The pseudokinase CaMKv is required for the activity-dependent maintenance of dendritic spines

Abstract

Dendritic spine stabilization depends on afferent synaptic input and requires changes in actin cytoskeleton dynamics and protein synthesis. However, the underlying molecular mechanism remains unclear. Here we report the identification of 'calmodulin kinase-like vesicle-associated' (CaMKv), a pseudokinase of the CaMK family with unknown function, as a synaptic protein crucial for dendritic spine maintenance. CaMKv mRNA localizes at dendrites, and its protein synthesis is regulated by neuronal activity. CaMKv function is inhibited upon phosphorylation by cyclin-dependent kinase 5 (Cdk5) at Thr345. Furthermore, CaMKv knockdown in mouse hippocampal CA1 pyramidal neurons impairs synaptic transmission and plasticity in vivo, resulting in hyperactivity and spatial memory impairment. These findings collectively indicate that the precise regulation of CaMKv through activity-dependent synthesis and post-translational phosphorylation is critical for dendritic spine maintenance, revealing an unusual signalling pathway in the regulation of synaptic transmission and brain function that involves a pseudokinase.

Figure 1. CaMKv protein synthesis at synapses is regulated by neuronal activity.

(a) Western blot analysis of CaMKv expression in the mouse forebrain at the indicated stages. (b) Co-localization between CaMKv (red) and PSD-95 (green) puncta (arrowheads) in the dendrites of cultured hippocampal neurons (23 days in vitro (DIV)). Scale bars, 20 μm (upper); 10 μm (lower panels). (c) CaMKv expression in the mouse visual cortex before (postnatal day (P) 10) and after eye-opening (P13). Arc protein increase upon eye-opening served as a positive control. (d) Quantification of CaMKv protein normalized to that of actin (n=3 mice per condition, *P<0.05, Student's t-test). (e) CaMKv expression in the mouse visual cortex was reduced after dark rearing (DR). (f) CaMKv protein was quantified after normalization to that of GAPDH (n=11 mice per condition, *P<0.05, Student's t-test). (g,h) Cortical neurons (14–16 DIV) were treated with Bic (40 μM) or TTX (2 μM) for the indicated durations. (i) CaMKv expression (normalized to actin) was increased by Bic and decreased by TTX (n=3 independent experiments). (j,k) Bic-induced CaMKv expression required protein synthesis. Cortical neurons (14 DIV) were pretreated with anisomycin (40 μM) for 1 h and incubated with Bic (40 μM) or vehicle (Veh) for 1 h (*P<0.05, one-way analysis of variance, Bonferroni's multiple comparison test; n=3 independent experiments). (l) camkv mRNA was localized at the distal dendrites (>100 μm from cell bodies) of dissociated hippocampal neurons (23 DIV) as indicated by fluorescence in situ hybridization. Arrowheads indicate camkv mRNA (green) along MAP2-positive dendrites (red). Scale bars, 20 μm (upper); 10 μm (lower). (m) The postsynaptic marker PSD-95 was highly enriched in the synaptoneurosome fraction (SNS) but absent in the cytosolic fraction (Cyto). (n) CaMKv was locally translated in the synaptoneurosome as revealed by metabolic labelling; its synthesis was abolished by cycloheximide (CHX, 10 ng μl⁻¹).

Figure 2. CaMKv is required for dendritic spine maintenance and synaptic transmission.

