Biosensors and Bioelectronics 64 (2015) 359–366
Contents lists available at ScienceDirect
Biosensors and Bioelectronics journal homepage: www.elsevier.com/locate/bios
A disposable evanescent wave fiber optic sensor coated with a molecularly imprinted polymer as a selective fluorescence probe Xuan-Anh Ton a, Victor Acha b, Paolo Bonomi a, Bernadette Tse Sum Bui a,n, Karsten Haupt a,n a b
CNRS Enzyme and Cell Engineering Laboratory, Compiègne University of Technology, Rue Roger Couttolenc, CS 60319, 60203 Compiègne Cedex, France HydrISE Unit, Institut Polytechnique LaSalle Beauvais, BP 30313, 60026 Beauvais, France
art ic l e i nf o
a b s t r a c t
Article history: Received 11 July 2014 Received in revised form 2 September 2014 Accepted 4 September 2014 Available online 8 September 2014
We have developed a disposable evanescent wave fiber optic sensor by coating a molecularly imprinted polymer (MIP) containing a fluorescent signaling group on a 4-cm long polystyrene optical waveguide. The MIP is composed of a naphthalimide-based fluorescent monomer, which shows fluorescence enhancement upon binding with carboxyl-containing molecules. The herbicide 2,4-dichlorophenoxyacetic acid and the mycotoxin citrinin were used as model analytes. The coating of the MIP was either performed ex-situ, by dip-coating the fiber with MIP particles synthesized beforehand, or in-situ by evanescent-wave photopolymerization on the fiber. The sensing element was interrogated with a fibercoupled spectrofluorimeter. The fiber optic sensor detects targets in the low nM range and exhibits specific and selective recognition over structural analogs and non-related carboxyl-containing molecules. This technology can be extended to other carboxyl-containing analytes, and to a broader spectrum of targets using different fluorescent monomers. & 2014 Elsevier B.V. All rights reserved.
Keywords: Molecularly imprinted polymer Signaling monomer Evanescent wave fiber optic sensor Fluorescence Mycotoxin
1. Introduction Molecularly imprinted polymers (MIPs) are biomimetic synthetic receptors possessing specific cavities designed for a target molecule. Produced by a templating process at the molecular level, MIPs are capable of recognizing and binding their targets with specificities and affinities comparable to those of natural receptors (Arshady and Mosbach, 1981; Alexander et al., 2006). These tailormade synthetic receptors have considerable advantages over biological receptors due to their greater chemical and physical stability, and have become an interesting alternative as recognition elements for biosensors (Haupt and Mosbach, 2000; Wulff, 1995). In particular, MIPs have been widely used for optical sensing, especially using fluorescence (Dickert and Thierer, 1996; Henry et al., 2005; Ton et al., 2012; Wolfbeis, 2005). Target binding can be monitored by directly measuring a fluorescent analyte, or, if the analyte is not fluorescent, by a competitive assay with a fluorescent analog as a probe. Alternatively, non-fluorescent analytes can be monitored if a fluorescent monomer is incorporated into the MIP that shows a change in its fluorescence properties when analyte binding occurs (Haupt and Mosbach, 2000), namely fluorescence quenching (Takeuchi et al., 2005; Wagner et al., n
Corresponding authors. Fax: þ 33 3 4420 3910. E-mail addresses: [email protected] (B. Tse Sum Bui), [email protected] (K. Haupt). http://dx.doi.org/10.1016/j.bios.2014.09.017 0956-5663/& 2014 Elsevier B.V. All rights reserved.
2013), fluorescence shift (Matsui et al., 2000) or fluorescence enhancement (Leung et al., 2001; Kubo et al., 2003; Ton et al., 2013; Wan et al., 2013). Among these, fluorescence enhancement is the most interesting, as it is more specific and less prone to false-positive results. There are only a few examples of fluorescence sensors with signaling MIPs, most of them in fiber optic format. Optical fibers are of interest as they offer many advantages such as ease of miniaturization and integration, limited loss of light even over long distances, and low cost (Marazuela and Moreno-Bondi, 2002). One example used lanthanide luminescence, combined with a MIP-based fiber optic sensor to measure the hydrolysis product of the nerve agent Soman (Jenkins et al., 1999). More recently, Nguyen et al. reported a MIP-based fiber optic sensor for the detection of cocaine. The signaling monomer was acrylamidofluorescein and the limit of detection (LOD) of the sensor was 2 μM (Nguyen et al., 2012). In these examples, the polymer coating procedure was complex, requiring multistep pretreatments of the fibers (polishing, surface activation, silanization) before the polymer could be covalently attached. Very recently, we reported the development of a new type of fluorimetric fiber optic sensor that carries a micrometer-sized MIP tip photopolymerized in situ at one end (Ton et al., 2013). No pretreatment of the fiber was necessary and the tip was fabricated in just a few seconds. The signaling MIP contained a naphthalimide fluorescent monomer (Doussineau et al., 2009; Niu et al., 2004), specially designed to exhibit fluorescence enhancement when interacting with carboxyl groups.
360
X.-A. Ton et al. / Biosensors and Bioelectronics 64 (2015) 359–366
This sensor could detect the herbicide 2,4-dichlorophenoxyacetic acid in the nM range. With the objective of developing an economic, disposable but sensitive fiber optic sensor based on a MIP, we turned towards fiber optic waveguides allowing for evanescent-wave measurements (Taitt et al., 2005; Wolfbeis, 2004). An evanescent field is generated when total internal reflection of light occurs in an optical fiber coated with a lower refractive index material. This field propagates perpendicularly to the fiber surface into the medium, decaying exponentially with distance. If a fluorescent recognition element like a MIP is coated onto the fiber, the fluorophores immediately adjacent to the surface (typically within 100–1000 nm) (Rogers et al., 1992) are excited by the evanescent wave, and part of the emitted fluorescent light is coupled back into the fiber and can be measured. This system has the advantage of a low background even in complex samples, as the molecules not bound to the MIP contribute very little to the measured signal. For example, a fiber optic immunosensor based on evanescent wave measurements was developed for the detection of the mycotoxin fumonisin B1 (FB1) (Thompson and Maragos, 1996). Anti-FB1 antibodies were covalently attached to the surface of an 800 μm core fiber, and monitoring of FB1 was performed by a displacement assay with fluorescently labeled FB1. Evanescent-wave fiber optic
biosensors have also been used for the detection of Escherichia coli in food samples, cocaine in urine, trinitrotoluene in environmental waters, ricin in water and urine, hormones in plasma, and others (Taitt et al., 2005). In the present work, we use 4-cm long injection-molded tapered polystyrene waveguides (Geng et al., 2006; Lim, 2003), combined with a MIP, as a disposable fiber optic sensor, to detect the herbicide 2,4-dichlorophenoxyacetic acid (2,4-D) and the mycotoxin citrinin (Fig. 1A). Monitoring of 2,4-D is of major concern as it is widely used and represents a threat for human health due to its endocrine disruption properties (Bus and Hammond, 2007). Citrinin is a nephrotoxic mycotoxin produced by several species of Aspergillus, Penicillium and Monascus fungi and occurs mainly in stored grains but also in other plant products (Bazin et al., 2013). The polystyrene fibers are generally coated with capture antibodies for the sensing of microorganisms and toxins in a competitive format and are incorporated in a multichannel evanescent wave fiber optic biosensor packaged in a portable instrument called RAPTOR, developed by the U.S. Naval Research Laboratory (Anderson et al., 2000; Lim, 2003). Here, we coated the polystyrene fibers with ‘plastic antibodies′, that is, 2,4D and citrinin-specific MIPs (Fig. 1B). The MIPs not only have specific recognition sites for capturing their respective template
Fig. 1. (A) Structures of compounds used in this study. (B) Schematic drawing of the polystyrene evanescent wave fiber optic waveguide coated with fluorescent MIP particles. Excitation with the fiber optic probe of a spectrofluorimeter and collection of emitted fluorescence light use the same fiber.
X.-A. Ton et al. / Biosensors and Bioelectronics 64 (2015) 359–366
but also contain a fluorescent monomer N-(2-(6-4-methylpiperazin-1-yl)-1,3-dioxo-1 H-benzo[de]isoquinolin-2(3H)-yl-ethyl)acrylamide (FIM), for sensing (Ton et al., 2013, Fig. 1). FIM is excited by the evanescent wave, and fluorescence enhancement occurs upon analyte binding to the MIP, the fluorescence intensity being proportional to the analyte concentration. This yields a robust, cheap and disposable optical sensor, suitable for real-time and onsite monitoring of environmental and biological analytes.
