Comparison of Electrochemical Immunosensors and Aptasensors for Detection of Small Organic Molecules in Environment, Food Safety, Clinical and Public Security

We review here the most frequently reported targets among the electrochemical immunosensors and aptasensors: antibiotics, bisphenol A, cocaine, ochratoxin A and estradiol. In each case, the immobilization procedures are described as well as the transduction schemes and the limits of detection. It is shown that limits of detections are generally two to three orders of magnitude lower for immunosensors than for aptasensors, due to the highest affinities of antibodies. No significant progresses have been made to improve these affinities, but transduction schemes were improved instead, which lead to a regular improvement of the limit of detections corresponding to ca. five orders of magnitude over these last 10 years. These progresses depend on the target, however.


Introduction
There is a need to detect, for environment, security or medical purposes, various small organic molecules including antibiotics, drugs, toxins, pollutants, and so on. Although highly sensitive and selective, conventional chromatographic and spectroscopic analytical methods are time-consuming and laborious. Moreover, these techniques require expensive equipment, trained operators and tedious pretreatments. The need for disposable tools for monitoring these molecules encourages the development of simple, efficient, continuous, reliable, cost-effective and field-portable screening methods for analysis of environmental contaminants. Biosensors, and particularly electrochemical ones, appear to be an optimal technology. Since 1962 and the first enzyme-based glucose sensing device, there has been explosive development in biosensors in terms of scientific publications.
The figures of merit for biosensors first depend on the bioprobes which are chosen. Herein we reviewed, among the electrochemical biosensors, the most frequently reported bioprobes (antibodies, aptamers, and peptides, when applicable) for frequently reported targets: antibiotics, bisphenol A, cocaine, ochratoxin A and estradiol. In each case, the immobilization procedure was described, as well as the transduction scheme and the limit of detection. We will compare the results from one probe to another and discuss their efficiency for each target, respectively.
Biosensors 2016, 6, 7; doi:10.3390/bios6010007 www.mdpi.com/journal/biosensors Figure 1. Immunoglobulin G (IgG) consists of two heavy chains (VH, CH1 and CH2) and two light chains (VL and CL). C are constant regions (which do not bring specificity to a given antigen) whereas V are variable regions, which bring specificity. As a consequence, antigens bind to the variable regions VL and VH. In order to decrease the size of IgG, only the Fab (antigen binding fragment, composed of the VH, VL, CH1 and CL regions) can be employed. Reproduced from [3].

Antibodies
In 2007, Garifallou et al. [5] described a label-free immunosensor for ciprofloxacin based on antibody-modified screen-printed carbon electrodes (SPEs), using electrochemical impedance spectroscopy (EIS) for detection. The antibody (Ab) was immobilized through biotin-avidin interactions. The impedance was reported to increase with ciprofloxacin concentration, with a linearity domain between 1 and 100 ng·mL −1 (3 nM-0.3 nM). The same year, another group proposed a similar approach, but the Ab was immobilized by peptide coupling, presenting must less steric hindrance than the biotin-avidin coupling, which allowed to decrease the LoD down to 10 pg·mL −1 (30 pM) [6]. In 2009, the same group published an innovative immunosensor where ciprofloxacin was first grafted on the electrode, onto which the anti-ciprofloxacin Ab was complexed ( Figure 3). In the presence of ciprofloxacin, the antibody was displaced in solution inducing strong changes in the impedance of the electrode; the LoD was of ca. 1 pg·mL −1 (3 pM) [7].

Antibodies
In 2007, Garifallou et al. [5] described a label-free immunosensor for ciprofloxacin based on antibody-modified screen-printed carbon electrodes (SPEs), using electrochemical impedance spectroscopy (EIS) for detection. The antibody (Ab) was immobilized through biotin-avidin interactions. The impedance was reported to increase with ciprofloxacin concentration, with a linearity domain between 1 and 100 ng¨mL´1 (3 nM-0.3 nM). The same year, another group proposed a similar approach, but the Ab was immobilized by peptide coupling, presenting must less steric hindrance than the biotin-avidin coupling, which allowed to decrease the LoD down to 10 pg¨mL´1 (30 pM) [6]. In 2009, the same group published an innovative immunosensor where ciprofloxacin was first grafted on the electrode, onto which the anti-ciprofloxacin Ab was complexed ( Figure 3). In the presence of ciprofloxacin, the antibody was displaced in solution inducing strong changes in the impedance of the electrode; the LoD was of ca. 1 pg¨mL´1 (3 pM) [7].

Antibodies
In 2007, Garifallou et al. [5] described a label-free immunosensor for ciprofloxacin based on antibody-modified screen-printed carbon electrodes (SPEs), using electrochemical impedance spectroscopy (EIS) for detection. The antibody (Ab) was immobilized through biotin-avidin interactions. The impedance was reported to increase with ciprofloxacin concentration, with a linearity domain between 1 and 100 ng·mL −1 (3 nM-0.3 nM). The same year, another group proposed a similar approach, but the Ab was immobilized by peptide coupling, presenting must less steric hindrance than the biotin-avidin coupling, which allowed to decrease the LoD down to 10 pg·mL −1 (30 pM) [6]. In 2009, the same group published an innovative immunosensor where ciprofloxacin was first grafted on the electrode, onto which the anti-ciprofloxacin Ab was complexed ( Figure 3). In the presence of ciprofloxacin, the antibody was displaced in solution inducing strong changes in the impedance of the electrode; the LoD was of ca. 1 pg·mL −1 (3 pM) [7].  Detection of enrofloxacin on SAM-modified electrodes gave similar results. For example, Wu et al., (2009) reported EIS characterization of interactions between enrofloxacin and its Ab immobilized on a SAM-modified Au electrode; this interaction caused an increase of electron transfer resistance (Rct) with a LoD of 1 ng¨mL´1 (3 nM). [8] Zhang et al., (2013) described a competitive enzyme-amplified immunosensor (using the horseradish peroxidase enzyme, HRP) based on gold nanoclusters/polypyrrole (PPy)-functionalized electrodes, for sensitive detection of ofloxacin [9], which exhibited a response in the range 0.08-410 ng¨mL´1, with a LoD of 30 pg¨mL´1 (0.1 nM).
The use of magnetic beads (MBs) can improve sensitivity. For example, Zacco et al. [10] described an electrochemical immunosensor for the detection of sulfonamide antibiotics based on magnetic beads, amplified with the use of HRP as label. The LoD was ca. 1 µg¨L´1 (6 nM). Centi et al. described a more complicated transduction architecture [11], using protein A immobilized on magnetic microbeads (MBs) to bind the specific Ab. A competition assay between the target analyte and the HRP-labeled sulfonamide target was carried out. The LoD was ca. 1 ng¨mL´1 (4 nM). Conzuelo et al. (2012) [12] used a selective capture Ab immobilized on the surface of MBs for tetracycline detection. A direct competitive immunoassay using HRP for the enzymatic labeling was performed, with hydroquinone (HQ) as redox mediator. They obtained low LoD (in the ng¨mL´1 range, i.e., the nM range) for 4 tetracycline antibiotics. The same group described more recently a similar procedure where Abs were immobilized by the diazonium route [13]. Que et al. [14] nanostructured their electrodes for tetracycline detection using platinum NPs deposited on graphene nanosheets (GN). This way, retaining a competitive immunoassay format, they significantly lowered the LoD down to 6 pg¨mL´1 (13 pM).
Even if, as it is generally the case for immunosensors, enzyme-labeling is the dominant strategy for improving sensitivity, it appears that non-enzymatic amplification (through increase of the active surface area or use of catalytic reactions) are often more efficient than classical (HRP) enzymatic amplification.

