Bio-Mimetic Sensors Based on Molecularly Imprinted Membranes

1. Introduction

Specific recognition at a molecular level represents a key necessity of living systems. Owing to this aspect, one of the most important challenges of many researchers is how to imitate the recognition ability of biological systems towards specific ligands. From this perspective, many efforts have been focused on the development of man-made materials with bio-mimetic properties. Among the different approaches used to produce these artificial systems, the “molecular imprinting technique” has received much attention in numerous fields due to its high selectivity for molecules of specific interest. The imprinting technology is based on the synthesis of a polymeric material endowed with specific recognition sites towards the molecule which must be recognized (template) [1]. This objective is accomplished by the addition of the template to a reaction mixture, which is constituted of a functional monomer, a cross-linking agent and a solvent. During the polymerization, the template is incorporated into the polymeric matrix and chemical groups of functional monomers will be arranged according to the shape and chemical properties of the template molecules. The extraction of the template from the obtained polymeric matrix will allow the formation of the template complementary recognition sites with high substrate selectivity and specificity [2,3]. In this way, selective detection and separation properties are introduced in the nascent polymer. In order to achieve efficient interactions between imprinted material and template, several aspects such as functional and shape complementarities, as well as the contributions due to the surrounding solvents have to be taken into account [4–6]. Imprinted materials are highly stable and useful for the development of new analytical methods that also work under relatively harsh operating conditions [7]. Up to now the mechanisms of recognition and the rules of template-monomer interactions are not well understood. Anyway, is logically possible to foresee how the change of one factor (e.g., the monomer/template/cross-linker ratio, polymerization method and temperature, etc.) has an effect on the recognition performance of such systems.

In the last decades the imprinting technique was employed to build up molecularly imprinted membranes (MIMs) able to selectively recognize target molecules from solutions by simple static adsorption or permeation through membrane devices [8]. A membrane is defined as a selective barrier interposed between two adjacent phases that regulates the transport of chemical species between them. Membrane technology had started to be developed a long time before of the birth of the imprinting technique and was in use in some research areas. Today it is one of the most developed technologies and it is widely used in numerous fields. The idea to introduce specific recognition sites into artificial membranes provided an unexpected impulse to the research activity focused on the development of new bio-mimetic recognition devices exhibiting high stability, sensitivity and specificity. In addition, compared to the traditional applications of imprinted polymers, MIMs can operate in a continuous way exploiting the features of membrane and molecular imprinting technologies [9]. In this context, Székely and co-workers [10], developed molecularly imprinted organic solvent nanofiltration membranes using polybenzimidazole (PBI) as functional polymer. Owing to the excellent chemical and solvent stability of PBI, the prepared membranes exhibited both shape-specific adsorption and size exclusion properties.

Different strategies were exploited for the development of MIMs in both, flat-sheet and hollow fiber configurations. Flat-sheet membranes have a paper-like configuration. They are thin films with porous or dense structures. Hollow fiber membranes have a long spaghetti-like tube configuration. Their diameter varies over a wide range from to 50 to 3000 micrometers. Fibers can have a macroporopus structure with a dense selective layer on the outside surface. As examples, Wang et al. prepared flat-sheet MIMs for the specific recognition of curcumin using the photo-copolymerization method [11]. In another approach, the phase inversion technique was applied for the development of flat-sheet MIMs for the recognition of Rhodamine B [12] and uric acid [13]. Hollow fiber composite membranes were also prepared via thermal co-polymerization. The antibiotic levofloxacin was used as template [14].

Using the product of an enzyme reaction as template an imprinted membrane can be utilized for assisting in synthetic reactions by continuously removing the product from the bulk solution by complexation. In a similar manner, MIMs can be used for recovering chemical substances coming from fermentation broths such as antibiotics and organic acids, as well in chiral resolution of enantiomers, substances utilized in pharmaceutical field.

Recently, bio-sensing systems using MIMs have received much attention from scientists in various areas. Biosensors are tools of molecular dimensions that are engineered for sensing different analytes on the basis of bio-specific recognition enabling the detection and quantification in a single operation step. In a sensor a MIM is used as bio-mimetic recognition element which is integrated with a transducer component. The detection is accomplished on the base of specific characteristics of the target analyte (such as its IR spectrum/fluorescence/electrochemical activity) or modification of physico-chemical parameters of the system in response to the binding with the template. A chemical or physical signal is generated upon the binding of the template to the membrane, which is translated into a quantifiable output signal by transducer. The direct and rapid determination of an interaction between the recognition element and the template represents an encouraging factor for the development of such systems in alternative to the traditional methods of bio-assay. This is because they do not need the addition of secondary reagent and the separation of free and bound reactants [15]. Bio-mimetic sensors-based membranes have been developed for environmental, food, and clinical uses. This review deals with the different preparation methods of molecularly imprinted membranes. Referring to the last decades, the application of these membranes as bio-mimetic sensor devices will be also reported.

2. Molecularly Imprinted Polymers

Although the bio-molecules used as sensing elements in molecular recognition processes are highly specific and sensitive, they suffer the disadvantages of being fragile and expensive. In addition, they possess low density of recognition sites; their regeneration as well as their storage is not easy and restricted. The technique of molecular imprinting is an advantageous alternative to surmount problems linked with bio-molecules such as enzymes, antibodies, cells, animal or plant tissues. In fact, imprinted materials can be synthesized in a short time, are robust and highly thermostable. They can be also used in a wide range of solvents, pH and metal ion solutions with different ionic strength. Furthermore, they are easily regenerated and stored at room temperature without loss of efficiency. In this perspective, new polymeric receptors may be designed towards a wide variety of target analytes over traditional biological receptors [16–18]. The concept of specific recognition was pioneered in 1955 by Dickey who prepared “specific adsorbents” of silica gel by simply acidifying commercial silicate solutions containing methyl orange or one of its homologs [19,20]. However, the imprinting on inorganic silica was not quite popular due to the lack of repeatability in materials preparation and their low selective properties. In 1972, Wulff and Sarhan [21] and Klots and Takagishi [22], in parallel, reported the first examples of molecular imprinting of organic network polymers.

