Progressive limb ataxia following inferior olive lesions


A. R. Gibson: Division of Neurobiology, Barrow Neurological Institute, 350 West Thomas Road, Phoenix, AZ 85013, USA. Email:

Key points

  • • The inferior olive (IO) provides climbing fibre input to the cerebellum. Kainic acid lesions of rostral IO of the cat produce several distinct movement deficits.
  • • Vestibular disturbances are apparent the first few days after injection but rapidly recover.
  • • Disturbances of grasping occur immediately and do not recover over months of testing.
  • • After a brief delay, limb movements during reaching and locomotion show a progressive development of ataxia that becomes severe over months of testing.
  • • The decomposition of movement during reaching and alterations in reach trajectories are similar to those seen in humans with cerebellar ataxias – degeneration of the IO may contribute to the progressive nature of these diseases.

Abstract  Cerebellar climbing fibres originate in the inferior olive (IO). Temporary IO inactivation produces movement deficits. Does permanent inactivation produce similar deficits and, if so, do they recover? The excitotoxin, kainic acid, was injected into the rostral IO of three cats. Behaviour was measured during reaching and locomotion. Two cats were injected during the reaching task. Within minutes, grasping became difficult and the trajectories of the reaches showed higher arcing than normally seen. During locomotion, both cats showed head and trunk deviation to the injected side, walking paths curved to the injected side, and the paws were lifted higher than normal. Limbs contralateral to the injections became rigid. Within 1 day, posture had normalized, locomotion was unsteady and high lifting of the paws had reversed to a tendency to drag the dorsum of the paws. Passive body movement produced vestibular signs. Over a few days, locomotion normalized and vestibular signs disappeared. Reach trajectories were normal but grasping deficits persisted. Over the first week, the amplitude of limb lift during reaching and locomotion began to increase. The increase continued over time and, after several months, limb movements became severely ataxic. The effects followed the somatotopy of the rostral IO: a loss of cells in medial rostral IO only affected the forelimb, whereas a loss of cells in medial and lateral IO affected both forelimb and hindlimb. Deficits produced by IO lesions involve multiple mechanisms; some recover rapidly, some appear stable, and some worsen over time. The nature of the progressive deficit suggests a gradual loss of Purkinje cell inhibition on cerebellar nuclear cells.


climbing fibre


dorsal lamellae of principal olive






inferior olive


kainic acid


post day


principal olive


rostral dorsal accessory olive


rostral medial accessory olive


magnocellular red nucleus


ventral lamellae of principal olive


Lesions of the cerebellum, as well as other brain structures, produce movement deficits that typically show partial initial recovery followed by a residual deficit (Konczak et al. 2010). Cats with cerebellar cortical or nuclear lesions are unable to walk on a treadmill after injury but show rapid improvement and, in a relatively short time, walk normally (Yu & Eidelberg, 1983). More extensive lesions of cerebellar cortex and/or nuclei show varying degrees of compensation with the greatest compensation occurring over the first few weeks, while some deficits persist for many months (Chambers & Sprague, 1955a,b).

The cerebellum is critical for motor adaptation (Reisman et al. 2010); given this, it is surprising that motor deficits after cerebellar lesions show a significant amount of recovery. One possible explanation is that the anatomical organization of the cerebellum makes it unlikely that a lesion will include the entire neural circuitry serving a given function: climbing fibre (CF) terminations from inferior olive (IO) neurons define narrow parasagittal zones of Purkinje cells in cerebellar cortex that include the entire anterior–posterior extent of the cortex (Groenewegen & Voogd, 1977; Groenewegen et al. 1979). Typically, lesions do not include an entire parasagittal zone, and remaining portions of the zone may mediate recovery.

Temporary inactivations of subdivisions of the IO produce movement deficits unique to the specific subdivision (Horn et al. 2010). In the absence of CF input, Purkinje cells initially discharge simple spikes at a high rate (Colin et al. 1980; Rawson & Tilokskulchai, 1981; Montarolo et al. 1982; Strata & Montarolo, 1982). Simple spikes inhibit nuclear cells (Ito et al. 1964), so a high rate of discharge effectively turns off cerebellar output (Benedetti et al. 1983). Presumably, the distinct deficits seen with IO inactivation are due to disruption of entire cerebellar processing modules (Voogd & Bigare, 1980). Therefore, permanent destruction of the IO might lead to specific movement deficits that show little or no recovery.

In the present study, we made cellular lesions of the IO by placing injections of kainic acid (KA), a glutamate agonist excitotoxin, into forelimb-related regions of the rostral medial accessory olive (rMAO). Surprisingly, the resulting movement deficits, while initially sharing some similarity with those produced by temporary inactivation, rapidly change in nature and new deficits emerge (Horn et al. 2010). The deficits are also different from those reported for electrolytic or surgical lesions of the IO (see Discussion). The post-lesion behavioural deficits suggest that several different neural mechanisms are involved, and the various mechanisms can have entirely different time courses.

