New molecularly imprinted polymers for reducing negative volatile phenols in red wine with low impact on wine colour

Journal Pre-proofs New molecularly imprinted polymers for reducing negative volatile phenols in red wine with low impact on wine colour Luís Filipe-Ribeiro, Fernanda Cosme, Fernando M. Nunes PII: DOI: Reference:

S0963-9969(19)30741-0 https://doi.org/10.1016/j.foodres.2019.108855 FRIN 108855

To appear in:

Food Research International

Received Date: Revised Date: Accepted Date:

20 February 2019 16 November 2019 20 November 2019

Please cite this article as: Filipe-Ribeiro, L., Cosme, F., Nunes, F.M., New molecularly imprinted polymers for reducing negative volatile phenols in red wine with low impact on wine colour, Food Research International (2019), doi: https://doi.org/10.1016/j.foodres.2019.108855

This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

© 2019 Published by Elsevier Ltd.

New molecularly imprinted polymers for reducing negative volatile phenols in red wine with low impact on wine colour

Luís Filipe-Ribeiro*1, Fernanda Cosme2, Fernando M. Nunes1

CQ-VR, Chemistry Research Centre, Food and Wine Chemistry Lab., 1Chemistry Department, 2

Biology and Environmental Department, University of Trás-os-Montes and Alto Douro,

School of Life Sciences and Environment, 5000-801 Vila Real, Portugal.

*Corresponding author: Email: [email protected] Phone: +351 918624138

1

Abstract 4-Ethylphenol (4-EP) and 4-ethylguaiacol (4-EG) formation in red wines by Dekkera/Brettanomyces yeasts reduce significantly wine consumer’s acceptability. Polymers with specific adsorption for volatile phenols (VPs) could be a valuable tool for wine producers for removing this negative sensory defect. In this work, a new molecularly imprinted polymer (MIP) was synthesised using ethylene glycol dimethacrylate (EDMA) as cross-linker and ethylene glycol methyl ether acrylate as functional monomers. Although there was observed a competitive binding of the more abundant structurally related phenolic compounds of the wine matrix, it was still able to reduce 38 to 63% the wine VPs, depending on the wine VPs levels, presenting higher performance than the respective non-imprinted polymers (NIP). Sensory analysis of the MIP treated wine resulted in a significant decrease in the phenolic attribute and significant increase of the fruity and floral attributes, with no significant differences in the wine colour perceived by the expert panel. The sensory improvement of the MIP was significantly higher than that observed for the correspondent NIP.

Keywords: Red wine, 4-Ethylphenol, 4-Ethylguaiacol, Molecularly imprinted polymers, Phenolic compounds, Wine colour, Sensory quality

Acronyms: 4-EG - 4-ethylguaiacol; 4-EP - 4-ethylphenol; Ce - concentration in sample solution; C-index - panellists consistence; Dp - average pore sizes; EDMA - ethylene glycol dimethacrylate; FA - frontal analysis; FTIR - Fourier transform infrared spectroscopy; HA headspace abundance; HS-SPME - headspace solid phase microextraction; K0 - median binding affinity; KL - Langmuir constant; KLxQm - binding capacity; MIP - molecular imprinted polymer; MIP1EG - molecularly imprinted polymer produced with methacrylamide monomer and 4-ethylguaiacol template; MIP1EP - molecularly imprinted polymer produced with methacrylamide monomer and 4-ethylphenol template; MIP2EG - molecularly imprinted polymer produced with ethyl methacrylate monomer and 4-ethylguaiacol template; MIP2EP molecularly imprinted polymer ethyl methacrylate monomer and 4-ethylphenol template; MIP3EG - molecularly imprinted polymer produced with ethylene glycol methyl ether

3

acrylate monomer and 4-ethylguaiacol template; MIP3EP - molecularly imprinted polymer produced with ethylene glycol methyl ether acrylate monomer and 4-ethylphenol template; MIPs - molecularly imprinted polymers; NIP - non imprinting polymer; NIP1 - non-imprinted polymer produced with methacrylamide monomer; NIP2 non-imprinted polymer produced with ethyl methacrylate monomer; NIP3 - non-imprinted polymer produced with ethylene glycol methyl ether acrylate monomer; ODT - olfactory detection threshold; Qe - the concentration of VP in equilibrium; Qm- total number of binding sites; Qmax - theoretical maximum monolayer capacity; SBET - specific surface area; Smeso - mesopore area; Vp - specific pore volume; VPs - volatile phenols; Vri - retention volume.

4

1. Introduction Volatile phenols, namely 4-ethylphenol and 4-ethylguaiacol, when present in wines above their olfactory detection threshold (ODT) confer phenolic, animal, and stable aromatic attributes (Chatonnet, Dubourdieu, Boidron, & Pons, 1992; Tempère, Schaaper, Cuzange, de Lescar, de Revel, & Sicard, 2016). This sensory defect is easily detected by consumers (Schumaker, Chandra, Malfeito-Ferreira, & Ross, 2017), especially for consumers with medium and high wine knowledge levels and considered negative to the wine quality. The presence of these volatile phenols reduces wine fruity aroma, although wine aroma background and style may influence the perception of this sensory defect (Schumaker, Diako, Castura, Edwards, & Ross, 2019). The appearance of volatile phenols in wine is mainly due to the activity of the spoilage yeast Dekkera/Brettanomyces bruxellensis (Chatonnet, Viala, & Dubourdieu, 1997) that can decarboxylate the hydroxycinnamic acid precursors, p-coumaric and ferulic acids, to the correspondent vinylphenols by the hydroxycinnamate decarboxylase followed by reduction of vinylphenols to the correspondent ethylphenols by the vinylphenol reductase (Chatonnet et al., 1997). Dekkera/Brettanomyces are widely present in wineries originating from the vineyard (Oro, Canonico, Marinelli, Ciani, & Comitini, 2019). Due to the serious impact on wine quality and consequent economic losses to the wineries, preventive actions have been developed and implemented during winemaking (Milheiro, Filipe-Ribeiro, Vilela, Cosme, & Nunes, 2017a). The most common method to prevent and/or control B. bruxellensis spoilage in winemaking is the addition of sulfur dioxide into must and wine. However, recently, it was reported that some B. bruxellensis strains can tolerate commonly used doses of sulphur dioxide (Avramova, Vallet-Courbin, Maupeu, MasneufPomarède, & Albertin, 2018). These differences in sulphur dioxide resistance are

5

probably due to preferential transcription of a BbSSU1 allele that encodes a more efficient Ssu1p transporter (Varela, Bartel, Roach, Borneman, & Curtin, 2019). The wide variability in B. bruxellensis in response to wine limiting factors in terms of the accumulation of volatile phenols, besides the strain resistance to environmental stress factors, is also closely related to chemical composition of wine, making difficult a priori evaluation of risk of wine alteration (Crauwels et al., 2017; Guzzon, Larcher, Guarcello, Francesca, Settanni, & Moschetti, 2018; Dimopoulou, Hatzikamari, Masneuf-Pomarede, & Albertin, 2019). Other preventive measures also available to winemakers include the use of chitosan (Portugal et al., 2014) and dimethyldicarbonate (Renouf, Strehaiano, & Lonvaud-Funel, 2008). Nevertheless, today there are wines still commercialised worldwide with high levels of these VPs (Milheiro et al., 2017a). In a recent study performed in Portuguese finished red wines, levels of 4-EP and 4-EG ranging from 4.5– 5604 μg/L and 2.3–831.2 μg/L, respectively were found, with 40% of the industrially produced wines containing 4-ethylphenol, 4-ethylguaiacol or both at levels above their threshold level (Filipe-Ribeiro, Milheiro, Ferreira, Correia, Cosme, & Nunes, 2019). Due to the difficulties in dealing with Dekkera/Brettanomyces wine contamination, several experimental treatments have been recently explored for its elimination from the grape surface, must, wine and oak barrels. Pulsed electric field (PEF) technology has been explored in red wines for inactivating Brettanomyces bruxellensis (van Wyk, Silva, & Farid, 2019), although when compared to high pressure processing (HPP), another tested non-thermal treatment, this last treatment was found to be more effective in maintaining wine quality regarding the development of “Brett Character” off-flavour after one year storage (van Wyk, & Silva, 2017; van Wyk, Farid, & Silva, 2018). The thermosonication of wine has also shown good results in the reduction of the B. bruxellensis population, although only in early stages of wine contamination (Križanović, Tomašević, Ćurko,

6

Gracin, Lukić, & Kovačević Ganić, 2019). Other studied options for reducing wine contamination with this yeast include the use gaseous ozone for reducing grape surface Brettanomyces contamination (Cravero et al., 2018; Englezos et al., 2019). For aged wine, used oak barrels are one of the main source of Dekkera/Brettanomyces contamination and therefore several treatments have also been devised for reducing their populations, for example, application of heated water, to reduce the risk of wine spoilage by barrels contaminated with B. bruxellensis. (at least at 70°C for a minimum of 30 minutes, Edwards, & Cartwright, 2019), or the use of steam during 9 minutes to inactivated yeasts present up to 4 mm cross sections and 12 min for a depth of 5 to 9 mm (Cartwright,. Glawe, &, Edwards, 2018). Meanwhile, while the Dekkera/Brettanomyces contamination of wines is not efficiently prevented or eliminated, several remediation treatments have been developed to reduce or eliminate the unpleasant VPs in the wine and their negative sensory impacts. These treatments can be divided into two main groups: those intended to decrease the headspace contents by decreasing their partition coefficients to the gas phase without changing the total VPs contents of wines by using additives like isinglass, carboxymethylcellulose (CMC) and chitosan (non-extractive techniques; (Filipe-Ribeiro, Cosme, & Nunes, 2018; Milheiro, Filipe-Ribeiro, Cosme, & Nunes, 2017; Petrozziello et al., 2014), and those intended to remove the VPs from wines, decreasing their content and their headspace concentration by using fining agents (extractive techniques). The oenological products already authorised by the OIV and tested in red wines include activated carbons (Milheiro et al., 2017a; Milheiro, Filipe-Ribeiro, Cosme, & Nunes, 2017b; Filipe-Ribeiro, Milheiro, Matos, Cosme, & Nunes, 2017a; Lisanti, Gambuti, Genovese, Piombino, & Moio, 2017; Milheiro et al., 2017) yeast cell walls (Chassagne, Guilloux-Benatier, Alexandre, & Voilley, 2005; Nieto-Rojo, Ancín-Azpilicueta, &