To produce molecularly imprinted polymers (MIPs) three essential elements are required: (1) the target molecule (or template) which is the substance that will be imprinted in the polymer; (2) the functional monomer which is a compound having chemical and shape complementarity to the template and will polymerize in order to form polymer matrix; (3) the cross-linking agent (or cross-linker) which is a multifunctional molecule containing two or more reactive split ends able to interact via chemical bound with specific functional groups present on other chemical substances (e.g., linear polymer chains).

The synthesis entails the following steps: (1) the template molecules form pre-polymerization complexes with polymerizable functional monomers capable of interacting with them. In this phase, the functional monomers arrange around the template in order to create the recognition sites. (2) The pre-formed complexes are polymerized with the cross-linker molecules into a rigid polymer to maintain the location of the functional groups for binding the template. (3) After the polymerization step, the template is removed from the polymeric matrix allowing to generate the specific recognition sites which possess the molecular memory of the template. In fact, they are capable to selectively rebind the template molecules with respect other compounds, including their structural analogues. In Scheme 1, a representation of the molecular imprinting process is reported.

The heart of the process is the template, which should be chemically inert under the polymerization conditions i.e., polymerization temperature, UV irradiation, etc. [6]. The polymerization is started by the addition of an initiator to the pre-polymerization mixture constituted of the three abovementioned elements. An appropriate solvent (porogen) is used to dissolve them. The nature and volume of solvent are important to control the morphology and the pore volume of the polymer [23,24]. In the high diluted reaction environment, the growing polymer is not able to live in the entire available volume and more small polymer particles/powders having higher surface area (and more accessible recognition sites) are obtained. The final structure and performance of the polymer depends also on other factors such as the selection of appropriate functional monomer and cross-linker. Their concentration is also critical for ensuring the interactions with the reactive groups of the template [6,25–34]. To exploit the complex formation and thus the imprinting effect it is extremely important to combine in a complementary way the functionality of the template with that of the functional monomer. As a rule, this latter is used in excess in relation to the total number of target molecules. In addition, if two or more functional monomers are simultaneously used in a co-polymerization is important to keep in mind their reactivity ratio in order to guarantee the feasibility of the synthesis.

In general, the functional monomer possesses two functional groups. At one end it forms non-covalent or reversible covalent bonds with the template. At the other end of the molecule there is a chemical group covalently interacting with the cross-linker or other monomers [16]. The presence of the cross-linker confers sufficient rigidity to the synthesized polymer, which is necessary for the formation and stabilization of the specific recognition sites. In fact, the cross-linker freezes the polymeric network in a suitable configuration to interact with the template. Furthermore, in addition to the solvent, the cross-linker is important in controlling the morphology of the polymer matrix (whether it is gel-type, macroporous or a microgel powder) [6]. To be able to generate materials with adequate mechanical stability, high cross-linker/functional monomer ratios are usually used. Yungerman and Srebnik [35] investigated the imprinting quality of cross-linked polymer networks using molecular dynamics simulations. The authors carried out a topological analysis of the imprinted network configuration before and after template removal. They concluded that the formation of distinct individual cavities retaining the size and shape of the template is enhanced by high degrees of cross-linking and low template concentrations. This is because of the reduction of the templates molecules aggregation in the pre-polymerization mixture and the deformation of the recognition sites (due to relaxation of the polymer network once the template has been removed). Polymers with cross-linker molar ratios in excess of 80% with respect to the monomer are usually used to obtain materials with adequate mechanical stability and good recognition performance [36]. However, an excessive cross-linker ratio will result in an extreme polymer rigidity, which adversely affects the interactions between the polymeric matrix and the template [26,37,38]. In addition, the number of recognition sites per unit mass of polymer are reduced [6,39]. On the other hand, using an insufficient cross-linker ratio, the polymer cannot maintain stable the configuration of the recognition sites because of the degree of low cross-linking [39].

Another important aspect to be taken into account to produce imprinted materials exhibiting good performance is the polymerization temperature. In fact, this parameter also affects the polymerization process (e.g., reaction completeness and reaction rate) and polymer structure (e.g., pore structure and swelling properties), which in turn influence the quality and quantity of recognition sites. Normally, lower temperatures will stabilize the template–functional monomers complexes. In 1988, Sellergren et al. demonstrated that a decrease of the polymerization temperature allowed one to increase the concentration of the pre-polymerization complex and consequently the number of recognition sites (creating during polymerization) compared with the same polymer synthesized at higher temperature [40]. Other synthesis, performed in a temperature range from 0 °C to 60 °C, using different initiators, demonstrated that polymers produced at low temperature exhibited better specificity with respect to the polymers synthesized at higher temperature [41,42]. On the other hand, a weak point of the polymerization performed at low temperature is the necessity of a longer time for the completion of the reaction [43,44]. In addition, when radical polymerization is used, another important parameter to be taken into account is the initiator system. In fact, when conventional initiators (including peroxides, azo compounds and thermal iniferters) are used a high polymerization temperature is typically required to guarantee the rapid decomposition of initiator avoiding the formation of toxic chemical side products [45]. In 2010 Cirillo et al. [46] performed polymerization processes at lower temperatures by using Fenton reagents as the redox initiator system. The advantage of this kind of initiator system were the low work temperature, the short reaction time (2 h for this synsthesis vs 24 h for the synthesis of conventional MIP synthesized by azo-initiators) and the absence of any kind of toxic reaction products [46]. In addition to those already mentioned, other interesting papers have discussed the influence of the polymerization conditions on the performance of MIPs [47–51].

Two main types of chemical interactions can be used in the preparation of molecularly imprinted materials: covalent and non-covalent binding [27,52,53]. The former, pioneered by Wulff [21,54–56], is based on a reversible covalent attachment strategy between the template and the functional monomer in the pre-polymerization mixture. After polymerization the bonds are chemically cleaved in order to set free the template and remove it. The interactions template/functional monomer takes place via the formation of covalent bonds also in the rebinding action. The second type of interaction, proposed by Mosbach and Sellergren [57–59], is based on non-covalent self-assembly (i.e., hydrogen, hydrophobic and Van der Waals interactions, metal chelation, etc.) of the template with functional monomers to position them in a precise spatial arrangement prior to polymerization. Subsequent free radical polymerization with the cross-linking agent allows one to obtain an imprinted polymeric network.

