Current Concepts of Climbing Fiber Function

118 THE ANATOMICAL RECORD (NEW ANAT.)

REVIEW

JAMES R. BLOEDEL AND VLASTISLAV BRACHA

This review examines several of the current postulates regarding the function of one of the most intriguing afferent systems in the brain, the climbing fiber system. The fact that these afferents are activated under a variety of conditions has contributed substantially to the diversity of postulates that have been proposed. In part because of the unique anatomical relationship between individual climbing fibers and the dendritic tree of Purkinje cells, these afferents have been proposed as a key input in establishing long-term plastic changes in the cerebellar cortex. This concept is contrasted with other postulates proposing that the heterosynaptic action of this system produces a short-lasting enhancement rather than a long-term depression of Purkinje cell responsiveness. Although a generally accepted view regarding climbing fiber function does not exist, this review emphasizes the extensive functional insights that have been reported and supports the notion that progress toward a complete understanding of these afferents will require an integration of their morphological characteristics with the fundamental physiological properties of their responses assessed in a variety of contexts and conditions. Anat. Rec. (New Anat.) 253:118–126, 1998. ௠ 1998 Wiley-Liss, Inc.

KEY WORDS: cerebellum; climbing ﬁber; cerebellar circuits; Purkinje cells; olivocerebellar

Cajal’s elegant description of one of is not graded and which occurs at a tions, Eccles and colleagues (for over- the most unique anatomical relation- relatively low frequency. view, see 1) formulated one of the first ships in the brain, the extensive inter- From the onset of the electrophysi- postulates regarding the functional ac- face between the dendritic tree of a ological studies characterizing the re- tion of this system, the read-out hy- Purkinje cell and a single climbing sponses of Purkinje cells to these in- pothesis. This view proposed that the fiber, triggered a century-long debate puts, the responses evoked by these depolarization evoked by the climbing regarding the function of this unique fibers were found to be as unique as fiber ‘‘read out’’ the level of excitability afferent system. This relationship is their morphology. Primarily as a con- produced via the combined action of illustrated in Figure 1, which also sequence of the classical experiments the mossy fiber–parallel fiber input shows the organization of the other by Eccles, Llinas, and Sasaki (re- directly to Purkinje cells and to the major input to these neurons, the viewed in 1), these afferents were dendrites of the inhibitory interneu- mossy fiber–granule cell–parallel fiber shown to evoke a characteristic, large rons. Although this postulate is no projection. In general, the mossy fiber depolarization of Purkinje cells and an longer considered tenable, it focused input is responsible for modulating associated complex spike response that the attention of physiologists on the unique role these fibers may play in the discharge of Purkinje cells in re- is distinctively different from the ac- regulating the output of the cerebellar sponse to activity in descending path- tion potentials modulated by the other cortex and more generally in regulat- ways as well as to a wide variety of major afferent system to the Purkinje ing the functional properties of the sensory inputs. In contrast, the climb- cell dendrites, the parallel fibers. An cerebellum itself. ing fiber input evokes a response which example of both the intracellularly While not intended to be compre- and extracellularly recorded response hensive, this review will provide an of Purkinje cells to a climbing fiber encapsulation of several theories of input is shown in Figure 2. These climbing fiber function that are rel- responses were produced by stimulat- evant to current discussions regarding Dr. Bloedel is currently the Chair of the ing the cerebellar white matter and this system. Interestingly, none of these Division of Neurobiology at the Barrow consequently contain both an anti- Neurological Institute in Phoenix, AZ. theories have been disproved, and all He and Dr. Bracha, a faculty member in dromically evoked action potential (the have at least some experimental sup- the same Division, codirect a laboratory initial spike in both records evoked by port. Perhaps because critical ques- focused on the learning and control of a variety of movements, with an empha- stimulating the axon of the neuron tions remain unanswered, controversy sis on understanding the role of the in the cerebellar white matter) and still abounds on this topic, contribut- cerebellum in these processes. His fax the complex spike response evoked by ing to the fact that there are almost as number is 602–406–4172; his e-mail address is [email protected] the climbing fiber input. Based on many viewpoints on this topic as there their electrophysiological observa- are investigators in the field!

௠ 1998 Wiley-Liss, Inc. REVIEW THE ANATOMICAL RECORD (NEW ANAT.) 119

Figure 1. Illustration of the mammalian cerebellar cortex. A single folium is sectioned to reveal the planar organization of the Purkinje cell dendrites, the relationship of the climbing ﬁber to these structures, the interneurons (basket, Golgi, and stellate cells), and the mossy ﬁber system. (Reproduced from Bloedel28 with permission of the publisher).

