Potassium currents dynamically set the recruitment and firing properties of F-type motoneurons in neonatal mice

Abstract

In neonatal mice, fast- and slow-type motoneurons display different patterns of discharge. In response to a long liminal current pulse, the discharge is delayed up to several seconds in fast-type motoneurons and their firing frequency accelerates. In contrast, slow-type motoneurons discharge immediately, and their firing frequency decreases at the beginning of the pulse. Here, we identify the ionic currents that underlie the delayed firing of fast-type motoneurons. We find that the firing delay is caused by a combination of an A-like potassium current that transiently suppresses firing on a short time scale and a slowly-inactivating potassium current that inhibits the discharge over a much longer time scale. We then show how these intrinsic currents dynamically shape the discharge threshold and the frequency-input function of fast-type motoneurons. These currents contribute to the orderly recruitment of motoneurons in neonates and might play a role in the postnatal maturation of motor units.

the second postnatal week is a milestone for the development of the mouse neuromuscular system. During this week, postural activity is acquired, allowing the locomotor behavior to switch from crawling to walking (Jiang et al. 1999). The firing behavior of motoneurons is heterogeneous, and two modes of discharge initiation have been observed during this week (Leroy et al. 2014; Pambo-Pambo et al. 2009; Russier et al. 2003). For liminal current pulses, the discharge starts at the current onset in one-third of spinal motoneurons (immediate-firing pattern) whereas it is delayed in the two other thirds (Leroy et al. 2014; Pambo-Pambo et al. 2009).

We recently showed that motoneurons innervating slow-contracting fibers (S-type motoneurons) selectively display the immediate-firing pattern in neonates, whereas those innervating fast-contracting fibers (F-type motoneurons) display the delayed-firing pattern (Leroy et al. 2014). Indeed, the immediate-firing motoneurons have a smaller input conductance, a lower rheobase, and a longer afterhyperpolarization (AHP) than the delayed-firing motoneurons (Leroy et al. 2014). This is in keeping with electrical differences between S- and F-type motoneurons in adult cats (Burke 1981; Zengel et al. 1985), rats (Beaumont and Gardiner 2002; Button et al. 2006), and mice (Manuel and Heckman 2011). Moreover, the delayed-firing motoneurons display a larger dendritic tree with more branching points and longer dendritic paths than the immediate-firing motoneurons. The same morphological differences have been observed in adult S- and F-type motoneurons (Cullheim et al. 1987). Finally, the immediate-firing motoneurons express ERRβ but do not express chondrolectin. This was suggested to be the molecular signature of S-type motoneurons (Enjin et al. 2010). In sharp contrast, none of the delayed-firing motoneurons expresses ERRβ, suggesting that they are F-type motoneurons, and the largest ones, i.e., those with the largest rheobase, do express chondrolectin. In the same line, we showed that MMP-9, known to be selectively expressed by the most vulnerable motoneurons in amyotrophic lateral sclerosis, i.e., those innervating the fast-contracting and fatigable motor units (Kaplan et al. 2014), is expressed in the largest delayed-firing motoneurons and not in the immediate-firing ones.

The aim of the present work was to identify the ionic currents that underlie the delayed-firing pattern and to investigate their functional consequences on the recruitment and the discharge properties of F-type motoneurons. We demonstrated that the delayed-firing pattern is caused by a combination of an A-like potassium current that acts on a short time scale and a slowly inactivating potassium current that delays the discharge on a much longer time scale. We also showed that these two potassium currents dynamically set the recruitment threshold of F-type motoneurons in neonates and shape their frequency-input (F-I) function. Potassium currents allow the recruitment of S-type motoneurons prior to the F-type ones during the critical second postnatal week. Additionally, they might also play a role in the postnatal maturation of motor units.

MATERIALS AND METHODS

Animals and slices preparation.

The experiments were performed in accordance with European directives (86/609/CEE and 2010-63-UE) and the French legislation. They were approved by Paris Descartes University ethics committee. Slices were prepared as described by Lamotte d'Incamps et al. (2012). Six- to 10-day-old C57BL/6J mice from both sexes (Janvier labs) were anesthetized with an intraperitoneal injection of 0.1 ml of pentobarbital sodium (25 mM). An intracardiac perfusion was performed using ice-cold low-Na solution (concentrations in mM: 3 KCl, 1 NaH₂PO₄, 230 sucrose, 26 NaHCO₃, 0.8 CaCl₂, 8 MgCl₂, 25 glucose, 0.4 ascorbic acid, 1 kynurenic acid, 2 Na-pyruvate) bubbled with 95% O₂ and 5% CO₂ (pH 7.4). After decapitation, a laminectomy was performed at about 4°C in the same solution. After sectioning the roots, the spinal cord was transferred to a K-gluconate solution (130 K-gluconate, 15 KCl, 0.05 EGTA, 20 HEPES, 25 glucose, 1 kynurenic acid, 2 Na-pyruvate, adjusted to pH 7.4 with KOH), then embedded into a 2% agar solution at 38°C (prepared with K-gluconate solution), which was then cooled down as quickly as possible. After solidification, the agar block containing the spinal cord was glued in the chamber of the slicer, and 400-μm-thick slices of the lumbar spinal cord were cut in K-gluconate solution. The slices were cut obliquely (35°) relative to the axis of the spinal cord to conserve the integrity of the motoneuron axons and thus allow the antidromic identification of the motoneurons. The slices were transferred into artificial cerebrospinal fluid (ACSF) containing (in mM) 130 NaCl, 2.5 KCl, 2 CaCl₂, 1 MgCl₂, 1 NaH₂PO₄, 26 NaHCO₃, 25 glucose, 0.4 ascorbic acid, 2 Na-pyruvate, bubbled with 95% O₂ and 5% CO₂ (pH 7.4). The transfer was made in ACSF at 34°C. After 30 min the slices were brought to room temperature (18–24°C).