Like all the things, both synthetic strategies possess advantages and disadvantages. Owing to the possibility to control the stoichiometry of the imprint materials, the covalent approach allows obtaining more homogeneous recognition sites than the non-covalent one. Covalent-imprinted polymers are very stable and selective. To their disadvantage, they have a limited number of functional groups to interact with target molecules and exhibit slow binding kinetics. In addition, their repetitive use in cleaving and rebinding templates can be difficult, because of the restricted interactions. Finally, the covalent interaction is different from the natural recognition at molecular level that occurs in biological systems.

The non-covalent imprinting approach had the advantage that the monomer/template complex can be formed by self-assembly simply by mixing the interacting molecules, yet, the template is removed under mild conditions and a high number of recognition sites may be generated. A weak point of this method is that the template-functional monomer interactions are less strong than covalent bonds. In addition, in this approach, the interactions are governed by an equilibrium process and therefore the functional monomers are used in excess with respect to the template to move the equilibrium versus the formation of the template-monomer complexes. As a consequence, the excess of free monomer is randomly distributed in the polymeric matrix allowing one to also generate non-homogeneous and/or non-selective recognition sites [47,60,61].

Sometimes, polymers were synthesized by means a combination of covalent and non-covalent imprinting to surmount the problems related to the independent application of each method. In this case, the monomers and the print molecules, were covalently joined during the polymerization and the subsequent rebinding took place by non-covalent interactions. This approach is called the sacrificial spacer method and was introduced by Whitcombe [62].

At present, the most used technique to produce imprinted materials is non-covalent imprinting. This route offers some advantages over the covalent approach: it is easy dissolve monomer and template in a solvent and there exist a variety of monomers having chemical functionalities able to interact with any desired template molecule [29]. Besides, the removal of the template is easier with respect to the covalent binding approach. In fact, non-covalent interactions like hydrogen bonding are reversed by washing with methanol, methanol/acetic acid, or in aqueous solutions of an acid or a base. This wash allows removal of the template from the polymeric network after the polymerization process [28]. Finally, non-covalent interactions characterize the basis of the molecular recognition in living systems. In effect, like non-covalent molecularly imprinted materials, biological receptors such as antibodies and enzymes, recognize their target molecule via complementary ionic, hydrogen-bonding and hydrophobic interactions.

However, the removal of the template is a critical step in the preparation of imprinted polymers. In fact, like in the case of non-covalent binding, the interactions between polymer and template can sometimes make hard the removal of the last traces of template, even after washing the material many times. This is a problem which may leads to the bleeding of the residual template when the polymer is used [45,63]. In addition, it must be take into account that if there are residual template molecules in the imprinted matrix, less recognition sites will be available for rebinding, thus lowering the polymer efficiency [64]. Some researchers have investigated the possibility of achieving complete template removal. In 2001, Ellwanger et al. [65] employed different post-polymerization treatments (thermal annealing, Soxlet extraction, microwave-assisted extraction and supercritical fluid template desorption) aiming to a complete removal of clenbuterol and and L-phenylalanine anilide from their respective imprinted polymers. Referring to clenbuterol, the lowest bleeding level was observed after microwave-assisted extraction. In the case of the polymer imprinted with L-phenylalanine anilide, the lowest bleeding was found after extensive on-line washing in solvents containing acid or base additives and after thermal annealing treatment of the polymer. Later, Lulinski et al. [66] obtained a low level of template bleeding from a dopamine-imprinted polymer applying a Soxhlet extraction followed by a microwave-assisted extraction. More recently, other studies were performed to optimize the template removal from molecularly imprinted polymers [64,67].

To obtain high purification levels is also important to optimize the selective removal of the template fraction which is non-specifically bound to the imprinted matrix. An interesting study in this context was recently published by Székely et al. [68]. The authors combined the advantages of molecular imprinting and organic solvent nanofiltration for removal of 1,3-diisopropylurea (IPU, a potential genotoxic impurity) from mometasone furoate (an active pharmaceutical ingredient). In particular, a novel 1,3-diisopropylurea imprinted polymer was employed to remove residual IPU after nanofiltration of a mixture of IPU and API. The recovery of the API was achieved using two elution steps after the use of the imprinted polymer. The solvent methyl isobutyl ketone allowed recovery of the non-specifically bound API. Dichloromethane and methanol were used as secondary elution solvents to wash out IPU and regenerate the imprinted polymer [68].

The binding energies between the template molecule and the polymer/copolymer non-covalent interactions were studied by molecular dynamics [69–72] and quantum mechanical calculations, as the case of the frame of density functional theory (DFT) [73–75]. Recently, Prasad and Rai [76,77] used the second order Moller Plesset theory (MP2)/DFT to screen best functional monomers for specified template molecules. Theoretical results were in agreement with experimental data. In a different paper [78], a computational study was applied to investigate intermolecular interactions in the pre-polymerization mixture during the synthesis of an imprinted polymer towards 3,4-methylenedioxy-methamphetamine. The results of this study indicated that the best functional monomer and polymerization solvent for the preparation of the imprinted polymer were methacrylic acid and chloroform, respectively. Theoretical studies were also performed to investigate the selectivity of a cobalt imprinted polymer calculating the formation energies of the complexes between similar transition metal ions (Fe²⁺, Co²⁺, Cu²⁺ and Ni²⁺) and vinylbenzyl iminodiacetate as functional ligand [79].

Combinatorial chemistry was also adopted in order to accelerate the optimization of MIPs to attain the desired performance [80–82]. In combinatorial approaches the elements of the imprinting process (particularly the kind and molar ratio of the functional monomers) are varied using automated procedures [17]. In 2002, Chianella et al. [83] reported the synthesis of a computationally designed imprinted polymer for microcystin-LR. Later, a library of imprinted polymers was developed by Cederfur et al. that used the penicillin G as template molecule [84]. Assortment of the library was obtained by combining various functionalized monomers and cross-linkers and by changing the stoichiometry and the concentration of the components in the pre-polymerization mixtures. Recently, Khan and co-workers used a combinatorial screening procedure aiming the selection of polymer precursors in the preparation of the benzo[α]pyrene imprinted polymer [85].

