In an attempt to explain the enzyme–substrate interactions, the lock-and-key mechanism was originally proposed by Nobel laureate Emil Fischer back in the 1970s. Although this theory was later disproved by X-ray diffraction studies, the idea of a matrix-designed material that recognizes a particular substrate remained the cornerstone of molecular imprinting [1]. Over decades of research, molecularly imprinted polymers (MIPs) have gained more and more acceptance with respect to their application as polymeric antibodies in analytical chemistry and separation science [2,3,4]. MIPs are solid materials that can be synthesized via a molecular imprinting process, in which a template molecule (the target compound) is present during polymerization. Thus, template molecules are imprinted into the polymer via the molecular imprinting process. After the removal of the template from the polymer, the end-product polymer is called a MIP, in which specific binding cavities with a shape and functional groups complementary to the template are created within the polymer matrix [5]. Due to the presence of specific binding cavities in MIPs, they exhibit specific selectivity for the template molecule, or group selectivity for a class of structurally related molecules. Thus, MIPs mimic the function of natural antibodies (monoclonal or polyclonal) [6,7]. They often afford high adsorption capacities (~50 mg/g), good site accessibilities, fast binding kinetics (~10 min), and good recoveries (~90%). Many applications using MIPs have been reported, such as the removal of 2,4-dichlorophenol from contaminated water [8], the sensing of amino acids via a quartz-crystal microbalance (QCM) [9], the detection of dansylated amino acids via a surface plasmon resonance sensor [10], the sensing of zearalenone with fluorescent probes [11], the detection of caffeine with molecularly imprinted quantum dot photoluminescence [12], the detection of amino acids via a capacitive sensor [13], food analysis [14,15,16], and the recognition of the tobacco mosaic virus [17], just to name a few. The application of MIPs as sorbents allows not only pre-concentration and cleanup of the sample, but also selective extraction of the target analyte, which is particularly important when the sample matrix (such as environmental and biological samples) is complex and impurities can interfere with quantification [18].

Mycotoxins are secondary metabolites that molds produce naturally. Ochratoxin A (OTA) is a carcinogenic mycotoxin of wide natural abundance, as produced by several species of Aspergillus (e.g., A. ochraceus) and Penicillium (e.g., P. verrucosum) fungi [19,20,21]. It is suspected to cause the Balkan Endemic Nephropathy, which is a fatal kidney disease observed in rural areas of southeast Europe [22]. OTA has been found as a contaminant in food and feed and exhibits multiple toxicities in animals and mankind, including nephrotoxic, hepatotoxic, immunotoxic, teratogenic, and carcinogenic effects, which represent serious health risks to livestock and the general population [23]. The widespread occurrence of OTA in cereals, wheat [24], maize, rice, beans, nuts, raisins, and beverages (such as milk, coffee, grape juice [25,26], and wine [27,28,29]) has prompted health regulation authorities to define maximal tolerable daily intake levels (5 ng/kg body weight) [30]. Epidemiology studies in Bulgaria, Romania, Spain, the Czech Republic, Turkey, Italy, Egypt, Algeria, and Tunisia have also found significantly higher serum or plasma levels of OTA in patients with certain kidney disorders compared to healthy people, although the association may not be a causal one [31]. Due to their ubiquitous presence in foodstuffs and their potential risk for human health, prompt detection is deemed important. An excellent review on analytical methods for the determination of mycotoxins has recently been published by Turner and co-workers [32]. The development of analytical methods for OTA identification generally involved liquid–liquid extraction, clean-up by an immunoaffinity column (IAC), and identification by HPLC with fluorescence detection (HPLC-FLD) [33]. Various extraction and clean-up procedures for the determination of OTA by HPLC-FD in musts, wine, and beer were compared: (1) dilution with polyethylene glycol 8000 and NaHCO₃ solution and clean-up on an IAC; (2) extraction with chloroform and IAC clean-up; (3) solid phase extraction (SPE) on C18; (4) SPE on reverse-phase phenylsilane; and (5) SPE on Oasis HLB cartridges [34]. The former IAC procedure was simple, rapid, and provided flat baselines that were free from most impurity peaks, high OTA recoveries and quite repeatable results. Unfortunately, the IACs are relatively costly and have a short shelf life. For the assay of real samples (cereals, oat, corn, etc.), denaturation of antibodies by the organic co-solvent needed to dissolve hydrophobic analytes in aqueous solution is a serious problem. Thus, the design and the synthesis of biomimetic antibodies that can specifically bind a target molecule has long been a research goal [35]. The use of MIPs as artificial antibodies for sample pretreatment was described, first in a review on emerging sorbent materials for SPE [36], and next in a report on immuno-based sample preparation for trace analysis [37]. Over the next few years, MIPs have increasingly attracted attention as substitutes for immunoanalysis (e.g., binding assays, biosensors, and solid-phase immune-extraction) [38]. The development of a new sorbent for selective SPE that is capable of OTA preconcentration prior to HPLC analysis is an important issue. An affordable SPE sorbent, compared to the use of IAC columns, will make the screening of foodstuff more frequent in both developed and poor countries, thus protecting human health [39]. An overview of conventional and emerging analytical methods for the determination of mycotoxins is written by Cigić and Prosen [40].