2. Materials and methods 2.1. Reagents and materials All chemicals and solvents were of analytical grade and purchased from Sigma-Aldrich (St-Quentin Fallavier, France), unless otherwise stated. 2,2′-azobis(2,4-dimethylvaleronitrile) (ABDV) was obtained from DuPont Chemicals (Wilmington, USA). The initiator, bis(2,4,6-trimethylbenzoyl)-phenylphosphineoxide (Irgacure 819) was generously provided by Ciba Specialty Chemicals (Saint Fons, France). 4-vinylpyridine (4-VPY) was distilled under reduced pressure before use. The cross-linker ethylene glycol dimethacrylate (EGDMA) was employed as received. Water was purified using a Milli-Q system (Millipore, Molsheim, France). The polystyrene optical waveguides were from Research International, Inc. (Monroe, Washington, USA) 2.2. Synthesis of the fluorescent monomer N-(2-(6-4methylpiperazin-1-yl)-1,3-dioxo-1H-benzo[de]isoquinolin-2(3H)-ylethyl)acrylamide The monomer FlM was synthesized as described previously (Ton et al., 2013). 2.3. Synthesis of MIPs by precipitation polymerization 2,4-D MIP. 0.025 mmol 2,4-D, 0.05 mmol FlM, 0.05 mmol 4-VPY, 0.5 mmol EGDMA and 0.016 mmol ABDV were dissolved in 3 mL of methanol/water (4/1), in a glass vial fitted with an airtight septum. The mixture was sonicated at room temperature for 5 min and purged with nitrogen for 2 min on ice. Polymerization was done overnight at 50 °C in a water-bath. The polymers were then washed with 3 rounds of methanol/acetic acid (7/3) and rinsed 3 times with methanol, before drying overnight under vacuum. The non-imprinted polymer (NIP) was prepared in the same way but in the absence of 2,4-D. Citrinin MIP. 0.02 mmol citrinin, 0.04 mmol FlM, 0.04 mmol 4-VPY, 0.4 mmol EGDMA and 0.0044 mmol ABDV were dissolved in 700 μL of methanol/chloroform (5/2), in a glass vial fitted with an airtight septum. The mixture was then sonicated at room temperature for 15 min and purged with nitrogen for 2 min on ice. Polymerization was done overnight at 40 °C in a water-bath. The polymers were then washed with 3 rounds of methanol/acetic acid (7/3) and rinsed 3 times with methanol, before drying overnight under vacuum. The non-imprinted polymer (NIP) was prepared according to the same protocol but in the absence of citrinin. 2.4. Binding studies with MIPs obtained by precipitation polymerization 2,4-D MIP. A stock solution of 500 nM of 2,4-D in methanol was prepared and stored in the dark at 20 °C. Polymer particles at a concentration of 0.5 mg/mL were suspended by sonication in methanol/water (4/1). From this stock, a polymer concentration of 5 μg/mL was pipetted into 1.5 mL polypropylene
361
microcentrifuge tubes and a volume ranging from 5 μL to 200 μL of 2,4-D solution was added so as to obtain final 2,4-D concentrations ranging from 2.5 nM to 100 nM. The final volume was adjusted to 1 mL with methanol/water (4/1). The tubes were incubated in the dark for 1 h at ambient temperature on a tube rotator. The solutions were then transferred into a quartz cuvette and fluorescence spectra were recorded using a Horiba Jobin-Yvon Fluorolog-3 spectrofluorimeter (Longjumeau, France). The excitation/emission wavelengths were set at 390/515 nm respectively. Fluorescence enhancement was calculated as follows: [(F F0)/F0) x100, with F and F0 the emitted fluorescence intensity of MIP respectively in presence and in absence of the analyte. Citrinin MIP. A stock solution of 10 μg/mL of citrinin in methanol was prepared and stored in the dark at 20 °C. The polymer particles, at a concentration of 2 mg/mL, were suspended by sonication in methanol. From this stock, a polymer concentration of 0.04 mg/mL was pipetted in 1.5 mL polypropylene microcentrifuge tubes and a volume ranging from 50 μL to 250 μL of citrinin solution was added so as to obtain final citrinin concentrations ranging from 0.5 to 2.5 μg/mL (2–10 mM). The final volume was adjusted to 1 mL with methanol. The same procedure as for the 2,4-D MIP was then followed. 2.5. Dip-coating of polystyrene optical waveguides with MIP particles, and binding studies A schematic representation describing the procedures of dipcoating, sensing, and the experimental setup is shown in Fig. SI-1, SI (Supporting information). The waveguide was immersed in a suspension of 20 mg/mL of 2,4-D MIP or 5 mg/mL of citrinin MIP in water containing 1.5% polyvinyl alcohol (PVA), in a 4-mL glass vial. The mixture was heated to 80 °C in a water-bath under gentle magnetic agitation. After 30 min, the vial was allowed to cool to room temperature under gentle agitation. The fiber was then removed from the vial and dried on air, resulting in an immobilization of MIP particles on the fiber surface, held in place by the PVA acting as a glue (insoluble in cold solvent). For binding studies, the MIP-coated optical fibers were placed in ambercolored 4-mL glass vials containing – 2,4-D at different concentrations in methanol/water (4/1), for the 2,4-D MIP (10 min incubation), or – citrinin at different concentrations in methanol, for the citrinin MIP (30 min). After incubation, the bifurcated fiber optic probe of the spectrofluorimeter was used to inject excitation light (λEX ¼410 nm) through the lens of the MIP-coated fiber optic waveguide (Fig. 1B, Fig. SI-1). λEX ¼410 nm instead of λEX ¼390 nm was chosen in order to minimize absorption by polystyrene, and fluorescence measurements were performed in the dark, with the fiber remaining immersed in the analyte solution, through the same fiber optic probe. 2.6. In-situ polymerization of MIPs on fiber optic waveguides The polystyrene fiber was immersed into a 2 mL-glass vial containing the polymerization mixture of MIP, consisting of: 0.1 mmol 2,4-D, 0.2 mmol FlM, 0.2 mmol 4-VPY, 2 mmol EGDMA, 1.5 mol% Irgacure 819 and 1.5 mL methanol/water (4/1). Photopolymerization was performed by injecting light (λ ¼410 nm) through the lens of the polystyrene fiber, using the fiber optic bundle of the spectrofluorimeter, during 20 min at room temperature and protected from outside light, resulting in a MIP-coated fiber optic waveguide. After polymerization, the MIP-coated fiber was washed three times by immersion during 15 min in a solution of ethanol/acetic acid (4/1), and rinsed three times by immersion
362
X.-A. Ton et al. / Biosensors and Bioelectronics 64 (2015) 359–366
during 15 min in methanol. Binding studies were performed as described in Section 2.5.
3. Results and discussion 3.1. Synthesis and characterization of the 2,4-D MIP Prior to the preparation of the fluorescent 2,4-D imprinted polymers, the signaling monomer FlM (Ton et al., 2013), was characterized and titrated with increasing concentrations of 2,4D (2.5–100 nM in methanol/water (4/1)), to verify whether fluorescence enhancement and thus the interaction was taking place. The maximum excitation and emission wavelengths of FlM were 390 nm and 515 nm, respectively. The fluorescence enhancement of FlM was observed to be concentration-dependent and increased with increasing concentrations of 2,4-D (Fig. SI-2, SI), confirming that FlM could be used as a functional monomer. The mechanism of the fluorescence enhancement of FIM can be explained as follows: the piperazinyl nitrogen atom of the monomer interacts with the carboxylic acid group of the analyte by acid-base ion-pair formation, which decreases the ability of the nitrogen to donate an electron, thus blocking the photoinduced electron transfer (PET) responsible for the quenching of fluorescence. Consequently, fluorescence enhancement is observed (De Silva et al., 1997). FlM was then incorporated into a 2,4-D MIP formulation adapted from a previous publication (Haupt et al., 1998), where it replaces partially the monomer 4-VPY. The MIP was synthesized by precipitation polymerization, and nanoparticles of size 300 nm were obtained, as deduced from scanning electron microscopy (SEM) (Fig. SI-3, SI). The binding properties of the polymers were evaluated by spectrofluorimetric titrations with 2,4-D. For this study, suspensions of 5 μg/mL of MIP and NIP were incubated with 2.5–100 nM of 2,4-D in methanol/water (4/1). As shown in Fig. 2A, the fluorescence response of the MIP increased with increasing concentrations of 2,4-D; the limit of quantification (LOQ) was 2.5 nM. Fig. 2B shows that the MIP exhibited a stronger fluorescence response than the NIP. The concentration-dependence of the MIP fluorescence enhancement was then used to determine the affinity of the MIP for 2,4-D. The dissociation constant KD was calculated as follows (Feng et al., 1998): the change in the MIP fluorescence intensity is attributed to the formation of the complex between FlM and 2,4-D described by the equilibrium (1):
FlM+n2, 4− D↔ complex[FlM−2, 4−D]
(1)
The association constant of the complex is thus defined as:
KA = [complex]/([FlM]2,4−D]n )
(2)
If [FlM]0 is the total amount of fluorescent monomer, then [FlM]0 ¼[complex] þ[FlM], with [FlM] being the unbound amount of fluorescent monomer. We also define F0 and F as the emitted fluorescence intensity of FlM respectively in the absence and in the presence of 2,4-D. The change in fluorescence intensity (F F0) is contributed by the formation of [complex]. This gives the relationship between the fluorescence intensity and concentrations of FlM that is [complex]/[FlM] ¼(F F0)/F0, with the condition [complex] o o[FlM]0, so we can attribute F0 to [FlM]. Finally this leads to the following equation:
log((F − F0)/F0) = logKA + n log[2, 4−D]
Fig. 2. (A) Fluorescence spectra (λEX/EM ¼390/515 nm) of MIP particles (5 μg/mL) after incubation with 2,4-D in methanol/water (4/1). (B) Fluorescence response of MIP and NIP particles (5 μg/mL in methanol/water (4/1)) with increasing concentrations of 2,4-D. Measurements were done in triplicate and the error bars represent SD. (C) Fluorescence enhancement response of the MIP particles (5 μg /mL) after incubation with 25 nM 2,4-D, the analogs 2,4-D-OMe and POAc and the non-related compounds Z-L-Phe and GlcA, in methanol/water (4/1).
