Self‐Calibrating Ratiometric DNA Aptamer Probe for Quantitative ATP Imaging in Living Cells

ABSTRACT

Accurate monitoring of adenosine triphosphate (ATP)—the universal energy currency of cells—is essential for elucidating cellular metabolism and disease progression. However, the high basal concentration of intracellular ATP (1–10 mM) and variable probe uptake during imaging have hampered the development of reliable fluorescent sensors. Although DNA aptamer‐based probes provide excellent selectivity, conventional turn‐on designs often lack internal calibration, and Förster resonance energy transfer‐based ratiometric probes typically exhibit limited signal changes. We report a ratiometric DNA duplex sensor comprising a Cy5‐labeled ATP aptamer and a thiazole orange (TO)‐labeled exciton‐controlled hybridization‐sensitive fluorescent oligonucleotide (ECHO) probe. Upon ATP binding, the aptamer structure switches and releases the reporter strand, resulting in a pronounced decrease in TO fluorescence while the Cy5 signal remains constant. Rational insertion of a polythymidine spacer effectively suppressed undesired TO‐to‐Cy5 energy transfer, enabling a reliable ratiometric Cy5/ECHO readout. The sensor operates robustly across physiological ATP concentrations, exhibits high nucleotide selectivity and satisfactory serum stability, and shows minimal cytotoxicity. Live‐cell flow cytometry and confocal imaging further confirmed that cancer cells displayed significantly higher Cy5/ECHO ratios than normal fibroblasts. This internally self‐calibrating aptamer sensor thus provides a powerful platform for intracellular ATP imaging and cancer diagnostics.

We developed the RFA–ECHO probe, an internally calibrated ATP sensor that remains responsive across millimolar ATP levels. Its self‐correcting Cy5/ECHO ratio ensures accurate quantification independent of probe uptake. Compared with FRET‐based sensors, it yields a robust ratiometric shift, distinguishing high‐ATP cancer cells from normal ones, enabling reliable metabolic imaging and diagnostics.

1. Introduction

Adenosine triphosphate (ATP)—the universal energy currency of cells—serves as a central indicator of metabolic state. Intracellular ATP is typically maintained at millimolar concentrations (1–10 mM in mammalian cells) [1, 2], a level that supports numerous energy‐intensive processes, from biosynthesis to signal transduction. Despite this high baseline, ATP levels fluctuate dynamically in response to nutrient availability, stress, and disease. For example, cancer cells often maintain markedly elevated ATP pools—up to 2‐ to 8‐fold higher nuclear ATP than normal cells [3, 4, 5, 6]—reflecting their hyperactive glycolysis and anabolic demands. Conversely, ATP depletion is a hallmark of apoptosis, ischemia, and drug‐induced cytotoxicity. These observations underscore the conclusion that ATP homeostasis is tightly linked to cell fate decisions and pathophysiology [7]. Thus, real‐time monitoring of ATP dynamics in living cells is indispensable not only for elucidating fundamental bioenergetic processes but also for identifying metabolic hallmarks of disease and evaluating therapeutic interventions [8, 9].

A major challenge lies in developing live‐cell probes that can faithfully report ATP fluctuations within the physiological millimolar range while avoiding artifacts arising from cellular heterogeneity and variable probe uptake. Numerous fluorescent ATP sensors have been developed, including genetically encoded protein indicators, small‐molecule probes, and DNA aptamer‐based nanodevices [10]. Each approach presents trade‐offs. Protein‐based ATP sensors often require genetic transfection and can become saturated under high‐ATP conditions [11]. Some small‐molecule “turn‐on” probes exhibit submicromolar affinities and large signal changes but saturate well below physiological ATP levels. For instance, a recently developed nuclear‐targeted probe (NucATP) showed a linear fluorescence response only up to 1.2 mM ATP [12], far below the typical intracellular concentration. DNA aptamer‐based sensors offer excellent selectivity, high biocompatibility, and detection windows well aligned with physiological ATP levels [13]. However, most conventional “turn‐on” aptamer probes lack an internal reference channel [5, 14, 15] making them prone to imaging artifacts. Variations in probe delivery or uneven intracellular distribution can reduce fluorescence intensity, making it unclear whether a weak signal reflects genuinely low ATP levels or poor probe uptake [16]. To mitigate this, aptamer probes based on Förster resonance energy transfer (FRET) have been developed to provide ratiometric readouts. Nevertheless, such designs typically yield only modest fold‐changes, limiting their sensitivity to subtle ATP fluctuations [14]. Ideally, a live‐cell ATP sensor should combine a broad millimolar detection range with a robust ratiometric output, enabling quantitative ATP imaging that is independent of experimental variability.