Electrophysiology.

The recording chamber was continuously perfused with ACSF at a rate of 1–2 ml/min, at room temperature. The slices used were those containing a ventral rootlet of sufficient length to be mounted on a suction stimulation electrode, a glass pipette with a tip size adapted to the diameter of the rootlet (40–170 μm) and filled with ACSF. Single biphasic stimulation of the VR (1–50 V, 0.1–0.3 ms) was used to elicit antidromic action potential in the recorded cells of the ventral cord. The motoneurons were found in the ventral horn and patched in whole cell configuration under visual control using a video-camera (Scientifica, Uckfield, UK). We targeted large cells in the ventral horn with the long soma axis larger than 20 μm. Patch pipettes had an initial open-tip resistance of 3–6 MΩ. The internal solution contained (in mM) 140 K-gluconate, 6 KCl, 10 HEPES, 1 EGTA, 0.1 CaCl₂, 4 Mg-ATP, 0.3 Na₂GTP. The pH was adjusted to 7.3 with KOH, and the osmolarity to 285–295 mOsm.

An AxoClamp 700B (Molecular Devices) amplifier was used for data acquisition. Whole cell recordings were filtered at 3 kHz, digitized at 10 kHz using a CED 1401 acquisition system and monitored using Signal 5 software (Cambridge Electronic Design, Cambridge, UK). In current-clamp mode, bridge resistance and capacitance compensations were applied. In voltage-clamp mode, series resistance compensation was used (50–80%). Identification of the cell as a motoneuron was based on the recording of an antidromic action potential following stimulation of the ventral root.

Prior to any other drug, synaptic currents were suppressed by adding 2 μM 2,3-dioxo-6-nitro-1,2,3,4-tetrahydrobenzo(f)quinoxaline-7-sulfonamide (NBQX) to block the AMPA receptors, 50 μM d-(−)-2-amino-5-phosphonopentanoic acid (d-APV) to block NMDA receptors, 1 μM strychnine to block glycinergic receptors and 3 μM gabazine to block GABAergic receptors. All synaptic blockers were purchased from Abcam (Cambridge) as well as TTX, oxotremorine, and XE991 dihydrochloride. Tetraethylamonium (TEA), 4-aminopyridine (4-AP), and Cd²⁺ were acquired from Sigma (St. Louis).

Analysis.

We retained for analysis motoneurons exhibiting a resting potential equal or below −50 mV and an overshooting action potential. Access resistance ranged from 8.5 to 20 MΩ. We discarded motoneurons whose resting potential or series resistance varied more than 5 mV and 5 MΩ, respectively. Liquid junction potential was not corrected to allow comparison with previous studies. Voltage-clamp recordings were leak-subtracted as follows. Each voltage jump was preceded by a 10 times smaller jump (leak pulse) to measure the leak current. Five successive records were averaged. The response to the leak pulse was then multiplied by 10 and subtracted from the voltage-jump response. Input conductance is the inverse of the slope of the current-voltage (I–V) curve obtained by injecting small 500 ms pulses of currents (−100 pA to +20 pA, 30-pA steps repeated 10 times). The rheobase was searched by applying a series of 5-s square pulses of increasing intensity 30 s apart. The pulse intensity was increased by 50 pA up to the intensity that elicited at least one action potential. We also measured the recruitment (I_on) and derecruitment (I_off) currents to which, respectively, the first and the last action potential was fired during a 0.1 nA/s current ramp (average of 3 trials separated by 40 s). Single action potentials were elicited by 1-ms square pulses (1–10 nA). The amplitude and relaxation time constant of the AHP were measured on the averaged trace of 30 successive trials. The relaxation time constant of the AHP was determined using a monoexponential fit on the first half of the relaxation. Several properties of the action potential throughout trains elicited just above rheobase were also measured (width at half-height, maximum velocity of the repolarization, and peak of the AHP). Values in the text are displayed as average ± SD. We used Wilcoxon matched-pairs signed rank tests to calculate P values displayed in the text.

