Capture‐SELEX of DNA Aptamers for Highly Selective Binding of Folate Authors: Yachen Xie, Yifan Chen, Juewen Liu Published: 2026 ABSTRACT AbstractFolate (vitamin B9) is an essential micronutrient for human health, particularly during pregnancy. Aptamers have emerged as attractive recognition elements for small‐molecule detection, as demonstrated by numerous natural riboswitches. Although an RNA aptamer for folate has been reported, the development of a DNA aptamer is highly desirable for biosensing applications due to its superior chemical stability and ease of modification. Herein, we employed a library‐immobilization‐based selection strategy to isolate a family of high‐affinity DNA aptamers for folate. Isothermal titration calorimetry (ITC) measurements yielded a dissociation constant of 640 nM. Based on the selected aptamer (FA1), we developed a signal‐on, strand‐displacement fluorescent biosensor by hybridizing a 5′‐FAM‐labeled aptamer with a short quencher‐labeled complementary strand. The sensor achieved a limit of detection (LOD) of 8.5 nM in selection buffer and 31 nM in 10% serum. Selectivity studies demonstrated negligible responses to structurally related compounds, including guanine, adenine, theophylline, xanthine, and caffeine. Tetrahydrofolate generated a measurable signal only at elevated concentrations; however, ITC confirmed no direct binding, indicating that the observed response likely arises from its oxidation to folate. These results establish a robust DNA aptamer for folate for both fundamental studies and applied contexts, such as clinical monitoring and food fortification quality control.Keywords: biosensor, DNA aptamer, folate, in vitro selection, SELEX Introduction Folate (vitamin B9) is an essential member of the B‐vitamin family. It was first isolated in 1941 from spinach leaves and named from the Latin folium—‘leaf’ [1]. Folate is widely distributed in nature, with particularly high concentrations in animal livers and kidneys as well as dark‐green leafy vegetables. However, processing and canning can markedly reduce its content since folate is sensitive to various conditions, such as temperature, oxygen, light, and pH [2]. As a water‐soluble vitamin, folate has limited body storage with a rapid turnover, necessitating regular intake to maintain a healthy level. Folate lies at the core of one‐carbon metabolism, and folate deficiency is a common vitamin deficiency. In its active coenzyme forms, folate donates single‐carbon units required for the synthesis of purine bases (adenine and guanine) and thymidylate, thereby supporting DNA synthesis, repair, and methylation [3]. In U.S. adults, the average serum folate concentration is ≈11.9 ng/mL (∼0.027 µM), while intracellular folate is about 254 ng/mL (∼0.58 µM) [4]. Serum concentrations can drop to around 0.009 µM in female adults [5]. Folate deficiency increases the risk of neural tube defects, a common severe congenital malformations [6]. During pregnancy, an additional 0.2 mg/day intake is associated with a 23% reduction in neural tube defects risk [7]. Folate is also relevant in oncology: many tumors have a high demand for folate and may overexpress folate receptor‐α to facilitate uptake, a feature that has been exploited for targeted tumor imaging and drug delivery [8]. Accurate assessment of folate status is therefore essential for both individual health and population‐level monitoring [9, 10]. Numerous analytical approaches have been developed for folate quantification [11, 12, 13]. Among these, immunoassays are the most widely established and typically employ an indirect competitive format [14, 15]. In such assays, the signal decreases as the target folate concentration increases, which can complicate data interpretation and reduce analytical robustness [16]. Electrochemical methods have also been explored, leveraging the intrinsic redox activity of folate. Although nanomolar sensitivity can be achieved through signal amplification strategies [17], these approaches are often susceptible to interference from other redox‐active species in complex sample matrices. Consequently, there remains a strong need for folate detection methods that combine high selectivity with reproducibility and robustness. Aptamers are single‐stranded DNA or RNA that fold into defined tertiary structures upon binding to their target molecules [18, 19]. These binding structures engage targets often through nonconical base pairing interactions, hydrogen bonding and stacking, enabling high‐specificity recognition [20, 21]. Moreover, the binding process is reversible, a property that supports regeneration and real‐time monitoring of target concentrations under varying conditions [22, 23]. An RNA aptamer for folate was reported by Terasaka et al. The authors used a so‐called smart‐SELEX (Systematic Evolution of Ligands by EXponential enrichment) strategy and identified a human precursor microRNA motif, known as hsa‐pre‐miR‐125a, whose terminal loop region interacts with folate [24]. In their workflow, folate was immobilized on