Anal Bioanal Chem DOI 10.1007/s00216-016-9348-8

RESEARCH PAPER

Development of magnetic molecularly imprinted polymers for selective extraction: determination of citrinin in rice samples by liquid chromatography with UV diode array detection Javier L. Urraca 1,2 & José F. Huertas-Pérez 3 & Guillermo Aragoneses Cazorla 1 & Jesus Gracia-Mora 4 & Ana M. García-Campaña 3 & María Cruz Moreno-Bondi 1

Received: 10 November 2015 / Revised: 13 January 2016 / Accepted: 19 January 2016 # Springer-Verlag Berlin Heidelberg 2016

Abstract In this work, we report the synthesis of novel magnetic molecularly imprinted polymers (m-MIPs) and their application to the selective extraction of the mycotoxin citrinin (CIT) from food samples. The polymers were prepared by surface imprinting of Fe3O4 nanoparticles, using 2-naphtholic acid (2NA) as template molecule, N-3,5-bis(trifluoromethyl)phenyl-N'4-vinylphenyl urea and methacrylamide as functional monomers and ethyleneglycol dimethacrylate as cross-linker. The resulting material was characterized by transmission electron microscopy (TEM), and X-ray diffraction (XRD) and Fourier

Published in the topical collection featuring Young Investigators in Analytical and Bioanalytical Science with guest editors S. Daunert, A. Baeumner, S. Deo, J. Ruiz Encinar and L. Zhang. Javier L. Urraca and José F. Huertas-Pérez contributed equally to this work.

transform infrared spectroscopies (FT-IR). The polymers were used to develop a solid-phase extraction method (m-MISPE) for the selective recovery of CIT from rice extracts prior to its determination by HPLC with UV diode array detection. The method involves ultrasound-assisted extraction of the mycotoxin from rice samples with (7:3, v/v) methanol/water, followed by sample cleanup and preconcentration with m-MIP. The extraction (washing and elution) conditions were optimized and their optimal values found to provide CIT recoveries of 94– 98 % with relative standard deviations (RSD) less than 3.4 % (n = 3) for preconcentrated sample extracts (5 mL) fortified with the analyte at concentrations over the range 25–100 μg kg−1. Based on the results, the application of the m-MIPs facilitates the accurate and efficient determination of CIT in rice extracts. Keywords Citrinin . Mycotoxins . Molecular imprinting . Cleanup . Rice

Electronic supplementary material The online version of this article (doi:10.1007/s00216-016-9348-8) contains supplementary material, which is available to authorized users.

Introduction * Javier L. Urraca [email protected] * María Cruz Moreno-Bondi [email protected]

1

Chemical Optosensors and Applied Photochemistry Group (GSOLFA), Department of Analytical Chemistry, Faculty of Chemistry, Complutense University, 28040 Madrid, Spain

2

CEI Campus Moncloa, UCM-UPM, Avda. Complutense s/n, 28040 Madrid, Spain

3

Department of Analytical Chemistry, Faculty of Sciences, University of Granada, Campus Fuentenueva s/n, 18071 Granada, Spain

4

Department of Inorganic Chemistry, National Autonomous University of Mexico (UNAM), Ciudad Universitaria, 04510 Mexico, Mexico

Citrinin [CIT; C13H14O5; IUPAC Name: (3R,4S)-8-hydroxy-3, 4,5-trimethyl-6-oxo-4,6-dihydro-3H-isochromene7-carboxylic acid] (Fig. 1) is a toxic secondary metabolite produced by various filamentous species of Penicillium, Aspergillus and Monascus fungi [1–3]. This mycotoxin is generally formed post-harvest and occurs mainly in stored grains such as wheat, oats, barley, rye and rice but can also be found in beans, fruits, vegetables, herbs and spices, as well as in spoiled dairy products [4]. Citrinin is nephrotoxic, hepatotoxic, immunotoxic and carcinogenic [5] and frequently occurs together with ochratoxin A and aflatoxin B1 in grains, particularly rice, and grains-based products [4]. Therefore, CIT contamination entails a real risk for human and livestock health in addition to economic losses [4]. China and Japan have set a maximum limit of 50 and

J.L. Urraca et al.

Fig. 1 Chemical structures of the target molecule (CIT), surrogates (2NA and 1H2NA), functional monomers (MAM and VPU) and cross-linker (EDMA) used to synthesize the imprinted polymers

200 μg kg−1, respectively, for CIT in fermented red rice [6]. Also, European authorities recently established a maximum level of 2000 μg kg−1 in food supplements based on rice fermented with the red yeast Monascus purpureus [7]. However, the acceptable levels of this mycotoxin in other food and feed commodities have not yet been regulated in Europe or elsewhere [8, 9]. Therefore, accurate, sensitive analytical methodologies for quantifying CIT in food and feed are needed in order to ensure consumers’ and livestock’s health, and also to promote international trade [10]. Citrinin is usually determined (Table 1S, Supplemental Material) by enzyme-linked immunosorbent assay (ELISA) [11, 12] or, especially, by chromatographic separation, mainly by thin-layer chromatography (TLC) or high-performance liquid chromatography (HPLC) with UV, fluorescence (FLD) or mass spectrometry (MS) detection [4, 13–15]. HPLC with tandem mass spectrometry detection (HPLC-MS/MS) is a powerful technique which has lately received much attention, especially for the purposes such as the simultaneous identification and quantitation of multi-class mycotoxins [16–18]. Whichever analytical technique is used, the analysis of complex matrices usually requires thorough sample preparation (especially when the analyte is present at very low concentrations). At present, solidphase extraction (SPE) is by far the most popular technique for routine analyses of mycotoxins [19] including CIT, which has been extracted with C8, aminopropyl and polyamide sorbents, among others [20–22]. However, alternative materials capable of selectively binding the target analyte have been found to provide cleaner, more concentrated extracts [14]. Thus, immunoaffinity columns (IACs) have proved highly selective in this respect by virtue of the specificity of antibody–antigen (target analyte) binding, which has boosted its use to determine CIT in various types of matrices [6, 23, 24] despite the associated shortcomings (mainly a high cost, complexity and susceptibility to changes in pH or ionic strength).

