Receiving editor: Prof. S. Grillner
correspondence: Prof. E. De Schutter

EJN/1998/090846

A Patchy Horizontal Organisation of the Somatosensory Activation of the Rat Cerebellum Demonstrated by Functional MRI

¹Bio Imaging Lab, University of Antwerp, RUCA, Belgium
²Vision Lab, University of Antwerp, RUCA, Belgium
³Born-Bunge Foundation, University of Antwerp - UIA, B2610 Antwerp, Belgium

ABSTRACT

Introduction

Functional brain mapping methods (Ogawa et al., 1990a,b) allow one to study the activation of specific projections to the entire brain at once. The most commonly used fMRI (functional Magnetic Resonance Imaging) technique is BOLD (Blood Oxygenation Level Dependent contrast), where a change in concentration of the blood's own deoxyhemoglobin manifests itself in a small change in signal intensity in T₂^*-weighted images (Ogawa et al., 1990a; Turner et al., 1991). Due to its non-invasiveness, BOLD-fMRI studies have become very popular to study the human brain. This includes both high resolution imaging of the functional topography to visual (Tootell et al., 1996) and somatosensory cortex (Sakai et al., 1995) and the determination of brain areas involved in specific cognitive tasks, like learning of motor control (Karni et al., 1995) or speech production (Binder, 1997).

Reports on animal fMRI studies on the contrary are sparse and have mostly used simple stimulation protocols, measuring well defined input projections to the cortex in the rat (Hyder et al., 1994; Gyngell et al., 1996; Kerskens et al., 1996; Yang et al., 1996, 1997), the mouse (Huang et al., 1996) or the cat (Jezzard et al., 1997). While the cerebellum is receiving particular attention in fMRI studies of the human brain (Ellerman et al., 1994; Gao et al., 1996; Allen et al., 1997; Desmond and Fiez, 1998), imaging studies in animals have been limited to the optical recordings of voltage changes at the surface of the cerebellum only (Ebner and Chen, 1995; Chen et al., 1996; Cohen and Yarom, 1998). Nevertheless, as the function of the cerebellum is at present quite controversial (De Schutter and Maex, 1996; Raymond et al., 1996; Bower, 1997; Schmahmann, 1997), it is important to image the functional activation of the cerebellum in animals, where the results can be compared with the extensive available anatomy and electrophysiology.

Because of the limited sampling typical of anatomical or electrophysiological techniques, projections to only a few parts of the rat cerebellum have been mapped in detail (Welker, 1987; Voogd, 1995). FMRI imaging could be an efficient method to map the functional projections everywhere in the cerebellum at once, provided it can achieve the necessary resolution to image patches which often have a diameter of only a few 100 micro m. In this study we demonstrate the possibility of mapping functional activation in the rat cerebellum using fMRI in a 7 Tesla magnet, which allows for a resolution of about 200 micro m (Hyder et al., 1994). We demonstrate that the observed patchy activation patterns have an overall horizontal organisation.

Our results have been reported previously in abstract form (Peeters et al., 1998).

Materials and methods

Nine adult female wistar rats (200-300 g) were initially sedated with a subcutaneous injection containing a mixture of Ketamine (35 mg/kg, Ketalar, Parke-Davis, Belgium) and Xylazine (5mg/kg, Rompun, Bayer, Germany). After 10 minutes, alpha-chloralose (40 mg/kg, Acros Organics, Belgium) was administrated intraperitoneally (IP) and a supplemental dose (30 mg/kg IP) was given 2 hours later. Anaesthesia waned after 3 hours, often leading to movement artifacts.

The animal was fixed in a plexiglass stereotactic head holder, consisting of an incisor bar and blunt earplugs, to enable accurate positioning within the magnet and immobilisation of the animal. During the experiment, the respiration rate of the animal was monitored and the body temperature was maintained using a warm water blanket (37 °C +/- 1 °C).

A copper wire which was covered with a conductive electrode paste (GRASS, Astro-Med, Inc West Warwick, USA) was wrapped with a single turn around the left forepaw and a second wire was attached similarly to the left hindpaw. Square electrical pulses (WPI stimulator, Sarasota Florida USA) were delivered with a frequency of 2.5 Hz and a duration of 0.5 ms, resulting in a duty cycle of 0.125 %. The current amplitude was 200 micro A, which is considered to be innocuous (Handwerker and Kobal, 1993).

MRI methods

MR-imaging was performed at 300 MHz on a SMIS MR microscope (SMIS, Guilford England) with a horizontal 7-T magnet and 8 cm aperture self-shielded gradient coils with a gradient strength of 0.1 T/m (Oxford Instruments, England). A circular RF surface coil of 16 mm diameter was placed on top of the skull and was used for both transmitting and receiving the MR signal.

To image the cerebral cortex the surface coil was positioned with its centre approximately 1 mm posterior to the bregma and 12 coronal images with a slice thickness of 1mm were taken from 4 mm anterior to 7 mm posterior to the bregma. To image the cerebellum the surface coil was placed over the cerebellum with the centre of the coil approximately 3 mm posterior to the interaural line. Twelve coronal or sagittal slices of 1mm comprising the complete cerebellum were imaged.

All images were acquired with a multislice gradient echo (GE) sequence. After tuning and shimming (¹H linewidth of approx. 40 Hz), high resolution sagittal and coronal images were taken, with a Field Of View (FOV) of 20 mm, a gradient echo time (TE) of 8 ms, a repetition time (TR) of 500 ms and an acquisition matrix of 256x128. Two averages were taken and all images were reconstructed after zerofilling the acquisition matrix to 256x256 data points.

Coronal and sagittal multislice functional imaging was performed at the same position of the 6 inner slices of the high resolution coronal images or the 8 inner slices of the high resolution sagittal ones. The GE-sequence was T₂^*-weighted (TE = 16 ms, slice thickness = 1 mm, FOV = 20 mm, image matrix = 128x128) with a TR of 150 ms for the coronal and 200 ms for the sagittal images. The total acquisition time per image set was 19.2 s for the coronal and 25.6 s for the sagittal images. The spatial resolution of the functional images was 160x160 micro m.

