Point-of-Care Diagnostics: Molecularly Imprinted Polymers and Nanomaterials for Enhanced Biosensor Selectivity and Transduction

The fundamental aspect of MIP operation is the creation of a selective pore for some analyte within a polymer’s matrix. Ideally, the MIP pore readily binds the analyte and retains it until transduction can take place. It is desirable if the MIP is easily refreshed or recycled for future use through template removal (95). In general, MIP pores are realized via polymerization of a functional monomer in the presence of the target analyte. After fabrication the analyte should be easily removed in a process like elution leaving a custom binding site for the intended application. The result is an epitope having a physical shape that corresponds to the analyte’s shape as well as chemical affinity for the analyte. (46)

MIP analyte binding can be covalent or noncovalent. Epitopes depending solely on covalent binding tend to be significantly more selective. Unfortunately, covalent epitopes are rigid in their production in that only a few suitable functional monomers are available. In noncovalent analyte recognition, a MIP and target can bind through π-π interactions, van der Waals forces, hydrogen bonding, and ionic interactions. (47)

A bulk MIP can be produced via free-radical polymerization and milled down to smaller sizes for simple fabrication. Unfortunately, the strategy results in a low selectivity MIP and excessive target analyte consumption. Noncovalent recognition is more accessible to most researchers due to facile execution. In addition, noncovalent binding and release tends to be faster, which is important in POCT. On the other hand, noncovalent recognition tends to result in less selective biosensors since the interactions are easier to disturb. (96)

Advanced MIP fabrication employs techniques like seed, suspension, precipitation, and emulsion polymerization (46). In particular, there is significant motivation to realize surface imprinting on some nanocarrier or substrate to maximize mass transfer (95, 97) (Fig. 2). This achieves effective diffusion of the analyte through the MIP matrix for rapid recognition and sensor refreshment. In addition, state-of-the-art MIP-based biosensors tend to realize a semicovalent strategy, or covalent templating followed by noncovalent sensor operation (96). Stimuli-responsive polymers are leveraged for the need for more functional monomers, protein recognition, and temperature responsive functionality (48, 52, 95, 98, 99, 100, 101, 102). The limited number of functional monomers is also addressed through computational prediction of monomers complementary to particular analytes and through novel chemical solutions, like repurposing silica as a monomer.

Figure 2

Before leaving the topic of MIP fabrication, it is instructive to highlight an unusual phenomenon in the literature. Efforts to realize a nanobiosensor with MIP binding are obligated to demonstrate analyte selectivity. This is often achieved by comparison of the device evaluation amongst samples containing the target analyte, chemical analogs, and potential interferents. Convincing data is characterized by high signals for only the target analyte compared to other sample additives. Another important control experiment evaluates the device with non-imprinted polymers (NIPs). The resulting data typically reveals high MIP signals and low NIP signals. This literature review only discovered one research group that templates NIPs against an analog molecule to the template (39, 103). Otherwise, the field is dominated by reports that assess their technique with an NIP having no template. This laxity results in MIPs and NIPs with significantly different porosity, specific surface area, and pore volume, thus detracting from a rigorous persuasion of selectivity (104).

MIPs are applied to a wide variety of biomedical applications. For example, graphene has been composited with MIPs to quantify dopamine (DA) in human serum and urine (105). Similarly, tryptophan has been measured in human serum with MIPs composited with carbon nanotubes (106). Lastly, cancer detection has been evaluated through a comparison of MIP-based optical and ECM biosensors (107). More examples are contained throughout this review.

MIP-based biosensors are also routinely developed for the assessment of environmental samples. River water samples were evaluated for the relative amount of the insecticide cyfluthrin by compositing MNPs with MIPs in an optical biosensor (108). Soil samples were tested for the herbicide diuron with a MIP-based HPLC strategy (59). Air streams have been monitored for the abundance of formaldehyde using a MIP-based gravimetric biosensor (109).

Similar to environmental monitoring, food safety is an important application for MIP-based biosensors. One such example involves furfural monitoring in alcoholic and caffeinated beverages by compositing nano silver with MIPs (110). Another example achieved the selective detection of the insecticide imidacloprid in vegetable samples (111). Lastly, an antibiotic has been quantified in samples of fish and pork by compositing MNPs with MIPs (112). While some of the case studies highlighted in this review deal with environmental and food safety, their strategies are amenable to biomedical analyte monitoring.

As a nanomaterial, silver exhibits a wide array of useful attributes that lends it to many diverse biomedical applications (117, 118, 119). In this section, a discussion of compositing AgNMs with MIPs is presented. AgNMs alone contribute several benefits to biosensor applications, such as its low cost, adaptable plasmonic properties, and high electrical conductivity. For the most part AgNMs do not present significant or unique challenges to biosensing. However, like most nanomaterials, some attention must be paid to effective dispersal as well as toxicity. While the biosensors reviewed here are not intended for internal use, the component materials may still find their way into the environment or come into contact with the user. In that case, nano silver’s environmental toxicity (120) and its potential to enter the body (121) cannot be discounted.