The most striking effect after IO lesion is a progressive increase in limb flexion during reaching and locomotion. The effect begins after a short post-lesion delay and progresses for an extended time over many months. With prolonged survival, limb movements become severely ataxic, although other body movements and posture appear normal. The effects are specific to the somatotopy of the IO: a lesion confined to medial rostral IO only affects forelimb movements, whereas lesions involving medial and lateral regions of rostral IO affect hindlimb as well as forelimb movements. Although the rostral dorsal accessory olive (rDAO) has the clearest somatotopy, both the rostral medial accessory olive (rMAO) and principal olive (PO) cells related to forelimb stimulation are predominately located in medial regions and cells related to hindlimb stimulation are predominately located in lateral regions (Boesten & Voogd, 1975; Berkley & Hand, 1978; Gellman et al. 1983; Gellman et al. 1985). At least for rDAO, the sensory somatotopy corresponds directly to motor output (Gibson et al. 1987). It is likely that studies designed to understand the mechanisms underlying this progressive deficit will provide a deeper understanding of cerebellar control of movement as well as IO function. The findings may also help elucidate progressive motor deficits seen with neurological diseases that involve the cerebellum.


Ethical approval

The research was approved by the Institutional Animal Care and Use Committee and was conducted in accordance with both the National Institutes of Health's Principles of Laboratory Animal Care (86-23, revised 1985) and the American Physiological Society's Guiding Principles in the Care and Use of Animals. The experiments comply with the policies and principles outlined by UK legislation and The Journal of Physiology (Drummond, 2009).


Three adult male cats (average weight 4.3 kg with a range of 4.1–4.4 kg; average age at the beginning of the experiment was 2.9 years with a range of 2.4–3.3 years) from a US Class B licensed supplier served as subjects. The methodology outlined in the current study is very similar to that used by Horn et al. (2010).

Experimental procedures

The cats were trained to reach out and grasp a handle to receive a food reward upon presentation of a tone. The cats kept their forelimbs placed on pedestals equipped with pressure transducers during a variable inter-trial interval (4–8 s). The cats sat in a nylon vest (Chatham Medical Arts). Wrist position was recorded in three-dimensional space using a reflective band (1.5 cm width) attached to the cat's wrist that was monitored using an infrared reflective system (Dynasight, Origin Instruments). The tracking system was modified (device designed and builty by Dr. Robert Scobey) to produce analog signals of the horizontal, vertical and lateral axes at 64 Hz. Signals were acquired, stored and analysed with a CED 1401+ (Cambridge Electronics Device).

The cats were also trained to walk along a 3 m runway to receive a food reward. Obstacles (2.5 cm square bars) were placed across the runway for blocks of 10–20 trials. The forelimb position during walking was analysed using individual digital video frames (720 × 480 pixels) acquired from video records sampled at 30 frames s-1 (Canon XL1 and GL2 video camcorders). Higher quality images of the cats during both the reach-to-grasp and locomotion tasks were made using a SLR digital camera (Canon EOS-1X). Vestibular responses were tested by picking the cats up by the trunk and gently swinging side-to-side laterally in an arc (about 75 cm, 0.5 Hz) for about 5 s. Although the vestibular testing was subjective, normal cats are undisturbed by the swinging, whereas cats with vestibular dysfunction react strongly and are unsteady when returned to the floor.


After the training period (2–8 weeks), a Narishige-type recording chamber and a head restraint device were fastened to the skull with titanium screws and dental acrylic during a recovery surgery. Surgery was performed in an American Association of Laboratory Animal Care approved surgical suite using aseptic techniques. Each cat was anaesthetised with ketamine (15 mg kg−1, intramuscular injection) followed by supplemental intravenous injections of sodium pentobarbital (5 mg as needed based upon level of anaesthesia). The anaesthetic level was monitored by a technician and included regular recordings of respiration rate, cardiac rate, blood oxygen levels, pupillary dilation and withdrawal responses. Local infusions of a long-lasting local anaesthetic (bupivacaine, 2 mg kg−1) were made at 1 cm intervals into the scalp prior to the incision. A second series of bupivacaine injections was made at the conclusion of the surgery into the edges of the incision and retracted muscles. A midsagittal scalp incision was made, the muscles on the dorsal surface of the skull were retracted, and the skull overlying the cerebellum was removed. Bone channels in the skull were sealed with bone wax and the recording/injection chamber was sealed to the skull with dental acrylic. Post-operatively, the cat received antibiotics (ampicillin, 10–20 mg kg−1) and steroids (dexamethasone, 10 mg initial with tapering amounts) for 5 days.

Localisation of injection site

Accurate localisation of injection sites requires physiological verification. To do this, we localized the appropriate IO region using tungsten microelectrodes. IO cells are readily identified by their characteristic action potentials and discharge patterns (Armstrong et al. 1968; Crill, 1970; Gellman et al. 1983). The most dorsal IO layer is the rDAO, which contains cells sensitive to light cutaneous stimulation (Gellman et al. 1983). rDAO cells are arranged so that their receptive fields provide a detailed somatotopy of the contralateral body surface. Therefore, rDAO receptive fields provide a guide to the underlying rMAO. Although receptive fields in rMAO can be complex, cells in the medial half of the subnucleus respond to squeezes and taps of the forelimb (Gellman et al. 1985). Once the coordinates of forelimb rMAO were determined, the metal electrode was replaced with a pipette electrode that could be used for injection as well as recording.

