3.1 Introduction

During the past 25 or so years, our studies, as well as those of many colleagues, have revealed ever-greater complexity in the input–output properties of motoneurons. It is thus no longer tenable to consider motoneurons to be simple “threshold and fire” followers of motor commands. This complexity is puzzling though—why is such a wide range of motoneuron input–output states needed to implement motor behaviors? The guiding concept for our ongoing work is that this multiplicity of motoneuron states is in fact an adaptation to match the great diversity of the normal movement repertoire.

We propose the following correspondences between motoneuron states and motor behaviors: For posture, motoneurons may act as integrators to provide memory of brief inputs; for a wide range of voluntary movements, motoneurons may act as variable gain amplifiers; and for repetitive movements driven by central pattern generators, motoneurons may act as oscillators. Admittedly, these correspondences are overly simplified, but they potentially provide clear guides for further studies of this fundamental issue. The first two states are highly dependent on neuromodulatory input from the brainstem, mediated via actions of serotonin (5HT) and norepinephrine (NE) while the oscillatory state appears to depend primarily on glutamatergic N-methyl-d-aspartate (NMDA) inputs, adding a whole new dimension to motoneuronal behavior. The data supporting these correspondences are presented in the following sections, after a brief review of the fundamental role of neuromodulation.

3.2 Motoneuron Input–Output Processing Is Profoundly Influenced by Brainstem Neuromodulatory Systems

Neuromodulation is the foundation of the diversity of motoneuronal states. The two most powerful neuromodulatory inputs to motoneurons are mediated by axons descending from the brainstem and releasing either 5HT or NE. Unlike ionotropic inputs, which open ion channels to generate excitatory postsynaptic potentials (EPSPs) or inhibitory postsynaptic potentials (IPSPs), neuromodulatory inputs use G protein-coupled receptors to activate signal transduction pathways (Hille 2001). The effects of 5HT and NE on motoneuron excitability are extraordinary (Hultborn 2002; Heckman and Enoka 2012). Recruitment thresholds are greatly reduced by depolarization of the rest potential and hyperpolarization of the spike voltage threshold. Rate modulation is transformed due to facilitation of persistent inward currents (PICs) mediated by voltage-sensitive Na and Ca channels. PICs are a primary focus of this proposal. Undoubtedly, motoneurons are subject to neuromodulation by other neurotransmitters in addition to 5HT and NE (e.g., Svirskis and Hounsgaard 1998). The dopaminergic input from the brainstem to the cord increases motoneuronal excitability, but has yet only been studied in neonatal preparations (Garraway and Hochman 2001; Zhu et al. 2007). It is likely weaker than the 5HT and NE systems. Cervical motoneurons in the cat have over 1500 synapses containing 5HT and over 1200 synapses containing NE and these synapses cover their extensive dendritic trees (Montague et al. 2013). Local spinal circuits are known to induce neuromodulatory effects on motoneurons as well. Cholinergic interneurons projecting to C-terminals on motoneurons also reduce the afterhyperpolarization (AHP) (Zagoraiou et al. 2009) and possibly this is the source of the reduced AHP that occurs during the scratch reflex in spinalized animals (Power et al. 2010). In contrast, 5HT and NE have very little effect on the AHP in the adult (Lee and Heckman 1999; Li et al. 2007). Spinal circuits also lower spike voltage threshold (Power et al. 2010). In this review, our focus is on neuromodulation by 5HT and NE for two reasons: their effects on motoneurons are remarkably powerful and reasonably well understood. Further states of motoneurons will likely become apparent as our understanding of the full diversity of neuromodulatory inputs increases.

3.2.1 Neuromodulatory Inputs are an Essential Component of Normal Motor Commands

Numerous studies indicate that the brainstem neuromodulatory input is essential for a wide range of motor behaviors (Heckman and Enoka 2012). To summarize: (1) There exists steady neuromodulatory drive from the brainstem to the cord during the waking state, which is increased during sustained motor output (Aston-Jones et al. 2001; Jacobs et al. 2002). (2) Motoneurons lacking this neuromodulatory drive are severely hypoexcitable (Hounsgaard et al. 1988; Miller et al. 1996). Excitability is so low that even simultaneous, intense activation of all known inputs is far from sufficient to produce the maximal motoneuronal output required for high forces (Powers and Binder 2001; Cushing et al. 2005). (3) Motor unit firing patterns in humans during slow increases in force closely resemble the firing patterns generated by equally slow activation of cat motoneurons with moderate to large amplitude PICs (Hornby et al. 2002; Heckman and Enoka 2012). In summary, normal motor behavior cannot be achieved without a strong neuromodulatory input to motoneurons. This input has a profound impact on motoneuron input–output processing, i.e., on its “state.”

