Real-time label-free detection of dynamic aptamer–small molecule interactions using a nanopore nucleic acid conformational sensor

Significance

Fast, low-cost, accurate detection of time-resolved dynamic nucleic acid conformational variation in response to small molecule binding: 1) enriches our understanding of nucleic acid structure, and regulation of structure by small molecules as a key mechanism in their biological functions; 2) enables discovery of small molecule drug compounds that target nucleic acid structures as a therapeutic strategy; 3) supports development of new tools for synthetic biology based on gene expression networks regulated by small molecules; and 4) allows development of new sensors of small molecules such as neurotransmitters for studies in neurochemistry and disease diagnostics.

Abstract

Nucleic acids can undergo conformational changes upon binding small molecules. These conformational changes can be exploited to develop new therapeutic strategies through control of gene expression or triggering of cellular responses and can also be used to develop sensors for small molecules such as neurotransmitters. Many analytical approaches can detect dynamic conformational change of nucleic acids, but they need labeling, are expensive, and have limited time resolution. The nanopore approach can provide a conformational snapshot for each nucleic acid molecule detected, but has not been reported to detect dynamic nucleic acid conformational change in response to small -molecule binding. Here we demonstrate a modular, label-free, nucleic acid-docked nanopore capable of revealing time-resolved, small molecule-induced, single nucleic acid molecule conformational transitions with millisecond resolution. By using the dopamine-, serotonin-, and theophylline-binding aptamers as testbeds, we found that these nucleic acids scaffolds can be noncovalently docked inside the MspA protein pore by a cluster of site-specific charged residues. This docking mechanism enables the ion current through the pore to characteristically vary as the aptamer undergoes conformational changes, resulting in a sequence of current fluctuations that report binding and release of single ligand molecules from the aptamer. This nanopore tool can quantify specific ligands such as neurotransmitters, elucidate nucleic acid–ligand interactions, and pinpoint the nucleic acid motifs for ligand binding, showing the potential for small molecule biosensing, drug discovery assayed via RNA and DNA conformational changes, and the design of artificial riboswitch effectors in synthetic biology.

Understanding small-molecule regulation of nucleic acid conformation is important to reveal biological mechanisms and develop biotechnological applications. Many native regulatory nucleic acids such as riboswitches can control gene expression via a conformational transition upon binding with metabolites (1), serving as new targets for antibiotic design (2, 3). New therapeutic compounds can be discovered by screening small molecules that bind the target nucleic acid motifs, change their conformation, and modulate their biological functions (4–11). Furthermore, in vitro-selected nucleic acid aptamers and engineered riboswitches (12–14) can change their conformation upon ligand binding. Utilizing this property, biosensors can be designed to detect biologically important small molecules, including neurotransmitters (14–19) and hormones (20), metabolites (12, 13, 21), antibiotics (22), and anticancer drugs (23), for biological mechanism exploration, disease diagnostics, enzyme profiling, and pharmacokinetics studies. In addition, small molecule-sensing aptamers, such as the theophylline aptamer, can be engineered into gene circuits (24–27) and activated through a ligand-triggered conformational transition to program gene expression and gene editing (27, 28). To advance the use of small molecule-sensitive nucleic acids as sensors, it is important to develop sensitive, rapid, and low-cost tools that can discriminate different conformations of single nucleic acid molecules and reveal their dynamic transitions in response to small-molecule binding.

Nanopore single-molecule–based biosensing techniques have been applied to sequencing (29–36) and various genetic (37, 38), epigenetic (39–45), and proteomic (46–50) analyses. By measuring dynamic changes of current through the nanopore, this technique has also demonstrated great potential to detect biomolecular structures. When a protein (51), DNA (30, 52, 53), RNA (54–56), or nucleic acid/protein complex (57, 58) occludes the nanopore under a transmembrane voltage, their structures can characteristically modulate the ion current through the pore. The resulting nanopore current pattern or “signature” can be analyzed to discriminate the molecular structure (59–64). However, these nanopore measurements to study biomolecular structure are often limited to providing a conformational “snapshot” and do not reveal the dynamic conformational variation of the detected molecule. This limitation can be overcome by immobilizing the target molecules in the nanopore in a mechanism such that its conformational variation can characteristically modulate the nanopore current. This strategy can be realized by attaching a nucleic acid aptamer (65) or engineering a polypeptide (66, 67) probe to the nanopore to detect reversible binding of a protein ligand outside the nanopore lumen. However, such approaches are generally not sensitive enough to detect small conformational changes of a nucleic acid scaffold upon the binding of a small ligand. In addition, custom fabrication of a nanopore with an attached molecular probe can be complicated. As a noncovalent strategy, we have encapsulated a 16-nt compact G-quadruplex in the α-hemolysin pore to elucidate its metal ion-regulated folding mechanism (68, 69), but this pore is not wide enough to accommodate most nucleic acid motifs. Recently, the ClyA protein pore has been used to trap a protein molecule in its large cylinder cavity, enabling the nanopore to elucidate protein–protein interactions (70) and detect protein-binding metabolites (71–73). This work points the way toward a new strategy for noncovalent, label-free detection of biomolecular conformational changes confined in a large nanopore. But this wide protein pore has not been explored for detecting for detecting conformational dynamics of other bio-molecular interactions.