(3)
Thus, according to (3), the association constant KA and the ratio 2,4-D:FlM in the complex can be deduced. Fig. SI-3 shows the plot of log ((F F0)/F0) vs. log[2,4-D]. Two distinct linear portions, indicating two types of binding sites in the MIP: high-affinity sites
and low-affinity sites are observed. High-affinity sites are the specific imprinted sites while the low-affinity sites are source of non-specific binding. Values of KA and n were calculated from the intercepts and the slopes. For high-affinity sites, KA was found to be 8.5×106 M 1, giving a dissociation constant KD ¼0.12 μM, and n¼ 0.92, which means an approximate ratio FlM:2,4-D ¼ 1:1. For low-affinity sites, the values were found to be KA ¼510 M 1 (KD ¼2 mM) and n ¼0.41, resulting in an approximate ratio FlM:2,4D¼ 2:1. At low concentrations of 2,4-D (o10 nM), only highaffinity sites are occupied. For high concentrations of 2,4-D, the analyte can also bind to low-affinity sites, which explains the increasing non-specific binding for the NIP. To conclude, the MIP displayed a very high affinity for 2,4-D, which is consistent with the low limit of quantification of 2.5 nM, far below the maximum concentration of 2,4-D in drinking water recommended by the
X.-A. Ton et al. / Biosensors and Bioelectronics 64 (2015) 359–366
363
World Health Organization (135 nM; World Health Organization, 2011). To assess the selectivity of the MIP, fluorescence binding assays were performed to compare the binding of 2,4-D (pKa of 5.09, methanol/water (4/1)), with that of two structurally related compounds, 2,4-D methyl ester (2,4-D-OMe), whose carboxyl group is blocked, and phenoxyacetic acid (POAc, pKa ¼5.17, methanol/water (4/1)), lacking the two ring chlorines; and with two non-related compounds containing a carboxyl group, N-carbobenzyloxy-L-phenylalanine (Z-L-Phe, pKa ¼5.06, methanol/water (4/1)) and β-D-glucuronic acid (GlcA, pKa ¼ 5.22, methanol/water (4/1)) (Fig. 1A). As described above, suspensions of 5 μg/mL of MIP particles were incubated with 25 nM of each analyte. For the four molecules tested, no or little fluorescence enhancement (o 20%) was observed (Fig. 2C), much less than with 2,4-D. These results confirm the selectivity of our MIP. 3.2. Coating fiber optic waveguides with MIP particles and sensing of 2,4-D The MIP will now be integrated into a fiber optic device. For this, we used 4-cm long injection-molded tapered polystyrene optical fibers with a lens on one end and a mirror at the other end. The light is injected and collected through the lens (Fig. 1B). In our study, MIP particles were coated onto the fiber by using PVA as a glue (Fig. 3A) (Surugiu et al., 2001). PVA dissolves in hot water for the preparation of the polymer suspension, but after drying does not dissolve under the conditions of the assay. SEM images (Fig. 3C) show a rather homogeneous coating of the MIP particles on the fiber. Excitation was performed through the fiber optic bundle of the spectrofluorimeter and fluorescence emission of the fluorescent MIP particles was collected via the same fiber bundle. Fluorescence spectra were recorded after 10 min incubation in 2,4D solution. Concentration-dependence of the MIP fluorescence response upon binding to 2,4-D is observed, the larger fluorescence enhancement with the MIP compared to the NIP indicating specific binding (Fig. 4A). The MIP-coated fiber optic device exhibited a LOQ of 1 nM, indicating a very high sensitivity of our
Fig. 4. (A) Fluorescence spectra (λEX ¼ 410 nm) from the MIP-coated fiber optic waveguide after incubation with 2,4-D (1–50 nM) in methanol/water (4/1). Inset: fluorescence response of MIP and NIP-coated fiber optic waveguides as a function of 2,4-D concentration. (B) Fluorescence enhancement response of the MIP-coated optical fiber after incubation with 5 nM each of 2,4-D, 2,4-D-OMe, POAc, a mixture of Z-L-Phe and GlcA.
sensor. The results obtained with the fiber were congruent with those obtained with the MIP in batch. To assess the selectivity of the sensor, the MIP-coated fiber was incubated with related and unrelated molecules (see Fig. 1 for structures): 5 nM 2,4-D-OMe,
Fig. 3. (A) Photograph of the polystyrene evanescent wave fiber optic waveguide coated with fluorescent 2,4-D MIP particles. (B) Excitation (λEX ¼ 410 nm) with the fiber optic bundle of the Horiba Fluorolog spectrofluorimeter marked by a red arrow, and fluorescence emission (λEM ¼515 nm) of the fluorescent MIP particles. (C) and (D) Scanning electron microscope images of the surface of the 2,4-D-MIP-coated optical fiber. The scale bars correspond respectively to 200 μm and to 2 μm. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
364
X.-A. Ton et al. / Biosensors and Bioelectronics 64 (2015) 359–366
5 nM POAc or a mixture of 5 nM of Z-L-Phe and 5 nM of GlcA in methanol/water (4/1) (Fig. 4B). Smaller fluorescence enhancement (o25%) is observed compared to the one obtained with 2,4-D (67%), which confirms that the sensor is selective. 3.3. Sensing of the mycotoxin citrinin In order to test the versatility of our system, a MIP for another target molecule, the mycotoxin citrinin (Fig. 1A), was synthesized and applied to the sensor. At the moment, no maximum residue limit for citrinin has been fixed by the European Union but according to a recent report, the characterization of the risk of citrinin as a food contaminant is based on the estimate of the citrinin concentrations in grains and grains-derived products that would result in an exposure equal to the level of no concern for nephrotoxicity, that is 19–100 μg/kg for average consumers (EFSA Panel on Contaminants in the Food Chain, 2012). Citrinin possesses a carboxyl group and is likely to interact with the signaling monomer FlM. Prior to the preparation of the fluorescent citrinin-imprinted polymer, the fluorescence enhancement of FlM with increasing concentrations of citrinin was verified by spectrofluorimetric titrations. The fluorescence enhancement of FlM was observed to be concentration-dependent and increased with increasing concentrations of citrinin (Fig. SI-4, SI). The signal enhancement of FlM was observed for concentrations Z25 ng/mL (0.1 μM) of citrinin.
Citrinin MIP particles were then synthesized by precipitation polymerization. A similar MIP formulation as for 2,4-D was employed, containing FIM, 4-VPY as co-functional monomer and EGDMA as cross-linker in a mixture of methanol/chloroform (5/2) as porogen. Citrinin MIP nanoparticles of a size between 150 and 500 nm were obtained (Fig. SI-5, SI). The binding properties of the citrinin MIP were evaluated by spectrofluorimetric titrations with citrinin, as described for the 2,4-D MIP. Fluorescence enhancement was observed for concentrations Z0.5 μg/mL (2 μM) (Fig. 5A). It has to be noted that the enhancement was due to the complex citrinin FIM and not to the intrinsic fluorescence of citrinin (λEM ¼500 nm) (Zhou et al., 2012), which accounted for only 2% of the total fluorescence intensity observed. To assess the selectivity of this MIP, it was incubated with two non-related compounds possessing carboxyl groups, namely 2,4-D and the mycotoxin fumonisin B1 (FB1). Fig. SI-6, SI) shows that a much lower fluorescence enhancement was observed for 2,4-D and FB1, as compared to citrinin. These results confirm the selectivity of the sensor for citrinin. A citrinin-MIP for the detection of citrinin has already been reported (Guo et al., 2010). It was prepared using a different synthesis protocol and employed for the solid-phase extraction of citrinin from rice, followed by quantification with HPLC. The imprinting factor, ratio of bound citrinin of MIP upon bound citrinin of NIP, as deduced from their chromatograms, was 2.3 but no selectivity studies were reported. Our MIP and NIP particles were then coated on the polystyrene waveguide (see Fig. 5B for an SEM image of the coated fiber). For
Fig.5. (A) (Left) Fluorescence spectra (λEX ¼390 nm; λEM ¼515 nm) of MIP particles, (right) fluorescence enhancement responses of citrinin MIP and NIP particles (0.04 mg/mL) after incubation with increasing concentrations of citrinin in methanol. Measurements were done in triplicate, the error bars represent SD. (B) SEM image of the citrinin MIP-coated fiber. Scale bar: 200 μm. (C) Fluorescence enhancement (λEX ¼ 410 nm; λEM ¼ 515 nm) of the MIP and NIP-coated fibers after incubation with 10 μM (2.5 mg/mL) citrinin, and of the MIP coated fiber after incubation with 10 μM of the non-related compounds, 2,4-D and FB1, in methanol.
X.-A. Ton et al. / Biosensors and Bioelectronics 64 (2015) 359–366
sensing experiments, the fibers were incubated with 2.5 μg/mL (10 μM) of citrinin in methanol for 30 min. Fig. 5C shows that the binding of citrinin to the MIP-coated fiber produced a larger signal than to the NIP-coated fiber. The selectivity was verified by incubation with 10 μM 2,4-D and 10 μM FB1. The fluorescence responses were 3 times lower as compared to those obtained with citrinin. It is worth noting that the sensing is done in methanol, a solvent currently used for the extraction of mycotoxins from food matrices (Kralj Cigić and Prosen, 2009). 3.4. In-situ MIP photopolymerization by evanescent waves on the polystyrene waveguide Having demonstrated that evanescent-wave fiber-optic sensing with a fluorescent signaling MIP is feasible, we wanted to know whether the evanescent field could be used not only for detection but also to directly photopolymerize the MIP in-situ on the fiber. In this way, the fabrication of the sensor would be much faster and more straightforward. In an earlier publication, we showed that polymer objects can be synthesized by evanescent wave-
365
photopolymerization using a laser beam reflected on a flat surface (Fuchs et al., 2011). In the present work, the actinic light source came from the optical fiber probe of the spectrofluorimeter, and polymerization was initiated at 410 nm with an irradiation time of 20 min using the spectrometer light source. The same 2,4-D MIP formulation as described above was used except for the amount of solvent and the initiator (the photoinitiator Irgacure 819 was used instead of ABDV). Irgacure 819 has the advantage to allow for very fast polymerization, as it undergoes fast photolysis under light exposure, resulting in very reactive radicals (Decker et al., 2001). Moreover, it is very reactive towards acrylates and methacrylates (Szablan et al., 2007). As shown in Fig. 6B, a homogeneous polymer coating was obtained that covers the fiber completely. After washing, the MIP-coated fiber was incubated for 15 min with 25 nM 2,4-D in methanol/water (4/1), which yielded a fluorescence enhancement of around 66% (Fig. 6C). The binding was specific, as a lower enhancement was observed for the corresponding NIP-coated fiber. For the moment, the MIP-coated device obtained by in-situ polymerization is somehow less sensitive than the one obtained with the MIP particles synthesized beforehand, and some additional optimization will be necessary. Nevertheless, these preliminary results show that sensitive detection is possible (nM range, below the maximal concentration of 135 nM 2,4-D in drinking water recommended by the WHO) (World Health Organization, 2011).