SPECIFIC VIEWS OF CLIMBING the body. These experiments as well as flecting a discrepancy or mismatch FIBER FUNCTION earlier electrophysiological studies between the intended and the actual characterizing the inputs to a variety movement. Other proponents of this Climbing Fibers as an Event or of olivocerebellar projections2 led to concept argued that this sensory infor- Error Signal the proposal that the climbing fiber mation is critical to the operation of Several studies in unanesthetized de- may be signaling the occurrence of an the modules consisting of cerebellar cerebrate animals revealed that climb- unexpected sensory stimulus during cortical sagittal zones and the topo- ing fibers were capable of responding the course of a movement. This view graphically organized output from re- to very low intensity stimuli applied to was extended by postulating that these lated regions of the cerebellar nuclei.3 their receptive field on the surface of signals may represent error signals re- At least three lines of experimenta- 120 THE ANATOMICAL RECORD (NEW ANAT.) REVIEW tion in awake behaving animals pro- employed experimental conditions in vide strong evidence supporting the which the animal encountered an un- view that the climbing fiber input can expected sensory input whose charac- signal an unexpected interruption of teristics were difficult to quantify or intended and/or ongoing movement. simply were not measured accurately First, Gellman et al.4 demonstrated as a systematic component of the pro- that olivary neurons (neurons originat- tocol. Simpson and colleagues9 have ing in the inferior olive) that were obviated this problem by employing a highly responsive to low intensity cuta- paradigm in which a very specific, neous stimuli applied passively were quantifiable sensory stimulus was em- not activated when their receptive ployed, this being the movement of fields came in contact with the floor multiple spots of light across the retina during ongoing locomotion. Second, in a specific direction. This visual in- in other experiments,5 climbing fibers put provides a directionally specific were activated by a unique, unex- retinal slip signal which these investi- pected perturbation. Cats were trained gators found to be very effective in to walk on the rungs of a ladder. A activating climbing fiber inputs to spe- rung was permitted to give way unex- cific locations within the cerebellar pectedly when contacted by the cat, flocculus of the rabbit, a region having Figure 2. An extracellularly recorded (A) and numerous interconnections with re- producing a substantial deviation from intracellularly recorded (B) response evoked the expected movement. Under this in two different Purkinje cells by a stimulus gions of the brainstem important in condition, climbing fiber inputs to Pur- applied to the cerebellar white matter (WM). kinje cells were activated. Third, experiments in our laboratory employed Together these data multiple single unit recordings from ample, Ojakangas and Ebner8 ob- up to five sagittally oriented Purkinje served that climbing fiber inputs to suggest that this general cells in the intermediate cerebellar Purkinje cells were activated in trials class of events, namely a cortex to examine their responses in which a monkey made errors in when the forelimb of cats and ferrets moving a joystick to a target position mismatch between ambulating encountered a bar inter- after the relationship between the expected and mittently placed in the trajectory of movement of the joystick and the swing phase (reviewed in 6). These movement of the cursor was changed. unexpected behaviors unexpected perturbations were very Together these data suggest that this or external stimuli, is one effective in evoking the synchronous general class of events, namely a mis- activation of climbing fiber inputs to match between expected and unex- condition in which neighboring Purkinje cells within a pected behaviors or external stimuli, climbing fibers frequently sagittal zone. is one condition in which climbing The results of these studies indicate fibers frequently are activated. are activated. that climbing fibers can be activated in However, climbing fibers can be ac- response to unexpected external tivated in the absence of any apparent stimuli encountered during move- unexpected event or mismatch during controlling eye movements. Function- ment as well as to low intensity pas- volitional movements, and, as will be ally, these stimuli are comparable to sive stimuli applied to the body sur- discussed below, they may be capable the movement of images on the retina face. Although this condition is an of encoding specific aspects of a sen- resulting from a discrepancy between adequate one for activating these affer- sory stimulus rather than merely indi- ent fibers, other studies indicate that cating the occurrence of a specific the velocity of the eye movement actu- this is not the only condition under condition or event. Interestingly, even ally performed and the movement re- which climbing fibers respond during when specific features of sensory quired to hold the image stationary. the performance of limb movements. stimuli are encoded, their activation An example of one of these experi- 9 For example, studies such as those still may result from an error signal ments is shown in Figure 3. Spots of performed in Mano’s laboratory7 have such as retinal slip (see the next sec- light are moved across the visual field shown that climbing fibers can be tion). Consequently, hypotheses based in specific directions by rotating a activated consistently during the ex- on error or event detection have sub- sphere either clockwise (CW) or coun- ecution of volitional, goal-directed stantial support in a wide variety of terclockwise (CCW) on the horizontal movements in the absence of exter- studies. axes indicated around the rabbit’s nally applied perturbations. However, head. The histograms at the end of even during these behaviors there is Climbing Fibers as Encoders of each axis illustrate the responses to 15 some indication that these afferents rotations of the globe. As indicated in are responding to a mismatch be- Sensory Information the key, the first half of the histogram tween intended movement and the Studies supporting the event hypoth- displays responses to the CW rotation, movement actually performed. For ex- esis reviewed in the previous section and the second half to the CCW rota- REVIEW THE ANATOMICAL RECORD (NEW ANAT.) 121

