Notice: Wiley Online Library will be unavailable on Saturday 30th July 2016 from 08:00-11:00 BST / 03:00-06:00 EST / 15:00-18:00 SGT for essential maintenance. Apologies for the inconvenience.
Corresponding author M. Garwicz: Department of Physiology and Neuroscience, Lund University, Sölvegatan 19, S-223 62 Lund, Sweden. Email: firstname.lastname@example.org
1The topographical organization of mossy fibre input to the forelimb area of the paravermal C3 zone in cerebellar lobules IV and V was investigated in barbiturate-anaesthetized cats and compared with the previously described microzonal organization of climbing fibre input to the same part of the cortex. Recordings were made in the Purkinje cell and granule cell layers from single climbing fibre and mossy fibre units, respectively, and the organization of cutaneous receptive fields was assessed for both types of afferents.
2Based on spatial characteristics, receptive fields of single mossy fibres could be systematized into ten classes and a total of thirty-two subclasses, mainly in accordance with a scheme previously used for classification of climbing fibres. Different mossy fibres displayed a substantial range of sensitivity to natural peripheral stimulation, responded preferentially to phasic or tonic stimuli and were activated by brushing of hairs or light tapping of the skin.
3Overall, mossy fibres to any given microzone had receptive fields resembling the climbing fibre receptive field defining that microzone. However, compared with the climbing fibre input, the mossy fibre input had a more intricate topographical organization. Mossy fibres with very similar receptive fields projected to circumscribed cortical regions, with a specific termination not only in the mediolateral, but also in some cases in the rostrocaudal and dorsoventral, dimensions of the zone. On the other hand, mossy fibre units with non-identical, albeit usually similar, receptive fields were frequently found in the same microelectrode track.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
The cerebellum receives two main types of afferents, the olivocerebellar climbing fibres and the mossy fibres (for references see Eccles et al. 1967). Both theoretical and experimental accounts of cerebellar function suggest that the main throughput of the cortex is due to activity in mossy fibres, contacting Purkinje cells via the granule cell- parallel fibre route (for references see Ito, 1984). The key role of the climbing fibres would be in regulating this throughput by employing various mechanisms of modulatory control, ranging from short term changes of Purkinje cell excitability to induction of long term plasticity in the parallel fibre-Purkinje cell synapses (see Simpson et al. 1996).
Several mossy fibre systems have been characterized with respect to the type of information that they transmit (Oscarsson, 1973). Some pathways to the cerebellar anterior lobe, such as the dorsal spinocerebellar tract and the cuneocerebellar tract, appear to convey information predominantly representing the state of the motor apparatus, as encoded by the activity of peripheral receptors. Accordingly, mossy fibres to the intermediate cerebellum of the monkey have been shown to provide position, velocity and direction information about movement of individual joints in the ipsilateral forelimb (van Kan et al. 1993). By contrast, other mossy fibre systems seem to mediate different types of efference copy signals, or rather a combination of input from the periphery and efference copy signals, from structures at various levels in the motor system hierarchy (e.g. Arshavsky et al. 1986).
Little is known about how input from the different mossy fibre systems is integrated by the cerebellar cortex and the cerebellar nuclei and how it is transformed into the efferent cerebellar output which is so crucial for the co-ordination of movement. However, given the presumed roles of mossy fibres and climbing fibres, the functional relationship and interaction between these two inputs must be of crucial importance for the sensorimotor transformations carried out by the cerebellum. Therefore, a detailed description of mossy fibre input topography and its relationship to the organization of climbing fibre input lies at the core of understanding the mode of operation of cerebellar circuitry.
In the cat, the cerebellar anterior lobe is divisible on each side into nine sagittally oriented cortical zones, defined by their climbing fibre input from circumscribed regions of the inferior olive and their efferent projections to circumscribed regions of the cerebellar output nuclei (Oscarsson, 1980; Voogd & Bigaré, 1980). The fact that the afferents to the olive and the efferents from the deep cerebellar nuclei also have a highly specific organization has been taken as evidence that each zone, or in some cases an ensemble of zones, serves a distinct function in motor control. In some zones, e.g. the C3 zone, a finer grain microzonal organization has also been demonstrated. The C3 zone is part of a system divisible into olivo-cortico-nuclear modules, each defined by input from climbing fibres with similar cutaneous receptive fields (Ekerot et al. 1991; Garwicz et al. 1996) and a projection to a specific group of cells in nucleus interpositus anterior (Garwicz & Ekerot, 1994). Each module has output acting on a small set of muscles in the forelimb and the action of this group of muscles overall tends to withdraw, from an external stimulus, the skin area approximately corresponding to the climbing fibre receptive field of the module itself (Ekerot et al. 1995).