Stimulation protocol

The stimulation protocol for all the animals consisted of three cycles of 6 images without and 6 images with stimulation of one of the paws, resulting in 36 functional image sets. Figure 1 shows the protocol and the resulting signal intensity changes in different areas. This protocol was used for both forepaw and hindpaw stimulation. In most rats two consecutive protocols were executed: either one orientation of slicing for both paws or both orientations for one paw. In a few rats all four stimulation protocols were executed, but movement artifacts in the latter parts of the experiment ruined the last set of images in all except one (rat 5).

Fig. 1: Time course of the signal change in different ROI's, during the fMRI experiments with stimulation of the rat's forepaw or hindpaw. The square wave shows the stimulation paradigm, with the stimulation switching on/off every 6 images. Three different traces of different ROI's are shown. For activated ROI's in the cerebellum during hindpaw (l) and forepaw (s) stimulation, we observed a 6 to 9% signal change following stimulation. The time course of the signal in the whole cerebellum (n) during stimulation, demonstrated no significant signal drift.

Data analysis

The fMRI data were analysed off-line using both routines in MEDx (Version 3.0, Sensor Systems, Inc, Sterling, USA) and custom developed routines and procedures in IDL (Interactive Data Language, RSI, Boulder Colorado, USA). The following steps were taken to obtain the activation maps:

• A motion detection algorithm, detecting motion of the centre of intensity in 3 directions was used to assess for major motion artefacts. The motion occurring in the time series was generally found to be at subpixel level, so that no motion correction was necessary. If the motion was higher some images were discarded from the dataset (Fig. 4A only).
• A 3 x 3 pixel Gaussian convolution filter was applied which resulted in noise reduction of the images.
• The activated regions were defined by using a paired t-test comparison of the "stimulation" and "non stimulation" images. This resulted in statistical maps giving the t-values of each voxel.
• This statistical map was interpolated to a 256x256 matrix.
• The activated regions sustained when thresholding the t-values maps at a level of t(34)=2 (p=0.05, giving a confidence level of 95%) were overlaid on the corresponding high resolution images, resulting in high resolution images showing in colour the regions with a significant difference between the images acquired with and without stimulation (Figs. 2, 4 and 5).
• Activation sites were identified according to the stereotaxic atlas of the rat brain of Paxinos and Watson (1986).

Results

Activations in somatosensory cortex

Fig. 2: Cerebral activation maps of two consecutive coronal slices resulting from electrical stimulation of the left forepaw. Activation is shown as brighter spots superposed on the underlying anatomical images. The activation seen in the middle between the 2 hemispheres is due to a large venous blood vessel: the sagittal sinus and in the first image a large contributing vene at the left side. This was confirmed independently by MRI angiography.

Similarly, during stimulation of the left hindpaw, activation was observed corresponding to the hindpaw area in the right parietal cortex, as expected (Chapin and Lin, 1990) more medial and posterior than the forepaw activation (not shown). These activation patterns were observed in all animals, with small variations probably due to differences in the positioning of the slices between animals.

Activations in the cerebellar cortex

The cerebellar activation patterns caused by stimulation of the hindpaw or forepaw were quite complex: it consisted of many isolated patches, generally organized in a medio-lateral fashion. Before describing the activations in detail, we will first consider the validity of our fMRI protocol. The observed activations had t-values in the range of 2-4.5, indicating high significance levels. Moreover, almost all of the activation spots consisted of several pixels and most of them could be found in both the coronal and sagittal slices.

Figure 3 shows a composite of all the cerebellar activations observed during either left forepaw or left hindpaw activation in a total of 9 different rats. This figure demonstrates two important points. First, the pattern of activations we observed was quite reproducible from animal to animal, but individual patches had often slightly different localisations. Second, the activations caused by either front or hindpaw stimulation were almost completely non-overlapping, both in the coronal and sagittal slices. It is thus unlikely that these cerebellar activations were caused by large venous blood vessels (with an exception possibly in Fig. 5D, see further), like we found in the cortical images (Fig. 2). The reproducibility and stimulus-dependency of the activations both argue for the validity of our results.

Fig. 3: Composite maps of activated areas during hindpaw and forepaw stimulation of all the imaged rat cerebella. A. Coronal images. B. Sagittal images. The red tinted spots show activated areas resulting from hindpaw stimulation and the blue-green tinted those resulting from forepaw stimulation. Several common activated areas of different animals resulting from hind or forepaw stimulation are present with a clear separation of the fore and hindpaw areas in the rostral cerebellum. The activated areas shown were thresholded at a t-value level of 2.0.

Finally, the original data (Figs. 4 and 5) show that the cerebellar activation could not be imaged completely with the surface coil we used. In effect, the area which could be observed reliably was limited to a distance of about 6 mm deep in the coronal slices and 3 mm lateral of the midline in the sagittal slices. Beyond these distances some activations could still be recorded in a reproducible way, but not resolved accurately (e.g. Fig. 5 A,C).

While almost no overlapping activations were found for the stimulation of respectively hind or forepaw, they did show many common features. In each case the activations were bilateral, showed a predominant medio-lateral organisation and were confined to similar horizontal levels in the anterior and posterior cerebellum. We will next describe the patchy structure in greater detail, based on the original data sets of three rats (Figs. 4 and 5). Additional original data can be viewed at .

Cerebellar activation caused by left hindpaw stimulation

As this is the simpler of the two response patterns, we will first consider the response to hindpaw activation in two rats in detail (Fig. 4). In every animal we observed bilateral signal changes in all the coronal slices through the cerebellum during electrical stimulation. These were situated in a broad horizontal rostral band between 1.5 and 3 mm below the upper surface of the cerebellum (top of VIa), with additionally a few activations very low (about 7 mm below).

Fig. 4: Cerebellar activation maps resulting from electrical stimulation of the left hindpaw superimposed on high resolution images of the cerebellum. Figures A and B are from rat #5 while C and D are from rat #4. Coronal images (A and C) show 5 consecutive 1 mm thick slices located from -1 mm to -5 mm interaurally, sagittal images (B and D) show 6 slices of 1 mm thickness. Several of the activated areas in 4B,D match with the position of the activations in the coronal images in 4A,C. All activations are computed from a functional imageset of 36 images (except 4A: 32 images) and are shown in a colour code representing t-values ranging from 2.0 to 4.5.