Silver nanoparticles (AgNPs) and MIPs have been combined in an optical biosensor to detect furfural, a carcinogenic and hepatotoxic flavor/aroma additive used in the food industry (110). Compared to a conventional surface plasmon resonance (SPR) biosensor, this work presents a facile strategy for quantifying analyte concentration changes. Researchers performed an analyte concentration study of intensity versus wavelength. A polychromatic light source was passed through a sample chamber containing linear AgNP-patterning achieved via controlled inkjet printing. For selective analyte binding a MIP layered onto the substrate within the sample chamber was prepared (Fig. 2d). The researchers report a LOD of 0.03 ppm for furfural in caffeinated and alcoholic samples. Although the results are encouraging, this report does not include a sensitivity range nor any selectivity control experiments such as assessment with NIPs or analogous compounds to furfural.

Another instance utilizing AgNPs takes advantage of their enhanced electrical conductivity to overcome a MIP’s insulating nature. The reported ECM sensor also makes use of MNPs and reduced GO resulting in a nanocomposite modification to the electrode (122). The MIPs were arranged in a thin layer (Fig. 2d). Compared to other ECM sensors tabulated by the authors their device exhibits a better than average linear detection range from 20 nM to 250 μM and an LOD equal to 13 nM. Where it surpasses the competition is the fact that its electrode is disposable and quickly renewed via external magnetic field application. The linear detection range is determined in phosphate buffer (PB) and the strategy is further assessed by quantifying quercetin in dietary supplement tablets.

A compelling application of AgNMs involves the novel construction of a paper sample collector for enhanced and specific imidacloprid monitoring, a neonicotinoid insecticide (111). The device’s chip is comprised of a paper substrate upon which dendritic silver was produced. These were layered with a molecularly imprinted membrane via electropolymerization (Fig. 2d). Additional AgNPs were decorated upon the molecularly imprinted membrane. This last step effected an increase in the signal. The authors reported a linear detection range from 0.200 to 800 ng mL^-1. The sensor is capable of quantifying imidacloprid in real samples prepared from vegetables and grains. It demonstrated stability by outputting consistent data from paper chips stored at varying temperatures. Selectivity was demonstrated in an assessment for the target pesticide when potentially interfering pesticides were present.

Like silver, AuNMs have long demonstrated their merit toward many biomedical applications (80, 123, 124). Regarding bio-sensors, AuNMs are valued for their capability to enhance ECM and optical transduction strategies. It follows that MIP-based biosensors also benefit from incorporating AuNMs (125). The major disadvantages associated with AuNMs, namely dispersibility and toxicity, are manageable. Many conventional biosensors employ AuNMs for antibody binding, taking advantage of thiol linkages (79, 126). Having eliminated those unreliable and expensive binding strategies, MIP biosensors often still make use of AuNMs. Here, several state-of-the-art biosensors reports are highlighted, which combine AuNMs for sensitivity enhancement with MIPs for selectivity.

First, researchers describe an ECM strategy leveraging AuNPs for their high electronic conductivity to increase the sensitivity of the biosensor (127). This research provides a rare report of salicylic acid (SA) detection in plant matter, which is critical for accurately determining the phenomena governing the defense, development, and growth of grains. The strategy boasts a sensitivity range from 0.0005 μmol L^-1 to 50 μmol L^-1. The strategy was demonstrated in real biological samples by quantifying SA added to wheat. MIPs generated more signal than the control NIPs. These MIPs were layered onto an electrode via precipitation polymerization in acetonitrile for selective SA binding (Fig. 2d).

In another recent compositing of AuNMs with MIPs, rhodamine 6G (R6G), a toxic fluorescent dye not yet banned from food stuffs in some countries, is quantified (128). An array of gold nanoparticles (AuNPs) achieves a homogeneous arrangement of “hot spots” resulting in increased surface enhanced Raman spectroscopy (SERS) signal reproducibility. A porous MIP achieves improved mass transfer enabling the target molecule to interact with the hot spots. Figure 2(d) depicts this MIP conformation. The sensor exhibits linear sensitivity between 10^-10 to 10^-4 mol L^-1 in deionized water. It was also selective for the analyte in orange juice. Moreover, it was effective for R6G sensing even in the presence of the following chemical analogs: rhodamine 123, crystal violet, and rhodamine B.

One final combined use of MIPs and AuNMs investigates the application of SERS sensing for protein disease markers like neuron-specific enolase (NSE) (129). This is a particularly impactful contribution to the field as it directly attacks the well-known problem of detecting proteins with MIPs (100, 130). MIPs were employed for the dual purposes of analyte isolation and labeling. This is achieved by preparing two specific MIPs for two corresponding epitopes on the same target protein. A layer of AuNPs and MIP for one epitope is oriented on a substrate for analyte capture. Then AgNPs, composited with MIP for a different epitope, are added to the sample for analyte labelling. The described configuration is comparable to Fig. 2c, where one MIPNP contains silver in its core and the other MIPNP is replaced with a MIP layered onto a gold substrate. This sandwich immunoassay strategy, along with the synergistic effects of Au and Ag, result in a highly selective and sensitive sensor. The strategy demonstrated a linear detection range in standard samples from 100 pg mL^-1 to 10 μg mL^-1. For comparison purposes enzyme-linked immunosorbent assay (ELISA) calibration was also determined with real human serum samples, but the conventional technique yielded a significantly narrower sensitivity range. ELISA took 6 times longer to complete, included more than double the number of steps, and required 20 times the sample volume. The report also included checks for selectivity by assessing analogs to NSE like bovine serum albumin (BSA), horseradish peroxidase, ribonuclease A, and ribonuclease B with no more than 7% cross-reactivity. Evaluation of NSE in healthy patient human serum and small cell lung cancer patient human serum demonstrates the strategy’s potential to real world applications.