With these considerations in mind, the aim of the present study was to characterize the mossy fibre input to the forelimb area of the paravermal C3 zone with respect to cutaneous receptive fields, zonal topography and the relationship to climbing fibre microzones. The present findings indicate that the C3 zone receives mossy fibre input characterized by a wide range of discrete receptive fields on the ipsilateral forelimb and that the representation of these receptive fields within the zone has a high degree of topographical organization. Overall, the mossy fibre input revealed many similarities with the climbing fibre input to the same parts of the zone, both in terms of spatial characteristics of receptive fields and intrazonal topography. However, with respect to the latter, the mossy fibre input displayed intricacies not found for the climbing fibre input.
The experiments were performed on eight adult cats (3–4 kg) anaesthetized with pentobarbitone (40 mg kg−1i.p.; supplementary doses 3–4 mg kg−1i.v., as required). The animals were purpose bred and all experimental procedures were approved in advance by the local Swedish Ethical Committee. The level of anaesthesia was characterized by complete muscle atonia, constricted pupils and a stable blood pressure, which remained constant during noxious stimulation of the skin. In order to obtain stable recording conditions, the animals were paralysed with alcuronium (0.1 mg kg−1i.v., as required) and a bilateral pneumothorax was made. Following the pneumothorax the cats were artificially ventilated. The muscle relaxant was allowed to wear off at regular intervals, in order to further check the level of anaesthesia. Cannulae were inserted into the trachea and the right femoral artery and vein. The mean arterial blood pressure was 90–120 mmHg. The end expiratory CO2 concentration and rectal temperature were monitored throughout the experiment and maintained at physiological levels. A continuous infusion of 5 % glucose in Ringer acetate (Kabi Pharmacia) was given to compensate for water and mineral losses.
The left cerebellar anterior lobe was exposed following craniotomy and a resection of the occipital lobe. The head of the cat was placed in a stereotaxic frame. A hole in the dura mater over the caudal brainstem was made in order to increase the mechanical stability of the cerebellum. All wound areas were infiltrated with lignocaine (5 %) (lidocaine). The hair on the ipsilateral forelimb was trimmed in order to facilitate mapping of cutaneous receptive fields.
At the end of the experiment the animals were killed by an overdose of barbiturate and perfused with formalin (10 %) in saline, the microelectrode (see below) being left in the tissue to facilitate histological reconstruction of recording locations.
Stimulation and recording
At the outset of each experiment the location of the paravermal sagittal zones (see Andersson et al. 1987) was delineated by mapping at the cerebellar surface the climbing fibre field potentials evoked on electrical stimulation of the skin (bipolar needle electrodes; pulse width, 0.1 ms; intensity, 0.8 mA; interstimulus interval, 3 s). Activity in single climbing fibres and mossy fibres was then recorded, respectively, as complex spikes in the Purkinje cell layer and mossy fibre potentials in the granular layer, using a glass-coated tungsten microelectrode (exposed tip, 10–15 μm). The intracortical electrical activity was monitored both visually, on an oscilloscope and a computer screen, and acoustically, via a pair of loudspeakers. Cortical layers were identified by electrophysiological features of spontaneous and evoked neuronal activity and depth below the cerebellar surface. In order to facilitate topographical analysis, the investigation was limited to the cortex at the dorsal surface of the folia. The recording depth did not exceed 300 μm below the Purkinje cell layer. For each unit, the mediolateral, rostrocaudal and dorsoventral (depth) co-ordinates were noted (resolution, 10 μm) and the location of each microelectrode track was indicated on a large-scale video image of the cerebellar surface.
As a search stimulus for mossy fibres, the skin of the ipsilateral forelimb was stimulated either electrically (see above) or manually, by brushing of hairs and tapping. Mossy fibre units were identified by criteria described in the first section of the Results. The cutaneous receptive fields of individual climbing fibres and mossy fibres were delineated using manually applied natural stimulation including touch, tapping and noxious pinch (see Ekerot et al. 1991). The cutaneous input to single mossy fibres was characterized also with respect to modality and the sensitivity and dynamics of the response. As a rule, it was relatively easy to determine the receptive field and usually only one or two of the experimenters were engaged in the mapping. In more difficult cases, all three authors independently mapped the receptive field.
Data analysis and presentation
Mossy fibre input is described both systematically and topographically, the latter with an emphasis on the relationship to climbing fibre input. The analysis shown in Fig. 6 was carried out using home-made computer software, operating on a database containing the cutaneous receptive field of each mossy fibre unit recorded from. The software was used to select and pool mossy fibre receptive fields according to their association with given climbing fibre reference units. The pooled receptive fields were then superimposed according to the principle outlined in Fig. 3.
