A ratiometric fluorogenic probe for the real-time detection of SO₃²⁻ in aqueous medium: application in a cellulose paper based device and potential to sense SO₃²⁻ in mitochondria

Abstract

A new water soluble and fluorogenic probe (L) that can demonstrate a specific ratiometric detection of a SO₂ derivative (SO₃²⁻) in 100% aqueous medium and live cells has been designed and synthesized. The detection process can be visualized by the naked eye, as the orange-red fluorescence of L turns into a strong blue fluorescence upon interaction with SO₃²⁻. L displayed several beneficial attributes such as detection in complete aqueous medium, extremely fast response time along with high selectivity and sensitivity. The ratiometric sensing was attributed to the selective nucleophilic addition reaction of SO₃²⁻ with L. The probe was further used to develop a low cost microfluidic sensor device (μPAD). The probe was biocompatible and its potential to sense SO₃²⁻ in mitochondria was captured in live HeLa cells.

---

Introduction

Sulfur dioxide (SO₂) is a well-known air pollutant, which is prevalent in the atmosphere owing to the extensive combustion of fossil fuels across the globe. From a toxicological point of view, increasing SO₂ levels are detrimental to the environment and climate and according to the US-EPA, continuous release of sulfur dioxide into the atmosphere is a major reason for increasing acid rain, which causes severe damage to crops, trees, lakes, rivers, and animals. SO₂ is highly soluble in water (40 g L⁻¹) and the SO₂ related toxicity can be mainly attributed to its derivatives, sulfite (SO₃²⁻) and bisulfite (HSO₃⁻) (3 : 1 M/M, in neutral fluid).^1,2 It is very likely that sulfurous acid produced in the respiratory tract from inhaled hydrated SO₂ may trigger the generation of these two derivatives. Epidemiological studies suggest that even though SO₂ renders physiological benefits such as vasodilatation and lowering of blood pressure,^3,4 prolonged SO₂ exposure may lead to respiratory ailments,⁵ lung cancer, cardiovascular diseases^6,7 and various neurological disorders including migraine headaches, stroke and brain cancer.⁸ Recent studies have revealed that SO₂ gas can be generated endogenously from sulfur-containing amino acids⁹ through biosynthetic pathways^10,11 like transamination by aspartate amino-transferase (AAT).¹² So, SO₂ has been identified as the fourth gasotransmitter after nitric oxide (NO), carbon monoxide (CO) and hydrogen sulfide (H₂S), respectively. However, the exact role of SO₂ in the context of tumour cell physiology and in normal cells has not been resolved yet. Thus the rapid detection of SO₂ and its hydrated derivatives (sulfite and bisulfite) in biological, environmental and food/beverage samples has sparked immense research interest. Fluorescence-based sensing probes can facilitate the rapid, non-invasive and sensitive detection of target analytes, which underlines their utility and superiority over other sensing systems for the detection of biological species.^13–19 However, a fluorescence based sensor must overcome the influence of various factors such as the localization of the probe, changes of the environment around the probe (pH, polarity, temperature, and so forth), emission collection efficiency, effective cell thickness in the optical beam and changes in the excitation intensity in order to demonstrate efficient cellular imaging.^20,21 In this regard, developing a ratiometric fluorescent probe is ideal as it not only reduces the influence of such factors but also offers precise quantification of the target. Very little progress has been made in developing efficient fluorescent sensors for detecting SO₂ (or sulfite and bisulfite) in biological and environmental samples as most of these reported detection systems suffer from drawbacks in terms of selectivity, detection limit, response time (5 min to 1 h) and background signal inside a cell.^22–29 More recently, some new and reliable sensors for the detection of SO₂ in live cells have been reported.^30–35 However, these probes require a considerable amount of organic solvents (10% DMF, 30% DMF, 40% glycerol, 30% DMF, 20% EtOH and 15% EtOH) as co-solvents along with water, which can curb their application potential in biological samples. Moreover, even though some of the reported probes exhibited the intracellular detection of SO₂ derivatives, the possibility of engineering low cost devices using these chemo-sensors has seldom been explored.

In order to address these issues, herein we report a versatile fluorogenic probe that can rapidly sense SO₃²⁻ through a highly selective and ratiometric response in 100% aqueous medium and live cells. This study also describes the engineering of a cellulose paper-based microfluidic device for the facile detection of SO₃²⁻.

Experimental

General information and materials

All the materials for synthesis were purchased from commercial suppliers and used without further purification. The absorption spectra were recorded on a PerkinElmer Lambda-25 UV-Vis spectrophotometer using 10 mm path length quartz cuvettes in the range of 300–800 nm wavelength, whereas fluorescence measurements were performed on a Horiba Fluoromax-4 spectrofluorometer using 10 mm path length quartz cuvettes with a slit width of 3 nm at 298 K. The mass spectrum of the probe L was obtained using a Waters Q-ToF Premier mass spectrometer. Nuclear magnetic resonance (NMR) spectra were recorded on a Bruker Advance 600 MHz instrument. Chemical shifts were recorded in parts per million (ppm) on the scale. The following abbreviations are used to describe spin multiplicities in ¹H NMR spectra: s = singlet; d = doublet; t = triplet; q = quartet, and m = multiplet.

