An optical sensor for reactive oxygen species: encapsulation of functionalised silica nanoparticles into silicate nanoprobes to reduce fluorophore leaching

Abstract

Sol–gel nanoprobes, also known as Photonic Explorer for Bioanalysis with Biologically Localised Embedding (PEBBLE), capable of performing in-vitrointracellular monitoring of reactive oxygen species have been developed using a modified form of 5(6)-carboxyfluorescein diacetate. A sol–gel matrix was selected for the design of the probes as it is photostable, optically transparent and chemically inert, and to minimise leaching of the dye from the porous matrix it was covalently immobilised to silica nanoparticles (15 nm). Using this approach, 0.1% of the dye was found to leach over a typical analysis time of 5 h and minimal photobleaching was observed. In addition, the nanoprobes were shown to respond to hydrogen peroxide, hydroxyl anions, nitric oxide, peroxynitrile and superoxide anions, obtaining limits of detection of 2.2, 1.1, 3.2, 1.1 and 1.1 nM respectively. The nanoprobes were subsequently introduced into bovine oviducts using a lipid transfection reagent (Escort IV™) and fluorescence was observed.

---

Introduction

To gain a true understanding of biological processes it is necessary to perform intracellular measurements of viable cells; consequently, stable and selective analytical techniques are required. One area of measurement, which is of particular interest in clinical research, is the determination of reactive oxygen species (ROS). These species were thought only to be released in host defence roles by phagocytic cells; however, it is now clear that ROS have a role in cell signalling for many biological systems.¹ They are easily inter-converted and can subsequently react with larger biological molecules, causing chain reactions to occur, which can lead to changes in both the structure and function of cellular components. To date, intracellular sensing of these species has involved the use of free fluorescent dyes;² this technique is, however, limited in its application due to problems such as the occurrence of non-specific protein binding, which can lead to enhanced fluorescent signals, along with the cytotoxic effects of the dyes used.³

Minimally invasive nanoprobes or nanosensors (20–300 nm), which are small enough to be internalised within the cell without perturbation, have also been used to perform intracellular measurements. These sensors include those based on gold nanoparticles⁴ and quantum dots,⁵ where the analyte recognition component is immobilised on the external surface of the nanoparticle. As the active molecule remains exposed to species within the cell, this can lead to interferences with the chemical measurement. An alternative approach is to immobilise the analytical reagent within a porous matrix, thus protecting the cell from any cytotoxic effects and the sensing element from any interferences. This technique affords a nanosensor termed a Photonic Explorer for Bioanalysis with Biologically Localised Embedding or PEBBLE.⁶ PEBBLE are typically fabricated from silicate sol–gels,⁶polyacrylamide⁷ or liquid polymers;⁸ however, silicate sol–gel matrices are well suited for the measurement of reactive species as they are photostable, optically transparent and chemically inert and as such were the material selected for use herein. The versatility of such sensors is illustrated through the variety of intracellular measurements performed which include the determination of glucose, oxygen, calcium, singlet oxygen, hydroxyl and potassium.^9–14 As well as being minimally invasive, the matrix provides the sensing element with a controlled environment in which to operate, not only protecting it from high molecular mass components such as proteins but also from being sequestered by cellular organelles.⁶ The incorporation of reference dyes within the nanoprobe also allows ratiometric measurements to be performed, which take into account any fluctuations in light source intensity and enable the sensors to be calibrated ex-vitro and used in-vitro as the same fluorescent response is obtained in both environments.⁹ It has also been observed that the dye remains active within the cell for longer when this mode of encapsulation is employed, due to random dye interactions being minimised by the polymeric coating.¹¹

The porosity of the PEBBLE matrix, through which analytes diffuse and interact with the entrapped dye, can, however, be problematic as the molecular probes employed are often smaller than the pores and have a tendency to leach. Leaching of the dye negates the positive effects of encapsulation and the cell may now experience the cytotoxic effects observed with free dyes. To some extent this has been addressed by the incorporation of dextran-supported dyes, an approach that relies on an increase in the overall size to retain the dye conjugate within the matrix; this technique is, however, limited by the small number of commercially available dyes. More recently we reported a simple and efficient technique to reduce leaching of a chemiluminescent dye from silicate sol–gel matrices via the covalent attachment of the dye to silica nanoparticles (15 nm);¹⁵ hereby the insolubility of the silica nanoparticles facilitated separation of the residual free dye. With this in mind, the aim of this investigation was to extend the methodology to enable the design of a novel silicate sol–gel nanoprobe capable of performing intracellular measurements of ROS.