In the topographical analysis of mossy fibre input, it was important to classify correctly the rostrocaudal level of folia recorded from, so that comparisons and pooling of data from different experiments could be made according to consistent rules. Since previous studies indicate that discrepancies may exist between the strictly anatomical and the functional representation of the zone across folia (Ekerot & Larson, 1979), we used physiological criteria, i.e. patterns of climbing fibre input, as indicators of the rostrocaudal zonal location of folia recorded from (Ekerot et al. 1991). Accordingly, folia were denoted as ‘rostral’ (presence of climbing fibre input from the hindlimb medially or laterally in the zone), ‘middle’ (climbing fibre input from the ulnar forearm present in two adjacent tracks) or ‘caudal’ (not ‘rostral’ and not ‘middle’).
General features of the material
The present material comprised a total of 226 mossy fibres and 143 climbing fibres activated by natural stimulation of the ipsilateral forelimb skin and terminating in the paravermal C3 zone in lobules IV and V. An additional twenty mossy fibres, found in the same regions of the zone but with receptive fields not including the forelimb, are shown only in Figs 2 and 5.
In Fig. 1, an example mossy fibre unit is presented with respect to the location of its receptive field on the forelimb (Fig. 1A) and to its discharge characteristics (Fig. 1B). Mossy fibre units were identified by their location in the granule cell layer and by the duration and shape of their potentials. In addition, they were notable by their often very high maximal firing frequencies (up to 1200 Hz within bursts). The unitary potentials typically consisted of a fast bi- (positive-negative) or triphasic (positive-negative-positive) waveform followed by a slower negative wave (Fig. 1C). The two components have previously been attributed to the incoming action potential and the glomerular synaptic potential, respectively (Walsh et al. 1974; van Kan et al. 1993).
A gradual decrease in the amplitude of the slow wave was regularly observed when units discharged tonically and/or at high frequencies (not shown). This finding (cf. Walsh et al. 1974) further adds to the consistency between defining characteristics of mossy fibre potentials in the present study and the more elaborate criteria for identification of mossy fibre units established in previous investigations from other laboratories. A direct comparison between raw data samples leaves little doubt as to the identity of units such as those presented below (cf. Fig. 1 in van Kan et al. 1993).
On electrical stimulation of the skin (up to 2 mA; see Methods), the units tested (n= 19) responded at relatively short latencies (4.3–9.6 ms), often with a burst of spikes (Fig. 1D) and many units displayed similar bursting on natural stimulation of the skin (Fig. 1B and C, bottom traces). Low spontaneous activity was typical of units in the present sample and many were completely silent in the absence of peripheral stimulation.
The raw data from an example experiment, in which a total of twenty-three climbing fibres and forty-one mossy fibres were sampled in the C3 zone, are illustrated in Fig. 2. In each animal at least two cerebellar folia were investigated, usually throughout the mediolateral width of the zone. The spacing between microelectrode tracks was routinely 150 μm in the mediolateral dimension and rostrocaudal co-ordinates were adjusted such that the sequence of tracks followed the long axis of the folium. A cutaneous climbing fibre receptive field of the Purkinje cells in the track could be determined in most cases. The electrode was then advanced into the granule cell layer and, with few exceptions, one to four mossy fibres were sampled in each track.
Characteristics of input to individual mossy fibre units
Spatial organization of cutaneous receptive fields. The spatial organization of individual mossy fibre receptive fields resembled that of climbing fibres projecting to the same parts of the C3 zone. However, whereas the overall climbing fibre input provided an uneven representation of the forelimb such that receptive fields on the ulnar forearm outnumbered those with receptive fields on the radial side (see also Ekerot et al. 1991), the overall mossy fibre input was biased to the radial forearm (Fig. 4, inset).
Based on the location of their proximal borders and the area of maximal sensitivity, receptive fields of climbing fibres in this zone have previously been divided into eight classes and tentatively into a total of thirty subclasses (Ekerot et al. 1991). A modified version of this scheme is shown in Fig. 4, providing a systematic presentation of the cutaneous receptive fields of mossy fibres found in the present study. Most of the corresponding climbing fibre receptive fields are illustrated in Fig. 6, upper panels. The principles for superimposing receptive fields to obtain the figurines in Fig. 4, and later also those in the lower panels of Fig. 6, are illustrated in Fig. 3.
As some previously unidentified climbing fibre receptive fields were found in the present study and these were matched by similar mossy fibre receptive fields, some new subclasses and classes have been introduced in the scheme in Fig. 4. However, as much as possible of the original nomenclature has been preserved in order to facilitate comparison with previous data. A total of six (6/226) mossy fibres with receptive fields on the forelimb did not fit the classification scheme.