4. Conclusions
Fig. 6. (A) Photograph and (B) scanning electron microscope image of a waveguide coated with MIP by evanescent wave polymerization. (C) Fluorescence enhancement (λEX ¼ 410 nm) of MIP and NIP-coated waveguides after incubation with 2,4-D in methanol/water (4/1).
We have developed a chemical sensor based on a disposable fiber-optical waveguide coated with a MIP, and on evanescentwave spectroscopy. The MIP was coated on the waveguide either in-situ by direct polymerization using the evanescent wave or exsitu by dip-coating the waveguide with the MIP particles synthesized beforehand. Sensing within the MIP was achieved with a naphthalimide probe, which exhibits fluorescence enhancement when interacting with carboxyl groups of the analytes. Thus the herbicide 2,4-D and the mycotoxin citrinin could be selectively determined in the nM and μM range respectively. Though the MIP sensor displays a certain degree of crossreactivity for related molecules and may therefore not be able to discriminate the target herbicide or mycotoxin in a very complex mixture, as well as it would with SPE followed by LC/MS–MS, it is nevertheless selective enough and can undoubtedly serve as an initial, rapid, sensitive and on-site screening tool, to detect and monitor their presence. As shown in this work, the MIP has great potential for real applications. Frequently monitored mycotoxins that co-exist in grains and grain-derived products are ochratoxin A, fumonisins, zearalenone, trichothecenes (e.g. deoxynivalenol), T2-toxin and aflatoxins (Lerda, 2010). The latter four cannot be detected using a MIP based on the FIM monomer because they do not possess a carboxyl group. For ochratoxin A and fumonisins, which bear a carboxyl group, it should be possible to synthesize MIPs using the same protocol as for citrinin, and to detect them selectively. Thus, an array of waveguides with different MIPs could be a solution for more complex samples containing different target mycotoxins. In addition, we believe that this sensor can also be applied for the detection of other carboxyl-containing analytes, for example in the biomedical field. Using different fluorescent monomers, analytes with other functional groups should also be detectable. To further increase the sensitivity, fine-tuning the thickness of the MIP layer and the inclusion of gold nanoparticles aiding to reflect more of the fluorescence light back into the waveguide (Ton et al., 2013) may be possible solutions.. We believe that this disposable portable device can be useful for cheap and
366
X.-A. Ton et al. / Biosensors and Bioelectronics 64 (2015) 359–366
rapid on-site monitoring of environmental, food and biomedical analytes.
Acknowledgments The authors gratefully acknowledge financial support by the European Regional Development Fund (ERDF), and by the Regional Council of Picardy (Projects CBI-PEM, NANOTOXISCREEN, and cofunding of equipment under CPER 2007–2013).
Appendix A. Supplementary information Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.bios.2014.09.017.
References Alexander, C., Andersson, H.S., Andersson, L.I., Ansell, R.J., Kirsch, N., Nicholls, I.A., O’Mahony, J., Whitcombe, M.J., 2006. J. Mol. Recognit. 19, 106–180. Anderson, G.P., King, K.D., Gaffney, K.L., Johnson, L.H., 2000. Biosens. Bioelectron. 14, 771–777. Arshady, R., Mosbach, K., 1981. Makromol. Chem. 182, 687–692. Bazin, I., Faucet-Marquis, V., Monje, M.-C., El Khoury, M., Marty, J.-L., PfohlLeszkowicz, A., 2013. Toxins 5, 2324–2340. Bus, J.S., Hammond, L.E., 2007. Crop Prot. 26, 266–269. De Silva, A.P., Gunaratne, H.Q.N., Gunnlaugsson, T., Huxley, A.J.M., McCoy, C.P., Rademacher, J.T., Rice, T.E., 1997. Chem. Rev. 28, 1515–1566. Decker, C., Zahouily, K., Decker, D., Nguyen, T., Viet, T., 2001. Polymer 42, 7551–7560. Dickert, F.L., Thierer, S., 1996. Adv. Mater. 8, 987–990. Doussineau, T., Trupp, S., Mohr, G.J., 2009. J. Colloid Interface Sci. 339, 266–270. EFSA Panel on Contaminants in the Food Chain (CONTAM), 2012. EFSA J. 10 (3), 2605–2686. Feng, X.Z., Lin, Z., Yang, L.J., Wang, C., Bai, C., 1998. Talanta 47, 1223–1229.
Fuchs, Y., Linares, A.V., Mayes, A.G., Haupt, K., Soppera, O., 2011. Chem. Mater. 23, 3645–3651. Geng, T., Uknalis, J., Tu, S., Bhunia, A.K., 2006. Sensors 6, 796–807. Guo, B.-Y., Wang, S., Ren, B., Li, X., Qin, F., Li, J., 2010. J. Sep. Sci. 33, 1156–1160. Haupt, K., Dzgoev, A., Mosbach, K., 1998. Anal. Chem. 70, 628–631. Haupt, K., Mosbach, K., 2000. Chem. Rev. 100, 2495–2504. Henry, O.Y.F., Cullen, D.C., Piletsky, S.A., 2005. Anal. Bioanal. Chem. 382, 947–956. Jenkins, A.L., Uy, O.M., Murray, G.M., 1999. Anal. Chem. 71, 373–378. Kralj Cigić, I., Prosen, H., 2009. Int. J. Mol. Sci. 10, 62–115. Kubo, H., Nariai, H., Takeuchi, T., 2003. Chem. Commun., 2792–2793. Leung, M.K.-P., Chow, C.-F., Lam, M.H.-W., 2001. J. Mater. Chem. 11, 2985–2991. Lerda, D., JRC Technical Notes: Mycotoxins factsheet, 3rd ed., 2010. Available online: 〈https://irmm.jrc.ec.europa.eu/EURLs/eurl_mycotoxins/Documents/JRC% 2060040_Mycotoxin%20factsheet_3rd%20edition.pdf〉. Lim, D.V., 2003. Proc. IEEE 91, 902–907. Marazuela, M.D., Moreno-Bondi, M.C., 2002. Anal. Bioanal. Chem. 372, 664–682. Matsui, J., Kubo, H., Takeuchi, T., 2000. Anal. Chem. 72, 3286–3290. Nguyen, T.H., Hardwick, S.A., Sun, T., Grattan, K.T.V., 2012. IEEE Sens. J. 12, 255–260. Niu, C.G., Zeng, G.M., Chen, L.X., Shen, G.L., Yu, R.Q., 2004. Analyst 129, 20–24. Rogers, K.R., Anis, N.A., Valdes, J.J., Eldefrawi, M.E., 1992. In: Mathewson, P.R., Finley, J.W. (Eds.), Biosensor Design and Application. American Chemical Society, Washington D.C., USA, pp. 165–173. Surugiu, I., Danielsson, B., Ye, L., Mosbach, K., Haupt, K., 2001. Anal. Chem. 73, 487–491. Szablan, Z., Junkers, T., Koo, S., Lovestead, T., Davis, T., Stenzel, M., Barner-Kowollik, C., 2007. Macromolecules 40, 6820–6833. Taitt, C.R., Anderson, G.P., Ligler, F.S., 2005. Biosens. Bioelectron. 20, 2470–2487. Takeuchi, T., Mukawa, T., Shinmori, H., 2005. Chem. Rec. 5, 263–275. Thompson, V.S., Maragos, C.M., 1996. J. Agric. Food Chem. 44, 1041–1046. Ton, X.A., Acha, V., Haupt, K., Tse Sum Bui, B., 2012. Biosens. Bioelectron. 36, 22–28. Ton, X.A., Tse Sum Bui, B., Resmini, M., Bonomi, P., Dika, I., Soppera, O., Haupt, K., 2013. Angew. Chem. Int. Ed. 52, 8317–8321. Wagner, R., Wan, W., Biyikal, M., Benito-Peñ a, E., Moreno-Bondi, M.C., Lazraq, I., Rurack, K., Sellergren, B., 2013. J. Org. Chem. 78, 1377-–11389. Wan, W., Biyikal, M., Wagner, R., Sellergren, B., Rurack, K., 2013. Angew. Chem. Int. Ed. 52, 7023–7027. Wolfbeis, O.S., 2004. Anal. Chem. 76, 3269–3284. Wolfbeis, O.S., 2005. J. Mater. Chem. 15, 2657–2669. World Health Organization Guidelines for Drinking Water Quality 4th ed. 2011 (Available online). Wulff, G., 1995. Angew. Chem. Int. Ed. 34, 1812–1832. Zhou, Y., Chen, J., Dong, L., Lu, L., Chen, F., Hub, D., Wanga, X., 2012. J. Lumin. 132, 1437–1445.
