Organization of common synaptic drive to motoneurones during fictive locomotion in the spinal cat
1 Division of Neurophysiology, Department of Medical Physiology, The Panum Institute, University of Copenhagen, Denmark
2 Institute of Physical Exercise and Sports Science, University of Copenhagen, Nrre Allé 55, 2100 Copenhagen , Denmark
3 Bioengineering Unit, University of Strathclyde, Glasgow, G4 0NW, UK
4 The Department of Electronics, University of York, UK
, http://www.100md.com 5 Department of Physiology, University of Oslo (Domus Medica), Oslo, Norway
Abstract
The basic locomotor rhythm in the cat is generated by a neuronal network in the spinal cord. The exact organization of this network and its drive to the spinal motoneurones is unknown. The purpose of the present study was to use time (cumulant density) and frequency domain (coherence) analysis to examine the organization of the last order drive to motoneurones during fictive locomotion (evoked by application of nialamide and dihydroxyphenylalanine (DOPA)) in the spinal cat. In all cats, narrow central synchronization peaks (half-width < 3 ms) were observed in cumulants estimated between electroneurograms (ENGs) of close synergists, but not between nerves belonging to muscles acting on different joints or to antagonistic muscles. Coherence was not observed at frequencies above 100 Hz and was mainly observed between synergists. Intracellular recording was obtained from a population of 70 lumbar motoneurones. Significant short-term synchronization was observed between the individual intracellular recordings and the ENGs recorded from nerves of the same pool and of close synergists. Recordings from 34 pairs of motoneurones (10 pairs belonged to the same motor pool, 11 pairs to close synergists and 13 pairs to antagonistic pools) failed to reveal any short-lasting synchronization. These data demonstrate that short-term synchronization during fictive locomotion is relatively weak and is restricted to close synergists. In addition, coherence analysis failed to identify any specific rhythmic component in the locomotor drive that could be associated with this synchronization. These results resemble findings obtained during human treadmill walking and imply that the spinal interneurones participating in the generation of the locomotor rhythm are themselves weakly synchronized. The restricted synchronization within the locomotor drive to motoneuronal pools may be a feature of the locomotor generating networks that is related to the ability of these networks to produce highly adaptive patterns of muscle activity during locomotion.
, http://www.100md.com
Introduction
It is well documented that a spinal neuronal network (central pattern generator; CPG) is responsible for generating the basic locomotor rhythm in vertebrates (Grillner, 1985; Rossignol, 1996; Burke, 2001). The periodicity of the step cycle and the patterns of muscle activity are therefore determined by elements within the networks generating and distributing the locomotor drive to the motor pools of the limbs.
In conceptual models of the CPG the locomotor drive to coactive motor pools is generally assumed to involve a common presynaptic drive (Graham Brown, 1911; Grillner, 1985; Yakovenko et al. 2005). Common or shared synaptic drive to populations of motoneurones may increase the probability of synchronized firing within a short time interval (short-term synchrony). In the event of synchronized motor activity during locomotion a statistical analysis has the potential to reveal features on the strength, frequency content and distribution of the common synaptic drive arising from the CPG (Kirkwood & Sears, 1978; Kirkwood et al. 1982; Farmer et al. 1993).
, http://www.100md.com
In previous experiments on cat fictive locomotion (induced by stimulation of the mesencephalic locomotor region; MLR), it has been demonstrated that coherence over a broad frequency range (up to 300 Hz) exists between ENGs recorded from both ipsi- and contralateral muscle nerves (Hamm & McCurdy, 1995; Trank et al. 1998; Hamm et al. 1999, 2001). This would suggest that the network responsible for the pattern of muscle activity is highly divergent. However, a combined time and frequency domain analysis of EMG data from healthy and spinal cord-injured human subjects walking on a treadmill failed to demonstrate any coherence above 50 Hz, and in healthy subjects short-term synchrony was restricted to synergists acting on the same joint (Hansen et al. 2001, 2005; Halliday et al. 2003). While this discrepancy between animal and human findings might reflect differences in the organization of the networks producing human (bipedal) and cat (quadrupedal) walking, it is also conceivable that the differences reflect the methodology used to generate locomotor-like behaviour in the decerebrate cat. During ongoing stimulation of the MLR the ENGs are likely to be strongly coupled to the stimulation, thereby generating significant coherence between the ENGs.
, http://www.100md.com
In the present study we investigated the coupling that exists between ENG activities recorded during fictive locomotion induced by the application of nialamide and DOPA in the spinal cat. In contrast to the results reported during MLR-induced locomotion (Hamm & McCurdy, 1995; Trank et al. 1998; Hamm et al. 1999, 2001), we found little evidence of high frequency coherence. Importantly, short-term synchrony was weak and restricted to close synergists. These findings are not consistent with a strongly synchronized and widely distributed last order drive to spinal motoneurones during locomotion. The analysis conducted demonstrates that the pattern of short-term synchrony in motor output during locomotion in the acute spinal cat is remarkably similar to that observed in humans.
, 百拇医药
Methods
Preparation
Data were obtained from nine adult cats (1.9–3.5 kg). All surgery and experimental protocols were conducted according to the National Institutes of Health Guide for the Care and Use of Laboratory Animals (National Institutes of Health publication no. 86–23, revised 1985). The experiments were approved by the Animal Experiments Inspectorate in Denmark. After anaesthesia with halothane–nitrous oxide (2–3% halothane in 70% N2O and 30% O2), the animals were intubated, and cannulas were inserted in the carotid artery and jugular vein for blood pressure monitoring and administration of fluid and drugs. Atropine (0.1 mg kg–1, S.C.) and dexamethasone (1 mg kg–1, I.V.) were given at the beginning of the experiment while buffer solution (10% dextrose and 1.7% NaHCO3) was infused continuously (4.5 ml h–1). During anaesthesia the cats were anaemically decerebrated by ligating the basilar and both common carotid arteries (and their collaterals), a procedure that has been shown to produce a decerebration that involves all cortical tissue above the pons and the anterior part of the cerebellum (Pollock & Davis, 1930; Crone et al. 1988). The anaesthetic was removed 5–6 h after ligating the vessels, and the decerebration was verified to be clinically complete by the development of tonic extensor muscle tone (alpha rigidity) in the forelimbs, lack of spontaneous movements, and large non-reactive pupils (Crone et al. 1988). Following these procedures and tests, pancuronium bromide (0.6 mg h–1) was given to block neuromuscular transmission, and artificial respiration was initiated. At the end of the experiments the animals were killed by an overdose of pentobarbital.
, http://www.100md.com
Prior to decerebration the following nerves from the left hindlimb were cut and dissected for recording: quadriceps (Q), sartorius (Sart), semimembranosus (Sm), anterior biceps (AB), posterior biceps (PB), semitendinosus (St), medial gastrocnemius (MG), lateral gastrocnemius plus soleus (LGS), tibialis anterior (TA), flexor digitorum and hallucis longus (FDHL), plantaris (Pl) and extensor digitorum longus (EDL). The Q and Sart nerves were placed in cuff electrodes, and the other nerves were mounted on bipolar silver–silver chloride hook electrodes. From the right limb, gastrocnemii + soleus (GS) and TA or deep peroneal (DP = TA + EDL) nerves were dissected. To expose the lumbar spinal cord a laminectomy (L4–S1 spinal cord segments) was performed and the animals transferred to a rigid frame, which secured and stabilized the spinal column. A complete spinalization was performed at thoracic segment 10–11 following local application of lidocaine (lignocaine). The expired CO2 was maintained between 3.0 and 5.0% by artificial ventilation and the blood pressure maintained by infusion of Dextran when needed. The temperature of the animal was kept at 38°C by a servo-controlled heating system.
, http://www.100md.com
Fictive locomotion was evoked by I.V. administration of nialamide (50 mg kg–1; Sigma-Aldrich, St Louis, MO, USA) and DOPA (50–100 mg kg–1; Pfizer, New York, NY, USA; Grillner & Zangger, 1979). Naloxone (100 μg kg–1; Sigma-Aldrich) was given to enhance the locomotor activity if necessary (Pearson et al. 1992). In two cats in which locomotor activity induced by nialamide and DOPA had waned, application of noradrenaline directly to the lumbar spinal cord re-established sustained locomotor activity (Kiehn et al. 1992).
, http://www.100md.com
Intracellular recordings were obtained from single or pairs of antidromically identified motoneurones innervating the left hindlimb. Glass micropipettes (tip diameter, 1.2–2 μm; resistance, 2–4 M) filled with the lidocaine derivative QX-314 in potassium acetate (2 M) were used. QX-314 blocks sodium spikes and may, in addition, reduce calcium currents. Thus, QX-314 limits the extent to which sodium and calcium currents could affect postsynaptic potentials, thereby facilitating the study of presynaptic drive.
, 百拇医药
Data analysis
Electroneurograms (ENGs), AC-coupled cord dorsum potentials, and DC-coupled intracellular recordings were digitized at a rate of 10 kHz with software developed by the Winnipeg Spinal Cord Research Center to run under real time QNX. The ENGs were band-pass filtered (high pass: 1 Hz; low pass 5 kHz). The theoretical framework for analysis of the present data represents a development of that presented in Halliday et al. (1995) to include time dependency, and to deal with analysis based on variable length segments. Time dependency is incorporated using the concept of a sliding data window (Gabor, 1946), discussed below. First we review briefly the calculation of the finite Fourier transform of time series data, and the periodogram-based approach to second order spectral estimation used in Halliday et al. (1995). For a stationary stochastic time series, x(t), the finite Fourier transform of a segment of length T, denoted by dTx (), is defined as:
, http://www.100md.com
at frequency . In discrete form this can be written as
for sample values xt. The j are the Fourier frequencies defined as (j= 0, . . . , T/2). Here corresponds to the Nyquist frequency. The quantity dTx(j) can be readily calculated using an Fast Fourier Transformation (FFT) algorithm.
For a record consisting of L sections of length T, the second order spectrum between two time series x(t) and y(t) can be estimated as:
, http://www.100md.com
The hat symbol indicates a parameter estimate; the overbar indicates a complex conjugate. The autospectra, fxx() and fyy() require only the finite Fourier transform of segments from their respective individual time series. Coherence, phase and cumulant density (cross-covariance) estimates can be constructed from the cross-spectral and auto-spectral estimates using the methods described in Halliday et al. (1995). Construction of spectral estimates using the average over L successive non-overlapping segments as in eqn (3) assumes that the signals are stationary across the entire record.
, 百拇医药
Equation (3) represents a periodogram approach to spectral estimation (Schuster, 1898). The periodogram itself is not a reliable estimate of the spectrum; statistical stability is achieved through averaging across the L periodograms. This approach is based on the assumption that data records are stationary, an assumption which does not apply to the present data (see Fig. 1). A periodogram approach does, however, allow task dependency in neural data to be explored through the use of averages constructed over repeat trials, in conjunction with an assumption of local stationarity (see below). Each segment of data is selected from a different trial, and, in the general case, may contain a different number of data points. Periodogram estimates require the same Fourier frequencies, which can be achieved for unequal numbers of data points through the use of zero padding, where the data in each segment is extended by the addition of zeros.
