An A–π–A′ structural ratiometric fluorescent probe based on benzo[e]indolium for bisulfite and its application in sugar samples and living cells

Abstract

An A–π–A′ structural ratiometric fluorescent probe 1 bearing a benzo[e]indolium moiety was developed for the detection of HSO₃⁻ in PBS buffer solution. HSO₃⁻ was expected to react with probe 1 via a Michael addition reaction, causing dramatic changes in its emission spectrum (from 578 nm to 466 nm) and an obvious color change from orange red to blue which could be observed by the naked eye. The reaction could be complete in 30 seconds in the sensing system, and displayed a high selectivity and sensitivity for HSO₃⁻ with a detection limit of 15 nM. The practical value of the probe was confirmed for application to detect the level of HSO₃⁻ in sugar samples with good recovery and in living cells.

---

1. Introduction

Bisulfite (HSO₃⁻), as an effective preservative, has been widely used in foods, beverages and pharmaceutical products.¹ It has been reported that extensive HSO₃⁻ intake would cause asthmatic attacks and allergic reactions, such as difficulty in breathing, wheezing, hives, and gastrointestinal distress in some individuals.^2,3 Due to the above mentioned harmful effects towards people, the threshold levels of HSO₃⁻ in food and medicine have been controlled rigorously in many countries. For instance, the total content of sulfur (calculated by SO₂) in white granulated sugar is strictly regulated as <30 mg kg⁻¹ in China.⁴ Hence, considerable attention has been attracted to the development of rapid, facile, and reliable techniques for HSO₃⁻ detection because of the food quality assurance and human healthy.

Fluorescent probes have become powerful tools for sensing and imaging trace amounts of samples due to their simplicity, sensitivity, low-toxicity and easy of operation in recent years.^5,6 To date, some fluorescent probes have been developed for monitoring HSO₃⁻ on the basis of complexation with amines,^7,8 the selective de-protection of a levulinate group,^9–12 selective nucleophilic reaction with aldehyde^13–20 and N=N in metal complexes,^21,22 and Michael-type additions.^23–39 However, most of these probes may suffer from the risk of slow response time (2 min to 10 h) or an organic co-solvent, which limited their utility. Recently, Song and co-workers reported a fluorescent turn-on probe and a ratiometric fluorescent probe, respectively.^31,32 Both of them were based on a near-infrared benzopyrylium dye which could detect HSO₃⁻ in aqueous solution (HEPES, pH = 7.4) with an incredible response time (<15 s) and a satisfactory detection limit. Compared to the intensity-based fluorescent probes, the ratiometric fluorescent probes enable the measurement of emission intensities at two different wavelengths, which increase the dynamic range and provide a built-in correction for environmental effects.⁴⁰ Thereby, it has remained a great challenge to construct new sensitive and selective ratiometric fluorescent probes for the rapid HSO₃⁻ detection in pure aqueous environments and in real samples.

Herein, we introduce a new ratiometric fluorescent probe 1 which is constructed by a benzo[e]indolium fluorophore (an electron acceptor A) and a malononitrile group (another electron acceptor A′), that is, an A–π–A′ type compound. The A–π–A′ type compound could effectively avoid the severe emission quenching which were frequently encountered in D–π–A type compounds owing to dipole-induced intermolecular interactions.⁴¹ Probe 1 has shown a selective response to HSO₃⁻ over other anions in PBS buffer (pH = 7.4, 10 mM), and a distinguished fluorescent color change from orange red to cyan can be observed. HSO₃⁻ is expected to undergo a Michael addition reaction with the C-4 atom in the ethylene group, which is activated by the electron-withdrawing feature of the positively-charged benzo[e]indolium fragment and the malononitrile group (Scheme 1).

Scheme 1 Probe 1 and its synthesis route.