magnetic beads to facilitate aptamer selection. By using biolayer interferometry, the aptamer’s dissociation constant (K d) was determined to be 1.9 μM, which is substantially weaker than what is expected to detect physiological folate levels (0.027 µM in serum). Consequently, its utility for bioanalytical assays in serum is limited. In addition, selectivity against related metabolites, such as tetrahydrofolate, was not reported, leaving the application feasibility uncertain. Finally, the poor stability of RNA also made it less attractive for detection applications outside cells. For extracellular sensing applications, DNA aptamers offer distinct advantages over their RNA counterparts, including significantly greater chemical stability (DNA is approximately six orders of magnitude more stable than RNA) and lower synthesis cost [25, 26, 27]. Through the SELEX process, numerous DNA aptamers targeting small molecules have been isolated with high affinity and specificity [28, 29]. In addition, real‐time sensing formats have been demonstrated, including highly selective detection of ATP [30], in vivo electrochemical detection of neurotransmitters [31], and wearable aptamer‐based platforms for continuous drug monitoring in clinical settings [32]. In this work, using a capture‐SELEX strategy, we selected a DNA aptamer that binds folate with nanomolar affinity. The aptamer was further characterized using spectroscopic and calorimetric techniques, confirming both high affinity and excellent selectivity for folate. Building on these properties, we engineered a turn‐on fluorescent biosensor capable of detecting folate in serum samples. Materials and Methods Chemicals All of the DNA oligonucleotides were purchased from Integrated DNA Technologies (Coralville, IA). See Table S1 for the DNA sequence. Streptavidin‐coated agarose resin possessed a loading capacity of 1–3 mg biotinylated BSA per milliliter was purchased from Thermo Scientific (IL, USA).10 K Ultra‐0.5 centrifugal filter units and human serum were purchased from Millipore‐Sigma (Oakville, ON, Canada). 10x PBS, Micro bio‐spin chromatography columns and SsoFast EvaGreen supermix were purchased from Bio‐Rad Laboratories (Hercules, CA, American). The dNTP mix, Taq DNA polymerase with ThermoPol buffer, and the low‐molecular‐weight DNA ladder were from New England Biolabs (Ipswich, MA). Folate, adenine, guanine, theophylline, caffeine, xanthine, NaCl, MgCl2, and other chemicals were purchased from Sigma–Aldrich (MO, USA). Tetrahydrofolate was purchased from Cayman Chemical (MI, USA). Milli‐Q water (18.2 MΩ·cm−1) was used in all experiments. All the sequences used in this study are shown in Table S1. Capture‐SELEX for Folate Aptamer The aptamer selection was performed by a library immobilized SELEX (so‐called capture‐SELEX) method following established protocols with minor modifications [33, 34, 35, 36]. Briefly, a single‐stranded DNA (ssDNA) library was immobilized on streptavidin agarose resin through hybridization with a biotinylated capture strand. The mixture was washed with selection buffer (1× PBS, containing 2 mM MgCl2, pH 7.5) to remove unbound ssDNA. Subsequently, the same buffer containing 50 µM freshly prepared folate was added, and the eluted DNA was concentrated and amplified by PCR. The resulting biotinylated duplex PCR products were bound to streptavidin agarose beads, and the beads were washed with separation buffer (1× PBS, no Mg2+, pH 7.5). The PCR products were denatured with 0.2 M NaOH at room temperature for 10 min to denature the dsDNA, and the resulting ssDNA was purified to yield the library for the subsequent selection round. Except for the first round (500 pmol), the amount of the DNA pool was 100 pmol, and the folate concentration was 50 µM for the entire 12 rounds. The round 12 PCR products were deep sequenced. Thioflavin T (ThT) Fluorescence‐Based Binding Assays ThT fluorescence assays were performed on a Varian Eclipse or a Horiba FluoroMax 4 fluorometer. An aptamer (100 μM, 5 μL) and ThT (1 mM, 5 μL) were added to selection buffer (490 μL). Then, the solution was transferred to a quartz cuvette, and folate was gradually titrated. The fluorescence emission values at 490 nm were recorded upon excitation at 440 nm. The K d was determined by fitting the data using F = F 0 + aK d/(K d + x), where x denotes the concentration of the titrated target, and a represents the maximum fluorescence change upon saturated binding. Each experiment was repeated three times. Isothermal Titration Calorimetry (ITC) The ITC measurements were performed using a MicroCal ITC 200. The selection buffer was used to prepare all aptamers and target solutions. DNA aptamers were first heated at 95°C, cooled to room temperature, and then degassed for 5 min before loading. Fresh folate or tetrahydrofolate solutions were prepared to ensure accurate titration. The syringe held 38.5 μL of a target solution, and the cell chamber held 209 µL of an aptamer solution. Except for a 0.5 μL initial injection, the subsequent 19 injections with 2 μL each