Here, we introduce a ratiometric fluorescent aptamer–ECHO (RFA–ECHO) probe for ATP sensing in living cells. The probe consists of a Cy5‐labeled ATP aptamer (recognition strand) and a thiazole orange (TO)‐labeled reporter strand (Figure 1a). The aptamer, originally developed by Huizenga and Szostak [17], selectively binds ATP and undergoes a structure‐switching transition. The reporter is an exciton‐controlled hybridization‐sensitive oligonucleotide (ECHO; Figure 1b) that carries a TO homodimer that is nonfluorescent in the single‐stranded state but strongly fluorescent when hybridized [18]. The sequences of the recognition strand and the reporter strand are shown in Figure 1c. In the duplex form, the RFA–ECHO probe exhibits a low Cy5/ECHO ratio, whereas ATP binding induces aptamer folding and reporter release, quenching the TO fluorescence and increasing the ratio. The probe shows more than a fivefold change in fluorescence between low‐ and high‐ATP concentrations, with a dynamic range of 1–10 mM matching physiological ATP levels. This design enables accurate ratiometric imaging of intracellular ATP fluctuations without relying on FRET.

FIGURE 1.

Schematic of the RFA–ECHO probe for ATP detection. (a) The ratiometric fluorescence mechanism for ATP detection by the RFA–ECHO probe. (b) Chemical structure of the thiazole orange (TO) homodimer‐labeled nucleotide incorporated into the RFA–ECHO probe. (c) Sequences of the recognition and reporter strands. Underlined nucleotides indicate the reported ATP aptamer sequence. X = Cy5; D = TO homodimer‐modified thymidine.

While fluorophore labeling, spacer engineering, and ratiometric formats have been individually explored in previous aptamer‐based probes, the novelty of our work lies in the integration of an ECHO reporter with a structure‐switching aptamer that is specifically optimized for ATP detection in the millimolar range, as well as in the elimination of FRET‐based dependency through a nonoverlapping and photophysically decoupled Cy5/ECHO architecture. The two fluorophores, Cy5 as the reference and TO as the reporter, are spectrally and spatially isolated (TO: λ _ex = 488 nm, λ _em = 530 nm; Cy5: λ _ex = 642 nm, λ _em = 660 nm), which minimizes crosstalk and removes the distance and orientation constraints inherent to traditional FRET systems. Second, although not designed as a FRET pair, we discovered and quantitatively characterized unintended TO to Cy5 quenching caused by duplex‐induced proximity, and we systematically resolved this effect by introducing a tunable polythymidine spacer. The resulting Förster‐type behavior was validated through both theoretical and experimental analyses. Third, the RFA–ECHO probe provides more than a fivefold ratiometric change across 1 to 10 mM ATP, covering the entire intracellular concentration range, while offering internal signal calibration through the nonresponsive Cy5 channel. Although ECHO probes have previously been applied to nucleic acid detection, this study represents, to our knowledge, the first use of an ECHO and Cy5 pair within a structure‐switching aptamer platform for live‐cell metabolite imaging.

2. Results and Discussion

2.1. Synthesis and Optimization of the RFA–ECHO Probe

The recognition strand contains a reported DNA aptamer sequence that binds two nonequivalent ATP molecules via intercalation between purine bases with the formation of identical G·A mismatches (Figure 2a) [17]. The reporter strand is partially complementary to the recognition strand and was designed as a doubly TO‐labeled probe. The reporter strand was prepared as follows. Diamino‐modified DNA was synthesized on an automated DNA synthesizer using standard phosphoramidite chemistry. The TO N‐hydroxysuccinimide ester was subsequently conjugated, and the resulting products were purified by HPLC and characterized by MALDI–TOF MS (Figure S1) [19]. A Cy5‐free version of the RFA–ECHO probe (composed of the ATP‐binding DNA aptamer and the reporter strand) was first designed to examine its intrinsic fluorescence properties. The fluorescence of the ECHO reporter strand is highly dependent on the local duplex structure, as the TO homodimer emits only when stacked within a rigid, hybridized environment. In our design, the TO‐modified thymidine (TO–T) was positioned near the center of the reporter strand. This central placement ensures that, upon hybridization with the aptamer strand, the TO dimer is flanked by multiple base pairs, forming a thermodynamically stable duplex that promotes efficient exciton coupling and strong fluorescence emission in the absence of ATP [20]. As the ATP concentration increased, the reporter fluorescence (Ex/Em = 488/530 nm) decreased, consistent with ATP‐induced duplex dissociation. The quenching of fluorescence was also visually observable to the naked eye (Figures 2b and S2). These results established the feasibility of the probe design. Next, the aptamer was labeled with Cy5 at the 3′‐end and hybridized with the reporter strand to form the complete RFA–ECHO probe. PAGE analysis again demonstrated ATP‐dependent duplex dissociation (Figure 2c, d). However, the fluorescence behavior deviated from expectation: both the Cy5 (reference) channel and the reporter (green) channel remained essentially constant across ATP concentrations. Notably, under 488 nm excitation, an emission at 660 nm was detected that decreased with increasing ATP (Figure 2e), suggesting energy transfer from TO (donor) to Cy5 (acceptor) [21]. Although the dyes are located on separate strands, duplex formation brings them into close proximity; given the R⁻⁶ distance dependence of FRET, such cross‐channel transfer can suppress the apparent TO signal (Figure 2f).