In general, lesions of the olivocerebellar projection resulted in an enhanced discharge rate of Purkinje cells and an associated decrease in the spontaneous discharge rate of nuclear neurons. Interestingly, this effect is relatively immediate and could be produced shortly after reversibly suppressing the output of the inferior olive using a cooling probe. The functional role of this tonic action of the climbing fiber system has not been fully elucidated. Subsequent studies of Strata’s group demonstrated that the tonic effects of olivary cooling are correlated with a change in the dynamic characteristics of the ves- tibulo-ocular reflex (VOR), the reflex mediating conjugate eye movements in response to movement of the head. At the very least, this observation and the fact that a cerebellar syndrome can be produced by chronic olivary lesions13 indicate that the climbing

At present it is not known whether climbing fibers Figure 3. The response properties of complex spikes recorded in the rabbit flocculus in responding during limb response to moving visual stimuli. (Reproduced from Leonard et al.9 with permission of the publisher). movements also encode specific aspects of a tion. Notice that the magnitude of the passive cutaneous stimuli, it certainly sensory signal. response to either stimulus as well as is feasible that some degree of encod- its selectivity to the CW and CCW ing exists related to a specific aspect of rotations is very dependent on the axis the stimulus (e.g., direction) despite of rotation. These data provide evi- the all-or-nothing character of the Pur- fiber system is critical for the normal dence that the responses of a given kinje cell responses to brief cutaneous function of the cerebellum even though climbing fiber input to a Purkinje cell stimuli. Population encoding also may the afferent system responsible for can reflect a very specific feature of a play a role. It also is possible that the modulating the simple spike activity sensory stimulus. However, as pointed spatial features of the receptive fields of Purkinje cells, the mossy fiber sys- out above, the stimuli themselves con- and their relationship on the body tem, is intact. These profound func- stitute a type of error signal, and surface convey important information tional effects produced by reversible consequently these data are quite con- during the generation of limb move- and irreversible olivary lesions could sistent with the general concepts ad- ments.11 be due to a disruption of the heterosy- dressed in the previous section. naptic actions of the climbing fibers, which will be discussed more exten- At present it is not known whether Climbing Fibers as Regulators of climbing fibers responding during limb sively below. movements also encode specific as- Tonic Activity pects of a sensory signal. For example, Intriguing insights into a somewhat Climbing Fibers in the although a somatotopic organization unique action of climbing fibers were Generation of Complex Motor of the climbing fiber input to the cer- generated by the laboratories of Strata ebellar cortex has been well estab- and Batini (reviewed in reference 12). Sequences lished,10 the quantitative relationship Their studies demonstrated that per- The concept that the climbing fiber between stimulus parameters and the manent or temporary lesions of the system can participate directly in the activation of these afferents has not inferior olive produced clear modifica- generation of phasic movements and been examined systematically. Based tions in the tonic or background dis- motor sequences is in part based on on the extensive literature characteriz- charge rate of Purkinje cells and their early studies showing that the admin- ing the responses of climbing fibers to target neurons in the cerebellar nuclei. istration of harmaline, a drug that 122 THE ANATOMICAL RECORD (NEW ANAT.) REVIEW

Figure 4. The distribution of correlated complex spike activity among 25 simultaneously recorded Purkinje cells, indicated as dots of varying sizes, aligned within several saggital zones. The diameter of the dot is proportional to the magnitude of correlation of discharge between cell M and other neurons. Recordings were obtained during successive 1 min epochs (A–F) and after the administration of harmaline (G). (Reproduced from Sasaki et al.14 with permission of the publisher.)