The MspA protein pore has been developed for sequencing of nucleic acid (42, 74–77) and polypeptide (78, 79), biomolecular mechanistic study (80), RNA discrimination (81), and single-molecule chemistry (82). This nanopore encloses a 3 to 5-nm–wide goblet-shaped cavity in the cis vestibule that can host nucleic acid scaffolds. Here we used the dopamine-, serotonin- and theophylline-binding aptamers as testbeds to explore a MspA-based, noncovalent, nucleic acid-docked nanopore capable of discriminating and continuously recording small molecule-regulated nucleic acid dynamic conformational variation (Fig. 1), offering a sensor platform for elucidating nucleic acid–small molecule interactions, screening nucleic acid-targeted small molecule regulators, designing synthetic biology gene switches and sensing biological small molecules.

Fig. 1.

Principle and applications of a nanopore sensor capable of discriminating nucleic acids conformational transitions in response to the small-molecule binding.

Results

Dopamine-Induced Aptamer Conformational Transitions in the Nanopore.

MspA forms an octameric protein pore (83) in the lipid bilayer. In this study, we utilized the mutant M2 pore that has been applied in sequencing and biomolecular detection (75, 84, 85). This pore generated an ion current of I₀ = 393 ± 7 pA at a transmembrane voltage of 180 mV in 1 M KCl (conductance of 2.21 ± 0.04 nS). Driven by the voltage, the dopamine-binding aptamer (15) presented in the cis solution produced a long-duration nanopore signature block with a duration of τ_A = 1.09 ± 0.19 s (Fig. 2A). The expanded signature (Fig. 2A and SI Appendix, Fig. S1) reveals two stable blocking states A₁ and A₂, identified from their distinct blocking levels, I/I₀ = 53.0 ± 0.3% for A₁ and 47.6 ± 0.4% for A₂. Most A₁ and A₂ states do not directly transition to each other. Instead, transitions are mediated by a cluster of rapidly transitioning, unstable intermediate states A_Int, which have blocking levels that span a wide range above and below A₁ and A₂ (greyed intervals in Fig. 2 A–C). The same current patterns were also identified at a high bandwidth of 100 kHz/400 kHz sampling rate (SI Appendix, Fig. S1), and no new stable conductance states were revealed in high-bandwidth experiments that were not evident at 5 kHz bandwidth. Since discrete conductance levels are not resolvable during the transition interval, we chose to treat each A_Int cluster as a single state (Fig. 2 A–C, grey-highlighted intervals) and describe the aptamer conformation pathway as A₁↔A_Int↔ A₂ (Fig. 2A, kinetic model). From the duration of A₁, A₂, and A_Int, as well as their transition frequencies, we can derive all the transition rates in the kinetic pathway (SI Appendix, section S2). This kinetic pathway is independent of the voltage, indirectly supporting the notion that the observed current substates are caused by conformational changes in the aptamer and not the binding substates of the aptamer to the pore interior, which would otherwise display voltage-dependent kinetics (SI Appendix, section S2). We also identified that many aptamer signatures are terminated with a short-lived deeper block A₃ (τ_A3 = 0.39 ± 0.08 ms and I/I₀ = 16.4 ± 0.9%, SI Appendix, Fig. S1C). This signal likely represents transient disruption of the aptamer structure followed by translocation to the trans side of the pore (86–88). In signatures without terminal A₃, our hypothesis is that the aptamer returns to the cis solution before unfolding. In addition, the relative frequencies of A₃ transitions from other states were characterized (SI Appendix, Fig. S1C). Overall, we conclude that the dopamine aptamer can be captured by the MspA pore from the cis entrance and stably reside in the pore while transitioning among multiple conformations.

Fig. 2.

