Antioxidants L-carnitine and D-methionine modulate neuronal activity through GABAergic inhibition.

Antioxidants are well known for their neuroprotective properties against reactive oxygen species in cortical neurons and auditory cells. We recently identified L-carnitine and D-methionine to be among agents that provide such protection. Here, we investigated their neuronal modulatory actions. We used cultured neuronal networks grown on microelectrode arrays to assess the effects of L-carnitine and D-methionine on network function. Spike production and burst properties of neuronal networks were used as parameters to monitor pharmacological responses. L-Carnitine and D-methionine reduced spike activity with 100% efficacy with EC50 values of 0.22 (± 0.01) mM and 1.06 (± 0.05) mM, respectively. In the presence of 1.0-40 µM of the GABAA antagonist bicuculline, the sigmoidal concentration-response curves of both compounds exhibited stepwise shifts, without a change in efficacy. Under a maximal bicuculline concentration of 40 µM, the EC50 increased to 3.57 (± 0.26) mM for L-carnitine and to 10.52 (± 0.97) mM for D-methionine, more than a tenfold increase. The agonist-antagonist interactions with bicuculline were estimated by Lineweaver-Burk plot analyses to be competitive, corroborated by the computed dissociation constants of bicuculline. For both compounds, the effects on the network burst pattern, activity reversibility, and bicuculline antagonism resembled that elicited by the GABAA agonist muscimol. We showed that the antioxidants L-carnitine and D-methionine modulate cortical electrical spike activity primarily through GABAA receptor activation. Our findings suggest the involvement of GABAergic mechanisms that perhaps contribute to the protective actions of these compounds.

Simultaneous electrophysiological and morphological assessment of functional damage to neural networks in vitro after 30-300 g impacts.

An enigma of mild traumatic brain injury are observations of substantial behavior and performance deficits in the absence of bleeding or other observable structural damage. Altered behavior and performance reflect changes in action potential (AP) patterns within neuronal networks, which could result from subtle subcellular responses that affect synaptic efficacy and AP production. The aim of this study was to investigate and quantify network activity changes after simulated concussions in vitro and therewith develop a platform for simultaneous and direct observations of morphological and electrophysiological changes in neural networks. We used spontaneously active networks grown on microelectrode arrays (MEAs) to allow long-term multisite monitoring with simultaneous optical observations before and after impacts delivered by a ballistic pendulum (30 to 300 g accelerations). The monitoring of AP waveshape templates for long periods before and after impact provided an internal control for cell death or loss of cell-electrode coupling in the observed set of neurons. Network activity patterns were linked in real-time to high power phase contrast microscopy. There was no overt loss of glial or neuronal adhesion, even at high-g impacts. All recording experiments showed repeatable spike production responses: a loss of activity with recovery to near reference in 1 hr, followed by a slow activity decay to a stable, level plateau approximately 30-40% below reference. The initial recovery occurred in two steps: a rapid return of activity to an average 24% below reference, forming a level plateau lasting from 5 to 20 min, followed by a climb to within 10% of reference where a second plateau was established for 1 to 2 hrs. Cross correlation profiles revealed changes in firing hierarchy as well as in Phase 1 in spontaneous network oscillations that were reduced by as much as 20% 6-8 min post impact with only a partial recovery at 30 min. We also observed that normally stable nuclei developed irregular rotational motion after impact in 27 out of 30 networks. The evolution of network activity deficits and recovery can be linked with microscopically observable changes in the very cells that are generating the activity. The repeatable electrophysiological impact response profiles and oscillation changes can provide a quantitative basis for systematic evaluations of pharmacological intervention strategies. Future expansion to include fluorescent microscopy should allow detailed investigations of damage mechanisms on the subcellular level.

Spontaneous coordinated activity in cultured networks: analysis of multiple ignition sites, primary circuits, and burst phase delay distributions.

All higher order central nervous systems exhibit spontaneous neural activity, though the purpose and mechanistic origin of such activity remains poorly understood. We quantitatively analyzed the ignition and spread of collective spontaneous electrophysiological activity in networks of cultured cortical neurons growing on microelectrode arrays. Leader neurons, which form a mono-synaptically connected primary circuit, and initiate a majority of network bursts were found to be a small subset of recorded neurons. Leader/follower firing delay times formed temporally stable positively skewed distributions. Blocking inhibitory synapses usually resulted in shorter delay times with reduced variance. These distributions are characterizations of general aspects of internal network dynamics and provide estimates of pair-wise synaptic distances. The resulting analysis produced specific quantitative constraints and insights into the activation patterns of collective neuronal activity in self-organized cortical networks, which may prove useful for models emulating spontaneously active systems.

