In vivo evidence for reduced ion channel expression in motor axons of patients with amyotrophic lateral sclerosis

Abstract

Key points

• The progressive loss of motor units in amyotrophic lateral sclerosis (ALS) is initially compensated for by the reinnervation of denervated muscle fibres by surviving motor axons.

• A disruption in protein homeostasis is thought to play a critical role in the pathogenesis of ALS.

• The changes in surviving motor neurons were studied by comparing the nerve excitability properties of moderately and severely affected single motor axons from patients with ALS with those from single motor axons in control subjects.

• A mathematical model indicated that approximately 99% of the differences between the ALS and control units could be explained by a non‐selective reduction in the expression of all ion channels.

• These changes in ALS patients are best explained by a failure in the supply of ion channel and other membrane proteins from the diseased motor neuron.

Abstract

Amyotrophic lateral sclerosis (ALS) is characterised by a progressive loss of motor units and the reinnervation of denervated muscle fibres by surviving motor axons. This reinnervation preserves muscle function until symptom onset, when some 60–80% of motor units have been lost. We have studied the changes in surviving motor neurons by comparing the nerve excitability properties of 31 single motor axons from patients with ALS with those from 21 single motor axons in control subjects. ALS motor axons were classified as coming from moderately or severely affected muscles according to the compound muscle action potential amplitude of the parent muscle. Compared with control units, thresholds were increased, and there was reduced inward and outward rectification and greater superexcitability following a conditioning impulse. These abnormalities were greater in axons from severely affected muscles, and were correlated with loss of fine motor skills. A mathematical model indicated that 99.1% of the differences between the moderately affected ALS and control units could be explained by a reduction in the expression of all ion channels. For the severely affected units, modelling required, in addition, an increase in the current leak through and under the myelin sheath. This might be expected if the anchoring proteins responsible for the paranodal seal were reduced. We conclude that changes in axonal excitability identified in ALS patients are best explained by a failure in the supply of ion channel and other membrane proteins from the diseased motor neuron, a conclusion consistent with recent animal and in vitro human data.

Key points

• The progressive loss of motor units in amyotrophic lateral sclerosis (ALS) is initially compensated for by the reinnervation of denervated muscle fibres by surviving motor axons.

• A disruption in protein homeostasis is thought to play a critical role in the pathogenesis of ALS.

• The changes in surviving motor neurons were studied by comparing the nerve excitability properties of moderately and severely affected single motor axons from patients with ALS with those from single motor axons in control subjects.

• A mathematical model indicated that approximately 99% of the differences between the ALS and control units could be explained by a non‐selective reduction in the expression of all ion channels.

• These changes in ALS patients are best explained by a failure in the supply of ion channel and other membrane proteins from the diseased motor neuron.

Introduction

Amyotrophic lateral sclerosis (ALS) is a heterogeneous disease with the site of symptom onset and the rate of disease progression varying widely between individuals. In ‘classical’ ALS both upper and lower motor neurons are affected. Additionally there are two dominant hypotheses in which the disease process follows a dying‐forward or dying‐back course, but in either case lower motor neurons are progressively lost from affected motor pools (Eisen & Weber, 2001; Dadon‐Nachum et al. 2011; Kiernan et al. 2011; Simon et al. 2014; Vucic et al. 2014; Maglemose et al. 2017). However, denervated muscle fibres attract reinnervation by the surviving motor axons, so that motor units become larger as they become fewer. Indeed, the mechanism of compensation by reinnervation of denervated muscle fibres is so effective that it has been estimated that up to 80% of motor units are lost before a deficit becomes apparent clinically (McComas et al. 1971; Hansen & Ballantyne, 1978; Daube, 2000; de Carvalho & Swash, 2016). This process of denervation and reinnervation may be expected to increase the metabolic demands on surviving neurons as the motor pool becomes more reliant on rate coding as a means of grading force (Daube, 2000; de Carvalho et al. 2014).

Several mechanisms could disrupt the supply of essential proteins to the membrane of the distal axon. Cytoplasmic inclusions in axons are a pathological hallmark of ALS, and provide strong evidence for dysfunction of protein homeostasis (Kabashi & Durham, 2006; Bilsland et al. 2010; Sundaramoorthy et al. 2015; Webster et al. 2017; Brenner et al. 2018).

Deficits of axonal transport have been documented in mouse models of neurodegenerative disease including ALS (Millecamps & Julien, 2013; Brady & Morfini, 2017; De Vos & Hafezparast, 2017), and transport abnormalities have been identified in vitro in cell culture lines from familial ALS carrying C‐terminal KIF5A splice site mutations (Brenner et al. 2018). Furthermore, postmortem observations of the accumulation of phosphorylated neurofilaments in ALS indicate a role for axonal transport in the pathogenesis of ALS (Xiao et al. 2006; De Vos & Hafezparast, 2017). Such changes in the production and maintenance of axonal proteins would be expected to affect the excitability of the distal motor axon.

Non‐invasive axonal excitability studies have been used to study the mechanisms underlying pathophysiology in patients in vivo with various neuromuscular and neurodegenerative conditions including ALS (Vucic & Kiernan, 2006; Krishnan et al. 2009; Park et al. 2017). Using axonal excitability, an early study found evidence for a reduction in K⁺ conductances in patients with ALS, and this was reproduced in the rat in vitro using the K⁺ channel blockers tetraethylammonium and 4‐aminopyridine (Bostock et al. 1995). Several studies have reported prolongation of the strength–duration time constant, a finding attributed to an up‐regulation of persistent‐Na⁺ currents (Mogyoros et al. 1998; Kanai et al. 2006; Tamura et al. 2006; Vucic & Kiernan, 2006, 2010; Cheah et al. 2012). Furthermore, Kanai et al. (2012) reported that a prolonged strength–duration time constant was strongly associated with a shorter survival. The changes in axonal excitability evident in ALS were probably related to disease stage with the earliest changes responsible for increases in persistent‐Na⁺ currents, and then subsequent loss of K⁺ currents (Kanai et al. 2006). These studies put a focus on axonal ion channels as a source of abnormality, clinical features and potentially novel therapies (Park et al. 2015, 2017).

These previous studies of nerve excitability changes in ALS have been restricted to recordings of compound action potentials, making it difficult to relate them to changes at the level of the individual motor neuron, since the disease can alter both axonal threshold and motor unit size. The present study was undertaken to avoid this difficulty by studying single motor axons.

It will be argued that, while there are distinctive ion channel abnormalities in ALS axons, they are likely to result principally from a deficit in the supply and maintenance of ion channel proteins to the distal axon.