Synthesis of the probe L

3-Ethyl-1,1,2-trimethyl-1H-benzo[e]indol-3-ium was prepared following the procedure reported in our previously published literature work (step 1 of Scheme S1, ESI†).³⁶ In the next step, 2.0 mmol of 3-ethyl-1,1,2-trimethyl-1H-benzo[e]indol-3-ium obtained in the first step was dissolved in dry EtOH by gentle warming. Then, 1.0 mmol of isophthalaldehyde was added to it and the resulting mixture was refluxed for 10 h to obtain a brown-red solid L. Calculated yield: 68%. ¹H NMR [600 MHz, DMSO-d6, J (Hz), δ (ppm)]: 8.97 (1H, s), 8.65 (2H, d, J = 16.8), 8.55 (2H, d, J = 7.8), 8.49 (2H, d, J = 8.4), 8.36 (2H, d, J = 9.0), 8.26 (2H, d, J = 7.8), 8.22 (2H, d, J = 9.0), 7.94 (2H, d, J = 16.2), 7.87–7.84 (3H, m), 7.78 (2H, t, J = 7.8), 4.98 (4H, q, J = 7.2), 2.10 (12H, s), 1.60 (6H, t, J = 7.2). ¹³C NMR [150 MHz, DMSO-d6, TMS, δ (ppm)]: 182.02, 151.01, 139.19, 138.14, 135.39, 133.63, 133.43, 131.27, 130.11, 130.04, 128.60, 127.58, 126.73, 123.34, 113.61, 113.57, 113.42, 54.19, 43.08, 25.36, 14.21. ESI-MS (positive mode, m/z) calculated for C₄₂H₄₂N₂²⁺: 287.1674. Found: 287.1726.

Crystallization of L

A small amount of brown-red solid L was taken in a test tube containing a DCM–EtOH mixture and sonicated for 5 min to obtain a clear solution. Then the solution was allowed to evaporate slowly at room temperature and after one week X-ray mountable plate shaped orange-red colored crystals appeared at the bottom of the test tube.

Crystallographic refinement details

In this case, a crystal of suitable size was selected from the mother liquor and immersed in silicone oil, then mounted on the tip of a glass fiber and cemented using epoxy resin. Intensity data for all the crystals were collected using Mo-Kα radiation (λ = 0.71073 Å) at 298(2) K, with increasing ω (width of 0.3° per frame) at a scan speed of 6 s per frame on a Bruker SMART APEX diffractometer equipped with a CCD area detector. Data integration and reduction were processed with SAINT³⁷ software. An empirical absorption correction was applied to the collected reflections with SADABS.³⁸ The structures were solved by direct methods using SHELXTL³⁹ and were refined on F² by the full-matrix least-squares technique using the SHELXL-97 program package.⁴⁰ Graphics were generated using MERCURY 3.0.⁴¹ In all cases, non-hydrogen atoms are treated anisotropically. The hydrogen atoms were located on a difference Fourier map and refined. In other cases, the hydrogen atoms are geometrically fixed.

UV-vis and fluorescence spectroscopic studies

Stock solutions of various analytes (1 × 10⁻¹ mol L⁻¹) were prepared in methanol or Millipore water depending on the preferred solubility of the analytes chosen. A stock solution of L (5 × 10⁻³ mol L⁻¹) was prepared in DMSO. The solution of L was then diluted to 10 × 10⁻⁶ mol L⁻¹ for spectral studies by taking only 4.0 μL stock solution of L and making the final volume up to 2.0 mL by adding aqueous phosphate buffered saline (PBS) solution (pH 7.4). In the fluorescence selectivity experiment, the test samples were prepared by placing appropriate amounts of the stock solutions of the respective anions/analytes into 2.0 mL of probe solution (10 × 10⁻⁶ mol L⁻¹). For UV-visible and fluorescence titration experiments, another set of SO₃²⁻ standard solution having 2.5 mM and 5.0 mM concentrations was prepared by diluting the previously prepared stock solutions (1 × 10⁻¹ mol L⁻¹) in Millipore water. Quartz optical cells of 1.0 cm path length were filled with 1.0 mL and 2.0 mL solutions of L for UV-visible and fluorescence titration experiments, respectively, to which the newly prepared stock solutions of the analytes (2.5 mM and 5.0 mM) were gradually added using a micropipette as necessary. For fluorescence measurements, excitation was provided at 380 nm and emission was acquired from 400 nm to 750 nm. Spectral data were recorded within 1 min after addition of the ions except for kinetic studies.