A detailed description of the different polymerization methods is beyond the scope of this review. However, it is important to highlight that free radical polymerization is usually the preferred method for preparing molecularly imprinted polymers. Methacrylic acid (MAA) and 4-vinylpyridine (4-VPY) are the most widely used functional monomers. The acrylic monomer MAA interacts via ionic interactions with amines and via hydrogen bonds with amides, carbamates and carboxyl groups. Owing to their strong force, the ionic interactions allow to obtain better performance of the MIP materials. The vinyl monomer 4-VPY is able to interact with the carboxyl functionality of template molecules. Other used functional monomers are 2-vinylpyridine, itaconic acid, acrylic acid, methyl methacrylate, 4-ethylstyrene, 1-vinylimidazole, etc. Two or more functional monomers may be also used together to improve the interactions with the template and, therefore the selectivity in recognition process. Figure 1 shows the chemical structure of some functional monomers.

The most used cross-linker is ethylene glycol dimethacrylate (EDMA) which contains two vinyl groups. Other cross-linkers, containing three or four vinyl groups were also exploited. It was found that they permit to obtain polymer exhibiting better load capacity and selectivity than the polymer prepared using EDMA [6,17,61]. Figure 2 shows the chemical structure of some cross-linkers used in MIP synthesis.

The morphology of imprinted polymers is very sensitive also to small alterations in the polymerization conditions. A useful method which permits control of the polymer's morphology is the “hierarchical imprinting approach” [86,87]. The target analyte is immobilized on the surface of a porous solid which acts as molds to create a desired porosity. Polymerization in the pores of the support and its subsequent removal allows to the creation of a porous solid containing surface-confined recognition sites specific for the template. This approach was used to imprint small molecules [86,88] and peptides [89,90].

Imprinted polymers are today well recognized in many areas which exploit their molecular selectivity. As an example, MIPs were applied in chromatography and chiral separation [91–93], in solid phase extraction [94–101], in enzyme catalysis [102], as antibodies and receptors mimics [103–105], drug delivery [106,107] and for macromolecules (e.g., proteins) recognition [108–111].

Even though the development of imprinted polymers has assumed great importance in all these fields, they have restricted processability due to the high cross-linking level (necessary to preserve the specificity) which renders them very hard and fragile materials. Anyway, owing to the integration of imprinting and membrane technologies, flexibility and specificity are achieved in a development of a special MIP format i.e., “Molecularly Imprinted Membranes” which are stable and robust materials.

3. Molecularly Imprinted Membranes

A membrane is defined as a selective barrier interposed between two neighboring phases and regulates the transport chemical species amongst the two phases. The development of membrane technology dates back to many years before the coming of the molecular imprinting technique and was applied in many research sectors. For instance, membranes were used in water treatment [112], enzymatic catalysis [113], controlled drug release [114], oil refinement [115] and gas separations [116], for the development of biosensors [117] and so much more [118–120].

Taking advantage of the “bio-inspired” molecular imprinting technique, numerous research efforts were devoted to the production of a new generation of highly selective membranes named “Molecularly Imprinted Membranes”. Owing to the introduction of specific molecular recognition sites in its matrix, an imprinted membrane is able to discern between template and other analytes. These novel systems made valuable contributions in the development of innovative bio-mimetic molecular recognition devices. In fact, MIMs marked a new path for the detection, transport or retention of targeted chemical and biological compounds. Membrane-based imprinting processes do not require additives and can be performed at low temperature, thus reducing the energy consumption costs. In addition, compared to the traditional applications of imprinted polymers, MIMs can operate in a continuous mode by exploiting the characteristics of membrane and molecular imprinting technologies [11]. In comparison with a traditional membrane, a MIM exhibits an improved specific selectivity maintaining at the same time the separation efficiency.

Two diverse transport mechanisms of target molecules can be distinguished in a molecularly imprinted membrane [121]: the “retarded permeation” and the “facilitated permeation”. In the first case, the transport of the template through the membrane is retarded owing to the binding affinity with the imprinted sites distributed on the surface and bulk structure of the membrane. These kind of MIMs are generally macroporous and used as “adsorber systems” towards the analyte of interest allowing its separation from a mixture or microenvironment also containing other compounds such as contaminants, structural homologues, etc. The separation efficiency is strictly related with the binding capacity exhibited by the recognition sites. One example of imprinting of these systems is the development of imprinted nylon membranes for the selective adsorption of L-phenylalanine from L/D mixtures of this amino acid [122]. Yet, Malaisamy and Ulbricht prepared imprinted membranes for the selective adsorption of the biomarker Rhodamine B [12].

In the second type of transport (typical of micro-porous membranes), the passage of the template through the membrane is quicker and its perm-selective separation is achieved. This is due to the presence of a preferential path for the target molecules, which is produced via binding to and dissociating from neighbored recognition sites in the membrane facilitating their permeation. On the opposite, other solutes are subjected to the slow non-specific transport. An example of this transport mechanism was given from Chen et al. that developed an imprinted membrane exhibiting perm-selective properties toward the protein lysozyme with respect to bovine hemoglobin and cytocrom c [123]. Another example reports the preparation of luteolin-imprinted membranes, which exhibited very high selective transport of the template with respect to the similar rutin [124]. The selectivity factor α was 14.12.

This ability of MIMs to produce a selective transport or retention of specific molecules makes them good candidates for the development of highly innovative membrane processes. A wide variety of MIMs was successfully prepared exploiting the main following approaches: (1) the contemporary production of recognition sites and membrane structure of a self-supported membrane; (2) the synthesis of a tailored imprinted polymer to use in successive membrane preparation step; (3) the preparation of a composite imprinted membrane (by surface imprinting, thermal/photo-copolymerization of a pre-existing membrane).

The contemporary formation of membrane structure and molecular recognition sites is mainly accomplished by means of the “in situ cross-linking polymerization” and the so-called “alternative molecular imprinting”. The first method allows the formation of a cross-linked polymeric network produced by thermal or UV initiated polymerization of a mixture solution of template, functional monomer, cross-linker and initiator in a suitable solvent. Plasticizer agents are also added to the pre-polymerization mixture to obtain membranes that are more flexible. In 1990, Piletsky et al. [125] prepared the first flat-sheet imprinted membrane via in situ bulk cross-linking polymerization of acrylate monomers. In diffusion experiments, the membrane exhibited selective transport of adenosine monophosphate, a nucleotide that was chosen as the template.

