Motor learning in the VOR: the cerebellar component.

This paper reviews results that support a model in which memory for VOR gain is initially encoded in the flocculus, and in which cerebellar LTD and LTP are responsible for gain increases and gain decreases, respectively. We also review data suggesting that after it is encoded, motor memory can either be disrupted, possibly by a local mechanism, or else consolidated. We show that consolidation can be rapid, in which case the frequency dependence of learning is unchanged and we will argue that this is consistent with a local mechanism of consolidation. In the longer term, however, the available evidence supports the transfer of memory out of the flocculus. In new experiments reported here, we address the mechanism of memory encoding. Pharmacological evidence shows that both mGluR1 and GABA(B) receptors in the flocculus are necessary for gain-up, but not for gain-down learning. Immunohistochemical experiments show that the two receptors are largely segregated on different dendritic spines on Purkinje cells. Together with what is already known of the mechanisms of cerebellar LTD and LTP, our data suggest that the direction of learning may be determined by interactions among groups of spines. Our results also provide new evidence for the existence of frequency channels for vestibular signals within the cerebellar cortex.

The bidirectionality of motor learning in the vestibulo-ocular reflex is a function of cerebellar mGluR1 receptors.

Bidirectional changes in synaptic transmission have the potential to optimize the control of movement. However, it can be difficult to establish a causal relationship between the bidirectionality of synaptic plasticity and bidirectional changes in the speed of actual movements. We asked whether metabotropic glutamate receptor 1 (mGluR1) receptors, which participate in cerebellar long-term depression (LTD), are necessary for bidirectional motor learning in the vestibulo-ocular reflex (VOR). Cerebellar LTD and long-term potentiation (LTP) are thought to cause increases and decreases, respectively, in the gain of the VOR; the direction of learning depends on the behavioral protocol. We injected either the mGluR1 agonist (S)-DHPG or the antagonist YM 298198 bilaterally into the flocculus of alert cats, and then induced motor learning. In the presence of YM 298198, the VOR gain decreased in gain-up, as well as in gain-down protocols. (S)-DHPG augmented gain-up learning. Gain-down learning was not significantly affected by either drug. These results supported the hypothesis that gain-up learning relies on cerebellar LTD, but gain-down learning relies on a different mechanism. In the absence of mGluR1 activity, cerebellar LTD may be replaced with LTP, permitting learning in only one direction.

Expression of calcium-binding proteins and nNOS in the human vestibular and precerebellar brainstem.

Information about the position and movement of the head in space is coded by vestibular receptors and relayed to four nuclei that comprise the vestibular nuclear complex (VNC). Many additional brainstem nuclei are involved in the processing of vestibular information, receiving signals either directly from the eighth nerve or indirectly via projections from the VNC. In cats, squirrel monkeys, and macaque monkeys, we found neurochemically defined subdivisions within the medial vestibular nucleus (MVe) and within the functionally related nucleus prepositus hypoglossi (PrH). In humans, different studies disagree about the borders, sizes, and possible subdivisions of the vestibular brainstem. In an attempt to clarify this organization, we have begun an analysis of the neurochemical characteristics of the human using brains from the Witelson Normal Brain Collection and standard techniques for antigen retrieval and immunohistochemistry. Using antibodies to calbindin, calretinin, parvalbumin, and nitric oxide synthase, we find neurochemically defined subdivisions within the MVe similar to the subdivisions described in cats and monkeys. The neurochemical organization of PrH is different. We also find unique neurochemical profiles for several structures that suggest reclassification of nuclei. These data suggest both quantitative and qualitative differences among cats, monkeys, and humans in the organization of the vestibular brainstem. These results have important implications for the analysis of changes in that organization subsequent to aging, disease, or loss of input.

Consolidation and disruption of motor memory generalize across stimulus conditions in the vestibulo-ocular reflex.

The vestibulo-ocular reflex (VOR) exhibits motor learning that initially depends on synaptic plasticity in the cerebellar cortex. Learned decreases in VOR gain can be disrupted by rotation in darkness immediately following learning, but consolidate rapidly if the disruption stimulus is delayed. Disruption may simply reverse the synaptic changes that have recently occurred, or it may reflect new learning at other sites. The alternative to disruption, rapid consolidation, also may take place by altering the existing memory trace or may require changes at other locations. To test these possibilities, we induced decreases in the gain of the VOR in cats that wore miniaturizing goggles. Using a range of frequencies of rotation, we investigated the patterns of generalization for disruption and for rapid consolidation of the learned changes in gain. Learning was most effective at the particular frequencies that were used during training. However, disruption and rapid consolidation were not more effective at the rotation frequencies that were used during training. Instead, after consolidation, the memory retained the frequency tuning that had been established during the learning process. We conclude that disruption and rapid consolidation may not require new learning.

Dynamics of glutamatergic synapses in the medial vestibular nucleus of the mouse.