induces a synchronous rhythmic dis- among the complex spike responses of Purkinje cells recorded during a vari- charge among olivary neurons, pro- Purkinje cells oriented sagittally over ety of motor behaviors in monkeys. duces phasic, tremor-like movements the successive time intervals. The ad- Future studies undoubtedly will re- of the extremities. Furthermore, there ministration of harmaline (Fig. 4G) solve this controversy. was a high correlation between the substantially enhanced the degree of rhythmic discharge of motor units and correlation along the same sagittal The Heterosynaptic Action of the complex spike responses of Pur- band without disrupting the sagittal kinje cells. This fundamental property organization of the correlated neu- Climbing Fibers of the olivocerebellar system was pur- rons. Perhaps the most interesting and sued further by the laboratory of More recently, Llina´s, Welsh, and widely discussed concepts of climbing Llinas, who used multiple single unit collaborators15 discovered a high cor- fiber function are those pertaining to recording techniques to demonstrate relation between a naturally rhythmic the action of these afferents on modu- directly that Purkinje cells located movement, licking in the rat, and the lating the simple spike activity of Pur- within a sagittal zone exhibit highly complex spike discharge of Purkinje kinje cells. An interest in this general correlated, synchronous activity.14 cells. Based on these findings and their issue appeared with the first discus- These effects are very clear in the previous studies of the olivocerebellar sions of climbing fiber function based experiment shown in Figure 4. In this system, these authors proposed that on the properties of the inactivation experiment, the spontaneous discharge the inferior olive contributes to the response of Purkinje cells and the of 25 Purkinje cells was recorded over specification of the combinations of pause in simple spike activity that successive 1 min intervals, and the fundamental output units of the mo- often follows it. The read-out hypoth- correlation in the discharge between tor system required for the execution esis of Eccles and colleagues also was cell M and the other neurons was of complex movements and move- based on the effects of the climbing measured and plotted (Fig. 4A–F). The ment sequences. Thach and collabora- fiber on features of the responses magnitude of the correlations is indi- tors16 have challenged aspects of this evoked by the mossy fiber–parallel fi- cated by the diameter of the circles view by reporting that the rhythmic ber system. representing the location of each cell. behavior of complex spike discharges More recently, heterosynaptic ef- Notice the consistent correlation was not present in the discharge of fects of climbing fiber inputs have REVIEW THE ANATOMICAL RECORD (NEW ANAT.) 123 been proposed as the basis for their input to the same population of cells quence of stimulus paradigms re- action on the simple spike activity of resulted in a long-lasting depression of stricted to the parallel fibers. These Purkinje cells. The hypotheses based Purkinje cell excitability, a phenom- studies demonstrate that the action of on these effects assume that these enon termed long-term depression or the climbing fibers per se is not essen- afferents are capable of inducing a LTD.19 To further support this hypoth- tial for LTD to occur. change in the synaptic effectiveness of esis, these same investigators demon- The behavioral evidence supporting the parallel fiber input to Purkinje cell strated that the depression was selec- LTD in learning also is equivocal. Al- dendritic trees, resulting in a change tive for parallel fiber inputs that were though initial reports by Gilbert and in their responsiveness to inputs medi- activated during the application of the Thach22 argued that changes in com- ated by the mossy fiber–parallel fiber conjunctive stimuli. Following the es- plex and simple spike modulation oc- system. Two classes of heterosynaptic tablishment of LTD of Purkinje cells to curred in a manner consistent with effects have been proposed: (1) a long- the parallel fiber input activated dur- the Marr-Albus hypothesis, the more lasting change related to the establishing conjunctive stimulation, no compa- recent quantitative studies of Ojakan- ment of memory traces in the cerebel- rable depression was found for the gus and Ebner8 did not support this lar cortex and (2) a short-lasting responses to parallel fibers that were view. These investigators examined ex- change in the responsiveness