Aptamers
Aptamers have been used for tetracycline detection as well, by Kim et al., in 2009. They used interdigitated gold electrodes modified with a 76-mer thiol-ssDNA aptamer. The dissociation constant (KD) of this aptamer for oxytetracycline was 11 nM. Transduction was explained by the fact that the redox probe diffusing in solution (Fe(CN) 6 3´{4´) was influenced by interactions between the aptamers and the target molecules. The dynamic range was 1-100 nM (LoD of 1 nM, i.e., 0.3 ng¨mL´1) [15]. In 2010, the same group [16] described similar results but using a streptavidin-modified screen-printed gold electrode onto which biotinylated ssDNA aptamers were immobilized. Still using CV and SWV with Fe(CN) 6 3´{4´a s diffusing probe, they found a LoD for tetracycline of 10 nM (i.e., 3 ng¨mL´1). In another similar example, Zhou et al. [17], still using diffusing Fe(CN) 6 3´{4´a s redox probe, described an electrochemical tetracycline aptasensor involving multiwalled carbon nanotubes (MWCNTs), the electroactivity of which being decreased due to the formation of aptamer-tetracycline complexes. The peak current changes obtained by differential pulse voltammetry (DPV) increased linearly with tetracycline concentrations from 5 pM (1.5 ng¨mL´1) to 50 µM. Chen et al., in 2014 [18], used EIS with a 13 bases-long aptamer; a linear relationship between the log concentration of tetracycline and the charge transfer resistance was found from 5 ng¨mL´1 to 5 µg¨mL´1; the LoD was 1 ng¨mL´1 (ca. 2 nM). This LoD was significantly lower than the K D determined by the authors using calorimetry (K D of 50 µM). Shen et al. [19] used differential pulse voltammetry (DPV) with Prussian blue immobilized onto AuNPs modified with a 79 bases long aptamer having a KD of 63 nM for tetracycline. The detection range was linear from 10´9 to 10´5 M, with a LoD of 0.3 nM (i.e., 0.1 ng¨mL´1). The first label-free electrochemical RNA-based aptasensor for detection of an antibiotic was described in 2007 for neomycin ( Figure 4) [20]. The authors described a competitive displacement assay and used EIS. The selectivity of the RNA aptamer allowed determination of neomycin even in biological samples of high protein content. The sensitivity was not excellent, around 1 µM, close to the K D of the aptamers-neomycin complex. The previous examples are mostly based on the use of a redox probe diffusing in solution, which brings sensitivity but is obviously difficult to integrate when a system is really applied. In 2010, the group of Plaxco showed that classical RNA aptamers, even if sensitive to nuclease, can be used in biological samples providing that these samples were previously ultra-filtrated. They demonstrated the detection of tobramycin in the range 2-6 μM (4-10 μg mL −1 ) in blood using a methylene blue (MB)-modified RNA aptamer sequence 5′-HS-GGG ACU UGG UUU AGG UAA UGA GUC CC-NH-MB-3′, having a dissociation constant of 13.2 μM for tobramycin (measured in homogenous conditions) [21]. They have shown that the KD increased at least by one order of magnitude in heterogeneous conditions. A similar approach has been reported in several following papers, with variable results. For example, Liu et al., in 2014 [22], reported an electrochemical DNA aptamer-based tobramycin sensor (5′-HS-C6-GGG ACT TGG TTT AGG TAA TGA GTC CC-MB-3′ sequence, immobilized on gold microelectrodes). Despite the electrochemical deposition of dendritic gold nanostructures, they reported a high LoD of ca. 1 mM (KD in the mM range as well).
The reagentless detection of small molecules using aptamers is not straightforward because the spatial reorganization triggered by the analyte is often small. To address this issue, Gonzales-Fernandes et al. described a strategy based on competition between tobramycin immobilized on magnetic microbeads (MBs) and free tobramycin diffusing in solution. Detection was made by postlabeling aptamers with an enzyme (alkaline phosphatase, AlkP) through the biotin-avidin conjugate. Using a 27-mer anti-tobramycin RNA aptamer (5′-GGC ACG AGG UUU AGC UAC ACU CGU GCC-3′), they found a linear response in the range 5-500 μM [23]. The same authors described later a slightly modified architecture for tobramycin detection using an antibody-antigen association for introducing the enzyme conjugate on the aptamers [24]. Compared with the biotin-avidin approach, the limit of detection was lowered down to 0.1 μM. The authors explained this behavior by the steric hindrance of avidin, which impeded detection.
As shown, reported LoD stay too high for practical applications when aptamers are labeled with redox probes. Another way to detect a molecule through the use of aptamers is to us them as preconcentrators immobilized on the electrode surface, followed by electro-oxidation or electroreduction of the target molecule directly on the electrode, without any other intermediate. Zhu et al. [25] proposed such an approach applied to kanamycin and involving AuNPs. The authors used a 22-bases DNA aptamer (5′-TGG GGG TTG AGG CTA AGC CGA C-3′) showing a high binding affinity to kanamycin (KD = 79 nM), immobilized onto AuNPs modified with poly-[2,5-di-(2-thienyl)-1H-pyrrole-1-(pbenzoic acid)]. Kanamycin detection was achieved by CV and linear sweep voltammetry (LSV). The calibration plot showed a linear range from 0.05 μM to 9.0 μM with a LoD significantly lowered (10 The previous examples are mostly based on the use of a redox probe diffusing in solution, which brings sensitivity but is obviously difficult to integrate when a system is really applied. In 2010, the group of Plaxco showed that classical RNA aptamers, even if sensitive to nuclease, can be used in biological samples providing that these samples were previously ultra-filtrated. They demonstrated the detection of tobramycin in the range 2-6 µM (4-10 µg¨mL´1) in blood using a methylene blue (MB)-modified RNA aptamer sequence 5 1 -HS-GGG ACU UGG UUU AGG UAA UGA GUC CC-NH-MB-3 1 , having a dissociation constant of 13.2 µM for tobramycin (measured in homogenous conditions) [21]. They have shown that the K D increased at least by one order of magnitude in heterogeneous conditions. A similar approach has been reported in several following papers, with variable results. For example, Liu et al., in 2014 [22], reported an electrochemical DNA aptamer-based tobramycin sensor (5 1 -HS-C6-GGG ACT TGG TTT AGG TAA TGA GTC CC-MB-3 1 sequence, immobilized on gold microelectrodes). Despite the electrochemical deposition of dendritic gold nanostructures, they reported a high LoD of ca. 1 mM (K D in the mM range as well).
The reagentless detection of small molecules using aptamers is not straightforward because the spatial reorganization triggered by the analyte is often small. To address this issue, Gonzales-Fernandes et al. described a strategy based on competition between tobramycin immobilized on magnetic microbeads (MBs) and free tobramycin diffusing in solution. Detection was made by post-labeling aptamers with an enzyme (alkaline phosphatase, AlkP) through the biotin-avidin conjugate. Using a 27-mer anti-tobramycin RNA aptamer (5 1 -GGC ACG AGG UUU AGC UAC ACU CGU GCC-3 1 ), they found a linear response in the range 5-500 µM [23]. The same authors described later a slightly modified architecture for tobramycin detection using an antibody-antigen association for introducing the enzyme conjugate on the aptamers [24]. Compared with the biotin-avidin approach, the limit of detection was lowered down to 0.1 µM. The authors explained this behavior by the steric hindrance of avidin, which impeded detection.
As shown, reported LoD stay too high for practical applications when aptamers are labeled with redox probes. Another way to detect a molecule through the use of aptamers is to us them as pre-concentrators immobilized on the electrode surface, followed by electro-oxidation or electroreduction of the target molecule directly on the electrode, without any other intermediate. Zhu et al. [25] proposed such an approach applied to kanamycin and involving AuNPs. The authors used a 22-bases DNA aptamer (5 1 -TGG GGG TTG AGG CTA AGC CGA C-3 1 ) showing a high binding affinity to kanamycin (K D = 79 nM), immobilized onto AuNPs modified with poly-[2,5-di-(2-thienyl)-1H-pyrrole-1-(p-benzoic acid)]. Kanamycin detection was achieved by CV and linear sweep voltammetry (LSV). The calibration plot showed a linear range from 0.05 µM to 9.0 µM with a LoD significantly lowered (10 nM, i.e., 5 ng¨mL´1) compared to the previously reported architectures. Compared to the other amplification strategies used with aptamers, this approach of preconcentration followed by stripping (adsorptive stripping voltammetry-ASV) is certainly the one which leads to the highest sensitivities after EIS.
Lastly, it should also be stressed that attempts to make all-polymer, all-printed, electrochemical sensors were reported. For example, Dapra et al. [26] presented an all-polymer electrochemical biosensor based on conductive bilayer consisting of tosylate-doped poly(3,4-ethylenedioxythiophene) (PEDOT-TsO) and its hydroxymethyl derivative which was covalently functionalized with two short aptamer probes with affinity to ampicillin or kanamycin A, respectively. The ampicillin DNA aptamer sequence was 5 1 -GCG GGC GGT TGT ATA GCG G-3 1 (K D = 13 nM) and the kanamycin one was 5 1 -TGG GGG TTG AGG CTA AGC CGA-3 1 (K D = 79 nM). Using EIS, they were able to detect ampicillin in a concentration range from 100 pM to 1 µM and kanamycin A from 10 nM to 1 mM. As discussed by the authors, the low LoD and wide dynamic range of their biosensor could be attributed not only to the high affinity of the DNA probes but also to their device properties. Indeed, for a similar probe applied in different techniques, LoDs can change over three orders of magnitude.
All these results are summarized in Table 1. As shown, tetracycline is a common target for Abs and aptamers. LoDs in the nM range were obtained for both capture probes, without significant differences. For a same probe, it appears that electrode nanostructuration leads to an improvement of the detection limit of at least one order of magnitude. It is also interesting to note that methods involving enzyme amplifications applied with aptamer probes do not appear to be more sensitive than non-amplified methods.   [26] Note: LoD (limit of detection); HRP (horseradish peroxidase); PPy (polypyrrole); HER (hydrogen evolution reaction), CV (cyclic voltammetry); SWV (square wave voltammetry); MBs (magnetic beads); CNTs (carbon nanotubes); PB (prussian blue); MB (methylene blue); AuNPs (gold nanoparticles).