Molecularly imprinted polymers (MIPs) are synthetic materials of use for antibody-like molecular recognition. These materials are synthesized by polymerization of selected functional and cross-linking monomers in the presence of a template molecule that is subsequently removed to generate binding cavities based on shape-matching and functional interactions. As a result, MIPs exhibit specificity for the template molecule and structurally related analogues [25] and are thus excellent choices for molecularly imprinted polymer solid-phase extraction (MISPE). Molecularly imprinted polymers are easier to prepare than immunoaffinity sorbents; also, they are relatively inexpensive, chemically and thermally stable, compatible with organic solvents, reusable and storable over long periods without degradation [26]. Guo et al. [27] developed the first selective MIP for CIT and used it successfully for the analysis of rice samples. Their MIP was prepared by bulk polymerization, which is usually subject to incomplete template removal, slow mass transfer and shape unevenness in the resulting polymer. Polymers obtained as surface-bound thin films on an appropriate support allow more efficient extraction of the template molecules after polymerization, produce binding sites that are easily accessible to the target species and expedite mass transfer and binding relative to bulk polymers. Magnetic nanoparticles (particularly magnetite, Fe3O4) have been widely used to prepare core-shell hybrids that combine the magnetic response of the core with the tailored selectivity of the MIP shell [28–30]. Magnetic MIPs (m-MIPs) are especially useful for analytical separations in crude samples containing suspended solid particles, as well as for biological particulates and large-scale operation. Also, they afford easy template removal, re-binding and use as solid-phase extraction sorbents [30]. This paper reports the development of novel m-MIPs by surface molecular imprinting of solvothermally prepared Fe3O4 nanoparticles. The polymers were synthetized by using a Bstoichiometric^ non-covalent imprinting approach involving the interaction of a urea-based monomer with carboxylcontaining templates [31], N-3,5-bis(trifluoromethyl)phenylN'-4-vinylphenyl urea (VPU) and methacrylamide (MAM) as functional monomers, and ethyleneglycol dimethacrylate (EDMA) as cross-linker (see structures in Fig. 1). Two analyte mimics, namely 2-naphtholic acid (2NA) and 1H2NA, which bear some structural similarities to CIT, were tested as templates. The resulting materials were evaluated chromatographically for their ability to retain the template molecule and CIT. The stoichiometry of non-covalent interactions between 2NA and VPU in the polymerization solvent was determined by 1 H-NMR spectroscopy. The composition and morphology of the m-MIPs was investigated by X-ray diffraction (XRD), Fourier transform infrared spectrometry (FT-IR) and transmission electron microscopy (TEM). The polymers were used to develop an MISPE method for extracting CIT from rice samples, followed by HPLC with UV diode array detection (DAD).

Development of magnetic MIPs for the determination of citrinin in rice samples

Experimental Chemicals and materials The urea-based functional monomer N-3,5-bis (trifluoromethyl)phenyl-N'-4-vinylphenyl urea (VPU) was prepared as described elsewhere [32]. 2-[4-(2-Hydroxyethyl)-1piperazinyl] ethanesulphonic acid (HEPES), 2-naphthoic acid (2NA), 1-hydroxy-2-naphthoic acid (1H2NA), citrinin (CIT), ethyleneglycol dimethacrylate (EDMA) and polyvinylpyrrolidone (PVP) were purchased from Sigma-Aldrich (St. Louis, MO, USA). 15-Crown ether was purchased from Acros (Geel, Belgium) and methacrylamide (MAM) from Fluka (SigmaAldrich, Steinheim, Germany). The polymerization initiator, 2, 2′-azobis(2,4-dimethylvaleronitrile) (ABDV), was obtained from Wako Chemicals (Neuss, Germany). Tetra-n-butylammonium hydrogen sulphate (TBA) (98 %) and sodium chloride were obtained from Merck (Darmstadt, Germany), and oleic acid, 1, 2,2,6,6-pentamethylpiperidine (PMP) and trifluoroacetic acid (TFA, 99 %) were supplied by Alfa Aesar (Karlsruhe, Germany). HPLC-grade acetonitrile (AcN), methanol (MeOH) and dimethylsulphoxide (DMSO) were purchased from Sds (Peypin, France), and HPLC water was purified by passage through a Milli-Q system from Millipore (Bedford, MA). All solutions for HPLC were passed through a 0.45-μm nylon filter before use.

Particle morphology was examined by transmission electron microscopy (TEM) on a JEM 2100 microscope operating at 300 kV. Samples were dispersed in n-butanol and deposited over copper grids coated with holey carbon films. FT-IR spectra were acquired with a Fourier Transform Tensor 27 infrared spectrophotometer from Bruker (Billerica, MA, USA) equipped with a DLATGS detector from Microwatt (Stuart, FL, USA). Synthesis of bulk imprinted and non-imprinted polymers Bulk polymers were prepared as follows: the template molecule (0.5 mmol, 86 mg) was mixed with an equivalent amount of PMP (0.5 mmol, 89 μL) in AcN (5.6 mL) and then with VPU (186 mg, 0.5 mmol), MAM (84 mg, 1 mmol) and EDMA (3.8 mL, 20 mmol). After addition of the azo initiator (1 % w/w total monomers), the mixture was transferred to a glass tube, purged with N2 for 10 min and allowed to polymerize at 40 °C for 48 h. The polymers thus obtained were broken into smaller fragments. The template molecule was extracted by sequential washing with MeOH (100 mL), (9:1, v/v, 100 mL) MeOH/0.1 M HCl (aq) and MeOH (100 mL). The resulting MIP was crushed and sieved, and particles 25– 50 μm in size were collected and allowed to sediment in (80:20, v/v) MeOH/water to remove fines prior to SPE use. A non-imprinted polymer (NIP) was prepared in the same manner, in the absence of a template molecule, for use as a control.