Some years later Krotz and Shea prepared similar membranes by thermally initiated in situ cross-linking copolymerization using other nucleotides and nucleosides [126]. These membranes were freestanding, but fragile. Other authors attained an important improvement in terms of fragility and mechanical stability of freestanding membranes by adding to the “in situ imprinting” polymerization mixture an oligo-urethane-acrylate macro-monomer [127]. Enhanced permeability properties were also obtained simply by adding to the system a macromolecular pore former [128]. Imprinted membranes possessing enantio-selective recognition properties towards different amino acids [117,129] and pesticides [130] were also produced by using the same in situ polymerization method.

The alternative molecular imprinting is an expansion of a bio-imprinting in which existing recognition sites in an enzyme may be modified through the presence of a target molecule. According to this approach, candidate materials not having molecular recognition sites are transformed in molecular recognition membranes. The molecular memory of the template is induced in the membrane at the same time as the membrane is forming from its polymer solution. The process is mostly used for the preparation of membranes having flat-sheet configuration. The membrane is prepared in the presence of target molecules via the phase inversion technique. Briefly, a specific synthesized polymer (without cross-linker), which contains functional groups complementary to the template is used as membrane forming material. The process consists in the transformation of the polymer from the liquid state to a solid phase either by the “dry” or by the “wet” method. In the first method, the polymer solution is cast on a glass or Teflon plate and the polymer solidification is accomplished by the simple evaporation of the solvent. In the “wet” method, the film cast solution is immersed in a non-solvent and polymer solidification is attained by a precipitation induced through contact with the non-solvent [119,131]. Sometimes, a combination of the two methods is also carried out. Scheme 2 shows the preparation of a flat-sheet membrane via the “wet” phase inversion.

Normally, the “dry” phase inversion method allows obtaining membranes exhibiting a dense structure due to the progressive increase of the polymer concentration in the forming membrane owing to the solvent evaporation. Conversely, the “wet” phase inversion method leads to the formation of a porous membrane structure due to the quick liquid-liquid de-mixing typical of the non-solvent induced phase separation. Membranes can be prepared in a flat-sheet or hollow fibre configuration. Figure 3 shows the cross section of flat membranes prepared using the two different methods.

The “dry” phase inversion, was pioneered by Yoshikawa et al. that reported the preparation of molecularly imprinted membranes bearing oligopeptide derivatives, derivatives of natural polymers and other synthetic polymers as molecular recognition elements. They were applied in enantiomeric separation of amino acids and selective transport of nucleic acid components [132–141].

The “wet” phase inversion method was developed by Kobayashi et al. who prepared MIMs towards molecules of pharmaceutical interest such as the anti-asthmatic agent theophylline from blends of acrylate copolymers. The authors studied the influence of the coagulation temperature on the efficiency of the formation of the template-membrane's functional groups during the phase inversion process and on membrane selectivity [142–145].

Many researchers have successfully developed MIMs using the phase inversion imprinting method introduced by Kobayashi, which is now one of the most employed in this field. As well as acrylic copolymers [146,147] different polymers were used as membrane forming material, such as polyamide, polystyrene and polysulphone [148–150], cellulose acetate/sulphonated polysulphone blends [12], polyviniylidene fluoride and polyethersuphone [151], chitosan [152].

Most common prepared MIMs have flat-sheet configurations and are used for recognition of a wide range of compounds like pesticides [153], flavonoids [48,146,154], biomolecules like uric acid [13], vitamins [155] optical isomers [156–159] drugs [160] proteins [161] and much other. For example, Trotta et al. developed MIMs for the recognition of the antibiotic tetracycline hydrochloride. Membranes were able to selectively bound the template with respect of the structural analogue choramphenicol [162]. Fan et al. produced an imprinted membrane specific for the chemotherapeutic agent trimetophrim by blending polysulfone with a polymer imprinted with the target molecule [163]. Wang et al. prepared a metal ion imprinted membrane for the recognition and recovery of silver from waste solutions [164]. UI-Haq used the wet-phase inversion to prepared D-phenylalanine (D-Phe)- and L-phenylalanine (L-Phe)-imprinted membranes, which had a nanoporous structure. Applying these membranes for chiral resolution of phenylalanine a selective solute rejection was observed for the first time [165]. Wang et al. [166] developed MIMs with efficient perm-selective binding towards the template uracil.

The phase inversion process was also employed to introduce both size exclusion properties and molecular recognition sites during membrane forming step of molecularly imprinted nanofiltration membranes [10]. During nanofiltration in organic solvent, these membranes exhibited good recognition capacity and rejection performance.

Due to its feasibility the phase inversion technique was also exploited for the development of membranes by means of the so-called “hybrid molecular imprinting”. This strategy entails the dispersion of a pre-synthesized cross-linked imprinted polymer (i.e., powders or micro/nano particles) into a normally used polymer matrix and the subsequent membrane formation via phase inversion. By using this method, Borrelli et al. incorporated imprinted polymer particles in a polymeric matrix. The obtained hybrid membrane was used for the selective depletion of riboflavin from beer [161]. Recently, Donato et al. developed hybrid imprinted membranes for the genotoxin 4,4′-methylene-dianiline [167]. Membranes were prepared by dispersing a pre-synthesized poly(acylonitrile-co-acrylic acid) imprinted polymer into polyvinylidene fluoride matrix. EDMA was used as cross-linker during polymer synthesis. The authors demonstrated an increase of the membrane performance with respect to previously developed non-hybrid membranes prepared directly with the imprinted polymer [168]. These membranes exhibited recognition properties in studies performed in organic environment. Wu et al. developed hybrid membranes for the separation of phenylalanine isomers incorporating an imprinted inorganic powder into a sodium alginate matrix [169]. To enhance the mechanical stability of sodium alginate and the compatibility between the two phases 3-aminopropyltriethoxysilane was used as precursor and cross-linking agent. Other publications deal with the preparation of hybrid imprinted membranes for the recognition in aqueous medium [170–180].