Methods

Ethical approval

Patients with a confirmed diagnosis of ALS (Awaji definite) were recruited prospectively from the NHMRC Sydney Health Partners ForeFront Motor Neurone Disease Clinic at the University of Sydney. Those with other conditions that could affect peripheral nerve function were excluded. Patients provided informed written consent prior to commencement of studies. Data for 31 single motor units from 21 patients with ALS (17 male, 4 female) were compared with our data for 21 single motor units from 13 healthy age‐matched controls (see Results), 17 from Trevillion et al. (2010). Ethical approval was obtained by the Human Research Ethics committee of the University of Sydney (no. 2014/056, 1 April 2014) and the studies conformed to the requirements of the Declaration of Helsinki (except for registration in a database).

Clinical assessment

All patients underwent a neurological assessment with formal testing of muscle power. Muscle strength was assessed using the MRC score (Medical Research Council, 1976). Thumb and finger abduction were used to test the abductor pollicis brevis (APB) and abductor digiti minimi (ADM) muscles, respectively. All patients with or without wasting had weakness of the tested hand.

The revised ALS functional rating scale (ALSFRS) was used as an overall measure of disease burden and measured out of 48, with lower scores denoting higher disease burden (Cedarbaum et al. 1999). Components that rely on hand function (the ‘Handwriting’, ‘Feeding’ and ‘Dressing and hygiene’ sections) were combined to provide a fine motor ALSFRS subscore out of 12. The progression rate was calculated using the formula: (48 – ALSFRS)/(Disease duration in months) (Labra et al. 2016).

ALS motor units were considered to come from moderately and severely affected muscles on the basis of the size of the compound muscle action potential (CMAP) of the parent muscle, >1 mV and <1 mV, respectively, in accordance with Kanai et al. (2006).

Electrophysiology

Recording set‐up

Motor unit recordings were obtained from the APB and ADM muscles of ALS patients. Disposable Ag–AgCl ECG‐type electrodes (WhiteSensor WS; Ambu A/S, Ballerup, Denmark) were secured to the skin over the muscle belly and on the corresponding proximal phalanx. The same type of electrode was used on the dorsum of the hand for signal ground. Wet‐gel Ag–AgCl electrodes (WhiteSensor 4500M‐H; Ambu A/S) were used for stimulation. The optimal location for the cathode was sought with a cloth‐covered Ag–AgCl searching electrode and the anode was sited approximately 10 cm proximal to the cathode and off the nerve under study. Skin temperature was monitored continuously throughout each experiment adjacent to the stimulus site at the wrist, and was kept at >32°C (Burke et al. 1999; Kiernan et al. 2001 a).

Single motor unit action potentials and compound muscle action potentials were amplified using a purpose‐built low‐noise amplifier (Howells et al. 2018; gain ×250; filter 2 Hz–2 kHz) before having mains frequency noise removed using a 50 Hz noise eliminator (Humbug noise eliminator; Quest Scientific, Vancouver, Canada). Isolation of single motor units is difficult in healthy subjects using Ag–AgCl surface electrodes, and the 17 units from Trevillion et al. (2010) were recorded using a double differential recording electrode with an active sensor (DE3.1, Delsys Inc., Natick, MA, USA). The additional four control units were isolated (with difficulty) using the same technique as in patients. These waveforms were sampled at 10 kHz by a data acquisition system (NIDAQ USB‐6221) which was under the control of a recording script within the QtracS software (TRONDNF.QRP; © ION, UCL, UK). This recording script also provided the command waveforms for control of a bipolar constant‐current stimulator (DS5; Digitimer Ltd, Welwyn Garden City, UK).

Motor unit number estimation

An estimate of the number of motor units was made using a CMAP scan and the MSCANFit method (Bostock, 2016; Farschtschi et al. 2017; Jacobsen et al. 2017). The CMAP scan is a finely graded stimulus–response curve, which plots the size of the compound muscle action potential in response to stimuli of decreasing intensity. The 200 μs stimuli began at a supramaximal intensity and each subsequent stimulus was reduced by 0.2% until no response could be detected. The MSCANFit procedure is a statistical method which estimates the number of motor units by assessing the size and variability in recruitment threshold of all of the motor units contributing to the compound response (Bostock, 2016).

Axonal excitability

Conventional axonal excitability recordings track changes in the stimulus required to produce a fixed fraction of the entire motor pool under study. The present study, however, focused on the properties of individual motor units. The threshold of the single motor unit was continually tracked by adjusting the 1 ms stimulus to produce an all‐or‐none response. Single units could be used only if their threshold was clearly distinguishable from those of other units. For threshold‐tracking purposes clear discrimination of a single unit was obtained when the next‐recruited units had thresholds that were substantially higher, or they were obviously smaller or of a different latency from the unit under study (Fig. 1). It was generally easier to discriminate the low‐threshold units in ALS subjects than in control subjects, because there were fewer units in ALS and those units were relatively large.

Figure 1. Threshold‐tracking an ALS single motor unit.

A, traces of single motor unit ‘all‐or‐none’ response to small changes in stimulus amplitude above and below the motor unit's threshold. Note the F‐wave at 40 ms, of identical shape to the direct response. B, the black horizontal line represents the unconditioned stimulus threshold for the single unit. The single motor unit recording consists of four parts: QT, charge–threshold relationship from which strength–duration time constant can be calculated (the red line in this section represents the stimulus intensity for stimuli of different durations); TE, threshold electrotonus (the red, blue, grey and green lines refer to the conditioned thresholds tested at various intervals of current strengths; +40, +20, −20, −40%, respectively); I–V, current–threshold relationship (the blue line refers to the stimulus threshold tested at the end of a 200 ms‐long conditioning current of varying strength (+50 to −100% of the unconditioned threshold); RC, recovery cycle (the light blue line refers to the test stimulus and response following activation of a single motor unit at various conditioning–test intervals). C, the coloured circles represent the response or non‐response to the stimuli in B. In this example the single unit response had an amplitude of ∼0.9 mV. The colours are the same as those in B (e.g. black circles are the responses for the unconditioned stimuli, and the red circles in QT are the responses to stimuli of different durations).

Threshold related measures

The stimulus required to elicit the all‐or‐none response of the single motor unit was tracked throughout the entire excitability recording and provided the control threshold for calculation of threshold changes and for setting conditioning stimulus intensities.

Strength–duration properties were recorded by measuring the unit threshold using rectangular pulses of different duration (0.2, 0.4, 0.6, 0.8 and 1 ms). The strength–duration time constant and rheobase were calculated using Weiss's law plotting the stimulus charge versus stimulus duration (Weiss, 1901; Bostock, 1983; Mogyoros et al. 1996).