Discrimination of dopamine-induced aptamer conformational changes in an MspA protein pore. (A–C) Single-pore current recordings at 180 mV showing signature blocks for aptamer captured from the cis side in the absence of ligands (A), in the presence of dopamine in different concentrations (B), and in the presence of nontarget serotonin or norepinephrine on the trans side of the pore (C). Ligand binding events are marked with red lines for dopamine and empty invert triangles for serotonin and norepinephrine. Expanded signatures, current amplitude histograms and proposed kinetic pathway are shown for identifying different aptamer conformations and their transition mechanisms. Greyed intervals depict the unstable intermediate states, A_Int, that occur during transitions between the A₁ and A₂ states. Additional signatures at different bandwidths are shown in SI Appendix, Fig. S1 for the dopamine aptamer without ligand, SI Appendix, Fig. S2 with dopamine, and SI Appendix, Fig. S5 with serotonin and norepinephrine. Terminal A₃ transitions in a and b are investigated in SI Appendix, Figs. S1 and Fig. S2; (D) Aptamer residence time τ_A in the absence of dopamine at different voltages and in the presence of 25 µM dopamine at optimum 180 mV; (E–H) Frequency f (E and G) and duration τ_off (F and H) of the AL blocks in different dopamine concentrations (E and F), and at different voltages between 150 and 180 mV (G and H); (I and J)f (I) and τ_off (J) for binding of the dopamine aptamer with dopamine (25 µM), serotonin (50 µM), and norepinephrine (50 µM). The nanopore was recorded in 1 M KCl and 10 mM Tris (pH7.4), with 100 nM aptamer in cis solution and different concentrations of ligand in the trans solution. Selected histograms for τ_A, τ_on and τ_off are shown in SI Appendix, Figs. S9–S11.

When dopamine was added in the trans solution, a sequence of single-level blocks AL immediately appeared in the aptamer signature (red intervals in Fig. 2B). Their blocking level (I/I₀ = 49.5 ± 0.5%) and duration (τ_off = 152 ± 18 ms) are both distinct from the free aptamer states. Notably, their frequency f, calculated from the elapsed time between consecutive AL blocks τ_on (f = 1/τ_on), monotonically increased with the dopamine concentration (Fig. 2 B and E), while their τ_off was independent of the dopamine concentration (Fig. 2F). These properties together suggest that the AL blocks are generated by single dopamine molecules that enter the pore from the trans side and bind to the aptamer, resulting in a sequence of dynamic transitions between the free (A) and ligand-bound (AL) aptamer conformations (A ↔ AL). The kinetics of this pathway can be obtained from the frequency and duration of the AL blocks. Specifically, the association rate constant is k_on = 0.24 ± 0.03 µM⁻¹ s⁻¹, calculated from f = k_on[L] ([L] = 25 µM dopamine); the dissociation rate constant is k_off = 6.6 ± 0.8 s⁻¹, calculated from k_off = 1/τ_off; and the dissociation constant is K_d = 27 ± 3 µM, calculated from k_off/k_on (SI Appendix, section S2). We found that addition of dopamine on the trans side is a prerequisite for generating sequential dopamine-binding events. Dopamine added in the cis solution where the aptamer is presented cannot produce the AL blocks (SI Appendix, Fig. S3), presumably because the positive voltage applied to promote entry of the anionic aptamer into the pore repels cationic dopamine from entering from the cis side.

We found that the optimal voltage to detect sequential dopamine binding events is ~180 mV. The aptamer residence time τ_A increases with the voltage (Fig. 2D), but saturates for voltages near 180 mV, which is also near the limit to maintain a stable lipid bilayer. Aptamer with long τ_A can capture many dopamine molecules, whereas that with too short τ_A at low voltages such as 120 mV cannot capture dopamine (SI Appendix, Fig. S4). Over the range of 150 to 180 mV, we found that the frequency (f, Fig. 2G) and duration (τ_off, Fig. 2H) of the AL blocks are weakly voltage-dependent, a phenomenon probably resulting from the uneven field distribution in the pore (87, 89). Presumably, a large fraction of the voltage falls on the short restrictive sensing zone at the trans end of the pore, whereas a small fraction of the voltage forms a weak field across the large cavity where dopamine binds to the aptamer, reducing the voltage impact on the aptamer–ligand interactions.

We further analyzed the aptamer signatures to elucidate the native properties of the aptamer–ligand interactions. First, the expanded signatures (Fig. 2B and SI Appendix, Fig. S2) show that the AL block was generated from the transition levels A_IL, a subtype of A_Int with a greater block level than A₂. After dopamine dissociation, the aptamer returns to a similar A_IL state before resuming free kinetics. This finding suggests that, among multiple conformations adopted by the free aptamer, dopamine selectively binds to the aptamer occupying an unstable (A_Int), rather than one of the two stable states (A₁ and A₂), demonstrating a conformation-selectivity for ligand binding. Second, the aptamer’s residence time τ_A was almost doubled from 1.02 ± 0.19 s in the absence of dopamine to 1.90 ± 0.29 s in the presence of 25 μM dopamine (Fig. 2D). Also, the aptamer residing in the pore does not unfold at the AL state (no terminal A₃ block, SI Appendix, Fig. S2C). Both findings demonstrate the stabilization effect of ligand binding on the aptamer structure. Third, we screened chemically related neurotransmitters on the trans side, and found that both serotonin and norepinephrine can also bind the dopamine aptamer (Fig. 2C and SI Appendix, Figs. S5 and S6). However, the binding time, τ_off = 16 ± 4 ms for serotonin and 18 ± 4 ms for norepinephrine, were distinctly shorter than for dopamine (τ_off = 157 ms, Fig. 2J), suggesting a ~10-fold faster dissociation process for the two nontarget ligands. Meanwhile, both neurotransmitters bind to the aptamer at much lower frequencies, with f = 0.27 s⁻¹ for serotonin and 0.20 s⁻¹ for norepinephrine (50 µM, Fig. 2I), compared with dopamine (f = 6.1 s⁻¹), resulting in a ~25-fold slower association process. With these kinetic parameters, we estimate that ~10% of the binding events will be shorter than 2 ms and therefore escape detection as indistinguishable from the free aptamer intermediates (SI Appendix, Fig. S6). Therefore, this will not introduce fundamental errors to the association rate estimation. We conclude that the aptamer is dopamine-selective for both the association and dissociation processes, which together result in a 230- and 260-fold decrease in the affinity (increase in K_d) for serotonin and norepinephrine.

