Activity-dependent hyperpolarization of human motor axons produced by natural activity


Corresponding author Professor David Burke: Prince of Wales Medical Research Institute, High Street, Randwick, NSW 2031, Australia. Email:


  • 1The changes in excitability of motor axons produced by natural activity were measured in six healthy subjects using voluntary contractions lasting 15 s, 30 s and 1 min, by recording the changes in stimulus current required to produce a compound muscle action potential of ≈60 % of maximum.
  • 2On cessation of the contractions there was a prominent increase in the current required to produce the target potential, accompanied by an increase in rheobase, a decrease in strength-duration time constant, and an increase in axonal supernormality. These changes indicate that the hypoexcitability was due to axonal hyperpolarization.
  • 3The activity-dependent hypoexcitability increased in depth and duration the longer the contraction. Following a 1 min contraction, it produced a 24 % increase in threshold, waning over 15 min. The hypoexcitability was greater than in cutaneous afferents tetanized to produce an equivalent rate-dependent stress.
  • 4It is concluded that natural activity results in substantial hyperpolarization of active axons and that, for similar discharge rates, the degree of hyperpolarization is greater in motor axons than cutaneous afferents. The greater effect of activity on the excitability of motor axons could be due to less inward rectification and less persistent Na+ conductance than in sensory axons. It is suggested that motor axons may therefore be more susceptible than cutaneous afferents to conduction block at sites of impaired safety margin for impulse conduction.

When axons conduct trains of impulses, they undergo hyperpolarization (Gasser, 1935), and this can lead to conduction block at sites of impaired safety margin for impulse conduction (Bostock & Grafe, 1985; Kaji & Sumner, 1989). Under these circumstances the conduction failure may be ameliorated by interfering with the activity of the electrogenic Na+-K+ pump, activation of which is largely responsible for the activity-dependent hyperpolarization (Bostock & Grafe, 1985; Gordon, Kocsis & Waxman, 1990; Morita, David, Barrett & Barrett, 1993). The safety margin for impulse conduction is normally high and the activity-dependent hyperpolarization is probably insufficient to jeopardize conduction in healthy axons, but focal demyelination can produce critical impairment of the safety margin, such that conduction failure occurs when the axon conducts trains of impulses (Bostock & Grafe, 1985). Indeed, the phenomenon of activity-dependent conduction block may explain fatiguability in patients with multiple sclerosis (McDonald, 1977; Waxman, 1988), and is probably important in patients with demyelinating peripheral nerve diseases such as Guillain-Barré syndrome, chronic inflammatory neuropathy and multi-focal motor neuropathy.

Activity-dependent hypoexcitability occurs in human cutaneous afferents when they conduct trains of impulses (e.g. Guisset, 1968; Lehmann & Tackmann, 1974; Applegate & Burke, 1989; Taylor, Burke & Heywood, 1992; Miller, Kiernan, Mogyoros & Burke, 1995; Kiernan, Mogyoros, Hales, Gracies & Burke, 1997b), and there have been studies of whether such trains can induce conduction failure in patients with a focal nerve lesion (Miller, Kiernan, Mogyoros & Burke, 1996; Kiernan, Mogyoros & Burke, 1997a). However, there are a number of biophysical differences between human motor and cutaneous afferent axons; for example, sensory axons have greater inward rectification (Bostock, Burke & Hales, 1994), lesser superexcitability and late subnormality following a single discharge (Kiernan, Mogyoros & Burke, 1996), longer strength-duration time constants (Panizza, Nilsson, Roth, Rothwell & Hallett, 1994; Mogyoros, Kiernan & Burke, 1996; Bostock & Rothwell, 1997) and a greater depolarizing threshold conductance, probably due to persistent Na+ channels (Bostock & Rothwell, 1997; for review see Burke, Kiernan, Mogyoros & Bostock, 1997). Some of these differences would tend to protect sensory axons from excessive hyperpolarization. Accordingly, it is likely that motor axons would undergo conduction failure more readily than cutaneous afferents when subjected to the same pathology (Waxman, 1996; Burke et al. 1997). There have, however, been few systematic studies of the changes in excitability of human motor axons produced by trains of impulses (Bergmans, 1970; Bostock & Bergmans, 1994), and most, if not all, have involved single motor units (indeed, largely one particular single motor unit that could be tracked faithfully despite large changes in its excitability). In addition, these studies have involved electrical stimulation, commonly at a high rate: the extent of the hyperpolarization produced by natural activity has not been documented.