Contents lists available at ScienceDirect
Biosensors and Bioelectronics journal homepage: www.elsevier.com/locate/bios
A disposable evanescent wave fiber optic sensor coated with a molecularly imprinted polymer as a selective fluorescence probe Xuan-Anh Ton a, Victor Acha b, Paolo Bonomi a, Bernadette Tse Sum Bui a,n, Karsten Haupt a,n a b
CNRS Enzyme and Cell Engineering Laboratory, Compiègne University of Technology, Rue Roger Couttolenc, CS 60319, 60203 Compiègne Cedex, France HydrISE Unit, Institut Polytechnique LaSalle Beauvais, BP 30313, 60026 Beauvais, France
art ic l e i nf o
a b s t r a c t
Article history: Received 11 July 2014 Received in revised form 2 September 2014 Accepted 4 September 2014 Available online 8 September 2014
We have developed a disposable evanescent wave fiber optic sensor by coating a molecularly imprinted polymer (MIP) containing a fluorescent signaling group on a 4-cm long polystyrene optical waveguide. The MIP is composed of a naphthalimide-based fluorescent monomer, which shows fluorescence enhancement upon binding with carboxyl-containing molecules. The herbicide 2,4-dichlorophenoxyacetic acid and the mycotoxin citrinin were used as model analytes. The coating of the MIP was either performed ex-situ, by dip-coating the fiber with MIP particles synthesized beforehand, or in-situ by evanescent-wave photopolymerization on the fiber. The sensing element was interrogated with a fibercoupled spectrofluorimeter. The fiber optic sensor detects targets in the low nM range and exhibits specific and selective recognition over structural analogs and non-related carboxyl-containing molecules. This technology can be extended to other carboxyl-containing analytes, and to a broader spectrum of targets using different fluorescent monomers. & 2014 Elsevier B.V. All rights reserved.
Keywords: Molecularly imprinted polymer Signaling monomer Evanescent wave fiber optic sensor Fluorescence Mycotoxin
1. Introduction Molecularly imprinted polymers (MIPs) are biomimetic synthetic receptors possessing specific cavities designed for a target molecule. Produced by a templating process at the molecular level, MIPs are capable of recognizing and binding their targets with specificities and affinities comparable to those of natural receptors (Arshady and Mosbach, 1981; Alexander et al., 2006). These tailormade synthetic receptors have considerable advantages over biological receptors due to their greater chemical and physical stability, and have become an interesting alternative as recognition elements for biosensors (Haupt and Mosbach, 2000; Wulff, 1995). In particular, MIPs have been widely used for optical sensing, especially using fluorescence (Dickert and Thierer, 1996; Henry et al., 2005; Ton et al., 2012; Wolfbeis, 2005). Target binding can be monitored by directly measuring a fluorescent analyte, or, if the analyte is not fluorescent, by a competitive assay with a fluorescent analog as a probe. Alternatively, non-fluorescent analytes can be monitored if a fluorescent monomer is incorporated into the MIP that shows a change in its fluorescence properties when analyte binding occurs (Haupt and Mosbach, 2000), namely fluorescence quenching (Takeuchi et al., 2005; Wagner et al., n
Corresponding authors. Fax: þ 33 3 4420 3910. E-mail addresses: [email protected] (B. Tse Sum Bui), [email protected] (K. Haupt). http://dx.doi.org/10.1016/j.bios.2014.09.017 0956-5663/& 2014 Elsevier B.V. All rights reserved.
2013), fluorescence shift (Matsui et al., 2000) or fluorescence enhancement (Leung et al., 2001; Kubo et al., 2003; Ton et al., 2013; Wan et al., 2013). Among these, fluorescence enhancement is the most interesting, as it is more specific and less prone to false-positive results. There are only a few examples of fluorescence sensors with signaling MIPs, most of them in fiber optic format. Optical fibers are of interest as they offer many advantages such as ease of miniaturization and integration, limited loss of light even over long distances, and low cost (Marazuela and Moreno-Bondi, 2002). One example used lanthanide luminescence, combined with a MIP-based fiber optic sensor to measure the hydrolysis product of the nerve agent Soman (Jenkins et al., 1999). More recently, Nguyen et al. reported a MIP-based fiber optic sensor for the detection of cocaine. The signaling monomer was acrylamidofluorescein and the limit of detection (LOD) of the sensor was 2 μM (Nguyen et al., 2012). In these examples, the polymer coating procedure was complex, requiring multistep pretreatments of the fibers (polishing, surface activation, silanization) before the polymer could be covalently attached. Very recently, we reported the development of a new type of fluorimetric fiber optic sensor that carries a micrometer-sized MIP tip photopolymerized in situ at one end (Ton et al., 2013). No pretreatment of the fiber was necessary and the tip was fabricated in just a few seconds. The signaling MIP contained a naphthalimide fluorescent monomer (Doussineau et al., 2009; Niu et al., 2004), specially designed to exhibit fluorescence enhancement when interacting with carboxyl groups.
360
X.-A. Ton et al. / Biosensors and Bioelectronics 64 (2015) 359–366
This sensor could detect the herbicide 2,4-dichlorophenoxyacetic acid in the nM range. With the objective of developing an economic, disposable but sensitive fiber optic sensor based on a MIP, we turned towards fiber optic waveguides allowing for evanescent-wave measurements (Taitt et al., 2005; Wolfbeis, 2004). An evanescent field is generated when total internal reflection of light occurs in an optical fiber coated with a lower refractive index material. This field propagates perpendicularly to the fiber surface into the medium, decaying exponentially with distance. If a fluorescent recognition element like a MIP is coated onto the fiber, the fluorophores immediately adjacent to the surface (typically within 100–1000 nm) (Rogers et al., 1992) are excited by the evanescent wave, and part of the emitted fluorescent light is coupled back into the fiber and can be measured. This system has the advantage of a low background even in complex samples, as the molecules not bound to the MIP contribute very little to the measured signal. For example, a fiber optic immunosensor based on evanescent wave measurements was developed for the detection of the mycotoxin fumonisin B1 (FB1) (Thompson and Maragos, 1996). Anti-FB1 antibodies were covalently attached to the surface of an 800 μm core fiber, and monitoring of FB1 was performed by a displacement assay with fluorescently labeled FB1. Evanescent-wave fiber optic
biosensors have also been used for the detection of Escherichia coli in food samples, cocaine in urine, trinitrotoluene in environmental waters, ricin in water and urine, hormones in plasma, and others (Taitt et al., 2005). In the present work, we use 4-cm long injection-molded tapered polystyrene waveguides (Geng et al., 2006; Lim, 2003), combined with a MIP, as a disposable fiber optic sensor, to detect the herbicide 2,4-dichlorophenoxyacetic acid (2,4-D) and the mycotoxin citrinin (Fig. 1A). Monitoring of 2,4-D is of major concern as it is widely used and represents a threat for human health due to its endocrine disruption properties (Bus and Hammond, 2007). Citrinin is a nephrotoxic mycotoxin produced by several species of Aspergillus, Penicillium and Monascus fungi and occurs mainly in stored grains but also in other plant products (Bazin et al., 2013). The polystyrene fibers are generally coated with capture antibodies for the sensing of microorganisms and toxins in a competitive format and are incorporated in a multichannel evanescent wave fiber optic biosensor packaged in a portable instrument called RAPTOR, developed by the U.S. Naval Research Laboratory (Anderson et al., 2000; Lim, 2003). Here, we coated the polystyrene fibers with ‘plastic antibodies′, that is, 2,4D and citrinin-specific MIPs (Fig. 1B). The MIPs not only have specific recognition sites for capturing their respective template
Fig. 1. (A) Structures of compounds used in this study. (B) Schematic drawing of the polystyrene evanescent wave fiber optic waveguide coated with fluorescent MIP particles. Excitation with the fiber optic probe of a spectrofluorimeter and collection of emitted fluorescence light use the same fiber.
X.-A. Ton et al. / Biosensors and Bioelectronics 64 (2015) 359–366
but also contain a fluorescent monomer N-(2-(6-4-methylpiperazin-1-yl)-1,3-dioxo-1 H-benzo[de]isoquinolin-2(3H)-yl-ethyl)acrylamide (FIM), for sensing (Ton et al., 2013, Fig. 1). FIM is excited by the evanescent wave, and fluorescence enhancement occurs upon analyte binding to the MIP, the fluorescence intensity being proportional to the analyte concentration. This yields a robust, cheap and disposable optical sensor, suitable for real-time and onsite monitoring of environmental and biological analytes.