or gain not activated during the conjunctive plicitly the changes in simple and com- of the Purkinje cells to parallel fiber stimulation paradigm. These observa- plex spike modulation during the inputs. tions established LTD as a potential learning of goal-directed reaching mechanism for generating long-term movements requiring that the animal plastic changes in the cerebellar cor- learn an altered relationship between Long-term heterosynaptic action tex during the learning of motor tasks. the movement of a manipulandum of climbing fibers—long-term These observations led to detailed and the movement of a cursor on a depression investigations into the cellular mecha- screen. In brief, the changes in the Interest in the possibility that climb- modulation of Purkinje cells were not ing fibers could induce a long-lasting, found to be consistent with the Marr- persistent change in Purkinje cell excit- Albus hypothesis. In another recent ability through the action of climbing More recently, behavioral experiment, Schreurs and fibers was stimulated by the early theo- heterosynaptic effects of colleagues23 demonstrated that LTD- related changes did not occur as a retical work of David Marr in the late climbing fiber inputs 1960s17 and led to the well-known consequence of the classical condition- Marr-Albus hypothesis. Marr proposed have been proposed as ing of the eye-blink reflex in the rabbit. that, when climbing fibers activate the the basis for their action In fact, a persistent decrease in excit- dendrites of Purkinje cells coincident ability (i.e., increased responsiveness) with the arrival of parallel fiber inputs on the simple spike of Purkinje cells was noted within a responding to behaviorally relevant in- activity of Purkinje cells. specific sagittal zone in lobule HVI of puts from the mossy fiber projections, the cerebellar cortex. Interestingly, al- a memory trace is established. Func- though these findings fail to support a tionally, this interaction was proposed contribution of LTD to the classical to reduce the responsiveness of Pur- nisms responsible for LTD (reviewed conditioning of this reflex, they do kinje cells to the same subset of paral- in references 20,21). Although many support a long-lasting change in excitability of Purkinje cell dendrites asso- lel fibers when these fibers were reacti- interesting phenomena related to a ciated with learning. vated repeatedly during the learning variety of second messenger systems Recent studies employing genetic of a movement. This reduction in re- have been identified and described, sponsiveness was believed to be persis- knockout techniques in mice have not the relationship of LTD to memory tent, representing the consequence of been successful in providing uniform processes still is not clear. The funda- establishing an engram in the cerebel- support for a causal link between LTD mental studies on the cellular mecha- lar cortex required for the retention and the learning of specific tasks. A nisms underlying long-term depres- and eventual recall of previously recent review by Llinas et al.24 notes sion certainly have provided strong learned motor patterns. that studies employing this technique This postulate was the basis for an evidence that associative processes, do not show parallel effects between extensive series of experiments per- such as those occurring as a conse- the magnitude of changes in LTD and formed by Ito and his colleagues.18 quence of pairing conditioned and un- the extent of the learning assessed in These investigators designed a para- conditioned stimuli in paradigms such the experiment. In summary, although digm, the conjunctive stimulation as classical conditioning, are not nec- available data provide initial evidence paradigm, which they felt imple- essary to produce this phenomenon. favoring the existence of long-term mented the conditions required to test For example, it can be evoked by bath- modifications in the excitability of Pur- adequately the provocative proposal ing the dendrites of the Purkinje cell in kinje cells associated with learning a of Marr. These studies demonstrated a slice preparation with putative exci- motor task, additional evidence will be that the application of a stimulus acti- tatory transmitters that result in a required before LTD can be consid- vating the granule cell–parallel fiber massive depolarization of that cell. ered to play a fundamental role in input together with the climbing fiber LTD even has been reported as a conse- acquisition and retention processes. 124 THE ANATOMICAL RECORD (NEW ANAT.) REVIEW

