*Article* **Molecularly Imprinted Solid Phase Extraction Strategy for Quinic Acid**

**Sarah H. Megahed <sup>1</sup> , Mohammad Abdel-Halim <sup>1</sup> , Amr Hefnawy <sup>2</sup> , Heba Handoussa 3, Boris Mizaikoff 4,5,\* and Nesrine A. El Gohary 1,\***


**Abstract:** Quinic acid (QA) and its ester conjugates have been subjected to in-depth scientific investigations for their antioxidant properties. In this study, molecularly imprinted polymers (MIPs) were used for selective extraction of quinic acid (QA) from coffee bean extract. Computational modelling was performed to optimize the process of MIP preparation. Three different functional monomers (allylamine, methacrylic acid (MAA) and 4-vinylpyridine (4-VP)) were tested for imprinting. The ratio of each monomer to template chosen was based on the optimum ratio obtained from computational studies. Equilibrium rebinding studies were conducted and MIP C, which was prepared using 4-VP as functional monomer with template to monomer ratio of 1:5, showed better binding performance than the other prepared MIPs. Accordingly, MIP C was chosen to be applied for selective separation of QA using solid-phase extraction. The selectivity of MIP C towards QA was tested versus its analogues found in coffee (caffeic acid and chlorogenic acid). Molecularly imprinted solid-phase extraction (MISPE) using MIP C as sorbent was then applied for selective extraction of QA from aqueous coffee extract. The applied MISPE was able to retrieve 81.918 ± 3.027% of QA with a significant reduction in the amount of other components in the extract.

**Keywords:** computational modelling; molecularly imprinted polymers; solid-phase extraction; quinic acid; *Coffea arabica*

#### **1. Introduction**

Oxidative stress is the main cause for altering numerous signaling pathways that eventually promote cellular damage. It is considered a key player mediator in the pathophysiology of several health complications [1]. Intracellular antioxidant enzymes and intake of dietary antioxidants may help maintain the utmost antioxidant balance in the body. Epidemiological studies have proven that the consumption of nutraceuticals with potential antioxidant impacts reduces the risk of several diseases, including neurodegenerative diseases, cardiovascular diseases and cancer, through apoptosis-mediated cytotoxicity. They are also known to reduce inflammation by different mechanisms such as the inhibition of pro-inflammatory transcription factor, nuclear factor Kappa B (NF-κB) [2].

Quinic acid (QA) and its ester conjugates (caffeoylquinic acids) are present in various food products [3] and are the major constituents of coffee [4]. Many reports support the efficacy of nutritional QA in the enhancement of several biological processes via its paramount antioxidant effects [5]. QA has been previously reported as a potent antioxidant due to its capability to lower the intracellular ROS levels in H2O2 pre-treated cells and

**Citation:** Megahed, S.H.; Abdel-Halim, M.; Hefnawy, A.; Handoussa, H.; Mizaikoff, B.; El Gohary, N.A. Molecularly Imprinted Solid Phase Extraction Strategy for Quinic Acid. *Polymers* **2022**, *14*, 3339. https://doi.org/10.3390/ polym14163339

Academic Editor: Michał Cegłowski

Received: 3 July 2022 Accepted: 10 August 2022 Published: 16 August 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

inhibition of lipid peroxidation [6]. QA was also found to protect against oxidative stress by increasing the antioxidant capacity as well as decreasing the levels of MDA and nitrite in an in vitro study by Khorasgani et al. [3]. Furthermore, QA upregulated *daf-16 sod-3* expression and downregulated reactive oxygen species (ROS) levels in a *C.elegans* in vivo model [5].

Noteworthily, the ingested QA is used by the human body as a precursor for the synthesis of many important compounds such as nicotinamide [7], which plays a major role in neuronal development and survival [8]. Nicotinamide is also used in the synthesis of two important co-enzymes, nicotinamide adenine dinucleotide (NAD) and nicotinamide adenine dinucleotide phosphate (NADP). Both NAD and NADP are essential in many processes in the human body such as DNA repair, energy production and cell death regulation [9].

Several old techniques have been developed to isolate QA from its original natural source, yet, they are still reports with several limitations [10]. Many extraction protocols and chromatographic methods have been developed to optimize the extraction of QA, such as column chromatography [11], alkaline hydrolysis [12] and liquid–liquid extraction while using an amine as an extractant [13]. However, most of the reported methods are solventand time-consuming and none of these methods are considered as having high selectivity towards QA.

Molecular imprinting is a rapidly growing technique used to create synthetic receptors with recognition sites that have the ability to bind specifically to a wide variety of molecules ranging from small drug molecules to large peptides or proteins. The molecularly imprinted polymer (MIP) technique can be described by analogy to the "lock and key model" described by Emil Fischer [14]. The synthesized molecularly imprinted polymers (MIPs) have many cavities complementary to their template molecules in shape, size and chemical functionality, causing them to be particularly selective towards those target molecules [15].

In the past few years, several research articles and reviews have been published, showing the current advances and diversity in the synthesis and application of MIPs [16–18]. These studies show the increasing importance of the use of MIPs in the field of analytical chemistry and their application in sensors, extraction and chromatography [19].

The use of MIPs as a replacement for biological material in optical [20] and electrochemical [21] biosensors has attracted much attention throughout the years. This is attributed to their superiority over the biological components in terms of cost, stability and reusability [17]. They also offer outstanding recognition ability, selectivity, specificity and robustness [17,22]. Accordingly, molecular imprinting has become one of the most important techniques for fabricating synthetic ligands on sensor surfaces [17,19]. MIPs have been successfully coupled to surface plasmon resonance (SPR) sensors [23], quartz crystal microbalance sensors [24], luminescence probes [25] and electrochemical sensors [26,27]. Molecular-imprinted fluorescent sensors (MIFS) have been used for detection of several organic molecules and metal ions including proteins [28], caffeine [29] and cocaine [30] in addition to silver [31] and aluminum ions [32]. Moreover, MIP-based SPR sensors have been applied for detection of biomarkers [33], biomolecules [34], pesticides [35] and banned additives [36].

The use of MIPs as sorbents has become one of the most commonly used methods for SPE [37]. They have attracted much attention owing to their numerous advantages such as high selectivity, ease of preparation, low cost, reusability and their potential application to a wide range of target molecules [38]. MISPE has been applied for the extraction of different analytes from biological fluids such as blood, urine and bile [39] in addition to environmental samples and plant tissues [18].

Different techniques of molecular imprinting have been reported for the extraction of antioxidants from natural resources [40,41] such as the isolation of oleuropein from olive leaf extracts [42] and the concentration of tannins from Brazilian natural sources [43].

MIPs were also applied on different plant extracts, including coffee, for isolation of QA derivatives [12]. In a recent study by Kanao et al. [44], poly(ethylene glycol) hydrogels prepared by molecular imprinting were used for selective extraction of quinic acid gammalactone (QAGL) from coffee. The synthesized MIPs successfully removed QAGL from

freshly brewed coffee at high speed with high yield, which resulted in better-tasting coffee. Moreover, the prepared MIPS were highly selective towards QAGL, which prevented non-specific adsorption of other components in coffee.

Bulk imprinting is considered the most widely used method for preparation of MIPs [45]. It is the method of choice for the imprinting of small molecules as it allows fast and reversible adsorption and the release of the template molecule [46]. Bulk imprinting has been successfully used for small molecules comparable to QA such as sinapic acid [47], gallic acid [48], caffeic acid and *p*-hydroxybenzoic acid [49]. However, the preparation of MIP using QA as a template has not been reported before.

In this work, three bulk MIPs were synthesized using three different monomers. Their binding performance and their ability to be used as sorbents for SPE of QA from coffee beans have been examined.

#### **2. Materials and Methods**

#### *2.1. Reagents and Materials*

Standard QA (98%) was purchased from Alfa Aesar. Caffeic acid (CA) (98%), chlorogenic acid (CLA) (95%), acetonitrile (CAN) (HPLC grade; ≥9.99%), methanol (99.8%), absolute ethanol (EtOH) (≥99.5%), formic acid (reagent grade; ≥95%), glacial acetic acid (≥99.7%), methacrylic acid (MAA) (stabilized with hydroquinone monomethyl ether; ≥90.0%), 4-vinylpyridine (4-VP) (contains 100 ppm hydroquinone as inhibitor; 95%), ethylene glycol and dimethacrylate (EGDMA) (contains 90–100 ppm hydroquinone monomethyl ether as inhibitor; 98%) were obtained from Sigma Aldrich (Darmstadt, Germany). Ethyl acetate (EtOAc) (99.5%) was purchased from Alfa Chemical (India) and a purelab UHQ (ELGA) water purification system (High Wycombe, Buckinghamshire, UK) was used to obtain ultra-pure water. Empty polypropylene SPE 3 mL tubes with PE frits of 20 μm porosity were obtained from Supelco Inc. (Bellefonte, PA, USA).

Green coffee (*C. arabica* L.) beans were kindly supplied by Misr Coffee (10th of Ramadan Ind. City, Cairo, Egypt Industrial Company). The beans were mechanically ground and milled into size (40 mesh) for extraction and application steps. The obtained granules were completely dried using a hot air oven at a temperature of 38 ◦C for 2 h.