2. Materials and methods 2.1. Reagents and materials All chemicals and solvents were of analytical grade and purchased from Sigma-Aldrich (St-Quentin Fallavier, France), unless otherwise stated. 2,2′-azobis(2,4-dimethylvaleronitrile) (ABDV) was obtained from DuPont Chemicals (Wilmington, USA). The initiator, bis(2,4,6-trimethylbenzoyl)-phenylphosphineoxide (Irgacure 819) was generously provided by Ciba Specialty Chemicals (Saint Fons, France). 4-vinylpyridine (4-VPY) was distilled under reduced pressure before use. The cross-linker ethylene glycol dimethacrylate (EGDMA) was employed as received. Water was purified using a Milli-Q system (Millipore, Molsheim, France). The polystyrene optical waveguides were from Research International, Inc. (Monroe, Washington, USA) 2.2. Synthesis of the fluorescent monomer N-(2-(6-4methylpiperazin-1-yl)-1,3-dioxo-1H-benzo[de]isoquinolin-2(3H)-ylethyl)acrylamide The monomer FlM was synthesized as described previously (Ton et al., 2013). 2.3. Synthesis of MIPs by precipitation polymerization 2,4-D MIP. 0.025 mmol 2,4-D, 0.05 mmol FlM, 0.05 mmol 4-VPY, 0.5 mmol EGDMA and 0.016 mmol ABDV were dissolved in 3 mL of methanol/water (4/1), in a glass vial fitted with an airtight septum. The mixture was sonicated at room temperature for 5 min and purged with nitrogen for 2 min on ice. Polymerization was done overnight at 50 °C in a water-bath. The polymers were then washed with 3 rounds of methanol/acetic acid (7/3) and rinsed 3 times with methanol, before drying overnight under vacuum. The non-imprinted polymer (NIP) was prepared in the same way but in the absence of 2,4-D. Citrinin MIP. 0.02 mmol citrinin, 0.04 mmol FlM, 0.04 mmol 4-VPY, 0.4 mmol EGDMA and 0.0044 mmol ABDV were dissolved in 700 μL of methanol/chloroform (5/2), in a glass vial fitted with an airtight septum. The mixture was then sonicated at room temperature for 15 min and purged with nitrogen for 2 min on ice. Polymerization was done overnight at 40 °C in a water-bath. The polymers were then washed with 3 rounds of methanol/acetic acid (7/3) and rinsed 3 times with methanol, before drying overnight under vacuum. The non-imprinted polymer (NIP) was prepared according to the same protocol but in the absence of citrinin. 2.4. Binding studies with MIPs obtained by precipitation polymerization 2,4-D MIP. A stock solution of 500 nM of 2,4-D in methanol was prepared and stored in the dark at 20 °C. Polymer particles at a concentration of 0.5 mg/mL were suspended by sonication in methanol/water (4/1). From this stock, a polymer concentration of 5 μg/mL was pipetted into 1.5 mL polypropylene
361
microcentrifuge tubes and a volume ranging from 5 μL to 200 μL of 2,4-D solution was added so as to obtain final 2,4-D concentrations ranging from 2.5 nM to 100 nM. The final volume was adjusted to 1 mL with methanol/water (4/1). The tubes were incubated in the dark for 1 h at ambient temperature on a tube rotator. The solutions were then transferred into a quartz cuvette and fluorescence spectra were recorded using a Horiba Jobin-Yvon Fluorolog-3 spectrofluorimeter (Longjumeau, France). The excitation/emission wavelengths were set at 390/515 nm respectively. Fluorescence enhancement was calculated as follows: [(F F0)/F0) x100, with F and F0 the emitted fluorescence intensity of MIP respectively in presence and in absence of the analyte. Citrinin MIP. A stock solution of 10 μg/mL of citrinin in methanol was prepared and stored in the dark at 20 °C. The polymer particles, at a concentration of 2 mg/mL, were suspended by sonication in methanol. From this stock, a polymer concentration of 0.04 mg/mL was pipetted in 1.5 mL polypropylene microcentrifuge tubes and a volume ranging from 50 μL to 250 μL of citrinin solution was added so as to obtain final citrinin concentrations ranging from 0.5 to 2.5 μg/mL (2–10 mM). The final volume was adjusted to 1 mL with methanol. The same procedure as for the 2,4-D MIP was then followed. 2.5. Dip-coating of polystyrene optical waveguides with MIP particles, and binding studies A schematic representation describing the procedures of dipcoating, sensing, and the experimental setup is shown in Fig. SI-1, SI (Supporting information). The waveguide was immersed in a suspension of 20 mg/mL of 2,4-D MIP or 5 mg/mL of citrinin MIP in water containing 1.5% polyvinyl alcohol (PVA), in a 4-mL glass vial. The mixture was heated to 80 °C in a water-bath under gentle magnetic agitation. After 30 min, the vial was allowed to cool to room temperature under gentle agitation. The fiber was then removed from the vial and dried on air, resulting in an immobilization of MIP particles on the fiber surface, held in place by the PVA acting as a glue (insoluble in cold solvent). For binding studies, the MIP-coated optical fibers were placed in ambercolored 4-mL glass vials containing – 2,4-D at different concentrations in methanol/water (4/1), for the 2,4-D MIP (10 min incubation), or – citrinin at different concentrations in methanol, for the citrinin MIP (30 min). After incubation, the bifurcated fiber optic probe of the spectrofluorimeter was used to inject excitation light (λEX ¼410 nm) through the lens of the MIP-coated fiber optic waveguide (Fig. 1B, Fig. SI-1). λEX ¼410 nm instead of λEX ¼390 nm was chosen in order to minimize absorption by polystyrene, and fluorescence measurements were performed in the dark, with the fiber remaining immersed in the analyte solution, through the same fiber optic probe. 2.6. In-situ polymerization of MIPs on fiber optic waveguides The polystyrene fiber was immersed into a 2 mL-glass vial containing the polymerization mixture of MIP, consisting of: 0.1 mmol 2,4-D, 0.2 mmol FlM, 0.2 mmol 4-VPY, 2 mmol EGDMA, 1.5 mol% Irgacure 819 and 1.5 mL methanol/water (4/1). Photopolymerization was performed by injecting light (λ ¼410 nm) through the lens of the polystyrene fiber, using the fiber optic bundle of the spectrofluorimeter, during 20 min at room temperature and protected from outside light, resulting in a MIP-coated fiber optic waveguide. After polymerization, the MIP-coated fiber was washed three times by immersion during 15 min in a solution of ethanol/acetic acid (4/1), and rinsed three times by immersion
362
X.-A. Ton et al. / Biosensors and Bioelectronics 64 (2015) 359–366
during 15 min in methanol. Binding studies were performed as described in Section 2.5.
3. Results and discussion 3.1. Synthesis and characterization of the 2,4-D MIP Prior to the preparation of the fluorescent 2,4-D imprinted polymers, the signaling monomer FlM (Ton et al., 2013), was characterized and titrated with increasing concentrations of 2,4D (2.5–100 nM in methanol/water (4/1)), to verify whether fluorescence enhancement and thus the interaction was taking place. The maximum excitation and emission wavelengths of FlM were 390 nm and 515 nm, respectively. The fluorescence enhancement of FlM was observed to be concentration-dependent and increased with increasing concentrations of 2,4-D (Fig. SI-2, SI), confirming that FlM could be used as a functional monomer. The mechanism of the fluorescence enhancement of FIM can be explained as follows: the piperazinyl nitrogen atom of the monomer interacts with the carboxylic acid group of the analyte by acid-base ion-pair formation, which decreases the ability of the nitrogen to donate an electron, thus blocking the photoinduced electron transfer (PET) responsible for the quenching of fluorescence. Consequently, fluorescence enhancement is observed (De Silva et al., 1997). FlM was then incorporated into a 2,4-D MIP formulation adapted from a previous publication (Haupt et al., 1998), where it replaces partially the monomer 4-VPY. The MIP was synthesized by precipitation polymerization, and nanoparticles of size 300 nm were obtained, as deduced from scanning electron microscopy (SEM) (Fig. SI-3, SI). The binding properties of the polymers were evaluated by spectrofluorimetric titrations with 2,4-D. For this study, suspensions of 5 μg/mL of MIP and NIP were incubated with 2.5–100 nM of 2,4-D in methanol/water (4/1). As shown in Fig. 2A, the fluorescence response of the MIP increased with increasing concentrations of 2,4-D; the limit of quantification (LOQ) was 2.5 nM. Fig. 2B shows that the MIP exhibited a stronger fluorescence response than the NIP. The concentration-dependence of the MIP fluorescence enhancement was then used to determine the affinity of the MIP for 2,4-D. The dissociation constant KD was calculated as follows (Feng et al., 1998): the change in the MIP fluorescence intensity is attributed to the formation of the complex between FlM and 2,4-D described by the equilibrium (1):
FlM+n2, 4− D↔ complex[FlM−2, 4−D]
(1)
The association constant of the complex is thus defined as:
KA = [complex]/([FlM]2,4−D]n )
(2)
If [FlM]0 is the total amount of fluorescent monomer, then [FlM]0 ¼[complex] þ[FlM], with [FlM] being the unbound amount of fluorescent monomer. We also define F0 and F as the emitted fluorescence intensity of FlM respectively in the absence and in the presence of 2,4-D. The change in fluorescence intensity (F F0) is contributed by the formation of [complex]. This gives the relationship between the fluorescence intensity and concentrations of FlM that is [complex]/[FlM] ¼(F F0)/F0, with the condition [complex] o o[FlM]0, so we can attribute F0 to [FlM]. Finally this leads to the following equation:
log((F − F0)/F0) = logKA + n log[2, 4−D]
Fig. 2. (A) Fluorescence spectra (λEX/EM ¼390/515 nm) of MIP particles (5 μg/mL) after incubation with 2,4-D in methanol/water (4/1). (B) Fluorescence response of MIP and NIP particles (5 μg/mL in methanol/water (4/1)) with increasing concentrations of 2,4-D. Measurements were done in triplicate and the error bars represent SD. (C) Fluorescence enhancement response of the MIP particles (5 μg /mL) after incubation with 25 nM 2,4-D, the analogs 2,4-D-OMe and POAc and the non-related compounds Z-L-Phe and GlcA, in methanol/water (4/1).