Threshold electrotonus and the current–threshold relationship

Threshold electrotonus measures the excitability before, during and after 100 ms‐long subthreshold conditioning currents. The strength of the conditioning currents was set to ±20% and ±40% of the control threshold. The depolarizing currents (+20, +40%) probe the delayed outward rectification attributed to fast and slow K⁺ channels of the K_v1 and K_v7 families, respectively. The hyperpolarizing currents probe the inwardly rectifying current (I _h) which passes through hyperpolarization‐activated cyclic nucleotide gated (HCN) channels. The accommodation to depolarization responsible for the ‘sag’ in excitability of depolarizing threshold electrotonus, termed S2 accommodation, is primarily due to outward rectifying slow K⁺ channels. TEd_90–100 ms and TEh_90–100 ms represent the threshold reduction at the end of the 40% depolarizing and −40% hyperpolarizing conditioning currents, respectively, and are sensitive to changes in membrane potential (Kiernan & Bostock, 2000).

The current–threshold relationship is the threshold‐tracking analogue of traditional current–voltage measurements from ion channel electrophysiology (Kiernan et al. 2000, 2001 b). The threshold for activation of a single motor unit was tested at the end of 200 ms‐long conditioning currents which varied in strength from +50% (depolarizing) to −100% (hyperpolarizing) of the control threshold. The slope of the curve reflects accommodation to the current injection; the steeper the curve with depolarizing currents the greater the accommodation due to K⁺ currents, and the steeper the curve with hyperpolarizing currents the greater the accommodation due to I _h. The slope of the current–threshold relationship as the current passes from hyperpolarization to depolarization is denoted as resting current–threshold (I–V) slope and is the threshold analogue of resting input conductance.

Recovery of excitability following the discharge of a single motor unit

Following the discharge of a single motor unit, there is a characteristic pattern of excitability changes known as the recovery cycle (RC), with phases of refractoriness, superexcitability and late subexcitability. Immediately following a discharge the axon is absolutely refractory (inexcitable), then relatively refractory where the threshold is greater than the control threshold. Refractoriness gives way to a period of superexcitability where the threshold is lower than the control threshold, and then a period of late subexcitability where again the threshold is greater than the control. These fluctuations in excitability are generally complete within 200 ms. To characterise better the activity of slow K⁺ channels, the recovery of excitability following a double suprathreshold conditioning pulse was also recorded (Shibuta et al. 2010 b). From this, a measure that accurately reflects slow K⁺ activity (Shibuta et al. 2010 b) was calculated as the difference between the subexcitability following the double pulse and the subexcitability following a single pulse. This difference between subexcitability measured using double and single supramaximal stimuli is denoted here as RC(2‐1). The period of subexcitability following activation typically occurs some 30–100 ms after the conditioning pulse(s).

Abnormal Excitability Index

To quantify the extent of abnormality of the ALS single motor units, an Abnormal Excitability Index (AEI) was developed. The index reflects the key differences in excitability parameters in the present study between ALS and normal control (NC) units: resting I–V slope, TEh_{90–100 ms,%}, TEd_{90–100 ms,%} and superexcitability:

$$\mathrm{AEI}=\frac{\begin{array}{c}{Z}_{{\mathrm{TEd}}_{90\text{--}100\mathrm{ms},\%}}-Z_{\mathrm{resting}IV\mathrm{slope}}\hfill \\ -Z_{{\mathrm{TEh}}_{90\text{--}100\mathrm{ms},\%}}-Z_{\mathrm{superexcitability}}\hfill \end{array}}{4}$$

where $Z_{\mathrm{restingIVslope}}=\frac{\mathrm{restingIVslop}{\mathrm{e}}_{\mathrm{ALS}}-{\mu }_{\mathrm{restingIVslop}{\mathrm{e}}_{\mathrm{NC}}}}{{\sigma }_{\mathrm{restingIVslop}{\mathrm{e}}_{\mathrm{NC}}}}$ and ${\mu }_{\mathrm{restingIVslop}{\mathrm{e}}_{\mathrm{NC}}}\mathrm{and}{\sigma }_{\mathrm{restingIVslop}{\mathrm{e}}_{\mathrm{NC}}}$ are the mean and standard deviation of the resting I–V slope values for the control subjects, respectively. $Z_{{\mathrm{TEh}}_{90\text{--}100\mathrm{ms},\%}},Z_{\mathrm{superexcitability}}\mathrm{and}{Z}_{{\mathrm{TEd}}_{90\text{--}100\mathrm{ms},\%}}$ were calculated in a similar manner. The AEI was constructed such that membrane hyperpolarization or depolarization produced positive or negative AEI values, respectively. Essentially the AEI represents the average number of standard deviations that each of these key measures is away from the mean of the control data.

Mathematical modelling

The biophysical basis for the underlying changes in ALS motor units was explored using the ‘Bostock’ model of a human motor axon as implemented in the QtracP software (Kiernan et al. 2005; Bostock, 2006). A complete description of the model is presented in Howells et al. (2012). Briefly, the model consists of a node and internode which are coupled by the ‘Barrett–Barrett’ pathways through and under the myelin sheath (Barrett & Barrett, 1982). Threshold‐tracking studies probe the excitability of the nodal membrane as influenced by the internode, incorporating the properties of key ion channels, pumps and their interconnections. Transient and persistent Na⁺ currents are modelled at the node, while slow and fast K⁺ currents are included at both the node and the internode, and the inwardly rectifying current, I _h, is internodally located. The Na⁺/K⁺‐ATPase pump and ‘leak’ currents are also represented at both the node and internode.

The model was first fitted to the control single motor unit data to create a normal control model and then changes from the normal control model were explored by minimising the discrepancy between the model and the ALS single motor unit data.

Statistical analysis

Normality was assessed using the Shapiro–Wilk test. For normally distributed data, group data are presented as the mean ± standard error of the mean. Data that followed a log‐normal distribution are presented as the geometric mean ×/÷ geometric standard error of the mean (expressed as a factor). The remaining non‐parametric data are presented as the median (interquartile range; IQR).

Differences between the data for the control and patient groups (moderately and severely affected) were assessed using a one‐way ANOVA. Specifically, parametric data were analysed using an ANOVA with either Dunnett's or Tukey's multiple comparisons tests, and the non‐parametric data with the Kruskal–Wallis and Dunn's post hoc tests.

The MRC, ALS, ALSFRS‐fine scales are ordinal and as such were correlated with axonal excitability data using Spearman rank‐order analysis.

Results

All 21 patients had an Awaji‐definite diagnosis of ALS (de Carvalho et al. 2008). The mean disease duration from symptom onset (25.9 ×/÷ 1.1 months) and clinical phenotypes were consistent with a representative ALS cohort (upper‐limb onset, 62%; lower‐limb onset, 14%; bulbar onset, 24%) (Kiernan et al. 2011). At the time of measurement, 95% of patients were receiving riluzole, the effects of which on nerve excitability have been documented elsewhere (Vucic et al. 2013).

