Differential Contribution of Sodium Channel Subtypes to Action Potential Generation in Unmyelinated Human C-type Nerve Fibers

The study was conducted according to the Helsinki Declaration and was approved by the local ethics committee (Munich, Germany). The patients gave written informed consent. The human nerves were obtained from four patients with biopsy of a sural nerve segment for histologic studies (neuropathy of unknown etiology) and from one patient who underwent an amputation above the knee due to peripheral vascular disease. None of the five patients reported (abnormal) pain states. The nerve segments (length 12–25 mm) were immediately put into fresh artificial cerebrospinal fluid solution containing 117 mm NaCl, 3.6 mm KCl, 2.5 mm CaCl², 1.2 mm MgCl², 1.2 mm NaH²PO⁴, 25 mm NaHCO³, and 11 mm d-glucose, pH 7.4 (bubbled with 95% O²–5% CO²). Several fascicles from the nerve segments were desheathed under a microscope and then held at each end by suction electrodes in an organ bath. One suction electrode was used to elicit action potentials, while the other one was used as a recording electrode. The distance between stimulating and recording electrodes was approximately 3 mm. The organ bath (volume approximately 2 ml) was continuously perfused with artificial cerebrospinal fluid solution at a flow rate of 8 ml/min and a temperature of 32°C. Drugs were applied for several minutes by addition to the perfusion system followed by a washout period. Lidocaine and mexiletine were purchased from Sigma-Aldrich (Taufkirchen, Germany); tetrodotoxin was purchased from ICN Biomedicals (Meckenheim, Germany).

Axonal excitability was measured using the threshold tracking technique, making use of the QTRAC program.§QTRAC is a flexible, stimulus–response data-acquisition program, originally written for studies of human nerves in vivo  16,17but also suitable for electrophysiologic recordings from isolated human peripheral nerve segments in vitro .18–20In the current study, QTRAC was used to record compound action potentials from peripheral C fibers, to generate stimuli, and to display the results. Isolated fascicles were stimulated with a linear stimulus isolator (A395; WPI, Sarasota, FL) with a maximal output of 500 μA. The stimulator was controlled by a computer via  a data-acquisition board (National Instruments NI PCI-6221; Munich, Germany). Two types of stimulation sequences (A and B) were used in the experiments (fig. 1). In both conditions, successive sweeps were delivered every 2 s.

• Stimulation sequence A: (no ramp) a supramaximal 1-ms stimulus (to monitor the maximal peak response amplitude); (ramp) a supramaximal 1-ms stimulus after a preceding ramp current of 300 ms duration and with 20% of the current amplitude used in (no ramp).

• Stimulation sequence B: (1) a supramaximal 1-ms stimulus (to monitor the maximal peak response amplitude); (2) a test stimulus automatically adjusted to maintain the C-fiber compound action potential at a constant amplitude (40% of the maximum); (3) a test stimulus after a preceding ramp current of 300 ms duration and with 50% of the current determined by stimulus parameter (2).

All data are given as mean ± SEM and represent changes from baseline (in percentages). The statistical uncertainty of our findings was quantified using the 95% confidence intervals (CIs).

Electrophysiologic studies were performed on isolated fascicles from short segments of human sural and tibial nerves. Electrical stimulation with current pulses of 0.1 or 1 ms in duration produced A- and C-fiber compound action potentials.20The C-fiber component had a peak amplitude of 0.6 ± 0.1 mV (mean ± SEM; n = 16; 95% CI, 0.5–0.8). Nerve fascicles were stimulated by a series of different current pulses (fig. 1). Within a few minutes after the onset of the stimulation protocol, stable compound C-fiber action potential amplitudes were elicited by the different stimulus parameters, and test substances were added to the bathing solution.