Identifying Dopamine-Binding Motifs by Screening Aptamer Variant–Ligand Interactions.

The secondary structure of the dopamine aptamer consists of a hairpin loop L1 and a multibranch loop L2 (Fig. 2A). In addition, the two single nucleotide internal loops L3 and L4 function as a joint, allowing the rigid main helix to bend and twist in the tertiary structure. How the aptamer forms a tertiary structure for dopamine binding and the key motifs participating in dopamine binding are not known. Here we used the aptamer-docked nanopore to screen a group of aptamer variants (Fig. 3). Through comparison of their conformations and their interactions with the ligand, we can pinpoint the key motifs on the aptamer structure for dopamine binding.

Fig. 3.

Identifying dopamine binding motifs by screening aptamer variants. (A–D) Dopamine aptamer variants: ΔL1/L2 (L1 and L2 deleted, A), ΔL1 (L1 deleted, B), ΔL2 (L2 deleted, C), and GG > GA (G substituted with A, D), their single-pore currents in the absence (Left) and presence (Right) of dopamine. The nanopore was recorded at +180 mV in 1 M KCl and 10 mM Tris (pH7.4), with 100 nM aptamer variants in the cis solution and 25 µM concentrations of dopamine in the trans solution; (E) A model showing a conformational mechanism for dopamine binding to the aptamer.

We first investigated the functions of loops L1 and L2. The variant ΔL1/L2 deletes both loops L1 and L2 (Fig. 3A), and ΔL1 and ΔL2 only deletes L1 (Fig. 3B) or L2 (Fig. 3C), respectively. Unlike the original aptamer that produced multi-level current blocks in the absence of dopamine (Fig. 2A), all the three variants produced long-duration single-level blocking events, with I/I₀ = 54.2 ± 1.2% and τ_A = 34 ± 5 ms for ΔL1/L2, I/I₀ = 56.9 ± 1.1% and τ_A = 46.3 ± 1.6 ms for ΔL1, and I/I₀ = 53.1 ± 0.8% and τ_A = 71.3 ± 5.9 ms for ΔL2 (Fig. 3 A–C, left traces). Addition of trans dopamine did not generate any AL binding events (Fig. 3 A–C, right traces). The loop screening indicates that unlike the original aptamer that rapidly transitions between multiple conformations, the variants with deletion of both loops L1 and L2 (ΔL1/L2) or either L1 (ΔL1) or L2 (ΔL2) only adopts a single resolvable current-blocking conformation that loses the sensitivity to dopamine, presumably because dopamine binding requires both loops L1 and L2. Without the tertiary folding capability, these loop variants should maintain a hairpin conformation, generating single-level blocks in a characteristic current range (I/I₀ = 53 to 57%).

It has been speculated that the binding of dopamine may induce the aptamer to form a G-quadruplex (15). If so, we anticipated that this should be a compact G-quadruplex consisting of two quartets contributed by four GG motifs: G16/G17 in L1, G28/G29 and G33/G34 in L2, and G9/G36 in L3, and the G-quadruplex formation can be disabled in the partial absence of these guanine bases. We therefore targeted these GG motifs in the loops by testing the variant GG > GA that carries the G > A single nucleotide polymorphisms (SNPs) at G17, G29 and G34 (Fig. 3D). This variant produced blocking patterns very similar to those produced by the loop-deletion variants with a single blocking level of I/I₀ = 57.3 ± 0.8% and duration of τ_A = 76 ± 5 ms (Fig. 3D, left trace). The presence of trans dopamine did not produce any apparent dopamine binding events, and the signature’s blocking level and duration were not significantly changed, with I/I₀ = 56.7 ± 0.7% and τ_A = 44 ± 5 ms (Fig. 3D, right trace). Therefore, without the three guanine bases distributed in L1 and L2 that enable the G-quadruplex formation, the aptamer can only adopt a single resolvable dopamine-insensitive blocking conformation. Its blocking level (57%) is similar to that of the three loop deletion variants (I/I₀ = 53 to 57%), in consistence with that this SNP variant could form a hairpin. In summary, the selected guanines in the loops are among the key motifs to the formation of the G-quadruplex upon dopamine binding.

