Free Solution Assay Signal Modulation in Variable-Stem-Length Hairpin Aptamers.

Interferometric measurements of free solution assays (FSAs) quantify changes in molecular conformation and hydration upon binding. Here, we demonstrate that aptamer probes designed to undergo varying levels of conformational change upon binding produce corresponding variations in FSA signals. A series of hairpin aptamers were synthesized for the small molecule (tenofovir) with identical loop regions that contain the binding pocket, with between 2 and 10 self-associating base pairings in the stem region. Aptamers selected for tenofovir showed a decrease in the FSA signal and binding affinity (increase in K D) with increasing stem length. Thermodynamic calculations of the Gibbs free energy (ΔG) reported a decrease in ΔG with respect to a corresponding increase in the aptamer stem length. Collectively, these observations provide an expanded understanding of FSA and demonstrate the potential for the rational design of label-free aptamer beacons using FSA as readout.

Introduction

The free solution assay (FSA) is a label-free, solution-phase methodology that enables characterization of molecular interactions for a broad range of species, rapidly, and in complex matrices.[1] The FSA method capitalizes upon changes in the solution dipole moment upon binding, which is measured using a compensated interferometric reader (CIR).[2] Because the FSA signal is a product of inherent properties of the binding interaction, there is no need for surface immobilization or labeling of one of the interacting species.

The FSA signal arises from changes in molecular conformation and hydration upon binding, as described by the free solution response function (FreeSRF) given below which states that the observed interferometric signal (ρ) is a function of the magnitude of conformation and hydration change upon binding, molar refractivity (χ), the instrument response function (β), and the concentration of binding events (C).[3] Using quality structural information about the binding system, such as X-ray crystal structures of the unbound constituents and bound product, the magnitude of conformation and hydration changes can be calculated, allowing the estimation of the binding signal.

An advantageous family of probe molecules for FSAs are aptamers, short segments of single-stranded DNA or RNA molecules that can be rapidly and inexpensively designed to bind to a wide range of targets.[4] Aptamers provide inexpensive production, no cold-chain requirement, high-affinity binding to their target, and excellent specificity to the desired target molecule, proving to be an attractive alternative to antibodies for many applications. Once identified, aptamers can be produced by purely synthetic means, requiring no organisms, cell culture, or biological expression system. Therefore, aptamer reagents exhibit a minimized lot-to-lot variability. Further, facile modifications can be made to aptamers during synthesis that can confer nuclease stability of the molecules in serum for in vivo and diagnostic applications.

Due to their ability to form a variety of three-dimensional (3D) structures, such as helices and single-stranded loops, aptamers can bind to diverse targets in a wide range of situations. They can serve as sensors,[4] therapeutics,[5] cellular process regulators,[6] and drug-targeting guides.[7] Aptamers are also unique because they can be selected to bind, manipulate, inhibit, control, or detect small molecules in the complex matrix, which will be used for the assay.[8]

Recently, there has been a renewed interest in the design of aptamers that bind small molecules due to the inherent limitations encountered when attempting to prepare small-molecule-specific antibodies.[9] Small molecules can be highly bioactive compounds and include harmful toxins, beneficial drugs, antibiotics, and nutrients and can serve as intercellular signal mediators or neurotransmitters. Numerous approaches to aptamer-based biosensing have been reported, including electrochemical,[10] colorimetric,[11] surface-enhanced Raman spectroscopy (SERS),[12] refractive index (RI) detection,[13] and others.[14] All of these techniques exhibit advantages and limitations, but aptamer-based biosensors, particularly the label-free strategies based on surface plasmon resonance (SPR) and bio-layer interferometry (BLI) that require immobilization, are complicated by chemical assembly and are limited by background and surface fowling when attempting to make determinations at high sensitivity. Yet, aptamer-based FSAs have been recently shown to be useful for the high-sensitivity measurement of several small-molecule targets.[8] For example, FSA has been used recently to quantify opioids and/or their metabolites in urine at the pg/mL level[15] and to measure the protein biomarker, CYFRA 21.1 at ∼60 pg/mL, which is nearly 10-fold more sensitive than electrochemiluminesence.[16] In particular, it has been illustrated that the aptamer–FSA–CIR method provides both highly sensitive quantification of opioids and their primary metabolites in urine, as well as pharmacodynamic insights about these important molecules.[15]

