Programming Fluorogenic DNA Probes for Rapid Detection of Steroids

Abstract

The ability of aptamers to recognize a variety of different molecules has fueled their emergence as recognition agents to probe complex media and cells. Many detection strategies require aptamer binding to its target to result in a dramatic change in structure, typically from an unfolded to a folded state. Here, we report a strategy based on forced intercalation (FIT) that increases the scope of aptamer recognition by transducing subtle changes in aptamer structures into fluorescent readouts. By screening a library of green-fluorescent FIT-aptamers whose design is guided by computational modeling, we could identify hits that sense steroids like dehydroepiandrosterone sulfate (DHEAS) down to 1.3 μM with no loss in binding affinity compared to the unmodified aptamer. This enabled us to study DHEAS in clinical serum samples with several advantages over gold standard methods, including rapid readout (< 30 min), simple instrumentation (plate-reader), and low sample volumes (10 μL).

Fluorogenic probes that selectively turn on in the presence of their target analytes have revolutionized chemical and biological analysis.^[1–10] Specifically, probes based on aptamers, nucleic acid sequences capable of binding to targets of interest, have recently emerged as effective tools for probing complex media and cells.^[11–18] By coupling aptamer-target binding to a fluorescence readout, several different classes of analytes ranging from simple ions to complex proteins can be detected and quantified.^[6,19–21] Aptamers are attractive as they can be rapidly synthesized, exhibit high stability, are amenable to various chemical modifications, and can be used as recognition moieties in strategies that involve simple sample processing and the use of widely available instrumentation.^[22] However, conventional aptamer-based strategies suffer from limitations such as kinetically slow responses, the presence of false-positive signals, and poor aptamer performance in complex milieu.^[19]

Recently, we introduced the concept of forced intercalation (FIT)-aptamers by leveraging structures that switch from an unfolded to a folded state or by designing structures with appended bases such that hybridization is deliberately induced upon target binding.^[23,24] FIT-aptamers contain a visco-sensitive dye as a base surrogate. Target binding leads to structural changes in the aptamer resulting in the forced intercalation of the dye between nucleic acid base pairs, turning on the dye’s fluorescence. Here, steroid-binding aptamers are used as examples to show that a FIT-strategy allows one to transduce subtle binding-induced structural changes to significant fluorescence enhancement. Such local structural changes are difficult to monitor by conventional techniques that rely on strand displacement.^[5,20] Importantly, the FIT strategy enabled us to develop the first fluorogenic aptamer capable of detecting dehydroepiandrosterone sulfate (DHEAS), a circulating adrenal androgen, in human serum and clinical samples.

We focus on steroids as these biomolecules play a central role in various physiological processes, including controlling gene transcription, regulating metabolism, and attenuating inflammatory response.^[25,26] Therefore, their dysregulation can lead to many clinical indications. For example, elevated levels of DHEAS are found in patients with adrenal tumors, congenital adrenal hyperplasia, and polycystic ovary syndrome.^[27] Steroids are typically detected using gas or liquid chromatography in tandem with mass spectrometry, immunoassays (e.g. Abbott Architect DHEAS test), or radioimmunoassays.^[27] However, these methods require bulky, expensive, or specialized equipment, entail long turnaround times, necessitate highly trained personnel for sample preparation, instrument operation, and data analysis, or involve multistep reaction procedures (Supporting Information, Section 3, Table S3).^[28–32] Consequently, rapid, sensitive, and simple fluorescence tests for studying steroids in physiological media remain highly sought after.

To address this challenge, we developed steroid-binding FIT-aptamers using thiazole orange (TO) as the visco-sensitive dye whose fluorescence (excitation: 485 nm, emission: 528 nm) is turned on when rotation about its methine bridge is restricted.^[33,34] We began by selecting a previously evolved aptamer for DHEAS.^[35] The minimum energy structure predicted by Mfold^[36] indicates that the free aptamer exists in a folded conformation at room temperature, containing three double-stranded regions (Figure 1; Supporting Information, Figure S1). These results are consistent with experimentally observed differential scanning calorimetry (DSC) traces (Figure S40). The junction of the double-stranded regions forms the binding pocket. Therefore, a strategy that could detect the subtle structural changes that would accompany DHEAS binding (Figure S35) needed to be developed.

Figure 1.

A) Secondary structure of the aptamer for DHEAS showing predicted fluorescence turn-on when FIT-aptamers are designed by substituting TO at different locations on the sequence. The FIT-aptamer library was generated by synthesizing 20 different sequences such that in each sequence one of the bases circled in black was substituted with TO. B) Chemical structure of an oligonucleotide with TO as a base surrogate.