, 百拇医药
Electroneurogram (ENG) was recorded from a number of motor nerves in the left hindlimb (A). Recordings were also made from the right hindlimb (not shown). In order to determine burst onset, burst offset, burst duration, cycle duration and the phasing of the activity in the different ENGs, all ENGs were first rectified and smoothed and then the onset and offset of each burst of activity was determined automatically by setting a threshold in the computer software. The cycle duration was determined from the onset of one burst of activity in the posterior biceps (PB) nerve until the onset of the next burst. St, semitendinosus; Sm, semimembranosus; AB, anterior biceps; LGS, lateral gastrocnemius plus soleus; MG, medial gastrocnemius; Pl, plantaris; FDHL, flexor digitorum and hallucis longus; TA, tibialis anterior; EDL, extensor digitorum longus. B, for each fictive step of the locomotor bout partly shown in A, the burst onset in St (black circles), TA (red circles), EDL (green triangles), LGS (yellow triangles) and MG (blue squares) was plotted relative to the locomotor cycle duration (which was calculated from the PB burst onsets).
, 百拇医药
We now use Tl to refer to the number of data points in the lth section of data, and S to refer to the length of the finite Fourier transform, where S max{Tl; l= 1, . . . , L}. The periodogram of the lth section of data is
where the Fourier frequencies are (j= 0, . . . , S/2). If necessary, zeros are appended to each data segment, so that the Fourier transform is performed on a sequence of length S. The spectral estimate,, constructed from l segments is
, 百拇医药
where wl are the weights for each section, and . The optimum weights, in the sense of minimizing the variance of the spectral estimate, are given by (Bloomfield, 2000):
Implicit in the averaging of periodograms across segments of data from repeat trials is the assumption of local stationarity across trials. Here the qualifier ‘local’ is used to distinguish the assumption of stationarity within a short segment of data from the non-stationarity highlighted in Fig. 1 (see Results), which occurs on a longer time scale. The variance of the spectral estimate eqn (5) with weights given by eqn (6) can be approximated by (Bloomfield, 2000):
, 百拇医药
One advantage of the periodogram-based approach outlined above is that the setting of confidence limits generally involves simple expressions, often in conjunction with the use of a suitable transformation to stabilize the variance (Jenkins & Watts, 1968; Bloomfield, 2000). One parameter which plays a key role in the setting of confidence limits for periodogram estimates for stationary records based on equal length disjoint sections is the number of sections, L (Halliday et al. 1995). To extend this approach to the present analysis we define a quantity, L', as
, 百拇医药
where wl are the weights used to construct the periodogram estimate. We refer to L' as the number of equivalent sections in the periodogram estimate. This will not necessarily be an integer for an analysis based on unequal length segments. This quantity allows confidence limits for spectral parameters to be set using the expressions given in Halliday et al. (1995). For example, the magnitude of a 95% confidence limit on a plot of log10 of an auto-spectral estimate is given by ± 0.851/L', and the upper 95% confidence limit for coherence estimates, based on the assumption of independence, is given by the quantity 1 – (0.05)1/(L'–1). Using eqns (5), (6) and (8), time/frequency multivariate parameters can be estimated and confidence limits set using the methods described in (Halliday et al. 1995).
, 百拇医药
ENG records were full wave rectified prior to analysis, intracellular records were not rectified, but an additional pre-processing in the form of a 4 Hz high pass filter was applied to intracellular records prior to analysis. This was necessary to remove the large low frequency trend induced by the locomotor drive potential (see Fig. 5). Three different methods were used to analyse the correlation between pairwise ENG and intracellular recordings (ENG to ENG, motoneurone to ENG and motoneurone to motoneurone). In the first method, two channels were selected as reference channels, one for extensors and one for flexors. Triggers were placed at the start of each burst of activity in these reference channels. The timing of individual trigger points was automatically performed in the QNX software, all trigger points were subsequently verified by visual inspection. These trigger points were used to calculate spike-triggered averages of every neurogram record, from which activity maps of the ENG were constructed (see Fig. 1). The activity maps indicate the relative timing of ENG bursts within each locomotor cycle (in an average sense) with reference to the trigger times. From these activity maps, periods of coactivity between pairs of ENG and intracellular motoneuronal records were determined by visual inspection of the spike-triggered averages. These periods of coactivity were used to obtain the offset and length of data segment from each trial to include in the spectral analysis. In this method the offset and segment length were the same for each trial. Estimates of auto- and cross-spectra were then formed according to eqn (5), by averaging across segments taken from each trial. The results in Figs 2, 3, 5 and 6 were obtained using this approach.
, 百拇医药
Intracellular recording was obtained simultaneously from two individual PB motoneurones. A, intracellular recording from the two motoneurones together with the ENG from the PB nerve at three different time scales. B, the coherence (upper graph) and cumulant density function (two lower graphs at different time scales) between the two intracellular recordings.
Coherence (upper row of graphs in A and B) and the cumulant density function (lower two rows in A and B) was calculated for the following pairs of ENG : PB and St (A; 1st column), PB and TA (A; 2nd column), TA and EDL (A; 3rd column), MG and LGS (B; 1st column), MG and plantaris (Pl) (B; 2nd column) and MG and TA (B; 3rd column). The cumulant was plotted with a long (middle row) and short time base (lower row). The horizontal lines in the middle row of graphs designate the confidence limits.
, 百拇医药
Intracellular measurements were obtained from an MG motoneurone together with MG, LGS, FDHL, Sm and TA ENGs. The cumulant density estimate between the motoneurone and the ENG activities (A, C, E, G, I) was calculated for segments of the ENG recording where a burst was observed. In the computer software a level was set to generate triggers corresponding to individual ENG spikes (roughly the 30% largest spikes). An average of the intracellular recording for a time period of 100 ms around the time of the trigger events was obtained (B: MG (37 505 triggers), D: LGS (40 642 triggers), F: FDHL (75 000 triggers), H: Sm (52 679 triggers) and J: TA (18 805 triggers)).
, 百拇医药
The second method of analysis was used to characterize changes in the correlation structure over different phases of the locomotor cycle, and was based on the same sequences of triggers as in the first method, above. A number of different spectral estimates were constructed, using a range of different offset values from each trigger, and a fixed segment length of 200 ms of data from each trial (see Fig. 4A). This is similar to the analysis carried out in Halliday et al. (2003) on human locomotion data. A segment length of 200 ms was used; this defines the frequency resolution of spectral estimates to be around 5 Hz. The results in Fig. 4 were obtained using this approach.
, http://www.100md.com
The data are from the same cat as used for Fig. 3. Coherence (B) and cumulant density estimates (C, long time base; D, short time base) were calculated for 200 ms segments starting at different times (400 and 200 ms before onset, at onset and 200, 400 and 600 ms after burst onset) in relation to the onset of PB burst activity.
The third method of analysis was introduced to investigate the effects of using a fixed length data segment from each burst (method 1) when the actual duration of bursts is not constant. In this method two triggers were used to mark each burst, one at the start and one at the end. These pairs of triggers were then used to select a different segment length from each trial, such that only data samples lying between the start and end trigger points for each burst were included from each segment in eqn (5). In this method the wl in eqn (6) will vary according to the relative length of each segment. No results from this method of analysis are illustrated; however, the Results section includes a description of results obtained using this method of analysis.
, 百拇医药
The correlation between rectified ENG signals and/or high pass filtered intracellular signals was assessed in the time and frequency domains using the functions described below. In the frequency domain, correlation was assessed through coherence functions (Brillinger, 1981; Rosenberg et al. 1989; Halliday et al. 1995). The coherence function between two signals, x(t) and y(t), is defined at frequency as:
Coherence functions provide normative measures of linear association on a scale from 0 to 1. For the present data, the coherence provides a measure, at each Fourier frequency , of the fraction of the activity in one signal which can be predicted by the activity in the second signal.
, 百拇医药
In the time domain, estimates of the cumulant density function were used to characterize the correlation between pairs of signals. The cumulant density function, denoted by qxy(u), is defined as the inverse Fourier transform of the cross-spectrum
)
For two uncorrelated signals the cumulant has an expected value of zero; deviations from this indicate a correlation between the two signals at a particular time lag, u. In the present experiments, which are exploring aspects of the common drive to motoneurone pools during locomotion, the dominant feature in cumulant density estimates is the central peak around zero lag, reflecting the presence of common synaptic input. Rhythmic inputs will induce symmetrical oscillatory components in the cumulant (Perkel et al. 1967), the frequency and strength of these components can be quantified from the corresponding coherence estimates. Cumulant density functions are analogous to cross-correlation functions often used to quantify spike train data, and have a similar interpretation (Halliday et al. 1995).
, http://www.100md.com
The interpretation of the above time and frequency domain measures of correlation does depend on which of the three specific methods of spectral estimation was used. Correlation analysis based on method 1 can be interpreted as providing an indication of correlation between the two signals at periods when the two signals are coactive. This assumes that the signals have local stationarity, i.e. that the correlation structure does not change markedly during the burst. The third method of spectral analysis (using both start and end triggers for each burst) can be viewed in a similar manner. The second method of spectral analysis introduces a greater degree of time dependence into the correlation estimates, which should now be interpreted as a measure of the correlation within a narrow time window (200 ms in this case) that is offset from the trigger point marking the start of each burst.
, 百拇医药
The present analysis is concerned with identification of the characteristics of common synaptic inputs to motoneuronal pools during fictive locomotion. Our interpretation of coherence and cumulant density estimates is similar to that used by Farmer et al. (1993), who used coherence estimates to quantify the strength and frequency of rhythmic synaptic inputs to a motoneurone pool during isometric contractions. A distinct band of coherence in a particular frequency range is taken as a marker for common rhythmic inputs to the motoneurones over this frequency range. The magnitude of coherence is used as an indication of the strength of the common input. A broad band of monotonically decreasing coherence is interpreted as an indication of common synaptic inputs which do not contain any specific rhythmic components. A theoretical perspective and further discussion of the common input model is given in Rosenberg et al. (1998). Similar inferences can often be drawn from time domain measures of association, thus time and frequency analyses should be considered as complementary approaches (see also Myers et al. 2004).
, http://www.100md.com
Cumulant density estimates provide a generalized measure of correlation, and as such can be viewed as statistical parameters. The intracellular records were further analysed by performing an average with respect to a multi-unit trigger signal derived from the homonymous nerve signal. This parameter has been termed the average common excitatory (ACE) potential (Kirkwood & Sears, 1978). Construction of ACE estimates allows direct comparison between the present data and those derived from respiratory CPG studies (Kirkwood & Sears, 1978).
, 百拇医药
Results
ENG data
Fictive locomotion was induced by the combination of nialamide and DOPA in nine cats. In five cats the locomotion consisted of slow alternating flexor and extensor bursts with phase-related modulation in the contralateral ENGs, consistent with a left–right reciprocating extensor–flexor locomotor rhythm. This pattern of motor activity persisted for several hours in these preparations. In the four remaining cats rhythmic bursts in flexor ENGs were observed, but without a clear or consistent reciprocal bursting in extensor ENGs. Note, that rhythmic activity in extensors was never observed in the absence of rhythmic flexor activity. This is a common feature of the fictive locomotor pattern in nialamide and DOPA-treated spinal cats (Grillner & Zangger, 1979).
, 百拇医药
An example of the ENG activity associated with fictive locomotion induced by nialamide and DOPA in the spinal cat is illustrated in Fig. 1A. In the example shown, rhythmic alternating extensor and flexor activity is apparent in most of the ENG records. The duration of a single fictive step ranged from 2 to 4 s (mean: 2.6 ± 0.6 s), producing a step frequency of 0.4 Hz (114 steps in 300 s). The activity of the different flexor nerves was largely in-phase and the mean burst duration was 0.9 ± 0.012 s. The extensor nerves displayed in-phase activity that was reciprocally organized with respect to the flexor burst activity. The mean burst duration in extensors was 1.6 ± 0.03 s (for LGS). Figure 1B shows the onset of the ENG activity in both flexors (St, TA and EDL) and extensors (LGS and MG) relative to the onset of PB activity. During this particular bout of locomotion, no activity was recorded from LGS and MG during several cycles (30 and 41 cycles out of 115, respectively) while the activity in St, TA and EDL failed in only a few cycles (6–8 cycles).