Recordings from 31 single motor units were made from APB (48%) and ADM (52%), to allow insights into potential differences related to the ‘split‐hand syndrome’ in ALS, a pathognomonic feature of ALS (Wilbourn, 2000; Eisen & Kuwabara, 2012; Menon et al. 2013, 2014). The excitability of ALS single motor units was compared to 21 single motor units from age‐matched healthy controls (ALS, median age, 54 years (IQR: 44–65 years); normal control (NC), median age, 40 years (IQR: 35–63.5 years); P = 0.13, Mann–Whitney U‐test). There was no significant difference in the temperature between ALS and control recordings (ALS, 33.8 ± 0.2°C; NC, 34.2 ± 0.2°C; P = 0.24).

The ALS single motor units were obtained from a muscle that had undergone significant atrophy when compared to healthy controls, as shown in Fig. 2 A (maximal CMAP: NC, 8.8 ×/÷ 1.1 mV; moderate ALS, 2.5 ×/÷ 1.1 mV, P = 3.8 × 10⁻¹⁰; severe ALS, 0.5 ×/÷ 1.3 mV, P = 2.2 × 10⁻⁹). This was also reflected in the number of surviving motor units in each muscle studied with an estimate obtained using the CMAP scan, of 12.6 ± 1.7 and 2.7 ± 0.8 motor units in the moderately and severely affected ALS patients groups, respectively (range 1–33) compared to normative data for this technique (greater than 70 units; Farschtschi et al. 2017). The average size of motor units was 699 ± 107 μV (moderately affected ALS) and 639 ± 193 μV (severely affected ALS). As noted in Methods, only four healthy control motor units were recorded using the same electrode montage as for the ALS units, and these were smaller than in ALS (30, 31, 50 and 220 μV). These findings indicate a substantial loss of motor units with larger surviving units, presumably due to reinnervation of denervated muscle fibres.

Figure 2. Maximal CMAP and single motor unit threshold‐related measures in ALS.

A, the maximal CMAP obtained from stimulation of the same nerve that the single units came from. ALS data were allocated into moderately (green, n = 20) and severely (blue, n = 11) affected groups on the basis of CMAP size, >1 mV and <1 mV, respectively. Control data (n = 21) are shown in red, and the CMAPs were significantly greater than the moderate and severely affected groups (P = 3.8 × 10⁻¹⁰ and P = 2.2 × 10⁻⁹, respectively). B, threshold for activation of single motor units was significantly higher in ALS subjects than in controls; P = 0.001. C, the rheobase of single motor units was significantly higher in ALS subjects than in controls; P = 0.002. D, there was no significant difference in strength–duration time constant between single motor units in ALS and controls (P = 0.8), unlike findings in earlier reports, indicating that the changes in rheobasic threshold are those expected for the change in threshold to a 1 ms stimulus, indicating that the changes in rheobasic threshold probably reflect the change in threshold to a 1 ms stimulus shown in B.

Axonal function

Recordings for single motor units were usually complete within 30 min, and were more readily achievable in ALS patients due to the presence of fewer (but larger) motor units, whose thresholds differed markedly from those of any adjacent units in the CMAP scan, making them easier to isolate than units in healthy controls. A typical threshold tracking recording for an ALS single motor unit is shown in Fig. 1. The ‘all‐or‐none’ response is clearly evident in Fig. 1 A, and the unconditioned and conditioned stimulus thresholds for activation are shown as black and coloured lines, respectively (Fig. 1 B). Some motor axons were excited by the 40% of threshold conditioning current used for depolarizing threshold electrotonus, something that does not normally happen with motor recordings from healthy nerve (Trevillion et al. 2007), suggesting axonal hyperexcitability.

Differences across the ALS ‘split hand’

Across the ALS cohort, there was a trend for a greater number of motor units in ADM than APB motor pools when assessed using the CMAP scan (APB, 4.1 ×/÷ 1.3; ADM, 8.3 ×/÷ 1.3; P = 0.06). Similarly, the maximal CMAP was larger in ADM than APB, though this was also not significant (APB, 1.1 ×/÷ 1.3 mV; ADM 1.7 ×/÷ 1.3 mV; P = 0.3). As such, four of the 16 ADM motor units and six of the 15 APB units were allocated to the ‘severely’ affected’ group. These trends are consistent with the ALS ‘split hand’ phenomenon, with preferential wasting of thenar muscles when compared to hypothenar muscles (Wilbourn, 2000; Eisen & Kuwabara, 2012; Menon et al. 2013). However, there were no significant differences in threshold, rheobase or strength–duration time constant of motor units recorded in the APB and ADM muscles. In addition, there were no significant differences in other axonal excitability measures between the APB and ADM muscles (Fig. 3). These findings are consistent with an earlier report on the absence of systematic differences in axonal excitability for different intrinsic muscles of the hand in studies using compound potentials (Menon et al. 2014).

Figure 3. Axonal excitability of ALS single motor units in ADM and APB.

Continuous lines represent the mean, and dashed lines the standard error of the mean. APB units (n = 15 units) are shown as filled blue squares and the ADM data (n = 16 units) as open blue squares. A, there were no significant differences in any of the parameters derived from the threshold electrotonus recordings of APB and ADM units. B, there was a trend toward greater superexcitability in the single motor units measured in APB than in ADM (P = 0.055) (B), but not when a multiple comparison correction was applied to the data (P = 0.69). C, none of the parameters associated with the current–threshold relationship were significantly different, as can be seen better in D.

The motor units of APB and ADM were therefore combined in subsequent analyses.

Threshold‐related measures

Rheobase and threshold for activation of single motor units were significantly increased in ALS when compared to controls (P = 0.002 and P = 0.001, respectively; Figs. 2 and 4). In contrast to earlier studies that used compound targets (Mogyoros et al. 1998; Kanai et al. 2006; Tamura et al. 2006; Vucic & Kiernan, 2006, 2010; Cheah et al. 2012), the strength–duration time constant was not longer in the single motor units of ALS subjects than in controls (P = 0.8; Figs. 2 and 4 and Table 1).

Figure 4. Relationship of threshold and strength–duration properties.

