(a) Representative images (Golgi-Cox staining) of distal Purkinje cell dendrites in the distal molecular layer (DML), quantification of spine density (WT, n=97/4, dendrites per mice; Shank2^−/−, n=89/4, P=0.2, MWU-test) and cumulative frequency plots of spine length (P=0.4, MWU-test) and thickness (P=1, MWU-test) in WT (n=748/4 spines per mice) and Shank2^−/− mice (n=639/4) as indicated. Scale bar, 1 μm. (b,c) Representative images (electron microscopy) of spine synapses in the DML and cumulative frequency plots of PSD length (WT, n=226/4, PSDs per mice; Shank2^−/−, n=243/4, P=0.3) and thickness (WT, n=223/4; Shank2^−/−, n=233/4, P=0.9) as indicated. Scale bar, 100 nm. (d) Biochemical analysis of Shank2 (WT, n=12 synaptosomes; Shank2^−/−, n=6), GluA1 (WT, n=12 synaptosomes; Shank2^−/−, n=6, P=0.041) GluA2 (WT, n=11; Shank2^−/−, n=5, P=0.014) and Nlgn3 (WT, n=12; Shank2^−/−, n=6, P=0.4) in cerebellar synaptosomes from WT and Shank2^−/− mice as indicated. Data in bar graphs are presented as mean±s.e.m.; single asterisks indicates P<0.05. Two-sided t-tests were used, unless stated otherwise.

(a) Recording configuration for voltage-clamp recordings of PF–PC synaptic transmission. Inset: an example PF-EPSC. (b–e) With comparable holding current (at −65 mV) (P=1) and input resistance (Ri) (P=0.8), PC EPSC rise time (P=0.2) and EPSC decay time (P=0.3) are not different between WT (n=9/6, cells per animals) and Shank2^−/− (n=7/6). (f) Example EPSCs in response to 3, 6, 9, 12 and 15 μA stimulation. (g,h) Varying stimulation strength (P=0.9, repeated-measures ANOVA) and inter-stimulus interval (P=0.2, repeated-measures ANOVA) evoked comparable EPSC amplitude or facilitation (WT, n=11/3; Shank2^−/− 15/3). (i) Recording configuration for whole-cell recording. Inset: an example action potential. (j–m) Action potential threshold (P=0.8), amplitude (P=0.1), half-width (P=0.7) and after-hyperpolarization (P=0.2) were not different (WT, n=10/6; Shank2^−/−, n=11/6). (n) Example traces of intrinsic Purkinje cell excitability as apparent from action potential firing evoked by 300 pA current injections. (o) No difference in evoked firing frequency relative to various levels of current injections (WT, n=10/5; Shank2^−/−, n=11/5, P=0.1, repeated-measures ANOVA). Inset barplot shows average slope of firing rate per current step (P=1). Data are represented as mean±s.e.m. Two-sided t-tests were used, unless stated otherwise.

(a,b) Example of spontaneous firing inhibitory post synaptic currents (sIPSCs) in lobules I–V and X. (c,e) A higher frequency of sIPSCs is found in both anterior (I–V) (P=0.0295) and posterior (X) lobules (P=0.0079) in Shank2^−/− PCs (anterior: WT, n=25/3, cells per animals; Shank2^−/−, n=20/3; posterior: WT, n=19/3; Shank2^−/−, n=19/3). (d,f) There were no significant differences in sIPSC amplitudes anteriorly (P=0.1) or posteriorly (P=0.5). (g,h) Extracellular traces of PCs recorded in anterior (left) and posterior (right) lobules in the cerebellum, in WT and Shank2^−/−. Asterisks denote complex spikes. (i–k) Simple spike (SS) firing frequency was significantly lower (P=0.0096) in Shank2^−/− (n=26/3) compared with wild type (n=23/3), whereas the coefficient of variation (CV) (P=0.7) and CV2 (P=0.1) did not differ. (l–n) In posterior lobule X, while SS firing frequency was similar (P=0.6), CV (P=0.0086) and CV2 (P=0.0003) were significantly higher in Shank2^−/− (n=27/3) compared with WT (n=16/3). Data are represented as mean±s.e.m. Single and double asterisks indicate P<0.05 and P<0.01, respectively. Two-sided t-tests were used, unless stated otherwise.