7

Garrido, 2014), potassium caseinate and egg albumin (Milheiro et al., 2017). Other nonconventional alternative solutions have been tested as the use of air depleted and solvent impregnated cork powder (250 g/hL) (Filipe-Ribeiro, Cosme, & Nunes, 2019a; FilipeRibeiro, Cosme, & Nunes, 2019b); esterified cellulose (200g/hL) (Larcher, Puecher, Rohregger, Malacarne, & Nicolini. 2012), suberin adsorbed on glass beads (dose not supplied, Gallardo-Chacón, & Karbowiak, 2015) and polyaniline (PANI) based adsorbent in base form (1000 g/hL) (Marican, Carrasco-Sanchez, John, Laurie, & Santos, 2014; Carrasco-Sánchez, John, Marican, Santos, & Laurie, 2015) all of them allowed a significant removal of 4-EP and 4-EG from red wines. Nevertheless, although in some treatments it has been shown a significant improvement of the wine quality by decreasing the aroma phenolic attribute and increasing the fruity and floral attributes (Filipe-Ribeiro, Cosme, & Nunes, 2019a; Filipe-Ribeiro, Cosme, & Nunes, 2019b; Filipe-Ribeiro, Cosme, & Nunes, 2017b; Filipe-Ribeiro, Milheiro, Matos, Cosme, & Nunes, 2017a; Filipe-Ribeiro, Cosme, & Nunes, 2018), the lack of specificity of these treatments for the volatile phenols have some side-effects on the wine phenolic composition and/or overall headspace aroma abundance. Production of molecularly imprinted polymers (MIPs) that could specifically and efficiently reduce or even remove VPs from wine could be a solution to deal with this negative and defective aromatic character. MIPs are synthetic materials with template induced binding sites to recognise the substance of interest, usually named template molecule, in preference to other structural analogues (Chen, & Li, 2012; Haupt, 2012). MIPs are synthesised by copolymerisation of a functional monomer and a cross-linker monomer in the presence of template molecules. Removal of the template molecules from the highly cross-linked polymer results in the formation of recognition cavities. These specific cavities can selectively bind the template molecules from complex matrices

8

(Vasapollo et al., 2011). MIPs have been developed and used in various applications, from stationary phases to be used in chromatography, solid-phase extraction, chemical sensors, drug-delivery and enzyme-like catalysis, due to their obvious advantages of predictable specific recognition, comparatively low cost, easy preparation, chemical stability to harsh chemical and physical conditions and excellent reusability (Chen, & Li, 2012; Vasapollo et al., 2011). Two works are described in the literature aiming to develop MIPs for removing VPs from wine (Garde-Cerdán, Lorenzo, Carot, Jabaloyes, Esteve, & Salinas, 2008; Teixeira et al., 2015). Both works used vinylpyridine as a functional monomer for MIP synthesis, but with different cross-linkers, EDMA in the work of Teixeira et al. (2015) and divinylbenzene-80 in the work of Garde-Cerdán et al. (2008). Also, different template molecules were used in both works, 4-EP and 4-EG in the work of Teixeira et al. (2015) and penthachlorophenol in the work of Garde-Cerdán et al. (2008). Their application doses were 400g/hL (Garde-Cerdán et al., 2008) and 200g/hL (Teixeira et al., 2015). In both works the non-imprinted polymers (NIPs), produced without the presence of the template molecule, showed already high adsorption for the two VPs - in average 45% (Teixeira et al., 2015) and 59% (Garde-Cerdán et al., 2008) but the MIP showed increased adsorption values - 55% (Teixeira et al., 2015) and 91% (Garde-Cerdán et al., 2008). Nevertheless, in both works, significant amounts of volatile compounds were removed and Teixeira et al. (2015) described a significant reduction of phenolic compounds and anthocyanins. The treatment impact on the overall wine sensory profile was not evaluated. Therefore, more work is needed to decrease the undesirable side effects of these polymers in the wine. Also, the sensory impact of the treatment with these polymers should be studied to access the suitability of this treatment to deal with this sensory defect.

9

Therefore, the purpose of this work was to develop new MIPs and evaluate their efficiency in the wine VPs removal. Their impact on wine quality, including the wine sensory profile, was also determined, as it will ultimately dictate the suitability of this approach for reduction of the negative red wine “Brett Character” sensory profile.

2. Materials and Methods 2.1. MIPs production MIPs were prepared by precipitation polymerisation by adaptation of the method described by Wang, Cormack, Sherrington and Khoshde (2003). This is a simple method without the need of stabilisers or other additives, with high degrees of crosslinking agents and use of polar aprotic solvents as porogens, which are capable of preserving noncovalent interactions between template and monomers and suitable particle sizes (Pardeshi, & Singh, 2016), (Figure 1). Briefly 20 mmol of EDMA (MW=198.22 g/mol, Sigma-Aldrich, Spain) as cross-linker and 4 mmol of three different functional monomers: methacrylamide (1, MW=85.10 g/mol, Sigma-Aldrich, Spain); ethyl methacrylate (2, MW=114.14 g/mol, Sigma-Aldrich, Spain) and ethylene glycol methyl ether acrylate (3, MW=130.14 g/mol, Sigma-Aldrich, Spain) were transferred in 50 mL of an acetonitrile:water (4:1 v/v) mixture. 2,2′-Azobis(2-methylpropionamidine) dihydrochloride (MW=271,19 g/mol, Sigma-Aldrich, Spain) was used as a free-radical initiator (AAPH, 0.33 mmol) and 4-ethylphenol (MW=122.16 g/mol, Sigma-Aldrich, Spain) or 4-ethylguaiacol (MW=152.19 g/mol, Sigma-Aldrich, Spain) (1 mmol) as templates. Initially, all reagents and solvents, except for the free radical initiator, were added to a 50 mL conical flask and the solution was degassed in an ultrasonic bath for 5 min then sparged with oxygen-free nitrogen for 10 min while cooling on an ice bath. After this treatment, AAPH was added and the tubes were rotated at 5-10 rpm to minimise

10

turbulence in a rotor-arm shaker in a controlled temperature oven (60 ºC) during 24 h. At the end of the reaction, the polymers were separated by vacuum filtration and washed in a Soxhlet with methanol/acetic acid (80/20 v/v) followed by methanol and dried overnight at 50 ºC in a forced-air oven. Non-imprinted control polymers were prepared under identical conditions to MIPs except that the template was omitted. Yields were determined by gravimetric analysis after drying. 2.2. NIPs and MIPs physicochemical characterisation 2.2.1. Adsorption isotherms and pore size analysis The adsorption isotherms for N2 (purity > 99.998%) at -196 ºC were determined using a semiautomatic adsorption apparatus Quantachrome Nova4200e and polymers pore size was determined according to the method described by Filipe-Ribeiro et al. (2017a). Analyses were performed in duplicate. 2.2.2. Scanning electron microscopy (SEM) NIPs and MIPs morphology was analysed using the FEI Quanta 400 Scanning Electron Microscope (FEI Company, USA) in environmental mode at 6 mbar using a Large Field Detector. NIPs and MIPs powders were applied on carbon glue. An accelerating voltage of 30 kV was used. 2.2.3. Fourier transform infrared (FTIR) spectroscopy For acquiring NIPs and MIPs FTIR spectra, a Unicam Research Series FTIR spectrometer was used. The spectra were recorded in the range of wavenumbers 4000450 cm-1 and 128 scans were taken at 2 cm-1 resolution. Pellets were prepared by thoroughly mixing polymers and KBr at a 1:400 sample/KBr weight ratio in a small size agate mortar. The resulting mixture was placed in a manual hydraulic press, and a force of 15 tons was applied for 10 min. Analyses were performed in duplicate. 2.2.4. X-ray diffraction analysis of NIPs and MIPs

11

Powder X-ray diffraction (XRD) data were recorded on NIPs and MIPs using a PANalytical X’Pert Pro X-ray diffractometer equipped with an X’Celerator detector and secondary monochromator. The measurements were performed using Cu Kα radiation (40kV; 30 mA) in Bragg-Bentano geometry at a 7-60° 2θ angular range. Analyses were performed in duplicate 2.3. Adsorption isotherm measurement Frontal analysis (FA) was used to determine the adsorption isotherms of 4-EP and 4-EG of MIPs and NIPs synthesised as is the most accurate chromatographic method for this determination (Sajonz, Zhong, & Guiochon, 1996). 2.3.1. Column void time determination The column void time, t0, was measured by injecting a non-retained compound (sodium nitrate at 1 mg/mL), and its retention time was taken as the column void time. 2.3.2. System delay time determination The system delay time, τ, represents the amount of time required for the compound to travel from the solvent reservoir through the HPLC system to the beginning of the column. The system delay time was measured by replacing the column with a zerovolume union and determining the breakthrough time of solutions of 4-EP from 2.5 to 100mg/L. The measured retention time of the fronts in FA was then corrected by subtraction of the system delay time. 2.3.3. FA method This method determines the isotherm through a series of stepwise increases in the concentration of 4-EP and 4-EG from 2.5 to 100 mg/L in a model wine solution (12% v/v ethanol, 3.5 g/L tartaric acid adjusted to pH 3.60 with NaOH). Using the HPLC system, one line was inserted into a reservoir of 100% mobile phase (model wine solution), and the other was inserted into a reservoir containing a predetermined concentration of one or

12

the other VPs in the mobile phase at 100 mg/L (line B). The amount pumped through the system from line B was increased step-wise at intervals to obtain 2.5mg/L, 5mg/L, 7.5mg/L, 10mg/L, 15mg/L, 20mg/L, 30mg/L, 40mg/L, 50mg/L, 75mg/L and 100mg/L and maintained during 90 min. The retention time of the breakthrough curve was recorded. The concentration of VP, Qe, in equilibrium with its concentration in sample solution Ce, was calculated using Equation 1: 𝑄𝑒 =

𝐶𝑒 − 𝐶𝑒−1 ×

𝑉𝑟𝑖 − 𝑉0 + 𝑄𝑒−1 (Equation Equation 4) 1 𝑉𝑎

where Ce-1 and Ce are the VP concentrations in the sample solutions at the beginning and the end of frontal development, respectively, Qe-1 and Qe are the VP concentrations accumulated on the stationary phase in equilibrium with Ce-1 and Ce, respectively, Vri is the retention volume of the ith frontal development, which is obtained by the half-height method (Rudzinske, & Everett, 1992). 2.3.4. Curve fitting and modelling of the adsorption isotherms To characterise the adsorption systems, three isotherm models, Langmuir, Freundlich and Langmuir-Freundlich isotherms, were used to analyse the equilibrium data obtained experimentally. The Langmuir model assumes that adsorption takes place on a homogeneous surface with identical active sites and uniform energies. The Langmuir model is expressed as (Equation 2) (García-Calzón, & Díaz-García, 2007). 𝑄𝑒 =

𝐾𝐿 𝑄𝑚𝑎𝑥 𝐶𝑒 (Equation 6) 1 + 𝐾𝐿 𝐶𝑒 Equation 2

where Ce is the equilibrium concentration, Qe is the amount of VP adsorbed at equilibrium, Qmax is the theoretical maximum monolayer capacity, and KL is the Langmuir constant related to the affinity of the active sites. Oppositely, the Freundlich model assumes that adsorption occurs on a heterogeneous surface with an exponential

13

distribution of active sites and energies (García-Calzón, & Díaz-García, 2007), is expressed as (Equation 3): 1/𝑛

𝑄𝑒 = 𝐾𝐹 𝐶𝑒

(Equation Equation 37)

where Ce and Qe are defined as above and n and KF are Freundlich constants, which are related to the adsorption favourability and adsorption capacity, respectively. The Langmuir-Freundlich isotherm (Equation 4), also known as the Sips equation, is capable of modelling both homogeneous and heterogeneous binding surfaces and is expressed as (Umpleby, Baxter, Chen, Shah, & Shimizu, 2001). 𝑄𝑚 𝑎𝐶𝑒𝑛 𝑄𝑒 = (Equation Equation 8) 4 1 + 𝑎𝐶𝑒𝑛

where Qe and Ce are described as above and Qm is the total number of binding sites, a is a parameter related to the median binding affinity (K0) via K0 = a1/n and n is the heterogeneity index, which varies from 0 to 1. For a homogeneous material, n = 1 and heterogeneous materials n < 1. In contrast to the heterogeneous Freundlich isotherm, the Langmuir-Freundlich model has the advantage that it does not require an independent measure of the total number of binding sites (Qm), which is very difficult to measure in heterogeneous MIPs (Umpleby et al., 2001). The Langmuir-Freundlich isotherm is a composite of the Langmuir and Freundlich isotherms and can reduce to either at its limits, when n = 1, the Langmuir-Freundlich isotherm (Equation 2) reduces to the Langmuir isotherm (Equation 6) in which a corresponds directly to binding affinity (KL). Alternatively, as either Ce or a approaches 0, the Langmuir-Freundlich isotherm reduces to the Freundlich isotherm (Equation 3). Also, the Langmuir-Freundlich isotherm reduces for all systems to the Freundlich isotherm at low concentrations. 2.4. Experimental design

