Multianalyte Physiological Microanalytical Devices

Abstract

Advances in scientific instrumentation have allowed experimentalists to evaluate well known systems in new ways as well as gain insight into previously unexplored or poorly understood phenomena. Within the growing field of multianalyte physiometry (MAP), the microphysiometer development is building, instruments capable of electrochemically measuring changes in the concentration of various metabolites in real time. By simultaneously quantifying multiple analytes, these devices have begun to unravel the complex pathways that govern biological responses to ischemia and oxidative stress while contributing to basic science discoveries in bioenergetics and neurology. Patients and clinicians have also benefited from the highly translational nature of MAP, and the continued expansion of the repertoire of analytes that can be measured with multianalyte microphysiometers will undoubtedly play a role in the automation and personalization of medicine. This is perhaps most evident with the recent advent of fully integrated sensor arrays that can continuously monitor changes in analytes linked to specific disease states using non-invasive sensors and can deliver a therapeutic agent as needed without the need for patient action.

1. Introduction

The development of analytical devices incorporating sensors for multiple analytes has enabled new studies of physiological systems. For example, although electrochemical glucose sensors(1–3) have a well-established clinical presence in monitoring diabetics (1) and pre- and post- op surgical patients (4), fundamental studies on aerobic and anaerobic respiration are better accomplished through the detection of lactate, oxygen, and pH in addition to glucose (Figure 1) (5). When observing physiological effects with complex methodologies varying from monoculture microfluidic devices to clinical samples, sensing strategies often need to accommodate small devices, moderate-throughput experiments, and limited volumes of patient samples.

Figure 1.

Schematic of a cell showing cellular bioenergetics pathways. Metabolites commonly detected with multianalyte physiometry (lactate, glucose, oxygen, and acid) are highlighted. Abbreviations: ADP, adenosine diphosphate; ATP, adenosine triphosphate; G6P, glucose 6-phosphate; NAD, nicotinamide adenine dinucleotide; NADH, nicotinamide adenine dinucleotide hydrate; TCA, tricarboxylic acid. Adapted with permission from Reference 10. Copyright 2006, Elsevier.

The subfield of multianalyte physiometry (MAP) has emerged to broaden the understanding of physiological pathways with multianalyte analysis using either a sensor array or single multiplexed detector to probe complex biochemical processes in higher dimensions (spatial and temporal) of phase space. Researched applications of electrochemical MAP sensors include observations of biological and chemical responses to stimuli, determination of the health status of organs on chips (OoCs), and development of novel clinical and point of care (POC) devices. When approaching the more complex task of detecting multiple analytes in living organisms, implantable ultramicroelectrodes are desirable due to decreased disruption in animal models and fast scan cyclic voltammetry (FSCV) minimizes adverse effects to the subject by minimizing the diffusion layer (6, 7). MAP devices’ breadth of applications includes fundamental cellular studies, clinical and point of care testing platforms, and in vivo/in situ research.

2. In Vitro Microfluidic Studies

Multianalyte physiological microanalytical devices currently encompass many different types of experimental methodologies and techniques. Microfluidic devices that focus on electrochemical measurements within the device are the focus of this review, as the electrode sensors are built within these self-contained systems. Examples of the different types of these devices include microphysiometers, microclinical analyzers, microfluidic electrophysiology, multielectrode arrays, and microchip electrophoresis coupled to electrochemical detection. Each of these sub-types is reviewed in the sections below.

2.1. Multianalyte Microphysiometer

Conventional microphysiometry was developed by Harden McConnell and co-workers leading to the Cytosensor Microphysiometer from Molecular Devices, an instrument that measured the extracellular acidification (i.e. pH changes) produced by the metabolism of approximately 100,000 to 1,000,000 live cells in a 3 μL microfluidic chamber using a light-addressable potentiometric sensors (LAPS) (Figure 2a). The Cytosensor has been modified (Figure 2b) into the multianalyte microphysiometer (MAMP) for detecting glucose, oxygen, and lactate in addition to pH inside the cell chamber. These additional analytes are detected amperometrically with immobilized glucose oxidase (GOx), immobilized lactate oxidase (LOx), and O₂ is directly detected (Figure 2b) (5, 8–16). Investigations with the MAMP instrument include the metabolic effects on cellular bioenergetics of nutrient concentrations (12), chemical toxins (5, 8–11), harmful proteins (10, 11) non-invasive fluorescent chemical probes (13), and oxidative burst (14–18). Major shifts in cellular bioenergetics observed in the MAMP demonstrate shifts in extenet of utilization of aerobic or anaerobic respiration pathways (5, 8).

Figure 2.