The present study was undertaken to determine whether a voluntary contraction would alter the excitability of the active motor axons significantly and, if so, to determine its extent. A secondary goal was to demonstrate that any decrease in excitability was indeed due to axonal hyperpolarization. The results indicate that even relatively brief contractions can produce a significant increase in the threshold of motor axons due to activity-dependent hyperpolarization, and that this is greater than would occur in cutaneous afferents for the same discharge rates.


Experiments were performed in six healthy adult subjects of both sexes, aged 22–52 years, all of whom gave informed consent to the experimental procedures, which had the approval of the Committee on Experimental Procedures Involving Human Subjects, University of New South Wales.

A computerized threshold-tracking procedure (QTRAC, Institute of Neurology, London) was used to follow the excitability of motor axons in the median nerve innervating the thenar eminence before and after maximal voluntary abduction of the thumb. The median nerve was stimulated at the wrist and the intensity adjusted for successive trials by 2 % in order to keep the compound motor action potential (CMAP) of the thenar muscles at either 50, 60 or 70 % of maximum. For any one subject, the target CMAP size was set as large as was reasonable on the steeply rising phase of the stimulus-response curve, given that it would be necessary to remain well below the maximal output of the current source (50 mA) when axons were hypoexcitable (see below). Different stimulus combinations were used to allow measurements of threshold and supernormality, and off-line calculation of strength-duration time constant (τSD) and rheobase current (Irh).

The stimulus intensity required to produce the test CMAP (i.e. the ‘threshold’ for the CMAP) was measured using single stimuli of 0.1 and 1.0 ms duration (I0.1 and I1.0, respectively). From these data, τSD and Irh were calculated, using the formulae below (Weiss, 1901; Bostock & Bergmans, 1994; Mogyoros, Kiernan, Burke & Bostock, 1997). Strength-duration time constant reflects the rate of decrease of threshold current as stimulus duration is increased, and equates to chronaxie in Weiss's formulation (1901). Rheobase is the threshold current if the stimulus duration could be infinitely long.

display math

Supernormality reflects the phase of increased axonal excitability that follows immediately after the refractory periods, when the test stimulus is preceded by a single supramaximal conditioning stimulus and was sampled at a conditioning-test interval of 7 ms. The duration of the test stimulus was 0.1 ms. Because the maximal CMAP produced by the supramaximal conditioning stimulus was superimposed on the test CMAP, the conditioned potential was measured by the computer after on-line subtraction of the CMAP produced by the conditioning stimulus when given by itself. Measurement of the maximal CMAP also allowed the size of the target test potential to be adjusted for activity-dependent changes in the maximal CMAP, so that the test potential always remained the specified percentage of maximum. However, contractions resulted in minimal or no change in the CMAP produced by supramaximal stimuli. Stimuli were delivered once every second, rotating in sequence through the different stimulus channels.

Subjects performed maximal isometric voluntary efforts of 15 s, 30 s and 1 min duration against resistance provided by one of the authors. Adequate time was allowed between contractions for the measured parameters to have returned to control levels for 5–10 min before the next contraction was performed. Maximal contractions were used to ensure that all relevant motoneurones were active so that the submaximal test stimuli would then sample axons that had undergone excitability changes. (If the electrical stimuli had sampled larger axons preferentially while the contractions had been confined to small motoneurones, the study might have failed to demonstrate the true extent of the excitability change. For this reason, also, a larger-than-usual test CMAP was used, 50–70 % of maximum, mean 61 %.) Whether the efforts were truly maximal was not critical, and the experiment was therefore not complicated further by controls to measure this. In all experiments, skin temperature was measured at the wrist and over the second metacarpophalangeal joint and was kept > 32°C at both sites.