, http://www.100md.com
Figure 2 illustrates a time and frequency analysis of paired ENG recordings where epochs of coactivity were extracted and used to calculate the coherence (uppermost graphs in Fig. 2A and B) and the cumulant density functions (middle and lowermost graphs in Fig. 2A and B), respectively.
Significant coherence was seen in all the pairs of ENGs (synergists and antagonists) but only at frequencies below 50 Hz. Significant differences were observed in the cumulant density estimates where a broad peak of synchronization was observed for synergist ENG pairs and a trough for antagonistic ENG (e.g. TA and MG) pairs, reflecting the phase relationships that exist between these muscle groups. The broad peak is considered to reflect features of the envelope of the ENG bursts and its duration is similar to the duration of the individual ENG bursts.
, 百拇医药
For synergist ENGs the cumulant displays an additional feature, which is recognizable as a narrow (half-width < 3 ms) central peak (Fig. 2A and B, bottom row of graphs showing the cumulant on an expanded scale). This central peak is taken as a positive marker for the presence of short-term synchronization between ENG recordings (Vaughan & Kirkwood, 1997). Such peaks were only observed for close synergists such as PB and St, TA and EDL, and MG and LGS. In no animal was short-term synchronization identifiable between synergist nerves acting on different joints (PB and TA or AB and LGS) or between antagonist (TA and LGS) muscle nerves (Fig. 2).
, 百拇医药
In all nine cats the coherence estimates were characterized by the absence of any significant coherence for frequencies above 100 Hz. All analyses showed a large coherence at low frequencies monotonically decreasing towards 20–100 Hz. This is considered in part to echo the co-modulation of the ENG records by the locomotor drive as expressed through the activity envelope of the ENG epochs, but it may also reflect an asynchronous activation arising from the CPG. It is important to note that this low frequency coherence is not equivalent to the frequency of the locomotor rhythm as this frequency component would only become manifest if the coherence had been estimated from non-segmented data sets. Although significant coherence was observed at the same frequencies for synergists working at the same joint as for synergists working at different joints, the magnitude of the coherence was always smaller in the latter cases. In all cases the coherence lacked any evidence of discrete spectral components.
, 百拇医药
For all nine cats, short-term synchronization between flexor motor pools could be identified for paired PB and St and paired TA and EDL ENG recordings, but not for synergists working at different joints (PB and TA, PB and EDL, St and TA, St and EDL). Similarly, in the five cats displaying consistent extensor ENG bursting, short-term synchrony was seen for synergists working on the same joint, but not for synergists working on different joints. In these five cats, no short-term synchrony was observed between antagonistic flexor and extensor ENG pairs. Similarly, no short-term synchronization was observed between bilateral pairs of coactive extensors and flexors in any animal.
, 百拇医药
High frequency oscillations during noradrenaline-induced fictive locomotion
In the two animals that received noradrenaline (NA) to boost the fading DOPA locomotion, the coherence and cumulant estimates revealed an additional feature during episodes of locomotor activity. Sample locomotor data from one of these cats is shown in Fig. 3A. Comparison with Fig. 1A reveals a difference in the temporal characteristics of the burst activity. Following NA, the cycle frequency was approximately 0.5 Hz (155 steps in 300 s; cycle duration: 1.9 ± 0.6 s), almost twice as high as the cycle frequency for the cat used for Fig. 1. Furthermore, the magnitude of the burst activity was much more variable following administration of NA (compare Fig. 3A and Fig. 1A) and short-lasting bursts of activity, which appear to be unrelated to the ongoing locomotor activity occurred simultaneously in both extensors and flexors (marked by arrows).
, 百拇医药
In A the burst activity in the recorded motor nerves is shown. The arrows indicate bursts which occurred synchronously in extensor and flexor nerves and which appeared to be separate from the locomotor burst activity. Coherence (upper row of graphs in B and C) and the cumulant density function (lower two rows in B and C) was calculated for the following pairs of ENGs: PB and St (B; 1st column), PB and TA (B; 2nd column), TA and EDL (B; 3rd column), MG and LGS (C; 1st column), LGS and FDL (C; 2nd column) and LGS and TA (C; 3rd column). The cumulant was plotted with a long (middle row) and short time base (lower row).
, 百拇医药
In agreement with the data from Fig. 2, short-term synchrony was only present in this preparation for close synergists and absent for synergists working at different joints or for antagonist muscle pairs. However, in contrast to the findings of the coherence analysis reported in Fig. 2 an additional high frequency band of coherence peaking between 100 and 300 Hz was seen for pairs of close synergists (PB : St and TA : EDL). This high frequency coupling was not observed between synergists working at different joints (PB and TA see Fig. 2B). Findings consistent with those shown in Fig. 3 were also seen in the second cat in which NA was used to evoked rhythmic activity.
, 百拇医药
Modulation of coupling during the locomotor cycle
To determine the coupling pattern within the locomotor cycle an analysis based on short (200 ms) fixed data segments and variable (ENG burst duration) length segments (see Methods) was used. The advantage of this approach is that it minimizes potential effects of the cyclic modulation of the ENG during each burst. Figure 4 illustrates data from one of the cats where coherence was observed above 100 Hz (after application of NA), but apart from this essentially similar findings were observed in all nine cats. Figure 4A shows the ENG recordings from the PB and St motor nerves and Fig. 4B–D shows the coherence and cumulant density estimates for the different 200 ms segments starting at 400 ms prior to the onset of the burst until 600 ms after burst onset. A low frequency coherence component is present in all sections of the segmented data, but is greatest in segments sampled around the onset of the ENG bursts. During that period, an additional 100–300 Hz band of coherence also became apparent in this experiment. Note that the low frequency coupling is accompanied by short-term synchronization within the early part of the burst and that clear secondary features at short lags (± 4 ms) are observed in the cumulant for the segments where 100–300 Hz coherence appeared.
, http://www.100md.com
Due to variability in burst intervals and durations (see Fig. 1B) a further analysis based on the use of triggers to identify the start and end of individual bursts was conducted. This allowed the analysis to be restricted only to ENG data occurring during bursts. The results from this analysis closely parallel those obtained by using the fixed average segment length. In both types of analyses, the presence of the large low frequency coherence (e.g. the centre portion of the burst in Fig. 4) suggests that the low frequency component is not simply an artefact reflecting the frequency of the rhythmic modulation of the burst activity in synergists, but is a signature of a common modulation of motoneuronal drive possibly derived from the CPG.
, 百拇医药
Intracellular data
Simultaneous intracellular recordings were obtained from 34 pairs of motoneurones during fictive locomotion in six of the cats. Ten of these pairs belonged to the same motor pool (7 extensor pairs, 3 flexor pairs), 5 belonged to two different close synergistic pools (2 extensor and 3 flexor pairs) and 8 belonged to synergists working at different joints (6 extensor and 2 flexor pairs). The remaining 11 pairs belonged to antagonistic motor pools.
, http://www.100md.com
Figure 5 demonstrates a representative example of the observations that were obtained. The recordings are from a pair of PB motoneurones. The action potentials in the two motoneurones were blocked by QX-314. As expected, the membrane potential of the two motoneurones is depolarized during the burst of activity in the PB motor nerve, reflecting the locomotor drive to the two motoneurones during the flexor phase of the locomotor cycle. As shown on expanded time scales, synaptic noise in the two motoneurones is quite different despite the similar time course of the locomotor drive potential. This is also reflected in the coherence and the cumulant density estimates, which did not reveal any statistically significant coupling of the synaptic activity other than that which could be related to the envelope of the depolarizing locomotor drive potential. For all pairs of intracellular recordings, coherence was only observed at very low frequency and there was no evidence of short-term synchrony from cumulant estimates. It is worth noting that when making paired intracellular recordings, the microelectrodes were placed 3–5 mm apart to avoid mechanical interference during tracking.
, http://www.100md.com
To investigate whether the synaptic activity in individual motoneurones was coupled to the burst activity in the nerves, coherence and cumulant densities were estimated from each intracellular recording (70 motoneurones) and the ENGs. Figure 6 shows data from a single MG motoneurone. The cumulant density function revealed significant coupling between the non-spiking motoneurone and the ENG activities from the MG and LGS nerves (Fig. 6A and C), but not from synergists working at different joints (FDHL and Sm; Fig. 6E and G) or antagonists (TA; Fig. 6I). Short-term synchrony between the intracellular recording and the corresponding motor nerve was observed in 50 of the 70 motoneurones sampled. Of these 50 cells short-term synchrony with close synergistic motor nerves was observed in 30 cases. No high frequency coherence was ever observed.
, 百拇医药
When the membrane potential recorded from the motoneurone was averaged using the ENG activity from the respective nerve as a trigger, an averaged common excitatory potential (ACE potential; Kirkwood & Sears, 1978) was observed for the MG and LGS nerves (Fig. 6B and D), but not for the other nerves (Fig. 6F, H and J). In 39 of the 50 motoneurones which showed short-term synchrony a similar ACE potential was found. The size of the potential varied from 100 μV to around 600 μV. The size of the short-term synchrony peak in the cumulant density function varied from 0.001 to 0.005, with a highly significant correlation to the size of the ACE potential (Spearman's rank; P < 0.001). An ACE potential was never observed without a significant short-term synchrony peak in the cumulant density function.
, 百拇医药
Discussion
It is well established that the rhythmic activities of synergistic as well as antagonistic motoneurones during fictive locomotion are functionally coupled (Grillner & Zangger, 1979; Schomburg & Steffens, 1995; cf. also Figs 1 and 3 of the present study). This reflects a common source of activation to the muscles acting on the different joints. A consequence of this organized burst-like motoneuronal behaviour is the appearance of a broad synchronization feature in cumulant density estimates that reflects the underlying temporal synergies between muscle groups (see Figs 1 and 2). The question we have addressed in the present study is to what extent a common synaptic drive from branches of last order neurones (Kirkwood & Sears, 1978) participate in forming the coupling of motoneuronal activities observed during fictive locomotion. In the time domain, this common synaptic drive (short-term synchronization) is identified by the presence of a narrow central peak with a half-width shorter than 3 ms (Vaughan & Kirkwood, 1997).
, http://www.100md.com
Short-term synchronization was observed between close synergists, but never for synergists working at different joints or for antagonists. A similar brief peak was also observed between intracellular recordings from single (non-spiking) motoneurones and their corresponding motor nerve (and in some cases also the nerves of their close synergists). This suggests that there is a difference in the organization of the synaptic drive from the spinal CPG to close synergists as compared to synergists working on different joints. This view is affirmed when considering data sets obtained from the two cats in which intrathecal noradrenaline was used to re-establish locomotor behaviour. In these cats short-term synchronization between close synergists was (following noradrenaline administration) accompanied by high frequency (200 Hz) coherence. High frequency coherence was never observed without short-term synchrony being present in the respective cumulant. The most likely mechanism for generating coherence between ENGs at such high frequencies is a reflection of the frequency content in the discharges of the last order neurones responsible for the short-term synchrony. In such cases the coherence does not reflect motoneuronal discharge frequencies, but reveals that a common high frequency modulation of output can be detected within the coupled motor pools (see Farmer et al. 1993). In nialamide- and DOPA-induced fictive locomotion short-term synchrony was apparent without accompanying high frequency coherence. We interpret this as an indication that during locomotion significant coupling of motoneuronal discharges by last order interneurones can occur without significant coupling in motoneuronal discharge frequencies even when the locomotor rhythm is itself stable. This situation will arise when the interneuronal drive from last order interneurones during different locomotor bursts does not display a distinct and robust spectral signature. Only in cases where a significant part of the last order interneuronal discharge contains reproducible spectral features, occurring within a specific frequency band, will motoneuronal coherence reflect the modulatory influence of the interneuronal drive.
