Perspective—Electrochemical Sensors to Monitor Endocrine Disrupting Pollutants

Endocrine disrupting compounds are ubiquitous in the environment and have been implicated in a variety of health problems. Despite awareness of their dangers at trace concentrations, options for analytical detection are limited, as these chemically dissimilar compoundsarecategorizedbytheiractivityratherthantheirstructure.Electrochemicalsensorshavemadesigniﬁcantstrides,bringingtestingtopoint-of-exposuresettingsandimprovingdetectionfromcomplexmatrices.Withadvancesindisposablesensorsandapushtowardactivity-basedratherthanstructure-baseddetection,theﬁeldispoisedtotransformmonitoringofendocrinedisruptorsforenvironmentalsurveillanceandelucidationofbiologicalmodesofaction.©TheAuthor(s)2019.PublishedbyECS.ThisisanopenaccessarticledistributedunderthetermsoftheCreativeCommonsAttribution4.0License(CCBY,http://creativecommons.org/licenses/by/4.0/),whichpermitsunrestrictedreuseoftheworkinanymedium,providedtheoriginalworkisproperlycited.[DOI:10.1149/2.0242003JES]

Endocrine disrupting compounds (EDCs) interfere with natural hormonal feedback loops that control many aspects of development, often at sub-nanomolar concentrations. 1 Numerous synthetic chemicals present as trace environmental contaminants are classified as EDCs, including xenoestrogens, polychlorobiphenyls (PCBs), phthalates, and perfluoroalkyl substances (PFAS) (Figure 1a). Importantly, these compounds are defined by their biological activity, not their specific chemical structure. EDCs have been implicated in many diseases, namely cancer, metabolic disorders (including obesity), and reproductive disorders. 2,3 These chemicals are known to similarly affect wildlife, causing lasting damage to reproductive systems upon exposure. Consequently, regulatory agencies including the United States Environmental Protection Agency (US EPA) 4 and the World Health Organization (WHO) 5 have set exposure limits and minimum allowable contaminant levels (MCL), which are often very low due to the sensitivity of native hormone systems to these compounds. Enforcement of MCL standards and elucidation of the biological effects of known endocrine disruptors in addition to newly-introduced chemicals require accurate and sensitive detection at biologically relevant concentrations.
Rapid, sensitive EDC detection is essential for environmental monitoring. To date, the majority of detection efforts have focused on biological and chemical assays for pollutant sensing, many of which require laboratory settings. 2 As the number of common chemicals that are found to interfere with endocrine signaling increases, 5 there is an urgent need for improved detection methods that are portable, rapid, and inexpensive, and that elucidate modes of action. Electrochemical sensors exhibit high sensitivity and are amenable to low-cost, miniaturized designs (Table I) and wearable sensors for real-time human health monitoring. 6 Importantly, electrochemical sensors are sufficiently flexible for both periodic field testing and continuous, remote monitoring.
These techniques are sensitive but often require centralized facilities and trained personnel to run. Fluorescence detection, in contrast, is portable and has been successfully used to detect EDCs by native emission (phenolic xenoestrogens including BPA 8 ), by derivatization with a fluorescent label (BPA, 9 PFAS 10 ), or by energy transfer to fluorogenic substrates (PBCs). 11 Derivatization and coupling with chromatographic methods improve sensitivity 13 but impose cost and equipment constraints that limit the point-of-exposure applicability of these methods. Immunoassays are well-suited for EDC detection because EDCs naturally interact with biomolecules, but these assays are often relatively slow and expensive. Kits using competitive enzyme-linked immunosorbent assays (ELISAs) are commercially available for PCBs, phthalates, and BPA. 19 These kits involve an immobilized antigen bound to an antibody against the target. The target is quantified by a signal decrease upon competition with the labeled analyte. Immunoassays are sensitive but require multiple steps and longer incubation times than are practical for real-time monitoring. The majority of analytical methods are impractical for real-time monitoring of trace concentrations in the field, and few provide biological insight into the modes of action of these chemicals.