Experimental

Reagents

Unless otherwise stated all reagents were used as received. All water used had a purity of 5 MΩ cm^–1 and was prepared by reverse osmosis and ion exchange using an Elgast Option 4 water purifier (Elga Ltd., High Wicombe, UK). Triethylamine (Et₃N) 1 (99.5%), 3-aminopropyltriethoxysilane (APTES) 2 (99.0%), triethylorthosilicate (TEOS) (98.0%), diisopropylethylamine (ⁱPr₂EtN) 3 (99.5%), Triton^® X-100, sodium hydroxide (NaOH) and silica nanoparticles (15 nm) were supplied by Aldrich (Gillingham, UK). 5(6)-Carboxyfluorescein 4, thionyl chloride (SOCl₂) 5 and anhydrous N,N-dimethylformamide (DMF) (<0.001% H₂O) were purchased from Fluka (Gillingham, UK) and ammonium hydroxide (LR, s.g. 0.88) obtained from Prime Chemicals (Rotherham, UK). Cyclohexane, hexanol, dichloromethane (DCM), acetic anhydride6, ethanol, sulfuric acid (98.0%) and toluene were supplied by Fisher Scientific (Loughborough, UK) and deuteriochloroform (>99.8% D) was purchased from Apollo Scientific Ltd. (Stockport, UK).

Derivatisation of 5(6)-carboxyfluorescein 4 to enable covalent attachment to silica nanoparticles

Synthesis of 5(6)-carboxyfluorescein diacetate 7¹⁶. To a solution of 5(6)-carboxyfluorescein 4 (0.50 g, 1.33 mmol) in anhydrous DMF (10 ml) was added ⁱPr₂EtN 3 (0.93 ml, 5.32 mmol) followed by acetic anhydride6 (0.28 ml, 2.93 mmol), the reaction mixture was then stirred for 72 h, under N₂ and protected from light. The reaction mixture was subsequently concentrated in vacuo and the yellow gum dissolved in DCM (50 ml) and washed with deionised (DI) water (50 ml). The aqueous layer was further extracted with DCM (2 × 50 ml) and the combined extracts dried over MgSO₄ to afford 5(6)-carboxyfluorescein diacetate 7 as a pale yellow, glassy foam (see ESI† for ¹H, ¹³C NMR and MS characterisation).

Derivatisation of 5(6)-carboxyfluorescein diacetate 8. 5(6)-Carboxyfluorescein diacetate 7 (0.10 g, 0.22 mmol) was heated to reflux in SOCl₂ 5 (0.04 ml, 0.55 mmol) for 3 h prior to concentrating in vacuo. The crude acid chloride 9, in toluene (10 ml), was then added dropwise (over a period of 1 h), to a stirred solution of APTES 2 (0.06 ml, 0.26 mmol) and Et₃N1 (0.04 ml, 0.26 mmol) in toluene (10 ml) under N₂. The reaction mixture was stirred for a further 16 h prior to filtration, to remove any Et₃N·HCl, and concentrated in vacuo to afford an orange product. Recrystallisation of the product from absolute ethanol afforded the desired product as an orange, crystalline solid 8 (full characterisation in ESI† ).

Preparation of functionalised silica nanoparticles 10

Derivatised 5(6)-carboxyfluorescein diacetate 8 (0.50 g, 0.75 mmol) was added to a stirred solution of silica nanoparticles (1.00 g, 15 nm) in aqueous sulfuric acid (50%) and stirred at room temperature for 24 h. The resulting reaction mixture was centrifuged (3000 rpm), the aqueous portion decanted and the nanoparticle pellet redispersed in DI water. The nanoparticles were subsequently washed with acetone, DCM and ethanol prior to drying over acetone.