, 百拇医药
Relation to MLR-induced fictive locomotion
In part, our data confirm the findings of Hamm and coworkers for fictive locomotion induced by MLR stimulation (Hamm & McCurdy, 1995; Trank et al. 1998; Hamm et al. 1999, 2001) in that significant coupling (in both the frequency and time domains) was observed for extensors and flexors acting on all limb segments as judged by the low frequency coherence and broad synchronization. However, the widespread broad band coherence reported in MLR preparations (Hamm & McCurdy, 1995; Trank et al. 1998; Hamm et al. 1999, 2001) was not seen in our data. In the majority of our data, coherence was restricted to frequencies well below 100 Hz and in the two cases where coherence was also seen at frequencies between 100 and 300 Hz, this high frequency coupling had a restricted organization that was linked to the distribution of short-term synchrony between close synergists. The combined time and frequency domain approach we have used therefore allows more precise descriptions to be made of the nature of the coupling process associated with the appearance of significant coherence between coactive motor pools. It seems to us a likely possibility that the stimulus train during MLR-evoked locomotion in the studies by Hamm and colleagues (Hamm & McCurdy, 1995; Trank et al. 1998; Hamm et al. 1999, 2001) may have induced widespread and high frequency coherence in all lumbar motor pools independently of the CPG activity. However, it should be noted that widespread high frequency coherence was also observed by Hamm & McCurdy (1995) during spontaneous locomotion and by Hamm et al. (2001) during fictive scratching – without any MLR stimulation. Other factors may thus also contribute to their findings.
, http://www.100md.com
Intracellular coupling
In the present study we did not see short-term synchronization between pairs of intracellular recordings, although all received a locomotor drive potential. To achieve stable paired motoneuronal recordings in this preparation the microelectrodes were inserted at some distance from each other. The lack of short-term synchronization in the sampled motoneurones may therefore be a consequence of the anatomical projections of the interneurones being highly localized. Thus, only motoneurones lying close to one another would demonstrate mutual short-term synchrony. This anatomical framework would also explain why short-term synchrony between intracellular recordings and the population ENG recordings can be observed. Related to this, the low level of synchrony across a motor pool would also suggest that the population of interneurones are themselves weakly coupled.
, 百拇医药
Interpretation in relation to the spinal CPG
The findings in the present study suggest that part of the output from the spinal rhythm generator is conveyed through a local segmental network of last order spinal interneurones with limited capacity for coupling motoneurones arising in different spinal segments. Previously, Grillner & Zangger (1979) noticed that flexor and extensors working at different joints appeared to be less tightly coupled than flexors and extensors working on the same joint. Our data extend this to include the distribution of short-term synchonization.
, http://www.100md.com
Importantly, our data indicate that the synergy of muscle activities during locomotion is not determined by the last order neurones. Rather the pattern of activity appears to be determined by a distributed input, to which the last order neurones studied here do not have sufficient divergence to define. Accordingly, we suggest that the last order neurones that generate the short-term synchonization revealed in this study have no specific role in determining the rate, timing or phasing of the locomotor bursts. However, these neurones could play an important role in setting the magnitude of the bursts in relation to functional demands. If this is the case, the group of interneurones may participate in reflexes that fail to entrain CPG activity, but lead to an alteration in the level of ongoing ENG bursts. The data obtained from the two cats, in which locomotion was enhanced by application of NA, highlight that changing the state of the spinal cord can alter the characteristics of the short-term synchrony by adding a high frequency component (coherence at 100–300 Hz) to it, but does not alter its distribution.
, 百拇医药
Functional relevance
In related studies on the pattern of motor-unit synchronization during locomotion in humans, Hansen et al. (2001) and Halliday et al. (2003) observed weak short-term synchronization restricted to close synergist muscles. This observation is similar to the findings from the acute spinal cat in the present study. However, in a group of ambulatory incomplete spinal cord-injured subjects Hansen et al. (2005) found a reduction or lack of short-term synchronization. This probably reflects an altered drive to spinal interneuronal networks, spinal motoneurones or both and requires further investigation. The present study reveals that, at least in the cat spinal cord, a population of last order interneurones may be brought into activity by the spinal CPG (independent of any descending system) and generate short-term synchrony in the motor output.
, http://www.100md.com
From a functional perspective Halliday et al. (2003) considered the restricted nature of the coupling in their study to confer on human locomotor rhythm-generating networks considerable scope for generating adaptable gait sequences. It is well known that the spinal cat preparation can generate different patterns of fictive locomotion in which the phase relations between muscle groups can change (Schomburg & Steffens, 1995). To achieve such adaptability in its output, hard wiring in the underlying locomotor network must be limited. The restricted nature of the divergence in the projection of the last order interneurones discussed in this report could be central to the provision of this adaptability.
, http://www.100md.com
References
Bloomfield P (2000). Fourier Analysis of Time Series: an Introduction, 2nd edn. Wiley, New York.
Brillinger DR (1981). Time Series – Data Analysis and Theory, 2nd edn. Holden Day, San Francisco.
Burke RE (2001). The central pattern generator for locomotion in mammals. Adv Neurol 87, 11–24.
Crone C, Hultborn H, Kiehn O, Mazieres L & Wigstrom H (1988). Maintained changes in motoneuronal excitability by short-lasting synaptic inputs in the decerebrate cat. J Physiol 405, 321–343.
, http://www.100md.com
Farmer SF, Bremner FD, Halliday DM, Rosenberg JR & Stephens JA (1993). The frequency content of common synaptic inputs to motoneurones studied during voluntary isometric contraction in man. J Physiol 470, 127–155.
Gabor D (1946). Theory of Communication. J Inst Elec Eng 93, 429–457.
Graham Brown T (1911). The intrinsic factors in the act of progression in the mammal. Proc R Soc Lond B Biol Sci 84, 308–319.
, 百拇医药
Grillner S (1985). Neurobiological bases of rhythmic motor acts in vertebrates. Science 228, 143–148.
Grillner S & Zangger P (1979). On the central generation of locomotion in the low spinal cat. Exp Brain Res 34, 241–261.
Halliday DM, Conway BA, Christensen LO, Hansen NL, Petersen NP & Nielsen JB (2003). Functional coupling of motor units is modulated during walking in human subjects. J Neurophysiol 89, 960–968.
, http://www.100md.com
Halliday DM, Rosenberg JR, Amjad AM, Breeze P, Conway BA & Farmer SF (1995). A framework for the analysis of mixed time series/point process data – theory and application to the study of physiological tremor, single motor unit discharges and electromyograms. Prog Biophys Mol Biol 64, 237–278.
Hamm TM & McCurdy ML (1995). The use of coherence spectra to determine common synaptic inputs to motoneurone pools of the cat during fictive locomotion. In Alpha and Gamma Motor Systems, ed. Taylor A, Gladden MH & Durbaba R, pp. 309–315. Plenum Press, New York, London.
, 百拇医药
Hamm TM, McCurdy ML, Trank TV & Turkin VV (2001). The use of correlational methods to investigate the organization of spinal networks for pattern generation. In Motor Neurobiology of the Spinal Cord, ed. Cope TC, pp. 135–171. CRC Press, New York.
Hamm TM, Trank TV & Turkin VV (1999). Correlations between neurograms and locomotor drive potentials in motoneurons during fictive locomotion: implications for the organization of locomotor commands. Prog Brain Res 123, 331–339.
, 百拇医药
Hansen NL, Conway BA, Halliday DM, Hansen S, Pyndt HS, Biering-Sorensen F & Nielsen JB (2005). Reduction of common synaptic drive to ankle dorsiflexor motoneurones during walking in patients with spinal cord lesion. J Neurophysiol 94, 934–942.
Hansen NL, Hansen S, Christensen LO, Petersen NT & Nielsen JB (2001). Synchronization of lower limb motor unit activity during walking in human subjects. J Neurophysiol 86, 1266–1276.
, 百拇医药
Jenkins GM & Watts DG (1968). Spectral Analysis and its Applications. Holden-Day, San Francisco.
Kiehn O, Hultborn H & Conway BA (1992). Spinal locomotor activity in acutely spinalized cats induced by intrathecal application of noradrenaline. Neurosci Lett 143, 243–246.
Kirkwood PA & Sears TA (1978). The synaptic connexions to intercostal motoneurones as revealed by the average common excitation potential. J Physiol 275, 103–134.
, http://www.100md.com
Kirkwood PA, Sears TA, Tuck DL & Westgaard RH (1982). Variations in the time course of the synchronization of intercostal motoneurones in the cat. J Physiol 327, 105–135.
Myers LJ, Erim Z & Lowery MM (2004). Time and frequency domain methods for quantifying common modulation of motor unit firing patterns. J Neuroengineering Rehabil 1, 2.
Pearson KG, Jiang W & Ramirez JM (1992). The use of naloxone to facilitate the generation of the locomotor rhythm in spinal cats. J Neurosci Meth 42, 75–81.
, 百拇医药
Perkel DH, Gerstein GL & Moore GP (1967). Neuronal spike trains and stochastic point processes. II. Simultaneous spike trains. Biophys J 7, 419–440.
Pollock LJ & Davis L (1930). The reflex activities of a decerebrate animal. J Comp Neurol 50, 377–411.
Rosenberg JR, Amjad AM, Breeze P, Brillinger DR & Halliday DM (1989). The Fourier approach to the identification of functional coupling between neuronal spike trains. Prog Biophys Mol Biol 53, 1–31.
, 百拇医药
Rosenberg JR, Halliday DM, Breeze P & Conway BA (1998). Identification of patterns of neuronal activity – partial spectra, partial coherence, and neuronal interactions. J Neurosci Meth 83, 57–72.
Rossignol S (1996). Neural control of stereotypic limb movements. In Handbook of Physiology, section 12, Exercise: Regulation and Integration of Multiple Systems, chap. 5, ed. Rowell L & Shepard J, pp. 173–216. The American Physiological Society, New York.
, 百拇医药
Schomburg ED & Steffens H (1995). Influence of opioids and naloxone on rhythmic motor activity in spinal cats. Exp Brain Res 103, 333–343.
Schuster A (1898). On the investigation of hidden periodicities with application to a supposed 26 day period of meteorological phenomenon. Terr Mag 3, 13–41.
Trank TV, Turkin VV & Hamm TM (1998). Coherence between locomotor drive potentials and neurograms of motor pools with variable patterns of locomotion. Ann N Y Acad Sci 860, 448–451.
Vaughan CW & Kirkwood PA (1997). Evidence from motoneurone synchronization for disynaptic pathways in the control of inspiratory motoneurones in the cat. J Physiol 503, 673–689.