Multianalyte physiometers. (a) Side cross section and (b) bottom view of a CytosensorTM modified into a multianalyte microphysiometer by electrode addition. The four added platinum electrodes include three working electrodes and one counter electrode. The working electrodes detect glucose, lactate, and oxygen. Panels a and b adapted with permission from Reference 5. Copyright 2004, American Chemical Society. (c) Samples are stop-flowed into the device, and pH is measured through light-emitting diode (LED)-illuminated light-addressable potentiometric sensors (LAPS). Panel adapted with permission from Reference10. Copyright 2006, Elsevier. (d) Schematic of a multianalyte physiometer based on a glass chip that combines a cell cultivation chamber, microfluidics, and metabolic monitoring. Oxygen and pH are measured in the cell culture area, and biosensors for lactate and glucose are connected downstream by microfluidics. The wafer-level fabrication features thin-film platinum and iridium oxide microelectrodes on a glass chip, microfluidics in an epoxy resist, a hybrid assembly, and an on-chip reference electrode. Panel adapted with permission from Reference 21. Copyright 2014, Royal Society of Chemistry.

Multiplexing cellular bioenergetics and insulin detection can offer valuable information on the complicated physiology and metabolism of primary islets. For example, excised pancreatic islet physiology and metabolism were studied by combining an insulin selective electrode with electrodes for the previously mentioned analytes in the MAMP. This allowed the time-dependent study of insulin release resulting from a hyperglycemic environment (19).

Cellular bioenergetics studies of neurons using the MAMP has also revealed a strong correlation between neuronal adaptation and survival by developing a new model of transient ischemic attack (TIA) (12). Metabolic adaptation to stress is critical for cell survival but poorly understood. The metabolic compensation and survival of nutrient-deprived neurons relies on neuronal-glial chemical and physical communication. In this model of transient ischemic attack (TIA), primary neurons primed with short deprivations of oxygen and glucose in the days leading up to an otherwise lethal deprivation achieved increased survival rates over unprimed neurons (20). It was found that after priming the neurons, cellular ATP and neuronal oxygen consumption increased above baseline. After treating these primed neurons with an oxygen deprivation triggering an anaerobic pathway shift that would normally be lethal to the neurons, these primed neurons recovered normal lactate production in 30 min signalling a return to aerobic sufficiency. The rapid increase in aerobic respiration (monitored via oxygen consumption) observed after priming, coupled with the increase in cellular ATP indicates that the protective pathways in this ischemic model include the immediate increase in production of energy stores through aerobic respiration. This increase in energy production may help to prevent cellular starvation upon subsequent stress. This provides the first dynamic measurement used to identify essential events mediating neuronal injury in vitro and can aid in the identification of predictive biomarkers in injury. These microphysiometry data reveal that the greatest single predictor of neuronal survival is extracellular acidification; however, lactate levels, which are currently a key clinical indicator of injury, were not in fact, correlated with neuronal cell fate (12, 21).

2.2. Microclinical Analyzer (MCA).

Unlike in multianalyte microphysiometry where the sensors are embedded into the microfluidic cellular chamber, in a microclinical analyzer, the electrochemical measurements are in performed in separate chamber downstream. Thus, MCAs measurements can be multiplexed between multiple organs on a chip using a valve to select which organ effluent is to be measured. Thus, the MCAs trade off temporal and spatial resolution for increased resistance to biofouling of the electrochemical sensors. Our MCA is a self-calibrating device consisting of a microfluidic pump and valve that directs calibrant and sample solutions to a 26μL sample chamber where an array of platinum screen-printed electrodes (SPEs) amperometrically detects hydrogen peroxide produced by spatially separated enzyme films and potentiometrically detects pH via open circuit potential (OCP) shifts using electrodeposited iridium oxide. SPEs are robust, highly reproducible, and low-cost, resulting in a device that is easily customizable with interchangeable sensors capable of measuring glutamate, acetylcholine, calcium, potassium and/or sodium (18, 54). Custom multichannel potentiostats were created to accommodate simultaneous MCA measurements (56). Our investigations with the MCA instrument have focused on monitoring tissues and organs on chips (OoCs), (52, 53).

Other researchers have developed similar MCAs that are useful in toxicological testing of healthy or cancer cells (57, 58). Combined with other metabolites of bioenergetics, oxygen sensing inside their 10μL device increases the likelihood that changes are a result of cellular aerobic respiration. Running multiple simultaneous experiments with multi-sensor microsystems allows for a moderate-throughput screening of potential drug candidates (58). An electrolyte-focused MCA has been developed with four ion selective and three partially selective electrodes, specifically selective for potassium, sodium, hydrogen, calcium, amines, cations, and anions. These electrolyte MCAs are primarily concerned with signaling flux from ionic gradients. While direct pathway elucidations are not always exact, using chemometric analysis, training sets can be used to define a particular physiological state of interest. The combination of electrolyte sensing and multiplexed analysis was used to determine the ratio of dead to live cells in a cancer therapeutics screening (57).