RESULTS

Delayed-firing pattern is due to two potassium currents that act at two time scales.

Figure 1A1 shows a motoneuron that displays the delayed-firing pattern in response to a long square pulse (5 s) at an intensity (4 nA) just above rheobase. The motoneuron depolarized slowly before reaching its spiking threshold 2.5 s after the pulse onset. Then the discharge frequency slowly increased from 4 to 21 Hz. This is due to the slow underlying membrane depolarization. To understand what are the underlying currents responsible for the delayed firing, we switched to voltage-clamp mode (0.2 μM TTX was added in the bath to block the sodium current) and applied voltage steps from a holding potential of −95 to −50 mV (the voltage reached at the pulse onset in current-clamp mode, Fig. 1A1). During the voltage jump, the motoneuron displayed an outward current with two distinct components (Fig. 1A2, dark trace). The first component quickly activated then inactivated and developed about 0.3 nA at its peak (filled arrowhead). This fast outward current likely explains why the motoneuron was not able to discharge in current-clamp at the pulse onset. The second component of the outward current was weaker (only 0.1 nA at the peak, open arrowhead). It displayed a slow activation and an even slower inactivation. This slow inactivation accounts for the slow depolarization that progressively brings the motoneuron towards its firing threshold. It is therefore likely responsible for the long delay before firing onset. It is also likely responsible for the increase in firing frequency throughout the rest of the pulse. When the voltage step was further increased to −20 mV (above the spiking threshold), the peaks of the fast and slow components increased and they were superimposed on a steady outward current (dashed line). This steady current corresponds to the activation of the delayed rectifier current at a higher voltage than the threshold voltage for spiking.

Fig. 1.

Delayed-firing motoneurons exhibit a fast and a slow outward currents. A1 and B1: voltage-responses of a delayed (A1)- and an immediate-firing (B1) motoneuron to a long-lasting current-pulse at liminal intensity (current-clamp mode). Bottom: injected current; top: voltage response. A2 and B2: leak-subtracted current responses of the same delayed (A2) and immediate (B2) firing motoneurons to voltage-pulses from −95 mV to either −50 mV (black trace) or −20 mV (gray trace). Bottom: voltage command; top: current response. Note the fast (filled arrowhead) and the slow (open arrowhead) outward currents in A2. They are virtually lacking in B2 (black trace).

Figure 1B1 shows another motoneuron that displays the immediate-firing pattern. This motoneuron discharged immediately at pulse onset in response to a current intensity (1.4 nA) close to rheobase. In addition the discharge displayed a different frequency adaptation: after an initial doublet at the pulse onset, the firing frequency decreased. When switching to voltage-clamp, virtually no outward current was visible at the threshold (−50 mV) (Fig. 1B2, dark trace). This explains why this motoneuron fired at pulse onset. When the voltage step was increased to −20 mV, a strong delayed rectifier was again present (4.9 nA, plateau on gray trace). We made similar observations in 15 motoneurons (12 delayed-firing and 3 immediate-firing motoneurons).

The fast and slow components observed in the delayed-firing motoneurons proved to be two independent currents. Only the fast component was sensitive to the level of the holding potential preceding the depolarizing jump as illustrated in Fig. 2A. A voltage step from −95 mV to −30 mV produced a fast outward current of more than 0.6 nA (dark trace), while a step from −50 mV to −30 mV elicited a fast outward current of less than 0.2 nA (dashed trace). This observation was consistently made in three delayed-firing motoneurons. To investigate the impact of the fast current on the discharge of the delayed-firing motoneurons, we induced firing with a current pulse that was just larger than the rheobase (1 nA in the example shown in Fig. 2, B1 and B2, current-clamp recording). In these conditions, the first spike latency was longer (77 ms, Fig. 2B1) following a hyperpolarizing prepulse (voltage held at −80 mV) than following a depolarizing prepulse (8 ms, Fig. 2B2, voltage held at −50 mV). The longer latency was due to the fact that more fast outward current was available because of its prior deinactivation by the hyperpolarizing prepulse. This observation was made in five delayed-firing motoneurons: following a prepulse to −80 mV, the first spike latency was 82 ± 15 ms whereas it decreased to 14 ± 8 ms (n = 5, P = 0.03) following a prepulse to −50 mV (Fig. 2B3). Moreover, in voltage-clamp mode, only the fast outward current was reduced (to 35 ± 14%, n = 5), after a bath application of 5 mM 4-AP (Fig. 2C). The slow current remained the same (94 ± 5%, n = 5). The functional effect of bath application of 5 mM 4-AP was also analyzed in current-clamp mode. In the example shown in Fig. 2D, the latency of the first spike decreased from 100 ms (Fig. 2D1) to 5 ms (Fig. 2D2) after 4-AP application. Overall, following 5 mM 4-AP application (Fig. 2D3), the latency of the first spike decreased from 33 ± 10 to 8 ± 2 ms (n = 5, P = 0.04). In addition, the width of the spikes was increased (see insets in Fig. 2, D1 and D2) as expected when the A-current is blocked (Viana et al. 1995). All together, the deinactivation of the fast current upon hyperpolarization and its suppression by 5 mM 4-AP strongly suggest that the fast current is a potassium A-current (Lape and Nistri 1999; Russier et al. 2003).