were performed at 25°C. Each injection lasted 20 s, and the intervals between each injection were fixed at 200 s. Using Origin software, the titration curves were fitted with a one‐site binding model to determine the binding constants and thermodynamic parameters. The fitting is based on the Wiseman isotherm, where the heat per injection is expressed as a function of the molar ratio, stoichiometry (n), association constant (K a), and enthalpy change (ΔH), with the entropy change (ΔS) derived from ΔG = ΔH−TΔS = −RTlnK a [37]. Fluorescent DNA Strand‐Displacement Assays A sensor complex was formed by annealing 1 µM FAM‐labeled aptamer and 2 µM quencher‐labeled short complementary strand in the selection buffer at 95°C and cooled to room temperature. This solution then was diluted to 20 nM in selection buffer for use. The kinetic measurements were carried out with a Tecan F200 Pro microplate reader. In each well, 98 μL of 20 nM sensor complex was added. The excitation and emission wavelengths were set at 485 nm and 535 nm, respectively. When the background fluorescence was monitored for 5 min and stabilized, a 2 μL target solution was added, and the sample was continued to be monitored for at least 25 more minutes with 30 s intervals. Each experiment was repeated three times. For the test in serum, the sensor complex was diluted to selection buffer with 10% human serum. Results and Discussion Library Immobilization‐Based Aptamer Selection The structure of folate contains three parts as shown in Figure 1A, a pteridine ‘head,’ a p‐aminobenzoate spacer, and an L‐glutamate ‘tail.’ The pterin’s large, planar aromatic π‐system, its ring nitrogen atoms, and the p‐aminobenzoate ring with a para‐amino group and two amide carbonyls present well‐defined hydrogen bonding and stacking motifs that can be engaged by aptamers. Accordingly, we expected that high affinity aptamers for folate can be generated. In this work, a DNA library containing a 30‐nucleotide randomized region flanked by primer binding sequences was first hybridized to a biotinylated capture strand, and the resulting duplex was immobilized on streptavidin agarose beads [38, 39]. The two flanking primer binding regions were designed to form a short base‐paired stem that promotes aptamer folding and release from the capture strand upon target binding. After incubation with a folate solution, the released DNA sequences were collected and amplified by PCR to prepare a library pool for the next round of SELEX. Selections were performed for 12 rounds using 50 µM freshly prepared folate each time, and the last round PCR products was subjected to deep sequencing. (A) The structure of folate. (B) Alignment of the top 10 sequences from round 12 of the folate selection. The pink region is the conserved region. The underline represents the possible pairing in the secondary structure. (C) The predicted secondary structures of FA1 aptamer. (D) The predicted secondary structures of FA2 aptamer. A total of 38,274 sequencing reads were obtained, and Figure 1B shows the top 10 sequences ranked by abundance. Interestingly, all these 10 sequences belong to the same family with a 9‐nucleotide G/T rich conserved region marked in pink. Interestingly, these conserved nucleotides are either at the left or right side of the library region. The nucleotides outside the conserved region can form a hairpin structure (colored blue, with the stem regions underlined). We then used the conserved motif to search the entire library, and this family of sequence occupied 67.7% of the final library. FA1 and FA2 were chosen as representatives of the two arrangements based on their high abundance. According to the secondary structure predicted by mFold [40], both FA1 (Figure 1C) and FA2 (Figure 1D) adopt a structure with conserved region residing in a loop flanked by two nonconserved long stems. From the secondary structures, it is easy to see that the internal hairpin and the conserved motif can be circular permutated [25, 41]. ThT Fluorescence Spectroscopy Based Binding Assays To evaluate these two aptamers, a ThT dye‐displacement assay was used. ThT is a benzothiazole dye that becomes strongly fluorescent upon nonspecific binding with ssDNA. Upon target binding, the aptamer may release some pre‐associated ThT dyes, leading to a decrease in fluorescence [42]. Therefore, ThT binding assay provides a convenient label‐free screen for apparent binding affinity [43]. As shown in Figure S1, the addition of folate to ThT‐FA1 complex produced a concentration‐dependent decrease in fluorescence. Fitting the response to a one‐site binding model yielded an apparent K d of 1.3 µM. The ThT‐FA2 complex also responded to folate, with an apparent K d of 3.8 µM. Since FA1 and FA2 belong to the same sequence family, we picked FA1 for subsequent characterization. Aptamer‐Binding Assay Using ITC We then used ITC to quantify FA1aptamer binding affinity. ITC measures the heat released or absorbed upon binding interactions and can be used to determine thermodynamic parameters of binding [44, 