Dissociation constants for GABA(A) receptor antagonists determined with neuronal networks on microelectrode arrays.

Changes in spontaneous spike activities from murine frontal cortex networks grown on microelectrode arrays were used to determine the dissociation constants of three GABA(A) antagonists: gabazine, bicuculline, and trimethylolpropane phosphate (TMPP). Networks were treated with fixed concentrations of antagonists and titrated with the GABA(A) receptor agonist muscimol. Muscimol decreased spike activity in a concentration-dependent manner with full efficacy (100% spike inhibition). A sigmoidal curve fit provided a 50% inhibitory concentration (IC(50)) of 0.14+/-0.05muM (mean+/-S.D., n=5). Increasing concentrations of the three antagonists shifted the muscimol concentration response curves (CRCs) to the right with the same efficacy. Schild plot analyses with linear regressions resulted in slopes that are statistically not different from unity and provided X-intercepts (dissociation constants) of 0.23, 0.61, and 3.98muM for gabazine, bicuculline, and TMPP, respectively. Corresponding pA2 values (-logK(B)) were 6.64, 6.21, and 5.40. The dissociation constants for gabazine and bicuculline agree well with those obtained with other methods. The TMPP K(B) has not yet been reported in the literature. The data suggest that spontaneously active networks on microelectrode arrays can be used as reliable platforms for rapid quantitative pharmacological investigations.

Functional structure of cortical neuronal networks grown in vitro.

We apply an information-theoretic treatment of action potential time series measured with microelectrode arrays to estimate the connectivity of mammalian neuronal cell assemblies grown in vitro. We infer connectivity between two neurons via the measurement of the mutual information between their spike trains. In addition we measure higher-point multi-information between any two spike trains, conditional on the activity of a third cell, as a means to identify and distinguish classes of functional connectivity among three neurons. The use of a conditional three-cell measure removes some interpretational shortcomings of the pairwise mutual information and sheds light on the functional connectivity arrangements of any three cells. We analyze the resultant connectivity graphs in light of other complex networks and demonstrate that, despite their ex vivo development, the connectivity maps derived from cultured neural assemblies are similar to other biological networks and display nontrivial structure in clustering coefficient, network diameter, and assortative mixing. Specifically we show that these networks are weakly disassortative small-world graphs, which differ significantly in their structure from randomized graphs with the same degree. We expect our analysis to be useful in identifying the computational motifs of a wide variety of complex networks, derived from time series data.

Pharmacological response sensitization in nerve cell networks exposed to the antibiotic gentamicin.

Gentamicin is an aminoglycoside antibiotic that is used in clinical, organismic, and agricultural applications to combat gram-negative, aerobic bacteria. The clinical use of gentamicin is widely linked to various toxicities, but there is a void in our knowledge about the neuromodulatory or neurotoxicity effects of gentamicin. This investigation explored the electrophysiologic effects of gentamicin on GABAergic pharmacological profiles in spontaneously active neuronal networks in vitro derived from auditory cortices of E16 mouse embryos and grown on microelectrode arrays. Using the GABAA agonist muscimol as the test substance, responses from networks to dose titrations of muscimol were compared in the presence and absence of 100µM gentamicin (the recommended concentration for cell culture conditions). Spike-rate based EC50 values were generated using sigmoidal fit concentration response curves (CRCs). Exposure to 100µM gentamicin exhibited a muscimol EC50±S.E.M. of 80±6nM (n=10). The EC50 value obtained in the absence of gentamicin was 124±11nM (n=10). The 35% increase in potency suggests network sensitization to muscimol in the presence of gentamicin. Action potential (AP) waveform analyses of neurons exposed to gentamicin demonstrated a concentration-dependent decrease in AP amplitudes (extracellular recordings), possibly reflecting gentamicin effects on voltage-gated ion channels. These in vitro results reveal alteration of pharmacological responses by antibiotics that could have significant influence on the behavior and performance of animals.