Equipment Preparation of Fe3O4 nanoparticles All buffer solutions and samples were pH-adjusted with an ORION 710A pH/ISE meter (Beverly, MA, USA). The chromatographic system consisted of an HP-1200 series highperformance liquid chromatograph from Agilent Technologies (Palo Alto, CA, USA) equipped with a binary pump, on-line degasser, autosampler, automatic injector, column thermostat and diode array detector (DAD). Chromatographic separation of CIT was performed on an ACE Excel 2 C18-PFP (2) (100 × 2.1 mm, 2 μm) HPLC column from ACE (Aberdeen, Scotland). Column oven temperature was kept at 45 °C, and the analysis were performed at a flow rate of 0.4 mL min−1 using acetonitrile–water (40/60, v/v) as mobile phase. The injected volume was 100 μL and the detection wavelength 331 nm. Peak identification was based on retention time and spectral information. Quantitation was performed by external calibration, using peak area as the analytical signal. The calibration graphs for CIT were linear over the concentration range 5–1000 μg L−1 (r2 > 0.999). The crystallographic structure of the polymer particles was examined on an X’Pert MPD X-ray diffractometer from Philips (Altmelo, The Netherlands) using CuKα radiation (λ = 0.154 nm). Diffraction data were acquired over the 2θ range 5–75°, using room temperature, a step size of 0.015° and a speed of 1 s step−1.

The Fe 3 O 4 nanoparticles used were synthesized by solvothermal reaction [33]. Thus, FeCl3·6H2O (2.43 g) and sodium acetate anhydrous (6.48 g) were mixed with 1.58 mL of polyethyleneglycol and 72 mL of ethyleneglycol at room temperature under shaking for 30 min. Then, the solution was transferred to a Teflon-lined stainless steel reactor and heated at 190 °C for 24 h. The resulting Fe3O4 nanoparticles were recovered with a magnet and washed twice with methanol (2 × 250 mL) and water (2 × 250 mL) before drying under vacuum at 50 °C. Magnetic measurements were made with a VSM 7304 vibrating sample magnetometer from Lake Shore Cyrotronics (Westerville, OH, USA). Synthesis of magnetic molecularly imprinted polymers The synthetic procedure was as follows: the template molecule, 2NA (8.79 mg, 0.05 mmol), was dissolved in 1.5 mL of (93:7, v/v) AcN:DMSO containing an identical number of PMP equivalents (0.05 mmol). Then, 0.1 mmol (8.7 mg) of MAM functional monomer and 0.05 mmol (18.7 mg) of VPU were added, and the mixture (1) was stirred for 30 min. In parallel, Fe3O4 nanoparticles (50 mg) were mixed with

J.L. Urraca et al.

50 μL of oleic acid in a glass vial and stirred for 5 min, followed by addition of 2 mmol (396 mg) of EDMA and (1). The dispersing agent, PVP (20 mg) dissolved in 5 mL of (80 %, v/v) ethanol/water, was then added to the prepolymerization mixture, and the resulting solution was dispersed by sonication in a water bath for 1 h before adding the radical initiator, ABDV (0.35 mg). The mixture was homogenized by sonication, purged with nitrogen (10 min) and placed in an oven at 60 °C for 24 h. The template molecule was extracted by sequentially washing the polymers with MeOH (250 mL), (97:3, v/v) MeOH/ TFA (2 × 250 mL) and MeOH (250 mL). Magnetic nonimprinted polymers (m-NIPs) were prepared and processed similarly but in the absence of 2NA. 1

H-NMR titration

1

H-NMR spectra were obtained on a Bruker Avance DPX500 spectrometer operating at 500 MHz. The template molecule, VPU (6.5 mg), was dissolved in 0.7 mL of anhydrous AcN-d3 (25 mM final concentration). A stock solution of 2NA (1 M in AcN-d3) was prepared by dissolving 175.7 mg in 1 mL of anhydrous AcN-d3 and added in increasing amounts to the VPU solution to a final VPU/2NA mole ratio of 10, 5.0, 3.3, 2.0, 1.43, 1.25, 1.00, 0.67, 0.50, 0.40, 0.33, 0.25, 0.20, 0.13 or 0.10 for recording and processing of 1H-NMR spectra. Chromatographic evaluation of imprinted and non-imprinted polymers Both types of polymers were slurry-packed in methanol into stainless steel HPLC columns (150 mm × 4.6 mm I.D.), using MeOH at a flow rate of 2 mL min−1 as pushing solvent. The columns were equilibrated with mobile phase (HEPES or binary mixtures of acetonitrile/HEPES at pH 7.5) for 15 min. A 0.1 mg mL−1 analyte concentration (CIT, 2NA, 1H2NA), 1 mL min−1 flow rate and 10 μL sample volume were used in each run. The DAD wavelength was set at 331 nm and each elution step repeated three times. Acetone was used as a void volume marker and the temperature kept at 25 °C throughout. The retention factor for each analyte was calculated as k = (t − t0)/ t0, where t and t0 are the retention times of the analyte and void marker (methanol), respectively. The imprinting factor was calculated as IF = kMIP/kNIP (namely, the ratio of the retention factor for each analyte in the MIP column to that in the NIP column). Extraction of CIT from rice samples Rice samples were obtained at a local market. Grains were ground in an Ika M20 mill from Ike-Werke (Staufern, Germany) and passed through a US std. No. 20 sieve (Filtra, Barcelona, Spain). Grain sub-samples (1.0 g) were extracted with 5 mL of (7:3, v/v) methanol/water in a Vibra-cell 72434