Fig. 2.

The fast outward current is an A-like current. A: voltage jumps to −30 mV from different holding potentials (−95 mV: solid lines; −50 mV: dashed lines) in a delayed-firing motoneuron (voltage-clamp mode). Bottom: voltage command; top: leak subtracted current-response. B1 and B2: voltage responses from the same motoneuron following a hyperpolarizing (−0.8 nA, B1) or a depolarizing (+0.2 nA, B2) prepulse (note that the pulse inducing the discharge remains unchanged at 1 nA). Bottom traces: injected current; top traces, voltage response. Arrowheads point at the first spike latency. B3: latency of the first spike in response to a suprathreshold current injection in two situations: when the membrane potential of the motoneuron prior to the pulse was hyperpolarized to −80 mV or maintained at −50 mV (dashed line: motoneuron in B1 and B2). C: voltage-clamp responses before (dashed line) and after (solid line) 15 min of 5 mM 4-AP. D1 and D2: voltage responses from the motoneuron to the same 0.8 nA pulse before (D1) and after (D2) 5 mM 4-AP application. Arrowheads point at the first spike latency. The first spike in D1 and D2 is enlarged to show the effect of 5 mM TEA. D3: plot of the first spike latency of five different motoneurons before and after 5 mM 4-AP application (dashed line: motoneuron in D1 and D2).

While the A-like current was selectively abolished by 4-AP, the remaining slow component was strongly reduced by 20 mM TEA (Fig. 3A), indicating that the latter component was also carried by potassium ions. This experiment was repeated in four delayed-firing motoneurons in which bath application of 20 mM TEA reduced the slow potassium current by 70 ± 21% (n = 4). Neither 50 μM oxotremorine nor 50 μM XE991 (not shown) had an effect indicating that this current was not a M-current. It resembled rather the slowly activating and even more slowly inactivating potassium current described by Luthi et al. (1996) in CA3 pyramidal cells of rat hippocampus. It was responsible for the long delay of the discharge that we observed in our experiment in current-clamp mode (Fig. 3B1) since this long delay and the increasing firing frequency were totally abolished by bath application of 20 mM TEA (Fig. 3B2). Such a high dose of TEA blocked not only the slow-inactivating potassium current, but also the A-current (Aguayo and Weight 1988; Denton and Leiter 2002) explaining why there was no short delay left at the pulse onset. Indeed the delayed-firing pattern was somehow converted into immediate-firing pattern [no delay in the first spike, firing frequency that does not increase anymore but rather tends to decrease, enlarged spikes—compare inset in Fig. 3, B1 and B2—consistent with our previous observation that immediate-firing motoneurons display larger spikes than delayed-firing motoneurons (Leroy et al. 2014)]. Moreover, the rheobase was lowered from 1 nA to less than 0.4 nA following the application of TEA (Fig. 3B2), indicating that these potassium currents largely set the excitability of F-type motoneuron. TEA application was tested in five cells where it suppressed the delay of firing and lowered the rheobase in all tested cells (1.35 ± 0.32 to 0.75 ± 0.15 nA, n = 5, P = 0.03).

Fig. 3.

The slow outward current is blocked by TEA. A: slow outward current activated by a long-lasting voltage step from −65 to −15 mV in a delayed-firing motoneuron (voltage-clamp recording). Gray line: control recording with 5 mM 4-AP to abolish the fast outward current. Black line: recording after application of 20 mM TEA. Bottom: voltage command; top: leak subtracted current responses. B1: response of a delayed-firing motoneuron to a threshold stimulation in control condition (no drugs, resting potential −67 mV). Note the long delay before firing onset and the increasing firing frequency. Bottom: injected current; middle: voltage response; top: instantaneous firing frequency (in Hz). B2: same motoneuron recorded during bath application of 20 mM TEA (same arrangement as B1). The resting membrane potential was unchanged (−67 mV) but the rheobase current was decreased. The motoneuron discharged now at the pulse onset and the discharge frequency decreased during the pulse. The first spike in B1 and B2 is enlarged to show the effect of 5 mM TEA.

Potassium currents dynamically shape the F-I function of F-type motoneurons.