The aptamer and its variant studies suggest a conformation model for the aptamer–dopamine interaction (Fig. 3E). The A₁ state of the free aptamer can be assigned to a hairpin, as its blocking level (I/I₀ = 53.0%) is similar to that of the four hairpin variants (I/I₀ = 53 to 57%), including the point-mutation variant (GG > GA) without changing the aptamer length. A₂ of the free aptamer could adopt a short-lived G-quadruplex that is folded from the hairpin (A₁) via unstable intermediates A_Int. The intermediates with low blocking levels can form a docking site accessible to dopamine. Dopamine is selectively docked in these intermediates, inducing the aptamer to form a stable G-quadruplex (AL).

Docking a Dopamine Aptamer with a Cationic Ring Engineered in the MspA Nanopore.

The ability to continuously observe aptamer dynamic conformational transitions originates from the aptamer docking configuration in the nanopore, which should not only stably hold the aptamer in the pore, but also enables the nanopore current to be sensitively modulated by the aptamer conformation. Here we use a group of mutant pores to explore where and how the aptamer is docked in the nanopore. Supposing that positively charged amino acid residues in the lumen participate in the nanopore–aptamer interaction, mutations were selected which altered the nanopore charge distribution. The candidate aptamer locations in the M2 pore include the R118 ring in the middle of the pore and the R134 ring near the cis entrance (83) (Fig. 4A). R165 in between the two rings was not considered because its side chain does not project to the lumen and its charge may be balanced by surrounding E63 and E127.

Fig. 4.

Understanding the aptamer docking mechanism in the MspA pore by changing charge distribution in the lumen. (A) Structure of the MspA-M2 protein pore. Positively (blue) and negatively (red) charged amino acid residues in the lumen are marked. The aptamer docking site at the R118 ring is highlighted; (B–D) Charge-altering mutations in the M2 pore were made, including M2-R118N/R134N (B), M2-R118N (C) and M2-R134N (D), and corresponding single-pore current signatures for the dopamine aptamer in the absence (Left) and presence (Right) of dopamine. The nanopore was recorded at +180 mV in 1 M KCl and 10 mM Tris (pH7.4), with 100 nM aptamer variants in the cis solution and 25 µM dopamine in the trans solution.

We first tested the mutant M2-R118N/R134N pore, which replaces both R118 and R134 rings with neutral asparagine (Fig. 4B model). We found that the cis aptamer no longer produced the M2-like prolonged block at 180 mV (Fig. 4B, left trace), but only short-lived partial blocks (I/I₀ = 61.2 ± 0.9%, τ_A = 1.4 ± 0.4 ms), indicating that the nanopore without both positively charge rings cannot effectively capture the aptamer. We further tested the mutant M2-R118N pore, which removes the R118 ring but retains the R134 ring (Fig. 4C). Again, only brief blocking events were observed (Fig. 4C, left trace, I/I₀ = 60.7 ± 0.8%, τ_A = 0.7 ± 0.3 ms). As neither pore can effectively capture the aptamer, it is unsurprising that there were no dopamine binding events from the trans solution (Fig. 4 B and C, right traces). We finally tested the mutant M2-R134N pore, which retains the R118 ring but removes the R134 ring (Fig. 4D model). Interestingly, this pore recapitulates the functional interactions with the aptamer found in the native M2 pore. The aptamer generated the M2-like blocking signature currents (Fig. 4 D, Left), which feature a long residence time (τ_A = 1.5 ± 0.2 s) and transitions between two main conformations, A₁ at I/I₀ = 50.7 ± 0.6% and A₂ at I/I₀ = 47.9 ± 1.0%, via a cluster of unstable transitions. Dopamine can also bind the aptamer in the pore from the trans side (Fig. 4D, right trace, marked by red lines), with similar kinetics (k_on = 0.14 ± 0.03 µM⁻¹ s⁻¹ and k_off = 4.9 ± 0.7 s⁻¹) as native M2.

In conclusion, by screening the charge distribution in the pore lumen, we identified that the R118 ring, rather than the R134 ring, plays a key role in docking the aptamer. This finding supports that the aptamer is very likely suspended in the middle of the lumen cavity, where it is coordinated by the positive charges on the R118 ring. This docking configuration has several functions: i) the “suspended” aptamer is fully exposed to the surrounding ion pathway, enabling the ion current to sensitively change with the aptamer conformation; ii) the aptamer does not block the ligand pathway at the narrow trans entrance, allowing the ligand to flow through the pore from the trans side to interact with the suspended aptamer from different directions, regardless of the orientation of the aptamer's ligand binding site; and iii) the multiple blocking levels in the aptamer signature (e.g., A₁ and A₂) are confirmed to be generated by different conformations, rather than different locations of the aptamer in the pore.