14

To study the effect of NIPs and MIPs in the removal of VPs from wines, polymers were used at 250g/hL. The wine was previously spiked at two levels of 4-EP (750µg/L and 1500µg/L) and 4-EG (150µg/L and 300µg/L) according to the ranges usually found in the literature (Chatonnet et al., 1992; Milheiro et al., 2017a). NIPs and MIPs were added to the 250 mL of wine contained in 250 mL graduated cylinders. After 6 days, the wine was centrifuged at 10,956 g, 10 min, and 20 °C before analysis. Fining treatments were performed in duplicate. Each treatment was analysed in duplicate unless otherwise stated. 2.5. Wine samples A red wine from Douro Valley (vintage 2016) was used in this work. The main characteristics of the wine were as follows: alcohol content 13.2% (v/v), specific gravity (20°C) 0.9914 g/mL, titratable acidity 6.0 g/L tartaric acid, pH 3.68, and volatile acidity 0.58 g/L acetic acid, free sulphur dioxide 35 mg/L and total sulphur dioxide 96 mg/L. Oenological parameters were analysed using an FTIR Bacchus Micro (Microderm, France). 2.6. Headspace wine aroma composition by solid-phase microextraction (HS-SPME) For the determination of red wines headspace aroma composition, a validated SPME method using Divinylbenzene/Carboxen/Polydimethylsiloxane 50/30 μm coated fibre (Vás, Gál, Harangi, Dobó, & Vékey, 1998), confirmed in our laboratory, was used (Filipe-Ribeiro et al., 2017b). To 10 mL of wine, 2 g of NaCl was added and the extraction was performed at 35ºC for 60 min before thermal desorption in the GC-MS injector at 270ºC for 3 min in splitless mode. Identification of the substances present in the headspace aroma of wines was performed by comparison of the retention times and mass spectra with those of authentic standards. Additionally, Kovat’s retention indexes were calculated by analysing a mixture of alkanes using the same chromatographic method.

15

The area was used for the relative quantification of wine headspace aroma compounds. All analyses were performed in quadruplicate. 2.7. 4-EP and 4-EG determination by liquid-liquid extraction and GC-MS analysis The extractions were performed by using 2mL of pentane/diethyl ether (2:1) to 20 mL of wine following the methodology described by Milheiro et al. (2017b). The internal standard method, using 3,4-dimethylphenol as internal standard, was used for quantification. Analyses were performed in quadruplicate. 2.8. Colour and chromatic characterisation Colour intensity and hue were determined according to OIV (2009). Wine chromatic characteristics were calculated using the CIELab method according to OIV (2009). 2.9. High-performance liquid chromatography (HPLC) analysis of anthocyanins and phenolic acids Analyses were performed according to Guise et al. (2014), and quantification was performed according to Filipe-Ribeiro et al. (2017b). 2.10. Sensory evaluation For sensory analysis nine attributes were selected: visual (limpidity, hue, colour intensity and oxidised), aroma (fruity, floral, vegetable, phenolic and oxidised aroma) using an adapted tasting sheet based on that recommended by the OIV (http://www.oiv.int/public/medias/3307/review-onsensory-

analysis-of-wine.pdf).

Attributes were quantified using a five-point intensity scale (ISO 4121, 2003), that was anchored with the terms “low intensity” for score one and “high intensity” for score five. For visual hue attribute, the five-point scale was anchored with violet-purple for score one and brick red-brown for score five. Panellists scored only integer values and two different tasting sessions were performed. The panel was composed of six experts (ISO

16

6658, 1985). Evaluations were performed in an individual booth from 10:00 a.m.to 12:00 p.m. (ISO 8589, 2007) using the recommended glassware according to ISO 3591 (1977). Wines were presented in random order (ISO 6658, 1985) and a volume of 50 mL of wine was used for experts to evaluate the wine (ISO 3591, 1977). Panellists consistency (Cindex) was evaluated by consonance analysis (Dijksterhuis, 1995). 2.11. Statistical treatment Statistically significant differences between means were determined by analysis of variance (ANOVA, one-way) followed by Tukey’s honestly significant difference (HSD, 5% level) post-hoc test for the physicochemical data and a post-hoc Duncan test for sensory data. All analyses were performed using Statistica 7 Software (StatSoft, Tulsa, OK U.S.A.). Non-linear regression of the experimental adsorption data for fitting the tested adsorption models was performed using GraphPad Prism Version 6.05 (Graph-Pad Software Inc., USA). The fitting performance was accessed by testing the residuals normal distribution using the D’Agostino-Pearson omnibus K2 normality test. The skewness of the parameters obtained for each adsorption model obtained by nonlinear regression was measured by Hougaard’s method (Hougaard, 1985) with values less than 0.10 being ideal. For model comparison the extra-sum of squares F test was used (Equation 5): 𝐹=

(𝑆𝑆1 − 𝑆𝑆2 )/𝑆𝑆2 (Equation 5) (𝐷𝐹1 − 𝐷𝐹2 )/𝐷𝐹2 Equation 5

with DF1-DF2 degrees of freedom for the numerator and DF2 degrees of freedom for the denominator, where 1 and 2 correspond to the simpler and more complex models to be evaluated. The F ratio quantifies the relationship between the relative increase in sum-ofsquares and the relative increase in degrees of freedom.

3. Results and Discussion 17

3.1. Physicochemical characterisation of NIPs and MIPs The physisorption data obtained from liquid nitrogen adsorption/desorption curves for MIPs and NIPs are presented in Figure S.1. and the corresponding specific surface area, SBET, mesopore area, Smeso, specific pore volume, Vp, and average pore size, Dp, are listed in Table 1. Most of the samples follow a type II nitrogen adsorption/desorption isotherm, featured by a hysteresis loop attributed to capillary condensation in mesopores (Figure S.1.). The polymers SBET surface was mainly attributed to the presence of mesopores (Table 1) formed due to phase separation between the polymer and the solvents during the polymerisation process (Nabavi, Vladisavljevic, Eguagie, Li, Georgiadou, & Manovic, 2016). Although the role of template in the creation of nanocavities is known, the effect of a template on the structure of mesopores is not well understood (Nabavi et al., 2016). Comparing the total pore volume and SBET of MIP samples and their NIP counterparts, no specific trend was observed, and both smaller and larger pore volumes of NIPs were observed (Table 1), in accordance with previous works (Dirion, Cobb, Schillinger, Andersson, & Sellergren, 2003; Nabavi et al., 2016). Therefore, it might not be possible to evaluate the separation performance of MIP and NIP particles by comparing their total pore volume and SBET. The morphology of the particles obtained with the different functional monomers was observed by SEM (Figure 2). NIPs and MIPs produced with 4-EG as a template showed similar morphologies presenting segmented cauliflower-like microparticles. On the other hand, the particles synthesised using 4-EP as a template presented a different morphology, although also presenting agglomerated particles, their shape is less smooth and more needle-like. The absence of distinct microspheres in our system can be explained either by the relative amount of cross-linker used but also because the solvent

18

system polarity used (acetonitrile:water) is strong, resulting in a low solubility of the molecules used for the synthesis of the polymers, resulting in the formation of particle aggregates (Cai, & Gupta, 2004). This result is similar to those previously reported, being known that the concentrations and chemical structure of the template, functional and crosslinking monomers greatly affect the polymer configuration (Jiang, & Tong, 2004). The NIPs FTIR spectra were identical to those of the corresponding MIPs (Figure 3). Also, the FTIR spectrum of the three NIPs were similar, except for the band intensity at 3469 cm-1 and near 1675 cm-1 for NIP1, attributed to the N-H stretching vibrations of the amide group of methacrylamide. The peak at around 3465 cm-1 on NIP2 and NIP3 spectra can be attributed to O–H stretching vibrations of adsorbed water and contribute also to the absorption at this wavenumber for NIP1. The two peaks at 1163 cm-1 and 1265 cm-1 were attributed to C–O–C stretching vibrations. The peak at 1730 cm-1 was attributed to C=O bonds present in all functional monomers and EDMA and the peaks at 2962 and 2999 cm-1 were attributed to C–H stretching vibrations. There is no clear peak in the 1680–1640 cm-1 range, corresponding to C=C stretch in EDMA and functional monomers, meaning that there was no monomer or cross-linker left in the polymer particles suggesting that all carbon-carbon double bonds were consumed during the reaction or removed by the cleaning process. From the XRD spectra (Figure 4) of the synthesised NIPs and MIPs it can be observed that the polymers obtained were a mixture of crystalline and amorphous phases (Yanti, Nurhayati, Royani, Widayani, & Khairurrijal, 2016). The polymer synthesis in the presence of different templates resulted in polymers with different crystallinities, depending on the functional monomer used for the synthesis. For the polymers synthesised with methacrylamide as the functional monomer, the synthesis in the presence of 4-EG as a template resulted in an increase in the crystallinity of MIP1EG in

19

comparison to NIP1 and MIP1EP. The MIP1EP presented a similar crystallinity to NIP1. For the polymers synthesised with EDMA as a functional monomer, the synthesis in the presence of both templated increased the crystallinity of the polymers (MIP3EG and MIP3EP) when compared to NIP3, and especially the MIP produced in the presence of 4-EG that was the polymer with the highest crystallinity of all the polymers synthesised (MIP3EG). For all the polymers synthesised with ethyl methacrylate as a functional monomer, there were no differences in the crystallinity.