During sinusoidal rotation or translation, primary vestibular afferents modulate their discharge rates at the frequency of motion, effectively transmitting frequency-modulated (FM) signals. This study indicates a possible role for excitatory synapses in the processing of FM signals by vestibular brainstem pathways. Inputs to medial vestibular neurons were activated with FM pulse trains, while inhibitory transmission was blocked. The relationship between the presynaptic pulse rate and the postsynaptic membrane potential was found to be linear within a range of pulse rates. Short-term plasticity was a factor contributing to sensitivity at higher modulating frequencies. The amount of low-pass filtering was correlated with excitatory postsynaptic potential (EPSP) shape, which affected temporal summation during the train. Although the NMDA component of glutamatergic transmission affected EPSP shape, it made only a minor contribution to the dynamics of synaptic transmission. Most responses showed low-pass filtering over the entire 1-16 Hz range. Overall, excitatory synapses in the medial vestibular nucleus contribute a low-pass filter to central vestibular processing and complement the high-pass filtering that is introduced both by peripheral vestibular dynamics and by the intrinsic dynamics of secondary vestibular neurons.

Rapid consolidation of motor memory in the vestibuloocular reflex.

Motor memory is relatively labile immediately after learning but can become more stable through consolidation. We investigated consolidation of motor memory in the vestibuloocular reflex (VOR). Cats viewed the world through telescopic lenses during 60 min of passive rotation. Learned decreases (gain-down learning) and increases in the VOR gain (gain-up learning) were measured during sinusoidal rotation at 2 Hz. We found that if rotation in darkness immediately followed learning, the gain of the VOR reverted toward its prelearning value, indicating that expression of the memory was disrupted. If after gain-down learning the cat spent another 60 min stationary without form vision, subsequent disruption did not occur, suggesting that memory had consolidated. Consolidation was less robust for gain-up learning. We conclude that memory in the VOR is initially labile but consolidates rapidly and consistently after gain-down learning.

Adaptive rescaling of central sensorimotor signals is preserved after unilateral vestibular damage.

Adaptive rescaling is a widespread phenomenon that dynamically adjusts the input-output relationship of a sensory system in response to changes in the ambient stimulus conditions. Rescaling has been described in the central vestibular neurons of normal cats. After recovery from unilateral vestibular damage, the vestibulo-ocular reflex (VOR) remains nonlinear for rotation toward the damaged side. Therefore, rescaling in the VOR pathway may be especially important after damage. Here, we demonstrate that central vestibular neurons adjust their input-output relationships depending on the input velocity range, suggesting that adaptive rescaling is preserved after vestibular damage and can contribute to the performance of the VOR. We recorded from isolated vestibular neurons in alert cats that had recovered from unilateral vestibular damage. The peak velocity of 1-Hz sinusoidal rotation was varied from 10 to 120 degrees/s and the sensitivities and dynamic ranges of vestibular neurons were measured. Most neuronal responses showed significant nonlinearities even at the lowest peak velocity that we tested. Significant rescaling was seen in the responses of neurons both ipsilateral and contralateral to chronic unilateral damage. On the average, when the peak rotational velocity increased by a factor of 8, the average sensitivity to rotation decreased by roughly a factor of 2. Rescaling did not depend on eye movement signals. Our results suggest that the dynamic ranges of central neurons are extended by rescaling and that, after vestibular damage, adaptive rescaling may act to reduce nonlinearities in the response of the VOR to rotation at high speeds.

The site of a motor memory shifts with consolidation.

The basis for the consolidation of memory is a controversial topic, particularly in the case of motor memory. One view is that motor memory is transferred, partially or completely, to a new location during the consolidation process ("systems consolidation"). We investigated this possibility in a primitive motor system, the vestibulo-ocular reflex (VOR). In the simple circuitry of the VOR, there are relatively few possible storage sites for memory. We partially blocked excitatory neurotransmission in the cerebellar cortex of cats with the glutamate antagonist 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX). If CNQX was injected immediately after 60 min of rotation under conditions that induced a learned decrease in the gain of the VOR, gain was returned to its baseline value. Expression of the new memory could also be disrupted by rotation in darkness, suggesting that consolidation had not taken place; however, after learning had continued for 3 d, expression of the learned change was diminished only slightly by blockade and was unaffected by rotation in darkness. Our interpretation of these results is that learning may take place initially in the cerebellar cortex and that during consolidation, motor memories are converted to a more distributed representation that includes the cerebellar cortex and another site.

Learning in a simple motor system.