Detection limit

Detection limits were calculated on the basis of fluorescence titration. The fluorescence emission spectrum of L was recorded 10 times, and the standard deviation of the blank measurement was estimated (λ_em = 470 nm). To measure the slope, the fluorescence emission at 470 nm was plotted as a function of the concentration of SO₃²⁻ from the titration experiment. The detection limit was then calculated using the following equation:

Detection limit = 3σ/k (1)

where σ is the standard deviation of the blank measurement and k is the slope between the fluorescence emission intensity versus [SO₃²⁻].

Microfluidic device (μPAD) fabrication

A microfluidic device (μPAD) was developed from simple Whatman 42 filter paper based on the inbuilt capillary behavior of cellulose papers. We have followed a slightly modified device fabrication process compared to several reported systems.^42,43 Firstly, a filter paper was immersed into a 1% polystyrene solution (in toluene) for several hours (Scheme S2, ESI†). The paper was then air-dried and a polystyrene coated hydrophobic cellulose paper was obtained. Subsequently, the paper was cut into a square (2.5 cm × 2.5 cm) to which another normal filter paper (properly cut) having a sampling area and a detection area connected by a straight micro-channel was pasted (see the graphical representation in Scheme S2, ESI†). Then, the probe L was carefully dropcast on the detection area by a drop on demand (DOD) method. The device was then used in application studies.

Cytotoxicity studies with L, L–SO₃²⁻ adduct and Na₂SO₃

HeLa cells (human cervical carcinoma cells) were propagated in a 25 cm² tissue culture flask in DMEM medium supplemented with 10% (v/v) FBS, penicillin (100 μg mL⁻¹) and streptomycin (100 μg mL⁻¹) under a humidified atmosphere of 5% CO₂ until the cells were approximately 80% confluent. Prior to the MTT assay, cells were trypsinized and seeded into 96 well tissue culture plates at 10⁴ cells per well and incubated with varying concentrations (1.0 μM–64 μM) of L made in DMEM for a period of 24 h. Following incubation, growth media were carefully aspirated and fresh DMEM containing MTT solution was added to the wells. The plate was subsequently incubated for 4 h at 37 °C. Following incubation, the supernatant was discarded and the insoluble colored formazan product was solubilized in DMSO and its absorbance was measured in a microtiter plate reader (Infinite M200, TECAN, Switzerland) at 550 nm with a reference of 670 nm. For each concentration of the test samples, the MTT assay was performed in six independent sets. Data analysis and determination of the standard deviation were accomplished using Microsoft Excel 2016 (Microsoft Corporation). In the MTT assay, the absorbance for the control cells was considered as 100% cell viability and the absorbance obtained for the treated cells was used to ascertain % cell viability with respect to the control. The MTT assay was also conducted in a parallel experiment to determine the cytotoxic potential of the L–SO₃²⁻ adduct in cells. To this end, HeLa cells seeded into 96 well tissue culture plates (10⁴ cells per well) were incubated with varying concentrations of 1.0 μM–64 μM of L made in DMEM for 30 min under 5% CO₂. Subsequently, sodium sulfite salt was added to HeLa cells pre-incubated with L so as to achieve an L–sulfite adduct in the ratios of 1 : 5, 1 : 10 and 1 : 20. The cells were then incubated for a period of 24 h under 5% CO₂ and subjected to the MTT assay as mentioned before. In a separate experiment, HeLa cells treated with Na₂SO₃ salt alone (5.0 μM–1280 μM) were also subjected to the MTT assay.

Detection of sulfite in HeLa cells by cell imaging

HeLa cells (human cervical carcinoma cells) were initially cultured in a 25 cm² tissue culture flask containing DMEM medium supplemented with 10% FBS, penicillin (100 μg mL⁻¹) and streptomycin (100 μg mL⁻¹) in a CO₂ incubator. Prior to imaging studies, HeLa cells were seeded onto a 6 well plate and grown in DMEM medium at 37 °C until 80% confluency in a CO₂ incubator. Subsequently, the cells were washed once with sterile PBS and incubated with 10 μM of L in DMEM at 37 °C for 5.0 min in a CO₂ incubator. The cells were further washed with sterile PBS in order to remove excess L and their images were acquired using a fluorescence microscope (Eclipse Ti-U, Nikon, USA) with filters that allowed blue and red light emission. In a separate experiment, HeLa cells pre-treated with 100 μM of sodium sulfite (prepared in sterile PBS) for 5 min were further incubated with 10 μM of L for 5.0 min to achieve the final 1 : 10 ratio of L–SO₃²⁻. Subsequently, the cells were washed with sterile PBS in order to remove excess salt and then their images were acquired using a fluorescence microscope (Eclipse Ti-U, Nikon, USA) with filters that allowed blue and red light emission. In order to ascertain cellular localization of the L–SO₃²⁻ adduct, HeLa cells were stained with 1.0 μM MitoTracker dye in DMEM at 37 °C for 30 min in a CO₂ incubator. The cells were again washed with sterile PBS and treated with 200 μM of sodium sulfite in sterile DMEM for a period of 5.0 min followed by treatment with 10 μM of L and incubated further for 5.0 min to facilitate L–SO₃²⁻ adduct formation. The cells were again washed with sterile PBS in order to remove excess dye and their images were acquired using a fluorescence microscope with a filter that allowed red light emission. The co-localization of the images for the MitoTracker and the L–SO₃²⁻ adduct was performed using ImageJ software and the JACoP plugin.⁴⁴