3.2. Adsorption isotherms of NIPs and MIPs and evaluation of the imprinting factor As shown in Figure 5, the MIPs and NIPs adsorption capacity increased in the whole concentration range assayed (2.5 to 100 mg/L of 4-EP and 4-EG in model wine solution). Comparing MIP 1, 2 and 3 for the adsorption of 4-EP and 4-EG, it can be observed that in the concentration range analysed, MIP 1 presented a lower adsorption capacity for both compounds when compared to MIP2 and MIP3. For all MIPs the adsorption capacity of 4-EG was lower than the adsorption of 4-EP (Tables S.1 to S.3. and Table 2). A comparison of the isotherm models for both VPs adsorption for MIPs and NIPs using nonlinear regression (Equations 6, 7 and 8) are given in Tables S.1., S.2., and S.3. For all models and the three polymers, the curves obtained were not satisfactorily described by one curve (p<0.0001), showing that the adsorption curves obtained for the NIP and MIPs were significantly different for each functional monomer and the two imprinting molecules. All models yielded high correlation coefficients (>0.999) and therefore for model comparison the extra-sum of squares F test was used (Table S.1., S.2., and S.3.). Except for the adsorption isotherms of NIP3 for both VPs, the LangmuirFreundlich isotherm always yielded a significantly higher fit of the experimental data than that obtained when using the Langmuir and Freundlich isotherms. As can be observed

20

also the n values obtained in the Langmuir-Freundlich model were significantly lower than 1, except for NIP3 for 4-EP adsorption, showing that the active site distribution of the NIPs and MIPs were heterogeneous. The binding capacity (KLxQm) of each NIP and MIP synthesised were calculated (Table 2) and used to compare the NIP and MIPs performance and calculate the imprinting factor (Baggiani, Giraudi, Giovannoli, Tozzi, & Anfossi, 2004). The polymer that yielded a higher imprinting factor was MIP3 either using 4-EP or 4-EG as templates, although for this MIP the 4-EG yielded a higher imprinting effect. The polymers synthesised with the functional monomer 1 also yielded an imprinting factor that ranged from 1.41 to 1.78, but significantly lower than the imprinting obtained with the functional monomer 3. Also, the polymers synthesised with monomer 2 showed an imprinting effect when using 4-EP as a template although with lower values that obtained for MIP3 and MIP1 (1.21 and 1.35), and contrarily to the other two MIPs, when 4-EG was used as a template molecule no imprinting effect was observed. The effect of the template used on the different imprinting efficiency of the MIPs obtained can be explained because template molecules with different steric structures have different affinities with the functional monomer with the same amounts of added template molecule and functional monomer (Jiang, & Tong, 2004). There was no correlation between the NIPs and MIPs binding capacity and the surface area of the polymers (SBET). This lack of correlation between the adsorption of the NIPs and MIPs and the surface area has been observed in other works (Dirion et al., 2003; Nabavi et al., 2016).

3.3. Effect of NIPs and MIPs in the removal of VPs in spiked red wine To evaluate the performance of the synthesised MIPs and compare their efficiency to the corresponding NIPs, their efficiency in VPs removal in red wine was determined.

21

The dosage used was selected to remove completely VPs, according to the adsorption isotherms determined by FA (Table S.4.), considering a linear dependence of the adsorption capacity on the level of adsorbent used. The VPs removal efficiency by MIPs and NIPs are presented in Table S.4. and Figure 6. The removal of VPs from wine was dependent on the VPs contamination level, with higher contamination levels resulting in higher adsorption on MIPs and NIPs, as expected. MIP1 presented a lower removal efficiency when compared to MIP2 and MIP1 and MIP2 presented a lower removal efficiency than MIP3 (Figure 6). Nevertheless, the VPs removal observed in wine was significantly lower than that expected from the adsorption curves for each NIP and MIP. This decrease in the adsorption efficiency of 4-EP and 4-EG by the synthesised materials can be explained by the adsorption of the structurally related phenolic compounds present in higher amounts in the wine such as coutaric acid, p-coumaric acid, p-coumaric acid ethyl ester and in lower extension caftaric acid, caffeic acid and caffeic acid ethyl ester (Table 5, Supplementary material Figure S.2.). Although the adsorption of VPs in wine by the MIPs and NIPs was lower than that predicted from the adsorption isotherms, there could still be observed an effect of the imprinting of MIPs dependent on the VPs spiking level (Table S.4.). The adsorption of VPs observed for NIP3 and MIP3 in red wine presented interesting high values (29-40%; 54-63% for the low and high spiking levels, respectively) in the range of that observed for activated carbons (Filipe-Ribeiro et al., 2017a) and other adsorbents assayed previously for dealing with this problem (CarrascoSanchez et al., 2015; Larcher et al., 2012; Marican et al., 2014; Milheiro et al., 2017b). The removal efficiency was in the range of that obtained by Teixeira et al. (2015) (5456% for MIP;40-45% for NIP for 1659 g/L 4-EP and 149 g/L 4-EG) using a MIP synthesised with vinylpiridine as a functional monomer and EDMA as cross-linker and 4-EP as a template and lower than that obtained by Garde-Cerdán et al. (2008) using a

22

polymer synthesised also with vinylpiridine, divinylbenzene as a cross-linker and trichloroanisol as a template (89-92% for MIP; 55-62% for NIP), nevertheless in this work the dose applied was higher than that used in this work (400 g/hL vs 250 g/hL), and the initial wine contamination level was not shown. Due to the water-insoluble nature of the polymers obtained and their higher density when compared to wine, the polymers were settled at the bottom of the recipient at the end of the treatment time. As they were removed by centrifugation at 10956 g for 10 min, no solid should have remained in the wine and this was supported by the lack of turbidity when the visible spectrum was acquired. At the industrial scale, these polymers can be handled similarly to the widely used synthetic water-insoluble polymer PVPP, usually removed by filtration or centrifugation, and therefore no safety issues are anticipated.

3.4. Effect of NIPs and MIPs on the instrumental colour intensity, phenolic compounds, and monomeric anthocyanins All NIPs and MIPs resulted in a decrease of the instrumental colour intensity (Table 3), nevertheless, the decrease observed for the NIPs and MIPs were limited (14 to 19% reduction). This decrease in colour intensity was in line with the decrease observed in total monomeric anthocyanins of wine treated with NIPs and MIPs (5% decrease for NIP3 to 8% for MIP2EG, Table 4). A significant correlation was observed between colour intensity and monomeric anthocyanins levels in wines (r=0.902, p<0.00036). In line with the colour intensity change, there was also observed an increase in wine lightness and redness (Table 3). No significant correlation between the reduction in monomeric anthocyanins and SBET, Smeso and Vp was observed, nevertheless for the colour intensity there was observed a significant correlation with the polymers Vp (r=0.723, p<0.028). For

23

phenolic acids and catechin there was also observed a decrease in their content in wines by application of NIPs and MIPs ranging from 7.7% for NIP1EG to 23% for MIP2EP, although gallic acid (8.0% for MIP3EG to 28% for MIP5EP), catechin (0.2% for NIP1 to 50% to MIP3EG) and p-coumaric acid ethyl ester (38% for NIP1 to 50% NIP1EG) were more affected than the other phenols (Table 5). There was no correlation between the reduction in the total phenols determined by HPLC and SBET, Smeso and Vp of the polymers used. Teixeira et al. (2015) observed a 29% (NIP) to 44% (MIP) reduction in the phenolic compounds determined by HPLC. Also, they observed a decrease in monomeric anthocyanins for the wines treated with NIP and MIPs from 18% to 25%, respectively (Table 4). Garde-Cerdán et al. (2008) didn’t measure the impact of their MIP on the phenolic compounds and monomeric anthocyanins. None of the two-previous works measured the impact of the wine treatment with MIPs on the colour intensity and chromatic characteristics of the MIP-treated wines.

3.5. Effect of NIPs and MIPs on the headspace abundance (HA) of aroma compounds As observed in Table 6, all NIPs and MIPs decreased the HA of the wine aroma compounds including the HA of VPs that for MIP3EG reached 68% and 63% for 4-EP and 4-EG, respectively. There was observed a significant correlation between the decrease in the HA of the VPs and the removal efficiency of the different NIPs and MIPs (r=0.832, p<0.054 4-EP; r=0.917, p<0.00050 4-EG). For the high spiking levels, the same trend in HA decrease was observed (r=0.951, p<0.001 4-EP; r=0.971, p<0.0001 4-EG), as expected from the removal observed for the total amount of 4-EP and 4-EG (Table 6, data not shown). Although there was observed a decrease in all the aroma compounds by application of NIPs and MIPs, some aromas were more affected than others. For example, ethyl decanoate abundance on the headspace aroma was decrease to undetectable levels

24

for all NIPs and MIPs and ethyl dodecanoate was not detected in the headspace aroma of wines treated with NIP3 and MIP3. Also, the decrease in the HA aroma of 4-EP and 4EG was higher than that observed for ethyl acetate (relative HA reduction 1.9-3.0), 3methylbutan-1-ol acetate (1.4-2.7), ethyl hexanoate (1.1-1.5), diethylsuccinate (1.3-2.0); 2-phenylethanol (1.6-1.9). Oppositely, higher reductions in HA were observed for ethyloctanoate (0.6-1.1), and octanoic acid (0.6-0.9). Except for ethyl acetate and 3-methylbutyl acetate where it was observed a significant correlation between the reduction in HA and the SBET of the polymers (r=0.731, p<0.034 and r=-0.834, p<0.0051, respectively), no correlation was observed for the HA of the other aroma compounds and the NIPs and MIPs SBET. The reduction in the total HA of aroma compounds of the NIPs and MIPs synthesised in our work ranged from 25% for NIP1 to 50% for MIP3EP, in the range of that observed by Teixeira et al. (2015) for the wines treated with a 4-vinylpiridine-EDMA molecularly imprinted polymer (36% reduction), and lower than that observed by GardeCerdán et al. (2008) where the reduction observed in the aroma compounds analysed ranged from 31 to 93%.

3.6. Effect of NIPs and MIPs treatment on wine sensory attributes To gain insights into the impact of the wine chemical changes observed by the application of MIPs and NIPs on their sensory profile, wines treated with 250 g/hL of NIP 2 and NIP3 and of MIP2EP and MIP3EG were subjected to sensory analysis by an expert panel. These MIPs were selected as they presented the higher reduction efficiency of 4-EP and 4-EG when applied to wine and also presented a higher adsorption capacity when analysed by FA. The correspondent NIPs were also analysed to ascertain if the observed increase in the absorption capacity of MIPs in relation to NIPs could be detected

25

sensory. For ethical reasons, only the visual and aroma sensory wine attributes were evaluated, as these polymers are not yet approved for food use. Only wines spiked with lower levels of VPs were subjected to sensory analysis as in these wines the tested MIPs were able to reduce the amount of these contaminants below their ODT reported in the literature (605 g/L for 4-EP; 110 g/L for 4-EG, Chatonnet et al., 1992). The presence of VPs in wines impacted significantly and negatively on their aroma profile (TF, Table 7), as the phenolic attribute was significantly increased and the wine fruity and floral attributes decreased significantly. A good panel consensus on the analysed sensory attributes was obtained (C-index >1; Table 7). For wines treated with NIPs 2 and 3 and MIPs 2 and 3, there was a significant decrease in the phenolic attribute, significantly different from TF (Table 7), although also significantly different from T0. Although on average MIPs presented a lower phenolic attribute than NIPs, only for the wine treated with MIP3 there was a significantly lower phenolic attribute when compared to the correspondent NIP3. These results were in accordance with the removal efficiency of VPs by these NIPs and MIPs (Figure 6). The treatment of wines with NIPs and MIPs also decrease the oxidised attribute of wines imparted by the addition of VPs to wine (Table 7). Wine treated with MIP3EG was the only one with an oxidised attribute significantly different from TF. The decrease in the phenolic attribute of the wines treated with NIPs and MIPs increased the wines’ fruity and floral attributes (Table 7). For the fruity attribute, all treated wines presented a significantly higher value than the wine spiked with VPs (TF), although significantly lower than that observed for the control wine (T0). Again, wines treated with the MIPs presented on average a higher fruity score than the wines treated with the correspondent NIPs, and again only for MIP3 this difference was significant (Table 7). For the floral attribute although all NIPs and MIPs presented an average higher score, only for the

26

MIP2EP the score was significantly higher than that of the spiked wine (TF), although the other wines treated with MIP3EG, NIP2 and NIP3 were not significantly different from MIP2EP. For the colour intensity, hue, limpidity and oxidised attributes the NIPs and MIPs didn’t change significantly these wine attributes (Table 7).