Due to the low level of spontaneous activity of mossy fibres (see Fig. 1) and the specific location of the search stimulus, one cannot altogether exclude the presence of input from other body parts to the forelimb area of the C3 zone, as defined by climbing fibre input. However, when mapping single mossy fibres, it appeared that units contributing to the background noise had receptive fields extending on the forelimb, making it unlikely that mossy fibres with small non-forelimb cutaneous receptive fields constituted a substantial portion of the local input. Furthermore, in tracks without a significant input from the forelimb (see Fig. 2), the whole body surface was routinely explored in the search for input.
Modalities, sensitivity and dynamics of cutaneous input. The vast majority of individual mossy fibre units were found to be activated either by brushing of hairs, brushing of hairs and touch/light tap of the skin or touch/light tap of the skin. The level of sensitivity to natural stimulation and the dynamics of the response varied between different units. Mossy fibres with the highest sensitivity gave clear, sometimes tonic responses to displacement of single hairs and in a few cases responded even to weak airflow, while units with the lowest sensitivity required highly phasic stimulation of patches of hair or skin. Units with a low sensitivity to natural stimulation had a relative dominance of input from the upper arm, while high sensitivity was associated with input from distal radial parts of the paw, especially the skin between digits I and II.
Topographical organization of mossy fibre input
As illustrated in Fig. 2, the topographical organization of mossy fibre input to the forelimb area of the C3 zone was not a simple one. While mossy fibre units in a sequence of tracks along the mediolateral axis of the zone displayed systematic changes of receptive field location on the forelimb, different receptive fields were regularly encountered within tracks.
Figure 5 shows the mediolateral, rostrocaudal and dorsoventral distribution of mossy fibres with receptive fields belonging to different classes. In this analysis, the folia recorded from in individual experiments were first categorized as ‘rostral’, ‘middle’ or ‘caudal’, based on characteristics of their climbing fibre input (see Methods), and zone width was normalized separately for each rostrocaudal level. To produce Fig. 5, the normalized plots of individual folia were then pooled according to the rostral-to-caudal categorization. For clarity of illustration, the zone has been divided mediolaterally and dorsoventrally into, respectively, quarters and thirds (except rostral folia). Classes with similar receptive fields are presented in the same diagrams, while in some cases different sets of subclasses are shown with symbols of different shade.
The overall sample of mossy fibres was rather evenly distributed throughout the granular layer (Fig. 5, extreme right-hand panel). The small number of units in the middle of the caudal folia of the zone was not due to sampling bias, but represented instead the limited number of responsive units encountered in this part of the zone (see Fig. 2). The scarcity of activated units was confirmed also by the minute size or absence of synaptic field potentials in these regions.
In order to limit the account of Fig. 5 below, only departures from an even distribution in all three dimensions will be pointed out. Note that, overall, the relationships between termination territories of different receptive field classes evident from the pooled material were observed also within single experiments. The relationship to climbing fibre input (square symbols) will be dealt with in the next section.
The distribution of units with distal receptive fields on the radial side of the paw (classes 1a-c, 2a-c) was restricted mediolaterally, peaking around the boundary between the 3rd and 4th quarters in caudal folia and in the 3rd quarter of the middle folia. In the dorsoventral dimension, these units terminated mainly in the two superficial thirds of the granule cell layer. Mossy fibres with distal receptive fields on the ulnar side of the paw (classes 1d and e, 2d and e) appeared to have a lateral bias in rostral and middle folia compared with those with radial receptive fields. In caudal folia, the distribution was suggestive of a ‘double’ representation, with one lateral and one medial population, the latter overlapping with the distribution of classes 1a-c and 2a-c. Interestingly, this pattern fitted well with the distribution of the corresponding climbing fibres, as will be elaborated below (see Ekerot & Larson, 1979; Ekerot et al. 1991).
The zonal distribution of dorsal receptive fields (class 3) had a caudal bias. Class 3a-c (dorsal paw) displayed two peaks mediolaterally, one in the 2nd quarter and one in the 4th quarter. Partly overlapping with and lateral to the laterally located population of units in class 3a-c was the rather distinctly circumscribed population of mossy fibres in class 3d-f (dorsal forearm), with a centre of gravity around the boundary between the 1st and 2nd quarters. Mossy fibres with ventral receptive fields (classes 4 and 5) were located predominantly in the medial half of the zone. Both classes 4, 5b-e (ventral radial) and class 5f and g (ventral ulnar) were rather dispersed mediolaterally in caudal folia, but had distinct and separate patterns of termination in the middle folia, respectively, around the boundary between the 3rd and 4th quarters and in the medial part of the 4th quarter.
