Non-uniform olivocerebellar conduction time in the vermis of the rat cerebellum
1 Department of Anatomy, Downing Street, Cambridge CB2 3DY, UK
Abstract
It has been proposed that the conduction velocities of cerebellar climbing fibre (olivocerebellar) axons are tuned according to length, in order to precisely fix the conduction time between the inferior olive and cerebellar cortex. Some data conflict with this view. We have re-evaluated this issue using the climbing fibre reflex. The white matter of the tip of one folium in lobule VI or VII was stimulated electrically 0.5–1 mm below the surface and recordings were made from Purkinje cells in lobules VIII and IX. Reflex evoked climbing fibre (CF) responses (33 units) were recorded at different depths from Purkinje cells found in a narrow sagittal zone of cortex as complex spikes. The responses had latencies ranging from 4.3 ms to 11.3 ms. A consistent trend was that Purkinje cell responses recorded at greater depth had shorter CF reflex latencies than those recorded more superficially, both in individual experiments and in grouped data. These data show that the CF reflex latency is not constant, but is directly proportional to the distance an action potential has to travel along a CF. These data are not consistent with tuning of CF conduction velocities to normalize olivocerebellar conduction time, but are consistent with a CF conduction velocity in the cortex of approximately 0.6 m s–1. This suggests that climbing fibres projecting to different parts of the cerebellar cortex may have differences in spike conduction time of a few milliseconds, and that submillisecond precision is not an important element of the climbing fibre signal.
, 百拇医药
Introduction
Precise timing of neural signals is essential to the function of any neural network; for instance learning rules usually require the temporal coincidence of more than one synaptic input. Most models of cerebellar function assume that the timing of climbing fibre (CF) activation is important (for reviews see Lang, 2003; Kitazawa & Wolpert, 2005), but the precision of this timing is not established. One particular view, based on the remarkable properties of the inferior olivary nucleus, argues that synchronous activity in the olivo-cerebellar system provides precise timing signals essential to the generation of movements (Llinas & Welsh, 1993; Welsh et al. 1995; Llinas, 1997). In support of this idea, multiple single unit recordings have shown synchronous CF activity within small areas of the cerebellar cortex (Sasaki et al. 1989; Lang, 2003). The cerebellar cortex is deeply folded and CF axons terminating in different regions of the cerebellum have different lengths. If all CF axons had similar conduction velocities then activity that is synchronous at the level of the inferior olive will become desynchronized at the level of the cortex, the longer conduction path having a longer conduction time. It has been proposed that CF conduction velocity is tuned such that longer CFs, which project to more superficial parts of the cerebellar folia, transmit action potentials with a conduction velocity that is greater than that of shorter CFs that terminate deeper in the cortex (Sugihara et al. 1993; Lang & Rosenbluth, 2003). Under this scheme, despite the foliation of the cerebellum and different lengths of climbing fibres, synchronous activity at the level of the inferior olive is maintained in the CF activation times of Purkinje cells (Sugihara et al. 1993; Lang & Rosenbluth, 2003).
, 百拇医药
Experiments seeking evidence for the same phenomenon in the cat, in which the cerebellar cortex is much larger so that a greater degree of conduction velocity tuning would be required to maintain synchrony, produced results suggesting that olivocerebellar conduction velocities are relatively uniform (Aggelopoulos et al. 1995). This allowed small differences in the time of arrival of CF action potentials, suggesting that synchronous activation of inferior olivary neurones would generate Purkinje cell CF responses that were desynchronized by a few milliseconds. Very recently a study of the olivocerebellar system in the turtle has suggested that turtle CFs have a constant conduction velocity in the brainstem, but that the velocity is tuned within the cerebellar cortex, such that Purkinje cells are activated synchronously (Ariel, 2004). However, the responses measured were population field potentials, which may have spread some distance through the cortex, rather than single units. Furthermore, the latencies that were used to assess conduction time were peak latencies of the field potentials. The raw data shown in that study clearly showed variation in onset latencies of the field potentials; the shortest latencies were at sites closest to the brainstem and the longest were at more lateral sites (see Fig. 8B, Ariel, 2004). In all of the experiments in mammals (Sugihara et al. 1993; Aggelopoulos et al. 1995; Lang & Rosenbluth, 2003) electrical stimulation in the ventral medulla was used to activate CFs close to their origin in the inferior olive. Different current strengths and stimulating electrode locations could therefore have activated the olivocerebellar fibres at different sites.
, 百拇医药
In this study we have re-examined the olivo-cerebellar conduction times in climbing fibres projecting to the vermis in the rat, using a method that avoids the need for electrical stimulation in the medulla. Our approach has been to use the CF reflex, in which antidromic activation of inferior olive neurones evoked by stimulation at one site in the cerebellar cortex activates neighbouring inferior olive cells, resulting in orthodromic activation of CFs. This was first described by Eccles and colleagues, using juxta-fastigial stimulation in the cat (Eccles et al. 1966). Electrical coupling between inferior olivary cells mediates the CF reflex (Llinas & Sasaki, 1989), which can occur in the absence of chemical synaptic transmission (Llinas & Yarom, 1981). The reflex activation of local groups of inferior olive neurones generates CF reflex activation of Purkinje cells in a thin sagittally orientated strip of cerebellar cortex aligned to the stimulating electrode. If conduction velocities are tuned to give a fixed conduction time, then the latency of reflex activation should be independent of the location (depth) of the Purkinje cells. If conduction velocities are not tuned then conduction time would vary: the longer the conduction distance, the longer the latency of the reflex.
, 百拇医药
By applying this technique we show that superficially recorded complex spikes are evoked at significantly longer latencies than complex spikes recorded deep within the cerebellar cortex. These findings are not consistent with a tuned CF conduction velocity.
Methods
The experimental rationale is illustrated in Fig. 1. A CF reflex set up by antidromic activation of fibres from one lobule will involve an antidromic conduction delay that traverses a given distance (AD in Fig. 1) which should involve a fixed time. Purkinje cells located at different depths within the cortex will have different orthodromic conduction paths (e.g. in Fig. 1, x for deep units and x+y for superficial units). If there is a uniform conduction time in the olivocerebellar system then all CF reflex responses should have similar latencies, irrespective of depth below the cerebellar surface. In Fig. 1, the time taken for an action potential to travel the distance AD +x for a deep Purkinje cell must equal the time taken to travel the distance AD +x+y for a superficially located Purkinje cell. If the olivocerebellar conduction time is not tuned, then the latencies of neurones recorded deeper within the cortex should have shorter latencies.