#### *2.2. Computational Modelling: Monomers Molar Ratio Screening*

Gaussian 03 package was used to determine the optimum template to monomer molar ratio for bulk polymers. Gaussview 5.0 software (Gaussian Inc., Pittsburgh, USA) was used first to draw 3D structures of the template, QA, monomers, MAA, allylamine and 4-VP, in addition to template-monomer complexes. All the obtained structures were then optimized to the lowest energy conformation using Hartree-Fock theory with the (6–31 G(D)) basis set. Hartree-Fock is an accurate method for large systems, which makes it easier to screen different monomer ratios for specific templates [50,51]. Different template to monomer molar ratios were screened for each of the used monomers and Equation (1) was used to calculate the binding energies of the complexes.

$$
\Delta E = E(\text{templete-monomer complex}) - \left[ E(\text{templete}) + nE(\text{monomer}) \right] \tag{1}
$$

where Δ*E* refers to the binding energy of the complex and *n* refers to the monomer number in the template–monomer complexes.

The calculations of the binding energies were conducted in the solvent phase (DMSO) using the polarizable continuum model (PCM) to mimic experimental conditions. In this model, the solvent effect is considered during calculations as it affects the stability and the energy of the template–monomer complexes [52], where the solvent is modelled as a polarizable continuum rather than individual molecules [53].

#### *2.3. Bulk Polymers Preparation*

Different bulk MIPs were prepared via the non-covalent approach, introduced by K. Mosbach et al. [54], using thermal free radical polymerization. The reaction was performed

in a glass vial, by dissolving 0.5 mmol of QA in 6 mL of the porogen, DMSO. This was followed by the addition of suitable amount of monomer and the pre-polymerization mixture was stirred at room temperature for 30 min. Afterwards, the cross-linker ethylene glycol dimethacrylate (EGDMA) was added and the solution was left to stir for 5 min. Following which, 75 mg of the free radical initiator was added and the solution was purged with argon for 3 min to remove oxygen and create inert conditions. The glass vial was sealed and left in an oil bath at 60 ◦C for 24 h to allow polymerization. For each MIP, a non-imprinted polymer (NIP) was prepared using the same procedure without adding the template. The glass vials were then smashed to obtain the bulk polymers, which were then subjected to crushing, grinding and sieving. The fraction with a particle size of 40–100 μm was collected. The full composition of the prepared polymers is described in Table 1.


**Table 1.** Chemical composition of prepared MIPs.

#### *2.4. Morphology Characterization*

The surface morphology of the MIPs and their corresponding NIPs was examined using FEI Quanta 650 environmental scanning electron microscope (ESEM) under high vacuum at a high voltage of 10 kV with a spot size of 3.5 and working distance set to around 10 mm. N2 adsorption–desorption isotherms were used to analyze the surface area, pore volume and pore size of all polymers at 77 K via a Quantachrome TouchWin v.1.2 instrument (FL, USA). The polymers were first degassed at 150 ◦C for 24 h to remove the adsorbed gasses and moisture. The specific surface areas were calculated using the Brunauer–Emmett–Teller (BET) method, while the Barrett–Joyner–Halenda (BJH) method was used to calculate the volume and pore size.

#### *2.5. Equilibrium Rebinding Studies*

The binding studies were conducted at room temperature by modifying the protocol previously described by Saad et al. [45]. Ten mg of the imprinted and non-imprinted polymers were added to 2 mL of 0.1 mM QA solution prepared in water, methanol or ACN: water (4: 1 *v*/*v*). The suspensions were then left to shake at room temperature for 2 h at 200 rpm using a Thermo ScientificTM MaxQ mini 4000 Benchtop Orbital Shaker (Waltham, MA, USA). This was followed by a centrifugation step at 14,000 rpm for 15 min and the supernatants were filtered through 0.22 polytetrafluoroethylene (PTFE) syringe filters. The concentration of the unbound QA was then quantified using UHPLC-MS/MS. The amount of the rebound QA was calculated using Equation (2)

$$B = \frac{\left(\mathbb{C}\_i - \mathbb{C}\_f\right) \times V \times 1000}{W} \tag{2}$$

where *B* is the amount of rebound template in μmol/g polymer, *Ci* and *Cf* represent the initial and final concentrations in mM, respectively, *V* is the volume of the solution in ml and *W* is the weight of used polymer in mg.

The imprinting factor was then calculated using Equation (3)

$$IF = \frac{B\_{MIP}}{B\_{NIR}}\tag{3}$$

where *IF* is the imprinting factor and *BMIP* is the amount of template bound in μmol/g of the MIP, while *BNIP* is the amount of template bound in μmol/g of the NIP.

#### *2.6. Adsorption Kinetics*

The uptake profiles of MIP C and its corresponding NIP were studied over 2 h. This was achieved by incubating 10 mg of the polymer with 2 mL of 0.1 mM QA in methanol for 5, 15, 30, 60 and 120 min. This was followed by a centrifugation step and the supernatants were analyzed using UHPLC-MS/MS and the amount of bound template was determined using Equation (2).

The obtained data were further analyzed to determine adsorption kinetics. The pseudofirst order and pseudo-second order kinetics were used to investigate the mechanism of adsorption of MIP C. The pseudo-first order rate can be expressed in Equation (4)

$$
\ln(q\_\varepsilon - q\_t) = \ln q\_\varepsilon - \mathcal{K}\_1 t \tag{4}
$$

where *qe* and *qt* are the binding capacities at equilibrium and at time *t* (μmol/g), respectively, *K*<sup>1</sup> is the rate constant of pseudo-first order in min−<sup>1</sup> and *t* is time in min [55].

Pseudo-second order is expressed in Equation (5)

$$\frac{1}{q\_t} = \frac{1}{\mathcal{K}\_2 q\_\varepsilon^2} + \frac{t}{q\_\varepsilon} \tag{5}$$

where *K*<sup>2</sup> is the rate constant of pseudo-second order in g/μmol.min [56].

#### *2.7. Binding Isotherm*

Ten mg of MIP C and its corresponding NIP were incubated with 2 mL of QA in methanol over the concentration range of (0.01–0.2 mM) for 2 h. The binding isotherms of both polymers were then obtained by plotting the binding capacity (B) versus the initial QA concentration (Ci). The results were further analyzed using the Freundlich isotherm model [57] expressed by Equation (6).

$$
\text{Log B} = m \text{Log C}\_f + \text{Loga} \tag{6}
$$

where *B* represents the binding capacity in μmol/g, *m* represents the Freundlich constant or heterogenicity factor ranging from 0 to 1, *Cf* represents the equilibrium concentration in mM and the constant α represents maximum binding capacity in μmol/g [45].

#### *2.8. MISPE Procedure Optimization*

MIP C and NIP C were used as sorbent materials for offline-mode solid-phase extraction. Forty mg of each polymer was packed into a 3 mL polypropylene SPE cartridge with a 0.22 PTFE frit placed below the polymer and another similar frit placed above the polymer for secure packing. All trials were performed in triplicate and the analytical measurements were obtained using UHPLC-MS/MS.

A systematic one-factor-at-a-time (OFAT) approach was used to investigate different parameters affecting the extraction procedure including loading amount, loading volume, washing solvent and elution volume. Water: acetic acid (9:1 *v*/*v*) was used as the elution solvent in all trials.

#### *2.9. UHPLC-MS/MS Measurements*

A new UHPLC-MS/MS method was applied for quantification of QA using ferulic acid as an internal standard. A seven-point calibration curve for QA was prepared in methanol over the concentration range 0.001–0.2 mM.

UHPLC-MS/MS measurements were done using ACQUITY Xevo TQD system (Waters), which is composed of ACQUITY UPLC H-Class system and a XevoTQD triplequadrupole tandem mass spectrometer with an electrospray ionization (ESI) interface (Waters Corp., Milford, MA, USA). The column used for separation was an Aquity UPLC BEH C18 (Waters, Wexford, Ireland), with dimensions of 100 mm × 2.1 mm and stationary phase particle size of 1.7 μm. MassLynx 4.1 software (Waters, Milford, MA, USA) was used

for system operation and data acquisition. The TargetLynx quantification program was used to process the acquired data (Waters, Milford, MA, USA). A gradient program was used for chromatographic separation using 0.01% formic acid in water (A) and acetonitrile (B) at a flow rate of 0.3 mL/min, injection volume of 10 μL and column temperature of 40 ◦C. The gradient was run as follows: 0 min, 90% A, 10% B; 0.75 min, 90% A, 10% B; 2.5 min, 1% A, 99% B; 4 min, 1% A, 99% B; 4.5 min, 90% A, 10% B; 6 min, 90% A, 10% B. The desolvation and cone gas flow rate were 800 and 20 L/h, respectively (nitrogen was used in both cases). The collision gas (argon) was applied at a pressure of 3.67 <sup>×</sup> <sup>10</sup>−<sup>3</sup> mbar approx. The MS parameters were as follows: radio frequency (RF) lens voltage 2.5 V, capillary voltage 4 kV, source temperature 150 ◦C and desolvation gas temperature 300 ◦C. Cone voltage was 45 V and 28 V for QA and ferulic acid, respectively. The ESI source was operated in negative mode. Quantification was performed using multiple reaction monitoring (MRM) of the transitions of m/z 191 > 85 with collision energy of 18 V for QA and m/z 192.89 > 133.95 with collision energy of 14 V for ferulic acid. Dwell time was set automatically by MassLynx 4.1 software.