Results and discussion

Design, synthesis and structure of the probe L

SO₂ derivatives (SO₃²⁻ and HSO₃⁻) are regarded as good nucleophiles that can readily react with α,β-unsaturated compounds very rapidly in aqueous solution.⁴⁵ Hence, the fundamental principle of the probe design was based on a nucleophilic addition reaction, which can subsequently be monitored by a rapid change in fluorescence. The probe L was synthesized by the condensation reaction of isophthalaldehyde with 3-ethyl-1,1,2-trimethyl-1H-benzo[e]indol-3-ium in dry ethanol (Scheme S1, ESI†), wherein 1H-benzo[e]indol-3-ium was used not only as a signalling moiety but also to improve the solubility of the fluorescent probe in aqueous medium. It is worth mentioning that in our previous report even though we were able to acquire a ratiometric fluorescence response by exciting the sensor at two different wavelengths, under UV light the probe³⁶ demonstrated only a TURN-ON response (conspicuous to the naked eye) as the native probe exhibited very low fluorescence due to the existing twisted intramolecular charge transfer (TICT) process. As a result, in our previous study we were unable to achieve a dual channel ratiometric bio-imaging and the probe was able to exhibit only a TURN-ON fluorescence response inside live HeLa cells. Hence, unlike our previous work,³⁶ in the present probe design we have purposely avoided involvement of the intramolecular charge transfer (ICT) process by designing an acceptor–π-acceptor fluorogenic system rather than a donor–π-acceptor system (Scheme 1), so that a dual channel ratiometric intracellular detection of SO₃²⁻ is feasible.

Scheme 1 Design principle of probe L.

It should also be highlighted here that when benzaldehyde was used in place of isophthalaldehyde in the probe design (by Suna et al.),⁴⁶ it only rendered a TURN-ON response rather than a ratiometric response owing to the non-fluorescent character of the free probe, which justifies the introduction of an acceptor–π-acceptor fluorogenic system in the present study by incorporating isophthalaldehyde in the probe structure. The single X-ray crystal structure clearly revealed that L has a ‘bat’ like structural orientation (Fig. 1A) and an almost planar geometry, which negates the possibility of TICT. The packing diagram of L around the ‘c’ axis manifested a ‘honeycomb’ like structure (Fig. 1B).

Fig. 1 (A) X-ray crystal structure of L and its structural resemblance to a ‘bat’. (B) Packing diagram shown in the ‘c’ axis.

Colorimetric response of the probe L towards SO₃²⁻

The probe displayed excellent aqueous solubility as anticipated, and its increased water solubility can perhaps be ascribed to the incorporation of an extra 1H-benzo[e]indol-3-ium moiety in the probe design (Scheme 1). The UV-vis spectra of the probe L recorded in almost 100% (actual 99.8% aqueous medium containing 0.2% DMSO) aqueous medium (PBS, pH 7.4) revealed two well defined strong absorbance maxima at 370 nm and 435 nm, which might be attributed to π–π* and n–π* transitions, respectively (Fig. 2A).

Fig. 2 (A) UV-visible spectra of L (10 μM) in ∼100% aqueous buffer (PBS, pH 7.4) in the presence of excess (20 equivalents) of various analytes. Inset: Visual change in the color of the solution of L in the presence of different analytes under daylight. (B) UV-visible spectra of L (10 μM) in the presence of varying concentrations of SO₃²⁻. Inset: Changes in the absorbance at 435 nm with the addition of equivalents of SO₃²⁻.

The addition of excess (20 equivalents) of SO₃²⁻ to L resulted in a significant obliteration of the absorbance peak at 435 nm, while the absorbance maximum at 370 nm (Fig. 2A) too had concurrently reduced substantially. This spectral change was complemented by a conspicuous change in the color of the experimental solution from orange-red (only L) to colorless (L⁺, SO₃²⁻) (Fig. 2A). Interestingly, no other analytes (H₂PO₄⁻, SO₄²⁻, HSO₄⁻, F⁻, NO₃⁻, ClO₄⁻ and SH⁻) were able to produce any significant change in the absorbance spectra upon interacting with L (Fig. 2A). A titration experiment with an incremental addition of SO₃²⁻ resulted in a systematic reduction of the absorbance peaks of L at 435 nm and 370 nm (Fig. 2B). Hence, the UV-visible spectral studies essentially suggested that the probe L could have potentially undergone nucleophilic attack by SO₃²⁻ to exhibit such colorimetric changes owing to the loss of extended conjugation in L. A detailed fluorescence study was pursued to further investigate the analytical prospect of L.

