Design and synthesis of ultrasensitive off–on fluoride detecting fluorescence probe via autoinductive signal amplification

Abstract

We prepared an off–on fluorometric probe, DPF₁, by incorporating the concept of autoinductive signal amplification into its molecular design. In the presence of fluoride, DPF₁ undergoes a cascade of self-immolative reactions concomitant with unmasking fluorogenic coumarin, which results in the ejection of two fluoride ions. These fluoride ions are continuously activating the cascade reaction and accumulating coumarins, which leads to exponentially amplifying the signal with high sensitivity. The fluorescence signal generated by this cascade reaction is rapid, specific and insensitive to other anions. Its limit of detection was 0.5 pM, considerably lower than other current methods of fluoride detection. In addition, DCC, a long wavelength fluorometric probe, was prepared. Interestingly, an assay platform coupling DPF₁ and DCC showed an outstanding sensing ability at higher wavelengths, suggesting that this can be a promising method for the sensitive and selective detection of fluoride in biological samples. The practical applicability of the proposed approach has been demonstrated in urine and water samples.

---

1. Introduction

Most currently utilized techniques capable of assaying analytes with high sensitivity and selectivity,¹ including enzyme-linked immunosorbent assays (ELISA),² polymerase chain reaction (PCR),³ and bio-barcode assays,⁴ rely on antibodies, enzymes and biomolecules.⁵ The unique chemical structures of these biomolecules serve as recognition sites to distinguish the analyte from similar molecules, thus enhancing the specificity of these assays. Moreover, the concept of continuous signal revealing reactions, triggered by analyte–probe recognition embedded within these methods, amplifies the signal for high sensitivity. In most cases, however, the reagents used in these assays are either thermally unstable or are not stable for prolonged periods of time. Thus, practical difficulties are encountered while trying to use these reagents in analyses at ambient temperature. To circumvent this limitation and achieve high sensitivity and specificity, new approaches have been used to design autoinductive small molecular probes that mimic these biomolecules (Scheme 1a).^6–12 These autoinductive molecular probes are equipped with unique triggering groups, which selectively interact with the analyte of interest via a designated chemical reactions pathway. These reactions trigger a cascade of self-immolative reactions, which simultaneously unmask the signal molecules (chromogenic or fluorogenic) and spontaneously liberate the signal transduction molecules.¹² The latent generated signal transduction molecules continuously induce the autoinducible molecular probes to undergo self-immolative signal revealing reactions, resulting in the accumulation of signal molecules. Thus, this cyclic reaction process results in the signal amplification required for high sensitivity.¹³

Scheme 1 Pathway by which DPF₁ detects fluoride and releases two equivalents of fluoride to propagate the autoinductive signal amplification process.

Fluoride ion chemosensors are in high demand due to the importance of fluoride in a variety of healthcare and environmental contexts.^14–17 The US Environmental Protection Agency (EPA) has recommended an allowed upper limit of 2 ppm fluoride in water.¹⁸ Nevertheless, the detection of low concentrations of fluoride in polar and aqueous solutions remains challenging, without an expensive analytical equipment. Several chromogenic and fluorogenic probes, which rely on hydrogen bonds or Lewis acid coordination, have been used to detect low fluoride concentrations.^19,20 Most of these probes can only be utilized in organic solvents to detect tetrabutylammonium (TBA⁺) fluoride, but they cannot be used to detect inorganic fluoride salts.^21–23 Other fluoride sensors based on the chemical affinity between fluoride and silicon have been used to detect inorganic fluoride salts in polar solvents; however, the sensitivity of these sensors can be influenced by increasing the water concentration in the detection system.^24–28