Previous studies involving nucleotide double-helix manipulation have further indicated that the magnitude of structural/conformational changes correlates with free solution interferometric signal.[17] Capitalizing on the unique property that the FSA signal scales with changes in conformation and hydration, the research presented here investigates a method to potentially amplify the FreeSRF through the rational design of the aptamer. This observation, combined with our recent FreeSRF modeling efforts, led us to postulate that hairpin aptamers that unfold upon binding to a target and elicit a large conformational change would produce a large FSA response. To test this hypothesis, we designed a series of hairpin aptamers with identical “loop” regions that bind the small-molecule tenofovir, accompanied by a variable-length, self-associating stem region (Figure ).[18] Then the FSA binding response was measured as a function of stem length. We demonstrate that the binding signal is inversely correlated with the number of self-associating base pairs in the aptamer stem region. Using this knowledge, we anticipate the potential to guide the aptamer design to improve the signal-to-noise ratio (S/N) for small-molecule–aptamer–FSA assays.

Figure 1

Hairpin aptamer showing the two regions of the structure: the loop region, which is constant across all structures, contains the binding region and the stem region, which contains self-associating base pairs. The cutoff for each aptamer stem length is denoted by the blue line.

Materials and Methods

Single-stranded DNA aptamers were selected by Base Pair Biotechnologies (Pearland, TX) using a modified SELEX process.[19] Tenofovir, phosphate-buffered saline (PBS), sodium chloride, and magnesium chloride were obtained from Sigma-Aldrich. All PBS used was modified to include 150 mM NaCl and 1 mM MgCl2.

Prior to measurement by FSA, aptamers were refolded by incubation at 80 °C for 5 min and then cooled to an ambient temperature. All aptamers were screened against the small-molecule target by incubating two concentrations of each aptamer (0 and 50 μM) with tenofovir (327.5 μM). For these screening experiments, we used very large concentrations of both the aptamer and the small molecule to ensure a robust signal. We used a relatively higher concentration of the small molecule (327.5 μM tenofovir versus 50 μM of the aptamer) to ensure all aptamer present in the solution is bound despite differences in affinity between stem lengths. A blank consisting of tenofovir mixed with buffer was used to correct for the bulk RI. The only molecules present in the sample solutions are the aptamer, tenofovir, and phosphate-buffered saline (used as purchased from Sigma-Aldrich). With nothing else present in the solution, when there is no target present, there should be no nonspecific interaction.

All binding affinity determinations were performed in an end-point format as previously described.[1b,20] Briefly, a constant concentration of aptamer (1 nM) was incubated with increasing concentrations of tenofovir (0–40 nM). The concentration of aptamer was kept at or below the expected KD to ensure an accurate measurement of the KD. The concentration of the aptamer was kept constant across the five different stem lengths to obtain a more direct comparison for BMAX. Reference samples were prepared by incubating the range of tenofovir concentrations with PBS, devoid of the aptamer. Samples were incubated for 2 h at 22 °C before being measured using an interferometer in the sample-reference format described previously.[8] Measurements were made in triplicate. Dissociation constants (KD) were determined by fitting the concentration-dependent signal to a single-site saturation isotherm (Y = X*Bmax/(KD + X)), where X is the concentration of the analyte and Y is the instrument signal, using GraphPad Prism (GraphPad Software, Inc., La Jolla, CA).

Thermodynamic values were calculated using the mfold web server.[21] The free energies used are from the laboratory of SantaLucia Jr,[22] and salt correction was used with a 150 mM Na+ and 1 mM Mg2+ concentrations; all other parameters were set to default. Aptamer 3D structures were simulated using RNA Composer.[23] The simulated 3D structures were used to perform ligand-docking simulations using Autodock VINA.[24]

Results and Discussion

Figure depicts the general structure for the hairpin aptamers developed for the small-molecule tenofovir. This target was chosen because aptamers were available that assumed the hairpin shape and could be easily modified without altering the binding region. Furthermore, we have previously characterized the 9-stem aptamer by both backscattering interferometry (BSI) (9.0 nM) and MST (1.0 nM).[18] Additionally, cross-reactivity measurements against a similarly structured small molecule (ampicillin) were shown to elicit minimal off-target binding.[18]

Enabled by the mix-and-read FSA methodology and our compensated interferometer,[8,16] an end-point screening study was performed on aptamers with varying stem lengths (2, 4, 6, 8, and 9 base pairings; Figure ). Here, we use a fixed concentration of the small-molecule target (327.5 μM of tenofovir) and hairpin aptamer (50 μM). Figure A illustrates that the binding signal of the five unique hairpin aptamers to tenofovir showed a “decrease” in the FSA signal as the length of the self-associating stem length was “increased”. Interestingly, the aptamer with fewer self-associating stem base pairs (the “2-stem” aptamer) yielded the largest FSA signal (60 milliradians (mrad)), providing a ∼6-fold signal increase over the aptamer with nine self-associating stem pairings (the “9-stem”). This observation confirmed our suspicion that “a longer stem structure” does not necessarily lead to a larger FSA signal.