4. Conclusions New MIPs were synthesised by precipitation polymerisation using 4-EP or 4-EG as templates. The MIP synthesised using ethylene glycol methyl ether acrylate as functional monomer and 4-EG as a template, presented a 3.2-4.7-fold higher 4-EP and 4-EG adsorption capacity when compared to the correspondent NIP. Application to red wine resulted in 38-63% removal of VPs, depending on the wine VPs levels. Although this MIP decreased wine colour intensity (14%) and aroma HA (51%), sensory analysis of the VPs spiked wine treated with this MIP resulted in a significant decrease of the phenolic attribute and increase of the fruity and floral attributes, with no significant differences in the wine colour perceived by the expert panel. The sensory improvement of the MIP was significantly higher than that observed for the correspondent NIP. These results show that the design of highly selective specific adsorption polymers by imprinting can be a good solution to deal with the negative sensory impact of VPs. The use of molecularly imprinted polymers is a promising technology for wine fining as well as for other food applications enabling the removal of sensory unwanted or toxic components, not limited to volatile compounds, without affecting or with low impact in other components present and in this way improving the wine quality and safety.

Acknowledgements

27

We acknowledge Aveleda S.A. for supplying the wine used in this study and SAI Enology for providing FTIR analyses. We appreciate the financial support provided to the Research Unit in Vila Real (PEst-OE/QUI/UI0616/2014) by FCT and COMPETE

5. References Avramova, M., Vallet-Courbin, A., Maupeu, J., Masneuf-Pomarède, I., & Albertin, W. (2018). Molecular diagnosis of Brettanomyces bruxellensis' sulfur dioxide sensitivity through genotype specific method. Frontiers in Microbiology, 9, (1260), 1-9. Baggiani, C., Giraudi, G., Giovannoli, C., Tozzi, C., & Anfossi, L. (2004). Adsorption isotherms of a molecular imprinted polymer prepared in the presence of a polymerisable template: Indirect evidence of the formation of template clusters in the binding site. Analytica Chimica Acta, 504, 43–52. Bailley, V., Jerkovic, J., Marchand-Brynaert, S., & Collin, (2006) Aroma extraction dilution analysis of sauternes wines. Key of role polyfunctional thiols. Journal of Agricultural and Food Chemistry, 54, 7227–7234. Cai, W.S., & Gupta, R. B. (2004). Molecularly-imprinted polymers selective for tetracycline binding. Separation and Purification Technology, 35, 215-221. Carrasco-Sánchez, V., John, A., Marican, A., Santos, L.S., & Laurie, V.F. (2015). Removal of 4-ethylphenol and 4-ethylguaiacol with polyaniline-based compounds in wine-like model solutions and red wine. Molecules, 20, 14312-14325. Cartwright, Z.M., Glawe, D.A., & Edwards, C.G. (2018). Reduction of brettanomyces bruxellensis populations from oak barrel staves using steam. American Journal of Enology and Viticulture, 69 (4), 400-409. Chassagne, D., Guilloux-Benatier, M., Alexandre, H., & Voilley, A. (2005). Sorption of wine volatile phenols by yeast lees. Food Chemistry, 91(1), 39–44.

28

Chatonnet, P., Dubourdieu, D., Boidron, J., & Poins, M. (1992). The origin of ethylphenols in wines. Journal of the Science of Food and Agriculture, 60 (2), 165178. Chatonnet, P., Viala, C., & Dubourdieu, D. (1997). Influence of polyphenolic components of red wines on the microbial synthesis of volatile phenols. American Journal of Enology and Viticulture, 48, 443-448. Chen, L., & Li, B. (2012). Application of magnetic molecularly imprinted polymers in analytical chemistry. Analytical Methods, 4, 2613-2621. Crauwels, S., Van Opstaele, F., Jaskula-Goiris, B., Steensels, J., Verreth, C., Bosmans, L., Paulussen, C., Herrera-Malaver, B., de Jonge, R., De Clippeleer, J., Marchal, K., De Samblanx, G., Willems, K.A., Verstrepen, K.J., Aerts, G., & Lievens, B. (2017). Fermentation assays reveal differences in sugar and (off-) flavor metabolism across different Brettanomyces bruxellensis strains. FEMS Yeast Research. 17 (1), 1-10. Cravero, F., Englezos, V., Rantsiou, K., Torchio, F., Giacosa, S., Río Segade, S., Gerbi, V., Rolle, L., & Cocolin, L. (2018). Control of Brettanomyces bruxellensis on wine grapes by post-harvest treatments with electrolyzed water, ozonated water and gaseous ozone. Innovative Food Science and Emerging Technologies 47, 309-316. Czerny, M., Brueckner, R., Kirchoff, E., Schmitt, R., & Buettner, A. (2011). The influence of molecular structure on odor qualities and odor detection thresholds of volatile alkylated phenols. Chemical Senses, 36, 539–553. Dijksterhuis, G. (1995). Assessing panel consonance. Food Quality and Preference, 6, 714. Dimopoulou, M., Hatzikamari, M., Masneuf-Pomarede, I., & Albertin, W. (2019). Sulfur dioxide response of Brettanomyces bruxellensis strains isolated from Greek wine. Food Microbiology, 78, 155-163.

29

Dirion, B., Cobb, Z., Schillinger, E., Andersson, L.I., & Sellergren, B. (2003). Water compatible molecularly imprinted polymers obtained via high-throughput synthesis and experimental design. Journal of the American Chemical Society, 125, 15101– 15109. Dragone, G., Mussato, S. I., Oliveira, J.M., &Teixeira, J. A. (2009). Characterization of volatile compounds in an alcoholic beverage produced by whey fermentation. Food Chemistry, 112, 929-935. Edwards, C.G., & Cartwright, Z.M. (2019). Application of heated water to reduce populations of Brettanomyces bruxellensis present in oak barrel staves. South African Journal of Enology and Viticulture, 40 (1), 8p. Englezos, V., Rantsiou, K., Torchio, F., Pollon, M., Giacosa, S., Río Segade, S., Gerbi, V., Rolle, L., & Cocolin, L. (2019). Efficacy of ozone against different strains of Brettanomyces bruxellensis on winegrapes postharvest and impact on wine composition. American Journal of Enology and Viticulture, 70 (3), 249-258. Filipe-Ribeiro, L., Cosme F., & Nunes, FM. (2019a). A simple method to improve cork powder waste adsorption properties: valorisation as a new sustainable wine fining agent. ACS Sustainable Chemistry & Engineering, 7, 1105-1112. Filipe-Ribeiro, L., Cosme, F., & Nunes, F.M. (2018). Reducing the negative sensory impact of volatile phenols in red wine with different chitosans: Effect of structure on efficiency. Food Chemistry, 242, 591–600. Filipe-Ribeiro, L., Cosme, F., & Nunes, F.M. (2019b). Air Depleted and Solvent Impregnated Cork Powder as a New Natural and Sustainable Wine Fining Agent, Advances in Grape and Wine Biotechnology, Antonio Morata and Iris Loira, IntechOpen,

30

Filipe-Ribeiro, L., Milheiro, J., Ferreira, L.C., Correia, E., Cosme, F., & Nunes, F.M. (2019). Biogenic amines and polyamines in wines: Does Dekkera/Brettanomyces red wine spoilage increases the risk of intake by consumers? LWT- Food Science and Technology, 115, 108488. Filipe-Ribeiro, L., Milheiro, J., Matos, C.C., Cosme, F., & Nunes, F.M. (2017a). Reduction of 4-ethylphenol and 4-ethylguaiacol in red wine by activated carbons with different physicochemical characteristics: Impact on wine quality. Food Chemistry, 242-251. Filipe-Ribeiro, L., Milheiro, J., Matos, C.C., Cosme, F., & Nunes, F.M. (2017b). Data on changes in red wine phenolic compounds, headspace aroma compounds and sensory profile after treatment of red wines with activated carbons with different physicochemical characteristics. Data in Brief, 12, 188-202. Gallardo-Chacón, J. J., & Karbowiak, T. (2015). Sorption of 4-ethylphenol and 4ethylguaiacol by suberin from cork. Food Chemistry, 181, 222−226. García-Calzón, J.A., & Díaz-García, M.E. (2007). Characterization of binding sites in molecularly imprinted polymers. Sensors and Actuators B, 123, 1180–1194. Garde-Cerdán, T., Zalacain-Aramburu, A., Lorenzo-Tendero, C., Serrano, J. L. A., & Salinas, M. R. F. (2008). Molecularly imprinted polymer-assisted simple clean-up of 2,4,6-trichloroanisole and ethylphenols from aged red wines. American Journal of Enology and Viticulture, 59, 396-400. Guise, R., Filipe-Ribeiro, L., Nascimento, D., Bessa, O., Nunes, F.M., & Cosme, F. (2014). Comparison between different types of carboxylmethylcellulose and other oenological additives used for white wine tartaric stabilization. Food Chemistry, 156, 250-257.

31

Guzzon, R., Larcher, R., Guarcello, R., Francesca, N., Settanni, L., & Moschetti, G. (2018). Spoilage potential of brettanomyces bruxellensis strains isolated from Italian wines. Food Research International, 105, 668-677. Haupt, K. (2012). Molecular Imprinting, 325, SpringerVerlag Berlin Heidelberg, 1-28. Hougaard, P. (1985). The appropriateness of the asymptotic distribution in a nonlinear regression model in relation to curvature. Journal of the Royal Statistical Society. Series B (Methodological), 103-114. ISO International Organization for Standardization. Switzerland. Norm: ISO 3591. (1977). Sensory analysis-apparatus Wine-tasting glass; ISO 6658. (1985). Sensory analysis – Methodology – General guidance; ISO 4121. (2003). Sensory analysis guidelines for the use of quantitative response scales; ISO 8589. (2007). Sensory analysis general guidance for the design of test rooms. Jiang, B., & Zhang, Z. (2010). Volatile compounds of young wines from Cabernet Sauvignon, Cabernet Gernischet and Chardonnay varieties grown in the Loess Plateau Region of China. Molecules, 15, 9184-9196. Jiang, Y., & Tong, A.J.J. (2004). Synthesis of molecularly imprinted microspheres for recognition of trans-aconitic acid. Journal of Applied Polymer Science, 94, 542-547. Križanović, S., Tomašević, M., Režek Jambrak, A., Ćurko, N., Gracin, L., Lukić, K., & Kovačević Ganić, K. (2019). Effect of thermosonication and physicochemical properties of wine on culturability, viability, and metabolic activity of Brettanomyces bruxellensis yeast in red wines. Journal of Agricultural and Food Chemistry doi.org/10.1021/acs.jafc.9b03661 Larcher, R., Puecher, C., Rohregger, S., Malacarne, M., & Nicolini. G. (2012). 4Ethylphenol and 4-ethylguaiacol depletion in wine using esterified cellulose. Food Chemistry, 132, 2126–2130.