Designing latent self-immolative ratiometric chemical probes and exploring their applications are ongoing research interests of our research group.^29–32 We recently used a quinone-methide type of rearrangement reaction to successfully design an off–on colorimetric probe, which detects fluoride; however, the limit of detection (LOD) of this probe was judged unsatisfactory by the EPA.¹⁸ To improve the LOD and sensitivity of this probe, we sought to expand its signal-revealing mechanism by incorporating an exponential signal amplification approach. Using this method, we have designed a new autoinductive off–on fluorogenic probe, DPF₁, as a ratiometric sensor for the ultrasensitive detection of inorganic fluoride (Scheme 1a). The chemical structure of the DPF₁ and schematic illustration of its autoinductive signal-revealing amplification mechanism induced by fluoride are briefly outlined in Scheme 1b. The DPF₁ probe is composed of a 4-(tert-butyldiphenylsilyloxy) trigger group attached to the latent reporter coumarin, which carries two fluoride moieties as signal amplifiers (Scheme 1a). Fluoride ions induce the removal of a silyl protecting trigger group on DPF₁, which results in a cascade of self-immolative reactions through a quinone-methide type of rearrangement reaction to generate a fluorogenic reporter, coumarin (b), along with the ejection of two additional fluorides. The latent generated fluorides then continuously induce self-immolative reactions of the unreacted DPF₁. This cyclic reaction process leads to the accumulation of fluorogenic coumarins and results in signal amplification. The fluorogenic probe, DPF₁, achieved very low LOD of 0.5 pM, lower than any other current known method used in the determination of fluoride.^14–17,33 To our knowledge, the lowest LOD for the detection of inorganic fluoride with currently known methods is in the low nanomolar ranges. A few recent studies have reported closely related examples of signal amplification for the detection of fluoride; however, they used colorimetric and absorption detection methods, and the LOD of their probes were in the low nanomolar range.^13,34 In contrast, our fluorescence probe method can detect fluoride in the upper picomolar range with a LOD of 0.5 pM, which surpasses the LODs of existing methods. The conceptual idea of Baker et al. and Perry-Feigenbaum et al. furnished us an interesting impression to design a unique fluorescence probe to enhance the signal and push the LOD from the nanomolar to picomolar level by employing fluorescence spectroscopy. In addition, an assay platform, coupling DPF₁ and DCC (a long wavelength fluorometric probe), showed an outstanding sensing ability at higher wavelengths, suggesting that this could be a promising method for the selective detection of fluoride in biological samples. Both DPF₁ and DCC are highly stable for prolonged periods when stored above 0 °C, and their syntheses involve simple and straightforward procedures.

2. Experimental section

2.1. Synthesis of 7-((4-(tert-butyldiphenylsilyloxy)benzyl)oxy)-8-formylcoumarin (DPA₁)

A solution of 4-(tert-butyldiphenylsilyloxy)benzyl chloride (a) (1 g, 2.62 mmol), 7-hydroxy-8-formylcoumarin (0.4 g, 2.1 mmol), KI (0.87 g, 5.24 mmol) and K₂CO₃ (2.9 g, 21 mmol) in dry DMF (20 mL) was stirred overnight at room temperature under an argon atmosphere (Scheme S1†).^29,33 The resulting mixture was diluted with water (150 mL). The organic layer was extracted with ethylacetate (EtOAc, 3 × 150 mL), dried with MgSO₄ and concentrated in vacuo, and then purified by column chromatography on silica gel (MeOH/toluene = 1/9), yielding the title compound (54%, 0.609 g) as a milky solid. ¹H NMR (300 MHz, CDCl₃, ppm): δ = 1.08 (s, 9H), 5.11 (s, 2H), 6.29 (d, 1H, J = 9.6), 6.76 (dd, 2H, J = 6.45, J = 1.95 Hz), 6.92 (d, 1H, J = 9 Hz), 7.16 (d, J = 8.7 Hz), 7.32–7.41 (m, 6H), 7.54 (d, 1H, J = 9 Hz), 7.60 (d, 1H, J = 9.6 Hz), 7.67–7.70 (m, 4H), 10.61 (s, 1H) (Fig. S1†). ¹³C NMR (75 MHz, CDCl₃, ppm): δ = 19.47, 26.48, 71.14, 109.60, 112.69, 113.15, 114.13, 120.01, 127.52, 127.84, 128.45, 130.00, 132.63, 133.95, 135.50, 143.01, 155.65, 155.79, 159.55, 162.59, 186.89 (Fig. S2†). MS (ESI+): calcd for [C₃₃H₃₀O₅Si + Na] = 557.17, found = 557.1; calcd for [C₃₃H₃₀O₅Si + K] = 573.14, found = 573.1 (Fig. S3†). HRMS (TOF MS AP+): calcd for [C₃₃H₃₀O₅Si+] = 534.1863, found = 534.1862.