Kainic acid injection

In two cats, CIO8 and CIO10, KA in saline (250 nl, 4.5 mm) was injected into the identified forelimb portion of rMAO using injection/recording pipettes that were fabricated from polyamide insulated quartz tubing (670 μm outside diameter, 10 μm tip opening). The injections were made over 5 min periods while the cats were performing the reach-to-grasp task. Injection volumes were monitored by observing movement of the KA meniscus within a short piece of calibrated tubing attached to the pipettes. Injection pipettes were left in place for 5 min after the completion of the injections.

The only previous report of KA injections into the cat IO (Pompeiano et al. 1981) used 10.8 mm concentrations; these injections produced a 25% mortality rate. Therefore, we chose a lower concentration. We have experienced no deaths using 4.5 mm KA, and it has proven very effective for destruction of IO neurons in the awake cat. We chose to inject rMAO rather than rDAO so that dorsal spread of the KA would be less likely to damage overlying reticular neurons. Outputs from forelimb regions of rMAO and rDAO converge onto regions of the magnocellular red nucleus (RNm) that project to the cervical spinal cord (Gibson et al. 1987; Robinson et al. 1987).

Behavioural testing on both tasks was done immediately after the completion of the KA injections. KA led to increased behavioural activity, and the resulting brainstem movement precluded recording IO activity after the injection. During locomotion testing, the effects of the KA injection increased and eventually led to an inability to maintain an upright posture. At this point, acepromazine (1.0 mg kg−1, intramuscular injection) was given to sedate the cats.

In a third cat, CIO9, we attempted to produce a more selective lesion using a smaller volume of less concentrated KA (125 nl, 2.5 mm). To gain more precise control of injection volume, the injection was made while the cat was anaesthetised, which allowed microscopic observation and measurement of meniscus movement in the injection pipette. The same physiological identification procedure was used to determine the injection site in this cat as was used in the behaving cats. Following injection, IO discharge increased to high rates, and the cells became extremely responsive to sensory stimulation.


At the conclusion of the testing periods (CIO8 = PD72; CIO9 = PD28; CIO10 = PD133, where PD is post-day), the cats were anaesthetised (ketamine 15 mg kg−1, intramuscular injection) and killed with an intravenous injection of sodium pentobarbital (25 mg kg−1, as a rapid bolus injection). The cats were perfused with 1 l of 9.25% sucrose at 300 mmHg followed by 2 l of 10% Formalin at 120 mmHg. The brains were embedded in gelatin and sectioned sagittally at 40 μm. Sequential brainstem sections were stained with a Nissl stain (cresyl violet or thionine). Digital images were used to reconstruct cell loss within each of the olivary subdivisions.

Cerebellar cortices were sectioned and stained with cresyl violet for examination with brightfield microscopy and, for case CIO10, also with DAPI (4′,6-diamidino-2-phenylindole) and red fluorescent Nissl (Life Technologies) for fluorescent microscopy. Sections from CIO10 were also stained with fluoro jade (Krinke et al. 2001) in an attempt to detect degenerating neurons.


Extent of IO cell loss

Three cats received unilateral injections of KA into physiologically identified forelimb regions of rMAO. IO neurons are more sensitive to KA injections than neurons in the surrounding brainstem reticular areas (Pompeiano et al. 1981); despite extensive IO cell loss in two of our cases, histological examination revealed no obvious cell loss in surrounding brainstem areas (Fig. 1). Connections of the IO are crossed so lesions of the left IO affect movements of the right limbs and lesions of the right IO affect movements of the left limbs.

Figure 1.

Parasagittal section through the IO of cat CIO8
The section is near the centre of the KA injection, and approximately 60% of the IO has been destroyed. In the lower left inset, the outline of the missing rDAO is shown by dashed lines. The reticular neurons immediately dorsal to rDAO appear to be normal. The lower right inset illustrates a region of cMAO with normal cell density. Perfusion was done on PD73.

Two cats, CIO8 and CIO10, were injected with 250 nl of 4.5 mm KA while they were performing the reaching and grasping task. Cat CIO8 had extensive cell loss in the rostral half of the left IO. Measurements indicated approximately 60% of the left IO was destroyed (Figs 1 and 2A). The right IO showed no cell loss – behavioural effects were limited to the forelimb and hindlimb contralateral to the injection (right limbs).

Figure 2.

Horizontal reconstructions of damaged IO regions with a frontal plane reconstruction at mid-rostral IO level shown at the centre
Arrows indicate the respective nuclei in the horizontal drawings. A, cat CIO8. Cell loss was limited to the rostral half of the left IO. B, cat CIO10 was perfused on PD136. On the right side, the entire rostral IO was destroyed with only the caudal 50% of cMAO intact. On the left side, medial regions of the rostral IO were destroyed. The left caudal IO and lateral regions of rostral IO were intact. The medial (M) –lateral (L) orientation of the reconstructions is noted on the upper right horizontal and frontal drawings.

Cat CIO10 was injected on the right side; the area of cell loss included essentially the entire rostral IO and about 50% of caudal IO (Fig. 2B). Measurements indicated that approximately 73% of the IO on the injected side (right) was destroyed; the contralateral (left) forelimb and hindlimb showed movement deficits.