Measuring synchronization in neuronal networks for biosensor applications.

Cultures of neurons can be grown on microelectrode arrays (MEAs), so that their spike and burst activity can be monitored. These activity patterns are quite sensitive to changes in the environment, such as chemical exposure, and hence the cultures can be used as biosensors. One key issue in analyzing the data from neuronal networks is how to quantify the level of synchronization among different units, which represent different neurons in the network. In this paper, we propose a synchronization metric, based on the statistical distribution of unit-to-unit correlation coefficients. We show that this synchronization metric changes significantly when the networks are exposed to bicuculline, strychnine, or 2,3-dioxo-6-nitro-l,2,3,4-tetrahydrobenzoquinoxaline-7-sulphonamide (NBQX). For that reason, this metric can be used to characterize pharmacologically induced changes in a network, either for research or for biosensor applications.

A portable microelectrode array recording system incorporating cultured neuronal networks for neurotoxin detection.

Cultured neuronal networks, which have the capacity to respond to a wide range of neuroactive compounds, have been suggested to be useful for both screening known analytes and unknown compounds for acute neuropharmacologic effects. Extracellular recording from cultured neuronal networks provides a means for extracting physiologically relevant activity, i.e. action potential firing, in a noninvasive manner conducive for long-term measurements. Previous work from our laboratory described prototype portable systems capable of high signal-to-noise extracellular recordings from cardiac myocytes. The present work describes a portable system tailored to monitoring neuronal extracellular potentials that readily incorporates standardized microelectrode arrays developed by and in use at the University of North Texas. This system utilizes low noise amplifier and filter boards, a two-stage thermal control system with integrated fluidics and a graphical user interface for data acquisition and control implemented on a personal computer. Wherever possible, off-the-shelf components have been utilized for system design and fabrication. During use with cultured neuronal networks, the system typically exhibits input referred noise levels of only 4-6 microVRMS, such that extracellular potentials exceeding 40 microV can be readily resolved. A flow rate of up to 1 ml/min was achieved while the cell recording chamber temperature was maintained within a range of 36-37 degrees C. To demonstrate the capability of this system to resolve small extracellular potentials, pharmacological experiments with cultured neuronal networks have been performed using ion channel blockers, tetrodotoxin and tityustoxin. The implications of the experiments for neurotoxin detection are discussed.

Differential acute effects of fluoxetine on frontal and auditory cortex networks in vitro.

Primary cultures of neuronal networks grown on microelectrode arrays were used to quantify acute effects of fluoxetine (Prozac) on spontaneous spike and burst activity. For frontal cortex cultures, fluoxetine showed consistent inhibitory effects and terminated activity at 10-16 microM. IC(50) mean+/-S.E. for spike rates was 5.4+/-0.7 microM (n=15). For auditory cortex cultures, fluoxetine caused excitation at 1-10 microM, initial inhibition at 15 microM, and activity cessation at 20-25 microM. The spike rate IC(50) was 15.9+/-1.0 microM (n=11). Fluoxetine did not change the action potential waveform shape. However, at high concentrations, it caused total cessation of spike activity on all channels. The inhibition caused by fluoxetine was reversible for both tissues. Based on the results, we conclude that cultures showed repeatable, concentration-dependent sensitivities to fluoxetine but demonstrated tissue-specific responses for frontal and auditory cortex networks. These responses may not be due to the interference with serotonin reuptake, but may be due to a secondary effect on ionic channels.

Acute neuropharmacologic action of chloroquine on cortical neurons in vitro.

Chloroquine, a common quinolone derivative used in the treatment of malaria, has been associated with neurologic side-effects including depression, psychosis and delirium. The neuropharmacologic effects of chloroquine were examined on cultured cortical neurons using microelectrode array (MEA) recording and the whole-cell patch clamp technique. Whole-cell patch clamp records under current-clamp mode also showed a chloroquine-induced depression of the firing rate of spontaneous action potentials by approximately 40%, consistent with the observations with the MEA recording, although no changes in either the baseline membrane potential or input resistance were observed. Voltage clamp recordings of spontaneous post-synaptic currents, recorded in the presence of tetrodotoxin, revealed no obvious changes in either the amplitude or rate of occurrence of inward currents with application of chloroquine at 10 microM, suggesting that the fundamental molecular mechanisms underlying spontaneous synaptic transmission may not be affected by acute application of the drug. In contrast, a concentration-dependent inhibition of whole-cell calcium current was observed in the presence of chloroquine. These acute neuropharmacologic changes were not accompanied by cytotoxic actions of the compound, even after exposure of up to 500 microM chloroquine for 7 h. These data suggest that chloroquine can depress in vitro neuronal activity, perhaps through inhibition of membrane calcium channels.

