Ethics

Protocol and consent forms were approved by the Committee for Ethics of the University of Munich and ethical approval to study drug effects in healthy volunteers was provided by the German Federal Institute for Drugs and Medical Devices. In accord with the declaration of Helsinki (Seoul 2008) all participants provided their written informed consent prior to participation in the study. ClinicalTrials registration is NCT01450865.

Clinical trial

Study design Volunteers were recruited for a randomised, double-blind, placebo-controlled, two-way, cross-over clinical phase I trial (RCT) to investigate the effect of oral flupirtine on the electrical excitability of peripheral myelinated axons. Exclusion criteria were: ongoing medication, prevailing organic disease, previous forearm trauma, primary organ failure, pregnancy or breast-feeding. All female participants were subject to a pregnancy test (Cyclotest, UEBE, Wertheim, Germany) prior to inclusion.

Subjects were assigned randomly to a study arm in accordance with an algorithm developed by the Institute of Medical Information Technology, Biometry and Epidemiology, University of Munich, Germany (IBE), stratifying for gender. Subjects received a set of sealed sequentially numbered pill-boxes each containing either flupirtine or placebo tablets to be taken prior to the recording sessions. Both the subjects and study physicians were blind to the group allocation.

Preliminary results indicated that a sample size of 12 would be needed to detect a decrease in the primary outcome measure (RRP) with 80% power at a significance level of 5%. With provision for drop-outs, 20 subjects were recruited, with an allocation ratio of 1:1.

Participants were evaluated on seven occasions in which motor axon excitability parameters and ischemic EMG signals were recorded (Additional file 1: Figure S1). The first 3 recording sessions, scheduled not less than 48 hours apart, were used to establish baseline values. The 4^th recording session comprised 2 sequential recordings of axonal excitability, one immediately prior to pharmacologic intervention and a second two hours after. EMG during ischemia was only recorded after medication. The 5^th session was a control recording performed 48 hours after the 4^th. The 6^th recording session was scheduled 7 ± 1 days after the 5th and comprised 2 sequential recordings immediately before and 2 hours after medication. The final (7^th) recording session was a control recording performed 48 hours after the 6^th. A pain questionnaire was assessed after the 3^rd, the 4^th and the 6^th session.

At the time of examination, subjects lay comfortably in a temperature controlled room.

Outcome measures The primary outcome measure was the relative refractory period (RRP) as determined with threshold tracking. All other indices were considered secondary outcome measures.

Threshold tracking All experiments were performed by the same examiner. Skin temperature was determined at the wrist at the end of each recording session using a thermocouple (Voltcraft, Hirschau, Germany).

The median nerve of the right hand was used for nerve excitability studies. A gel disc electrode (H92SG, Kendall-Arbo, Neustadt, Germany) positioned at the wrist between palmaris longus and flexor carpi radialis tendons served as the cathode. The anode was placed over the course of the median nerve approximately 2 cm proximal to the cathode. EMG from abductor pollicis brevis (APB) was recorded from one electrode situated over the motor point and a second electrode over the proximal interphalangeal joint.

Constant current pulses (DS5, Digitimer, UK) were used to determine electrical excitability in QTRAC using the TROND^XM4 protocol with same measures as for in vitro experiments (see above). The EMG signal was amplified (gain 200-500x), filtered (bandpass 0.3 Hz to 10 kHz) and sampled at 50 kHz (BNC-2120, National Instruments, Austin, USA).

Ischemia A cuff around the upper arm inflated to >200mmHg was used to examine flupirtine and placebo effects on ischemic and post-ischemic EMG from APB. Muscle fasciculations are known to appear in some subjects upon recovery from a 10 minute period of ischemia of the lower arm with the intensity and incidence increasing with the period of ischemia [23]. EMG signals were filtered, amplified and sampled at 20 kHz (ADInstruments GmbH, Spechbach, Germany). EMG was recorded continuously over a 21 minute period comprising one minute baseline followed by 10 minutes of ischemia and a further 10 minutes post-ischemia.

Volunteers rated the intensity of sensations before, during and after ischemia on a visual analogue scale (0–100 mm with 0 being no pain and 100 being maximal pain) and completed the Short Form McGill Pain Questionnaire (SF-MPQ) for sensations experienced during the first 5 minutes following release of the pressure cuff, i.e. post-ischemia. The SF-MPQ comprises eleven sensory and four affective pain descriptors that can be ranked in intensity from 0 = none to 3 = severe[24]. The sum of ranked values provides a sensory (SPRI; 0–33), an affective (APRI; 0–12) and a total pain score (TPRI; 0–45).