In cat CIO10 medial portions of the left IO also showed some cell loss (reticular neurons between the olives appeared normal). The medial 25% of the left rDAO had complete cell loss with partial loss extending to approximately the medial 50%. The lateral 50% of the left rDAO appeared undamaged. About 50% of the left rMAO was totally destroyed with the lateral half having partial cell loss. The most medial parts of the lamellae of the principal olive (PO) also had cell loss (Fig. 2B). Movement deficits contralateral to this lesion were limited to the forelimb.

The third cat, CIO9, was injected with 125 nl of 2.5 mm KA while anaesthetised. Although gliosis was clearly visible along the recording tracks, no loss of IO cells could be attributed to the KA injection. The absence of damage may have been due to the lower amount injected and/or due to anaesthetic effects. Over 28 days of testing, CIO9 showed no behavioural changes, and limb movements during reaching and locomotion were unchanged from pre-injection testing.

Despite extensive cell loss in the IO of CIO8 and CIO10, light microscopic examinations of cerebellar sections failed to reveal any degenerative changes throughout the Purkinje cell layer or in cerebellar nuclear regions.

Immediate behavioural effects of KA injection into rostral IO

The two cats injected with KA during the reaching task, CIO8 and CIO10, allowed observation of immediate behavioural effects. For each cat, the injection was placed into a region of rMAO that responded to squeeze and release of the contralateral forelimb: many of the immediate behavioural effects were similar for the two cats.

Two minutes after the KA injection, each cat had difficulty grasping the handle with the forelimb contralateral to the injection (Fig. 3). After about 3 min, the cats ceased working on the task. The trajectories of the reaches also showed changes. The reach trajectories for both cats showed an upward curvature with a higher end point (Fig. 3). Only a few reaches were recorded before the cats quit working, and it was not possible to collect records of reaching with the forelimbs ipsilateral to the injections. On cessation of reaching, the cats were removed from the apparatus to observe effects on locomotion.

Figure 3.

Over-plotted wrist trajectory traces for 4 reaches immediately before and 2 min after KA injection for cats CIO8 and CIO10
The grey box at the lower left of the over-plotted traces indicates the position of the pedestal and the grey narrow rectangle at the upper right of the over-plotted traces indicates the location of the handle (see photographs in Fig. 7).

While standing or sitting, both cats showed a strong curvature of the trunk and head toward the side of injection. The cats were hesitant to walk, but when they did, the cats lifted their feet higher than normally seen, and the walking paths showed curvature toward the injected side. Cat CIO10 often held one or the other (more often the right) forelimb in the air with the elbow flexed while sitting on the hindlimbs. As the effects of the injections increased, limbs contralateral to the injection became extended and showed a high degree of tone and rigidity. Cat CIO10 also developed high tone in the forelimb ipsilateral to the injection. After 1 h the effects of the KA injections had increased to the extent that the cats were unable to maintain an upright posture, and they were sedated with injections of acepromazine (1.0 mg kg−1, intramuscular injection).

At 4.5 h following the KA injection, CIO8 no longer showed neck torsion and was able to walk, although the walk had a crouched posture with the tail curved to the left. The right forelimb and hindlimb still showed extension and high muscle tone, while the left limbs were flexed with normal muscle tone. Eye movements appeared to be small, and the head had a pronounced tremor. At 18 h post-KA, CIO8 walked with a wide stance and higher lift of the paws on the right side. Head tremor was still present, although the tail was held straight and eye movements appeared to be normal.

At 2 h post-KA injection, cat CIO10 walked with curvature to the right. Both forelimbs showed increased tone, eye movements appeared normal and no vestibular effects were noted.

Post day 1 behavioural effects

Of the two cats with IO lesions, only CIO8 worked in the reaching task in the week following the KA injection. On post day 1 (PD1), reach trajectories using the limb contralateral to the injection were normal, but grasping the handle required multiple attempts in comparison with pre-injection performance.

The strong activation of musculature immediately following the KA injection was entirely absent by testing on PD1. Although the cats objected to passive movement during carrying, they did not show strong vestibular disturbance: walking paths were straight, and despite some tendency to tip to the side, neither cat fell nor used the room wall to aid in body support. Both cats showed normal appetites. The curvature of the head and trunk seen shortly after injection was absent, and the body was held straight during locomotion. In contrast to high paw lifting seen during locomotion immediately after KA injection, both cats showed a tendency to drag their paws at the initial phase of the swing cycle on PD1. Cat CIO8 dragged the forelimb and hindlimb on the left side but not the right. Cat CIO10 dragged all four paws.

PD2 to ∼PD12

Throughout the first week, cat CIO8 showed normal reach trajectories and foot placement with the right limb but was unable to grasp the handle efficiently (Figs 4A and 5A). Neither grasping, reaching nor placement of the left limb showed any abnormality. Cat CIO10 only began reaching consistently on PD12; at this point, cat CIO10 had difficulty with grasping the handle when using either the left or right limb (Fig. 4B). However, the trajectories of the reaches were relatively normal although reaches with either limb showed a tendency to be somewhat higher than pre-injection reaches (Figs 4B and 5B). The trajectories of the limb during return to placement onto the pedestal, while variable, did not appear to be different from those seen pre-injection (Fig. 4).