Yakovenko S, McCrea D, Stecina K & Prochazka A (2005). Control of locomotor cycle durations. J Neurophysiol 94, 1057–1065., 百拇医药(J. B. Nielsen,, B. A. Con)
2 Institute of Physical Exercise and Sports Science, University of Copenhagen, Nrre Allé 55, 2100 Copenhagen , Denmark
3 Bioengineering Unit, University of Strathclyde, Glasgow, G4 0NW, UK
4 The Department of Electronics, University of York, UK
, http://www.100md.com 5 Department of Physiology, University of Oslo (Domus Medica), Oslo, Norway
Abstract
The basic locomotor rhythm in the cat is generated by a neuronal network in the spinal cord. The exact organization of this network and its drive to the spinal motoneurones is unknown. The purpose of the present study was to use time (cumulant density) and frequency domain (coherence) analysis to examine the organization of the last order drive to motoneurones during fictive locomotion (evoked by application of nialamide and dihydroxyphenylalanine (DOPA)) in the spinal cat. In all cats, narrow central synchronization peaks (half-width < 3 ms) were observed in cumulants estimated between electroneurograms (ENGs) of close synergists, but not between nerves belonging to muscles acting on different joints or to antagonistic muscles. Coherence was not observed at frequencies above 100 Hz and was mainly observed between synergists. Intracellular recording was obtained from a population of 70 lumbar motoneurones. Significant short-term synchronization was observed between the individual intracellular recordings and the ENGs recorded from nerves of the same pool and of close synergists. Recordings from 34 pairs of motoneurones (10 pairs belonged to the same motor pool, 11 pairs to close synergists and 13 pairs to antagonistic pools) failed to reveal any short-lasting synchronization. These data demonstrate that short-term synchronization during fictive locomotion is relatively weak and is restricted to close synergists. In addition, coherence analysis failed to identify any specific rhythmic component in the locomotor drive that could be associated with this synchronization. These results resemble findings obtained during human treadmill walking and imply that the spinal interneurones participating in the generation of the locomotor rhythm are themselves weakly synchronized. The restricted synchronization within the locomotor drive to motoneuronal pools may be a feature of the locomotor generating networks that is related to the ability of these networks to produce highly adaptive patterns of muscle activity during locomotion.
, http://www.100md.com
Introduction
It is well documented that a spinal neuronal network (central pattern generator; CPG) is responsible for generating the basic locomotor rhythm in vertebrates (Grillner, 1985; Rossignol, 1996; Burke, 2001). The periodicity of the step cycle and the patterns of muscle activity are therefore determined by elements within the networks generating and distributing the locomotor drive to the motor pools of the limbs.
In conceptual models of the CPG the locomotor drive to coactive motor pools is generally assumed to involve a common presynaptic drive (Graham Brown, 1911; Grillner, 1985; Yakovenko et al. 2005). Common or shared synaptic drive to populations of motoneurones may increase the probability of synchronized firing within a short time interval (short-term synchrony). In the event of synchronized motor activity during locomotion a statistical analysis has the potential to reveal features on the strength, frequency content and distribution of the common synaptic drive arising from the CPG (Kirkwood & Sears, 1978; Kirkwood et al. 1982; Farmer et al. 1993).
, http://www.100md.com
In previous experiments on cat fictive locomotion (induced by stimulation of the mesencephalic locomotor region; MLR), it has been demonstrated that coherence over a broad frequency range (up to 300 Hz) exists between ENGs recorded from both ipsi- and contralateral muscle nerves (Hamm & McCurdy, 1995; Trank et al. 1998; Hamm et al. 1999, 2001). This would suggest that the network responsible for the pattern of muscle activity is highly divergent. However, a combined time and frequency domain analysis of EMG data from healthy and spinal cord-injured human subjects walking on a treadmill failed to demonstrate any coherence above 50 Hz, and in healthy subjects short-term synchrony was restricted to synergists acting on the same joint (Hansen et al. 2001, 2005; Halliday et al. 2003). While this discrepancy between animal and human findings might reflect differences in the organization of the networks producing human (bipedal) and cat (quadrupedal) walking, it is also conceivable that the differences reflect the methodology used to generate locomotor-like behaviour in the decerebrate cat. During ongoing stimulation of the MLR the ENGs are likely to be strongly coupled to the stimulation, thereby generating significant coherence between the ENGs.
, http://www.100md.com
In the present study we investigated the coupling that exists between ENG activities recorded during fictive locomotion induced by the application of nialamide and DOPA in the spinal cat. In contrast to the results reported during MLR-induced locomotion (Hamm & McCurdy, 1995; Trank et al. 1998; Hamm et al. 1999, 2001), we found little evidence of high frequency coherence. Importantly, short-term synchrony was weak and restricted to close synergists. These findings are not consistent with a strongly synchronized and widely distributed last order drive to spinal motoneurones during locomotion. The analysis conducted demonstrates that the pattern of short-term synchrony in motor output during locomotion in the acute spinal cat is remarkably similar to that observed in humans.
, 百拇医药
Methods
Preparation
Data were obtained from nine adult cats (1.9–3.5 kg). All surgery and experimental protocols were conducted according to the National Institutes of Health Guide for the Care and Use of Laboratory Animals (National Institutes of Health publication no. 86–23, revised 1985). The experiments were approved by the Animal Experiments Inspectorate in Denmark. After anaesthesia with halothane–nitrous oxide (2–3% halothane in 70% N2O and 30% O2), the animals were intubated, and cannulas were inserted in the carotid artery and jugular vein for blood pressure monitoring and administration of fluid and drugs. Atropine (0.1 mg kg–1, S.C.) and dexamethasone (1 mg kg–1, I.V.) were given at the beginning of the experiment while buffer solution (10% dextrose and 1.7% NaHCO3) was infused continuously (4.5 ml h–1). During anaesthesia the cats were anaemically decerebrated by ligating the basilar and both common carotid arteries (and their collaterals), a procedure that has been shown to produce a decerebration that involves all cortical tissue above the pons and the anterior part of the cerebellum (Pollock & Davis, 1930; Crone et al. 1988). The anaesthetic was removed 5–6 h after ligating the vessels, and the decerebration was verified to be clinically complete by the development of tonic extensor muscle tone (alpha rigidity) in the forelimbs, lack of spontaneous movements, and large non-reactive pupils (Crone et al. 1988). Following these procedures and tests, pancuronium bromide (0.6 mg h–1) was given to block neuromuscular transmission, and artificial respiration was initiated. At the end of the experiments the animals were killed by an overdose of pentobarbital.
, http://www.100md.com
Prior to decerebration the following nerves from the left hindlimb were cut and dissected for recording: quadriceps (Q), sartorius (Sart), semimembranosus (Sm), anterior biceps (AB), posterior biceps (PB), semitendinosus (St), medial gastrocnemius (MG), lateral gastrocnemius plus soleus (LGS), tibialis anterior (TA), flexor digitorum and hallucis longus (FDHL), plantaris (Pl) and extensor digitorum longus (EDL). The Q and Sart nerves were placed in cuff electrodes, and the other nerves were mounted on bipolar silver–silver chloride hook electrodes. From the right limb, gastrocnemii + soleus (GS) and TA or deep peroneal (DP = TA + EDL) nerves were dissected. To expose the lumbar spinal cord a laminectomy (L4–S1 spinal cord segments) was performed and the animals transferred to a rigid frame, which secured and stabilized the spinal column. A complete spinalization was performed at thoracic segment 10–11 following local application of lidocaine (lignocaine). The expired CO2 was maintained between 3.0 and 5.0% by artificial ventilation and the blood pressure maintained by infusion of Dextran when needed. The temperature of the animal was kept at 38°C by a servo-controlled heating system.
, http://www.100md.com
Fictive locomotion was evoked by I.V. administration of nialamide (50 mg kg–1; Sigma-Aldrich, St Louis, MO, USA) and DOPA (50–100 mg kg–1; Pfizer, New York, NY, USA; Grillner & Zangger, 1979). Naloxone (100 μg kg–1; Sigma-Aldrich) was given to enhance the locomotor activity if necessary (Pearson et al. 1992). In two cats in which locomotor activity induced by nialamide and DOPA had waned, application of noradrenaline directly to the lumbar spinal cord re-established sustained locomotor activity (Kiehn et al. 1992).
, http://www.100md.com
Intracellular recordings were obtained from single or pairs of antidromically identified motoneurones innervating the left hindlimb. Glass micropipettes (tip diameter, 1.2–2 μm; resistance, 2–4 M) filled with the lidocaine derivative QX-314 in potassium acetate (2 M) were used. QX-314 blocks sodium spikes and may, in addition, reduce calcium currents. Thus, QX-314 limits the extent to which sodium and calcium currents could affect postsynaptic potentials, thereby facilitating the study of presynaptic drive.
, 百拇医药
Data analysis
Electroneurograms (ENGs), AC-coupled cord dorsum potentials, and DC-coupled intracellular recordings were digitized at a rate of 10 kHz with software developed by the Winnipeg Spinal Cord Research Center to run under real time QNX. The ENGs were band-pass filtered (high pass: 1 Hz; low pass 5 kHz). The theoretical framework for analysis of the present data represents a development of that presented in Halliday et al. (1995) to include time dependency, and to deal with analysis based on variable length segments. Time dependency is incorporated using the concept of a sliding data window (Gabor, 1946), discussed below. First we review briefly the calculation of the finite Fourier transform of time series data, and the periodogram-based approach to second order spectral estimation used in Halliday et al. (1995). For a stationary stochastic time series, x(t), the finite Fourier transform of a segment of length T, denoted by dTx (), is defined as:
, http://www.100md.com
at frequency . In discrete form this can be written as
for sample values xt. The j are the Fourier frequencies defined as (j= 0, . . . , T/2). Here corresponds to the Nyquist frequency. The quantity dTx(j) can be readily calculated using an Fast Fourier Transformation (FFT) algorithm.
For a record consisting of L sections of length T, the second order spectrum between two time series x(t) and y(t) can be estimated as:
, http://www.100md.com
The hat symbol indicates a parameter estimate; the overbar indicates a complex conjugate. The autospectra, fxx() and fyy() require only the finite Fourier transform of segments from their respective individual time series. Coherence, phase and cumulant density (cross-covariance) estimates can be constructed from the cross-spectral and auto-spectral estimates using the methods described in Halliday et al. (1995). Construction of spectral estimates using the average over L successive non-overlapping segments as in eqn (3) assumes that the signals are stationary across the entire record.
, 百拇医药
Equation (3) represents a periodogram approach to spectral estimation (Schuster, 1898). The periodogram itself is not a reliable estimate of the spectrum; statistical stability is achieved through averaging across the L periodograms. This approach is based on the assumption that data records are stationary, an assumption which does not apply to the present data (see Fig. 1). A periodogram approach does, however, allow task dependency in neural data to be explored through the use of averages constructed over repeat trials, in conjunction with an assumption of local stationarity (see below). Each segment of data is selected from a different trial, and, in the general case, may contain a different number of data points. Periodogram estimates require the same Fourier frequencies, which can be achieved for unequal numbers of data points through the use of zero padding, where the data in each segment is extended by the addition of zeros.