Fabrication of silicate sol–gel nanoprobes containing functionalised silica nanoparticles 10

The sol–gel nanoprobes were formed using a water-in-oil microemulsion method.¹⁴ The microemulsion was formed by mixing Triton^® X-100 (32.0 ml), cyclohexane (134.4 ml), hexanol (32.0 ml) and nanoparticles (0.49 ml) (consisting of 0.50 g of functionalised nanoparticles in 1.0 ml of DMSO and 4.0 ml of DI water), with continuous stirring. TEOS (6.4 ml) was subsequently added to a portion of the microemulsion (160 ml), followed by ammonium hydroxide (3.8 ml); the microemulsion was then stirred continuously for 24 h prior to the addition of acetone. The resulting solution was centrifuged (3000 rpm), the acetone decanted off and the PEBBLE pellet redispersed in acetone, this process being repeated several times to ensure all residual reagents were removed. The nanoprobes were then washed with ethanol and water, filtered and dried over acetone to afford a free-flowing yellow powder. Prior to evaluation of the resulting nanoprobes, the encapsulated dye was firstly activated by stirring in an aqueous solution of sodium hydroxide (0.04 M) for 2 h (see ESI† for details on the leaching investigation).

Evaluation of nanoprobes ex-vitro

Preparation of ROS species was performed as follows, where all water used for these experiments was pre-treated by passing though a column containing immobilised potassium permanganate, to remove any H₂O₂ already contained in the water, and stock solutions of hydrogen peroxide were prepared as described by Santra and co-workers.¹⁷Hydroxyl stock solutions prepared by dispersing iron(II) perchlorate in phosphate buffer (1 mM) and diluted to give a solution of 1 × 10^–4 M.¹⁸Nitric oxide was formed upon photoactivation of N-nitrosoethylaniline in phosphate buffer at 308 nm; assuming 100% release upon photoactivation the concentration of N-nitrosoethylaniline would be equivalent to the concentration of nitric oxide based on the findings of Cabail et al.¹⁹Sodium nitrite was reacted with hydrogen peroxide in acidified water with stirring and the reaction immediately quenched with sodium hydroxide. Excess H₂O₂ was removed from the solution with manganese dioxide and the solution twice filtered through a Millipore syringe filter (0.02 µm) to remove the manganese dioxide. The concentration of peroxynitrite was determined by the UV/Vis absorbance of the solution at 302 nm, based on the literature that states the molar extinction coefficient to be ε₃₀₂ = 1670 M^–1 cm^–1.²⁰Superoxide anions were obtained from standard solutions of potassium superoxide, by dissolving in DI in accordance with the literature.²¹ Once standardised, ROS solutions were diluted to afford working solutions of 5 × 10^–8 and 1 × 10^–8 M respectively (see ESI† for details on determination of LOD).

Evaluation of nanoprobes in-vitro

Bovine oviduct cells were seeded onto glass slides with an optical density of 1, in small culture dishes at a density of 5 × 10⁵ ml^–1 in culture medium. Cultures were incubated at 3 °C in a humidified atmosphere of 5% CO₂ in air. An aliquot of Escort IV™ (lipid transfection reagent) was diluted 1 : 50 with a suspension of nanoprobes to give a final concentration of 5 × 10^–3 g ml^–1. The prepared solution was added to a culture dish containing bovine oviduct cells, at 80% confluence, and incubated for 12 h at 37 °C in a humidified atmosphere of 5% CO₂ in air. At the end of the incubation period the cells were washed in phosphate buffered saline (PBS) to remove the transfection medium, along with any nanoprobes that had not been internalised by the cells. A negative control plate of oviduct cells was prepared along with a second control plate that contained cells that had been incubated with blank nanosensors. The fluorescence of the cells and nanoprobes was monitored at an excitation wavelength of 495 nm and an emission wavelength of 515 nm using a Zeiss LSM 510 Meta Axiovert 200 M confocal microscope.