A, there was a strong inverse correlation of strength–duration time constant and rheobase for single motor units of controls (red), but this relationship was more variable for units from ALS subjects. Dashed ellipse indicates 95% confidence limit of the data and the line of best fit is shown unbroken. ALS moderately affected (green), y = 0.534 − 0.014x, R = −0.189, P = 0.43; ALS severely affected (blue), y = 0.533 − 0.02x, R = −0.706, P = 0.015; Controls (red), y = 0.79 − 0.157x, R = −0.810, P < 4 × 10⁻⁵. B, unit threshold vs. rheobase in single motor units of ALS subjects (moderately affected, n = 20; severely affected, n = 11) and controls (n = 21 units). Threshold is linearly correlated to rheobase in both ALS and NC and subjects: ALS moderately affected (green), y = −0.173 + 1.009x, R = 0.989, P = 6.4 × 10⁻¹⁴; ALS severely affected (blue), y = 0.199 + 0.951x, R = 0.991, P = 8.3 × 10⁻⁸; Controls (red), y = 0.219 + 0.800x, R = 0.977, P = 1.5 × 10⁻¹¹.

Table 1.

Excitability differences between single motor units from ALS and normal control (NC) subjects

		ALS		ANOVA/{Kruskal–Wallis} P
Parameter	NC (n = 21)	Moderate (n = 20)	Severe (n = 11)	NC moderate	NC severe
Threshold for single unit (log mA)*	2.60 [1.07]	3.70 [1.11]	5.17 [1.22]	0.013	0.0001
Rheobase (log mA)*	1.76 [1.09]	2.51 [1.11]	3.48 [1.23]	0.050	0.001
Strength–duration time constant (μs)	482 (32)	493 (26)	431 (26)	0.95	0.44
RRP (log ms)*	3.34 [1.05]	2.78 [1.03]	2.51 [1.04]	0.0078	0.0003
Superexcitability (%)	−22.0 (1.8)	−31.3 (2.5)	−42.1 (3.8)	0.013	0.0001
Subexcitability (%)	11.4 (0.8)	10.8 (1.3)	10.1 (1.6)	0.89	0.71
TEdpeak,%	62.5 (1.8)	68.7 (1.9)	73.5 (3.7)	0.060	0.0055
TEd40–60 ms,% †	46.6 {44.9–50.1}	55.7 {50.4–59.1}	57.5 {54.8–62.8}	{0.0004}	{<0.0001}
TEd90–100 ms,% †	42.0 {39.8–44.9}	48.2 {44.0–50.6}	50.9 {47.2–54.0}	{0.0046}	{0.0002}
TEh90–100 ms,%	−88.6 (4.1)	−114.6 (4.8)	−144.6 (14.6)	0.0097	0.0001
Resting I–V slope	0.667 (0.02)	0.547 (0.02)	0.491 (0.05)	0.0021	0.0002
Minimum I–V slope	0.341 (0.018)	0.270 (0.015)	0.251 (0.026)	0.010	0.0056
Hyperpolarizing I–V slope†	0.396 {0.343–0.439}	0.341 {0.269–0.443}	0.415 {0.332–0.552}	{0.27}	{1}

Data that are normally distributed are presented as mean (standard error of the mean). ^*Data that are log‐normally distributed (threshold, rheobase and RRP) are presented as geometric mean [geometric standard error of the mean]. ^†Non‐parametric data (TEd_peak,%, TEd_{40–60 ms,%}, TEd_{90–100 ms,%}, Hyperpolarizing I–V slope) are presented as median {interquartile range}. One‐way analysis of variance was performed for the normally distributed data using an ANOVA with Dunnett's post hoc test. Similarly, ANOVA was performed for the non‐parametric data using the Kruskal–Wallis test with Dunn's post hoc test.

Consistent with previous findings, rheobase was inversely correlated with the strength–duration properties for the control motor units. The relationship for the moderately affected ALS group was not significant, possibly reflecting a more heterogeneous population of motor units (Fig. 4 A; see Mogyoros et al. 1998). This will be addressed in the Discussion.

Polarization of the nodal and internodal membranes

The response to long‐lasting polarization is best observed during threshold electrotonus and current–threshold measurements (Fig. 5 A; Table 1), and they identified a reduction in outwardly rectifying conductances in the ALS units, more so for the more severely affected group (compare upper‐most blue with green and red traces in Fig. 5 A). This finding is supported by significantly greater threshold reductions across a number of measures in depolarizing threshold electrotonus: the peak of depolarizing threshold electrotonus (TEd_peak,%), 40–60 ms after the onset of depolarization (TEd_{40–60 ms,%}) and at the end of the 100 ms depolarization (TEd_{90–100 ms,%}). Similarly, in the current–threshold relationship there was a greater threshold change to depolarizing currents (see top right quadrant in Fig. 5 C). With depolarizing currents +50% of the control threshold, the threshold change was 56.0 ± 1.7% and 59.7 ± 2.2% for the moderate and severe ALS groups, and 48.4% ± 1.0 (P = 3.8 × 10⁻⁵) in control units.

Figure 5. Excitability of single motor axons in ALS.

Continuous lines represent the mean, and dashed lines the standard error of the mean. Data from the ALS severely affected, ALS moderately affected and control groups are shown in blue, green and red, respectively. A, threshold electrotonus recordings were significantly different for both depolarizing and hyperpolarizing current pulses (±20% and ±40% of control threshold). B, recovery cycle. The relative refractory period was significantly shorter and superexcitability significantly greater in ALS motor units. C, current–threshold relationship. The resting and minimal current–threshold slopes were significantly reduced in ALS motor units. The slope of the current–threshold relationship (C) is shown in D, and more clearly demonstrates the reduction in resting and minimum current–threshold slopes with disease progression (shown as Resting I–V slope and Min. I–V slope, respectively). There was a greater threshold change to depolarizing currents in the ALS data (top right in D). The threshold change to depolarizing currents +50% of the control threshold was 56.0 ± 1.7% and 59.7 ± 2.2% for the moderate and severe ALS groups, respectively, and 48.4 ± 1.0% (P = 3.8 × 10⁻⁵) in control units.

Similarly, inward rectification was reduced in the ALS units, with significant threshold increases in response to hyperpolarization evident during threshold electrotonus, TEh_{90–100 ms,%}; and in the current–threshold relationship (see lower left quadrant in Fig. 5 C). Hyperpolarizing currents −100% of control threshold produced threshold changes of −308.5 ± 15.3% and −339.8 ± 37.6% in the moderate and severe ALS groups, and −243 ± 9.7% in control units (P = 0.001). The slope of the current–threshold plot is plotted against threshold reduction in Fig. 5 D for each group. This is the threshold analogue of a conductance versus membrane potential plot. It revealed a progressive reduction in minimum and resting conductances (minimum and resting I–V slopes, respectively) and at all levels of polarization (from +50% depolarization through to −100% hyperpolarization), suggesting a reduction in multiple voltage‐dependent conductances across the range of membrane potentials studied.