In three of the six subjects the changes in threshold of cutaneous afferents in the digital nerves of the index finger were measured following supramaximal stimulation of the digital nerves at 20 Hz for 1 min and, after full recovery, at 50 Hz for 1 min. The technical details are as described elsewhere (Kiernan et al. 1997a, b). In brief, the digital nerves were stimulated using ring electrodes around the proximal phalanx and the evoked compound sensory action potential was recorded using bipolar surface electrodes taped to the skin 4 cm apart, the distal electrode being at the distal crease of the wrist. The test stimulus was of 0.1 ms duration. The stimulating rates were chosen to encompass the discharge rates of motor axons in maximal voluntary contractions (see Discussion).


For each subject, threshold data were normalized to the pre-contraction threshold, the post-contraction increase above 1.0 indicating the extent by which it was necessary to increase stimulus current to produce the test CMAP. Supernormality was expressed as the decrease in stimulus current below the unconditioned intensity necessary to produce the target CMAP, normalized to the unconditioned test stimulus (i.e. the difference between the conditioned and the unconditioned test stimulus intensities, divided by unconditioned intensity). Accordingly, supernormality is plotted as a negative value (with a more negative value implying greater supernormality). The precontraction values for supernormality and strength-duration time constant were within those previously reported from this laboratory for CMAPs of equivalent size (Kiernan et al. 1996; Mogyoros et al. 1996): −17.2 ± 2.1 % and 384 ± 26 ms, respectively. Precontraction rheobase values were 3.36 ± 0.35 mA, of necessity greater than those for CMAPs of 30 % (2.46 ± 0.6 mA; see Mogyoros et al. 1996).

Changes in threshold

In all subjects, performance of a voluntary contraction reduced the excitability of motor axons, whether this was tested with stimuli of 0.1 or 1 ms duration. Figure 1 shows the data for one subject for the 1 min contraction and Fig. 2 shows mean data (±s.e.m.) for all six subjects. The discrete steps in the traces of Fig. 1 represent the tracking steps of 2 %. It is likely that the threshold change was maximal immediately after the cessation of each contraction. The gradual development of the threshold changes in Fig. 1A, B and C is a distortion due to the tracking steps used in the study: successive increments (or decrements) in threshold could be no more than 2 %. Accordingly, the extent of the post-contraction hypoexcitability is underestimated in this study.

Figure 1.

Changes in excitability in a single subject

The median nerve was stimulated at the wrist and a 70 % CMAP was tracked over the thenar muscles, using increments and decrements in stimulus intensity of 2 %. At 5 min, a 60 s voluntary contraction was performed by the subject. A, threshold changes measured using test stimuli of 1.0 and 0.1 ms duration, normalized to 1.0. B, threshold changes during the supernormal period measured using a 0.1 ms test stimulus, delivered 7 ms after a supramaximal conditioning stimulus. Supernormality is expressed as the decrease in current below the control threshold necessary to produce the target CMAP. C, strength-duration time constant, calculated from the threshold changes in A.

Figure 2.

Changes in excitability indices following contraction for 1 min

Post-contraction changes in threshold (A), supernormality (B), rheobase (C), and time constant (D). C and D were calculated using the 0.1 ms and 1.0 ms data from A. All data points represent mean ±s.e.m. for six subjects. For this figure, five consecutive data points were averaged.

The reduction in excitability (measured as an increase in threshold for the test CMAP) was greater the longer the contraction, there being both a higher peak threshold and a slower return to control levels (Fig. 3). Using the 0.1 ms test stimulus, the peak threshold increases were 29, 19 and 15 % for the 1 min, 30 s and 15 s contractions, respectively, and it took > 15 min, < 10 min and < 10 min for threshold to recover to the control level.

Figure 3.

Dependence of the decrease in excitability on the duration of contraction

Threshold changes, measured using 1.0 and 0.1 ms test stimuli (○, continuous line, respectively), for contractions lasting 15, 30 and 60 s. Each trace represents mean data for six subjects. The data in C are the same as in Fig. 2A, but without the smoothing created by averaging consecutive data points.