Electrochemical detection of endocrine disruptors.-Electrochemical detection methods are rapid, sensitive, and require small sample volumes. Additionally, electrochemical sensors are readily integrated into small, portable devices, 20 enabling their point-ofexposure deployment. Many are compatible with complex background matrices, 21 reducing the need for sample pre-treatment and simplifying sampling protocols. Because of the versatility of such sensors, many strategies have been adopted for electrochemical EDC detection. Electroactive EDCs, including xenoestrogens and phthalates, can be directly detected (Figure 1b). 22 Electrochemically-inactive EDCs, including PCBs and PFAS, can be monitored by changes to electroactive species that are affected by their presence. 23,24 Biorecognition with an antibody, aptamer, or hormone receptor integrated into an electrochemical sensor facilitates specific detection and can provide insight into biological modes of action of the EDC. The majority of recent advances in electrochemical detection of EDCs have relied on novel nanostructuring of materials to enhance selectivity for specific analytes and enable lower detection limits at times approaching biological relevance. 25 A lesser-used strategy is to use the bioactivity of endocrine disrupting compounds to monitor the total biological activity of a sample, the readout of interest for most point-of-care monitoring. Below we discuss analyte-specific and activity-based approaches for the most prevalent EDCs and argue that the field is poised to further develop the latter technologies with the potential to vastly improve public health monitoring. BPA and other phenolic xenoestrogens.-BPA is a component of many plastics and resins often used for food storage, making it especially problematic as an EDC. [26][27][28] It is also electroactive, undergoing a 2-electron, 2-proton oxidation, which can be directly monitored. However, direct detection of BPA by oxidation is limited by electrode fouling and a high overpotential. Consequently, many sensors incorporate carbon nanotubes (CNTs), 29 graphene, or metal nanoparticles (NPs) to increase the electroactive surface area and lower the overpotential. Selectivity is improved by attaching a recognition element to the electrode, most often molecularly imprinted polymers (MIP) or aptamers. 30 Fouling can also be reduced by simply adding surfactants to samples with vigorous stirring. 31 Most methods exhibit optimal performance at pH 8-8.5, well below the pKa of BPA. 22 Though BPA can be detected directly, the incorporation of a biorecognition agent can increase the specificity and sensitivity of a device. We developed an electrochemical aptasensor based on a BPA-specific DNA aptamer. 32 A turn-off sensor was developed based upon changes in the permittivity of the DNA layer in the absence and presence of BPA. Without BPA, ferricyanide diffused to the surface, providing a large signal by differential pulse voltammetry (DPV). As BPA bound to the aptamer, the surface became more shielded and the signal decreased. The limit of detection (LOD) of this sensor was 10 nM, and it was highly specific for BPA; the sensor did not respond to other structurally similar bisphenol EDCs. By incorporating a specific biorecognition element, electrochemical sensors for bisphenol compounds can be made more specific and do not require the harsh conditions of direct oxidative detection.
Enzymatic detection has also been used extensively for phenolic xenoestrogen sensing. Because the enzymes employed are catalytic, they provide inherent signal amplification, allowing for highly sensitive detection of these compounds. Tyrosinase, a copper-based enzyme that catalyzes the oxidation and hydroxylation of phenols to orthoquinones, is often used in electrochemical sensors due to its affinity for BPA, the large signal produced, and the ease with which it is incorporated into materials. 33 Moscone et al. incorporated tyrosinase into carbon pastes with the addition of single wall carbon nanotubes. Their amperometric device had a linear range of 0.1-12 μM, a response time of 6 minutes, and an LOD of 0.02 μM for BPA. 34 Chao et al. reported an even lower detection limit (5 nM) and expanded dynamic range (0.01-1 μM) for BPA using a tyrosinase/carbon fiber paper (Tyr-CFP) electrode. 35 Tyrosinase mixed with polyvinyl alcohol functionalized with pyridinium methyl sulfate (PVA-SbQ) was coated on a CFP electrode, followed by crosslinking and photopolymerization. Despite improvements in sensitivity, CFP is relatively fragile and expensive, which limits its utility in point-of-exposure devices. As seen with the aforementioned examples, a tradeoff is often required for tyrosinase-based devices between sensitivity and simplicity, making the application of the device (field testing versus laboratory monitoring) an important consideration.