One of these low activations was in the anterior cerebellum, where a bilateral activation is observed in lobule I in the sagittal images of most animals (B2-4 and D2-4). It is not observed in the coronal slices probably because of the large distance from the surface coil. At a similar depth, but more laterally, a large contralateral activation is seen in one animal (B1). In the coronal slices several corresponding activations are situated in the anterior interpositus and lateral nuclei (A2 and C2).

A second series of anterior cerebellum activations is situated around the preculminate fissure. The most anterior one consists in rat 5 of a number of ipsilateral (three spots in A1 and one in each of B4 and B5) and one contralateral activation in lobule III (A1 and B3); in rat 4 similar activations could be found in the coronal slice C1 but in the sagittal slices only a contralateral activation is visible in D1-2. At the other side of the preculminate fissure a series of bilateral activations is observed deep in lobule IV. In rat 5 these are discrete spots just behind and at about the same height as the activations in lobule III (B2-6 but mostly contralateral in A2). In rat 4 the corresponding activations are much more extensive and seem to consist of several confluent patches in the sagittal images (D1-6) and in the coronal image (C2), with a medial extension into lobule V which is only visible in the sagittal slices (D3-4).

Several more lateral spots of activation seen in the coronal slices were outside the field of view for the sagittal slices, so they could not be located accurately. They were probably located in the lobulus simplex or crus I (ipsilateral in C1 and bilateral in A2 and C2, also ipsilateral in B6 and D5) or the top of crus II (bilateral A3 and C3), though a location in the lateral nuclei cannot be excluded. The most noticeable aspect of these activations, combined with the ones described before in lobules III and IV, was their medio-lateral organisation along a single band (Fig. 3A).

In the posterior cerebellum a spot is seen deep in lobule VIa close to the primary fissure in the sagittal images, mostly ipsilateral central in rat 5 (B3-4) and more extensive in rat 4 (D2-4, with a second higher spot of activation in D3), but these activations are not significant in the corresponding coronal images (A3 and C3) except for the ipsilateral spot in A3 (it was also visible in the corresponding images of rats 2 and 3, cf. Fig. 3A). Similarly an activation is observed deep in VIb in the sagittal images of rat 4 only (D3-4).

Another activation in the posterior cerebellum is deep in lobules VII and VIII, close to the prepyramidal fissure. We find this as extensive bilateral spots in sagittal images B3-5 and D2-5 and as two symmetric activations in A4 or as two contralateral spots in C4. More laterally, several spots of activation are visible on each side of the coronal images A4-5 and C4-5 in the copula pyramis and lower paramedian lobule; these activations are mostly outside the field of view of the sagittal images except for a small ipsilateral activations in B6 and a contralateral one in D1.

Finally, one (C5, D3-4) to two (B3-4) medial activations are seen in lobule X close to the posterolateral fissure.

Cerebellar activation caused by left forepaw stimulation

Stimulation of the left forepaw (Fig. 5) gave rise to signal changes in parts of the cerebellum different from those seen during the hindpaw stimulation. This stimulation resulted also in a bilateral pattern of activation, which was confined to two horizontal levels: a rostral series of spots 0.5 to 1.5 mm below the surface of the rostral vermis and paravermis and extensive activations in the caudal cerebellum and brainstem. The latter activation was typical for the forepaw stimulation paradigm: it was seen in all 7 rats, while no brainstem activation was found for the hindpaw stimulation in any of the same rats. Though this brainstem activation was very reproducible, it did not seem very reliable: the activations were very large, often spanning several structures and were sometimes located in unlikely places (e.g. the vestibular nucleus and cerebellar peduncles). As these activations were specific to one stimulation paradigm they were probably genuine, but they could not be resolved accurately because of the large distances from the surface coil. We will therefore describe the cerebellar locations only.

Fig. 5: Cerebellar activation maps resulting from electrical stimulation of the left forepaw. Figures A and B are from rat 5 while C and D are from rat 6. Same conventions as in Fig. 4.

In coronal slice A1 several spots of activation are seen in a medio-lateral band, more pronounced activation ipsilaterally. These activations were common to all rats (Fig. 3a) except for rat 6, presumably because of the more anterior positioning of the first coronal slice of rat 6 (Fig. 5C). These activations could not be found in the sagittal images, except for an activation deep in lobule IV in rat 5 (B1-2).

Additional activations in the anterior cerebellum were located Superficially in lobules IV and V of the vermis: several symmetrically disposed patches of activation span mediolaterally along the upper part of coronal slice A2 and C2. In the sagittal images the contralateral activations can be located more easily: superficial (B2-3) in lobule IV and several spots in lobule V close to the primary fissure (B2, D1-3,6).

In the posterior lobe several patches were located in lobule VI: in VIa close to the primary fissure (B2-4), in VIa/b at the fissure (B2-4, D2-6) and in VIb (B2-3, D3-6). Note that on sagittal slices B2 and B3, right of the midline, these activations seemed to form a continuous sagittal band (t-value of 3.4), but this is not the case on D2-3. All these activations were in olivary termination zone A (Buisseret-Delmas and Angaut, 1993). The coronal images revealed activations corresponding to the possible sagittal band just described in the midline region of lobule VI of the vermis (A3-4, C5). Additionally, several more lateral spots of activation were found in lobule VI (A3-5, C3-4). Here more patches were activated ipsilaterally and most contralateral ones had a symmetric ipsilateral homologue. All the sagittal images from rat 6 show, additionally, a highly significant activation at the upper surface of lobule VIb (D1-6, striped patch in Fig. 3B) which is likely to be a venous artifact caused by one of the veins covering the cerebellum.

As mentioned before, the forepaw stimulation also activated large parts of the lower cerebellum which could not be resolved accurately. These included parts of the lower vermis, with several activations in the midline of lobule I (A2-3, B3-4, C2-3) and seemingly even more extensive in lobule X (A4-5, B2-5, C5, D2-5) close to the posterolateral fissure bilaterally. More laterally, activations were found in the copula pyramis and the paramedian lobule, mostly contralaterally (A4-5, B1,6, C4-5). The forepaw stimulation may also have activated the cerebellar nuclei, but these activations are not very well delineated. Specifically, in coronal slices A2/C2 and A3/C3 most of the caudal activity may be located in the medial (B3-4, D3), interpositus (B1-2,5-6, D1-2) and lateral nuclei.