FIGURE 2.

Cy5 labeling induces FRET‐mediated signal loss. (a) Sequences of the recognition and reporter strands. Underlined nucleotides indicate the reported ATP aptamer sequence. X = Cy5; D = TO homodimer‐modified thymidine. (b) Fluorescence response of the DNA duplex lacking Cy5 in the presence of increasing ATP concentrations. The binding of ATP to the aptamer induces a conformational switch that releases the recognition strand, resulting in a progressive decrease in reporter fluorescence. Fluorescence of the Cy5‐free DNA duplex (25 nM) was measured in buffer (20 mM HEPES, 150 mM NaCl, 5 mM MgCl₂, pH 7.4) containing ATP at 37°C (λ _ex = 488 nm, λ _em = 530 nm). (c) Native polyacrylamide gel electrophoresis (PAGE) analysis to determine the RFA–ECHO probe stability. Lanes 1–7 correspond to the RFA–ECHO probe incubated with 0, 0.25, 0.5, 2, 3, 5, and 8 mM ATP, respectively. Lane 8 shows the Cy5‐modified aptamer. Similar to the nonlabeled RFA–ECHO probe, increasing ATP concentrations caused a gradual reduction in the duplex band and a concomitant increase in the free aptamer band, confirming ATP‐induced strand displacement. (d) Densitometric analysis (gray values) of duplex and single‐stranded DNA bands in each lane. (e) Fluorescence spectra of the Cy5‐labeled DNA duplex in response to varying ATP concentrations. Red curves represent the emission spectra of the reporter strand (ECHO probe), whereas blue curves represent those of the recognition strand (Cy5 aptamer). Spectra were obtained using a 25 nM RFA–ECHO probe in buffer (150 mM NaCl, 5 mM MgCl₂, 20 mM HEPES, pH 7.4) with ATP (0, 0.25, 0.5, 2.0, 3.0, 5.0 mM) at 37°C (λ _ex(ECHO) = 488 nm, λ _ex(Cy5) = 642 nm). (f) Schematic representation of TO‐to‐Cy5 FRET in the duplex. Hybridization places the donor and acceptor in close proximity, resulting in efficient quenching of TO fluorescence.

To restore the intended TO readout and suppress donor‐to‐acceptor energy transfer, we introduced a polythymidine (polyT) spacer between the 3′‐end of the aptamer and Cy5 to increase the donor–acceptor separation [22, 23]. Recognition strands containing 5, 10, 15, or 25 thymidine residues were hybridized with the reporter strand to form RFA–ECHO probes 5T, 10T, 15T, and 25T, respectively. The fluorescence intensities at 530 nm (F_D, donor) and 660 nm (F_A, acceptor) of each RFA–ECHO probe in the absence of ATP were measured (Figure 3a–e). The experimental data obeyed the Förster equation

$$\frac{F_{\mathrm{D}}}{F_{\mathrm{A}}}=\gamma {\left(\frac{r}{R_0}\right)}^6,$$

where the Förster critical radius (R ₀) for the TO–Cy5 pair was previously determined to be 5.1 nm [21]. The donor–acceptor distance (r) for each polyT linker was estimated using the worm‐like chain model [23, 24]. Regression analysis yielded a slope of γ = 1.2924 and a near‐zero intercept (0.1179) with excellent goodness of fit (R ² = 0.9991), satisfying both the sixth‐power distance law and the distance‐independence of γ (Figure 3f; see Supporting Information for details of the calculation). These results confirm that donor quenching at 0T arises predominantly from FRET. To determine the optimal spacer length, the fluorescence responses of the RFA–ECHO probes were also examined at 2 mM ATP (Figure 3a–e). Native PAGE analysis confirmed ATP‐induced duplex dissociation for all the variants (Figure S3). Short spacers (5T and 10T) partially relieved FRET and elevated the baseline fluorescence at 0 mM ATP, resulting in relatively small signal changes (ca twofold). By contrast, probes with 15T or 25T spacers produced comparable dynamic changes (ca 2.5‐fold), indicating that beyond approximately 15T, residual FRET no longer perturbs TO emission. Therefore, a 15T spacer was selected as the minimal and optimal design for subsequent experiments.