2.2. Synthesis of 7-((4-(tert-butyldiphenylsilyloxy)benzyl)oxy)-8-(difluoromethyl)-coumarin (DPF₁)

DPA₁ (0.4 g, 0.75 mmol) solution was dissolved in dichloromethane (DCM, 15 mL) and cooled to −20 °C. N,N-Diethylaminosulfur trifluoride (DAST, 0.36 g, 2.25 mmol) was added to DPA₁, and the reaction mixture was warmed to −5 °C (Scheme S1†). The reaction mixture was stirred under these conditions for 6 h, while the reaction progress was monitored by TLC (methanol (MeOH)/toluene = 1 : 9). Upon completion of the reaction, the reaction mixture was cooled to −78 °C and quenched with 5 drops of water. The resulting residue was dried with MgSO₄ and concentrated in vacuo, and then purified by column chromatography on silica gel (MeOH/toluene = 3/97), yielding the title compound (31%, 0.13 g) as a milky yellow solid. ¹H NMR (300 MHz, (CD₃)₂CO, ppm): δ = 1.09 (s, 9H), 5.22 (s, 2H), 6.29 (d, 1H, J = 9.6 Hz), 6.82 (d, 2H, J = 8.4 Hz), 7.21 (d, 1H, J = 8.7 Hz), 7.26 (t, 1H, J = 45.3 Hz), 7.29 (d, 2H, J = 8.4 Hz), 7.39–7.50 (m, 6H), 7.73–7.79 (m, 5H), 7.93 (d, 1H, J = 9.6 Hz) (Fig. S4†). ¹³C NMR (75 MHz, CDCl₃, ppm): δ = 19.83, 26.74, 71.46, 109.84, 110.14, 110.49, 113.94, 114.32, 120.48, 128.74, 129.69, 129.87, 130.97, 133.04, 133.30, 136.20, 144.48, 154.37, 156.40, 159.63, 160.64 (Fig. S5†). MS (ESI+): calcd for [C₃₃H₃₀F₂O₄Si + Na] = 579.17, found = 579.2 (Fig. S6†). HRMS (TOF MS AP+): calcd for [C₃₃H₃₀F₂O₄Si+] = 556.1881, found = 556.1882.

2.3. Synthesis of 3-(benzothiazol-2-yl)-4-carbonitrile-7-((4-(tert-butyldiphenylsilyl oxy)benzyl)oxy)coumarin (DCC)

A solution of 4-(tert-butyldiphenylsilyloxy)benzyl chloride (a) (0.7 g, 1.84 mmol), 3-(2′-benzothiazolyl)-4-carbonitrile-7-hydroxycoumarin (c) (0.88 g, 2.76 mmol), KI (0.92 g, 5.52 mmol) and K₂CO₃ (2.54 g, 18.4 mmol) in dry DMF (15 mL) was stirred overnight at room temperature under an argon atmosphere (Scheme S2†).³² The resulting mixture was diluted with water (150 mL). The organic layer was extracted with EtOAc (3 × 150 mL), dried with MgSO₄ and concentrated in vacuo, and then purified by column chromatography on silica gel (EtOAc/toluene = 0.5/9.5), yielding the title compound (67%, 1.22 g) as an orange solid. ¹H NMR (300 MHz, CDCl₃, ppm): δ = 1.083 (s, 9H), 5.03 (s, 2H), 6.78 (d, 1H, J = 8.4 Hz), 6.93 (d, 2H, J = 2.4 Hz), 7.07 (dd, 1H, J = 9, J = 2.4 Hz), 7.15 (d, 2H, J = 84 Hz), 7.325–7.554 (m, 8H), 7.697 (dd, 4H, J = 7.8, J = 1.5 Hz), 7.97–8.02 (m, 2H), 8.214 (d, 1H, J = 8.1 Hz) (Fig. S7†). ¹³C NMR (75 MHz, CDCl₃, ppm): δ = 19.42, 26.43, 70.89, 101.76, 110.97, 113.57, 115.32, 120.03, 120.68, 121.46, 122.15, 124.12, 126.45, 126.77, 127.25, 127.81, 128.89, 129.12, 129.98, 132.56, 135.46, 137.24, 152.16, 154.42, 156.01, 156.63, 158.81, 164.02 (Fig. S8†). MS (ESI+): calcd for [C₄₀H₃₂N₂O₄SSi+]: = 664.19, found = 664.2 (Fig. S9†). HRMS (TOF MS AP+): calcd for [C₄₀H₃₂N₂O₄SSi+]: = 664.1852, found = 664.1852.