#### *2.10. Method Validation*

The applied UPLC-MS/MS method was validated according to the ICH guidelines in terms of linearity, limit of detection (LOD), limit of quantification (LOQ), inter- and intra-day precision and accuracy. More details are provided in the supplementary material.

#### *2.11. MIP Cartridge Reusability*

MIP cartridge reusability was tested over 10 adsorption–desorption cycles, where the SPE cartridge was filled with 40 mg of MIP C. This was followed by a conditioning step using 2 mL of absolute ethanol, then loading with 2 mL of 0.1 mM QA in ethanol, a washing step using 2 mL of acetonitrile and finally an elution step using 2 mL of water: acetic acid (9:1 *v*/*v*). After the elution step, the cartridge was subjected to 5 washing steps; 2 steps of washing using 3 mL of water: acetic acid (9:1 *v*/*v*) each, then once using 3 mL of water and finally 2 washing steps using 3 mL of absolute ethanol each. The elution fractions were analyzed using the validated UHPLC-MS/MS method and QA recovery percentage was calculated after each elution.

#### *2.12. Selectivity Study*

Two mL of equimolar mixture of QA, caffeic acid and chlorogenic acid (Figure 1) (0.05 mM) in ethanol was percolated through SPE cartridges packed with 40 mg of MIP C and NIP C. The cartridges were then washed using 2 mL of acetonitrile. This was followed by the elution step, using 2 mL of 10% acetic acid in UPW.

**Figure 1.** Structures of quinic acid, caffeic acid and chlorogenic acid.

The obtained elution fraction was evaporated and reconstituted in methanol. The amount of QA in the elution solvent was measured using UHPLC-MS/MS, while caffeic acid and chlorogenic acid were quantified using UHPLC-UV at λmax 325 nm.

#### *2.13. MISPE Application on Coffee Extract*

#### 2.13.1. UHPLC Method for QA Quantification in Coffee Extract

UHPLC-PDA-ESI- MS and MS/MS analyses were done using the ACQUITY Xevo TQD system (Waters), which is composed of the ACQUITY UPLC H-Class system and a XevoTQD triple-quadrupole tandem mass spectrometer with an electrospray ionization (ESI) interface (Waters Corp., Milford, MA, USA). The column used for separation was an Aquity UPLC BEH C18 (Waters, Wexford, Ireland), with dimensions of 100 mm × 2.1 mm and stationary phase particle size of 1.7 μm. MassLynx 4.1 software (Waters, Milford, MA, USA) was used for system operation and data acquisition. The solvent system consisted of 0.01% formic acid in water (A) and acetonitrile (B) by applying the following gradient program: 0 min, 8% B; 30 min, 45% B; 31 min, 8% B; and 33 min, 8% B. The flow rate was 0.2 mL/min and the injection volume was 10 μL. The samples were dissolved in ethanol then filtered through a filter of pore size 0.2 μm. The eluted compounds were detected at mass ranges from 100 to 1000 m/z. The MS scan was carried out at the following conditions: capillary voltage, 3.5 kV; detection at cone voltages, (20 V–95 V); radio frequency (RF) lens voltage, 2.5 V; source temperature, 150 ◦C and desolvation gas temperature 500 ◦C. The desolvation and cone gas flow rate were 800 and 20 L/h, respectively (nitrogen was used in both cases). QA was detected through the MRM of the transition m/z 191 > 85 with collision energy of 18 V and cone voltage of 45 V.

#### 2.13.2. Method Validation

The method was validated according to the ICH guidelines in terms of linearity, limit of detection (LOD), limit of quantification (LOQ), inter- and intra-day precision and accuracy. More details are found in the supplementary data.

#### 2.13.3. Preparation of Aqueous Coffee Extract

Fifty grams of roasted coffee beans were subjected to fine grinding and placed in a conical flask, then 1 L of ultrapure water was added. The mixture was heated at 60 ◦C for 1 h and was left to macerate overnight. This was followed by a centrifugation and a filtration step. Then, the supernatant was concentrated using a rotary vacuum evaporator at 40 ◦C. The dried residue was stored in an opaque glass bottle for further studies.

#### 2.13.4. Application of MISPE for Extraction of QA from Total Aqueous Coffee Extract

The optimized SPE method was used for the extraction of QA from total aqueous coffee extract. The extract was reconstituted in ethanol: water (97:3 *v*/*v*) and 2 concentrations were prepared, 0.25 mg/mL and 0.5 mg/mL. Two ml of each concentration was loaded to SPE cartridge containing 40 mg of MIP C. This was followed by a washing step using 2 mL of acetonitrile and an elution step using 2 mL of 10% acetic acid in water.

#### **3. Results and Discussion**

#### *3.1. Computational Modelling: Monomers Molar Ratio Screening*

In this study, computational modelling was used to optimize the pre-polymerization complex by determining the most suitable functional monomer molar ratio for each of the chosen monomers, since the self-assembly of the template and functional monomer is the most crucial step in polymer preparation [58]. The study was conducted in the solvent phase and DMSO was the solvent of choice, which was used as the porogen during polymer preparation [15]. The influence of the cross-linker was not considered to simplify the calculations [59]. The three functional monomers used in this study were allylamine, MAA and 4-VP. For each of the chosen monomers, different template: functional monomer ratios were examined. For allylamine, the studied template: monomer ratios were (1:1, 1:2, 1:3, 1:4, 1:5 and 1:6), for MAA, the ratios were (1:1, 1:2, 1:3 and 1:4) and for 4-VP, the studied ratios were (1:1, 1:2, 1:3, 1:4 and 1:5). The optimized structures of QA, functional monomers and pre-polymerization complexes are shown in Figure 2 and Figures S1–S13 in the supplementary data.

**Figure 2.** Computer-modelled structures of the best conformations for (**A**) QA–(allylamine)6, (**B**) QA–(MAA)4 and (**C**) QA–(4-VP)5 complexes.

Energies of the most stable conformations were then determined and the binding energies of the formed complexes were calculated according to Equation (1) and the results are shown in supplementary data. Based on the binding energies; the best template: monomer ratio was determined for each functional monomer. Results revealed that by increasing the number of monomers used, the calculated binding energies increased, which indicates the formation of more stable complexes [60]. For allylamine, the best ratio was 1:6 (*E* = −175.909 kJ/mol). For MAA, it was 1:4 (*E* = −1633.061 kJ/mol). Finally, for 4-VP the optimum ratio was 1:5 (*E* = −136.5265 kJ/mol) (Figure 2). Accordingly, these ratios were chosen for the synthesis of MIPs and their corresponding NIPs to be used for further applications.

#### *3.2. Morphology Characterization*

The surface morphology of MIPs and their corresponding NIPs of particle size range 40–100 μm were analyzed using scanning electron microscopy (SEM) as shown in Figure 3 and Figures S14 and S15 in the supplementary material. The SEM images showed irregular shapes and sizes, which agrees with the nature of bulk MIPs previously reported in literature [45].

ǻǼȱ

**Figure 3.** SEM images of (**A**) MIP C and (**B**) NIP C with increasing magnification from left to right.

Nitrogen adsorption–desorption isotherms were performed and BET analysis was used to determine surface areas, while BJH analysis was used to determine the average pore size diameter and pore volume, as these parameters may have a strong impact on the efficiency of adsorption (Figure 4 and Figure S16 in the supplementary material).

The BET results, shown in Table 2 revealed that the MIPs have lower surface areas compared to their corresponding NIPs. This most probably could be attributed to the heterogeneity and roughness of the surface of NIPs which were prepared in the absence of the template, unlike the MIP imprinting process that follows a certain degree of order during the polymerization step [45]. MIP C exhibited the highest surface area (31.41 m2/g) compared to MIP A and MIP B that exhibited surface areas of 21.51 m2/g and 23.80 m2/g, respectively.

The data derived from BJH (Table 2) revealed that all the polymers exhibited a welldeveloped pore structure. They were all mesoporous with a pore radius of 1.64–1.77 nm, which provides good recognition properties for interaction with the template molecule. These results suggest that the synthesized polymers can be used as sorbents for SPE, since the mesoporous structures are more permeable for solvents compared to micropores and do not require the application of high pressure [61]. All the MIPs and the corresponding NIPs have comparable pore radii, while the pore volumes of all the NIPs are generally larger than the MIPs.

**Figure 4.** BET isotherms of (**A**) MIP C and (**B**) NIP C.

The overall results reveal that all the NIPs showed higher surface areas and porosities compared to the corresponding MIPs. Thus, it may be concluded that the binding performance of the polymers would be attributed to the imprinting process rather than the surface area and porosity of the particles [45].


**Table 2.** Surface area, pore volume and pore size of synthesized polymers using BET and BJH methods.

#### *3.3. Rebinding Studies*

The synthesized polymers were subjected to batch rebinding studies to evaluate their affinity to QA using 0.1 mM QA solution prepared in three different solvents: water, methanol and acetonitrile: water (4:1 *v*/*v*) as shown in Table 3.