sonifier from Bioblock-Scientific (Illkrich, France) at 43 W for 5 min. Then, the samples were passed through a 10-μm filter from Varian (Palo Alto, CA, USA), and the extract was evaporated in a rotary evaporator at 40 °C. Finally, the extract was re-dissolved in 5 mL of HEPES buffer (0.1 M, pH 7.5) and spiked with the mycotoxin at concentrations from 25 to 100 μg kg−1 before extraction with the m-MIPs/NIPs. The experiments were performed in triplicate. The aim of the study was to validate the molecularly imprinted solid-phase extraction (m-MISPE) cleanup procedure and not the extraction method; thus, the rice samples were spiked after extraction to prevent variations from that step. For quantitation, matrix-matched calibration standards were prepared by diluting appropriate amounts of CIT standard with the rice blank extracts obtained under the optimum m-MISPE conditions. The calibration graphs were linear over the CIT concentration range 5–200 μg L−1 (r2 > 0.997). Molecularly imprinted solid-phase extraction An amount of 200 mg of m-MIP was equilibrated with 10 mL of MeOH and 10 mL of HEPES buffer (0.1 M, pH 7.5). Then, a volume of 5 mL of the rice extract in HEPES buffer (0.1 M, pH 7.5) (prepared as described previously) was mixed with magnetic particles for 15 min. After the polymers were washed with 1 mL of 5:95 (v/v) AcN/water to remove nonspecifically retained compounds, CIT was eluted with 1 mL of a methanolic solution containing 50 mM TBA and reequilibrated with 10 mL of MeOH and HEPES (0.1 M, pH 7.5) before a new application. In all steps, the nanoparticles were recovered with a magnet. A volume of 250 μL of the final extract from the m-MISPE procedure was mixed with water up to 1 mL and injected (100 μL) into the HPLC system. All the analyses were run in triplicate.

Results and discussion 1

H-NMR titration

The surrogate–monomer binding stoichiometry was determined by examining the interaction of the functional monomer (VPU) with the surrogate molecule (2NA) by 1H-NMR in AcN-d3. The mixture was prepared by using a constant VPU concentration and increasing amounts of template. Figure S1 (see Electronic Supplementary Material (ESM)) shows the 1H-NMR spectra obtained after each titration step. H-N and H-N' protons underwent the largest changes in chemical shift, namely 4.99 ppm (H-N) and 4.42 ppm (H-N'). Monitoring their shifts allowed the Δδobs values shown in the Job plot of Fig. 2 to be calculated. Shielding of the aromatic protons in the VPU monomer in the presence of 2NA is compatible with a hydrogen donor–acceptor

Development of magnetic MIPs for the determination of citrinin in rice samples 1.8

Fig. 2 Variation of the H-N and H-N' chemical shifts as a function of the mole fraction of VPU-to2NA in AcN-d3 ([VPU] = 0.025 mol L−1). The blue circles and red squares represent experimental data for HN' and H-N protons, respectively

1.6 1.4

χ2NA · Δδ

1.2 1.0 0.8 0.6 0.4 0.2 0.0

0.1

0.2

0.3

0.4

0.5

χ

interaction between the template and the monomer molecules. The stoichiometry of the monomer–template interaction as calculated from the plot of Fig. 2—which peaked at 0.51—was consistent with a 1:1 VPU–2NA complex of the association constant of which was obtained from Eqs. 1 and 2 [34]. The difference in chemical shift of the monitored proton (H-N or H-N') in the absence and presence of template (Δδobs) was calculated from Eq. 1: Δδobs ¼

Δδ011 K 11 ½A 1 þ K 11 ½A

where Δδ011 denotes difference in proton chemical shifts between the free, 1:1 complexed species, K11 is the corresponding association constant and [A] the concentration of free VPU. The last parameter was estimated from the total concentrations of template and VPU ([T]T and [A]T), and K11, at each point in the experimental curves (Eq. 2): ½A ¼

½AT 1 þ K 11 ½T T

Variables Δδobs and [A]T were measured experimentally, whereas Δδ011 and K11 were estimated by fitting the curves to Eq. 1. Also, the intermediate variable [A] was iteratively calculated from Eq. 2 at each point in the experimental curve. Nonlinear least-squares fitting of the curves allowed the binding constant (K11) to be calculated: 3.4 × 103 M−1 for H-N and 3.4 × 103 M−1 for H-N'. Chromatographic evaluation The results of the 1H-NMR experiment were used as the basis for preparing an MIP in bulk format, using 2NA as surrogate

0.6

0.7

0.8

0.9

1.0

VPU

molecule and VPU as functional monomer in a 1:1 mole ratio. A second polymer was synthetized from 1H2NA as a template as described elsewhere [27]. Both surrogates were structurally similar to CIT and possessed the carboxylate moiety needed for successful imprinting. The pre-polymerization mixture was prepared in the presence of methacrylamide as comonomer and EDMA as cross-linker, using a constant 0.5:0.5:1:20 template:VPU:MAM:EDMA mole proportion. After the template was removed, the polymers were used as HPLC sorbents in order to evaluate their ability to retain the template molecule. Based on previous experience in handling MIPs prepared from VPU as functional monomer [31], we used binary mixtures of AcN and HEPES buffer (0.1 M, pH 7.5) containing 75 to 100 % (v/v) of the latter as mobile phases. The buffer pH used was intended to ensure deprotonation of carboxyl groups in the template and facilitate formation of hydrogen bonds with the urea group in VPU. As can be seen from Fig. 3, retention of CIT, 2NA and 1H2NA by the MIP (kMIP) was maximal with 100 % HEPES buffer (0.1 M, pH 7.5) as mobile phase. However, the highest imprinting factors for CIT (Fig. 3b), and also for the template molecules (2NA and 1H2NA), were obtained by using a 5:95 or 10:90 (v/v) AcN/HEPES binary mixture. The polymers prepared from 2NA as template were more selective for CIT than were those prepared with 1H2NA as surrogate—which exhibited poor recognition capabilities even for the template. This result can be ascribed to intramolecular hydrogen bonding between phenolic and carboxyl groups in 1H2NA precluding interaction with the functional monomer in the prepolymerization mixture and leading to poor imprinting. Based on the foregoing, m-MIPs were prepared by using 2NA as template.