, 百拇医药
The drawing shows two cerebellar folia innervated by inferior olivary neurones that are coupled via gap junctions. A stimulating electrode in the cerebellar lobule on the left will antidromically activate some olivary neurones which, through electrical coupling, produces orthodromic activity in the CF projections to the folium on the right. If the olivocerebellar conduction time is tuned to compensate for conduction distance, then the reflex activation of Purkinje cells should occur at a similar time, regardless of the location of the Purkinje cell (e.g. the time taken for an action potential to travel the distance (AD +x) to a deep Purkinje cell should be same as the time taken to travel the distance (AD +x+y) to a superficially located Purkinje cell). If the olivocerebellar conduction time is not tuned to compensate for conduction distance, then the reflex activation of Purkinje cells should not depend on depth.
, 百拇医药
Experiments were performed on 13 female rats weighing between 0.25 kg and 0.3 kg, which were anaesthetized via an intraperitoneal injection of urethane (1 g kg–1), with supplemental doses of barbiturate (Sagatal, 10 mg kg–1, I.V.) to maintain deep anaesthesia, as indicated by the absence of limb withdrawal reflexes. All experimental procedures were performed under UK Home Office regulations and were approved by the local Ethical Committees. A cannula in the femoral vein allowed intravenous administration of fluids and anaesthetics. The airway was secured via a tracheal cannula. Core temperature was maintained at 37°C with a heating blanket. The head was fixed in a stereotaxic frame and the posterior lobe of the cerebellum exposed by removing part of the occipital bone. The dura mater over the vermis was opened and an agar pool created around the exposure and filled with mineral oil to insulate the cortical surface and prevent drying.
, 百拇医药
Stimulation and recording
A fine monopolar stainless steel stimulating electrode (60–110 k) was inserted perpendicularly into the white matter of one lobule of the cerebellar cortex, usually lobule VI b or c, to a depth of between 0.5 and 1 mm below the surface. Single stimuli delivered through this electrode (0.2 ms, current intensities between 50 and 600 μA) evoked CF reflex responses. Penetrations were made perpendicular to the surface of the cerebellar folium under investigation (angle 30–40 deg, tip rostral) and the depths of single unit recordings below the cerebellar surface were measured from the microdrive (Burleigh Inchworm; EXFO Life Sciences Group, Mississauga, Ontario, Canada). Recordings of CF responses were made using glass micropipettes (1–5 M), filled with 1 M potassium citrate, initially as climbing fibre field potentials, which were used to locate the sagittal strip of cortex in which climbing fibre reflex responses were recorded, then as single unit Purkinje cell complex spikes, for measurement of latencies. Signals were recorded using an Axoprobe-1A microelectrode amplifier (Axon Instruments, Union City, CA, USA), amplified (x1000), filtered with a low frequency cut of between 100 Hz and 1 kHz and a high frequency cut 10 kHz, digitized at 25 kHz and sampled (CED 1401, Cambridge Electronic Design, Cambridge, UK), as well as being recorded to digital audio tape.
, http://www.100md.com
Identification of climbing fibre reflex responses
Purkinje cells were identified by the presence of both climbing fibre evoked complex spikes and the more frequent simple spikes. Climbing fibre responses were identified by their typical spike shape (Thach, 1968; Lang & Rosenbluth, 2003), infrequent occurrence and pauses in simple spike firing that followed them. Confirmation that responses were climbing fibre mediated was obtained by stimulating antidromically with paired pulses (Armstrong & Harvey, 1968): inferior olive cells are refractory to antidromic activation 30–50 ms after antidromic activation, so a stimulus at this time does not evoke a reflex. Temporal jitter was also assessed in the evoked responses. This has previously been shown to be of the order of 0.5 ms for CF reflex evoked responses in the cat (Eccles et al. 1967). Units with minimal or no jitter were assumed to be evoked via an axon reflex (Armstrong et al. 1973) and were therefore excluded. Such units also had short latencies (< 4 ms).
, 百拇医药
Analysis
Response latencies were measured from the onset of the stimulus artefact to the onset of the initial fast spike component of the complex spike. Responses to multiple stimulus presentations (at least 20) were analysed in order to estimate the minimum response latency and its variability. Linear regression analysis was used to estimate the conduction velocity of CFs encountered during a single penetration. Grouped data for all penetrations were used to determine the correlation coefficient (coefficient of determination, or adjusted R2 statistic) and F-test for the data ascertained.
, http://www.100md.com
Results
A total of 33 unit recordings of CF reflex responses were made. Example records are illustrated in Fig. 2. The CF responses had characteristic, complex shapes. When they occurred spontaneously they were typically followed by a pause in simple spike activity. In some cases the CF responses showed multiple spikes but more usually in our recordings Purkinje cell complex spikes comprised an initial fast negative spike followed by a large, relatively long duration positive potential that included smaller longer duration elements, whereas simple spikes consisted of only the fast negative spike followed by a brief positive component (as has been described before, e.g. Lang & Rosenbluth, 2003). This is shown in the unit of Fig. 2, recorded at a depth of 2.02 mm: compare the evoked complex spikes, which have large secondary components, with the simple spikes that are visible earlier in the record (arrows). Paired pulse stimulation was used to confirm that the climbing fibre responses resulted from the CF reflex (see Methods). Latencies of the CF reflex responses ranged from 4.3 to 11.3 ms in different Purkinje cells. Although the CF reflex responses were consistently evoked, a small amount of jitter was observed in the recorded onset latencies, as can be seen in the records in Fig. 2. The mean jitter was 0.32 ± 0.21 ms (mean ±S.D.).
, 百拇医药
Climbing fibre reflex responses recorded from a Purkinje cell (2.02 mm deep). Stimulation of a more rostral folium (130 μA) evoked CF responses at a minimal latency of 8.4 ms. As previously described, these show some jitter in latency and sometimes fail (lowermost trace). The climbing fibre responses show a complex form, and can be compared to the simple spikes of the same neurone (arrows).