CNMs are varied in their conformations, such as sheets of graphene oxide (GO), fullerenes, carbon nanotubes, and more. They manifest several beneficial properties making them desirable for diverse biomedical applications (131, 132, 133, 134). Aside from biosensors, recent reports indicate the potential for composites of GO with MIPs for drug delivery (135, 136). CNMs are employed to take advantage of electrical, mechanical, or optical properties.

An interesting work that combines the advantages of MIPs and GO succeeds in quantifying both dopamine (DA) and chlorpromazine (CPZ) at the same time (105). This is achieved through electropolymerization of MIPs in the presence of both analytes. The study achieves enhanced ECM signal via employment of QDs comprised of GO and doped with nitrogen. These optimize the electroactive region on the sensor’s surface and improve the rate of charge transfer. Two linear detection ranges for DA were found: the first from 0.05 to 8 μM and the second from 8 to 40 μM. For CPZ, the linear detection range was from 0.005 to 2 μM. These ranges are either more sensitive or wider than other recent and comparable efforts to quantify these analytes. Other efforts are not capable of assessing both DA and CPZ simultaneously. The report demonstrates amenability to real world applications by assessing both analytes in media like pharmaceutical samples, urine, and human serum. Evaluation of this ECM sensor proved selective when potentially interfering substances were present, like glucose, urea, creatine, and others.

Another innovative compositing of CNMs with MIPs is presented, wherein the researchers report the first chlorambucil detection via MIP binding (137). Insulating chlorambucil imprinted micelles were composited with water insoluble carbon 60 to realize ECM detection of the cancer fighting compound. The analyte’s linear detection range was between 1.47 and 247.20 ng mL^-1 for aqueous samples. The sensor was capable of reliably detecting the analyte in pharmaceutics, urine, and blood plasma. The strategy was reliable in the presence of interferents like other cancer fighting compounds (i.e. melphalan, 5-flurouracil, and temozolomide) as well as some biologically relevant compounds (i.e. DA, uric acid, and glucose).

Multi-walled carbon nanotubes (MWCNTs) have been layered onto an ECM sensor electrode with inexpensive, biocompatible chitosan MIPs for determination of tryptophan (106) (Fig. 2d). The MWCNTs nullify the MIP’s natural inhibition of the electrode’s conductivity. The strategy was capable of detecting tryptophan in PBS in three linear ranges: 2.0 nM to 0.2 μM, 0.2 to 10 μM, and 10 to 100 μM. Commonly encountered interferents like tyrosine, uric acid, DA, and ascorbic acid are inconsequential. A real sample of human serum was assessed with good results.

Of all the fields of study using MNPs, biomedical applications are the most prevalent and encouraging (138, 139, 140). Besides nanobiosensors, they have demonstrated an efficacy toward hyperthermia cancer treatment, imaging contrast, drug delivery, as well as cell, nucleic acid, or protein isolation (141, 142, 143). MNPs coupled with conventional binding strategies (e.g. antibodies) (144, 145), allow for various magnetic transduction strategies like giant magnetoresistance (GMR), Hall effect, superconducting quantum interference device (SQUID), etc. (29, 146, 147, 148, 149, 150). Unfortunately, no reports (Fig. 1) describe MIP binding coupled with magnetic transduction; none could be found at the time of writing and this omission is acknowledged elsewhere (43). While there exists a paucity of reports on true magnetic biosensors utilizing MIPs for analyte binding, there are many MIP reports that leverage MNPs for sample heterogenization, thereby enhancing analyte concentration (99, 151).

SERS biodetection is advanced through a nanocomposite comprised of an MNP core for sample purification as well as GO layer and AgNPs for signal enhancement (108). This nanocomposite was further functionalized with an MIP via precipitation polymerization of the functional monomer acrylamide within the shell of the construct (Fig. 2b). SERS enables non-destructive sample processing, favorable sensitivity, and rapid operation. The method detects the insecticide cyfluthrin in river water samples from 10^-8 to 10^-3 mol L^-1 with good linearity. The technique is specific only for the target analyte when compared against the chemically similar compounds bifenthrin and fenvalerate. The Raman signal intensity was more than two times stronger for the target analyte when compared to the analogs.

A separate report also uses MNPs and MIPs to detect cyfluthrin. Again, MNPs in this work only serve to realize analyte purification (152). MIP functionalized MNPs efficiently extract the pesticide cyfluthrin from honeysuckle samples. The conformation of this nanocomposite is similar to Fig. 2b. Rather than quantify cyfluthrin using a novel nanobiosensor, conventional high performance liquid chromatography (HPLC) is employed. Linear sensitivity is demonstrated from 30 to 3000 ng mL^-1. The researchers demonstrate their selectivity toward cyfluthrin through a binding investigation of MIP and NIP performance compared to other insecticides.