Although molecular imprinting has been around for over 30 years, recently, this technology has made rapid developments. Due to the variety of structures of mycotoxins, it is possible to use the molecular imprinting technique for their analysis and selective detection. As early as 2004, a two-dimensional extraction procedure employed solid phase extraction (SPE) and MIP for the extraction of OTA [41]. Here direct sample loading onto the MIP resulted in low recoveries, thus prior removal of interfering acidic matrix compounds by C18 SPE was deemed necessary. Unfortunately, a similar result was observed in a control experiment, in which the MIP was replaced by the corresponding non-imprinted polymer (NIP). These findings suggested that specific binding to molecularly imprinted sites played a minor role in OTA enrichment. Recent developments in imprinting technology have made possible the practical application of MIPs in mycotoxin detection. The structure activity relationships of reported MIPs for OTA, deoxynivalenol (DON), zearalenone (ZEA), and moniliformin (MON) have been reviewed by Appell, Maragos, and Kendra [42]. MIPs offer quite a promising tool for the future development of antibody-based methods, in which compatibility between sample extract and antibody is a major limiting factor. The current trend in OTA determination is largely towards the powerful combination of liquid chromatography with tandem mass spectroscopy (LC-MS/MS) [43]. An emerging focus of LC-MS/MS is in the field of multi-residue methods for the simultaneous determination of mycotoxins that are being considered by the EU legislation in force (i.e., aflatoxins, fumonisins, ochratoxin A, trichothecenes, and zearalenone). This topic has been investigated by several research groups, although major problems with extraction and cleanup steps have not been fully resolved [44]. LC-MS/MS is a powerful tool for the identification and quantification of masked mycotoxins, which are mycotoxins conjugated to more polar compounds (e.g., glucose) that are not detected by routine analytical methods.

Although most validated detection methods are chromatographic, alternative strategies based on biosensing principles are emerging. The factors which worked against unattended, continuous monitoring of toxins included the inherent instability of suitable protein receptors and antibodies, and the irreversible nature of the binding event (which necessitates a continuous supply of reagents for sequential measurements). Nevertheless, biosensors were hopefully capable of being used for on-site measurement of contamination by specific toxins. Methods for improving the stability, extending the range, and for altering the binding characteristics of sensing molecules would be essential, as discussed previously by Paddle [45]. An evaluation of MIP films for coulometry used an applied positive potential to induce adsorption of the target molecules [46]. The resultant sensors showed a high degree of sensitivity, selectivity, and a broad linear range. Imprinting a polymer matrix with binding sites located at the surface has been shown to be advantageous for use as the sensor interface. The binding sites are more accessible, the mass transfer is faster, and the binding kinetics are faster [47]. A molecularly imprinted polypyrrole (MIPPy) film was synthesized on the Spreeta sensor, a miniaturized surface plasmon resonance (SPR) device, for the detection of OTA [48]. The MIPPy was electrochemically polymerized on the sensor surface from a solution of pyrrole and OTA in ethanol/water (1:9 v/v). The film growth was monitored in situ by an increasing SPR angle. The binding properties of the MIPPy film were investigated by loading OTA standard solutions into the integrated 20-μL flow cell. After 300 s, nonlinear regression was used to determine the maximum binding signal. Spreeta results showed that the signal was measurable for OTA concentrations down to 0.05 ppm. Pulsed elution with 1% acetic acid in methanol/water (1:9 v/v) was found to be efficient for the regeneration of the MIPPy film surface. Interference by the matrices of wheat and wine extracts was evaluated. No significant binding of the wheat extract with MIPPy was observed when acetonitrile/water (1:1 v/v) was used as the mobile phase. Biosensors and sensor arrays provided selective, sensitive, and accurate measurements. The feasibility of miniaturizing biosensors and sensor arrays, so that they are portable, makes them useful as screening bio-tools meant to ensure the correct assessment of mycotoxins in food so as to reassure the consumer [49]. The interfacing of a suitable transducer to MIPs is still growing and is expected to have a more significant impact in the field of biochemical sensors. A rapid and highly sensitive SPR assay of OTA has recently been reported, using Au nanoparticles for signal enhancement on a mixed, self-assembled monolayer surface, in a competitive immunoassay format [50]. Although an enormous effort is being put into developing biosensors, relatively few toxic analytes can yet be measured by commercially available devices.