A Platform for Assaying Nucleic Acid–Small Molecule Interactions.

We also investigated a serotonin-binding aptamer (15) and a theophylline riboswitch RNA aptamer (26) (SI Appendix, Table S1) to demonstrate broader applications of this nanopore platform for detecting small molecule-induced nucleic acid conformation changes. The study of the serotonin aptamer further enhances its potential for neurotransmitter detection, and the study of the theophylline aptamer expands the target to small molecule-sensitive RNA scaffolds.

Similar to the dopamine aptamer, the serotonin aptamer can be stably docked in the M2 pore from the cis side at 120 mV, producing a long signature at a blocking level of I/I₀ = 42.0 ± 0.6% (120 mV, Fig. 5A). The binding of serotonin from the trans side characteristically increased the blocking level to I/I₀ = 44.8 ± 0.8%, generating a sequence of AL blocks with τ_off = 5.9 ± 0.7 s for the ligand-bound aptamer conformation (Fig. 5B). As the serotonin concentration increased, the frequency f of the AL blocks increased (Fig. 5 B and I), and their duration τ_off was not changed (Fig. 5J). Both f and τ_off are not significantly influenced by the voltage applied (Fig. 5 K and L). Serotonin binding can stabilize the aptamer, increasing its residence time τ_A by 1.6 fold from 22.4 ± 2.7 s in the absence of serotonin to 35.5 ± 4.5 s in 25 μM serotonin (Fig. 5G). Dopamine rarely binds to the serotonin aptamer (Fig. 5C). f for dopamine binding was sevenfold lower (Fig. 5M) and τ_off was 23-fold faster (Fig. 5N) than serotonin binding, suggesting that the aptamer is serotonin-selective (15).

Fig. 5.

Characterization of serotonin-binding aptamer and theophylline riboswitch aptamer conformational change upon ligand binding in the nanopore. (A–E) Single pore current recordings at 120 mV showing signature blocks for the serotonin aptamer alone (A), in the presence of 25 µM and 50 µM serotonin (B), and 50 µM dopamine in the trans solution (C), signature blocks for the theophylline aptamer alone (D) and in 25 µM theophylline in the trans solution (E). Expanded signature in 50 µM serotonin (B) is also shown in SI Appendix, Fig. S14 for identifying free aptamer domains between consecutive AL blocks. Expanded signatures, current amplitude histograms and kinetic model are shown. More signatures at different bandwidths are shown in SI Appendix, Fig. S7 for the serotonin aptamer. Identification of serotonin binding events at 180 mV is shown in SI Appendix, Fig. S8; (F) Kinetic scheme for ligands binding to the serotonin and theophylline aptamers; (G and H) Residence time (τ_A) of the serotonin aptamer (G) and theophylline aptamer (H) in the absence ligand at different voltages, and in the presence of 25 µM ligand at 120 mV (G) and 180 mV (H); (I–L) f (I, K) and τ_off (J, L) of the serotonin and theophylline binding blocks with different ligand concentrations (I and J), and at different voltages (K and L); M and N. f (M) and τ_off (N) for binding of the serotonin aptamer with 25 µM serotonin or 50 µM dopamine at 120 mV. Selected duration histograms for both aptamers are shown in SI Appendix, Figs. S12 and S13.

The serotonin and dopamine aptamers also show significant differences in blocking conformations and kinetics. i) In contrast to the dopamine aptamer that transitions between two resolvable conformations (A₁ and A₂), the serotonin aptamer folds into one resolvable conformation A (Fig. 5A). In addition, the aptamer also generates several types of sub-millisecond to millisecond intermediates with small probabilities (Fig. 5A and SI Appendix, Fig. S7). ii) Serotonin directly binds to the main conformation A (Fig. 5B and SI Appendix, Fig. S7), in contrast to dopamine that binds to an intermediate state. iii) From f (Fig. 5K) and τ_off (Fig. 5L), the kinetic constants were calculated to be k_on = 0.0185 ± 0.0037 µM⁻¹ s⁻¹, k_off = 0.167 ± 0.022 s⁻¹, and K_d = 9.0 ± 1.3 µM. The kinetic analysis suggests that the binding of serotonin to its aptamer is ninefold slower in the association process and 40-fold slower in the dissociation process compared to dopamine binding to its aptamer. iv) In contrast to the dopamine aptamer that reaches the highest stability and sensitivity at 180 mV, the serotonin aptamer is more stable in the nanopore at lower voltages (Fig. 5G) and therefore can generate more serotonin binding events at 120 mV than at 180 mV (SI Appendix, Fig. S8).