**Table 3.** Binding capacities and imprinting factors of bulk polymers in different solvents.

It was observed that when water was used as the rebinding medium, all the polymers showed relatively low binding. This might be attributed to the high solubility of QA in water. This high affinity between QA and water molecules might decrease its interaction with the polymers. Moreover, there was a significant difference between the binding of the MIPs and the corresponding NIPs, pronouncing the specific interaction with the imprinted polymers, where MIP C showed the highest binding capacity of 4.88 ± 0.32 μmol/g, while the binding of its corresponding NIP was 2.28 ± 0.38 μmol/g, with an imprinting factor of 2.14. Although aqueous medium is known to disrupt hydrogen bonding interactions between template and monomer, in the rebinding results of QA a pronounced difference between the binding of QA to the MIPs and the corresponding NIPs was observed. This suggests that hydrogen bonding is not the only factor behind the MIP selectivity towards QA. It can be concluded that, during the imprinting process, different interactions took place based on the size, shape and functionality of the template [62]. During NIP preparation, no proper cavities or recognition sites were formed, therefore the NIP binding to QA was only through non-specific adsorption [63]. As a result, the amount of QA adsorbed by the NIP was lower than by MIP.

It was observed that there was a significant increase in the binding capacity in all polymers when methanol was used as the rebinding medium. This might be because QA has lower solubility in methanol [64], thus a lower affinity to the binding solvent, which increases the chance of interaction between QA and the polymers. It was still observed in this solvent that the binding of QA to MIPs is higher than its binding to the corresponding NIPs, which indicates the success of the imprinting process.

The binding capacities of MIP A, MIP B and MIP C were 7.52 ± 0.64 μmol/g, 3.14 ± 0.36 μmol/g and 9.05 ± 0.80 μmol/g, respectively, while NIP A, NIP B and NIP C showed binding capacities of 4.70 ± 0.28 μmol/g, 1.68 ± 0.32 μmol/g and 5.58 ± 0.53 μmol/g, respectively.

The third binding solvent chosen was ACN, an example of polar aprotic solvent, water was added to acetonitrile with a ratio of ACN:H2O (4:1) to ensure the solubility of polar QA, which is insoluble in pure acetonitrile. On comparing the results of rebinding to results obtained in UPW, it was found that the use of acetonitrile increased the interaction between QA and the synthesized polymers, causing an increase in the binding capacity in most of the polymers. It could be argued that the addition of a polar aprotic solvent enhances the hydrogen bond formation between QA and the polymers. Additionally, the low solubility of QA in acetonitrile probably enhanced the interaction between QA and the polymers [65]. It was observed that MIP B and its corresponding NIP prepared using MAA as functional monomer showed lower binding in ACN compared to UPW. This most probably indicates that, in the case of this polymer, the hydrophobic interactions are the main interactions that take place between QA- and MAA-based polymers and this type of interaction is more pronounced when using only UPW as a binding solvent.

MIP C synthesized using 4-VP as functional monomer showed the highest binding capacity in all the rebinding solvents. Although the calculated binding energy of the QA– MAA complex was the highest during computational studies, practically, the polymers prepared with 4-VP showed better overall performance. This can be attributed to the extra interaction between the basic monomer and the acidic template, the pyridine ring of the monomer could promote adsorption because it can form both acid–base interactions and strong hydrogen bonds with the template. Thus, a more stable complex between the template and the functional monomer was formed during the imprinting process [55]. This agrees with what was reported in some studies where 4-VP monomer showed superior results during the imprinting of acidic templates when compared to other acidic or neutral monomers. In a study by Zhang et al., salicylic acid was imprinted using 4-VP and acrylamide as functional monomers, where 4-VP showed superior imprinting effect compared to acrylamide [66]. BET results also revealed that MIP C has the highest surface area (31.41 m2/g) when compared to MIP A and MIP B, which might contribute to its higher binding capacity. The binding of the allylamine MIP was higher than the MAA MIP, which also might be attributed to its basic nature.

#### *3.4. Adsorption Kinetics*

In order to study the interactions between the 4-VP polymers and the template during the 2 h equilibrium rebinding study in methanol, an effect of time experiment was conducted on MIP C and NIP C. Ten mg of each polymer was incubated with 0.1 mM QA solution for definite time intervals. The uptake profile shown in Figure 5 revealed that the uptake of both MIP C and NIP C gradually increased during the course of the experiment to reach its maximum after 1 h, after which no significant improvement in the binding performance was observed.

**Figure 5.** Uptake profiles of MIP C and NIP C over 2 h equilibrium rebinding study in methanol.

The pseudo-first order and pseudo-second order kinetics were used to investigate the mechanism of adsorption of MIP C (Figure S17 in the supplementary material). The results showed better fitting in pseudo-second order equation where R2 was found to be 0.9967 versus R<sup>2</sup> of 0.8683 obtained from the pseudo-first order equation. This suggests that MIP C binding follows pseudo-second order kinetics with a calculated rate constant of 6.33 <sup>×</sup> <sup>10</sup>−<sup>3</sup> in g/μmol min.

#### *3.5. Binding Isotherm*

A two-hour equilibrium rebinding study was carried out by incubating 10 mg polymer with 2 mL QA solution in methanol with different concentrations ranging from 0.01 to 0.2 mM. A binding isotherm was conducted by plotting the binding capacity against initial concentration (Ci) as in Figure 6. The amount of QA bound increased with the increase of the initial concentration up to 0.1 mM of QA, while there was only a slight difference in the binding capacity between 0.1 and 0.2 mM QA. The results also revealed that the difference between the binding performance of MIP and NIP was more pronounced in the higher concentration ranges, mostly 0.1 and 0.2 mM.

The obtained results were further analyzed using the Freundlich isotherm for MIP C (Figure S18 in the supplementary material). The model was fitted with a high degree of correlation, R<sup>2</sup> of 0.954. MIP C showed a heterogenicity factor of 0.5389, suggesting heterogenicity of the binding surface, while the α value was 35.3 μmol/g.

**Figure 6.** Binding isotherm of MIP C and NIP C.

#### *3.6. MISPE Procedure Optimization*

3.6.1. Loading Step Optimization

Different amounts of QA were loaded in SPE cartridges, using methanol as the loading solvent. These amounts were 5, 10, 15 and 20 μmol QA/g polymers. This was followed by a washing step using 2 mL acetonitrile and an elution step using 2 mL of 10% acetic acid in water. The amount of QA in the eluted fraction was calculated and the recovery percent was determined, and the results are shown in Table 4.

The loading amount that showed the highest recovery percent was 5 μmol/g, where the recovery percentage was 72.53 ± 2.68 for the MIP and 56.55 ± 2.77 for the NIP. As the loading amount increased, a decrease in the recovery percentage was attained. This is most probably attributed to occupation of binding sites in the polymer. Therefore, as the loaded amount increased, the fraction of QA that binds to the polymer decreased.

In the following experiment, 5 μmol/g of QA was loaded in different solvents. These solvents were methanol, water, ethanol, ethanol water (1:1 *v*/*v*) and acetonitrile: water (4:1 *v*/*v*). The recovery percent of QA was then calculated and the results are shown in Table 4. It was observed that when water was used as the loading solvent, there was a significant decrease in the recovery % of QA, which agrees with the previously conducted rebinding studies. This confirms that using water decreases the interaction between QA and the polymers.


**Table 4.** Percent recoveries of QA during SPE optimization.

There was no significant difference between the recovery percent of the MIP and the NIP when using ultrapure water as loading solvent, which is opposite to what was observed during equilibrium rebinding studies. This could be attributed to the short contact time between the template and the polymer in addition to the short length of the polymer column. It could be concluded that increasing the contact time may have a positive impact on the recovery percentage [67].

It was observed that when a mixture of water and an organic solvent was used, (EtOH:H2O) or (ACN:H2O), only a slight increase in the recovery percent was observed compared to pure water. However, the use of absolute ethanol caused a remarkable increase in the recovery percent, reaching 101.76 ± 1.96.

The corresponding NIP showed a recovery percent of 63.49 ± 5.84. Comparing the two loading solvents, methanol and ethanol, methanol showed lower recovery percent (72.53 ± 1.68), suggesting that the use of less polar organic solvent (with lower dielectric constant, where the dielectric constant for methanol is 32.70 and 24.55 for ethanol) enhances the binding between QA and the MIP. Thus, ethanol was chosen as the loading solvent in further steps.

#### 3.6.2. Washing Step Optimization

The washing step is a crucial step during SPE to maximize the specific interactions between the analyte and the binding sites of the MIP and decrease non-specific interactions [68]. Different washing solvents were used, including acetonitrile, water, ethanol and ethyl acetate. Acetonitrile showed significant increase in the recovery percentage (101.76 ± 1.96) compared to other solvents. It revealed the best selectivity as well as the best retention ability. Acetonitrile is an organic aprotic solvent, where QA is insoluble. This probably enhances the QA–polymer interaction, decreasing the amount of QA lost during the washing step. On the other hand, using water as the washing solvent caused a significant decrease in the recovery percent (44.46 ± 2.55). This could be attributed to the fact that QA is a polar molecule that is highly soluble in water. Additionally, water may disrupt the hydrogen bonds formed between QA and the polymer. For both reasons, some of the QA is lost during the washing step, decreasing its recovery percent. The use of another organic aprotic solvent, ethyl acetate, showed a higher recovery percent

(68.38 ± 3.98) compared to water and ethanol, a polar protic solvent. From the obtained results, it was concluded acetonitrile is the optimum washing solvent.