(a) Western blot analysis of CaMKv expression in cortical neurons transfected with pSUPER vector (Vec), shRNAs (shCaMKv) or the corresponding scrambled control (Scr). (b,c) CaMKv knockdown significantly decreased dendritic spine density. The spine loss was rescued by co-expressing the RNAi-resistant wild-type (WT) CaMKv but not the A316R mutant (scale bar, 10 μm; n=39 dendrites per condition, ***P<0.001, one-way analysis of variance (ANOVA), Tukey's multiple comparison test). (d) The pull-down experiment indicated that WT CaMKv but not the A316R mutant interacted with calmodulin (CaM). (e–h) CaMKv was required for synaptic transmission. (e) Representative miniature excitatory postsynaptic current (mEPSC) traces. (f,h) CaMKv knockdown significantly decreased mEPSC frequency, which was restored by co-expressing WT CaMKv. (g) No significant difference in mEPSC amplitude was observed among conditions (Scr: n=27 neurons, shCaMKv: n=29 neurons, WT CaMKv rescue: n=28 neurons, **P<0.01, ***P<0.001, one-way ANOVA, Tukey's multiple comparison test).

Figure 3. CaMKv regulates spine morphogenesis via RhoA inhibition.

(a–c) CaMKv knockdown dramatically increased RhoA activity but not Rac1 or Cdc42. (d) The RhoA-GEF Lfc, but not the other GEFs TIAM1 or Ephexin1, was pulled down from the synaptosome by FLAG-tagged CaMKv. (e) CaMKv was co-immunoprecipitated with Lfc in brain homogenate and synaptosome fractions. Immunoprecipitation by IgG served as the negative control. (f) FLAG-tagged Lfc and HA-tagged RhoA were co-expressed in HEK293T cells, and the lysate was subsequently incubated with recombinant GST-tagged CaMKv CaM-binding domain (CaM) or the CaMKv PEST domain. Lfc–RhoA interaction was specifically disrupted by the CaMKv CaM-binding domain but not the PEST domain. (g,h) Hippocampal neurons transfected with CaMKv shRNA or Scr were treated with the ROCK inhibitor Y27632 (10 μM) or vehicle (Veh) at 17 DIV for 6 h. Y27632 abolished the spine loss induced by CaMKv shRNA (scale bar, 10 μm; n=30 dendrites per condition, ***P<0.001, one-way analysis of variance (ANOVA), Tukey's multiple comparison test). (i,j) Cultured hippocampal neurons were treated with the AMPA receptor blocker NBQX (20 μM) from 9 to 16 DIV. NBQX reduced CaMKv expression (n=3 independent experiments, *P<0.05, Student's t-test). (k,l) Hippocampal neurons transfected at 7 DIV with the indicated constructs were treated with NBQX (20 μM) or vehicle (Veh) from 9 to 16 DIV. The spine loss on long-term AMPA receptor blockade was rescued by CaMKv overexpression (scale bar, 10 μm; n=30 dendrites per condition, ***P<0.001, one-way ANOVA, Tukey's multiple comparison test).

Figure 4. CaMKv phosphorylation at Thr345 by Cdk5 and its role in spine morphogenesis.

(a) Schematic diagram showing the different phospho-serine and phospho-threonine residues of CaMKv. Thr345 is conserved in the human, mouse, and rat orthologues. (b) Phospho-specific antibody against CaMKv Thr345 (P-CaMKv) revealed the induction of Thr345 phosphorylation of FLAG-tagged CaMKv by recombinant Cdk5/p35. The signal was abolished in the T345A phospho-deficient CaMKv mutant. (c,d) Thr345 phosphorylation was significantly reduced in p35^−/− adult mouse brains. (p35^+/+: n=4 mice, p35^−/−: n=5 mice, *P<0.05, Student's t-test). (e) CaMKv phosphorylation and expression levels in the mouse forebrain at different developmental stages. (f,g) CaMKv phosphorylation decreased significantly in the visual cortex after eye-opening (P13). CaMKv phosphorylation levels were normalized to that of total CaMKv (n=3 mice per condition, *P<0.05, Student's t-test). (h,i) CaMKv phosphorylation significantly increased in the mouse visual cortex after DR (n=11 mice per condition, **P<0.01, Student's t-test). (j,k) CaMKv phosphorylation increased on activity blockade. Cortical neurons (14–16 DIV) were treated with TTX (2 μM) for the indicated durations and quantified after normalization to total CaMKv (n=3 independent experiments). (l,m) The phospho-mimetic CaMKv-T345E mutant failed to rescue the spine defects on CaMKv knockdown by shRNA (scale bar, 10 μm; n=33 dendrites per condition, ***P<0.001, one-way analysis of variance, Tukey's multiple comparison test).