Many other enzymes have similarly proved useful for phenolic xenoestrogen detection. Laccase is a copper-based oxidase that oxidizes diphenols, polyphenols, aminophenols, and polyamines. Lepore et al. developed a laccase-based sensor for BPA. They layered a screenprinted electrode successively with carbon black (to improve activity), thionine (as a mediator), laccase (the oxidative enzyme), and Nafion (to seal the modified electrode and avoid enzyme leaking). 36 BPA was detected in water over a linear range of 0.5-50 μM with an LOD of 0.2 μM. It was also successfully detected in spiked tomato juice at 10 μM. Importantly, as this sensor was developed on a screen-printed electrode and tested in realistic samples, it has the potential to be a useful method of monitoring BPA in point-of-exposure settings. Comparable performance was obtained using human cytochrome P450 2C9 (CYP2C9) as an oxidase. 37 When CYP2C9 was immobilized in a polyacrylamide hydrogel film on a glassy carbon electrode, BPA was detected over a linear range of 1.25-10 μM with a detection limit of 0.58 μM. However, because this enzyme is promiscuous, the CYP2C9 sensor has limited selectivity for BPA. Enzymatic detection of phenolic xenoestrogens can afford selectivity unavailable with direct detection of bisphenols, though this selectivity is highly dependent on the specificity of the enzyme. Critically, enzymatic methods have inherent signal amplification, which is not always available for biorecognition-based platforms. Overall, numerous complementary strategies are available for bisphenol detection, and often, the most suitable sensor for a particular application is dependent on that application.
Phthalates.-Phthalates are broadly-used small-molecule plasticizers. These compounds are so pervasive that the Centers for Disease Control (CDC) found phthalate metabolic products in urine from the majority of Americans tested, 38 making detection and monitoring of these compounds vital. Phthalates can be directly reduced at negative potentials, which necessitates the use of mercury or silver amalgam electrodes. The application of these electrode materials can be impractical for real-world detection. 26 Thus, more practical platforms have been developed, especially for the most common phthalate plasticizer (di(2-ethylhexyl)phthalate (DEHP)). DEHP was detected by frequency response analysis (FRA) using a thin film interdigitated electrode. 39 Selectivity was imparted by molecularly imprinting DEHP in a polyacrylamide coating. Despite this selective design, the sensor's linear range (0.1 to 100 ppm) fails to meet the guidelines Figure 1. Electrochemical detection strategies for endocrine disrupting compounds (EDCs). a) Common EDCs. b) BPA can be detected through the direct electrochemical oxidation of its phenolic groups. c) A competition-based immunoassay for PCBs. An enzyme-linked PCB (green with gold star) is prebound to surface. Following sample exposure, the enzyme-linked signal decreases based upon the PCB concentration in the analyte solution. d) A sandwich assay based on bacterial binding to an electrode is used to detect EDCs. E. coli modified with a mammalian estrogen receptor only bind to gold electrodes modified with a monobody in the presence of xenoextrogen pollutants. This assay is a measure of the bioactivity of an analyte solution rather than the concentration of a particular EDC.
for safe drinking water. Unfortunately, other voltammetric methods also fail to meet useful limits of detection, performing as much as six orders of magnitude worse than chromatographic methods. 26 Because of the limited number of electrochemical sensors available for phthalates, this area is ripe for the development of portable, specific, and sensitive sensors.
PFAS.-Perfluoroalkyl substances (PFAS) are increasingly cited as the main concern for EDC pollution because of their pervasiveness and persistence. Perfluorooctane sulfonic acid (PFOS) and perfluorooctanoic acid (PFOA) are on the EPA list of emerging contaminants, 40 and animal studies increasingly implicate them in endocrine disorders. The properties that lend these compounds their utility -hydrophobicity, oleophobicity and stability -also present challenges in their analysis and remediation. They are not electrochemically active and therefore require indirect detection methods in electroanalytical applications. Recent work by Luo et al. exploited the surfactant properties of these compounds to detect PFOS and PFOA at 30 μg/L. 27 The presence of a surfactant affects electrochemical bubble nucleation by reducing the surface tension of the gas-liquid interface, thereby decreasing the gas supersaturation level required for a bubble to form. PFAS concentrations were determined by a decrease in the bubble nucleation barrier, as measured by changes in peak current in HClO 4 solution. This method was robust to the presence of PEG at 1000-fold excess, lysozyme, and humic acid, and the platform exhibited a wide linear range of 0.1 mg/L-10 g/L. However, the system required pre-concentration to meet the EPA drinking water guideline of 70 ng/L. Measurements depend on the surfactant activity of each compound, yielding native LODs of 72 and 160 nM for PFOA and PFOS, respectively. Because the dangers of PFAS have only recently emerged, effective point-of-exposure detection strategies remain extremely limited. Additionally, the mechanism of action and biological target of these compounds remain unclear, emphasizing the importance of platforms that can elucidate these pathways by developing sensors based on the likely human hormone receptor targets themselves. 32,41 The unreactive nature of these compounds demands creative strategies for detection at trace levels.