Changes in other measures of axonal excitability

In absolute terms, the increase in threshold was greater with the 0.1 ms test stimulus than with the 1.0 ms test stimulus but, when the precontraction thresholds were normalized to unity, the increase in threshold was proportionally greater with the 1.0 ms test stimulus (Figs 1–3). This indicates that rheobase current (another measure of axonal excitability, but one sensitive to changes in internodal properties) underwent qualitatively similar changes to the other threshold measures (Fig. 2C).

There were changes in the other excitability parameters that were sensitive to membrane potential. There was an increase in supernormality and a decrease in τSD (Figs 1 and 2), changes that are expected for axonal hyperpolarization (see Bostock & Bergmans, 1994; Kiernan et al. 1997b; Mogyoros et al. 1997). Importantly, these changes followed similar time courses as the threshold increased. The changes in these parameters were greater the longer the contraction, paralleling the greater change in threshold.

Changes in threshold of cutaneous afferents

To compare the activity-dependent hyperpolarization of motor axons with that of cutaneous afferents, cutaneous afferents were tetanized for 1 min in three of the six subjects, and the resulting change in threshold measured using identical threshold tracking techniques. Figure 4 contrasts the activity-dependent increase in threshold of motor axons, maximally 18 % in these three subjects (Fig. 4A), with that for cutaneous afferents stimulated at 20 Hz for 1 min (Fig. 4B) and 50 Hz for 1 min (Fig. 4C). In all three subjects the post-contraction hypoexcitability was less in cutaneous afferents with both stimulus rates. The increase in threshold for cutaneous afferents in Fig. 4B was maximally 10 % with return to control levels over 5 min, and in Fig. 4C it was maximally 13.6 % with return to control levels over 10–15 min. To illustrate the greater threshold change for motor axons the data from Fig. 4A are re-plotted in Fig. 4D, superimposed on the threshold for cutaneous afferents discharging at 50 Hz for 1 min (i.e. the data from Fig. 4C).

Figure 4.

Activity-dependent changes in threshold for motor axons and cutaneous afferents in three subjects

A, mean threshold change following a 60 s voluntary contraction of the thenar muscles. B and C, mean changes in threshold of cutaneous afferents after tetanic stimulation at 20 Hz for 60 s (B) and at 50 Hz for 60 s (C). D, comparison of the threshold change for motor axons (from A) with that for cutaneous afferents following the 50 Hz tetanic train (from C). Error bars represent ±s.e.m.


The present study has documented the threshold changes that occur in motor axons following a voluntary contraction, and has demonstrated that natural activity produces hypoexcitability that increases in depth and duration with the duration of the contraction. It could be expected that activity would result in axonal hyperpolarization, but the extent of this change has never previously been documented and comparison with the change in sensory axons has not been undertaken. These issues will be discussed below, together with the rationale for the conclusion that the hypoexcitability was due to axonal hyperpolarization.

The activity-dependent subexcitability is due to hyperpolarization

The excitability of axons can be altered by mechanisms other than those that directly change membrane potential: hyperventilation hyperexcitability is a classic example (Macefield & Burke, 1991; Mogyoros et al. 1997). However, supernormality and strength-duration time constant are sensitive to membrane potential, though for different reasons, and the fact that they underwent appropriate changes, paralleling those in threshold, indicates that the threshold changes were, indeed, due to a change in membrane potential - in this case, hyperpolarization.

In myelinated axons, the supernormal period is probably due to a capacitive discharge of current stored on the internodal membrane (Barrett & Barrett, 1982; Blight & Someya, 1985; Bowe, Kocsis & Waxman, 1987). The extent of supernormality varies inversely with membrane potential, increasing with hyperpolarization and decreasing with depolarization, because shifts in membrane potential alter internodal resistance, largely through effects on paranodal fast K+ channels. The strength-duration time constant is a nodal property that changes with membrane potential because of its dependence on a voltage-dependent conductance that is active at threshold (Bostock & Rothwell, 1997; Mogyoros et al. 1997), presumably a persistent Na+ current (Baker & Bostock, 1997).