With few exceptions, mossy fibres with receptive fields on the ulnar forearm and paw (class 6) were confined mediolaterally to the medial part of the 4th quarter, while in the rostrocaudal dimension there was a heavy bias for middle folia. As a population, the medially located mossy fibres of class 6 displayed a bias for the deep third of the granular layer, especially relative to the present sample as a whole (extreme right-hand panel). However, given the small sample of units, the latter result is no more than tentative. Similarly, the number of receptive fields located mainly on the lateral side of the upper arm (classes 7 and 8) was small and results concerning topography are therefore not conclusive.
Units belonging to class 9, with receptive fields on the radial side of the forearm and paw, had a bias for caudal folia and displayed a clear bimodal distribution in the mediolateral dimension, peaking in the 2nd and 4th quarters. Mossy fibres in class 10 terminated mainly, but not exclusively, in the lateral half of the zone in caudal folia, where they peaked in the 1st quarter lateral to units from class 9. Mossy fibres with receptive fields on the face and neck were found in a circumscribed region of the zone, namely in the 1st mediolateral quarter, middle dorsoventral third of caudal folia. Units with receptive fields on the trunk had a similar location in caudal folia, but were also found laterally in middle and rostral folia of the zone.
Relationship between mossy fibre and climbing fibre input
Several of the above findings have alluded to the relationship between receptive fields of mossy fibres and climbing fibres. Similarities in the existing types of receptive fields are implicit by allowing a corresponding classification (Fig. 4). Similarities in the patterns of termination were found in the raw data (Fig. 2) and by comparing the intrazonal mediolateral and rostrocaudal topography of mossy fibre and climbing fibre input to the C3 zone (Fig. 5; see also Ekerot et al. 1991). The distribution across receptive field classes in the present material was rather similar for mossy fibres and climbing fibres and the main discrepancy was seen in class 2, whereas classes 3, 6, 9 and 10 differed to a lesser extent (Table 1). The sample of climbing fibre receptive fields differed somewhat from samples in previous studies, discrepancies again occurring mainly in class 2 and 3, but also in class 1. Note also that although classes 9 and 10 were introduced only in the present investigation, subclass 9a is identical to former subclass 5a, while the few climbing fibre receptive fields resembling those in subclass 9b and class 10 have previously been presented in the ‘not classified’ group.
Table 1. Proportions of mossy and climbing fibre units in present and previous studies
Proportions of mossy fibre (MF) and climbing fibre (CF) units in different receptive classes sampled in the present study (a) and by Joörntell et al. (1996) (b) and Ekerot et al. (1991) (c). Numbers in parentheses indicate the total number of units. *Note that subclass 9a in the present study was identical to subclass 5 a in previous studies.
The relationship between mossy fibre and climbing fibre input is further evaluated in Fig. 6, which shows the receptive fields of mossy fibre units terminating in given climbing fibre microzones (see Introduction). The typical receptive field of the different reference climbing fibre units is shown in the upper panels. The figurines in the lower panels were produced by superimposing the receptive fields of all mossy fibre units terminating in the same microelectrode tracks as the search reference units (see Fig. 3).
For several subclasses, the overall population of mossy fibres terminating underneath a particular reference climbing fibre had receptive fields on average strongly resembling the receptive field of that climbing fibre. Microzones 1c, 2a, 2b, 2c, 3d and 6c (cf. Fig. 4) provided the clearest examples in favour of such an organization. In most cases, this was in spite of the relatively low proportion of mossy fibres belonging to the subclass corresponding to the reference climbing fibre (compare filled and hatched bars in Fig. 6, lower panels). The sparse input from the corresponding subclass rarely reflected a small sample of that subclass in the material as a whole (open bars), but appeared to be instead a true organizational feature, representing a modest degree of convergence between climbing and mossy fibres with identical receptive fields. The only exception was microzone 6c, which actually contained the majority of mossy fibres in subclass 6c (compare filled and open bars in Fig. 6).
In some cases, e.g. climbing fibre subclasses 3e, 4a, 5b and 10a, the associated mossy fibres had receptive fields located on adjacent areas of the forelimb skin, resembling other but still rather similar subclasses. On the other hand, a somewhat different principle of organization seemed to apply to subclasses such as 1e, 2e and 3a, converging consistently with quite dissimilar mossy fibres, or to subclass 5e, in which the organization of receptive fields did not give a coherent, easily interpretable picture. The lack of mossy fibre units in microzone 1a reflects the location of this climbing fibre subclass close to the centre of the zone in caudal folia (cf. also other subclasses in class 1). In many experiments, this region was devoid of peripherally activated mossy fibre input altogether. The lack of units with receptive fields on the forelimb in microzone 10b is accounted for in the Discussion.
