Plateau Potentials and Rhythmic firing in Motoneurons

University of Colorado

Boulder, Colorado
June 15-17, 2000


Meeting Organizers

C. J. Heckman
Department of Physiology
Northwestern University Medical School
M211 303 E. Chicago Ave
Chicago, IL 60611, USA
Phone: 312-503-2164
Fax: 312-503-5101


Roger M. Enoka
Department of Kinesiology and Applied Physiology
University of Colorado
Boulder, CO 80309-0354, USA
Phone: 303-492-7232
Fax: 303-492-6778



The discovery by Hultborn, Hounsgaard, Kiehn and colleagues that the monoamines serotonin and norepinephrine allow spinal motoneurons in the adult cat to generate sustained plateau potentials initiated a remarkable transformation in our understanding of how these cells shape motor output. The discovery that most of the plateau is generated by L-type calcium currents by Hounsgaard, Kiehn and colleagues opened a new window into the relations between cellular properties of motoneurons and motor behavior. The initial demonstrations that plateaus exist in normal animals and humans by Kiehn and colleagues exphasized that plateaus may have a fundamental role in normal motor behavior. Recent studies in several labs have shown that the L-type calcium channels and the plateau potential are primarily generated in the dendritic tree of the motoneuron. As a result of continued intensive research along these lines of inquiry, our understanding of synaptic integration in motoneurons is undergoing a rapid transformation. This symposium was organized to discuss the most recent findings on plateau potential and to consider their significance for generation of motor unit firing patterns in human subjects. The following figure defines the basic phenomena:

The L-type calcium channels in the dendrites of motoneurons are voltage-sensitive and highly persistent — that is, once activated, these channels tend to stay open. This produces a persistent inward current. Thus, when a synaptic input is applied to the motoneuron during voltage clamp, the persistent inward current amplifies the synaptic current while the input is maintained and then continues to generate current on its own after the input ceases (see the bottom trace in the figure). When the cell is not voltage clamped, the persistent inward current produces a sustained plateau potential (middle trace — spikes in this cell have been eliminated by the intracellular sodium channel blocker QX-314). When the cell is allowed to discharge normally, the persistent inward current produces strong firing during the input and self-sustained firing for a long duration after the input ends. In all three conditions (voltage clamp, QX-314, and normal firing), the cell acts in a bistable manner. That is, the cell can be returned from the excited state generated by the persistent inward current to its rest level by a brief period of hyperpolarization (The synaptic input for all traces in this figure was generated by steady firing in muscle spindle Ia afferents. The upper and lower traces are taken from the same cell; the middle trace is from a different cell and experiment).



Converging intracellular pathways for metabotropic modulation of intrinsic response properties in spinal motoneurons

J.F, Perrier and J.Hounsgaard

Panum Institue, Copenhagen, Denmark

In the turtle L-type calcium channels in spinal motoneurons are facilitated by serotonin, glutamate and acetylcholine acting on specific metabotropic receptors. We have found that this facilitation for all 3 transmitters is mediated by an increase in the concentration of intracellular calcium and involves activation of calmodulin. We also show that calcium influx during calcium spikes also facilitate L-channels in a calmodulin dependent manner. These results suggest that intracellular calcium and calmodulin are central elements in controlling motoneuronal response properties and excitability on time scales of seconds and minutes.

Experiments on explant cultures of transverse sections of the spinal cord of adult turtles show that motoneurons loose functional L-type calcium channels and acquire T-type calcium channels and A-type potassium channels after few days in culture. The resulting electrophysiological phenotype resembles the properties found in embryonic and neonatal spinal motoneurons in other vertebrates. We suggest that this induced dedifferentiation may represent an extreme case of regulated expression of the ion channels responsible for the intrinsic response properties of motoneurons. We speculate that regulation of channel expression could contribute to normal function and pathology on time scales of hours and days.

Perrier JF, Mejia-Gervacio S, Hounsgaard J: Facilitation of plateau potentials in turtle motoneurons by a pathway depending on calcium and calmodulin. J. Physiology In press.

Perrier JF, Hounsgaard J: Development and regulation of flexible response properties in motoneurons. Brain Res Bull In press.

Perrier JF, Noraberg J, Simon M & Hounsgaard J: Intrinsic Response Properties Of Motoneurons In Organotypic Cultures Of The Spinal Cord Of The Adult Turtle. Eup J Neurosci In Press.

Delgado-Lezama R, Hounsgaard J: Adapting motoneurons for motor behavior. Progress in Brain Research 23: 57-63, 2000.


DendritIc plateau potentials in functionally mature motoneurons in vitro

K. Carlen and R. Brownstone

University of Manitoba, Winnepeg, Canada

In motor functionally mature mouse spinal motoneurones, studied using whole cell patch clamp techniques in slice, there are various types of voltage-activated calcium currents. Of these, one is a dihydropyridine-sensitive current originating in the dendrites. Immunohistochemical evidence shows alpha-1D subunits (a component of one type of "L-type" or dihydropyridine-sensitive, current) on dendrites, and modelling evidence supports this location for the observed currents (Carlin et al, EJN 12:1635-1646, 2000). Plateau potentials can be demonstrated in these cells, which is consistent with a non-inactivating dendritic inward current. Furthermore, in many cells, these currents may activate at slightly different times and with different time constants (non-orderly time constants) within a single voltage step, indicating that the current may activate "independently" in different dendrites.


Dendritic conductances and synaptic integration in motoneurons

Randall K Powers and Marc D. Binder

University of Washington, Seattle, USA

There is now a growing body of evidence for the presence of voltage- and calcium-sensitive conductances in the dendrites of motoneurons. However, the role that these conductances play in synaptic integration is not well understood. Persistent voltage-activated inward currents are likely to result in amplification of the synaptic current reaching the soma and the adjacent axonal site of spike initiation. Different channel types may contribute to these inward currents in different types of motoneurons. In turtle,mouse and possibly cat spinal motoneurons, L-type calcium channels appear be the primary source of this inward current. However, recent work in rat hypoglossal motoneurons indicates that in these cells most of the low threshold inward current is due to a persistent sodium current. Computer simulations indicate that at low levels of excitatory synaptic input, persistent inward currents on the dendrites can amplify the synaptic current transmitted to the soma. In contrast, at higher levels of excitatory input, activation of voltage-sensitive potassium currents can reduce the transmitted synaptic current to below the levels transmitted by passive dendrites. This latter feature should manifest itself as less-than-linear summation of the effects of two separate synaptic inputs. It is difficult to reconcile this simulation result with recent work indicating both amplification of synaptic currents and linear summation of the effects of two synaptic inputs.


All dendrites are not created equal: insights gained from simulations of anatomically realistic models of spinal motoneurons

Ken Rose, Sharon Cushing, Tuan Bui, Maria Ter-Mikalean, Monica Nueber-Hess

Queens University, Kingston, Canada

There is a wide consensus that motoneuron input/output properties are governed, to a large extent, by the presence of a large and complex dendritic tree. However, it is generally assumed that details of the dendritic structure, eg. the direction followed by dendrites en route to their termination, are of little functional significance. The goal of my talk was to convince the assembled experts that this view is simplistic and mis-leading. 

Compartmental models were constructed based on detailed measurements of the geometry of the dendritic trees of 3 electrophyiologically identified and intracellularly stained motoneurons. All motoneurons supplied neck extensor muscles of the adult cat. Simulations which assumed uniform passive membrane properties indicated that the current delivered to the soma from synapses on different dendrites depended on the trajectory followed by the dendrites, eg. rostro-caudal versus dorsolateral. A cluster analysis revealed that the 10 morphological classes of dendrites (based on their trajectory) could be divided into 2 functional subdivisions, based on their ability to deliver synaptic current to the soma. At modest levels of backgound synaptic activity (eg. Rm=4000 ohms-cm2), there was a 2 to 1 difference in the ability of the subdivisions to deliver current to the soma. Thus, at least from the perspective of dendritic trajectory, all dendrites of motoneurons are not functionally equivalent. 