Data analysis

Axonal excitability Electrical excitability of myelinated axons determined both in vitro and in vivo were analyzed off-line with the MEM routine in QTRAC. Six parameters were taken directly from the QTRAC analysis, namely the strength-duration time constant, rheobase current, refractoriness at 2 and 2.5 ms, superexcitability at 5 and 7 ms and during the 90-100 ms hyperpolarizing and depolarizing current pulses of the threshold electrotonus. The relative refractory period was determined using custom written software in Igor Pro (WaveMetrics, Lake Oswego, USA) by determining the first zero crossing of an exponential fit to recovery cycle data points less than 5 ms.

Electromyography Rectified and integrated EMG was used to quantify continuous signals. EMG power was calculated by discrete fast Fourier transform (FFT) of EMG recorded in the time domain.

Statistics

Mean and standard deviation are used for population descriptors while mean and standard error of the mean are indicated for comparisons between groups. Normality was tested using Kolmogorov-Smirnov analysis. For parametric datasets Student’s paired t-test was used for pairwise comparisons. Unpaired datasets were compared using one-way ANOVA with post hoc Tukey-Kramer. Non-parametric datasets were analysed using Wilcoxon ranked sign test for paired data, and Kruskal-Wallis ANOVA with post hoc Mann–Whitney-U tests for non-paired data. The level of statistical significance is indicated throughout with * for p < 0.05 and ** for p < 0.01.

Chemicals and drugs

Triton 100, natural goat serum and the monoclonal antibody against panNaV were purchased from Sigma-Aldrich (Munich, Germany), the monoclonal antibody against peripherin was obtained from Chemicon (EMD Millipore Corporation, Billerica, USA). Anti-KCNQ2/Kv7.2 was kindly provided by M. Schwake (Kiel, Germany). Fluorescein-and rhodamine-conjugated donkey cross-affinity-purified secondary antibodies were purchased from Invitrogen (Paisley, UK). Aqua-Poly/Mount Coverslipping Medium was purchased from Polysciences Europe (Eppelheim, Germany). Flupirtine (in vitro) was purchased from Sigma (Taufkirchen, Germany) and XE991 from Biotrend (Wangen, Switzerland). The final concentration of test substances used in vitro was achieved by diluting stock solutions in the perfusing solution on the day of the experiment.

Blinded oral preparations of flupirtine 100 mg (AWD.pharma GmbH & Co.KG, Radebeul, Germany) and placebo tablets (7 mm capsules (Winthrop Arzneimittel GmbH, Frankfurt, Germany) were performed by the Pharmaceutical Department, University of Munich encapsulating both in gelatine (size 00, white opaque; Shionogl Qualicaps S.A., Alcobendas / Madrid, Spain) containing lactose, ferric oxide and food colouring E172 (all Fagron GmbH, Barsbüttel, Germany).

Immunohistochemistry of human A-fibre

Using polyclonal antibodies, Kv7.2 subunits could be identified at nodes of Ranvier in teased fibre preparations of human sural nerve (Figure 1). Nodes of Ranvier were identified using DIC images to delineate myelinating Schwann cells and co-labeling with mouse PanNa_V (voltage-gated sodium channels, Figure 1 vi-viii). Notably, Kv7.2 was also detected along unmyelinated fibres co-labeled with peripherin (Figure 1 iv, v).

In vitro assessment of human A-fibre excitability

Recordings were made from twenty-one individual fascicles of human sural nerve. The average compound A-fibre response to supra-maximal electrical stimulation (0.1 ms, 11.03 ± 5.43 mA) was 4.97 ± 1.65 mV and had a latency to half-maximum of 0.82 ± 0.05 ms.

