Journal of Neurophysiology

Essential Role of a Fast Persistent Inward Current in Action Potential Initiation and Control of Rhythmic Firing

R. H. Lee, C. J. Heckman


In spinal motoneurons in an in vivo preparation, we investigated the relationship between a fast persistent inward current located in or near the soma and the capacity of these cells to fire rhythmically. The fast persistent current could be markedly reduced by prolonged depolarization. Modest reductions resulted in profound changes in the slope of the frequency-current relationship. At greater reduction levels, rhythmic firing failed and could not be restored by increasing injected current. However, fully formed spikes still occurred in a slow, uncoordinated fashion, suggesting that the fast inactivating Na+ currents that generate the spike itself remained unchanged. Consequently, the fast persistent inward current, which may be primarily generated by persistent Na+ channels, appears to be essential for initiation of spikes during rhythmic firing. Additionally, it appears that the fast persistent current plays a major role in setting the frequency-current gain.


A fundamental behavior of neurons is the generation of trains of action potentials in response to slow excitatory inputs. Action potentials are generated primarily by rapidly opening sodium channels that subsequently rapidly close due to inactivation, resulting in a sharp but very brief depolarization (Hodgkin and Huxley 1952). However, the rapidity of the inactivation of these channels poses a problem. Each spike in a train is preceded by an afterhyperpolarization (AHP) from the previous spike. Consequently, the voltage trajectory prior to spike initiation has a slow rate of rise due to the slow decay of the AHP. At these slow rates, fast Na+ channel inactivation should readily keep pace with its own activation and therefore should prevent spike initiation from occurring. However, a small proportion of the Na+ channels fails to rapidly inactivate, providing a fast but persistent inward current (Crill 1996). Although previous work has emphasized the role of this persistent Na+ current in amplifying synaptic input (Schwindt and Crill 1995; Stuart and Sakmann 1995), we propose that persistent Na+ current serves a more basic role as the spike initiator during rhythmic firing. Our recent computer simulations showed that a realistic motoneuron model that lacked persistent Na+ current still generated normal spikes in response to rapidly rising inputs but could not sustain rhythmic firing regardless of the magnitude of the applied steady current (Lee and Heckman 1998b). In this paper, experimental evidence in support of this hypothesis is presented.


Experiments were performed in spinal motoneurons in four adult cats deeply anesthetized with pentobarbital sodium, using standard procedures in our lab (Lee and Heckman 1998a). Intracellular recordings of triceps surae motoneurons in the lumbar spinal cord were obtained with sharp microelectrodes. Voltage clamp was applied by discontinuous, single-electrode methods (Finkel and Redman 1983; Lee and Heckman 1998a). Clamp control of the underlying currents was assessed by comparing ascending versus descending ramp commands, with a lack of hysteresis during these slow ramps indicating good clamp control (see Lee and Heckman 1998a). All procedures were fully approved by the animal care committee at Northwestern University.

The time constants of the channels active in the threshold region of motoneurons exhibit a wide degree of separation (Binder et al. 1996). Consequently, fast and slowly activating currents were separated based on their responses during voltage clamp to a voltage command in which a fast, small amplitude sine wave was superimposed on a slow (8 mV/s), large amplitude (40 mV) triangular waveform. The frequency (125–200 Hz) and amplitude (0.1–0.25 mV) of the sine wave were chosen to provide rates of rise of voltage that were too rapid to activate slow currents like the Ca2+-mediated K+ current but still slow enough to allow full inactivation of the inactivating Na+ current. Preliminary computer simulations indicated that this approach did provide good separation of fast and slow currents.

Separation of the fast and slow currents in the response to this dual input required several steps and is illustrated in Fig.1. Low-pass filtering (typically at 25 Hz) the total voltage input and current output yielded the ramp input and response [i.e., “Total” current-voltage (I-V) function of Fig. 1, B and C]. The residual traces (i.e., raw minus smoothed) yielded the sine wave input and response (Fig. 1 A). Examination of this residual response revealed that the apparent decrease and subsequent increase in magnitude of the sine wave response at higher voltage was due to the net conductance becoming negative, resulting in a current response that was out of phase with the voltage command (see Fig. 1 A, insets).

Fig. 1.