(a) Recording configuration for PF-LTP experiments. Inset: example of five averaged EPSCs for WT (blue) and Shank2^−/− (red) before LTP induction (at 10 min) and after LTP induction (at 40 min). (b) LTP experiment with 5 min PF stimulation at 1 Hz inducing LTP in WTs (n=7/6, cells per animals, P=0.0027) but not in Shank2^−/− PCs (n=12/6, P=0.3), which is reflected in the difference between genotypes (P=0.0066). (c) Recording configuration for PF-LTD experiments. Inset: example traces as in a. (d) LTD is induced in both WT (n=9/6, P<0.0001) and Shank2^−/− (n=7/6, P=0.0009) PCs, to a similar degree (P=1). (e) Example of traces for intrinsic plasticity with current injections of 300 pA. (f) LTP induction protocol induced enhanced spike output in WT PCs (n=5/4, P=0.0053), but not in Shank2^−/− PCs (n=5/4, P=0.5), as reflected in their difference (P=0.0201). Data are represented as mean±s.e.m. Single and double asterisks indicate P<0.05 and P<0.01, respectively. All tests were repeated-measures ANOVAs. NS, not significant.

(a) Sagittal cryosection of hippocampal SHANK2 (red) and nucleus staining (DAPI; blue) in a WT (left) L7-Shank2^−/−. Scale bar, 500 μm. Sagittal cryosection of cerebellar SHANK2 staining (right). Scale bar, 200 μm. (b) Staining for SHANK2 in the hippocampus and cerebellum of the global Shank2^−/− shows absence of expression. (c) The L7-Shank2^−/− shows expression in the hippocampus, but not in the molecular layer of the cerebellum.

(a,b) After 3 days of training, WT (n=9) but not L7-Shank2^−/− mice (n=6) (P=0.018) learned to increase their speed during a conditioned ErasmusLadder test using tone-cued rung displacements. (c,d) In vestibulo-ocular reflex (VOR) gain increase training, L7-Shank2^−/− mice (n=8) were not able to adapt their gain like WTs did (n=7) (P=0.006). (e,f) L7-Shank2^−/− mutants (n=9) did not adapt their VOR phase following a reversal training paradigm, whereas WT (n=9) did (second day, P=0.047; third, P=0.0013; fourth, P<0.0001; fifth, P=0.0003). (g–i) Impaired percentage (P=0.0013) and amplitude (P=0.0009) of conditioned responses (CRs) in L7-Shank2^−/− mice (n=10) compared with WT (n=11) in an eye-blink conditioning paradigm (200 paired trials daily). Data are represented as mean±s.e.m.. Single and double asterisks indicate P<0.05 and P<0.01, respectively. All tests were repeated-measures ANOVAs.

(a) Extracellular PC traces recorded from anterior lobules (I–V) in WT (top) and L7-Shank2^−/− (bottom) mice. (b–d) Firing characteristics in the anterior lobules reveal a difference in CV2 (P=0.0092) between L7-Shank2^−/− (n=25/3, cells per animals) and WT (n=23/3). (e) Example PC traces from posterior lobules (IX–X). (f–h) PCs from the posterior cerebellum in L7-Shank2^−/− (n=21/3) showed significantly higher CV (P<0.0001) and CV2 (P<0.0001) values compared with WT (n=25/3). (i) Example PC traces from posteriorly located flocculus, which is responsible for VOR learning. (j–l) Again, L7-Shank2^−/− (n=19/2) shows significantly higher CV (P<0.0001) and CV2 (P<0.0001) values than WT (n=23/2). Double asterisks denote P<0.01. All tests were two-sided t-tests.

(a) Three-chamber social interaction evaluated by relative time spent in each chamber. WTs (n=16) prefer to spend time in the room with the stranger 1 mouse (S1), compared with the empty room (P=0.0002, MWU-test). (b) This was not the case for L7-Shank2^−/− (n=17) mice (P=0.7, MWU-test). (c) The preference index (S1-empty) confirms the difference between genotypes (P=0.021). (d) Following the introduction of a second stranger (S2), WTs (n=16) prefer to spend time in the chamber with S2 compared with that with S1 (P=0.0001, MWU-test). (e) The L7-Shank2^−/− mice (n=17) did not show a preference for newly introduced S2 (P=0.1, MWU-test). (f) The S1−S2 preference index indicates that WTs prefer S2 more than L7-Shank2^−/− do (P=0.013). (g) No difference was found in a marble-burying task indicative of anxious and/or repetitive behaviour (WT, n=16; L7-Shank2^−/−, n=17, P=1.0). (h) L7-Shank2^−/− (n=16) seemed to trend towards less grooming than WTs (n=13) (P=0.054). (i) T-maze paradigm showed less consecutive alternations in L7-Shank2^−/− (n=17) compared with WT (n=16) (P=0.0023, MWU-test) indicating repetitive decision-making. Data are presented as mean±s.e.m. Single and double asterisks indicate P<0.05 and P<0.01, respectively. Two-sided t-tests were used, unless stated otherwise. NS, not significant.