45]. Because it is a label‐free method performed in solution without any additional components, ITC provides an unbiased determination of affinity. In our experiment, 200 µM folate was titrated into 20 µM FA1 aptamer, and the ITC results are show in Figure 2A. The downward injection peaks indicated an exothermic process, and their progressively decreasing amplitudes with successive injection reflected saturation of binding sites. After subtracting heats of folate dilution (Figure S2), integration and one‐site model fitting gave a K d of 0.64 µM, which was in good agreement with the results obtained from the ThT‐based assay. This reaction has a ΔH = −18.3 kcal/mol and a ΔS = −33 cal/mol/K. Such a large heat release suggested that folate has established extensive contact with the aptamer with an accompanying loss of flexibility. These results establish FA1 as a high affinity aptamer for folate. The ITC thermogram and integrated heat (after background subtraction) for titrating 150 µM folate into 20 μM (A) FA1 aptamer or (C) FA1m mutant at 25°C in selection buffer. (B) Predicted secondary structure of FA1m, the mutation bases are highlighted in red. As a negative control, we designed a mutant, named FA1m, by switching two GT dinucleotides to TG within the conserved region (Figure 2B). Upon titrating 150 µM folate into 20 µM FA1m, the ITC trace showed no detectable exothermic binding signal. Instead, only very small endothermic heats were observed, consistent with the heat of dilution of folate (Figure 2C and Figure S2). Thus, FA1m does not binding folate, confirming that the conserved region in FA1 is required for recognition. It is interesting that given the large size of folate, we only obtained a single family of its aptamer with nanomolar affinity. Likely, the binding of the aptamers involved the pteridine ring, which can interact with the aptamer with extensive hydrogen bonding and pi–pi stacking. The other end of the molecule, containing a glutamate moiety, is unlikely to support such nanomolar binding affinity, as a previously reported DNA aptamer for glutamate exhibited a K d of only ~12 µM [46], nearly 30‐fold weaker than the affinity observed here. Fluorescent Sensing of Folate After confirming the binding affinity of the FA1 aptamer, we next designed a turn‐on biosensor that exploits its strand‐displacement property [47]. The FA1 aptamers was extended by 5‐nt at the 5′‐ends and labeled with a 6‐carboxyfluorescein (FAM) fluorophore. It was then hybridized to a 12‐mer quencher‐labeled short ssDNA to yield a quenched complex (Figure 3A). FAM is a widely used green fluorophore with ∼495/520 nm excitation/emission, readily conjugated to oligonucleotides at a low cost, and a high quantum yield. Upon the addition of folate, target binding induces a conformation change of FAM‐aptamer quencher complex that releases the quencher strand, producing enhanced fluorescence [47]. (A) Schematic illustration of the fluorescent strand‐displacement biosensor for folate detection. (B) Real‐time fluorescence traces of the FAm‐FA1/quencher–DNA complex following the addition of folate at 5 min. (C) Calibration curve of the sensor for the detection of folate. Inset: the linear response at low folate concentrations. Error bars represent standard deviation (SD) from three independent experiments (n = 3). To quantitatively evaluate the sensor response, we first monitored the baseline fluorescence of 20 nM FAM‐FA1/quencher–DNA complex, which was low (~7,500 arbitrary units). Sequential additions of folate produced immediate, concentration dependent increases, approaching a plateau within almost 3 min (Figure 3B). Using the fluorescence change at 25 min, we constructed a calibration curve of Δ F versus folate concentration (Figure 3C). By fitting to a one site model, an apparent K d of 6.3 µM folate was obtained. The apparent K d of biosensor is 10‐fold higher than the K d value obtained from ITC (0.64 µM), consistent with competition from the quencher‐DNA [48, 49]. The inset of Figure 3C shows the linear fluorescence increase at low folate concentrations. Based on this, the LOD was calculated to be 8.5 nM based on the signal being higher than three times the standard deviation of background variation. This LOD compares favorably with magnetic‐bead electrochemical immunoassays for folate (~13 nM in milk matrices) and remains within one order of magnitude of chemiluminescent immunoassays (~1 nM) [14, 16]. More importantly, the strand‐displacement aptamer biosensor operates without the need of enzymatic signal amplification and provides a direct and reliable readout. By using fluorophore that are even more stable, even better reproducibility can be achieved. Selectivity of the FA1 Aptamer Biosensor To study the selectivity of the FA1 aptamer, we challenged the FA1 sensor with compounds bearing related fused N‐heteroaromatic scaffolds, such as tetrahydrofolate, guanine, adenine, theophylline, xanthine, and caffeine (Figure 4A). As shown in Figure 4B, tetrahydrofolate produced a positive response at both 