Over the past two decades, enormous activity has taken place in the field of sensor technology. Biosensors, in particular, have attracted considerable attention because of their extraordinary sensitivities and specificities. However, such devices often lack storage and operational stability because they are based on a fragile biological recognition element: an enzyme or antibody. For this reason, biosensors have not become quite the commercial success expected in the early euphoric development phase.
An emerging technology called molecular imprinting, however, could provide an alternative. This technique leads to highly stable synthetic polymers that possess selective molecular recognition properties because of recognition sites within the polymer matrix that are complementary to the analyte in the shape and positioning of functional groups. Some of these polymers have high selectivities and affinity constants, comparable with naturally occurring recognition systems such as monoclonal antibodies or receptors, which make them especially suitable as constituents in chemical (biomimetic) sensors for analytical chemistry.
Chemical recognition systems, however, have been developed mainly by rational design in the laboratory, although combinatorial approaches, which can generate recognition systems by selection from large libraries, will become increasingly important in the future. The principles of host-guest chemistry were set out by Cram, Lehn, and Pedersen in the 1960s and 1970s and have been developed for a multitude of synthetic or semisynthetic systems such as crown ethers, cyclodextrins, and cyclophanes (1-3). Such host-guest systems are potentially very useful as recognition elements in analytical applications and have been used in separation and isolation processes.
One of the most intriguing areas for host-guest chemistry is the development of biomimetic recognition systems. A wide range of analytical procedures depend on reliable and sensitive biological recognition elements such as antibodies and enzymes. Because such biomolecules can suffer from stability problems, synthetic counterparts are desirable. One such approach to biomimetic recognition is the fabrication of molecularly imprinted polymers (MIPs) (4).
Molecular imprinting is a powerful method for preparing synthetic recognition sites with predetermined selectivity for various substances. Although the concept of molecular imprinting has existed for many years (5), it is only recently - accelerated in part by a report on "plastic antibodies" (6) - that we have witnessed a surge in interest. Molecular imprinting can be approached in two ways: the self-assembly approach (7) and the preorganized approach (8) (Figure 1). These two approaches, which differ with respect to the interaction mechanism in prepolymerization, follow common molecular recognition terminology (1, 9).
The self-assembly molecular imprinting approach involves host-guest complexes produced from weak intermolecular interactions (such as ionic or hydrophobic interactions, hydrogen bonding, and metal coordinations) between the analyte and the monomer precursors. These self-assembled complexes are spontaneously established in the liquid phase and are then sterically fixed by polymerization with a high degree of crosslinking. After removal of the print molecules from the resulting macroporous matrix, vacant recognition sites that are specific to the print molecule are established. The shape of the sites, maintained by the polymer backbone and the arrangement of the functional groups in the recognition sites, results in affinity for the analyte.
Figure 2 shows an example of the polymerization chemistry used to prepare MIPs that are selective for dansyl-L-phenylalanine. A combination of carboxylate- and pyridinyl-containing monomers (methacrylic acid and vinylpyridine) prearranges with the amino acid derivative in acetonitrile solution before polymerization. Following crosslinking with ethylene glycol dimethacrylate by free-radical polymerization, a rigid polymer is produced that, after extensive washing to remove the print molecule, retains recognition sites specific for dansyl-L-phenylalanine.
Click here to see
Figure 2.The preorganized molecular imprinting approach involves formation of strong, reversible, covalent arrangements (e.g., boronate esters, imines, and ketals) of the monomers with the print molecules before polymerization. Thus, the print molecules need to be "derivatized" with the monomers before the actual imprinting is performed. After cleaving the covalent bonds that hold the print molecules to the macroporous polymer matrix, recognition sites complementary to the analyte remain in the polymer.
The recognition properties of MIPs are highly attractive. In several cases, the selectivities and affinities acquired from the molecular imprinting process are on par with natural binding entities, such as antibodies. In self-assembly imprinting protocols using only noncovalent interactions, which have been predominantly used to produce biomimetic matrices, the principal means for exerting specificity are ionic interactions and hydrogen bonding between the analytes and the polymer functional groups, much akin to natural systems. In addition to these polar interactions, the three-dimensional geometry of the sites contributes to the overall quality of the specificity. Until recently, hydrophobic interactions - of paramount importance in biological systems - have been used rather haphazardly. Although hydrophobic forces are potentially more difficult to master because they are less specific and have less direction, combinations of polar and hydrophobic interactions may, in the future, be used to generate strong and selective binding. In this way, even more improved biomimetic matrices may be produced.
So far, MIPs have been prepared with affinities for proteins, amino acid derivatives, sugars and their derivatives, vitamins, nucleotide bases, pesticides, and pharmaceuticals (e.g., theophylline, morphine, diazepam, naproxen, cortisol, and pentamidine) (8). The binding of some of these MIPs (also referred to as antibody binding mimics) has been comparable with the binding of some natural monoclonal antibodies (6, 10). One of the advantages of molecular imprinting is that imprints can be made of compounds against which it is difficult, if not impossible, to raise antibodies. The use of animals - often necessary with antibody production - is avoided, and the scale-up for bulk manufacture is easily done. Thus, benefits are reaped from practical, ethical, and economical points of view.
Other advantages of MIPs are their long-term stability and resistance to chemically harsh environments (11). So far, MIPs have been used primarily as stationary phases in HPLC. Recently, however, they have been used in TLC (12), CE (13), heterogeneous binding assays (6, 14), and in biomimetic affinity sensors. A challenge that has yet to be met is improving the rather slow binding kinetics that have been observed, particularly in sensor applications.