The next investigation realizes detection of the antibiotic chloramphenicol (CAP) in food with a sandwich assay and personal glucose meter (PGM) quantification (112). CAP detection was possible in the presence of interferents like tetracycline, streptomycin, and oxytetracycline. The strategy boasts of an LOD of 0.16 ng mL^-1 and a detection range from 0.5 ng mL-1 to 50 ng mL^-1. The combination of PGM and MIPs contribute to fast, portable, facile, inexpensive, reusable, and stable implementation. These benefits can be attributed to the PGM’s half century of optimization (153) and MIP substitution of antibodies as target analyte binding strategy. These MIPs were created from fragment imprinting where only part of the target is contained within the polymer leaving other parts exposed to other binding sites. AuNPs were used to adhere invertase to β-cyclodextrin, which was used due to its unique attributes. β-cyclodextrin couples with nitrobenzene through its hydrophobic cavity. EnVision polymeric chain was used to load many invertase tags. Hydrolysis of sucrose to glucose through invertase action amplified the signal for PGM readout. The m-MIP allowed for magnetic washing upon analyte binding. The analyte was also captured by the invertase entity for tagging. The described sandwich immunoassay is comparable to Fig. 2c, but with one NP replaced by the β-cyclodextrin-invertase tag.

Another report describes silica-coated MNPs for surface polymerization imprinting (Fig. 2b). Silica ameliorates agglomeration and paves the way for surface polymerization imprinting with methacrylic acid (MAA) as functional monomer (59). Diuron, a known water contaminant, spiked into water and methanol was assessed with HPLC. The method obviates conventional, and time-intensive, sample pre-treatment. This strategy resulted in a linear sensitivity for diuron from 0.1 mg L^-1 to 10.0 mg L^-1 and an LOD equal to 0.012 mg L^-1. The authors demonstrated selectivity for diuron by detecting the target in paddy water, paddy soil, and grain seedlings also spiked with chemically similar analogs (fluometuron, chlorotoluron, and isoproturon) in addition to several reference compounds (altrazine and terbutylazine).

Besides sample manipulation, some strategies take advantage of an increased electrical conductivity native to iron oxide MNPs. Also, MNPs can be produced and functionalized in a facile manner to be biocompatible and dispersible in any medium. However, MNPs have much more to offer. Since biological samples exhibit negligible magnetism, magnetic transduction with MNPs suffers from little background signal resulting in high sensitivity (154, 155). Therefore, it is expected that magnetic transduction researchers will soon realize the potential in abandoning conventional analyte binding (e.g. antibodies) in favor of a MIPs. Consider that an MNP’s magnetic susceptibility, or the extent of magnetization versus applied magnetic field, can be correlated with the amount of analyte bound to a surrounding MIP layer (142). In that case, magnetic transduction strategies relying on magnetic relaxation, like SQUID (156, 157) or Fluxgate biosensors (158, 159), are expected to perform well with MIPs. Similarly, MIP functionalized MNPs should be amenable to a variety of magnetoresistance transduction strategies like GMR (160, 161). Figure 3(a) is a depiction of a magnetoresistance transducer employing an MIP, for selectivity, layered on an MNP, for signal generation dependent on the amount of bound target analyte.

Figure 3

Quantum dots are also leveraged to advance fluorescence and luminescence devices (162, 163). The virtue of QDs has been demonstrated by their propensity to avoid photobleaching, size-dependent light emission, enhanced brightness, relatively large absorption coefficient, and tight emission spectrum. Conventionally, QDs were toxic detracting from their application in biomedicine. However, advances in fabrication have mitigated toxicity and improved solubility by realizing production in water. Furthermore, some of these can be made not to contain heavy metals.

An interesting incorporation of QDs comprised of zinc selenide, or ZnSeQDs, into an iron-based metal organic framework (MOF) was reported for the determination of squamous cell carcinoma antigen (SCCA) in human serum (164). The report details the realization of biosensor interface protection from the peroxydisulfate transducing species. The fabricated MIP thin layer is comprised of polydopamine for biocompatibility and amenability to tailored absorption. This is possible with the MOF acting as accelerator of the co-reaction with ZnSeQDs. These luminophores were selected for their excellent luminescence and relevance to similar analyte detection, namely carcinoembryonic antigen (CEA). Sensitivity in PBS from 0.0001 to 100 ng mL^-1 was demonstrated with an LOD equal to 31 fg mL^{- 1}. The electrode was recyclable up to five times and the electro-chemiluminescence (ECL) strategy was specific to SCCA when interferants like CEA, BSA, α-fetoprotein, and cancer antigen 153 (CA-153) were present. NIP formulations exhibited a negligible signal.