(3)
Thus, according to (3), the association constant KA and the ratio 2,4-D:FlM in the complex can be deduced. Fig. SI-3 shows the plot of log ((F F0)/F0) vs. log[2,4-D]. Two distinct linear portions, indicating two types of binding sites in the MIP: high-affinity sites
and low-affinity sites are observed. High-affinity sites are the specific imprinted sites while the low-affinity sites are source of non-specific binding. Values of KA and n were calculated from the intercepts and the slopes. For high-affinity sites, KA was found to be 8.5×106 M 1, giving a dissociation constant KD ¼0.12 μM, and n¼ 0.92, which means an approximate ratio FlM:2,4-D ¼ 1:1. For low-affinity sites, the values were found to be KA ¼510 M 1 (KD ¼2 mM) and n ¼0.41, resulting in an approximate ratio FlM:2,4D¼ 2:1. At low concentrations of 2,4-D (o10 nM), only highaffinity sites are occupied. For high concentrations of 2,4-D, the analyte can also bind to low-affinity sites, which explains the increasing non-specific binding for the NIP. To conclude, the MIP displayed a very high affinity for 2,4-D, which is consistent with the low limit of quantification of 2.5 nM, far below the maximum concentration of 2,4-D in drinking water recommended by the
X.-A. Ton et al. / Biosensors and Bioelectronics 64 (2015) 359–366
363
World Health Organization (135 nM; World Health Organization, 2011). To assess the selectivity of the MIP, fluorescence binding assays were performed to compare the binding of 2,4-D (pKa of 5.09, methanol/water (4/1)), with that of two structurally related compounds, 2,4-D methyl ester (2,4-D-OMe), whose carboxyl group is blocked, and phenoxyacetic acid (POAc, pKa ¼5.17, methanol/water (4/1)), lacking the two ring chlorines; and with two non-related compounds containing a carboxyl group, N-carbobenzyloxy-L-phenylalanine (Z-L-Phe, pKa ¼5.06, methanol/water (4/1)) and β-D-glucuronic acid (GlcA, pKa ¼ 5.22, methanol/water (4/1)) (Fig. 1A). As described above, suspensions of 5 μg/mL of MIP particles were incubated with 25 nM of each analyte. For the four molecules tested, no or little fluorescence enhancement (o 20%) was observed (Fig. 2C), much less than with 2,4-D. These results confirm the selectivity of our MIP. 3.2. Coating fiber optic waveguides with MIP particles and sensing of 2,4-D The MIP will now be integrated into a fiber optic device. For this, we used 4-cm long injection-molded tapered polystyrene optical fibers with a lens on one end and a mirror at the other end. The light is injected and collected through the lens (Fig. 1B). In our study, MIP particles were coated onto the fiber by using PVA as a glue (Fig. 3A) (Surugiu et al., 2001). PVA dissolves in hot water for the preparation of the polymer suspension, but after drying does not dissolve under the conditions of the assay. SEM images (Fig. 3C) show a rather homogeneous coating of the MIP particles on the fiber. Excitation was performed through the fiber optic bundle of the spectrofluorimeter and fluorescence emission of the fluorescent MIP particles was collected via the same fiber bundle. Fluorescence spectra were recorded after 10 min incubation in 2,4D solution. Concentration-dependence of the MIP fluorescence response upon binding to 2,4-D is observed, the larger fluorescence enhancement with the MIP compared to the NIP indicating specific binding (Fig. 4A). The MIP-coated fiber optic device exhibited a LOQ of 1 nM, indicating a very high sensitivity of our
Fig. 4. (A) Fluorescence spectra (λEX ¼ 410 nm) from the MIP-coated fiber optic waveguide after incubation with 2,4-D (1–50 nM) in methanol/water (4/1). Inset: fluorescence response of MIP and NIP-coated fiber optic waveguides as a function of 2,4-D concentration. (B) Fluorescence enhancement response of the MIP-coated optical fiber after incubation with 5 nM each of 2,4-D, 2,4-D-OMe, POAc, a mixture of Z-L-Phe and GlcA.
sensor. The results obtained with the fiber were congruent with those obtained with the MIP in batch. To assess the selectivity of the sensor, the MIP-coated fiber was incubated with related and unrelated molecules (see Fig. 1 for structures): 5 nM 2,4-D-OMe,
Fig. 3. (A) Photograph of the polystyrene evanescent wave fiber optic waveguide coated with fluorescent 2,4-D MIP particles. (B) Excitation (λEX ¼ 410 nm) with the fiber optic bundle of the Horiba Fluorolog spectrofluorimeter marked by a red arrow, and fluorescence emission (λEM ¼515 nm) of the fluorescent MIP particles. (C) and (D) Scanning electron microscope images of the surface of the 2,4-D-MIP-coated optical fiber. The scale bars correspond respectively to 200 μm and to 2 μm. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
364
X.-A. Ton et al. / Biosensors and Bioelectronics 64 (2015) 359–366
5 nM POAc or a mixture of 5 nM of Z-L-Phe and 5 nM of GlcA in methanol/water (4/1) (Fig. 4B). Smaller fluorescence enhancement (o25%) is observed compared to the one obtained with 2,4-D (67%), which confirms that the sensor is selective. 3.3. Sensing of the mycotoxin citrinin In order to test the versatility of our system, a MIP for another target molecule, the mycotoxin citrinin (Fig. 1A), was synthesized and applied to the sensor. At the moment, no maximum residue limit for citrinin has been fixed by the European Union but according to a recent report, the characterization of the risk of citrinin as a food contaminant is based on the estimate of the citrinin concentrations in grains and grains-derived products that would result in an exposure equal to the level of no concern for nephrotoxicity, that is 19–100 μg/kg for average consumers (EFSA Panel on Contaminants in the Food Chain, 2012). Citrinin possesses a carboxyl group and is likely to interact with the signaling monomer FlM. Prior to the preparation of the fluorescent citrinin-imprinted polymer, the fluorescence enhancement of FlM with increasing concentrations of citrinin was verified by spectrofluorimetric titrations. The fluorescence enhancement of FlM was observed to be concentration-dependent and increased with increasing concentrations of citrinin (Fig. SI-4, SI). The signal enhancement of FlM was observed for concentrations Z25 ng/mL (0.1 μM) of citrinin.
Citrinin MIP particles were then synthesized by precipitation polymerization. A similar MIP formulation as for 2,4-D was employed, containing FIM, 4-VPY as co-functional monomer and EGDMA as cross-linker in a mixture of methanol/chloroform (5/2) as porogen. Citrinin MIP nanoparticles of a size between 150 and 500 nm were obtained (Fig. SI-5, SI). The binding properties of the citrinin MIP were evaluated by spectrofluorimetric titrations with citrinin, as described for the 2,4-D MIP. Fluorescence enhancement was observed for concentrations Z0.5 μg/mL (2 μM) (Fig. 5A). It has to be noted that the enhancement was due to the complex citrinin FIM and not to the intrinsic fluorescence of citrinin (λEM ¼500 nm) (Zhou et al., 2012), which accounted for only 2% of the total fluorescence intensity observed. To assess the selectivity of this MIP, it was incubated with two non-related compounds possessing carboxyl groups, namely 2,4-D and the mycotoxin fumonisin B1 (FB1). Fig. SI-6, SI) shows that a much lower fluorescence enhancement was observed for 2,4-D and FB1, as compared to citrinin. These results confirm the selectivity of the sensor for citrinin. A citrinin-MIP for the detection of citrinin has already been reported (Guo et al., 2010). It was prepared using a different synthesis protocol and employed for the solid-phase extraction of citrinin from rice, followed by quantification with HPLC. The imprinting factor, ratio of bound citrinin of MIP upon bound citrinin of NIP, as deduced from their chromatograms, was 2.3 but no selectivity studies were reported. Our MIP and NIP particles were then coated on the polystyrene waveguide (see Fig. 5B for an SEM image of the coated fiber). For
Fig.5. (A) (Left) Fluorescence spectra (λEX ¼390 nm; λEM ¼515 nm) of MIP particles, (right) fluorescence enhancement responses of citrinin MIP and NIP particles (0.04 mg/mL) after incubation with increasing concentrations of citrinin in methanol. Measurements were done in triplicate, the error bars represent SD. (B) SEM image of the citrinin MIP-coated fiber. Scale bar: 200 μm. (C) Fluorescence enhancement (λEX ¼ 410 nm; λEM ¼ 515 nm) of the MIP and NIP-coated fibers after incubation with 10 μM (2.5 mg/mL) citrinin, and of the MIP coated fiber after incubation with 10 μM of the non-related compounds, 2,4-D and FB1, in methanol.
X.-A. Ton et al. / Biosensors and Bioelectronics 64 (2015) 359–366
sensing experiments, the fibers were incubated with 2.5 μg/mL (10 μM) of citrinin in methanol for 30 min. Fig. 5C shows that the binding of citrinin to the MIP-coated fiber produced a larger signal than to the NIP-coated fiber. The selectivity was verified by incubation with 10 μM 2,4-D and 10 μM FB1. The fluorescence responses were 3 times lower as compared to those obtained with citrinin. It is worth noting that the sensing is done in methanol, a solvent currently used for the extraction of mycotoxins from food matrices (Kralj Cigić and Prosen, 2009). 3.4. In-situ MIP photopolymerization by evanescent waves on the polystyrene waveguide Having demonstrated that evanescent-wave fiber-optic sensing with a fluorescent signaling MIP is feasible, we wanted to know whether the evanescent field could be used not only for detection but also to directly photopolymerize the MIP in-situ on the fiber. In this way, the fabrication of the sensor would be much faster and more straightforward. In an earlier publication, we showed that polymer objects can be synthesized by evanescent wave-
365
photopolymerization using a laser beam reflected on a flat surface (Fuchs et al., 2011). In the present work, the actinic light source came from the optical fiber probe of the spectrofluorimeter, and polymerization was initiated at 410 nm with an irradiation time of 20 min using the spectrometer light source. The same 2,4-D MIP formulation as described above was used except for the amount of solvent and the initiator (the photoinitiator Irgacure 819 was used instead of ABDV). Irgacure 819 has the advantage to allow for very fast polymerization, as it undergoes fast photolysis under light exposure, resulting in very reactive radicals (Decker et al., 2001). Moreover, it is very reactive towards acrylates and methacrylates (Szablan et al., 2007). As shown in Fig. 6B, a homogeneous polymer coating was obtained that covers the fiber completely. After washing, the MIP-coated fiber was incubated for 15 min with 25 nM 2,4-D in methanol/water (4/1), which yielded a fluorescence enhancement of around 66% (Fig. 6C). The binding was specific, as a lower enhancement was observed for the corresponding NIP-coated fiber. For the moment, the MIP-coated device obtained by in-situ polymerization is somehow less sensitive than the one obtained with the MIP particles synthesized beforehand, and some additional optimization will be necessary. Nevertheless, these preliminary results show that sensitive detection is possible (nM range, below the maximal concentration of 135 nM 2,4-D in drinking water recommended by the WHO) (World Health Organization, 2011).
