Memory consolidation in the cerebellar cortex

Abstract

Several forms of learning, including classical conditioning of the eyeblink, depend upon the cerebellum. In examining mechanisms of eyeblink conditioning in rabbits, reversible inactivations of the control circuitry have begun to dissociate aspects of cerebellar cortical and nuclear function in memory consolidation. It was previously shown that post-training cerebellar cortical, but not nuclear, inactivations with the GABAA agonist muscimol prevented consolidation but these findings left open the question as to how final memory storage was partitioned across cortical and nuclear levels. Memory consolidation might be essentially cortical and directly disturbed by actions of the muscimol, or it might be nuclear, and sensitive to the raised excitability of the nuclear neurons following the loss of cortical inhibition. To resolve this question, we simultaneously inactivated cerebellar cortical lobule HVI and the anterior interpositus nucleus of rabbits during the post-training period, so protecting the nuclei from disinhibitory effects of cortical inactivation. Consolidation was impaired by these simultaneous inactivations. Because direct application of muscimol to the nuclei alone has no impact upon consolidation, we can conclude that post-training, consolidation processes and memory storage for eyeblink conditioning have critical cerebellar cortical components. The findings are consistent with a recent model that suggests the distribution of learning-related plasticity across cortical and nuclear levels is task-dependent. There can be transfer to nuclear or brainstem levels for control of high-frequency responses but learning with lower frequency response components, such as in eyeblink conditioning, remains mainly dependent upon cortical memory storage.

Figure 1. Effects of local inactivations on olivocorticonuclear neuronal activity and memory consolidation.

Simplified views of olivocorticonuclear circuitry involved in motor memory formation, with cortical interneurons, multiple mossy fiber inputs, and some brainstem circuits omitted for clarity. Each panel shows how information transmission and excitabilities within the system may change after different interventions. Excitability increases (↑) and decreases (↓) are indicated. Asterisks denote synapses at which muscimol may be acting. Post-training muscimol infusions to cerebellar cortex (A) prevent consolidation, but it is uncertain whether consolidation processes are disrupted directly in the targeted structure or indirectly through disturbance of the OCN loop. In particular, the excitability of cerebellar nuclear neurons will increase as a consequence of cortical inactivation. However, post-training muscimol infusions to the cerebellar nuclei (B) do not affect consolidation. Nucleo-olivary inhibition is depressed, so olivary excitability will be increased. At Purkinje cells, increased climbing fiber activity increases complex spike activity with a corollary reduction in simple spike activity, indicated by ↑↓, but this does not impair consolidation processes. It remains possible that consolidation occurs entirely in the cerebellar nuclei only if these processes are disrupted by excitability increases, but insensitive deep neuronal inhibition with muscimol. Hence in the present investigation (C), post-training muscimol infusions to both cortex and nuclei cause a deep inhibition of cortical neurons, whilst protecting the cerebellar nuclei from disinhibition. (D) A key to panels A–C and a model of cerebellar pathways engaged in NMR conditioning. CS- and US-related information converges within the cerebellar cortex and within the cerebellar nuclei through mossy fiber and climbing fiber inputs, respectively. The inhibitory olivo-cortico-nuclear loop (OCN) is indicated by dashed arrows. Conventions: excitatory neurons and synapses are shown in white; inhibitory neurons and synapses in black. Abbreviations: Ba, basket cell; cf, climbing fiber; Go, Golgi cell; Grc, granule cell; mf, mossy fibers; NV, trigeminal nucleus; Pc, Purkinje cell; pf, parallel fibers; RN, red nucleus; St, Stellate cell.

Figure 2. Experimental design and effects of post-training muscimol infusions to cerebellar cortex and nuclei on consolidation.

Experimental design: each daily session is shown as an open rectangle. Solid vertical lines indicate 3 day rest periods. Post-training cortical and nuclear infusions of muscimol (closed arrows) or vehicle (open arrows) are indicated. Behavioral data: daily, mean session %CRs (±1 SEM) for the Control (cortical and nuclear vehicle, n = 4) and Cortex+Nucleus (cortical and nuclear muscimol, n = 4) groups. Control subjects acquired asymptotic CRs during Phase 1, but Cortex+Nucleus subjects did not. Cortex+Nucleus subjects developed robust CRs during Phase 2, when muscimol was not given. Post-training infusions of muscimol given to the Control subjects during Phase 3 had no consequences for the maintained expression of CRs during Phases 3 and 4.

Figure 3. Incomplete and off-target infusions confirm a critical role of cerebellar cortex in consolidation.

Expanded view of CR acquisition in Phase 1 (mean CR frequencies for successive 10-trial blocks of Phase 1 in all experimental groups, followed by mean session frequencies for each session of Phase 2 and 3). Rate of acquisition is related to depth of cortical inactivation with muscimol: Cortex+Nucleus (n = 4) subjects fail to acquire robust CRs during Phase 1, whilst Incomplete (n = 6) subjects do acquire CRs, but not as rapidly as Nucleus Only (n = 7) and Off-Target (n = 10) subjects, which acquire CRs at a similar rate to Controls (n = 4).

Figure 4. Effects of cortical and nuclear muscimol and CNQX on performance of established CRs.

Mean (±1 SEM) effects of muscimol (7 nmol; left panels) and CNQX (6 nmol; right panels) on CR performance when infused into cortical (2 µl; top panels) or nuclear cannulae (1 µl; bottom panels). Only the Cortex+Nucleus subjects satisfy all the criteria for inclusion in this principal group.

Figure 5. Histological reconstruction of cortical and nuclear infusion sites.

Cannula tip locations are shown for all subjects in each group on a series of 8 standard transverse sections at levels from −1.0 mm to −4.5 mm relative to skull lambda. For each subject in the five groups, two matching symbols indicates the location of the cortical and nuclear cannula tips. For example, one subject in the Cortex+Nucleus group has cannula positions indicated by filled circles – the cortical placement is seen at level −1.0 mm and the nuclear placement at level −3.0 mm. The scale bar indicates 5 mm. The separation of the two cannula tip locations may be judged by reference to distances in the rostro-caudal axis and transverse planes. Abbreviations: crI and crII - crus 1 and 2 (of ansiform lobe); DPFL - dorsal paraflocculus; FL - flocculus; HIV-V, HVI - hemispheral lobules 4–5 and 6 (of Larsell); ND - dentate nucleus; NF - fastigial nucleus; NI - interpositus nucleus; PM - paramedian lobe; VPFL - ventral paraflocculus; II-X vermis lobules 2–10 (of Larsell).

Figure 6. Histological identification of cannula tip and spread of infusion.

Photomicrographs of 50 µm transverse cerebellar sections from a experimental subject from the Cortex+Nucleus group. Sections are stained with cresyl violet, showing location of cortical (HVI) and nuclear (AIP) cannula tips (arrowhead). Positions relative to skull λ are indicated (mm). The approximate extent of diffusion of pontamine sky blue dye (2 µl for cortex, 1 µl for nuclei) was plotted on adjacent neutral red-stained sections, and superimposed on the current sections (dotted lines). Calibration bar represents 2 mm.

Figure 7. Conditioned response profiles.

Mean (±1 SEM) CR latency to peak, and magnitude (amplitude above threshold of 0.5 mm), in Control and Cortex+Nucleus subjects.