2. Experimental

2.1 Materials and chemicals

All reagents and solvents were purchased from commercial source and used without further purification, if not stated. All reactions were carried out on the magnetic stirrers and their reaction process was monitored on thin layer chromatography (TLC). Stock solutions (100 mM) of HSO₃⁻, HS⁻, F⁻, Cl⁻, Br⁻, I⁻, AcO⁻, ClO₄⁻, NO₃⁻, N₃⁻, SO₄²⁻, HSO₄⁻, SCN⁻, PO₄³⁻, HPO₄²⁻, H₂PO₄⁻, Cys, Hcy, and GSH were prepared by direct dissolution of proper amounts of their sodium salts. Absorption and fluorescence spectra were taken on a Shimadzu UV-1800 spectrophotometer and a Hitachi F-2700 fluorescence spectrometer, respectively. ¹H NMR and ¹³C NMR measurements were recorded at 600 and 150 MHz on a Brucker Avance 600 MHz spectrometer. Dimethyl sulfoxide (DMSO-d₆) was solvent, and tetramethylsilane (TMS) was used as internal standard. HRMS (ESI) were taken on a Fourier transform ion cyclotron resonance mass spectrometry (Varian 7.0T). An Olympus FV-1000 confocal fluorescence microscope and HepG2 cells were used in the living cell experiment.

2.2 Preparation of probe 1

In a 25 mL round-bottom flask, terephthalaldehyde (200 mg, 1.49 mmol) and malononitrile (98 mg, 1.49 mmol) were dissolved in ethanol (15 mL) and the mixture was stirred at 80 °C for 12 h. After cooling to room temperature, the precipitate was collected by filtration, washed with cold ethanol to afford the crude product. The pure product was obtained by further purification using silica gel chromatography as a light yellow solid 2 (190 mg, 70%). ¹H NMR (600 MHz, CDCl₃) δ 9.87 (s, 1H), 8.65 (s, 1H), 8.48 (d, 2H), 8.45 (d, 2H).

To a solution of 1, 2, 3, 3-tetramethylbenz[e]indolium iodide (70 mg, 0.2 mmol) in ethanol (10 mL) was added the prepared 2 (36 mg, 0.2 mmol) and the solution was refluxed for 12 h. After cooling to room temperature, the precipitate was collected by filtration, washed with cold ethanol. The pure product was obtained by recrystallization from ethanol as dark red solid (88 mg, 77%). ¹H NMR (600 MHz, DMSO-d₆): δ 8.65 (s, 1H), 8.49 (m, 4H), 8.35 (d, 1H), 8.25 (d, 1H), 8.17 (d, 1H), 8.12 (d, 2H), 7.90 (d, 1H), 7.88 (t, 1H), 7.76 (t, 1H), 4.34 (s, 3H), 2.03 (s, 6H). ¹³C NMR (150 MHz, DMSO-d₆): δ 186.0, 176.5, 167.2, 152.5, 142.4, 141.9, 137.8, 134.4, 134.0, 131.8, 131.0, 128.9, 127.2, 118.8, 116.9, 57.6, 39.2, 28.0. HRMS (ESI) calcd. for [M]⁺ 388.1806, found 388.1808 (Fig. S1–S3†).

2.3 Preparation of solutions of probe 1 and analytes

Deionized water was used as the solvent for titration experiment. The titrations were carried out in 10 mm quartz cuvettes at room temperature. Probe 1 was dissolved in DMSO (spectroscopic grade) to afford a concentration of 10 mM stock solution, and then diluted with PBS buffer (pH = 7.4, 10 mM) to a concentration of 20 μM. Anions (as their Na⁺ salt, 10 mM and 100 mM) in deionized water were added to the diluted probe solution and used for the sensing behavior experiment. The resulting solution was shaken well and kept at room temperature for 30 seconds before measuring the spectra. The excitation wavelength was 400 nm, and the PMT voltage was 400 V. The excitation and emission slit width were 5 nm and 5 nm, respectively.

2.4 ¹H NMR analysis experiments of probe 1

The solution of probe 1 (3 × 10⁻³ M) in DMSO-d₆ (450 μL) and D₂O (50 μL) was placed in the NMR tube, and sodium bisulfite powder was added. All the sodium bisulfite was soluble.

2.5 Sugar samples test

Granulated sugar, soft sugar and crystal sugar purchased from supermarket were used in the sample analysis. Sample solution was prepared by dissolving 5.0 g of sugar in deionized water and diluting to 10 mL. Aliquots of the sugar solution were added directly to the PBS buffer (pH = 7.4, 10 mM) containing probe 1 (20 μM), and the emission intensity at 466 nm and 578 nm were recorded.

2.6 Preparation of test papers and detection

The neutral filter papers were dipped in the stock solution of 1 (10 mM) and dried. Then various solutions of anions (1 mM, 2.5 μL) were seriatim dropped. The concentrations of sugar samples for test paper experiment were 1, 0.7, 0.4, 0.1 g mL⁻¹.