2.3. Electrophysiological Event Monitoring

Exocytosis of redox active species can be measured potentiometrically or amperometrically (22) and has led to numerous discoveries in vesicle transport. For example, the measurement of the relative concentrations of specific analytes, including octopamine, following release have been reported (23). While differentiating the molecules released is not always possible, the timing and frequency of events marked by the release of neurotransmitters can be recorded (24) and microelectrode arrays allow for both single cell and population analysis (25). Additionally, using experimentally determined parameters, the initial pore size before full release has been mathematically modeled (24). Nanopipette potentiometric electrodes and carbon fiber nanoelectrodes can be used separately or in combination to measure postsynaptic potential shifts and neurotransmitter release events (22, 24, 26). Measurements of synaptic events were performed in a microfluidic device which co-cultured superior cervical ganglion neurons and smooth muscle cells (Figure 2c) to confirm that synaptic events mimic in vivo conditions, in addition to improved axon orientation (26).

2.4. Multiplexed Reactive Oxygen and Nitrogen Species Sensing.

Oxidative stress is marked by the production of reactive oxygen species (ROS) and reactive nitrogen species (RNS), specifically hydrogen peroxide (H₂O₂), nitric oxide (NO^-), nitrite (NO₂^-), and peroxynitrite (ONOO^-). By amperometrically measuring at four different potentials, the concentrations of ROS and RNS are determined with a physiological model consisting of four linear equations delimiting current contributions of each compound. The detection of ROS and RNS concentrations has been used to quantify anti-oxidant capacity, to help determine basic cellular function of macrophages, and to dcompare the effectiveness of cancer therapies. The antioxidant capacity of a superoxide dismutase mimicking manganese complex (27) was determined with amperometric detection of ROS/RNS produced by macrophage cells. Utilizing the redox nature of ROS/RNS reactions, a new cell-free method for determining antioxidant capacity showed that trolox, ascorbic acid, gallic acid, and caffeic acid, compounds with known antioxidant properties, electrochemically simulated the redox effect of ROS in a microfluidic device (28). In a study of amacrophage’s ability to control ROS/RNS release and implement protective measures, two platinum electrodes were used to observe dynamic intracellular and extracellular ROS/RNS concentrations. By inserting and sealing one electrode into a macrophage cell while simultaneously measuring outside the same cell, ROS/RNS leakage through phagolysosmes and neutralization to avoid premature oxidative damage were observed (29). Experimental drug molecules of the ferrocifen class were tested on two breast cancer cell lines to determine how treatment with each effected mechanical depolarization of the cells. As the multiplexed detection of ROS/RNS species allows for the concentration of each species to be quantified, the analysis of each drug’s effectiveness showed that H₂O₂, NO^-, NO₂^-, and ONOO^- production was not uniformly affected and that the differences in the mechanism of action could be determined with more study (30). After being established as an adaptable method, the multiplexed amperometric detection of ROS/RNS at four potentials was integrated into a microfluidic device that separates each analytical measurement into individual 38μL cell chambers (31).

2.5. Microchip Electrophoresis with Amperometric Detection

As an alternative approach to multielectrode formats, microchip electrophoresis methods have been applied to multianalyte physiometry. By adding a separation step prior to electrochemical detection, these devices can increase the number of analytes detected on a single electrode. The challenge in pairing microchip electrophoresis (ME) and amperometric detection is noise from high ME operating potentials, but noise reduction techniques, such as in-channel detection and an electrically isolated potentiostat, can be implemented (32). ME with amperometric detection (Figure 2d) has been used to detect redox active biomolecules, interfering species that are electroactive at the same potentials as target molecules (32–40), and glucose using a glucose oxidase modified working electrode (34). Methods have been optimized to observe nitric oxide production in macrophages stimulated with lipopolysaccharide (LPS) by analyzing nitrite ions (35) and dopamine metabolites from rat brain slices (39). ME with amperometric detection is a versatile technique that can detect multiple analytes, and does not required specific modification of each electrode sensor. This enables ME to be more generic as the same basic device does not need specific analyte modifications to detect another set of species.

3. Clinical Devices and Testing

Clinical testing and point of care (POC) devices incorporates multianalyte physiometry to monitor, diagnose, or treat a patient. With hundreds of possible tests available, clinical MAP devices focus on patient monitoring, organ function status, or disease diagnosis These focused assays significantly narrow the number and type of tests performed. Typically, a physician starts with a few dozen analytes in clinical blood tests that are used for monitoring a patient during recovery, treatments and procedures; and transitions to organ function and disease biomarker specific tests if irregularities in the basic tests are present (41–43).