Figure 5. CaMKv knockdown in vivo reduces spine density and impairs basal synaptic neurotransmission and L-LTP.

(a) Representative immunostained coronal sections from mice subjected to bilateral lentivirus injections (scale bar, 500 μm). (b,c) The indicated GFP-positive CA1 region was dissected to examine CaMKv levels. CaMKv knockdown (shCaMKv) in the hippocampal CA1 region significantly reduced CaMKv expression (Con: n=4 mice, shCaMKv: n=6 mice, *P<0.05, Student's t-test). (d) Hippocampal CA1 pyramidal neurons were transduced by control or lentiviral vectors expressing CaMKv shRNA. CaMKv knockdown in vivo reduced CaMKv expression (red) in CA1 neurons (scale bar, 25 μm; arrowheads: GFP-positive neurons; dotted area: magnified images of infected neurons). (e,f) CaMKv knockdown in vivo significantly reduced dendritic spine density. (scale bar, 20 μm; Con: n=34 dendrites, shCaMKv: n=21 dendrites, ***P<0.001, Student's t-test; green; GFP, red: CaMKv, blue: 4,6-diamidino-2-phenylindole (DAPI)). (g) CaMKv knockdown in hippocampal CA1 neurons inhibited basal synaptic neurotransmission. The input–output curve was altered in the infected CA1 region after CaMKv knockdown (Con: n=8 mice, shCaMKv: n=9 mice). (h) CaMKv knockdown impaired the L-LTP. (i) The field excitatory postsynaptic potential (fEPSP) amplitude (normalized to baseline) was significantly reduced in hippocampi injected with CaMKv shRNA (Con: n=8 mice, shCaMKv: n=9 mice, *P<0.05, Student's t-test).

Figure 6. CaMKv knockdown in hippocampal CA1 neurons leads to hyperactivity and reduced MWM performance.

(a–c) CaMKv knockdown mice exhibited hyperlocomotion in the open-field test, significantly increased total running distance and central/total duration (Con: n=9 mice, shCaMKv: n=12 mice, **P<0.01, Student's t-test). (d–g) CaMKv knockdown reduced MWM performance. (d) Mice were trained in the MWM and tested in a probe trial on day 8. The acquisition over 7 days (four trials per day) was similar between groups. (e) Representative swimming paths of a control mouse and a CaMKv knockdown mouse are shown. (f) In the probe trial, control mice spent significantly more time in the target quadrant (T) than any other quadrant (T versus L, T versus O or T versus R, ***P<0.001, one-way analysis of variance (ANOVA), Tukey's multiple comparison test). The CaMKv knockdown mice spent comparable time in T as the other quadrants at the level of chance (T versus L, T versus O or T versus R, P>0.05, one-way ANOVA, Tukey's multiple comparison test). Time spent in T was significantly lower in the CaMKv knockdown mice than the controls (Con: n=14 mice, shCaMKv: n=18 mice, ***P<0.001, two-way ANOVA, genotype × platform: F_(3,90)=5.49, P<0.01, genotype: F_(1,30)=−0.4835, platform: F_(3,90)=16.05, P<0.0001, Bonferroni's multiple comparison test). (g) The cumulative distance of the swimming path to the platform was greater in CaMKv knockdown mice than the controls (Con: n=14 mice, shCaMKv: n=18 mice, ***P<0.001, two-way ANOVA, genotype × platform: F_(3,120)=6.48, P<0.001, genotype: F_(1,120)=−2.16, platform: F_(3,120)=17.92, P<0.0001, Bonferroni's multiple comparison test).