2.7 Cell incubation and fluorescence imaging

The HepG2 cells were provided by of Shanxi Medical University. Cells were seeded onto the cover slips and cultured in DMEM in an incubator (37 °C, 5% CO₂ and 20% O₂). After 24 hours, the cover slips were washed 3 times with PBS to remove the media and then cultured in PBS for use. Probe 1 (10 μM) was added to above cellular samples and incubated for 15 min, and then washed 3 times with PBS. In a further experiment, HSO₃⁻ (15 μM) was added to the cells and the fluorescence change was observed under the confocal fluorescence microscope.

2.8 Cytotoxicity assays

HepG2 cells were seeded onto the cover slips and cultured in DMEM in an incubator (37 °C, 5% CO₂ and 20% O₂). Before the experiment, the cells were placed in a 96-well plate, followed by addition of various concentrations of probe 1. Then the cells were incubated at 37 °C in an atmosphere of 5% CO₂ and 20% O₂ for 12 hours, followed by MTT assays (n = 5).

3. Results and discussion

3.1 Spectral studies of probe 1

The detection environment was investigated at first. As shown in Fig. S4,† the probe was stable in a pH range of 1–8 and displayed the satisfied response for HSO₃⁻ in the physiological pH region. Thus, the UV-vis absorption and emission spectra of probe 1 towards HSO₃⁻ were measured in PBS buffer (pH = 7.4, 10 mM). Fig. 1 shows the absorption (a) and emission (b) responses of probe 1 (20 μM) with different amounts of HSO₃⁻ (0–60 μM). Probe 1 exhibited a yellow color with an obvious band at 380 nm and a shoulder band at 445 nm (Fig. 1a), which was ascribed to the typical absorption induced by the π–π conjugation. Upon addition of HSO₃⁻, the absorption bands were gradually attenuated and a new band at 268 nm was appeared, suggesting that the π–π conjugation is interrupted due to the nucleophilic attack of HSO₃⁻. The addition also induced an obvious color change from yellow to colorless (inset of Fig. 1a), which could be observed by naked eye. Gratifyingly, the free probe 1 showed an emission band centered at 578 nm. In contrast, upon addition of HSO₃⁻ (Fig. 1b), the emission band at 578 nm was decreased and a new band at 466 nm increased remarkably, which could be attributed to the formation of 2-methylene benzo[e]indoline moiety. A well-defined isoemissive point at 536 nm was also observed. The ratiometric change in fluorescence intensity indicated that the addition reaction between probe 1 and HSO₃⁻ interrupted the π-conjugation, and the fluorescence of 2-methylenebenzo[e]indoline was recovered. Accordingly, the fluorescence color changed from orange red to cyan (inset of Fig. 1b). The distinct emission shift (about 112 nm) was favorable for ratiometric fluorescence detection. Moreover, the ratio of the emission intensities at 466 and 578 nm (I₄₆₅/I₅₇₁) was increased (25-fold, from 0.10 to 2.5) with the addition of HSO₃⁻ and became constant when the amount of HSO₃⁻ added reached 2.4 equiv. (48 μM) (Fig. S5†). Under similar conditions, the ratio changes produced an excellent linear function with HSO₃⁻ concentration between 0 and 24 μM, and the detection limit for HSO₃⁻ was determined to be 0.15 μM (R² = 0.9930) based on S/N = 3 (Fig. S6†), which is sufficient to probe the HSO₃⁻ concentration in cells (ca. 16 μM).²³

Fig. 1 Absorption (a) and emission spectra (b) of probe 1 (20 μM) in PBS buffer (pH 7.4, 10 mM) treated with HSO₃⁻ and their photographs in the absence or presence of HSO₃⁻ (60 μM) under sunlight (a) and UV light (b, 365 nm). λ_ex = 400 nm. Slits: 5 nm/5 nm. The concentrations of HSO₃⁻ were varied in the following order: 0, 4, 8, 12, 16, 20, 24, 28, 32, 36, 40, 44, 50 and 60 μM.