32

Lisanti, M. T., Gambuti, A., Genovese, A., Piombino, P. & Moio, L. (2017). Treatment by fining agents of red wine affected by phenolic off odour. European Food Research and Technology 242(3), 501–510. Marican, A., Carrasco-Sanchez, V., John, A., Laurie, V. F., & Santos. L. S. (2014). The binding of 4-ethylguaiacol with polyaniline-based materials in wines. Food Chemistry, 159, 486–492. Milheiro, J., Filipe-Ribeiro, L., Cosme, F., & Nunes, M.F. (2017b). A simple, cheap and reliable method for control of 4-ethylphenol and 4-ethylguaiacol in red wines. Screening of fining agents for reducing volatile phenols levels in red wines. Journal of Chromatography B, 1041-1042, 183-190. Milheiro, J., Filipe-Ribeiro, L., Vilela, A., Cosme, F., & Nunes, F. M. (2017a). 4Ethylphenol, 4-ethylguaiacol and 4-ethylcatechol in red wines: Microbial formation, prevention, remediation and overview of analytical approaches. Critical Reviews in Food Science and Nutrition, DOI: 10.1080/10408398.2017.1408563. Nabavi, S.A., Vladisavljevic, G.T., Eguagie, E.M., Li, B., Georgiadou, S., & Manovic, V. (2016). Production of spherical mesoporous molecularly imprinted polymer particles containing tunable amine decorated nanocavities with CO2 molecule recognition properties. Chemical Engineering Journal, 306, 214–225. Nieto-Rojo, R., Ancın-Azpilicueta, C., & Garrido. J. J. (2014). Sorption of 4ethylguaiacol and 4-ethylphenol on yeast cell walls, using a synthetic wine. Food Chemistry , 152, 399–406. OIV (2009). Récueil de Méthodes Internationeiles d’Analyse des Vins et des Moûts: Method OIV-MA-AS2-11:R2006 and OIV-MA-AS2-07B:R2009 Paris: Edition Officielle.

33

OIV. (2015) updated in 2015 http://www.oiv.int/public/medias/3307/review-on-sensoryanalysis-of-wine.pdf. Oro, L., Canonico, L., Marinelli, V., Ciani, M., & Comitini, F. (2019). Occurrence of Brettanomyces bruxellensis on grape berries and in related winemaking cellar. Frontiers in Microbiology, 10 (415), 1-10. Pardeshi, S., & Singh, S. K. (2016). Precipitation polymerization: A versatile tool for preparing Molecularly Imprinted Polymer beads for chromatography applications. RSC Advances, 28, 23094-23980. Perestrelo, A. Fernandes, F. F. Albuquerque, J. C. Marques, J. S., & Câmara, J. (2006). Analytical characterization of the aroma of Tinta Negra Mole red wine: Identification of the main odorants compounds. Analytical Chimical Acta, 563, 154-164. Petrozziello, M., Asproudi, A., Guaita, M., Borsa, D., Motta, S., Panero, L., & Bosso, A., 2014. Influence of the matrix composition on the volatility and sensory perception of 4-ethylphenol and 4-ethylguaiacol in model wine solutions. Food Chemistry, 149, 197–202. Portugal, C., Saenz, Y., Rojo-Bezares, B., Zarazaga, M., Torres, C., Cacho, J., & RuizLarrea, F. (2014). Brettanomyces susceptibility to antimicrobial agents used in winemaking: in vitro and practical approaches. European Food Research and Technology, 238, 641–652. Renouf, V., Strehaiano, P., & Lonvaud-Funel, A. (2008). Effectiveness of dimethlydicarbonate to prevent Brettanomyces bruxellensis growth in wine. Food Control, 19, 208–216. Rudzinske, W., & Everett, D.H. (1992). Adsorption of gases on the heterogeneous surface, Academic Press, New York.

34

Sajonz, P., Zhong, G., & Guiochon, G. (1996). Influence of the concentration dependence of the mass transfer properties on chromatographic band profiles II. Accuracy of the determination of isotherm data by frontal analysis. Journal of Chromatography A, 731, 1-25. Schumaker, M. R., Chandra, M., Malfeito-Ferreira, M., & Ross, C. F. (2017). Influence of Brettanomyces ethylphenols on red wine aroma evaluated by consumers in the United States and Portugal. Food Research International, 100, 161-167. Schumaker, M. R., Diako, C., Castura, J. C., Edwards, C. G., & Ross, C. F. (2019). Influence of wine composition on consumer perception and acceptance of Brettanomyces metabolites using temporal check-all-that-apply methodology. Food Research International, 116, 963-972. Teixeira, R., Dopico-García, S., Andrade, P.B., Valentão, P., López-Vilariño, J.M., González-Rodríguez V., Cela-Pérez, C., & Silva, L.R (2015). Volatile phenols depletion in red wine using molecular imprinted polymers. Journal of Food Science & Technology, 52(12), 7735-7746. Tempère, S., Schaaper, M.H., Cuzange, E., de Lescar, R., de Revel, G., & Sicard, G. (2016). The olfactory masking effect of ethylphenols: Characterization and elucidation of its origin. Food Quality and Preference, 50, 135-144. Umpleby, R, J.II, Baxter, S.C., Chen, Y., Shah, R.N., & Shimizu, K.D. (2001). Characterization of Molecularly Imprinted Polymers with the Langmuir-Freundlich Isotherm. Analytical Chemistry, 73, 4584-4591. van Wyk, S., & Silva, F.V.M (2017). High pressure processing inactivation of Brettanomyces bruxellensis in seven different table wines. Food Control, 81, 1-8. van Wyk, S., Farid, M.M., & Silva, F.V.M. (2018). SO2 high pressure processing and pulsed electric field treatments of red wine: Effect on sensory, Brettanomyces

35

inactivation and other quality parameters during one year storage. Innovative Food Science and Emerging Technologies, 48, 204-211. van Wyk, S., Silva, F.V.M., & Farid, M.M. (2019). Pulsed electric field treatment of red wine: Inactivation of Brettanomyces and potential hazard caused by metal ion dissolution. Innovative Food Science and Emerging Technologies, 52, 57-65. Varela, C., Bartel, C., Roach, M., Borneman, A., & Curtin, C. (2019). Brettanomyces bruxellensis SSU1 haplotypes confer different levels of sulfite tolerance when expressed in a Saccharomyces cerevisiae SSU1 null mutant. Applied and Environmental Microbiology, 85 (4), e02429-18. Vás, G., Gál, L., Harangi, J., Dobó, A., & Vékey, K. (1998). Determination of volatile compounds of Blaufrankisch wines extracted by solid-phase microextraction. Journal of Chromatographic Science, 36, 505–510. Vasapollo, G., Del Sole, R., Mergola, L., Lazzoi, M.R., Scardino, A., Scorrano, S., & Mele, G. (2011). Molecularly Imprinted Polymers: Present and Future Prospective. International Journal of Molecular, 12, 5908–5945. Wang, J., Cormack, P.A., Sherrington, D.C., & Khoshde, E. (2003). Monodisperse, molecularly imprinted polymer microspheres prepared by precipitation polymerization for affinity separation applications. Angewandte Chemie International Edition England. 10, 5336-5338. Yanti, Nurhayati, T., Royani, I., Widayani, & Khairurrijal. (2016). Synthesis and characterization of MAA-based molecularly-imprinted polymer (MIP) with D-glucose template. Journal of Physics: Conference Series, 739, 012143.

36

Figure Captions Figure 1 – Production scheme of molecularly imprinted polymers and the structure of the functional monomers (1-3) and cross-linker (4) used to produce molecularly imprinted polymers. 1 – methacrylamide; 2 - ethyl methacrylate; 3- ethylene glycol methyl ether acrylate; 4 - ethylene glycol dimethacrylate (EDMA).

Figure 2. Particles obtained in ethylene glycol dimethacrylate / methacrylamide / acetonitrile:water (4:1 v/v) system (MIP1); ethylene glycol dimethacrylate / ethyl methacrylate / acetonitrile:water (4:1 v/v) system (MIP2); ethylene glycol dimethacrylate / ethylene glycol methyl ether acrylate / acetonitrile:water (4:1 v/v) system (MIP3) visualised by SEM measurements. The bars correspond to 5 μm.

Figure 3. FTIR spectra of non-imprinted polymers synthesised by precipitation polymerisation.

Figure 4. X-ray diffraction pattern of NIP1 (black), MIP1EP (red) and MIP1EG (green) polymers (A), X-ray diffraction pattern of NIP2 (black), MIP2EP (red) and MIP2EG (green) polymers (B), X-ray diffraction pattern of NIP3 (black), MIP3EP (red) and MIP3EG (green) polymers (C)

Figure 5. Freundlich-Langmuir adsorption isotherms of non-imprinted and molecularly imprinted polymers for 4-EP and 4-EG synthesised using functional monomers 1, 2 and 3. Non-imprinted polymer (), for EP template (■) and for EG template (▲).

37

Figure 6. Amount of 4-ethylphenol (4-EP), 4-ethylguaiacol (4-EG) and total volatile phenols (Total) removal by application of 250g/hL of NIPs and MIPs to volatile phenols spiked wines at two levels (750 and 1500 g/L of 4-EP and 150 and 300 g/L of 4-EG) obtained for NIP1 and MIP1 (a), NIP2 and MIP2 (b), NIP3 and MIP3 (c). NIP (yellow), MIPEP (red) and MIPEG (green). Amount of 4-EP or 4-EG needed to be removed from wines in order reach the olfactory detection threshold of 605 g/L and 110 g/L for 4-EP or 4-EG (Chatonnet et al., 1992) (- - -).

38

Table 1. Physicochemical characteristics of non-imprinted and molecularly imprinted polymers synthesised. Smeso

Vmicro

(m2/g)

(cm3/g)

Vp

p/p0=0.95

Dp BJH (mode)

Samples

Yield

SBET (m2/g)

NIP1

76.8%

22310c

15010cd 0.0170.01a 0.1880.03ab

3.600.08a

MIP1EP

73.4%

24010c

16610bc 0.0120.01a 0.2010.02ab

3.600.08a

MIP1EG

74.8%

24210c

16210bc 0.0200.02a 0.1980.03ab

3.630.08a

NIP2

75.2%

33810a

20810a 0.0270.01a 0.2470.03a

3.610.08a

MIP2EP

82.7%

33210a

21310a 0.0320.02a 0.2590.04a

3.610.08a

MIP2EG

73.4%

29110b

19410ab 0.0200.01a 0.2230.03ab

3.610.08a

NIP3

87.1%

16110d

11510d 0.0040.01a 0.1420.03ab

3.570.08a

MIP3EP

75.6%

23910c

10010e 0.0750.01b 0.2340.01ab

3.570.08a

MIP3EG

66.9%

20210c

7510e

3.630.08a

(cm3/g)

0.0750.01b 0.1940.03ab

(nm)

Values are presented as mean±standard deviation (n=2); SBET: Brunauer-Emmett-Teller (BET) surface area; Smeso: surface area of mesopores; Vp: total volume of pores; Vmicro: micropore volume; Dp: average pore diameter. Means within a column followed by the same letter are not significantly different (Tukey p ˂ 0.05). 1 – methacrylamide; 2 - ethyl methacrylate; 3 - ethylene glycol methyl ether acrylate.