Relation of recording depth to CF reflex latency
, 百拇医药
A general finding in all experiments was that superficially recorded units produced longer latency CF reflex responses than those recorded deep within the cerebellar cortex.
A specific aim of each experiment was to record CF reflex responses in several Purkinje cells at different depths in single penetrations, all evoked with the same stimulus location and strength, as illustrated in Fig. 1. Responses from experiments in which more than one Purkinje cells were recorded at different depths in a single electrode track are shown in Fig. 3. The insets show the entry points for stimulating electrodes (which were in lobule VI or VII) and the recording electrodes (which were in lobules VIII and IX) in the two experiments where three Purkinje cells were recorded in one track. The graph plots the CF reflex latency against depth of recording for two tracks in which three Purkinje cells were recorded and for three other tracks in which two Purkinje cells were recorded in one track. These illustrate the consistent finding that the most superficially recorded Purkinje cells had the longest latency reflex responses, the deeper units having shorter reflex response latencies. There is a significant inverse relationship between recording depth and response latency (adjusted R2 values for each graph > 0.987, F-statistics for each regression line were significant P < 0.05). Grouped data across the experiments are summarized in Fig. 4, which demonstrates that there is a strong correlation between the depth at which a Purkinje cell was recorded and its CF reflex latency. Superficially recorded cells (< 1 mm) had CF latencies greater than 6 ms, whereas about half of those recorded at depths greater than 1 mm had latencies shorter than 6 ms. This inverse relationship is clear despite the fact that responses were pooled from different experiments (i.e. the stimulation sites and intensities were different so that the ‘AD’ component as illustrated in Fig. 1 was not constant). The pooling may contribute to the apparent scatter of latencies at each depth. Nevertheless there is a significant inverse relationship between depth and latency (t statistic, P < 0.001; F= 31.2, P < 0.001). The mean regression coefficient (–1.724) suggests a conduction velocity of the climbing fibre terminal branches in the cortex of approximately 0.6 m s–1.
, 百拇医药
This shows data from experiments in which more than one Purkinje cell with CF reflex activation from a single stimulus site could be recorded in one electrode penetration. In two tracks three cells could be recorded, the relative positions of the stimulating electrode (filled circle) and the entry point of the recording electrode (unfilled circle) are shown on the diagrams to the left. The plot on the right shows the relationship between latencies of the CF reflexes and recording depths for these tracks (filled diamonds and filled circles), as well as data from 3 other penetrations in which 2 Purkinje cells were found (filled squares).
, 百拇医药
The scatterplot shows latencies of CF reflex responses against depth of recording. There is a significant relationship: the depth of the recorded unit was significantly predictive of the latency of the recorded unit (t statistic, P < 0.001; F= 31.2, P < 0.001). Least squares regression analysis revealed an inverse relationship (y= 8.635 – 1.724x). The continuous line through the data points represents the least squares regression; the dashed lines delimit the prediction interval and the dotted lines the 95% confidence interval (n= 33). These data were obtained from 11 different experiments, which may explain some of the variation in latency at each depth.
, 百拇医药
Discussion
The results show a statistically significant inverse relationship between Purkinje cell depth in the cerebellar cortex and the latency of climbing fibre reflex responses in rats. This is not consistent with a precisely tuned conduction velocity to give a constant conduction delay.
We do not know exactly where action potentials were initiated by the stimuli in the climbing fibres in these experiments, as in experiments with electrical activation of the climbing fibres in the brainstem. The stimuli that set up the CF reflex are complex in that individual climbing fibres may have multiple terminal branches in different folia or across a folium (e.g. Sugihara et al. 2001). The reflex may thus be set up by antidromic activation of a combination of stem axons and terminal branches. Depending on the extent of electrical coupling in the inferior olive, this may introduce some variance in the conduction time of the antidromic limb of the reflex (AD in Fig. 1) in different Purkinje cells. If electrical coupling is strong then the antidromic activation from the stem axons should be dominant. If the coupling is less widespread then the reflex latencies in individual cells may vary dependent on whether stem axons or terminal branches activated the reflex. However, the experiment was designed only to address the issue of uniform conduction velocity and the conclusions do not depend on the site of activation of the reflex: all of the responses involved an antidromic and an orthodromic limb and if the conduction time were tuned to give a uniform conduction time, then the orthodromic limb should have a similar latency regardless of the depth of the Purkinje cell in which the reflex was recorded. The consistent finding, in both individual experiments and in the grouped data, was that CF reflex latency was longer in more superficially recorded Purkinje cells and briefer in deeper Purkinje cells. If olivocerebellar conduction time were tuned to climbing fibre length then our results could only be obtained if the stimulus somehow activated superficially projecting (i.e. faster conducting) climbing fibres at longer latency than deeper projecting (i.e. slower conducting) climbing fibres. This could not occur from a fixed stimulus location and intensity.
, http://www.100md.com
Our results depend on the CF reflex responses representing climbing fibre activation of Purkinje cells. The spike form and the presence of characteristic climbing fibre discharge pauses in Purkinje cell simple spike activity were consistent with this, the climbing fibre responses being characterized by a large positive potential following the initial fast spike (Lang & Rosenbluth, 2003). The restriction of the responses to a zone in the plane of the stimulating electrode is as expected for mediation from climbing fibre zonation in the cerebellar cortex (Apps & Garwicz, 2005): if they had been mediated via the direct activation of mossy fibres, a larger effective recording area should have been observed, and characteristic climbing fibre responses would not have been seen. Recordings were made in tracks 2 or 3 surface folia from the stimulating electrode, so it is unlikely that mossy fibres were directly activated at such a distance. There is little evidence for synaptic coupling between mossy fibres outside one lobule. In general the stimulating electrode was positioned in lobule 6 and the recording electrode in lobule 7 or 8; the fissures between these lobules are deep, limiting any possibility of current spread from stimulation site to recording site. The reflex responses had some variability in latency (jitter), comparable to the 0.5 ms variation previously reported for the CF reflex in the cat (Eccles et al. 1967). This may in part be a consequence of the transmission through gap junction connections and the threshold properties of inferior olive cells. In addition, there was a strong depression of the responses 30–50 ms after a stimulus, as described for the CF reflex in the cat (Armstrong et al. 1973).