Recovery cycle

Finally, the recovery of excitability following the discharge of a single motor unit was markedly different between ALS patients and controls with a significantly shorter relative refractory period and greater superexcitability in ALS (Fig. 5 B and Table 1).

Abnormal Excitability Index

Taken together, the pattern of change of excitability across single motor units in ALS resembled a pattern that can be induced by hyperpolarization of membrane potential (Kiernan & Bostock, 2000), with increased thresholds, ‘fanning out’ of threshold electrotonus, reduced inward and outward rectification in the current–threshold relationship and an increase in superexcitability following passage of an action potential. Indeed 87% of the ALS single motor unit recordings exhibited the ‘fanned out’ pattern of excitability, and the mean Abnormal Excitability Index for all groups was significantly different from each other with the moderately and severely affected axons progressively more abnormal than the normal control group (AEI: NC, 0.0 ± 0.2; moderate, 1.1 ± 0.2; severe, 2.5 ± 0.6. ANOVA: NC vs. moderate, P = 0.0056; NC vs. severe, P < 0.0001; moderate vs. severe, P = 0.0052; Fig. 6 A). The most positive values of the Abnormal Excitability Index represent the most abnormal recordings, which were from the more affected muscles.

Figure 6. Correlation of clinical variables with Abnormal Excitability Index (AEI) in ALS motor units.

Normal control (NC), moderately and severely affected ALS units are shown as red, green and blue circles, respectively. A, the Abnormal Excitability Index values for all groups were significantly different from each other with the moderately and severely affected axons progressively more abnormal than the normal control group (AEI: NC, 0.0 ± 0.2; moderate, 1.1 ± 0.2; severe, 2.5 ± 0.6. ANOVA: F(2,49) = 18.75; NC vs. moderate, P = 0.0056, NC vs. severe, P < 0.0001; moderate vs. severe, P = 0.0052). The most positive values of AEI represent the most ‘fanned‐out’ or abnormal recordings and roughly correspond to recordings from the more affected muscles. B, strength measured using the Medical Research Council scale was negatively correlated with AEI, with the more abnormal single motor units innervating weaker muscles (Spearman rank correlation of all ALS units: ρ = −0.51, P = 0.0034). C, the ALSFRS‐fine subscore was also negatively correlated with AEI, and the more abnormal the motor unit AEI the more impaired were fine motor skills (Spearman rank correlation of all ALS units: ρ = −0.55, P = 0.0019).

The Abnormal Excitability Index was negatively correlated with specific measures of hand function, such as the number of motor units, MRC score and ALS fine motor subscore (Table 2). Specifically, the more abnormal the excitability index, the greater the loss of fine motor skills (Fig. 6 B and C). However the Abnormal Excitability Index was not correlated with maximal CMAP size, the amplitude or threshold of the single motor unit studied. Similarly, it was not correlated with disease duration, total ALSFRS or progression rate.

Table 2.

Correlation of Axonal Excitability Index with clinical measures

	Axonal Excitability Index
Clinical measure	ρ	P
MUNE (log number of motor units)	−0.44	0.016
MRC	−0.511	0.0033
ALSFRS‐fine	−0.547	0.0019
Maximal CMAP (log mV)	−0.258	0.16
Unit size (log mV)	0.04	0.81
Unit threshold (log mA)	0.235	0.20
Symptom duration (months)	0.132	0.49
ALSFRS	−0.203	0.27
Progression rate	0.103	0.59

Correlation of axonal excitability index using Spearman rank correlation. MUNE, motor unit number estimation.

Within‐subject comparisons

In six of the ALS subjects, multiple recordings were made in different muscles. Specifically, in four subjects recordings from units in the APB and ADM muscles were made ipsilaterally. In another subject recordings from both APB and ADM were made bilaterally, and in the final subject, recordings from ADM were made bilaterally (see Table 3).

Table 3.

Within‐subject comparisons of ALS single motor units

Subject	Muscle	Symptom (months)	ALSFRS (max 48)	ALSFRS‐fine (max 12)	MRC (max 5)	CMAP (mV)	AEI
1	Left APB	17	39	9	3	0.15	0.59
	Left ADM	21	22	2	0	0.09	2.39
2	Left APB	14	43	10	3	1.38	1.12
	Left ADM	14	43	10	3	0.98	0.55
3	Left APB	108	36	5	4	5.03	1.32
	Left ADM	108	36	5	2	4.20	1.43
4	Right APB	8	45	10	4	2.64	2.38
	Right ADM	8	45	10	4	2.04	2.23
5	Right APB	25	38	4	3	0.86	4.44
	Right ADM	30	36	4	0	0.47	2.78
	Left APB	31	36	4	1	1.92	4.53
	Left ADM	31	36	4	1	0.999	2.55
6	Left ADM	11	42	8	4	4.88	−0.37
	Right ADM	17	36	4	3	1.92	−0.03

Multiple recordings were made from different muscles in six ALS subjects. In subjects 1–4, recordings from the APB and ADM muscles were made ipsilaterally. In subject 5 both APB and ADM were recorded bilaterally. In subject 6 single motor units from the ADM muscles were recorded bilaterally.

There were no significant differences in the excitability variables or the Abnormal Excitability Index for the APB and ADM muscles (n = 6) in the same subjects. Three APB and ADM muscle pairs had the same muscle strength as assessed by MRC; neither of these pairs nor the remaining pairs with different strengths demonstrated significant differences in the excitability variables or the Abnormal Excitability Index. Similarly, no significant differences were evident when the APB/ADM muscle pairs were assessed with similar and different ALSFRS fine motor subscores, though, for each comparison, the number of unit pairs was small.

Insights from mathematical modelling

Differences in the excitability of single motor units in ALS patients were explored using a mathematical model of the behaviour of human motor axons. The model was first fitted to the control single motor unit data (red filled circles in Fig. 7) to create a normal control model (red lines in Fig. 7). The changes from the normal control model necessary to reproduce the ALS recordings were then used to explore the underlying mechanisms of the abnormal ALS single motor units.

Figure 7. Mathematical model of the excitability of ALS single motor units.

The mean excitability recordings are shown as filled circles (ALS moderately affected, green; ALS severely affected, blue; controls, red). The blue, green and red lines represent the best‐fit outputs of the mathematical model for ALS severe, ALS moderate and control single motor units. A, threshold electrotonus. B, recovery cycle. C, current–threshold relationship. D, I–V slope (derived from the slope of the current–threshold relationship plot shown in C).

Given the overall pattern of changes, we started with whether membrane hyperpolarization could reproduce the differences between ALS and control units. The best reduction in discrepancy between ALS single motor unit data and controls was 37.3% (moderate) and 29.2% (severe), and this was achieved by hyperpolarization of resting membrane potential (from −81.6 to −83.7 and −84.2 mV). This suggests that hyperpolarization cannot explain the ALS data.