#### 3.6.3. Elution Step Optimization

The desorption of the analyte is achieved using a solvent that develops interactions with the sorbent in order to desorb the analytes retained on the MIP [69]. Water: acetic acid (9:1 *v*/*v*) was chosen as the elution solvent as QA is soluble in water, while the addition of small amount of acetic acid enhanced the disruption of the hydrogen bond between QA and the polymer without causing a major change in the polymer morphology [70].

Different volumes (1 mL, 2 mL, 3 mL and 4 mL) of the elution solvent were used in this step. There was a significant increase in the recovery percentage upon shifting from 1 mL to 2 mL elution solvent. However, on using 3 mL of the elution solvent, there was no significant increase in the recovery percentage, while using 4 mL of the elution solvent decreased the difference in the recovery between the MIP and the NIP. Accordingly, 2 mL of water: acetic acid (9:1 *v*/*v*) was used for elution as it was the lowest amount of solvent to achieve the highest recovery of the analyte.

#### *3.7. UHPLC-MS/MS Method Validation*

The validated method showed good linearity, LOD, LOQ, precision and accuracy. Further details are provided in the supplementary material.

#### *3.8. MIP Cartridge Reusability*

MIP C reusability was studied over ten adsorption–desorption cycles, following the optimized SPE protocol, and QA recovery percentage was calculated after each elution (Figure 7). The results revealed that in cycles 1–4, QA recovery was more than 93%. However, a significant decrease in the recovery was observed in fifth cycle, where the recovery percent reached 73.15 ± 4.77. Cycles 5–8 had comparable results with QA recovery ranging between 73 and 78%, then a small drop was observed in cycles nine and ten, where QA recovery percent reached 69.87± 5.20 and 68.58 ± 3.73, respectively.

**Figure 7.** Percent recovery of QA during different cycles of reusing MIP C.

#### *3.9. MISPE Selectivity*

The selective recognition and retaining properties of the MISPE and NISPE were evaluated. Two compounds were chosen in this study: caffeic acid (CA), having a comparable size to QA, and chlorogenic acid (CLA), an ester of QA and caffeic acid. Both compounds are found in high concentrations in coffee [71]. The results are shown in Figure 8.

Both MISPE and NISPE showed a higher recovery of QA compared to CA and CLA. This difference is probably due to the difference in polarity, where QA is highly hydrophilic with low solubility in acetonitrile, the washing solvent, while CA and CLA exhibit higher solubility in organic solvents. Therefore, some of the CA and CLA may have been removed during the washing step. The recovery percent of QA from MISPE was 82.30 ± 5.58, while it was 53.58 ± 2.77 in case of NISPE. It was observed that for both CA and CLA, the recovery of MISPE was higher than NISPE. For CA, the recovery percent was 23.71 ± 2.85 for MISPE and 14.28 ± 1.84 for NISPE, while for CLA, the recovery percent was 33.41 ± 0.90 for MISPE and 17.46 ± 3.28 for NISPE. This might be due to the structural similarities between QA and the two compounds, where CA is a small molecule with comparable size to QA, it also possesses carboxyl and hydroxyl groups, so it is assumed to bind with some of the functionalities present in the imprinted cavities of the MIP. As for CLA, it has a QA moiety that can fit in the MIP cavities, since CLA is an ester of QA and CA.

**Figure 8.** Recovery percentages of QA, CA and CLA upon loading an equimolar mixture of the three compounds to MISPE and NISPE, (*n* = 3), \* indicates *p* value ≤ 0.05 and \*\* indicates *p* value ≤ 0.01.

#### *3.10. MISPE Application on Coffee Extract* 3.10.1. UHPLC-MS/MS Method Validation

The validated method showed good linearity, sensitivity, precision and accuracy. Further details are provided in the supplementary material.

#### 3.10.2. MISPE Application on Coffee Extract

MIP C was tested for its ability to selectively extract QA from coffee extract. The optimized SPE procedure was applied to the aqueous extract of coffee beans. Two ml of coffee extract (0.25 mg/mL and 0.5 mg/mL) was loaded onto the SPE cartridge. This was followed by a washing step using 2 mL of acetonitrile and elution step using 2 mL of 10% acetic acid in water. It was noticed that when 2 mL of 0.5 mg/mL coffee extract was loaded onto MISPE, the recovery percent was only 36.50 ± 1.19 for MISPE and 28.47 ± 1.22 for NISPE. However, decreasing the concentration of the loaded extract to 0.25 mg/mL showed a significant increase in the recovery percent to reach 81.92 ± 3.03 for MISPE, while the NISPE showed a much lower recovery percent of 37.26 ± 0.84 using the same concentration of the extract (Figure 9). This concludes that the low recovery percent observed while using a higher concentration of the extract could be attributed to the saturation of the binding cavities within the MIP. These results prove that the optimized MISPE is superior to reported conventional methods for QA isolation, such as liquid–liquid extraction previously reported by Tuyun et al. [13], where the maximum QA recovery was found to be 66.906%. The UV chromatogram for aqueous coffee extract before and after loading to MISPE and NISPE shown in Figure 10 revealed that neither the MISPE nor the NISPE were able to bind significantly to any of the other components of the extract, while there was a significant decrease in the amount of the extract components in the elution fractions of both MISPE and NISPE, compared to the original amounts found in the loaded extract.

**Figure 9.** Mass chromatograms of QA (**A**) in 0.25 μg/mL coffee extract dissolved in ethanol: water (97:3 *v*/*v*) before loading to MISPE C, (**B**) the elution fraction obtained from MISPE procedure, (**C**) the elution fraction obtained from NISPE procedure.

**Figure 10.** UV chromatograms of QA (**A**) in 0.25 μg/mL coffee extract dissolved in ethanol: water (97:3 *v*/*v*) before loading to MISPE C, (**B**) the elution fraction obtained from MISPE procedure, (**C**) the elution fraction obtained from NISPE procedure.

#### **4. Conclusions**

The current study represents the use of cheap, selective and simple MISPE procedure for extraction of QA from coffee beans. Three bulk polymers based on three different functional monomers (allylamine, MAA and 4-VP) were synthesized and the molar ratio of each monomer to QA was optimized via computational studies. The 4-VP polymer showed better overall performance in comparison to the other two polymers, thus it was the polymer of choice for SPE application. MIP reusability was tested over ten adsorption– desorption cycles and showed a high recovery of QA (more than 93%) up to the fourth cycle. Selective extraction of QA was observed upon using the optimized MISPE procedure on an equimolar mixture of QA, CA and CGA. The recovery percent of QA was 82.30 ± 5.58, compared to 23.71 ± 2.85 and 33.41 ± 0.90 for CA and CLA, respectively. The application of MISPE for extraction of QA from aqueous coffee extract showed a recovery percent of 81.92 ± 3.03, with a significant reduction in the amounts of other components in the extract. The developed MISPE procedure represents a promising approach for selective extraction of QA from different complex herbal extracts that may be scaled to industrial applications. It can also be applied in the food and beverage industry to decrease the concentration of QA in coffee and enhance its taste. In conclusion, this study succeeded in the isolation of an important nutraceutical in a cost-effective, rapid, robust and reliable method.

**Supplementary Materials:** The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/polym14163339/s1. Figure S1: Computer modeled structures of the best conformations for (a) QA, (b) 4-VP, (c) allylamine, (d)MAA. Figure S2–S13: Computer modeled structures of the best conformations for QA-FM complexes. Figure S14: SEM images of (a) MIP A and (b) NIP A. Figure S15: SEM images of (a) MIP B and (b) NIP B. Figure S16: BET isotherms of (A) MIP A, (B) NIP A, (C) MIP B and (D) NIP B. Figure S17: (A) Pseudo-first order kinetics and (B) pseudo-second order kinetics for MIP C. Figure S18: Freundlich isotherm for MIP C. Figure S19: Calibration curve of QA in methanol over the concentration range of 0.001–0.2 mM. Figure S20: Calibration curve for QA in ethanol over concentration range of 0.2–40 μg/mL. Table S1: The calculated binding energies for complexes prepared in solvent phase. Table S2: Intra-day and Inter-day precision of QA determination in UPLC-MS/MS method. Table S3: Accuracy of QA determination in UHPLC-MS/MS method. Table S4: %RSD of inter-day and intra-day precision assay for UHPLC measurements. Table S5: Recovery % of spiked QA amount 1x, 2x, and 3x the amount of QA present in coffee extract (10, 20, and 30 μg/mL).

