A noradrenergic mechanism of motor memory consolidation in the cerebellar cortex

Daniel Kellett, Christopher Yeo. UCL, London, United Kingdom.

Classical conditioning of the rabbit eyeblink/nictitating membrane (NM) response is critically dependent upon the cerebellum (see Yeo & Hesslow, 1998). Normal activity in an eyeblink control region of cerebellar cortical lobule HVI is essential for acquisition, consolidation, and performance of NM conditioning. Consolidation is prevented by post-training inactivation of HVI with the GABA A receptor agonist muscimol (Attwell et al. 2002) but the mechanisms are unknown. An influential theory of cerebellar learning suggests a noradrenergic signalling mechanism is important for consolidation (Gilbert, 1975), similar to mechanisms known to be important for memory formation in other brain regions. Here, the noradrenergic theory of cerebellar consolidation was tested using post-training infusions of the β1 selective adrenoceptor antagonist atenolol in cortical lobule HVI.

Male New Zealand white rabbits (2.4-2.8 kg) were implanted with a guide cannula in cerebellar cortical lobule HVI under isoflurane anaesthesia. Each subject received 6 daily sessions of NM conditioning (tone conditional stimulus paired with periocular electrical unconditional stimulus, 50 trials/day; see Attwell et al., 2002). Immediately after Sessions 1 and 2, each subject received 2 µl infusions into cerebellar cortex. Control subjects received saline infusions (SAL), experimental subjects received either muscimol (MUS; 14 nmol), atenolol (ATEN; 20 nmol), or atenolol delayed by 2h (ATEN-2H; 20 nmol). In Sessions 3-6 all subjects were trained to asymptote in the absence of infusions. At the end of the protocol, the non-NMDA receptor antagonist CNQX (12 nmol) was infused into cortex. Subjects with conditioned responses (CRs) abolished by CNQX were judged to have had effecive cannula placements and, if histology indicated no incidental cortical damage, they were admitted to the study.

Learning was significantly impaired in both the MUS and ATEN groups, compared to SAL (Kruskal-Wallis ANOVA on ranks followed by Dunn’s test). On Session 3, CR frequency was 38±14% in MUS (P<0.05, n=4), and 36±11% in ATEN (P<0.05, n=8), compared to 94±4% in SAL (n=7). However, the ATEN-2H group was not significantly different from SAL (87±8%, n=5). Likewise, ATEN subjects with off-target infusions (as judged by the CNQX test) were not significantly different from SAL (92±3%, n=6). Lastly, ATEN had no effect on established CRs over 3 additional sessions, when infused post-training on Sessions 7 and 8 (Friedman ANOVA on ranks, P=0.4).

The data indicate that noradrenaline, acting at cerebellar cortical β1 adrenoceptors, provides a major signal for the consolidation of motor memory, and this process occurs within 2h of the end of training.


Attwell PJE, Cooke SF & Yeo CH (2002). Neuron 34: 1011-1020
Gilbert P (1975). Nature 254: 688-9
Yeo CH & Hesslow G (1998). Trends Cogn. Sci. 2: 322–330


Supported by: BBSRC ( UK)