The i-STAT ® handheld blood analyzer (Abbott Point of Care) has been a staple in hospitals and doctor’s offices around the world for almost 25 years and frequently a featured point of care device in clinical chemistry literature (41, 43–47). The self-contained testing cartridges (Figure 3a) have nineteen commercially available variations that contain either single-analyte or multianalyte electrode arrays (41, 48). Low operating volumes (17–95μL) accommodate comparative validation of a newly developed MAP device or improved detection schemes, especially if the amount of sample is limited. The cartridges contain a calibrant solution and one or more electrode arrays with exposed contact pads to easily connect to the handheld device. Quality control during the analysis includes a pseudo-sample statistical treatment to determine validity of the cartridge, followed by the sample, which is evaluated for concentrations, bubbles, and volume collected. If the calibrant and sample measurements meet the statistical demands, a digital output reports concentrations determined via single concentration point calibration (41, 47). In most circumstances, pseudosample analysis is not ideal for treating data sets. To limit variability in the experimental conditions and determine data skews within the population, high precision commercial microfabrication of cartridges and regular clinical chemistry evaluations are relied upon (45–47, 49, 50). As a result of the reliability and immunoassay availability, an i-STAT device and cartridge has been modified to detect point mutations in genetic diseases and strain differentiation in viral and bacterial infections after an eight minute on-device PCR (51). The small, rugged nature of the i-STAT device lends to use in the emergency room and by care providers who travel between resource-constrained hospitals (46). Even though the i-STAT has low operating volumes conducive to some of the larger OoCs and bioreactors, the i-STAT lacks automated fluid handling and is targeted to physiometry measurements of macroscale samples.

Figure 3.

Photograph of a microclinical analyzer inset with a schematic of a screen-printed electrode. The pump and valve work together to flow 26μL of buffer, calibrants, and/or sample into the sample chamber containing the electrodes. From left to right, electrodes are modified to detect pH (blue), glucose (yellow), oxygen(middle), and lactate (pink). The far right electrode is an Ag/AgCl quasi-reference. Photograph courtesy of Dmitry Markov.

3.1. Multiplexed Electrolyte Detection

A prominent category of physiological tests is monitoring of electrolytes, specifically sodium, potassium, calcium, magnesium, and chloride (41). Electrochemical ion detection is most commonly achieved via ion selective electrodes and potentiometric techniques (56). While many ion selective membranes designed for ISEs are commercially available, innovative multiplexed, low-interferent, ion selective sensors for ecological and physiological purposes could reduce the number of electrodes needed in an array potentially leading to reduced detection volumes. Recently described layered devices utilized cyclic voltammetry to measure concentrations of multiple ions based on a thin membrane embedded with multiple ionophores covering an ion-to-electron transducing material. Ion-transfer at the membrane surface occurs at specific potentials for different ions depending on the ionophore used (59–62). While not all materials utilizing the thin membrane sensors are robust enough for some biological samples, a recent polyurethane ionophore was able to detect lithium and potassium ions in a single scan in undiluted plasma and serum (61). ISEs enhanced with ion-exchange exclusion membranes have improved the separation efficiencies and sensor lifetimes enabling them to be used in complex clinical samples (63–66). In addition to multiple electrolyte detection on a single sensor, fully automated ISE arrays have detected monoatomic and polyatomic ions for analysis of the carbon cycle in freshwater ecosystems (67).

3.2. Additional Biofluids

POC sensors need to be able to handle measurements in various biological matrices, such as blood, cerebrospinal fluid, urine, intraocular fluid and saliva, yet handle a wide range of concentrations and potential interferences that vary between matrix fluid types (42–44, 68–70). Many new devices are being designed to avoid finger-stick blood draws by optimizing testing in fluids requiring less invasive sampling (70, 71). One such MAP device was shown capable of measuring glucose, lactate, and cholesterol in blood and saliva with organic electrochemical transistors. In this device, analyte specific oxidase enzymes were immobilized using chitosan modified with a ferrrocene redox mediator, onto poly(3,4-ethylene-dioxythiophene):poly(styrene sulfonic acid) (PEDOT:PSS) coated gold electrodes, which effectively lowered the operating potential to 100–200mV and reduced the number of redox interferences. An additional electrode was included for the purposes of background subtraction and was coated with all the same materials except bovine serum albumin (BSA) replaced the enzymes. The microfluidic flow of the sample in PBS through the device was controlled by manual compression of the poly(dimethylsiloxane) housing. Sample injection was accomplished by manual release of the pressure on the device resulting in sample flow to electrodes upon channel re-expansion (Figure 3d) (71).