In a series of experiments, we tested the effects of tetrodotoxin on the peak height of action potentials recorded either by supramaximal stimulation with a 1-ms current pulse alone or by such a current pulse at the end of a slow current ramp (see fig. 1Afor stimulus protocol). A representative example of the observations is illustrated in figure 2. There was no clear difference in the peak height of the two types of action potentials in the normal bathing solution. However, in the presence of tetrodotoxin (90 nm), a strong reduction of the action potential peak height elicited at the end of the current ramp was observed (−73.6 ± 1.8%, mean ± SEM; n = 6; 95% CI, −69.1 to −78.1). In contrast, the peak height of the action potential produced by the 1-ms current pulse only (no ramp) was less affected (−22.8 ± 1.6%, mean ± SEM; n = 6; 95% CI, −18.7 to −26.9). We also observed that tetrodotoxin produced an increase in the latency to half of the maximal C-fiber compound action potential peak height determined by the stimulus without a ramp (fig. 2B, bottom panel).

In further experiments, we determined the maximal compound C-fiber action potential peak height, and the currents necessary to maintain the C-fiber action potential at 40% of the maximum by using either a 1-ms current pulse alone or by such a current pulse at the end of a slow current ramp (see fig. 1Bfor stimulus protocol). A representative experiment that demonstrates the effects of tetrodotoxin on these stimulus conditions is illustrated in figure 3. There was a clear difference in the sensitivity to tetrodotoxin (30 nm) of the maximal peak amplitude, elicited by a 1-ms supramaximal current pulse (fig. 3A, middle panel/test 100%; fig. 3B, left panel) and on the membrane threshold determined at the end of a 300 ms ramp current (fig. 3A, top panel/ramp). Quantitatively, only a slight reduction in the maximal peak amplitude was observed (−5.4 ± 3.5%, mean ± SEM; n = 10; 95% CI, −2.5 to −13.4). In contrast, the current necessary to produce an action potential with 40% of the maximal amplitude after a depolarizing ramp stimulation (fig. 3A, top panel/ramp) was elevated by 54.4 ± 11.8% (mean ± SEM; n = 10; 95% CI, 27.7–81.1). We also tested the effects of tetrodotoxin on a 1-ms current pulse necessary to maintain the C-fiber compound action potential peak amplitude at 40% of the maximum (fig. 3A, top panel/test 40%). This parameter, which is sensitive to changes in the axonal membrane potential,19,20remained almost unchanged (< 5%).

In another series of experiments, we tested the effects of lidocaine and mexiletine on the maximal C-fiber peak amplitude produced by a 1-ms current pulse and on the membrane threshold at the end of a depolarizing ramp current. A representative experiment is illustrated in figure 4. Both local anesthetics differentially affected these two parameters. Peak height (fig. 4A, middle panel/test 100%; fig. 4B, top panel) was slightly reduced by −7.0 ± 1.3% (mean ± SEM; n = 9; 95% CI, −3.9 to −10.0) in the presence of lidocaine (50 μm) and by −9.8 ± 1.4% (mean ± SEM; n = 5; 95% CI, −6.0 to −13.5) during application of mexiletine (50 μm). In contrast, the threshold current at the end of a depolarizing current ramp stimulation was increased by 40.3 ± 5.3% (mean ± SEM; n = 9; 95% CI, 28.0–52.6) during application of 50 μm lidocaine (fig. 4A, top panel/ramp). Likewise, a rise in threshold by 44.2 ± 11.9% (mean ± SEM; n = 5; 95% CI, 11.1–77.3) was observed in the presence of mexiletine (50 μm). We also tested the effects of lidocaine and mexiletine on a 1-ms current pulse necessary to maintain the C-fiber compound action potential peak amplitude at 40% of the maximum (fig. 4A, top panel/test 40%). This parameter remained almost unchanged (< 5%).