Bisphenol A
Bisphenol-A (BPA) ( Figure 5) is a well-known and much studied contaminant causing a wide range of health problems to living beings, especially the young. In 2013, Ragavan et al. reviewed the various principles, mechanisms and performances of BPA biosensors available in the literature [27], among which are found electrochemical immuno-and apta-sensors.

Bisphenol A
Bisphenol-A (BPA) ( Figure 5) is a well-known and much studied contaminant causing a wide range of health problems to living beings, especially the young. In 2013, Ragavan et al. reviewed the various principles, mechanisms and performances of BPA biosensors available in the literature [27], among which are found electrochemical immuno-and apta-sensors.  [28,29]. The immunosensor was fabricated by covalent binding of the Ab onto COOH-functionalized polythiophene. It showed specific recognition of BPA on a linear dynamic range between 1 and 100 ng mL −1 . The LoD was 0.3 ng·mL −1 (1.3 nM). Wang et al. published recently a label-free electrochemical competitive immunosensor able to detect BPA directly with a limit of detection of 2 pg·mL −1 (9 pM). Characterizations made by SWV showed that the polymer film presented a current decrease upon anti-BPA binding and an opposite current increase upon BPA addition in solution [30]. The difference in sensitivity between the results of Rahman et al., Piao et al. and Wang et al. could be explained by the nature of transduction scheme. Indeed, the latter reported the use of a competitive immunodisplacement approach whereas the two first articles reported a classical Ab-Ag interaction where the Ab was covalently immobilized on the electrode.

Aptamers
Even if strategies to develop antibodies towards bisphenol A, including antibodies with good affinities, have been reported (see above), DNA aptamers specific to BPA were reported to be of higher affinity than antibodies (in the nM range). For this reason, lower LoDs have been reached. Electrochemical transduction was investigated by Xue et al. [31]. They used a specific DNA aptamer against BPA (5′-CCG GTG GGT GGT CAG GTG GGA TAG CGT TCC GCG TAT GGC CCA GCG CAT CAC GGG TTC GCA CCA-3′), immobilized on the surface of a gold electrode via self-assembly and hybridized with its complementary strand. A redox intercalator was added. The detection of BPA was based on the competitive recognition of BPA by the immobilized aptamer on the surface of the electrode, which dehybridized the complementary strand and therefore freed the redox intercalator into solution. The redox signal was recorded by CV. The LoD was 0.3 pg·mL −1 (1 pM).
Graphene-modified glassy carbon (GC) electrodes were also investigated, which confers higher specific active areas. Zhou et al. [32] immobilized AuNPs on GR-modified electrodes and used freely diffusing Fe(CN)6 3−/4− as electrochemical probe. The DPV peak current of Fe(CN)6 3−/4− changed linearly with the concentration of BPA in the range from 10 nM to 10 μM, with a LoD of 5 nM (ca. 1 ng·mL −1 ). They used the same aptamer sequence as the previous reference.

Peptides
Specific peptides were also reported for BPA. Yang et al. [33] described an electrochemical sensor using the peptide sequence Cys-Lys-Ser-Leu-Glu-Asn-Ser-Tyr-Cys (CKSLENSYC) capable of recognizing BPA with high specificity (isolated using a phage display technique). Peptides were immobilized by thiol adsorption on Au electrodes and BPA was sensed using DPV with Fe(CN)6 3−/4− as redox probe, with a broad detection range between 1 nM and 5 μM (LoD of 1 nM, i.e., 0.2 ng·mL −1 ). Kim et al. also reported very recently the electrochemical detection of BPA with a protein immobilized on RGO Indeed, the latter reported the use of a competitive immunodisplacement approach whereas the two first articles reported a classical Ab-Ag interaction where the Ab was covalently immobilized on the electrode.

Aptamers
Even if strategies to develop antibodies towards bisphenol A, including antibodies with good affinities, have been reported (see above), DNA aptamers specific to BPA were reported to be of higher affinity than antibodies (in the nM range). For this reason, lower LoDs have been reached. Electrochemical transduction was investigated by Xue et al. [31]. They used a specific DNA aptamer against BPA (5 1 -CCG GTG GGT GGT CAG GTG GGA TAG CGT TCC GCG TAT GGC CCA GCG CAT CAC GGG TTC GCA CCA-3 1 ), immobilized on the surface of a gold electrode via self-assembly and hybridized with its complementary strand. A redox intercalator was added. The detection of BPA was based on the competitive recognition of BPA by the immobilized aptamer on the surface of the electrode, which dehybridized the complementary strand and therefore freed the redox intercalator into solution. The redox signal was recorded by CV. The LoD was 0.3 pg¨mL´1 (1 pM).
Graphene-modified glassy carbon (GC) electrodes were also investigated, which confers higher specific active areas. Zhou et al. [32] immobilized AuNPs on GR-modified electrodes and used freely diffusing Fe(CN) 6 3´{4´a s electrochemical probe. The DPV peak current of Fe(CN) 6 3´{4´c hanged linearly with the concentration of BPA in the range from 10 nM to 10 µM, with a LoD of 5 nM (ca. 1 ng¨mL´1). They used the same aptamer sequence as the previous reference.

Peptides
Specific peptides were also reported for BPA. Yang et al. [33] described an electrochemical sensor using the peptide sequence Cys-Lys-Ser-Leu-Glu-Asn-Ser-Tyr-Cys (CKSLENSYC) capable of recognizing BPA with high specificity (isolated using a phage display technique). Peptides were immobilized by thiol adsorption on Au electrodes and BPA was sensed using DPV with Fe(CN) 6 3´{4´a s redox probe, with a broad detection range between 1 nM and 5 µM (LoD of 1 nM, i.e., 0.2 ng¨mL´1). Kim et al. also reported very recently the electrochemical detection of BPA with a protein immobilized on RGO electrodes [34]. The protein was a recombinant one obtained by fusing the disulfide-constrained high affinity BPA binding peptide (CKSLENSYC, as reported by Yang et al.) to the C-terminus of the Lac repressor (Lad), which is a large protein (140 Åˆ60 Åˆ45 Å). This modified protein was immobilized on RGO by heat-denaturation, so that it can adsorb by π-stacking interactions. EIS was used (the redox reaction is not given in their manuscript) to probe the presence of BPA down to 100 fM (23 fg¨mL´1), without interference with bisphenol S and bisphenol F. All these results are summarized in Table 2. Considering that BPA, Abs, aptamers and peptides have yet been poorly reported in the literature, an unbiased comparison is not possible.