Measurement of fast persistent currents. A: sinusoidal voltage command and current response. Bottom: sine wave command over the course of the ramp. Top: current response to the sine wave command. Insets: brief samples of the sinusoidal modulation of voltage (thin trace) and the resulting current (bold trace). B: components of current-voltage (I-V) function. Total I-V function and data fromA shown after processing (see methods), resulting in I Leak,I Fast, andI Slow. C: comparison of Fast vs. Total I-V functions. Fast curve constructed from the I Leak andI Fast traces of B. Onset denotes maximum in Fast curve.

The net sinusoidal response obtained by the above procedure included not only the conductances from fast activating ion channels but also the leak conductance and cell capacitance. The active component of this response can be separated by a procedure analogous to leak compensation in steady inputs. The effective conductance generated by the sinusoid (G sin) at each membrane potential was calculated with the formula: G sin =I sin *V sin/(V sin* V sin), whereV sin andI sin were the sinusoidal input and current response, respectively. Systematic variations in the frequency and amplitude of the sine wave showed that the leak and capacitance effects simply generated a net bias inG sin (5 cells). Their contribution was thus independent of voltage and could be removed by subtracting the amplitude of G sin at the most hyperpolarized level from the values of the overallG sin response. The remaining component of G sin, containing only the sinusoidal responses of fast voltage-sensitive currents (I Fast), was smoothed (low-pass filter at 5 Hz) and then converted back into a current trace by integration with respect to voltage (“I Fast” trace in Fig. 1 B). Since the goal of this work was to assess the functional impact of I Fast on rhythmic firing, a total “Fast” I-V function (i.e., the current response to fast moving voltage inputs) was reconstructed by combining I Fast andI Leak.

The Fast I-V function usually had a negative slope region due to the negative effective conductance ofI Fast. The onset ofI Fast was defined as the point of zero slope of the Fast I-V function (see Fig. 1 C). From a theoretical perspective, reaching the onset point should induce spike initiation. In cases where there was no negative slope region in the Fast I-V function, onset was defined as the point of minimal slope. The amplitude of I Fastwas measured as the difference in current between onset and 4 mV above onset.

Spike and rhythmic firing properties were assessed by application of slow (10 s) triangular injected currents, applied using discontinuous current clamp. Spikes properties were characterized based on the first spike in each train. Spike overshoot was defined as the voltage at the peak of the spike, while spike threshold was measured as the voltage where the first peak of the second derivative of the voltage trace occurred. The quality of rhythmic firing was assessed by the slope (i.e., gain) of the frequency-current (F-I) relationship slope, obtained by triangular injected currents in discontinuous current clamp.


The results shown in Fig. 1 C were typical for the 25 cells studied. The Fast I-V function evoked by the sinusoid contained a negative slope region due to activation of a fast persistent inward current, I Fast. The impact of I Fast on the TotalI-V can be seen as a reduced slope region (Fig.1 C). Note that the effects were not due to poor clamp control as ascending and descending ramps produced the same result (seemethods).

Long duration suprathreshold depolarization leads to a temporary loss of the capacity for rhythmic firing in motoneurons (Coombs et al. 1955). If I Fast plays an essential role in spike initiation, then it should fade as rhythmic firing declines. This idea was tested by alternating the I-Vprotocols illustrated in Fig. 1, which provided sustained periods of suprathreshold depolarization, with the F-I protocol, to evaluate the quality of rhythmic firing. Figure2 illustrates this test on a cell initially exhibiting good rhythmic firing (early trace) that subsequently deteriorated (middle trace) and then failed completely (late trace). Deterioration of rhythmic firing tended to be marked by two characteristics, “missing spikes” and a very asymmetrical response to the ascending versus descending ramps (middle trace, Fig. 2 A). When firing deteriorated, there were no changes evident in the voltage trajectory following each spike. Instead, subsequent spikes simply failed to initiate (Fig. 2 B). Individual spikes for both good and poor firing were similar in shape (Fig. 2 C), indicating that the fast inactivating Na+ channels generating the primary action potential current did not significantly change. What did change was I Fast, which markedly decayed in concert with the deterioration in firing. The decay inI Fast is evident in the fastI-V function from a depolarizing shift inI Fast onset and the disappearance of the negative slope region (Fig. 2 D).