Figure 4.

Over-plots of wrist trajectories during reaching out (Reach) and return (Return) of the paw to the stance platform comparing pre-injection and post-KA injection movements
Over time, the reach trajectories of the affected limbs developed a high arc and became highly variable. Returns to the stance platform also became highly variable. Grasping was compromised for the affected limbs on all post-lesion days, although limb trajectories were normal for ∼10 days post-injection. The left forelimb of CIO8 showed no changes for reaching and return.

Figure 5.

Percent increases in reach path length compared with straight reaches
Before the KA injections, the reach paths for both left and right forelimbs (continuous and dashed lines) were close to a straight path drawn from stance platform to device handle. Each entry shows the average of 5 reaches and the standard error. Day of injection is not included, and the horizontal lines indicate the pre-injection averages for the right (dashed lines) and left forelimb (continuous lines). Notice the progressive increase in reach curvature for the affected limbs over the entire testing period.

At the end of the first week, neither cat showed signs of vestibular disturbance and tolerated passive body movement well. Locomotion was efficient, and neither cat showed the foot drag seen on PD1 (foot drag had disappeared by PD2). However, measurements of foot lift during locomotion showed that both cats were lifting the limbs contralateral to the injection side somewhat higher than pre-injection (Fig. 6). For cat CIO8, the limbs ipsilateral to the injection showed no increase in lift. For cat CIO10, the forelimb ipsilateral to the injection as well as the limbs contralateral to the injection showed exaggerated lifting during locomotion. The hindlimb ipsilateral to the injection showed no exaggerated lifting (Fig. 6B).

Figure 6.

Average step height measurements during locomotion over the entire testing period
Day of injection is not included, and the horizontal lines mark the average forelimb (dashed lines) and hindlimb (continuous lines) step height pre-KA injection. Notice the progressive increase in height for the affected limbs over the entire testing period.

In summary, approximately 10 days post lesion, the cats no longer showed signs of vestibular problems and body postures were normal. Both cats showed a persistent deficit in grasping the handle during reaching, but only cat CIO10 showed some disturbance in limb trajectory during reaching. Locomotion appeared normal, but the cats lifted their feet higher than before lesion and, when stepping over obstacles, CIO10 showed an abnormal degree of lift for the left (contralateral) forelimb and hindlimb and the right forelimb.

Progressive changes from ∼PD13 to termination

From PD13 to the termination of testing, the cats showed progressively increasing deficits in limb movements. By the end of testing, day 72, reaches with the right limb for cat CIO8 showed strong curvature with a high reach (Figs 4A and 5A) and poor grasp control. The return paths for the right forelimb also became highly erratic (Fig. 4A). In contrast, the left limb showed no changes in reach trajectory, grasp or placement.

Cat CIO10, which was tested for a longer time than CIO8, showed progressive deficits for both the left and right forelimbs (Figs 4B and 5B) throughout the entire survival period. Return paths became very erratic, as if the cat had no sense of limb position, and successful placement of the paw only occurred when the cat touched the stance platform, which seemed to provide a position reference (Fig 4B). The right forelimb developed a similar progressive ataxia in reach and placement, although not as strong as that seen for the left, which probably reflects the significantly smaller size of the lesion in the left IO (Fig. 1).

An analysis of joint angles during reaching showed a fragmentation of reach trajectories into upward and outward components (Fig. 7). The reach trajectories of cat CIO10 became so distorted that the cat would strike himself on the muzzle and then drop the limb and project it outward to grasp the handle.

Figure 7.

Plots of shoulder and elbow joint angles during reaching for CIO10 pre-injection (A) and PD116 post-injection (B)
Normally, elbow and shoulder joints moved together to produce a straight trajectory; after the lesion, the shoulder joint angle did not change until maximum elbow flexion was reached. The dashed white lines on the images show the approximate wrist paths for the reaches.

Plots of average maximum step height during locomotion demonstrate a progressive increase over time and provide additional information about changes in the hindlimb as well as the forelimb. Figure 6 illustrates average maximum paw height throughout the entire testing period during locomotion with no obstacles. Cat CIO8 showed a progressive increase in step height for the right forelimb and hindlimb over the entire 72 day testing period (Fig. 6A). Neither the left forelimb nor left hindlimb showed any increase in step height over the testing period. The increase in step height was more dramatic for cat CIO10. CIO10 showed a rapid increase in paw height over the first 20 days post lesion with a more gradual increase throughout the remainder of the 133 day testing period (Fig. 6B). The absolute height reached by PD133 was over twice that seen for cat CIO8. The left forelimb and hindlimb showed the same increase in step height. The right forelimb showed somewhat less absolute size of increase, whereas the right hindlimb showed no increase in step height (Fig. 6B). Comparing the corresponding graphs for forelimb step height (Fig. 6) and forelimb reach path length (Fig. 5) reveals a strong similarity between plots, which suggests a common underlying cause affecting reaching and stepping during locomotion.

