Ad hoc hybrid synaptic junctions to detect nerve stimulation and its application to detect onset of diabetic polyneuropathy

Highlights

• A minimally invasive, synaptic transistor-based construct to monitor in vivo neuronal activity.

• Measure excitability of sciatic nerves due to a stimulation of the footpad in cohorts of m+/db and db/db mice.

• Detecting loss in sensitivity and onset of polyneuropathy.

• Predict the onset of polyneuropathy

Abstract

We report a minimally invasive, synaptic transistor-based construct to monitor in vivo neuronal activity via a longitudinal study in mice and use depolarization time from measured data to predict the onset of polyneuropathy. The synaptic transistor is a three-terminal device in which ionic coupling between pre- and post-synaptic electrodes provides a framework for sensing low-power (sub μW) and high-bandwidth (0.1–0.5 kHz) ionic currents. A validated first principles-based approach is discussed to demonstrate the significance of this sensing framework and we introduce a metric, referred to as synaptic efficiency to quantify structural and functional properties of the electrodes in sensing. The application of this framework for in vivo neuronal sensing requires a post-synaptic electrode and its reference electrode and the tissue becomes the pre-synaptic signal. The ionic coupling resembles axo-axonic junction and hence we refer to this framework as an ad hoc synaptic junction. We demonstrate that this arrangement can be applied to measure excitability of sciatic nerves due to a stimulation of the footpad in cohorts of m⁺/db and db/db mice for detecting loss in sensitivity and onset of polyneuropathy. The signal attributes were subsequently integrated with machine learning-based framework to identify the probability of polyneuropathy and to detect the onset of diabetic polyneuropathy.

Introduction

Neural membrane opens to allow positively charged ions inside the cell and negatively charged ions out giving rise to action potential. As part of the sequelae, the positive charge of the nerve fiber rises sharply to reach +40 mV allowing the impulse to conduct down the nerve fiber. A series of action potentials allows the electrical impulse to propagate down the nerve. Under conditions of pathophysiology such as diabetic polyneuropathy, varying degrees of nerve conduction changes appear in the early asymptomatic stage (Erdogan et al., 2011). Electrophysiological abnormality as manifested by abnormal nerve conduction signal parameters i.e., conduction velocity vs. amplitude of the compound action potential is a hallmark of diabetic polyneuropathy. In db/db mice, weakening of compound motor action potential (CMAP) in the hind paw has been reported in response to stimulation of sciatic nerves (Heise et al., 2012). Analytically, nerve conduction studies are performed by applying depolarizing square wave electrical pulses to the skin over a peripheral nerve. Such exogenous supraphysiological stimulation generates sensory nerve action potential that is recorded at a point further along that nerve.

The diagnosis of structural damage in neuropathy can be classified into two broad groups – (a) quantitative sensory testing and (b) imaging. Quantitative sensory testing, also referred to as nerve conductivity study (NCS) performed via Semmes-Weinstein monofilament (SWM) or pressure-specified sensory device (PSSD), is subjective and influenced by various attributes such as location of damage, body fat, skin surface preparation and does not provide sufficient resolution to detect early onset of polyneuropathy (Zhang et al., 2014). Although often combined with electromyography (EMG) that uses global response to detect localized defects in a nerve, such NCS technique has poor reproducibility and sensitivity to small structural damage. Imaging methods to detect structural damage in nerve fibers include ultrasound (US) imaging and confocal microscopy. Among the two imaging methods, US imaging offers a non-invasive point of care approach with little or no patient education required prior to testing. Despite this advantage, US imaging suffers from poor resolution and cannot detect damage to small nerve fibers that are essential to detect the onset of peripheral neuropathy. Confocal imaging performed with molecular probes, is inapplicable at a point-of-care setting, and hence cannot be employed to routinely detect the onset, progression and repair of peripheral neuropathy (De Gregorio et al., 2018; Ruhdorfer et al., 2015). Nerve-interfacing electrodes such as the microelectrode array, flat interface nerve electrode and electrophysiology electrodes directly penetrate the nerve and result in tissue damage, repair and regeneration before being ready for use. In long term clinical studies, it is established that electrical signal attenuates over time and hence cannot be used in a longitudinal study for the detection of onset and progression of neuropathy (Gallardo et al., 2015; Kerasnoudis et al., 2015; Christensen et al., 2014). With estimates of about 5–10% of the general population and 50% of diabetics suffering from some form of neuropathy (retinal, peripheral) it is imperative to develop broadly applicable techniques that can detect damages to peripheral nerves to aid with early treatment and the discovery of novel therapeutics (Alshelh et al., 2016). The technical challenges in direct measurement of neuronal currents are due to the low power and high frequency structural components that attenuate rapidly within few 10s of microns. As discussed earlier, direct contact poses challenges due to tissue injury and immediacy of performing these measurements. Thus, any neuronal sensing framework should meet these contradicting requirements and provide high precision temporally accurate signal that mimics the action potential traveling through the nerve.

In this report, we demonstrate a sensing framework based on a three-terminal circuit where ionic currents between the input terminal and reference electrode affects redox current in a partially reduced conducting polymer used as the output terminal. This framework allows for measuring very small perturbations in the input terminal as a chronoamperometric response in the output terminal coupled by ionic capacitance of the volume enclosed between the three electrodes. By adjusting the geometrical attributes of the electrodes, ionic composition of the volume and capacitance of the conducting polymer in the output terminal, the coupling between the input and output terminal can be tailored to meet specified sensitivity, detection limit and spatial constraint. Owing to its structural resemblance to synaptic junctions, we refer to this architecture as a synaptic transistor, input and output terminals as pre-synaptic and post-synaptic terminals, and the volume enclosed between the electrodes as synaptic transistor cleft. For in vivo applications, the tissue under observation will become the pre-synaptic terminal and the chemoelectrical perturbations in the tissue will be pre-synaptic voltage and currents. The arrangement of components in the synaptic transistor and its application for in vivo sensing is shown in Fig. 1 (A-C). Our implementation of the synaptic transistor in this article uses conducting polymer electrodes for both the pre- and post-synaptic terminals as shown in Fig. 1A. A small voltage signal in the pre-synaptic electrode results in a current in the post-synaptic conducting polymer held at its partially reduced state. The post-synaptic current is observed due to the rapid ingress/egress of ions into/out of the conducting polymer to maintain electrochemical equilibrium with its environment. Owing to the properties of the synaptic transistor and partially reduced state of the post-synaptic terminal, we demonstrate the feasibility to detect cation concentration changes at a high bandwidth. (Fig. 1A, orange curve). A detailed physics of operation of this sensor is provided in SI 1.1. If this post-synaptic electrode is placed in vivo, the compound action potential passing through the sciatic nerve in response to a thermomechanical stimulus becomes the pre-synaptic input and the corresponding current in the conducting polymer electrode becomes the post-synaptic activity.