**Author Contributions:** Conceptualization: S.H.M., H.H., M.A.-H., B.M. and N.A.E.G.; Methodology: S.H.M., H.H., M.A.-H., B.M. and N.A.E.G.; Investigation: S.H.M., A.H. and M.A.-H.; Formal Analysis: S.H.M., M.A.-H. and N.A.E.G.; Validation: S.H.M., M.A.-H. and N.A.E.G.; Supervision: H.H., M.A.- H., B.M. and N.A.E.G.; Writing—original draft: S.H.M.; Writing—review and editing: S.H.M., H.H., M.A.-H., B.M. and N.A.E.G. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


**Monika Sobiech , Dorota Maciejewska and Piotr Luli ´nski \***

Department of Organic Chemistry, Faculty of Pharmacy, Medical University of Warsaw, Banacha 1, 02-097 Warsaw, Poland; monika.sobiech@wum.edu.pl (M.S.); dorota.maciejewska@wum.edu.pl (D.M.) **\*** Correspondence: piotr.lulinski@wum.edu.pl; Tel.: +48-22-5720643

**Abstract:** The paper describes the formation of six aromatic *N*-(2-arylethyl)-2-methylprop-2-enamides with various substituents in benzene ring, viz., 4-F, 4-Cl, 2,4-Cl2, 4-Br, 4-OMe, and 3,4-(OMe)2 from 2-arylethylamines and methacryloyl chloride in ethylene dichloride with high yields (46–94%). The structure of the compounds was confirmed by 1H NMR, 13C NMR, IR, and HR-MS. Those compounds were obtained to serve as functionalized templates for the fabrication of molecularly imprinted polymers followed by the hydrolysis of an amide linkage. In an exemplary experiment, the imprinted polymer was produced from N-(2-(4-bromophenyl)ethyl)-2-methylprop-2-enamide and divinylbenzene, acting as cross-linker. The hydrolysis of 2-(4-bromophenyl)ethyl residue proceeded and the characterization of material including SEM, EDS, 13C CP MAS NMR, and BET on various steps of preparation was carried out. The adsorption studies proved that there was a high affinity towards the target biomolecules tyramine and L-norepinephrine, with imprinting factors equal to 2.47 and 2.50, respectively, when compared to non-imprinted polymer synthesized from methacrylic acid and divinylbenzene only.

**Keywords:** *N*-acylation; phenethylamines; molecularly imprinted polymers; semi-covalent imprinting; tyramine

#### **1. Introduction**

Molecular-imprinting technology is engaged in searching for advanced selective materials with great potential for environmental, food, or biomedical analyses [1–4]. This technique forms polymers with desired selectivity towards a template, being a result of the interactions between the functional groups of template and monomer(s) prior to the polymerization process. The orientation of molecules is fixed through chemical crosslinking during the polymerization and then the removal of the template is undertaken to obtain a cavity in the molecularly imprinted polymer (MIP). Covalent or non-covalent imprinting strategies employ either chemical bonds or various weak interactions in the formation of template–monomer prepolymerization moieties [5].

The formation of stable prepolymerization structures is a critical step during the imprinting process. The use of a template with covalently bound polymerizable units (a functionalized template) prior to the polymerization resulted in the formation of welldefined binding sites in the polymer matrix. Chemical cleavage is required at the final stage of the process. The fabrication of MIPs, applying the monomer covalently bound to the template, is followed by different rebinding/adsorption approaches, e.g., the rebinding of the target analyte to the polymer matrix by covalent bonds [6] or adsorption of the target analyte via non-covalent intermolecular interactions [7].

The advancement of the use of a functionalized template resulted in a more homogeneous population of binding sites in the resultant MIP and greater binding-site integrity when compared to the MIP synthesized with a non-covalent strategy [8–10]. Hashim and co-workers [8] compared the covalent and non-covalent imprinting strategies for the

**Citation:** Sobiech, M.; Maciejewska, D.; Luli ´nski, P. N-(2-Arylethyl)-2 methylprop-2-enamides as Versatile Reagents for Synthesis of Molecularly Imprinted Polymers. *Polymers* **2022**, *14*, 2738. https://doi.org/10.3390/ polym14132738

Academic Editor: Michał Cegłowski

Received: 22 May 2022 Accepted: 2 July 2022 Published: 4 July 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

synthesis of stigmasterol imprinted polymers. In the non-covalent imprinting strategy, stigmasterol was selected as the template and methacrylic acid or 4-vinylpyridine were used as the functional monomers to form different MIPs. In the covalent imprinting strategy, stigmasteryl-3-*O*-methacrylate was synthesized prior to its application as the functionalized template. It was found that the non-covalent imprinting method showed insufficient binding affinity and low selectivity towards stigmasterol. In contrast, the application of covalent imprinting in the formation of MIP followed by the chemical cleavage of ester bonds resulted with highly selective imprinted polymer. Similar results were described by Tang and co-workers [9]. Here, the simultaneous reaction of *N*- and *O*-acylation of ractopamine was provided to obtain a novel functionalized template. The results were compared with previously published studies of non-covalent imprinting of ractopamine [10]. It was found that MIPs obtained by the covalent imprinting strategy possessed significantly higher binding capacity and selectivity. The homogeneous population of binding sites towards ractopamine in the covalently produced MIP was confirmed by isotherm equilibrium-binding experiments [9], providing a substantial difference between it and the heterogeneous population of binding sites observed in the non-covalently formed MIP [10]. More recently, the covalent approach was used to synthesize more advanced and selective materials integrated with metal-organic frameworks [11], magnetic cores [12], surfaces of microtiter plates [13], or dendritic fibrous silica [14].

The current investigations of our group aim to explore the ability of MIPs produced by the covalent imprinting strategy to selectively recognize biogenic amines with 2 phenylethylamine system (phenethylamine system) [15,16]. The group of compounds that contain a backbone of phenethylamine play a very important role in the human nervous system. These molecules act as neurotransmitters and neuromodulators or psychotropic agents, causing neurological disorders related to mood, emotion, attention, and cognition when their levels are irregular or produce hallucinations, illusions, or mental disorders when prolonged or overdosed [17,18]. Moreover, the presence of these compounds, predominantly at low levels in highly complex samples, hampers their analysis. Thus, investigations of selective materials with satisfactory clean-up capabilities for the separation of phenethylamines are completely justified. Here, advanced polymeric materials could improve the detection of the above-mentioned biomolecules as well as contribute to explaining some aspects of neurological diseases or drug addiction.

The aim of the study was to synthesize and characterize various N-(2-arylethyl)-2 methylprop-2-enamides. The potential of those reagents for the fabrication of specific MIPs was proved in an exemplary synthesis of a molecularly imprinted polymer, using one of synthesized compounds as a functionalized template, followed by the analysis and characterization of the resultant material. In this paper, the synthesis of compounds that possess a template fragment of phenethylamine covalently bound with a polymerizable unit is presented, as well as the primary verification of their ability to synthesize an imprinted sorbent for the analysis of phenethylamines.

#### **2. Materials and Methods**

#### *2.1. Materials*

#### 2.1.1. Reagents

The following are the relevant ethylamines: 2-(4-fluorophenyl)ethylamine (**1a**), 2-(4 chlorophenyl)ethylamine (**1b**), 2-(2,4-dichlorophenyl)ethylamine (**1c**), 2-(4-bromophenyl)eth ylamine (**1d**), 2-(4-methoxyphenyl)ethylamine (**1e**), 2-(3,4-dimethoxyphenyl)ethylamine (**1f**). The target analytes are as follows: tyramine; L-norepinephrine or 3,4-dihydroxyphenyl acetic acid; and methacrylic acid, methacryloyl chloride, divinylbenzene were sourced from Sigma-Aldrich (Steinheim, Germany). The polymerization reaction initiator, 1,1 azobiscyclohexanecarbonitrile, was from Merck (Darmstadt, Germany). The relevant solvents (ethylene dichloride, hexane, petroleum ether, toluene, methanol, hydrochloric acid (36%), triethylamine, and salts used in the synthesis and the post-polymerization

treatment) were delivered from POCh (Gliwice, Poland). Ultra-pure water delivered from a Milli-Q purification system (Millipore, France) was used to prepare water solutions.

#### 2.1.2. Synthesis of N-(2-arylethyl)-2-methylprop-2-enamides

The selected amines **1a**–**1f** (10 mmol) were added to ethylene dichloride (30 mL) and triethylamine (1.67 mL, 12 mmol) under stirring. Subsequently, methacryloyl chloride (1.18 mL, 12 mmol, 20% excess) was slowly added dropwise to the reaction mixture for 15 min. A white precipitate of triethylammonium chloride was formed and the reaction mixture was stirred for 3 h in room temperature. The precipitated salt was filtered off and the remaining organic layer was washed with saturated NaHCO3 and water, prior to drying over anhydrous MgSO4. Following this, the layer was evaporated under vacuum yielding crude solid products which were recrystallized from hexane or petroleum ether to obtain pure products **2a**–**2f** (Scheme 1).

; )&+**D**&O&+**E**&O&+**F**%U&+**G** 

0H2&+**H**0H2&+**I** 

**Scheme 1.** Synthesis of *N*-(2-arylethyl)-2-methylprop-2-enamides.

#### 2.1.3. Preparation of Polymers

The synthesis of the imprinted material, coded MIP**ft**, was carried out. The functionalized template (ft) was used to form an imprinted material. A functionalized template could be defined as the template that possesses one or more polymerizable units that are attached by covalent bonds to form a template–monomer structure by a chemical step independent of polymer formation. In the synthesis of MIP**ft**, the compound **2d** (175.39 mg; 0.8 mmol) was added to toluene (2.056 mL). In the synthesis of NIP, methacrylic acid (69 mg; 0.8 mmol) was added to the same volume of toluene. Following this, divinylbenzene (570 μL; 4 mmol) and 1,1 -azobiscyclohexanecarbonitrile (10 mg) were added to the prepolymerization mixture prior to purging the mixture with nitrogen for 3–5 min. The polymerization process was carried out in 88–92 ◦C for 24 h. Subsequently, the bulk rigid polymers were ground in a mortar with a pestle and wet-sieved into particles below 45 μm diameter prior to discarding the fine particles by repeated decantation in acetone. Following this, the imprinted particles were treated under reflux with 1 mol L−<sup>1</sup> hydrochloric acid for 3 h (50 mL) in order to hydrolyze the amide linkage and to remove 4-bromophenylethylamine residue. For comparison, NIP was treated in the same way. Finally, the particles were extensively washed with methanol and were dried prior to the analysis.