Previous studies have suggested abnormal expression of axonal membrane ion channels, and accordingly each of these conductances was explored individually. Variation of the expression of Na⁺ channels on the nodal membrane was unable to reduce the discrepancy between the normal control model and ALS motor units. Variation of the percentage of persistent Na⁺ channels alone did not account for any of the changes recorded in ALS patients. A reduction in K⁺ channel expression alone had a minor effect on the discrepancy between ALS motor units and controls. The maximal improvement in the fit produced by reducing slow or fast K⁺ channel activity was 5.5% (though when both slow and fast K⁺ were decreased together the maximal improvement was 19.5%). Taken together, these findings indicate that alterations in only Na⁺ or only K⁺ channel function cannot explain the observed changes in ALS.

An alternative mechanism is that diseased motor neurons might be incapable of supplying sufficient membrane proteins to the distal axon. Therefore, the expression of all modelled conductances was allowed to vary together. Unexpectedly, an equal reduction of Na⁺, slow and fast K⁺, HCN and ‘leak’ channels by 28.9% and 44.3% accounted for 96.1% and 92.5% of the discrepancy between the moderate and severe groups and the normal control model. However it is unlikely that each channel would be affected to the same degree (see Discussion), and when the expression of all modelled conductances and the pump current was allowed to vary independently, the discrepancy between normal control model and ALS data was reduced by 99.1% (moderate) and 98.5% (severe). The modelling results are shown in Figure 7, where the filled circles represent the mean data and the lines represent the best‐fit outputs for the mathematical model. The model was able to demonstrate the progressive abnormality in ALS axons (normal to moderate to severe, shown as red, green and blue, respectively) – increased ‘fanning out’ of threshold electrotonus (Fig. 7 A), increased superexcitability (Fig. 7 B), and decreased inward and outward rectification (lower left in Fig. 7 C and D and upper right in Fig. 7 C and D, respectively). Importantly, a reduction in all channel conductances, apart from the Barrett–Barrett pathway under the myelin sheath, contributed to the observed abnormal excitability in the ALS patients (moderate group: Na⁺, 24%; slow K⁺, 17%; fast K⁺, 26%; ‘leak’ channels, 35%; HCN channels, 32%; severe group: Na⁺, 22%; slow K⁺, 35%; fast K⁺, 35%; ‘leak’ channels, 43%; HCN channels, 34%). In contrast to the reductions above, there was an increase (16%) in the Barrett–Barrett conductance, but this occurred only in the severe group. An increase in the Barrett–Barrett conductance can be explained if there is a reduction in expression of axoglial proteins that anchor the myelin sheath to the axonal membrane. There were slight increases in the pump current (moderate, 13.3 pA; severe, 7.7 pA), but this was insufficient to offset the depolarizing effect of reducing K⁺ expression and the overall effect on resting membrane potential was a depolarization of 0.4 mV (moderate) and 0.8 mV (severe).

Reinnervation and the activity of individual ion channels

ALS is characterised not only by axonal degeneration, but also by reinnervation of denervated muscle fibres by surviving axons. On the assumption that the size of the individual motor unit is an indication of the extent of sprouting and reinnervation of denervated muscle fibres, correlations were sought with measures related to individual ion channels. There were no correlations of unit size with measures dependent on Na⁺, fast K⁺, or I _h channels. However, there was a clear correlation between unit size and measures of slow K⁺ activity. S2 accommodation and RC(2‐1) are the most specific axonal excitability measures of slow K⁺ function, and were strongly correlated in the ALS single units (Fig. 8 A). Both S2 accommodation and RC(2‐1) were positively correlated with the logarithm of unit size (Fig. 8 B and C), suggesting greater slow K⁺ activity with increasing unit size.

Figure 8. Correlation of measures of slow K⁺ activity with single motor unit size.

The moderately and severely affected ALS units are represented by green and blue circles, respectively. A, RC(2‐1) and S2 accommodation are the most specific measures of slow K⁺ function and are strongly correlated in the ALS single units (linear regression of all ALS units: y = 0.677x + 2.845, R = 0.89, P = 9.8 × 10⁻⁸). B, S2 accommodation was correlated with the single motor unit size (y = 10.15 × log(x) + 24.03, R = 0.49, P = 0.0087). C, RC(2‐1) was also correlated with single motor unit size (y = 10.52 × log(x) + 22.3, R = 0.57, P = 0.0033). Linear regression and 95% confidence intervals are represented as continuous and dashed lines respectively.

Discussion

The present study has established the behaviour and functional properties of single motor axons in patients diagnosed with ALS. The markedly abnormal alterations in excitability and thereby function cannot be attributed to relatively normal axons from small surviving motor neurons (Shibuta et al. 2010 a; Trevillion et al. 2010). These changes can, however, be explained by a failure of supply to the peripheral axon of channel and other membrane proteins from a diseased motor neuron. It is relevant that, in motor neurons derived from human induced pluripotent stem cells harbouring TARDBP and C9orf72 ALS mutations, there is a progressive decrease in Na⁺ and K⁺ currents (Devlin et al. 2015). A unique aspect of the present study is the evidence for such a defect in ALS from in vivo studies. The Barrett–Barrett conductance (while increased) is effectively a leakage conductance, dependent on the integrity of the paranodal seal and the supply of adhesion molecules. As such a reduction in the supply of proteins to the axonal membrane can account for the abnormalities identified in the present study.

Axonal excitability and ion channel homeostasis

Several mechanisms of ALS neurodegeneration could limit the turnover of ion channels and associated proteins on the membrane of the distal motor axon. Synthesis of ready‐made transmembrane proteins, packaging for transport, axonal transport, local synthesis, insertion into the membrane and failure of protein quality control could all be affected in ALS. Evidence for limited protein turnover may be seen in the cytoplasmic inclusions in upper and lower motor neurons that are a neuropathological hallmark of ALS (Al‐Chalabi et al. 2012; Blokhuis et al. 2013).

Decreased axoplasmic transport has been documented in mouse models of ALS and other neurodegenerative diseases (Bilsland et al. 2010; Millecamps & Julien, 2013; Brady & Morfini, 2017; De Vos & Hafezparast, 2017), and defects in intracellular transport have been identified in vitro in familial ALS (Brenner et al. 2018).

Local mechanisms have also been implicated in ALS, with a failure of protein quality control, local synthesis and insertion all impacting on protein homeostasis (Kabashi & Durham, 2006; Sundaramoorthy et al. 2015; Cestra et al. 2017; Webster et al. 2017).