To further evaluate the specific nature of probe 1 towards HSO₃⁻, the influence of other anions were examined under the same conditions. As shown in Fig. 2 and S7,† after adding all the other anions into the solution of probe 1 (20 μM), they did not induce any significant fluorescence change, and the I₄₆₅/I₅₇₁ values varied in a limited range between 0.12 and 0.25. By contrast, treatment of probe 1 with HSO₃⁻ resulted in an obvious ratiometric fluorescent response. Competition experiments of probe 1 were also determined by various species mentioned previously in the same condition. The signaling of probe 1 toward HSO₃⁻ was not affected by the presence of 60 μM of coexisting species (Fig. S8†). The phenomenon suggests that probe 1 could be used as an efficient signaling probe for HSO₃⁻ detection in aqueous media.

Fig. 2 Emission ratio I₄₆₆/I₅₇₈ of 1 (20 μM) in PBS buffer (pH 7.4, 10 mM) in the presence of various species. λ_ex = 400 nm. Slits: 5 nm/5 nm. (Inset) Corresponding fluorescent color changes under a handle UV lamp (365 nm): 1: probe 1 alone; 2–20: probe 1 + HSO₃⁻, HS⁻, F⁻, Cl⁻, Br⁻, I⁻, AcO⁻, ClO₄⁻, NO₃⁻, N₃⁻, SO₄²⁻, HSO₄⁻, SCN⁻, PO₄³⁻, HPO₄²⁻, H₂PO₄⁻, Cys, Hcy and GSH (anions: 60 μM; Cys and Hcy: 200 μM; GSH: 10 mM).

3.2 Sensing mechanism

A well-known sensing process takes place between 1 and HSO₃⁻ with a Michael addition reaction. ¹H NMR analysis was carried out to demonstrate the sensing mechanism. As anticipated, after addition of HSO₃⁻ into the solution of 1, the nucleophilic attack of HSO₃⁻ toward C-4 disturbed the π–π conjugation and caused all the proton signals shifted to up-field (Fig. 3). It was noticed that the proton of the methyl group (Ha) connected with N⁺ was dramatically shifted from δ 4.34 to 3.09 ppm. Another irrefutable evidence was the equivalent proton of the two methyl groups (Hb) at δ 2.03 ppm, then also shifted to up-field and splitted into two signals (Hb′, at δ 1.89 and 1.56 ppm). This adduct was further confirmed and characterized by HRMS analysis (Fig. S9†), where the peak at m/z 468.1398 (calc. = 468.1387) corresponding to [M − SO₃]⁻ was clearly observed.

Fig. 3 ¹H NMR spectral change of 1 (3 × 10⁻³ M) in the absence (A) and presence of 1 equiv. (B), of NaHSO₃ in DMSO-d₆ : D₂O = 9 : 1 (v/v).

3.3 Real samples analysis

To investigate the applicability of the probe to real samples, the levels of HSO₃⁻ in granulated sugar, soft sugar and crystal sugar were examined by using the proposed method.¹³ The recoveries were good enough for practical use and determination values are listed in Table 1. The HSO₃⁻ levels in these samples were 16.96, 10.16, and 14.46 mg kg⁻¹, respectively.

Table 1 Results for the determination of HSO₃⁻ in various samples

Samples	HSO3^− level/(μmol L^−1)	Added/(μmol L^−1)	Found/(μmol L^−1)	Recovery/%
Granulated sugar	4.07	4	7.93	98.26
		6	10.02	99.50
Soft sugar	2.44	4	6.61	102.64
		6	8.76	103.79
Crystal sugar	3.47	4	7.34	98.26
		6	9.31	98.31

---

3.4 Test papers detection

A preliminary paper test strip system was constructed as shown in Fig. S10.† After dropping various anions to the prepared test papers, only HSO₃⁻ made the color changed from orange to blue under the UV light (365 nm). Moreover, the test papers could also detect the HSO₃⁻ in real sugar samples as shown in Fig. 4 and S11.† It can be seen that glucose without HSO₃⁻ did not cause any color change; while the sugar solution made the color changing at different concentrations.

Fig. 4 The photos of the filter paper containing 1 treated with real sugar samples under UV light (365 nm, bottom): probe 1, HSO₃⁻, glucose, soft sugar of 1 g mL⁻¹, soft sugar of 0.7 g mL⁻¹, soft sugar of 0.4 g mL⁻¹, and soft sugar of 0.1 g mL⁻¹.