, http://www.100md.com
The range of latencies we found for the CF reflex is consistent with previous reports of olivocerebellar conduction time (Sugihara et al. 1993; Aggelopoulos et al. 1995; Lang & Rosenbluth, 2003), and includes both the time taken for action potentials to travel antidromically from the point of activation to the inferior olive and the time for an orthodromically propagated action potential to travel from the inferior olive to the Purkinje cell from which the recording was made, plus the inferior olivary transit time. Therefore considering a range of latencies of between 2 and 5 ms for conduction between inferior olive and cerebellar cortex for the rat (Sugihara et al. 1993; Lang & Rosenbluth, 2003), a range of approximately 4–11 ms for CF reflex evoked responses could be expected.
, 百拇医药
In all cases in which an electrode penetration sampled single units of the same sagittal zone at different depths, the trend for deeper recordings to have shorter reflex activation latencies was seen (Fig. 3). Calculations from the mean of the regression coefficients obtained from the single penetration data and from the grouped data suggest a mean conduction velocity for the climbing fibres within the cerebellar cortex of approximately 0.6 m s–1. This estimation is substantially lower than the conduction velocities reported for the entire olivocerebellar pathway (2.37–4.22 m s–1, Sugihara et al. 1993), most of which is located in the brainstem. This indicates that there can be a very considerable slowing of conduction velocity in the branches of climbing fibres to the more superficial parts of the cerebellar cortex. Anatomical studies have shown that within the cortex many climbing fibre collaterals have diameters averaging less than 1 μm (Sugihara et al. 1993, 1999). However, we specifically sought recordings from the most superficial layer of the cortex and these were those with the longest CF reflex latencies; four units had reflex activation latencies around 10 ms and three of these were located in the first Purkinje cell layer encountered (< 0.5 mm deep). It may be that the activation times of these units were particularly long. Since the CF reflex activation was only found in Purkinje cells in a narrow sagittal zone of cortex, in the same plane as the stimulating electrode, we were unable to target activated units in the deepest parts of the fissures.
, 百拇医药
Functional significance
These results indicate that there is a significant inverse relationship between recording depth and latency of climbing fibre reflex responses evoked in the vermis in the rats. This is not consistent with a conduction velocity in climbing fibres that is precisely tuned to give a fixed conduction time, but is more in keeping with a conduction time that varies such that more superficial Purkinje cells are likely to activated later than deep Purkinje cells. However, the difference between the timing of arrival of climbing fibre activity between the most superficial parts of the folia and deeper levels is of the order of a few milliseconds. This suggests that the function of climbing fibres does not require timing precision at a submillisecond level, but that differences of a few milliseconds exist. These data do not argue directly against the importance of synchronous climbing fibre activity, but do argue against a submillisecond precision of the synchrony.
, 百拇医药
References
Aggelopoulos NC, Duke C & Edgley SA (1995). Non-uniform conduction time in the olivo-cerebellar pathway in the anaesthetised cat. J Physiol 486, 763–768.
Apps R & Garwicz M (2005). Anatomical and physiological foundations of cerebellar information processing. Nat Rev Neurosci 6, 297–311.
Ariel M (2004). Latencies of climbing fiber inputs to turtle cerebellar cortex. J Neurophysiol 93, 1042–1054.
, http://www.100md.com
Armstrong DM & Harvey RJ (1968). Responses to a spino-olivo-cerebellar pathway in the cat. J Physiol 194, 147–168.
Armstrong DM, Harvey RJ & Schild RF (1973). Cerebello-cerebellar responses mediated via climbing fibres. Exp Brain Res 18, 19–39.
Eccles JC, Ito M & Szentagothai J (1967). The Cerebellum as a Neuronal Machine. Springer-Verlag, Berlin.
Eccles JC, Llinas R & Sasaki K (1966). The excitatory synaptic action of climbing fibres on the Purkinje cells of the cerebellum. J Physiol 182, 268–296.
, http://www.100md.com
Kitazawa S & Wolpert DM (2005). Rhythmicity, randomness and synchrony in climbing fiber signals. Trends Neurosci 28, 611–619.
Lang EJ (2003). Excitatory afferent modulation of complex spike synchrony. Cerebellum 2, 165–170.
Lang EJ & Rosenbluth J (2003). Role of myelination in the development of a uniform olivocerebellar conduction time. J Neurophysiol 89, 2259–2270.
Llinas R & Sasaki K (1989). The Functional Organization of the Olivo-Cerebellar System as Examined by Multiple Purkinje Cell Recordings. Eur J Neurosci 1, 587–602.
, 百拇医药
Llinas R & Welsh JP (1993). On the cerebellum and motor learning. Curr Opin Neurobiol 3, 958–965.
Llinas R & Yarom Y (1981). Electrophysiology of mammalian inferior olivary neurones in vitro. Different types of voltage-dependent ionic conductances. J Physiol 315, 549–567.
Sasaki K, Bower JM & Llinas R (1989). Multiple Purkinje cell recording in rodent cerebellar cortex. Eur J Neurosci 1, 572–586.
, http://www.100md.com
Sugihara I, Lang EJ & Llinas R (1993). Uniform olivo-cerebellar conduction time underlies Purkinje cell complex spike synchronicity in the rat cerebellum. J Physiol 470, 243–271.
Sugihara I, Wu H-S & Shinoda Y (1999). Morphology of single olivocerebellar. axons labeled with biotinylated dextran amine in the rat. J Comp Neurol 414, 131–148.
Sugihara I, Wu H-S & Shinoda Y (2001). The entire trajectories of single olivocerebellar axons in the cerebellar cortex and their contribution to cerebellar compartmentalization. J Neurosci 21, 7715–7723.
, 百拇医药
Thach WT (1968). Discharge of Purkinje and cerebellar nuclear neurons during rapidly alternating arm movements in the monkey. J Neurophysiol 31, 785–797.
Welsh JP, Lang EJ, Sugihara I & Llinas R (1995). Dynamic organization of motor control within the olivocerebellar system. Nature 374, 453–457.