Short-term heterosynaptic teristics of this effect on the responsive- action of climbing fibers ness of Purkinje cells are quite different than those of LTD. Rather than Our laboratory discovered that a spon- producing a long-term depression, a taneous climbing fiber input not only short-lasting enhancement of the evokes a pause in the simple spike simple spike responses was observed. activity but also may result in an in- These observations led to the ‘‘gain crease or decrease in the discharge change’’ hypothesis, which proposed rate of Purkinje cells once its simple that the occurrence of climbing fiber spike activity resumes (for review see inputs could produce a short-lasting 25 Bloedel and Ebner ). Furthermore, increase in the gain of a Purkinje cell’s our laboratory demonstrated that the response to its other primary input, action of these afferents can modify the mossy fiber–parallel fiber projec- the autocorrelation of the simple spike tion. In order to address this concept discharge after the pause. in a more general framework, Bloedel Observing these changes in simple and Kelly6 proposed the ‘‘dynamic se- spike discharge rate following the lection’’ hypothesis. This viewpoint was pause suggested to Ebner and Bloedel intended to synthesize the specific ob- that the climbing fiber input actually servations on which the gain change may be capable of modifying the short- hypothesis was based with the funda- term responsiveness of Purkinje cells mental morphological and physiologi- to parallel fiber inputs. Based on this cal organization of the climbing fiber premise, we performed an extensive and mossy fiber projections to the series of studies demonstrating that cerebellar cortex. the modulation of simple spike re- The features of this concept are illus- sponses evoked after the occurrence of trated in Figure 6. In addition to incor- the cell’s climbing fiber input is en- porating the gain change hypothesis, hanced and that this enhancement is this view addresses an interesting and short-lasting, decaying over a few hun- Figure 5. The effect of climbing fiber inputs sometimes enigmatic feature of cer- dred milliseconds at maximum. These on the simple spike activity of a Purkinje cell ebellar organization. In brief, climb- changes in the responses were very evoked by a natural cutaneous stimulus. A,B: ing fiber inputs from specific locations unusual compared to effects that had Simple spike histograms constructed only from within the inferior olive are well orga- been described previously, since they trials in which climbing fiber inputs were not or nized in sagittal bands that are somato- were evoked, respectively. C: Histogram of all did not substantially change the pro- the complex spike responses across all trials. topically organized. In contrast, al- file of the response and could consist D: Histogram of the simple spike responses though there is some banding of mossy of an enhancement of both excitatory across all trials. E: Simple spike responses fiber inputs, receptive fields represent- and inhibitory response components. aligned on the time of occurrence of the ing a given region of the body tend to An example of this effect is shown in complex spikes in each trial in which they be distributed over multiple sites on occurred. n, number of trials from which the Figure 5. The responses shown in these the cerebellar cortical surface extend- indicated histogram is constructed; RE, rela- histograms were evoked by a natural tive change in the amplitude of the excita- ing across different sagittal zones. As a cutaneous stimulus applied on the tory response in trials in which the climbing consequence, it remains unclear how forelimb. Note that the response con- fiber input occurred compared to the ampli- information pertinent to a given body sisted of both an excitatory and an tude of the same response component in region is processed via the mossy fiber trials in which no climbing fiber inputs were inhibitory component. The simple system in a manner that implements evoked; RI, relative change in the amplitude spike responses first were sorted based of the inhibitory response component in trials the strict topographical organization on whether or not a complex spike was in which the climbing fiber input was evoked. of the cerebellar nuclei. This is impor- evoked by the stimulus. Next, the am- Pairs of arrows along the x-axis indicate the tant because this afferent projection is plitude of both the excitatory and the response windows in which the response am- responsible for generating the graded inhibitory response components in the plitudes were measured. (Reproduced from Bloedel and Kelly6 with permission of the pub- modulation of Purkinje cell activity trials in which complex spikes were lisher.) during the control of a variety of mo- evoked (Fig. 5B) was calculated rela- tor behaviors. tive to the amplitude of the same The distribution of activated mossy components in the histogram con- was activated. The temporal relation- fiber and climbing fiber inputs is structed from trials in which no com- ship between the complex spike input shown for three different conditions plex spikes were evoked by the stimu- and the simple spike response is shown (Fig. 6) to illustrate how the dynamic lus (Fig. 5A). The fact that both RE and in Figure 5E, a histogram constructed selection hypothesis addresses this RI are greater than 100% indicates by aligning all trials on the occurrence problem. It is assumed that a stimulus that the excitatory as well as the inhibi- of the complex spike rather than on encountered by the paw activates the tory response components were en- the time at which the cutaneous stimu- same distribution of mossy fiber in- hanced in those trials in which the lus was applied. puts in each condition. However, be- climbing fiber input to the same cell These data indicate that the charac- cause of the well-known convergence REVIEW THE ANATOMICAL RECORD (NEW ANAT.) 125