Mandani et al. report a state-of-the-art utilization of QDs for fluorescent probe labels in determination of methamphetamine (MA) (165). The report details the facile coating of eco-friendly carbon QDs (CQDs) onto a core-shell mesoporous silica nanoparticle (MSN) with MIP layer coinciding with CQD layer. The described configuration is like Fig. 2b, except that CQDs are in proximity to the MIP binding sites. This sol-gel fabrication, which realizes proximate positioning of fluorescent label with MIP recognition, allows for increased analyte binding and easier removal due to improved mass transfer. Linear detection ranges from 5.0 to 250 μM with an LOD equal to 1.6 μM as measured in PBS. The device was reusable up to five times without the need for refreshment. MA selectivity was demonstrated in the presence of interferants like ascorbic acid, uric acid, glycine, alanine, DA, caffeine, and morphine. Also, ions like K⁺, Ca²⁺, and Cl^- were allowed to interfere with negligible impact. Selectivity was further demonstrated by the fact that NIPs exhibited significantly lower fluorescence compared to MIPs. This strategy demonstrated its potential toward real world applications in biological samples such as human urine and blood plasma.

Many MIP-based nanobiosensors adopt silica as the functional monomer matrix for containing selective binding sites for the target analyte (166, 167). Silica readily adapts itself to this application as a result of its biocompatibility, facile functionalization through silanization, potential for periodic porosity, high surface area to volume ratio, tunable pore size, and chemical stability. These attributes also make SNMs suitable for drug delivery applications (168, 169, 170). The challenges in working with silica-based MIPs are in mitigating irregular porosity and short crosslink distances. These lead to insignificant swelling, meaning special care is required to precisely manipulate crosslinking density and related mass transfer.

The authors of a recent investigation report efforts to enhance electrical conductivity of MIP binding in an ECM serotonin biosensor (171). Nanocomposites prepared in this study were of the core-shell variety (Fig. 2b). The MIP shell was comprised of serotonin-specific binding sites embedded in a silica layer. Within the core, they employ magnetite and gold nanocomposites for improved electrical conductivity. Besides improving the rate of electron transfer, the magnetite MNPs enable magnetic washing. The ECM biosensor detects serotonin with an LOD of 0.002 μM and boasts a linear detection range from 0.01 to 1000 μM. Serotonin detection was achieved in solutions derived from either over-the-counter pharmaceutical products or human urine. Selectivity was assessed with NIPs, interferants, and structural analogs, like magnesium sulfate and creatinine. Another recent contribution to the use of silica as MIP for biosensor applications details their novel fabrication via microwave-enabled synthesis (172). This heating method, rather than hydrothermal synthesis, is credited with the rapid manifestation of monodisperse and controlled nanomaterial. The authors attribute this to the higher quantum yield, which relates to the strategy’s improved selectivity and sensitivity. The core-shell nanocomposite (Fig. 2b), employs a mesoporous silica layer for adherence to the QD core, inherent biocompatibility, and regular porosity. The cadmium telluride QDs enable optical fluorescence determination of tyrosine phosphopeptide (pTyr). This strategy demonstrates a linear detection range between 0.07 and 230 μM as well as an LOD equal to 34 nM. The potential application toward real sample evaluation was demonstrated in solutions comprised primarily of β-casein tryptic digest spiked with pTyr. Selectivity was addressed by investigation of several interferants including multiple ions, ascorbic acid, uric acid, fructose, glucose, various proteins, and some peptides chemically similar to pTyr. The device’s detection signal with NIPs was significantly diminished as compared with MIPs.

One final contribution to the adoption of mesoporous silica for molecularly imprinted materials utilizes thiol-ene click chemistry to realize a novel functional monomer (173). The one-pot formulation realizes custom binding sites on the favorable specific surface area of mesoporous silica. The strategy exhibits linear sensitivity in PBS and human serum for sparfloxacin from 0.05 to 2.0 μg mL^-1 and an LOD equal to 0.012 μg mL^-1. The mechanism of operation is optical fluorescence made possible with zinc sulfide QDs doped with manganese. Selectivity was demonstrated by comparing the performance of the strategy with the chemically similar compounds gatifloxacin, ciprofloxacin (CPX), clindamycin, and norfloxacin.

TNMs often serve as support upon which other nanomaterials are assembled. Additionally, TNMs possess distinctive properties making them appropriate for certain MIP biosensors strategies. They exhibit good catalytic activity and electron transfer, are biocompatible, have a large surface area to volume ratio, and are chemically stable. Also, TNMs are inexpensive and exhibit good optical properties. For these reasons, there are many relevant investigations into TNMs applied to biosensors using MIP binding, as discussed in this section (174, 175, 176). The most significant challenge associated with TNMs, unlike SNMs, is that they possess no intrinsic analyte binding affinity.

To begin, researchers describe a successful ECM biosensor for point-of-use testing of toltrazuril contamination in animal food products (177). The device benefited from an MIP-layered electrode (Fig. 2d), for selective template binding. The MIP reported here is synthesized from beta-cyclodextrin as monomer for host-guest interaction between the template and monomer, thus realizing non-covalent binding. TiO₂ enhances the conductivity of the platinum substrate for increased sensitivity. The researchers demonstrate linear detection from 0.43 to 42.54 μg L^-1 using differential pulse voltammetry more sensitively than both UPLC and LC-MS/MS. However, the sensor cannot detect the template at higher concentrations, meaning that conventional detection is not wholly replaced. Not only does the MIP exhibit obvious template affinity as compared to NIP performance, but the biosensor is not subject to interferants like glucose and ions like sodium or potassium.