Unique responses of auditory cortex networks in vitro to low concentrations of quinine.

The anti-malarial drug quinine has several side effects including tinnitus. The aim of the study was to determine if cultured auditory networks growing on microelectrode arrays exhibited unique dynamic states when exposed to quinine. Eight auditory cortex networks (ACN), eight frontal cortex networks (FCN), and five inferior colliculus networks (ICN) were used in this study. Response of ACNs to quinine was biphasic, with an excitatory phase followed by inhibition. FCNs and ICNs revealed only inhibitory responses. The concentrations at which the spike rate was inhibited by 50% (IC50 mean +/- SE) were 42.5 +/- 3.9, 28.7 +/- 4.8 and 23.9 +/- 2.1 microM for ACNs, FCNs, and ICNs, respectively. Quinine increased the regularity and coordination of bursting in all three tissues. The increased burst pattern regularity of ICNs coupled with the excitatory responses seen only in ACNs between 1 and 10 microM show a unique susceptibility of auditory tissues to quinine that may be related to the underlying mechanism that triggers tinnitus-like activity.

Histiotypic electrophysiological responses of cultured neuronal networks to ethanol.

Embryonic murine neuronal networks cultured on substrate-integrated microelectrode arrays were used to quantify acute electrophysiological effects of ethanol by using extracellular, multichannel recording of action potentials. Spontaneously active frontal cortex cultures showed repeatable, concentration-dependent sensitivities to ethanol, with initial inhibition at 20 mM and a spike rate 50% effective concentration (EC50) of 48.8+/-5.4 mM. Ethanol concentrations of greater than 100 mM led to cessation of activity. The ethanol inhibitions up to the maximum tested 160 mM were reversible. Although ethanol did not change the shape of action potentials, unit-specific spike pattern effects were found. At 40 mM, ethanol decreased neuronal firing in 71%, increased firing in 20%, and generated no effect in 9% of all units observed (14 cultures, 200 discriminated units). The effects of combined application of ethanol and fluoxetine were additive. Excellent agreement with findings obtained from experimental studies with animals validates the use of these in vitro systems for alcohol research.

Botulinum toxin suppression of CNS network activity in vitro.

The botulinum toxins are potent agents which disrupt synaptic transmission. While the standard method for BoNT detection and quantification is based on the mouse lethality assay, we have examined whether alterations in cultured neuronal network activity can be used to detect the functional effects of BoNT. Murine spinal cord and frontal cortex networks cultured on substrate integrated microelectrode arrays allowed monitoring of spontaneous spike and burst activity with exposure to BoNT serotype A (BoNT-A). Exposure to BoNT-A inhibited spike activity in cultured neuronal networks where, after a delay due to toxin internalization, the rate of activity loss depended on toxin concentration. Over a 30 hr exposure to BoNT-A, the minimum concentration detected was 2 ng/mL, a level consistent with mouse lethality studies. A small proportion of spinal cord networks, but not frontal cortex networks, showed a transient increase in spike and burst activity with exposure to BoNT-A, an effect likely due to preferential inhibition of inhibitory synapses expressed in this tissue. Lastly, prior exposure to human-derived antisera containing neutralizing antibodies prevented BoNT-A induced inhibition of network spike activity. These observations suggest that the extracellular recording from cultured neuronal networks can be used to detect and quantify functional BoNT effects.

Pharmacodynamics of potassium channel openers in cultured neuronal networks.