Welsh JP & Llinas R (1997). Some organizing principles for the control of movement based on olivocerebellar physiology. Prog Brain Res 114, 449–461., 百拇医药(M. R. Baker and S. A. Edgley)
Abstract
It has been proposed that the conduction velocities of cerebellar climbing fibre (olivocerebellar) axons are tuned according to length, in order to precisely fix the conduction time between the inferior olive and cerebellar cortex. Some data conflict with this view. We have re-evaluated this issue using the climbing fibre reflex. The white matter of the tip of one folium in lobule VI or VII was stimulated electrically 0.5–1 mm below the surface and recordings were made from Purkinje cells in lobules VIII and IX. Reflex evoked climbing fibre (CF) responses (33 units) were recorded at different depths from Purkinje cells found in a narrow sagittal zone of cortex as complex spikes. The responses had latencies ranging from 4.3 ms to 11.3 ms. A consistent trend was that Purkinje cell responses recorded at greater depth had shorter CF reflex latencies than those recorded more superficially, both in individual experiments and in grouped data. These data show that the CF reflex latency is not constant, but is directly proportional to the distance an action potential has to travel along a CF. These data are not consistent with tuning of CF conduction velocities to normalize olivocerebellar conduction time, but are consistent with a CF conduction velocity in the cortex of approximately 0.6 m s–1. This suggests that climbing fibres projecting to different parts of the cerebellar cortex may have differences in spike conduction time of a few milliseconds, and that submillisecond precision is not an important element of the climbing fibre signal.
, 百拇医药
Introduction
Precise timing of neural signals is essential to the function of any neural network; for instance learning rules usually require the temporal coincidence of more than one synaptic input. Most models of cerebellar function assume that the timing of climbing fibre (CF) activation is important (for reviews see Lang, 2003; Kitazawa & Wolpert, 2005), but the precision of this timing is not established. One particular view, based on the remarkable properties of the inferior olivary nucleus, argues that synchronous activity in the olivo-cerebellar system provides precise timing signals essential to the generation of movements (Llinas & Welsh, 1993; Welsh et al. 1995; Llinas, 1997). In support of this idea, multiple single unit recordings have shown synchronous CF activity within small areas of the cerebellar cortex (Sasaki et al. 1989; Lang, 2003). The cerebellar cortex is deeply folded and CF axons terminating in different regions of the cerebellum have different lengths. If all CF axons had similar conduction velocities then activity that is synchronous at the level of the inferior olive will become desynchronized at the level of the cortex, the longer conduction path having a longer conduction time. It has been proposed that CF conduction velocity is tuned such that longer CFs, which project to more superficial parts of the cerebellar folia, transmit action potentials with a conduction velocity that is greater than that of shorter CFs that terminate deeper in the cortex (Sugihara et al. 1993; Lang & Rosenbluth, 2003). Under this scheme, despite the foliation of the cerebellum and different lengths of climbing fibres, synchronous activity at the level of the inferior olive is maintained in the CF activation times of Purkinje cells (Sugihara et al. 1993; Lang & Rosenbluth, 2003).
, 百拇医药
Experiments seeking evidence for the same phenomenon in the cat, in which the cerebellar cortex is much larger so that a greater degree of conduction velocity tuning would be required to maintain synchrony, produced results suggesting that olivocerebellar conduction velocities are relatively uniform (Aggelopoulos et al. 1995). This allowed small differences in the time of arrival of CF action potentials, suggesting that synchronous activation of inferior olivary neurones would generate Purkinje cell CF responses that were desynchronized by a few milliseconds. Very recently a study of the olivocerebellar system in the turtle has suggested that turtle CFs have a constant conduction velocity in the brainstem, but that the velocity is tuned within the cerebellar cortex, such that Purkinje cells are activated synchronously (Ariel, 2004). However, the responses measured were population field potentials, which may have spread some distance through the cortex, rather than single units. Furthermore, the latencies that were used to assess conduction time were peak latencies of the field potentials. The raw data shown in that study clearly showed variation in onset latencies of the field potentials; the shortest latencies were at sites closest to the brainstem and the longest were at more lateral sites (see Fig. 8B, Ariel, 2004). In all of the experiments in mammals (Sugihara et al. 1993; Aggelopoulos et al. 1995; Lang & Rosenbluth, 2003) electrical stimulation in the ventral medulla was used to activate CFs close to their origin in the inferior olive. Different current strengths and stimulating electrode locations could therefore have activated the olivocerebellar fibres at different sites.
, 百拇医药
In this study we have re-examined the olivo-cerebellar conduction times in climbing fibres projecting to the vermis in the rat, using a method that avoids the need for electrical stimulation in the medulla. Our approach has been to use the CF reflex, in which antidromic activation of inferior olive neurones evoked by stimulation at one site in the cerebellar cortex activates neighbouring inferior olive cells, resulting in orthodromic activation of CFs. This was first described by Eccles and colleagues, using juxta-fastigial stimulation in the cat (Eccles et al. 1966). Electrical coupling between inferior olivary cells mediates the CF reflex (Llinas & Sasaki, 1989), which can occur in the absence of chemical synaptic transmission (Llinas & Yarom, 1981). The reflex activation of local groups of inferior olive neurones generates CF reflex activation of Purkinje cells in a thin sagittally orientated strip of cerebellar cortex aligned to the stimulating electrode. If conduction velocities are tuned to give a fixed conduction time, then the latency of reflex activation should be independent of the location (depth) of the Purkinje cells. If conduction velocities are not tuned then conduction time would vary: the longer the conduction distance, the longer the latency of the reflex.
, 百拇医药
By applying this technique we show that superficially recorded complex spikes are evoked at significantly longer latencies than complex spikes recorded deep within the cerebellar cortex. These findings are not consistent with a tuned CF conduction velocity.
Methods
The experimental rationale is illustrated in Fig. 1. A CF reflex set up by antidromic activation of fibres from one lobule will involve an antidromic conduction delay that traverses a given distance (AD in Fig. 1) which should involve a fixed time. Purkinje cells located at different depths within the cortex will have different orthodromic conduction paths (e.g. in Fig. 1, x for deep units and x+y for superficial units). If there is a uniform conduction time in the olivocerebellar system then all CF reflex responses should have similar latencies, irrespective of depth below the cerebellar surface. In Fig. 1, the time taken for an action potential to travel the distance AD +x for a deep Purkinje cell must equal the time taken to travel the distance AD +x+y for a superficially located Purkinje cell. If the olivocerebellar conduction time is not tuned, then the latencies of neurones recorded deeper within the cortex should have shorter latencies.