Cocaine
Cocaine ( Figure 6) is the most frequently used illegal drug after cannabis (ca. 15 million users around the world) and directly provokes several thousands of deaths each year. Its detection, e.g., in airports, is routinely made with trained dogs, but more automatic systems would be extremely useful. electrodes [34]. The protein was a recombinant one obtained by fusing the disulfide-constrained high affinity BPA binding peptide (CKSLENSYC, as reported by Yang et al.) to the C-terminus of the Lac repressor (Lad), which is a large protein (140 Å × 60 Å × 45 Å). This modified protein was immobilized on RGO by heat-denaturation, so that it can adsorb by π-stacking interactions. EIS was used (the redox reaction is not given in their manuscript) to probe the presence of BPA down to 100 fM (23 fg·mL −1 ), without interference with bisphenol S and bisphenol F. All these results are summarized in Table 2. Considering that BPA, Abs, aptamers and peptides have yet been poorly reported in the literature, an unbiased comparison is not possible.

Cocaine
Cocaine ( Figure 6) is the most frequently used illegal drug after cannabis (ca. 15 million users around the world) and directly provokes several thousands of deaths each year. Its detection, e.g., in airports, is routinely made with trained dogs, but more automatic systems would be extremely useful.

Antibodies
There are very few examples of immunosensors dedicated to narcotics. The two very first articles were published in 1998 and dealt with cocaine detection. In 1998, Bauer et al. [35] proposed an automated amplified flow immunoassay for cocaine using the alkaline phosphatase (AlkP) enzyme label for amplification and transduction. The LoD was 380 pM. The same year, Suleiman et al. [36] described an amperometric immunosensor for cocaine using HRP and, as in the case of Bauer et al., an oxygen electrode. Cocaine was quantified in the range 10 −7 -10 −5 M, i.e., at too high concentrations for being of practical use.
More than 15 years passed before another example of immunosensor for cocaine or morphine was published. It is only recently that, considering the good results obtained by aptamer-based cocaine sensors, researchers began again to study cocaine-specific antibodies. The literature remains very limited, however. Yang et al. [37] described an immunosensor based on a SAM-modified Au electrode for detection of morphine and methamphetamine (MA), using EIS, with a low LoD of 10 pg·L −1 (33 fM). For more information, a very recent (2015) book chapter published by Ozkan et al., is available, dealing with electrochemical biosensors for drug analysis [38].

Antibodies
There are very few examples of immunosensors dedicated to narcotics. The two very first articles were published in 1998 and dealt with cocaine detection. In 1998, Bauer et al. [35] proposed an automated amplified flow immunoassay for cocaine using the alkaline phosphatase (AlkP) enzyme label for amplification and transduction. The LoD was 380 pM. The same year, Suleiman et al. [36] described an amperometric immunosensor for cocaine using HRP and, as in the case of Bauer et al., an oxygen electrode. Cocaine was quantified in the range 10´7-10´5 M, i.e., at too high concentrations for being of practical use.
More than 15 years passed before another example of immunosensor for cocaine or morphine was published. It is only recently that, considering the good results obtained by aptamer-based cocaine sensors, researchers began again to study cocaine-specific antibodies. The literature remains very limited, however. Yang et al. [37] described an immunosensor based on a SAM-modified Au electrode for detection of morphine and methamphetamine (MA), using EIS, with a low LoD of 10 pg¨L´1 (33 fM). For more information, a very recent (2015) book chapter published by Ozkan et al., is available, dealing with electrochemical biosensors for drug analysis [38].

Aptamers
By contrast, aptamer-based sensors for cocaine were more frequently reported. The first significant work on this topic was published in 2006 by Baker et al. [39]. They demonstrated a rapid, label-free, electrochemical method for the detection of small molecules in general, based on a target-induced conformational change, unaffected by non-specific contaminants, and illustrated by the detection of cocaine in blood serum, saliva, and other complex samples. Two DNA aptamers were investigated (GAC AAG GAA AAT CCT TCA ATG AAG TGG GTC and AGA CAA GGA AAA TCC TTC AAT GAA GTG GGT CG) with K D of approximately 90 µM each, which allowed a LoD below 10 µM ( Figure 7); however, these were too high to allow for practical applications.

Aptamers
By contrast, aptamer-based sensors for cocaine were more frequently reported. The first significant work on this topic was published in 2006 by Baker et al. [39]. They demonstrated a rapid, label-free, electrochemical method for the detection of small molecules in general, based on a targetinduced conformational change, unaffected by non-specific contaminants, and illustrated by the detection of cocaine in blood serum, saliva, and other complex samples. Two DNA aptamers were investigated (GAC AAG GAA AAT CCT TCA ATG AAG TGG GTC and AGA CAA GGA AAA TCC TTC AAT GAA GTG GGT CG) with KD of approximately 90 μM each, which allowed a LoD below 10 μM (Figure 7); however, these were too high to allow for practical applications. The same group reported later [40] a sandwich assay based on single aptamer sequences labeled with MB and able to detect cocaine. The principle relied on the formation of a complex made of twostrand aptamers which was stabilized together by the presence of the target (Figure 8). Even though the KD was estimated at around 1 mM, the authors reported a significantly lower LoD of ca.  The same group reported later [40] a sandwich assay based on single aptamer sequences labeled with MB and able to detect cocaine. The principle relied on the formation of a complex made of two-strand aptamers which was stabilized together by the presence of the target (Figure 8). Even though the K D was estimated at around 1 mM, the authors reported a significantly lower LoD of ca.