39

Table 2. Binding Capacity (BC) (L/g) of NIP and MIP and imprinting factor of the polymers synthesised. 4-EP

4-EG

(L/g)

(L/g)

Ratio t-

BCMIP/BCNIP

test

NIP1

0.164±0.040a

0.116±0.017a 1.41

Sig

4-EP

4-EG

MIP1EP

0.276±0.064b

0.164±0.036b 1.68

Sig

1.68±0.57

1.41±0.37

MIP1EG

0.292±0.049b

0.198±0.035c 1.47

Sig

1.78±0.56

1.71±0.39

NIP2

0.242±0.013a

0.224±0.011a 1.08

Sig

MIP2EP

0.324±0.035b

0.271±0.053b 1.20

Sig

1.35±0.16

1.21±0.24

MIP2EG

0.263±0.049a

0.262±0.044a -

Ns

1.08±0.21ns

1.17±0.20ns

NIP3

0.061±0.024a

0.074±0.023a -

Ns

MIP3EP

0.264±0.032b

0.199±0.039b 1.33

Sig

4.33±1.78

2.69±0.99

MIP3EG

0.285±0.028b

0.233±0.033c 1.22

Sig

4.66±1.88

3.15±1.08

Within the same column for each MIP, means with the same letters are not significantly different (t-test, p<0.05); ns – not significantly different from one (t-test, p<0.05); sig – significantly different from one (t-test, p<0.05).

40

Table 3. Chromatic characteristics of red wines spiked with volatile phenols (TF) and after treatment with NIPs and MIPs produced from different functional monomers and templates. Samples

L*

a*

b*

◦h

C*

ΔE

Colour intensity

Hue

A.U. …

11.01±0.02a

0.66±0.00a

0.70±0.00bc

5.19±0.30a

9.40±0.03bc

0.65±0.00a

58.68±0.54b

0.69±0.00d

5.82±0.62a

9.06±0.09e

0.65±0.00a

37.36±0.18bc

58.75±0.37b

0.69±0.00cd

5.76±0.43a

9.12±0.06de

0.65±0.00a

45.37±0.02bc

38.03±0.05bc

59.20±0.04b

0.70±0.00bc

5.87±0.03a

9.29±0.00cd

0.65±0.00a

14.99±0.17c

46.36±0.28c

38.03±0.32bc

59.97±0.42b

0.69±0.00d

7.17±0.38a

8.93±0.01e

0.65±0.00a

MIP2EG

14.18±0.12bc

45.48±0.16bc

38.14±0.07c

59.36±0.17b

0.70±0.00bc

6.03±0.21a

9.28±0.03cd

0.66±0.00a

NIP3

13.64±0.05b

44.83±0.08b

38.11±0.03c

58.84±0.04b

0.70±0.00bc

5.21±0.08a

9.52±0.00b

0.66±0.01a

MIP3EP

13.86±0.68b

45.08±0.79bc

37.38±1.13bc

58.56±1.33b

0.69±0.01cd

5.41±1.20a

9.29±0.09cd

0.65±0.01a

a

41.01±0.13

a

36.54±0.16

a

54.93±0.21

a

TF

10.46±0.07

NIP1

13.72±0.16b

44.93±0.25b

37.53±0.06bc

58.54±0.23b

MIP1EP

14.25±0.36bc

45.38±0.47bc

37.20±0.29bc

MIP1EG

14.17±0.24bc

45.34±0.33bc

NIP2

14.11±0.01bc

MIP2EP

0.73±0.00

a

13.97±0.16bc 45.31±0.23bc 38.35±0.15c 59.36±0.27b 0.70±0.00cd 5.83±0.31a 9.45±0.03bc 0.65±0.00a MIP3EG Values are presented as mean ± standard deviation; Means within a column followed by the same letter are not significantly different (Tuckey, p˂0.05). L* – lightness, a* redness, b* - yellowness, ΔE* –colour difference. The values corresponding to ΔE* were obtained taking as a reference the untreated wine (T), and wine treated with Mips (NIP1, MIPEP, MIP1EG, NIP2, MIP2EP, MIP2EG, NIP3, MIP3EP, and MIP3EG). A.U. – absorbance units

41

Table 4. Monomeric anthocyanin composition of red wines spiked with volatile phenols (TF) and after treatment with NIPs and MIPs produced from different functional monomers and templates. Samples

Del-3-Glc

Cya-3-Glc

Pet-3-Glc

(mg/L)

(mg/L)

(mg/L)

a

a

TF

0.90±0.04

9.78±0.46

15.87±1.14

NIP1

0.83±0.03a

9.06±1.10a

MIP1EP

0.79±0.02a

MIP1EG

a

Peo-3-Glc

Mal-3-Glc

(mg/L)

(mg/L) a

6.05±0.80

Del-3AcGlc

Cya-3AcGlc

(mg/L) a

(mg/L) a

a

Pet-3AcGlc

Peo-3AcGlc

(mg/L)

(mg/L)

Mal-3AcGlc

Del-3CoGlc

(mg/L) a

23.12±0.64

Cya-3CoGlc

Pet-3CoGlc

(mg/L) (mg/L)

(mg/L)

a

n.d

n.d

Peo-3CoGlc

Mal-3CoGlc

(mg/L) a

(mg/L)

1.20±0.11

a

1.08±0.08

10.70±0.46a

142.03±4.33

3.03±0.11

0.51±0.02

n.d.

1.33±0.00

14.36±0.23a

5.00±0.06ab 139.48±1.15a

2.94±0.18a

0.38±0.09a

n.d.

1.22±0.00ab 22.87±0.84a

n.d

n.d

0.78±0.14b

0.62±0.05c

6.27±0.42cd

9.21±0.43a

14.30±0.32a

4.83±0.31b 139.42±1.23a

2.77±0.06a

0.38±0.01a

n.d.

1.20±0.03b

23.02±0.63a

n.d

n.d

0.70±0.01b

0.57±0.00c

5.99±0.03cd

0.80±0.05a

9.20±0.42a

15.08±1.07a

4.96±0.05ab 137.50±1.73a

2.83±0.06a

0.44±0.02a

n.d.

1.21±0.00ab 19.40±4.45a

n.d

n.d

0.79±0.01b

0.56±0.01c

5.55±0.25d

NIP2

0.77±0.30a

8.24±0.09a

14.39±0.03a

5.20±0.10ab 138.22±0.03a

2.97±0.05a

0.50±0.01a

n.d.

1.23±0.03ab 22.73±0.06a

n.d

n.d

0.85±0.12b

0.62±0.00c

6.24±0.02cd

MIP2EP

0.93±0.09a

8.77±0.30a

14.20±0.52a

5.09±0.22ab 138.43±0.70a

2.83±0.09a

0.40±0.07a

n.d.

1.27±0.04ab 22.86±0.40a

n.d

n.d

0.86±0.01ab

0.55±0.00c

5.52±0.05d

MIP2EG

0.73±0.01a

8.69±0.82a

14.14±0.17a

5.10±0.11ab 136.51±0.09a

2.84±0.03a

0.48±0.08a

n.d.

1.22±0.03ab 22.52±0.31a

n.d

n.d

0.81±0.09b

0.54±0.02c

5.58±0.02d

NIP3

0.80±0.14a

8.87±0.63a

14.30±0.31a

5.24±0.10ab 138.24±0.91a

2.92±0.04a

0.41±0.03a

n.d.

1.25±0.00ab 23.61±0.28a

n.d

n.d

0.97±0.10ab

0.78±0.04b

7.65±0.22b

MIP3EP

0.80±0.14a

8.85±0.63a

14.35±0.08a

5.17±0.18ab 137.36±1.15a

2.89±0.07a

0.40±0.09a

n.d.

1.25±0.00ab 23.18±0.13a

n.d

n.d

0.86±0.10ab 0.66±0.06bc

6.66±0.15c

MIP3EG

0.73±0.02a

9.09±0.26a

14.47±0.30a

5.12±0.02ab 136.67±0.74a

2.92±0.01a

0.48±0.05a

n.d.

1.26±0.03ab 23.66±0.57a

n.d

n.d

0.93±0.00ab 0.65±0.08bc

6.71±0.03c

Values are presented as mean ± standard deviation; Del-3-Glc-Delphinidin-3-glucoside, Cya-3-Glc-Cyanidin-3-glucoside, Pet-3-Glc-Petunidin-3-glucoside, Peo-3-GlcPeonidin-3-glucoside, Mal-3-Glc-Malvidin-3-glucoside, Del-3-AcGlc-Delphinidin-3-acetylglucoside, Cya-3-AcGlc-Cyanidin-3-acetylglucoside, Pet-3-AcGlc-Petunidin-3acetylglucoside, Peo-3-AcGlc-Peonidin-3-acetylglucoside, Mal-3-AcGlc-Malvidin-3-acetylglucoside, Del-3-CoGlc-Delphidin-3-coumaryl-glucoside, Cya-3-CoGlc-Cyanidin3-coumaroylglucoside, Pet-3-CoGlc-Petunidin-3-coumaroylglucoside, Peo-3-CoGlc-Peonidin-3-coumaroylglucoside; Mal-3-CoGlc-Malvidin-3-coumaroylglucoside. Means within a column followed by the same letter are not significantly different ANOVA and Tuckey post-hoc test (p˂0.05). n.d. non detected.

42

Table 5. Phenolic acids and catechin of red wines spiked with volatile phenols (TF) and after treatment with NIPs and MIPs produced from different functional monomers and templates. Samples

Gallic acid (mg/L)

Catechin (mg/L)

trans-Caftaric acid (mg/L)

GRP (mg/L)

Coutaric acid (mg/L)

Caffeic acid (mg/L)

p-Coumaric acid (mg/L)

Ferulic acid (mg/L)

Caffeic acid ethyl ester (mg/L)

Coumaric acid ethyl ester (mg/L)

TF

60.71±2.70a

15.69±1.68a

28.81±0.02a

n.d

15.46±0.41a

1.54±0.13a

0.50±0.08a

0.34±0.09a

0.52±0.19a

3.02±0.01a

NIP1

55.83±1.60a

15.66±0.36a

27.24±0.42a

n.d

11.49±0.27b

0.79±0.01c

0.48±0.03a

0.23±0.00a

0.18±0.02b

1.88±0.16bc

MIP1EP

55.37±0.26a

15.41±0.50a

27.71±0.79a

n.d

11.15±0.65b 0.87±0.29bc

0.30±0.04a

0.16±0.01a

0.07±0.10b

1.56±0.07d

MIP1EG

55.84±0.80a

13.64±2.40ab

27.59±0.94a

n.d

14.02±0.66a 1.08±0.02abc

0.35±0.16a

0.34±0.13a

0.15±0.01b

1.50±0.08d

NIP2

55.14±0.00a

7.92±0.01c

28.70±0.18a

n.d

15.40±1.13a

0.86±0.01c

0.40±0.13a

0.26±0.03a

0.14±0.00b

1.62±0.11cd

MIP2EP

43.66±0.51b

9.75±1.49bc

26.91±0.34a

n.d

11.32±0.25b

0.71±0.06c

0.37±0.16a

0.19±0.05a

0.19±0.05b

1.57±0.08cd

MIP2EG

57.32±4.55a

10.09±0.12bc

27.01±0.64a

n.d

11.45±0.26b 1.12±0.11abc

0.33±0.12a

0.15±0.01a

0.17±0.05b

1.54±0.03d

NIP3

54.58±0.05ab

9.84±1.25bc

28.81±1.43a

n.d

11.62±0.07b 1.43±0.29ab

0.40±0.06a

0.16±0.02a

0.13±0.00b

2.09±0.04b

MIP3EP

49.59±7.08ab

8.85±1.01c

27.44±0.71a

n.d

11.88±0.40b 1.24±0.02abc

0.26±0.01a

0.16±0.01a

0.22±0.03b

1.77±0.04cd

55.87±0.81a 8.63±0.70c 27.06±1.08a n.d 11.25±0.41b 1.21±0.06abc 0.23±0.00a 0.15±0.00a 0.12±0.01b MIP3EG Values are presented as mean ± standard deviation; Means within a column followed by the same letter are not significantly different (Tuckey, p˂0.05). GRP - 2-S-glutathionyl caftaric acid. n.d. non detected.