The introduction of voltage-dependent channels on the dendrites of spinal motoneurons may subdivide the dendritic tree into additional functional subunits. Using experimental data published by David Bennett, Monica Gorassini, Brent Fedirchuk and Hans Hultborn, we attempted to identify the location of the voltage-dependent channels responsible for plateau potentials/currents. Once again, we took advantage of simulations obtained from compartmental models of motoneuron dendritic trees. The membrane potential throughout the dendritic tree was determined under 2 conditions. In the first condition, the cell was assumed to be 'at rest' (ie a low level of background synaptic activity) and the soma membrane potential was set at a value just subthreshold for the initiation of plateau potentials. In the second condition, a uniformly distributed set of synapses were activated (total curent reaching the soma approximately 5 nA) and the soma membrane potential was once again set at a value just subthreshold for the initiation of plateau potentials. Assuming that the threshold for activation of the voltage-dependent channels is the same throughout the neuron, the intersection of plots of membrane potential versus distance along each dendrite provided a measure of the most likely location of the channels responsible for plateau currents. These locations were 150 to 250 microns from the soma, on second or third order dendrites. This arrangement provides a mechanism for selective amplification of those synaptic inputs that are located more distally. Hence, the voltage-dependent channels responsible for the plateau potential may dynamically fractionate the dendritic tree depending on the level and distribution of the synaptic input.



Mechanisms underlying increased motoneurone excitability in the spastic mouse

R.J. Callister

Univesity of Newcastle, Callagan, Australia

The spastic mouse trembles violently and exhibits an impaired righting reflex when frightened or disturbed. These symptoms are somewhat similar to those observed in human spasticity, a condition thought to result from the uncontrolled discharge of motoneurones. The primary defect in spastic has been traced to a mutation in the glycine receptor (GlyR) that causes a reduction in transcriptional efficiency of the b subunit of the receptor. While much is understood about the physiological properties of recombinant GlyRs, expressed in isolated cell systems, little is known about the properties of the native receptor (i.e., the a 1/b -subunit heteromer) in the spastic mutant. In order to address this issue we are using a brainstem slice preparation to examine the physiological properties of native GlyRs in the spastic mutant. Glycinergic miniature inhibitory postsynaptic currents (mIPSCs) recorded from hypoglossal motoneurones in the spastic mutant are reduced by two- to three-fold, but the decay of the mIPSC is unchanged. These data suggest that the spastic mutation cause a large reduction in the number of functional GlyRs whose properties remain unchanged. Data obtained from the responses of excised outside-out patches to low concentrations (10-50 m M) of glycine suggest that the spastic mutant is assembling a 1 homomeric receptors, but their location is restricted to sites outside the synaptic cleft. We also present evidence for a compensatory upregulation of GABAAergic receptors in the spastic mutant. This finding confirms the close historical and functional association between glycinergic and GABA-mediated transmission on motoneurones in the spinal cord and brainstem.


Location of Renshaw cell input on motoneuron dendrites

Thomas M. Hamm and Mitchell G. Maltenfort

Barrows Institute, Phoenix, USA

The location of a tonic dendritic conductance can be localized from changes in the shape in the impedance function, as shown by Fox (1985) and Fox and Chan (1985). We have developed and applied this technique to the investigation of dendritic conductances in spinal motoneurons. Simulations were performed using compartmental models based on the six motoneurons identified by motor unit type whose morphology and electrotonic properties had been characterized (Cullheim et al. 1987; Fleshman et al. 1988). These simulations demonstrated that the location and relative magnitude of a tonic dendritic conductance can be determined based on two parameters, the reversal frequency and the relative change in impedance at low frequency, respectively. The reversal frequency describes a characteristic of the change in the magnitude of the impedance function produced by a dendritic conductance. As frequency increases, the decrease in impedance becomes less until it actually reverses sign and becomes an increase, i.e., the impedance magnitude with the additional dendritic conductance exceeds that without the conductance. The frequency at which the change in impedance magnitude is zero is the reversal frequency, and it is less, the more distal the location of the added conductance.

This method was applied to determining the dendritic location and magnitude of the conductance produced by recurrent inhibition. Experiments were performed in pentobarbital anesthetized cats with sectioned dorsal roots. Current was injected into each motoneuron in the form of a sum of sinusoids of different frequencies to determine impedance. Recurrent inhibition was produced by stimulation of muscle nerves at 200 Hz in alternate trials. According to these determinations, the average location of Renshaw synapses was 0.22 length constants from the soma, in good agreement with the estimate of 0.25 by Fyffe (1991), based on morphological measurements. The average conductance magnitude was 75% of the resting conductance in the dendritic regions containing the synapses. Based on the measurements in these experiments, we estimate that a third of the inhibitory effect of recurrent inhibition on a motoneuron at threshold should derive from its conductance. More recently we examined the extent by which Ia EPSPs are reduced in amplitude by concurrent recurrent inhibition. These values were compared with the reduction expected from impedance-derived estimates of the location and magnitude of recurrent inhibition and published information on the dendritic distribution of Ia synapses (Burke and Glenn 1996). These preliminary data show a reasonable agreement between the reduction in Ia EPSP amplitude found experimentally and the expected value obtained from simulations using the data on synaptic conductance magnitudes and locations.


Neuromodulation of the current (I) - frequency (f) relation of turtle motoneurons

T. George Hornby, Jennifer C. McDonagh, Robert M. Reinking, and Douglas G. Stuart

University of Arizona, Tucson, USA

This presentation summarized a study that quantified the effects of excitatory and inhibitory neuromodulation on selected passive, transitional, and active (repetitive-firing) properties of turtle spinal motoneurons (MNs) in slices of spinal cord (SC). Responses were noted under control conditions. and following application of one of four excitatory modulatory conditions (serotonin [5-HT], muscarine, trans-1-amino-1,3-cyclopentanedicarboxylic acid [tACPD], all combined). A sample of 44 MNs was divided into two groups on the basis of whether MNs demonstrated (28/44) or did not demonstrate (16/44) an excitatory modulator engendered,nifedipine-sensitive acceleration of discharge during constant current stimuli indicative of plateau potential (PP) behavior. The most pronounced of the differences between non-PP and PP MNs was in their I-f relation. To describe this difference, it was necessary to measure a two-phase relation, in addition to the conventionally measured one phase relation. In PP MNs, excitatory modulation considerably steepened the first (initial) phase of the I-f relation and flattened the second (later) phase. This latter result bore similarities to the results obtained previously in a study on the decerebrate cat by Brownstone et al. in which MNs were studied during fictive locomotion (Exp. Brain Res. 90: 441-455, 1992).

Also addressed in this presentation were comparative aspects of the "preferred firing range" of turtle MNs. This term refers to a pattern of human motor unit (MU) discharge in which the profile of the spike-frequency of a MU's extracellulary (EMG) recorded compound action potentials is dissociated from the profile of the presumed depolarizing pressure exerted on the unit's spinal motoneuron (MN) during a voluntary contraction.Recently, such a dissociation has been been attributed by inference to the presence of a PP in the active MN (Kiehn and Eken J. Neurophysiol. 78:3061-3068, 1997). This inference was evaluated qualitatively by comparing the preferred firing range of human MUs to the analogous behavior of turtle MNs recorded IC in in vitro SC slices, and previously reported IC-recorded cat MNs in an in vivo decerebrate preparation (Lee and Heckman J.Neurophysiol. 78: 3061-3068, 1998). In both latter cases, the MNs were known to be generating PPs, as evoked by excitatory neuromodulation. The qualitative similarities in the MN/MU responses of the three species gave credence to the proposition that PPs may indeed underlie the preferred firing range of human MNs. This conclusion was further supported by comparing the rate-limiting behavior of IC-recorded turtle MNs while generating PPs in the same SC slice preparation to that of human MUs, again during the elaboration of a slowly rising voluntary contraction (Monster and Chan J. Neurophysiol. 40: 1432-1443, 1977). For this latter comparison, rate-limiting refers to a progressively lessening increase in spike-frequency response as stimulus intensity is increased: i.e., it is a special form of preferred firing range behavior. Again, there were sufficient similarities in the rate-limiting behavior of MNs/MUs in the two species to support the inference of PP generation in human MNs.  In the discussion of the above results and comparisons, a variety of open issues concerning the PP were summarized, together with selected findings on the potential functional significance of PPs across several vertebrate species.