Aptamers
By contrast, aptamer-based sensors for cocaine were more frequently reported. The first significant work on this topic was published in 2006 by Baker et al. [39]. They demonstrated a rapid, label-free, electrochemical method for the detection of small molecules in general, based on a targetinduced conformational change, unaffected by non-specific contaminants, and illustrated by the detection of cocaine in blood serum, saliva, and other complex samples. Two DNA aptamers were investigated (GAC AAG GAA AAT CCT TCA ATG AAG TGG GTC and AGA CAA GGA AAA TCC TTC AAT GAA GTG GGT CG) with KD of approximately 90 μM each, which allowed a LoD below 10 μM (Figure 7); however, these were too high to allow for practical applications. The same group reported later [40] a sandwich assay based on single aptamer sequences labeled with MB and able to detect cocaine. The principle relied on the formation of a complex made of twostrand aptamers which was stabilized together by the presence of the target (Figure 8). Even though the KD was estimated at around 1 mM, the authors reported a significantly lower LoD of ca.  On the same principle, Golub et al. [41] described two anti-cocaine aptamer subunits, where one subunit was assembled on an Au electrode, and the second aptamer subunit was labeled with Pt-NPs. The supramolecular Pt-NPs-aptamer subunits-cocaine complex allowed the detection of cocaine by the electrocatalyzed reduction of H 2 O 2 on the Pt-NPs. The detection limit was in the range of 10´6-10´5 M (Figure 9).
On the same principle, Golub et al. [41] described two anti-cocaine aptamer subunits, where one subunit was assembled on an Au electrode, and the second aptamer subunit was labeled with Pt-NPs. The supramolecular Pt-NPs-aptamer subunits-cocaine complex allowed the detection of cocaine by the electrocatalyzed reduction of H2O2 on the Pt-NPs. The detection limit was in the range of 10 −6 -10 −5 M (Figure 9). Again regarding the use of a two-strand aptamers, Du et al. [42] constructed a label-free electrochemical aptasensor using layer-by-layer self-assembly of Fc-modified poly(ethyleneimine), with a LoD of 0.1 μM. A similar approach was described by Zhang et al. [43] using a primary immobilized aptamer and a secondary aptamer labeled with various quantum dots used in the same way than in bar-code assays, i.e., different QDs yield distinct electrochemical signatures. A LoD of 50 nM was obtained. Another approach using multiple stems was described by Hua et al. [44] (with the cocaine DNA aptamer 5′-GGG AGA CAA GGA TAA ATC CTT CAA TGA AGT GGG TCT CCC-3′), where the transduction step was also based on a hairpin structure. The linear range was from 10 −9 M to 10 −6 M. Again using the multiple stem approach, Zhang et al. [45] investigated the formation of the corresponding supramolecular aptamer fragments/cocaine complex by EIS in the presence of Fe(CN)6 3−/4− . Cocaine concentrations as low as 100 nM were detected. Very recently, Roushani and Shahdost-Fard [46] described a sensor based on the same principle with aptamer-functionalized AuNPs. The interaction of cocaine with the aptamer caused the AuNPs to come close to the electrode surface and impede diffusion of the redox probe Fe(CN)6 3−/4− . Monitored with DPV, the detection limit was 100 pM. The same authors described a similar approach using AgNPs [47] and riboflavin as the redox probe; the LoD was equivalent, i.e., 150 pM.
Jiang et al. [48] used the same supramolecular aptamer sequences for detection of cocaine using an alkaline phosphatase-modified binding aptamer, leading to a LoD of 1 nM. A similar approach was used by Zhang et al. [49] but using HRP as the amplifying label. DPV was used to obtain a LoD down to 20 nM. Lastly, Shen et al. [50] added rolling circle amplification (RCA) in addition to alkaline phosphatase amplification, and used DPV for detection. This dual amplification strategy strongly decreased the detection limit down to 1 nM.
All these results are summarized in Table 3. Due to very few examples of Ab sensors for cocaine, a comparison between Abs and aptamers cannot be objective. As for other targets, nanostructuration of the electrodes led to a gain in terms of LoD, of one to three orders of magnitude. Again, as for antibiotics, enzymatic amplifications applied to aptamers do not bring lower detection limits, as it could be expected. Lastly, even if it is an obvious observation, these results show that the sensitivity largely depends on the affinity of the aptamer for its target: the lowest LoD were obtained with aptamers having low KD. Again regarding the use of a two-strand aptamers, Du et al. [42] constructed a label-free electrochemical aptasensor using layer-by-layer self-assembly of Fc-modified poly(ethyleneimine), with a LoD of 0.1 µM. A similar approach was described by Zhang et al. [43] using a primary immobilized aptamer and a secondary aptamer labeled with various quantum dots used in the same way than in bar-code assays, i.e., different QDs yield distinct electrochemical signatures. A LoD of 50 nM was obtained. Another approach using multiple stems was described by Hua et al. [44] (with the cocaine DNA aptamer 5 1 -GGG AGA CAA GGA TAA ATC CTT CAA TGA AGT GGG TCT CCC-3 1 ), where the transduction step was also based on a hairpin structure. The linear range was from 10´9 M to 10´6 M. Again using the multiple stem approach, Zhang et al. [45] investigated the formation of the corresponding supramolecular aptamer fragments/cocaine complex by EIS in the presence of Fe(CN) 6 3´{4´. Cocaine concentrations as low as 100 nM were detected. Very recently, Roushani and Shahdost-Fard [46] described a sensor based on the same principle with aptamer-functionalized AuNPs. The interaction of cocaine with the aptamer caused the AuNPs to come close to the electrode surface and impede diffusion of the redox probe Fe(CN) 6 3´{4´. Monitored with DPV, the detection limit was 100 pM. The same authors described a similar approach using AgNPs [47] and riboflavin as the redox probe; the LoD was equivalent, i.e., 150 pM. Jiang et al. [48] used the same supramolecular aptamer sequences for detection of cocaine using an alkaline phosphatase-modified binding aptamer, leading to a LoD of 1 nM. A similar approach was used by Zhang et al. [49] but using HRP as the amplifying label. DPV was used to obtain a LoD down to 20 nM. Lastly, Shen et al. [50] added rolling circle amplification (RCA) in addition to alkaline phosphatase amplification, and used DPV for detection. This dual amplification strategy strongly decreased the detection limit down to 1 nM.
All these results are summarized in Table 3. Due to very few examples of Ab sensors for cocaine, a comparison between Abs and aptamers cannot be objective. As for other targets, nanostructuration of the electrodes led to a gain in terms of LoD, of one to three orders of magnitude. Again, as for antibiotics, enzymatic amplifications applied to aptamers do not bring lower detection limits, as it could be expected. Lastly, even if it is an obvious observation, these results show that the sensitivity largely depends on the affinity of the aptamer for its target: the lowest LoD were obtained with aptamers having low K D .

Ochratoxin A
A toxin is a poisonous substance produced by living cells or organisms. Toxins can be small organic molecules, peptides, or proteins. The most studied toxin in electrochemical biosensors is ochratoxin A (Figure 10). It is one of the most abundant food-contaminating mycotoxins.

Ochratoxin A
A toxin is a poisonous substance produced by living cells or organisms. Toxins can be small organic molecules, peptides, or proteins. The most studied toxin in electrochemical biosensors is ochratoxin A (Figure 10). It is one of the most abundant food-contaminating mycotoxins.

Antibodies
Prieto-Simon et al. described, in 2008 [51], two indirect electrochemical competitive enzyme-linked immunosorbent assays (ELISA) for ochratoxin A (OTA). Performances of polyclonal (pAb) and monoclonal (mAb) antibodies against OTA were compared, showing at least one order of magnitude higher affinity values when working with mAb. Alkaline phosphatase (AlkP) and horseradish peroxidase (HRP)-labeled secondary antibodies were evaluated. Similar limits of detection of 0.7 and 0.3 ng·mL −1 (0.8 nM) were obtained for HRP-and AlkP-labeled immunosensors, respectively. Later, in 2009, Radi et al. [52] described an electrochemical immunosensor for detection of OTA on screenprinted gold electrodes modified by electroreduction of 4-nitrophenyl diazonium salts and anti-OTA. A competition between OTA and HRP-labeled OTA (OTA-HRP) for the immobilized antibodies was held. The activity of the bound OTA-HRP was electrochemically measured by chronoamperometry using 3,3,5,5-tetramethylbenzidine (TMB) as substrate. A LoD of 12 ng·mL −1 (30 nM) was obtained. The same authors [53] also investigated a similar architecture using EIS, with a LoD of 0.5 ng·mL −1 (1.3 nM).
Zamfir et al. described a label-free immunosensor based on magnetic nanoparticles for ochratoxin A [54]. The OTA antibodies were attached to these magnetic nanoparticles (MBs) and afterwards immobilized on a gold electrode pretreated with bovine serum albumin (BSA) under a magnetic field. The impedance variations due to the specific antibody-OTA interactions were correlated with the OTA concentration in the samples. The increase in electron transfer resistance was proportional to the concentration of OTA on a linear range between 0.01 and 5 ng mL −1 (25 pM-13 nM).