A recent incorporation of TiO₂ nanoparticles and MIPs in an ECM biosensor realizes facile implementation of p-NP quantification in food samples (178). MIPs were layered on a glassy carbon electrode (Fig. 2d) for selective analyte binding. The strategy exhibited a linear sensitivity between 0.01 and 80 μmol L^-1, as well as an LOD equal to 3.91 nmol L^-1 as measured in spiked samples of isopropanol and PBS. Target selectivity was evident upon significantly reduced ECM signal generated from the chemically similar compounds 4-aminophenol, phenol, bisphenol A, 4-bromophenol, and 4-octylphenol. Application to real samples was demonstrated with solutions containing milk powder. The sensor was stable after as many as four weeks.

One unusual case study, departing from both ECM and optical biosensor transduction, makes use of TiO₂ nanoparticles for the enhancement of surface area. Investigators report a surface imprinting of polypyrrole onto colloidal red mud (179). Besides enhancing surface area, red mud, comprised of Al2O3, SiO₂, Fe₂O₃, and TiO₂, also leverages good adsorption and mechanical attributes, despite no innate recognition capability. The template was bilirubin, for which the sensor exhibited linear sensitivity between 0.01 and 120 μM and an LOD equal to 0.003 μM in PBS. This gravimetric biosensor employed a quartz crystal microbalance for mass determination, meaning the MIP was layered onto a substrate (Fig. 2d). The device exhibited reproducibility by way of about 3% performance loss after 20 uses and about 4% loss after 5 months post-fabrication. Selectivity was established with good serum sample performance and minimal response to the chemical analogs cholesterol, testosterone, and biliverdin as well as minimal response to the prepared NIP formulation.

While optical transduction strategies employing MIPs for nanobiosensors are not as prominent in the literature, they are a close second to ECM strategies. Of the reports reviewed here (Tables 3 and 4) nearly 31% of them are optical by design (Fig. 1d). These strategies can be further subcategorized into biosensors employing optical resonators, SPR, optical fibers, optical waveguides, photonic crystals, and others. The advantages associated with optical transduction are numerous (183, 184, 185). They can be miniaturized, are reliable in that their operation is not susceptible to the environment, and are renowned for their sensitivity. Optical biosensors are capable of facile sample handling like minimal pretreatment and the non-destructive nature of biosensor operation on the sample (27). When nanomaterials are incorporated into optical biosensor design the product is improved. Several researchers are reporting the use of nanoparticles tagged to bioanalyte molecules to improve sensitivity (Fig. 3b). However, this solution requires the binding of a target analyte to a label, which is generally undesirable since it complicates sample processing and detracts from POCT (186, 187).

One report details the employment of casein protein-facilitated emulsion synthesis of surface MIPs for bovine hemoglobin (BHb) biosensing (188). The authors reason that Pickering emulsion is wanted since it enables surface imprinting. By using the casein nanoparticle as stabilizer, the process is safer and less expensive by excluding traditional chemical surfactants. The authors address the difficulties with imprinting for relatively large macromolecules. Protein template removal is problematic due to size and potential denaturation associated with harsh polymerization conditions (i.e. pH, temperature, solvent). By employing the casein nanoparticle Pickering emulsion, surface-based imprinting is achieved so that the diffusion of the analyte into the polymeric matrix is inconsequential. In order to demonstrate selectivity, the control target proteins include BSA, lysozyme, and trypsin. The reported LOD is 0.911 μg mL^-1. This work employs ultraviolet and visible spectroscopic characterization of samples to assess BHb concentrations dispersed in PBS.

In a separate investigation of surface imprinting, the authors describe a detection system for tramadol comprised of magnetite for sample manipulation, silicon dioxide envelope for further functionalization, and MIP envelope for specific tramadol selectivity (189). With this first nanocomposite the tramadol can be purified from a sample (urine or plasma) and detected via chemiluminescence (CL). This CL reaction was realized thorough reaction of R6G with potassium permanganate. The initially weak CL reaction was enhanced via copper nanoclusters (CuNCs) (15 x) and even more so with the inclusion of copper metal organic framework (CuNCs@CuMOF) (48 x). Extracted tramadol further enhanced CL in proportion to the analyte’s concentration. Linear detection ranged from 0.0030 to 2.5 μM and an LOD of 0.80 nM was reported. The system did not suffer from interference with natural biological compounds nor from other drugs. NIPs were also evaluated with tramadol samples to demonstrate selectivity with good results. While the sensitivities reported here are impressive, the described analyte labelling involves magnetic sample washing for purification purposes, which complicates the process for prospective non-technical users.

A valuable report was published recently in which the authors disclose the comparison of an ECM biosensor with an optical biosensor, namely SPR, both of which were functionalized with MIPs(107). The fabrication of the sensors followed similar designs. Both sensor chips were constructed via analyte templating during the polymerization of pyrrole realized by cyclic voltammetry. Each sensor was used to quantify samples of PBS containing varying concentrations of cancer antigen 125 (CA-125). The ECM sensor exhibited a linear sensitivity ranging from 0.01 to 500 U mL^-1, while the SPR sensor’s sensitivity was between 0.1 to 300 U mL^-1. It was found that the adsorption capacity of the ECM sensor was an order of magnitude greater than the SPR sensor. The improved performance of the ECM technique was explained in terms of the inconsistencies in surface roughness in the Au-layer between the film used in the ECM technique versus disk in the SPR technique. The authors also explain that in the ECM technique, analyte molecules far from the electrode can be detected, but in the SPR technique the target must permeate the polymer layer so as to be adsorbed onto the electrode. While the work is valuable as a comparison of ECM transduction to the SPR strategy, it does raise the question of how other optical sensor strategies would compare. Of course, this may necessitate the use of a label, like QDs or AuNPs, that was otherwise avoided throughout the report.