, 百拇医药
The drawing shows two cerebellar folia innervated by inferior olivary neurones that are coupled via gap junctions. A stimulating electrode in the cerebellar lobule on the left will antidromically activate some olivary neurones which, through electrical coupling, produces orthodromic activity in the CF projections to the folium on the right. If the olivocerebellar conduction time is tuned to compensate for conduction distance, then the reflex activation of Purkinje cells should occur at a similar time, regardless of the location of the Purkinje cell (e.g. the time taken for an action potential to travel the distance (AD +x) to a deep Purkinje cell should be same as the time taken to travel the distance (AD +x+y) to a superficially located Purkinje cell). If the olivocerebellar conduction time is not tuned to compensate for conduction distance, then the reflex activation of Purkinje cells should not depend on depth.
, 百拇医药
Experiments were performed on 13 female rats weighing between 0.25 kg and 0.3 kg, which were anaesthetized via an intraperitoneal injection of urethane (1 g kg–1), with supplemental doses of barbiturate (Sagatal, 10 mg kg–1, I.V.) to maintain deep anaesthesia, as indicated by the absence of limb withdrawal reflexes. All experimental procedures were performed under UK Home Office regulations and were approved by the local Ethical Committees. A cannula in the femoral vein allowed intravenous administration of fluids and anaesthetics. The airway was secured via a tracheal cannula. Core temperature was maintained at 37°C with a heating blanket. The head was fixed in a stereotaxic frame and the posterior lobe of the cerebellum exposed by removing part of the occipital bone. The dura mater over the vermis was opened and an agar pool created around the exposure and filled with mineral oil to insulate the cortical surface and prevent drying.
, 百拇医药
Stimulation and recording
A fine monopolar stainless steel stimulating electrode (60–110 k) was inserted perpendicularly into the white matter of one lobule of the cerebellar cortex, usually lobule VI b or c, to a depth of between 0.5 and 1 mm below the surface. Single stimuli delivered through this electrode (0.2 ms, current intensities between 50 and 600 μA) evoked CF reflex responses. Penetrations were made perpendicular to the surface of the cerebellar folium under investigation (angle 30–40 deg, tip rostral) and the depths of single unit recordings below the cerebellar surface were measured from the microdrive (Burleigh Inchworm; EXFO Life Sciences Group, Mississauga, Ontario, Canada). Recordings of CF responses were made using glass micropipettes (1–5 M), filled with 1 M potassium citrate, initially as climbing fibre field potentials, which were used to locate the sagittal strip of cortex in which climbing fibre reflex responses were recorded, then as single unit Purkinje cell complex spikes, for measurement of latencies. Signals were recorded using an Axoprobe-1A microelectrode amplifier (Axon Instruments, Union City, CA, USA), amplified (x1000), filtered with a low frequency cut of between 100 Hz and 1 kHz and a high frequency cut 10 kHz, digitized at 25 kHz and sampled (CED 1401, Cambridge Electronic Design, Cambridge, UK), as well as being recorded to digital audio tape.
, http://www.100md.com
Identification of climbing fibre reflex responses
Purkinje cells were identified by the presence of both climbing fibre evoked complex spikes and the more frequent simple spikes. Climbing fibre responses were identified by their typical spike shape (Thach, 1968; Lang & Rosenbluth, 2003), infrequent occurrence and pauses in simple spike firing that followed them. Confirmation that responses were climbing fibre mediated was obtained by stimulating antidromically with paired pulses (Armstrong & Harvey, 1968): inferior olive cells are refractory to antidromic activation 30–50 ms after antidromic activation, so a stimulus at this time does not evoke a reflex. Temporal jitter was also assessed in the evoked responses. This has previously been shown to be of the order of 0.5 ms for CF reflex evoked responses in the cat (Eccles et al. 1967). Units with minimal or no jitter were assumed to be evoked via an axon reflex (Armstrong et al. 1973) and were therefore excluded. Such units also had short latencies (< 4 ms).
, 百拇医药
Analysis
Response latencies were measured from the onset of the stimulus artefact to the onset of the initial fast spike component of the complex spike. Responses to multiple stimulus presentations (at least 20) were analysed in order to estimate the minimum response latency and its variability. Linear regression analysis was used to estimate the conduction velocity of CFs encountered during a single penetration. Grouped data for all penetrations were used to determine the correlation coefficient (coefficient of determination, or adjusted R2 statistic) and F-test for the data ascertained.
, http://www.100md.com
Results
A total of 33 unit recordings of CF reflex responses were made. Example records are illustrated in Fig. 2. The CF responses had characteristic, complex shapes. When they occurred spontaneously they were typically followed by a pause in simple spike activity. In some cases the CF responses showed multiple spikes but more usually in our recordings Purkinje cell complex spikes comprised an initial fast negative spike followed by a large, relatively long duration positive potential that included smaller longer duration elements, whereas simple spikes consisted of only the fast negative spike followed by a brief positive component (as has been described before, e.g. Lang & Rosenbluth, 2003). This is shown in the unit of Fig. 2, recorded at a depth of 2.02 mm: compare the evoked complex spikes, which have large secondary components, with the simple spikes that are visible earlier in the record (arrows). Paired pulse stimulation was used to confirm that the climbing fibre responses resulted from the CF reflex (see Methods). Latencies of the CF reflex responses ranged from 4.3 to 11.3 ms in different Purkinje cells. Although the CF reflex responses were consistently evoked, a small amount of jitter was observed in the recorded onset latencies, as can be seen in the records in Fig. 2. The mean jitter was 0.32 ± 0.21 ms (mean ±S.D.).
, 百拇医药
Climbing fibre reflex responses recorded from a Purkinje cell (2.02 mm deep). Stimulation of a more rostral folium (130 μA) evoked CF responses at a minimal latency of 8.4 ms. As previously described, these show some jitter in latency and sometimes fail (lowermost trace). The climbing fibre responses show a complex form, and can be compared to the simple spikes of the same neurone (arrows).