Antibodies
Prieto-Simon et al. described, in 2008 [51], two indirect electrochemical competitive enzyme-linked immunosorbent assays (ELISA) for ochratoxin A (OTA). Performances of polyclonal (pAb) and monoclonal (mAb) antibodies against OTA were compared, showing at least one order of magnitude higher affinity values when working with mAb. Alkaline phosphatase (AlkP) and horseradish peroxidase (HRP)-labeled secondary antibodies were evaluated. Similar limits of detection of 0.7 and 0.3 ng¨mL´1 (0.8 nM) were obtained for HRP-and AlkP-labeled immunosensors, respectively. Later, in 2009, Radi et al. [52] described an electrochemical immunosensor for detection of OTA on screen-printed gold electrodes modified by electroreduction of 4-nitrophenyl diazonium salts and anti-OTA. A competition between OTA and HRP-labeled OTA (OTA-HRP) for the immobilized antibodies was held. The activity of the bound OTA-HRP was electrochemically measured by chronoamperometry using 3,3,5,5-tetramethylbenzidine (TMB) as substrate. A LoD of 12 ng¨mL´1 (30 nM) was obtained. The same authors [53] also investigated a similar architecture using EIS, with a LoD of 0.5 ng¨mL´1 (1.3 nM).
Zamfir et al. described a label-free immunosensor based on magnetic nanoparticles for ochratoxin A [54]. The OTA antibodies were attached to these magnetic nanoparticles (MBs) and afterwards immobilized on a gold electrode pretreated with bovine serum albumin (BSA) under a magnetic field. The impedance variations due to the specific antibody-OTA interactions were correlated with the OTA concentration in the samples. The increase in electron transfer resistance was proportional to the concentration of OTA on a linear range between 0.01 and 5 ng¨mL´1 (25 pM-13 nM).
An indirect competitive enzyme-linked immunosorbent assay format was also constructed by immobilizing ochratoxin A on gold electrodes [55]. Electrochemical detection was performed by chronoamperometry using TMB and hydrogen peroxide with HRP as label. A LoD of 0.05 µg¨L´1 (0.12 nM) was achieved.
Additionally, boron doped diamond (BDD) was employed [56] for sensitive immunodetection of ochratoxin A using EIS. Abs were grafted through diazonium functionalization. The increase in electron-transfer resistance presented a sigmoidal shape versus log concentration of OTA, with a dynamic range between 7 pg¨mL´1 and 25 ng¨mL´1. A LoD of 7 pg¨mL´1 (17 pM) was obtained. These good results could be attributed to the low background and capacitive currents allowed by BDD compared to other more classical electrode materials.
Lastly, a label-free photoelectrochemical sensor was recently reported by Yang et al. [57], by assembly of CdSe nanoparticles on TiO 2 electrodes. Ascorbic acid (AA) was used as an electron donor for scavenging photogenerated holes under visible-light irradiation. The photocurrent response of the CdSe NPs-modified electrode was significantly enhanced as a result of the band alignment of CdSe and TiO 2 in electrolyte. OTA antibodies were immobilized on CdSe. Even if these values cannot strictly be compared to the previous ones because the transduction mechanism is not of the same nature, one may underline the good sensitivity of this technique, with a LoD of 2 pg¨mL´1 (5 pM) and a linear dependence with OTA concentration from 10 pg¨mL´1 to 50 ng¨mL´1.

Aptamers
Aptamers were also frequently reported in the literature, for ochratoxin A detection. A first electrochemical platform was described in 2010 using the DNA sequence 5 1 -GAT CGG GTG TGG GTG GCG TAA AGG GAG CAT CGG ACA-3 1 as aptamer probe [58]. In this article, the aptamer was immobilized on the surface of a glassy carbon electrode treated with PCl 5 -activated sulfonic acid groups able to bind amino-modified DNA strands. These primary strands were then sandwiched with two other DNA strands carrying methylene blue (MB) as redox label. Binding of the OTA target to the aptamer disassembled the DNA strands and lowered the MB redox current. The low LoD (30 pg¨mL´1, i.e., 80 pM) could be attributed to the functionalization strategy (direct coupling of the DNA strands on the electrode without any intermediate bulky groups). One year later, a Langmuir-Blodgett (polyaniline (PANI)-stearic acid (SA)) impedimetric OTA aptasensor was developed by Prabhakar et al. [59]. The DNA aptamer was covalently immobilized onto the LB monolayer deposited onto ITO plates. The dissociation constant (K D ) was found to be ca. 0.8ˆ10´7 M and the LoD was 0.1 ng¨mL´1 (0.25 nM), i.e., higher than with the direct coupling strategy of Kuang et al. Tong et al., in 2011 [60], described an original signal-on aptasensor based on exonuclease-catalyzed target recycling. To construct the aptasensor, a Fc-labeled probe DNA (5 1 -Fc-AAA GAT CGG GTG TGG GTG GCG TAA AGG GAG CAT CGG ACA-3 1 -SH), carrying the Fc label at its 5 1 -distal end was immobilized on Au electrode through 3 1 -thiol chemisorption, and hybridized with the complementary ochratoxin A aptamer. In the presence of ochratoxin A, formation of the complex resulted in the transformation of the probe DNA into a hairpin structure where the Fc group comes close to the electrode surface, therefore producing a current increase (signal-on). OTA was then decomplexed by the use of the exonuclease, which digest the aptamer, so that OTA is recycled and can come again to bind to another aptamer. Due to the amplification step, this original method gave a LoD of 1 pg¨mL´1 (2.5 pM). Zhang et al. [61] described a sensor based on a similar hairpin-shaped aptamer and site-specific DNA cleavage of restriction endonuclease TaqaI, with HRP instead of Fc. In their approach, the endonuclease is able to cut DNA (and release HRP in solution, far from the electrode surface) when OTA is absent; if OTA is present, there is no hairpin structure, and, therefore, the TaqI is inactive and HRP can be used as enzymatic amplification label. Due to this amplification, the reported LoD was 0.4 pg¨mL´1 (1 pM).
Lastly, Hayat et al. reported, in 2013 [62], a strategy for the fabrication of an electrochemical label-free (but needing a redox molecule diffusing in solution) aptasensor where long spacer chains of polyethyleneglycol were used to create diffusion channels for the redox probe, while aptamers acted as gate of the tunnels, depending on their conformation. The LoD was 0.12 ng¨L´1 (0.3 pM). The aptamer sequence was the same as Tong et al.

Peptides
Even though it has not yet been applied to electrochemical sensors, peptides were also described in a few works as a probe for ochratoxin A. Indeed, Bazin et al. [63] reported in 2013 the peptide NF04 (12-mer sequence: N 1 -Lys-Cys-Cys-Lys-Tyr-Tyr-Lys-Arg-Asn-Met-Tyr-Val-C 1 ) for specific binding to ochratoxin A, which was evaluated using a peptide-based ELISA assay. Heurich et al., described a computational approach to design peptide ligands for Ochratoxin A [64]. Two peptides were identified: N 1 -Cys-Ser-Ile-Val-Glu-Asp-Gly-Lys-C 1 (octapeptide) and N'-Gly-Pro-Ala-Gly-Ile-Asp-Gly-Pro-Ala-Gly-Ile-Arg-Cys-C' (13-mer). SPR analysis confirmed that the peptides bind to ochratoxin A with KD values of 10 to 15 µM. More recently, Solerai et al. [65] reported another peptide conjugated chitosan foam for detection of OTA using enzymatic chemiluminescence. There is therefore no reason for electrochemical OTA sensors based on peptide probes not to appear in the literature in the forthcoming years.