Serotonin 5-HT2C receptor activation induces a long-lasting amplification of reflex gain in the isolated rat spinal cord

D.W. Machacek, S.M. Garraway, B.L. Shay and S. Hochman

Emory University, Atlanta, USA

C-fiber activation induces an LTP in the spinal flexion reflex in mammals, via mGLUR1 and NMDA receptor dependent mechanisms, presumably to provide enhanced reflexive protection of damaged tissue from further injury. Descending monoaminergic pathways, while generally thought to depress sensory input, may also amplify spinal reflexes, but the details within spinal cord remain to be elucidated. Here, we demonstrate that 5-HT is capable of inducing a long-lasting (hours) increase in reflex gain in the isolated lumbar spinal cord of rat maintained in vitro at all ages examined (postnatal days 2- 12). Pharmacological analyses indicate an essential requirement for activation of 5-HT2C receptors while 5-HT1A, 5-HT7 and 5- HT2A receptors were not involved. High threshold afferents were most facilitated by 5-HT, but low-threshold and APV-insensitive short-latency reflexes were also facilitated. In addition, sensory- evoked synaptic potentials recorded in a subpopulation of laminae IV-VII spinal neurons were similarly facilitated. Thus 5- HT receptor-evoked facilitatory actions are complex, involving alterations in neuronal integrative properties at both motoneuronal and pre-motoneuronal levels. This study provides the first demonstration of a descending transmitter, acting via a specific receptor subtype, producing a long-lasting amplification in reflex gain, and suggests that brain modulatory systems interact with spinal segmental pathways to control the reflex gain.



Discharge and membrane properties of prenatal motoneurones studied in tissue culture

Daniel Kernell, Rob Bakels, Nieske Brouwer, Sjef Copray, JanMulder, Bert Otten, Erik Boddeke

University of Groningen, The Netherlands

Repetitive firing properties and membrane characteristics were investigated in immature dissociated motoneurones (MNs), grown for 6-7 days in tissue culture in the vicinity of muscle fibres. This preparation was explored as a model for future work concerning the plasticity and developmental biology of MN properties. The cells were obtained from the spinal cord of E15 rat embryos. Presumed MNs were separated from other cells using trypsin incubation, metrizamide gradient centrifugation and immuno-panning for p75 membrane receptors. Electrical recordings were obtained using whole-cell current- and voltage-clamp techniques.  

When stimulated with long-lasting steady currents, many of the cells were capable of repetitive impulse firing lasting several sec. Firing rates were low, on average between about 9 and 14 Hz, and the relation between discharge rate and current intensity was roughly linear. In comparison to adult MNs, the cells were highly excitable with a low threshold current (about 0.04 nA) and a steep slope for the frequency-current relation (f-I slope; around 300-600 Hz/nA). All cells showed a frequency-adaptation similar to the "late" adaptation in adult MNs (gradual decline of rate during constant stimulation). In contrast, the "initial" phase (marked initial drop of rate and f-I slope) was clearly seen in only about 1/3 of the cells. Action potentials were long-lasting (mean about 12 ms) and never succeeded by a delayed depolarization. The afterhyperpolarization (AHP) lasted ca. 40 - 130 ms.

Voltage clamp measurements performed in combination with various blocking agents revealed the presence of several seemingly normal kinds of ion channels: an inactivating sodium channel; two types of potassium channels (delayed and transient rectifier); two to three kinds of high-voltage-activated calcium channels (L- and N-type in all cells, P-type in some). The AHP was relatively insensitive to the blocking actions of apamin (only partial or no block of initial portion) and most of the gradual decline of this afterpotential was passive; a direct proportionality was found between the duration of AHP and the membrane time constant. Model studies supported the view that the long duration of the action potentials and the high f-I slopes were (partly) caused by a low density of the appropriate ion channels. The high input resistance and low current-threshold were, in addition, related to the small cell size and the absence of well-developed dendrites.


Development of tonic firing in rat soleus muscle

Torsten Eken, Geoff Elder, and Terje Lømo

University of Oslo, Norway

During long-lasting tonic firing in rat soleus muscle, one or a few slow motor units fire at remarkably stable frequencies around 20 Hz for up to several minutes (1). This preferred firing range corresponds to 80% of maximal tetanic muscle tension, and is probably optimal for energy efficient use of the muscle fibres. Activity is cycled between different motor units in different tonic activity periods. We have previously argued that this characteristic long-lasting tonic firing may be due to the activation of monoamine-dependent plateau potentials that generate self-sustained firing: Abrupt jumps from lower frequencies or inactivity to long-lasting firing in the preferred firing range can often be seen (1, 2); they can be reproduced by excitatory inputs to soleus motoneurones via Ia afferents (2); jumps from this range to lower frequencies or inactivity can be induced by inhibition of motoneurones through cutaneous afferents (2); and long-lasting tonic activity is considerably reduced after depletion of spinal serotonin and noradrenaline (3).

Previous investigators have concluded that development of tonic firing in rat soleus muscle is completed by 20 days after birth (4, 5). However, the adult pattern and density of serotonin immunoreactivity in the lumbar cord does not appear until 21 days (6, 7), while the innervation pattern of noradrenergic fibres is similar to adult at 31 days (8). Thus, it is likely that further development of firing activity may take place after 20 days.

We have studied single-motor-unit and gross EMG in rats from 7 days to adult. There was no significant change of motor-unit firing rates, at least from 10 days to adult. Long-lasting tonic gross-EMG segments with adult shape appeared at the end of the third week. However, the duration of individual tonic activity periods increased dramatically until well after two months, the 90-percentile value increasing almost 20-fold. Awareness of this prolonged development might be important when selecting experimental models for studies of motoneurone properties.

1. Eken T. J. Neurophysiol. 80:365-377 (1998).

2. Eken T. and O. Kiehn Acta Physiol. Scand. 136:383-394 (1989).

3. Kiehn O., J. Erdal, T. Eken, and T. Bruhn J. Physiol. (Lond.) 492:173-184 (1996).

4. Navarrete R. and G. Vrbová Dev. Brain Res. 8:11-19 (1983).

5. Westerga J. and A. Gramsbergen Dev. Brain Res. 80:233-243 (1994).

6. Bregman B. S. Dev. Brain Res. 34:245-263 (1987).

7. Rajaofetra N., F. Sandillon, M. Geffard, and A. Privat J. Neurosci. Res. 22:305-321 (1989).

8. Rajaofetra N., P. Poulat, L. Marlier, M. Geffard, and A. Privat Dev. Brain Res. 67:237-246 (1992).


Fast and slow persistent inward currents in spinal motoneurons function as a dual amplification system for synaptic inputs

R. H. Lee and C.J. Heckman

Northwestern University, Chicago, USA

In spinal motoneurons in the adult cat preparation, a strong link was demonstrated 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 was isolated during voltage clamp by superimposing a fast sine wave (typically 125 Hz, 0.5 mV) on a linearly rising and falling voltage command. These parameters were found to be fast enough to exclude activation of slow dendritic currents, such as that mediated by the L-type Ca channel, but still slow enough to allow full inactivation of the inactivating Na+ current that generates the action potential. Consequently, the fast persistent current responding to the sinusoid was probably due to the persistent component of the Na current. The fast persistent current appeared to have a somatic location, as indicated by good clamp control and lack of hysteresis on ascending and descending voltage ramps. When the fast persistent current was markedly reduced by prolonged depolarization, rhythmic firing failed and could not be restored by increasing the injected current. More modest reductions in the amplitude of the fast persistent current produced proportional reductions in the gain of the motoneuron frequency-current function. Thus, this fast persistent current, which may be primarily due to persistent Na channels, is essential for initiation of spikes in rhythmic firing and plays a major role in setting frequency-current gain.