During the course of locomotion testing, cat CIO8 showed exaggerated step height when stepping over objects, and an obstacle test was added to the locomotion testing protocol. In the second half of testing of CIO8, the right limbs showed exaggerated flexion when stepping over a 25 mm barrier. The degree of flexion grew gradually throughout the remaining test period for the right limbs (left IO lesion) while showing a gradual decline for the left limbs (Fig. 8A). CIO10 showed a rapid increase in step height over the first 10–20 days, followed by a gradual increase over the remaining testing time except for the right hindlimb. The right hindlimb showed a modest initial increase in step height followed by a gradual decrease to or slightly below pre-KA levels (Fig. 8B). The right hindlimb received input from regions of the IO with minimal damage (Fig. 2B). The rapid rise and flattening of limb height during obstacle avoidance may reflect mechanical constraints on limb movement. Figure 9 illustrates the extreme limb flexion seen during stepping over the 25 mm barrier for cat CIO10.

Figure 8.

Step height when avoiding a barrier
Step height when avoiding a barrier during locomotion for the second half of testing for CIO8 (A) and for the entire testing period of CIO10 (B).

Figure 9.

Stepping over a barrier
Cat CIO10 stepping over a 25 mm barrier with the forelimb (A) and hindlimb (B) pre-KA injection and post-KA (PD116) injection (C and D).


At the outset of this study, we anticipated finding a spectrum of movement deficits that would be most severe shortly after IO lesion and would either remain constant over time or show some degree of recovery. Indeed, deficits in grasping were present on the first day following KA injection and showed no recovery over time, and signs of vestibular disturbance were strongest on the first day of testing after KA injection but rapidly decreased over the first week. A delayed long-term progressive deficit in limb movements was not anticipated and has not been reported previously.

Although the possibility that other brainstem regions contribute to the observed deficits cannot be ruled out, several observations indicate that they are due to IO damage. First, cell loss was extensive in the IO but could not be seen in other regions of the brainstem. Reticular cells adjacent to the IO appeared normal, although they must have been excited by the KA injections – the IO appears to be exquisitely sensitive to excitotoxic damage. Second, the affected limbs correspond to the somatotopic relations of the damaged IO regions. Finally, the long-term deficits are similar to deficits seen after damage to cerebellar cortex.

Previous IO lesion studies

Only one other study is directly comparable to the present report: Pompeiano et al. (1981) made large unilateral lesions using KA injections into the cat IO; the lesions included caudal IO. The cats were unable to walk for 4–15 days after injection but showed minimal motor deficits 27–60 days later. In contrast, our cats walked the day following KA injection and had developed severe limb ataxia by 27–60 days. The ability to walk and rapid recovery of vestibular deficits in our cats may be due to intact regions of caudal IO that were destroyed in the Pompeiano et al. (1981) study.

It is not clear why Pompeiano et al. (1981) did not report the progressive development of limb ataxia, but cats with lesions that included rostral IO may not have been observed for sufficiently long survival times. Their experimental design included decerebration and cerebellar lesions, and these additional procedures may have confounded effects and truncated observation times. Cats with lesions restricted to rostral IO were only followed for 7 days; at this time point our cats had relatively normal limb movements, and only grasping was severely affected.

Several studies have observed movements after making surgical lesions of the IO. The studies are problematical, since surgical lesions interrupt a variety of pathways in addition to olivocerebellar pathways (many decussating IO fibres cross through the opposite IO, so surgical lesions are not purely unilateral). All of the studies noted deficits in limb movement, but the details vary among the studies as well as in comparison with our results. Similar to our findings, Wilson & Magoun (1945) noted that 1 month after IO lesion, cats showed decomposition of movement and hyperflexion of the contralateral limbs during locomotion. Murphy & O’Leary (1971) stressed a remarkable degree of recovery, but the cats showed pronounced extension of the hindlimb during locomotion 65 days post lesion. Voneida et al. (1990) reported that cats with lesions of the caudal IO showed a high step during locomotion, but the cats appeared normal 1 month later.

Other studies of behaviour following IO lesion have used systemic injections of 3-acetylpyridine (3AP), which is toxic to the rat IO (Desclin & Escubi, 1974) as well as to some other structures (Balaban, 1985). Rats with 3AP destruction of the IO display some of the effects that we observed with KA injections. For the first few days after 3AP injection, the rats show hypotonia, which may correspond to the period of foot drag that we observed the day after KA injection. The hypotonic period occurs during the period that Purkinje cells show a high simple spike discharge rate. After a few days, simple spike rates begin to decline and gradually return to pre-injection rates (Batini & Billard, 1985). However, the pattern of discharge is abnormal, and the rats display hypertonia, tremor and other abnormalities (Llinas et al. 1975; Batini et al. 1985; Lopiano & Savio, 1986). One deficit, ‘mud walking,’ involves hyperflexion of the limbs during locomotion. The mud walking may be a bilateral form of the limb ataxia that we observed, which also included hyperflexion during limb movement (although mud walking in the rat developed in only 4–7 days after 3-AP treatment (Sukin et al. 1987)).

