the N-terminus of the monomeric β-barrel nanopore tFhuA via a flexible (GGS)2 linker. Monobodies (FN3 scaffolds) offer small size, lack of disulfide bonds, straightforward production, and wide target specificity, enabling a general strategy to detect distinct proteins while maintaining sensor architecture, specificity, and sensitivity.

The study builds on extensive literature demonstrating the capabilities of nanopore sensors for single-molecule peptide and protein analyses, including discrimination at single–amino acid resolution, detection of protein shape and size, post-translational modifications, enzymatic activities, mechanical unfolding, and efforts toward protein sequencing. Prior approaches often required protein partitioning into the pore interior, imposing steric constraints. External recognition strategies (e.g., aptamer-equipped nanopores and protein fragment binders) had challenges generalizing across targets due to binder heterogeneity, steric hindrance at the pore entrance, and limited accessible interfaces. Monobodies (FN3 domain-based antibody mimetics) have been shown to be versatile, small, stable, and free of disulfide bonds, with broad targetability, making them promising candidates for creating generalizable, tethered external binding interfaces on nanopores. The authors aim to overcome limitations of existing techniques by using monobody–tFhuA fusions to enable specific, label-free, real-time detection of diverse protein targets outside the pore lumen with distinct electrical signatures.

Design and sensor construction: Three monobodies (FN3SUMO against hSUMO1, Mb4 against WDR5, and Adnectin1 against the EGFR ectodomain) were genetically fused to the N-terminus of the monomeric β-barrel nanopore tFhuA via a flexible (GGS)2 peptide tether, forming single-polypeptide fusion sensors (FN3SUMO-tFhuA, Mb4-tFhuA, Adnectin1-tFhuA). AlphaFold2 was used for in silico structural predictions of the fusion proteins to visualize monobody orientation and assess potential pore occlusion. Synthetic gene construction and protein production: cDNAs encoding fn3sumo, mb4, and adnectin1 were fused 5' to tfhua using restriction-free cloning. Constructs were expressed in E. coli BL21(DE3), purified from inclusion bodies using detergent solubilization, anion-exchange chromatography, and size-exclusion chromatography. Analytes were prepared as follows: hSUMO1 expressed in E. coli and purified by ion exchange and size exclusion; WDR5 expressed in ROSETTA cells, purified via IMAC, TEV cleavage, nuclease treatment, and polishing; EGFR ectodomain expressed and secreted from Expi293F cells, purified by IMAC and buffer exchange. Refolding: Fusion sensors were refolded in 1% DDM detergent followed by extensive dialysis into 200 mM KCl, 20 mM Tris-HCl, pH 8, at 4 °C to obtain functional proteomicelles. Single-channel electrophysiology: Planar lipid bilayers (1,2-diphytanoyl-sn-glycero-phosphatidylcholine) were formed across a ~100 µm aperture. Fusion nanopores and analytes were added to the cis side. Currents were recorded with an Axopatch 200B amplifier, sampled at 50 kHz, typically at +40 mV (Adnectin1-tFhuA also at +20 mV due to noise considerations), with additional analog and digital Bessel filtering. Buffer: 300 mM KCl, 10 mM Tris-HCl, pH 8.0, with TCEP as appropriate for each analyte. Event detection and kinetic analysis employed maximum likelihood estimation and logarithm likelihood ratio (LLR) tests to determine single- versus multi-exponential distributions, yielding association (kon) and dissociation (koff) rates and KD. Orthogonal and validation assays: Biolayer interferometry (BLI) was conducted using biotinylated Mb4-tFhuA proteomicelles immobilized on streptavidin sensors to qualitatively assess fast WDR5 binding kinetics (beyond BLI time resolution). Steady-state fluorescence polarization (FP) anisotropy measured binding of fluorescein-labeled hSUMO1 and rhodamine-labeled WDR5 to fusion sensors across concentration series to estimate KD in solution. EGFR detection in complex matrices was tested by spiking 20 nM EGFR into 5% (v/v) fetal bovine serum (FBS) and analyzing event signatures, noise spectra (PSD), and kinetics for quantification using Ton-2 and kon-2.