A novel class of drugs - potassium (K(+)) channel openers or activators - has recently been shown to cause anticonvulsive and neuroprotective effects by activating hyperpolarizing K(+) currents, and therefore, may show efficacy for treating tinnitus. This study presents measurements of the modulatory effects of four K(+) channel openers on the spontaneous activity and action potential waveforms of neuronal networks. The networks were derived from mouse embryonic auditory cortices and grown on microelectrode arrays. Pentylenetetrazol was used to create hyperactivity states in the neuronal networks as a first approximation for mimicking tinnitus or tinnitus-like activity. We then compared the pharmacodynamics of the four channel activators, retigabine and flupirtine (voltage-gated K(+) channel KV7 activators), NS1619 and isopimaric acid ("big potassium" BK channel activators). The EC50 of retigabine, flupirtine, NS1619, and isopimaric acid were 8.0, 4.0, 5.8, and 7.8µM, respectively. The reduction of hyperactivity compared to the reference activity was significant. The present results highlight the notion of re-purposing the K(+) channel activators for reducing hyperactivity of spontaneously active auditory networks, serving as a platform for these drugs to show efficacy toward target identification, prevention, as well as treatment of tinnitus.

d-Methionine protects against cisplatin-induced neurotoxicity in cortical networks.

Cisplatin is a platinum-based chemotherapeutic agent widely used for the treatment of various types of cancer. Patients undergoing cisplatin treatment often suffer from a condition known as "chemobrain", ototoxicity, peripheral neuropathy, weight loss, nausea, vomiting, nephrotoxicity, seizures, hearing loss and tinnitus. d-Methionine (d-Met), a sulfur-containing nucleophilic antioxidant, has been shown to prevent cisplatin-induced side effects in animals without antitumor interference. In this study, we have used an in vitro model of cortical networks (CNs), enriched in auditory cortex cells; to quantify cisplatin neurotoxicity and the protective effects of d-Met. Dissociated neurons from auditory cortices of mouse embryos were grown on microelectrode arrays with 64 transparent indium-tin oxide electrodes, which enabled continuous optical and electrophysiological monitoring of network neurons. Cisplatin at 0.10-0.25 mM induced up to a 200% increase in spontaneous spiking activity, while concentrations at or above 0.5mM caused irreversible loss of neuronal activity, accompanied by cell death. Pretreatment with d-Met, at a concentration of 1.0mM, prevented the cisplatin-induced excitation at 0.10-0.25 mM, caused sustained excitation without occurrence of cell death at 0.5mM, and delayed cell death at 0.75 mM cisplatin. l-Methionine, the optical isomer, showed lower potency and less efficacy than d-Met, was less protective against 0.1mM cisplatin, and proved ineffective at a concentration of 0.5mM cisplatin. Pre-exposure time of d-Met was associated with the protective effects at 0.1 and 0.5mM cisplatin, with longer pre-exposure times exhibiting better protection. This study quantifies as a function of concentration and time that d-Met protects central nervous system tissue from acute cisplatin toxicity.

An in vitro model for testing drugs to treat tinnitus.

Tinnitus affects approximately 50 million people in the USA alone, with 10 million being highly debilitated. Pharmacotherapy for tinnitus is still in emerging stages due to time consuming clinical trials and/or animal experiments. We tested a new cellular model where induced rapid neuronal firing or spiking was used as a mimic for the type of aberrant activity that may occur in tinnitus. Spontaneously active auditory cortical networks growing on microelectrode arrays were exposed to pentylenetetrazol (PTZ), a proconvulsant and an antagonist of GABA(A) receptor, which is implicated in tinnitus. Auditory cortical networks were then exposed to experimental tinnitus drugs linopirdine (Dup966, a potassium channel blocker), L-carnitine (an antioxidant), or selective Ca(2+) channel antagonists pregabalin (Lyrica), or gabapentin (Neurontin) at various concentrations. PTZ increased spike rate by 139.6±27% and burst rate by 129.7±28% in auditory cortical networks with a phenotypic high firing of excitable neurons. Reductions of increased activity were observed to varying degrees using the experimental tinnitus drugs. The potency of the drugs was linopirdine (EC(50): 176±7.0 µM)>L-carnitine (EC(50): 1569±41 µM)>pregabalin (EC(50): 8360±340 µM), >gabapentin, with 34.2±7.5% efficacy (EC(50): 2092±980 µM). These studies provide proof of principle for the use of auditory cortical networks on microelectrode array as a feasible platform for semi-high throughput application for screening of drugs that might be used for the treatment of tinnitus.

Assessment of styrene oxide neurotoxicity using in vitro auditory cortex networks.