Fluorescence spectral response of the probe L towards various analytes

The free probe L (10 μM) showed a well-defined strong emission peak at 573 nm (λ_ex = 380 nm, Fig. 3) and yielded a bright orange fluorescence in ∼100% aqueous PBS (pH 7.4) which can be accredited to the existing uninterrupted extended π conjugation in the system. The selectivity of the probe was ascertained in the presence of an excess of various analytes which included F⁻, Cl⁻, Br⁻, I⁻, NO₂⁻, NO₃⁻, CH₃COO⁻, H₂PO₄⁻, PF₆⁻, PPi, HCO₃⁻, SO₄²⁻, HSO₄⁻, ClO₄⁻, SCN⁻, S₂O₃²⁻, S₂O₈²⁻, C₆H₇O₆⁻ (ascorbate), SO₃²⁻, SH⁻, Cys, Hcy and GSH. A fluorescence selectivity experiment clearly depicted that no noteworthy change could be observed in the spectral nature of L in the presence of the aforementioned analytes except for the SO₂ derivative, SO₃²⁻ (Fig. 3). Interestingly, the addition of SO₃²⁻ to L resulted in the generation of a high intensity new emission peak at 470 nm, whereas the original emission maximum of L at 573 nm had disappeared completely (Fig. 3).

Fig. 3 Fluorescence spectra of L (10 μM) in ∼100% aqueous buffer (PBS, pH 7.4) in the presence of excess (20 equivalents) of various anions and nucleophilic reagents (Cys, Hcy, and GSH). Visual changes in the fluorescence of L in the presence of different analytes under UV-light (bottom panel).

This selective fluorogenic response of L towards SO₃²⁻ was also evident visually as the orange emitting solution of L turned into a strong bluish-green fluorescent solution upon interacting with SO₃²⁻ (Fig. 3, bottom panel). It should be mentioned here that the fluorescence response of L towards SO₃²⁻ in solution was very rapid as all the fluorescence spectra were recorded within 1 min of the addition of the analytes. Moreover, the fluorescence kinetic studies of L with and without the addition of SO₃²⁻ indicated that a significant increase in the emission intensity could be observed within seconds after the addition of SO₃²⁻ to L, which underlines the real-time detection ability of L (Fig. S8, ESI†) as compared to several reported probes.^22–29 Interestingly, this instantaneous fluorescence response could be witnessed by the naked eye also (a supporting video is provided in the ESI† of the manuscript). L showed a similar but diminished fluorescence response towards HSO₃⁻ like SO₃²⁻ (Fig. S9, ESI†). However, it may be mentioned that this experiment was conducted in ∼100% aqueous buffer at pH 7.4 and at this pH, the HSO₃⁻ species is likely to exist as SO₃²⁻ in solution. Also, a variable pH study suggested that L can sense SO₃²⁻ in the pH range of 6.0 to 8.0 through TURN-ON fluorescence responses (Fig. S10, ESI†) as SO₃²⁻ is the predominant species in this pH range.

Meanwhile, the addition of ammonia to L only serves as the addition of a base (like NaOH) to the probe and does not interfere with the sensing of SO₃²⁻ (Fig. S11, ESI†). It is worth mentioning that even the addition of huge excess (200 equivalents) of some inorganic sulfur compounds (SCN⁻, S₂O₃²⁻, and S₂O₈²⁻), reactive sulfur species (HS⁻, Cys, Hcy and GSH) and a reducing agent (sodium ascorbate) to L induced a negligible change in the emission behavior of L, which reaffirmed the selectivity of L towards SO₃²⁻. We have also checked the selectivity of the probe L towards CN⁻ and OCl⁻ as some of the recent literature studies^47–49 suggested that these reactive species could also potentially undergo a reaction with cyanine based compounds. It is encouraging to note that OCl⁻ and CN⁻ rendered a very insignificant change in the fluorescence when added in excess to L (Fig. S12, ESI†). Hence, the selectivity of L is solely specific towards SO₃²⁻.