By analyzing the structure of the DHEAS aptamer, we noted that there are four types of sites (Figure 1): i) single-stranded regions away from the binding site, ii) double-stranded regions away from the binding site, iii) double-stranded regions close to the binding site, and iv) bases in the binding site. We reasoned that if TO is placed in regions (i) and (ii), no fluorescence turn-on will be observed because the local environment of the dye should not change significantly upon DHEAS binding. On the other hand, placing TO in regions (iii) and (iv) may result in a fluorescence turn-on due to restricted internal rotation of the dye upon target binding. The aptamer structure after TO substitution was predicted by computational modeling using Mfold, where a base mismatch was used to simulate the incorporation of TO. From our initial set of simulations, we anticipated that regions within 2 base pairs of the binding site may undergo local conformational changes upon DHEAS binding sufficient for fluorescence turn-on (Supporting Information, Section 2.2).

To test our hypothesis, we synthesized a library of 20 variants of the DHEAS aptamer, each with a single point mutation (Figure 1, Table S1) where TO is used as a nucleo-base surrogate. Approximately 1500 points were screened in high throughput corresponding to selection buffers with different NaCl and MgCl₂ concentrations (Figures S5,S6). At each condition, 100 μM DHEAS was added and the absolute change in fluorescence was calculated as a percentage, ΔI (%), relative to the fluorescence of the aptamer alone (Figure 2, Figures S5,S6). The initial screen resulted in the selection of 5 sequences that turn on (sequences b, f, o, r, and s), which can be rationalized based on Mfold simulations (Figures S7–S27). ΔI (%) increases with increasing salt concentrations, reaching up to 100%. These results suggest that conditions in which the secondary structure of the aptamer is preserved to a greater extent facilitate DHEAS binding (Figures S28). The generality of this approach was further validated by using simulation guided design to program fluorescence into FIT-aptamers for other steroids such as deoxycorticosterone-21 glucoside and deoxycholic acid (Figures S36–39).

Figure 2.

Screening data showing percent fluorescence change (ΔI (%)) for 4 representative FIT-aptamers with TO substitution at different locations along the sequence. In all cases, 0.1 μM probe and 100 μM DHEAS were used. A, D, G, J) Schemes for sequence t, j, f, and s, respectively. TO substitution is denoted by a grey circle. B, E, H, K) Corresponding fluorescence enhancement heatmaps. C, F, I, L) ΔI (%) shown as bar graphs.

We next measured the fluorescence response of the 3 FIT-aptamers with > 50% fluorescence increase (sequences f, o, and s) as a function of DHEAS concentration (Figure 3). By fitting the data to the Hill equation, the apparent dissociation constants, K_d, were calculated to be 7, 23, and 21 μM, respectively. We note that DSC experiments suggest that in probe f, the TO-containing region is destabilized relative to the unmodified aptamer. The stability is partially recovered in the presence of DHEAS (Figure S40). However, the K_d of probe f is not significantly changed compared to the unmodified aptamer, as ascertained by isothermal titration calorimetry (Figure S34). In contrast, probes o and s, both of which have TO in the binding pocket, have higher dissociation constants, suggesting that modifications outside of the binding pocket are more desirable for creating probes with higher sensitivity. Notably, the apparent K_d of probe f designed using a FIT-strategy is significantly smaller than that observed when a fluorophore-quencher-based strategy that requires strand displacement is used.^[35]

Figure 3.

Sensitivity and selectivity of the best FIT-aptamers from the initial library screen. A, E, H) Schemes of sequence o, s, and f, respectively with the dye denoted by a grey circle. B, F, I) ΔI (%) of 0.1 μM of FIT-aptamers o, s, and f, respectively with varying DHEAS concentration. C, G, J) Selectivity of FIT-aptamers o, s, and f, respectively. Gray bars denote ΔI (%) after addition of 25 μM of different analytes to 0.1 μM probe. Colored bars represent ΔI (%) after DHEAS is added to this mixture. “Probe” denotes FIT-aptamers with no added analyte. D) Structures of the biomarkers used for selectivity studies.