In general, these concepts fall into two categories: those related to the online processing of the cerebellar cortex and those pertaining to the establishment of long-term modifications in responsiveness relevant to the establishment of memory engrams. The data clearly indicate that the climbing fiber system is relevant to the online processing performed by the cerebellum. At present, the data implicating these afferents in the establishment of memory traces is yet not conclusive. The studies reviewed here strongly suggest that long-term changes in ex- Figure 6. Explanation of the dynamic selection hypothesis. Oval-like patches on the cerebel- citability can occur in Purkinje cells in lum represent the distribution of mossy fiber inputs evoked by a stimulus applied to the same association with learning. However, location on the paw under each of the three different functional conditions indicated above these changes were not consistent with each panel. Stripe-like areas indicate the location of climbing fiber inputs activated under the a role for LTD, and the findings on this same three conditions. (Reproduced from Bloedel and Kelly6 with permission of the publisher.) issue from the gene knockout experiments remain equivocal. It remains of ascending and descending projec- selection hypothesis should be re- tions to the inferior olive, it also is garded as a conceptual framework that assumed that a somewhat different undoubtedly will require modification distribution of climbing fiber inputs is as additional information regarding Like most current activated in each, reflecting the spe- the climbing fiber system is acquired. viewpoints, the dynamic cific behavioral conditions under It also must be acknowledged that the selection hypothesis which they are evoked. This condition- short-term enhancement described specific distribution of activated climb- above has not been observed under all should be regarded as a ing fiber inputs in turn determines the functional conditions. Nevertheless, conceptual framework location of those Purkinje cells most short-term enhancement has been responsive to the ongoing modulation demonstrated in response to passive that undoubtedly will of their inputs from parallel fibers and stimuli and in response to stimuli en- require modification as inhibitory interneurons. This is de- countered during active movements, picted in Figure 6 as the differences in as reviewed above, and the possibility additional information the regions of overlapping climbing that the occurrence and even the rel- regarding the climbing and mossy fiber inputs under the three evance of these interactions may be conditions. Given the strict topo- task-dependent should not be over- fiber system is acquired. graphical organization of the cortico- looked. nuclear projection, this in turn would specify the organization of the output projections from the cerebellar nuclei CONCLUSIONS feasible that long-term changes re- that would be most highly modulated To borrow a phrase commonly em- lated to learning occur in the cerebel- in relation to each behavior. ployed in Lake Wobegon, the ‘‘horse lum but that they are not established This view accounts for the fact that remains out of the barn’’ with regard as a consequence of the climbing fi- the climbing fibers can be activated in to a generally accepted and well- bers’ actions. In fact, the strongest a number of functional contexts, and substantiated viewpoint characteriz- data for learning-related plastic it provides one of the few postulates ing the precise function of the climb- changes in the cerebellum implicate that integrates the topography of the ing fibers. Nevertheless, the spectrum the target nuclei of the Purkinje cells corticonuclear projection, the recep- of studies reviewed above provides a (the vestibular and cerebellar nuclei) tive field distribution of mossy fiber basis for optimism, since the data rather than the cerebellar cortex. Inter- inputs, and the organization of the reveal important fundamental fea- estingly, the projections of the climb- olivocerebellar system with several tures of this system as well as critical ing fibers to the deep nuclei have been functional properties of this projec- interrelationships among a variety of largely overlooked in most discussions tion. It also integrates the short-term experimental observations and even regarding the function of these affer- heterosynaptic action of Purkinje cells among the multiple concepts address- ents. with the tendency for these afferents ing the function of this system. In our It also should be emphasized that to be activated synchronously in a view, these postulates and the findings the occurrence of plastic changes of sagittal distribution under specific be- on which they are based reveal the any type may be highly task-depen- havioral conditions. However, like spectrum of relevant conditions in dent, as emphasized by several of our most current viewpoints, the dynamic which climbing fibers are activated. recent studies.26,27 For example, block- 126 THE ANATOMICAL RECORD (NEW ANAT.) REVIEW ing protein synthesis in the ipsilateral vermis (b Zone) of the cat cerebellum dur- 17 Marr D (1969) A theory of cerebellar cerebellar interposed nuclei in the rab- ing locomotion. J Physiol 385:107–134. cortex. J Physiol (Lond), 202:437–470. bit suppresses the acquisition of 6 Bloedel JR, Kelly TM (1992) The dynamic 18 Ito M (1993) Synaptic plasticity in the memory traces required for the reten- selection hypothesis: A proposed function cerebellar cortex and its role in motor for cerebellar sagittal zones. In Llinas R, Can J Neurol Sci 20(Suppl. 