1.79±0.04bcd

43

Table 6. Headspace aroma profile of red wines before (volatile phenols free T0 and volatile phenols spiked TF) and after treatment with nonimprinted (NIPs) and molecularly imprinted polymers (MIPs) produced from different functional monomers and templates. Compounds

ID$

Ethyl acetate

…

711

715

Fruity, sweet

49.10±4.79

3-Methylbutyl acetate

std

1152

1126

Banana

158.96±4.39abc

Ethyl hexanoate

std

RI Calculated

1231

RI*

1238

Odour descriptor

Green apple, anise

T0

TF a

13.32±0.50

a

a

Ethyl octanoate

std

1414

1429

Sweet, fruity, fresh

54.06±1.17

Ethyl decanoate

std

1623

1646

Flowery, fruity

74.30±3.96a

Diethyl succinate

std

1681

1698

Light fruity

120.01±1.30

Ethyl dodecanoate

std

1824

1850

Flowery, fruity

9.26±0.27a

42.27±3.53

NIP1 abc

162.52±9.04ab 12.85±2.17

a

50.69±3.33

a

112.21±7.76

ab

331.92±3.27

43.77±5.57

142.10±3.58abcde b

20.73±1.03

9.71±2.17a a

47.31±5.44

9.33±0.68

79.21±10.86a

a

MIP1EP ab

b

n.d bc

136.90±6.13bcde 7.52±0.26

bcd

17.40±1.98

bc

n.d

2.41±0.23bc

2.04±0.12bc

131.73±0.90cde 7.79±0.30

ab

13.67±0.79

e

6.78±0.31

c

bcd

13.94±0.32

c

78.56±1.58

bc

80.71±0.92

164.94±1.33a 5.68±0.23

cd

16.81±0.97

77.94±0.88

bcd

20.43±0.68

bc

n.d de

4.98±0.52d 18.03±0.08bc n.d

de

n.d. d

150.44±0.77abcd

72.80±5.53e n.d. 191.94±16.43d

std

1970

1989

Smoke

n.d.

28.67±2.22a

16.82±0.50b 41,31

53,27

50,23

57,04

66,69

58,00

67,02

65,86

68,38

Octanoic acid

std

2019

2030

Fatty acid, rancid

12.06±0.62a

9.62±1.49b

5.23±0.63c

3.35±0.16cd

3.80±0.11d

3.47±0.42cd

2.32±0.11d

2.60±0.06d

2.52±0.26d

1.85±0.16d

1.85±0.46d

4-Ethylphenol

std

2113

2142

Musty, spicy, phenolic

n.d

34.01±3.36a

21.19±1.72a

18.17±4.11a

18.03±1.21a

15.68±0.47a

12.57±0.93ab

15.15±1.57a

13.65±0.20a

13.56±0.72a

12.80±3.11ab

46,59

47,00

53,92

63,03

55,46

59,87

60,13

62,36

37,71 Total area - VPs

833.23±31.42

Reduction (%) SPME

…

a

873.69±48.87 …

a

649.73±34.65 25.63

b

535.01±42.17 38.76

c,d

523.79±53.50 40.05

c,d

545.78±28.64 37.53

b,c

453.68±17.61 48.07

c,d

501.82±37.64 42.56

c,d

9.45±0.29d

465.04±16.98 46.77

208.75±9.97

d

Ethylguaiacol

12.04±1.39cd

204.61±2.23

77.72±3.58

39.08±2.65abc

342.16±4.21

9.55±0.31d

257.13±23.82

158.41±3.51abc 7.09±0.40

bc

n.d. bc

38.38±4.56

MIP3EG bc

Roses, sweet

12.32±0.53cd

223.49±11.72

36.45±0.91

n.d cde

2.30±0.11bc cd

MIP3EP c

1911

14.27±2.34bc

273.27±15.31

7.56±0.42

bcd

n.d cde

1.43±0.21c b

118.96±15.33e

17.10±1.01

n.d bcd

1.60±0.06c

bc

127.60±3.75de

46.27±5.38

NIP3

abc

1872

13.40±2.19c

256.65±6.98

96.89±6.21

44.13±5.03

MIP2EG

abc

std

Reduction SPME(%)

256.52±3.81

46.42±3.64

n.d

75.06±8.02

bc

MIP2EP abc

2-Phenylethanol

Reduction SPME(%)

326.57±22.81

a

42.35±2.85

NIP2

abc

e

73.93±1.96

2.99±0.15b a

bcd

23.98±1.40

n.d 98.78±9.32

145.93±1.75abcde 6.82±0.32

bc

MIP1EG

abc

9.79±0.50d

c,d

470.12±9.74 46.19

c,d

9.06±1.74d

432.50±32.74d 50.50

Results expressed in absolute area (area*105). Values are presented as mean ± standard deviation; $ ID – Identification; std – Standard; * RI (retention index) from: Vás, Gál, Harangi, Dobó and Vékey (1998); Bailley, Jerkovic, Marchand-Brynaert and Collin (2006); Czerny, Brueckner, Kirchoff, Schmitt and Buettner (2011). MW (molecular weight). Odour descriptor from: Perestrelo, Fernandes, Albuquerque, Marques and Câmara (2006); Dragone, Mussato, Oliveira and Teixeira (2009). Jiang and Zhang (2010). Means within a column followed by the same letter are not significantly different ANOVA and Tukey post-hoc test (p˂0.05). n.d., not detected; Uncontaminated (T0) spiked red wine (TF) and wines treated with NIPs and MIPs. n.d. non detected.

44

Table 7. Mean scores of each attribute after sensory analysis of the volatile phenols free (T0) and volatile phenols spiked (TF) red wine and after treatment with non-imprinted (NIPs) and molecularly imprinted polymers (MIPs) T0 TF NIP2 MIP2EP NIP3 MIP3EG ANOVA C-index1 Colour Intensity

3.7±0.2a

3.8±0.2a

3.7±0.3a

3.4±0.2a

3.3±0.2a

3.6±0.2a

p<0.742

3.6

Hue

3.7±0.1a

3.8±0.1a

3.5±0.2a

3.8±0.1a

3.7±0.1a

3.5±0.2a

p<0.762

8.2

Limpidity

3.8±0.2a

3.8±0.2a

3.8±0.2a

3.8±0.2a

3.8±0.2a

3.7±0.2a

p<0.992

25.8

Oxidised

1.3±0.2a

1.3±0.1a

1.3±0.1a

1.2±0.1a

1.3±0.1a

1.3±0.1a

p<0.950

3.7

Fruity

4.3±0.2a

1.6±0.1b

2.4±0.2c,d

2.6±0.3c,d

2.3±0.2c

2.9±0.2d

p<0.000001

1.3

Floral

3.1±0.2a

1.3±0.2b

1.8±0.2b,c

2.0±0.2c

1.7±0.1b,c

1.8±0.2b,c

p<0.000002

3.5

Vegetable

1.5±0.2

2.2±0.2

1.9±0.1

1.8±0.1

1.9±0.2

1.8±0.2

p<0.190

2.3

Phenolic

1.1±0.1a

4.5±0.2b

2.8±0.2c,d

2.6±0.3c,d

3.0±0.3d

2.3±0.1c

p<0.000001

1.1

Oxidised aroma

1.2±0.1a

1.9±0.2b

1.6±0.2b

1.4±0.1a,b

1.5±0.2ª,b

1.2±0.1a

p<0.0261

3.6

Consonance analysis results – C-index values for attributes; Values are presented as the mean ± standard deviation (n = 12); Means within a line followed by the same letter are not significantly different (Duncan p ˂ 0.05). 1

45

Figure 1

46

MIP1

MIP2

MIP3

None

EP

EG

47

NIP3

NIP2

NIP1 4000

3500

3000

2500

2000

1500

1000

500

Wavenumber (cm-1)

Figure 3

49

A

2500

Intensity (cps)

2000 1500 1000 500 0 5

15

25

35

45

55

35

45

55

35

45

55

2 (º)

B

2000 1800

Intensity (cps)

1600 1400 1200 1000 800 600 400 200 0 5

15

25

2 (º)

C

3000

Intensity (cps)

2500 2000

1500 1000 500 0 5

15

25

2 (º)

Figure 4

50

4 - EP 4 - EG Polymer 1 - Ethylene glycol dimethacrylate + methacrylamide 0 .2 0

0 .2 5

q e ( m m o l/g )

q e ( m m o l/g )

0 .2 0

0 .1 5

0 .1 0

0 .1 0

0 .0 5

0 .0 5

0 .0 0 0 .0

0 .1 5

0 .0 0 0 .2

0 .4

0 .6

0 .8

1 .0

0 .0

0 .2

m m o l/L

0 .4

0 .6

0 .8

0 .6

0 .8

m m o l/L

Polymer 2 - Ethylene glycol dimethacrylate + ethyl methacrylate 0 .2 5

0 .4

q e ( m m o l/g )

q e ( m m o l/g )

0 .2 0

0 .3

0 .2

0 .1

0 .1 5

0 .1 0

0 .0 5

0 .0 0

0 .0 0 .0

0 .2

0 .4

0 .6

0 .8

0 .0

1 .0

0 .2

0 .4

m m o l/L

m m o l/L

Polymer 3 - Ethylene glycol dimethacrylate + ethylene glycol methyl ether acrylate 0 .2 5

0 .2 0

q e ( m m o l/g )

q e ( m m o l/g )

0 .2 0

0 .1 5

0 .1 0

0 .1 0

0 .0 5

0 .0 5

0 .0 0 0 .0

0 .1 5

0 .0 0 0 .2

0 .4

0 .6

0 .8

1 .0

m m o l/L

0 .0

0 .2

0 .4

0 .6

0 .8

m m o l/L

Figure 5

51

400

800

a

b b

a

600

 g /L

a b b

100

a a

l

ta

o

b

800

a

b 600

b

200

b

a

a

a

400

a b b

a a a 200

0

a

l

ta

T

4

o

-E

-E 4

b b 1000

200

34% 38% 33%

1500

 g /L

a

High

b b

400

G

l

ta

T

Medium

300

P

32% 36% 37% o

G -E 4

4

-E

P

0

a a a

a

500

0

b b

a

b b

a a a

100

Medium

High

o

ta

l

63% 60% 54% T

G

P -E 4

o

ta

l

38% 40% 29% T

G -E 4

-E

P

0

4

 g /L

34% 27% 19% T

T

4

4

-E

-E

G

P

l

ta

-E 4

a a

100

c

b b

a

High

 g /L

 g /L

34% 31% 26% o

G

P -E 4

400

300

a

a

0

Medium b

b

400

200

0

b

c

-E

 g /L

200

c

b b

300

4

a

Figure 6

52

4

Fl or al

Fr ui ty

Ph en ol ic

H ue

C ol ou r

Sensory score

Graphical Abstract

5

TF NIP MIP

3

2

1

0

53

Highlights

New molecularly imprinted polymers were synthesised to remove volatile phenols (VPs) Ethylene glycol methyl ether acrylate was the best functional monomer Removal up to 63% of wine VPs were obtained depending on their concentration in wine The new MIP didn´t change significantly the red wine colour The new MIP improved significantly the sensory quality of the red wine

54