Discussion

Comparison to other animal fMRI studies

The regions of signal elevation in the somatosensory cortex during electrical stimulation of the left fore and hindpaw in this study corresponded to the front and hindlimb regions in rat cortex (Chapin and Lin, 1990) and to activations described in other fMRI studies (Hyder et al., 1994; Gyngell et al., 1996; Kerskens et al., 1996). Overall, the activated areas were smaller than those found in other studies in which electrical stimulation of the forepaw was used. The discrepancy might be due to the lower current used. Gyngell et al. (1996) used a current of 0.5mA, which is 2.5 times higher than the one we used, while Hyder et al. (1994) used a voltage controlled stimulator, so that they did not report the stimulation current. Gygnell et al. also reported a size discrepancy between these two studies, which was attributed to the stimulation current they used.

We used a surface coil because preliminary fMRI tests with a volume birdcage coil, which has a lower signal to noise ratio than the surface coil, resulted in no detectable signal changes above threshold after stimulation. Disadvantages of using the surface coil is that the detectable area is relatively small and the presence of a signal drop: the further away from the midpoint of the sphere the larger the signal drop (see Figs. 4 and 5). This can be partially solved by the use of adiabatic pulses. For these reasons it was not possible to image the cortex and the cerebellum at the same time, as there is a 12 mm distance between both structures. As we expected that hindlimb and forelimb projections could be found in the deep cerebellum (Voogd, 1995), we used a relatively large surface coil. In the cortex, however, the activated zones are positioned near the cortical surface, which makes it possible to use a smaller surface coil (e.g. Hyder et al. (1994) used a surface coil with a diameter of 8 mm) with a better signal to noise ratio.

Validity and reproducibility of the observed activations

With functional MR imaging of neural responses the signal change observed may reflect the presence of a large venous blood vessel in the neighbourhood instead of being truly representative of neural activation (Ogawa et al., 1990a,b; Segebarth et al., 1994). In Fig. 2 the activations at the cortical surfaces centrally between the two hemispheres were due to such a large venous blood vessel (the sagittal sinus). Similar signals of venous origin have been reported in other studies of rat cortical responses (Hyder et al., 1994; Kerskens et al., 1996). In the cerebellum this problem is less likely to occur because of the absence of large venes on the top of or inside the cerebellum (Scremin, 1995). We had also the advantage of using a 7T high field strength, which gives a relatively smaller contribution of large vessels in the T₂^*-signal changes as compared to the contribution of capillaries than at the lower fields used in human studies (Menon et al., 1995). Because we did not see any cerebellar activation common to the forepaw and hindpaw stimulation paradigms, we assume that most cerebellar activations reflect neural activity. An exception is the very superficial activation in Fig 5D. It can also not be excluded that the activations caused by forepaw or hindpaw stimulation which we attributed to lobule X were in fact due to the venous plexus beneath this lobule.

The reproducibility of the cerebellar activations was not perfect. The composite images in Fig. 3 demonstrate that while most activation spots were present in all of the rats imaged, they were not always localised in exactly the same place. Some of these differences could be explained by the variability in positioning of the slices between rats and by the volume effects due to the relative thick slices (1 mm, this is also the most likely cause of differences between the sagittal and coronal images), but it is likely that genuine differences were present between individual rats. Most of these differences may be caused by the normal variability in the gross anatomy of the brain, which is already obvious when the two rat brains shown in Fig 4 and 5 each are compared (see also Perciavalle et al. (1998) for variability in sizes of the cat cerebellum and Steinmetz and Seitz (1991); Woods et al. (1994); Rombouts et al. (1997); Strother et al. (1997) for variability among human brains). In humans these differences can be corrected for by transforming each image to a standard human brain (Talairach and Tournoux, 1988), but such techniques are not available for the rat brain. As a consequence the exact anatomical localisations as shown in the composites in Fig. 3 have to be considered approximative and it is unknown if the small differences in patch localisations between rats are genuine or would disappear after normalization.

Anatomical projections to the cerebellum

The fMRI images usually had not sufficient resolution to distinguish between activations in the molecular versus granular layer of the cerebellar cortex. This is important because such localisation would identify the source of the activation: granular layer activations can be caused by mossy fiber input only, while an isolated molecular layer activation would suggest input from climbing fibers. Some isolated evidence for both can be found in the sagittal images of Fig. 5B, where the rostral activations in the anterior lobe (slices 2-4) had both very superficial (presumably molecular layer) and deeper (presumably granular layer) components, while the activations in lobule VI were often only in the superficial layers (e.g. slice 4).

For now we can only compare the activated regions with the known somatosensory projections to the cerebellum. Of these, only the mossy fiber projections by the spinocerebellar and cuneocerebellar pathways and an indirect spino-cerebellar path through the lateral reticular nucleus have been described in detail in the rat (Voogd, 1995). These projections have sometimes been described as following the parasagittal banding revealed by zebrin II immunocytochemistry (Ji and Hawkes, 1994), but many studies (reviewed by Voogd (1995)) have found a less strict organisation.

The spinocerebellar pathways which were activated during hindpaw stimulation project to the vermis of lobules II, III, IV and V (Tolbert et al., 1993; Voogd, 1995) like we found, but also to lobule II (Matsushita et al., 1991; Tolbert et al., 1993; Alisky and Tolbert, 1997), where we detected no significant activation. Similarly, in the posterior lobe projections to lobule VIII have been described (Gravel and Hawkes, 1990; Matsushita et al., 1991) in consistence with our observations. Conversely, we also observed activations in lobules VI and in lobule X which are not known projection zones of the direct spinocerebellar pathway (see also the remarks above about the veracity of the lobule X activations). These activations could be caused by other projections arising from many different brain structures, though the most likely ones are the lateral nucleus (Clendenin et al., 1974b; Russchen et al., 1976; Chan-Palay et al., 1977), the olivary nuclei (Voogd, 1995) and the corticopontine projection for the more lateral activations (Bjaalie and Brodal, 1997). The topography of the corticopontine input to the rat cerebellum has not been described in detail but in the case of facial inputs it is known to be activated during somatosensory activation (Bower et al., 1981) and to be more bilateral in its projections than the trigeminal mossy fiber input (Morissette and Bower, 1996).