Figure 2

Tenofovir hairpin aptamers. Five hairpin aptamers with varying stem lengths bind to tenofovir with varying FSA signal magnitudes (A) and KD (B) despite identical target binding. (A) Error bars represent the standard deviation of three replicate trials. The dotted line is a fitted slope = −7.5 ± 0.9 mrad/base pair, R2 = 0.96. (B) KD fitted to single-site saturation isotherm (full data in the Supporting Information Figure S1). Error bars represent the standard error of the fitted KD. Thermodynamic calculations show a decrease in the Gibbs free energy (C) and an increase in melting temperature (D) correlating with the increasing stem length.

Further analysis of these observations is enlightening regarding the FSA mechanism. First, Figure A illustrates that the FSA signal exhibits a monotonic reduction in signal (slope of −7.5 ± 0.9 mrad/base pair, R2 = 0.96) with increasing stem length. Second, this “inverse” relationship for stem length versus binding signal is not necessarily intuitive. One would assume more broken bonds, and a larger conformation change would give a larger FSA signal. Third, while conformation and hydration changes dominate the FSA signal,[3] here it appears that the prediction of the binding response of the hairpin configuration requires the consideration of the need to overcome the energy for the Watson–Crick pairing to break bonds in the stem distant from the loop. This energy boundary must be balanced by the ability of the aptamer to form the target–probe complex, thereby undergoing structural and hydration changes. When paired bases in the hairpin are separated, the aptamer is allowed to unfold, presumably resulting in a subsequent increase in the FreeSRF signal. Yet, as demonstrated below, there are other considerations when estimating the value of FreeSRF for aptamer–small-molecule systems containing stem structures, including the “strength” of the hairpin structure.

It has been previously shown that long hairpin tails can produce very stable structures, with the 8-stem duplex species having up to picomolar binding affinities.[25] It is interesting that even the 9-stem tenofovir aptamer studied here gave KD measurements in the nanomolar range.[18]

This observation is likely attributable to the fact that aptamer structures with longer stem regions are more likely to form stable complexes because they undergo base pair binding, making them less likely to dissociate. Yet, this observation may not apply to FSA measurements, which depends on conformational changes to report the aptamer binding to the target. In other words, it is not necessarily expected that the interaction producing the highest affinity complex would produce the largest FreeSRF signal. To test this hypothesis, we performed affinity determinations for each of the five aptamers binding to tenofovir. Taking into account that we started with the 9-stem aptamer and then snipped off the stem sections prior to performing the binding experiments, the KD reported in this manuscript for the 9-stem aptamer corresponds quite well to that previously obtained and reported in our Analyst paper several years ago (∼9 nM in both cases for the 9-stem species). Figure B and Table present the results from these KD determinations, with a representative binding curve presented in Figure . The error in fitting the KD (Figure B) is a result of how well the line fits the data (R2), but also the number and distribution of the analyzed data points (the measured ligand concentrations) and the error in replicate measurements for these samples. Much to our surprise, as we analyzed aptamers with shorter length stems, the measured binding affinities increased (a decrease in KD) with the stem length, implying a “weaker” binding interaction between the aptamer and tenofovir. Interestingly, we also found that BMAX scaled inversely with the stem length. This observation is important because, for analyte quantification, the upper limit of the binding signal is bound by BMAX. We hypothesize that aptamers with fewer stem region base pairings are more likely to dissociate upon binding the target than those with more base pairings, for small-molecule–aptamer interactions. The larger FSA signal for shorter stems is somewhat counterintuitive and is presumably due to less structural rearrangement for the long stem probe being allowed.

Table 1

Affinity Measurements and Thermodynamic Values for the Tenofovir Aptamers

stem length	KD (nM)	R2	BMAX (mrad)	melting temp (°C)	ΔG (kcal/mol)
2	1.71	0.991	44.2	37.1	–0.01
4	6.64	0.980	30.2	85.3	–5.23
6	4.17	0.888	18.1	96.8	–7.7
8	4.88	0.980	10.1	105.5	–11.2
9	9.04	0.988	15.3	117.3	–14.82

Figure 3

Representative binding curve of tenofovir binding to the aptamer; stem length 2 shown. Error bars represent the standard deviation of three independent trials.

