In their initial description of persistent inwards currents (PICs) in cat spinal motoneurones, Schwindt & Crill (1977) proposed that these currents could enable motoneurones to remain depolarized and to fire repetitively after the source of excitatory input was removed. Further, they noted that PICs would paradoxically allow synaptic currents to produce larger postsynaptic potentials in motoneurones as the cells became more depolarized. Whereas much of the attention of subsequent work has focused on the capacity of PICs in motoneurone dendrites to generate ‘plateau potentials’, bistable discharge and self-sustained firing (e.g. Hounsgaard et al. 1988; Lee & Heckman, 1996; Kiehn & Eken, 1998), it is not yet clear how common these ‘all-or-none’ behaviours are in normal motor control (reviewed in Powers & Binder, 2001). What has become increasingly clear, however, as illustrated by the study of Hultborn and colleagues (2003b) in this issue of The Journal of Physiology is that the PICs generated in motoneurone dendrites produce an enormous amplification of excitatory inputs and thereby contribute a major intrinsic source of depolarizing drive to motoneurones (Binder, 2002).
Since the 1960s, cat spinal motoneurones have provided the ‘model system’ for understanding how neurones in the mammalian central nervous system transform the synaptic inputs they receive into spike-train outputs (reviewed in Powers & Binder, 2001; Hultborn et al. 2003a). Whether excited with real synaptic inputs or equivalently with current injected through a somatic microelectrode, spinal motoneurones increase their steady-state discharge linearly with a gain of 1-3 impulses s−1 nA−1 in the primary range of firing rates and 3-6 impulses s−1 nA−1 in the secondary range (when present). More recently, quantitative measurements have been made of the amount of synaptic current that reaches the spike-generating conductances in the initial segment and axon of a motoneurone (i.e. the ‘effective synaptic current’) in response to stimulating different input systems (e.g. stretch reflex, rubrospinal, pyramidal tract, etc.) and the effects of these synaptic currents on repetitive firing (reviewed in Powers & Binder, 2001). Perhaps the most surprising finding to emerge from this latter work was how little current these afferent inputs supply to the motoneurone with respect to the amount of current required to produce the normal range of firing rates. High-threshold motoneurones in particular must be supplied with more than 50 nA of current to generate the requisite firing rates for maximal activation of the muscle fibres they innervate. However, the actual measurements of effective synaptic currents in motoneurones in anesthetized cats indicate that the major excitatory input systems can generally only deliver < 5 nA of current to the soma. This problem of insufficient ‘depolarizing drive’ is compounded by the finding that when two or more synaptic input systems are activated concurrently in these cells, the net effective synaptic currents measured are generally equal to, or slightly less than, the linear sum of their individual effects (Powers & Binder, 2000).
In the unanaesthetized decerebrate cat, however, providing adequate depolarizing drive to motoneurones does not appear to be a problem as illustrated by Hultborn et al. (2003b) and other recent studies (Lee & Heckman, 2000; Prather et al. 2001; Lee et al. 2003). Monoaminergic modulation of conductances on the dendrites of motoneurones in this preparation enhances the expression of PICs in the cells leading to a nearly a sixfold amplification in the effective synaptic currents generated by excitatory inputs (Lee & Heckman, 2000). In effect, a small excitatory synaptic input of < 2 nA can ‘trigger’ the activation of an intrinsic dendritic current that may exceed 20 nA. However, as shown here by Hultborn and colleagues (2003b) and others (Lee & Heckman, 2000; Prather et al. 2001; Lee et al. 2003) the PICs can be activated in a graded fashion, rather than in an exclusively all-or-none mode characteristic of plateau potentials and bi-stable firing behaviour, and thus, can contribute to well-controlled motor outputs.
Hultborn and colleagues (2003b) demonstrate that dendritic PICs can be activated by several different sources of synaptic inputs (stretch reflex, crossed extension reflex and pyramidal tract stimulation). By combining a triangular ramp of current injected through a somatic microelectrode with superimposed synaptic stimulation, they compare the profile of motoneurone firing rate modulation with that predicted from the product of the effective synaptic current produced by the synaptic input and the slope of the frequency-current relation (Powers & Binder, 1995). The activation of PICs is manifested in their records by a clear departure from the expected linear increase in firing rate as the level of injected current increases. Further, the slope of the frequency-current relation increased about three- to fourfold at higher firing rates, reflecting a progressive activation of a PIC as the level or spatial distribution of depolarization within the dendritic tree was increased (see also Lee at al. 2003).
Hultborn and colleagues (2003b) have also confirmed an earlier finding that the presence of PICs in the dendrites of motoneurones can dramatically enhance the efficacy of inhibition and lead to non-linear summation of synaptic inputs (Powers & Binder, 2000). They show that when PICs are activated and produce ‘secondary range’ firing, inhibitory synaptic inputs (Renshaw cells, pyramidal tract stimulation, and contralateral high-threshold afferents) decrease motoneurone discharge threefold more than expected based on the amount of effective synaptic current generated by the inhibitory inputs. Presumably, this enhanced efficacy of inhibitory inputs arises from their capacity to maintain the dendritic membrane below the threshold for PIC activation. In support of this hypothesis, they demonstrate that recurrent inhibition is substantially more effective in reducing the discharge evoked by pyramidal tract stimulation versus somatic current injection.
The principal findings reported by Hultborn and colleagues in this issue of The Journal of Physiology (2003b) are probably not unique to motoneurones, as substantial amplification of synaptic inputs by dendritic PICs has been reported in cortex and hippocampus as well (e.g. Schwindt & Crill, 1995). While it has been evident for some time that the extensive dendritic trees of many types of neurones do much more than receive, filter and transfer synaptic inputs to the soma, dendrites may also be responsible for generating a substantial portion of the input that drives neural output in the CNS.