4. In Vivo Multianalyte Physiometry using Microanalytical Devices.

Due to the failure of in vitro models to accurately predict the efficacy of certain drugs in vivo, the need for animal and human trials remains. Several factors may contribute to the discrepancy between predicted vs observed therapeutic effects, including poor bioavailability of the compound and off-target toxicity of the drug or its metabolites. Although reductionist approaches have provided insight into the complex biochemical pathways that exist within the human body, it can be particularly difficult to predict off-target effects prior to in vivo testing. Additionally, the shift of the medical field toward personalized medicine makes it advantageous to develop methodologies that allow for continuous patient monitoring for an extended period of time.

4.1. Strategies for Avoiding Electrochemical Interference

Measurements made in the undefined environmental conditions within an organism may encounter electrochemical interference. This interference can come from both direct sources, such as redox active species, and indirect sources, such as biofouling, and commonly manifests as artificial changes in current or potential, plaguing electrochemical investigations. The need to measure signal coming only from analytes of interest has led investigators to develop several methods capable reducing the impacts of electrochemical interference, including 1) using electrostatic polymer barriers (6, 76, 81, 82, 84, 91, 95), 2) shifting the overpotential through electrode modifications (4, 77, 78, 89, 90, 96, 97), and 3) limiting electrode exposure time to the interfering species (98).

Polymer barriers such as polypyrrole and Nafion films can reduce the interference of redox active species, such as dopamine (DA) and ascorbic acid (AA), commonly encountered when making electrochemical measurements in vivo. These polymers are charged, and as such they form an electrostatic barrier around the electrode that can prevent the approach of charged species. For example, 3,4-dihydroxyphenyl-L-alanine (L-DOPA) is used to treat Parkinson’s disease but polymerizes on the surface of carbon microelectrodes forming melanin and acting as an interferent (84). This biofouling is largely mitigated at high concentrations of L-DOPA by electrochemically preconditioning a carbon microfiber electrode and electrodepositing a thin film of Nafion onto the surface of the electrode, maintaining the sub-second response times associated with fast-scan cyclic voltammetry (FSCV) (84). Without the Nafion layer, L-DOPA treatment appeared to inhibit the release of DA, but with the Nafion layer, L-DOPA treatment was observed to enhance DA release in neural pathways associated with Parkinson’s disease, which highlights the role of electrode design in understanding biochemical pathways in vivo (84).

Direct interference can also be avoided by shifting the potential at which amperometric measurements are made (4, 77, 78, 89, 90, 96, 97). Shifting of the needed overpotential can be accomplished by either coating or impregnating the working electrode with dyes such as Meldola’s blue (MB) and Prussian blue (PB). For example, carbon electrodes modified with MB can be used to monitor enzymatic reactions involving the production of NADH, but AA interferes with these measurements (97). NADH is frequently detected at ~0.6 V, the same potential as AA. Incorporation of MB into the electrode shifts the needed potential down to +0.1 V 24 where AA does not become oxidized and thus does not contribute to the measured current. Measuring NADH has been used to diagnose and monitor the intrahepatic cholestasis of pregnancy (97). When diagnosing and monitoring this it is imperative that biosensors utilizing 3-HSD respond only to analytes of interest where artificially large currents resulting in a false positive could lead to unnecessarily inducing labor.

4.2. MAP Utilizing Microdialysate.

Monitoring metabolic changes within the brain can help elucidate the biochemical basis of a variety of neural pathologies. Because AA is a known interferent in experiments such as these, shifting the potential used can help signal overlap with analytes of interest. Secondary enzymes such as horseradish peroxidase (HRP) can be included to shift the potential needed for amperometric detection of glucose and lactate down to −0.1 V by adding ferrocene to the buffer used to carry the dialysate sample through FIA systems (4, 76–80). The use of rapid sampling microdialysis (rsMD) allows for on-line measurements of patient samples in FIA systems (4, 76–80). When coupled with other methods of monitoring brain activity such as electrocorticography, correlations can be made between extracellular concentrations of glucose and lactate and spreading depolarizations (SDs) in perilesional tissue following traumatic brain injury (76, 78). After the SD wave passes through the perilesional tissue, significant decreases in glucose and increases in lactate concentrations are noted, likely the result of cellular recovery (78). The observation that the changes in extracellular glucose and lactate concentrations occur after the SD wave has passed successfully demonstrates that the SD wave is likely the cause and not the effect of changes in cellular metabolism. For patients experiencing frequent SDs, the extracellular glucose concentration fails to reach pre-SD levels before the next wave again depolarizes the cells, suggesting that a prolonged hypoglycemic milieu may contribute to the poorer outcome for these patients (78).