The present study provides the first account of the functional organization of cutaneous mossy fibre input to identified microzones in the paravermal cortex of the cerebellar anterior lobe of the cat. The relationship between mossy fibre input and the climbing fibre input to the same cortical regions was assessed, revealing a high degree of similarity in spatial organization of receptive fields and sequences of termination within the zone for the two types of afferents. However, the mossy fibre input had a more intricate topographical organization and units with non-identical, albeit usually similar, receptive fields were frequently found in the same microelectrode track.
Given the characteristics of input to single mossy fibres and their patterns of termination (for references see below), the most likely candidate pathway for mediating the input seems to be the exteroceptive component of the cuneocerebellar tract (E-CCT; for references see Oscarsson, 1973). Although one cannot exclude contributions from other systems, such as the rostral spinocerebellar tract or pathways mediated via the lateral reticular nucleus (LRN), the E-CCT would be favoured also by the fact that transmission in at least the LRN is suppressed under barbiturate anaesthesia (e.g. Crichlow & Kennedy, 1967).
Characteristics of input mediated by single mossy fibres
Modalities, sensitivity and dynamics. Also, in terms of modality, sensitivity and dynamics of mossy fibre input, our data are compatible with previous studies. They reported that activity in mossy fibre pathways was mainly evoked by stimulation of hair and by touch, but also by stimulation of hair only or by touch only (Holmqvist et al. 1963; E-CCT), gentle deflection of fur or light contact with glabrous skin (Kassel et al. 1984) or light cutaneous stimulation (van Kan et al. 1993). Differences in dynamics (adaptation) of responses similar to those reported here have been noted previously (Holmqvist et al. 1963; Kassel et al. 1984).
Comparison with climbing fibre input. In some respects, the characteristics of input to single mossy fibres differed from input to climbing fibres. Mossy fibres had a higher sensitivity to non-noxious cutaneous stimulation, while nociceptive input, which is so prominent in climbing fibres projecting to the C3 zone (Ekerot et al. 1991), was found only in a few units. Mossy fibre units activated exclusively by nociceptive input may have been overlooked due to the low level of spontaneous activity in the granular layer in combination with the non-nociceptive nature of the search stimulus. However, if present, such units would be expected to have receptive fields not overlapping with those reported here, otherwise they would have been detected by their contribution to the background noise when nociceptive stimulation was applied during mapping of isolated units.
In terms of spatial organization, a large number of single mossy fibre receptive fields were almost indistinguishable from their climbing fibre counterparts and could easily be fitted into a similar classification scheme. However, some mossy fibres had receptive fields that fell between classes or subclasses in this scheme, resulting in an overall less definite classification. A further difference between the two inputs was the presence of very small, rather proximal mossy fibre receptive fields, e.g. in classes 5, 6 and 9. Such receptive fields did not in principle appear to be tailored in the same way as climbing fibre receptive fields, which typically have proximal borders close to joints and eccentrically located areas of maximal sensitivity, features suggestive of a relationship to movement (see Ekerot et al. 1991).
Given the high sensitivity of a large proportion of mossy fibres activated from hair and skin and the crucial role for cutaneous input in proprioception (e.g. Edin & Johansson, 1995), it is not unlikely that mossy fibres activated from muscle and skin provide complementary information about different parameters of movement per se (van Kan et al. 1993; see our Introduction). Movement detection would be aided also by the airflow-sensitive mossy fibres, which probably would be activated by mere limb movement. In addition, since input from the mossy fibres studied reaches the cerebellum at short latencies and is finely graded, with both phasic and tonic components, it would be expected to reliably convey information about characteristics of objects inspected during exploratory movements or about objects unexpectedly encountered during fast, goal-directed movements.
Topographical aspects of mossy fibre input
Distribution of mossy fibres with different receptive fields.Ekerot & Larson (1980) found that mossy fibre input to the forelimb area of the C3 zone had a topographical organization such that stimulation of different peripheral nerves evoked potentials in different sagittally oriented cortical strips. The present results are compatible with the previous findings in demonstrating an overall similar zonal topography, based on a coherent termination of units with similar receptive fields. However, while discrepancies in gross topography are few, the variety of receptive fields within single microelectrode tracks in the present study (Fig. 2), suggests an organization that is different from the one described by Ekerot & Larson (1980) not only in the level of detail, but also in principle.