Preliminary data suggest that the fast persistent current is not altered by the monoamines, being of similar amplitude in pentobarbital anesthetized preparations and decerebrate preparations. In contrast, a slow persistent current with a dendritic origin is strongly potentiated by the monoamines. This current, perhaps predominatly generated by L-type Ca channels, markedly enhances synaptic input in the decerebrate preparation. This enhancement was seen in all cells, regardless of whether they exhibited bistable behavior. Preliminary experiments with linearly rising currents generated by muscle stretch indicate that this dendritic persistent current acts as an amplifier — that is, it multiplies the gain of the frequency-current function. The gain from synaptic current to rhythmic firing was found to be 3 to 6 times larger than that for injected current. This difference likely arises from the ineffectiveness of somatic current injection in activating a dendritic persistent inward current. A simple steady state model of the motoneuron might be constructed from two amplifiers coupled together, one for the dendrites (due mainly to L-type channels) and one for the soma (due mainly to persistent Na channels). Both amplify the synaptic current, but the somatic one is essential for generation of rhythmic firing while the dendritic one is under neuromodulatory control.


Amplification and summation of synaptic input in motoneurons under tonic monoaminergic input

T. Cope and J. Prather

Emory University, Atlanta, USA

Knowledge of synaptic physiology alone can be misleading as to the effects of a synaptic input on motoneuron (MN) firing output. For example, medial gastrocnemius (MG) muscle stretch afferents (Burke et al. 1976; Fleshman et al. 1981) and caudal cutaneous sural nerve afferents (Kanda et al. 1977; Pinter et al. 1982; cf. Heckman et al. 1994) have very different organizations across the MG pool, yet they elicit the same MN recruitment sequence (  and Clark 1993). Therefore in the current study, we compare patterns of repetitive firing elicited by these two inputs in MG MNs to determine the functionally relevant differences in the organization of those inputs.

All data were recorded intra-axonally so that the site of synaptic integration was left undisturbed. MG muscle stretch and sural nerve synaptic inputs both caused firing rates to be modulated in MG MNs of decerebrate cats. We quantified that modulation by comparing instantaneous firing rates (sampled using the algorithm of Berger et al. 1986) during periods of coactivity within pairs of MNs. In those rate-rate plots, clusters of points were apparent. Those clusters were linear in shape, permitting them to be analyzed using a reduced major axis regression.

In the pair of MNs given as an example, the regression slopes elicited by muscle stretch and sural nerve inputs were indistinguishable. In addition, the magnitude of that slope was 1.6, indicating that the firing rate of the second-recruited MN was modulated more than that of the first-recruited MN.

Across a sample of 20 MN pairs, each pair was represented as a single point of a graph of the average slope elicited by stretch input plotted against the average slope due to sural input. Those points tended to lie along the line of identity, indicating that the effect of those two inputs was indistinguishable across the population of MG MNs. In 17/20 pairs, both inputs elicited a slope greater than 1, and in no case did both inputs elicit slopes less than 1.

These phenomena were considered in light of various pre- and post-synaptic mechanisms that might explain the results. One possible interpretation is that all post-synaptic integration properties were the same across MNs, but both inputs were arranged to give systematically greater drive onto less excitable MNs. Given the organization of stretch input described in previously published results, that possibility is not likely to be correct.

Alternatively, the synaptic input could be essentially uniform across the MN pool, but post-synaptic integration properties could be different across cells. Specifically, the recently described phenomenon of synaptic input amplification was considered. It has been shown recently that both muscle stretch and sural inputs are amplified and are usually amplified similarly in the same cell. Additionally, linear summation of those inputs is maintained even in the presence of that amplification (Prather et al. in press). It has also been shown that amplification magnitude is larger in less excitable MNs than in the more excitable cells (Lee and Heckman, in press).

These amplification mechanisms may underlie the present findings.  Synaptic inputs from muscle stretch and sural nerve afferents are equally  efficacious in activating amplification. Combined with a difference in amplification magnitude across MNs, these mechanisms may explain the finding that MG muscle stretch and sural nerve synaptic inputs caused indistinguishable patterns of firing rate modulation across the MG pool and that the pattern was to elicit faster firing in the second-recruited MN than in the first.

Berger RD. Akselrod S. Gordon D. Cohen RJ. An efficient algorithm for spectral analysis of heart rate variability. IEEE Transactions on Biomedical Engineering 33: 900-4, 1986.

Burke RE. Rymer WZ. Walsh JV. Relative strength of synaptic input from short-latency pathways to motor units of defined type in cat medial gastrocnemius. Journal of Neurophysiology 39: 447-458, 1976.

Cope TC. Clark BD. Motor-unit recruitment in self-reinnervated muscle. Journal of Neurophysiology 70: 1787-96, 1993.

Fleshman JW. Munson JB. Sypert GW. Homonymous projection of individual group Ia-fibers to physiologically characterized medial gastrocnemius motoneurons in the cat. Journal of Neurophysiology 46: 1339-48, 1981.

Heckman CJ. Miller JF. Munson M. Paul KD. Rymer WZ. Reduction in postsynaptic inhibition during maintained electrical stimulation of different nerves in the cat hindlimb. Journal of Neurophysiology 1: 2281-93, 1994.

Kanda K. Burke RE. Walmsley B. Differential control of fast and slow witch motor units in the decerebrate cat. Experimental Brain Research 9: 57-74, 1977.

Pinter MJ. Burke RE. O'Donovan MJ. Dum RP. Supraspinal facilitation of cutaneous polysynaptic EPSPs in cat medical gastrocnemius motoneurons. Experimental Brain Research 45: 133-43, 1982.



Single and doublet firing of human motoneurones: two patterns or two mechanisms?

Lydia Kudina

Institute for Information Transmission Problems, Moscow, Russia

Analysis of interspike intervals and firing pattern of motoneurones capable and uncapable of firing doublets as well as testing excitability recovery after a discharge suggest the leading role of the delayed depolarization in the origin of human doublets. It was also suggested that two distinct mechanisms operated by different synaptic inputs ("common drive" inputs and specifical dendritic inputs evoking the dendritic plateau depolarization) account for single-spike and doublet firing of human motoneurones. 


Motoneuron Firing Patterns and AHP

M. Piotrkiewicz and L. Kudina

Institute of Biocybernetics and Biomedical Engineering, Warsaw, Poland


Institute for Information Transmission Problems, Moscow, Russia

Discharge characteristics of human motoneurons (MNs) supplying fast (brachial biceps, BB) and slow (soleus, SOL) muscles were investigated in nine healthy volunteers. Motor unit (MU) potentials were recorded by means of intramuscular selective electrodes. Two kinds of experiments were performed. During the first one, voluntary MU activity during weak and moderate muscle effort was analyzed. EMG was decomposed to constituent single MU action potential trains (MUAPTs) from which standard deviation (SD) and mean value (Tm) of interspike intervals (ISIs) were calculated and presented as SD(Tm) plots. The second experiment was based on the method developed by Kudina and Alexeeva (1992). The tibialis posterior nerve was stimulated with paired stimuli of variable interpulse interval and single MU responses from a relaxed muscle were analyzed. The recovery of their motoneurons’ excitability after the conditioning stimulus was evaluated and the time of complete recovery was taken as an indirect measure of afterhyperpolarization (AHP) duration.