Historically, ECM transduction has been the most prominent biosensor operational mechanism. Nearly 38% (Fig. 1d) of the devices reviewed in this work (Tables 3 and 4) utilize ECM transduction, a clear majority. Consider the success of the PGM as an example of this fact (153, 190, 191, 192). ECM nanobiosensors are those that quantify a bioanalyte via measurement of an electrical signal occurring after recognition of that target with an MIP. A particular device’s monitored signal subcategorizes ECM biosensors into those that rely on conductometry, potentiometry, impedance, and amperometry. The ECM strategy (Fig. 3c) benefits from a wide variety of benefits. Such devices are easy to miniaturize, inexpensive, selective, sensitive, capable of multiplex sensing, and easy to assemble (25, 180, 193). However, when MIPs are applied to ECM transduction strategies a fundamental handicap becomes apparent. The insulating nature of most functional monomers causes MIPs to inhibit ECM signal generation (167, 171, 194, 195), thus sacrificing sensitivity. Molecularly imprinted electrodes have been shown to suffer from the adhesion of large, dynamic molecules, like enzymes. An ECM device lacking the benefit of MIPs or other nanomaterials may suffer from prohibitive linear detection range, unacceptable selectivity, and poor LOD (106).

First, investigators report on a novel ECM sensor that quantitates cardiac troponin I (cTnI) (196). It achieves this through surface-based analyte detection onto an electrode comprised of an MIP embedded with boron nitride QDs, which lies on a glassy carbon electrode similar to Figs. 2(d) or 3(c). Compared to others this strategy is cost-effective, less wasteful, eco-friendly, and more efficient. The authors address reuse considerations; the sensor can be washed and reused less than 25 times. Regarding the fabrication of the nanobiosensor, QDs were made first. Then, pyrrole containing the target analyte was prepared in a voltammetric cell. The sensor was not sensitive to non-targets including myoglobin, BSA, and cardiac troponin T. It was sensitive to the target, cTnI, in the linear range from 0.01 to 5.00 ng mL-1 with an LOD equal to 0.0005 ng mL^-1. Samples were dispersed in PBS for calibration and in plasma (serum) for selectivity demonstration.

The next ECM case study details a new application of MIPs for cocaine quantification using a potentiometric device (103). MIPNPs were synthesized via solid-phase imprinting technique. The authors narrowed the choices of functional monomers through computational modeling of target analyte and monomer interactions. The three functional monomers with the highest binding affinity toward cocaine were progressed to an experiment optimizing their relative composition with respect to each other. Photopolymerization in organic solvent yielded nearly 20 times more MIPNPs and about 60 times faster compared to chemical polymerization in water. When detecting cocaine in PBS the sensor exhibited a sensitivity between 1 nM to 1 mM. Besides PBS, the nanobiosensor demonstrated selectivity by detecting cocaine in human serum. This research group distinguishes itself in selectivity demonstration by employing NIPs that are actually templated against a non-target molecule, thus ensuring a compelling control experiment. The sensor’s response corresponded directly to cocaine or benzoylecgonine concentrations and was insensitive to the control, galantamine. Since cocaine presented no functional groups for the purposes of templating the MIPs against, the researchers used benzoylecgonine for templating.

A biosensor was recently reported that boasts a wearable device for monitoring sweat via ECL transduction. The design adopts ingenious adhesion of luminescent ruthenium composite nanoparticles on a polymer and gold nanotube substrate for flexibility as well as MIP layer coating for recognition (197). This solution differs from the conventional ECM strategy in that it does not suffer from the same challenges with large target molecule (i.e. enzyme) adhesion to electrodes or the fact that the target need not be electroactive. The researchers demonstrate linear sensitivities for lactate between 50 μM and 1.0 mM as well as between 2.5 mM and 20.0 mM. In addition, linear detection of urea was possible between 45 μM and 19.5 mM. Detection was carried out in PBS, methanol, artificial sweat, and human sweat. Unfortunately, the device was not capable of continuous monitoring due to a need for refreshment via elution.