Results and discussion

As previously discussed, the low molecular weight of those fluorescent dyes employed for the determination of ROS leads to leaching from the silicate sol–gel PEBBLE. This was illustrated by the incorporation of free 5(6)-carboxyfluorescein 4 into silicate sol–gel nanoprobes whereby no useful results were obtained, owing to leaching of the dye at the final washing stage of the nanosensor. Consequently, we proposed that by attaching the fluorescent dye to an insoluble anchor, such that the size of the dye conjugate would be increased, leaching from the larger sol–gel structure could be prevented, or at least minimised (Fig. 1). Silica nanoparticles were selected as the anchor as, unlike dextran supports, silica nanoparticles can be readily separated from any residual free dye by filtration. Furthermore they are available in a range of well defined particle sizes, enabling optimisation of the nanoprobe size and loading, whilst allowing the use of well established derivatisation protocols developed in solid-supported chemistry.

Fig. 1 Schematic illustrating a silica nanoparticle functionalised with the derivatised fluorescent dye 10.

As Scheme 1 illustrates, to prepare the functionalised nanoparticles protection of the dye 4, as the diacetate 7 (98%), was required to ensure that the interaction of ROS and the fluorescent properties of the fluorophore were not affected by the derivatisation; this does, however, mean that the dye must be deprotected prior to use [this was achieved via treatment of the dye with aqueous NaOH (2 h)] in order to enable oxidation of the fluorescein moiety by the ROS. Treatment of the resulting carboxylic acid 7 with SOCl₂ 5 afforded the respective acid chloride 9, which was reacted with APTES2 in the presence of Et₃N1 to afford compound 8 (96%). Therefore by synthesising a compound that possesses a triethoxysilane linker, the dye could be covalently bound to the silica nanoparticles via a facile acid-catalysed hydrolysis to afford functionalised silica nanoparticles 10 (Fig. 1). The functionalised silica nanoparticles 10 were subsequently characterised by fluorescence spectroscopy. After incubation with aq. NaOH, the suspension was neutralised with aq. HCl and diluted with phosphate buffer, prior to fluorescence analysis of the supernatant. As would be expected, the fluorescence spectra for the functionalised silica particles was different to that of the free dye, with the excitation spectra affording two possible excitation wavelengths (451 and 469 nm), with that at 451 nm giving the emission spectrum with the highest fluorescent intensity. This excitation wavelength was red-shifted by 42 nm compared to the spectrum of the free dye (493 nm) and the emission wavelength had moved from 519 to 525 nm. Prior to incorporating the fluorescently tagged nanoparticles 10 into the silicate sol–gel PEBBLE, a series of controls was prepared including blank PEBBLE (containing no dye) and those doped with the free dye, 5(6)-carboxyfluorescein 4. Using these enabled any changes in size and behaviour to be attributed to the incorporation of functionalised silica nanoparticles 10 within the sol–gel structure. As Fig. 2 illustrates, in both cases a homogeneous distribution of PEBBLE was produced; however, inclusion of the functionalised silica nanoparticles 10 led to a slight increase in the final PEBBLE size (150 ± 11 nm) compared with the blank PEBBLE (120 ± 10 nm).

Scheme 1 Reaction scheme illustrating the synthesis of derivatised 5(6)-carboxyfluorescein 8.

Fig. 2 SEM images of (a) blank nanoprobes and (b) nanoprobes containing functionalised silica nanoparticles 10 (15 nm).

Importantly however, the PEBBLE was still of a suitable size to be employed for cellular investigations. The fluorescence spectra of the PEBBLE containing functionalised silica nanoparticles 10 was then re-evaluated and, as expected, encapsulation of the particles 10 was seen to have no additional effect on the fluorescence spectrum.

Having fabricated a series of PEBBLE containing the silica functionalised dye 10, the next step was to evaluate the percentage of dye leached from the nanoprobes. To do this, fluorescent measurements were performed on a solution of the activated nanoprobes every 2 h for 10 h. It can be seen from Fig. 3 that only 0.2% of the dye was found to leach from the sol–gel PEBBLE matrix over 10 h. It is also useful to note that the majority of cellular measurements are performed within a relatively short timescale (<5 h) of the nanoprobes being dispersed, activated and introduced into the cell; over this timescale, 0.1% of the dye was found to leach.