Figure 4, A and B, shows the most typical responses observed in immediate-firing motoneurons (S-type motoneurons) that are not endowed with the slow potassium current described above. At rheobase current, this motoneuron discharged at the pulse onset with a few spikes (Fig. 4A1). At higher frequencies, it discharged more regularly and displayed spike frequency adaptation (Fig. 4, A2 and A3). The F-I curve derived from long pulses (triangles in Fig. 4B) superimposed with the F-I curve derived from the ascending branch of a slow triangular ramp of current (gray open squares in Fig. 4B). In the example shown in Fig. 4B, the response to the triangular ramp displayed a clockwise hysteresis: the recruitment current (0.06 nA) is lower than derecruitment current (0.32 nA, Fig. 4B2). The majority of immediate-firing motoneurons (12 of 20) displayed such a clockwise hysteresis (I_on = 0.6 ± 0.5 nA vs. I_off = 0.7 ± 0.6 nA; n = 12, P = 0.003). The remaining eight immediate-firing motoneurons do not display any hysteresis in their F-I function as exemplified in Fig. 4C: the recruitment and derecruitment currents were less than 50 pA apart (0.8 ± 0.4 vs. 0.8 ± 0.4 nA, n = 8, P = 0.2) and the ascending and descending branches were superimposed (Fig. 4C2).

Fig. 4.

Patterns of discharge of immediate-firing motoneurons. A1–A3: responses of an immediate-firing motoneuron to long-lasting pulses of increasing amplitude. Bottom: injected current (square pulses in nA); middle: voltage response (in mV); top (A2, A3 only): instantaneous firing frequency (Hz). Close to rheobase (A1), the motoneuron discharged a doublet at the pulse onset and then emitted a few spikes during the pulse. When the current intensity was increased to 0.3 nA (A2), the discharge was irregular. For higher intensities (0.5 nA in A3), after an initial frequency adaptation, the discharge was very regular. B: response to a slow triangular ramp of current in the same motoneuron (slope ± 0.1 nA/s). B1: bottom, injected current; middle, voltage response; top, instantaneous firing frequency (Hz). The dashed vertical lines indicate the recruitment (I_on) and the derecruitment currents (I_off). B2: instantaneous firing frequency plotted in function of the injected current during the ascending (empty squares) and descending (filled circles) ramps. The stationary F-I curve calculated from the square-pulses (Fig. 1) is also plotted (red triangles). For each pulse, the frequency is averaged over the last second. C1 and C2 same as B1 and B2 but for an immediate-firing motoneuron that did not display an hysteric F-I function in response to a slow triangular current ramp.

The potassium currents play a major role in setting the recruitment threshold and the F-I function of F-type motoneurons. Figure 5 illustrates the responses of a delayed-firing motoneuron to long rectangular pulses (Fig. 5A) and to a slow triangular ramp of current (Fig. 5C). In response to a rectangular subthreshold current (1.3 nA), the motoneuron displayed the slow depolarization (arrowhead) due to the inactivation of the slow potassium current (Fig. 5A1). Just above rheobase (1.4 nA, Fig. 5A2), the discharge started after a delay and the firing frequency increased throughout the rest of the pulse. The firing frequency acceleration is accounted for by the slow depolarization of the membrane that results from the slow inactivation of the potassium current. At higher intensity (2.0 nA, Fig. 5A3) the firing started at pulse onset and the discharge frequency progressively decreased (frequency adaptation) indicating that the slow inactivation of the potassium current does not control the firing pattern anymore.

Fig. 5.

Most typical pattern of discharge of delayed-firing motoneurons. A1–A3: responses of a delayed-firing motoneuron to long lasting pulses (5 s) of increasing amplitude. Bottom: injected current (square pulses); middle: voltage response; top (A2, A3 only): instantaneous firing frequency. In A1, the horizontal dashed line and the arrowhead point out to the progressive depolarization of the membrane potential. B: evolution of action potential properties in function of the spike number in the train shown in A2. B1: action potential (AP) width at half-height. B2: action potential maximum speed of repolarization. B3: membrane potential at the peak of afterhyperpolarization (AHP). B4: adaptation indexes for each property: average of the values for the 3 last action potentials divided by the values for the first 3 ones. The adaptation indexes were computed for 14 (AP width), 15 (AP repolarization), and 11 (voltage at the peak of the AHP) motoneurons. C1 and C2: response to a slow triangular ramp of current in the same motoneuron (slope ± 0.1 nA/s). C1: bottom, injected current; middle, voltage response; top, instantaneous firing frequency (Hz). The dashed vertical lines indicate the recruitment (I_on) and the derecruitment currents (I_off). C2: instantaneous firing frequency plotted in function of the injected current during the ascending (empty squares) and descending (filled circles) ramps. Open arrow, inflexion point. The stationary F-I curve calculated from the square pulses is also plotted (triangles). For each pulse, the frequency is averaged over the last second.