PCBs.-Polychlorinated biphenyls (PCBs) are used in many items with which we come into regular contact, including lubricants, plasticizers, and components of electrical systems. PCBs are not inherently electrochemically active. Thus, detection often relies on using these chemicals to perturb the interaction of an electroactive molecule with an electrode. In a targeted approach, a photoelectrochemical (PEC) aptasensor was developed for one member of the PCB family, PCB77, using an aptamer attached to N-doped TiO 2 nanotubes (NT). Singlestranded DNA modified with CdS quantum dots (DNA-CdS) was initially hybridized to the aptamer. Upon addition of PCB77, the DNA-CdS was displaced, resulting in a signal decrease. 42 Though this aptasensor is sensitive, the ability to detect only a single compound can be limiting for environmental monitoring where bioactivity may be more relevant than targeted analyte detection.
As an alternative strategy, several methods have been developed to detect PCBs as a class of chemicals without differentiating between specific PCBs. In one example, PCBs displaced a guest molecule in a host-guest complex. An electroactive ferrocene acted as the guest molecule to be displaced in a host-guest complex with beta-cyclodextrin that was attached to a modified pyrolytic graphite electrode. 43 PCBs were indirectly measured by the loss of a ferrocene signal, as determined by DPV. The sensor exhibited selectivity for PCBs in the presence of other EDCs. The detection limit, 0.5 pM, is well below the EPA's 2001 enforceable MCL, and PCB concentrations at the MCL were detected in lake sediments, demonstrating the utility of the sensor for rapid screening of environmental samples. A more recent example of a beta-cyclodextrin-based sensor for planar PCBs demonstrated that other redox indicators such as methylene blue can effectively be used, although the detection limits were orders of magnitude higher than the previous sample. 44 Similarly, an immunosensor was designed to detect PCBs in complex matrices, again without differentiation between specific chemicals. Screen-printed electrodes were modified with PCB-specific antibodies pre-bound with enzymelabeled PCB. Upon analyte binding, enzyme-PCB was competed off the electrode (Figure 1c). 45 Importantly, this sensor is sensitive to a group of chemicals (PCBs) in complex solutions including food matrices, without requiring independent monitoring of each unique chemical. The reported linear range of detection was 0.01−50 ppm in standard solutions, with successful detection at the tolerance levels established by the European commission. These platforms represent a key improvement in the electrochemical detection of EDCs because they can detect families of compounds from complex solutions, rather than individual chemicals.

Receptor-based bioactivity sensor for xenoestrogens.-Many
chemically dissimilar compounds act on a single hormone pathway. Thus, monitoring concentrations of single analytes can be less important than determining the total bioactivity of a mixture for a particular pathway. We recently reported an electrochemical biosensor for the class of chemically dissimilar molecules that interfere with native estrogen signaling (xenoestrogens) by binding to the human estrogen receptor ERα (Figure 1d). 46 The platform is based on an electrochemical sandwich assay, with one half being ERα expressed on the surface of E. coli. The second half is a small, fibronectin-based monobody immobilized on an electrode that binds to ERα only in the presence of a xenoestrogen. This sensor successfully detected sub-ppb levels of the native hormone, estradiol, as well as the activity of nM concentrations of combinations of xenoestrogens. Because the sensor reports on the total bioactivity of a solution, combinations of chemically dissimilar molecules were quantifiable over five orders of magnitude of concentration. Importantly, this feature also enabled the determination of the bioactivity of an unknown compound; significant estrogenic activity was detected from a microwaved BPA-free baby bottle, equivalent to 100 nM estradiol. As the sensor was developed on disposable electrodes, it is low-cost and portable. The E. coli-bound hormone receptor is most stable when the cells on which it is expressed are lyophilized. Those E. coli are stable when stored frozen for at least six months, which would enable storage and transport of these sensors for point-ofexposure monitoring. As regulatory agencies move away from simple quantification to studying the bioactivity of pollutants and analyzing environmental samples for unknown targets, assays based on bioactivity will become increasingly important.