The cuneocerebellar pathways which were activated by the forepaw stimulation project to the anterior vermis in an ipsilateral sagittal band spanning lobules II to V, consisting of several separated patches (Alisky and Tolbert, 1997; Tolbert and Gutting, 1997). We found bilateral activations in lobules III-V. This is not in contradiction with the anatomical data, as weak staining was also found at the contralateral side (Tolbert and Gutting, 1997). Moreover, similar to the mentioned anatomical study, we observed a medio-lateral band of activations in the posterior part of lobule V (Fig. 5A2, C2) which was more pronounced ipsilaterally. Finally, the bilateral midline activations in lobule I again match the anatomical projections (Tolbert and Gutting, 1997) where, similar to our observation of rostral and caudal horizontal planes of activation, the terminals in lobule I are completely separated from the projections to the more rostral lobules.

The cuneocerebellar projection to the posterior lobe forms transversely oriented bands (Tolbert and Gutting, 1997) which fits our observations. Nevertheless, in lobule VI only an ipsilateral band has been described in VIa, while we found bilateral activations in both VIa and VIb. One explanation for this discrepancy could be of course the possible contralateral parasagittal band observed in several rats (Fig. 3B), suggestive of the activation of climbing fibers originating in medial subnucleus C of the medial accessory olive (Buisseret-Delmas and Angaut, 1993). Another possible source of bilateral projections is the lateral reticular nucleus, especially if it was activated by the bilateral ventral flexor reflex tract (Clendenin et al., 1974a, c). More laterally the lobule VI activations were mainly ipsilateral, like the mossy fiber projections (Tolbert and Gutting, 1997). These authors also describe two more posterior transverse bands of mossy fiber projections, one extending in the paramedian lobule and another one into the copula pyramis. The ipsilateral forelimb project to the paramedian lobule has also been described in electrophysiological studies (Welker, 1987; the copula pyramis was not investigated). We found extensive activations in both the paramedian lobule and copula pyramis, but again these were bilateral while the known anatomical projections are to the ipsilateral hemisphere and the bilateral vermis only.

The more extensive activations in the cerebellar nuclei during forelimb compared to hindlimb stimulation (Fig. 3B) fit with electrophysiological studies that have described an extensive representation of the forelimb (and the face) in the rat dentate nucleus, with only a small representation of the hindlimb (Angaut and Cicirata, 1990). Unfortunately, as was mentioned before, the lower resolution of the images at this distance from the surface coil did not allow for an accurate anatomical localisation of the nuclear activity.

A final point concerning the correspondence with anatomical studies of the mossy fiber system is that these investigations have shown a clear separation between the spino and cuneocerebellar projections to the anterior lobe (Ji and Hawkes, 1994; Alisky and Tolbert, 1997). Similarly, we found completely distinct activation sites during hindpaw and forepaw stimulation respectively (Fig. 3), but while the anatomical studies found a medio-lateral separation we mostly observed a rostro-caudal one. Careful inspection of the data of rat 5 (Figs 4A, 5A) shows, however, that the forepaw and hindpaw activations were often also separated medio-laterally.

The patchy medio-lateral organisation in horizontal planes

Both transverse and longitudinal patterns of organisation are present in the anatomy of the cerebellum (Ito, 1984; Voogd and Ruigrok, 1997). The transverse axis is determined by the parallel fibers and their relationship with the dendritic arbors in the molecular layer (Braitenberg and Atwood, 1958). The parallel fibers form in most of the cerebellum an uninterrupted transverse band from the vermis to the most lateral hemisphere (Voogd, 1995). This has led Eccles et al. (1967) to propose that beams of parallel fiber activation would be the fundamental mode of cerebellar operation. Later authors were struck by the longitudinal zonal organisation of the corticonuclear projections, which was found to match the parasagittal organisation of the climbing fiber projections (Voogd and Bigare, 1980; Voogd, 1995). This strong parasagittal organisation of one afferent pathway and of the only efferent pathway of cortex has been proposed to be further subdivided into microzones (Oscarsson, 1979). The parasagittal organisation can also be found in the banding of the cerebellum produced with several staining techniques (Hawkes and Gravel, 1991). Whether the second afferent system, the mossy fibers, follow a similar pattern of parasagittal organisation is controversial (Tolbert et al., 1993; Ji and Hawkes, 1994; Voogd, 1995; Tolbert and Gutting, 1997).

Neither of these two forms of organisation were found, however, in careful electrophysiological studies of granular layer activation by somatosensory mossy fiber input (Shambes et al., 1978; Welker, 1987; Bower and Kassel, 1990). These authors found instead what they called a fractured somatotopy: each body location was represented multiple times in a mosaic of patches which were mixed in seemingly random order. Moreover, when Purkinje cell responses were recorded similar patch-like receptive fields were found, instead of the expected beam-like responses due to parallel fiber activation (Bower and Woolston, 1983). This patchy activation of cerebellar cortex by mossy fiber stimulation has also been observed in voltage imaging of the isolated guinea pig cerebellum (Cohen and Yarom, 1998).

The few human fMRI studies of the somatotopic organisation of the cerebellum have revealed a mixture of activation patterns during motor activity, including parasagittal bands (Ellerman et al., 1994), medio-lateral bands (Ellerman et al., 1994; Nitschke et al., 1996) and many isolated patches. In our study we found all of these activation patterns during low amplitude electrical stimulation of a rat paw too, with a predominance of isolated patches and medio-lateral bands. The most striking pattern, however, was a rostro-caudal horizontal organisation of these responses. This is most obvious in the sagittal images (e.g. Fig. 3B) where most activations lie in either a rostral or caudal plane with a width of 1 to 1.5 mm. The rostral planes were more obvious, because they were completely separate for fore and hindpaw stimulation. They comprise activations in both anterior (lobules III-V) and posterior (VI-VIII) cerebellum. The caudal planes were not separated and included probably nuclear activations, but, as mentioned before, the image resolution was poor at this depth.