Relation of recording depth to CF reflex latency
, 百拇医药
A general finding in all experiments was that superficially recorded units produced longer latency CF reflex responses than those recorded deep within the cerebellar cortex.
A specific aim of each experiment was to record CF reflex responses in several Purkinje cells at different depths in single penetrations, all evoked with the same stimulus location and strength, as illustrated in Fig. 1. Responses from experiments in which more than one Purkinje cells were recorded at different depths in a single electrode track are shown in Fig. 3. The insets show the entry points for stimulating electrodes (which were in lobule VI or VII) and the recording electrodes (which were in lobules VIII and IX) in the two experiments where three Purkinje cells were recorded in one track. The graph plots the CF reflex latency against depth of recording for two tracks in which three Purkinje cells were recorded and for three other tracks in which two Purkinje cells were recorded in one track. These illustrate the consistent finding that the most superficially recorded Purkinje cells had the longest latency reflex responses, the deeper units having shorter reflex response latencies. There is a significant inverse relationship between recording depth and response latency (adjusted R2 values for each graph > 0.987, F-statistics for each regression line were significant P < 0.05). Grouped data across the experiments are summarized in Fig. 4, which demonstrates that there is a strong correlation between the depth at which a Purkinje cell was recorded and its CF reflex latency. Superficially recorded cells (< 1 mm) had CF latencies greater than 6 ms, whereas about half of those recorded at depths greater than 1 mm had latencies shorter than 6 ms. This inverse relationship is clear despite the fact that responses were pooled from different experiments (i.e. the stimulation sites and intensities were different so that the ‘AD’ component as illustrated in Fig. 1 was not constant). The pooling may contribute to the apparent scatter of latencies at each depth. Nevertheless there is a significant inverse relationship between depth and latency (t statistic, P < 0.001; F= 31.2, P < 0.001). The mean regression coefficient (–1.724) suggests a conduction velocity of the climbing fibre terminal branches in the cortex of approximately 0.6 m s–1.
, 百拇医药
This shows data from experiments in which more than one Purkinje cell with CF reflex activation from a single stimulus site could be recorded in one electrode penetration. In two tracks three cells could be recorded, the relative positions of the stimulating electrode (filled circle) and the entry point of the recording electrode (unfilled circle) are shown on the diagrams to the left. The plot on the right shows the relationship between latencies of the CF reflexes and recording depths for these tracks (filled diamonds and filled circles), as well as data from 3 other penetrations in which 2 Purkinje cells were found (filled squares).
, 百拇医药
The scatterplot shows latencies of CF reflex responses against depth of recording. There is a significant relationship: the depth of the recorded unit was significantly predictive of the latency of the recorded unit (t statistic, P < 0.001; F= 31.2, P < 0.001). Least squares regression analysis revealed an inverse relationship (y= 8.635 – 1.724x). The continuous line through the data points represents the least squares regression; the dashed lines delimit the prediction interval and the dotted lines the 95% confidence interval (n= 33). These data were obtained from 11 different experiments, which may explain some of the variation in latency at each depth.
, 百拇医药
Discussion
The results show a statistically significant inverse relationship between Purkinje cell depth in the cerebellar cortex and the latency of climbing fibre reflex responses in rats. This is not consistent with a precisely tuned conduction velocity to give a constant conduction delay.
We do not know exactly where action potentials were initiated by the stimuli in the climbing fibres in these experiments, as in experiments with electrical activation of the climbing fibres in the brainstem. The stimuli that set up the CF reflex are complex in that individual climbing fibres may have multiple terminal branches in different folia or across a folium (e.g. Sugihara et al. 2001). The reflex may thus be set up by antidromic activation of a combination of stem axons and terminal branches. Depending on the extent of electrical coupling in the inferior olive, this may introduce some variance in the conduction time of the antidromic limb of the reflex (AD in Fig. 1) in different Purkinje cells. If electrical coupling is strong then the antidromic activation from the stem axons should be dominant. If the coupling is less widespread then the reflex latencies in individual cells may vary dependent on whether stem axons or terminal branches activated the reflex. However, the experiment was designed only to address the issue of uniform conduction velocity and the conclusions do not depend on the site of activation of the reflex: all of the responses involved an antidromic and an orthodromic limb and if the conduction time were tuned to give a uniform conduction time, then the orthodromic limb should have a similar latency regardless of the depth of the Purkinje cell in which the reflex was recorded. The consistent finding, in both individual experiments and in the grouped data, was that CF reflex latency was longer in more superficially recorded Purkinje cells and briefer in deeper Purkinje cells. If olivocerebellar conduction time were tuned to climbing fibre length then our results could only be obtained if the stimulus somehow activated superficially projecting (i.e. faster conducting) climbing fibres at longer latency than deeper projecting (i.e. slower conducting) climbing fibres. This could not occur from a fixed stimulus location and intensity.
, http://www.100md.com
Our results depend on the CF reflex responses representing climbing fibre activation of Purkinje cells. The spike form and the presence of characteristic climbing fibre discharge pauses in Purkinje cell simple spike activity were consistent with this, the climbing fibre responses being characterized by a large positive potential following the initial fast spike (Lang & Rosenbluth, 2003). The restriction of the responses to a zone in the plane of the stimulating electrode is as expected for mediation from climbing fibre zonation in the cerebellar cortex (Apps & Garwicz, 2005): if they had been mediated via the direct activation of mossy fibres, a larger effective recording area should have been observed, and characteristic climbing fibre responses would not have been seen. Recordings were made in tracks 2 or 3 surface folia from the stimulating electrode, so it is unlikely that mossy fibres were directly activated at such a distance. There is little evidence for synaptic coupling between mossy fibres outside one lobule. In general the stimulating electrode was positioned in lobule 6 and the recording electrode in lobule 7 or 8; the fissures between these lobules are deep, limiting any possibility of current spread from stimulation site to recording site. The reflex responses had some variability in latency (jitter), comparable to the 0.5 ms variation previously reported for the CF reflex in the cat (Eccles et al. 1967). This may in part be a consequence of the transmission through gap junction connections and the threshold properties of inferior olive cells. In addition, there was a strong depression of the responses 30–50 ms after a stimulus, as described for the CF reflex in the cat (Armstrong et al. 1973).