Flupirtine (10-30 μM) did however reduce the RRP (ANOVA F (3:48) 8.61, p < 0.01) determined in human A-fibres in vitro and this in a concentration-dependent manner (see Figure 2B&C, Table 1; post-hoc Tukey-Kramer p < 0.01for 10 μM and 3 μM). Importantly, these effects could be reversed by bath application of the specific Kv7 channel blocker XE911 (10 μM; Figure 2 B&C and Table 1; ANOVA F (3:36) 7.46, p < 0.01; post-hoc Tukey-Kramer p < 0.05 for flupirtine 30 μM). Flupirtine (30 μM) also reduced refractoriness at 2.5 ms (Figure 2D; ANOVA F (3:48) 5.39, p < 0.01; post-hoc Tukey-Kramer p < 0.01). Similarly, flupirtine produced a concentration-dependent increase in the magnitude of post-spike superexcitability at 5 ms (Figure 2F and Table 1; ANOVA F (3:48) 11.56, p < 0.01; post-hoc Tukey-Kramer p < 0.01 for flupirtine 10 μM and 30 μM) and at 7 ms (Figure 2G; ANOVA F (3:48) 7.31, p < 0.01; post-hoc Tukey-Kramer p < 0.01 for flupirtine 30 μM). Since the RRP is determined from a fit to 3 or more data points, we consider this the most reliable index of changes in excitability.

In vivo assessment of human A-fibre excitability

Twenty volunteers (10 male, 10 female; age 28.3 ± 4.4 years) were randomly allocated to a study arm and received both placebo and flupirtine (200 mg p.o.) capsules to be taken in a cross-over design. The trial was enroled between October 2009 and August 2010. None of the subjects dropped out, all subjects received the allocated interventions and the follow-up was complete.

Statistical analysis of electrical excitability datasets from the initial three recording sessions revealed no time dependent differences in any of the excitability parameters and further comparisons between this ‘baseline’ dataset and each of the two pre interventional sessions also showed no differences amongst excitability parameters. Consequently, all five baseline datasets were pooled and used as the baseline values for statistical comparisons.

Consistent with observations made in vitro, RRP in motoaxons to APB was reduced 2 hours after oral flupirtine (200 mg p.o.; baseline 3.40 ± 0.07 ms versus flupirtine 3.23 ± 0.09 ms; p < 0.01 paired t-test; see Figure 3B&C and Table 2) but not placebo (3.45 ± 0.12 ms, p = 0.53). Oral flupirtine also reduced refractoriness at 2 ms and 2.5 ms (Figure 3D&E and Table 2). In contrast to the effect of flupirtine in vitro, the magnitude of superexcitability at both 5 and 7 ms determined in vivo was not affected by flupirtine (200 mg p.o; Figure 3F&G). Flupirtine and placebo were also both without effect on de- and hyperpolarizing threshold electrotonus in vivo (Table 2).

Recordings from APB during ischemia

Cuff-induced (200 mmHg) ischemia, induced a substantial increase in surface EMG activity in all 20 subjects (Figure 4A). EMG activity began on average 68.25 ± 8.72 s after cuff inflation and persisted for 288.65 ± 22.93 s. The total rectified integrated EMG during the baseline period of ischemia was 1.08 ± 0.12 V.s and this was reduced to 0.80 ± 0.08 V.s (p < 0.01, paired t-test) 2 hours after flupirtine (200 mg) but not placebo (0.98 ± 0.13 V.s; p = 0.29).

Interestingly, in eleven subjects an increase in surface EMG was also observed in the post-ischemic period (Figure 4A). Post-ischemic EMG activity began on average 125.80 ± 21.11 s after release of the cuff, persisted for 295.20 ± 27.85 s and the total rectified integrated EMG in the 10 minute period following release of the cuff was 0.85 ± 0.10 V.s. Consistent with its effect on EMG activity in the peri-ischemic period, flupirtine also reduced total integrated EMG activity in the post-ischemic period (200 mg p.o.; 0.58 ± 0.05 V.s; p = 0.023 paired t-test, see Figure 4D&E) and this effect was not significant following placebo (0.72 ± 0.09 V.s; p = 0.24).

Under baseline conditions, the surface EMG signal recorded in the post-ischemic period exhibited pronounced high frequency motor unit action potential discharges (MUAP; see insets in Figure 4A&D). The uniformity of MUAP shape and grouping of these discharges suggest that they arose from repetitive discharges in single motor units. The incidence of such high frequency burst discharges was higher in the post-ischemic period than during ischemia. This is evident in the running power spectral density plots shown below the raw EMG traces in Figure 4A&D. Following flupirtine, post-ischemic high frequency repetitive discharges are also evident in surface EMG recordings however the length of each burst was curtailed (Figure 4D, inset). The tendency for flupirtine to curtail high frequency motor unit discharge is shown in the normalised power spectral density plots (PSD; Figure 4C&F) where the relative power in the 150-200Hz band for post-ischemic discharge is reduced following flupirtine (Figure 4F).