FIGURE 3.

Optimization of polyT spacer length to minimize donor–acceptor FRET in the RFA–ECHO probe. Fluorescence spectra of DNA duplexes (25 nM) containing polyT linkers of varying lengths: (a) 0 nt, (b) 5 nt, (c) 10 nt, (d) 15 nt, and (e) 25 nt. Red curves: fluorescence spectra of duplexes excited at 488 nm (ECHO channel); blue curves: spectra of duplexes excited at 642 nm (Cy5 channel) (λ _ex(ECHO) = 488 nm, λ _ex(Cy5) = 642 nm). (f) Correlation between donor–acceptor distance, estimated from the worm‐like chain model, and the experimental fluorescence ratio (F_D/F_A), showing the distance‐dependent decay of FRET efficiency with increasing spacer length.

2.2. Ratiometric Fluorescence Properties of the RFA–ECHO Probe

Fluorescence responses of the RFA–ECHO probe were characterized across a gradient of ATP concentrations (Figure 4a). As the ATP concentration increased, the TO fluorescence of the duplex progressively decreased, indicating target‐induced unwinding. In titration experiments, the Cy5/ECHO fluorescence ratio reached a plateau between 1–3 mM ATP, followed by a further increase from 3–10 mM (Figure 4b). The fluorescence plateau observed between 1–3 mM ATP is attributed to the two‐step binding mechanism of the 27‐nt ATP aptamer. Previous studies have shown that this aptamer possesses two adenosine‐binding sites with distinct affinities [25]. Binding of the first ATP molecule induces partial folding of the recognition strand and dissociation of the duplex, resulting in a linear fluorescence increase in the 0–1 mM range. Between 1–3 mM, the system is dominated by the singly bound intermediate. This intermediate is sufficiently stable to produce minimal signal change, which appears as a plateau. At higher ATP concentrations (>3 mM), the second site becomes occupied, converting the intermediate into the doubly bound state and promoting complete dissociation of the reporting and recognition strands, leading to a renewed fluorescence change. Native PAGE analysis supported this proposed mechanism (Figure S4). Such nonlinear titration profiles are commonly observed in aptamer‐based sensors and do not compromise their biological utility. Indeed, many reported ATP aptamer probes exhibit linear responses only within the low‐millimolar range before plateauing, yet have been successfully applied to intracellular ATP imaging [26, 27]. Therefore, although the RFA–ECHO probe displays a local plateau, its effective detection window (1–10 mM) still covers the physiological intracellular ATP range, ensuring practical applicability.

FIGURE 4.

Ratiometric fluorescence properties of the RFA–ECHO probe. (a) Fluorescence spectra of the RFA–ECHO probe in response to increasing ATP concentrations. Red curves: emission spectra of the reporter strand (ECHO probe); blue curves: emission spectra of the recognition strand (Cy5‐modified aptamer). The spectra of the Cy5–T15 DNA duplex (25 nM) were measured in buffer (20 mM HEPES, 150 mM NaCl, 5 mM MgCl₂, pH 7.4) containing ATP (0–10 mM) at 37°C (λ _ex(ECHO) = 488 nm, λ _ex(Cy5) = 642 nm). (b) Fluorescence spectra and ratiometric plots (FI_Cy5/FI_ECHO) of the Cy5–T15 duplex (25 nM) at varying ATP concentrations in vitro (37°C). (c) Selectivity of the probe toward ATP over other nucleoside triphosphates ([analyte] = 5 mM). (d) PAGE analysis of potential off‐target reactions. Different nucleoside triphosphates (5 mM) were added to the RFA–ECHO probe (100 nM) in buffer (20 mM HEPES, 150 mM NaCl, 5 mM MgCl₂, pH 7.4). Reaction mixtures were incubated at 37°C, analyzed with 10% native PAGE, and visualized using a Gel Doc EZ Imager (Bio‐Rad).

As shown in Figure 4c, the addition of other nucleoside triphosphates (UTP, CTP, or GTP) produced minimal fluorescence changes compared with ATP, demonstrating that the aptamer probe retains high selectivity in its duplex form. Figure 4d further supports the sensing mechanism: only in the presence of ATP does the RFA–ECHO probe unwind upon target binding, releasing the single‐stranded reporter and turning off fluorescence. Notably, the probe exhibited a similar response to ADP as to ATP, consistent with the known binding specificity of this aptamer for the adenosine (adenine) moiety rather than the phosphate tail of ATP [25]. Many classical DNA aptamers selected against adenosine‐containing ligands primarily recognize the adenine base; thus, partial binding to ADP or AMP is expected. Nevertheless, this cross‐reactivity does not compromise ATP sensing, because intracellular ATP concentrations are typically one to two orders of magnitude higher than those of ADP, AMP, and other adenine metabolites [28]. Consequently, the fluorescence readout of the RFA–ECHO probe predominantly reflects ATP dynamics, even if adenine‐based recognition is intrinsic to the aptamer. This interpretation is consistent with prior studies using the same aptamer sequence, which consistently demonstrated reliable intracellular ATP detection [10].