, 百拇医药
Electroneurogram (ENG) was recorded from a number of motor nerves in the left hindlimb (A). Recordings were also made from the right hindlimb (not shown). In order to determine burst onset, burst offset, burst duration, cycle duration and the phasing of the activity in the different ENGs, all ENGs were first rectified and smoothed and then the onset and offset of each burst of activity was determined automatically by setting a threshold in the computer software. The cycle duration was determined from the onset of one burst of activity in the posterior biceps (PB) nerve until the onset of the next burst. St, semitendinosus; Sm, semimembranosus; AB, anterior biceps; LGS, lateral gastrocnemius plus soleus; MG, medial gastrocnemius; Pl, plantaris; FDHL, flexor digitorum and hallucis longus; TA, tibialis anterior; EDL, extensor digitorum longus. B, for each fictive step of the locomotor bout partly shown in A, the burst onset in St (black circles), TA (red circles), EDL (green triangles), LGS (yellow triangles) and MG (blue squares) was plotted relative to the locomotor cycle duration (which was calculated from the PB burst onsets).
, 百拇医药
We now use Tl to refer to the number of data points in the lth section of data, and S to refer to the length of the finite Fourier transform, where S max{Tl; l= 1, . . . , L}. The periodogram of the lth section of data is
where the Fourier frequencies are (j= 0, . . . , S/2). If necessary, zeros are appended to each data segment, so that the Fourier transform is performed on a sequence of length S. The spectral estimate,, constructed from l segments is
, 百拇医药
where wl are the weights for each section, and . The optimum weights, in the sense of minimizing the variance of the spectral estimate, are given by (Bloomfield, 2000):
Implicit in the averaging of periodograms across segments of data from repeat trials is the assumption of local stationarity across trials. Here the qualifier ‘local’ is used to distinguish the assumption of stationarity within a short segment of data from the non-stationarity highlighted in Fig. 1 (see Results), which occurs on a longer time scale. The variance of the spectral estimate eqn (5) with weights given by eqn (6) can be approximated by (Bloomfield, 2000):
, 百拇医药
One advantage of the periodogram-based approach outlined above is that the setting of confidence limits generally involves simple expressions, often in conjunction with the use of a suitable transformation to stabilize the variance (Jenkins & Watts, 1968; Bloomfield, 2000). One parameter which plays a key role in the setting of confidence limits for periodogram estimates for stationary records based on equal length disjoint sections is the number of sections, L (Halliday et al. 1995). To extend this approach to the present analysis we define a quantity, L', as
, 百拇医药
where wl are the weights used to construct the periodogram estimate. We refer to L' as the number of equivalent sections in the periodogram estimate. This will not necessarily be an integer for an analysis based on unequal length segments. This quantity allows confidence limits for spectral parameters to be set using the expressions given in Halliday et al. (1995). For example, the magnitude of a 95% confidence limit on a plot of log10 of an auto-spectral estimate is given by ± 0.851/L', and the upper 95% confidence limit for coherence estimates, based on the assumption of independence, is given by the quantity 1 – (0.05)1/(L'–1). Using eqns (5), (6) and (8), time/frequency multivariate parameters can be estimated and confidence limits set using the methods described in (Halliday et al. 1995).
, 百拇医药
ENG records were full wave rectified prior to analysis, intracellular records were not rectified, but an additional pre-processing in the form of a 4 Hz high pass filter was applied to intracellular records prior to analysis. This was necessary to remove the large low frequency trend induced by the locomotor drive potential (see Fig. 5). Three different methods were used to analyse the correlation between pairwise ENG and intracellular recordings (ENG to ENG, motoneurone to ENG and motoneurone to motoneurone). In the first method, two channels were selected as reference channels, one for extensors and one for flexors. Triggers were placed at the start of each burst of activity in these reference channels. The timing of individual trigger points was automatically performed in the QNX software, all trigger points were subsequently verified by visual inspection. These trigger points were used to calculate spike-triggered averages of every neurogram record, from which activity maps of the ENG were constructed (see Fig. 1). The activity maps indicate the relative timing of ENG bursts within each locomotor cycle (in an average sense) with reference to the trigger times. From these activity maps, periods of coactivity between pairs of ENG and intracellular motoneuronal records were determined by visual inspection of the spike-triggered averages. These periods of coactivity were used to obtain the offset and length of data segment from each trial to include in the spectral analysis. In this method the offset and segment length were the same for each trial. Estimates of auto- and cross-spectra were then formed according to eqn (5), by averaging across segments taken from each trial. The results in Figs 2, 3, 5 and 6 were obtained using this approach.
, 百拇医药
Intracellular recording was obtained simultaneously from two individual PB motoneurones. A, intracellular recording from the two motoneurones together with the ENG from the PB nerve at three different time scales. B, the coherence (upper graph) and cumulant density function (two lower graphs at different time scales) between the two intracellular recordings.
Coherence (upper row of graphs in A and B) and the cumulant density function (lower two rows in A and B) was calculated for the following pairs of ENG : PB and St (A; 1st column), PB and TA (A; 2nd column), TA and EDL (A; 3rd column), MG and LGS (B; 1st column), MG and plantaris (Pl) (B; 2nd column) and MG and TA (B; 3rd column). The cumulant was plotted with a long (middle row) and short time base (lower row). The horizontal lines in the middle row of graphs designate the confidence limits.
, 百拇医药
Intracellular measurements were obtained from an MG motoneurone together with MG, LGS, FDHL, Sm and TA ENGs. The cumulant density estimate between the motoneurone and the ENG activities (A, C, E, G, I) was calculated for segments of the ENG recording where a burst was observed. In the computer software a level was set to generate triggers corresponding to individual ENG spikes (roughly the 30% largest spikes). An average of the intracellular recording for a time period of 100 ms around the time of the trigger events was obtained (B: MG (37 505 triggers), D: LGS (40 642 triggers), F: FDHL (75 000 triggers), H: Sm (52 679 triggers) and J: TA (18 805 triggers)).
, 百拇医药
The second method of analysis was used to characterize changes in the correlation structure over different phases of the locomotor cycle, and was based on the same sequences of triggers as in the first method, above. A number of different spectral estimates were constructed, using a range of different offset values from each trigger, and a fixed segment length of 200 ms of data from each trial (see Fig. 4A). This is similar to the analysis carried out in Halliday et al. (2003) on human locomotion data. A segment length of 200 ms was used; this defines the frequency resolution of spectral estimates to be around 5 Hz. The results in Fig. 4 were obtained using this approach.
, http://www.100md.com
The data are from the same cat as used for Fig. 3. Coherence (B) and cumulant density estimates (C, long time base; D, short time base) were calculated for 200 ms segments starting at different times (400 and 200 ms before onset, at onset and 200, 400 and 600 ms after burst onset) in relation to the onset of PB burst activity.
The third method of analysis was introduced to investigate the effects of using a fixed length data segment from each burst (method 1) when the actual duration of bursts is not constant. In this method two triggers were used to mark each burst, one at the start and one at the end. These pairs of triggers were then used to select a different segment length from each trial, such that only data samples lying between the start and end trigger points for each burst were included from each segment in eqn (5). In this method the wl in eqn (6) will vary according to the relative length of each segment. No results from this method of analysis are illustrated; however, the Results section includes a description of results obtained using this method of analysis.
, 百拇医药
The correlation between rectified ENG signals and/or high pass filtered intracellular signals was assessed in the time and frequency domains using the functions described below. In the frequency domain, correlation was assessed through coherence functions (Brillinger, 1981; Rosenberg et al. 1989; Halliday et al. 1995). The coherence function between two signals, x(t) and y(t), is defined at frequency as:
Coherence functions provide normative measures of linear association on a scale from 0 to 1. For the present data, the coherence provides a measure, at each Fourier frequency , of the fraction of the activity in one signal which can be predicted by the activity in the second signal.
, 百拇医药
In the time domain, estimates of the cumulant density function were used to characterize the correlation between pairs of signals. The cumulant density function, denoted by qxy(u), is defined as the inverse Fourier transform of the cross-spectrum
)
For two uncorrelated signals the cumulant has an expected value of zero; deviations from this indicate a correlation between the two signals at a particular time lag, u. In the present experiments, which are exploring aspects of the common drive to motoneurone pools during locomotion, the dominant feature in cumulant density estimates is the central peak around zero lag, reflecting the presence of common synaptic input. Rhythmic inputs will induce symmetrical oscillatory components in the cumulant (Perkel et al. 1967), the frequency and strength of these components can be quantified from the corresponding coherence estimates. Cumulant density functions are analogous to cross-correlation functions often used to quantify spike train data, and have a similar interpretation (Halliday et al. 1995).
, http://www.100md.com
The interpretation of the above time and frequency domain measures of correlation does depend on which of the three specific methods of spectral estimation was used. Correlation analysis based on method 1 can be interpreted as providing an indication of correlation between the two signals at periods when the two signals are coactive. This assumes that the signals have local stationarity, i.e. that the correlation structure does not change markedly during the burst. The third method of spectral analysis (using both start and end triggers for each burst) can be viewed in a similar manner. The second method of spectral analysis introduces a greater degree of time dependence into the correlation estimates, which should now be interpreted as a measure of the correlation within a narrow time window (200 ms in this case) that is offset from the trigger point marking the start of each burst.
, 百拇医药
The present analysis is concerned with identification of the characteristics of common synaptic inputs to motoneuronal pools during fictive locomotion. Our interpretation of coherence and cumulant density estimates is similar to that used by Farmer et al. (1993), who used coherence estimates to quantify the strength and frequency of rhythmic synaptic inputs to a motoneurone pool during isometric contractions. A distinct band of coherence in a particular frequency range is taken as a marker for common rhythmic inputs to the motoneurones over this frequency range. The magnitude of coherence is used as an indication of the strength of the common input. A broad band of monotonically decreasing coherence is interpreted as an indication of common synaptic inputs which do not contain any specific rhythmic components. A theoretical perspective and further discussion of the common input model is given in Rosenberg et al. (1998). Similar inferences can often be drawn from time domain measures of association, thus time and frequency analyses should be considered as complementary approaches (see also Myers et al. 2004).
, http://www.100md.com
Cumulant density estimates provide a generalized measure of correlation, and as such can be viewed as statistical parameters. The intracellular records were further analysed by performing an average with respect to a multi-unit trigger signal derived from the homonymous nerve signal. This parameter has been termed the average common excitatory (ACE) potential (Kirkwood & Sears, 1978). Construction of ACE estimates allows direct comparison between the present data and those derived from respiratory CPG studies (Kirkwood & Sears, 1978).
, 百拇医药
Results
ENG data
Fictive locomotion was induced by the combination of nialamide and DOPA in nine cats. In five cats the locomotion consisted of slow alternating flexor and extensor bursts with phase-related modulation in the contralateral ENGs, consistent with a left–right reciprocating extensor–flexor locomotor rhythm. This pattern of motor activity persisted for several hours in these preparations. In the four remaining cats rhythmic bursts in flexor ENGs were observed, but without a clear or consistent reciprocal bursting in extensor ENGs. Note, that rhythmic activity in extensors was never observed in the absence of rhythmic flexor activity. This is a common feature of the fictive locomotor pattern in nialamide and DOPA-treated spinal cats (Grillner & Zangger, 1979).
, 百拇医药
An example of the ENG activity associated with fictive locomotion induced by nialamide and DOPA in the spinal cat is illustrated in Fig. 1A. In the example shown, rhythmic alternating extensor and flexor activity is apparent in most of the ENG records. The duration of a single fictive step ranged from 2 to 4 s (mean: 2.6 ± 0.6 s), producing a step frequency of 0.4 Hz (114 steps in 300 s). The activity of the different flexor nerves was largely in-phase and the mean burst duration was 0.9 ± 0.012 s. The extensor nerves displayed in-phase activity that was reciprocally organized with respect to the flexor burst activity. The mean burst duration in extensors was 1.6 ± 0.03 s (for LGS). Figure 1B shows the onset of the ENG activity in both flexors (St, TA and EDL) and extensors (LGS and MG) relative to the onset of PB activity. During this particular bout of locomotion, no activity was recorded from LGS and MG during several cycles (30 and 41 cycles out of 115, respectively) while the activity in St, TA and EDL failed in only a few cycles (6–8 cycles).