Our findings provide evidence for a specific mossy fibre termination also in the dorsoventral dimension, although the data suggest only relative differences and only for some receptive field classes. Such subtleties may not be easily detectable anatomically (e.g. Cajal, 1911) and, as far as we know, depth-specific termination within a mossy fibre pathway has not been previously described. However, it is noteworthy that Arshavsky et al. (1981) reported a difference in termination depth between peripheral and cerebral mossy fibre input to the granular layer.
Based on the assumption that a relationship exists between depth of the granule cell and the depth of the corresponding parallel fibre in the molecular layer (see Eccles et al. 1967), Arshavsky et al. (1981) proposed that deeply located granule cells would depolarize Purkinje cell dendrites closer to the soma. Note, however, that although there is some evidence for such a granule cell-parallel fibre organization with respect to depth (Fig. 2E in Mellor et al. 1998; see also Arshavsky et al. 1981), the issue remains unclear (J. Rawson, personal communication). One can only speculate that a functional compartmentalization of input to the granular layer may result in a compartmentalization of input to different parts of the dendritic tree of the overlying Purkinje cells.
Sagittal organization versus‘fractured somatotopy’. Welker and colleagues investigated the cutaneous input to the granule cell layer of cerebellar crus II and paramedian lobule in cat (Kassel et al. 1984) and rat (also crus I; e.g. Shambes et al. 1978) using natural stimulation of the skin. From their high-density mapping they concluded that the mossy fibre input has a ‘fractured’ or ‘patchy’ somatotopy, with projections within patches being somatotopically organized and projections to adjacent patches often coming from non-contiguous body regions. In addition, particular regions of the body often had multiple representations, even within a single folium (reviewed by Welker, 1987).
The fractured somatotopy is presented as differing in principle from the sagittal zonation (Welker, 1987). However, given methodological differences between investigations from Welker's group and those from our laboratory and the fact that different regions of the cerebellar cortex were mapped, it seems that the studies complement each other and should not be considered as contradictory in terms of raw data. For example, in the rat a comparison between the study of Shambes et al. (1978) and a recent combined electrophysiological and anatomical study of olivocerebellar input to partly overlapping regions of the cerebellum (Atkins & Apps, 1997) suggests that, in some cases, somatotopically organized ‘patches’ may correspond to entire zones. The ‘fracturing’ would correspond to the characteristic, often abrupt shifts of representation at zonal boundaries (see Andersson et al. 1987). Analogous similarities can be found in the cat by comparing the rostral-most folium of the paramedian lobule in Kassel et al. (1984) and Trott & Apps (1993).
Relationship between mossy fibre and climbing fibre topography. As expected from the variety of mossy fibre receptive fields encountered in individual microelectrode tracks, there was not a one-to-one relationship between the climbing fibre receptive field defining a microzone and the average receptive field of the associated mossy fibres, not even in terms of receptive field focus. Instead, this relationship appeared to differ in different microzones, ranging from strong resemblance to no resemblance at all (Fig. 6).
The high degree of resemblance between mossy fibre and climbing fibre receptive fields in microzones 1c and 6c was based on a large sample of mossy fibre units (Fig. 4) with a rather circumscribed termination (see Fig. 5). Climbing fibre microzones 2a and 2c received instead a predominant input from mossy fibres in the ‘analogous’ subclasses in class 1. By contrast, several mossy fibre subclasses contained a relatively large number of units terminating in a relatively clustered fashion, but in microzones with non-identical receptive fields. This was seen to different extents in climbing fibre microzones 2e, 3a, 4a, 5b and 10a, which received substantial input from, respectively, mossy fibre subclasses 3b, 9b, 1a, 1a and 10c.
The receptive field mismatch in microzones 10a (mossy fibre input mainly from subclass 10c) and 10b (mossy fibre input mainly from face and neck; see Fig. 5), may be due to the fact that the angle of the microelectrode remained the same throughout each experiment. Since the curvature of the cortex is greatly increased in lateral, compared with medial, parts of the C3 zone, recordings from the granular layer will in effect represent a more lateral zonal region than recordings from the molecular layer in the same track. This is illustrated in the lateral most three or four tracks in the caudal folium of Fig. 2, where the matching between climbing fibre and mossy fibre receptive fields occurs obliquely rather than perpendicularly to the surface of the cortex. This caveat may apply also to microzones 1d and 1e, which were represented both in medial and rather lateral parts of the zone. Mossy fibre receptive fields in the lateral tracks were predominantly located proximal to the wrist.
The lateral shift between Purkinje cell layer and granular layer recordings can be quite accurately estimated, as the curvature in paravermal parts of the folia has been measured in some previous studies in our laboratory (e.g. Garwicz & Andersson, 1992). Assuming that the angle between the microelectrode and the cerebellar surface ranged from 15 to 45 deg in individual experiments, the range of lateral shifts for the deepest recording sites (300 μm below the Purkinje cell layer, where the shift is maximal) would be 80–210 μm. Such a shift could potentially correspond to one or two microzones (see Ekerot et al. 1991).