Single MU SD(Tm) plots were obtained for 46 MUs from BB and 77 MUs from SOL. Usually, on each plot two ranges of ISI could be distinguished: short-interval part with low SD and long-interval part where SD was higher and dependent on Tm. The position of each plot with respect to Tm axis was quantified, whenever possible, by estimating its break-point as an intersection of two straight lines fitted to short- and long-interval parts of the data. The break-points for BB (range 70.9—130.6 ms) were located at ISIs shorter than those for SOL (range 99.0—199.5 ms).

The range of presumed AHP duration estimated in 23 SOL MNs was 150—240 ms. Also, for six MNs for which both break-point and AHP duration was determined, the latter was systematically higher.

It is concluded, that the features of the relationship SD(Tm) reflect the features of interspike potential trajectory of a MN. The position of this relationship with respect to the mean ISI axis, which can be roughly estimated by the break-point, can be considered as a measure of the „fastness" of a motoneuron. The slope of the short-interval part is related to the degree of the convergence between the trajectories for different mean ISI. The results indicate that the trajectories for most of MNs are fan-shaped although about 20% of MNs may be judged as having parallel trajectories. The considerable range of break points and slopes of both parts of SD(Tm) relationships for each muscle indicate that the properties of the MN pools are not uniform.


Different modes of inter-spike interval shortenings induced by juxtathresholds Ia inputs on steadily discharging human motoneurones

B. Mattei and A. Schmied

DPM -CNRS. Marseilles, France

Data obtained with neocortical neurons tested in vitro have revealed that a transient excitatory input delivered in the first half of the inter-spike interval (ISI) are liable to shorten the timing of the next firing with a delayed effect which differs from the shortening associated to the direct threshold-crossings which occur later in the ISI (Reyes and Fetz, 1993). To investigate the possible existence of a similar phenomenon with motoneurones, the effect of a transient excitatory input has been tested on the firing time of tonically discharging human motor units tested in the extensor carpi radialis muscles during isometric contraction. The excitatory input was generated by 100 mm-tendon taps delivered randomly with a mean frequency of 0.5 Hz. The changes in the ISI during which the stimulation occurred and in the following ISI were evaluated in percent of the value of the ISI preceding the stimulation. The significance (P < 0.001) of the ISI changes was determined in comparison with the spontaneous ISI fluctuations associated to a fictive stimulation generated 0.5 s after each tendon tap. When the stimulation was delivered in the second half of the ISI, it induced an advanced firing time-locked to the stimulation at a monosynaptic-like latency compatible with a direct threshold-crossing. The following ISI was slightly lengthened consistently with previous observations. When the stimulation was delivered in the first half of the ISI, the next firing was also advanced but this occurred with no clear coupling with the stimulus over a broad period ranging from 40 to 80 ms after the monosynaptic responses detected in the post-stimulus time histograms (latencies 20-30 ms, duration 2.75 - 6 ms). Most surprisingly, the ISI following the stimulation delivered in the first half of the ISI tended to be also shortened suggesting a rather prolonged effect clearly differing from a simple threshold-crossing process. Further experimentation is awaited in order to determine its origin.


How reliable are the estimates of AHP parameters and common synaptic drive for human motoneurons based on statistical analysis of motor unit spike trains?

Marc D. Binder and Randall K. Powers

University of Washington, Seattle, USA

We made intracellular recordings from cat lumbar motoneurons and induced them to fire repetitively by injecting "noisy" current waveforms through the microelectrode.  The resultant  spike trains were comparable to those recorded from human subjects during sustained, voluntary contractions with respect to their interspike interval distributions.  We found that the relationship between mean firing rate and its standard deviation did not always provide a reliable estimate of the AHP duration as previously proposed (e.g., Piotrkiewicz.  J. Physiol. Paris 93: 125-133, 1999).  However, the transformation of the interval death rate into a voltage-time plot as described by Matthews (J. Physiol. Lond. 492: 597-628, 1996) did provide a good estimate of the distance-to- threshold throughout much of the interspike trajectory.  We also explored the relationship between the central peak in a crosscorrelation histogram and the amount of common input used to drive the repetitive discharge of a motoneuron on successive trials.  Although we found a non-linear parabolic relationship between the amount of common input and the degree of synchrony in motoneuron discharge, estimates of the amount of common input responsible for the synchrony observed in human studies (25-70%) appear to be quite reasonable (e.g. Bremner et al. J. Physiol. Lond. 432: 381-399, 1991; Nordstrom et al.  J. Physiol. Lond. 432: 381-399, 1992).  The apparent relationship between synchronization and the discharge rate of the motoneurons depended on which "index of synchronization" was used.  This remains a complex problem that requires more systematic analysis.



Properties of motor units in hand muscles: implications for cortical control of finger movement

Andrew J. Fuglevand and Douglas A. Keen

University of Arizona, Tucson, USA

The ability to move the fingers relatively independently provides humans and non-human primates with a sophisticated repertoire of motor behaviors. Given such exquisite control, one might expect the contributions of single motor units in the multi-tendoned muscles that supply each of the fingers to be highly selective for a particular digit. Instead, we found that spike-triggered average forces for motor units in extensor digitorum (ED) were broadly distributed over the four fingers. Indeed, we never observed a motor unit that had all of its force concentrated on a single digit.  Similar results (albeit with somewhat greater selectivity) have been obtained in preliminary experiments in the main flexor muscles of the fingers. To test the role that physical linkages between the distal tendons might play in distributing force across the fingers, weak electrical stimulation was used to directly activated small bundles of muscle fibers in different compartments of ED. The stimulating microelectrodes were inserted into the distal portion of ED to minimize the possibility of activating the motor nerve, which enters the muscle at a mid-proximal level. Forces evoked by electrical stimulation were markedly more localized to individual fingers than was found for single motor units. Therefore, the broad distribution of force seen in ED motor units was likely due to factors other than mechanical coupling among tendons. Furthermore, during voluntary contraction, little synchrony was seen for pairs of motor units residing in different compartments of ED. Consequently, it seems likely that the main reason for the broad distribution of motor unit force in ED is that single motor axons, while primarily innervating muscle fibers in one compartment, probably supply muscle fibers in other compartments of ED as well.

Given the apparent absence of independent actuators for the fingers, it was of interest to examine how multiple muscles might be coordinated in the elaboration of individual finger movements. We recorded EMG activity from 24 muscle/muscle compartments in the forearm and hand during unloaded flexion-extension movements of each of the fingers. For most of these movements, several muscles showed significant activity. Therefore, these "simple" movements actually required a delicate and complicated counterbalance among multiple muscles in order to achieve isolated movement of an individual finger. Furthermore, an individual muscle was active in many finger movements, not just those in which it would be considered to be a prime mover. This finding might help explain a puzzling feature of neural activity in primary motor cortex (M1). Namely, most neurons in the hand/finger region of monkey M1 are active during individuated movements of several digits (Schieber & Hibbard, 1993). Accordingly, pyramidal neurons in M1 might chiefly encode the activity of single muscles or sets of muscles. However, because individual muscles participate in several types of movements, this factor might tend to blur functional identification of M1 neurons, particularly when viewed from a body segment, rather than from a muscle, frame of reference.


Acute and chronic adaptations of motor unit firing rate in humans

Jacques Duchateau

Free University of Brussels, Belgium.

The purpose of the presentation was to provide examples of acute and chronic adaptations in motor unit (MU) behavior. The example of an acute adaptation was muscle fatigue, whereas the example of chronic adaptation was several weeks of physical training to improve power production.