Styrene oxide (SO) (C8H8O), the major metabolite of styrene (C6H5CH=CH2), is widely used in industrial applications. Styrene and SO are neurotoxic and cause damaging effects on the auditory system. However, little is known about their concentration-dependent electrophysiological and morphological effects. We used spontaneously active auditory cortex networks (ACNs) growing on microelectrode arrays (MEA) to characterize neurotoxic effects of SO. Acute application of 0.1 to 3.0 mM SO showed concentration-dependent inhibition of spike activity with no noticeable morphological changes. The spike rate IC50 (concentration inducing 50% inhibition) was 511 ± 60 µM (n = 10). Subchronic (5 hr) single applications of 0.5 mM SO also showed 50% activity reduction with no overt changes in morphology. The results imply that electrophysiological toxicity precedes cytotoxicity. Five-hour exposures to 2 mM SO revealed neuronal death, irreversible activity loss, and pronounced glial swelling. Paradoxical "protection" by 40 µM bicuculline suggests binding of SO to GABA receptors.

Microelectrode arrays: a physiologically based neurotoxicity testing platform for the 21st century.

Microelectrode arrays (MEAs) have been in use over the past decade and a half to study multiple aspects of electrically excitable cells. In particular, MEAs have been applied to explore the pharmacological and toxicological effects of numerous compounds on spontaneous activity of neuronal and cardiac cell networks. The MEA system enables simultaneous extracellular recordings from multiple sites in the network in real time, increasing spatial resolution and thereby providing a robust measure of network activity. The simultaneous gathering of action potential and field potential data over long periods of time allows the monitoring of network functions that arise from the interaction of all cellular mechanisms responsible for spatio-temporal pattern generation. In these functional, dynamic systems, physical, chemical, and pharmacological perturbations are holistically reflected by the tissue responses. Such features make MEA technology well suited for the screening of compounds of interest, and also allow scaling to high throughput systems that can record from multiple, separate cell networks simultaneously in multi-well chips or plates. This article is designed to be useful to newcomers to this technology as well as those who are currently using MEAs in their research. It explains how MEA systems operate, summarizes what systems are available, and provides a discussion of emerging mathematical schemes that can be used for a rapid classification of drug or chemical effects. Current efforts that will expand this technology to an influential, high throughput, electrophysiological approach for reliable determinations of compound toxicity are also described and a comprehensive review of toxicological publications using MEAs is provided as an appendix to this publication. Overall, this article highlights the benefits and promise of MEA technology as a high throughput, rapid screening method for toxicity testing.

Magnetic stimulation and depression of mammalian networks in primary neuronal cell cultures.

For transcranial magnetic stimulation (TMS), the coupling of induced electric fields with neurons in gray matter is not well understood. There is little information on optimal stimulation parameters and on basic cellular mechanisms. For this reason, magnetic stimulation of spontaneously active neuronal networks, grown on microelectrode arrays in culture, was employed as a test environment. This allowed use of smaller coils and the continual monitoring of network action potential (AP) activity before, during, and for long periods after stimulation. Biphasic, rectangular, and 500 micros long pulses were used at mean pulse frequencies (MPFs) ranging from 3 to 100 Hz on both spinal cord (SC) and frontal cortex (FC) cultures. Contrary to stimulation of organized fiber bundles, APs were not elicited directly. Responses were predominantly inhibitory, dose dependent, with onset times between 10 s and several minutes. Spinal networks showed a greater sensitivity to activity suppression. Under pharmacological disinhibition, some excitation was seen at low pulse frequencies. FC cultures showed greater excitatory responses than SC networks. The observed primary inhibitory responses imply interference with synaptic exocytosis mechanisms. With 20,000 pulses at 10 Hz, 40% inhibition was maintained for over 30 min with full recovery, suggesting possible application to nonchemical, noninvasive pain management.

Carbon nanotube coating improves neuronal recordings.

Implanting electrical devices in the nervous system to treat neural diseases is becoming very common. The success of these brain-machine interfaces depends on the electrodes that come into contact with the neural tissue. Here we show that conventional tungsten and stainless steel wire electrodes can be coated with carbon nanotubes using electrochemical techniques under ambient conditions. The carbon nanotube coating enhanced both recording and electrical stimulation of neurons in culture, rats and monkeys by decreasing the electrode impedance and increasing charge transfer. Carbon nanotube-coated electrodes are expected to improve current electrophysiological techniques and to facilitate the development of long-lasting brain-machine interface devices.