Assessment of sensations during and after ischemia

Cuff-induced ischemia affects both motor and sensory axons. To assess the sensory component, a visual analogue scale (VAS) was used to rate the intensity of painful sensations during and after ischemia. The VAS comprised a 0 point anchor indicating no pain and an anchor at 100 as the most intense pain imaginable. Averaged across three baseline recordings, pain was rated at 31.1 ± 20.9 mm during ischemia and 33.2 ± 25.4 mm in the postischemic period (Figure 5A). Two hours after oral flupirtine (200 mg) the average ischemic pain rating was 23.2 ± 16.8 mm VAS, (p = 0.12 paired t-test) and 19.6 ± 20.3 mm in the post-ischemic period (-41%, p = 0.03). Oral placebo did not change either the ischemic (30.5 ± 20.6 mm VAS, -1.7%, p = 0.70 paired t-test) nor the post-ischemic pain rating (32.8 ± 27.1 mm VAS, -1.2%, p = 0.69; Figure 5A).

The SF-MPQ was used to assess the intensity and quality of sensations arising in the 10 minute period following release of the cuff. Under baseline conditions, total rating index in the postischemic period was 14.1 ± 1.5 with a sensory and affective component scores of 11.0 ± 1.2 and 3.1 ± 3.0 respectively (Figure 5B). Two hours after flupirtine (200 mg p.o.) the total intensity rating (−45% to 7.8 ± 1.0, p < 0.01, paired t-test) and sensory component scores were reduced (-48%, 5.8 ± 0.7, p < 0.01, paired t-test). In contrast, total intensity ratings were not affected 2 hours after oral placebo (12.0 ± 1.3, p = 0.25, Figure 5B). The SF-MPQ offers descriptors to qualitatively describe sensations. Under baseline conditions, the average rating of ‘throbbing’, ‘shooting’, ’stabbing’, ‘sharp’ and ‘hot-burning’ during ischemia was mild-moderate (Figure 5C). For the affective descriptors, only ‘sickening’ was rated as moderate. Two hours after oral flupirtine (200 mg p.o.) the ratings of ‘hot-burning’ (p < 0.01 Kruskal-Wallis) and ‘throbbing’(p = 0.04) were reduced from moderate to mild.

Strength and limitations

The primary strength of this study examining drug-induced effects on axonal excitability in vivo and in vitro is the ability to compare and thus judge target engagement and the efficacy of pharmacological compounds on peripheral myelinated axons. In particular, the ability to expose nerve segments to high concentrations provides a basis for the interpretation of subtle changes in excitability detected in people in situ, where dose is limited.

Segments of human nerve material were obtained at biopsy and are thus potentially pathological. Previous experiments indicate however that the electrical excitability of nerve fascicles obtained from patients at biopsy is not correlated in a systematic way with the underlying pathology [25, 26].

Generalizability

A common deficit encountered during the clinical and pre-clinical development phases of pharmacological compounds targeting peripheral nerve is the inability to determine directly the axonal effects in people. To address this, standardized measures of axonal excitability were used to examine the effect of flupritine on human peripheral nerve in vitro and in situ. Demonstrable changes in the excitability of myelinated axons were observed in vitro (Figure 2) and in motoraxons in situ (Figure 3). This confirmation of target engagement was shown to be therapeutically relevant with flupirtine suppressing ectopic axonal activity evoked by ischemia (Figure 4).