4. Conclusions
Fig. 6. (A) Photograph and (B) scanning electron microscope image of a waveguide coated with MIP by evanescent wave polymerization. (C) Fluorescence enhancement (λEX ¼ 410 nm) of MIP and NIP-coated waveguides after incubation with 2,4-D in methanol/water (4/1).
We have developed a chemical sensor based on a disposable fiber-optical waveguide coated with a MIP, and on evanescentwave spectroscopy. The MIP was coated on the waveguide either in-situ by direct polymerization using the evanescent wave or exsitu by dip-coating the waveguide with the MIP particles synthesized beforehand. Sensing within the MIP was achieved with a naphthalimide probe, which exhibits fluorescence enhancement when interacting with carboxyl groups of the analytes. Thus the herbicide 2,4-D and the mycotoxin citrinin could be selectively determined in the nM and μM range respectively. Though the MIP sensor displays a certain degree of crossreactivity for related molecules and may therefore not be able to discriminate the target herbicide or mycotoxin in a very complex mixture, as well as it would with SPE followed by LC/MS–MS, it is nevertheless selective enough and can undoubtedly serve as an initial, rapid, sensitive and on-site screening tool, to detect and monitor their presence. As shown in this work, the MIP has great potential for real applications. Frequently monitored mycotoxins that co-exist in grains and grain-derived products are ochratoxin A, fumonisins, zearalenone, trichothecenes (e.g. deoxynivalenol), T2-toxin and aflatoxins (Lerda, 2010). The latter four cannot be detected using a MIP based on the FIM monomer because they do not possess a carboxyl group. For ochratoxin A and fumonisins, which bear a carboxyl group, it should be possible to synthesize MIPs using the same protocol as for citrinin, and to detect them selectively. Thus, an array of waveguides with different MIPs could be a solution for more complex samples containing different target mycotoxins. In addition, we believe that this sensor can also be applied for the detection of other carboxyl-containing analytes, for example in the biomedical field. Using different fluorescent monomers, analytes with other functional groups should also be detectable. To further increase the sensitivity, fine-tuning the thickness of the MIP layer and the inclusion of gold nanoparticles aiding to reflect more of the fluorescence light back into the waveguide (Ton et al., 2013) may be possible solutions.. We believe that this disposable portable device can be useful for cheap and
366
X.-A. Ton et al. / Biosensors and Bioelectronics 64 (2015) 359–366
rapid on-site monitoring of environmental, food and biomedical analytes.
Acknowledgments The authors gratefully acknowledge financial support by the European Regional Development Fund (ERDF), and by the Regional Council of Picardy (Projects CBI-PEM, NANOTOXISCREEN, and cofunding of equipment under CPER 2007–2013).
Appendix A. Supplementary information Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.bios.2014.09.017.
References Alexander, C., Andersson, H.S., Andersson, L.I., Ansell, R.J., Kirsch, N., Nicholls, I.A., O’Mahony, J., Whitcombe, M.J., 2006. J. Mol. Recognit. 19, 106–180. Anderson, G.P., King, K.D., Gaffney, K.L., Johnson, L.H., 2000. Biosens. Bioelectron. 14, 771–777. Arshady, R., Mosbach, K., 1981. Makromol. Chem. 182, 687–692. Bazin, I., Faucet-Marquis, V., Monje, M.-C., El Khoury, M., Marty, J.-L., PfohlLeszkowicz, A., 2013. Toxins 5, 2324–2340. Bus, J.S., Hammond, L.E., 2007. Crop Prot. 26, 266–269. De Silva, A.P., Gunaratne, H.Q.N., Gunnlaugsson, T., Huxley, A.J.M., McCoy, C.P., Rademacher, J.T., Rice, T.E., 1997. Chem. Rev. 28, 1515–1566. Decker, C., Zahouily, K., Decker, D., Nguyen, T., Viet, T., 2001. Polymer 42, 7551–7560. Dickert, F.L., Thierer, S., 1996. Adv. Mater. 8, 987–990. Doussineau, T., Trupp, S., Mohr, G.J., 2009. J. Colloid Interface Sci. 339, 266–270. EFSA Panel on Contaminants in the Food Chain (CONTAM), 2012. EFSA J. 10 (3), 2605–2686. Feng, X.Z., Lin, Z., Yang, L.J., Wang, C., Bai, C., 1998. Talanta 47, 1223–1229.
Fuchs, Y., Linares, A.V., Mayes, A.G., Haupt, K., Soppera, O., 2011. Chem. Mater. 23, 3645–3651. Geng, T., Uknalis, J., Tu, S., Bhunia, A.K., 2006. Sensors 6, 796–807. Guo, B.-Y., Wang, S., Ren, B., Li, X., Qin, F., Li, J., 2010. J. Sep. Sci. 33, 1156–1160. Haupt, K., Dzgoev, A., Mosbach, K., 1998. Anal. Chem. 70, 628–631. Haupt, K., Mosbach, K., 2000. Chem. Rev. 100, 2495–2504. Henry, O.Y.F., Cullen, D.C., Piletsky, S.A., 2005. Anal. Bioanal. Chem. 382, 947–956. Jenkins, A.L., Uy, O.M., Murray, G.M., 1999. Anal. Chem. 71, 373–378. Kralj Cigić, I., Prosen, H., 2009. Int. J. Mol. Sci. 10, 62–115. Kubo, H., Nariai, H., Takeuchi, T., 2003. Chem. Commun., 2792–2793. Leung, M.K.-P., Chow, C.-F., Lam, M.H.-W., 2001. J. Mater. Chem. 11, 2985–2991. Lerda, D., JRC Technical Notes: Mycotoxins factsheet, 3rd ed., 2010. Available online: 〈https://irmm.jrc.ec.europa.eu/EURLs/eurl_mycotoxins/Documents/JRC% 2060040_Mycotoxin%20factsheet_3rd%20edition.pdf〉. Lim, D.V., 2003. Proc. IEEE 91, 902–907. Marazuela, M.D., Moreno-Bondi, M.C., 2002. Anal. Bioanal. Chem. 372, 664–682. Matsui, J., Kubo, H., Takeuchi, T., 2000. Anal. Chem. 72, 3286–3290. Nguyen, T.H., Hardwick, S.A., Sun, T., Grattan, K.T.V., 2012. IEEE Sens. J. 12, 255–260. Niu, C.G., Zeng, G.M., Chen, L.X., Shen, G.L., Yu, R.Q., 2004. Analyst 129, 20–24. Rogers, K.R., Anis, N.A., Valdes, J.J., Eldefrawi, M.E., 1992. In: Mathewson, P.R., Finley, J.W. (Eds.), Biosensor Design and Application. American Chemical Society, Washington D.C., USA, pp. 165–173. Surugiu, I., Danielsson, B., Ye, L., Mosbach, K., Haupt, K., 2001. Anal. Chem. 73, 487–491. Szablan, Z., Junkers, T., Koo, S., Lovestead, T., Davis, T., Stenzel, M., Barner-Kowollik, C., 2007. Macromolecules 40, 6820–6833. Taitt, C.R., Anderson, G.P., Ligler, F.S., 2005. Biosens. Bioelectron. 20, 2470–2487. Takeuchi, T., Mukawa, T., Shinmori, H., 2005. Chem. Rec. 5, 263–275. Thompson, V.S., Maragos, C.M., 1996. J. Agric. Food Chem. 44, 1041–1046. Ton, X.A., Acha, V., Haupt, K., Tse Sum Bui, B., 2012. Biosens. Bioelectron. 36, 22–28. Ton, X.A., Tse Sum Bui, B., Resmini, M., Bonomi, P., Dika, I., Soppera, O., Haupt, K., 2013. Angew. Chem. Int. Ed. 52, 8317–8321. Wagner, R., Wan, W., Biyikal, M., Benito-Peñ a, E., Moreno-Bondi, M.C., Lazraq, I., Rurack, K., Sellergren, B., 2013. J. Org. Chem. 78, 1377-–11389. Wan, W., Biyikal, M., Wagner, R., Sellergren, B., Rurack, K., 2013. Angew. Chem. Int. Ed. 52, 7023–7027. Wolfbeis, O.S., 2004. Anal. Chem. 76, 3269–3284. Wolfbeis, O.S., 2005. J. Mater. Chem. 15, 2657–2669. World Health Organization Guidelines for Drinking Water Quality 4th ed. 2011 (Available online). Wulff, G., 1995. Angew. Chem. Int. Ed. 34, 1812–1832. Zhou, Y., Chen, J., Dong, L., Lu, L., Chen, F., Hub, D., Wanga, X., 2012. J. Lumin. 132, 1437–1445.