As with traumatic brain injuries, localized physiological monitoring after surgical procedures can provide insight into recovery mechanisms as well as provide a personalized approach to medicine to predict patient outcome and alter the course of treatment as necessary. One such procedure is surgical anastomosis, the tethering of tubes or channels such as blood vessels or intestines. Following an anastomosis, the lactate to glucose (L:G) ratio determined through FIA using rsMD may continue to increase near the surgical site indicating local hypoxia (4, 77). Ischemic conditions, which often require surgical intervention if prolonged, have been evaluated in porcine bowels (77). Using an intramurally inserted microdialysis probe into porcine bowels, glucose and lactate were measured both before and after repeated simulated ischemic conditions (77). As ischemia set in, the concentration of glucose decreased with a simultaneous rise in lactate concentration. Upon reperfusion, the levels returned to baseline, but the effects of a secondary ischemic event were more severe than those seen in the first, suggesting that the tissue did not fully recover from the first ischemic event or was more susceptible to future events (77). In experiments such as these, the ratio of the analytes provides a better picture of the metabolic processes as slight changes in the position of the microdialysis probe can alter the measurements dramatically. This is in part due to the flushing effect of the microdialysis perfusion fluid over time causing artificially decreased absolute concentrations of analytes being measured under conditions where nutrients are not being replaced as seen in ischemia.

Monitoring of free flap surgical procedures through FIA with rsMD predicts patient outcome using the L:G ratio (4). After anastomosis of the arteries and veins of the free flap, the L:G ratio is expected to return to near pre-operative baselines as the nutrient supply is restored and wastes are carried away. Cases in which the ratio did not return to baseline but instead continued to steadily increase were those where the anastomosis had failed or a thrombosis developed (4). Interestingly, the effect of the topical vasodilator papaverine was also observed in real time as absolute concentrations of glucose and lactate both increased due to the increase in blood flow to the area (4).

The microdialysis platforms used in these studies have limited temporal resolution in part due to the downstream injection system. However, the use of microdialysis probes is not limited to FIA platforms. Continuous measurements of the microdialysate using traditional enzymatic biosensors can improve temporal resolution (76). In the case of SD wave passages, where the L:G ratio increases as discussed above (78), on-line sensors are capable of continuously measuring glucose and potassium ion concentrations. Using a potassium ion selective electrode and the glucose biosensor to monitor induced SD in mice, temporal resolution was improved by a factor of 60, down to 1 second (76). Using this method, the decrease in extracellular glucose was temporally resolved from the increase in potassium ion concentration, demonstrating that the drop is glucose is in response to the increased metabolic demands of the neurons undergoing repolarization (76).

4.3. MAP Using Implantable Electrodes

Electrodes may also be directly implanted in vivo to probe complex biochemical pathways that are difficult or impossible to reproduce in vitro, such as those seen in the brain. These in vivo electrodes can therefore preserve information about sub-second cellular responses, a daunting task for microdialysis and flow injection methods. However, implanted electrodes present their own set of challenges: the sensitivity of the electrodes must remain constant and they must be very small or risk a trade-off of spatial resolution for temporal resolution.

Two broad categories of electrodes have proven efficient in maintaining spatial resolution for in vivo physiological measurements: metallic microelectrode arrays (6, 81, 82) and carbon fiber microelectrodes (7, 83–88). Metal microelectrode arrays are slightly larger than carbon fiber electrodes because they spatially separate the surface at which each analyte is detected by electrode-specific functionalization such as enzymatic or ion-selective films. However, this separation allows for more specific detection of each analyte of interest by decreasing the degree of interference between different analytes while maintaining high sensitivity for individual analytes (6, 81, 82, 89–92). Some metal electrodes are not suitable for implantation despite their ubiquity in in vitro devices. For example, Ag/AgCl film electrodes have been shown to elicit inflammatory responses in vivo (81). Alternatively, IrO₂ is a biocompatible reference electrode that exhibits long-term stability and low noise in vivo. This electrode is, however, sensitive to pH changes and may only be suitable in regions of the body where pH changes are minimal, such as the brain (81). Metal electrodes have been used to measure physiological changes in the brain in response to mechanical and chemical stimuli. For example, the release of neurotransmitters in the striatum of rats in response to electrical stimuli and physical stresses can be monitored with sub-second temporal resolution using platinum electrodes (6). The second category of implantable electrodes, carbon fiber microelectrodes, has superior spatial resolution relative to metal electrodes. Carbon fiber microelectrodes are frequently used with FSCV to quantitate catecholamines such as DA (7, 84–87). Because DA adsorbs well to the surface of carbon fiber microelectrodes, faster scan rates are used to maximize signal. Traditionally, this is accomplished by using a triangular waveform, but novel waveforms such as the “sawhorse” scan which briefly holds the potential above +1.0 V further improve sensitivity by opening up sites on the electrode for DA to bind (85). This can be used to deconvolute signals arising from mixtures of redox active molecules such as adenosine, ATP, and H₂O₂ (93).