There are several reports in the literature on the behavior of low-threshold MUs during a fatiguing contraction (Enoka et al. J. Neurophysiol. 62 : 1344-1359,1989; Maton and Gamet Eur. J. Appl. Physiol 58 : 369-374, 1989; Garland et al. J. Appl. Physiol. 76 : 2411-2419, 1994). These studies found that the recruitment threshold and discharge rate of low-threshold MUs do not change substantially during a fatiguing contraction. Because the subjects experienced fatigue, however, some MUs must have altered their behavior during the task. To examine this possibility, we studied the contractile properties, firing rate and recruitment threshold of low- and high-threshold MUs in the first dorsal interosseous muscle during intermittent fatiguing contractions. The task consisted of reaching 50% of MVC in 3 s, maintaining this level for 10 s, and then slowly returning to base line in 3 s, which was repeated until exhaustion. A rest period of 4 s was allowed between two successive contractions. The endurance limit ranged from 6 to 15 min and a total of 63 units with recruitment thresholds between 0.5 and 55.2% of MVC were tested. Different behaviors were observed as a function of the recruitment threshold of the unit. Low-threshold (<25% MVC) MUs exhibited either no change or an increase in recruitment threshold, which was accompanied by a small decrease in discharge rate during the force plateau (10 s hold). Conversely, high-threshold (> 25% MVC) units showed a systematic decrease in activation threshold and a more drastic reduction in discharge rate. In addition, some units not active at the beginning of the test (threshold > 50% MVC) were recruited during the course of the fatigue test and all showed a progressive decline in recruitment threshold with an initial increase in discharge rate, which decreased toward the end of the test. Similarly, the changes in spike-triggered averaged force varied with recruitment threshold; it increased for low-threshold units and decreased for high-threshold units. An important feature of these findings was that despite a progressive enhancement of activation during the fatiguing contraction, the discharge rate of MUs either did not change, decreased, or increased depending of the recruitment threshold of the MU.

As with acute adjustments in MU activity, long-term interventions can involve adaptations in the discharge rate of MUs (Van Cutsem et al. J. Physiol. 513:295-305,1998). Whereas numerous studies have found that the contractile properties of a muscle can change with physical training, there are few reports on the accompanying neural adaptations. Therefore, two groups of 5 subjects trained for 12 weeks by performing 10 sets of 10 fast contractions of the ankle dorsiflexor muscles 5 times a week. One group trained with isometric contractions whereas the second group trained by moving a load that was 1/3 of the subject’s MVC force. Both training methods were found to increase the rate of tension development of an isometric contraction but the dynamic training was more effective than isometric training. After training, the instantaneous discharge rate of MUs in tibialis anterior was increased at the beginning of the fast isometric contraction (first 3 interspike intervals). In addition, the percentage of MUs that discharged with fast doublets (< 5 ms) was increased substantially (33% vs 5% in control condition) with dynamic training. Although these changes were also observed after isometric training, they were more substantial after dynamic training, which explained the smaller increase in the rate of tension development after isometric training condition.

These findings emphasize the adaptability of the discharge rate of human MUs in response to short- and long-term contractile activity. In one example (muscle fatigue), the modulation of discharge rate varied with recruitment threshold, whereas in the other example (training), the adaptation was exhibited by all MUs.


Transient change in EMG activity after immobilization

John G. Semmler, Devin V. Kutzscher and Roger M. Enoka

University of Colorado, Boulder, USA

This presentation was based on our previous unexpected observation that 4 weeks of limb immobilization caused an intensity-dependent effect on fatigability, where an increase in the endurance time of the elbow flexor muscles was observed for low-force contractions (Yue et al. Exp. Physiol. 82: 567-592, 1997). The increased endurance appeared to be caused by an immobilization-induced improvement in the efficacy of excitation-contraction coupling. The purpose of the study was to identify the mechanisms responsible for the increased endurance time for low-force contractions after 4 weeks of elbow joint immobilization.

Twelve subjects (6 men and 6 women) participated in a protocol that required immobilization of the elbow joint in a fiberglass cast for 4 weeks, and 4 subjects (1 woman and 3 men) acted as controls. Measurements of elbow flexor strength, contractile properties, and fatigability were performed before and after 4 weeks of limb immobilization, and after 4 weeks of recovery. For these measurements, the task was to exert an upward-directed force with the wrist by activating the elbow flexor muscles. Muscle activity during the fatiguing contraction was measured with surface (biceps brachii, brachioradialis and triceps brachii) and intramuscular (brachialis) electrodes. Muscle fatigue was assessed as the duration that individuals could sustain an isometric contraction with the elbow flexor muscles at a force that was 15% of the maximum contraction force.

The average ( SD) endurance time when the isometric contraction was sustained at 15% of maximum was 897 416 s before immobilization, 2035 1418 s immediately after immobilization, and 1136 527 s after 4 weeks of recovery for the subjects in the immobilization group. These endurance times were not statistically different due to the large variability between subjects. After participation in the 4-week intervention that reduced the activity of the elbow flexor muscles, some subjects experienced an unusual pattern of muscle activity during the fatiguing contraction (Semmler et al. J. Neurophysiol. 82: 3590-3593, 1999). The unusual pattern of muscle activity comprised intermittent rather than continuous EMG activity and no increase in the amplitude of the EMG throughout the fatiguing contraction. In those individuals who exhibited this behavior, the novel pattern of muscle activity was only present immediately after the 4 weeks of limb immobilization and not prior to the intervention or after 4 weeks of recovery. Furthermore, the intermittent EMG activity was present in all the women who participated in the study, but only one man, and it was associated with a substantial increase (~220%) in the endurance time of the fatiguing contraction. No session differences were observed in the force-frequency relationship of the biceps brachii muscle in these subjects, suggesting that changes in the contractile characteristics of the biceps brachii muscle were not responsible for the large increase in endurance time (Semmler et al. Muscle Nerve 23: in press, 2000).

The absence of a progressive enhancement of EMG during the fatiguing contraction after immobilization suggests that the post-immobilization increase in endurance time involved neither the recruitment of additional motor units nor a gradual increase in the discharge rate of the active motor units. Rather, the EMG activity appeared to be comprised of bursts of motor unit activity, consistent with motor unit rotation. The factors that could account for these findings include muscle strength, the level of activity during immobilization, and central modulation of motor neuron activity.


Motor unit firing patterns during submaximal fatiguing contractions

S. Jayne Garland

University of Western Ontario, London, Canada

Firing rate, in the majority of motor units, has been shown to decrease during low-force fatiguing isometric contractions of the elbow flexor and extensor muscles (Garland et al 1994; Garland et al. 1997). However, when the isometric contraction is interrupted by arm movement, the minority of motor units display a decline in firing rate during fatigue (Miller et al. 1996; Griffin et al. 1998). We speculated that input from muscle spindle afferents during the movement would facilitate the motoneurone pool. This facilitation would offset any inhibition or late-adaptation process that may be causing the firing rates to decline in isometric conditions. To test this hypothesis, muscle vibration was applied to the distal part of triceps brachii muscle for 2 s every 10 s throughout a sustained isometric fatiguing contraction of the elbow extensors. Motor unit firing rate in the lateral head of triceps brachii muscle was recorded using a fine-wire subcutaneous electrode. Firing rates were determined during both the 8 s holding phases and during the 2 s vibration periods. In a second set of experiments, subjects performed a sustained isometric contraction for 2 minutes first and then intermittent muscle vibration was applied for the remainder of the fatiguing contraction. Application of intermittent vibration resulted in few motor units demonstrating a decline in firing rate during both the vibration and the holding phases of the fatiguing isometric contraction. Furthermore, the decline in firing rate that occurred in the first 2 minutes of the sustained isometric contraction in the second set of experiments could be reversed with the application of intermittent vibration. Motor units were affected by the vibration in a similar way. That is, the relationship between the initial firing rate of the motor unit and the change in firing rate with fatigue (without vibration) shifted upwards with the same slope (with vibration). To conclude, vibration or movement superimposed on a fatiguing isometric contraction resulted in a similar change in motor unit discharge rate. This suggests that the decrease of motor unit discharge rate in isometric fatiguing contractions could be attributed, at least in part, to declining spindle input.