To further unravel our observations, we performed thermodynamic calculations of the Gibbs free energy (ΔG) and melting temperature (Tm) with the stem length of each aptamer using the mfold software (Figure C,D). Figure C illustrates that the 2-stem base-pairing aptamer has a self-association ΔG of −0.01 kcal/mol, which means that the molecule has a very small energy barrier to overcome to dissociate the stem. Therefore, the stem is easily forced apart upon binding to tenofovir. Meanwhile, the 9-stem aptamer exhibited a ΔG of −15 kcal/mol, meaning that self-association of the stem structure of the aptamer has an energy state that is considerably more thermodynamically favorable. In this case, it would require substantially more energy to dissociate stem pairs. Remembering that in the free solution assay, we start with the aptamer in the associated “hairpin” form, and upon binding the small molecule the stem is presumably disrupted (or some portion of it) enabling binding of the probe to the target. This observation is important because it correlates with the general character of the ΔG plot that predicts the aptamer–tenofovir disassociation to be more favorable for short stem length aptamers.

The melting point of a DNA strand provides another metric for assessing the stability of a hybridization event.[26] Here, we calculated Tm, finding that it increased with the stem length. These observations provide further support for the hypothesis that fewer base pairings in the stem region are more likely to lead to dissociation upon binding. As such, an aptamer with a short stem, residing in a higher resting thermodynamic energy state, requires less energy to undergo a significant conformation change upon binding, resulting in a large FSA signal. Conversely, the aptamers with longer hairpins have a low ΔG and high Tm, making it energetically undesirable for the hairpin part of the probe to dissociate upon binding to tenofovir.

In the case of longer stems, the target still binds to the probe but with a reduced FSA signal and lower affinity (higher KD). In this case, the longer stem aptamer likely binds to the target in the “absence” of breaking some of the self-associating stem bonds, thus mitigating the magnitude of conformational/hydration change that would otherwise contribute to a larger FSA signal. This observation is also consistent with the FreeSRF model, which predicts that the binding signal magnitude scales with conformation and hydration changes.[3]

As shown in Figure , the results of Autodock VINA ligand-docking simulations demonstrate that the binding location is within the loop region of the aptamer. Figure A illustrates that when tenofovir binds to the 2-stem aptamer within the loop region, it appears to place strain on the stem, presumably leading to a forced dissociation. Figure B shows that tenofovir can potentially slip into the binding pocket without disturbing the stem region, largely because of the inherent stability of the multiple base-paired stem.

Figure 4

Results of the ligand-docking simulation show tenofovir binding to the loop region of the hairpin aptamer. (A) Tenofovir binding to the 2-stem hairpin aptamer fits within the loop region and can easily force dissociation of the stem. (B) Tenofovir binding to the 9-stem hairpin aptamer can fit into the loop binding pocket without disturbing the stem region.

Collectively, we hypothesize that because the small-molecule tenofovir binds in the loop region, there is reduced likelihood, or need, for the hairpin to fully disassociate during the binding event. It is probable that the stronger (longer) stems make it more difficult for the binding loop portion of the aptamer to make slight rearrangements necessary to let tenofovir slip into the pocket. Due to entropic drivers, it also appears that shorter hairpins dissociate more frequently, resulting in more binding events and a collectively larger conformational change, with the overall result of a larger FSA signal. Therefore, the tenofovir could bind inside the hairpin loop without adding enough stress to the stem region to force the longer stems apart.

In a report on the development of an ATP aptamer for the aim of exploring enantioselective sensor molecules for adenosine,[27] a temperature and stem length dependency was observed for the fluorescently labeled aptamers. While it is interesting to note similar observations with other techniques, given the difference in the signal transduction mechanism (fluorescence versus structure and hydration changes), it is not clear to us that these observations provide additional insight into our stem length observations. We hope to explore these potential insights in the future.

Conclusions

Hairpin aptamers likely undergo large conformational changes upon binding their target, and knowledge of the relative magnitude of these changes can enable optimization of the free solution assay for a target. For a given aptamer, the obligate region for binding can be either the stem or the loop (or, less conveniently both) portion of the hairpin. Once the structure–activity relationship of the aptamer is minimally understood, the rational design of the stem length can be used to obtain greater FSA signal upon binding, which can then be used to improve assay S/N and increase assay sensitivity/performance.

Here, we observed that the FSA signal from small-molecule binding hairpin aptamers is highest when the product results from easily dissociable hairpin bonds. For tenofovir, as the length of the hairpin is increased beyond favorable ΔG and Tm conditions, the FSA signal magnitude degrades. Future work is focused on quantifying the binding affinity, specificity, limit of detection, and limit of quantification for hairpin aptamers to HIV p24, other proteins, and an array of small molecules, while implementing the findings presented here to aid in probe design to improve the sensitivity of quantitative assays.