Antibodies
Estradiol (E2, Figure 11) immunosensors were extensively reported in the literature. In 1998, Padeste et al. [66] reported an amperometric immunosensor for estradiol using microperoxidase MP-11 antibody conjugates for amplification. However, in this preliminary work, the authors provided neither LoD nor sensitivity. Five year later, in 2003, Kuramitz et al. [67] described a 17β-estradiol-modified electrode for non-labeled immunoassay. When the antibody was bound to the estradiol self-assembled monolayer on the gold electrode surface, they observed a decrease of the electroactivity of the redox probes added in solution (BQ or Fe(CN) 6 3´{4´) , attributed to the steric hindrance between the antibody on the electrode surface and the redox marker. The authors detected 17β-estradiol, 17β-estradiol-6-one (1,3,5-estratriene-3,17β-diol-6-one 6-O-carboxymethyloxime) and diethylstilbestrol using competition for the antibody between the analyte in solution and 17β-estradiol immobilized on the electrode surface, and found that the affinities ranked as follow: 17β-estradiol-6-one > 17β-estradiol > DES. The LoD for 17β-estradiol was ca. 0.1 nM (the KD was estimated around 10´7 M). Biosensors 2016, 6, 7 14 of 21 Figure 11. Structure of estradiol (E2).
In 2005, Pemberton et al. [68] investigated an electrochemical immunosensor for estradiol on carbon SPE electrodes. A competitive immunoassay was performed using an AlkP-labeled estradiol conjugate. Electrochemical measurements were performed using differential pulse voltammetry (DPV) following the production of 1-naphthol from 1-naphthyl phosphate. The LoD was of 50 pg·mL −1 (0.2 nM). In 2006, Butler et al. [69] also reported very similar work. The sensor detected 17-beta-estradiol in an environmentally relevant range with a LoD of 0.25 pg·mL −1 (0.9 pM). The same year, Volpe et al. [70] reported a similar work with a LoD of 40 pg·mL −1 (0.15 nM) in bovine serum.
As for the other analytes, nanostructuration of the electrodes improved sensitivity and LoD. For example, Liu et al. (2009) [71] reported the use of AuNPs immobilized on gold electrodes in order to increase the surface area and therefore to improve sensitivity. For amplification, they used a HRP-labeled 17 beta-estradiol conjugate and benzoquinone (BQ) as redox reporter. They reported a LoD of 6 pg·mL −1 (20 pM). In 2010, the same authors compared SWV and EIS on a SAM-modified Au electrode carrying AuNPs vertical bar thiolated protein G-scaffold to facilitate the immobilization of an enhanced quantity of an almost uprightly aligned anti-estradiol capture antibody [72]. They used Fe(CN)6 3−/4− as redox probe. They obtained a LoD of 18 pg·mL −1 (65 pM) for SWV, and 26 pg·mL −1 for EIS (95 pM), e.g., not better than for their previous work. Again using nano/microstructuration, Martinez et al. [73] developed a competitive direct immunoassay using magnetic microspheres and an estradiol-HRP conjugate. The detection of estradiol was carried out using a competitive direct immunoassay format using anti-estradiol polyclonal Abs immobilized on 3-aminopropyl-modified magnetic microspheres. Estradiol present in the sample competed with an estradiol-HRP conjugate for the immobilized antibody. The current obtained from the enzymatic reaction was inversely proportional to the amount of estradiol in the sample. The LoD was 0.32 ng L-1 (ca. 1 pM).
More recently, in 2012, Kim et al. [74] described an impedimetric sensor by immobilizing a monolayer of estrogen receptor-alpha on Au electrodes. The binding of 17β-estradiol increased the electron-transfer resistance of the electrode which was directly monitored by EIS in the presence of Fe(CN)6 3−/4− . Despite the small size of estradiol compared to that of the immobilized antibody, the authors reported a KD of ca. 5 pM and a LoD of ca. 0.1 pM. Liu et al. [75] published an original copper monolayer-based sensor to construct a Cu/vertical bar protein G immunosensor for 17 beta-estradiol. Copper minimized the non-specific adsorption of biological molecules on the immunosensor surface and enhanced the binding efficiency between immunosensor surface and protein G. SWV was employed to monitor the electrochemical reduction current of ferrocenemethanol; SWV current decreased with the increase of estradiol-BSA conjugate concentration. The LoD was 12 pg·mL −1 (45 pM).
In 2012, Ojeda et al., described a more classical approach [76] based on the surface modification of a carbon SPE with grafted p-aminobenzoic acid (PABA), followed by covalent binding of streptavidin and immobilization of biotinylated anti-estradiol antibodies labeled with HRP. Hydroquinone was In 2005, Pemberton et al. [68] investigated an electrochemical immunosensor for estradiol on carbon SPE electrodes. A competitive immunoassay was performed using an AlkP-labeled estradiol conjugate. Electrochemical measurements were performed using differential pulse voltammetry (DPV) following the production of 1-naphthol from 1-naphthyl phosphate. The LoD was of 50 pg¨mL´1 (0.2 nM). In 2006, Butler et al. [69] also reported very similar work. The sensor detected 17-beta-estradiol in an environmentally relevant range with a LoD of 0.25 pg¨mL´1 (0.9 pM). The same year, Volpe et al. [70] reported a similar work with a LoD of 40 pg¨mL´1 (0.15 nM) in bovine serum.
As for the other analytes, nanostructuration of the electrodes improved sensitivity and LoD. For example, Liu et al. (2009) [71] reported the use of AuNPs immobilized on gold electrodes in order to increase the surface area and therefore to improve sensitivity. For amplification, they used a HRP-labeled 17 beta-estradiol conjugate and benzoquinone (BQ) as redox reporter. They reported a LoD of 6 pg¨mL´1 (20 pM). In 2010, the same authors compared SWV and EIS on a SAM-modified Au electrode carrying AuNPs vertical bar thiolated protein G-scaffold to facilitate the immobilization of an enhanced quantity of an almost uprightly aligned anti-estradiol capture antibody [72]. They used Fe(CN) 6 3´{4´a s redox probe. They obtained a LoD of 18 pg¨mL´1 (65 pM) for SWV, and 26 pg¨mL´1 for EIS (95 pM), e.g., not better than for their previous work. Again using nano/microstructuration, Martinez et al. [73] developed a competitive direct immunoassay using magnetic microspheres and an estradiol-HRP conjugate. The detection of estradiol was carried out using a competitive direct immunoassay format using anti-estradiol polyclonal Abs immobilized on 3-aminopropyl-modified magnetic microspheres. Estradiol present in the sample competed with an estradiol-HRP conjugate for the immobilized antibody. The current obtained from the enzymatic reaction was inversely proportional to the amount of estradiol in the sample. The LoD was 0.32 ng L-1 (ca. 1 pM). More recently, in 2012, Kim et al. [74] described an impedimetric sensor by immobilizing a monolayer of estrogen receptor-alpha on Au electrodes. The binding of 17β-estradiol increased the electron-transfer resistance of the electrode which was directly monitored by EIS in the presence of Fe(CN) 6 3´{4´. Despite the small size of estradiol compared to that of the immobilized antibody, the authors reported a K D of ca. 5 pM and a LoD of ca. 0.1 pM. Liu et al. [75] published an original copper monolayer-based sensor to construct a Cu/vertical bar protein G immunosensor for 17 beta-estradiol. Copper minimized the non-specific adsorption of biological molecules on the immunosensor surface and enhanced the binding efficiency between immunosensor surface and protein G. SWV was employed to monitor the electrochemical reduction current of ferrocenemethanol; SWV current decreased with the increase of estradiol-BSA conjugate concentration. The LoD was 12 pg¨mL´1 (45 pM). In 2012, Ojeda et al., described a more classical approach [76] based on the surface modification of a carbon SPE with grafted p-aminobenzoic acid (PABA), followed by covalent binding of streptavidin and immobilization of biotinylated anti-estradiol antibodies labeled with HRP. Hydroquinone was used as redox mediator. A LoD of 0.77 pg¨mL´1 (2.8 pM) was achieved. Kanso et al. [77] described two methods using MBs onto which synthetic estrogen derivatives were bound. Using a classical primary antibody and a HRP-labeled secondary antibody, SWV was used for detection of the enzyme product, with LoDs between 1 and 10 ng¨L´1 (3.6-36 pM) ( Figure 12). Still using nanoparticles, Li et al. [79] described an estradiol immunosensor based on graphenepolyaniline (GR-PANI) composites and graphene oxide (GO). HRP-GO-antibody conjugates were used to improve sensitivity, through a competitive immunoassay. The LoD was of 0.02 ng mL −1 (75 pM). Very recently (2015), again using nanostructures, Zhang et al. [80] reported a sensor using Cu2S nanostructures as labels. Cu2S NPs were covalently conjugated to BSA-estradiol. Even without using enzyme-label and acid dissolution, the Cu2S generated an intense electrochemical SWV signal giving a LoD of 7.5 pg·mL −1 (28 pM). The same authors [81] described a graphene-modified electrode modified with mesoporous Fe3O4 and loaded with Pb 2+ or Cd 2+ cations and estradiol antibodies. In this case, square wave adsorptive stripping voltammetry (SWASV) was used, in order to detect Pb 2+ or Cd 2+ . SWASV peak currents were proportional to the concentrations of estradiol down to a LoD of 0.015 pg·mL −1 (55 fM). Lastly, Jimena et al. [82] developed an immunosensor by immobilization of the anti-17 beta-estradiol monoclonal antibody on an Au disk modified with AuNPs on a cysteamine SAM. The limit of detection was 0.84 pg·mL −1 (3 pM). To conclude this section and as already remarked in Section 2.1 concerning antibiotics, the most pertinent strategy to reach the lowest limits of detection is to use adsorptive stripping methods [78,81].