2 and 20 µM, whereas the other analytes produced no detectable fluorescence change. (A) Structures of compounds with N‐heteroaromatic scaffolds. (B) Selectivity of the FAM‐labeled FA1 biosensor toward folate and related biomolecules at 2 and 20 µM. (C) Real‐time fluorescence traces in response to freshly prepared tetrahydrofolate at the indicated concentrations. (D) Concentration‐dependent fluorescence response for fresh tetrahydrofolate solutions (ΔF/F 0 at 20 min). Error bars represent standard deviation (SD) from three independent experiments (n = 3). Because tetrahydrofolate is the active form of folate and is readily oxidized in air, we prepared a fresh tetrahydrofolate solution for a careful titration (Figure 4C). The kinetic measurements showed no significant signals at low tetrahydrofolate concentrations (0 to 5 µM) and the maximum signal change was only 483 a.u., whereas higher concentrations (20 or 50 µM) led to pronounced increase (up to 5629 a.u.). The resulting binding curve did not reach saturation (Figure 4D). We further tested FA1 aptamer binding to tetrahydrofolate using ITC. Titrating 200 μM tetrahydrofolate into FA1 produced no substantial binding heat; the integrated heats were small and matched the tetrahydrofolate dilution heat (∼4 kcal·mol−1; Figure S3). Therefore, we reasoned that the fluorescence responses to tetrahydrofolate were due to its oxidation at low tetrahydrofolate concentrations. For ITC experiments, a high concentration of tetrahydrofolate was used and this sample was sealed in a syringe, the fraction of oxidation by dissolved oxygen is much lower. Similar observations were also made with the measurement of bilirubin aptamer binding to bilirubin, which can be oxidized to biliverdin [29]. Collectively, these results show that FA1 is highly selective for folate: structurally related heterocycles do not activate the sensor, and even the folate metabolite tetrahydrofolate can be discriminated against. Detection of Folate in Serum To test the potential application of the folate biosensor in complex sample matrix, we then evaluated it in human serum. Because undiluted serum strongly scatters light and elevated background, 10% (v/v) serum diluted in selection buffer was used. Concentration‐dependent responses recorded in 10% serum (Figure 5A,B) closely matched those obtained in buffer, and the sensor exhibits a dynamic range from 31 nM to ~20 µM in 10% serum. The LOD was determined to be 31 nM, which was slightly higher than that in buffer (8 nM). These results indicate minimal matrix interference at this dilution and confirm that aptamer’s function is mostly preserved in serum. At its current sensitivity, the sensor is suitable for applications involving elevated folate levels, such as intracellular folate quantification (~0.58 µM) and food‐fortification quality control. The clinical folate deficiency threshold is ~9 nM for measurement in blood [5], and this should be able to be achieved by designing advanced signaling methods or signal amplification. Future adoption of red‐emitting fluorophores [50] signal amplification by enzymes [51], and nanomaterials [52] may enable direct detection of folate at physiological concentrations. (A) Real‐time fluorescence signaling kinetic traces of the FAM‐FA1/quencher–DNA complex in 10% serum following the addition of folate at 5 min. (B) Calibration curve of the biosensor for the detection of folate in 10% serum. Inset: the linear response at low folate concentrations. Error bars represent standard deviation (SD) from three independent experiments (n = 3). Conclusion In summary, we selected and validated a folate‐binding DNA aptamer, FA1, using the capture‐SELEX strategy that immobilizes a DNA library rather than the target, thereby minimizing selection bias caused by target structural changes. The FA1 aptamer has a highly conserved loop and mutations inside the loop abolished binding. ITC yielded a K d of 0.64 μM with binding predominantly driven by a favorable enthalpy change. A turn‐on fluorescence sensor was designed by pairing a 5′‐FAM–labeled aptamer with a short quencher labeled strand. The sensor reports folate in real time, reaching near‐steady signals within ~ 3 min, with an LOD of 8.5 nM in buffer and 31 nM in 10% serum. Selectivity studies showed no response to competing molecules, except for tetrahydrofolate, whose response was attributed to its oxidation to produce folate in situ. Together, these results provide a chemically stable, cost‐effective DNA aptamer and a simple, regenerable turn‐on biosensor capable of folate detection in serum with a dynamic range suitable for applications such as intracellular folate quantification and food fortification quality control. Funding This study was supported by the Natural Sciences and Engineering Research Council of Canada (NSERC) and the Canada Research Chairs program. Conflicts of Interest The authors declare no conflicts of interest. Supporting information Supplementary Material References Associated Data Supplementary Materials Supplementary Material