Effects of aging on motor unit control properties

Zeynep Erim

Boston University, Boston, USA

The effects of aging on the motor unit control properties in the First Dorsal Interosseous muscle of the hand and the Vastus Lateralis muscle of the leg was discussed. Intramuscular EMG was recorded from male subjects during voluntary isometric abduction of the index finger and knee extension. EMG signals detected via the quadrifillar needle or fine-wire electrodes were processed using the Precision Decomposition Technique to identify firing times of individual motor units.

Data from the FDI reveal three main findings: 1) The highly correlated common fluctuations in the firing rates of concurrently active motor units in the young are not as correlated in the elderly, i.e., common drive is decreased in the elderly; 2) The phenomenon of onion skin observed in the young whereby motor units recruited earlier during increasing force have a higher average firing rate than motor units recruited later which have a lower firing rate is not observed in the elderly in whom the firing rate plots of motor units "crossover".; 3) Motor unit firing rates which show decreases in the young during short constant-force contractions as a result of the potentiation in twitch forces, do not display any decreases in the elderly, suggesting a decrease in the capacity of aged muscle fibers to potentiate. In addition to these novel findings, our results provide support for two previous observations: 1) Depression of the firing rate of motor units in the elderly was observed as compared to the young; 2) There was a shift in the recruitment thresholds toward lower values suggesting a fiber type change toward Type I fibers. Taken together, these observations suggest an altered control behavior of motor units in the elderly as compared to the young.

Our preliminary results suggest that there are age-induced changes in the control of motor units in the VL muscle similar to those observed in the FDI; specifically, the high correlation among motor unit firing rates of concurrently active motor units is diminished in the elderly, with motor units frequently displaying unrelated trends in firing rates. It remains to be determined whether . the phenomena of onion skin and progressively decreasing firing rates observed in the young hold true in the aged VL muscle.


Motor unit behavior following spinal cord injury

Inge Zijdewind, Lisa Griffin, and Christine Thomas

University of Miami, Miami USA

Skeletal muscles are often paralyzed completely or partially by injury to the human spinal cord. Here we present three aspects of the motor unit activity we have recorded in these muscles. For example, there is an increase in the incidence of doublets during both weak and strong contractions in triceps brachii muscles partially paralyzed by cervical spinal cord injury. This change may reflect alterations in the intrinsic properties of the motoneurons, their dendrites or dendritic growth following the death of some motoneurons. Together these factors may enhance the size and number of delayed depolarizations occurring in motoneurons, thereby increasing the incidence of doublets.

Second, all motoneurons in a pool can die after spinal cord injury inducing complete muscle denervation. Our strategy for neuron replacement in adult rats involves severance of the tibial nerve and transplantation of embryonic day 14-15 ventral spinal cord cells into the distal stump. Muscle reinnervation from the transplanted cells was assessed physiologically 10 weeks later. Many reinnervated motor units showed regular tonic firing or produced a few potentials in the absence of any obvious inputs. Each motor unit tended to fire at a particular frequency, but the range for different units varied widely (2 — 57 Hz). The onset and offset of unit firing was abrupt. A unit could change its firing behavior without any apparent influence on another active unit. Similar activity in the same motor units could also be evoked by a brief duration (10 m s) electrical stimulus.

In human thenar muscles paralyzed by spinal cord injury, many motor units also displayed tonic or sporadic activity without any apparent inputs to their motoneurons. This ongoing motor unit activity could start and stop abruptly or in a slow graded way. These changes in firing behavior often occurred independently of the firing of already active motor units. The mechanism behind spontaneous motor unit activity in paralyzed human muscles and reinnervated rat muscles may be intrinsic to the motoneuron. Other explanations are also possible.


Evidence for activation of Ipics from firing behaviour of motor unit pairs in uninjured and spinal cord injured human subjects

Monica Gorassini, David Bennett, Jaynie Yang and Philip Harvey

University of Alberta, Edmonton, Canada

The potential contribution of Ipics (re plateau potentials) in recruiting and sustaining the discharge of human motoneurons during voluntary contractions was presented. Pairs of human motor units were recorded from either the tibialis anterior or soleus muscle during low-level (20-30% MVC), triangular isometric torque contractions. The firing rate of the lower-threshold unit of the pair (control unit) was used as a measure of the effective synaptic excitation or drive to the motoneuron pool and, more specifically, to the higher-threshold motor unit of the pair (test unit; for rationale, see following abstract). The control unit rate was compared at recruitment and de-recruitment of the test unit to determine if the estimated synaptic drive needed to recruit a motor unit was less than the amount needed to sustain firing.

After test unit recruitment, the firing rate of the control unit could be decreased significantly (by » 40% on average) before the test unit was de-recruited during the falling phase of the contraction. This decrease represents a » 40% reduction in the estimated synaptic drive needed to maintain firing of a motor unit, compared to that needed to recruit the unit initially. In all motor units, the firing rate at recruitment was significantly higher than the minimum rate obtained at de-recruitment (by 48% on average), either during gradual volitional contractions or in response to slow, sinusoidal muscle stretch. These results are consistent with the behaviour of decerebrate cat motoneurons where activation of neuromodulatory-dependent Ipics at recruitment boost the initial firing rate of the motoneuron and help sustain firing during contractions. In addition, we also observed time-dependent facilitation of motor unit excitability that had a similar time-course (< 4-6 s) to that described for warm-up or wind-up of Ipics in motoneurons of reduced animal preparations (Russo & Hounsgaard Neurosci. 61: 191, 1994, Bennett et al. J. Neurophyisol. 80: 2038, 1998).

Motor unit activity was also recorded in subjects following chronic spinal cord injury (i.e., in subjects whose descending neuromodulatory inputs are possibly disrupted), in order to examine the potential role of Ipics in sustaining unit discharge during involuntary muscle spasms. During low levels of background activity (as monitored by the firing rate of a tonically-discharging control unit), vibration of the homonymous muscle produced brief (< 1 s) or no self-sustained unit activity in both functionally complete and incomplete subjects. In contrast, when the dorsum of the foot was vibrated or when a cold stimulus was paced on the skin, involuntary muscle spasms were readily induced and the firing rate of the control units were able to increase to 10-30 Hz. In addition, these diffuse "cutaneous" stimuli recruited new units that were able to fire in a self-sustained manner, i.e., they continued to fire at estimated levels of synaptic input that were much lower that the level needed to recruit the unit initially (by 60-80%). The specific contribution of cutaneous inputs in facilitating self-sustained unit activity (as opposed to attainment of critical levels of synaptic drive) is being pursued. Finally, in functionally motor-complete subjects, prolonged, involuntary unit firing was possible in a small proportion of units following the cessation of a muscle spasm (i.e., during little or no surface-EMG activity). The firing profiles of these units were unique compared to non-injured controls in that the spike-to-spike firing variability was extremely low (» 1%) and units pairs exhibited dissociated rate modulation. This type of firing behaviour may be produced by activation of Ipics under little or no extrinsic synaptic drive where the kinetics of the channels responsible for such steady, prolonged firing (e.g., persistent L-type Ca++ and Na+ channels) may have changed in response to the chronic dennervation of descending inputs.