- Sensor architecture and baseline conductance: AlphaFold2 predicted monobodies oriented to partially occlude tFhuA, reducing conductance relative to unmodified tFhuA (1.5 ± 0.1 nS). Measured unitary conductances (mean ± s.d.): FN3SUMO-tFhuA 0.81 ± 0.03 nS; Mb4-tFhuA 0.99 ± 0.04 nS; Adnectin1-tFhuA 0.90 ± 0.02 nS. - hSUMO1 detection (FN3SUMO-tFhuA, +40 mV): Nanomolar hSUMO1 produced frequent reversible blockades between Oon and Ooff with normalized amplitude A/I0 = (91.5 ± 0.7)%. Event frequency scaled linearly with [hSUMO1], Ton decreased with increasing [hSUMO1], and Toff was independent of [hSUMO1], indicating bimolecular association and unimolecular dissociation. Kinetics (mean ± s.e.m.): kon = (1.12 ± 0.02) × 10^5 M^-1 s^-1; koff = 74.5 ± 2.4 s^-1; KD = 665 ± 24 nM. - WDR5 detection (Mb4-tFhuA, +40 mV): WDR5 induced modest-amplitude blockades (normalized amplitude (14 ± 1)%) absent in controls, with single-exponential Ton and Toff distributions. Frequency scaled with [WDR5]; Toff independent of [WDR5]. Kinetics (mean ± s.e.m.): kon = (0.83 ± 0.01) × 10^6 M^-1 s^-1; koff = 72.4 ± 3.7 s^-1; KD = 872 ± 45 nM. BLI qualitatively confirmed rapid association/dissociation beyond BLI time resolution. - EGFR detection (Adnectin1-tFhuA, +20 mV): EGFR yielded reversible blockades with two amplitude populations and two kinetic subpopulations. Normalized blockade peaks centered at ~65% and ~85% with probabilities ~0.72 and ~0.28, independent of [EGFR]. Ton followed single-peak distributions; Toff showed two components: short-lived (Toff-1 ≈ 0.07–0.08 s) and long-lived (Toff-2 ≈ 0.8–1.2 s). Association rates (mean ± s.e.m.): kon-1 = (6.62 ± 0.21) × 10^7 M^-1 s^-1; kon-2 = (2.89 ± 0.10) × 10^7 M^-1 s^-1. Dissociation rates: koff-1 = 12.0 ± 0.4 s^-1; koff-2 = 1.01 ± 0.01 s^-1. Equilibrium dissociation constants: KD1 = 181 ± 8 nM (short-lived); KD2 = 34 ± 2 nM (long-lived). Interconversion between captured substates was not supported by LLR tests. - Specificity and orientation controls: No significant blockades from analytes on unmodified tFhuA or when added to the trans side; Adnectin1-tFhuA showed only nonspecific brief spikes with noncognate analytes (hSUMO1, WDR5), indicating high specificity and single orientation insertion. - Detection in biofluid: In 5% FBS, EGFR events persisted with two blockade levels; PSD indicated transition to 1/f noise due to serum constituents. Long-lived EGFR blockades (Toff-2 = 0.93 ± 0.14 s) were distinguishable from brief FBS-induced closures. Using Ton-2 = 1.7 ± 0.5 s and kon-2 from buffer, EGFR concentration in FBS was estimated as 22.2 ± 5.9 nM, close to the actual 20 nM. kon values showed small, same-order changes in FBS; koff values were largely unchanged. - FP validation: Solution-phase KD values: hSUMO1 with FN3SUMO-tFhuA 186 ± 16 nM; WDR5 with Mb4-tFhuA 455 ± 59 nM, consistent with single-channel results. EGFR unsuitable for FP due to large size; long-lived KD aligned with prior literature.

The work demonstrates a generalizable, modular nanopore sensing platform where only the binding interface (monobody) is swapped while preserving the pore scaffold and tether architecture. This overcomes steric limitations of lumen-confined detection by placing the recognition site outside the pore yet sufficiently proximal to transduce binding as electrical blockades. The platform successfully detects three structurally and functionally distinct proteins (hSUMO1, WDR5, EGFR) with characteristic single-molecule electrical signatures and kinetics, validating specificity and orientation control. For EGFR, the sensor reveals two binding modes (bimodal recognition) with distinct amplitudes and lifetimes, likely reflecting different conformers of the flexible EGFR ectodomain—features that are difficult to resolve by ensemble methods. The sensor operates in complex biofluids (5% FBS), distinguishing analyte-induced events from nonspecific background via amplitude and kinetics, enabling quantitative estimation of analyte concentration. Compared to ensemble techniques (ELISA, BLI, SPR, ITC), this approach provides time-resolved, single-molecule measurements across a wide dynamic range, capturing fast association rates (up to 10^7–10^8 M^-1 s^-1) and dissociation rates not readily accessible by BLI/SPR, with potential for multiplexing by signature differentiation. Overall, the findings address the need for specific, sensitive, and adaptable protein detection suitable for biomarker analysis in clinically relevant ranges.

The study introduces single-polypeptide, monobody-tethered tFhuA nanopore sensors that generalize to diverse protein targets while maintaining architecture, sensitivity, and specificity. Using FN3SUMO-, Mb4-, and Adnectin1-based sensors, the authors achieved label-free, real-time detection of hSUMO1, WDR5, and EGFR at single-molecule precision, quantified association/dissociation kinetics and KD, and demonstrated accurate quantification of EGFR in serum. A key discovery is EGFR’s bimodal recognition by Adnectin1, revealing distinct conformational binding substates. The genetically encodable, modular design enables straightforward substitution of binders and suggests scalability to libraries of tethered binders, with prospects for multiplex detection and integration into high-throughput platforms. Future directions include employing higher-affinity binders (e.g., affibodies) to push detection sensitivity to pg/ml levels, expanding target panels, and developing parallelized arrays for clinical proteomic diagnostics.

- Immobilization/tethering effects can reduce apparent affinity by up to an order of magnitude compared with solution-phase measurements, influencing KD on the bilayer versus FP assays. - BLI lacked the time resolution to quantitatively resolve the rapid WDR5–Mb4 kinetics observed by nanopores, providing only qualitative confirmation. - The Adnectin1-tFhuA sensor exhibited higher noise at +40 mV, necessitating operation at +20 mV for cleaner traces. - In complex matrices (FBS), low-frequency 1/f noise and brief nonspecific closures occur; while distinguishable from analyte events by amplitude and kinetics, they can modestly perturb measured kon values. - Detection relies on correct, single-orientation insertion of the nanopore and on accessibility of the binder’s interface; very large binders or sterically hindered interfaces could reduce signal transduction.