Reinnervation and axonal sprouting

We had supposed that the extra membrane requirements for the expansion of surviving motor units in ALS might be a significant factor in the loss of ion channel function, but there was no relationship between the Abnormal Excitability Index and motor unit size. Membrane dysfunction was related to measures of fine motor control, but not to the expansion of the particular motor unit. However, measures of slow K⁺ function were correlated with the size of individual motor units. It is conceivable that this is a secondary phenomenon, with the larger ALS motor units compensating for the spontaneous activity of immature terminal sprouts by attempting to upregulate slow K⁺ currents. There are previous studies of axonal excitability in regenerating nerve by Moldovan and Krarup (2004a,b ) and (Moldovan et al. 2016), who presented evidence for axonal hyperpolarization which they attributed to increased activity of the electrogenic Na⁺/K⁺ pump, to maintain Na⁺ balance in the regenerating axons. We found no such evidence in ALS axons, and the critical difference is likely to be that their studies involved the properties of actively regrowing axons, not the properties of intact axons that have sprouted distally to reinnervate denervated muscle fibres.

Why was there no difference in the strength–duration time constant?

The absence of a change in the strength–duration time constant in the present study contrasts with the findings in earlier reports (Mogyoros et al. 1998; Kanai et al. 2006; Tamura et al. 2006; Vucic & Kiernan, 2006, 2010; Cheah et al. 2012). There is likely to be no single explanation for this discrepancy, but a number of issues need to be considered. First, the CMAP studies looked at a population of motor units, not just one, and there is likely to be heterogeneity in the properties of the pool. Secondly, the changes in the strength–duration time constant were difficult to demonstrate in some previous studies (e.g. Mogyoros et al. 1998; Tamura et al. 2006), and was reported to be significant only for patients with a preserved CMAP (Kanai et al. 2006). Thirdly, all patients in this study were on riluzole, and this is a Na⁺ channel blocker, preferentially affecting their persistent behaviour (Xie et al. 2011). Finally, the strength–duration time constant depends on the balance between persistent Na⁺, K⁺, leak conductances and passive membrane properties. It is conceivable that the turnover of K⁺ channels is faster than that of Na⁺ channels, so that initially only K⁺ channels are reduced, resulting in depolarization and an increase in the strength–duration time constant. This may provide further insight into the findings of Kanai et al. (2012) who found that the higher strength–duration time constants correlated with shorter survival in ALS patients. It is reasonable to presume that more rapidly degenerating units would have a greater impairment of axonal supply mechanisms (protein synthesis, axonal transport, etc.) and therefore a greater imbalance in axonal ion channels underlying the strength–duration time constant.

This study provides no simple explanation for hyperexcitability responsible for fasciculations. Indeed it provides evidence against a favoured hypothesis that persistent Na⁺ currents are increased. However, a simple decrease in K⁺ currents would be expected to produce increased excitability. We note however that the ectopic activity causing fasciculations occurs mainly in the motor axon terminals and sometimes proximally near the cell body (Roth, 1982; de Carvalho & Swash, 2013), not mid‐axon (Brigant & Mallart, 1982; Mallart, 1985), but we have no insights into how they might change to produce the hyperexcitability.

Clinical correlations

While the patients had a split‐hand syndrome clinically (and there was some neurophysiological support for this), motor axons innervating the thenar and hypothenar muscles did not differ significantly and there was no correlation between the abnormality in axonal excitability and the rate of disease progression.

Importantly, however, there were strong correlations with fine motor skills for the involved hand, but only when all of the motor unit data were included in the analysis. This will be discussed below.

The ‘Abnormal Excitability Index’ may represent a useful index for following the progression of disease within individual motor units.

The correlation with severity is consistent with a progressive dysfunction of the surviving motor neurons, and confirms the value of axonal excitability studies in following disease progress and any response to therapy (Turner et al. 2009).

Heterogeneity of moderately affected ALS motor units

There are several reasons why correlations between the commonly used clinical variables and axonal excitability parameters were not evident in the moderately affected ALS single motor units. First, these widely used clinical parameters may not accurately reflect the state of the motor pool: maximal CMAP is confounded by the reinnervation of denervated muscle fibres and will not accurately reflect the number of motoneurons, while MRC and ALSFRS‐fine are affected by loss of descending drive from upper motor neurons. Perhaps the cleanest functional measure of the state of the lower motor neuron pool is the motor unit number estimate. Secondly, it is likely that the moderately affected ALS group have a heterogeneous population of single motor units, some relatively normal, some undergoing degenerative changes. This might explain why the expected correlation between strength–duration time constant and rheobase was not evident in the moderate group. Further work needs to be performed to follow individual motor units longitudinally to see how they transition from normal to abnormal excitability. The application of high‐density surface EMG using electrode arrays to measure axonal excitability measurements for a number of units should make this task easier to perform (Sleutjes et al. 2018).

Conclusion

A disruption in protein homeostasis is thought to play a critical role in the pathogenesis of ALS, and this study presents the first in vivo evidence for a non‐selective reduction in ion channel expression in patients. It is not surprising that a diseased motor neuron may have difficulty maintaining a supply of axonal proteins to the membrane surface throughout its length.

Additional information

Competing interests

H.B. receives royalties from University College London from sales of the Qtrac software used in this study.

Author contributions

J.H., M.C.K. and D.B. conceived and designed the study. All authors contributed to critical analysis and interpretation of the data, and drafting of the manuscript. All authors have read and approved the final version of this manuscript and agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. All persons designated as authors qualify for authorship, and all those who qualify for authorship are listed.

Funding

This work was supported by a National Health and Medical Research Council of Australia Project grant (no. 1107749), funding to Forefront Program Grant (no. 1037746) from the National Health and Medical Research Council of Australia (NHMRC). J.H. is the recipient of a Bill Gole MND Fellowship from the Motor Neurone Disease Research Institute of Australia. J.M. was supported by the International Federation of Clinical Neurophysiology (IFCN) Research Scholarship, Becas‐Chile Scholarship (CONICYT) and Clinica Alemana de Santiago. N.G. is supported by a postgraduate scholarship from the National Health and Medical Research Council (Australia), Australian and New Zealand Association of Neurologists and Muscular Dystrophy NSW.

Biography

James Howells received his PhD from the University of Sydney in 2014 for studies on the biophysical mechanisms underlying the behaviour of human myelinated axons, and in particular the role of the hyperpolarization‐activated current (I _H) in axonal excitability. Since then he has focused on mechanisms underlying neurodegeneration in motor neuron disease and related conditions. His other interests include mathematical modelling of the peripheral nervous system, assessment of upper motor neuron dysfunction using transcranial magnetic stimulation, and developing novel techniques to study the physiology and pathophysiology of the human nervous system.