J.L. Urraca et al. Fig. 3 (a) Influence of the composition of the mobile phase on the retention of CIT, 2NA and 1H2NA in imprinted and nonimprinted polymers. (b) Imprinting factors (IF) for CIT, 2NA and 1H2NA as a function of the composition of the mobile phase. Sample volume = 10 μL; [Compound] = 0.1 mM; λ(DAD) = 331 nm; temperature = 25 °C; flow rate = 1.0 mL min−1 (RSD < 4.7 %, n = 3)

As can be seen from Fig. 3b, the retention factor for CIT in the 2NA-MIP was higher in 5:95 (v/v) AcN/HEPES (0.1 M, pH 7.5) (kMIP-CIT = 17.7, RSD < 5.4, n = 3) than it was in 10:90 (v/v) mixture (kMIP-CIT = 7.4, RSD < 5.5, n = 3), although their IF values were similar (4.4 and 4.9, respectively). Characterization of m-MIPs The magnetic nanoparticles and m-MIPs prepared by using 2NA as surrogate molecule were characterized by XRD, FTIR and TEM. The XRD patterns for the magnetic nanoparticles before and after polymerization are shown in Fig. S2 (ESM). The patterns are suggestive of a crystalline structure with diffraction peaks at 2θ values of 30.2°, 35.5°, 43.2°, 53.5°, 57.2° and 62.6 o, which can be indexed as the (220), (311), (400), (422), (511) and (440) lattice planes of cubic magnetite. These values are consistent with those in the reference file of the Joint Committee on Powder Diffraction Standards for magnetite (JCPDS card no. 19-629) [35].

Although the XRD peaks for the m-MIPs were weaker, their positions were not modified by polymerization, which suggests that the crystal structure remained essentially unchanged. The spacing parameter for the cubic lattice estimated from the XRD patterns for the magnetic nanoparticles was 0.839 ± 0.002, which compares well with the reference parameter for magnetite (amagnetite = 0.8396 nm) [36], and also with the calculated value for the m-MIPs (am-MIP = 0.836 ± 0.002). The average crystallite size (D) was calculated by applying the Debye-Scherer equation (Eq. 3) to the (311) plane refraction peak: D¼

0:9λ β cos θ

where λ is the wavelength of the Cu Kα radiation used (1.540598 Å), β the half-width of the XRD diffraction line and θ the peak position in degrees. Crystallite size was smaller before polymerization (10.8 nm) than after it (12.1 nm).

Development of magnetic MIPs for the determination of citrinin in rice samples

Optimization of the m-MISPE method The conditions for the m-MISPE procedure were optimized in terms of the composition of the eluting and washing solvent and also of the mass of polymer.

100

80

60

R (%)

Figure 4 shows the FT-IR spectra for the magnetic nanoparticles before and after polymerization. The peak at 592.2 cm−1 was assigned to Fe–O bond vibrations in Fe3O4 and the band at ∼3432 cm−1 to stretching vibrations in O–H bonds due to the presence of hydroxyl groups absorbed in Fe3O4 nanoparticles. The FT-IR spectra of Fig. 4 confirm successful coating of the MIP onto the surface of the magnetic nanoparticles. The peak at ∼2963.1 cm−1 corresponds to stretching vibrations in CH2 and CH3 groups and that at 1467 cm−1 to bending vibrations in CH and CH2 groups. The absorption bands at ca. 1732, 1261 and 1150 cm−1 were assigned to C = O stretching vibrations in carboxyl groups and C–O stretching vibrations in symmetric and asymmetric esters, respectively [37]. Based on the transmission electron micrographs of Fig. S3 (ESM), the uncoated Fe3O4 particles were roughly spherical and covered with MIP after polymerization, which is consistent with the XRD results. The magnetic properties of the synthetic Fe3O4 particles and m-MIPs were studied by vibrating sample magnetometry (VSM). Figure S4 (ESM) shows the magnetic hysteresis loop for the dried samples at room temperature. Saturation magnetization (Ms) was 104.5 emu g−1 for the Fe3O4 nanoparticles and 8.2 emu g−1 for the m-MIPs. The decreased Ms value for the m-MIPs is consistent with shielding of magnetite by the polymeric coating. The particles exhibited superparamagnetic properties at room temperature, but no residual magnetization in the absence of an external magnetic field. In the presence of one, however, the particles were attracted to the vial walls, which facilitated their removal from the suspension and their use as a sorbent for solid-phase extraction.