2.3. Ratiometric Fluorescence Detection of ATP in Living Cells

Before applying the RFA–ECHO probe in living cells, we first examined its nuclease resistance (Figure S5a). The probe was incubated in buffer containing 10% fetal bovine serum at 37°C, and aliquots were analyzed over time by PAGE. The duplex remained largely intact for several hours, although most degradation occurred by approximately 6 h. This stability was sufficient for our experiments, as the recognition and unwinding reactions were completed within 10 min (Figure S5b)—well before substantial degradation took place. The fluorescence of RFA–ECHO was measured after the solution was thoroughly mixed with ATP; therefore, we believe that the reaction timescale is primarily governed by the aptamer folding and binding kinetics upon ATP recognition. Once ATP is present, the aptamer undergoes a structural change that promotes strand displacement and subsequent signal generation. This folding process is expected to take several minutes, which is consistent with the observed reaction time. These results indicate that the duplex aptamer possesses adequate nuclease stability to recognize ATP and generate a fluorescent response within the experimental timeframe. We next assessed the biocompatibility of the RFA–ECHO probe to ensure its safe use in live‐cell assays. Cell viability was evaluated in three cell lines (MCF‐7, HeLa, and MRC‐5) following incubation with the probe at concentrations up to 50 nM for 3 h, followed by an additional 21 h culture in standard medium. No significant cytotoxicity was observed in HeLa or MRC‐5 cells, and MCF‐7 cells maintained over 90% viability (Figure S6). These results demonstrate that the RFA–ECHO probe exhibits low cytotoxicity and good biocompatibility, supporting its applicability for intracellular sensing.

To verify that the RFA–ECHO probe remains intact and functional inside cells, MCF‐7 cells were transfected separately with the Cy5‐labeled recognition strand, the TO‐labeled reporter strand, or the complete duplex probe. The cells were analyzed by fluorescence‐activated cell sorting (FACS) and confocal microscopy (Figure 5). Cells transfected with only the recognition strand displayed strong fluorescence in the Cy5 channel (λ _ex = 642 nm, λ _em = 664 nm), whereas the ECHO channel (λ _ex = 488 nm, λ _em = 525 nm) remained at baseline levels comparable to control cells. Similarly, transfection with the reporter strand alone did not increase the fluorescence in either channel, indicating negligible background from the unhybridized reporter. By contrast, cells transfected with the RFA–ECHO duplex exhibited a strong Cy5 signal—comparable to the recognition‐strand‐only group—together with a clear increase in the ECHO channel fluorescence. These findings confirm that a green fluorescence signal is generated only upon intracellular hybridization of the two strands, validating the structural integrity and functionality of the RFA–ECHO probe inside living cells.

FIGURE 5.

Intracellular validation of DNA duplex assembly and fluorescence readout in MCF‐7 cells. (a) Flow cytometry histograms for four conditions (untreated cells, ECHO reporter strand only, Cy5 recognition strand only, and RFA–ECHO duplex) recorded in both Cy5 (642/664 nm) and ECHO (488/512 nm) channels. (b) Corresponding ratiometric analysis of Cy5/ECHO signals across the four conditions. Data are averaged from three independent experiments; error bars represent standard deviations. 1: untreated, 2: reporter strand, 3: recognition strand, 4: RFA–ECHO probe. (c) Confocal microscopy images of the same four groups under four detection modes: bright field, ECHO channel (green), Cy5 channel (red), and merged images including Hoechst nuclear staining (blue). Scale bar: 20 µm.

To further validate the ratiometric design and its ability to compensate for variations in probe uptake, we investigated the “self‐calibrating” property of the RFA–ECHO probe. In this context, “self‐calibrating” refers to the probe's intrinsic ability to normalize for differences in intracellular delivery. Specifically, the Cy5 fluorophore—covalently linked to the aptamer strand—remains unaffected by ATP binding and thus serves as an internal reference, while the TO‐labeled reporter strand provides an ATP‐responsive signal. The resulting Cy5/ECHO fluorescence ratio inherently corrects for fluctuations in absolute signal intensity caused by variable probe uptake or distribution, thereby enabling accurate ATP detection across heterogeneous cell populations. To experimentally support this property, we transfected MCF‐7 cells with two different concentrations of the RFA–ECHO duplex (25 nM and 50 nM). Flow cytometry analysis showed that the absolute Cy5 fluorescence intensity in the 25 nM group decreased by approximately 50% (Figure S7a), indicating reduced cellular uptake. However, the Cy5/ECHO fluorescence ratio remained nearly constant between the two conditions (Figure S7b), demonstrating that the ratiometric output is preserved even under significant variations in probe loading. These results confirm that the RFA–ECHO probe operates as a self‐calibrating system and supports its application in quantitative ATP imaging independent of transfection efficiency.