A chemical sensor selectively recognizes a target molecule in a complex matrix and generates an output signal using a transducer that correlates to the concentration of the analyte (Figure 3). The recognition element is responsible for the selective binding (and in some cases, conversion) of the analyte in a matrix containing both related and unrelated compounds. Upon binding, the transducer translates the chemical event into a quantifiable output signal. When the analyte interacts with the recognition element, a change in one or more physicochemical parameters associated with the interaction occurs. This change may produce ions, electrons, gases, heat, mass changes, or light, and the transducer converts these parameters into an electrical output signal that can be amplified, processed, and displayed in a suitable form. The successful performance of a chemical sensor depends on the appropriate choice of recognition element and transducer.
Sensor performance is characterized by selectivity, sensitivity, stability, and reusability. Selectivity, a measure of how well a chemical sensor discriminates between the analyte and compounds of similar, or different, chemical structure, is principally determined by the recognition component within the sensor device. In this context, recognition elements of biological origin, such as enzymes and antibodies, are especially promising.
Sensitivity is determined by the recognition element and the transducer. Depending on the S/N, additional amplification steps can enhance sensitivity and lower the detection limit of the analyte. Maintaining long-term stability, withstanding harsh chemical environments, and operating at high temperatures and/or pressures are severe challenges for sensors, particularly when the recognition element is of biological origin. This, in part, slowed down the application of biosensors.
We proposed a "real" biosensor based on an MIP in 1991 (19). This was followed by the first reported attempt to make a biomimetic sensor based on capacitance measurements on a field-effect transistor coated with a phenylalanine anilide-imprinted polymer (20). However, the results were qualitative. A subsequently described amperometric morphine sensor showed more quantitative results, enabling morphine detection in the concentration range of 0.1-10 µg/mL (11). It also showed long-term stability, resistance to harsh chemical environments, and the ability to be autoclaved - highly attractive features not normally associated with conventional biosensors.
In another approach based on conductometric measurements (21), a direct signal is obtained on binding (due to the increased local concentration) of the positively charged species to the negatively charged imprints placed on the conductometric transducer. The difference in signal between the sensor and a reference sensor correlated well with the concentration of the analyte. This type of sensor arrangement is useful only in well-defined (pure) matrices in which the interference caused by conductivity of the solution can be controlled. Atrazine sensing by MIP membrane permeation measurements has also been reported (22).
To date, the most convincing demonstration of the usefulness of a "real" biomimetic sensor based on molecular imprints is an optical-fiber-like device in which a fluorescent amino acid derivative (dansyl-L-phenylalanine) binds to the polymer particles, resulting in fluorescent signals that vary as a function of the concentration of the derivative (Figure 4) (23). Chiral selectivity was shown by using the corresponding D-enantiomer as a control.
A problem when making measurements with MIP-based biomimetic sensors is the long response time (15-60 min). This delay could be minimized by optimizing the kinetics and selectivity of polymers. It is believed that the use of highly rigid polymers favors selectivity (because of the higher in-out energy barriers to exchange the analyte) and increasing response time. Similarly, polymer porosity increases polymer-binding capacity and response time. Using smaller polymer particles or thin polymer layers should improve diffusion rates and thus the apparent binding kinetics, giving far lower response times. Alternatively, the initial binding (pre-equilibrium) rate could be used to determine analyte concentration.
Biosensors based on enzymes have, in some cases, been shown to be superior in sensitivity to affinity-based biosensors. This tendency is explained by analyte conversion, which occurs after the initial binding step in conjunction with turnover amplification and makes it possible to obtain highly sensitive amperometric transducers that are also less sensitive to unspecific binding interference. Biomimetic sensors containing catalytically active polymers should exhibit promising sensor characteristics, although so far only modest catalytic activity has been reported. Molecular imprinting of substances that resemble reaction transition states has led to polymers exhibiting some esterolytic activity (24); other examples of catalytic reactions include the ß-elimination of HF from 4-fluoro-4-(p-nitrophenyl)-2-butanone (25, 26) and aldol condensation (27).
Conducting polymers have been used as rudimentary selective partitioning phases on electrodes and have been shown to retain a "memory" for the anion that was used for doping. This ion-sieving effect has been studied amperometrically and potentiometrically and can be correlated to the ionic radius and the charge of the tested anion (28). Materials exhibiting predetermined molecular recognition selectivity in combination with electrical conductivity could be used in electrochemical sensors and would provide the basis for a new line of sensor development - introducing new fusion materials constituting an integrated recognition element and transducer.
Along these lines, the preparation and characterization of composite particles containing an electrically conducting polymer (polypyrrole) and an MIP for morphine have been reported (29). The morphine-specific molecular recognition properties were not significantly altered by the manufacturing procedure, which involved rather harsh treatment; and the composite particles were shown to be electrically conductive when examined as dry layers on an interdigitated finger array device. Such particles were immobilized by simultaneous electropolymerization of pyrrole onto gold-covered silica substrates and the topography of the substrates studied by atomic force microscopy. This demonstrates how readily the composite particles can be electrically connected to an electrode, thereby obtaining integration between the transducer and the recognition sites within the polymer.
The current generation of MIP biomimetic sensors is 100- to 1000-fold less sensitive than other types of biosensors. Although further improvements in MIPs are likely to decrease the sensitivity gap and lead to useful applications, biomimetic and other types of biosensors will most probably find their own niches in the future.
We are grateful to Richard Ansell and Peter Cormack for their linguistic advice.
ChemCenter
Pubs Div. Home Page