3.5 Cellular imaging

The potential application of probe 1 for living cell detection of HSO₃⁻ was investigated. Confocal fluorescence imaging in living HepG2 cells was carried out under selective excitation at 405 nm. First, the MTT assay for probe 1 showed that the probe with a concentration range between 5–15 μM has only minimal cytotoxicity after 12 h (Fig. S12†), thus the probe at 10 μM was selected for imaging experiments. The cells incubated with probe 1 (10 μM) for 15 min displayed strong fluorescence in the red channel and weak fluorescence in the green channel, confirming that probe 1 is cell-permeable (Fig. 5). Since a certain concentration of HSO₃⁻ (15 μM) has been indicated in cells, the observed strong fluorescence in the green channel is probably a result of the fact that probe 1 binds HSO₃⁻ in cells. These cell experiments indicated that probe 1 could competent for imaging HSO₃⁻ in living cells.

Fig. 5 Confocal fluorescence images in Hala cells. (A) Bright field imaging of the cells with adding probe 1 (10 μM) for 15 min; (B) fluorescent imaging of (A) from the green channel; (C) fluorescent imaging of (A) from the red channel; (D) overlap of (A) and (C). (E) Bright field imaging of the cells were treated with probe 1 (10 μM) for 15 min and subsequently treated with HSO₃⁻ (15 μM) for another 15 min; (F) fluorescent imaging of (E) from the green channel; (G) fluorescent imaging of (E) from the red channel; (H) overlap of (E) and (F). λ_ex = 405 nm.

4. Conclusion

In summary, we have developed a new ratiometric fluorescent probe bearing a benzo[e]indolium group for HSO₃⁻ detection in PBS buffer under physiology conditions. The probe displayed a dramatic fluorescent change and a color change after exposure to HSO₃⁻ with a very short response time (within 30 s). Probe 1 allowed the detection of HSO₃⁻ with high sensitivity at a very low concentration. These advantages made the probe a potential sensor for HSO₃⁻ detection in real sugar samples and in living cells.

Acknowledgements

This work was supported by the National Natural Science Foundation of china (No. 21301126).