Similar to the dopamine aptamer, the theophylline aptamer can be docked in the pore from the cis solution (Fig. 5D) with τ_A that is prolonged to 1.4 ± 0.3 s as the voltage increases to 180 mV (Fig. 5H). Like the serotonin aptamer, this aptamer’s signature is a single-level current block at I/I₀ = 61.2 ± 0.9% with 100-μs-scale, low-probability blocking flickers, indicating that the aptamer without theophylline binding adopts a single resolvable main conformation (A) that transitions with short-lived intermediates at lower blocking level than A. Theophylline can bind the docked aptamer from the trans solution, generating consecutive single-level blocking events with τ_off = 51 ± 9 ms (Fig. 5E), which slightly increase the current level from I/I₀ = 61.2 ± 0.9% for free aptamer conformation (A) to I/I₀ = 62.4 ± 0.7% for the theophylline-bound state (AL), with its f increasing with the theophylline concentration (Fig. 5I). Resembling the dopamine and serotonin aptamer binding kinetics, f and τ_off for theophylline binding to its aptamer are weakly voltage-dependent (Fig. 5 K and L). From f and τ_off (180 mV, 25 uM theophylline), we calculated k_on = 0.269 ± 0.043 µM⁻¹ s⁻¹, k_off = 19.6 ± 2.7 s⁻¹, and K_d = 72 ± 17 µM. In addition, the theophylline binding can also stabilize its aptamer, increasing τ_A by 29% from 1.21 ± 0.32 s without ligand to 1.58 ± 0.27 s with ligand. Overall, we verified that this nanopore can detect conformational changes of an RNA scaffold interacting with its small molecule ligand.

Finally, we summarize two common conformational properties of the three aptamer–ligand interactions revealed by the characteristic nanopore-blocking levels. One common feature is that the aptamer alone transitions among multiple conformations, whereas ligand binding transforms the aptamer into a single stable ligand-bound conformation. This property markedly reduces the nanopore current fluctuations (I_SD, SD of the nanopore current) upon ligand binding to the aptamer (Fig. 6A). For the dopamine aptamer, the reduction in current fluctuations is primarily due to the stabilization of the aptamer such that it no longer fluctuates between the A1 and A2 conformations, which block the current to different extents. For the serotonin and theophylline aptamers, fluctuations are also reduced upon their ligands binding to their respective aptamers, even though only a single current-blocking conformation is resolved before ligand binding. The reduction in current fluctuation is presumably because the aptamer assumes a more stable configuration in the pore upon ligand binding. Therefore, detection of a ligand-binding event is possible by observing changes in current fluctuations for all the aptamers. A second notable common feature is that the three aptamers’ ligand-bound conformations block the current to a lesser extent compared to the ligand-free conformation (Fig. 6B). For the dopamine aptamer, dopamine binds to an A_IL intermediate to form a ligand-bound conformation AL. AL can be distinguished from the A_IL as its current level is higher than A_IL by ΔI/I₀ = 4.3% (~19 pA). The serotonin and theophylline aptamers follow the same trend: the current level for the ligand-bound conformation (AL) is higher than the free aptamer (A) by ΔI/I₀ = 2.8% (~7 pA) for the serotonin aptamer and ΔI/I₀ = 1.2% (~5 pA) for the theophylline aptamer. This common property provides insight into the aptamer structure variation upon ligand binding. We interpret that ligand binding causes the aptamer to form a more compact structure that slightly decrease in volume, due to the coordination between the ligand and docking site. The smaller molecular volume decreases the occlusion of the ion pathway, thereby increasing the rate of ion flow through the nanopore.

Fig. 6.

(A and B) Comparison of nanopore ion current SDs (I_SD, A) and blocking levels (I/I₀, B) for the dopamine-, serotonin- and theophylline-binding aptamers in conformations without and with ligand binding. Data values are provided in SI Appendix, Table S3. Calculation of I_SD and the t test to compare difference in I_SD between free and ligand-bound aptamer conformations are described in SI Appendix, Methods. *P < 0.05 and **P < 0.001 from the t test. For the dopamine aptamer, the reduction in current fluctuations is primarily due to the stabilization of the aptamer such that it no longer fluctuates between multiple conformations including A₁ and A₂. On the other hand, fluctuations are also reduced upon binding of serotonin or theophylline to their respective aptamers, even though only a single main current-blocking conformation is resolved before ligand binding. This is presumably because the aptamer assumes a more stable configuration in the pore upon ligand binding.

Discussion

We have investigated a rapid, sensitive and label-free platform to study dynamic nucleic acid–small molecule interactions at high temporal resolution. The ligand screening experiment for aptamer selectivity demonstrates the potential of the platform in screening for small molecule regulators or potential therapeutic agents targeting nucleic acid motifs. Small molecules are regulators of various nucleic acid structures and functions, such as riboswitch aptamers, HIV trans-activation response (TAR) element (90), Hepatitis C virus (HCV) internal ribosome entry site (8), SARS-CoV-2 frameshifting element (91), human microRNA (92) and RNA repeats (93). The aptamer motif screening experiment for ligand binding demonstrates the potential of the platform in gene switch design for programming cellular functions (24) by in vitro screening for ligand induced conformation changes. Although the three ligands in this study are cationic compounds, anionic ligands could also be tested. Theoretically, anionic ligands can be presented in the cis solution, to be electrically dragged into the pore from the cis entrance by a positive voltage to bind the aptamer residing in the pore.