In order to acquire a quantitative understanding of the fluorescence spectral change of L upon interaction with varying levels of SO₃²⁻, a fluorescence titration experiment was carried out. With incremental addition of SO₃²⁻ to the solution of L in ∼100% aqueous PBS there was a systematic enhancement in its emission intensity at 470 nm and a concurrent decrease in the original emission peak of L at 573 nm (Fig. 4). This concomitant change in the emission intensities at two different wavelengths (470 nm and 573 nm) provided an additional handle to track SO₃²⁻ concentration. Essentially, the ratio of the fluorescence emission intensities at two different wavelengths (I₄₇₀/I₅₇₃) varied from 0.0094 to 12.06 up to the addition of 10 equivalents of SO₃²⁻. This significant change (∼1282 fold increase) in the emission signal ratios not only introduced high sensitivity into the detection system but also provided a wide dynamic range for the ratiometric sensing of SO₃²⁻ under physiological conditions by measuring the signal ratios at two wavelengths. Further, it is notable that the two emission bands are widely spaced (over 103 nm), which also enhances the feasibility of a dual emission ratiometric imaging in cells. The detection limit for SO₃²⁻ was determined from the fluorescence titration experiment (the slope was obtained from the change in the intensity at 470 nm only) and it was found to be 1.056 × 10⁻⁷ M or 8.45 ppb (Fig. S13, ESI†), which is much lower compared to several reported fluorescent probes for SO₂ derivatives.^22–30,33

Fig. 4 Fluorescence spectra of L (10 μM) in the presence of varying concentrations of SO₃²⁻. Inset: Changes in the emission intensity ratio (I₄₇₀/I₅₇₃) with the addition of equivalents of SO₃²⁻.

Engineering cellulose paper based sensor devices

The ultra-sensitive, rapid, ratiometric and naked eye detecting ability of the probe L provided us the opportunity to develop a user-friendly and low-cost detection device for the on-site analysis of SO₂ derivatives. Cellulose paper (Whatman 42 filter paper), being a low-cost, abundantly available, and disposable substrate, has been preferentially selected by us as the core material of the device over glass or a polymer.^42,43 A preliminary experiment revealed that L (10 μM) could retain its fluorogenic behavior by exhibiting a visible orange fluorescence, when dropcast on a cellulose paper (Fig. 5A). Subsequent addition of one drop of SO₃²⁻ (50 μM in water) rapidly altered the orange fluorescence of L into a highly bluish-green emitting spot, which underlines the effectiveness of utilizing the probe in a low cost device fabrication. Hence, we took a step further to engineer a cellulose paper-based microfluidic analytical device wherein liquid transport would be entirely driven by capillary flow and there will be no additional requirement for external pumping. A simple polystyrene coated microfluidic device (μPAD) was developed (Fig. 5B) by dropcasting L in the detection area (Experimental section and Scheme S2, ESI†). It was remarkable to note that the device could successfully sense the trace SO₂ derivative (SO₃²⁻) imported in the sampling part of the μPAD through a rapid and conspicuous change in fluorescence (Fig. 5C).

Fig. 5 (A) L dropcast on cellulose paper with and without the addition of SO₃²⁻. (B) Design of the developed μPAD. (C) μPAD device before and after treatment with SO₃²⁻.

Plausible sensing mechanism

Even though it is well established that α,β-unsaturated compounds can undergo nucleophilic attack by reactive anionic species like SO₃²⁻ in aqueous medium,⁴⁵ there are some contradicting claims about which probe–SO₃²⁻ adduct is formed. Some of the reports suggested that nucleophilic attack takes place at the ‘C2’ site whereas the majority of the reports asserted the reaction took place at the ‘C4’ site. A few reports even advocated the subsequent ‘C=C’ reduction after nucleophilic addition at the ‘C4’ site. Hence, even though ratiometric fluorescence based sensing of the anionic SO₃²⁻ by L in the present case is primarily attributed to the nucleophilic addition reaction, in order to gain insight into the mechanism of sensing, we have isolated the pure colorless product (L–SO₃²⁻ adduct) and characterized the same by NMR and mass spectroscopy experiments (Fig. S14 and S15, ESI†).

Both the ¹H-NMR and mass spectral studies validated the detection mechanism depicted in Scheme 2 by confirming the formation of the presented L–SO₃ adduct. So, the present study also supported the notion of the majority of the reports that the reaction at the ‘C4’ site (without further reduction) was the key phenomenon for the observed fluorescence response.

Scheme 2 Plausible sensing mechanism of L.

Dual channel ratiometric live cell imaging

Encouraged by solution-based studies, our next endeavour was to ascertain the potential of L in the detection of sulfite in cells by fluorescence-based imaging. To this end, evaluation of cytotoxic potential by an MTT assay illustrated that neither L alone nor the L–SO₃²⁻ adduct (in ratios of 1 : 5, 1 : 10 and 1 : 20) could exert any toxic effect on live HeLa cells up to a concentration of 64 μM (Fig. S16, ESI†), which validated the biocompatibility of the probe. In cell imaging studies, HeLa cells incubated with L for just 5 minutes showed negligible fluorescence in the blue emission channel and a bright red fluorescence in the red channel, which was perinuclear and essentially localized in the cytoplasm (Fig. 6). This suggested that L could rapidly traverse across the cell membrane. However, when HeLa cells pre-treated with Na₂SO₃ (100 μM) for 5 minutes were incubated with L for less than 5 minutes, they displayed an intense blue fluorescence in the blue channel with a concomitant and a dramatic drop of red channel fluorescence (Fig. 6). In a co-localization experiment, the merged images obtained with the mitochondrial dye MitoTracker and L–SO₃²⁻ adduct revealed that the fluorescence of MitoTracker and L–SO₃²⁻ co-localized well with a high Pearson coefficient of 0.88 (Fig. 7), which suggested the potential of the probe in detecting SO₃²⁻ in mitochondria. It may be highlighted that unlike several reported probes, the biocompatible probe L exhibited rapid cellular uptake and rendered an extremely fast ratiometric intracellular sensing of SO₃²⁻.