We are not aware of any previous description of such a horizontal organisation in the cerebellar cortex. As we pointed out above, many of the observed activations correspond to known anatomical projection sites of spinocerebellar and cuneocerebellar mossy fibers, which also have a patchy organisation (Tolbert et al., 1993; Tolbert and Gutting, 1997). These studies all rely on the injection of tracers into the spinal cord, which even in the case of very small injections is not expected to label specifically those fibers which carry sensory information from a small body surface like a single paw. Therefore it is not surprising that anatomical tracer studies show a much wider projection area than observed in our fMRI study. Similarly, most electrophysiological studies would never have found a horizontal organisation because they either mapped activity at the surface of the hemispheres (Shambes et al., 1978) or vermis (Joseph et al., 1978) only, or made penetrations in one lobule only (e.g. the many studies summarised by Ito (1984)).

Obviously our observation of a horizontal organisation needs confirmation, both by recording from the sites of activation in electrophysiological studies and by demonstrating that this pattern is not specific to the electrical activation of the paws. Our study also shows the importance of taking slices of the cerebellum at multiple orientations, as some activations in the coronal slices seemed to be parasagittal bands but were really isolated patches, while the sagittal slices did not demonstrate the medio-lateral bands clearly.

The functional significance of the observed pattern of activation raises interesting questions. The medio-lateral organisation of bands and patches within one coronal slice probably follows the corresponding parallel fiber beams. Several theories of cerebellar function suggest that multiple activation sites along one beam would allow mixtures of different input which may be important towards Purkinje cell functioning (Bower, 1997; Braitenberg et al., 1997; De Schutter, 1998). Whether the organisation of these medio-lateral bands within one broad horizontal plane throughout different lobules has a functional significance, or reflects the developmental program (Oberdick et al., 1998) only, is still unknown.

ABBREVIATIONS

BOLD: Blood Oxygenation Level Dependent contrast
fMRI: functional Magnetic Resonance Imaging
MRI: Magnetic Resonance Imaging

ACKNOWLEDGEMENTS

RRP is supported by the IWT (Instituut Wetenschap en Technologie, Flanders), EDS and BPV are supported by the FWO (Fund for Scientific Research Flanders). This research was partially supported by the research project G.0113.96 of the FWO (Flanders).

REFERENCES

Alisky, J. M. & Tolbert, D. L. (1997) Quantitative-analysis of converging spinal and cuneate mossy fiber afferent-projections to the rat cerebellar anterior lobe. Neurosci, 80, 373-388.

Allen, G., Buxton, R. B., Wong, E. C. & Courchesne, E. (1997) Attentional activation of the cerebellum independent of motor involvement. Science, 275, 1940-1943.

Angaut, P. & Cicirata, F. (1990) Dentate control pathways of cortical motor activity. Anatomical and physiological studies in rat: comparative considerations. Arch Ital Biol, 128, 315-330.

Binder, J. R. (1997) Neuroanatomy of language processing studied with functional MRI. Clin Neurosci, 4, 87-94.

Bjaalie, J. G. & Brodal, P. (1997) Salient anatomic features of the cortico-ponto-cerebellar pathway. Prog Brain Res, 14, 227-249.

Bower, J. M. (1997) Is the cerebellum sensory for motor's sake or motor for sensory's sake: the view from the whiskers of a rat? Prog Brain Res, 114, 463-496.

Bower, J. M., Beerman, D. H., Gibson, J. M., Shambes, G. M. & Welker, W. (1981) Principles of organization of a cerebro-cerebellar circuit. Micromapping the projections from cerebral (SI) to cerebellar (granule cell layer) tactile areas of rats. Brain Behav Evol, 18, 1-18.

Bower, J. M. & Kassel, J. (1990) Variability in tactile projection patterns to cerebellar folia crus IIA of the Norway rat. J Comp Neurol, 302, 768-778.

Bower, J. M. & Woolston, D. C. (1983) Congruence of spatial organization of tactile projections to granule cell and Purkinje cell layers of cerebellar hemispheres of the albino rat: vertical organization of cerebellar cortex. J Neurophysiol, 49, 745-766.

Braitenberg, V. & Atwood, R. P. (1958) Morphological observations on the cerebellar cortex. J Comp Neurol, 109, 1-33.

Braitenberg, V., Heck, D. & Sultan, F. (1997) The detection and generation of sequences as a key to cerebellar function. Experiments and theory. Behav Brain Sci, 20, 229-245.

Buisseret-Delmas, C. & Angaut, P. (1993) The cerebellar olivo-corticonuclear connections in the rat. Prog Neurobiol, 40, 63-87.

Chan-Palay, V., Palay, S. L., Brown, J. T. & Van Itallie, C. (1977) Sagittal organization of olivocerebellar and reticulocerebellar projections: autoradiographic studies with 35S-methionine. Exp Brain Res, 30, 561-576.

Chapin, J. K. & Lin, C.-S. (1990) The somatic sensory cortex of the rat. In Kolb, B. and Tees, R. C. (eds), The cerebral cortex of the rat. The MIT Press, Cambridge, MA, pp. 341-380.

Chen, G., Hanson, C. L. & Ebner, T. J. (1996) Functional parasagittal compartments in the rat cerebellar cortex: an in vivo imaging study using neutral red. J Neurophysiol, 76, 4169-4174.

Clendenin, M., Ekerot, C. F. & Oscarsson, O. (1974a) The lateral reticular nucleus in the cat. III. Organization of component activated from ipsilateral forelimb tract. Exp Brain Res, 21, 501-513.

Clendenin, M., Ekerot, C. F., Oscarsson, O. & Rosen, I. (1974b) The lateral reticular nucleus in the cat. I. Mossy fibre distribution in cerebellar cortex. Exp Brain Res, 21, 473-486.

Clendenin, M., Ekerot, C. F., Oscarsson, O. & Rosen, I. (1974c) The lateral reticular nucleus in the cat. II. Organization of component activated from bilateral ventral flexor reflex tract (bVFRT). Exp Brain Res, 21, 487-500.