3.2.2 The PIC and Bistable Behavior as a Pattern Generator for Postural Control

Bistable behavior reflects the inherent tendency of PICs to prolong inputs. Figure 3.1 shows an example of how the PIC transforms a synaptic current. The PIC converts a brief excitatory current (green trace) into a prolonged “tail” current that persists for many seconds (red trace; inward, depolarizing current is upward). The sustained PIC tail current is the basis of self-sustained firing, which requires little or no tonic input from descending or sensory systems. Recordings of electromyogram (EMG) and force in response to brief inputs have documented prolonged motoneuronal output lasting minutes (Hounsgaard et al. 1988; unpublished data, Heckman lab). Bistable behavior has long been thought to be important for posture (Hounsgaard et al. 1988). Consistent with this role, PICs with low-voltage thresholds and long-lasting tail currents only exist in type S motoneurons that innervate fatigue resistant muscle fibers, while PICs in type F motoneurons have much higher thresholds and faster decays (Lee and Heckman 1998a, 1998b). This PIC “tuning” makes good functional sense: Type F motoneurons do not participate in the low forces needed for steady posture and their extremely low fatigue resistance seems well matched to a more rapidly decaying PIC. The match between PIC properties and the fatigue resistance of motor units was a key insight leading to our concept that different motoneuronal states match different motor behaviors. In a sense, bistable behavior can be considered a pattern generator for the steady forces required for posture. Remarkably, this pattern generator is intrinsic to motoneurons themselves.

Fig. 3.1
figure 1

Amplification and prolongation of synaptic input by persistent inward current. Intracellular recording of a medial gastrocnemius motoneuron in the decerebrate cat preparation. The synaptic current was generated by vibration of the Achilles tendon to steadily activate muscle spindle Ia afferents for ~ 0.75 s. The cell was voltage clamped using the switching single electrode technique. The green trace shows the current generated by the steady Ia input when the cell was voltage clamped to a very hyperpolarized level, preventing this input from activating the motoneuron’s persistent inward current. The red trace shows the effect of the very same input when the cell was clamped near its threshold for firing. Because the clamp is only effective at the soma, the Ia input activated the PIC in dendritic regions. As a result, the input was both strongly amplified and prolonged—the tail current persisted for many seconds (not shown). (Data from Lee and Heckman 1996)

Full size image

3.2.3 Does the Bistable State Provide Integration or Memory of Transient Inputs?

Figure 3.1 suggests that a bistable motoneuron is not an integrator in the strict mathematical sense—the prolonged current is less than the initial current during the pulse. This history-dependent behavior, however, does have a strong memory-like aspect to it. For posture, the key functional issue is how a motoneuron in the self-sustained firing state responds to further transients for postural corrections. Previous studies of the effects of PICs on synaptic inputs have been done using a combination of synaptic and injected currents (e.g., Bennett et al. 1998; Lee and Heckman 2000) and have ignored how the cell behaves when primarily driven by the PIC to produce self-sustained firing. Further studies are needed to define the behavior of the motoneuron in the postural state. If the response during self-sustained firing to further transients is prolonged, then the motoneuron is indeed acting much like an integrator or memory device. If not, then bistable behavior may provide a steady background that nonetheless allows for transient corrections.