3):S70– tion of the classically conditioned eye- learning. Sotelo C (eds): The Cerebellum Revisited. S74. blink reflex. In contrast, cats can ac- New York: Springer-Verlag pp 267–282. 19 Ito M (1989) Long-term depression. quire complex, volitional forelimb 7 Mano N-I, Kanazawa I, Yamamoto K-I Annu Rev Neurosci 12:85–102. movements while the ipsilateral den- (1986) Complex spike activity of cerebellar 20 Linden DJ, Connor JA (1995) Long-term tate and interposed nuclei are inacti- Purkinje cells related to wrist-tracking synaptic depression. Annu Rev Neurosci vated by the injection of muscimol. movement in monkey. J Neurophysiol 56: 18:319–357. 137–158. Thus, although these data support the 21 Crepel F, Hemart N, Jaillard D, Daniel H feasibility of plastic changes in the 8 Ojakangas CL, Ebner TJ (1992) Purkinje (1996) Cellular mechanisms of long-term cell complex and simple spike changes depression in the cerebellum. Behavioral cerebellar nuclei related to eye-blink during a voluntary arm movement learning and Brain Sciences 19:347–353. conditioning in the rabbit, they do not task in the monkey. J Neurophysiol 68:2222– support a critical intracerebellar stor- 2236. 22 Gilbert PFC, Thach WT (1977) Purkinje cell activity during motor learning. Brain age site related to the learning of the 9 Leonard CS, Simpson JI, Graf W (1988) Res 128:309–328. type of complex forelimb movements Spatial organization of visual messages of 23 Schreurs BG, Tomsic D, Gusev PA, Al- employed in our experiments. the rabbit’s cerebellar flocculus. I. Topology kon DL (1997) Dendritic excitability micro- In summary, we believe that a con- of inferior olive neurons of the dorsal cap of Kooy. J Neurophysiol 60:2073–2089. zones and occluded long-term depression tinued focus on integrating the funda- 10 Robertson LT (1987) Organization of after classical conditioning of the rabbit’s mental morphological and functional climbing fiber representation in the ante- nictitating membrane response. J Neuro- properties of the climbing fibers in rior lobe. In: New Concepts in Cerebellar physiol 77:86–92. experiments using a variety of behav- Neurobiology. New York: Alan R. Liss, Inc., 24 Llina´s R, Lang EJ, Welsh JP (1997) The ioral models will result in substantial pp 281–320. cerebellum, LTD, and memory: Alternative progress in uncovering the elusive 11 Robertson LT, McCollum G (1991) views. Learning and Memory 3:445–455. function of this intriguing system. Stimulus classification by ensembles of 25 Bloedel JR, Ebner TJ (1985) Climbing climbing fiber receptive fields. Trends Neu- fiber function: Regulation of Purkinje cell roscI 14:248–254. responsiveness. In Bloedel JR, Dichgans J, LITERATURE CITED 12 Strata P (1985) Inferior olive: Func- Precht W (eds): Cerebellar Functions. Ber- tional aspects. In Bloedel JR, Dichgans J, lin, Heidelberg, New York: Springer-Verlag, 1 Eccles JC, Ito M, Szentagothai J (1967) Precht W (eds): Cerebellar Functions. Ber- pp 247–259. The Cerebellum as a Neuronal Machine. lin, Heidelberg, New York, Tokyo: Springer, 26 Bloedel JR, Bracha V, Shimansky Y, Berlin: Springer-Verlag. pp 230–246. Milak MS (1996) The role of the cerebellum 2 Oscarsson O (1980) Functional organiza- 13 Carrea RME, Reissig M, Mettler FA in the acquisition of complex volitional tion of olivary projection to the cerebellar (1947) The climbing fiber of the simian forelimb movements. In Bloedel JR, Ebner anterior lobe. In Courville J, De Montigny cerebellum. J Comp Neurol 87:321–367. TJ, Wise SP (eds): The Acquisition of Motor C, LaMarre Y (eds): The Inferior Olivary Behavior in Vertebrates. Cambridge: MIT Nucleus: Anatomy and Physiology. New York: 14 Sasaki K, Bower JM, Llinas R (1989) Raven Press, pp 279–289. Multiple Purkinje cell recording in rodent Press, pp 319–342. 3 Jo¨ rntell H, Garwicz M, Ekerot CF (1996) cerebellar cortex. Eur J Neurosci 1:572– 27 Bracha V, Bloedel JR (1996) The mul- Relation between cutaneous receptive fields 586. tiple-pathway model of circuits subserving and muscle afferent input to climbing fi- 15 Welsh JP, Lang EJ, Sugihara I, Llina´s R the classical conditioning of withdrawal bres projecting to the cerebellar C3 zone in (1995) Dynamic organization of motor con- reflexes. In Bloedel JR, Ebner TJ, Wise SP the cat. Eur J Neurosci 8:1769–1779. trol within the olivocerebellar system. Na- (eds): The Acquisition of Motor Behavior in 4 Gellman R, Gibson AR, Houk JC (1985) ture 374:453–457. Vertebrates. Cambridge: MIT Press, pp 175– Inferior olivary neurons in the awake cat: 16 Keating JG, Thach WT (1995) Nonclock 204. detection of contact and passive body move- behavior of inferior olive neurons: Inter- 28 Bloedel JR (1986) A review of cerebellar ment. J Neurophysiol 54:40–60. spike interval of Purkinje cell complex spike anatomy and physiology and their relation- 5 Andersson G, Armstrong DM (1987) Com- discharge in the awake behaving monkey is ship to cerebellar deficits. BNI Quarterly, plex spikes in Purkinje cells in the Lateral random. J Neurophysiol 73:1329–1340. 2:24–29.