Interpretation

Kv7.2 subunits were identified nodally in large diameter myelinated axons in human sural nerve (Figure 1) consistent with previous reports using myelinated rat axons [6, 8, 9]. Kv7.2 expression was also evident along unmyelinated axons [27] where it is also postulated to regulate excitability (6) and suppress axonal discharge [28]. From voltage clamp studies performed on single nodes of Ranvier of myelinated axons from human nerve, it is possible that up to 30% of the slow potassium current may be active at rest [6]. Since, flupirtine shifts the voltage-dependent activation of cell-expressed Kv7.2 in the hyperpolarizing direction [29] it would be expected to hyperpolarize axons by increasing the number of open Kv7 channels at resting membrane potentials around −60 to −70 mV. For the human myelinated axons examined here, flupirtine (3-30μM) produced a concentration-dependent shortening of the relative refractory period and an increase in post-spike superexcitability in vitro, as well as a comparable reduction in the RRP in vivo (2 hours after oral flupirtine).. The increase in post-spike superexcitability seen in vitro (Figure 2C) was only apparent for flupirtine concentrations above 10 μM (Figure 2F,G). Previous reports indicate that the peak plasma concentration of flupirtine reaches values in the range of 5 to 6.5 μM, 2 hours after oral flupirtine 200 mg; [30–32]. This concentration is below that at which effects can be detected in vitro and potentially accounts for the lack of effect of flupirtine on the superexcitable phase in motor axons as well as the modest effect of flupirtine on the RRP in vivo (Figure 3B,C).

The modest reduction in the value of the relative refractory period in motoraxons to APB may well be secondary to a change in temperature, specifically a warming of peripheral nerve. Propofol was recently reported to shorten the relative refractory period in motoaxons and this effect was attributed to a non-specific increase in skin temperature [33]. Although no global differences in skin temperature were detected between baseline, placebo and flupirtine recording sessions, flupirtine could potentially lead to a warming of peripheral nerve either directly by relaxing vascular smooth muscle [34] or indirectly through suppression of activity in sympathetic neurones [35].

The only other study examining the effects of flupirtine on peripheral axons in people used a 400 mg oral dose and they reported no change in soleus H-reflex latency but a reduction in medium latency flexor reflex responses evoked with 5 pulses at 200 Hz [36]. It is possible that this effect of flupirtine reflects the ability of slow axonal potassium channels to suppress high frequency discharge.

Flupirtine is however also reported to act as an agonist at GABA_A receptors in DRG neurones [37]. While an agonist action of flupritine at GABA_A receptors may contribute to flupritine’s ability to alter spinal reflex latency, peripheral axons maintain a high intra-axonal chloride concentration and thus activation of GABA_A receptors alone cannot account for an increase in post-spike superexcitability or a shortening of the relative refractory period in peripheral myelinated axons.

Interestingly, post-spike recovery cycles recorded from myelinated sural nerve axons in vitro (Figure 1B) appear to lack the late phase of sub-excitability (between ca. 7 and 100 ms) which is otherwise prominent in recovery cycles from median nerve motor axons in vivo (Figure 2B). Recordings in people indicate that late sub-normality is typically smaller in sural nerve than it is in median nerve [38] and that, sural axons thus have less slow nodal potassium channels (Kv7) [38].

Post-ischemic EMG activity arises axonally [39, 40] and for a 10 minute period of ischemia affects predominantly myelinated axons [41, 42]. In myelinated sensory axons, ectopic activity produces paraesthesias and in motor axons fasciculations are observed [42], with the former being more susceptible due to a more pronounced persistent nodal sodium current [43]. During the period of ischemia, ectopic activity in myelinated axons is typically attributed to a protracted axonal depolarization arising chiefly from a reduction in electrogenic Na⁺/K⁺-ATPase activity and possibly a resultant accumulation of extracellular potassium, particularly in the adaxonal space [44]. Activation of slowly activating, non-inactivating Kv7 channels by flupirtine should adequately counter slow depolarization and indeed flupirtine effectively reduced total EMG activity during ischemia (Figure 4). Post-ischemic EMG activity comprises high frequency components (Figure 4) and microneurographic recordings from myelinated axons in people following ischemia also show prominent high frequency burst discharges ca. 200 Hz; [41, 42]. Although high frequency bursts still occurred in the post-ischemic period following flupirtine, there was a reduction in the burst duration, i.e. number of impulses (Figure 4). Flupirtine’s (200 mg p.o) curtailing of action potential burst length in human motoneurones is consistent with previous observations in rat peripheral myelinated axons using retigabine [6] and hippocampal CA1 neurones [45] and suggests that flupirtine may be beneficial in limiting the duration of ectopic bursts in peripheral axons.

Changes in the excitability of peripheral myelinated axons in people (Figures 2 and 3) were comparable with those seen in vitro using human sural nerve segments, suggesting that flupirtine affects Kv7 subunits expressed in peripheral nerve.