2.4. Validation of Ratiometric ATP Imaging Under Metabolic Perturbations

To verify that the observed fluorescence changes originated from intracellular ATP fluctuations, we modulated ATP levels in MCF‐7 cells using iodoacetic acid (IAA) and Ca^{2 +}, followed by ratiometric ATP imaging with the RFA–ECHO probe. IAA is a glycolysis inhibitor that markedly depletes cellular ATP [29], whereas elevated Ca^{2 +} stimulates mitochondrial metabolism and enhances ATP synthesis [30]. As expected, ATP depletion or elevation also affected cellular uptake of the DNA–liposome sensor, resulting in different absolute Cy5 fluorescence intensities among the three conditions (control, IAA, and Ca^{2 +}) [2] (Figure 6a). Because of this uptake variability, a single‐channel ATP probe would be unreliable—cells with higher ATP tend to internalize more probe and exhibit stronger fluorescence simply due to uptake rather than actual ATP levels [31]. Indeed, the green ECHO fluorescence (reference channel) increased from low‐ATP to high‐ATP conditions, reflecting enhanced probe uptake under higher ATP. Without ratiometric correction, this effect would lead to false‐positive ATP readouts. We therefore analyzed the ratio of Cy5 (reference) to ECHO (ATP‐responsive) fluorescence, which intrinsically normalizes for uptake differences (Figure 6b). Consistent with ATP‐dependent behavior, IAA‐treated cells exhibited a significantly lower Cy5/ECHO ratio, whereas Ca^{2 +}‐treated cells showed a modestly higher ratio compared with the untreated controls.

FIGURE 6.

Intracellular validation of DNA duplex assembly and ATP‐responsive fluorescence in MCF‐7 cells. (a) Flow cytometry histograms for the three conditions (untreated, 100 µM IAA‐treated, and 5 mM Ca^{2 +}‐treated) recorded in both Cy5 (Ex/Em = 642/664 nm) and ECHO (Ex/Em = 488/512 nm) channels. (b) Corresponding ratiometric analysis of Cy5/ECHO signals across the three conditions. Data are an average of three independent experiments; error bars represent standard deviations. *p < 0.05, unpaired Student's t test (n = 3). (c) Confocal images showing ATP‐dependent changes in fluorescence after treatment: untreated (top), 100 µM IAA (middle), and 5 mM Ca^{2 +} (bottom), followed by incubation with the DNA duplex at 37°C for 3 h. Scale bar: 20 µm.

Confocal microscopy further supported these results. In ATP‐depleted cells (IAA treatment), the red Cy5 fluorescence was almost entirely colocalized with the green ECHO signal, with the Pearson correlation coefficient increasing from 0.605 in untreated cells to 0.793 under IAA treatment. This strong colocalization (Figures 6c and S8) indicates that, under low‐ATP conditions, the RFA–ECHO probe remains in its hybridized (aggregated) state, keeping Cy5 and turn‐on TO in close proximity. By contrast, in the control and Ca^{2 +}‐stimulated cells, red fluorescence was more diffusely distributed throughout the cytoplasm (while the green signal remained largely unchanged), consistent with ATP‐induced duplex unwinding of the RFA–ECHO probe and release of the Cy5‐labeled recognition strand. These results demonstrate that the RFA–ECHO probe enables accurate ratiometric ATP imaging in living cells by correcting for probe uptake and providing a reliable readout of endogenous ATP dynamics.