Long-term studies with implantable electrodes are limited by their decreased sensitivity over time. Loss of sensitivity is primarily due to biofouling and is compounded by the inability to recalibrate electrodes once implanted (82). Biofouling of GOx, LOx, and pyruvate oxidase (POx) coated electrodes results in a diffusion barrier shown to decrease sensitivity in vivo (82). However, the purposeful use of diffusion barriers, such as microdialysis membranes has been shown to extend the linear range of electrodes as well as reduce biofouling. For example, LOx films protected by a microdialysis membrane better retained sensitivity compared to GOx and POx films without the protective membrane (82). Alternatively, FSCV can be used to extend the lifetime of implanted electrodes by eliminating the assumption that the electrode sensitivity is constant during the course of an experiment. It does so by fitting the total background current and switching potential to a model with four regression coefficients (83). This method has accurately predicted the sensitivity of a carbon fiber microelectrode for a variety of analytes in vivo, including DA, AA, H₂O₂, and H⁺ (83).

Spatial mapping can be accomplished by electrochemically evaluating how cells respond to different stimuli. By determining how cells respond to DA, medium spiny neurons can be subtyped, mapping the nucleus accumbens in behavioral studies (87). When iontophoresed stimulants are not redox active in the potential range being swept with FSCV, redox active compounds such as DOPAC (88) and acetaminophen (87) can be added to the iontophoresed solution to act as internal standards. Glutamate, an iontophoresed stimulant that is not redox active, can be accurately quantified by such internal standards (88).

By simultaneously using electrodes at different locations, responses in different regions of the brain may be seen as a result of a single stimulus (7). For example, DA can be tracked in the nucleus accumbens while 5-HT is tracked in the substantia nigra pars reticulata to investigate the mechanisms governing their release. Upon stimulation, DA release in the nucleus accumbens is 300 times greater than 5-HT release in the substantia nigra pars reticulata despite them being similar in overall concentration within their respective tissue (7). By selectively inhibiting enzymes responsible for the synthesis, packaging, release, uptake, and metabolism of these two neurotransmitters, the dependence of DA transmission on synthesis and repackaging and the tight regulation of 5-HT transmission by reuptake and degradation pathways were revealed. Additionally, the severe neurological consequences resulting from the co-administration of 5-HT transporter and monoamine oxidase inhibitors were observed (7).

4.4. Wearable Devices for MAP

As manageable chronic conditions such as diabetes mellitus, hypertension, and hyperlipidemia become more prevalent, the demand for minimally invasive continuous monitoring devices increases. Wearable devices are now capable of monitoring many different analytes simultaneously, have readouts that interface with mobile phones, and can deliver a therapeutic agent (89). Devices that utilize bodily fluids other than serum and cerebral spinal fluid such as sweat and urine are becoming increasingly common in part due to the ease by which samples can be obtained (89–92). The ‘diabetes patch’ sits on the surface of the skin, monitors temperature, humidity, pH, and glucose levels, and delivers Metformin as needed. In the device, glucose is measured by the amperometric detection of peroxide on gold-doped graphene modified with PB and GOx at −0.05 V. By using additional sensors to assess pH and temperature, corrections can be made for the activity of GOx, yielding more accurate glucose readings. When hyperglycemic conditions are detected, a thermal actuator will melt polymeric microneedles incorporating Metformin, releasing it into the bloodstream. This patch is capable of continuously monitoring glucose levels over an entire day, and such non-invasive platforms may someday replace current methods of monitoring and managing blood glucose levels.

Wearable MAP microanalytical devices have also been utilized in exercise training regimens to gain insights into individual performance. Two of the most promising are a 3D-printed microfluidic device and ‘smart bands.’ The 3D-printed microfluidic devices have a subcutaneous microdialysis probe that allows external needle-based electrodes to continuously monitor glucose and lactate levels (94). The needle electrodes are removable from the device housing and can be easily modified to measure different analytes as desired. Although technically wearable, this device is still reliant upon an external potentiostat and requires a clinician to subcutaneously insert the microdialysis probe (94). On the other hand, a recently developed ‘smart headband/smart wristband’ can simultaneously and non-invasively measure glucose, lactate, Na⁺, and K⁺ levels in sweat (90). In these smart bands, GOx- and LOx-chitosan-SWCNT films on PB are used for chronoamerpometric detection of glucose and lactate, respectively, whereas Na⁺ and K⁺ levels are determined via open circuit potential using ionophores (Na ionophore X and valinomycin) in ion-selective membranes atop PEDOT:PSS. When access to water is restricted during exercise while wearing the flexible integrated sensor array (FISA), dehydration can be clearly seen when concentrations of sodium and potassium begin to significantly increase. In combination with the real-time profiles of glucose and lactate, athletes undergoing intense training can avoid over-exertion and gain physiological insights into individual performance.