Motor learning is a very basic, essential form of learning that appears to share common mechanisms across different motor systems. We evaluate and compare a few conceptual models for learning in a relatively simple neural system, the vestibulo-ocular reflex (VOR) of vertebrates. We also compare the different animal models that have been used to study the VOR. In the VOR, a sensory signal from the semicircular canals is transformed into a motor signal that moves the eyes. The VOR can modify the transformation under the guidance of vision. The changes are persistent and share some characteristics with other types of associative learning. The cerebellar cortex is directly linked to the VOR reflex circuitry in a partnership that is present in all vertebrates, and which is necessary for motor learning. Early theories of Marr, Albus, and Ito, in which motor memories are stored solely in the cerebellar cortex, have not explained the bulk of the experimental data. Many studies appear to indicate a site of learning in the vestibular nuclei, and the most successful models have incorporated long-term memory storage in both the cerebellar cortex and the brainstem. Plausible cellular mechanisms for learning have been identified in both structures. We propose that short-term motor memory is initially stored in the cerebellar cortex, and that during consolidation of the motor memory the locus of storage shifts to include a brainstem site. We present experimental results that support our hypothesis.

Asymmetric responses to rotation at high frequencies in central vestibular neurons of the alert cat.

The horizontal rotatory vestibulo-ocular reflex (VOR) stabilizes gaze by moving the eyes at an angular velocity proportional to head velocity, and can accomplish this for a broad range of frequencies and amplitudes of head motion. Rotation at 5 Hz and above may be processed differently than lower frequencies by the VOR network. We recorded discharges and calculated spike densities of a small sample of vestibular neurons in alert cats during low-velocity rotation at frequencies up to 8 Hz. At high frequencies, we found both vestibular-only (V-only) and eye-movement-sensitive (EM) cells that generated asymmetric output signals. Asymmetry was primarily of the cutoff type, i.e., changes in spike density were smallest for rotation in the inhibitory direction. Most cells were identified as secondary neurons. The mean spike density was 23 sp/s, which was lower than previously reported in vestibular neurons of monkeys. A few neurons had very high sensitivities, associated with phase-locking, to rotation at high frequencies. In general, vestibular neurons carried a high-pass-filtered version of rotational signals. When synaptic inputs from the vestibular commissure were quantified, we found that the immediate change in probability of firing due to commissural vestibular input was inversely correlated with the degree of high-pass filtering. At high frequencies, increased asymmetry and phase-locking occurred in some neurons. A small number of neurons responded with increased probability of firing to both directions of rotation. Together, these observations suggest that high frequencies of rotation may be encoded differently than low frequencies by central vestibular neurons in alert animals.

Commissural inputs to secondary vestibular neurons in alert cats after canal plugs.

Gaze is stabilized during head movements primarily by the vestibuloocular reflex (VOR). After a unilateral canal plug, the VOR's response is reduced. Recovery of the VOR may be brought about by changes in the efficacy of brain stem synapses or by other mechanisms. We measured the responses of horizontal secondary vestibular neurons (HSNs) to stimulation of the contralateral labyrinth. HSN responses in normal alert cats were compared with those in cats that had recovered from unilateral horizontal semicircular canal (HSCC) plugs. After recovery, excitatory commissural inputs to HSNs on the plugged side elicited significantly smaller responses than in normal cats with no change in mean discharge rates. However, mean discharge rates tended to be higher after recovery for cells receiving inhibitory commissural inputs. The change in resting rate invalidates any direct comparison of inhibitory inputs. These results are interpreted in terms of possible mechanisms for recovery from unilateral vestibular loss by the VOR neural network. We conclude that after unilateral HSCC plugs, changes in brain stem excitatory synapses and/or excitability of secondary vestibular neurons may participate in the restoration of normal vestibular reflexes.

The response of vestibulo-ocular reflex pathways to electrical stimulation after canal plugging.

The vestibulo-ocular reflex (VOR) allows clear vision during head movements by generating compensatory eye movements. Its response to horizontal rotation is reduced after one horizontal semicircular canal is plugged, but recovers partially over time. The majority of VOR interneurons contribute to the shortest VOR pathway, the so-called three-neuron arc, which includes only two synapses in the brainstem. After a semicircular canal is plugged, transmission of signals by the three-neuron arc originating from the undamaged side may be altered during recovery. We measured the oculomotor response to single current pulses delivered to the vestibular labyrinth of alert cats between 9 h and 1 month after plugging the contralateral horizontal canal. The same response was also measured after motor learning induced by continuously-worn telescopes (optically induced motor learning). Optically induced learning did not change the peak velocity of the evoked eye movement (PEEV) significantly but, after a canal plug, the PEEV increased significantly, reaching a maximum during the first few post-plug days and then decreasing. VOR gain also showed transient changes during recovery. Because the PEEV occurred early in the eye movement evoked by a current pulse, we think the observed increase in PEEV represented changes in transmission by the three-neuron arc. Sham surgery did not result in significant changes in the response to electrical stimulation or in VOR gain. Our data suggest that different pathways and processes may underlie optically induced motor learning and recovery from plugging of the semicircular canals.