40

20

0 100/0

95/5

90/10

80/20

50/50

20/80

0/100

Water/Acetonitrile (v/v)

Fig 5 Influence of the composition of the washing solvent (water/AcN) on the extraction recovery (%) of CIT from the MIP (blue bars) and the NIP (red bars)

Elution solvent To optimize the elution solvent, 1 mL of extract from a blank rice sample spiked with a 0.5 mg L−1 concentration of CIT dissolved in HEPES buffer (0.1 M, pH 7.5) was mixed with 100 mg of MIP/NIP magnetic particles and eluted with (a) 1 mL of methanol, (b) methanol containing increasing concentrations of TFA (1–3 %, v/v) or (c) 1 mL of a methanolic solution of 0.05 M tetran-butyl ammonium hydrogen sulphate (TBA), an ion-pairing reagent. The concentration of CIT in each fraction was evaluated by HPLC-DAD. In order to focus the analyte at the head of the chromatographic column and prevent peak diffusion, the eluates were diluted (1:3) with Milli-Q water, as non-eluting focusing solvent, prior to injection [38]. As can be seen from Fig. S5 (ESM), extraction recoveries increased by up to 84 % (RSD = 4 %, n = 3), with the addition of TFA to the eluting solution (97:3, v/v, methanol/TFA). This 50

40

A (mAU)

30

20

CIT

10

0 0.0

Fig 4 FT-IR spectra for Fe3O4 particles before (red line) and after coating with MIP (blue line)

0.5

1.0

1.5

2.0

2.5 3.0 3.5 Time (min)

4.0

4.5

5.0

5.5

6.0

Fig. 6 Chromatograms for a sample fortified with 25 μg CIT kg−1 before (solid black line) and after MISPE (blue line), and also after NISPE (red line) and MISPE of a non-fortified sample (dashed black line)

[27] 5–100 μg L−1

Technique

ELISA UPLC-FLD HPLC-FLD HPLC-FLD HPLC-MS/MS

HPLC-DAD

Preconcentration and cleanup

– – SPE (aminopropyl column) SPE (polyamide column) Inmunoaffinity column

MISPE

Sample

Foodstuff Rice Barley, rye, wheat Wheat, rye, barley, oats, maize Read yeast rice, medicinal plants, others Ground rice

The amount of polymer used in the m-MISPE procedure was optimized for maximal extraction efficiency. For this purpose, 10 mL of a rice extract containing 0.05 mg L−1 CIT dissolved in HEPES buffer (0.1 M, pH 7.5) was mixed with increasing amounts of MIP or NIP magnetic particles (5–250 mg), followed by washing with 1 mL of 5:95 (v/v) AcN/water and elution with 1 mL of 0.05 M TBA in methanol. As can be seen from Fig. S6 (ESM), recovery from the mMIP and m-NIP increased with increasing amount of

Table 2

Amount of m-MIP

Comparison of the analytical characteristics of selected analytical methods for CIT analysis in food samples

Washing solvent Based on the results of the chromatographic evaluation of the polymers, non-specific binding of CIT to the MIPs was minimized by using AcN-water mixture with a low organic solvent concentration. The usefulness of such mixtures as washing solvents in the m-MISPE method was assessed by using one of 100 mg of m-MIP/m-NIP and 10 mL of rice extract dissolved in HEPES buffer (0.1 M, pH 7.5) containing 0.05 mg CIT L−1 for 15 min. This was followed by washing of the polymers in 1 mL of AcN/water (AcN 0–100 %, v/v) to remove nonspecifically retained compounds and, after magnetic separation, by elution of the mycotoxin with 1 mL of 0.05 M TBA in MeOH. The resulting extracts were analysed by HPLC-DAD. As can be seen from Fig. 5, retention of CIT in the m-MIP and m-NIP increased with decreasing concentration of AcN in the washing solvent. Also, recovery from the m-MIP exceeded recovery from the NIP throughout the studied range of AcN concentration in the mixture. Consistent with the chromatographic data, the greatest differences were obtained with 5:95 (v/v) AcN/water. Under these conditions, CIT was recovered by 73 % (RSD = 3.5 %, n = 3) from the m-MIP and 23 % (RSD = 2.1 %, n = 3) from the m-NIP. These result led us to choose this solvent mixture for the washing step.

[11] [15] [21] [22] [23]

–

Precision (RSD %)

was a result of protonation of the carboxyl group in the mycotoxin disrupting interactions with the binding site and increasing the elution yield as a result. However, using 1 mL of a methanolic solution of 0.05 M TBA ensured quantitative recovery (R = 99 %, RSD = 4 %, n = 3). This led us to use it in subsequent tests.

0.5 μg kg−1

3.4 0.9

30 μg L−1 5 μg kg−1 1.7–3.3 μgkg−1 3–5 μg kg−1 2 μg kg−1

1.1

95.8 94.4

10 μg L−1 1.5 μg kg−1 0.6–0.9 μg kg−1 1–2 μg kg−1 0.8 μg kg−1

98.2

47.9 94.4

LOQ

24.5

50 100

LOD

25

1.3–3.8

Fortified level (μg kg−1) Found (μg kg−1) Recovery (%) RSD (%)

20–640 μg L−1 10–100 μg kg−1 1–15 μg kg−1 1–8 μg kg−1 25–200 μg kg−1

Concentration range studied

Table 1 Mean recoveries (R%) and relative standard deviations (RSD, %) for the analysis of CIT in rice samples spiked at three different concentration levels (n = 3)

6.9–13.0 2.9–8.7 4.8–5.5 2.0–13.7 1.4–7.9

Ref

J.L. Urraca et al.