, http://www.100md.com
Figure 2 illustrates a time and frequency analysis of paired ENG recordings where epochs of coactivity were extracted and used to calculate the coherence (uppermost graphs in Fig. 2A and B) and the cumulant density functions (middle and lowermost graphs in Fig. 2A and B), respectively.
Significant coherence was seen in all the pairs of ENGs (synergists and antagonists) but only at frequencies below 50 Hz. Significant differences were observed in the cumulant density estimates where a broad peak of synchronization was observed for synergist ENG pairs and a trough for antagonistic ENG (e.g. TA and MG) pairs, reflecting the phase relationships that exist between these muscle groups. The broad peak is considered to reflect features of the envelope of the ENG bursts and its duration is similar to the duration of the individual ENG bursts.
, 百拇医药
For synergist ENGs the cumulant displays an additional feature, which is recognizable as a narrow (half-width < 3 ms) central peak (Fig. 2A and B, bottom row of graphs showing the cumulant on an expanded scale). This central peak is taken as a positive marker for the presence of short-term synchronization between ENG recordings (Vaughan & Kirkwood, 1997). Such peaks were only observed for close synergists such as PB and St, TA and EDL, and MG and LGS. In no animal was short-term synchronization identifiable between synergist nerves acting on different joints (PB and TA or AB and LGS) or between antagonist (TA and LGS) muscle nerves (Fig. 2).
, 百拇医药
In all nine cats the coherence estimates were characterized by the absence of any significant coherence for frequencies above 100 Hz. All analyses showed a large coherence at low frequencies monotonically decreasing towards 20–100 Hz. This is considered in part to echo the co-modulation of the ENG records by the locomotor drive as expressed through the activity envelope of the ENG epochs, but it may also reflect an asynchronous activation arising from the CPG. It is important to note that this low frequency coherence is not equivalent to the frequency of the locomotor rhythm as this frequency component would only become manifest if the coherence had been estimated from non-segmented data sets. Although significant coherence was observed at the same frequencies for synergists working at the same joint as for synergists working at different joints, the magnitude of the coherence was always smaller in the latter cases. In all cases the coherence lacked any evidence of discrete spectral components.
, 百拇医药
For all nine cats, short-term synchronization between flexor motor pools could be identified for paired PB and St and paired TA and EDL ENG recordings, but not for synergists working at different joints (PB and TA, PB and EDL, St and TA, St and EDL). Similarly, in the five cats displaying consistent extensor ENG bursting, short-term synchrony was seen for synergists working on the same joint, but not for synergists working on different joints. In these five cats, no short-term synchrony was observed between antagonistic flexor and extensor ENG pairs. Similarly, no short-term synchronization was observed between bilateral pairs of coactive extensors and flexors in any animal.
, 百拇医药
High frequency oscillations during noradrenaline-induced fictive locomotion
In the two animals that received noradrenaline (NA) to boost the fading DOPA locomotion, the coherence and cumulant estimates revealed an additional feature during episodes of locomotor activity. Sample locomotor data from one of these cats is shown in Fig. 3A. Comparison with Fig. 1A reveals a difference in the temporal characteristics of the burst activity. Following NA, the cycle frequency was approximately 0.5 Hz (155 steps in 300 s; cycle duration: 1.9 ± 0.6 s), almost twice as high as the cycle frequency for the cat used for Fig. 1. Furthermore, the magnitude of the burst activity was much more variable following administration of NA (compare Fig. 3A and Fig. 1A) and short-lasting bursts of activity, which appear to be unrelated to the ongoing locomotor activity occurred simultaneously in both extensors and flexors (marked by arrows).
, 百拇医药
In A the burst activity in the recorded motor nerves is shown. The arrows indicate bursts which occurred synchronously in extensor and flexor nerves and which appeared to be separate from the locomotor burst activity. Coherence (upper row of graphs in B and C) and the cumulant density function (lower two rows in B and C) was calculated for the following pairs of ENGs: PB and St (B; 1st column), PB and TA (B; 2nd column), TA and EDL (B; 3rd column), MG and LGS (C; 1st column), LGS and FDL (C; 2nd column) and LGS and TA (C; 3rd column). The cumulant was plotted with a long (middle row) and short time base (lower row).
, 百拇医药
In agreement with the data from Fig. 2, short-term synchrony was only present in this preparation for close synergists and absent for synergists working at different joints or for antagonist muscle pairs. However, in contrast to the findings of the coherence analysis reported in Fig. 2 an additional high frequency band of coherence peaking between 100 and 300 Hz was seen for pairs of close synergists (PB : St and TA : EDL). This high frequency coupling was not observed between synergists working at different joints (PB and TA see Fig. 2B). Findings consistent with those shown in Fig. 3 were also seen in the second cat in which NA was used to evoked rhythmic activity.
, 百拇医药
Modulation of coupling during the locomotor cycle
To determine the coupling pattern within the locomotor cycle an analysis based on short (200 ms) fixed data segments and variable (ENG burst duration) length segments (see Methods) was used. The advantage of this approach is that it minimizes potential effects of the cyclic modulation of the ENG during each burst. Figure 4 illustrates data from one of the cats where coherence was observed above 100 Hz (after application of NA), but apart from this essentially similar findings were observed in all nine cats. Figure 4A shows the ENG recordings from the PB and St motor nerves and Fig. 4B–D shows the coherence and cumulant density estimates for the different 200 ms segments starting at 400 ms prior to the onset of the burst until 600 ms after burst onset. A low frequency coherence component is present in all sections of the segmented data, but is greatest in segments sampled around the onset of the ENG bursts. During that period, an additional 100–300 Hz band of coherence also became apparent in this experiment. Note that the low frequency coupling is accompanied by short-term synchronization within the early part of the burst and that clear secondary features at short lags (± 4 ms) are observed in the cumulant for the segments where 100–300 Hz coherence appeared.
, http://www.100md.com
Due to variability in burst intervals and durations (see Fig. 1B) a further analysis based on the use of triggers to identify the start and end of individual bursts was conducted. This allowed the analysis to be restricted only to ENG data occurring during bursts. The results from this analysis closely parallel those obtained by using the fixed average segment length. In both types of analyses, the presence of the large low frequency coherence (e.g. the centre portion of the burst in Fig. 4) suggests that the low frequency component is not simply an artefact reflecting the frequency of the rhythmic modulation of the burst activity in synergists, but is a signature of a common modulation of motoneuronal drive possibly derived from the CPG.
, 百拇医药
Intracellular data
Simultaneous intracellular recordings were obtained from 34 pairs of motoneurones during fictive locomotion in six of the cats. Ten of these pairs belonged to the same motor pool (7 extensor pairs, 3 flexor pairs), 5 belonged to two different close synergistic pools (2 extensor and 3 flexor pairs) and 8 belonged to synergists working at different joints (6 extensor and 2 flexor pairs). The remaining 11 pairs belonged to antagonistic motor pools.
, http://www.100md.com
Figure 5 demonstrates a representative example of the observations that were obtained. The recordings are from a pair of PB motoneurones. The action potentials in the two motoneurones were blocked by QX-314. As expected, the membrane potential of the two motoneurones is depolarized during the burst of activity in the PB motor nerve, reflecting the locomotor drive to the two motoneurones during the flexor phase of the locomotor cycle. As shown on expanded time scales, synaptic noise in the two motoneurones is quite different despite the similar time course of the locomotor drive potential. This is also reflected in the coherence and the cumulant density estimates, which did not reveal any statistically significant coupling of the synaptic activity other than that which could be related to the envelope of the depolarizing locomotor drive potential. For all pairs of intracellular recordings, coherence was only observed at very low frequency and there was no evidence of short-term synchrony from cumulant estimates. It is worth noting that when making paired intracellular recordings, the microelectrodes were placed 3–5 mm apart to avoid mechanical interference during tracking.
, http://www.100md.com
To investigate whether the synaptic activity in individual motoneurones was coupled to the burst activity in the nerves, coherence and cumulant densities were estimated from each intracellular recording (70 motoneurones) and the ENGs. Figure 6 shows data from a single MG motoneurone. The cumulant density function revealed significant coupling between the non-spiking motoneurone and the ENG activities from the MG and LGS nerves (Fig. 6A and C), but not from synergists working at different joints (FDHL and Sm; Fig. 6E and G) or antagonists (TA; Fig. 6I). Short-term synchrony between the intracellular recording and the corresponding motor nerve was observed in 50 of the 70 motoneurones sampled. Of these 50 cells short-term synchrony with close synergistic motor nerves was observed in 30 cases. No high frequency coherence was ever observed.
, 百拇医药
When the membrane potential recorded from the motoneurone was averaged using the ENG activity from the respective nerve as a trigger, an averaged common excitatory potential (ACE potential; Kirkwood & Sears, 1978) was observed for the MG and LGS nerves (Fig. 6B and D), but not for the other nerves (Fig. 6F, H and J). In 39 of the 50 motoneurones which showed short-term synchrony a similar ACE potential was found. The size of the potential varied from 100 μV to around 600 μV. The size of the short-term synchrony peak in the cumulant density function varied from 0.001 to 0.005, with a highly significant correlation to the size of the ACE potential (Spearman's rank; P < 0.001). An ACE potential was never observed without a significant short-term synchrony peak in the cumulant density function.
, 百拇医药
Discussion
It is well established that the rhythmic activities of synergistic as well as antagonistic motoneurones during fictive locomotion are functionally coupled (Grillner & Zangger, 1979; Schomburg & Steffens, 1995; cf. also Figs 1 and 3 of the present study). This reflects a common source of activation to the muscles acting on the different joints. A consequence of this organized burst-like motoneuronal behaviour is the appearance of a broad synchronization feature in cumulant density estimates that reflects the underlying temporal synergies between muscle groups (see Figs 1 and 2). The question we have addressed in the present study is to what extent a common synaptic drive from branches of last order neurones (Kirkwood & Sears, 1978) participate in forming the coupling of motoneuronal activities observed during fictive locomotion. In the time domain, this common synaptic drive (short-term synchronization) is identified by the presence of a narrow central peak with a half-width shorter than 3 ms (Vaughan & Kirkwood, 1997).
, http://www.100md.com
Short-term synchronization was observed between close synergists, but never for synergists working at different joints or for antagonists. A similar brief peak was also observed between intracellular recordings from single (non-spiking) motoneurones and their corresponding motor nerve (and in some cases also the nerves of their close synergists). This suggests that there is a difference in the organization of the synaptic drive from the spinal CPG to close synergists as compared to synergists working on different joints. This view is affirmed when considering data sets obtained from the two cats in which intrathecal noradrenaline was used to re-establish locomotor behaviour. In these cats short-term synchronization between close synergists was (following noradrenaline administration) accompanied by high frequency (200 Hz) coherence. High frequency coherence was never observed without short-term synchrony being present in the respective cumulant. The most likely mechanism for generating coherence between ENGs at such high frequencies is a reflection of the frequency content in the discharges of the last order neurones responsible for the short-term synchrony. In such cases the coherence does not reflect motoneuronal discharge frequencies, but reveals that a common high frequency modulation of output can be detected within the coupled motor pools (see Farmer et al. 1993). In nialamide- and DOPA-induced fictive locomotion short-term synchrony was apparent without accompanying high frequency coherence. We interpret this as an indication that during locomotion significant coupling of motoneuronal discharges by last order interneurones can occur without significant coupling in motoneuronal discharge frequencies even when the locomotor rhythm is itself stable. This situation will arise when the interneuronal drive from last order interneurones during different locomotor bursts does not display a distinct and robust spectral signature. Only in cases where a significant part of the last order interneuronal discharge contains reproducible spectral features, occurring within a specific frequency band, will motoneuronal coherence reflect the modulatory influence of the interneuronal drive.