#### 2.1.4. Binding Studies

Empty 1 mL solid-phase extraction cartridges were filled with 25 mg of MIP**ft** or NIP and secured by fiberglass frits. The polymers were then conditioned with methanol–water (85:15 *v*/*v*, 1 mL), and loaded with a standard solution of **1d** (conc. 50 μmol L−1, methanol– water, 85:15 *v*/*v*, 5 mL) or tyramine (conc. 20 μmol L−1, methanol–water, 85:15 *v*/*v*, 5 mL). The unbound amounts of **1d** or tyramine were determined with reference to their respective calibration lines and UV spectroscopy was used for detection. The bound amounts were calculated by subtracting the unbound amount of **1d** or tyramine from initial amount of **1d** or tyramine, respectively. For selectivity tests, L-norepinephrine (conc. 20 μmol L−1) or 3,4-dihydroxyphenylacetic acid (conc. 20 μmol L−1) were used and the analysis was carried out in the same way as described above. The binding capacities and specificity

of materials were evaluated [5,19]. The binding capacities (*B*, μmol g−1) were calculated according to Equation (1):

$$B = \frac{\left(\mathbb{C}\_i - \mathbb{C}\_f\right)V}{M} \tag{1}$$

where *V* represents the volume of portion (L) in each loading step, *Ci* represents the initial analyte concentration (μmol L−1), *Cf* represents the analyte concentration in solution after adsorption (μmol L−1), and *M* is the mass of polymer particles (g). The binding capacities of MIP**ft** and NIP were compared by the determination of the imprinting factor (IF) calculated according to Equation (2):

$$\text{IF} = \frac{B\_{MIP}}{B\_{NIP}} \tag{2}$$

For isotherm analysis, equilibrium-binding experiments were applied. The polypropylene tubes were filled with 10 mg MIP**ft** or NIP. Next, a volume of 50 mL of methanol– water (85:15 *v/v*) standard solution of tyramine was added (concentration range between 10–200 μmol L−1). The tubes were sealed and oscillated by a shaker at room temperature for 24 h prior to centrifugation for 10 min. The aliquots were then used to analyze the unbound amount of tyramine. The amount of tyramine bound to the polymer was calculated according to Equation (1). The detailed analyses of adsorption on MIP**ft** and NIP were provided using Lineweaver–Burk model [20] represented by Equation (3):

$$\frac{1}{B} = \frac{1}{B\_{\text{max}}} + \frac{1}{K\_{\text{L}} B\_{\text{max}} F} \tag{3}$$

where *B*max is the maximum binding capacity and *K*<sup>L</sup> is the equation constant. The Freundlich model [21] represented by Equation (4) was also employed:

$$B = aF^m\tag{4}$$

where *B* is the adsorbed amount of analyte, *F* is the unbound amount of analyte, *a* is the measure of the capacity (*B*max) and *m* is a heterogeneity index.

#### 2.1.5. Physicochemical Characterization

The 1H NMR, 13C NMR spectra of **2a**–**2e** and the 13C CP/MAS NMR spectrum of MIPft in solid state were recorded with a Bruker Avance DMX 400 spectrometer (Bruker, Germany) at the Faculty of Pharmacy, Medical University of Warsaw, Poland. For the 13C CP/MAS NMR, a powdered sample was contained in 4 mm ZrO2 rotors and was spun at 8 kHz. The 90◦ pulse length was 2.15 μs. A contact time of 4 ms and a repetition time of 10 s were used for the accumulation of 1900 scans. The chemical shifts, δ ppm, were referenced to tetramethylsilane.

Spectroscopic analyses were carried out using a UV-1605PC spectrophotometer (Shimadzu, Germany). The calibration lines as a function of absorbance (y) versus concentration (x) were constructed at λmax of the analyzed compounds. Each point was measured in triplicate. The linearities of calibration lines were good, with correlation coefficients r<sup>2</sup> > 0.997.

Scanning electron microscopy (SEM) and energy-dispersive X-ray spectrophotometry (EDS) for MIP**ft** were studied on Merlin FE-SEM (Zeiss, Germany) combined with an EDS X-ray detector (Brucker, Germany). The samples were Au/Pd sputter-coated (SEM) or carbon-coated (EDS) before analysis. The analyses were performed at the Faculty of Chemistry, University of Warsaw, Poland.

The porosity data for MIP**ft** were determined using the adsorption isotherm of N2 at 77 K (BET) on an ASAP 2420 system (Micromeritics Inc., Norcross, GA, USA) at the Faculty of Chemistry, Maria Curie-Skłodowska University, Lublin, Poland.

#### **3. Results and Discussion**

#### *3.1. Synthesis and Identification of N-(2-arylethyl)-2-methylprop-2-enamides*

In order to obtain reagents for the fabrication of imprinted polymers, the procedure for the synthesis of six *N*-(2-arylethyl)-2-methylprop-2-enamides from *N*-(2-arylethyl)amines with various substituents (methoxy or halogens) in benzene ring was described (Scheme 1). The respective *N*-(2-arylethyl)-2-methylprop-2-enamides were isolated with high yields (Table 1) and their chemical structures were determined by high-resolution mass spectrometry, IR and NMR analyses.

**Table 1.** Amines (**1a**–**1f**) used in synthesis followed by yields and melting points of respective *N*-(2-arylethyl)-2-methylprop-2-enamides (**2a**–**2f**).


The conversion of phenethylamine to phenethylamide is usually a straightforward matter, involving reactions with acyl chloride [22] or acetyl anhydride [23]. Weiner and co-workers [22] used methacryloyl chloride as acylating agent to obtain *N*-(2-(4 hydroxyphenyl)ethyl)-2-methylprop-2-enamide. The corresponding *N*-acylations were made in ethyl ether or in benzene, but are inconvenient because of the low boiling point and high toxicity, respectively. Rathelot and co-workers [23] used the *N*-acylation reaction of phenethylamine to obtain 2-phenylethylacetamide, a substrate in the multistep synthesis of novel antimalarial drugs. In the above-mentioned reaction, acetic anhydride was applied as the *N*-acylation reagent.

Here, ethylene dichloride was used as a solvent and methacryloyl chloride as acylating agents to obtain derivatives of *N*-(2-arylethyl)-2-methylprop-2-enamides. The presence of triethylamine provides a satisfactory method of neutralizing the hydrogen halide for

amide synthesis. The reaction was finished within 3 h at room temperature. A series of 2-arylethylamines with various substituents in benzene ring, viz., 4-F, 4-Cl, 2,4-Cl2, 4-Br, 4- OMe, 3,4-(OMe)2, **1a**–**1f** were then used as substrates (Table 1). These amines were selected because of their structural similarity to the biogenic amines or synthetic psychoactive substances used in designer drugs. The conversion of substrates was monitored by TLC. The formation of any other products apart from the main products of the *N*-acylation reaction was not observed. The structures of compounds **2a**–**2f** were confirmed by spectroscopy (Table 2).

**Table 2.** Spectral data of synthesized compounds.



It has to be underlined that the screening of chemical databases revealed only the presence of *N*-(2-(4-iodophenyl)ethyl)-2-methylprop-2-enamide [24] and *N*-(2-(2-bromophenyl)e thyl)-2-methylprop-2-enamide [25,26].

Hence, it could be assumed that the *N*-acylation reaction of phenethylamines with methacryloyl chloride is effective for phenethylamines with the strong electron-donating group –OCH3 or with the halogens F, Cl, or Br in a benzene ring.

#### *3.2. Preparation of Polymer*

In an antecedent paper, Weiner and co-workers [22] described the effect of the attachments of various phenethylamines, viz., phenethylamine, tyramine, ephedrine (2(methylamino)-1-phenylpropan-1-ol), and amphetamine into synthetic or natural polymers by an amide or carbamate linkage as a method for increasing their duration of action. However, the imprinting technique was not studied in the above-mentioned paper.

Here, we employed the covalent imprinting approach to synthesize the polymers but the adsorption process on imprinted material was entirely non-covalent in nature [27]. Thus, so-called 'semi-covalent imprinting' was applied. In order to confirm the structure, to analyze the morphology, and to prove the selectivity of the resultant imprinted polymer, the synthesis of MIP**ft** from *N*-(2-(4-bromophenyl)ethyl)-2-methylprop-2-enamide, **2d**, was carried out in the presence of divinylbenzene (cross-linker) in toluene (porogen). Radical thermal polymerization was applied to obtain bulk material. The schematic idea of the synthesis, employing a covalent strategy for the imprinting process and an adsorption process on the resultant imprinted polymer that is based on non-covalent interactions of the target analytes, is presented in Figure 1. The NIP was prepared with the employment of methacrylic acid as the functional monomer and divinylbenzene as the cross-linker, omitting the addition of any template, and it was fabricated to compare sorption behavior**.**

**Figure 1.** Schematic idea of the synthesis, employing a covalent strategy for the imprinting process and an adsorption process on the resultant imprinted polymer that is based on non-covalent interactions of the target analytes.