Fig. 2.

Relationship between fast persistent currents and rhythmic firing in a single cell. Three runs of I-V and frequency-current (F-I) protocols illustrate the relationship between rhythmic firing and I Fast Onset.A: decay of rhythmic firing due to prolonged depolarization. B: afterhyperpolarizations (AHPs) of 1st spikes of the early (thin trace) andmiddle (thick trace) traces inA. C: 1st action potentials of theearly and middle traces inA. D: Fast I-V functions corresponding to the traces in A. Dots represent measured onset point for each trace.

The close relationship between I Fastand rhythmic firing shown in Fig. 2 was also apparent across the full sample of 68 F-I protocols made in 25 cells. Despite the scatter from pooling data across cells and experiments, there was a strong correlation (r = 0.74, P < 1E-11) between the onset voltage ofI Fast and spike threshold voltage (Fig. 3 A), with the relationship being near unity (dashed line). This result supports a fundamental role for I Fast in spike initiation. Equally important, the F-I gains were strongly correlated with I Fast voltage onset (Fig. 3 B; r = −0.64, P < 1E-7) and with I Fast amplitude (not shown, r = 0.6, P < 1E-6). Thus asI Fast declined so too did rhythmic firing gain. In contrast, spike overshoot, a measure of the fast inactivating Na+ current, did not significantly correlate with I Fast onset (Fig.3 C; r = −0.05, P = 0.68) orF-I gain (r = 0.16, P = 0.24; not shown).

Fig. 3.

Overall relationship between fast persistent current and rhythmic firing. Properties for 68 runs from 25 cells. A: relationship between Spike Threshold andI Fast Onset. Solid line, linear regression (r = 0.74, P < 1E-11); dashed line, Unity relationship. B: relationship betweenF-I gain and I Fast Onset. Solid line, linear regression (r = −0.64,P < 1E-7). C: relationship between spike overshoot and I Fast onset. Solid line, linear regression (r = −0.05,P = 0.7)


The fact that I Fast was under good clamp control is a strong indication that it represents a current in or near the soma, with the most likely location being the initial segment (Safronov et al. 1997). Additionally,I Fast had characteristics that are consistent with those of persistent Na+ currents observed in other preparations (Fleidervish and Gutnick 1996; Pan and Beam 1999). Previous studies in spinal motoneurons suggest that most of their persistent inward currents are carried by calcium (Hounsgaard and Kiehn 1985; Schwindt and Crill 1980). However, these studies did not eliminate the possibility of a contribution from persistent Na+ currents, which have been clearly demonstrated in cranial motoneurons (Chandler et al. 1994).

Another type of Na+ current, resurgent Na+ current (Raman and Bean 1997) may also play a role for spike initiation some cells but appears not to exist in motoneurons (Fleidervish and Gutnick 1996;Pan and Beam 1999). Reasonably fast, low-threshold calcium currents may also contribute toI Fast, such as the T-type Ca2+ current (Barish 1991) or the L-type Ca2+ current (Magee et al. 1996). It is unlikely that reductions in potassium currents could be responsible for the observed inward current as required conductance of such a channel would be larger than the measured leak in some cells. However, potassium channels could attenuate the measured fast net inward current.

These results support our hypothesis that a fast persistent inward current is necessary for spike initiation during rhythmic firing. If this current only served to amplify synaptic and injected currents, normal rhythmic firing could have been attained simply by injecting more current to make up for the reduction inI Fast. Instead, rhythmic firing failed and could not be restored by increased injected current whenI Fast became small enough to eliminate the negative slope region in the fast I-V function. Furthermore, changes in I Fast closely correlated with changes in the gain of the F-I function. Although a reduction in F-I gain as rhythmic firing failed would appear to be an almost foregone conclusion, this relationship also held at F-I values well within the range of values typically associated with normal neuron function (Kernell 1979), suggesting that I Fastis the dominant determinant of this fundamental cell behavior.


This work was supported by National Institute of Neurological Disorders and Stroke Grant NS-34382.


  • Address for reprint requests: R. H. Lee, Physiology, M211, Northwestern University Medical School, 303 E. Chicago Ave., Chicago, IL 60611.


View Abstract