Cohen, D. & Yarom, Y. (1998) Patches of synchronized activity in the cerebellar cortex evoked by mossy-fiber stimulation: Questioning the role of parallel fibers. Proc Natl Acad Sci USA, 98, 15032-15036.

De Schutter, E. (1998) Dendritic voltage and calcium-gated channels amplify the variability of postsynaptic responses in a Purkinje cell model. J Neurophysiol, 80, 504-519.

De Schutter, E. & Maex, R. (1996) The cerebellum: cortical processing and theory. Curr Opin Neurobiol, 6, 759-764.

De Zeeuw, C. I., Lang, E. J., Sugihara, I., Ruigrok, T. J. H., Eisenman, L. M., Mugnaini, E. & Llinás, R. (1996) Morphological correlates of bilateral synchrony in the rat cerebellar cortex. J Neurosci, 16, 3412-3426.

Desmond, J. E. & Fiez, J. A. (1998) Neuroimaging studies of the cerebellum: language, learning and memory. Trends Cogn Sci, 2, 355-362.

Ebner, T. J. & Chen, G. (1995) Use of voltage-sensitive dyes and optical recordings in the central-nervous-system. Prog Neurobiol, 46, 463-506.

Eccles, J. C., Ito, M. & Szentagothai, J. (1967) The cerebellum as a neuronal machine. Springer-Verlag, Berlin.

Ellerman, J. M., Flament, D., Kim, S.-G., Fu, Q.-G., Merkle, H., Ebner, T. J. & Ugurbil, K. (1994) Spatial patterns of functional activation of the cerebellum investigated using high field (4T) MRI. NMR Biomed, 7, 63-68.

Gao, J. H., Parsons, L. M., Bower, J. M., Xiong, J. H., Li, J. Q. & Fox, P. T. (1996) Cerebellum implicated in sensory acquisition and discrimination rather than motor control. Science, 272, 545-547.

Gravel, C. & Hawkes, R. (1990) Parasagittal organization of the rat cerebellar cortex: direct comparison of Purkinje cell compartments and the organization of the spinocerebellar projection. J Comp Neurol, 291, 79-102.

Gyngell, M. L., Bock, C., Schmitz, B., Hoehn-Berlage, M. & Hossmann, K.-A. (1996) Variation of functional MRI signal in response to frequency of somatosensory stimulation in a-chloralose anesthetized rats. Magn Res Med, 36, 13-15.

Handwerker, H. O. & Kobal, G. (1993) Psychophysiology of experimentally induced pain. Physiol Rev, 73, 639-671.

Hawkes, R. (1997) An anatomical model of cerebellar modules. Prog Brain Res, 114, 39-52.

Hawkes, R. & Gravel, C. (1991) The modular cerebellum. Prog Neurobiol, 36, 309-327.

Huang, W., Palyka, I., Li, H., Eisenstein, E. M., Volkow, N. D. & Springer, C. S., Jr (1996) Magnetic resonance imaging (MRI) detection of the murine brain response to light: Temporal differentiation and negative functional MRI changes. Proc Natl Acad Sci USA, 93, 6037-6042.

Hyder, F., Behar, K. L., Martin, M. A., Blamire, A. M. & Shulman, R. G. (1994) Dynamic magnetic resonance imaging of the rat brain during forepaw stimulation. J Cereb Blood Flow Metab, 14, .

Ito, M. (1984) The cerebellum and neural control. Raven Press, New York.

Jezzard, P., Rauschecker, J. P. & Malonek, D. (1997) An in vivo model for functional MRI in cat visual cortex. Magn Res Med, 38, 699-705.

Ji, Z. & Hawkes, R. (1994) Topography of Purkinje cell compartments and mossy fiber terminal fields in lobule II and lobule III of the rat cerebellar cortex: spinocerebellar and cuneocerebellar projections. Neurosci, 61, 935-954.

Joseph, J. W., Shambes, G. M., Gibson, J. M. & Welker, W. (1978) Tactile projections to granule cells in caudal vermis of the rat's cerebellum. Brain Behav Evol, 15, 141-149.

Karni, A., Meyer, G., Jezzard, P., Adams, M. M., Turner, R. & Ungerleider, L. G. (1995) Functional MRI evidence for adult motor cortex plasticity during motor skill learning. Nature, 377, 155-158.

Kerskens, C. M., Hoehn-Berlage, M., Schmitz, B., Bush, E., Bock, C., Gyngell, M. L. & Hossmann, K.-A. (1996) Ultrafast perfusion-weighted MRI of functional brain activation in rats during forepaw stimulation: comparison with T2^*-weigthed MRI. NMR Biomed, 8, 20-23.

Matsushita, M., Ragnarson, B. & Grant, G. (1991) Topographic relationship between sagittal Purkinje cell bands revealed by a monoclonal antibody to zebrin i and spinocerebellar projections arising from the central cervical nucleus in the rat. Exp Brain Res, 84, 133-141.

Menon, R. S., Kim, S. G., Hu, X., Ogawa, M. & Ugurbil, K. (1995) Functional MR imaging using the BOLD approach: Field strength and sequence issues. In Le Bihan, D. (ed), Diffusion and Perfusion Magnetic Resonance Imaging. Raven Press Ltd, New York, pp. 327-344.

Morissette, J. & Bower, J. M. (1996) Contribution of somatosensory cortex to responses in the rat cerebellar cortex granule cell layer following peripheral tactile stimulation. Exp Brain Res, 109, 240-250.

Nitschke, M. F., Kleinschmidt, A., Wessel, K. & Frahm, J. (1996) Somatotopic motor representation in the human anterior cerebellum. A high-resolution functional MRI study. Brain, 119, 1023-9.

Oberdick, J., Baader, S. L. & Schilling, K. (1998) From zebra stripes to postal zones: deciphering patterns of gene expression in the cerebellum. Trends Neurosci, 21, 383-390.

Ogawa, S., Lee, T. M., Kay, A. R. & Tank, D. W. (1990a) Brain magnetic resonance imaging with contrast dependent on blood oxygenation. Proc Natl Acad Sci USA, 87, 9868-9882.