Imprinted microsphere and/or nano-particles may be also used for the preparation of composite MIMs by covering the surface of a pre-existing membrane via simple deposition or by means cross-flow filtration. Lehmann et al. [181] used imprinted microspheres as a filter cake between two polyamide microfiltration membranes, thus forming an enantio-selective multilayered composite membrane. Another example of membranes loaded with imprinted particles was reported by Silvestri et al. [182] that developed poly(acrylic acid-co-methylmethacrylate) membranes superficially modified by deposition of poly(methylmethacrylate-co-methacrylic acid)-based nanospheres imprinted with theophylline and caffeine. In a simultaneous application of the electrospray deposition and the phase inversion techniques, molecularly imprinted nanofiber membranes were also prepared for application in enantiomeric separation [183,184], selective separation of copper ions [185] and the alkaloid (−)-cinchonidine [186].

Composite MIMs were also prepared by other surface imprinting strategies, which entail the functionalization of the surface of a support membrane with a thin layer of molecularly imprinted polymer. In the presence of a template molecule, binding sites are introduced into the membrane support without damaging its pore structure. This method has the advantage to combine the mechanical integrity of the membrane support with the selectivity of the imprinted polymer. Two different polymerization routes may be followed: the surface grafting via photo-polymerization and the surface coating via thermal polymerization. The Scheme 3 shows the two different polymerization methods.

The membrane preparation process is accomplished by the immersion of a membrane to be modified in the pre-polymerization mixture and the subsequent polymerization. Depending of the polymerization method, photo or redox initiators are used to activate membrane surface by means of the formation of radical sites. Many flat sheet and hollow fibres membranes were developed by applying this method. First imprinted composite membrane was prepared by Wang et al. in 1997 [187] that used as support matrix a poly(acrylonitrile) membrane containing a photosensitive dithiocarbamate group. UV initiated graft copolymerization of acrylic acid and N,N methylene bis-acrylamide in the presence of theophylline as template allowed the formation of a molecularly imprinted layer on the membrane surface. Since then, many researchers prepared composite MIMs exhibiting high recognition performance. As example, Piletsky et al. [188] photographed flat-sheet polypropylene microfiltration membranes which showed recognition properties towards the herbicide desmetryn in water. Other membranes exhibiting selective permeation towards the template 4-amino-pyridine were prepared using the same method [189]. Recently, composite MIMs were produced by Liu et al. [190] for the selective recognition of the herbicide nicosulfuron for food safety detection. Koster and co-workers [191] coated silica fibers with a thin layer of methacrylate polymer for solid-phase microextraction. Donato et al. [52] prepared flat-sheet composite MIM by photo-copolymerization of polypropylene support with the functional monomer 4-vinylpiridine. This membrane is the first example of composite MIM which exhibited enantioselective permeation properties toward the anti-inflammatory drug S-naproxen. Composite membranes imprinted with S-naproxen were more recently developed via thermal copolymerization of the same functional monomer and the surface of hollow fibre polyvinylidene fluoride [192]. The separation properties of these novel imprinted membranes were determined by on-line HPLC membrane technology. Also in this case a selective permeation of the target molecules was observed. S-Propranolol-imprinted composite cellulose membranes were also prepared and employed as enantioselective-controlled release systems [193]. Polyvinylidene fluoride hollow fiber ultrafiltration membranes were also used as support for the deposition of a thin imprinted layer of poly(methacrylic acid). Applying the “template analogue strategy” composite MIMs were developed for the recognition of lovastatin acid in aqueous medium [194]. The process consists in the use of analogues complementary substructures of the template as “dummy template” exploiting the cross-reactivity of imprinted materials. This strategy avoids the problem of template leakage during its extraction in the imprinting process and/or template leakage in solid phase extraction [195]. It is also possible use this approach when target analytes cannot be directly used as template due to their toxicity or high costs and low availability. Zhai et al. [196] used the thermal polymerization method to prepare metal ion-composite imprinted membranes using as support material polyvinylidene fluoride. More recently, Zhu and co-workers used the same polymerization method to prepare composite MIMs able to selectively recognize magnolol, which is one of the most popular traditional Chinese medicines [197].

MIMs have been also produced by polymerizing thin imprinted polymer films on inorganic microporous glass-fiber membranes and Teflon filters [198] or on multiwalled carbon nanotubes [199]. The development of imprinted ultrathin titania gel films via surface sol–gel process was well reviewed by Kunitake and Lee in 2004 [200].

Some problems related to the preparation of MIMs are the limited accessibility of recognition sites owing to random distribution in the bulk and on the surface of the polymeric material. In addition, it is difficult to combine high yield imprinted sites with pore structure appropriate for efficient membrane permeability and separation.

In this section only some selected examples concerning the production of molecularly imprinted membranes are reported. Nevertheless, they symbolize the wide range of their application. In the next sections more attention will be given to the development of MIMs as molecular recognition elements for application in biosensors technology. For other information on molecular imprinting technology and applications, the reader is directed to some relevant publications [7,103,121,131,147,201–203].

4. Biosensors

In the last two decades, biosensors have become very important tools for the detection of chemical and biological compounds for clinical [204,205] environmental [206,207] and food [208] monitoring. The wide range of applications in different fields of these devices is due to their excellent high specificity, sensitivity and rapid response. Biosensors comprise a biological recognition element and a transducer to convert the biological response into a measurable signal proportional to the analyte concentration [209,210]. In Scheme 4, a representation of a biosensor is shown. The biological recognition elements are classified in biocatatalytic and affinity [211]. The former type, (including enzymes, microbes, plant and animal cells and plant or animal tissue) provides a binding between the analyte and the biological species resulting in a change detected by the transducer. The latter type (including antibodies, receptors, nucleic acids and molecularly imprinted polymers) offers selective interactions of the sample with a specific ligand to form a thermodynamically stable complex [211,212].