Fig. 6 Fluorescence microscopy images of HeLa cells treated with 10 μM of L only and HeLa cells pre-treated with 100 μM sulfite followed by the addition of L (10 μM). The scale bar for the images is 20 μm.

Fig. 7 (A) Fluorescence microscopy images of HeLa cells co-stained by MitoTracker and the L–SO₃²⁻ adduct. The scale bar for the images is 20 μm. (B) Co-localization of the mitochondrial dye MitoTracker (red channel emission) and L–SO₃²⁻ adduct (blue channel emission). The co-localization coefficient (Pearson coefficient) is 0.88.

Conclusions

In summary, the rationally designed probe L demonstrated specific ratiometric sensing of the SO₂ derivative (SO₃²⁻), which can be visualized even by the naked eye. L could provide several unique advantages like sensing in completely aqueous medium and extremely fast response time along with very high selectivity and sensitivity. The probe was successfully employed in developing a low cost μPAD sensor device for detecting the trace SO₂ derivative (SO₃²⁻), which might be useful for preliminary environmental monitoring. In addition, the excellent biocompatibility, cell permeability as well as fast response time of the probe could be leveraged for the ratiometric intracellular imaging of SO₃²⁻ in live HeLa cells.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

We thank the Council of Scientific and Industrial Research (01/2727/13/EMR-II), the Science & Engineering Research Board (SR/S1/OC-62/2011) and the Department of Biotechnology (BT/PR13560/COE/34/44/2015) for research grants and the Central Instruments Facility (CIF), IIT Guwahati for providing analytical facilities. SS, SH, PD and UM thank IIT Guwahati for research fellowships.