Ogawa, S., Lee, T. M., Nayak, A. S. & Glynn, P. (1990b) Oxygenation-sensitive contrast in magnetic resonance imaging of rodent brain at high magnetic fields. Magn Res Med, 14, 68-78.

Oscarsson, O. (1979) Functional units of the cerebllum: sagittal zones and microzones. Trends Neurosci, 2, 143-145.

Paxinos, G. & Watson, C. (1986) The rat brain in stereotaxic coordinates. Academic Press, Inc., San Diego.

Peeters, R. R., Verhoye, M., Van Der Linden, A., Van Dyck, D., Vos, B. P. & De Schutter, E. (1998) Functional imaging resonance imaging of the rat cerebellum during electrical stimulation of the fore- and hindpaw at 7 tesla. Eur J Neurosci, 10, Suppl. 10, 438.

Perciavalle, V., Bosco, G. & Poppele, R. E. (1998) Spatial organization of proprioception in the cat spinocerebellum. Purkinje cell responses to passive foot rotation. Eur J Neurosci, 10, 1975-1985.

Raymond, J. L., Lisberger, S. G. & Mauk, M. D. (1996) The cerebellum: a neuronal learning machine? Science, 272, 1126-1131.

Rombouts, S. A., Barkhof, F., Hoogenraad, F. G., Sprenger, M., Valk, J. & Scheltens, P. (1997) Test-retest analysis with functional MR of the activated area in the human visual cortex. Amer J Neuroradiol, 18, 1317-1322.

Russchen, F. T., Groenewegen, H. J. & Voogd, J. (1976) Reticulocerebellar connections in the cat. An autoradiographic study. Acta Morphol Neerl Scand, 14, 245-246.

Sakai, K., Watanabe, E., Onodera, Y., Itagaki, H., Yamamoto, E., Koizumi, H. & Miyashita, Y. (1995) Functional mapping of the human somatosensory cortex with echo-planar MRI. Magn Res Med, 33, 736-743.

Sasaki, K., Bower, J. M. & Llinás, R. R. (1989) Multiple Purkinje cell recording in rodent cerebellar cortex. Eur J Neurosci, 1, 572-586.

Schmahmann, J. D. (1997) The cerebellum and cognition. (Vol. 41). Academic Press, San Diego.

Scremin, O. U. (1995) Cerebral vascular system. In Paxinos, G. (ed), The rat nervous system, (2nd ed.). Academic Press Inc., San Diego,

Segebarth, C., Belle, V., Delon, C., Massarelli, R., Decety, J., Lebas, J. F., Decorps, M. & Benabid, A. L. (1994) Functional MRI of the human brain: predominance of signals from extracerebral veins. NeuroReport, 5, 813-816.

Shambes, G. M., Gibson, J. M. & Welker, W. (1978) Fractured somatotopy in granule cell tactile areas of rat cerebellar hemispheres revealed by micromapping. Brain Behav Evol, 15, 94-140.

Steinmetz, H. & Seitz, R. J. (1991) Functional anatomy of language processing: neuroimaging and the problem of individual variability. Neuropsychol, 29, 1149-1161.

Strother, S. C., Lange, N., Anderson, J. R., Schaper, K. A., Rehm, K., Hansen, L. K. & Rottenberg, D. A. (1997) Activation pattern reproducibility: measuring the effects of group size and data analysis models. Hum Brain Mapp, 5, 312-316.

Talairach, J. & Tournoux, P. (1988) Co-planar stereotaxic atlas of the human brain: 3-dimensional proportional system: an approach to cerebral imaging. Thieme, Stuttgart.

Tolbert, D. L., Alisky, J. M. & Clark, B. R. (1993) Lower thoracic upper lumbar spinocerebellar projections in rats: a complex topography revealed in computer reconstructions of the unfolded anterior lobe. Neurosci, 55, 755-774.

Tolbert, D. L. & Gutting, J. C. (1997) Quantitative analysis of cuneocerebellar projections in rats: differential topography in the anterior and posterior lobes. Neurosci, 80, 359-371.

Tootell, R. B. H., Dale, A. M., Sereno, M. I. & Malach, R. (1996) New images from human visual cortex. Trends Neurosci, 19, 481-489.

Turner, R., Le Bihan, D., Moonen, C. T. W., DesPres, D. & Franck, J. (1991) Echo-planar time course MRI of cat brain oxygenation changes. Magn Res Med, 22, 159-166.

Voogd, J. (1995) Cerebellum. In Paxinos, G. (ed), The rat nervous system, (2nd ed.). Academic Press Inc., San Diego, pp. 309-350.

Voogd, J. & Bigare, F. (1980) Topographical distribution of olivary and cortico nuclear fibers in the cerebellum: A review. In Courville, J., de Montigny, C. and Lamarre, Y. (eds), The inferior olivary nucleus: Anatomy and physiology. Raven Press, New York, pp. 207-235.

Voogd, J. & Ruigrok, T. J. (1997) Transverse and longitudinal patterns in the mammalian cerebellum. Prog Brain Res, 114, 21-37.

Welker, W. (1987) Spatial organization of somatosensory projections to granule ceII cerebellar cortex: Functional and connectional implications of fractured somatotopy (summary of Wisconsin studies). In King, J. S. (ed), New concepts in cerebellar neurobiology. Alan R. Liss, Inc., New York, pp. 239-280.

Woods, R. P., Mazziotta, J. C. & Cherry, S. R. (1994) Optimizing activation methods: tomographic mapping of functional cerebral activity. In Thacher, R. W., Hallet, M., Zeffiro, T., John, E. R. and Huerta, M. (eds), Functional neuroimaging: Technical foundations. Academic Press, San Diego, pp. 47-58.

Yang, X., Hyder, F. & Shulman, R. G. (1996) Activation of single whisker barrel in rat brain localized by functional magnetic resonance imaging. Proc Natl Acad Sci USA, 93, 475-478.

Yang, X., Hyder, F. & Shulman, R. G. (1997) Functional MRI BOLD signal coincides with electrical activity in the rat whisker barrels. Magn Res Med, 38, 874-877.