The amide linkage, present in the MIP**ft**, was hydrolyzed prior to the removal of 2-(4-bromophenyl)ethylamine residue in order to form specific cavities in the imprinted polymer. The hydrolysis was carried out in 1 mol L−<sup>1</sup> hydrochloric acid under reflux. The hydrolysis lasted up to 3 h.

#### *3.3. Characterization of Material*

#### 3.3.1. Adsorption Behavior

The binding capacities were determined for MIP**ft** as well as for NIP in one μmol of **1d** for one gram of polymer particles. The selectivity of the imprinted polymer (imprinting factor, IF) was expressed as the ratio of the amount of **1d** bound to MIP**ft** in comparison to NIP.

The binding capacities of **1d** and IF were as follows: MIP**ft**: 8.09 <sup>±</sup> 0.08 <sup>μ</sup>mol g−1, NIP: 4.59 <sup>±</sup> 0.04 <sup>μ</sup>mol g−1, IF = 1.76. These results proved that the functionalized template **2d** could be used as reagent for covalent imprinting and the resulting MIP**ft** was possessed of selectivity when compared to NIP.

Sorption behavior is responsible for the effective separation of target analytes on MIPs. In order to prove that the novel imprinted material possessed an affinity towards biogenic amines with a phenethylamine system, the adsorption of tyramine on MIP**ft** was examined. The binding capacities of tyramine on MIP**ft** and on NIP were as follows: 2.74 <sup>±</sup> 0.03 <sup>μ</sup>mol g−<sup>1</sup>

and 1.11 <sup>±</sup> 0.01 <sup>μ</sup>mol g−1, respectively (IF = 2.47). In order to analyze adsorption data, the Langmuir isotherm was applied [28]. Here, various linearized modifications of the model could be used to determine the most proper fit [29]. It was found that the Lineweaver– Burk modification represented by Equation (3) was characterized by the highest regression coefficients for analyzed data with values of *r*<sup>2</sup> = 0.974 and *r*<sup>2</sup> = 0.971 for MIP**ft** and NIP, respectively. Moreover, to reveal the homogeneity of MIP**ft**, the Freundlich model described by Equation (4) was applied. That model fits well to MIP adsorption data in the low-concentration regions and allows for the surface-homogeneity determination of the material tested. The straight line of log *B* versus log *F* is the evidence that adsorption can be described by the Freundlich equation. The correlation coefficients, *r*2, for MIP**ft** and NIP were equal to 0.999 and 0.972, respectively, and the estimated values of *m* were equal to 0.70 and 0.93 for MIP**ft** and NIP, respectively (Figure 2).

**Figure 2.** Lineweaver–Burk (**a**) and Freundlich (**b**) models for tyramine adsorption on MIP**ft** and NIP.

It was found that MIP**ft** provided affinity towards tyramine. In order to analyze the selectivity, two other biomolecules were used for studies, viz., L-norepinephrine and 3,4-dihydroxyphenylacetic acid. L-norepinephrine possesses a phenethylamine system. In contrast, 3,4-dihydroxyphenylacetic acid (a metabolite in the dopamine system) does not possess a phenethylamine system. The binding capacities for MIP**ft** and NIP were as follows: for L-norepinephrine, 14.5 <sup>±</sup> 1.4 <sup>μ</sup>mol g−<sup>1</sup> and 5.80 <sup>±</sup> 0.59 <sup>μ</sup>mol g−1, respectively (IF = 2.50); and for 3,4-dihydroxyphenylacetic acid, 0.579 <sup>±</sup> 0.070 <sup>μ</sup>mol g−<sup>1</sup> and 0.549 <sup>±</sup> 0.060 <sup>μ</sup>mol g−1, respectively (IF = 1.06). The results show the selectivity of MIP**ft** to L-norepinephrine, a biomolecule with a phenethylamine system, and its lack of selectivity to 3,4-dihydroxyphenylacetic acid, a biomolecule that not possess a phenethylamine system. The higher binding capacity of L-norepinephrine when compared to the binding capacity of tyramine could be explained by the presence of two strong electron-donating hydroxy groups in positions 3 and 4 of the aromatic ring and one hydroxy group in the aliphatic ethylamine chain, enhancing strongly the basicity of L-norepinephrine. Apart from being an exemplary experiment, the results are very promising for the possible application of such reagents in the preparation of sorbents for biomedical purposes.

#### 3.3.2. Morphology Characterization

In order to provide morphological characterization, scanning electron microscopy was employed and the surface of MIP**ft** was analyzed. Figure 3 presents a micrograph of MIP**ft** after the hydrolysis process. The particles possessed the morphology of the bulk materials that were composed from spherical entities agglomerated into bigger forms of 10–20 μm. The diameter of single entity varied from 500 nm to 2 μm and was similar for MIP**ft** and NIP (Figure 3a–d). However, further magnification revealed substantial difference between MIP**ft** and NIP (Figure 3e–h). The MIP**ft** was characterized by significant surface extension with numerous macropores clearly detected on the particle's surface. On the contrary, the NIP possessed a smoother surface. The difference could be related to the presence of the functionalized template in the prepolymerization mixture.

#### 3.3.3. Structural Evaluation

In order to confirm the structure of the resultant polymers, EDS was used to prove that the functionalized template was polymerized (Figure 4a). The MIP**ft** was prepared in order to confirm that the functionalized template was built up into the polymer matrix because the presence of bromine atoms in the structure of **2d** allowed us to detect heteroatom during the analysis of the materials. It has to be underlined that the analysis of the polymer, viz., MIP**ft**, was carried out omitting the process of the hydrolysis of the amide linkage. The atoms of bromine were detected in the MIP**ft** structure in the region of 1.50 keV.

**Figure 3.** *Cont*.

**Figure 3.** SEM micrographs of MIP**ft** (**a**,**c**,**e**,**g**) and NIP (**b**,**d**,**f**,**h**).

13C CP/MAS NMR spectroscopy was then applied. This is a versatile tool to confirm the composition of polymer materials. For the purpose of our analysis, the MIP**ft** was post-treated to remove 2-(4-bromophenyl)ethylamine residue from the polymer matrix (Figure 4b). In the 13C CP/MAS NMR spectrum of MIP**ft**, strong resonances in the aromatic region, representing quaternary benzene C atoms at 137.2 and 144.3 ppm, could be seen. The tertiary –CH atoms at 127.0 ppm originated from the cross-linker. In the aliphatic region, various methyl groups were represented by broad peaks located between 15 and 30 ppm with a narrower sharp peak at 28.7 ppm. Methylene groups in C–CH2–C were found in approximately 44.4 ppm and methylene groups in Φ–CH–C in 39.8 ppm. Carboxyl group, –COOH, atoms were represented by broad resonances in the region of 177.1–182.7 ppm. Low intensity resonance at 111.7 ppm could originate from unreacted double bonds in Φ–(CH=CH2)2.

#### 3.3.4. Porosity Data

Finally, the nitrogen-adsorption isotherm (Brunauer–Emmett–Teller) for MIP**ft** was analyzed. As it can be seen (Figure 4c) the material revealed physisorption isotherms of type IV with a hysteresis loop. The shape of the hysteresis loop is related to the specific pore structure. Here, type H3 loops characterized MIP**ft**, indicating the slit-shaped structure of its pores. However, the deformation of the desorption-hysteresis line of MIP**ft** in the region of 0.50 P/Po could be related to the expulsion of adsorbate from larger-volume mesopores through narrower pore necks. Thus, a more complicated pore system could exist in MIP**ft**. The total specific surface area of MIP**ft** was equal to 89.88 m<sup>2</sup> g−<sup>1</sup> and the

external surface area was equal to 77.99 m2 g−1. The plots of pore volume versus diameter for MIP**ft** showed a peak for a pore diameter of 56 nm (Figure 4d).

#### **4. Conclusions**

In conclusion, it should be emphasized that a series of compounds, *N*-(2-arylethyl)-2 methylprop-2-enamides, were obtained with the synthetic procedure presented here with high yields. These compounds possessed fragments of a template covalently bound to polymerizable units and could be used as reagents for the covalent imprinting of polymers. In the control experiment, one of synthesized compounds was used to produce an imprinted polymer (MIP**ft**). Binding-capacity analysis revealed that a molecular imprinting process took place and the polymer, MIP**ft**, possessed selectivity towards biomolecules of tyramine and L-norepinephrine.

**Author Contributions:** Conceptualization, P.L.; methodology, P.L.; validation, P.L.; formal analysis, M.S. and P.L.; investigation, P.L.; resources, P.L.; data curation, P.L. and D.M.; writing—original draft preparation, P.L., M.S. and D.M.; writing—review and editing, P.L. and M.S.; supervision, P.L. and D.M. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Not applicable.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**