We noted that, in MCF‐7 cells, the ratiometric fluorescence ratio increased only modestly upon Ca^{2 +} stimulation relative to untreated conditions. This is likely because the treatment for 30 min with 5 mM Ca^{2 +} induces only a modest ATP increase [32], and because MCF‐7 cells (a cancer cell line) inherently maintain a high basal ATP level [33]. Because the RFA–ECHO probe is more responsive to lower ATP concentrations, even Ca^{2 +}‐induced ATP elevation in these ATP‐rich cells results in only subtle ratiometric changes. To validate further the performance of the sensor in a low‐ATP context, we applied the same ATP‐modulating treatments to MRC‐5 cells, a normal human fibroblast line with a lower baseline ATP content [34, 35]. In MRC‐5 cells, Ca^{2 +} stimulation caused a pronounced increase in the Cy5/ECHO ratio relative to the control, whereas IAA treatment led to a marked decrease (Figure S9). Confocal imaging corroborated these findings: after IAA exposure, nearly all red fluorescence overlapped with the green reference signal (Pearson's coefficient increasing from 0.645 to 0.768), indicating that these low‐ATP cells had been driven into an extremely ATP‐depleted state—accompanied by morphological signs of apoptosis. Conversely, Ca^{2 +}‐treated MRC‐5 cells exhibited more dispersed red fluorescence (less colocalization with green), reflecting the RFA–ECHO probe's response to ATP elevation. Collectively, these findings confirm that the RFA–ECHO probe is particularly sensitive in cells with lower ATP levels and highlight the value of its ratiometric design for accurate detection of ATP fluctuations across different metabolic states.

2.5. Discrimination of Cancerous and Normal Cells by Ratiometric ATP Imaging

We compared the ratiometric ATP readouts between cancerous and nonmalignant cell lines. Both MCF‐7 and HeLa (cervical cancer) cells exhibited substantially higher Cy5/ECHO ratios than normal MRC‐5 fibroblasts (Figure 7a), consistent with the generally elevated ATP levels characteristic of rapidly proliferating cancer cells [36]. In glycolytic tumors, increased ATP production—driven by enhanced glycolysis and nutrient uptake—promotes greater aptamer–ATP binding, thereby boosting the Cy5 signal relative to the constant ECHO reference and increasing the overall ratio. The ratiometric format inherently compensates for variations in probe uptake: any differences in total probe internalization affect both fluorescence channels equally, so the ratio reflects ATP binding rather than the absolute probe amount. This interpretation is supported by the flow cytometry data (Figure 7b), which show marked differences in Cy5 fluorescence intensities among MCF‐7, HeLa, and MRC‐5 cells, indicating cell‐type‐dependent variability in probe uptake. Because Cy5 fluorescence does not change upon ATP binding, its intensity reliably reflects the amount of internalized probe. The RFA–ECHO duplex is most likely internalized through endocytosis, a well‐established uptake pathway for DNA aptamers. By using Cy5 as an internal reference and ECHO as the ATP‐responsive reporter, the ratiometric Cy5/ECHO readout provides a self‐calibrated signal that inherently compensates for differences in probe delivery, illumination, and cell size. Thus, the Cy5/ECHO ratio provides a reliable measure of cellular metabolic status, clearly distinguishing high‐ATP cancer cells from lower‐ATP normal cells. The Cy5/ECHO fluorescence ratio in the cancer cell lines MCF‐7 and HeLa was 4.4‐fold and 2.7‐fold higher, respectively, than that in the normal fibroblast cell line MRC‐5. The HeLa data illustrate some practical considerations. Although HeLa cells also showed an elevated ratio, their absolute fluorescence intensities were lower than those of MCF‐7 cells (Figure 7b, c). The lower fluorescence intensity in HeLa cells is likely derived from differences in the cellular uptake of the RFA–ECHO probe. However, fluorescence signals observed in living cells can also be influenced by variations in extra‐ and intracellular environments, including autofluorescence, ATP metabolic heterogeneity, pH effects, and nonspecific probe binding. From Figure 5a, we can conclude that both TO and Cy5 exhibit much stronger turn‐on fluorescence than the cellular autofluorescence, indicating that autofluorescence is not responsible for the reduced signal. Furthermore, because RFA–ECHO shows very high selectivity toward adenosine (Figure 4c), nonspecific probe binding is also unlikely to account for the observed differences. This discrepancy may arise from differences in intracellular trafficking or degradation of the probe. DNA aptamers are often internalized via endocytosis and can be exposed to nucleases within endosomes or lysosomes. Although some probe degradation may occur, our results indicate that the majority of the RFA–ECHO probe remains intact long enough to report intracellular ATP levels, as evidenced by the preserved Cy5/ECHO signal. Importantly, even if partial degradation does take place, the ratiometric design minimizes its quantitative impact because both fluorescence channels would be affected proportionally.

FIGURE 7.

ATP detection by the RFA–ECHO probe in different cell lines. (a) Flow cytometry histograms of the three cell lines (MRC‐5, HeLa, and MCF‐7) recorded in both Cy5 (Ex/Em = 642/664 nm) and ECHO (Ex/Em = 488/512 nm) channels. (b) Corresponding ratiometric analysis of Cy5/ECHO signals across the three cell lines. Data are averaged from three independent experiments; error bars represent standard deviations. (c) Confocal images of the three cell lines after incubation with the DNA duplex at 37°C for 3 h. Scale bar: 20 µm.