The development of the biosensors depends on the immobilization methods of the biological macromolecules for solving several problems such as their loss, preservation of their stability and shelf life [213,214]. To reach these objectives, different immobilization techniques were investigated such as physical entrapment, adsorption, covalent binding and covalent cross-linking [215]. It is necessary to mention that the selected immobilization method depends on the nature of the biological element, type of the transducer and operating conditions of the biosensors [216]. Physical entrapment is the oldest and a very simple method, but with different drawbacks as the loss of the biological element as the pH varies [217]. The most used methods are those that involve the formation of a chemical bond. In particular, the bond is formed between the functional group of the biological species and the reactive group of the support. The biosensors can be also classified in different types in relation to the transducers: electrochemical (potentiometric, amperometric, impedimentric, conductimetric), calorimetric, piezoelectric and optical transducers. Electrochemical approaches are extensively used in the development of these devices. Potentiometric transducers measure the potential difference between a working electrode and a reference electrode, and the signal is correlated to the concentration of analyte [218]. The sensitivity and selectivity of these biosensors are excellent. However, a reference electrode very stable and accurate is required and consequently it could represent a limitation. The amperometric biosensor-type measures the current produced by the reduction or oxidation of an electroactive species present on the surface of a working electrode [219]. They are accurate and precise. Nevertheless, they are less sensitive because the measured current is only governed by the redox potential of the electroactive species so the contributions of other chemical species can be included in the measurement result [220]. Conductimetric sensors work based on the change of conductivity associated with many reactions that imply a change in ionic species [221]. They are less used because are not specific and at the same time present a very low signal/noise ratio [220]. Recent studies on biosensors have demonstrated that electrochemical impedance spectroscopy (EIS) detection is much preferred to other electrochemical methods including amperometric and potentiometric ones [222,223]. EIS is a characterization technique that offers electrical information in the frequency domain [224,225]. With this technique, a process takes place in an electrochemical cell and it can be modelled using a combination of resistors and capacitors [226]. However, a major disadvantage of EIS methods is their low sensitive detection limit compared to the other technologies for the lack of a signal amplification method adaptable to the EIS [227]. Biosensors with thermal transducers exploit the change of heat that occurs during chemical reactions [228]. Anyway, the heat is not confined in an adiabatic system and this determines a loss of information due to any partially loss of the heat by irradiation, conduction or convection [228]. Piezoelectric biosensors have a biological recognition agent immobilized on a surface of a piezoelectric element (for example a quartz crystal) [229]. The presence of the analyte can be detected by immersing the piezoelectric crystal in the reaction solution to allow a bond between the target species and the immobilized biological element [230]. This determines a modification of the crystal mass with a consequent change in the vibrational frequency [231]. The drawbacks of a piezoelectric sensor are the lack of an exact correlation between mass addition and frequency change for solution phase and sensitivity to environmental conditions. Is important to note as optical biosensors are receiving much attention due to the developments in the optical fiber technology. The variation of the optical properties such as UV-Vis absorption, bio-and chemo-luminescence, reflectance and fluorescence is caused by the interaction of the analyte with the biological element [231]. They are compact, flexible and present a small probe size.

Following are reported some recent and interesting examples of biosensors enzyme-based for the detection of pesticides. Acetylcholinesterase (AChE) is an enzyme that stabilizes the levels of the neurotransmitter acetylcholine for the catalytic hydrolysis of acetylcholine to thiocholine [232]. The AChE catalytic activity is inhibited by trace amounts of organophosphorus pesticides (OPs). Biosensors based on AChE are applied for the detection of pollutants such as OPs and carbamates. Recently, Guan et al. [233] developed a novel (AChE) biosensor based on multilayer films containing chitosan and AChE liposome bioreactor. The biosensor was not expensive; it showed high sensitivity, and good reproducibility resulting capable to detect pesticides.

A very large number of enzyme biosensors is studied for the detection of pesticides. In particular, as it is shown in Figure 4, about 70% of these devices is employed for the determination of pesticides, the 20% is applied in the determination of heavy metals and the rest (10%) for other chemicals (benzoic acid, nitric oxide, cyanide and so on) [214].

Creatinine is an important clinical analyte for the diagnosis of renal and muscular dysfunction [234]. Creatinine is detected by using both the colorimetric method by means of the Jaffe reaction [235] and enzymatic colorimetric methods [236]. However, the two different methods present different drawbacks [237]. The biosensing method presents different advantages with respect to the traditional techniques in terms of time reduction, simplicity, sensitivity and analysis costs. For example, Yadav and co-workers [238] demonstrated the possibility of using iron oxide nanoparticles/chitosan graft-polyaniline (Fe₃O₄-NPs/CHIT-g-PANI) composite film for construction of an amperometric creatinine biosensor. The biosensor showed rapid response, higher sensitivity, good reproducibility and long-term stability. The very expensive and time-consuming enzyme purification process hinders the application of the biosensors-enzyme based at large scale. Microbes (algae, bacteria, and yeast and so on) offer an alternative in the manufacturing of biosensors because they can be massively produced by cell culturing. They present other advantages as the ability to identify a large variety of analytes and the possibility to work at wide range of temperature and pH [239]. Singh and Mittal [208] demonstrated as a biosensor obtained by immobilizing Chlorella sp. microbes over a glassy carbon electrode are highly selective for mercury without interference from alkali, alkaline earth and other transition metal ions. The experimental results showed high reproducibility with a lifetime of 14 days. An amperometric biosensor based on the immobilization of a microbe (Pseudomonas alcaligenes MTCC 5264) on a cellophane membrane detected caffeine in solution over a concentration range of 0.1 to 1 mg·mL⁻¹ in a very short time (3 min) at pH of 6.8 and temperature of 30 °C [240]. However, in this study sugars like glucose and sucrose gave interference in analysis of caffeine. An atrazine imprinted polymer was prepared by Lavignac et al. using a non-covalent approach [98]. More recently, Yaqub et al. [241] developed artificial receptors via in situ MIP synthesis directly on gold electrodes of piezoelectric transducers using as template the same pesticide.

The principal drawbacks present in the biosensors field are the chemical and physical properties of the biological recognition elements being unstable when are not in their natural environment and this results in difficulties to fine-tune them for a specific application [242]. In this context, molecular imprinted polymers, which were meticulously discussed in Section 2, can represent valid substitutes for the biological agents owing to their peculiar characteristics [243]. MIPs are combined with different type of transducers (optical electrochemical, acoustic, etc.). Recently, a molecularly imprinted polymer was developed and integrated in an optical sensor for the detection of the antibiotic enrofloxacin [244]. The authors demonstrated the capability of the MIP to selectively recognize the target molecule. Gonzalo et al. [245] prepared an opto-sensor based on an imprinted polymer synthesized using toluene as template. The sensor was used for the screening of a mixture of toluene, ethylbenzene and xylene in drinking water. During the test the contaminated samples were identified quickly (81 s) with a cut-off level of 700 μg·L⁻¹ for ethylbenzene, which is the maximum level established by US Environmental Protection Agency [245]. In recent times, the possibility to determine 2,4-dinitrophenol in water samples by using an electrochemical sensor based on an hydrophilic imprinted polymer was demonstrated [246]. The novelty of the work is represented by the presence of a hydrophilic imprinted polymer that could enhance the accessibility of the target species to the imprinted cavities and thus improve the selectivity of the MIPs in a water medium.