Development of magnetic MIPs for the determination of citrinin in rice samples

magnetic polymer used for MISPE up to about 200 mg. Under these conditions, recoveries from the MIP (R = 95 %; RSD = 2.8 %, n = 3) were 2.6 times higher than recoveries from the NIP (R = 36 %; RSD = 2.6 %, n = 3) and the maximum capacity of the imprinted polymers was estimated to be 2.5 μg g−1 m-MIP. An amount of 200 mg of m-MIP was selected for method development, however, for samples, contaminated with higher amounts of mycotoxin, the sample has to be diluted or the amount of m-MIP must be increased [39]. The reproducibility between m-MIP batches prepared in different days was evaluated by analysing rice extracts containing 0.05 mg L−1 CIT dissolved in HEPES buffer (0.1 M, pH 7.5), in the optimized MISPE conditions. CIT recoveries were in the range of 94–97 % with RSDs < 4 % (n = 3 batches, 3 replicates per batch) demonstrating the reproducibility of the optimized method. Application of m-MISPE to the analysis of CIT in rice extracts The proposed m-MISPE method was used in combination with HPLC-DAD to analyse CIT in rice samples. For this purpose, a matrix-matched calibration curve was constructed by spiking rice sample extracts previously found to contain undetectable levels of CIT with the mycotoxin at five different concentration levels over the range 5–200 μg L−1. The absence of matrix interference peaks at the retention time of CIT (Fig. 6) confirms the specificity of the proposed methodology [40]. Due to the lack of certificate reference materials of CIT in rice samples, trueness was assessed by fortifying the rice extracts with 25, 50 and 100 μg CIT kg−1. Previously, we checked that the samples were free of the analyte, at least at a concentration lower than the LOQ. The experiments were repeated three times. As it can be seen from Table 1, CIT recoveries were in the range of 94–98 % with RSDs < 3.4 % (n = 3). The limits of detection and quantitation for CIT in the rice matrix were calculated by using blank extracts at the lowest analyte concentrations giving a signal-to-noise (s/n) ratio of 3 and 10, respectively, and found to be LOD = 0.7 μg kg−1 and LOQ = 2.3 μg kg−1. The proposed method provides cleaner extracts and higher specific retention in the imprinted material [27] than existing alternatives (Table 2); also, it requires no solvent evaporation before HPLC analysis. Moreover, the m-MIPs were reused for sample analysis, totaling at least 30 extraction cycles, without significant loss in performance or reproducibility (Fig. S7, ESM).

Conclusions A novel m-MISPE method for the selective extraction of the mycotoxin citrinin (CIT) from rice sample extracts was developed. Using 2-NA as surrogate molecule in combination with

VPU as functional monomer allowed the synthesis of m-MIP sorbents affording highly selective re-binding of the mycotoxin in complex samples relative to non-imprinted material. The proposed method was used to determine the mycotoxin in rice sample extracts containing it at low (microgram-per-kilogram) concentration levels by HPLC-DAD with excellent recoveries. The m-MIPs exhibited excellent reusability and, in comparison to traditional MISPE, the method is faster, avoiding the need of SPE column packing or filtration operations, and operationally simple thanks to the ease with which the magnetic particles can be removed. Therefore, it provides a promising choice for the determination of CIT in food matrices. Acknowledgments This work has been supported by MINECO (CTQ2012-37573-C02-02). J.L. Urraca thanks the CEI-Moncloa for a post-doctoral contract. J Gracia-Mora thanks PASPA-UNAM and CONACyT for a grant for his sabbatical leave. Compliance with ethical standards Conflicts of interest The authors declare that they have no competing interests.

References 1.

Balajee SA, Baddley JW, Peterson SW, Nickle D, Varga J, Boey A, et al. Aspergillus alabamensis, a new clinically relevant species in the section Terrei. Eukaryot Cell. 2009;8:713–22. 2. Samson RA, Frisvad JC. Penicillium subgenus Penicillium: new taxonomic schemes and mycotoxins and other extrolites. Stud Mycol. 2004;49:1–257. 3. Udagawa S, Baba H. Monascus lunisporas, a new species isolated from mouldy feeds. Cryptogam Mycol. 1998;19:269–76. 4. EFSA Panel on Contaminants in the Food Chain (CONTAM); Scientific Opinion on the risks for public and animal health related to the presence of citrinin in food and feed. EFSA J. 2012;10:2605– 2687. http://www.efsa.europa.eu/en/efsajournal/pub/2605. Accessed 3 Nov 2015 5. Flajs D, Peraica M. Toxicological properties of citrinin. Arh Hig Rada Toksikol. 2009;60:457–64. 6. Zhu D, Zhang H, Bing X. Preparation of an immunoaffinity column for the clean-up of fermented food samples contaminated with citrinin. Food Addit Contam Part A. 2013;30:389–94. 7. Commission Regulation (EU) No 212/2014 of 6 March 2014 amending Regulation (EC) No 1881/2006 as regards maximum levels of the contaminant citrinin in food supplements based on rice fermented with red yeast Monascus purpureus. Off J Eur Union. L67:3–4. 8. Commission Regulation (EC) No 1881/2006 of 19 December 2006 setting maximum levels for certain contaminants in foodstuffs. Off J Eur Union. L364:5–24. 9. Liao CD, Chen YC, Lin HY, Chiueh LC, Shih DYC. Incidence of citrinin in red yeast rice and various commercial Monascus products in Taiwan from 2009 to 2012. Food Control. 2014;38:178–83. 10. http://ec.europa.eu/food/food/chemicalsafety/contaminants/docs/ monitoring_recommandations_en.pdf. Accessed 3 November 2015.

J.L. Urraca et al. 11.

12.

13. 14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

25. 26.