Notes and references

1. Z. Q. Meng, G. H. Qin, B. Zhang and J. L. Bai, Mutagenesis, 2004, 19, 465.
2. X. Shi, J. Inorg. Biochem., 1994, 56, 155.
3. Z. Meng and H. Zhang, Inhalation Toxicol., 2007, 19, 979.
4. S. Du, H. Jin, D. Bu, X. Zhao, B. Geng, C. Tang and J. Du, Acta Pharmacol. Sin., 2008, 29, 923.
5. S. Iwasawa, Y. Kikuchi, Y. Nishiwaki, M. Nakano, T. Michikawa, T. Tsuboi, S. Tanaka, T. Uemura, A. Ishigami, H. Nakashima, T. Takebayashi, M. Adachi, A. Morikawa, K. Maruyama, S. Kudo, I. Uchiyama and K. Omae, J. Occup. Health, 2009, 51, 38.
6. G. Li and N. Sang, Ecotoxicol. Environ. Saf., 2009, 72, 236.
7. J. Li, R. Li and Z. Meng, Eur. J. Pharmacol., 2010, 645, 143.
8. N. Sang, Y. Yun, H. Li, L. Hou, M. Han and G. K. Li, Toxicol. Sci., 2010, 114, 226.
9. M. H. Stipanuk, J. E. Dominy Jr., J. Lee and R. M. Coloso, J. Nutr., 2006, 136, 1652S.
10. M. H. Stipanuk, Annu. Rev. Nutr., 2004, 24, 539.
11. H. Kimura, Exp. Physiol., 2011, 96, 833.
12. D. Liu, Y. Huang, D. Bu, A. D. Liu, L. Holmberg, Y. Jia, C. Tang, J. Du and H. Jin, Cell Death Dis., 2014, 5, e1251.
13. S. Samanta, S. Goswami, Md. N. Hoque, A. Ramesh and G. Das, Chem. Commun., 2014, 50, 11833.
14. S. Samanta, T. Ray, F. Haque and G. Das, J. Lumin., 2016, 171, 13–18.
15. S. Samanta, S. Goswami, A. Ramesh and G. Das, J. Photochem. Photobiol., A, 2015, 310, 45.
16. S. Samanta, S. Goswami, A. Ramesh and G. Das, Sens. Actuators, B, 2014, 194, 120.
17. S. Samanta, C. Kar and G. Das, Anal. Chem., 2015, 87, 9002–9008.
18. S. Samanta, U. Manna, T. Ray and G. Das, Dalton Trans., 2015, 44, 18902.
19. S. Samanta, B. K. Datta, M. Boral, A. Nandan and G. Das, Analyst, 2016, 141, 4388.
20. T. Kowada, H. Maeda and K. Kikuchi, Chem. Soc. Rev., 2015, 44, 4953.
21. K. Kikuchi, Chem. Soc. Rev., 2010, 39, 2048.
22. X. Cheng, H. Jia, J. Feng, J. Qin and Z. Li, Sens. Actuators, B, 2013, 184, 274.
23. X. Ma, C. Liu, Q. Shan, W. Guo, D. Wei and Y. Du, Sens. Actuators, B, 2013, 188, 1196.
24. M. G. Choi, J. Hwang, S. Eor and S. K. Chang, Org. Lett., 2010, 12, 5624.
25. H. Tian, J. Qian, Q. Sun, H. Bai and W. Zhang, Anal. Chim. Acta, 2013, 788, 165.
26. Y. Q. Sun, J. Liu, J. Zhang, T. Yang and W. Guo, Chem. Commun., 2013, 49, 2637.
27. J. Xu, K. Liu, D. Di, S. Shao and Y. Guo, Inorg. Chem. Commun., 2007, 10, 681.
28. Y. Sun, C. Zhong, R. Gong, H. Mu and E. Fu, J. Org. Chem., 2009, 74, 7943.
29. L. E. Santos-Figueroa, C. Gimenez, A. Agostini, E. Aznar, M. D. Marcos, F. Sancenon, R. Martinez-Manez and P. Amoros, Angew. Chem., Int. Ed., 2013, 52, 13712.
30. X. Zhu, L. Zhu, H.-W. Liu, X. Hu, R.-Z. Peng, J. Zhang, X.-B. Zhang and W. Tan, Anal. Chim. Acta, 2016, 937, 136.
31. U. Diwan, V. Kumar, R. K. Mishra, N. K. Rana, B. Koch, M. K. Singh and K. K. Upadhyay, Anal. Chim. Acta, 2016, 929, 39.
32. L. Zhu, J. Xu, Z. Sun, B. Fu, C. Qin, L. Zeng and X. Hu, Chem. Commun., 2015, 51, 1154.
33. Y. Liu, K. Li, M. Y. Wu, Y. H. Liu, Y. M. Xie and X. Q. Yu, Chem. Commun., 2015, 51, 10236.
34. X. Yang, Y. Zhou, X. Zhang, S. Yang, Y. Chen, J. Guo, X. Li, Z. Qing and R. Yang, Chem. Commun., 2016, 52, 10289.
35. W. Xu, C. L. Teoh, J. Peng, D. Su, L. Yuan and Y.-T. Chang, Biomaterials, 2015, 56, 1.
36. S. Samanta, P. Dey, A. Ramesh and G. Das, Chem. Commun., 2016, 52, 10381.
37. G. M. Sheldrick, SAINT and XPREP, 5.1, Siemens Industrial Automation Inc., Madison, WI, 1995.
38. G. M. Sheldrick, SADABS, software for Empirical Absorption Correction, University of Gottingen, Institute fur Anorganische Chemie der Universitat, Tammanstrasse 4, D-3400 Gottingen, Germany, 1999–2003.
39. G. M. Sheldrick, SHELXTL Reference Manual: Version 5.1, Bruker AXS, Madison, WI, 1997.
40. G. M. Sheldrick, SHELXL-97: Program for Crystal Structure Refinement, University of Gottingen, Gottingen, Germany, 1997.
41. Mercury 2.3 Supplied with Cambridge Structural Database, CCDC, Cambridge, UK, 2011–2012.
42. K. Yamada, T. G. Henares, K. Suzuki and D. Citterio, Angew. Chem., Int. Ed., 2015, 54, 5294.
43. K. Yamada, S. Takaki, N. Komuro, K. Suzuki and D. Citterio, Analyst, 2014, 139, 1637.
44. S. Bolte and F.P. Cordelieres, J. Microsc., 2006, 224, 213.
45. M. Morton and H. Landfield, J. Am. Chem. Soc., 1952, 74, 3523.
46. Y. Suna, S. Fan, S. Zhang, D. Zhao, L. Duan and R. Li, Sens. Actuators, B, 2014, 193, 173.
47. X. Lv, J. Liu, Y. Liu, Y. Zhao, Y. Q. Sun, P. Wang and W. Guo, Chem. Commun., 2011, 47, 12843.
48. Z. Yang, Z. Liu, Y. Chen, X. Wang, W. He and Y. Lu, Org. Biomol. Chem., 2012, 10, 5073.
49. Z. Lou, P. Li, P. Song and K. Han, Analyst, 2013, 138, 6291.

---

Footnote

† Electronic supplementary information (ESI) available: Experimental section and supporting plots and figures. CCDC 1534470. For ESI and crystallographic data in CIF or other electronic format see DOI: 10.1039/c7an01368j

---