3.2.4 PIC Amplification as the Basis for Gain Control of Motor Output via the Brainstem

Figure 3.1 shows that the PIC not only prolongs input but also greatly enhances it (note the large increase in current while the input is active). This enhancement is generally thought to provide input amplification (Lee and Heckman 2000; Prather et al. 2001; Hultborn et al. 2003) and this amplification is essential in increasing the gain of motoneurons sufficiently so that they can be activated effectively by descending and sensory inputs (see above). PICs are highly sensitive to inhibitory inputs (Kuo et al. 2003; Bui et al. 2008b) and so motor commands can easily decouple PIC amplification from PIC prolongation by using inhibition to turn off the PIC when it is not needed. The amplitude of the PIC is proportional to the level of neuromodulatory input (Lee and Heckman 2000). Thus, brainstem neuromodulatory control of PICs potentially provides motor commands with a mechanism to control the input–output gain of motoneurons. We have speculated that a low motoneuronal gain is advantageous for low-force, high-precision tasks but that gain should be progressively increased as the force requirements of the task increase (Johnson and Heckman 2010). Matching of motoneuronal gain to the widely varying forces of voluntary motor tasks would constitute a clear example of the diversity of motoneuronal behavior matching the diversity of motor behaviors, but this match has yet to be clearly demonstrated.

3.2.5 Does Synaptic Inhibition of PICs Afford a Local, Highly Specific Gain Control of Motoneurons?

Inhibitory synaptic input hyperpolarizes motoneurons, thereby increasing their threshold for activation and opposing the depolarization from excitatory inputs. Probably, most considerations of the role of inhibition in motor control assume this fundamental model. Yet the effect of inhibition on the PIC suggests that it may also alter motoneuron gain, because we have shown that inhibition can reduce the amplitude of the PIC in a proportional manner (Kuo et al. 2003). Thus, inhibitory input may provide a gain control mechanism that opposes the gain control from the brainstem. The brainstem system is remarkably diffuse—in fact some of these axons give off branches along the entire length of the cord (Holstege and Kuypers 1987). In contrast, there are several highly focused sources of spinal inhibition—e.g., reciprocal inhibition, recurrent inhibition, and low threshold cutaneous inhibition (Baldissera et al. 1981; Jankowska 2001). A previous suggestion that recurrent inhibition may provide gain control (Hultborn et al. 1979) faltered because of its relatively small ionotropic currents (Lindsay and Binder 1991), but the effect of inhibition on the PIC potentially provides a much larger impact on gain (Bui et al. 2008a). If inhibitory inputs can reduce gain by reducing PIC amplitude, then local spinal circuits would have the capacity for a focused, specific control of motoneuron gain (Heckman et al. 2008; Johnson and Heckman 2010).

3.2.6 Is the Oscillatory Motoneuron State Linked to CPGs for Repetitive Behaviors?

NMDA receptors are ionotropic but their behavior is unique in that they are voltage-sensitive because depolarization relieves the Mg2+ block of the channel (Hille 2001). The resulting synaptic currents behave much like a PIC and can generate large plateau potentials. Unlike PICs, however, the NMDA plateau potential is self-terminating, probably because of activation of an outward current (Manuel et al. 2012; Wang et al. 2013). The resulting NMDA-mediated oscillations have been consistently demonstrated in neonatal interneurons and motoneurons (Harris-Warrick 2011). These oscillations are intrinsic to the cell, in that they persist in the presence of tetrodotoxin (TTX) to block all inputs (Hochman et al. 1994; MacLean et al. 1997). We have recently demonstrated that the response of motoneurons to NMDA does not fade with maturation. Instead, adult rodent motoneurons exhibit very strong intrinsic oscillations in response to bath administration of NMDA—exactly like the neonatal results (Manuel et al. 2012). Thus, NMDA-mediated input may place the adult motoneuron in an entirely new, oscillatory state. The actual input system that provides the NMDA input to spinal motoneurons is unknown, but a recent study in phrenic motoneurons (Enriquez Denton et al. 2012) supports the concept that spinal central pattern generators (CPGs) for repetitive behaviors are a major source. If NMDA mediated oscillations in spinal motoneurons are in fact driven from CPGs, then this previously unknown “oscillation” state would match the electrical behavior of the motoneuron to the demands of repetitive motor tasks like scratch and locomotion.

3.3 Summary

Thus far, considerable data indicate that the electrical properties of motoneurons are reconfigured by the interactions between neuromodulation, inhibition, and NMDA-mediated inputs. We present here the hypothesis that these different states are required to match the properties of motoneurons to the extreme diversity of the normal movement repertoire. Considerable further work is required to establish this hypothesis more firmly but it appears to make good functional sense.