2.4. Assay conditions for the detection of fluoride using DPF₁

A stock solution of DPF₁ was prepared in acetonitrile, whereas stock solutions of all other reagents were prepared in water. In a typical assay, DPF₁ (50 μM) was incubated in an acetonitrile–pyridine–water (94 : 1 : 5 [v/v/v]; APW) solution at the respective temperature and time. The assay conditions such as the ratio between the solvents, temperature and time were optimized to get maximum fluorescence response for the detection of fluoride. Various amounts of NaF or other anions (in the case of selectivity studies) were added, and the release of 7-hydroxy-8-formylcoumarin (b) was monitored by recording fluorescence spectra at λ_ex = 360 nm and λ_em = 445 nm.

2.5. Assay conditions for the detection of fluoride using two probes approach (DPF₁ and DCC)

Stock solutions of both DPF₁ and DCC were prepared in acetonitrile, whereas stock solutions of all the other reagents were prepared in water. In a typical assay, DPF₁ (50 μM) and DCC (5 μM) were incubated in APW solution at the corresponding temperature and time. Various amounts of NaF or other anions were added, and the release of 3-(2′-benzothiazolyl)-4-carbonitrile-7-hydroxycoumarin (c) was monitored by recording fluorescence spectra at λ_ex = 500 nm and λ_em = 595 nm.

3. Results and discussion

DPF₁ was prepared in two sequential steps, with the coupling of two known synthons, 4-(tert-butyldiphenylsilyloxy)benzyl chloride and 7-hydroxy-8-formylcoumarin, through a simple SN₂ reaction to yield 7-((4-(tert-butyldiphenylsilyloxy)benzyl)oxy)-8-formylcoumarin (DPA₁).^29,33 Fluorides were transferred to DPA₁ by treatment with (diethylamino)sulfur trifluoride (DAST), yielding DPF₁ (Scheme S1†). Interestingly, fluoride transfer was successful in the presence of 4-tert-butyldiphenylsilyl functionality at a low temperature, without interference from the special chemical affinity between fluoride and silicon, which would have resulted in the self-immolative disassembly of DPF₁. The overall yield of the two steps was 16%. The chemical structures of the synthetic intermediates and the final products were determined by ¹H and ¹³C NMR and by mass spectrometry (ESI†).

3.1. Fluoride detection at DPF₁

The sensing ability of the DPF₁ was determined by recording its fluorescence spectra (λ_ex = 360 nm, λ_em = 445 nm) at different fluoride concentrations. In each analysis, DPF₁ was incubated in an APW solution for 1 h at 60 °C (Fig. 1a). In the absence of fluoride, the optical switch, DPF₁, produced little fluorescence (Fig. 1a). The introduction of 0.5 pM fluoride, however, resulted in highly enhanced fluorescence, corresponding to the emission spectrum of free coumarin (b).³³ Fluorescence intensity was increased as the fluoride concentration increased from 0.5 pM to 50 μM (Fig. 1a). A plot between logarithms of fluorescence intensity versus logarithm of fluoride concentration exhibited a linear relationship in a wide linear concentration range from 0.5 pM to 50 μM (Fig. 1b). The LOD of this probe was sufficiently sensitive to detect the EPA mandated upper limit of fluoride concentration (2 ppm or 106 μM) in drinking water.

Fig. 1 (a) Fluorescence emission changes (λ_ex = 360 nm) of DPF₁ (50 μM) upon incubation with fluoride (a = 0, b = 0.5, c = 5, d = 5 × 10¹, e = 5 × 10², f = 5 × 10³, g = 5 × 10⁴, h = 5 × 10⁵, i = 5 × 10⁶ and j = 5 × 10⁷ pM) in APW solution for 1 h at 60 °C. (b) A log–log calibration curve of the reaction of DPF₁ with fluoride. (c) Kinetic analysis of fluorescence emission following the reaction of DPF₁ (50 μM) with 0 to 50 μM fluoride in APW solution at 40 °C via autoinductive signal amplification.

3.2. Kinetic analysis for the fluoride detection at DPF₁

Kinetic analysis showed that the reaction of the fluoride with DPF₁ (50 μM) in APW solution at 40 °C was characteristic of an exponential progress of disassembly (Fig. 1c). The relationship between percent fluorescence intensity ([I_exp/I_max] × 100) of the released reporter b and time is expressed as a sigmoidal curve for each concentration of fluoride, which confirmed the exponential amplification of the signal expected for an autoinductive process.^7,35 Here, I_exp is the fluorescence signal obtained at a particular time for each concentration of fluoride, whereas I_max is the maximum fluorescence signal following complete exposure to DPF₁. In the presence of 0.5 pM fluoride, reporter b begins to liberate from DPF₁ within 30 min and reaches a maximum fluorescence signal at 700 min, which indicates the complete disassembly of DPF₁ at 700 min. In the absence of fluoride, however, DPF₁ does not emit fluorescence (Fig. 1c). This was not due to the instability of DPF₁, since the latter remained stable even after incubation for 35 h (data not shown). In the presence of low concentrations of fluoride, the probe requires long time for the complete disassembly. Thus, the maximum autoinductive amplification process shows slower kinetics at lower concentrations.

3.3. Selectivity

The selectivity of DPF₁ to the fluoride ions was investigated by testing several other common anions, including Cl⁻, Br⁻, I⁻, N₃⁻, CH₃COO⁻, HSO₄⁻, NO₃⁻ and H₂PO₄⁻. Of these ions, only fluoride elicited a strong fluorescence signal from DPF₁, whereas all the other anions were unable to produce significant signals above background (Fig. 2). These results reveal that the proposed system is highly specific for fluoride over other anions. Moreover, our results are similar to those of other fluoride probes, which utilize specific fluoride-induced silyl deprotection to uncloak masked fluorogenic probes.^24–28

Fig. 2 Fluorescence emissions of DPF₁ upon incubation with fluoride (0.5 pM, 50 μM) and other anions (1 equivalent) in APW solution for 1 h at 60 °C. The control sample refers to the absence of any anions.

3.4. The role of autoinductive signal amplification

To investigate the role of the autoinductive signal amplification mechanism embedded within DPF₁, aiming to improve its LOD, we prepared an analogous DPF₁, which lacked the ability to undergo the autoinductive amplification reaction by substituting the geminal difluoride coumarin with the long fluorogenic molecule, 3-(2′-benzothiazolyl)-4-carbonitrile-7-hydroxycoumarin (DCC) (Scheme 2). The fluoride-induced signal revealing mechanism of DCC is the same as that of DPF₁. DCC revealed the fluorescence signals after incubation at 60 °C for 5 h. However, the LOD of this system to detect fluoride was found to be 238 nM, consistent with other fluoride detecting latent fluorescent probes (i.e., a LOD in the upper nanomolar range).²⁵ Interestingly, the LOD of DPF₁ towards fluoride detection was at least 10⁵ fold more sensitive than DCC, which revealed the crucial role of the autoinductive amplification mechanism in improving the LOD from nanomolar to picomolar concentrations. Therefore, incorporating autoinductive signal amplification into the design of an analyte detecting platform can boost the sensitivity of the assay system for determining very low concentrations of analyte.

Scheme 2 A long-wave fluorogenic probe (DCC) that uses an autoinductive fluoride signal amplifier (DPF₁) to detect fluoride.

3.5. The role of geminal fluorides as signal transduction agents

We also sought to demonstrate that the two geminal fluorides in DPF₁, released after the initiation of the self-immolative reaction, are effective signal transduction agents. Co-incubation of DPF₁ with DCC in the presence of fluoride should unlock coumarin, the moiety with long wavelength fluorescence in DCC (Scheme 2). The fluorescence spectrum of the signal transduction system for DPF₁ (50 μM) coupled with DCC (5 μM) was tested by the addition of fluoride in an APW solution; we deliberately designed this assay platform such that DPF₁ was present at a 10-fold higher concentration than DCC. Therefore, the initial fluoride would more likely interact with DPF₁ rather than with DCC. Fig. 3(a) presents the fluorescence spectra of the two-probe system (5 μM DCC and 50 μM DPF₁) in the presence and absence of fluoride at different λ_ex and λ_em. As shown in Fig. 3a (left), the fluorescence spectrum (λ_ex = 360 nm, λ_em = 445 nm) was characterized by the strong fluorescence signal of coumarin (b) released from DPF₁ in the presence of fluoride (50 pM) after incubation for 1 h at 60 °C. These findings suggested that the geminal fluorides in DPF₁ were released into the medium as two free fluoride ions. Upon approaching DCC, these fluorides could induce the removal of silyl ether protecting groups, initiating a cascade of self-immolative reactions to eject the long-wavelength coumarin (c) (λ_ex = 500 nm, λ_em = 595 nm) (Fig. 3a, right).

Fig. 3 Changes in fluorescence emission through λ_ex = 360 nm (left) and λ_ex = 500 nm (right) of two probes, DCC (5 μM) and DPF₁ (50 μM), in the presence of fluoride (50 pM) at 1 h (left) and 2 h (right) in APW solution at 60 °C. (b) Kinetic analysis of fluorescence emissions (λ_ex = 500 nm) of DPF₁ (50 μM), DCC (5 μM), and a two-probe system, consisting of DCC (5 μM) and DPF₁ (50 μM), in the presence of low concentrations of fluoride (0.5 pM and 500 nM) in APW solution at 40 °C.

3.6. Fluoride detection using the two probes approach (DPF₁ and DCC)

Kinetic analysis indicated that neither DPF₁ nor DCC alone was able to emit fluorescence (λ_ex, 500 nm) in the presence of a low concentration (0.5 pM) of fluoride, even after incubation for 300 min (Fig. 3b). In contrast, the incubation of both probes in APW solution at 40 °C with 0.5 pM and 50 pM fluoride increased the fluorescence emission spectra within 50 and 90 min, respectively (Fig. 3b). Only when the concentration of DPF₁ exceeded that of DCC, we were able to detect these low concentrations of fluoride with the two-probe coupling assay method. Furthermore, the LOD of DCC for the fluoride detection was 238 nM, but the LOD was 10⁵ fold lowered when DCC is coupled with DPF₁, indicating the outstanding sensitivity of this two probes method. These results reveal that the two geminal fluorides in DPF₁, available after activation, are effective signal transduction agents that can remove other silyl ether trigger probes.

Fluorescence emission spectra (λ_ex = 500 nm) of the two-probe system [DCC (5 μM) and DPF₁ (50 μM)] were recorded in the presence of various concentrations of fluoride, following incubation in APW solution for 2 h at 60 °C (Fig. 4a). Fluorescence increased with the fluoride concentration (Fig. 4a). A plot between logarithms of fluorescence intensity versus logarithm of fluoride exhibited a linear relationship over a wide concentration range, from 0.5 pM to 50 μM (Fig. 4b). Kinetic analysis of the two-probe system [DCC (5 μM), DPF₁ (50 μM)] in the presence of various concentrations of fluoride showed an exponential progression of disassembly, similar to that of DPF₁ (Fig. 4c). Notably, on comparing Fig. 1c and 4c, a delay is observed in the disassembly of DCC in the two-probe coupling assay, whereas rapid signal unmasking was observed for DPF₁. The time required for the complete disassembly of DCC in the two-probe system was nearly twice that of DPF₁ at the same fluoride concentration, with the delay in the former likely due to the concentration differences in DPF₁ and DCC. The initially ejected latent fluorides from DPF₁ would be more liable to react with unreacted DPF₁ than with DCC due to the 10-fold difference in their concentrations. As the reaction progresses, most of the unreacted DPF₁ becomes depleted, resulting in the increase of free fluoride ions. These ions can catalyze the self-immolative reaction of DCC, unmasking long wavelength fluorogenic molecules.

Fig. 4 (a) Fluorescence emission changes (λ_ex = 500 nm) of a two-probe system, consisting of DCC (5 μM) and DPF₁ (50 μM), in the presence of various concentrations of fluoride (a = 0, b = 0.5, c = 5, d = 5 × 10¹, e = 5 × 10², f = 5 × 10³, g = 5 × 10⁴, h = 5 × 10⁵, i = 5 × 10⁶ and j = 5 × 10⁷ pM) in APW solution after 2 h at 60 °C. (b) log–log calibration curve of the reaction of the two-probe system, as mentioned above, with various concentrations of fluoride. (c) Kinetic analysis of fluorescence emission following the reaction of the two-probe system with various concentrations of fluoride in APW solution at 40 °C.

3.7. Advantages of the proposed approach over other methods

The LOD of our DPF₁ fluoride assay platform was considerably better than the LODs of existing methods of fluoride detection.^21–29 In addition, the synthesis of DPF₁ is easy and straightforward. Moreover, we are able to extend this fluoride detection platform to incorporate a long wavelength fluorescence probe, DCC. Numerous biological samples show some fluorescence of their own, typically in the blue region of the spectrum. As this would interfere with the measurement of fluoride fluorescence, it is desirable to enhance the sensitivity of the detection using marker dyes that fluoresce in a low-energy region (≥600 nm) of the electromagnetic spectrum.³⁶ Our two-probe coupled amplification strategy resulted in a long wavelength reporter with sensitivity in the sub-picomolar range with a detection time as short as 2.5 h. Our findings indicate that this method will constitute a novel and advanced platform for fluoride analysis in biological samples.

3.8. Real sample analysis and repeatability studies

We have demonstrated the real sample analysis of the proposed sensor (two probes approach) towards the determination of fluoride present in human urine samples and water samples (tap, rain and pond). The urine sample was collected from a healthy man, filtered with Whatman filter paper and diluted to the ratio of 1 : 50 with the addition of APW solution. The spiked fluoride concentrations are 5 nM and 5 μM. The found and recovery values are given in Table 1. The acceptable recoveries obtained for the water and urine samples reveal the promising practical feasibilty of the sensor. Moreover, the sensor has offered appreciable repeatability towards the determination of 5 nM fluoride with an R.S.D of 3.36% for five repeated measurements.

Table 1 Determination of fluoride present in various water and urine samples using the two probes approach (DPF₁ and DCC)

Samples	Added	Found	Recovery/%	RSD^a/%
a Relative standard deviation of three individual measurements.
Tap water	5 nM	4.81 nM	96.2	2.83
	5 μM	4.85 μM	97.0	3.57
Rain water	5 nM	4.80 nM	96.0	2.33
	5 μM	4.91 μM	98.2	3.28
Pond water	5 nM	4.79 nM	95.8	2.80
	5 μM	4.90 μM	98.0	2.88
Urine sample	5 nM	4.92 nM	98.4	2.25
	5 μM	4.90 μM	98.0	2.47

---

4. Conclusions

In summary, we have successfully implemented a quinone-methide type rearrangement reaction as an off–on fluorometric switch and incorporated the concept of signal amplification into the design to prepare an ultrasensitive latent fluorogenic probe, DPF₁, for the sensitive detection of fluoride. DPF₁ in the presence of fluoride undergoes a cascade of self-immolative reactions with concomitant ejection of fluorogenic coumarin and two additional fluorides, leading to a continuous signal revealing process to achieve signal amplification with high sensitivity. The LOD of this probe surpasses the LODs of existing methods for fluoride detection. The fluorescence signal generated by this tandem reaction is highly specific and insensitive to other anions. Furthermore, DPF₁ coupled with the long wavelength DCC probe can act as a sensitive fluorometric indicator to quantitatively measure fluoride in long wavelength spectra, thus avoiding any interference by biological samples. The practical applicability of the proposed approach has been demonstrated in water samples with appreciable recoveries. The assay platform coupling DPF₁ and DCC should be applicable to measure fluoride in biological samples. In future, this assay system may be used to construct fiber-optic sensors.

Acknowledgements

This work was supported by the Ministry of Science and Technology, Taiwan (NSC 103-2811-M-027-002 and 102-2113-M-027-002-MY3).

References

1. D. A. Giljohann and C. A. Mirkin, Nature, 2009, 462, 461–464.
2. J. P. Gosling, Clin. Chem., 1990, 36, 1408–1427.
3. S. GreenER, Nat. Methods, 2007, 4, 869.
4. J.-M. Nam, C. S. Thaxton and C. A. Mirkin, Science, 2003, 301, 1884–1886.
5. P. Scrimin and L. J. Prins, Chem. Soc. Rev., 2011, 40, 4488–4505.
6. N. Karton-Lifshin and D. Shabat, New J. Chem., 2012, 36, 386–393.
7. M. S. Baker and S. T. Phillips, J. Am. Chem. Soc., 2011, 133, 5170–5173.
8. E. Sella, A. Lubelski, J. Klafter and D. Shabat, J. Am. Chem. Soc., 2010, 132, 3945–3952.
9. E. Sella, R. Weinstain, R. Erez, N. Z. Burns, P. S. Baran and D. Shabat, Chem. Commun., 2010, 46, 6575–6577.
10. M. Avital-Shmilovici and D. Shabat, Bioorg. Med. Chem., 2010, 18, 3643–3647.
11. E. Sella and D. Shabat, J. Am. Chem. Soc., 2009, 131, 9934–9936.
12. L. Zhu and E. V. Anslyn, Angew. Chem., Int. Ed., 2006, 45, 1190–1196.
13. M. S. Baker and S. T. Phillips, Org. Biomol. Chem., 2012, 10, 3595–3599.
14. Y. Wu, X. Peng, J. Fan, S. Gao, M. Tian, J. Zhao and S. Sun, J. Org. Chem., 2007, 72, 62–70.
15. H. Matsui, M. Morimoto, K. Horimoto and Y. Nishimura, Toxicol. In Vitro, 2007, 21, 1113–1120.
16. P. Singh, M. Barjatiya, S. Dhing, R. Bhatnagar, S. Kothari and V. Dhar, Urol. Res., 2001, 29, 238–244.
17. M.-L. Cittanova, B. Lelongt, M.-C. Verpont, M. Geniteau-Legendre, F. Wahbe, D. Prie, P. Coriat and P. M. Ronco, Anesthesiology, 1996, 84, 428–435.
18. U. EPA, Total Coliforms (Including Fecal Coliforms and E. Coli), 2009, 54.
19. Y. Zhou, J. F. Zhang and J. Yoon, Chem. Rev., 2014, 114, 5511–5571.
20. R. Martínez-Máñez and F. Sancenón, Chem. Rev., 2003, 103, 4419–4476.
21. S. Rivadehi, E. F. Reid, C. F. Hogan, S. V. Bhosale and S. J. Langford, Org. Biomol. Chem., 2012, 10, 705–709.
22. Y. Qu, J. Hua and H. Tian, Org. Lett., 2010, 12, 3320–3323.
23. M. Cametti and K. Rissanen, Chem. Commun., 2009, 2809–2829.
24. J. Du, M. Hu, J. Fan and X. Peng, Chem. Soc. Rev., 2012, 41, 4511–4535.
25. S. Y. Kim and J.-I. Hong, Org. Lett., 2007, 9, 3109–3112.
26. S. Elsayed, A. Agostini, L. E. Santos-Figueroa, R. Martínez-Máñez and F. Sancenón, ChemistryOpen, 2013, 2, 58–62.
27. R. Hu, J. Feng, D. Hu, S. Wang, S. Li, Y. Li and G. Yang, Angew. Chem., Int. Ed., 2010, 49, 4915–4918.
28. S. Y. Kim, J. Park, M. Koh, S. B. Park and J.-I. Hong, Chem. Commun., 2009, 4735–4737.
29. J.-A. Gu, Y.-J. Lin, Y.-M. Chia, H.-Y. Lin and S.-T. Huang, Microchim. Acta, 2013, 180, 801–806.
30. H.-C. Huang, K.-L. Wang, S.-T. Huang, H.-Y. Lin and C.-M. Lin, Biosens. Bioelectron., 2011, 26, 3511–3516.
31. S.-T. Huang, C.-J. Teng, Y.-H. Lee, J.-Y. Wu, K.-L. Wang and C.-M. Lin, Anal. Chem., 2010, 82, 7329–7334.
32. S.-T. Huang, K.-N. Ting and K.-L. Wang, Anal. Chim. Acta, 2008, 620, 120–126.
33. D. H. Kwan, H. M. Chen, K. Ratananikom, S. M. Hancock, Y. Watanabe, P. T. Kongsaeree, A. L. Samuels and S. G. Withers, Angew. Chem., 2011, 123, 314–317.
34. R. Perry-Feigenbaum, E. Sella and D. Shabat, Chem.–Eur. J., 2011, 17, 12123–12128.
35. M. H. Todd, Chem. Soc. Rev., 2002, 31, 211–222.
36. A. Gómez-Hens and M. Aguilar-Caballos, TrAC, Trends Anal. Chem., 2004, 23, 127–136.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c4an01723d

---