Accordingly, the threshold increase was accompanied by appropriate changes in another measure of nodal properties (τSD) and a measure of internodal properties (supernormality) and it is a reasonable conclusion that it was due to axonal hyperpolarization. The mechanism responsible for hyperpolarization was not elucidated in the present study, but was presumably activation of the electrogenic Na+-K+ pump, as was the case in animal studies (e.g. Bostock & Grafe, 1985; Gordon et al. 1990; Morita et al. 1993). The long duration of the hyperpolarization would be surprising, but it is consistent with other human studies in which hyperpolarization, presumably due to pump activity, was produced by release of ischaemia (Bostock et al. 1994; Mogyoros et al. 1997) or by tetanization (Applegate & Burke, 1989; Bostock & Bergmans, 1994; Kiernan et al. 1997a, b).

Extent of the activity-dependent hyperpolarization

The threshold for cutaneous afferents increased by 10 and 13.6 % following stimulation at 20 and 50 Hz, respectively, for 1 min (Fig. 4). In comparison, the threshold increase for motor axons (15–29 %) would seem to be disproportionately large. The average peak discharge rate of motor axons innervating adductor pollicis is about 30 Hz in maximal voluntary contractions (Bellemare, Woods, Johansson & Bigland-Ritchie, 1983), and that for first dorsal interosseous is ∼21 Hz (Gandevia, Macefield, Burke & McKenzie, 1990). However, such rates are not sustained, falling to ∼15 Hz within 60 s in adductor pollicis (Bigland-Ritchie, Johansson, Lippold, Smith & Woods, 1983; Bigland-Ritchie, Dawson, Johansson & Lippold, 1986) and to ∼13 Hz within 30 s in first dorsal interosseous (Gandevia et al. 1990). Assuming that thenar motor axons behave like motor axons innervating other intrinsic muscles of the hand, the average discharge rate of motor axons over a 1 min maximal voluntary contraction would not have exceeded 20 Hz. It seems that some factor limits the extent of the hyperpolarization in cutaneous afferents: likely factors could be the greater inward rectification (Ih) in human cutaneous afferents than motor axons (Bostock et al. 1994), and the greater expression of ‘threshold’ channels on cutaneous afferents (Bostock & Rothwell, 1997; Baker & Bostock, 1997). Whatever the explanation, the extent of the activity-dependent hyperpolarization in motor axons following their voluntary activation is quite prominent and could be a significant limiting factor for impulse conduction if the safety margin was impaired.

Clinical implications

Conduction block in motor axons may be a prominent feature of acquired neuropathies such as the Guillain-Barré syndrome, chronic inflammatory neuropathy and multifocal motor neuropathy. Indeed, in multifocal motor neuropathy, clinical evidence of cutaneous afferent involvement is generally considered an exclusion. Nevertheless, in this condition and its variants, sural nerve biopsies sometimes reveal pathology in cutaneous afferents (Lewis, Sumner, Brown & Asbury, 1982; Bradley, Bennett, Good & Little, 1988; Pestronk et al. 1988), and the occasional biopsies of mixed peripheral nerve reveal widespread demyelination, without clear sparing of any axons, sensory or motor (Prestronk et al. 1988; Auer, Bell & Lee, 1989; Kaji et al. 1993). As argued elsewhere (Waxman, 1996; Burke et al. 1997), the biophysical differences between human cutaneous afferents and motor axons could protect cutaneous afferents from conduction block, but at the expense of a greater disposition to ectopic activity. The present finding that natural activity will result in substantial hyperpolarization of motor axons supports this view.

That natural activity can result in significant hyperpolarization and thereby lead to conduction block would explain why the extent of conduction block demonstrated in routine nerve conduction studies may not correlate well with clinical symptoms or response to therapy in patients with multifocal motor neuropathy. Stimulation of motor axons at 1 Hz could fail to reveal impaired conduction in some axons unable to maintain the higher discharge rates associated with voluntary effort.


This study was supported by the National Health and Medical Research Council of Australia.