Next, the specificity of aptamers f, o, and s for DHEAS was investigated (Figure 3). Due to the structural similarity of DHEA and DHEAS and the evolutionary promiscuity of these aptamers towards deoxycorticosterone-21 glucoside,^[35] fluorescence turn-on is observed in the presence of these two steroids. However, physiological concentrations of DHEAS are 1–2 orders of magnitude higher than those of deoxycorticosterone-21 glucoside and DHEA, alleviating challenges with cross-reactivity in biologically relevant media.^[37–39] Moreover, other off-target biomarkers like deoxycholic acid, neuropeptide Y, and β-estradiol cause negligible turn-on, and the presence of off-target biomarkers in a mixture of the aptamer and DHEAS does not inhibit binding. Importantly, these results show that for the structures studied, the selectivities of FIT-aptamers are the same as that of the original aptamer.^[35] Given that all three show comparable selectivity, we identified probe f to be the best FIT-aptamer in the library since it has the lowest K_d (and consequently, a limit of detection of 1.3 μM). Therefore, we next evaluated its ability to sense DHEAS in serum.

Studies where DHEAS was spiked into steroid-free human serum showed a significant loss in binding affinity (K_d = 198 μM) of the aptamer (Figure S30), likely due to interaction of DHEAS with serum proteins.^[40] We therefore developed a plasma crash procedure to break these interactions and remove proteins from serum (Figure 4A, Figure S31). Briefly, the proteins were denatured with ethylenediaminetetraacetic acid (EDTA) and sodium dodecyl sulfate (SDS) and precipitated with acetonitrile (ACN). The supernatant containing DHEAS was dried and resuspended in buffer. Thereafter, the FIT-aptamer was added for taking fluorescence readings. Serum samples with DHEAS spiked in before and after the treatment showed that this procedure yields fluorescence turn-ons comparable to those observed in buffer (Figure S32), indicating that the treatment is effective in extracting DHEAS (Figure 4B).

Figure 4.

DHEAS measurement in serum. A) Plasma crash procedure. Created using biorender.com. B) ΔI (%) of FIT-aptamer f after spiking in DHEAS in steroid-free serum. C) Measured raw fluorescence of target and control probes in clinical serum samples. D) ΔI (%) in clinical samples plotted against concentration reported by the Abbott Architect assay.

We next assessed the ability of the FIT-aptamer to make measurements of DHEAS in clinical serum samples. Specimens from 13 different individuals with known DHEAS levels, measured using the Abbott Architect chemiluminescent assay, were acquired and tested blind using probe f. A control probe (sequence n) that does not result in fluorescence turn-on upon binding DHEAS was used as a baseline to enable comparisons between the different samples. The observed fluorescence values were sorted based on the known DHEAS concentrations (Figure 4C). The target probe yields a Spearman’s rank correlation coefficient of 0.90 (p < 2.2E-16) between the average fluorescence and the concentration, whereas the control probe yields a coefficient of 0.42 (p = 0.16). These results show that the signal from probe f increases monotonically with increasing concentrations of DHEAS whereas that from the control probe does not. Moreover, a χ² goodness-of-fit test verified that the percent fluorescence change relative to the control probe observed as a function of concentration is well-described by the Hill-equation (Figure 4D). Taken together, these results show that FIT-aptamers can be utilized for the measurement of clinically relevant concentrations of steroids in serum. Importantly, the FIT strategy enabled us to use a plate reader-based format to make high throughput measurements with sample volumes as low as 10 μL in 30 min (Figure S33). This reduces the volume of serum required by up to 20-fold compared to routinely used electrochemiluminescence immunoassays and yields immediate results.^[41] We note that the limit of detection can be improved by using other FIT-dyes^[34] or a different aptamer sequence with a lower K_d.

In summary, this work shows that fluorogenic aptamers for steroids can be programmed using a FIT strategy with the aid of simulation-guided design. Importantly, these results show that global structural transitions from an unfolded to a folded state are not necessary for the design of FIT-aptamers. If designed correctly, these structures can retain the binding affinity of the original aptamers. This is particularly important considering that commonly used strand displacement strategies require partial blocking of the aptamer recognition site and, therefore, result in slower binding and increased K_d, both of which impede sensitive target detection in complex biological media.^[20,23,35] Specifically, for steroids, the FIT strategy allows rapid (< 30 min), high throughput detection using low sample volumes (10 μL), without the use of complex instrumentation. These results bode well for the use of FIT-aptamers as a new and general strategy for sensing other important biological markers whose binding may lead to subtle changes in aptamer structure. Moreover, given that a broad range of FIT-dyes spanning different wavelengths have been previously developed,^[34] one can envision multiplexed detection of various analytes.