Multiple voltage-dependent sodium channels (Na^v) have been described in primary afferent, including nociceptive, neurons.1–4,21Suitable preparations for detailed electrophysiologic studies of Na^vin primary nociceptive neurons are dorsal root ganglion cells. Voltage-dependent sodium currents in such neurons can be observed with intracellular recording techniques. The aim of the current study was to characterize Na^vin the peripheral axons of unmyelinated human nerve fibers. These fibers are of very small diameter and are not accessible with intracellular recording techniques. A useful technique for functional studies on human C fibers is microneurography. This method has helped to gain much insight into the regulation of membrane excitability of nociceptive fibers in normal and pathophysiologic conditions.22Alternatively, information about the presence of voltage- or ligand-gated channels in the axonal membrane of human C-type nerve fibers can be obtained by a combination of threshold tracking techniques with pharmacologic tools using isolated segments of peripheral human nerve segments.10,20,23 

The nerve segments used in the current study were obtained from patients, and we do not have “control” preparations from healthy volunteers. However, data from patients with chronic ischemia and with neuropathy of unknown etiology apparently do not differ from each other. This indicates that the disease condition does not seem to affect our findings. However, a definite conclusion would require experiments on nerve segments from healthy human subjects. Compound C-fiber action potentials are composed of action potentials from afferent and efferent fibers. However, the contribution of efferent sympathetic axons to the total number of fibers is small (7–19%).24Also, tetrodotoxin-resistant Na^v1.8 have not been described for postganglionic sympathetic neurons or fibers,25and it is unlikely that the key observation, i.e. , different sensitivity to tetrodotoxin of action potentials evoked with or without a preceding ramp stimulus, could be related to electrogenesis of sympathetic efferent fibers only.

In previous electrophysiologic studies of action potentials in isolated human nerve segments, the compound C-fiber action potential produced by a 1-ms current pulse was found to be rather insensitive to tetrodotoxin even in concentrations of several micromolars.10,11This observation in human unmyelinated C-fiber axons is similar to the presence of tetrodotoxin-resistant action potentials in C-type nerve fibers of other mammalian species.26–29It can be explained by the functional activity of the sodium channel subtype Na^v1.830,31because only graded action potentials can be elicited in about 80% of knockout Na^v1.8 −/− mouse small C-type dorsal root ganglion neurons.32The current study describes a ramp stimulation protocol that reveals that also tetrodotoxin-sensitive sodium channels are functionally expressed in unmyelinated C-type peripheral human nerve fibers. It is possible that the molecular identity of this subtype of voltage-dependent sodium channel is Na^v1.7 because other investigated sodium channels, such as the Na^v1.4 and Na^v1.6 sodium channels, show faster kinetics of closed-state inactivation.9,12,33However, we cannot exclude that further subtypes of Na^vexpressed in dorsal root ganglion cells1,4contribute to membrane excitability after a ramp stimulus. The molecular identity could be deduced with more certainty by the use of subtype-specific blockers of tetrodotoxin-sensitive sodium channels. However, to our knowledge, such blockers are not publicly available at present.

It has been demonstrated that the adequate stimulus for activation of Na^v1.7 is a small, slow membrane depolarization, e.g. , a “generator” potential because Na^v1.7 is very slow in the kinetics of closed-state inactivation.9,12,33“Generator” potentials are produced by the transduction of (noxious) stimuli in the peripheral nerve endings. Furthermore, preferential immunoreactivity of peripheral nerve terminals to an antibody against Na^v1.7 has been described.34These facts may indicate that the functional expression of Na^v1.7 is restricted to nerve endings. However, in our studies over the past years on isolated human nerve segments, many voltage- and ligand-gated ion channels considered to be specific for terminals of unmyelinated axons were found to be present in the axonal membrane, too.19,20Therefore, we have used the axonal membrane of unmyelinated nerve fibers as a model for peripheral nerve terminals and have imitated a “generator” potential in the axonal membrane by a slow current ramp. Excitability of peripheral nerve fibers can be studied with high sensitivity using a computer-controlled threshold tracking technique (QTRAC).§We have used this program to generate a current ramp with duration of 300 ms and small current amplitude.9,12Both membrane threshold and action potential peak height determined at the end of this current ramp were strongly altered in the presence of tetrodotoxin (figs. 2 and 3).