, 百拇医药
Relation to MLR-induced fictive locomotion
In part, our data confirm the findings of Hamm and coworkers for fictive locomotion induced by MLR stimulation (Hamm & McCurdy, 1995; Trank et al. 1998; Hamm et al. 1999, 2001) in that significant coupling (in both the frequency and time domains) was observed for extensors and flexors acting on all limb segments as judged by the low frequency coherence and broad synchronization. However, the widespread broad band coherence reported in MLR preparations (Hamm & McCurdy, 1995; Trank et al. 1998; Hamm et al. 1999, 2001) was not seen in our data. In the majority of our data, coherence was restricted to frequencies well below 100 Hz and in the two cases where coherence was also seen at frequencies between 100 and 300 Hz, this high frequency coupling had a restricted organization that was linked to the distribution of short-term synchrony between close synergists. The combined time and frequency domain approach we have used therefore allows more precise descriptions to be made of the nature of the coupling process associated with the appearance of significant coherence between coactive motor pools. It seems to us a likely possibility that the stimulus train during MLR-evoked locomotion in the studies by Hamm and colleagues (Hamm & McCurdy, 1995; Trank et al. 1998; Hamm et al. 1999, 2001) may have induced widespread and high frequency coherence in all lumbar motor pools independently of the CPG activity. However, it should be noted that widespread high frequency coherence was also observed by Hamm & McCurdy (1995) during spontaneous locomotion and by Hamm et al. (2001) during fictive scratching – without any MLR stimulation. Other factors may thus also contribute to their findings.
, http://www.100md.com
Intracellular coupling
In the present study we did not see short-term synchronization between pairs of intracellular recordings, although all received a locomotor drive potential. To achieve stable paired motoneuronal recordings in this preparation the microelectrodes were inserted at some distance from each other. The lack of short-term synchronization in the sampled motoneurones may therefore be a consequence of the anatomical projections of the interneurones being highly localized. Thus, only motoneurones lying close to one another would demonstrate mutual short-term synchrony. This anatomical framework would also explain why short-term synchrony between intracellular recordings and the population ENG recordings can be observed. Related to this, the low level of synchrony across a motor pool would also suggest that the population of interneurones are themselves weakly coupled.
, 百拇医药
Interpretation in relation to the spinal CPG
The findings in the present study suggest that part of the output from the spinal rhythm generator is conveyed through a local segmental network of last order spinal interneurones with limited capacity for coupling motoneurones arising in different spinal segments. Previously, Grillner & Zangger (1979) noticed that flexor and extensors working at different joints appeared to be less tightly coupled than flexors and extensors working on the same joint. Our data extend this to include the distribution of short-term synchonization.
, http://www.100md.com
Importantly, our data indicate that the synergy of muscle activities during locomotion is not determined by the last order neurones. Rather the pattern of activity appears to be determined by a distributed input, to which the last order neurones studied here do not have sufficient divergence to define. Accordingly, we suggest that the last order neurones that generate the short-term synchonization revealed in this study have no specific role in determining the rate, timing or phasing of the locomotor bursts. However, these neurones could play an important role in setting the magnitude of the bursts in relation to functional demands. If this is the case, the group of interneurones may participate in reflexes that fail to entrain CPG activity, but lead to an alteration in the level of ongoing ENG bursts. The data obtained from the two cats, in which locomotion was enhanced by application of NA, highlight that changing the state of the spinal cord can alter the characteristics of the short-term synchrony by adding a high frequency component (coherence at 100–300 Hz) to it, but does not alter its distribution.
, 百拇医药
Functional relevance
In related studies on the pattern of motor-unit synchronization during locomotion in humans, Hansen et al. (2001) and Halliday et al. (2003) observed weak short-term synchronization restricted to close synergist muscles. This observation is similar to the findings from the acute spinal cat in the present study. However, in a group of ambulatory incomplete spinal cord-injured subjects Hansen et al. (2005) found a reduction or lack of short-term synchronization. This probably reflects an altered drive to spinal interneuronal networks, spinal motoneurones or both and requires further investigation. The present study reveals that, at least in the cat spinal cord, a population of last order interneurones may be brought into activity by the spinal CPG (independent of any descending system) and generate short-term synchrony in the motor output.
, http://www.100md.com
From a functional perspective Halliday et al. (2003) considered the restricted nature of the coupling in their study to confer on human locomotor rhythm-generating networks considerable scope for generating adaptable gait sequences. It is well known that the spinal cat preparation can generate different patterns of fictive locomotion in which the phase relations between muscle groups can change (Schomburg & Steffens, 1995). To achieve such adaptability in its output, hard wiring in the underlying locomotor network must be limited. The restricted nature of the divergence in the projection of the last order interneurones discussed in this report could be central to the provision of this adaptability.
, http://www.100md.com
References
Bloomfield P (2000). Fourier Analysis of Time Series: an Introduction, 2nd edn. Wiley, New York.
Brillinger DR (1981). Time Series – Data Analysis and Theory, 2nd edn. Holden Day, San Francisco.
Burke RE (2001). The central pattern generator for locomotion in mammals. Adv Neurol 87, 11–24.
Crone C, Hultborn H, Kiehn O, Mazieres L & Wigstrom H (1988). Maintained changes in motoneuronal excitability by short-lasting synaptic inputs in the decerebrate cat. J Physiol 405, 321–343.
, http://www.100md.com
Farmer SF, Bremner FD, Halliday DM, Rosenberg JR & Stephens JA (1993). The frequency content of common synaptic inputs to motoneurones studied during voluntary isometric contraction in man. J Physiol 470, 127–155.
Gabor D (1946). Theory of Communication. J Inst Elec Eng 93, 429–457.
Graham Brown T (1911). The intrinsic factors in the act of progression in the mammal. Proc R Soc Lond B Biol Sci 84, 308–319.
, 百拇医药
Grillner S (1985). Neurobiological bases of rhythmic motor acts in vertebrates. Science 228, 143–148.
Grillner S & Zangger P (1979). On the central generation of locomotion in the low spinal cat. Exp Brain Res 34, 241–261.
Halliday DM, Conway BA, Christensen LO, Hansen NL, Petersen NP & Nielsen JB (2003). Functional coupling of motor units is modulated during walking in human subjects. J Neurophysiol 89, 960–968.
, http://www.100md.com
Halliday DM, Rosenberg JR, Amjad AM, Breeze P, Conway BA & Farmer SF (1995). A framework for the analysis of mixed time series/point process data – theory and application to the study of physiological tremor, single motor unit discharges and electromyograms. Prog Biophys Mol Biol 64, 237–278.
Hamm TM & McCurdy ML (1995). The use of coherence spectra to determine common synaptic inputs to motoneurone pools of the cat during fictive locomotion. In Alpha and Gamma Motor Systems, ed. Taylor A, Gladden MH & Durbaba R, pp. 309–315. Plenum Press, New York, London.
, 百拇医药
Hamm TM, McCurdy ML, Trank TV & Turkin VV (2001). The use of correlational methods to investigate the organization of spinal networks for pattern generation. In Motor Neurobiology of the Spinal Cord, ed. Cope TC, pp. 135–171. CRC Press, New York.
Hamm TM, Trank TV & Turkin VV (1999). Correlations between neurograms and locomotor drive potentials in motoneurons during fictive locomotion: implications for the organization of locomotor commands. Prog Brain Res 123, 331–339.
, 百拇医药
Hansen NL, Conway BA, Halliday DM, Hansen S, Pyndt HS, Biering-Sorensen F & Nielsen JB (2005). Reduction of common synaptic drive to ankle dorsiflexor motoneurones during walking in patients with spinal cord lesion. J Neurophysiol 94, 934–942.
Hansen NL, Hansen S, Christensen LO, Petersen NT & Nielsen JB (2001). Synchronization of lower limb motor unit activity during walking in human subjects. J Neurophysiol 86, 1266–1276.
, 百拇医药
Jenkins GM & Watts DG (1968). Spectral Analysis and its Applications. Holden-Day, San Francisco.
Kiehn O, Hultborn H & Conway BA (1992). Spinal locomotor activity in acutely spinalized cats induced by intrathecal application of noradrenaline. Neurosci Lett 143, 243–246.
Kirkwood PA & Sears TA (1978). The synaptic connexions to intercostal motoneurones as revealed by the average common excitation potential. J Physiol 275, 103–134.
, http://www.100md.com
Kirkwood PA, Sears TA, Tuck DL & Westgaard RH (1982). Variations in the time course of the synchronization of intercostal motoneurones in the cat. J Physiol 327, 105–135.
Myers LJ, Erim Z & Lowery MM (2004). Time and frequency domain methods for quantifying common modulation of motor unit firing patterns. J Neuroengineering Rehabil 1, 2.
Pearson KG, Jiang W & Ramirez JM (1992). The use of naloxone to facilitate the generation of the locomotor rhythm in spinal cats. J Neurosci Meth 42, 75–81.
, 百拇医药
Perkel DH, Gerstein GL & Moore GP (1967). Neuronal spike trains and stochastic point processes. II. Simultaneous spike trains. Biophys J 7, 419–440.
Pollock LJ & Davis L (1930). The reflex activities of a decerebrate animal. J Comp Neurol 50, 377–411.
Rosenberg JR, Amjad AM, Breeze P, Brillinger DR & Halliday DM (1989). The Fourier approach to the identification of functional coupling between neuronal spike trains. Prog Biophys Mol Biol 53, 1–31.
, 百拇医药
Rosenberg JR, Halliday DM, Breeze P & Conway BA (1998). Identification of patterns of neuronal activity – partial spectra, partial coherence, and neuronal interactions. J Neurosci Meth 83, 57–72.
Rossignol S (1996). Neural control of stereotypic limb movements. In Handbook of Physiology, section 12, Exercise: Regulation and Integration of Multiple Systems, chap. 5, ed. Rowell L & Shepard J, pp. 173–216. The American Physiological Society, New York.
, 百拇医药
Schomburg ED & Steffens H (1995). Influence of opioids and naloxone on rhythmic motor activity in spinal cats. Exp Brain Res 103, 333–343.
Schuster A (1898). On the investigation of hidden periodicities with application to a supposed 26 day period of meteorological phenomenon. Terr Mag 3, 13–41.
Trank TV, Turkin VV & Hamm TM (1998). Coherence between locomotor drive potentials and neurograms of motor pools with variable patterns of locomotion. Ann N Y Acad Sci 860, 448–451.
Vaughan CW & Kirkwood PA (1997). Evidence from motoneurone synchronization for disynaptic pathways in the control of inspiratory motoneurones in the cat. J Physiol 503, 673–689.
Yakovenko S, McCrea D, Stecina K & Prochazka A (2005). Control of locomotor cycle durations. J Neurophysiol 94, 1057–1065., 百拇医药(J. B. Nielsen,, B. A. Con)