The slow depolarization of the membrane also induced a progressive broadening of the action potentials (Fig. 5, B1 and B4) that is caused by a slowing of the spike repolarization (Figs. 5, B2 and B4). However, the voltage at the peak of the AHP was progressively depolarizing (Fig. 5, B3 and B4). This suggests that the spike enlargement did not result in significant larger calcium currents and in larger SK-calcium activated currents. In these conditions, the SK current cannot counteract the increase of firing frequency caused by the membrane depolarization (Manuel et al. 2006).

The stationary F-I relationship of delayed-firing motoneurons can be fairly well approximated on ramps (Fig. 5C2) provided that the ramp velocity was as low as 0.1 nA/s (Fig. 5C1). At this very slow velocity, the recruitment current during the ramp (1.0 ± 0.6 nA, n = 50) was close to the rheobase measured on long rectangular pulses (1.2 ± 0.6 nA, n = 56, P = 0.2). Moreover, the ascending branch of the F-I function was superimposed to the stationary F-I curve obtained with long rectangular current pulses (triangles on Fig. 5C2). Prior to the inflexion point (open arrow on Fig. 5C2), the pulse response (Fig. 5A2) displayed an acceleration of the discharge throughout the pulse. After the inflexion point, the pulse responses no longer displayed this feature (see Fig. 5A3). It is worth noting that the response to the slow triangular ramp was hysteretic. In this example, the recruitment current (I_on = 1.25 nA) was larger than the derecruitment current (I_off = 0.89 nA, Fig. 5B2).

Thirty-seven of 50 delayed-firing motoneurons (76%) had a recruitment current higher than the derecruitment current as in Fig. 5 (I_on = 1.1 ± 0.6 nA vs. I_off = 0.9 ± 0.6 nA, n = 37, P < 0.0001). This type of hysteresis, which was observed only on delayed-firing motoneurons, was most likely due to the slow inactivating potassium currents that delayed the recruitment of the motoneuron (see below). In the remaining delayed-firing motoneurons, the F-I function in response to a slow triangular current ramp looked more like the responses observed in immediate-firing motoneurons. This is likely because slow potassium currents are weaker in these delayed-firing motoneurons. In seven delayed-firing motoneurons (12%), the recruitment current was lower than the derecruitment current (I_on = 0.6 ± 0.3 nA vs. I_off = 0.7 ± 0.3 nA, n = 7, P = 0.03). In the remaining six motoneurons (12%), the response was not hysteretic (recruitment and derecruitment currents were the same, ascending and descending branches superimposed) (I_on = 1.4 ± 0.8 nA vs. I_off = 1.4 ± 0.8 nA, n = 6, P = 1). It is noteworthy that delayed-firing motoneurons with I_on > I_off have a rheobase (Fig. 6A), an input conductance (Fig. 6B), an AHP amplitude (Fig. 6C), and an AHP relaxation time (Fig. 6D) that were distributed over the full ranges of these parameters. There is no relationship between the type of hysteresis and the rheobase, the input conductance, the AHP amplitude, and the AHP relaxation time (Fig. 6).

Fig. 6.

Delayed-firing motoneurons with recruitment current higher than derecruitment current do not display specific rheobase, input conductance, AHP amplitude, and AHP relaxation time. Recruitment current of delayed-firing motoneurons during the slow ramp against the rheobase measured on long rectangular pulses (A), the input conductance (B), the AHP amplitude (C), and the AHP relaxation time constant (D). AHP amplitude and AHP relaxation time constant were measured on single action potentials elicited by 1-ms square pulses (1–10 nA) (see materials and methods). I_on, recruitment current; I_off, derecruitment current. Symbols differ depending whether I_on is higher than I_off (red triangles), equal to I_off (green circles), or smaller than I_off (blue squares).

Dynamic effects observed on the F-I function further emphasize the role of the slow-inactivating potassium currents in the F-type motoneurons. Figure 7A shows another delayed-firing motoneuron for which the recruitment current was again substantially higher (0.56 nA) than the derecruitment current (0.15 nA). However, in response to a second triangular ramp that started 2 s after the end of the first one (Fig. 7B), the recruitment current was much lower (0.10 nA) and was close to the derecruitment current. As a consequence, the motoneuron is more excitable at the onset of the second ramp than at the onset of the first one. It is likely that at the end of the first triangular ramp the slow potassium current had fully inactivated and that the 2-s delay before the next ramp was not long enough to allow it to deinactivate, thus accounting for the excitability increase. This memory effect depends on the time between two successive ramps and the number of ramps as shown in Fig. 8 (only delayed-firing motoneurons with I_on > I_off were used for quantification). The recruitment current was lower (89 ± 9%, n = 20) when the second ramp occurred only 10 s after the completion of the first one. It was even lower after a third ramp (81 ± 13%, n = 20, one-way ANOVA P < 0.0001) that occurred another 10 s after the second one (Fig. 8, red curve). With 40-s intervals between two successive ramps the dynamic effect was no longer visible (Fig. 8, black curve, 100 ± 26% on second ramp, 99 ± 23% on third ramp, n = 63, one-way ANOVA P = 0.2). In sharp contrast, the recruitment current of immediate-firing motoneurons showed no adaptation even with 10-s intervals (Fig. 8, blue curve 101 ± 14% on second ramp, 99 ± 11% on third ramp, n = 44, one-way ANOVA P = 0.7). All together, our results show that the slow potassium current plays a critical role in dynamically setting the recruitment current of F-type motoneurons.