, http://www.100md.com
The range of latencies we found for the CF reflex is consistent with previous reports of olivocerebellar conduction time (Sugihara et al. 1993; Aggelopoulos et al. 1995; Lang & Rosenbluth, 2003), and includes both the time taken for action potentials to travel antidromically from the point of activation to the inferior olive and the time for an orthodromically propagated action potential to travel from the inferior olive to the Purkinje cell from which the recording was made, plus the inferior olivary transit time. Therefore considering a range of latencies of between 2 and 5 ms for conduction between inferior olive and cerebellar cortex for the rat (Sugihara et al. 1993; Lang & Rosenbluth, 2003), a range of approximately 4–11 ms for CF reflex evoked responses could be expected.
, 百拇医药
In all cases in which an electrode penetration sampled single units of the same sagittal zone at different depths, the trend for deeper recordings to have shorter reflex activation latencies was seen (Fig. 3). Calculations from the mean of the regression coefficients obtained from the single penetration data and from the grouped data suggest a mean conduction velocity for the climbing fibres within the cerebellar cortex of approximately 0.6 m s–1. This estimation is substantially lower than the conduction velocities reported for the entire olivocerebellar pathway (2.37–4.22 m s–1, Sugihara et al. 1993), most of which is located in the brainstem. This indicates that there can be a very considerable slowing of conduction velocity in the branches of climbing fibres to the more superficial parts of the cerebellar cortex. Anatomical studies have shown that within the cortex many climbing fibre collaterals have diameters averaging less than 1 μm (Sugihara et al. 1993, 1999). However, we specifically sought recordings from the most superficial layer of the cortex and these were those with the longest CF reflex latencies; four units had reflex activation latencies around 10 ms and three of these were located in the first Purkinje cell layer encountered (< 0.5 mm deep). It may be that the activation times of these units were particularly long. Since the CF reflex activation was only found in Purkinje cells in a narrow sagittal zone of cortex, in the same plane as the stimulating electrode, we were unable to target activated units in the deepest parts of the fissures.
, 百拇医药
Functional significance
These results indicate that there is a significant inverse relationship between recording depth and latency of climbing fibre reflex responses evoked in the vermis in the rats. This is not consistent with a conduction velocity in climbing fibres that is precisely tuned to give a fixed conduction time, but is more in keeping with a conduction time that varies such that more superficial Purkinje cells are likely to activated later than deep Purkinje cells. However, the difference between the timing of arrival of climbing fibre activity between the most superficial parts of the folia and deeper levels is of the order of a few milliseconds. This suggests that the function of climbing fibres does not require timing precision at a submillisecond level, but that differences of a few milliseconds exist. These data do not argue directly against the importance of synchronous climbing fibre activity, but do argue against a submillisecond precision of the synchrony.
, 百拇医药
References
Aggelopoulos NC, Duke C & Edgley SA (1995). Non-uniform conduction time in the olivo-cerebellar pathway in the anaesthetised cat. J Physiol 486, 763–768.
Apps R & Garwicz M (2005). Anatomical and physiological foundations of cerebellar information processing. Nat Rev Neurosci 6, 297–311.
Ariel M (2004). Latencies of climbing fiber inputs to turtle cerebellar cortex. J Neurophysiol 93, 1042–1054.
, http://www.100md.com
Armstrong DM & Harvey RJ (1968). Responses to a spino-olivo-cerebellar pathway in the cat. J Physiol 194, 147–168.
Armstrong DM, Harvey RJ & Schild RF (1973). Cerebello-cerebellar responses mediated via climbing fibres. Exp Brain Res 18, 19–39.
Eccles JC, Ito M & Szentagothai J (1967). The Cerebellum as a Neuronal Machine. Springer-Verlag, Berlin.
Eccles JC, Llinas R & Sasaki K (1966). The excitatory synaptic action of climbing fibres on the Purkinje cells of the cerebellum. J Physiol 182, 268–296.
, http://www.100md.com
Kitazawa S & Wolpert DM (2005). Rhythmicity, randomness and synchrony in climbing fiber signals. Trends Neurosci 28, 611–619.
Lang EJ (2003). Excitatory afferent modulation of complex spike synchrony. Cerebellum 2, 165–170.
Lang EJ & Rosenbluth J (2003). Role of myelination in the development of a uniform olivocerebellar conduction time. J Neurophysiol 89, 2259–2270.
Llinas R & Sasaki K (1989). The Functional Organization of the Olivo-Cerebellar System as Examined by Multiple Purkinje Cell Recordings. Eur J Neurosci 1, 587–602.
, 百拇医药
Llinas R & Welsh JP (1993). On the cerebellum and motor learning. Curr Opin Neurobiol 3, 958–965.
Llinas R & Yarom Y (1981). Electrophysiology of mammalian inferior olivary neurones in vitro. Different types of voltage-dependent ionic conductances. J Physiol 315, 549–567.
Sasaki K, Bower JM & Llinas R (1989). Multiple Purkinje cell recording in rodent cerebellar cortex. Eur J Neurosci 1, 572–586.
, http://www.100md.com
Sugihara I, Lang EJ & Llinas R (1993). Uniform olivo-cerebellar conduction time underlies Purkinje cell complex spike synchronicity in the rat cerebellum. J Physiol 470, 243–271.
Sugihara I, Wu H-S & Shinoda Y (1999). Morphology of single olivocerebellar. axons labeled with biotinylated dextran amine in the rat. J Comp Neurol 414, 131–148.
Sugihara I, Wu H-S & Shinoda Y (2001). The entire trajectories of single olivocerebellar axons in the cerebellar cortex and their contribution to cerebellar compartmentalization. J Neurosci 21, 7715–7723.
, 百拇医药
Thach WT (1968). Discharge of Purkinje and cerebellar nuclear neurons during rapidly alternating arm movements in the monkey. J Neurophysiol 31, 785–797.
Welsh JP, Lang EJ, Sugihara I & Llinas R (1995). Dynamic organization of motor control within the olivocerebellar system. Nature 374, 453–457.
Welsh JP & Llinas R (1997). Some organizing principles for the control of movement based on olivocerebellar physiology. Prog Brain Res 114, 449–461., 百拇医药(M. R. Baker and S. A. Edgley)