Future Needs and Prospects
As the list of prevalent chemicals that act on native hormone pathways increases daily, the need for novel and creative detection strategies becomes ever more pressing. EDC detection comes with specific challenges not associated with other groups of small molecules, as they are classified based on their biological activity rather than their chemical similarity. Disparate structures that all target a single biological pathway pose challenges for conventional detection strategies, and further complications arise because EDC monitoring is so varied and application-dependent. Monitoring ranges from batch or continuous monitoring of environmental samples, in-home monitoring of consumer products, fundamental studies of the biological targets of EDCs, to, most challenging, evaluating whether EDCs are present without knowing the specific compounds present.
Moving forward, three key areas necessitate significant efforts in sensor development for EDCs. Electrochemical sensors are ideally suited to provide improvements in all of these areas, including: 1) environmental and in-home monitoring of specific compounds, 2) monitoring the bioactivity of polluted systems with potentially unknown contaminants, and 3) elucidation of the modes of action of emerging contaminants.
Improved sensors for point-of-exposure EDC detection are key, especially as contamination occurs in such a broad spectrum of locations, from public lands to private households. EDCs are increasingly found in household products, from bisphenols in kitchen plastics and the lining of aluminum cans to PCBs in household electronics. With the omnipresence of these chemicals comes a need for in-home testing. Thus, developing sensors for existing contaminants that are amenable to commercialization is an area primed for significant growth in the next few years.
Similarly, there are limited examples of strategies for effective, sensitive, inexpensive sensors for newly-emerging contaminants, including phthalates and PFAS. In the case of phthalates, their use in so many common products points to their pervasiveness as environmental pollutants. However, current strategies to monitor these compounds remain limited. Portable, inexpensive sensors will be invaluable for remediation efforts, especially as governments and businesses increase these efforts. Current detection strategies 47 are insufficient for the broad implementation that is required to detect these compounds and elucidate their biological modes of action. As direct electrochemical detection of these classes of compounds is either impractical (phthalates) or impossible (PFAS), electrochemical strategies will require additional creativity, and the bioactivity of these compounds suggests that biomolecules can be successfully exploited in sensor development. With the rate at which industrial compounds are found to be endocrine disruptors, we expect new targets to emerge in the future that will similarly require unique sensors for environmental monitoring.
As it is increasingly apparent that we do not know all of the chemicals in our environment that may act as endocrine disruptors, sensors based on bioactivity rather than specific concentrations of chemicals become imperative. Though monitoring specific pollutants will always be important, complementary strategies that do not require knowledge of the components of a polluted system are equally important. Expanding on the electrochemical sandwich assay developed for estrogenic compounds will be an important thrust in the next few years. Developing equivalent systems for xeno-testosterones and chemicals that target the thyroid hormone receptor PPARγ will be the next steps in the development of a screening platform based on a bioactivity panel for human hormone receptors. Similarly, as many of these compounds have been found to also target the brain, it is vital to develop sensors that report on the neurological activity of EDCs. Because electrochemical platforms are modular and readily multiplexed, they are ideal for the development of such screens.
Finally, as researchers continue to develop new compounds that could interfere with native hormonal signaling, platforms to elucidate their mode of action will be essential. In the near future, such platforms will illuminate the biological targets of compounds that have only recently been identified as problematic, such as PFAS. We envision such electrochemical sensors being similar in nature to those described above for monitoring total bioactivity. However, as these sensors will likely be limited to laboratory settings, more complicated protein targets can be incorporated for screening. Long term, this technology will be essential as a pre-screen for industrial and pharmaceutical compounds prior to their broad implementation to ensure new compounds have no adverse effects on endocrine systems.
Overall, environmental monitoring of endocrine disruptors is a rapidly growing field. Electrochemical sensors offer low cost, versatility, portability, and modularity unavailable with other analytical methods. As creative methods to ensure biologically-relevant sensing improves, electrochemical sensors for EDCs will move to the forefront of detection strategies.

Conclusions
Endocrine disrupting compounds pose major threats to human health, and monitoring them with knowledge of their biological targets is key for efficient containment. Electrochemical sensors exist for many of these compounds, with both direct and biorecognition-based detection finding success. However, as the number of newly-identified endocrine disruptors is constantly increasing, there is a pressing need for sensors to: 1) directly detect specific chemical contaminants in point-of-exposure settings, 2) determine the total bioactivity of a combination of compounds based on their bio-interactions, negating the need for knowledge of the specific components, and 3) elucidate modes of action of newly-identified endocrine disruptors and evaluate the potential biological implications of new industrial chemicals. Better electrochemical sensors will be vital to the success of these efforts. Technological advances to address each need will significantly improve environmental monitoring and enable remediation.