Among xenobiotics, applications of MS techniques for the analysis of toxins, pesticides, drug residues, amines, and migrants from packaging were previously overviewed [105]. OTA in grape was determined by nano-HPLC coupled with ESI-mass spectrometry [106]. Currently, there is a trend towards developing multi-mycotoxin methods for the simultaneous analysis of several Fusarium mycotoxins belonging to different chemical families that is best achieved by LC-MS/MS (liquid chromatography with tandem mass spectrometry). Krska et al. gave an overview of the commonly used methodology for the analysis of fumonisins (FBs), moniliformin (MON), zearalenone (ZON), and trichothecenes in feeds [107]. Good limits of detection for OTA in wine samples (1.3–3.4 μg/kg) by LC/MS/MS have also been demonstrated by Reinsch et al. [108]. In order to reduce the ion suppression effect on MS/MS analysis, the use of MISPE for sample clean-up will be a promising development in the future.

The major mycotoxin-producing fungi are species of Aspergillus, Fusarium, and Penicillium and the important mycotoxins are aflatoxins, fumonisins, ochratoxins, cyclopiazonic acid, deoxynivalenol/nivalenol, patulin, and zearalenone [109]. In 2006, MIP Technologies signed an exclusive distribution agreement with Supelco [110]. SPE cartridges that were prepared using MIP technology were termed SupelMIP SPE by the company. This commercialization of MIP materials offered tailor-made selectivity for the extraction of trace analytes in complex matrixes. The MISPE products provided faster sample preparation and better MS compatibility (reduced ion suppression), allowing analysts to achieve lower detection limits and improved sensitivity. By the end of 2009, there were 20 different SupelMIP SPE cartridges available in the market. Most target analytes were pesticides and drugs. Unfortunately, no SupelMIP SPE cartridge was made for ochratoxins. However, consultation on the preparation of new SPE phases is available from the company. In the future, development and research for other specific SupelMIP SPE cartridges will be expected.

To prepare MIPs with good recognition properties toward target compounds, proper selection of functional monomers has always been the first task for most researchers. The complexity of multiple interactions among the pre-polymerization components makes the selection of functional monomers still a big challenge. A simplified computational approach is often employed to develop a prediction of the resulting MIP. Predictive modeling of functional monomer interactions and the resulting binding site performance provides a more rational design of synthetic strategies for new MIPs. It can be expected that upon establishment of more complex models that provide better resemblance of the actual MIP preparation conditions, computational modeling will develop into a useful complementary tool, efficiently reducing the large number of MIPs that are currently screened to achieve optimal recognition properties. However, the accuracy of predictive simulations will strongly depend on the complexity of the model and on experimental conditions. Hopefully, refined simulations, along with supporting spectroscopic data, would lead to more accurate modeling of pre-polymerization complexes and binding sites. Spectroscopic studies can gather evidence for self-assembly based on ionic and π–π stacking interactions (rather than previously assumed hydrogen bridge bonding), leading to a refined model of the pre-polymerization complex [111]. In Table 1, a functional monomer with high affinity towards OTA, a mix of functional monomers, and a novel conductive polymer, are all feasible for the preparation of MIPs for OTA extraction and detection.

Direct imprinting of OTA was found to be not ideal for the preparation of OTA MIPs [69] due to its slow release of OTA template molecules. Structure analogs, such as N-(4-chloro-1-hydroxy-2-naphthoylamido)-(L)-phenylalanine, have been used as a surrogate in the preparation of MIPs for OTA extraction [80]. New desorption chemistry (such as electrochemical, electrospray desorption) with fast kinetics and efficiency for template removal should be investigated in the future. Ideally, slow release of template can be eliminated to prevent any potential carry-overs. As for the recovery of OTA from different samples matrices, possible binding of OTA to proteins or other sample matrix components was tested by acid treatment before extraction, but no significant differences with controls appeared [112]. It is in the best interest of researchers to find new ways for the ease of MIP preparation, ease of OTA removal from MIP, and ease of use for real-world samples. These remaining issues will definitely need a broad investigation of novel polymer chemistry.

MIPs are synthetic receptors with high-affinity sites that can selectively recognize a target analyte, based on its shape, size, or functional group distribution. These receptors are promising due to their easy preparation, thermal stability, chemical inertness, and long shelf life at room temperature and humidity. From the point of view of analytical chemistry, this protocol is very promising for applications in the extraction and analysis of ochratoxins. Recent investigations have led to the synthesis of new MIPs for a wider range of mycotoxins. Currently, MIPs are not selective enough in the aqueous environment to compete with natural antibodies [113], and better shape selectivity must be achieved in future development. In addition, mycotoxins are costly for the large-scale preparation of MIPs. The concept of template mimics, surrogate templates, or dummy templates must be used to prepare MIPs that are both affordable and selective for the target mycotoxins [105,114]. A couple of peptide receptors have recently been selected for ochratoxin A using computational methods by screening de novo designed peptide libraries. Affinity characterization resulted in K_A = 63 mM^-1 for the 13-mer peptide and K_A = 84 mM^-1 for the 8-mer peptide [115]. Last, researchers must be able to determine a more proper wash solution to remove OTA template molecules from the strongest-binding MIP cavities, especially in ultra-trace analysis (by LC-MS/MS).