3.1. NanoMIPs Characterization Study
Several batches of MIP nanoparticles specific for cocaine, synthesized using the solid phase approach with the same method and monomers composition, were received from the University of Leicester, and these were first characterized by TEM and DLS (Figure 1
). As summarized in Figure 1
a, DLS analysis showed that the size of the nanoMIPs varied from batch to batch, with an average of the hydrodynamic diameter (dH
) across all the batches of 168.80 ± 68.73 nm. Overall, the one-way ANOVA analysis revealed that the cocaine nanoMIPs dH
was different across the batches (F
(3, 74) = 1004.02; p
< 0.00001). Such a size difference may be caused by the variability occurring during the several manual nanoMIPs syntheses, and this could be easily eliminated by automation of the production process. The TEM analysis, performed on one batch, revealed a small degree of nanoMIPs size heterogeneity within the batch itself (see Figure 1
b). According to the TEM analysis, the cocaine nanoMIPs size ranged from 128.98 to 193.09 nm, with an average (±SD) diameter size equal to 148.35 ± 24.69 nm, which is smaller compared to the values achieved by DLS, but in agreement with the size reported in the literature (size range 90–130 nm) [25
]. The disparity between the TEM and DLS values might be due to the difference in the samples state during the measurements (dry for TEM; solvated for DLS).
3.2. EIS NanoMIPs Sensor Construction
The nanoMIPs were functionalized with primary amino groups in order to achieve a covalent attachment to the gold sensor surface. Electrodes (DPR C220AT, DropSens, Spain) were cleaned and then functionalized by directly attaching the cocaine nanoMIPs onto the gold sensor chip via amine coupling. The typical EIS spectra are shown in Figure 2
a. Briefly, a MUDA SAM was first formed through a thermodynamically favored chemisorption of the thiol groups onto the gold surface [28
]. In addition, at a neutral/basic pH, a deprotonation of the MUDA interfacial carboxylic acid groups occurred. Consequently, the electrostatic repulsion between the negatively charged interface and the anionic redox probe induced an increase in Rct value compared to the bare electrode signal [29
]. Although there are other thiol compounds that can be used to provide carboxylic groups on gold sensor surface, MUDA was chosen here, as it has been shown previously in our work to be efficient in attaching sensing elements without affecting their binding affinity to the target analytes [30
]. After MUDA immobilization on the sensor surface, the carboxylic groups were activated by EDC NHS, which decreased the negative charges of the SAM and, hence, induced a drop of the Rct value [31
]. The activated carboxylic groups reacted with the primary amino groups of the cocaine nanoMIPs, thus enabling their covalent attachment. Hence, the Rct value increased due to the nanoMIPs size and insulating properties. Finally, ethanolamine was used to block any unreacted and activated carboxylic groups, thus minimizing the nonspecific binding occurrence. By attaching to the unreacted MUDA, the ethanolamine reduced the negative charges and introduced hydrophilic groups, thus producing a drop of Rct value. Overall the EIS revealed that the nanoMIPs were successfully attached to the electrode surface, although nanoMIPs adsorption onto the surface cannot be completely excluded. Overall, the nanoMIPs immobilization was found to be reproducible, as shown in Figure 2
b. Once functionalized, the electrodes were characterized by using AFM analysis, which was conducted on both the bare and the functionalized sensors. Figure 2
c,d show the changes in roughness of the surface topography for the bare electrode and for the nanoMIPs functionalized sensor surface, respectively.
3.3. EIS NanoMIPs Sensor Optimization
The sensitivity of the EIS nanoMIPs sensor, fabricated and blocked with ethanolamine (1 M, pH 8.5), was tested by performing a cumulative concentration assay, using cocaine in water (100 pg mL−1
to 50 ng mL−1
). The EIS experimental data (Figure 3
a) were fitted in the Randles equivalent circuit, and the Rct values were expressed as Δ % Rct, considering the blank signal value (distilled water) as the starting point. Figure 3
b shows the response curve of the sensor when exposed to increasing concentrations of cocaine. The limit of detection (LOD) was calculated as three times the standard deviation of the blank signals and found to be 0.52 ng mL−1
. No change in Rct values was observed when cocaine was incubated on a control sensor that was functionalized with the same method, but without the nanoMIPs (data not shown).
Nevertheless, to reduce the error bars and enhance the sensor reproducibility, the nanoMIPs sensor was optimized by changing the working solution and optimizing the blocking agents [32
] Therefore, MOPS at pH 7.4 was selected as the working buffer diluent. Then, several blocking reagents were investigated (BSA, milk proteins, Tween 20 and PVA), alone or in combination, as explained above. By comparing the results achieved from these experiments (data not shown), ethanolamine combined with BSA + Tween 20 (pH 7.4) was identified as the optimal blocking condition and was therefore chosen as the best strategy to minimize nonspecific binding and enhance sensor reproducibility.