There are two possible explanations for the high sensitivity to tetrodotoxin of action potentials elicited after a ramp stimulus. First, tetrodotoxin blocks a ramp current which enhances membrane depolarization during a small current ramp.9,12Such an effect can contribute to the rise in “threshold” current necessary to maintain the compound action potential at a constant amplitude (fig. 3). Second, it has been described that, compared with tetrodotoxin-sensitive channels, slow inactivation of tetrodotoxin-resistant Na^v(e.g. , Na^v1.8) is more complete at voltages reached with physiologic stimuli.35Therefore, it is likely that the sodium current during the action potential at the end of the ramp stimulus is carried by tetrodotoxin-sensitive channels (possibly Na^v1.7). This view can explain the strong reduction of the action potential peak height by tetrodotoxin as illustrated in figure 2. Another finding in the current study is that tetrodotoxin produced an increase in the latency to the action potential, although the peak height was not changed much (figs. 2 and 3). This effect is also explainable by the assumption that, similar to the cell bodies in dorsal root ganglia,36multiple sodium currents contribute to action potentials in unmyelinated, including nociceptive, nerve fibers. In nociceptive sensory neurons, the tetrodotoxin-sensitive sodium conductance activates quickly and returns to zero at the peak of the action potential. In contrast, tetrodotoxin-resistant channels account for the main sodium conductance at the peak of the action potential.36 

We also describe effects of lidocaine and mexiletine, which are used in the treatment of various types of neuropathic pain,37,38including erythromelalgia.13–15Lidocaine in concentrations of 50 μm reduced the axonal excitability at the end of a slow voltage ramp (fig. 4). This concentration is above the effective plasma concentrations of approximately 7.4–18.5 μm reached in the therapy of erythromelalgia15and above 6.2–33.9 μm used for patients with neuropathic pain after nerve injury39but far below the concentration of lidocaine necessary to induce a nerve conduction block (> 1 mm).40The actual concentration of lidocaine at the axonal membrane in the human nerve segments may be below 50 μm because of diffusion barriers. It is also possible that voltage ramps with longer duration could increase the efficacy of lidocaine. The low concentration of lidocaine necessary to reduce axonal excitability at the end of the slow voltage ramp is in accordance with the observation that analgesic concentrations of lidocaine suppress spontaneous injury discharge in corneal C fibers but do not block action potential conduction.41This finding is plausible under the assumption that injury of the nerve terminal produces slow depolarizing “generator” potentials. It has been demonstrated that the steady state inactivation and the development of slow inactivation of Na^v1.7 channels have a higher sensitivity to lidocaine as compared with blockade of the open channel.42Mexiletine is a structural analog of lidocaine, and the explanation of its effect in the current study seems to be identical to lidocaine.

In summary, our data indicate that the use of a slow current ramp, as suggested from observations using voltage clamp techniques,12is a useful parameter in electrophysiologic studies with the aim to reveal the function of Na^v1.7 in the axonal membrane of unmyelinated, including nociceptive, peripheral nerve fibers. The findings also support the point of view16,43that the analysis of action potentials and membrane threshold of peripheral nerve fibers will benefit from the use of multiple measures of axonal excitability. Finally, the presence of several subtypes of Na^vin human nociceptive nerve fibers suggests that the knowledge about their role in different pain states and the availability of specific sodium channel blockers may result in a mechanism-based pain therapy.

The authors thank Christa Müller (Technician) for expert technical assistance, Angelika Lampert, M.D. (Research Associate), and Gerrit ten Bruggencate M.D., Ph.D. (Emeritus; all from the Department of Physiology, University of Munich, Munich, Germany), for helpful discussions and critical reading of the manuscript.