Whereas transmitters such as dopamine and serotonin can be detected by electron transfer to an electrochemical electrode, this approach is restricted to transmitters that are readily oxidizable and is further complicated by interference from other oxidizable compounds in the sample. Aptamers can target both oxidizable and nonoxidizable neurotransmitters, and its ligand selectivity diminishes nontarget interference. Neurotransmitters can also be detected by binding with a labeled reporter, which is mixed with the target in the solution to reveal its spatial and temporal distribution (94, 95). However, the mixture needs time to get to equilibrium, causing limited applications in fast detection of neurotransmitter. In the nanopore method, the aptamer is separated from the ligand by the membrane. They do not mix in the same solution and are instead driven by the imposed electric field to interact within the nanopore. This configuration allows detecting the ligand immediately when it binds to the aptamer without the need to come to equilibrium with the aptamer, thus allowing rapid dynamic neurotransmitter detection. We acknowledge that the cis aptamer/trans ligand configuration is not applicable to anionic ligands as they will be prevented from entering the pore from the trans side by the voltage that drives the aptamer into the pore from the cis side. In spite of this limitation, the approach is still applicable to most cationic small molecules, including common cationic neurotransmitters. Recently, an aptamer-immobilized field effect transistor (FET) has been applied in neurotransmitter detection by electrically measuring the aptamer conformation change caused by ligand binding (15). By comparison, the nanopore has a potential to detect neurotransmitters by discriminating and counting single aptamer molecule conformation transitions with a high signal-to-noise ratio.

Currently, the limitation of this nanopore sensor for real-time small molecule sensing is the limit of detection (LOD). LOD is the lowest ligand concentration at which the ligand binding events counted by the nanopore reaches a resolvable frequency. In dopamine detection, we have observed that the frequency of AL blocks f in 100 nM dopamine was 0.11 s⁻¹ (average of 1 AL block every 9 s, Fig. 2F). If the minimal observable frequency, for example, is set at 0.1 s⁻¹, in order to observe dozens of events in a ~10-min recording, then the LOD is 100 nM. This LOD, however, is not low enough to detect neurotransmitters that are in the 1 to 10 nM range in tissue (15, 26). The limiting factor is the low frequency of dopamine binding events at low concentrations. This limitation is related to high dissociation constants (K_d) measured by the nanopore for the three aptamers, which are higher than what was found in free solution by two orders of magnitude (SI Appendix, Table S3) (15, 26). At ligand concentrations well below the K_d value, the principal determinant of the observable event frequency is the ligand association rate, k_on. k_on in the nanopore environment is likely much smaller than in free solution to account for why the K_d value calculated in our nanopore system (k_off/k_on) is much higher than reported in free solution. This may be attributed to two possible factors. Firstly, confinement within the nanopore lumen may reduce the accessibility or exposure of the aptamer's binding site to its ligand. Secondly, the high salt concentration (1 M KCl) could diminish the attraction between the negatively charged aptamers and cationic ligands. Future work to increase sensitivity will focus on increasing the frequency of nanopore ligand-binding events. An effective strategy to increase the flow of cationic ligands from the trans entrance involves modifying the pH and ionic strength conditions, as well as manipulating the local charge distribution. For instance, one possible approach is to replace the positively charged amino acid R96 with neutral or negatively charged amino acids. These factors are believed to regulate ligand flow through electrostatic interaction, ion selectivity, and/or electroosmotic flow. Another strategy is to fine tune the kinetics of aptamer–ligand interactions to enhance the ligand binding rate (k_on) by site-specific sequence alteration and nucleotide modification. Finally, the optimized sensor can be integrated within nanopore arrays in commercial devices (96) for high-throughput neurochemical detection or drug screening; and the protein pore can also be incorporated into the tips of micropipette probes (prefilled with aptamer) amenable to micromanipulation (97) to achieve high spatial and temporal resolution in quantifying ligand concentrations.

Methods

The Materials and Methods section for this work can be found in SI Appendix, Methods, including 1) Preparation of the MspA and variant proteins, including gene sequence design, plasmid synthesis, protein expression and purification for MspA and various variants; 2) Aptamers and small-molecule ligands, including aptamers and ligands sources and stock solutions, and aptamer refolding process; 3) Nanopore single-channel recording, including lipid, device, lipid bilayer formation, protein pore formation, electric recording and bandwidth setting; 4) Data analysis, including methods for analyzing events durations, current amplitudes, and current SD, and statistical significance.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix.