Design principles of chemiluminescence (CL) chemodosimeter for self-signaling detection: luminol protective approach

Abstract

Chemiluminescence (CL) sensors can provide convenience and high sensitivity because they do not require an external excitation light source to produce a fluorescence signal. However, most CL based detection systems do not have a built-in self-signaling process, leading to inefficient and complex protocols due to the required multistep cascade reactions. Here, we develop a CL based sensory system with a built-in self-signaling feature by adapting the chemodosimeter concept. We found that a masking group incorporated to luminol efficiently suppresses the CL of luminol and that selective removal of the masking group by a target analyte can turn on the CL process, generating a sensitive fluorescence turn-on signal. Through systematic studies on newly devised TBS-luminol and TIPS-luminol, we optimized the molecular design parameters to achieve a highly sensitive and selective CL chemodosimeter. The optimized conditions rendered highly sensitive (Limit of Detection (LOD) = 18 nM) and selective fluoride sensing in aqueous environments. We anticipate that our new sensor system offers an efficient way to achieve highly sensitive, selective, and convenient CL turn-on detection of various important analytes.

---

Introduction

Chemiluminescence (CL) is the emission of light as a result of a chemical reaction.¹ Since an excitation light source is not required for sample radiation in CL, common problems associated with photoluminescence (PL) such as light scattering, light source instability, and photo-bleaching are absent in CL. Possible high background signals from non-specific photo-excitation are also excluded and thus CL transducers can provide extremely high sensitivity. In this context, CL has been applied broadly, serving in immunoassay, non-immunoassay diagnostic, DNA sequencing, combustion analysis, monitoring techniques, and biosensors as well as in chemosensor developments.^2–13

Self-signaling sensors allow for convenient and effective detection of target analytes.^14–16 Unfortunately however, most CL based detection systems reported so far do not have a built-in self-signaling feature because CL molecules such as luminol and luciferin only play a role as a signal transducer. Therefore, they usually require another crucial component as a part of the receptor including (1) metallic or semiconducting nanoparticles as labels of oligonucleotide probes^17–20 or graphene,²¹ (2) a hemin/G-quadruplex probe,^22,23 or (3) a fluorescent probe or surface functionalized quantum dots which participate in chemiluminescence resonance energy transfer (CRET) with the CL molecule.^24–26 Although these CL sensor systems attained very high sensitivity down to picomole levels because of the merits of CL techniques, these methods lead to inefficient and complex protocols with high cost due to the required multistep cascade reactions.

Here, we present a CL based sensor with a built-in self-signaling feature by adapting the chemodosimeter concept (Scheme 1).²⁷ Based on the analysis of the CL mechanism of luminol, we designed a CL chemodosimeter system with a proper masking group incorporated into luminol which efficiently suppresses its CL of luminol.^28,29 Selective removal of the masking group by a target analyte can subsequently turn on the CL process. Similar masking strategies by means of acetyl group or 2-nitrobenzyl group have been reported and the latter case was used to develop light sensitive CL generation from luminol.^28,29 As a target analyte, we chose fluoride anion (F⁻). Fluoride anions serve an important role in a wide range of chemical, biological, and medical processes, such as water supply treatment, dental care, osteoporosis treatment, and even in chemical warfare agents.³⁰ Thus, the detection of F⁻ has been attracting much interest. Although a number of 1,2-dioxetane based chemiluminophores for F⁻ detection have been designed to date,^31–33 luminol based sensor has not been developed.

Scheme 1 Our CL chemodosimeter design and its CL mechanism are illustrated. TBS or TIPS group attached to the oxygen atom of luminol could act as a blocker to prevent the CL process. Chemical transformation to the monoanion adduct by fluoride anion revitalizes the CL process.

In this contribution, TBS-luminol (1) and TIPS-luminol (2) were newly developed as CL chemodosimeters for detecting F⁻. Through thorough studies on CL properties of TBS-luminol and TIPS-luminol, we established an optimized molecular design to achieve a highly sensitive and selective CL chemodosimeter. The optimized conditions rendered highly sensitive (LOD = 18 nM) and selective fluoride sensing in aqueous environments. Furthermore, we contributed to establishing selective synthesis of the silyl substituted luminol. Since luminol bears multiple reactive functional groups in a single molecule, the selective substitution of luminol is very difficult and the origin of selectivity is not well understood. The unambiguous characterization of the structure of TBS-luminol via NMR analyses and the density functional theory (DFT) calculations for the reaction mechanism allows for clear understanding of the origin of selectivity in luminol substitution reaction.

Results and discussion

Luminol is a representative of an important class of organic CL molecules.^34,35 Accordingly this particular molecule has been thoroughly investigated, illuminating the details of its CL property and CL conditions as well as its CL mechanism.^4,36 The CL mechanism of luminol generally accepted is illustrated in Scheme 1.³⁶ α-Hydroxyperoxide anion (4) is considered as a key intermediate for luminol CL, which can be obtained by the oxidation of the monoanion intermediate (3). We envisioned that a proper masking group incorporated to one side of luminol could prevent the formation of the monoanion intermediate that leads to α-hydroxyperoxide anion, eventually shutting down the luminol CL. As in the case for TBS-luminol, the tert-butyldimethylsilyl (TBS) group attached to the oxygen atom of luminol could act as a blocker to prevent the CL process, and its chemical transformation to the monoanion adduct by adding a fluoride anion (F⁻) could revitalize the CL process.

The target compound was synthesized by TBS protection of luminol. The reaction using tert-butyldimethylsilyl chloride (TBSCl) and pyridine gave TBS-luminol as a single product among the listed possible reaction products in Scheme 2. Detailed structure elucidation of the resulting product was performed by the combined analysis of 1D and 2D NMR. We could rule out the possible formation of (3)-N-silylation through careful comparison of ¹H-NMR data of both luminol and the resulting product (Fig. S1 in the ESI†); in the ¹H-NMR spectrum of the product registered in DMSO-d₆, the integration value of protons at δ 7.35 assigned to H-1 is still 2, but the integration of the proton at δ 11.35 assigned to H-5 is 1. The regioselectivity of the reaction was established by two dimensional heteronuclear multiple bond correlation (2D HMBC) NMR analysis of the resulting product (Fig. 1b). The HMBC correlations of the proton at δ 11.35 with carbons C-6, C-7, and C-8 confirmed that this proton was located at position N-1; if the proton is placed on N-2, correlation of the proton at δ 11.35 with C-5, C-6, and C-7 should be observed in the HMBC spectrum. Finally, we determined whether the resulting product is O-silylated or N-silylated product via one dimensional selective nuclear overhauser effect spectroscopy (1D selective NOESY) NMR analysis (Fig. 1c). While we did not observe any protons which correlate with H-5, the protons located on tert-butyl group (H-6) showed strong correlation with H-4 and protons placed on methyl group (H-7). This result clearly indicates that the resulting product is O-alkylated product. The origin of reaction selectivity was well rationalized by DFT (B3LYP) calculation. Luminol can exist in four different keto-enol tautomeric forms (k,k-, k,e-, e,k- and e,e-luminol). In the gas phase, the k,e-tautomer was shown to be the most stable form by DFT (B3LYP) calculations,³⁷ which is also the case for the BH and HLYP calculations performed here. Optimization in water within the polarizable continuum model (PCM) scheme however yields a significant stabilization of the k,k-form (Fig. 1a) against k,e-form in terms of Gibbs free enthalpies ΔG (Fig. 1a). The possible intermediate states (IM1, IM2), created by proton abstraction through the presence of pyridine, can be both formed from the k,k-tautomer. IM1 (i.e. proton abstraction in the N-1 position), is thermodynamically only little below IM2 (proton abstraction in the N-2), ΔG = −0.2 kJ mol⁻¹, and can thus be easily thermally populated, so that TBS substitution at the O-1 position can occur. In fact, from theory, O-1 substitution is largely favored against O-2 (−15.4 kJ mol⁻¹) and N-1, N-2 (−49.3 kJ mol⁻¹), confirming the experimental results.

Scheme 2 Reaction condition and possible reaction products for the synthesis of TBS-luminol are described. (2)-O-silylation indicated with square box was selectively obtained in this condition.

Fig. 1 (a) The origin of selective synthesis of TBS-luminol was explained by DFT calculation. ΔG values were calculated for the starting material, possible reaction intermediates and products. (b) 2D HMBC NMR spectrum, and (c) 1D selective NOE spectra of TBS-luminol.

We investigated its CL properties of TBS-luminol to confirm whether TBS protection indeed suppresses CL of luminol. We used a luminol–H₂O₂ based CL system and chose cobalt as a catalyst for hydrogen peroxide decomposition.³⁸ The reaction conditions and injection sequence were optimized for the TBS-luminol and luminol such as follows: Addition of tetrahydrofuran (THF) solution of TBS-luminol or luminol (1.7 × 10⁻³ M, 200 μl) to a mixed water solution of Co(NO₃)₂ (5 × 10⁻⁴ M, 400 μl), H₂O₂ (2 × 10⁻² M, 400 μl) and Na₂CO₃/NaHCO₃ (1.0 × 10⁻¹ M, 400 μl). As expected, TBS protection resulted in a reduction of CL intensity (Fig. 2a). However, the reduction (CL_TBS-luminol/CL_Luminol = 0.21) was considerably lower than expected. If all luminol molecules are well protected by the TBS group under these reaction conditions, the CL intensity should be close to zero due to the suppression of the monoanion intermediate formation by TBS protection, shutting down the CL process. We attributed this non-zero background signal to the instability of the TBS protection group in basic conditions. The comparison of the spectral characteristics of the CL emission (Fig. 2b) and the CL kinetics (Fig. 2a) for TBS-luminol and luminol support our reasoning. The spectral characteristics of the CL emission for TBS-luminol is the same as luminol with a maximum at 420 nm (Fig. 2b), indicating that emission originates from the same emitting species, i.e. the monoanion intermediate. However their CL kinetics analyzed within the first minutes were distinctively different (Fig. 2a). While luminol shows a rather prompt CL emission and reaches the maximum emission intensity after 12 seconds of the injection, TBS-luminol displays a long rise towards the maximum, i.e. 40 second after the injection. Apparently, TBS-luminol converts to unprotected luminol with a considerable rate, presumably by Si–O bond cleavage in the presence of base; in fact, it is known that TBS protection of enol groups is rather labile under basic conditions.³⁹

Fig. 2 CL intensity curves of luminol (black line), TBS-luminol (blue line), and TBAF treated TBS-luminol (orange line) were plotted versus (a) time and (b) wavelength after they were introduced. CL intensity curves of luminol (balck line), TIPS-luminol (green line), and TBAF treated TIPS-luminol (purple line) were plotted versus (c) time and (d) wavelength. For (a) and (c), the CL at 420 nm was monitored against time.

To obtain high sensitivity, background signal should be minimized. However, unfortunately basic conditions cannot be avoided, since the type and amount of base play a crucial role in the luminol CL process, having a strong influence on the kinetics of the monoanion generation, and the formation of hydroxyl radicals and superoxide anions. In fact, the pH value should be higher than the pK_a value of α-hydroxyperoxide (pK_a = 11–12), otherwise a dark reaction can occur instead of the CL emission process.³⁶ Indeed, reduction in the amount of base is accompanied with a considerable decrease in CL intensity of TBS-luminol as well as luminol (Fig. S2†), affecting both in a similar way. Since the CL_TBS-lumol/CL_Luminol ratio cannot be further reduced in that way, we decided to modify the molecular structure in such a way to make it more robust against undesirable bond cleavage by base.

A variety of silicon-based protecting groups have been developed in the past including trimethylsilyl (TMS), triethylsilyl (TES), tert-butyldimethylsilyl (TBS), triisopropylsilyl (TIPS), and tert-butyldiphenylsilyl (TBDPS), showing different reactivity and/or stability under basic conditions. Wuts et al. studied the stability of various silyl ethers and reported that the TIPS group is the most robust under basic environments,⁴⁰ being 5 times more stable than the TBS group. Therefore, we selected the TIPS moiety as a new masking group instead of the TBS group. TIPS-luminol was easily prepared through the same procedure as TBS-luminol synthesis. ¹H-NMR and ¹³C-NMR of TIPS-luminol look essentially the same as those of TBS-luminol. The measurement of CL intensity of TIPS-luminol was conducted in the same condition as TBS-luminol. As expected, TIPS protection almost completely blocked the CL process, i.e. essentially no emission was observed (Fig. 2c and d and S3†). These results indicate that changing the masking group greatly enhances the stability of the protected luminol under basic conditions and that this would increase the sensor property, i.e. the sensitivity by minimizing false alarms (Fig. S4†).

We further evaluated the performance of TBS-luminol and TIPS-luminol as a sensor of F⁻. We used tetrabutylammonium fluoride (TBAF, 1 × 10⁻¹ M, 100 μl) as a F⁻ source and added TBAF into the THF solution of TBS-luminol (3.4 × 10⁻³ M, 100 μl) or TIPS-luminol (3.4 × 10⁻³ M, 100 μl). Subsequent exposure of the resulting solution to the mixed water solution of Co(NO₃)₂ (5 × 10⁻⁴ M, 400 μl), H₂O₂ (2 × 10⁻² M, 400 μl) and Na₂CO₃/NaHCO₃ (1.0 × 10⁻¹ M, 400 μl) produced strong blue CL emission, which was clearly observable by naked eye in both cases (Fig. 2c inset and Movie 1†). Time integrated CL intensity of the F⁻ treated samples is 4.7 times higher for TBS-luminol and 110 times higher for TIPS-luminol than the CL of untreated samples, which is consistent with our visual observation (Fig. 2). The CL spectra and time-resolved curves of the F⁻ treated TBS-luminol and TIPS-luminol are identical to those of luminol (Fig. 2), implying that the monoanions are successfully formed by the cleavage of Si–O bonds by F⁻, leading to the revitalization of the CL processes.

The sensing mechanism of TIPS-luminol was further confirmed through the combined analysis of UV-Vis spectroscopy and ¹H-NMR (note: limited here to TIPS-luminol, since we had showed that TBS-luminol had exactly same sensing mechanism). The deprotection of TIPS-luminol with F⁻ was verified by UV-Vis. The absorption spectrum of TIPS-luminol in THF (1.1 × 10⁻⁵ M) exhibited strong peaks at 295 nm and 356 nm, closely resembling those of luminol (Fig. 3a). Addition of F⁻ (THF, 3.3 × 10⁻⁴ M) resulted in a notable change from colorless to light yellow, which is associated with a decrease in the absorption band at 295 nm and 357 nm and simultaneous appearance of new bands at 329 nm and 370 nm (see Fig. 3a, purple line). The observations are consistent with the formation of the monoanion species upon cleavage of the TIPS group by F⁻. The 1 : 1 stoichiometric TIPS deprotection was confirmed in the ¹H-NMR spectra (Fig. 3b). For TIPS-luminol, two peaks at δ = 1.36 and 1.08 were observed, assigned to the TIPS group. Adding one mole equivalent of F⁻, the peaks shifted about δ = 0.06 downfield, attributed to the stoichiometric formation of triisopropylsilyl fluoride (TIPSF).

Fig. 3 (a) UV-Vis spectra of THF solution of luminol (1.1 × 10⁻⁵ M, black line), TIPS-luminol (1.1 × 10⁻⁵ M, green line), and TBAF treated TIPS-luminol (purple line) and (b) upfield ¹H-NMR spectra of TIPS-luminol (upper) and 1.0 equiv. of TBAF treated TIPS-luminol (lower) registered in DMSO-d₆.

We explored the sensitivity of TBS-luminol and TIPS-luminol under optimized experimental condition, monitoring the CL intensity versus the amount of F⁻ (Fig. S5 and S6†). For this, the CL time traces were integrated at different concentrations of F⁻. As shown in Fig. 4a, the integrated CL intensity was almost perfectly proportional to the F⁻ concentration for both TBS-luminol (Fig. S5†) and TIPS-luminol, giving a limit of detection (LOD) was 362 nM for TBS-luminol and 18 nM for TIPS-luminol at a signal-to-noise (S/N) ratio of 3. TIPS-luminol is thus shown to be extremely sensitive compared to other reported chemodosimeters.^29,30 We also examined the selectivity of the TIPS-luminol sensor system. Addition of various tetrabultyammonium salts of anions such as Cl⁻, Br⁻, I⁻, NO₃⁻, HSO₄⁻, and CN⁻ had negligible effect on the CL of TIPS-luminol, confirming the highly selective nature of TIPS-luminol for the F⁻ (Fig. 4b).

Fig. 4 (a) Plots of integrated CL intensity versus TBAF concentration for TIPS-luminol (b) relative integrated CL intensity of TIPS-luminol (1.1 × 10⁻⁵ M) after treatment with tetrabutylammonium salt of various anions (7.1 × 10⁻⁴ M) in THF and water solutions.

Importantly the response time of our sensor is also very fast. Chemodosimeters generally show rather a slow response time because they are reaction based sensors.⁴¹ For instance, in certain systems, the time necessary to reach equilibrium was 40 min to 2 h, and one of the chemodosimeters needed 1400 equivalents of F⁻ to reach saturation of the signals.⁴² The UV-Vis spectra were employed to measure the response time of TIPS-luminol. As shown in Fig. S6,† the absorption spectrum is saturated within 10 s upon addition of TBAF, even at very low concentrations of TIPS-luminol (THF, 1.1 × 10⁻⁵ M) and TBAF (THF, 1.1 × 10⁻⁵ M). Apparently, labile nature of TIPS-enolether leads to a remarkable enhancement in the reaction rate.⁴⁰

Finally, we investigated the fluorescence response of TIPS-luminol with common inorganic fluorides such as KF in aqueous solution. This study is very important because most of the common fluorides are inorganic in nature. For this, an excess of KF (1.0 × 10⁻² M, 100 μl) was added into a THF solution of TIPS-luminol (3.4 × 10⁻³ M, 100 μl). Subsequent exposure of the resulting solution to a mixed water solution gave CL emission; however, its CL intensity is 10 times lower than the TBAF treated one (Fig. S8†). We attributed the reduced CL intensity to the reduced formation of the monoanion intermediate arising from the high ion pairing energy of KF in the THF–water mixture system. Therefore, we used a DMSO–water mixture as a new solvent system to enhance CL intensity. As expected, the more polar DMSO–water mixture efficiently diminished cation interference and increased the CL intensity by 5 times compared to the THF–water system (Fig. S8†).

Conclusions

In summary, we successfully developed a CL based sensor with a built-in self-signaling feature by adapting the chemodosimeter concept. We found that a masking group incorporated to luminol efficiently suppresses the CL route of luminol and that the selective removal of the masking group by a target analyte can turn on the CL process. Through systematic studies of TBS-luminol and TIPS-luminol, we optimized the molecular design to achieve a highly sensitive and selective CL chemodosimeter for F⁻ detection. The optimized conditions rendered highly sensitive (LOD = 18 nM) and selective fluoride sensing in aqueous environments. We also contributed to understanding the origin of the selective synthesis of the silylsubstitued luminol through the combined NMR analyses and DFT calculations. We anticipate that our new sensor system offers an efficient way to achieve highly sensitive, selective, and convenient detection of various important analytes.

Acknowledgements

This work was partly supported by the Converging Research Center Program funded by the Ministry of Science, ICT and Future Planning (Project no. 2013K000314). The work in Madrid was supported by the Spanish Ministry for Science and Innovation (MICINN), project CONMOL, grant no. CTQ2011-27317, and by the Campus of International Excellence (CEI) UAM+CSIC. JG and BMM thank M. Wykes (Madrid) and P. Trouillas (Limoges) for helpful discussions.

Notes and references

1. I. Weeks, Chemiluminescence Immunoassay Wilson and Wilson's Comprehensive Analytical Chemistry, Elsevier, Amsterdam, 1992, vol. 29, p. 13.
2. K. Robards, Anal. Chim. Acta, 1992, 266, 147.
3. I. Bronstein, J. Fortin, P. E. Stanley, G. S. A. B. Stewart and L. J. Kricka, Anal. Biochem., 1994, 219, 169.
4. C. Dodeigne, L. Thunus and R. Lejeune, Talanta, 2000, 51, 415.
5. K. A. Fähnrich, M. Pravda and G. G. Guilbault, Talanta, 2001, 54, 531.
6. A. Roda, P. Pasini, M. Mirasoli, E. Michelini and M. Guardigli, Trends Biotechnol., 2004, 22, 295.
7. Z. F. Zhang, H. Cui, C. Lai and L. Liu, Anal. Chem., 2005, 77, 3324.
8. W. Miao, Chem. Rev., 2008, 108, 2506.
9. L. Hu and G. Xu, Chem. Soc. Rev., 2010, 39, 3275.
10. G. C. Van de Bittner, E. A. Dubikovskaya, C. R. Bertozzi and C. J. Chang, Proc. Natl. Acad. Sci. U. S. A., 2010, 107, 21316.
11. A. Roda and M. Guardigli, Anal. Bioanal. Chem., 2012, 402, 69.
12. T. S. Bailey and M. D. Pluth, J. Am. Chem. Soc., 2013, 135, 16697.
13. J. A. Ocaña-González, M. Ramos-Payán, R. Fernández-Torres, M. V. Navarro and M. Á. Bello-López, Talanta, 2014, 122, 214.
14. C. J. Yang, M. Pinto, K. Schanze and W. Tan, Angew. Chem., Int. Ed., 2005, 44, 2572.
15. K. Lee, L. K. Povlich and J. Kim, Adv. Funct. Mater., 2007, 17, 2580.
16. J. Lee, S. Seo and J. Kim, Adv. Funct. Mater., 2012, 22, 1632.
17. C. Liu, Z. Li, B. Du, X. Duan and Y. Wang, Anal. Chem., 2006, 78, 3738.
18. S. Cai, K. Lao, C. Lau and J. Lu, Anal. Chem., 2011, 9702.
19. S. Bi, S. Hao, L. Li and S. Zhang, Chem. Commun., 2010, 46, 6093.
20. Z. Yang, Z. Xie, H. Liu, F. Yan and H. Ju, Adv. Funct. Mater., 2008, 48, 2308–2312.
21. P. Yang, S. Jin, Q. Xu and S. Yu, Small, 2013, 9, 199.
22. S. Bi, J. Zhang and S. Zhang, Chem. Commun., 2010, 5509.
23. Y. Gao and B. Li, Anal. Chem., 2013, 85, 11494.
24. R. Freeman, X. Liu and I. Willner, J. Am. Chem. Soc., 2011, 133, 11597.
25. X. Liu, R. Freeman and I. Willner, ACS Nano, 2011, 5, 7648.
26. S. Zhao, G. Qin, Y. Huang, S. Li, X. Lu, J. Jiang and F. Ye, Anal. Methods, 2012, 4, 1927.
27. Y. Yang, Q. Zhao, W. Feng and F. Li, Chem. Rev., 2013, 113, 192.
28. Y. Omote, T. Miyake, S. Ohmori and N. Sugiyama, Bull. Chem. Soc. Jpn., 1967, 40, 899.
29. M. Nakazono, M. Asechi and K. Zaitsu, J. Photochem. Photobiol., A, 2004, 163, 149.
30. P. Connett, Fluoride, 2007, 40, 155.
31. L. F. M. L. Ciscato, D. Weiss, R. Beckert and W. J. Baader, J. Photochem. Photobiol., A, 2011, 218, 41.
32. N. Hoshiya, N. Fukuda, H. Maeda, N. Watanabe and M. Matsumoto, Tetrahedron, 2006, 62, 5808.
33. I. S. Turan and E. U. Akkaya, Org. Lett., 2014, 16, 1680.
34. H. O. Z. Albrecht, Phys. Chem., 1928, 136, 321.
35. E. H. Huntress, L. N. Stanley and A. S. Parker, J. Am. Chem. Soc., 1934, 56, 241.
36. G. Merényi, J. Lind and T. E. Eriksen, J. Biolumin. Chemilumin., 1990, 5, 53.
37. N. S. Moyon, A. K. Chandra and S. Mitra, J. Phys. Chem. A, 2010, 114, 60.
38. T. G. Burdo and W. R. Seitz, Anal. Chem., 1975, 47, 1639.
39. Z. A. Fataftah, A. M. Rawashdeh and C. Sotiriou-Leventis, Synth. Commun., 2001, 31, 2379.
40. T. W. Greene and P. G. M. Wuts, Protective Groups in Organic Synthesis, John Wiley & Sons, New York, 1999, p. 114.
41. L. Fu, F. Jiang, D. Fortin, P. D. Harvey and Y. Liu, Chem. Commun., 2011, 47, 5503.
42. S. Y. Kim and J.-I. Hong, Org. Lett., 2007, 9, 3109.

---

Footnote

† Electronic supplementary information (ESI) available: Details of synthesis and characterization of TBS-luminol and TIPS-luminol was included. See DOI: 10.1039/c4ra08182j

---