Wearable FISAs have also incorporated ion-selective membranes to continuously measure Ca²⁺ and pH levels to determine hydration and electrolyte levels. Ca²⁺ concentration and pH are measured through films incorporating calcium ionophore II on PEDOT:PSS and electropolymerized polyaniline.(92) As with sodium and potassium, calcium ion concentration can be monitored with open circuit potential measurements in sweat during the course of a workout, and the in vivo stability of the reference electrode can be enhanced by the incorporation of a poly-vinyl butyral layer to maintain chloride ion saturation at the Ag/AgCl film with little interference from other common cations, including NH₄⁺, Mg²⁺, K⁺, and Na⁺. The results obtained with the FISA for both pH and Ca²⁺ aligned well with the results from commercial pH meters and ICP-MS, the gold-standard for Ca²⁺ measurement, indicating that these wearable platforms can be used in clinically relevant conditions such as hyperparathyroidism and kidney stones.

The quantitation of other metal species bears clinical relevance due conditions such as Wilson’s disease and acute heavy metal poisoning. Heavy metals can be detected by square wave anodic stripping voltammetry relying on the inherent redox potentials of the metal species being evaluated. The redox potentials of zinc, copper, cadmium, lead, and mercury are sufficiently separated to allow their quantitation in a complex mixture (91). The strong correlation between values obtained from a sweat sample using ICP-MS and the wearable sensor (91) indicates that these wearable sensors can also be used reliably for heavy metal detection in addition to the analytes previously discussed (90, 92). These advances significantly expand the number of cationic species that can be continuously evaluated in biofluids such as sweat and urine with wearable sensors, thus enlarging the diagnostic toolbox in the field.

5. Conclusions

The in-house modification of commercially available instruments to convert single analyte detection systems into those capable of measuring multiple analytes has largely become obsolete in part due to the advent of screen printed electrodes and 3-D fabrication techniques. The inclusion of customizable sensors into microfluidic devices has enabled multianalyte detection under physiological conditions, allowing researchers to make advances in the fields of bioenergetics, toxicology, and neurology. These devices also show great promise in clinical settings due to their capabilities of continuously monitoring vitals and predicting patient outcome. As the medical field shifts toward more personalized approaches, the widespread adoption of devices that can perform multianalyte detection is inevitable as evidenced by the continued success of the i-STAT. Outside of research and clinical settings, non-invasive smart bands are beginning to allow athletes to optimize training regimens, and the self-contained nature of these robust fully integrated sensor arrays in combination with their low power usage makes them ideal for use in low resource monitoring settings as well. Overall, multianalyte investigations will continue to grow in number as the monitoring of a panel of analytes is more broadly recognized for its utility in gaining a better understanding of fundamental cellular processes and disease states.

Figure 4.

Clinical and wearable devices. (a) i-STAT handheld device. Adapted with permission from Reference 48. Copyright 1998, American Chemical Society. (b) Photograph and schematic of the selective multianalyte detection in complex media using the finger-powered OECT array. Photograph shows a red-colored solution that was pressure driven from the inlet through the sensing areas, as indicated by the vertical arrow. Adapted with permission from Reference 70. Copyright 2016, John Wiley & Sons. (c) A fully integrated wearable multiplexed sensing system on a subject’s arm. Adapted with permission from Reference 91. Copyright 2013, Royal Society of Chemistry. Abbreviations: BSA, bovine serum albumin; OECT, organic electrochemical transistor

Table 1:

Summary of physiological systems and their multianalyte readouts in microanalytical devices.

Systems	Analytes/Event Detection Method	References
Cellular Bioenergetics	Glucose, Lactate, Oxygen, pH and ATP	(5, 8–17, 19, 21, 34, 54, 58, 73)
Oxidative Stress	RNS, ROS, Glutathione, AA, Tyr	(29–32, 35)
Vesicular Exocytosis	Histamine, Dopamine, Serotonin	(22, 24, 25)
Antioxidant Capacity	Caffeic, Gallic, Ascorbic, and Trolox T acids	(27, 28, 36)
Neurotransmitters	Dopamine, Adenosine, Serotonin	(25, 26, 33, 35)
Cytotoxicity	K+, Na+, H+, Ca2+−	(57)
Patient Monitoring & Diagnosis	Ions, biomolecules, glucose, urea	(40, 51, 59–67, 71)