References

1. T. Fazio and C. R. Warner, Food Addit. Contam., 1990, 7, 433–454.
2. S. L. Taylor, N. A. Higley and R. K. Bush, Adv. Food Res., 1986, 30, 1–76.
3. H. Vally, N. L. Misso and V. Madan, Clin. Exp. Allergy, 2009, 39, 1643–1651.
4. White granulated sugar GB317-2006, General administration of quality supervision, Inspection and Quarantine of the People's Republic of China, 2006.
5. J. Chan, S. Dodani and C. Chang, Nat. Chem., 2012, 4, 973–984.
6. M. Schäferling, Angew. Chem., Int. Ed., 2012, 15, 3532–3554.
7. Y. Sun, C. Zhong, R. Gong, H. Mu and E. Fu, J. Org. Chem., 2009, 74, 7943–7946.
8. C. Wang, S. Feng, L. Wu, S. Yan, C. Zhong, P. Guo, R. Huang, X. Weng and X. Zhou, Sens. Actuators, B, 2014, 190, 792–799.
9. M. G. Choi, J. Hwang, S. Eor and S.-K. Chang, Org. Lett., 2010, 12, 5624–5627.
10. X. Gu, C. Liu, Y.-C. Zhu and Y.-Z. Zhu, J. Agric. Food Chem., 2011, 59, 11935–11939.
11. S. Chen, P. Hou, J. Wang and X. Song, RSC Adv., 2012, 2, 10869–10873.
12. H. Zhang, S. Xue and G. Feng, Sens. Actuators, B, 2016, 231, 752–758.
13. K. Chen, Y. Guo, Z. Lu, B. Yang and Z. Shi, Chin. J. Chem., 2010, 28, 55–60.
14. X.-F. Yang, M. Zhao and G. Wang, Sens. Actuators, B, 2011, 152, 8–13.
15. Y. Yang, F. Huo, J. Zhang, Z. Xie, J. Chao, C. Yin, H. Tong, D. Liu, S. Jin, F. Cheng and X. Yan, Sens. Actuators, B, 2012, 166–167, 665–670.
16. Y.-Q. Sun, P. Wang, J. Liu, J. Zhang and W. Guo, Analyst, 2012, 137, 3430–3433.
17. G. Wang, H. Qi and X.-F. Yang, Luminescence, 2013, 28, 97–101.
18. X. Cheng, H. Jia, J. Feng, J. Qin and Z. Li, Sens. Actuators, B, 2013, 184, 274–280.
19. H. Xie, X. Jiang, F. Zeng, C. Yu and S. Wu, Sens. Actuators, B, 2014, 203, 504–510.
20. X. Su, X. Li, T. Ding, G. Zheng and Z. Liu, J. Organomet. Chem., 2015, 781, 59–64.
21. G. Li, Y. Chen, J. Wang, Q. Lin, J. Zhao, L. Ji and H. Chao, Chem. Sci., 2013, 4, 4426–4433.
22. G. Li, Y. Chen, J. Wang, J. Wu, G. Gasser, L. Ji and H. Chao, Biomaterials, 2015, 63, 128–136.
23. Y.-Q. Sun, J. Liu, J. Zhang, T. Yang and W. Guo, Chem. Commun., 2013, 49, 2637–2639.
24. H. Tian, J. Qian, Q. Sun, H. Bai and W. Zhang, Anal. Chim. Acta, 2013, 788, 165–170.
25. M.-Y. Wu, T. He, K. Li, M.-B. Wu, Z. Huang and X.-Q. Yu, Analyst, 2013, 138, 3018–3025.
26. M.-Y. Wu, K. Li, C.-Y. Li, J.-T. Hou and X.-Q. Yu, Chem. Commun., 2014, 50, 183–185.
27. L. Zhu, J. Xu, Z. Sun, B. Fu, C. Qin, L. Zeng and X. Hu, Chem. Commun., 2015, 51, 1154–1156.
28. Q. Zhang, Y. Zhang, S. Ding, H. Zhang and G. Feng, Sens. Actuators, B, 2015, 211, 377–384.
29. X. Dai, T. Zhang, Z.-F. Du, X.-J. Cao, M.-Y. Chen, S.-W. Hu, J.-Y. Miao and B.-X. Zhao, Anal. Chim. Acta, 2015, 888, 138–145.
30. G. Xu, H. Wu, X. Liu, R. Feng and Z. Liu, Dyes Pigm., 2015, 120, 322–327.
31. W. Chen, Q. Fang, D. Yang, H. Zhang, X. Song and J. Foley, Anal. Chem., 2015, 87, 609–616.
32. W. Chen, X. Liu, S. Chen, X. Song and J. Kang, RSC Adv., 2015, 5, 25409–25415.
33. Y. Chen, X. Wang, X.-F. Yang, Y. Zhong, Z. Li and H. Li, Sens. Actuators, B, 2015, 206, 268–275.
34. Y. Liu, K. Li, M.-Y. Wu, Y.-H. Liu, Y.-M. Xie and X.-Q. Yu, Chem. Commun., 2015, 51, 10236–10239.
35. Y. Liu, K. Li, K.-X. Xie, L.-L. Li, K.-K. Yu, X. Wang and X.-Q. Yu, Chem. Commun., 2016, 52, 3430–3433.
36. H. Zhang, Z. Huang and G. Feng, Anal. Chim. Acta, 2016, 920, 72–79.
37. Y. Zhang, L. Guan, H. Yu, Y. Yan, L. Du, Y. Liu, M. Sun, D. Huang and S. Wang, Anal. Chem., 2016, 88, 4426–4431.
38. D.-P. Li, Z.-Y. Wang, X.-J. Cao, J. Cui, X. Wang, H.-Z. Cui, J.-Y. Miao and B.-X. Zhao, Chem. Commun., 2016, 52, 2760–2763.
39. J. Chao, Z. Li, Y. Zhang, F. Huo, C. Yin, Y. Liu, Y. Li and J. Wang, J. Mater. Chem., 2016, 4, 3703–3712.
40. L. Y. Niu, Y. S. Guan, Y. Z. Chen, L. Z. Wu, C. H. Tung and Q. Z. Yang, J. Am. Chem. Soc., 2012, 134, 18928–18931.
41. M. Shimizu, R. Kaki, Y. Takeda, T. Hiyama, N. Nagai, H. Yamagishi and H. Furutani, Angew. Chem., Int. Ed., 2012, 51, 4095–4099.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra15605c

---