3.4. Sensor Sensitivity and Specificity
The cocaine cumulative assay was repeated under the optimized assay and blocking conditions. The experimental data were fitted into the simplified Randles equivalent circuit, and the Rct values were extrapolated with an average of error fitting equal to 2.44% (±1.55%). The Rct values were then expressed as −Δ % Rct and were used to plot the calibration curve. As shown in Figure 4
, using these optimized conditions, the sensor was able to detect cocaine at lower concentrations. The R2
of the linear calibration curve was equal to 0.984 (p
-value = 0.00001), and the LOD was found to be 0.24 ng mL−1
The EIS nanoMIPs sensor was then tested against morphine, as another commonly trafficked drug of abuse, and levamisole as a common cutting agent [35
]. Increasing concentrations of each analyte (from 100 pg mL−1
to 50 ng mL−1
) were incubated on the sensor surface, and the EIS spectra were recorded. The data were processed, and the average (±SD) % error of Rct values, as well as the main statistics are reported in Table 1
The results show that the sensor was not responding to increasing concentrations of morphine (Figure 5
). A low response in the opposite direction to that of cocaine was recorded for levamisole. However, although in real drug samples, the cutting agent might be present in lower concentration than cocaine [39
], in this study, it was tested in a similar concentration range (ng mL−1
) (i.e., 1.06–1.67 times higher in molarity), thus assessing a worst-case scenario. Furthermore, the specificity of the nanoMIPs used in this work was also tested against paracetamol and caffeine, using a “Pseudo” ELISA assay, and no cross-reactivity was reported [41
]. Nevertheless, it should be noted that cutting agents change over time and across different countries, as well as across different regions in the same countries, as reported in retrospective studies [36
]. Therefore, the nanoMIPs EIS sensor specificity will need to be reassessed periodically, and further blocking agent optimization studies may be required accordingly.
Overall, the developed EIS nanoMIPs sensor promises to be a valuable alternative to the current onsite screening methods, as a cost-effective, portable, highly sensitive and specific technique. The achieved LOD is slightly above the estimated LOD of the canine olfactory system for cocaine detection [42
], but still below the ppb (ng mL−1
) range required for trace analysis. In addition, the ability of dogs to detect illicit drugs varies according to the breed, training and environmental interference, with the event of cocaine incorrect indication reported to be as high as 26% [4
shows a comparison between the sensor developed in this work with other available analytical tools.
Although mass spectrometry can achieve very low LOD, our EIS nanoMIPs sensor can provide faster results compared to Raman and mass spectroscopy analysis, which require long and tedious sample preparation procedures. The developed sensor can also compete with the extensively used IMS and immunoassays. Indeed, the commercially available IMS has an LOD in the sub ng range (IONSCAN 500DT®
, Smiths Group plc, London, UK), but it is prone to false-positive results due to the competitive ionization of cutting agents or other environmental compounds [60
]. The sensor developed within this work is based on a direct assay format, which decreases the likelihood of false-positive results often linked to the use of competitive inhibition immunoassay screening kits (ELISA kit or lateral flow device) [7
]. Furthermore, the achieved LOD is well beyond the cocaine cut-off (10–50 ng mL−1
) of most immuno-based devices in use [9
]. Biosensor platforms for drugs of abuse detection have been explored previously, and these have shown low LOD, while being faster and cheaper compared to commercially available analytical tools [62
]. As shown in Table 2
, most of the developed biosensors are able to detect cocaine in trace concentration and in a wide range of sample matrices (mostly biological samples). Recently, sub µM limit of detection has been achieved by using bioderived [55
] or hybrid nanozyme receptors [53
]. However, the nanoMIPs used in this work guarantees higher stability against environmental factors that usually affect bioderived receptors (protein, antibodies and aptamers), such as high or low temperatures and enzymatic degradation. In addition, the sensor response here is due to the detection of direct binding between the nanoMIPs and cocaine at the electrodes interface. On the other hand, aptamers-based sensors mostly rely upon the folding of the aptamers, which occurs when the aptamer binds the target [46
], although nonspecific aptamer folding cannot be excluded. Recently, Oliveira and co-workers [52
] demonstrated the development of a holographic sensor based on a biomimetic affinity ligand for the detection of cocaine. Although promising preliminary results have been reported, the specificity toward cocaine cutting agents was not assessed.
Compared to the MIP-film-based optical sensors developed by Wren et al. [50
] and Nguyen et al. [49
], the LOD achieved in our work is far lower (0.24 ng mL−1
or 0.70 nM), and the EIS nanoMIPs sensor specificity has been investigated and confirmed against levamisole and morphine. Compared to optical and piezoelectric platforms, the EIS analyzer used in this work is cheaper and easier to miniaturize to obtain a portable device. Furthermore, the faradic EIS technique used here enhances the sensitivity in a sub ppb range, making the developed sensor a promising platform to detect traces of cocaine in environmental samples with high sensitivity and specificity.