Phases post-KA injection

The immediate effects of KA injection are characterized by over-activation of musculature, with the flexor muscles of the limb being the first affected. The effects are most likely due to the increased IO discharge caused by glutamate receptor activation. Olivary discharge provides some direct excitation of the nuclei via collateral input, but high nuclear cell activity is largely due to inhibitory release from reduced simple spike discharge. Climbing fibre activity reduces simple spike discharge, and even a modest increase in complex spikes can completely inhibit simple spikes (Colin et al. 1980; Rawson & Tilokskulchai, 1981). Several hours after KA injection, some musculature, such as that of the cat's head and trunk, normalized, which suggests that these olivary regions were no longer active. In the rat, IO neurons become silent in as little as 2–3 h after a KA injection (Batini et al. 1987).

One day after KA injection, signs of high muscle activation were absent. The cats now dragged their paws when walking rather than high stepping, suggesting that the limb regions of the IO were now silent. The initial response of the Purkinje cell to a loss of climbing fibre input is a high steady rate of simple spike discharge (Colin et al. 1980; Leonard & Simpson, 1985; Cerminara & Rawson, 2004). High simple spike discharge rates inhibit nuclear cells (Benedetti et al. 1983), and the resulting behavioural effects on reaching and locomotion are similar to those seen after temporary inactivation of the rDAO (Horn et al. 2010). The foot dragging in our cats and hypotonia in 3AP rats is relatively short in duration (1 day for our cats, 2–4 days for 3AP rats, Sukin et al. (1987)), which is probably due to the fact that the high rate of simple spike discharge after losing climbing fibre input changes rapidly and begins to decrease after 1 day (Batini & Billard, 1985).

The third phase of behavioural changes consists of a progressive increase in limb ataxia that begins sometime in the first week and continues to the end of testing. The ataxia is characterized by hyperflexion of the limb during reaching and walking. The high stepping and curvature in reaching resemble that seen immediately after KA injection, which indicates that there may be a progressive loss of inhibitory control over nuclear cells following IO destruction.

Comparison with cerebellar lesions

If the progressive ataxia is due to a loss of inhibitory control over cerebellar nuclei, then lesions of cerebellar cortex should produce a similar limb ataxia, which is the case. Chambers & Sprague (1955b) studied the effects of lesions of paravermal cortex in the cat. During locomotion the lesions produced a marked hyperflexion of the limb joints ipsilateral to the lesion with instability at the wrist and ankle. The stepping sequence between forelimb and hindlimb was not affected. At rest, the forelimb was frequently held off the ground in a flexed posture, which we observed after the KA injection in case CIO10. Lesions of paravermal cortex of the anterior lobe produced long-lasting effects: one cat was followed for 58 days and showed no recovery. Another cat, with a less complete lesion, did show recovery after 89 days. Lesions of the paravermal cortex of the posterior lobe produced similar motor effects, but these recovered in about 1 week.

It is interesting that the anterior lobe lesions were more enduring than those of the posterior lobe, and this might account for the reports of rapid recovery following cerebellar lesions (see Introduction). Of course, our IO lesions removed climbing fibre input to both anterior and posterior lobes (Groenewegen et al. 1979). Climbing fibres bifurcate to provide input to both lobes, but interfolial branching is more extensive on the anterior lobe (Rosina & Provini, 1987). The additional anterior lobe branching might sustain IO cells following a posterior lobe lesion thereby allowing behavioural recovery.

Possible mechanisms of progressive symptoms

Perhaps the simplest explanation for progressive deficits would be a gradual and progressive degeneration of Purkinje cells, which would progressively release cerebellar nuclear cells from inhibition. However, Nissl- and DAPI-stained sections as well as fluoro jade staining showed no signs of degenerative changes in cerebellar cortex or nuclei, and Purkinje cells continue to discharge simple spikes long after destruction of the IO (Batini et al. 1985). Despite the lack of visible degeneration with light microscopy, a loss of climbing fibres alters both Purkinje cells and deep nuclear neurons (Desclin & Colin, 1980; Anderson & Flumerfelt, 1986). After treatment with 3AP, nuclear cells appear to lose sensitivity to GABAA (Billard & Batini, 1991; Vigot et al. 1993), which might effectively remove Purkinje cell inhibition on nuclear cells. A post-synaptic change in the cerebellar nuclei is consistent with the finding that glutamic acid decarboxylase levels show a long-term increase after 3AP treatment (Sukin et al. 1987), which suggests that Purkinje cell terminals continue to release GABA after climbing fibre loss. A loss of nuclear cell response to GABA may be similar to that seen after spinal damage, which may contribute to the development of spasticity (Lu et al. 2008; Boulenguez et al. 2010).

Locus of action

Our lesions included all subdivisions of the rostral IO, and the observed deficits may be due to action of the combined lesion or may consist of subcomponents associated with specific IO regions. Inactivation of different subdivisions of the IO produces very different behavioural effects (Horn et al. 2010) and it is likely that the overall behavioural deficit represents a combination of different components.

The lesion of the left IO in CIO10 produced little damage of the left PO, yet use of the corresponding right forelimb showed grasping deficits and progressive ataxia in reaching and walking, which suggests that a loss of climbing fibre input to dentate and lateral cortex is not necessary to produce these deficits. This is consistent with the findings of Chambers & Sprague (1955a,b) and Martin et al. (2000), who failed to observe disruption of normal limb movements with lesions or inactivation of the dentate nucleus.