Fig. 7.

Dynamic of the recruitment current in delayed-firing motoneurons. A1 and B1: two consecutive current injections (triangular ramp, slope ± 0.1 nA/s) in the same motoneuron. The ramp in B started 2 s after the completion of the ramp in A. Bottom: injected current; top: voltage response. A2 and B2: instantaneous firing frequency during the ascending (gray squares) and descending (black circles) ramps. Arrow points to the decrease in recruitment current between A and B.

Fig. 8.

Quantification of the dynamic of the recruitment current. We injected 3 consecutive ramps at different intervals in delayed-firing motoneurons with recruitment currents superior to derecruitment currents. Recruitment currents for each successive ramps were normalized to the value for the first ramp. Red curve: 10 s between two successive ramps; black curve: 40 s between two successive ramps; blue curve: immediate-firing motoneurons (hysteretic and nonhysteretic motoneurons pooled), 10 s between two successive ramps. Values are the average ± SE.

DISCUSSION

We showed that the delayed discharge of F-type motoneurons in neonates is caused by the combined action of a fast A-like current and a slow-inactivating potassium current. These potassium currents dynamically control the excitability of F-type motoneurons by setting their recruitment threshold and shaping their F-I function.

Expression of A-like and slow-inactivating potassium currents during postnatal development.

During liminal pulses, we observed delays before firing that could last several seconds after the pulse onset (Leroy et al. 2014). Here we showed that the firing delay is due to two potassium currents that act at two different time scales. The stronger transient A-like current acts by blocking the initial discharge over a relatively short time (about 30 ms in our experimental conditions). A similar current has been observed in spinal motoneurons of neonatal mice (Gao and Ziskind-Conhaim 1998; Perrier and Hounsgaard 2000) as well as in brain stem motoneurons (Lape and Nistri 1999; Russier et al. 2003). The slow-inactivating potassium current is responsible for the long delay of the discharge (several seconds). Indeed, the slow potassium inactivation induces a slow membrane depolarization that reaches the voltage threshold long after the pulse onset. Such slow currents have also been observed in other types of neurons such as the adult rat hippocampus neurons (Luthi et al. 1996; Storm 1988) and the thalamic relay neurons (Huguenard and Prince 1991) where they also coexist with the A current. However, the delayed-firing pattern was transiently observed between P4 and P9 in rat abducens motoneurons (Russier et al. 2003). It has not been reported in adult spinal motoneurons so far (Kernell 2006; Manuel et al. 2009; Meehan et al. 2010). A-current and slow-inactivating potassium current have not been observed in lumbar motoneurons of adult mammals (Barrett et al. 1980; Kernell 2006).

Potassium currents determine the firing properties of F-type motoneurons.

The potassium currents not only set the delay before firing in response to liminal pulses (injected current just above rheobase), but they also shape the firing properties of F-type motoneurons. First, the inactivation of the slow potassium current causes an acceleration of the discharge during rectangular pulses. Second, the slow potassium current shapes the F-I function during triangular ramp of currents. It is likely however that the slow inactivation of the sodium current also contributes to shaping the F-I function (Iglesias et al. 2011). The slow inactivating component of the sodium current counteracts the slow-inactivating potassium current. This might give rise to a symmetrical F-I function or to a F-I function that displays a clockwise hysteresis. The F-I functions of the immediate-firing motoneurons are more stereotyped since the slow potassium current is lacking in these cells. In adult animals, spinal motoneurons were also found to display a variety of F-I functions (Bennett et al. 2001; Button et al. 2006; Hounsgaard et al. 1988; Lee and Heckman 1998). A classification in four types was proposed for adult spinal motoneurons by Bennett et al. (2001). However, this classification may not readily apply to motoneurons of neonatal mice since it is not clear whether adult motoneurons are endowed with the same panoply of currents as the delayed-firing motoneurons. Moreover, adult motoneurons display calcium persistent inward currents (Perrier and Hounsgaard 2000) which are weak in motoneurons of neonatal mice (Quinlan et al. 2011). The present work suggests that, in neonates, higher recruitment current compared with derecruitment current is likely caused by potassium currents, which increase the recruitment threshold, rather than by calcium persistent inward currents, which decrease the derecruitment current.

Impact of the potassium currents on the recruitment of delayed-firing motoneurons.