To evaluate whether the RFA–ECHO system remains reliable under asymmetric degradation of the two dyes, we performed a control experiment simulating preferential degradation by partially replacing the fluorophore‐labeled strands with unlabeled equivalents of the same sequence. Specifically, five conditions were tested: full labeling of both Cy5 and ECHO, and partial labeling where either Cy5 or ECHO was reduced to 75% or 50%. These configurations mimic situations where one fluorophore degrades more than the other. For each condition, we measured the fluorescence spectrum under 0 mM and 2 mM ATP (Figure S10). When Cy5 was reduced to 75% or 50%, its fluorescence intensity decreased proportionally, as expected. However, this had little effect on the TO (ECHO) signal. ATP‐dependent quenching of ECHO fluorescence remained clearly visible, and the overall sensing function of the probe was preserved. We also noticed that the unwanted energy transfer from TO to Cy5 decreased with lower Cy5 levels, which further improved ECHO channel isolation. In contrast, reducing ECHO labeling to 75% still allowed ATP‐dependent signal change, but at 50% the probe largely lost its responsiveness. This is likely due to competition between labeled and unlabeled reporter strands, where the nonemissive duplexes dominate and reduce the overall fluorescence signal. The system is relatively robust to Cy5 degradation, since Cy5 serves as a passive reference. Moderate loss of ECHO is also tolerated, but more substantial degradation of ECHO compromises detection.

The RFA–ECHO probe offers several notable advantages. Thousands of experimentally validated aptamer sequences are available in public databases for a wide range of targets, meaning that new sensors can often be designed simply by selecting an appropriate aptamer—eliminating the need for complex organic synthesis required for many small‐molecule probes [37]. Conventional aptamer‐based sensors that rely on a single fluorescence output are often susceptible to artifacts arising from variations in probe concentration, cellular uptake, or microenvironmental factors [38]. By contrast, our ratiometric design incorporates an internal Cy5 reference signal that corrects for these variations, greatly improving measurement accuracy—particularly in heterogeneous cell populations. Unlike many dual‐channel sensors that depend on precise FRET pairs, which require stringent control of distance and orientation [39], our design achieves ratiometric signaling without FRET. Simply labeling the aptamer with Cy5 and hybridizing it with a TO‐labeled reporter strand generates a stable, distance‐independent dual‐color signal. This straightforward architecture simplifies probe construction and facilitates adaptation to new targets: one can attach a Cy5 label to any chosen aptamer and pair it with a complementary TO reporter strand without extensive optimization. The sensor employs commonly used fluorophores (Cy5 and TO) and can be delivered using standard transfection methods, making it fully compatible with conventional confocal microscopy and flow cytometry (FACS). Its single‐digit micromolar sensitivity, high nucleotide selectivity, and adequate serum stability were confirmed in vitro, and its live‐cell performance clearly reflects metabolic differences among cell types. Collectively, these features make the RFA–ECHO probe a robust and versatile platform for quantitative intracellular ATP imaging—and potentially for sensing other analytes through simple substitution of the aptamer sequence.

3. Conclusion

In summary, we present the RFA–ECHO probe, which effectively overcomes the key limitations of existing ATP sensors. Unlike small‐molecule “turn‐on” indicators that saturate below physiological ATP concentrations, our aptamer–reporter duplex remains responsive across the biologically relevant millimolar range. In contrast to conventional aptamer probes that lack internal calibration and are thus affected by variable cellular uptake or distribution, our design integrates a built‐in fluorescence reference, enabling a self‐correcting Cy5/ECHO ratio that quantitatively reflects ATP levels independent of delivery efficiency. Furthermore, compared with previous FRET‐based ratiometric sensors that typically exhibit modest ratio changes, the RFA–ECHO duplex produces a robust and easily discernible signal shift, allowing clear differentiation between high‐ATP (malignant) and lower‐ATP (normal) cells (e.g., MCF‐7/HeLa vs MRC‐5). The modularity, simplicity, and artifact resistance of this approach provide a practical foundation for quantitative metabolic profiling, imaging‐guided cancer diagnostics, and the development of next‐generation ratiometric aptamer sensors targeting diverse cellular metabolites.

Conflicts of Interest

The authors declare no conflict of interest.

Supporting information

The authors have cited additional references within the Supporting Information [30, 31].

Xu Y., Morihiro K., Cong R., Patil S. M., Wille I., and Okamoto A., “Self‐Calibrating Ratiometric DNA Aptamer Probe for Quantitative ATP Imaging in Living Cells.” Chemistry – A European Journal 32, no. 11 (2026): e03293. 10.1002/chem.202503293

Data Availability Statement

The data that support the findings of this study are available in the supplementary material of this article.