Aptamers
Comparatively, aptamer-based estradiol sensors were less reported than their immunosensor counterparts and their use appeared more recently. Olowu et al. [83] described in 2010 an electrochemical DNA aptasensor for 17-beta-estradiol based on poly(3,4-ethylenedioxythiophene) (PEDOT) doped with AuNPs. Streptavidin was covalently attached to the electrode and the aptamer immobilized via streptavidin-biotin interaction. The electrochemical signal generated from the aptamer-target molecule interaction was monitored by standard CV and SWV using Fe(CN)6 3−/4− as redox probe. The signal observed showed a current decrease due to steric hindrance from the bound 17 beta-estradiol. The LoD was 0.1 nM. In 2012, Lin et al. [84] reported also a conventional transduction approach based on EIS with Au electrodes modified by a thiolated aptamer. Upon formation of the estradiol/aptamer complex on the electrode surface, the interfacial electron transfer resistance (Ret) increased, with a detection limit of 2 pM. This Ret increased was explained by the fact Chaisuwan et al. described also an original procedure [78] using estradiol-modified CdSe quantum dots (QDs). A bismuth-coated electrode was used for detecting the cadmium ions (Cd 2+ ) released during the acid dissolution of the QDs, after the recognition step that occurred in solution. The LoD was ca. 52 pg¨mL´1 (0.2 nM).
Still using nanoparticles, Li et al. [79] described an estradiol immunosensor based on graphene-polyaniline (GR-PANI) composites and graphene oxide (GO). HRP-GO-antibody conjugates were used to improve sensitivity, through a competitive immunoassay. The LoD was of 0.02 ng¨mL´1 (75 pM). Very recently (2015), again using nanostructures, Zhang et al. [80] reported a sensor using Cu 2 S nanostructures as labels. Cu 2 S NPs were covalently conjugated to BSA-estradiol. Even without using enzyme-label and acid dissolution, the Cu 2 S generated an intense electrochemical SWV signal giving a LoD of 7.5 pg¨mL´1 (28 pM). The same authors [81] described a graphene-modified electrode modified with mesoporous Fe 3 O 4 and loaded with Pb 2+ or Cd 2+ cations and estradiol antibodies. In this case, square wave adsorptive stripping voltammetry (SWASV) was used, in order to detect Pb 2+ or Cd 2+ . SWASV peak currents were proportional to the concentrations of estradiol down to a LoD of 0.015 pg¨mL´1 (55 fM). Lastly, Jimena et al. [82] developed an immunosensor by immobilization of the anti-17 beta-estradiol monoclonal antibody on an Au disk modified with AuNPs on a cysteamine SAM. The limit of detection was 0.84 pg¨mL´1 (3 pM). To conclude this section and as already remarked in Section 2.1 concerning antibiotics, the most pertinent strategy to reach the lowest limits of detection is to use adsorptive stripping methods [78,81].

Aptamers
Comparatively, aptamer-based estradiol sensors were less reported than their immunosensor counterparts and their use appeared more recently. Olowu et al. [83] described in 2010 an electrochemical DNA aptasensor for 17-beta-estradiol based on poly(3,4-ethylenedioxythiophene) (PEDOT) doped with AuNPs. Streptavidin was covalently attached to the electrode and the aptamer immobilized via streptavidin-biotin interaction. The electrochemical signal generated from the aptamer-target molecule interaction was monitored by standard CV and SWV using Fe(CN) 6 3´{4á s redox probe. The signal observed showed a current decrease due to steric hindrance from the bound 17 beta-estradiol. The LoD was 0.1 nM. In 2012, Lin et al. [84] reported also a conventional transduction approach based on EIS with Au electrodes modified by a thiolated aptamer. Upon formation of the estradiol/aptamer complex on the electrode surface, the interfacial electron transfer resistance (Ret) increased, with a detection limit of 2 pM. This Ret increased was explained by the fact that the negatively-charged aptamer probe hindered the electron transfer reaction of the redox probe Fe(CN) 6 3´{4´o n the Au electrode surface.
More recently, in 2014, Huang et al. [85] described a more original nanostructured sensor using flower-like vanadium disulfide, showing ordered nanosheets of several nanometers of thickness onto which the aptamers were immobilized. Very classically, Fe(CN) 6 3´{4´w as used as a diffusing redox probe and DPV was used to sense change in apparent diffusion coefficient when 17-beta-estradiol is associated with the grafted aptamers. A detection limit of ca. 1 pM was reported. Again exploiting nanostructuration, Ke et al. [86] reported a dendritic gold microstructure immobilized on boron doped diamond (BDD) electrode. Estradiol aptamers were immobilized on the surface of the dendritic Au/BDD electrode through Au-S interaction. Once estradiol reacts with aptamers, it enlarges the interfacial electron transfer resistance (sensed by EIS, again using the redox probe Fe(CN) 6 3´/ Fe(CN) 6 4´) on the electrode surface thus cause the increase of impedance) down to a LoD of 5 fM of estradiol. In 2015, Zhu et al. [87] reported a nanoporous conducting polymer electrode whose surface potential, modified upon interaction between estradiol and its aptamer, was still probed via EIS using Fe(CN) 6 3´{4´a s redox probe. Transduction was explained by the redistribution of negative charges in the electrode double-layer region when the aptamer adopts a folded conformation around the small neutral target molecule. The LoD was in the fM range. Additionally, in 2015, Fan et al. [88] used nickel hexacyanoferrate NPs as signaling probe, immobilized on the electrode. AuNPs were put on the NiFe(CN) 6 NPs, and the thiolated aptamer was immobilized. Upon formation of estradiol-aptamer complex, the interfacial electron transfer reaction (still probed via EIS using Fe(CN) 6 3´{4´) of the probe was lowered, resulting in the decrease of the electrochemical signal. The LoD was ca. 1 pM. At last, Jimenez et al. [89] described the selection of aptamers having dissociation constants for estradiol in the low nM range (17 nM for the best one). They described an EIS aptasensor based on the conformational change of the aptamer immobilized on a gold electrode by self-assembly. It was shown that the signal of the Fe(CN) 6 3´{4´r edox probe was optimized upon hybridization of the aptamer with a short complementary sequence at some specific sites, which was attributed to the more significant conformational change of the aptamer/DNA duplex than the single-stranded aptamer upon binding with the target. The LoD was 0.90 ng¨mL´1 (3 nM). All these results are summarized in Table 5. LoDs appear to be lower for estradiol than for the other targets dealt with in this review. Conversely to pollutants or other toxic small organic molecules, antibodies are available and lead to low LoDs, even if aptamers allow to obtain the lower ones (for similar transduction methods). If we compared LoD for a same probe, similarly to the others targets discussed above, the enzyme-amplified strategy applied with aptamer probes do not lead to significant improvements; this a obviously not the case for antibodies, for which enzyme-labelling has long proven its efficiency. Concerning the sensing strategies used with aptamers, EIS coupled to the Fe(CN) 6 3´{4ŕ edox probe is dominant. Efforts could be made to improve this step.