Another case study in ECL transduction utilized AuNPs embedded in an organic matrix as well as upconversion nanoparticles to achieve synergistic effects with MIP detection (198). AuNPs were chosen in this work for their superior surface area to volume ratio as well as their beneficial electrical conductivity. The AuNPs were embedded in a matrix of cross-linking oligoaniline to achieve electrical transmission in three dimensions, since the spacing between AuNPs is spanned by the organic material, regardless of where a recognition site might be located. This MIP matrix was polymerized from the functional monomer aminothiophenol. A biocompatible, lanthanide-doped nanoparticle was used to produce optical light upon near-infrared laser irradiation. The operation of this up-conversion nanoparticle was intensified via employment of covalent organic framework for its constant porosity as well as favorable selective recognition, surface area, stability, and density. These different nanomaterials obviated the usual disadvantage of poor MIP conductivity and resulted in a highly sensitive and specific device. This ECL sensor detects DA in the linear range from 10^-14 M to 10^-6 M with an LOD equal to 2 x 10^-15 M in PBS. The device proved capable of distinguishing DA from chemically similar compounds like caffeic acid, adrenaline, and α-phenylethylamine, as well as the interfering compounds ascorbic acid and uric acid. When all these compounds were present in rat serum DA elicited a signal approximately three times stronger than any of the others.

Gravimetric nanobiosensors take advantage of the unique mass of the target analyte to achieve selectivity. Examples of this strategy include the piezoelectric cantilever, the quartz crystal microbalance (QCM), mirco electromechanical systems (MEMS), and surface acoustic wave (SAW) devices (199, 200, 201, 202). In general, gravimetric biosensors benefit from many attributes: fast to yield results, sensitive to analyte concentration, inexpensive to construct, and selective for the intended targets. The strategy enjoys facile sample handling due to the label-free operation which can occur in air or liquid. However, gravimetric biosensors attract less attention from researchers compared to other strategies. Only about 13% of the papers reviewed in this work (Tables 3 and 4) utilize gravimetric transduction (Fig. 1d). This is despite the fact that gravimetric nanobiosensors exhibit comparable performance versus ECM devices (203). Unfortunately, since the analyte must come into direct contact with the transducing surface, sample homogenization is required which can add time or steps to the process.

Gravimetric biosensors are experiencing renewed interest due to the potential of MIPs to eliminate wet operation required by antibodies for analyte binding. MIPNPs have been functionalized onto a microcantilever for sensitive mass detection of analyte (204). Investigators describe the preparation of MIPNPs with double phase emulsion polymerization. Specific binding is possible via templated cavities wherein the carboxyl groups of the functional monomer, MAA, noncovalently attracted certain parts of the target analyte. MIPNPs were decorated onto a microcantilever and dynamic sensing mode detected resonant frequency changes after analyte binding. The sensor exhibited a linear detection range for CPX between 1.5 and 150.9 μM. Sensor selectivity for CPX was demonstrated through negligible frequency changes during NIP functionalized microcantilever experiments as well as separate interferent enrofloxacin detection. This strategy is facile, inexpensive, and time efficient compared to conventional nano electromechanical systems (NEMS).

Investigators recently described a unique sensor fabrication obviating the tedium of polymerization (205). This is achieved with commercially available poly(vinylidene difluoride) (PVDF) and chemically modifying it from its native hydrophobicity for compatibility with the hydrophilic trichlorfon (TCF) analyte, a dangerous pesticide requiring food monitoring. PVDF’s long chains knot similar to crosslinking networks. The constructed device exhibits linear detection from 0 to 250 ppb and an LOD equal to 4.63 ppb. An imprinting factor, ratio of MIP selectivity to NIP selectivity, of 3.2 was reported after optimizing the PVDF to TCF amounts. Selectivity was further demonstrated through sensor performance with TCF versus chemically similar pesticides. Biosensor performance was evaluated in water and iceberg lettuce samples. Evaluation of the adsorption isotherms revealed the MIP layer was comprised of homogenous and heterogenous binding sites, after data fitting best matched the Redlich-Peterson model. The researchers conclude their facile synthesis lowers costs.

Another relevant contribution to gravimetric biosensor transduction was made by Tang et al. The researchers report novel construction of a Love-type SAW biosensor on which an MIP recognizes prostate-specific membrane antigen (PSMA) (206). This MIP was fabricated via reversible addition/fragmentation chain transfer (RAFT) precipitation polymerization so that resulting polymer brushes enhanced hydrophilicity. Acrylamide and 2(dimethylamino) ethyl methacrylate were the functional monomers. The synergistic combination of Love sensor with modified MIP results in sensitive aqueous analyte detection. The device exhibits an LOD equal to 0.013 ng mL^-1 and good linear performance from 0.01 to 100 ng mL^-1. The device was selective when assessed against three other analytes. Calibration of PSMA detection was performed in spiked PBS. Trials in mouse serum were conducted to demonstrate applicability to real samples.

One final gravimetric case study describes incorporation of MIPs for NEMS devices. Researchers use MIPs to assess their resistance to damage under formidable wet-etching conditions, like soft lithography, microspotting, photolithography, and two-photon stereolithography, which is characteristic of MEMS fabrication (207). Such processes are detrimental in conventional binding strategies. NEMS devices access ultra-sensitive performance via excellent mass resolution, label-free detection, and highly-integrated operation. The investigators address NEMS obstacles like inefficient transduction integration, low response time associated with miniaturization, and poor multiplexed functionalization. MIP coated cantilevers are selective for the target analyte, the fluorescent amino acid dansyl-L-phenylalanine (DLP), compared to NIP coated cantilevers. The detection performance of eight cantilevers was narrowly uniform. A sensitivity range was not reported due to the low molecular weight of the target and poorly controlled cantilever functionalization. The analyte spiked samples were dispersed in acetonitrile.