In some cases, e.g. microzone 2e, the receptive mismatch between the two inputs reflects in part quantitative differences in the present sample - whereas mossy fibre class 2 had a very modest representation compared with its climbing fibre counterpart, the reverse was true for class 3 (Table 1). Receptive fields of classes 9 and 10 displayed an analogous mismatch between numbers of mossy fibre and climbing fibre units, reflected in the presence of mossy fibre classes 9 and 10 in other microzones, e.g. those of class 3. However, for many other microzones, e.g. 4a and 5b, it appears that the sometimes subtle discrepancies between the two inputs have no other explanation than that they reflect true principles of organization.
Spread of activity along parallel fibres. Further complexity is added to the functional relationship between mossy fibre and climbing fibre input by the fact that the axons of the granule cells, the parallel fibres, extend for several millimetres along the folia (e.g. Brand et al. 1976; Pichitpornchai et al. 1994), thus traversing many different microzones, or even different zones. Therefore, except for the associations between the two inputs suggested by the present findings, a variety of potential combinations of climbing fibre and mossy-parallel fibre receptive fields could be envisaged at the point of convergence between the two inputs onto single Purkinje cells.
However, the issue of spread of activity along the parallel fibres is a matter of some controversy and reports on the distance of spread range from up to 1.5 mm (Garwicz & Andersson, 1992) to no spread at all (Bower & Woolston, 1983). The latter findings suggested a ‘radial’ organization of mossy fibre input, emphasizing the importance of the synaptic contacts between the ascending portion of the granule cell axon and the Purkinje cell (Llinás, 1982). Such an organization is supported in part also by the distribution and size of parallel fibre varicosities and their associated postsynaptic densities (Pichitpornchai et al. 1994). Indeed, the high degree of congruence between complex spike and simple spike receptive fields in single Purkinje cells (e.g. Thach, 1967; Eccles et al. 1972; see, however, Edgley & Lidierth, 1988) would appear compatible with a predominantly radial organization. On the other hand, as suggested by Braitenberg et al. (1997), it is possible that activation of Purkinje cells outside the mossy fibre projection area depends on particular dynamic aspects of the peripheral input to granule cells, which is not easily mimicked under experimental conditions.
Mode of operation of cerebellar circuitry
The present findings add to a series of studies (see Garwicz et al. 1998) suggesting that the cerebellar control system encompassing the cortical C1, C3 and Y zones and the intracerebellar nucleus interpositus anterior consists of more than thirty parallel modules. As illustrated in Fig. 7, each module receives from the forelimb skin a homogeneous climbing fibre input and a similar, albeit more heterogeneous, mossy fibre input and in turn controls, via the rubrospinal tract, a small group of muscles in the limb. The relationship between input and output is roughly such that the movement caused by the muscles would tend to withdraw the associated cutaneous climbing fibre receptive field from a stimulus. It appears, however, that the mossy fibre input, with its distribution on more than one forelimb segment, mirrors more closely the multijoint nature of the output from a given module.
According to this organization, skin contact with an object would activate Purkinje cells via the mossy fibres, thereby inhibiting the output of the module and therefore reducing the drive on a group of muscles that prevent further contact with the object. In simplistic terms, this would constitute a cerebellar cortical loop providing positive feedback for the original movement. If the climbing fibres signal motor errors and in the process depress the efficacy of synaptic transmission between parallel fibres and Purkinje cells (Ito, 1989), ‘unintended’, ‘unexpected’ or ‘erroneous’ contact with an object would activate the climbing fibres to the module and in the long term cause a diminished positive feedback via the cortical loop.
However, speculations about the mode of operation of this circuitry are seriously hampered by the lack of information on the spatial and temporal organization of mossy fibre input to nucleus interpositus anterior. This point is underscored by the fact that a number of studies have questioned the existence of nuclear collaterals from mossy fibres belonging to the E-CCT (Dietrichs & Walberg, 1987; see also Gerrits et al. 1985), and therefore it cannot be taken for granted that the nuclear cells in a given module receive the same input as the overlying cortex. If this were the case, however, the adjustable positive feedback loop via the cortex would be balancing the activity in a negative feedback loop via the nucleus.
This study was supported by the Magn. Bergvall Foundation, the Swedish Society for Medical Research, the Swedish Medical Research Council (project no. 8291), Greta and Johan Kocks Stiftelser and the Royal Physiographic Society in Lund.