Fig. 3 Graph illustrating the proportion of dye leached from activated nanoprobes containing functionalised nanoparticles 10.

Having demonstrated the efficient entrapment of a fluorescent dye and its stability in solution, the next step was to evaluate the PEBBLE response to ROS. As Table 1 illustrates, the PEBBLE were calibrated, ex-vitro, for a range of ROS including hydrogen peroxide, superoxide anion, hydroxyl, peroxynitrile and nitric oxide, demonstrating the capacity of the sensors to monitor changes in concentration. Most importantly, the sensors operated well over the range of concentrations that would be expected in an intracellular environment. The effect of photobleaching was also investigated for the free and encapsulated dye with continuous exposure to light for 1 h, with fluorescence measurements taken every minute. No observable difference between the dyes was detected and photobleaching was found to be minimal in both cases.

Table 1 Response of PEBBLE containing fluorescently tagged nanoparticles 10 obtained for a range of reactive oxygen species [where y is fluorescence response (au) and x is concentration (nM)]

Reactive oxygen species	Equation of line	LOD/nM	R ^2
Hydrogen peroxide (2–30 nM)	y = 0.114x – 0.121	2.2	0.996
Hydroxyl (2–28 nM)	y = 0.042x – 0.040	1.1	0.998
Nitric oxide (22–33 nM)	y = 0.194x – 2.500	3.2	0.996
Peroxynitrile (2–28 nM)	y = 0.045x – 0.015	1.1	0.998
Superoxide anion (1–35 nM)	y = 0.124x – 4.114	1.1	0.997

---

Prior to conducting in-vitro measurements with the PEBBLE, the effect of protein interference on the sensors was evaluated using bovine serum albumin (BSA), a protein commonly used in cell culture media. In addition to the nanosensors, a series of controls was also evaluated: these included the non-functionalised dye 4 and the functionalised silica nanoparticles 10. The effect of the protein on the fluorescence intensity was monitored every minute over a period of 1 h. It was found that the free dye and the silica-attached dye demonstrated a greater response to protein interference than the encapsulated dye, which is in accord with the literature.⁷

Having demonstrated the increased stability of the PEBBLE, containing fluorescently tagged nanoparticles 10 to protein interference, the ability to perform in-vitro monitoring with the sensors was subsequently investigated. Employing bovine oviducts, the performance of the PEBBLE in a cellular environment was evaluated. To achieve this, the nanoprobes were introduced into the cells using a lipid transfection reagent, Escort IV™, using a method previously reported.^7,21 A negative control plate of bovine oviduct cells incubated with Escort IV™ with no nanoprobe addition was also prepared along with a second control plate containing cells that had been incubated with blank nanosensors. As expected, no fluorescence was observed for either of the controls, only for those cells incubated with the sol–gel-encapsulated, functionalised silica-nanoparticles 10 (Fig. 4).

Fig. 4 Confocal fluorescence microscope images of bovine oviduct cells incubated with PEBBLE containing functionalised silica nanoparticles 10 illustrating the fluorescence at 515 nm (top left), the phase contrast (top right) and the merged image (bottom); excitation at 495 nm, emission at 515 nm.

Conclusions

By covalently attaching 5(6)-carboxyfluorescein diacetate to silica nanoparticles, a marked reduction in the proportion of dye leaching from the silicate sol–gel nanoprobes was observed, without significantly affecting the overall size of the nanoprobe.

In comparison to the use of a dextran ‘anchor’, the use of silica nanoparticles proved advantageous as along with ease of purification (due to the insolubility of the silica nanoparticles), the uniform particle size and shape also provides an estimate of the nanoprobes pore size, in this case they were found to be <15 nm. Using the technique described herein we have developed a novel nanoprobe capable of performing intracellular measurements of ROS and have demonstrated their qualitative evaluation in an intracellular environment. Compared to the use of free dyes, this technique is advantageous as not only are the cells protected from the cytotoxic effects of the dye, but the dye is also guarded against any cellular interferences, leading to an increased sensor lifetime. We therefore believe that this method of dye encapsulation is suited to an array of applications where improved sensor function is required. Work is currently underway within our laboratories to increase the scope of the technique with respect to the intracellular measurement of numerous analytes and the sensitivity of the technique. The production of ratiometric nanoprobes is also in progress through the co-immobilisation of a reference dye.