On average, F-type motoneurons have larger input conductances than the S-type ones, but there is a large overlap between the distribution of input conductances in the two motoneuron populations (Leroy et al. 2014). As a consequence the orderly recruitment of motoneurons (S-type before F-type) cannot solely rely on their input conductance in neonates. However, the potassium currents are responsible for the more depolarized voltage thresholds for spiking of F-type motoneurons compared with S-type ones (Leroy et al. 2014). These currents decrease the excitability of F-type motoneurons and increase the separation of recruitment thresholds. The A-like current will increase the recruitment threshold for brief synaptic inputs (several tens of milliseconds) whereas the slow potassium current will increase the threshold during steady synaptic inputs. Potassium currents might therefore help ensure an orderly recruitment of neonatal motoneurons for a wide range of synaptic inputs.

However, fast and slow potassium currents are dynamically changing. For example, the activation level of the A-like current by excitatory synaptic inputs depends on the membrane voltage just before the inputs. Membrane hyperpolarization deinactivates the A-like current reducing the probability of brief excitatory inputs to reach the threshold for firing. Conversely, depolarization inactivates the A-current and increases the probability that the same excitatory inputs reach the firing threshold. Moreover, the recruitment threshold of the delayed-firing motoneurons may be lowered when the intervals between the slow current ramps are short. This is likely due to the fact that the slow potassium current remains inactivated, thus enhancing motoneuron excitability. Slow synaptic activation of the delayed-firing motoneurons repeated at short intervals is then likely to decrease their recruitment threshold and to increase their firing probability. This might be a mechanism to recruit F-type units, which develop more force than S-type units, during postural activities in neonates. In contrast, when the F-type motoneurons are excited at long intervals, the slow potassium current has time to deinactivate. Their recruitment threshold is then higher. In these conditions F-type units remain available for recruitment during movements that require an important amount of force output.

Potassium currents might contribute to the maturation of motor units in neonates.

This decrease in the recruitment probability of F-type motoneurons by the potassium currents might also contribute to the maturation of their motor units. Indeed, the motoneurons activity has an impact on the maturation of the myosin heavy chain isoforms (Goldspink et al. 1992; Lowrie et al. 1989; Picquet et al. 1998). Since the slow muscle fibers already express their mature myosin heavy chain isoform at the end of the first week whereas the fast fibers express theirs later (Agbulut et al. 2003; Schiaffino and Reggiani 2011), one may speculate that this differential rate of maturation originates from the differential activity between S- and F-type motoneurons. Moreover, the motoneurons activity has also an impact on the rate of elimination of supernumerary neuromuscular junctions (Thompson 1985). Indeed, increasing the neuromuscular activity by chronically applying electrical stimulation dramatically increases the rate of loss of synaptic contacts (O'Brien et al. 1978). Conversely, reducing proprioceptive inputs to motoneurons (Benoit and Changeux 1975) or reducing inputs to muscle fibers by partial axotomy (Brown et al. 1976) delays the elimination of poly-innervation. As a result, motor units whose activity was reduced by TTX had larger territories than those whose axons remained active (Callaway et al. 1989). This was further supported by experimental and theoretical works suggesting that motoneurons with less active neuromuscular junctions have larger motor units than motoneurons with more active ones (Barber and Lichtman 1999; Callaway et al. 1987; Nowik et al. 2012). One may wonder to what extent the larger F-type motor unit territories might result from a lower activity of their motoneurons induced by the potassium currents. A lower activity of F-type motoneurons compared with S-type motoneurons is expected because of their larger average input conductance and their higher rheobase (Leroy et al. 2014). However, the present findings suggest that the activity of F-type motoneurons might be further reduced, compared with S-type motoneurons, if they are recruited after a period of inactivity (inactivation of the slow potassium current) or after an hyperpolarization of the membrane (inactivation of the A-current). This would not be the case if F-type motoneurons were frequently recruited or if their membrane potential was depolarized. Depending on their degree of activation, the potassium currents will reduce more or less the activity of F-type motoneurons with respect to S-type motoneurons and will have or not an impact in the maturation of the motor units.

GRANTS

Financial support was provided by the Agence Nationale pour la Recherche (HYPER-MND, ANR-2010-BLAN-1429-01), the National Institute of Neurological Disorders and Stroke (R01-NS-077863), the Thierry Latran Fundation (OHEX Project), and Target ALS are gratefully acknowledged. F. Leroy was the recipient of a “Contrat Doctoral” from the Ecole Normale Supérieure, Cachan.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

Author contributions: F.L., B.L.d., and D.Z. conception and design of research; F.L. performed experiments; F.L. analyzed data; F.L., B.L.d., and D.Z. interpreted results of experiments; F.L. prepared figures; F.L. and D.Z. drafted manuscript; F.L., B.L.d., and D.Z. edited and revised manuscript; F.L., B.L.d., and D.Z. approved final version of manuscript.