Duan ZH, Lin ZS, Yao HR, Gao YH, Zhang K, Zhao SQ, et al. Preparation of artificial antigen and egg yolk-derived immunoglobulin (IgY) of citrinin for enzyme-linked immunosorbent assay. Biomed Environ Sci. 2009;22:237–43. Li YN, Wang YY, Zheng YQ, Guo YH. Preparation and characterization of the high specificity monoclonal antibodies against citrinin. Prog Biochem Biophys. 2010;37:1248–53. Xu B, Jia X, Gu L, Sung C. Review on the qualitative and quantitative analysis of the mycotoxin citrinin. Food Control. 2006;17:271–85. Cigiae IK, Prosen H. An overview of conventional and emerging analytical methods for the determination of mycotoxins. Int J Mol Sci. 2009;10:62–115. Huertas-Pérez JF, Arroyo-Manzanares N, García-Campaña AM, Gámiz-Gracia L. High-throughput determination of citrinin in rice by ultra-high-performance liquid chromatography and fluorescence detection (UHPLC-FL). Food Add Cont Part A. 2015;32:1352–7. Garrido-Frenich A, Romero-González R, Gómez-Pérez ML, Martínez-Vidal JL. Multi-mycotoxin analysis in eggs using a QuEChERS-based extraction procedure and ultra-high-pressure liquid chromatography coupled to triple quadrupole mass spectrometry. J Chromatogr A. 2011;1218:4349–56. Arroyo-Manzanares N, Huertas-Pérez JF, Gámiz-Gracia L, GarcíaCampaña AM. A new approach in sample treatment combined with UHPLC-MS/MS for the determination of multiclass mycotoxins in edible nuts and seeds. Talanta. 2013;115:61–7. Arroyo-Manzanares N, Huertas-Pérez JF, García-Campaña AM, Gámiz-Gracia L. Simple methodology for the determination of mycotoxins in pseudocereals, spelt and rice. Food Control. 2014;36: 94–101. Turner NW, Subrahmanyam S, Piletsky SA. Analytical methods for determination of mycotoxins: a review. Anal Chim Acta. 2009;632: 168–80. Tuomi T, Johnsson T, Hintikka EL, Reijula K. Detection of aflatoxins (G1-2, B1-2), sterigmatocystin, citrinine and ochratoxin A in samples contaminated by microbes. Analyst. 2001;126:1545–50. Hartl A, Stenzel WR. Development of a method for the determination of citrinin in barley, rye and wheat by solid phase extraction on aminopropyl columns and HPLC-FLD. Mycotoxin Res. 2007;23: 127–31. Meister U. New method of citrinin determination by HPLC after polyamide column clean-up. Eur Food Res Technol. 2004;218: 394–9. Li Y, Zhou YC, Yang MH, Ou-Yang Z. Natural occurrence of citrinin in widely consumed traditional Chinese food red yeast rice, medicinal plants and their related products. Food Chem. 2012;132:1040–5. Sato T, Higashihara K, Sasaki A, Toth D, Goto T. Development and single laboratory validation of a method for citrinin. World Mycotoxin J. 2010;3:129–34. Haupt K. Editor. Molecular imprinting. Heidelberg: Springer; 2012. Urraca JL, Marazuela MD, Merino ER, Orellana G, Moreno-Bondi MC. Molecularly imprinted polymers with a streamlined mimic for zearalenone analysis. J Chromatogr A. 2006;1116:127–34.

27.

28.

29.

30.

31.

32.

33.

34. 35.

36.

37.

38.

39.

40.

Guo BY, Wang S, Ren B, Li X, Qin F, Li J. Citrinin selective molecularly imprinted polymers for SPE. J Sep Sci. 2010;33: 1156–60. Zhou Y, Zhou T, Jin H, Jing T, Song B, Mei S, et al. Rapid and selective extraction of multiple macrolide antibiotics in food stuff samples based on magnetic molecularly imprinted polymers. Talanta. 2015;137:1–10. Miao SS, Wu MS, Zuo HG, Jiang C, Jin SF, Lu YC, et al. Coreshell magnetic molecularly imprinted polymers as sorbent for sulfonylurea herbicide residues. J Agric Food Chem. 2015;63:3634– 45. Li XS, Zhu GT, Luo YB, Yuan BF, Feng YQ. Synthesis and applications of functionalized magnetic materials in sample preparation. TrAC Trends Anal Chem. 2013;45:233–47. Urraca JL, Hall AJ, Moreno-Bondi MC, Sellergren B. A stoichometric molecularly imprinted polymer for the classselective recognition of antibiotics in aqueous media. Angew Chem Int Ed. 2006;45:5158–61. Hall AJ, Manesiotis P, De Lorenzi E, Emgenbroich M, Sellergren B. Urea host monomers for stoichiometric molecular imprinting of oxyanions. J Org Chem. 2005;70:1732–6. Liu X, Yu D, Yu Y, Ji S. Preparation of a magnetic imprinted polymer for selective recognition of rhodamine B. Appl Surf Sci. 2014;320:138–45. Fielding L. Determination of association constants (Ka) from solution NMR data. Tetrahedron. 2000;56:6151–70. American Society for Testing and Materials, JCPDS—Joint Committee on Powder Diffraction Standards. Mineral power diffraction files data book. Pennsylvania: Swarthmore; 1980. American Society for Testing and Materials, JCPDS—Joint Committee on Powder Diffraction Standards. Selected Powder Diffraction Data for Minerals: Data Book. 1st ed. Pennsylvania: Swarthmore; 1974. Yoshimatsu K, Reimhult K, Krozer A, Mosbach K, Sode K, Ye L. Uniform molecularly imprinted microspheres and nanoparticles prepared by precipitation polymerization: the control of particle size suitable for different analytical applications. Anal Chim Acta. 2007;584:112–21. Stephen R, Bakalyar SR, Phipps C, Spruce B, Olsen K. Choosing sample volume to achieve maximum detection sensitivity and resolution with high-performance liquid chromatography columns of 1.0, 2.1 and 4.6 mm I.D. J Chromatogr A. 1997;762:167–85. Chapuis F, Pichon V, Lanza F, Sellergren B, Hennion MC. Retention mechanism of analytes in the solid-phase extraction process using molecularly imprinted polymers: application to the extraction of triazines from complex matrices. J Chromatogr B. 2004;804:93–101. European Commission Decision 2002/657/EC of 12 August implementing Council Directive 96/23/EC concerning the performance of analytical methods and the interpretation of results. Off J Eur Union. 2002; L221:8–36.