Acknowledgements

The authors would like to acknowledge the EPSRC and the Royal Society of Chemistry (V. H.) for the studentship funding this project.

References

1. J. T. Hancock, R. Desikan and S. J. Neill, Biochem. Soc. Trans., 2001, 29, 345–350.
2. O. Kann and R. Kovacs, Am. J. Physiol. Cell Physiol., 2007, 292, C641–C657.
3. R. Rudolf, M. Mongillo, R. Rizzuto and T. Pozzan, Nat. Rev. Mol. Cell Biol., 2003, 4, 579–586.
4. (a) I. Tokareva, S. Minko, J. H. Fendler and E. Hutter, J. Am. Chem. Soc., 2004, 126, 15950–15951; (b) C.-C. Huang, Z. Cao, W. Tan, Y.-F. Huang and H.-T. Chang, Anal. Chem., 2005, 77, 5735–5741.
5. N. G. Portney and M. Ozkan, Anal. Bioanal. Chem., 2006, 384, 620–630.
6. S. M. Buck, Y. E. L. Koo, E. Park, H. Xu, M. A. Philbert, M. Brasuel and R. Kopelman, Curr. Opin. Chem. Biol., 2004, 8, 540–546.
7. H. A. Clark, M. Hoyer, M. A. Philbert and R. Kopelman, Anal. Chem., 1999, 71, 4831–4836.
8. M. Brasuel, R. Kopelman, T. J. Miller, R. Tjalkens and M. A. Philbert, Anal. Chem., 2001, 73, 2221–2228.
9. M. Brasuel, R. Kopelman, J. W. Aylott, H. Clark, H. Xu, M. Hoyer, T. J. Miller, R. Tjalkens and M. A. Philbert, Sens. Mater., 2002, 14, 309–338.
10. M. King and R. Kopelman, Sens. Actuators, B, 2003, 90, 76–81.
11. A. Webster, S. J. Compton and J. W. Aylott, Analyst, 2005, 130, 163–170.
12. M. J. Moreno, E. Monson, R. G. Reddy, A. Rehemtulla, B. D. Ross, M. Philbert, R. J. Schneider and R. Kopelman, Sens. Actuators, B, 2003, 90, 82–89.
13. H. Xu, J. W. Aylott and R. Kopelman, Analyst, 2002, 127, 1471–1477.
14. H. Xu, J. W. Aylott, R. Kopelman, T. J. Miller and M. A. Philbert, Anal. Chem., 2001, 73, 4124–4133.
15. N. Baker, G. M. Greenway, R. A. Wheatley and C. Wiles, Analyst, 2007, 132, 104–106.
16. M. Adamczyk, C. M. Chan, J. R. Fino and P. G. Mattingly, J. Org. Chem., 2000, 65, 596–601.
17. S. Santra, P. Zhang, K. M. Wang, R. Tapec and W. H. Tan, Anal. Chem., 2001, 73, 4988–4993.
18. A. I. Vogel, A Text-Book Of Quantitative Inorganic Analysis, Including Elementary Instrumental Analysis, Longmans, London, 1961.
19. M. Z. Cabail, P. J. Lace, J. Uselding and A. A. Pacheco, J. Photochem. Photobiol., A, 2002, 152, 109–121.
20. S. L. Hempel, G. R. Buettner, Y. Q. O'Malley, D. A. Wessels and D. M. Flaherty, Free Radical Biol. Med., 1999, 27, 146–159.
21. A. Webster, P. Coupland, F. D. Houghton, H. J. Leese and J. W. Aylott, Biochem. Soc. Trans., 2007, 35, 538–543.

---

Footnote

† Electronic supplementary information (ESI) available: detailed procedures for nanoparticle derivatisation and PEBBLE production. See DOI: 10.1039/b711995j

---