Plateaus and spasticity in chronic sacral spinal rats

David Bennett, Monica Gorassini, Yunru Li, and Leo Sanelli

University of Alberta, Edmonton, Canada

We presented evidence that plateau potentials in motoneurons contribute substantially to spasticity after chronic spinal cord injury. The sacral spinal cord of adult rats was transected and this led to the gradual development of exaggerated, long-lasting, low threshold reflexes associated with a general spasticity syndrome in the tail (Bennett et al. J. Neurotrauma, 1998). Conveniently, this long-lasting spastic reflex could also be seen when the whole sacrocaudal spinal cord of the chronic spinal rat (> month post injury) was acutely isolated and maintained in vitro (Bennett et al. Soc. Neuroscience Abst., 1999). That is, in vitro there were exaggerated ventral root responses to brief dorsal root stimuli, which were blocked by the NMDA antagonist APV. Intracellular recordings were made from motoneurons in acute and chronic spinal rats in normal CSF in vitro.  Most cells in spastic chronic spinal rats exhibited sustained depolarization and firing (plateaus) in response to brief intracellular activation, whereas acute spinal rats (not spastic) did not exhibit plateaus. The long-lasting reflexes with brief dorsal root stimulation seen in chronic spinal rats were significantly reduced in duration by hyperpolarizing the motoneuron (from 5 to 1 sec), indicating that a plateau potential intrinsic to the motoneuron was largely responsible for these long-lasting spastic reflexes. However, even at hyperpolarized levels there was an EPSP that outlasted the stimuli by 1 sec, suggesting that there is a moderately long synaptic event, probably NMDA mediated, that normally triggers a plateau.rats (w/o plateaus), suggesting that the plateau was not facilitated by a reduction in the K+ currents associated with the AHP, in contrast to how 5-HT is thought to facilitate plateaus. Thus we suggest that the plateaus occurred in chronic spinal rats because of a direct facilitation or upregulation of intrinsic L-type calcium channels, though this hypothesis needs to be tested pharmacologically. 

In chronic spinal rats when the plateau was activated during slowly increasing current ramps it was usually activated just before or at least within a second of the initiation of firing, and this contributed to boosting the initial firing rate achieved at recruitment. Following this, the firing rate usually increased linearly with current, even though a plateau was present. However, the slope of the F-I relationship (5Hz/nA; I/O gain) was about half that in acute spinal rats without plateaus, probably because of the increased conductance provided by the IPIC (plateau). Firing rate adaptation was usually small with the slow, low amplitude activations used. Because of the relatively linear F-I relation, the firing rate is a good indicator of the input the neuron. Thus, assuming that there is a common input to a given motoneuron pool, then the firing rate of one low threshold motoneuron (control cell) can be used to indicate the input to a second higher threshold cell (test), if they are recorded simultaneously. We have thus recorded pairs of motor units in the tail of awake spastic chronic spinal rats, to verify our observations of plateau potentials seen in vitro. When graded contractions were evoked by skin stimulation the low threshold control unit could be made to fire with an approximately triangular ramp profile, mimicking the injected current described above during intracellular recording, and providing a measure of the effective synaptic current input to the cells of the motoneuron pool.  During this ramp the test unit was recruited when the control unit had moderately high rates (10-20 Hz), but after recruitment its firing was sustained even when the control unit rate (effective synaptic input current) was reduced by 5- 10 Hz. This suggests that the test unit had a plateau potential in its motoneuron that produced self-sustained firing. The amplitude of the plateau could be estimated by dividing the drop in control unit rate at de-recruitment by the slope of the F-I relation for these cells (5Hz/5Hz/nA = 1 nA), which agrees nicely with IPIC found with intracellular recording. In summary, evidence from both intracellular recording in vitro and extracellular recording in awake rats indicates that plateau potentials occur after chronic spinal cord injury and contribute to spasticity.


Discussion Session

Peter Matthews (Oxford University, Oxford, UK)


Zev Rymer (Northwestern University, Chicago, USA)

Introduction by Matthews: The meeting came at an important transition point for understanding motoneurons. The plateau potential is now increasingly being accepted as the normal mode of function for the motoneuron, instead of an oddity occurring under extreme levels of serotonin or norepinephrine. The ability of higher centers to regulate the plateau potential is an important behavior that needs to be ‘sold’ to our neuroscience colleagues who have tended to neglect the motoneuron.

The first discussion point was whether the motoneuron was a single functional entity or whether sub-calculations could be performed in fractions of the dendritic tree. Overall, the classical or single functional entity view prevailed because the motoneuron has but a single output.

The issue of the definition of a doublet was considered, revealing little agreement. One important point may be that the duration of the doublet should be considered in relation to the average firing rate for the spike train. Perhaps definitions should be linked to mechanisms, because doublets produced by the delayed depolarization following the action potential are fundamentally different than doublets produced by synaptic noise.

The definitions of bistable, self-sustained firing, plateau potential, and persistent inward current (IPIC or just PIC) as given in the introduction were generally agreed upon. It was felt that the most appropriate term for electrophysiological use was "persistent inward current" but that "plateau potential" was also acceptable.

Several criteria for recognition of plateau potentials in human motor unit firing patterns were suggested, but it was generally agreed that the paired recording technique (which uses one unit to show that synaptic input is constant) was the most convincing method. Other criteria include the following: (1) Low threshold motor units in humans often exhibit steep phase of rate modulation just after recruitment followed by a transition to a low rate of increase once a firing rate of 10 to 20 Hz is attained. This "rate limiting" is similar to that seen in intracellular recordings of motoneurons during activation of a plateau potential; (2) Another effect of dendritic plateau potentials is activation/deactivation hysteresis — deactivation of both the persistent inward current and the plateau potential requires a much more hyperpolarized level than their activation. In human motor unit recordings, this may be reflected in the de-recruitment force being much lower than the recruitment force; (3) When a motoneuron is close to the threshold for activation of the plateau, repeated steps become more effective in turning on the full plateau. This ‘warm-up’ phenomenon can also be detected in human motor unit firing patterns.

The issue of ‘preferred’ firing rate was also brought up, it being argued that activation of the plateau potential brings the motoneuron to its preferred rate and that this preferred rate is low because the plateau potential is activated close to or below the threshold for recruitment of the cell. The relationship between preferred rate and rate limiting was not clear. It was acknowledged that cells could easily increase their firing rates past the preferred level — this is just normal rate modulation.

This led to a general discussion of the relation between firing rate and synaptic drive. Perhaps the most important point was that the dendritic plateau potential is activated more readily by synaptic current than by injected current, which enters the cell only at the soma. This efficacy of the synaptic current plays a major role in bringing the activation of the plateau down to the recruitment threshold of the cell (see above). In addition, this means that the input-output function of the cell has a much higher gain for synaptic current than injected current. The claim that injected current as a probe of motoneuron function was a ‘waste of time’ was considered to be an exaggeration; a more precise statement would be that injected current measures the contribution of the somatic compartment but that synaptic current is required to fully probe the contribution of the plateau potential in dendritic regions.

However, there remain significant differences between the electrical properties of motoneurons as measured in the decerebrate cat and the firing patterns in humans. One clear example is the tendency of the rates of early recruited units to stay above those of late recruited units. This tendency is not apparent in intracellular records from cat motoneurons, where strong plateau potentials in high threshold motoneurons generate rates exceeding those of low threshold units. However, in turtle motoneurons, the pattern from intracellular recordings resembles those from human subjects to a remarkable degree. Further work is obviously required both in animal preparations and in human subjects.

Rymer then summarized several points regarding the behavior of motor units during stroke or spinal cord. The role of the plateau potential in chronic spinal injury was thought to be very important, in light of the recent results from the chronic rat-tail injury model and certain behaviors from single motor units in spinal cord injured humans. These behaviors include a strong tendency for units to discharge steadily at low rates (~6 Hz) and to experience potent effects from cutaneous stimulation, which often produces spasms that generate relatively high firing rates (10-20 Hz) for several seconds. While these phenomena are consistent with the presence of plateau potentials, further work is clearly needed.