Our lesions produced some damage in caudal IO (especially CIO10); however, inactivation of caudal IO affected balance and walking but not reaching and grasping (Horn et al. 2010). Similarly, lesions of vermal cortex or the fastigial nucleus (Chambers & Sprague, 1955a,b) or inactivation of the fastigial nucleus (Martin et al. 2000) affect balance but do not disturb limb movements. Therefore it is unlikely that the damage to caudal IO contributes to the progressive ataxia, although it may well be involved in the early vestibular disturbance, which rapidly recovered.

The KA lesions produced severe damage to rDAO and rMAO, which project to paravermal cortex and interpositus. Lesions or inactivation of paravermal cortex or interpositus have been reported to affect locomotion, reaching and grasping (Chambers & Sprague, 1955b; Milak et al. 1997; Cooper et al. 2000; Martin et al. 2000) and it is likely that the progressive ataxia and grasping deficits are due to involvement of rDAO and rMAO.

Inactivation of rDAO or rMAO suggest that these regions of the IO may contribute to different aspects of the movement deficits. Inactivation of the cat rDAO produces a strong deficit on grasping with a relatively minor effect on locomotion (Horn et al. 2010). The effect on grasping is probably mediated via interpositus projections to the magnocellular red nucleus (RNm), which has selective projections to motoneurons innervating digit musculature (McCurdy et al. 1987; Robinson et al. 1987). Immediately after KA injections, our cats had difficulty in grasping the handle and this deficit persisted throughout the entire testing period. The immediate and severe grasping effect may be due to the direct RNm projections to digit motoneurons. Although the deficits in grasping appeared constant after KA injection, its underlying cause may not have been the same throughout the testing period. For example, the immediate effects were probably due to over-activation of interpositus, whereas the effects at PD1 were probably due to inhibition of interpositus; either could produce deficits in using the toes to grasp. Stimulation of interpositus affects both distal and proximal forelimb musculature (Ekerot et al. 1995), and upper limb deficits may have contributed also to grasping deficits.

Inactivation of rMAO has a severe effect on locomotion with less effect on grasping than inactivation of rDAO (Horn et al. 2010). Therefore, it is possible that the progressive ataxia during reaching and locomotion may largely be due to damage of rMAO. In monkeys, inactivation of anterior interpositus (associated with rDAO) produces an inability to properly form the digits for grasping, whereas inactivation of posterior interpositus (associated with rMAO) produces deficits in proximal limb control during reaching and grasping (Mason et al. 1998). (However, Martin et al. (2000) reported opposing effects of anterior and posterior interpositus inactivation on reaching in the cat with no effect on grasping.) Understanding differential actions of IO subdivisions will require additional experiments with selective lesions.

It is possible that alterations in cerebellar output induce changes in additional regions of the brain that receive cerebellar output directly or indirectly. For example, small lesions of a cerebellar module in the rat can alter spinal reflexes without affecting locomotion (Pijpers et al. 2008). Large lesions of the cerebellar nuclei (Wolpaw & Chen, 2006) or IO (Chen et al. 2011) contribute to a large and delayed (40 day) increase in the spinal H-reflex (however, the increase was only seen in rats that had undergone down-conditioning and not in control subjects). Changes in spinal reflexes might produce high stepping and alterations in reaching, but it is unlikely that they would account for the hyperflexion that we saw immediately after KA injection.

Potential clinical relevance

Many neurological diseases involving cerebellar circuits are progressive and often involve degenerative changes in the IO (Koeppen et al. 1999). Interestingly, the spinal cerebellar ataxias produce changes in reaching movements that are similar to the changes that we observed in the cat reaching. When patients with cerebellar ataxias point to a target, the normal coordination across shoulder and elbow joints decomposes so that the joints tend to move sequentially rather than together. The trajectories of the reaches have high arcing initial components followed by outward components, and the trajectories are variable from trial to trial (Day et al. 1998). The progressive nature may be due to a progressive loss of GABA-ergic synapses in the cerebellar nuclei (Koeppen et al. 2011), which is consistent with the reported loss of GABA sensitivity after IO destruction (Billard & Batini, 1991).

Remaining questions

The behavioural effects of IO destruction leave many questions unanswered: Why is grasping immediately impaired whereas ataxic limb movements require days to develop? Why is progressive ataxia limited to limb movements? Trunk posture appears normal, despite being severely affected in the initial period after KA injection. What pathways are responsible for the ataxic limb movements? Why does limb ataxia favour hyperflexion? Experiments designed to answer these questions will provide deeper insight into the function of the inferior olive as well as the cerebellum.


Author contributions

Experiments were performed by K.M.H., A.R.G. and A.D. at the Division of Neurobiology, Barrow Neurological Institute, St Joseph's Hospital and Medical Center. Study conception, design, data collection, analysis, interpretation and manuscript writing: K.M.H. and A.R.G. Data collection, data analysis, manuscript editing and revision: A.D. All authors approved the final version of the manuscript. The authors have no conflicts of interest to declare.


This work was supported by the Barrow Neurological Foundation.