5. MIMs-Based Bio-Mimetic Sensors

Biosensor technology is an area characterized by novel testing approaches. Since the first biosensor was developed by Updike and Hicks in 1967 [247] many biosensors have been studied and developed. Classical molecular biorecognition materials are based on the use of biological molecules such as antibodies, enzymes, microorganisms, as recognition elements. Despite their high selectivity, biosensors represent today only a small proportion of the market due to problems directly related with the chemical and physical properties of the biomolecules employed. In fact, these biological receptors possess high selectivity but owing to their poor stability, it is difficult to incorporate them into devices for developing biosensors. In addition, they suffer of availability, environmental intolerance and reproducibility. In the last decades, biosensing systems using molecularly imprinted materials as recognition elements instead of biomolecules have received attention of scientists in various fields due to their potential facility to overcome some of these problems. Traditional applications of imprinted materials in sensor technology involved the use of imprinted polymer particles. Nevertheless, the interface adhesion between the particles and the transducer surface can be poor. In addition, the response time is extremely long, owing to the slow mass transfer [9], so as an alternative to imprinted polymers, a special format of them like thin films or molecularly imprinted membranes are attracting growing interest.

In fact, the direct and rapid determination of an interaction between the recognition agent and the target analyte represents an encouraging factor for the development of MIMs as alternatives to the traditional bioassay methods. This is because they don't need the addition of secondary reagents and the separation of free and bound reactants [15]. In addition, MIMs-based sensors take advantage of their high selectivity and stability, which enable them to long-term operation under conditions not tolerated by biomolecules without losing sensitiveness. They can operate in acid/basic solutions, high temperature, organic environment, etc. Furthermore, MIMs can work in a continuous way and are therefore promising for an industrial application.

Hedborg et al. [248] that developed an L-phenylalanine-imprinted polymeric membrane in combination with a field-effect capacitance device performed first studies on this topic. The authors demonstrate that the change in capacitance of the system was similar for the template and its structural analogue tyrosinanilide, while the similar phenylalaninol determined only a small change of capacitance in the system. The recognized template was detected by means of selective binding. Alternatively, Piletsky et al. developed a sensor device based on the selective permeation of an analyte through an imprinted membrane, which was able to recognize low-weight organic molecules [249].

In some MIMs-based sensors, also developed by Piletsky et al., an opposite response between covalent and non-covalent bonding-based sensors was observed. In particular, an increase of template concentration caused a reduction in conductivity signal in covalently-imprinted device, while opposite behavior was observed for the non-covalent one. This different response was attributed to the presence of more numerous and homogeneous recognition sites in the covalent systems, which shrank more reducing the micropores when the template was added. Probably this phenomenon determined a decrease of the electroconductivity based on ion transfer [250].

Up to now, different sensor-based imprinted membranes were developed. The most common fields of their application are the detection of contaminants in water and food, the analytical sensing of bioactive molecules as well as metals detection. In the following sections, some relevant examples are discussed.

6. Conclusions

The production of materials exhibiting biomimetic properties was one of the most important challenges of many researchers during the last decades. A considerable boost in this direction was given to the development of the molecular imprinting technology. In fact, since their invention imprinted materials were employed as elements capable of imitating the recognition aptitude of biological systems towards specific ligands. Owing to their high sensitivity, selectivity and stability, imprinted materials were successfully employed in different areas like separation, assay and catalysis. In particular, the requirement of novel testing approaches in substitution of the classical bio-recognition elements has determined a rapid development of materials in biosensor technology. Classical biorecognition materials are based on the use of biological molecules such as antibodies, enzymes, microorganisms, as recognition elements. Although these biological receptors are highly specific and sensitive, they suffer the disadvantages of being fragile and expensive. In addition, they possess low density of recognition sites; their regeneration as well as their storage is not easy and restricted. In addition, they suffer from availability, environmental intolerance and reproducibility issues. Biosensing systems using molecularly imprinted materials as recognition elements instead of biomolecules offer the possibility of overcoming some of these problems owing to their potential facility. The direct and rapid determination of an interaction between the recognition element and the template represents an encouraging factor for the development of such systems. This is because they do not need the addition of secondary reagents and the separation of free and bound reactants.

Traditional application of imprinted materials in sensor technology involved the use of imprinted polymer particles. However, the interface adhesion between the particles and the transducer surface can be weak. In addition, the response time is extremely long, owing to the slow mass transfer. Therefore, as an alternative to imprinted polymers, special formats of them like thin films or molecularly imprinted membranes are also under development. They are usually prepared by UV irradiation, thermal polymerization and via phase inversion techniques. MIMs-based sensors take advantage of their high selectivity and stability, which enable their long-term operation under conditions not tolerated by biomolecules without losing sensitivity.

Up to now MIMs-based sensors were developed for sensing water contaminants, drugs, food additives and some molecules in biological fluids. Despite the great application potential of these materials in sensor technology, we are aware that many efforts still need to be made to produce recognition systems optimized for each analyte. As an example, to simplify the imprinting of larger biological molecules there is a need for materials with more flexibility and a higher degree of accessibility. A great challenge for research in this field is the production of membranes imprinted with virus, which could be used in diagnosis and therapeutic treatments. The future of research in this area will have a thriving development benefitting from the integration of different disciplines. A contribution in this direction is to be found in the application of molecular modeling and combinatorial chemistry, which offer the possibility of optimizing the choice of template-functional monomer couples and to be able to predict the actions of the imprinted material at a molecular level. The literature data encourage the authors to believe that the different contribution offered by scientists in this field will have real importance in health care, environmental protection and industrial development.